{
  "groups": {
    "final": [
      {
        "label": "LayerNorm + SiLU + plateau LR",
        "id": "14985df6b90f"
      }
    ]
  },
  "runs": {
    "14985df6b90f_33": {
      "history": [
        {
          "epoch": 1,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.23888491606339812,
          "train_loss": 0.08166317641735077,
          "validation_loss": 0.06555181741714478,
          "train_metric": 0.00822481606155634,
          "validation_metric": 0.0066021401435136795,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.013900837000619504
        },
        {
          "epoch": 2,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.061784587102010846,
          "train_loss": 0.054036371409893036,
          "validation_loss": 0.045189712196588516,
          "train_metric": 0.005442344583570957,
          "validation_metric": 0.004551341757178307,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.025444443001106265
        },
        {
          "epoch": 3,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.04121296224184334,
          "train_loss": 0.03482135012745857,
          "validation_loss": 0.027817262336611748,
          "train_metric": 0.003507079090923071,
          "validation_metric": 0.0028016529977321625,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03746818600120605
        },
        {
          "epoch": 4,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.029260389623232186,
          "train_loss": 0.02433099038898945,
          "validation_loss": 0.01896427944302559,
          "train_metric": 0.002450528321787715,
          "validation_metric": 0.001910012448206544,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.04976458700002695
        },
        {
          "epoch": 5,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.024199075298383832,
          "train_loss": 0.02122872695326805,
          "validation_loss": 0.021520046517252922,
          "train_metric": 0.0021380793768912554,
          "validation_metric": 0.0021674202289432287,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.060992492000877974
        },
        {
          "epoch": 6,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.021335106168407947,
          "train_loss": 0.017451364547014236,
          "validation_loss": 0.01547304354608059,
          "train_metric": 0.0017576378304511309,
          "validation_metric": 0.0015583883505314589,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07279244300116261
        },
        {
          "epoch": 7,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.02976817957824096,
          "train_loss": 0.03284674137830734,
          "validation_loss": 0.034943778067827225,
          "train_metric": 0.0033082030713558197,
          "validation_metric": 0.0035194093361496925,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08426426799996989
        },
        {
          "epoch": 8,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.02878994884667918,
          "train_loss": 0.02708299085497856,
          "validation_loss": 0.019958991557359695,
          "train_metric": 0.002727699698880315,
          "validation_metric": 0.002010196214541793,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09517702599987388
        },
        {
          "epoch": 9,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.016897937923204154,
          "train_loss": 0.01903846114873886,
          "validation_loss": 0.01356436312198639,
          "train_metric": 0.0019174840999767184,
          "validation_metric": 0.0013661530101671815,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.16749019800045062
        },
        {
          "epoch": 10,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.014107360038906336,
          "train_loss": 0.011092609725892544,
          "validation_loss": 0.009211700409650803,
          "train_metric": 0.001117207109928131,
          "validation_metric": 0.0009277687640860677,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.17898638900078367
        },
        {
          "epoch": 11,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.011751949205063283,
          "train_loss": 0.008849341422319412,
          "validation_loss": 0.008775648660957813,
          "train_metric": 0.0008912733173929155,
          "validation_metric": 0.0008838512003421783,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.19018887300080678
        },
        {
          "epoch": 12,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.010980227962136269,
          "train_loss": 0.025530699640512466,
          "validation_loss": 0.02131485566496849,
          "train_metric": 0.0025713585782796144,
          "validation_metric": 0.0021467541810125113,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20112885600065056
        },
        {
          "epoch": 13,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.02549975086003542,
          "train_loss": 0.01695833168923855,
          "validation_loss": 0.01574423722922802,
          "train_metric": 0.0017079809913411736,
          "validation_metric": 0.0015857020625844598,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21240808200127503
        },
        {
          "epoch": 14,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.019903491061995737,
          "train_loss": 0.013715945184230804,
          "validation_loss": 0.014348994940519333,
          "train_metric": 0.0013814197154715657,
          "validation_metric": 0.0014451781753450632,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2231803470003797
        },
        {
          "epoch": 15,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.013678477756911889,
          "train_loss": 0.019475849345326424,
          "validation_loss": 0.013384152203798294,
          "train_metric": 0.0019615362398326397,
          "validation_metric": 0.001348002697341144,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.23413419400094426
        },
        {
          "epoch": 16,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.011750767356716096,
          "train_loss": 0.011588453315198421,
          "validation_loss": 0.008362028747797012,
          "train_metric": 0.0011671464890241623,
          "validation_metric": 0.0008421927923336625,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2456109180002386
        },
        {
          "epoch": 17,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.00787850258348044,
          "train_loss": 0.007302626967430115,
          "validation_loss": 0.0058487593196332455,
          "train_metric": 0.0007354938425123692,
          "validation_metric": 0.000589065661188215,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25702081600138627
        },
        {
          "epoch": 18,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.00889447663212195,
          "train_loss": 0.006209243554621935,
          "validation_loss": 0.004950473550707102,
          "train_metric": 0.0006253722822293639,
          "validation_metric": 0.0004985934938304126,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.26890690100117354
        },
        {
          "epoch": 19,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.011325097439112142,
          "train_loss": 0.0055108158849179745,
          "validation_loss": 0.004341572523117065,
          "train_metric": 0.0005550291971303523,
          "validation_metric": 0.00043726732837967575,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.280951841001297
        },
        {
          "epoch": 20,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.007683465169975534,
          "train_loss": 0.004882046952843666,
          "validation_loss": 0.004007525276392698,
          "train_metric": 0.0004917018231935799,
          "validation_metric": 0.0004036233003716916,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2928252150013577
        },
        {
          "epoch": 21,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.008050136239035055,
          "train_loss": 0.005788047332316637,
          "validation_loss": 0.004216156434267759,
          "train_metric": 0.0005829509464092553,
          "validation_metric": 0.0004246358585078269,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30420762399990053
        },
        {
          "epoch": 22,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.01719522225903347,
          "train_loss": 0.023246873170137405,
          "validation_loss": 0.01762869581580162,
          "train_metric": 0.0023413393646478653,
          "validation_metric": 0.0017754975706338882,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3153400340015651
        },
        {
          "epoch": 23,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.02437688660575077,
          "train_loss": 0.009796607308089733,
          "validation_loss": 0.009899741038680077,
          "train_metric": 0.0009866784093901515,
          "validation_metric": 0.0009970655664801598,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3260839770009625
        },
        {
          "epoch": 24,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.01053882839914877,
          "train_loss": 0.006641744635999203,
          "validation_loss": 0.005747511517256498,
          "train_metric": 0.00066893215989694,
          "validation_metric": 0.0005788682610727847,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.33704265500091424
        },
        {
          "epoch": 25,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.010213554996880703,
          "train_loss": 0.007118855603039265,
          "validation_loss": 0.007156596519052982,
          "train_metric": 0.0007169851451180875,
          "validation_metric": 0.0007207861635833979,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.34799529400152096
        },
        {
          "epoch": 26,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.008563331473851576,
          "train_loss": 0.005285867024213076,
          "validation_loss": 0.004842318128794432,
          "train_metric": 0.0005323732038959861,
          "validation_metric": 0.0004877005412708968,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3585524840000289
        },
        {
          "epoch": 27,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.012568902275233995,
          "train_loss": 0.022204900160431862,
          "validation_loss": 0.018985990434885025,
          "train_metric": 0.0022363958414644003,
          "validation_metric": 0.0019121996592730284,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.36988097700123035
        },
        {
          "epoch": 28,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.01679351064376533,
          "train_loss": 0.006613657809793949,
          "validation_loss": 0.005368424113839865,
          "train_metric": 0.0006661033257842064,
          "validation_metric": 0.0005406880518421531,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.38084648299991386
        },
        {
          "epoch": 29,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.007585612416733056,
          "train_loss": 0.007600780110806227,
          "validation_loss": 0.007061448413878679,
          "train_metric": 0.0007655228255316615,
          "validation_metric": 0.000711203261744231,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3915948780013423
        },
        {
          "epoch": 30,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.013993913438753225,
          "train_loss": 0.01605561003088951,
          "validation_loss": 0.010954132303595543,
          "train_metric": 0.001617062371224165,
          "validation_metric": 0.001103260088711977,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.40269336700112035
        },
        {
          "epoch": 31,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.008528002261300571,
          "train_loss": 0.006476330105215311,
          "validation_loss": 0.006094194017350674,
          "train_metric": 0.0006522722542285919,
          "validation_metric": 0.00061378488317132,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41379783900083567
        },
        {
          "epoch": 32,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.008905707880330738,
          "train_loss": 0.010734885931015015,
          "validation_loss": 0.008449658751487732,
          "train_metric": 0.0010811785468831658,
          "validation_metric": 0.0008510186453349888,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.42443362700032594
        },
        {
          "epoch": 33,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.007577283075079322,
          "train_loss": 0.003922881092876196,
          "validation_loss": 0.0034323823638260365,
          "train_metric": 0.00039509820635430515,
          "validation_metric": 0.0003456970152910799,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.43595894900136045
        },
        {
          "epoch": 34,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.00360892678872915,
          "train_loss": 0.0032230280339717865,
          "validation_loss": 0.0027567893266677856,
          "train_metric": 0.0003246115520596504,
          "validation_metric": 0.0002776537439785898,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4479956370014406
        },
        {
          "epoch": 35,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.003924632663256489,
          "train_loss": 0.003139584558084607,
          "validation_loss": 0.002995756920427084,
          "train_metric": 0.00031620750087313354,
          "validation_metric": 0.00030172168044373393,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4596729900003993
        },
        {
          "epoch": 36,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.004978348712029401,
          "train_loss": 0.00343205314129591,
          "validation_loss": 0.0031588103156536818,
          "train_metric": 0.00034566380782052875,
          "validation_metric": 0.00031814377871342003,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.47072933100025693
        },
        {
          "epoch": 37,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.003140195731248241,
          "train_loss": 0.0024505648761987686,
          "validation_loss": 0.00236724060960114,
          "train_metric": 0.00024681194918230176,
          "validation_metric": 0.00023841980146244168,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4823110800007271
        },
        {
          "epoch": 38,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.002831874182447791,
          "train_loss": 0.0023875823244452477,
          "validation_loss": 0.0018939138390123844,
          "train_metric": 0.0002404685947112739,
          "validation_metric": 0.00019074806186836213,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4937046900013229
        },
        {
          "epoch": 39,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.006535253283800557,
          "train_loss": 0.028644394129514694,
          "validation_loss": 0.018808677792549133,
          "train_metric": 0.0028849586378782988,
          "validation_metric": 0.0018943409668281674,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5049098350009444
        },
        {
          "epoch": 40,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.016351179765479174,
          "train_loss": 0.01290273480117321,
          "validation_loss": 0.012090597301721573,
          "train_metric": 0.001299516181461513,
          "validation_metric": 0.0012177207972854376,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5157842560001882
        },
        {
          "epoch": 41,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.008869607874657959,
          "train_loss": 0.007582868915051222,
          "validation_loss": 0.0061017367988824844,
          "train_metric": 0.0007637187372893095,
          "validation_metric": 0.0006145446677692235,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5264339340010338
        },
        {
          "epoch": 42,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.0058883669480565,
          "train_loss": 0.005375618115067482,
          "validation_loss": 0.0037933329585939646,
          "train_metric": 0.000541412562597543,
          "validation_metric": 0.00038205055170692503,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5372545509999327
        },
        {
          "epoch": 43,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.0042082878717337735,
          "train_loss": 0.0029756950680166483,
          "validation_loss": 0.0022573231253772974,
          "train_metric": 0.0002997011470142752,
          "validation_metric": 0.00022734931553713977,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5483863160006877
        },
        {
          "epoch": 44,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.0032150754559552297,
          "train_loss": 0.0034050822723656893,
          "validation_loss": 0.002708876971155405,
          "train_metric": 0.00034294743090867996,
          "validation_metric": 0.0002728281542658806,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5593902000000526
        },
        {
          "epoch": 45,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.003295619786513271,
          "train_loss": 0.003672040067613125,
          "validation_loss": 0.0038277199491858482,
          "train_metric": 0.0003698343934956938,
          "validation_metric": 0.0003855139366351068,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.570432016000268
        },
        {
          "epoch": 46,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.003078979418205563,
          "train_loss": 0.0032184228766709566,
          "validation_loss": 0.0022972053848206997,
          "train_metric": 0.0003241478116251528,
          "validation_metric": 0.00023136608069762588,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5811459619999368
        },
        {
          "epoch": 47,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.003616947193222586,
          "train_loss": 0.0023462269455194473,
          "validation_loss": 0.002109962049871683,
          "train_metric": 0.00023630340001545846,
          "validation_metric": 0.00021250764257274568,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5918211860007432
        },
        {
          "epoch": 48,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.0023250239974004216,
          "train_loss": 0.0038655574899166822,
          "validation_loss": 0.002962973900139332,
          "train_metric": 0.00038932482129894197,
          "validation_metric": 0.00029841987998224795,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6028080040014174
        },
        {
          "epoch": 49,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.0029477911593858153,
          "train_loss": 0.0029394871089607477,
          "validation_loss": 0.0022885631769895554,
          "train_metric": 0.00029605437885038555,
          "validation_metric": 0.00023049568699207157,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6133957060010289
        },
        {
          "epoch": 50,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.0034806174007826485,
          "train_loss": 0.00380074349232018,
          "validation_loss": 0.0030071388464421034,
          "train_metric": 0.0003827969776466489,
          "validation_metric": 0.0003028680512215942,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6238946740013489
        },
        {
          "epoch": 51,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.002605419354949845,
          "train_loss": 0.0018213038565590978,
          "validation_loss": 0.0016792693641036749,
          "train_metric": 0.00018343504052609205,
          "validation_metric": 0.00016912985302042216,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6351675450005132
        },
        {
          "epoch": 52,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.004080665821675211,
          "train_loss": 0.002199184149503708,
          "validation_loss": 0.0019179549999535084,
          "train_metric": 0.00022149377036839724,
          "validation_metric": 0.000193169413250871,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6460666100010712
        },
        {
          "epoch": 53,
          "updates": 848,
          "samples_seen": 54272,
          "online_loss": 0.002720065040193731,
          "train_loss": 0.0019235775107517838,
          "validation_loss": 0.0016753959935158491,
          "train_metric": 0.00019373567192815244,
          "validation_metric": 0.00016873975982889533,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6573194360007619
        },
        {
          "epoch": 54,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0023779427974659484,
          "train_loss": 0.0017264370108023286,
          "validation_loss": 0.0013534598983824253,
          "train_metric": 0.00017388039850629866,
          "validation_metric": 0.0001363155315630138,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6688595330015232
        },
        {
          "epoch": 55,
          "updates": 880,
          "samples_seen": 56320,
          "online_loss": 0.0020156774007773492,
          "train_loss": 0.0016833902336657047,
          "validation_loss": 0.0012908440548926592,
          "train_metric": 0.00016954491729848087,
          "validation_metric": 0.0001300090953009203,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6805982020014198
        },
        {
          "epoch": 56,
          "updates": 896,
          "samples_seen": 57344,
          "online_loss": 0.0036628420784836635,
          "train_loss": 0.0045376671478152275,
          "validation_loss": 0.0038465620018541813,
          "train_metric": 0.0004570171586237848,
          "validation_metric": 0.0003874116519000381,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.6918478420011525
        },
        {
          "epoch": 57,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.002620893563289428,
          "train_loss": 0.0019287827890366316,
          "validation_loss": 0.00144688761793077,
          "train_metric": 0.00019425991922616959,
          "validation_metric": 0.00014572522195521742,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7030011480001122
        },
        {
          "epoch": 58,
          "updates": 928,
          "samples_seen": 59392,
          "online_loss": 0.0022101903650764143,
          "train_loss": 0.002622372005134821,
          "validation_loss": 0.0017477951478213072,
          "train_metric": 0.0002641156897880137,
          "validation_metric": 0.00017603152082301676,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7142333170013444
        },
        {
          "epoch": 59,
          "updates": 944,
          "samples_seen": 60416,
          "online_loss": 0.004260587873432087,
          "train_loss": 0.0030077865812927485,
          "validation_loss": 0.003109072567895055,
          "train_metric": 0.0003029332438018173,
          "validation_metric": 0.0003131343692075461,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7249585350000416
        },
        {
          "epoch": 60,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.0022365374134096783,
          "train_loss": 0.002216682769358158,
          "validation_loss": 0.00154267274774611,
          "train_metric": 0.00022325613826978952,
          "validation_metric": 0.0001553723559482023,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7359872670003824
        },
        {
          "epoch": 61,
          "updates": 976,
          "samples_seen": 62464,
          "online_loss": 0.0018296962225576863,
          "train_loss": 0.0020384720992296934,
          "validation_loss": 0.001905615208670497,
          "train_metric": 0.00020530739857349545,
          "validation_metric": 0.00019192656327504665,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7469786540004861
        },
        {
          "epoch": 62,
          "updates": 992,
          "samples_seen": 63488,
          "online_loss": 0.0017216375963471364,
          "train_loss": 0.0014483369886875153,
          "validation_loss": 0.0011452959151938558,
          "train_metric": 0.00014587122132070363,
          "validation_metric": 0.00011535001249285415,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7581891780009755
        },
        {
          "epoch": 63,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.002812097238347633,
          "train_loss": 0.014162757433950901,
          "validation_loss": 0.010514027439057827,
          "train_metric": 0.0014264211058616638,
          "validation_metric": 0.0010589343728497624,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7693170570000802
        },
        {
          "epoch": 64,
          "updates": 1024,
          "samples_seen": 65536,
          "online_loss": 0.007845269508834463,
          "train_loss": 0.0029648931231349707,
          "validation_loss": 0.002757398411631584,
          "train_metric": 0.0002986131876241416,
          "validation_metric": 0.00027771503664553165,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7805627470006584
        },
        {
          "epoch": 65,
          "updates": 1040,
          "samples_seen": 66560,
          "online_loss": 0.004364003838418284,
          "train_loss": 0.008987737819552422,
          "validation_loss": 0.007578937336802483,
          "train_metric": 0.0009052121895365417,
          "validation_metric": 0.0007633228087797761,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.7916399320001801
        },
        {
          "epoch": 66,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.004419900164066348,
          "train_loss": 0.001808779314160347,
          "validation_loss": 0.0013286637840792537,
          "train_metric": 0.00018217363685835153,
          "validation_metric": 0.0001338181464234367,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8029007130007813
        },
        {
          "epoch": 67,
          "updates": 1072,
          "samples_seen": 68608,
          "online_loss": 0.0031441365481441608,
          "train_loss": 0.0021038760896772146,
          "validation_loss": 0.001514333882369101,
          "train_metric": 0.00021189471590332687,
          "validation_metric": 0.00015251815784722567,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.813838908001344
        },
        {
          "epoch": 68,
          "updates": 1088,
          "samples_seen": 69632,
          "online_loss": 0.002240926074591698,
          "train_loss": 0.0018392369383946061,
          "validation_loss": 0.0014090447220951319,
          "train_metric": 0.00018524122424423695,
          "validation_metric": 0.00014191381342243403,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8244913570015342
        },
        {
          "epoch": 69,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.001856082963058725,
          "train_loss": 0.0015386184677481651,
          "validation_loss": 0.0014156489633023739,
          "train_metric": 0.0001549640146549791,
          "validation_metric": 0.0001425789960194379,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8352947990006214
        },
        {
          "epoch": 70,
          "updates": 1120,
          "samples_seen": 71680,
          "online_loss": 0.0017993034671235364,
          "train_loss": 0.002309999894350767,
          "validation_loss": 0.0018170294351875782,
          "train_metric": 0.00023265468189492822,
          "validation_metric": 0.0001830045657698065,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8462635380001302
        },
        {
          "epoch": 71,
          "updates": 1136,
          "samples_seen": 72704,
          "online_loss": 0.0013365714221436065,
          "train_loss": 0.0015817810781300068,
          "validation_loss": 0.0012222325894981623,
          "train_metric": 0.00015931121015455574,
          "validation_metric": 0.0001230988127645105,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8570036270011769
        },
        {
          "epoch": 72,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0014941570025257533,
          "train_loss": 0.0011872350005432963,
          "validation_loss": 0.0010791559470817447,
          "train_metric": 0.00011957396054640412,
          "validation_metric": 0.00010868861863855273,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8685079130009399
        },
        {
          "epoch": 73,
          "updates": 1168,
          "samples_seen": 74752,
          "online_loss": 0.0013925560051575303,
          "train_loss": 0.0013072044821456075,
          "validation_loss": 0.000965601415373385,
          "train_metric": 0.00013165685231797397,
          "validation_metric": 9.72518464550376e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8800403880013619
        },
        {
          "epoch": 74,
          "updates": 1184,
          "samples_seen": 75776,
          "online_loss": 0.0014475377829512581,
          "train_loss": 0.001517817610874772,
          "validation_loss": 0.0014482211554422975,
          "train_metric": 0.00015286901907529682,
          "validation_metric": 0.00014585955068469048,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8911960250006814
        },
        {
          "epoch": 75,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.001872633376478916,
          "train_loss": 0.0019470937550067902,
          "validation_loss": 0.0014618865679949522,
          "train_metric": 0.00019610414165072143,
          "validation_metric": 0.0001472358708269894,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9020515360007266
        },
        {
          "epoch": 76,
          "updates": 1216,
          "samples_seen": 77824,
          "online_loss": 0.0054850954984431155,
          "train_loss": 0.0013190304161980748,
          "validation_loss": 0.001257438096217811,
          "train_metric": 0.0001328479265794158,
          "validation_metric": 0.00012664457608480006,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9126316510009929
        },
        {
          "epoch": 77,
          "updates": 1232,
          "samples_seen": 78848,
          "online_loss": 0.0064657146322133485,
          "train_loss": 0.0052958386950194836,
          "validation_loss": 0.004091070499271154,
          "train_metric": 0.000533377518877387,
          "validation_metric": 0.00041203771252185106,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9231785810006841
        },
        {
          "epoch": 78,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.004608973162248731,
          "train_loss": 0.004421045072376728,
          "validation_loss": 0.003347221063449979,
          "train_metric": 0.0004452714929357171,
          "validation_metric": 0.00033711985452100635,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9338110340013372
        },
        {
          "epoch": 79,
          "updates": 1264,
          "samples_seen": 80896,
          "online_loss": 0.0037155583704588935,
          "train_loss": 0.001600156887434423,
          "validation_loss": 0.0012908277567476034,
          "train_metric": 0.00016116192273329943,
          "validation_metric": 0.0001300074509344995,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9444226600007823
        },
        {
          "epoch": 80,
          "updates": 1280,
          "samples_seen": 81920,
          "online_loss": 0.0019795540592895122,
          "train_loss": 0.0014186494518071413,
          "validation_loss": 0.0011992478976026177,
          "train_metric": 0.00014288116653915495,
          "validation_metric": 0.00012078385043423623,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9547999840015109
        },
        {
          "epoch": 81,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.0026520797437115107,
          "train_loss": 0.007639104966074228,
          "validation_loss": 0.004448641091585159,
          "train_metric": 0.0007693827501498163,
          "validation_metric": 0.00044805085053667426,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9654744370000117
        },
        {
          "epoch": 82,
          "updates": 1312,
          "samples_seen": 83968,
          "online_loss": 0.003698776454257313,
          "train_loss": 0.0014480710960924625,
          "validation_loss": 0.0010283075971528888,
          "train_metric": 0.0001458444312447682,
          "validation_metric": 0.00010356736311223358,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.976033199000085
        },
        {
          "epoch": 83,
          "updates": 1328,
          "samples_seen": 84992,
          "online_loss": 0.0013876126940886024,
          "train_loss": 0.0011812495067715645,
          "validation_loss": 0.0009725693962536752,
          "train_metric": 0.0001189711329061538,
          "validation_metric": 9.79536198428832e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9868358700005047
        },
        {
          "epoch": 84,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.0019841021130559966,
          "train_loss": 0.0021492906380444765,
          "validation_loss": 0.001738726976327598,
          "train_metric": 0.00021646871755365282,
          "validation_metric": 0.00017511819896753877,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9972936230005871
        },
        {
          "epoch": 85,
          "updates": 1360,
          "samples_seen": 87040,
          "online_loss": 0.0028180437220726162,
          "train_loss": 0.004712448921054602,
          "validation_loss": 0.0034331746865063906,
          "train_metric": 0.00047462055226787925,
          "validation_metric": 0.0003457767888903618,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.008104739999908
        },
        {
          "epoch": 86,
          "updates": 1376,
          "samples_seen": 88064,
          "online_loss": 0.002345712906389963,
          "train_loss": 0.0016680278349667788,
          "validation_loss": 0.001376044237986207,
          "train_metric": 0.0001679976558079943,
          "validation_metric": 0.00013859011232852936,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0187951470015832
        },
        {
          "epoch": 87,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0012541514261101838,
          "train_loss": 0.0010155471973121166,
          "validation_loss": 0.0007942775264382362,
          "train_metric": 0.00010228219616692513,
          "validation_metric": 7.999671652214602e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0298647569998138
        },
        {
          "epoch": 88,
          "updates": 1408,
          "samples_seen": 90112,
          "online_loss": 0.001138269546572701,
          "train_loss": 0.0010479723569005728,
          "validation_loss": 0.0008517660899087787,
          "train_metric": 0.00010554794425843284,
          "validation_metric": 8.578676352044567e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0413164060009876
        },
        {
          "epoch": 89,
          "updates": 1424,
          "samples_seen": 91136,
          "online_loss": 0.0009965488243324216,
          "train_loss": 0.00098379363771528,
          "validation_loss": 0.0008038965752348304,
          "train_metric": 9.90840999293141e-05,
          "validation_metric": 8.096552483038977e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0521208430000115
        },
        {
          "epoch": 90,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0010749639877758455,
          "train_loss": 0.0010404292261227965,
          "validation_loss": 0.0009060397278517485,
          "train_metric": 0.00010478822514414787,
          "validation_metric": 9.125300857704133e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0627626510013215
        },
        {
          "epoch": 91,
          "updates": 1456,
          "samples_seen": 93184,
          "online_loss": 0.001460254143239581,
          "train_loss": 0.0019443311030045152,
          "validation_loss": 0.001366279786452651,
          "train_metric": 0.00019582589447963983,
          "validation_metric": 0.00013760670844931155,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.073672421000083
        },
        {
          "epoch": 92,
          "updates": 1472,
          "samples_seen": 94208,
          "online_loss": 0.0012967810544068925,
          "train_loss": 0.001923888921737671,
          "validation_loss": 0.001730398740619421,
          "train_metric": 0.0001937670458573848,
          "validation_metric": 0.00017427944112569094,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0842904140008613
        },
        {
          "epoch": 93,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0019328592552483315,
          "train_loss": 0.0038160188123583794,
          "validation_loss": 0.002962038852274418,
          "train_metric": 0.00038433546433225274,
          "validation_metric": 0.0002983256708830595,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0946668360011245
        },
        {
          "epoch": 94,
          "updates": 1504,
          "samples_seen": 96256,
          "online_loss": 0.0017400491506123217,
          "train_loss": 0.0011966134188696742,
          "validation_loss": 0.0009719364461489022,
          "train_metric": 0.00012051851081196219,
          "validation_metric": 9.788988245418295e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1055937340006494
        },
        {
          "epoch": 95,
          "updates": 1520,
          "samples_seen": 97280,
          "online_loss": 0.0017479341368016321,
          "train_loss": 0.003015235997736454,
          "validation_loss": 0.0029419991187751293,
          "train_metric": 0.00030368351144716144,
          "validation_metric": 0.0002963073493447155,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1163765080000303
        },
        {
          "epoch": 96,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0018786116306728218,
          "train_loss": 0.0011003109393641353,
          "validation_loss": 0.0007198821404017508,
          "train_metric": 0.00011081926641054451,
          "validation_metric": 7.250389171531424e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1273855680010456
        },
        {
          "epoch": 97,
          "updates": 1552,
          "samples_seen": 99328,
          "online_loss": 0.001118400596169522,
          "train_loss": 0.0010641854023560882,
          "validation_loss": 0.000821062596514821,
          "train_metric": 0.00010718085832195356,
          "validation_metric": 8.26944233267568e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1381356600013532
        },
        {
          "epoch": 98,
          "updates": 1568,
          "samples_seen": 100352,
          "online_loss": 0.0014151977466099197,
          "train_loss": 0.0010325670009478927,
          "validation_loss": 0.0008353047305718064,
          "train_metric": 0.00010399638267699629,
          "validation_metric": 8.412883471464738e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1485741830001643
        },
        {
          "epoch": 99,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0022217766472749645,
          "train_loss": 0.0025818648282438517,
          "validation_loss": 0.0016593376640230417,
          "train_metric": 0.0002600359730422497,
          "validation_metric": 0.00016712241631466895,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1589974819999043
        },
        {
          "epoch": 100,
          "updates": 1600,
          "samples_seen": 102400,
          "online_loss": 0.0020864171729044756,
          "train_loss": 0.0021261502988636494,
          "validation_loss": 0.0019838493317365646,
          "train_metric": 0.00021413809736259282,
          "validation_metric": 0.00019980604702141136,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1694154890010395
        },
        {
          "epoch": 101,
          "updates": 1616,
          "samples_seen": 103424,
          "online_loss": 0.0014630054174631368,
          "train_loss": 0.0009938398143276572,
          "validation_loss": 0.000675144896376878,
          "train_metric": 0.0001000959164230153,
          "validation_metric": 6.799811671953648e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1805657479999354
        },
        {
          "epoch": 102,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.001151106131146662,
          "train_loss": 0.0012100051390007138,
          "validation_loss": 0.0010119632352143526,
          "train_metric": 0.0001218672696268186,
          "validation_metric": 0.00010192122863372788,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.191175395000755
        },
        {
          "epoch": 103,
          "updates": 1648,
          "samples_seen": 105472,
          "online_loss": 0.0018337278197577689,
          "train_loss": 0.002508733654394746,
          "validation_loss": 0.0015389523468911648,
          "train_metric": 0.0002526704629417509,
          "validation_metric": 0.00015499764413107187,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.202335996000329
        },
        {
          "epoch": 104,
          "updates": 1664,
          "samples_seen": 106496,
          "online_loss": 0.0011815213028967264,
          "train_loss": 0.0011719854082912207,
          "validation_loss": 0.0011014487827196717,
          "train_metric": 0.00011803808592958376,
          "validation_metric": 0.00011093389184679836,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.213338487999863
        },
        {
          "epoch": 105,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0009116000474023167,
          "train_loss": 0.0008642240427434444,
          "validation_loss": 0.0006143442005850375,
          "train_metric": 8.704147330718115e-05,
          "validation_metric": 6.187448889249936e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2243091580003238
        },
        {
          "epoch": 106,
          "updates": 1696,
          "samples_seen": 108544,
          "online_loss": 0.0008887615640560398,
          "train_loss": 0.001306250924244523,
          "validation_loss": 0.0010509275598451495,
          "train_metric": 0.00013156080967746675,
          "validation_metric": 0.00010584556730464101,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.235104258001229
        },
        {
          "epoch": 107,
          "updates": 1712,
          "samples_seen": 109568,
          "online_loss": 0.0014119302104518283,
          "train_loss": 0.0033885352313518524,
          "validation_loss": 0.0023170323111116886,
          "train_metric": 0.0003412808873690665,
          "validation_metric": 0.00023336302547249943,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2456709550006053
        },
        {
          "epoch": 108,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.001290787953621475,
          "train_loss": 0.0008294096332974732,
          "validation_loss": 0.0006559633184224367,
          "train_metric": 8.353509474545717e-05,
          "validation_metric": 6.606622628169134e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2562811020015943
        },
        {
          "epoch": 109,
          "updates": 1744,
          "samples_seen": 111616,
          "online_loss": 0.0009402629475516733,
          "train_loss": 0.00081358797615394,
          "validation_loss": 0.0006201669457368553,
          "train_metric": 8.194160182029009e-05,
          "validation_metric": 6.246093835216016e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2670265560009284
        },
        {
          "epoch": 110,
          "updates": 1760,
          "samples_seen": 112640,
          "online_loss": 0.0016191280592465773,
          "train_loss": 0.0017546454910188913,
          "validation_loss": 0.0013112009037286043,
          "train_metric": 0.0001767214562278241,
          "validation_metric": 0.0001320593582931906,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2775595750008506
        },
        {
          "epoch": 111,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.002896518899433431,
          "train_loss": 0.0031807313207536936,
          "validation_loss": 0.0036292378790676594,
          "train_metric": 0.0003203515661880374,
          "validation_metric": 0.00036552356323227286,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2880646890007483
        },
        {
          "epoch": 112,
          "updates": 1792,
          "samples_seen": 114688,
          "online_loss": 0.002246349489723798,
          "train_loss": 0.0020763855427503586,
          "validation_loss": 0.0011937487870454788,
          "train_metric": 0.00020912595209665596,
          "validation_metric": 0.0001202300118166022,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2986701520003407
        },
        {
          "epoch": 113,
          "updates": 1808,
          "samples_seen": 115712,
          "online_loss": 0.001525375646451721,
          "train_loss": 0.0011201828019693494,
          "validation_loss": 0.0008319831686094403,
          "train_metric": 0.00011282070772722363,
          "validation_metric": 8.379429345950484e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3092201939998631
        },
        {
          "epoch": 114,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.001263603935512947,
          "train_loss": 0.0016336179105564952,
          "validation_loss": 0.0013926696265116334,
          "train_metric": 0.00016453202988486737,
          "validation_metric": 0.00014026457211002707,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3196873699998832
        },
        {
          "epoch": 115,
          "updates": 1840,
          "samples_seen": 117760,
          "online_loss": 0.001392403897625627,
          "train_loss": 0.0013046482345089316,
          "validation_loss": 0.0009021891164593399,
          "train_metric": 0.00013139938528183848,
          "validation_metric": 9.086518548429012e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3302140350006084
        },
        {
          "epoch": 116,
          "updates": 1856,
          "samples_seen": 118784,
          "online_loss": 0.0015674877722631209,
          "train_loss": 0.0010566188720986247,
          "validation_loss": 0.0006733585614711046,
          "train_metric": 0.00010641875996952876,
          "validation_metric": 6.781821139156818e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3408250480006245
        },
        {
          "epoch": 117,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.0010538383594393963,
          "train_loss": 0.0007539812941104174,
          "validation_loss": 0.0005744522204622626,
          "train_metric": 7.593823102070019e-05,
          "validation_metric": 5.785672692582011e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.3518289790008566
        },
        {
          "epoch": 118,
          "updates": 1888,
          "samples_seen": 120832,
          "online_loss": 0.0009630692547943909,
          "train_loss": 0.0007359401206485927,
          "validation_loss": 0.000599755730945617,
          "train_metric": 7.412119884975255e-05,
          "validation_metric": 6.0405207477742806e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.3624146050005947
        },
        {
          "epoch": 119,
          "updates": 1904,
          "samples_seen": 121856,
          "online_loss": 0.0007728676646365784,
          "train_loss": 0.0007416592561639845,
          "validation_loss": 0.0006294507766142488,
          "train_metric": 7.469720003427938e-05,
          "validation_metric": 6.339597894111648e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.3733450640011142
        },
        {
          "epoch": 120,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0007630875843460672,
          "train_loss": 0.0007877536118030548,
          "validation_loss": 0.0007380867609754205,
          "train_metric": 7.933966116979718e-05,
          "validation_metric": 7.433738937834278e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.3840767910005525
        },
        {
          "epoch": 121,
          "updates": 1936,
          "samples_seen": 123904,
          "online_loss": 0.0013088174546282971,
          "train_loss": 0.0016405705828219652,
          "validation_loss": 0.0012900199508294463,
          "train_metric": 0.00016523225349374115,
          "validation_metric": 0.00012992610572837293,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.3946487800003524
        },
        {
          "epoch": 122,
          "updates": 1952,
          "samples_seen": 124928,
          "online_loss": 0.0012000570604868699,
          "train_loss": 0.0007895866874605417,
          "validation_loss": 0.0007408070960082114,
          "train_metric": 7.952428131829947e-05,
          "validation_metric": 7.461137283826247e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4055446010006563
        },
        {
          "epoch": 123,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0008329929696628824,
          "train_loss": 0.00073792296461761,
          "validation_loss": 0.0004965517437085509,
          "train_metric": 7.432090933434665e-05,
          "validation_metric": 5.001086901756935e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.416965460000938
        },
        {
          "epoch": 124,
          "updates": 1984,
          "samples_seen": 126976,
          "online_loss": 0.0008067919116001576,
          "train_loss": 0.0007859061006456614,
          "validation_loss": 0.0005208409857004881,
          "train_metric": 7.915359310572967e-05,
          "validation_metric": 5.24571914866101e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4274190520009142
        },
        {
          "epoch": 125,
          "updates": 2000,
          "samples_seen": 128000,
          "online_loss": 0.0008896638810256263,
          "train_loss": 0.0007983163231983781,
          "validation_loss": 0.0005726370727643371,
          "train_metric": 8.040350076043978e-05,
          "validation_metric": 5.7673903938848525e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4382929889998195
        },
        {
          "epoch": 126,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.0009440040212211898,
          "train_loss": 0.0006803326541557908,
          "validation_loss": 0.000577577156946063,
          "train_metric": 6.852061051176861e-05,
          "validation_metric": 5.8171452110400423e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4486920070012275
        },
        {
          "epoch": 127,
          "updates": 2032,
          "samples_seen": 130048,
          "online_loss": 0.0008375902725674678,
          "train_loss": 0.0006913222023285925,
          "validation_loss": 0.0006149025866761804,
          "train_metric": 6.962744373595342e-05,
          "validation_metric": 6.193073932081461e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4594846370000596
        },
        {
          "epoch": 128,
          "updates": 2048,
          "samples_seen": 131072,
          "online_loss": 0.0007404214065900305,
          "train_loss": 0.0006538117304444313,
          "validation_loss": 0.0005017479415982962,
          "train_metric": 6.58495191601105e-05,
          "validation_metric": 5.053421773482114e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.470404124000197
        },
        {
          "epoch": 129,
          "updates": 2064,
          "samples_seen": 132096,
          "online_loss": 0.0007067936621751869,
          "train_loss": 0.0007418058812618256,
          "validation_loss": 0.0005111866048537195,
          "train_metric": 7.471197750419378e-05,
          "validation_metric": 5.148484342498705e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.4808147940002527
        },
        {
          "epoch": 130,
          "updates": 2080,
          "samples_seen": 133120,
          "online_loss": 0.0007714972107351059,
          "train_loss": 0.0007295311079360545,
          "validation_loss": 0.0005250779213383794,
          "train_metric": 7.347569771809503e-05,
          "validation_metric": 5.288392276270315e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.491676453000764
        },
        {
          "epoch": 131,
          "updates": 2096,
          "samples_seen": 134144,
          "online_loss": 0.0007271570793818682,
          "train_loss": 0.0006390250637196004,
          "validation_loss": 0.0005283074569888413,
          "train_metric": 6.436026887968183e-05,
          "validation_metric": 5.3209190809866413e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5027506679998623
        },
        {
          "epoch": 132,
          "updates": 2112,
          "samples_seen": 135168,
          "online_loss": 0.0008120228048937861,
          "train_loss": 0.0006319767562672496,
          "validation_loss": 0.0004922047373838723,
          "train_metric": 6.36503827990964e-05,
          "validation_metric": 4.95730600960087e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5136105600013252
        },
        {
          "epoch": 133,
          "updates": 2128,
          "samples_seen": 136192,
          "online_loss": 0.0007135446003303514,
          "train_loss": 0.0006187251419760287,
          "validation_loss": 0.0005095007363706827,
          "train_metric": 6.23157320660539e-05,
          "validation_metric": 5.131504076416604e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5244184769999265
        },
        {
          "epoch": 134,
          "updates": 2144,
          "samples_seen": 137216,
          "online_loss": 0.0006890664699312765,
          "train_loss": 0.000653604045510292,
          "validation_loss": 0.0005648205406032503,
          "train_metric": 6.582860078196973e-05,
          "validation_metric": 5.688665260095149e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.534945424000398
        },
        {
          "epoch": 135,
          "updates": 2160,
          "samples_seen": 138240,
          "online_loss": 0.0006978870278544491,
          "train_loss": 0.000668891763780266,
          "validation_loss": 0.0004657748795580119,
          "train_metric": 6.7368324380368e-05,
          "validation_metric": 4.6911140088923275e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.546115369999825
        },
        {
          "epoch": 136,
          "updates": 2176,
          "samples_seen": 139264,
          "online_loss": 0.0008951014151534764,
          "train_loss": 0.0009096621070057154,
          "validation_loss": 0.0005645695491693914,
          "train_metric": 9.16178323677741e-05,
          "validation_metric": 5.686137592419982e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5568407790015044
        },
        {
          "epoch": 137,
          "updates": 2192,
          "samples_seen": 140288,
          "online_loss": 0.0010316480838810094,
          "train_loss": 0.0007710973732173443,
          "validation_loss": 0.0007840674370527267,
          "train_metric": 7.766210910631344e-05,
          "validation_metric": 7.896839088061824e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5674824550005724
        },
        {
          "epoch": 138,
          "updates": 2208,
          "samples_seen": 141312,
          "online_loss": 0.0012377157199807698,
          "train_loss": 0.0009587572421878576,
          "validation_loss": 0.0009327777661383152,
          "train_metric": 9.656252223066986e-05,
          "validation_metric": 9.394596418133005e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.5780805089998466
        },
        {
          "epoch": 139,
          "updates": 2224,
          "samples_seen": 142336,
          "online_loss": 0.000721626689482946,
          "train_loss": 0.0007017616298981011,
          "validation_loss": 0.0004969692090526223,
          "train_metric": 7.067886326694861e-05,
          "validation_metric": 5.005292041460052e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.588731658001052
        },
        {
          "epoch": 140,
          "updates": 2240,
          "samples_seen": 143360,
          "online_loss": 0.0007075288331179763,
          "train_loss": 0.0006003317539580166,
          "validation_loss": 0.00044900289503857493,
          "train_metric": 6.04632223257795e-05,
          "validation_metric": 4.5221924665383995e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6001096640011383
        },
        {
          "epoch": 141,
          "updates": 2256,
          "samples_seen": 144384,
          "online_loss": 0.0009192513298330596,
          "train_loss": 0.0011007862631231546,
          "validation_loss": 0.0008914555073715746,
          "train_metric": 0.00011086717859143391,
          "validation_metric": 8.978414552984759e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6109966770000028
        },
        {
          "epoch": 142,
          "updates": 2272,
          "samples_seen": 145408,
          "online_loss": 0.0007254140746226767,
          "train_loss": 0.0006168916006572545,
          "validation_loss": 0.0005967763718217611,
          "train_metric": 6.213106098584831e-05,
          "validation_metric": 6.010513607179746e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6224901350014989
        },
        {
          "epoch": 143,
          "updates": 2288,
          "samples_seen": 146432,
          "online_loss": 0.0006394125721271848,
          "train_loss": 0.0005713716964237392,
          "validation_loss": 0.0004783312324434519,
          "train_metric": 5.7546461903257295e-05,
          "validation_metric": 4.817576700588688e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6337869570015755
        },
        {
          "epoch": 144,
          "updates": 2304,
          "samples_seen": 147456,
          "online_loss": 0.0005919034520047717,
          "train_loss": 0.0005738321342505515,
          "validation_loss": 0.0004173588240519166,
          "train_metric": 5.7794277381617576e-05,
          "validation_metric": 4.203485150355846e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6455149620014708
        },
        {
          "epoch": 145,
          "updates": 2320,
          "samples_seen": 148480,
          "online_loss": 0.0005623415563604794,
          "train_loss": 0.0008316182065755129,
          "validation_loss": 0.0008234010310843587,
          "train_metric": 8.375753532163799e-05,
          "validation_metric": 8.29299315228127e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.657148539001355
        },
        {
          "epoch": 146,
          "updates": 2336,
          "samples_seen": 149504,
          "online_loss": 0.0007150471265049418,
          "train_loss": 0.0006346557638607919,
          "validation_loss": 0.0005659281741827726,
          "train_metric": 6.392020441126078e-05,
          "validation_metric": 5.699821485904977e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.668300060000547
        },
        {
          "epoch": 147,
          "updates": 2352,
          "samples_seen": 150528,
          "online_loss": 0.0005973811012154329,
          "train_loss": 0.0005512955831363797,
          "validation_loss": 0.0004147751023992896,
          "train_metric": 5.552446600631811e-05,
          "validation_metric": 4.177462687948719e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6801074060003884
        },
        {
          "epoch": 148,
          "updates": 2368,
          "samples_seen": 151552,
          "online_loss": 0.0005734397309424821,
          "train_loss": 0.0005500433035194874,
          "validation_loss": 0.0004248306795489043,
          "train_metric": 5.539834819501266e-05,
          "validation_metric": 4.278739288565703e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6914174749999802
        },
        {
          "epoch": 149,
          "updates": 2384,
          "samples_seen": 152576,
          "online_loss": 0.0005412529753812123,
          "train_loss": 0.000643218751065433,
          "validation_loss": 0.0006517860456369817,
          "train_metric": 6.478264549514279e-05,
          "validation_metric": 6.56455013086088e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.702458989000661
        },
        {
          "epoch": 150,
          "updates": 2400,
          "samples_seen": 153600,
          "online_loss": 0.001003230032438296,
          "train_loss": 0.0010066234972327948,
          "validation_loss": 0.0008854818297550082,
          "train_metric": 0.00010138342622667551,
          "validation_metric": 8.918248204281554e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7132826739998563
        },
        {
          "epoch": 151,
          "updates": 2416,
          "samples_seen": 154624,
          "online_loss": 0.0008094745244306978,
          "train_loss": 0.0007328369538299739,
          "validation_loss": 0.0006760487449355423,
          "train_metric": 7.380865281447768e-05,
          "validation_metric": 6.808913894928992e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.72414553300041
        },
        {
          "epoch": 152,
          "updates": 2432,
          "samples_seen": 155648,
          "online_loss": 0.000795601328718476,
          "train_loss": 0.0019932317081838846,
          "validation_loss": 0.001785092637874186,
          "train_metric": 0.00020075100474059582,
          "validation_metric": 0.00017978799587581307,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7347855790012545
        },
        {
          "epoch": 153,
          "updates": 2448,
          "samples_seen": 156672,
          "online_loss": 0.003982776901466423,
          "train_loss": 0.005400818772614002,
          "validation_loss": 0.0036137900315225124,
          "train_metric": 0.0005439506494440138,
          "validation_metric": 0.0003639676724560559,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.745403418000933
        },
        {
          "epoch": 154,
          "updates": 2464,
          "samples_seen": 157696,
          "online_loss": 0.0022029654173820745,
          "train_loss": 0.001263036043383181,
          "validation_loss": 0.0009959045564755797,
          "train_metric": 0.0001272083754884079,
          "validation_metric": 0.00010030385601567104,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.755966669999907
        },
        {
          "epoch": 155,
          "updates": 2480,
          "samples_seen": 158720,
          "online_loss": 0.0013887511922803242,
          "train_loss": 0.0009063013130798936,
          "validation_loss": 0.0009328422020189464,
          "train_metric": 9.127935481956229e-05,
          "validation_metric": 9.395246161147952e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.766401635000875
        },
        {
          "epoch": 156,
          "updates": 2496,
          "samples_seen": 159744,
          "online_loss": 0.0007176695453381399,
          "train_loss": 0.0006058876169845462,
          "validation_loss": 0.0004526523989625275,
          "train_metric": 6.102277620811947e-05,
          "validation_metric": 4.558949513011612e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7772247690008953
        },
        {
          "epoch": 157,
          "updates": 2512,
          "samples_seen": 160768,
          "online_loss": 0.0006477558908954961,
          "train_loss": 0.0005650977254845202,
          "validation_loss": 0.0004101101076230407,
          "train_metric": 5.691457772627473e-05,
          "validation_metric": 4.130478191655129e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7889184020004905
        },
        {
          "epoch": 158,
          "updates": 2528,
          "samples_seen": 161792,
          "online_loss": 0.000670060535412631,
          "train_loss": 0.0007781878812238574,
          "validation_loss": 0.0004984083352610469,
          "train_metric": 7.837622979423031e-05,
          "validation_metric": 5.0197857490275055e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7998955300008674
        },
        {
          "epoch": 159,
          "updates": 2544,
          "samples_seen": 162816,
