{
  "config": {
    "label": "LayerNorm + silu + plateau LR",
    "batch": 64,
    "lr": 0.003,
    "schedule": "plateau",
    "normalization": "layer",
    "residual": false,
    "concatenate": false,
    "dropout": 0.0,
    "activation": "silu",
    "loss": "mse",
    "outliers": false,
    "max_updates": null
  },
  "seeds": [
    11,
    22,
    33
  ],
  "epochs": 50000,
  "patience": 20,
  "subsets": [
    [
      0,
      11
    ],
    [
      0,
      4,
      7,
      11
    ],
    [
      0,
      1,
      4,
      6,
      7,
      11
    ],
    [
      0,
      1,
      3,
      4,
      6,
      7,
      10,
      11
    ]
  ],
  "validation_profiles": [
    2,
    8
  ],
  "test_profiles": [
    5,
    9
  ],
  "workers": 12,
  "threads": 1,
  "runs": [
    {
      "history": [
        {
          "epoch": 1,
          "updates": 4,
          "samples_seen": 256,
          "online_loss": 0.26354655623435974,
          "train_loss": 0.11158597469329834,
          "validation_loss": 0.20771658420562744,
          "train_metric": 0.02118718810379505,
          "validation_metric": 0.039439812302589417,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.006345158999465639
        },
        {
          "epoch": 2,
          "updates": 8,
          "samples_seen": 512,
          "online_loss": 0.07558783888816833,
          "train_loss": 0.06384824216365814,
          "validation_loss": 0.2421005815267563,
          "train_metric": 0.01212307158857584,
          "validation_metric": 0.04596841707825661,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.010617313999318867
        },
        {
          "epoch": 3,
          "updates": 12,
          "samples_seen": 768,
          "online_loss": 0.06913704890757799,
          "train_loss": 0.05132748559117317,
          "validation_loss": 0.13149908185005188,
          "train_metric": 0.00974571518599987,
          "validation_metric": 0.024968154728412628,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.015013031999842497
        },
        {
          "epoch": 4,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.05067810509353876,
          "train_loss": 0.04211435839533806,
          "validation_loss": 0.12362726032733917,
          "train_metric": 0.007996387779712677,
          "validation_metric": 0.023473504930734634,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.01931339700058743
        },
        {
          "epoch": 5,
          "updates": 20,
          "samples_seen": 1280,
          "online_loss": 0.035924747586250305,
          "train_loss": 0.03954773768782616,
          "validation_loss": 0.22560101747512817,
          "train_metric": 0.007509056478738785,
          "validation_metric": 0.04283558949828148,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02304335699955118
        },
        {
          "epoch": 6,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.03350058011710644,
          "train_loss": 0.024498717859387398,
          "validation_loss": 0.21155890822410583,
          "train_metric": 0.004651650320738554,
          "validation_metric": 0.0401693657040596,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.026668726999560022
        },
        {
          "epoch": 7,
          "updates": 28,
          "samples_seen": 1792,
          "online_loss": 0.02388045541010797,
          "train_loss": 0.023508066311478615,
          "validation_loss": 0.18819287419319153,
          "train_metric": 0.004463551566004753,
          "validation_metric": 0.03573278710246086,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.030284817999927327
        },
        {
          "epoch": 8,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.022532728500664234,
          "train_loss": 0.018397139385342598,
          "validation_loss": 0.1849251538515091,
          "train_metric": 0.0034931241534650326,
          "validation_metric": 0.035112328827381134,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.033996321999438806
        },
        {
          "epoch": 9,
          "updates": 36,
          "samples_seen": 2304,
          "online_loss": 0.0177878700196743,
          "train_loss": 0.015261538326740265,
          "validation_loss": 0.23226062953472137,
          "train_metric": 0.002897757338359952,
          "validation_metric": 0.044100068509578705,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0376183949992992
        },
        {
          "epoch": 10,
          "updates": 40,
          "samples_seen": 2560,
          "online_loss": 0.015301399631425738,
          "train_loss": 0.013095943257212639,
          "validation_loss": 0.2587611973285675,
          "train_metric": 0.0024865688756108284,
          "validation_metric": 0.049131814390420914,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.041237273000660934
        },
        {
          "epoch": 11,
          "updates": 44,
          "samples_seen": 2816,
          "online_loss": 0.011688546044752002,
          "train_loss": 0.0102589325979352,
          "validation_loss": 0.22660449147224426,
          "train_metric": 0.0019478964386507869,
          "validation_metric": 0.043026119470596313,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.044767205999960424
        },
        {
          "epoch": 12,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.01065474800998345,
          "train_loss": 0.011588916182518005,
          "validation_loss": 0.23540210723876953,
          "train_metric": 0.0022004246711730957,
          "validation_metric": 0.04469655454158783,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.048445506999996724
        },
        {
          "epoch": 13,
          "updates": 52,
          "samples_seen": 3328,
          "online_loss": 0.009741075104102492,
          "train_loss": 0.007796064950525761,
          "validation_loss": 0.2632593512535095,
          "train_metric": 0.001480264007113874,
          "validation_metric": 0.04998590424656868,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05208403799952066
        },
        {
          "epoch": 14,
          "updates": 56,
          "samples_seen": 3584,
          "online_loss": 0.00970686855725944,
          "train_loss": 0.008643693290650845,
          "validation_loss": 0.264565646648407,
          "train_metric": 0.0016412058612331748,
          "validation_metric": 0.05023392662405968,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.055696520999845234
        },
        {
          "epoch": 15,
          "updates": 60,
          "samples_seen": 3840,
          "online_loss": 0.007812251336872578,
          "train_loss": 0.0059364126063883305,
          "validation_loss": 0.25229817628860474,
          "train_metric": 0.0011271657422184944,
          "validation_metric": 0.047904662787914276,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.059176504000788555
        },
        {
          "epoch": 16,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.005885066871996969,
          "train_loss": 0.006794051732867956,
          "validation_loss": 0.27221158146858215,
          "train_metric": 0.0012900084257125854,
          "validation_metric": 0.05168568715453148,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.06274970699996629
        },
        {
          "epoch": 17,
          "updates": 68,
          "samples_seen": 4352,
          "online_loss": 0.006255954620428383,
          "train_loss": 0.0076976646669209,
          "validation_loss": 0.26513153314590454,
          "train_metric": 0.0014615803956985474,
          "validation_metric": 0.050341375172138214,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.066290910999669
        },
        {
          "epoch": 18,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.007168185256887227,
          "train_loss": 0.005037018563598394,
          "validation_loss": 0.2705925703048706,
          "train_metric": 0.0009563948842696846,
          "validation_metric": 0.0513782761991024,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.06985851699937484
        },
        {
          "epoch": 19,
          "updates": 76,
          "samples_seen": 4864,
          "online_loss": 0.008660748484544456,
          "train_loss": 0.009393628686666489,
          "validation_loss": 0.2845909297466278,
          "train_metric": 0.0017835983308032155,
          "validation_metric": 0.05403619632124901,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.07342422399960924
        },
        {
          "epoch": 20,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.006885100854560733,
          "train_loss": 0.00598621740937233,
          "validation_loss": 0.266573429107666,
          "train_metric": 0.0011366222752258182,
          "validation_metric": 0.05061515420675278,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.07712231400000746
        },
        {
          "epoch": 21,
          "updates": 84,
          "samples_seen": 5376,
          "online_loss": 0.006367947440594435,
          "train_loss": 0.005663607735186815,
          "validation_loss": 0.28769171237945557,
          "train_metric": 0.0010753674432635307,
          "validation_metric": 0.054624952375888824,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.08073029300066992
        },
        {
          "epoch": 22,
          "updates": 88,
          "samples_seen": 5632,
          "online_loss": 0.007753567537292838,
          "train_loss": 0.003788123605772853,
          "validation_loss": 0.2762261927127838,
          "train_metric": 0.0007192632183432579,
          "validation_metric": 0.052447959780693054,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.08440332099962689
        },
        {
          "epoch": 23,
          "updates": 92,
          "samples_seen": 5888,
          "online_loss": 0.005607030703686178,
          "train_loss": 0.005773313343524933,
          "validation_loss": 0.25922077894210815,
          "train_metric": 0.001096197753213346,
          "validation_metric": 0.049219079315662384,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.0879320549993281
        },
        {
          "epoch": 24,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.006908231240231544,
          "train_loss": 0.006482541561126709,
          "validation_loss": 0.277430921792984,
          "train_metric": 0.0012308609439060092,
          "validation_metric": 0.052676692605018616,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.09154165899963118
        }
      ],
      "best_validation_metric": 0.023473504930734634,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 22,
      "parameters": 2369,
      "train_samples": 256,
      "validation_rmse_m_s": 0.15321065540860607,
      "training_profiles": [
        0,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 4,
          "samples_seen": 256,
          "online_loss": 0.34945591166615486,
          "train_loss": 0.14827919006347656,
          "validation_loss": 0.18102659285068512,
          "train_metric": 0.028154246509075165,
          "validation_metric": 0.034372102469205856,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.006246491999263526
        },
        {
          "epoch": 2,
          "updates": 8,
          "samples_seen": 512,
          "online_loss": 0.125831987708807,
          "train_loss": 0.0816517025232315,
          "validation_loss": 0.10363718122243881,
          "train_metric": 0.015503470785915852,
          "validation_metric": 0.01967792585492134,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.011288498999419971
        },
        {
          "epoch": 3,
          "updates": 12,
          "samples_seen": 768,
          "online_loss": 0.0800209492444992,
          "train_loss": 0.06523114442825317,
          "validation_loss": 0.08534522354602814,
          "train_metric": 0.012385647743940353,
          "validation_metric": 0.016204774379730225,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.015782152999236132
        },
        {
          "epoch": 4,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.05782860051840544,
          "train_loss": 0.05137048661708832,
          "validation_loss": 0.15350016951560974,
          "train_metric": 0.009753878228366375,
          "validation_metric": 0.02914556860923767,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.019561068000257364
        },
        {
          "epoch": 5,
          "updates": 20,
          "samples_seen": 1280,
          "online_loss": 0.04507517535239458,
          "train_loss": 0.03705363720655441,
          "validation_loss": 0.12380605190992355,
          "train_metric": 0.007035492919385433,
          "validation_metric": 0.02350745163857937,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.023409523999362136
        },
        {
          "epoch": 6,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.04752524849027395,
          "train_loss": 0.034233346581459045,
          "validation_loss": 0.09908599406480789,
          "train_metric": 0.006499995943158865,
          "validation_metric": 0.01881377585232258,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.027206974000364426
        },
        {
          "epoch": 7,
          "updates": 28,
          "samples_seen": 1792,
          "online_loss": 0.033731522504240274,
          "train_loss": 0.03983556479215622,
          "validation_loss": 0.0895775631070137,
          "train_metric": 0.007563706953078508,
          "validation_metric": 0.017008382827043533,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0309853929993551
        },
        {
          "epoch": 8,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.031507430132478476,
          "train_loss": 0.03111933544278145,
          "validation_loss": 0.058544013649225235,
          "train_metric": 0.0059087276458740234,
          "validation_metric": 0.011115938425064087,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.035179440999854705
        },
        {
          "epoch": 9,
          "updates": 36,
          "samples_seen": 2304,
          "online_loss": 0.03332523675635457,
          "train_loss": 0.02264220267534256,
          "validation_loss": 0.07009594142436981,
          "train_metric": 0.004299147054553032,
          "validation_metric": 0.013309339992702007,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03901757300081954
        },
        {
          "epoch": 10,
          "updates": 40,
          "samples_seen": 2560,
          "online_loss": 0.024769559502601624,
          "train_loss": 0.023663539439439774,
          "validation_loss": 0.09466235339641571,
          "train_metric": 0.00449307169765234,
          "validation_metric": 0.017973843961954117,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.042750693999551004
        },
        {
          "epoch": 11,
          "updates": 44,
          "samples_seen": 2816,
          "online_loss": 0.019565835362300277,
          "train_loss": 0.027453485876321793,
          "validation_loss": 0.10709361732006073,
          "train_metric": 0.005212681367993355,
          "validation_metric": 0.020334210246801376,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04651395700057037
        },
        {
          "epoch": 12,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.02343587949872017,
          "train_loss": 0.017563357949256897,
          "validation_loss": 0.07485462725162506,
          "train_metric": 0.0033348111901432276,
          "validation_metric": 0.014212887734174728,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05025952300093195
        },
        {
          "epoch": 13,
          "updates": 52,
          "samples_seen": 3328,
          "online_loss": 0.01647886075079441,
          "train_loss": 0.01522070076316595,
          "validation_loss": 0.07231438159942627,
          "train_metric": 0.002890003379434347,
          "validation_metric": 0.013730565086007118,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05393945499963593
        },
        {
          "epoch": 14,
          "updates": 56,
          "samples_seen": 3584,
          "online_loss": 0.014921918511390686,
          "train_loss": 0.0128402728587389,
          "validation_loss": 0.07743097841739655,
          "train_metric": 0.002438024152070284,
          "validation_metric": 0.014702067710459232,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05777225699966948
        },
        {
          "epoch": 15,
          "updates": 60,
          "samples_seen": 3840,
          "online_loss": 0.014008042169734836,
          "train_loss": 0.011728323996067047,
          "validation_loss": 0.08329734206199646,
          "train_metric": 0.002226894488558173,
          "validation_metric": 0.01581593230366707,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.061520888999439194
        },
        {
          "epoch": 16,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.012540597002953291,
          "train_loss": 0.01054874062538147,
          "validation_loss": 0.08816200494766235,
          "train_metric": 0.0020029230508953333,
          "validation_metric": 0.016739601269364357,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06516934600040258
        },
        {
          "epoch": 17,
          "updates": 68,
          "samples_seen": 4352,
          "online_loss": 0.010051020537503064,
          "train_loss": 0.009182686917483807,
          "validation_loss": 0.07351073622703552,
          "train_metric": 0.0017435464542359114,
          "validation_metric": 0.013957719318568707,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06877114000053552
        },
        {
          "epoch": 18,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.011653052875772119,
          "train_loss": 0.011279248632490635,
          "validation_loss": 0.07188692688941956,
          "train_metric": 0.0021416267845779657,
          "validation_metric": 0.013649402186274529,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07234540200079209
        },
        {
          "epoch": 19,
          "updates": 76,
          "samples_seen": 4864,
          "online_loss": 0.008179530035704374,
          "train_loss": 0.014460936188697815,
          "validation_loss": 0.09922724962234497,
          "train_metric": 0.0027457447722554207,
          "validation_metric": 0.018840596079826355,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07591458500064618
        },
        {
          "epoch": 20,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.011979538190644234,
          "train_loss": 0.015786610543727875,
          "validation_loss": 0.07074040919542313,
          "train_metric": 0.0029974542558193207,
          "validation_metric": 0.013431708328425884,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.07961707300091803
        },
        {
          "epoch": 21,
          "updates": 84,
          "samples_seen": 5376,
          "online_loss": 0.012286808108910918,
          "train_loss": 0.012815357185900211,
          "validation_loss": 0.09662756323814392,
          "train_metric": 0.0024332930333912373,
          "validation_metric": 0.018346985802054405,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.08319318699977885
        },
        {
          "epoch": 22,
          "updates": 88,
          "samples_seen": 5632,
          "online_loss": 0.007371140003670007,
          "train_loss": 0.012352189980447292,
          "validation_loss": 0.07640940696001053,
          "train_metric": 0.002345350105315447,
          "validation_metric": 0.014508099295198917,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.0869915380008024
        },
        {
          "epoch": 23,
          "updates": 92,
          "samples_seen": 5888,
          "online_loss": 0.00878836796618998,
          "train_loss": 0.010488348081707954,
          "validation_loss": 0.10201094299554825,
          "train_metric": 0.00199145614169538,
          "validation_metric": 0.01936914399266243,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.0906596629993146
        },
        {
          "epoch": 24,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.010697930585592985,
          "train_loss": 0.009655779227614403,
          "validation_loss": 0.07697345316410065,
          "train_metric": 0.0018333739135414362,
          "validation_metric": 0.014615197665989399,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.0943846919999487
        },
        {
          "epoch": 25,
          "updates": 100,
          "samples_seen": 6400,
          "online_loss": 0.011633498594164848,
          "train_loss": 0.008023960515856743,
          "validation_loss": 0.09209376573562622,
          "train_metric": 0.0015235350001603365,
          "validation_metric": 0.01748613640666008,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.09806475100049283
        },
        {
          "epoch": 26,
          "updates": 104,
          "samples_seen": 6656,
          "online_loss": 0.02070760406786576,
          "train_loss": 0.008661627769470215,
          "validation_loss": 0.081355020403862,
          "train_metric": 0.0016446111258119345,
          "validation_metric": 0.015447139739990234,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.10179420200074674
        },
        {
          "epoch": 27,
          "updates": 108,
          "samples_seen": 6912,
          "online_loss": 0.02298506151419133,
          "train_loss": 0.013646839186549187,
          "validation_loss": 0.07091948390007019,
          "train_metric": 0.0025911694392561913,
          "validation_metric": 0.013465709052979946,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.10534247900068294
        },
        {
          "epoch": 28,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.01783348130993545,
          "train_loss": 0.025529149919748306,
          "validation_loss": 0.1053474172949791,
          "train_metric": 0.004847301170229912,
          "validation_metric": 0.02000265195965767,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.10906140000042797
        }
      ],
      "best_validation_metric": 0.011115938425064087,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 33,
      "parameters": 2369,
      "train_samples": 256,
      "validation_rmse_m_s": 0.10543215081304226,
      "training_profiles": [
        0,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 4,
          "samples_seen": 256,
          "online_loss": 0.6527862399816513,
          "train_loss": 0.19003473222255707,
          "validation_loss": 0.12282265722751617,
          "train_metric": 0.03608250617980957,
          "validation_metric": 0.02332073263823986,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0058399319987074705
        },
        {
          "epoch": 2,
          "updates": 8,
          "samples_seen": 512,
          "online_loss": 0.122270118445158,
          "train_loss": 0.07979883253574371,
          "validation_loss": 0.14065200090408325,
          "train_metric": 0.01515166088938713,
          "validation_metric": 0.026706047356128693,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.009973287998946034
        },
        {
          "epoch": 3,
          "updates": 12,
          "samples_seen": 768,
          "online_loss": 0.0701860673725605,
          "train_loss": 0.05998780205845833,
          "validation_loss": 0.22623123228549957,
          "train_metric": 0.011390075087547302,
          "validation_metric": 0.04295525327324867,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.013525894999474986
        },
        {
          "epoch": 4,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.05411594454199076,
          "train_loss": 0.05357860401272774,
          "validation_loss": 0.17196278274059296,
          "train_metric": 0.010173140093684196,
          "validation_metric": 0.03265112638473511,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.017041395998603548
        },
        {
          "epoch": 5,
          "updates": 20,
          "samples_seen": 1280,
          "online_loss": 0.04498360585421324,
          "train_loss": 0.03222450613975525,
          "validation_loss": 0.18090739846229553,
          "train_metric": 0.006118570454418659,
          "validation_metric": 0.0343494676053524,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02050747199973557
        },
        {
          "epoch": 6,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.03534803818911314,
          "train_loss": 0.03162136301398277,
          "validation_loss": 0.14750738441944122,
          "train_metric": 0.006004049442708492,
          "validation_metric": 0.02800769917666912,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.024036998000156018
        },
        {
          "epoch": 7,
          "updates": 28,
          "samples_seen": 1792,
          "online_loss": 0.03182708052918315,
          "train_loss": 0.02336534485220909,
          "validation_loss": 0.18291668593883514,
          "train_metric": 0.0044364528730511665,
          "validation_metric": 0.034730978310108185,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.027475896999021643
        },
        {
          "epoch": 8,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.02129579638130963,
          "train_loss": 0.018848640844225883,
          "validation_loss": 0.20334237813949585,
          "train_metric": 0.0035788516979664564,
          "validation_metric": 0.03860926628112793,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03088500000012573
        },
        {
          "epoch": 9,
          "updates": 36,
          "samples_seen": 2304,
          "online_loss": 0.01743072271347046,
          "train_loss": 0.02214619144797325,
          "validation_loss": 0.19185110926628113,
          "train_metric": 0.004204968921840191,
          "validation_metric": 0.03642738610506058,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03435864199855132
        },
        {
          "epoch": 10,
          "updates": 40,
          "samples_seen": 2560,
          "online_loss": 0.018380904104560614,
          "train_loss": 0.017128944396972656,
          "validation_loss": 0.24245275557041168,
          "train_metric": 0.0032523281406611204,
          "validation_metric": 0.0460352897644043,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03779825099991285
        },
        {
          "epoch": 11,
          "updates": 44,
          "samples_seen": 2816,
          "online_loss": 0.013994618318974972,
          "train_loss": 0.011172102764248848,
          "validation_loss": 0.24323999881744385,
          "train_metric": 0.002121283207088709,
          "validation_metric": 0.046184755861759186,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.041345781999552855
        },
        {
          "epoch": 12,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.011776979546993971,
          "train_loss": 0.011524202302098274,
          "validation_loss": 0.26932522654533386,
          "train_metric": 0.0021881372667849064,
          "validation_metric": 0.05113764479756355,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04498770999998669
        },
        {
          "epoch": 13,
          "updates": 52,
          "samples_seen": 3328,
          "online_loss": 0.010419686790555716,
          "train_loss": 0.010910877026617527,
          "validation_loss": 0.25689929723739624,
          "train_metric": 0.0020716830622404814,
          "validation_metric": 0.04877829551696777,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.04858351099937863
        },
        {
          "epoch": 14,
          "updates": 56,
          "samples_seen": 3584,
          "online_loss": 0.0118070108583197,
          "train_loss": 0.008800256066024303,
          "validation_loss": 0.26684582233428955,
          "train_metric": 0.001670932862907648,
          "validation_metric": 0.05066687613725662,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.05207753600006981
        },
        {
          "epoch": 15,
          "updates": 60,
          "samples_seen": 3840,
          "online_loss": 0.012664735142607242,
          "train_loss": 0.008803189732134342,
          "validation_loss": 0.265213280916214,
          "train_metric": 0.0016714901430532336,
          "validation_metric": 0.050356894731521606,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.055605703999390244
        },
        {
          "epoch": 16,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.016566382721066475,
          "train_loss": 0.014035027474164963,
          "validation_loss": 0.24247747659683228,
          "train_metric": 0.002664875937625766,
          "validation_metric": 0.04603997617959976,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.0590739819999726
        },
        {
          "epoch": 17,
          "updates": 68,
          "samples_seen": 4352,
          "online_loss": 0.016224535647779703,
          "train_loss": 0.010416223667562008,
          "validation_loss": 0.26715144515037537,
          "train_metric": 0.0019777617417275906,
          "validation_metric": 0.050724901258945465,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.06251270999928238
        },
        {
          "epoch": 18,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.013322917511686683,
          "train_loss": 0.008675090968608856,
          "validation_loss": 0.26777949929237366,
          "train_metric": 0.0016471671406179667,
          "validation_metric": 0.05084415525197983,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.06614383899977838
        },
        {
          "epoch": 19,
          "updates": 76,
          "samples_seen": 4864,
          "online_loss": 0.010003146366216242,
          "train_loss": 0.009864618070423603,
          "validation_loss": 0.26035594940185547,
          "train_metric": 0.0018730266019701958,
          "validation_metric": 0.04943462088704109,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.0697104840000975
        },
        {
          "epoch": 20,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.009505402151262388,
          "train_loss": 0.010588476434350014,
          "validation_loss": 0.2808798551559448,
          "train_metric": 0.002010467927902937,
          "validation_metric": 0.05333155766129494,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.07336694900004659
        },
        {
          "epoch": 21,
          "updates": 84,
          "samples_seen": 5376,
          "online_loss": 0.011064890713896602,
          "train_loss": 0.011730536818504333,
          "validation_loss": 0.2564202845096588,
          "train_metric": 0.002227314980700612,
          "validation_metric": 0.048687342554330826,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.07695982800032652
        }
      ],
      "best_validation_metric": 0.02332073263823986,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 11,
      "parameters": 2369,
      "train_samples": 256,
      "validation_rmse_m_s": 0.15271127213876473,
      "training_profiles": [
        0,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 8,
          "samples_seen": 512,
          "online_loss": 0.30074516870081425,
          "train_loss": 0.1375461220741272,
          "validation_loss": 0.12516865134239197,
          "train_metric": 0.015804002061486244,
          "validation_metric": 0.01438183430582285,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.00895799699901545
        },
        {
          "epoch": 2,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.120384331792593,
          "train_loss": 0.08434943854808807,
          "validation_loss": 0.06685061752796173,
          "train_metric": 0.009691720828413963,
          "validation_metric": 0.007681114133447409,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.016752358000303502
        },
        {
          "epoch": 3,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.06904806802049279,
          "train_loss": 0.052915215492248535,
          "validation_loss": 0.06376462429761887,
          "train_metric": 0.006079940125346184,
          "validation_metric": 0.0073265330865979195,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.023958736999702523
        },
        {
          "epoch": 4,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.04713609116151929,
          "train_loss": 0.03653869405388832,
          "validation_loss": 0.025614794343709946,
          "train_metric": 0.004198283422738314,
          "validation_metric": 0.002943130675703287,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.030978188000517548
        },
        {
          "epoch": 5,
          "updates": 40,
          "samples_seen": 2560,
          "online_loss": 0.03814950375817716,
          "train_loss": 0.02738906815648079,
          "validation_loss": 0.023487582802772522,
          "train_metric": 0.0031469943933188915,
          "validation_metric": 0.002698715543374419,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03811112299990782
        },
        {
          "epoch": 6,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.025003344286233187,
          "train_loss": 0.028793571516871452,
          "validation_loss": 0.022008638828992844,
          "train_metric": 0.003308371640741825,
          "validation_metric": 0.0025287852622568607,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04520998099906137
        },
        {
          "epoch": 7,
          "updates": 56,
          "samples_seen": 3584,
          "online_loss": 0.030110429623164237,
          "train_loss": 0.034755460917949677,
          "validation_loss": 0.025477871298789978,
          "train_metric": 0.003993390128016472,
          "validation_metric": 0.0029273987747728825,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.051630859999932
        },
        {
          "epoch": 8,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.020155279780738056,
          "train_loss": 0.017706774175167084,
          "validation_loss": 0.018749792128801346,
          "train_metric": 0.0020345021039247513,
          "validation_metric": 0.0021543444599956274,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05866560600043158
        },
        {
          "epoch": 9,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.017685099388472736,
          "train_loss": 0.014036674052476883,
          "validation_loss": 0.011367387138307095,
          "train_metric": 0.001612809021025896,
          "validation_metric": 0.0013061088975518942,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06579331600005389
        },
        {
          "epoch": 10,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.013075668772216886,
          "train_loss": 0.011340843513607979,
          "validation_loss": 0.019624534994363785,
          "train_metric": 0.0013030590489506721,
          "validation_metric": 0.0022548523265868425,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07220123899969622
        },
        {
          "epoch": 11,
          "updates": 88,
          "samples_seen": 5632,
          "online_loss": 0.011291996401268989,
          "train_loss": 0.02612718567252159,
          "validation_loss": 0.027500901371240616,
          "train_metric": 0.0030020044650882483,
          "validation_metric": 0.0031598438508808613,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0784484510004404
        },
        {
          "epoch": 12,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.014433524571359158,
          "train_loss": 0.007938946597278118,
          "validation_loss": 0.011367585510015488,
          "train_metric": 0.0009121822076849639,
          "validation_metric": 0.0013061317149549723,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0846564280000166
        },
        {
          "epoch": 13,
          "updates": 104,
          "samples_seen": 6656,
          "online_loss": 0.009756303043104708,
          "train_loss": 0.011849142611026764,
          "validation_loss": 0.010633177123963833,
          "train_metric": 0.0013614624040201306,
          "validation_metric": 0.001221748418174684,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09149576900017564
        },
        {
          "epoch": 14,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.013497430423740298,
          "train_loss": 0.007487600203603506,
          "validation_loss": 0.012599858455359936,
          "train_metric": 0.0008603226742707193,
          "validation_metric": 0.0014477192889899015,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09792003000075056
        },
        {
          "epoch": 15,
          "updates": 120,
          "samples_seen": 7680,
          "online_loss": 0.012905386625789106,
          "train_loss": 0.010294072329998016,
          "validation_loss": 0.01522810384631157,
          "train_metric": 0.001182785490527749,
          "validation_metric": 0.0017497038934379816,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10408400899905246
        },
        {
          "epoch": 16,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.015175218344666064,
          "train_loss": 0.015756279230117798,
          "validation_loss": 0.011996247805655003,
          "train_metric": 0.001810391084291041,
          "validation_metric": 0.0013783648610115051,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1103101249991596
        },
        {
          "epoch": 17,
          "updates": 136,
          "samples_seen": 8704,
          "online_loss": 0.021520191221497953,
          "train_loss": 0.015210301615297794,
          "validation_loss": 0.010078600607812405,
          "train_metric": 0.0017476582434028387,
          "validation_metric": 0.0011580277932807803,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.11748800699933781
        },
        {
          "epoch": 18,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.014935841201804578,
          "train_loss": 0.008425836451351643,
          "validation_loss": 0.014130564406514168,
          "train_metric": 0.0009681258234195411,
          "validation_metric": 0.0016235967632383108,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.19557799199901638
        },
        {
          "epoch": 19,
          "updates": 152,
          "samples_seen": 9728,
          "online_loss": 0.011382441851310432,
          "train_loss": 0.011215923354029655,
          "validation_loss": 0.01762605458498001,
          "train_metric": 0.0012887058546766639,
          "validation_metric": 0.0020252277608960867,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20196621600007347
        },
        {
          "epoch": 20,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.015102793870028108,
          "train_loss": 0.0081098061054945,
          "validation_loss": 0.013917739503085613,
          "train_metric": 0.0009318139054812491,
          "validation_metric": 0.0015991433756425977,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20832397199956176
        },
        {
          "epoch": 21,
          "updates": 168,
          "samples_seen": 10752,
          "online_loss": 0.007822331041097641,
          "train_loss": 0.009408475831151009,
          "validation_loss": 0.008206742815673351,
          "train_metric": 0.0010810308158397675,
          "validation_metric": 0.0009429519413970411,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21530654900016089
        },
        {
          "epoch": 22,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.007424273586366326,
          "train_loss": 0.0070803058333694935,
          "validation_loss": 0.007396696135401726,
          "train_metric": 0.000813524704426527,
          "validation_metric": 0.0008498778915964067,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22221136900043348
        },
        {
          "epoch": 23,
          "updates": 184,
          "samples_seen": 11776,
          "online_loss": 0.006253229308640584,
          "train_loss": 0.005634869448840618,
          "validation_loss": 0.008943471126258373,
          "train_metric": 0.0006474445690400898,
          "validation_metric": 0.001027601771056652,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22846429499986698
        },
        {
          "epoch": 24,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.006718260323395953,
          "train_loss": 0.011751797050237656,
          "validation_loss": 0.008959870785474777,
          "train_metric": 0.0013502773363143206,
          "validation_metric": 0.0010294860694557428,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.23455712999930256
        },
        {
          "epoch": 25,
          "updates": 200,
          "samples_seen": 12800,
          "online_loss": 0.019003745284862816,
          "train_loss": 0.008439859375357628,
          "validation_loss": 0.010085606947541237,
          "train_metric": 0.0009697368368506432,
          "validation_metric": 0.0011588328052312136,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24075379500027339
        },
        {
          "epoch": 26,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.010407768655568361,
          "train_loss": 0.007757967337965965,
          "validation_loss": 0.013250709511339664,
          "train_metric": 0.0008913878118619323,
          "validation_metric": 0.0015225018141791224,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24696051799946872
        },
        {
          "epoch": 27,
          "updates": 216,
          "samples_seen": 13824,
          "online_loss": 0.010404666420072317,
          "train_loss": 0.007505074609071016,
          "validation_loss": 0.010106694884598255,
          "train_metric": 0.0008623304893262684,
          "validation_metric": 0.0011612557573243976,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2529987140005687
        },
        {
          "epoch": 28,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.008165845240000635,
          "train_loss": 0.010806885547935963,
          "validation_loss": 0.017140494659543037,
          "train_metric": 0.0012417075922712684,
          "validation_metric": 0.0019694368820637465,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2590252429999964
        },
        {
          "epoch": 29,
          "updates": 232,
          "samples_seen": 14848,
          "online_loss": 0.012696578167378902,
          "train_loss": 0.006709414999932051,
          "validation_loss": 0.00726989796385169,
          "train_metric": 0.0007709094206802547,
          "validation_metric": 0.0008353087469004095,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2659321129995078
        },
        {
          "epoch": 30,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.00832796114264056,
          "train_loss": 0.012591690756380558,
          "validation_loss": 0.007583585102111101,
          "train_metric": 0.0014467807486653328,
          "validation_metric": 0.0008713513379916549,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27237175700065563
        },
        {
          "epoch": 31,
          "updates": 248,
          "samples_seen": 15872,
          "online_loss": 0.007434035942424089,
          "train_loss": 0.007631327491253614,
          "validation_loss": 0.005734886042773724,
          "train_metric": 0.0008768367697484791,
          "validation_metric": 0.0006589364493265748,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27926502999980585
        },
        {
          "epoch": 32,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.00991964913555421,
          "train_loss": 0.02519202046096325,
          "validation_loss": 0.012654587626457214,
          "train_metric": 0.0028945545200258493,
          "validation_metric": 0.001454007695429027,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.28560659699905955
        },
        {
          "epoch": 33,
          "updates": 264,
          "samples_seen": 16896,
          "online_loss": 0.011001617764122784,
          "train_loss": 0.008067553862929344,
          "validation_loss": 0.006965528707951307,
          "train_metric": 0.0009269590955227613,
          "validation_metric": 0.0008003368275240064,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.29182181599935575
        },
        {
          "epoch": 34,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.006825832882896066,
          "train_loss": 0.004899003077298403,
          "validation_loss": 0.006366224493831396,
          "train_metric": 0.0005628938088193536,
          "validation_metric": 0.0007314769900403917,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2980436830002873
        },
        {
          "epoch": 35,
          "updates": 280,
          "samples_seen": 17920,
          "online_loss": 0.005557318334467709,
          "train_loss": 0.00438224570825696,
          "validation_loss": 0.010910240933299065,
          "train_metric": 0.000503518502227962,
          "validation_metric": 0.0012535830028355122,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30429975100014417
        },
        {
          "epoch": 36,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.004718813754152507,
          "train_loss": 0.005574928130954504,
          "validation_loss": 0.010733566246926785,
          "train_metric": 0.0006405572639778256,
          "validation_metric": 0.0012332829646766186,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.31061725099971227
        },
        {
          "epoch": 37,
          "updates": 296,
          "samples_seen": 18944,
          "online_loss": 0.007335039030294865,
          "train_loss": 0.0076481676660478115,
          "validation_loss": 0.01689104177057743,
          "train_metric": 0.0008787717670202255,
          "validation_metric": 0.001940774847753346,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3168053310000687
        },
        {
          "epoch": 38,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.005912370019359514,
          "train_loss": 0.004610916133970022,
          "validation_loss": 0.004790410399436951,
          "train_metric": 0.0005297926836647093,
          "validation_metric": 0.0005504164728336036,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3236033769990172
        },
        {
          "epoch": 39,
          "updates": 312,
          "samples_seen": 19968,
          "online_loss": 0.0049631731963017955,
          "train_loss": 0.00799610186368227,
          "validation_loss": 0.009100187569856644,
          "train_metric": 0.0009187493706122041,
          "validation_metric": 0.0010456083109602332,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.32979898499979754
        },
        {
          "epoch": 40,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.008546646975446492,
          "train_loss": 0.025019288063049316,
          "validation_loss": 0.013416511937975883,
          "train_metric": 0.0028747073374688625,
          "validation_metric": 0.001541552715934813,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3358583949993772
        },
        {
          "epoch": 41,
          "updates": 328,
          "samples_seen": 20992,
          "online_loss": 0.012891968828625977,
          "train_loss": 0.009016279131174088,
          "validation_loss": 0.008403605781495571,
          "train_metric": 0.0010359674924984574,
          "validation_metric": 0.0009655713220126927,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3419996609991358
        },
        {
          "epoch": 42,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.00699252198683098,
          "train_loss": 0.011650266125798225,
          "validation_loss": 0.009572231210768223,
          "train_metric": 0.0013386114733293653,
          "validation_metric": 0.0010998460929840803,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3481147669990605
        },
        {
          "epoch": 43,
          "updates": 344,
          "samples_seen": 22016,
          "online_loss": 0.006516897527035326,
          "train_loss": 0.005723107140511274,
          "validation_loss": 0.007527239620685577,
          "train_metric": 0.0006575830630026758,
          "validation_metric": 0.0008648771909065545,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.35433405799994944
        },
        {
          "epoch": 44,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.00543459746404551,
          "train_loss": 0.006403577513992786,
          "validation_loss": 0.008180567994713783,
          "train_metric": 0.0007357688737101853,
          "validation_metric": 0.000939944467972964,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.36053962699952535
        },
        {
          "epoch": 45,
          "updates": 360,
          "samples_seen": 23040,
          "online_loss": 0.004850206911214627,
          "train_loss": 0.003463214263319969,
          "validation_loss": 0.007493342738598585,
          "train_metric": 0.00039792212191969156,
          "validation_metric": 0.0008609825163148344,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3666967610006395
        },
        {
          "epoch": 46,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.002958843127998989,
          "train_loss": 0.0025560674257576466,
          "validation_loss": 0.005094943568110466,
          "train_metric": 0.0002936912642326206,
          "validation_metric": 0.0005854072514921427,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37287859999923967
        },
        {
          "epoch": 47,
          "updates": 376,
          "samples_seen": 24064,
          "online_loss": 0.003576395283744205,
          "train_loss": 0.004842943977564573,
          "validation_loss": 0.010268238373100758,
          "train_metric": 0.0005564525490626693,
          "validation_metric": 0.0011798168998211622,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37907219099906797
        },
        {
          "epoch": 48,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.009436110980459489,
          "train_loss": 0.004969988949596882,
          "validation_loss": 0.005674723070114851,
          "train_metric": 0.0005710499244742095,
          "validation_metric": 0.0006520237657241523,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.38510769799904665
        },
        {
          "epoch": 49,
          "updates": 392,
          "samples_seen": 25088,
          "online_loss": 0.0071722702705301344,
          "train_loss": 0.003176075406372547,
          "validation_loss": 0.008715260773897171,
          "train_metric": 0.0003649299615062773,
          "validation_metric": 0.001001380500383675,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3911907199999405
        },
        {
          "epoch": 50,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.003271451612818055,
          "train_loss": 0.002266546245664358,
          "validation_loss": 0.004971424117684364,
          "train_metric": 0.00026042541139759123,
          "validation_metric": 0.0005712149431928992,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.3971721660000185
        },
        {
          "epoch": 51,
          "updates": 408,
          "samples_seen": 26112,
          "online_loss": 0.0026167549222009256,
          "train_loss": 0.0026973236817866564,
          "validation_loss": 0.006599913351237774,
          "train_metric": 0.0003099215100519359,
          "validation_metric": 0.0007583277765661478,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4032043759998487
        },
        {
          "epoch": 52,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.0023430222063325346,
          "train_loss": 0.00224121636711061,
          "validation_loss": 0.005235291086137295,
          "train_metric": 0.00025751502835191786,
          "validation_metric": 0.0006015332182869315,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4094804310007021
        },
        {
          "epoch": 53,
          "updates": 424,
          "samples_seen": 27136,
          "online_loss": 0.0031622624956071377,
          "train_loss": 0.00548495864495635,
          "validation_loss": 0.007096599321812391,
          "train_metric": 0.0006302198744378984,
          "validation_metric": 0.0008153968374244869,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.41575217100034934
        },
        {
          "epoch": 54,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.002953265437099617,
          "train_loss": 0.0023827541153877974,
          "validation_loss": 0.0045664748176932335,
          "train_metric": 0.0002737776085268706,
          "validation_metric": 0.0005246864166110754,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4224863199997344
        },
        {
          "epoch": 55,
          "updates": 440,
          "samples_seen": 28160,
          "online_loss": 0.002266659852466546,
          "train_loss": 0.0021113352850079536,
          "validation_loss": 0.006734838709235191,
          "train_metric": 0.00024259167548734695,
          "validation_metric": 0.0007738306885585189,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4286584759993275
        },
        {
          "epoch": 56,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.002333877950150054,
          "train_loss": 0.0016040229238569736,
          "validation_loss": 0.005611204542219639,
          "train_metric": 0.00018430167983751744,
          "validation_metric": 0.0006447255727835,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4348893099995621
        },
        {
          "epoch": 57,
          "updates": 456,
          "samples_seen": 29184,
          "online_loss": 0.0021095820848131552,
          "train_loss": 0.010888640768826008,
          "validation_loss": 0.01603577472269535,
          "train_metric": 0.001251101028174162,
          "validation_metric": 0.00184250483289361,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.441194209999594
        },
        {
          "epoch": 58,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.0062746111798333,
          "train_loss": 0.006719579454511404,
          "validation_loss": 0.014649978838860989,
          "train_metric": 0.0007720773573964834,
          "validation_metric": 0.0016832774272188544,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4473468579999462
        },
        {
          "epoch": 59,
          "updates": 472,
          "samples_seen": 30208,
          "online_loss": 0.003750429918000009,
          "train_loss": 0.0025552676524966955,
          "validation_loss": 0.008983115665614605,
          "train_metric": 0.0002935993834398687,
          "validation_metric": 0.0010321568697690964,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4535470009996061
        },
        {
          "epoch": 60,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.002583240559033584,
          "train_loss": 0.004177934490144253,
          "validation_loss": 0.009188644587993622,
          "train_metric": 0.0004800432361662388,
          "validation_metric": 0.0010557720670476556,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.45966869699987
        },
        {
          "epoch": 61,
          "updates": 488,
          "samples_seen": 31232,
          "online_loss": 0.004548664452158846,
          "train_loss": 0.006176943890750408,
          "validation_loss": 0.012443192303180695,
          "train_metric": 0.0007097286870703101,
          "validation_metric": 0.001429718337021768,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.46583663700039324
        },
        {
          "epoch": 62,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.0049194840103155,
          "train_loss": 0.00379566615447402,
          "validation_loss": 0.011418895795941353,
          "train_metric": 0.0004361207247711718,
          "validation_metric": 0.00131202710326761,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4721319719992607
        },
        {
          "epoch": 63,
          "updates": 504,
          "samples_seen": 32256,
          "online_loss": 0.005605347658274695,
          "train_loss": 0.001952821621671319,
          "validation_loss": 0.007548854686319828,
          "train_metric": 0.00022437852749135345,
          "validation_metric": 0.0008673607371747494,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4784048250003252
        },
        {
          "epoch": 64,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.0032026648113969713,
          "train_loss": 0.001945051597431302,
          "validation_loss": 0.004661613143980503,
          "train_metric": 0.00022348576749209315,
          "validation_metric": 0.0005356177571229637,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.484564426000361
        },
        {
          "epoch": 65,
          "updates": 520,
          "samples_seen": 33280,
          "online_loss": 0.002099170371366199,
          "train_loss": 0.0018912985688075423,
          "validation_loss": 0.006646021734923124,
          "train_metric": 0.00021730955631937832,
          "validation_metric": 0.0007636257214471698,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4908123250006611
        },
        {
          "epoch": 66,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.0018178019527113065,
          "train_loss": 0.0020367216784507036,
          "validation_loss": 0.005112410988658667,
          "train_metric": 0.00023401863290928304,
          "validation_metric": 0.000587414251640439,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.49699776799934625
        },
        {
          "epoch": 67,
          "updates": 536,
          "samples_seen": 34304,
          "online_loss": 0.0016093706362880766,
          "train_loss": 0.0022721539717167616,
          "validation_loss": 0.007837805896997452,
          "train_metric": 0.00026106968289241195,
          "validation_metric": 0.000900561164598912,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5031705450001027
        },
        {
          "epoch": 68,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.002148630315787159,
          "train_loss": 0.0014555695233866572,
          "validation_loss": 0.006003613583743572,
          "train_metric": 0.0001672444341238588,
          "validation_metric": 0.00068981311051175,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5094827339999028
        },
        {
          "epoch": 69,
          "updates": 552,
          "samples_seen": 35328,
          "online_loss": 0.0016519010096089914,
          "train_loss": 0.0014561038697138429,
          "validation_loss": 0.00510204304009676,
          "train_metric": 0.00016730581410229206,
          "validation_metric": 0.000586222973652184,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.51558956900044
        },
        {
          "epoch": 70,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.0014530223415931687,
          "train_loss": 0.0016437903977930546,
          "validation_loss": 0.006655781529843807,
          "train_metric": 0.00018887092301156372,
          "validation_metric": 0.0007647470920346677,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5216159039991908
        },
        {
          "epoch": 71,
          "updates": 568,
          "samples_seen": 36352,
          "online_loss": 0.0015760475944261998,
          "train_loss": 0.0019474448636174202,
          "validation_loss": 0.00533836567774415,
          "train_metric": 0.0002237607113784179,
          "validation_metric": 0.0006133763235993683,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5276330710003094
        },
        {
          "epoch": 72,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.0025215489222318865,
          "train_loss": 0.0019024752546101809,
          "validation_loss": 0.004586474969983101,
          "train_metric": 0.00021859374828636646,
          "validation_metric": 0.0005269843386486173,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.533681629000057
        },
        {
          "epoch": 73,
          "updates": 584,
          "samples_seen": 37376,
          "online_loss": 0.00257299840086489,
          "train_loss": 0.002105321967974305,
          "validation_loss": 0.007563198450952768,
          "train_metric": 0.00024190076510421932,
          "validation_metric": 0.0008690089453011751,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5398373629996058
        },
        {
          "epoch": 74,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.0016764251122367568,
          "train_loss": 0.0014961946289986372,
          "validation_loss": 0.006679362617433071,
          "train_metric": 0.00017191223741974682,
          "validation_metric": 0.000767456425819546,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.545980671999132
        }
      ],
      "best_validation_metric": 0.0005246864166110754,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 22,
      "parameters": 2369,
      "train_samples": 512,
      "validation_rmse_m_s": 0.02290603450209301,
      "training_profiles": [
        0,
        4,
        7,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 8,
          "samples_seen": 512,
          "online_loss": 0.33874210715293884,
          "train_loss": 0.15275239944458008,
          "validation_loss": 0.1439913660287857,
          "train_metric": 0.017551196739077568,
          "validation_metric": 0.016544558107852936,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.008851160999256535
        },
        {
          "epoch": 2,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.10888769291341305,
          "train_loss": 0.061448223888874054,
          "validation_loss": 0.04393427073955536,
          "train_metric": 0.007060379721224308,
          "validation_metric": 0.00504803191870451,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.016435510999144753
        },
        {
          "epoch": 3,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.0539900881703943,
          "train_loss": 0.05038190633058548,
          "validation_loss": 0.05085129290819168,
          "train_metric": 0.005788863170892,
          "validation_metric": 0.0058427960611879826,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02298817699920619
        },
        {
          "epoch": 4,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.04435880621895194,
          "train_loss": 0.03541881591081619,
          "validation_loss": 0.024074433371424675,
          "train_metric": 0.004069609101861715,
          "validation_metric": 0.002766143996268511,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0301731559993641
        },
        {
          "epoch": 5,
          "updates": 40,
          "samples_seen": 2560,
          "online_loss": 0.03399531508330256,
          "train_loss": 0.029082821682095528,
          "validation_loss": 0.022443704307079315,
          "train_metric": 0.003341606119647622,
          "validation_metric": 0.002578773768618703,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03738358999908087
        },
        {
          "epoch": 6,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.029298198060132563,
          "train_loss": 0.025052469223737717,
          "validation_loss": 0.02044091932475567,
          "train_metric": 0.0028785199392586946,
          "validation_metric": 0.0023486544378101826,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.044611156999962986
        },
        {
          "epoch": 7,
          "updates": 56,
          "samples_seen": 3584,
          "online_loss": 0.031684620305895805,
          "train_loss": 0.04241085797548294,
          "validation_loss": 0.032621435821056366,
          "train_metric": 0.004872992634773254,
          "validation_metric": 0.003748191986232996,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05118701199899078
        },
        {
          "epoch": 8,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.04396696831099689,
          "train_loss": 0.08762447535991669,
          "validation_loss": 0.07266072928905487,
          "train_metric": 0.010068021714687347,
          "validation_metric": 0.008348692208528519,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0574666119991889
        },
        {
          "epoch": 9,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.06814928981475532,
          "train_loss": 0.0414615236222744,
          "validation_loss": 0.036158010363578796,
          "train_metric": 0.004763915203511715,
          "validation_metric": 0.004154542461037636,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06374291499923856
        },
        {
          "epoch": 10,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.04521411994937807,
          "train_loss": 0.03689274936914444,
          "validation_loss": 0.03160436078906059,
          "train_metric": 0.004238964524120092,
          "validation_metric": 0.003631330095231533,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06999199299934844
        },
        {
          "epoch": 11,
          "updates": 88,
          "samples_seen": 5632,
          "online_loss": 0.03321108070667833,
          "train_loss": 0.026114460080862045,
          "validation_loss": 0.018952295184135437,
          "train_metric": 0.003000542288646102,
          "validation_metric": 0.002177612390369177,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07685992500046268
        },
        {
          "epoch": 12,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.030252517550252378,
          "train_loss": 0.02160101756453514,
          "validation_loss": 0.01811099238693714,
          "train_metric": 0.0024819497484713793,
          "validation_metric": 0.0020809462293982506,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08379243899980793
        },
        {
          "epoch": 13,
          "updates": 104,
          "samples_seen": 6656,
          "online_loss": 0.034894408541731536,
          "train_loss": 0.035564906895160675,
          "validation_loss": 0.025647375732660294,
          "train_metric": 0.004086395725607872,
          "validation_metric": 0.0029468745924532413,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09022407900010876
        },
        {
          "epoch": 14,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.029745517356786877,
          "train_loss": 0.024186627939343452,
          "validation_loss": 0.025039203464984894,
          "train_metric": 0.0027790353633463383,
          "validation_metric": 0.0028769958298653364,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09676283699991473
        },
        {
          "epoch": 15,
          "updates": 120,
          "samples_seen": 7680,
          "online_loss": 0.023717874311842024,
          "train_loss": 0.023796284571290016,
          "validation_loss": 0.016862567514181137,
          "train_metric": 0.002734184730798006,
          "validation_metric": 0.0019375032279640436,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10392751300059899
        },
        {
          "epoch": 16,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.019807393313385546,
          "train_loss": 0.017166638746857643,
          "validation_loss": 0.013067662715911865,
          "train_metric": 0.0019724410958588123,
          "validation_metric": 0.0015014699893072248,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.11103722500047297
        },
        {
          "epoch": 17,
          "updates": 136,
          "samples_seen": 8704,
          "online_loss": 0.015457500354386866,
          "train_loss": 0.014416879042983055,
          "validation_loss": 0.013176307082176208,
          "train_metric": 0.0016564945690333843,
          "validation_metric": 0.0015139529714360833,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.11749066499942273
        },
        {
          "epoch": 18,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.014127449947409332,
          "train_loss": 0.015894349664449692,
          "validation_loss": 0.016338564455509186,
          "train_metric": 0.0018262553494423628,
          "validation_metric": 0.0018772950861603022,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.19661781799914024
        },
        {
          "epoch": 19,
          "updates": 152,
          "samples_seen": 9728,
          "online_loss": 0.013395574991591275,
          "train_loss": 0.013818839564919472,
          "validation_loss": 0.015216388739645481,
          "train_metric": 0.0015877799596637487,
          "validation_metric": 0.001748357666656375,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20334831199943437
        },
        {
          "epoch": 20,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.021007888251915574,
          "train_loss": 0.05866473540663719,
          "validation_loss": 0.04881053790450096,
          "train_metric": 0.006740557495504618,
          "validation_metric": 0.00560831418260932,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2095685570002388
        },
        {
          "epoch": 21,
          "updates": 168,
          "samples_seen": 10752,
          "online_loss": 0.03897643066011369,
          "train_loss": 0.024562597274780273,
          "validation_loss": 0.03251754492521286,
          "train_metric": 0.0028222338296473026,
          "validation_metric": 0.003736254759132862,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2157170779992157
        },
        {
          "epoch": 22,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.03296506387414411,
          "train_loss": 0.014949101954698563,
          "validation_loss": 0.01200372539460659,
          "train_metric": 0.0017176464898511767,
          "validation_metric": 0.0013792237732559443,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22254310300013458
        },
        {
          "epoch": 23,
          "updates": 184,
          "samples_seen": 11776,
          "online_loss": 0.018007585080340505,
          "train_loss": 0.015199336223304272,
          "validation_loss": 0.011989936232566833,
          "train_metric": 0.0017463983967900276,
          "validation_metric": 0.0013776395935565233,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22953648200018506
        },
        {
          "epoch": 24,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.013226119452156126,
          "train_loss": 0.01365899108350277,
          "validation_loss": 0.011390596628189087,
          "train_metric": 0.0015694133471697569,
          "validation_metric": 0.0013087758561596274,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.23638013799973123
        },
        {
          "epoch": 25,
          "updates": 200,
          "samples_seen": 12800,
          "online_loss": 0.011350761633366346,
          "train_loss": 0.009608411230146885,
          "validation_loss": 0.010620858520269394,
          "train_metric": 0.0011040031677111983,
          "validation_metric": 0.0012203329242765903,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24326278599983198
        },
        {
          "epoch": 26,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.014300157956313342,
          "train_loss": 0.009373362176120281,
          "validation_loss": 0.011989312246441841,
          "train_metric": 0.0010769959772005677,
          "validation_metric": 0.001377567881718278,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24974816699977964
        },
        {
          "epoch": 27,
          "updates": 216,
          "samples_seen": 13824,
          "online_loss": 0.012993209762498736,
          "train_loss": 0.017109155654907227,
          "validation_loss": 0.022792045027017593,
          "train_metric": 0.0019658361561596394,
          "validation_metric": 0.002618798054754734,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25606856800004607
        },
        {
          "epoch": 28,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.016498312004841864,
          "train_loss": 0.008104851469397545,
          "validation_loss": 0.009930779226124287,
          "train_metric": 0.0009312446345575154,
          "validation_metric": 0.001141043147072196,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2630535989992495
        },
        {
          "epoch": 29,
          "updates": 232,
          "samples_seen": 14848,
          "online_loss": 0.009896628966089338,
          "train_loss": 0.007669086568057537,
          "validation_loss": 0.0073914313688874245,
          "train_metric": 0.0008811754523776472,
          "validation_metric": 0.0008492728811688721,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2700545389998297
        },
        {
          "epoch": 30,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.009097594331251457,
          "train_loss": 0.006528021767735481,
          "validation_loss": 0.00910231750458479,
          "train_metric": 0.0007500674109905958,
          "validation_metric": 0.0010458531323820353,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2764433110005484
        },
        {
          "epoch": 31,
          "updates": 248,
          "samples_seen": 15872,
          "online_loss": 0.01054160637431778,
          "train_loss": 0.00876102689653635,
          "validation_loss": 0.008132996968925,
          "train_metric": 0.0010066390968859196,
          "validation_metric": 0.0009344785357825458,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2826715259998309
        },
        {
          "epoch": 32,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.006633295852225274,
          "train_loss": 0.00705468887463212,
          "validation_loss": 0.00602885102853179,
          "train_metric": 0.0008105812012217939,
          "validation_metric": 0.0006927127833478153,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2896738950003055
        },
        {
          "epoch": 33,
          "updates": 264,
          "samples_seen": 16896,
          "online_loss": 0.008123699168208987,
          "train_loss": 0.0048049152828752995,
          "validation_loss": 0.009248670190572739,
          "train_metric": 0.0005520831327885389,
          "validation_metric": 0.0010626689763739705,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2962950919991272
        },
        {
          "epoch": 34,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.0072242166206706315,
          "train_loss": 0.008463036268949509,
          "validation_loss": 0.013247063383460045,
          "train_metric": 0.0009723998955450952,
          "validation_metric": 0.0015220828354358673,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3025024429989571
        },
        {
          "epoch": 35,
          "updates": 280,
          "samples_seen": 17920,
          "online_loss": 0.007407268596580252,
          "train_loss": 0.00784765463322401,
          "validation_loss": 0.013155612163245678,
          "train_metric": 0.0009016926633194089,
          "validation_metric": 0.0015115754213184118,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30873556900041876
        },
        {
          "epoch": 36,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.0076035889505874366,
          "train_loss": 0.006654494442045689,
          "validation_loss": 0.01557324267923832,
          "train_metric": 0.0007645991863682866,
          "validation_metric": 0.0017893603071570396,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.31488908900064416
        },
        {
          "epoch": 37,
          "updates": 296,
          "samples_seen": 18944,
          "online_loss": 0.005219005804974586,
          "train_loss": 0.0072506205178797245,
          "validation_loss": 0.009493772871792316,
          "train_metric": 0.0008330938289873302,
          "validation_metric": 0.0010908311232924461,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.32108176499968977
        },
        {
          "epoch": 38,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.00807711633387953,
          "train_loss": 0.007687176577746868,
          "validation_loss": 0.010831372812390327,
          "train_metric": 0.0008832538733258843,
          "validation_metric": 0.00124452100135386,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3271311730004527
        },
        {
          "epoch": 39,
          "updates": 312,
          "samples_seen": 19968,
          "online_loss": 0.010322583897504956,
          "train_loss": 0.01164124347269535,
          "validation_loss": 0.019251560792326927,
          "train_metric": 0.0013375749113038182,
          "validation_metric": 0.0022119979839771986,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3333337099993514
        },
        {
          "epoch": 40,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.012872602499555796,
          "train_loss": 0.006928805727511644,
          "validation_loss": 0.011990775354206562,
          "train_metric": 0.0007961173541843891,
          "validation_metric": 0.001377735985442996,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3396373539999331
        },
        {
          "epoch": 41,
          "updates": 328,
          "samples_seen": 20992,
          "online_loss": 0.014497727155685425,
          "train_loss": 0.008550217375159264,
          "validation_loss": 0.010716531425714493,
          "train_metric": 0.0009824171429499984,
          "validation_metric": 0.0012313257902860641,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.34585950700056856
        },
        {
          "epoch": 42,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.01017152820713818,
          "train_loss": 0.006522581912577152,
          "validation_loss": 0.009163047187030315,
          "train_metric": 0.0007494425517506897,
          "validation_metric": 0.00105283095035702,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3520902149994072
        },
        {
          "epoch": 43,
          "updates": 344,
          "samples_seen": 22016,
          "online_loss": 0.007715357322013006,
          "train_loss": 0.0075792428106069565,
          "validation_loss": 0.007299985270947218,
          "train_metric": 0.0008708523819223046,
          "validation_metric": 0.0008387657580897212,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3583701530005783
        },
        {
          "epoch": 44,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.006303383852355182,
          "train_loss": 0.005337366834282875,
          "validation_loss": 0.009282140992581844,
          "train_metric": 0.0006132615962997079,
          "validation_metric": 0.0010665147565305233,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.36458384599973215
        },
        {
          "epoch": 45,
          "updates": 360,
          "samples_seen": 23040,
          "online_loss": 0.0050206883024657145,
          "train_loss": 0.003515695920214057,
          "validation_loss": 0.006681270897388458,
          "train_metric": 0.00040395226096734405,
          "validation_metric": 0.0007676758104935288,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.37076693900053215
        },
        {
          "epoch": 46,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.005680825503077358,
          "train_loss": 0.00340977287851274,
          "validation_loss": 0.00542603200301528,
          "train_metric": 0.00039178173756226897,
          "validation_metric": 0.0006234491593204439,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.377498244999515
        },
        {
          "epoch": 47,
          "updates": 376,
          "samples_seen": 24064,
          "online_loss": 0.0049199053610209376,
          "train_loss": 0.004894658457487822,
          "validation_loss": 0.010761183686554432,
          "train_metric": 0.0005623945035040379,
          "validation_metric": 0.001236456329934299,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.38372643399998196
        },
        {
          "epoch": 48,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.002886471847887151,
          "train_loss": 0.003012551460415125,
          "validation_loss": 0.0066860150545835495,
          "train_metric": 0.00034614113974384964,
          "validation_metric": 0.0007682208670303226,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.389996453999629
        },
        {
          "epoch": 49,
          "updates": 392,
          "samples_seen": 25088,
          "online_loss": 0.0028647197177633643,
          "train_loss": 0.0035250182263553143,
          "validation_loss": 0.008822726085782051,
          "train_metric": 0.0004050233692396432,
          "validation_metric": 0.0010137282079085708,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.3962852530003147
        },
        {
          "epoch": 50,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.003412011908949353,
          "train_loss": 0.006242081988602877,
          "validation_loss": 0.005947654601186514,
          "train_metric": 0.0007172132027335465,
          "validation_metric": 0.0006833834340795875,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4026798979994055
        },
        {
          "epoch": 51,
          "updates": 408,
          "samples_seen": 26112,
          "online_loss": 0.004760742711368948,
          "train_loss": 0.0068327453918755054,
          "validation_loss": 0.005502735730260611,
          "train_metric": 0.0007850800757296383,
          "validation_metric": 0.0006322625558823347,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.408861464000438
        },
        {
          "epoch": 52,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.00774645947967656,
          "train_loss": 0.0025272781495004892,
          "validation_loss": 0.007225301116704941,
          "train_metric": 0.00029038338107056916,
          "validation_metric": 0.0008301846100948751,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4150631440006691
        },
        {
          "epoch": 53,
          "updates": 424,
          "samples_seen": 27136,
          "online_loss": 0.010391998370323563,
          "train_loss": 0.006662149913609028,
          "validation_loss": 0.012603695504367352,
          "train_metric": 0.0007654787041246891,
          "validation_metric": 0.0014481603866443038,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4214307160000317
        },
        {
          "epoch": 54,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.0036065784515812993,
          "train_loss": 0.0032307219225913286,
          "validation_loss": 0.00784010998904705,
          "train_metric": 0.00037120882188901305,
          "validation_metric": 0.0009008258348330855,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4276075409998157
        },
        {
          "epoch": 55,
          "updates": 440,
          "samples_seen": 28160,
          "online_loss": 0.0038629205664619803,
          "train_loss": 0.003249124623835087,
          "validation_loss": 0.004287763498723507,
          "train_metric": 0.0003733233315870166,
          "validation_metric": 0.0004926625988446176,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4345128109998768
        },
        {
          "epoch": 56,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.0032905418192967772,
          "train_loss": 0.002736113267019391,
          "validation_loss": 0.007430641911923885,
          "train_metric": 0.0003143784706480801,
          "validation_metric": 0.0008537781541235745,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4408974540001509
        },
        {
          "epoch": 57,
          "updates": 456,
          "samples_seen": 29184,
          "online_loss": 0.0028744826995534822,
          "train_loss": 0.0029304574709385633,
          "validation_loss": 0.004741361364722252,
          "train_metric": 0.00033670858829282224,
          "validation_metric": 0.0005447808071039617,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.44713652199970966
        },
        {
          "epoch": 58,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.002833608421497047,
          "train_loss": 0.00268420297652483,
          "validation_loss": 0.00631787208840251,
          "train_metric": 0.0003084140189457685,
          "validation_metric": 0.0007259213016368449,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.45345597099912993
        },
        {
          "epoch": 59,
          "updates": 472,
          "samples_seen": 30208,
          "online_loss": 0.002542818183428608,
          "train_loss": 0.001965518342331052,
          "validation_loss": 0.0061184014193713665,
          "train_metric": 0.00022583735699299723,
          "validation_metric": 0.0007030022097751498,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.45982151799944404
        },
        {
          "epoch": 60,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.002980183213367127,
          "train_loss": 0.003831483656540513,
          "validation_loss": 0.011494156904518604,
          "train_metric": 0.0004402361228130758,
          "validation_metric": 0.0013206746662035584,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.46612441099932767
        },
        {
          "epoch": 61,
          "updates": 488,
          "samples_seen": 31232,
          "online_loss": 0.004958357450959738,
          "train_loss": 0.007808977738022804,
          "validation_loss": 0.012509914115071297,
          "train_metric": 0.000897248974069953,
          "validation_metric": 0.0014373848680406809,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4724114679993363
        },
        {
          "epoch": 62,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.0052762756022275425,
          "train_loss": 0.0026829810813069344,
          "validation_loss": 0.004980673547834158,
          "train_metric": 0.0003082736220676452,
          "validation_metric": 0.0005722776404581964,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.478603371000645
        },
        {
          "epoch": 63,
          "updates": 504,
          "samples_seen": 32256,
          "online_loss": 0.004856336672673933,
          "train_loss": 0.0049470881931483746,
          "validation_loss": 0.004627629183232784,
          "train_metric": 0.0005684187053702772,
          "validation_metric": 0.0005317130126059055,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.48483836199920916
        },
        {
          "epoch": 64,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.010370406627771445,
          "train_loss": 0.006312601733952761,
          "validation_loss": 0.0070257484912872314,
          "train_metric": 0.0007253157673403621,
          "validation_metric": 0.0008072560885921121,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.49122136599908117
        },
        {
          "epoch": 65,
          "updates": 520,
          "samples_seen": 33280,
          "online_loss": 0.006157988638733514,
          "train_loss": 0.0033145705237984657,
          "validation_loss": 0.004450424108654261,
          "train_metric": 0.0003808430046774447,
          "validation_metric": 0.0005113522056490183,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.49739583100017626
        },
        {
          "epoch": 66,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.005276798925478943,
          "train_loss": 0.003296098904684186,
          "validation_loss": 0.007606136612594128,
          "train_metric": 0.00037872063694521785,
          "validation_metric": 0.000873942393809557,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5034603419990162
        },
        {
          "epoch": 67,
          "updates": 536,
          "samples_seen": 34304,
          "online_loss": 0.004063361837324919,
          "train_loss": 0.004084300249814987,
          "validation_loss": 0.008221114054322243,
          "train_metric": 0.0004692847141996026,
          "validation_metric": 0.0009446031763218343,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5095416509993811
        },
        {
          "epoch": 68,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.002697712028748356,
          "train_loss": 0.002691656583920121,
          "validation_loss": 0.005264964886009693,
          "train_metric": 0.0003092704282607883,
          "validation_metric": 0.0006049427320249379,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5156280070004868
        },
        {
          "epoch": 69,
          "updates": 552,
          "samples_seen": 35328,
          "online_loss": 0.002031738666119054,
          "train_loss": 0.0019753770902752876,
          "validation_loss": 0.005678872112184763,
          "train_metric": 0.00022697013628203422,
          "validation_metric": 0.0006525004282593727,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5217760149989772
        },
        {
          "epoch": 70,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.0021291924713295884,
          "train_loss": 0.0018141502514481544,
          "validation_loss": 0.005989687982946634,
          "train_metric": 0.00020844521350227296,
          "validation_metric": 0.0006882131565362215,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5279509499996493
        },
        {
          "epoch": 71,
          "updates": 568,
          "samples_seen": 36352,
          "online_loss": 0.002488472717232071,
          "train_loss": 0.004485276993364096,
          "validation_loss": 0.004687621723860502,
          "train_metric": 0.0005153567180968821,
          "validation_metric": 0.0005386060802266002,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5340202909992513
        },
        {
          "epoch": 72,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.00279061273613479,
          "train_loss": 0.0029226422775536776,
          "validation_loss": 0.009920493699610233,
          "train_metric": 0.0003358105896040797,
          "validation_metric": 0.0011398611823096871,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5399672719995579
        },
        {
          "epoch": 73,
          "updates": 584,
          "samples_seen": 37376,
          "online_loss": 0.002789540092635434,
          "train_loss": 0.0018113856203854084,
          "validation_loss": 0.007273186929523945,
          "train_metric": 0.0002081275888485834,
          "validation_metric": 0.0008356866310350597,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5459845989989844
        },
        {
          "epoch": 74,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.0023265380004886538,
          "train_loss": 0.0016725758323445916,
          "validation_loss": 0.005903279408812523,
          "train_metric": 0.00019217838416807353,
          "validation_metric": 0.0006782847340218723,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5520495609998761
        },
        {
          "epoch": 75,
          "updates": 600,
          "samples_seen": 38400,
          "online_loss": 0.0018011822467087768,
          "train_loss": 0.0019265555310994387,
          "validation_loss": 0.007764906156808138,
          "train_metric": 0.00022136056213639677,
          "validation_metric": 0.0008921850239858031,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5581973970001854
        }
      ],
      "best_validation_metric": 0.0004926625988446176,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 33,
      "parameters": 2369,
      "train_samples": 512,
      "validation_rmse_m_s": 0.022196004118863775,
      "training_profiles": [
        0,
        4,
        7,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 8,
          "samples_seen": 512,
          "online_loss": 0.43743024580180645,
          "train_loss": 0.14514164626598358,
          "validation_loss": 0.13035011291503906,
          "train_metric": 0.016676723957061768,
          "validation_metric": 0.01497718133032322,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.009033632999489782
        },
        {
          "epoch": 2,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.13397582713514566,
          "train_loss": 0.07702149450778961,
          "validation_loss": 0.06441806256771088,
          "train_metric": 0.008849742822349072,
          "validation_metric": 0.007401613052934408,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.017122669998570927
        },
        {
          "epoch": 3,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.07483507785946131,
          "train_loss": 0.06088092550635338,
          "validation_loss": 0.05047868564724922,
          "train_metric": 0.006995196919888258,
          "validation_metric": 0.005799983162432909,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02430468300008215
        },
        {
          "epoch": 4,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.055466287303715944,
          "train_loss": 0.04980260878801346,
          "validation_loss": 0.05694778263568878,
          "train_metric": 0.005722302943468094,
          "validation_metric": 0.006543280091136694,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.030828433998976834
        },
        {
          "epoch": 5,
          "updates": 40,
          "samples_seen": 2560,
          "online_loss": 0.05081207235343754,
          "train_loss": 0.04125731810927391,
          "validation_loss": 0.03261909261345863,
          "train_metric": 0.004740451462566853,
          "validation_metric": 0.003747922368347645,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.038278301999525866
        },
        {
          "epoch": 6,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.04801018023863435,
          "train_loss": 0.03397592157125473,
          "validation_loss": 0.031210457906126976,
          "train_metric": 0.00390382157638669,
          "validation_metric": 0.003586071077734232,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04570856999998796
        },
        {
          "epoch": 7,
          "updates": 56,
          "samples_seen": 3584,
          "online_loss": 0.033210801309905946,
          "train_loss": 0.03439726680517197,
          "validation_loss": 0.028376713395118713,
          "train_metric": 0.0039522345177829266,
          "validation_metric": 0.0032604746520519257,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.053116443999897456
        },
        {
          "epoch": 8,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.03156958078034222,
          "train_loss": 0.03783886507153511,
          "validation_loss": 0.02976493537425995,
          "train_metric": 0.0043476722203195095,
          "validation_metric": 0.003419980639591813,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0596025129998452
        },
        {
          "epoch": 9,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.027924984227865934,
          "train_loss": 0.021938549354672432,
          "validation_loss": 0.018254486843943596,
          "train_metric": 0.0025207316502928734,
          "validation_metric": 0.0020974341314285994,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06681779700011248
        },
        {
          "epoch": 10,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.022839101613499224,
          "train_loss": 0.02055368386209011,
          "validation_loss": 0.017211874946951866,
          "train_metric": 0.002361610997468233,
          "validation_metric": 0.0019776385743170977,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07411360199876071
        },
        {
          "epoch": 11,
          "updates": 88,
          "samples_seen": 5632,
          "online_loss": 0.02101140870945528,
          "train_loss": 0.020499657839536667,
          "validation_loss": 0.017041459679603577,
          "train_metric": 0.0023554032668471336,
          "validation_metric": 0.001958057750016451,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08126533499853394
        },
        {
          "epoch": 12,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.019150728825479746,
          "train_loss": 0.017757480964064598,
          "validation_loss": 0.015658756718039513,
          "train_metric": 0.0020403284579515457,
          "validation_metric": 0.0017991856439039111,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08847362799861003
        },
        {
          "epoch": 13,
          "updates": 104,
          "samples_seen": 6656,
          "online_loss": 0.01553925801999867,
          "train_loss": 0.012755809351801872,
          "validation_loss": 0.012175877578556538,
          "train_metric": 0.001465638168156147,
          "validation_metric": 0.0013990041334182024,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0957017389991961
        },
        {
          "epoch": 14,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.020522375009022653,
          "train_loss": 0.06612290441989899,
          "validation_loss": 0.0552271343767643,
          "train_metric": 0.007597498130053282,
          "validation_metric": 0.0063455794006586075,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10239024799921026
        },
        {
          "epoch": 15,
          "updates": 120,
          "samples_seen": 7680,
          "online_loss": 0.056935574393719435,
          "train_loss": 0.03704623132944107,
          "validation_loss": 0.033624254167079926,
          "train_metric": 0.0042565991170704365,
          "validation_metric": 0.0038634154479950666,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10869265899964375
        },
        {
          "epoch": 16,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.04619953688234091,
          "train_loss": 0.028750553727149963,
          "validation_loss": 0.02352171018719673,
          "train_metric": 0.0033034286461770535,
          "validation_metric": 0.0027026364114135504,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1150740439989022
        },
        {
          "epoch": 17,
          "updates": 136,
          "samples_seen": 8704,
          "online_loss": 0.02146038122009486,
          "train_loss": 0.01732872985303402,
          "validation_loss": 0.015235017985105515,
          "train_metric": 0.001991064753383398,
          "validation_metric": 0.0017504983115941286,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.12140835199897992
        },
        {
          "epoch": 18,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.017121856741141528,
          "train_loss": 0.014472289010882378,
          "validation_loss": 0.01260227058082819,
          "train_metric": 0.0016628610901534557,
          "validation_metric": 0.0014479964738711715,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20075410899880808
        },
        {
          "epoch": 19,
          "updates": 152,
          "samples_seen": 9728,
          "online_loss": 0.014139877632260323,
          "train_loss": 0.017885847017169,
          "validation_loss": 0.013132980093359947,
          "train_metric": 0.0020550771150738,
          "validation_metric": 0.00150897481944412,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2073431119988527
        },
        {
          "epoch": 20,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.022730416269041598,
          "train_loss": 0.020630812272429466,
          "validation_loss": 0.015708697959780693,
          "train_metric": 0.002370473463088274,
          "validation_metric": 0.001804923638701439,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2137400149986206
        },
        {
          "epoch": 21,
          "updates": 168,
          "samples_seen": 10752,
          "online_loss": 0.037655188876669854,
          "train_loss": 0.0367177315056324,
          "validation_loss": 0.02626103349030018,
          "train_metric": 0.004218854941427708,
          "validation_metric": 0.0030173840932548046,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22014615499938373
        },
        {
          "epoch": 22,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.02984950807876885,
          "train_loss": 0.019520802423357964,
          "validation_loss": 0.015376284718513489,
          "train_metric": 0.002242933725938201,
          "validation_metric": 0.0017667298670858145,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22640595299890265
        },
        {
          "epoch": 23,
          "updates": 184,
          "samples_seen": 11776,
          "online_loss": 0.020379040739499032,
          "train_loss": 0.01635872572660446,
          "validation_loss": 0.016761481761932373,
          "train_metric": 0.0018796119838953018,
          "validation_metric": 0.001925888704136014,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2327110009991884
        },
        {
          "epoch": 24,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.01360907033085823,
          "train_loss": 0.011752190999686718,
          "validation_loss": 0.011654357425868511,
          "train_metric": 0.0013503226218745112,
          "validation_metric": 0.0013390814419835806,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2397499019989482
        },
        {
          "epoch": 25,
          "updates": 200,
          "samples_seen": 12800,
          "online_loss": 0.010955725272651762,
          "train_loss": 0.017219705507159233,
          "validation_loss": 0.011570574715733528,
          "train_metric": 0.001978538231924176,
          "validation_metric": 0.0013294550590217113,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24680834699938714
        },
        {
          "epoch": 26,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.0186813700129278,
          "train_loss": 0.03824961185455322,
          "validation_loss": 0.025828778743743896,
          "train_metric": 0.004394866526126862,
          "validation_metric": 0.0029677178245037794,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2532647949992679
        },
        {
          "epoch": 27,
          "updates": 216,
          "samples_seen": 13824,
          "online_loss": 0.027657047379761934,
          "train_loss": 0.02549782209098339,
          "validation_loss": 0.0186738483607769,
          "train_metric": 0.002929690992459655,
          "validation_metric": 0.002145618898794055,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25946362799913913
        },
        {
          "epoch": 28,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.023223139462061226,
          "train_loss": 0.010247087106108665,
          "validation_loss": 0.00807290617376566,
          "train_metric": 0.0011773869628086686,
          "validation_metric": 0.000927574117667973,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2662370119996922
        },
        {
          "epoch": 29,
          "updates": 232,
          "samples_seen": 14848,
          "online_loss": 0.021911024843575433,
          "train_loss": 0.024809909984469414,
          "validation_loss": 0.027485931292176247,
          "train_metric": 0.0028506501112133265,
          "validation_metric": 0.003158124163746834,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27274378899892326
        },
        {
          "epoch": 30,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.01741655037039891,
          "train_loss": 0.013061138801276684,
          "validation_loss": 0.01670430786907673,
          "train_metric": 0.0015007203910499811,
          "validation_metric": 0.0019193191546946764,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2790087299999868
        },
        {
          "epoch": 31,
          "updates": 248,
          "samples_seen": 15872,
          "online_loss": 0.014726106193847954,
          "train_loss": 0.008558278903365135,
          "validation_loss": 0.007268024142831564,
          "train_metric": 0.0009833433432504535,
          "validation_metric": 0.0008350934949703515,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.28592397299871664
        },
        {
          "epoch": 32,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.009918647294398397,
          "train_loss": 0.007458877749741077,
          "validation_loss": 0.00782529916614294,
          "train_metric": 0.0008570224163122475,
          "validation_metric": 0.0008991241920739412,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2923247499984427
        },
        {
          "epoch": 33,
          "updates": 264,
          "samples_seen": 16896,
          "online_loss": 0.007600008539156988,
          "train_loss": 0.007633129134774208,
          "validation_loss": 0.006630163639783859,
          "train_metric": 0.0008770438726060092,
          "validation_metric": 0.0007618035888299346,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2994258689996059
        },
        {
          "epoch": 34,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.00799759806250222,
          "train_loss": 0.005954699125140905,
          "validation_loss": 0.005748753435909748,
          "train_metric": 0.0006841928116045892,
          "validation_metric": 0.000660529884044081,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30655024599946046
        },
        {
          "epoch": 35,
          "updates": 280,
          "samples_seen": 17920,
          "online_loss": 0.008666988287586719,
          "train_loss": 0.012372590601444244,
          "validation_loss": 0.01474210899323225,
          "train_metric": 0.0014216064009815454,
          "validation_metric": 0.0016938630724325776,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.31299674000001687
        },
        {
          "epoch": 36,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.01142809633165598,
          "train_loss": 0.007589329034090042,
          "validation_loss": 0.0056255776435136795,
          "train_metric": 0.0008720112964510918,
          "validation_metric": 0.000646376924123615,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3200336479985708
        },
        {
          "epoch": 37,
          "updates": 296,
          "samples_seen": 18944,
          "online_loss": 0.009837308738497086,
          "train_loss": 0.006467379629611969,
          "validation_loss": 0.006987590808421373,
          "train_metric": 0.0007430996629409492,
          "validation_metric": 0.000802871712949127,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3265483459999814
        },
        {
          "epoch": 38,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.008615338942036033,
          "train_loss": 0.010672320611774921,
          "validation_loss": 0.007270217873156071,
          "train_metric": 0.0012262460077181458,
          "validation_metric": 0.0008353455341421068,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.33287861999997403
        },
        {
          "epoch": 39,
          "updates": 312,
          "samples_seen": 19968,
          "online_loss": 0.00850760523462668,
          "train_loss": 0.01758895069360733,
          "validation_loss": 0.010196945630013943,
          "train_metric": 0.0020209643989801407,
          "validation_metric": 0.0011716256849467754,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3392513709986815
        },
        {
          "epoch": 40,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.016755231597926468,
          "train_loss": 0.017493652179837227,
          "validation_loss": 0.012459956109523773,
          "train_metric": 0.002010014606639743,
          "validation_metric": 0.0014316446613520384,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.34553971599962097
        },
        {
          "epoch": 41,
          "updates": 328,
          "samples_seen": 20992,
          "online_loss": 0.015337337972596288,
          "train_loss": 0.015248069539666176,
          "validation_loss": 0.008565226569771767,
          "train_metric": 0.0017519979737699032,
          "validation_metric": 0.0009841416031122208,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3520866439994279
        },
        {
          "epoch": 42,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.008155822259141132,
          "train_loss": 0.005729135125875473,
          "validation_loss": 0.0066061257384717464,
          "train_metric": 0.000658275734167546,
          "validation_metric": 0.0007590416935272515,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3581964959994366
        },
        {
          "epoch": 43,
          "updates": 344,
          "samples_seen": 22016,
          "online_loss": 0.005368857993744314,
          "train_loss": 0.0046254713088274,
          "validation_loss": 0.005517695564776659,
          "train_metric": 0.0005314651061780751,
          "validation_metric": 0.0006339812534861267,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3650023819991475
        },
        {
          "epoch": 44,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.00783945299917832,
          "train_loss": 0.006188864354044199,
          "validation_loss": 0.005281725898385048,
          "train_metric": 0.0007110983715392649,
          "validation_metric": 0.000606868474278599,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37191564199929417
        },
        {
          "epoch": 45,
          "updates": 360,
          "samples_seen": 23040,
          "online_loss": 0.005777406768174842,
          "train_loss": 0.004472400527447462,
          "validation_loss": 0.00492186238989234,
          "train_metric": 0.0005138773121871054,
          "validation_metric": 0.0005655202548950911,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37894975499875727
        },
        {
          "epoch": 46,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.006344806955894455,
          "train_loss": 0.0053038084879517555,
          "validation_loss": 0.006339316722005606,
          "train_metric": 0.000609405804425478,
          "validation_metric": 0.0007283854065462947,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.38561482799923397
        },
        {
          "epoch": 47,
          "updates": 376,
          "samples_seen": 24064,
          "online_loss": 0.005885673803277314,
          "train_loss": 0.007229005452245474,
          "validation_loss": 0.007370363920927048,
          "train_metric": 0.0008306101663038135,
          "validation_metric": 0.0008468522573821247,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3918779599989648
        },
        {
          "epoch": 48,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.00805703501100652,
          "train_loss": 0.004510094877332449,
          "validation_loss": 0.006522946059703827,
          "train_metric": 0.0005182083114050329,
          "validation_metric": 0.0007494842866435647,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.39825941099843476
        },
        {
          "epoch": 49,
          "updates": 392,
          "samples_seen": 25088,
          "online_loss": 0.0047606244625058025,
          "train_loss": 0.005234242882579565,
          "validation_loss": 0.003474057652056217,
          "train_metric": 0.0006014127284288406,
          "validation_metric": 0.0003991679986938834,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.405193064998457
        },
        {
          "epoch": 50,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.004231025028275326,
          "train_loss": 0.003573098685592413,
          "validation_loss": 0.005039774347096682,
          "train_metric": 0.00041054777102544904,
          "validation_metric": 0.0005790683208033442,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4116150369991374
        },
        {
          "epoch": 51,
          "updates": 408,
          "samples_seen": 26112,
          "online_loss": 0.0036496774264378473,
          "train_loss": 0.00525596272200346,
          "validation_loss": 0.00315229594707489,
          "train_metric": 0.0006039082654751837,
          "validation_metric": 0.0003621977230068296,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41839534700011427
        },
        {
          "epoch": 52,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.003577427225536667,
          "train_loss": 0.002961893565952778,
          "validation_loss": 0.0035922157112509012,
          "train_metric": 0.0003403205191716552,
          "validation_metric": 0.00041274435352534056,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4246236580002005
        },
        {
          "epoch": 53,
          "updates": 424,
          "samples_seen": 27136,
          "online_loss": 0.0034465471107978374,
          "train_loss": 0.004723128397017717,
          "validation_loss": 0.0035725829657167196,
          "train_metric": 0.0005426858551800251,
          "validation_metric": 0.00041048851562663913,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4307220659993618
        },
        {
          "epoch": 54,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.004424020924489014,
          "train_loss": 0.006950423587113619,
          "validation_loss": 0.008883067406713963,
          "train_metric": 0.0007986013079062104,
          "validation_metric": 0.0010206613223999739,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4372092839985271
        },
        {
          "epoch": 55,
          "updates": 440,
          "samples_seen": 28160,
          "online_loss": 0.008784937541349791,
          "train_loss": 0.018442578613758087,
          "validation_loss": 0.009595907293260098,
          "train_metric": 0.002119045704603195,
          "validation_metric": 0.0011025663698092103,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.44357015499917907
        },
        {
          "epoch": 56,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.011634398979367688,
          "train_loss": 0.029351893812417984,
          "validation_loss": 0.018073150888085365,
          "train_metric": 0.0033725223038345575,
          "validation_metric": 0.002076598582789302,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4499228309996397
        },
        {
          "epoch": 57,
          "updates": 456,
          "samples_seen": 29184,
          "online_loss": 0.03055432834662497,
          "train_loss": 0.03055228851735592,
          "validation_loss": 0.02043328434228897,
          "train_metric": 0.003510447684675455,
          "validation_metric": 0.002347777131944895,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.45609399899876735
        },
        {
          "epoch": 58,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.023079946753568947,
          "train_loss": 0.014560774900019169,
          "validation_loss": 0.019172267988324165,
          "train_metric": 0.0016730278730392456,
          "validation_metric": 0.002202887088060379,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4623015619999933
        },
        {
          "epoch": 59,
          "updates": 472,
          "samples_seen": 30208,
          "online_loss": 0.014290992345195264,
          "train_loss": 0.015208314172923565,
          "validation_loss": 0.016823237761855125,
          "train_metric": 0.001747430069372058,
          "validation_metric": 0.0019329842180013657,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.46881898699939484
        },
        {
          "epoch": 60,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.008599012100603431,
          "train_loss": 0.0057950629852712154,
          "validation_loss": 0.00976387970149517,
          "train_metric": 0.0006658507627435029,
          "validation_metric": 0.0011218664003536105,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4749962579990097
        },
        {
          "epoch": 61,
          "updates": 488,
          "samples_seen": 31232,
          "online_loss": 0.006280872621573508,
          "train_loss": 0.005610454827547073,
          "validation_loss": 0.006292941048741341,
          "train_metric": 0.0006446393672376871,
          "validation_metric": 0.0007230567280203104,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4813678489990707
        },
        {
          "epoch": 62,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.005373255931772292,
          "train_loss": 0.004074385389685631,
          "validation_loss": 0.0037280942779034376,
          "train_metric": 0.0004681454156525433,
          "validation_metric": 0.00042835675412788987,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.48758952999924077
        },
        {
          "epoch": 63,
          "updates": 504,
          "samples_seen": 32256,
          "online_loss": 0.0039472369535360485,
          "train_loss": 0.0035181662533432245,
          "validation_loss": 0.006566083990037441,
          "train_metric": 0.00040423611062578857,
          "validation_metric": 0.0007544408435933292,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4937794950001262
        },
        {
          "epoch": 64,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.003684538387460634,
          "train_loss": 0.003128425683826208,
          "validation_loss": 0.003517630510032177,
          "train_metric": 0.0003594550071284175,
          "validation_metric": 0.0004041745269205421,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5002730369997153
        },
        {
          "epoch": 65,
          "updates": 520,
          "samples_seen": 33280,
          "online_loss": 0.0035367781383683905,
          "train_loss": 0.002759945346042514,
          "validation_loss": 0.004741136450320482,
          "train_metric": 0.00031711682095192373,
          "validation_metric": 0.000544755021110177,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5066363219993946
        },
        {
          "epoch": 66,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.005084812066343147,
          "train_loss": 0.0026917962823063135,
          "validation_loss": 0.003149355761706829,
          "train_metric": 0.0003092865226790309,
          "validation_metric": 0.0003618599148467183,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5135322389996873
        },
        {
          "epoch": 67,
          "updates": 536,
          "samples_seen": 34304,
          "online_loss": 0.004863250913331285,
          "train_loss": 0.014343133196234703,
          "validation_loss": 0.008681048639118671,
          "train_metric": 0.001648021163418889,
          "validation_metric": 0.000997449504211545,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5200140149991057
        },
        {
          "epoch": 68,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.008030031764064915,
          "train_loss": 0.008325565606355667,
          "validation_loss": 0.0048237270675599575,
          "train_metric": 0.0009566046646796167,
          "validation_metric": 0.0005542445578612387,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5262682960001257
        },
        {
          "epoch": 69,
          "updates": 552,
          "samples_seen": 35328,
          "online_loss": 0.007316313654882833,
          "train_loss": 0.0028002250473946333,
          "validation_loss": 0.003626888385042548,
          "train_metric": 0.00032174488296732306,
          "validation_metric": 0.0004167282022535801,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.532733220999944
        },
        {
          "epoch": 70,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.0049945510108955204,
          "train_loss": 0.005808508023619652,
          "validation_loss": 0.009432906284928322,
          "train_metric": 0.0006673955358564854,
          "validation_metric": 0.0010838375892490149,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5391596519984887
        },
        {
          "epoch": 71,
          "updates": 568,
          "samples_seen": 36352,
          "online_loss": 0.003949519275920466,
          "train_loss": 0.004071845207363367,
          "validation_loss": 0.006498308386653662,
          "train_metric": 0.0004678536206483841,
          "validation_metric": 0.0007466532406397164,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5453345149999222
        },
        {
          "epoch": 72,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.0038599338731728494,
          "train_loss": 0.0029519156087189913,
          "validation_loss": 0.004874020349234343,
          "train_metric": 0.000339174090186134,
          "validation_metric": 0.000560023239813745,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5515818659987417
        },
        {
          "epoch": 73,
          "updates": 584,
          "samples_seen": 37376,
          "online_loss": 0.003597781207645312,
          "train_loss": 0.0027764819096773863,
          "validation_loss": 0.0035501401871442795,
          "train_metric": 0.0003190167772118002,
          "validation_metric": 0.0004079098580405116,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.557640620998427
        },
        {
          "epoch": 74,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.004310092292143963,
          "train_loss": 0.002634630771353841,
          "validation_loss": 0.0035545022692531347,
          "train_metric": 0.0003027181373909116,
          "validation_metric": 0.0004084110842086375,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5637394499990478
        },
        {
          "epoch": 75,
          "updates": 600,
          "samples_seen": 38400,
          "online_loss": 0.0026406601537019014,
          "train_loss": 0.0038536770734936,
          "validation_loss": 0.006648094393312931,
          "train_metric": 0.00044278617133386433,
          "validation_metric": 0.000763863732572645,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5699792389987124
        },
        {
          "epoch": 76,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.0030431088962359354,
          "train_loss": 0.0021944104228168726,
          "validation_loss": 0.00369885191321373,
          "train_metric": 0.0002521370188333094,
          "validation_metric": 0.00042499680421315134,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5761675559988362
        },
        {
          "epoch": 77,
          "updates": 616,
          "samples_seen": 39424,
          "online_loss": 0.00223932813241845,
          "train_loss": 0.0020697563886642456,
          "validation_loss": 0.004152796231210232,
          "train_metric": 0.0002378142817178741,
          "validation_metric": 0.00047715488472022116,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5823331289993803
        },
        {
          "epoch": 78,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.0023557063104817644,
          "train_loss": 0.0019346317276358604,
          "validation_loss": 0.002946852007880807,
          "train_metric": 0.00022228850866667926,
          "validation_metric": 0.0003385922755114734,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.588968242998817
        },
        {
          "epoch": 79,
          "updates": 632,
          "samples_seen": 40448,
          "online_loss": 0.002449618885293603,
          "train_loss": 0.0024707058910280466,
          "validation_loss": 0.005132647696882486,
          "train_metric": 0.0002838832151610404,
          "validation_metric": 0.0005897395312786102,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.5951786400000856
        },
        {
          "epoch": 80,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.0035082791800959967,
          "train_loss": 0.0023203592281788588,
          "validation_loss": 0.005576199386268854,
          "train_metric": 0.00026660849107429385,
          "validation_metric": 0.0006407033652067184,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6013429039994662
        },
        {
          "epoch": 81,
          "updates": 648,
          "samples_seen": 41472,
          "online_loss": 0.004852604717598297,
          "train_loss": 0.005567458923906088,
          "validation_loss": 0.003356220433488488,
          "train_metric": 0.0006396990502253175,
          "validation_metric": 0.00038562851841561496,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6076368389985873
        },
        {
          "epoch": 82,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.003926236167899333,
          "train_loss": 0.0020537625532597303,
          "validation_loss": 0.0035311749670654535,
          "train_metric": 0.00023597662220709026,
          "validation_metric": 0.00040573079604655504,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6138167979988793
        },
        {
          "epoch": 83,
          "updates": 664,
          "samples_seen": 42496,
          "online_loss": 0.002816940024786163,
          "train_loss": 0.0019625085406005383,
          "validation_loss": 0.004377186764031649,
          "train_metric": 0.00022549157438334078,
          "validation_metric": 0.0005029371823184192,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6199337639991427
        },
        {
          "epoch": 84,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.003014773756149225,
          "train_loss": 0.002841282170265913,
          "validation_loss": 0.0025438701268285513,
          "train_metric": 0.00032646238105371594,
          "validation_metric": 0.0002922897692769766,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6265622269984306
        },
        {
          "epoch": 85,
          "updates": 680,
          "samples_seen": 43520,
          "online_loss": 0.002948745750472881,
          "train_loss": 0.002251275349408388,
          "validation_loss": 0.0027017320971935987,
          "train_metric": 0.00025867079966701567,
          "validation_metric": 0.0003104280913248658,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6332355539998389
        },
        {
          "epoch": 86,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.0028906647639814764,
          "train_loss": 0.002267702016979456,
          "validation_loss": 0.005331465974450111,
          "train_metric": 0.0002605581539683044,
          "validation_metric": 0.0006125835352577269,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6393305629990209
        },
        {
          "epoch": 87,
          "updates": 696,
          "samples_seen": 44544,
          "online_loss": 0.0035766819200944155,
          "train_loss": 0.0022343378514051437,
          "validation_loss": 0.005445471964776516,
          "train_metric": 0.0002567246265243739,
          "validation_metric": 0.0006256828783079982,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6455104849992495
        },
        {
          "epoch": 88,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.004252103550243191,
          "train_loss": 0.0027079705614596605,
          "validation_loss": 0.0027616824954748154,
          "train_metric": 0.0003111449477728456,
          "validation_metric": 0.00031731638591736555,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6516581599989877
        },
        {
          "epoch": 89,
          "updates": 712,
          "samples_seen": 45568,
          "online_loss": 0.0025054748548427597,
          "train_loss": 0.0018370583420619369,
          "validation_loss": 0.0031061177141964436,
          "train_metric": 0.00021107736392877996,
          "validation_metric": 0.00035689183278009295,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6576279979999526
        },
        {
          "epoch": 90,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.002935472075478174,
          "train_loss": 0.0045126271434128284,
          "validation_loss": 0.006443794351071119,
          "train_metric": 0.0005184992915019393,
          "validation_metric": 0.0007403896888718009,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6637029439989419
        },
        {
          "epoch": 91,
          "updates": 728,
          "samples_seen": 46592,
          "online_loss": 0.003100978792645037,
          "train_loss": 0.001767540699802339,
          "validation_loss": 0.0036707911640405655,
          "train_metric": 0.00020308983221184462,
          "validation_metric": 0.0004217726527713239,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6698135629994795
        },
        {
          "epoch": 92,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.002269730786792934,
          "train_loss": 0.0016738187987357378,
          "validation_loss": 0.0027581953909248114,
          "train_metric": 0.00019232119666412473,
          "validation_metric": 0.00031691568437963724,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6759256739987904
        },
        {
          "epoch": 93,
          "updates": 744,
          "samples_seen": 47616,
          "online_loss": 0.0019559365828172304,
          "train_loss": 0.002434520283713937,
          "validation_loss": 0.002739533083513379,
          "train_metric": 0.00027972550014965236,
          "validation_metric": 0.000314771372359246,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6821189089987456
        },
        {
          "epoch": 94,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.002306606445927173,
          "train_loss": 0.005058960989117622,
          "validation_loss": 0.008614689111709595,
          "train_metric": 0.0005812729359604418,
          "validation_metric": 0.000989824766293168,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6880731299988838
        },
        {
          "epoch": 95,
          "updates": 760,
          "samples_seen": 48640,
          "online_loss": 0.003052995823964011,
          "train_loss": 0.0018902214942499995,
          "validation_loss": 0.0030864374712109566,
          "train_metric": 0.00021718577772844583,
          "validation_metric": 0.00035463061067275703,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6940930199998547
        },
        {
          "epoch": 96,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.0017470008970121853,
          "train_loss": 0.0016654421342536807,
          "validation_loss": 0.004412233363837004,
          "train_metric": 0.00019135873299092054,
          "validation_metric": 0.0005069641047157347,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7001725079990138
        },
        {
          "epoch": 97,
          "updates": 776,
          "samples_seen": 49664,
          "online_loss": 0.002061699007754214,
          "train_loss": 0.0015359499957412481,
          "validation_loss": 0.003681780304759741,
          "train_metric": 0.00017648012726567686,
          "validation_metric": 0.00042303523514419794,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7063850109989289
        },
        {
          "epoch": 98,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.001864918798673898,
          "train_loss": 0.0015909664798527956,
          "validation_loss": 0.002969433320686221,
          "train_metric": 0.00018280149379279464,
          "validation_metric": 0.00034118685289286077,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7126069149999239
        },
        {
          "epoch": 99,
          "updates": 792,
          "samples_seen": 50688,
          "online_loss": 0.0015938741453283,
          "train_loss": 0.0016194606432691216,
          "validation_loss": 0.004282553680241108,
          "train_metric": 0.000186075470992364,
          "validation_metric": 0.0004920639912597835,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7188389189996087
        },
        {
          "epoch": 100,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.0020789401241927408,
          "train_loss": 0.0017267869552597404,
          "validation_loss": 0.0027634548023343086,
          "train_metric": 0.00019840721506625414,
          "validation_metric": 0.00031751999631524086,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.724988186999326
        },
        {
          "epoch": 101,
          "updates": 808,
          "samples_seen": 51712,
          "online_loss": 0.0016082122238003649,
          "train_loss": 0.001531347050331533,
          "validation_loss": 0.0038013209123164415,
          "train_metric": 0.00017595123790670186,
          "validation_metric": 0.00043677043868228793,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7311556199983897
        },
        {
          "epoch": 102,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.0015052543312776834,
          "train_loss": 0.0013916699681431055,
          "validation_loss": 0.0031472723931074142,
          "train_metric": 0.00015990238171070814,
          "validation_metric": 0.0003616205067373812,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7372495780000463
        },
        {
          "epoch": 103,
          "updates": 824,
          "samples_seen": 52736,
          "online_loss": 0.0014825258258497342,
          "train_loss": 0.0013610243331640959,
          "validation_loss": 0.0033344714902341366,
          "train_metric": 0.00015638125478290021,
          "validation_metric": 0.0003831296053249389,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7433969779995095
        },
        {
          "epoch": 104,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.001658487002714537,
          "train_loss": 0.00153192353900522,
          "validation_loss": 0.004244194366037846,
          "train_metric": 0.00017601749277673662,
          "validation_metric": 0.0004876564780715853,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.7497308969996084
        }
      ],
      "best_validation_metric": 0.0002922897692769766,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 11,
      "parameters": 2369,
      "train_samples": 512,
      "validation_rmse_m_s": 0.01709648412033821,
      "training_profiles": [
        0,
        4,
        7,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 12,
          "samples_seen": 768,
          "online_loss": 0.36001816391944885,
          "train_loss": 0.14616499841213226,
          "validation_loss": 0.10772781074047089,
          "train_metric": 0.0141739621758461,
          "validation_metric": 0.010446619242429733,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.012839053000789136
        },
        {
          "epoch": 2,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.10526151830951373,
          "train_loss": 0.06448564678430557,
          "validation_loss": 0.0525534488260746,
          "train_metric": 0.006253324914723635,
          "validation_metric": 0.005096231587231159,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.022971910999331158
        },
        {
          "epoch": 3,
          "updates": 36,
          "samples_seen": 2304,
          "online_loss": 0.05363008612766862,
          "train_loss": 0.03873616084456444,
          "validation_loss": 0.03909844905138016,
          "train_metric": 0.003756336634978652,
          "validation_metric": 0.0037914684507995844,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03371739300018817
        },
        {
          "epoch": 4,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.035925685738523803,
          "train_loss": 0.0332755483686924,
          "validation_loss": 0.03017473965883255,
          "train_metric": 0.0032268075738102198,
          "validation_metric": 0.0029261154122650623,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04460076100076549
        },
        {
          "epoch": 5,
          "updates": 60,
          "samples_seen": 3840,
          "online_loss": 0.03103002921367685,
          "train_loss": 0.03872064873576164,
          "validation_loss": 0.03829406946897507,
          "train_metric": 0.003754831850528717,
          "validation_metric": 0.0037134659942239523,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.054750458999478724
        },
        {
          "epoch": 6,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.045119710344200335,
          "train_loss": 0.03137197718024254,
          "validation_loss": 0.025607585906982422,
          "train_metric": 0.0030422143172472715,
          "validation_metric": 0.002483227988705039,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06553443899974809
        },
        {
          "epoch": 7,
          "updates": 84,
          "samples_seen": 5376,
          "online_loss": 0.02456587568546335,
          "train_loss": 0.02545652724802494,
          "validation_loss": 0.023000773042440414,
          "train_metric": 0.002468579448759556,
          "validation_metric": 0.0022304391022771597,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07698435600104858
        },
        {
          "epoch": 8,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.027729810991634924,
          "train_loss": 0.022236905992031097,
          "validation_loss": 0.023192159831523895,
          "train_metric": 0.0021563649643212557,
          "validation_metric": 0.002248998498544097,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08668783699977212
        },
        {
          "epoch": 9,
          "updates": 108,
          "samples_seen": 6912,
          "online_loss": 0.02025076715896527,
          "train_loss": 0.018535012379288673,
          "validation_loss": 0.02099686674773693,
          "train_metric": 0.0017973833018913865,
          "validation_metric": 0.0020361156202852726,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09700683300070523
        },
        {
          "epoch": 10,
          "updates": 120,
          "samples_seen": 7680,
          "online_loss": 0.016009064003204305,
          "train_loss": 0.01469621155411005,
          "validation_loss": 0.012097318656742573,
          "train_metric": 0.0014251264510676265,
          "validation_metric": 0.0011731055565178394,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1079768409999815
        },
        {
          "epoch": 11,
          "updates": 132,
          "samples_seen": 8448,
          "online_loss": 0.013676253496669233,
          "train_loss": 0.014117958955466747,
          "validation_loss": 0.011958019807934761,
          "train_metric": 0.0013690516352653503,
          "validation_metric": 0.0011595971882343292,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.18768086300042341
        },
        {
          "epoch": 12,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.013550634573524198,
          "train_loss": 0.009796878322958946,
          "validation_loss": 0.008441264741122723,
          "train_metric": 0.0009500263840891421,
          "validation_metric": 0.0008185693877749145,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.19881286100098805
        },
        {
          "epoch": 13,
          "updates": 156,
          "samples_seen": 9984,
          "online_loss": 0.01822738240783413,
          "train_loss": 0.019743459299206734,
          "validation_loss": 0.025192083790898323,
          "train_metric": 0.0019145695259794593,
          "validation_metric": 0.0024429357144981623,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20899766399998043
        },
        {
          "epoch": 14,
          "updates": 168,
          "samples_seen": 10752,
          "online_loss": 0.013955099508166313,
          "train_loss": 0.008341564796864986,
          "validation_loss": 0.008913151919841766,
          "train_metric": 0.0008089011535048485,
          "validation_metric": 0.0008643293404020369,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21860088799985533
        },
        {
          "epoch": 15,
          "updates": 180,
          "samples_seen": 11520,
          "online_loss": 0.012566304513408491,
          "train_loss": 0.017348241060972214,
          "validation_loss": 0.014685286208987236,
          "train_metric": 0.00168229965493083,
          "validation_metric": 0.0014240667223930359,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22842678900087776
        },
        {
          "epoch": 16,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.02482462431847428,
          "train_loss": 0.016594216227531433,
          "validation_loss": 0.018128234893083572,
          "train_metric": 0.0016091801226139069,
          "validation_metric": 0.0017579374834895134,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.23815173200091522
        },
        {
          "epoch": 17,
          "updates": 204,
          "samples_seen": 13056,
          "online_loss": 0.015107512978526453,
          "train_loss": 0.00944579392671585,
          "validation_loss": 0.0073835705406963825,
          "train_metric": 0.0009159808396361768,
          "validation_metric": 0.0007160023087635636,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2489191580007173
        },
        {
          "epoch": 18,
          "updates": 216,
          "samples_seen": 13824,
          "online_loss": 0.015132936396791289,
          "train_loss": 0.009491121396422386,
          "validation_loss": 0.009548623114824295,
          "train_metric": 0.0009203764493577182,
          "validation_metric": 0.0009259524522349238,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25817279200055054
        },
        {
          "epoch": 19,
          "updates": 228,
          "samples_seen": 14592,
          "online_loss": 0.01091807750829806,
          "train_loss": 0.007873625494539738,
          "validation_loss": 0.006839514244347811,
          "train_metric": 0.0007635238580405712,
          "validation_metric": 0.0006632438162341714,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2683467879996897
        },
        {
          "epoch": 20,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.015766989323310554,
          "train_loss": 0.00969535019248724,
          "validation_loss": 0.009057085029780865,
          "train_metric": 0.0009401810239069164,
          "validation_metric": 0.0008782868972048163,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27818790300079854
        },
        {
          "epoch": 21,
          "updates": 252,
          "samples_seen": 16128,
          "online_loss": 0.010086959809996188,
