{
  "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": 12,
    "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_s12.json",
    "output_lr": 0.1,
    "pert_directions": 1,
    "pert_every": 4,
    "pert_sigma": 0.01,
    "perturb_seed": 1000,
    "predictor_every": 0,
    "predictor_mode": "nlms",
    "predictor_warmup_steps": 0,
    "residual_scale": null,
    "seed": 12,
    "split_seed": 2027,
    "traffic_calibration_examples": 64,
    "traffic_ratio": 4.0,
    "traffic_rule": "innovation",
    "traffic_seed": 4000,
    "train_limit": 0,
    "use_residual": 1,
    "val_examples": 0,
    "vectorizer_mode": "spatial_template",
    "warmup_epochs": 0,
    "weight_decay": 0.0001,
    "weight_scale": 1.0,
    "width": 16
  },
  "calibration_metric_space": "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": 0.9995090001159244,
    "feedback_forward_cosine": [
      0.999492347240448,
      0.9993671178817749,
      0.9994820356369019,
      0.9994138479232788,
      0.9995945692062378,
      0.9995646476745605,
      0.9993652105331421,
      0.9993512034416199,
      0.9994674921035767,
      0.9993777275085449,
      0.9995577931404114,
      0.9995392560958862,
      0.9995492696762085,
      0.9994874000549316,
      0.999705970287323,
      0.9996337294578552,
      0.9997380971908569,
      0.9995861649513245,
      0.9996955990791321,
      0.9997649192810059,
      0.9995940923690796,
      0.9995415210723877,
      0.9996059536933899,
      0.9995549917221069,
      0.9996305704116821,
      0.9995378255844116,
      0.9996811151504517,
      0.9995619654655457,
      0.9997225999832153,
      0.9996045827865601,
      0.9996874928474426,
      0.999572217464447,
      0.9996489882469177,
      0.9994925260543823,
      0.9996796250343323,
      0.9995327591896057,
      0.9997528195381165,
      0.9998396635055542,
      0.9998064041137695,
      0.999785304069519,
      0.9998205304145813,
      0.9997357130050659,
      0.999761700630188,
      0.999605655670166,
      0.9997472763061523,
      0.9995254278182983,
      0.9993584752082825,
      0.9989626407623291,
      0.9987921714782715,
      0.9983224868774414,
      0.9982134103775024,
      0.9976545572280884,
      0.9983347058296204,
      0.997771680355072,
      0.9999784231185913
    ],
    "feedback_forward_norm_ratio": [
      1.000089406967163,
      1.0017415285110474,
      0.9985799789428711,
      1.0014532804489136,
      0.9992918968200684,
      1.000859022140503,
      0.9995941519737244,
      1.001286268234253,
      0.9982871413230896,
      1.000485897064209,
      0.9979104399681091,
      1.000905156135559,
      0.9982880353927612,
      1.001328706741333,
      0.9984083771705627,
      1.0006762742996216,
      0.9997565746307373,
      1.0016711950302124,
      0.9993085265159607,
      1.0002381801605225,
      0.9990171790122986,
      1.0010908842086792,
      0.9994428157806396,
      1.0014185905456543,
      0.9996382594108582,
      1.0020818710327148,
      0.9992923140525818,
      1.0016818046569824,
      0.9992808699607849,
      1.0008541345596313,
      0.9991425275802612,
      1.0014350414276123,
      0.9993600249290466,
      1.0013771057128906,
      0.999732494354248,
      1.0014729499816895,
      0.9994292259216309,
      1.000601053237915,
      1.0000132322311401,
      1.0007882118225098,
      0.9997448325157166,
      1.0010770559310913,
      0.9994710087776184,
      1.0012922286987305,
      0.9996737241744995,
      1.0013598203659058,
      0.9986661076545715,
      1.0019320249557495,
      0.9978883266448975,
      1.0027220249176025,
      0.9968799948692322,
      1.0033371448516846,
      0.997822105884552,
      1.0025888681411743,
      1.0013923645019531
    ],
    "feedback_forward_relative_error": [
      0.03186540678143501,
      0.03565174341201782,
      0.03219625726342201,
      0.034293968230485916,
      0.028475996106863022,
      0.029532907530665398,
      0.03562766686081886,
      0.03606780245900154,
      0.03265102952718735,
      0.03528871014714241,
      0.029782500118017197,
      0.030384186655282974,
      0.030047781765460968,
      0.032069332897663116,
      0.02427857369184494,
      0.02707965299487114,
      0.022884933277964592,
      0.028844140470027924,
      0.02467583678662777,
      0.021682925522327423,
      0.02849097177386284,
      0.03031676448881626,
      0.02807055599987507,
      0.029889831319451332,
      0.027179697528481483,
      0.030506731942296028,
      0.02525198645889759,
      0.029669566079974174,
      0.023555569350719452,
      0.028146808966994286,
      0.025006014853715897,
      0.029306400567293167,
      0.026494760066270828,
      0.031911738216876984,
      0.025312673300504684,
      0.030629947781562805,
      0.022243095561861992,
      0.017927750945091248,
      0.01967710815370083,
      0.02074536494910717,
      0.018943214789032936,
      0.023028654977679253,
      0.02183547243475914,
      0.02812952548265457,
      0.022486107423901558,
      0.03086264058947563,
      0.03581830486655235,
      0.04563266783952713,
      0.04914337396621704,
      0.05806482210755348,
      0.05976491793990135,
      0.06868523359298706,
      0.05769040808081627,
      0.06689479947090149,
      0.006721368059515953
    ],
    "innovation_negative_gradient_cosine": [
      0.9995743632316589,
      0.9992213249206543,
      0.9995816946029663,
      0.9994047284126282,
      0.9995815753936768,
      0.9994869828224182,
      0.9995934963226318,
      0.9992918968200684,
      0.9995976090431213,
      0.9993172883987427,
      0.9996025562286377,
      0.9994415044784546,
      0.9996123909950256,
      0.9994906187057495,
      0.9996334314346313,
      0.9995494484901428,
      0.9996424913406372,
      0.9995386004447937,
      0.9996543526649475,
      0.999616265296936,
      0.9996732473373413,
      0.9994845390319824,
      0.9996790885925293,
      0.9995281100273132,
      0.999685525894165,
      0.9995770454406738,
      0.9996961355209351,
      0.9995557069778442,
      0.9997034072875977,
      0.9996285438537598,
      0.9997194409370422,
      0.9996160268783569,
      0.9997323751449585,
      0.9995778203010559,
      0.9997451305389404,
      0.9996391534805298,
      0.9997557401657104,
      0.9997762441635132,
      0.9997713565826416,
      0.9997782707214355,
      0.9997873902320862,
      0.9997929930686951,
      0.9997757077217102,
      0.9997889995574951,
      0.9998126029968262,
      0.9998635053634644,
      0.9997594356536865,
      0.999828577041626,
      0.9996746778488159,
      0.9996174573898315,
      0.9996563792228699,
      0.9995285272598267,
      0.9997157454490662,
      0.9997916221618652,
      0.6614387035369873
    ],
    "mean_feedback_forward_cosine": 0.9994391322135925,
    "mean_feedback_forward_relative_error": 0.031407567316835576,
    "normalization_state": "training_batch_stats_without_running_update",
    "raw_negative_gradient_cosine": [
      0.9995743632316589,
      0.9992213249206543,
      0.9995816946029663,
      0.9994047284126282,
      0.9995815753936768,
      0.9994869828224182,
      0.9995934963226318,
      0.9992918968200684,
      0.9995976090431213,
      0.9993172883987427,
      0.9996025562286377,
      0.9994415044784546,
      0.9996123909950256,
      0.9994906187057495,
      0.9996334314346313,
      0.9995494484901428,
      0.9996424913406372,
      0.9995386004447937,
      0.9996543526649475,
      0.999616265296936,
      0.9996732473373413,
      0.9994845390319824,
      0.9996790885925293,
      0.9995281100273132,
      0.999685525894165,
      0.9995770454406738,
      0.9996961355209351,
      0.9995557069778442,
      0.9997034072875977,
      0.9996285438537598,
      0.9997194409370422,
      0.9996160268783569,
      0.9997323751449585,
      0.9995778203010559,
      0.9997451305389404,
      0.9996391534805298,
      0.9997557401657104,
      0.9997762441635132,
      0.9997713565826416,
      0.9997782707214355,
      0.9997873902320862,
      0.9997929930686951,
      0.9997757077217102,
      0.9997889995574951,
      0.9998126029968262,
      0.9998635053634644,
      0.9997594356536865,
      0.999828577041626,
      0.9996746778488159,
      0.9996174573898315,
      0.9996563792228699,
      0.9995285272598267,
      0.9997157454490662,
      0.9997916221618652,
      0.6614387035369873
    ],
    "teaching_negative_gradient_cosine": [
      0.9995743632316589,
      0.9992213249206543,
      0.9995816946029663,
      0.9994047284126282,
      0.9995815753936768,
      0.9994869828224182,
      0.9995934963226318,
      0.9992918968200684,
      0.9995976090431213,
      0.9993172883987427,
      0.9996025562286377,
      0.9994415044784546,
      0.9996123909950256,
      0.9994906187057495,
      0.9996334314346313,
      0.9995494484901428,
      0.9996424913406372,
      0.9995386004447937,
      0.9996543526649475,
      0.999616265296936,
      0.9996732473373413,
      0.9994845390319824,
      0.9996790885925293,
      0.9995281100273132,
      0.999685525894165,
      0.9995770454406738,
      0.9996961355209351,
      0.9995557069778442,
      0.9997034072875977,
      0.9996285438537598,
      0.9997194409370422,
      0.9996160268783569,
      0.9997323751449585,
      0.9995778203010559,
      0.9997451305389404,
      0.9996391534805298,
      0.9997557401657104,
      0.9997762441635132,
      0.9997713565826416,
      0.9997782707214355,
      0.9997873902320862,
      0.9997929930686951,
      0.9997757077217102,
      0.9997889995574951,
      0.9998126029968262,
      0.9998635053634644,
      0.9997594356536865,
      0.999828577041626,
      0.9996746778488159,
      0.9996174573898315,
      0.9996563792228699,
      0.9995285272598267,
      0.9997157454490662,
      0.9997916221618652,
      0.6614387035369873
    ],
    "wall_s": 0.191314697265625
  },
  "epochs": [
    {
      "epoch": 1,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3244383335113525,
        "mean_feedback_forward_cosine": 0.4198355341499502,
        "mean_feedback_forward_relative_error": 1.0560421732338992,
        "min_feedback_forward_cosine": 0.127158984541893
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 391,
      "train_examples": 50000,
      "train_loss": 2.4138239181900025
    },
    {
      "epoch": 2,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3209409713745117,
        "mean_feedback_forward_cosine": 0.4342506834051826,
        "mean_feedback_forward_relative_error": 1.0433166081255132,
        "min_feedback_forward_cosine": 0.1320267617702484
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 782,
      "train_examples": 50000,
      "train_loss": 1.7105659466934204
    },
    {
      "epoch": 3,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.3159164190292358,
        "mean_feedback_forward_cosine": 0.4584254817529158,
        "mean_feedback_forward_relative_error": 1.0215907725420865,
        "min_feedback_forward_cosine": 0.13956084847450256
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1173,
      "train_examples": 50000,
      "train_loss": 1.4826833779144286
    },
    {
      "epoch": 4,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.308376431465149,
        "mean_feedback_forward_cosine": 0.4871980583125895,
        "mean_feedback_forward_relative_error": 0.9948613367297433,
        "min_feedback_forward_cosine": 0.14930999279022217
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1564,
      "train_examples": 50000,
      "train_loss": 1.27878708026886
    },
    {