          "online_loss": 0.0007500628398702247,
          "train_loss": 0.0005620337906293571,
          "validation_loss": 0.00041411624988541007,
          "train_metric": 5.660597889800556e-05,
          "validation_metric": 4.1708270146045834e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8107346750002762
        },
        {
          "epoch": 160,
          "updates": 2560,
          "samples_seen": 163840,
          "online_loss": 0.0005836155305587454,
          "train_loss": 0.0005472018383443356,
          "validation_loss": 0.00038942720857448876,
          "train_metric": 5.511216659215279e-05,
          "validation_metric": 3.9221678889589384e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.822011582000414
        },
        {
          "epoch": 161,
          "updates": 2576,
          "samples_seen": 164864,
          "online_loss": 0.000623301673840615,
          "train_loss": 0.0005659356247633696,
          "validation_loss": 0.0005244302446953952,
          "train_metric": 5.6998957006726414e-05,
          "validation_metric": 5.28186974406708e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8326705260005838
        },
        {
          "epoch": 162,
          "updates": 2592,
          "samples_seen": 165888,
          "online_loss": 0.0006820034814154496,
          "train_loss": 0.0006371367489919066,
          "validation_loss": 0.0006004131864756346,
          "train_metric": 6.417008262360469e-05,
          "validation_metric": 6.047141505405307e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8433689749999758
        },
        {
          "epoch": 163,
          "updates": 2608,
          "samples_seen": 166912,
          "online_loss": 0.00057037858368858,
          "train_loss": 0.0005716446903534234,
          "validation_loss": 0.0005395550979301333,
          "train_metric": 5.75739613850601e-05,
          "validation_metric": 5.434201375464909e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8541810240003542
        },
        {
          "epoch": 164,
          "updates": 2624,
          "samples_seen": 167936,
          "online_loss": 0.0005781241870863596,
          "train_loss": 0.0005461816326715052,
          "validation_loss": 0.00045858975499868393,
          "train_metric": 5.500942643266171e-05,
          "validation_metric": 4.61874769825954e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8647983020000538
        },
        {
          "epoch": 165,
          "updates": 2640,
          "samples_seen": 168960,
          "online_loss": 0.000547896164789563,
          "train_loss": 0.0013942391378805041,
          "validation_loss": 0.0011870099697262049,
          "train_metric": 0.00014042267866898328,
          "validation_metric": 0.00011955130321439356,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8754541220005194
        },
        {
          "epoch": 166,
          "updates": 2656,
          "samples_seen": 169984,
          "online_loss": 0.0007566760159534169,
          "train_loss": 0.0005671782419085503,
          "validation_loss": 0.000506838783621788,
          "train_metric": 5.7124121667584404e-05,
          "validation_metric": 5.104693991597742e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8863147890006076
        },
        {
          "epoch": 167,
          "updates": 2672,
          "samples_seen": 171008,
          "online_loss": 0.0005494132219610037,
          "train_loss": 0.0004830503021366894,
          "validation_loss": 0.0004159896052442491,
          "train_metric": 4.8651058023096994e-05,
          "validation_metric": 4.189694664091803e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8968394550010999
        },
        {
          "epoch": 168,
          "updates": 2688,
          "samples_seen": 172032,
          "online_loss": 0.0005338625578588108,
          "train_loss": 0.0007496257312595844,
          "validation_loss": 0.0007941900985315442,
          "train_metric": 7.549955626018345e-05,
          "validation_metric": 7.998792716534808e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9074443300014536
        },
        {
          "epoch": 169,
          "updates": 2704,
          "samples_seen": 173056,
          "online_loss": 0.000815714140117052,
          "train_loss": 0.0006566544179804623,
          "validation_loss": 0.0003837425902020186,
          "train_metric": 6.613582809222862e-05,
          "validation_metric": 3.8649151974823326e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9185039380008675
        },
        {
          "epoch": 170,
          "updates": 2720,
          "samples_seen": 174080,
          "online_loss": 0.0006869183707749471,
          "train_loss": 0.0006060826708562672,
          "validation_loss": 0.00037917433655820787,
          "train_metric": 6.104243220761418e-05,
          "validation_metric": 3.818904951913282e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9297200100008922
        },
        {
          "epoch": 171,
          "updates": 2736,
          "samples_seen": 175104,
          "online_loss": 0.0005823457850056002,
          "train_loss": 0.0005655809072777629,
          "validation_loss": 0.0004009166150353849,
          "train_metric": 5.696323569281958e-05,
          "validation_metric": 4.037885082652792e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9403851850001956
        },
        {
          "epoch": 172,
          "updates": 2752,
          "samples_seen": 176128,
          "online_loss": 0.0005157726445759181,
          "train_loss": 0.00046410958748310804,
          "validation_loss": 0.0004119148652534932,
          "train_metric": 4.674341471400112e-05,
          "validation_metric": 4.1486557165626436e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9509138820012595
        },
        {
          "epoch": 173,
          "updates": 2768,
          "samples_seen": 177152,
          "online_loss": 0.0005552327647819766,
          "train_loss": 0.0006200886564329267,
          "validation_loss": 0.00037273025372996926,
          "train_metric": 6.2453058490064e-05,
          "validation_metric": 3.754002682399005e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.961954123000396
        },
        {
          "epoch": 174,
          "updates": 2784,
          "samples_seen": 178176,
          "online_loss": 0.0004995061026420444,
          "train_loss": 0.0004813708656001836,
          "validation_loss": 0.00045683892676606774,
          "train_metric": 4.848191383644007e-05,
          "validation_metric": 4.6011140511836857e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9727575610013446
        },
        {
          "epoch": 175,
          "updates": 2800,
          "samples_seen": 179200,
          "online_loss": 0.0006330435107884114,
          "train_loss": 0.0007435770821757615,
          "validation_loss": 0.0007447457755915821,
          "train_metric": 7.48903548810631e-05,
          "validation_metric": 7.500806532334536e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9833230650001497
        },
        {
          "epoch": 176,
          "updates": 2816,
          "samples_seen": 180224,
          "online_loss": 0.0006641903128183912,
          "train_loss": 0.0008348728297278285,
          "validation_loss": 0.000778575602453202,
          "train_metric": 8.408531721215695e-05,
          "validation_metric": 7.841528713470325e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9936100220002118
        },
        {
          "epoch": 177,
          "updates": 2832,
          "samples_seen": 181248,
          "online_loss": 0.0006401800219464349,
          "train_loss": 0.00112800276838243,
          "validation_loss": 0.0009005216415971518,
          "train_metric": 0.0001136082864832133,
          "validation_metric": 9.069724183063954e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.00437679100105
        },
        {
          "epoch": 178,
          "updates": 2848,
          "samples_seen": 182272,
          "online_loss": 0.0008201971122616669,
          "train_loss": 0.000792064645793289,
          "validation_loss": 0.000698256422765553,
          "train_metric": 7.977385394042358e-05,
          "validation_metric": 7.032582652755082e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0150610900000174
        },
        {
          "epoch": 179,
          "updates": 2864,
          "samples_seen": 183296,
          "online_loss": 0.001020786776280147,
          "train_loss": 0.0011857454665005207,
          "validation_loss": 0.0006720059318467975,
          "train_metric": 0.00011942394485231489,
          "validation_metric": 6.768197636120021e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.025438043001486
        },
        {
          "epoch": 180,
          "updates": 2880,
          "samples_seen": 184320,
          "online_loss": 0.0010134124404430622,
          "train_loss": 0.001211515860632062,
          "validation_loss": 0.000975983974058181,
          "train_metric": 0.00012201943900436163,
          "validation_metric": 9.82975252554752e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0361470420011756
        },
        {
          "epoch": 181,
          "updates": 2896,
          "samples_seen": 185344,
          "online_loss": 0.0007228523900266737,
          "train_loss": 0.0008792271837592125,
          "validation_loss": 0.0007564472034573555,
          "train_metric": 8.855254418449476e-05,
          "validation_metric": 7.618659583386034e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0467011830005504
        },
        {
          "epoch": 182,
          "updates": 2912,
          "samples_seen": 186368,
          "online_loss": 0.0006868507862236584,
          "train_loss": 0.0005417423672042787,
          "validation_loss": 0.00043250180897302926,
          "train_metric": 5.456230428535491e-05,
          "validation_metric": 4.355999772087671e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0571188180001627
        },
        {
          "epoch": 183,
          "updates": 2928,
          "samples_seen": 187392,
          "online_loss": 0.0005358506059565116,
          "train_loss": 0.000732689630240202,
          "validation_loss": 0.0004578224034048617,
          "train_metric": 7.379380986094475e-05,
          "validation_metric": 4.611019539879635e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.067619839001054
        },
        {
          "epoch": 184,
          "updates": 2944,
          "samples_seen": 188416,
          "online_loss": 0.00058493761025602,
          "train_loss": 0.000660723599139601,
          "validation_loss": 0.0008176888804882765,
          "train_metric": 6.654565368080512e-05,
          "validation_metric": 8.235462155425921e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0781941120003466
        },
        {
          "epoch": 185,
          "updates": 2960,
          "samples_seen": 189440,
          "online_loss": 0.0006076570989534957,
          "train_loss": 0.00046097167069092393,
          "validation_loss": 0.0004909292911179364,
          "train_metric": 4.642737621907145e-05,
          "validation_metric": 4.94446067023091e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0885324190003303
        },
        {
          "epoch": 186,
          "updates": 2976,
          "samples_seen": 190464,
          "online_loss": 0.0004607835016940953,
          "train_loss": 0.00043785443995147943,
          "validation_loss": 0.0003484237822704017,
          "train_metric": 4.40990952483844e-05,
          "validation_metric": 3.509196903905831e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0997657400002936
        },
        {
          "epoch": 187,
          "updates": 2992,
          "samples_seen": 191488,
          "online_loss": 0.0004744569878312177,
          "train_loss": 0.0004554018669296056,
          "validation_loss": 0.0003306979197077453,
          "train_metric": 4.586640716297552e-05,
          "validation_metric": 3.330668550916016e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1111960720008938
        },
        {
          "epoch": 188,
          "updates": 3008,
          "samples_seen": 192512,
          "online_loss": 0.00046554185973946005,
          "train_loss": 0.00045311416033655405,
          "validation_loss": 0.00035722393658943474,
          "train_metric": 4.563600305118598e-05,
          "validation_metric": 3.597828617785126e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1217208160014707
        },
        {
          "epoch": 189,
          "updates": 3024,
          "samples_seen": 193536,
          "online_loss": 0.0005746487495343899,
          "train_loss": 0.00047855067532509565,
          "validation_loss": 0.0003148105461150408,
          "train_metric": 4.819787136511877e-05,
          "validation_metric": 3.1706564186606556e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1327009240012558
        },
        {
          "epoch": 190,
          "updates": 3040,
          "samples_seen": 194560,
          "online_loss": 0.0004498844546105829,
          "train_loss": 0.00041475496254861355,
          "validation_loss": 0.0003851274377666414,
          "train_metric": 4.1772600525291637e-05,
          "validation_metric": 3.87886211683508e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1434641720006766
        },
        {
          "epoch": 191,
          "updates": 3056,
          "samples_seen": 195584,
          "online_loss": 0.0005899745683564106,
          "train_loss": 0.0005434637423604727,
          "validation_loss": 0.0005797285703010857,
          "train_metric": 5.4735675803385675e-05,
          "validation_metric": 5.838813740410842e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1547913450012857
        },
        {
          "epoch": 192,
          "updates": 3072,
          "samples_seen": 196608,
          "online_loss": 0.00046762269448663574,
          "train_loss": 0.0004214799264445901,
          "validation_loss": 0.00038602884160354733,
          "train_metric": 4.244991214363836e-05,
          "validation_metric": 3.887941056746058e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.165906452000854
        },
        {
          "epoch": 193,
          "updates": 3088,
          "samples_seen": 197632,
          "online_loss": 0.0005697526703443145,
          "train_loss": 0.00041759078158065677,
          "validation_loss": 0.00040157060720957816,
          "train_metric": 4.205821824143641e-05,
          "validation_metric": 4.044472370878793e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.176965055999972
        },
        {
          "epoch": 194,
          "updates": 3104,
          "samples_seen": 198656,
          "online_loss": 0.00046158706845744746,
          "train_loss": 0.0004591844044625759,
          "validation_loss": 0.000461281044408679,
          "train_metric": 4.624736902769655e-05,
          "validation_metric": 4.6458539145532995e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1879695990010077
        },
        {
          "epoch": 195,
          "updates": 3120,
          "samples_seen": 199680,
          "online_loss": 0.0004429160389918252,
          "train_loss": 0.00040060980245471,
          "validation_loss": 0.0003516266297083348,
          "train_metric": 4.0347949834540486e-05,
          "validation_metric": 3.541454134392552e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1990625790003833
        },
        {
          "epoch": 196,
          "updates": 3136,
          "samples_seen": 200704,
          "online_loss": 0.00041881856213876745,
          "train_loss": 0.0005992746446281672,
          "validation_loss": 0.0006487033679150045,
          "train_metric": 6.0356749600032344e-05,
          "validation_metric": 6.53350361972116e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2098169900000357
        },
        {
          "epoch": 197,
          "updates": 3152,
          "samples_seen": 201728,
          "online_loss": 0.0005972120961814653,
          "train_loss": 0.0006366533925756812,
          "validation_loss": 0.0006278879591263831,
          "train_metric": 6.412139191525057e-05,
          "validation_metric": 6.323857815004885e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.220747895000386
        },
        {
          "epoch": 198,
          "updates": 3168,
          "samples_seen": 202752,
          "online_loss": 0.0006097866862546653,
          "train_loss": 0.0005167157505638897,
          "validation_loss": 0.0005880665848962963,
          "train_metric": 5.204171611694619e-05,
          "validation_metric": 5.9227910242043436e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2313252769999963
        },
        {
          "epoch": 199,
          "updates": 3184,
          "samples_seen": 203776,
          "online_loss": 0.00043761560846178327,
          "train_loss": 0.00038862760993652046,
          "validation_loss": 0.0003382504510227591,
          "train_metric": 3.91411485907156e-05,
          "validation_metric": 3.406735049793497e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.24182221600131
        },
        {
          "epoch": 200,
          "updates": 3200,
          "samples_seen": 204800,
          "online_loss": 0.000437162455455109,
          "train_loss": 0.0003931705141440034,
          "validation_loss": 0.0003335249493829906,
          "train_metric": 3.959869354730472e-05,
          "validation_metric": 3.3591415558476e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.2527105110002594
        },
        {
          "epoch": 201,
          "updates": 3216,
          "samples_seen": 205824,
          "online_loss": 0.0003986622359661851,
          "train_loss": 0.0003838477423414588,
          "validation_loss": 0.000369026034604758,
          "train_metric": 3.865973849315196e-05,
          "validation_metric": 3.71669557353016e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.263224018000983
        },
        {
          "epoch": 202,
          "updates": 3232,
          "samples_seen": 206848,
          "online_loss": 0.0003895975423802156,
          "train_loss": 0.000380951096303761,
          "validation_loss": 0.0003599485207814723,
          "train_metric": 3.8368001696653664e-05,
          "validation_metric": 3.625269528129138e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.273667737999858
        },
        {
          "epoch": 203,
          "updates": 3248,
          "samples_seen": 207872,
          "online_loss": 0.0004010049442513264,
          "train_loss": 0.00038437574403360486,
          "validation_loss": 0.0003351276973262429,
          "train_metric": 3.871291482937522e-05,
          "validation_metric": 3.375283267814666e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.2842025340014516
        },
        {
          "epoch": 204,
          "updates": 3264,
          "samples_seen": 208896,
          "online_loss": 0.00041521532875776757,
          "train_loss": 0.000396302726585418,
          "validation_loss": 0.0004246995667926967,
          "train_metric": 3.991415724158287e-05,
          "validation_metric": 4.2774187022587284e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.2947580960008054
        },
        {
          "epoch": 205,
          "updates": 3280,
          "samples_seen": 209920,
          "online_loss": 0.0004157355169809307,
          "train_loss": 0.00045669457176700234,
          "validation_loss": 0.0003152732679154724,
          "train_metric": 4.599660314852372e-05,
          "validation_metric": 3.175317033310421e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.364355761999832
        },
        {
          "epoch": 206,
          "updates": 3296,
          "samples_seen": 210944,
          "online_loss": 0.0004196903528281837,
          "train_loss": 0.000386418221751228,
          "validation_loss": 0.00039148260839283466,
          "train_metric": 3.891862797900103e-05,
          "validation_metric": 3.942869443562813e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.375100368000858
        },
        {
          "epoch": 207,
          "updates": 3312,
          "samples_seen": 211968,
          "online_loss": 0.0004131152199988719,
          "train_loss": 0.000426965591032058,
          "validation_loss": 0.00047346146311610937,
          "train_metric": 4.3002411985071376e-05,
          "validation_metric": 4.7685309255030006e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.385875671001486
        },
        {
          "epoch": 208,
          "updates": 3328,
          "samples_seen": 212992,
          "online_loss": 0.00040633550634083804,
          "train_loss": 0.00038776747533120215,
          "validation_loss": 0.0003243548271711916,
          "train_metric": 3.9054521039361134e-05,
          "validation_metric": 3.2667830964783207e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.396520395001062
        },
        {
          "epoch": 209,
          "updates": 3344,
          "samples_seen": 214016,
          "online_loss": 0.0003861074146698229,
          "train_loss": 0.00041193337528966367,
          "validation_loss": 0.00044810122926719487,
          "train_metric": 4.1488423448754475e-05,
          "validation_metric": 4.513111343840137e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.40713766600129
        }
      ],
      "best": {
        "epoch": 189,
        "updates": 3024
      },
      "best_validation_metric": 3.1706564186606556e-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": 0.005630858210486795
    },
    "14985df6b90f_11": {
      "history": [
        {
          "epoch": 1,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.2918598894029856,
          "train_loss": 0.08864232152700424,
          "validation_loss": 0.07799669355154037,
          "train_metric": 0.008927728049457073,
          "validation_metric": 0.007855541072785854,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.013911497999288258
        },
        {
          "epoch": 2,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.07088851742446423,
          "train_loss": 0.06481601297855377,
          "validation_loss": 0.0761234313249588,
          "train_metric": 0.006528032012283802,
          "validation_metric": 0.007666872814297676,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.025641720998464734
        },
        {
          "epoch": 3,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.042878184583969414,
          "train_loss": 0.037330325692892075,
          "validation_loss": 0.03774002939462662,
          "train_metric": 0.003759773913770914,
          "validation_metric": 0.0038010375574231148,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.03714907099856646
        },
        {
          "epoch": 4,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.035872515523806214,
          "train_loss": 0.03026277758181095,
          "validation_loss": 0.029498307034373283,
          "train_metric": 0.0030479563865810633,
          "validation_metric": 0.0029709613882005215,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.048540646999754244
        },
        {
          "epoch": 5,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.03364978532772511,
          "train_loss": 0.04825786501169205,
          "validation_loss": 0.0387478843331337,
          "train_metric": 0.004860354587435722,
          "validation_metric": 0.0039025451987981796,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.059454226999150706
        },
        {
          "epoch": 6,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.021768706443253905,
          "train_loss": 0.01536557637155056,
          "validation_loss": 0.012449638918042183,
          "train_metric": 0.0015475646359845996,
          "validation_metric": 0.001253881840966642,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.070891921999646
        },
        {
          "epoch": 7,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.016361093265004456,
          "train_loss": 0.03012162074446678,
          "validation_loss": 0.022713040933012962,
          "train_metric": 0.003033739048987627,
          "validation_metric": 0.0022875743452459574,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08170923399848107
        },
        {
          "epoch": 8,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.015725061297416687,
          "train_loss": 0.02052845060825348,
          "validation_loss": 0.0129659753292799,
          "train_metric": 0.002067550551146269,
          "validation_metric": 0.0013058854965493083,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0925400569994963
        },
        {
          "epoch": 9,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.03412104261224158,
          "train_loss": 0.0281328447163105,
          "validation_loss": 0.02257581613957882,
          "train_metric": 0.0028334371745586395,
          "validation_metric": 0.0022737535182386637,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1667573809991154
        },
        {
          "epoch": 10,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.0245996072771959,
          "train_loss": 0.014161459170281887,
          "validation_loss": 0.01212075911462307,
          "train_metric": 0.001426290487870574,
          "validation_metric": 0.001220758305862546,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.17872288299986394
        },
        {
          "epoch": 11,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.01971524575492367,
          "train_loss": 0.013194844126701355,
          "validation_loss": 0.01390079502016306,
          "train_metric": 0.0013289365451782942,
          "validation_metric": 0.0014000372029840946,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.18999181899926043
        },
        {
          "epoch": 12,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.016061629517935216,
          "train_loss": 0.009087281301617622,
          "validation_loss": 0.00776623236015439,
          "train_metric": 0.0009152375860139728,
          "validation_metric": 0.000782186514697969,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.201420336999945
        },
        {
          "epoch": 13,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.016754603464505635,
          "train_loss": 0.011239910498261452,
          "validation_loss": 0.008611350320279598,
          "train_metric": 0.0011320423800498247,
          "validation_metric": 0.0008673035772517323,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21231112699933874
        },
        {
          "epoch": 14,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.014856431516818702,
          "train_loss": 0.011729576624929905,
          "validation_loss": 0.00934813916683197,
          "train_metric": 0.0011813601013273,
          "validation_metric": 0.0009415103122591972,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.222987885999828
        },
        {
          "epoch": 15,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.022869698019349016,
          "train_loss": 0.03948310762643814,
          "validation_loss": 0.036755409091711044,
          "train_metric": 0.003976593725383282,
          "validation_metric": 0.003701870096847415,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.23404979899896716
        },
        {
          "epoch": 16,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.021364092215662822,
          "train_loss": 0.018701668828725815,
          "validation_loss": 0.017138896510004997,
          "train_metric": 0.001883563818410039,
          "validation_metric": 0.0017261671600863338,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.24503307599843538
        },
        {
          "epoch": 17,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.011553974603884853,
          "train_loss": 0.008492782711982727,
          "validation_loss": 0.006388961803168058,
          "train_metric": 0.0008553619845770299,
          "validation_metric": 0.000643473002128303,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25633676999859745
        },
        {
          "epoch": 18,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.011279565369477496,
          "train_loss": 0.01520159374922514,
          "validation_loss": 0.010612648911774158,
          "train_metric": 0.0015310489106923342,
          "validation_metric": 0.0010688671609386802,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2670029879991489
        },
        {
          "epoch": 19,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.013181203394196928,
          "train_loss": 0.0082817068323493,
          "validation_loss": 0.009010737761855125,
          "train_metric": 0.0008341032662428916,
          "validation_metric": 0.0009075284469872713,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.27795643599893083
        },
        {
          "epoch": 20,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.008930652780691162,
          "train_loss": 0.01041437592357397,
          "validation_loss": 0.012514843605458736,
          "train_metric": 0.0010488976258784533,
          "validation_metric": 0.0012604491785168648,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.28836089099968376
        },
        {
          "epoch": 21,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.007346003723796457,
          "train_loss": 0.005623728036880493,
          "validation_loss": 0.004434825852513313,
          "train_metric": 0.0005664012278430164,
          "validation_metric": 0.00044665936729870737,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.29939259299862897
        },
        {
          "epoch": 22,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.01089118086383678,
          "train_loss": 0.013976730406284332,
          "validation_loss": 0.010552308522164822,
          "train_metric": 0.0014076848747208714,
          "validation_metric": 0.0010627898154780269,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3104401479995431
        },
        {
          "epoch": 23,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.01286614901619032,
          "train_loss": 0.012942020781338215,
          "validation_loss": 0.01053493283689022,
          "train_metric": 0.001303472905419767,
          "validation_metric": 0.0010610398603603244,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3213545399994473
        },
        {
          "epoch": 24,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.01680598038365133,
          "train_loss": 0.007113293278962374,
          "validation_loss": 0.005545926280319691,
          "train_metric": 0.0007164248963817954,
          "validation_metric": 0.0005585654289461672,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3319948319985997
        },
        {
          "epoch": 25,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.009716997010400519,
          "train_loss": 0.007082410156726837,
          "validation_loss": 0.0067711095325648785,
          "train_metric": 0.0007133144536055624,
          "validation_metric": 0.0006819613045081496,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3427876189998642
        },
        {
          "epoch": 26,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.007690569633268751,
          "train_loss": 0.005337899550795555,
          "validation_loss": 0.004222970921546221,
          "train_metric": 0.0005376136396080256,
          "validation_metric": 0.0004253221850376576,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.35413525399962964
        },
        {
          "epoch": 27,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.015077051655680407,
          "train_loss": 0.024322792887687683,
          "validation_loss": 0.020289147272706032,
          "train_metric": 0.0024497027043253183,
          "validation_metric": 0.0020434486214071512,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.36470553599974664
        },
        {
          "epoch": 28,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.01614713718299754,
          "train_loss": 0.010776746086776257,
          "validation_loss": 0.009896081872284412,
          "train_metric": 0.0010853944113478065,
          "validation_metric": 0.0009966972284018993,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3757704419986112
        },
        {
          "epoch": 29,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.009654254477936774,
          "train_loss": 0.01019070390611887,
          "validation_loss": 0.007867228239774704,
          "train_metric": 0.0010263704461976886,
          "validation_metric": 0.0007923583616502583,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.38664076199893316
        },
        {
          "epoch": 30,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.009363326942548156,
          "train_loss": 0.007015187293291092,
          "validation_loss": 0.005033033899962902,
          "train_metric": 0.0007065440295264125,
          "validation_metric": 0.0005069086910225451,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.39756105599917646
        },
        {
          "epoch": 31,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.01186647597933188,
          "train_loss": 0.036512408405542374,
          "validation_loss": 0.031734712421894073,
          "train_metric": 0.0036773961037397385,
          "validation_metric": 0.0031962040811777115,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.408449684999141
        },
        {
          "epoch": 32,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.01852813793811947,
          "train_loss": 0.00820240005850792,
          "validation_loss": 0.007192524150013924,
          "train_metric": 0.0008261156617663801,
          "validation_metric": 0.0007244047010317445,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41947455599984096
        },
        {
          "epoch": 33,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.008491960179526359,
          "train_loss": 0.005819373764097691,
          "validation_loss": 0.004808956757187843,
          "train_metric": 0.0005861060344614089,
          "validation_metric": 0.00048434047494083643,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.43006015199898684
        },
        {
          "epoch": 34,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.008550145721528679,
          "train_loss": 0.007180631626397371,
          "validation_loss": 0.00788583792746067,
          "train_metric": 0.0007232069037854671,
          "validation_metric": 0.000794232648331672,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.44073110399949655
        },
        {
          "epoch": 35,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.009529343107715249,
          "train_loss": 0.013074944727122784,
          "validation_loss": 0.012273103930056095,
          "train_metric": 0.0013168605510145426,
          "validation_metric": 0.0012361020781099796,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4516353879989765
        },
        {
          "epoch": 36,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.026141693306271918,
          "train_loss": 0.037616487592458725,
          "validation_loss": 0.03295835480093956,
          "train_metric": 0.0037885946221649647,
          "validation_metric": 0.0033194448333233595,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4620645669983787
        },
        {
          "epoch": 37,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.01746085361810401,
          "train_loss": 0.013114810921251774,
          "validation_loss": 0.012586081400513649,
          "train_metric": 0.0013208757154643536,
          "validation_metric": 0.0012676240876317024,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4729169610000099
        },
        {
          "epoch": 38,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.010132260547834449,
          "train_loss": 0.007585165090858936,
          "validation_loss": 0.007952830754220486,
          "train_metric": 0.0007639501127414405,
          "validation_metric": 0.0008009800221771002,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4837115949994768
        },
        {
          "epoch": 39,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.007146771589759737,
          "train_loss": 0.0056703523732721806,
          "validation_loss": 0.00408181780949235,
          "train_metric": 0.0005710971308872104,
          "validation_metric": 0.0004111057205591351,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.49506772399945476
        },
        {
          "epoch": 40,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.006550184145453386,
          "train_loss": 0.007307811640202999,
          "validation_loss": 0.005809319671243429,
          "train_metric": 0.0007360160816460848,
          "validation_metric": 0.0005850933375768363,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5062074749985186
        },
        {
          "epoch": 41,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.004882334527792409,
          "train_loss": 0.0038950606249272823,
          "validation_loss": 0.0029174494557082653,
          "train_metric": 0.00039229626418091357,
          "validation_metric": 0.0002938348043244332,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5177029389997188
        },
        {
          "epoch": 42,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.00438209627463948,
          "train_loss": 0.003544508945196867,
          "validation_loss": 0.002924832981079817,
          "train_metric": 0.00035698997089639306,
          "validation_metric": 0.0002945784945040941,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5284329139994952
        },
        {
          "epoch": 43,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.004237340333929751,
          "train_loss": 0.003243435872718692,
          "validation_loss": 0.0024500698782503605,
          "train_metric": 0.0003266669518779963,
          "validation_metric": 0.0002467620652168989,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5398825879983633
        },
        {
          "epoch": 44,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.0033664343573036604,
          "train_loss": 0.0030133433174341917,
          "validation_loss": 0.0030486444011330605,
          "train_metric": 0.0003034929104615003,
          "validation_metric": 0.00030704832170158625,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5511595299994951
        },
        {
          "epoch": 45,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.004683814971940592,
          "train_loss": 0.004118543118238449,
          "validation_loss": 0.002932007657364011,
          "train_metric": 0.0004148045554757118,
          "validation_metric": 0.00029530105530284345,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5623510339992208
        },
        {
          "epoch": 46,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.002896019388572313,
          "train_loss": 0.0023314848076552153,
          "validation_loss": 0.0018630980048328638,
          "train_metric": 0.00023481860989704728,
          "validation_metric": 0.00018764441483654082,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.573694146998605
        },
        {
          "epoch": 47,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.0027481613724376075,
          "train_loss": 0.0026821109931916,
          "validation_loss": 0.0030792420729994774,
          "train_metric": 0.0002701324119698256,
          "validation_metric": 0.00031013000989332795,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5845444099995802
        },
        {
          "epoch": 48,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.002649345431564143,
          "train_loss": 0.003925068769603968,
          "validation_loss": 0.004198695998638868,
          "train_metric": 0.0003953185223508626,
          "validation_metric": 0.0004228772595524788,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5952686509990599
        },
        {
          "epoch": 49,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.003079888985666912,
          "train_loss": 0.002275755163282156,
          "validation_loss": 0.0015838653780519962,
          "train_metric": 0.000229205732466653,
          "validation_metric": 0.0001595211069798097,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6066119709994382
        },
        {
          "epoch": 50,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.002326218116650125,
          "train_loss": 0.00522620091214776,
          "validation_loss": 0.003146749921143055,
          "train_metric": 0.0005263639031909406,
          "validation_metric": 0.00031692913034930825,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6173132709991478
        },
        {
          "epoch": 51,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.00480363740643952,
          "train_loss": 0.004983453545719385,
          "validation_loss": 0.0034358741249889135,
          "train_metric": 0.0005019151722081006,
          "validation_metric": 0.0003460486768744886,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6280319429988594
        },
        {
          "epoch": 52,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.0028441566828405485,
          "train_loss": 0.002820125548169017,
          "validation_loss": 0.0023110825568437576,
          "train_metric": 0.00028403272153809667,
          "validation_metric": 0.00023276376305148005,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6387468709999666
        },
        {
          "epoch": 53,
          "updates": 848,
          "samples_seen": 54272,
          "online_loss": 0.001936301625391934,
          "train_loss": 0.0018195986049249768,
          "validation_loss": 0.0015617499593645334,
          "train_metric": 0.00018326332792639732,
          "validation_metric": 0.00015729373262729496,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6502632629999425
        },
        {
          "epoch": 54,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0017413092791684903,
          "train_loss": 0.0031077933963388205,
          "validation_loss": 0.003102565184235573,
          "train_metric": 0.0003130055556539446,
          "validation_metric": 0.00031247898004949093,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.661091749998377
        },
        {
          "epoch": 55,
          "updates": 880,
          "samples_seen": 56320,
          "online_loss": 0.004894323144981172,
          "train_loss": 0.004866339731961489,
          "validation_loss": 0.0039683375507593155,
          "train_metric": 0.0004901199135929346,
          "validation_metric": 0.000399676471715793,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6719401959999232
        },
        {
          "epoch": 56,
          "updates": 896,
          "samples_seen": 57344,
          "online_loss": 0.0041966439530369826,
          "train_loss": 0.00346235278993845,
          "validation_loss": 0.0025353501550853252,
          "train_metric": 0.00034871548996306956,
          "validation_metric": 0.00025535112945362926,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6825067559984745
        },
        {
          "epoch": 57,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.0034974187510670163,
          "train_loss": 0.0026655583642423153,
          "validation_loss": 0.0022552325390279293,
          "train_metric": 0.00026846525724977255,
          "validation_metric": 0.00022713873477187008,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6932855209997797
        },
        {
          "epoch": 58,
          "updates": 928,
          "samples_seen": 59392,
          "online_loss": 0.0020061367249581963,
          "train_loss": 0.0017767900135368109,
          "validation_loss": 0.001756118144840002,
          "train_metric": 0.00017895179917104542,
          "validation_metric": 0.00017686979845166206,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7037866119990213
        },
        {
          "epoch": 59,
          "updates": 944,
          "samples_seen": 60416,
          "online_loss": 0.0017039553422364406,
          "train_loss": 0.0014474795898422599,
          "validation_loss": 0.0011961733689531684,
          "train_metric": 0.00014578485570382327,
          "validation_metric": 0.00012047417840221897,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7149263100000098
        },
        {
          "epoch": 60,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.0017270449061470572,
          "train_loss": 0.0019031697884202003,
          "validation_loss": 0.0018503154860809445,
          "train_metric": 0.00019168028666172177,
          "validation_metric": 0.00018635702144820243,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7256728109987307
        },
        {
          "epoch": 61,
          "updates": 976,
          "samples_seen": 62464,
          "online_loss": 0.002180475123168435,
          "train_loss": 0.002288784831762314,
          "validation_loss": 0.0016458157915621996,
          "train_metric": 0.0002305180241819471,
          "validation_metric": 0.00016576053167227656,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.73626491199866
        },
        {
          "epoch": 62,
          "updates": 992,
          "samples_seen": 63488,
          "online_loss": 0.0015423729964822996,
          "train_loss": 0.0018165878718718886,
          "validation_loss": 0.001704301335848868,
          "train_metric": 0.00018296006601303816,
          "validation_metric": 0.00017165098688565195,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7467538459986827
        },
        {
          "epoch": 63,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.0032234261998382863,
          "train_loss": 0.0023911548778414726,
          "validation_loss": 0.001842659548856318,
          "train_metric": 0.00024082837626338005,
          "validation_metric": 0.0001855859300121665,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7572705779984972
        },
        {
          "epoch": 64,
          "updates": 1024,
          "samples_seen": 65536,
          "online_loss": 0.006601787816180149,
          "train_loss": 0.006433246657252312,
          "validation_loss": 0.006822941824793816,
          "train_metric": 0.0006479330477304757,
          "validation_metric": 0.000687181658577174,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7679025219986215
        },
        {
          "epoch": 65,
          "updates": 1040,
          "samples_seen": 66560,
          "online_loss": 0.0035965959759778343,
          "train_loss": 0.0028520552441477776,
          "validation_loss": 0.0024737175554037094,
          "train_metric": 0.0002872485783882439,
          "validation_metric": 0.0002491437771823257,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7787064289987029
        },
        {
          "epoch": 66,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.0028004393861920107,
          "train_loss": 0.004272925201803446,
          "validation_loss": 0.004459010437130928,
          "train_metric": 0.00043035339331254363,
          "validation_metric": 0.0004490951541811228,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7895096759984881
        },
        {
          "epoch": 67,
          "updates": 1072,
          "samples_seen": 68608,
          "online_loss": 0.0028591335758392233,
          "train_loss": 0.0016040296759456396,
          "validation_loss": 0.0011391775915399194,
          "train_metric": 0.00016155197226908058,
          "validation_metric": 0.00011473381164250895,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8007892059995356
        },
        {
          "epoch": 68,
          "updates": 1088,
          "samples_seen": 69632,
          "online_loss": 0.0040647276655363385,
          "train_loss": 0.003866623854264617,
          "validation_loss": 0.004257454536855221,
          "train_metric": 0.00038943218532949686,
          "validation_metric": 0.00042879529064521194,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8120255709982303
        },
        {
          "epoch": 69,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.0020060224014741834,
          "train_loss": 0.002261676359921694,
          "validation_loss": 0.0018569206586107612,
          "train_metric": 0.00022778772108722478,
          "validation_metric": 0.000187022247700952,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8230429539999022
        },
        {
          "epoch": 70,
          "updates": 1120,
          "samples_seen": 71680,
          "online_loss": 0.0018406348826829344,
          "train_loss": 0.0015012867515906692,
          "validation_loss": 0.0011148540070280433,
          "train_metric": 0.00015120411990210414,
          "validation_metric": 0.00011228400398977101,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8341817499986064
        },
        {
          "epoch": 71,
          "updates": 1136,
          "samples_seen": 72704,
          "online_loss": 0.002162911870982498,
          "train_loss": 0.011232945136725903,
          "validation_loss": 0.00887205172330141,
          "train_metric": 0.0011313408613204956,
          "validation_metric": 0.0008935605874285102,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8451375029999326
        },
        {
          "epoch": 72,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0084548872691812,
          "train_loss": 0.005071002524346113,
          "validation_loss": 0.004630617331713438,
          "train_metric": 0.0005107328179292381,
          "validation_metric": 0.0004663788713514805,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8558823769999435
        },
        {
          "epoch": 73,
          "updates": 1168,
          "samples_seen": 74752,
          "online_loss": 0.005415471823653206,
          "train_loss": 0.003343538148328662,
          "validation_loss": 0.0032748656813055277,
          "train_metric": 0.0003367489262018353,
          "validation_metric": 0.0003298324591014534,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8664562009998917
        },
        {
          "epoch": 74,
          "updates": 1184,
          "samples_seen": 75776,
          "online_loss": 0.003593408804590581,
          "train_loss": 0.004045362584292889,
          "validation_loss": 0.004012070596218109,
          "train_metric": 0.0004074340686202049,
          "validation_metric": 0.000404081103624776,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8775285289993917
        },
        {
          "epoch": 75,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.0027095369332528207,
          "train_loss": 0.0019463020144030452,
          "validation_loss": 0.0013752930099144578,
          "train_metric": 0.00019602442625910044,
          "validation_metric": 0.0001385144714731723,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8885188999993261
        },
        {
          "epoch": 76,
          "updates": 1216,
          "samples_seen": 77824,
          "online_loss": 0.004001264758699108,
          "train_loss": 0.003718052990734577,
          "validation_loss": 0.0027752479072660208,
          "train_metric": 0.0003744686837308109,
          "validation_metric": 0.0002795127802528441,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8993121699986659
        },
        {
          "epoch": 77,
          "updates": 1232,
          "samples_seen": 78848,
          "online_loss": 0.005240593753114808,
          "train_loss": 0.008636370301246643,
          "validation_loss": 0.009356452152132988,
          "train_metric": 0.00086982361972332,
          "validation_metric": 0.0009423475712537766,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9102277479996701
        },
        {
          "epoch": 78,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.006110473608714528,
          "train_loss": 0.006646636873483658,
          "validation_loss": 0.008010126650333405,
          "train_metric": 0.0006694248877465725,
          "validation_metric": 0.0008067506132647395,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9209552519987483
        },
        {
          "epoch": 79,
          "updates": 1264,
          "samples_seen": 80896,
          "online_loss": 0.003902298129105475,
          "train_loss": 0.0018766896100714803,
          "validation_loss": 0.0012020838912576437,
          "train_metric": 0.00018901328439824283,
          "validation_metric": 0.00012106947542633861,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.931554310998763
        },
        {
          "epoch": 80,
          "updates": 1280,
          "samples_seen": 81920,
          "online_loss": 0.0020461318617890356,
          "train_loss": 0.0015792911872267723,
          "validation_loss": 0.0011573329102247953,
          "train_metric": 0.00015906042244751006,
          "validation_metric": 0.000116562339826487,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.942072546999043
        },
        {
          "epoch": 81,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.002321494046555017,
          "train_loss": 0.006457476411014795,
          "validation_loss": 0.0064075966365635395,
          "train_metric": 0.0006503733457066119,
          "validation_metric": 0.0006453496171161532,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.9529811349984811
        },
        {
          "epoch": 82,
          "updates": 1312,
          "samples_seen": 83968,
          "online_loss": 0.003204850427209749,
          "train_loss": 0.002415082883089781,
          "validation_loss": 0.0018170818220824003,
          "train_metric": 0.00024323830439243466,
          "validation_metric": 0.00018300981901120394,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9635431459992105
        },
        {
          "epoch": 83,
          "updates": 1328,
          "samples_seen": 84992,
          "online_loss": 0.0031883777355687926,
          "train_loss": 0.001261584460735321,
          "validation_loss": 0.0008865423733368516,
          "train_metric": 0.00012706215784419328,
          "validation_metric": 8.928930765250698e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9749581909982226
        },
        {
          "epoch": 84,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.002409536733466666,
          "train_loss": 0.001181971631012857,
          "validation_loss": 0.0009367477032355964,
          "train_metric": 0.00011904386337846518,
          "validation_metric": 9.434579988010228e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9856884399996488
        },
        {
          "epoch": 85,
          "updates": 1360,
          "samples_seen": 87040,
          "online_loss": 0.001390066232488607,
          "train_loss": 0.0012464605970308185,
          "validation_loss": 0.0008673122501932085,
          "train_metric": 0.00012553896522149444,
          "validation_metric": 8.735251321922988e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9972077729999
        },
        {
          "epoch": 86,
          "updates": 1376,
          "samples_seen": 88064,
          "online_loss": 0.002419913191261003,
          "train_loss": 0.001348488382063806,
          "validation_loss": 0.001102409209124744,
          "train_metric": 0.0001358148001600057,
          "validation_metric": 0.00011103061842732131,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0080171959998552
        },
        {
          "epoch": 87,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0017408568201062735,
          "train_loss": 0.0011814890895038843,
          "validation_loss": 0.0009288333239965141,
          "train_metric": 0.00011899525998160243,
          "validation_metric": 9.354869689559564e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0184560879988567
        },
        {
          "epoch": 88,
          "updates": 1408,
          "samples_seen": 90112,
          "online_loss": 0.0017324551154160872,
          "train_loss": 0.0011425481643527746,
          "validation_loss": 0.0009018828277476132,
          "train_metric": 0.00011507324961712584,
          "validation_metric": 9.083432814804837e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0289086629982194
        },
        {
          "epoch": 89,
          "updates": 1424,
          "samples_seen": 91136,
          "online_loss": 0.0012147173438279424,
          "train_loss": 0.0013103218516334891,
          "validation_loss": 0.0009822866413742304,
          "train_metric": 0.00013197080988902599,
          "validation_metric": 9.893231617752463e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.039463100998546
        },
        {
          "epoch": 90,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0011784885446104454,
          "train_loss": 0.000960516685154289,
          "validation_loss": 0.0007663265569135547,
          "train_metric": 9.673971362644807e-05,
          "validation_metric": 7.718160486547276e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.050435662999007
        },
        {
          "epoch": 91,
          "updates": 1456,
          "samples_seen": 93184,
          "online_loss": 0.0010833909072971437,
          "train_loss": 0.0010734724346548319,
          "validation_loss": 0.000848552561365068,
          "train_metric": 0.0001081162117770873,
          "validation_metric": 8.546310709789395e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.060893160998603
        },
        {
          "epoch": 92,
          "updates": 1472,
          "samples_seen": 94208,
          "online_loss": 0.0012365112434054026,
          "train_loss": 0.0009756602230481803,
          "validation_loss": 0.0008622216410003603,
          "train_metric": 9.826493624132127e-05,
          "validation_metric": 8.683979831403121e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0715561429988156
        },
        {
          "epoch": 93,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0025003928276419174,
          "train_loss": 0.0013205186696723104,
          "validation_loss": 0.001320013077929616,
          "train_metric": 0.00013299781130626798,
          "validation_metric": 0.0001329468796029687,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.0818730989994947
        },
        {
          "epoch": 94,
          "updates": 1504,
          "samples_seen": 96256,
          "online_loss": 0.0016845814006956061,
          "train_loss": 0.0014334963634610176,
          "validation_loss": 0.0012157424353063107,
          "train_metric": 0.0001443765067961067,
          "validation_metric": 0.00012244514073245227,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.092531329999474
        },
        {
          "epoch": 95,
          "updates": 1520,
          "samples_seen": 97280,
          "online_loss": 0.0014987075119279325,
          "train_loss": 0.0009721279493533075,
          "validation_loss": 0.0006941115716472268,
          "train_metric": 9.790917101781815e-05,
          "validation_metric": 6.990836845943704e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1035058869983914
        },
        {
          "epoch": 96,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0016023708831198746,
          "train_loss": 0.0011439751833677292,
          "validation_loss": 0.0009673596941865981,
          "train_metric": 0.00011521699343575165,
          "validation_metric": 9.742892871145159e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.114146168998559
        },
        {
          "epoch": 97,
          "updates": 1552,
          "samples_seen": 99328,
          "online_loss": 0.001111484376451699,
          "train_loss": 0.0007552597671747208,
          "validation_loss": 0.0006622183136641979,
          "train_metric": 7.6067000918556e-05,
          "validation_metric": 6.669620051980019e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1252529489993321
        },
        {
          "epoch": 98,
          "updates": 1568,
          "samples_seen": 100352,
          "online_loss": 0.0009331957116955891,
          "train_loss": 0.0008972304640337825,
          "validation_loss": 0.0007863245555199683,
          "train_metric": 9.036575647769496e-05,
          "validation_metric": 7.919573545223102e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1357918149988109
        },
        {
          "epoch": 99,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0009515590008959407,
          "train_loss": 0.0007746305782347918,
          "validation_loss": 0.0006479760631918907,
          "train_metric": 7.80179543653503e-05,
          "validation_metric": 6.526177457999438e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.147135299999718
        },
        {
          "epoch": 100,
          "updates": 1600,
          "samples_seen": 102400,
          "online_loss": 0.0010599801080388715,
          "train_loss": 0.0018618018366396427,
          "validation_loss": 0.0014919815585017204,
          "train_metric": 0.00018751384050119668,
          "validation_metric": 0.0001502669183537364,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1579815659988526
        },
        {
          "epoch": 101,
          "updates": 1616,
          "samples_seen": 103424,
          "online_loss": 0.0008684580097906291,
          "train_loss": 0.0007200469262897968,
          "validation_loss": 0.0005225190543569624,
          "train_metric": 7.252048817463219e-05,
          "validation_metric": 5.262620106805116e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1692190999983723
        },
        {
          "epoch": 102,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0007888901418482419,
          "train_loss": 0.0010073018493130803,
          "validation_loss": 0.0008439203957095742,
          "train_metric": 0.00010145175474463031,
          "validation_metric": 8.499657269567251e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1799525779988471
        },
        {
          "epoch": 103,
          "updates": 1648,
          "samples_seen": 105472,
          "online_loss": 0.000710979684299673,
          "train_loss": 0.0007370702805928886,
          "validation_loss": 0.0006512306863442063,
          "train_metric": 7.423501665471122e-05,
          "validation_metric": 6.558956374647096e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.1904557579982793
        },
        {
          "epoch": 104,
          "updates": 1664,
          "samples_seen": 106496,
          "online_loss": 0.0008429464542132337,
          "train_loss": 0.0007028619293123484,
          "validation_loss": 0.0004955179756507277,
          "train_metric": 7.078967610141262e-05,
          "validation_metric": 4.990675370208919e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.20133758099837
        },
        {
          "epoch": 105,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0010533402873988962,
          "train_loss": 0.0013825575588271022,
          "validation_loss": 0.0009572035050950944,
          "train_metric": 0.00013924615632276982,
          "validation_metric": 9.6406030934304e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.212283363998722
        },
        {
          "epoch": 106,
          "updates": 1696,
          "samples_seen": 108544,
          "online_loss": 0.001499497830081964,
          "train_loss": 0.0007349012885242701,
          "validation_loss": 0.0006142314523458481,
          "train_metric": 7.401655602734536e-05,
          "validation_metric": 6.186313839862123e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2230238469983306
        },
        {
          "epoch": 107,
          "updates": 1712,
          "samples_seen": 109568,
          "online_loss": 0.0008827261553960852,
          "train_loss": 0.0007867153035476804,
          "validation_loss": 0.0006472402019426227,
          "train_metric": 7.923508383100852e-05,
          "validation_metric": 6.51876543997787e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2334814459991321
        },
        {