          "train_loss": 0.011243380606174469,
          "validation_loss": 0.012268530204892159,
          "train_metric": 0.001090297126211226,
          "validation_metric": 0.0011897081276401877,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.28772376400047506
        },
        {
          "epoch": 22,
          "updates": 264,
          "samples_seen": 16896,
          "online_loss": 0.012835896360532692,
          "train_loss": 0.013347680680453777,
          "validation_loss": 0.013009093701839447,
          "train_metric": 0.0012943560723215342,
          "validation_metric": 0.0012615224113687873,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2971021949997521
        },
        {
          "epoch": 23,
          "updates": 276,
          "samples_seen": 17664,
          "online_loss": 0.011469504291502139,
          "train_loss": 0.010631428100168705,
          "validation_loss": 0.013344145379960537,
          "train_metric": 0.0010309546487405896,
          "validation_metric": 0.001294013112783432,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3065368469997338
        },
        {
          "epoch": 24,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.008787387690972537,
          "train_loss": 0.006246078759431839,
          "validation_loss": 0.006140440236777067,
          "train_metric": 0.0006056969286873937,
          "validation_metric": 0.0005954529624432325,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3167592080008035
        },
        {
          "epoch": 25,
          "updates": 300,
          "samples_seen": 19200,
          "online_loss": 0.0069632845697924495,
          "train_loss": 0.005772631149739027,
          "validation_loss": 0.004762131720781326,
          "train_metric": 0.0005597855779342353,
          "validation_metric": 0.0004617951053660363,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.32680960000107007
        },
        {
          "epoch": 26,
          "updates": 312,
          "samples_seen": 19968,
          "online_loss": 0.013450541805165509,
          "train_loss": 0.020967215299606323,
          "validation_loss": 0.023937836289405823,
          "train_metric": 0.0020332399290055037,
          "validation_metric": 0.0023213080130517483,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3360323970009631
        },
        {
          "epoch": 27,
          "updates": 324,
          "samples_seen": 20736,
          "online_loss": 0.009748808476918688,
          "train_loss": 0.007887272164225578,
          "validation_loss": 0.006984096951782703,
          "train_metric": 0.0007648475002497435,
          "validation_metric": 0.0006772642373107374,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3469446319995768
        },
        {
          "epoch": 28,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.006296804155378292,
          "train_loss": 0.005476027261465788,
          "validation_loss": 0.005459879525005817,
          "train_metric": 0.0005310233100317419,
          "validation_metric": 0.0005294573493301868,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3560702510003466
        },
        {
          "epoch": 29,
          "updates": 348,
          "samples_seen": 22272,
          "online_loss": 0.00540052274785315,
          "train_loss": 0.006997639779001474,
          "validation_loss": 0.004806365817785263,
          "train_metric": 0.0006785776349715889,
          "validation_metric": 0.0004660846316255629,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.36559514200052945
        },
        {
          "epoch": 30,
          "updates": 360,
          "samples_seen": 23040,
          "online_loss": 0.008590366613740722,
          "train_loss": 0.020363297313451767,
          "validation_loss": 0.02030143328011036,
          "train_metric": 0.0019746769685298204,
          "validation_metric": 0.001968677621334791,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37446543099940754
        },
        {
          "epoch": 31,
          "updates": 372,
          "samples_seen": 23808,
          "online_loss": 0.014410206892838081,
          "train_loss": 0.006264426279813051,
          "validation_loss": 0.005423746071755886,
          "train_metric": 0.0006074761622585356,
          "validation_metric": 0.0005259534227661788,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3844755570007692
        },
        {
          "epoch": 32,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.008409995876718313,
          "train_loss": 0.017086081206798553,
          "validation_loss": 0.020528778433799744,
          "train_metric": 0.0016568773426115513,
          "validation_metric": 0.0019907241221517324,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3937802990003547
        },
        {
          "epoch": 33,
          "updates": 396,
          "samples_seen": 25344,
          "online_loss": 0.015825824074757595,
          "train_loss": 0.028833605349063873,
          "validation_loss": 0.033277757465839386,
          "train_metric": 0.0027960622683167458,
          "validation_metric": 0.0032270224764943123,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.40293548599947826
        },
        {
          "epoch": 34,
          "updates": 408,
          "samples_seen": 26112,
          "online_loss": 0.023287200446551044,
          "train_loss": 0.015789106488227844,
          "validation_loss": 0.011928249150514603,
          "train_metric": 0.0015311067691072822,
          "validation_metric": 0.001156710204668343,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41268694599966693
        },
        {
          "epoch": 35,
          "updates": 420,
          "samples_seen": 26880,
          "online_loss": 0.014149945850173632,
          "train_loss": 0.014434496872127056,
          "validation_loss": 0.013024196960031986,
          "train_metric": 0.0013997472124174237,
          "validation_metric": 0.0012629871489480138,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.42175818599935155
        },
        {
          "epoch": 36,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.011120203960066041,
          "train_loss": 0.007562019396573305,
          "validation_loss": 0.007736825849860907,
          "train_metric": 0.0007333068060688674,
          "validation_metric": 0.0007502582157030702,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.43117044199971133
        },
        {
          "epoch": 37,
          "updates": 444,
          "samples_seen": 28416,
          "online_loss": 0.006600356990626703,
          "train_loss": 0.005882056429982185,
          "validation_loss": 0.00550543749704957,
          "train_metric": 0.0005703968345187604,
          "validation_metric": 0.0005338751943781972,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4405096610007604
        },
        {
          "epoch": 38,
          "updates": 456,
          "samples_seen": 29184,
          "online_loss": 0.006956126793132474,
          "train_loss": 0.010768398642539978,
          "validation_loss": 0.01038109790533781,
          "train_metric": 0.0010442370548844337,
          "validation_metric": 0.0010066793765872717,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4497445119995973
        },
        {
          "epoch": 39,
          "updates": 468,
          "samples_seen": 29952,
          "online_loss": 0.00932320910699976,
          "train_loss": 0.007937480695545673,
          "validation_loss": 0.007455973420292139,
          "train_metric": 0.0007697162218391895,
          "validation_metric": 0.0007230232586152852,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.45854533900092065
        },
        {
          "epoch": 40,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.005154521223933746,
          "train_loss": 0.00537374010309577,
          "validation_loss": 0.004768969491124153,
          "train_metric": 0.0005211041425354779,
          "validation_metric": 0.00046245817793533206,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4678851360004046
        },
        {
          "epoch": 41,
          "updates": 492,
          "samples_seen": 31488,
          "online_loss": 0.005257466216183578,
          "train_loss": 0.004224447067826986,
          "validation_loss": 0.004069863818585873,
          "train_metric": 0.0004096544871572405,
          "validation_metric": 0.00039466426824219525,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.47806812499948137
        },
        {
          "epoch": 42,
          "updates": 504,
          "samples_seen": 32256,
          "online_loss": 0.010605045700989043,
          "train_loss": 0.01451094076037407,
          "validation_loss": 0.012075112201273441,
          "train_metric": 0.0014071600744500756,
          "validation_metric": 0.001170951989479363,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.48694341800000984
        },
        {
          "epoch": 43,
          "updates": 516,
          "samples_seen": 33024,
          "online_loss": 0.01275613585797449,
          "train_loss": 0.015462311916053295,
          "validation_loss": 0.013425647281110287,
          "train_metric": 0.001499416772276163,
          "validation_metric": 0.0013019165489822626,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.49615804399945773
        },
        {
          "epoch": 44,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.00978561797334502,
          "train_loss": 0.007318509742617607,
          "validation_loss": 0.006099899299442768,
          "train_metric": 0.000709693122189492,
          "validation_metric": 0.000591521558817476,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5053400779997901
        },
        {
          "epoch": 45,
          "updates": 540,
          "samples_seen": 34560,
          "online_loss": 0.005767857073806226,
          "train_loss": 0.004456784576177597,
          "validation_loss": 0.004436961840838194,
          "train_metric": 0.0004321848100516945,
          "validation_metric": 0.00043026264756917953,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5146523010007513
        },
        {
          "epoch": 46,
          "updates": 552,
          "samples_seen": 35328,
          "online_loss": 0.004681354466204842,
          "train_loss": 0.004746620077639818,
          "validation_loss": 0.005901013035327196,
          "train_metric": 0.00046029093209654093,
          "validation_metric": 0.0005722350906580687,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5234647110009973
        },
        {
          "epoch": 47,
          "updates": 564,
          "samples_seen": 36096,
          "online_loss": 0.004560082141930859,
          "train_loss": 0.004610855598002672,
          "validation_loss": 0.005193167366087437,
          "train_metric": 0.00044712552335113287,
          "validation_metric": 0.0005035936483182013,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5331548129997827
        },
        {
          "epoch": 48,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.0042814031864206,
          "train_loss": 0.003304655896499753,
          "validation_loss": 0.0032732998952269554,
          "train_metric": 0.00032046023989096284,
          "validation_metric": 0.0003174196172039956,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.543831389000843
        },
        {
          "epoch": 49,
          "updates": 588,
          "samples_seen": 37632,
          "online_loss": 0.003387956152437255,
          "train_loss": 0.002921576611697674,
          "validation_loss": 0.003291332395747304,
          "train_metric": 0.00028331216890364885,
          "validation_metric": 0.00031916823354549706,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5533605100008572
        },
        {
          "epoch": 50,
          "updates": 600,
          "samples_seen": 38400,
          "online_loss": 0.0029919377799766758,
          "train_loss": 0.004141738638281822,
          "validation_loss": 0.006180264987051487,
          "train_metric": 0.00040163411176763475,
          "validation_metric": 0.0005993148079141974,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5624530950008193
        },
        {
          "epoch": 51,
          "updates": 612,
          "samples_seen": 39168,
          "online_loss": 0.003602073416307879,
          "train_loss": 0.0026182124856859446,
          "validation_loss": 0.003092031227424741,
          "train_metric": 0.0002538942208047956,
          "validation_metric": 0.000299841514788568,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5722941799995169
        },
        {
          "epoch": 52,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.0032110883427473405,
          "train_loss": 0.009683830663561821,
          "validation_loss": 0.01239205151796341,
          "train_metric": 0.0009390639024786651,
          "validation_metric": 0.0012016864493489265,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5817516950010031
        },
        {
          "epoch": 53,
          "updates": 636,
          "samples_seen": 40704,
          "online_loss": 0.004659959221802031,
          "train_loss": 0.002704775892198086,
          "validation_loss": 0.0036525491159409285,
          "train_metric": 0.0002622884640004486,
          "validation_metric": 0.00035419632331468165,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5903577740009496
        },
        {
          "epoch": 54,
          "updates": 648,
          "samples_seen": 41472,
          "online_loss": 0.0055504945339635015,
          "train_loss": 0.00543349189683795,
          "validation_loss": 0.004864016082137823,
          "train_metric": 0.0005268985405564308,
          "validation_metric": 0.00047167512821033597,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6010791560001962
        },
        {
          "epoch": 55,
          "updates": 660,
          "samples_seen": 42240,
          "online_loss": 0.0053878054895903915,
          "train_loss": 0.0033380845561623573,
          "validation_loss": 0.003889328334480524,
          "train_metric": 0.0003237019118387252,
          "validation_metric": 0.00037715735379606485,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6117431340007897
        },
        {
          "epoch": 56,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.005892007291549817,
          "train_loss": 0.0026670435909181833,
          "validation_loss": 0.002293698024004698,
          "train_metric": 0.00025862950133159757,
          "validation_metric": 0.00022242532577365637,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6252927879995696
        },
        {
          "epoch": 57,
          "updates": 684,
          "samples_seen": 43776,
          "online_loss": 0.005589792543711762,
          "train_loss": 0.0038044799584895372,
          "validation_loss": 0.0035325000062584877,
          "train_metric": 0.0003689293807838112,
          "validation_metric": 0.0003425548493396491,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6367920640004741
        },
        {
          "epoch": 58,
          "updates": 696,
          "samples_seen": 44544,
          "online_loss": 0.004441961752794062,
          "train_loss": 0.007416464854031801,
          "validation_loss": 0.010594288818538189,
          "train_metric": 0.0007191920303739607,
          "validation_metric": 0.0010273531079292297,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6475436479995551
        },
        {
          "epoch": 59,
          "updates": 708,
          "samples_seen": 45312,
          "online_loss": 0.009632765655017769,
          "train_loss": 0.004489933606237173,
          "validation_loss": 0.005037371069192886,
          "train_metric": 0.00043539947364479303,
          "validation_metric": 0.0004884856753051281,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6599026719995891
        },
        {
          "epoch": 60,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.006528164958581328,
          "train_loss": 0.012993056327104568,
          "validation_loss": 0.011633559130132198,
          "train_metric": 0.0012599671026691794,
          "validation_metric": 0.0011281338520348072,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.673489664999579
        },
        {
          "epoch": 61,
          "updates": 732,
          "samples_seen": 46848,
          "online_loss": 0.005781339238941048,
          "train_loss": 0.006288773845881224,
          "validation_loss": 0.005773259326815605,
          "train_metric": 0.0006098371814005077,
          "validation_metric": 0.0005598465795628726,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6846945460001734
        },
        {
          "epoch": 62,
          "updates": 744,
          "samples_seen": 47616,
          "online_loss": 0.005249933252343908,
          "train_loss": 0.005894376430660486,
          "validation_loss": 0.007198994047939777,
          "train_metric": 0.0005715915467590094,
          "validation_metric": 0.0006981033948250115,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6960327640008472
        },
        {
          "epoch": 63,
          "updates": 756,
          "samples_seen": 48384,
          "online_loss": 0.006563239769699673,
          "train_loss": 0.004540777299553156,
          "validation_loss": 0.004231176804751158,
          "train_metric": 0.000440329808043316,
          "validation_metric": 0.0004103071114514023,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7077284469996812
        },
        {
          "epoch": 64,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.004758686904096976,
          "train_loss": 0.0027177641168236732,
          "validation_loss": 0.0029213011730462313,
          "train_metric": 0.00026354799047112465,
          "validation_metric": 0.0002832854224834591,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7192847959995561
        },
        {
          "epoch": 65,
          "updates": 780,
          "samples_seen": 49920,
          "online_loss": 0.002797718387834417,
          "train_loss": 0.003627736819908023,
          "validation_loss": 0.00361327501013875,
          "train_metric": 0.00035179019323550165,
          "validation_metric": 0.0003503877960611135,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.730626017000759
        },
        {
          "epoch": 66,
          "updates": 792,
          "samples_seen": 50688,
          "online_loss": 0.0034438446552182236,
          "train_loss": 0.0027970701921731234,
          "validation_loss": 0.0022933906875550747,
          "train_metric": 0.00027123847394250333,
          "validation_metric": 0.00022239550889935344,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7433619479998015
        },
        {
          "epoch": 67,
          "updates": 804,
          "samples_seen": 51456,
          "online_loss": 0.002430729626212269,
          "train_loss": 0.0019524655072018504,
          "validation_loss": 0.0028174743056297302,
          "train_metric": 0.00018933515821117908,
          "validation_metric": 0.0002732170978561044,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7552710969994223
        },
        {
          "epoch": 68,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.0025255550378157445,
          "train_loss": 0.0028323724400252104,
          "validation_loss": 0.005043208133429289,
          "train_metric": 0.0002746617828961462,
          "validation_metric": 0.0004890518030151725,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7669383180000295
        },
        {
          "epoch": 69,
          "updates": 828,
          "samples_seen": 52992,
          "online_loss": 0.003990700638193327,
          "train_loss": 0.004789565224200487,
          "validation_loss": 0.004160535521805286,
          "train_metric": 0.0004644554283004254,
          "validation_metric": 0.00040345691377297044,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7781795780010725
        },
        {
          "epoch": 70,
          "updates": 840,
          "samples_seen": 53760,
          "online_loss": 0.008445646220934577,
          "train_loss": 0.007298097014427185,
          "validation_loss": 0.009759149514138699,
          "train_metric": 0.0007077135960571468,
          "validation_metric": 0.0009463677415624261,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7887118889993872
        },
        {
          "epoch": 71,
          "updates": 852,
          "samples_seen": 54528,
          "online_loss": 0.009709758994479975,
          "train_loss": 0.00499199191108346,
          "validation_loss": 0.0070982021279633045,
          "train_metric": 0.00048408517614006996,
          "validation_metric": 0.0006883293390274048,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.800442371999452
        },
        {
          "epoch": 72,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.006771543441573158,
          "train_loss": 0.004062929656356573,
          "validation_loss": 0.005490388721227646,
          "train_metric": 0.0003939918242394924,
          "validation_metric": 0.0005324159283190966,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.811696247999862
        },
        {
          "epoch": 73,
          "updates": 876,
          "samples_seen": 56064,
          "online_loss": 0.003673273129000639,
          "train_loss": 0.008434396237134933,
          "validation_loss": 0.008290943689644337,
          "train_metric": 0.0008179032593034208,
          "validation_metric": 0.000803992326837033,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8222207769995293
        },
        {
          "epoch": 74,
          "updates": 888,
          "samples_seen": 56832,
          "online_loss": 0.005156466814999779,
          "train_loss": 0.0032787572126835585,
          "validation_loss": 0.0038037195336073637,
          "train_metric": 0.0003179487830493599,
          "validation_metric": 0.00036885563167743385,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8333913910009869
        },
        {
          "epoch": 75,
          "updates": 900,
          "samples_seen": 57600,
          "online_loss": 0.003437641920754686,
          "train_loss": 0.002124935854226351,
          "validation_loss": 0.0026013576425611973,
          "train_metric": 0.00020606002362910658,
          "validation_metric": 0.0002522597787901759,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.844658616999368
        },
        {
          "epoch": 76,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.002456110572287192,
          "train_loss": 0.00215168041177094,
          "validation_loss": 0.002359754405915737,
          "train_metric": 0.00020865352416876704,
          "validation_metric": 0.00022883093333803117,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.855297824000445
        },
        {
          "epoch": 77,
          "updates": 924,
          "samples_seen": 59136,
          "online_loss": 0.0024540410328578823,
          "train_loss": 0.0036898453254252672,
          "validation_loss": 0.0029742266051471233,
          "train_metric": 0.00035781299811787903,
          "validation_metric": 0.00028841779567301273,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.86674991000109
        },
        {
          "epoch": 78,
          "updates": 936,
          "samples_seen": 59904,
          "online_loss": 0.002443255618951904,
          "train_loss": 0.0018921002047136426,
          "validation_loss": 0.0018696602201089263,
          "train_metric": 0.00018348141747992486,
          "validation_metric": 0.0001813053386285901,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8792528249996394
        },
        {
          "epoch": 79,
          "updates": 948,
          "samples_seen": 60672,
          "online_loss": 0.0021463779363936433,
          "train_loss": 0.0016621713293716311,
          "validation_loss": 0.0026148478500545025,
          "train_metric": 0.0001611846237210557,
          "validation_metric": 0.0002535679377615452,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8908654379993095
        },
        {
          "epoch": 80,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.001712515571853146,
          "train_loss": 0.0016540484502911568,
          "validation_loss": 0.0015960750170052052,
          "train_metric": 0.00016039697220548987,
          "validation_metric": 0.00015477514534723014,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9033268250004767
        },
        {
          "epoch": 81,
          "updates": 972,
          "samples_seen": 62208,
          "online_loss": 0.0017457167171717931,
          "train_loss": 0.0014172778464853764,
          "validation_loss": 0.0018994284328073263,
          "train_metric": 0.00013743675663135946,
          "validation_metric": 0.00018419206026010215,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9152465919996757
        },
        {
          "epoch": 82,
          "updates": 984,
          "samples_seen": 62976,
          "online_loss": 0.001605705149510565,
          "train_loss": 0.0013655894435942173,
          "validation_loss": 0.002112779999151826,
          "train_metric": 0.00013242442219052464,
          "validation_metric": 0.00020488124573603272,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9263042639995547
        },
        {
          "epoch": 83,
          "updates": 996,
          "samples_seen": 63744,
          "online_loss": 0.00148640233722593,
          "train_loss": 0.0012803287245333195,
          "validation_loss": 0.002058779587969184,
          "train_metric": 0.0001241564896190539,
          "validation_metric": 0.00019964472448918968,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9376902559997689
        },
        {
          "epoch": 84,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.0017294882563874125,
          "train_loss": 0.0018386529991403222,
          "validation_loss": 0.003126174211502075,
          "train_metric": 0.00017829850548878312,
          "validation_metric": 0.0003031524538528174,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9495659340009297
        },
        {
          "epoch": 85,
          "updates": 1020,
          "samples_seen": 65280,
          "online_loss": 0.0015964844254388784,
          "train_loss": 0.0014719831524416804,
          "validation_loss": 0.0027605786453932524,
          "train_metric": 0.00014274167187977582,
          "validation_metric": 0.00026769982650876045,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9610915089997434
        },
        {
          "epoch": 86,
          "updates": 1032,
          "samples_seen": 66048,
          "online_loss": 0.001551562309032306,
          "train_loss": 0.001721731387078762,
          "validation_loss": 0.0017441929085180163,
          "train_metric": 0.0001669603370828554,
          "validation_metric": 0.00016913848230615258,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9718574700000318
        },
        {
          "epoch": 87,
          "updates": 1044,
          "samples_seen": 66816,
          "online_loss": 0.001780737914183798,
          "train_loss": 0.0019086068496108055,
          "validation_loss": 0.0035034583415836096,
          "train_metric": 0.0001850820699473843,
          "validation_metric": 0.0003397386462893337,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9832299720001174
        },
        {
          "epoch": 88,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.0019031500317699586,
          "train_loss": 0.002550264587625861,
          "validation_loss": 0.003953557461500168,
          "train_metric": 0.000247305171797052,
          "validation_metric": 0.00038338577724061906,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9951393379997171
        },
        {
          "epoch": 89,
          "updates": 1068,
          "samples_seen": 68352,
          "online_loss": 0.0018450517382007092,
          "train_loss": 0.004250623751431704,
          "validation_loss": 0.006414574105292559,
          "train_metric": 0.00041219303966499865,
          "validation_metric": 0.0006220363429747522,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0058133319998888
        },
        {
          "epoch": 90,
          "updates": 1080,
          "samples_seen": 69120,
          "online_loss": 0.003344223543535918,
          "train_loss": 0.0019180349772796035,
          "validation_loss": 0.0032470175065100193,
          "train_metric": 0.00018599635222926736,
          "validation_metric": 0.0003148708783555776,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0171791390002909
        },
        {
          "epoch": 91,
          "updates": 1092,
          "samples_seen": 69888,
          "online_loss": 0.001598705295085286,
          "train_loss": 0.001276237890124321,
          "validation_loss": 0.0016742832958698273,
          "train_metric": 0.0001237597898580134,
          "validation_metric": 0.00016235918155871332,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0285353649996978
        },
        {
          "epoch": 92,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.0013737176729288574,
          "train_loss": 0.0012430668575689197,
          "validation_loss": 0.002373984083533287,
          "train_metric": 0.00012054311810061336,
          "validation_metric": 0.00023021084780339152,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0393902329997218
        },
        {
          "epoch": 93,
          "updates": 1116,
          "samples_seen": 71424,
          "online_loss": 0.0013181955146137625,
          "train_loss": 0.001088335644453764,
          "validation_loss": 0.0017233631806448102,
          "train_metric": 0.00010553846368566155,
          "validation_metric": 0.00016711857460904866,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0508506290007062
        },
        {
          "epoch": 94,
          "updates": 1128,
          "samples_seen": 72192,
          "online_loss": 0.0011298864798542734,
          "train_loss": 0.0013387900544330478,
          "validation_loss": 0.0024669941049069166,
          "train_metric": 0.00012982562475372106,
          "validation_metric": 0.00023923021217342466,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0621602480005095
        },
        {
          "epoch": 95,
          "updates": 1140,
          "samples_seen": 72960,
          "online_loss": 0.0013312733887384336,
          "train_loss": 0.001190770766697824,
          "validation_loss": 0.002231081947684288,
          "train_metric": 0.00011547183385118842,
          "validation_metric": 0.00021635329176206142,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0732054090003658
        },
        {
          "epoch": 96,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0012960533689086635,
          "train_loss": 0.0015609449474141002,
          "validation_loss": 0.0015079067088663578,
          "train_metric": 0.000151368512888439,
          "validation_metric": 0.0001462252694182098,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0848561439997866
        },
        {
          "epoch": 97,
          "updates": 1164,
          "samples_seen": 74496,
          "online_loss": 0.0012363122320190694,
          "train_loss": 0.0011186940828338265,
          "validation_loss": 0.00198384583927691,
          "train_metric": 0.0001084823816199787,
          "validation_metric": 0.00019237821106798947,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0967609340004856
        },
        {
          "epoch": 98,
          "updates": 1176,
          "samples_seen": 75264,
          "online_loss": 0.001350310369161889,
          "train_loss": 0.0011610575020313263,
          "validation_loss": 0.002020321786403656,
          "train_metric": 0.00011259048915235326,
          "validation_metric": 0.0001959153451025486,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1082569749996765
        },
        {
          "epoch": 99,
          "updates": 1188,
          "samples_seen": 76032,
          "online_loss": 0.0011956856809168432,
          "train_loss": 0.001372601487673819,
          "validation_loss": 0.002771687461063266,
          "train_metric": 0.00013310440408531576,
          "validation_metric": 0.000268777075689286,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1200411209993035
        },
        {
          "epoch": 100,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.001279629228520207,
          "train_loss": 0.0010911704739555717,
          "validation_loss": 0.001947138924151659,
          "train_metric": 0.00010581337846815586,
          "validation_metric": 0.0001888186379801482,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1317645649996848
        },
        {
          "epoch": 101,
          "updates": 1212,
          "samples_seen": 77568,
          "online_loss": 0.0011709248234789509,
          "train_loss": 0.001071164500899613,
          "validation_loss": 0.002036710735410452,
          "train_metric": 0.00010387334623374045,
          "validation_metric": 0.0001975046470761299,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1433614600009605
        },
        {
          "epoch": 102,
          "updates": 1224,
          "samples_seen": 78336,
          "online_loss": 0.0012743909392156638,
          "train_loss": 0.0009614255395717919,
          "validation_loss": 0.0018018918344751,
          "train_metric": 9.323168342234567e-05,
          "validation_metric": 0.00017473370826337487,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1544745450009941
        },
        {
          "epoch": 103,
          "updates": 1236,
          "samples_seen": 79104,
          "online_loss": 0.0011685731345399593,
          "train_loss": 0.0009579839534126222,
          "validation_loss": 0.0018366186413913965,
          "train_metric": 9.289797162637115e-05,
          "validation_metric": 0.00017810123972594738,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1662325849993067
        },
        {
          "epoch": 104,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.0012055577972205356,
          "train_loss": 0.0011498035164549947,
          "validation_loss": 0.0014122490538284183,
          "train_metric": 0.00011149916099384427,
          "validation_metric": 0.00013694909284822643,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1783543340006872
        },
        {
          "epoch": 105,
          "updates": 1260,
          "samples_seen": 80640,
          "online_loss": 0.00126185478681388,
          "train_loss": 0.0011268415255472064,
          "validation_loss": 0.0013920936034992337,
          "train_metric": 0.00010927247785730287,
          "validation_metric": 0.0001349945814581588,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1900319239994133
        },
        {
          "epoch": 106,
          "updates": 1272,
          "samples_seen": 81408,
          "online_loss": 0.001103895408353613,
          "train_loss": 0.0009826868772506714,
          "validation_loss": 0.001537040458060801,
          "train_metric": 9.529344970360398e-05,
          "validation_metric": 0.00014905042189639062,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2017631439994148
        },
        {
          "epoch": 107,
          "updates": 1284,
          "samples_seen": 82176,
          "online_loss": 0.001126975873679233,
          "train_loss": 0.001554375165142119,
          "validation_loss": 0.0014091478660702705,
          "train_metric": 0.00015073141548782587,
          "validation_metric": 0.00013664837752003223,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2133952730000601
        },
        {
          "epoch": 108,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.0020425095572136343,
          "train_loss": 0.004377005156129599,
          "validation_loss": 0.0034095318987965584,
          "train_metric": 0.00042444848804734647,
          "validation_metric": 0.00033063028240576386,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2249695860009524
        },
        {
          "epoch": 109,
          "updates": 1308,
          "samples_seen": 83712,
          "online_loss": 0.0026573062326254635,
          "train_loss": 0.00121176871471107,
          "validation_loss": 0.0017550258198753,
          "train_metric": 0.00011750807607313618,
          "validation_metric": 0.00017018895596265793,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2365964670007088
        },
        {
          "epoch": 110,
          "updates": 1320,
          "samples_seen": 84480,
          "online_loss": 0.0012176332044570397,
          "train_loss": 0.0009443266317248344,
          "validation_loss": 0.0018301644595339894,
          "train_metric": 9.157357999356464e-05,
          "validation_metric": 0.0001774753473000601,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2471030890010297
        },
        {
          "epoch": 111,
          "updates": 1332,
          "samples_seen": 85248,
          "online_loss": 0.0013042358865883823,
          "train_loss": 0.0009540454484522343,
          "validation_loss": 0.0016542848898097873,
          "train_metric": 9.251603478332981e-05,
          "validation_metric": 0.00016041989147197455,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.257201525999335
        },
        {
          "epoch": 112,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.0010479378834133968,
          "train_loss": 0.000875449797604233,
          "validation_loss": 0.001762478263117373,
          "train_metric": 8.489442552672699e-05,
          "validation_metric": 0.00017091167683247477,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2674652139994578
        },
        {
          "epoch": 113,
          "updates": 1356,
          "samples_seen": 86784,
          "online_loss": 0.0010813559420057572,
          "train_loss": 0.0008531611529178917,
          "validation_loss": 0.0016177588840946555,
          "train_metric": 8.273304410977289e-05,
          "validation_metric": 0.00015687785344198346,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2781634780003515
        },
        {
          "epoch": 114,
          "updates": 1368,
          "samples_seen": 87552,
          "online_loss": 0.0009125209714208419,
          "train_loss": 0.0010555543703958392,
          "validation_loss": 0.0012743850238621235,
          "train_metric": 0.0001023595905280672,
          "validation_metric": 0.00012358011736068875,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.288608208000369
        },
        {
          "epoch": 115,
          "updates": 1380,
          "samples_seen": 88320,
          "online_loss": 0.0009562112342488641,
          "train_loss": 0.0008818550850264728,
          "validation_loss": 0.0019652587361633778,
          "train_metric": 8.551557402824983e-05,
          "validation_metric": 0.00019057576719205827,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2983758520003903
        },
        {
          "epoch": 116,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0011956367137220998,
          "train_loss": 0.0010070550488308072,
          "validation_loss": 0.002151855267584324,
          "train_metric": 9.765650611370802e-05,
          "validation_metric": 0.00020867046259809285,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3076475329999084
        },
        {
          "epoch": 117,
          "updates": 1404,
          "samples_seen": 89856,
          "online_loss": 0.0009103348226441691,
          "train_loss": 0.0011277225567027926,
          "validation_loss": 0.002099621808156371,
          "train_metric": 0.000109357904875651,
          "validation_metric": 0.0002036052756011486,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3174484560004203
        },
        {
          "epoch": 118,
          "updates": 1416,
          "samples_seen": 90624,
          "online_loss": 0.0012877163583956037,
          "train_loss": 0.000914131524041295,
          "validation_loss": 0.0015008057234808803,
          "train_metric": 8.864547271514311e-05,
          "validation_metric": 0.00014553667278960347,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3270013439996546
        },
        {
          "epoch": 119,
          "updates": 1428,
          "samples_seen": 91392,
          "online_loss": 0.0014986447034364876,
          "train_loss": 0.0009618885815143585,
          "validation_loss": 0.0023498358204960823,
          "train_metric": 9.327660518465564e-05,
          "validation_metric": 0.00022786908084526658,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4087674370002787
        },
        {
          "epoch": 120,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0013322100130608305,
          "train_loss": 0.0008547147735953331,
          "validation_loss": 0.0016876833979040384,
          "train_metric": 8.28837146400474e-05,
          "validation_metric": 0.00016365863848477602,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4182434020003711
        },
        {
          "epoch": 121,
          "updates": 1452,
          "samples_seen": 92928,
          "online_loss": 0.001018462295178324,
          "train_loss": 0.0011819627834483981,
          "validation_loss": 0.0011545532615855336,
          "train_metric": 0.00011461772373877466,
          "validation_metric": 0.00011195974366273731,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4285540639993997
        },
        {
          "epoch": 122,
          "updates": 1464,
          "samples_seen": 93696,
          "online_loss": 0.0012612209514675972,
          "train_loss": 0.0007976984488777816,
          "validation_loss": 0.001698874868452549,
          "train_metric": 7.735469989711419e-05,
          "validation_metric": 0.0001647439057705924,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4378971900005126
        },
        {
          "epoch": 123,
          "updates": 1476,
          "samples_seen": 94464,
          "online_loss": 0.0017485249578991595,
          "train_loss": 0.0009174130391329527,
          "validation_loss": 0.001585477963089943,
          "train_metric": 8.896370854927227e-05,
          "validation_metric": 0.00015374753274954855,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4473586340009206
        },
        {
          "epoch": 124,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.001251981428746755,
          "train_loss": 0.0008344541420228779,
          "validation_loss": 0.0018709490541368723,
          "train_metric": 8.091898780548945e-05,
          "validation_metric": 0.00018143032502848655,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4565819620002003
        },
        {
          "epoch": 125,
          "updates": 1500,
          "samples_seen": 96000,
          "online_loss": 0.0008806413970887661,
          "train_loss": 0.0008247832302004099,
          "validation_loss": 0.0011820349609479308,
          "train_metric": 7.998117507668212e-05,
          "validation_metric": 0.00011462471593404189,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4660184399999707
        },
        {
          "epoch": 126,
          "updates": 1512,
          "samples_seen": 96768,
          "online_loss": 0.0009662121980606267,
          "train_loss": 0.0007864015060476959,
          "validation_loss": 0.0017469810554757714,
          "train_metric": 7.625920261489227e-05,
          "validation_metric": 0.00016940887144301087,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4750612560001173
        },
        {
          "epoch": 127,
          "updates": 1524,
          "samples_seen": 97536,
          "online_loss": 0.0008390548343110519,
          "train_loss": 0.0009555017459206283,
          "validation_loss": 0.002067721448838711,
          "train_metric": 9.26572538446635e-05,
          "validation_metric": 0.00020051180035807192,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4840134250007395
        },
        {
          "epoch": 128,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0008898682669193173,
          "train_loss": 0.0007316937553696334,
          "validation_loss": 0.001594569068402052,
          "train_metric": 7.095406908774748e-05,
          "validation_metric": 0.00015462911687791348,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4929392160011048
        },
        {
          "epoch": 129,
          "updates": 1548,
          "samples_seen": 99072,
          "online_loss": 0.0008667937023953224,
          "train_loss": 0.0007942856755107641,
          "validation_loss": 0.0012456366093829274,
          "train_metric": 7.702375296503305e-05,
          "validation_metric": 0.00012079231237294152,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5020680140005425
        },
        {
          "epoch": 130,
          "updates": 1560,
          "samples_seen": 99840,
          "online_loss": 0.0010568377232023825,
          "train_loss": 0.0007713933591730893,
          "validation_loss": 0.00138824840541929,
          "train_metric": 7.480382919311523e-05,
          "validation_metric": 0.00013462171773426235,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5118181230009213
        },
        {
          "epoch": 131,
          "updates": 1572,
          "samples_seen": 100608,
          "online_loss": 0.0016440613496039684,
          "train_loss": 0.0017173265805467963,
          "validation_loss": 0.001351380837149918,
          "train_metric": 0.00016653318016324192,
          "validation_metric": 0.00013104658864904195,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5205931649998092
        },
        {
          "epoch": 132,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.002607045418699272,
          "train_loss": 0.0015870212810114026,
          "validation_loss": 0.003723864909261465,
          "train_metric": 0.00015389718464575708,
          "validation_metric": 0.0003611119755078107,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5295263930001965
        },
        {
          "epoch": 133,
          "updates": 1596,
          "samples_seen": 102144,
          "online_loss": 0.0014234744400406878,
          "train_loss": 0.0011004310799762607,
          "validation_loss": 0.002023775829002261,
          "train_metric": 0.000106711384432856,
          "validation_metric": 0.0001962503301911056,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5384599379995052
        },
        {
          "epoch": 134,
          "updates": 1608,
          "samples_seen": 102912,
          "online_loss": 0.0009285625298313486,
          "train_loss": 0.0008946345187723637,
          "validation_loss": 0.0013588534202426672,
          "train_metric": 8.675479330122471e-05,
          "validation_metric": 0.00013177123037166893,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5487668759997177
        },
        {
          "epoch": 135,
          "updates": 1620,
          "samples_seen": 103680,
          "online_loss": 0.0008321670126557971,
          "train_loss": 0.0007723316084593534,
          "validation_loss": 0.0017405570251867175,
          "train_metric": 7.489482231903821e-05,
          "validation_metric": 0.00016878588940016925,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5574697669999296
        },
        {
          "epoch": 136,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0008267391773794467,
          "train_loss": 0.0007134290062822402,
          "validation_loss": 0.001285068690776825,
          "train_metric": 6.918289727764204e-05,
          "validation_metric": 0.00012461614096537232,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5667302910005674
        },
        {
          "epoch": 137,
          "updates": 1644,
          "samples_seen": 105216,
          "online_loss": 0.00076663684497665,
          "train_loss": 0.0006915437406860292,
          "validation_loss": 0.001511060749180615,
          "train_metric": 6.706061685690656e-05,
          "validation_metric": 0.0001465311215724796,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5757572359998449
        },
        {
          "epoch": 138,
          "updates": 1656,
          "samples_seen": 105984,
          "online_loss": 0.000803527562917831,
          "train_loss": 0.0008607963100075722,
          "validation_loss": 0.001975064864382148,
          "train_metric": 8.347343100467697e-05,
          "validation_metric": 0.00019152667664457113,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5843833729995822
        },
        {
          "epoch": 139,
          "updates": 1668,
          "samples_seen": 106752,
          "online_loss": 0.0008398939777786533,
          "train_loss": 0.0007016838644631207,
          "validation_loss": 0.001544189639389515,
          "train_metric": 6.804392614867538e-05,
          "validation_metric": 0.00014974370424170047,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5929995440001221
        },
        {
          "epoch": 140,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0009289673010547025,
          "train_loss": 0.000984752899967134,
          "validation_loss": 0.0011594112729653716,
          "train_metric": 9.549380047246814e-05,
          "validation_metric": 0.00011243083281442523,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6020216300003085
        },
        {
          "epoch": 141,
          "updates": 1692,
          "samples_seen": 108288,
          "online_loss": 0.0008204657933674753,
          "train_loss": 0.0006873189122416079,
          "validation_loss": 0.0015863195294514298,
          "train_metric": 6.665092951152474e-05,
          "validation_metric": 0.000153829125338234,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6110860310000135
        }
      ],
      "best_validation_metric": 0.00011195974366273731,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 11,
      "parameters": 2369,
      "train_samples": 768,
      "validation_rmse_m_s": 0.010581103140161583,
      "training_profiles": [
        0,
        1,
        4,
        6,
        7,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 12,
          "samples_seen": 768,
          "online_loss": 0.2511904550095399,
          "train_loss": 0.13154064118862152,
          "validation_loss": 0.08343078196048737,
          "train_metric": 0.012755806557834148,
          "validation_metric": 0.008090480230748653,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.012346806001005461
        },
        {
          "epoch": 2,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.09532567113637924,
          "train_loss": 0.06220892071723938,
          "validation_loss": 0.07160463184118271,
          "train_metric": 0.006032545585185289,
          "validation_metric": 0.006943670101463795,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.022273592001511133
        },
        {
          "epoch": 3,
          "updates": 36,
          "samples_seen": 2304,
          "online_loss": 0.04797139142950376,
          "train_loss": 0.035091135650873184,
          "validation_loss": 0.029242627322673798,
          "train_metric": 0.003402870148420334,
          "validation_metric": 0.0028357268311083317,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03205673000047682
        },
        {
          "epoch": 4,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.033876446075737476,
          "train_loss": 0.04056554660201073,
          "validation_loss": 0.035264600068330765,
          "train_metric": 0.003933736588805914,
          "validation_metric": 0.0034196909982711077,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04151817199999641
        },
        {
          "epoch": 5,
          "updates": 60,
          "samples_seen": 3840,
          "online_loss": 0.0330599796337386,
          "train_loss": 0.02281113900244236,
          "validation_loss": 0.034328412264585495,
          "train_metric": 0.0022120496723800898,
          "validation_metric": 0.003328906372189522,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05061173100148153
        },
        {
          "epoch": 6,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.024056872120127082,
          "train_loss": 0.025601709261536598,
          "validation_loss": 0.033650048077106476,
          "train_metric": 0.002482658252120018,
          "validation_metric": 0.0032631244976073503,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05969350600025791
        },
        {
          "epoch": 7,
          "updates": 84,
          "samples_seen": 5376,
          "online_loss": 0.01932806397477786,
          "train_loss": 0.018813295289874077,
          "validation_loss": 0.021309569478034973,
          "train_metric": 0.00182436965405941,
          "validation_metric": 0.0020664390176534653,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.06952851400092186
        },
        {
          "epoch": 8,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.01455846766475588,
          "train_loss": 0.011791516095399857,
          "validation_loss": 0.014370099641382694,
          "train_metric": 0.0011434510815888643,
          "validation_metric": 0.0013935023453086615,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07953588400050648
        },
        {
          "epoch": 9,
          "updates": 108,
          "samples_seen": 6912,
          "online_loss": 0.011374087810205916,
          "train_loss": 0.009941109456121922,
          "validation_loss": 0.010973168537020683,
          "train_metric": 0.0009640126954764128,
          "validation_metric": 0.0010640941327437758,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08942125100111298
        },
        {
          "epoch": 10,
          "updates": 120,
          "samples_seen": 7680,
          "online_loss": 0.013775227591395378,
          "train_loss": 0.018101343885064125,
          "validation_loss": 0.02145976573228836,
          "train_metric": 0.0017553298966959119,
          "validation_metric": 0.0020810042042285204,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09907888600173465
        },
        {
          "epoch": 11,
          "updates": 132,
          "samples_seen": 8448,
          "online_loss": 0.017668677299904328,
          "train_loss": 0.020909836515784264,
          "validation_loss": 0.02071322128176689,
          "train_metric": 0.0020276757422834635,
          "validation_metric": 0.0020086097065359354,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.17260619400076394
        },
        {
          "epoch": 12,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.012610101994747916,
          "train_loss": 0.013613268733024597,
          "validation_loss": 0.016237512230873108,
          "train_metric": 0.001320110633969307,
          "validation_metric": 0.0015745898708701134,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.182195399000193
        },
        {
          "epoch": 13,
          "updates": 156,
          "samples_seen": 9984,
          "online_loss": 0.009174551426743468,
          "train_loss": 0.0072700814343988895,
          "validation_loss": 0.008091682568192482,
          "train_metric": 0.0007049969281069934,
          "validation_metric": 0.0007846695370972157,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.19204986700060545
        },
        {
          "epoch": 14,
          "updates": 168,
          "samples_seen": 10752,
          "online_loss": 0.010802283921899894,
          "train_loss": 0.008508020080626011,
          "validation_loss": 0.010389652103185654,
          "train_metric": 0.0008250426617451012,
          "validation_metric": 0.0010075089521706104,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20137009700010822
        },
        {
          "epoch": 15,
          "updates": 180,
          "samples_seen": 11520,
          "online_loss": 0.025700842845253646,
          "train_loss": 0.016843529418110847,
          "validation_loss": 0.023374591022729874,
          "train_metric": 0.0016333566745743155,
          "validation_metric": 0.0022666892036795616,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21065414200165833
        },
        {
          "epoch": 16,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.012233484730434915,
          "train_loss": 0.00833966489881277,
          "validation_loss": 0.011191297322511673,
          "train_metric": 0.0008087169844657183,
          "validation_metric": 0.0010852463310584426,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21956563500134507
        },
        {
          "epoch": 17,
          "updates": 204,
          "samples_seen": 13056,
          "online_loss": 0.01711975986836478,
          "train_loss": 0.011135059408843517,
          "validation_loss": 0.010028447955846786,
          "train_metric": 0.0010797928553074598,
          "validation_metric": 0.0009724821429699659,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22854169700076454
        },
        {
          "epoch": 18,
          "updates": 216,
          "samples_seen": 13824,
          "online_loss": 0.013414324067222575,
          "train_loss": 0.007255332078784704,
          "validation_loss": 0.00543588912114501,
          "train_metric": 0.0007035667076706886,
          "validation_metric": 0.0005271310219541192,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.23846616500122764
        },
        {
          "epoch": 19,
          "updates": 228,
          "samples_seen": 14592,
          "online_loss": 0.009498179657384753,
          "train_loss": 0.007903389632701874,
          "validation_loss": 0.005120044108480215,
          "train_metric": 0.0007664103177376091,
          "validation_metric": 0.0004965027910657227,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24858441400101583
        },
        {
          "epoch": 20,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.010719567692528168,
          "train_loss": 0.008326740004122257,
          "validation_loss": 0.0069522312842309475,
          "train_metric": 0.0008074634824879467,
          "validation_metric": 0.000674174225423485,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25773597700026585
        },
        {
          "epoch": 21,
          "updates": 252,
          "samples_seen": 16128,
          "online_loss": 0.005903703102376312,
          "train_loss": 0.005004848353564739,
          "validation_loss": 0.007521722000092268,
          "train_metric": 0.00048533189692534506,
          "validation_metric": 0.0007293990929611027,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.26691146100165497
        },
        {
          "epoch": 22,
          "updates": 264,
          "samples_seen": 16896,
          "online_loss": 0.0075640799865747494,
          "train_loss": 0.015136943198740482,
          "validation_loss": 0.014762640930712223,
          "train_metric": 0.0014678650768473744,
          "validation_metric": 0.00143156829290092,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27603693500168447
        },
        {
          "epoch": 23,
          "updates": 276,
          "samples_seen": 17664,
          "online_loss": 0.010708646480149278,
          "train_loss": 0.00773077504709363,
          "validation_loss": 0.0106160594150424,
          "train_metric": 0.0007496715406887233,
          "validation_metric": 0.0010294642997905612,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.28540185100064264
        },
        {
          "epoch": 24,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.010221113916486502,
          "train_loss": 0.011129074729979038,
          "validation_loss": 0.01731887459754944,
          "train_metric": 0.0010792125249281526,
          "validation_metric": 0.0016794519033282995,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2943969530006143
        },
        {
          "epoch": 25,
          "updates": 300,
          "samples_seen": 19200,
          "online_loss": 0.02818506381784876,
          "train_loss": 0.021593628451228142,
          "validation_loss": 0.028050735592842102,
          "train_metric": 0.0020939845126122236,
          "validation_metric": 0.0027201459743082523,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3035882780004613
        },
        {
          "epoch": 26,
          "updates": 312,
          "samples_seen": 19968,
          "online_loss": 0.015275806421414018,
          "train_loss": 0.01030733622610569,
          "validation_loss": 0.01169421337544918,
          "train_metric": 0.000999526702798903,
          "validation_metric": 0.0011340152705088258,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3129245780000929
        },
        {
          "epoch": 27,
          "updates": 324,
          "samples_seen": 20736,
          "online_loss": 0.014392147849624356,
          "train_loss": 0.009285742416977882,
          "validation_loss": 0.01004004292190075,
          "train_metric": 0.0009004603489302099,
          "validation_metric": 0.0009736065985634923,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3221212930002366
        },
        {
          "epoch": 28,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.008007664737912515,
          "train_loss": 0.009245752356946468,
          "validation_loss": 0.011212890967726707,
          "train_metric": 0.000896582321729511,
          "validation_metric": 0.0010873404098674655,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3312456630010274
        },
        {
          "epoch": 29,
          "updates": 348,
          "samples_seen": 22272,
          "online_loss": 0.005769369483459741,
          "train_loss": 0.004467126447707415,
          "validation_loss": 0.006840132176876068,
          "train_metric": 0.0004331878153607249,
          "validation_metric": 0.0006633037701249123,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3405917380005121
        },
        {
          "epoch": 30,
          "updates": 360,
          "samples_seen": 23040,
          "online_loss": 0.009877877659164369,
          "train_loss": 0.012157230637967587,
          "validation_loss": 0.010646713897585869,
          "train_metric": 0.0011789152631536126,
          "validation_metric": 0.001032436965033412,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3495134050008346
        },
        {
          "epoch": 31,
          "updates": 372,
          "samples_seen": 23808,
          "online_loss": 0.008846909855492413,
          "train_loss": 0.007929042913019657,
          "validation_loss": 0.008182764984667301,
          "train_metric": 0.0007688979967497289,
          "validation_metric": 0.0007935019675642252,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.35829004700099176
        },
        {
          "epoch": 32,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.0051671583399487036,
          "train_loss": 0.006096261087805033,
          "validation_loss": 0.005064744036644697,
          "train_metric": 0.0005911687621846795,
          "validation_metric": 0.0004911401774734259,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.36797903000115184
        },
        {
          "epoch": 33,
          "updates": 396,
          "samples_seen": 25344,
          "online_loss": 0.006313897979756196,
          "train_loss": 0.008074836805462837,
          "validation_loss": 0.006967291235923767,
          "train_metric": 0.0007830359390936792,
          "validation_metric": 0.0006756347138434649,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.3773772890017426
        },
        {
          "epoch": 34,
          "updates": 408,
          "samples_seen": 26112,
          "online_loss": 0.006375687972952922,
          "train_loss": 0.010554282926023006,
          "validation_loss": 0.009317183867096901,
          "train_metric": 0.0010234735673293471,
          "validation_metric": 0.0009035091497935355,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.38640528500036453
        },
        {
          "epoch": 35,
          "updates": 420,
          "samples_seen": 26880,
          "online_loss": 0.0054377950145863,
          "train_loss": 0.0045149195939302444,
          "validation_loss": 0.0036008497700095177,
          "train_metric": 0.00043782233842648566,
          "validation_metric": 0.00034918286837637424,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.39610673600145674
        },
        {
          "epoch": 36,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.003834146239872401,
          "train_loss": 0.003323223674669862,
          "validation_loss": 0.0037125837989151478,
          "train_metric": 0.0003222608065698296,
          "validation_metric": 0.00036001799162477255,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4054658700006257
        },
        {
          "epoch": 37,
          "updates": 444,
          "samples_seen": 28416,
          "online_loss": 0.004057133919559419,
          "train_loss": 0.0027651668060570955,
          "validation_loss": 0.003563077189028263,
          "train_metric": 0.00026814473676495254,
          "validation_metric": 0.000345520005794242,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4152201030010474
        },
        {
          "epoch": 38,
          "updates": 456,
          "samples_seen": 29184,
          "online_loss": 0.004968616160719345,
          "train_loss": 0.005969271529465914,
          "validation_loss": 0.006734903901815414,
          "train_metric": 0.0005788542912341654,
          "validation_metric": 0.0006530995015054941,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4241689590016904
        },
        {
          "epoch": 39,
          "updates": 468,
          "samples_seen": 29952,
          "online_loss": 0.005158620561511877,
          "train_loss": 0.003824077546596527,
          "validation_loss": 0.0035781534388661385,
          "train_metric": 0.00037082983180880547,
          "validation_metric": 0.00034698194940574467,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.43355644500115886
        },
        {
          "epoch": 40,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.0034718913423906392,
          "train_loss": 0.002663401886820793,
          "validation_loss": 0.0045485966838896275,
          "train_metric": 0.00025827635545283556,
          "validation_metric": 0.0004410881665535271,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4425244410012965
        },
        {
          "epoch": 41,
          "updates": 492,
          "samples_seen": 31488,
          "online_loss": 0.008616760654452568,
          "train_loss": 0.012083497829735279,
          "validation_loss": 0.01383018959313631,
          "train_metric": 0.001171765266917646,
          "validation_metric": 0.0013411459513008595,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4513220500011812
        },
        {
          "epoch": 42,
          "updates": 504,
          "samples_seen": 32256,
          "online_loss": 0.009134823029550413,
          "train_loss": 0.007482882589101791,
          "validation_loss": 0.009675348177552223,
          "train_metric": 0.0007256327662616968,
          "validation_metric": 0.000938241311814636,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.46016286900157866
        },
        {
          "epoch": 43,
          "updates": 516,
          "samples_seen": 33024,
          "online_loss": 0.004393384365054469,
          "train_loss": 0.003573810448870063,
          "validation_loss": 0.003043525153771043,
          "train_metric": 0.00034656081697903574,
          "validation_metric": 0.0002951377537101507,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.47004877400104306
        },
        {
          "epoch": 44,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.003525933949276805,
          "train_loss": 0.0025292467325925827,
          "validation_loss": 0.0037555352319031954,
          "train_metric": 0.00024526697234250605,
          "validation_metric": 0.0003641830990090966,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4793796060002933
        },
        {
          "epoch": 45,
          "updates": 540,
          "samples_seen": 34560,
          "online_loss": 0.007999241788638756,
          "train_loss": 0.009280218742787838,
          "validation_loss": 0.013010505586862564,
          "train_metric": 0.0008999244892038405,
          "validation_metric": 0.001261659199371934,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4881280420013354
        },
        {
          "epoch": 46,
          "updates": 552,
          "samples_seen": 35328,
          "online_loss": 0.004519151261774823,
          "train_loss": 0.002915630117058754,
          "validation_loss": 0.005076534114778042,
          "train_metric": 0.0002827354765031487,
          "validation_metric": 0.0004922834923490882,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.497084550001091
        },
        {
          "epoch": 47,
          "updates": 564,
          "samples_seen": 36096,
          "online_loss": 0.0036087261008409164,
          "train_loss": 0.002414036775007844,
          "validation_loss": 0.0034697270020842552,
          "train_metric": 0.00023409479763358831,
          "validation_metric": 0.00033646757947281003,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5061404710013448
        },
        {
          "epoch": 48,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.004947332189961647,
          "train_loss": 0.0024769629817456007,
          "validation_loss": 0.0022617015056312084,
          "train_metric": 0.00024019689590204507,
          "validation_metric": 0.00021932255185674876,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5160311660001753
        },
        {
          "epoch": 49,
          "updates": 588,
          "samples_seen": 37632,
          "online_loss": 0.006455134047428146,
          "train_loss": 0.008811206556856632,
          "validation_loss": 0.008519665338099003,
          "train_metric": 0.0008544434676878154,
          "validation_metric": 0.0008261718903668225,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5254318960014643
        },
        {
          "epoch": 50,
          "updates": 600,
          "samples_seen": 38400,
          "online_loss": 0.007599750048636149,
          "train_loss": 0.003092320403084159,
          "validation_loss": 0.00465333741158247,
          "train_metric": 0.0002998695708811283,
          "validation_metric": 0.00045124508324079216,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5344912150012533
        },
        {
          "epoch": 51,
          "updates": 612,
          "samples_seen": 39168,
          "online_loss": 0.004772083484567702,
          "train_loss": 0.00610625883564353,
          "validation_loss": 0.006431185640394688,
          "train_metric": 0.0005921382107771933,
          "validation_metric": 0.0006236472981981933,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5436953370008268
        },
        {
          "epoch": 52,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.00636286722146906,
          "train_loss": 0.007133313920348883,
          "validation_loss": 0.007389704696834087,
          "train_metric": 0.000691734254360199,
          "validation_metric": 0.0007165970746427774,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5526524520009843
        },
        {
          "epoch": 53,
          "updates": 636,
          "samples_seen": 40704,
          "online_loss": 0.00508090085349977,
          "train_loss": 0.003700885223224759,
          "validation_loss": 0.005924556870013475,
          "train_metric": 0.0003588835534173995,
          "validation_metric": 0.0005745182279497385,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5616901790017437
        },
        {
          "epoch": 54,
          "updates": 648,
          "samples_seen": 41472,
          "online_loss": 0.003739459721449142,
          "train_loss": 0.004007634241133928,
          "validation_loss": 0.0040937731973826885,
          "train_metric": 0.0003886297345161438,
          "validation_metric": 0.0003969828539993614,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5711059850000311
        },
        {
          "epoch": 55,
          "updates": 660,
          "samples_seen": 42240,
          "online_loss": 0.0033377038344042376,
          "train_loss": 0.004337516613304615,
          "validation_loss": 0.005376025568693876,
          "train_metric": 0.00042061915155500174,
          "validation_metric": 0.0005213259137235582,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5803940850000799
        },
        {
          "epoch": 56,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.003426018044895803,
          "train_loss": 0.00283797481097281,
          "validation_loss": 0.003656539134681225,
          "train_metric": 0.000275205064099282,
          "validation_metric": 0.00035458322963677347,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5893295990008482
        },
        {
          "epoch": 57,
          "updates": 684,
          "samples_seen": 43776,
          "online_loss": 0.0025459659712699554,
          "train_loss": 0.003098939312621951,
          "validation_loss": 0.004393997136503458,
          "train_metric": 0.00030051139765419066,
          "validation_metric": 0.0004260962305124849,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5985132950008847
        },
        {
          "epoch": 58,
          "updates": 696,
          "samples_seen": 44544,
          "online_loss": 0.002815911313518882,
          "train_loss": 0.003730800934135914,
          "validation_loss": 0.005225213244557381,
          "train_metric": 0.00036178456502966583,
          "validation_metric": 0.0005067011807113886,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.607777271001396
        },
        {
          "epoch": 59,
          "updates": 708,
          "samples_seen": 45312,
          "online_loss": 0.0024978099099826068,
          "train_loss": 0.0015975037822499871,
          "validation_loss": 0.002198752947151661,
          "train_metric": 0.00015491370868403465,
          "validation_metric": 0.0002132182416971773,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6175569580009324
        },
        {
          "epoch": 60,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.0021090276325897626,
          "train_loss": 0.0017432126915082335,
          "validation_loss": 0.0027182912454009056,
          "train_metric": 0.00016904341464396566,
          "validation_metric": 0.0002635990967974067,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6268830110002455
        },
        {
          "epoch": 61,
          "updates": 732,
          "samples_seen": 46848,
          "online_loss": 0.002094323567386406,
          "train_loss": 0.00344483251683414,
          "validation_loss": 0.005004082806408405,
          "train_metric": 0.0003340534458402544,
          "validation_metric": 0.0004852576821576804,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6362590120006644
        },
        {
          "epoch": 62,
          "updates": 744,
          "samples_seen": 47616,
          "online_loss": 0.0032645040967812142,
          "train_loss": 0.0027567141223698854,
          "validation_loss": 0.002885209396481514,
          "train_metric": 0.0002673250564839691,
          "validation_metric": 0.0002797855413518846,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6456162550002773
        },
        {
          "epoch": 63,
          "updates": 756,
          "samples_seen": 48384,
          "online_loss": 0.002662694915973892,
          "train_loss": 0.0016349215293303132,
          "validation_loss": 0.0017710793763399124,
          "train_metric": 0.00015854218509048223,
          "validation_metric": 0.00017174574895761907,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6553146790010942
        },
        {
          "epoch": 64,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.0021779704256914556,
          "train_loss": 0.0014819590141996741,
          "validation_loss": 0.001781704369932413,
          "train_metric": 0.00014370903954841197,
          "validation_metric": 0.00017277608276344836,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6646205209999607
        },
        {
          "epoch": 65,
          "updates": 780,
          "samples_seen": 49920,
          "online_loss": 0.00458336253844512,
          "train_loss": 0.0034844938199967146,
          "validation_loss": 0.0027086609043180943,
          "train_metric": 0.0003378995752427727,
          "validation_metric": 0.000262665213085711,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6734981750014413
        },
        {
          "epoch": 66,
          "updates": 792,
          "samples_seen": 50688,
          "online_loss": 0.003088721103267744,
          "train_loss": 0.0032313901465386152,
          "validation_loss": 0.0017329123802483082,
          "train_metric": 0.0003133554710075259,
          "validation_metric": 0.00016804460028652102,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6830278480010747
        },
        {
          "epoch": 67,
          "updates": 804,
          "samples_seen": 51456,
          "online_loss": 0.0020577630784828216,
          "train_loss": 0.0017137081595137715,
          "validation_loss": 0.0014174843672662973,
          "train_metric": 0.00016618230438325554,
          "validation_metric": 0.00013745680917054415,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6926906360004068
        },
        {
          "epoch": 68,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.004123110459962239,
          "train_loss": 0.0052749053575098515,
          "validation_loss": 0.005401821341365576,
          "train_metric": 0.0005115200183354318,
          "validation_metric": 0.0005238273879513144,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.701979798001048
        },
        {
          "epoch": 69,
          "updates": 828,
          "samples_seen": 52992,
          "online_loss": 0.00305564269850341,
          "train_loss": 0.003509216709062457,
          "validation_loss": 0.003824003040790558,
          "train_metric": 0.0003402969741728157,
          "validation_metric": 0.0003708226140588522,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.710984676001317
        },
        {
          "epoch": 70,
          "updates": 840,
          "samples_seen": 53760,
          "online_loss": 0.004466748874013622,
          "train_loss": 0.004428642336279154,
          "validation_loss": 0.004919473081827164,
          "train_metric": 0.0004294558602850884,
          "validation_metric": 0.0004770528757944703,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7199599890009267
        },
        {
          "epoch": 71,
          "updates": 852,
          "samples_seen": 54528,
          "online_loss": 0.0025258194364141673,
          "train_loss": 0.0018246088875457644,
          "validation_loss": 0.0018495621625334024,
          "train_metric": 0.0001769366062944755,
          "validation_metric": 0.0001793563860701397,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.728781155001343
        },
        {
          "epoch": 72,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0019184504684138421,
          "train_loss": 0.0013054691953584552,
          "validation_loss": 0.0018491734517738223,
          "train_metric": 0.00012659441563300788,
          "validation_metric": 0.00017931871116161346,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7376223920000484
        },
        {
          "epoch": 73,
          "updates": 876,
          "samples_seen": 56064,
          "online_loss": 0.005203293180481221,
          "train_loss": 0.003304019570350647,
          "validation_loss": 0.0039947740733623505,
          "train_metric": 0.00032039853977039456,
          "validation_metric": 0.00038738263538107276,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7467189410017454
        },
        {
          "epoch": 74,
          "updates": 888,
          "samples_seen": 56832,
          "online_loss": 0.004315338107213999,
          "train_loss": 0.0030470332130789757,
          "validation_loss": 0.0032025810796767473,
          "train_metric": 0.00029547797748818994,
          "validation_metric": 0.0003105617652181536,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7555359100006172
        },
        {
          "epoch": 75,
          "updates": 900,
          "samples_seen": 57600,
          "online_loss": 0.001730291328082482,
          "train_loss": 0.0013992368476465344,
          "validation_loss": 0.0011933003552258015,
          "train_metric": 0.0001356872817268595,
          "validation_metric": 0.00011571714276215062,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7653028670010826
        },
        {
          "epoch": 76,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.0017486786042960982,
          "train_loss": 0.0012273163301870227,
          "validation_loss": 0.0016560895601287484,
          "train_metric": 0.00011901575635420159,
          "validation_metric": 0.00016059490735642612,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7744748600016464
        },
        {
          "epoch": 77,
          "updates": 924,
          "samples_seen": 59136,
          "online_loss": 0.0013245318550616503,
          "train_loss": 0.00107875547837466,
          "validation_loss": 0.0014658300206065178,
          "train_metric": 0.00010460944031365216,
          "validation_metric": 0.00014214498514775187,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7838483700015786
        },
        {
          "epoch": 78,
          "updates": 936,
          "samples_seen": 59904,
          "online_loss": 0.0019079161866102368,
          "train_loss": 0.0011712664272636175,
          "validation_loss": 0.0015031921211630106,
          "train_metric": 0.00011358045594533905,
          "validation_metric": 0.00014576807734556496,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7928691240012995
        },
        {
          "epoch": 79,
          "updates": 948,
          "samples_seen": 60672,
          "online_loss": 0.001920730312122032,
          "train_loss": 0.0019168645376339555,
          "validation_loss": 0.002192594576627016,
          "train_metric": 0.00018588289094623178,
          "validation_metric": 0.00021262103109620512,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8019702520014107
        },
        {
          "epoch": 80,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.0032405064897223688,
          "train_loss": 0.0031909116078168154,
          "validation_loss": 0.002799591049551964,
          "train_metric": 0.0003094301500823349,
          "validation_metric": 0.00027148291701450944,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8110724930011202
        },
        {
          "epoch": 81,
          "updates": 972,
          "samples_seen": 62208,
          "online_loss": 0.002520648888700331,
          "train_loss": 0.0014571825740858912,
          "validation_loss": 0.0015431693755090237,
          "train_metric": 0.000141306416480802,
          "validation_metric": 0.00014964478032197803,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8200414990005811
        },
        {
          "epoch": 82,
          "updates": 984,
          "samples_seen": 62976,
          "online_loss": 0.002459905355256827,
          "train_loss": 0.004202689975500107,
          "validation_loss": 0.0034652003087103367,
          "train_metric": 0.0004075446631759405,
          "validation_metric": 0.00033602866460569203,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8291537660006725
        },
        {
          "epoch": 83,
          "updates": 996,
          "samples_seen": 63744,
          "online_loss": 0.002740089277115961,
          "train_loss": 0.0027196481823921204,
          "validation_loss": 0.0023767005186527967,
          "train_metric": 0.00026373067521490157,
          "validation_metric": 0.00023047423746902496,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8381977270000789
        },
        {
          "epoch": 84,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.002573590628647556,
          "train_loss": 0.0014781575882807374,
          "validation_loss": 0.0020896641071885824,
          "train_metric": 0.0001433404249837622,
          "validation_metric": 0.00020263963961042464,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8473930950003705
        },
        {
          "epoch": 85,
          "updates": 1020,
          "samples_seen": 65280,
          "online_loss": 0.0018270278960699216,
          "train_loss": 0.0016784296603873372,
          "validation_loss": 0.001762857660651207,
          "train_metric": 0.00016276126552838832,
          "validation_metric": 0.00017094844952225685,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8563925320013368
        },
        {
          "epoch": 86,
          "updates": 1032,
          "samples_seen": 66048,
          "online_loss": 0.0031862108735367656,
          "train_loss": 0.0017340313643217087,
          "validation_loss": 0.0015965865459293127,
          "train_metric": 0.0001681530848145485,
          "validation_metric": 0.00015482476737815887,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8655523490015185
        },
        {
          "epoch": 87,
          "updates": 1044,
          "samples_seen": 66816,
          "online_loss": 0.0014021295210113749,
          "train_loss": 0.001180364633910358,
          "validation_loss": 0.0010184801649302244,
          "train_metric": 0.00011446272401371971,
          "validation_metric": 9.876441617961973e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8751899280014186
        },
        {
          "epoch": 88,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.0013390548338065855,
          "train_loss": 0.001650824211537838,
          "validation_loss": 0.0017865711124613881,
          "train_metric": 0.00016008430975489318,
          "validation_metric": 0.0001732480013743043,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8844244330011861
        },
        {
          "epoch": 89,
          "updates": 1068,
          "samples_seen": 68352,
          "online_loss": 0.0012718464398252156,
          "train_loss": 0.0009074225090444088,
          "validation_loss": 0.0013275539968162775,
          "train_metric": 8.799490024102852e-05,
          "validation_metric": 0.00012873603554908186,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.893567487000837
        },
        {
          "epoch": 90,
          "updates": 1080,
          "samples_seen": 69120,
          "online_loss": 0.0013062623280954238,
          "train_loss": 0.0014206873020157218,
          "validation_loss": 0.001376548083499074,
          "train_metric": 0.00013776739069726318,
          "validation_metric": 0.00013348710490390658,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9025344500005303
        },
        {
          "epoch": 91,
          "updates": 1092,
          "samples_seen": 69888,
          "online_loss": 0.0014066787625779398,
          "train_loss": 0.002320053754374385,
          "validation_loss": 0.0016051025595515966,
          "train_metric": 0.0002249810640932992,
          "validation_metric": 0.00015565057401545346,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9113787280002725
        },
        {
          "epoch": 92,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.0010993925485915195,
          "train_loss": 0.0009524838533252478,
          "validation_loss": 0.0012167191598564386,
          "train_metric": 9.23645930015482e-05,
          "validation_metric": 0.00011798810010077432,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9204742810015887
        },
        {
          "epoch": 93,
          "updates": 1116,
          "samples_seen": 71424,
          "online_loss": 0.0009938013536157086,
          "train_loss": 0.0008273790590465069,
          "validation_loss": 0.0012189442059025168,
          "train_metric": 8.023289410630241e-05,
          "validation_metric": 0.00011820389772765338,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9294672910000372
        },
        {
          "epoch": 94,
          "updates": 1128,
          "samples_seen": 72192,
          "online_loss": 0.0012290003845312942,
          "train_loss": 0.002055492950603366,
          "validation_loss": 0.002495750319212675,
          "train_metric": 0.00019932597933802754,
          "validation_metric": 0.00024201876658480614,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9384339790012746
        },
        {
          "epoch": 95,
          "updates": 1140,
          "samples_seen": 72960,
          "online_loss": 0.001269701814938647,
          "train_loss": 0.0018912433879449964,
          "validation_loss": 0.0024886042810976505,
          "train_metric": 0.00018339836969971657,
          "validation_metric": 0.00024132580438163131,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9474561060014821
        },
        {
          "epoch": 96,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0021592856549735493,
          "train_loss": 0.003234334522858262,
          "validation_loss": 0.003472167532891035,
          "train_metric": 0.0003136410377919674,
          "validation_metric": 0.0003367042518220842,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9563918100011506
        },
        {
          "epoch": 97,
          "updates": 1164,
          "samples_seen": 74496,
          "online_loss": 0.002074315406692525,
          "train_loss": 0.001826103194616735,
          "validation_loss": 0.002371723297983408,
          "train_metric": 0.00017708154337015003,
          "validation_metric": 0.00022999160864856094,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9654383960005362
        },
        {
          "epoch": 98,
          "updates": 1176,
          "samples_seen": 75264,
          "online_loss": 0.0012250215125580628,
          "train_loss": 0.0014616836560890079,
          "validation_loss": 0.0009535670978948474,
          "train_metric": 0.0001417429157299921,
          "validation_metric": 9.246963600162417e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.97491339500084
        },
        {
          "epoch": 99,
          "updates": 1188,
          "samples_seen": 76032,
          "online_loss": 0.0012916851968232852,
          "train_loss": 0.0010109024588018656,
          "validation_loss": 0.001167483627796173,
          "train_metric": 9.80295953922905e-05,
          "validation_metric": 0.00011321362399030477,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9841794220010343
        },
        {
          "epoch": 100,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.0015356332345011954,
          "train_loss": 0.0016848471714183688,
          "validation_loss": 0.0015371604822576046,
          "train_metric": 0.00016338359273504466,
          "validation_metric": 0.00014906207798048854,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9931907200007117
        },
        {
          "epoch": 101,
          "updates": 1212,
          "samples_seen": 77568,
          "online_loss": 0.001984987611649558,
          "train_loss": 0.0015993998385965824,
          "validation_loss": 0.002736792666837573,
          "train_metric": 0.00015509755758102983,
          "validation_metric": 0.0002653932315297425,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0024432440004603
        },
        {
          "epoch": 102,
          "updates": 1224,
          "samples_seen": 78336,
          "online_loss": 0.0016098143241833895,
          "train_loss": 0.0025502245407551527,
          "validation_loss": 0.002105201594531536,
          "train_metric": 0.00024730124277994037,
          "validation_metric": 0.00020414634491316974,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0115561120001075
        },
        {
          "epoch": 103,
          "updates": 1236,
          "samples_seen": 79104,
          "online_loss": 0.0019818225652367496,
          "train_loss": 0.0013092132285237312,
          "validation_loss": 0.001453681499697268,
          "train_metric": 0.00012695750046987087,
          "validation_metric": 0.00014096690574660897,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0204719910016138
        },
        {
          "epoch": 104,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.0009877275952021591,
          "train_loss": 0.0007758014253340662,
          "validation_loss": 0.0007881926139816642,
          "train_metric": 7.52312844269909e-05,
          "validation_metric": 7.643290155101568e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.030256209000072
        },
        {
          "epoch": 105,
          "updates": 1260,
          "samples_seen": 80640,
          "online_loss": 0.0011080331569246482,
          "train_loss": 0.001190915354527533,
          "validation_loss": 0.0021707857958972454,
          "train_metric": 0.00011548586917342618,
          "validation_metric": 0.00021050620125606656,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0395434260008187
        },
        {
          "epoch": 106,
          "updates": 1272,
          "samples_seen": 81408,
          "online_loss": 0.0009602078838118663,
          "train_loss": 0.0006873794482089579,
          "validation_loss": 0.0009948831284418702,
          "train_metric": 6.665680120931938e-05,
          "validation_metric": 9.647614933783188e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0488169190011831
        },
        {
          "epoch": 107,
          "updates": 1284,
          "samples_seen": 82176,
          "online_loss": 0.001008193357847631,
          "train_loss": 0.0015476475236937404,
          "validation_loss": 0.0014772303402423859,
          "train_metric": 0.00015007902402430773,
          "validation_metric": 0.0001432504941476509,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0576895460017113
        },
        {
          "epoch": 108,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.0012735306845570449,
          "train_loss": 0.0008165697217918932,
          "validation_loss": 0.0011297406163066626,
          "train_metric": 7.91846978245303e-05,
          "validation_metric": 0.00010955359903164208,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.066919612001584
        },
        {
          "epoch": 109,
          "updates": 1308,
          "samples_seen": 83712,
          "online_loss": 0.0009625005889877988,
          "train_loss": 0.0021629049442708492,
          "validation_loss": 0.002163512632250786,
          "train_metric": 0.00020974199287593365,
          "validation_metric": 0.00020980089902877808,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0759496780010522
        },
        {
          "epoch": 110,
          "updates": 1320,
          "samples_seen": 84480,
          "online_loss": 0.002493541719256124,
          "train_loss": 0.001571450848132372,
          "validation_loss": 0.0018242094665765762,
          "train_metric": 0.00015238729247357696,
          "validation_metric": 0.00017689786909613758,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0848718310007825
        },
        {
          "epoch": 111,
          "updates": 1332,
          "samples_seen": 85248,
          "online_loss": 0.0017277702475742747,
          "train_loss": 0.0008866467396728694,
          "validation_loss": 0.0010397626319900155,
          "train_metric": 8.598021668149158e-05,
          "validation_metric": 0.00010082822700496763,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0939725170010206
        },
        {
          "epoch": 112,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.0017223413378815167,
          "train_loss": 0.0013361731544137,
          "validation_loss": 0.001461500534787774,
          "train_metric": 0.0001295718684559688,
          "validation_metric": 0.00014172514784149826,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1031732020001073
        },
        {
          "epoch": 113,
          "updates": 1356,
          "samples_seen": 86784,
          "online_loss": 0.0012447384409218405,
          "train_loss": 0.000938901852350682,
          "validation_loss": 0.0009335944778285921,
          "train_metric": 9.104751370614395e-05,
          "validation_metric": 9.05328561202623e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.111972639000669
        },
        {
          "epoch": 114,
          "updates": 1368,
          "samples_seen": 87552,
          "online_loss": 0.0007956004944086695,
          "train_loss": 0.0006750769098289311,
          "validation_loss": 0.000915661221370101,
          "train_metric": 6.54637988191098e-05,
          "validation_metric": 8.87938222149387e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1209131560008245
        },
        {
          "epoch": 115,
          "updates": 1380,
          "samples_seen": 88320,
          "online_loss": 0.0007000157711445354,
          "train_loss": 0.0007226886227726936,
          "validation_loss": 0.0012111185351386666,
          "train_metric": 7.008082320680842e-05,
          "validation_metric": 0.00011744500079657882,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1301828200012096
        },
        {
          "epoch": 116,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0005950076701992657,
          "train_loss": 0.0005448927986435592,
          "validation_loss": 0.0007337626302614808,
          "train_metric": 5.2839535783277825e-05,
          "validation_metric": 7.115468906704336e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1396937410008832
        },
        {
          "epoch": 117,
          "updates": 1404,
          "samples_seen": 89856,
          "online_loss": 0.0005711073414810622,
          "train_loss": 0.0005388027639128268,
          "validation_loss": 0.0009728710865601897,
          "train_metric": 5.2248971769586205e-05,
          "validation_metric": 9.434160165255889e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1487680600002932
        },
        {
          "epoch": 118,
          "updates": 1416,
          "samples_seen": 90624,
          "online_loss": 0.0005441506170124436,
          "train_loss": 0.0007178044761531055,
          "validation_loss": 0.0007864832878112793,
          "train_metric": 6.960718746995553e-05,
          "validation_metric": 7.626714068464935e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1576466170008644
        },
        {
          "epoch": 119,
          "updates": 1428,
          "samples_seen": 91392,
          "online_loss": 0.0006645112225669436,
          "train_loss": 0.0005046960432082415,
          "validation_loss": 0.000725881545804441,
          "train_metric": 4.894155790680088e-05,
          "validation_metric": 7.039043703116477e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2306369990001258
        },
        {
          "epoch": 120,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0005681107395503204,
          "train_loss": 0.0005636601708829403,
          "validation_loss": 0.0009245898691006005,
          "train_metric": 5.465944923344068e-05,
          "validation_metric": 8.96596466191113e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.23973283900159
        },
        {
          "epoch": 121,
          "updates": 1452,
          "samples_seen": 92928,
          "online_loss": 0.0005943374271737412,
          "train_loss": 0.0005450450698845088,
          "validation_loss": 0.0007848478271625936,
          "train_metric": 5.285430233925581e-05,
          "validation_metric": 7.610853936057538e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2488427330008562
        },
        {
          "epoch": 122,
          "updates": 1464,
          "samples_seen": 93696,
          "online_loss": 0.0005374046450015157,
          "train_loss": 0.0006482838070951402,
          "validation_loss": 0.0011155846295878291,
          "train_metric": 6.286559801083058e-05,
          "validation_metric": 0.00010818087321240455,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.258013121001568
        },
        {
          "epoch": 123,
          "updates": 1476,
          "samples_seen": 94464,
          "online_loss": 0.0006744020417196831,
          "train_loss": 0.0005386933335103095,
          "validation_loss": 0.0009370011393912137,
          "train_metric": 5.223835978540592e-05,
          "validation_metric": 9.086319914786145e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2671371740016184
        },
        {
          "epoch": 124,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0008952070468997894,
          "train_loss": 0.0006283597904257476,
          "validation_loss": 0.0009609180269762874,
          "train_metric": 6.093352931202389e-05,
          "validation_metric": 9.318248339695856e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.276056441000037
        },
        {
          "epoch": 125,
          "updates": 1500,
          "samples_seen": 96000,
          "online_loss": 0.000676295860709312,
          "train_loss": 0.0004621773259714246,
          "validation_loss": 0.0007079680799506605,
          "train_metric": 4.481841824599542e-05,
          "validation_metric": 6.865333125460893e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2854684640005871
        },
        {
          "epoch": 126,
          "updates": 1512,
          "samples_seen": 96768,
          "online_loss": 0.000526751835423056,
          "train_loss": 0.0009695104672573507,
          "validation_loss": 0.0007484802044928074,
          "train_metric": 9.4015704235062e-05,
          "validation_metric": 7.258188270498067e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2944383840003866
        },
        {
          "epoch": 127,
          "updates": 1524,
          "samples_seen": 97536,
          "online_loss": 0.0005046879353661401,
          "train_loss": 0.000589536561165005,
          "validation_loss": 0.0009531544637866318,
          "train_metric": 5.7168752391589805e-05,
          "validation_metric": 9.242962551070377e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3031221510009345
        },
        {
          "epoch": 128,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0005373472110174286,
          "train_loss": 0.00045035124639980495,
          "validation_loss": 0.0006790979532524943,
          "train_metric": 4.3671618186635897e-05,
          "validation_metric": 6.58537246636115e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3124619590016664
        },
        {
          "epoch": 129,
          "updates": 1548,
          "samples_seen": 99072,
          "online_loss": 0.0006866527461776665,
          "train_loss": 0.00043663650285452604,
          "validation_loss": 0.0006389272166416049,
          "train_metric": 4.234166772221215e-05,
          "validation_metric": 6.19582788203843e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3218920070012246
        },
        {
          "epoch": 130,
          "updates": 1560,
          "samples_seen": 99840,
          "online_loss": 0.0008087148332075836,
          "train_loss": 0.0005564946332015097,
          "validation_loss": 0.0010092125739902258,
          "train_metric": 5.3964595281286165e-05,
          "validation_metric": 9.786572627490386e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3310137750013382
        },
        {
          "epoch": 131,
          "updates": 1572,
          "samples_seen": 100608,
          "online_loss": 0.000531775431833618,
          "train_loss": 0.0004930224386043847,
          "validation_loss": 0.0007873100694268942,
          "train_metric": 4.780954259331338e-05,
          "validation_metric": 7.634731446160004e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3398380070011626
        },
        {
          "epoch": 132,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0004954904482777541,
          "train_loss": 0.0004447285027708858,
          "validation_loss": 0.0006464358302764595,
          "train_metric": 4.3126365198986605e-05,
          "validation_metric": 6.268639845075086e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3487751360007678
        },
        {
          "epoch": 133,
          "updates": 1596,
          "samples_seen": 102144,
          "online_loss": 0.0006270406920521054,
          "train_loss": 0.0004855840525124222,
          "validation_loss": 0.0006487875361926854,
          "train_metric": 4.708823325927369e-05,
          "validation_metric": 6.291445606620982e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3574771260009584
        },
        {
          "epoch": 134,
          "updates": 1608,
          "samples_seen": 102912,
          "online_loss": 0.0007324018248861345,
          "train_loss": 0.0008177866111509502,
          "validation_loss": 0.000778470013756305,
          "train_metric": 7.930267747724429e-05,
          "validation_metric": 7.549006841145456e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.366274460000568
        },
        {
          "epoch": 135,
          "updates": 1620,
          "samples_seen": 103680,
          "online_loss": 0.0007679707220328661,
          "train_loss": 0.0006256338092498481,
          "validation_loss": 0.00064184790244326,
          "train_metric": 6.0669186495942995e-05,
          "validation_metric": 6.224151002243161e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3748562270011462
        },
        {
          "epoch": 136,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0006421468594150307,
          "train_loss": 0.0004227127355989069,
          "validation_loss": 0.0006704851984977722,
          "train_metric": 4.09914500778541e-05,
          "validation_metric": 6.5018524765037e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3834500110006047
        },
        {
          "epoch": 137,
          "updates": 1644,
          "samples_seen": 105216,
          "online_loss": 0.0004210517957593159,
          "train_loss": 0.0004961538361385465,
          "validation_loss": 0.0010088697308674455,
          "train_metric": 4.811320104636252e-05,
          "validation_metric": 9.783246787264943e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3919845330001408
        },
        {
          "epoch": 138,
          "updates": 1656,
          "samples_seen": 105984,
          "online_loss": 0.0006122063617416037,
          "train_loss": 0.0006702628452330828,
          "validation_loss": 0.0012837594840675592,
          "train_metric": 6.499696610262617e-05,
          "validation_metric": 0.00012448917550500482,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.400683908001156
        },
        {
          "epoch": 139,
          "updates": 1668,
          "samples_seen": 106752,
          "online_loss": 0.0009007527987705544,
          "train_loss": 0.000538319640327245,
          "validation_loss": 0.0005961274728178978,
          "train_metric": 5.220211824052967e-05,
          "validation_metric": 5.780789069831371e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4100656520004122
        },
        {
          "epoch": 140,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0005947646689795268,
          "train_loss": 0.000446458492660895,
          "validation_loss": 0.000751371611841023,
          "train_metric": 4.329414150561206e-05,
          "validation_metric": 7.286226173164323e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4187902550002036
        },
        {
          "epoch": 141,
          "updates": 1692,
          "samples_seen": 108288,
          "online_loss": 0.00040726659184050124,
          "train_loss": 0.0003947397635784,
          "validation_loss": 0.000769853126257658,
          "train_metric": 3.827884211204946e-05,
          "validation_metric": 7.465446833521128e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.427342041000884
        },
        {
          "epoch": 142,
          "updates": 1704,
          "samples_seen": 109056,
          "online_loss": 0.0004025532228600544,
          "train_loss": 0.000528551172465086,
          "validation_loss": 0.0008561407448723912,
          "train_metric": 5.125485404278152e-05,
          "validation_metric": 8.302198693854734e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4360589030002302
        },
        {
          "epoch": 143,
          "updates": 1716,
          "samples_seen": 109824,
          "online_loss": 0.0007007550520938821,
          "train_loss": 0.0007571654859930277,
          "validation_loss": 0.001480599632486701,
          "train_metric": 7.34241257305257e-05,
          "validation_metric": 0.00014357722830027342,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4447362970004178
        },
        {
          "epoch": 144,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0010747103030250098,
          "train_loss": 0.000922336068470031,
          "validation_loss": 0.0011185682378709316,
          "train_metric": 8.944109868025407e-05,
          "validation_metric": 0.00010847019439097494,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.453319365000425
        },
        {
          "epoch": 145,
          "updates": 1740,
          "samples_seen": 111360,
          "online_loss": 0.0016783019382273778,
          "train_loss": 0.0004497575282584876,
          "validation_loss": 0.0008668588125146925,
          "train_metric": 4.361404353403486e-05,
          "validation_metric": 8.40613356558606e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.461934161001409
        },
        {
          "epoch": 146,
          "updates": 1752,
          "samples_seen": 112128,
          "online_loss": 0.000779424225280915,
          "train_loss": 0.0005081407143734396,
          "validation_loss": 0.000434798130299896,
          "train_metric": 4.927559712086804e-05,
          "validation_metric": 4.216339948470704e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4713460960010707
        },
        {
          "epoch": 147,
          "updates": 1764,
          "samples_seen": 112896,
          "online_loss": 0.0004884779179216517,
          "train_loss": 0.0004171680484432727,
          "validation_loss": 0.000713606714271009,
          "train_metric": 4.045377136208117e-05,
          "validation_metric": 6.920012674527243e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4800060010002198
        },
        {
          "epoch": 148,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.00048282406593595323,
          "train_loss": 0.0003923643671441823,
          "validation_loss": 0.0006805182783864439,
          "train_metric": 3.804849984589964e-05,
          "validation_metric": 6.599146581720561e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4883832720006467
        },
        {
          "epoch": 149,
          "updates": 1788,
          "samples_seen": 114432,
          "online_loss": 0.0004616328199820903,
          "train_loss": 0.00034149238490499556,
          "validation_loss": 0.0005450371536426246,
          "train_metric": 3.3115316909970716e-05,
          "validation_metric": 5.2853534725727513e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4969125250008801
        },
        {
          "epoch": 150,
          "updates": 1800,
          "samples_seen": 115200,
          "online_loss": 0.0004004111760878004,
          "train_loss": 0.00036369755980558693,
          "validation_loss": 0.000577154045458883,
          "train_metric": 3.526861110003665e-05,
          "validation_metric": 5.5967986554605886e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5056187690006482
        },
        {
          "epoch": 151,
          "updates": 1812,
          "samples_seen": 115968,
          "online_loss": 0.00041694815930289525,
          "train_loss": 0.0005085829761810601,
          "validation_loss": 0.0005310350679792464,
          "train_metric": 4.9318488891003653e-05,
          "validation_metric": 5.149572098162025e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5142406530012522
        },
        {
          "epoch": 152,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.0004932774403035486,
          "train_loss": 0.00033342026290483773,
          "validation_loss": 0.0006454507820308208,
          "train_metric": 3.233254756196402e-05,
          "validation_metric": 6.259087967919186e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5226763260016014
        },
        {
          "epoch": 153,
          "updates": 1836,
          "samples_seen": 117504,
          "online_loss": 0.00037321556616613333,
          "train_loss": 0.0003231654118280858,
          "validation_loss": 0.000560184009373188,
          "train_metric": 3.13381060550455e-05,
          "validation_metric": 5.4322365031111985e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5312163840007997
        },
        {
          "epoch": 154,
          "updates": 1848,
          "samples_seen": 118272,
          "online_loss": 0.0003918784107857694,
          "train_loss": 0.000617407204117626,
          "validation_loss": 0.0005467073060572147,
          "train_metric": 5.9871425037272274e-05,
          "validation_metric": 5.301549390424043e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5398600890002854
        },
        {
          "epoch": 155,
          "updates": 1860,
          "samples_seen": 119040,
          "online_loss": 0.0004981982049988195,
          "train_loss": 0.0003945249191019684,
          "validation_loss": 0.0006583290523849428,
          "train_metric": 3.8258007407421246e-05,
          "validation_metric": 6.383971049217507e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5483922030016402
        },
        {
          "epoch": 156,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.00039439527548286907,
          "train_loss": 0.000372993788914755,
          "validation_loss": 0.0007604914135299623,
          "train_metric": 3.617008405853994e-05,
          "validation_metric": 7.3746639827732e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5568213680016925
        },
        {
          "epoch": 157,
          "updates": 1884,
          "samples_seen": 120576,
          "online_loss": 0.00039965197841714445,
          "train_loss": 0.0004750888328999281,
          "validation_loss": 0.00044853848521597683,
          "train_metric": 4.6070486860116944e-05,
          "validation_metric": 4.349583468865603e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5654311190010048
        },
        {
          "epoch": 158,
          "updates": 1896,
          "samples_seen": 121344,
          "online_loss": 0.0004174661298748106,
          "train_loss": 0.00031985019450075924,
          "validation_loss": 0.0005977883702144027,
          "train_metric": 3.101662514382042e-05,
          "validation_metric": 5.796894402010366e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5740392210009304
        },
        {
          "epoch": 159,
          "updates": 1908,
          "samples_seen": 122112,
          "online_loss": 0.00041412788050365634,
          "train_loss": 0.000568514980841428,
          "validation_loss": 0.00044111619354225695,
          "train_metric": 5.5130236432887614e-05,
          "validation_metric": 4.2776075133588165e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.582716259001245
        },
        {
          "epoch": 160,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0006039725849404931,
          "train_loss": 0.00037592448643408716,
          "validation_loss": 0.0008614215184934437,
          "train_metric": 3.645427932497114e-05,
          "validation_metric": 8.35340833873488e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.591124184000364
        },
        {
          "epoch": 161,
          "updates": 1932,
          "samples_seen": 123648,
          "online_loss": 0.00034897423878040473,
          "train_loss": 0.0004570375895127654,
          "validation_loss": 0.0007801408064551651,
          "train_metric": 4.432000787346624e-05,
          "validation_metric": 7.565209671156481e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5997655640003359
        },
        {
          "epoch": 162,
          "updates": 1944,
          "samples_seen": 124416,
          "online_loss": 0.0004180806499789469,
          "train_loss": 0.00038267471245490015,
          "validation_loss": 0.0008354574674740434,
          "train_metric": 3.710886448970996e-05,
          "validation_metric": 8.10162746347487e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6083811150001566
        },
        {
          "epoch": 163,
          "updates": 1956,
          "samples_seen": 125184,
          "online_loss": 0.0004465232353444056,
          "train_loss": 0.0004417561867740005,
          "validation_loss": 0.0008622388122603297,
          "train_metric": 4.2838131776079535e-05,
          "validation_metric": 8.361331856576726e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6170641020016774
        },
        {
          "epoch": 164,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0003978675473869468,
          "train_loss": 0.0005344499368220568,
          "validation_loss": 0.00048790290020406246,
          "train_metric": 5.1826868002535775e-05,
          "validation_metric": 4.7313093091361225e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6256018490003044
        },
        {
          "epoch": 165,
          "updates": 1980,
          "samples_seen": 126720,
          "online_loss": 0.0004294335315838301,
          "train_loss": 0.00029988668393343687,
          "validation_loss": 0.0005434091435745358,
          "train_metric": 2.9080720196361654e-05,
          "validation_metric": 5.269566099741496e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6342639240010612
        },
        {
          "epoch": 166,
          "updates": 1992,
          "samples_seen": 127488,
          "online_loss": 0.00035393861374662566,
          "train_loss": 0.0003218671481590718,
          "validation_loss": 0.0004456362221390009,
          "train_metric": 3.121221379842609e-05,
          "validation_metric": 4.321439701016061e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.642790274001527
        }
      ],
      "best_validation_metric": 4.216339948470704e-05,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 22,
      "parameters": 2369,
      "train_samples": 768,
      "validation_rmse_m_s": 0.0064933350048112445,
      "training_profiles": [
        0,
        1,
        4,
        6,
        7,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 12,
          "samples_seen": 768,
          "online_loss": 0.29002902284264565,
          "train_loss": 0.09428044408559799,
          "validation_loss": 0.05948938801884651,
          "train_metric": 0.00914259534329176,
          "validation_metric": 0.005768825765699148,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.017172539999592118
        },
        {
          "epoch": 2,
          "updates": 24,
          "samples_seen": 1536,
          "online_loss": 0.07214350967357556,
          "train_loss": 0.05306924879550934,
          "validation_loss": 0.05686318874359131,
          "train_metric": 0.005146249663084745,
          "validation_metric": 0.0055141570046544075,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03133294799954456
        },
        {
          "epoch": 3,
          "updates": 36,
          "samples_seen": 2304,
          "online_loss": 0.04378137434832752,
          "train_loss": 0.0330018512904644,
          "validation_loss": 0.027254100888967514,
          "train_metric": 0.003200267441570759,
          "validation_metric": 0.002642894396558404,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.045535790999565506
        },
        {
          "epoch": 4,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.03397426071266333,
          "train_loss": 0.026375597342848778,
          "validation_loss": 0.021885929629206657,
          "train_metric": 0.0025577035266906023,
          "validation_metric": 0.0021223300136625767,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05956409899954451
        },
        {
          "epoch": 5,
          "updates": 60,
          "samples_seen": 3840,
          "online_loss": 0.026903835823759437,
          "train_loss": 0.024096980690956116,
          "validation_loss": 0.017812691628932953,
          "train_metric": 0.0023367407266050577,
          "validation_metric": 0.0017273386474698782,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07391328499943484
        },
        {
          "epoch": 6,
          "updates": 72,
          "samples_seen": 4608,
          "online_loss": 0.025824298655303817,
          "train_loss": 0.024137111380696297,
          "validation_loss": 0.021672656759619713,
          "train_metric": 0.002340632490813732,
          "validation_metric": 0.0021016483660787344,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.086971265000102
        },
        {
          "epoch": 7,
          "updates": 84,
          "samples_seen": 5376,
          "online_loss": 0.02777112027009328,
          "train_loss": 0.02187364362180233,
          "validation_loss": 0.018283039331436157,
          "train_metric": 0.002121138619259,
          "validation_metric": 0.0017729493556544185,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10010796200003824
        },
        {
          "epoch": 8,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.023254066007211804,
          "train_loss": 0.01665104739367962,
          "validation_loss": 0.013666925020515919,
          "train_metric": 0.0016146908747032285,
          "validation_metric": 0.0013253139331936836,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.11397699400004058
        },
        {
          "epoch": 9,
          "updates": 108,
          "samples_seen": 6912,
          "online_loss": 0.024514219102760155,
          "train_loss": 0.018493181094527245,
          "validation_loss": 0.015074401162564754,
          "train_metric": 0.0017933272756636143,
          "validation_metric": 0.0014618000714108348,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.12657364299957408
        },
        {
          "epoch": 10,
          "updates": 120,
          "samples_seen": 7680,
          "online_loss": 0.02864822947109739,
          "train_loss": 0.016237718984484673,
          "validation_loss": 0.014288689009845257,
          "train_metric": 0.0015746097778901458,
          "validation_metric": 0.0013856076402589679,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1391871319992788
        },
        {
          "epoch": 11,
          "updates": 132,
          "samples_seen": 8448,
          "online_loss": 0.023393176961690187,
          "train_loss": 0.024054594337940216,
          "validation_loss": 0.019684527069330215,
          "train_metric": 0.002332630567252636,
          "validation_metric": 0.0019088545814156532,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2363100399998075
        },
        {
          "epoch": 12,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.019519270087281864,
          "train_loss": 0.015371005982160568,
          "validation_loss": 0.019079474732279778,
          "train_metric": 0.001490562572143972,
          "validation_metric": 0.001850181259214878,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24962929999855987
        },
        {
          "epoch": 13,
          "updates": 156,
          "samples_seen": 9984,
          "online_loss": 0.015608443529345095,
          "train_loss": 0.022176846861839294,
          "validation_loss": 0.018709391355514526,
          "train_metric": 0.002150540938600898,
          "validation_metric": 0.0018142935587093234,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.262706776999039
        },
        {
          "epoch": 14,
          "updates": 168,
          "samples_seen": 10752,
          "online_loss": 0.016685141056465607,
          "train_loss": 0.02102743834257126,
          "validation_loss": 0.02358662150800228,
          "train_metric": 0.0020390800200402737,
          "validation_metric": 0.002287250244989991,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2760037049993116
        },
        {
          "epoch": 15,
          "updates": 180,
          "samples_seen": 11520,
          "online_loss": 0.017610900026435655,
          "train_loss": 0.00959115382283926,
          "validation_loss": 0.008525596931576729,
          "train_metric": 0.0009300767560489476,
          "validation_metric": 0.0008267472730949521,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2903322339989245
        },
        {
          "epoch": 16,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.013167609654677411,
          "train_loss": 0.010257936082780361,
          "validation_loss": 0.008794249035418034,
          "train_metric": 0.0009947360958904028,
          "validation_metric": 0.00085279910126701,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3040274329996464
        },
        {
          "epoch": 17,
          "updates": 204,
          "samples_seen": 13056,
          "online_loss": 0.012862154243824383,
          "train_loss": 0.02518528141081333,
          "validation_loss": 0.02846336178481579,
          "train_metric": 0.0024422763381153345,
          "validation_metric": 0.002760159084573388,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3168775670001196
        },
        {
          "epoch": 18,
          "updates": 216,
          "samples_seen": 13824,
          "online_loss": 0.03809180776200568,
          "train_loss": 0.025959523394703865,
          "validation_loss": 0.022583402693271637,
          "train_metric": 0.0025173560716211796,
          "validation_metric": 0.0021899656858295202,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.329414177000217
        },
        {
          "epoch": 19,
          "updates": 228,
          "samples_seen": 14592,
          "online_loss": 0.017926283568764727,
          "train_loss": 0.014807590283453465,
          "validation_loss": 0.013446171768009663,
          "train_metric": 0.0014359267661347985,
          "validation_metric": 0.0013039067853242159,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.342087625998829
        },
        {
          "epoch": 20,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.01589225814677775,
          "train_loss": 0.011337365955114365,
          "validation_loss": 0.009379109367728233,
          "train_metric": 0.0010994108160957694,
          "validation_metric": 0.0009095143177546561,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3540878530002374
        },
        {
          "epoch": 21,
          "updates": 252,
          "samples_seen": 16128,
          "online_loss": 0.014422519326520463,
          "train_loss": 0.016910897567868233,
          "validation_loss": 0.016507305204868317,
          "train_metric": 0.0016398894367739558,
          "validation_metric": 0.001600752235390246,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3663216309996642
        },
        {
          "epoch": 22,
          "updates": 264,
          "samples_seen": 16896,
          "online_loss": 0.01116227872747307,
          "train_loss": 0.011010355316102505,
          "validation_loss": 0.009668802842497826,
          "train_metric": 0.0010677000973373652,
          "validation_metric": 0.0009376064408570528,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3786790659996768
        },
        {
          "epoch": 23,
          "updates": 276,
          "samples_seen": 17664,
          "online_loss": 0.008756835954651857,
          "train_loss": 0.005679578986018896,
          "validation_loss": 0.006653133314102888,
          "train_metric": 0.0005507621099241078,
          "validation_metric": 0.0006451699882745743,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3913184349985386
        },
        {
          "epoch": 24,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.014709203775661686,
          "train_loss": 0.01250026747584343,
          "validation_loss": 0.015631068497896194,
          "train_metric": 0.001212180475704372,
          "validation_metric": 0.0015157817397266626,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4039996869996685
        },
        {
          "epoch": 25,
          "updates": 300,
          "samples_seen": 19200,
          "online_loss": 0.016998211503960192,
          "train_loss": 0.007794936653226614,
          "validation_loss": 0.00684234919026494,
          "train_metric": 0.000755893241148442,
          "validation_metric": 0.0006635186728090048,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41601335799896333
        },
        {
          "epoch": 26,
          "updates": 312,
          "samples_seen": 19968,
          "online_loss": 0.010065267095342278,
          "train_loss": 0.00705099618062377,
          "validation_loss": 0.008078569546341896,
          "train_metric": 0.000683751713950187,
          "validation_metric": 0.0007833978161215782,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.427772486998947
        },
        {
          "epoch": 27,
          "updates": 324,
          "samples_seen": 20736,
          "online_loss": 0.006216007634066045,
          "train_loss": 0.007955552078783512,
          "validation_loss": 0.0074818735010921955,
          "train_metric": 0.0007714685052633286,
          "validation_metric": 0.0007255348609760404,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4401301789985155
        },
        {
          "epoch": 28,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.019594372832216322,
          "train_loss": 0.0347214974462986,
          "validation_loss": 0.031855542212724686,
          "train_metric": 0.0033670254051685333,
          "validation_metric": 0.003089107107371092,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.451813506999315
        },
        {
          "epoch": 29,
          "updates": 348,
          "samples_seen": 22272,
          "online_loss": 0.026691250658283632,
          "train_loss": 0.02383790910243988,
          "validation_loss": 0.022574584931135178,
          "train_metric": 0.002311618300154805,
          "validation_metric": 0.002189110731706023,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4636400279996451
        },
        {
          "epoch": 30,
          "updates": 360,
          "samples_seen": 23040,
          "online_loss": 0.016825956486475963,
          "train_loss": 0.015650125220417976,
          "validation_loss": 0.01328558661043644,
          "train_metric": 0.001517629250884056,
          "validation_metric": 0.0012883343733847141,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4760232079988782
        },
        {
          "epoch": 31,
          "updates": 372,
          "samples_seen": 23808,
          "online_loss": 0.01750721347828706,
          "train_loss": 0.02119598351418972,
          "validation_loss": 0.0217406265437603,
          "train_metric": 0.0020554240327328444,
          "validation_metric": 0.002108239568769932,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4885067059985886
        },
        {
          "epoch": 32,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.014690780236075321,
          "train_loss": 0.009121040813624859,
          "validation_loss": 0.009419845417141914,
          "train_metric": 0.0008844888652674854,
          "validation_metric": 0.0009134645806625485,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.500804885999969
        },
        {
          "epoch": 33,
          "updates": 396,
          "samples_seen": 25344,
          "online_loss": 0.011931013374123722,
          "train_loss": 0.007728725206106901,
          "validation_loss": 0.008461584337055683,
          "train_metric": 0.0007494727615267038,
          "validation_metric": 0.0008205397753044963,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5131859590001113
        },
        {
          "epoch": 34,
          "updates": 408,
          "samples_seen": 26112,
          "online_loss": 0.007524920782695214,
          "train_loss": 0.01615307480096817,
          "validation_loss": 0.013480676338076591,
          "train_metric": 0.001566401682794094,
          "validation_metric": 0.0013072526780888438,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.524567690999902
        },
        {
          "epoch": 35,
          "updates": 420,
          "samples_seen": 26880,
          "online_loss": 0.007177371667542805,
          "train_loss": 0.012361480854451656,
          "validation_loss": 0.009695271030068398,
          "train_metric": 0.0011987219331786036,
          "validation_metric": 0.0009401732822880149,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5335618219996832
        },
        {
          "epoch": 36,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.0068014582502655685,
          "train_loss": 0.004277374129742384,
          "validation_loss": 0.004799574613571167,
          "train_metric": 0.0004147870640736073,
          "validation_metric": 0.0004654260410461575,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5431164159999753
        },
        {
          "epoch": 37,
          "updates": 444,
          "samples_seen": 28416,
          "online_loss": 0.005687303511270632,
          "train_loss": 0.005223054904490709,
          "validation_loss": 0.006365083158016205,
          "train_metric": 0.0005064919241704047,
          "validation_metric": 0.0006172371795400977,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5521718729996792
        },
        {
          "epoch": 38,
          "updates": 456,
          "samples_seen": 29184,
          "online_loss": 0.006040199630660936,
          "train_loss": 0.003740445477887988,
          "validation_loss": 0.004320237319916487,
          "train_metric": 0.0003627197875175625,
          "validation_metric": 0.0004189435567241162,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5615531349994853
        },
        {
          "epoch": 39,
          "updates": 468,
          "samples_seen": 29952,
          "online_loss": 0.005634621493906404,
          "train_loss": 0.006491262000054121,
          "validation_loss": 0.005237582605332136,
          "train_metric": 0.0006294729537330568,
          "validation_metric": 0.0005079007241874933,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5705731689995446
        },
        {
          "epoch": 40,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.004086642671609297,
          "train_loss": 0.0040804436430335045,
          "validation_loss": 0.004347433336079121,
          "train_metric": 0.0003956901782657951,
          "validation_metric": 0.00042158080032095313,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.579322820998641
        },
        {
          "epoch": 41,
          "updates": 492,
          "samples_seen": 31488,
          "online_loss": 0.0038403300810993337,
          "train_loss": 0.005962299648672342,
          "validation_loss": 0.007003446109592915,
          "train_metric": 0.0005781781510449946,
          "validation_metric": 0.0006791406776756048,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5881845530002465
        },
        {
          "epoch": 42,
          "updates": 504,
          "samples_seen": 32256,
          "online_loss": 0.004211619748578717,
          "train_loss": 0.004014142323285341,
          "validation_loss": 0.005231286399066448,
          "train_metric": 0.00038926079287193716,
          "validation_metric": 0.000507290125824511,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5971698159992229
        },
        {
          "epoch": 43,
          "updates": 516,
          "samples_seen": 33024,
          "online_loss": 0.011961099633481354,
          "train_loss": 0.00611898535862565,
          "validation_loss": 0.005877905525267124,
          "train_metric": 0.0005933723878115416,
          "validation_metric": 0.0005699943867512047,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6060894359998201
        },
        {
          "epoch": 44,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.007012881833361462,
          "train_loss": 0.004585783462971449,
          "validation_loss": 0.006082321051508188,
          "train_metric": 0.0004446942184586078,
          "validation_metric": 0.0005898170056752861,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6151668849997805
        },
        {
          "epoch": 45,
          "updates": 540,
          "samples_seen": 34560,
          "online_loss": 0.007694741342371951,
          "train_loss": 0.004565877839922905,
          "validation_loss": 0.0051443129777908325,
          "train_metric": 0.00044276390690356493,
          "validation_metric": 0.0004988561267964542,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.623776590999114
        },
        {
          "epoch": 46,
          "updates": 552,
          "samples_seen": 35328,
          "online_loss": 0.006851335191943993,
          "train_loss": 0.004872740246355534,
          "validation_loss": 0.005231840535998344,
          "train_metric": 0.0004725210601463914,
          "validation_metric": 0.0005073438514955342,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6326956079992669
        },
        {
          "epoch": 47,
          "updates": 564,
          "samples_seen": 36096,
          "online_loss": 0.005948746557502697,
          "train_loss": 0.0032231684308499098,
          "validation_loss": 0.0034720709081739187,
          "train_metric": 0.00031255820067599416,
          "validation_metric": 0.0003366948803886771,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6418543130002945
        },
        {
          "epoch": 48,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.0030182625147669264,
          "train_loss": 0.00549061456695199,
          "validation_loss": 0.005412179045379162,
          "train_metric": 0.0005324378143996,
          "validation_metric": 0.0005248317611403763,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6506675459986582
        },
        {
          "epoch": 49,
          "updates": 588,
          "samples_seen": 37632,
          "online_loss": 0.004435097983029361,
          "train_loss": 0.003988981246948242,
          "validation_loss": 0.004735975060611963,
          "train_metric": 0.00038682090234942734,
          "validation_metric": 0.0004592586192302406,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6592759769991972
        },
        {
          "epoch": 50,
          "updates": 600,
          "samples_seen": 38400,
          "online_loss": 0.006563680435647257,
          "train_loss": 0.004198857583105564,
          "validation_loss": 0.0050694989040493965,
          "train_metric": 0.0004071730945724994,
          "validation_metric": 0.0004916011821478605,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6680972059984924
        },
        {
          "epoch": 51,
          "updates": 612,
          "samples_seen": 39168,
          "online_loss": 0.003876876192710673,
          "train_loss": 0.003438076935708523,
          "validation_loss": 0.002914109267294407,
          "train_metric": 0.00033339837682433426,
          "validation_metric": 0.00028258803649805486,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6773638969989406
        },
        {
          "epoch": 52,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.003393791198808079,
          "train_loss": 0.0035884513054043055,
          "validation_loss": 0.0044036428444087505,
          "train_metric": 0.0003479805018287152,
          "validation_metric": 0.00042703162762336433,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6861478589999024
        },
        {
          "epoch": 53,
          "updates": 636,
          "samples_seen": 40704,
          "online_loss": 0.0032623732986394316,
          "train_loss": 0.0037320435512810946,
          "validation_loss": 0.0035389030817896128,
          "train_metric": 0.0003619050548877567,
          "validation_metric": 0.0003431757795624435,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6948197530000471
        },
        {
          "epoch": 54,
          "updates": 648,
          "samples_seen": 41472,
          "online_loss": 0.008034991507884115,
          "train_loss": 0.010650355368852615,
          "validation_loss": 0.013361840508878231,
          "train_metric": 0.001032790052704513,
          "validation_metric": 0.001295729074627161,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7033600579998165
        },
        {
          "epoch": 55,
          "updates": 660,
          "samples_seen": 42240,
          "online_loss": 0.008265311693927893,
          "train_loss": 0.007518466096371412,
          "validation_loss": 0.010024549439549446,
          "train_metric": 0.0007290833164006472,
          "validation_metric": 0.0009721041424199939,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7120411369996873
        },
        {
          "epoch": 56,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.004119166444676618,
          "train_loss": 0.0029487276915460825,
          "validation_loss": 0.0033481973223388195,
          "train_metric": 0.00028594498871825635,
          "validation_metric": 0.0003246825945097953,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7206033969996497
        },
        {