      "epoch": 5,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2998486757278442,
        "mean_feedback_forward_cosine": 0.5155361040072007,
        "mean_feedback_forward_relative_error": 0.9674677659164775,
        "min_feedback_forward_cosine": 0.1607993245124817
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 1955,
      "train_examples": 50000,
      "train_loss": 1.0934885717010498
    },
    {
      "epoch": 6,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2901713848114014,
        "mean_feedback_forward_cosine": 0.5428229207342321,
        "mean_feedback_forward_relative_error": 0.9398032789880579,
        "min_feedback_forward_cosine": 0.1737784445285797
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2346,
      "train_examples": 50000,
      "train_loss": 0.9438204985427856
    },
    {
      "epoch": 7,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2803062200546265,
        "mean_feedback_forward_cosine": 0.567431657964533,
        "mean_feedback_forward_relative_error": 0.9138331499966708,
        "min_feedback_forward_cosine": 0.1865416318178177
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 2737,
      "train_examples": 50000,
      "train_loss": 0.8270858850860596
    },
    {
      "epoch": 8,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2699190378189087,
        "mean_feedback_forward_cosine": 0.5901177953590047,
        "mean_feedback_forward_relative_error": 0.8889968286861073,
        "min_feedback_forward_cosine": 0.19944104552268982
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3128,
      "train_examples": 50000,
      "train_loss": 0.7495757960128784
    },
    {
      "epoch": 9,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2612634897232056,
        "mean_feedback_forward_cosine": 0.610681467977437,
        "mean_feedback_forward_relative_error": 0.865493153442036,
        "min_feedback_forward_cosine": 0.20822472870349884
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3519,
      "train_examples": 50000,
      "train_loss": 0.679872678489685
    },
    {
      "epoch": 10,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2559030055999756,
        "mean_feedback_forward_cosine": 0.6305192985317923,
        "mean_feedback_forward_relative_error": 0.8422314649278467,
        "min_feedback_forward_cosine": 0.21694274246692657
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 3910,
      "train_examples": 50000,
      "train_loss": 0.6310647740745544
    },
    {
      "epoch": 11,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2501100301742554,
        "mean_feedback_forward_cosine": 0.6483281940221787,
        "mean_feedback_forward_relative_error": 0.8205998350273479,
        "min_feedback_forward_cosine": 0.22608375549316406
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4301,
      "train_examples": 50000,
      "train_loss": 0.5876526419639587
    },
    {
      "epoch": 12,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2440547943115234,
        "mean_feedback_forward_cosine": 0.6655318002809177,
        "mean_feedback_forward_relative_error": 0.7990088679573752,
        "min_feedback_forward_cosine": 0.23492339253425598
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 4692,
      "train_examples": 50000,
      "train_loss": 0.5574879420471192
    },
    {
      "epoch": 13,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2371782064437866,
        "mean_feedback_forward_cosine": 0.6813188051635569,
        "mean_feedback_forward_relative_error": 0.7786574320359664,
        "min_feedback_forward_cosine": 0.24500702321529388
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5083,
      "train_examples": 50000,
      "train_loss": 0.5174666587638855
    },
    {
      "epoch": 14,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.229864478111267,
        "mean_feedback_forward_cosine": 0.6966934740543366,
        "mean_feedback_forward_relative_error": 0.7583921773867174,
        "min_feedback_forward_cosine": 0.2564243674278259
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5474,
      "train_examples": 50000,
      "train_loss": 0.4944457879543304
    },
    {
      "epoch": 15,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2204899787902832,
        "mean_feedback_forward_cosine": 0.7118308224461295,
        "mean_feedback_forward_relative_error": 0.7377489366314628,
        "min_feedback_forward_cosine": 0.26965150237083435
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 5865,
      "train_examples": 50000,
      "train_loss": 0.4681345001411438
    },
    {
      "epoch": 16,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.21138334274292,
        "mean_feedback_forward_cosine": 0.7262426392598585,
        "mean_feedback_forward_relative_error": 0.7175799703056162,
        "min_feedback_forward_cosine": 0.28238144516944885
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 6256,
      "train_examples": 50000,
      "train_loss": 0.45293981085777285
    },
    {
      "epoch": 17,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.2019273042678833,
        "mean_feedback_forward_cosine": 0.7394660765474493,
        "mean_feedback_forward_relative_error": 0.6985175696286288,
        "min_feedback_forward_cosine": 0.2950320839881897
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 6647,
      "train_examples": 50000,
      "train_loss": 0.4310007293605804
    },
    {
      "epoch": 18,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1909077167510986,
        "mean_feedback_forward_cosine": 0.7525688263502988,
        "mean_feedback_forward_relative_error": 0.6793072627349334,
        "min_feedback_forward_cosine": 0.30850374698638916
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7038,
      "train_examples": 50000,
      "train_loss": 0.4120278524303436
    },
    {
      "epoch": 19,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1793495416641235,
        "mean_feedback_forward_cosine": 0.7650805679234591,
        "mean_feedback_forward_relative_error": 0.6604170032522895,
        "min_feedback_forward_cosine": 0.3222784996032715
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7429,
      "train_examples": 50000,
      "train_loss": 0.3957406766605377
    },
    {
      "epoch": 20,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1666817665100098,
        "mean_feedback_forward_cosine": 0.7772322437979958,
        "mean_feedback_forward_relative_error": 0.6416691671718251,
        "min_feedback_forward_cosine": 0.33726298809051514
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 7820,
      "train_examples": 50000,
      "train_loss": 0.3890600772857666
    },
    {
      "epoch": 21,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.15311861038208,
        "mean_feedback_forward_cosine": 0.7886456803842025,
        "mean_feedback_forward_relative_error": 0.6236146463589235,
        "min_feedback_forward_cosine": 0.353469580411911
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8211,
      "train_examples": 50000,
      "train_loss": 0.36867258852005
    },
    {
      "epoch": 22,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1378247737884521,
        "mean_feedback_forward_cosine": 0.7999685997312719,
        "mean_feedback_forward_relative_error": 0.6052621527151628,
        "min_feedback_forward_cosine": 0.3707684278488159
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8602,
      "train_examples": 50000,
      "train_loss": 0.3646633823204041
    },
    {
      "epoch": 23,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.121764898300171,
        "mean_feedback_forward_cosine": 0.8108454075726595,
        "mean_feedback_forward_relative_error": 0.5872186479243365,
        "min_feedback_forward_cosine": 0.38909459114074707
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 8993,
      "train_examples": 50000,
      "train_loss": 0.3533440400314331
    },
    {
      "epoch": 24,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.1063214540481567,
        "mean_feedback_forward_cosine": 0.821115351265127,
        "mean_feedback_forward_relative_error": 0.5697222119027918,
        "min_feedback_forward_cosine": 0.40641462802886963
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9384,
      "train_examples": 50000,
      "train_loss": 0.3383803585338592
    },
    {
      "epoch": 25,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0895359516143799,
        "mean_feedback_forward_cosine": 0.8312289080836556,
        "mean_feedback_forward_relative_error": 0.552090187235312,
        "min_feedback_forward_cosine": 0.42520034313201904
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 9775,
      "train_examples": 50000,
      "train_loss": 0.3309914978790283
    },
    {
      "epoch": 26,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.073138952255249,
        "mean_feedback_forward_cosine": 0.8406367025592111,
        "mean_feedback_forward_relative_error": 0.5351539647037332,
        "min_feedback_forward_cosine": 0.4425792098045349
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10166,
      "train_examples": 50000,
      "train_loss": 0.320999254283905
    },
    {
      "epoch": 27,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0543512105941772,
        "mean_feedback_forward_cosine": 0.8497189771045338,
        "mean_feedback_forward_relative_error": 0.5185014180161737,
        "min_feedback_forward_cosine": 0.462225079536438
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10557,
      "train_examples": 50000,
      "train_loss": 0.31371587188243866
    },
    {
      "epoch": 28,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.0364478826522827,
        "mean_feedback_forward_cosine": 0.8583152397112412,
        "mean_feedback_forward_relative_error": 0.5023321167989211,
        "min_feedback_forward_cosine": 0.48109471797943115
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 10948,
      "train_examples": 50000,
      "train_loss": 0.3024843281364441
    },
    {
      "epoch": 29,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 1.017153024673462,
        "mean_feedback_forward_cosine": 0.8665038390593095,
        "mean_feedback_forward_relative_error": 0.48660044859756124,
        "min_feedback_forward_cosine": 0.5008649826049805
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11339,
      "train_examples": 50000,
      "train_loss": 0.2939609637641907
    },
    {
      "epoch": 30,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9980878829956055,
        "mean_feedback_forward_cosine": 0.8744109912352128,
        "mean_feedback_forward_relative_error": 0.4709355667233467,
        "min_feedback_forward_cosine": 0.5197159051895142
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 11730,
      "train_examples": 50000,
      "train_loss": 0.28885329203605653
    },
    {
      "epoch": 31,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9772093892097473,
        "mean_feedback_forward_cosine": 0.8819113514640115,
        "mean_feedback_forward_relative_error": 0.4558220600539988,
        "min_feedback_forward_cosine": 0.5398478507995605
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12121,
      "train_examples": 50000,
      "train_loss": 0.2821906175947189
    },
    {
      "epoch": 32,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9556038975715637,
        "mean_feedback_forward_cosine": 0.8891913132234053,
        "mean_feedback_forward_relative_error": 0.4406657814979553,
        "min_feedback_forward_cosine": 0.5601248741149902
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12512,
      "train_examples": 50000,
      "train_loss": 0.2786395811319351
    },
    {