          "epoch": 108,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0007599805849167751,
          "train_loss": 0.0008101004641503096,
          "validation_loss": 0.0006061627063900232,
          "train_metric": 8.159034769050777e-05,
          "validation_metric": 6.105047941673547e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2438893639991875
        },
        {
          "epoch": 109,
          "updates": 1744,
          "samples_seen": 111616,
          "online_loss": 0.0023320287582464516,
          "train_loss": 0.002897440455853939,
          "validation_loss": 0.002189994091168046,
          "train_metric": 0.00029181959689594805,
          "validation_metric": 0.0002205682103522122,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2543114049985888
        },
        {
          "epoch": 110,
          "updates": 1760,
          "samples_seen": 112640,
          "online_loss": 0.001986946204851847,
          "train_loss": 0.004696398973464966,
          "validation_loss": 0.0042628576047718525,
          "train_metric": 0.00047300412552431226,
          "validation_metric": 0.00042933941585943103,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2646344899985706
        },
        {
          "epoch": 111,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.002292122386279516,
          "train_loss": 0.00269507453776896,
          "validation_loss": 0.001467134919948876,
          "train_metric": 0.0002714380680117756,
          "validation_metric": 0.00014776448369957507,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2749868219998461
        },
        {
          "epoch": 112,
          "updates": 1792,
          "samples_seen": 114688,
          "online_loss": 0.001585295156473876,
          "train_loss": 0.002505150157958269,
          "validation_loss": 0.002000041538849473,
          "train_metric": 0.0002523095754440874,
          "validation_metric": 0.00020143689471296966,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.285433318998912
        },
        {
          "epoch": 113,
          "updates": 1808,
          "samples_seen": 115712,
          "online_loss": 0.0019906311317754444,
          "train_loss": 0.0011033102637156844,
          "validation_loss": 0.0008188696810975671,
          "train_metric": 0.00011112137144664302,
          "validation_metric": 8.247356163337827e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.2958432149989676
        },
        {
          "epoch": 114,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.0008147057196765672,
          "train_loss": 0.0007392544648610055,
          "validation_loss": 0.0005941683775745332,
          "train_metric": 7.445500523317605e-05,
          "validation_metric": 5.984247036394663e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3066842869993707
        },
        {
          "epoch": 115,
          "updates": 1840,
          "samples_seen": 117760,
          "online_loss": 0.0010260439630656037,
          "train_loss": 0.002080790465697646,
          "validation_loss": 0.0016173674957826734,
          "train_metric": 0.00020956959633622319,
          "validation_metric": 0.00016289533232338727,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 1.3175874699991255
        },
        {
          "epoch": 116,
          "updates": 1856,
          "samples_seen": 118784,
          "online_loss": 0.0013931800640420988,
          "train_loss": 0.0025870027020573616,
          "validation_loss": 0.00205914955586195,
          "train_metric": 0.0002605534391477704,
          "validation_metric": 0.0002073900104733184,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3279459889999998
        },
        {
          "epoch": 117,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.0011363845806044992,
          "train_loss": 0.0006742316763848066,
          "validation_loss": 0.0005440751556307077,
          "train_metric": 6.79061486152932e-05,
          "validation_metric": 5.479726314661093e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3383766129991272
        },
        {
          "epoch": 118,
          "updates": 1888,
          "samples_seen": 120832,
          "online_loss": 0.000714739719114732,
          "train_loss": 0.0006696152267977595,
          "validation_loss": 0.000547104689758271,
          "train_metric": 6.744118581991643e-05,
          "validation_metric": 5.510238770511933e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3486693989998457
        },
        {
          "epoch": 119,
          "updates": 1904,
          "samples_seen": 121856,
          "online_loss": 0.000726640264474554,
          "train_loss": 0.0005994715029373765,
          "validation_loss": 0.000443046708824113,
          "train_metric": 6.037657294655219e-05,
          "validation_metric": 4.4622043787967414e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3596276139996917
        },
        {
          "epoch": 120,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.000699035513207491,
          "train_loss": 0.0006095705321058631,
          "validation_loss": 0.0004978777142241597,
          "train_metric": 6.139372271718457e-05,
          "validation_metric": 5.014442285755649e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3703141509995476
        },
        {
          "epoch": 121,
          "updates": 1936,
          "samples_seen": 123904,
          "online_loss": 0.0006894093639857601,
          "train_loss": 0.0005615675472654402,
          "validation_loss": 0.0004286050098016858,
          "train_metric": 5.6559027143521234e-05,
          "validation_metric": 4.316752892918885e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3813012399987201
        },
        {
          "epoch": 122,
          "updates": 1952,
          "samples_seen": 124928,
          "online_loss": 0.0009189539723593043,
          "train_loss": 0.0007903330842964351,
          "validation_loss": 0.0006748268497176468,
          "train_metric": 7.959945651236922e-05,
          "validation_metric": 6.796608795411885e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.391904648999116
        },
        {
          "epoch": 123,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0015095192238732125,
          "train_loss": 0.0013989565195515752,
          "validation_loss": 0.0014995195670053363,
          "train_metric": 0.00014089775504544377,
          "validation_metric": 0.00015102613542694598,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4023788839986082
        },
        {
          "epoch": 124,
          "updates": 1984,
          "samples_seen": 126976,
          "online_loss": 0.0016194925956369843,
          "train_loss": 0.0017759758047759533,
          "validation_loss": 0.0012059382861480117,
          "train_metric": 0.00017886979912873358,
          "validation_metric": 0.00012145769142080098,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4127398329983407
        },
        {
          "epoch": 125,
          "updates": 2000,
          "samples_seen": 128000,
          "online_loss": 0.0013028637367824558,
          "train_loss": 0.0005991756916046143,
          "validation_loss": 0.0004957999917678535,
          "train_metric": 6.034678153810091e-05,
          "validation_metric": 4.993515904061496e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4231793439994362
        },
        {
          "epoch": 126,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.0008587142992837471,
          "train_loss": 0.0006913767429068685,
          "validation_loss": 0.0007160301320254803,
          "train_metric": 6.96329225320369e-05,
          "validation_metric": 7.211593037936836e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4335403789991688
        },
        {
          "epoch": 127,
          "updates": 2032,
          "samples_seen": 130048,
          "online_loss": 0.0008113021776807727,
          "train_loss": 0.0007229626644402742,
          "validation_loss": 0.0005817274213768542,
          "train_metric": 7.281415309989825e-05,
          "validation_metric": 5.8589463151292875e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4440562169984332
        },
        {
          "epoch": 128,
          "updates": 2048,
          "samples_seen": 131072,
          "online_loss": 0.0005925541572651127,
          "train_loss": 0.0005365508841350675,
          "validation_loss": 0.00039055911474861205,
          "train_metric": 5.403943941928446e-05,
          "validation_metric": 3.9335685869446024e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4549925779992918
        },
        {
          "epoch": 129,
          "updates": 2064,
          "samples_seen": 132096,
          "online_loss": 0.0006207470269146143,
          "train_loss": 0.0008947577443905175,
          "validation_loss": 0.0007931837462820113,
          "train_metric": 9.011672955239192e-05,
          "validation_metric": 7.988657307578251e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4654908729989984
        },
        {
          "epoch": 130,
          "updates": 2080,
          "samples_seen": 133120,
          "online_loss": 0.0009856000460786163,
          "train_loss": 0.0006760033429600298,
          "validation_loss": 0.0006067940266802907,
          "train_metric": 6.808457692386582e-05,
          "validation_metric": 6.111407856224105e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4761261659987213
        },
        {
          "epoch": 131,
          "updates": 2096,
          "samples_seen": 134144,
          "online_loss": 0.001077648053978919,
          "train_loss": 0.0011006571585312486,
          "validation_loss": 0.0006227483972907066,
          "train_metric": 0.00011085417645517737,
          "validation_metric": 6.272093742154539e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4867619069991633
        },
        {
          "epoch": 132,
          "updates": 2112,
          "samples_seen": 135168,
          "online_loss": 0.0010348358591727447,
          "train_loss": 0.001040260773152113,
          "validation_loss": 0.000762815005145967,
          "train_metric": 0.00010477126488694921,
          "validation_metric": 7.682792056584731e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.497243844998593
        },
        {
          "epoch": 133,
          "updates": 2128,
          "samples_seen": 136192,
          "online_loss": 0.0006579512391908793,
          "train_loss": 0.0005078170797787607,
          "validation_loss": 0.00042929136543534696,
          "train_metric": 5.1145478209946305e-05,
          "validation_metric": 4.3236657802481204e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.508031266999751
        },
        {
          "epoch": 134,
          "updates": 2144,
          "samples_seen": 137216,
          "online_loss": 0.000606933069320803,
          "train_loss": 0.0006670717848464847,
          "validation_loss": 0.0004781965399160981,
          "train_metric": 6.71850357321091e-05,
          "validation_metric": 4.8162204620894045e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5187116249999235
        },
        {
          "epoch": 135,
          "updates": 2160,
          "samples_seen": 138240,
          "online_loss": 0.0005396958949859254,
          "train_loss": 0.0004736183909699321,
          "validation_loss": 0.00040575317689217627,
          "train_metric": 4.770111263496801e-05,
          "validation_metric": 4.0865976188797504e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5290115079988027
        },
        {
          "epoch": 136,
          "updates": 2176,
          "samples_seen": 139264,
          "online_loss": 0.0004927599802613258,
          "train_loss": 0.00046927310177125037,
          "validation_loss": 0.0003673181345220655,
          "train_metric": 4.7263471060432494e-05,
          "validation_metric": 3.699494118336588e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5400440399989748
        },
        {
          "epoch": 137,
          "updates": 2192,
          "samples_seen": 140288,
          "online_loss": 0.0005609894415101735,
          "train_loss": 0.0005332434084266424,
          "validation_loss": 0.00046150796697475016,
          "train_metric": 5.3706320613855496e-05,
          "validation_metric": 4.648139656637795e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5505390539983637
        },
        {
          "epoch": 138,
          "updates": 2208,
          "samples_seen": 141312,
          "online_loss": 0.0009738629332787241,
          "train_loss": 0.0008304921793751419,
          "validation_loss": 0.0005294853472150862,
          "train_metric": 8.364413952222094e-05,
          "validation_metric": 5.332782893674448e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5611900559997594
        },
        {
          "epoch": 139,
          "updates": 2224,
          "samples_seen": 142336,
          "online_loss": 0.0008640666528663132,
          "train_loss": 0.000529735698364675,
          "validation_loss": 0.0004919539205729961,
          "train_metric": 5.335303285392001e-05,
          "validation_metric": 4.954779433319345e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.571949245999349
        },
        {
          "epoch": 140,
          "updates": 2240,
          "samples_seen": 143360,
          "online_loss": 0.0007753005484119058,
          "train_loss": 0.0006802050047554076,
          "validation_loss": 0.0005046048318035901,
          "train_metric": 6.850775389466435e-05,
          "validation_metric": 5.082194911665283e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5822298849998333
        },
        {
          "epoch": 141,
          "updates": 2256,
          "samples_seen": 144384,
          "online_loss": 0.0009336260773125105,
          "train_loss": 0.00047524814726784825,
          "validation_loss": 0.0003975180152337998,
          "train_metric": 4.7865250962786376e-05,
          "validation_metric": 4.0036553400568664e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5926514599996153
        },
        {
          "epoch": 142,
          "updates": 2272,
          "samples_seen": 145408,
          "online_loss": 0.0005038759554736316,
          "train_loss": 0.00046782390563748777,
          "validation_loss": 0.00040473323315382004,
          "train_metric": 4.711750807473436e-05,
          "validation_metric": 4.0763250581221655e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.603733342999476
        },
        {
          "epoch": 143,
          "updates": 2288,
          "samples_seen": 146432,
          "online_loss": 0.000945818953368871,
          "train_loss": 0.0008276710286736488,
          "validation_loss": 0.0005718684988096356,
          "train_metric": 8.335999154951423e-05,
          "validation_metric": 5.759651321568526e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.614061298998422
        },
        {
          "epoch": 144,
          "updates": 2304,
          "samples_seen": 147456,
          "online_loss": 0.0010849764021259034,
          "train_loss": 0.00048262818017974496,
          "validation_loss": 0.00037242466351017356,
          "train_metric": 4.860854824073613e-05,
          "validation_metric": 3.750924952328205e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6244276299985358
        },
        {
          "epoch": 145,
          "updates": 2320,
          "samples_seen": 148480,
          "online_loss": 0.0005612868144453387,
          "train_loss": 0.0004631209303624928,
          "validation_loss": 0.0003701538371387869,
          "train_metric": 4.664385051000863e-05,
          "validation_metric": 3.7280533433659e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6348030159988411
        },
        {
          "epoch": 146,
          "updates": 2336,
          "samples_seen": 149504,
          "online_loss": 0.0011360869084455771,
          "train_loss": 0.0013377949362620711,
          "validation_loss": 0.00128399848472327,
          "train_metric": 0.00013473779836203903,
          "validation_metric": 0.00012931961100548506,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6452945499986527
        },
        {
          "epoch": 147,
          "updates": 2352,
          "samples_seen": 150528,
          "online_loss": 0.0015311713959818007,
          "train_loss": 0.0012828262988477945,
          "validation_loss": 0.0009347639279440045,
          "train_metric": 0.00012920155131723732,
          "validation_metric": 9.41460020840168e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6557513559982908
        },
        {
          "epoch": 148,
          "updates": 2368,
          "samples_seen": 151552,
          "online_loss": 0.0007702884449827252,
          "train_loss": 0.0005837542121298611,
          "validation_loss": 0.0004150537424720824,
          "train_metric": 5.8793586504179984e-05,
          "validation_metric": 4.180269024800509e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.666154712998832
        },
        {
          "epoch": 149,
          "updates": 2384,
          "samples_seen": 152576,
          "online_loss": 0.0005370584503907594,
          "train_loss": 0.0004953020252287388,
          "validation_loss": 0.0003903490141965449,
          "train_metric": 4.9885005864780396e-05,
          "validation_metric": 3.931452374672517e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6766173869982595
        },
        {
          "epoch": 150,
          "updates": 2400,
          "samples_seen": 153600,
          "online_loss": 0.0007483099707314977,
          "train_loss": 0.0010151524329558015,
          "validation_loss": 0.0006455963593907654,
          "train_metric": 0.000102242425782606,
          "validation_metric": 6.502209726022556e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6870235859987588
        },
        {
          "epoch": 151,
          "updates": 2416,
          "samples_seen": 154624,
          "online_loss": 0.0006709594654239481,
          "train_loss": 0.0004956838674843311,
          "validation_loss": 0.0003564115904737264,
          "train_metric": 4.992346657672897e-05,
          "validation_metric": 3.5896471672458574e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6980193169983977
        },
        {
          "epoch": 152,
          "updates": 2432,
          "samples_seen": 155648,
          "online_loss": 0.0009004975090647349,
          "train_loss": 0.0004513422609306872,
          "validation_loss": 0.0003502455074340105,
          "train_metric": 4.545753836282529e-05,
          "validation_metric": 3.5275446862215176e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.709462712999084
        },
        {
          "epoch": 153,
          "updates": 2448,
          "samples_seen": 156672,
          "online_loss": 0.0006481930049631046,
          "train_loss": 0.0008068275637924671,
          "validation_loss": 0.00048712673014961183,
          "train_metric": 8.126072498271242e-05,
          "validation_metric": 4.9061614845413715e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.720233598998675
        },
        {
          "epoch": 154,
          "updates": 2464,
          "samples_seen": 157696,
          "online_loss": 0.0005660286406055093,
          "train_loss": 0.0005851404275745153,
          "validation_loss": 0.0004207146994303912,
          "train_metric": 5.893320121685974e-05,
          "validation_metric": 4.237284520058893e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7312852649993147
        },
        {
          "epoch": 155,
          "updates": 2480,
          "samples_seen": 158720,
          "online_loss": 0.0006385936303558992,
          "train_loss": 0.0006696440977975726,
          "validation_loss": 0.0005013167392462492,
          "train_metric": 6.744410347891971e-05,
          "validation_metric": 5.049079118180089e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.742328884998642
        },
        {
          "epoch": 156,
          "updates": 2496,
          "samples_seen": 159744,
          "online_loss": 0.0005787678583146771,
          "train_loss": 0.00042351576848886907,
          "validation_loss": 0.0003361297131050378,
          "train_metric": 4.265495590516366e-05,
          "validation_metric": 3.385375384823419e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7538155919992278
        },
        {
          "epoch": 157,
          "updates": 2512,
          "samples_seen": 160768,
          "online_loss": 0.0005533422490771045,
          "train_loss": 0.0006642754888162017,
          "validation_loss": 0.0004976690979674459,
          "train_metric": 6.690339796477929e-05,
          "validation_metric": 5.012340625398792e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7647797699992225
        },
        {
          "epoch": 158,
          "updates": 2528,
          "samples_seen": 161792,
          "online_loss": 0.00048564630378677975,
          "train_loss": 0.0004120109078940004,
          "validation_loss": 0.0003079709713347256,
          "train_metric": 4.149623055127449e-05,
          "validation_metric": 3.101770926150493e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.77649019799901
        },
        {
          "epoch": 159,
          "updates": 2544,
          "samples_seen": 162816,
          "online_loss": 0.00043557440039876383,
          "train_loss": 0.0003972981357946992,
          "validation_loss": 0.0003044064505957067,
          "train_metric": 4.0014412661548704e-05,
          "validation_metric": 3.065870623686351e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.788308284998493
        },
        {
          "epoch": 160,
          "updates": 2560,
          "samples_seen": 163840,
          "online_loss": 0.00042027176641568076,
          "train_loss": 0.0003937441506423056,
          "validation_loss": 0.0003121639310847968,
          "train_metric": 3.9656464650761336e-05,
          "validation_metric": 3.144000220345333e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7990917679999257
        },
        {
          "epoch": 161,
          "updates": 2576,
          "samples_seen": 164864,
          "online_loss": 0.0004471573001865181,
          "train_loss": 0.0005128788761794567,
          "validation_loss": 0.00045719838817603886,
          "train_metric": 5.165527545614168e-05,
          "validation_metric": 4.6047356590861455e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8098619589982263
        },
        {
          "epoch": 162,
          "updates": 2592,
          "samples_seen": 165888,
          "online_loss": 0.0005911839343752945,
          "train_loss": 0.0008396205957978964,
          "validation_loss": 0.000600555504206568,
          "train_metric": 8.456351497443393e-05,
          "validation_metric": 6.0485755966510624e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.820491285998287
        },
        {
          "epoch": 163,
          "updates": 2608,
          "samples_seen": 166912,
          "online_loss": 0.0007345873282247339,
          "train_loss": 0.000823371927253902,
          "validation_loss": 0.0006545680807903409,
          "train_metric": 8.29270138638094e-05,
          "validation_metric": 6.592571298824623e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8309509079990676
        },
        {
          "epoch": 164,
          "updates": 2624,
          "samples_seen": 167936,
          "online_loss": 0.0006214027580426773,
          "train_loss": 0.00041622709250077605,
          "validation_loss": 0.0003520985774230212,
          "train_metric": 4.1920869989553466e-05,
          "validation_metric": 3.546207517501898e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.841533367998636
        },
        {
          "epoch": 165,
          "updates": 2640,
          "samples_seen": 168960,
          "online_loss": 0.0005430116298157373,
          "train_loss": 0.00045219185994938016,
          "validation_loss": 0.0003464340406935662,
          "train_metric": 4.5543107262346894e-05,
          "validation_metric": 3.489156370051205e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8518747219986835
        },
        {
          "epoch": 166,
          "updates": 2656,
          "samples_seen": 169984,
          "online_loss": 0.0004981085967301624,
          "train_loss": 0.0007995185442268848,
          "validation_loss": 0.0006866967887617648,
          "train_metric": 8.05245726951398e-05,
          "validation_metric": 6.91615859977901e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8621752719991491
        },
        {
          "epoch": 167,
          "updates": 2672,
          "samples_seen": 171008,
          "online_loss": 0.0005443296240628115,
          "train_loss": 0.0003774147480726242,
          "validation_loss": 0.0003044845652766526,
          "train_metric": 3.801183629548177e-05,
          "validation_metric": 3.066657154704444e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8725537529990106
        },
        {
          "epoch": 168,
          "updates": 2688,
          "samples_seen": 172032,
          "online_loss": 0.0004987593783880584,
          "train_loss": 0.0005878182710148394,
          "validation_loss": 0.0003850880602840334,
          "train_metric": 5.920289913774468e-05,
          "validation_metric": 3.8784663047408685e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.883087356998658
        },
        {
          "epoch": 169,
          "updates": 2704,
          "samples_seen": 173056,
          "online_loss": 0.0004682969520217739,
          "train_loss": 0.0003782364074140787,
          "validation_loss": 0.0003154903242830187,
          "train_metric": 3.809458576142788e-05,
          "validation_metric": 3.177503094775602e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8940173279988812
        },
        {
          "epoch": 170,
          "updates": 2720,
          "samples_seen": 174080,
          "online_loss": 0.00042609466800058726,
          "train_loss": 0.0004531600570771843,
          "validation_loss": 0.0003786420274991542,
          "train_metric": 4.564061600831337e-05,
          "validation_metric": 3.813544026343152e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.904572413999631
        },
        {
          "epoch": 171,
          "updates": 2736,
          "samples_seen": 175104,
          "online_loss": 0.00043618015206448035,
          "train_loss": 0.0004061362415086478,
          "validation_loss": 0.00034454429987818,
          "train_metric": 4.090455695404671e-05,
          "validation_metric": 3.4701246477197856e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.91518582199933
        },
        {
          "epoch": 172,
          "updates": 2752,
          "samples_seen": 176128,
          "online_loss": 0.00038705534552718746,
          "train_loss": 0.0003860268625430763,
          "validation_loss": 0.0003176046593580395,
          "train_metric": 3.8879214116605e-05,
          "validation_metric": 3.198797639925033e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.9257174279991887
        },
        {
          "epoch": 173,
          "updates": 2768,
          "samples_seen": 177152,
          "online_loss": 0.00038788578149251407,
          "train_loss": 0.00036845181602984667,
          "validation_loss": 0.00029599666595458984,
          "train_metric": 3.710911187226884e-05,
          "validation_metric": 2.98117029160494e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.936577024998769
        },
        {
          "epoch": 174,
          "updates": 2784,
          "samples_seen": 178176,
          "online_loss": 0.0003742994495041785,
          "train_loss": 0.0003607433463912457,
          "validation_loss": 0.00030182398040778935,
          "train_metric": 3.6332749004941434e-05,
          "validation_metric": 3.0398603485082276e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.947147072000007
        },
        {
          "epoch": 175,
          "updates": 2800,
          "samples_seen": 179200,
          "online_loss": 0.0003831854255622602,
          "train_loss": 0.0004304691683501005,
          "validation_loss": 0.00034296681405976415,
          "train_metric": 4.3355281377444044e-05,
          "validation_metric": 3.4542357752798125e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.9577774749996024
        },
        {
          "epoch": 176,
          "updates": 2816,
          "samples_seen": 180224,
          "online_loss": 0.00040377465575147653,
          "train_loss": 0.0003839514101855457,
          "validation_loss": 0.00032372953137382865,
          "train_metric": 3.867018313030712e-05,
          "validation_metric": 3.260485391365364e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.9682293809983094
        },
        {
          "epoch": 177,
          "updates": 2832,
          "samples_seen": 181248,
          "online_loss": 0.0003805406122410204,
          "train_loss": 0.00041513118776492774,
          "validation_loss": 0.0003306072030682117,
          "train_metric": 4.18104900745675e-05,
          "validation_metric": 3.329754690639675e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.9786885989997245
        },
        {
          "epoch": 178,
          "updates": 2848,
          "samples_seen": 182272,
          "online_loss": 0.00035288317758386256,
          "train_loss": 0.00038229173514992,
          "validation_loss": 0.00031757738906890154,
          "train_metric": 3.8503021642100066e-05,
          "validation_metric": 3.198523336322978e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.9892502189995867
        },
        {
          "epoch": 179,
          "updates": 2864,
          "samples_seen": 183296,
          "online_loss": 0.0003871398585033603,
          "train_loss": 0.00039162556640803814,
          "validation_loss": 0.00033871171763166785,
          "train_metric": 3.944308991776779e-05,
          "validation_metric": 3.411380748730153e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 1.9997892919982405
        },
        {
          "epoch": 180,
          "updates": 2880,
          "samples_seen": 184320,
          "online_loss": 0.0003805838987318566,
          "train_loss": 0.00035200314596295357,
          "validation_loss": 0.00029398815240710974,
          "train_metric": 3.545246727298945e-05,
          "validation_metric": 2.9609409466502257e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.010749680999652
        },
        {
          "epoch": 181,
          "updates": 2896,
          "samples_seen": 185344,
          "online_loss": 0.00036845328668277944,
          "train_loss": 0.0003470314550213516,
          "validation_loss": 0.0002906238369178027,
          "train_metric": 3.495173586998135e-05,
          "validation_metric": 2.9270569939399138e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0216984339986084
        },
        {
          "epoch": 182,
          "updates": 2912,
          "samples_seen": 186368,
          "online_loss": 0.0003569741484170663,
          "train_loss": 0.00034633715404197574,
          "validation_loss": 0.0002855377388186753,
          "train_metric": 3.488181027933024e-05,
          "validation_metric": 2.8758317057508975e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0327033259982272
        },
        {
          "epoch": 183,
          "updates": 2928,
          "samples_seen": 187392,
          "online_loss": 0.0003653632238638238,
          "train_loss": 0.0003448373463470489,
          "validation_loss": 0.0002897440572269261,
          "train_metric": 3.4730754123302177e-05,
          "validation_metric": 2.9181963327573612e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.043372719999752
        },
        {
          "epoch": 184,
          "updates": 2944,
          "samples_seen": 188416,
          "online_loss": 0.0003668381232273532,
          "train_loss": 0.00038811040576547384,
          "validation_loss": 0.00030609112582169473,
          "train_metric": 3.908905273419805e-05,
          "validation_metric": 3.082837429246865e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0536967089992686
        },
        {
          "epoch": 185,
          "updates": 2960,
          "samples_seen": 189440,
          "online_loss": 0.0003705934750541928,
          "train_loss": 0.00034400628646835685,
          "validation_loss": 0.0002883311826735735,
          "train_metric": 3.464705514488742e-05,
          "validation_metric": 2.903966378653422e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0639328099987324
        },
        {
          "epoch": 186,
          "updates": 2976,
          "samples_seen": 190464,
          "online_loss": 0.000358027868969657,
          "train_loss": 0.0003427918709348887,
          "validation_loss": 0.00028842175379395485,
          "train_metric": 3.4524746297392994e-05,
          "validation_metric": 2.9048780561424792e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.074459768999077
        },
        {
          "epoch": 187,
          "updates": 2992,
          "samples_seen": 191488,
          "online_loss": 0.00035015714456676506,
          "train_loss": 0.00034095620503649116,
          "validation_loss": 0.0002925486187450588,
          "train_metric": 3.4339860576437786e-05,
          "validation_metric": 2.9464428735082038e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.084750821999478
        },
        {
          "epoch": 188,
          "updates": 3008,
          "samples_seen": 192512,
          "online_loss": 0.0003717328590937541,
          "train_loss": 0.0003448535571806133,
          "validation_loss": 0.00028974839369766414,
          "train_metric": 3.473238757578656e-05,
          "validation_metric": 2.9182399885030463e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.095032089999222
        },
        {
          "epoch": 189,
          "updates": 3024,
          "samples_seen": 193536,
          "online_loss": 0.00037997526305844076,
          "train_loss": 0.00035468279384076595,
          "validation_loss": 0.0002983494778163731,
          "train_metric": 3.572235073079355e-05,
          "validation_metric": 3.0048668122617528e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.105547047998698
        },
        {
          "epoch": 190,
          "updates": 3040,
          "samples_seen": 194560,
          "online_loss": 0.00036374804403749295,
          "train_loss": 0.0006160202901810408,
          "validation_loss": 0.0004455654416233301,
          "train_metric": 6.204330566106364e-05,
          "validation_metric": 4.487572005018592e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.11599406699861
        },
        {
          "epoch": 191,
          "updates": 3056,
          "samples_seen": 195584,
          "online_loss": 0.0004721450022771023,
          "train_loss": 0.00035751413088291883,
          "validation_loss": 0.0003169221745338291,
          "train_metric": 3.6007513699587435e-05,
          "validation_metric": 3.191924042766914e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.126359183999739
        },
        {
          "epoch": 192,
          "updates": 3072,
          "samples_seen": 196608,
          "online_loss": 0.00035820324046653695,
          "train_loss": 0.0004252365615684539,
          "validation_loss": 0.00039024968282319605,
          "train_metric": 4.282827285351232e-05,
          "validation_metric": 3.930451930500567e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1367111969993857
        },
        {
          "epoch": 193,
          "updates": 3088,
          "samples_seen": 197632,
          "online_loss": 0.00037599009101541014,
          "train_loss": 0.00033787806751206517,
          "validation_loss": 0.0002902658306993544,
          "train_metric": 3.402984293643385e-05,
          "validation_metric": 2.9234513931442052e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.147046200998375
        },
        {
          "epoch": 194,
          "updates": 3104,
          "samples_seen": 198656,
          "online_loss": 0.00035646352625917643,
          "train_loss": 0.00034618392237462103,
          "validation_loss": 0.00028491095872595906,
          "train_metric": 3.486637433525175e-05,
          "validation_metric": 2.8695183573290706e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1579095119996055
        },
        {
          "epoch": 195,
          "updates": 3120,
          "samples_seen": 199680,
          "online_loss": 0.0003508990780574095,
          "train_loss": 0.00033606382203288376,
          "validation_loss": 0.00029277519206516445,
          "train_metric": 3.384711453691125e-05,
          "validation_metric": 2.948724795714952e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.168339122999896
        },
        {
          "epoch": 196,
          "updates": 3136,
          "samples_seen": 200704,
          "online_loss": 0.000341857502462517,
          "train_loss": 0.00033858270035125315,
          "validation_loss": 0.0002810624719131738,
          "train_metric": 3.41008126270026e-05,
          "validation_metric": 2.8307582397246733e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1792936339988955
        },
        {
          "epoch": 197,
          "updates": 3152,
          "samples_seen": 201728,
          "online_loss": 0.00035830794877256267,
          "train_loss": 0.00032971007749438286,
          "validation_loss": 0.00028371921507641673,
          "train_metric": 3.320719406474382e-05,
          "validation_metric": 2.857516119547654e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1897360309994838
        },
        {
          "epoch": 198,
          "updates": 3168,
          "samples_seen": 202752,
          "online_loss": 0.0003608262413763441,
          "train_loss": 0.00033694604644551873,
          "validation_loss": 0.00028370440122671425,
          "train_metric": 3.3935972169274464e-05,
          "validation_metric": 2.8573669624165632e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.200109835999683
        },
        {
          "epoch": 199,
          "updates": 3184,
          "samples_seen": 203776,
          "online_loss": 0.0003329938381284592,
          "train_loss": 0.00033036223612725735,
          "validation_loss": 0.0002812645398080349,
          "train_metric": 3.3272866858169436e-05,
          "validation_metric": 2.83279314317042e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.210728604999531
        },
        {
          "epoch": 200,
          "updates": 3200,
          "samples_seen": 204800,
          "online_loss": 0.00033863157841551583,
          "train_loss": 0.0003271506866440177,
          "validation_loss": 0.00027439850964583457,
          "train_metric": 3.294941416243091e-05,
          "validation_metric": 2.7636411687126383e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.2216653509985917
        },
        {
          "epoch": 201,
          "updates": 3216,
          "samples_seen": 205824,
          "online_loss": 0.0003613564049373963,
          "train_loss": 0.0003503981570247561,
          "validation_loss": 0.0003054361732210964,
          "train_metric": 3.529081732267514e-05,
          "validation_metric": 3.076241409871727e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.232043686999532
        },
        {
          "epoch": 202,
          "updates": 3232,
          "samples_seen": 206848,
          "online_loss": 0.0003534037914505461,
          "train_loss": 0.0003285643470007926,
          "validation_loss": 0.00028213229961693287,
          "train_metric": 3.309179737698287e-05,
          "validation_metric": 2.8415330234565772e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.242576365999412
        },
        {
          "epoch": 203,
          "updates": 3248,
          "samples_seen": 207872,
          "online_loss": 0.00034584171135065844,
          "train_loss": 0.00033882088609971106,
          "validation_loss": 0.00027868844335898757,
          "train_metric": 3.412479782127775e-05,
          "validation_metric": 2.8068479878129438e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.31529945299917
        },
        {
          "epoch": 204,
          "updates": 3264,
          "samples_seen": 208896,
          "online_loss": 0.0003491957268124679,
          "train_loss": 0.00034122608485631645,
          "validation_loss": 0.00028471730183809996,
          "train_metric": 3.436703991610557e-05,
          "validation_metric": 2.86756876448635e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3263754489998973
        },
        {
          "epoch": 205,
          "updates": 3280,
          "samples_seen": 209920,
          "online_loss": 0.0003388841496416717,
          "train_loss": 0.00032751221442595124,
          "validation_loss": 0.00028702145209535956,
          "train_metric": 3.298583396826871e-05,
          "validation_metric": 2.8907745218020864e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3369008029985707
        },
        {
          "epoch": 206,
          "updates": 3296,
          "samples_seen": 210944,
          "online_loss": 0.00036009566429129336,
          "train_loss": 0.0003371155180502683,
          "validation_loss": 0.0002805476833600551,
          "train_metric": 3.395304156583734e-05,
          "validation_metric": 2.8255733923288062e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3473006039985194
        },
        {
          "epoch": 207,
          "updates": 3312,
          "samples_seen": 211968,
          "online_loss": 0.00033929487381101353,
          "train_loss": 0.00032389455009251833,
          "validation_loss": 0.00027900704299099743,
          "train_metric": 3.262147220084444e-05,
          "validation_metric": 2.8100565032218583e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.357923524999933
        },
        {
          "epoch": 208,
          "updates": 3328,
          "samples_seen": 212992,
          "online_loss": 0.0003414490684008342,
          "train_loss": 0.00034779153065755963,
          "validation_loss": 0.0003101510810665786,
          "train_metric": 3.502828985801898e-05,
          "validation_metric": 3.123728674836457e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3682906169997295
        },
        {
          "epoch": 209,
          "updates": 3344,
          "samples_seen": 214016,
          "online_loss": 0.00037752490698039765,
          "train_loss": 0.0003540367470122874,
          "validation_loss": 0.0002878240484278649,
          "train_metric": 3.5657285479828715e-05,
          "validation_metric": 2.8988586564082652e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.378597186998377
        },
        {
          "epoch": 210,
          "updates": 3360,
          "samples_seen": 215040,
          "online_loss": 0.00032694958008505637,
          "train_loss": 0.00032493629259988666,
          "validation_loss": 0.00028570802533067763,
          "train_metric": 3.272639514761977e-05,
          "validation_metric": 2.8775466489605606e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.388933805999841
        },
        {
          "epoch": 211,
          "updates": 3376,
          "samples_seen": 216064,
          "online_loss": 0.00034163308646384394,
          "train_loss": 0.00032465136609971523,
          "validation_loss": 0.00026979678659699857,
          "train_metric": 3.269769513281062e-05,
          "validation_metric": 2.7172945920028724e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3999884899985773
        },
        {
          "epoch": 212,
          "updates": 3392,
          "samples_seen": 217088,
          "online_loss": 0.0003464586839072581,
          "train_loss": 0.0003325994766782969,
          "validation_loss": 0.0002850258897524327,
          "train_metric": 3.34982032654807e-05,
          "validation_metric": 2.8706766897812486e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4105190499994933
        },
        {
          "epoch": 213,
          "updates": 3408,
          "samples_seen": 218112,
          "online_loss": 0.00034477428926038556,
          "train_loss": 0.00033490030909888446,
          "validation_loss": 0.0002992725057993084,
          "train_metric": 3.3729935239534825e-05,
          "validation_metric": 3.0141631214064546e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4211087279982166
        },
        {
          "epoch": 214,
          "updates": 3424,
          "samples_seen": 219136,
          "online_loss": 0.00036395677580003394,
          "train_loss": 0.0003955359279643744,
          "validation_loss": 0.00030799192609265447,
          "train_metric": 3.983693022746593e-05,
          "validation_metric": 3.1019819289213046e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4319813219990465
        },
        {
          "epoch": 215,
          "updates": 3440,
          "samples_seen": 220160,
          "online_loss": 0.0003637172485468909,
          "train_loss": 0.00033946134499274194,
          "validation_loss": 0.00029862605151720345,
          "train_metric": 3.418929918552749e-05,
          "validation_metric": 3.0076520488364622e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.443367293999472
        },
        {
          "epoch": 216,
          "updates": 3456,
          "samples_seen": 221184,
          "online_loss": 0.0003453317731327843,
          "train_loss": 0.0003265762934461236,
          "validation_loss": 0.0002872067561838776,
          "train_metric": 3.2891570299398154e-05,
          "validation_metric": 2.8926409868290648e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.454041289998713
        },
        {
          "epoch": 217,
          "updates": 3472,
          "samples_seen": 222208,
          "online_loss": 0.0003766615209315205,
          "train_loss": 0.0003248974389862269,
          "validation_loss": 0.0002694205613806844,
          "train_metric": 3.272247704444453e-05,
          "validation_metric": 2.713505455176346e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4652115069984575
        },
        {
          "epoch": 218,
          "updates": 3488,
          "samples_seen": 223232,
          "online_loss": 0.0003424898759476491,
          "train_loss": 0.00032872549491003156,
          "validation_loss": 0.00027292314916849136,
          "train_metric": 3.310803003842011e-05,
          "validation_metric": 2.748782208072953e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4757862179994845
        },
        {
          "epoch": 219,
          "updates": 3504,
          "samples_seen": 224256,
          "online_loss": 0.0003638399357441813,
          "train_loss": 0.00031779910204932094,
          "validation_loss": 0.000267805386101827,
          "train_metric": 3.20075596391689e-05,
          "validation_metric": 2.6972378691425547e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.486829901999954
        },
        {
          "epoch": 220,
          "updates": 3520,
          "samples_seen": 225280,
          "online_loss": 0.0003240369351260597,
          "train_loss": 0.00032727583311498165,
          "validation_loss": 0.00029358992469497025,
          "train_metric": 3.2962019758997485e-05,
          "validation_metric": 2.9569302569143474e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4972949879993394
        },
        {
          "epoch": 221,
          "updates": 3536,
          "samples_seen": 226304,
          "online_loss": 0.0003389974217498093,
          "train_loss": 0.0003199097700417042,
          "validation_loss": 0.0002637380675878376,
          "train_metric": 3.22201412927825e-05,
          "validation_metric": 2.65627313638106e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.508636125998237
        },
        {
          "epoch": 222,
          "updates": 3552,
          "samples_seen": 227328,
          "online_loss": 0.0003186224976161611,
          "train_loss": 0.0003257069911342114,
          "validation_loss": 0.00028936899616383016,
          "train_metric": 3.280401142546907e-05,
          "validation_metric": 2.9144188374630176e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.51926786399963
        },
        {
          "epoch": 223,
          "updates": 3568,
          "samples_seen": 228352,
          "online_loss": 0.0003394426112208748,
          "train_loss": 0.0003246062551625073,
          "validation_loss": 0.00028636588831432164,
          "train_metric": 3.2693151297280565e-05,
          "validation_metric": 2.884172681660857e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.529788033998557
        },
        {
          "epoch": 224,
          "updates": 3584,
          "samples_seen": 229376,
          "online_loss": 0.0003623662250902271,
          "train_loss": 0.0003415491955820471,
          "validation_loss": 0.00028102510259486735,
          "train_metric": 3.439958891249262e-05,
          "validation_metric": 2.8303822546149604e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.540270186998896
        },
        {
          "epoch": 225,
          "updates": 3600,
          "samples_seen": 230400,
          "online_loss": 0.0003272360008850228,
          "train_loss": 0.00031415163539350033,
          "validation_loss": 0.00027221255004405975,
          "train_metric": 3.164020017720759e-05,
          "validation_metric": 2.7416248485678807e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5506591809989914
        },
        {
          "epoch": 226,
          "updates": 3616,
          "samples_seen": 231424,
          "online_loss": 0.0003188829805367277,
          "train_loss": 0.00031163834501057863,
          "validation_loss": 0.00026662443997338414,
          "train_metric": 3.1387073249788955e-05,
          "validation_metric": 2.6853436793317087e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5620425269989937
        },
        {
          "epoch": 227,
          "updates": 3632,
          "samples_seen": 232448,
          "online_loss": 0.0003325373509142082,
          "train_loss": 0.0003209039568901062,
          "validation_loss": 0.0002712107670959085,
          "train_metric": 3.232027302146889e-05,
          "validation_metric": 2.7315356419421732e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5727853309999773
        },
        {
          "epoch": 228,
          "updates": 3648,
          "samples_seen": 233472,
          "online_loss": 0.00033897991215781076,
          "train_loss": 0.0003217901394236833,
          "validation_loss": 0.0002717422612477094,
          "train_metric": 3.2409523555543274e-05,
          "validation_metric": 2.736888563958928e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5830188889995043
        },
        {
          "epoch": 229,
          "updates": 3664,
          "samples_seen": 234496,
          "online_loss": 0.00031959877560439054,
          "train_loss": 0.00031945749651640654,
          "validation_loss": 0.00027865232550539076,
          "train_metric": 3.2174590160138905e-05,
          "validation_metric": 2.8064840080332942e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.59333383899866
        },
        {
          "epoch": 230,
          "updates": 3680,
          "samples_seen": 235520,
          "online_loss": 0.0003142996829410549,
          "train_loss": 0.00031227697036229074,
          "validation_loss": 0.0002636255230754614,
          "train_metric": 3.145138907711953e-05,
          "validation_metric": 2.65513954218477e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6042496079990087
        },
        {
          "epoch": 231,
          "updates": 3696,
          "samples_seen": 236544,
          "online_loss": 0.00031671516717324266,
          "train_loss": 0.00031198651413433254,
          "validation_loss": 0.00027528300415724516,
          "train_metric": 3.1422136089531705e-05,
          "validation_metric": 2.772549123619683e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.614626775999568
        },
        {
          "epoch": 232,
          "updates": 3712,
          "samples_seen": 237568,
          "online_loss": 0.0003694421702675754,
          "train_loss": 0.0004324360634200275,
          "validation_loss": 0.0003207287809345871,
          "train_metric": 4.35533765994478e-05,
          "validation_metric": 3.23026288242545e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6248900729988236
        },
        {
          "epoch": 233,
          "updates": 3728,
          "samples_seen": 238592,
          "online_loss": 0.00036274780995881883,
          "train_loss": 0.0003121584886685014,
          "validation_loss": 0.00025871407706290483,
          "train_metric": 3.1439460144611076e-05,
          "validation_metric": 2.6056735805468634e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6357014319983136
        },
        {
          "epoch": 234,
          "updates": 3744,
          "samples_seen": 239616,
          "online_loss": 0.00031668919382354943,
          "train_loss": 0.00030838060774840415,
          "validation_loss": 0.00027612410485744476,
          "train_metric": 3.105896394117735e-05,
          "validation_metric": 2.7810208848677576e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6461583849995804
        },
        {
          "epoch": 235,
          "updates": 3760,
          "samples_seen": 240640,
          "online_loss": 0.00031027929799165577,
          "train_loss": 0.0003066586796194315,
          "validation_loss": 0.000258448941167444,
          "train_metric": 3.0885537853464484e-05,
          "validation_metric": 2.6030029403045774e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6571296979982435
        },
        {
          "epoch": 236,
          "updates": 3776,
          "samples_seen": 241664,
          "online_loss": 0.0003137228522973601,
          "train_loss": 0.0003081993490923196,
          "validation_loss": 0.00027267757104709744,
          "train_metric": 3.1040712201502174e-05,
          "validation_metric": 2.7463083824841306e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.66750392099857
        },
        {
          "epoch": 237,
          "updates": 3792,
          "samples_seen": 242688,
          "online_loss": 0.0003109598001174163,
          "train_loss": 0.0003258029173593968,
          "validation_loss": 0.00026580135454423726,
          "train_metric": 3.281368117313832e-05,
          "validation_metric": 2.677053998922929e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6779352269986703
        },
        {
          "epoch": 238,
          "updates": 3808,
          "samples_seen": 243712,
          "online_loss": 0.0003349289954712731,
          "train_loss": 0.0003251967136748135,
          "validation_loss": 0.00028535755700431764,
          "train_metric": 3.275262133684009e-05,
          "validation_metric": 2.8740170819219202e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6884171239998977
        },
        {
          "epoch": 239,
          "updates": 3824,
          "samples_seen": 244736,
          "online_loss": 0.0002926884708358557,
          "train_loss": 0.0003303444245830178,
          "validation_loss": 0.0002721581840887666,
          "train_metric": 3.3271080610575154e-05,
          "validation_metric": 2.741077696555294e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6987695899988466
        },
        {
          "epoch": 240,
          "updates": 3840,
          "samples_seen": 245760,
          "online_loss": 0.000341374528943561,
          "train_loss": 0.00030995564884506166,
          "validation_loss": 0.000260973465628922,
          "train_metric": 3.121759800706059e-05,
          "validation_metric": 2.628429137985222e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.709747228998822
        },
        {
          "epoch": 241,
          "updates": 3856,
          "samples_seen": 246784,
          "online_loss": 0.0003193309721609694,
          "train_loss": 0.0003030627849511802,
          "validation_loss": 0.00026728754164651036,
          "train_metric": 3.05233734252397e-05,
          "validation_metric": 2.6920220989268273e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7206331309989764
        },
        {
          "epoch": 242,
          "updates": 3872,
          "samples_seen": 247808,
          "online_loss": 0.00031283381576940883,
          "train_loss": 0.0003009643405675888,
          "validation_loss": 0.0002557363186497241,
          "train_metric": 3.0312026865431108e-05,
          "validation_metric": 2.5756828108569607e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7317543989993283
        },
        {
          "epoch": 243,
          "updates": 3888,
          "samples_seen": 248832,
          "online_loss": 0.0003069946960749803,
          "train_loss": 0.00030916795367375016,
          "validation_loss": 0.0002738151524681598,
          "train_metric": 3.113826096523553e-05,
          "validation_metric": 2.757766196737066e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.742772244999287
        },
        {
          "epoch": 244,
          "updates": 3904,
          "samples_seen": 249856,
          "online_loss": 0.00032497337815584615,
          "train_loss": 0.0003222412196919322,
          "validation_loss": 0.0002728543186094612,
          "train_metric": 3.2454954634886235e-05,
          "validation_metric": 2.748088627413381e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7535126819984725
        },
        {
          "epoch": 245,
          "updates": 3920,
          "samples_seen": 250880,
          "online_loss": 0.0003645034021246829,
          "train_loss": 0.00030244130175560713,
          "validation_loss": 0.00027259031776338816,
          "train_metric": 3.046078018087428e-05,
          "validation_metric": 2.745429810602218e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7640042859984533
        },
        {
          "epoch": 246,
          "updates": 3936,
          "samples_seen": 251904,
          "online_loss": 0.0003299778154541855,
          "train_loss": 0.00034823533496819437,
          "validation_loss": 0.000281727930996567,
          "train_metric": 3.507298606564291e-05,
          "validation_metric": 2.8374604880809784e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7746195509989775
        },
        {
          "epoch": 247,
          "updates": 3952,
          "samples_seen": 252928,
          "online_loss": 0.0003257705930082011,
          "train_loss": 0.0002967376494780183,
          "validation_loss": 0.0002608137729112059,
          "train_metric": 2.9886325137340464e-05,
          "validation_metric": 2.6268206056556664e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.785156852
        },
        {
          "epoch": 248,
          "updates": 3968,
          "samples_seen": 253952,
          "online_loss": 0.00033581560137463384,
          "train_loss": 0.0002968608750961721,
          "validation_loss": 0.0002637257566675544,
          "train_metric": 2.9898736102040857e-05,
          "validation_metric": 2.656149081303738e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7954989799982286
        },
        {
          "epoch": 249,
          "updates": 3984,
          "samples_seen": 254976,
          "online_loss": 0.0003072070112466463,
          "train_loss": 0.00031248119194060564,
          "validation_loss": 0.0002829036093316972,
          "train_metric": 3.147196548525244e-05,
          "validation_metric": 2.8493013815023005e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.805973818998609
        },
        {
          "epoch": 250,
          "updates": 4000,
          "samples_seen": 256000,
          "online_loss": 0.00033135287503682775,
          "train_loss": 0.00030972628155723214,
          "validation_loss": 0.00026931913453154266,
          "train_metric": 3.1194496841635555e-05,
          "validation_metric": 2.7124842745251954e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8162059409987705
        },
        {
          "epoch": 251,
          "updates": 4016,
          "samples_seen": 257024,
          "online_loss": 0.00030214559910746175,
          "train_loss": 0.000305742840282619,
          "validation_loss": 0.00026913327747024596,
          "train_metric": 3.079330053878948e-05,
          "validation_metric": 2.710611624934245e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8266362019985536
        },
        {
          "epoch": 252,
          "updates": 4032,
          "samples_seen": 258048,
          "online_loss": 0.000301774186482362,
          "train_loss": 0.00030750257428735495,
          "validation_loss": 0.0002623066829983145,
          "train_metric": 3.097053195233457e-05,
          "validation_metric": 2.6418569177621976e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.836946025998259
        },
        {
          "epoch": 253,
          "updates": 4048,
          "samples_seen": 259072,
          "online_loss": 0.0003078587742493255,
          "train_loss": 0.0002937580575235188,
          "validation_loss": 0.0002528800687287003,
          "train_metric": 2.9586235541501082e-05,
          "validation_metric": 2.546915493439883e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8478871669994987
        },
        {
          "epoch": 254,
          "updates": 4064,
          "samples_seen": 260096,
          "online_loss": 0.00031305045285989763,
          "train_loss": 0.000296320446068421,
          "validation_loss": 0.0002502925635781139,
          "train_metric": 2.984431012009736e-05,
          "validation_metric": 2.520855559851043e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.859309718998702
        },
        {
          "epoch": 255,
          "updates": 4080,
          "samples_seen": 261120,
          "online_loss": 0.00035689121796167456,
          "train_loss": 0.00036366612766869366,
          "validation_loss": 0.0003043339238502085,
          "train_metric": 3.662711969809607e-05,
          "validation_metric": 3.065139753744006e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.869824957999299
        },
        {
          "epoch": 256,
          "updates": 4096,
          "samples_seen": 262144,
          "online_loss": 0.0003201622275810223,
          "train_loss": 0.0003124268550891429,
          "validation_loss": 0.0002555179235059768,
          "train_metric": 3.146649032714777e-05,
          "validation_metric": 2.573482743173372e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.880591307999566
        },
        {
          "epoch": 257,
          "updates": 4112,
          "samples_seen": 263168,
          "online_loss": 0.00034286915069969837,
          "train_loss": 0.0003283314872533083,
          "validation_loss": 0.00029740872560068965,
          "train_metric": 3.3068346965592355e-05,
          "validation_metric": 2.9953918783576228e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.8913705309987563
        },
        {
          "epoch": 258,
          "updates": 4128,
          "samples_seen": 264192,
          "online_loss": 0.00030623881684732623,
          "train_loss": 0.0003323867276776582,
          "validation_loss": 0.00029227513005025685,
          "train_metric": 3.347677557030693e-05,
          "validation_metric": 2.943688195955474e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9019528329990862
        },
        {
          "epoch": 259,
          "updates": 4144,
          "samples_seen": 265216,
          "online_loss": 0.00030923632129997713,
          "train_loss": 0.0002960569690912962,
          "validation_loss": 0.00025182176614180207,
          "train_metric": 2.981777288368903e-05,
          "validation_metric": 2.536256579332985e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9124689119998948
        },
        {
          "epoch": 260,
          "updates": 4160,
          "samples_seen": 266240,
          "online_loss": 0.000300221171528392,
          "train_loss": 0.00029016818734817207,
          "validation_loss": 0.00026339205214753747,
          "train_metric": 2.9224676836747676e-05,