          "epoch": 57,
          "updates": 684,
          "samples_seen": 43776,
          "online_loss": 0.002710560593792858,
          "train_loss": 0.0020056224893778563,
          "validation_loss": 0.002827393589541316,
          "train_metric": 0.00019448988314252347,
          "validation_metric": 0.00027417903766036034,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7301083509992168
        },
        {
          "epoch": 58,
          "updates": 696,
          "samples_seen": 44544,
          "online_loss": 0.0024318137972538048,
          "train_loss": 0.001964898081496358,
          "validation_loss": 0.0030390513129532337,
          "train_metric": 0.0001905407989397645,
          "validation_metric": 0.00029470393201336265,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7390682400000514
        },
        {
          "epoch": 59,
          "updates": 708,
          "samples_seen": 45312,
          "online_loss": 0.006196321036744242,
          "train_loss": 0.031439170241355896,
          "validation_loss": 0.024229299277067184,
          "train_metric": 0.0030487303156405687,
          "validation_metric": 0.0023495720233768225,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7480484059997252
        },
        {
          "epoch": 60,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.014706009843697151,
          "train_loss": 0.0074784234166145325,
          "validation_loss": 0.006602554582059383,
          "train_metric": 0.0007252002251334488,
          "validation_metric": 0.0006402653525583446,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7567885539992858
        },
        {
          "epoch": 61,
          "updates": 732,
          "samples_seen": 46848,
          "online_loss": 0.0053559984371531755,
          "train_loss": 0.004610145930200815,
          "validation_loss": 0.004642090760171413,
          "train_metric": 0.0004470566345844418,
          "validation_metric": 0.00045015450450591743,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7655969669995102
        },
        {
          "epoch": 62,
          "updates": 744,
          "samples_seen": 47616,
          "online_loss": 0.00544705677505893,
          "train_loss": 0.0041727772913873196,
          "validation_loss": 0.0033050612546503544,
          "train_metric": 0.0004046440590173006,
          "validation_metric": 0.00032049958826974034,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7743210749995342
        },
        {
          "epoch": 63,
          "updates": 756,
          "samples_seen": 48384,
          "online_loss": 0.009797165868803859,
          "train_loss": 0.006461258511990309,
          "validation_loss": 0.0073476010002195835,
          "train_metric": 0.000626563502009958,
          "validation_metric": 0.0007125141564756632,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7831811079995532
        },
        {
          "epoch": 64,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.004941670660628006,
          "train_loss": 0.003159311367198825,
          "validation_loss": 0.0038617567624896765,
          "train_metric": 0.00030636583687737584,
          "validation_metric": 0.00037448364309966564,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7918546499986405
        },
        {
          "epoch": 65,
          "updates": 780,
          "samples_seen": 49920,
          "online_loss": 0.0032159942493308336,
          "train_loss": 0.0022203889675438404,
          "validation_loss": 0.0024744917172938585,
          "train_metric": 0.00021531636593863368,
          "validation_metric": 0.00023995728406589478,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8014024629992491
        },
        {
          "epoch": 66,
          "updates": 792,
          "samples_seen": 50688,
          "online_loss": 0.002441603661281988,
          "train_loss": 0.0024428176693618298,
          "validation_loss": 0.003056779270991683,
          "train_metric": 0.00023688579676672816,
          "validation_metric": 0.00029642306617461145,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8102626429990778
        },
        {
          "epoch": 67,
          "updates": 804,
          "samples_seen": 51456,
          "online_loss": 0.0026724563989167414,
          "train_loss": 0.00256863865070045,
          "validation_loss": 0.002401223173364997,
          "train_metric": 0.00024908690829761326,
          "validation_metric": 0.00023285226779989898,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8196813159993326
        },
        {
          "epoch": 68,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.004229429929788846,
          "train_loss": 0.0030211296398192644,
          "validation_loss": 0.004369927570223808,
          "train_metric": 0.0002929660549852997,
          "validation_metric": 0.0004237621324136853,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.828826412998751
        },
        {
          "epoch": 69,
          "updates": 828,
          "samples_seen": 52992,
          "online_loss": 0.004892203394168367,
          "train_loss": 0.005909258034080267,
          "validation_loss": 0.004303823225200176,
          "train_metric": 0.000573034689296037,
          "validation_metric": 0.000417351839132607,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8374691949993576
        },
        {
          "epoch": 70,
          "updates": 840,
          "samples_seen": 53760,
          "online_loss": 0.004949509015811297,
          "train_loss": 0.005550771951675415,
          "validation_loss": 0.0061132158152759075,
          "train_metric": 0.0005382714443840086,
          "validation_metric": 0.0005928129539825022,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8463440539999283
        },
        {
          "epoch": 71,
          "updates": 852,
          "samples_seen": 54528,
          "online_loss": 0.0031401295612643785,
          "train_loss": 0.002523032482713461,
          "validation_loss": 0.0036811884492635727,
          "train_metric": 0.00024466440663672984,
          "validation_metric": 0.00035697349812835455,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8551863949996914
        },
        {
          "epoch": 72,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0022878074378240854,
          "train_loss": 0.0017730769468471408,
          "validation_loss": 0.002539434004575014,
          "train_metric": 0.00017193944950122386,
          "validation_metric": 0.0002462548727635294,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8642661449994193
        },
        {
          "epoch": 73,
          "updates": 876,
          "samples_seen": 56064,
          "online_loss": 0.002008546425107246,
          "train_loss": 0.0025046791415661573,
          "validation_loss": 0.0033342319075018167,
          "train_metric": 0.0002428846200928092,
          "validation_metric": 0.0003233283059671521,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8728525569986232
        },
        {
          "epoch": 74,
          "updates": 888,
          "samples_seen": 56832,
          "online_loss": 0.008393679735794043,
          "train_loss": 0.005218779668211937,
          "validation_loss": 0.0066897934302687645,
          "train_metric": 0.000506077311001718,
          "validation_metric": 0.0006487249629572034,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8816385979989718
        },
        {
          "epoch": 75,
          "updates": 900,
          "samples_seen": 57600,
          "online_loss": 0.004004422400612384,
          "train_loss": 0.0023751321714371443,
          "validation_loss": 0.0025608635041862726,
          "train_metric": 0.00023032212629914284,
          "validation_metric": 0.0002483329735696316,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8903943099994649
        },
        {
          "epoch": 76,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.0035383511373462775,
          "train_loss": 0.004982635844498873,
          "validation_loss": 0.004687522538006306,
          "train_metric": 0.00048317795153707266,
          "validation_metric": 0.00045456006773747504,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8993519179985014
        },
        {
          "epoch": 77,
          "updates": 924,
          "samples_seen": 59136,
          "online_loss": 0.00852877602543837,
          "train_loss": 0.002477366477251053,
          "validation_loss": 0.003408300457522273,
          "train_metric": 0.00024023604055400938,
          "validation_metric": 0.00033051095670089126,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9080687730001955
        },
        {
          "epoch": 78,
          "updates": 936,
          "samples_seen": 59904,
          "online_loss": 0.008560598274925724,
          "train_loss": 0.011688685975968838,
          "validation_loss": 0.009780239313840866,
          "train_metric": 0.0011334792943671346,
          "validation_metric": 0.0009484128095209599,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9169707159999234
        },
        {
          "epoch": 79,
          "updates": 948,
          "samples_seen": 60672,
          "online_loss": 0.009230770859479284,
          "train_loss": 0.004604403395205736,
          "validation_loss": 0.005357181653380394,
          "train_metric": 0.00044649982010014355,
          "validation_metric": 0.000519498426001519,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9255680919995939
        },
        {
          "epoch": 80,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.004366299438212688,
          "train_loss": 0.0034405142068862915,
          "validation_loss": 0.0036006539594382048,
          "train_metric": 0.00033363481634296477,
          "validation_metric": 0.00034916389267891645,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9344272979997186
        },
        {
          "epoch": 81,
          "updates": 972,
          "samples_seen": 62208,
          "online_loss": 0.0031679988072331375,
          "train_loss": 0.002258377382531762,
          "validation_loss": 0.0025307456962764263,
          "train_metric": 0.00021900016872677952,
          "validation_metric": 0.00024541240418329835,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9430787969995436
        },
        {
          "epoch": 82,
          "updates": 984,
          "samples_seen": 62976,
          "online_loss": 0.002112242946168408,
          "train_loss": 0.003192078322172165,
          "validation_loss": 0.004114796873182058,
          "train_metric": 0.00030954336398281157,
          "validation_metric": 0.0003990215191151947,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9520022189990414
        },
        {
          "epoch": 83,
          "updates": 996,
          "samples_seen": 63744,
          "online_loss": 0.0024978475994430482,
          "train_loss": 0.0018271369626745582,
          "validation_loss": 0.0023278449662029743,
          "train_metric": 0.00017718173330649734,
          "validation_metric": 0.00022573659953195602,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9613417029995617
        },
        {
          "epoch": 84,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.0028018373996019363,
          "train_loss": 0.0038681570440530777,
          "validation_loss": 0.004141743760555983,
          "train_metric": 0.0003751043404918164,
          "validation_metric": 0.00040163463563658297,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9702884559992526
        },
        {
          "epoch": 85,
          "updates": 1020,
          "samples_seen": 65280,
          "online_loss": 0.0036701376763327667,
          "train_loss": 0.0025298420805484056,
          "validation_loss": 0.003364563686773181,
          "train_metric": 0.00024532474344596267,
          "validation_metric": 0.00032626965548843145,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9791565999985323
        },
        {
          "epoch": 86,
          "updates": 1032,
          "samples_seen": 66048,
          "online_loss": 0.0027346254889077195,
          "train_loss": 0.003156165825203061,
          "validation_loss": 0.003624719101935625,
          "train_metric": 0.0003060607996303588,
          "validation_metric": 0.00035149758332408965,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9877856819985027
        },
        {
          "epoch": 87,
          "updates": 1044,
          "samples_seen": 66816,
          "online_loss": 0.0018795519523943465,
          "train_loss": 0.0016854549758136272,
          "validation_loss": 0.0023473373148590326,
          "train_metric": 0.00016344255709555,
          "validation_metric": 0.0002276268060086295,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9967524809999304
        },
        {
          "epoch": 88,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.001877006938836227,
          "train_loss": 0.002386544132605195,
          "validation_loss": 0.0030499990098178387,
          "train_metric": 0.0002314287849003449,
          "validation_metric": 0.0002957655815407634,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0053911649993097
        },
        {
          "epoch": 89,
          "updates": 1068,
          "samples_seen": 68352,
          "online_loss": 0.0022857370434697564,
          "train_loss": 0.0018952401587739587,
          "validation_loss": 0.0025768799241632223,
          "train_metric": 0.00018378591630607843,
          "validation_metric": 0.00024988609948195517,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0142784269992262
        },
        {
          "epoch": 90,
          "updates": 1080,
          "samples_seen": 69120,
          "online_loss": 0.002535228525327208,
          "train_loss": 0.0035303488839417696,
          "validation_loss": 0.0038264666218310595,
          "train_metric": 0.00034234626218676567,
          "validation_metric": 0.00037106149829924107,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0227540939995379
        },
        {
          "epoch": 91,
          "updates": 1092,
          "samples_seen": 69888,
          "online_loss": 0.0021511318651998104,
          "train_loss": 0.003306981176137924,
          "validation_loss": 0.003521275706589222,
          "train_metric": 0.00032068570726551116,
          "validation_metric": 0.0003414663951843977,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0317009760001383
        },
        {
          "epoch": 92,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.002017543842763795,
          "train_loss": 0.001789654023014009,
          "validation_loss": 0.0022580339573323727,
          "train_metric": 0.00017354694136884063,
          "validation_metric": 0.00021896688849665225,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0409543989990198
        },
        {
          "epoch": 93,
          "updates": 1116,
          "samples_seen": 71424,
          "online_loss": 0.0032891271378806173,
          "train_loss": 0.0014185700565576553,
          "validation_loss": 0.0024707941338419914,
          "train_metric": 0.00013756206317339092,
          "validation_metric": 0.0002395987685304135,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.049794139998994
        },
        {
          "epoch": 94,
          "updates": 1128,
          "samples_seen": 72192,
          "online_loss": 0.008546398845889295,
          "train_loss": 0.002935129916295409,
          "validation_loss": 0.0043343473225831985,
          "train_metric": 0.0002846264105755836,
          "validation_metric": 0.00042031187331303954,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0585707259997434
        },
        {
          "epoch": 95,
          "updates": 1140,
          "samples_seen": 72960,
          "online_loss": 0.005715192062780261,
          "train_loss": 0.0019572994206100702,
          "validation_loss": 0.003032069653272629,
          "train_metric": 0.0001898038899526,
          "validation_metric": 0.0002940269187092781,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.067512617999455
        },
        {
          "epoch": 96,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0018539197917561978,
          "train_loss": 0.002797790803015232,
          "validation_loss": 0.002869995543733239,
          "train_metric": 0.0002713083813432604,
          "validation_metric": 0.0002783101808745414,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.076177576000191
        },
        {
          "epoch": 97,
          "updates": 1164,
          "samples_seen": 74496,
          "online_loss": 0.0038266731814170876,
          "train_loss": 0.004266688134521246,
          "validation_loss": 0.004040713422000408,
          "train_metric": 0.0004137507639825344,
          "validation_metric": 0.0003918374713975936,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0849426459990354
        },
        {
          "epoch": 98,
          "updates": 1176,
          "samples_seen": 75264,
          "online_loss": 0.0021164588930939012,
          "train_loss": 0.0014802945079281926,
          "validation_loss": 0.002701780991628766,
          "train_metric": 0.00014354765880852938,
          "validation_metric": 0.0002619980659801513,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.093634707000092
        },
        {
          "epoch": 99,
          "updates": 1188,
          "samples_seen": 76032,
          "online_loss": 0.0018656378694383118,
          "train_loss": 0.0013211170444265008,
          "validation_loss": 0.0019616647623479366,
          "train_metric": 0.00012811183114536107,
          "validation_metric": 0.00019022723427042365,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1030657829996926
        },
        {
          "epoch": 100,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.001515732285042759,
          "train_loss": 0.001388448872603476,
          "validation_loss": 0.002204256597906351,
          "train_metric": 0.00013464115909300745,
          "validation_metric": 0.0002137519622920081,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1117865319993143
        },
        {
          "epoch": 101,
          "updates": 1212,
          "samples_seen": 77568,
          "online_loss": 0.0018741773625758167,
          "train_loss": 0.0020106008742004633,
          "validation_loss": 0.002880915068089962,
          "train_metric": 0.00019497268658597022,
          "validation_metric": 0.00027936912374570966,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.120321499000056
        },
        {
          "epoch": 102,
          "updates": 1224,
          "samples_seen": 78336,
          "online_loss": 0.0019072198144082602,
          "train_loss": 0.0010784048354253173,
          "validation_loss": 0.0020109955221414566,
          "train_metric": 0.00010457545431563631,
          "validation_metric": 0.00019501095812302083,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1290590209991933
        },
        {
          "epoch": 103,
          "updates": 1236,
          "samples_seen": 79104,
          "online_loss": 0.002429688383320657,
          "train_loss": 0.001825698302127421,
          "validation_loss": 0.0030187892261892557,
          "train_metric": 0.0001770422823028639,
          "validation_metric": 0.00029273907421156764,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1375508309993165
        },
        {
          "epoch": 104,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.003182121601033335,
          "train_loss": 0.008067446760833263,
          "validation_loss": 0.0064222607761621475,
          "train_metric": 0.0007823192863725126,
          "validation_metric": 0.0006227816920727491,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1463957219984877
        },
        {
          "epoch": 105,
          "updates": 1260,
          "samples_seen": 80640,
          "online_loss": 0.006934160483069718,
          "train_loss": 0.009573355317115784,
          "validation_loss": 0.00856721680611372,
          "train_metric": 0.0009283507242798805,
          "validation_metric": 0.0008307831594720483,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1552289409992227
        },
        {
          "epoch": 106,
          "updates": 1272,
          "samples_seen": 81408,
          "online_loss": 0.014771038084290922,
          "train_loss": 0.008449922315776348,
          "validation_loss": 0.01208472065627575,
          "train_metric": 0.0008194088004529476,
          "validation_metric": 0.0011718837777152658,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1642370509998727
        },
        {
          "epoch": 107,
          "updates": 1284,
          "samples_seen": 82176,
          "online_loss": 0.006293876135411362,
          "train_loss": 0.0029367778915911913,
          "validation_loss": 0.003145145485177636,
          "train_metric": 0.0002847862197086215,
          "validation_metric": 0.000304992136079818,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1727464649993635
        },
        {
          "epoch": 108,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.0031976084186074636,
          "train_loss": 0.00257515674456954,
          "validation_loss": 0.003250878071412444,
          "train_metric": 0.0002497189852874726,
          "validation_metric": 0.000315245270030573,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1815663420002238
        },
        {
          "epoch": 109,
          "updates": 1308,
          "samples_seen": 83712,
          "online_loss": 0.0027098086332747093,
          "train_loss": 0.001768671558238566,
          "validation_loss": 0.002419746946543455,
          "train_metric": 0.000171512205270119,
          "validation_metric": 0.00023464854166377336,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1900949470000342
        },
        {
          "epoch": 110,
          "updates": 1320,
          "samples_seen": 84480,
          "online_loss": 0.002124264559824951,
          "train_loss": 0.002581943990662694,
          "validation_loss": 0.0026127893943339586,
          "train_metric": 0.00025037716841325164,
          "validation_metric": 0.00025336831458844244,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1988689720001275
        },
        {
          "epoch": 111,
          "updates": 1332,
          "samples_seen": 85248,
          "online_loss": 0.002416476466654179,
          "train_loss": 0.0013532597804442048,
          "validation_loss": 0.001991728087887168,
          "train_metric": 0.0001312287786277011,
          "validation_metric": 0.00019314256496727467,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2074320419997093
        },
        {
          "epoch": 112,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.002854122295199583,
          "train_loss": 0.0027605444192886353,
          "validation_loss": 0.0034177997149527073,
          "train_metric": 0.0002676965086720884,
          "validation_metric": 0.0003314320638310164,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2162749049984996
        },
        {
          "epoch": 113,
          "updates": 1356,
          "samples_seen": 86784,
          "online_loss": 0.002362228241205836,
          "train_loss": 0.0017500288086012006,
          "validation_loss": 0.002620877232402563,
          "train_metric": 0.000169704420841299,
          "validation_metric": 0.0002541526046115905,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2251853870002378
        },
        {
          "epoch": 114,
          "updates": 1368,
          "samples_seen": 87552,
          "online_loss": 0.001845449010337082,
          "train_loss": 0.0013218718813732266,
          "validation_loss": 0.002083045896142721,
          "train_metric": 0.00012818504183087498,
          "validation_metric": 0.0002019978710450232,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2341453949993593
        },
        {
          "epoch": 115,
          "updates": 1380,
          "samples_seen": 88320,
          "online_loss": 0.0015416380944467771,
          "train_loss": 0.0011728736571967602,
          "validation_loss": 0.00183727010153234,
          "train_metric": 0.00011373631423339248,
          "validation_metric": 0.0001781643950380385,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2433513959986158
        },
        {
          "epoch": 116,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0011269722890574485,
          "train_loss": 0.0012573885032907128,
          "validation_loss": 0.002032689517363906,
          "train_metric": 0.00012193193106213585,
          "validation_metric": 0.00019711469940375537,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.252211905999502
        },
        {
          "epoch": 117,
          "updates": 1404,
          "samples_seen": 89856,
          "online_loss": 0.0014185532393942897,
          "train_loss": 0.0010762239107862115,
          "validation_loss": 0.0018364309798926115,
          "train_metric": 0.00010436397133162245,
          "validation_metric": 0.0001780830352799967,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2613712049987953
        },
        {
          "epoch": 118,
          "updates": 1416,
          "samples_seen": 90624,
          "online_loss": 0.0011933474064183731,
          "train_loss": 0.0023666417691856623,
          "validation_loss": 0.0028631933964788914,
          "train_metric": 0.00022949882259126753,
          "validation_metric": 0.00027765060076490045,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2700549000001047
        },
        {
          "epoch": 119,
          "updates": 1428,
          "samples_seen": 91392,
          "online_loss": 0.0018337491322502804,
          "train_loss": 0.0014723940985277295,
          "validation_loss": 0.0021413895301520824,
          "train_metric": 0.0001427815150236711,
          "validation_metric": 0.0002076555829262361,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3403346439990855
        },
        {
          "epoch": 120,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.002419664978030293,
          "train_loss": 0.0030884428415447474,
          "validation_loss": 0.0027967384085059166,
          "train_metric": 0.00029949357849545777,
          "validation_metric": 0.00027120631420984864,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3491960419996758
        },
        {
          "epoch": 121,
          "updates": 1452,
          "samples_seen": 92928,
          "online_loss": 0.001955955294154895,
          "train_loss": 0.001583156525157392,
          "validation_loss": 0.0024679561611264944,
          "train_metric": 0.0001535224000690505,
          "validation_metric": 0.00023932350450195372,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3576240649999818
        },
        {
          "epoch": 122,
          "updates": 1464,
          "samples_seen": 93696,
          "online_loss": 0.001491563688129342,
          "train_loss": 0.0012136843288317323,
          "validation_loss": 0.0019435854628682137,
          "train_metric": 0.00011769382399506867,
          "validation_metric": 0.00018847403407562524,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3662589909999951
        },
        {
          "epoch": 123,
          "updates": 1476,
          "samples_seen": 94464,
          "online_loss": 0.001314627552346792,
          "train_loss": 0.001018908224068582,
          "validation_loss": 0.0019173134351149201,
          "train_metric": 9.880592551780865e-05,
          "validation_metric": 0.00018592638662084937,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3747540049989766
        },
        {
          "epoch": 124,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0010487203665737372,
          "train_loss": 0.001040746341459453,
          "validation_loss": 0.0018737027421593666,
          "train_metric": 0.00010092362208524719,
          "validation_metric": 0.0001816973672248423,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3833317890002945
        },
        {
          "epoch": 125,
          "updates": 1500,
          "samples_seen": 96000,
          "online_loss": 0.001028748765141548,
          "train_loss": 0.0010038751643151045,
          "validation_loss": 0.0019693709909915924,
          "train_metric": 9.734814375406131e-05,
          "validation_metric": 0.00019097451877314597,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3917247299996234
        },
        {
          "epoch": 126,
          "updates": 1512,
          "samples_seen": 96768,
          "online_loss": 0.0010775102685632494,
          "train_loss": 0.0012792149791494012,
          "validation_loss": 0.002154537010937929,
          "train_metric": 0.00012404848530422896,
          "validation_metric": 0.000208930519875139,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.400316588998976
        },
        {
          "epoch": 127,
          "updates": 1524,
          "samples_seen": 97536,
          "online_loss": 0.0012890245197922923,
          "train_loss": 0.00097424047999084,
          "validation_loss": 0.001970033161342144,
          "train_metric": 9.447438787901774e-05,
          "validation_metric": 0.00019103875092696398,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.408752026000002
        },
        {
          "epoch": 128,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0016443585288167621,
          "train_loss": 0.0010877068853005767,
          "validation_loss": 0.002029133029282093,
          "train_metric": 0.00010547748388489708,
          "validation_metric": 0.00019676980446092784,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4172987009987992
        },
        {
          "epoch": 129,
          "updates": 1548,
          "samples_seen": 99072,
          "online_loss": 0.0010309138500209276,
          "train_loss": 0.000898548576515168,
          "validation_loss": 0.001898170099593699,
          "train_metric": 8.713436545804143e-05,
          "validation_metric": 0.00018407002789899707,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4256728499985911
        },
        {
          "epoch": 130,
          "updates": 1560,
          "samples_seen": 99840,
          "online_loss": 0.0009131575498031452,
          "train_loss": 0.0008628915529698133,
          "validation_loss": 0.0018811107147485018,
          "train_metric": 8.367662667296827e-05,
          "validation_metric": 0.00018241572252009064,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.434503708000193
        },
        {
          "epoch": 131,
          "updates": 1572,
          "samples_seen": 100608,
          "online_loss": 0.0009605050484727448,
          "train_loss": 0.0009196228347718716,
          "validation_loss": 0.0018098040018230677,
          "train_metric": 8.917798550100997e-05,
          "validation_metric": 0.00017550094344187528,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.443492296999466
        },
        {
          "epoch": 132,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0008938777682487853,
          "train_loss": 0.0008525503217242658,
          "validation_loss": 0.001845165970735252,
          "train_metric": 8.267380326287821e-05,
          "validation_metric": 0.00017893011681735516,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.452150402999905
        },
        {
          "epoch": 133,
          "updates": 1596,
          "samples_seen": 102144,
          "online_loss": 0.0008747425769494536,
          "train_loss": 0.0008275905274786055,
          "validation_loss": 0.0018046863842755556,
          "train_metric": 8.025339775485918e-05,
          "validation_metric": 0.0001750046940287575,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.461169553000218
        },
        {
          "epoch": 134,
          "updates": 1608,
          "samples_seen": 102912,
          "online_loss": 0.000875832275293457,
          "train_loss": 0.0008211400709114969,
          "validation_loss": 0.0018348322482779622,
          "train_metric": 7.962788367876783e-05,
          "validation_metric": 0.00017792799917515367,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4698758899994573
        },
        {
          "epoch": 135,
          "updates": 1620,
          "samples_seen": 103680,
          "online_loss": 0.0008555346090967456,
          "train_loss": 0.0008124329033307731,
          "validation_loss": 0.001816204166971147,
          "train_metric": 7.878352334955707e-05,
          "validation_metric": 0.00017612158262636513,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.478484342000229
        },
        {
          "epoch": 136,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0008882233887561597,
          "train_loss": 0.0008748589898459613,
          "validation_loss": 0.0018013989320024848,
          "train_metric": 8.483712008455768e-05,
          "validation_metric": 0.00017468587611801922,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4874441100000695
        },
        {
          "epoch": 137,
          "updates": 1644,
          "samples_seen": 105216,
          "online_loss": 0.0008764027661527507,
          "train_loss": 0.0007989767473191023,
          "validation_loss": 0.0018563021440058947,
          "train_metric": 7.747865311102942e-05,
          "validation_metric": 0.00018000995623879135,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4961055379990285
        },
        {
          "epoch": 138,
          "updates": 1656,
          "samples_seen": 105984,
          "online_loss": 0.0008468620411197966,
          "train_loss": 0.0007998240180313587,
          "validation_loss": 0.0018331741448491812,
          "train_metric": 7.756082050036639e-05,
          "validation_metric": 0.00017776720051188022,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5045134629999666
        },
        {
          "epoch": 139,
          "updates": 1668,
          "samples_seen": 106752,
          "online_loss": 0.0009052258368077067,
          "train_loss": 0.0012433849042281508,
          "validation_loss": 0.0021664744708687067,
          "train_metric": 0.0001205739681608975,
          "validation_metric": 0.0002100881392834708,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.513049682998826
        },
        {
          "epoch": 140,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0009743516250940351,
          "train_loss": 0.0011761932400986552,
          "validation_loss": 0.002157545182853937,
          "train_metric": 0.00011405823897803202,
          "validation_metric": 0.00020922222756780684,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5214144740002666
        },
        {
          "epoch": 141,
          "updates": 1692,
          "samples_seen": 108288,
          "online_loss": 0.0009240408786960567,
          "train_loss": 0.0007942933007143438,
          "validation_loss": 0.0018363379640504718,
          "train_metric": 7.702448783675209e-05,
          "validation_metric": 0.0001780740130925551,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5299713379990862
        },
        {
          "epoch": 142,
          "updates": 1704,
          "samples_seen": 109056,
          "online_loss": 0.0008776430549914949,
          "train_loss": 0.0007832914707250893,
          "validation_loss": 0.0018825443694368005,
          "train_metric": 7.595761417178437e-05,
          "validation_metric": 0.00018255475151818246,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5383693650001078
        },
        {
          "epoch": 143,
          "updates": 1716,
          "samples_seen": 109824,
          "online_loss": 0.0008264871236557761,
          "train_loss": 0.0007725483155809343,
          "validation_loss": 0.001851251581683755,
          "train_metric": 7.491582800867036e-05,
          "validation_metric": 0.0001795202260836959,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5469512719992053
        },
        {
          "epoch": 144,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0007880241585856614,
          "train_loss": 0.0007790662930347025,
          "validation_loss": 0.0018452268559485674,
          "train_metric": 7.554789772257209e-05,
          "validation_metric": 0.0001789359812391922,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5552016219990037
        },
        {
          "epoch": 145,
          "updates": 1740,
          "samples_seen": 111360,
          "online_loss": 0.0008669625434170788,
          "train_loss": 0.0008416804485023022,
          "validation_loss": 0.0018899922724813223,
          "train_metric": 8.161973528331146e-05,
          "validation_metric": 0.00018327699217479676,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.563764832999368
        },
        {
          "epoch": 146,
          "updates": 1752,
          "samples_seen": 112128,
          "online_loss": 0.0007932912825102297,
          "train_loss": 0.0007403486524708569,
          "validation_loss": 0.0018701528897508979,
          "train_metric": 7.179336535045877e-05,
          "validation_metric": 0.00018135311256628484,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5720521559996996
        },
        {
          "epoch": 147,
          "updates": 1764,
          "samples_seen": 112896,
          "online_loss": 0.0008294261278933845,
          "train_loss": 0.0008483826532028615,
          "validation_loss": 0.0018839306430891156,
          "train_metric": 8.226966019719839e-05,
          "validation_metric": 0.00018268916755914688,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5807163949993992
        },
        {
          "epoch": 148,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.0009317469666711986,
          "train_loss": 0.0007292867521755397,
          "validation_loss": 0.001819215016439557,
          "train_metric": 7.072064909152687e-05,
          "validation_metric": 0.00017641355225350708,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.5890268579987605
        },
        {
          "epoch": 149,
          "updates": 1788,
          "samples_seen": 114432,
          "online_loss": 0.0007635130896233022,
          "train_loss": 0.0007266986067406833,
          "validation_loss": 0.0018015996320173144,
          "train_metric": 7.046967220958322e-05,
          "validation_metric": 0.00017470536113251,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.5976403829990886
        },
        {
          "epoch": 150,
          "updates": 1800,
          "samples_seen": 115200,
          "online_loss": 0.000755441093739743,
          "train_loss": 0.0007332779350690544,
          "validation_loss": 0.001784464344382286,
          "train_metric": 7.110767910489812e-05,
          "validation_metric": 0.00017304369248449802,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6065228850002313
        },
        {
          "epoch": 151,
          "updates": 1812,
          "samples_seen": 115968,
          "online_loss": 0.0007503637122378374,
          "train_loss": 0.0007254843949340284,
          "validation_loss": 0.0018526611384004354,
          "train_metric": 7.035192538751289e-05,
          "validation_metric": 0.0001796568976715207,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6152249249989836
        },
        {
          "epoch": 152,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.0008030267830084389,
          "train_loss": 0.0007850281544961035,
          "validation_loss": 0.0018758215010166168,
          "train_metric": 7.612602348672226e-05,
          "validation_metric": 0.00018190282571595162,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.623394616999576
        },
        {
          "epoch": 153,
          "updates": 1836,
          "samples_seen": 117504,
          "online_loss": 0.0007804220755739758,
          "train_loss": 0.0007469438132829964,
          "validation_loss": 0.0018553336849436164,
          "train_metric": 7.243290019687265e-05,
          "validation_metric": 0.00017991606728173792,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6320067149990791
        },
        {
          "epoch": 154,
          "updates": 1848,
          "samples_seen": 118272,
          "online_loss": 0.0007609379924057672,
          "train_loss": 0.0007141666137613356,
          "validation_loss": 0.0018493701936677098,
          "train_metric": 6.925441994098946e-05,
          "validation_metric": 0.00017933777417056262,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6404349479998928
        },
        {
          "epoch": 155,
          "updates": 1860,
          "samples_seen": 119040,
          "online_loss": 0.0007341985498593809,
          "train_loss": 0.0007080016075633466,
          "validation_loss": 0.0018372952472418547,
          "train_metric": 6.865656905574724e-05,
          "validation_metric": 0.00017816683975979686,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6488247309989674
        },
        {
          "epoch": 156,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.000750015560091318,
          "train_loss": 0.0007088083657436073,
          "validation_loss": 0.001843789592385292,
          "train_metric": 6.873480742797256e-05,
          "validation_metric": 0.00017879661754705012,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6569269229985366
        },
        {
          "epoch": 157,
          "updates": 1884,
          "samples_seen": 120576,
          "online_loss": 0.0008388965070480481,
          "train_loss": 0.000788176606874913,
          "validation_loss": 0.0018736420897766948,
          "train_metric": 7.643132994417101e-05,
          "validation_metric": 0.00018169148825109005,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6656328150002082
        },
        {
          "epoch": 158,
          "updates": 1896,
          "samples_seen": 121344,
          "online_loss": 0.0007832920115712719,
          "train_loss": 0.000739190960302949,
          "validation_loss": 0.001887176651507616,
          "train_metric": 7.168108277255669e-05,
          "validation_metric": 0.00018300396914128214,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6739164969985723
        },
        {
          "epoch": 159,
          "updates": 1908,
          "samples_seen": 122112,
          "online_loss": 0.0007977167139567124,
          "train_loss": 0.0007076705223880708,
          "validation_loss": 0.0018516841810196638,
          "train_metric": 6.862446753075346e-05,
          "validation_metric": 0.00017956216470338404,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6822716759998002
        },
        {
          "epoch": 160,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0007786821261106525,
          "train_loss": 0.0007170437020249665,
          "validation_loss": 0.0018366083968430758,
          "train_metric": 6.953341653570533e-05,
          "validation_metric": 0.00017810023564379662,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.6905235670001275
        },
        {
          "epoch": 161,
          "updates": 1932,
          "samples_seen": 123648,
          "online_loss": 0.0007129430183946776,
          "train_loss": 0.0007019462645985186,
          "validation_loss": 0.0018303284887224436,
          "train_metric": 6.806938472436741e-05,
          "validation_metric": 0.0001774912525434047,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.69891452399861
        },
        {
          "epoch": 162,
          "updates": 1944,
          "samples_seen": 124416,
          "online_loss": 0.0007227408156419793,
          "train_loss": 0.0006919806473888457,
          "validation_loss": 0.0018358879024162889,
          "train_metric": 6.71030065859668e-05,
          "validation_metric": 0.00017803037189878523,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.707239204999496
        },
        {
          "epoch": 163,
          "updates": 1956,
          "samples_seen": 125184,
          "online_loss": 0.000701831534873539,
          "train_loss": 0.0006878965068608522,
          "validation_loss": 0.001841826131567359,
          "train_metric": 6.67069325572811e-05,
          "validation_metric": 0.00017860620573628694,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.7155868829995597
        },
        {
          "epoch": 164,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0007632947599631734,
          "train_loss": 0.0006875737453810871,
          "validation_loss": 0.001820572535507381,
          "train_metric": 6.66756386635825e-05,
          "validation_metric": 0.0001765452034305781,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.723752357000194
        },
        {
          "epoch": 165,
          "updates": 1980,
          "samples_seen": 126720,
          "online_loss": 0.000715406822564546,
          "train_loss": 0.0007275768439285457,
          "validation_loss": 0.0018577596638351679,
          "train_metric": 7.055483001749963e-05,
          "validation_metric": 0.00018015132809523493,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.7321435439989727
        },
        {
          "epoch": 166,
          "updates": 1992,
          "samples_seen": 127488,
          "online_loss": 0.0007222166859719437,
          "train_loss": 0.0006873186212033033,
          "validation_loss": 0.001851139240898192,
          "train_metric": 6.665090040769428e-05,
          "validation_metric": 0.00017950934125110507,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.740562349999891
        },
        {
          "epoch": 167,
          "updates": 2004,
          "samples_seen": 128256,
          "online_loss": 0.0007232340382567296,
          "train_loss": 0.0006907111965119839,
          "validation_loss": 0.0018201371422037482,
          "train_metric": 6.697989738313481e-05,
          "validation_metric": 0.00017650301742833108,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.749094630999025
        },
        {
          "epoch": 168,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.0007105857390949192,
          "train_loss": 0.0006862692534923553,
          "validation_loss": 0.0018407009774819016,
          "train_metric": 6.654914614045992e-05,
          "validation_metric": 0.00017849708092398942,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.7572281609991478
        },
        {
          "epoch": 169,
          "updates": 2028,
          "samples_seen": 129792,
          "online_loss": 0.000753417630524685,
          "train_loss": 0.0006880022119730711,
          "validation_loss": 0.0018390612676739693,
          "train_metric": 6.67171916575171e-05,
          "validation_metric": 0.0001783381012501195,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.765457293999134
        },
        {
          "epoch": 170,
          "updates": 2040,
          "samples_seen": 130560,
          "online_loss": 0.0007475274614989758,
          "train_loss": 0.0006803810247220099,
          "validation_loss": 0.0018376903608441353,
          "train_metric": 6.597815081477165e-05,
          "validation_metric": 0.00017820516950450838,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 1.7737881159991957
        }
      ],
      "best_validation_metric": 0.00017304369248449802,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 33,
      "parameters": 2369,
      "train_samples": 768,
      "validation_rmse_m_s": 0.013154607272149862,
      "training_profiles": [
        0,
        1,
        4,
        6,
        7,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.2388849128037691,
          "train_loss": 0.08166318386793137,
          "validation_loss": 0.06555182486772537,
          "train_metric": 0.008224816992878914,
          "validation_metric": 0.006602139677852392,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02078287999938766
        },
        {
          "epoch": 2,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.06178459106013179,
          "train_loss": 0.05403638631105423,
          "validation_loss": 0.04518972337245941,
          "train_metric": 0.005442346446216106,
          "validation_metric": 0.0045513431541621685,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03856002599968633
        },
        {
          "epoch": 3,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.041212972428184,
          "train_loss": 0.034821365028619766,
          "validation_loss": 0.027817271649837494,
          "train_metric": 0.0035070800222456455,
          "validation_metric": 0.0028016536962240934,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.056507965999117005
        },
        {
          "epoch": 4,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.029260385723318905,
          "train_loss": 0.024330992251634598,
          "validation_loss": 0.018964281305670738,
          "train_metric": 0.0024505285546183586,
          "validation_metric": 0.001910013030283153,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07470617099897936
        },
        {
          "epoch": 5,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.02419908356387168,
          "train_loss": 0.0212287288159132,
          "validation_loss": 0.021520044654607773,
          "train_metric": 0.0021380798425525427,
          "validation_metric": 0.0021674204617738724,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09149877199888579
        },
        {
          "epoch": 6,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.021335104072932154,
          "train_loss": 0.017451366409659386,
          "validation_loss": 0.015473037958145142,
          "train_metric": 0.0017576378304511309,
          "validation_metric": 0.001558387652039528,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10919745699902705
        },
        {
          "epoch": 7,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.029768219159450382,
          "train_loss": 0.032846707850694656,
          "validation_loss": 0.03494375944137573,
          "train_metric": 0.003308200044557452,
          "validation_metric": 0.0035194074735045433,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.12598374499975762
        },
        {
          "epoch": 8,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.028789970820071176,
          "train_loss": 0.02708311378955841,
          "validation_loss": 0.019959069788455963,
          "train_metric": 0.0027277120389044285,
          "validation_metric": 0.0020102045964449644,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.14254694699957327
        },
        {
          "epoch": 9,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.016897958033951,
          "train_loss": 0.019038455560803413,
          "validation_loss": 0.013564357534050941,
          "train_metric": 0.0019174834014847875,
          "validation_metric": 0.0013661523116752505,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24352568899848848
        },
        {
          "epoch": 10,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.014107381663052365,
          "train_loss": 0.011092585511505604,
          "validation_loss": 0.009211717173457146,
          "train_metric": 0.001117204548791051,
          "validation_metric": 0.0009277703939005733,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25576088599882496
        },
        {
          "epoch": 11,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.011751928148441948,
          "train_loss": 0.00884933490306139,
          "validation_loss": 0.00877564586699009,
          "train_metric": 0.0008912726771086454,
          "validation_metric": 0.0008838509675115347,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2680297199985944
        },
        {
          "epoch": 12,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.010980222083162516,
          "train_loss": 0.025530735030770302,
          "validation_loss": 0.021314870566129684,
          "train_metric": 0.0025713620707392693,
          "validation_metric": 0.002146755810827017,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2792779459996382
        },
        {
          "epoch": 13,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.02549984192592092,
          "train_loss": 0.016958437860012054,
          "validation_loss": 0.01574435457587242,
          "train_metric": 0.00170799158513546,
          "validation_metric": 0.0015857137041166425,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.29074730799948156
        },
        {
          "epoch": 14,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.01990346507227514,
          "train_loss": 0.013715934939682484,
          "validation_loss": 0.01434900052845478,
          "train_metric": 0.001381418784148991,
          "validation_metric": 0.0014451788738369942,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3020225569998729
        },
        {
          "epoch": 15,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.013678442890523002,
          "train_loss": 0.019475840032100677,
          "validation_loss": 0.013384134508669376,
          "train_metric": 0.0019615350756794214,
          "validation_metric": 0.0013480009511113167,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3131267519984249
        },
        {
          "epoch": 16,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.011750799196306616,
          "train_loss": 0.011588438414037228,
          "validation_loss": 0.008362023159861565,
          "train_metric": 0.0011671452084556222,
          "validation_metric": 0.0008421923848800361,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3248903879994032
        },
        {
          "epoch": 17,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.007878501550294459,
          "train_loss": 0.007302616257220507,
          "validation_loss": 0.005848750472068787,
          "train_metric": 0.0007354928529821336,
          "validation_metric": 0.0005890646716579795,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.33698337099849596
        },
        {
          "epoch": 18,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.008894446160411462,
          "train_loss": 0.006209285464137793,
          "validation_loss": 0.004950490314513445,
          "train_metric": 0.0006253764149732888,
          "validation_metric": 0.0004985951818525791,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.34881288099859376
        },
        {
          "epoch": 19,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.01132512217736803,
          "train_loss": 0.005510804243385792,
          "validation_loss": 0.004341584164649248,
          "train_metric": 0.0005550279747694731,
          "validation_metric": 0.00043726846342906356,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.36077058799855877
        },
        {
          "epoch": 20,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.0076834783831145614,
          "train_loss": 0.004882063716650009,
          "validation_loss": 0.00400752667337656,
          "train_metric": 0.0004917035694234073,
          "validation_metric": 0.0004036234167870134,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37307920700004615
        },
        {
          "epoch": 21,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.00805011972261127,
          "train_loss": 0.0057880692183971405,
          "validation_loss": 0.004216188099235296,
          "train_metric": 0.0005829531000927091,
          "validation_metric": 0.0004246389726176858,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3845658869995532
        },
        {
          "epoch": 22,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.017195425942190923,
          "train_loss": 0.023246794939041138,
          "validation_loss": 0.017628690227866173,
          "train_metric": 0.002341332146897912,
          "validation_metric": 0.0017754969885572791,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3958467180000298
        },
        {
          "epoch": 23,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.024376829678658396,
          "train_loss": 0.009796548634767532,
          "validation_loss": 0.009899702854454517,
          "train_metric": 0.0009866725886240602,
          "validation_metric": 0.0009970617247745395,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4073029969986237
        },
        {
          "epoch": 24,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.010538687332882546,
          "train_loss": 0.006641799118369818,
          "validation_loss": 0.005747571121901274,
          "train_metric": 0.0006689376314170659,
          "validation_metric": 0.0005788743146695197,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41848068599938415
        },
        {
          "epoch": 25,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.010213399669737555,
          "train_loss": 0.007118729408830404,
          "validation_loss": 0.0071565331891179085,
          "train_metric": 0.0007169725140556693,
          "validation_metric": 0.0007207797607406974,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4299903020000784
        },
        {
          "epoch": 26,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.008562687020457815,
          "train_loss": 0.005284890998154879,
          "validation_loss": 0.0048418063670396805,
          "train_metric": 0.0005322748911567032,
          "validation_metric": 0.00048764896928332746,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.441464122999605
        },
        {
          "epoch": 27,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.012570571947435383,
          "train_loss": 0.022207990288734436,
          "validation_loss": 0.018989352509379387,
          "train_metric": 0.002236707601696253,
          "validation_metric": 0.0019125380786135793,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4527701839997462
        },
        {
          "epoch": 28,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.016797543328721076,
          "train_loss": 0.006614943500608206,
          "validation_loss": 0.005369212944060564,
          "train_metric": 0.0006662329542450607,
          "validation_metric": 0.000540767447091639,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.464211473999967
        },
        {
          "epoch": 29,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.007585178536828607,
          "train_loss": 0.007600062061101198,
          "validation_loss": 0.007060759235173464,
          "train_metric": 0.0007654504734091461,
          "validation_metric": 0.0007111338782124221,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.47551581699917733
        },
        {
          "epoch": 30,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.013991023384733126,
          "train_loss": 0.0160601194947958,
          "validation_loss": 0.01095736026763916,
          "train_metric": 0.00161751639097929,
          "validation_metric": 0.0011035852367058396,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4868968259997928
        },
        {
          "epoch": 31,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.008522318326868117,
          "train_loss": 0.006465114187449217,
          "validation_loss": 0.006082844454795122,
          "train_metric": 0.0006511426181532443,
          "validation_metric": 0.0006126418593339622,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4983977279989631
        },
        {
          "epoch": 32,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.008892199999536388,
          "train_loss": 0.010725112631917,
          "validation_loss": 0.008439010009169579,
          "train_metric": 0.0010801941389217973,
          "validation_metric": 0.0008499461691826582,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5099244679986441
        },
        {
          "epoch": 33,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.007572439593786839,
          "train_loss": 0.003920089919120073,
          "validation_loss": 0.003430972807109356,
          "train_metric": 0.00039481709245592356,
          "validation_metric": 0.00034555495949462056,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5217283669990138
        },
        {
          "epoch": 34,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.003607242339057848,
          "train_loss": 0.0032221139408648014,
          "validation_loss": 0.0027554009575396776,
          "train_metric": 0.0003245194675400853,
          "validation_metric": 0.00027751390007324517,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5336500229986996
        },
        {
          "epoch": 35,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.003925277222151635,
          "train_loss": 0.003137432038784027,
          "validation_loss": 0.0029943687841296196,
          "train_metric": 0.0003159906773362309,
          "validation_metric": 0.00030158180743455887,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5449208809986885
        },
        {
          "epoch": 36,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.004980146637535654,
          "train_loss": 0.003434479469433427,
          "validation_loss": 0.003160941181704402,
          "train_metric": 0.0003459081635810435,
          "validation_metric": 0.00031835847767069936,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.556132099000024
        },
        {
          "epoch": 37,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.0031407287751790136,
          "train_loss": 0.0024505085311830044,
          "validation_loss": 0.0023674406111240387,
          "train_metric": 0.00024680624483153224,
          "validation_metric": 0.00023843992676120251,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5678585810001096
        },
        {
          "epoch": 38,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.002832180165569298,
          "train_loss": 0.002387761604040861,
          "validation_loss": 0.001893980661407113,
          "train_metric": 0.0002404865954304114,
          "validation_metric": 0.0001907547703012824,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5797591399987141
        },
        {
          "epoch": 39,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.006534582149470225,
          "train_loss": 0.028642652556300163,
          "validation_loss": 0.018807820975780487,
          "train_metric": 0.0028847833164036274,
          "validation_metric": 0.0018942547030746937,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5910179749989766
        },
        {
          "epoch": 40,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.016350389065337367,
          "train_loss": 0.012905722483992577,
          "validation_loss": 0.012093516997992992,
          "train_metric": 0.0012998171150684357,
          "validation_metric": 0.0012180147459730506,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.602279531000022
        },
        {
          "epoch": 41,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.008869165219948627,
          "train_loss": 0.0075831967405974865,
          "validation_loss": 0.006102235056459904,
          "train_metric": 0.0007637517992407084,
          "validation_metric": 0.00061459478456527,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6134276319990022
        },
        {
          "epoch": 42,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.005887303752388107,
          "train_loss": 0.0053742434829473495,
          "validation_loss": 0.0037914335262030363,
          "train_metric": 0.0005412740865722299,
          "validation_metric": 0.00038185931043699384,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6246910049994767
        },
        {
          "epoch": 43,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.004209444236039417,
          "train_loss": 0.0029743139166384935,
          "validation_loss": 0.0022562851663678885,
          "train_metric": 0.00029956200160086155,
          "validation_metric": 0.00022724476002622396,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6360145829985413
        },
        {
          "epoch": 44,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.003216043609427288,
          "train_loss": 0.0034065672662109137,
          "validation_loss": 0.0027101044543087482,
          "train_metric": 0.0003430969372857362,
          "validation_metric": 0.0002729517873376608,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6475017429984291
        },
        {
          "epoch": 45,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.003296306458651088,
          "train_loss": 0.0036727366968989372,
          "validation_loss": 0.003828497603535652,
          "train_metric": 0.000369904562830925,
          "validation_metric": 0.00038559225504286587,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6587989999989077
        },
        {
          "epoch": 46,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.0030792666475463193,
          "train_loss": 0.003218308789655566,
          "validation_loss": 0.002296805614605546,
          "train_metric": 0.0003241363156121224,
          "validation_metric": 0.00023132585920393467,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6701348299993697
        },
        {
          "epoch": 47,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.003617135480453726,
          "train_loss": 0.0023462858516722918,
          "validation_loss": 0.002109959488734603,
          "train_metric": 0.00023630933719687164,
          "validation_metric": 0.00021250738063827157,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6814103189990419
        },
        {
          "epoch": 48,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.0023250978156283963,
          "train_loss": 0.0038655956741422415,
          "validation_loss": 0.0029629087075591087,
          "train_metric": 0.0003893286339007318,
          "validation_metric": 0.0002984133316203952,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.692846264999389
        },
        {
          "epoch": 49,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.0029478614742401987,
          "train_loss": 0.002939426340162754,
          "validation_loss": 0.0022886001970618963,
          "train_metric": 0.0002960482379421592,
          "validation_metric": 0.0002304994413862005,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7041283620001195
        },
        {
          "epoch": 50,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.003480549858068116,
          "train_loss": 0.0038005749229341745,
          "validation_loss": 0.0030069262720644474,
          "train_metric": 0.0003827800101134926,
          "validation_metric": 0.0003028465434908867,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7154740859987214
        },
        {
          "epoch": 51,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.0026053215842694044,
          "train_loss": 0.0018212493741884828,
          "validation_loss": 0.0016792260576039553,
          "train_metric": 0.0001834295690059662,
          "validation_metric": 0.00016912550199776888,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7275673639996967
        },
        {
          "epoch": 52,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.004080366339621833,
          "train_loss": 0.002198778325691819,
          "validation_loss": 0.0019176234491169453,
          "train_metric": 0.00022145290859043598,
          "validation_metric": 0.0001931360166054219,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7389295309985755
        },
        {
          "epoch": 53,
          "updates": 848,
          "samples_seen": 54272,
          "online_loss": 0.002719874453759985,
          "train_loss": 0.0019235246581956744,
          "validation_loss": 0.001675290521234274,
          "train_metric": 0.00019373034592717886,
          "validation_metric": 0.00016872913693077862,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7510472559988557
        },
        {
          "epoch": 54,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0023776644957251847,
          "train_loss": 0.0017265448113903403,
          "validation_loss": 0.0013534065801650286,
          "train_metric": 0.00017389126878697425,
          "validation_metric": 0.00013631016190629452,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7633015269984753
        },
        {
          "epoch": 55,
          "updates": 880,
          "samples_seen": 56320,
          "online_loss": 0.002015912083152216,
          "train_loss": 0.001683213165961206,
          "validation_loss": 0.0012907448690384626,
          "train_metric": 0.00016952706209849566,
          "validation_metric": 0.0001299990835832432,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7753521069989802
        },
        {
          "epoch": 56,
          "updates": 896,
          "samples_seen": 57344,
          "online_loss": 0.003662852759589441,
          "train_loss": 0.004538101144134998,
          "validation_loss": 0.0038468309212476015,
          "train_metric": 0.00045706084347330034,
          "validation_metric": 0.00038743874756619334,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7867808129994955
        },
        {
          "epoch": 57,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.0026208025628875475,
          "train_loss": 0.0019290491472929716,
          "validation_loss": 0.0014470829628407955,
          "train_metric": 0.00019428676750976592,
          "validation_metric": 0.0001457449106965214,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7982121539989748
        },
        {
          "epoch": 58,
          "updates": 928,
          "samples_seen": 59392,
          "online_loss": 0.002210343316619401,
          "train_loss": 0.0026221219450235367,
          "validation_loss": 0.0017475809436291456,
          "train_metric": 0.0002640905440784991,
          "validation_metric": 0.00017600995488464832,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8095573509999667
        },
        {
          "epoch": 59,
          "updates": 944,
          "samples_seen": 60416,
          "online_loss": 0.004260339141183067,
          "train_loss": 0.003006784012541175,
          "validation_loss": 0.0031082017812877893,
          "train_metric": 0.0003028322826139629,
          "validation_metric": 0.00031304670847021043,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8207289049987594
        },
        {
          "epoch": 60,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.0022360518960340414,
          "train_loss": 0.0022161505185067654,
          "validation_loss": 0.001542267156764865,
          "train_metric": 0.00022320254356600344,
          "validation_metric": 0.00015533147961832583,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8319809879994864
        },
        {
          "epoch": 61,
          "updates": 976,
          "samples_seen": 62464,
          "online_loss": 0.0018291905835212674,
          "train_loss": 0.0020380057394504547,
          "validation_loss": 0.0019052597926929593,
          "train_metric": 0.00020526051230262965,
          "validation_metric": 0.0001918907801155001,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8432648699999845
        },
        {
          "epoch": 62,
          "updates": 992,
          "samples_seen": 63488,
          "online_loss": 0.001721346408885438,
          "train_loss": 0.0014482982223853469,
          "validation_loss": 0.0011452032485976815,
          "train_metric": 0.00014586729230359197,
          "validation_metric": 0.00011534068471519276,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8551766559994576
        },
        {
          "epoch": 63,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.002811607537296368,
          "train_loss": 0.014163834974169731,
          "validation_loss": 0.010514729656279087,
          "train_metric": 0.0014265296049416065,
          "validation_metric": 0.0010590049205347896,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8666904559995601
        },
        {
          "epoch": 64,
          "updates": 1024,
          "samples_seen": 65536,
          "online_loss": 0.007845600135624409,
          "train_loss": 0.0029643867164850235,
          "validation_loss": 0.0027569790836423635,
          "train_metric": 0.0002985621977131814,
          "validation_metric": 0.00027767286519519985,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8779966169986437
        },
        {
          "epoch": 65,
          "updates": 1040,
          "samples_seen": 66560,
          "online_loss": 0.004363848398497794,
          "train_loss": 0.008987226523458958,
          "validation_loss": 0.007578655611723661,
          "train_metric": 0.0009051605011336505,
          "validation_metric": 0.0007632944034412503,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8893629249996593
        },
        {
          "epoch": 66,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.0044192369969096035,
          "train_loss": 0.00180892925709486,
          "validation_loss": 0.00132869859226048,
          "train_metric": 0.00018218872719444335,
          "validation_metric": 0.0001338216388830915,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9006472649998614
        },
        {
          "epoch": 67,
          "updates": 1072,
          "samples_seen": 68608,
          "online_loss": 0.0031426198074768763,
          "train_loss": 0.002104674931615591,
          "validation_loss": 0.0015148133970797062,
          "train_metric": 0.00021197512978687882,
          "validation_metric": 0.00015256645565386862,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9119634639992
        },
        {
          "epoch": 68,
          "updates": 1088,
          "samples_seen": 69632,
          "online_loss": 0.0022411084501072764,
          "train_loss": 0.0018391741905361414,
          "validation_loss": 0.00140914018265903,
          "train_metric": 0.00018523490871302783,
          "validation_metric": 0.00014192343223839998,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9232122649991652
        },
        {
          "epoch": 69,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.001856512335507432,
          "train_loss": 0.001539024873636663,
          "validation_loss": 0.0014161864528432488,
          "train_metric": 0.0001550049491925165,
          "validation_metric": 0.00014263308548834175,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9345114799998555
        },
        {
          "epoch": 70,
          "updates": 1120,
          "samples_seen": 71680,
          "online_loss": 0.001799609774025157,
          "train_loss": 0.002310500480234623,
          "validation_loss": 0.0018174133729189634,
          "train_metric": 0.00023270517704077065,
          "validation_metric": 0.00018304323020856827,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9456593079994491
        },
        {
          "epoch": 71,
          "updates": 1136,
          "samples_seen": 72704,
          "online_loss": 0.0013366869679884985,
          "train_loss": 0.0015817340463399887,
          "validation_loss": 0.0012221885845065117,
          "train_metric": 0.0001593064662301913,
          "validation_metric": 0.00012309435987845063,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9570890439990762
        },
        {
          "epoch": 72,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0014940660093998304,
          "train_loss": 0.001187215093523264,
          "validation_loss": 0.0010790928499773145,
          "train_metric": 0.000119571945106145,
          "validation_metric": 0.00010868227400351316,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9688974609998695
        },
        {
          "epoch": 73,
          "updates": 1168,
          "samples_seen": 74752,
          "online_loss": 0.0013925510174885858,
          "train_loss": 0.0013070175191387534,
          "validation_loss": 0.0009654491441324353,
          "train_metric": 0.00013163800758775324,
          "validation_metric": 9.72365087363869e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9808702339996671
        },
        {
          "epoch": 74,
          "updates": 1184,
          "samples_seen": 75776,
          "online_loss": 0.0014476400829153135,
          "train_loss": 0.001518074655905366,
          "validation_loss": 0.0014484759885817766,
          "train_metric": 0.00015289493603631854,
          "validation_metric": 0.00014588522026315331,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9924009709993697
        },
        {
          "epoch": 75,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.0018728342784015695,
          "train_loss": 0.0019472280982881784,
          "validation_loss": 0.0014619478024542332,
          "train_metric": 0.00019611768948379904,
          "validation_metric": 0.00014724199718330055,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0035358049990464
        },
        {
          "epoch": 76,
          "updates": 1216,
          "samples_seen": 77824,
          "online_loss": 0.005485426874656696,
          "train_loss": 0.0013190505560487509,
          "validation_loss": 0.0012575038708746433,
          "train_metric": 0.0001328499347437173,
          "validation_metric": 0.00012665119720622897,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0146091999995406
        },
        {
          "epoch": 77,
          "updates": 1232,
          "samples_seen": 78848,
          "online_loss": 0.006465837079304038,
          "train_loss": 0.005295605398714542,
          "validation_loss": 0.004090891219675541,
          "train_metric": 0.0005333539447747171,
          "validation_metric": 0.0004120195226278156,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0258175330000086
        },
        {
          "epoch": 78,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.004608969215041725,
          "train_loss": 0.00442119175568223,
          "validation_loss": 0.0033473202493041754,
          "train_metric": 0.0004452863067854196,
          "validation_metric": 0.0003371298953425139,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0368832929998462
        },
        {
          "epoch": 79,
          "updates": 1264,
          "samples_seen": 80896,
          "online_loss": 0.0037160005704208743,
          "train_loss": 0.0016003415221348405,
          "validation_loss": 0.0012910048244521022,
          "train_metric": 0.0001611805200809613,
          "validation_metric": 0.00013002527703065425,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0482820149991312
        },
        {
          "epoch": 80,
          "updates": 1280,
          "samples_seen": 81920,
          "online_loss": 0.0019796651122305775,
          "train_loss": 0.0014186376938596368,
          "validation_loss": 0.0011992004001513124,
          "train_metric": 0.00014287998783402145,
          "validation_metric": 0.00012077909195795655,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0596090509989153
        },
        {
          "epoch": 81,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.002652116774697788,
          "train_loss": 0.007639347109943628,
          "validation_loss": 0.00444882083684206,
          "train_metric": 0.0007694070227444172,
          "validation_metric": 0.0004480689240153879,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0707471499990788
        },
        {
          "epoch": 82,
          "updates": 1312,
          "samples_seen": 83968,
          "online_loss": 0.0036989080435887445,
          "train_loss": 0.001447979360818863,
          "validation_loss": 0.001028208527714014,
          "train_metric": 0.0001458351907785982,
          "validation_metric": 0.00010355739505030215,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0819141559986747
        },
        {
          "epoch": 83,
          "updates": 1328,
          "samples_seen": 84992,
          "online_loss": 0.0013876320444978774,
          "train_loss": 0.0011812099255621433,
          "validation_loss": 0.0009725422132760286,
          "train_metric": 0.000118967131129466,
          "validation_metric": 9.795087680686265e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.093157526000141
        },
        {
          "epoch": 84,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.001984124124646769,
          "train_loss": 0.0021492214873433113,
          "validation_loss": 0.0017386440886184573,
          "train_metric": 0.00021646171808242798,
          "validation_metric": 0.00017510986072011292,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1041930969986424
        },
        {
          "epoch": 85,
          "updates": 1360,
          "samples_seen": 87040,
          "online_loss": 0.002818073497110163,
          "train_loss": 0.004713023547083139,
          "validation_loss": 0.003433603560552001,
          "train_metric": 0.00047467846889048815,
          "validation_metric": 0.0003458199498709291,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1155557299989596
        },
        {
          "epoch": 86,
          "updates": 1376,
          "samples_seen": 88064,
          "online_loss": 0.002345837292523356,
          "train_loss": 0.0016680110711604357,
          "validation_loss": 0.0013760486617684364,
          "train_metric": 0.0001679959532339126,
          "validation_metric": 0.00013859057798981667,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1269072649993177
        },
        {
          "epoch": 87,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0012541318610601593,
          "train_loss": 0.0010155250784009695,
          "validation_loss": 0.0007942498195916414,
          "train_metric": 0.00010227995517197996,
          "validation_metric": 7.99939371063374e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1387562769996293
        },
        {
          "epoch": 88,
          "updates": 1408,
          "samples_seen": 90112,
          "online_loss": 0.00113826784217963,
          "train_loss": 0.0010479639749974012,
          "validation_loss": 0.0008517428068444133,
          "train_metric": 0.00010554708569543436,
          "validation_metric": 8.578442066209391e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.14995078299944
        },
        {
          "epoch": 89,
          "updates": 1424,
          "samples_seen": 91136,
          "online_loss": 0.0009965281915356172,
          "train_loss": 0.0009837758261710405,
          "validation_loss": 0.0008038682281039655,
          "train_metric": 9.908231004374102e-05,
          "validation_metric": 8.096267265500501e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1610552539987111
        },
        {
          "epoch": 90,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0010749436787591549,
          "train_loss": 0.0010403982596471906,
          "validation_loss": 0.0009059951407834888,
          "train_metric": 0.000104785111034289,
          "validation_metric": 9.124851931119338e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.17209726100009
        },
        {
          "epoch": 91,
          "updates": 1456,
          "samples_seen": 93184,
          "online_loss": 0.0014602425108023454,
          "train_loss": 0.001944422721862793,
          "validation_loss": 0.001366357202641666,
          "train_metric": 0.00019583514949772507,
          "validation_metric": 0.00013761449372395873,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1831690269991668
        },
        {
          "epoch": 92,
          "updates": 1472,
          "samples_seen": 94208,
          "online_loss": 0.0012968079518032027,
          "train_loss": 0.001923979725688696,
          "validation_loss": 0.0017304642824456096,
          "train_metric": 0.00019377619901206344,
          "validation_metric": 0.0001742860331432894,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1945190890000958
        },
        {
          "epoch": 93,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0019329253354953835,
          "train_loss": 0.0038161128759384155,
          "validation_loss": 0.0029621021822094917,
          "train_metric": 0.0003843448939733207,
          "validation_metric": 0.00029833210282959044,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2057015310001589
        },
        {
          "epoch": 94,
          "updates": 1504,
          "samples_seen": 96256,
          "online_loss": 0.0017400501765223453,
          "train_loss": 0.0011966590536758304,
          "validation_loss": 0.0009719710797071457,
          "train_metric": 0.00012052310921717435,
          "validation_metric": 9.789337491383776e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2168115219992615
        },
        {
          "epoch": 95,
          "updates": 1520,
          "samples_seen": 97280,
          "online_loss": 0.0017478972440585494,
          "train_loss": 0.0030153461266309023,
          "validation_loss": 0.002942101564258337,
          "train_metric": 0.0003036946291103959,
          "validation_metric": 0.0002963176812045276,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.227869855998506
        },
        {
          "epoch": 96,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0018786856162478216,
          "train_loss": 0.0011003551771864295,
          "validation_loss": 0.0007198831881396472,
          "train_metric": 0.00011082374840043485,
          "validation_metric": 7.250400085467845e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.239419341998655
        },
        {
          "epoch": 97,
          "updates": 1552,
          "samples_seen": 99328,
          "online_loss": 0.0011183844508195762,
          "train_loss": 0.001064203679561615,
          "validation_loss": 0.0008210770902223885,
          "train_metric": 0.00010718268458731472,
          "validation_metric": 8.269587124232203e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.25037069300015
        },
        {
          "epoch": 98,
          "updates": 1568,
          "samples_seen": 100352,
          "online_loss": 0.0014152042913337937,
          "train_loss": 0.0010325449984520674,
          "validation_loss": 0.0008352725417353213,
          "train_metric": 0.00010399413440609351,
          "validation_metric": 8.412559691350907e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.261418258998674
        },
        {
          "epoch": 99,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0022218364392756484,
          "train_loss": 0.0025818808935582638,
          "validation_loss": 0.0016593331238254905,
          "train_metric": 0.0002600376319605857,
          "validation_metric": 0.00016712193610146642,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2723686899989843
        },
        {
          "epoch": 100,
          "updates": 1600,
          "samples_seen": 102400,
          "online_loss": 0.0020865266160399187,
          "train_loss": 0.0021263211965560913,
          "validation_loss": 0.0019840055610984564,
          "train_metric": 0.00021415529772639275,
          "validation_metric": 0.00019982174853794277,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2832367760001944
        },
        {
          "epoch": 101,
          "updates": 1616,
          "samples_seen": 103424,
          "online_loss": 0.001463073738705134,
          "train_loss": 0.0009938577422872186,
          "validation_loss": 0.000675154326017946,
          "train_metric": 0.00010009770630858839,
          "validation_metric": 6.799906986998394e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2948230749989307
        },
        {
          "epoch": 102,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.001151071428466821,
          "train_loss": 0.0012099450686946511,
          "validation_loss": 0.0010119028156623244,
          "train_metric": 0.00012186124513391405,
          "validation_metric": 0.00010191513138124719,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3059829089997947
        },
        {
          "epoch": 103,
          "updates": 1648,
          "samples_seen": 105472,
          "online_loss": 0.0018336924040340818,
          "train_loss": 0.002508704550564289,
          "validation_loss": 0.0015389361651614308,
          "train_metric": 0.00025266752345487475,
          "validation_metric": 0.0001549960143165663,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3169170639994263
        },
        {
          "epoch": 104,
          "updates": 1664,
          "samples_seen": 106496,
          "online_loss": 0.00118149135960266,
          "train_loss": 0.0011720415204763412,
          "validation_loss": 0.0011014898773282766,
          "train_metric": 0.00011804373934864998,
          "validation_metric": 0.00011093801003880799,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.327946235998752
        },
        {
          "epoch": 105,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0009116013006860157,
          "train_loss": 0.0008642507018521428,
          "validation_loss": 0.0006143507780507207,
          "train_metric": 8.704415813554078e-05,
          "validation_metric": 6.187517283251509e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3395082669994736
        },
        {
          "epoch": 106,
          "updates": 1696,
          "samples_seen": 108544,
          "online_loss": 0.000888768217919278,
          "train_loss": 0.0013063357910141349,
          "validation_loss": 0.0010510067222639918,
          "train_metric": 0.0001315693516517058,
          "validation_metric": 0.00010585353447822854,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3504776039990247
        },
        {
          "epoch": 107,
          "updates": 1712,
          "samples_seen": 109568,
          "online_loss": 0.0014120061932771932,
          "train_loss": 0.003388883313164115,
          "validation_loss": 0.0023172940127551556,
          "train_metric": 0.0003413159283809364,
          "validation_metric": 0.000233389378990978,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3616147630000341
        },
        {
          "epoch": 108,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0012908653625345323,
          "train_loss": 0.0008294145809486508,
          "validation_loss": 0.0006559605244547129,
          "train_metric": 8.353559678653255e-05,
          "validation_metric": 6.606595707125962e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3725343109999812
        },
        {
          "epoch": 109,
          "updates": 1744,
          "samples_seen": 111616,
          "online_loss": 0.0009402670984854922,
          "train_loss": 0.0008135782554745674,
          "validation_loss": 0.0006201523938216269,
          "train_metric": 8.194064139388502e-05,
          "validation_metric": 6.245948316063732e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.383432847000222
        },
        {
          "epoch": 110,
          "updates": 1760,
          "samples_seen": 112640,
          "online_loss": 0.001619107904843986,
          "train_loss": 0.0017546056769788265,
          "validation_loss": 0.0013111615553498268,
          "train_metric": 0.0001767174544511363,
          "validation_metric": 0.00013205538562033325,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3944083289989067
        },
        {
          "epoch": 111,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.002896530555517529,
          "train_loss": 0.003180895699188113,
          "validation_loss": 0.0036294134333729744,
          "train_metric": 0.0003203681844752282,
          "validation_metric": 0.00036554120015352964,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4052632900002209
        },
        {
          "epoch": 112,
          "updates": 1792,
          "samples_seen": 114688,
          "online_loss": 0.0022463543555204524,
          "train_loss": 0.0020763238426297903,
          "validation_loss": 0.0011937252711504698,
          "train_metric": 0.00020911972387693822,
          "validation_metric": 0.00012022763257846236,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4160931629994593
        },
        {
          "epoch": 113,
          "updates": 1808,
          "samples_seen": 115712,
          "online_loss": 0.0015253144647431327,
          "train_loss": 0.001120115863159299,
          "validation_loss": 0.0008319179760292172,
          "train_metric": 0.00011281396291451529,
          "validation_metric": 8.378773054573685e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4270993249992898
        },
        {
          "epoch": 114,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.001263531874428736,
          "train_loss": 0.00163357297424227,
          "validation_loss": 0.0013926455285400152,
          "train_metric": 0.0001645274751354009,
          "validation_metric": 0.00014026215649209917,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4379814419989998
        },
        {
          "epoch": 115,
          "updates": 1840,
          "samples_seen": 117760,
          "online_loss": 0.0013923459482612088,
          "train_loss": 0.001304728677496314,
          "validation_loss": 0.0009021886507980525,
          "train_metric": 0.0001314075052505359,
          "validation_metric": 9.086514182854444e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4488795649995154
        },
        {
          "epoch": 116,
          "updates": 1856,
          "samples_seen": 118784,
          "online_loss": 0.0015674962633056566,
          "train_loss": 0.0010565624106675386,
          "validation_loss": 0.0006733196787536144,
          "train_metric": 0.00010641310655046254,
          "validation_metric": 6.781428237445652e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4599267349985894
        },
        {
          "epoch": 117,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.001053830659657251,
          "train_loss": 0.000753972795791924,
          "validation_loss": 0.000574449310079217,
          "train_metric": 7.593737245770171e-05,
          "validation_metric": 5.7856424973579124e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4714356839995162
        },
        {
          "epoch": 118,
          "updates": 1888,
          "samples_seen": 120832,
          "online_loss": 0.0009630814711272251,
          "train_loss": 0.00073592847911641,
          "validation_loss": 0.000599745602812618,
          "train_metric": 7.412002014461905e-05,
          "validation_metric": 6.040418520569801e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4825745909984107
        },
        {
          "epoch": 119,
          "updates": 1904,
          "samples_seen": 121856,
          "online_loss": 0.0007728607251920039,
          "train_loss": 0.0007416520384140313,
          "validation_loss": 0.0006294539198279381,
          "train_metric": 7.469647971447557e-05,
          "validation_metric": 6.339629180729389e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4935906329992577
        },
        {
          "epoch": 120,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0007630780673935078,
          "train_loss": 0.000787758850492537,
          "validation_loss": 0.0007381044561043382,
          "train_metric": 7.934019231470302e-05,
          "validation_metric": 7.433917926391587e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5043919759991695
        },
        {
          "epoch": 121,
          "updates": 1936,
          "samples_seen": 123904,
          "online_loss": 0.0013088295363559155,
          "train_loss": 0.0016406559152528644,
          "validation_loss": 0.0012900856090709567,
          "train_metric": 0.00016524086822755635,
          "validation_metric": 0.00012993268319405615,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.515163340998697
        },
        {
          "epoch": 122,
          "updates": 1952,
          "samples_seen": 124928,
          "online_loss": 0.001200091845021234,
          "train_loss": 0.0007895908784121275,
          "validation_loss": 0.0007408249657601118,
          "train_metric": 7.952468877192587e-05,
          "validation_metric": 7.461317727575079e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5260353719986597
        },
        {
          "epoch": 123,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0008329938664246583,
          "train_loss": 0.0007379018934443593,
          "validation_loss": 0.0004965397529304028,
          "train_metric": 7.431877020280808e-05,
          "validation_metric": 5.000965757062659e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5375124919992231
        },
        {
          "epoch": 124,
          "updates": 1984,
          "samples_seen": 126976,
          "online_loss": 0.0008067749749898212,
          "train_loss": 0.0007858728640712798,
          "validation_loss": 0.0005208254442550242,
          "train_metric": 7.915023888926953e-05,
          "validation_metric": 5.245563079370186e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5483567639985267
        },
        {
          "epoch": 125,
          "updates": 2000,
          "samples_seen": 128000,
          "online_loss": 0.0008896477411326487,
          "train_loss": 0.0007982951356098056,
          "validation_loss": 0.0005726222298108041,
          "train_metric": 8.04013543529436e-05,
          "validation_metric": 5.767241236753762e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5590806179989158
        },
        {
          "epoch": 126,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.000943997143622255,
          "train_loss": 0.0006803205469623208,
          "validation_loss": 0.0005775721510872245,
          "train_metric": 6.851939542684704e-05,
          "validation_metric": 5.817095370730385e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5698153389985237
        },
        {
          "epoch": 127,
          "updates": 2032,
          "samples_seen": 130048,
          "online_loss": 0.0008375897577934666,
          "train_loss": 0.0006913229590281844,
          "validation_loss": 0.0006149114924483001,
          "train_metric": 6.962751649552956e-05,
          "validation_metric": 6.193162698764354e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5805900689992995
        },
        {
          "epoch": 128,
          "updates": 2048,
          "samples_seen": 131072,
          "online_loss": 0.0007404171828966355,
          "train_loss": 0.0006537981098517776,
          "validation_loss": 0.0005017401999793947,
          "train_metric": 6.584815128007904e-05,
          "validation_metric": 5.053343920735642e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5912539099990681
        },
        {
          "epoch": 129,
          "updates": 2064,
          "samples_seen": 132096,
          "online_loss": 0.0007067837232170859,
          "train_loss": 0.0007417864981107414,
          "validation_loss": 0.0005111728678457439,
          "train_metric": 7.471001299563795e-05,
          "validation_metric": 5.1483457355061546e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6021040939995146
        },
        {
          "epoch": 130,
          "updates": 2080,
          "samples_seen": 133120,
          "online_loss": 0.000771488137615961,
          "train_loss": 0.0007295109098777175,
          "validation_loss": 0.0005250639514997602,
          "train_metric": 7.347366772592068e-05,
          "validation_metric": 5.288252214086242e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6127919009995821
        },
        {
          "epoch": 131,
          "updates": 2096,
          "samples_seen": 134144,
          "online_loss": 0.0007271536069310969,
          "train_loss": 0.0006390198832377791,
          "validation_loss": 0.000528299598954618,
          "train_metric": 6.43597450107336e-05,
          "validation_metric": 5.320839409250766e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6233395099989139
        },
        {
          "epoch": 132,
          "updates": 2112,
          "samples_seen": 135168,
          "online_loss": 0.0008120074180624215,
          "train_loss": 0.0006319672684185207,
          "validation_loss": 0.000492200197186321,
          "train_metric": 6.364942964864895e-05,
          "validation_metric": 4.957260171067901e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.634832154999458
        },
        {
          "epoch": 133,
          "updates": 2128,
          "samples_seen": 136192,
          "online_loss": 0.0007135384621506091,
          "train_loss": 0.0006187153048813343,
          "validation_loss": 0.0005094959633424878,
          "train_metric": 6.231473525986075e-05,
          "validation_metric": 5.1314560550963506e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6458464079987607
        },
        {
          "epoch": 134,
          "updates": 2144,
          "samples_seen": 137216,
          "online_loss": 0.0006890583208587486,
          "train_loss": 0.000653602706734091,
          "validation_loss": 0.0005648237420246005,
          "train_metric": 6.582847709069028e-05,
          "validation_metric": 5.688697274308652e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.656406418998813
        },
        {
          "epoch": 135,
          "updates": 2160,
          "samples_seen": 138240,
          "online_loss": 0.0006978905785217648,
          "train_loss": 0.0006688819848932326,
          "validation_loss": 0.0004657699028030038,
          "train_metric": 6.736733485013247e-05,
          "validation_metric": 4.69106380478479e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6678560829986964
        },
        {
          "epoch": 136,
          "updates": 2176,
          "samples_seen": 139264,
          "online_loss": 0.0008951053096097894,
          "train_loss": 0.0009096726425923407,
          "validation_loss": 0.0005645743221975863,
          "train_metric": 9.161889465758577e-05,
          "validation_metric": 5.6861856137402356e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6786011879994476
        },
        {
          "epoch": 137,
          "updates": 2192,
          "samples_seen": 140288,
          "online_loss": 0.0010316569769202033,
          "train_loss": 0.0007711026701144874,
          "validation_loss": 0.0007840702310204506,
          "train_metric": 7.766264025121927e-05,
          "validation_metric": 7.896868191892281e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6890335349999077
        },
        {
          "epoch": 138,
          "updates": 2208,
          "samples_seen": 141312,
          "online_loss": 0.0012377412185742287,
          "train_loss": 0.0009587740059942007,
          "validation_loss": 0.0009327976731583476,
          "train_metric": 9.656421025283635e-05,
          "validation_metric": 9.394797234563157e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6994623489990772
        },
        {
          "epoch": 139,
          "updates": 2224,
          "samples_seen": 142336,
          "online_loss": 0.0007216270441858796,
          "train_loss": 0.0007017640746198595,
          "validation_loss": 0.0004969612346030772,
          "train_metric": 7.067911064950749e-05,
          "validation_metric": 5.005210914532654e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.710456967999562
        },
        {
          "epoch": 140,
          "updates": 2240,
          "samples_seen": 143360,
          "online_loss": 0.0007075251769492752,
          "train_loss": 0.0006003204616717994,
          "validation_loss": 0.00044899224303662777,
          "train_metric": 6.0462076362455264e-05,
          "validation_metric": 4.522085509961471e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7223194379985216
        },
        {
          "epoch": 141,
          "updates": 2256,
          "samples_seen": 144384,
          "online_loss": 0.0009192591514874948,
          "train_loss": 0.0011008214205503464,
          "validation_loss": 0.0008914764039218426,
          "train_metric": 0.0001108707147068344,
          "validation_metric": 8.978623372968286e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7336596449986246
        },
        {
          "epoch": 142,
          "updates": 2272,
          "samples_seen": 145408,
          "online_loss": 0.0007254173269757302,
          "train_loss": 0.0006168790860101581,
          "validation_loss": 0.0005967639153823256,
          "train_metric": 6.212980224518105e-05,
          "validation_metric": 6.01038736931514e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7452175990001706
        },
        {
          "epoch": 143,
          "updates": 2288,
          "samples_seen": 146432,
          "online_loss": 0.0006394045212800847,
          "train_loss": 0.0005713648861274123,
          "validation_loss": 0.00047831638948991895,
          "train_metric": 5.7545785239199176e-05,
          "validation_metric": 4.8174275434575975e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7564800399995875
        },
        {
          "epoch": 144,
          "updates": 2304,
          "samples_seen": 147456,
          "online_loss": 0.0005918962287978502,
          "train_loss": 0.0005738257896155119,
          "validation_loss": 0.00041735140257515013,
          "train_metric": 5.7793629821389914e-05,
          "validation_metric": 4.203409844194539e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.767940979998457
        },
        {
          "epoch": 145,
          "updates": 2320,
          "samples_seen": 148480,
          "online_loss": 0.0005623338911391329,
          "train_loss": 0.0008316052844747901,
          "validation_loss": 0.0008233815897256136,
          "train_metric": 8.375624020118266e-05,
          "validation_metric": 8.292796701425686e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.779802639999616
        },
        {
          "epoch": 146,
          "updates": 2336,
          "samples_seen": 149504,
          "online_loss": 0.0007150329911382869,
          "train_loss": 0.0006346395239233971,
          "validation_loss": 0.0005659150192514062,
          "train_metric": 6.391856732079759e-05,
          "validation_metric": 5.6996890634763986e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.791108916999292
        },
        {
          "epoch": 147,
          "updates": 2352,
          "samples_seen": 150528,
          "online_loss": 0.0005973669340164633,
          "train_loss": 0.0005512838251888752,
          "validation_loss": 0.0004147668951191008,
          "train_metric": 5.552328730118461e-05,
          "validation_metric": 4.177380105829798e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.802611100998547
        },
        {
          "epoch": 148,
          "updates": 2368,
          "samples_seen": 151552,
          "online_loss": 0.0005734298038078123,
          "train_loss": 0.0005500311963260174,
          "validation_loss": 0.0004248238110449165,
          "train_metric": 5.5397125834133476e-05,
          "validation_metric": 4.2786694393726066e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8135992719999194
        },
        {
          "epoch": 149,
          "updates": 2384,
          "samples_seen": 152576,
          "online_loss": 0.0005412477839854546,
          "train_loss": 0.0006432041991502047,
          "validation_loss": 0.000651777139864862,
          "train_metric": 6.478116847574711e-05,
          "validation_metric": 6.564461364177987e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8247457599991321
        },
        {
          "epoch": 150,
          "updates": 2400,
          "samples_seen": 153600,
          "online_loss": 0.001003211755232769,
          "train_loss": 0.001006593112833798,
          "validation_loss": 0.0008854628540575504,
          "train_metric": 0.00010138037032447755,
          "validation_metric": 8.918058301787823e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8355498169985367
        },
        {
          "epoch": 151,
          "updates": 2416,
          "samples_seen": 154624,
          "online_loss": 0.0008094539552985225,
          "train_loss": 0.0007328043575398624,
          "validation_loss": 0.0006760237738490105,
          "train_metric": 7.380535680567846e-05,
          "validation_metric": 6.808663601987064e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8466711419987405
        },
        {
          "epoch": 152,
          "updates": 2432,
          "samples_seen": 155648,
          "online_loss": 0.0007955831188155571,
          "train_loss": 0.001993192359805107,
          "validation_loss": 0.0017850699368864298,
          "train_metric": 0.00020074706117156893,
          "validation_metric": 0.00017978569667320698,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.858446283999001
        },
        {
          "epoch": 153,
          "updates": 2448,
          "samples_seen": 156672,
          "online_loss": 0.003982760665167007,
          "train_loss": 0.0054007223807275295,
          "validation_loss": 0.003613719018176198,
          "train_metric": 0.0005439409869723022,
          "validation_metric": 0.000363960542017594,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8695069379991764
        },
        {
          "epoch": 154,
          "updates": 2464,
          "samples_seen": 157696,
          "online_loss": 0.0022029061219654977,
          "train_loss": 0.0012630055425688624,
          "validation_loss": 0.000995897687971592,
          "train_metric": 0.0001272052904823795,
          "validation_metric": 0.00010030315752374008,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8806440339994879
        },
        {
          "epoch": 155,
          "updates": 2480,
          "samples_seen": 158720,
          "online_loss": 0.0013886513515899424,
          "train_loss": 0.0009062718017958105,
          "validation_loss": 0.0009327945881523192,
          "train_metric": 9.12763862288557e-05,
          "validation_metric": 9.394765220349655e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8920966099994985
        },
        {
          "epoch": 156,
          "updates": 2496,
          "samples_seen": 159744,
          "online_loss": 0.0007176380549935857,
          "train_loss": 0.0006058666622266173,
          "validation_loss": 0.0004526329576037824,
          "train_metric": 6.102068073232658e-05,
          "validation_metric": 4.558754153549671e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9030568229991331
        },
        {
          "epoch": 157,
          "updates": 2512,
          "samples_seen": 160768,
          "online_loss": 0.0006477354745584307,
          "train_loss": 0.0005650856764987111,
          "validation_loss": 0.00041010184213519096,
          "train_metric": 5.691335900337435e-05,
          "validation_metric": 4.130394881940447e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.915006449999055
        },
        {
          "epoch": 158,
          "updates": 2528,
          "samples_seen": 161792,
          "online_loss": 0.0006700488283968298,
          "train_loss": 0.0007781725144013762,
          "validation_loss": 0.0004983928520232439,
          "train_metric": 7.83746872912161e-05,
          "validation_metric": 5.019630407332443e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9270784909986105
        },
        {
          "epoch": 159,
          "updates": 2544,
          "samples_seen": 162816,
          "online_loss": 0.000750048615373089,
          "train_loss": 0.0005620175506919622,
          "validation_loss": 0.0004141048120800406,
          "train_metric": 5.660434908349998e-05,
          "validation_metric": 4.170712054474279e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9386950489988521
        },
        {
          "epoch": 160,
          "updates": 2560,
          "samples_seen": 163840,
          "online_loss": 0.0005836112577526364,
          "train_loss": 0.0005471915937960148,
          "validation_loss": 0.0003894175752066076,
          "train_metric": 5.511113704415038e-05,
          "validation_metric": 3.92207111872267e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9501222210001288
        },
        {
          "epoch": 161,
          "updates": 2576,
          "samples_seen": 164864,
          "online_loss": 0.0006232939740584698,
          "train_loss": 0.0005659210728481412,
          "validation_loss": 0.0005244111525826156,
          "train_metric": 5.699749817722477e-05,
          "validation_metric": 5.281677658786066e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9619293819996528
        },
        {
          "epoch": 162,
          "updates": 2592,
          "samples_seen": 165888,
          "online_loss": 0.0006819831742177485,
          "train_loss": 0.0006371060735546052,
          "validation_loss": 0.0006003886228427291,
          "train_metric": 6.416700489353389e-05,
          "validation_metric": 6.046894486644305e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.973615131999395
        },
        {
          "epoch": 163,
          "updates": 2608,
          "samples_seen": 166912,
          "online_loss": 0.0005703677043129574,
          "train_loss": 0.0005716239684261382,
          "validation_loss": 0.000539539847522974,
          "train_metric": 5.757187318522483e-05,
          "validation_metric": 5.4340478527592495e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9847741829998995
        },
        {
          "epoch": 164,
          "updates": 2624,
          "samples_seen": 167936,
          "online_loss": 0.0005781191048299661,
          "train_loss": 0.0005461715627461672,
          "validation_loss": 0.000458584981970489,
          "train_metric": 5.500840052263811e-05,
          "validation_metric": 4.618700040737167e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9960279619990615
        },
        {
          "epoch": 165,
          "updates": 2640,
          "samples_seen": 168960,
          "online_loss": 0.0005478869752550963,
          "train_loss": 0.0013942476361989975,
          "validation_loss": 0.001187005895189941,
          "train_metric": 0.00014042349357623607,
          "validation_metric": 0.00011955088120885193,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0076061569998274
        },
        {
          "epoch": 166,
          "updates": 2656,
          "samples_seen": 169984,
          "online_loss": 0.000756671734052361,
          "train_loss": 0.0005671640392392874,
          "validation_loss": 0.000506828015204519,
          "train_metric": 5.71226846659556e-05,
          "validation_metric": 5.1045855798292905e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0193788689994108
        },
        {
          "epoch": 167,
          "updates": 2672,
          "samples_seen": 171008,
          "online_loss": 0.0005493979097082047,
          "train_loss": 0.0004830397665500641,
          "validation_loss": 0.00041597665403969586,
          "train_metric": 4.8649992095306516e-05,
          "validation_metric": 4.18956478824839e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.031072955998752
        },
        {
          "epoch": 168,
          "updates": 2688,
          "samples_seen": 172032,
          "online_loss": 0.000533851578438771,
          "train_loss": 0.000749601807910949,
          "validation_loss": 0.0007941676885820925,
          "train_metric": 7.549714791821316e-05,
          "validation_metric": 7.998566434253007e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0423927909996564
        },
        {
          "epoch": 169,
          "updates": 2704,
          "samples_seen": 173056,
          "online_loss": 0.0008156785697792657,
          "train_loss": 0.0006566474330611527,
          "validation_loss": 0.0003837323165498674,
          "train_metric": 6.613512960029766e-05,
          "validation_metric": 3.864811878884211e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0544040059994586
        },
        {
          "epoch": 170,
          "updates": 2720,
          "samples_seen": 174080,
          "online_loss": 0.0006869065728096757,
          "train_loss": 0.0006060584564693272,
          "validation_loss": 0.00037916682776995003,
          "train_metric": 6.103998748585582e-05,
          "validation_metric": 3.818830009549856e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.066276810999625
        },
        {
          "epoch": 171,
          "updates": 2736,
          "samples_seen": 175104,
          "online_loss": 0.0005823257406518678,
          "train_loss": 0.0005655634449794888,
          "validation_loss": 0.00040090386755764484,
          "train_metric": 5.6961478549055755e-05,
          "validation_metric": 4.037757025798783e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.077817067000069
        },
        {
          "epoch": 172,
          "updates": 2752,
          "samples_seen": 176128,
          "online_loss": 0.0005157605664862785,
          "train_loss": 0.00046410353388637304,
          "validation_loss": 0.0004119082004763186,
          "train_metric": 4.674281444749795e-05,
          "validation_metric": 4.148588777752593e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.089433145998555
        },
        {
          "epoch": 173,
          "updates": 2768,
          "samples_seen": 177152,
          "online_loss": 0.0005552261400225689,
          "train_loss": 0.0006200824864208698,
          "validation_loss": 0.0003727242583408952,
          "train_metric": 6.24524473096244e-05,
          "validation_metric": 3.753941928152926e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.10122387499905
        },
        {
          "epoch": 174,
          "updates": 2784,
          "samples_seen": 178176,
          "online_loss": 0.0004994977880414808,
          "train_loss": 0.00048135916586034,
          "validation_loss": 0.00045683211646974087,
          "train_metric": 4.8480735131306574e-05,
          "validation_metric": 4.601046384777874e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.112861507999696
        },
        {
          "epoch": 175,
          "updates": 2800,
          "samples_seen": 179200,
          "online_loss": 0.0006330383484964841,
          "train_loss": 0.0007435468723997474,
          "validation_loss": 0.0007447195239365101,
          "train_metric": 7.488732080673799e-05,
          "validation_metric": 7.500542415073141e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1243892699985736
        },
        {
          "epoch": 176,
          "updates": 2816,
          "samples_seen": 180224,
          "online_loss": 0.0006641671316174325,
          "train_loss": 0.0008348234696313739,
          "validation_loss": 0.000778529210947454,
          "train_metric": 8.408035500906408e-05,
          "validation_metric": 7.841062324587256e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1356192269995518
        },
        {
          "epoch": 177,
          "updates": 2832,
          "samples_seen": 181248,
          "online_loss": 0.0006401525552064413,
          "train_loss": 0.0011279433965682983,
          "validation_loss": 0.0009004672174341977,
          "train_metric": 0.00011360232747392729,
          "validation_metric": 9.069175575859845e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1472573470000498
        },
        {
          "epoch": 178,
          "updates": 2848,
          "samples_seen": 182272,
          "online_loss": 0.0008201586751965806,
          "train_loss": 0.0007919699419289827,
          "validation_loss": 0.0006981806945987046,
          "train_metric": 7.976431515999138e-05,
          "validation_metric": 7.031820132397115e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1588825519993406
        },
        {
          "epoch": 179,
          "updates": 2864,
          "samples_seen": 183296,
          "online_loss": 0.0010207331397396047,
          "train_loss": 0.0011858254438266158,
          "validation_loss": 0.0006720465607941151,
          "train_metric": 0.0001194320066133514,
          "validation_metric": 6.7686072725337e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.170294996998564
        },
        {
          "epoch": 180,
          "updates": 2880,
          "samples_seen": 184320,
          "online_loss": 0.001013374057947658,
          "train_loss": 0.001211507129482925,
          "validation_loss": 0.0009759780950844288,
          "train_metric": 0.00012201855861349031,
          "validation_metric": 9.829693590290844e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1816111659991293
        },
        {
          "epoch": 181,
          "updates": 2896,
          "samples_seen": 185344,
          "online_loss": 0.0007228452850540634,
          "train_loss": 0.0008792683947831392,
          "validation_loss": 0.000756480498239398,
          "train_metric": 8.855669148033485e-05,
          "validation_metric": 7.618994277436286e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1928848739989917
        },
        {
          "epoch": 182,
          "updates": 2912,
          "samples_seen": 186368,
          "online_loss": 0.0006868579002912156,
          "train_loss": 0.0005417202482931316,
          "validation_loss": 0.0004324711626395583,
          "train_metric": 5.4560085118282586e-05,
          "validation_metric": 4.355690907686949e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.2039794499996788
        },
        {
          "epoch": 183,
          "updates": 2928,
          "samples_seen": 187392,
          "online_loss": 0.000535844459591317,
          "train_loss": 0.0007326724007725716,
          "validation_loss": 0.0004578099469654262,
          "train_metric": 7.37920927349478e-05,
          "validation_metric": 4.610894393408671e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.2153193699996336
        },
        {
          "epoch": 184,
          "updates": 2944,
          "samples_seen": 188416,
          "online_loss": 0.0005849303443028475,
          "train_loss": 0.0006607443792745471,
          "validation_loss": 0.0008177367271855474,
          "train_metric": 6.654774188064039e-05,
          "validation_metric": 8.235944551415741e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.2266958259988314
        },
        {
          "epoch": 185,
          "updates": 2960,
          "samples_seen": 189440,
          "online_loss": 0.0006076543013477931,
          "train_loss": 0.00046097577433101833,
          "validation_loss": 0.0004909384879283607,
          "train_metric": 4.6427794586634263e-05,
          "validation_metric": 4.944552347296849e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2379689749996032
        },
        {
          "epoch": 186,
          "updates": 2976,
          "samples_seen": 190464,
          "online_loss": 0.0004607771388691617,
          "train_loss": 0.0004378472513053566,
          "validation_loss": 0.00034841932938434184,
          "train_metric": 4.409837129060179e-05,
          "validation_metric": 3.5091517929686233e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2495943549984077
        },
        {
          "epoch": 187,
          "updates": 2992,
          "samples_seen": 191488,
          "online_loss": 0.000474450613182853,
          "train_loss": 0.0004554066399578005,
          "validation_loss": 0.00033069102209992707,
          "train_metric": 4.586689101415686e-05,
          "validation_metric": 3.330598337925039e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.261669258999973
        },
        {
          "epoch": 188,
          "updates": 3008,
          "samples_seen": 192512,
          "online_loss": 0.00046553830998163903,
          "train_loss": 0.00045311148278415203,
          "validation_loss": 0.00035721538006328046,
          "train_metric": 4.5635722926817834e-05,
          "validation_metric": 3.5977427614852786e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.272971954998866
        },
        {
          "epoch": 189,
          "updates": 3024,
          "samples_seen": 193536,
          "online_loss": 0.0005746538845414761,
          "train_loss": 0.0004785491619259119,
          "validation_loss": 0.00031480257166549563,
          "train_metric": 4.8197718570008874e-05,
          "validation_metric": 3.1705763831268996e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2847402739989775
        },
        {
          "epoch": 190,
          "updates": 3040,
          "samples_seen": 194560,
          "online_loss": 0.00044987810542806983,
          "train_loss": 0.0004147505678702146,
          "validation_loss": 0.0003851095389109105,
          "train_metric": 4.1772163967834786e-05,
          "validation_metric": 3.878682036884129e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2958144599997468
        },
        {
          "epoch": 191,
          "updates": 3056,
          "samples_seen": 195584,
          "online_loss": 0.0005899688985664397,
          "train_loss": 0.0005434472113847733,
          "validation_loss": 0.000579702144023031,
          "train_metric": 5.4734013247070834e-05,
          "validation_metric": 5.838548167957924e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3068063069986238
        },
        {
          "epoch": 192,
          "updates": 3072,
          "samples_seen": 196608,
          "online_loss": 0.0004676149674196495,
          "train_loss": 0.00042147201020270586,
          "validation_loss": 0.00038602171116508543,
          "train_metric": 4.244911906425841e-05,
          "validation_metric": 3.887869024765678e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3177049049991183
        },
        {
          "epoch": 193,
          "updates": 3088,
          "samples_seen": 197632,
          "online_loss": 0.0005697513915947638,
          "train_loss": 0.0004175797803327441,
          "validation_loss": 0.00040155305759981275,
          "train_metric": 4.205710501992144e-05,
          "validation_metric": 4.044294837513007e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3291603210000176
        },
        {
          "epoch": 194,
          "updates": 3104,
          "samples_seen": 198656,
          "online_loss": 0.00046158086570358137,
          "train_loss": 0.00045918222167529166,
          "validation_loss": 0.00046127763926051557,
          "train_metric": 4.6247150748968124e-05,
          "validation_metric": 4.6458189899567515e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.340011901998878
        },
        {
          "epoch": 195,
          "updates": 3120,
          "samples_seen": 199680,
          "online_loss": 0.0004429052250998211,
          "train_loss": 0.0004005955415777862,
          "validation_loss": 0.00035162654239684343,
          "train_metric": 4.03465201088693e-05,
          "validation_metric": 3.5414534067967907e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.35059368399925
        },
        {
          "epoch": 196,
          "updates": 3136,
          "samples_seen": 200704,
          "online_loss": 0.00041881307424773695,
          "train_loss": 0.0005992762744426727,
          "validation_loss": 0.0006487118662334979,
          "train_metric": 6.035690603312105e-05,
          "validation_metric": 6.533588020829484e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3612627779984905
        },
        {
          "epoch": 197,
          "updates": 3152,
          "samples_seen": 201728,
          "online_loss": 0.0005972049766569398,
          "train_loss": 0.0006366623565554619,
          "validation_loss": 0.0006278923247009516,
          "train_metric": 6.412230140995234e-05,
          "validation_metric": 6.323900015559047e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.37174387199957
        },
        {
          "epoch": 198,
          "updates": 3168,
          "samples_seen": 202752,
          "online_loss": 0.0006097884006521781,
          "train_loss": 0.0005167159251868725,
          "validation_loss": 0.0005880731041543186,
          "train_metric": 5.204173794481903e-05,
          "validation_metric": 5.922856871620752e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3823260799999844
        },
        {
          "epoch": 199,
          "updates": 3184,
          "samples_seen": 203776,
          "online_loss": 0.00043761266533692833,
          "train_loss": 0.00038861887878738344,
          "validation_loss": 0.0003382425056770444,
          "train_metric": 3.914027183782309e-05,
          "validation_metric": 3.40665465046186e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3929213439987507
        },
        {
          "epoch": 200,
          "updates": 3200,
          "samples_seen": 204800,
          "online_loss": 0.0004371615241325344,
          "train_loss": 0.00039316638140007854,
          "validation_loss": 0.00033351342426612973,
          "train_metric": 3.959828245569952e-05,
          "validation_metric": 3.359025140525773e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.4035588209990237
        },
        {
          "epoch": 201,
          "updates": 3216,
          "samples_seen": 205824,
          "online_loss": 0.00039865503549663117,
          "train_loss": 0.00038384110666811466,
          "validation_loss": 0.0003690176527015865,
          "train_metric": 3.865906910505146e-05,
          "validation_metric": 3.716610444826074e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4142277569990256
        },
        {
          "epoch": 202,
          "updates": 3232,
          "samples_seen": 206848,
          "online_loss": 0.00038959028643148486,
          "train_loss": 0.00038094230694696307,
          "validation_loss": 0.0003599387709982693,
          "train_metric": 3.836712130578235e-05,
          "validation_metric": 3.6251713027013466e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4251287809984206
        },
        {
          "epoch": 203,
          "updates": 3248,
          "samples_seen": 207872,
          "online_loss": 0.0004009947606391506,
          "train_loss": 0.0003843689919449389,
          "validation_loss": 0.0003351160848978907,
          "train_metric": 3.87122381653171e-05,
          "validation_metric": 3.375166124897078e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4358430020001833
        },
        {
          "epoch": 204,
          "updates": 3264,
          "samples_seen": 208896,
          "online_loss": 0.0004152111459916341,
          "train_loss": 0.0003962987393606454,
          "validation_loss": 0.00042470009066164494,
          "train_metric": 3.991375706391409e-05,
          "validation_metric": 4.2774237954290584e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4462782369992055
        },
        {
          "epoch": 205,
          "updates": 3280,
          "samples_seen": 209920,
          "online_loss": 0.00041572927511879243,
          "train_loss": 0.0004566879360936582,
          "validation_loss": 0.0003152646531816572,
          "train_metric": 4.599593376042321e-05,
          "validation_metric": 3.175230085616931e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.517632222999964
        },
        {
          "epoch": 206,
          "updates": 3296,
          "samples_seen": 210944,
          "online_loss": 0.00041968090135924285,
          "train_loss": 0.0003864103928208351,
          "validation_loss": 0.0003914682602044195,
          "train_metric": 3.891783853759989e-05,
          "validation_metric": 3.94272428820841e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5289016970000375
        },
        {
          "epoch": 207,
          "updates": 3312,
          "samples_seen": 211968,
          "online_loss": 0.00041310251981485635,
          "train_loss": 0.000426959537435323,
          "validation_loss": 0.00047346390783786774,
          "train_metric": 4.300180444261059e-05,
          "validation_metric": 4.768554572365247e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.539761541998814
        },
        {
          "epoch": 208,
          "updates": 3328,
          "samples_seen": 212992,
          "online_loss": 0.0004063301612404757,
          "train_loss": 0.00038775792927481234,
          "validation_loss": 0.00032435194589197636,
          "train_metric": 3.9053556974977255e-05,
          "validation_metric": 3.2667543564457446e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.550140321998697
        },
        {
          "epoch": 209,
          "updates": 3344,
          "samples_seen": 214016,
          "online_loss": 0.0003860991464534891,
          "train_loss": 0.0004119300574529916,
          "validation_loss": 0.00044809235259890556,
          "train_metric": 4.14880778407678e-05,
          "validation_metric": 4.5130222133593634e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5609139329990285
        }
      ],
      "best_validation_metric": 3.1705763831268996e-05,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 33,
      "parameters": 2369,
      "train_samples": 1024,
      "validation_rmse_m_s": 0.0056307871413567925,
      "training_profiles": [
        0,
        1,
        3,
        4,
        6,
        7,
        10,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.29185990057885647,
          "train_loss": 0.08864232152700424,
          "validation_loss": 0.07799669355154037,
          "train_metric": 0.008927728980779648,
          "validation_metric": 0.007855541072785854,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.014395173999218969
        },
        {
          "epoch": 2,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.07088851719163358,
          "train_loss": 0.06481600552797318,
          "validation_loss": 0.0761234238743782,
          "train_metric": 0.006528031080961227,
          "validation_metric": 0.007666872814297676,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02679132999946887
        },
        {
          "epoch": 3,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.04287818376906216,
          "train_loss": 0.037330321967601776,
          "validation_loss": 0.03774002194404602,
          "train_metric": 0.0037597729824483395,
          "validation_metric": 0.003801036858931184,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03911430699918128
        },
        {
          "epoch": 4,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.035872509120963514,
          "train_loss": 0.030262794345617294,
          "validation_loss": 0.029498286545276642,
          "train_metric": 0.0030479575507342815,
          "validation_metric": 0.002970959758386016,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05161909500020556
        },
        {
          "epoch": 5,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.03364982106722891,
          "train_loss": 0.04825809225440025,
          "validation_loss": 0.03874806687235832,
          "train_metric": 0.004860378336161375,
          "validation_metric": 0.0039025633595883846,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0632947209996928
        },
        {
          "epoch": 6,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.02176874369615689,
          "train_loss": 0.015365611761808395,
          "validation_loss": 0.012449651956558228,
          "train_metric": 0.0015475681284442544,
          "validation_metric": 0.0012538832379505038,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07560123000075691
        },
        {
          "epoch": 7,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.016361077869078144,
          "train_loss": 0.03012143261730671,
          "validation_loss": 0.022712912410497665,
          "train_metric": 0.0030337204225361347,
          "validation_metric": 0.0022875615395605564,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08761624799990386
        },
        {
          "epoch": 8,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.015724998142104596,
          "train_loss": 0.020528338849544525,
          "validation_loss": 0.012965889647603035,
          "train_metric": 0.002067538909614086,
          "validation_metric": 0.001305876881815493,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09913177599992196
        },
        {
          "epoch": 9,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.03412122480222024,
          "train_loss": 0.02813313901424408,
          "validation_loss": 0.022576000541448593,
          "train_metric": 0.0028334667440503836,
          "validation_metric": 0.002273772144690156,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.17523870299919508
        },
        {
          "epoch": 10,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.024599767348263413,
          "train_loss": 0.014161516912281513,
          "validation_loss": 0.012120835483074188,
          "train_metric": 0.0014262963086366653,
          "validation_metric": 0.0012207662221044302,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1880602019991784
        },
        {
          "epoch": 11,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.019715111528057605,
          "train_loss": 0.013194785453379154,
          "validation_loss": 0.0139006981626153,
          "train_metric": 0.0013289304915815592,
          "validation_metric": 0.0014000274240970612,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.19967590700071014
        },
        {
          "epoch": 12,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.016061457456089556,
          "train_loss": 0.009087180718779564,
          "validation_loss": 0.007766084745526314,
          "train_metric": 0.0009152275160886347,
          "validation_metric": 0.000782171671744436,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21198688299955393
        },
        {
          "epoch": 13,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.016754681695601903,
          "train_loss": 0.011239994317293167,
          "validation_loss": 0.008611376397311687,
          "train_metric": 0.00113205099478364,
          "validation_metric": 0.0008673063130117953,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22368015800020657
        },
        {
          "epoch": 14,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.0148566659190692,
          "train_loss": 0.011730100959539413,
          "validation_loss": 0.00934852659702301,
          "train_metric": 0.0011814128374680877,
          "validation_metric": 0.0009415495442226529,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2352660119995562
        },
        {
          "epoch": 15,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.022869505191920325,
          "train_loss": 0.03948159143328667,
          "validation_loss": 0.036753736436367035,
          "train_metric": 0.003976441454142332,
          "validation_metric": 0.003701701993122697,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24688067699935345
        },
        {
          "epoch": 16,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.021363710562582128,
          "train_loss": 0.018701840192079544,
          "validation_loss": 0.0171390101313591,
          "train_metric": 0.0018835806986317039,
          "validation_metric": 0.0017261783359572291,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25848770000084187
        },
        {
          "epoch": 17,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.011553934862604365,
          "train_loss": 0.008492785505950451,
          "validation_loss": 0.006388943642377853,
          "train_metric": 0.0008553622174076736,
          "validation_metric": 0.0006434709066525102,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27068972699998994
        },
        {
          "epoch": 18,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.011279518745141104,
          "train_loss": 0.015201610513031483,
          "validation_loss": 0.010612578131258488,
          "train_metric": 0.001531050307676196,
          "validation_metric": 0.001068859826773405,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2825050299998111
        },
        {
          "epoch": 19,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.013181072135921568,
          "train_loss": 0.008281592279672623,
          "validation_loss": 0.009010563604533672,
          "train_metric": 0.0008340917993336916,
          "validation_metric": 0.0009075109846889973,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.29391695000049367
        },
        {
          "epoch": 20,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.008930531999794766,
          "train_loss": 0.01041443645954132,
          "validation_loss": 0.012514898553490639,
          "train_metric": 0.001048903912305832,
          "validation_metric": 0.0012604545336216688,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30521190599938564
        },
        {
          "epoch": 21,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.0073460168787278235,
          "train_loss": 0.005623689852654934,
          "validation_loss": 0.004434810020029545,
          "train_metric": 0.000566397444345057,
          "validation_metric": 0.0004466577956918627,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.31727446900003997
        },
        {
          "epoch": 22,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.010891402227571234,
          "train_loss": 0.013976411893963814,
          "validation_loss": 0.01055215671658516,
          "train_metric": 0.0014076530933380127,
          "validation_metric": 0.0010627746814861894,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3286185230008414
        },
        {
          "epoch": 23,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.012866138393292204,