      "epoch": 33,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9322633147239685,
        "mean_feedback_forward_cosine": 0.8961453188549389,
        "mean_feedback_forward_relative_error": 0.4259107045151971,
        "min_feedback_forward_cosine": 0.5821226239204407
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 12903,
      "train_examples": 50000,
      "train_loss": 0.27072911348342893
    },
    {
      "epoch": 34,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.9096354842185974,
        "mean_feedback_forward_cosine": 0.9028071836991743,
        "mean_feedback_forward_relative_error": 0.41126837974244895,
        "min_feedback_forward_cosine": 0.6022782921791077
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13294,
      "train_examples": 50000,
      "train_loss": 0.26704496616363527
    },
    {
      "epoch": 35,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8879315257072449,
        "mean_feedback_forward_cosine": 0.9087696888230063,
        "mean_feedback_forward_relative_error": 0.39780180914835495,
        "min_feedback_forward_cosine": 0.6216888427734375
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 13685,
      "train_examples": 50000,
      "train_loss": 0.25724610758781435
    },
    {
      "epoch": 36,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8641149401664734,
        "mean_feedback_forward_cosine": 0.9146937760439786,
        "mean_feedback_forward_relative_error": 0.38405355621467935,
        "min_feedback_forward_cosine": 0.6411852836608887
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14076,
      "train_examples": 50000,
      "train_loss": 0.257418413066864
    },
    {
      "epoch": 37,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.840953528881073,
        "mean_feedback_forward_cosine": 0.9202366232872009,
        "mean_feedback_forward_relative_error": 0.37086766606027427,
        "min_feedback_forward_cosine": 0.6600959300994873
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14467,
      "train_examples": 50000,
      "train_loss": 0.24879705251693726
    },
    {
      "epoch": 38,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.8156360387802124,
        "mean_feedback_forward_cosine": 0.9256891088052229,
        "mean_feedback_forward_relative_error": 0.3574880524115129,
        "min_feedback_forward_cosine": 0.6800718307495117
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 14858,
      "train_examples": 50000,
      "train_loss": 0.24602906025886537
    },
    {
      "epoch": 39,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7926400899887085,
        "mean_feedback_forward_cosine": 0.9306280038573526,
        "mean_feedback_forward_relative_error": 0.3450457442890514,
        "min_feedback_forward_cosine": 0.6978662014007568
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15249,
      "train_examples": 50000,
      "train_loss": 0.2412323922920227
    },
    {
      "epoch": 40,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7697350978851318,
        "mean_feedback_forward_cosine": 0.9352672685276379,
        "mean_feedback_forward_relative_error": 0.33294801833954724,
        "min_feedback_forward_cosine": 0.7149748206138611
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 15640,
      "train_examples": 50000,
      "train_loss": 0.2362298415184021
    },
    {
      "epoch": 41,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7469183206558228,
        "mean_feedback_forward_cosine": 0.9396627296100963,
        "mean_feedback_forward_relative_error": 0.32107209251685576,
        "min_feedback_forward_cosine": 0.7316209077835083
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16031,
      "train_examples": 50000,
      "train_loss": 0.23337628873825073
    },
    {
      "epoch": 42,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.7224966287612915,
        "mean_feedback_forward_cosine": 0.9438528407703747,
        "mean_feedback_forward_relative_error": 0.3094679505987601,
        "min_feedback_forward_cosine": 0.749053955078125
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16422,
      "train_examples": 50000,
      "train_loss": 0.23098030502319336
    },
    {
      "epoch": 43,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6987288594245911,
        "mean_feedback_forward_cosine": 0.9477465358647433,
        "mean_feedback_forward_relative_error": 0.29833831367167557,
        "min_feedback_forward_cosine": 0.7652150392532349
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 16813,
      "train_examples": 50000,
      "train_loss": 0.2250031017637253
    },
    {
      "epoch": 44,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6747825145721436,
        "mean_feedback_forward_cosine": 0.9515370965003968,
        "mean_feedback_forward_relative_error": 0.2871929381381382,
        "min_feedback_forward_cosine": 0.7807072401046753
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17204,
      "train_examples": 50000,
      "train_loss": 0.2254936946439743
    },
    {
      "epoch": 45,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6516567468643188,
        "mean_feedback_forward_cosine": 0.955012834072113,
        "mean_feedback_forward_relative_error": 0.2764912331646139,
        "min_feedback_forward_cosine": 0.7951079607009888
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17595,
      "train_examples": 50000,
      "train_loss": 0.22084568507671357
    },
    {
      "epoch": 46,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.628865659236908,
        "mean_feedback_forward_cosine": 0.9582318782806396,
        "mean_feedback_forward_relative_error": 0.2662767881696874,
        "min_feedback_forward_cosine": 0.8090265989303589
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 17986,
      "train_examples": 50000,
      "train_loss": 0.21951218831062316
    },
    {
      "epoch": 47,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.6079846620559692,
        "mean_feedback_forward_cosine": 0.9611279325051741,
        "mean_feedback_forward_relative_error": 0.2566930052908984,
        "min_feedback_forward_cosine": 0.8213279247283936
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 18377,
      "train_examples": 50000,
      "train_loss": 0.21334740138053895
    },
    {
      "epoch": 48,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5870092511177063,
        "mean_feedback_forward_cosine": 0.9639066804539074,
        "mean_feedback_forward_relative_error": 0.24725833399729294,
        "min_feedback_forward_cosine": 0.8334741592407227
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 18768,
      "train_examples": 50000,
      "train_loss": 0.2046345087623596
    },
    {
      "epoch": 49,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5655739903450012,
        "mean_feedback_forward_cosine": 0.9666054834019054,
        "mean_feedback_forward_relative_error": 0.23775543746623126,
        "min_feedback_forward_cosine": 0.8452627062797546
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19159,
      "train_examples": 50000,
      "train_loss": 0.21139758206367493
    },
    {
      "epoch": 50,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5459854602813721,
        "mean_feedback_forward_cosine": 0.9689616495912725,
        "mean_feedback_forward_relative_error": 0.22913335771723228,
        "min_feedback_forward_cosine": 0.8555338978767395
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19550,
      "train_examples": 50000,
      "train_loss": 0.2025952133321762
    },
    {
      "epoch": 51,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.526841938495636,
        "mean_feedback_forward_cosine": 0.9711984493515708,
        "mean_feedback_forward_relative_error": 0.2206283477896994,
        "min_feedback_forward_cosine": 0.8654561042785645
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 19941,
      "train_examples": 50000,
      "train_loss": 0.20171171308994293
    },
    {
      "epoch": 52,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.5071823596954346,
        "mean_feedback_forward_cosine": 0.9733563986691561,
        "mean_feedback_forward_relative_error": 0.21216394819996573,
        "min_feedback_forward_cosine": 0.8751754760742188
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20332,
      "train_examples": 50000,
      "train_loss": 0.20021353316307067
    },
    {
      "epoch": 53,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.4873877167701721,
        "mean_feedback_forward_cosine": 0.9753846992145885,
        "mean_feedback_forward_relative_error": 0.20394220799207688,
        "min_feedback_forward_cosine": 0.8846782445907593
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 20723,
      "train_examples": 50000,
      "train_loss": 0.19994908579826354
    },
    {
      "epoch": 54,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.46815577149391174,
        "mean_feedback_forward_cosine": 0.9772642558271235,
        "mean_feedback_forward_relative_error": 0.19602264843203804,
        "min_feedback_forward_cosine": 0.8935234546661377
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21114,
      "train_examples": 50000,
      "train_loss": 0.19712803175926208
    },
    {
      "epoch": 55,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.45061376690864563,
        "mean_feedback_forward_cosine": 0.9789202657612888,
        "mean_feedback_forward_relative_error": 0.18874934566291895,
        "min_feedback_forward_cosine": 0.9012744426727295
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21505,
      "train_examples": 50000,
      "train_loss": 0.1897450712442398
    },
    {
      "epoch": 56,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.43402156233787537,
        "mean_feedback_forward_cosine": 0.9804969830946488,
        "mean_feedback_forward_relative_error": 0.1815253614702008,
        "min_feedback_forward_cosine": 0.9083393812179565
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 21896,
      "train_examples": 50000,
      "train_loss": 0.1948779441356659
    },
    {
      "epoch": 57,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.41747668385505676,
        "mean_feedback_forward_cosine": 0.9819663166999817,
        "mean_feedback_forward_relative_error": 0.17460938387296415,
        "min_feedback_forward_cosine": 0.9150904417037964
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22287,
      "train_examples": 50000,
      "train_loss": 0.1941714857006073
    },
    {
      "epoch": 58,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.4008060693740845,
        "mean_feedback_forward_cosine": 0.983362311666662,
        "mean_feedback_forward_relative_error": 0.1677873245017095,
        "min_feedback_forward_cosine": 0.9216729402542114
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 22678,
      "train_examples": 50000,
      "train_loss": 0.19348405673980712
    },
    {
      "epoch": 59,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3854062855243683,
        "mean_feedback_forward_cosine": 0.9845780177549882,
        "mean_feedback_forward_relative_error": 0.16157968430356545,
        "min_feedback_forward_cosine": 0.9275352954864502
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23069,
      "train_examples": 50000,
      "train_loss": 0.18416994456768035
    },
    {
      "epoch": 60,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.36981651186943054,
        "mean_feedback_forward_cosine": 0.9857558510520241,
        "mean_feedback_forward_relative_error": 0.15531217523596502,
        "min_feedback_forward_cosine": 0.9332382678985596
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23460,
      "train_examples": 50000,