          "validation_metric": 2.6527883164817467e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9228361639998184
        },
        {
          "epoch": 261,
          "updates": 4176,
          "samples_seen": 267264,
          "online_loss": 0.0002994460319314385,
          "train_loss": 0.0002939927508123219,
          "validation_loss": 0.00025224703131243587,
          "train_metric": 2.9609875127789564e-05,
          "validation_metric": 2.5405397536815144e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.933341820998976
        },
        {
          "epoch": 262,
          "updates": 4192,
          "samples_seen": 268288,
          "online_loss": 0.00033164243450301,
          "train_loss": 0.0003124473732896149,
          "validation_loss": 0.0002621021121740341,
          "train_metric": 3.146855306113139e-05,
          "validation_metric": 2.6397963665658608e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9439516379989072
        },
        {
          "epoch": 263,
          "updates": 4208,
          "samples_seen": 269312,
          "online_loss": 0.00033032565715984674,
          "train_loss": 0.00034601055085659027,
          "validation_loss": 0.00030469210469163954,
          "train_metric": 3.484891203697771e-05,
          "validation_metric": 3.068747173529118e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.954180657998222
        },
        {
          "epoch": 264,
          "updates": 4224,
          "samples_seen": 270336,
          "online_loss": 0.0003304289530206006,
          "train_loss": 0.0003196413745172322,
          "validation_loss": 0.0002732042921707034,
          "train_metric": 3.2193107472267e-05,
          "validation_metric": 2.7516136469785124e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.9646956299984595
        },
        {
          "epoch": 265,
          "updates": 4240,
          "samples_seen": 271360,
          "online_loss": 0.0003552678067535453,
          "train_loss": 0.00029717120924033225,
          "validation_loss": 0.0002789603022392839,
          "train_metric": 2.992999361595139e-05,
          "validation_metric": 2.809585930663161e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.975159193998479
        },
        {
          "epoch": 266,
          "updates": 4256,
          "samples_seen": 272384,
          "online_loss": 0.0003006809479302319,
          "train_loss": 0.0002843642432708293,
          "validation_loss": 0.0002464940771460533,
          "train_metric": 2.8640124583034776e-05,
          "validation_metric": 2.4825982109177858e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9861627829995996
        },
        {
          "epoch": 267,
          "updates": 4272,
          "samples_seen": 273408,
          "online_loss": 0.00028516339534689905,
          "train_loss": 0.0002873997436836362,
          "validation_loss": 0.0002527337346691638,
          "train_metric": 2.8945849408046342e-05,
          "validation_metric": 2.54544174822513e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.996833893999792
        },
        {
          "epoch": 268,
          "updates": 4288,
          "samples_seen": 274432,
          "online_loss": 0.0003097429280387587,
          "train_loss": 0.00029408864793367684,
          "validation_loss": 0.00026207900373265147,
          "train_metric": 2.9619533961522393e-05,
          "validation_metric": 2.6395637178211473e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.007324891999815
        },
        {
          "epoch": 269,
          "updates": 4304,
          "samples_seen": 275456,
          "online_loss": 0.0002954398050860618,
          "train_loss": 0.0002899955725297332,
          "validation_loss": 0.0002481213887222111,
          "train_metric": 2.9207294574007392e-05,
          "validation_metric": 2.4989874873426743e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0178418989999045
        },
        {
          "epoch": 270,
          "updates": 4320,
          "samples_seen": 276480,
          "online_loss": 0.0002994037713506259,
          "train_loss": 0.00028623436810448766,
          "validation_loss": 0.00026034004986286163,
          "train_metric": 2.882848093577195e-05,
          "validation_metric": 2.622049578349106e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0282685559996025
        },
        {
          "epoch": 271,
          "updates": 4336,
          "samples_seen": 277504,
          "online_loss": 0.00029006394743191777,
          "train_loss": 0.0002908404276240617,
          "validation_loss": 0.00024368715821765363,
          "train_metric": 2.9292381441337056e-05,
          "validation_metric": 2.4543276595068164e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0393342209990806
        },
        {
          "epoch": 272,
          "updates": 4352,
          "samples_seen": 278528,
          "online_loss": 0.00029972615266160574,
          "train_loss": 0.0003048560465686023,
          "validation_loss": 0.0002750589628703892,
          "train_metric": 3.070398815907538e-05,
          "validation_metric": 2.7702932129614055e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0503382409988262
        },
        {
          "epoch": 273,
          "updates": 4368,
          "samples_seen": 279552,
          "online_loss": 0.0002992797808474279,
          "train_loss": 0.0002811932936310768,
          "validation_loss": 0.0002455832145642489,
          "train_metric": 2.832075915648602e-05,
          "validation_metric": 2.473423955962062e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0608761529983894
        },
        {
          "epoch": 274,
          "updates": 4384,
          "samples_seen": 280576,
          "online_loss": 0.0002886028169086785,
          "train_loss": 0.0002810312435030937,
          "validation_loss": 0.000249172851908952,
          "train_metric": 2.8304437364568003e-05,
          "validation_metric": 2.509578007447999e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.071238751999772
        },
        {
          "epoch": 275,
          "updates": 4400,
          "samples_seen": 281600,
          "online_loss": 0.00028631320674321614,
          "train_loss": 0.0002806431439239532,
          "validation_loss": 0.0002468460297677666,
          "train_metric": 2.8265349101275206e-05,
          "validation_metric": 2.4861428755684756e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0816989869999816
        },
        {
          "epoch": 276,
          "updates": 4416,
          "samples_seen": 282624,
          "online_loss": 0.00029083454592182534,
          "train_loss": 0.00028549862327054143,
          "validation_loss": 0.00024313713947776705,
          "train_metric": 2.8754377126460895e-05,
          "validation_metric": 2.4487882910761982e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0927558289986337
        },
        {
          "epoch": 277,
          "updates": 4432,
          "samples_seen": 283648,
          "online_loss": 0.0002854735312212142,
          "train_loss": 0.0002917969541158527,
          "validation_loss": 0.00026523947599343956,
          "train_metric": 2.938872239610646e-05,
          "validation_metric": 2.6713949409895577e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1036398539999936
        },
        {
          "epoch": 278,
          "updates": 4448,
          "samples_seen": 284672,
          "online_loss": 0.00028765653041773476,
          "train_loss": 0.0002849719312507659,
          "validation_loss": 0.00024690895224921405,
          "train_metric": 2.8701329938485287e-05,
          "validation_metric": 2.486776611476671e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.114415129999543
        },
        {
          "epoch": 279,
          "updates": 4464,
          "samples_seen": 285696,
          "online_loss": 0.00029106334204698214,
          "train_loss": 0.0002809399738907814,
          "validation_loss": 0.00025174670736305416,
          "train_metric": 2.8295249649090692e-05,
          "validation_metric": 2.5355006073368713e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1247706149988517
        },
        {
          "epoch": 280,
          "updates": 4480,
          "samples_seen": 286720,
          "online_loss": 0.00029081501634209417,
          "train_loss": 0.0002843899419531226,
          "validation_loss": 0.0002404970582574606,
          "train_metric": 2.8642714823945425e-05,
          "validation_metric": 2.422198303975165e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1359319609982776
        },
        {
          "epoch": 281,
          "updates": 4496,
          "samples_seen": 287744,
          "online_loss": 0.00028927766652486753,
          "train_loss": 0.00027867837343364954,
          "validation_loss": 0.0002491250925231725,
          "train_metric": 2.8067461244063452e-05,
          "validation_metric": 2.509096884750761e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1466384659997857
        },
        {
          "epoch": 282,
          "updates": 4512,
          "samples_seen": 288768,
          "online_loss": 0.00028298138295212993,
          "train_loss": 0.0002783779054880142,
          "validation_loss": 0.00024231565475929528,
          "train_metric": 2.803720417432487e-05,
          "validation_metric": 2.4405147996731102e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.157321165999747
        },
        {
          "epoch": 283,
          "updates": 4528,
          "samples_seen": 289792,
          "online_loss": 0.0002878287150451797,
          "train_loss": 0.0002798881323542446,
          "validation_loss": 0.00024249406123999506,
          "train_metric": 2.8189308068249375e-05,
          "validation_metric": 2.4423114155069925e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1677903389991116
        },
        {
          "epoch": 284,
          "updates": 4544,
          "samples_seen": 290816,
          "online_loss": 0.00028392759486450814,
          "train_loss": 0.00027817708905786276,
          "validation_loss": 0.00024998915614560246,
          "train_metric": 2.801697701215744e-05,
          "validation_metric": 2.5177991119562648e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.178809132999959
        },
        {
          "epoch": 285,
          "updates": 4560,
          "samples_seen": 291840,
          "online_loss": 0.0002904873435909394,
          "train_loss": 0.0002776071196421981,
          "validation_loss": 0.00024106430646497756,
          "train_metric": 2.7959576982539147e-05,
          "validation_metric": 2.4279113858938217e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.189631411998562
        },
        {
          "epoch": 286,
          "updates": 4576,
          "samples_seen": 292864,
          "online_loss": 0.0002813718583638547,
          "train_loss": 0.0002765085664577782,
          "validation_loss": 0.00024408330500591546,
          "train_metric": 2.784893149510026e-05,
          "validation_metric": 2.458317430864554e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2002545229988755
        },
        {
          "epoch": 287,
          "updates": 4592,
          "samples_seen": 293888,
          "online_loss": 0.0002823004938363738,
          "train_loss": 0.0002775885513983667,
          "validation_loss": 0.0002506068267393857,
          "train_metric": 2.79577052424429e-05,
          "validation_metric": 2.5240204195142724e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2109102379999968
        },
        {
          "epoch": 288,
          "updates": 4608,
          "samples_seen": 294912,
          "online_loss": 0.0002820948593580397,
          "train_loss": 0.00027573652914725244,
          "validation_loss": 0.00024263524392154068,
          "train_metric": 2.777117333607748e-05,
          "validation_metric": 2.4437333195237443e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2214402059998974
        },
        {
          "epoch": 289,
          "updates": 4624,
          "samples_seen": 295936,
          "online_loss": 0.0002801358159558731,
          "train_loss": 0.00027799836243502796,
          "validation_loss": 0.00024058643612079322,
          "train_metric": 2.799897629301995e-05,
          "validation_metric": 2.42309870372992e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2317952729990793
        },
        {
          "epoch": 290,
          "updates": 4640,
          "samples_seen": 296960,
          "online_loss": 0.000282075419818284,
          "train_loss": 0.0002759960188996047,
          "validation_loss": 0.00024539497098885477,
          "train_metric": 2.779730493784882e-05,
          "validation_metric": 2.4715282052056864e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.242437181999776
        },
        {
          "epoch": 291,
          "updates": 4656,
          "samples_seen": 297984,
          "online_loss": 0.0002856978644558694,
          "train_loss": 0.0002763209631666541,
          "validation_loss": 0.00024383788695558906,
          "train_metric": 2.783003219519742e-05,
          "validation_metric": 2.455845969961956e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2529247559996293
        },
        {
          "epoch": 292,
          "updates": 4672,
          "samples_seen": 299008,
          "online_loss": 0.0002747498747339705,
          "train_loss": 0.00027604750357568264,
          "validation_loss": 0.00024373290943913162,
          "train_metric": 2.7802492695627734e-05,
          "validation_metric": 2.4547887733206153e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.2635228729996015
        },
        {
          "epoch": 293,
          "updates": 4688,
          "samples_seen": 300032,
          "online_loss": 0.00027761002002080204,
          "train_loss": 0.00027400668477639556,
          "validation_loss": 0.00023930254974402487,
          "train_metric": 2.759695053100586e-05,
          "validation_metric": 2.4101675080601126e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.2751172429998405
        },
        {
          "epoch": 294,
          "updates": 4704,
          "samples_seen": 301056,
          "online_loss": 0.000285005483419809,
          "train_loss": 0.0002742321230471134,
          "validation_loss": 0.00024352263426408172,
          "train_metric": 2.7619656975730322e-05,
          "validation_metric": 2.452670560160186e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.2858578879986453
        },
        {
          "epoch": 295,
          "updates": 4720,
          "samples_seen": 302080,
          "online_loss": 0.00028594226751010865,
          "train_loss": 0.0002741378848440945,
          "validation_loss": 0.0002398512588115409,
          "train_metric": 2.7610169127001427e-05,
          "validation_metric": 2.4156943254638463e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.2968100089983636
        },
        {
          "epoch": 296,
          "updates": 4736,
          "samples_seen": 303104,
          "online_loss": 0.0002819721685227705,
          "train_loss": 0.0002733570581767708,
          "validation_loss": 0.00024400949769187719,
          "train_metric": 2.753152330114972e-05,
          "validation_metric": 2.4575745555921458e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.307700200999534
        },
        {
          "epoch": 297,
          "updates": 4752,
          "samples_seen": 304128,
          "online_loss": 0.00027763874913944164,
          "train_loss": 0.00027317757485434413,
          "validation_loss": 0.00024152151308953762,
          "train_metric": 2.7513442546478473e-05,
          "validation_metric": 2.432516157568898e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.318164615999194
        },
        {
          "epoch": 298,
          "updates": 4768,
          "samples_seen": 305152,
          "online_loss": 0.0002758106274995953,
          "train_loss": 0.0002728643885347992,
          "validation_loss": 0.00024154726997949183,
          "train_metric": 2.7481901270220987e-05,
          "validation_metric": 2.432775727356784e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.3286571009994077
        },
        {
          "epoch": 299,
          "updates": 4784,
          "samples_seen": 306176,
          "online_loss": 0.0002749607137957355,
          "train_loss": 0.0002726663660723716,
          "validation_loss": 0.00023977234377525747,
          "train_metric": 2.7461954232421704e-05,
          "validation_metric": 2.4148994270944968e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.339139913998224
        },
        {
          "epoch": 300,
          "updates": 4800,
          "samples_seen": 307200,
          "online_loss": 0.00028007099808746716,
          "train_loss": 0.000275230297120288,
          "validation_loss": 0.00024510512594133615,
          "train_metric": 2.7720188882085495e-05,
          "validation_metric": 2.4686089091119356e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.3503211909992388
        },
        {
          "epoch": 301,
          "updates": 4816,
          "samples_seen": 308224,
          "online_loss": 0.00029470625213434687,
          "train_loss": 0.0002735276357270777,
          "validation_loss": 0.00024236779427155852,
          "train_metric": 2.7548700018087402e-05,
          "validation_metric": 2.441039396217093e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.361105325999233
        },
        {
          "epoch": 302,
          "updates": 4832,
          "samples_seen": 309248,
          "online_loss": 0.0002922776229752344,
          "train_loss": 0.00027732003945857286,
          "validation_loss": 0.0002437353687128052,
          "train_metric": 2.793065687001217e-05,
          "validation_metric": 2.4548135115765035e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.3715171169988025
        },
        {
          "epoch": 303,
          "updates": 4848,
          "samples_seen": 310272,
          "online_loss": 0.0002839570715877926,
          "train_loss": 0.00027429594774730504,
          "validation_loss": 0.00023943181440699846,
          "train_metric": 2.7626081646303646e-05,
          "validation_metric": 2.4114695406751707e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.3819198129986034
        },
        {
          "epoch": 304,
          "updates": 4864,
          "samples_seen": 311296,
          "online_loss": 0.0002729250904849323,
          "train_loss": 0.00027512433007359505,
          "validation_loss": 0.00024721992667764425,
          "train_metric": 2.7709515052265488e-05,
          "validation_metric": 2.489908911229577e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.3925456079996366
        },
        {
          "epoch": 305,
          "updates": 4880,
          "samples_seen": 312320,
          "online_loss": 0.000280314130577608,
          "train_loss": 0.00027441204292699695,
          "validation_loss": 0.00024080461298581213,
          "train_metric": 2.7637775929179043e-05,
          "validation_metric": 2.425296224828344e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4030910649998987
        },
        {
          "epoch": 306,
          "updates": 4896,
          "samples_seen": 313344,
          "online_loss": 0.0002786257182378904,
          "train_loss": 0.0002725808590184897,
          "validation_loss": 0.00024123919138219208,
          "train_metric": 2.7453346774564125e-05,
          "validation_metric": 2.429672713333275e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4138444589989376
        },
        {
          "epoch": 307,
          "updates": 4912,
          "samples_seen": 314368,
          "online_loss": 0.00027625629900285276,
          "train_loss": 0.00027110741939395666,
          "validation_loss": 0.0002407257561571896,
          "train_metric": 2.730494634306524e-05,
          "validation_metric": 2.4245018721558154e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4243167629992968
        },
        {
          "epoch": 308,
          "updates": 4928,
          "samples_seen": 315392,
          "online_loss": 0.0002755088808044093,
          "train_loss": 0.0002711908018682152,
          "validation_loss": 0.00024186646624002606,
          "train_metric": 2.7313340979162604e-05,
          "validation_metric": 2.435990609228611e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.434753916999398
        },
        {
          "epoch": 309,
          "updates": 4944,
          "samples_seen": 316416,
          "online_loss": 0.00027456917359813815,
          "train_loss": 0.00027084044995717704,
          "validation_loss": 0.00023932175827212632,
          "train_metric": 2.7278059860691428e-05,
          "validation_metric": 2.41036104853265e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4453650069990545
        },
        {
          "epoch": 310,
          "updates": 4960,
          "samples_seen": 317440,
          "online_loss": 0.00027777546893048566,
          "train_loss": 0.0002707895473577082,
          "validation_loss": 0.00023839109053369612,
          "train_metric": 2.7272932129562832e-05,
          "validation_metric": 2.400987796136178e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.456342866998966
        },
        {
          "epoch": 311,
          "updates": 4976,
          "samples_seen": 318464,
          "online_loss": 0.000273360552455415,
          "train_loss": 0.00027143629267811775,
          "validation_loss": 0.0002419180382275954,
          "train_metric": 2.733807014010381e-05,
          "validation_metric": 2.4365099307033233e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4672279599999456
        },
        {
          "epoch": 312,
          "updates": 4992,
          "samples_seen": 319488,
          "online_loss": 0.0002775834364001639,
          "train_loss": 0.00027023107395507395,
          "validation_loss": 0.00023897012579254806,
          "train_metric": 2.7216685339226387e-05,
          "validation_metric": 2.406819476163946e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4777522569984285
        },
        {
          "epoch": 313,
          "updates": 5008,
          "samples_seen": 320512,
          "online_loss": 0.00027342289286025334,
          "train_loss": 0.0002701384946703911,
          "validation_loss": 0.00023800313647370785,
          "train_metric": 2.7207363018533215e-05,
          "validation_metric": 2.397080243099481e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.488851727999645
        },
        {
          "epoch": 314,
          "updates": 5024,
          "samples_seen": 321536,
          "online_loss": 0.00027528961072675884,
          "train_loss": 0.0002702079073060304,
          "validation_loss": 0.00023951199545990676,
          "train_metric": 2.721434793784283e-05,
          "validation_metric": 2.4122769900714047e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.4995991910000157
        },
        {
          "epoch": 315,
          "updates": 5040,
          "samples_seen": 322560,
          "online_loss": 0.0002702428573684301,
          "train_loss": 0.00027124330517835915,
          "validation_loss": 0.00024271670554298908,
          "train_metric": 2.731863241933752e-05,
          "validation_metric": 2.4445540475426242e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.510188919999564
        },
        {
          "epoch": 316,
          "updates": 5056,
          "samples_seen": 323584,
          "online_loss": 0.0002799539997795364,
          "train_loss": 0.0002698727184906602,
          "validation_loss": 0.00023843967937864363,
          "train_metric": 2.7180591132491827e-05,
          "validation_metric": 2.401477104285732e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.520620676999897
        },
        {
          "epoch": 317,
          "updates": 5072,
          "samples_seen": 324608,
          "online_loss": 0.0002789899990602862,
          "train_loss": 0.0002695750736165792,
          "validation_loss": 0.00024000764824450016,
          "train_metric": 2.7150612368131988e-05,
          "validation_metric": 2.417269024590496e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5314060319997225
        },
        {
          "epoch": 318,
          "updates": 5088,
          "samples_seen": 325632,
          "online_loss": 0.00027191757180844434,
          "train_loss": 0.0002708336687646806,
          "validation_loss": 0.00023666996276006103,
          "train_metric": 2.7277375920675695e-05,
          "validation_metric": 2.3836531909182668e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5429241099991486
        },
        {
          "epoch": 319,
          "updates": 5104,
          "samples_seen": 326656,
          "online_loss": 0.0002718412133617676,
          "train_loss": 0.0002695312141440809,
          "validation_loss": 0.00024173219571821392,
          "train_metric": 2.7146197680849582e-05,
          "validation_metric": 2.4346380087081343e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5536044999989826
        },
        {
          "epoch": 320,
          "updates": 5120,
          "samples_seen": 327680,
          "online_loss": 0.0002758473692665575,
          "train_loss": 0.0002695018774829805,
          "validation_loss": 0.0002393302711425349,
          "train_metric": 2.714324182306882e-05,
          "validation_metric": 2.4104469048324972e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5644986839997728
        },
        {
          "epoch": 321,
          "updates": 5136,
          "samples_seen": 328704,
          "online_loss": 0.0002715141390581266,
          "train_loss": 0.0002695636940188706,
          "validation_loss": 0.0002400495868641883,
          "train_metric": 2.714947004278656e-05,
          "validation_metric": 2.41769175772788e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5749830679997103
        },
        {
          "epoch": 322,
          "updates": 5152,
          "samples_seen": 329728,
          "online_loss": 0.0002743819204624742,
          "train_loss": 0.000268483447143808,
          "validation_loss": 0.00023669222719036043,
          "train_metric": 2.7040670829592273e-05,
          "validation_metric": 2.3838772904127836e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.5853386339986173
        },
        {
          "epoch": 323,
          "updates": 5168,
          "samples_seen": 330752,
          "online_loss": 0.0002718880068641738,
          "train_loss": 0.0002686669467948377,
          "validation_loss": 0.00023696868447586894,
          "train_metric": 2.705914812395349e-05,
          "validation_metric": 2.3866621631896123e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.596048957999301
        },
        {
          "epoch": 324,
          "updates": 5184,
          "samples_seen": 331776,
          "online_loss": 0.00027112455154565396,
          "train_loss": 0.00026862978120334446,
          "validation_loss": 0.00023902300745248795,
          "train_metric": 2.7055404643760994e-05,
          "validation_metric": 2.4073522581602447e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.606718669998372
        },
        {
          "epoch": 325,
          "updates": 5200,
          "samples_seen": 332800,
          "online_loss": 0.00026972691284754546,
          "train_loss": 0.00026796324527822435,
          "validation_loss": 0.00023909012088552117,
          "train_metric": 2.6988276658812538e-05,
          "validation_metric": 2.408028012723662e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6174358209991624
        },
        {
          "epoch": 326,
          "updates": 5216,
          "samples_seen": 333824,
          "online_loss": 0.00027255030090600485,
          "train_loss": 0.00026840963982976973,
          "validation_loss": 0.00023933091142680496,
          "train_metric": 2.7033234800910577e-05,
          "validation_metric": 2.41045345319435e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.628026131998922
        },
        {
          "epoch": 327,
          "updates": 5232,
          "samples_seen": 334848,
          "online_loss": 0.0002733673009061022,
          "train_loss": 0.0002692776033654809,
          "validation_loss": 0.00023745084763504565,
          "train_metric": 2.7120651793666184e-05,
          "validation_metric": 2.391517955402378e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6385985529996105
        },
        {
          "epoch": 328,
          "updates": 5248,
          "samples_seen": 335872,
          "online_loss": 0.00027217966180614894,
          "train_loss": 0.0002676650765351951,
          "validation_loss": 0.000238885753788054,
          "train_metric": 2.6958248781738803e-05,
          "validation_metric": 2.4059700081124902e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6492359839994606
        },
        {
          "epoch": 329,
          "updates": 5264,
          "samples_seen": 336896,
          "online_loss": 0.0002710372182264109,
          "train_loss": 0.0002688634558580816,
          "validation_loss": 0.00024018404656089842,
          "train_metric": 2.7078946004621685e-05,
          "validation_metric": 2.4190458134398796e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6603693749984814
        },
        {
          "epoch": 330,
          "updates": 5280,
          "samples_seen": 337920,
          "online_loss": 0.00027726231473934604,
          "train_loss": 0.0002711945853661746,
          "validation_loss": 0.0002362659142818302,
          "train_metric": 2.7313724785926752e-05,
          "validation_metric": 2.379583747824654e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6720595359984145
        },
        {
          "epoch": 331,
          "updates": 5296,
          "samples_seen": 338944,
          "online_loss": 0.0002715609689403209,
          "train_loss": 0.00026964335120283067,
          "validation_loss": 0.0002444300043862313,
          "train_metric": 2.7157489967066795e-05,
          "validation_metric": 2.461809344822541e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.6831694999982574
        },
        {
          "epoch": 332,
          "updates": 5312,
          "samples_seen": 339968,
          "online_loss": 0.00026891781544691185,
          "train_loss": 0.0002683680795598775,
          "validation_loss": 0.00023472360044252127,
          "train_metric": 2.702905476326123e-05,
          "validation_metric": 2.3640501240151934e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.694699016999948
        },
        {
          "epoch": 333,
          "updates": 5328,
          "samples_seen": 340992,
          "online_loss": 0.0002703698519326281,
          "train_loss": 0.0002665034553501755,
          "validation_loss": 0.0002394305047346279,
          "train_metric": 2.684125502128154e-05,
          "validation_metric": 2.4114564439514652e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7054288899998937
        },
        {
          "epoch": 334,
          "updates": 5344,
          "samples_seen": 342016,
          "online_loss": 0.00027155924635735573,
          "train_loss": 0.0002662785700522363,
          "validation_loss": 0.0002383617975283414,
          "train_metric": 2.6818603146239184e-05,
          "validation_metric": 2.400692756054923e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7162466529989615
        },
        {
          "epoch": 335,
          "updates": 5360,
          "samples_seen": 343040,
          "online_loss": 0.0002708536567297415,
          "train_loss": 0.0002660230966284871,
          "validation_loss": 0.00023712099937256426,
          "train_metric": 2.679286990314722e-05,
          "validation_metric": 2.3881955712568015e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.727153906998865
        },
        {
          "epoch": 336,
          "updates": 5376,
          "samples_seen": 344064,
          "online_loss": 0.00027165929077455075,
          "train_loss": 0.0002666132932063192,
          "validation_loss": 0.00023616170801687986,
          "train_metric": 2.68523144768551e-05,
          "validation_metric": 2.3785343728377484e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7380458649986394
        },
        {
          "epoch": 337,
          "updates": 5392,
          "samples_seen": 345088,
          "online_loss": 0.0002698196858545998,
          "train_loss": 0.0002663419581949711,
          "validation_loss": 0.0002365994150750339,
          "train_metric": 2.6824985980056226e-05,
          "validation_metric": 2.382942875556182e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.748525326998788
        },
        {
          "epoch": 338,
          "updates": 5408,
          "samples_seen": 346112,
          "online_loss": 0.0002708152851482737,
          "train_loss": 0.0002654524287208915,
          "validation_loss": 0.00023554908693768084,
          "train_metric": 2.6735397113952786e-05,
          "validation_metric": 2.372364360780921e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.7593368879988702
        },
        {
          "epoch": 339,
          "updates": 5424,
          "samples_seen": 347136,
          "online_loss": 0.00026948354388878215,
          "train_loss": 0.00026543281273916364,
          "validation_loss": 0.00023772615531925112,
          "train_metric": 2.6733421691460535e-05,
          "validation_metric": 2.3942908228491433e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.770243136999852
        },
        {
          "epoch": 340,
          "updates": 5440,
          "samples_seen": 348160,
          "online_loss": 0.00026969504688167945,
          "train_loss": 0.00026533694472163916,
          "validation_loss": 0.00023634526587557048,
          "train_metric": 2.6723766495706514e-05,
          "validation_metric": 2.3803831936675124e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.780947836999985
        },
        {
          "epoch": 341,
          "updates": 5456,
          "samples_seen": 349184,
          "online_loss": 0.00026800653540703934,
          "train_loss": 0.00026491592871025205,
          "validation_loss": 0.00023472581233363599,
          "train_metric": 2.6681362214731053e-05,
          "validation_metric": 2.3640723156859167e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.791785016999711
        },
        {
          "epoch": 342,
          "updates": 5472,
          "samples_seen": 350208,
          "online_loss": 0.00026938664177578175,
          "train_loss": 0.00026703099138103426,
          "validation_loss": 0.0002392071473877877,
          "train_metric": 2.6894384063780308e-05,
          "validation_metric": 2.4092065359582193e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8025261889997637
        },
        {
          "epoch": 343,
          "updates": 5488,
          "samples_seen": 351232,
          "online_loss": 0.00026967869234795216,
          "train_loss": 0.0002645159256644547,
          "validation_loss": 0.0002348959824303165,
          "train_metric": 2.664107523742132e-05,
          "validation_metric": 2.3657859856029972e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8134048739993887
        },
        {
          "epoch": 344,
          "updates": 5504,
          "samples_seen": 352256,
          "online_loss": 0.00027090293951914646,
          "train_loss": 0.0002646867069415748,
          "validation_loss": 0.00023625153698958457,
          "train_metric": 2.665827560122125e-05,
          "validation_metric": 2.379439138167072e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8243282309995266
        },
        {
          "epoch": 345,
          "updates": 5520,
          "samples_seen": 353280,
          "online_loss": 0.0002703910467971582,
          "train_loss": 0.00026496429927647114,
          "validation_loss": 0.00023360968043562025,
          "train_metric": 2.6686235287343152e-05,
          "validation_metric": 2.352831324969884e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.835536645998218
        },
        {
          "epoch": 346,
          "updates": 5536,
          "samples_seen": 354304,
          "online_loss": 0.00027175219383934746,
          "train_loss": 0.0002645400236360729,
          "validation_loss": 0.00023702629550825804,
          "train_metric": 2.664350176928565e-05,
          "validation_metric": 2.3872424208093435e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.846672199999375
        },
        {
          "epoch": 347,
          "updates": 5552,
          "samples_seen": 355328,
          "online_loss": 0.00026981903192790924,
          "train_loss": 0.0002640002057887614,
          "validation_loss": 0.00023389857960864902,
          "train_metric": 2.6589133995003067e-05,
          "validation_metric": 2.3557411623187363e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.857469216998652
        },
        {
          "epoch": 348,
          "updates": 5568,
          "samples_seen": 356352,
          "online_loss": 0.00026866670577874174,
          "train_loss": 0.0002643539337441325,
          "validation_loss": 0.0002372950839344412,
          "train_metric": 2.6624760721460916e-05,
          "validation_metric": 2.38994925894076e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8680193569998664
        },
        {
          "epoch": 349,
          "updates": 5584,
          "samples_seen": 357376,
          "online_loss": 0.00028005411422782345,
          "train_loss": 0.0002643967163749039,
          "validation_loss": 0.00023401233192998916,
          "train_metric": 2.6629068088368513e-05,
          "validation_metric": 2.356886579946149e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.878737236998859
        },
        {
          "epoch": 350,
          "updates": 5600,
          "samples_seen": 358400,
          "online_loss": 0.00026443156002642354,
          "train_loss": 0.0002662420447450131,
          "validation_loss": 0.00023388979025185108,
          "train_metric": 2.681492333067581e-05,
          "validation_metric": 2.3556523956358433e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.889335834999656
        },
        {
          "epoch": 351,
          "updates": 5616,
          "samples_seen": 359424,
          "online_loss": 0.0002721841428865446,
          "train_loss": 0.00026524069835431874,
          "validation_loss": 0.00023964268621057272,
          "train_metric": 2.671406946319621e-05,
          "validation_metric": 2.413593392702751e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.8998330049998913
        },
        {
          "epoch": 352,
          "updates": 5632,
          "samples_seen": 360448,
          "online_loss": 0.0002772553225440788,
          "train_loss": 0.0002672830887604505,
          "validation_loss": 0.00023227022029459476,
          "train_metric": 2.6919771698885597e-05,
          "validation_metric": 2.3393404262606055e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9113990799996827
        },
        {
          "epoch": 353,
          "updates": 5648,
          "samples_seen": 361472,
          "online_loss": 0.0002653417914189049,
          "train_loss": 0.00026584367151372135,
          "validation_loss": 0.00024211424170061946,
          "train_metric": 2.6774800062412396e-05,
          "validation_metric": 2.4384860807913356e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.921893682998416
        },
        {
          "epoch": 354,
          "updates": 5664,
          "samples_seen": 362496,
          "online_loss": 0.00027549337573873345,
          "train_loss": 0.0002636306162457913,
          "validation_loss": 0.00023507422883994877,
          "train_metric": 2.6551910195848905e-05,
          "validation_metric": 2.3675816919421777e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.932296573999338
        },
        {
          "epoch": 355,
          "updates": 5680,
          "samples_seen": 363520,
          "online_loss": 0.0002705913561840134,
          "train_loss": 0.00026316079311072826,
          "validation_loss": 0.0002322601794730872,
          "train_metric": 2.6504592824494466e-05,
          "validation_metric": 2.3392392904497683e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.9434550319983828
        },
        {
          "epoch": 356,
          "updates": 5696,
          "samples_seen": 364544,
          "online_loss": 0.0002696225319596124,
          "train_loss": 0.0002620183222461492,
          "validation_loss": 0.00023568261531181633,
          "train_metric": 2.6389529011794366e-05,
          "validation_metric": 2.373708957748022e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.954011713998625
        },
        {
          "epoch": 357,
          "updates": 5712,
          "samples_seen": 365568,
          "online_loss": 0.0002733710662141675,
          "train_loss": 0.0002640981983859092,
          "validation_loss": 0.00023404693638440222,
          "train_metric": 2.6598998374538496e-05,
          "validation_metric": 2.3572350983158685e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.96451228099977
        },
        {
          "epoch": 358,
          "updates": 5728,
          "samples_seen": 366592,
          "online_loss": 0.0002692597545319586,
          "train_loss": 0.0002641433384269476,
          "validation_loss": 0.0002400643570581451,
          "train_metric": 2.660354584804736e-05,
          "validation_metric": 2.4178403691621497e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.975099991999741
        },
        {
          "epoch": 359,
          "updates": 5744,
          "samples_seen": 367616,
          "online_loss": 0.0002674385241334676,
          "train_loss": 0.00026132521452382207,
          "validation_loss": 0.00023382084327749908,
          "train_metric": 2.6319719836465083e-05,
          "validation_metric": 2.3549579054815695e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.985579563999636
        },
        {
          "epoch": 360,
          "updates": 5760,
          "samples_seen": 368640,
          "online_loss": 0.0002653846331668319,
          "train_loss": 0.00026194177917204797,
          "validation_loss": 0.00023478030925616622,
          "train_metric": 2.6381814677733928e-05,
          "validation_metric": 2.364621286687907e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 3.996085332999428
        },
        {
          "epoch": 361,
          "updates": 5776,
          "samples_seen": 369664,
          "online_loss": 0.00026623856592777884,
          "train_loss": 0.0002630311355460435,
          "validation_loss": 0.0002322755753993988,
          "train_metric": 2.649153429956641e-05,
          "validation_metric": 2.339394632144831e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.006561880998561
        },
        {
          "epoch": 362,
          "updates": 5792,
          "samples_seen": 370688,
          "online_loss": 0.00026671996693039546,
          "train_loss": 0.000260830856859684,
          "validation_loss": 0.0002339642815059051,
          "train_metric": 2.6269930458511226e-05,
          "validation_metric": 2.3564025468658656e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.0171413569987635
        },
        {
          "epoch": 363,
          "updates": 5808,
          "samples_seen": 371712,
          "online_loss": 0.000262675005615165,
          "train_loss": 0.00026092497864738107,
          "validation_loss": 0.0002332895528525114,
          "train_metric": 2.6279409212293103e-05,
          "validation_metric": 2.3496069843531586e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.02764918199864
        },
        {
          "epoch": 364,
          "updates": 5824,
          "samples_seen": 372736,
          "online_loss": 0.0002721541995924781,
          "train_loss": 0.0002628135262057185,
          "validation_loss": 0.00023436201445292681,
          "train_metric": 2.646961365826428e-05,
          "validation_metric": 2.3604083253303543e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.03801724099867
        },
        {
          "epoch": 365,
          "updates": 5840,
          "samples_seen": 373760,
          "online_loss": 0.0002688583444978576,
          "train_loss": 0.00026435934705659747,
          "validation_loss": 0.00023365889501292259,
          "train_metric": 2.6625308237271383e-05,
          "validation_metric": 2.3533269995823503e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.048361784000008
        },
        {
          "epoch": 366,
          "updates": 5856,
          "samples_seen": 374784,
          "online_loss": 0.0002719034455367364,
          "train_loss": 0.0002669675159268081,
          "validation_loss": 0.0002355187461944297,
          "train_metric": 2.6887992135016248e-05,
          "validation_metric": 2.372058770561125e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.058891689999655
        },
        {
          "epoch": 367,
          "updates": 5872,
          "samples_seen": 375808,
          "online_loss": 0.0002634163192851702,
          "train_loss": 0.000259811436990276,
          "validation_loss": 0.00023256872373167425,
          "train_metric": 2.6167255782638676e-05,
          "validation_metric": 2.342346851946786e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.069251191998774
        },
        {
          "epoch": 368,
          "updates": 5888,
          "samples_seen": 376832,
          "online_loss": 0.0002631803681651945,
          "train_loss": 0.00025977869518101215,
          "validation_loss": 0.00023375138698611408,
          "train_metric": 2.6163957954850048e-05,
          "validation_metric": 2.354258504055906e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.079623025998444
        },
        {
          "epoch": 369,
          "updates": 5904,
          "samples_seen": 377856,
          "online_loss": 0.0002638494424900273,
          "train_loss": 0.00026008536224253476,
          "validation_loss": 0.00023314366990234703,
          "train_metric": 2.619484621391166e-05,
          "validation_metric": 2.3481377866119146e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.09011749399906
        },
        {
          "epoch": 370,
          "updates": 5920,
          "samples_seen": 378880,
          "online_loss": 0.0002655949492691434,
          "train_loss": 0.00025972776347771287,
          "validation_loss": 0.0002323188236914575,
          "train_metric": 2.615882840473205e-05,
          "validation_metric": 2.33983009820804e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.10032708899962
        },
        {
          "epoch": 371,
          "updates": 5936,
          "samples_seen": 379904,
          "online_loss": 0.00026654675684767426,
          "train_loss": 0.00025997153716161847,
          "validation_loss": 0.0002351532457396388,
          "train_metric": 2.6183379304711707e-05,
          "validation_metric": 2.368377499806229e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1113120969985175
        },
        {
          "epoch": 372,
          "updates": 5952,
          "samples_seen": 380928,
          "online_loss": 0.00026287039145245217,
          "train_loss": 0.0002591254888102412,
          "validation_loss": 0.00022889056708663702,
          "train_metric": 2.6098170565092005e-05,
          "validation_metric": 2.3053018594509922e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.122224432998337
        },
        {
          "epoch": 373,
          "updates": 5968,
          "samples_seen": 381952,
          "online_loss": 0.00026484439285923145,
          "train_loss": 0.00025946152163669467,
          "validation_loss": 0.0002317569451406598,
          "train_metric": 2.6132012862944975e-05,
          "validation_metric": 2.3341710402746685e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.132574718998512
        },
        {
          "epoch": 374,
          "updates": 5984,
          "samples_seen": 382976,
          "online_loss": 0.0002607154956422164,
          "train_loss": 0.00025868095690384507,
          "validation_loss": 0.00023195584071800113,
          "train_metric": 2.605339977890253e-05,
          "validation_metric": 2.3361742933047935e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.143377263999355
        },
        {
          "epoch": 375,
          "updates": 6000,
          "samples_seen": 384000,
          "online_loss": 0.00026354819965490606,
          "train_loss": 0.0002587641356512904,
          "validation_loss": 0.00023245534976013005,
          "train_metric": 2.6061774406116456e-05,
          "validation_metric": 2.3412052541971207e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.15387873599866
        },
        {
          "epoch": 376,
          "updates": 6016,
          "samples_seen": 385024,
          "online_loss": 0.0002646450557222124,
          "train_loss": 0.0002608666254673153,
          "validation_loss": 0.0002341375802643597,
          "train_metric": 2.627352841955144e-05,
          "validation_metric": 2.358148412895389e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.164215582999532
        },
        {
          "epoch": 377,
          "updates": 6032,
          "samples_seen": 386048,
          "online_loss": 0.00026917399463854963,
          "train_loss": 0.0002586048503872007,
          "validation_loss": 0.00022835202980786562,
          "train_metric": 2.604573455755599e-05,
          "validation_metric": 2.2998778149485588e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.175225380999109
        },
        {
          "epoch": 378,
          "updates": 6048,
          "samples_seen": 387072,
          "online_loss": 0.0002701408111533965,
          "train_loss": 0.0002577376726549119,
          "validation_loss": 0.00023392564617097378,
          "train_metric": 2.5958397600334138e-05,
          "validation_metric": 2.3560136469313875e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.185803845999544
        },
        {
          "epoch": 379,
          "updates": 6064,
          "samples_seen": 388096,
          "online_loss": 0.00025950855615519686,
          "train_loss": 0.0002595665864646435,
          "validation_loss": 0.00023237142886500806,
          "train_metric": 2.6142595743294805e-05,
          "validation_metric": 2.340359969821293e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1961359179986175
        },
        {
          "epoch": 380,
          "updates": 6080,
          "samples_seen": 389120,
          "online_loss": 0.0002673240851436276,
          "train_loss": 0.00026249559596180916,
          "validation_loss": 0.00022949042613618076,
          "train_metric": 2.6437592168804258e-05,
          "validation_metric": 2.31134381465381e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.206670240999301
        },
        {
          "epoch": 381,
          "updates": 6096,
          "samples_seen": 390144,
          "online_loss": 0.0002651559852893115,
          "train_loss": 0.0002573154051788151,
          "validation_loss": 0.00023504179262090474,
          "train_metric": 2.591586599010043e-05,
          "validation_metric": 2.367255001445301e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.216982594000001
        },
        {
          "epoch": 382,
          "updates": 6112,
          "samples_seen": 391168,
          "online_loss": 0.00025831529001152376,
          "train_loss": 0.0002575234102550894,
          "validation_loss": 0.0002305128873558715,
          "train_metric": 2.5936815291061066e-05,
          "validation_metric": 2.321641477465164e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.227299782998671
        },
        {
          "epoch": 383,
          "updates": 6128,
          "samples_seen": 392192,
          "online_loss": 0.00026110030103154713,
          "train_loss": 0.00025707497843541205,
          "validation_loss": 0.00022784173779655248,
          "train_metric": 2.5891651603160426e-05,
          "validation_metric": 2.2947384422877803e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.238271991998772
        },
        {
          "epoch": 384,
          "updates": 6144,
          "samples_seen": 393216,
          "online_loss": 0.0002647615438036155,
          "train_loss": 0.00025667654699645936,
          "validation_loss": 0.00023174354282673448,
          "train_metric": 2.5851524696918204e-05,
          "validation_metric": 2.334036435058806e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.248848747000011
        },
        {
          "epoch": 385,
          "updates": 6160,
          "samples_seen": 394240,
          "online_loss": 0.0002598435030449764,
          "train_loss": 0.0002560750872362405,
          "validation_loss": 0.0002298634935868904,
          "train_metric": 2.5790941435843706e-05,
          "validation_metric": 2.315100937266834e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.259268654999687
        },
        {
          "epoch": 386,
          "updates": 6176,
          "samples_seen": 395264,
          "online_loss": 0.000262467518041376,
          "train_loss": 0.0002585109032224864,
          "validation_loss": 0.00023112143389880657,
          "train_metric": 2.6036272174678743e-05,
          "validation_metric": 2.3277705622604117e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.269644762998723
        },
        {
          "epoch": 387,
          "updates": 6192,
          "samples_seen": 396288,
          "online_loss": 0.0002596759650259628,
          "train_loss": 0.0002555929240770638,
          "validation_loss": 0.0002306670357938856,
          "train_metric": 2.574238351371605e-05,
          "validation_metric": 2.3231936211232096e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.280098366998573
        },
        {
          "epoch": 388,
          "updates": 6208,
          "samples_seen": 397312,
          "online_loss": 0.0002603826651466079,
          "train_loss": 0.00025551888393238187,
          "validation_loss": 0.00022861396428197622,
          "train_metric": 2.5734927476150915e-05,
          "validation_metric": 2.3025158952805214e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.290537179998864
        },
        {
          "epoch": 389,
          "updates": 6224,
          "samples_seen": 398336,
          "online_loss": 0.0002595939959064708,
          "train_loss": 0.0002563301532063633,
          "validation_loss": 0.0002301435306435451,
          "train_metric": 2.5816632842179388e-05,
          "validation_metric": 2.3179216441349126e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.300934244000018
        },
        {
          "epoch": 390,
          "updates": 6240,
          "samples_seen": 399360,
          "online_loss": 0.00026338983570894925,
          "train_loss": 0.00025612657191231847,
          "validation_loss": 0.00022898943279869854,
          "train_metric": 2.579612919362262e-05,
          "validation_metric": 2.3062975742504932e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.311903314999654
        },
        {
          "epoch": 391,
          "updates": 6256,
          "samples_seen": 400384,
          "online_loss": 0.00025776290840440197,
          "train_loss": 0.0002548526390455663,
          "validation_loss": 0.00022676629305351526,
          "train_metric": 2.5667823138064705e-05,
          "validation_metric": 2.283906906086486e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.323095804998957
        },
        {
          "epoch": 392,
          "updates": 6272,
          "samples_seen": 401408,
          "online_loss": 0.0002632235464261612,
          "train_loss": 0.00025607773568481207,
          "validation_loss": 0.00023016842897050083,
          "train_metric": 2.579121064627543e-05,
          "validation_metric": 2.318172118975781e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.333457978998922
        },
        {
          "epoch": 393,
          "updates": 6288,
          "samples_seen": 402432,
          "online_loss": 0.0002720615912039648,
          "train_loss": 0.00025931483833119273,
          "validation_loss": 0.00023628240160178393,
          "train_metric": 2.6117242668988183e-05,
          "validation_metric": 2.379750003456138e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.3442936960000225
        },
        {
          "epoch": 394,
          "updates": 6304,
          "samples_seen": 403456,
          "online_loss": 0.00026276859352947213,
          "train_loss": 0.00025586396805010736,
          "validation_loss": 0.0002273078280268237,
          "train_metric": 2.576967926870566e-05,
          "validation_metric": 2.2893609639140777e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.35475312699964
        },
        {
          "epoch": 395,
          "updates": 6320,
          "samples_seen": 404480,
          "online_loss": 0.00025751723023859086,
          "train_loss": 0.0002555543906055391,
          "validation_loss": 0.0002301423082826659,
          "train_metric": 2.5738501790328883e-05,
          "validation_metric": 2.3179090931080282e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.365228748998561
        },
        {
          "epoch": 396,
          "updates": 6336,
          "samples_seen": 405504,
          "online_loss": 0.00025738307795108994,
          "train_loss": 0.00025378927239216864,
          "validation_loss": 0.00022697419626638293,
          "train_metric": 2.5560726498952135e-05,
          "validation_metric": 2.2860009266878478e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.375839626000015
        },
        {
          "epoch": 397,
          "updates": 6352,
          "samples_seen": 406528,
          "online_loss": 0.0002581362659839215,
          "train_loss": 0.0002542499278206378,
          "validation_loss": 0.00022782931046094745,
          "train_metric": 2.5607119823689573e-05,
          "validation_metric": 2.294613113917876e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.386205578999579
        },
        {
          "epoch": 398,
          "updates": 6368,
          "samples_seen": 407552,
          "online_loss": 0.00025357987306051655,
          "train_loss": 0.000254241400398314,
          "validation_loss": 0.000231480342335999,
          "train_metric": 2.5606263079680502e-05,
          "validation_metric": 2.331385258003138e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.397221812998396
        },
        {
          "epoch": 399,
          "updates": 6384,
          "samples_seen": 408576,
          "online_loss": 0.0002580408608992002,
          "train_loss": 0.00025311848730780184,
          "validation_loss": 0.00022698396060150117,
          "train_metric": 2.549316741351504e-05,
          "validation_metric": 2.2860993340145797e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.408037589999367
        },
        {
          "epoch": 400,
          "updates": 6400,
          "samples_seen": 409600,
          "online_loss": 0.0002580701752776804,
          "train_loss": 0.00025304447626695037,
          "validation_loss": 0.00022841767349746078,
          "train_metric": 2.5485713194939308e-05,
          "validation_metric": 2.3005393813946284e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.418651623998812
        },
        {
          "epoch": 401,
          "updates": 6416,
          "samples_seen": 410624,
          "online_loss": 0.0002567024175732513,
          "train_loss": 0.00025238661328330636,
          "validation_loss": 0.00022789582726545632,
          "train_metric": 2.541945650591515e-05,
          "validation_metric": 2.2952834115130827e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.429570654998315
        },
        {
          "epoch": 402,
          "updates": 6432,
          "samples_seen": 411648,
          "online_loss": 0.00025825490047282074,
          "train_loss": 0.0002535106032155454,
          "validation_loss": 0.00023032049648463726,
          "train_metric": 2.5532661311444826e-05,
          "validation_metric": 2.319703889952507e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.440423548998297
        },
        {
          "epoch": 403,
          "updates": 6448,
          "samples_seen": 412672,
          "online_loss": 0.0002606564075904316,
          "train_loss": 0.00025214202469214797,
          "validation_loss": 0.00022613992041442543,
          "train_metric": 2.5394823751412332e-05,
          "validation_metric": 2.2775984689360484e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.451929894999921
        },
        {
          "epoch": 404,
          "updates": 6464,
          "samples_seen": 413696,
          "online_loss": 0.0002550288536440348,
          "train_loss": 0.000252456491580233,
          "validation_loss": 0.00022695297957397997,
          "train_metric": 2.5426492356928065e-05,
          "validation_metric": 2.2857873773318715e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.463303882999753
        },
        {
          "epoch": 405,
          "updates": 6480,
          "samples_seen": 414720,
          "online_loss": 0.0002558641863288358,
          "train_loss": 0.00025183428078889847,
          "validation_loss": 0.00022520491620525718,
          "train_metric": 2.5363826352986507e-05,
          "validation_metric": 2.2681813788949512e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.535913697998694
        },
        {
          "epoch": 406,
          "updates": 6496,
          "samples_seen": 415744,
          "online_loss": 0.0002616329102238524,
          "train_loss": 0.00025239645037800074,
          "validation_loss": 0.00023026498092804104,
          "train_metric": 2.5420444217161275e-05,
          "validation_metric": 2.319144732609857e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.5474103939995985
        },
        {
          "epoch": 407,
          "updates": 6512,
          "samples_seen": 416768,
          "online_loss": 0.00025631273092585616,
          "train_loss": 0.00025159132201224566,
          "validation_loss": 0.00022358509886544198,
          "train_metric": 2.5339355488540605e-05,
          "validation_metric": 2.2518670448334888e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.558741573999214
        },
        {
          "epoch": 408,
          "updates": 6528,
          "samples_seen": 417792,
          "online_loss": 0.0002567614164945553,
          "train_loss": 0.0002513397776056081,
          "validation_loss": 0.00022632356558460742,
          "train_metric": 2.5314024242106825e-05,
          "validation_metric": 2.279448017361574e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.569831700999202
        },
        {
          "epoch": 409,
          "updates": 6544,
          "samples_seen": 418816,
          "online_loss": 0.0002545785982874804,
          "train_loss": 0.0002529876655898988,
          "validation_loss": 0.0002253263519378379,
          "train_metric": 2.5479990654275753e-05,
          "validation_metric": 2.2694044673698954e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.580613446998541
        },
        {
          "epoch": 410,
          "updates": 6560,
          "samples_seen": 419840,
          "online_loss": 0.0002595522446426912,
          "train_loss": 0.0002505349984858185,
          "validation_loss": 0.00022706927848048508,
          "train_metric": 2.5232970074284822e-05,
          "validation_metric": 2.2869586246088147e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.591259770999386
        },
        {
          "epoch": 411,
          "updates": 6576,
          "samples_seen": 420864,
          "online_loss": 0.0002562366735219257,
          "train_loss": 0.00025080458726733923,
          "validation_loss": 0.00022806492052040994,
          "train_metric": 2.5260118491132744e-05,
          "validation_metric": 2.2969863493926823e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.601920088998668
        },
        {
          "epoch": 412,
          "updates": 6592,
          "samples_seen": 421888,
          "online_loss": 0.00025529582580929855,