          "train_loss": 0.012941645458340645,
          "validation_loss": 0.01053465623408556,
          "train_metric": 0.0013034349540248513,
          "validation_metric": 0.0010610120370984077,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.339838306999809
        },
        {
          "epoch": 24,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.016805645514978096,
          "train_loss": 0.007113191299140453,
          "validation_loss": 0.0055458382703363895,
          "train_metric": 0.0007164146518334746,
          "validation_metric": 0.0005585565813817084,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.35134470499906456
        },
        {
          "epoch": 25,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.00971692027815152,
          "train_loss": 0.007082209922373295,
          "validation_loss": 0.006770888809114695,
          "train_metric": 0.0007132942555472255,
          "validation_metric": 0.0006819391273893416,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.362572538000677
        },
        {
          "epoch": 26,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.007690543658100069,
          "train_loss": 0.005337786860764027,
          "validation_loss": 0.004222960211336613,
          "train_metric": 0.0005376023473218083,
          "validation_metric": 0.0004253211081959307,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37443675699978485
        },
        {
          "epoch": 27,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.01507712859165622,
          "train_loss": 0.024322228506207466,
          "validation_loss": 0.02028871327638626,
          "train_metric": 0.002449645660817623,
          "validation_metric": 0.0020434048492461443,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3858348739995563
        },
        {
          "epoch": 28,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.01614690272253938,
          "train_loss": 0.010776660405099392,
          "validation_loss": 0.009896017611026764,
          "train_metric": 0.001085385913029313,
          "validation_metric": 0.0009966908255591989,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3971037930004968
        },
        {
          "epoch": 29,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.009654062814661302,
          "train_loss": 0.010190442204475403,
          "validation_loss": 0.007867034524679184,
          "train_metric": 0.0010263440199196339,
          "validation_metric": 0.0007923389784991741,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4082314030001726
        },
        {
          "epoch": 30,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.009363177072373219,
          "train_loss": 0.0070151640102267265,
          "validation_loss": 0.005032956600189209,
          "train_metric": 0.000706541643012315,
          "validation_metric": 0.0005069009494036436,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4196489890000521
        },
        {
          "epoch": 31,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.011866637709317729,
          "train_loss": 0.0365130640566349,
          "validation_loss": 0.03173527494072914,
          "train_metric": 0.003677461761981249,
          "validation_metric": 0.0031962606590241194,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4308054159992025
        },
        {
          "epoch": 32,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.01852836227044463,
          "train_loss": 0.008202547207474709,
          "validation_loss": 0.007192766759544611,
          "train_metric": 0.000826130504719913,
          "validation_metric": 0.0007244290900416672,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.44199610500072595
        },
        {
          "epoch": 33,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.008492099703289568,
          "train_loss": 0.0058194552548229694,
          "validation_loss": 0.0048090312629938126,
          "train_metric": 0.0005861141835339367,
          "validation_metric": 0.0004843481001444161,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4534566879992781
        },
        {
          "epoch": 34,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.008550057507818565,
          "train_loss": 0.007180484477430582,
          "validation_loss": 0.007885686121881008,
          "train_metric": 0.0007231921190395951,
          "validation_metric": 0.0007942174561321735,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4648561689991766
        },
        {
          "epoch": 35,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.009529258735710755,
          "train_loss": 0.013074851594865322,
          "validation_loss": 0.012272978201508522,
          "train_metric": 0.0013168511213734746,
          "validation_metric": 0.0012360895052552223,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.47626831699926697
        },
        {
          "epoch": 36,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.02614188826555619,
          "train_loss": 0.03761672228574753,
          "validation_loss": 0.032958563417196274,
          "train_metric": 0.0037886181380599737,
          "validation_metric": 0.0033194657880812883,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4877344680007809
        },
        {
          "epoch": 37,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.01746096447459422,
          "train_loss": 0.013114865869283676,
          "validation_loss": 0.012586108408868313,
          "train_metric": 0.0013208811869844794,
          "validation_metric": 0.0012676266487687826,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.49909554599980765
        },
        {
          "epoch": 38,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.010132286930456758,
          "train_loss": 0.0075851790606975555,
          "validation_loss": 0.007952858693897724,
          "train_metric": 0.0007639515097253025,
          "validation_metric": 0.0008009828743524849,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5104762169994501
        },
        {
          "epoch": 39,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.007146781266783364,
          "train_loss": 0.005670350976288319,
          "validation_loss": 0.004081818275153637,
          "train_metric": 0.0005710969562642276,
          "validation_metric": 0.000411105778766796,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5227894159997959
        },
        {
          "epoch": 40,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.006550196587340906,
          "train_loss": 0.007307831197977066,
          "validation_loss": 0.005809362046420574,
          "train_metric": 0.000736017944291234,
          "validation_metric": 0.0005850976449437439,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5344240510003146
        },
        {
          "epoch": 41,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.004882345449004788,
          "train_loss": 0.003895065514370799,
          "validation_loss": 0.002917467150837183,
          "train_metric": 0.0003922967880498618,
          "validation_metric": 0.000293836637865752,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5465595850000682
        },
        {
          "epoch": 42,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.0043821022409247234,
          "train_loss": 0.003544519655406475,
          "validation_loss": 0.002924846950918436,
          "train_metric": 0.0003569910186342895,
          "validation_metric": 0.0002945798623841256,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.558052807000422
        },
        {
          "epoch": 43,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.004237342283886392,
          "train_loss": 0.0032434312161058187,
          "validation_loss": 0.0024500787258148193,
          "train_metric": 0.00032666654442436993,
          "validation_metric": 0.0002467629383318126,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5702095010001358
        },
        {
          "epoch": 44,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.003366431839822326,
          "train_loss": 0.0030133426189422607,
          "validation_loss": 0.0030486395116895437,
          "train_metric": 0.0003034928231500089,
          "validation_metric": 0.000307047797832638,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5819624529995053
        },
        {
          "epoch": 45,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.00468380991515005,
          "train_loss": 0.004118538927286863,
          "validation_loss": 0.002932014875113964,
          "train_metric": 0.00041480411891825497,
          "validation_metric": 0.0002953018120024353,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.593661486000201
        },
        {
          "epoch": 46,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.0028960167910554446,
          "train_loss": 0.0023314799182116985,
          "validation_loss": 0.0018630936974659562,
          "train_metric": 0.0002348181005800143,
          "validation_metric": 0.00018764397827908397,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.605653925000297
        },
        {
          "epoch": 47,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.0027481615761644207,
          "train_loss": 0.002682116348296404,
          "validation_loss": 0.003079255111515522,
          "train_metric": 0.0002701329649426043,
          "validation_metric": 0.0003101313195656985,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6171105350003927
        },
        {
          "epoch": 48,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.0026493419536564033,
          "train_loss": 0.0039250608533620834,
          "validation_loss": 0.004198679234832525,
          "train_metric": 0.00039531770744360983,
          "validation_metric": 0.00042287560063414276,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6283273249991908
        },
        {
          "epoch": 49,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.0030798777152085677,
          "train_loss": 0.002275746315717697,
          "validation_loss": 0.0015838630497455597,
          "train_metric": 0.0002292048593517393,
          "validation_metric": 0.00015952088870108128,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6401451420006197
        },
        {
          "epoch": 50,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.002326219477254199,
          "train_loss": 0.005226192530244589,
          "validation_loss": 0.0031467454973608255,
          "train_metric": 0.000526362971868366,
          "validation_metric": 0.0003169286355841905,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6515991539999959
        },
        {
          "epoch": 51,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.004803637842996977,
          "train_loss": 0.004983479622751474,
          "validation_loss": 0.0034358843695372343,
          "train_metric": 0.0005019177915528417,
          "validation_metric": 0.0003460496664047241,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6628561090001313
        },
        {
          "epoch": 52,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.0028441618123906665,
          "train_loss": 0.002820131601765752,
          "validation_loss": 0.002311078831553459,
          "train_metric": 0.00028403333271853626,
          "validation_metric": 0.00023276341380551457,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6739926440004638
        },
        {
          "epoch": 53,
          "updates": 848,
          "samples_seen": 54272,
          "online_loss": 0.0019362981438462157,
          "train_loss": 0.0018195963930338621,
          "validation_loss": 0.0015617486787959933,
          "train_metric": 0.0001832630659919232,
          "validation_metric": 0.00015729361621197313,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6858090000005177
        },
        {
          "epoch": 54,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0017413093628420029,
          "train_loss": 0.0031078129541128874,
          "validation_loss": 0.0031025882344692945,
          "train_metric": 0.00031300753471441567,
          "validation_metric": 0.0003124813374597579,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6971883860005619
        },
        {
          "epoch": 55,
          "updates": 880,
          "samples_seen": 56320,
          "online_loss": 0.004894348105153767,
          "train_loss": 0.00486629968509078,
          "validation_loss": 0.003968322649598122,
          "train_metric": 0.0004901158972643316,
          "validation_metric": 0.0003996749292127788,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.708444824000253
        },
        {
          "epoch": 56,
          "updates": 896,
          "samples_seen": 57344,
          "online_loss": 0.004196633861283772,
          "train_loss": 0.0034623865503817797,
          "validation_loss": 0.0025353729724884033,
          "train_metric": 0.000348718895111233,
          "validation_metric": 0.00025535348686389625,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7198115539995342
        },
        {
          "epoch": 57,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.003497446079563815,
          "train_loss": 0.0026656114496290684,
          "validation_loss": 0.002255259547382593,
          "train_metric": 0.00026847064145840704,
          "validation_metric": 0.00022714148508384824,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7313098619997618
        },
        {
          "epoch": 58,
          "updates": 928,
          "samples_seen": 59392,
          "online_loss": 0.00200615671928972,
          "train_loss": 0.001776804099790752,
          "validation_loss": 0.0017561281565576792,
          "train_metric": 0.00017895321070682257,
          "validation_metric": 0.00017687080253381282,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7426123770001141
        },
        {
          "epoch": 59,
          "updates": 944,
          "samples_seen": 60416,
          "online_loss": 0.0017039558988471981,
          "train_loss": 0.001447477494366467,
          "validation_loss": 0.001196163590066135,
          "train_metric": 0.00014578463742509484,
          "validation_metric": 0.0001204732179758139,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7546893189992261
        },
        {
          "epoch": 60,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.0017270408061449416,
          "train_loss": 0.0019031675765290856,
          "validation_loss": 0.0018503068713471293,
          "train_metric": 0.00019168008293490857,
          "validation_metric": 0.00018635614833328873,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7662055599994346
        },
        {
          "epoch": 61,
          "updates": 976,
          "samples_seen": 62464,
          "online_loss": 0.0021804726056871004,
          "train_loss": 0.0022888125386089087,
          "validation_loss": 0.001645817537792027,
          "train_metric": 0.00023052080359775573,
          "validation_metric": 0.0001657607062952593,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7775990080008341
        },
        {
          "epoch": 62,
          "updates": 992,
          "samples_seen": 63488,
          "online_loss": 0.001542376310680993,
          "train_loss": 0.0018165635410696268,
          "validation_loss": 0.0017042809631675482,
          "train_metric": 0.00018295763584319502,
          "validation_metric": 0.00017164893506560475,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7889910119993147
        },
        {
          "epoch": 63,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.003223351130145602,
          "train_loss": 0.0023911765310913324,
          "validation_loss": 0.0018426412716507912,
          "train_metric": 0.00024083054449874908,
          "validation_metric": 0.00018558405281510204,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8003568280000763
        },
        {
          "epoch": 64,
          "updates": 1024,
          "samples_seen": 65536,
          "online_loss": 0.006601887587748934,
          "train_loss": 0.00643350463360548,
          "validation_loss": 0.006823212839663029,
          "train_metric": 0.000647959066554904,
          "validation_metric": 0.0006872090161778033,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8117608899992774
        },
        {
          "epoch": 65,
          "updates": 1040,
          "samples_seen": 66560,
          "online_loss": 0.003596680260670837,
          "train_loss": 0.002852036850526929,
          "validation_loss": 0.002473704516887665,
          "train_metric": 0.0002872466866392642,
          "validation_metric": 0.00024914246750995517,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8229551189997437
        },
        {
          "epoch": 66,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.002800502745230915,
          "train_loss": 0.004273152444511652,
          "validation_loss": 0.004459229297935963,
          "train_metric": 0.0004303762107156217,
          "validation_metric": 0.00044911730219610035,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.834310173000631
        },
        {
          "epoch": 67,
          "updates": 1072,
          "samples_seen": 68608,
          "online_loss": 0.00285920418536989,
          "train_loss": 0.001604099990800023,
          "validation_loss": 0.001139243715442717,
          "train_metric": 0.00016155907360371202,
          "validation_metric": 0.00011474044003989547,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8464262569996208
        },
        {
          "epoch": 68,
          "updates": 1088,
          "samples_seen": 69632,
          "online_loss": 0.004064946930157021,
          "train_loss": 0.003867120249196887,
          "validation_loss": 0.004258006811141968,
          "train_metric": 0.00038948218571022153,
          "validation_metric": 0.0004288508207537234,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8577790660001483
        },
        {
          "epoch": 69,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.002006018410611432,
          "train_loss": 0.0022614330518990755,
          "validation_loss": 0.0018566781654953957,
          "train_metric": 0.00022776328842155635,
          "validation_metric": 0.0001869978295871988,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8692520460008382
        },
        {
          "epoch": 70,
          "updates": 1120,
          "samples_seen": 71680,
          "online_loss": 0.0018405326991342008,
          "train_loss": 0.0015012529911473393,
          "validation_loss": 0.0011148350313305855,
          "train_metric": 0.0001512007147539407,
          "validation_metric": 0.0001122821049648337,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8813471749999735
        },
        {
          "epoch": 71,
          "updates": 1136,
          "samples_seen": 72704,
          "online_loss": 0.002162839788070414,
          "train_loss": 0.011232448741793633,
          "validation_loss": 0.008871671743690968,
          "train_metric": 0.0011312910355627537,
          "validation_metric": 0.0008935224614106119,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8928422919998411
        },
        {
          "epoch": 72,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.008454457944026217,
          "train_loss": 0.005071519874036312,
          "validation_loss": 0.004630960524082184,
          "train_metric": 0.0005107849137857556,
          "validation_metric": 0.00046641338849440217,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9043190280008275
        },
        {
          "epoch": 73,
          "updates": 1168,
          "samples_seen": 74752,
          "online_loss": 0.0054149773277458735,
          "train_loss": 0.00334286130964756,
          "validation_loss": 0.0032741427421569824,
          "train_metric": 0.00033668079413473606,
          "validation_metric": 0.0003297596413176507,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9157365350001783
        },
        {
          "epoch": 74,
          "updates": 1184,
          "samples_seen": 75776,
          "online_loss": 0.003592736851715017,
          "train_loss": 0.004045060370117426,
          "validation_loss": 0.004011746495962143,
          "train_metric": 0.0004074036260135472,
          "validation_metric": 0.000404048478230834,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9269981920006103
        },
        {
          "epoch": 75,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.0027096719823020976,
          "train_loss": 0.0019461190095171332,
          "validation_loss": 0.0013754535466432571,
          "train_metric": 0.0001960059889825061,
          "validation_metric": 0.00013853065320290625,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9384447419997741
        },
        {
          "epoch": 76,
          "updates": 1216,
          "samples_seen": 77824,
          "online_loss": 0.004002389523520833,
          "train_loss": 0.0037197202909737825,
          "validation_loss": 0.0027766721323132515,
          "train_metric": 0.0003746365837287158,
          "validation_metric": 0.0002796562621369958,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9499005430006946
        },
        {
          "epoch": 77,
          "updates": 1232,
          "samples_seen": 78848,
          "online_loss": 0.005241013979684794,
          "train_loss": 0.008637193590402603,
          "validation_loss": 0.009357202798128128,
          "train_metric": 0.0008699065074324608,
          "validation_metric": 0.0009424232412129641,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9612497979996988
        },
        {
          "epoch": 78,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.006111270442488603,
          "train_loss": 0.006648399867117405,
          "validation_loss": 0.008012400940060616,
          "train_metric": 0.0006696025957353413,
          "validation_metric": 0.0008069797186180949,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9725154600000678
        },
        {
          "epoch": 79,
          "updates": 1264,
          "samples_seen": 80896,
          "online_loss": 0.0039026886952342466,
          "train_loss": 0.0018770559690892696,
          "validation_loss": 0.0012023645685985684,
          "train_metric": 0.0001890502026071772,
          "validation_metric": 0.0001210977352457121,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9839911779999966
        },
        {
          "epoch": 80,
          "updates": 1280,
          "samples_seen": 81920,
          "online_loss": 0.002046189096290618,
          "train_loss": 0.0015796102816239,
          "validation_loss": 0.001157616963610053,
          "train_metric": 0.00015909253852441907,
          "validation_metric": 0.00011659094161586836,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9953219260005426
        },
        {
          "epoch": 81,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.002322138292583986,
          "train_loss": 0.006456043571233749,
          "validation_loss": 0.006406559608876705,
          "train_metric": 0.0006502292235381901,
          "validation_metric": 0.0006452453089877963,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 1.00653086300008
        },
        {
          "epoch": 82,
          "updates": 1312,
          "samples_seen": 83968,
          "online_loss": 0.003204882908903528,
          "train_loss": 0.0024143466725945473,
          "validation_loss": 0.0018164805369451642,
          "train_metric": 0.00024316417693626136,
          "validation_metric": 0.0001829492684919387,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0179646069991577
        },
        {
          "epoch": 83,
          "updates": 1328,
          "samples_seen": 84992,
          "online_loss": 0.003188109960319707,
          "train_loss": 0.0012616388266906142,
          "validation_loss": 0.0008864965057000518,
          "train_metric": 0.00012706764391623437,
          "validation_metric": 8.928468741942197e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0297563650001393
        },
        {
          "epoch": 84,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.002409474840533221,
          "train_loss": 0.001182090723887086,
          "validation_loss": 0.0009368669707328081,
          "train_metric": 0.00011905584688065574,
          "validation_metric": 9.435780521016568e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.041239811000196
        },
        {
          "epoch": 85,
          "updates": 1360,
          "samples_seen": 87040,
          "online_loss": 0.0013900829162594164,
          "train_loss": 0.0012465763138607144,
          "validation_loss": 0.0008673874544911087,
          "train_metric": 0.00012555062130559236,
          "validation_metric": 8.736010204302147e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0532113080007548
        },
        {
          "epoch": 86,
          "updates": 1376,
          "samples_seen": 88064,
          "online_loss": 0.002419645657937508,
          "train_loss": 0.001348462188616395,
          "validation_loss": 0.0011023604311048985,
          "train_metric": 0.00013581218081526458,
          "validation_metric": 0.00011102570715593174,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0646770029998152
        },
        {
          "epoch": 87,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.001740726176649332,
          "train_loss": 0.0011814566096290946,
          "validation_loss": 0.0009288092260248959,
          "train_metric": 0.00011899197124876082,
          "validation_metric": 9.35462667257525e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0761027829994418
        },
        {
          "epoch": 88,
          "updates": 1408,
          "samples_seen": 90112,
          "online_loss": 0.0017324087930319365,
          "train_loss": 0.001142566092312336,
          "validation_loss": 0.0009018928394652903,
          "train_metric": 0.00011507508315844461,
          "validation_metric": 9.083533223019913e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0873686039994936
        },
        {
          "epoch": 89,
          "updates": 1424,
          "samples_seen": 91136,
          "online_loss": 0.0012147196648584213,
          "train_loss": 0.0013103290693834424,
          "validation_loss": 0.0009823008440434933,
          "train_metric": 0.00013197155203670263,
          "validation_metric": 9.8933742265217e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.098588025999561
        },
        {
          "epoch": 90,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.001178481132228626,
          "train_loss": 0.0009605168015696108,
          "validation_loss": 0.0007663269643671811,
          "train_metric": 9.673972090240568e-05,
          "validation_metric": 7.718164852121845e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1103842289994645
        },
        {
          "epoch": 91,
          "updates": 1456,
          "samples_seen": 93184,
          "online_loss": 0.001083401159121422,
          "train_loss": 0.0010734888492152095,
          "validation_loss": 0.0008485697326250374,
          "train_metric": 0.0001081178561435081,
          "validation_metric": 8.546484605176374e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1217603069999313
        },
        {
          "epoch": 92,
          "updates": 1472,
          "samples_seen": 94208,
          "online_loss": 0.0012365160764602479,
          "train_loss": 0.0009756420040503144,
          "validation_loss": 0.0008622088353149593,
          "train_metric": 9.826308814808726e-05,
          "validation_metric": 8.683853229740635e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1333303749997867
        },
        {
          "epoch": 93,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0025003322880365886,
          "train_loss": 0.0013204857241362333,
          "validation_loss": 0.0013199702370911837,
          "train_metric": 0.0001329944934695959,
          "validation_metric": 0.0001329425722360611,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.144757746000323
        },
        {
          "epoch": 94,
          "updates": 1504,
          "samples_seen": 96256,
          "online_loss": 0.0016845805330376606,
          "train_loss": 0.0014335045125335455,
          "validation_loss": 0.0012157566379755735,
          "train_metric": 0.0001443773362552747,
          "validation_metric": 0.00012244656682014465,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1560114069998235
        },
        {
          "epoch": 95,
          "updates": 1520,
          "samples_seen": 97280,
          "online_loss": 0.0014987282702350058,
          "train_loss": 0.0009721152018755674,
          "validation_loss": 0.0006941081956028938,
          "train_metric": 9.790788317332044e-05,
          "validation_metric": 6.990803376538679e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.167908964000162
        },
        {
          "epoch": 96,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0016023859243432526,
          "train_loss": 0.0011439736699685454,
          "validation_loss": 0.0009673684253357351,
          "train_metric": 0.00011521682608872652,
          "validation_metric": 9.742980182636529e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1793030000008002
        },
        {
          "epoch": 97,
          "updates": 1552,
          "samples_seen": 99328,
          "online_loss": 0.0011114807894045953,
          "train_loss": 0.0007552664028480649,
          "validation_loss": 0.0006622230284847319,
          "train_metric": 7.606767030665651e-05,
          "validation_metric": 6.669667345704511e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1910959380002168
        },
        {
          "epoch": 98,
          "updates": 1568,
          "samples_seen": 100352,
          "online_loss": 0.0009332133922725916,
          "train_loss": 0.0008972630021162331,
          "validation_loss": 0.0007863689679652452,
          "train_metric": 9.036904521053657e-05,
          "validation_metric": 7.920021744212136e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2026548689991614
        },
        {
          "epoch": 99,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0009515583333268296,
          "train_loss": 0.0007746415212750435,
          "validation_loss": 0.0006479820003733039,
          "train_metric": 7.801905303495005e-05,
          "validation_metric": 6.526237848447636e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.214869973999157
        },
        {
          "epoch": 100,
          "updates": 1600,
          "samples_seen": 102400,
          "online_loss": 0.0010600112036627252,
          "train_loss": 0.001861897180788219,
          "validation_loss": 0.001492048497311771,
          "train_metric": 0.00018752345931716263,
          "validation_metric": 0.00015027365589048713,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2262999619997572
        },
        {
          "epoch": 101,
          "updates": 1616,
          "samples_seen": 103424,
          "online_loss": 0.0008684765998623334,
          "train_loss": 0.0007200572872534394,
          "validation_loss": 0.0005225206259638071,
          "train_metric": 7.252152863657102e-05,
          "validation_metric": 5.262636113911867e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2381311200006166
        },
        {
          "epoch": 102,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0007888988984632306,
          "train_loss": 0.0010073272278532386,
          "validation_loss": 0.0008439382072538137,
          "train_metric": 0.00010145430132979527,
          "validation_metric": 8.499836985720322e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2493367980005132
        },
        {
          "epoch": 103,
          "updates": 1648,
          "samples_seen": 105472,
          "online_loss": 0.000710986972990213,
          "train_loss": 0.0007370981038548052,
          "validation_loss": 0.0006512574036605656,
          "train_metric": 7.423782517435029e-05,
          "validation_metric": 6.559227040270343e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2605428049992042
        },
        {
          "epoch": 104,
          "updates": 1664,
          "samples_seen": 106496,
          "online_loss": 0.0008429623976553557,
          "train_loss": 0.0007028741529211402,
          "validation_loss": 0.0004955228650942445,
          "train_metric": 7.079089846229181e-05,
          "validation_metric": 4.990725210518576e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2723517609992996
        },
        {
          "epoch": 105,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0010533280910749454,
          "train_loss": 0.0013825237983837724,
          "validation_loss": 0.0009571844711899757,
          "train_metric": 0.00013924273662269115,
          "validation_metric": 9.640411735745147e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2838351980008156
        },
        {
          "epoch": 106,
          "updates": 1696,
          "samples_seen": 108544,
          "online_loss": 0.0014994533794379095,
          "train_loss": 0.0007348996587097645,
          "validation_loss": 0.0006142295314930379,
          "train_metric": 7.401639595627785e-05,
          "validation_metric": 6.186295649968088e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2951453940004285
        },
        {
          "epoch": 107,
          "updates": 1712,
          "samples_seen": 109568,
          "online_loss": 0.0008827261717669899,
          "train_loss": 0.000786723627243191,
          "validation_loss": 0.0006472434615716338,
          "train_metric": 7.923592056613415e-05,
          "validation_metric": 6.518798909382895e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3064476330000616
        },
        {
          "epoch": 108,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0007599814889545087,
          "train_loss": 0.0008101108251139522,
          "validation_loss": 0.0006061692256480455,
          "train_metric": 8.159139542840421e-05,
          "validation_metric": 6.105114152887836e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3175455959990359
        },
        {
          "epoch": 109,
          "updates": 1744,
          "samples_seen": 111616,
          "online_loss": 0.0023320349646382965,
          "train_loss": 0.0028975699096918106,
          "validation_loss": 0.0021900818683207035,
          "train_metric": 0.0002918326645158231,
          "validation_metric": 0.0002205770433647558,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.328484835999916
        },
        {
          "epoch": 110,
          "updates": 1760,
          "samples_seen": 112640,
          "online_loss": 0.0019870266332873143,
          "train_loss": 0.0046964893117547035,
          "validation_loss": 0.004262920934706926,
          "train_metric": 0.0004730131768155843,
          "validation_metric": 0.00042934578959830105,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3394884329991328
        },
        {
          "epoch": 111,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.0022922488060430624,
          "train_loss": 0.002695011906325817,
          "validation_loss": 0.00146711606066674,
          "train_metric": 0.0002714316942729056,
          "validation_metric": 0.00014776256284676492,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3505050789990491
        },
        {
          "epoch": 112,
          "updates": 1792,
          "samples_seen": 114688,
          "online_loss": 0.0015852987507969374,
          "train_loss": 0.0025053692515939474,
          "validation_loss": 0.002000194974243641,
          "train_metric": 0.00025233160704374313,
          "validation_metric": 0.00020145233429502696,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3614528449998033
        },
        {
          "epoch": 113,
          "updates": 1808,
          "samples_seen": 115712,
          "online_loss": 0.0019907157711713808,
          "train_loss": 0.0011034869821742177,
          "validation_loss": 0.0008189959335140884,
          "train_metric": 0.00011113918299088255,
          "validation_metric": 8.248627273133025e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3723401940005715
        },
        {
          "epoch": 114,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.0008147495500452351,
          "train_loss": 0.0007392112747766078,
          "validation_loss": 0.000594095210544765,
          "train_metric": 7.445065421052277e-05,
          "validation_metric": 5.983509254292585e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.383356599999388
        },
        {
          "epoch": 115,
          "updates": 1840,
          "samples_seen": 117760,
          "online_loss": 0.001026008390908828,
          "train_loss": 0.0020808393601328135,
          "validation_loss": 0.0016173629555851221,
          "train_metric": 0.0002095745294354856,
          "validation_metric": 0.00016289486666209996,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3943253729994467
        },
        {
          "epoch": 116,
          "updates": 1856,
          "samples_seen": 118784,
          "online_loss": 0.0013932040892541409,
          "train_loss": 0.0025870290119200945,
          "validation_loss": 0.0020591802895069122,
          "train_metric": 0.0002605561167001724,
          "validation_metric": 0.00020739308092743158,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4052654040006018
        },
        {
          "epoch": 117,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.0011363911507942248,
          "train_loss": 0.0006742279510945082,
          "validation_loss": 0.000544068228919059,
          "train_metric": 6.790577026549727e-05,
          "validation_metric": 5.4796557378722355e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4164055879991793
        },
        {
          "epoch": 118,
          "updates": 1888,
          "samples_seen": 120832,
          "online_loss": 0.0007147571122914087,
          "train_loss": 0.0006696413620375097,
          "validation_loss": 0.0005471266340464354,
          "train_metric": 6.744382699253038e-05,
          "validation_metric": 5.510459595825523e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4272358430007444
        },
        {
          "epoch": 119,
          "updates": 1904,
          "samples_seen": 121856,
          "online_loss": 0.0007266733236974687,
          "train_loss": 0.0005994972889311612,
          "validation_loss": 0.0004430650733411312,
          "train_metric": 6.0379177739378065e-05,
          "validation_metric": 4.4623891881201416e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4387410430008458
        },
        {
          "epoch": 120,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0006990397778281476,
          "train_loss": 0.0006095686694607139,
          "validation_loss": 0.0004978736396878958,
          "train_metric": 6.13935335422866e-05,
          "validation_metric": 5.0144008127972484e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4498956780007575
        },
        {
          "epoch": 121,
          "updates": 1936,
          "samples_seen": 123904,
          "online_loss": 0.0006894023917993763,
          "train_loss": 0.0005615713889710605,
          "validation_loss": 0.0004286060284357518,
          "train_metric": 5.65594200452324e-05,
          "validation_metric": 4.316762715461664e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4615427419994376
        },
        {
          "epoch": 122,
          "updates": 1952,
          "samples_seen": 124928,
          "online_loss": 0.0009189566262648441,
          "train_loss": 0.000790310325101018,
          "validation_loss": 0.0006748128216713667,
          "train_metric": 7.959716458572075e-05,
          "validation_metric": 6.796466186642647e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4726800759999605
        },
        {
          "epoch": 123,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0015094928075995995,
          "train_loss": 0.001398970140144229,
          "validation_loss": 0.0014995381934568286,
          "train_metric": 0.0001408991520293057,
          "validation_metric": 0.00015102799807209522,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4836809160005942
        },
        {
          "epoch": 124,
          "updates": 1984,
          "samples_seen": 126976,
          "online_loss": 0.0016194682430068497,
          "train_loss": 0.0017759579932317138,
          "validation_loss": 0.0012059161672368646,
          "train_metric": 0.00017886799469124526,
          "validation_metric": 0.00012145545042585582,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4946305670000584
        },
        {
          "epoch": 125,
          "updates": 2000,
          "samples_seen": 128000,
          "online_loss": 0.0013028527355345432,
          "train_loss": 0.000599171151407063,
          "validation_loss": 0.0004957971978001297,
          "train_metric": 6.034631951479241e-05,
          "validation_metric": 4.9934878916246817e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5055570309996256
        },
        {
          "epoch": 126,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.0008587150177845615,
          "train_loss": 0.000691378372721374,
          "validation_loss": 0.0007160317036323249,
          "train_metric": 6.963308260310441e-05,
          "validation_metric": 7.211609045043588e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5165385239997704
        },
        {
          "epoch": 127,
          "updates": 2032,
          "samples_seen": 130048,
          "online_loss": 0.0008113045205391245,
          "train_loss": 0.0007229594048112631,
          "validation_loss": 0.0005817307974211872,
          "train_metric": 7.281383295776322e-05,
          "validation_metric": 5.8589786931406707e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5275317230007204
        },
        {
          "epoch": 128,
          "updates": 2048,
          "samples_seen": 131072,
          "online_loss": 0.0005925567929807585,
          "train_loss": 0.0005365528049878776,
          "validation_loss": 0.000390558474464342,
          "train_metric": 5.403962859418243e-05,
          "validation_metric": 3.933561674784869e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5390493039994908
        },
        {
          "epoch": 129,
          "updates": 2064,
          "samples_seen": 132096,
          "online_loss": 0.0006207485857885331,
          "train_loss": 0.0008947639726102352,
          "validation_loss": 0.0007931815925985575,
          "train_metric": 9.01173334568739e-05,
          "validation_metric": 7.98863620730117e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5503059059992665
        },
        {
          "epoch": 130,
          "updates": 2080,
          "samples_seen": 133120,
          "online_loss": 0.0009855967346084071,
          "train_loss": 0.0006760189426131546,
          "validation_loss": 0.0006068037473596632,
          "train_metric": 6.80861558066681e-05,
          "validation_metric": 6.111504626460373e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5614233519991103
        },
        {
          "epoch": 131,
          "updates": 2096,
          "samples_seen": 134144,
          "online_loss": 0.0010776455128507223,
          "train_loss": 0.0011006335262209177,
          "validation_loss": 0.0006227392004802823,
          "train_metric": 0.00011085177538916469,
          "validation_metric": 6.272000609897077e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5723661600004561
        },
        {
          "epoch": 132,
          "updates": 2112,
          "samples_seen": 135168,
          "online_loss": 0.0010348218611397897,
          "train_loss": 0.0010402373736724257,
          "validation_loss": 0.0007628019666299224,
          "train_metric": 0.00010476889292476699,
          "validation_metric": 7.682662544539198e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5834617460004665
        },
        {
          "epoch": 133,
          "updates": 2128,
          "samples_seen": 136192,
          "online_loss": 0.0006579443906957749,
          "train_loss": 0.00050782086327672,
          "validation_loss": 0.0004292949161026627,
          "train_metric": 5.1145852921763435e-05,
          "validation_metric": 4.32370143244043e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5944194959993183
        },
        {
          "epoch": 134,
          "updates": 2144,
          "samples_seen": 137216,
          "online_loss": 0.0006069316559660365,
          "train_loss": 0.0006670592119917274,
          "validation_loss": 0.0004781922325491905,
          "train_metric": 6.718374788761139e-05,
          "validation_metric": 4.8161764425458387e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.605144453000321
        },
        {
          "epoch": 135,
          "updates": 2160,
          "samples_seen": 138240,
          "online_loss": 0.0005396970027504722,
          "train_loss": 0.00047361961333081126,
          "validation_loss": 0.00040575434104539454,
          "train_metric": 4.7701232688268647e-05,
          "validation_metric": 4.0866088966140524e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6161923309991835
        },
        {
          "epoch": 136,
          "updates": 2176,
          "samples_seen": 139264,
          "online_loss": 0.00049276243771601,
          "train_loss": 0.0004692736256401986,
          "validation_loss": 0.0003673183673527092,
          "train_metric": 4.726352199213579e-05,
          "validation_metric": 3.6994963011238724e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6275661510007922
        },
        {
          "epoch": 137,
          "updates": 2192,
          "samples_seen": 140288,
          "online_loss": 0.0005609916406683624,
          "train_loss": 0.0005332444561645389,
          "validation_loss": 0.000461512798210606,
          "train_metric": 5.3706433391198516e-05,
          "validation_metric": 4.648188769351691e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6383466740007862
        },
        {
          "epoch": 138,
          "updates": 2208,
          "samples_seen": 141312,
          "online_loss": 0.0009738688058860134,
          "train_loss": 0.0008304863586090505,
          "validation_loss": 0.0005294893635436893,
          "train_metric": 8.364355016965419e-05,
          "validation_metric": 5.332822183845565e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6492266219993326
        },
        {
          "epoch": 139,
          "updates": 2224,
          "samples_seen": 142336,
          "online_loss": 0.0008640583346277708,
          "train_loss": 0.0005297320312820375,
          "validation_loss": 0.0004919485654681921,
          "train_metric": 5.3352669056039304e-05,
          "validation_metric": 4.954725591233e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6599221010001202
        },
        {
          "epoch": 140,
          "updates": 2240,
          "samples_seen": 143360,
          "online_loss": 0.0007752852634439478,
          "train_loss": 0.0006801997078582644,
          "validation_loss": 0.0005046036094427109,
          "train_metric": 6.850723730167374e-05,
          "validation_metric": 5.08218290633522e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6707297100001597
        },
        {
          "epoch": 141,
          "updates": 2256,
          "samples_seen": 144384,
          "online_loss": 0.0009336123821412912,
          "train_loss": 0.00047524768160656095,
          "validation_loss": 0.00039751274744048715,
          "train_metric": 4.786520730704069e-05,
          "validation_metric": 4.0036025893641636e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6815172639999219
        },
        {
          "epoch": 142,
          "updates": 2272,
          "samples_seen": 145408,
          "online_loss": 0.0005038714643887943,
          "train_loss": 0.00046782250865362585,
          "validation_loss": 0.00040473006083630025,
          "train_metric": 4.711736983153969e-05,
          "validation_metric": 4.076293043908663e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6922659989995736
        },
        {
          "epoch": 143,
          "updates": 2288,
          "samples_seen": 146432,
          "online_loss": 0.0009458073336645612,
          "train_loss": 0.0008276672451756895,
          "validation_loss": 0.0005718614556826651,
          "train_metric": 8.33596131997183e-05,
          "validation_metric": 5.759578925790265e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7029974130000483
        },
        {
          "epoch": 144,
          "updates": 2304,
          "samples_seen": 147456,
          "online_loss": 0.0010849819245777326,
          "train_loss": 0.00048262931522913277,
          "validation_loss": 0.00037242763210088015,
          "train_metric": 4.860865738010034e-05,
          "validation_metric": 3.750954783754423e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7137908570002764
        },
        {
          "epoch": 145,
          "updates": 2320,
          "samples_seen": 148480,
          "online_loss": 0.0005612927716356353,
          "train_loss": 0.00046312014455907047,
          "validation_loss": 0.00037015456473454833,
          "train_metric": 4.664376683649607e-05,
          "validation_metric": 3.728061346919276e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.72483215200009
        },
        {
          "epoch": 146,
          "updates": 2336,
          "samples_seen": 149504,
          "online_loss": 0.0011360682037775405,
          "train_loss": 0.001337773515842855,
          "validation_loss": 0.0012839667033404112,
          "train_metric": 0.00013473565923050046,
          "validation_metric": 0.00012931643868796527,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7356347479999386
        },
        {
          "epoch": 147,
          "updates": 2352,
          "samples_seen": 150528,
          "online_loss": 0.0015311433526221663,
          "train_loss": 0.0012828308390453458,
          "validation_loss": 0.0009347671875730157,
          "train_metric": 0.00012920201697852463,
          "validation_metric": 9.414632222615182e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7462717599992175
        },
        {
          "epoch": 148,
          "updates": 2368,
          "samples_seen": 151552,
          "online_loss": 0.000770281876611989,
          "train_loss": 0.0005837510107085109,
          "validation_loss": 0.00041505650733597577,
          "train_metric": 5.879326272406615e-05,
          "validation_metric": 4.180297037237324e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7572341649993177
        },
        {
          "epoch": 149,
          "updates": 2384,
          "samples_seen": 152576,
          "online_loss": 0.0005370593953557545,
          "train_loss": 0.0004953070310875773,
          "validation_loss": 0.0003903518372680992,
          "train_metric": 4.9885507905855775e-05,
          "validation_metric": 3.931481478502974e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.767629174000831
        },
        {
          "epoch": 150,
          "updates": 2400,
          "samples_seen": 153600,
          "online_loss": 0.0007483158806280699,
          "train_loss": 0.0010151672177016735,
          "validation_loss": 0.0006456064293161035,
          "train_metric": 0.00010224392462987453,
          "validation_metric": 6.502310861833394e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.777972662999673
        },
        {
          "epoch": 151,
          "updates": 2416,
          "samples_seen": 154624,
          "online_loss": 0.0006709673016302986,
          "train_loss": 0.0004956919583491981,
          "validation_loss": 0.00035641813883557916,
          "train_metric": 4.992427784600295e-05,
          "validation_metric": 3.589713014662266e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7891747939993365
        },
        {
          "epoch": 152,
          "updates": 2432,
          "samples_seen": 155648,
          "online_loss": 0.0009005180236272281,
          "train_loss": 0.00045134377432987094,
          "validation_loss": 0.0003502478648442775,
          "train_metric": 4.5457694795913994e-05,
          "validation_metric": 3.527567969285883e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8010521279993554
        },
        {
          "epoch": 153,
          "updates": 2448,
          "samples_seen": 156672,
          "online_loss": 0.0006482024982688017,
          "train_loss": 0.0008068403694778681,
          "validation_loss": 0.000487127632368356,
          "train_metric": 8.126201282721013e-05,
          "validation_metric": 4.90617094328627e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8125424600002589
        },
        {
          "epoch": 154,
          "updates": 2464,
          "samples_seen": 157696,
          "online_loss": 0.0005660322822222952,
          "train_loss": 0.0005851406021974981,
          "validation_loss": 0.00042071231291629374,
          "train_metric": 5.893321940675378e-05,
          "validation_metric": 4.237260509398766e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8238601670000207
        },
        {
          "epoch": 155,
          "updates": 2480,
          "samples_seen": 158720,
          "online_loss": 0.0006385993783624144,
          "train_loss": 0.0006696469499729574,
          "validation_loss": 0.0005013223271816969,
          "train_metric": 6.744439451722428e-05,
          "validation_metric": 5.049135143053718e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8354448600002797
        },
        {
          "epoch": 156,
          "updates": 2496,
          "samples_seen": 159744,
          "online_loss": 0.0005787700683868024,
          "train_loss": 0.00042351713636890054,
          "validation_loss": 0.00033613224513828754,
          "train_metric": 4.265509778633714e-05,
          "validation_metric": 3.3854012144729495e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8469146999996156
        },
        {
          "epoch": 157,
          "updates": 2512,
          "samples_seen": 160768,
          "online_loss": 0.0005533475232368801,
          "train_loss": 0.0006642845110036433,
          "validation_loss": 0.000497671018820256,
          "train_metric": 6.690430018352345e-05,
          "validation_metric": 5.01235990668647e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8585505129995
        },
        {
          "epoch": 158,
          "updates": 2528,
          "samples_seen": 161792,
          "online_loss": 0.0004856501327594742,
          "train_loss": 0.00041201323620043695,
          "validation_loss": 0.00030797123326919973,
          "train_metric": 4.1496467019896954e-05,
          "validation_metric": 3.101773472735658e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8705901979992632
        },
        {
          "epoch": 159,
          "updates": 2544,
          "samples_seen": 162816,
          "online_loss": 0.00043557483331824187,
          "train_loss": 0.00039729964919388294,
          "validation_loss": 0.0003044081386178732,
          "train_metric": 4.00145654566586e-05,
          "validation_metric": 3.065887722186744e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8826042429991503
        },
        {
          "epoch": 160,
          "updates": 2560,
          "samples_seen": 163840,
          "online_loss": 0.0004202742566121742,
          "train_loss": 0.000393747235648334,
          "validation_loss": 0.00031216494971886277,
          "train_metric": 3.965678479289636e-05,
          "validation_metric": 3.1440111342817545e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8936261300004844
        },
        {
          "epoch": 161,
          "updates": 2576,
          "samples_seen": 164864,
          "online_loss": 0.00044715742842527106,
          "train_loss": 0.000512880040332675,
          "validation_loss": 0.00045719899935647845,
          "train_metric": 5.165539914742112e-05,
          "validation_metric": 4.6047407522564754e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9045802340006048
        },
        {
          "epoch": 162,
          "updates": 2592,
          "samples_seen": 165888,
          "online_loss": 0.0005911850130360108,
          "train_loss": 0.0008396121556870639,
          "validation_loss": 0.0006005523609928787,
          "train_metric": 8.456267096335068e-05,
          "validation_metric": 6.04854358243756e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9156270709991077
        },
        {
          "epoch": 163,
          "updates": 2608,
          "samples_seen": 166912,
          "online_loss": 0.0007345862450165441,
          "train_loss": 0.0008233780390582979,
          "validation_loss": 0.0006545708747580647,
          "train_metric": 8.2927625044249e-05,
          "validation_metric": 6.592596764676273e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9266290239993396
        },
        {
          "epoch": 164,
          "updates": 2624,
          "samples_seen": 167936,
          "online_loss": 0.0006213968390511582,
          "train_loss": 0.0004162225523032248,
          "validation_loss": 0.0003520961618050933,
          "train_metric": 4.192041160422377e-05,
          "validation_metric": 3.546183506841771e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9377729890002229
        },
        {
          "epoch": 165,
          "updates": 2640,
          "samples_seen": 168960,
          "online_loss": 0.0005430035589597537,
          "train_loss": 0.000452179869171232,
          "validation_loss": 0.000346427463227883,
          "train_metric": 4.554189945338294e-05,
          "validation_metric": 3.489090158836916e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9495215619990631
        },
        {
          "epoch": 166,
          "updates": 2656,
          "samples_seen": 169984,
          "online_loss": 0.0004981083984603174,
          "train_loss": 0.0007995184860192239,
          "validation_loss": 0.000686694635078311,
          "train_metric": 8.052456541918218e-05,
          "validation_metric": 6.916137499501929e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9606575180005166
        },
        {
          "epoch": 167,
          "updates": 2672,
          "samples_seen": 171008,
          "online_loss": 0.0005443294812721433,
          "train_loss": 0.0003774148935917765,
          "validation_loss": 0.0003044860786758363,
          "train_metric": 3.8011850847397e-05,
          "validation_metric": 3.066672070417553e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9721131419992162
        },
        {
          "epoch": 168,
          "updates": 2688,
          "samples_seen": 172032,
          "online_loss": 0.0004987595257262001,
          "train_loss": 0.0005878148949705064,
          "validation_loss": 0.0003850861394312233,
          "train_metric": 5.9202568081673235e-05,
          "validation_metric": 3.87844665965531e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9832434559993999
        },
        {
          "epoch": 169,
          "updates": 2704,
          "samples_seen": 173056,
          "online_loss": 0.00046830081282678293,
          "train_loss": 0.0003782385610975325,
          "validation_loss": 0.0003154940204694867,
          "train_metric": 3.8094804040156305e-05,
          "validation_metric": 3.1775402021594346e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9946787249991758
        },
        {
          "epoch": 170,
          "updates": 2720,
          "samples_seen": 174080,
          "online_loss": 0.00042610081891325535,
          "train_loss": 0.00045316043542698026,
          "validation_loss": 0.0003786468005273491,
          "train_metric": 4.564065966405906e-05,
          "validation_metric": 3.813592411461286e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.006281845999183
        },
        {
          "epoch": 171,
          "updates": 2736,
          "samples_seen": 175104,
          "online_loss": 0.00043618199833872495,
          "train_loss": 0.00040613749297335744,
          "validation_loss": 0.00034454328124411404,
          "train_metric": 4.0904677007347345e-05,
          "validation_metric": 3.470113733783364e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.017804936000175
        },
        {
          "epoch": 172,
          "updates": 2752,
          "samples_seen": 176128,
          "online_loss": 0.00038705698079866124,
          "train_loss": 0.0003860287833958864,
          "validation_loss": 0.00031760489218868315,
          "train_metric": 3.887940329150297e-05,
          "validation_metric": 3.198800186510198e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0292790349994902
        },
        {
          "epoch": 173,
          "updates": 2768,
          "samples_seen": 177152,
          "online_loss": 0.00038788706115155946,
          "train_loss": 0.00036845417344011366,
          "validation_loss": 0.00029599620029330254,
          "train_metric": 3.710935197887011e-05,
          "validation_metric": 2.9811653803335503e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0409527110005
        },
        {
          "epoch": 174,
          "updates": 2784,
          "samples_seen": 178176,
          "online_loss": 0.00037430060729093384,
          "train_loss": 0.00036074419040232897,
          "validation_loss": 0.0003018274437636137,
          "train_metric": 3.6332832678454e-05,
          "validation_metric": 3.039895455003716e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.052644359999249
        },
        {
          "epoch": 175,
          "updates": 2800,
          "samples_seen": 179200,
          "online_loss": 0.0003831859849015018,
          "train_loss": 0.0004304730973672122,
          "validation_loss": 0.0003429660282563418,
          "train_metric": 4.335567427915521e-05,
          "validation_metric": 3.454227771726437e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0643758080004773
        },
        {
          "epoch": 176,
          "updates": 2816,
          "samples_seen": 180224,
          "online_loss": 0.0004037767839690787,
          "train_loss": 0.00038395292358472943,
          "validation_loss": 0.00032373025896959007,
          "train_metric": 3.8670332287438214e-05,
          "validation_metric": 3.2604926673229784e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.075685940000767
        },
        {
          "epoch": 177,
          "updates": 2832,
          "samples_seen": 181248,
          "online_loss": 0.00038054288506828016,
          "train_loss": 0.00041513709584251046,
          "validation_loss": 0.0003306087455712259,
          "train_metric": 4.181108306511305e-05,
          "validation_metric": 3.329770333948545e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.086869016999117
        },
        {
          "epoch": 178,
          "updates": 2848,
          "samples_seen": 182272,
          "online_loss": 0.00035288369144836906,
          "train_loss": 0.0003822901926469058,
          "validation_loss": 0.00031757724354974926,
          "train_metric": 3.850287612294778e-05,
          "validation_metric": 3.1985215173335746e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0986151699999027
        },
        {
          "epoch": 179,
          "updates": 2864,
          "samples_seen": 183296,
          "online_loss": 0.0003871391490974929,
          "train_loss": 0.00039162763278000057,
          "validation_loss": 0.0003387137548997998,
          "train_metric": 3.9443300920538604e-05,
          "validation_metric": 3.4114011214114726e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.110506212000473
        },
        {
          "epoch": 180,
          "updates": 2880,
          "samples_seen": 184320,
          "online_loss": 0.00038058572226873366,
          "train_loss": 0.0003520049504004419,
          "validation_loss": 0.0002939888508990407,
          "train_metric": 3.545264189597219e-05,
          "validation_metric": 2.9609480407088995e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1226859069993225
        },
        {
          "epoch": 181,
          "updates": 2896,
          "samples_seen": 185344,
          "online_loss": 0.00036845102931692963,
          "train_loss": 0.00034703267738223076,
          "validation_loss": 0.0002906251756940037,
          "train_metric": 3.495185956126079e-05,
          "validation_metric": 2.9270704544615e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1346188410007017
        },
        {
          "epoch": 182,
          "updates": 2912,
          "samples_seen": 186368,
          "online_loss": 0.0003569739719750942,
          "train_loss": 0.0003463371249381453,
          "validation_loss": 0.0002855375350918621,
          "train_metric": 3.488179936539382e-05,
          "validation_metric": 2.8758297048625536e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.146808048000821
        },
        {
          "epoch": 183,
          "updates": 2928,
          "samples_seen": 187392,
          "online_loss": 0.000365363885066472,
          "train_loss": 0.0003448390634730458,
          "validation_loss": 0.0002897469385061413,
          "train_metric": 3.473092874628492e-05,
          "validation_metric": 2.9182250727899373e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1584906009993574
        },
        {
          "epoch": 184,
          "updates": 2944,
          "samples_seen": 188416,
          "online_loss": 0.00036684070437331684,
          "train_loss": 0.00038811645936220884,
          "validation_loss": 0.0003060937742702663,
          "train_metric": 3.908966391463764e-05,
          "validation_metric": 3.082864350290038e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1696119619991805
        },
        {
          "epoch": 185,
          "updates": 2960,
          "samples_seen": 189440,
          "online_loss": 0.0003705961344167008,
          "train_loss": 0.0003440105065237731,
          "validation_loss": 0.00028833633405156434,
          "train_metric": 3.464747715042904e-05,
          "validation_metric": 2.9040180379524827e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.181143300000258
        },
        {
          "epoch": 186,
          "updates": 2976,
          "samples_seen": 190464,
          "online_loss": 0.0003580278589652153,
          "train_loss": 0.00034279286046512425,
          "validation_loss": 0.0002884228597395122,
          "train_metric": 3.452484088484198e-05,
          "validation_metric": 2.9048895157757215e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.192705252999076
        },
        {
          "epoch": 187,
          "updates": 2992,
          "samples_seen": 191488,
          "online_loss": 0.0003501597257127287,
          "train_loss": 0.00034095506998710334,
          "validation_loss": 0.0002925455628428608,
          "train_metric": 3.433974416111596e-05,
          "validation_metric": 2.9464124963851646e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.204143832999762
        },
        {
          "epoch": 188,
          "updates": 3008,
          "samples_seen": 192512,
          "online_loss": 0.0003717329191204044,
          "train_loss": 0.00034485282958485186,
          "validation_loss": 0.00028974845190532506,
          "train_metric": 3.473231481621042e-05,
          "validation_metric": 2.9182399885030463e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.215501303000565
        },
        {
          "epoch": 189,
          "updates": 3024,
          "samples_seen": 193536,
          "online_loss": 0.00037997535491740564,
          "train_loss": 0.0003546858497429639,
          "validation_loss": 0.0002983511658385396,
          "train_metric": 3.572266359697096e-05,
          "validation_metric": 3.004883365065325e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2270849969991104
        },
        {
          "epoch": 190,
          "updates": 3040,
          "samples_seen": 194560,
          "online_loss": 0.0003637500758486567,
          "train_loss": 0.0006160378106869757,
          "validation_loss": 0.0004455748130567372,
          "train_metric": 6.204507371876389e-05,
          "validation_metric": 4.4876662286696956e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2386928430005355
        },
        {
          "epoch": 191,
          "updates": 3056,
          "samples_seen": 195584,
          "online_loss": 0.00047215270024025813,
          "train_loss": 0.00035751756513491273,
          "validation_loss": 0.0003169278788845986,
          "train_metric": 3.600785930757411e-05,
          "validation_metric": 3.191981159034185e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.250143355000546
        },
        {
          "epoch": 192,
          "updates": 3072,
          "samples_seen": 196608,
          "online_loss": 0.00035820600987790385,
          "train_loss": 0.0004252410726621747,
          "validation_loss": 0.00039025270962156355,
          "train_metric": 4.282873123884201e-05,
          "validation_metric": 3.9304824895225465e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.261719202000677
        },
        {
          "epoch": 193,
          "updates": 3088,
          "samples_seen": 197632,
          "online_loss": 0.0003759943174372893,
          "train_loss": 0.00033787995926104486,
          "validation_loss": 0.00029026702395640314,
          "train_metric": 3.403003211133182e-05,
          "validation_metric": 2.9234633984742686e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.272820631000286
        },
        {
          "epoch": 194,
          "updates": 3104,
          "samples_seen": 198656,
          "online_loss": 0.0003564622729754774,
          "train_loss": 0.0003461863088887185,
          "validation_loss": 0.0002849132288247347,
          "train_metric": 3.4866618079831824e-05,
          "validation_metric": 2.8695414584944956e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.2844825079992006
        },
        {
          "epoch": 195,
          "updates": 3120,
          "samples_seen": 199680,
          "online_loss": 0.00035090030814899364,
          "train_loss": 0.0003360644623171538,
          "validation_loss": 0.0002927777823060751,
          "train_metric": 3.384718002052978e-05,
          "validation_metric": 2.948750443465542e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.29610682800012
        },
        {
          "epoch": 196,
          "updates": 3136,
          "samples_seen": 200704,
          "online_loss": 0.0003418601163502899,
          "train_loss": 0.00033858398091979325,
          "validation_loss": 0.00028106357785873115,
          "train_metric": 3.410093631828204e-05,
          "validation_metric": 2.8307691536610946e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3081191689998377
        },
        {
          "epoch": 197,
          "updates": 3152,
          "samples_seen": 201728,
          "online_loss": 0.0003583097204682417,
          "train_loss": 0.0003297113289590925,
          "validation_loss": 0.00028372122324071825,
          "train_metric": 3.3207321394002065e-05,
          "validation_metric": 2.8575363103300333e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.319361354000648
        },
        {
          "epoch": 198,
          "updates": 3168,
          "samples_seen": 202752,
          "online_loss": 0.0003608279766922351,
          "train_loss": 0.00033694779267534614,
          "validation_loss": 0.00028370757354423404,
          "train_metric": 3.39361431542784e-05,
          "validation_metric": 2.8573987947311252e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.330532599000435
        },
        {
          "epoch": 199,
          "updates": 3184,
          "samples_seen": 203776,
          "online_loss": 0.00033299516962870257,
          "train_loss": 0.00033036296372301877,
          "validation_loss": 0.0002812651509884745,
          "train_metric": 3.327294689370319e-05,
          "validation_metric": 2.8327995096333325e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.341785546999745
        },
        {
          "epoch": 200,
          "updates": 3200,
          "samples_seen": 204800,
          "online_loss": 0.00033863281805679435,
          "train_loss": 0.0003271525783929974,
          "validation_loss": 0.0002743976656347513,
          "train_metric": 3.2949610613286495e-05,
          "validation_metric": 2.763632801361382e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3536158579991024
        },
        {
          "epoch": 201,
          "updates": 3216,
          "samples_seen": 205824,
          "online_loss": 0.00036135858408670174,
          "train_loss": 0.0003503996704239398,
          "validation_loss": 0.00030543902539648116,
          "train_metric": 3.529097739374265e-05,
          "validation_metric": 3.0762697861064225e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.364998552000543
        },
        {
          "epoch": 202,
          "updates": 3232,
          "samples_seen": 206848,
          "online_loss": 0.0003534067673172103,
          "train_loss": 0.00032856446341611445,
          "validation_loss": 0.0002821308153215796,
          "train_metric": 3.3091811928898096e-05,
          "validation_metric": 2.8415188353392296e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3761934480007767
        },
        {
          "epoch": 203,
          "updates": 3248,
          "samples_seen": 207872,
          "online_loss": 0.00034584438526508166,
          "train_loss": 0.00033882010029628873,
          "validation_loss": 0.00027868893812410533,
          "train_metric": 3.4124717785743997e-05,
          "validation_metric": 2.8068530809832737e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4441968589999306
        },
        {
          "epoch": 204,
          "updates": 3264,
          "samples_seen": 208896,
          "online_loss": 0.0003491980323815369,
          "train_loss": 0.00034123024670407176,
          "validation_loss": 0.0002847195719368756,
          "train_metric": 3.4367458283668384e-05,
          "validation_metric": 2.8675916837528348e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.455282922999686
        },
        {
          "epoch": 205,
          "updates": 3280,
          "samples_seen": 209920,
          "online_loss": 0.00033888541383930715,
          "train_loss": 0.00032751428079791367,
          "validation_loss": 0.0002870220923796296,
          "train_metric": 3.298604133306071e-05,
          "validation_metric": 2.8907808882649988e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4659237150008266
        },
        {
          "epoch": 206,
          "updates": 3296,
          "samples_seen": 210944,
          "online_loss": 0.00036009812174597755,
          "train_loss": 0.00033711380092427135,
          "validation_loss": 0.00028054608264937997,
          "train_metric": 3.3952870580833405e-05,
          "validation_metric": 2.8255575671209954e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4765785920008057
        },
        {
          "epoch": 207,
          "updates": 3312,
          "samples_seen": 211968,
          "online_loss": 0.00033929739674931625,
          "train_loss": 0.00032389763509854674,
          "validation_loss": 0.00027900884742848575,
          "train_metric": 3.262178142904304e-05,
          "validation_metric": 2.810074875014834e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.487273640999774
        },
        {
          "epoch": 208,
          "updates": 3328,
          "samples_seen": 212992,
          "online_loss": 0.00034144965229643276,
          "train_loss": 0.0003477930149529129,
          "validation_loss": 0.0003101509064435959,
          "train_metric": 3.502844265312888e-05,
          "validation_metric": 3.123726855847053e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.497933765000198
        },
        {
          "epoch": 209,
          "updates": 3344,
          "samples_seen": 214016,
          "online_loss": 0.0003775256982407882,
          "train_loss": 0.0003540377947501838,
          "validation_loss": 0.0002878256782423705,
          "train_metric": 3.565739098121412e-05,
          "validation_metric": 2.8988746635150164e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5086897059991315
        },
        {
          "epoch": 210,
          "updates": 3360,
          "samples_seen": 215040,
          "online_loss": 0.0003269510334575898,
          "train_loss": 0.0003249368746764958,
          "validation_loss": 0.0002857105282600969,
          "train_metric": 3.272645335528068e-05,
          "validation_metric": 2.877572478610091e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.519566771999962
        },
        {
          "epoch": 211,
          "updates": 3376,
          "samples_seen": 216064,
          "online_loss": 0.0003416344461584231,
          "train_loss": 0.00032465183176100254,
          "validation_loss": 0.000269800191745162,
          "train_metric": 3.269773878855631e-05,
          "validation_metric": 2.7173280614078976e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5305752389995178
        },
        {
          "epoch": 212,
          "updates": 3392,
          "samples_seen": 217088,
          "online_loss": 0.0003464596366029582,
          "train_loss": 0.00033260113559663296,
          "validation_loss": 0.0002850257442332804,
          "train_metric": 3.3498366974527016e-05,
          "validation_metric": 2.870674870791845e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5412354899999627
        },
        {
          "epoch": 213,
          "updates": 3408,
          "samples_seen": 218112,
          "online_loss": 0.0003447743747528875,
          "train_loss": 0.00033490086207166314,
          "validation_loss": 0.0002992729714605957,
          "train_metric": 3.372999344719574e-05,
          "validation_metric": 3.0141682145767845e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5524565510004322
        },
        {
          "epoch": 214,
          "updates": 3424,
          "samples_seen": 219136,
          "online_loss": 0.00036395462575455895,
          "train_loss": 0.00039553173701278865,
          "validation_loss": 0.0003079881425946951,
          "train_metric": 3.983651185990311e-05,
          "validation_metric": 3.10194373014383e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.563397473000805
        },
        {
          "epoch": 215,
          "updates": 3440,
          "samples_seen": 220160,
          "online_loss": 0.00036371466012496967,
          "train_loss": 0.00033946154871955514,
          "validation_loss": 0.00029862625524401665,
          "train_metric": 3.418932465137914e-05,
          "validation_metric": 3.0076542316237465e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5739284150004096
        },
        {
          "epoch": 216,
          "updates": 3456,
          "samples_seen": 221184,
          "online_loss": 0.00034533067355368985,
          "train_loss": 0.000326578039675951,
          "validation_loss": 0.00028720925911329687,
          "train_metric": 3.2891744922380894e-05,
          "validation_metric": 2.8926660888828337e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.584852530000717
        },
        {
          "epoch": 217,
          "updates": 3472,
          "samples_seen": 222208,
          "online_loss": 0.0003766623303818051,
          "train_loss": 0.0003248975845053792,
          "validation_loss": 0.0002694190479815006,
          "train_metric": 3.2722498872317374e-05,
          "validation_metric": 2.713489629968535e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.596422246999282
        },
        {
          "epoch": 218,
          "updates": 3488,
          "samples_seen": 223232,
          "online_loss": 0.0003424904334679013,
          "train_loss": 0.00032872575684450567,
          "validation_loss": 0.0002729239931795746,
          "train_metric": 3.3108051866292953e-05,
          "validation_metric": 2.7487907573231496e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.607312161999289
        },
        {
          "epoch": 219,
          "updates": 3504,
          "samples_seen": 224256,
          "online_loss": 0.0003638398893599515,
          "train_loss": 0.0003178014885634184,
          "validation_loss": 0.0002678050077520311,
          "train_metric": 3.200780338374898e-05,
          "validation_metric": 2.6972336854669265e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.618841624000197
        },
        {
          "epoch": 220,
          "updates": 3520,
          "samples_seen": 225280,
          "online_loss": 0.00032403825935034547,
          "train_loss": 0.0003272750764153898,
          "validation_loss": 0.00029358951724134386,
          "train_metric": 3.296195063740015e-05,
          "validation_metric": 2.9569264370366e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6296739150002395
        },
        {
          "epoch": 221,
          "updates": 3536,
          "samples_seen": 226304,
          "online_loss": 0.00033900026210176293,
          "train_loss": 0.0003199098282493651,
          "validation_loss": 0.00026373742730356753,
          "train_metric": 3.222015220671892e-05,
          "validation_metric": 2.656266588019207e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.640668199999709
        },
        {
          "epoch": 222,
          "updates": 3552,
          "samples_seen": 227328,
          "online_loss": 0.00031862364721746417,
          "train_loss": 0.0003257113858126104,
          "validation_loss": 0.000289373827399686,
          "train_metric": 3.280445889686234e-05,
          "validation_metric": 2.9144675863790326e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6516500630004884
        },
        {
          "epoch": 223,
          "updates": 3568,
          "samples_seen": 228352,
          "online_loss": 0.0003394467985344818,
          "train_loss": 0.0003246061969548464,
          "validation_loss": 0.0002863643749151379,
          "train_metric": 3.269314765930176e-05,
          "validation_metric": 2.8841568564530462e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6619523389999813
        },
        {
          "epoch": 224,
          "updates": 3584,
          "samples_seen": 229376,
          "online_loss": 0.00036236820960766636,
          "train_loss": 0.00034154916647821665,
          "validation_loss": 0.0002810238511301577,
          "train_metric": 3.439958527451381e-05,
          "validation_metric": 2.8303693397901952e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6724044069997035
        },
        {
          "epoch": 225,
          "updates": 3600,
          "samples_seen": 230400,
          "online_loss": 0.0003272373960498953,
          "train_loss": 0.00031415181001648307,
          "validation_loss": 0.00027221080381423235,
          "train_metric": 3.164022200508043e-05,
          "validation_metric": 2.7416073862696066e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6832938970001123
        },
        {
          "epoch": 226,
          "updates": 3616,
          "samples_seen": 231424,
          "online_loss": 0.00031888331614027265,
          "train_loss": 0.0003116403822787106,
          "validation_loss": 0.0002666247310116887,
          "train_metric": 3.138728061458096e-05,
          "validation_metric": 2.6853467716136947e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.693631400999948
        },
        {
          "epoch": 227,
          "updates": 3632,
          "samples_seen": 232448,
          "online_loss": 0.0003325392799524707,
          "train_loss": 0.0003209073911421001,
          "validation_loss": 0.00027121370658278465,
          "train_metric": 3.232061499147676e-05,
          "validation_metric": 2.731565291469451e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7039542780003103
        },
        {
          "epoch": 228,
          "updates": 3648,
          "samples_seen": 233472,
          "online_loss": 0.00033898102628882043,
          "train_loss": 0.0003217904013581574,
          "validation_loss": 0.0002717441529966891,
          "train_metric": 3.240955265937373e-05,
          "validation_metric": 2.736907481448725e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7147631459993136
        },
        {
          "epoch": 229,
          "updates": 3664,
          "samples_seen": 234496,
          "online_loss": 0.0003195978260919219,
          "train_loss": 0.00031945868977345526,
          "validation_loss": 0.00027865287847816944,
          "train_metric": 3.217471021343954e-05,
          "validation_metric": 2.806489465001505e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7251241489993845
        },
        {
          "epoch": 230,
          "updates": 3680,
          "samples_seen": 235520,
          "online_loss": 0.0003143031490253634,
          "train_loss": 0.00031227851286530495,
          "validation_loss": 0.00026362627977505326,
          "train_metric": 3.145154914818704e-05,
          "validation_metric": 2.6551475457381457e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.735929540000143
        },
        {
          "epoch": 231,
          "updates": 3696,
          "samples_seen": 236544,
          "online_loss": 0.000316715641929477,
          "train_loss": 0.00031198744545690715,
          "validation_loss": 0.0002752849832177162,
          "train_metric": 3.1422234314959496e-05,
          "validation_metric": 2.7725694963010028e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7464793439994537
        },
        {
          "epoch": 232,
          "updates": 3712,
          "samples_seen": 237568,
          "online_loss": 0.00036944311250408646,
          "train_loss": 0.0004324381588958204,
          "validation_loss": 0.00032073090551421046,
          "train_metric": 4.3553583964239806e-05,
          "validation_metric": 3.2302843465004116e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.756980479000049
        },
        {
          "epoch": 233,
          "updates": 3728,
          "samples_seen": 238592,
          "online_loss": 0.00036275170805311063,
          "train_loss": 0.0003121591289527714,
          "validation_loss": 0.00025871366960927844,
          "train_metric": 3.1439521990250796e-05,
          "validation_metric": 2.6056693968712352e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7678273189994798
        },
        {
          "epoch": 234,
          "updates": 3744,
          "samples_seen": 239616,
          "online_loss": 0.0003166898959534592,
          "train_loss": 0.0003083817136939615,
          "validation_loss": 0.00027612425037659705,
          "train_metric": 3.105907671852037e-05,
          "validation_metric": 2.781022703857161e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7790581909994216
        },
        {
          "epoch": 235,
          "updates": 3760,
          "samples_seen": 240640,
          "online_loss": 0.0003102806931565283,
          "train_loss": 0.00030665844678878784,
          "validation_loss": 0.0002584474568720907,
          "train_metric": 3.088551602559164e-05,
          "validation_metric": 2.6029882064904086e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7901093649998074
        },
        {
          "epoch": 236,
          "updates": 3776,
          "samples_seen": 241664,
          "online_loss": 0.0003137214871458127,
          "train_loss": 0.0003082009789068252,
          "validation_loss": 0.0002726767270360142,
          "train_metric": 3.1040872272569686e-05,
          "validation_metric": 2.7463005608296953e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.800643979999222
        },
        {
          "epoch": 237,
          "updates": 3792,
          "samples_seen": 242688,
          "online_loss": 0.00031096237762540113,
          "train_loss": 0.00032580186962150037,
          "validation_loss": 0.00026580438134260476,
          "train_metric": 3.281357203377411e-05,
          "validation_metric": 2.6770843760459684e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.811260948999916
        },
        {
          "epoch": 238,
          "updates": 3808,
          "samples_seen": 243712,
          "online_loss": 0.0003349291619088035,
          "train_loss": 0.00032519770320504904,
          "validation_loss": 0.00028535741148516536,
          "train_metric": 3.2752719562267885e-05,
          "validation_metric": 2.874015444831457e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.821669667999231
        },
        {
          "epoch": 239,
          "updates": 3824,
          "samples_seen": 244736,
          "online_loss": 0.0002926894421761972,
          "train_loss": 0.0003303418925497681,
          "validation_loss": 0.0002721563505474478,
          "train_metric": 3.327082595205866e-05,
          "validation_metric": 2.741059142863378e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8322169889997895
        },
        {
          "epoch": 240,
          "updates": 3840,
          "samples_seen": 245760,
          "online_loss": 0.00034137490729335696,
          "train_loss": 0.0003099536697845906,
          "validation_loss": 0.0002609715156722814,
          "train_metric": 3.12173979182262e-05,
          "validation_metric": 2.6284093110007234e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.842565208000451
        },
        {
          "epoch": 241,
          "updates": 3856,
          "samples_seen": 246784,
          "online_loss": 0.0003193330639987835,
          "train_loss": 0.000303062261082232,
          "validation_loss": 0.0002672867849469185,
          "train_metric": 3.05233224935364e-05,
          "validation_metric": 2.692014822969213e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.852977942000507
        },
        {
          "epoch": 242,
          "updates": 3872,
          "samples_seen": 247808,
          "online_loss": 0.0003128346779703861,
          "train_loss": 0.00030096498085185885,
          "validation_loss": 0.0002557356492616236,
          "train_metric": 3.031209053006023e-05,
          "validation_metric": 2.5756758986972272e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.864333778999935
        },
        {
          "epoch": 243,
          "updates": 3888,
          "samples_seen": 248832,
          "online_loss": 0.0003069950907956809,
          "train_loss": 0.00030916984542272985,
          "validation_loss": 0.00027381573454476893,
          "train_metric": 3.11384501401335e-05,
          "validation_metric": 2.7577721994020976e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.87468108900066
        },
        {
          "epoch": 244,
          "updates": 3904,
          "samples_seen": 249856,
          "online_loss": 0.00032497277334186947,
          "train_loss": 0.00032224037568084896,
          "validation_loss": 0.0002728531835600734,
          "train_metric": 3.245487096137367e-05,
          "validation_metric": 2.7480775315780193e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8852165970001806
        },
        {
          "epoch": 245,
          "updates": 3920,
          "samples_seen": 250880,
          "online_loss": 0.0003645023243734613,
          "train_loss": 0.00030244255322031677,
          "validation_loss": 0.00027258932823315263,
          "train_metric": 3.046090751013253e-05,
          "validation_metric": 2.7454199880594388e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.895717641000374
        },
        {
          "epoch": 246,
          "updates": 3936,
          "samples_seen": 251904,
          "online_loss": 0.00032997755170072196,
          "train_loss": 0.00034823836176656187,
          "validation_loss": 0.000281729269772768,
          "train_metric": 3.50732880178839e-05,
          "validation_metric": 2.8374737667036243e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9059288789994753
        },
        {
          "epoch": 247,
          "updates": 3952,
          "samples_seen": 252928,
          "online_loss": 0.0003257692196712014,
          "train_loss": 0.00029673828976228833,
          "validation_loss": 0.0002608155773486942,
          "train_metric": 2.98863942589378e-05,
          "validation_metric": 2.6268384317518212e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.916491775000395
        },
        {
          "epoch": 248,
          "updates": 3968,
          "samples_seen": 253952,
          "online_loss": 0.000335813636866078,
          "train_loss": 0.0002968614862766117,
          "validation_loss": 0.00026372328284196556,
          "train_metric": 2.9898796128691174e-05,
          "validation_metric": 2.65612434304785e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9269532519992936
        },
        {
          "epoch": 249,
          "updates": 3984,
          "samples_seen": 254976,
          "online_loss": 0.0003072061545026372,
          "train_loss": 0.00031248119194060564,
          "validation_loss": 0.0002829070435836911,
          "train_metric": 3.147196548525244e-05,
          "validation_metric": 2.8493368517956696e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.937297586000568
        },
        {
          "epoch": 250,
          "updates": 4000,
          "samples_seen": 256000,
          "online_loss": 0.0003313509032523143,
          "train_loss": 0.00030972494278103113,
          "validation_loss": 0.00026932026958093047,
          "train_metric": 3.119435496046208e-05,
          "validation_metric": 2.7124951884616166e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9477295500000764
        },
        {
          "epoch": 251,
          "updates": 4016,
          "samples_seen": 257024,
          "online_loss": 0.00030214689923013793,
          "train_loss": 0.0003057458670809865,
          "validation_loss": 0.00026913610054180026,
          "train_metric": 3.079360249103047e-05,
          "validation_metric": 2.7106401830678806e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9581495050006197
        },
        {
          "epoch": 252,
          "updates": 4032,
          "samples_seen": 258048,
          "online_loss": 0.00030177288135746494,
          "train_loss": 0.0003075057757087052,
          "validation_loss": 0.0002623066247906536,
          "train_metric": 3.09708557324484e-05,
          "validation_metric": 2.6418560082674958e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.968564452999999
        },
        {
          "epoch": 253,
          "updates": 4048,
          "samples_seen": 259072,
          "online_loss": 0.00030786022398388013,
          "train_loss": 0.0002937604731414467,
          "validation_loss": 0.00025288318283855915,
          "train_metric": 2.958648110507056e-05,
          "validation_metric": 2.5469465981586836e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9797018360004586
        },
        {
          "epoch": 254,
          "updates": 4064,
          "samples_seen": 260096,
          "online_loss": 0.0003130514187432709,
          "train_loss": 0.0002963223378174007,
          "validation_loss": 0.0002502918359823525,
          "train_metric": 2.9844501113984734e-05,
          "validation_metric": 2.5208477381966077e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9910348970006453
        },
        {
          "epoch": 255,
          "updates": 4080,
          "samples_seen": 261120,
          "online_loss": 0.0003568906686268747,
          "train_loss": 0.0003636638866737485,
          "validation_loss": 0.0003043344768229872,
          "train_metric": 3.6626897781388834e-05,
          "validation_metric": 3.065145210712217e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.001755773999321
        },
        {
          "epoch": 256,
          "updates": 4096,
          "samples_seen": 262144,
          "online_loss": 0.00032016441491578007,
          "train_loss": 0.0003124268550891429,
          "validation_loss": 0.0002555205428507179,
          "train_metric": 3.146648668916896e-05,
          "validation_metric": 2.5735093004186638e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.0124676180003007
        },
        {
          "epoch": 257,
          "updates": 4112,
          "samples_seen": 263168,
          "online_loss": 0.00034287093421880854,
          "train_loss": 0.00032833489240147173,
          "validation_loss": 0.00029740887111984193,
          "train_metric": 3.3068685297621414e-05,
          "validation_metric": 2.9953933335491456e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.0228135130000737
        },
        {
          "epoch": 258,
          "updates": 4128,
          "samples_seen": 264192,
          "online_loss": 0.0003062403420699411,
          "train_loss": 0.00033238757168874145,
          "validation_loss": 0.0002922728599514812,
          "train_metric": 3.347685924381949e-05,
          "validation_metric": 2.9436654585879296e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.033227595999051
        },
        {
          "epoch": 259,
          "updates": 4144,
          "samples_seen": 265216,
          "online_loss": 0.0003092354263571906,
          "train_loss": 0.0002960572892334312,
          "validation_loss": 0.00025182217359542847,
          "train_metric": 2.9817805625498295e-05,
          "validation_metric": 2.5362609449075535e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.04407847100083
        },
        {
          "epoch": 260,
          "updates": 4160,
          "samples_seen": 266240,
          "online_loss": 0.00030022571354493266,
          "train_loss": 0.0002901693223975599,
          "validation_loss": 0.0002633916446939111,
          "train_metric": 2.9224789614090696e-05,
          "validation_metric": 2.6527841328061186e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.054539971999475
        },
        {
          "epoch": 261,
          "updates": 4176,
          "samples_seen": 267264,
          "online_loss": 0.0002994463939103298,
          "train_loss": 0.0002939965925179422,
          "validation_loss": 0.00025224947603419423,
          "train_metric": 2.961026257253252e-05,
          "validation_metric": 2.540564128139522e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.0655267349993665
        },
        {
          "epoch": 262,
          "updates": 4192,
          "samples_seen": 268288,
          "online_loss": 0.00033164439810207114,
          "train_loss": 0.0003124475770164281,
          "validation_loss": 0.0002621029852889478,
          "train_metric": 3.146857488900423e-05,
          "validation_metric": 2.6398052796139382e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.0764015759996255
        },
        {
          "epoch": 263,
          "updates": 4208,
          "samples_seen": 269312,
          "online_loss": 0.00033032877945515793,
          "train_loss": 0.0003460099105723202,
          "validation_loss": 0.00030469344346784055,
          "train_metric": 3.484885019133799e-05,
          "validation_metric": 3.068760997848585e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.0867232469990995
        },
        {
          "epoch": 264,
          "updates": 4224,
          "samples_seen": 270336,
          "online_loss": 0.0003304314786873874,
          "train_loss": 0.0003196419565938413,
          "validation_loss": 0.0002732021384872496,
          "train_metric": 3.219317295588553e-05,
          "validation_metric": 2.7515920010046102e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.0972933449993434
        },
        {
          "epoch": 265,
          "updates": 4240,
          "samples_seen": 271360,
          "online_loss": 0.00035526589408618747,
          "train_loss": 0.0002971721114590764,
          "validation_loss": 0.00027895928360521793,
          "train_metric": 2.9930084565421566e-05,
          "validation_metric": 2.8095755624235608e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 3.107646428999942
        },
        {
          "epoch": 266,
          "updates": 4256,
          "samples_seen": 272384,
          "online_loss": 0.0003006805445693317,
          "train_loss": 0.00028436470893211663,
          "validation_loss": 0.00024649687111377716,
          "train_metric": 2.864017187675927e-05,
          "validation_metric": 2.4826262233546004e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.119112813999891
        },
        {
          "epoch": 267,
          "updates": 4272,
          "samples_seen": 273408,
          "online_loss": 0.00028516436032077763,
          "train_loss": 0.0002874014317058027,
          "validation_loss": 0.00025273548089899123,
          "train_metric": 2.8946018574060872e-05,
          "validation_metric": 2.5454592105234042e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.130004304999602
        },
        {
          "epoch": 268,
          "updates": 4288,
          "samples_seen": 274432,
          "online_loss": 0.00030974162291386165,
          "train_loss": 0.00029408925911411643,
          "validation_loss": 0.0002620774903334677,
          "train_metric": 2.961959398817271e-05,
          "validation_metric": 2.6395484383101575e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.140285503999621
        },
        {
          "epoch": 269,
          "updates": 4304,
          "samples_seen": 275456,
          "online_loss": 0.0002954405745185795,
          "train_loss": 0.0002899968530982733,
          "validation_loss": 0.0002481217961758375,
          "train_metric": 2.920742190326564e-05,
          "validation_metric": 2.498992034816183e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1506185800008097
        },
        {
          "epoch": 270,
          "updates": 4320,
          "samples_seen": 276480,
          "online_loss": 0.0002994064898302895,
          "train_loss": 0.0002862360852304846,
          "validation_loss": 0.0002603398752398789,
          "train_metric": 2.882865010178648e-05,
          "validation_metric": 2.6220477593597025e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.160916171000281
        },
        {
          "epoch": 271,
          "updates": 4336,
          "samples_seen": 277504,
          "online_loss": 0.00029006306976953056,
          "train_loss": 0.0002908403694164008,
          "validation_loss": 0.0002436892973491922,
          "train_metric": 2.929237780335825e-05,
          "validation_metric": 2.4543493054807186e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.171719784999368
        },
        {
          "epoch": 272,
          "updates": 4352,
          "samples_seen": 278528,
          "online_loss": 0.00029972698393976316,
          "train_loss": 0.00030485718161799014,
          "validation_loss": 0.0002750572457443923,
          "train_metric": 3.0704104574397206e-05,
          "validation_metric": 2.7702757506631315e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1821376850002707
        },
        {
          "epoch": 273,
          "updates": 4368,
          "samples_seen": 279552,
          "online_loss": 0.0002992809131683316,
          "train_loss": 0.000281194515991956,
          "validation_loss": 0.0002455851063132286,
          "train_metric": 2.8320886485744268e-05,
          "validation_metric": 2.4734432372497395e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.192435592000038
        },
        {
          "epoch": 274,
          "updates": 4384,
          "samples_seen": 280576,
          "online_loss": 0.0002886055826820666,
          "train_loss": 0.0002810319128911942,
          "validation_loss": 0.00024917369592003524,
          "train_metric": 2.8304506486165337e-05,
          "validation_metric": 2.5095863747992553e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2026583349997964
        },
        {
          "epoch": 275,
          "updates": 4400,
          "samples_seen": 281600,
          "online_loss": 0.0002863162681023823,
          "train_loss": 0.0002806441916618496,
          "validation_loss": 0.0002468477177899331,
          "train_metric": 2.826545460266061e-05,
          "validation_metric": 2.4861596102709882e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2129740550008137
        },
        {
          "epoch": 276,
          "updates": 4416,
          "samples_seen": 282624,
          "online_loss": 0.00029083712797728367,
          "train_loss": 0.00028549833223223686,
          "validation_loss": 0.0002431374159641564,
          "train_metric": 2.8754348022630438e-05,
          "validation_metric": 2.4487910195603035e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2237329830004455
        },
        {
          "epoch": 277,
          "updates": 4432,
          "samples_seen": 283648,
          "online_loss": 0.00028547483634611126,
          "train_loss": 0.0002917977108154446,
          "validation_loss": 0.00026524224085733294,
          "train_metric": 2.9388798793661408e-05,
          "validation_metric": 2.671422771527432e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.234073350999097
        },
        {
          "epoch": 278,
          "updates": 4448,
          "samples_seen": 284672,
          "online_loss": 0.0002876583002944244,
          "train_loss": 0.0002849741722457111,
          "validation_loss": 0.0002469107275828719,
          "train_metric": 2.8701553674181923e-05,
          "validation_metric": 2.4867944375728257e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.244529326999327
        },
        {
          "epoch": 279,
          "updates": 4464,
          "samples_seen": 285696,
          "online_loss": 0.00029106591864547227,
          "train_loss": 0.0002809411089401692,
          "validation_loss": 0.0002517470857128501,
          "train_metric": 2.8295358788454905e-05,
          "validation_metric": 2.5355042453156784e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.254802767000001
        },
        {
          "epoch": 280,
          "updates": 4480,
          "samples_seen": 286720,
          "online_loss": 0.00029081726552249165,
          "train_loss": 0.00028439215384423733,
          "validation_loss": 0.0002404974220553413,
          "train_metric": 2.8642940378631465e-05,
          "validation_metric": 2.4222017600550316e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2655395300007513
        },
        {
          "epoch": 281,
          "updates": 4496,
          "samples_seen": 287744,
          "online_loss": 0.00028927896437380696,
          "train_loss": 0.0002786803524941206,
          "validation_loss": 0.0002491260238457471,
          "train_metric": 2.8067661332897842e-05,
          "validation_metric": 2.5091059796977788e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2759587740001734
        },
        {
          "epoch": 282,
          "updates": 4512,
          "samples_seen": 288768,
          "online_loss": 0.0002829828654284938,
          "train_loss": 0.0002783781383186579,
          "validation_loss": 0.0002423157711746171,
          "train_metric": 2.8037226002197713e-05,
          "validation_metric": 2.4405158910667524e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2862313330006145
        },
        {
          "epoch": 283,
          "updates": 4528,
          "samples_seen": 289792,
          "online_loss": 0.00028783053039660444,
          "train_loss": 0.0002798904897645116,
          "validation_loss": 0.00024249321722891182,
          "train_metric": 2.8189542717882432e-05,
          "validation_metric": 2.4423026843578555e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2964253819991427
        },
        {
          "epoch": 284,
          "updates": 4544,
          "samples_seen": 290816,
          "online_loss": 0.00028392694457579637,
          "train_loss": 0.0002781774091999978,
          "validation_loss": 0.00024999139714054763,
          "train_metric": 2.8017011572956108e-05,
          "validation_metric": 2.5178220312227495e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3067296289991646
        },
        {
          "epoch": 285,
          "updates": 4560,
          "samples_seen": 291840,
          "online_loss": 0.00029048812575638294,
          "train_loss": 0.0002776077308226377,
          "validation_loss": 0.00024106580531224608,
          "train_metric": 2.795963519020006e-05,
          "validation_metric": 2.4279263016069308e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3169857089997095
        },
        {
          "epoch": 286,
          "updates": 4576,
          "samples_seen": 292864,
          "online_loss": 0.0002813727314787684,
          "train_loss": 0.0002765097888186574,
          "validation_loss": 0.0002440841926727444,
          "train_metric": 2.7849051548400894e-05,
          "validation_metric": 2.4583265258115716e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3272017290000804
        },
        {
          "epoch": 287,
          "updates": 4592,
          "samples_seen": 293888,
          "online_loss": 0.0002823010781867197,
          "train_loss": 0.0002775891334749758,
          "validation_loss": 0.0002506060409359634,
          "train_metric": 2.79577579931356e-05,
          "validation_metric": 2.5240124159608968e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3373906159995386
        },
        {
          "epoch": 288,
          "updates": 4608,
          "samples_seen": 294912,
          "online_loss": 0.0002820954759954475,
          "train_loss": 0.0002757364127319306,
          "validation_loss": 0.0002426313585601747,
          "train_metric": 2.7771160603151657e-05,
          "validation_metric": 2.4436940293526277e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.347768413999802
        },
        {
          "epoch": 289,
          "updates": 4624,
          "samples_seen": 295936,
          "online_loss": 0.00028013640530843986,
          "train_loss": 0.00027800147654488683,
          "validation_loss": 0.000240588778979145,
          "train_metric": 2.7999290978186764e-05,
          "validation_metric": 2.4231223505921662e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.357955449000656
        },
        {
          "epoch": 290,
          "updates": 4640,
          "samples_seen": 296960,
          "online_loss": 0.0002820764639182016,
          "train_loss": 0.0002759972121566534,
          "validation_loss": 0.00024539511650800705,
          "train_metric": 2.779743226710707e-05,
          "validation_metric": 2.4715298422961496e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.368084387999261
        },
        {
          "epoch": 291,
          "updates": 4656,
          "samples_seen": 297984,
          "online_loss": 0.0002856972914742073,
          "train_loss": 0.0002763202355708927,
          "validation_loss": 0.00024383573327213526,
          "train_metric": 2.782996125461068e-05,
          "validation_metric": 2.4558241420891136e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3787022439992143
        },
        {
          "epoch": 292,
          "updates": 4672,
          "samples_seen": 299008,
          "online_loss": 0.000274751044344157,
          "train_loss": 0.00027604782371781766,
          "validation_loss": 0.00024373414635192603,
          "train_metric": 2.78025272564264e-05,
          "validation_metric": 2.4548013243474998e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.388838176000718
        },
        {
          "epoch": 293,
          "updates": 4688,
          "samples_seen": 300032,
          "online_loss": 0.00027760981174651533,
          "train_loss": 0.0002740061318036169,
          "validation_loss": 0.00023930447059683502,
          "train_metric": 2.7596892323344946e-05,
          "validation_metric": 2.4101867893477902e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.399620321999464
        },
        {
          "epoch": 294,
          "updates": 4704,
          "samples_seen": 301056,
          "online_loss": 0.00028501123597379774,
          "train_loss": 0.00027423372375778854,
          "validation_loss": 0.00024352135369554162,
          "train_metric": 2.7619817046797834e-05,
          "validation_metric": 2.452657645335421e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.410327707999386
        },
        {
          "epoch": 295,
          "updates": 4720,
          "samples_seen": 302080,
          "online_loss": 0.0002859432152035879,
          "train_loss": 0.000274140911642462,
          "validation_loss": 0.00023985271400306374,
          "train_metric": 2.7610465622274205e-05,
          "validation_metric": 2.4157088773790747e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.4210600040005374
        },
        {
          "epoch": 296,
          "updates": 4736,
          "samples_seen": 303104,
          "online_loss": 0.00028197501524118707,
          "train_loss": 0.00027335455524735153,
          "validation_loss": 0.0002440084790578112,
          "train_metric": 2.7531270461622626e-05,
          "validation_metric": 2.457563823554665e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.431302909999431
        },
        {
          "epoch": 297,
          "updates": 4752,
          "samples_seen": 304128,
          "online_loss": 0.0002776388218990178,
          "train_loss": 0.00027317876811139286,
          "validation_loss": 0.00024152071273420006,
          "train_metric": 2.751356441876851e-05,
          "validation_metric": 2.4325081540155225e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.4416169279993483
        },
        {
          "epoch": 298,
          "updates": 4768,
          "samples_seen": 305152,
          "online_loss": 0.0002758104856184218,
          "train_loss": 0.00027286563999950886,
          "validation_loss": 0.00024154811399057508,
          "train_metric": 2.7482028599479236e-05,
          "validation_metric": 2.4327842766069807e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.451900733000002
        },
        {
          "epoch": 299,
          "updates": 4784,
          "samples_seen": 306176,
          "online_loss": 0.0002749607601799653,
          "train_loss": 0.00027266520191915333,
          "validation_loss": 0.0002397731877863407,
          "train_metric": 2.746183963608928e-05,
          "validation_metric": 2.414907794445753e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.462181805
        },
        {
          "epoch": 300,
          "updates": 4800,
          "samples_seen": 307200,
          "online_loss": 0.00028007028413412627,
          "train_loss": 0.0002752329455688596,
          "validation_loss": 0.000245107599766925,
          "train_metric": 2.7720454454538412e-05,
          "validation_metric": 2.468634193064645e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.4724600730005477
        },
        {
          "epoch": 301,
          "updates": 4816,
          "samples_seen": 308224,
          "online_loss": 0.0002947076964119333,
          "train_loss": 0.0002735284506343305,
          "validation_loss": 0.0002423688129056245,
          "train_metric": 2.7548783691599965e-05,
          "validation_metric": 2.4410499463556334e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.4829138179993606
        },
        {
          "epoch": 302,
          "updates": 4832,
          "samples_seen": 309248,
          "online_loss": 0.00029227892082417384,
          "train_loss": 0.0002773213491309434,
          "validation_loss": 0.00024373696942348033,
          "train_metric": 2.793078965623863e-05,
          "validation_metric": 2.4548295186832547e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.4932161500000802
        },
        {
          "epoch": 303,
          "updates": 4848,
          "samples_seen": 310272,
          "online_loss": 0.0002839570597643615,
          "train_loss": 0.0002742986544035375,
          "validation_loss": 0.00023943251289892942,
          "train_metric": 2.762635631370358e-05,
          "validation_metric": 2.4114766347338445e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.5035318110003573
        },
        {
          "epoch": 304,
          "updates": 4864,
          "samples_seen": 311296,
          "online_loss": 0.0002729262337197724,
          "train_loss": 0.00027512715314514935,
          "validation_loss": 0.00024721951922401786,
          "train_metric": 2.7709796995623037e-05,
          "validation_metric": 2.4899039999581873e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.514013698999406
        },
        {
          "epoch": 305,
          "updates": 4880,
          "samples_seen": 312320,
          "online_loss": 0.0002803133493216592,
          "train_loss": 0.0002744154480751604,
          "validation_loss": 0.0002408068539807573,
          "train_metric": 2.7638121537165716e-05,
          "validation_metric": 2.425318416499067e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.524322029999894
        },
        {
          "epoch": 306,
          "updates": 4896,
          "samples_seen": 313344,
          "online_loss": 0.00027862631668540416,
          "train_loss": 0.00027258097543381155,
          "validation_loss": 0.00024123917683027685,
          "train_metric": 2.7453357688500546e-05,
          "validation_metric": 2.4296725314343348e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.534566619000543
        },
        {
          "epoch": 307,
          "updates": 4912,
          "samples_seen": 314368,
          "online_loss": 0.00027625834081845824,
          "train_loss": 0.000271108525339514,
          "validation_loss": 0.0002407278079772368,
          "train_metric": 2.7305060939397663e-05,
          "validation_metric": 2.424522244837135e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5450728319992777
        },
        {
          "epoch": 308,
          "updates": 4928,
          "samples_seen": 315392,
          "online_loss": 0.00027551024777494604,
          "train_loss": 0.0002711904526222497,
          "validation_loss": 0.00024186656810343266,
          "train_metric": 2.7313310056342743e-05,
          "validation_metric": 2.4359913368243724e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.555329681999865
        },
        {
          "epoch": 309,
          "updates": 4944,
          "samples_seen": 316416,
          "online_loss": 0.00027456748375698226,
          "train_loss": 0.0002708402171265334,
          "validation_loss": 0.00023931893520057201,
          "train_metric": 2.7278032575850375e-05,
          "validation_metric": 2.4103326722979546e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5659252559999004
        },
        {
          "epoch": 310,
          "updates": 4960,
          "samples_seen": 317440,
          "online_loss": 0.0002777746722131269,
          "train_loss": 0.00027079018764197826,
          "validation_loss": 0.00023839061032049358,
          "train_metric": 2.7272995794191957e-05,
          "validation_metric": 2.400982702965848e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5771871550005017
        },
        {
          "epoch": 311,
          "updates": 4976,
          "samples_seen": 318464,
          "online_loss": 0.00027336153925716644,
          "train_loss": 0.0002714385627768934,
          "validation_loss": 0.00024191818374674767,
          "train_metric": 2.733829569478985e-05,
          "validation_metric": 2.4365112039959058e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.587711655000021
        },
        {
          "epoch": 312,
          "updates": 4992,
          "samples_seen": 319488,
          "online_loss": 0.0002775843495328445,
          "train_loss": 0.00027023121947422624,
          "validation_loss": 0.00023897243954706937,
          "train_metric": 2.7216694434173405e-05,
          "validation_metric": 2.406842577329371e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5980290160005097
        },
        {
          "epoch": 313,
          "updates": 5008,
          "samples_seen": 320512,
          "online_loss": 0.0002734246004365559,
          "train_loss": 0.00027013872750103474,
          "validation_loss": 0.00023800545022822917,
          "train_metric": 2.720738666539546e-05,
          "validation_metric": 2.397103889961727e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.609023557999535
        },
        {
          "epoch": 314,
          "updates": 5024,
          "samples_seen": 321536,
          "online_loss": 0.0002752914360826253,
          "train_loss": 0.00027020814013667405,
          "validation_loss": 0.00023951265029609203,
          "train_metric": 2.721437340369448e-05,
          "validation_metric": 2.412283356534317e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.619371193999541
        },
        {
          "epoch": 315,
          "updates": 5040,
          "samples_seen": 322560,
          "online_loss": 0.0002702430992940208,
          "train_loss": 0.00027124170446768403,
          "validation_loss": 0.00024271468282677233,
          "train_metric": 2.7318472348270006e-05,
          "validation_metric": 2.444533492962364e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6298989730003086
        },
        {
          "epoch": 316,
          "updates": 5056,
          "samples_seen": 323584,
          "online_loss": 0.00027995614345854847,
          "train_loss": 0.0002698745811358094,
          "validation_loss": 0.00023844004317652434,
          "train_metric": 2.7180780307389796e-05,
          "validation_metric": 2.4014805603655986e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6403529700000945
        },
        {
          "epoch": 317,
          "updates": 5072,
          "samples_seen": 324608,
          "online_loss": 0.00027899135966436006,
          "train_loss": 0.00026957568479701877,
          "validation_loss": 0.00024000924895517528,
          "train_metric": 2.715067967073992e-05,
          "validation_metric": 2.417285395495128e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.650724008999532
        },
        {
          "epoch": 318,
          "updates": 5088,
          "samples_seen": 325632,
          "online_loss": 0.0002719196313591965,
          "train_loss": 0.0002708344254642725,
          "validation_loss": 0.00023666951165068895,
          "train_metric": 2.7277452318230644e-05,
          "validation_metric": 2.3836484615458176e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.661469348000537
        },
        {
          "epoch": 319,
          "updates": 5104,
          "samples_seen": 326656,
          "online_loss": 0.00027184252030565403,
          "train_loss": 0.00026953272754326463,
          "validation_loss": 0.00024173426209017634,
          "train_metric": 2.714635047595948e-05,
          "validation_metric": 2.4346587451873347e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.671898037999199
        },
        {
          "epoch": 320,
          "updates": 5120,
          "samples_seen": 327680,
          "online_loss": 0.00027584856525209034,
          "train_loss": 0.00026950277970172465,
          "validation_loss": 0.0002393302711425349,
          "train_metric": 2.71433345915284e-05,
          "validation_metric": 2.4104470867314376e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6823076710006717
        },
        {
          "epoch": 321,
          "updates": 5136,
          "samples_seen": 328704,
          "online_loss": 0.0002715148557399516,
          "train_loss": 0.00026956514921039343,
          "validation_loss": 0.00024004990700632334,
          "train_metric": 2.7149615561938845e-05,
          "validation_metric": 2.417694850009866e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6925914809999085
        },
        {
          "epoch": 322,
          "updates": 5152,
          "samples_seen": 329728,
          "online_loss": 0.0002743826717050979,
          "train_loss": 0.0002684843784663826,
          "validation_loss": 0.00023669289657846093,
          "train_metric": 2.7040765417041257e-05,
          "validation_metric": 2.383884202572517e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7029505749997043
        },
        {
          "epoch": 323,
          "updates": 5168,
          "samples_seen": 330752,
          "online_loss": 0.00027188849344383925,
          "train_loss": 0.00026866537518799305,
          "validation_loss": 0.0002369661960983649,
          "train_metric": 2.7058991690864787e-05,
          "validation_metric": 2.386636879236903e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7136018769997463
        },
        {
          "epoch": 324,
          "updates": 5184,
          "samples_seen": 331776,
          "online_loss": 0.00027112548195873387,
          "train_loss": 0.0002686300140339881,
          "validation_loss": 0.000239024157053791,
          "train_metric": 2.7055430109612644e-05,
          "validation_metric": 2.407363717793487e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.724168089000159
        },
        {
          "epoch": 325,
          "updates": 5200,
          "samples_seen": 332800,
          "online_loss": 0.0002697275886021089,
          "train_loss": 0.0002679638273548335,
          "validation_loss": 0.0002390907029621303,
          "train_metric": 2.698833486647345e-05,
          "validation_metric": 2.4080340153886937e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7344466720005585
        },
        {
          "epoch": 326,
          "updates": 5216,
          "samples_seen": 333824,
          "online_loss": 0.0002725511030803318,
          "train_loss": 0.00026840984355658293,
          "validation_loss": 0.000239329194300808,
          "train_metric": 2.7033258447772823e-05,
          "validation_metric": 2.4104356270981953e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.744958114999463
        },
        {
          "epoch": 327,
          "updates": 5232,
          "samples_seen": 334848,
          "online_loss": 0.0002733694136622944,
          "train_loss": 0.0002692787966225296,
          "validation_loss": 0.00023745233193039894,
          "train_metric": 2.7120775484945625e-05,
          "validation_metric": 2.3915330530144274e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7551839480001945
        },
        {
          "epoch": 328,
          "updates": 5248,
          "samples_seen": 335872,
          "online_loss": 0.0002721811297305976,
          "train_loss": 0.000267664494458586,
          "validation_loss": 0.0002388850407442078,
          "train_metric": 2.695819057407789e-05,
          "validation_metric": 2.4059625502559356e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7653770319993782
        },
        {
          "epoch": 329,
          "updates": 5264,
          "samples_seen": 336896,
          "online_loss": 0.00027103784668724984,
          "train_loss": 0.00026886421255767345,
          "validation_loss": 0.0002401810052106157,
          "train_metric": 2.7079016945208423e-05,
          "validation_metric": 2.4190150725189596e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7759109159997024
        },
        {
          "epoch": 330,
          "updates": 5280,
          "samples_seen": 337920,
          "online_loss": 0.0002772664392978186,
          "train_loss": 0.0002711968554649502,
          "validation_loss": 0.00023626760230399668,
          "train_metric": 2.73139539785916e-05,
          "validation_metric": 2.379600664426107e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.786896452000292
        },
        {
          "epoch": 331,
          "updates": 5296,
          "samples_seen": 338944,
          "online_loss": 0.00027156145461049164,
          "train_loss": 0.00026964585413224995,
          "validation_loss": 0.00024443003349006176,
          "train_metric": 2.7157742806593888e-05,
          "validation_metric": 2.4618097086204216e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.797442331999264
        },
        {
          "epoch": 332,
          "updates": 5312,
          "samples_seen": 339968,
          "online_loss": 0.00026891781180893304,
          "train_loss": 0.00026837055338546634,
          "validation_loss": 0.00023472269822377712,
          "train_metric": 2.702930214582011e-05,
          "validation_metric": 2.3640408471692353e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8084687129994563
        },
        {
          "epoch": 333,
          "updates": 5328,
          "samples_seen": 340992,
          "online_loss": 0.0002703705240492127,
          "train_loss": 0.00026650441577658057,
          "validation_loss": 0.0002394300972810015,
          "train_metric": 2.6841344151762314e-05,
          "validation_metric": 2.4114520783768967e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.818859421000525
        },
        {
          "epoch": 334,
          "updates": 5344,
          "samples_seen": 342016,
          "online_loss": 0.00027156094347446924,
          "train_loss": 0.00026627990882843733,
          "validation_loss": 0.00023836237960495055,
          "train_metric": 2.681873957044445e-05,
          "validation_metric": 2.4006985768210143e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.829188042000169
        },
        {
          "epoch": 335,
          "updates": 5360,
          "samples_seen": 343040,
          "online_loss": 0.0002708553693082649,
          "train_loss": 0.00026602353318594396,
          "validation_loss": 0.00023712134861852974,
          "train_metric": 2.6792913558892906e-05,
          "validation_metric": 2.388199391134549e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8396785050008475
        },
        {
          "epoch": 336,
          "updates": 5376,
          "samples_seen": 344064,
          "online_loss": 0.0002716608069022186,
          "train_loss": 0.0002666140499059111,
          "validation_loss": 0.00023616189719177783,
          "train_metric": 2.685239269339945e-05,
          "validation_metric": 2.378536191827152e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8500302089996694
        },
        {
          "epoch": 337,
          "updates": 5392,
          "samples_seen": 345088,
          "online_loss": 0.0002698207490539062,
          "train_loss": 0.0002663444320205599,
          "validation_loss": 0.0002366007538512349,
          "train_metric": 2.6825235181604512e-05,
          "validation_metric": 2.3829563360777684e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8603022049992433
        },
        {
          "epoch": 338,
          "updates": 5408,
          "samples_seen": 346112,
          "online_loss": 0.0002708168221943197,
          "train_loss": 0.00026545175933279097,
          "validation_loss": 0.00023554801009595394,
          "train_metric": 2.6735329811344855e-05,
          "validation_metric": 2.3723532649455592e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8707384249992174
        },
        {
          "epoch": 339,
          "updates": 5424,
          "samples_seen": 347136,
          "online_loss": 0.00026948647246172186,
          "train_loss": 0.0002654335694387555,
          "validation_loss": 0.00023772659187670797,
          "train_metric": 2.673349627002608e-05,
          "validation_metric": 2.3942951884237118e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8814077280003403
        },
        {
          "epoch": 340,
          "updates": 5440,
          "samples_seen": 348160,
          "online_loss": 0.00026969690588884987,
          "train_loss": 0.0002653380506671965,
          "validation_loss": 0.00023634372337255627,
          "train_metric": 2.6723879273049533e-05,
          "validation_metric": 2.3803670046618208e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.891854711999258
        },
        {
          "epoch": 341,
          "updates": 5456,
          "samples_seen": 349184,
          "online_loss": 0.00026800766863743775,
          "train_loss": 0.0002649170346558094,
          "validation_loss": 0.00023472621978726238,
          "train_metric": 2.668147317308467e-05,
          "validation_metric": 2.364076499361545e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9021829579996847
        },
        {
          "epoch": 342,
          "updates": 5472,
          "samples_seen": 350208,
          "online_loss": 0.00026938806786347413,
          "train_loss": 0.00026703139883466065,
          "validation_loss": 0.00023920778767205775,
          "train_metric": 2.689442590053659e-05,
          "validation_metric": 2.4092132662190124e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.912894278000749
        },
        {
          "epoch": 343,
          "updates": 5488,
          "samples_seen": 351232,
          "online_loss": 0.00026967837493430125,
          "train_loss": 0.00026451662415638566,
          "validation_loss": 0.00023489570594392717,
          "train_metric": 2.6641144359018654e-05,
          "validation_metric": 2.365783802815713e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.923583007999696
        },
        {
          "epoch": 344,
          "updates": 5504,
          "samples_seen": 352256,
          "online_loss": 0.00027090375624538865,
          "train_loss": 0.0002646867069415748,
          "validation_loss": 0.00023625178437214345,
          "train_metric": 2.6658279239200056e-05,
          "validation_metric": 2.379441139055416e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.934025673999713
        },
        {
          "epoch": 345,
          "updates": 5520,
          "samples_seen": 353280,
          "online_loss": 0.000270392389211338,
          "train_loss": 0.00026496374630369246,
          "validation_loss": 0.00023360757040791214,
          "train_metric": 2.6686178898671642e-05,
          "validation_metric": 2.352809860894922e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9456896890005737
        },
        {
          "epoch": 346,
          "updates": 5536,
          "samples_seen": 354304,
          "online_loss": 0.0002717540537560126,
          "train_loss": 0.00026454130420461297,
          "validation_loss": 0.00023702613543719053,
          "train_metric": 2.6643629098543897e-05,
          "validation_metric": 2.3872404199209996e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9563070189997234
        },
        {
          "epoch": 347,
          "updates": 5552,
          "samples_seen": 355328,
          "online_loss": 0.0002698200987651944,
          "train_loss": 0.0002640010497998446,
          "validation_loss": 0.00023389875423163176,
          "train_metric": 2.6589215849526227e-05,
          "validation_metric": 2.35574298130814e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9671844500007865
        },
        {
          "epoch": 348,
          "updates": 5568,
          "samples_seen": 356352,
          "online_loss": 0.00026866844928008504,
          "train_loss": 0.00026435533072799444,
          "validation_loss": 0.00023729568056296557,
          "train_metric": 2.662490078364499e-05,
          "validation_metric": 2.389955443504732e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.977773820999573
        },
        {
          "epoch": 349,
          "updates": 5584,
          "samples_seen": 357376,
          "online_loss": 0.0002800543416014989,
          "train_loss": 0.0002643961925059557,
          "validation_loss": 0.0002340125647606328,
          "train_metric": 2.662901533767581e-05,
          "validation_metric": 2.3568887627334334e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9884086940001
        },
        {
          "epoch": 350,
          "updates": 5600,
          "samples_seen": 358400,
          "online_loss": 0.0002644327087182319,
          "train_loss": 0.00026624248130246997,
          "validation_loss": 0.00023389109992422163,
          "train_metric": 2.68149688054109e-05,
          "validation_metric": 2.3556653104606085e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9988491059993976
        },
        {
          "epoch": 351,
          "updates": 5616,
          "samples_seen": 359424,
          "online_loss": 0.00027218630248171394,
          "train_loss": 0.00026524244458414614,
          "validation_loss": 0.00023964318097569048,
          "train_metric": 2.671424954314716e-05,
          "validation_metric": 2.4135981220752e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.009432619999643
        },
        {
          "epoch": 352,
          "updates": 5632,
          "samples_seen": 360448,
          "online_loss": 0.00027725698691938305,
          "train_loss": 0.0002672834089025855,
          "validation_loss": 0.00023226825578603894,
          "train_metric": 2.691980444069486e-05,
          "validation_metric": 2.339320781175047e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.0206240139996225
        },
        {
          "epoch": 353,
          "updates": 5648,
          "samples_seen": 361472,
          "online_loss": 0.00026534360858931905,
          "train_loss": 0.00026584326406009495,
          "validation_loss": 0.00024211322306655347,
          "train_metric": 2.6774758225656115e-05,
          "validation_metric": 2.4384757125517353e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.0312682090007
        },
        {
          "epoch": 354,
          "updates": 5664,
          "samples_seen": 362496,
          "online_loss": 0.0002754954548436217,
          "train_loss": 0.0002636321005411446,
          "validation_loss": 0.0002350745489820838,
          "train_metric": 2.6552057533990592e-05,
          "validation_metric": 2.3675847842241637e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.041797623000093
        },
        {
          "epoch": 355,
          "updates": 5680,
          "samples_seen": 363520,
          "online_loss": 0.00027058964769821614,
          "train_loss": 0.00026316079311072826,
          "validation_loss": 0.0002322608488611877,
          "train_metric": 2.6504591005505063e-05,
          "validation_metric": 2.3392462026095018e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.052727894000782
        },
        {
          "epoch": 356,
          "updates": 5696,
          "samples_seen": 364544,
          "online_loss": 0.00026962410720443586,
          "train_loss": 0.0002620198065415025,
          "validation_loss": 0.000235685525694862,
          "train_metric": 2.6389678168925457e-05,
          "validation_metric": 2.373738243477419e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.063195077000273
        },
        {
          "epoch": 357,
          "updates": 5712,
          "samples_seen": 365568,
          "online_loss": 0.0002733711608016165,
          "train_loss": 0.00026410006103105843,
          "validation_loss": 0.00023404884268529713,
          "train_metric": 2.659918936842587e-05,
          "validation_metric": 2.3572541977046058e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.0735375250005745
        },
        {
          "epoch": 358,
          "updates": 5728,
          "samples_seen": 366592,
          "online_loss": 0.00026926136615657015,
          "train_loss": 0.0002641441242303699,
          "validation_loss": 0.00024006437161006033,
          "train_metric": 2.6603631340549327e-05,
          "validation_metric": 2.4178403691621497e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.083872248000262
        },
        {
          "epoch": 359,
          "updates": 5744,
          "samples_seen": 367616,
          "online_loss": 0.0002674398583621951,
          "train_loss": 0.0002613264077808708,
          "validation_loss": 0.0002338223857805133,
          "train_metric": 2.631983625178691e-05,
          "validation_metric": 2.35497354879044e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.094599191999805
        },
        {
          "epoch": 360,
          "updates": 5760,
          "samples_seen": 368640,
          "online_loss": 0.00026538619113125606,
          "train_loss": 0.0002619453298393637,
          "validation_loss": 0.0002347814734093845,
          "train_metric": 2.6382176656625234e-05,
          "validation_metric": 2.3646332920179702e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.104920376999871
        },
        {
          "epoch": 361,
          "updates": 5776,
          "samples_seen": 369664,
          "online_loss": 0.00026623955000104615,
          "train_loss": 0.00026303285267204046,
          "validation_loss": 0.00023227743804454803,
          "train_metric": 2.6491705284570344e-05,
          "validation_metric": 2.3394131858367473e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.115506620000815
        },
        {
          "epoch": 362,
          "updates": 5792,
          "samples_seen": 370688,
          "online_loss": 0.000266721474872611,
          "train_loss": 0.0002608317299745977,
          "validation_loss": 0.00023396409233100712,
          "train_metric": 2.6270017770002596e-05,
          "validation_metric": 2.3564005459775217e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.126032997999573
        },
        {
          "epoch": 363,
          "updates": 5808,
          "samples_seen": 371712,
          "online_loss": 0.0002626758941914886,
          "train_loss": 0.0002609268412925303,