      "train_loss": 0.18529998634815217
    },
    {
      "epoch": 61,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3562355637550354,
        "mean_feedback_forward_cosine": 0.9868420557542281,
        "mean_feedback_forward_relative_error": 0.14927968003533104,
        "min_feedback_forward_cosine": 0.9379853010177612
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 23851,
      "train_examples": 50000,
      "train_loss": 0.18291772512435914
    },
    {
      "epoch": 62,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3428122401237488,
        "mean_feedback_forward_cosine": 0.9878508589484475,
        "mean_feedback_forward_relative_error": 0.14348044361580503,
        "min_feedback_forward_cosine": 0.942572832107544
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24242,
      "train_examples": 50000,
      "train_loss": 0.18351415759563447
    },
    {
      "epoch": 63,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.32837072014808655,
        "mean_feedback_forward_cosine": 0.9888015866279602,
        "mean_feedback_forward_relative_error": 0.1378271481530233,
        "min_feedback_forward_cosine": 0.9472540020942688
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 24633,
      "train_examples": 50000,
      "train_loss": 0.1803669850063324
    },
    {
      "epoch": 64,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.31547480821609497,
        "mean_feedback_forward_cosine": 0.9896420522169633,
        "mean_feedback_forward_relative_error": 0.13261645124717192,
        "min_feedback_forward_cosine": 0.9512698650360107
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25024,
      "train_examples": 50000,
      "train_loss": 0.17748355750083925
    },
    {
      "epoch": 65,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.3037189543247223,
        "mean_feedback_forward_cosine": 0.9904208053242076,
        "mean_feedback_forward_relative_error": 0.12756685563786463,
        "min_feedback_forward_cosine": 0.9548318386077881
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25415,
      "train_examples": 50000,
      "train_loss": 0.17406793964862824
    },
    {
      "epoch": 66,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2912445664405823,
        "mean_feedback_forward_cosine": 0.9911755648526278,
        "mean_feedback_forward_relative_error": 0.12248950502411886,
        "min_feedback_forward_cosine": 0.958398699760437
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 25806,
      "train_examples": 50000,
      "train_loss": 0.17906529505729676
    },
    {
      "epoch": 67,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2801404893398285,
        "mean_feedback_forward_cosine": 0.9918416305021807,
        "mean_feedback_forward_relative_error": 0.11781480559571222,
        "min_feedback_forward_cosine": 0.9614805579185486
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26197,
      "train_examples": 50000,
      "train_loss": 0.173457589635849
    },
    {
      "epoch": 68,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.27006998658180237,
        "mean_feedback_forward_cosine": 0.9924406051635742,
        "mean_feedback_forward_relative_error": 0.1134418520737778,
        "min_feedback_forward_cosine": 0.9641738533973694
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26588,
      "train_examples": 50000,
      "train_loss": 0.16886799769878388
    },
    {
      "epoch": 69,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.2597082555294037,
        "mean_feedback_forward_cosine": 0.9930091619491577,
        "mean_feedback_forward_relative_error": 0.10910528938878666,
        "min_feedback_forward_cosine": 0.9668391942977905
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 26979,
      "train_examples": 50000,
      "train_loss": 0.16880864161491393
    },
    {
      "epoch": 70,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.24950578808784485,
        "mean_feedback_forward_cosine": 0.9935406565666198,
        "mean_feedback_forward_relative_error": 0.10489608872343194,
        "min_feedback_forward_cosine": 0.969358503818512
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27370,
      "train_examples": 50000,
      "train_loss": 0.16825597145080567
    },
    {
      "epoch": 71,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.23873963952064514,
        "mean_feedback_forward_cosine": 0.9940642356872559,
        "mean_feedback_forward_relative_error": 0.10061055208471688,
        "min_feedback_forward_cosine": 0.9719176292419434
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 27761,
      "train_examples": 50000,
      "train_loss": 0.1729454252386093
    },
    {
      "epoch": 72,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.22891759872436523,
        "mean_feedback_forward_cosine": 0.9945134357972578,
        "mean_feedback_forward_relative_error": 0.09677447625859217,
        "min_feedback_forward_cosine": 0.9741685390472412
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28152,
      "train_examples": 50000,
      "train_loss": 0.16978507235527038
    },
    {
      "epoch": 73,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.21945972740650177,
        "mean_feedback_forward_cosine": 0.9949410059235313,
        "mean_feedback_forward_relative_error": 0.0929661270908334,
        "min_feedback_forward_cosine": 0.9762440919876099
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28543,
      "train_examples": 50000,
      "train_loss": 0.16630286949157716
    },
    {
      "epoch": 74,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.21060435473918915,
        "mean_feedback_forward_cosine": 0.9953264767473394,
        "mean_feedback_forward_relative_error": 0.08938674618574706,
        "min_feedback_forward_cosine": 0.9781172871589661
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 28934,
      "train_examples": 50000,
      "train_loss": 0.16400834189891816
    },
    {
      "epoch": 75,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.20204050838947296,
        "mean_feedback_forward_cosine": 0.9956956917589361,
        "mean_feedback_forward_relative_error": 0.08582023791968822,
        "min_feedback_forward_cosine": 0.9798524379730225
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29325,
      "train_examples": 50000,
      "train_loss": 0.16743798432826995
    },
    {
      "epoch": 76,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.19352245330810547,
        "mean_feedback_forward_cosine": 0.9960233428261497,
        "mean_feedback_forward_relative_error": 0.08252717652781442,
        "min_feedback_forward_cosine": 0.9815119504928589
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 29716,
      "train_examples": 50000,
      "train_loss": 0.1595230025625229
    },
    {
      "epoch": 77,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1849481463432312,
        "mean_feedback_forward_cosine": 0.9963419870896773,
        "mean_feedback_forward_relative_error": 0.07919774959710511,
        "min_feedback_forward_cosine": 0.9830996990203857
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30107,
      "train_examples": 50000,
      "train_loss": 0.16649277725219727
    },
    {
      "epoch": 78,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1773487776517868,
        "mean_feedback_forward_cosine": 0.9966184778646989,
        "mean_feedback_forward_relative_error": 0.07619481232355942,
        "min_feedback_forward_cosine": 0.9844546318054199
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30498,
      "train_examples": 50000,
      "train_loss": 0.15958376306056976
    },
    {
      "epoch": 79,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1702713668346405,
        "mean_feedback_forward_cosine": 0.9968827930363742,
        "mean_feedback_forward_relative_error": 0.07318445687944239,
        "min_feedback_forward_cosine": 0.9856667518615723
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 30889,
      "train_examples": 50000,
      "train_loss": 0.16138536895275116
    },
    {
      "epoch": 80,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.16331738233566284,
        "mean_feedback_forward_cosine": 0.9971186236901717,
        "mean_feedback_forward_relative_error": 0.0704044618728486,
        "min_feedback_forward_cosine": 0.9868065118789673
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 31280,
      "train_examples": 50000,
      "train_loss": 0.15965634570121764
    },
    {
      "epoch": 81,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1560080200433731,
        "mean_feedback_forward_cosine": 0.9973478187214244,
        "mean_feedback_forward_relative_error": 0.06760258132761175,
        "min_feedback_forward_cosine": 0.9879544973373413
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 31671,
      "train_examples": 50000,
      "train_loss": 0.15754901772499086
    },
    {
      "epoch": 82,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.14889590442180634,
        "mean_feedback_forward_cosine": 0.9975626056844538,
        "mean_feedback_forward_relative_error": 0.06485398079861295,
        "min_feedback_forward_cosine": 0.9890235662460327
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32062,
      "train_examples": 50000,
      "train_loss": 0.1604923850774765
    },
    {
      "epoch": 83,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1428852528333664,
        "mean_feedback_forward_cosine": 0.9977468035437844,
        "mean_feedback_forward_relative_error": 0.06238936851309104,
        "min_feedback_forward_cosine": 0.9898895025253296
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32453,
      "train_examples": 50000,
      "train_loss": 0.15494182635784148
    },
    {
      "epoch": 84,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.13744902610778809,
        "mean_feedback_forward_cosine": 0.9979130354794589,
        "mean_feedback_forward_relative_error": 0.06006247954951091,
        "min_feedback_forward_cosine": 0.9906392693519592
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 32844,
      "train_examples": 50000,
      "train_loss": 0.1530601090836525
    },
    {
      "epoch": 85,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.13181516528129578,
        "mean_feedback_forward_cosine": 0.9980757865038785,
        "mean_feedback_forward_relative_error": 0.057702659544619644,
        "min_feedback_forward_cosine": 0.991387128829956
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33235,
      "train_examples": 50000,
      "train_loss": 0.15579026302337645
    },
    {
      "epoch": 86,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.1261598914861679,
        "mean_feedback_forward_cosine": 0.9982293020595204,
        "mean_feedback_forward_relative_error": 0.05538785102015192,
        "min_feedback_forward_cosine": 0.9921053647994995
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 33626,
      "train_examples": 50000,
      "train_loss": 0.15822976639270783
    },
    {
      "epoch": 87,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.12078189104795456,
        "mean_feedback_forward_cosine": 0.9983660004355691,
        "mean_feedback_forward_relative_error": 0.05325097621164539,
        "min_feedback_forward_cosine": 0.992760419845581
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34017,
      "train_examples": 50000,
      "train_loss": 0.15028054527759552
    },
    {
      "epoch": 88,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.11607860028743744,
        "mean_feedback_forward_cosine": 0.9984901482408697,
        "mean_feedback_forward_relative_error": 0.05120365558700128,
        "min_feedback_forward_cosine": 0.9933149814605713