          "train_loss": 0.00025203675613738596,
          "validation_loss": 0.0002249135432066396,
          "train_metric": 2.538421904318966e-05,
          "validation_metric": 2.2652468032902107e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.61255867699947
        },
        {
          "epoch": 413,
          "updates": 6608,
          "samples_seen": 422912,
          "online_loss": 0.00025476348855590913,
          "train_loss": 0.00025288210599683225,
          "validation_loss": 0.00022877840092405677,
          "train_metric": 2.5469362299190834e-05,
          "validation_metric": 2.3041722670313902e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.6231377919993975
        },
        {
          "epoch": 414,
          "updates": 6624,
          "samples_seen": 423936,
          "online_loss": 0.00025398011075594695,
          "train_loss": 0.0002509344485588372,
          "validation_loss": 0.00022532459115609527,
          "train_metric": 2.5273198843933642e-05,
          "validation_metric": 2.269386823172681e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.6336685699989175
        },
        {
          "epoch": 415,
          "updates": 6640,
          "samples_seen": 424960,
          "online_loss": 0.00025598683350835927,
          "train_loss": 0.00025041672051884234,
          "validation_loss": 0.00022629999148193747,
          "train_metric": 2.5221057512681e-05,
          "validation_metric": 2.279210639244411e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.644064997999521
        },
        {
          "epoch": 416,
          "updates": 6656,
          "samples_seen": 425984,
          "online_loss": 0.00025328393621748546,
          "train_loss": 0.0002492129569873214,
          "validation_loss": 0.00022403067850973457,
          "train_metric": 2.5099816411966458e-05,
          "validation_metric": 2.256354855489917e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.654486124998584
        },
        {
          "epoch": 417,
          "updates": 6672,
          "samples_seen": 427008,
          "online_loss": 0.000254068047979672,
          "train_loss": 0.00024921860313043,
          "validation_loss": 0.0002264536451548338,
          "train_metric": 2.5100385755649768e-05,
          "validation_metric": 2.2807584173278883e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.664882758999738
        },
        {
          "epoch": 418,
          "updates": 6688,
          "samples_seen": 428032,
          "online_loss": 0.00025411651495232945,
          "train_loss": 0.0002490640035830438,
          "validation_loss": 0.00022378542053047568,
          "train_metric": 2.5084811568376608e-05,
          "validation_metric": 2.2538850316777825e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.675307435998548
        },
        {
          "epoch": 419,
          "updates": 6704,
          "samples_seen": 429056,
          "online_loss": 0.0002511036045689252,
          "train_loss": 0.00024870934430509806,
          "validation_loss": 0.00022657298541162163,
          "train_metric": 2.5049095711437985e-05,
          "validation_metric": 2.281960223626811e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.685681809000016
        },
        {
          "epoch": 420,
          "updates": 6720,
          "samples_seen": 430080,
          "online_loss": 0.0002566969696999877,
          "train_loss": 0.00024984381161630154,
          "validation_loss": 0.00022605742560699582,
          "train_metric": 2.5163357349811122e-05,
          "validation_metric": 2.2767679183743894e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.696160958999826
        },
        {
          "epoch": 421,
          "updates": 6736,
          "samples_seen": 431104,
          "online_loss": 0.000256795726272685,
          "train_loss": 0.0002517452521715313,
          "validation_loss": 0.00022587529383599758,
          "train_metric": 2.5354856916237622e-05,
          "validation_metric": 2.2749331037630327e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.706548959999054
        },
        {
          "epoch": 422,
          "updates": 6752,
          "samples_seen": 432128,
          "online_loss": 0.0002528074483052478,
          "train_loss": 0.00024804830900393426,
          "validation_loss": 0.00022393940889742225,
          "train_metric": 2.49825170612894e-05,
          "validation_metric": 2.2554357201443054e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.71710970499953
        },
        {
          "epoch": 423,
          "updates": 6768,
          "samples_seen": 433152,
          "online_loss": 0.00025176377766911173,
          "train_loss": 0.0002489887410774827,
          "validation_loss": 0.00022445936338044703,
          "train_metric": 2.5077233658521436e-05,
          "validation_metric": 2.2606724087381735e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.727454668998689
        },
        {
          "epoch": 424,
          "updates": 6784,
          "samples_seen": 434176,
          "online_loss": 0.00025125254342128756,
          "train_loss": 0.00024763308465480804,
          "validation_loss": 0.00022260608966462314,
          "train_metric": 2.4940696675912477e-05,
          "validation_metric": 2.2420068489736877e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.7385124919983355
        },
        {
          "epoch": 425,
          "updates": 6800,
          "samples_seen": 435200,
          "online_loss": 0.0002525818654248724,
          "train_loss": 0.00024780421517789364,
          "validation_loss": 0.00022300104319583625,
          "train_metric": 2.4957933419500478e-05,
          "validation_metric": 2.2459846150013618e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.748977800998546
        },
        {
          "epoch": 426,
          "updates": 6816,
          "samples_seen": 436224,
          "online_loss": 0.0002553794693085365,
          "train_loss": 0.0002471363986842334,
          "validation_loss": 0.00022326149337459356,
          "train_metric": 2.4890672648325562e-05,
          "validation_metric": 2.2486077796202153e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.759391408999363
        },
        {
          "epoch": 427,
          "updates": 6832,
          "samples_seen": 437248,
          "online_loss": 0.0002557467232691124,
          "train_loss": 0.00024989983648993075,
          "validation_loss": 0.00022882450139150023,
          "train_metric": 2.5168998035951518e-05,
          "validation_metric": 2.3046362912282348e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.769612050999058
        },
        {
          "epoch": 428,
          "updates": 6848,
          "samples_seen": 438272,
          "online_loss": 0.00026638318195182364,
          "train_loss": 0.00024844740983098745,
          "validation_loss": 0.00022270716726779938,
          "train_metric": 2.5022713089128956e-05,
          "validation_metric": 2.2430249373428524e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.779976665999129
        },
        {
          "epoch": 429,
          "updates": 6864,
          "samples_seen": 439296,
          "online_loss": 0.00025108300451393006,
          "train_loss": 0.0002505504817236215,
          "validation_loss": 0.00022633522166870534,
          "train_metric": 2.523452349123545e-05,
          "validation_metric": 2.2795657059759833e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.790230095999505
        },
        {
          "epoch": 430,
          "updates": 6880,
          "samples_seen": 440320,
          "online_loss": 0.0002515218893677229,
          "train_loss": 0.00024927471531555057,
          "validation_loss": 0.00022109737619757652,
          "train_metric": 2.5106037355726585e-05,
          "validation_metric": 2.2268115571932867e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.801097878998917
        },
        {
          "epoch": 431,
          "updates": 6896,
          "samples_seen": 441344,
          "online_loss": 0.00025083118180191377,
          "train_loss": 0.0002463195123709738,
          "validation_loss": 0.00022495683515444398,
          "train_metric": 2.4808399757603183e-05,
          "validation_metric": 2.265682996949181e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.811698952998995
        },
        {
          "epoch": 432,
          "updates": 6912,
          "samples_seen": 442368,
          "online_loss": 0.0002489158950993442,
          "train_loss": 0.0002486792509444058,
          "validation_loss": 0.00022445002105087042,
          "train_metric": 2.504606163711287e-05,
          "validation_metric": 2.26057818508707e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.822079714998836
        },
        {
          "epoch": 433,
          "updates": 6928,
          "samples_seen": 443392,
          "online_loss": 0.00025777418704819866,
          "train_loss": 0.0002453245979268104,
          "validation_loss": 0.00022483532666228712,
          "train_metric": 2.470819526934065e-05,
          "validation_metric": 2.2644591808784753e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.832671696998659
        },
        {
          "epoch": 434,
          "updates": 6944,
          "samples_seen": 444416,
          "online_loss": 0.000255186371759919,
          "train_loss": 0.00024525323533453047,
          "validation_loss": 0.00022153605823405087,
          "train_metric": 2.470100844220724e-05,
          "validation_metric": 2.2312298824544996e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.843470077999882
        },
        {
          "epoch": 435,
          "updates": 6960,
          "samples_seen": 445440,
          "online_loss": 0.00025591357461962616,
          "train_loss": 0.00024547689827159047,
          "validation_loss": 0.00022698049724567682,
          "train_metric": 2.472353844495956e-05,
          "validation_metric": 2.2860644094180316e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.853840377998495
        },
        {
          "epoch": 436,
          "updates": 6976,
          "samples_seen": 446464,
          "online_loss": 0.00025445380106248194,
          "train_loss": 0.00025435624411329627,
          "validation_loss": 0.00023080295068211854,
          "train_metric": 2.5617828214308247e-05,
          "validation_metric": 2.324562956346199e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.864331812999808
        },
        {
          "epoch": 437,
          "updates": 6992,
          "samples_seen": 447488,
          "online_loss": 0.00025663148608146,
          "train_loss": 0.0002463319106027484,
          "validation_loss": 0.00022070457634981722,
          "train_metric": 2.4809647584334016e-05,
          "validation_metric": 2.2228556190384552e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.8755275049988995
        },
        {
          "epoch": 438,
          "updates": 7008,
          "samples_seen": 448512,
          "online_loss": 0.00024839380967023317,
          "train_loss": 0.00024473731173202395,
          "validation_loss": 0.00022324742167256773,
          "train_metric": 2.464904355292674e-05,
          "validation_metric": 2.2484664441435598e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.886303456998576
        },
        {
          "epoch": 439,
          "updates": 7024,
          "samples_seen": 449536,
          "online_loss": 0.00024884587219276,
          "train_loss": 0.00024397825472988188,
          "validation_loss": 0.00021899644343648106,
          "train_metric": 2.4572595066274516e-05,
          "validation_metric": 2.2056519810575992e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.8982878389997495
        },
        {
          "epoch": 440,
          "updates": 7040,
          "samples_seen": 450560,
          "online_loss": 0.0002469018836563919,
          "train_loss": 0.00024413158826064318,
          "validation_loss": 0.00022317908587865531,
          "train_metric": 2.458803828631062e-05,
          "validation_metric": 2.247778138553258e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.909456974999557
        },
        {
          "epoch": 441,
          "updates": 7056,
          "samples_seen": 451584,
          "online_loss": 0.00024681415288796416,
          "train_loss": 0.00024355246569029987,
          "validation_loss": 0.0002214512787759304,
          "train_metric": 2.4529714210075326e-05,
          "validation_metric": 2.2303760488284752e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.980733143998805
        },
        {
          "epoch": 442,
          "updates": 7072,
          "samples_seen": 452608,
          "online_loss": 0.0002505246930013527,
          "train_loss": 0.00024546924396418035,
          "validation_loss": 0.00022198972874321043,
          "train_metric": 2.4722761736484244e-05,
          "validation_metric": 2.2357990019372664e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.99161817499953
        },
        {
          "epoch": 443,
          "updates": 7088,
          "samples_seen": 453632,
          "online_loss": 0.0002471850193614955,
          "train_loss": 0.00024356185167562217,
          "validation_loss": 0.00022631019237451255,
          "train_metric": 2.4530658265575767e-05,
          "validation_metric": 2.2793134121457115e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.001950861998921
        },
        {
          "epoch": 444,
          "updates": 7104,
          "samples_seen": 454656,
          "online_loss": 0.00025673872187326197,
          "train_loss": 0.0002441309334244579,
          "validation_loss": 0.0002181273594032973,
          "train_metric": 2.4587972802692093e-05,
          "validation_metric": 2.1968988221487962e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.013042035998296
        },
        {
          "epoch": 445,
          "updates": 7120,
          "samples_seen": 455680,
          "online_loss": 0.00025187253049807623,
          "train_loss": 0.00024366004799958318,
          "validation_loss": 0.00022422299662139267,
          "train_metric": 2.4540548110962845e-05,
          "validation_metric": 2.258291897305753e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.0235480289993575
        },
        {
          "epoch": 446,
          "updates": 7136,
          "samples_seen": 456704,
          "online_loss": 0.0002468663915351499,
          "train_loss": 0.00024260548525489867,
          "validation_loss": 0.00021822591952513903,
          "train_metric": 2.44343355007004e-05,
          "validation_metric": 2.197891444666311e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.034249056998306
        },
        {
          "epoch": 447,
          "updates": 7152,
          "samples_seen": 457728,
          "online_loss": 0.00025453120997553924,
          "train_loss": 0.000244297698372975,
          "validation_loss": 0.00022106835967861116,
          "train_metric": 2.460477116983384e-05,
          "validation_metric": 2.2265196093940176e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.044732881999153
        },
        {
          "epoch": 448,
          "updates": 7168,
          "samples_seen": 458752,
          "online_loss": 0.00024798310005280655,
          "train_loss": 0.00024659177870489657,
          "validation_loss": 0.00022127592819742858,
          "train_metric": 2.4835819203872234e-05,
          "validation_metric": 2.2286103558144532e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.055223401999683
        },
        {
          "epoch": 449,
          "updates": 7184,
          "samples_seen": 459776,
          "online_loss": 0.0002496212100595585,
          "train_loss": 0.00024177647719625384,
          "validation_loss": 0.00022306665778160095,
          "train_metric": 2.4350843887077644e-05,
          "validation_metric": 2.2466456357506104e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.065891867998289
        },
        {
          "epoch": 450,
          "updates": 7200,
          "samples_seen": 460800,
          "online_loss": 0.00024581837624282343,
          "train_loss": 0.00024160621978808194,
          "validation_loss": 0.00021947291679680347,
          "train_metric": 2.4333690817002207e-05,
          "validation_metric": 2.2104506570030935e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.076404840998293
        },
        {
          "epoch": 451,
          "updates": 7216,
          "samples_seen": 461824,
          "online_loss": 0.00025127689877990633,
          "train_loss": 0.00024122977629303932,
          "validation_loss": 0.0002190226223319769,
          "train_metric": 2.4295779439853504e-05,
          "validation_metric": 2.2059157345211133e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.086903690998952
        },
        {
          "epoch": 452,
          "updates": 7232,
          "samples_seen": 462848,
          "online_loss": 0.0002546472296671709,
          "train_loss": 0.00024171368568204343,
          "validation_loss": 0.000222994145588018,
          "train_metric": 2.4344519260921516e-05,
          "validation_metric": 2.2459151296061464e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.097361836998971
        },
        {
          "epoch": 453,
          "updates": 7248,
          "samples_seen": 463872,
          "online_loss": 0.0002520781654311577,
          "train_loss": 0.00024094183754641563,
          "validation_loss": 0.00021750424639321864,
          "train_metric": 2.4266781110782176e-05,
          "validation_metric": 2.190623308706563e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.108341815999665
        },
        {
          "epoch": 454,
          "updates": 7264,
          "samples_seen": 464896,
          "online_loss": 0.00024910039610404056,
          "train_loss": 0.0002430659660603851,
          "validation_loss": 0.00021726700651925057,
          "train_metric": 2.4480714273522608e-05,
          "validation_metric": 2.1882335204281844e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.119432461999168
        },
        {
          "epoch": 455,
          "updates": 7280,
          "samples_seen": 465920,
          "online_loss": 0.0002448938503221143,
          "train_loss": 0.00024196022422984242,
          "validation_loss": 0.00022498948965221643,
          "train_metric": 2.4369348466279916e-05,
          "validation_metric": 2.2660116883344017e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.130027602999689
        },
        {
          "epoch": 456,
          "updates": 7296,
          "samples_seen": 466944,
          "online_loss": 0.0002485010900272755,
          "train_loss": 0.00024125579511746764,
          "validation_loss": 0.0002144592144759372,
          "train_metric": 2.429839878459461e-05,
          "validation_metric": 2.159954783564899e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.1411474689994066
        },
        {
          "epoch": 457,
          "updates": 7312,
          "samples_seen": 467968,
          "online_loss": 0.0002519672761991387,
          "train_loss": 0.00024119170848280191,
          "validation_loss": 0.0002203293115599081,
          "train_metric": 2.4291946829180233e-05,
          "validation_metric": 2.2190766685525887e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.151704483998401
        },
        {
          "epoch": 458,
          "updates": 7328,
          "samples_seen": 468992,
          "online_loss": 0.0002574521158749121,
          "train_loss": 0.00024958536960184574,
          "validation_loss": 0.00023144636361394078,
          "train_metric": 2.5137323973467574e-05,
          "validation_metric": 2.3310429241973907e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.162340502000006
        },
        {
          "epoch": 459,
          "updates": 7344,
          "samples_seen": 470016,
          "online_loss": 0.0002615091807456338,
          "train_loss": 0.00024007535830605775,
          "validation_loss": 0.00021834659855812788,
          "train_metric": 2.4179509637178853e-05,
          "validation_metric": 2.1991068933857605e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.172699267999633
        },
        {
          "epoch": 460,
          "updates": 7360,
          "samples_seen": 471040,
          "online_loss": 0.000248198484769091,
          "train_loss": 0.0002398690558038652,
          "validation_loss": 0.0002185445191571489,
          "train_metric": 2.4158733140211552e-05,
          "validation_metric": 2.2011008695699275e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.183246743999916
        },
        {
          "epoch": 461,
          "updates": 7376,
          "samples_seen": 472064,
          "online_loss": 0.0002524703040762688,
          "train_loss": 0.00024108801153488457,
          "validation_loss": 0.00022021975019015372,
          "train_metric": 2.428150037303567e-05,
          "validation_metric": 2.2179727238835767e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.193597689998569
        },
        {
          "epoch": 462,
          "updates": 7392,
          "samples_seen": 473088,
          "online_loss": 0.0002605724748718785,
          "train_loss": 0.00024089802172966301,
          "validation_loss": 0.0002167889615520835,
          "train_metric": 2.4262368242489174e-05,
          "validation_metric": 2.183418837375939e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.204109649999737
        },
        {
          "epoch": 463,
          "updates": 7408,
          "samples_seen": 474112,
          "online_loss": 0.00024321168712049257,
          "train_loss": 0.00024645126541145146,
          "validation_loss": 0.00023056699137669057,
          "train_metric": 2.4821672923280858e-05,
          "validation_metric": 2.322186264791526e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.214548009998907
        },
        {
          "epoch": 464,
          "updates": 7424,
          "samples_seen": 475136,
          "online_loss": 0.00024907284296205034,
          "train_loss": 0.0002416962233837694,
          "validation_loss": 0.00021875512902624905,
          "train_metric": 2.434275666018948e-05,
          "validation_metric": 2.2032214474165812e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.224916238999867
        },
        {
          "epoch": 465,
          "updates": 7440,
          "samples_seen": 476160,
          "online_loss": 0.0002394622570136562,
          "train_loss": 0.00024198743631131947,
          "validation_loss": 0.0002230655518360436,
          "train_metric": 2.4372087864321657e-05,
          "validation_metric": 2.2466349037131295e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.23530847299844
        },
        {
          "epoch": 466,
          "updates": 7456,
          "samples_seen": 477184,
          "online_loss": 0.00024803845553833526,
          "train_loss": 0.00023803295334801078,
          "validation_loss": 0.0002176499110646546,
          "train_metric": 2.3973807401489466e-05,
          "validation_metric": 2.1920903236605227e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.24580159299876
        },
        {
          "epoch": 467,
          "updates": 7472,
          "samples_seen": 478208,
          "online_loss": 0.00024251118611573474,
          "train_loss": 0.0002380351652391255,
          "validation_loss": 0.0002157790877390653,
          "train_metric": 2.3974027499207295e-05,
          "validation_metric": 2.173247776227072e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.256086901999879
        },
        {
          "epoch": 468,
          "updates": 7488,
          "samples_seen": 479232,
          "online_loss": 0.0002428985735605238,
          "train_loss": 0.0002403146354481578,
          "validation_loss": 0.00021994607232045382,
          "train_metric": 2.420360760879703e-05,
          "validation_metric": 2.215216045442503e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.26642587599963
        },
        {
          "epoch": 469,
          "updates": 7504,
          "samples_seen": 480256,
          "online_loss": 0.0002435552282804565,
          "train_loss": 0.00023731793044134974,
          "validation_loss": 0.0002164396137231961,
          "train_metric": 2.3901791792013682e-05,
          "validation_metric": 2.1799005480716005e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.276946014999339
        },
        {
          "epoch": 470,
          "updates": 7520,
          "samples_seen": 481280,
          "online_loss": 0.0002399859640718205,
          "train_loss": 0.0002400726662017405,
          "validation_loss": 0.00021945305343251675,
          "train_metric": 2.417923678876832e-05,
          "validation_metric": 2.2102507500676438e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.2872752959992795
        },
        {
          "epoch": 471,
          "updates": 7536,
          "samples_seen": 482304,
          "online_loss": 0.00024353738808713388,
          "train_loss": 0.0002368013228988275,
          "validation_loss": 0.00021718896459788084,
          "train_metric": 2.3849761419114657e-05,
          "validation_metric": 2.1874475351069123e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.297544767998261
        },
        {
          "epoch": 472,
          "updates": 7552,
          "samples_seen": 483328,
          "online_loss": 0.0002474320763212745,
          "train_loss": 0.00023918639635667205,
          "validation_loss": 0.00022031775733921677,
          "train_metric": 2.408998079772573e-05,
          "validation_metric": 2.2189597075339407e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.308191596999677
        },
        {
          "epoch": 473,
          "updates": 7568,
          "samples_seen": 484352,
          "online_loss": 0.00024488479903084226,
          "train_loss": 0.00023648090427741408,
          "validation_loss": 0.0002138920535799116,
          "train_metric": 2.3817492547095753e-05,
          "validation_metric": 2.154242611140944e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.319079311999303
        },
        {
          "epoch": 474,
          "updates": 7584,
          "samples_seen": 485376,
          "online_loss": 0.00023940912069519982,
          "train_loss": 0.00023625064932275563,
          "validation_loss": 0.000217477761907503,
          "train_metric": 2.3794296794221736e-05,
          "validation_metric": 2.1903564629610628e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.329617129998951
        },
        {
          "epoch": 475,
          "updates": 7600,
          "samples_seen": 486400,
          "online_loss": 0.0002503120504115941,
          "train_loss": 0.00024381413822993636,
          "validation_loss": 0.00021598492457997054,
          "train_metric": 2.4556065909564495e-05,
          "validation_metric": 2.1753212422481738e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.340266975999839
        },
        {
          "epoch": 476,
          "updates": 7616,
          "samples_seen": 487424,
          "online_loss": 0.0002473856875440106,
          "train_loss": 0.0002401381207164377,
          "validation_loss": 0.00022867113875690848,
          "train_metric": 2.418583244434558e-05,
          "validation_metric": 2.3030919692246243e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.350923115998739
        },
        {
          "epoch": 477,
          "updates": 7632,
          "samples_seen": 488448,
          "online_loss": 0.00025141156220342964,
          "train_loss": 0.00023595949460286647,
          "validation_loss": 0.00021429748449008912,
          "train_metric": 2.376497650402598e-05,
          "validation_metric": 2.1583256966550834e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.418103551999593
        },
        {
          "epoch": 478,
          "updates": 7648,
          "samples_seen": 489472,
          "online_loss": 0.00024166373532352736,
          "train_loss": 0.00023811522987671196,
          "validation_loss": 0.00021812939667142928,
          "train_metric": 2.398209471721202e-05,
          "validation_metric": 2.1969195586279966e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.428786154998306
        },
        {
          "epoch": 479,
          "updates": 7664,
          "samples_seen": 490496,
          "online_loss": 0.0002454465347909718,
          "train_loss": 0.000235527622862719,
          "validation_loss": 0.0002157227136194706,
          "train_metric": 2.372147901041899e-05,
          "validation_metric": 2.1726802515331656e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.439207963998342
        },
        {
          "epoch": 480,
          "updates": 7680,
          "samples_seen": 491520,
          "online_loss": 0.0002426730634397245,
          "train_loss": 0.00023560243425890803,
          "validation_loss": 0.00021652186114806682,
          "train_metric": 2.372901508351788e-05,
          "validation_metric": 2.1807289158459753e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.449684126999273
        },
        {
          "epoch": 481,
          "updates": 7696,
          "samples_seen": 492544,
          "online_loss": 0.0002398188853476313,
          "train_loss": 0.0002346825785934925,
          "validation_loss": 0.0002123607264366001,
          "train_metric": 2.3636368496227078e-05,
          "validation_metric": 2.1388193999882787e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.460612682998544
        },
        {
          "epoch": 482,
          "updates": 7712,
          "samples_seen": 493568,
          "online_loss": 0.00023898583276604768,
          "train_loss": 0.00023544311989098787,
          "validation_loss": 0.00022057932801544666,
          "train_metric": 2.3712967958999798e-05,
          "validation_metric": 2.221594149887096e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.4711149239992665
        },
        {
          "epoch": 483,
          "updates": 7728,
          "samples_seen": 494592,
          "online_loss": 0.00023989304008864565,
          "train_loss": 0.0002349127025809139,
          "validation_loss": 0.00021437412942759693,
          "train_metric": 2.3659547878196463e-05,
          "validation_metric": 2.1590978576568887e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.4815490179989865
        },
        {
          "epoch": 484,
          "updates": 7744,
          "samples_seen": 495616,
          "online_loss": 0.00023794150820322102,
          "train_loss": 0.0002343103988096118,
          "validation_loss": 0.00021267225383780897,
          "train_metric": 2.359888458158821e-05,
          "validation_metric": 2.141956974810455e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.491978233998452
        },
        {
          "epoch": 485,
          "updates": 7760,
          "samples_seen": 496640,
          "online_loss": 0.00023863735896156868,
          "train_loss": 0.0002335377794224769,
          "validation_loss": 0.00021363870473578572,
          "train_metric": 2.3521068214904517e-05,
          "validation_metric": 2.1516905690077692e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.502277904999573
        },
        {
          "epoch": 486,
          "updates": 7776,
          "samples_seen": 497664,
          "online_loss": 0.0002436536951790913,
          "train_loss": 0.00023468861763831228,
          "validation_loss": 0.00021921748702879995,
          "train_metric": 2.363697421969846e-05,
          "validation_metric": 2.207878242188599e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.513053886999842
        },
        {
          "epoch": 487,
          "updates": 7792,
          "samples_seen": 498688,
          "online_loss": 0.00023669697384320898,
          "train_loss": 0.0002335273165954277,
          "validation_loss": 0.0002119339187629521,
          "train_metric": 2.3520016839029267e-05,
          "validation_metric": 2.1345207642298192e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.524013278998609
        },
        {
          "epoch": 488,
          "updates": 7808,
          "samples_seen": 499712,
          "online_loss": 0.00023586877341585932,
          "train_loss": 0.00023355499433819205,
          "validation_loss": 0.000213694351259619,
          "train_metric": 2.3522805349784903e-05,
          "validation_metric": 2.1522513634408824e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.534501900998293
        },
        {
          "epoch": 489,
          "updates": 7824,
          "samples_seen": 500736,
          "online_loss": 0.00023510616938438034,
          "train_loss": 0.00023280017194338143,
          "validation_loss": 0.0002143159363185987,
          "train_metric": 2.3446777049684897e-05,
          "validation_metric": 2.1585114154731855e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.545118197998818
        },
        {
          "epoch": 490,
          "updates": 7840,
          "samples_seen": 501760,
          "online_loss": 0.00023514507392974338,
          "train_loss": 0.00023246798082254827,
          "validation_loss": 0.0002155234251404181,
          "train_metric": 2.3413320377585478e-05,
          "validation_metric": 2.170673178625293e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.555597561999093
        },
        {
          "epoch": 491,
          "updates": 7856,
          "samples_seen": 502784,
          "online_loss": 0.0002392593482909433,
          "train_loss": 0.0002325962414033711,
          "validation_loss": 0.0002122546429745853,
          "train_metric": 2.3426242478308268e-05,
          "validation_metric": 2.137750925612636e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.566367247998642
        },
        {
          "epoch": 492,
          "updates": 7872,
          "samples_seen": 503808,
          "online_loss": 0.0002409491607977543,
          "train_loss": 0.00023401349608320743,
          "validation_loss": 0.00021852264762856066,
          "train_metric": 2.3568984033772722e-05,
          "validation_metric": 2.2008802261552773e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.576995026998702
        },
        {
          "epoch": 493,
          "updates": 7888,
          "samples_seen": 504832,
          "online_loss": 0.0002421278095425805,
          "train_loss": 0.00023515689827036113,
          "validation_loss": 0.00021656809258274734,
          "train_metric": 2.3684140614932403e-05,
          "validation_metric": 2.1811943952343427e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.587311897999825
        },
        {
          "epoch": 494,
          "updates": 7904,
          "samples_seen": 505856,
          "online_loss": 0.00023464534115191782,
          "train_loss": 0.00023738195886835456,
          "validation_loss": 0.00021885767637286335,
          "train_metric": 2.390824374742806e-05,
          "validation_metric": 2.204254451498855e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.597671358998923
        },
        {
          "epoch": 495,
          "updates": 7920,
          "samples_seen": 506880,
          "online_loss": 0.00023688714099989738,
          "train_loss": 0.0002334969467483461,
          "validation_loss": 0.00021582149202004075,
          "train_metric": 2.3516957298852503e-05,
          "validation_metric": 2.173675238736905e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.608245581999654
        },
        {
          "epoch": 496,
          "updates": 7936,
          "samples_seen": 507904,
          "online_loss": 0.0002507097760826582,
          "train_loss": 0.00024182419292628765,
          "validation_loss": 0.0002201594616053626,
          "train_metric": 2.4355649657081813e-05,
          "validation_metric": 2.2173651814227924e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.619296962999215
        },
        {
          "epoch": 497,
          "updates": 7952,
          "samples_seen": 508928,
          "online_loss": 0.00024907765691750683,
          "train_loss": 0.0002320482162758708,
          "validation_loss": 0.0002110652712872252,
          "train_metric": 2.3371047063847072e-05,
          "validation_metric": 2.125772152794525e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.6302898449994245
        },
        {
          "epoch": 498,
          "updates": 7968,
          "samples_seen": 509952,
          "online_loss": 0.0002537641566959792,
          "train_loss": 0.00023467684513889253,
          "validation_loss": 0.00022009828535374254,
          "train_metric": 2.363579005759675e-05,
          "validation_metric": 2.216749453509692e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.641213931998209
        },
        {
          "epoch": 499,
          "updates": 7984,
          "samples_seen": 510976,
          "online_loss": 0.0002370052980040782,
          "train_loss": 0.00023164160666055977,
          "validation_loss": 0.00020934856729581952,
          "train_metric": 2.3330096155405045e-05,
          "validation_metric": 2.1084821128170006e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.6521041769992735
        },
        {
          "epoch": 500,
          "updates": 8000,
          "samples_seen": 512000,
          "online_loss": 0.00024100852715491783,
          "train_loss": 0.00023226338089443743,
          "validation_loss": 0.00021634223230648786,
          "train_metric": 2.3392718503600918e-05,
          "validation_metric": 2.178919930884149e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.662806092999745
        },
        {
          "epoch": 501,
          "updates": 8016,
          "samples_seen": 513024,
          "online_loss": 0.00023843573490012204,
          "train_loss": 0.0002310812269570306,
          "validation_loss": 0.00021092120732646435,
          "train_metric": 2.327365473320242e-05,
          "validation_metric": 2.124320963048376e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.67329751699981
        },
        {
          "epoch": 502,
          "updates": 8032,
          "samples_seen": 514048,
          "online_loss": 0.00023479516403313028,
          "train_loss": 0.00023117168166209012,
          "validation_loss": 0.00021796082728542387,
          "train_metric": 2.328276605112478e-05,
          "validation_metric": 2.1952213501208462e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.683632230999137
        },
        {
          "epoch": 503,
          "updates": 8048,
          "samples_seen": 515072,
          "online_loss": 0.0002355616807108163,
          "train_loss": 0.00022954911401029676,
          "validation_loss": 0.00021202080824878067,
          "train_metric": 2.3119344405131415e-05,
          "validation_metric": 2.135395880031865e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.693962349998401
        },
        {
          "epoch": 504,
          "updates": 8064,
          "samples_seen": 516096,
          "online_loss": 0.00024124945048242807,
          "train_loss": 0.00023293129925150424,
          "validation_loss": 0.00021439726697281003,
          "train_metric": 2.3459986550733447e-05,
          "validation_metric": 2.1593306883005425e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.7043617439994705
        },
        {
          "epoch": 505,
          "updates": 8080,
          "samples_seen": 517120,
          "online_loss": 0.00023301072724279948,
          "train_loss": 0.00023167167091742158,
          "validation_loss": 0.00021825706062372774,
          "train_metric": 2.3333122953772545e-05,
          "validation_metric": 2.1982052203384228e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.714871008998671
        },
        {
          "epoch": 506,
          "updates": 8096,
          "samples_seen": 518144,
          "online_loss": 0.00023623658853466623,
          "train_loss": 0.00022977670596446842,
          "validation_loss": 0.0002121766156051308,
          "train_metric": 2.3142269128584303e-05,
          "validation_metric": 2.1369653040892445e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.725462693999361
        },
        {
          "epoch": 507,
          "updates": 8112,
          "samples_seen": 519168,
          "online_loss": 0.0002562661666161148,
          "train_loss": 0.00023827115364838392,
          "validation_loss": 0.00021264373208396137,
          "train_metric": 2.3997798052732833e-05,
          "validation_metric": 2.141669756383635e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.735817519998818
        },
        {
          "epoch": 508,
          "updates": 8128,
          "samples_seen": 520192,
          "online_loss": 0.00022617682861891808,
          "train_loss": 0.00023243222676683217,
          "validation_loss": 0.0002173142711399123,
          "train_metric": 2.3409718778566457e-05,
          "validation_metric": 2.1887097318540327e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.74654886999997
        },
        {
          "epoch": 509,
          "updates": 8144,
          "samples_seen": 521216,
          "online_loss": 0.00023879124910308747,
          "train_loss": 0.00022931929561309516,
          "validation_loss": 0.00021019010455347598,
          "train_metric": 2.3096199583960697e-05,
          "validation_metric": 2.1169576939428225e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.756905077998454
        },
        {
          "epoch": 510,
          "updates": 8160,
          "samples_seen": 522240,
          "online_loss": 0.0002323232238268247,
          "train_loss": 0.00022893337882123888,
          "validation_loss": 0.00021008892508689314,
          "train_metric": 2.3057333237375133e-05,
          "validation_metric": 2.1159388779778965e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.767190643999129
        },
        {
          "epoch": 511,
          "updates": 8176,
          "samples_seen": 523264,
          "online_loss": 0.0002393927970842924,
          "train_loss": 0.00022799393627792597,
          "validation_loss": 0.00021230039419606328,
          "train_metric": 2.2962714865570888e-05,
          "validation_metric": 2.138211675628554e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.778008482999212
        },
        {
          "epoch": 512,
          "updates": 8192,
          "samples_seen": 524288,
          "online_loss": 0.00024143975861079525,
          "train_loss": 0.0002316619793418795,
          "validation_loss": 0.00021173972345422953,
          "train_metric": 2.333214615646284e-05,
          "validation_metric": 2.1325649868231267e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.788749501998609
        },
        {
          "epoch": 513,
          "updates": 8208,
          "samples_seen": 525312,
          "online_loss": 0.00023783872893545777,
          "train_loss": 0.00023524863354396075,
          "validation_loss": 0.00021802508854307234,
          "train_metric": 2.3693379262113012e-05,
          "validation_metric": 2.195868728449568e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.855991164999068
        },
        {
          "epoch": 514,
          "updates": 8224,
          "samples_seen": 526336,
          "online_loss": 0.00024523209413018776,
          "train_loss": 0.00022991569130681455,
          "validation_loss": 0.00020871420565526932,
          "train_metric": 2.3156268071033992e-05,
          "validation_metric": 2.1020929125370458e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.867208364999897
        },
        {
          "epoch": 515,
          "updates": 8240,
          "samples_seen": 527360,
          "online_loss": 0.0002333168877157732,
          "train_loss": 0.00022716994863003492,
          "validation_loss": 0.00021310690499376506,
          "train_metric": 2.2879725293023512e-05,
          "validation_metric": 2.1463345547090285e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.877861812999981
        },
        {
          "epoch": 516,
          "updates": 8256,
          "samples_seen": 528384,
          "online_loss": 0.0002295985314049176,
          "train_loss": 0.00022848779917694628,
          "validation_loss": 0.00021125130297150463,
          "train_metric": 2.3012453311821446e-05,
          "validation_metric": 2.1276458937791176e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.8883938519993535
        },
        {
          "epoch": 517,
          "updates": 8272,
          "samples_seen": 529408,
          "online_loss": 0.00023123260598367779,
          "train_loss": 0.00022784408065490425,
          "validation_loss": 0.0002124339807778597,
          "train_metric": 2.2947620891500264e-05,
          "validation_metric": 2.1395573639892973e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.898629669000002
        },
        {
          "epoch": 518,
          "updates": 8288,
          "samples_seen": 530432,
          "online_loss": 0.0002308665907548857,
          "train_loss": 0.00022643311240244657,
          "validation_loss": 0.00020851512090303004,
          "train_metric": 2.2805514163337648e-05,
          "validation_metric": 2.1000880224164575e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.909960129998581
        },
        {
          "epoch": 519,
          "updates": 8304,
          "samples_seen": 531456,
          "online_loss": 0.00023099557438399643,
          "train_loss": 0.0002288590621901676,
          "validation_loss": 0.00020903699623886496,
          "train_metric": 2.3049846276990138e-05,
          "validation_metric": 2.1053439922980033e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.920426359998601
        },
        {
          "epoch": 520,
          "updates": 8320,
          "samples_seen": 532480,
          "online_loss": 0.0002309805749973748,
          "train_loss": 0.0002271516277687624,
          "validation_loss": 0.00020863008103333414,
          "train_metric": 2.2877880837768316e-05,
          "validation_metric": 2.101245627272874e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.930895986999531
        },
        {
          "epoch": 521,
          "updates": 8336,
          "samples_seen": 533504,
          "online_loss": 0.00023318480407397146,
          "train_loss": 0.00022657093359157443,
          "validation_loss": 0.00021094865223858505,
          "train_metric": 2.2819394871476106e-05,
          "validation_metric": 2.1245976313366555e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.941254828998353
        },
        {
          "epoch": 522,
          "updates": 8352,
          "samples_seen": 534528,
          "online_loss": 0.00022971294038143242,
          "train_loss": 0.00022710012854076922,
          "validation_loss": 0.00020981873967684805,
          "train_metric": 2.2872691261e-05,
          "validation_metric": 2.1132174879312515e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.951577601999816
        },
        {
          "epoch": 523,
          "updates": 8368,
          "samples_seen": 535552,
          "online_loss": 0.0002315738065590267,
          "train_loss": 0.00022850465029478073,
          "validation_loss": 0.0002110269124386832,
          "train_metric": 2.3014152247924358e-05,
          "validation_metric": 2.1253859813441522e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.961736439998276
        },
        {
          "epoch": 524,
          "updates": 8384,
          "samples_seen": 536576,
          "online_loss": 0.00023359804799838457,
          "train_loss": 0.00022736108803655952,
          "validation_loss": 0.00020679582667071372,
          "train_metric": 2.289897747687064e-05,
          "validation_metric": 2.082771788991522e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.972753947999081
        },
        {
          "epoch": 525,
          "updates": 8400,
          "samples_seen": 537600,
          "online_loss": 0.0002319719296792755,
          "train_loss": 0.00022554403403773904,
          "validation_loss": 0.0002073128125630319,
          "train_metric": 2.271596713399049e-05,
          "validation_metric": 2.087978646159172e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.983199013999183
        },
        {
          "epoch": 526,
          "updates": 8416,
          "samples_seen": 538624,
          "online_loss": 0.00022922424250282347,
          "train_loss": 0.00022479906328953803,
          "validation_loss": 0.00021023293083999306,
          "train_metric": 2.2640939278062433e-05,
          "validation_metric": 2.1173891582293436e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.993637959998523
        },
        {
          "epoch": 527,
          "updates": 8432,
          "samples_seen": 539648,
          "online_loss": 0.00023007459731161362,
          "train_loss": 0.00022512713621836156,
          "validation_loss": 0.00020659089204855263,
          "train_metric": 2.2673983039567247e-05,
          "validation_metric": 2.0807077817153186e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.00453428399851
        },
        {
          "epoch": 528,
          "updates": 8448,
          "samples_seen": 540672,
          "online_loss": 0.0002443976736685727,
          "train_loss": 0.0002403864054940641,
          "validation_loss": 0.00022059230832383037,
          "train_metric": 2.421083627268672e-05,
          "validation_metric": 2.2217252990230918e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.015079142998729
        },
        {
          "epoch": 529,
          "updates": 8464,
          "samples_seen": 541696,
          "online_loss": 0.00023642882661079057,
          "train_loss": 0.00023284881899598986,
          "validation_loss": 0.00021337861835490912,
          "train_metric": 2.3451679226127453e-05,
          "validation_metric": 2.1490714061656035e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.025460469998507
        },
        {
          "epoch": 530,
          "updates": 8480,
          "samples_seen": 542720,
          "online_loss": 0.00023438706102751894,
          "train_loss": 0.00022408923541661352,
          "validation_loss": 0.00020519607642199844,
          "train_metric": 2.2569443899556063e-05,
          "validation_metric": 2.0666599084506743e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.036574148998625
        },
        {
          "epoch": 531,
          "updates": 8496,
          "samples_seen": 543744,
          "online_loss": 0.00022934597382118227,
          "train_loss": 0.0002252784324809909,
          "validation_loss": 0.0002133663510903716,
          "train_metric": 2.2689218894811347e-05,
          "validation_metric": 2.1489477148861624e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.0470545459993446
        },
        {
          "epoch": 532,
          "updates": 8512,
          "samples_seen": 544768,
          "online_loss": 0.00022741025804862147,
          "train_loss": 0.000224228497245349,
          "validation_loss": 0.00020773708820343018,
          "train_metric": 2.2583470126846805e-05,
          "validation_metric": 2.0922519979649223e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.057542722999642
        },
        {
          "epoch": 533,
          "updates": 8528,
          "samples_seen": 545792,
          "online_loss": 0.00022864219045004575,
          "train_loss": 0.00022345472825691104,
          "validation_loss": 0.00020688402582891285,
          "train_metric": 2.25055446207989e-05,
          "validation_metric": 2.0836600015172735e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.067824189998646
        },
        {
          "epoch": 534,
          "updates": 8544,
          "samples_seen": 546816,
          "online_loss": 0.0002309929163857305,
          "train_loss": 0.00022351107327267528,
          "validation_loss": 0.0002112788351951167,
          "train_metric": 2.2511216229759157e-05,
          "validation_metric": 2.127923107764218e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.078318104999198
        },
        {
          "epoch": 535,
          "updates": 8560,
          "samples_seen": 547840,
          "online_loss": 0.00022649334096058737,
          "train_loss": 0.00022269895998761058,
          "validation_loss": 0.0002062367129838094,
          "train_metric": 2.2429423552239314e-05,
          "validation_metric": 2.0771405615960248e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.0885781009983475
        },
        {
          "epoch": 536,
          "updates": 8576,
          "samples_seen": 548864,
          "online_loss": 0.00022820413960289443,
          "train_loss": 0.00022321179858408868,
          "validation_loss": 0.00020755958394147456,
          "train_metric": 2.2481073756353e-05,
          "validation_metric": 2.090464113280177e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.098919215999558
        },
        {
          "epoch": 537,
          "updates": 8592,
          "samples_seen": 549888,
          "online_loss": 0.000241616178300319,
          "train_loss": 0.00023169720952864736,
          "validation_loss": 0.00020961443078704178,
          "train_metric": 2.333569500478916e-05,
          "validation_metric": 2.11115966521902e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.109468056998594
        },
        {
          "epoch": 538,
          "updates": 8608,
          "samples_seen": 550912,
          "online_loss": 0.00023086393684934592,
          "train_loss": 0.0002277009771205485,
          "validation_loss": 0.00021614067372865975,
          "train_metric": 2.293320903845597e-05,
          "validation_metric": 2.1768897568108514e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.119945015998383
        },
        {
          "epoch": 539,
          "updates": 8624,
          "samples_seen": 551936,
          "online_loss": 0.00023150101696955971,
          "train_loss": 0.00022275725496001542,
          "validation_loss": 0.0002055877266684547,
          "train_metric": 2.2435297069023363e-05,
          "validation_metric": 2.0706043869722635e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.130238669999017
        },
        {
          "epoch": 540,
          "updates": 8640,
          "samples_seen": 552960,
          "online_loss": 0.0002294355545018334,
          "train_loss": 0.0002260667533846572,
          "validation_loss": 0.0002101197896990925,
          "train_metric": 2.276861596328672e-05,
          "validation_metric": 2.1162493794690818e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.140706126998339
        },
        {
          "epoch": 541,
          "updates": 8656,
          "samples_seen": 553984,
          "online_loss": 0.00022597721272177296,
          "train_loss": 0.00022655792417936027,
          "validation_loss": 0.00021551332611124963,
          "train_metric": 2.281808338011615e-05,
          "validation_metric": 2.1705713152186945e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.151099360999069
        },
        {
          "epoch": 542,
          "updates": 8672,
          "samples_seen": 555008,
          "online_loss": 0.00023540334677818464,
          "train_loss": 0.00022199036902748048,
          "validation_loss": 0.0002032608026638627,
          "train_metric": 2.235805368400179e-05,
          "validation_metric": 2.0471685274969786e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.162070615999255
        },
        {
          "epoch": 543,
          "updates": 8688,
          "samples_seen": 556032,
          "online_loss": 0.00022683625957142795,
          "train_loss": 0.00022422766778618097,
          "validation_loss": 0.00020774368022102863,
          "train_metric": 2.2583393729291856e-05,
          "validation_metric": 2.0923183910781518e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.17258146399945
        },
        {
          "epoch": 544,
          "updates": 8704,
          "samples_seen": 557056,
          "online_loss": 0.00022369316684489604,
          "train_loss": 0.00022213972988538444,
          "validation_loss": 0.0002055553050013259,
          "train_metric": 2.237310036434792e-05,
          "validation_metric": 2.0702776964753866e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.183035962998474
        },
        {
          "epoch": 545,
          "updates": 8720,
          "samples_seen": 558080,
          "online_loss": 0.0002225188136435463,
          "train_loss": 0.00022086525859776884,
          "validation_loss": 0.00020825125102419406,
          "train_metric": 2.22447397391079e-05,
          "validation_metric": 2.097430478897877e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.19332445299915
        },
        {
          "epoch": 546,
          "updates": 8736,
          "samples_seen": 559104,
          "online_loss": 0.00022721147888660198,
          "train_loss": 0.00022293921210803092,
          "validation_loss": 0.0002071884518954903,
          "train_metric": 2.2453621568274684e-05,
          "validation_metric": 2.0867260900558904e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.203726433999691
        },
        {
          "epoch": 547,
          "updates": 8752,
          "samples_seen": 560128,
          "online_loss": 0.00022873446960147703,
          "train_loss": 0.00022194586927071214,
          "validation_loss": 0.0002075220545521006,
          "train_metric": 2.2353577151079662e-05,
          "validation_metric": 2.09008594538318e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.214443713999572
        },
        {
          "epoch": 548,
          "updates": 8768,
          "samples_seen": 561152,
          "online_loss": 0.00022548446577275172,
          "train_loss": 0.00022349283972289413,
          "validation_loss": 0.0002090524067170918,
          "train_metric": 2.250938086945098e-05,
          "validation_metric": 2.1054995158920065e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.225209599999289
        },
        {
          "epoch": 549,
          "updates": 8784,
          "samples_seen": 562176,
          "online_loss": 0.0002263002697873162,
          "train_loss": 0.0002277857274748385,
          "validation_loss": 0.0002123151789419353,
          "train_metric": 2.2941741917748004e-05,
          "validation_metric": 2.138360832759645e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.292309688999012
        },
        {
          "epoch": 550,
          "updates": 8800,
          "samples_seen": 563200,
          "online_loss": 0.0002216749380750116,
          "train_loss": 0.00022273004287853837,
          "validation_loss": 0.0002050532930297777,
          "train_metric": 2.243255585199222e-05,
          "validation_metric": 2.0652216335292906e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.303025628998512
        },
        {
          "epoch": 551,
          "updates": 8816,
          "samples_seen": 564224,
          "online_loss": 0.00024257604400190758,
          "train_loss": 0.00023151890491135418,
          "validation_loss": 0.00021393875067587942,
          "train_metric": 2.3317734303418547e-05,
          "validation_metric": 2.1547128199017607e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.313902545998644
        },
        {
          "epoch": 552,
          "updates": 8832,
          "samples_seen": 565248,
          "online_loss": 0.00022798421650804812,
          "train_loss": 0.00021948126959614456,
          "validation_loss": 0.0002024807472480461,
          "train_metric": 2.210535058111418e-05,
          "validation_metric": 2.0393119484651834e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.325083841999003
        },
        {
          "epoch": 553,
          "updates": 8848,
          "samples_seen": 566272,
          "online_loss": 0.00022231323964660987,
          "train_loss": 0.00021958185243420303,
          "validation_loss": 0.00020965764997527003,
          "train_metric": 2.211548235209193e-05,
          "validation_metric": 2.111595131282229e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.335498109998298
        },
        {
          "epoch": 554,
          "updates": 8864,
          "samples_seen": 567296,
          "online_loss": 0.00022241363058128627,
          "train_loss": 0.00021877991093788296,
          "validation_loss": 0.0002059911930700764,
          "train_metric": 2.2034710127627477e-05,
          "validation_metric": 2.0746676455019042e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.346034100999532
        },
        {
          "epoch": 555,
          "updates": 8880,
          "samples_seen": 568320,
          "online_loss": 0.00022424058533943025,
          "train_loss": 0.0002203940093750134,
          "validation_loss": 0.00020251856767572463,
          "train_metric": 2.2197276848601177e-05,
          "validation_metric": 2.0396930267452262e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.356379448998268
        },
        {
          "epoch": 556,
          "updates": 8896,
          "samples_seen": 569344,
          "online_loss": 0.00022258014814724447,
          "train_loss": 0.00021855728118680418,
          "validation_loss": 0.0002053672360489145,
          "train_metric": 2.2012285626260564e-05,
          "validation_metric": 2.068383400910534e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.36671971999931
        },
        {
          "epoch": 557,
          "updates": 8912,
          "samples_seen": 570368,
          "online_loss": 0.00022824970255896915,
          "train_loss": 0.00022228489979170263,
          "validation_loss": 0.00020546370069496334,
          "train_metric": 2.2387719582184218e-05,
          "validation_metric": 2.069355105049908e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.377263067999593
        },
        {
          "epoch": 558,
          "updates": 8928,
          "samples_seen": 571392,
          "online_loss": 0.00022309806263365317,
          "train_loss": 0.00021790021855849773,
          "validation_loss": 0.00020469010632950813,
          "train_metric": 2.1946110791759565e-05,
          "validation_metric": 2.0615638277377002e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.387678553999649
        },
        {
          "epoch": 559,
          "updates": 8944,
          "samples_seen": 572416,
          "online_loss": 0.00022321122560242657,
          "train_loss": 0.00021791484323330224,
          "validation_loss": 0.00020641772425733507,
          "train_metric": 2.1947582354187034e-05,
          "validation_metric": 2.078963916574139e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.397988611999608
        },
        {
          "epoch": 560,
          "updates": 8960,
          "samples_seen": 573440,
          "online_loss": 0.00022126959356683074,
          "train_loss": 0.00021769048180431128,
          "validation_loss": 0.00020138865511398762,
          "train_metric": 2.1924986867816187e-05,
          "validation_metric": 2.0283130652387626e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.409255978998772
        },
        {
          "epoch": 561,
          "updates": 8976,
          "samples_seen": 574464,
          "online_loss": 0.0002226910737590515,
          "train_loss": 0.0002187678765039891,
          "validation_loss": 0.00020524696446955204,
          "train_metric": 2.2033498680684716e-05,
          "validation_metric": 2.0671721358667128e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.4197442199983925
        },
        {
          "epoch": 562,
          "updates": 8992,
          "samples_seen": 575488,
          "online_loss": 0.0002209194062743336,
          "train_loss": 0.00021864291920792311,
          "validation_loss": 0.00020208288333378732,
          "train_metric": 2.2020913093001582e-05,
          "validation_metric": 2.0353048967081122e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.430160897998576
        },
        {
          "epoch": 563,
          "updates": 9008,
          "samples_seen": 576512,
          "online_loss": 0.00022220050777832512,
          "train_loss": 0.00022042685304768384,
          "validation_loss": 0.0002070000336971134,
          "train_metric": 2.2200585590326227e-05,
          "validation_metric": 2.0848287022090517e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.440945104999628