          "validation_loss": 0.00023329233226832002,
          "train_metric": 2.627959656820167e-05,
          "validation_metric": 2.3496351786889136e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.136588686000323
        },
        {
          "epoch": 364,
          "updates": 5824,
          "samples_seen": 372736,
          "online_loss": 0.00027215438421990257,
          "train_loss": 0.0002628134097903967,
          "validation_loss": 0.00023436199990101159,
          "train_metric": 2.6469606382306665e-05,
          "validation_metric": 2.360408143431414e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.147028340999896
        },
        {
          "epoch": 365,
          "updates": 5840,
          "samples_seen": 373760,
          "online_loss": 0.000268860911091906,
          "train_loss": 0.00026435955078341067,
          "validation_loss": 0.00023366129607893527,
          "train_metric": 2.6625324608176015e-05,
          "validation_metric": 2.3533511921414174e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.157464412999616
        },
        {
          "epoch": 366,
          "updates": 5856,
          "samples_seen": 374784,
          "online_loss": 0.00027190419859834947,
          "train_loss": 0.00026696850545704365,
          "validation_loss": 0.00023551841150037944,
          "train_metric": 2.6888092179433443e-05,
          "validation_metric": 2.372055132582318e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.167771616999744
        },
        {
          "epoch": 367,
          "updates": 5872,
          "samples_seen": 375808,
          "online_loss": 0.0002634167476571747,
          "train_loss": 0.0002598127175588161,
          "validation_loss": 0.00023256841814145446,
          "train_metric": 2.6167386749875732e-05,
          "validation_metric": 2.3423439415637404e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1787377189994
        },
        {
          "epoch": 368,
          "updates": 5888,
          "samples_seen": 376832,
          "online_loss": 0.00026318259187974036,
          "train_loss": 0.00025977936456911266,
          "validation_loss": 0.0002337516489205882,
          "train_metric": 2.616402525745798e-05,
          "validation_metric": 2.354261050641071e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.189016828000604
        },
        {
          "epoch": 369,
          "updates": 5904,
          "samples_seen": 377856,
          "online_loss": 0.0002638514597492758,
          "train_loss": 0.00026008576969616115,
          "validation_loss": 0.00023314352438319474,
          "train_metric": 2.619488259369973e-05,
          "validation_metric": 2.348136513319332e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.199269249000281
        },
        {
          "epoch": 370,
          "updates": 5920,
          "samples_seen": 378880,
          "online_loss": 0.00026559491743682884,
          "train_loss": 0.0002597294223960489,
          "validation_loss": 0.00023232086095958948,
          "train_metric": 2.6158995751757175e-05,
          "validation_metric": 2.3398506527883e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.209588674999395
        },
        {
          "epoch": 371,
          "updates": 5936,
          "samples_seen": 379904,
          "online_loss": 0.0002665466340658895,
          "train_loss": 0.0002599720610305667,
          "validation_loss": 0.00023515429347753525,
          "train_metric": 2.618343205540441e-05,
          "validation_metric": 2.3683878680458292e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.219897204999143
        },
        {
          "epoch": 372,
          "updates": 5952,
          "samples_seen": 380928,
          "online_loss": 0.00026287146556569496,
          "train_loss": 0.0002591258380562067,
          "validation_loss": 0.00022888905368745327,
          "train_metric": 2.6098205125890672e-05,
          "validation_metric": 2.3052867618389428e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.230752441000732
        },
        {
          "epoch": 373,
          "updates": 5968,
          "samples_seen": 381952,
          "online_loss": 0.00026484437148610596,
          "train_loss": 0.00025946320965886116,
          "validation_loss": 0.00023175655223894864,
          "train_metric": 2.6132187485927716e-05,
          "validation_metric": 2.3341666747001e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.241197090999776
        },
        {
          "epoch": 374,
          "updates": 5984,
          "samples_seen": 382976,
          "online_loss": 0.00026071622960444074,
          "train_loss": 0.00025868063676171005,
          "validation_loss": 0.00023195792164187878,
          "train_metric": 2.6053370675072074e-05,
          "validation_metric": 2.336195029783994e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.25174651099951
        },
        {
          "epoch": 375,
          "updates": 6000,
          "samples_seen": 384000,
          "online_loss": 0.00026354993042332353,
          "train_loss": 0.0002587648050393909,
          "validation_loss": 0.0002324573724763468,
          "train_metric": 2.606184352771379e-05,
          "validation_metric": 2.3412258087773807e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.262159936999524
        },
        {
          "epoch": 376,
          "updates": 6016,
          "samples_seen": 385024,
          "online_loss": 0.0002646459806783241,
          "train_loss": 0.00026086674188263714,
          "validation_loss": 0.00023413619783241302,
          "train_metric": 2.627354479045607e-05,
          "validation_metric": 2.358134042879101e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.272501723999085
        },
        {
          "epoch": 377,
          "updates": 6032,
          "samples_seen": 386048,
          "online_loss": 0.0002691737236091285,
          "train_loss": 0.00025860569439828396,
          "validation_loss": 0.00022835178242530674,
          "train_metric": 2.604581823106855e-05,
          "validation_metric": 2.2998759959591553e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.2834882700008166
        },
        {
          "epoch": 378,
          "updates": 6048,
          "samples_seen": 387072,
          "online_loss": 0.00027014105398848187,
          "train_loss": 0.00025773936067707837,
          "validation_loss": 0.00023392662114929408,
          "train_metric": 2.595856312836986e-05,
          "validation_metric": 2.3560232875752263e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.293931096999586
        },
        {
          "epoch": 379,
          "updates": 6064,
          "samples_seen": 388096,
          "online_loss": 0.00025950999497581506,
          "train_loss": 0.00025956801255233586,
          "validation_loss": 0.0002323719672858715,
          "train_metric": 2.614273580547888e-05,
          "validation_metric": 2.3403652448905632e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.304360750000342
        },
        {
          "epoch": 380,
          "updates": 6080,
          "samples_seen": 389120,
          "online_loss": 0.0002673243025128613,
          "train_loss": 0.00026249722577631474,
          "validation_loss": 0.00022949191043153405,
          "train_metric": 2.6437759515829384e-05,
          "validation_metric": 2.3113583665690385e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.314944674999424
        },
        {
          "epoch": 381,
          "updates": 6096,
          "samples_seen": 390144,
          "online_loss": 0.00026515687022765633,
          "train_loss": 0.00025731619098223746,
          "validation_loss": 0.0002350400754949078,
          "train_metric": 2.5915946025634184e-05,
          "validation_metric": 2.367237539147027e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.325363324000136
        },
        {
          "epoch": 382,
          "updates": 6112,
          "samples_seen": 391168,
          "online_loss": 0.0002583154755484429,
          "train_loss": 0.0002575215185061097,
          "validation_loss": 0.0002305114030605182,
          "train_metric": 2.59366279351525e-05,
          "validation_metric": 2.321626561752055e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.335632538999562
        },
        {
          "epoch": 383,
          "updates": 6128,
          "samples_seen": 392192,
          "online_loss": 0.0002611028930914472,
          "train_loss": 0.0002570757642388344,
          "validation_loss": 0.00022784208704251796,
          "train_metric": 2.5891728000715375e-05,
          "validation_metric": 2.294741898367647e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.346555358999467
        },
        {
          "epoch": 384,
          "updates": 6144,
          "samples_seen": 393216,
          "online_loss": 0.00026476228686078684,
          "train_loss": 0.0002566763723734766,
          "validation_loss": 0.0002317421167390421,
          "train_metric": 2.585150105005596e-05,
          "validation_metric": 2.3340217012446374e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.357055300999491
        },
        {
          "epoch": 385,
          "updates": 6160,
          "samples_seen": 394240,
          "online_loss": 0.00025984318108385196,
          "train_loss": 0.0002560752909630537,
          "validation_loss": 0.000229863086133264,
          "train_metric": 2.5790965082705952e-05,
          "validation_metric": 2.3150969354901463e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.3673301540002285
        },
        {
          "epoch": 386,
          "updates": 6176,
          "samples_seen": 395264,
          "online_loss": 0.00026246705874655163,
          "train_loss": 0.0002585130278021097,
          "validation_loss": 0.00023112079361453652,
          "train_metric": 2.603648681542836e-05,
          "validation_metric": 2.3277638319996186e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.377768964999632
        },
        {
          "epoch": 387,
          "updates": 6192,
          "samples_seen": 396288,
          "online_loss": 0.00025967482451960677,
          "train_loss": 0.0002555942628532648,
          "validation_loss": 0.00023066604626365006,
          "train_metric": 2.5742519937921315e-05,
          "validation_metric": 2.3231839804793708e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.3882403719999274
        },
        {
          "epoch": 388,
          "updates": 6208,
          "samples_seen": 397312,
          "online_loss": 0.00026038423857244197,
          "train_loss": 0.0002555190003477037,
          "validation_loss": 0.00022861352772451937,
          "train_metric": 2.5734936571097933e-05,
          "validation_metric": 2.3025117116048932e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.398579255999721
        },
        {
          "epoch": 389,
          "updates": 6224,
          "samples_seen": 398336,
          "online_loss": 0.0002595979876787169,
          "train_loss": 0.00025633253972046077,
          "validation_loss": 0.00023014475300442427,
          "train_metric": 2.5816872948780656e-05,
          "validation_metric": 2.3179334675660357e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.408984829000474
        },
        {
          "epoch": 390,
          "updates": 6240,
          "samples_seen": 399360,
          "online_loss": 0.0002633906851770007,
          "train_loss": 0.0002561289584264159,
          "validation_loss": 0.0002289885887876153,
          "train_metric": 2.5796367481234483e-05,
          "validation_metric": 2.3062890250002965e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.419380281000485
        },
        {
          "epoch": 391,
          "updates": 6256,
          "samples_seen": 400384,
          "online_loss": 0.0002577640516392421,
          "train_loss": 0.00025485199876129627,
          "validation_loss": 0.00022676507069263607,
          "train_metric": 2.5667757654446177e-05,
          "validation_metric": 2.283894718857482e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.430190209999637
        },
        {
          "epoch": 392,
          "updates": 6272,
          "samples_seen": 401408,
          "online_loss": 0.0002632240639286465,
          "train_loss": 0.0002560772409196943,
          "validation_loss": 0.00023016917111817747,
          "train_metric": 2.579116335255094e-05,
          "validation_metric": 2.3181795768323354e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.4406134790006035
        },
        {
          "epoch": 393,
          "updates": 6288,
          "samples_seen": 402432,
          "online_loss": 0.0002720643278735224,
          "train_loss": 0.0002593165554571897,
          "validation_loss": 0.00023628229973837733,
          "train_metric": 2.611741547298152e-05,
          "validation_metric": 2.379748912062496e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.450945344000502
        },
        {
          "epoch": 394,
          "updates": 6304,
          "samples_seen": 403456,
          "online_loss": 0.00026276839889760595,
          "train_loss": 0.00025586612173356116,
          "validation_loss": 0.00022730795899406075,
          "train_metric": 2.5769899366423488e-05,
          "validation_metric": 2.289362601004541e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.461232808000204
        },
        {
          "epoch": 395,
          "updates": 6320,
          "samples_seen": 404480,
          "online_loss": 0.00025751984048838494,
          "train_loss": 0.0002555557293817401,
          "validation_loss": 0.0002301420026924461,
          "train_metric": 2.5738636395544745e-05,
          "validation_metric": 2.3179061827249825e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.471476164999331
        },
        {
          "epoch": 396,
          "updates": 6336,
          "samples_seen": 405504,
          "online_loss": 0.00025738440854183864,
          "train_loss": 0.0002537926484365016,
          "validation_loss": 0.00022697466192767024,
          "train_metric": 2.55610666499706e-05,
          "validation_metric": 2.286005656060297e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.481795638999756
        },
        {
          "epoch": 397,
          "updates": 6352,
          "samples_seen": 406528,
          "online_loss": 0.00025813790216488997,
          "train_loss": 0.00025425024796277285,
          "validation_loss": 0.00022783121676184237,
          "train_metric": 2.5607152565498836e-05,
          "validation_metric": 2.2946327590034343e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.492041437999433
        },
        {
          "epoch": 398,
          "updates": 6368,
          "samples_seen": 407552,
          "online_loss": 0.00025358176117151743,
          "train_loss": 0.0002542404399719089,
          "validation_loss": 0.00023148047330323607,
          "train_metric": 2.5606168492231518e-05,
          "validation_metric": 2.3313865312957205e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.502299187999597
        },
        {
          "epoch": 399,
          "updates": 6384,
          "samples_seen": 408576,
          "online_loss": 0.0002580402478997712,
          "train_loss": 0.00025311685749329627,
          "validation_loss": 0.00022698097745887935,
          "train_metric": 2.5493001885479316e-05,
          "validation_metric": 2.286069320689421e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.512646860999666
        },
        {
          "epoch": 400,
          "updates": 6400,
          "samples_seen": 409600,
          "online_loss": 0.0002580706077424111,
          "train_loss": 0.0002530453202780336,
          "validation_loss": 0.00022841652389615774,
          "train_metric": 2.5485798687441275e-05,
          "validation_metric": 2.3005275579635054e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.523094149000826
        },
        {
          "epoch": 401,
          "updates": 6416,
          "samples_seen": 410624,
          "online_loss": 0.0002567022011135123,
          "train_loss": 0.00025238803937099874,
          "validation_loss": 0.0002278953033965081,
          "train_metric": 2.541960020607803e-05,
          "validation_metric": 2.2952777726459317e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.533390414000678
        },
        {
          "epoch": 402,
          "updates": 6432,
          "samples_seen": 411648,
          "online_loss": 0.00025825459124462213,
          "train_loss": 0.00025351170916110277,
          "validation_loss": 0.0002303222136106342,
          "train_metric": 2.5532768631819636e-05,
          "validation_metric": 2.319721352250781e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.5438035179995495
        },
        {
          "epoch": 403,
          "updates": 6448,
          "samples_seen": 412672,
          "online_loss": 0.00026065575457323575,
          "train_loss": 0.0002521442365832627,
          "validation_loss": 0.00022614044428337365,
          "train_metric": 2.539504384913016e-05,
          "validation_metric": 2.2776035621063784e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.554649714000334
        },
        {
          "epoch": 404,
          "updates": 6464,
          "samples_seen": 413696,
          "online_loss": 0.0002550293547756155,
          "train_loss": 0.0002524571609683335,
          "validation_loss": 0.00022695385268889368,
          "train_metric": 2.5426557840546593e-05,
          "validation_metric": 2.2857961084810086e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.5650219790004485
        },
        {
          "epoch": 405,
          "updates": 6480,
          "samples_seen": 414720,
          "online_loss": 0.0002558659916758188,
          "train_loss": 0.00025183570687659085,
          "validation_loss": 0.0002252075501019135,
          "train_metric": 2.5363973691128194e-05,
          "validation_metric": 2.2682081180391833e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.630814169000587
        },
        {
          "epoch": 406,
          "updates": 6496,
          "samples_seen": 415744,
          "online_loss": 0.00026163421898672823,
          "train_loss": 0.0002523981384001672,
          "validation_loss": 0.0002302671637153253,
          "train_metric": 2.5420618840144016e-05,
          "validation_metric": 2.319166378583759e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.641573945000346
        },
        {
          "epoch": 407,
          "updates": 6512,
          "samples_seen": 416768,
          "online_loss": 0.0002563128537076409,
          "train_loss": 0.00025159146753139794,
          "validation_loss": 0.00022358589922077954,
          "train_metric": 2.533937367843464e-05,
          "validation_metric": 2.2518750483868644e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.6525105189994065
        },
        {
          "epoch": 408,
          "updates": 6528,
          "samples_seen": 417792,
          "online_loss": 0.0002567639157859958,
          "train_loss": 0.0002513417275622487,
          "validation_loss": 0.0002263235073769465,
          "train_metric": 2.5314218873973005e-05,
          "validation_metric": 2.279447653563693e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.662947741999233
        },
        {
          "epoch": 409,
          "updates": 6544,
          "samples_seen": 418816,
          "online_loss": 0.00025458076106588123,
          "train_loss": 0.0002529885387048125,
          "validation_loss": 0.00022532526054419577,
          "train_metric": 2.548007614677772e-05,
          "validation_metric": 2.269393917231355e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.67325068900027
        },
        {
          "epoch": 410,
          "updates": 6560,
          "samples_seen": 419840,
          "online_loss": 0.0002595549149191356,
          "train_loss": 0.00025053753051906824,
          "validation_loss": 0.000227069846005179,
          "train_metric": 2.5233224732801318e-05,
          "validation_metric": 2.286964445374906e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.683603554000001
        },
        {
          "epoch": 411,
          "updates": 6576,
          "samples_seen": 420864,
          "online_loss": 0.00025623679630371043,
          "train_loss": 0.00025080726481974125,
          "validation_loss": 0.00022806529887020588,
          "train_metric": 2.526038770156447e-05,
          "validation_metric": 2.2969901692704298e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.693851234000249
        },
        {
          "epoch": 412,
          "updates": 6592,
          "samples_seen": 421888,
          "online_loss": 0.00025529536014801124,
          "train_loss": 0.0002520361158531159,
          "validation_loss": 0.0002249130338896066,
          "train_metric": 2.5384153559571132e-05,
          "validation_metric": 2.265241892018821e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.7040834550007276
        },
        {
          "epoch": 413,
          "updates": 6608,
          "samples_seen": 422912,
          "online_loss": 0.0002547648973632022,
          "train_loss": 0.0002528832119423896,
          "validation_loss": 0.00022877962328493595,
          "train_metric": 2.5469469619565643e-05,
          "validation_metric": 2.304184454260394e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.714485344000423
        },
        {
          "epoch": 414,
          "updates": 6624,
          "samples_seen": 423936,
          "online_loss": 0.00025398148409294663,
          "train_loss": 0.00025093561271205544,
          "validation_loss": 0.00022532457660418004,
          "train_metric": 2.5273318897234276e-05,
          "validation_metric": 2.269386823172681e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.7247660320008436
        },
        {
          "epoch": 415,
          "updates": 6640,
          "samples_seen": 424960,
          "online_loss": 0.00025598970751161687,
          "train_loss": 0.00025041800108738244,
          "validation_loss": 0.0002262973866891116,
          "train_metric": 2.5221183022949845e-05,
          "validation_metric": 2.2791846276959404e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.735263730999577
        },
        {
          "epoch": 416,
          "updates": 6656,
          "samples_seen": 425984,
          "online_loss": 0.00025328546689706855,
          "train_loss": 0.0002492136263754219,
          "validation_loss": 0.00022403168259188533,
          "train_metric": 2.5099881895584986e-05,
          "validation_metric": 2.2563652237295173e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.7462981839998974
        },
        {
          "epoch": 417,
          "updates": 6672,
          "samples_seen": 427008,
          "online_loss": 0.0002540697996664676,
          "train_loss": 0.00024922064039856195,
          "validation_loss": 0.0002264553913846612,
          "train_metric": 2.5100593120441772e-05,
          "validation_metric": 2.2807760615251027e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.756810199000029
        },
        {
          "epoch": 418,
          "updates": 6688,
          "samples_seen": 428032,
          "online_loss": 0.0002541177227612934,
          "train_loss": 0.0002490659535396844,
          "validation_loss": 0.0002237866574432701,
          "train_metric": 2.5085009838221595e-05,
          "validation_metric": 2.2538974008057266e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.7671735910007556
        },
        {
          "epoch": 419,
          "updates": 6704,
          "samples_seen": 429056,
          "online_loss": 0.00025110384331128444,
          "train_loss": 0.0002487126039341092,
          "validation_loss": 0.00022657596855424345,
          "train_metric": 2.5049419491551816e-05,
          "validation_metric": 2.2819900550530292e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.777552370000194
        },
        {
          "epoch": 420,
          "updates": 6720,
          "samples_seen": 430080,
          "online_loss": 0.00025669911883596797,
          "train_loss": 0.00024984433548524976,
          "validation_loss": 0.0002260576147818938,
          "train_metric": 2.5163406462525018e-05,
          "validation_metric": 2.276769737363793e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.787879887000599
        },
        {
          "epoch": 421,
          "updates": 6736,
          "samples_seen": 431104,
          "online_loss": 0.00025679569989733864,
          "train_loss": 0.0002517466782592237,
          "validation_loss": 0.00022587651619687676,
          "train_metric": 2.5355006073368713e-05,
          "validation_metric": 2.2749458366888575e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.798322122000172
        },
        {
          "epoch": 422,
          "updates": 6752,
          "samples_seen": 432128,
          "online_loss": 0.0002528088652979932,
          "train_loss": 0.00024804958957247436,
          "validation_loss": 0.00022394010738935322,
          "train_metric": 2.4982642571558245e-05,
          "validation_metric": 2.255442632304039e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.8087050070007535
        },
        {
          "epoch": 423,
          "updates": 6768,
          "samples_seen": 433152,
          "online_loss": 0.00025176413601002423,
          "train_loss": 0.0002489910111762583,
          "validation_loss": 0.00022445988724939525,
          "train_metric": 2.5077464670175686e-05,
          "validation_metric": 2.2606780476053245e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.819114236999667
        },
        {
          "epoch": 424,
          "updates": 6784,
          "samples_seen": 434176,
          "online_loss": 0.00025125335560005624,
          "train_loss": 0.0002476354711689055,
          "validation_loss": 0.00022260508558247238,
          "train_metric": 2.494093860150315e-05,
          "validation_metric": 2.2419970264309086e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.830054519999976
        },
        {
          "epoch": 425,
          "updates": 6800,
          "samples_seen": 435200,
          "online_loss": 0.000252582278335467,
          "train_loss": 0.000247805321123451,
          "validation_loss": 0.00022300246928352863,
          "train_metric": 2.4958046196843497e-05,
          "validation_metric": 2.2459993488155305e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.8404700930004765
        },
        {
          "epoch": 426,
          "updates": 6816,
          "samples_seen": 436224,
          "online_loss": 0.00025538272893754765,
          "train_loss": 0.0002471373591106385,
          "validation_loss": 0.00022326175530906767,
          "train_metric": 2.4890767235774547e-05,
          "validation_metric": 2.248610690003261e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.850961779999125
        },
        {
          "epoch": 427,
          "updates": 6832,
          "samples_seen": 437248,
          "online_loss": 0.00025574690698704217,
          "train_loss": 0.00024990126257762313,
          "validation_loss": 0.00022882511257193983,
          "train_metric": 2.5169141736114398e-05,
          "validation_metric": 2.3046432033879682e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.861316308999449
        },
        {
          "epoch": 428,
          "updates": 6848,
          "samples_seen": 438272,
          "online_loss": 0.0002663822824615636,
          "train_loss": 0.00024844854488037527,
          "validation_loss": 0.00022270911722443998,
          "train_metric": 2.5022829504450783e-05,
          "validation_metric": 2.243044764327351e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.8715724689991475
        },
        {
          "epoch": 429,
          "updates": 6864,
          "samples_seen": 439296,
          "online_loss": 0.0002510840577087947,
          "train_loss": 0.0002505520242266357,
          "validation_loss": 0.0002263374044559896,
          "train_metric": 2.5234685381292365e-05,
          "validation_metric": 2.2795873519498855e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.882055313000819
        },
        {
          "epoch": 430,
          "updates": 6880,
          "samples_seen": 440320,
          "online_loss": 0.0002515238784326357,
          "train_loss": 0.0002492751518730074,
          "validation_loss": 0.00022109670680947602,
          "train_metric": 2.510608101147227e-05,
          "validation_metric": 2.2268048269324936e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.892989556999964
        },
        {
          "epoch": 431,
          "updates": 6896,
          "samples_seen": 441344,
          "online_loss": 0.0002508325278540724,
          "train_loss": 0.0002463200653437525,
          "validation_loss": 0.00022495554003398865,
          "train_metric": 2.4808457965264097e-05,
          "validation_metric": 2.2656699002254754e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.903306973999861
        },
        {
          "epoch": 432,
          "updates": 6912,
          "samples_seen": 442368,
          "online_loss": 0.0002489170874468982,
          "train_loss": 0.0002486813464201987,
          "validation_loss": 0.0002244488860014826,
          "train_metric": 2.5046272639883682e-05,
          "validation_metric": 2.2605667254538275e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.913954575000389
        },
        {
          "epoch": 433,
          "updates": 6928,
          "samples_seen": 443392,
          "online_loss": 0.0002577755867605447,
          "train_loss": 0.00024532637326046824,
          "validation_loss": 0.0002248362434329465,
          "train_metric": 2.47083753492916e-05,
          "validation_metric": 2.2644679120276123e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.924278802000117
        },
        {
          "epoch": 434,
          "updates": 6944,
          "samples_seen": 444416,
          "online_loss": 0.0002551874094933737,
          "train_loss": 0.0002452560293022543,
          "validation_loss": 0.00022153632016852498,
          "train_metric": 2.4701288566575386e-05,
          "validation_metric": 2.2312327928375453e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.934680301999833
        },
        {
          "epoch": 435,
          "updates": 6960,
          "samples_seen": 445440,
          "online_loss": 0.0002559131085035915,
          "train_loss": 0.0002454764908179641,
          "validation_loss": 0.0002269811520818621,
          "train_metric": 2.4723496608203277e-05,
          "validation_metric": 2.2860709577798843e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.9451189060000615
        },
        {
          "epoch": 436,
          "updates": 6976,
          "samples_seen": 446464,
          "online_loss": 0.00025445509345445316,
          "train_loss": 0.00025435732095502317,
          "validation_loss": 0.00023080417304299772,
          "train_metric": 2.561793735367246e-05,
          "validation_metric": 2.3245751435752027e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.955509225999776
        },
        {
          "epoch": 437,
          "updates": 6992,
          "samples_seen": 447488,
          "online_loss": 0.00025663355518190656,
          "train_loss": 0.00024633461725898087,
          "validation_loss": 0.00022070475097279996,
          "train_metric": 2.4809920432744548e-05,
          "validation_metric": 2.2228574380278587e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.966348641999502
        },
        {
          "epoch": 438,
          "updates": 7008,
          "samples_seen": 448512,
          "online_loss": 0.0002483962971382425,
          "train_loss": 0.0002447384758852422,
          "validation_loss": 0.00022324666497297585,
          "train_metric": 2.464915996824857e-05,
          "validation_metric": 2.2484589862870052e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.977028389999759
        },
        {
          "epoch": 439,
          "updates": 7024,
          "samples_seen": 449536,
          "online_loss": 0.00024884636604838306,
          "train_loss": 0.00024397972447331995,
          "validation_loss": 0.0002189964579883963,
          "train_metric": 2.4572744223405607e-05,
          "validation_metric": 2.2056519810575992e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.987925614999767
        },
        {
          "epoch": 440,
          "updates": 7040,
          "samples_seen": 450560,
          "online_loss": 0.00024690398004167946,
          "train_loss": 0.0002441327815176919,
          "validation_loss": 0.00022318126866593957,
          "train_metric": 2.4588160158600658e-05,
          "validation_metric": 2.247800148325041e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.998646531999839
        },
        {
          "epoch": 441,
          "updates": 7056,
          "samples_seen": 451584,
          "online_loss": 0.00024681600461917697,
          "train_loss": 0.00024355374625883996,
          "validation_loss": 0.00022145146795082837,
          "train_metric": 2.452983972034417e-05,
          "validation_metric": 2.230378049716819e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.064626302000761
        },
        {
          "epoch": 442,
          "updates": 7072,
          "samples_seen": 452608,
          "online_loss": 0.0002505270886103972,
          "train_loss": 0.00024546930217184126,
          "validation_loss": 0.00022199051454663277,
          "train_metric": 2.4722772650420666e-05,
          "validation_metric": 2.2358073692885228e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.075272781999956
        },
        {
          "epoch": 443,
          "updates": 7088,
          "samples_seen": 453632,
          "online_loss": 0.00024718602617213037,
          "train_loss": 0.0002435624919598922,
          "validation_loss": 0.0002263114001834765,
          "train_metric": 2.4530723749194294e-05,
          "validation_metric": 2.2793255993747152e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.085615042999052
        },
        {
          "epoch": 444,
          "updates": 7104,
          "samples_seen": 454656,
          "online_loss": 0.00025673923028080026,
          "train_loss": 0.00024413276696577668,
          "validation_loss": 0.00021812711202073842,
          "train_metric": 2.4588158339611255e-05,
          "validation_metric": 2.196896639361512e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.096505719000561
        },
        {
          "epoch": 445,
          "updates": 7120,
          "samples_seen": 455680,
          "online_loss": 0.0002518738356229733,
          "train_loss": 0.00024365987337660044,
          "validation_loss": 0.00022422228357754648,
          "train_metric": 2.4540531740058213e-05,
          "validation_metric": 2.2582848032470793e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.106920407999496
        },
        {
          "epoch": 446,
          "updates": 7136,
          "samples_seen": 456704,
          "online_loss": 0.0002468684415362077,
          "train_loss": 0.00024260720238089561,
          "validation_loss": 0.0002182260504923761,
          "train_metric": 2.443451012368314e-05,
          "validation_metric": 2.197892899857834e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.117441382000834
        },
        {
          "epoch": 447,
          "updates": 7152,
          "samples_seen": 457728,
          "online_loss": 0.0002545328952692216,
          "train_loss": 0.000244297698372975,
          "validation_loss": 0.00022106721007730812,
          "train_metric": 2.460477116983384e-05,
          "validation_metric": 2.2265077859628946e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.127699648999624
        },
        {
          "epoch": 448,
          "updates": 7168,
          "samples_seen": 458752,
          "online_loss": 0.0002479843024048023,
          "train_loss": 0.0002465937868691981,
          "validation_loss": 0.00022127605916466564,
          "train_metric": 2.483602293068543e-05,
          "validation_metric": 2.2286116291070357e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.13827652100008
        },
        {
          "epoch": 449,
          "updates": 7184,
          "samples_seen": 459776,
          "online_loss": 0.00024962385668914067,
          "train_loss": 0.00024177940213121474,
          "validation_loss": 0.00022306609025690705,
          "train_metric": 2.4351136744371615e-05,
          "validation_metric": 2.2466401787823997e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.148716383999272
        },
        {
          "epoch": 450,
          "updates": 7200,
          "samples_seen": 460800,
          "online_loss": 0.00024582092510172515,
          "train_loss": 0.00024160662724170834,
          "validation_loss": 0.00021947463392280042,
          "train_metric": 2.433373265375849e-05,
          "validation_metric": 2.2104681193013676e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.159006956000667
        },
        {
          "epoch": 451,
          "updates": 7216,
          "samples_seen": 461824,
          "online_loss": 0.0002512770952307619,
          "train_loss": 0.00024122963077388704,
          "validation_loss": 0.00021902332082390785,
          "train_metric": 2.429576670692768e-05,
          "validation_metric": 2.205922828579787e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.169483080999271
        },
        {
          "epoch": 452,
          "updates": 7232,
          "samples_seen": 462848,
          "online_loss": 0.0002546466566855088,
          "train_loss": 0.00024171313270926476,
          "validation_loss": 0.00022299264674074948,
          "train_metric": 2.4344461053260602e-05,
          "validation_metric": 2.2459002138930373e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.180281640999965
        },
        {
          "epoch": 453,
          "updates": 7248,
          "samples_seen": 463872,
          "online_loss": 0.00025207752241840353,
          "train_loss": 0.00024094281252473593,
          "validation_loss": 0.00021750698215328157,
          "train_metric": 2.4266877517220564e-05,
          "validation_metric": 2.1906504116486758e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.1914143659996626
        },
        {
          "epoch": 454,
          "updates": 7264,
          "samples_seen": 464896,
          "online_loss": 0.0002491037566869636,
          "train_loss": 0.00024306841078214347,
          "validation_loss": 0.0002172693348256871,
          "train_metric": 2.4480959837092087e-05,
          "validation_metric": 2.18825698539149e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.202684997000688
        },
        {
          "epoch": 455,
          "updates": 7280,
          "samples_seen": 465920,
          "online_loss": 0.0002448939958412666,
          "train_loss": 0.00024196076265070587,
          "validation_loss": 0.00022499123588204384,
          "train_metric": 2.436940121697262e-05,
          "validation_metric": 2.266029332531616e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.213370858000417
        },
        {
          "epoch": 456,
          "updates": 7296,
          "samples_seen": 466944,
          "online_loss": 0.00024850158206390915,
          "train_loss": 0.00024125934578478336,
          "validation_loss": 0.00021445975289680064,
          "train_metric": 2.4298758944496512e-05,
          "validation_metric": 2.1599602405331098e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.224352253999314
        },
        {
          "epoch": 457,
          "updates": 7312,
          "samples_seen": 467968,
          "online_loss": 0.00025196949536621105,
          "train_loss": 0.0002411938039585948,
          "validation_loss": 0.00022032976266928017,
          "train_metric": 2.429215601296164e-05,
          "validation_metric": 2.2190806703292765e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.234697178000715
        },
        {
          "epoch": 458,
          "updates": 7328,
          "samples_seen": 468992,
          "online_loss": 0.0002574528243712848,
          "train_loss": 0.00024958461290225387,
          "validation_loss": 0.00023144482111092657,
          "train_metric": 2.5137247575912625e-05,
          "validation_metric": 2.3310274627874605e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.2452626489994145
        },
        {
          "epoch": 459,
          "updates": 7344,
          "samples_seen": 470016,
          "online_loss": 0.00026151274960284354,
          "train_loss": 0.00024007449974305928,
          "validation_loss": 0.000218346351175569,
          "train_metric": 2.4179424144676886e-05,
          "validation_metric": 2.199104164901655e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.255628335000438
        },
        {
          "epoch": 460,
          "updates": 7360,
          "samples_seen": 471040,
          "online_loss": 0.0002482004374542157,
          "train_loss": 0.00023987195163499564,
          "validation_loss": 0.00021854501392226666,
          "train_metric": 2.4159022359526716e-05,
          "validation_metric": 2.201105235144496e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.265983654000593
        },
        {
          "epoch": 461,
          "updates": 7376,
          "samples_seen": 472064,
          "online_loss": 0.00025247150824725395,
          "train_loss": 0.00024108972866088152,
          "validation_loss": 0.0002202192263212055,
          "train_metric": 2.428167499601841e-05,
          "validation_metric": 2.2179674488143064e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.276342646000558
        },
        {
          "epoch": 462,
          "updates": 7392,
          "samples_seen": 473088,
          "online_loss": 0.0002605710315037868,
          "train_loss": 0.00024089850194286555,
          "validation_loss": 0.00021678987832274288,
          "train_metric": 2.426241735520307e-05,
          "validation_metric": 2.183428114221897e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.286799659999815
        },
        {
          "epoch": 463,
          "updates": 7408,
          "samples_seen": 474112,
          "online_loss": 0.00024321271666849498,
          "train_loss": 0.00024645260418765247,
          "validation_loss": 0.00023056553618516773,
          "train_metric": 2.482180207152851e-05,
          "validation_metric": 2.322171894775238e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.29770973199993
        },
        {
          "epoch": 464,
          "updates": 7424,
          "samples_seen": 475136,
          "online_loss": 0.0002490734377715853,
          "train_loss": 0.0002416962233837694,
          "validation_loss": 0.00021875754464417696,
          "train_metric": 2.4342758479178883e-05,
          "validation_metric": 2.2032461856724694e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.308368993999466
        },
        {
          "epoch": 465,
          "updates": 7440,
          "samples_seen": 476160,
          "online_loss": 0.00023946064447954996,
          "train_loss": 0.0002419857628410682,
          "validation_loss": 0.00022306408209260553,
          "train_metric": 2.4371918698307127e-05,
          "validation_metric": 2.2466196242021397e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.318894904999979
        },
        {
          "epoch": 466,
          "updates": 7456,
          "samples_seen": 477184,
          "online_loss": 0.0002480372040736256,
          "train_loss": 0.00023803462681826204,
          "validation_loss": 0.00021765210840385407,
          "train_metric": 2.3973976567503996e-05,
          "validation_metric": 2.1921123334323056e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.329538682999555
        },
        {
          "epoch": 467,
          "updates": 7472,
          "samples_seen": 478208,
          "online_loss": 0.00024251102604466723,
          "train_loss": 0.00023803451040294021,
          "validation_loss": 0.00021577956795226783,
          "train_metric": 2.3973965653567575e-05,
          "validation_metric": 2.1732526874984615e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.33988039399992
        },
        {
          "epoch": 468,
          "updates": 7488,
          "samples_seen": 479232,
          "online_loss": 0.00024289827888424043,
          "train_loss": 0.00024031559587456286,
          "validation_loss": 0.00021994342387188226,
          "train_metric": 2.4203707653214224e-05,
          "validation_metric": 2.215189851995092e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.350447017999613
        },
        {
          "epoch": 469,
          "updates": 7504,
          "samples_seen": 480256,
          "online_loss": 0.00024355653067686944,
          "train_loss": 0.0002373198076384142,
          "validation_loss": 0.00021644163643941283,
          "train_metric": 2.3901982785901055e-05,
          "validation_metric": 2.17992073885398e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.360832902999391
        },
        {
          "epoch": 470,
          "updates": 7520,
          "samples_seen": 481280,
          "online_loss": 0.00023998689948712126,
          "train_loss": 0.0002400734811089933,
          "validation_loss": 0.00021945423213765025,
          "train_metric": 2.4179320462280884e-05,
          "validation_metric": 2.2102627553977072e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.3711468069996045
        },
        {
          "epoch": 471,
          "updates": 7536,
          "samples_seen": 482304,
          "online_loss": 0.0002435387405057554,
          "train_loss": 0.00023680258891545236,
          "validation_loss": 0.00021719082724303007,
          "train_metric": 2.38498869293835e-05,
          "validation_metric": 2.187466452596709e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.381599976999496
        },
        {
          "epoch": 472,
          "updates": 7552,
          "samples_seen": 483328,
          "online_loss": 0.0002474314560458879,
          "train_loss": 0.00023918677470646799,
          "validation_loss": 0.00022031820844858885,
          "train_metric": 2.4090015358524397e-05,
          "validation_metric": 2.2189642550074495e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.392050653999831
        },
        {
          "epoch": 473,
          "updates": 7568,
          "samples_seen": 484352,
          "online_loss": 0.0002448858558636857,
          "train_loss": 0.00023648201022297144,
          "validation_loss": 0.00021389173343777657,
          "train_metric": 2.3817601686459966e-05,
          "validation_metric": 2.1542391550610773e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.402851635000843
        },
        {
          "epoch": 474,
          "updates": 7584,
          "samples_seen": 485376,
          "online_loss": 0.0002394119901509839,
          "train_loss": 0.0002362522791372612,
          "validation_loss": 0.00021747779101133347,
          "train_metric": 2.3794465960236266e-05,
          "validation_metric": 2.1903568267589435e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.41361754799982
        },
        {
          "epoch": 475,
          "updates": 7600,
          "samples_seen": 486400,
          "online_loss": 0.0002503128789612674,
          "train_loss": 0.00024381466209888458,
          "validation_loss": 0.00021598575403913856,
          "train_metric": 2.455612411722541e-05,
          "validation_metric": 2.1753294277004898e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.4242776350001805
        },
        {
          "epoch": 476,
          "updates": 7616,
          "samples_seen": 487424,
          "online_loss": 0.0002473877102602273,
          "train_loss": 0.00024014040536712855,
          "validation_loss": 0.00022866988729219884,
          "train_metric": 2.418606345599983e-05,
          "validation_metric": 2.3030792362987995e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.434635612999045
        },
        {
          "epoch": 477,
          "updates": 7632,
          "samples_seen": 488448,
          "online_loss": 0.00025141084825008875,
          "train_loss": 0.00023596077517140657,
          "validation_loss": 0.00021429872140288353,
          "train_metric": 2.3765105652273633e-05,
          "validation_metric": 2.1583384295809083e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.499207041000773
        },
        {
          "epoch": 478,
          "updates": 7648,
          "samples_seen": 489472,
          "online_loss": 0.00024166587445506593,
          "train_loss": 0.00023811818391550332,
          "validation_loss": 0.0002181323798140511,
          "train_metric": 2.3982389393495396e-05,
          "validation_metric": 2.196949571953155e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.509928105000654
        },
        {
          "epoch": 479,
          "updates": 7664,
          "samples_seen": 490496,
          "online_loss": 0.0002454477162245894,
          "train_loss": 0.00023552845232188702,
          "validation_loss": 0.00021572207333520055,
          "train_metric": 2.3721562683931552e-05,
          "validation_metric": 2.172673885070253e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.520217019999109
        },
        {
          "epoch": 480,
          "updates": 7680,
          "samples_seen": 491520,
          "online_loss": 0.00024267354092444293,
          "train_loss": 0.00023560434055980295,
          "validation_loss": 0.00021652018767781556,
          "train_metric": 2.3729207896394655e-05,
          "validation_metric": 2.180711817345582e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.530823105000309
        },
        {
          "epoch": 481,
          "updates": 7696,
          "samples_seen": 492544,
          "online_loss": 0.00023982019320101244,
          "train_loss": 0.00023468451399821788,
          "validation_loss": 0.0002123627346009016,
          "train_metric": 2.363656494708266e-05,
          "validation_metric": 2.1388399545685388e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.541869361999488
        },
        {
          "epoch": 482,
          "updates": 7712,
          "samples_seen": 493568,
          "online_loss": 0.00023898644030850846,
          "train_loss": 0.00023544294526800513,
          "validation_loss": 0.00022057777096051723,
          "train_metric": 2.3712949769105762e-05,
          "validation_metric": 2.221578688477166e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.552350952000779
        },
        {
          "epoch": 483,
          "updates": 7728,
          "samples_seen": 494592,
          "online_loss": 0.0002398948654445121,
          "train_loss": 0.00023491412866860628,
          "validation_loss": 0.00021437527902889997,
          "train_metric": 2.3659693397348747e-05,
          "validation_metric": 2.159109317290131e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.562871635000192
        },
        {
          "epoch": 484,
          "updates": 7744,
          "samples_seen": 495616,
          "online_loss": 0.00023794296339474386,
          "train_loss": 0.0002343086525797844,
          "validation_loss": 0.00021266998373903334,
          "train_metric": 2.3598708139616065e-05,
          "validation_metric": 2.1419342374429107e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.5732760509999935
        },
        {
          "epoch": 485,
          "updates": 7760,
          "samples_seen": 496640,
          "online_loss": 0.00023863820206315722,
          "train_loss": 0.0002335377357667312,
          "validation_loss": 0.00021363691485021263,
          "train_metric": 2.3521066395915113e-05,
          "validation_metric": 2.1516731067094952e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.583631081999556
        },
        {
          "epoch": 486,
          "updates": 7776,
          "samples_seen": 497664,
          "online_loss": 0.0002436552858853247,
          "train_loss": 0.00023468797735404223,
          "validation_loss": 0.00021921850566286594,
          "train_metric": 2.3636914193048142e-05,
          "validation_metric": 2.2078882466303185e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.594126904999939
        },
        {
          "epoch": 487,
          "updates": 7792,
          "samples_seen": 498688,
          "online_loss": 0.00023669637630518992,
          "train_loss": 0.0002335289173061028,
          "validation_loss": 0.00021193661086726934,
          "train_metric": 2.352017691009678e-05,
          "validation_metric": 2.1345476852729917e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.605205764999482
        },
        {
          "epoch": 488,
          "updates": 7808,
          "samples_seen": 499712,
          "online_loss": 0.0002358684550927137,
          "train_loss": 0.0002335557364858687,
          "validation_loss": 0.0002136917901225388,
          "train_metric": 2.352287629037164e-05,
          "validation_metric": 2.1522253518924117e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.616098172999045
        },
        {
          "epoch": 489,
          "updates": 7824,
          "samples_seen": 500736,
          "online_loss": 0.0002351069661017391,
          "train_loss": 0.00023280117602553219,
          "validation_loss": 0.00021431647473946214,
          "train_metric": 2.3446882551070303e-05,
          "validation_metric": 2.1585170543403365e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.626545360000819
        },
        {
          "epoch": 490,
          "updates": 7840,
          "samples_seen": 501760,
          "online_loss": 0.00023514648182754172,
          "train_loss": 0.0002324692322872579,
          "validation_loss": 0.0002155251568183303,
          "train_metric": 2.341344952583313e-05,
          "validation_metric": 2.170690095226746e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.637215628999911
        },
        {
          "epoch": 491,
          "updates": 7856,
          "samples_seen": 502784,
          "online_loss": 0.00023925869800223154,
          "train_loss": 0.00023259714362211525,
          "validation_loss": 0.0002122539735864848,
          "train_metric": 2.342633160878904e-05,
          "validation_metric": 2.1377441953518428e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.647607974000493
        },
        {
          "epoch": 492,
          "updates": 7872,
          "samples_seen": 503808,
          "online_loss": 0.00024094892341963714,
          "train_loss": 0.00023401560611091554,
          "validation_loss": 0.0002185224584536627,
          "train_metric": 2.356919321755413e-05,
          "validation_metric": 2.2008784071658738e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.658124448000308
        },
        {
          "epoch": 493,
          "updates": 7888,
          "samples_seen": 504832,
          "online_loss": 0.0002421275021333713,
          "train_loss": 0.00023515849898103625,
          "validation_loss": 0.00021656892204191536,
          "train_metric": 2.3684300685999915e-05,
          "validation_metric": 2.1812029444845393e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.668365372999688
        },
        {
          "epoch": 494,
          "updates": 7904,
          "samples_seen": 505856,
          "online_loss": 0.00023464633795811096,
          "train_loss": 0.00023738527670502663,
          "validation_loss": 0.00021886106696911156,
          "train_metric": 2.390857662248891e-05,
          "validation_metric": 2.2042884666007012e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.678784673000337
        },
        {
          "epoch": 495,
          "updates": 7920,
          "samples_seen": 506880,
          "online_loss": 0.00023688687724643387,
          "train_loss": 0.00023349770344793797,
          "validation_loss": 0.00021582338376902044,
          "train_metric": 2.3517035515396856e-05,
          "validation_metric": 2.173694156226702e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.689027395999801
        },
        {
          "epoch": 496,
          "updates": 7936,
          "samples_seen": 507904,
          "online_loss": 0.0002507131521269912,
          "train_loss": 0.00024182646302506328,
          "validation_loss": 0.00022016058210283518,
          "train_metric": 2.4355877030757256e-05,
          "validation_metric": 2.217376822954975e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.699301995000496
        },
        {
          "epoch": 497,
          "updates": 7952,
          "samples_seen": 508928,
          "online_loss": 0.00024908079012675444,
          "train_loss": 0.0002320497005712241,
          "validation_loss": 0.00021106566418893635,
          "train_metric": 2.337119440198876e-05,
          "validation_metric": 2.1257761545712128e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.710129549999692
        },
        {
          "epoch": 498,
          "updates": 7968,
          "samples_seen": 509952,
          "online_loss": 0.00025376389385201037,
          "train_loss": 0.00023467594292014837,
          "validation_loss": 0.00022009726671967655,
          "train_metric": 2.3635700927115977e-05,
          "validation_metric": 2.216739085270092e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.720688592000442
        },
        {
          "epoch": 499,
          "updates": 7984,
          "samples_seen": 510976,
          "online_loss": 0.00023700636029388988,
          "train_loss": 0.0002316425961907953,
          "validation_loss": 0.00020934913482051343,
          "train_metric": 2.3330192561843432e-05,
          "validation_metric": 2.108487933583092e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.73148637200029
        },
        {
          "epoch": 500,
          "updates": 8000,
          "samples_seen": 512000,
          "online_loss": 0.00024100870450638467,
          "train_loss": 0.00023226147459354252,
          "validation_loss": 0.00021634047152474523,
          "train_metric": 2.3392525690724142e-05,
          "validation_metric": 2.1789019228890538e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.74199932800002
        },
        {
          "epoch": 501,
          "updates": 8016,
          "samples_seen": 513024,
          "online_loss": 0.00023843707822379656,
          "train_loss": 0.00023108444293029606,
          "validation_loss": 0.0002109222114086151,
          "train_metric": 2.327397851331625e-05,
          "validation_metric": 2.1243315131869167e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.752350544000365
        },
        {
          "epoch": 502,
          "updates": 8032,
          "samples_seen": 514048,
          "online_loss": 0.0002347957897654851,
          "train_loss": 0.00023117400996852666,
          "validation_loss": 0.00021796076907776296,
          "train_metric": 2.3282998881768435e-05,
          "validation_metric": 2.1952211682219058e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.762924218999615
        },
        {
          "epoch": 503,
          "updates": 8048,
          "samples_seen": 515072,
          "online_loss": 0.00023556344513053773,
          "train_loss": 0.00022955125314183533,
          "validation_loss": 0.00021202018251642585,
          "train_metric": 2.3119559045881033e-05,
          "validation_metric": 2.1353895135689527e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.773220327000672
        },
        {
          "epoch": 504,
          "updates": 8064,
          "samples_seen": 516096,
          "online_loss": 0.0002412493658994208,
          "train_loss": 0.00023293211415875703,
          "validation_loss": 0.0002143997116945684,
          "train_metric": 2.3460068405256607e-05,
          "validation_metric": 2.159355608455371e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.78362140200079
        },
        {
          "epoch": 505,
          "updates": 8080,
          "samples_seen": 517120,
          "online_loss": 0.00023301126748265233,
          "train_loss": 0.00023167232575360686,
          "validation_loss": 0.00021825700241606683,
          "train_metric": 2.3333188437391073e-05,
          "validation_metric": 2.1982046746416017e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.793856557000254
        },
        {
          "epoch": 506,
          "updates": 8096,
          "samples_seen": 518144,
          "online_loss": 0.00023623783727089176,
          "train_loss": 0.00022977714252192527,
          "validation_loss": 0.0002121769794030115,
          "train_metric": 2.3142312784329988e-05,
          "validation_metric": 2.136968578270171e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.804190332999497
        },
        {
          "epoch": 507,
          "updates": 8112,
          "samples_seen": 519168,
          "online_loss": 0.0002562676672823727,
          "train_loss": 0.00023827403492759913,
          "validation_loss": 0.0002126454346580431,
          "train_metric": 2.39980890910374e-05,
          "validation_metric": 2.1416870367829688e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.8147018019990355
        },
        {
          "epoch": 508,
          "updates": 8128,
          "samples_seen": 520192,
          "online_loss": 0.0002261760173496441,
          "train_loss": 0.00023243355099111795,
          "validation_loss": 0.0002173159155063331,
          "train_metric": 2.340985884075053e-05,
          "validation_metric": 2.1887264665565453e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.825207002999377
        },
        {
          "epoch": 509,
          "updates": 8144,
          "samples_seen": 521216,
          "online_loss": 0.0002387923959759064,
          "train_loss": 0.00022932117281015962,
          "validation_loss": 0.0002101906866300851,
          "train_metric": 2.3096388758858666e-05,
          "validation_metric": 2.116963514708914e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.835599775000446
        },
        {
          "epoch": 510,
          "updates": 8160,
          "samples_seen": 522240,
          "online_loss": 0.00023232292687680456,
          "train_loss": 0.00022893484856467694,
          "validation_loss": 0.00021008857584092766,
          "train_metric": 2.3057476937538013e-05,
          "validation_metric": 2.115935058100149e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.846583204000126
        },
        {
          "epoch": 511,
          "updates": 8176,
          "samples_seen": 523264,
          "online_loss": 0.00023939364291436505,
          "train_loss": 0.0002279944601468742,
          "validation_loss": 0.00021229895355645567,
          "train_metric": 2.2962765797274187e-05,
          "validation_metric": 2.138197305612266e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.857256159000826
        },
        {
          "epoch": 512,
          "updates": 8192,
          "samples_seen": 524288,
          "online_loss": 0.00024143904556694906,
          "train_loss": 0.00023166104801930487,
          "validation_loss": 0.00021174087305553257,
          "train_metric": 2.333205338800326e-05,
          "validation_metric": 2.132576446456369e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.867739871000595
        },
        {
          "epoch": 513,
          "updates": 8208,
          "samples_seen": 525312,
          "online_loss": 0.00023783742017258191,
          "train_loss": 0.00023524776042904705,
          "validation_loss": 0.0002180242445319891,
          "train_metric": 2.369329013163224e-05,
          "validation_metric": 2.1958601791993715e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.934620072999678
        },
        {
          "epoch": 514,
          "updates": 8224,
          "samples_seen": 526336,
          "online_loss": 0.0002452291864756262,
          "train_loss": 0.00022991413425188512,
          "validation_loss": 0.00020871202286798507,
          "train_metric": 2.3156109818955883e-05,
          "validation_metric": 2.1020710846642032e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.946730972000296
        },
        {
          "epoch": 515,
          "updates": 8240,
          "samples_seen": 527360,
          "online_loss": 0.00023331855027208803,
          "train_loss": 0.00022717221872881055,
          "validation_loss": 0.00021310763258952647,
          "train_metric": 2.287995084770955e-05,
          "validation_metric": 2.1463421944645233e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.957268703999944
        },
        {
          "epoch": 516,
          "updates": 8256,
          "samples_seen": 528384,
          "online_loss": 0.0002295988379046321,
          "train_loss": 0.00022849004017189145,
          "validation_loss": 0.00021125358762219548,
          "train_metric": 2.3012678866507486e-05,
          "validation_metric": 2.1276688130456023e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.967682974000127
        },
        {
          "epoch": 517,
          "updates": 8272,
          "samples_seen": 529408,
          "online_loss": 0.00023123232040234143,
          "train_loss": 0.00022784611792303622,
          "validation_loss": 0.0002124350139638409,
          "train_metric": 2.294782461831346e-05,
          "validation_metric": 2.139567550329957e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.9781860259990935
        },
        {
          "epoch": 518,
          "updates": 8288,
          "samples_seen": 530432,
          "online_loss": 0.00023086742930900073,
          "train_loss": 0.00022643382544629276,
          "validation_loss": 0.00020851544104516506,
          "train_metric": 2.2805585103924386e-05,
          "validation_metric": 2.1000909327995032e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.989188137000383
        },
        {
          "epoch": 519,
          "updates": 8304,
          "samples_seen": 531456,
          "online_loss": 0.00023099630652723135,
          "train_loss": 0.00022885799990035594,
          "validation_loss": 0.00020903769473079592,
          "train_metric": 2.3049742594594136e-05,
          "validation_metric": 2.105351086356677e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.999649105000572
        },
        {
          "epoch": 520,
          "updates": 8320,
          "samples_seen": 532480,
          "online_loss": 0.00023098153906175867,
          "train_loss": 0.00022715155500918627,
          "validation_loss": 0.00020862935343757272,
          "train_metric": 2.2877873561810702e-05,
          "validation_metric": 2.1012385332142003e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.010081280999657
        },
        {
          "epoch": 521,
          "updates": 8336,
          "samples_seen": 533504,
          "online_loss": 0.0002331855675947736,
          "train_loss": 0.00022657244699075818,
          "validation_loss": 0.00021095058764331043,
          "train_metric": 2.2819544028607197e-05,
          "validation_metric": 2.124616912624333e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.020321706999312
        },
        {
          "epoch": 522,
          "updates": 8352,
          "samples_seen": 534528,
          "online_loss": 0.00022971414182393346,
          "train_loss": 0.0002270995610160753,
          "validation_loss": 0.00020981641137041152,
          "train_metric": 2.287263487232849e-05,
          "validation_metric": 2.113194204866886e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.030701517000125
        },
        {
          "epoch": 523,
          "updates": 8368,
          "samples_seen": 535552,
          "online_loss": 0.0002315737037861254,
          "train_loss": 0.00022850645473226905,
          "validation_loss": 0.00021102720347698778,
          "train_metric": 2.301433232787531e-05,
          "validation_metric": 2.1253887098282576e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.041218956999728
        },
        {
          "epoch": 524,
          "updates": 8384,
          "samples_seen": 536576,
          "online_loss": 0.00023359904389508301,
          "train_loss": 0.0002273638965561986,
          "validation_loss": 0.00020679517183452845,
          "train_metric": 2.289925942022819e-05,
          "validation_metric": 2.0827652406296693e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.052430943000218
        },
        {
          "epoch": 525,
          "updates": 8400,
          "samples_seen": 537600,
          "online_loss": 0.00023197216887638206,
          "train_loss": 0.00022554321913048625,
          "validation_loss": 0.00020731217227876186,
          "train_metric": 2.2715888917446136e-05,
          "validation_metric": 2.0879726434941404e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.063112656000158
        },
        {
          "epoch": 526,
          "updates": 8416,
          "samples_seen": 538624,
          "online_loss": 0.00022922535572433844,
          "train_loss": 0.00022480118786916137,
          "validation_loss": 0.0002102326980093494,
          "train_metric": 2.264115391881205e-05,
          "validation_metric": 2.1173866116441786e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.07361382199997
        },
        {
          "epoch": 527,
          "updates": 8432,
          "samples_seen": 539648,
          "online_loss": 0.00023007799609331414,
          "train_loss": 0.0002251287951366976,
          "validation_loss": 0.0002065918961307034,
          "train_metric": 2.267414856760297e-05,
          "validation_metric": 2.0807181499549188e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.08454456400068
        },
        {
          "epoch": 528,
          "updates": 8448,
          "samples_seen": 540672,
          "online_loss": 0.00024439917615382,
          "train_loss": 0.00024038837000261992,
          "validation_loss": 0.0002205943310400471,
          "train_metric": 2.4211038180510513e-05,
          "validation_metric": 2.2217451260075904e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.095108506999168
        },
        {
          "epoch": 529,
          "updates": 8464,
          "samples_seen": 541696,
          "online_loss": 0.00023642989617655985,
          "train_loss": 0.00023284924100153148,
          "validation_loss": 0.00021337790531106293,
          "train_metric": 2.3451721062883735e-05,
          "validation_metric": 2.1490643121069297e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.105360723000558
        },
        {
          "epoch": 530,
          "updates": 8480,
          "samples_seen": 542720,
          "online_loss": 0.00023438974767486798,
          "train_loss": 0.00022408782388083637,
          "validation_loss": 0.00020519312238320708,
          "train_metric": 2.2569305656361394e-05,
          "validation_metric": 2.0666298951255158e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.116478452000592
        },
        {
          "epoch": 531,
          "updates": 8496,
          "samples_seen": 543744,
          "online_loss": 0.00022934681783226551,
          "train_loss": 0.0002252784906886518,
          "validation_loss": 0.0002133647503796965,
          "train_metric": 2.2689222532790154e-05,
          "validation_metric": 2.148931525880471e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.126908013999127
        },
        {
          "epoch": 532,
          "updates": 8512,
          "samples_seen": 544768,
          "online_loss": 0.00022741146040061722,
          "train_loss": 0.00022422811889555305,
          "validation_loss": 0.00020773590949829668,
          "train_metric": 2.2583433747058734e-05,
          "validation_metric": 2.0922401745337993e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.1374881510000705
        },
        {
          "epoch": 533,
          "updates": 8528,
          "samples_seen": 545792,
          "online_loss": 0.0002286408371219295,
          "train_loss": 0.00022345561592373997,
          "validation_loss": 0.00020688280346803367,
          "train_metric": 2.2505630113300867e-05,
          "validation_metric": 2.0836476323893294e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.147912411999641
        },
        {
          "epoch": 534,
          "updates": 8544,
          "samples_seen": 546816,
          "online_loss": 0.00023099386407920974,
          "train_loss": 0.00022350961808115244,
          "validation_loss": 0.00021127602667547762,
          "train_metric": 2.2511070710606873e-05,
          "validation_metric": 2.1278945496305823e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.158501518000776
        },
        {
          "epoch": 535,
          "updates": 8560,
          "samples_seen": 547840,
          "online_loss": 0.00022649440597888315,
          "train_loss": 0.0002227000513812527,
          "validation_loss": 0.00020623406453523785,
          "train_metric": 2.242953451059293e-05,
          "validation_metric": 2.0771138224517927e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.169139379000626
        },
        {
          "epoch": 536,
          "updates": 8576,
          "samples_seen": 548864,
          "online_loss": 0.00022820549202151597,
          "train_loss": 0.0002232138649560511,
          "validation_loss": 0.0002075588854495436,
          "train_metric": 2.2481286578113213e-05,
          "validation_metric": 2.0904572011204436e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.179638298999635
        },
        {
          "epoch": 537,
          "updates": 8592,
          "samples_seen": 549888,
          "online_loss": 0.00024161817964341026,
          "train_loss": 0.00023170049826148897,
          "validation_loss": 0.00020961354312021285,
          "train_metric": 2.3336026060860604e-05,
          "validation_metric": 2.111150934069883e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.190095908999865
        },
        {
          "epoch": 538,
          "updates": 8608,
          "samples_seen": 550912,
          "online_loss": 0.00023086456258170074,
          "train_loss": 0.00022770196665078402,
          "validation_loss": 0.00021613991702906787,
          "train_metric": 2.2933309082873166e-05,
          "validation_metric": 2.1768821170553565e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.200738146000731
        },
        {
          "epoch": 539,
          "updates": 8624,
          "samples_seen": 551936,
          "online_loss": 0.0002315025030839024,
          "train_loss": 0.0002227568911621347,
          "validation_loss": 0.00020558711548801512,
          "train_metric": 2.243525887024589e-05,
          "validation_metric": 2.0705983843072318e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.211275084999215
        },
        {
          "epoch": 540,
          "updates": 8640,
          "samples_seen": 552960,
          "online_loss": 0.00022943700287214597,
          "train_loss": 0.0002260658220620826,
          "validation_loss": 0.00021011930948588997,
          "train_metric": 2.276852319482714e-05,
          "validation_metric": 2.1162446500966325e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.22171441700084
        },
        {
          "epoch": 541,
          "updates": 8656,
          "samples_seen": 553984,
          "online_loss": 0.00022597804127144627,
          "train_loss": 0.00022655718203168362,
          "validation_loss": 0.0002155129477614537,
          "train_metric": 2.2818008801550604e-05,
          "validation_metric": 2.1705676772398874e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.2324904649995005
        },
        {
          "epoch": 542,
          "updates": 8672,
          "samples_seen": 555008,
          "online_loss": 0.00023540533857158152,
          "train_loss": 0.00022199172235559672,
          "validation_loss": 0.00020326243247836828,
          "train_metric": 2.2358195565175265e-05,
          "validation_metric": 2.047185080300551e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.243877225999313
        },
        {
          "epoch": 543,
          "updates": 8688,
          "samples_seen": 556032,
          "online_loss": 0.00022683641327603254,
          "train_loss": 0.00022422903566621244,
          "validation_loss": 0.0002077413082588464,
          "train_metric": 2.258352833450772e-05,
          "validation_metric": 2.0922941985190846e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.254448568999578
        },
        {
          "epoch": 544,
          "updates": 8704,
          "samples_seen": 557056,
          "online_loss": 0.0002236938635178376,
          "train_loss": 0.00022213862393982708,
          "validation_loss": 0.0002055538643617183,
          "train_metric": 2.2372991224983707e-05,
          "validation_metric": 2.070263508358039e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.26502406399959
        },
        {
          "epoch": 545,
          "updates": 8720,
          "samples_seen": 558080,
          "online_loss": 0.00022251989503274672,
          "train_loss": 0.00022086701937951148,
          "validation_loss": 0.0002082511637127027,
          "train_metric": 2.2244916181080043e-05,
          "validation_metric": 2.097429387504235e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.275388240999746
        },
        {
          "epoch": 546,
          "updates": 8736,
          "samples_seen": 559104,
          "online_loss": 0.00022721250979884644,
          "train_loss": 0.00022293911024462432,
          "validation_loss": 0.00020718935411423445,
          "train_metric": 2.2453612473327667e-05,
          "validation_metric": 2.0867355488007888e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.285884667999198
        },
        {
          "epoch": 547,
          "updates": 8752,
          "samples_seen": 560128,
          "online_loss": 0.0002287342504132539,
          "train_loss": 0.00022194848861545324,
          "validation_loss": 0.00020752254931721836,
          "train_metric": 2.2353835447574966e-05,
          "validation_metric": 2.0900912204524502e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.296224329000324
        },
        {
          "epoch": 548,
          "updates": 8768,
          "samples_seen": 561152,
          "online_loss": 0.00022548519154952373,
          "train_loss": 0.00022349377104546875,
          "validation_loss": 0.00020905247947666794,
          "train_metric": 2.250947363791056e-05,
          "validation_metric": 2.105499879689887e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.306590463000248
        },
        {
          "epoch": 549,
          "updates": 8784,
          "samples_seen": 562176,
          "online_loss": 0.00022630008152191294,
          "train_loss": 0.00022778604761697352,
          "validation_loss": 0.00021231372375041246,
          "train_metric": 2.294177647854667e-05,
          "validation_metric": 2.138346098945476e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.369852968000487
        },
        {
          "epoch": 550,
          "updates": 8800,
          "samples_seen": 563200,
          "online_loss": 0.00022167702900333097,
          "train_loss": 0.00022273110516835004,
          "validation_loss": 0.00020505243446677923,
          "train_metric": 2.2432661353377625e-05,
          "validation_metric": 2.065213084279094e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.3806993609996425
        },
        {
          "epoch": 551,
          "updates": 8816,
          "samples_seen": 564224,
          "online_loss": 0.00024257876611954998,
          "train_loss": 0.00023152359062805772,
          "validation_loss": 0.0002139390999218449,
          "train_metric": 2.3318210878642276e-05,
          "validation_metric": 2.1547162759816274e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.391246314000455
        },
        {
          "epoch": 552,
          "updates": 8832,
          "samples_seen": 565248,
          "online_loss": 0.00022798625877840095,
          "train_loss": 0.0002194834523834288,
          "validation_loss": 0.0002024822897510603,
          "train_metric": 2.2105568859842606e-05,
          "validation_metric": 2.039327227976173e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.402163082000698
        },
        {
          "epoch": 553,
          "updates": 8848,
          "samples_seen": 566272,
          "online_loss": 0.00022231549610296497,
          "train_loss": 0.00021958215802442282,
          "validation_loss": 0.00020965709700249135,
          "train_metric": 2.2115511455922388e-05,
          "validation_metric": 2.1115896743140183e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.412879131999944
        },
        {
          "epoch": 554,
          "updates": 8864,
          "samples_seen": 567296,
          "online_loss": 0.00022241489477892173,
          "train_loss": 0.00021878021652810276,
          "validation_loss": 0.00020599138224497437,
          "train_metric": 2.203474286943674e-05,
          "validation_metric": 2.0746698282891884e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.423276309000357
        },
        {
          "epoch": 555,
          "updates": 8880,
          "samples_seen": 568320,
          "online_loss": 0.00022424118378694402,
          "train_loss": 0.00022039428586140275,
          "validation_loss": 0.00020252136164344847,
          "train_metric": 2.219730959041044e-05,
          "validation_metric": 2.0397208572831005e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.433542007000142
        },
        {
          "epoch": 556,
          "updates": 8896,
          "samples_seen": 569344,
          "online_loss": 0.00022258096032601316,
          "train_loss": 0.00021855896920897067,
          "validation_loss": 0.0002053679636446759,
          "train_metric": 2.2012460249243304e-05,
          "validation_metric": 2.0683908587670885e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.44427255499977
        },
        {
          "epoch": 557,
          "updates": 8912,
          "samples_seen": 570368,
          "online_loss": 0.000228251287353487,
          "train_loss": 0.00022228475427255034,
          "validation_loss": 0.00020546428277157247,
          "train_metric": 2.2387708668247797e-05,
          "validation_metric": 2.069360743917059e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.454631724000137
        },
        {
          "epoch": 558,
          "updates": 8928,
          "samples_seen": 571392,
          "online_loss": 0.00022309924224828137,
          "train_loss": 0.00021790160099044442,
          "validation_loss": 0.00020469063019845635,
          "train_metric": 2.1946252672933042e-05,
          "validation_metric": 2.06156892090803e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.464966306999486
        },
        {
          "epoch": 559,
          "updates": 8944,
          "samples_seen": 572416,
          "online_loss": 0.0002232149317933363,
          "train_loss": 0.00021791673498228192,
          "validation_loss": 0.0002064180443994701,
          "train_metric": 2.1947776986053213e-05,
          "validation_metric": 2.0789671907550655e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.475315595000211
        },
        {
          "epoch": 560,
          "updates": 8960,
          "samples_seen": 573440,
          "online_loss": 0.0002212697791037499,
          "train_loss": 0.0002176917187171057,
          "validation_loss": 0.00020139082334935665,
          "train_metric": 2.1925108740106225e-05,
          "validation_metric": 2.0283345293137245e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.4866242759999295
        },
        {
          "epoch": 561,
          "updates": 8976,
          "samples_seen": 574464,
          "online_loss": 0.00022269063993007876,
          "train_loss": 0.0002187686914112419,
          "validation_loss": 0.00020524623687379062,
          "train_metric": 2.2033580535207875e-05,
          "validation_metric": 2.0671648599090986e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.4973037090003345
        },
        {
          "epoch": 562,
          "updates": 8992,
          "samples_seen": 575488,
          "online_loss": 0.00022092193648859393,
          "train_loss": 0.00021864469454158098,
          "validation_loss": 0.00020208675414323807,
          "train_metric": 2.202109135396313e-05,
          "validation_metric": 2.0353440049802884e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.507863685999837
        },
        {
          "epoch": 563,
          "updates": 9008,
          "samples_seen": 576512,
          "online_loss": 0.00022219910306375823,
          "train_loss": 0.00022042894852347672,
          "validation_loss": 0.00020700063032563776,
          "train_metric": 2.220079659309704e-05,
          "validation_metric": 2.084834522975143e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.5182563079997635
        },
        {
          "epoch": 564,
          "updates": 9024,
          "samples_seen": 577536,
          "online_loss": 0.00022528965473611606,
          "train_loss": 0.0002210839302279055,
          "validation_loss": 0.0002074540825560689,
          "train_metric": 2.2266764062806033e-05,
          "validation_metric": 2.0894014596706256e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.5285884810000425
        },
        {
          "epoch": 565,
          "updates": 9040,
          "samples_seen": 578560,
          "online_loss": 0.00022145724778965814,
          "train_loss": 0.00021650223061442375,
          "validation_loss": 0.00020288994710426778,
          "train_metric": 2.18053119169781e-05,
          "validation_metric": 2.0434332327567972e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.538890458999958
        },
        {
          "epoch": 566,
          "updates": 9056,
          "samples_seen": 579584,
          "online_loss": 0.00022066616747906664,
          "train_loss": 0.0002166076737921685,
          "validation_loss": 0.00020292542467359453,
          "train_metric": 2.1815929358126596e-05,
          "validation_metric": 2.0437908460735343e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.549326975999065
        },
        {
          "epoch": 567,
          "updates": 9072,
          "samples_seen": 580608,
          "online_loss": 0.00022545907813764643,
          "train_loss": 0.00022129106218926609,
          "validation_loss": 0.00020797905744984746,
          "train_metric": 2.2287624233285896e-05,
          "validation_metric": 2.094689079967793e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.55961913200008
        },
        {
          "epoch": 568,
          "updates": 9088,
          "samples_seen": 581632,
          "online_loss": 0.0002209957365266746,
          "train_loss": 0.00022485270164906979,
          "validation_loss": 0.00020556795061565936,
          "train_metric": 2.2646341676590964e-05,
          "validation_metric": 2.0704050257336348e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.56988491499942
        },
        {
          "epoch": 569,
          "updates": 9104,
          "samples_seen": 582656,
          "online_loss": 0.0002249714671052061,
          "train_loss": 0.0002158237184630707,
          "validation_loss": 0.00020481969113461673,
          "train_metric": 2.1736974304076284e-05,
          "validation_metric": 2.0628689526347443e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.580366246000267
        },
        {
          "epoch": 570,
          "updates": 9120,
          "samples_seen": 583680,
          "online_loss": 0.00022436923427449074,
          "train_loss": 0.00021855480736121535,
          "validation_loss": 0.00019878115563187748,
          "train_metric": 2.2012036424712278e-05,
          "validation_metric": 2.0020510419271886e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.591247553000358
        },
        {
          "epoch": 571,
          "updates": 9136,
          "samples_seen": 584704,
          "online_loss": 0.0002222157800133573,
          "train_loss": 0.00021896605903748423,
          "validation_loss": 0.00021154248679522425,
          "train_metric": 2.2053458451409824e-05,
          "validation_metric": 2.130578286596574e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.601674585999717
        },
        {
          "epoch": 572,
          "updates": 9152,
          "samples_seen": 585728,
          "online_loss": 0.00021947515870124334,
          "train_loss": 0.00021590334654320031,
          "validation_loss": 0.00020122250134591013,
          "train_metric": 2.1744992409367114e-05,
          "validation_metric": 2.02663941308856e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.6124126840004465
        },
        {
          "epoch": 573,
          "updates": 9168,
          "samples_seen": 586752,
          "online_loss": 0.00022360592538461788,
          "train_loss": 0.00021485456090886146,
          "validation_loss": 0.00020183117885608226,
          "train_metric": 2.1639363694703206e-05,
          "validation_metric": 2.03276958927745e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.622683642999618
        },
        {
          "epoch": 574,
          "updates": 9184,
          "samples_seen": 587776,
          "online_loss": 0.00022452507619163953,
          "train_loss": 0.00021455447131302208,
          "validation_loss": 0.00019964331295341253,
          "train_metric": 2.1609143004752696e-05,
          "validation_metric": 2.0107347154407762e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.633049858999584
        },
        {
          "epoch": 575,
          "updates": 9200,
          "samples_seen": 588800,
          "online_loss": 0.00022833094317320501,
          "train_loss": 0.00021446440950967371,
          "validation_loss": 0.00020373292500153184,
          "train_metric": 2.160006988560781e-05,
          "validation_metric": 2.051923183898907e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.643362409999099
        },
        {
          "epoch": 576,
          "updates": 9216,
          "samples_seen": 589824,
          "online_loss": 0.00022028651164873736,
          "train_loss": 0.0002167651691706851,
          "validation_loss": 0.00020688785298261791,
          "train_metric": 2.1831794583704323e-05,
          "validation_metric": 2.083698200294748e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.653722111999741
        },
        {
          "epoch": 577,
          "updates": 9232,
          "samples_seen": 590848,
          "online_loss": 0.00022383227178579546,
          "train_loss": 0.00022411371173802763,
          "validation_loss": 0.00020745278743561357,
          "train_metric": 2.2571912268176675e-05,
          "validation_metric": 2.0893885448458605e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.663961746999121
        },
        {
          "epoch": 578,
          "updates": 9248,
          "samples_seen": 591872,
          "online_loss": 0.00023615153986611404,
          "train_loss": 0.00023451984452549368,
          "validation_loss": 0.00021435697271954268,
          "train_metric": 2.3619983039679937e-05,
          "validation_metric": 2.1589250536635518e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.67436492699926
        },
        {
          "epoch": 579,
          "updates": 9264,
          "samples_seen": 592896,
          "online_loss": 0.00022056183206586866,
          "train_loss": 0.00022638225345872343,
          "validation_loss": 0.00020689447410404682,
          "train_metric": 2.2800393708166666e-05,
          "validation_metric": 2.0837653210037388e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.684702475000449
        },
        {
          "epoch": 580,
          "updates": 9280,
          "samples_seen": 593920,
          "online_loss": 0.0002159655105060665,
          "train_loss": 0.00021962709433864802,
          "validation_loss": 0.00021131138782948256,
          "train_metric": 2.2120035282569006e-05,
          "validation_metric": 2.1282507077557966e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.6951179890002095
        },
        {
          "epoch": 581,
          "updates": 9296,
          "samples_seen": 594944,
          "online_loss": 0.00021996686791680986,
          "train_loss": 0.00021967686188872904,
          "validation_loss": 0.00020360600319691002,
          "train_metric": 2.212504841736518e-05,
          "validation_metric": 2.050645161943976e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.705332524999903
        },
        {
          "epoch": 582,
          "updates": 9312,
          "samples_seen": 595968,
          "online_loss": 0.00022395726136892335,
          "train_loss": 0.00021303312678355724,
          "validation_loss": 0.00019838070147670805,
          "train_metric": 2.14559149753768e-05,
          "validation_metric": 1.998017614823766e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.716420631000801
        },
        {
          "epoch": 583,
          "updates": 9328,
          "samples_seen": 596992,
          "online_loss": 0.00021654546162608312,
          "train_loss": 0.00021263858070597053,
          "validation_loss": 0.00019965379033237696,
          "train_metric": 2.1416180970845744e-05,
          "validation_metric": 2.0108400349272415e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.726797248000366
        },
        {
          "epoch": 584,
          "updates": 9344,
          "samples_seen": 598016,
          "online_loss": 0.00021557659420068376,
          "train_loss": 0.00021245940297376364,
          "validation_loss": 0.00020091420446988195,
          "train_metric": 2.1398134776973166e-05,
          "validation_metric": 2.0235340343788266e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.73708639600045
        },
        {
          "epoch": 585,
          "updates": 9360,
          "samples_seen": 599040,
          "online_loss": 0.00021552663474722067,
          "train_loss": 0.00021284265676513314,
          "validation_loss": 0.00020164737361483276,
          "train_metric": 2.14367300941376e-05,
          "validation_metric": 2.0309184037614614e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.8022074199998315
        },
        {
          "epoch": 586,
          "updates": 9376,
          "samples_seen": 600064,
          "online_loss": 0.00021566920349869179,
          "train_loss": 0.00021198936155997217,
          "validation_loss": 0.0002007188886636868,
          "train_metric": 2.1350790120777674e-05,
          "validation_metric": 2.021567343035713e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.813255207000111
        },
        {
          "epoch": 587,
          "updates": 9392,
          "samples_seen": 601088,
          "online_loss": 0.00021592319990304532,
          "train_loss": 0.00021449911582749337,
          "validation_loss": 0.000202338385861367,
          "train_metric": 2.1603567802230828e-05,
          "validation_metric": 2.0378782210173085e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.823588392999227
        },
        {
          "epoch": 588,
          "updates": 9408,
          "samples_seen": 602112,
          "online_loss": 0.00021611878764815629,
          "train_loss": 0.00021382384875323623,
          "validation_loss": 0.00019670107576530427,
          "train_metric": 2.1535559426411055e-05,
          "validation_metric": 1.98110119526973e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.834483877000821
        },
        {
          "epoch": 589,
          "updates": 9424,
          "samples_seen": 603136,
          "online_loss": 0.00022174400464791688,
          "train_loss": 0.00021525911870412529,
          "validation_loss": 0.00020750534895341843,
          "train_metric": 2.1680110876332037e-05,
          "validation_metric": 2.0899178707622923e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.845349555000212
        },
        {
          "epoch": 590,
          "updates": 9440,
          "samples_seen": 604160,
          "online_loss": 0.0002175085901399143,
          "train_loss": 0.00021434217342175543,
          "validation_loss": 0.00019950488058384508,
          "train_metric": 2.158775896532461e-05,
          "validation_metric": 2.0093400962650776e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.855754514999717
        },
        {
          "epoch": 591,
          "updates": 9456,
          "samples_seen": 605184,
          "online_loss": 0.00021718900825362653,
          "train_loss": 0.00021124460909049958,
          "validation_loss": 0.00020163327280897647,
          "train_metric": 2.127578409272246e-05,
          "validation_metric": 2.0307765225879848e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.866015976000199
        },
        {
          "epoch": 592,
          "updates": 9472,
          "samples_seen": 606208,
          "online_loss": 0.00022344023318510153,
          "train_loss": 0.00021814543288201094,
          "validation_loss": 0.00020629604114219546,
          "train_metric": 2.1970807210891508e-05,
          "validation_metric": 2.0777380996150896e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.876477959000113
        },
        {
          "epoch": 593,
          "updates": 9488,
          "samples_seen": 607232,
          "online_loss": 0.00023262546801561257,
          "train_loss": 0.000214391402550973,
          "validation_loss": 0.0002004406414926052,
          "train_metric": 2.1592717530438676e-05,
          "validation_metric": 2.0187646441627294e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.886708276000718
        },
        {
          "epoch": 594,
          "updates": 9504,
          "samples_seen": 608256,
          "online_loss": 0.0002176347325075767,
          "train_loss": 0.00021113290858920664,
          "validation_loss": 0.00020281241449993104,
          "train_metric": 2.1264531824272126e-05,
          "validation_metric": 2.042652340605855e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.897096606000559
        },
        {
          "epoch": 595,
          "updates": 9520,
          "samples_seen": 609280,
          "online_loss": 0.00021630244282278,
          "train_loss": 0.0002103425213135779,
          "validation_loss": 0.00019940061611123383,
          "train_metric": 2.118492739100475e-05,
          "validation_metric": 2.0082899936824106e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.907367943000281
        },
        {
          "epoch": 596,
          "updates": 9536,
          "samples_seen": 610304,
          "online_loss": 0.00022342030933941714,
          "train_loss": 0.00021590714459307492,
          "validation_loss": 0.00020035530906170607,
          "train_metric": 2.174537439714186e-05,
          "validation_metric": 2.017905171669554e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.918095487000755
        },
        {
          "epoch": 597,
          "updates": 9552,
          "samples_seen": 611328,
          "online_loss": 0.0002162690016120905,
          "train_loss": 0.00021001201821491122,
          "validation_loss": 0.00020008425053674728,
          "train_metric": 2.1151641703909263e-05,
          "validation_metric": 2.0151752323727123e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.928656564999983
        },
        {
          "epoch": 598,
          "updates": 9568,
          "samples_seen": 612352,
          "online_loss": 0.00021906851543462835,
          "train_loss": 0.0002101954014506191,
          "validation_loss": 0.000195107699255459,
          "train_metric": 2.117011354130227e-05,
          "validation_metric": 1.9650531612569466e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.939674993000153
        },
        {
          "epoch": 599,
          "updates": 9584,
          "samples_seen": 613376,
          "online_loss": 0.0002155480997316772,
          "train_loss": 0.00021549759549088776,
          "validation_loss": 0.00020690940436907113,
          "train_metric": 2.170412517443765e-05,
          "validation_metric": 2.0839159333263524e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.95017721899967
        },
        {
          "epoch": 600,
          "updates": 9600,
          "samples_seen": 614400,
          "online_loss": 0.00021160183132451493,
          "train_loss": 0.0002096555253956467,
          "validation_loss": 0.00019603998225647956,
          "train_metric": 2.1115738491062075e-05,
          "validation_metric": 1.974443148355931e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.960666835999291
        },
        {
          "epoch": 601,
          "updates": 9616,
          "samples_seen": 615424,
          "online_loss": 0.00021372365335992072,
          "train_loss": 0.00021013992954976857,
          "validation_loss": 0.00020115308871027082,
          "train_metric": 2.116452196787577e-05,
          "validation_metric": 2.025940193561837e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.970906250000553
        },
        {
          "epoch": 602,
          "updates": 9632,
          "samples_seen": 616448,
          "online_loss": 0.00021390489109762711,
          "train_loss": 0.00020905140263494104,
          "validation_loss": 0.00019476082525216043,
          "train_metric": 2.1054893295513466e-05,
          "validation_metric": 1.9615597921074368e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.982100155999433
        },
        {
          "epoch": 603,
          "updates": 9648,
          "samples_seen": 617472,
          "online_loss": 0.00021265736040732008,
          "train_loss": 0.00021140584431122988,
          "validation_loss": 0.00020174554083496332,
          "train_metric": 2.1292024030117318e-05,
          "validation_metric": 2.031907206401229e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.992720186999577
        },
        {
          "epoch": 604,
          "updates": 9664,
          "samples_seen": 618496,
          "online_loss": 0.00022262344282353297,
          "train_loss": 0.0002084403531625867,
          "validation_loss": 0.00019610821618698537,
          "train_metric": 2.0993349608033895e-05,
          "validation_metric": 1.9751301806536503e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.00315556500027
        },
        {
          "epoch": 605,
          "updates": 9680,
          "samples_seen": 619520,
          "online_loss": 0.00021412935757325613,
          "train_loss": 0.00020834877796005458,
          "validation_loss": 0.00019740991410799325,
          "train_metric": 2.0984127331757918e-05,
          "validation_metric": 1.9882401829818264e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.0136650999993435
        },
        {
          "epoch": 606,
          "updates": 9696,
          "samples_seen": 620544,
          "online_loss": 0.00021424275018944172,
          "train_loss": 0.0002082789142150432,
          "validation_loss": 0.00019743751909118146,
          "train_metric": 2.0977091480745003e-05,
          "validation_metric": 1.9885186702595092e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.024046217000432
        },
        {
          "epoch": 607,
          "updates": 9712,
          "samples_seen": 621568,
          "online_loss": 0.0002128615642504883,
          "train_loss": 0.0002089588379021734,
          "validation_loss": 0.00019829283701255918,
          "train_metric": 2.1045567336841486e-05,
          "validation_metric": 1.9971330402768217e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.03425259100004
        },
        {
          "epoch": 608,
          "updates": 9728,
          "samples_seen": 622592,
          "online_loss": 0.00021916880632488756,
          "train_loss": 0.00021210969134699553,
          "validation_loss": 0.00019943257211707532,
          "train_metric": 2.1362908228184097e-05,
          "validation_metric": 2.0086121367057785e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.044765455999368
        },
        {
          "epoch": 609,
          "updates": 9744,
          "samples_seen": 623616,
          "online_loss": 0.00022139563498058124,
          "train_loss": 0.0002076963719446212,
          "validation_loss": 0.0001970895827980712,
          "train_metric": 2.0918416339554824e-05,
          "validation_metric": 1.9850142052746378e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.055025267000019
        },
        {
          "epoch": 610,
          "updates": 9760,
          "samples_seen": 624640,
          "online_loss": 0.00022044798333809013,
          "train_loss": 0.00021163097699172795,
          "validation_loss": 0.00020159143605269492,
          "train_metric": 2.131469955202192e-05,
          "validation_metric": 2.0303552446421236e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.065349791999324
        },
        {
          "epoch": 611,
          "updates": 9776,
          "samples_seen": 625664,
          "online_loss": 0.00021710503824579064,
          "train_loss": 0.0002084254374494776,
          "validation_loss": 0.00019384874030947685,
          "train_metric": 2.0991847122786567e-05,
          "validation_metric": 1.952373349922709e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.076178150999112
        },
        {
          "epoch": 612,
          "updates": 9792,
          "samples_seen": 626688,
          "online_loss": 0.00021235502208583057,
          "train_loss": 0.0002125677710864693,
          "validation_loss": 0.0002032410557148978,
          "train_metric": 2.140904871339444e-05,
          "validation_metric": 2.0469693481572904e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.086768691000543
        },
        {
          "epoch": 613,
          "updates": 9808,
          "samples_seen": 627712,
          "online_loss": 0.00021585676768154372,
          "train_loss": 0.00020736279839184135,
          "validation_loss": 0.0001971286692423746,
          "train_metric": 2.088482324325014e-05,
          "validation_metric": 1.985407834581565e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.09717324399935
        },
        {
          "epoch": 614,
          "updates": 9824,
          "samples_seen": 628736,
          "online_loss": 0.0002102493403981498,
          "train_loss": 0.00020717000006698072,
          "validation_loss": 0.00019553344463929534,
          "train_metric": 2.086540189338848e-05,
          "validation_metric": 1.9693412468768656e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.107816215999264
        },
        {
          "epoch": 615,
          "updates": 9840,
          "samples_seen": 629760,
          "online_loss": 0.0002106421143253101,
          "train_loss": 0.00020621035946533084,
          "validation_loss": 0.00019626092398539186,
          "train_metric": 2.0768751710420474e-05,
          "validation_metric": 1.976667954295408e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.118422252999153
        },
        {
          "epoch": 616,
          "updates": 9856,
          "samples_seen": 630784,
          "online_loss": 0.000213593651096744,
          "train_loss": 0.00020840355136897415,
          "validation_loss": 0.0001954051258508116,
          "train_metric": 2.0989644326618873e-05,
          "validation_metric": 1.9680488549056463e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.128795336000621
        },
        {
          "epoch": 617,
          "updates": 9872,
          "samples_seen": 631808,
          "online_loss": 0.00022370687474904116,
          "train_loss": 0.00021426769671961665,
          "validation_loss": 0.00020350819977466017,
          "train_metric": 2.1580255634034984e-05,
          "validation_metric": 2.049659815384075e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.139057682999919
        },
        {
          "epoch": 618,
          "updates": 9888,
          "samples_seen": 632832,
          "online_loss": 0.00021799770456709666,
          "train_loss": 0.0002056584198726341,
          "validation_loss": 0.0001947115088114515,
          "train_metric": 2.0713163394248113e-05,
          "validation_metric": 1.9610630261013284e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.1497510409990355
        },
        {
          "epoch": 619,
          "updates": 9904,
          "samples_seen": 633856,
          "online_loss": 0.00020985473747714423,
          "train_loss": 0.0002054746146313846,
          "validation_loss": 0.00019628820882644504,
          "train_metric": 2.0694651539088227e-05,
          "validation_metric": 1.9769428035942838e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.160036335000768
        },
        {
          "epoch": 620,
          "updates": 9920,
          "samples_seen": 634880,
          "online_loss": 0.00021084542095195502,
          "train_loss": 0.0002052763884421438,
          "validation_loss": 0.0001954495965037495,
          "train_metric": 2.067468813038431e-05,
          "validation_metric": 1.9684966900967993e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.170432200000505
        },
        {
          "epoch": 621,
          "updates": 9936,
          "samples_seen": 635904,
          "online_loss": 0.00020752628824993735,
          "train_loss": 0.000205091098905541,
          "validation_loss": 0.00019505934324115515,
          "train_metric": 2.0656023480114527e-05,
          "validation_metric": 1.9645662177936174e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.235230046000652
        },
        {
          "epoch": 622,
          "updates": 9952,
          "samples_seen": 636928,
          "online_loss": 0.00021128500247868942,
          "train_loss": 0.000207969278562814,
          "validation_loss": 0.00019831876852549613,
          "train_metric": 2.094590490742121e-05,
          "validation_metric": 1.9973937014583498e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.245981523999944
        },
        {
          "epoch": 623,
          "updates": 9968,
          "samples_seen": 637952,
          "online_loss": 0.00020812678030779352,
          "train_loss": 0.00020619345013983548,
          "validation_loss": 0.00019466911908239126,
          "train_metric": 2.0767047317349352e-05,
          "validation_metric": 1.9606357454904355e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.256292804999248
        },
        {
          "epoch": 624,
          "updates": 9984,
          "samples_seen": 638976,
          "online_loss": 0.0002107072168655577,
          "train_loss": 0.0002083910658257082,
          "validation_loss": 0.00020219477300997823,
          "train_metric": 2.098838558595162e-05,
          "validation_metric": 2.0364315787446685e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.266659716999129
        },
        {
          "epoch": 625,
          "updates": 10000,
          "samples_seen": 640000,
          "online_loss": 0.00021091149255880737,
          "train_loss": 0.0002123698650393635,
          "validation_loss": 0.00020146270981058478,
          "train_metric": 2.138911440852098e-05,
          "validation_metric": 2.0290584870963357e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.277001563999875
        },
        {
          "epoch": 626,
          "updates": 10016,
          "samples_seen": 641024,
          "online_loss": 0.00021874860976822674,
          "train_loss": 0.00020603887969627976,
          "validation_loss": 0.00019856060680467635,
          "train_metric": 2.075148222502321e-05,
          "validation_metric": 1.9998298739665188e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.28729920799924
        },
        {
          "epoch": 627,
          "updates": 10032,
          "samples_seen": 642048,
          "online_loss": 0.00021113369439262897,
          "train_loss": 0.00020584311278071254,
          "validation_loss": 0.0001940116926562041,
          "train_metric": 2.0731764379888773e-05,
          "validation_metric": 1.9540149878594093e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.297674288000053
        },
        {
          "epoch": 628,
          "updates": 10048,
          "samples_seen": 643072,
          "online_loss": 0.00021302531922628987,
          "train_loss": 0.0002087680040858686,
          "validation_loss": 0.00019567395793274045,
          "train_metric": 2.1026351532782428e-05,
          "validation_metric": 1.970756784430705e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.308062632999281
        },
        {
          "epoch": 629,
          "updates": 10064,
          "samples_seen": 644096,
          "online_loss": 0.00021391169411799638,
          "train_loss": 0.0002036687801592052,
          "validation_loss": 0.0001964800903806463,
          "train_metric": 2.0512774426606484e-05,
          "validation_metric": 1.9788756617344916e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.318426666999585
        },
        {
          "epoch": 630,
          "updates": 10080,
          "samples_seen": 645120,
          "online_loss": 0.0002083407389363856,
          "train_loss": 0.00020339389448054135,
          "validation_loss": 0.0001949461584445089,
          "train_metric": 2.0485089407884516e-05,
          "validation_metric": 1.963426075235475e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.328834044999894
        },
        {
          "epoch": 631,
          "updates": 10096,
          "samples_seen": 646144,
          "online_loss": 0.00021215002561802976,
          "train_loss": 0.0002077083190670237,
          "validation_loss": 0.00019916711607947946,
          "train_metric": 2.0919622329529375e-05,
          "validation_metric": 2.0059380403836258e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.339313282000148
        }
      ],
      "best_validation_metric": 1.952373349922709e-05,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 11,
      "parameters": 2369,
      "train_samples": 1024,
      "validation_rmse_m_s": 0.004418566905595873,
      "training_profiles": [
        0,
        1,
        3,
        4,
        6,
        7,
        10,
        11
      ],
      "stopped_early": true
    },
    {
      "history": [
        {
          "epoch": 1,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.18389458395540714,
          "train_loss": 0.06916050612926483,
          "validation_loss": 0.06209269165992737,
          "train_metric": 0.006965593434870243,
          "validation_metric": 0.0062537482008337975,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.02097310599856428
        },
        {
          "epoch": 2,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.045738676795735955,
          "train_loss": 0.03489450737833977,
          "validation_loss": 0.0327359214425087,
          "train_metric": 0.0035144470166414976,
          "validation_metric": 0.0032970423344522715,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03949523099981889
        },
        {
          "epoch": 3,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.02863602078286931,
          "train_loss": 0.023731259629130363,
          "validation_loss": 0.017781881615519524,
          "train_metric": 0.002390125300735235,
          "validation_metric": 0.0017909262096509337,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05715862499891955
        },
        {
          "epoch": 4,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.03047498583327979,
          "train_loss": 0.018814608454704285,
          "validation_loss": 0.018376579508185387,
          "train_metric": 0.0018949381774291396,
          "validation_metric": 0.0018508217763155699,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07376260299861315
        },
        {
          "epoch": 5,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.018881817115470767,
          "train_loss": 0.01885741390287876,
          "validation_loss": 0.016206376254558563,
          "train_metric": 0.001899249735288322,
          "validation_metric": 0.0016322467708960176,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09160129799965944
        },
        {
          "epoch": 6,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.016196074313484132,
          "train_loss": 0.011927200481295586,
          "validation_loss": 0.011007371358573437,
          "train_metric": 0.001201263745315373,
          "validation_metric": 0.0011086221784353256,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.10946310899998934
        },
        {
          "epoch": 7,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.01251865629456006,
          "train_loss": 0.010872822254896164,
          "validation_loss": 0.011406886391341686,
          "train_metric": 0.0010950708528980613,
          "validation_metric": 0.0011488599702715874,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1263439260001178
        },
        {
          "epoch": 8,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.009969748003641143,
          "train_loss": 0.009192896075546741,
          "validation_loss": 0.009333268739283085,
          "train_metric": 0.0009258748032152653,
          "validation_metric": 0.0009400126291438937,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.14393017499969574
        },
        {
          "epoch": 9,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.010474060924025252,
          "train_loss": 0.006564396433532238,
          "validation_loss": 0.006587806157767773,
          "train_metric": 0.000661141995806247,
          "validation_metric": 0.000663499697111547,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2486123869985022
        },
        {
          "epoch": 10,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.01304644881747663,
          "train_loss": 0.023848678916692734,
          "validation_loss": 0.01683354564011097,
          "train_metric": 0.00240195170044899,
          "validation_metric": 0.0016954130260273814,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.26620208699932846
        },
        {
          "epoch": 11,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.015088012762134895,
          "train_loss": 0.017693638801574707,
          "validation_loss": 0.01363987848162651,
          "train_metric": 0.0017820383654907346,
          "validation_metric": 0.0013737586559727788,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2820141149986739
        },
        {
          "epoch": 12,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.013406075217062607,
          "train_loss": 0.01004302129149437,
          "validation_loss": 0.010694079101085663,
          "train_metric": 0.0010114964097738266,
          "validation_metric": 0.001077068387530744,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.29772903799857886
        },
        {
          "epoch": 13,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.007156433042837307,
          "train_loss": 0.006388344801962376,
          "validation_loss": 0.005151539575308561,
          "train_metric": 0.0006434107199311256,
          "validation_metric": 0.0005188441718928516,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30967994699858536
        },
        {
          "epoch": 14,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.006110931179136969,
          "train_loss": 0.005998085718601942,
          "validation_loss": 0.008657893165946007,
          "train_metric": 0.0006041052401997149,
          "validation_metric": 0.0008719912730157375,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3211942479993013
        },
        {
          "epoch": 15,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.006247513956623152,
          "train_loss": 0.01197152677923441,
          "validation_loss": 0.007577700540423393,
          "train_metric": 0.001205728272907436,
          "validation_metric": 0.0007631981279700994,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3324999439992098
        },
        {
          "epoch": 16,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.011391179024940357,
          "train_loss": 0.008418243378400803,
          "validation_loss": 0.005429280921816826,
          "train_metric": 0.0008478547097183764,
          "validation_metric": 0.0005468173185363412,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.34382212799937406
        },
        {
          "epoch": 17,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.0069479590019909665,
          "train_loss": 0.00622050603851676,
          "validation_loss": 0.004411760251969099,
          "train_metric": 0.0006265066331252456,
          "validation_metric": 0.00044433632865548134,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.35553008599890745
        },
        {
          "epoch": 18,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.015920921810902655,
          "train_loss": 0.0055329278111457825,
          "validation_loss": 0.006054013967514038,
          "train_metric": 0.0005572562222369015,
          "validation_metric": 0.000609738053753972,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.366961668998556
        },
        {
          "epoch": 19,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.006680482503725216,
          "train_loss": 0.005352706182748079,
          "validation_loss": 0.006394482683390379,
          "train_metric": 0.0005391049198806286,
          "validation_metric": 0.0006440288852900267,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.37813198499861755
        },
        {
          "epoch": 20,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.006032157325535081,
          "train_loss": 0.00705934502184391,
          "validation_loss": 0.006349105853587389,
          "train_metric": 0.000710991327650845,
          "validation_metric": 0.0006394586525857449,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3893320609986404
        },
        {
          "epoch": 21,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.011214781610760838,
          "train_loss": 0.010435477830469608,
          "validation_loss": 0.007207105867564678,
          "train_metric": 0.0010510231368243694,
          "validation_metric": 0.0007258732803165913,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4011310389996652
        },
        {
          "epoch": 22,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.006948110181838274,
          "train_loss": 0.0041479915380477905,
          "validation_loss": 0.004911016672849655,
          "train_metric": 0.0004177704977337271,
          "validation_metric": 0.0004946195986121893,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4129084589985723
        },
        {
          "epoch": 23,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.005094576001283713,
          "train_loss": 0.0054644146002829075,
          "validation_loss": 0.007467461749911308,
          "train_metric": 0.0005503557622432709,
          "validation_metric": 0.0007520954241044819,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4246364619993983
        },
        {
          "epoch": 24,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.005976622007437982,
          "train_loss": 0.006017579697072506,
          "validation_loss": 0.004652527626603842,
          "train_metric": 0.0006060685263946652,
          "validation_metric": 0.00046858552377671003,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4364342039989424
        },
        {
          "epoch": 25,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.008278040666482411,
          "train_loss": 0.004652044270187616,
          "validation_loss": 0.004609555471688509,
          "train_metric": 0.00046853689127601683,
          "validation_metric": 0.0004642575513571501,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.44797699699847726
        },
        {
          "epoch": 26,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.005110999627504498,
          "train_loss": 0.004341036546975374,
          "validation_loss": 0.003443229477852583,
          "train_metric": 0.0004372132825665176,
          "validation_metric": 0.0003467893984634429,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.46069998999882955
        },
        {
          "epoch": 27,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.004839688001084141,
          "train_loss": 0.00404869532212615,
          "validation_loss": 0.0022409860976040363,
          "train_metric": 0.0004077698104083538,
          "validation_metric": 0.00022570393048226833,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4734746879985323
        },
        {
          "epoch": 28,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.006637361788307317,
          "train_loss": 0.003133957739919424,
          "validation_loss": 0.0033039322588592768,
          "train_metric": 0.0003156407328788191,
          "validation_metric": 0.0003327599843032658,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.48527665799883835
        },
        {
          "epoch": 29,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.0044007540855091065,
          "train_loss": 0.007334485650062561,
          "validation_loss": 0.007427752483636141,
          "train_metric": 0.000738702539820224,
          "validation_metric": 0.0007480959757231176,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4972409530000732
        },
        {
          "epoch": 30,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.007312072368222289,
          "train_loss": 0.011192835867404938,
          "validation_loss": 0.009809882380068302,
          "train_metric": 0.0011273014824837446,
          "validation_metric": 0.0009880154393613338,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5089793609986373
        },
        {
          "epoch": 31,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.011944417288759723,
          "train_loss": 0.015252713114023209,
          "validation_loss": 0.009867140091955662,
          "train_metric": 0.001536197611130774,
          "validation_metric": 0.0009937823051586747,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5209320159992785
        },
        {
          "epoch": 32,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.01918676812783815,
          "train_loss": 0.008229783736169338,
          "validation_loss": 0.00664070900529623,
          "train_metric": 0.0008288737153634429,
          "validation_metric": 0.000668827909976244,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5329846789991279
        },
        {
          "epoch": 33,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.007924737190478481,
          "train_loss": 0.009503010660409927,
          "validation_loss": 0.008739456534385681,
          "train_metric": 0.0009571083937771618,
          "validation_metric": 0.0008802061202004552,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5448937269993621
        },
        {
          "epoch": 34,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.0060406165139283985,
          "train_loss": 0.004433082416653633,
          "validation_loss": 0.003141037654131651,
          "train_metric": 0.0004464838420972228,
          "validation_metric": 0.00031635383493267,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5565852299987455
        },
        {
          "epoch": 35,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.0071589397120988,
          "train_loss": 0.0048681641928851604,
          "validation_loss": 0.0044341073371469975,
          "train_metric": 0.0004903037333860993,
          "validation_metric": 0.0004465870442800224,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5682474690001982
        },
        {
          "epoch": 36,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.006210321422258858,
          "train_loss": 0.008831135928630829,
          "validation_loss": 0.006625158712267876,
          "train_metric": 0.0008894396596588194,
          "validation_metric": 0.0006672616000287235,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5797820049992879
        },
        {
          "epoch": 37,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.01327749777919962,
          "train_loss": 0.006252899765968323,
          "validation_loss": 0.0056560649536550045,
          "train_metric": 0.0006297691143117845,
          "validation_metric": 0.0005696581793017685,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.5914724719987134
        },
        {
          "epoch": 38,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.01251587161095813,
          "train_loss": 0.010483533143997192,
          "validation_loss": 0.008579114452004433,
          "train_metric": 0.0010558629874140024,
          "validation_metric": 0.0008640570449642837,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.6031375319998915
        },
        {
          "epoch": 39,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.005335541412932798,
          "train_loss": 0.004233999643474817,
          "validation_loss": 0.003688322613015771,
          "train_metric": 0.00042643293272703886,
          "validation_metric": 0.0003714743070304394,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.614583386999584
        },
        {
          "epoch": 40,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.003953272498620208,
          "train_loss": 0.0034779412671923637,
          "validation_loss": 0.002764059929177165,
          "train_metric": 0.00035028555430471897,
          "validation_metric": 0.0002783860545605421,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6261559249996935
        },
        {
          "epoch": 41,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.00377638908685185,
          "train_loss": 0.0027444881852716208,
          "validation_loss": 0.0030336780473589897,
          "train_metric": 0.0002764147939160466,
          "validation_metric": 0.0003055409179069102,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6377412949987047
        },
        {
          "epoch": 42,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.0037896673165960237,
          "train_loss": 0.002957987366244197,
          "validation_loss": 0.002368975430727005,
          "train_metric": 0.0002979176351800561,
          "validation_metric": 0.00023859455541241914,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6493589289984811
        },
        {
          "epoch": 43,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.0032661419681971893,
          "train_loss": 0.0026191635988652706,
          "validation_loss": 0.002050513867288828,
          "train_metric": 0.0002637925499584526,
          "validation_metric": 0.00020652025705203414,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6617554049989849
        },
        {
          "epoch": 44,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.003126035513560055,
          "train_loss": 0.0018938423600047827,
          "validation_loss": 0.001551566761918366,
          "train_metric": 0.00019074087322223932,
          "validation_metric": 0.00015626814274583012,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6744706760000554
        },
        {
          "epoch": 45,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.0022818444849690422,
          "train_loss": 0.0027080366853624582,
          "validation_loss": 0.0018924244213849306,
          "train_metric": 0.00027274349122308195,
          "validation_metric": 0.00019059803162235767,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6863097479999851
        },
        {
          "epoch": 46,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.003363342890224885,
          "train_loss": 0.0049658408388495445,
          "validation_loss": 0.004808166064321995,
          "train_metric": 0.0005001412937417626,
          "validation_metric": 0.0004842609050683677,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6979689660001895
        },
        {
          "epoch": 47,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.004253091574355494,
          "train_loss": 0.003261780831962824,
          "validation_loss": 0.0023286305367946625,
          "train_metric": 0.000328514666762203,
          "validation_metric": 0.00023453116591554135,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7100036189985985
        },
        {
          "epoch": 48,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.002463635250023799,
          "train_loss": 0.0020019318908452988,
          "validation_loss": 0.0021933219395577908,
          "train_metric": 0.00020162724831607193,
          "validation_metric": 0.00022090337006375194,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7217264099999738
        },
        {
          "epoch": 49,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.003193930879206164,
          "train_loss": 0.0037988610565662384,
          "validation_loss": 0.0037622912786900997,
          "train_metric": 0.0003826073952950537,
          "validation_metric": 0.0003789242182392627,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7335833349989116
        },
        {
          "epoch": 50,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.0044996889228059445,
          "train_loss": 0.0018775853095576167,
          "validation_loss": 0.0010179852833971381,
          "train_metric": 0.00018910352082457393,
          "validation_metric": 0.00010252775246044621,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7459286579996842
        },
        {
          "epoch": 51,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.0016705868511053268,
          "train_loss": 0.0014612646773457527,
          "validation_loss": 0.001000102492980659,
          "train_metric": 0.00014717323938384652,
          "validation_metric": 0.00010072665463667363,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7587148989987327
        },
        {
          "epoch": 52,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.0021061545085103717,
          "train_loss": 0.002590892603620887,
          "validation_loss": 0.0015706499107182026,
          "train_metric": 0.00026094523491337895,
          "validation_metric": 0.0001581901015015319,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7704920950000087
        },
        {
          "epoch": 53,
          "updates": 848,
          "samples_seen": 54272,
          "online_loss": 0.0020384202143759467,
          "train_loss": 0.0013220999389886856,
          "validation_loss": 0.0012417752295732498,
          "train_metric": 0.00013315706746652722,
          "validation_metric": 0.00012506706116255373,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7826929349994316
        },
        {
          "epoch": 54,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0028513810466392897,
          "train_loss": 0.004793196450918913,
          "validation_loss": 0.005338828079402447,
          "train_metric": 0.00048275318113155663,
          "validation_metric": 0.0005377072375267744,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7944788099994184
        },
        {
          "epoch": 55,
          "updates": 880,
          "samples_seen": 56320,
          "online_loss": 0.004428972246387275,
          "train_loss": 0.004048469942063093,
          "validation_loss": 0.004571325145661831,
          "train_metric": 0.00040774705121293664,
          "validation_metric": 0.0004604071145877242,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8061789969997335
        },
        {
          "epoch": 56,
          "updates": 896,
          "samples_seen": 57344,
          "online_loss": 0.0031326928292401135,
          "train_loss": 0.0032271670643240213,
          "validation_loss": 0.0019543946254998446,
          "train_metric": 0.00032502846443094313,
          "validation_metric": 0.00019683946447912604,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8180586679991393
        },
        {
          "epoch": 57,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.005380890012020245,
          "train_loss": 0.006661044433712959,
          "validation_loss": 0.006052902899682522,
          "train_metric": 0.0006708760047331452,
          "validation_metric": 0.0006096262368373573,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8298659669999324
        },
        {
          "epoch": 58,
          "updates": 928,
          "samples_seen": 59392,
          "online_loss": 0.0034078115058946423,
          "train_loss": 0.002100005280226469,
          "validation_loss": 0.0021339955274015665,
          "train_metric": 0.00021150482643861324,
          "validation_metric": 0.00021492820815183222,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8416298289994302
        },
        {
          "epoch": 59,
          "updates": 944,
          "samples_seen": 60416,
          "online_loss": 0.0018083594404743053,
          "train_loss": 0.001776599558070302,
          "validation_loss": 0.0014196888078004122,
          "train_metric": 0.00017893261974677444,
          "validation_metric": 0.00014298585301730782,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8533137519989396
        },
        {
          "epoch": 60,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.0014302956406027079,
          "train_loss": 0.0011902791447937489,
          "validation_loss": 0.0008654524572193623,
          "train_metric": 0.00011988055484835058,
          "validation_metric": 8.716520096641034e-05,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8657990599986078
        },
        {
          "epoch": 61,
          "updates": 976,
          "samples_seen": 62464,
          "online_loss": 0.001970822188013699,
          "train_loss": 0.001578174065798521,
          "validation_loss": 0.001249500666745007,
          "train_metric": 0.0001589479361427948,
          "validation_metric": 0.00012584512296598405,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8777598919987213
        },
        {
          "epoch": 62,
          "updates": 992,
          "samples_seen": 63488,
          "online_loss": 0.005406915570347337,
          "train_loss": 0.0030568751972168684,
          "validation_loss": 0.003068486461415887,
          "train_metric": 0.00030787725700065494,
          "validation_metric": 0.00030904673621989787,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8894871109987434
        },
        {
          "epoch": 63,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.0031975126257748343,
          "train_loss": 0.0018979760352522135,
          "validation_loss": 0.0013291232753545046,
          "train_metric": 0.00019115718896500766,
          "validation_metric": 0.00013386443606577814,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9013774859995465
        },
        {
          "epoch": 64,
          "updates": 1024,
          "samples_seen": 65536,
          "online_loss": 0.0016776107077021152,
          "train_loss": 0.0017805370735004544,
          "validation_loss": 0.0013566658599302173,
          "train_metric": 0.00017932915943674743,
          "validation_metric": 0.00013663839490618557,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9129767899994476
        },
        {
          "epoch": 65,
          "updates": 1040,
          "samples_seen": 66560,
          "online_loss": 0.003178376646246761,
          "train_loss": 0.005163970869034529,
          "validation_loss": 0.005070366896688938,
          "train_metric": 0.0005200962768867612,
          "validation_metric": 0.0005106688477098942,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9248368069984281
        },
        {
          "epoch": 66,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.0031665079477534164,
          "train_loss": 0.0015181401977315545,
          "validation_loss": 0.0010476965690031648,
          "train_metric": 0.000152901528053917,
          "validation_metric": 0.00010552018648013473,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9365838089997851
        },
        {
          "epoch": 67,
          "updates": 1072,
          "samples_seen": 68608,
          "online_loss": 0.002006137241551187,
          "train_loss": 0.002555121900513768,
          "validation_loss": 0.00190651323646307,
          "train_metric": 0.0002573425299488008,
          "validation_metric": 0.00019201700342819095,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9480639489993337
        },
        {
          "epoch": 68,
          "updates": 1088,
          "samples_seen": 69632,
          "online_loss": 0.0016639358691463713,
          "train_loss": 0.0015188218094408512,
          "validation_loss": 0.0011548068141564727,
          "train_metric": 0.0001529701694380492,
          "validation_metric": 0.00011630791414063424,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.959880478998457
        },
        {
          "epoch": 69,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.0015187839053396601,
          "train_loss": 0.0010285762837156653,
          "validation_loss": 0.0009928204817697406,
          "train_metric": 0.00010359444422647357,
          "validation_metric": 9.999324538512155e-05,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9714502099996025
        },
        {
          "epoch": 70,
          "updates": 1120,
          "samples_seen": 71680,
          "online_loss": 0.0015248895615513902,
          "train_loss": 0.0022571294102817774,
          "validation_loss": 0.00145188276655972,
          "train_metric": 0.00022732981597073376,
          "validation_metric": 0.0001462282962165773,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9830525039997156
        },
        {
          "epoch": 71,