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34408,
      "train_examples": 50000,
      "train_loss": 0.15129416436910628
    },
    {
      "epoch": 89,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.11128923296928406,
        "mean_feedback_forward_cosine": 0.9986075195399198,
        "mean_feedback_forward_relative_error": 0.049207691560414706,
        "min_feedback_forward_cosine": 0.9938527941703796
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 34799,
      "train_examples": 50000,
      "train_loss": 0.14803420697927475
    },
    {
      "epoch": 90,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.10640254616737366,
        "mean_feedback_forward_cosine": 0.998720792206851,
        "mean_feedback_forward_relative_error": 0.047186382792212746,
        "min_feedback_forward_cosine": 0.9943797588348389
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35190,
      "train_examples": 50000,
      "train_loss": 0.153843375749588
    },
    {
      "epoch": 91,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.10187916457653046,
        "mean_feedback_forward_cosine": 0.9988207470286976,
        "mean_feedback_forward_relative_error": 0.0453362681648948,
        "min_feedback_forward_cosine": 0.994846522808075
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35581,
      "train_examples": 50000,
      "train_loss": 0.14836268123626709
    },
    {
      "epoch": 92,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.09747994691133499,
        "mean_feedback_forward_cosine": 0.99891292181882,
        "mean_feedback_forward_relative_error": 0.04355126805603504,
        "min_feedback_forward_cosine": 0.9952813386917114
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 35972,
      "train_examples": 50000,
      "train_loss": 0.1528516566991806
    },
    {
      "epoch": 93,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.09351164847612381,
        "mean_feedback_forward_cosine": 0.9989956454797224,
        "mean_feedback_forward_relative_error": 0.041882835277779536,
        "min_feedback_forward_cosine": 0.9956561326980591
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 36363,
      "train_examples": 50000,
      "train_loss": 0.14722164509296418
    },
    {
      "epoch": 94,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.08970483392477036,
        "mean_feedback_forward_cosine": 0.9990740808573636,
        "mean_feedback_forward_relative_error": 0.0402288471263918,
        "min_feedback_forward_cosine": 0.9960024356842041
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 36754,
      "train_examples": 50000,
      "train_loss": 0.14886850304365157
    },
    {
      "epoch": 95,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.08588581532239914,
        "mean_feedback_forward_cosine": 0.9991466793147,
        "mean_feedback_forward_relative_error": 0.03864135615188967,
        "min_feedback_forward_cosine": 0.9963352084159851
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 37145,
      "train_examples": 50000,
      "train_loss": 0.14826063853263854
    },
    {
      "epoch": 96,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.08223635703325272,
        "mean_feedback_forward_cosine": 0.9992124145681208,
        "mean_feedback_forward_relative_error": 0.037147075022486126,
        "min_feedback_forward_cosine": 0.9966393113136292
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 37536,
      "train_examples": 50000,
      "train_loss": 0.14491665635108947
    },
    {
      "epoch": 97,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0788571760058403,
        "mean_feedback_forward_cosine": 0.9992724895477295,
        "mean_feedback_forward_relative_error": 0.03572304660285061,
        "min_feedback_forward_cosine": 0.9969098567962646
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 37927,
      "train_examples": 50000,
      "train_loss": 0.14346269236564635
    },
    {
      "epoch": 98,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.07558047026395798,
        "mean_feedback_forward_cosine": 0.9993296547369523,
        "mean_feedback_forward_relative_error": 0.034297426011074675,
        "min_feedback_forward_cosine": 0.9971612691879272
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 38318,
      "train_examples": 50000,
      "train_loss": 0.14584156653642655
    },
    {
      "epoch": 99,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.07237829267978668,
        "mean_feedback_forward_cosine": 0.9993817773732272,
        "mean_feedback_forward_relative_error": 0.03295629530806433,
        "min_feedback_forward_cosine": 0.9973962903022766
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 38709,
      "train_examples": 50000,
      "train_loss": 0.14407524058818816
    },
    {
      "epoch": 100,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06947639584541321,
        "mean_feedback_forward_cosine": 0.999428694898432,
        "mean_feedback_forward_relative_error": 0.03169208192689852,
        "min_feedback_forward_cosine": 0.9975995421409607
      },
      "lr": 0.1,
      "output_lr": 0.1,
      "step": 39100,
      "train_examples": 50000,
      "train_loss": 0.14411247708320618
    },
    {
      "epoch": 101,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06945092976093292,
        "mean_feedback_forward_cosine": 0.999429096958854,
        "mean_feedback_forward_relative_error": 0.031680582599206406,
        "min_feedback_forward_cosine": 0.997601330280304
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39491,
      "train_examples": 50000,
      "train_loss": 0.07716866317510605
    },
    {
      "epoch": 102,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06943776458501816,
        "mean_feedback_forward_cosine": 0.9994292963634838,
        "mean_feedback_forward_relative_error": 0.03167427863756364,
        "min_feedback_forward_cosine": 0.9976023435592651
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 39882,
      "train_examples": 50000,
      "train_loss": 0.051544788405299184
    },
    {
      "epoch": 103,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0694250762462616,
        "mean_feedback_forward_cosine": 0.9994294773448598,
        "mean_feedback_forward_relative_error": 0.03166841171173887,
        "min_feedback_forward_cosine": 0.9976032972335815
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40273,
      "train_examples": 50000,
      "train_loss": 0.04210718454018235
    },
    {
      "epoch": 104,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06941260397434235,
        "mean_feedback_forward_cosine": 0.999429668079723,
        "mean_feedback_forward_relative_error": 0.031662963059815495,
        "min_feedback_forward_cosine": 0.9976043701171875
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 40664,
      "train_examples": 50000,
      "train_loss": 0.03675717756778002
    },
    {
      "epoch": 105,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06940004974603653,
        "mean_feedback_forward_cosine": 0.9994298501448198,
        "mean_feedback_forward_relative_error": 0.03165755365890535,
        "min_feedback_forward_cosine": 0.9976052045822144
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41055,
      "train_examples": 50000,
      "train_loss": 0.031186733311414718
    },
    {
      "epoch": 106,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0693858414888382,
        "mean_feedback_forward_cosine": 0.9994300560517745,
        "mean_feedback_forward_relative_error": 0.031651697697287255,
        "min_feedback_forward_cosine": 0.9976061582565308
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41446,
      "train_examples": 50000,
      "train_loss": 0.0309412691372633
    },
    {
      "epoch": 107,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06937147676944733,
        "mean_feedback_forward_cosine": 0.9994302467866377,
        "mean_feedback_forward_relative_error": 0.03164597680284218,
        "min_feedback_forward_cosine": 0.9976072311401367
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 41837,
      "train_examples": 50000,
      "train_loss": 0.028081531404852866
    },
    {
      "epoch": 108,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06935874372720718,
        "mean_feedback_forward_cosine": 0.9994304375215011,
        "mean_feedback_forward_relative_error": 0.03164071908051318,
        "min_feedback_forward_cosine": 0.9976081848144531
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42228,
      "train_examples": 50000,
      "train_loss": 0.02442640140235424
    },
    {
      "epoch": 109,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0693446695804596,
        "mean_feedback_forward_cosine": 0.999430613084273,
        "mean_feedback_forward_relative_error": 0.031635197048837486,
        "min_feedback_forward_cosine": 0.99760901927948
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 42619,
      "train_examples": 50000,
      "train_loss": 0.023568274050354956
    },
    {
      "epoch": 110,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06933095306158066,
        "mean_feedback_forward_cosine": 0.9994308016516945,
        "mean_feedback_forward_relative_error": 0.031630029351535166,
        "min_feedback_forward_cosine": 0.9976098537445068
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43010,
      "train_examples": 50000,
      "train_loss": 0.020625956691354513
    },
    {
      "epoch": 111,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06931696087121964,
        "mean_feedback_forward_cosine": 0.9994309826330705,
        "mean_feedback_forward_relative_error": 0.031624852019277484,
        "min_feedback_forward_cosine": 0.9976110458374023
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43401,
      "train_examples": 50000,
      "train_loss": 0.020318002663254738
    },
    {
      "epoch": 112,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06930254399776459,
        "mean_feedback_forward_cosine": 0.9994311701167714,
        "mean_feedback_forward_relative_error": 0.03161953128027645,
        "min_feedback_forward_cosine": 0.9976118803024292
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 43792,
      "train_examples": 50000,
      "train_loss": 0.018820216268151998
    },
    {
      "epoch": 113,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0692884624004364,
        "mean_feedback_forward_cosine": 0.9994313651865179,
        "mean_feedback_forward_relative_error": 0.0316144339824942,
        "min_feedback_forward_cosine": 0.9976129531860352
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44183,
      "train_examples": 50000,
      "train_loss": 0.017537187362760307
    },
    {
      "epoch": 114,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06927390396595001,
        "mean_feedback_forward_cosine": 0.9994315515864979,
        "mean_feedback_forward_relative_error": 0.03160912807692181,
        "min_feedback_forward_cosine": 0.9976140856742859
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44574,
      "train_examples": 50000,
      "train_loss": 0.01681633037008345
    },
    {
      "epoch": 115,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06925929337739944,
        "mean_feedback_forward_cosine": 0.9994317444888028,
        "mean_feedback_forward_relative_error": 0.03160384031520649,
        "min_feedback_forward_cosine": 0.9976149201393127
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 44965,
      "train_examples": 50000,
      "train_loss": 0.016025505062937737
    },
    {
      "epoch": 116,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06924420595169067,