        },
        {
          "epoch": 564,
          "updates": 9024,
          "samples_seen": 577536,
          "online_loss": 0.0002252889580631745,
          "train_loss": 0.00022108282428234816,
          "validation_loss": 0.00020745419897139072,
          "train_metric": 2.2266653104452416e-05,
          "validation_metric": 2.0894025510642678e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.451803061998362
        },
        {
          "epoch": 565,
          "updates": 9040,
          "samples_seen": 578560,
          "online_loss": 0.00022145706316223368,
          "train_loss": 0.00021650176495313644,
          "validation_loss": 0.00020289079111535102,
          "train_metric": 2.1805262804264203e-05,
          "validation_metric": 2.0434419639059342e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.462405457999921
        },
        {
          "epoch": 566,
          "updates": 9056,
          "samples_seen": 579584,
          "online_loss": 0.00022066401561460225,
          "train_loss": 0.00021660566562786698,
          "validation_loss": 0.00020292526460252702,
          "train_metric": 2.1815729269292206e-05,
          "validation_metric": 2.043789208983071e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.472842690998732
        },
        {
          "epoch": 567,
          "updates": 9072,
          "samples_seen": 580608,
          "online_loss": 0.00022545658885064768,
          "train_loss": 0.00022129135322757065,
          "validation_loss": 0.00020797926117666066,
          "train_metric": 2.2287655156105757e-05,
          "validation_metric": 2.0946907170582563e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.483194853999521
        },
        {
          "epoch": 568,
          "updates": 9088,
          "samples_seen": 581632,
          "online_loss": 0.00022099486068327678,
          "train_loss": 0.00022485062072519213,
          "validation_loss": 0.00020556851814035326,
          "train_metric": 2.2646130673820153e-05,
          "validation_metric": 2.0704108464997262e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.493529343999398
        },
        {
          "epoch": 569,
          "updates": 9104,
          "samples_seen": 582656,
          "online_loss": 0.00022497078680316918,
          "train_loss": 0.00021582232147920877,
          "validation_loss": 0.00020481963292695582,
          "train_metric": 2.1736832422902808e-05,
          "validation_metric": 2.062868225038983e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.503854702999888
        },
        {
          "epoch": 570,
          "updates": 9120,
          "samples_seen": 583680,
          "online_loss": 0.00022436836479755584,
          "train_loss": 0.0002185538032790646,
          "validation_loss": 0.00019877978775184602,
          "train_metric": 2.2011936380295083e-05,
          "validation_metric": 2.002037399506662e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.514917181999408
        },
        {
          "epoch": 571,
          "updates": 9136,
          "samples_seen": 584704,
          "online_loss": 0.00022221530070964945,
          "train_loss": 0.00021896524413023144,
          "validation_loss": 0.00021154218120500445,
          "train_metric": 2.2053376596886665e-05,
          "validation_metric": 2.130575194314588e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.525493306999124
        },
        {
          "epoch": 572,
          "updates": 9152,
          "samples_seen": 585728,
          "online_loss": 0.0002194751359638758,
          "train_loss": 0.00021590424876194447,
          "validation_loss": 0.00020122274872846901,
          "train_metric": 2.174508335883729e-05,
          "validation_metric": 2.026641959673725e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.535880472998542
        },
        {
          "epoch": 573,
          "updates": 9168,
          "samples_seen": 586752,
          "online_loss": 0.00022360493494488765,
          "train_loss": 0.0002148550993297249,
          "validation_loss": 0.00020183229935355484,
          "train_metric": 2.1639420083374716e-05,
          "validation_metric": 2.032780867011752e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.546321944999363
        },
        {
          "epoch": 574,
          "updates": 9184,
          "samples_seen": 587776,
          "online_loss": 0.00022452432767749997,
          "train_loss": 0.0002145532052963972,
          "validation_loss": 0.00019964312377851456,
          "train_metric": 2.1609010218526237e-05,
          "validation_metric": 2.0107323507545516e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.557152086999849
        },
        {
          "epoch": 575,
          "updates": 9200,
          "samples_seen": 588800,
          "online_loss": 0.00022833050206827465,
          "train_loss": 0.0002144646568922326,
          "validation_loss": 0.00020373446750454605,
          "train_metric": 2.160009535145946e-05,
          "validation_metric": 2.051939009106718e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.567455527998391
        },
        {
          "epoch": 576,
          "updates": 9216,
          "samples_seen": 589824,
          "online_loss": 0.0002202861724072136,
          "train_loss": 0.0002167656202800572,
          "validation_loss": 0.00020688990480266511,
          "train_metric": 2.1831838239450008e-05,
          "validation_metric": 2.0837196643697098e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.578052682998532
        },
        {
          "epoch": 577,
          "updates": 9232,
          "samples_seen": 590848,
          "online_loss": 0.00022383155146599165,
          "train_loss": 0.0002241115435026586,
          "validation_loss": 0.00020745056099258363,
          "train_metric": 2.257169398944825e-05,
          "validation_metric": 2.0893663531751372e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.5883455029998
        },
        {
          "epoch": 578,
          "updates": 9248,
          "samples_seen": 591872,
          "online_loss": 0.00023615048303327058,
          "train_loss": 0.00023451955348718911,
          "validation_loss": 0.00021435922826640308,
          "train_metric": 2.3619950297870673e-05,
          "validation_metric": 2.158947791031096e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.598641423999652
        },
        {
          "epoch": 579,
          "updates": 9264,
          "samples_seen": 592896,
          "online_loss": 0.00022056024772609817,
          "train_loss": 0.00022638072550762445,
          "validation_loss": 0.0002068956382572651,
          "train_metric": 2.2800239094067365e-05,
          "validation_metric": 2.083777144434862e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.6091054739990795
        },
        {
          "epoch": 580,
          "updates": 9280,
          "samples_seen": 593920,
          "online_loss": 0.00021596410169877345,
          "train_loss": 0.0002196278510382399,
          "validation_loss": 0.0002113106456818059,
          "train_metric": 2.212011349911336e-05,
          "validation_metric": 2.1282434317981824e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.619440385999042
        },
        {
          "epoch": 581,
          "updates": 9296,
          "samples_seen": 594944,
          "online_loss": 0.0002199667101194791,
          "train_loss": 0.00021967505745124072,
          "validation_loss": 0.0002036033693002537,
          "train_metric": 2.2124866518424824e-05,
          "validation_metric": 2.0506187865976244e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.62975079899843
        },
        {
          "epoch": 582,
          "updates": 9312,
          "samples_seen": 595968,
          "online_loss": 0.00022395455471269088,
          "train_loss": 0.00021303298126440495,
          "validation_loss": 0.00019838038133457303,
          "train_metric": 2.145590042346157e-05,
          "validation_metric": 1.99801452254178e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.640884124999502
        },
        {
          "epoch": 583,
          "updates": 9328,
          "samples_seen": 596992,
          "online_loss": 0.00021654383635905106,
          "train_loss": 0.00021263922099024057,
          "validation_loss": 0.00019965441606473178,
          "train_metric": 2.1416244635474868e-05,
          "validation_metric": 2.0108460375922732e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.651365816998805
        },
        {
          "epoch": 584,
          "updates": 9344,
          "samples_seen": 598016,
          "online_loss": 0.00021557457603194052,
          "train_loss": 0.00021245841344352812,
          "validation_loss": 0.00020091391343157738,
          "train_metric": 2.1398031094577163e-05,
          "validation_metric": 2.0235313058947213e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.662242613998387
        },
        {
          "epoch": 585,
          "updates": 9360,
          "samples_seen": 599040,
          "online_loss": 0.0002155259271603427,
          "train_loss": 0.00021284034301061183,
          "validation_loss": 0.00020164794113952667,
          "train_metric": 2.1436500901472755e-05,
          "validation_metric": 2.0309240426286124e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.729446051998821
        },
        {
          "epoch": 586,
          "updates": 9376,
          "samples_seen": 600064,
          "online_loss": 0.0002156682312488556,
          "train_loss": 0.00021198851754888892,
          "validation_loss": 0.00020072032930329442,
          "train_metric": 2.1350708266254514e-05,
          "validation_metric": 2.0215815311530605e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.740175448998343
        },
        {
          "epoch": 587,
          "updates": 9392,
          "samples_seen": 601088,
          "online_loss": 0.00021592250868707197,
          "train_loss": 0.00021449672931339592,
          "validation_loss": 0.00020233732357155532,
          "train_metric": 2.1603325876640156e-05,
          "validation_metric": 2.0378673070808873e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.750539174998266
        },
        {
          "epoch": 588,
          "updates": 9408,
          "samples_seen": 602112,
          "online_loss": 0.00021611786178254988,
          "train_loss": 0.00021382185514084995,
          "validation_loss": 0.00019670100300572813,
          "train_metric": 2.1535353880608454e-05,
          "validation_metric": 1.981100285775028e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.761443147999671
        },
        {
          "epoch": 589,
          "updates": 9424,
          "samples_seen": 603136,
          "online_loss": 0.00022174405103214667,
          "train_loss": 0.00021525719785131514,
          "validation_loss": 0.00020750268595293164,
          "train_metric": 2.1679916244465858e-05,
          "validation_metric": 2.0898911316180602e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.771853464999367
        },
        {
          "epoch": 590,
          "updates": 9440,
          "samples_seen": 604160,
          "online_loss": 0.00021750746691395761,
          "train_loss": 0.00021433965594042093,
          "validation_loss": 0.00019950389105360955,
          "train_metric": 2.1587504306808114e-05,
          "validation_metric": 2.0093302737222984e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.782303159998264
        },
        {
          "epoch": 591,
          "updates": 9456,
          "samples_seen": 605184,
          "online_loss": 0.00021718819971283665,
          "train_loss": 0.00021124391059856862,
          "validation_loss": 0.00020163478620816022,
          "train_metric": 2.1275713152135722e-05,
          "validation_metric": 2.0307918020989746e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.792643841999961
        },
        {
          "epoch": 592,
          "updates": 9472,
          "samples_seen": 606208,
          "online_loss": 0.00022343960336002056,
          "train_loss": 0.000218145563849248,
          "validation_loss": 0.0002062989806290716,
          "train_metric": 2.197082358179614e-05,
          "validation_metric": 2.0777677491423674e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.803001263999249
        },
        {
          "epoch": 593,
          "updates": 9488,
          "samples_seen": 607232,
          "online_loss": 0.00023262499053089414,
          "train_loss": 0.00021439208649098873,
          "validation_loss": 0.00020044177654199302,
          "train_metric": 2.159278665203601e-05,
          "validation_metric": 2.018776285694912e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.813448584998696
        },
        {
          "epoch": 594,
          "updates": 9504,
          "samples_seen": 608256,
          "online_loss": 0.00021763307904620888,
          "train_loss": 0.00021113075490575284,
          "validation_loss": 0.0002028125018114224,
          "train_metric": 2.1264315364533104e-05,
          "validation_metric": 2.042653431999497e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.823880467998606
        },
        {
          "epoch": 595,
          "updates": 9520,
          "samples_seen": 609280,
          "online_loss": 0.0002163026028938475,
          "train_loss": 0.00021034169185440987,
          "validation_loss": 0.00019940004858653992,
          "train_metric": 2.1184841898502782e-05,
          "validation_metric": 2.0082841729163192e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.834134256998368
        },
        {
          "epoch": 596,
          "updates": 9536,
          "samples_seen": 610304,
          "online_loss": 0.0002234193561889697,
          "train_loss": 0.00021590449614450336,
          "validation_loss": 0.0002003541449084878,
          "train_metric": 2.174511246266775e-05,
          "validation_metric": 2.017893348238431e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.844700732999627
        },
        {
          "epoch": 597,
          "updates": 9552,
          "samples_seen": 611328,
          "online_loss": 0.0002162678219974623,
          "train_loss": 0.00021000967535655946,
          "validation_loss": 0.00020008572028018534,
          "train_metric": 2.1151405235286802e-05,
          "validation_metric": 2.0151897842879407e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.855040731999907
        },
        {
          "epoch": 598,
          "updates": 9568,
          "samples_seen": 612352,
          "online_loss": 0.0002190677123508067,
          "train_loss": 0.0002101938589476049,
          "validation_loss": 0.0001951070735231042,
          "train_metric": 2.116995347023476e-05,
          "validation_metric": 1.965047158591915e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.865988009998546
        },
        {
          "epoch": 599,
          "updates": 9584,
          "samples_seen": 613376,
          "online_loss": 0.00021554636350629153,
          "train_loss": 0.00021549696975853294,
          "validation_loss": 0.0002069128240691498,
          "train_metric": 2.1704065147787333e-05,
          "validation_metric": 2.0839503122260794e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.876618230999156
        },
        {
          "epoch": 600,
          "updates": 9600,
          "samples_seen": 614400,
          "online_loss": 0.0002116007844961132,
          "train_loss": 0.00020965698058716953,
          "validation_loss": 0.00019604209228418767,
          "train_metric": 2.1115882191224955e-05,
          "validation_metric": 1.9744640667340718e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.887079879999874
        },
        {
          "epoch": 601,
          "updates": 9616,
          "samples_seen": 615424,
          "online_loss": 0.00021372173068812117,
          "train_loss": 0.00021013822697568685,
          "validation_loss": 0.0002011524047702551,
          "train_metric": 2.116435280186124e-05,
          "validation_metric": 2.0259332814021036e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.897342000998833
        },
        {
          "epoch": 602,
          "updates": 9632,
          "samples_seen": 616448,
          "online_loss": 0.00021390418260125443,
          "train_loss": 0.00020905036944895983,
          "validation_loss": 0.00019475954468362033,
          "train_metric": 2.1054789613117464e-05,
          "validation_metric": 1.9615468772826716e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.9083619069988345
        },
        {
          "epoch": 603,
          "updates": 9648,
          "samples_seen": 617472,
          "online_loss": 0.00021265441455398104,
          "train_loss": 0.0002114045637426898,
          "validation_loss": 0.0002017454826273024,
          "train_metric": 2.1291893062880263e-05,
          "validation_metric": 2.031906660704408e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.918836235998242
        },
        {
          "epoch": 604,
          "updates": 9664,
          "samples_seen": 618496,
          "online_loss": 0.0002226218639407307,
          "train_loss": 0.00020843806851189584,
          "validation_loss": 0.00019610777962952852,
          "train_metric": 2.0993118596379645e-05,
          "validation_metric": 1.975125451281201e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.92931105399839
        },
        {
          "epoch": 605,
          "updates": 9680,
          "samples_seen": 619520,
          "online_loss": 0.00021412815431176568,
          "train_loss": 0.00020834957831539214,
          "validation_loss": 0.00019741222786251456,
          "train_metric": 2.0984207367291674e-05,
          "validation_metric": 1.9882634660461918e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.93969765699876
        },
        {
          "epoch": 606,
          "updates": 9696,
          "samples_seen": 620544,
          "online_loss": 0.0002142415542039089,
          "train_loss": 0.0002082775317830965,
          "validation_loss": 0.0001974370825337246,
          "train_metric": 2.097695141856093e-05,
          "validation_metric": 1.9885135770891793e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.950102943999809
        },
        {
          "epoch": 607,
          "updates": 9712,
          "samples_seen": 621568,
          "online_loss": 0.0002128616661138949,
          "train_loss": 0.0002089583722408861,
          "validation_loss": 0.00019829395751003176,
          "train_metric": 2.1045521862106398e-05,
          "validation_metric": 1.9971441361121833e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.960453684998356
        },
        {
          "epoch": 608,
          "updates": 9728,
          "samples_seen": 622592,
          "online_loss": 0.000219169552110543,
          "train_loss": 0.00021210810518823564,
          "validation_loss": 0.0001994315389310941,
          "train_metric": 2.1362749976105988e-05,
          "validation_metric": 2.008601586567238e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.970892289999028
        },
        {
          "epoch": 609,
          "updates": 9744,
          "samples_seen": 623616,
          "online_loss": 0.0002213947500422364,
          "train_loss": 0.00020769712864421308,
          "validation_loss": 0.0001970896846614778,
          "train_metric": 2.0918494556099176e-05,
          "validation_metric": 1.9850151147693396e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.981358793998879
        },
        {
          "epoch": 610,
          "updates": 9760,
          "samples_seen": 624640,
          "online_loss": 0.00022044657771402854,
          "train_loss": 0.00021163056953810155,
          "validation_loss": 0.000201593735255301,
          "train_metric": 2.1314655896276236e-05,
          "validation_metric": 2.0303778001107275e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.991744342998572
        },
        {
          "epoch": 611,
          "updates": 9776,
          "samples_seen": 625664,
          "online_loss": 0.0002171055257349508,
          "train_loss": 0.00020842472440563142,
          "validation_loss": 0.0001938513305503875,
          "train_metric": 2.099177618219983e-05,
          "validation_metric": 1.9523997252690606e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.00258704099906
        },
        {
          "epoch": 612,
          "updates": 9792,
          "samples_seen": 626688,
          "online_loss": 0.00021235489748505643,
          "train_loss": 0.00021256755280774087,
          "validation_loss": 0.00020324139040894806,
          "train_metric": 2.140902324754279e-05,
          "validation_metric": 2.0469731680350378e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.013178007999159
        },
        {
          "epoch": 613,
          "updates": 9808,
          "samples_seen": 627712,
          "online_loss": 0.00021585675722235464,
          "train_loss": 0.00020736336591653526,
          "validation_loss": 0.00019713076471816748,
          "train_metric": 2.0884877812932245e-05,
          "validation_metric": 1.9854289348586462e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.023545726999146
        },
        {
          "epoch": 614,
          "updates": 9824,
          "samples_seen": 628736,
          "online_loss": 0.00021024835541538778,
          "train_loss": 0.00020716951985377818,
          "validation_loss": 0.00019553433230612427,
          "train_metric": 2.086535641865339e-05,
          "validation_metric": 1.969350159924943e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.033912468999915
        },
        {
          "epoch": 615,
          "updates": 9840,
          "samples_seen": 629760,
          "online_loss": 0.0002106426254613325,
          "train_loss": 0.00020621063595172018,
          "validation_loss": 0.0001962605892913416,
          "train_metric": 2.0768778995261528e-05,
          "validation_metric": 1.9766646801144816e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.044394578999345
        },
        {
          "epoch": 616,
          "updates": 9856,
          "samples_seen": 630784,
          "online_loss": 0.00021359280799515545,
          "train_loss": 0.00020840438082814217,
          "validation_loss": 0.0001954053877852857,
          "train_metric": 2.0989726181142032e-05,
          "validation_metric": 1.9680514014908113e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.054853935998835
        },
        {
          "epoch": 617,
          "updates": 9872,
          "samples_seen": 631808,
          "online_loss": 0.00022370849819708383,
          "train_loss": 0.00021426554303616285,
          "validation_loss": 0.00020350655540823936,
          "train_metric": 2.1580040993285365e-05,
          "validation_metric": 2.049643444479443e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.065119003998916
        },
        {
          "epoch": 618,
          "updates": 9888,
          "samples_seen": 632832,
          "online_loss": 0.0002179970615543425,
          "train_loss": 0.00020565735758282244,
          "validation_loss": 0.0001947120763361454,
          "train_metric": 2.0713056073873304e-05,
          "validation_metric": 1.961068483069539e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.075630109999111
        },
        {
          "epoch": 619,
          "updates": 9904,
          "samples_seen": 633856,
          "online_loss": 0.00020985466471756808,
          "train_loss": 0.00020547663734760135,
          "validation_loss": 0.00019628848531283438,
          "train_metric": 2.0694851627922617e-05,
          "validation_metric": 1.9769458958762698e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.085888204999719
        },
        {
          "epoch": 620,
          "updates": 9920,
          "samples_seen": 634880,
          "online_loss": 0.0002108441740347189,
          "train_loss": 0.00020527424931060523,
          "validation_loss": 0.00019544664246495813,
          "train_metric": 2.0674469851655886e-05,
          "validation_metric": 1.9684670405695215e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.096541967999656
        },
        {
          "epoch": 621,
          "updates": 9936,
          "samples_seen": 635904,
          "online_loss": 0.00020752555292347097,
          "train_loss": 0.000205088741495274,
          "validation_loss": 0.0001950591686181724,
          "train_metric": 2.065578883048147e-05,
          "validation_metric": 1.9645645807031542e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.164005002998238
        },
        {
          "epoch": 622,
          "updates": 9952,
          "samples_seen": 636928,
          "online_loss": 0.00021128537082404364,
          "train_loss": 0.0002079705591313541,
          "validation_loss": 0.00019832074758596718,
          "train_metric": 2.0946030417690054e-05,
          "validation_metric": 1.9974138922407292e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.175179758998638
        },
        {
          "epoch": 623,
          "updates": 9968,
          "samples_seen": 637952,
          "online_loss": 0.00020812689490412595,
          "train_loss": 0.00020619225688278675,
          "validation_loss": 0.0001946697011590004,
          "train_metric": 2.076693272101693e-05,
          "validation_metric": 1.960642111953348e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.188251608999053
        },
        {
          "epoch": 624,
          "updates": 9984,
          "samples_seen": 638976,
          "online_loss": 0.00021070721913929447,
          "train_loss": 0.00020839118224103004,
          "validation_loss": 0.00020219426369294524,
          "train_metric": 2.0988394680898637e-05,
          "validation_metric": 2.0364264855743386e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.201026759999877
        },
        {
          "epoch": 625,
          "updates": 10000,
          "samples_seen": 640000,
          "online_loss": 0.0002109100514644524,
          "train_loss": 0.00021236905013211071,
          "validation_loss": 0.00020146224414929748,
          "train_metric": 2.138903619197663e-05,
          "validation_metric": 2.0290541215217672e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.213600730998223
        },
        {
          "epoch": 626,
          "updates": 10016,
          "samples_seen": 641024,
          "online_loss": 0.00021874778030905873,
          "train_loss": 0.0002060384867945686,
          "validation_loss": 0.00019855952996294945,
          "train_metric": 2.0751442207256332e-05,
          "validation_metric": 1.9998189600300975e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.226490081999145
        },
        {
          "epoch": 627,
          "updates": 10032,
          "samples_seen": 642048,
          "online_loss": 0.00021113265211170074,
          "train_loss": 0.00020584423327818513,
          "validation_loss": 0.00019401359895709902,
          "train_metric": 2.0731877157231793e-05,
          "validation_metric": 1.9540339053492062e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.238690860998759
        },
        {
          "epoch": 628,
          "updates": 10048,
          "samples_seen": 643072,
          "online_loss": 0.00021302448431015364,
          "train_loss": 0.00020876876078546047,
          "validation_loss": 0.00019567356503102928,
          "train_metric": 2.1026426111347973e-05,
          "validation_metric": 1.9707524188561365e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.251198699999804
        },
        {
          "epoch": 629,
          "updates": 10064,
          "samples_seen": 644096,
          "online_loss": 0.00021391172958828975,
          "train_loss": 0.00020366741227917373,
          "validation_loss": 0.00019647736917249858,
          "train_metric": 2.0512636183411814e-05,
          "validation_metric": 1.978848194994498e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.2630335229987395
        },
        {
          "epoch": 630,
          "updates": 10080,
          "samples_seen": 645120,
          "online_loss": 0.00020833922008023364,
          "train_loss": 0.0002033926866715774,
          "validation_loss": 0.00019494789012242109,
          "train_metric": 2.0484967535594478e-05,
          "validation_metric": 1.963443537533749e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.275033860998519
        },
        {
          "epoch": 631,
          "updates": 10096,
          "samples_seen": 646144,
          "online_loss": 0.00021214825937931892,
          "train_loss": 0.0002077074022963643,
          "validation_loss": 0.00019916701421607286,
          "train_metric": 2.0919529561069794e-05,
          "validation_metric": 2.005937130888924e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.285829401998853
        }
      ],
      "best": {
        "epoch": 611,
        "updates": 9776
      },
      "best_validation_metric": 1.9523997252690606e-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": 0.004418596751536692
    },
    "14985df6b90f_22": {
      "history": [
        {
          "epoch": 1,
          "updates": 16,
          "samples_seen": 1024,
          "online_loss": 0.18389458605088294,
          "train_loss": 0.06916050612926483,
          "validation_loss": 0.06209268793463707,
          "train_metric": 0.006965593434870243,
          "validation_metric": 0.0062537482008337975,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0139888419998897
        },
        {
          "epoch": 2,
          "updates": 32,
          "samples_seen": 2048,
          "online_loss": 0.045738674933090806,
          "train_loss": 0.03489449992775917,
          "validation_loss": 0.032735906541347504,
          "train_metric": 0.0035144463181495667,
          "validation_metric": 0.003297040704637766,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.025725466999574564
        },
        {
          "epoch": 3,
          "updates": 48,
          "samples_seen": 3072,
          "online_loss": 0.02863602398429066,
          "train_loss": 0.02373126894235611,
          "validation_loss": 0.017781876027584076,
          "train_metric": 0.0023901264648884535,
          "validation_metric": 0.0017909255111590028,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.037212017999991076
        },
        {
          "epoch": 4,
          "updates": 64,
          "samples_seen": 4096,
          "online_loss": 0.03047496487852186,
          "train_loss": 0.018814587965607643,
          "validation_loss": 0.01837657205760479,
          "train_metric": 0.001894936547614634,
          "validation_metric": 0.001850820961408317,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.0480237939991639
        },
        {
          "epoch": 5,
          "updates": 80,
          "samples_seen": 5120,
          "online_loss": 0.01888181286631152,
          "train_loss": 0.01885739341378212,
          "validation_loss": 0.016206376254558563,
          "train_metric": 0.001899247639812529,
          "validation_metric": 0.0016322467708960176,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.05945236399929854
        },
        {
          "epoch": 6,
          "updates": 96,
          "samples_seen": 6144,
          "online_loss": 0.01619607725297101,
          "train_loss": 0.011927196756005287,
          "validation_loss": 0.011007381603121758,
          "train_metric": 0.0012012633960694075,
          "validation_metric": 0.001108623226173222,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.07095492699954775
        },
        {
          "epoch": 7,
          "updates": 112,
          "samples_seen": 7168,
          "online_loss": 0.012518655508756638,
          "train_loss": 0.010872814804315567,
          "validation_loss": 0.01140688918530941,
          "train_metric": 0.0010950700379908085,
          "validation_metric": 0.0011488600866869092,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.08177693200013891
        },
        {
          "epoch": 8,
          "updates": 128,
          "samples_seen": 8192,
          "online_loss": 0.009969741659006104,
          "train_loss": 0.009192890487611294,
          "validation_loss": 0.009333253838121891,
          "train_metric": 0.000925874279346317,
          "validation_metric": 0.00094001111574471,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.09319984399917303
        },
        {
          "epoch": 9,
          "updates": 144,
          "samples_seen": 9216,
          "online_loss": 0.010474069189513102,
          "train_loss": 0.00656439084559679,
          "validation_loss": 0.00658778939396143,
          "train_metric": 0.0006611414137296379,
          "validation_metric": 0.0006634980672970414,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.16778796399921703
        },
        {
          "epoch": 10,
          "updates": 160,
          "samples_seen": 10240,
          "online_loss": 0.013046462088823318,
          "train_loss": 0.023848779499530792,
          "validation_loss": 0.01683362014591694,
          "train_metric": 0.0024019619449973106,
          "validation_metric": 0.0016954205930233002,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.17914598700008355
        },
        {
          "epoch": 11,
          "updates": 176,
          "samples_seen": 11264,
          "online_loss": 0.0150880494620651,
          "train_loss": 0.01769361086189747,
          "validation_loss": 0.013639772310853004,
          "train_metric": 0.0017820355715230107,
          "validation_metric": 0.0013737480621784925,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.1901299809996999
        },
        {
          "epoch": 12,
          "updates": 192,
          "samples_seen": 12288,
          "online_loss": 0.013406007667072117,
          "train_loss": 0.010042965412139893,
          "validation_loss": 0.01069402415305376,
          "train_metric": 0.0010114905890077353,
          "validation_metric": 0.0010770629160106182,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.20088241600024048
        },
        {
          "epoch": 13,
          "updates": 208,
          "samples_seen": 13312,
          "online_loss": 0.007156405947171152,
          "train_loss": 0.0063883112743496895,
          "validation_loss": 0.0051514930091798306,
          "train_metric": 0.0006434072856791317,
          "validation_metric": 0.0005188394570723176,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.21218743300050846
        },
        {
          "epoch": 14,
          "updates": 224,
          "samples_seen": 14336,
          "online_loss": 0.006110915433964692,
          "train_loss": 0.005998062901198864,
          "validation_loss": 0.008657868020236492,
          "train_metric": 0.0006041027954779565,
          "validation_metric": 0.0008719888282939792,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.22292227800062392
        },
        {
          "epoch": 15,
          "updates": 240,
          "samples_seen": 15360,
          "online_loss": 0.0062475081358570606,
          "train_loss": 0.01197153702378273,
          "validation_loss": 0.007577677723020315,
          "train_metric": 0.0012057292042300105,
          "validation_metric": 0.0007631958578713238,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.234132633999252
        },
        {
          "epoch": 16,
          "updates": 256,
          "samples_seen": 16384,
          "online_loss": 0.0113911721855402,
          "train_loss": 0.008418227545917034,
          "validation_loss": 0.005429287441074848,
          "train_metric": 0.0008478531381115317,
          "validation_metric": 0.0005468179006129503,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2451009399992472
        },
        {
          "epoch": 17,
          "updates": 272,
          "samples_seen": 17408,
          "online_loss": 0.006947965404833667,
          "train_loss": 0.006220518611371517,
          "validation_loss": 0.004411776550114155,
          "train_metric": 0.0006265078554861248,
          "validation_metric": 0.00044433801667764783,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.25639663400033896
        },
        {
          "epoch": 18,
          "updates": 288,
          "samples_seen": 18432,
          "online_loss": 0.015920912410365418,
          "train_loss": 0.005532862152904272,
          "validation_loss": 0.0060539236292243,
          "train_metric": 0.0005572495865635574,
          "validation_metric": 0.0006097290315665305,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2670710799993685
        },
        {
          "epoch": 19,
          "updates": 304,
          "samples_seen": 19456,
          "online_loss": 0.0066804879024857655,
          "train_loss": 0.005352771375328302,
          "validation_loss": 0.006394522730261087,
          "train_metric": 0.0005391114973463118,
          "validation_metric": 0.0006440328434109688,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2779886839998653
        },
        {
          "epoch": 20,
          "updates": 320,
          "samples_seen": 20480,
          "online_loss": 0.006032283257809468,
          "train_loss": 0.007059706840664148,
          "validation_loss": 0.006349417380988598,
          "train_metric": 0.0007110278238542378,
          "validation_metric": 0.000639490142930299,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.2883354709993
        },
        {
          "epoch": 21,
          "updates": 336,
          "samples_seen": 21504,
          "online_loss": 0.011215290942345746,
          "train_loss": 0.010435744188725948,
          "validation_loss": 0.007207291200757027,
          "train_metric": 0.0010510500287637115,
          "validation_metric": 0.0007258919067680836,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.29867511299926264
        },
        {
          "epoch": 22,
          "updates": 352,
          "samples_seen": 22528,
          "online_loss": 0.006948341120732948,
          "train_loss": 0.004147992469370365,
          "validation_loss": 0.004910673946142197,
          "train_metric": 0.0004177706432528794,
          "validation_metric": 0.0004945851396769285,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.30947180300063337
        },
        {
          "epoch": 23,
          "updates": 368,
          "samples_seen": 23552,
          "online_loss": 0.005095057058497332,
          "train_loss": 0.005464919842779636,
          "validation_loss": 0.00746883312240243,
          "train_metric": 0.000550406810361892,
          "validation_metric": 0.0007522335508838296,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.320271190999847
        },
        {
          "epoch": 24,
          "updates": 384,
          "samples_seen": 24576,
          "online_loss": 0.005976300351903774,
          "train_loss": 0.006015682592988014,
          "validation_loss": 0.004650956951081753,
          "train_metric": 0.0006058774888515472,
          "validation_metric": 0.00046842731535434723,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3308839439996518
        },
        {
          "epoch": 25,
          "updates": 400,
          "samples_seen": 25600,
          "online_loss": 0.008279251887870487,
          "train_loss": 0.004652548581361771,
          "validation_loss": 0.004610558971762657,
          "train_metric": 0.0004685877065639943,
          "validation_metric": 0.00046435859985649586,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.34151806400041096
        },
        {
          "epoch": 26,
          "updates": 416,
          "samples_seen": 26624,
          "online_loss": 0.005110172278364189,
          "train_loss": 0.004339386243373156,
          "validation_loss": 0.003442443208768964,
          "train_metric": 0.0004370470705907792,
          "validation_metric": 0.0003467102360446006,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.352677388000302
        },
        {
          "epoch": 27,
          "updates": 432,
          "samples_seen": 27648,
          "online_loss": 0.004839025321416557,
          "train_loss": 0.004047211725264788,
          "validation_loss": 0.0022406503558158875,
          "train_metric": 0.000407620333135128,
          "validation_metric": 0.0002256701118312776,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.36376157799895736
        },
        {
          "epoch": 28,
          "updates": 448,
          "samples_seen": 28672,
          "online_loss": 0.006638483784627169,
          "train_loss": 0.0031345903407782316,
          "validation_loss": 0.0033052018843591213,
          "train_metric": 0.0003157044993713498,
          "validation_metric": 0.00033288789563812315,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.3747210669989727
        },
        {
          "epoch": 29,
          "updates": 464,
          "samples_seen": 29696,
          "online_loss": 0.004400948935654014,
          "train_loss": 0.007335529197007418,
          "validation_loss": 0.007428395561873913,
          "train_metric": 0.0007388076628558338,
          "validation_metric": 0.0007481608190573752,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.38564662899989344
        },
        {
          "epoch": 30,
          "updates": 480,
          "samples_seen": 30720,
          "online_loss": 0.007314472968573682,
          "train_loss": 0.01119297370314598,
          "validation_loss": 0.009809521026909351,
          "train_metric": 0.001127315335907042,
          "validation_metric": 0.000987979001365602,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.39659138299975893
        },
        {
          "epoch": 31,
          "updates": 496,
          "samples_seen": 31744,
          "online_loss": 0.011946768732741475,
          "train_loss": 0.015263335779309273,
          "validation_loss": 0.009868407621979713,
          "train_metric": 0.00153726723510772,
          "validation_metric": 0.0009939097799360752,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4074878949995764
        },
        {
          "epoch": 32,
          "updates": 512,
          "samples_seen": 32768,
          "online_loss": 0.01918425314943306,
          "train_loss": 0.008226007223129272,
          "validation_loss": 0.006636414676904678,
          "train_metric": 0.0008284933865070343,
          "validation_metric": 0.0006683953106403351,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.41843284199967457
        },
        {
          "epoch": 33,
          "updates": 528,
          "samples_seen": 33792,
          "online_loss": 0.007922696531750262,
          "train_loss": 0.00950158853083849,
          "validation_loss": 0.00873914361000061,
          "train_metric": 0.0009569652029313147,
          "validation_metric": 0.0008801745716482401,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.42899874199974875
        },
        {
          "epoch": 34,
          "updates": 544,
          "samples_seen": 34816,
          "online_loss": 0.0060403976967791095,
          "train_loss": 0.004429431166499853,
          "validation_loss": 0.0031370145734399557,
          "train_metric": 0.00044611605699174106,
          "validation_metric": 0.0003159486223012209,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4396755449997727
        },
        {
          "epoch": 35,
          "updates": 560,
          "samples_seen": 35840,
          "online_loss": 0.007136315063689835,
          "train_loss": 0.00483589805662632,
          "validation_loss": 0.004416401963680983,
          "train_metric": 0.00048705397057347,
          "validation_metric": 0.00044480382348410785,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.45059103299900016
        },
        {
          "epoch": 36,
          "updates": 576,
          "samples_seen": 36864,
          "online_loss": 0.006166164930618834,
          "train_loss": 0.008886346593499184,
          "validation_loss": 0.006666672416031361,
          "train_metric": 0.0008950002957135439,
          "validation_metric": 0.0006714428309351206,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.46103758699973696
        },
        {
          "epoch": 37,
          "updates": 592,
          "samples_seen": 37888,
          "online_loss": 0.013328372235264396,
          "train_loss": 0.006340267136693001,
          "validation_loss": 0.005754882004112005,
          "train_metric": 0.0006385684828273952,
          "validation_metric": 0.0005796106997877359,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.47163813699989987
        },
        {
          "epoch": 38,
          "updates": 608,
          "samples_seen": 38912,
          "online_loss": 0.012484577498980798,
          "train_loss": 0.010533424094319344,
          "validation_loss": 0.008602084591984749,
          "train_metric": 0.0010608879383653402,
          "validation_metric": 0.0008663704502396286,
          "lr_first": 0.003,
          "lr_last": 0.003,
          "seconds": 0.4824385150004673
        },
        {
          "epoch": 39,
          "updates": 624,
          "samples_seen": 39936,
          "online_loss": 0.005340583942597732,
          "train_loss": 0.004230243153870106,
          "validation_loss": 0.003673012601211667,
          "train_metric": 0.0004260546702425927,
          "validation_metric": 0.0003699323860928416,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.4932964939998783
        },
        {
          "epoch": 40,
          "updates": 640,
          "samples_seen": 40960,
          "online_loss": 0.003958030189096462,
          "train_loss": 0.0034799654968082905,
          "validation_loss": 0.0027653248980641365,
          "train_metric": 0.0003504893684294075,
          "validation_metric": 0.00027851335471495986,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.504016822000267
        },
        {
          "epoch": 41,
          "updates": 656,
          "samples_seen": 41984,
          "online_loss": 0.003779596743697766,
          "train_loss": 0.0027438648976385593,
          "validation_loss": 0.0030321215745061636,
          "train_metric": 0.0002763520460575819,
          "validation_metric": 0.0003053841646760702,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5147444209997047
        },
        {
          "epoch": 42,
          "updates": 672,
          "samples_seen": 43008,
          "online_loss": 0.003792784315010067,
          "train_loss": 0.002959120785817504,
          "validation_loss": 0.002368425950407982,
          "train_metric": 0.00029803180950693786,
          "validation_metric": 0.00023853919992689043,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5252152140001272
        },
        {
          "epoch": 43,
          "updates": 688,
          "samples_seen": 44032,
          "online_loss": 0.003266840103606228,
          "train_loss": 0.0026186981704086065,
          "validation_loss": 0.0020507124718278646,
          "train_metric": 0.0002637456636875868,
          "validation_metric": 0.00020654025138355792,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5364626569989923
        },
        {
          "epoch": 44,
          "updates": 704,
          "samples_seen": 45056,
          "online_loss": 0.0031263779928849544,
          "train_loss": 0.0018942949827760458,
          "validation_loss": 0.0015515806153416634,
          "train_metric": 0.00019078644982073456,
          "validation_metric": 0.00015626955428160727,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.548325379999369
        },
        {
          "epoch": 45,
          "updates": 720,
          "samples_seen": 46080,
          "online_loss": 0.002282231584104011,
          "train_loss": 0.002708063693717122,
          "validation_loss": 0.001892526401206851,
          "train_metric": 0.0002727462269831449,
          "validation_metric": 0.00019060831982642412,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5594365889992332
        },
        {
          "epoch": 46,
          "updates": 736,
          "samples_seen": 47104,
          "online_loss": 0.003364461757882964,
          "train_loss": 0.004969216883182526,
          "validation_loss": 0.004811704624444246,
          "train_metric": 0.0005004813428968191,
          "validation_metric": 0.0004846172232646495,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.570243620999463
        },
        {
          "epoch": 47,
          "updates": 752,
          "samples_seen": 48128,
          "online_loss": 0.0042546289405436255,
          "train_loss": 0.0032602050341665745,
          "validation_loss": 0.002327308990061283,
          "train_metric": 0.00032835587626323104,
          "validation_metric": 0.0002343980740988627,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5810083849992225
        },
        {
          "epoch": 48,
          "updates": 768,
          "samples_seen": 49152,
          "online_loss": 0.0024624920843052678,
          "train_loss": 0.0020010184962302446,
          "validation_loss": 0.002191733568906784,
          "train_metric": 0.00020153526565991342,
          "validation_metric": 0.00022074338630773127,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.5917792700001883
        },
        {
          "epoch": 49,
          "updates": 784,
          "samples_seen": 50176,
          "online_loss": 0.0031925522671372164,
          "train_loss": 0.0037992135621607304,
          "validation_loss": 0.003762789536267519,
          "train_metric": 0.00038264290196821094,
          "validation_metric": 0.00037897436413913965,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6022071369989135
        },
        {
          "epoch": 50,
          "updates": 800,
          "samples_seen": 51200,
          "online_loss": 0.004498012462136103,
          "train_loss": 0.0018756872741505504,
          "validation_loss": 0.0010174153139814734,
          "train_metric": 0.00018891235231421888,
          "validation_metric": 0.00010247033787891269,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6132182000001194
        },
        {
          "epoch": 51,
          "updates": 816,
          "samples_seen": 52224,
          "online_loss": 0.0016706399583199527,
          "train_loss": 0.0014610332436859608,
          "validation_loss": 0.0009999549947679043,
          "train_metric": 0.00014714991266373545,
          "validation_metric": 0.00010071179713122547,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6247349370005395
        },
        {
          "epoch": 52,
          "updates": 832,
          "samples_seen": 53248,
          "online_loss": 0.0021081952581880614,
          "train_loss": 0.0025933822616934776,
          "validation_loss": 0.001571556436829269,
          "train_metric": 0.0002611959644127637,
          "validation_metric": 0.0001582814147695899,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6354683949994069
        },
        {
          "epoch": 53,
          "updates": 848,
          "samples_seen": 54272,
          "online_loss": 0.0020395970896061044,
          "train_loss": 0.0013225739821791649,
          "validation_loss": 0.0012418730184435844,
          "train_metric": 0.00013320479774847627,
          "validation_metric": 0.0001250768982572481,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.646295745000316
        },
        {
          "epoch": 54,
          "updates": 864,
          "samples_seen": 55296,
          "online_loss": 0.0028531694151752163,
          "train_loss": 0.004799867980182171,
          "validation_loss": 0.005346137098968029,
          "train_metric": 0.0004834251303691417,
          "validation_metric": 0.0005384434480220079,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6570429990006232
        },
        {
          "epoch": 55,
          "updates": 880,
          "samples_seen": 56320,
          "online_loss": 0.004429804837855045,
          "train_loss": 0.004045144654810429,
          "validation_loss": 0.004567306488752365,
          "train_metric": 0.00040741218253970146,
          "validation_metric": 0.0004600024258252233,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6678255650003848
        },
        {
          "epoch": 56,
          "updates": 896,
          "samples_seen": 57344,
          "online_loss": 0.0031316953027271666,
          "train_loss": 0.0032288390211760998,
          "validation_loss": 0.0019552449230104685,
          "train_metric": 0.00032519683009013534,
          "validation_metric": 0.00019692510250024498,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6785011169995414
        },
        {
          "epoch": 57,
          "updates": 912,
          "samples_seen": 58368,
          "online_loss": 0.005380759339459473,
          "train_loss": 0.006660917773842812,
          "validation_loss": 0.0060523864813148975,
          "train_metric": 0.0006708631990477443,
          "validation_metric": 0.0006095741991885006,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6892094880004151
        },
        {
          "epoch": 58,
          "updates": 928,
          "samples_seen": 59392,
          "online_loss": 0.003408056953048799,
          "train_loss": 0.002101266523823142,
          "validation_loss": 0.0021351310424506664,
          "train_metric": 0.00021163185010664165,
          "validation_metric": 0.00021504257165361196,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.6997091199991701
        },
        {
          "epoch": 59,
          "updates": 944,
          "samples_seen": 60416,
          "online_loss": 0.0018077379900205415,
          "train_loss": 0.001777643570676446,
          "validation_loss": 0.0014206002233549953,
          "train_metric": 0.00017903775733429939,
          "validation_metric": 0.00014307766105048358,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7105396299994027
        },
        {
          "epoch": 60,
          "updates": 960,
          "samples_seen": 61440,
          "online_loss": 0.001430878310202388,
          "train_loss": 0.0011914899805560708,
          "validation_loss": 0.0008654165430925786,
          "train_metric": 0.00012000252172583714,
          "validation_metric": 8.716157753951848e-05,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7216877379996731
        },
        {
          "epoch": 61,
          "updates": 976,
          "samples_seen": 62464,
          "online_loss": 0.001971792684344109,
          "train_loss": 0.0015797116793692112,
          "validation_loss": 0.0012506500352174044,
          "train_metric": 0.00015910275396890938,
          "validation_metric": 0.0001259608834516257,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7322007179991488
        },
        {
          "epoch": 62,
          "updates": 992,
          "samples_seen": 63488,
          "online_loss": 0.005401995935244486,
          "train_loss": 0.003050468862056732,
          "validation_loss": 0.0030622242484241724,
          "train_metric": 0.00030723202507942915,
          "validation_metric": 0.00030841605621390045,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7427229879995139
        },
        {
          "epoch": 63,
          "updates": 1008,
          "samples_seen": 64512,
          "online_loss": 0.003195493158273166,
          "train_loss": 0.0018956264248117805,
          "validation_loss": 0.0013271392090246081,
          "train_metric": 0.00019092054571956396,
          "validation_metric": 0.00013366462371777743,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7532647999996698
        },
        {
          "epoch": 64,
          "updates": 1024,
          "samples_seen": 65536,
          "online_loss": 0.001677237029070966,
          "train_loss": 0.0017810814315453172,
          "validation_loss": 0.0013566982233896852,
          "train_metric": 0.00017938400560524315,
          "validation_metric": 0.00013664166908711195,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7636646619994281
        },
        {
          "epoch": 65,
          "updates": 1040,
          "samples_seen": 66560,
          "online_loss": 0.003182141444995068,
          "train_loss": 0.005172677803784609,
          "validation_loss": 0.005077451933175325,
          "train_metric": 0.0005209732335060835,
          "validation_metric": 0.0005113823572173715,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7743392089996632
        },
        {
          "epoch": 66,
          "updates": 1056,
          "samples_seen": 67584,
          "online_loss": 0.003169672407238977,
          "train_loss": 0.0015204192604869604,
          "validation_loss": 0.0010483849328011274,
          "train_metric": 0.00015313106996472925,
          "validation_metric": 0.00010558950452832505,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7849824959994294
        },
        {
          "epoch": 67,
          "updates": 1072,
          "samples_seen": 68608,
          "online_loss": 0.0020033447108289693,
          "train_loss": 0.0025511691346764565,
          "validation_loss": 0.0019035914447158575,
          "train_metric": 0.00025694441865198314,
          "validation_metric": 0.00019172276370227337,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.7958574539989058
        },
        {
          "epoch": 68,
          "updates": 1088,
          "samples_seen": 69632,
          "online_loss": 0.0016619744819763582,
          "train_loss": 0.0015163269126787782,
          "validation_loss": 0.0011533438228070736,
          "train_metric": 0.00015271891606971622,
          "validation_metric": 0.00011616056872298941,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8065515399994183
        },
        {
          "epoch": 69,
          "updates": 1104,
          "samples_seen": 70656,
          "online_loss": 0.0015179603487922577,
          "train_loss": 0.0010273667285218835,
          "validation_loss": 0.0009914685506373644,
          "train_metric": 0.00010347260831622407,
          "validation_metric": 9.985708311432973e-05,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8173644130001776
        },
        {
          "epoch": 70,
          "updates": 1120,
          "samples_seen": 71680,
          "online_loss": 0.0015253068613674259,
          "train_loss": 0.0022541654761880636,
          "validation_loss": 0.0014503425918519497,
          "train_metric": 0.00022703129798173904,
          "validation_metric": 0.00014607320190407336,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8279460230005498
        },
        {
          "epoch": 71,
          "updates": 1136,
          "samples_seen": 72704,
          "online_loss": 0.0014404388530238066,
          "train_loss": 0.0015009132912382483,
          "validation_loss": 0.0010611419565975666,
          "train_metric": 0.00015116647409740835,
          "validation_metric": 0.00010687434405554086,
          "lr_first": 0.0015,
          "lr_last": 0.0015,
          "seconds": 0.8385213929996098
        },
        {
          "epoch": 72,
          "updates": 1152,
          "samples_seen": 73728,
          "online_loss": 0.0014964083711674903,
          "train_loss": 0.000843506189994514,
          "validation_loss": 0.0005511737545020878,
          "train_metric": 8.495486690662801e-05,
          "validation_metric": 5.5512198741780594e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8500964169998042
        },
        {
          "epoch": 73,
          "updates": 1168,
          "samples_seen": 74752,
          "online_loss": 0.0020285771397539065,
          "train_loss": 0.0011159328278154135,
          "validation_loss": 0.0009043682948686182,
          "train_metric": 0.00011239266314078122,
          "validation_metric": 9.108466474572197e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8606943929989939
        },
        {
          "epoch": 74,
          "updates": 1184,
          "samples_seen": 75776,
          "online_loss": 0.00159594154865772,
          "train_loss": 0.002564321970567107,
          "validation_loss": 0.002573132049292326,
          "train_metric": 0.0002582691377028823,
          "validation_metric": 0.00025915642618201673,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8712535249997018
        },
        {
          "epoch": 75,
          "updates": 1200,
          "samples_seen": 76800,
          "online_loss": 0.001989463178688311,
          "train_loss": 0.0012428787304088473,
          "validation_loss": 0.0012404659064486623,
          "train_metric": 0.000125178208691068,
          "validation_metric": 0.00012493517715483904,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.88222170500012
        },
        {
          "epoch": 76,
          "updates": 1216,
          "samples_seen": 77824,
          "online_loss": 0.0011931037697650027,
          "train_loss": 0.0009971588151529431,
          "validation_loss": 0.0006950655952095985,
          "train_metric": 0.0001004301811917685,
          "validation_metric": 7.000446203164756e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.8933126110005105
        },
        {
          "epoch": 77,
          "updates": 1232,
          "samples_seen": 78848,
          "online_loss": 0.0009488010600762209,
          "train_loss": 0.0008507153252139688,
          "validation_loss": 0.0006065382622182369,
          "train_metric": 8.568093471694738e-05,
          "validation_metric": 6.108831439632922e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9040037220001977
        },
        {
          "epoch": 78,
          "updates": 1248,
          "samples_seen": 79872,
          "online_loss": 0.0008343484760189313,
          "train_loss": 0.0008572710212320089,
          "validation_loss": 0.0006751189939677715,
          "train_metric": 8.634119149064645e-05,
          "validation_metric": 6.799550465075299e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9147551370006113
        },
        {
          "epoch": 79,
          "updates": 1264,
          "samples_seen": 80896,
          "online_loss": 0.0009182748162857024,
          "train_loss": 0.0009832093492150307,
          "validation_loss": 0.0007327371276915073,
          "train_metric": 9.902524470817298e-05,
          "validation_metric": 7.37986047170125e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9253168029990775
        },
        {
          "epoch": 80,
          "updates": 1280,
          "samples_seen": 81920,
          "online_loss": 0.001212134913657792,
          "train_loss": 0.0017133249202743173,
          "validation_loss": 0.0011205439222976565,
          "train_metric": 0.00017255979764740914,
          "validation_metric": 0.00011285710206720978,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9359902790001797
        },
        {
          "epoch": 81,
          "updates": 1296,
          "samples_seen": 82944,
          "online_loss": 0.00227742695096822,
          "train_loss": 0.0024035379756242037,
          "validation_loss": 0.002715857233852148,
          "train_metric": 0.000242075533606112,
          "validation_metric": 0.0002735312154982239,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9466007339997304
        },
        {
          "epoch": 82,
          "updates": 1312,
          "samples_seen": 83968,
          "online_loss": 0.004072214724146761,
          "train_loss": 0.0035324934870004654,
          "validation_loss": 0.002653245348483324,
          "train_metric": 0.0003557798045221716,
          "validation_metric": 0.00026722511393018067,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.957169471999805
        },
        {
          "epoch": 83,
          "updates": 1328,
          "samples_seen": 84992,
          "online_loss": 0.002495613489372772,
          "train_loss": 0.0007845122599974275,
          "validation_loss": 0.0007261664140969515,
          "train_metric": 7.901319622760639e-05,
          "validation_metric": 7.313682726817206e-05,
          "lr_first": 0.00075,
          "lr_last": 0.00075,
          "seconds": 0.9676553150002292
        },
        {
          "epoch": 84,
          "updates": 1344,
          "samples_seen": 86016,