          "updates": 1136,
          "samples_seen": 72704,
          "online_loss": 0.001439062592908158,
          "train_loss": 0.0014999349368736148,
          "validation_loss": 0.0010605333372950554,
          "train_metric": 0.0001510679576313123,
          "validation_metric": 0.00010681303683668375,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9949239409997972
        },
        {
          "epoch": 72,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0014940136470613652,
          "train_loss": 0.0008434871560893953,
          "validation_loss": 0.0005514133372344077,
          "train_metric": 8.495293877786025e-05,
          "validation_metric": 5.553632945520803e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0071533310001541
        },
        {
          "epoch": 73,
          "updates": 1168,
          "samples_seen": 74752,
          "online_loss": 0.002026221026426356,
          "train_loss": 0.001112982165068388,
          "validation_loss": 0.0009021915611810982,
          "train_metric": 0.0001120954766520299,
          "validation_metric": 9.0865432866849e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.019059191999986
        },
        {
          "epoch": 74,
          "updates": 1184,
          "samples_seen": 75776,
          "online_loss": 0.0015926780961308395,
          "train_loss": 0.0025590313598513603,
          "validation_loss": 0.002567782998085022,
          "train_metric": 0.00025773627567104995,
          "validation_metric": 0.00025861768517643213,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.030651149998448
        },
        {
          "epoch": 75,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.001985373864954454,
          "train_loss": 0.0012435856042429805,
          "validation_loss": 0.0012410796480253339,
          "train_metric": 0.00012524938210844994,
          "validation_metric": 0.00012499700824264437,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0422395739988133
        },
        {
          "epoch": 76,
          "updates": 1216,
          "samples_seen": 77824,
          "online_loss": 0.0011924768241442507,
          "train_loss": 0.0009991909610107541,
          "validation_loss": 0.0006971396505832672,
          "train_metric": 0.00010063486115541309,
          "validation_metric": 7.021335477475077e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0538451289994555
        },
        {
          "epoch": 77,
          "updates": 1232,
          "samples_seen": 78848,
          "online_loss": 0.0009482882505835732,
          "train_loss": 0.0008499838295392692,
          "validation_loss": 0.0006060958839952946,
          "train_metric": 8.560725837014616e-05,
          "validation_metric": 6.104375643189996e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0654315849988052
        },
        {
          "epoch": 78,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.0008338067891600076,
          "train_loss": 0.0008567058830522001,
          "validation_loss": 0.0006747104926034808,
          "train_metric": 8.628427895018831e-05,
          "validation_metric": 6.795437366236001e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.077266984999369
        },
        {
          "epoch": 79,
          "updates": 1264,
          "samples_seen": 80896,
          "online_loss": 0.0009178394248010591,
          "train_loss": 0.000982388504780829,
          "validation_loss": 0.0007323910831473768,
          "train_metric": 9.894258255371824e-05,
          "validation_metric": 7.376374560408294e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0888361059987801
        },
        {
          "epoch": 80,
          "updates": 1280,
          "samples_seen": 81920,
          "online_loss": 0.0012114068758819485,
          "train_loss": 0.001715384772978723,
          "validation_loss": 0.001122105517424643,
          "train_metric": 0.0001727672788547352,
          "validation_metric": 0.00011301435733912513,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1000437579987192
        },
        {
          "epoch": 81,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.0022773641476305784,
          "train_loss": 0.0024052015505731106,
          "validation_loss": 0.0027181541081517935,
          "train_metric": 0.0002422430698061362,
          "validation_metric": 0.000273762532742694,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1112948069985578
        },
        {
          "epoch": 82,
          "updates": 1312,
          "samples_seen": 83968,
          "online_loss": 0.004071679988555843,
          "train_loss": 0.003537379205226898,
          "validation_loss": 0.0026577571406960487,
          "train_metric": 0.00035627189208753407,
          "validation_metric": 0.00026767957024276257,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1225092219992803
        },
        {
          "epoch": 83,
          "updates": 1328,
          "samples_seen": 84992,
          "online_loss": 0.0024950070492195664,
          "train_loss": 0.0007838337915018201,
          "validation_loss": 0.000725074321962893,
          "train_metric": 7.894486770965159e-05,
          "validation_metric": 7.302682206500322e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1338218429991684
        },
        {
          "epoch": 84,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.0008696256863913732,
          "train_loss": 0.0006963034393265843,
          "validation_loss": 0.00046888476936146617,
          "train_metric": 7.012912828940898e-05,
          "validation_metric": 4.7224355512298644e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1460500289995252
        },
        {
          "epoch": 85,
          "updates": 1360,
          "samples_seen": 87040,
          "online_loss": 0.0007062994345687912,
          "train_loss": 0.0006695411866530776,
          "validation_loss": 0.0006062244065105915,
          "train_metric": 6.743372796336189e-05,
          "validation_metric": 6.105670763645321e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1576108419994853
        },
        {
          "epoch": 86,
          "updates": 1376,
          "samples_seen": 88064,
          "online_loss": 0.0006381670700648101,
          "train_loss": 0.0005992556689307094,
          "validation_loss": 0.00046084614587016404,
          "train_metric": 6.035483966115862e-05,
          "validation_metric": 4.641473788069561e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1696713029996317
        },
        {
          "epoch": 87,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0006606417564398726,
          "train_loss": 0.0006323436973616481,
          "validation_loss": 0.000453388987807557,
          "train_metric": 6.368733011186123e-05,
          "validation_metric": 4.566367715597153e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.181559541999377
        },
        {
          "epoch": 88,
          "updates": 1408,
          "samples_seen": 90112,
          "online_loss": 0.0006196066569827963,
          "train_loss": 0.000547687232028693,
          "validation_loss": 0.00044589993194676936,
          "train_metric": 5.516105011338368e-05,
          "validation_metric": 4.490940409596078e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1933237279990863
        },
        {
          "epoch": 89,
          "updates": 1424,
          "samples_seen": 91136,
          "online_loss": 0.0006692092283628881,
          "train_loss": 0.0005220096209086478,
          "validation_loss": 0.0004208296013530344,
          "train_metric": 5.257489829091355e-05,
          "validation_metric": 4.238441397319548e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2050938579996
        },
        {
          "epoch": 90,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0006681594386463985,
          "train_loss": 0.0011662142351269722,
          "validation_loss": 0.0008668734226375818,
          "train_metric": 0.00011745680967578664,
          "validation_metric": 8.730831905268133e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2164107159987907
        },
        {
          "epoch": 91,
          "updates": 1456,
          "samples_seen": 93184,
          "online_loss": 0.0008924047433538362,
          "train_loss": 0.0008188040228560567,
          "validation_loss": 0.0007451470592059195,
          "train_metric": 8.246693323599175e-05,
          "validation_metric": 7.504848326789215e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2277953039993008
        },
        {
          "epoch": 92,
          "updates": 1472,
          "samples_seen": 94208,
          "online_loss": 0.0008474355963699054,
          "train_loss": 0.0005372262094169855,
          "validation_loss": 0.00041728292126208544,
          "train_metric": 5.410745507106185e-05,
          "validation_metric": 4.2027211748063564e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.239738114998545
        },
        {
          "epoch": 93,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0006022331899657729,
          "train_loss": 0.0005183502216823399,
          "validation_loss": 0.00039882780401967466,
          "train_metric": 5.220633465796709e-05,
          "validation_metric": 4.016848106402904e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.251692223999271
        },
        {
          "epoch": 94,
          "updates": 1504,
          "samples_seen": 96256,
          "online_loss": 0.000592401755056926,
          "train_loss": 0.0005056716036051512,
          "validation_loss": 0.00040047228685580194,
          "train_metric": 5.092938954476267e-05,
          "validation_metric": 4.0334103687200695e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2629664769992814
        },
        {
          "epoch": 95,
          "updates": 1520,
          "samples_seen": 97280,
          "online_loss": 0.0005336565418474493,
          "train_loss": 0.00048708217218518257,
          "validation_loss": 0.00035464862594380975,
          "train_metric": 4.905711830360815e-05,
          "validation_metric": 3.5718905564863235e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.274548268998842
        },
        {
          "epoch": 96,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.000528234390912985,
          "train_loss": 0.0005377856432460248,
          "validation_loss": 0.00046728926827199757,
          "train_metric": 5.416380372480489e-05,
          "validation_metric": 4.706365871243179e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2857353419985884
        },
        {
          "epoch": 97,
          "updates": 1552,
          "samples_seen": 99328,
          "online_loss": 0.000531343427610409,
          "train_loss": 0.00044532943866215646,
          "validation_loss": 0.0003487455251161009,
          "train_metric": 4.485195313463919e-05,
          "validation_metric": 3.512437251629308e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2973926499998925
        },
        {
          "epoch": 98,
          "updates": 1568,
          "samples_seen": 100352,
          "online_loss": 0.0005508964241016656,
          "train_loss": 0.0004510860308073461,
          "validation_loss": 0.0003458448627498001,
          "train_metric": 4.543173417914659e-05,
          "validation_metric": 3.4832228266168386e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3091069749989401
        },
        {
          "epoch": 99,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0006911301552463556,
          "train_loss": 0.0007764076581224799,
          "validation_loss": 0.0006253534811548889,
          "train_metric": 7.81969356467016e-05,
          "validation_metric": 6.298331572907045e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.320229589999144
        },
        {
          "epoch": 100,
          "updates": 1600,
          "samples_seen": 102400,
          "online_loss": 0.0005399152696554665,
          "train_loss": 0.0006966835353523493,
          "validation_loss": 0.00041482425876893103,
          "train_metric": 7.016742165433243e-05,
          "validation_metric": 4.1779581806622446e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3312237270001788
        },
        {
          "epoch": 101,
          "updates": 1616,
          "samples_seen": 103424,
          "online_loss": 0.0006984431165619753,
          "train_loss": 0.0008436291245743632,
          "validation_loss": 0.0008493774803355336,
          "train_metric": 8.496723603457212e-05,
          "validation_metric": 8.554617670597509e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3423596829998132
        },
        {
          "epoch": 102,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0007001500125625171,
          "train_loss": 0.0005392361781559885,
          "validation_loss": 0.0004396653384901583,
          "train_metric": 5.4309894039761275e-05,
          "validation_metric": 4.4281485315877944e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.353463751000163
        },
        {
          "epoch": 103,
          "updates": 1648,
          "samples_seen": 105472,
          "online_loss": 0.0004681957334469189,
          "train_loss": 0.0008502296986989677,
          "validation_loss": 0.0007560586091130972,
          "train_metric": 8.563201845390722e-05,
          "validation_metric": 7.614745118189603e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3644721459986613
        },
        {
          "epoch": 104,
          "updates": 1664,
          "samples_seen": 106496,
          "online_loss": 0.0008260307577074855,
          "train_loss": 0.0004480724746827036,
          "validation_loss": 0.00037234771298244596,
          "train_metric": 4.5128217607270926e-05,
          "validation_metric": 3.7501500628422946e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3753929909998988
        },
        {
          "epoch": 105,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0006001081610520487,
          "train_loss": 0.0004655866068787873,
          "validation_loss": 0.0003252593451179564,
          "train_metric": 4.6892175305401906e-05,
          "validation_metric": 3.2758936868049204e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.386933298999793
        },
        {
          "epoch": 106,
          "updates": 1696,
          "samples_seen": 108544,
          "online_loss": 0.00045485134160117013,
          "train_loss": 0.0005409763543866575,
          "validation_loss": 0.0004489580460358411,
          "train_metric": 5.448516094475053e-05,
          "validation_metric": 4.52174135716632e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3980645669998921
        },
        {
          "epoch": 107,
          "updates": 1712,
          "samples_seen": 109568,
          "online_loss": 0.0004908022056042682,
          "train_loss": 0.00040013869875110686,
          "validation_loss": 0.0002980868157465011,
          "train_metric": 4.03005069529172e-05,
          "validation_metric": 3.0022212740732357e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4097549859998253
        },
        {
          "epoch": 108,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0005982665807096055,
          "train_loss": 0.0005868626176379621,
          "validation_loss": 0.0004462937358766794,
          "train_metric": 5.910665640840307e-05,
          "validation_metric": 4.494907261687331e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4208948069990583
        },
        {
          "epoch": 109,
          "updates": 1744,
          "samples_seen": 111616,
          "online_loss": 0.00048280295686708996,
          "train_loss": 0.0003902291937265545,
          "validation_loss": 0.0002997904666699469,
          "train_metric": 3.9302460209000856e-05,
          "validation_metric": 3.0193798011168838e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4318553609991795
        },
        {
          "epoch": 110,
          "updates": 1760,
          "samples_seen": 112640,
          "online_loss": 0.0005666137949447148,
          "train_loss": 0.0011385675752535462,
          "validation_loss": 0.0008203402394428849,
          "train_metric": 0.00011467235162854195,
          "validation_metric": 8.26216783025302e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4428080050001881
        },
        {
          "epoch": 111,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.0006922585562278982,
          "train_loss": 0.0006016285624355078,
          "validation_loss": 0.000547126226592809,
          "train_metric": 6.059382576495409e-05,
          "validation_metric": 5.510454502655193e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.453745409999101
        },
        {
          "epoch": 112,
          "updates": 1792,
          "samples_seen": 114688,
          "online_loss": 0.0005104869715069071,
          "train_loss": 0.0003946213109884411,
          "validation_loss": 0.00029680764419026673,
          "train_metric": 3.974481296609156e-05,
          "validation_metric": 2.9893380997236818e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4653582249993633
        },
        {
          "epoch": 113,
          "updates": 1808,
          "samples_seen": 115712,
          "online_loss": 0.00048767466705612605,
          "train_loss": 0.0013670644257217646,
          "validation_loss": 0.0011792592704296112,
          "train_metric": 0.00013768573990091681,
          "validation_metric": 0.00011877068027388304,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4764686540002003
        },
        {
          "epoch": 114,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.0009408915257154149,
          "train_loss": 0.00034648869768716395,
          "validation_loss": 0.0002557392872404307,
          "train_metric": 3.489707160042599e-05,
          "validation_metric": 2.575712278485298e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.488091703999089
        },
        {
          "epoch": 115,
          "updates": 1840,
          "samples_seen": 117760,
          "online_loss": 0.00042846056385315023,
          "train_loss": 0.00041302188765257597,
          "validation_loss": 0.000299140257993713,
          "train_metric": 4.159804666414857e-05,
          "validation_metric": 3.012831075466238e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4992160479996528
        },
        {
          "epoch": 116,
          "updates": 1856,
          "samples_seen": 118784,
          "online_loss": 0.00039185746527437004,
          "train_loss": 0.00037092898855917156,
          "validation_loss": 0.00033607319346629083,
          "train_metric": 3.7358611734816805e-05,
          "validation_metric": 3.38480640493799e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5103453229985462
        },
        {
          "epoch": 117,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.0003776965177166858,
          "train_loss": 0.0003449218929745257,
          "validation_loss": 0.00025178369833156466,
          "train_metric": 3.473927063168958e-05,
          "validation_metric": 2.535873318265658e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5219591969998874
        },
        {
          "epoch": 118,
          "updates": 1888,
          "samples_seen": 120832,
          "online_loss": 0.00037904972396063386,
          "train_loss": 0.0003402087895665318,
          "validation_loss": 0.00026857058401219547,
          "train_metric": 3.426457988098264e-05,
          "validation_metric": 2.7049449272453785e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5330282279992389
        },
        {
          "epoch": 119,
          "updates": 1904,
          "samples_seen": 121856,
          "online_loss": 0.00041737469382496784,
          "train_loss": 0.00035983131965622306,
          "validation_loss": 0.0002504259755369276,
          "train_metric": 3.624089004006237e-05,
          "validation_metric": 2.522198519727681e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.544743901999027
        },
        {
          "epoch": 120,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0006039239233359694,
          "train_loss": 0.0003244675463065505,
          "validation_loss": 0.0002660635800566524,
          "train_metric": 3.267918509664014e-05,
          "validation_metric": 2.6796951715368778e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.555770331999156
        },
        {
          "epoch": 121,
          "updates": 1936,
          "samples_seen": 123904,
          "online_loss": 0.0003905162620867486,
          "train_loss": 0.00030449670157395303,
          "validation_loss": 0.0002486840239726007,
          "train_metric": 3.0667793907923624e-05,
          "validation_metric": 2.5046547307283618e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.567341841999223
        },
        {
          "epoch": 122,
          "updates": 1952,
          "samples_seen": 124928,
          "online_loss": 0.0003688499054987915,
          "train_loss": 0.000627216650173068,
          "validation_loss": 0.0004478181363083422,
          "train_metric": 6.317096995189786e-05,
          "validation_metric": 4.5102606236469e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.578270504000102
        },
        {
          "epoch": 123,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.000710779961082153,
          "train_loss": 0.0021248417906463146,
          "validation_loss": 0.001844349317252636,
          "train_metric": 0.00021400628611445427,
          "validation_metric": 0.00018575610010884702,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5890437429989106
        },
        {
          "epoch": 124,
          "updates": 1984,
          "samples_seen": 126976,
          "online_loss": 0.0010519336556171766,
          "train_loss": 0.00098179723136127,
          "validation_loss": 0.000862658373080194,
          "train_metric": 9.888302156468853e-05,
          "validation_metric": 8.688379602972418e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.599894584998765
        },
        {
          "epoch": 125,
          "updates": 2000,
          "samples_seen": 128000,
          "online_loss": 0.0007679610071136267,
          "train_loss": 0.00032956284121610224,
          "validation_loss": 0.0002776906476356089,
          "train_metric": 3.319236566312611e-05,
          "validation_metric": 2.796798617055174e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6107469529997616
        },
        {
          "epoch": 126,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.000348027449035726,
          "train_loss": 0.00031173796742223203,
          "validation_loss": 0.00024146659416146576,
          "train_metric": 3.139711043331772e-05,
          "validation_metric": 2.4319630028912798e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6219824649997463
        },
        {
          "epoch": 127,
          "updates": 2032,
          "samples_seen": 130048,
          "online_loss": 0.00036180705683364067,
          "train_loss": 0.0008208840154111385,
          "validation_loss": 0.0006034027901478112,
          "train_metric": 8.267643715953454e-05,
          "validation_metric": 6.077251964597963e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6330947970000125
        },
        {
          "epoch": 128,
          "updates": 2048,
          "samples_seen": 131072,
          "online_loss": 0.0004490423871175153,
          "train_loss": 0.0002998602867592126,
          "validation_loss": 0.0002167179191019386,
          "train_metric": 3.020083750016056e-05,
          "validation_metric": 2.182703792641405e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6444779140001629
        },
        {
          "epoch": 129,
          "updates": 2064,
          "samples_seen": 132096,
          "online_loss": 0.00030613996659667464,
          "train_loss": 0.0006652660667896271,
          "validation_loss": 0.0004243291332386434,
          "train_metric": 6.700315861962736e-05,
          "validation_metric": 4.273687954992056e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6552965659993788
        },
        {
          "epoch": 130,
          "updates": 2080,
          "samples_seen": 133120,
          "online_loss": 0.0005115826043038396,
          "train_loss": 0.0003118898603133857,
          "validation_loss": 0.00022277141397353262,
          "train_metric": 3.1412404496222734e-05,
          "validation_metric": 2.2436724975705147e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.666358295999089
        },
        {
          "epoch": 131,
          "updates": 2096,
          "samples_seen": 134144,
          "online_loss": 0.0003531270758685423,
          "train_loss": 0.000274205842288211,
          "validation_loss": 0.00020880613010376692,
          "train_metric": 2.7617006708169356e-05,
          "validation_metric": 2.1030185962445103e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.677901725999618
        },
        {
          "epoch": 132,
          "updates": 2112,
          "samples_seen": 135168,
          "online_loss": 0.000360027303941024,
          "train_loss": 0.0006244647083804011,
          "validation_loss": 0.0005000451346859336,
          "train_metric": 6.289379962254316e-05,
          "validation_metric": 5.036271613789722e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6885896929998125
        },
        {
          "epoch": 133,
          "updates": 2128,
          "samples_seen": 136192,
          "online_loss": 0.00033724552440617117,
          "train_loss": 0.0003035857807844877,
          "validation_loss": 0.00022145007096696645,
          "train_metric": 3.057604772038758e-05,
          "validation_metric": 2.230364225397352e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6995771789988794
        },
        {
          "epoch": 134,
          "updates": 2144,
          "samples_seen": 137216,
          "online_loss": 0.0004828611740776978,
          "train_loss": 0.0006174505688250065,
          "validation_loss": 0.00056999217486009,
          "train_metric": 6.218736234586686e-05,
          "validation_metric": 5.740753113059327e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7101690159997815
        },
        {
          "epoch": 135,
          "updates": 2160,
          "samples_seen": 138240,
          "online_loss": 0.00038734445752197644,
          "train_loss": 0.00026574768708087504,
          "validation_loss": 0.00019706643070094287,
          "train_metric": 2.6765133952721953e-05,
          "validation_metric": 1.9847811927320436e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7211185859996476
        },
        {
          "epoch": 136,
          "updates": 2176,
          "samples_seen": 139264,
          "online_loss": 0.0003963693893638265,
          "train_loss": 0.0008395526092499495,
          "validation_loss": 0.0006167516112327576,
          "train_metric": 8.455668285023421e-05,
          "validation_metric": 6.211696745594963e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7317254029985634
        },
        {
          "epoch": 137,
          "updates": 2192,
          "samples_seen": 140288,
          "online_loss": 0.0006073207496228861,
          "train_loss": 0.0002821187663357705,
          "validation_loss": 0.00019354865071363747,
          "train_metric": 2.8413971449481323e-05,
          "validation_metric": 1.9493509171297774e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7436760869986756
        },
        {
          "epoch": 138,
          "updates": 2208,
          "samples_seen": 141312,
          "online_loss": 0.0005030906449974282,
          "train_loss": 0.00042871065670624375,
          "validation_loss": 0.00032383500365540385,
          "train_metric": 4.3178166379220784e-05,
          "validation_metric": 3.261547681177035e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7552833859990642
        },
        {
          "epoch": 139,
          "updates": 2224,
          "samples_seen": 142336,
          "online_loss": 0.0003811306514762691,
          "train_loss": 0.0003264312690589577,
          "validation_loss": 0.00025608466239646077,
          "train_metric": 3.2876960176508874e-05,
          "validation_metric": 2.5791909138206393e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7666569819994038
        },
        {
          "epoch": 140,
          "updates": 2240,
          "samples_seen": 143360,
          "online_loss": 0.0003330643571644032,
          "train_loss": 0.0003936355351470411,
          "validation_loss": 0.0003640662180259824,
          "train_metric": 3.964552888646722e-05,
          "validation_metric": 3.666741758934222e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7782305860000633
        },
        {
          "epoch": 141,
          "updates": 2256,
          "samples_seen": 144384,
          "online_loss": 0.0004479325161810266,
          "train_loss": 0.0003910649102181196,
          "validation_loss": 0.00030627878732047975,
          "train_metric": 3.938662848668173e-05,
          "validation_metric": 3.08472772303503e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.7892740970000887
        },
        {
          "epoch": 142,
          "updates": 2272,
          "samples_seen": 145408,
          "online_loss": 0.00028675241128439666,
          "train_loss": 0.0006774336798116565,
          "validation_loss": 0.000534636084921658,
          "train_metric": 6.822864088462666e-05,
          "validation_metric": 5.384659016272053e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.800499270999353
        },
        {
          "epoch": 143,
          "updates": 2288,
          "samples_seen": 146432,
          "online_loss": 0.0006114945117587922,
          "train_loss": 0.0011111987987533212,
          "validation_loss": 0.0007578536169603467,
          "train_metric": 0.00011191586236236617,
          "validation_metric": 7.632823690073565e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.8121020189992123
        },
        {
          "epoch": 144,
          "updates": 2304,
          "samples_seen": 147456,
          "online_loss": 0.00039734360598231433,
          "train_loss": 0.00030875872471369803,
          "validation_loss": 0.0002604889450594783,
          "train_metric": 3.109704630332999e-05,
          "validation_metric": 2.6235491532133892e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.8234740959996998
        },
        {
          "epoch": 145,
          "updates": 2320,
          "samples_seen": 148480,
          "online_loss": 0.0006069539113013889,
          "train_loss": 0.0005058750975877047,
          "validation_loss": 0.0005799050559289753,
          "train_metric": 5.0949885917361826e-05,
          "validation_metric": 5.840590893058106e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.8344443979985954
        },
        {
          "epoch": 146,
          "updates": 2336,
          "samples_seen": 149504,
          "online_loss": 0.0003197610694769537,
          "train_loss": 0.00047973106848075986,
          "validation_loss": 0.00035565064172260463,
          "train_metric": 4.8316753236576915e-05,
          "validation_metric": 3.5819826734950766e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.8453237829999125
        },
        {
          "epoch": 147,
          "updates": 2352,
          "samples_seen": 150528,
          "online_loss": 0.0004995988101654802,
          "train_loss": 0.0003822286380454898,
          "validation_loss": 0.00034713040804490447,
          "train_metric": 3.8496669731102884e-05,
          "validation_metric": 3.496170393191278e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.8565968200000498
        },
        {
          "epoch": 148,
          "updates": 2368,
          "samples_seen": 151552,
          "online_loss": 0.0003516895776556339,
          "train_loss": 0.000310690636979416,
          "validation_loss": 0.00029425707180052996,
          "train_metric": 3.129162359982729e-05,
          "validation_metric": 2.9636494218721054e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.8674408569986554
        },
        {
          "epoch": 149,
          "updates": 2384,
          "samples_seen": 152576,
          "online_loss": 0.00028639693937293487,
          "train_loss": 0.0002947768080048263,
          "validation_loss": 0.0002647123474162072,
          "train_metric": 2.9688841095776297e-05,
          "validation_metric": 2.6660853109206073e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.878571514000214
        },
        {
          "epoch": 150,
          "updates": 2400,
          "samples_seen": 153600,
          "online_loss": 0.00025839478121270076,
          "train_loss": 0.0002267792442580685,
          "validation_loss": 0.00019103699014522135,
          "train_metric": 2.2840376914246008e-05,
          "validation_metric": 1.9240544133936055e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8908624129999225
        },
        {
          "epoch": 151,
          "updates": 2416,
          "samples_seen": 154624,
          "online_loss": 0.00038653478486594395,
          "train_loss": 0.0002742597716860473,
          "validation_loss": 0.00025079381885007024,
          "train_metric": 2.7622438210528344e-05,
          "validation_metric": 2.5259036192437634e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9022859309989144
        },
        {
          "epoch": 152,
          "updates": 2432,
          "samples_seen": 155648,
          "online_loss": 0.0005634970048049581,
          "train_loss": 0.0005468317540362477,
          "validation_loss": 0.0004913934390060604,
          "train_metric": 5.507489186129533e-05,
          "validation_metric": 4.949135472998023e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9136362219996954
        },
        {
          "epoch": 153,
          "updates": 2448,
          "samples_seen": 156672,
          "online_loss": 0.0004801420382136712,
          "train_loss": 0.00025247310986742377,
          "validation_loss": 0.0001886619720607996,
          "train_metric": 2.5428165827179328e-05,
          "validation_metric": 1.9001343389390968e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9256718969991198
        },
        {
          "epoch": 154,
          "updates": 2464,
          "samples_seen": 157696,
          "online_loss": 0.0003251306584388658,
          "train_loss": 0.0003476211568340659,
          "validation_loss": 0.0002626196655910462,
          "train_metric": 3.501112951198593e-05,
          "validation_metric": 2.645009408297483e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9371795199986082
        },
        {
          "epoch": 155,
          "updates": 2480,
          "samples_seen": 158720,
          "online_loss": 0.0004049230192322284,
          "train_loss": 0.0002491021587047726,
          "validation_loss": 0.0001966276322491467,
          "train_metric": 2.5088656911975704e-05,
          "validation_metric": 1.980361412279308e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9487230369995814
        },
        {
          "epoch": 156,
          "updates": 2496,
          "samples_seen": 159744,
          "online_loss": 0.00029620567920574103,
          "train_loss": 0.0002980631252285093,
          "validation_loss": 0.00020609823695849627,
          "train_metric": 3.00198244076455e-05,
          "validation_metric": 2.0757461243192665e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.960728169999129
        },
        {
          "epoch": 157,
          "updates": 2512,
          "samples_seen": 160768,
          "online_loss": 0.00041525469941916526,
          "train_loss": 0.00043503998313099146,
          "validation_loss": 0.000323075451888144,
          "train_metric": 4.3815634853672236e-05,
          "validation_metric": 3.253897375543602e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.972289176999766
        },
        {
          "epoch": 158,
          "updates": 2528,
          "samples_seen": 161792,
          "online_loss": 0.0004777056556122261,
          "train_loss": 0.0002131945948349312,
          "validation_loss": 0.00016686473099980503,
          "train_metric": 2.1472180378623307e-05,
          "validation_metric": 1.6806005078251474e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9841495539985772
        },
        {
          "epoch": 159,
          "updates": 2544,
          "samples_seen": 162816,
          "online_loss": 0.00022610973837799975,
          "train_loss": 0.00020942965056747198,
          "validation_loss": 0.00016011809930205345,
          "train_metric": 2.109298475261312e-05,
          "validation_metric": 1.6126508853631094e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9966314739995141
        },
        {
          "epoch": 160,
          "updates": 2560,
          "samples_seen": 163840,
          "online_loss": 0.000267871137111797,
          "train_loss": 0.00036317366175353527,
          "validation_loss": 0.00031119538471102715,
          "train_metric": 3.657751949504018e-05,
          "validation_metric": 3.134246071567759e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.008612279998488
        },
        {
          "epoch": 161,
          "updates": 2576,
          "samples_seen": 164864,
          "online_loss": 0.00041032591798284557,
          "train_loss": 0.0005074284854345024,
          "validation_loss": 0.00043789102346636355,
          "train_metric": 5.110634447191842e-05,
          "validation_metric": 4.4102780520915985e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0200082259998453
        },
        {
          "epoch": 162,
          "updates": 2592,
          "samples_seen": 165888,
          "online_loss": 0.0003357888708706014,
          "train_loss": 0.0002463503915350884,
          "validation_loss": 0.0001651784114073962,
          "train_metric": 2.481150659150444e-05,
          "validation_metric": 1.6636166037642397e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0312289819994476
        },
        {
          "epoch": 163,
          "updates": 2608,
          "samples_seen": 166912,
          "online_loss": 0.0002176628959205118,
          "train_loss": 0.0002547978365328163,
          "validation_loss": 0.00020531618793029338,
          "train_metric": 2.566230614320375e-05,
          "validation_metric": 2.0678695364040323e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.043054650999693
        },
        {
          "epoch": 164,
          "updates": 2624,
          "samples_seen": 167936,
          "online_loss": 0.00022047909533284837,
          "train_loss": 0.00020388004486449063,
          "validation_loss": 0.00015409996558446437,
          "train_metric": 2.0534052964649163e-05,
          "validation_metric": 1.552038338559214e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0557185109992133
        },
        {
          "epoch": 165,
          "updates": 2640,
          "samples_seen": 168960,
          "online_loss": 0.00025090998815358034,
          "train_loss": 0.00019999906362500042,
          "validation_loss": 0.00016247358871623874,
          "train_metric": 2.01431739697e-05,
          "validation_metric": 1.6363743270630948e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0672885320000205
        },
        {
          "epoch": 166,
          "updates": 2656,
          "samples_seen": 169984,
          "online_loss": 0.00024061698104560492,
          "train_loss": 0.00023256769054569304,
          "validation_loss": 0.00017444680270273238,
          "train_metric": 2.3423366656061262e-05,
          "validation_metric": 1.7569642295711674e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0784738860002108
        },
        {
          "epoch": 167,
          "updates": 2672,
          "samples_seen": 171008,
          "online_loss": 0.00024689674592082156,
          "train_loss": 0.0002822498499881476,
          "validation_loss": 0.0002001117100007832,
          "train_metric": 2.8427170036593452e-05,
          "validation_metric": 2.0154519006609917e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.090136166998491
        },
        {
          "epoch": 168,
          "updates": 2688,
          "samples_seen": 172032,
          "online_loss": 0.00021598573403025512,
          "train_loss": 0.0002560985158197582,
          "validation_loss": 0.0001911029394250363,
          "train_metric": 2.5793306122068316e-05,
          "validation_metric": 1.9247188902227208e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1016311409985065
        },
        {
          "epoch": 169,
          "updates": 2704,
          "samples_seen": 173056,
          "online_loss": 0.00028543329563035513,
          "train_loss": 0.00019256483938079327,
          "validation_loss": 0.00013713182124774903,
          "train_metric": 1.9394426999497227e-05,
          "validation_metric": 1.381141282763565e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.113414962999741
        },
        {
          "epoch": 170,
          "updates": 2720,
          "samples_seen": 174080,
          "online_loss": 0.00019954519621023792,
          "train_loss": 0.00019489307305775583,
          "validation_loss": 0.00016360488370992243,
          "train_metric": 1.962891656148713e-05,
          "validation_metric": 1.647768476686906e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1250980459990387
        },
        {
          "epoch": 171,
          "updates": 2736,
          "samples_seen": 175104,
          "online_loss": 0.00024219679107773118,
          "train_loss": 0.00022774015087634325,
          "validation_loss": 0.00019063890795223415,
          "train_metric": 2.2937152607482858e-05,
          "validation_metric": 1.9200453607481904e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.136752311998862
        },
        {
          "epoch": 172,
          "updates": 2752,
          "samples_seen": 176128,
          "online_loss": 0.00020744196945088333,
          "train_loss": 0.00019041087944060564,
          "validation_loss": 0.00014630316582042724,
          "train_metric": 1.917748522828333e-05,
          "validation_metric": 1.4735119293618482e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.14816282099855
        },
        {
          "epoch": 173,
          "updates": 2768,
          "samples_seen": 177152,
          "online_loss": 0.00021398054332166794,
          "train_loss": 0.0005512743373401463,
          "validation_loss": 0.00041323681944049895,
          "train_metric": 5.552233778871596e-05,
          "validation_metric": 4.161969991400838e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1595489100000123
        },
        {
          "epoch": 174,
          "updates": 2784,
          "samples_seen": 178176,
          "online_loss": 0.00036205332344252383,
          "train_loss": 0.0008014894556254148,
          "validation_loss": 0.0005903361598029733,
          "train_metric": 8.072307537076995e-05,
          "validation_metric": 5.9456499002408236e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1711266549991706
        },
        {
          "epoch": 175,
          "updates": 2800,
          "samples_seen": 179200,
          "online_loss": 0.00045071384965922334,
          "train_loss": 0.00029420721693895757,
          "validation_loss": 0.00023634529497940093,
          "train_metric": 2.963147016998846e-05,
          "validation_metric": 2.3803831936675124e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1827140749992395
        },
        {
          "epoch": 176,
          "updates": 2816,
          "samples_seen": 180224,
          "online_loss": 0.0003108865012109163,
          "train_loss": 0.0002256504085380584,
          "validation_loss": 0.00013945958926342428,
          "train_metric": 2.2726682800566778e-05,
          "validation_metric": 1.4045857824385166e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.1943208789998607
        },
        {
          "epoch": 177,
          "updates": 2832,
          "samples_seen": 181248,
          "online_loss": 0.0002869410263883765,
          "train_loss": 0.00020912213949486613,
          "validation_loss": 0.0001735149708110839,
          "train_metric": 2.1062014639028348e-05,
          "validation_metric": 1.747579335642513e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.205964745999154
        },
        {
          "epoch": 178,
          "updates": 2848,
          "samples_seen": 182272,
          "online_loss": 0.0003060315307266137,
          "train_loss": 0.00027214805595576763,
          "validation_loss": 0.00026041633100248873,
          "train_metric": 2.7409756512497552e-05,
          "validation_metric": 2.6228177375742234e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.217286555998726
        },
        {
          "epoch": 179,
          "updates": 2864,
          "samples_seen": 183296,
          "online_loss": 0.0006167707424538094,
          "train_loss": 0.0008305271039716899,
          "validation_loss": 0.0005309295374900103,
          "train_metric": 8.364764653379098e-05,
          "validation_metric": 5.347327532945201e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.2283388579999155
        },
        {
          "epoch": 180,
          "updates": 2880,
          "samples_seen": 184320,
          "online_loss": 0.00040679453650227515,
          "train_loss": 0.00021956348791718483,
          "validation_loss": 0.000149208412040025,
          "train_metric": 2.211363062087912e-05,
          "validation_metric": 1.5027725567051675e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.2400129409998044
        },
        {
          "epoch": 181,
          "updates": 2896,
          "samples_seen": 185344,
          "online_loss": 0.000234804612773587,
          "train_loss": 0.0001831582048907876,
          "validation_loss": 0.00014254706911742687,
          "train_metric": 1.8447024558554403e-05,
          "validation_metric": 1.4356819519889541e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2514328399993246
        },
        {
          "epoch": 182,
          "updates": 2912,
          "samples_seen": 186368,
          "online_loss": 0.00023824039635655936,
          "train_loss": 0.00017331031267531216,
          "validation_loss": 0.00013344008766580373,
          "train_metric": 1.745517693052534e-05,
          "validation_metric": 1.3439595932140946e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2632644139994227
        },
        {
          "epoch": 183,
          "updates": 2928,
          "samples_seen": 187392,
          "online_loss": 0.0001928984252117516,
          "train_loss": 0.00021224447118584067,
          "validation_loss": 0.00017227459466084838,
          "train_metric": 2.1376485165092163e-05,
          "validation_metric": 1.7350866983179003e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2745490620000055
        },
        {
          "epoch": 184,
          "updates": 2944,
          "samples_seen": 188416,
          "online_loss": 0.00019098663915428915,
          "train_loss": 0.00017354104784317315,
          "validation_loss": 0.00013615822535939515,
          "train_metric": 1.747841997712385e-05,
          "validation_metric": 1.3713357475353405e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2857246530002158
        },
        {
          "epoch": 185,
          "updates": 2960,
          "samples_seen": 189440,
          "online_loss": 0.00021953921577733126,
          "train_loss": 0.0001756008277880028,
          "validation_loss": 0.00013985330588184297,
          "train_metric": 1.768587389960885e-05,
          "validation_metric": 1.408551361237187e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2969654469998204
        },
        {
          "epoch": 186,
          "updates": 2976,
          "samples_seen": 190464,
          "online_loss": 0.00022213120155356592,
          "train_loss": 0.00018215185264125466,
          "validation_loss": 0.00013992443564347923,
          "train_metric": 1.8345665012020618e-05,
          "validation_metric": 1.4092675883148331e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.308318703999248
        },
        {
          "epoch": 187,
          "updates": 2992,
          "samples_seen": 191488,
          "online_loss": 0.00022063078586143092,
          "train_loss": 0.00022679660469293594,
          "validation_loss": 0.00016839355521369725,
          "train_metric": 2.2842123144073412e-05,
          "validation_metric": 1.695998253126163e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.31951652799944
        },
        {
          "epoch": 188,
          "updates": 3008,
          "samples_seen": 192512,
          "online_loss": 0.0002112316424245364,
          "train_loss": 0.000173715190612711,
          "validation_loss": 0.00014235774870030582,
          "train_metric": 1.7495958672952838e-05,
          "validation_metric": 1.433775105397217e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.330332528999861
        },
        {
          "epoch": 189,
          "updates": 3024,
          "samples_seen": 193536,
          "online_loss": 0.00019560464988899184,
          "train_loss": 0.00023972720373421907,
          "validation_loss": 0.00018801738042384386,
          "train_metric": 2.4144446797436103e-05,
          "validation_metric": 1.8936421838589013e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.341290124999432
        },
        {
          "epoch": 190,
          "updates": 3040,
          "samples_seen": 194560,
          "online_loss": 0.000191525438822282,
          "train_loss": 0.00017085624858736992,
          "validation_loss": 0.0001344804186373949,
          "train_metric": 1.7208016288350336e-05,
          "validation_metric": 1.35443760882481e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3525072669999645
        },
        {
          "epoch": 191,
          "updates": 3056,
          "samples_seen": 195584,
          "online_loss": 0.00018026886391453445,
          "train_loss": 0.0002040734252659604,
          "validation_loss": 0.00014773044676985592,
          "train_metric": 2.0553527065203525e-05,
          "validation_metric": 1.4878869478707202e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3638402910000877
        },
        {
          "epoch": 192,
          "updates": 3072,
          "samples_seen": 196608,
          "online_loss": 0.00017460730668972246,
          "train_loss": 0.0001725928159430623,
          "validation_loss": 0.00013374029367696494,
          "train_metric": 1.7382915757480077e-05,
          "validation_metric": 1.346983299299609e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.374529384998823
        },
        {
          "epoch": 193,
          "updates": 3088,
          "samples_seen": 197632,
          "online_loss": 0.00017643055889493553,
          "train_loss": 0.0001713516830932349,
          "validation_loss": 0.00013425358338281512,
          "train_metric": 1.7257912986679003e-05,
          "validation_metric": 1.3521529581339564e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.3851556469999196
        },
        {
          "epoch": 194,
          "updates": 3104,
          "samples_seen": 198656,
          "online_loss": 0.00017692869596430683,
          "train_loss": 0.00017892596952151507,
          "validation_loss": 0.00013431080151349306,
          "train_metric": 1.8020768038695678e-05,
          "validation_metric": 1.3527291230275296e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3958586569988256
        },
        {
          "epoch": 195,
          "updates": 3120,
          "samples_seen": 199680,
          "online_loss": 0.00019217124508941197,
          "train_loss": 0.00016558512288611382,
          "validation_loss": 0.00012956543650943786,
          "train_metric": 1.667712422204204e-05,
          "validation_metric": 1.3049355402472429e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.407168947000173
        },
        {
          "epoch": 196,
          "updates": 3136,
          "samples_seen": 200704,
          "online_loss": 0.00017881164785649162,
          "train_loss": 0.0001659259869484231,
          "validation_loss": 0.00013552335440181196,
          "train_metric": 1.6711457647033967e-05,
          "validation_metric": 1.3649416359839961e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4178642199985916
        },
        {
          "epoch": 197,
          "updates": 3152,
          "samples_seen": 201728,
          "online_loss": 0.00018966237621498294,
          "train_loss": 0.00016906450036913157,
          "validation_loss": 0.00013355855480767787,
          "train_metric": 1.702755798760336e-05,
          "validation_metric": 1.3451529412122909e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4286357299988595
        },
        {
          "epoch": 198,
          "updates": 3168,
          "samples_seen": 202752,
          "online_loss": 0.00016862961774677387,
          "train_loss": 0.00016443189815618098,
          "validation_loss": 0.0001334072439931333,
          "train_metric": 1.6560979929636233e-05,
          "validation_metric": 1.34362890094053e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.439242757998727
        },
        {
          "epoch": 199,
          "updates": 3184,
          "samples_seen": 203776,
          "online_loss": 0.00016909239184315084,
          "train_loss": 0.00016972127195913345,
          "validation_loss": 0.00012832667562179267,
          "train_metric": 1.7093705537263304e-05,
          "validation_metric": 1.2924592738272622e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4506431959998736
        },
        {
          "epoch": 200,
          "updates": 3200,
          "samples_seen": 204800,
          "online_loss": 0.00017533401933178538,
          "train_loss": 0.0001584481360623613,
          "validation_loss": 0.00012432810035534203,
          "train_metric": 1.5958315998432226e-05,
          "validation_metric": 1.2521870303316973e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4623219529985363
        },
        {
          "epoch": 201,
          "updates": 3216,
          "samples_seen": 205824,
          "online_loss": 0.00016807464180601528,
          "train_loss": 0.000158283204655163,
          "validation_loss": 0.00012784670980181545,
          "train_metric": 1.5941703168209642e-05,
          "validation_metric": 1.2876251275883988e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.473026079000192
        },
        {
          "epoch": 202,
          "updates": 3232,
          "samples_seen": 206848,
          "online_loss": 0.00017243929096366628,
          "train_loss": 0.00015899143181741238,
          "validation_loss": 0.00012511966633610427,
          "train_metric": 1.601303301868029e-05,
          "validation_metric": 1.2601594789884984e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4836038119992736
        },
        {
          "epoch": 203,
          "updates": 3248,
          "samples_seen": 207872,
          "online_loss": 0.000182997592673928,
          "train_loss": 0.00016827345825731754,
          "validation_loss": 0.00014098503743298352,
          "train_metric": 1.694788625172805e-05,
          "validation_metric": 1.4199496945366263e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5554799339988676
        },
        {
          "epoch": 204,
          "updates": 3264,
          "samples_seen": 208896,
          "online_loss": 0.00016298369655487477,
          "train_loss": 0.00016571949527133256,
          "validation_loss": 0.00012831522326450795,
          "train_metric": 1.6690659322193824e-05,
          "validation_metric": 1.2923439498990774e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.566886179998619
        },
        {
          "epoch": 205,
          "updates": 3280,
          "samples_seen": 209920,
          "online_loss": 0.0001905512876874127,
          "train_loss": 0.00015954556874930859,
          "validation_loss": 0.00012304174015298486,
          "train_metric": 1.606884688953869e-05,
          "validation_metric": 1.2392314602038823e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5779752700000245
        },
        {
          "epoch": 206,
          "updates": 3296,
          "samples_seen": 210944,
          "online_loss": 0.00017595411463844357,
          "train_loss": 0.0001748851063894108,
          "validation_loss": 0.00014834103058092296,
          "train_metric": 1.7613787349546328e-05,
          "validation_metric": 1.494036587246228e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.588771506998455
        },
        {
          "epoch": 207,
          "updates": 3312,
          "samples_seen": 211968,
          "online_loss": 0.00016647917982481886,
          "train_loss": 0.00016688248433638364,
          "validation_loss": 0.0001255495590157807,
          "train_metric": 1.6807791325845756e-05,
          "validation_metric": 1.2644890375668183e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.599542941999971
        },
        {
          "epoch": 208,
          "updates": 3328,
          "samples_seen": 212992,
          "online_loss": 0.00016493133443873376,
          "train_loss": 0.0001552967878524214,
          "validation_loss": 0.00012152847921242937,
          "train_metric": 1.564092599437572e-05,
          "validation_metric": 1.223990329890512e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.610385017000226
        },
        {
          "epoch": 209,
          "updates": 3344,
          "samples_seen": 214016,
          "online_loss": 0.00016110483875309,
          "train_loss": 0.0001626606099307537,
          "validation_loss": 0.00013521104119718075,
          "train_metric": 1.6382578905904666e-05,
          "validation_metric": 1.361795966658974e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.621351013000094
        },
        {
          "epoch": 210,
          "updates": 3360,
          "samples_seen": 215040,
          "online_loss": 0.00016141188416440855,
          "train_loss": 0.0001602611446287483,
          "validation_loss": 0.00012307052384130657,
          "train_metric": 1.6140915249707177e-05,
          "validation_metric": 1.2395212252158672e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6319500659992627
        },
        {
          "epoch": 211,
          "updates": 3376,
          "samples_seen": 216064,
          "online_loss": 0.0001674772738624597,
          "train_loss": 0.00016952371515799314,
          "validation_loss": 0.0001382957852911204,
          "train_metric": 1.707380943116732e-05,
          "validation_metric": 1.3928645785199478e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.642290019999564
        },
        {
          "epoch": 212,
          "updates": 3392,
          "samples_seen": 217088,
          "online_loss": 0.0001708744503048365,
          "train_loss": 0.00016218487871810794,
          "validation_loss": 0.00013251231575850397,
          "train_metric": 1.6334666725015268e-05,
          "validation_metric": 1.3346155355975498e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6527500829997734
        },
        {
          "epoch": 213,
          "updates": 3408,
          "samples_seen": 218112,
          "online_loss": 0.0001611292541383591,
          "train_loss": 0.00016285968013107777,
          "validation_loss": 0.00012247220729477704,
          "train_metric": 1.6402631445089355e-05,
          "validation_metric": 1.23349500427139e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.663375547999749
        },
        {
          "epoch": 214,
          "updates": 3424,
          "samples_seen": 219136,
          "online_loss": 0.0001695626538094075,
          "train_loss": 0.00016950479766819626,
          "validation_loss": 0.00012822796998079866,
          "train_metric": 1.7071903130272403e-05,
          "validation_metric": 1.2914651961182244e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6737057659993297
        },
        {
          "epoch": 215,
          "updates": 3440,
          "samples_seen": 220160,
          "online_loss": 0.0001626910743652843,
          "train_loss": 0.00015230618009809405,
          "validation_loss": 0.00012143290950916708,
          "train_metric": 1.5339721358031966e-05,
          "validation_metric": 1.2230277206981555e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.684509132999665
        },
        {
          "epoch": 216,
          "updates": 3456,
          "samples_seen": 221184,
          "online_loss": 0.00015857098787819268,
          "train_loss": 0.00024980303714983165,
          "validation_loss": 0.0002021436666836962,
          "train_metric": 2.5159244614769705e-05,
          "validation_metric": 2.0359169866424054e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.694898050998745
        },
        {
          "epoch": 217,
          "updates": 3472,
          "samples_seen": 222208,
          "online_loss": 0.00021555122066274635,
          "train_loss": 0.00015777362568769604,
          "validation_loss": 0.00012964487541466951,
          "train_metric": 1.5890382201178e-05,
          "validation_metric": 1.3057355317869224e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7055125319984654
        },
        {
          "epoch": 218,
          "updates": 3488,
          "samples_seen": 223232,
          "online_loss": 0.00015790864654263714,
          "train_loss": 0.00016161578241735697,
          "validation_loss": 0.0001232639333466068,
          "train_metric": 1.627734673093073e-05,
          "validation_metric": 1.2414690900186542e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7160525759991287
        },
        {
          "epoch": 219,
          "updates": 3504,
          "samples_seen": 224256,
          "online_loss": 0.0001570606000314001,
          "train_loss": 0.00016425146895926446,
          "validation_loss": 0.00013165062409825623,
          "train_metric": 1.6542806406505406e-05,
          "validation_metric": 1.325936864304822e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7271476659989276
        },
        {
          "epoch": 220,
          "updates": 3520,
          "samples_seen": 225280,
          "online_loss": 0.00016490133975821664,
          "train_loss": 0.00015115179121494293,
          "validation_loss": 0.0001221905113197863,
          "train_metric": 1.5223455193336122e-05,
          "validation_metric": 1.2306580174481496e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7377810639991367
        },
        {
          "epoch": 221,
          "updates": 3536,
          "samples_seen": 226304,
          "online_loss": 0.000160162423071597,
          "train_loss": 0.00015513069229200482,
          "validation_loss": 0.00012186459207441658,
          "train_metric": 1.562419674883131e-05,
          "validation_metric": 1.227375560119981e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7481923879986425
        },
        {
          "epoch": 222,
          "updates": 3552,
          "samples_seen": 227328,
          "online_loss": 0.0001593610022609937,
          "train_loss": 0.00015554016863461584,
          "validation_loss": 0.00012870540376752615,
          "train_metric": 1.56654368765885e-05,
          "validation_metric": 1.2962736946064979e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7588657950000197
        },
        {
          "epoch": 223,
          "updates": 3568,
          "samples_seen": 228352,
          "online_loss": 0.00015241781784425257,
          "train_loss": 0.00016039714682847261,
          "validation_loss": 0.00012112852709833533,
          "train_metric": 1.6154612239915878e-05,
          "validation_metric": 1.2199621778563596e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.769833207999909
        },
        {
          "epoch": 224,
          "updates": 3584,
          "samples_seen": 229376,
          "online_loss": 0.00017112223486037692,
          "train_loss": 0.00014896635548211634,
          "validation_loss": 0.00011947507300646976,
          "train_metric": 1.500334565207595e-05,
          "validation_metric": 1.2033092389174271e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7809178859988606
        },
        {
          "epoch": 225,
          "updates": 3600,
          "samples_seen": 230400,
          "online_loss": 0.00016251865099548013,
          "train_loss": 0.00015701024676673114,
          "validation_loss": 0.00012960960157215595,
          "train_metric": 1.5813497157068923e-05,
          "validation_metric": 1.3053806469542906e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7914688579985523
        },
        {
          "epoch": 226,
          "updates": 3616,
          "samples_seen": 231424,
          "online_loss": 0.00016824578278828994,
          "train_loss": 0.00015077809803187847,
          "validation_loss": 0.0001231583155458793,
          "train_metric": 1.5185817574092653e-05,
          "validation_metric": 1.2404054359649308e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.801999748999151
        },
        {
          "epoch": 227,
          "updates": 3632,
          "samples_seen": 232448,
          "online_loss": 0.00015441689492945443,
          "train_loss": 0.00015437767433468252,
          "validation_loss": 0.00012115563004044816,
          "train_metric": 1.554835580463987e-05,
          "validation_metric": 1.2202352081658319e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.81253433099846
        },
        {
          "epoch": 228,
          "updates": 3648,
          "samples_seen": 233472,
          "online_loss": 0.000163048504418839,
          "train_loss": 0.00015375835937447846,
          "validation_loss": 0.0001210931659443304,
          "train_metric": 1.5485979020013474e-05,
          "validation_metric": 1.2196062016300857e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8229456249991927
        },
        {
          "epoch": 229,
          "updates": 3664,
          "samples_seen": 234496,
          "online_loss": 0.00015593782200085116,
          "train_loss": 0.00015166889352258295,
          "validation_loss": 0.00012721148959826678,
          "train_metric": 1.5275534678949043e-05,
          "validation_metric": 1.2812275599571876e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8334856199999194
        },
        {
          "epoch": 230,
          "updates": 3680,
          "samples_seen": 235520,
          "online_loss": 0.00015730078894193866,
          "train_loss": 0.00015350633475463837,
          "validation_loss": 0.00012577083543874323,
          "train_metric": 1.546059502288699e-05,
          "validation_metric": 1.2667177543335129e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8438392089992703
        },
        {
          "epoch": 231,
          "updates": 3696,
          "samples_seen": 236544,
          "online_loss": 0.00016226153366005747,
          "train_loss": 0.00015488246572203934,
          "validation_loss": 0.0001299453288083896,
          "train_metric": 1.5599196558468975e-05,
          "validation_metric": 1.3087616935081314e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.854273432998525
        },
        {
          "epoch": 232,
          "updates": 3712,
          "samples_seen": 237568,
          "online_loss": 0.0001598654298504698,
          "train_loss": 0.00016794048133306205,
          "validation_loss": 0.00012635930033866316,
          "train_metric": 1.691435136308428e-05,
          "validation_metric": 1.2726444765576161e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8647497999991174
        },
        {
          "epoch": 233,
          "updates": 3728,
          "samples_seen": 238592,
          "online_loss": 0.00016343071138180676,
          "train_loss": 0.00015008437912911177,
          "validation_loss": 0.00011695097055053338,
          "train_metric": 1.5115949281607755e-05,
          "validation_metric": 1.1778873158618808e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8759940569998435
        },
        {
          "epoch": 234,
          "updates": 3744,
          "samples_seen": 239616,
          "online_loss": 0.00015241343817251618,
          "train_loss": 0.00015678480849601328,
          "validation_loss": 0.00013251737982500345,
          "train_metric": 1.579079071234446e-05,
          "validation_metric": 1.334666467300849e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8866619509990414
        },
        {
          "epoch": 235,
          "updates": 3760,
          "samples_seen": 240640,
          "online_loss": 0.00014189225794325466,
          "train_loss": 0.0002042912965407595,
          "validation_loss": 0.000143897530506365,
          "train_metric": 2.05754695343785e-05,
          "validation_metric": 1.449283354304498e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8969558979988506
        },
        {
          "epoch": 236,
          "updates": 3776,
          "samples_seen": 241664,
          "online_loss": 0.00021196849229454529,
          "train_loss": 0.00017813600425142795,
          "validation_loss": 0.0001434299920219928,
          "train_metric": 1.794120362319518e-05,
          "validation_metric": 1.4445743545365985e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9076341959989804
        },
        {
          "epoch": 237,
          "updates": 3792,
          "samples_seen": 242688,
          "online_loss": 0.00019249250999564538,
          "train_loss": 0.0001446689711883664,
          "validation_loss": 0.0001207150926347822,
          "train_metric": 1.4570528946933337e-05,
          "validation_metric": 1.2157983292127028e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.918190809999942
        },
        {
          "epoch": 238,
          "updates": 3808,
          "samples_seen": 243712,
          "online_loss": 0.00017839437487054965,
          "train_loss": 0.00015472497034352273,
          "validation_loss": 0.00012705284461844712,
          "train_metric": 1.5583333151880652e-05,
          "validation_metric": 1.2796296687156428e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.92858598499879
        },
        {
          "epoch": 239,
          "updates": 3824,
          "samples_seen": 244736,
          "online_loss": 0.00015269751429514145,
          "train_loss": 0.00015847105532884598,
          "validation_loss": 0.00013174554624129087,
          "train_metric": 1.596062611497473e-05,
          "validation_metric": 1.3268928341858555e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.939771921999636
        },
        {
          "epoch": 240,
          "updates": 3840,
          "samples_seen": 245760,
          "online_loss": 0.00015213133656288846,
          "train_loss": 0.00019425204664003104,
          "validation_loss": 0.00014245306374505162,
          "train_metric": 1.9564355170587078e-05,
          "validation_metric": 1.4347353499033488e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9502654049993
        },
        {
          "epoch": 241,
          "updates": 3856,
          "samples_seen": 246784,
          "online_loss": 0.0001904436135191645,
          "train_loss": 0.00016926544776652008,
          "validation_loss": 0.00013239558029454201,
          "train_metric": 1.7047794244717807e-05,
          "validation_metric": 1.3334398317965679e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.960673728999609
        },
        {
          "epoch": 242,
          "updates": 3872,
          "samples_seen": 247808,
          "online_loss": 0.00015411992626468418,
          "train_loss": 0.0001492748415330425,
          "validation_loss": 0.00012632975995074958,
          "train_metric": 1.5034415810077917e-05,
          "validation_metric": 1.2723469808406662e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9711002449985244
        },
        {
          "epoch": 243,
          "updates": 3888,
          "samples_seen": 248832,
          "online_loss": 0.00015034173429739894,
          "train_loss": 0.00016075866005849093,
          "validation_loss": 0.00012458190030883998,
          "train_metric": 1.6191022950806655e-05,
          "validation_metric": 1.2547432561405003e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.981455167999229
        },
        {
          "epoch": 244,
          "updates": 3904,
          "samples_seen": 249856,
          "online_loss": 0.00017096970941565814,
          "train_loss": 0.0001619139948161319,
          "validation_loss": 0.00012583975330926478,
          "train_metric": 1.6307383702951483e-05,
          "validation_metric": 1.2674120625888463e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9923768879998534
        },
        {
          "epoch": 245,
          "updates": 3920,
          "samples_seen": 250880,
          "online_loss": 0.00014667651794297853,
          "train_loss": 0.00016849774692673236,
          "validation_loss": 0.00014719455793965608,
          "train_metric": 1.6970474462141283e-05,
          "validation_metric": 1.4824895515630487e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.002637348999997
        },
        {
          "epoch": 246,
          "updates": 3936,
          "samples_seen": 251904,
          "online_loss": 0.00015449430566150113,
          "train_loss": 0.00015116820577532053,
          "validation_loss": 0.00011931882909266278,
          "train_metric": 1.5225107745209243e-05,
          "validation_metric": 1.2017355402349494e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.013757822998741
        },
        {
          "epoch": 247,
          "updates": 3952,
          "samples_seen": 252928,
          "online_loss": 0.00016086904224721366,
          "train_loss": 0.00014754265430383384,
          "validation_loss": 0.0001221745478687808,
          "train_metric": 1.4859955626889132e-05,
          "validation_metric": 1.2304974006838165e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0244625489995087
        },
        {
          "epoch": 248,
          "updates": 3968,
          "samples_seen": 253952,
          "online_loss": 0.00014649258400822873,
          "train_loss": 0.0001401475165039301,
          "validation_loss": 0.00011420431110309437,
          "train_metric": 1.4115146768745035e-05,
          "validation_metric": 1.150223943113815e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.035254177999377
        },
        {
          "epoch": 249,
          "updates": 3984,
          "samples_seen": 254976,
          "online_loss": 0.00014372996247402625,
          "train_loss": 0.0001395242434227839,
          "validation_loss": 0.00011414211621740833,
          "train_metric": 1.4052370715944562e-05,
          "validation_metric": 1.149597574112704e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0463938169996254
        },
        {
          "epoch": 250,
          "updates": 4000,
          "samples_seen": 256000,
          "online_loss": 0.00014502647127301316,
          "train_loss": 0.00014227819337975234,
          "validation_loss": 0.00011933577479794621,
          "train_metric": 1.4329739315144252e-05,
          "validation_metric": 1.2019060704915319e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0571168169990415
        },
        {
          "epoch": 251,
          "updates": 4016,
          "samples_seen": 257024,
          "online_loss": 0.00014490956436929991,
          "train_loss": 0.00013957818737253547,
          "validation_loss": 0.00011298245954094455,
          "train_metric": 1.4057804037292954e-05,
          "validation_metric": 1.1379180250514764e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.068546239999705
        },
        {
          "epoch": 252,
          "updates": 4032,
          "samples_seen": 258048,
          "online_loss": 0.00015026821984065464,
          "train_loss": 0.00014126377936918288,
          "validation_loss": 0.0001132687320932746,
          "train_metric": 1.422757122782059e-05,
          "validation_metric": 1.1408012142055668e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0792193899997073
        },
        {
          "epoch": 253,
          "updates": 4048,
          "samples_seen": 259072,
          "online_loss": 0.0001465811374146142,
          "train_loss": 0.00014323853247333318,
          "validation_loss": 0.0001209393740282394,
          "train_metric": 1.4426461348193698e-05,
          "validation_metric": 1.2180569683550857e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0895860669988906
        },
        {
          "epoch": 254,
          "updates": 4064,
          "samples_seen": 260096,
          "online_loss": 0.00014473692817773554,
          "train_loss": 0.00013872995623387396,
          "validation_loss": 0.00011294891010038555,
          "train_metric": 1.3972375199955422e-05,
          "validation_metric": 1.1375799658708274e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1004722469988337
        },
        {
          "epoch": 255,
          "updates": 4080,
          "samples_seen": 261120,
          "online_loss": 0.00014251713400881272,
          "train_loss": 0.00013828137889504433,
          "validation_loss": 0.00011400445509934798,
          "train_metric": 1.3927196050644852e-05,
          "validation_metric": 1.1482110494398512e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.110948637999172
        },
        {
          "epoch": 256,
          "updates": 4096,
          "samples_seen": 262144,
          "online_loss": 0.00014121386766419164,
          "train_loss": 0.0001384658389724791,
          "validation_loss": 0.00011529652692843229,
          "train_metric": 1.3945774298917968e-05,
          "validation_metric": 1.161224463430699e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.121409319999657
        },
        {
          "epoch": 257,
          "updates": 4112,
          "samples_seen": 263168,
          "online_loss": 0.00014410590620173025,
          "train_loss": 0.00015850580530241132,
          "validation_loss": 0.00012018813868053257,
          "train_metric": 1.596412403159775e-05,
          "validation_metric": 1.2104907909815665e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.13164893999965
        },
        {
          "epoch": 258,
          "updates": 4128,
          "samples_seen": 264192,
          "online_loss": 0.00015016438783277408,
          "train_loss": 0.00014074917999096215,
          "validation_loss": 0.00011781507055275142,
          "train_metric": 1.4175741853250656e-05,
          "validation_metric": 1.1865901797136758e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1420527499994932
        },
        {
          "epoch": 259,
          "updates": 4144,
          "samples_seen": 265216,
          "online_loss": 0.00014033798697710154,
          "train_loss": 0.00013742521696258336,
          "validation_loss": 0.00011333231668686494,
          "train_metric": 1.3840965038980357e-05,
          "validation_metric": 1.141441498475615e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.152358130999346
        },
        {
          "epoch": 260,
          "updates": 4160,
          "samples_seen": 266240,
          "online_loss": 0.00015493000910282717,
          "train_loss": 0.0001535312767373398,
          "validation_loss": 0.00012971414253115654,
          "train_metric": 1.5463108866242692e-05,
          "validation_metric": 1.3064334780210629e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1626335900000413
        },
        {
          "epoch": 261,
          "updates": 4176,
          "samples_seen": 267264,
          "online_loss": 0.00014479921856036526,
          "train_loss": 0.00014193460810929537,
          "validation_loss": 0.00011227439972572029,
          "train_metric": 1.4295133951236494e-05,
          "validation_metric": 1.1307865861454047e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.173501405999559
        },
        {
          "epoch": 262,
          "updates": 4192,
          "samples_seen": 268288,
          "online_loss": 0.00014164244385028724,
          "train_loss": 0.00013751753431279212,
          "validation_loss": 0.00011322373757138848,
          "train_metric": 1.385026189382188e-05,
          "validation_metric": 1.1403480129956733e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1839214249994257
        },
        {
          "epoch": 263,
          "updates": 4208,
          "samples_seen": 269312,
          "online_loss": 0.0001390476386404771,
          "train_loss": 0.00013651506742462516,
          "validation_loss": 0.00011141305003548041,
          "train_metric": 1.3749297977483366e-05,
          "validation_metric": 1.1221114618820138e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1948138209991157
        },
        {
          "epoch": 264,
          "updates": 4224,
          "samples_seen": 270336,
          "online_loss": 0.00014613789562645252,
          "train_loss": 0.0001441218046238646,
          "validation_loss": 0.00012422713916748762,
          "train_metric": 1.4515419024974108e-05,
          "validation_metric": 1.2511703062045854e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.205165218998445
        },
        {
          "epoch": 265,
          "updates": 4240,
          "samples_seen": 271360,
          "online_loss": 0.00015115717724256683,
          "train_loss": 0.00013674104411620647,
          "validation_loss": 0.00011197617277503014,
          "train_metric": 1.377205626340583e-05,
          "validation_metric": 1.1277828889433295e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2154537299993535
        },
        {
          "epoch": 266,
          "updates": 4256,
          "samples_seen": 272384,
          "online_loss": 0.0001450722552363004,
          "train_loss": 0.00013965056859888136,
          "validation_loss": 0.00011147263285238296,
          "train_metric": 1.4065094546822365e-05,
          "validation_metric": 1.1227114555367734e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2258176159994036
        },
        {
          "epoch": 267,
          "updates": 4272,
          "samples_seen": 273408,
          "online_loss": 0.0001423063479251141,
          "train_loss": 0.00013583367399405688,
          "validation_loss": 0.00011231889948248863,
          "train_metric": 1.3680670235771686e-05,
          "validation_metric": 1.1312347851344384e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2361106919997837
        },
        {
          "epoch": 268,
          "updates": 4288,
          "samples_seen": 274432,
          "online_loss": 0.00014962260320317,
          "train_loss": 0.0001399939792463556,
          "validation_loss": 0.00011644679761957377,
          "train_metric": 1.4099681720836088e-05,
          "validation_metric": 1.1728094250429422e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2462714279990905
        },
        {
          "epoch": 269,
          "updates": 4304,
          "samples_seen": 275456,
          "online_loss": 0.00014200556643118034,
          "train_loss": 0.00013516476610675454,
          "validation_loss": 0.0001113905746024102,
          "train_metric": 1.3613300325232558e-05,
          "validation_metric": 1.1218851796002127e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.257284133000212
        },
        {
          "epoch": 270,
          "updates": 4320,
          "samples_seen": 276480,
          "online_loss": 0.00014455318660111516,
          "train_loss": 0.00013767016935162246,
          "validation_loss": 0.00011204795737285167,
          "train_metric": 1.386563599226065e-05,
          "validation_metric": 1.1285061191301793e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2676949609995063
        },
        {
          "epoch": 271,
          "updates": 4336,
          "samples_seen": 277504,
          "online_loss": 0.0001484775038989028,
          "train_loss": 0.00013516232138499618,
          "validation_loss": 0.0001117814244935289,
          "train_metric": 1.3613053852168377e-05,
          "validation_metric": 1.125821472669486e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.277942257000177
        },
        {
          "epoch": 272,
          "updates": 4352,
          "samples_seen": 278528,
          "online_loss": 0.00014940226310500293,
          "train_loss": 0.00014775512681808323,
          "validation_loss": 0.00011814082972705364,
          "train_metric": 1.4881355127727147e-05,
          "validation_metric": 1.1898710909008514e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2884035739989486
        },
        {
          "epoch": 273,
          "updates": 4368,
          "samples_seen": 279552,
          "online_loss": 0.00013915538556830143,
          "train_loss": 0.000140376461786218,
          "validation_loss": 0.00011238725710427389,
          "train_metric": 1.4138202459434979e-05,
          "validation_metric": 1.1319233635731507e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2986194169989176
        },
        {
          "epoch": 274,
          "updates": 4384,
          "samples_seen": 280576,
          "online_loss": 0.000165505981840397,
          "train_loss": 0.00013461073103826493,
          "validation_loss": 0.00011027274013031274,
          "train_metric": 1.3557499187299982e-05,
          "validation_metric": 1.1106268175353762e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.309446630999446
        },
        {
          "epoch": 275,
          "updates": 4400,
          "samples_seen": 281600,
          "online_loss": 0.00016086143159554922,
          "train_loss": 0.00015204295050352812,
          "validation_loss": 0.00013236593804322183,
          "train_metric": 1.5313209587475285e-05,
          "validation_metric": 1.333141335635446e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.319977156999812
        },
        {
          "epoch": 276,
          "updates": 4416,
          "samples_seen": 282624,
          "online_loss": 0.00015379932801806717,
          "train_loss": 0.00013864149514120072,
          "validation_loss": 0.00011065570288337767,
          "train_metric": 1.3963464880362153e-05,
          "validation_metric": 1.1144836207677145e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.330444841998542
        },
        {
          "epoch": 277,
          "updates": 4432,
          "samples_seen": 283648,
          "online_loss": 0.00013846784395354916,
          "train_loss": 0.0001342454634141177,
          "validation_loss": 0.00011067427112720907,
          "train_metric": 1.3520712855097372e-05,
          "validation_metric": 1.1146707947773393e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3409689489999437
        },
        {
          "epoch": 278,
          "updates": 4448,
          "samples_seen": 284672,
          "online_loss": 0.00013807988602820842,
          "train_loss": 0.00013510875578504056,
          "validation_loss": 0.00011523431749083102,
          "train_metric": 1.3607659639092162e-05,
          "validation_metric": 1.1605980034801178e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.351353278998431
        },
        {
          "epoch": 279,
          "updates": 4464,
          "samples_seen": 285696,
          "online_loss": 0.0001392753069922037,
          "train_loss": 0.00013360947195906192,
          "validation_loss": 0.00011379043280612677,
          "train_metric": 1.3456658052746207e-05,
          "validation_metric": 1.1460555469966494e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3622881619994587
        },
        {
          "epoch": 280,
          "updates": 4480,
          "samples_seen": 286720,
          "online_loss": 0.0001359792659059167,
          "train_loss": 0.00013339577708393335,
          "validation_loss": 0.00011008049477823079,
          "train_metric": 1.3435132132144645e-05,
          "validation_metric": 1.1086905033153016e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3731886169989593
        },
        {
          "epoch": 281,
          "updates": 4496,
          "samples_seen": 287744,
          "online_loss": 0.0001377666535518074,
          "train_loss": 0.00013252950157038867,
          "validation_loss": 0.00010914961603702977,
          "train_metric": 1.3347886124392971e-05,
          "validation_metric": 1.0993149771820754e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3841401599984238
        },
        {
          "epoch": 282,
          "updates": 4512,
          "samples_seen": 288768,
          "online_loss": 0.0001348331470580888,
          "train_loss": 0.0001488279813202098,
          "validation_loss": 0.00011757273023249581,
          "train_metric": 1.4989410374255385e-05,
          "validation_metric": 1.1841495506814681e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3946209179994185
        },
        {
          "epoch": 283,
          "updates": 4528,
          "samples_seen": 289792,
          "online_loss": 0.00014001043427924742,
          "train_loss": 0.00015220881323330104,
          "validation_loss": 0.00013208248128648847,
          "train_metric": 1.532991518615745e-05,
          "validation_metric": 1.3302864317665808e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4050266699996428
        },
        {
          "epoch": 284,
          "updates": 4544,
          "samples_seen": 290816,
          "online_loss": 0.00015345648671427625,
          "train_loss": 0.0001355136773781851,
          "validation_loss": 0.00010830035898834467,
          "train_metric": 1.364844138151966e-05,
          "validation_metric": 1.0907615433097817e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.41583442099909
        },
        {
          "epoch": 285,
          "updates": 4560,
          "samples_seen": 291840,
          "online_loss": 0.00014417764941754285,
          "train_loss": 0.00015053828246891499,
          "validation_loss": 0.00013411187683232129,
          "train_metric": 1.516166503279237e-05,
          "validation_metric": 1.3507258699974045e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4264084930000536
        },
        {
          "epoch": 286,
          "updates": 4576,
          "samples_seen": 292864,
          "online_loss": 0.00014370754979609046,
          "train_loss": 0.0001324477925663814,
          "validation_loss": 0.0001080810179701075,
          "train_metric": 1.3339656106836628e-05,
          "validation_metric": 1.0885525625781156e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.437299274999532
        },
        {
          "epoch": 287,
          "updates": 4592,
          "samples_seen": 293888,
          "online_loss": 0.00013490470200849813,
          "train_loss": 0.00013322080485522747,
          "validation_loss": 0.00011118157271994278,
          "train_metric": 1.3417511581792496e-05,
          "validation_metric": 1.1197800631634891e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.447797177999746
        },
        {
          "epoch": 288,
          "updates": 4608,
          "samples_seen": 294912,
          "online_loss": 0.00013622440565086436,
          "train_loss": 0.00013350190420169383,
          "validation_loss": 0.00011414803157094866,
          "train_metric": 1.344582506135339e-05,
          "validation_metric": 1.1496570550662e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.458370500999081
        },
        {
          "epoch": 289,
          "updates": 4624,
          "samples_seen": 295936,
          "online_loss": 0.00013979328286950476,
          "train_loss": 0.00014373894373420626,
          "validation_loss": 0.00011274602002231404,
          "train_metric": 1.4476860087597743e-05,
          "validation_metric": 1.1355366950738244e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4687176480001654
        },
        {
          "epoch": 290,
          "updates": 4640,
          "samples_seen": 296960,
          "online_loss": 0.00013640696533911978,
          "train_loss": 0.00013496384781319648,
          "validation_loss": 0.00011543649452505633,
          "train_metric": 1.3593064977612812e-05,
          "validation_metric": 1.1626339983195066e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4789855199996964
        },
        {
          "epoch": 291,
          "updates": 4656,
          "samples_seen": 297984,
          "online_loss": 0.00013749026129517006,
          "train_loss": 0.00013379815209191293,
          "validation_loss": 0.00010906186798820272,
          "train_metric": 1.347566103504505e-05,
          "validation_metric": 1.0984311302308924e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.48952311299945
        },
        {
          "epoch": 292,
          "updates": 4672,
          "samples_seen": 299008,
          "online_loss": 0.00014342918893817114,
          "train_loss": 0.00013246152957435697,
          "validation_loss": 0.00010960525105474517,
          "train_metric": 1.3341040357772727e-05,
          "validation_metric": 1.1039039236493409e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4998081119993003
        },
        {
          "epoch": 293,
          "updates": 4688,
          "samples_seen": 300032,
          "online_loss": 0.00013921233176006353,
          "train_loss": 0.00013193002087064087,
          "validation_loss": 0.00010929667041637003,
          "train_metric": 1.3287509318615776e-05,
          "validation_metric": 1.1007959983544424e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.510393455999292
        },
        {
          "epoch": 294,
          "updates": 4704,
          "samples_seen": 301056,
          "online_loss": 0.0001479504107919638,
          "train_loss": 0.00013337898417375982,
          "validation_loss": 0.00010994198237312958,
          "train_metric": 1.3433440471999347e-05,
          "validation_metric": 1.107295338442782e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5210404049994395
        },
        {
          "epoch": 295,
          "updates": 4720,
          "samples_seen": 302080,
          "online_loss": 0.00013581448047261802,
          "train_loss": 0.00013146127457730472,
          "validation_loss": 0.00010771527013275772,
          "train_metric": 1.3240300177130848e-05,
          "validation_metric": 1.0848687452380545e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5321419479987526
        },
        {
          "epoch": 296,
          "updates": 4736,
          "samples_seen": 303104,
          "online_loss": 0.0001423387298018497,
          "train_loss": 0.00013131360174156725,
          "validation_loss": 0.00011248284135945141,
          "train_metric": 1.322542448178865e-05,
          "validation_metric": 1.1328860637149774e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5427196780001395
        },
        {
          "epoch": 297,
          "updates": 4752,
          "samples_seen": 304128,
          "online_loss": 0.00015040523567222408,
          "train_loss": 0.00013858570309821516,
          "validation_loss": 0.00011798551713582128,
          "train_metric": 1.39578460220946e-05,
          "validation_metric": 1.1883068509632722e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5531665540001995
        },
        {
          "epoch": 298,
          "updates": 4768,
          "samples_seen": 305152,
          "online_loss": 0.0001396254456267343,
          "train_loss": 0.00013235313235782087,
          "validation_loss": 0.00011026868014596403,
          "train_metric": 1.3330121873877943e-05,
          "validation_metric": 1.1105858902737964e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5634963779993996
        },
        {
          "epoch": 299,
          "updates": 4784,
          "samples_seen": 306176,
          "online_loss": 0.00013680278971150983,
          "train_loss": 0.00015203829389065504,
          "validation_loss": 0.00011813676246674731,
          "train_metric": 1.5312736650230363e-05,
          "validation_metric": 1.1898301636392716e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.573770252000031
        },
        {
          "epoch": 300,
          "updates": 4800,
          "samples_seen": 307200,
          "online_loss": 0.00017408247367711738,
          "train_loss": 0.0001554375485284254,
          "validation_loss": 0.00014005755656398833,
          "train_metric": 1.565509955980815e-05,
          "validation_metric": 1.4106083654041868e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.583999053998923
        },
        {
          "epoch": 301,
          "updates": 4816,
          "samples_seen": 308224,
          "online_loss": 0.00014712693837282131,
          "train_loss": 0.00013467077224049717,
          "validation_loss": 0.00011627336061792448,
          "train_metric": 1.3563548236561473e-05,
          "validation_metric": 1.1710627404681873e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5944479729987506
        },
        {
          "epoch": 302,
          "updates": 4832,
          "samples_seen": 309248,
          "online_loss": 0.00013461962407745887,
          "train_loss": 0.0001298649440286681,
          "validation_loss": 0.00010654916695784777,
          "train_metric": 1.3079521522740833e-05,
          "validation_metric": 1.0731243492045905e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.605298671000128
        },
        {
          "epoch": 303,
          "updates": 4848,
          "samples_seen": 310272,
          "online_loss": 0.00013459916272040573,
          "train_loss": 0.00013545926776714623,
          "validation_loss": 0.00011723775969585404,
          "train_metric": 1.3642960766446777e-05,
          "validation_metric": 1.1807757800852414e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6157683469991753
        },
        {
          "epoch": 304,
          "updates": 4864,
          "samples_seen": 311296,
          "online_loss": 0.00013728713247473934,
          "train_loss": 0.00012853047519456595,
          "validation_loss": 0.00010732278315117583,
          "train_metric": 1.2945119124196935e-05,
          "validation_metric": 1.0809158084157389e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6263407199985522
        },
        {
          "epoch": 305,
          "updates": 4880,
          "samples_seen": 312320,
          "online_loss": 0.00013609295319838566,
          "train_loss": 0.0001293580571655184,
          "validation_loss": 0.00010668102913768962,
          "train_metric": 1.3028469766140915e-05,
          "validation_metric": 1.074452211469179e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6366305699993973
        },
        {
          "epoch": 306,
          "updates": 4896,
          "samples_seen": 313344,
          "online_loss": 0.00014375580440173508,
          "train_loss": 0.00015459985297638923,
          "validation_loss": 0.0001353532716166228,
          "train_metric": 1.5570731193292886e-05,
          "validation_metric": 1.3632285117637366e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6469812229988747
        },
        {
          "epoch": 307,
          "updates": 4912,
          "samples_seen": 314368,
          "online_loss": 0.00014779975799683598,
          "train_loss": 0.00012820246047340333,
          "validation_loss": 0.00010573776671662927,
          "train_metric": 1.2912081729155034e-05,
          "validation_metric": 1.0649519936123397e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6581594309991488
        },
        {
          "epoch": 308,
          "updates": 4928,
          "samples_seen": 315392,
          "online_loss": 0.00013815798638461274,
          "train_loss": 0.00012753964983858168,
          "validation_loss": 0.00010566713172011077,
          "train_metric": 1.2845325727539603e-05,
          "validation_metric": 1.0642404959071428e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.669183838999743
        },
        {
          "epoch": 309,
          "updates": 4944,
          "samples_seen": 316416,
          "online_loss": 0.00012927117268191068,
          "train_loss": 0.00012766430154442787,
          "validation_loss": 0.00010803467012010515,
          "train_metric": 1.2857879482908174e-05,
          "validation_metric": 1.0880857189476956e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6795508729992434
        },
        {
          "epoch": 310,
          "updates": 4960,
          "samples_seen": 317440,
          "online_loss": 0.00013303574360179482,
          "train_loss": 0.00014282739721238613,
          "validation_loss": 0.00011318765609757975,
          "train_metric": 1.438505387341138e-05,
          "validation_metric": 1.1399845789128449e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.690150085998539
        },
        {
          "epoch": 311,
          "updates": 4976,
          "samples_seen": 318464,
          "online_loss": 0.00014722480182172148,
          "train_loss": 0.0001273398520424962,
          "validation_loss": 0.00010694967932067811,
          "train_metric": 1.2825204066757578e-05,
          "validation_metric": 1.0771579582069535e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.700398580998808
        },
        {
          "epoch": 312,
          "updates": 4992,
          "samples_seen": 319488,
          "online_loss": 0.00013766935262538027,
          "train_loss": 0.00016868063539732248,
          "validation_loss": 0.00012769208115059882,
          "train_metric": 1.6988897186820395e-05,
          "validation_metric": 1.286067799810553e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7106248479994974
        },
        {
          "epoch": 313,
          "updates": 5008,
          "samples_seen": 320512,
          "online_loss": 0.00013904897241445724,
          "train_loss": 0.00016081779904197901,
          "validation_loss": 0.00013171593309380114,
          "train_metric": 1.6196980141103268e-05,
          "validation_metric": 1.3265946108731441e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7212505409988808
        },
        {
          "epoch": 314,
          "updates": 5024,
          "samples_seen": 321536,
          "online_loss": 0.00015720445071565337,
          "train_loss": 0.00015726889250800014,
          "validation_loss": 0.00012458574201446027,
          "train_metric": 1.58395450853277e-05,
          "validation_metric": 1.2547819096653257e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7316301989994827
        },
        {
          "epoch": 315,
          "updates": 5040,
          "samples_seen": 322560,
          "online_loss": 0.0001325855782852159,
          "train_loss": 0.00013439096801448613,
          "validation_loss": 0.00011263898340985179,
          "train_metric": 1.3535364814742934e-05,
          "validation_metric": 1.134458671003813e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.742163144999722
        },
        {
          "epoch": 316,
          "updates": 5056,
          "samples_seen": 323584,
          "online_loss": 0.00014936621892047697,
          "train_loss": 0.00015855113451834768,
          "validation_loss": 0.00014129998453427106,
          "train_metric": 1.5968691513990052e-05,
          "validation_metric": 1.4231215573090594e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7524189539999497
        },
        {
          "epoch": 317,
          "updates": 5072,
          "samples_seen": 324608,
          "online_loss": 0.00014385988924914273,
          "train_loss": 0.0001262440491700545,
          "validation_loss": 0.00010445749649079517,
          "train_metric": 1.2714839613181539e-05,
          "validation_metric": 1.0520577234274242e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7632903029989393
        },
        {
          "epoch": 318,
          "updates": 5088,
          "samples_seen": 325632,
          "online_loss": 0.000133085852212389,
          "train_loss": 0.00012689022696577013,
          "validation_loss": 0.0001062937080860138,
          "train_metric": 1.27799185065669e-05,
          "validation_metric": 1.0705514796427451e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.773674845999267
        },
        {
          "epoch": 319,
          "updates": 5104,
          "samples_seen": 326656,
          "online_loss": 0.00013148469997759094,
          "train_loss": 0.00013118298375047743,
          "validation_loss": 0.00011629170330706984,
          "train_metric": 1.3212270459916908e-05,
          "validation_metric": 1.1712475497915875e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.783947358999285
        },
        {
          "epoch": 320,
          "updates": 5120,
          "samples_seen": 327680,
          "online_loss": 0.00012674563868131372,
          "train_loss": 0.00012516495189629495,
          "validation_loss": 0.00010491567081771791,
          "train_metric": 1.2606155905814376e-05,
          "validation_metric": 1.0566722266958095e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7943388749990845
        },
        {
          "epoch": 321,
          "updates": 5136,
          "samples_seen": 328704,
          "online_loss": 0.00013046736603428144,
          "train_loss": 0.00013503441005013883,
          "validation_loss": 0.00011888458539033309,
          "train_metric": 1.3600170859717764e-05,
          "validation_metric": 1.197362053062534e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8046761309997237
        },
        {
          "epoch": 322,
          "updates": 5152,
          "samples_seen": 329728,
          "online_loss": 0.00016160468567250064,
          "train_loss": 0.00013303336163517088,
          "validation_loss": 0.00011431337043177336,
          "train_metric": 1.3398632290773094e-05,
          "validation_metric": 1.151322248915676e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8150466219994996
        },
        {
          "epoch": 323,
          "updates": 5168,
          "samples_seen": 330752,
          "online_loss": 0.00015590645853080787,
          "train_loss": 0.00012416455138009042,
          "validation_loss": 0.00010542281233938411,
          "train_metric": 1.2505397535278462e-05,
          "validation_metric": 1.0617798579914961e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8258170489989425
        },
        {
          "epoch": 324,
          "updates": 5184,
          "samples_seen": 331776,
          "online_loss": 0.00013260137757242774,
          "train_loss": 0.00012375970254652202,
          "validation_loss": 0.00010516624024603516,
          "train_metric": 1.2464622159313876e-05,
          "validation_metric": 1.059195892594289e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.83632349599975
        },
        {
          "epoch": 325,
          "updates": 5200,
          "samples_seen": 332800,
          "online_loss": 0.00013327802525964216,
          "train_loss": 0.0001263628073502332,
          "validation_loss": 0.00010390799434389919,
          "train_metric": 1.2726799468509853e-05,
          "validation_metric": 1.0465232662681956e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.847255716998916
        },
        {
          "epoch": 326,
          "updates": 5216,
          "samples_seen": 333824,
          "online_loss": 0.00012694164593085588,
          "train_loss": 0.00012408234761096537,
          "validation_loss": 0.00010405171633465216,
          "train_metric": 1.2497118405008223e-05,
          "validation_metric": 1.0479707270860672e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8581040179997217
        },
        {
          "epoch": 327,
          "updates": 5232,
          "samples_seen": 334848,
          "online_loss": 0.00013363018888412626,
          "train_loss": 0.00013272103387862444,
          "validation_loss": 0.00010727838525781408,
          "train_metric": 1.3367175597522873e-05,
          "validation_metric": 1.0804687008203473e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8688081909986067
        },
        {
          "epoch": 328,
          "updates": 5248,
          "samples_seen": 335872,
          "online_loss": 0.00013394319876169902,
          "train_loss": 0.00012685940600931644,
          "validation_loss": 0.0001106051349779591,
          "train_metric": 1.2776815310644452e-05,
          "validation_metric": 1.1139743946841918e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8792582049991324
        },
        {
          "epoch": 329,
          "updates": 5264,
          "samples_seen": 336896,
          "online_loss": 0.00012658056084546843,
          "train_loss": 0.0001221651618834585,
          "validation_loss": 0.00010328803909942508,
          "train_metric": 1.2304028132348321e-05,
          "validation_metric": 1.0402793122921139e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.891027980000217
        },
        {
          "epoch": 330,
          "updates": 5280,
          "samples_seen": 337920,
          "online_loss": 0.00012858947502536466,
          "train_loss": 0.00013406049401964992,
          "validation_loss": 0.00011816909682238474,
          "train_metric": 1.3502080946636852e-05,
          "validation_metric": 1.1901559446414467e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.9016640489990095
        },
        {
          "epoch": 331,
          "updates": 5296,
          "samples_seen": 338944,
          "online_loss": 0.00013059111142865731,
          "train_loss": 0.00012271171726752073,
          "validation_loss": 0.00010599127563182265,
          "train_metric": 1.2359075299173128e-05,
          "validation_metric": 1.0675053999875672e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.912542845999269
        },
        {
          "epoch": 332,
          "updates": 5312,
          "samples_seen": 339968,
          "online_loss": 0.0001296580335292674,
          "train_loss": 0.00012250967847649008,
          "validation_loss": 0.00010825719800777733,
          "train_metric": 1.2338724445726257e-05,
          "validation_metric": 1.0903268048423342e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.923310299998775
        },
        {
          "epoch": 333,
          "updates": 5328,
          "samples_seen": 340992,
          "online_loss": 0.0001258404522559431,
          "train_loss": 0.00012144810170866549,
          "validation_loss": 0.00010239728726446629,
          "train_metric": 1.2231807886564638e-05,
          "validation_metric": 1.0313079656043556e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.934509906999665
        },
        {
          "epoch": 334,
          "updates": 5344,
          "samples_seen": 342016,
          "online_loss": 0.00012961058109794976,
          "train_loss": 0.00013333032256923616,
          "validation_loss": 0.00011248952068854123,
          "train_metric": 1.3428541933535598e-05,
          "validation_metric": 1.1329532753734384e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.945317323999916
        },
        {
          "epoch": 335,
          "updates": 5360,
          "samples_seen": 343040,
          "online_loss": 0.00014095213759901526,
          "train_loss": 0.00012153003626735881,
          "validation_loss": 0.00010337918502045795,
          "train_metric": 1.2240060641488526e-05,
          "validation_metric": 1.0411973562440835e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.9559146789997612
        },
        {
          "epoch": 336,
          "updates": 5376,
          "samples_seen": 344064,
          "online_loss": 0.00013410455039775115,
          "train_loss": 0.0001263267477042973,
          "validation_loss": 0.00011398817878216505,
          "train_metric": 1.2723166946670972e-05,
          "validation_metric": 1.1480471584945917e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.96656133399847
        },
        {
          "epoch": 337,
          "updates": 5392,
          "samples_seen": 345088,
          "online_loss": 0.0001305175942434289,
          "train_loss": 0.00012248072016518563,
          "validation_loss": 0.00010258236579829827,
          "train_metric": 1.2335809515207075e-05,
          "validation_metric": 1.0331720659451094e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.9771756879999884
        },
        {
          "epoch": 338,
          "updates": 5408,
          "samples_seen": 346112,
          "online_loss": 0.00013801121531287208,
          "train_loss": 0.00011976622772635892,
          "validation_loss": 0.00010370645031798631,
          "train_metric": 1.206241449835943e-05,
          "validation_metric": 1.0444934559927788e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.9876565220001794
        },
        {
          "epoch": 339,
          "updates": 5424,
          "samples_seen": 347136,
          "online_loss": 0.00013433646699922974,
          "train_loss": 0.00011986350000370294,
          "validation_loss": 0.00010430813563289121,
          "train_metric": 1.2072214303771034e-05,
          "validation_metric": 1.0505533282412216e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.9980361639991315
        },
        {
          "epoch": 340,
          "updates": 5440,
          "samples_seen": 348160,
          "online_loss": 0.00013222292636783095,
          "train_loss": 0.00012394317309372127,
          "validation_loss": 0.00010284287418471649,
          "train_metric": 1.24831021821592e-05,
          "validation_metric": 1.0357957762607839e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 4.008384022999962
        },
        {
          "epoch": 341,
          "updates": 5456,
          "samples_seen": 349184,
          "online_loss": 0.00012296349586904398,
          "train_loss": 0.00011903826816705987,
          "validation_loss": 0.0001031847350532189,
          "train_metric": 1.1989098311460111e-05,
          "validation_metric": 1.0392388503532857e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 4.018662316999325
        },
        {
          "epoch": 342,
          "updates": 5472,
          "samples_seen": 350208,
          "online_loss": 0.00013412029284154414,
          "train_loss": 0.00012149308167863637,
          "validation_loss": 0.00010519638453843072,
          "train_metric": 1.2236337170179468e-05,
          "validation_metric": 1.0594994819257408e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 4.029015297999649
        },
        {
          "epoch": 343,
          "updates": 5488,
          "samples_seen": 351232,
          "online_loss": 0.0001337773119303165,
          "train_loss": 0.00013309340283740312,
          "validation_loss": 0.00011445059499237686,
          "train_metric": 1.3404679521045182e-05,
          "validation_metric": 1.1527044989634305e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 4.039832093998484
        },
        {
          "epoch": 344,
          "updates": 5504,
          "samples_seen": 352256,
          "online_loss": 0.0001296600244131696,
          "train_loss": 0.00011939475371036679,
          "validation_loss": 0.00010602138354443014,
          "train_metric": 1.2025002433802001e-05,
          "validation_metric": 1.0678086255211383e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 4.0503018719991815
        },
        {
          "epoch": 345,
          "updates": 5520,
          "samples_seen": 353280,
          "online_loss": 0.0001237271385434724,
          "train_loss": 0.00011968764010816813,
          "validation_loss": 0.00010117265628650784,
          "train_metric": 1.2054501894454006e-05,
          "validation_metric": 1.0189740351052023e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.061454208998839
        },
        {
          "epoch": 346,
          "updates": 5536,
          "samples_seen": 354304,
          "online_loss": 0.00012726981503874413,
          "train_loss": 0.0001193219650303945,
          "validation_loss": 0.00010115310578839853,
          "train_metric": 1.2017670087516308e-05,
          "validation_metric": 1.0187771295022685e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.07286946000022
        },
        {
          "epoch": 347,
          "updates": 5552,
          "samples_seen": 355328,
          "online_loss": 0.00012165543012088165,
          "train_loss": 0.00011961548443650827,
          "validation_loss": 0.00010123111860593781,
          "train_metric": 1.2047233212797437e-05,
          "validation_metric": 1.01956275102566e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.083441218999724
        },
        {
          "epoch": 348,
          "updates": 5568,
          "samples_seen": 356352,
          "online_loss": 0.00011829691220555105,
          "train_loss": 0.00011902146798092872,
          "validation_loss": 0.00010563184332568198,
          "train_metric": 1.1987407560809515e-05,
          "validation_metric": 1.06388515632716e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.094050174999211
        },
        {
          "epoch": 349,
          "updates": 5584,
          "samples_seen": 357376,
          "online_loss": 0.0001192638751490449,
          "train_loss": 0.00011754577280953526,
          "validation_loss": 0.00010135030606761575,
          "train_metric": 1.1838780665129889e-05,
          "validation_metric": 1.0207630111835897e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.10449028399853
        },
        {
          "epoch": 350,
          "updates": 5600,
          "samples_seen": 358400,
          "online_loss": 0.00012010739055767772,
          "train_loss": 0.00011820317013189197,
          "validation_loss": 0.00010057311737909913,
          "train_metric": 1.1904989150934853e-05,
          "validation_metric": 1.0129355359822512e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.115441956999348
        },
        {
          "epoch": 351,
          "updates": 5616,
          "samples_seen": 359424,
          "online_loss": 0.00012036265525239287,
          "train_loss": 0.00012007384066237137,
          "validation_loss": 0.00010597250366117805,
          "train_metric": 1.2093397344870027e-05,
          "validation_metric": 1.0673162250895984e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.126545999999507
        },
        {
          "epoch": 352,
          "updates": 5632,
          "samples_seen": 360448,
          "online_loss": 0.00012298280171307852,
          "train_loss": 0.0001171836192952469,
          "validation_loss": 0.00010218765237368643,
          "train_metric": 1.1802306289609987e-05,
          "validation_metric": 1.0291965736541897e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.137149296999269
        },
        {
          "epoch": 353,
          "updates": 5648,
          "samples_seen": 361472,
          "online_loss": 0.00011946722861466696,
          "train_loss": 0.0001173227428807877,
          "validation_loss": 0.00010214500071015209,
          "train_metric": 1.1816317964985501e-05,
          "validation_metric": 1.0287671102560125e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.147439444999691
        },
        {
          "epoch": 354,
          "updates": 5664,
          "samples_seen": 362496,
          "online_loss": 0.00011852924626509775,
          "train_loss": 0.00011705458018695936,
          "validation_loss": 0.00010124135587830096,
          "train_metric": 1.178930870082695e-05,
          "validation_metric": 1.019665887724841e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.15780106999955
        },
        {
          "epoch": 355,
          "updates": 5680,
          "samples_seen": 363520,
          "online_loss": 0.0001190766283798439,
          "train_loss": 0.00011693686246871948,
          "validation_loss": 0.0001017967369989492,
          "train_metric": 1.1777452527894638e-05,
          "validation_metric": 1.0252594620396849e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.168187926999963
        },
        {
          "epoch": 356,
          "updates": 5696,
          "samples_seen": 364544,
          "online_loss": 0.00012351453051451244,
          "train_loss": 0.00011704645294230431,
          "validation_loss": 0.0001010304840747267,
          "train_metric": 1.1788490155595355e-05,
          "validation_metric": 1.0175420356972609e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.17841658999896
        },
        {
          "epoch": 357,
          "updates": 5712,
          "samples_seen": 365568,
          "online_loss": 0.00011838966793220607,
          "train_loss": 0.00011717320012394339,
          "validation_loss": 0.00010060489876195788,
          "train_metric": 1.1801256732724141e-05,
          "validation_metric": 1.0132557690667454e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.188780345999476
        },
        {
          "epoch": 358,
          "updates": 5728,
          "samples_seen": 366592,
          "online_loss": 0.00011967825321335113,
          "train_loss": 0.0001169246097560972,
          "validation_loss": 0.00010122948879143223,
          "train_metric": 1.1776219253079034e-05,
          "validation_metric": 1.0195462891715579e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.199307613998826
        },
        {
          "epoch": 359,
          "updates": 5744,
          "samples_seen": 367616,
          "online_loss": 0.00011897774425051466,
          "train_loss": 0.00011689231178024784,
          "validation_loss": 0.00010360529267927632,
          "train_metric": 1.1772966900025494e-05,
          "validation_metric": 1.0434746400278527e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.209705862998817
        },
        {
          "epoch": 360,
          "updates": 5760,
          "samples_seen": 368640,
          "online_loss": 0.00011939736987187644,
          "train_loss": 0.00011708695819834247,
          "validation_loss": 0.0001003022407530807,
          "train_metric": 1.1792570148827508e-05,
          "validation_metric": 1.0102074156748131e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.220702994998646
        },
        {
          "epoch": 361,
          "updates": 5776,
          "samples_seen": 369664,
          "online_loss": 0.00011978997372352751,
          "train_loss": 0.00011730557162081823,
          "validation_loss": 0.00010030972043750808,
          "train_metric": 1.1814587196568027e-05,
          "validation_metric": 1.01028281278559e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.231265901998995
        },
        {
          "epoch": 362,
          "updates": 5792,
          "samples_seen": 370688,
          "online_loss": 0.00012028237142658327,
          "train_loss": 0.00011797079787356779,
          "validation_loss": 0.00010332728561479598,
          "train_metric": 1.1881586942763533e-05,
          "validation_metric": 1.040674396790564e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.241610340999614
        },
        {
          "epoch": 363,
          "updates": 5808,
          "samples_seen": 371712,
          "online_loss": 0.00012107507791370153,
          "train_loss": 0.00011757001630030572,
          "validation_loss": 0.00010074879537569359,
          "train_metric": 1.1841220839414746e-05,
          "validation_metric": 1.0147050488740206e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.252031846999671
        },
        {
          "epoch": 364,
          "updates": 5824,
          "samples_seen": 372736,
          "online_loss": 0.00011826227910205489,
          "train_loss": 0.00011644248297670856,
          "validation_loss": 0.0001019073897623457,
          "train_metric": 1.1727660421456676e-05,
          "validation_metric": 1.0263739568472374e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.2625803829996585
        },
        {
          "epoch": 365,
          "updates": 5840,
          "samples_seen": 373760,
          "online_loss": 0.00011739013689293643,
          "train_loss": 0.0001168623857665807,
          "validation_loss": 0.00010102213855134323,
          "train_metric": 1.1769951925089117e-05,
          "validation_metric": 1.0174581802857574e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.273186248999991
        },
        {
          "epoch": 366,
          "updates": 5856,
          "samples_seen": 374784,
          "online_loss": 0.00011733543169611949,
          "train_loss": 0.0001166498550446704,
          "validation_loss": 0.00010210928303422406,
          "train_metric": 1.1748546967282891e-05,
          "validation_metric": 1.0284073141519912e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.283550518999618
        },
        {
          "epoch": 367,
          "updates": 5872,
          "samples_seen": 375808,
          "online_loss": 0.00012094336852896959,
          "train_loss": 0.00011620738951023668,
          "validation_loss": 0.00010255179950036108,
          "train_metric": 1.170398263639072e-05,
          "validation_metric": 1.0328642929380294e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.293925482999839
        },
        {
          "epoch": 368,
          "updates": 5888,
          "samples_seen": 376832,
          "online_loss": 0.00011794126294262242,
          "train_loss": 0.00011834742326755077,
          "validation_loss": 0.0001012603534036316,
          "train_metric": 1.1919520147785079e-05,
          "validation_metric": 1.0198573363595642e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.304259927999738
        },
        {
          "epoch": 369,
          "updates": 5904,
          "samples_seen": 377856,
          "online_loss": 0.00012068463274772512,
          "train_loss": 0.00012557209993246943,
          "validation_loss": 0.00011218780127819628,
          "train_metric": 1.2647163202927914e-05,
          "validation_metric": 1.1299145626253448e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.3144647869994515
        },
        {
          "epoch": 370,
          "updates": 5920,
          "samples_seen": 378880,
          "online_loss": 0.00012719286132778507,
          "train_loss": 0.00011909006570931524,
          "validation_loss": 0.0001010355117614381,
          "train_metric": 1.1994316082564183e-05,
          "validation_metric": 1.0175926945521496e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.324881291999191
        },
        {
          "epoch": 371,
          "updates": 5936,
          "samples_seen": 379904,
          "online_loss": 0.00011748610631912015,
          "train_loss": 0.00011724582145689055,
          "validation_loss": 0.0001043754382408224,
          "train_metric": 1.18085708891158e-05,
          "validation_metric": 1.051231174642453e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.335458861000006
        },
        {
          "epoch": 372,
          "updates": 5952,
          "samples_seen": 380928,
          "online_loss": 0.00011856416404043557,
          "train_loss": 0.00011894712952198461,
          "validation_loss": 0.00010152015602216125,
          "train_metric": 1.1979918781435117e-05,
          "validation_metric": 1.022473952616565e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.345818300998872
        },
        {
          "epoch": 373,
          "updates": 5968,
          "samples_seen": 381952,
          "online_loss": 0.00012102211439923849,
          "train_loss": 0.00011748926772270352,
          "validation_loss": 0.00010452418064232916,
          "train_metric": 1.1833089047286194e-05,
          "validation_metric": 1.0527293852646835e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.35633532799875
        },
        {
          "epoch": 374,
          "updates": 5984,
          "samples_seen": 382976,
          "online_loss": 0.00011697150239342591,
          "train_loss": 0.00011877315409947187,
          "validation_loss": 0.0001013026776490733,
          "train_metric": 1.1962398275500163e-05,
          "validation_metric": 1.020283525576815e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.366673257998627
        },
        {
          "epoch": 375,
          "updates": 6000,
          "samples_seen": 384000,
          "online_loss": 0.00011960955134782125,
          "train_loss": 0.00011621934390859678,
          "validation_loss": 0.00010371551616117358,
          "train_metric": 1.1705186807375867e-05,
          "validation_metric": 1.0445846783113666e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.376908682999783
        },
        {
          "epoch": 376,
          "updates": 6016,
          "samples_seen": 385024,
          "online_loss": 0.00011840307342936285,
          "train_loss": 0.00011451529280748218,
          "validation_loss": 0.00010078771447297186,
          "train_metric": 1.1533560609677806e-05,
          "validation_metric": 1.0150968591915444e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.3873212469989085
        },
        {
          "epoch": 377,
          "updates": 6032,
          "samples_seen": 386048,
          "online_loss": 0.00011892189377249451,
          "train_loss": 0.00011590467329369858,
          "validation_loss": 0.00010394146374892443,
          "train_metric": 1.1673494554997887e-05,
          "validation_metric": 1.0468603250046726e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.39750813899991
        },
        {
          "epoch": 378,
          "updates": 6048,
          "samples_seen": 387072,
          "online_loss": 0.00012360502978481236,
          "train_loss": 0.00011897484364453703,
          "validation_loss": 0.00010047470277640969,
          "train_metric": 1.198271093016956e-05,
          "validation_metric": 1.011944368656259e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.407864755999981
        },
        {
          "epoch": 379,
          "updates": 6064,
          "samples_seen": 388096,
          "online_loss": 0.00011992521058346028,
          "train_loss": 0.00011554827506188303,
          "validation_loss": 0.00010429363464936614,
          "train_metric": 1.1637599527603015e-05,
          "validation_metric": 1.0504072633921169e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.418229152999629
        },
        {
          "epoch": 380,
          "updates": 6080,
          "samples_seen": 389120,
          "online_loss": 0.000120104125016951,
          "train_loss": 0.00011597129196161404,
          "validation_loss": 9.914241672959179e-05,
          "train_metric": 1.1680203897412866e-05,
          "validation_metric": 9.98526047624182e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.429068489998826
        },
        {
          "epoch": 381,
          "updates": 6096,
          "samples_seen": 390144,
          "online_loss": 0.00011586855089262826,
          "train_loss": 0.00011505593283800408,
          "validation_loss": 0.00010121650120709091,
          "train_metric": 1.158801114797825e-05,
          "validation_metric": 1.019415503833443e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.439531695999904
        },
        {
          "epoch": 382,
          "updates": 6112,
          "samples_seen": 391168,
          "online_loss": 0.00011881084469678171,
          "train_loss": 0.00011484508286230266,
          "validation_loss": 0.00010025200026575476,
          "train_metric": 1.1566776265681256e-05,
          "validation_metric": 1.009701554721687e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.449837319998551
        },
        {
          "epoch": 383,
          "updates": 6128,
          "samples_seen": 392192,
          "online_loss": 0.00012187049014755758,
          "train_loss": 0.0001163627821370028,
          "validation_loss": 0.00010464109072927386,
          "train_metric": 1.1719633221218828e-05,
          "validation_metric": 1.0539067261561286e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.460186778998832
        },
        {
          "epoch": 384,
          "updates": 6144,
          "samples_seen": 393216,
          "online_loss": 0.00011840544266306097,
          "train_loss": 0.00011384084064047784,
          "validation_loss": 9.863925515674055e-05,
          "train_metric": 1.146563226939179e-05,
          "validation_metric": 9.934583431459032e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.471226645999195
        },
        {
          "epoch": 385,
          "updates": 6160,
          "samples_seen": 394240,
          "online_loss": 0.00011590456870180788,
          "train_loss": 0.00011374818859621882,
          "validation_loss": 0.00010104088869411498,
          "train_metric": 1.14563008537516e-05,
          "validation_metric": 1.0176468094869051e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.481583652999689
        },
        {
          "epoch": 386,
          "updates": 6176,
          "samples_seen": 395264,
          "online_loss": 0.00011585028778426931,
          "train_loss": 0.00011341803474351764,
          "validation_loss": 9.929826046572998e-05,
          "train_metric": 1.1423049727454782e-05,
          "validation_metric": 1.0000956535805017e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.492064005999055
        },
        {
          "epoch": 387,
          "updates": 6192,
          "samples_seen": 396288,
          "online_loss": 0.00011509480805216299,
          "train_loss": 0.00011331494170008227,
          "validation_loss": 9.950359526555985e-05,
          "train_metric": 1.141266602644464e-05,
          "validation_metric": 1.0021637535828631e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.502342615000089
        },
        {
          "epoch": 388,
          "updates": 6208,
          "samples_seen": 397312,
          "online_loss": 0.00011720658585545607,
          "train_loss": 0.00011447616998339072,
          "validation_loss": 9.859348938334733e-05,
          "train_metric": 1.1529621588124428e-05,
          "validation_metric": 9.929975021805149e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.513166883998565
        },
        {
          "epoch": 389,
          "updates": 6224,
          "samples_seen": 398336,
          "online_loss": 0.00011925509897992015,
          "train_loss": 0.00011402561358409002,
          "validation_loss": 0.00010088199633173645,
          "train_metric": 1.1484242349979468e-05,
          "validation_metric": 1.0160465535591356e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.52417018199958
        },
        {
          "epoch": 390,
          "updates": 6240,
          "samples_seen": 399360,
          "online_loss": 0.00011492077646835241,
          "train_loss": 0.0001138161460403353,
          "validation_loss": 9.950566163752228e-05,
          "train_metric": 1.1463145710877143e-05,
          "validation_metric": 1.0021846719610039e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.5349071149994415
        },
        {
          "epoch": 391,
          "updates": 6256,
          "samples_seen": 400384,
          "online_loss": 0.00011630545759544475,
          "train_loss": 0.0001131832686951384,
          "validation_loss": 9.853075607679784e-05,
          "train_metric": 1.139940559369279e-05,
          "validation_metric": 9.923656762111932e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.5465549519994966
        },
        {
          "epoch": 392,
          "updates": 6272,
          "samples_seen": 401408,
          "online_loss": 0.00011702376741595799,
          "train_loss": 0.00011439922673162073,
          "validation_loss": 0.00010209454922005534,
          "train_metric": 1.1521872693265323e-05,
          "validation_metric": 1.028258975566132e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.557264323999334
        },
        {
          "epoch": 393,
          "updates": 6288,
          "samples_seen": 402432,
          "online_loss": 0.00011478092210381874,
          "train_loss": 0.00011426406126702204,
          "validation_loss": 9.887603664537892e-05,
          "train_metric": 1.1508257557579782e-05,
          "validation_metric": 9.958433111023623e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.567676008999115
        },
        {
          "epoch": 394,
          "updates": 6304,
          "samples_seen": 403456,
          "online_loss": 0.00011755677405744791,
          "train_loss": 0.00011319108307361603,
          "validation_loss": 9.867896005744115e-05,
          "train_metric": 1.140019048762042e-05,
          "validation_metric": 9.938584298652131e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.5780628100001195
        },
        {
          "epoch": 395,
          "updates": 6320,
          "samples_seen": 404480,
          "online_loss": 0.00011484001379358233,
          "train_loss": 0.00011236337741138414,
          "validation_loss": 9.946000500349328e-05,
          "train_metric": 1.131682893174002e-05,
          "validation_metric": 1.0017247404903173e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.588442125999791
        },
        {
          "epoch": 396,
          "updates": 6336,
          "samples_seen": 405504,
          "online_loss": 0.00011860505856020609,
          "train_loss": 0.00011268845992162824,
          "validation_loss": 0.00010061914508696645,
          "train_metric": 1.1349568922014441e-05,
          "validation_metric": 1.013399196381215e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.598776281998653
        },
        {
          "epoch": 397,
          "updates": 6352,
          "samples_seen": 406528,
          "online_loss": 0.00011840731303891516,
          "train_loss": 0.00011282921332167462,
          "validation_loss": 9.75807779468596e-05,
          "train_metric": 1.1363744306436274e-05,
          "validation_metric": 9.82797791948542e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.6096006939988
        },
        {
          "epoch": 398,
          "updates": 6368,
          "samples_seen": 407552,
          "online_loss": 0.00011580668342503486,
          "train_loss": 0.00011248261580476537,
          "validation_loss": 9.811639756662771e-05,
          "train_metric": 1.1328836080792826e-05,
          "validation_metric": 9.88192368822638e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.620025831000021
        },
        {
          "epoch": 399,
          "updates": 6384,
          "samples_seen": 408576,
          "online_loss": 0.00011414849404900451,
          "train_loss": 0.00011192927195224911,
          "validation_loss": 9.94493966572918e-05,
          "train_metric": 1.1273104973952286e-05,
          "validation_metric": 1.001617874862859e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.630386909999288
        },
        {
          "epoch": 400,
          "updates": 6400,
          "samples_seen": 409600,
          "online_loss": 0.00011343072128511267,
          "train_loss": 0.00011194259423064068,
          "validation_loss": 9.870652866084129e-05,
          "train_metric": 1.1274447388132103e-05,
          "validation_metric": 9.941360076481942e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.640655131999665
        },
        {
          "epoch": 401,
          "updates": 6416,
          "samples_seen": 410624,
          "online_loss": 0.00011318064116494497,
          "train_loss": 0.00011199430446140468,
          "validation_loss": 9.778395178727806e-05,
          "train_metric": 1.1279656064289156e-05,
          "validation_metric": 9.848441550275311e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.650902849998602
        },
        {
          "epoch": 402,
          "updates": 6432,
          "samples_seen": 411648,
          "online_loss": 0.00011569590969884302,
          "train_loss": 0.00011213423567824066,
          "validation_loss": 9.793911158340052e-05,
          "train_metric": 1.129374959418783e-05,
          "validation_metric": 9.864069397735875e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.661237884000002
        },
        {
          "epoch": 403,
          "updates": 6448,
          "samples_seen": 412672,
          "online_loss": 0.00012106627491448307,
          "train_loss": 0.0001117376668844372,
          "validation_loss": 0.0001002274002530612,
          "train_metric": 1.125380822486477e-05,
          "validation_metric": 1.0094536264659837e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.727149788999668
        },
        {
          "epoch": 404,
          "updates": 6464,
          "samples_seen": 413696,
          "online_loss": 0.0001129121797021071,
          "train_loss": 0.00011140732385683805,
          "validation_loss": 9.888204658636823e-05,
          "train_metric": 1.1220537089684512e-05,
          "validation_metric": 9.959037925000302e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.73814139499882
        },
        {
          "epoch": 405,
          "updates": 6480,
          "samples_seen": 414720,
          "online_loss": 0.00011398386459404719,
          "train_loss": 0.00011434634507168084,
          "validation_loss": 9.831542411120608e-05,
          "train_metric": 1.1516543054312933e-05,
          "validation_metric": 9.901970770442858e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.74864000999878
        },
        {
          "epoch": 406,
          "updates": 6496,
          "samples_seen": 415744,
          "online_loss": 0.00011870123194057669,
          "train_loss": 0.00012070583761669695,
          "validation_loss": 0.00011245717905694619,
          "train_metric": 1.2157049241068307e-05,
          "validation_metric": 1.1326275853207335e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.759299284998633
        },
        {
          "epoch": 407,
          "updates": 6512,
          "samples_seen": 416768,