        "mean_feedback_forward_cosine": 0.9994319384748286,
        "mean_feedback_forward_relative_error": 0.03159833199937235,
        "min_feedback_forward_cosine": 0.9976159334182739
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45356,
      "train_examples": 50000,
      "train_loss": 0.01605200676083565
    },
    {
      "epoch": 117,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06923021376132965,
        "mean_feedback_forward_cosine": 0.9994321400469,
        "mean_feedback_forward_relative_error": 0.03159324789250439,
        "min_feedback_forward_cosine": 0.9976168870925903
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 45747,
      "train_examples": 50000,
      "train_loss": 0.013920427777171136
    },
    {
      "epoch": 118,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06921586394309998,
        "mean_feedback_forward_cosine": 0.9994323199445551,
        "mean_feedback_forward_relative_error": 0.0315880704247816,
        "min_feedback_forward_cosine": 0.9976179599761963
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46138,
      "train_examples": 50000,
      "train_loss": 0.013951102841943503
    },
    {
      "epoch": 119,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06920035928487778,
        "mean_feedback_forward_cosine": 0.9994325269352305,
        "mean_feedback_forward_relative_error": 0.031582749711180275,
        "min_feedback_forward_cosine": 0.997619092464447
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46529,
      "train_examples": 50000,
      "train_loss": 0.013470773504525423
    },
    {
      "epoch": 120,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06918603926897049,
        "mean_feedback_forward_cosine": 0.999432711167769,
        "mean_feedback_forward_relative_error": 0.03157752440734343,
        "min_feedback_forward_cosine": 0.9976199865341187
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 46920,
      "train_examples": 50000,
      "train_loss": 0.013048178313672543
    },
    {
      "epoch": 121,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06917116791009903,
        "mean_feedback_forward_cosine": 0.999432897567749,
        "mean_feedback_forward_relative_error": 0.031572245416993445,
        "min_feedback_forward_cosine": 0.9976210594177246
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47311,
      "train_examples": 50000,
      "train_loss": 0.011964423420876264
    },
    {
      "epoch": 122,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06915569305419922,
        "mean_feedback_forward_cosine": 0.9994331078095869,
        "mean_feedback_forward_relative_error": 0.03156682039526376,
        "min_feedback_forward_cosine": 0.9976219534873962
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 47702,
      "train_examples": 50000,
      "train_loss": 0.012266639451049268
    },
    {
      "epoch": 123,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06913956254720688,
        "mean_feedback_forward_cosine": 0.9994333267211915,
        "mean_feedback_forward_relative_error": 0.03156118229539557,
        "min_feedback_forward_cosine": 0.9976232647895813
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48093,
      "train_examples": 50000,
      "train_loss": 0.012148132819086313
    },
    {
      "epoch": 124,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0691244974732399,
        "mean_feedback_forward_cosine": 0.9994335142048922,
        "mean_feedback_forward_relative_error": 0.03155574777417562,
        "min_feedback_forward_cosine": 0.997624397277832
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48484,
      "train_examples": 50000,
      "train_loss": 0.011397822160273791
    },
    {
      "epoch": 125,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06910843402147293,
        "mean_feedback_forward_cosine": 0.999433725530451,
        "mean_feedback_forward_relative_error": 0.03155008294534954,
        "min_feedback_forward_cosine": 0.9976255297660828
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 48875,
      "train_examples": 50000,
      "train_loss": 0.011458082374185323
    },
    {
      "epoch": 126,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06909443438053131,
        "mean_feedback_forward_cosine": 0.9994339130141519,
        "mean_feedback_forward_relative_error": 0.0315452404066243,
        "min_feedback_forward_cosine": 0.9976264834403992
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49266,
      "train_examples": 50000,
      "train_loss": 0.009869020701497793
    },
    {
      "epoch": 127,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06907776743173599,
        "mean_feedback_forward_cosine": 0.9994341232559898,
        "mean_feedback_forward_relative_error": 0.03153970285572789,
        "min_feedback_forward_cosine": 0.9976276159286499
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 49657,
      "train_examples": 50000,
      "train_loss": 0.010547742835283279
    },
    {
      "epoch": 128,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06906381249427795,
        "mean_feedback_forward_cosine": 0.999434313990853,
        "mean_feedback_forward_relative_error": 0.03153472960164601,
        "min_feedback_forward_cosine": 0.9976285099983215
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50048,
      "train_examples": 50000,
      "train_loss": 0.009300732849575579
    },
    {
      "epoch": 129,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06904778629541397,
        "mean_feedback_forward_cosine": 0.9994345090605996,
        "mean_feedback_forward_relative_error": 0.031529135595668445,
        "min_feedback_forward_cosine": 0.9976296424865723
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50439,
      "train_examples": 50000,
      "train_loss": 0.009818964008167386
    },
    {
      "epoch": 130,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06902909278869629,
        "mean_feedback_forward_cosine": 0.999434745311737,
        "mean_feedback_forward_relative_error": 0.03152294444258918,
        "min_feedback_forward_cosine": 0.9976310133934021
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 50830,
      "train_examples": 50000,
      "train_loss": 0.010077409034781158
    },
    {
      "epoch": 131,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06901176273822784,
        "mean_feedback_forward_cosine": 0.9994349924000827,
        "mean_feedback_forward_relative_error": 0.031517018733376805,
        "min_feedback_forward_cosine": 0.9976321458816528
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51221,
      "train_examples": 50000,
      "train_loss": 0.009160347877591849
    },
    {
      "epoch": 132,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06899315118789673,
        "mean_feedback_forward_cosine": 0.9994351983070373,
        "mean_feedback_forward_relative_error": 0.031511178113181486,
        "min_feedback_forward_cosine": 0.9976334571838379
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 51612,
      "train_examples": 50000,
      "train_loss": 0.008961532037444413
    },
    {
      "epoch": 133,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06897760182619095,
        "mean_feedback_forward_cosine": 0.9994353966279463,
        "mean_feedback_forward_relative_error": 0.03150578120892698,
        "min_feedback_forward_cosine": 0.9976345300674438
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52003,
      "train_examples": 50000,
      "train_loss": 0.008542742440328002
    },
    {
      "epoch": 134,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06896280497312546,
        "mean_feedback_forward_cosine": 0.9994355873628096,
        "mean_feedback_forward_relative_error": 0.031500711986287074,
        "min_feedback_forward_cosine": 0.9976356029510498
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52394,
      "train_examples": 50000,
      "train_loss": 0.008353168544620275
    },
    {
      "epoch": 135,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06894773989915848,
        "mean_feedback_forward_cosine": 0.9994357878511603,
        "mean_feedback_forward_relative_error": 0.031495631257580085,
        "min_feedback_forward_cosine": 0.9976365566253662
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 52785,
      "train_examples": 50000,
      "train_loss": 0.008146675640046596
    },
    {
      "epoch": 136,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06893233209848404,
        "mean_feedback_forward_cosine": 0.9994359666650946,
        "mean_feedback_forward_relative_error": 0.03149048773233186,
        "min_feedback_forward_cosine": 0.9976376295089722
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 53176,
      "train_examples": 50000,
      "train_loss": 0.007538513074964285
    },
    {
      "epoch": 137,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06891530007123947,
        "mean_feedback_forward_cosine": 0.9994361877441407,
        "mean_feedback_forward_relative_error": 0.03148495936766267,
        "min_feedback_forward_cosine": 0.9976388216018677
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 53567,
      "train_examples": 50000,
      "train_loss": 0.007756935535706579
    },
    {
      "epoch": 138,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06889911741018295,
        "mean_feedback_forward_cosine": 0.999436401237141,
        "mean_feedback_forward_relative_error": 0.031479515169154516,
        "min_feedback_forward_cosine": 0.9976398944854736
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 53958,
      "train_examples": 50000,
      "train_loss": 0.007756191560998559
    },
    {
      "epoch": 139,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06888337433338165,
        "mean_feedback_forward_cosine": 0.9994366049766541,
        "mean_feedback_forward_relative_error": 0.031474347962913186,
        "min_feedback_forward_cosine": 0.9976409077644348
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 54349,
      "train_examples": 50000,
      "train_loss": 0.0069661939964443444
    },
    {
      "epoch": 140,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06886947154998779,
        "mean_feedback_forward_cosine": 0.9994367707859386,
        "mean_feedback_forward_relative_error": 0.03146959664300084,
        "min_feedback_forward_cosine": 0.9976418614387512
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 54740,
      "train_examples": 50000,
      "train_loss": 0.006709710410907865
    },
    {
      "epoch": 141,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0688507929444313,
        "mean_feedback_forward_cosine": 0.9994369961998679,
        "mean_feedback_forward_relative_error": 0.0314637197858908,
        "min_feedback_forward_cosine": 0.9976431131362915
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 55131,
      "train_examples": 50000,
      "train_loss": 0.007294634993895889
    },
    {
      "epoch": 142,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06883271783590317,
        "mean_feedback_forward_cosine": 0.9994372302835638,
        "mean_feedback_forward_relative_error": 0.03145776710219004,
        "min_feedback_forward_cosine": 0.9976444244384766
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 55522,
      "train_examples": 50000,
      "train_loss": 0.007054433014690876
    },
    {
      "epoch": 143,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06881632655858994,
        "mean_feedback_forward_cosine": 0.9994374318556352,
        "mean_feedback_forward_relative_error": 0.03145231891592795,
        "min_feedback_forward_cosine": 0.9976456165313721
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 55913,
      "train_examples": 50000,