          "online_loss": 0.0008701173646841198,
          "train_loss": 0.0006956751458346844,
          "validation_loss": 0.00046837294939905405,
          "train_metric": 7.006584928603843e-05,
          "validation_metric": 4.717280171462335e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.9792131209997024
        },
        {
          "epoch": 85,
          "updates": 1360,
          "samples_seen": 87040,
          "online_loss": 0.0007068669920045068,
          "train_loss": 0.0006698669749312103,
          "validation_loss": 0.0006065756315365434,
          "train_metric": 6.746654253220186e-05,
          "validation_metric": 6.109207606641576e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 0.9900582819991541
        },
        {
          "epoch": 86,
          "updates": 1376,
          "samples_seen": 88064,
          "online_loss": 0.0006383104282576824,
          "train_loss": 0.0005993851809762418,
          "validation_loss": 0.0004608859308063984,
          "train_metric": 6.036787817720324e-05,
          "validation_metric": 4.641874329536222e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0013690670002688
        },
        {
          "epoch": 87,
          "updates": 1392,
          "samples_seen": 89088,
          "online_loss": 0.0006607069499295903,
          "train_loss": 0.0006324349087662995,
          "validation_loss": 0.00045339783537201583,
          "train_metric": 6.369652692228556e-05,
          "validation_metric": 4.566457209875807e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0126733409997541
        },
        {
          "epoch": 88,
          "updates": 1408,
          "samples_seen": 90112,
          "online_loss": 0.000619731581537053,
          "train_loss": 0.0005477811791934073,
          "validation_loss": 0.0004459129413589835,
          "train_metric": 5.517050885828212e-05,
          "validation_metric": 4.491072104428895e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0237965649994294
        },
        {
          "epoch": 89,
          "updates": 1424,
          "samples_seen": 91136,
          "online_loss": 0.0006693748637189856,
          "train_loss": 0.0005220915772952139,
          "validation_loss": 0.00042084953747689724,
          "train_metric": 5.258315650280565e-05,
          "validation_metric": 4.238642213749699e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0347884950006119
        },
        {
          "epoch": 90,
          "updates": 1440,
          "samples_seen": 92160,
          "online_loss": 0.0006681859322270611,
          "train_loss": 0.0011663727927953005,
          "validation_loss": 0.0008670196402817965,
          "train_metric": 0.000117472794954665,
          "validation_metric": 8.732305286685005e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.045302800999707
        },
        {
          "epoch": 91,
          "updates": 1456,
          "samples_seen": 93184,
          "online_loss": 0.0008925436559366062,
          "train_loss": 0.0008190940716303885,
          "validation_loss": 0.0007454908918589354,
          "train_metric": 8.249613892985508e-05,
          "validation_metric": 7.508311682613567e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.05571532400063
        },
        {
          "epoch": 92,
          "updates": 1472,
          "samples_seen": 94208,
          "online_loss": 0.0008474731203023111,
          "train_loss": 0.0005371115985326469,
          "validation_loss": 0.0004170157480984926,
          "train_metric": 5.4095915402285755e-05,
          "validation_metric": 4.2000290704891086e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0665731110002525
        },
        {
          "epoch": 93,
          "updates": 1488,
          "samples_seen": 95232,
          "online_loss": 0.0006023637361067813,
          "train_loss": 0.0005184495821595192,
          "validation_loss": 0.00039884104626253247,
          "train_metric": 5.221634637564421e-05,
          "validation_metric": 4.0169812564272434e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.077753842000675
        },
        {
          "epoch": 94,
          "updates": 1504,
          "samples_seen": 96256,
          "online_loss": 0.0005925915520492708,
          "train_loss": 0.0005057367961853743,
          "validation_loss": 0.00040048666414804757,
          "train_metric": 5.093595609650947e-05,
          "validation_metric": 4.0335555240744725e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0884007370004838
        },
        {
          "epoch": 95,
          "updates": 1520,
          "samples_seen": 97280,
          "online_loss": 0.0005337636184776784,
          "train_loss": 0.00048719157348386943,
          "validation_loss": 0.00035465986002236605,
          "train_metric": 4.906815229333006e-05,
          "validation_metric": 3.5720044252229854e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.0993818779988942
        },
        {
          "epoch": 96,
          "updates": 1536,
          "samples_seen": 98304,
          "online_loss": 0.0005283006994432071,
          "train_loss": 0.0005378840724006295,
          "validation_loss": 0.00046729634050279856,
          "train_metric": 5.417371357907541e-05,
          "validation_metric": 4.70643826702144e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.110083745999873
        },
        {
          "epoch": 97,
          "updates": 1552,
          "samples_seen": 99328,
          "online_loss": 0.0005313867732184008,
          "train_loss": 0.0004453692235983908,
          "validation_loss": 0.00034873277763836086,
          "train_metric": 4.48559585493058e-05,
          "validation_metric": 3.5123088309774175e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1211519539992878
        },
        {
          "epoch": 98,
          "updates": 1568,
          "samples_seen": 100352,
          "online_loss": 0.0005509997390618082,
          "train_loss": 0.0004511236911639571,
          "validation_loss": 0.0003458083374425769,
          "train_metric": 4.543552495306358e-05,
          "validation_metric": 3.4828550269594416e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.132286063999345
        },
        {
          "epoch": 99,
          "updates": 1584,
          "samples_seen": 101376,
          "online_loss": 0.0006912787721375935,
          "train_loss": 0.0007766124908812344,
          "validation_loss": 0.000625491316895932,
          "train_metric": 7.82175557105802e-05,
          "validation_metric": 6.299718370428309e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1431184129996836
        },
        {
          "epoch": 100,
          "updates": 1600,
          "samples_seen": 102400,
          "online_loss": 0.0005400173922680551,
          "train_loss": 0.0006971806287765503,
          "validation_loss": 0.0004150439635850489,
          "train_metric": 7.021748024271801e-05,
          "validation_metric": 4.180170799372718e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1537866489998123
        },
        {
          "epoch": 101,
          "updates": 1616,
          "samples_seen": 103424,
          "online_loss": 0.0006987284850765718,
          "train_loss": 0.000843168469145894,
          "validation_loss": 0.0008488795720040798,
          "train_metric": 8.49208518047817e-05,
          "validation_metric": 8.549603808205575e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1641967989999102
        },
        {
          "epoch": 102,
          "updates": 1632,
          "samples_seen": 104448,
          "online_loss": 0.0007000443765718956,
          "train_loss": 0.0005398172652348876,
          "validation_loss": 0.00044017963227815926,
          "train_metric": 5.436841456685215e-05,
          "validation_metric": 4.433327922015451e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.1750440209998487
        },
        {
          "epoch": 103,
          "updates": 1648,
          "samples_seen": 105472,
          "online_loss": 0.000468253664621443,
          "train_loss": 0.0008504493744112551,
          "validation_loss": 0.000756210065446794,
          "train_metric": 8.565415191696957e-05,
          "validation_metric": 7.616270886501297e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.185322219000227
        },
        {
          "epoch": 104,
          "updates": 1664,
          "samples_seen": 106496,
          "online_loss": 0.0008260653530669515,
          "train_loss": 0.0004479802446439862,
          "validation_loss": 0.0003722138935700059,
          "train_metric": 4.5118937123334035e-05,
          "validation_metric": 3.748802555492148e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.19565630499892
        },
        {
          "epoch": 105,
          "updates": 1680,
          "samples_seen": 107520,
          "online_loss": 0.0005999941340633086,
          "train_loss": 0.0004657612007576972,
          "validation_loss": 0.00032526947325095534,
          "train_metric": 4.6909764932934195e-05,
          "validation_metric": 3.275994458817877e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2067018550005741
        },
        {
          "epoch": 106,
          "updates": 1696,
          "samples_seen": 108544,
          "online_loss": 0.00045484592374123167,
          "train_loss": 0.0005409568548202515,
          "validation_loss": 0.0004488911072257906,
          "train_metric": 5.448318916023709e-05,
          "validation_metric": 4.521067239693366e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2175885020005808
        },
        {
          "epoch": 107,
          "updates": 1712,
          "samples_seen": 109568,
          "online_loss": 0.0004908250612061238,
          "train_loss": 0.0004001584311481565,
          "validation_loss": 0.0002980621939059347,
          "train_metric": 4.0302493289345875e-05,
          "validation_metric": 3.0019733458175324e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.228448304000267
        },
        {
          "epoch": 108,
          "updates": 1728,
          "samples_seen": 110592,
          "online_loss": 0.0005983905011817114,
          "train_loss": 0.0005870725144632161,
          "validation_loss": 0.0004464038647711277,
          "train_metric": 5.912779306527227e-05,
          "validation_metric": 4.496015753829852e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2390604789998179
        },
        {
          "epoch": 109,
          "updates": 1744,
          "samples_seen": 111616,
          "online_loss": 0.0004827441807719879,
          "train_loss": 0.0003902669413946569,
          "validation_loss": 0.0002997760602738708,
          "train_metric": 3.930625462089665e-05,
          "validation_metric": 3.0192346457624808e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2494210109998676
        },
        {
          "epoch": 110,
          "updates": 1760,
          "samples_seen": 112640,
          "online_loss": 0.0005665363687512581,
          "train_loss": 0.0011384909739717841,
          "validation_loss": 0.0008203002507798374,
          "train_metric": 0.00011466463911347091,
          "validation_metric": 8.261763287009671e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.259623781999835
        },
        {
          "epoch": 111,
          "updates": 1776,
          "samples_seen": 113664,
          "online_loss": 0.0006922082320670597,
          "train_loss": 0.0006019834545440972,
          "validation_loss": 0.0005475224461406469,
          "train_metric": 6.0629568906733766e-05,
          "validation_metric": 5.514446093002334e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2699907869991875
        },
        {
          "epoch": 112,
          "updates": 1792,
          "samples_seen": 114688,
          "online_loss": 0.0005106444623379502,
          "train_loss": 0.0003945694479625672,
          "validation_loss": 0.000296755664749071,
          "train_metric": 3.9739588828524575e-05,
          "validation_metric": 2.988814412674401e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.280873061999955
        },
        {
          "epoch": 113,
          "updates": 1808,
          "samples_seen": 115712,
          "online_loss": 0.00048763505856186384,
          "train_loss": 0.001366328913718462,
          "validation_loss": 0.0011787258554250002,
          "train_metric": 0.00013761164154857397,
          "validation_metric": 0.00011871694005094469,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.2914107800006605
        },
        {
          "epoch": 114,
          "updates": 1824,
          "samples_seen": 116736,
          "online_loss": 0.0009404624097442138,
          "train_loss": 0.0003464885812718421,
          "validation_loss": 0.0002557314292062074,
          "train_metric": 3.4897064324468374e-05,
          "validation_metric": 2.575633152446244e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3026256970006216
        },
        {
          "epoch": 115,
          "updates": 1840,
          "samples_seen": 117760,
          "online_loss": 0.0004283728640075424,
          "train_loss": 0.00041304799378849566,
          "validation_loss": 0.00029907116550020874,
          "train_metric": 4.1600676922826096e-05,
          "validation_metric": 3.012135493918322e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3137869509992015
        },
        {
          "epoch": 116,
          "updates": 1856,
          "samples_seen": 118784,
          "online_loss": 0.0003918726147276175,
          "train_loss": 0.000370957946870476,
          "validation_loss": 0.0003360460978001356,
          "train_metric": 3.736152939382009e-05,
          "validation_metric": 3.3845331927295774e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3243138020006882
        },
        {
          "epoch": 117,
          "updates": 1872,
          "samples_seen": 119808,
          "online_loss": 0.00037772477480757516,
          "train_loss": 0.0003448660718277097,
          "validation_loss": 0.000251723948167637,
          "train_metric": 3.473364267847501e-05,
          "validation_metric": 2.535271596570965e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.335133151998889
        },
        {
          "epoch": 118,
          "updates": 1888,
          "samples_seen": 120832,
          "online_loss": 0.00037904694727330934,
          "train_loss": 0.0003402280271984637,
          "validation_loss": 0.0002685183717403561,
          "train_metric": 3.4266522561665624e-05,
          "validation_metric": 2.7044186936109327e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3456521650005016
        },
        {
          "epoch": 119,
          "updates": 1904,
          "samples_seen": 121856,
          "online_loss": 0.00041750468881218694,
          "train_loss": 0.0003597447066567838,
          "validation_loss": 0.00025031858240254223,
          "train_metric": 3.6232173442840576e-05,
          "validation_metric": 2.5211171305272728e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3567336140004045
        },
        {
          "epoch": 120,
          "updates": 1920,
          "samples_seen": 122880,
          "online_loss": 0.0006041453025318333,
          "train_loss": 0.00032441000803373754,
          "validation_loss": 0.0002659092715475708,
          "train_metric": 3.267338979640044e-05,
          "validation_metric": 2.6781406631926075e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3672663489996921
        },
        {
          "epoch": 121,
          "updates": 1936,
          "samples_seen": 123904,
          "online_loss": 0.0003904484010490705,
          "train_loss": 0.0003045335179194808,
          "validation_loss": 0.0002487014280632138,
          "train_metric": 3.067150100832805e-05,
          "validation_metric": 2.5048298994079232e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.378382139999303
        },
        {
          "epoch": 122,
          "updates": 1952,
          "samples_seen": 124928,
          "online_loss": 0.0003688784308906179,
          "train_loss": 0.0006275029736571014,
          "validation_loss": 0.00044795399298891425,
          "train_metric": 6.319979729596525e-05,
          "validation_metric": 4.5116285036783665e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3889048420005565
        },
        {
          "epoch": 123,
          "updates": 1968,
          "samples_seen": 125952,
          "online_loss": 0.0007111332306521945,
          "train_loss": 0.0021258313208818436,
          "validation_loss": 0.0018453336087986827,
          "train_metric": 0.00021410595218185335,
          "validation_metric": 0.00018585524230729789,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.3992214840000088
        },
        {
          "epoch": 124,
          "updates": 1984,
          "samples_seen": 126976,
          "online_loss": 0.0010522935954213608,
          "train_loss": 0.000982508179731667,
          "validation_loss": 0.0008631922537460923,
          "train_metric": 9.89546169876121e-05,
          "validation_metric": 8.693755808053538e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4096382560001075
        },
        {
          "epoch": 125,
          "updates": 2000,
          "samples_seen": 128000,
          "online_loss": 0.000768384636103292,
          "train_loss": 0.00032966729486361146,
          "validation_loss": 0.0002778651542030275,
          "train_metric": 3.320288669783622e-05,
          "validation_metric": 2.7985559427179396e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4199762899988855
        },
        {
          "epoch": 126,
          "updates": 2016,
          "samples_seen": 129024,
          "online_loss": 0.0003478840171737829,
          "train_loss": 0.0003117503074463457,
          "validation_loss": 0.00024146052601281554,
          "train_metric": 3.139835098409094e-05,
          "validation_metric": 2.431902066746261e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4308416429994395
        },
        {
          "epoch": 127,
          "updates": 2032,
          "samples_seen": 130048,
          "online_loss": 0.00036173600619804347,
          "train_loss": 0.000820610614027828,
          "validation_loss": 0.0006031462689861655,
          "train_metric": 8.264889765996486e-05,
          "validation_metric": 6.0746693634428084e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4413721000000805
        },
        {
          "epoch": 128,
          "updates": 2048,
          "samples_seen": 131072,
          "online_loss": 0.0004490610517677851,
          "train_loss": 0.00029991596238687634,
          "validation_loss": 0.00021668038971256465,
          "train_metric": 3.020643998752348e-05,
          "validation_metric": 2.182325624744408e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4523034499998175
        },
        {
          "epoch": 129,
          "updates": 2064,
          "samples_seen": 132096,
          "online_loss": 0.00030620830602856586,
          "train_loss": 0.0006654991302639246,
          "validation_loss": 0.00042452028719708323,
          "train_metric": 6.702663813484833e-05,
          "validation_metric": 4.2756124457810074e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.462775927999246
        },
        {
          "epoch": 130,
          "updates": 2080,
          "samples_seen": 133120,
          "online_loss": 0.0005116830016049789,
          "train_loss": 0.0003119233588222414,
          "validation_loss": 0.00022276115487329662,
          "train_metric": 3.141577690257691e-05,
          "validation_metric": 2.2435688151745126e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4731496469994454
        },
        {
          "epoch": 131,
          "updates": 2096,
          "samples_seen": 134144,
          "online_loss": 0.0003531554921210045,
          "train_loss": 0.00027423532446846366,
          "validation_loss": 0.00020882398530375212,
          "train_metric": 2.7619977117865346e-05,
          "validation_metric": 2.1031988580944017e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.4843473300006735
        },
        {
          "epoch": 132,
          "updates": 2112,
          "samples_seen": 135168,
          "online_loss": 0.00035998752719024196,
          "train_loss": 0.0006240218644961715,
          "validation_loss": 0.0004997733631171286,
          "train_metric": 6.284919800236821e-05,
          "validation_metric": 5.0335343985352665e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.495002611000018
        },
        {
          "epoch": 133,
          "updates": 2128,
          "samples_seen": 136192,
          "online_loss": 0.0003371636921656318,
          "train_loss": 0.0003034695109818131,
          "validation_loss": 0.0002213252300862223,
          "train_metric": 3.056434070458636e-05,
          "validation_metric": 2.229106758022681e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.50570178199996
        },
        {
          "epoch": 134,
          "updates": 2144,
          "samples_seen": 137216,
          "online_loss": 0.0004825937148780213,
          "train_loss": 0.0006174672162160277,
          "validation_loss": 0.0005699259927496314,
          "train_metric": 6.218903581611812e-05,
          "validation_metric": 5.7400859077461064e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.516371858000639
        },
        {
          "epoch": 135,
          "updates": 2160,
          "samples_seen": 138240,
          "online_loss": 0.00038741136359021766,
          "train_loss": 0.0002657816803548485,
          "validation_loss": 0.00019705368322320282,
          "train_metric": 2.6768560928758234e-05,
          "validation_metric": 1.9846524082822725e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5272835229989141
        },
        {
          "epoch": 136,
          "updates": 2176,
          "samples_seen": 139264,
          "online_loss": 0.0003964353427363676,
          "train_loss": 0.0008399318321608007,
          "validation_loss": 0.0006169424159452319,
          "train_metric": 8.459485979983583e-05,
          "validation_metric": 6.213619053596631e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5377467599992087
        },
        {
          "epoch": 137,
          "updates": 2192,
          "samples_seen": 140288,
          "online_loss": 0.0006074074481148273,
          "train_loss": 0.0002821312809828669,
          "validation_loss": 0.0001935452310135588,
          "train_metric": 2.8415230190148577e-05,
          "validation_metric": 1.9493167201289907e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.548590903999866
        },
        {
          "epoch": 138,
          "updates": 2208,
          "samples_seen": 141312,
          "online_loss": 0.0005031224463891704,
          "train_loss": 0.00042860067333094776,
          "validation_loss": 0.0003237599739804864,
          "train_metric": 4.316708873375319e-05,
          "validation_metric": 3.260792072978802e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5591612240004906
        },
        {
          "epoch": 139,
          "updates": 2224,
          "samples_seen": 142336,
          "online_loss": 0.0003811296173807932,
          "train_loss": 0.00032635184470564127,
          "validation_loss": 0.0002559703425504267,
          "train_metric": 3.2868963899090886e-05,
          "validation_metric": 2.5780393116292544e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5698462439995637
        },
        {
          "epoch": 140,
          "updates": 2240,
          "samples_seen": 143360,
          "online_loss": 0.0003331890820845729,
          "train_loss": 0.0003935746499337256,
          "validation_loss": 0.00036410827306099236,
          "train_metric": 3.9639402530156076e-05,
          "validation_metric": 3.667164855869487e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.580162621999989
        },
        {
          "epoch": 141,
          "updates": 2256,
          "samples_seen": 144384,
          "online_loss": 0.0004481005780689884,
          "train_loss": 0.0003914705594070256,
          "validation_loss": 0.00030650716507807374,
          "train_metric": 3.9427486626664177e-05,
          "validation_metric": 3.0870276532368734e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.5905316650005261
        },
        {
          "epoch": 142,
          "updates": 2272,
          "samples_seen": 145408,
          "online_loss": 0.0002869001518774894,
          "train_loss": 0.0006783499266020954,
          "validation_loss": 0.0005354427848942578,
          "train_metric": 6.832091457908973e-05,
          "validation_metric": 5.3927829867461696e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6013647769996169
        },
        {
          "epoch": 143,
          "updates": 2288,
          "samples_seen": 146432,
          "online_loss": 0.0006123257917352021,
          "train_loss": 0.0011126358294859529,
          "validation_loss": 0.000758785055950284,
          "train_metric": 0.0001120605957112275,
          "validation_metric": 7.642206765012816e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6118304419997003
        },
        {
          "epoch": 144,
          "updates": 2304,
          "samples_seen": 147456,
          "online_loss": 0.00039765636302036,
          "train_loss": 0.0003088544763159007,
          "validation_loss": 0.0002605428162496537,
          "train_metric": 3.1106694223126397e-05,
          "validation_metric": 2.624091757752467e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6220815090000542
        },
        {
          "epoch": 145,
          "updates": 2320,
          "samples_seen": 148480,
          "online_loss": 0.0006069705614208942,
          "train_loss": 0.0005056319059804082,
          "validation_loss": 0.0005796283949166536,
          "train_metric": 5.092539868201129e-05,
          "validation_metric": 5.837804565089755e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6324030539999512
        },
        {
          "epoch": 146,
          "updates": 2336,
          "samples_seen": 149504,
          "online_loss": 0.00031974642024579225,
          "train_loss": 0.00047997443471103907,
          "validation_loss": 0.00035590617335401475,
          "train_metric": 4.834127321373671e-05,
          "validation_metric": 3.5845572710968554e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.642806539999583
        },
        {
          "epoch": 147,
          "updates": 2352,
          "samples_seen": 150528,
          "online_loss": 0.0004999093880542205,
          "train_loss": 0.00038301540189422667,
          "validation_loss": 0.00034783780574798584,
          "train_metric": 3.8575912185478956e-05,
          "validation_metric": 3.5032950108870864e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.653141112999947
        },
        {
          "epoch": 148,
          "updates": 2368,
          "samples_seen": 151552,
          "online_loss": 0.0003517675531838904,
          "train_loss": 0.00031076959567144513,
          "validation_loss": 0.00029436880140565336,
          "train_metric": 3.129957622149959e-05,
          "validation_metric": 2.9647744668181986e-05,
          "lr_first": 0.000375,
          "lr_last": 0.000375,
          "seconds": 1.6635459219996847
        },
        {
          "epoch": 149,
          "updates": 2384,
          "samples_seen": 152576,
          "online_loss": 0.00028644320809689816,
          "train_loss": 0.0002948589390143752,
          "validation_loss": 0.00026483333203941584,
          "train_metric": 2.969711204059422e-05,
          "validation_metric": 2.6673042157199234e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6738083549989824
        },
        {
          "epoch": 150,
          "updates": 2400,
          "samples_seen": 153600,
          "online_loss": 0.0002584164990366844,
          "train_loss": 0.0002268178650410846,
          "validation_loss": 0.0001910840073833242,
          "train_metric": 2.284426591359079e-05,
          "validation_metric": 1.9245282601332292e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6847087650003232
        },
        {
          "epoch": 151,
          "updates": 2416,
          "samples_seen": 154624,
          "online_loss": 0.0003866503584504244,
          "train_loss": 0.0002743983641266823,
          "validation_loss": 0.00025097225443460047,
          "train_metric": 2.7636397135211155e-05,
          "validation_metric": 2.527700416976586e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.6952856570005679
        },
        {
          "epoch": 152,
          "updates": 2432,
          "samples_seen": 155648,
          "online_loss": 0.0005636369814965292,
          "train_loss": 0.0005463949637487531,
          "validation_loss": 0.0004908788832835853,
          "train_metric": 5.503089050762355e-05,
          "validation_metric": 4.943953172187321e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7058138789998338
        },
        {
          "epoch": 153,
          "updates": 2448,
          "samples_seen": 156672,
          "online_loss": 0.0004801316317752935,
          "train_loss": 0.00025242145056836307,
          "validation_loss": 0.000188563484698534,
          "train_metric": 2.5422963517485186e-05,
          "validation_metric": 1.8991420802194625e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7168316879997292
        },
        {
          "epoch": 154,
          "updates": 2464,
          "samples_seen": 157696,
          "online_loss": 0.0003250063446103013,
          "train_loss": 0.00034747799509204924,
          "validation_loss": 0.0002625536872074008,
          "train_metric": 3.499671583995223e-05,
          "validation_metric": 2.6443447495694272e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.728025845999582
        },
        {
          "epoch": 155,
          "updates": 2480,
          "samples_seen": 158720,
          "online_loss": 0.0004047630818604375,
          "train_loss": 0.00024908105842769146,
          "validation_loss": 0.00019663479179143906,
          "train_metric": 2.508653051336296e-05,
          "validation_metric": 1.9804336261586286e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7390287130001525
        },
        {
          "epoch": 156,
          "updates": 2496,
          "samples_seen": 159744,
          "online_loss": 0.00029615815856232075,
          "train_loss": 0.00029795951559208333,
          "validation_loss": 0.00020598665287252516,
          "train_metric": 3.000938886543736e-05,
          "validation_metric": 2.0746221707668155e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7497529390002455
        },
        {
          "epoch": 157,
          "updates": 2512,
          "samples_seen": 160768,
          "online_loss": 0.00041523096206219634,
          "train_loss": 0.0004350168746896088,
          "validation_loss": 0.000323029060382396,
          "train_metric": 4.38133065472357e-05,
          "validation_metric": 3.253430259064771e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7608517890002986
        },
        {
          "epoch": 158,
          "updates": 2528,
          "samples_seen": 161792,
          "online_loss": 0.00047772297830306343,
          "train_loss": 0.00021317134087439626,
          "validation_loss": 0.00016684246656950563,
          "train_metric": 2.146983752027154e-05,
          "validation_metric": 1.6803764083306305e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7721648450005887
        },
        {
          "epoch": 159,
          "updates": 2544,
          "samples_seen": 162816,
          "online_loss": 0.0002261233598801482,
          "train_loss": 0.00020944596326444298,
          "validation_loss": 0.00016011275874916464,
          "train_metric": 2.109463093802333e-05,
          "validation_metric": 1.6125970432767645e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.783757405999495
        },
        {
          "epoch": 160,
          "updates": 2560,
          "samples_seen": 163840,
          "online_loss": 0.00026793491997523233,
          "train_loss": 0.0003631651634350419,
          "validation_loss": 0.00031121482606977224,
          "train_metric": 3.657666457002051e-05,
          "validation_metric": 3.1344417948275805e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.7949910369989084
        },
        {
          "epoch": 161,
          "updates": 2576,
          "samples_seen": 164864,
          "online_loss": 0.0004104247436771402,
          "train_loss": 0.0005075770895928144,
          "validation_loss": 0.0004380839818622917,
          "train_metric": 5.1121311116730794e-05,
          "validation_metric": 4.412221460370347e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8055876349990285
        },
        {
          "epoch": 162,
          "updates": 2592,
          "samples_seen": 165888,
          "online_loss": 0.0003357457735546632,
          "train_loss": 0.00024625493097119033,
          "validation_loss": 0.00016511519788764417,
          "train_metric": 2.48018932325067e-05,
          "validation_metric": 1.6629797755740583e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8161048689999006
        },
        {
          "epoch": 163,
          "updates": 2608,
          "samples_seen": 166912,
          "online_loss": 0.00021767637326775002,
          "train_loss": 0.00025487912353128195,
          "validation_loss": 0.00020541901176329702,
          "train_metric": 2.567049341450911e-05,
          "validation_metric": 2.0689049051725306e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8267614119995415
        },
        {
          "epoch": 164,
          "updates": 2624,
          "samples_seen": 167936,
          "online_loss": 0.00022050567440601299,
          "train_loss": 0.0002039228711510077,
          "validation_loss": 0.00015409319894388318,
          "train_metric": 2.0538367607514374e-05,
          "validation_metric": 1.5519703083555214e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8377170799994929
        },
        {
          "epoch": 165,
          "updates": 2640,
          "samples_seen": 168960,
          "online_loss": 0.0002509558080419083,
          "train_loss": 0.00019998646166641265,
          "validation_loss": 0.00016245729057118297,
          "train_metric": 2.0141902496106923e-05,
          "validation_metric": 1.6362104361178353e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8481029340000532
        },
        {
          "epoch": 166,
          "updates": 2656,
          "samples_seen": 169984,
          "online_loss": 0.0002406021258138935,
          "train_loss": 0.00023250981757882982,
          "validation_loss": 0.00017436749476473778,
          "train_metric": 2.3417536795022897e-05,
          "validation_metric": 1.7561655113240704e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.858597483000267
        },
        {
          "epoch": 167,
          "updates": 2672,
          "samples_seen": 171008,
          "online_loss": 0.0002468565617164131,
          "train_loss": 0.0002822740643750876,
          "validation_loss": 0.00020007371495012194,
          "train_metric": 2.8429607482394204e-05,
          "validation_metric": 2.015069367189426e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.8688477149989922
        },
        {
          "epoch": 168,
          "updates": 2688,
          "samples_seen": 172032,
          "online_loss": 0.00021601728212772286,
          "train_loss": 0.00025612389435991645,
          "validation_loss": 0.00019108643755316734,
          "train_metric": 2.5795859983190894e-05,
          "validation_metric": 1.924552816490177e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.879355762999694
        },
        {
          "epoch": 169,
          "updates": 2704,
          "samples_seen": 173056,
          "online_loss": 0.00028545680879688007,
          "train_loss": 0.0001925661927089095,
          "validation_loss": 0.00013712367217522115,
          "train_metric": 1.939456160471309e-05,
          "validation_metric": 1.3810593372909352e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.890826176999326
        },
        {
          "epoch": 170,
          "updates": 2720,
          "samples_seen": 174080,
          "online_loss": 0.00019955384868808324,
          "train_loss": 0.0001948920835275203,
          "validation_loss": 0.00016359683650080115,
          "train_metric": 1.962881833605934e-05,
          "validation_metric": 1.6476875316584483e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9013585310003691
        },
        {
          "epoch": 171,
          "updates": 2736,
          "samples_seen": 175104,
          "online_loss": 0.0002421858771413099,
          "train_loss": 0.0002277054009027779,
          "validation_loss": 0.00019054840959142894,
          "train_metric": 2.2933656509849243e-05,
          "validation_metric": 1.919134047057014e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9120810980002716
        },
        {
          "epoch": 172,
          "updates": 2752,
          "samples_seen": 176128,
          "online_loss": 0.0002074286949209636,
          "train_loss": 0.00019043123757001013,
          "validation_loss": 0.00014628912322223186,
          "train_metric": 1.917953704833053e-05,
          "validation_metric": 1.4733704119862523e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9224386879996018
        },
        {
          "epoch": 173,
          "updates": 2768,
          "samples_seen": 177152,
          "online_loss": 0.00021398349417722784,
          "train_loss": 0.0005510141490958631,
          "validation_loss": 0.00041311574750579894,
          "train_metric": 5.549612615141086e-05,
          "validation_metric": 4.160750540904701e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9328219409999292
        },
        {
          "epoch": 174,
          "updates": 2784,
          "samples_seen": 178176,
          "online_loss": 0.0003620385537033144,
          "train_loss": 0.0008017574436962605,
          "validation_loss": 0.0005906072910875082,
          "train_metric": 8.07500837254338e-05,
          "validation_metric": 5.9483809309313074e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.943289275999632
        },
        {
          "epoch": 175,
          "updates": 2800,
          "samples_seen": 179200,
          "online_loss": 0.00045080598965796526,
          "train_loss": 0.00029394004377536476,
          "validation_loss": 0.00023609267373103648,
          "train_metric": 2.9604561859741807e-05,
          "validation_metric": 2.377838609390892e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9536720630003401
        },
        {
          "epoch": 176,
          "updates": 2816,
          "samples_seen": 180224,
          "online_loss": 0.0003109308054263238,
          "train_loss": 0.0002258072345284745,
          "validation_loss": 0.00013948159175924957,
          "train_metric": 2.274247890454717e-05,
          "validation_metric": 1.4048075172468089e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9640658730004361
        },
        {
          "epoch": 177,
          "updates": 2832,
          "samples_seen": 181248,
          "online_loss": 0.00028704806663881755,
          "train_loss": 0.00020912753825541586,
          "validation_loss": 0.00017350616690237075,
          "train_metric": 2.1062558516860008e-05,
          "validation_metric": 1.74749056895962e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.974589608000315
        },
        {
          "epoch": 178,
          "updates": 2848,
          "samples_seen": 182272,
          "online_loss": 0.00030609657187596895,
          "train_loss": 0.000272127625066787,
          "validation_loss": 0.0002603580942377448,
          "train_metric": 2.740769923548214e-05,
          "validation_metric": 2.6222309315926395e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9849412189996656
        },
        {
          "epoch": 179,
          "updates": 2864,
          "samples_seen": 183296,
          "online_loss": 0.0006169639491417911,
          "train_loss": 0.0008312284480780363,
          "validation_loss": 0.0005312629509717226,
          "train_metric": 8.371826697839424e-05,
          "validation_metric": 5.3506853873841465e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 1.9954935099995055
        },
        {
          "epoch": 180,
          "updates": 2880,
          "samples_seen": 184320,
          "online_loss": 0.00040703573904465884,
          "train_loss": 0.000219659399590455,
          "validation_loss": 0.0001492545852670446,
          "train_metric": 2.212328945461195e-05,
          "validation_metric": 1.503237581346184e-05,
          "lr_first": 0.0001875,
          "lr_last": 0.0001875,
          "seconds": 2.0058788919996005
        },
        {
          "epoch": 181,
          "updates": 2896,
          "samples_seen": 185344,
          "online_loss": 0.00023483772065446828,
          "train_loss": 0.0001831488189054653,
          "validation_loss": 0.00014253355038817972,
          "train_metric": 1.8446076865075156e-05,
          "validation_metric": 1.4355458915815689e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0161788940004044
        },
        {
          "epoch": 182,
          "updates": 2912,
          "samples_seen": 186368,
          "online_loss": 0.0002381908830102475,
          "train_loss": 0.00017332899733446538,
          "validation_loss": 0.000133434179588221,
          "train_metric": 1.7457061403547414e-05,
          "validation_metric": 1.343900294159539e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0271047659989563
        },
        {
          "epoch": 183,
          "updates": 2928,
          "samples_seen": 187392,
          "online_loss": 0.00019288523208160768,
          "train_loss": 0.0002122207952197641,
          "validation_loss": 0.00017226394265890121,
          "train_metric": 2.1374100469984114e-05,
          "validation_metric": 1.7349791960441507e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0376457429993025
        },
        {
          "epoch": 184,
          "updates": 2944,
          "samples_seen": 188416,
          "online_loss": 0.0001909866359710577,
          "train_loss": 0.00017354890587739646,
          "validation_loss": 0.00013615851639769971,
          "train_metric": 1.7479207599535584e-05,
          "validation_metric": 1.3713387488678563e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0478496390005603
        },
        {
          "epoch": 185,
          "updates": 2960,
          "samples_seen": 189440,
          "online_loss": 0.00021954392968837055,
          "train_loss": 0.0001755991397658363,
          "validation_loss": 0.00013984553515911102,
          "train_metric": 1.7685702914604917e-05,
          "validation_metric": 1.4084730537433643e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.058156711000265
        },
        {
          "epoch": 186,
          "updates": 2976,
          "samples_seen": 190464,
          "online_loss": 0.0002221235563411028,
          "train_loss": 0.00018217200704384595,
          "validation_loss": 0.00013991589366924018,
          "train_metric": 1.834769682318438e-05,
          "validation_metric": 1.4091816410655156e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0683905869991577
        },
        {
          "epoch": 187,
          "updates": 2992,
          "samples_seen": 191488,
          "online_loss": 0.00022061976324039279,
          "train_loss": 0.00022677393280901015,
          "validation_loss": 0.00016838945157360286,
          "train_metric": 2.2839838493382558e-05,
          "validation_metric": 1.6959569620667025e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0787321949992474
        },
        {
          "epoch": 188,
          "updates": 3008,
          "samples_seen": 192512,
          "online_loss": 0.0002112208667313098,
          "train_loss": 0.0001737186685204506,
          "validation_loss": 0.0001423424982931465,
          "train_metric": 1.749630791891832e-05,
          "validation_metric": 1.4336216736410279e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.0891005680005037
        },
        {
          "epoch": 189,
          "updates": 3024,
          "samples_seen": 193536,
          "online_loss": 0.00019560735563572962,
          "train_loss": 0.00023969030007719994,
          "validation_loss": 0.00018801490659825504,
          "train_metric": 2.4140726964105852e-05,
          "validation_metric": 1.893617445603013e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.099247366999407
        },
        {
          "epoch": 190,
          "updates": 3040,
          "samples_seen": 194560,
          "online_loss": 0.00019152779032083345,
          "train_loss": 0.00017086473235394806,
          "validation_loss": 0.00013447242963593453,
          "train_metric": 1.72088693943806e-05,
          "validation_metric": 1.3543572094931733e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1096963090003555
        },
        {
          "epoch": 191,
          "updates": 3056,
          "samples_seen": 195584,
          "online_loss": 0.00018027242867901805,
          "train_loss": 0.00020412685989867896,
          "validation_loss": 0.0001477296755183488,
          "train_metric": 2.0558909454848617e-05,
          "validation_metric": 1.487879126216285e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1201543909992324
        },
        {
          "epoch": 192,
          "updates": 3072,
          "samples_seen": 196608,
          "online_loss": 0.00017462019422964659,
          "train_loss": 0.00017260668391827494,
          "validation_loss": 0.00013375090202316642,
          "train_metric": 1.7384312741342e-05,
          "validation_metric": 1.347090073977597e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1302989489995525
        },
        {
          "epoch": 193,
          "updates": 3088,
          "samples_seen": 197632,
          "online_loss": 0.00017644088075030595,
          "train_loss": 0.00017135590314865112,
          "validation_loss": 0.00013424821372609586,
          "train_metric": 1.7258338630199432e-05,
          "validation_metric": 1.3520989341486711e-05,
          "lr_first": 9.375e-05,
          "lr_last": 9.375e-05,
          "seconds": 2.1405790579992754
        },
        {
          "epoch": 194,
          "updates": 3104,
          "samples_seen": 198656,
          "online_loss": 0.00017693703239274328,
          "train_loss": 0.0001789337256923318,
          "validation_loss": 0.00013429073442239314,
          "train_metric": 1.8021552023128606e-05,
          "validation_metric": 1.3525270333047956e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1508524939999916
        },
        {
          "epoch": 195,
          "updates": 3120,
          "samples_seen": 199680,
          "online_loss": 0.0001921771827255725,
          "train_loss": 0.00016558755305595696,
          "validation_loss": 0.00012955306738149375,
          "train_metric": 1.667737160460092e-05,
          "validation_metric": 1.3048110304225702e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1615393219999532
        },
        {
          "epoch": 196,
          "updates": 3136,
          "samples_seen": 200704,
          "online_loss": 0.00017881759094962035,
          "train_loss": 0.00016593669715803117,
          "validation_loss": 0.00013553421013057232,
          "train_metric": 1.671253630775027e-05,
          "validation_metric": 1.3650509572471492e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1719898930005
        },
        {
          "epoch": 197,
          "updates": 3152,
          "samples_seen": 201728,
          "online_loss": 0.00018966156767419307,
          "train_loss": 0.00016906003293115646,
          "validation_loss": 0.00013356014096643776,
          "train_metric": 1.7027108697220683e-05,
          "validation_metric": 1.3451687664201017e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.1823025869998673
        },
        {
          "epoch": 198,
          "updates": 3168,
          "samples_seen": 202752,
          "online_loss": 0.0001686354275989288,
          "train_loss": 0.00016444384527858347,
          "validation_loss": 0.00013341540761757642,
          "train_metric": 1.6562182281631976e-05,
          "validation_metric": 1.3437111192615703e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.192465349999111
        },
        {
          "epoch": 199,
          "updates": 3184,
          "samples_seen": 203776,
          "online_loss": 0.00016910136810110998,
          "train_loss": 0.0001697281695669517,
          "validation_loss": 0.0001283138699363917,
          "train_metric": 1.709439675323665e-05,
          "validation_metric": 1.2923302165290806e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.203402773000562
        },
        {
          "epoch": 200,
          "updates": 3200,
          "samples_seen": 204800,
          "online_loss": 0.00017533852724227472,
          "train_loss": 0.0001584547571837902,
          "validation_loss": 0.00012432481162250042,
          "train_metric": 1.5958983567543328e-05,
          "validation_metric": 1.252154015674023e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.2145884170004138
        },
        {
          "epoch": 201,
          "updates": 3216,
          "samples_seen": 205824,
          "online_loss": 0.00016808173631943646,
          "train_loss": 0.0001582900295034051,
          "validation_loss": 0.0001278498093597591,
          "train_metric": 1.5942392565193586e-05,
          "validation_metric": 1.28765650515561e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.2250916129996767
        },
        {
          "epoch": 202,
          "updates": 3232,
          "samples_seen": 206848,
          "online_loss": 0.0001724461462799809,
          "train_loss": 0.00015899755817372352,
          "validation_loss": 0.00012512155808508396,
          "train_metric": 1.6013651475077495e-05,
          "validation_metric": 1.2601783964782953e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.235354025000561
        },
        {
          "epoch": 203,
          "updates": 3248,
          "samples_seen": 207872,
          "online_loss": 0.00018300419424122083,
          "train_loss": 0.0001682780566625297,
          "validation_loss": 0.0001409952819813043,
          "train_metric": 1.6948350094025955e-05,
          "validation_metric": 1.4200530131347477e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.308338515998912
        },
        {
          "epoch": 204,
          "updates": 3264,
          "samples_seen": 208896,
          "online_loss": 0.00016299367416650057,
          "train_loss": 0.00016573297034483403,
          "validation_loss": 0.00012830815103370696,
          "train_metric": 1.6692018107278273e-05,
          "validation_metric": 1.2922726455144584e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.319221538999045
        },
        {
          "epoch": 205,
          "updates": 3280,
          "samples_seen": 209920,
          "online_loss": 0.00019057797499044682,
          "train_loss": 0.00015954651462379843,
          "validation_loss": 0.00012303152470849454,
          "train_metric": 1.6068941476987675e-05,
          "validation_metric": 1.2391284144541714e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3303365310002846
        },
        {
          "epoch": 206,
          "updates": 3296,
          "samples_seen": 210944,
          "online_loss": 0.00017596558882360114,
          "train_loss": 0.0001748956274241209,
          "validation_loss": 0.00014836477930657566,
          "train_metric": 1.761484418238979e-05,
          "validation_metric": 1.494275602453854e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3410236250001617
        },
        {
          "epoch": 207,
          "updates": 3312,
          "samples_seen": 211968,
          "online_loss": 0.00016648792143314495,
          "train_loss": 0.0001668957556830719,
          "validation_loss": 0.00012553718988783658,
          "train_metric": 1.6809128283057362e-05,
          "validation_metric": 1.2643646186916158e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3514963750003517
        },
        {
          "epoch": 208,
          "updates": 3328,
          "samples_seen": 212992,
          "online_loss": 0.00016494239071107586,
          "train_loss": 0.0001553035544930026,
          "validation_loss": 0.00012152733688708395,
          "train_metric": 1.5641604477423243e-05,
          "validation_metric": 1.2239789612067398e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.3625693369995133
        },
        {
          "epoch": 209,
          "updates": 3344,
          "samples_seen": 214016,
          "online_loss": 0.00016111069180624327,
          "train_loss": 0.00016266675083898008,
          "validation_loss": 0.00013522146036848426,
          "train_metric": 1.6383201000280678e-05,
          "validation_metric": 1.3619010132970288e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.373145444000329
        },
        {
          "epoch": 210,
          "updates": 3360,
          "samples_seen": 215040,
          "online_loss": 0.00016142015783771058,
          "train_loss": 0.00016027076344471425,
          "validation_loss": 0.00012306190910749137,
          "train_metric": 1.6141882952069864e-05,
          "validation_metric": 1.239434459421318e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.383410836000621
        },
        {
          "epoch": 211,
          "updates": 3376,
          "samples_seen": 216064,
          "online_loss": 0.0001674862496656715,
          "train_loss": 0.00016953803424257785,
          "validation_loss": 0.00013831765681970865,
          "train_metric": 1.7075250070774928e-05,
          "validation_metric": 1.3930848581367172e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.393819663000613
        },
        {
          "epoch": 212,
          "updates": 3392,
          "samples_seen": 217088,
          "online_loss": 0.00017088445247281925,
          "train_loss": 0.00016218898235820234,
          "validation_loss": 0.00013252333155833185,
          "train_metric": 1.6335081454599276e-05,
          "validation_metric": 1.3347263120522257e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4041738449996046
        },
        {
          "epoch": 213,
          "updates": 3408,
          "samples_seen": 218112,
          "online_loss": 0.00016113983565446688,
          "train_loss": 0.00016288187180180103,
          "validation_loss": 0.00012246874393895268,
          "train_metric": 1.640486516407691e-05,
          "validation_metric": 1.2334603525232524e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.414631121000639
        },
        {
          "epoch": 214,
          "updates": 3424,
          "samples_seen": 219136,
          "online_loss": 0.00016957834941422334,
          "train_loss": 0.00016950565623119473,
          "validation_loss": 0.00012820778647437692,
          "train_metric": 1.707198862277437e-05,
          "validation_metric": 1.2912620150018483e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4254105929994694
        },
        {
          "epoch": 215,
          "updates": 3440,
          "samples_seen": 220160,
          "online_loss": 0.00016269777097477345,
          "train_loss": 0.00015231393626891077,
          "validation_loss": 0.00012143931962782517,
          "train_metric": 1.534050352347549e-05,
          "validation_metric": 1.2230922948219813e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4370242099994357
        },
        {
          "epoch": 216,
          "updates": 3456,
          "samples_seen": 221184,
          "online_loss": 0.00015858231790844002,
          "train_loss": 0.0002497771929483861,
          "validation_loss": 0.0002021567488554865,
          "train_metric": 2.5156645278912038e-05,
          "validation_metric": 2.0360484995762818e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.4481144059991493
        },
        {
          "epoch": 217,
          "updates": 3472,
          "samples_seen": 222208,
          "online_loss": 0.00021554091608777526,
          "train_loss": 0.00015777067164890468,
          "validation_loss": 0.0001296417322009802,
          "train_metric": 1.589008570590522e-05,
          "validation_metric": 1.3057041542197112e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.458991554000022
        },
        {
          "epoch": 218,
          "updates": 3488,
          "samples_seen": 223232,
          "online_loss": 0.00015791271243870142,
          "train_loss": 0.00016162989777512848,
          "validation_loss": 0.0001232550130225718,
          "train_metric": 1.6278770999633707e-05,
          "validation_metric": 1.2413795047905296e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.469378918000075
        },
        {
          "epoch": 219,
          "updates": 3504,
          "samples_seen": 224256,
          "online_loss": 0.00015707138891229988,
          "train_loss": 0.00016427742957603186,
          "validation_loss": 0.0001316763664362952,
          "train_metric": 1.65454202942783e-05,
          "validation_metric": 1.3261961612442974e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.479702266999084
        },
        {
          "epoch": 220,
          "updates": 3520,
          "samples_seen": 225280,
          "online_loss": 0.00016491395854245638,
          "train_loss": 0.0001511566515546292,
          "validation_loss": 0.00012219144264236093,
          "train_metric": 1.5223943591990974e-05,
          "validation_metric": 1.2306675671425182e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.490673018999587
        },
        {
          "epoch": 221,
          "updates": 3536,
          "samples_seen": 226304,
          "online_loss": 0.00016016758536352427,
          "train_loss": 0.00015513176913373172,
          "validation_loss": 0.00012186868116259575,
          "train_metric": 1.562430406920612e-05,
          "validation_metric": 1.2274167602299713e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.50106339299964
        },
        {
          "epoch": 222,
          "updates": 3552,
          "samples_seen": 227328,
          "online_loss": 0.0001593631068317336,
          "train_loss": 0.00015553904813714325,
          "validation_loss": 0.00012870585487689823,
          "train_metric": 1.5665322280256078e-05,
          "validation_metric": 1.2962780601810664e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5116445330004353
        },
        {
          "epoch": 223,
          "updates": 3568,
          "samples_seen": 228352,
          "online_loss": 0.00015242234667312005,
          "train_loss": 0.00016039938782341778,
          "validation_loss": 0.00012111449905205518,
          "train_metric": 1.6154837794601917e-05,
          "validation_metric": 1.2198210242786445e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.522899236999365
        },
        {
          "epoch": 224,
          "updates": 3584,
          "samples_seen": 229376,
          "online_loss": 0.00017113809099100763,
          "train_loss": 0.00014897040091454983,
          "validation_loss": 0.00011947266466449946,
          "train_metric": 1.5003754015197046e-05,
          "validation_metric": 1.2032847735099494e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.533901505999893
        },
        {
          "epoch": 225,
          "updates": 3600,
          "samples_seen": 230400,
          "online_loss": 0.00016252553768936195,
          "train_loss": 0.0001570174645166844,
          "validation_loss": 0.0001296235859626904,
          "train_metric": 1.5814222933840938e-05,
          "validation_metric": 1.3055213457846548e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5445434150005894
        },
        {
          "epoch": 226,
          "updates": 3616,
          "samples_seen": 231424,
          "online_loss": 0.00016825611101012328,
          "train_loss": 0.0001507794513599947,
          "validation_loss": 0.00012316122592892498,
          "train_metric": 1.518595490779262e-05,
          "validation_metric": 1.240434812643798e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5550702450000244
        },
        {
          "epoch": 227,
          "updates": 3632,
          "samples_seen": 232448,
          "online_loss": 0.00015442142239407985,
          "train_loss": 0.00015438236005138606,
          "validation_loss": 0.00012114206765545532,
          "train_metric": 1.5548825103905983e-05,
          "validation_metric": 1.2200986020616256e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.566520873999252
        },
        {
          "epoch": 228,
          "updates": 3648,
          "samples_seen": 233472,
          "online_loss": 0.00016305746430589352,
          "train_loss": 0.0001537699718028307,
          "validation_loss": 0.00012110365787521005,
          "train_metric": 1.548714681121055e-05,
          "validation_metric": 1.219711521116551e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5769429270003457
        },
        {
          "epoch": 229,
          "updates": 3664,
          "samples_seen": 234496,
          "online_loss": 0.00015594661635986995,
          "train_loss": 0.00015167059609666467,
          "validation_loss": 0.00012721589882858098,
          "train_metric": 1.527570748294238e-05,
          "validation_metric": 1.281271852349164e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.58727714700035
        },
        {
          "epoch": 230,
          "updates": 3680,
          "samples_seen": 235520,
          "online_loss": 0.00015730635050204,
          "train_loss": 0.000153502551256679,
          "validation_loss": 0.00012577304732985795,
          "train_metric": 1.546021485410165e-05,
          "validation_metric": 1.2667399460042361e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.5974550689988973
        },
        {
          "epoch": 231,
          "updates": 3696,
          "samples_seen": 236544,
          "online_loss": 0.00016225741001107963,
          "train_loss": 0.0001548809086671099,
          "validation_loss": 0.00012994848657399416,
          "train_metric": 1.5599036487401463e-05,
          "validation_metric": 1.3087936167721637e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.607926998000039
        },
        {
          "epoch": 232,
          "updates": 3712,
          "samples_seen": 237568,
          "online_loss": 0.00015986752578101004,
          "train_loss": 0.00016795439296402037,
          "validation_loss": 0.000126347309560515,
          "train_metric": 1.691575198492501e-05,
          "validation_metric": 1.272523877560161e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.618162447999566
        },
        {
          "epoch": 233,
          "updates": 3728,
          "samples_seen": 238592,
          "online_loss": 0.00016344196683348855,
          "train_loss": 0.00015008471382316202,
          "validation_loss": 0.00011694519344018772,
          "train_metric": 1.511598293291172e-05,
          "validation_metric": 1.1778291082009673e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.628930863998903
        },
        {
          "epoch": 234,
          "updates": 3744,
          "samples_seen": 239616,
          "online_loss": 0.00015242089557432337,
          "train_loss": 0.00015679239004384726,
          "validation_loss": 0.0001325382909271866,
          "train_metric": 1.5791556506883353e-05,
          "validation_metric": 1.3348771062737796e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6394298969989904
        },
        {
          "epoch": 235,
          "updates": 3760,
          "samples_seen": 240640,
          "online_loss": 0.00014189978128342773,
          "train_loss": 0.000204338546609506,
          "validation_loss": 0.0001438927138224244,
          "train_metric": 2.0580229829647578e-05,