          "online_loss": 0.00011443787752796197,
          "train_loss": 0.00011471730977063999,
          "validation_loss": 9.958846203517169e-05,
          "train_metric": 1.1553906915651169e-05,
          "validation_metric": 1.0030184967035893e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.769705012999111
        },
        {
          "epoch": 408,
          "updates": 6528,
          "samples_seen": 417792,
          "online_loss": 0.00012598038438227377,
          "train_loss": 0.00012150929978815839,
          "validation_loss": 0.00010951738659059629,
          "train_metric": 1.2237972441653255e-05,
          "validation_metric": 1.1030188943550456e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.780112993999865
        },
        {
          "epoch": 409,
          "updates": 6544,
          "samples_seen": 418816,
          "online_loss": 0.00011841714240290457,
          "train_loss": 0.0001110338926082477,
          "validation_loss": 9.670728468336165e-05,
          "train_metric": 1.11829285742715e-05,
          "validation_metric": 9.740004315972328e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.791220084998713
        },
        {
          "epoch": 410,
          "updates": 6560,
          "samples_seen": 419840,
          "online_loss": 0.00011347975032549584,
          "train_loss": 0.00011088711471529678,
          "validation_loss": 9.838394908001646e-05,
          "train_metric": 1.116814382839948e-05,
          "validation_metric": 9.908872016239911e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.801753096999164
        },
        {
          "epoch": 411,
          "updates": 6576,
          "samples_seen": 420864,
          "online_loss": 0.00011534083569131326,
          "train_loss": 0.00011280691978754476,
          "validation_loss": 9.805866284295917e-05,
          "train_metric": 1.1361498764017597e-05,
          "validation_metric": 9.876109288597945e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.811996568998438
        },
        {
          "epoch": 412,
          "updates": 6592,
          "samples_seen": 421888,
          "online_loss": 0.00011416051825108298,
          "train_loss": 0.00011266819637967274,
          "validation_loss": 0.0001024789089569822,
          "train_metric": 1.1347528925398365e-05,
          "validation_metric": 1.0321300578652881e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.822382760999972
        },
        {
          "epoch": 413,
          "updates": 6608,
          "samples_seen": 422912,
          "online_loss": 0.0001121457239605661,
          "train_loss": 0.00011129664198961109,
          "validation_loss": 9.721643436932936e-05,
          "train_metric": 1.1209389413124882e-05,
          "validation_metric": 9.79128344624769e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.83267909999995
        },
        {
          "epoch": 414,
          "updates": 6624,
          "samples_seen": 423936,
          "online_loss": 0.00011567739556994638,
          "train_loss": 0.00011517266102600843,
          "validation_loss": 0.00010273220686940476,
          "train_metric": 1.1599767276493367e-05,
          "validation_metric": 1.0346811905037612e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.842927491999944
        },
        {
          "epoch": 415,
          "updates": 6640,
          "samples_seen": 424960,
          "online_loss": 0.00011426313812989974,
          "train_loss": 0.00011434233601903543,
          "validation_loss": 9.71053377725184e-05,
          "train_metric": 1.1516141967149451e-05,
          "validation_metric": 9.780094842426479e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.853316737999194
        },
        {
          "epoch": 416,
          "updates": 6656,
          "samples_seen": 425984,
          "online_loss": 0.0001135024826908193,
          "train_loss": 0.00011114495282527059,
          "validation_loss": 9.958970622392371e-05,
          "train_metric": 1.11941117211245e-05,
          "validation_metric": 1.0030310477304738e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.863753633999295
        },
        {
          "epoch": 417,
          "updates": 6672,
          "samples_seen": 427008,
          "online_loss": 0.00011340783748892136,
          "train_loss": 0.00011420169903431088,
          "validation_loss": 9.891929948935285e-05,
          "train_metric": 1.1501976587169338e-05,
          "validation_metric": 9.962789590645116e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.874189225998634
        },
        {
          "epoch": 418,
          "updates": 6688,
          "samples_seen": 428032,
          "online_loss": 0.00011326593448757194,
          "train_loss": 0.00011244750930927694,
          "validation_loss": 0.00010410656977910548,
          "train_metric": 1.1325300874887034e-05,
          "validation_metric": 1.048523154167924e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.884565881999151
        },
        {
          "epoch": 419,
          "updates": 6704,
          "samples_seen": 429056,
          "online_loss": 0.00011944591278734151,
          "train_loss": 0.00011217465362278745,
          "validation_loss": 9.694904292700812e-05,
          "train_metric": 1.1297821401967667e-05,
          "validation_metric": 9.764353308128193e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.894925037999201
        },
        {
          "epoch": 420,
          "updates": 6720,
          "samples_seen": 430080,
          "online_loss": 0.00011578724252103711,
          "train_loss": 0.00011240130697842687,
          "validation_loss": 0.00010127147834282368,
          "train_metric": 1.1320647899992764e-05,
          "validation_metric": 1.0199692042078823e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.905244429999584
        },
        {
          "epoch": 421,
          "updates": 6736,
          "samples_seen": 431104,
          "online_loss": 0.0001231396622642933,
          "train_loss": 0.0001108342403313145,
          "validation_loss": 9.909904474625364e-05,
          "train_metric": 1.1162817827425897e-05,
          "validation_metric": 9.980891263694502e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.915520855000068
        },
        {
          "epoch": 422,
          "updates": 6752,
          "samples_seen": 432128,
          "online_loss": 0.00011646618759186822,
          "train_loss": 0.00011044552957173437,
          "validation_loss": 9.649834828451276e-05,
          "train_metric": 1.112366771849338e-05,
          "validation_metric": 9.718961337057408e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.926710033998461
        },
        {
          "epoch": 423,
          "updates": 6768,
          "samples_seen": 433152,
          "online_loss": 0.00011163581643813814,
          "train_loss": 0.00011005523265339434,
          "validation_loss": 9.976114961318672e-05,
          "train_metric": 1.1084360266977455e-05,
          "validation_metric": 1.0047576324723195e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.937353733999771
        },
        {
          "epoch": 424,
          "updates": 6784,
          "samples_seen": 434176,
          "online_loss": 0.00011685105982905952,
          "train_loss": 0.00011418793292250484,
          "validation_loss": 9.773335477802902e-05,
          "train_metric": 1.1500589607749134e-05,
          "validation_metric": 9.843345651461277e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.947952409998834
        },
        {
          "epoch": 425,
          "updates": 6800,
          "samples_seen": 435200,
          "online_loss": 0.00011603702114371117,
          "train_loss": 0.00011044122220482677,
          "validation_loss": 9.927876817528158e-05,
          "train_metric": 1.1123234799015336e-05,
          "validation_metric": 9.998995665227994e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.958780740998918
        },
        {
          "epoch": 426,
          "updates": 6816,
          "samples_seen": 436224,
          "online_loss": 0.00011131071551062632,
          "train_loss": 0.00011077617818955332,
          "validation_loss": 9.655843314249068e-05,
          "train_metric": 1.1156970685988199e-05,
          "validation_metric": 9.725012205308303e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.96941994000008
        },
        {
          "epoch": 427,
          "updates": 6832,
          "samples_seen": 437248,
          "online_loss": 0.00011450535066614975,
          "train_loss": 0.00011127817560918629,
          "validation_loss": 0.00010172138718189672,
          "train_metric": 1.1207530405954458e-05,
          "validation_metric": 1.0245004887110554e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.97994347599888
        },
        {
          "epoch": 428,
          "updates": 6848,
          "samples_seen": 438272,
          "online_loss": 0.00011445959216871415,
          "train_loss": 0.00010904360533459112,
          "validation_loss": 9.809074981603771e-05,
          "train_metric": 1.0982474123011343e-05,
          "validation_metric": 9.879341632768046e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.990489490999607
        },
        {
          "epoch": 429,
          "updates": 6864,
          "samples_seen": 439296,
          "online_loss": 0.00011349502392477007,
          "train_loss": 0.00010991135786753148,
          "validation_loss": 9.613641304895282e-05,
          "train_metric": 1.106987019738881e-05,
          "validation_metric": 9.682507879915647e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.001609639999515
        },
        {
          "epoch": 430,
          "updates": 6880,
          "samples_seen": 440320,
          "online_loss": 0.00011276994564468623,
          "train_loss": 0.00010866645607165992,
          "validation_loss": 9.685270197223872e-05,
          "train_metric": 1.0944489076791797e-05,
          "validation_metric": 9.754649909154978e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.012086123999325
        },
        {
          "epoch": 431,
          "updates": 6896,
          "samples_seen": 441344,
          "online_loss": 0.0001119881080740015,
          "train_loss": 0.0001095044644898735,
          "validation_loss": 9.866686013992876e-05,
          "train_metric": 1.1028889275621623e-05,
          "validation_metric": 9.937364666257054e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.022545719999471
        },
        {
          "epoch": 432,
          "updates": 6912,
          "samples_seen": 442368,
          "online_loss": 0.00011496897468532552,
          "train_loss": 0.00010893051512539387,
          "validation_loss": 9.860083810053766e-05,
          "train_metric": 1.0971083611366339e-05,
          "validation_metric": 9.930715350492392e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.032931463998466
        },
        {
          "epoch": 433,
          "updates": 6928,
          "samples_seen": 443392,
          "online_loss": 0.00011217295059395838,
          "train_loss": 0.0001084788455045782,
          "validation_loss": 9.600172779755667e-05,
          "train_metric": 1.092559250537306e-05,
          "validation_metric": 9.668943675933406e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.043779239998912
        },
        {
          "epoch": 434,
          "updates": 6944,
          "samples_seen": 444416,
          "online_loss": 0.0001148508749793109,
          "train_loss": 0.00011088945029769093,
          "validation_loss": 0.0001011127678793855,
          "train_metric": 1.116838029702194e-05,
          "validation_metric": 1.0183708582189865e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.054476555998917
        },
        {
          "epoch": 435,
          "updates": 6960,
          "samples_seen": 445440,
          "online_loss": 0.00011428735342633445,
          "train_loss": 0.00010815459245350212,
          "validation_loss": 9.665234392741695e-05,
          "train_metric": 1.0892933460127097e-05,
          "validation_metric": 9.734469131217338e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.064780230999531
        },
        {
          "epoch": 436,
          "updates": 6976,
          "samples_seen": 446464,
          "online_loss": 0.00011846610414067982,
          "train_loss": 0.00011233019176870584,
          "validation_loss": 9.769882308319211e-05,
          "train_metric": 1.1313486538711004e-05,
          "validation_metric": 9.839867743721697e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.075188153999989
        },
        {
          "epoch": 437,
          "updates": 6992,
          "samples_seen": 447488,
          "online_loss": 0.00012190641382403555,
          "train_loss": 0.0001115719715016894,
          "validation_loss": 0.00010175359057029709,
          "train_metric": 1.1237121725571342e-05,
          "validation_metric": 1.0248248145217076e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.085590730999684
        },
        {
          "epoch": 438,
          "updates": 7008,
          "samples_seen": 448512,
          "online_loss": 0.00011214762571398751,
          "train_loss": 0.00010809136438183486,
          "validation_loss": 9.77258023340255e-05,
          "train_metric": 1.0886566997214686e-05,
          "validation_metric": 9.842585313890595e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.096018834999995
        },
        {
          "epoch": 439,
          "updates": 7024,
          "samples_seen": 449536,
          "online_loss": 0.0001135391348725534,
          "train_loss": 0.00011022020771633834,
          "validation_loss": 0.00010117008059751242,
          "train_metric": 1.1100974916189443e-05,
          "validation_metric": 1.0189478416577913e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.1618613339996955
        },
        {
          "epoch": 440,
          "updates": 7040,
          "samples_seen": 450560,
          "online_loss": 0.00011019483099516947,
          "train_loss": 0.00010998890502378345,
          "validation_loss": 9.584367217030376e-05,
          "train_metric": 1.1077679118898232e-05,
          "validation_metric": 9.653023880673572e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.173118777000127
        },
        {
          "epoch": 441,
          "updates": 7056,
          "samples_seen": 451584,
          "online_loss": 0.00011505452175697428,
          "train_loss": 0.00010751967784017324,
          "validation_loss": 9.487939678365365e-05,
          "train_metric": 1.0828989616129547e-05,
          "validation_metric": 9.555904398439452e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.184055101999547
        },
        {
          "epoch": 442,
          "updates": 7072,
          "samples_seen": 452608,
          "online_loss": 0.0001098728453143849,
          "train_loss": 0.00010785063204821199,
          "validation_loss": 9.776128717930987e-05,
          "train_metric": 1.0862320777960122e-05,
          "validation_metric": 9.84615871857386e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.194800221999685
        },
        {
          "epoch": 443,
          "updates": 7088,
          "samples_seen": 453632,
          "online_loss": 0.00010911557274084771,
          "train_loss": 0.00010933673183899373,
          "validation_loss": 9.691965533420444e-05,
          "train_metric": 1.1011996321030892e-05,
          "validation_metric": 9.761393812368624e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.205209883999487
        },
        {
          "epoch": 444,
          "updates": 7104,
          "samples_seen": 454656,
          "online_loss": 0.00011001448228853405,
          "train_loss": 0.00010765589104266837,
          "validation_loss": 9.764156129676849e-05,
          "train_metric": 1.0842706615221687e-05,
          "validation_metric": 9.834100637817755e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.21565108499999
        },
        {
          "epoch": 445,
          "updates": 7120,
          "samples_seen": 455680,
          "online_loss": 0.000113956657514791,
          "train_loss": 0.00010933279554592445,
          "validation_loss": 9.495459380559623e-05,
          "train_metric": 1.1011597962351516e-05,
          "validation_metric": 9.563477760821115e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.226146180999422
        },
        {
          "epoch": 446,
          "updates": 7136,
          "samples_seen": 456704,
          "online_loss": 0.0001116824446398823,
          "train_loss": 0.0001074162355507724,
          "validation_loss": 9.46416548686102e-05,
          "train_metric": 1.0818570444826037e-05,
          "validation_metric": 9.531960131425876e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.237390091999259
        },
        {
          "epoch": 447,
          "updates": 7152,
          "samples_seen": 457728,
          "online_loss": 0.00011627805497482768,
          "train_loss": 0.00010690557974157855,
          "validation_loss": 9.638891060603783e-05,
          "train_metric": 1.0767138519440778e-05,
          "validation_metric": 9.707937351777218e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.248194067999066
        },
        {
          "epoch": 448,
          "updates": 7168,
          "samples_seen": 458752,
          "online_loss": 0.00011660452946671285,
          "train_loss": 0.00011706669465638697,
          "validation_loss": 0.00010396848665550351,
          "train_metric": 1.179052924271673e-05,
          "validation_metric": 1.0471325367689133e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.258969264999905
        },
        {
          "epoch": 449,
          "updates": 7184,
          "samples_seen": 459776,
          "online_loss": 0.00011629217897279887,
          "train_loss": 0.00011912484478671104,
          "validation_loss": 9.934874105965719e-05,
          "train_metric": 1.1997816727671307e-05,
          "validation_metric": 1.0006040611187927e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.269436766999206
        },
        {
          "epoch": 450,
          "updates": 7200,
          "samples_seen": 460800,
          "online_loss": 0.00011389766859792871,
          "train_loss": 0.00011998203990515321,
          "validation_loss": 0.00010934146121144295,
          "train_metric": 1.2084152331226505e-05,
          "validation_metric": 1.1012471077265218e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.279957173999719
        },
        {
          "epoch": 451,
          "updates": 7216,
          "samples_seen": 461824,
          "online_loss": 0.00011605609870457556,
          "train_loss": 0.00010630337055772543,
          "validation_loss": 9.561385377310216e-05,
          "train_metric": 1.0706485227274243e-05,
          "validation_metric": 9.629877240513451e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.290544708999732
        },
        {
          "epoch": 452,
          "updates": 7232,
          "samples_seen": 462848,
          "online_loss": 0.00011007308194166399,
          "train_loss": 0.00010715151438489556,
          "validation_loss": 9.490308002568781e-05,
          "train_metric": 1.0791908607643563e-05,
          "validation_metric": 9.5582890935475e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.300985025000045
        },
        {
          "epoch": 453,
          "updates": 7248,
          "samples_seen": 463872,
          "online_loss": 0.00011169137769684312,
          "train_loss": 0.00011168624769197777,
          "validation_loss": 9.925783524522558e-05,
          "train_metric": 1.1248628652538173e-05,
          "validation_metric": 9.996886547014583e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.311276894999537
        },
        {
          "epoch": 454,
          "updates": 7264,
          "samples_seen": 464896,
          "online_loss": 0.00011043514405173482,
          "train_loss": 0.00010927925177384168,
          "validation_loss": 9.696290362626314e-05,
          "train_metric": 1.1006206477759406e-05,
          "validation_metric": 9.765747563506011e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.321910298998773
        },
        {
          "epoch": 455,
          "updates": 7280,
          "samples_seen": 465920,
          "online_loss": 0.0001081448845070554,
          "train_loss": 0.00010782533354358748,
          "validation_loss": 9.828033216763288e-05,
          "train_metric": 1.0859772373805754e-05,
          "validation_metric": 9.898435564537067e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.332209627998964
        },
        {
          "epoch": 456,
          "updates": 7296,
          "samples_seen": 466944,
          "online_loss": 0.00010884907214858686,
          "train_loss": 0.00010833804844878614,
          "validation_loss": 9.5937772130128e-05,
          "train_metric": 1.0911412573477719e-05,
          "validation_metric": 9.662500815466046e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.342807482998978
        },
        {
          "epoch": 457,
          "updates": 7312,
          "samples_seen": 467968,
          "online_loss": 0.00011384287586224673,
          "train_loss": 0.00012553301348816603,
          "validation_loss": 0.0001160645842901431,
          "train_metric": 1.2643226000363939e-05,
          "validation_metric": 1.1689599887176882e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.353174619998754
        },
        {
          "epoch": 458,
          "updates": 7328,
          "samples_seen": 468992,
          "online_loss": 0.0001236066489127552,
          "train_loss": 0.00010783059406094253,
          "validation_loss": 9.478263382334262e-05,
          "train_metric": 1.0860302609216888e-05,
          "validation_metric": 9.546160072204657e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.36400882199996
        },
        {
          "epoch": 459,
          "updates": 7344,
          "samples_seen": 470016,
          "online_loss": 0.00011013780931534711,
          "train_loss": 0.00010620239481795579,
          "validation_loss": 9.399185364600271e-05,
          "train_metric": 1.0696317076508421e-05,
          "validation_metric": 9.466515621170402e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.374865833999138
        },
        {
          "epoch": 460,
          "updates": 7360,
          "samples_seen": 471040,
          "online_loss": 0.00010856129483727273,
          "train_loss": 0.00010967427806463093,
          "validation_loss": 0.00010149968875339255,
          "train_metric": 1.1045991413993761e-05,
          "validation_metric": 1.0222678611171432e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.38538217399946
        },
        {
          "epoch": 461,
          "updates": 7376,
          "samples_seen": 472064,
          "online_loss": 0.0001209793085763522,
          "train_loss": 0.00010768861102405936,
          "validation_loss": 9.420850983588025e-05,
          "train_metric": 1.0846004443010315e-05,
          "validation_metric": 9.488335308560636e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.396031829999629
        },
        {
          "epoch": 462,
          "updates": 7392,
          "samples_seen": 473088,
          "online_loss": 0.00010642604615895834,
          "train_loss": 0.00010750236106105149,
          "validation_loss": 9.68564345384948e-05,
          "train_metric": 1.0827242476807442e-05,
          "validation_metric": 9.75502553046681e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.406710518998807
        },
        {
          "epoch": 463,
          "updates": 7408,
          "samples_seen": 474112,
          "online_loss": 0.00010628289442138339,
          "train_loss": 0.00010839497554115951,
          "validation_loss": 9.530565876048058e-05,
          "train_metric": 1.0917146028077696e-05,
          "validation_metric": 9.598837095836643e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.4171493939993525
        },
        {
          "epoch": 464,
          "updates": 7424,
          "samples_seen": 475136,
          "online_loss": 0.00011011548986061825,
          "train_loss": 0.00010934320744127035,
          "validation_loss": 0.00010295159154338762,
          "train_metric": 1.1012647519237362e-05,
          "validation_metric": 1.0368907169322483e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.427681582999867
        },
        {
          "epoch": 465,
          "updates": 7440,
          "samples_seen": 476160,
          "online_loss": 0.00010987049836330698,
          "train_loss": 0.00010961953375954181,
          "validation_loss": 9.450029756408185e-05,
          "train_metric": 1.1040478057111613e-05,
          "validation_metric": 9.517723810859025e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.438007041999299
        },
        {
          "epoch": 466,
          "updates": 7456,
          "samples_seen": 477184,
          "online_loss": 0.00010710201559049892,
          "train_loss": 0.00010534264583839104,
          "validation_loss": 9.534404671285301e-05,
          "train_metric": 1.0609724995447323e-05,
          "validation_metric": 9.602702448319178e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.448372544999074
        },
        {
          "epoch": 467,
          "updates": 7472,
          "samples_seen": 478208,
          "online_loss": 0.00010804299654409988,
          "train_loss": 0.00010699106496758759,
          "validation_loss": 9.503723413217813e-05,
          "train_metric": 1.0775748705782462e-05,
          "validation_metric": 9.571802365826443e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.458712522999122
        },
        {
          "epoch": 468,
          "updates": 7488,
          "samples_seen": 479232,
          "online_loss": 0.00011241150514251785,
          "train_loss": 0.00010716955148382112,
          "validation_loss": 9.894122194964439e-05,
          "train_metric": 1.0793723959068302e-05,
          "validation_metric": 9.96499784378102e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.46909850100019
        },
        {
          "epoch": 469,
          "updates": 7504,
          "samples_seen": 480256,
          "online_loss": 0.00011362899613232003,
          "train_loss": 0.00010440722689963877,
          "validation_loss": 9.489900548942387e-05,
          "train_metric": 1.051551316777477e-05,
          "validation_metric": 9.557880730426405e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.479298145999564
        },
        {
          "epoch": 470,
          "updates": 7520,
          "samples_seen": 481280,
          "online_loss": 0.00011332390113238944,
          "train_loss": 0.00011183299648109823,
          "validation_loss": 9.615831368137151e-05,
          "train_metric": 1.1263409760431387e-05,
          "validation_metric": 9.684714314062148e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.48972486799903
        },
        {
          "epoch": 471,
          "updates": 7536,
          "samples_seen": 482304,
          "online_loss": 0.00011111488674941938,
          "train_loss": 0.00010432344424771145,
          "validation_loss": 9.496009442955256e-05,
          "train_metric": 1.0507074875931721e-05,
          "validation_metric": 9.564032552589197e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.500036219998947
        },
        {
          "epoch": 472,
          "updates": 7552,
          "samples_seen": 483328,
          "online_loss": 0.00010664406363503076,
          "train_loss": 0.00010488947737030685,
          "validation_loss": 9.521848551230505e-05,
          "train_metric": 1.056408291333355e-05,
          "validation_metric": 9.590056833985727e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.510335334000047
        },
        {
          "epoch": 473,
          "updates": 7568,
          "samples_seen": 484352,
          "online_loss": 0.00010695802484406158,
          "train_loss": 0.00010534990724408999,
          "validation_loss": 9.677726484369487e-05,
          "train_metric": 1.061045713868225e-05,
          "validation_metric": 9.747051080921665e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.5208526859987614
        },
        {
          "epoch": 474,
          "updates": 7584,
          "samples_seen": 485376,
          "online_loss": 0.0001059731707755418,
          "train_loss": 0.00010396088327979669,
          "validation_loss": 9.439337736694142e-05,
          "train_metric": 1.047055957315024e-05,
          "validation_metric": 9.506955393590033e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.5312187419986
        },
        {
          "epoch": 475,
          "updates": 7600,
          "samples_seen": 486400,
          "online_loss": 0.00010691577881516423,
          "train_loss": 0.00010410737013444304,
          "validation_loss": 9.453072561882436e-05,
          "train_metric": 1.04853133962024e-05,
          "validation_metric": 9.520788808004e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.597276884000166
        },
        {
          "epoch": 476,
          "updates": 7616,
          "samples_seen": 487424,
          "online_loss": 0.00010493751005924423,
          "train_loss": 0.00010372613905929029,
          "validation_loss": 9.449889330426231e-05,
          "train_metric": 1.044691634888295e-05,
          "validation_metric": 9.517581929685548e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.607941041000231
        },
        {
          "epoch": 477,
          "updates": 7632,
          "samples_seen": 488448,
          "online_loss": 0.00010667936112440657,
          "train_loss": 0.00010782670869957656,
          "validation_loss": 9.338375821243972e-05,
          "train_metric": 1.0859910617000423e-05,
          "validation_metric": 9.405270247953013e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.618802136999875
        },
        {
          "epoch": 478,
          "updates": 7648,
          "samples_seen": 489472,
          "online_loss": 0.00010944792211375898,
          "train_loss": 0.0001147261937148869,
          "validation_loss": 0.00010492702858755365,
          "train_metric": 1.1554800948943011e-05,
          "validation_metric": 1.0567865501798224e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.629307405999498
        },
        {
          "epoch": 479,
          "updates": 7664,
          "samples_seen": 490496,
          "online_loss": 0.00011287820598226972,
          "train_loss": 0.00010388869122834876,
          "validation_loss": 9.421088907402009e-05,
          "train_metric": 1.0463287253514864e-05,
          "validation_metric": 9.488575415161904e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.639664899999843
        },
        {
          "epoch": 480,
          "updates": 7680,
          "samples_seen": 491520,
          "online_loss": 0.0001079137614397041,
          "train_loss": 0.0001039215421769768,
          "validation_loss": 9.483490430284292e-05,
          "train_metric": 1.0466595995239913e-05,
          "validation_metric": 9.551424227538519e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.649939526998423
        },
        {
          "epoch": 481,
          "updates": 7696,
          "samples_seen": 492544,
          "online_loss": 0.00010905129556704196,
          "train_loss": 0.00010403721535112709,
          "validation_loss": 9.3224058218766e-05,
          "train_metric": 1.0478248441359028e-05,
          "validation_metric": 9.38918674364686e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.660970023000118
        },
        {
          "epoch": 482,
          "updates": 7712,
          "samples_seen": 493568,
          "online_loss": 0.00010507694514672039,
          "train_loss": 0.0001036499161273241,
          "validation_loss": 9.310111636295915e-05,
          "train_metric": 1.0439240213599987e-05,
          "validation_metric": 9.376803973282222e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.671956428999692
        },
        {
          "epoch": 483,
          "updates": 7728,
          "samples_seen": 494592,
          "online_loss": 0.00011074671328969998,
          "train_loss": 0.00011201671441085637,
          "validation_loss": 0.000104084778286051,
          "train_metric": 1.128191524912836e-05,
          "validation_metric": 1.0483037840458564e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.682370262999029
        },
        {
          "epoch": 484,
          "updates": 7744,
          "samples_seen": 495616,
          "online_loss": 0.00011379742727513076,
          "train_loss": 0.00010336286504752934,
          "validation_loss": 9.421813592780381e-05,
          "train_metric": 1.041032919602003e-05,
          "validation_metric": 9.489307558396831e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.693039829999179
        },
        {
          "epoch": 485,
          "updates": 7760,
          "samples_seen": 496640,
          "online_loss": 0.00010725471020123223,
          "train_loss": 0.00010550353908911347,
          "validation_loss": 9.31389004108496e-05,
          "train_metric": 1.0625930372043513e-05,
          "validation_metric": 9.380608389619738e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.703357660999245
        },
        {
          "epoch": 486,
          "updates": 7776,
          "samples_seen": 497664,
          "online_loss": 0.00010886829159062472,
          "train_loss": 0.00010512267181184143,
          "validation_loss": 9.846898319665343e-05,
          "train_metric": 1.0587569704512134e-05,
          "validation_metric": 9.917433999362402e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.713673673999438
        },
        {
          "epoch": 487,
          "updates": 7792,
          "samples_seen": 498688,
          "online_loss": 0.00010652988430592814,
          "train_loss": 0.0001027185790007934,
          "validation_loss": 9.230487194145098e-05,
          "train_metric": 1.0345440387027338e-05,
          "validation_metric": 9.296609277953394e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.7248078389984585
        },
        {
          "epoch": 488,
          "updates": 7808,
          "samples_seen": 499712,
          "online_loss": 0.00010644184976627002,
          "train_loss": 0.00010293953528162092,
          "validation_loss": 9.227075497619808e-05,
          "train_metric": 1.0367693903390318e-05,
          "validation_metric": 9.293171387980692e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.735896091999166
        },
        {
          "epoch": 489,
          "updates": 7824,
          "samples_seen": 500736,
          "online_loss": 0.00010563074829406105,
          "train_loss": 0.00010244647273793817,
          "validation_loss": 9.409665653947741e-05,
          "train_metric": 1.0318032764189411e-05,
          "validation_metric": 9.477071216679178e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.746441336999851
        },
        {
          "epoch": 490,
          "updates": 7840,
          "samples_seen": 501760,
          "online_loss": 0.00010803757368194056,
          "train_loss": 0.00010374708654126152,
          "validation_loss": 9.217686601914465e-05,
          "train_metric": 1.0449026376591064e-05,
          "validation_metric": 9.283715371566359e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.757749012998829
        },
        {
          "epoch": 491,
          "updates": 7856,
          "samples_seen": 502784,
          "online_loss": 0.0001069821696546569,
          "train_loss": 0.00010471943824086338,
          "validation_loss": 9.677240450400859e-05,
          "train_metric": 1.0546958037593868e-05,
          "validation_metric": 9.746562682266813e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.76835365199986
        },
        {
          "epoch": 492,
          "updates": 7872,
          "samples_seen": 503808,
          "online_loss": 0.00010809682680701371,
          "train_loss": 0.00010331528028473258,
          "validation_loss": 9.559332102071494e-05,
          "train_metric": 1.040553797793109e-05,
          "validation_metric": 9.62780904956162e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.779007851999268
        },
        {
          "epoch": 493,
          "updates": 7888,
          "samples_seen": 504832,
          "online_loss": 0.00010803005875459348,
          "train_loss": 0.00012274317850824445,
          "validation_loss": 0.00010143686813535169,
          "train_metric": 1.2362243978714105e-05,
          "validation_metric": 1.0216349437541794e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.789556390998769
        },
        {
          "epoch": 494,
          "updates": 7904,
          "samples_seen": 505856,
          "online_loss": 0.00011516415906953625,
          "train_loss": 0.00010425147775094956,
          "validation_loss": 9.877266711555421e-05,
          "train_metric": 1.0499828931642696e-05,
          "validation_metric": 9.948022125172429e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.800102634999348
        },
        {
          "epoch": 495,
          "updates": 7920,
          "samples_seen": 506880,
          "online_loss": 0.00010308733499186928,
          "train_loss": 0.00010521707736188546,
          "validation_loss": 9.188619878841564e-05,
          "train_metric": 1.059707847161917e-05,
          "validation_metric": 9.254441465600394e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.811177926998425
        },
        {
          "epoch": 496,
          "updates": 7936,
          "samples_seen": 507904,
          "online_loss": 0.0001059672338215023,
          "train_loss": 0.0001017923787003383,
          "validation_loss": 9.323310223408043e-05,
          "train_metric": 1.025215442496119e-05,
          "validation_metric": 9.390096238348633e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.822190107999631
        },
        {
          "epoch": 497,
          "updates": 7952,
          "samples_seen": 508928,
          "online_loss": 0.00010409797710053681,
          "train_loss": 0.00010196382208960131,
          "validation_loss": 9.475023398408666e-05,
          "train_metric": 1.0269423000863753e-05,
          "validation_metric": 9.542896805214696e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.833247681999637
        },
        {
          "epoch": 498,
          "updates": 7968,
          "samples_seen": 509952,
          "online_loss": 0.00010681657363420527,
          "train_loss": 0.00010150383604923263,
          "validation_loss": 9.267851419281214e-05,
          "train_metric": 1.022309334075544e-05,
          "validation_metric": 9.334239621239249e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.844078375999743
        },
        {
          "epoch": 499,
          "updates": 7984,
          "samples_seen": 510976,
          "online_loss": 0.00010704524220273015,
          "train_loss": 0.00010727487097028643,
          "validation_loss": 9.352699271403253e-05,
          "train_metric": 1.0804333214764483e-05,
          "validation_metric": 9.419695743417833e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.854637057998843
        },
        {
          "epoch": 500,
          "updates": 8000,
          "samples_seen": 512000,
          "online_loss": 0.00011447143469922594,
          "train_loss": 0.00011085635924246162,
          "validation_loss": 0.00010582029790384695,
          "train_metric": 1.1165047908434644e-05,
          "validation_metric": 1.06578318082029e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.865174134998597
        },
        {
          "epoch": 501,
          "updates": 8016,
          "samples_seen": 513024,
          "online_loss": 0.00011612532944127452,
          "train_loss": 0.00010232867498416454,
          "validation_loss": 9.373285138281062e-05,
          "train_metric": 1.0306168405804783e-05,
          "validation_metric": 9.440428584639449e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.875598980999712
        },
        {
          "epoch": 502,
          "updates": 8032,
          "samples_seen": 514048,
          "online_loss": 0.00010692739419937425,
          "train_loss": 0.00010397427104180679,
          "validation_loss": 9.381899144500494e-05,
          "train_metric": 1.0471906534803566e-05,
          "validation_metric": 9.44910425459966e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.886549263999768
        },
        {
          "epoch": 503,
          "updates": 8048,
          "samples_seen": 515072,
          "online_loss": 0.00011093195780631504,
          "train_loss": 0.00010893950093304738,
          "validation_loss": 0.00010212049528490752,
          "train_metric": 1.0971987649099901e-05,
          "validation_metric": 1.0285201824444812e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.897160864999023
        },
        {
          "epoch": 504,
          "updates": 8064,
          "samples_seen": 516096,
          "online_loss": 0.00011024991363228764,
          "train_loss": 0.00010293058585375547,
          "validation_loss": 9.571696864441037e-05,
          "train_metric": 1.036679168464616e-05,
          "validation_metric": 9.640261851018295e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.907599067999399
        },
        {
          "epoch": 505,
          "updates": 8080,
          "samples_seen": 517120,
          "online_loss": 0.0001024493649310898,
          "train_loss": 0.00010135177581105381,
          "validation_loss": 9.096969733946025e-05,
          "train_metric": 1.0207778359472286e-05,
          "validation_metric": 9.162135029328056e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.918482293000125
        },
        {
          "epoch": 506,
          "updates": 8096,
          "samples_seen": 518144,
          "online_loss": 0.00010720245063566836,
          "train_loss": 0.00010868310346268117,
          "validation_loss": 0.00010169761662837118,
          "train_metric": 1.0946164366032463e-05,
          "validation_metric": 1.0242612006550189e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.929153586999746
        },
        {
          "epoch": 507,
          "updates": 8112,
          "samples_seen": 519168,
          "online_loss": 0.00010598144081086502,
          "train_loss": 0.00010088142880704254,
          "validation_loss": 9.119055175688118e-05,
          "train_metric": 1.0160408237425145e-05,
          "validation_metric": 9.184377631754614e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.939626223998857
        },
        {
          "epoch": 508,
          "updates": 8128,
          "samples_seen": 520192,
          "online_loss": 0.00011057373421863304,
          "train_loss": 0.00010142616520170122,
          "validation_loss": 9.254869655705988e-05,
          "train_metric": 1.0215270776825491e-05,
          "validation_metric": 9.321165634901263e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.950043103999633
        },
        {
          "epoch": 509,
          "updates": 8144,
          "samples_seen": 521216,
          "online_loss": 0.00010347069246563478,
          "train_loss": 0.00010039790504379198,
          "validation_loss": 9.127936937147751e-05,
          "train_metric": 1.0111710253113415e-05,
          "validation_metric": 9.193323421641253e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.9603833649998705
        },
        {
          "epoch": 510,
          "updates": 8160,
          "samples_seen": 522240,
          "online_loss": 0.00010163890237890882,
          "train_loss": 0.00010119518992723897,
          "validation_loss": 9.417054388904944e-05,
          "train_metric": 1.0192008630838245e-05,
          "validation_metric": 9.484512702329084e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.970743232999666
        },
        {
          "epoch": 511,
          "updates": 8176,
          "samples_seen": 523264,
          "online_loss": 0.00010620766670399462,
          "train_loss": 0.00010438034951221198,
          "validation_loss": 9.740484529174864e-05,
          "train_metric": 1.0512807421036996e-05,
          "validation_metric": 9.810260962694883e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.036096757999985
        },
        {
          "epoch": 512,
          "updates": 8192,
          "samples_seen": 524288,
          "online_loss": 0.00010405384955447516,
          "train_loss": 0.00010125037078978494,
          "validation_loss": 9.098532609641552e-05,
          "train_metric": 1.0197567462455481e-05,
          "validation_metric": 9.163708455162123e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.046884747998774
        },
        {
          "epoch": 513,
          "updates": 8208,
          "samples_seen": 525312,
          "online_loss": 0.00010459581585564592,
          "train_loss": 0.000107107414805796,
          "validation_loss": 9.884234896162525e-05,
          "train_metric": 1.0787465726025403e-05,
          "validation_metric": 9.95504069578601e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.057493624999552
        },
        {
          "epoch": 514,
          "updates": 8224,
          "samples_seen": 526336,
          "online_loss": 0.00010621772707963828,
          "train_loss": 0.00010015862062573433,
          "validation_loss": 9.099360613618046e-05,
          "train_metric": 1.0087608643516432e-05,
          "validation_metric": 9.164544280793052e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.067976258998897
        },
        {
          "epoch": 515,
          "updates": 8240,
          "samples_seen": 527360,
          "online_loss": 0.00010197801157119102,
          "train_loss": 0.00010009718243964016,
          "validation_loss": 9.162615606328472e-05,
          "train_metric": 1.0081421351060271e-05,
          "validation_metric": 9.228250746673439e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.078483646999302
        },
        {
          "epoch": 516,
          "updates": 8256,
          "samples_seen": 528384,
          "online_loss": 0.00010388433611296932,
          "train_loss": 0.00010047553951153532,
          "validation_loss": 9.347209561383352e-05,
          "train_metric": 1.0119528269569855e-05,
          "validation_metric": 9.414166015631054e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.08901382699878
        },
        {
          "epoch": 517,
          "updates": 8272,
          "samples_seen": 529408,
          "online_loss": 0.00010573495410426403,
          "train_loss": 0.00010348142677685246,
          "validation_loss": 9.251957089873031e-05,
          "train_metric": 1.0422270861454308e-05,
          "validation_metric": 9.318232514488045e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.099475965998863
        },
        {
          "epoch": 518,
          "updates": 8288,
          "samples_seen": 530432,
          "online_loss": 0.00010693675176298711,
          "train_loss": 9.960916213458404e-05,
          "validation_loss": 9.018505807034671e-05,
          "train_metric": 1.0032270438387059e-05,
          "validation_metric": 9.083109034691006e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.110765280000123
        },
        {
          "epoch": 519,
          "updates": 8304,
          "samples_seen": 531456,
          "online_loss": 0.00010514320456422865,
          "train_loss": 0.00010128736175829545,
          "validation_loss": 9.079965093405917e-05,
          "train_metric": 1.0201292752753943e-05,
          "validation_metric": 9.145007425104268e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.121382322999125
        },
        {
          "epoch": 520,
          "updates": 8320,
          "samples_seen": 532480,
          "online_loss": 0.00010459749501023907,
          "train_loss": 0.00010028770338976756,
          "validation_loss": 9.330650937044993e-05,
          "train_metric": 1.010061168926768e-05,
          "validation_metric": 9.397490430274047e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.131973642999583
        },
        {
          "epoch": 521,
          "updates": 8336,
          "samples_seen": 533504,
          "online_loss": 0.00010150673642783659,
          "train_loss": 9.985473298002034e-05,
          "validation_loss": 9.035200491780415e-05,
          "train_metric": 1.005700232781237e-05,
          "validation_metric": 9.099922863242682e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.1429261269986455
        },
        {
          "epoch": 522,
          "updates": 8352,
          "samples_seen": 534528,
          "online_loss": 0.00010141295388166327,
          "train_loss": 0.00010193789785262197,
          "validation_loss": 9.462924936087802e-05,
          "train_metric": 1.0266810932080261e-05,
          "validation_metric": 9.530713214189745e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.153518757999336
        },
        {
          "epoch": 523,
          "updates": 8368,
          "samples_seen": 535552,
          "online_loss": 0.00010056773112410156,
          "train_loss": 0.00010270232451148331,
          "validation_loss": 8.980184065876529e-05,
          "train_metric": 1.0343801477574743e-05,
          "validation_metric": 9.044512808031868e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.164617865999389
        },
        {
          "epoch": 524,
          "updates": 8384,
          "samples_seen": 536576,
          "online_loss": 0.00010359347197663737,
          "train_loss": 9.899048018269241e-05,
          "validation_loss": 9.126096847467124e-05,
          "train_metric": 9.969959137379192e-06,
          "validation_metric": 9.19146987143904e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.175603916999535
        },
        {
          "epoch": 525,
          "updates": 8400,
          "samples_seen": 537600,
          "online_loss": 0.00010382940899944515,
          "train_loss": 0.00010618480882840231,
          "validation_loss": 9.838561527431011e-05,
          "train_metric": 1.069454629032407e-05,
          "validation_metric": 9.909039363265038e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.186463226998967
        },
        {
          "epoch": 526,
          "updates": 8416,
          "samples_seen": 538624,
          "online_loss": 0.00010096061259901035,
          "train_loss": 0.00010063863737741485,
          "validation_loss": 9.027455234900117e-05,
          "train_metric": 1.0135954653378576e-05,
          "validation_metric": 9.092121217690874e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.197154984000008
        },
        {
          "epoch": 527,
          "updates": 8432,
          "samples_seen": 539648,
          "online_loss": 0.00010220915055469959,
          "train_loss": 9.867532935459167e-05,
          "validation_loss": 9.017170668812469e-05,
          "train_metric": 9.938217772287317e-06,
          "validation_metric": 9.081762982532382e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.2076517949990375
        },
        {
          "epoch": 528,
          "updates": 8448,
          "samples_seen": 540672,
          "online_loss": 0.00010039999233413255,
          "train_loss": 9.864813910098746e-05,
          "validation_loss": 9.059489093488082e-05,
          "train_metric": 9.935478374245577e-06,
          "validation_metric": 9.124385542236269e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.218133974000011
        },
        {
          "epoch": 529,
          "updates": 8464,
          "samples_seen": 541696,
          "online_loss": 0.00010067638095279108,
          "train_loss": 9.897027484839782e-05,
          "validation_loss": 9.126215445576236e-05,
          "train_metric": 9.967922778741922e-06,
          "validation_metric": 9.191589924739674e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.228573022999626
        },
        {
          "epoch": 530,
          "updates": 8480,
          "samples_seen": 542720,
          "online_loss": 0.00010170626887884282,
          "train_loss": 9.870663052424788e-05,
          "validation_loss": 9.110987593885511e-05,
          "train_metric": 9.941370990418363e-06,
          "validation_metric": 9.176253115583677e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.239083997999842
        },
        {
          "epoch": 531,
          "updates": 8496,
          "samples_seen": 543744,
          "online_loss": 9.920523416440119e-05,
          "train_loss": 9.832342766458169e-05,
          "validation_loss": 8.930850162869319e-05,
          "train_metric": 9.902774763759226e-06,
          "validation_metric": 8.994824383989908e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.249980704998961
        },
        {
          "epoch": 532,
          "updates": 8512,
          "samples_seen": 544768,
          "online_loss": 9.995780055760406e-05,
          "train_loss": 9.8175230959896e-05,
          "validation_loss": 8.996717224363238e-05,
          "train_metric": 9.887850865197834e-06,
          "validation_metric": 9.061164746526629e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.260732893999375
        },
        {
          "epoch": 533,
          "updates": 8528,
          "samples_seen": 545792,
          "online_loss": 0.00010022971900980338,
          "train_loss": 9.799389954423532e-05,
          "validation_loss": 9.0160989202559e-05,
          "train_metric": 9.869587302091531e-06,
          "validation_metric": 9.080685231310781e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.27140400099961
        },
        {
          "epoch": 534,
          "updates": 8544,
          "samples_seen": 546816,
          "online_loss": 0.00010021905177381996,
          "train_loss": 9.802330168895423e-05,
          "validation_loss": 8.98523285286501e-05,
          "train_metric": 9.872547707345802e-06,
          "validation_metric": 9.04959779290948e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.282137860000148
        },
        {
          "epoch": 535,
          "updates": 8560,
          "samples_seen": 547840,
          "online_loss": 9.857773966359673e-05,
          "train_loss": 0.00010239642870146781,
          "validation_loss": 9.68397653196007e-05,
          "train_metric": 1.0312993254046887e-05,
          "validation_metric": 9.753346603247337e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.292681253999035
        },
        {
          "epoch": 536,
          "updates": 8576,
          "samples_seen": 548864,
          "online_loss": 0.00010136696482732077,
          "train_loss": 9.79384349193424e-05,
          "validation_loss": 8.89160146471113e-05,
          "train_metric": 9.864001185633242e-06,
          "validation_metric": 8.955295015766751e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.303724039998997
        },
        {
          "epoch": 537,
          "updates": 8592,
          "samples_seen": 549888,
          "online_loss": 0.00010401398367321235,
          "train_loss": 0.00010079675848828629,
          "validation_loss": 9.550341928843409e-05,
          "train_metric": 1.0151881724596024e-05,
          "validation_metric": 9.618754120310768e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.314171957999861
        },
        {
          "epoch": 538,
          "updates": 8608,
          "samples_seen": 550912,
          "online_loss": 0.0001053552857683826,
          "train_loss": 0.0001095055922633037,
          "validation_loss": 9.791780757950619e-05,
          "train_metric": 1.1029003871954046e-05,
          "validation_metric": 9.861923899734393e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.324682872998892
        },
        {
          "epoch": 539,
          "updates": 8624,
          "samples_seen": 551936,
          "online_loss": 0.00010133970363312983,
          "train_loss": 9.970674727810547e-05,
          "validation_loss": 8.927696035243571e-05,
          "train_metric": 1.0042099347629119e-05,
          "validation_metric": 8.991648428491317e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.33518694599843
        },
        {
          "epoch": 540,
          "updates": 8640,
          "samples_seen": 552960,
          "online_loss": 0.00010281428922098712,
          "train_loss": 9.835541277425364e-05,
          "validation_loss": 8.919148240238428e-05,
          "train_metric": 9.905998012982309e-06,
          "validation_metric": 8.983040970633738e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.345564743998693
        },
        {
          "epoch": 541,
          "updates": 8656,
          "samples_seen": 553984,
          "online_loss": 9.940247991835349e-05,
          "train_loss": 9.844180021900684e-05,
          "validation_loss": 9.099477756535634e-05,
          "train_metric": 9.91469914879417e-06,
          "validation_metric": 9.164660696114879e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.356259891999798
        },
        {
          "epoch": 542,
          "updates": 8672,
          "samples_seen": 555008,
          "online_loss": 0.00010150665093533462,
          "train_loss": 9.735958155943081e-05,
          "validation_loss": 9.05490669538267e-05,
          "train_metric": 9.805699846765492e-06,
          "validation_metric": 9.119770766119473e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.3666750920001505
        },
        {
          "epoch": 543,
          "updates": 8688,
          "samples_seen": 556032,
          "online_loss": 0.00010181733068748144,
          "train_loss": 9.865777246886864e-05,
          "validation_loss": 9.169720578938723e-05,
          "train_metric": 9.936447895597667e-06,
          "validation_metric": 9.235406650986988e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.3769842789988616
        },
        {
          "epoch": 544,
          "updates": 8704,
          "samples_seen": 557056,
          "online_loss": 0.00010807589933392592,
          "train_loss": 0.00011369019193807617,
          "validation_loss": 0.00010784460027934983,
          "train_metric": 1.1450460078776814e-05,
          "validation_metric": 1.0861714144994039e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.387408290998792
        },
        {
          "epoch": 545,
          "updates": 8720,
          "samples_seen": 558080,
          "online_loss": 0.00010030310750153149,
          "train_loss": 0.000115053633635398,
          "validation_loss": 0.00010002759518101811,
          "train_metric": 1.1587780136324e-05,
          "validation_metric": 1.0074413694383111e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.39782711099906
        },
        {
          "epoch": 546,
          "updates": 8736,
          "samples_seen": 559104,
          "online_loss": 0.00010792704188133939,
          "train_loss": 0.00010297734115738422,
          "validation_loss": 9.952415712177753e-05,
          "train_metric": 1.037150104821194e-05,
          "validation_metric": 1.0023708455264568e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.4083006999990175
        },
        {
          "epoch": 547,
          "updates": 8752,
          "samples_seen": 560128,
          "online_loss": 0.0001017555960061145,
          "train_loss": 9.684026008471847e-05,
          "validation_loss": 9.06848581507802e-05,
          "train_metric": 9.753396625455935e-06,
          "validation_metric": 9.133446837950032e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.474248903999978
        },
        {
          "epoch": 548,
          "updates": 8768,
          "samples_seen": 561152,
          "online_loss": 0.00010144283533009002,
          "train_loss": 9.752303594723344e-05,
          "validation_loss": 8.813713066047058e-05,
          "train_metric": 9.822163519856986e-06,
          "validation_metric": 8.87684836925473e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.485567596999317
        },
        {
          "epoch": 549,
          "updates": 8784,
          "samples_seen": 562176,
          "online_loss": 0.00010067258153867442,
          "train_loss": 0.00010196025687037036,
          "validation_loss": 9.825703455135226e-05,
          "train_metric": 1.0269062840961851e-05,
          "validation_metric": 9.896089068206493e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.496164193998993
        },
        {
          "epoch": 550,
          "updates": 8800,
          "samples_seen": 563200,
          "online_loss": 0.00010539071035964298,
          "train_loss": 9.781535482034087e-05,
          "validation_loss": 8.815723413135856e-05,
          "train_metric": 9.851602953858674e-06,
          "validation_metric": 8.878872904460877e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.506641238998782
        },
        {
          "epoch": 551,
          "updates": 8816,
          "samples_seen": 564224,
          "online_loss": 9.688677300800919e-05,
          "train_loss": 9.94509900920093e-05,
          "validation_loss": 9.239687642548233e-05,
          "train_metric": 1.0016339729190804e-05,
          "validation_metric": 9.305876119469758e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.516975586999251
        },
        {
          "epoch": 552,
          "updates": 8832,
          "samples_seen": 565248,
          "online_loss": 0.0001029543482218287,
          "train_loss": 9.806543675949797e-05,
          "validation_loss": 9.105947538046166e-05,
          "train_metric": 9.876790500129573e-06,
          "validation_metric": 9.171177225653082e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.527535482999156
        },
        {
          "epoch": 553,
          "updates": 8848,
          "samples_seen": 566272,
          "online_loss": 9.898962821353052e-05,
          "train_loss": 9.609927656129003e-05,
          "validation_loss": 9.017703996505588e-05,
          "train_metric": 9.678767128207255e-06,
          "validation_metric": 9.082301403395832e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.537982391999321
        },
        {
          "epoch": 554,
          "updates": 8864,
          "samples_seen": 567296,
          "online_loss": 9.814698432819569e-05,
          "train_loss": 9.597797907190397e-05,
          "validation_loss": 8.914828504202887e-05,
          "train_metric": 9.66654988587834e-06,
          "validation_metric": 8.978689947980456e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.548440290998769
        },
        {
          "epoch": 555,
          "updates": 8880,
          "samples_seen": 568320,
          "online_loss": 0.00010348138630433823,
          "train_loss": 0.00010107370326295495,
          "validation_loss": 9.403412695974112e-05,
          "train_metric": 1.0179773198615294e-05,
          "validation_metric": 9.470772056374699e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.559312856999895
        },
        {
          "epoch": 556,
          "updates": 8896,
          "samples_seen": 569344,
          "online_loss": 9.805548324948177e-05,
          "train_loss": 9.764164133230224e-05,
          "validation_loss": 8.832261664792895e-05,
          "train_metric": 9.834108823270071e-06,
          "validation_metric": 8.895529390429147e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.5697102489993995
        },
        {
          "epoch": 557,
          "updates": 8912,
          "samples_seen": 570368,
          "online_loss": 0.00010220557715001632,
          "train_loss": 0.00010313518578186631,
          "validation_loss": 8.922803681343794e-05,
          "train_metric": 1.0387397196609527e-05,
          "validation_metric": 8.986721695691813e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.580027176998556
        },
        {
          "epoch": 558,
          "updates": 8928,
          "samples_seen": 571392,
          "online_loss": 0.00010295332367604715,
          "train_loss": 0.00011058709060307592,
          "validation_loss": 0.0001058220150298439,
          "train_metric": 1.1137925866933074e-05,
          "validation_metric": 1.0658005521690939e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.590545719000147
        },
        {
          "epoch": 559,
          "updates": 8944,
          "samples_seen": 572416,
          "online_loss": 0.00010693419380913838,
          "train_loss": 0.00010156451025977731,
          "validation_loss": 9.190604032482952e-05,
          "train_metric": 1.0229205145151354e-05,
          "validation_metric": 9.25644053495489e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.600829549999617
        },
        {
          "epoch": 560,
          "updates": 8960,
          "samples_seen": 573440,
          "online_loss": 0.00010257877238473156,
          "train_loss": 9.646772377891466e-05,
          "validation_loss": 8.760243508731946e-05,
          "train_metric": 9.715875421534292e-06,
          "validation_metric": 8.822995368973352e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.611674452999068
        },
        {
          "epoch": 561,
          "updates": 8976,
          "samples_seen": 574464,
          "online_loss": 9.921456148731522e-05,
          "train_loss": 0.00010446501983096823,
          "validation_loss": 9.576368029229343e-05,
          "train_metric": 1.0521333933866117e-05,
          "validation_metric": 9.644966667110566e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.6222051629993075
        },
        {
          "epoch": 562,
          "updates": 8992,
          "samples_seen": 575488,
          "online_loss": 0.00010238737718282209,
          "train_loss": 9.57221636781469e-05,
          "validation_loss": 8.827901910990477e-05,
          "train_metric": 9.640785719966516e-06,
          "validation_metric": 8.89114016899839e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.63260001999879
        },
        {
          "epoch": 563,
          "updates": 9008,
          "samples_seen": 576512,
          "online_loss": 9.883605525828898e-05,
          "train_loss": 9.722571849124506e-05,
          "validation_loss": 8.89346410986036e-05,
          "train_metric": 9.792218406801112e-06,
          "validation_metric": 8.957171303336509e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.642933891998837
        },
        {
          "epoch": 564,
          "updates": 9024,
          "samples_seen": 577536,
          "online_loss": 0.00010004336945712566,
          "train_loss": 0.00010476233728695661,
          "validation_loss": 9.625616803532466e-05,
          "train_metric": 1.055127904692199e-05,
          "validation_metric": 9.69456959865056e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.653306968999459
        },
        {
          "epoch": 565,
          "updates": 9040,
          "samples_seen": 578560,
          "online_loss": 9.855201801656222e-05,
          "train_loss": 0.00011120739509351552,
          "validation_loss": 9.649332787375897e-05,
          "train_metric": 1.1200401786481962e-05,
          "validation_metric": 9.718455658003222e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.663638093999907
        },
        {
          "epoch": 566,
          "updates": 9056,
          "samples_seen": 579584,
          "online_loss": 0.0001021572429635853,
          "train_loss": 0.00010590063902782276,
          "validation_loss": 0.0001038533664541319,
          "train_metric": 1.0665926311048679e-05,
          "validation_metric": 1.0459732038725633e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.673978908000208
        },
        {
          "epoch": 567,
          "updates": 9072,
          "samples_seen": 580608,
          "online_loss": 0.0001006641327876423,
          "train_loss": 9.618644980946556e-05,
          "validation_loss": 8.793924644123763e-05,
          "train_metric": 9.68754648056347e-06,
          "validation_metric": 8.85691861185478e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.684297799998603
        },
        {
          "epoch": 568,
          "updates": 9088,
          "samples_seen": 581632,
          "online_loss": 9.778699359230814e-05,
          "train_loss": 9.8609016276896e-05,
          "validation_loss": 9.513292752671987e-05,
          "train_metric": 9.9315402621869e-06,
          "validation_metric": 9.58144028118113e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.6949473610002315
        },
        {
          "epoch": 569,
          "updates": 9104,
          "samples_seen": 582656,
          "online_loss": 9.60718957685458e-05,
          "train_loss": 0.00010666450543794781,
          "validation_loss": 9.376578236697242e-05,
          "train_metric": 1.074285955837695e-05,
          "validation_metric": 9.443746421311516e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.705663409999033
        },
        {
          "epoch": 570,
          "updates": 9120,
          "samples_seen": 583680,
          "online_loss": 0.00010701073620111856,
          "train_loss": 9.52741174842231e-05,
          "validation_loss": 8.937480743043125e-05,
          "train_metric": 9.59566023084335e-06,
          "validation_metric": 9.001502803585026e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.716228299999784
        },
        {
          "epoch": 571,
          "updates": 9136,
          "samples_seen": 584704,
          "online_loss": 9.798870428312512e-05,
          "train_loss": 9.500834858044982e-05,
          "validation_loss": 8.894635539036244e-05,
          "train_metric": 9.568892892275471e-06,
          "validation_metric": 8.958350008470006e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.726757991998966
        },
        {
          "epoch": 572,
          "updates": 9152,
          "samples_seen": 585728,
          "online_loss": 0.00010367245363340771,
          "train_loss": 9.855251846602187e-05,
          "validation_loss": 8.887319563655183e-05,
          "train_metric": 9.925848644343205e-06,
          "validation_metric": 8.950982191890944e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.737209637998603
        },
        {
          "epoch": 573,
          "updates": 9168,
          "samples_seen": 586752,
          "online_loss": 9.866200662145275e-05,
          "train_loss": 9.698978828964755e-05,
          "validation_loss": 8.63712775753811e-05,
          "train_metric": 9.76845603872789e-06,
          "validation_metric": 8.698997589817736e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.74805247599943
        },
        {
          "epoch": 574,
          "updates": 9184,
          "samples_seen": 587776,
          "online_loss": 9.735655066833715e-05,
          "train_loss": 9.456570842303336e-05,
          "validation_loss": 8.875768980942667e-05,
          "train_metric": 9.524311280983966e-06,
          "validation_metric": 8.939349754655268e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.759093851998841
        },
        {
          "epoch": 575,
          "updates": 9200,
          "samples_seen": 588800,
          "online_loss": 9.62511603574967e-05,
          "train_loss": 9.422653238289058e-05,
          "validation_loss": 8.664513006806374e-05,
          "train_metric": 9.490150659985375e-06,
          "validation_metric": 8.726580745133106e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.769467938998787
        },
        {
          "epoch": 576,
          "updates": 9216,
          "samples_seen": 589824,
          "online_loss": 9.949994910130044e-05,
          "train_loss": 9.499763109488413e-05,
          "validation_loss": 8.838254871079698e-05,
          "train_metric": 9.567814231559169e-06,
          "validation_metric": 8.901566616259515e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.779865925000195
        },
        {
          "epoch": 577,
          "updates": 9232,
          "samples_seen": 590848,
          "online_loss": 9.572014619152469e-05,
          "train_loss": 9.620889613870531e-05,
          "validation_loss": 8.79936633282341e-05,
          "train_metric": 9.689807484392077e-06,
          "validation_metric": 8.862399226927664e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.7903657400001975
        },
        {
          "epoch": 578,
          "updates": 9248,
          "samples_seen": 591872,
          "online_loss": 9.777904733709875e-05,
          "train_loss": 9.529698581900448e-05,
          "validation_loss": 9.076823334908113e-05,
          "train_metric": 9.59796307142824e-06,
          "validation_metric": 9.1418432930368e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.800704055998722
        },
        {
          "epoch": 579,
          "updates": 9264,
          "samples_seen": 592896,
          "online_loss": 0.00010134031253983267,
          "train_loss": 9.903224417939782e-05,
          "validation_loss": 8.804093522485346e-05,
          "train_metric": 9.974166459869593e-06,
          "validation_metric": 8.867160431691445e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.811024417000226
        },
        {
          "epoch": 580,
          "updates": 9280,
          "samples_seen": 593920,
          "online_loss": 0.00010766671948658768,
          "train_loss": 0.00010309027129551396,
          "validation_loss": 9.098538430407643e-05,
          "train_metric": 1.038287427945761e-05,
          "validation_metric": 9.163713912130333e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.821483093999632
        },
        {
          "epoch": 581,
          "updates": 9296,
          "samples_seen": 594944,
          "online_loss": 9.665904030953243e-05,
          "train_loss": 0.00010277558612870052,
          "validation_loss": 9.638829942559823e-05,
          "train_metric": 1.0351180208090227e-05,
          "validation_metric": 9.7078764156322e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.831801873999211
        },
        {
          "epoch": 582,
          "updates": 9312,
          "samples_seen": 595968,
          "online_loss": 0.00011126339177280897,
          "train_loss": 9.574685100233182e-05,
          "validation_loss": 8.892495679901913e-05,
          "train_metric": 9.643272278481163e-06,
          "validation_metric": 8.956195415521506e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.842259961998934
        },
        {
          "epoch": 583,
          "updates": 9328,
          "samples_seen": 596992,
          "online_loss": 0.0001192656945931958,
          "train_loss": 0.00010848245437955484,
          "validation_loss": 9.42119222600013e-05,
          "train_metric": 1.0925954484264366e-05,
          "validation_metric": 9.488678188063204e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.908818936999523
        },
        {
          "epoch": 584,
          "updates": 9344,
          "samples_seen": 598016,
          "online_loss": 0.0001065383169134293,
          "train_loss": 9.506243804935366e-05,
          "validation_loss": 8.897275984054431e-05,
          "train_metric": 9.574340765539091e-06,
          "validation_metric": 8.961011189967394e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.919578850998732
        },
        {
          "epoch": 585,
          "updates": 9360,
          "samples_seen": 599040,
          "online_loss": 9.588773377799953e-05,
          "train_loss": 9.37131408136338e-05,
          "validation_loss": 8.568362682126462e-05,
          "train_metric": 9.438444976694882e-06,
          "validation_metric": 8.629742296761833e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.9305874799993035
        },
        {
          "epoch": 586,
          "updates": 9376,
          "samples_seen": 600064,
          "online_loss": 9.743571581566357e-05,
          "train_loss": 9.748277079779655e-05,
          "validation_loss": 8.838574285618961e-05,
          "train_metric": 9.81810808298178e-06,
          "validation_metric": 8.901887667889241e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.941400930998498
        },
        {
          "epoch": 587,
          "updates": 9392,
          "samples_seen": 601088,
          "online_loss": 9.615982435207115e-05,
          "train_loss": 9.911711094900966e-05,
          "validation_loss": 9.223037341143936e-05,
          "train_metric": 9.982712981582154e-06,
          "validation_metric": 9.289105037169065e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.951885458998731
        },
        {
          "epoch": 588,
          "updates": 9408,
          "samples_seen": 602112,
          "online_loss": 9.75485895651218e-05,
          "train_loss": 9.315174247603863e-05,
          "validation_loss": 8.59846913954243e-05,
          "train_metric": 9.381903510075063e-06,
          "validation_metric": 8.660063031129539e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.9625242659985815
        },
        {
          "epoch": 589,
          "updates": 9424,
          "samples_seen": 603136,
          "online_loss": 9.600329622116988e-05,
          "train_loss": 9.298672375734895e-05,
          "validation_loss": 8.552245708415285e-05,
          "train_metric": 9.365283403894864e-06,
          "validation_metric": 8.613508725829888e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.973392545000024
        },
        {
          "epoch": 590,
          "updates": 9440,
          "samples_seen": 604160,
          "online_loss": 9.68818526416726e-05,
          "train_loss": 9.401634451933205e-05,
          "validation_loss": 8.90676019480452e-05,
          "train_metric": 9.468983080296312e-06,
          "validation_metric": 8.970562703325413e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.983805460999065
        },
        {
          "epoch": 591,
          "updates": 9456,
          "samples_seen": 605184,
          "online_loss": 9.511011512586265e-05,
          "train_loss": 9.31921458686702e-05,
          "validation_loss": 8.573467493988574e-05,
          "train_metric": 9.385971679876093e-06,
          "validation_metric": 8.634881851321552e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.994250484998702
        },
        {
          "epoch": 592,
          "updates": 9472,
          "samples_seen": 606208,
          "online_loss": 0.00010178882894251728,
          "train_loss": 9.433701052330434e-05,
          "validation_loss": 9.013687667902559e-05,
          "train_metric": 9.50128014665097e-06,
          "validation_metric": 9.078256880457047e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.004617389999112
        },
        {
          "epoch": 593,
          "updates": 9488,
          "samples_seen": 607232,
          "online_loss": 0.00010605895340631832,
          "train_loss": 9.490599040873349e-05,
          "validation_loss": 8.594889368396252e-05,
          "train_metric": 9.558583769830875e-06,
          "validation_metric": 8.65645688463701e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.0148197540002
        },
        {
          "epoch": 594,
          "updates": 9504,
          "samples_seen": 608256,
          "online_loss": 9.567628285367391e-05,
          "train_loss": 9.642540680943057e-05,
          "validation_loss": 9.155106818070635e-05,
          "train_metric": 9.711613529361784e-06,
          "validation_metric": 9.220688298228197e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.025301505998868
        },
        {
          "epoch": 595,
          "updates": 9520,
          "samples_seen": 609280,
          "online_loss": 9.498875533608953e-05,
          "train_loss": 9.67456289799884e-05,
          "validation_loss": 8.688517846167088e-05,
          "train_metric": 9.743864211486652e-06,
          "validation_metric": 8.750756933295634e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.035694498999874
        },
        {
          "epoch": 596,
          "updates": 9536,
          "samples_seen": 610304,
          "online_loss": 9.779197262105299e-05,
          "train_loss": 9.23818806768395e-05,
          "validation_loss": 8.60963118611835e-05,
          "train_metric": 9.304364539275412e-06,
          "validation_metric": 8.671305295138154e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.0460333449991595
        },
        {
          "epoch": 597,
          "updates": 9552,
          "samples_seen": 611328,
          "online_loss": 9.791929960556445e-05,
          "train_loss": 9.647013212088495e-05,
          "validation_loss": 9.046117338584736e-05,
          "train_metric": 9.716118256619666e-06,
          "validation_metric": 9.110917744692415e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.05677127799936
        },
        {
          "epoch": 598,
          "updates": 9568,
          "samples_seen": 612352,
          "online_loss": 9.832452087721322e-05,
          "train_loss": 9.64096252573654e-05,
          "validation_loss": 8.613611134933308e-05,
          "train_metric": 9.710024642117787e-06,
          "validation_metric": 8.675313438288867e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.06723713700012
        },
        {
          "epoch": 599,
          "updates": 9584,
          "samples_seen": 613376,
          "online_loss": 0.0001010762416626676,
          "train_loss": 9.339855023426935e-05,
          "validation_loss": 8.473948400933295e-05,
          "train_metric": 9.406760909769218e-06,
          "validation_metric": 8.534650987712666e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.078110022999681
        },
        {
          "epoch": 600,
          "updates": 9600,
          "samples_seen": 614400,
          "online_loss": 9.834883940129657e-05,
          "train_loss": 9.327162115368992e-05,
          "validation_loss": 8.939808321883902e-05,
          "train_metric": 9.393976142746396e-06,
          "validation_metric": 9.003846571431495e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.088834481999584
        },
        {
          "epoch": 601,
          "updates": 9616,
          "samples_seen": 615424,
          "online_loss": 9.491023865848547e-05,
          "train_loss": 9.535873687127605e-05,
          "validation_loss": 8.856372733134776e-05,
          "train_metric": 9.604183105693664e-06,
          "validation_metric": 8.919814717955887e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.099257670999577
        },
        {
          "epoch": 602,
          "updates": 9632,
          "samples_seen": 616448,
          "online_loss": 9.94766896837973e-05,
          "train_loss": 9.197568579111248e-05,
          "validation_loss": 8.524537406628951e-05,
          "train_metric": 9.263453648600262e-06,
          "validation_metric": 8.585600880905986e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.109858002999317
        },
        {
          "epoch": 603,
          "updates": 9648,
          "samples_seen": 617472,
          "online_loss": 9.698363851384784e-05,
          "train_loss": 0.00010212189226876944,
          "validation_loss": 8.949295443017036e-05,
          "train_metric": 1.0285343705618288e-05,
          "validation_metric": 9.013404451252427e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.1202477179995185
        },
        {
          "epoch": 604,
          "updates": 9664,
          "samples_seen": 618496,
          "online_loss": 0.00010083824463436031,
          "train_loss": 9.345208673039451e-05,
          "validation_loss": 8.753201109357178e-05,
          "train_metric": 9.41215330385603e-06,
          "validation_metric": 8.815903129288927e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.130925928999204
        },
        {
          "epoch": 605,
          "updates": 9680,
          "samples_seen": 619520,
          "online_loss": 9.864989260677248e-05,
          "train_loss": 9.831655188463628e-05,
          "validation_loss": 9.280710946768522e-05,
          "train_metric": 9.902081728796475e-06,
          "validation_metric": 9.347192644781899e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.141706088999854
        },
        {
          "epoch": 606,
          "updates": 9696,
          "samples_seen": 620544,
          "online_loss": 9.679661116024363e-05,
          "train_loss": 9.178277105093002e-05,
          "validation_loss": 8.471666660625488e-05,
          "train_metric": 9.244025022780988e-06,
          "validation_metric": 8.532352694601286e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.15291546699882
        },
        {
          "epoch": 607,
          "updates": 9712,
          "samples_seen": 621568,
          "online_loss": 9.3042511252861e-05,
          "train_loss": 9.19140875339508e-05,
          "validation_loss": 8.492778579238802e-05,
          "train_metric": 9.257251804228872e-06,
          "validation_metric": 8.553615771234035e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.163724188998458
        },
        {
          "epoch": 608,
          "updates": 9728,
          "samples_seen": 622592,
          "online_loss": 9.828790507526719e-05,
          "train_loss": 0.00010627474694047123,
          "validation_loss": 9.745841089170426e-05,
          "train_metric": 1.0703603038564324e-05,
          "validation_metric": 9.815654266276397e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.174192253000001
        },
        {
          "epoch": 609,
          "updates": 9744,
          "samples_seen": 623616,
          "online_loss": 0.00010471865448380413,
          "train_loss": 9.268929716199636e-05,
          "validation_loss": 8.608611824456602e-05,
          "train_metric": 9.335325557913166e-06,
          "validation_metric": 8.670276656630449e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.184443361999001
        },
        {
          "epoch": 610,
          "updates": 9760,
          "samples_seen": 624640,
          "online_loss": 9.525343648419948e-05,
          "train_loss": 9.142708586296067e-05,
          "validation_loss": 8.489524043397978e-05,
          "train_metric": 9.208200935972854e-06,
          "validation_metric": 8.550337042834144e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.1948653330000525
        },
        {
          "epoch": 611,
          "updates": 9776,
          "samples_seen": 625664,
          "online_loss": 9.769161988515407e-05,
          "train_loss": 9.183924703393131e-05,
          "validation_loss": 8.545800665160641e-05,
          "train_metric": 9.249712093151174e-06,
          "validation_metric": 8.607017662143335e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.2051461589999235
        },
        {
          "epoch": 612,
          "updates": 9792,
          "samples_seen": 626688,
          "online_loss": 0.00010474980194885575,
          "train_loss": 0.00010191478941123933,
          "validation_loss": 9.706392302177846e-05,
          "train_metric": 1.0264484444633126e-05,
          "validation_metric": 9.775923899724148e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.215438507999352
        },
        {
          "epoch": 613,
          "updates": 9808,
          "samples_seen": 627712,
          "online_loss": 9.842583267527516e-05,
          "train_loss": 9.153712016995996e-05,
          "validation_loss": 8.58282801345922e-05,
          "train_metric": 9.219283128913958e-06,
          "validation_metric": 8.64430967340013e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.225886180000089
        },
        {
          "epoch": 614,
          "updates": 9824,
          "samples_seen": 628736,
          "online_loss": 9.284889165428467e-05,
          "train_loss": 9.201851935358718e-05,
          "validation_loss": 8.500272815581411e-05,
          "train_metric": 9.26776738197077e-06,
          "validation_metric": 8.561162758269347e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.236299882999447
        },
        {
          "epoch": 615,
          "updates": 9840,
          "samples_seen": 629760,
          "online_loss": 9.557313569530379e-05,
          "train_loss": 9.263878018828109e-05,
          "validation_loss": 8.492147026117891e-05,
          "train_metric": 9.330237844551448e-06,
          "validation_metric": 8.552980034437496e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.246574028999021
        },
        {
          "epoch": 616,
          "updates": 9856,
          "samples_seen": 630784,
          "online_loss": 9.731991735861811e-05,
          "train_loss": 9.239226346835494e-05,
          "validation_loss": 8.38581909192726e-05,
          "train_metric": 9.30541045818245e-06,
          "validation_metric": 8.445888852293137e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.257750098999168
        },
        {
          "epoch": 617,
          "updates": 9872,
          "samples_seen": 631808,
          "online_loss": 9.292201616517559e-05,
          "train_loss": 9.622938523534685e-05,
          "validation_loss": 9.282668179366738e-05,
          "train_metric": 9.6918711278704e-06,
          "validation_metric": 9.349162610305939e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.2682040499985305
        },
        {
          "epoch": 618,
          "updates": 9888,
          "samples_seen": 632832,
          "online_loss": 0.00010387021188762446,
          "train_loss": 9.084245539270341e-05,
          "validation_loss": 8.473193156532943e-05,
          "train_metric": 9.149318429990672e-06,
          "validation_metric": 8.533889740647282e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.278594584999155
        },
        {
          "epoch": 619,
          "updates": 9904,
          "samples_seen": 633856,
          "online_loss": 9.592817082193505e-05,
          "train_loss": 9.131933620665222e-05,
          "validation_loss": 8.538087422493845e-05,
          "train_metric": 9.197347026201896e-06,
          "validation_metric": 8.59924875840079e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.347570545000053
        },
        {
          "epoch": 620,
          "updates": 9920,
          "samples_seen": 634880,
          "online_loss": 0.00010034285560323042,
          "train_loss": 9.082058386411518e-05,
          "validation_loss": 8.346154208993539e-05,
          "train_metric": 9.147118362307083e-06,
          "validation_metric": 8.405940207012463e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.359202902000106
        },
        {
          "epoch": 621,
          "updates": 9936,
          "samples_seen": 635904,
          "online_loss": 0.00010604477006381785,
          "train_loss": 0.00011520222324179485,
          "validation_loss": 9.764075366547331e-05,
          "train_metric": 1.1602746781136375e-05,
          "validation_metric": 9.834020602284e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.3696711149987095
        },
        {
          "epoch": 622,
          "updates": 9952,
          "samples_seen": 636928,
          "online_loss": 0.00010502987333893543,
          "train_loss": 9.159048204310238e-05,
          "validation_loss": 8.660916500957683e-05,
          "train_metric": 9.224658242601436e-06,
          "validation_metric": 8.722958227735944e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.380128719998538
        },
        {
          "epoch": 623,
          "updates": 9968,
          "samples_seen": 637952,
          "online_loss": 9.598001543054124e-05,
          "train_loss": 9.220532956533134e-05,
          "validation_loss": 8.689041715115309e-05,
          "train_metric": 9.28658391785575e-06,
          "validation_metric": 8.75128353072796e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.390699005998613
        },
        {
          "epoch": 624,
          "updates": 9984,
          "samples_seen": 638976,
          "online_loss": 9.265096264243766e-05,
          "train_loss": 9.148653771262616e-05,
          "validation_loss": 8.377131598535925e-05,
          "train_metric": 9.214189049089327e-06,
          "validation_metric": 8.437140422756784e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.401101319999725
        },
        {
          "epoch": 625,
          "updates": 10000,
          "samples_seen": 640000,
          "online_loss": 9.2133748694323e-05,
          "train_loss": 9.213601151714101e-05,
          "validation_loss": 8.777822949923575e-05,
          "train_metric": 9.279600817535538e-06,
          "validation_metric": 8.840701411827467e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.411295434998465
        },
        {
          "epoch": 626,
          "updates": 10016,
          "samples_seen": 641024,
          "online_loss": 9.598222823115066e-05,
          "train_loss": 9.468539792578667e-05,
          "validation_loss": 8.988394256448373e-05,
          "train_metric": 9.536366633255966e-06,
          "validation_metric": 9.052781024365686e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.421601467000073
        },
        {
          "epoch": 627,
          "updates": 10032,
          "samples_seen": 642048,
          "online_loss": 9.50053988617583e-05,
          "train_loss": 9.067232895176858e-05,
          "validation_loss": 8.34305610624142e-05,
          "train_metric": 9.132184459303971e-06,
          "validation_metric": 8.402820640185382e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.432336695999766
        },
        {
          "epoch": 628,
          "updates": 10048,
          "samples_seen": 643072,
          "online_loss": 9.276616583520081e-05,
          "train_loss": 9.028283238876611e-05,
          "validation_loss": 8.410912414547056e-05,
          "train_metric": 9.092957043321803e-06,
          "validation_metric": 8.471163710055407e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.442757844999505
        },
        {
          "epoch": 629,
          "updates": 10064,
          "samples_seen": 644096,
          "online_loss": 9.513455142950988e-05,
          "train_loss": 9.094842243939638e-05,
          "validation_loss": 8.493693894706666e-05,
          "train_metric": 9.159992259810679e-06,
          "validation_metric": 8.554537089366931e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.4530842379990645
        },
        {
          "epoch": 630,
          "updates": 10080,
          "samples_seen": 645120,
          "online_loss": 9.106202264774765e-05,
          "train_loss": 9.025355393532664e-05,
          "validation_loss": 8.305413939524442e-05,
          "train_metric": 9.090008461498655e-06,
          "validation_metric": 8.364908353541978e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.463908985999296
        },
        {
          "epoch": 631,
          "updates": 10096,
          "samples_seen": 646144,
          "online_loss": 9.37704166972253e-05,
          "train_loss": 9.244180546374992e-05,
          "validation_loss": 8.43932357383892e-05,
          "train_metric": 9.310400855611078e-06,
          "validation_metric": 8.499777322867885e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.474247635000211
        },
        {
          "epoch": 632,
          "updates": 10112,
          "samples_seen": 647168,
          "online_loss": 9.448158471059287e-05,
          "train_loss": 9.383443830301985e-05,
          "validation_loss": 8.915494254324585e-05,
          "train_metric": 9.450660400034394e-06,
          "validation_metric": 8.979359336080961e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.4848675439989165
        },
        {
          "epoch": 633,
          "updates": 10128,
          "samples_seen": 648192,
          "online_loss": 9.39373148867162e-05,
          "train_loss": 8.974178490461782e-05,
          "validation_loss": 8.351962605956942e-05,
          "train_metric": 9.038463758770376e-06,
          "validation_metric": 8.411790986428969e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.495159614998556
        },
        {
          "epoch": 634,
          "updates": 10144,
          "samples_seen": 649216,
          "online_loss": 9.432840033696266e-05,
          "train_loss": 9.047886123880744e-05,
          "validation_loss": 8.286433148896322e-05,
          "train_metric": 9.112698535318486e-06,
          "validation_metric": 8.345792593900114e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.505956512999546
        },
        {
          "epoch": 635,
          "updates": 10160,
          "samples_seen": 650240,
          "online_loss": 9.184384862237494e-05,
          "train_loss": 8.954912482295185e-05,
          "validation_loss": 8.255658758571371e-05,
          "train_metric": 9.019060598802753e-06,
          "validation_metric": 8.314797014463693e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.516804900998977
        },
        {
          "epoch": 636,
          "updates": 10176,
          "samples_seen": 651264,
          "online_loss": 9.1315081817811e-05,
          "train_loss": 9.081331518245861e-05,
          "validation_loss": 8.376724872505292e-05,
          "train_metric": 9.146384400082752e-06,
          "validation_metric": 8.436731150140986e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.527277489998596
        },
        {
          "epoch": 637,
          "updates": 10192,
          "samples_seen": 652288,
          "online_loss": 9.093206017496414e-05,
          "train_loss": 8.929717296268791e-05,
          "validation_loss": 8.413958858000115e-05,
          "train_metric": 8.993682968139183e-06,
          "validation_metric": 8.474231435684487e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.537581365999358
        },
        {
          "epoch": 638,
          "updates": 10208,
          "samples_seen": 653312,
          "online_loss": 9.049907589542272e-05,
          "train_loss": 8.933862409321591e-05,
          "validation_loss": 8.324504597112536e-05,
          "train_metric": 8.997858458315022e-06,
          "validation_metric": 8.384135071537457e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.547855492999588
        },
        {
          "epoch": 639,
          "updates": 10224,
          "samples_seen": 654336,
          "online_loss": 9.142870339928777e-05,
          "train_loss": 8.949649054557085e-05,
          "validation_loss": 8.396644989261404e-05,
          "train_metric": 9.013758244691417e-06,
          "validation_metric": 8.456792784272693e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.558338438999272
        },
        {
          "epoch": 640,
          "updates": 10240,
          "samples_seen": 655360,
          "online_loss": 8.955285898082366e-05,
          "train_loss": 9.368506289320067e-05,
          "validation_loss": 8.480373799102381e-05,
          "train_metric": 9.435615538677666e-06,
          "validation_metric": 8.54112204251578e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.568645471999844
        },
        {
          "epoch": 641,
          "updates": 10256,
          "samples_seen": 656384,
          "online_loss": 9.973545957109309e-05,
          "train_loss": 9.183849761029705e-05,
          "validation_loss": 8.510695624863729e-05,
          "train_metric": 9.249637514585629e-06,
          "validation_metric": 8.57166014611721e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.578909381998528
        },
        {
          "epoch": 642,
          "updates": 10272,
          "samples_seen": 657408,
          "online_loss": 8.911237841857655e-05,
          "train_loss": 8.981207065517083e-05,
          "validation_loss": 8.258424350060523e-05,
          "train_metric": 9.045542356034275e-06,
          "validation_metric": 8.317582796735223e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.589240248998976
        },
        {
          "epoch": 643,
          "updates": 10288,
          "samples_seen": 658432,
          "online_loss": 9.282401651944383e-05,
          "train_loss": 8.929600880946964e-05,
          "validation_loss": 8.330932905664667e-05,
          "train_metric": 8.993567462312058e-06,
          "validation_metric": 8.39061067381408e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.5994854849996045
        },
        {
          "epoch": 644,
          "updates": 10304,
          "samples_seen": 659456,
          "online_loss": 9.268138319384889e-05,
          "train_loss": 9.706610580906272e-05,
          "validation_loss": 9.117137233261019e-05,
          "train_metric": 9.776142178452574e-06,
          "validation_metric": 9.18244677450275e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.609791939999923
        },
        {
          "epoch": 645,
          "updates": 10320,
          "samples_seen": 660480,
          "online_loss": 0.00010163447291233751,
          "train_loss": 9.058863361133263e-05,
          "validation_loss": 8.31601646495983e-05,
          "train_metric": 9.123756171902642e-06,
          "validation_metric": 8.375586730835494e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.620074540998758
        },
        {
          "epoch": 646,
          "updates": 10336,
          "samples_seen": 661504,
          "online_loss": 9.439125892640732e-05,
          "train_loss": 8.935660298448056e-05,
          "validation_loss": 8.367771079065278e-05,
          "train_metric": 8.999670171760954e-06,
          "validation_metric": 8.427711691183504e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.6303848899988225
        },
        {
          "epoch": 647,
          "updates": 10352,
          "samples_seen": 662528,
          "online_loss": 9.356080090583418e-05,
          "train_loss": 0.00010528571874601766,
          "validation_loss": 8.946030720835552e-05,
          "train_metric": 1.0603992450342048e-05,
          "validation_metric": 9.010115718410816e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.64055076600016
        },
        {
          "epoch": 648,
          "updates": 10368,
          "samples_seen": 663552,
          "online_loss": 9.167067469206813e-05,
          "train_loss": 9.921516902977601e-05,
          "validation_loss": 9.489526564721018e-05,
          "train_metric": 9.992589184548706e-06,
          "validation_metric": 9.557503290125169e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.650892894000208
        },
        {
          "epoch": 649,
          "updates": 10384,
          "samples_seen": 664576,
          "online_loss": 9.247571870218962e-05,
          "train_loss": 9.99697731458582e-05,
          "validation_loss": 8.876276842784137e-05,
          "train_metric": 1.0068589290312957e-05,
          "validation_metric": 8.939861800172366e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.6611405160001596
        },
        {
          "epoch": 650,
          "updates": 10400,
          "samples_seen": 665600,
          "online_loss": 9.219955609296449e-05,
          "train_loss": 9.272876923205331e-05,
          "validation_loss": 8.766191604081541e-05,
          "train_metric": 9.339302778244019e-06,
          "validation_metric": 8.828987120068632e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.671481391998896
        },
        {
          "epoch": 651,
          "updates": 10416,
          "samples_seen": 666624,
          "online_loss": 9.036754363478394e-05,
          "train_loss": 9.165847586700693e-05,
          "validation_loss": 8.291417907457799e-05,
          "train_metric": 9.231507647200488e-06,
          "validation_metric": 8.350812095159199e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.6819335330001195
        },
        {
          "epoch": 652,
          "updates": 10432,
          "samples_seen": 667648,
          "online_loss": 9.31405697883747e-05,
          "train_loss": 8.826474368106574e-05,
          "validation_loss": 8.302554488182068e-05,
          "train_metric": 8.889701348380186e-06,
          "validation_metric": 8.362027983821463e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.692343204998906
        },
        {
          "epoch": 653,
          "updates": 10448,
          "samples_seen": 668672,
          "online_loss": 9.223767165167374e-05,
          "train_loss": 8.813905151328072e-05,
          "validation_loss": 8.250293467426673e-05,
          "train_metric": 8.877042091626208e-06,
          "validation_metric": 8.30939370644046e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.703150846999051
        },
        {
          "epoch": 654,
          "updates": 10464,
          "samples_seen": 669696,
          "online_loss": 9.074792569663259e-05,
          "train_loss": 8.841247472446412e-05,
          "validation_loss": 8.313176658703014e-05,
          "train_metric": 8.90458068170119e-06,
          "validation_metric": 8.37272727949312e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.713625857999432
        },
        {
          "epoch": 655,
          "updates": 10480,
          "samples_seen": 670720,
          "online_loss": 9.435388483325369e-05,
          "train_loss": 9.547434456180781e-05,
          "validation_loss": 8.771822467679158e-05,
          "train_metric": 9.615826456865761e-06,
          "validation_metric": 8.834657819534186e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.778525064999485
        },
        {
          "epoch": 656,
          "updates": 10496,
          "samples_seen": 671744,
          "online_loss": 9.607305605641159e-05,
          "train_loss": 8.79459039424546e-05,
          "validation_loss": 8.27044787001796e-05,
          "train_metric": 8.857589818944689e-06,
          "validation_metric": 8.329691809194628e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.7893351599996095
        },
        {
          "epoch": 657,
          "updates": 10512,
          "samples_seen": 672768,
          "online_loss": 9.614861505724548e-05,
          "train_loss": 0.00011186923075001687,
          "validation_loss": 9.129571117227897e-05,
          "train_metric": 1.12670586531749e-05,
          "validation_metric": 9.194970516546164e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.799667766999846
        },
        {
          "epoch": 658,
          "updates": 10528,
          "samples_seen": 673792,
          "online_loss": 9.472121428188984e-05,
          "train_loss": 0.0001096968335332349,
          "validation_loss": 9.799226245377213e-05,
          "train_metric": 1.1048262422264088e-05,
          "validation_metric": 9.86942268355051e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.8100268829985
        },
        {
          "epoch": 659,
          "updates": 10544,
          "samples_seen": 674816,
          "online_loss": 0.0001112380650738487,
          "train_loss": 9.061401942744851e-05,
          "validation_loss": 8.776047616265714e-05,
          "train_metric": 9.126313671004027e-06,
          "validation_metric": 8.838914254738484e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.820399045000158
        },
        {
          "epoch": 660,
          "updates": 10560,
          "samples_seen": 675840,
          "online_loss": 0.00010510454330869834,
          "train_loss": 0.00011838987120427191,
          "validation_loss": 9.68739841482602e-05,
          "train_metric": 1.1923792953894008e-05,
          "validation_metric": 9.756793588167056e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.83075299399934
        },
        {
          "epoch": 661,
          "updates": 10576,
          "samples_seen": 676864,
          "online_loss": 0.00010064002526632976,
          "train_loss": 8.787005936028436e-05,
          "validation_loss": 8.292069105664268e-05,
          "train_metric": 8.8499518824392e-06,
          "validation_metric": 8.351468750333879e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.841017502998511
        },
        {
          "epoch": 662,
          "updates": 10592,
          "samples_seen": 677888,
          "online_loss": 8.831728973746067e-05,
          "train_loss": 9.010570647660643e-05,
          "validation_loss": 8.158280252246186e-05,
          "train_metric": 9.075117304746527e-06,
          "validation_metric": 8.216720743803307e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.852001127999756
        },
        {
          "epoch": 663,
          "updates": 10608,
          "samples_seen": 678912,
          "online_loss": 9.322040386905428e-05,
          "train_loss": 8.891762990970165e-05,
          "validation_loss": 8.255559077952057e-05,
          "train_metric": 8.95545872481307e-06,
          "validation_metric": 8.314696970046498e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.862400565998541
        },
        {
          "epoch": 664,
          "updates": 10624,
          "samples_seen": 679936,
          "online_loss": 9.106972311201389e-05,
          "train_loss": 8.827874262351543e-05,
          "validation_loss": 8.124974556267262e-05,
          "train_metric": 8.891111065167934e-06,
          "validation_metric": 8.183175850717817e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.873343084998851
        },
        {
          "epoch": 665,
          "updates": 10640,
          "samples_seen": 680960,
          "online_loss": 9.070437795344333e-05,
          "train_loss": 8.750853885430843e-05,
          "validation_loss": 8.144367893692106e-05,
          "train_metric": 8.813539352559019e-06,
          "validation_metric": 8.202709068427794e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.883782778999375
        },
        {
          "epoch": 666,
          "updates": 10656,
          "samples_seen": 681984,
          "online_loss": 9.283889744438056e-05,
          "train_loss": 8.772619185037911e-05,
          "validation_loss": 8.350580174010247e-05,
          "train_metric": 8.835460903355852e-06,
          "validation_metric": 8.410397640545852e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.894316047999382
        },
        {
          "epoch": 667,
          "updates": 10672,
          "samples_seen": 683008,
          "online_loss": 9.583088740328094e-05,
          "train_loss": 9.034868708113208e-05,
          "validation_loss": 8.530064951628447e-05,
          "train_metric": 9.099589078687131e-06,
          "validation_metric": 8.59116880747024e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.904574087999208
        },
        {
          "epoch": 668,
          "updates": 10688,
          "samples_seen": 684032,
          "online_loss": 8.845949105307227e-05,
          "train_loss": 8.743082435103133e-05,
          "validation_loss": 8.086969319265336e-05,
          "train_metric": 8.805712241155561e-06,
          "validation_metric": 8.144898856699001e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.915785706998577
        },
        {
          "epoch": 669,
          "updates": 10704,
          "samples_seen": 685056,
          "online_loss": 8.908990071176959e-05,
          "train_loss": 8.700168109498918e-05,
          "validation_loss": 8.212101965909824e-05,
          "train_metric": 8.762490324443206e-06,
          "validation_metric": 8.270928447018377e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.92617600199992
        },
        {
          "epoch": 670,
          "updates": 10720,
          "samples_seen": 686080,
          "online_loss": 9.084355883715034e-05,
          "train_loss": 8.822290692478418e-05,
          "validation_loss": 8.21357753011398e-05,
          "train_metric": 8.885487659426872e-06,
          "validation_metric": 8.272414561361074e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.936483945999498
        },
        {
          "epoch": 671,
          "updates": 10736,
          "samples_seen": 687104,
          "online_loss": 9.273981140722753e-05,
          "train_loss": 8.751778659643605e-05,
          "validation_loss": 8.194272231776267e-05,
          "train_metric": 8.814471584628336e-06,
          "validation_metric": 8.252970474131871e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.946709252999426
        },
        {
          "epoch": 672,
          "updates": 10752,
          "samples_seen": 688128,
          "online_loss": 8.918415323933004e-05,
          "train_loss": 8.849964069668204e-05,
          "validation_loss": 8.367568807443604e-05,
          "train_metric": 8.913359124562703e-06,
          "validation_metric": 8.427508873865008e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.957194732998687
        },
        {
          "epoch": 673,
          "updates": 10768,
          "samples_seen": 689152,
          "online_loss": 9.76404371613171e-05,
          "train_loss": 8.75411496963352e-05,
          "validation_loss": 8.341600914718583e-05,
          "train_metric": 8.816825356916524e-06,
          "validation_metric": 8.401354534726124e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.967417668998678
        },
        {
          "epoch": 674,
          "updates": 10784,
          "samples_seen": 690176,
          "online_loss": 9.624012568565377e-05,
          "train_loss": 0.00010501644283067435,
          "validation_loss": 8.612398960394785e-05,
          "train_metric": 1.0576872227829881e-05,
          "validation_metric": 8.674091986904386e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.977707995998571
        },
        {
          "epoch": 675,
          "updates": 10800,
          "samples_seen": 691200,
          "online_loss": 9.199659007208538e-05,
          "train_loss": 8.807062840787694e-05,
          "validation_loss": 8.356595935765654e-05,
          "train_metric": 8.870150850270875e-06,
          "validation_metric": 8.416456694249064e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.987924346
        },
        {
          "epoch": 676,
          "updates": 10816,
          "samples_seen": 692224,
          "online_loss": 8.704982792551164e-05,
          "train_loss": 8.82584135979414e-05,
          "validation_loss": 7.986623677425086e-05,
          "train_metric": 8.889063792594243e-06,
          "validation_metric": 8.043834895943291e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.998737936999532
        },
        {
          "epoch": 677,
          "updates": 10832,
          "samples_seen": 693248,
          "online_loss": 9.038331927513354e-05,
          "train_loss": 8.663946209708229e-05,
          "validation_loss": 8.219786104746163e-05,
          "train_metric": 8.726009582460392e-06,
          "validation_metric": 8.278667337435763e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.009100401999603
        },
        {
          "epoch": 678,
          "updates": 10848,
          "samples_seen": 694272,
          "online_loss": 8.751497375669715e-05,
          "train_loss": 8.699158934177831e-05,
          "validation_loss": 8.094516670098528e-05,
          "train_metric": 8.761473509366624e-06,
          "validation_metric": 8.152501322911121e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.019490086999213
        },
        {
          "epoch": 679,
          "updates": 10864,
          "samples_seen": 695296,
          "online_loss": 9.36655360419536e-05,
          "train_loss": 8.703681669430807e-05,
          "validation_loss": 8.264713687822223e-05,
          "train_metric": 8.766030077822506e-06,
          "validation_metric": 8.32391651783837e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.029731516999163
        },
        {
          "epoch": 680,
          "updates": 10880,
          "samples_seen": 696320,
          "online_loss": 8.834942195790063e-05,
          "train_loss": 8.665031054988503e-05,
          "validation_loss": 8.271248225355521e-05,
          "train_metric": 8.727101885597222e-06,
          "validation_metric": 8.3304985309951e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.03999445899899
        },
        {
          "epoch": 681,
          "updates": 10896,
          "samples_seen": 697344,
          "online_loss": 8.758838430367177e-05,
          "train_loss": 8.821618393994868e-05,
          "validation_loss": 8.094339864328504e-05,
          "train_metric": 8.884810995368753e-06,
          "validation_metric": 8.152323061949573e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.050238025998624
        },
        {
          "epoch": 682,
          "updates": 10912,
          "samples_seen": 698368,
          "online_loss": 9.081973053071124e-05,
          "train_loss": 8.811817679088563e-05,
          "validation_loss": 8.044327842071652e-05,
          "train_metric": 8.874940249370411e-06,
          "validation_metric": 8.101952516881283e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.060581048999666
        },
        {
          "epoch": 683,
          "updates": 10928,
          "samples_seen": 699392,
          "online_loss": 8.937556481214415e-05,
          "train_loss": 8.772571163717657e-05,
          "validation_loss": 8.333413279615343e-05,
          "train_metric": 8.835410881147254e-06,
          "validation_metric": 8.393109055759851e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.070834690999618
        },
        {
          "epoch": 684,
          "updates": 10944,
          "samples_seen": 700416,
          "online_loss": 9.208374081026705e-05,
          "train_loss": 8.803656237432733e-05,
          "validation_loss": 8.46111070131883e-05,
          "train_metric": 8.866720236255787e-06,
          "validation_metric": 8.52172070153756e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.081472975998622
        },
        {
          "epoch": 685,
          "updates": 10960,
          "samples_seen": 701440,
          "online_loss": 9.423208803127636e-05,
          "train_loss": 9.419109846930951e-05,
          "validation_loss": 8.244193304562941e-05,
          "train_metric": 9.486580893280916e-06,
          "validation_metric": 8.303250979224686e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.091775428998517
        },
        {
          "epoch": 686,
          "updates": 10976,
          "samples_seen": 702464,
          "online_loss": 8.987600790533179e-05,
          "train_loss": 8.58567000250332e-05,
          "validation_loss": 8.036709186853841e-05,
          "train_metric": 8.647173672216013e-06,
          "validation_metric": 8.094280019577127e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.102113531000214
        },
        {
          "epoch": 687,
          "updates": 10992,
          "samples_seen": 703488,
          "online_loss": 8.719352103980782e-05,
          "train_loss": 8.62419547047466e-05,
          "validation_loss": 8.088283357210457e-05,
          "train_metric": 8.685972716193646e-06,
          "validation_metric": 8.14622217149008e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.112280638999437
        },
        {
          "epoch": 688,
          "updates": 11008,
          "samples_seen": 704512,
          "online_loss": 8.84263654370443e-05,
          "train_loss": 8.561631693737581e-05,
          "validation_loss": 8.044406422413886e-05,
          "train_metric": 8.622962013760116e-06,
          "validation_metric": 8.102031642920338e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.122602923000159
        },
        {
          "epoch": 689,
          "updates": 11024,
          "samples_seen": 705536,
          "online_loss": 8.644610534247477e-05,
          "train_loss": 8.607091876911e-05,
          "validation_loss": 7.918578921817243e-05,
          "train_metric": 8.668746886542067e-06,
          "validation_metric": 7.975301741680596e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.1336626049997
        },
        {
          "epoch": 690,
          "updates": 11040,
          "samples_seen": 706560,
          "online_loss": 8.76921508279338e-05,
          "train_loss": 8.556728425901383e-05,
          "validation_loss": 8.04664523457177e-05,
          "train_metric": 8.618023457529489e-06,
          "validation_metric": 8.104286280286033e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.144240756999352
        },
        {
          "epoch": 691,
          "updates": 11056,
          "samples_seen": 707584,
          "online_loss": 8.698016517882934e-05,
          "train_loss": 8.666757639730349e-05,
          "validation_loss": 8.089203765848652e-05,
          "train_metric": 8.728842658456415e-06,
          "validation_metric": 8.147149856085889e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.210300765998909
        },
        {
          "epoch": 692,
          "updates": 11072,
          "samples_seen": 708608,
          "online_loss": 9.08592005544051e-05,
          "train_loss": 9.258360660169274e-05,
          "validation_loss": 8.234934648498893e-05,
          "train_metric": 9.324681741418317e-06,
          "validation_metric": 8.293923201563302e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.221077898999283
        },
        {
          "epoch": 693,
          "updates": 11088,
          "samples_seen": 709632,
          "online_loss": 8.90883904958173e-05,
          "train_loss": 8.676374272909015e-05,
          "validation_loss": 8.27413605293259e-05,
          "train_metric": 8.738526048546191e-06,
          "validation_metric": 8.33340709505137e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.231422682998527
        },
        {
          "epoch": 694,
          "updates": 11104,
          "samples_seen": 710656,
          "online_loss": 8.838870098770713e-05,
          "train_loss": 9.264113759854808e-05,
          "validation_loss": 8.250163227785379e-05,
          "train_metric": 9.330476132163312e-06,
          "validation_metric": 8.309262739203405e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.24178579999898
        },
        {
          "epoch": 695,
          "updates": 11120,
          "samples_seen": 711680,
          "online_loss": 8.833533979668573e-05,
          "train_loss": 8.61264270497486e-05,
          "validation_loss": 8.057681407080963e-05,
          "train_metric": 8.674338459968567e-06,
          "validation_metric": 8.1154012150364e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.252051335999568
        },
        {
          "epoch": 696,
          "updates": 11136,
          "samples_seen": 712704,
          "online_loss": 8.935788150665758e-05,
          "train_loss": 8.510936459060758e-05,
          "validation_loss": 8.020449604373425e-05,
          "train_metric": 8.571902981202584e-06,
          "validation_metric": 8.077903657977004e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.262396138999975
        },
        {
          "epoch": 697,
          "updates": 11152,
          "samples_seen": 713728,
          "online_loss": 8.86871350758156e-05,
          "train_loss": 8.734288712730631e-05,
          "validation_loss": 7.978232315508649e-05,
          "train_metric": 8.796856491244398e-06,
          "validation_metric": 8.03538478066912e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.272615926998697
        },
        {
          "epoch": 698,
          "updates": 11168,
          "samples_seen": 714752,
          "online_loss": 9.298218765252386e-05,
          "train_loss": 9.769885946298018e-05,
          "validation_loss": 9.372573549626395e-05,
          "train_metric": 9.839871381700505e-06,
          "validation_metric": 9.439712812309153e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.283177579998664
        },
        {
          "epoch": 699,
          "updates": 11184,
          "samples_seen": 715776,
          "online_loss": 9.642378699936671e-05,
          "train_loss": 8.745129889575765e-05,
          "validation_loss": 7.997373177204281e-05,
          "train_metric": 8.80777406564448e-06,
          "validation_metric": 8.054662430367898e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.293477809998876
        },
        {
          "epoch": 700,
          "updates": 11200,
          "samples_seen": 716800,
          "online_loss": 8.836420715852e-05,
          "train_loss": 8.56385231600143e-05,
          "validation_loss": 7.990656740730628e-05,
          "train_metric": 8.625198461231776e-06,
          "validation_metric": 8.047895789786708e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.303833385998587
        },
        {
          "epoch": 701,
          "updates": 11216,
          "samples_seen": 717824,
          "online_loss": 8.622316886430781e-05,
          "train_loss": 9.219624189427122e-05,
          "validation_loss": 8.472516492474824e-05,
          "train_metric": 9.285666237701662e-06,
          "validation_metric": 8.533207619620953e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.314003610999862
        },
        {
          "epoch": 702,
          "updates": 11232,
          "samples_seen": 718848,
          "online_loss": 9.479006871515594e-05,
          "train_loss": 8.88867216417566e-05,
          "validation_loss": 8.56277474667877e-05,
          "train_metric": 8.9523446149542e-06,
          "validation_metric": 8.624112524557859e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.324368671999764
        },
        {
          "epoch": 703,
          "updates": 11248,
          "samples_seen": 719872,
          "online_loss": 8.912397584026621e-05,
          "train_loss": 8.496652299072593e-05,
          "validation_loss": 7.885231025284156e-05,
          "train_metric": 8.557517503504641e-06,
          "validation_metric": 7.941715011838824e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.335217077999914
        },
        {
          "epoch": 704,
          "updates": 11264,
          "samples_seen": 720896,
          "online_loss": 8.77536572261306e-05,
          "train_loss": 8.500080002704635e-05,
          "validation_loss": 8.013458136701956e-05,
          "train_metric": 8.560969945392571e-06,
          "validation_metric": 8.070862349995878e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.345663116999276
        },
        {
          "epoch": 705,
          "updates": 11280,
          "samples_seen": 721920,
          "online_loss": 8.72603256993898e-05,
          "train_loss": 8.559451816836372e-05,
          "validation_loss": 8.030673780012876e-05,
          "train_metric": 8.620766493550036e-06,
          "validation_metric": 8.088200047495775e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.356000952999239
        },
        {
          "epoch": 706,
          "updates": 11296,
          "samples_seen": 722944,
          "online_loss": 8.810056647234887e-05,
          "train_loss": 8.581259316997603e-05,
          "validation_loss": 7.901657954789698e-05,
          "train_metric": 8.64272988110315e-06,
          "validation_metric": 7.958260539453477e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.36650391799958
        },
        {
          "epoch": 707,
          "updates": 11312,
          "samples_seen": 723968,
          "online_loss": 8.615558249402966e-05,
          "train_loss": 9.085652709472924e-05,
          "validation_loss": 8.559168054489419e-05,
          "train_metric": 9.150737241725437e-06,
          "validation_metric": 8.620480002718978e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.376722484999846
        },
        {
          "epoch": 708,
          "updates": 11328,
          "samples_seen": 724992,
          "online_loss": 9.44194503063045e-05,
          "train_loss": 8.439344674116e-05,
          "validation_loss": 7.881336205173284e-05,
          "train_metric": 8.499798241246026e-06,
          "validation_metric": 7.937792361190077e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.387542795999252
        },
        {
          "epoch": 709,
          "updates": 11344,
          "samples_seen": 726016,
          "online_loss": 8.769319219936733e-05,
          "train_loss": 8.494898793287575e-05,
          "validation_loss": 7.843633648008108e-05,
          "train_metric": 8.555751264793798e-06,
          "validation_metric": 7.899820957391057e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.398491100999308
        },
        {
          "epoch": 710,
          "updates": 11360,
          "samples_seen": 727040,
          "online_loss": 8.634813525532081e-05,
          "train_loss": 8.421693928539753e-05,
          "validation_loss": 7.84630246926099e-05,
          "train_metric": 8.482022167299874e-06,
          "validation_metric": 7.902508514234796e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.409031604998745
        },
        {
          "epoch": 711,
          "updates": 11376,
          "samples_seen": 728064,
          "online_loss": 8.635423569103295e-05,
          "train_loss": 8.582773443777114e-05,
          "validation_loss": 7.852212729630992e-05,
          "train_metric": 8.644255103718024e-06,
          "validation_metric": 7.9084611570579e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.419320606999463
        },
        {
          "epoch": 712,
          "updates": 11392,
          "samples_seen": 729088,
          "online_loss": 8.619549134891713e-05,
          "train_loss": 9.258577483706176e-05,
          "validation_loss": 8.431600144831464e-05,
          "train_metric": 9.324900020146742e-06,
          "validation_metric": 8.491998414683621e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.42973529199844
        },
        {
          "epoch": 713,
          "updates": 11408,
          "samples_seen": 730112,
          "online_loss": 8.837397081151721e-05,
          "train_loss": 8.566399628762156e-05,
          "validation_loss": 7.795145938871428e-05,
          "train_metric": 8.627764145785477e-06,
          "validation_metric": 7.85098563937936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.440569735999816
        },
        {
          "epoch": 714,
          "updates": 11424,
          "samples_seen": 731136,
          "online_loss": 8.804403364592872e-05,
          "train_loss": 8.415862976107746e-05,
          "validation_loss": 7.826994260540232e-05,
          "train_metric": 8.476147741021123e-06,
          "validation_metric": 7.883060789026786e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.45103939999899
        },
        {
          "epoch": 715,
          "updates": 11440,
          "samples_seen": 732160,
          "online_loss": 8.717912555766816e-05,
          "train_loss": 8.544746378902346e-05,
          "validation_loss": 7.735868712188676e-05,
          "train_metric": 8.605955372331664e-06,
          "validation_metric": 7.791282769176178e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.461896343998887
        },
        {
          "epoch": 716,
          "updates": 11456,
          "samples_seen": 733184,
          "online_loss": 8.93954334060254e-05,
          "train_loss": 8.430668094661087e-05,
          "validation_loss": 7.925957470433787e-05,
          "train_metric": 8.491059816151392e-06,
          "validation_metric": 7.982735041878186e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.472342457000195
        },
        {
          "epoch": 717,
          "updates": 11472,
          "samples_seen": 734208,
          "online_loss": 8.819798949843971e-05,
          "train_loss": 8.409189467784017e-05,
          "validation_loss": 7.782658940413967e-05,
          "train_metric": 8.469427484669723e-06,
          "validation_metric": 7.838408237148542e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.482864898998741
        },
        {
          "epoch": 718,
          "updates": 11488,
          "samples_seen": 735232,
          "online_loss": 9.015744126372738e-05,
          "train_loss": 9.53337294049561e-05,
          "validation_loss": 8.30617209430784e-05,
          "train_metric": 9.601663805369753e-06,
          "validation_metric": 8.365672329091467e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.493252869999196
        },
        {
          "epoch": 719,
          "updates": 11504,
          "samples_seen": 736256,
          "online_loss": 8.998456678455113e-05,
          "train_loss": 8.419642108492553e-05,
          "validation_loss": 8.006046118680388e-05,
          "train_metric": 8.479954885842744e-06,
          "validation_metric": 8.063397217483725e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.503524748999553
        },
        {
          "epoch": 720,
          "updates": 11520,
          "samples_seen": 737280,
          "online_loss": 8.899887325242162e-05,
          "train_loss": 8.460576646029949e-05,
          "validation_loss": 7.766841008560732e-05,
          "train_metric": 8.521182280674111e-06,
          "validation_metric": 7.822478437446989e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.513860203998775
        },
        {
          "epoch": 721,
          "updates": 11536,
          "samples_seen": 738304,
          "online_loss": 8.764768926994293e-05,
          "train_loss": 8.506630547344685e-05,
          "validation_loss": 7.904900121502578e-05,
          "train_metric": 8.567567419959232e-06,
          "validation_metric": 7.961525625432841e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.524155731000064
        },
        {
          "epoch": 722,
          "updates": 11552,
          "samples_seen": 739328,
          "online_loss": 8.58329328821128e-05,
          "train_loss": 8.634765981696546e-05,
          "validation_loss": 7.879541954025626e-05,
          "train_metric": 8.696620170667302e-06,
          "validation_metric": 7.935986104712356e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.53440441499879
        },
        {
          "epoch": 723,
          "updates": 11568,
          "samples_seen": 740352,
          "online_loss": 8.519807579432381e-05,
          "train_loss": 8.363337838090956e-05,
          "validation_loss": 7.765523332636803e-05,
          "train_metric": 8.4232469816925e-06,
          "validation_metric": 7.8211505751824e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.544686612000078
        },
        {
          "epoch": 724,
          "updates": 11584,
          "samples_seen": 741376,
          "online_loss": 8.724002577764622e-05,
          "train_loss": 8.774110028753057e-05,
          "validation_loss": 8.318317850353196e-05,
          "train_metric": 8.836962479108479e-06,
          "validation_metric": 8.377905032830313e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.555137960000138
        },
        {
          "epoch": 725,
          "updates": 11600,
          "samples_seen": 742400,
          "online_loss": 8.803850414551562e-05,
          "train_loss": 0.00010771026427391917,
          "validation_loss": 9.1374728071969e-05,
          "train_metric": 1.0848183592315763e-05,
          "validation_metric": 9.202927685691975e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.565559659999053
        },
        {
          "epoch": 726,
          "updates": 11616,
          "samples_seen": 743424,
          "online_loss": 9.375575905323785e-05,
          "train_loss": 8.350261487066746e-05,
          "validation_loss": 7.718340202700347e-05,
          "train_metric": 8.410076588916127e-06,
          "validation_metric": 7.773629477014765e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.576477356999021
        },
        {
          "epoch": 727,
          "updates": 11632,
          "samples_seen": 744448,
          "online_loss": 8.736398376640864e-05,
          "train_loss": 9.106638026423752e-05,
          "validation_loss": 8.472947956761345e-05,
          "train_metric": 9.171872989099938e-06,
          "validation_metric": 8.533643267583102e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.643175669998527
        },
        {
          "epoch": 728,
          "updates": 11648,
          "samples_seen": 745472,
          "online_loss": 8.74748109254142e-05,
          "train_loss": 8.546247408958152e-05,
          "validation_loss": 7.82194038038142e-05,
          "train_metric": 8.60746695252601e-06,
          "validation_metric": 7.877970347180963e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.654161333999582
        },
        {
          "epoch": 729,
          "updates": 11664,
          "samples_seen": 746496,
          "online_loss": 8.574129378757789e-05,
          "train_loss": 9.071378008229658e-05,
          "validation_loss": 8.575636456953362e-05,
          "train_metric": 9.136359949479811e-06,
          "validation_metric": 8.637067367089912e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.664479530998506
        },
        {
          "epoch": 730,
          "updates": 11680,
          "samples_seen": 747520,
          "online_loss": 8.788347486188286e-05,
          "train_loss": 9.297142969444394e-05,
          "validation_loss": 7.962052768561989e-05,
          "train_metric": 9.36374271987006e-06,
          "validation_metric": 8.019089364097454e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.674812609999208
        },
        {
          "epoch": 731,
          "updates": 11696,
          "samples_seen": 748544,
          "online_loss": 8.681368331053818e-05,
          "train_loss": 8.338096085935831e-05,
          "validation_loss": 7.893191650509834e-05,
          "train_metric": 8.397824785788544e-06,
          "validation_metric": 7.949733117129654e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.685107338998932
        },
        {
          "epoch": 732,
          "updates": 11712,
          "samples_seen": 749568,
          "online_loss": 8.768851034801628e-05,
          "train_loss": 8.353692828677595e-05,
          "validation_loss": 7.807341171428561e-05,
          "train_metric": 8.413531759288162e-06,
          "validation_metric": 7.863268365326803e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.695423436000056
        },
        {
          "epoch": 733,
          "updates": 11728,
          "samples_seen": 750592,
          "online_loss": 8.901797559701663e-05,
          "train_loss": 8.357372280443087e-05,
          "validation_loss": 7.886237290222198e-05,
          "train_metric": 8.41723976918729e-06,
          "validation_metric": 7.9427290984313e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.705723177999971
        },
        {
          "epoch": 734,
          "updates": 11744,
          "samples_seen": 751616,
          "online_loss": 8.654356315673795e-05,
          "train_loss": 8.660808089189231e-05,
          "validation_loss": 7.978660869412124e-05,
          "train_metric": 8.722849088371731e-06,
          "validation_metric": 8.035814971663058e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.716022124999654
        },
        {
          "epoch": 735,
          "updates": 11760,
          "samples_seen": 752640,
          "online_loss": 8.821515211820952e-05,
          "train_loss": 8.260380855062976e-05,
          "validation_loss": 7.695312524447218e-05,
          "train_metric": 8.319552762259264e-06,
          "validation_metric": 7.750437362119555e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.726881980999678
        },
        {
          "epoch": 736,
          "updates": 11776,
          "samples_seen": 753664,
          "online_loss": 8.545319928998651e-05,
          "train_loss": 8.264870848506689e-05,
          "validation_loss": 7.661136623937637e-05,
          "train_metric": 8.32407567941118e-06,
          "validation_metric": 7.716016625636257e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.737865223998597
        },
        {
          "epoch": 737,
          "updates": 11792,
          "samples_seen": 754688,
          "online_loss": 8.849779351294274e-05,
          "train_loss": 8.490374602843076e-05,
          "validation_loss": 7.745091716060415e-05,
          "train_metric": 8.551192877348512e-06,
          "validation_metric": 7.800572348060086e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.748175824999635
        },
        {
          "epoch": 738,
          "updates": 11808,
          "samples_seen": 755712,
          "online_loss": 8.59176434460096e-05,
          "train_loss": 8.295547740999609e-05,
          "validation_loss": 7.799749437253922e-05,
          "train_metric": 8.354971214430407e-06,
          "validation_metric": 7.855622243368998e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.758785172998614
        },
        {
          "epoch": 739,
          "updates": 11824,
          "samples_seen": 756736,
          "online_loss": 8.535009419574635e-05,
          "train_loss": 8.260030881501734e-05,
          "validation_loss": 7.663761789444834e-05,
          "train_metric": 8.319200787809677e-06,
          "validation_metric": 7.718660526734311e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.76905494699895
        },
        {
          "epoch": 740,
          "updates": 11840,
          "samples_seen": 757760,
          "online_loss": 8.86927600731724e-05,
          "train_loss": 8.757867180975154e-05,
          "validation_loss": 7.890746201155707e-05,
          "train_metric": 8.82060339790769e-06,
          "validation_metric": 7.947271114971954e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.779291233999174
        },
        {
          "epoch": 741,
          "updates": 11856,
          "samples_seen": 758784,
          "online_loss": 8.306190966322902e-05,
          "train_loss": 8.424230327364057e-05,
          "validation_loss": 7.792641554260626e-05,
          "train_metric": 8.484576028422453e-06,
          "validation_metric": 7.848463610571343e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.789555341998494
        },
        {
          "epoch": 742,
          "updates": 11872,
          "samples_seen": 759808,
          "online_loss": 8.407831296608492e-05,
          "train_loss": 8.251026156358421e-05,
          "validation_loss": 7.765749614918604e-05,
          "train_metric": 8.3101322161383e-06,
          "validation_metric": 7.821378858352546e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.799937456999032
        },
        {
          "epoch": 743,
          "updates": 11888,
          "samples_seen": 760832,
          "online_loss": 8.534931430403958e-05,
          "train_loss": 8.42254376038909e-05,