      "train_loss": 0.006983992516249418
    },
    {
      "epoch": 144,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06879774481058121,
        "mean_feedback_forward_cosine": 0.9994376886974682,
        "mean_feedback_forward_relative_error": 0.03144586983729492,
        "min_feedback_forward_cosine": 0.9976467490196228
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 56304,
      "train_examples": 50000,
      "train_loss": 0.007579090554155409
    },
    {
      "epoch": 145,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06878021359443665,
        "mean_feedback_forward_cosine": 0.9994379151951183,
        "mean_feedback_forward_relative_error": 0.0314398794604296,
        "min_feedback_forward_cosine": 0.9976479411125183
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 56695,
      "train_examples": 50000,
      "train_loss": 0.007095893447734415
    },
    {
      "epoch": 146,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06876447051763535,
        "mean_feedback_forward_cosine": 0.9994381048462607,
        "mean_feedback_forward_relative_error": 0.03143445560708642,
        "min_feedback_forward_cosine": 0.9976490139961243
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 57086,
      "train_examples": 50000,
      "train_loss": 0.006462747115045786
    },
    {
      "epoch": 147,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06874611973762512,
        "mean_feedback_forward_cosine": 0.999438354102048,
        "mean_feedback_forward_relative_error": 0.03142826100811362,
        "min_feedback_forward_cosine": 0.9976503252983093
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 57477,
      "train_examples": 50000,
      "train_loss": 0.006878289589509368
    },
    {
      "epoch": 148,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06872718781232834,
        "mean_feedback_forward_cosine": 0.9994386044415561,
        "mean_feedback_forward_relative_error": 0.0314219257963652,
        "min_feedback_forward_cosine": 0.9976516366004944
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 57868,
      "train_examples": 50000,
      "train_loss": 0.0068016401921957735
    },
    {
      "epoch": 149,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06870897114276886,
        "mean_feedback_forward_cosine": 0.9994388276880438,
        "mean_feedback_forward_relative_error": 0.031415897768668155,
        "min_feedback_forward_cosine": 0.9976528882980347
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 58259,
      "train_examples": 50000,
      "train_loss": 0.006183813361376524
    },
    {
      "epoch": 150,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06869076192378998,
        "mean_feedback_forward_cosine": 0.9994390563531356,
        "mean_feedback_forward_relative_error": 0.031410239746963435,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.010000000000000002,
      "output_lr": 0.010000000000000002,
      "step": 58650,
      "train_examples": 50000,
      "train_loss": 0.005665142865814269
    },
    {
      "epoch": 151,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06869052350521088,
        "mean_feedback_forward_cosine": 0.999439059604298,
        "mean_feedback_forward_relative_error": 0.0314101467163048,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59041,
      "train_examples": 50000,
      "train_loss": 0.006256562349833548
    },
    {
      "epoch": 152,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06869038194417953,
        "mean_feedback_forward_cosine": 0.9994390639391812,
        "mean_feedback_forward_relative_error": 0.03141007806936448,
        "min_feedback_forward_cosine": 0.997654139995575
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59432,
      "train_examples": 50000,
      "train_loss": 0.00507289287718013
    },
    {
      "epoch": 153,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06869025528430939,
        "mean_feedback_forward_cosine": 0.9994390574368563,
        "mean_feedback_forward_relative_error": 0.03141001376577399,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 59823,
      "train_examples": 50000,
      "train_loss": 0.0047390447936952115
    },
    {
      "epoch": 154,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06869012117385864,
        "mean_feedback_forward_cosine": 0.999439065022902,
        "mean_feedback_forward_relative_error": 0.031409954017197546,
        "min_feedback_forward_cosine": 0.9976540803909302
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60214,
      "train_examples": 50000,
      "train_loss": 0.004380627877153456
    },
    {
      "epoch": 155,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868996471166611,
        "mean_feedback_forward_cosine": 0.9994390661066229,
        "mean_feedback_forward_relative_error": 0.031409888883883304,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60605,
      "train_examples": 50000,
      "train_loss": 0.004598957079797983
    },
    {
      "epoch": 156,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868986040353775,
        "mean_feedback_forward_cosine": 0.9994390498508107,
        "mean_feedback_forward_relative_error": 0.031409834867173976,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 60996,
      "train_examples": 50000,
      "train_loss": 0.003838625859953463
    },
    {
      "epoch": 157,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868971884250641,
        "mean_feedback_forward_cosine": 0.9994390465996482,
        "mean_feedback_forward_relative_error": 0.03140976984392513,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61387,
      "train_examples": 50000,
      "train_loss": 0.004432131025716662
    },
    {
      "epoch": 158,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868959218263626,
        "mean_feedback_forward_cosine": 0.9994390661066229,
        "mean_feedback_forward_relative_error": 0.031409713769839566,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 61778,
      "train_examples": 50000,
      "train_loss": 0.0044609378554672004
    },
    {
      "epoch": 159,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868948042392731,
        "mean_feedback_forward_cosine": 0.9994390715252269,
        "mean_feedback_forward_relative_error": 0.03140965415672822,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62169,
      "train_examples": 50000,
      "train_loss": 0.004222902919054031
    },
    {
      "epoch": 160,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868936866521835,
        "mean_feedback_forward_cosine": 0.9994390617717396,
        "mean_feedback_forward_relative_error": 0.031409599987620655,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62560,
      "train_examples": 50000,
      "train_loss": 0.003845143410935998
    },
    {
      "epoch": 161,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868921965360641,
        "mean_feedback_forward_cosine": 0.9994390563531356,
        "mean_feedback_forward_relative_error": 0.031409540425308725,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 62951,
      "train_examples": 50000,
      "train_loss": 0.004378214224409312
    },
    {
      "epoch": 162,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868909299373627,
        "mean_feedback_forward_cosine": 0.9994390617717396,
        "mean_feedback_forward_relative_error": 0.031409482336179774,
        "min_feedback_forward_cosine": 0.9976541996002197
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63342,
      "train_examples": 50000,
      "train_loss": 0.004232344762831926
    },
    {
      "epoch": 163,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868896633386612,
        "mean_feedback_forward_cosine": 0.9994390661066229,
        "mean_feedback_forward_relative_error": 0.03140942423858426,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 63733,
      "train_examples": 50000,
      "train_loss": 0.004272423257436603
    },
    {
      "epoch": 164,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868885457515717,
        "mean_feedback_forward_cosine": 0.9994390736926686,
        "mean_feedback_forward_relative_error": 0.03140936523506587,
        "min_feedback_forward_cosine": 0.9976542592048645
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64124,
      "train_examples": 50000,
      "train_loss": 0.004210389293860644
    },
    {
      "epoch": 165,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868872046470642,
        "mean_feedback_forward_cosine": 0.999439076943831,
        "mean_feedback_forward_relative_error": 0.03140930999070406,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64515,
      "train_examples": 50000,
      "train_loss": 0.003977990410029888
    },
    {
      "epoch": 166,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868861615657806,
        "mean_feedback_forward_cosine": 0.9994390693577853,
        "mean_feedback_forward_relative_error": 0.031409247414293615,
        "min_feedback_forward_cosine": 0.9976542592048645
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 64906,
      "train_examples": 50000,
      "train_loss": 0.004078018093742431
    },
    {
      "epoch": 167,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868851184844971,
        "mean_feedback_forward_cosine": 0.9994390921159224,
        "mean_feedback_forward_relative_error": 0.031409196327017115,
        "min_feedback_forward_cosine": 0.9976542592048645
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65297,
      "train_examples": 50000,
      "train_loss": 0.003648907724171877
    },
    {
      "epoch": 168,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868840754032135,
        "mean_feedback_forward_cosine": 0.9994390834461558,
        "mean_feedback_forward_relative_error": 0.03140914492647756,
        "min_feedback_forward_cosine": 0.997654378414154
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 65688,
      "train_examples": 50000,
      "train_loss": 0.0033877091257832944
    },
    {
      "epoch": 169,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0686882957816124,
        "mean_feedback_forward_cosine": 0.9994390823624351,
        "mean_feedback_forward_relative_error": 0.03140909403393215,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66079,
      "train_examples": 50000,
      "train_loss": 0.0035693821069598198
    },
    {
      "epoch": 170,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868819147348404,
        "mean_feedback_forward_cosine": 0.9994390866973183,
        "mean_feedback_forward_relative_error": 0.03140904228626327,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66470,
      "train_examples": 50000,
      "train_loss": 0.0036892011823877694
    },
    {
      "epoch": 171,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868808716535568,
        "mean_feedback_forward_cosine": 0.9994390791112726,
        "mean_feedback_forward_relative_error": 0.03140899477187883,
        "min_feedback_forward_cosine": 0.997654378414154
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 66861,
      "train_examples": 50000,
      "train_loss": 0.003492035176092759
    },
    {
      "epoch": 172,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868799030780792,
        "mean_feedback_forward_cosine": 0.9994390877810392,
        "mean_feedback_forward_relative_error": 0.03140894459052519,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67252,
      "train_examples": 50000,
      "train_loss": 0.003594661014340818
    },
    {
      "epoch": 173,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868788599967957,
        "mean_feedback_forward_cosine": 0.9994390856135975,
        "mean_feedback_forward_relative_error": 0.031408889811824665,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 67643,
      "train_examples": 50000,
      "train_loss": 0.003707864111084491
    },
    {
      "epoch": 174,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868778169155121,
        "mean_feedback_forward_cosine": 0.99943908886476,
        "mean_feedback_forward_relative_error": 0.03140884155238217,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68034,
      "train_examples": 50000,
      "train_loss": 0.003334373738132417
    },
    {
      "epoch": 175,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868770718574524,
        "mean_feedback_forward_cosine": 0.9994390986182473,
        "mean_feedback_forward_relative_error": 0.031408793199807404,
        "min_feedback_forward_cosine": 0.997654378414154
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68425,
      "train_examples": 50000,
      "train_loss": 0.003135205378085375
    },
    {
      "epoch": 176,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868759542703629,
        "mean_feedback_forward_cosine": 0.9994390921159224,
        "mean_feedback_forward_relative_error": 0.031408739513294266,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 68816,
      "train_examples": 50000,
      "train_loss": 0.00379005764760077
    },
    {
      "epoch": 177,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868747621774673,
        "mean_feedback_forward_cosine": 0.9994390856135975,
        "mean_feedback_forward_relative_error": 0.03140868591991338,
        "min_feedback_forward_cosine": 0.997654378414154
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69207,
      "train_examples": 50000,
      "train_loss": 0.0037669762847572564
    },
    {
      "epoch": 178,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868737190961838,
        "mean_feedback_forward_cosine": 0.9994390921159224,
        "mean_feedback_forward_relative_error": 0.03140863081948324,
        "min_feedback_forward_cosine": 0.9976543188095093
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69598,
      "train_examples": 50000,
      "train_loss": 0.0033926852117478847
    },
    {
      "epoch": 179,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868726015090942,
        "mean_feedback_forward_cosine": 0.9994390975345265,
        "mean_feedback_forward_relative_error": 0.03140857703137127,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 69989,
      "train_examples": 50000,
      "train_loss": 0.003614906167127192
    },
    {
      "epoch": 180,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868716329336166,
        "mean_feedback_forward_cosine": 0.9994390931996432,
        "mean_feedback_forward_relative_error": 0.03140852423384786,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70380,
      "train_examples": 50000,
      "train_loss": 0.0034720032831281424
    },
    {
      "epoch": 181,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868704408407211,
        "mean_feedback_forward_cosine": 0.9994391029531305,
        "mean_feedback_forward_relative_error": 0.031408465721390465,
        "min_feedback_forward_cosine": 0.997654378414154
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 70771,
      "train_examples": 50000,
      "train_loss": 0.0035889510291069746
    },
    {
      "epoch": 182,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868693232536316,
        "mean_feedback_forward_cosine": 0.9994390997019681,
        "mean_feedback_forward_relative_error": 0.03140841539610516,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 71162,
      "train_examples": 50000,
      "train_loss": 0.003305433737561107
    },
    {
      "epoch": 183,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0686868354678154,
        "mean_feedback_forward_cosine": 0.9994391202926636,
        "mean_feedback_forward_relative_error": 0.03140837007456205,
        "min_feedback_forward_cosine": 0.997654378414154
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 71553,
      "train_examples": 50000,
      "train_loss": 0.003175204580221325
    },
    {
      "epoch": 184,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868674606084824,
        "mean_feedback_forward_cosine": 0.9994391170415011,
        "mean_feedback_forward_relative_error": 0.03140832256017761,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 71944,
      "train_examples": 50000,
      "train_loss": 0.0031294494361244142
    },
    {
      "epoch": 185,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868667155504227,
        "mean_feedback_forward_cosine": 0.9994391116228971,
        "mean_feedback_forward_relative_error": 0.03140827675604008,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 72335,
      "train_examples": 50000,
      "train_loss": 0.003196823445223272
    },
    {
      "epoch": 186,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0686865746974945,
        "mean_feedback_forward_cosine": 0.9994391029531305,
        "mean_feedback_forward_relative_error": 0.03140823346647349,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 72726,
      "train_examples": 50000,
      "train_loss": 0.0030139125097729267
    },
    {
      "epoch": 187,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868649274110794,
        "mean_feedback_forward_cosine": 0.9994390910322016,
        "mean_feedback_forward_relative_error": 0.03140818929638375,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 73117,
      "train_examples": 50000,
      "train_loss": 0.003099756380841136
    },
    {
      "epoch": 188,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868639588356018,
        "mean_feedback_forward_cosine": 0.9994391040368513,
        "mean_feedback_forward_relative_error": 0.031408142391592266,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 73508,
      "train_examples": 50000,
      "train_loss": 0.0031279313344694673
    },
    {
      "epoch": 189,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868629902601242,
        "mean_feedback_forward_cosine": 0.9994391029531305,
        "mean_feedback_forward_relative_error": 0.03140809452161193,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 73899,
      "train_examples": 50000,
      "train_loss": 0.0032867785268090666
    },
    {
      "epoch": 190,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868621706962585,
        "mean_feedback_forward_cosine": 0.9994391105391762,
        "mean_feedback_forward_relative_error": 0.03140805058858611,
        "min_feedback_forward_cosine": 0.9976544976234436
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 74290,
      "train_examples": 50000,
      "train_loss": 0.0030857411340996623
    },
    {
      "epoch": 191,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0686860904097557,
        "mean_feedback_forward_cosine": 0.9994391148740595,
        "mean_feedback_forward_relative_error": 0.03140799194066362,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 74681,
      "train_examples": 50000,
      "train_loss": 0.004001324234195054
    },
    {
      "epoch": 192,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868598610162735,
        "mean_feedback_forward_cosine": 0.9994391094554554,
        "mean_feedback_forward_relative_error": 0.031407940692522306,
        "min_feedback_forward_cosine": 0.9976544380187988
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 75072,
      "train_examples": 50000,
      "train_loss": 0.003275472015310079
    },
    {
      "epoch": 193,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868590414524078,
        "mean_feedback_forward_cosine": 0.9994391170415011,
        "mean_feedback_forward_relative_error": 0.03140789498151703,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 75463,
      "train_examples": 50000,
      "train_loss": 0.002768472644649446
    },
    {
      "epoch": 194,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868579983711243,
        "mean_feedback_forward_cosine": 0.9994391202926636,
        "mean_feedback_forward_relative_error": 0.03140784820372408,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 75854,
      "train_examples": 50000,
      "train_loss": 0.0030618526208773256
    },
    {
      "epoch": 195,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868570297956467,
        "mean_feedback_forward_cosine": 0.9994391040368513,
        "mean_feedback_forward_relative_error": 0.03140779734504494,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 76245,
      "train_examples": 50000,
      "train_loss": 0.00339243379432708
    },
    {
      "epoch": 196,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.0686856135725975,
        "mean_feedback_forward_cosine": 0.999439123543826,
        "mean_feedback_forward_relative_error": 0.031407752802426164,
        "min_feedback_forward_cosine": 0.9976546168327332
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 76636,
      "train_examples": 50000,
      "train_loss": 0.0028797079719975593
    },
    {
      "epoch": 197,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868550926446915,
        "mean_feedback_forward_cosine": 0.9994391224601052,
        "mean_feedback_forward_relative_error": 0.03140770508484407,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 77027,
      "train_examples": 50000,
      "train_loss": 0.0033768319156579673
    },
    {
      "epoch": 198,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868541985750198,
        "mean_feedback_forward_cosine": 0.9994391332973134,
        "mean_feedback_forward_relative_error": 0.03140765925530683,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 77418,
      "train_examples": 50000,
      "train_loss": 0.0029052105254959316
    },
    {
      "epoch": 199,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868532299995422,
        "mean_feedback_forward_cosine": 0.9994391322135925,
        "mean_feedback_forward_relative_error": 0.03140761311250654,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 77809,
      "train_examples": 50000,
      "train_loss": 0.003043961836770177
    },
    {
      "epoch": 200,
      "feedback_tracking": {
        "max_feedback_forward_relative_error": 0.06868523359298706,
        "mean_feedback_forward_cosine": 0.9994391322135925,
        "mean_feedback_forward_relative_error": 0.031407567316835576,
        "min_feedback_forward_cosine": 0.9976545572280884
      },
      "lr": 0.0010000000000000002,
      "output_lr": 0.0010000000000000002,
      "step": 78200,
      "train_examples": 50000,
      "train_loss": 0.0028966831861622633
    }
  ],
  "evaluation_protocol": {
    "test_evaluations": 1,
    "test_used_for_selection": false,
    "validation_evaluations": 0
  },
  "final": {
    "accuracy": 0.9265,
    "epoch": 200,
    "evaluation_split": "test",
    "finite": true,
    "loss": 0.3951418914794922,
    "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": 12,
    "normalization_mean": [
      0.49140000343322754,
      0.4821999967098236,
      0.4465000033378601
    ],
    "normalization_std": [
      0.24699999392032623,
      0.2434999942779541,
      0.26159998774528503
    ],
    "split_from_training_only": true,
    "split_seed": null,
    "test_examples": 10000,
    "train_examples": 50000,
    "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
    "validation_class_counts": {},
    "validation_examples": 0,
    "validation_index_sha256": null
  },
  "timing": {
    "apical_warmup_wall_s": 0.0,
    "evaluation_wall_s": 2.0679945945739746,
    "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": 18253.185290575027,
    "train_wall_s": 18247.471817970276
  },
  "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
  }
}