          "validation_loss": 7.912563160061836e-05,
          "train_metric": 8.482878001814242e-06,
          "validation_metric": 7.969243597472087e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.81024622699988
        },
        {
          "epoch": 744,
          "updates": 11904,
          "samples_seen": 761856,
          "online_loss": 8.478284871671349e-05,
          "train_loss": 8.311851706821471e-05,
          "validation_loss": 7.783448381815106e-05,
          "train_metric": 8.371392141270917e-06,
          "validation_metric": 7.839204045012593e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.82066308299909
        },
        {
          "epoch": 745,
          "updates": 11920,
          "samples_seen": 762880,
          "online_loss": 8.624346628494095e-05,
          "train_loss": 8.278652967419475e-05,
          "validation_loss": 7.77232417021878e-05,
          "train_metric": 8.337956387549639e-06,
          "validation_metric": 7.827999979781453e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.830968481999662
        },
        {
          "epoch": 746,
          "updates": 11936,
          "samples_seen": 763904,
          "online_loss": 8.783756493357942e-05,
          "train_loss": 9.00832746992819e-05,
          "validation_loss": 7.883559010224417e-05,
          "train_metric": 9.072858119907323e-06,
          "validation_metric": 7.940032446640544e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.841543370999716
        },
        {
          "epoch": 747,
          "updates": 11952,
          "samples_seen": 764928,
          "online_loss": 8.46287807689805e-05,
          "train_loss": 8.22760266601108e-05,
          "validation_loss": 7.654615910723805e-05,
          "train_metric": 8.286539014079608e-06,
          "validation_metric": 7.709449164394755e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.852439573998709
        },
        {
          "epoch": 748,
          "updates": 11968,
          "samples_seen": 765952,
          "online_loss": 8.432435743088718e-05,
          "train_loss": 8.280704059870914e-05,
          "validation_loss": 7.7019416494295e-05,
          "train_metric": 8.340021850017365e-06,
          "validation_metric": 7.757113053230569e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.8629681049988
        },
        {
          "epoch": 749,
          "updates": 11984,
          "samples_seen": 766976,
          "online_loss": 8.40856000650092e-05,
          "train_loss": 9.295046038459986e-05,
          "validation_loss": 7.981131784617901e-05,
          "train_metric": 9.36162905418314e-06,
          "validation_metric": 8.03830425866181e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.87330989899965
        },
        {
          "epoch": 750,
          "updates": 12000,
          "samples_seen": 768000,
          "online_loss": 9.178116602015507e-05,
          "train_loss": 8.297373278765008e-05,
          "validation_loss": 7.6922049629502e-05,
          "train_metric": 8.356811122212093e-06,
          "validation_metric": 7.747305971861351e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.883766414999627
        },
        {
          "epoch": 751,
          "updates": 12016,
          "samples_seen": 769024,
          "online_loss": 8.433233324467437e-05,
          "train_loss": 8.402848470723256e-05,
          "validation_loss": 7.857958553358912e-05,
          "train_metric": 8.463041922368575e-06,
          "validation_metric": 7.914248271845281e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.894211130998883
        },
        {
          "epoch": 752,
          "updates": 12032,
          "samples_seen": 770048,
          "online_loss": 8.240869101427961e-05,
          "train_loss": 8.26828763820231e-05,
          "validation_loss": 7.655887748114765e-05,
          "train_metric": 8.327517207362689e-06,
          "validation_metric": 7.710729732934851e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.90462218399989
        },
        {
          "epoch": 753,
          "updates": 12048,
          "samples_seen": 771072,
          "online_loss": 8.711444161235704e-05,
          "train_loss": 8.377858466701582e-05,
          "validation_loss": 7.70035112509504e-05,
          "train_metric": 8.437872565991711e-06,
          "validation_metric": 7.755512342555448e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.914933062998898
        },
        {
          "epoch": 754,
          "updates": 12064,
          "samples_seen": 772096,
          "online_loss": 9.468275948165683e-05,
          "train_loss": 8.692764822626486e-05,
          "validation_loss": 8.110557973850518e-05,
          "train_metric": 8.755034286878072e-06,
          "validation_metric": 8.168656677298713e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.925612514998647
        },
        {
          "epoch": 755,
          "updates": 12080,
          "samples_seen": 773120,
          "online_loss": 8.472583817820123e-05,
          "train_loss": 8.183417958207428e-05,
          "validation_loss": 7.586697029182687e-05,
          "train_metric": 8.242039257311262e-06,
          "validation_metric": 7.641043339390308e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.936626859998796
        },
        {
          "epoch": 756,
          "updates": 12096,
          "samples_seen": 774144,
          "online_loss": 8.453131249552825e-05,
          "train_loss": 8.194144902518019e-05,
          "validation_loss": 7.609745080117136e-05,
          "train_metric": 8.252843144873623e-06,
          "validation_metric": 7.66425728215836e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.947325035998801
        },
        {
          "epoch": 757,
          "updates": 12112,
          "samples_seen": 775168,
          "online_loss": 8.846113473737205e-05,
          "train_loss": 9.00153027032502e-05,
          "validation_loss": 8.339566556969658e-05,
          "train_metric": 9.066011443792377e-06,
          "validation_metric": 8.39930544316303e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.95800891899853
        },
        {
          "epoch": 758,
          "updates": 12128,
          "samples_seen": 776192,
          "online_loss": 8.658233105052204e-05,
          "train_loss": 8.194333349820226e-05,
          "validation_loss": 7.566095155198127e-05,
          "train_metric": 8.25303141027689e-06,
          "validation_metric": 7.620293672516709e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.969210223000118
        },
        {
          "epoch": 759,
          "updates": 12144,
          "samples_seen": 777216,
          "online_loss": 8.347729203705967e-05,
          "train_loss": 8.195987902581692e-05,
          "validation_loss": 7.558630022685975e-05,
          "train_metric": 8.254699423559941e-06,
          "validation_metric": 7.612775789311854e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.980143307999242
        },
        {
          "epoch": 760,
          "updates": 12160,
          "samples_seen": 778240,
          "online_loss": 8.350419284397503e-05,
          "train_loss": 8.116396202240139e-05,
          "validation_loss": 7.530360016971827e-05,
          "train_metric": 8.174536560545675e-06,
          "validation_metric": 7.584302238683449e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.991510836998714
        },
        {
          "epoch": 761,
          "updates": 12176,
          "samples_seen": 779264,
          "online_loss": 8.342398291460995e-05,
          "train_loss": 8.173155219992623e-05,
          "validation_loss": 7.666205056011677e-05,
          "train_metric": 8.231702850025613e-06,
          "validation_metric": 7.721121619397309e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.001943825998751
        },
        {
          "epoch": 762,
          "updates": 12192,
          "samples_seen": 780288,
          "online_loss": 8.376097980544728e-05,
          "train_loss": 8.511254418408498e-05,
          "validation_loss": 7.815275603206828e-05,
          "train_metric": 8.57222403283231e-06,
          "validation_metric": 7.87125918577658e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.012581977000082
        },
        {
          "epoch": 763,
          "updates": 12208,
          "samples_seen": 781312,
          "online_loss": 8.529042042937363e-05,
          "train_loss": 8.162672020262107e-05,
          "validation_loss": 7.529263530159369e-05,
          "train_metric": 8.221142707043327e-06,
          "validation_metric": 7.583198566862848e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.081497093999133
        },
        {
          "epoch": 764,
          "updates": 12224,
          "samples_seen": 782336,
          "online_loss": 8.363986671611201e-05,
          "train_loss": 8.183652244042605e-05,
          "validation_loss": 7.50030085328035e-05,
          "train_metric": 8.242274816439021e-06,
          "validation_metric": 7.5540283432928845e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.092977403999612
        },
        {
          "epoch": 765,
          "updates": 12240,
          "samples_seen": 783360,
          "online_loss": 8.305077267323213e-05,
          "train_loss": 8.095733937807381e-05,
          "validation_loss": 7.419029861921445e-05,
          "train_metric": 8.153727321769111e-06,
          "validation_metric": 7.4721756391227245e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.104242089999389
        },
        {
          "epoch": 766,
          "updates": 12256,
          "samples_seen": 784384,
          "online_loss": 8.389023378185811e-05,
          "train_loss": 8.113564399536699e-05,
          "validation_loss": 7.528159767389297e-05,
          "train_metric": 8.171685294655617e-06,
          "validation_metric": 7.582087164337281e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.114962915999058
        },
        {
          "epoch": 767,
          "updates": 12272,
          "samples_seen": 785408,
          "online_loss": 8.300461740873288e-05,
          "train_loss": 8.306486415676773e-05,
          "validation_loss": 7.625277794431895e-05,
          "train_metric": 8.365989742742386e-06,
          "validation_metric": 7.679900591028854e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.125563384999623
        },
        {
          "epoch": 768,
          "updates": 12288,
          "samples_seen": 786432,
          "online_loss": 8.284838213512558e-05,
          "train_loss": 8.85347108123824e-05,
          "validation_loss": 7.994394400157034e-05,
          "train_metric": 8.916892511479091e-06,
          "validation_metric": 8.051661097852048e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.135940753998511
        },
        {
          "epoch": 769,
          "updates": 12304,
          "samples_seen": 787456,
          "online_loss": 8.735825076655601e-05,
          "train_loss": 8.232748950831592e-05,
          "validation_loss": 7.459563494194299e-05,
          "train_metric": 8.291722224385012e-06,
          "validation_metric": 7.512999218306504e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.14639251799963
        },
        {
          "epoch": 770,
          "updates": 12320,
          "samples_seen": 788480,
          "online_loss": 9.150284654424468e-05,
          "train_loss": 0.00010168759763473645,
          "validation_loss": 8.933453500503674e-05,
          "train_metric": 1.024160155793652e-05,
          "validation_metric": 8.997447366709821e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.156966900998668
        },
        {
          "epoch": 771,
          "updates": 12336,
          "samples_seen": 789504,
          "online_loss": 9.28307092635805e-05,
          "train_loss": 8.146479376591742e-05,
          "validation_loss": 7.449717668350786e-05,
          "train_metric": 8.204835467040539e-06,
          "validation_metric": 7.503083452320425e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.167464586998904
        },
        {
          "epoch": 772,
          "updates": 12352,
          "samples_seen": 790528,
          "online_loss": 8.850588233144663e-05,
          "train_loss": 9.462736488785595e-05,
          "validation_loss": 8.710975816939026e-05,
          "train_metric": 9.530521310807671e-06,
          "validation_metric": 8.773375157034025e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.177922558999853
        },
        {
          "epoch": 773,
          "updates": 12368,
          "samples_seen": 791552,
          "online_loss": 9.156271448773623e-05,
          "train_loss": 8.530647028237581e-05,
          "validation_loss": 7.789024675730616e-05,
          "train_metric": 8.591755431552883e-06,
          "validation_metric": 7.844820174796041e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.188303939999969
        },
        {
          "epoch": 774,
          "updates": 12384,
          "samples_seen": 792576,
          "online_loss": 8.370040518457245e-05,
          "train_loss": 8.450982568319887e-05,
          "validation_loss": 7.88189863669686e-05,
          "train_metric": 8.511520718457177e-06,
          "validation_metric": 7.938359885883983e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.198997691999466
        },
        {
          "epoch": 775,
          "updates": 12400,
          "samples_seen": 793600,
          "online_loss": 8.260340609922423e-05,
          "train_loss": 8.270091348094866e-05,
          "validation_loss": 7.557665958302096e-05,
          "train_metric": 8.32933346828213e-06,
          "validation_metric": 7.611804448970361e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.209350264000022
        },
        {
          "epoch": 776,
          "updates": 12416,
          "samples_seen": 794624,
          "online_loss": 8.510504267178476e-05,
          "train_loss": 8.030535536818206e-05,
          "validation_loss": 7.505732355639338e-05,
          "train_metric": 8.088061804301105e-06,
          "validation_metric": 7.559498499176698e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.219679174999328
        },
        {
          "epoch": 777,
          "updates": 12432,
          "samples_seen": 795648,
          "online_loss": 8.20869529434276e-05,
          "train_loss": 8.052945486269891e-05,
          "validation_loss": 7.359348819591105e-05,
          "train_metric": 8.110631824820302e-06,
          "validation_metric": 7.4120680437772535e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.230591136998555
        },
        {
          "epoch": 778,
          "updates": 12448,
          "samples_seen": 796672,
          "online_loss": 8.462144137411087e-05,
          "train_loss": 8.136213000398129e-05,
          "validation_loss": 7.551572343800217e-05,
          "train_metric": 8.194495421776082e-06,
          "validation_metric": 7.605666723975446e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.241069835999951
        },
        {
          "epoch": 779,
          "updates": 12464,
          "samples_seen": 797696,
          "online_loss": 9.021118444252352e-05,
          "train_loss": 9.710730228107423e-05,
          "validation_loss": 8.457092917524278e-05,
          "train_metric": 9.780291293282062e-06,
          "validation_metric": 8.517674359609373e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.251401520999934
        },
        {
          "epoch": 780,
          "updates": 12480,
          "samples_seen": 798720,
          "online_loss": 8.63282912177965e-05,
          "train_loss": 8.158439595717937e-05,
          "validation_loss": 7.332907262025401e-05,
          "train_metric": 8.216882633860223e-06,
          "validation_metric": 7.385434855677886e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.262334126999122
        },
        {
          "epoch": 781,
          "updates": 12496,
          "samples_seen": 799744,
          "online_loss": 8.34313962059241e-05,
          "train_loss": 8.409674774156883e-05,
          "validation_loss": 7.643400022061542e-05,
          "train_metric": 8.469916792819276e-06,
          "validation_metric": 7.698151421209332e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.27273377399979
        },
        {
          "epoch": 782,
          "updates": 12512,
          "samples_seen": 800768,
          "online_loss": 8.255995408035233e-05,
          "train_loss": 8.365994290215895e-05,
          "validation_loss": 7.63007192290388e-05,
          "train_metric": 8.425922715105116e-06,
          "validation_metric": 7.684729098400567e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.283121441998446
        },
        {
          "epoch": 783,
          "updates": 12528,
          "samples_seen": 801792,
          "online_loss": 8.795135772743379e-05,
          "train_loss": 8.146707114065066e-05,
          "validation_loss": 7.634297799086198e-05,
          "train_metric": 8.205063750210684e-06,
          "validation_metric": 7.688984624110162e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.293366152998715
        },
        {
          "epoch": 784,
          "updates": 12544,
          "samples_seen": 802816,
          "online_loss": 8.270779221675184e-05,
          "train_loss": 0.00010120678780367598,
          "validation_loss": 8.788151171756908e-05,
          "train_metric": 1.0193176422035322e-05,
          "validation_metric": 8.851102393236943e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.303684470000007
        },
        {
          "epoch": 785,
          "updates": 12560,
          "samples_seen": 803840,
          "online_loss": 8.917780951378518e-05,
          "train_loss": 8.078332029981539e-05,
          "validation_loss": 7.563197868876159e-05,
          "train_metric": 8.136200449371245e-06,
          "validation_metric": 7.617376013513422e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.313901506999173
        },
        {
          "epoch": 786,
          "updates": 12576,
          "samples_seen": 804864,
          "online_loss": 8.23065224722086e-05,
          "train_loss": 7.972127787070349e-05,
          "validation_loss": 7.361244934145361e-05,
          "train_metric": 8.029235686990432e-06,
          "validation_metric": 7.413977527903626e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.324239260999093
        },
        {
          "epoch": 787,
          "updates": 12592,
          "samples_seen": 805888,
          "online_loss": 8.194752444978803e-05,
          "train_loss": 8.208037615986541e-05,
          "validation_loss": 7.35524226911366e-05,
          "train_metric": 8.266833901870996e-06,
          "validation_metric": 7.4079307523788884e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.334450755999569
        },
        {
          "epoch": 788,
          "updates": 12608,
          "samples_seen": 806912,
          "online_loss": 8.890657090887544e-05,
          "train_loss": 9.499090811004862e-05,
          "validation_loss": 8.297649765154347e-05,
          "train_metric": 9.56713756750105e-06,
          "validation_metric": 8.357087608601432e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.344860518000132
        },
        {
          "epoch": 789,
          "updates": 12624,
          "samples_seen": 807936,
          "online_loss": 0.00010118075874743226,
          "train_loss": 8.042350236792117e-05,
          "validation_loss": 7.397303852485493e-05,
          "train_metric": 8.099959813989699e-06,
          "validation_metric": 7.450293196598068e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.35522157000014
        },
        {
          "epoch": 790,
          "updates": 12640,
          "samples_seen": 808960,
          "online_loss": 0.00010096684422933322,
          "train_loss": 0.00011527358583407477,
          "validation_loss": 0.00010197285155300051,
          "train_metric": 1.1609932698775083e-05,
          "validation_metric": 1.0270332495565526e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.365643061999435
        },
        {
          "epoch": 791,
          "updates": 12656,
          "samples_seen": 809984,
          "online_loss": 9.040145982908143e-05,
          "train_loss": 8.419762889388949e-05,
          "validation_loss": 7.389661914203316e-05,
          "train_metric": 8.480077667627484e-06,
          "validation_metric": 7.442596142936964e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.375916764000067
        },
        {
          "epoch": 792,
          "updates": 12672,
          "samples_seen": 811008,
          "online_loss": 8.33344247439527e-05,
          "train_loss": 8.261035691248253e-05,
          "validation_loss": 7.569915032945573e-05,
          "train_metric": 8.32021305541275e-06,
          "validation_metric": 7.6241412898525596e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.386213208999834
        },
        {
          "epoch": 793,
          "updates": 12688,
          "samples_seen": 812032,
          "online_loss": 8.238267514570907e-05,
          "train_loss": 7.975276093930006e-05,
          "validation_loss": 7.413705316139385e-05,
          "train_metric": 8.032405276026111e-06,
          "validation_metric": 7.466813258361071e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.396396946000095
        },
        {
          "epoch": 794,
          "updates": 12704,
          "samples_seen": 813056,
          "online_loss": 8.32378207178408e-05,
          "train_loss": 8.493725181324407e-05,
          "validation_loss": 7.807785732438788e-05,
          "train_metric": 8.554568921681494e-06,
          "validation_metric": 7.863715836720075e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.406709544999103
        },
        {
          "epoch": 795,
          "updates": 12720,
          "samples_seen": 814080,
          "online_loss": 8.169417219505704e-05,
          "train_loss": 8.24305388960056e-05,
          "validation_loss": 7.374987762887031e-05,
          "train_metric": 8.302101377921645e-06,
          "validation_metric": 7.427817308780504e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.416909386000043
        },
        {
          "epoch": 796,
          "updates": 12736,
          "samples_seen": 815104,
          "online_loss": 8.319688890878751e-05,
          "train_loss": 7.944733079057187e-05,
          "validation_loss": 7.354881381615996e-05,
          "train_metric": 8.001643436728045e-06,
          "validation_metric": 7.407568318740232e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.427296926998679
        },
        {
          "epoch": 797,
          "updates": 12752,
          "samples_seen": 816128,
          "online_loss": 8.169091074705648e-05,
          "train_loss": 8.148322376655415e-05,
          "validation_loss": 7.453966100001708e-05,
          "train_metric": 8.206692655221559e-06,
          "validation_metric": 7.507362170144916e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.43748769099875
        },
        {
          "epoch": 798,
          "updates": 12768,
          "samples_seen": 817152,
          "online_loss": 8.388461105823808e-05,
          "train_loss": 8.506747690262273e-05,
          "validation_loss": 7.830331014702097e-05,
          "train_metric": 8.567684744775761e-06,
          "validation_metric": 7.88642319093924e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.447799165998731
        },
        {
          "epoch": 799,
          "updates": 12784,
          "samples_seen": 818176,
          "online_loss": 8.560820060665719e-05,
          "train_loss": 7.960816583363339e-05,
          "validation_loss": 7.289007771760225e-05,
          "train_metric": 8.017842446861323e-06,
          "validation_metric": 7.3412220444879495e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.512638067999433
        },
        {
          "epoch": 800,
          "updates": 12800,
          "samples_seen": 819200,
          "online_loss": 8.965862298282445e-05,
          "train_loss": 0.0001006269667414017,
          "validation_loss": 8.766332757659256e-05,
          "train_metric": 1.0134779586223885e-05,
          "validation_metric": 8.829128091747407e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.52339438999843
        },
        {
          "epoch": 801,
          "updates": 12816,
          "samples_seen": 820224,
          "online_loss": 8.826308635434543e-05,
          "train_loss": 7.905914389993995e-05,
          "validation_loss": 7.307754276553169e-05,
          "train_metric": 7.962547897477634e-06,
          "validation_metric": 7.360103609244106e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.53377390300011
        },
        {
          "epoch": 802,
          "updates": 12832,
          "samples_seen": 821248,
          "online_loss": 8.388051219299086e-05,
          "train_loss": 8.83464963408187e-05,
          "validation_loss": 7.6731012086384e-05,
          "train_metric": 8.897935913410038e-06,
          "validation_metric": 7.728066520940047e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.544135434000054
        },
        {
          "epoch": 803,
          "updates": 12848,
          "samples_seen": 822272,
          "online_loss": 8.373195169042447e-05,
          "train_loss": 7.883912621764466e-05,
          "validation_loss": 7.286524487426504e-05,
          "train_metric": 7.940388059068937e-06,
          "validation_metric": 7.338720479310723e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.554982708999887
        },
        {
          "epoch": 804,
          "updates": 12864,
          "samples_seen": 823296,
          "online_loss": 8.495100109939813e-05,
          "train_loss": 7.984909461811185e-05,
          "validation_loss": 7.360018935287371e-05,
          "train_metric": 8.042108674999326e-06,
          "validation_metric": 7.412742434098618e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.56558574599876
        },
        {
          "epoch": 805,
          "updates": 12880,
          "samples_seen": 824320,
          "online_loss": 8.320843153342139e-05,
          "train_loss": 7.912721775937825e-05,
          "validation_loss": 7.29600724298507e-05,
          "train_metric": 7.969402759044897e-06,
          "validation_metric": 7.348271992668742e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.57583348299886
        },
        {
          "epoch": 806,
          "updates": 12896,
          "samples_seen": 825344,
          "online_loss": 8.065042538873968e-05,
          "train_loss": 7.939356146380305e-05,
          "validation_loss": 7.341320451814681e-05,
          "train_metric": 7.996229214768391e-06,
          "validation_metric": 7.393908163066953e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.586315845999707
        },
        {
          "epoch": 807,
          "updates": 12912,
          "samples_seen": 826368,
          "online_loss": 8.423668145951524e-05,
          "train_loss": 9.09053924260661e-05,
          "validation_loss": 7.571271999040619e-05,
          "train_metric": 9.15565851755673e-06,
          "validation_metric": 7.625508260389324e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.596596309000233
        },
        {
          "epoch": 808,
          "updates": 12928,
          "samples_seen": 827392,
          "online_loss": 8.589278695581015e-05,
          "train_loss": 7.987357093952596e-05,
          "validation_loss": 7.540093793068081e-05,
          "train_metric": 8.044574315135833e-06,
          "validation_metric": 7.59410613682121e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.606877311998687
        },
        {
          "epoch": 809,
          "updates": 12944,
          "samples_seen": 828416,
          "online_loss": 8.897366842575138e-05,
          "train_loss": 9.456888074055314e-05,
          "validation_loss": 8.535937377018854e-05,
          "train_metric": 9.524632332613692e-06,
          "validation_metric": 8.597083251515869e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.617109830000118
        },
        {
          "epoch": 810,
          "updates": 12960,
          "samples_seen": 829440,
          "online_loss": 9.193022560793906e-05,
          "train_loss": 8.391131268581375e-05,
          "validation_loss": 7.968212594278157e-05,
          "train_metric": 8.45124031911837e-06,
          "validation_metric": 8.025292117963545e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.627463388998876
        },
        {
          "epoch": 811,
          "updates": 12976,
          "samples_seen": 830464,
          "online_loss": 9.270822556572966e-05,
          "train_loss": 7.886798994150013e-05,
          "validation_loss": 7.291545625776052e-05,
          "train_metric": 7.943294804135803e-06,
          "validation_metric": 7.343777724599931e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.637719103999189
        },
        {
          "epoch": 812,
          "updates": 12992,
          "samples_seen": 831488,
          "online_loss": 8.104466269287514e-05,
          "train_loss": 8.028298907447606e-05,
          "validation_loss": 7.29495077393949e-05,
          "train_metric": 8.085808985924814e-06,
          "validation_metric": 7.347207429120317e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.648006084998997
        },
        {
          "epoch": 813,
          "updates": 13008,
          "samples_seen": 832512,
          "online_loss": 9.039720316650346e-05,
          "train_loss": 8.087958849500865e-05,
          "validation_loss": 7.505932444473729e-05,
          "train_metric": 8.145896572386846e-06,
          "validation_metric": 7.559700861747842e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.658234957998502
        },
        {
          "epoch": 814,
          "updates": 13024,
          "samples_seen": 833536,
          "online_loss": 8.519106995663606e-05,
          "train_loss": 8.199753938242793e-05,
          "validation_loss": 7.546448614448309e-05,
          "train_metric": 8.258491106971633e-06,
          "validation_metric": 7.600506705784937e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.668497347000084
        },
        {
          "epoch": 815,
          "updates": 13040,
          "samples_seen": 834560,
          "online_loss": 8.818143396638334e-05,
          "train_loss": 9.145072544924915e-05,
          "validation_loss": 7.605653809150681e-05,
          "train_metric": 9.210581993102096e-06,
          "validation_metric": 7.660136361664627e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.678732579999632
        },
        {
          "epoch": 816,
          "updates": 13056,
          "samples_seen": 835584,
          "online_loss": 8.514035062034964e-05,
          "train_loss": 7.859821926103905e-05,
          "validation_loss": 7.423230272252113e-05,
          "train_metric": 7.916124559415039e-06,
          "validation_metric": 7.476406153728021e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.689227220998873
        },
        {
          "epoch": 817,
          "updates": 13072,
          "samples_seen": 836608,
          "online_loss": 8.300381773551635e-05,
          "train_loss": 8.097336831269786e-05,
          "validation_loss": 7.5287971412763e-05,
          "train_metric": 8.155340765370056e-06,
          "validation_metric": 7.582728358102031e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.699471657999311
        },
        {
          "epoch": 818,
          "updates": 13088,
          "samples_seen": 837632,
          "online_loss": 8.026441742003954e-05,
          "train_loss": 8.147361950250342e-05,
          "validation_loss": 7.413447019644082e-05,
          "train_metric": 8.205724952858873e-06,
          "validation_metric": 7.466551323886961e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.709925182998631
        },
        {
          "epoch": 819,
          "updates": 13104,
          "samples_seen": 838656,
          "online_loss": 8.475942922814284e-05,
          "train_loss": 8.004758274182677e-05,
          "validation_loss": 7.343559991568327e-05,
          "train_metric": 8.062098459049594e-06,
          "validation_metric": 7.396165528916754e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.720131673999276
        },
        {
          "epoch": 820,
          "updates": 13120,
          "samples_seen": 839680,
          "online_loss": 8.548808386876772e-05,
          "train_loss": 7.904895028332248e-05,
          "validation_loss": 7.269700290635228e-05,
          "train_metric": 7.961521077959333e-06,
          "validation_metric": 7.32177477402729e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.731002380998689
        },
        {
          "epoch": 821,
          "updates": 13136,
          "samples_seen": 840704,
          "online_loss": 8.389098638872383e-05,
          "train_loss": 8.192426321329549e-05,
          "validation_loss": 7.520155486417934e-05,
          "train_metric": 8.251111466961447e-06,
          "validation_metric": 7.574024493806064e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.741395429999102
        },
        {
          "epoch": 822,
          "updates": 13152,
          "samples_seen": 841728,
          "online_loss": 8.243788442996447e-05,
          "train_loss": 7.999318040674552e-05,
          "validation_loss": 7.312734669540077e-05,
          "train_metric": 8.056620572460815e-06,
          "validation_metric": 7.365119472524384e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.75178330899871
        },
        {
          "epoch": 823,
          "updates": 13168,
          "samples_seen": 842752,
          "online_loss": 8.420741392001219e-05,
          "train_loss": 8.795161556918174e-05,
          "validation_loss": 8.207687642425299e-05,
          "train_metric": 8.858162800606806e-06,
          "validation_metric": 8.266482836916111e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.762337624999418
        },
        {
          "epoch": 824,
          "updates": 13184,
          "samples_seen": 843776,
          "online_loss": 8.31050165288616e-05,
          "train_loss": 7.796350109856576e-05,
          "validation_loss": 7.221112173283473e-05,
          "train_metric": 7.852197995816823e-06,
          "validation_metric": 7.272840321093099e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.773200895999253
        },
        {
          "epoch": 825,
          "updates": 13200,
          "samples_seen": 844800,
          "online_loss": 8.035235759962234e-05,
          "train_loss": 8.191962842829525e-05,
          "validation_loss": 7.450606062775478e-05,
          "train_metric": 8.250644896179438e-06,
          "validation_metric": 7.503977030864917e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.783563363998837
        },
        {
          "epoch": 826,
          "updates": 13216,
          "samples_seen": 845824,
          "online_loss": 8.518218442077341e-05,
          "train_loss": 8.075580262811854e-05,
          "validation_loss": 7.292160444194451e-05,
          "train_metric": 8.133428309520241e-06,
          "validation_metric": 7.344397090491839e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.793965281000055
        },
        {
          "epoch": 827,
          "updates": 13232,
          "samples_seen": 846848,
          "online_loss": 8.35162707062409e-05,
          "train_loss": 8.701848855707794e-05,
          "validation_loss": 7.47852900531143e-05,
          "train_metric": 8.764181984588504e-06,
          "validation_metric": 7.532100426033139e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.804211173999647
        },
        {
          "epoch": 828,
          "updates": 13248,
          "samples_seen": 847872,
          "online_loss": 8.921714061216335e-05,
          "train_loss": 8.636401616968215e-05,
          "validation_loss": 7.698516128584743e-05,
          "train_metric": 8.698267265572213e-06,
          "validation_metric": 7.753664249321446e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.814612336998835
        },
        {
          "epoch": 829,
          "updates": 13264,
          "samples_seen": 848896,
          "online_loss": 7.993017720764328e-05,
          "train_loss": 8.16146784927696e-05,
          "validation_loss": 7.225028093671426e-05,
          "train_metric": 8.219932169595268e-06,
          "validation_metric": 7.276783435372636e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.824817220000114
        },
        {
          "epoch": 830,
          "updates": 13280,
          "samples_seen": 849920,
          "online_loss": 8.28249008009152e-05,
          "train_loss": 7.94652005424723e-05,
          "validation_loss": 7.296333933481947e-05,
          "train_metric": 8.003444236237556e-06,
          "validation_metric": 7.348600775003433e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.835191869999107
        },
        {
          "epoch": 831,
          "updates": 13296,
          "samples_seen": 850944,
          "online_loss": 7.932791800158157e-05,
          "train_loss": 7.785043271724135e-05,
          "validation_loss": 7.163645204855129e-05,
          "train_metric": 7.840812031645328e-06,
          "validation_metric": 7.214961897261674e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.845981425998616
        },
        {
          "epoch": 832,
          "updates": 13312,
          "samples_seen": 851968,
          "online_loss": 8.021102576094563e-05,
          "train_loss": 7.738944259472191e-05,
          "validation_loss": 7.100054062902927e-05,
          "train_metric": 7.794380508130416e-06,
          "validation_metric": 7.150913916120771e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.85717785199995
        },
        {
          "epoch": 833,
          "updates": 13328,
          "samples_seen": 852992,
          "online_loss": 8.179788756024209e-05,
          "train_loss": 8.586350304540247e-05,
          "validation_loss": 7.320818986045197e-05,
          "train_metric": 8.647857612231746e-06,
          "validation_metric": 7.373260814347304e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.867573354998967
        },
        {
          "epoch": 834,
          "updates": 13344,
          "samples_seen": 854016,
          "online_loss": 9.238059510607854e-05,
          "train_loss": 8.198369323508814e-05,
          "validation_loss": 7.334741530939937e-05,
          "train_metric": 8.257096851593815e-06,
          "validation_metric": 7.3872829489118885e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.877933056999609
        },
        {
          "epoch": 835,
          "updates": 13360,
          "samples_seen": 855040,
          "online_loss": 8.280760971501877e-05,
          "train_loss": 8.203073957702145e-05,
          "validation_loss": 7.85693118814379e-05,
          "train_metric": 8.261836228484754e-06,
          "validation_metric": 7.913213266874664e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.94381548199999
        },
        {
          "epoch": 836,
          "updates": 13376,
          "samples_seen": 856064,
          "online_loss": 8.088021172625304e-05,
          "train_loss": 7.739236752968282e-05,
          "validation_loss": 7.064123201416805e-05,
          "train_metric": 7.794676093908492e-06,
          "validation_metric": 7.114726486179279e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.955246673998772
        },
        {
          "epoch": 837,
          "updates": 13392,
          "samples_seen": 857088,
          "online_loss": 8.037310317376978e-05,
          "train_loss": 8.135141979437321e-05,
          "validation_loss": 7.36513247829862e-05,
          "train_metric": 8.19341676105978e-06,
          "validation_metric": 7.4178915383527055e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.96581678199982
        },
        {
          "epoch": 838,
          "updates": 13408,
          "samples_seen": 858112,
          "online_loss": 7.936909923955682e-05,
          "train_loss": 8.072147466009483e-05,
          "validation_loss": 7.164479029597715e-05,
          "train_metric": 8.129971320158802e-06,
          "validation_metric": 7.215800451376708e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.97626007100007
        },
        {
          "epoch": 839,
          "updates": 13424,
          "samples_seen": 859136,
          "online_loss": 8.062006804721022e-05,
          "train_loss": 7.734772225376219e-05,
          "validation_loss": 7.144559640437365e-05,
          "train_metric": 7.790179552102927e-06,
          "validation_metric": 7.195739271992352e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.986826970000038
        },
        {
          "epoch": 840,
          "updates": 13440,
          "samples_seen": 860160,
          "online_loss": 7.885368427196227e-05,
          "train_loss": 7.707863551331684e-05,
          "validation_loss": 7.012749119894579e-05,
          "train_metric": 7.763077519484796e-06,
          "validation_metric": 7.0629844231007155e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.997786186999292
        },
        {
          "epoch": 841,
          "updates": 13456,
          "samples_seen": 861184,
          "online_loss": 8.017402865334589e-05,
          "train_loss": 7.721847214270383e-05,
          "validation_loss": 7.214790093712509e-05,
          "train_metric": 7.777162863931153e-06,
          "validation_metric": 7.266472039191285e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.00848699099879
        },
        {
          "epoch": 842,
          "updates": 13472,
          "samples_seen": 862208,
          "online_loss": 7.896476290625287e-05,
          "train_loss": 7.747800555080175e-05,
          "validation_loss": 6.989630492171273e-05,
          "train_metric": 7.803301741660107e-06,
          "validation_metric": 7.039699539745925e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.019447574999504
        },
        {
          "epoch": 843,
          "updates": 13488,
          "samples_seen": 863232,
          "online_loss": 7.97321044956334e-05,
          "train_loss": 8.187975618056953e-05,
          "validation_loss": 7.516978803323582e-05,
          "train_metric": 8.246628567576408e-06,
          "validation_metric": 7.570825800939929e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.029876316999434
        },
        {
          "epoch": 844,
          "updates": 13504,
          "samples_seen": 864256,
          "online_loss": 8.198166074180335e-05,
          "train_loss": 7.797616126481444e-05,
          "validation_loss": 7.04093836247921e-05,
          "train_metric": 7.85347401688341e-06,
          "validation_metric": 7.09137566445861e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.040304667998498
        },
        {
          "epoch": 845,
          "updates": 13520,
          "samples_seen": 865280,
          "online_loss": 8.059911760938121e-05,
          "train_loss": 8.127765613608062e-05,
          "validation_loss": 7.270582136698067e-05,
          "train_metric": 8.185987098840997e-06,
          "validation_metric": 7.322664714592975e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.050550074998682
        },
        {
          "epoch": 846,
          "updates": 13536,
          "samples_seen": 866304,
          "online_loss": 8.216528635784925e-05,
          "train_loss": 7.773395191179588e-05,
          "validation_loss": 7.179489330155775e-05,
          "train_metric": 7.829078640497755e-06,
          "validation_metric": 7.230918527056929e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.06099927499963
        },
        {
          "epoch": 847,
          "updates": 13552,
          "samples_seen": 867328,
          "online_loss": 8.396743851335486e-05,
          "train_loss": 7.67203455325216e-05,
          "validation_loss": 6.972885603317991e-05,
          "train_metric": 7.726992407697253e-06,
          "validation_metric": 7.022834324743599e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.071894898999744
        },
        {
          "epoch": 848,
          "updates": 13568,
          "samples_seen": 868352,
          "online_loss": 7.802612503837736e-05,
          "train_loss": 7.816623110556975e-05,
          "validation_loss": 7.115124026313424e-05,
          "train_metric": 7.872616151871625e-06,
          "validation_metric": 7.166092018451309e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.082674592998956
        },
        {
          "epoch": 849,
          "updates": 13584,
          "samples_seen": 869376,
          "online_loss": 7.903155574240373e-05,
          "train_loss": 7.682358409510925e-05,
          "validation_loss": 7.021710916887969e-05,
          "train_metric": 7.737390660622623e-06,
          "validation_metric": 7.072009793773759e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.093057538999346
        },
        {
          "epoch": 850,
          "updates": 13600,
          "samples_seen": 870400,
          "online_loss": 7.946074629217037e-05,
          "train_loss": 7.647502206964418e-05,
          "validation_loss": 6.973810377530754e-05,
          "train_metric": 7.70228507462889e-06,
          "validation_metric": 7.023767466307618e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.103370674998587
        },
        {
          "epoch": 851,
          "updates": 13616,
          "samples_seen": 871424,
          "online_loss": 7.853804640944873e-05,
          "train_loss": 7.667799945920706e-05,
          "validation_loss": 6.992892303969711e-05,
          "train_metric": 7.722725968051236e-06,
          "validation_metric": 7.042984634608729e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.113638380998964
        },
        {
          "epoch": 852,
          "updates": 13632,
          "samples_seen": 872448,
          "online_loss": 8.089291145552124e-05,
          "train_loss": 8.682363841217011e-05,
          "validation_loss": 7.7156386396382e-05,
          "train_metric": 8.744559636397753e-06,
          "validation_metric": 7.77090826886706e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.124025338000138
        },
        {
          "epoch": 853,
          "updates": 13648,
          "samples_seen": 873472,
          "online_loss": 8.442301827926713e-05,
          "train_loss": 7.70115657360293e-05,
          "validation_loss": 7.022292993497103e-05,
          "train_metric": 7.756322702334728e-06,
          "validation_metric": 7.072595508361701e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.134280100999604
        },
        {
          "epoch": 854,
          "updates": 13664,
          "samples_seen": 874496,
          "online_loss": 9.526668577564124e-05,
          "train_loss": 7.975022890605032e-05,
          "validation_loss": 7.049446139717475e-05,
          "train_metric": 8.032150617509615e-06,
          "validation_metric": 7.099944014044013e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.144771966999542
        },
        {
          "epoch": 855,
          "updates": 13680,
          "samples_seen": 875520,
          "online_loss": 7.865859424782684e-05,
          "train_loss": 7.963724783621728e-05,
          "validation_loss": 7.010580156929791e-05,
          "train_metric": 8.020771019801032e-06,
          "validation_metric": 7.060799816827057e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.155048820999582
        },
        {
          "epoch": 856,
          "updates": 13696,
          "samples_seen": 876544,
          "online_loss": 8.2490023032733e-05,
          "train_loss": 8.397748752031475e-05,
          "validation_loss": 7.704988092882559e-05,
          "train_metric": 8.457905096292961e-06,
          "validation_metric": 7.76018168835435e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.165545279998696
        },
        {
          "epoch": 857,
          "updates": 13712,
          "samples_seen": 877568,
          "online_loss": 8.509134841006016e-05,
          "train_loss": 7.707850454607978e-05,
          "validation_loss": 6.939723971299827e-05,
          "train_metric": 7.76306478655897e-06,
          "validation_metric": 6.989435405557742e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.176397794999502
        },
        {
          "epoch": 858,
          "updates": 13728,
          "samples_seen": 878592,
          "online_loss": 8.329034335474716e-05,
          "train_loss": 9.445824980502948e-05,
          "validation_loss": 7.780315354466438e-05,
          "train_metric": 9.513488294032868e-06,
          "validation_metric": 7.836049007892143e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.18717492199903
        },
        {
          "epoch": 859,
          "updates": 13744,
          "samples_seen": 879616,
          "online_loss": 8.950663595896913e-05,
          "train_loss": 7.810819079168141e-05,
          "validation_loss": 7.270744390552863e-05,
          "train_metric": 7.86677082942333e-06,
          "validation_metric": 7.322828423639294e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.197527151998656
        },
        {
          "epoch": 860,
          "updates": 13760,
          "samples_seen": 880640,
          "online_loss": 7.866698047109821e-05,
          "train_loss": 7.616390939801931e-05,
          "validation_loss": 6.871702498756349e-05,
          "train_metric": 7.670950253668707e-06,
          "validation_metric": 6.9209277171466965e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.2088228349985
        },
        {
          "epoch": 861,
          "updates": 13776,
          "samples_seen": 881664,
          "online_loss": 7.790756035319646e-05,
          "train_loss": 7.642092532478273e-05,
          "validation_loss": 7.043153163976967e-05,
          "train_metric": 7.696835382375866e-06,
          "validation_metric": 7.093606200214708e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.21924589199989
        },
        {
          "epoch": 862,
          "updates": 13792,
          "samples_seen": 882688,
          "online_loss": 7.80213081270631e-05,
          "train_loss": 7.732375524938107e-05,
          "validation_loss": 6.989820394665003e-05,
          "train_metric": 7.78776484366972e-06,
          "validation_metric": 7.039890988380648e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.229687172999547
        },
        {
          "epoch": 863,
          "updates": 13808,
          "samples_seen": 883712,
          "online_loss": 7.813952515789424e-05,
          "train_loss": 7.616265793330967e-05,
          "validation_loss": 6.943296466488391e-05,
          "train_metric": 7.67082383390516e-06,
          "validation_metric": 6.993032911850605e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.24034319599923
        },
        {
          "epoch": 864,
          "updates": 13824,
          "samples_seen": 884736,
          "online_loss": 7.838004103177809e-05,
          "train_loss": 7.647788879694417e-05,
          "validation_loss": 6.913102697581053e-05,
          "train_metric": 7.70257247495465e-06,
          "validation_metric": 6.962624865991529e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.25066540699845
        },
        {
          "epoch": 865,
          "updates": 13840,
          "samples_seen": 885760,
          "online_loss": 7.800030357429932e-05,
          "train_loss": 7.607472798554227e-05,
          "validation_loss": 6.99730881024152e-05,
          "train_metric": 7.661968083993997e-06,
          "validation_metric": 7.047433427942451e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.260990812999808
        },
        {
          "epoch": 866,
          "updates": 13856,
          "samples_seen": 886784,
          "online_loss": 7.875507890275912e-05,
          "train_loss": 7.661843847017735e-05,
          "validation_loss": 7.005625229794532e-05,
          "train_metric": 7.716727850493044e-06,
          "validation_metric": 7.055808055156376e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.271284695998474
        },
        {
          "epoch": 867,
          "updates": 13872,
          "samples_seen": 887808,
          "online_loss": 7.799397758390114e-05,
          "train_loss": 7.875058508943766e-05,
          "validation_loss": 7.006989471847191e-05,
          "train_metric": 7.931470463518053e-06,
          "validation_metric": 7.057182756398106e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.281853333999607
        },
        {
          "epoch": 868,
          "updates": 13888,
          "samples_seen": 888832,
          "online_loss": 8.132922812364995e-05,
          "train_loss": 7.766344060655683e-05,
          "validation_loss": 7.138632645364851e-05,
          "train_metric": 7.821977305866312e-06,
          "validation_metric": 7.189768439275213e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.292251752998709
        },
        {
          "epoch": 869,
          "updates": 13904,
          "samples_seen": 889856,
          "online_loss": 8.209409020309977e-05,
          "train_loss": 7.678906695218757e-05,
          "validation_loss": 6.923187902430072e-05,
          "train_metric": 7.733913662377745e-06,
          "validation_metric": 6.972781648073578e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.302555984999344
        },
        {
          "epoch": 870,
          "updates": 13920,
          "samples_seen": 890880,
          "online_loss": 7.834162988729076e-05,
          "train_loss": 7.786852074787021e-05,
          "validation_loss": 6.895254773553461e-05,
          "train_metric": 7.842632840038277e-06,
          "validation_metric": 6.944648248463636e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.312887672000215
        },
        {
          "epoch": 871,
          "updates": 13936,
          "samples_seen": 891904,
          "online_loss": 7.71241591337457e-05,
          "train_loss": 7.809225644450635e-05,
          "validation_loss": 7.19929303159006e-05,
          "train_metric": 7.865165571274702e-06,
          "validation_metric": 7.250864655361511e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.379507337000177
        },
        {
          "epoch": 872,
          "updates": 13952,
          "samples_seen": 892928,
          "online_loss": 7.903168443590403e-05,
          "train_loss": 7.60187758714892e-05,
          "validation_loss": 6.802983261877671e-05,
          "train_metric": 7.65633194532711e-06,
          "validation_metric": 6.8517156250891276e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.391043711999373
        },
        {
          "epoch": 873,
          "updates": 13968,
          "samples_seen": 893952,
          "online_loss": 7.889807420724537e-05,
          "train_loss": 9.373709326609969e-05,
          "validation_loss": 7.672419451409951e-05,
          "train_metric": 9.440856956643984e-06,
          "validation_metric": 7.727379852440208e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.402004322999346
        },
        {
          "epoch": 874,
          "updates": 13984,
          "samples_seen": 894976,
          "online_loss": 8.892777054825274e-05,
          "train_loss": 8.094417717074975e-05,
          "validation_loss": 7.417950837407261e-05,
          "train_metric": 8.152400368999224e-06,
          "validation_metric": 7.471088338206755e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.412507551998715
        },
        {
          "epoch": 875,
          "updates": 14000,
          "samples_seen": 896000,
          "online_loss": 7.785534944559913e-05,
          "train_loss": 7.71455088397488e-05,
          "validation_loss": 6.925719935679808e-05,
          "train_metric": 7.769813237246126e-06,
          "validation_metric": 6.975331416469999e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.42300573600005
        },
        {
          "epoch": 876,
          "updates": 14016,
          "samples_seen": 897024,
          "online_loss": 7.811115710865124e-05,
          "train_loss": 7.556522177765146e-05,
          "validation_loss": 6.934798875590786e-05,
          "train_metric": 7.610653028677916e-06,
          "validation_metric": 6.984475476201624e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.433486587999141
        },
        {
          "epoch": 877,
          "updates": 14032,
          "samples_seen": 898048,
          "online_loss": 7.80775728799199e-05,
          "train_loss": 7.682599971303716e-05,
          "validation_loss": 6.940792081877589e-05,
          "train_metric": 7.737633495707996e-06,
          "validation_metric": 6.9905122472846415e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.443833377999908
        },
        {
          "epoch": 878,
          "updates": 14048,
          "samples_seen": 899072,
          "online_loss": 7.760563767078565e-05,
          "train_loss": 7.571930473204702e-05,
          "validation_loss": 6.775406654924154e-05,
          "train_metric": 7.626169917784864e-06,
          "validation_metric": 6.823942385381088e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.454863729999488
        },
        {
          "epoch": 879,
          "updates": 14064,
          "samples_seen": 900096,
          "online_loss": 7.919556014712725e-05,
          "train_loss": 8.119108679238707e-05,
          "validation_loss": 7.302690937649459e-05,
          "train_metric": 8.177268682629801e-06,
          "validation_metric": 7.355002708209213e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.465662081000119
        },
        {
          "epoch": 880,
          "updates": 14080,
          "samples_seen": 901120,
          "online_loss": 7.763971439089801e-05,
          "train_loss": 7.593737973365933e-05,
          "validation_loss": 6.861662404844537e-05,
          "train_metric": 7.648134669580031e-06,
          "validation_metric": 6.910815500305034e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.47609673199986
        },
        {
          "epoch": 881,
          "updates": 14096,
          "samples_seen": 902144,
          "online_loss": 7.63140180879418e-05,
          "train_loss": 7.956449553603306e-05,
          "validation_loss": 7.022968929959461e-05,
          "train_metric": 8.013444130483549e-06,
          "validation_metric": 7.073276719893329e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.486417250998784
        },
        {
          "epoch": 882,
          "updates": 14112,
          "samples_seen": 903168,
          "online_loss": 8.016216838768742e-05,
          "train_loss": 7.562004611827433e-05,
          "validation_loss": 6.966294313315302e-05,
          "train_metric": 7.616174116265029e-06,
          "validation_metric": 7.01619683241006e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.496845322999434
        },
        {
          "epoch": 883,
          "updates": 14128,
          "samples_seen": 904192,
          "online_loss": 7.799196123414731e-05,
          "train_loss": 8.331469871336594e-05,
          "validation_loss": 7.018677570158616e-05,
          "train_metric": 8.391151823161636e-06,
          "validation_metric": 7.068954801070504e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.507585820998429
        },
        {
          "epoch": 884,
          "updates": 14144,
          "samples_seen": 905216,
          "online_loss": 8.158623336385062e-05,
          "train_loss": 7.593088957946748e-05,
          "validation_loss": 6.854299863334745e-05,
          "train_metric": 7.647480742889456e-06,
          "validation_metric": 6.9034003900014795e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.51792293499966
        },
        {
          "epoch": 885,
          "updates": 14160,
          "samples_seen": 906240,
          "online_loss": 8.055641183091211e-05,
          "train_loss": 7.484399975510314e-05,
          "validation_loss": 6.803211726946756e-05,
          "train_metric": 7.538013505836716e-06,
          "validation_metric": 6.851946181996027e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.528271404999032
        },
        {
          "epoch": 886,
          "updates": 14176,
          "samples_seen": 907264,
          "online_loss": 7.735636859251827e-05,
          "train_loss": 7.486255344701931e-05,
          "validation_loss": 6.765782018192112e-05,
          "train_metric": 7.539882517448859e-06,
          "validation_metric": 6.814247171860188e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.539481253999838
        },
        {
          "epoch": 887,
          "updates": 14192,
          "samples_seen": 908288,
          "online_loss": 7.703382721047092e-05,
          "train_loss": 7.489717972930521e-05,
          "validation_loss": 6.828975165262818e-05,
          "train_metric": 7.543369974882808e-06,
          "validation_metric": 6.877893611090258e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.549909958999706
        },
        {
          "epoch": 888,
          "updates": 14208,
          "samples_seen": 909312,
          "online_loss": 7.681855367991375e-05,
          "train_loss": 7.602061668876559e-05,
          "validation_loss": 6.806624151067808e-05,
          "train_metric": 7.656518391740974e-06,
          "validation_metric": 6.855382707726676e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.560568888999114
        },
        {
          "epoch": 889,
          "updates": 14224,
          "samples_seen": 910336,
          "online_loss": 7.958477658576157e-05,
          "train_loss": 7.548482972197235e-05,
          "validation_loss": 6.827976903878152e-05,
          "train_metric": 7.602555342600681e-06,
          "validation_metric": 6.876888164697448e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.570835327998793
        },
        {
          "epoch": 890,
          "updates": 14240,
          "samples_seen": 911360,
          "online_loss": 7.978394728525018e-05,
          "train_loss": 8.722092752577737e-05,
          "validation_loss": 7.308286149054766e-05,
          "train_metric": 8.784571946307551e-06,
          "validation_metric": 7.360637937381398e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.581202437999309
        },
        {
          "epoch": 891,
          "updates": 14256,
          "samples_seen": 912384,
          "online_loss": 8.638417943984678e-05,
          "train_loss": 7.557529170298949e-05,
          "validation_loss": 7.010521221673116e-05,
          "train_metric": 7.611666660523042e-06,
          "validation_metric": 7.060740244924091e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.591522627999439
        },
        {
          "epoch": 892,
          "updates": 14272,
          "samples_seen": 913408,
          "online_loss": 8.660397406856646e-05,
          "train_loss": 7.457875472027808e-05,
          "validation_loss": 6.755055801477283e-05,
          "train_metric": 7.51129891796154e-06,
          "validation_metric": 6.803445103287231e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.602467644999706
        },
        {
          "epoch": 893,
          "updates": 14288,
          "samples_seen": 914432,
          "online_loss": 7.946522009660839e-05,
          "train_loss": 8.027235162444413e-05,
          "validation_loss": 7.095455657690763e-05,
          "train_metric": 8.084736691671424e-06,
          "validation_metric": 7.1462832238466945e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.612875750999592
        },
        {
          "epoch": 894,
          "updates": 14304,
          "samples_seen": 915456,
          "online_loss": 7.833379504518234e-05,
          "train_loss": 7.500010542571545e-05,
          "validation_loss": 6.806499732192606e-05,
          "train_metric": 7.5537359407462645e-06,
          "validation_metric": 6.855257652205182e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.623309623999376
        },
        {
          "epoch": 895,
          "updates": 14320,
          "samples_seen": 916480,
          "online_loss": 7.71054160395579e-05,
          "train_loss": 7.485339301638305e-05,
          "validation_loss": 6.686915003228933e-05,
          "train_metric": 7.538959835073911e-06,
          "validation_metric": 6.7348164520808496e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.634141104999799
        },
        {
          "epoch": 896,
          "updates": 14336,
          "samples_seen": 917504,
          "online_loss": 7.905460506663076e-05,
          "train_loss": 8.444619743386284e-05,
          "validation_loss": 7.188301242422312e-05,
          "train_metric": 8.505112418788485e-06,
          "validation_metric": 7.23979383110418e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.644633935999082
        },
        {
          "epoch": 897,
          "updates": 14352,
          "samples_seen": 918528,
          "online_loss": 8.314726278513263e-05,
          "train_loss": 7.545600965386257e-05,
          "validation_loss": 6.83740436215885e-05,
          "train_metric": 7.5996531450073235e-06,
          "validation_metric": 6.886382834636606e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.655037611999433
        },
        {
          "epoch": 898,
          "updates": 14368,
          "samples_seen": 919552,
          "online_loss": 7.987919298102497e-05,
          "train_loss": 7.903460209490731e-05,
          "validation_loss": 6.866087642265484e-05,
          "train_metric": 7.960074981383514e-06,
          "validation_metric": 6.9152724790910725e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.665670888000022
        },
        {
          "epoch": 899,
          "updates": 14384,
          "samples_seen": 920576,
          "online_loss": 8.38739397295285e-05,
          "train_loss": 7.483169611077756e-05,
          "validation_loss": 6.906017370056361e-05,
          "train_metric": 7.536774774052901e-06,
          "validation_metric": 6.955487151572015e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.675944366999829
        },
        {
          "epoch": 900,
          "updates": 14400,
          "samples_seen": 921600,
          "online_loss": 7.631465632584877e-05,
          "train_loss": 7.578475924674422e-05,
          "validation_loss": 6.710119487252086e-05,
          "train_metric": 7.632763299625367e-06,
          "validation_metric": 6.758186827937607e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.686606726998434
        },
        {
          "epoch": 901,
          "updates": 14416,
          "samples_seen": 922624,
          "online_loss": 7.867781209824898e-05,
          "train_loss": 7.608093437738717e-05,
          "validation_loss": 6.945880159037188e-05,
          "train_metric": 7.662592906854115e-06,
          "validation_metric": 6.995635885687079e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.696937250999326
        },
        {
          "epoch": 902,
          "updates": 14432,
          "samples_seen": 923648,
          "online_loss": 7.971733680278703e-05,
          "train_loss": 7.569448644062504e-05,
          "validation_loss": 6.769353058189154e-05,
          "train_metric": 7.623671990586445e-06,
          "validation_metric": 6.8178442234057e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.707332549998682
        },
        {
          "epoch": 903,
          "updates": 14448,
          "samples_seen": 924672,
          "online_loss": 7.57680541028094e-05,
          "train_loss": 7.55045548430644e-05,
          "validation_loss": 6.78300712024793e-05,
          "train_metric": 7.604542133776704e-06,
          "validation_metric": 6.831596692791209e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.717622610000035
        },
        {
          "epoch": 904,
          "updates": 14464,
          "samples_seen": 925696,
          "online_loss": 8.365713665625663e-05,
          "train_loss": 7.498206832678989e-05,
          "validation_loss": 6.740758544765413e-05,
          "train_metric": 7.551919679826824e-06,
          "validation_metric": 6.78904507367406e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.728061587999036
        },
        {
          "epoch": 905,
          "updates": 14480,
          "samples_seen": 926720,
          "online_loss": 8.29825148684904e-05,
          "train_loss": 7.883085345383734e-05,
          "validation_loss": 6.848965131212026e-05,
          "train_metric": 7.939554961922113e-06,
          "validation_metric": 6.898027550050756e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.738325124999392
        },
        {
          "epoch": 906,
          "updates": 14496,
          "samples_seen": 927744,
          "online_loss": 7.887501578807132e-05,
          "train_loss": 7.498794002458453e-05,
          "validation_loss": 6.873584061395377e-05,
          "train_metric": 7.552510851382976e-06,
          "validation_metric": 6.922821739863139e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.748673871999927
        },
        {
          "epoch": 907,
          "updates": 14512,
          "samples_seen": 928768,
          "online_loss": 7.809970497874019e-05,
          "train_loss": 7.655614899704233e-05,
          "validation_loss": 6.896265404066071e-05,
          "train_metric": 7.710454156040214e-06,
          "validation_metric": 6.945665063540218e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.813367304999701
        },
        {
          "epoch": 908,
          "updates": 14528,
          "samples_seen": 929792,
          "online_loss": 8.391148094233358e-05,
          "train_loss": 8.156875264830887e-05,
          "validation_loss": 7.139184890547767e-05,
          "train_metric": 8.215305570047349e-06,
          "validation_metric": 7.1903255047800485e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.824183696999171
        },
        {
          "epoch": 909,
          "updates": 14544,
          "samples_seen": 930816,
          "online_loss": 7.855291937630682e-05,
          "train_loss": 7.441233901772648e-05,
          "validation_loss": 6.662278610747308e-05,
          "train_metric": 7.494537840102566e-06,
          "validation_metric": 6.71000316287973e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.835310338999989
        },
        {
          "epoch": 910,
          "updates": 14560,
          "samples_seen": 931840,
          "online_loss": 7.491341625609493e-05,
          "train_loss": 7.404384814435616e-05,
          "validation_loss": 6.852884689578786e-05,
          "train_metric": 7.457424544554669e-06,
          "validation_metric": 6.9019743023091e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.845771382999374
        },
        {
          "epoch": 911,
          "updates": 14576,
          "samples_seen": 932864,
          "online_loss": 7.806828853063053e-05,
          "train_loss": 7.801364699844271e-05,
          "validation_loss": 6.836922693764791e-05,
          "train_metric": 7.857248419895768e-06,
          "validation_metric": 6.885898073960561e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.856427570999585
        },
        {
          "epoch": 912,
          "updates": 14592,
          "samples_seen": 933888,
          "online_loss": 7.914588854873728e-05,
          "train_loss": 7.665564044145867e-05,
          "validation_loss": 7.00067394063808e-05,
          "train_metric": 7.720474968664348e-06,
          "validation_metric": 7.050823114695959e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.867099523999059
        },
        {
          "epoch": 913,
          "updates": 14608,
          "samples_seen": 934912,
          "online_loss": 7.584432205476332e-05,
          "train_loss": 7.394242857117206e-05,
          "validation_loss": 6.737712828908116e-05,
          "train_metric": 7.4472109190537594e-06,
          "validation_metric": 6.785978257539682e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.877384822999375
        },
        {
          "epoch": 914,
          "updates": 14624,
          "samples_seen": 935936,
          "online_loss": 7.630530171809369e-05,
          "train_loss": 7.62977433623746e-05,
          "validation_loss": 6.823510921094567e-05,
          "train_metric": 7.684428965148982e-06,
          "validation_metric": 6.872389349155128e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.888000259999899
        },
        {
          "epoch": 915,
          "updates": 14640,
          "samples_seen": 936960,
          "online_loss": 7.709526698818081e-05,
          "train_loss": 7.367862417595461e-05,
          "validation_loss": 6.780413968954235e-05,
          "train_metric": 7.420640940836165e-06,
          "validation_metric": 6.828984624007717e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.898278081998797
        },
        {
          "epoch": 916,
          "updates": 14656,
          "samples_seen": 937984,
          "online_loss": 7.602981486343197e-05,
          "train_loss": 7.582905527669936e-05,
          "validation_loss": 6.803152791690081e-05,
          "train_metric": 7.637225280632265e-06,
          "validation_metric": 6.851887064840412e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.90865859599944
        },
        {
          "epoch": 917,
          "updates": 14672,
          "samples_seen": 939008,
          "online_loss": 7.557783237643889e-05,
          "train_loss": 7.427371019730344e-05,
          "validation_loss": 6.71110610710457e-05,
          "train_metric": 7.480575732188299e-06,
          "validation_metric": 6.759179996151943e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.918922462999035
        },
        {
          "epoch": 918,
          "updates": 14688,
          "samples_seen": 940032,
          "online_loss": 7.59074962388695e-05,
          "train_loss": 7.357020513154566e-05,
          "validation_loss": 6.743406265741214e-05,
          "train_metric": 7.40972200219403e-06,
          "validation_metric": 6.79171262163436e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.929198852998525
        },
        {
          "epoch": 919,
          "updates": 14704,
          "samples_seen": 941056,
          "online_loss": 7.88642689713015e-05,
          "train_loss": 8.030702883843333e-05,
          "validation_loss": 6.923655018908903e-05,
          "train_metric": 8.088229151326232e-06,
          "validation_metric": 6.973252766329097e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.939442136999787
        },
        {
          "epoch": 920,
          "updates": 14720,
          "samples_seen": 942080,
          "online_loss": 7.412860918520892e-05,
          "train_loss": 7.523383828811347e-05,
          "validation_loss": 6.673830648651347e-05,
          "train_metric": 7.577276392112253e-06,
          "validation_metric": 6.721637419104809e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.949773107999135
        },
        {
          "epoch": 921,
          "updates": 14736,
          "samples_seen": 943104,
          "online_loss": 7.643298476978089e-05,
          "train_loss": 7.347977953031659e-05,
          "validation_loss": 6.703980034217238e-05,
          "train_metric": 7.400613412755774e-06,
          "validation_metric": 6.752003173460253e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.960244848998627
        },
        {
          "epoch": 922,
          "updates": 14752,
          "samples_seen": 944128,
          "online_loss": 7.506827000725025e-05,
          "train_loss": 7.462475332431495e-05,
          "validation_loss": 6.732299516443163e-05,
          "train_metric": 7.51593142922502e-06,
          "validation_metric": 6.780525836802553e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.970533828000043
        },
        {
          "epoch": 923,
          "updates": 14768,
          "samples_seen": 945152,
          "online_loss": 7.632505298715841e-05,
          "train_loss": 7.373101107077673e-05,
          "validation_loss": 6.673391908407211e-05,
          "train_metric": 7.425917374348501e-06,
          "validation_metric": 6.721194949932396e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.980779572999381
        },
        {
          "epoch": 924,
          "updates": 14784,
          "samples_seen": 946176,
          "online_loss": 7.502211974497186e-05,
          "train_loss": 7.458971958840266e-05,
          "validation_loss": 6.624536763411015e-05,
          "train_metric": 7.512403044529492e-06,
          "validation_metric": 6.671989922324428e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.99199580899949
        },
        {
          "epoch": 925,
          "updates": 14800,
          "samples_seen": 947200,
          "online_loss": 7.476184669030772e-05,
          "train_loss": 8.370889554498717e-05,
          "validation_loss": 7.548963912995532e-05,
          "train_metric": 8.430853085883427e-06,
          "validation_metric": 7.603039648529375e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.002635025999552
        },
        {
          "epoch": 926,
          "updates": 14816,
          "samples_seen": 948224,
          "online_loss": 8.032576215555309e-05,
          "train_loss": 7.403362542390823e-05,
          "validation_loss": 6.693026807624847e-05,
          "train_metric": 7.4563945418049116e-06,
          "validation_metric": 6.740971457475098e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.013029189998633
        },
        {
          "epoch": 927,
          "updates": 14832,
          "samples_seen": 949248,
          "online_loss": 7.767990496176935e-05,
          "train_loss": 7.482081855414435e-05,
          "validation_loss": 6.674398173345253e-05,
          "train_metric": 7.535678832937265e-06,
          "validation_metric": 6.722209946019575e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.023327995999352
        },
        {
          "epoch": 928,
          "updates": 14848,
          "samples_seen": 950272,
          "online_loss": 8.069166869972833e-05,
          "train_loss": 7.348146027652547e-05,
          "validation_loss": 6.681020022369921e-05,
          "train_metric": 7.400783943012357e-06,
          "validation_metric": 6.728879270667676e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.033669241000098
        },
        {
          "epoch": 929,
          "updates": 14864,
          "samples_seen": 951296,
          "online_loss": 9.549233755024034e-05,
          "train_loss": 8.773433364694938e-05,
          "validation_loss": 7.885768718551844e-05,
          "train_metric": 8.83628035808215e-06,
          "validation_metric": 7.94225707068108e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.04393943499963
        },
        {
          "epoch": 930,
          "updates": 14880,
          "samples_seen": 952320,
          "online_loss": 8.561826598452171e-05,
          "train_loss": 9.522316395305097e-05,
          "validation_loss": 8.166671614162624e-05,
          "train_metric": 9.590528861735947e-06,
          "validation_metric": 8.225171768572181e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.054341208999176
        },
        {
          "epoch": 931,
          "updates": 14896,
          "samples_seen": 953344,
          "online_loss": 8.800855698609666e-05,
          "train_loss": 7.85302763688378e-05,
          "validation_loss": 6.715863128192723e-05,
          "train_metric": 7.909282430773601e-06,
          "validation_metric": 6.7639712142408825e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.064637694998964
        },
        {
          "epoch": 932,
          "updates": 14912,
          "samples_seen": 954368,
          "online_loss": 8.251408644355251e-05,
          "train_loss": 7.810549141140655e-05,
          "validation_loss": 6.847894110251218e-05,
          "train_metric": 7.8664988905075e-06,
          "validation_metric": 6.896948434587102e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.074949926000045
        },
        {
          "epoch": 933,
          "updates": 14928,
          "samples_seen": 955392,
          "online_loss": 7.585280650346249e-05,
          "train_loss": 7.369793456746265e-05,
          "validation_loss": 6.655055040027946e-05,
          "train_metric": 7.4225858952559065e-06,
          "validation_metric": 6.702727660012897e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.085252037999453
        },
        {
          "epoch": 934,
          "updates": 14944,
          "samples_seen": 956416,
          "online_loss": 7.51955451505637e-05,
          "train_loss": 7.339443254750222e-05,
          "validation_loss": 6.630348070757464e-05,
          "train_metric": 7.39201868782402e-06,
          "validation_metric": 6.67784388497239e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.095547323999199
        },
        {
          "epoch": 935,
          "updates": 14960,
          "samples_seen": 957440,
          "online_loss": 7.644639390491648e-05,
          "train_loss": 7.950230792630464e-05,
          "validation_loss": 7.050920248730108e-05,
          "train_metric": 8.00718134996714e-06,
          "validation_metric": 7.101429218892008e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.106073271999776
        },
        {
          "epoch": 936,
          "updates": 14976,
          "samples_seen": 958464,
          "online_loss": 8.021319627005141e-05,
          "train_loss": 8.068625902524218e-05,
          "validation_loss": 7.201392145361751e-05,
          "train_metric": 8.12642520031659e-06,
          "validation_metric": 7.252978775795782e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.11636081999859
        },
        {
          "epoch": 937,
          "updates": 14992,
          "samples_seen": 959488,
          "online_loss": 8.210882265302644e-05,
          "train_loss": 7.383407501038164e-05,
          "validation_loss": 6.599152402486652e-05,
          "train_metric": 7.436297892127186e-06,
          "validation_metric": 6.646424481004942e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.127214362999439
        },
        {
          "epoch": 938,
          "updates": 15008,
          "samples_seen": 960512,
          "online_loss": 7.621484610353946e-05,
          "train_loss": 7.256831304403022e-05,
          "validation_loss": 6.646791007369757e-05,
          "train_metric": 7.3088149292743765e-06,
          "validation_metric": 6.694403964502271e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.137680200999966
        },
        {
          "epoch": 939,
          "updates": 15024,
          "samples_seen": 961536,
          "online_loss": 7.49696489492635e-05,
          "train_loss": 7.550320151494816e-05,
          "validation_loss": 6.78665455779992e-05,
          "train_metric": 7.604406164318789e-06,
          "validation_metric": 6.8352701418916695e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.147920186000192
        },
        {
          "epoch": 940,
          "updates": 15040,
          "samples_seen": 962560,
          "online_loss": 7.880376688262913e-05,
          "train_loss": 7.691027713008225e-05,
          "validation_loss": 6.820017733843997e-05,
          "train_metric": 7.746120900264941e-06,
          "validation_metric": 6.86887142364867e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.158471329999884
        },
        {
          "epoch": 941,
          "updates": 15056,
          "samples_seen": 963584,
          "online_loss": 7.794726730026014e-05,
          "train_loss": 7.807306246832013e-05,
          "validation_loss": 6.82626196066849e-05,
          "train_metric": 7.863232895033434e-06,
          "validation_metric": 6.875161489006132e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.169001066999044
        },
        {
          "epoch": 942,
          "updates": 15072,
          "samples_seen": 964608,
          "online_loss": 7.814645391590602e-05,
          "train_loss": 7.314346294151619e-05,
          "validation_loss": 6.589767872355878e-05,
          "train_metric": 7.366741101577645e-06,
          "validation_metric": 6.636973466811469e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.179848275998665
        },
        {
          "epoch": 943,
          "updates": 15088,
          "samples_seen": 965632,
          "online_loss": 7.505125904572196e-05,
          "train_loss": 7.96968161012046e-05,
          "validation_loss": 6.851109355920926e-05,
          "train_metric": 8.026770956348628e-06,
          "validation_metric": 6.900186235725414e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.24454002299899
        },
        {
          "epoch": 944,
          "updates": 15104,
          "samples_seen": 966656,
          "online_loss": 8.057830996222037e-05,
          "train_loss": 7.247752364492044e-05,
          "validation_loss": 6.565974035765976e-05,
          "train_metric": 7.29966996004805e-06,
          "validation_metric": 6.613008736167103e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.255920114999753
        },
        {
          "epoch": 945,
          "updates": 15120,
          "samples_seen": 967680,
          "online_loss": 7.478670704585966e-05,
          "train_loss": 7.260705751832575e-05,
          "validation_loss": 6.463683530455455e-05,
          "train_metric": 7.312717116292333e-06,
          "validation_metric": 6.509985269076424e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.266933553999479
        },
        {
          "epoch": 946,
          "updates": 15136,
          "samples_seen": 968704,
          "online_loss": 7.429306469930452e-05,
          "train_loss": 7.213665958261117e-05,
          "validation_loss": 6.504652992589399e-05,
          "train_metric": 7.2653410825296305e-06,
          "validation_metric": 6.5512490436958615e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.27744319699923
        },
        {
          "epoch": 947,
          "updates": 15152,
          "samples_seen": 969728,
          "online_loss": 7.807294446138258e-05,
          "train_loss": 7.734986866125837e-05,
          "validation_loss": 6.90123051754199e-05,
          "train_metric": 7.790395102347247e-06,
          "validation_metric": 6.9506672843999695e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.28801819399996
        },
        {
          "epoch": 948,
          "updates": 15168,
          "samples_seen": 970752,
          "online_loss": 7.406125928355323e-05,
          "train_loss": 7.364678458543494e-05,
          "validation_loss": 6.662134546786547e-05,
          "train_metric": 7.417434517265065e-06,
          "validation_metric": 6.709858098474797e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.298401805999674
        },
        {
          "epoch": 949,
          "updates": 15184,
          "samples_seen": 971776,
          "online_loss": 7.62207021125505e-05,
          "train_loss": 7.256606477312744e-05,
          "validation_loss": 6.455295078922063e-05,
          "train_metric": 7.308588010346284e-06,
          "validation_metric": 6.501536063296953e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.309187990998907
        },
        {
          "epoch": 950,
          "updates": 15200,
          "samples_seen": 972800,
          "online_loss": 7.561472307315853e-05,
          "train_loss": 7.351852400461212e-05,
          "validation_loss": 6.473394751083106e-05,
          "train_metric": 7.404515599773731e-06,
          "validation_metric": 6.519765520351939e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.31968102799874
        },
        {
          "epoch": 951,
          "updates": 15216,
          "samples_seen": 973824,
          "online_loss": 7.411441083604586e-05,
          "train_loss": 7.200828258646652e-05,
          "validation_loss": 6.503587064798921e-05,
          "train_metric": 7.252410796354525e-06,
          "validation_metric": 6.550174475705717e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.330009266999696
        },
        {
          "epoch": 952,
          "updates": 15232,
          "samples_seen": 974848,
          "online_loss": 7.62489080443629e-05,
          "train_loss": 8.523510041413829e-05,
          "validation_loss": 7.552572787972167e-05,
          "train_metric": 8.58456678543007e-06,
          "validation_metric": 7.606674898852361e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.340323061998788
        },
        {
          "epoch": 953,
          "updates": 15248,
          "samples_seen": 975872,
          "online_loss": 8.25215188342554e-05,
          "train_loss": 7.208540773717687e-05,
          "validation_loss": 6.424429739126936e-05,
          "train_metric": 7.260178790602367e-06,
          "validation_metric": 6.4704495343903545e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.351457587999903
        },
        {
          "epoch": 954,
          "updates": 15264,
          "samples_seen": 976896,
          "online_loss": 7.393715372927545e-05,
          "train_loss": 7.406298391288146e-05,
          "validation_loss": 6.76278505125083e-05,
          "train_metric": 7.459352673322428e-06,
          "validation_metric": 6.811229468439706e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.362203511000189
        },
        {
          "epoch": 955,
          "updates": 15280,
          "samples_seen": 977920,
          "online_loss": 7.443520985361829e-05,
          "train_loss": 7.271544018294662e-05,
          "validation_loss": 6.419677083613351e-05,
          "train_metric": 7.323632416955661e-06,
          "validation_metric": 6.465663773269625e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.373102667999774
        },
        {
          "epoch": 956,
          "updates": 15296,
          "samples_seen": 978944,
          "online_loss": 7.367047419393202e-05,
          "train_loss": 7.63162097427994e-05,
          "validation_loss": 6.996074807830155e-05,
          "train_metric": 7.686289791308809e-06,
          "validation_metric": 7.046190603432478e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.383809376999125
        },
        {
          "epoch": 957,
          "updates": 15312,
          "samples_seen": 979968,
          "online_loss": 7.825466173017048e-05,
          "train_loss": 7.325434125959873e-05,
          "validation_loss": 6.64682374917902e-05,
          "train_metric": 7.377909241768066e-06,
          "validation_metric": 6.694438070553588e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.394112154999675
        },
        {
          "epoch": 958,
          "updates": 15328,
          "samples_seen": 980992,
          "online_loss": 7.487290326935181e-05,
          "train_loss": 7.475958409486338e-05,
          "validation_loss": 6.568591197719797e-05,
          "train_metric": 7.529511094617192e-06,
          "validation_metric": 6.615644451812841e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.404451159998644
        },
        {
          "epoch": 959,
          "updates": 15344,
          "samples_seen": 982016,
          "online_loss": 7.673389200135716e-05,
          "train_loss": 7.170491880970076e-05,
          "validation_loss": 6.478782597696409e-05,
          "train_metric": 7.221856321848463e-06,
          "validation_metric": 6.525192929984769e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.414677262999248
        },
        {
          "epoch": 960,
          "updates": 15360,
          "samples_seen": 983040,
          "online_loss": 7.288743722710933e-05,
          "train_loss": 7.275216921698302e-05,
          "validation_loss": 6.482894241344184e-05,
          "train_metric": 7.327332241402473e-06,
          "validation_metric": 6.52933340461459e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.425118903998737
        },
        {
          "epoch": 961,
          "updates": 15376,
          "samples_seen": 984064,
          "online_loss": 7.320410804823041e-05,
          "train_loss": 7.179746899055317e-05,
          "validation_loss": 6.565904914168641e-05,
          "train_metric": 7.231178642541636e-06,
          "validation_metric": 6.6129387050750665e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.43535032499858
        },
        {
          "epoch": 962,
          "updates": 15392,
          "samples_seen": 985088,
          "online_loss": 7.374557753792033e-05,
          "train_loss": 7.504324457840994e-05,
          "validation_loss": 6.499588198494166e-05,
          "train_metric": 7.558080142189283e-06,
          "validation_metric": 6.546148142660968e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.44573254900024
        },
        {
          "epoch": 963,
          "updates": 15408,
          "samples_seen": 986112,
          "online_loss": 7.461441805389768e-05,
          "train_loss": 7.197669037850574e-05,
          "validation_loss": 6.450626096921042e-05,
          "train_metric": 7.249229383887723e-06,
          "validation_metric": 6.496833975688787e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.456001525999454
        },
        {
          "epoch": 964,
          "updates": 15424,
          "samples_seen": 987136,
          "online_loss": 7.299625303858193e-05,
          "train_loss": 7.537579949712381e-05,
          "validation_loss": 6.607504474231973e-05,
          "train_metric": 7.5915741035714746e-06,
          "validation_metric": 6.654835942754289e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.466397967999
        },
        {
          "epoch": 965,
          "updates": 15440,
          "samples_seen": 988160,
          "online_loss": 7.609502335981233e-05,
          "train_loss": 7.153578189900145e-05,
          "validation_loss": 6.368795584421605e-05,
          "train_metric": 7.204821940831607e-06,
          "validation_metric": 6.414417839550879e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.477263486998709
        },
        {
          "epoch": 966,
          "updates": 15456,
          "samples_seen": 989184,
          "online_loss": 7.429022252836148e-05,
          "train_loss": 7.214024662971497e-05,
          "validation_loss": 6.442856101784855e-05,
          "train_metric": 7.265700787684182e-06,
          "validation_metric": 6.489009138022084e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.488071679999848
        },
        {
          "epoch": 967,
          "updates": 15472,
          "samples_seen": 990208,
          "online_loss": 7.635541805939283e-05,
          "train_loss": 7.178528903750703e-05,
          "validation_loss": 6.553016282850876e-05,
          "train_metric": 7.229950824694242e-06,
          "validation_metric": 6.599958851438714e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.49868032699851
        },
        {
          "epoch": 968,
          "updates": 15488,
          "samples_seen": 991232,
          "online_loss": 7.36730512471695e-05,
          "train_loss": 7.251853094203398e-05,
          "validation_loss": 6.431320070987567e-05,
          "train_metric": 7.3038004302361514e-06,
          "validation_metric": 6.477389433712233e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.509023007998621
        },
        {
          "epoch": 969,
          "updates": 15504,
          "samples_seen": 992256,
          "online_loss": 7.462928851964534e-05,
          "train_loss": 7.225203444249928e-05,
          "validation_loss": 6.488461076514795e-05,
          "train_metric": 7.2769607868394814e-06,
          "validation_metric": 6.534939984703669e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.519334378999702
        },
        {
          "epoch": 970,
          "updates": 15520,
          "samples_seen": 993280,
          "online_loss": 8.13144242783892e-05,
          "train_loss": 7.167836156440899e-05,
          "validation_loss": 6.457995914388448e-05,
          "train_metric": 7.219182862172602e-06,
          "validation_metric": 6.50425818093936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.529681506999623
        },
        {
          "epoch": 971,
          "updates": 15536,
          "samples_seen": 994304,
          "online_loss": 7.883156172283634e-05,
          "train_loss": 7.450897828675807e-05,
          "validation_loss": 6.510318053187802e-05,
          "train_metric": 7.504270797653589e-06,
          "validation_metric": 6.556954758707434e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.54021846499927
        },
        {
          "epoch": 972,
          "updates": 15552,
          "samples_seen": 995328,
          "online_loss": 7.52902001295297e-05,
          "train_loss": 7.39038732717745e-05,
          "validation_loss": 6.669804861303419e-05,
          "train_metric": 7.443327376677189e-06,
          "validation_metric": 6.717583346471656e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.550562385999001
        },
        {
          "epoch": 973,
          "updates": 15568,
          "samples_seen": 996352,
          "online_loss": 7.594141470690374e-05,
          "train_loss": 7.193755300249904e-05,
          "validation_loss": 6.445150938816369e-05,
          "train_metric": 7.245287179102888e-06,
          "validation_metric": 6.491320164059289e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.560912508999536
        },
        {
          "epoch": 974,
          "updates": 15584,
          "samples_seen": 997376,
          "online_loss": 7.966475641296711e-05,
          "train_loss": 8.153389353537932e-05,
          "validation_loss": 6.928161019459367e-05,
          "train_metric": 8.211794920498505e-06,
          "validation_metric": 6.977789780648891e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.571616551998886
        },
        {
          "epoch": 975,
          "updates": 15600,
          "samples_seen": 998400,
          "online_loss": 7.418691689053958e-05,
          "train_loss": 7.181548426160589e-05,
          "validation_loss": 6.356789526762441e-05,
          "train_metric": 7.232993084471673e-06,
          "validation_metric": 6.402325652743457e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.582562089999556
        },
        {
          "epoch": 976,
          "updates": 15616,
          "samples_seen": 999424,
          "online_loss": 7.352481816269574e-05,
          "train_loss": 7.214080687845126e-05,
          "validation_loss": 6.601227505598217e-05,
          "train_metric": 7.265758540597744e-06,
          "validation_metric": 6.648515409324318e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.593165184998725
        },
        {
          "epoch": 977,
          "updates": 15632,
          "samples_seen": 1000448,
          "online_loss": 7.347293353632267e-05,
          "train_loss": 7.161348185036331e-05,
          "validation_loss": 6.438978016376495e-05,
          "train_metric": 7.212648597487714e-06,
          "validation_metric": 6.485102403530618e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.603474654999445
        },
        {
          "epoch": 978,
          "updates": 15648,
          "samples_seen": 1001472,
          "online_loss": 7.391138819912157e-05,
          "train_loss": 7.095209730323404e-05,
          "validation_loss": 6.396170647349209e-05,
          "train_metric": 7.1460349317931104e-06,
          "validation_metric": 6.441988716687774e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.613776575999509
        },
        {
          "epoch": 979,
          "updates": 15664,
          "samples_seen": 1002496,
          "online_loss": 7.39097351925011e-05,
          "train_loss": 7.128747529350221e-05,
          "validation_loss": 6.340539403026924e-05,
          "train_metric": 7.179814019764308e-06,
          "validation_metric": 6.385959750332404e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.679174447999685
        },
        {
          "epoch": 980,
          "updates": 15680,
          "samples_seen": 1003520,
          "online_loss": 7.335517830142635e-05,
          "train_loss": 7.087569974828511e-05,
          "validation_loss": 6.359477993100882e-05,
          "train_metric": 7.138341061363462e-06,
          "validation_metric": 6.405032308975933e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.69032763299947
        },
        {
          "epoch": 981,
          "updates": 15696,
          "samples_seen": 1004544,
          "online_loss": 7.220797101581411e-05,
          "train_loss": 7.100271614035591e-05,
          "validation_loss": 6.382164428941905e-05,
          "train_metric": 7.151133104343899e-06,
          "validation_metric": 6.427882453863276e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.700715711000157
        },
        {
          "epoch": 982,
          "updates": 15712,
          "samples_seen": 1005568,
          "online_loss": 7.32128607978666e-05,
          "train_loss": 7.149855082388967e-05,
          "validation_loss": 6.327635492198169e-05,
          "train_metric": 7.201072548923548e-06,
          "validation_metric": 6.372962161549367e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.711613810999552
        },
        {
          "epoch": 983,
          "updates": 15728,
          "samples_seen": 1006592,
          "online_loss": 7.337524061767908e-05,
          "train_loss": 7.151491445256397e-05,
          "validation_loss": 6.395457603503019e-05,
          "train_metric": 7.202719643828459e-06,
          "validation_metric": 6.441269761126023e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.722051138000097
        },
        {
          "epoch": 984,
          "updates": 15744,
          "samples_seen": 1007616,
          "online_loss": 7.70060037211806e-05,
          "train_loss": 7.726701733190566e-05,
          "validation_loss": 6.55114054097794e-05,
          "train_metric": 7.78205048845848e-06,
          "validation_metric": 6.598068011953728e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.732536180999887
        },
        {
          "epoch": 985,
          "updates": 15760,
          "samples_seen": 1008640,
          "online_loss": 7.969387797857053e-05,
          "train_loss": 8.324725058628246e-05,
          "validation_loss": 6.839058914920315e-05,
          "train_metric": 8.384358807234094e-06,
          "validation_metric": 6.888049938424956e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.742835047998597
        },
        {
          "epoch": 986,
          "updates": 15776,
          "samples_seen": 1009664,
          "online_loss": 8.091998188319849e-05,
          "train_loss": 7.146580901462585e-05,
          "validation_loss": 6.48822751827538e-05,
          "train_metric": 7.19777472113492e-06,
          "validation_metric": 6.534704880323261e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.753243366998504
        },
        {
          "epoch": 987,
          "updates": 15792,
          "samples_seen": 1010688,
          "online_loss": 7.758594460938184e-05,
          "train_loss": 7.090099097695202e-05,
          "validation_loss": 6.357957317959517e-05,
          "train_metric": 7.140887191781076e-06,
          "validation_metric": 6.4035020841402e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.763592008999694
        },
        {
          "epoch": 988,
          "updates": 15808,
          "samples_seen": 1011712,
          "online_loss": 7.493296766369895e-05,
          "train_loss": 7.247464964166284e-05,
          "validation_loss": 6.384323933161795e-05,
          "train_metric": 7.299380740732886e-06,
          "validation_metric": 6.430056600947864e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.774098731999402
        },
        {
          "epoch": 989,
          "updates": 15824,
          "samples_seen": 1012736,
          "online_loss": 7.240479567371949e-05,
          "train_loss": 7.059981726342812e-05,
          "validation_loss": 6.2961844378151e-05,
          "train_metric": 7.110554179234896e-06,
          "validation_metric": 6.3412862800760195e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.78486048800005
        },
        {
          "epoch": 990,
          "updates": 15840,
          "samples_seen": 1013760,
          "online_loss": 7.441060074597772e-05,
          "train_loss": 7.084578101057559e-05,
          "validation_loss": 6.297542131505907e-05,
          "train_metric": 7.135327905416489e-06,
          "validation_metric": 6.342653250612784e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.795421106999129
        },
        {
          "epoch": 991,
          "updates": 15856,
          "samples_seen": 1014784,
          "online_loss": 7.242161177600792e-05,
          "train_loss": 7.622395060025156e-05,
          "validation_loss": 6.70918743708171e-05,
          "train_metric": 7.676998393435497e-06,
          "validation_metric": 6.757247319910675e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.806071805000101
        },
        {
          "epoch": 992,
          "updates": 15872,
          "samples_seen": 1015808,
          "online_loss": 7.179554586400627e-05,
          "train_loss": 8.600518049206585e-05,
          "validation_loss": 7.099456706782803e-05,
          "train_metric": 8.662126674607862e-06,
          "validation_metric": 7.150313649617601e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.816465793999669
        },
        {
          "epoch": 993,
          "updates": 15888,
          "samples_seen": 1016832,
          "online_loss": 7.707220606789633e-05,
          "train_loss": 7.050783460726961e-05,
          "validation_loss": 6.185167148942128e-05,
          "train_metric": 7.101290975697339e-06,
          "validation_metric": 6.229473001440056e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.827413924000211
        },
        {
          "epoch": 994,
          "updates": 15904,
          "samples_seen": 1017856,
          "online_loss": 7.444935226885718e-05,
          "train_loss": 7.071033178362995e-05,
          "validation_loss": 6.275850319070742e-05,
          "train_metric": 7.121685030142544e-06,
          "validation_metric": 6.320806278381497e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.838174038999568
        },
        {
          "epoch": 995,
          "updates": 15920,
          "samples_seen": 1018880,
          "online_loss": 7.294692272807879e-05,
          "train_loss": 7.138196087908e-05,
          "validation_loss": 6.34397947578691e-05,
          "train_metric": 7.189329153334256e-06,
          "validation_metric": 6.389424015651457e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.848458513999503
        },
        {
          "epoch": 996,
          "updates": 15936,
          "samples_seen": 1019904,
          "online_loss": 7.170927415245387e-05,
          "train_loss": 7.023498619673774e-05,
          "validation_loss": 6.242693052627146e-05,
          "train_metric": 7.073811502777971e-06,
          "validation_metric": 6.287411451921798e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.858875747999264
        },
        {
          "epoch": 997,
          "updates": 15952,
          "samples_seen": 1020928,
          "online_loss": 7.182576723607781e-05,
          "train_loss": 7.020396151347086e-05,
          "validation_loss": 6.248772842809558e-05,
          "train_metric": 7.070686024235329e-06,
          "validation_metric": 6.293535079748835e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.869435140999485
        },
        {
          "epoch": 998,
          "updates": 15968,
          "samples_seen": 1021952,
          "online_loss": 7.168858587647264e-05,
          "train_loss": 7.049647683743387e-05,
          "validation_loss": 6.266259151743725e-05,
          "train_metric": 7.100146831362508e-06,
          "validation_metric": 6.311146989901317e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.879757659999086
        },
        {
          "epoch": 999,
          "updates": 15984,
          "samples_seen": 1022976,
          "online_loss": 7.430301184285781e-05,
          "train_loss": 7.37609516363591e-05,
          "validation_loss": 6.50360670988448e-05,
          "train_metric": 7.428931894537527e-06,
          "validation_metric": 6.550194029841805e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.890067438000187
        },
        {
          "epoch": 1000,
          "updates": 16000,
          "samples_seen": 1024000,
          "online_loss": 7.283759873644158e-05,
          "train_loss": 7.365596684394404e-05,
          "validation_loss": 6.53347815386951e-05,
          "train_metric": 7.418358563882066e-06,
          "validation_metric": 6.5802796598291025e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.900375488999998
        },
        {
          "epoch": 1001,
          "updates": 16016,
          "samples_seen": 1025024,
          "online_loss": 7.188824270087935e-05,
          "train_loss": 7.031529821688309e-05,
          "validation_loss": 6.364216824294999e-05,
          "train_metric": 7.081899184413487e-06,
          "validation_metric": 6.4098048824234866e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.911060843998712
        },
        {
          "epoch": 1002,
          "updates": 16032,
          "samples_seen": 1026048,
          "online_loss": 7.23103833024652e-05,
          "train_loss": 7.030457345535979e-05,
          "validation_loss": 6.15759490756318e-05,
          "train_metric": 7.080820523697184e-06,
          "validation_metric": 6.201704309205525e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.921911698998883
        },
        {
          "epoch": 1003,
          "updates": 16048,
          "samples_seen": 1027072,
          "online_loss": 8.067673138612008e-05,
          "train_loss": 8.913515921449289e-05,
          "validation_loss": 7.217369420686737e-05,
          "train_metric": 8.97736845217878e-06,
          "validation_metric": 7.269070920301601e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.932310353999128
        },
        {
          "epoch": 1004,
          "updates": 16064,
          "samples_seen": 1028096,
          "online_loss": 7.146956295400742e-05,
          "train_loss": 7.075754547258839e-05,
          "validation_loss": 6.293033948168159e-05,
          "train_metric": 7.126440323190764e-06,
          "validation_metric": 6.338112598314183e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.942718042999331
        },
        {
          "epoch": 1005,
          "updates": 16080,
          "samples_seen": 1029120,
          "online_loss": 7.459605217263743e-05,
          "train_loss": 7.021127385087311e-05,
          "validation_loss": 6.253461469896138e-05,
          "train_metric": 7.071423169691116e-06,
          "validation_metric": 6.2982571762404405e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.952995997999096
        },
        {
          "epoch": 1006,
          "updates": 16096,
          "samples_seen": 1030144,
          "online_loss": 7.284181629074737e-05,
          "train_loss": 7.374577398877591e-05,
          "validation_loss": 6.609786942135543e-05,
          "train_metric": 7.427403943438549e-06,
          "validation_metric": 6.657135145360371e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.963319394999417
        },
        {
          "epoch": 1007,
          "updates": 16112,
          "samples_seen": 1031168,
          "online_loss": 7.221756186481798e-05,
          "train_loss": 6.972109258640558e-05,
          "validation_loss": 6.234452303033322e-05,
          "train_metric": 7.022052159300074e-06,
          "validation_metric": 6.2791114032734185e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.97374172599848
        },
        {
          "epoch": 1008,
          "updates": 16128,
          "samples_seen": 1032192,
          "online_loss": 7.23556047432794e-05,
          "train_loss": 7.644965808140114e-05,
          "validation_loss": 6.621515058213845e-05,
          "train_metric": 7.699729394516908e-06,
          "validation_metric": 6.668947207799647e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.984163572999023
        },
        {
          "epoch": 1009,
          "updates": 16144,
          "samples_seen": 1033216,
          "online_loss": 7.160960308283393e-05,
          "train_loss": 7.358190487138927e-05,
          "validation_loss": 6.465243495767936e-05,
          "train_metric": 7.410900707327528e-06,
          "validation_metric": 6.5115564211737365e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.994511005999811
        },
        {
          "epoch": 1010,
          "updates": 16160,
          "samples_seen": 1034240,
          "online_loss": 7.347858331741008e-05,
          "train_loss": 7.005979568930343e-05,
          "validation_loss": 6.30963477306068e-05,
          "train_metric": 7.056166396068875e-06,
          "validation_metric": 6.354833658406278e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.004885888998615
        },
        {
          "epoch": 1011,
          "updates": 16176,
          "samples_seen": 1035264,
          "online_loss": 7.155759931265493e-05,
          "train_loss": 7.131844904506579e-05,
          "validation_loss": 6.171075801830739e-05,
          "train_metric": 7.182932222349336e-06,
          "validation_metric": 6.215281246113591e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.015205465999315
        },
        {
          "epoch": 1012,
          "updates": 16192,
          "samples_seen": 1036288,
          "online_loss": 7.250188491525478e-05,
          "train_loss": 7.03338737366721e-05,
          "validation_loss": 6.27792687737383e-05,
          "train_metric": 7.083770469762385e-06,
          "validation_metric": 6.322898116195574e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.025583774999177
        },
        {
          "epoch": 1013,
          "updates": 16208,
          "samples_seen": 1037312,
          "online_loss": 7.133529925340554e-05,
          "train_loss": 7.162366819102317e-05,
          "validation_loss": 6.41307924524881e-05,
          "train_metric": 7.213674507511314e-06,
          "validation_metric": 6.459019459725823e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.03584844499892
        },
        {
          "epoch": 1014,
          "updates": 16224,
          "samples_seen": 1038336,
          "online_loss": 7.143635798456671e-05,
          "train_loss": 7.058730261633173e-05,
          "validation_loss": 6.372477218974382e-05,
          "train_metric": 7.109295438567642e-06,
          "validation_metric": 6.418125849450007e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.046257097999842
        },
        {
          "epoch": 1015,
          "updates": 16240,
          "samples_seen": 1039360,
          "online_loss": 7.800876096553111e-05,
          "train_loss": 8.391460141865537e-05,
          "validation_loss": 7.147832366172224e-05,
          "train_metric": 8.451573194179218e-06,
          "validation_metric": 7.199035280791577e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.111478840999553
        },
        {
          "epoch": 1016,
          "updates": 16256,
          "samples_seen": 1040384,
          "online_loss": 8.001542050806165e-05,
          "train_loss": 7.197377271950245e-05,
          "validation_loss": 6.426309118978679e-05,
          "train_metric": 7.248934252856998e-06,
          "validation_metric": 6.472342647612095e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.122292168000058
        },
        {
          "epoch": 1017,
          "updates": 16272,
          "samples_seen": 1041408,
          "online_loss": 7.280401041498408e-05,
          "train_loss": 7.039548654574901e-05,
          "validation_loss": 6.28041525487788e-05,
          "train_metric": 7.08997504261788e-06,
          "validation_metric": 6.325403774098959e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.132671702998778
        },
        {
          "epoch": 1018,
          "updates": 16288,
          "samples_seen": 1042432,
          "online_loss": 7.483497211069334e-05,
          "train_loss": 7.013286813162267e-05,
          "validation_loss": 6.13619340583682e-05,
          "train_metric": 7.0635251177009195e-06,
          "validation_metric": 6.180149284773506e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.143651570999282
        },
        {
          "epoch": 1019,
          "updates": 16304,
          "samples_seen": 1043456,
          "online_loss": 7.410644502670038e-05,
          "train_loss": 7.039764750516042e-05,
          "validation_loss": 6.133202987257391e-05,
          "train_metric": 7.0901928665989544e-06,
          "validation_metric": 6.177137947815936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.154587221999464
        },
        {
          "epoch": 1020,
          "updates": 16320,
          "samples_seen": 1044480,
          "online_loss": 7.081417015797342e-05,
          "train_loss": 6.974149437155575e-05,
          "validation_loss": 6.246801785891876e-05,
          "train_metric": 7.024108072073432e-06,
          "validation_metric": 6.291549652814865e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.165088906000165
        },
        {
          "epoch": 1021,
          "updates": 16336,
          "samples_seen": 1045504,
          "online_loss": 7.076422343743616e-05,
          "train_loss": 7.365406054304913e-05,
          "validation_loss": 6.468813808169216e-05,
          "train_metric": 7.418167569994694e-06,
          "validation_metric": 6.515152563224547e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.175590542999998
        },
        {
          "epoch": 1022,
          "updates": 16352,
          "samples_seen": 1046528,
          "online_loss": 7.230990649986779e-05,
          "train_loss": 6.962850602576509e-05,
          "validation_loss": 6.252997263800353e-05,
          "train_metric": 7.012727110122796e-06,
          "validation_metric": 6.297789695963729e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.185923757999262
        },
        {
          "epoch": 1023,
          "updates": 16368,
          "samples_seen": 1047552,
          "online_loss": 7.109917078196304e-05,
          "train_loss": 6.963130726944655e-05,
          "validation_loss": 6.159037002362311e-05,
          "train_metric": 7.0130099629750475e-06,
          "validation_metric": 6.203156772244256e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.196283893999862
        },
        {
          "epoch": 1024,
          "updates": 16384,
          "samples_seen": 1048576,
          "online_loss": 7.295528030226706e-05,
          "train_loss": 7.817296864232048e-05,
          "validation_loss": 6.645455141551793e-05,
          "train_metric": 7.87329554441385e-06,
          "validation_metric": 6.693058821838349e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.206583239998508
        },
        {
          "epoch": 1025,
          "updates": 16400,
          "samples_seen": 1049600,
          "online_loss": 7.379770886473125e-05,
          "train_loss": 6.910171214258298e-05,
          "validation_loss": 6.0935224610148e-05,
          "train_metric": 6.9596712819475215e-06,
          "validation_metric": 6.137173386377981e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.217414012999143
        },
        {
          "epoch": 1026,
          "updates": 16416,
          "samples_seen": 1050624,
          "online_loss": 7.152624402806396e-05,
          "train_loss": 7.683220610488206e-05,
          "validation_loss": 6.591975397896022e-05,
          "train_metric": 7.738258318568114e-06,
          "validation_metric": 6.639195817115251e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.227899612998954
        },
        {
          "epoch": 1027,
          "updates": 16432,
          "samples_seen": 1051648,
          "online_loss": 7.058426012918062e-05,
          "train_loss": 7.898705371189862e-05,
          "validation_loss": 7.070450374158099e-05,
          "train_metric": 7.95528649177868e-06,
          "validation_metric": 7.121098860807251e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.238243429999784
        },
        {
          "epoch": 1028,
          "updates": 16448,
          "samples_seen": 1052672,
          "online_loss": 7.279745682353678e-05,
          "train_loss": 6.926633068360388e-05,
          "validation_loss": 6.126326479716226e-05,
          "train_metric": 6.97625046086614e-06,
          "validation_metric": 6.1702107814198826e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.248604315998818
        },
        {
          "epoch": 1029,
          "updates": 16464,
          "samples_seen": 1053696,
          "online_loss": 7.1428092496717e-05,
          "train_loss": 7.325530896196142e-05,
          "validation_loss": 6.309845775831491e-05,
          "train_metric": 7.3780065577011555e-06,
          "validation_metric": 6.355045570671791e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.258947722999437
        },
        {
          "epoch": 1030,
          "updates": 16480,
          "samples_seen": 1054720,
          "online_loss": 8.309315876431356e-05,
          "train_loss": 7.380198076134548e-05,
          "validation_loss": 6.47261185804382e-05,
          "train_metric": 7.433065547957085e-06,
          "validation_metric": 6.5189778979402035e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.269260684999608
        },
        {
          "epoch": 1031,
          "updates": 16496,
          "samples_seen": 1055744,
          "online_loss": 7.560516951343743e-05,
          "train_loss": 7.284455932676792e-05,
          "validation_loss": 6.368773028953001e-05,
          "train_metric": 7.336637281696312e-06,
          "validation_metric": 6.414394647435984e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.279542360998676
        },
        {
          "epoch": 1032,
          "updates": 16512,
          "samples_seen": 1056768,
          "online_loss": 7.34767295398342e-05,
          "train_loss": 7.140666275518015e-05,
          "validation_loss": 6.342009146464989e-05,
          "train_metric": 7.191817985585658e-06,
          "validation_metric": 6.3874394982121885e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.289870142998552
        },
        {
          "epoch": 1033,
          "updates": 16528,
          "samples_seen": 1057792,
          "online_loss": 8.245347203228448e-05,
          "train_loss": 7.552628085250035e-05,
          "validation_loss": 6.568444223375991e-05,
          "train_metric": 7.606729923281819e-06,
          "validation_metric": 6.615496658923803e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.30014596899855
        },
        {
          "epoch": 1034,
          "updates": 16544,
          "samples_seen": 1058816,
          "online_loss": 8.040010402510234e-05,
          "train_loss": 7.976297638379037e-05,
          "validation_loss": 6.376943201757967e-05,
          "train_metric": 8.033434824028518e-06,
          "validation_metric": 6.422624664992327e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.310461410999778
        },
        {
          "epoch": 1035,
          "updates": 16560,
          "samples_seen": 1059840,
          "online_loss": 7.813853494553769e-05,
          "train_loss": 7.214368088170886e-05,
          "validation_loss": 6.233965541468933e-05,
          "train_metric": 7.266046850418206e-06,
          "validation_metric": 6.278621185629163e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.320703788998799
        },
        {
          "epoch": 1036,
          "updates": 16576,
          "samples_seen": 1060864,
          "online_loss": 7.982113925208978e-05,
          "train_loss": 7.709299825364724e-05,
          "validation_loss": 6.586438394151628e-05,
          "train_metric": 7.764524525555316e-06,
          "validation_metric": 6.633619705098681e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.331064977999631
        },
        {
          "epoch": 1037,
          "updates": 16592,
          "samples_seen": 1061888,
          "online_loss": 7.635314113940694e-05,
          "train_loss": 6.929959636181593e-05,
          "validation_loss": 6.192870205268264e-05,
          "train_metric": 6.9796014940948226e-06,
          "validation_metric": 6.2372319007408805e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.341267414998583
        },
        {
          "epoch": 1038,
          "updates": 16608,
          "samples_seen": 1062912,
          "online_loss": 7.018490146037948e-05,
          "train_loss": 7.464231748599559e-05,
          "validation_loss": 6.652812589891255e-05,
          "train_metric": 7.5177003964199685e-06,
          "validation_metric": 6.700469839415746e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.351606668999011
        },
        {
          "epoch": 1039,
          "updates": 16624,
          "samples_seen": 1063936,
          "online_loss": 7.335736518143676e-05,
          "train_loss": 8.994032396003604e-05,
          "validation_loss": 7.586210267618299e-05,
          "train_metric": 9.058459909283556e-06,
          "validation_metric": 7.640554031240754e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.3619475029991
        },
        {
          "epoch": 1040,
          "updates": 16640,
          "samples_seen": 1064960,
          "online_loss": 7.596357613692817e-05,
          "train_loss": 6.900925654917955e-05,
          "validation_loss": 5.936159868724644e-05,
          "train_metric": 6.95035987519077e-06,
          "validation_metric": 5.978683020657627e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.372884528000213
        },
        {
          "epoch": 1041,
          "updates": 16656,
          "samples_seen": 1065984,
          "online_loss": 7.180265674833208e-05,
          "train_loss": 6.849643978057429e-05,
          "validation_loss": 6.150660192361102e-05,
          "train_metric": 6.8987105805717874e-06,
          "validation_metric": 6.19472029939061e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.383408302999669
        },
        {
          "epoch": 1042,
          "updates": 16672,
          "samples_seen": 1067008,
          "online_loss": 6.93281174335425e-05,
          "train_loss": 7.224049477372319e-05,
          "validation_loss": 6.275354098761454e-05,
          "train_metric": 7.275798452610616e-06,
          "validation_metric": 6.320306965790223e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.39390274699872
        },
        {
          "epoch": 1043,
          "updates": 16688,
          "samples_seen": 1068032,
          "online_loss": 7.386400216091715e-05,
          "train_loss": 6.964182830415666e-05,
          "validation_loss": 6.11419600318186e-05,
          "train_metric": 7.014070433797315e-06,
          "validation_metric": 6.157993993838318e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.404206538998551
        },
        {
          "epoch": 1044,
          "updates": 16704,
          "samples_seen": 1069056,
          "online_loss": 9.215575641974283e-05,
          "train_loss": 8.262021583504975e-05,
          "validation_loss": 7.265160093083978e-05,
          "train_metric": 8.321205314132385e-06,
          "validation_metric": 7.317202744161477e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.414577976000146
        },
        {
          "epoch": 1045,
          "updates": 16720,
          "samples_seen": 1070080,
          "online_loss": 7.721793213022465e-05,
          "train_loss": 6.93499023327604e-05,
          "validation_loss": 6.060691885068081e-05,
          "train_metric": 6.984667379583698e-06,
          "validation_metric": 6.104106432758272e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.424911115000214
        },
        {
          "epoch": 1046,
          "updates": 16736,
          "samples_seen": 1071104,
          "online_loss": 7.665487396479875e-05,
          "train_loss": 7.251117494888604e-05,
          "validation_loss": 6.390197086147964e-05,
          "train_metric": 7.303060556296259e-06,
          "validation_metric": 6.435972409235546e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.435271829999692
        },
        {
          "epoch": 1047,
          "updates": 16752,
          "samples_seen": 1072128,
          "online_loss": 7.149889711399737e-05,
          "train_loss": 7.14559355401434e-05,
          "validation_loss": 6.198233313625678e-05,
          "train_metric": 7.196780643425882e-06,
          "validation_metric": 6.242633844522061e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.445528794998609
        },
        {
          "epoch": 1048,
          "updates": 16768,
          "samples_seen": 1073152,
          "online_loss": 7.236038550217927e-05,
          "train_loss": 7.068685226840898e-05,
          "validation_loss": 6.456564005929977e-05,
          "train_metric": 7.119320343917934e-06,
          "validation_metric": 6.502815267594997e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.455876322999757
        },
        {
          "epoch": 1049,
          "updates": 16784,
          "samples_seen": 1074176,
          "online_loss": 7.969520925144025e-05,
          "train_loss": 7.106882549123839e-05,
          "validation_loss": 6.255508924368769e-05,
          "train_metric": 7.157791515055578e-06,
          "validation_metric": 6.30031900072936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.466133324000111
        },
        {
          "epoch": 1050,
          "updates": 16800,
          "samples_seen": 1075200,
          "online_loss": 7.063134262352833e-05,
          "train_loss": 6.9908652221784e-05,
          "validation_loss": 6.049290095688775e-05,
          "train_metric": 7.040942819003249e-06,
          "validation_metric": 6.092623152653687e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.476434872998652
        },
        {
          "epoch": 1051,
          "updates": 16816,
          "samples_seen": 1076224,
          "online_loss": 7.940029308883823e-05,
          "train_loss": 8.24017115519382e-05,
          "validation_loss": 7.198166713351384e-05,
          "train_metric": 8.299199180328287e-06,
          "validation_metric": 7.249729605973698e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.542855306999627
        },
        {
          "epoch": 1052,
          "updates": 16832,
          "samples_seen": 1077248,
          "online_loss": 7.93176909610338e-05,
          "train_loss": 7.009041291894391e-05,
          "validation_loss": 6.302654946921393e-05,
          "train_metric": 7.059250037855236e-06,
          "validation_metric": 6.3478028096142225e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.553844152998863
        },
        {
          "epoch": 1053,
          "updates": 16848,
          "samples_seen": 1078272,
          "online_loss": 7.303495658561587e-05,
          "train_loss": 7.498742343159392e-05,
          "validation_loss": 6.693729665130377e-05,
          "train_metric": 7.552459010184975e-06,
          "validation_metric": 6.741678589605726e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.564156433998505
        },
        {
          "epoch": 1054,
          "updates": 16864,
          "samples_seen": 1079296,
          "online_loss": 7.469752654287731e-05,
          "train_loss": 6.828799814684317e-05,
          "validation_loss": 5.950841659796424e-05,
          "train_metric": 6.877716714370763e-06,
          "validation_metric": 5.993469585519051e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.574630334998801
        },
        {
          "epoch": 1055,
          "updates": 16880,
          "samples_seen": 1080320,
          "online_loss": 7.138480464163877e-05,
          "train_loss": 7.278681732714176e-05,
          "validation_loss": 6.23450760031119e-05,
          "train_metric": 7.330822427320527e-06,
          "validation_metric": 6.27916870143963e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.58491639299973
        },
        {
          "epoch": 1056,
          "updates": 16896,
          "samples_seen": 1081344,
          "online_loss": 7.250770727296185e-05,
          "train_loss": 6.801167182857171e-05,
          "validation_loss": 5.933843567618169e-05,
          "train_metric": 6.849886176496511e-06,
          "validation_metric": 5.976350166747579e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.59596538599908
        },
        {
          "epoch": 1057,
          "updates": 16912,
          "samples_seen": 1082368,
          "online_loss": 7.660177175239369e-05,
          "train_loss": 8.028004231164232e-05,
          "validation_loss": 6.934757402632385e-05,
          "train_metric": 8.085512490652036e-06,
          "validation_metric": 6.984433639445342e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.606323624999277
        },
        {
          "epoch": 1058,
          "updates": 16928,
          "samples_seen": 1083392,
          "online_loss": 7.495461113649071e-05,
          "train_loss": 6.981527985772118e-05,
          "validation_loss": 6.220853538252413e-05,
          "train_metric": 7.031540008028969e-06,
          "validation_metric": 6.26541532255942e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.616732421000052
        },
        {
          "epoch": 1059,
          "updates": 16944,
          "samples_seen": 1084416,
          "online_loss": 8.153875046446046e-05,
          "train_loss": 6.98268340784125e-05,
          "validation_loss": 6.019443753757514e-05,
          "train_metric": 7.032703251752537e-06,
          "validation_metric": 6.062562988518039e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.627152639999622
        },
        {
          "epoch": 1060,
          "updates": 16960,
          "samples_seen": 1085440,
          "online_loss": 8.105675306069315e-05,
          "train_loss": 7.591024041175842e-05,
          "validation_loss": 6.524596392409876e-05,
          "train_metric": 7.645400728506502e-06,
          "validation_metric": 6.571334779437166e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.637514407999333
        },
        {
          "epoch": 1061,
          "updates": 16976,
          "samples_seen": 1086464,
          "online_loss": 7.44382143693656e-05,
          "train_loss": 6.882743036840111e-05,
          "validation_loss": 6.0756523453164846e-05,
          "train_metric": 6.9320462898758706e-06,
          "validation_metric": 6.119174031482544e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.647808925999925
        },
        {
          "epoch": 1062,
          "updates": 16992,
          "samples_seen": 1087488,
          "online_loss": 7.149089924496366e-05,
          "train_loss": 7.03268960933201e-05,
          "validation_loss": 6.20207647443749e-05,
          "train_metric": 7.083067885105265e-06,
          "validation_metric": 6.246503289730754e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.658129732999441
        },
        {
          "epoch": 1063,
          "updates": 17008,
          "samples_seen": 1088512,
          "online_loss": 7.879304507696361e-05,
          "train_loss": 7.262374128913507e-05,
          "validation_loss": 6.275795021792874e-05,
          "train_metric": 7.314396953006508e-06,
          "validation_metric": 6.320750344457338e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.66861242899904
        },
        {
          "epoch": 1064,
          "updates": 17024,
          "samples_seen": 1089536,
          "online_loss": 7.263579277605459e-05,
          "train_loss": 6.774722714908421e-05,
          "validation_loss": 5.9399721067165956e-05,
          "train_metric": 6.823253443144495e-06,
          "validation_metric": 5.9825219977938104e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.678955462999511
        },
        {
          "epoch": 1065,
          "updates": 17040,
          "samples_seen": 1090560,
          "online_loss": 7.082455135787313e-05,
          "train_loss": 7.972081220941618e-05,
          "validation_loss": 6.618504994548857e-05,
          "train_metric": 8.02918839326594e-06,
          "validation_metric": 6.665914952463936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.689511269998548
        },
        {
          "epoch": 1066,
          "updates": 17056,
          "samples_seen": 1091584,
          "online_loss": 8.093991391433519e-05,
          "train_loss": 8.107673784252256e-05,
          "validation_loss": 6.401718565030023e-05,
          "train_metric": 8.165752660715953e-06,
          "validation_metric": 6.447577106882818e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.699939614998584
        },
        {
          "epoch": 1067,
          "updates": 17072,
          "samples_seen": 1092608,
          "online_loss": 7.141986543501844e-05,
          "train_loss": 6.824202137067914e-05,
          "validation_loss": 6.025601032888517e-05,
          "train_metric": 6.873086022096686e-06,
          "validation_metric": 6.068764832889428e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.71026573800009
        },
        {
          "epoch": 1068,
          "updates": 17088,
          "samples_seen": 1093632,
          "online_loss": 6.964569001866039e-05,
          "train_loss": 6.84926490066573e-05,
          "validation_loss": 6.070154995541088e-05,
          "train_metric": 6.8983299570390955e-06,
          "validation_metric": 6.113638846727554e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.720553174998713
        },
        {
          "epoch": 1069,
          "updates": 17104,
          "samples_seen": 1094656,
          "online_loss": 7.041955905151553e-05,
          "train_loss": 6.872820813441649e-05,
          "validation_loss": 5.9834208514075726e-05,
          "train_metric": 6.922053671587491e-06,
          "validation_metric": 6.026282790116966e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.730895534999945
        },
        {
          "epoch": 1070,
          "updates": 17120,
          "samples_seen": 1095680,
          "online_loss": 7.201937273748626e-05,
          "train_loss": 6.819014379289001e-05,
          "validation_loss": 6.036007107468322e-05,
          "train_metric": 6.867861884529702e-06,
          "validation_metric": 6.0792463045800105e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.741255787999762
        },
        {
          "epoch": 1071,
          "updates": 17136,
          "samples_seen": 1096704,
          "online_loss": 7.12157025191118e-05,
          "train_loss": 7.294875831576064e-05,
          "validation_loss": 6.235140608623624e-05,
          "train_metric": 7.347132395807421e-06,
          "validation_metric": 6.27980489298352e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.751595648998773
        },
        {
          "epoch": 1072,
          "updates": 17152,
          "samples_seen": 1097728,
          "online_loss": 6.983072171351523e-05,
          "train_loss": 6.874306563986465e-05,
          "validation_loss": 5.961714487057179e-05,
          "train_metric": 6.923548880877206e-06,
          "validation_metric": 6.004420356475748e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.762125272998674
        },
        {
          "epoch": 1073,
          "updates": 17168,
          "samples_seen": 1098752,
          "online_loss": 7.014102925495536e-05,
          "train_loss": 6.789717735955492e-05,
          "validation_loss": 6.079359809518792e-05,
          "train_metric": 6.8383542384253815e-06,
          "validation_metric": 6.122908871475374e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.772466088999863
        },
        {
          "epoch": 1074,
          "updates": 17184,
          "samples_seen": 1099776,
          "online_loss": 7.107835654096561e-05,
          "train_loss": 7.06521313986741e-05,
          "validation_loss": 6.073774420656264e-05,
          "train_metric": 7.115823336789617e-06,
          "validation_metric": 6.117283191997558e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.782899238000027
        },
        {
          "epoch": 1075,
          "updates": 17200,
          "samples_seen": 1100800,
          "online_loss": 6.989334383433743e-05,
          "train_loss": 6.831644714111462e-05,
          "validation_loss": 5.955057349638082e-05,
          "train_metric": 6.8805825321760494e-06,
          "validation_metric": 5.997715561534278e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.793181224998989
        },
        {
          "epoch": 1076,
          "updates": 17216,
          "samples_seen": 1101824,
          "online_loss": 6.956168863325729e-05,
          "train_loss": 6.782672426197678e-05,
          "validation_loss": 6.08654081588611e-05,
          "train_metric": 6.831259270256851e-06,
          "validation_metric": 6.130140263849171e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.803645019999749
        }
      ],
      "best_validation_metric": 5.976350166747579e-06,
      "config": {
        "label": "LayerNorm + silu + plateau LR",
        "batch": 64,
        "lr": 0.003,
        "schedule": "plateau",
        "normalization": "layer",
        "residual": false,
        "concatenate": false,
        "dropout": 0.0,
        "activation": "silu",
        "loss": "mse",
        "outliers": false,
        "max_updates": null
      },
      "seed": 22,
      "parameters": 2369,
      "train_samples": 1024,
      "validation_rmse_m_s": 0.002444657474319783,
      "training_profiles": [
        0,
        1,
        3,
        4,
        6,
        7,
        10,
        11
      ],
      "stopped_early": true
    }
  ],
  "seconds": 15.202893150000818,
  "source_sha256": {
    "training_tools.py": "73f9c91df0bdc2d0681a2025d3d554d11d289e1fbfed038047dec4277c3e6236",
    "training_strategies_experiments.py": "d45307e44b19857ad48314ab8933ed928537b17c540020af6681d6dcc062c9bd",
    "training_data_scaling.py": "7340fdf0beda376f24e6a3de96b8f7e0b6e19a55bf72b59d0a4201d570aefde1"
  }
}