          "validation_metric": 1.4492345144390129e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6497043359995587
        },
        {
          "epoch": 236,
          "updates": 3776,
          "samples_seen": 241664,
          "online_loss": 0.00021200962828515912,
          "train_loss": 0.00017812332953326404,
          "validation_loss": 0.0001433926518075168,
          "train_metric": 1.7939930330612697e-05,
          "validation_metric": 1.4441980965784751e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6601018130004377
        },
        {
          "epoch": 237,
          "updates": 3792,
          "samples_seen": 242688,
          "online_loss": 0.00019249887509431574,
          "train_loss": 0.0001446761016268283,
          "validation_loss": 0.00012072749086655676,
          "train_metric": 1.4571245628758334e-05,
          "validation_metric": 1.215923111885786e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.67031990400028
        },
        {
          "epoch": 238,
          "updates": 3808,
          "samples_seen": 243712,
          "online_loss": 0.00017839595921032014,
          "train_loss": 0.00015473015082534403,
          "validation_loss": 0.00012707163114100695,
          "train_metric": 1.5583853382850066e-05,
          "validation_metric": 1.2798189345630817e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6806376589993306
        },
        {
          "epoch": 239,
          "updates": 3824,
          "samples_seen": 244736,
          "online_loss": 0.0001526986850421963,
          "train_loss": 0.00015846433234401047,
          "validation_loss": 0.0001317514688707888,
          "train_metric": 1.5959945812937804e-05,
          "validation_metric": 1.3269523151393514e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.6911437309991015
        },
        {
          "epoch": 240,
          "updates": 3840,
          "samples_seen": 245760,
          "online_loss": 0.00015213426695481758,
          "train_loss": 0.00019425329810474068,
          "validation_loss": 0.0001424271467840299,
          "train_metric": 1.956447886186652e-05,
          "validation_metric": 1.4344740520755295e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.7015701649997936
        },
        {
          "epoch": 241,
          "updates": 3856,
          "samples_seen": 246784,
          "online_loss": 0.0001904490059132513,
          "train_loss": 0.0001693001395324245,
          "validation_loss": 0.0001324386103078723,
          "train_metric": 1.7051290342351422e-05,
          "validation_metric": 1.3338731150724925e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.712478215000374
        },
        {
          "epoch": 242,
          "updates": 3872,
          "samples_seen": 247808,
          "online_loss": 0.00015413447454193374,
          "train_loss": 0.0001492796145612374,
          "validation_loss": 0.0001263437297893688,
          "train_metric": 1.5034896932775155e-05,
          "validation_metric": 1.2724877706205007e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.723468017999039
        },
        {
          "epoch": 243,
          "updates": 3888,
          "samples_seen": 248832,
          "online_loss": 0.00015034742318675853,
          "train_loss": 0.00016077265900094062,
          "validation_loss": 0.0001245692983502522,
          "train_metric": 1.6192432667594403e-05,
          "validation_metric": 1.2546162906801328e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.733889003000513
        },
        {
          "epoch": 244,
          "updates": 3904,
          "samples_seen": 249856,
          "online_loss": 0.000170985103068233,
          "train_loss": 0.00016190570022445172,
          "validation_loss": 0.00012581454939208925,
          "train_metric": 1.6306546967825852e-05,
          "validation_metric": 1.267157949769171e-05,
          "lr_first": 4.6875e-05,
          "lr_last": 4.6875e-05,
          "seconds": 2.744924954000453
        },
        {
          "epoch": 245,
          "updates": 3920,
          "samples_seen": 250880,
          "online_loss": 0.0001466872513447015,
          "train_loss": 0.0001684899762040004,
          "validation_loss": 0.00014720993931405246,
          "train_metric": 1.6969692296697758e-05,
          "validation_metric": 1.482644620409701e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.7556389029996353
        },
        {
          "epoch": 246,
          "updates": 3936,
          "samples_seen": 251904,
          "online_loss": 0.00015450056071131257,
          "train_loss": 0.00015116909344214946,
          "validation_loss": 0.00011930397158721462,
          "train_metric": 1.522519869467942e-05,
          "validation_metric": 1.201586019305978e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.7660431009990134
        },
        {
          "epoch": 247,
          "updates": 3952,
          "samples_seen": 252928,
          "online_loss": 0.00016087346102722222,
          "train_loss": 0.00014754546282347292,
          "validation_loss": 0.00012218451593071222,
          "train_metric": 1.4860237570246682e-05,
          "validation_metric": 1.230597717949422e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.7765562669992505
        },
        {
          "epoch": 248,
          "updates": 3968,
          "samples_seen": 253952,
          "online_loss": 0.00014649552758783102,
          "train_loss": 0.00014015246415510774,
          "validation_loss": 0.00011420786177041009,
          "train_metric": 1.41156415338628e-05,
          "validation_metric": 1.1502595953061245e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.7876383469993016
        },
        {
          "epoch": 249,
          "updates": 3984,
          "samples_seen": 254976,
          "online_loss": 0.00014373373096532305,
          "train_loss": 0.0001395288563799113,
          "validation_loss": 0.0001141421525971964,
          "train_metric": 1.4052837286726572e-05,
          "validation_metric": 1.1495980288600549e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.7985513000003266
        },
        {
          "epoch": 250,
          "updates": 4000,
          "samples_seen": 256000,
          "online_loss": 0.00014503094780593528,
          "train_loss": 0.00014227893552742898,
          "validation_loss": 0.00011934055510209873,
          "train_metric": 1.4329812984215096e-05,
          "validation_metric": 1.201954364660196e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.8091533269998763
        },
        {
          "epoch": 251,
          "updates": 4016,
          "samples_seen": 257024,
          "online_loss": 0.00014491336423816392,
          "train_loss": 0.00013958195631857961,
          "validation_loss": 0.00011298169556539506,
          "train_metric": 1.4058184206078295e-05,
          "validation_metric": 1.1379102033970412e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.8201549109999178
        },
        {
          "epoch": 252,
          "updates": 4032,
          "samples_seen": 258048,
          "online_loss": 0.0001502725654063397,
          "train_loss": 0.00014126730093266815,
          "validation_loss": 0.00011326211097184569,
          "train_metric": 1.4227925930754282e-05,
          "validation_metric": 1.1407344572944567e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.830570952999551
        },
        {
          "epoch": 253,
          "updates": 4048,
          "samples_seen": 259072,
          "online_loss": 0.00014658597865491174,
          "train_loss": 0.0001432389981346205,
          "validation_loss": 0.00012094565317966044,
          "train_metric": 1.4426509551412892e-05,
          "validation_metric": 1.2181202691863291e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.8410455729990645
        },
        {
          "epoch": 254,
          "updates": 4064,
          "samples_seen": 260096,
          "online_loss": 0.0001447401227778755,
          "train_loss": 0.00013873544230591506,
          "validation_loss": 0.00011294475552858785,
          "train_metric": 1.3972925444249995e-05,
          "validation_metric": 1.137538220064016e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.851855896999041
        },
        {
          "epoch": 255,
          "updates": 4080,
          "samples_seen": 261120,
          "online_loss": 0.0001425223367732542,
          "train_loss": 0.00013828629744239151,
          "validation_loss": 0.00011400601943023503,
          "train_metric": 1.3927689906267915e-05,
          "validation_metric": 1.1482267836981919e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.862709572998938
        },
        {
          "epoch": 256,
          "updates": 4096,
          "samples_seen": 262144,
          "online_loss": 0.00014121801314104232,
          "train_loss": 0.00013846918591298163,
          "validation_loss": 0.00011529693438205868,
          "train_metric": 1.3946109902462922e-05,
          "validation_metric": 1.1612284652073868e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.8733010819996707
        },
        {
          "epoch": 257,
          "updates": 4112,
          "samples_seen": 263168,
          "online_loss": 0.00014411150596060907,
          "train_loss": 0.00015849996998440474,
          "validation_loss": 0.00012016612890874967,
          "train_metric": 1.5963538317009807e-05,
          "validation_metric": 1.210269419971155e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.8838918910005304
        },
        {
          "epoch": 258,
          "updates": 4128,
          "samples_seen": 264192,
          "online_loss": 0.0001501673004895565,
          "train_loss": 0.0001407518284395337,
          "validation_loss": 0.00011782220099121332,
          "train_metric": 1.4176009244692978e-05,
          "validation_metric": 1.1866620297951158e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.89485702499951
        },
        {
          "epoch": 259,
          "updates": 4144,
          "samples_seen": 265216,
          "online_loss": 0.00014034293599252123,
          "train_loss": 0.00013742971350438893,
          "validation_loss": 0.00011333494330756366,
          "train_metric": 1.3841417057847138e-05,
          "validation_metric": 1.1414679647714365e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9055522630005726
        },
        {
          "epoch": 260,
          "updates": 4160,
          "samples_seen": 266240,
          "online_loss": 0.00015493788259846042,
          "train_loss": 0.0001535358896944672,
          "validation_loss": 0.00012973256525583565,
          "train_metric": 1.5463572708540596e-05,
          "validation_metric": 1.3066190149402246e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.915899695999542
        },
        {
          "epoch": 261,
          "updates": 4176,
          "samples_seen": 267264,
          "online_loss": 0.0001448048747079156,
          "train_loss": 0.0001419407781213522,
          "validation_loss": 0.00011226838250877336,
          "train_metric": 1.42957524076337e-05,
          "validation_metric": 1.1307261047477368e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.927012702999491
        },
        {
          "epoch": 262,
          "updates": 4192,
          "samples_seen": 268288,
          "online_loss": 0.00014164484173306846,
          "train_loss": 0.00013752240920439363,
          "validation_loss": 0.00011323046055622399,
          "train_metric": 1.385075393045554e-05,
          "validation_metric": 1.1404158613004256e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9375950590001594
        },
        {
          "epoch": 263,
          "updates": 4208,
          "samples_seen": 269312,
          "online_loss": 0.00013905251512369432,
          "train_loss": 0.00013651896733790636,
          "validation_loss": 0.00011141415598103777,
          "train_metric": 1.374968996969983e-05,
          "validation_metric": 1.1221225577173755e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9485676229996898
        },
        {
          "epoch": 264,
          "updates": 4224,
          "samples_seen": 270336,
          "online_loss": 0.00014614360725317965,
          "train_loss": 0.00014413002645596862,
          "validation_loss": 0.00012424336455296725,
          "train_metric": 1.4516250303131528e-05,
          "validation_metric": 1.2513336514530238e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9591106269999727
        },
        {
          "epoch": 265,
          "updates": 4240,
          "samples_seen": 271360,
          "online_loss": 0.00015116511212909245,
          "train_loss": 0.00013674386718776077,
          "validation_loss": 0.000111979796201922,
          "train_metric": 1.3772341844742186e-05,
          "validation_metric": 1.1278194506303407e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9694776039996214
        },
        {
          "epoch": 266,
          "updates": 4256,
          "samples_seen": 272384,
          "online_loss": 0.00014507528976537287,
          "train_loss": 0.00013965215475764126,
          "validation_loss": 0.00011146543693030253,
          "train_metric": 1.4065254617889877e-05,
          "validation_metric": 1.1226390597585123e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.9799969420000707
        },
        {
          "epoch": 267,
          "updates": 4272,
          "samples_seen": 273408,
          "online_loss": 0.00014230932902137283,
          "train_loss": 0.0001358379377052188,
          "validation_loss": 0.00011232019460294396,
          "train_metric": 1.3681099517270923e-05,
          "validation_metric": 1.131247881858144e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 2.990357275999486
        },
        {
          "epoch": 268,
          "updates": 4288,
          "samples_seen": 274432,
          "online_loss": 0.0001496241957283928,
          "train_loss": 0.00013999527436681092,
          "validation_loss": 0.00011645749327726662,
          "train_metric": 1.4099811778578442e-05,
          "validation_metric": 1.1729172001651023e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0008060569998634
        },
        {
          "epoch": 269,
          "updates": 4304,
          "samples_seen": 275456,
          "online_loss": 0.00014200990244717104,
          "train_loss": 0.00013517004845198244,
          "validation_loss": 0.00011139468551846221,
          "train_metric": 1.3613830560643692e-05,
          "validation_metric": 1.121926379710203e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.011927014000321
        },
        {
          "epoch": 270,
          "updates": 4320,
          "samples_seen": 276480,
          "online_loss": 0.000144557284329494,
          "train_loss": 0.00013767494237981737,
          "validation_loss": 0.00011204244219698012,
          "train_metric": 1.3866116205463186e-05,
          "validation_metric": 1.1284504580544308e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.022411979000026
        },
        {
          "epoch": 271,
          "updates": 4336,
          "samples_seen": 277504,
          "online_loss": 0.00014848519208499056,
          "train_loss": 0.0001351681275991723,
          "validation_loss": 0.00011178967542946339,
          "train_metric": 1.361363956675632e-05,
          "validation_metric": 1.1259046914346982e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0329393390002224
        },
        {
          "epoch": 272,
          "updates": 4352,
          "samples_seen": 278528,
          "online_loss": 0.00014940693336029653,
          "train_loss": 0.00014775630552321672,
          "validation_loss": 0.00011815636389655992,
          "train_metric": 1.488147427153308e-05,
          "validation_metric": 1.1900277058884967e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0437369549999858
        },
        {
          "epoch": 273,
          "updates": 4368,
          "samples_seen": 279552,
          "online_loss": 0.00013915978433942655,
          "train_loss": 0.00014038632798474282,
          "validation_loss": 0.0001123807014664635,
          "train_metric": 1.4139196537144016e-05,
          "validation_metric": 1.1318572433083318e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.054221123999014
        },
        {
          "epoch": 274,
          "updates": 4384,
          "samples_seen": 280576,
          "online_loss": 0.00016552372244404978,
          "train_loss": 0.00013461442722473294,
          "validation_loss": 0.000110276771010831,
          "train_metric": 1.3557871170633007e-05,
          "validation_metric": 1.110667290049605e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.065289431000565
        },
        {
          "epoch": 275,
          "updates": 4400,
          "samples_seen": 281600,
          "online_loss": 0.00016087059066194342,
          "train_loss": 0.0001520407386124134,
          "validation_loss": 0.0001323819742538035,
          "train_metric": 1.531298585177865e-05,
          "validation_metric": 1.333302952843951e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0759220069994626
        },
        {
          "epoch": 276,
          "updates": 4416,
          "samples_seen": 282624,
          "online_loss": 0.00015380838885903358,
          "train_loss": 0.000138647184940055,
          "validation_loss": 0.00011065047874581069,
          "train_metric": 1.3964036043034866e-05,
          "validation_metric": 1.1144309610244818e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0864348779996362
        },
        {
          "epoch": 277,
          "updates": 4432,
          "samples_seen": 283648,
          "online_loss": 0.0001384719093948661,
          "train_loss": 0.0001342502946499735,
          "validation_loss": 0.0001106822892324999,
          "train_metric": 1.3521198525268119e-05,
          "validation_metric": 1.114751648856327e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.0969255109994265
        },
        {
          "epoch": 278,
          "updates": 4448,
          "samples_seen": 284672,
          "online_loss": 0.00013808483299726504,
          "train_loss": 0.00013511121505871415,
          "validation_loss": 0.00011524152068886906,
          "train_metric": 1.3607907931145746e-05,
          "validation_metric": 1.1606702173594385e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.107955891000529
        },
        {
          "epoch": 279,
          "updates": 4464,
          "samples_seen": 285696,
          "online_loss": 0.0001392785975440347,
          "train_loss": 0.00013361316814552993,
          "validation_loss": 0.00011379528586985543,
          "train_metric": 1.3457028217089828e-05,
          "validation_metric": 1.1461044778116047e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.118413240999871
        },
        {
          "epoch": 280,
          "updates": 4480,
          "samples_seen": 286720,
          "online_loss": 0.00013598259783975664,
          "train_loss": 0.00013339892029762268,
          "validation_loss": 0.00011007908324245363,
          "train_metric": 1.3435451364784967e-05,
          "validation_metric": 1.1086761332990136e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.129325911999331
        },
        {
          "epoch": 281,
          "updates": 4496,
          "samples_seen": 287744,
          "online_loss": 0.00013777102549283882,
          "train_loss": 0.0001325343910139054,
          "validation_loss": 0.00010915342136286199,
          "train_metric": 1.3348377251531929e-05,
          "validation_metric": 1.0993531759595498e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1407059809989732
        },
        {
          "epoch": 282,
          "updates": 4512,
          "samples_seen": 288768,
          "online_loss": 0.00013483711609296734,
          "train_loss": 0.00014883711992297322,
          "validation_loss": 0.00011756228923331946,
          "train_metric": 1.4990329873398878e-05,
          "validation_metric": 1.1840444130939431e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.151262362000125
        },
        {
          "epoch": 283,
          "updates": 4528,
          "samples_seen": 289792,
          "online_loss": 0.00014001809404362575,
          "train_loss": 0.0001522196689620614,
          "validation_loss": 0.00013210962060838938,
          "train_metric": 1.533100839878898e-05,
          "validation_metric": 1.3305597349244636e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1618196700001135
        },
        {
          "epoch": 284,
          "updates": 4544,
          "samples_seen": 290816,
          "online_loss": 0.0001534687953608227,
          "train_loss": 0.00013552197196986526,
          "validation_loss": 0.0001082973467418924,
          "train_metric": 1.3649276297655888e-05,
          "validation_metric": 1.0907313480856828e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1729916450003657
        },
        {
          "epoch": 285,
          "updates": 4560,
          "samples_seen": 291840,
          "online_loss": 0.00014417863758353633,
          "train_loss": 0.00015054599498398602,
          "validation_loss": 0.00013413565466180444,
          "train_metric": 1.5162443560257088e-05,
          "validation_metric": 1.3509653399523813e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1839548829993873
        },
        {
          "epoch": 286,
          "updates": 4576,
          "samples_seen": 292864,
          "online_loss": 0.00014371337601915002,
          "train_loss": 0.00013245172158349305,
          "validation_loss": 0.00010807799117174,
          "train_metric": 1.3340052646526601e-05,
          "validation_metric": 1.088522003556136e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.1954823579999356
        },
        {
          "epoch": 287,
          "updates": 4592,
          "samples_seen": 293888,
          "online_loss": 0.00013490851961250883,
          "train_loss": 0.00013322519953362644,
          "validation_loss": 0.00011118889960926026,
          "train_metric": 1.3417952686722856e-05,
          "validation_metric": 1.119853914133273e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.206245501000012
        },
        {
          "epoch": 288,
          "updates": 4608,
          "samples_seen": 294912,
          "online_loss": 0.00013622770029542153,
          "train_loss": 0.00013350480003282428,
          "validation_loss": 0.00011415423796279356,
          "train_metric": 1.3446114280668553e-05,
          "validation_metric": 1.1497197192511521e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.21657953400063
        },
        {
          "epoch": 289,
          "updates": 4624,
          "samples_seen": 295936,
          "online_loss": 0.00013980024368720478,
          "train_loss": 0.00014374463353306055,
          "validation_loss": 0.00011273646668996662,
          "train_metric": 1.447743306925986e-05,
          "validation_metric": 1.1354403795849066e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2272074989996327
        },
        {
          "epoch": 290,
          "updates": 4640,
          "samples_seen": 296960,
          "online_loss": 0.0001364125159852847,
          "train_loss": 0.0001349685771856457,
          "validation_loss": 0.00011545138113433495,
          "train_metric": 1.3593539733847138e-05,
          "validation_metric": 1.162783973995829e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.237568881999323
        },
        {
          "epoch": 291,
          "updates": 4656,
          "samples_seen": 297984,
          "online_loss": 0.00013749453819400514,
          "train_loss": 0.00013380381278693676,
          "validation_loss": 0.00010906182433245704,
          "train_metric": 1.347623037872836e-05,
          "validation_metric": 1.0984307664330117e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.247969263999039
        },
        {
          "epoch": 292,
          "updates": 4672,
          "samples_seen": 299008,
          "online_loss": 0.00014343515431392007,
          "train_loss": 0.00013246522576082498,
          "validation_loss": 0.00010960662621073425,
          "train_metric": 1.3341412341105752e-05,
          "validation_metric": 1.103917838918278e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.258504761999575
        },
        {
          "epoch": 293,
          "updates": 4688,
          "samples_seen": 300032,
          "online_loss": 0.0001392150088577182,
          "train_loss": 0.00013193566701374948,
          "validation_loss": 0.00010929920244961977,
          "train_metric": 1.3288076843309682e-05,
          "validation_metric": 1.1008215551555622e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.26923028300007
        },
        {
          "epoch": 294,
          "updates": 4704,
          "samples_seen": 301056,
          "online_loss": 0.0001479596717217646,
          "train_loss": 0.0001333853870164603,
          "validation_loss": 0.00010993811883963645,
          "train_metric": 1.3434088941721711e-05,
          "validation_metric": 1.1072565030190162e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.280161278999003
        },
        {
          "epoch": 295,
          "updates": 4720,
          "samples_seen": 302080,
          "online_loss": 0.00013581941266238573,
          "train_loss": 0.00013146169658284634,
          "validation_loss": 0.00010771235974971205,
          "train_metric": 1.324034201388713e-05,
          "validation_metric": 1.0848394595086575e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.2919523809996463
        },
        {
          "epoch": 296,
          "updates": 4736,
          "samples_seen": 303104,
          "online_loss": 0.0001423427138433908,
          "train_loss": 0.00013131849118508399,
          "validation_loss": 0.0001124943228205666,
          "train_metric": 1.322591651842231e-05,
          "validation_metric": 1.1330015695421025e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3026473600002646
        },
        {
          "epoch": 297,
          "updates": 4752,
          "samples_seen": 304128,
          "online_loss": 0.00015040443486213917,
          "train_loss": 0.00013858504826202989,
          "validation_loss": 0.00011799666390288621,
          "train_metric": 1.3957777809991967e-05,
          "validation_metric": 1.1884192645084113e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.313332163999803
        },
        {
          "epoch": 298,
          "updates": 4768,
          "samples_seen": 305152,
          "online_loss": 0.00013962771799924667,
          "train_loss": 0.0001323575124843046,
          "validation_loss": 0.00011027875007130206,
          "train_metric": 1.3330563888303004e-05,
          "validation_metric": 1.1106871170341037e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3239619840005616
        },
        {
          "epoch": 299,
          "updates": 4784,
          "samples_seen": 306176,
          "online_loss": 0.00013680866254617285,
          "train_loss": 0.00015203993825707585,
          "validation_loss": 0.00011811821605078876,
          "train_metric": 1.5312907635234296e-05,
          "validation_metric": 1.1896433534275275e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.334684303999893
        },
        {
          "epoch": 300,
          "updates": 4800,
          "samples_seen": 307200,
          "online_loss": 0.00017408837493348983,
          "train_loss": 0.0001554517657496035,
          "validation_loss": 0.00014009510050527751,
          "train_metric": 1.5656532923458144e-05,
          "validation_metric": 1.410986533301184e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3453704189996643
        },
        {
          "epoch": 301,
          "updates": 4816,
          "samples_seen": 308224,
          "online_loss": 0.00014712875554323546,
          "train_loss": 0.00013466521340887994,
          "validation_loss": 0.00011628001084318385,
          "train_metric": 1.3562987987825181e-05,
          "validation_metric": 1.171129770227708e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.3564835010001843
        },
        {
          "epoch": 302,
          "updates": 4832,
          "samples_seen": 309248,
          "online_loss": 0.00013461981234286213,
          "train_loss": 0.00012986778165213764,
          "validation_loss": 0.00010654962534317747,
          "train_metric": 1.307980710407719e-05,
          "validation_metric": 1.0731288966780994e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.367494598000121
        },
        {
          "epoch": 303,
          "updates": 4848,
          "samples_seen": 310272,
          "online_loss": 0.00013460632362694014,
          "train_loss": 0.00013546613627113402,
          "validation_loss": 0.00011725619697244838,
          "train_metric": 1.3643651072925422e-05,
          "validation_metric": 1.1809614989033435e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.378003276000527
        },
        {
          "epoch": 304,
          "updates": 4864,
          "samples_seen": 311296,
          "online_loss": 0.00013729173952015117,
          "train_loss": 0.00012853529187850654,
          "validation_loss": 0.00010732468217611313,
          "train_metric": 1.2945604794367682e-05,
          "validation_metric": 1.080934816855006e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.388402380000116
        },
        {
          "epoch": 305,
          "updates": 4880,
          "samples_seen": 312320,
          "online_loss": 0.00013610102450911654,
          "train_loss": 0.0001293627719860524,
          "validation_loss": 0.00010667955939425156,
          "train_metric": 1.3028944522375241e-05,
          "validation_metric": 1.0744374776550103e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.399019103999308
        },
        {
          "epoch": 306,
          "updates": 4896,
          "samples_seen": 313344,
          "online_loss": 0.0001437635605725518,
          "train_loss": 0.00015461626753676683,
          "validation_loss": 0.00013538904022425413,
          "train_metric": 1.5572382835671306e-05,
          "validation_metric": 1.3635890354635194e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.409762678000334
        },
        {
          "epoch": 307,
          "updates": 4912,
          "samples_seen": 314368,
          "online_loss": 0.00014781376739847474,
          "train_loss": 0.00012820462870877236,
          "validation_loss": 0.00010573989857221022,
          "train_metric": 1.2912300917378161e-05,
          "validation_metric": 1.0649735486367717e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4207749479992344
        },
        {
          "epoch": 308,
          "updates": 4928,
          "samples_seen": 315392,
          "online_loss": 0.00013816011687595164,
          "train_loss": 0.00012754365161526948,
          "validation_loss": 0.00010567060235189274,
          "train_metric": 1.2845731362176593e-05,
          "validation_metric": 1.0642756024026312e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4319418590002897
        },
        {
          "epoch": 309,
          "updates": 4944,
          "samples_seen": 316416,
          "online_loss": 0.0001292741494580696,
          "train_loss": 0.00012766889994964004,
          "validation_loss": 0.00010804507473949343,
          "train_metric": 1.2858345144195482e-05,
          "validation_metric": 1.0881904927373398e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4426200300003984
        },
        {
          "epoch": 310,
          "updates": 4960,
          "samples_seen": 317440,
          "online_loss": 0.000133040558466746,
          "train_loss": 0.00014282957999967039,
          "validation_loss": 0.00011317364987917244,
          "train_metric": 1.4385273971129209e-05,
          "validation_metric": 1.1398435162845999e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.45311492700057
        },
        {
          "epoch": 311,
          "updates": 4976,
          "samples_seen": 318464,
          "online_loss": 0.00014722616151630064,
          "train_loss": 0.00012734014308080077,
          "validation_loss": 0.00010695037053665146,
          "train_metric": 1.2825234080082737e-05,
          "validation_metric": 1.0771649613161571e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4637626299991098
        },
        {
          "epoch": 312,
          "updates": 4992,
          "samples_seen": 319488,
          "online_loss": 0.00013767142991127912,
          "train_loss": 0.00016868933744262904,
          "validation_loss": 0.00012767090811394155,
          "train_metric": 1.6989770301734097e-05,
          "validation_metric": 1.2858547961513977e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4741130160000466
        },
        {
          "epoch": 313,
          "updates": 5008,
          "samples_seen": 320512,
          "online_loss": 0.00013905380365031306,
          "train_loss": 0.00016082561342045665,
          "validation_loss": 0.00013174847117625177,
          "train_metric": 1.61977659445256e-05,
          "validation_metric": 1.3269223018141929e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4846495959991444
        },
        {
          "epoch": 314,
          "updates": 5024,
          "samples_seen": 321536,
          "online_loss": 0.0001572163250784797,
          "train_loss": 0.0001573003246448934,
          "validation_loss": 0.00012458661512937397,
          "train_metric": 1.584271012688987e-05,
          "validation_metric": 1.254790822713403e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.4952988120003283
        },
        {
          "epoch": 315,
          "updates": 5040,
          "samples_seen": 322560,
          "online_loss": 0.00013259726893011248,
          "train_loss": 0.00013440959446597844,
          "validation_loss": 0.00011265988723607734,
          "train_metric": 1.3537242011807393e-05,
          "validation_metric": 1.1346691280778032e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5057198680005968
        },
        {
          "epoch": 316,
          "updates": 5056,
          "samples_seen": 323584,
          "online_loss": 0.00014938825142962742,
          "train_loss": 0.00015855691162869334,
          "validation_loss": 0.00014133850345388055,
          "train_metric": 1.5969271771609783e-05,
          "validation_metric": 1.4235094567993656e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5161199399990437
        },
        {
          "epoch": 317,
          "updates": 5072,
          "samples_seen": 324608,
          "online_loss": 0.0001438673448319605,
          "train_loss": 0.00012625212548300624,
          "validation_loss": 0.00010445801308378577,
          "train_metric": 1.2715651791950222e-05,
          "validation_metric": 1.0520629075472243e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5275685569995403
        },
        {
          "epoch": 318,
          "updates": 5088,
          "samples_seen": 325632,
          "online_loss": 0.00013309009182194131,
          "train_loss": 0.00012689396680798382,
          "validation_loss": 0.00010629033931763843,
          "train_metric": 1.2780295037373435e-05,
          "validation_metric": 1.0705173735914286e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5383123499996145
        },
        {
          "epoch": 319,
          "updates": 5104,
          "samples_seen": 326656,
          "online_loss": 0.00013148992752576305,
          "train_loss": 0.00013118790229782462,
          "validation_loss": 0.00011630857625277713,
          "train_metric": 1.3212765225034673e-05,
          "validation_metric": 1.1714174434018787e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5490824329990573
        },
        {
          "epoch": 320,
          "updates": 5120,
          "samples_seen": 327680,
          "online_loss": 0.00012675005973505904,
          "train_loss": 0.0001251688227057457,
          "validation_loss": 0.00010491505963727832,
          "train_metric": 1.2606546079041436e-05,
          "validation_metric": 1.0566660421318375e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5599663150005654
        },
        {
          "epoch": 321,
          "updates": 5136,
          "samples_seen": 328704,
          "online_loss": 0.0001304715319747629,
          "train_loss": 0.00013503542868420482,
          "validation_loss": 0.00011890113819390535,
          "train_metric": 1.3600272723124363e-05,
          "validation_metric": 1.197528763441369e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5703274789993884
        },
        {
          "epoch": 322,
          "updates": 5152,
          "samples_seen": 329728,
          "online_loss": 0.00016161302119144239,
          "train_loss": 0.00013302997103892267,
          "validation_loss": 0.00011432532483013347,
          "train_metric": 1.3398291230259929e-05,
          "validation_metric": 1.151442756963661e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.5806726020000497
        },
        {
          "epoch": 323,
          "updates": 5168,
          "samples_seen": 330752,
          "online_loss": 0.0001559042711960501,
          "train_loss": 0.00012416653044056147,
          "validation_loss": 0.00010542784002609551,
          "train_metric": 1.2505597624112852e-05,
          "validation_metric": 1.0618306987453252e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.591111371999432
        },
        {
          "epoch": 324,
          "updates": 5184,
          "samples_seen": 331776,
          "online_loss": 0.00013259826801004237,
          "train_loss": 0.00012376398080959916,
          "validation_loss": 0.00010517526243347675,
          "train_metric": 1.2465054169297218e-05,
          "validation_metric": 1.0592868420644663e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6017420070002117
        },
        {
          "epoch": 325,
          "updates": 5200,
          "samples_seen": 332800,
          "online_loss": 0.00013328213617569418,
          "train_loss": 0.00012636510655283928,
          "validation_loss": 0.00010390681563876569,
          "train_metric": 1.2727030480164103e-05,
          "validation_metric": 1.0465113518876024e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.612769482000658
        },
        {
          "epoch": 326,
          "updates": 5216,
          "samples_seen": 333824,
          "online_loss": 0.00012694453107542358,
          "train_loss": 0.00012408499605953693,
          "validation_loss": 0.00010405022476334125,
          "train_metric": 1.2497384886955842e-05,
          "validation_metric": 1.0479558113729581e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.623952026999177
        },
        {
          "epoch": 327,
          "updates": 5232,
          "samples_seen": 334848,
          "online_loss": 0.00013363398238652735,
          "train_loss": 0.00013272430805955082,
          "validation_loss": 0.00010726852633524686,
          "train_metric": 1.3367506653594319e-05,
          "validation_metric": 1.0803691111505032e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6342714479997085
        },
        {
          "epoch": 328,
          "updates": 5248,
          "samples_seen": 335872,
          "online_loss": 0.00013394636062002974,
          "train_loss": 0.00012686391710303724,
          "validation_loss": 0.00011061879195040092,
          "train_metric": 1.2777269148500636e-05,
          "validation_metric": 1.1141119102830999e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6449522519997117
        },
        {
          "epoch": 329,
          "updates": 5264,
          "samples_seen": 336896,
          "online_loss": 0.0001265835498998058,
          "train_loss": 0.00012216928007546812,
          "validation_loss": 0.00010329276119591668,
          "train_metric": 1.2304441952437628e-05,
          "validation_metric": 1.0403268788650166e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6562522189997253
        },
        {
          "epoch": 330,
          "updates": 5280,
          "samples_seen": 337920,
          "online_loss": 0.00012859307798862574,
          "train_loss": 0.0001340635062661022,
          "validation_loss": 0.00011818881466751918,
          "train_metric": 1.3502384717867244e-05,
          "validation_metric": 1.1903544873348437e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6676216229989222
        },
        {
          "epoch": 331,
          "updates": 5296,
          "samples_seen": 338944,
          "online_loss": 0.00013059434786555357,
          "train_loss": 0.00012271448213141412,
          "validation_loss": 0.00010599750385154039,
          "train_metric": 1.2359352695057169e-05,
          "validation_metric": 1.0675679732230492e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.6788041870004236
        },
        {
          "epoch": 332,
          "updates": 5312,
          "samples_seen": 339968,
          "online_loss": 0.00012966050644536153,
          "train_loss": 0.00012251320003997535,
          "validation_loss": 0.00010826758807525039,
          "train_metric": 1.233908005815465e-05,
          "validation_metric": 1.0904315786319785e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.689725451000413
        },
        {
          "epoch": 333,
          "updates": 5328,
          "samples_seen": 340992,
          "online_loss": 0.00012584340811372385,
          "train_loss": 0.00012145077926106751,
          "validation_loss": 0.00010240143456030637,
          "train_metric": 1.2232077096996363e-05,
          "validation_metric": 1.0313496204616968e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7006840230005764
        },
        {
          "epoch": 334,
          "updates": 5344,
          "samples_seen": 342016,
          "online_loss": 0.00012961133415956283,
          "train_loss": 0.00013332809612620622,
          "validation_loss": 0.00011247523070778698,
          "train_metric": 1.3428318197838962e-05,
          "validation_metric": 1.132809484261088e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.711855816000025
        },
        {
          "epoch": 335,
          "updates": 5360,
          "samples_seen": 343040,
          "online_loss": 0.00014095412120695983,
          "train_loss": 0.00012153361603850499,
          "validation_loss": 0.00010338321590097621,
          "train_metric": 1.2240420801390428e-05,
          "validation_metric": 1.0412379197077826e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.722797438998896
        },
        {
          "epoch": 336,
          "updates": 5376,
          "samples_seen": 344064,
          "online_loss": 0.0001341061313269165,
          "train_loss": 0.0001263347949134186,
          "validation_loss": 0.00011401130177546293,
          "train_metric": 1.2723978215944953e-05,
          "validation_metric": 1.1482800800877158e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7335471530004725
        },
        {
          "epoch": 337,
          "updates": 5392,
          "samples_seen": 345088,
          "online_loss": 0.0001305275995946431,
          "train_loss": 0.00012248703569639474,
          "validation_loss": 0.00010258326074108481,
          "train_metric": 1.2336446161498316e-05,
          "validation_metric": 1.0331810699426569e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.744437889999972
        },
        {
          "epoch": 338,
          "updates": 5408,
          "samples_seen": 346112,
          "online_loss": 0.00013801654904455063,
          "train_loss": 0.00011976925452472642,
          "validation_loss": 0.00010371302050771192,
          "train_metric": 1.2062720998073928e-05,
          "validation_metric": 1.0445595762575977e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7550072710000677
        },
        {
          "epoch": 339,
          "updates": 5424,
          "samples_seen": 347136,
          "online_loss": 0.00013433379535854328,
          "train_loss": 0.00011986611207248643,
          "validation_loss": 0.00010431643022457138,
          "train_metric": 1.2072476238245144e-05,
          "validation_metric": 1.0506369108043145e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7658517449999636
        },
        {
          "epoch": 340,
          "updates": 5440,
          "samples_seen": 348160,
          "online_loss": 0.00013222292909631506,
          "train_loss": 0.00012394336226861924,
          "validation_loss": 0.00010283719166181982,
          "train_metric": 1.2483123100537341e-05,
          "validation_metric": 1.0357386599935126e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7767338329995255
        },
        {
          "epoch": 341,
          "updates": 5456,
          "samples_seen": 349184,
          "online_loss": 0.00012296951535972767,
          "train_loss": 0.00011904184066224843,
          "validation_loss": 0.00010319468128727749,
          "train_metric": 1.1989457561867312e-05,
          "validation_metric": 1.0393391676188912e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7872001539999474
        },
        {
          "epoch": 342,
          "updates": 5472,
          "samples_seen": 350208,
          "online_loss": 0.00013412264252110617,
          "train_loss": 0.00012149872054578736,
          "validation_loss": 0.00010521228250581771,
          "train_metric": 1.2236905604368076e-05,
          "validation_metric": 1.0596595529932529e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.7979317779991106
        },
        {
          "epoch": 343,
          "updates": 5488,
          "samples_seen": 351232,
          "online_loss": 0.00013377578989093308,
          "train_loss": 0.0001330860104644671,
          "validation_loss": 0.0001144636407843791,
          "train_metric": 1.340393464488443e-05,
          "validation_metric": 1.1528358299983665e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8088589959988894
        },
        {
          "epoch": 344,
          "updates": 5504,
          "samples_seen": 352256,
          "online_loss": 0.00012965663700015284,
          "train_loss": 0.00011939598334720358,
          "validation_loss": 0.00010602630209177732,
          "train_metric": 1.2025126125081442e-05,
          "validation_metric": 1.0678581020329148e-05,
          "lr_first": 2.34375e-05,
          "lr_last": 2.34375e-05,
          "seconds": 3.8196390220000467
        },
        {
          "epoch": 345,
          "updates": 5520,
          "samples_seen": 353280,
          "online_loss": 0.0001237312662851764,
          "train_loss": 0.0001196884986711666,
          "validation_loss": 0.00010117220517713577,
          "train_metric": 1.2054588296450675e-05,
          "validation_metric": 1.0189694876316935e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.831007590999434
        },
        {
          "epoch": 346,
          "updates": 5536,
          "samples_seen": 354304,
          "online_loss": 0.00012727184503091848,
          "train_loss": 0.00011932542838621885,
          "validation_loss": 0.00010115512122865766,
          "train_metric": 1.2018019333481789e-05,
          "validation_metric": 1.018797411234118e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.842528494000362
        },
        {
          "epoch": 347,
          "updates": 5552,
          "samples_seen": 355328,
          "online_loss": 0.00012165861789981136,
          "train_loss": 0.00011961902782786638,
          "validation_loss": 0.00010123422543983907,
          "train_metric": 1.2047591553709935e-05,
          "validation_metric": 1.019594037643401e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.853362156000003
        },
        {
          "epoch": 348,
          "updates": 5568,
          "samples_seen": 356352,
          "online_loss": 0.00011830036737592309,
          "train_loss": 0.00011902406549779698,
          "validation_loss": 0.00010564127296674997,
          "train_metric": 1.1987668585788924e-05,
          "validation_metric": 1.0639801985234953e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.863897834000454
        },
        {
          "epoch": 349,
          "updates": 5584,
          "samples_seen": 357376,
          "online_loss": 0.00011926632942049764,
          "train_loss": 0.00011754878505598754,
          "validation_loss": 0.00010135708726011217,
          "train_metric": 1.1839084436360281e-05,
          "validation_metric": 1.0208315870841034e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.874738585000159
        },
        {
          "epoch": 350,
          "updates": 5600,
          "samples_seen": 358400,
          "online_loss": 0.00012011145554424729,
          "train_loss": 0.0001182068299385719,
          "validation_loss": 0.00010057622421300039,
          "train_metric": 1.1905358405783772e-05,
          "validation_metric": 1.0129669135494623e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.8857110899989493
        },
        {
          "epoch": 351,
          "updates": 5616,
          "samples_seen": 359424,
          "online_loss": 0.00012036638008794398,
          "train_loss": 0.00012007556506432593,
          "validation_loss": 0.00010598552762530744,
          "train_metric": 1.2093571058358066e-05,
          "validation_metric": 1.0674473742255941e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.896325433999664
        },
        {
          "epoch": 352,
          "updates": 5632,
          "samples_seen": 360448,
          "online_loss": 0.00012298541514610406,
          "train_loss": 0.00011718643509084359,
          "validation_loss": 0.00010219415707979351,
          "train_metric": 1.1802587323472835e-05,
          "validation_metric": 1.0292621482221875e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9071364829997037
        },
        {
          "epoch": 353,
          "updates": 5648,
          "samples_seen": 361472,
          "online_loss": 0.0001194707319882582,
          "train_loss": 0.00011732635903172195,
          "validation_loss": 0.00010215462680207565,
          "train_metric": 1.1816680853371508e-05,
          "validation_metric": 1.0288639714417513e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9177013139997143
        },
        {
          "epoch": 354,
          "updates": 5664,
          "samples_seen": 362496,
          "online_loss": 0.00011853205614897888,
          "train_loss": 0.0001170575269497931,
          "validation_loss": 0.0001012493739835918,
          "train_metric": 1.1789605196099728e-05,
          "validation_metric": 1.0197465599048883e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.928203625999231
        },
        {
          "epoch": 355,
          "updates": 5680,
          "samples_seen": 363520,
          "online_loss": 0.00011907997168236761,
          "train_loss": 0.0001169405149994418,
          "validation_loss": 0.0001018055627355352,
          "train_metric": 1.1777820873248857e-05,
          "validation_metric": 1.0253483196720481e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.938660505998996
        },
        {
          "epoch": 356,
          "updates": 5696,
          "samples_seen": 364544,
          "online_loss": 0.0001235167715094576,
          "train_loss": 0.00011704774806275964,
          "validation_loss": 0.00010103399836225435,
          "train_metric": 1.1788619303843006e-05,
          "validation_metric": 1.0175774150411598e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9491189239997766
        },
        {
          "epoch": 357,
          "updates": 5712,
          "samples_seen": 365568,
          "online_loss": 0.00011839147100545233,
          "train_loss": 0.00011717536835931242,
          "validation_loss": 0.00010061082866741344,
          "train_metric": 1.1801475920947269e-05,
          "validation_metric": 1.0133153409697115e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9596851619990048
        },
        {
          "epoch": 358,
          "updates": 5728,
          "samples_seen": 366592,
          "online_loss": 0.00011968119861194282,
          "train_loss": 0.00011692762200254947,
          "validation_loss": 0.00010123484389623627,
          "train_metric": 1.1776523024309427e-05,
          "validation_metric": 1.0196003131568432e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9701373530006094
        },
        {
          "epoch": 359,
          "updates": 5744,
          "samples_seen": 367616,
          "online_loss": 0.00011898063576154527,
          "train_loss": 0.00011689562234096229,
          "validation_loss": 0.00010361374006606638,
          "train_metric": 1.1773299775086343e-05,
          "validation_metric": 1.0435596777824685e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9805926349999936
        },
        {
          "epoch": 360,
          "updates": 5760,
          "samples_seen": 368640,
          "online_loss": 0.0001194012506857689,
          "train_loss": 0.00011709009413607419,
          "validation_loss": 0.00010030525299953297,
          "train_metric": 1.1792886652983725e-05,
          "validation_metric": 1.010237610898912e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 3.9917787430003955
        },
        {
          "epoch": 361,
          "updates": 5776,
          "samples_seen": 369664,
          "online_loss": 0.00011979264854744542,
          "train_loss": 0.00011730838741641492,
          "validation_loss": 0.00010031589772552252,
          "train_metric": 1.1814870049420279e-05,
          "validation_metric": 1.0103450222231913e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.002210233998994
        },
        {
          "epoch": 362,
          "updates": 5792,
          "samples_seen": 370688,
          "online_loss": 0.00012028600212943275,
          "train_loss": 0.00011797250044764951,
          "validation_loss": 0.00010333836689824238,
          "train_metric": 1.1881758837262169e-05,
          "validation_metric": 1.0407862646388821e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.013150778999261
        },
        {
          "epoch": 363,
          "updates": 5808,
          "samples_seen": 371712,
          "online_loss": 0.00012107792417737073,
          "train_loss": 0.00011757243191823363,
          "validation_loss": 0.00010074990132125095,
          "train_metric": 1.1841465493489522e-05,
          "validation_metric": 1.0147163266083226e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.023556038000606
        },
        {
          "epoch": 364,
          "updates": 5824,
          "samples_seen": 372736,
          "online_loss": 0.000118265079890989,
          "train_loss": 0.00011644669575616717,
          "validation_loss": 0.00010191848559770733,
          "train_metric": 1.1728085155482404e-05,
          "validation_metric": 1.0264856427966151e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.033948992999285
        },
        {
          "epoch": 365,
          "updates": 5840,
          "samples_seen": 373760,
          "online_loss": 0.00011739341744032572,
          "train_loss": 0.00011686691141221672,
          "validation_loss": 0.00010102631495101377,
          "train_metric": 1.1770406672440004e-05,
          "validation_metric": 1.0175001989409793e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.044192186000146
        },
        {
          "epoch": 366,
          "updates": 5856,
          "samples_seen": 374784,
          "online_loss": 0.00011733908104361035,
          "train_loss": 0.000116652314318344,
          "validation_loss": 0.00010212054621661082,
          "train_metric": 1.1748794349841774e-05,
          "validation_metric": 1.028520637191832e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.054652032999002
        },
        {
          "epoch": 367,
          "updates": 5872,
          "samples_seen": 375808,
          "online_loss": 0.0001209467582157231,
          "train_loss": 0.00011621017620200291,
          "validation_loss": 0.00010256255336571485,
          "train_metric": 1.1704263670253567e-05,
          "validation_metric": 1.0329724318580702e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.064966051999363
        },
        {
          "epoch": 368,
          "updates": 5888,
          "samples_seen": 376832,
          "online_loss": 0.00011794371357609634,
          "train_loss": 0.00011834804899990559,
          "validation_loss": 0.00010125755215995014,
          "train_metric": 1.1919582902919501e-05,
          "validation_metric": 1.0198289601248689e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.075546141999439
        },
        {
          "epoch": 369,
          "updates": 5904,
          "samples_seen": 377856,
          "online_loss": 0.00012069015610904898,
          "train_loss": 0.00012557314767036587,
          "validation_loss": 0.00011220663145650178,
          "train_metric": 1.2647267794818617e-05,
          "validation_metric": 1.1301040103717241e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.085847612999714
        },
        {
          "epoch": 370,
          "updates": 5920,
          "samples_seen": 378880,
          "online_loss": 0.00012719532514893217,
          "train_loss": 0.00011909359454875812,
          "validation_loss": 0.00010103250679094344,
          "train_metric": 1.1994670785497874e-05,
          "validation_metric": 1.0175623174291104e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.0961590179995255
        },
        {
          "epoch": 371,
          "updates": 5936,
          "samples_seen": 379904,
          "online_loss": 0.00011748981705750339,
          "train_loss": 0.00011724825162673369,
          "validation_loss": 0.0001043897500494495,
          "train_metric": 1.1808815543190576e-05,
          "validation_metric": 1.0513754205021542e-05,
          "lr_first": 1.171875e-05,
          "lr_last": 1.171875e-05,
          "seconds": 4.107074072000614
        },
        {
          "epoch": 372,
          "updates": 5952,
          "samples_seen": 380928,
          "online_loss": 0.00011856806395371677,
          "train_loss": 0.00011894953058799729,
          "validation_loss": 0.00010151691822102293,
          "train_metric": 1.1980162526015192e-05,
          "validation_metric": 1.0224412108073011e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1175760490004905
        },
        {
          "epoch": 373,
          "updates": 5968,
          "samples_seen": 381952,
          "online_loss": 0.0001210263772009057,
          "train_loss": 0.00011749412806238979,
          "validation_loss": 0.00010454062430653721,
          "train_metric": 1.183357926493045e-05,
          "validation_metric": 1.052894822350936e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1277949399991485
        },
        {
          "epoch": 374,
          "updates": 5984,
          "samples_seen": 382976,
          "online_loss": 0.00011697524814735516,
          "train_loss": 0.00011877669749082997,
          "validation_loss": 0.0001013004221022129,
          "train_metric": 1.1962754797423258e-05,
          "validation_metric": 1.0202607882092707e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.138287903999299
        },
        {
          "epoch": 375,
          "updates": 6000,
          "samples_seen": 384000,
          "online_loss": 0.00011961349491684814,
          "train_loss": 0.00011622215970419347,
          "validation_loss": 0.000103728860267438,
          "train_metric": 1.1705469660228118e-05,
          "validation_metric": 1.0447191016282886e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1488379610000266
        },
        {
          "epoch": 376,
          "updates": 6016,
          "samples_seen": 385024,
          "online_loss": 0.00011840621846204158,
          "train_loss": 0.00011451840691734105,
          "validation_loss": 0.00010079416097141802,
          "train_metric": 1.1533874385349918e-05,
          "validation_metric": 1.0151619790121913e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.159268957000677
        },
        {
          "epoch": 377,
          "updates": 6032,
          "samples_seen": 386048,
          "online_loss": 0.00011892571501448401,
          "train_loss": 0.00011590697249630466,
          "validation_loss": 0.00010395294521003962,
          "train_metric": 1.1673725566652138e-05,
          "validation_metric": 1.046975921781268e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.169542481000462
        },
        {
          "epoch": 378,
          "updates": 6048,
          "samples_seen": 387072,
          "online_loss": 0.00012360786604403984,
          "train_loss": 0.00011897492368007079,
          "validation_loss": 0.00010046976240118966,
          "train_metric": 1.1982720025116578e-05,
          "validation_metric": 1.0118947102455422e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.180164645998957
        },
        {
          "epoch": 379,
          "updates": 6064,
          "samples_seen": 388096,
          "online_loss": 0.0001199271946461522,
          "train_loss": 0.00011555013770703226,
          "validation_loss": 0.00010430652764625847,
          "train_metric": 1.163778688351158e-05,
          "validation_metric": 1.0505370482860599e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.1905380909993255
        },
        {
          "epoch": 380,
          "updates": 6080,
          "samples_seen": 389120,
          "online_loss": 0.00012010771979475976,
          "train_loss": 0.00011597483535297215,
          "validation_loss": 9.914481779560447e-05,
          "train_metric": 1.1680560419335961e-05,
          "validation_metric": 9.985503311327193e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.201358006999726
        },
        {
          "epoch": 381,
          "updates": 6096,
          "samples_seen": 390144,
          "online_loss": 0.0001158722893705999,
          "train_loss": 0.00011505777365528047,
          "validation_loss": 0.00010122613457497209,
          "train_metric": 1.1588196684897412e-05,
          "validation_metric": 1.019512455968652e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.2120139180005935
        },
        {
          "epoch": 382,
          "updates": 6112,
          "samples_seen": 391168,
          "online_loss": 0.00011881462751262006,
          "train_loss": 0.00011484964488772675,
          "validation_loss": 0.00010026409290730953,
          "train_metric": 1.1567235560505651e-05,
          "validation_metric": 1.009823154163314e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.222344433999751
        },
        {
          "epoch": 383,
          "updates": 6128,
          "samples_seen": 392192,
          "online_loss": 0.00012187364950477786,
          "train_loss": 0.00011636513954726979,
          "validation_loss": 0.00010465316154295579,
          "train_metric": 1.171987059933599e-05,
          "validation_metric": 1.0540283255977556e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.2325568569995085
        },
        {
          "epoch": 384,
          "updates": 6144,
          "samples_seen": 393216,
          "online_loss": 0.00011840815204777755,
          "train_loss": 0.00011384519893908873,
          "validation_loss": 9.864476305665448e-05,
          "train_metric": 1.1466070645838045e-05,
          "validation_metric": 9.935139132721815e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.243587715000103
        },
        {
          "epoch": 385,
          "updates": 6160,
          "samples_seen": 394240,
          "online_loss": 0.00011590845042519504,
          "train_loss": 0.00011374989117030054,
          "validation_loss": 0.00010105074034072459,
          "train_metric": 1.1456471838755533e-05,
          "validation_metric": 1.0177460353588685e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.254136446999837
        },
        {
          "epoch": 386,
          "updates": 6176,
          "samples_seen": 395264,
          "online_loss": 0.00011585345691855764,
          "train_loss": 0.00011342180368956178,
          "validation_loss": 9.930602391250432e-05,
          "train_metric": 1.1423428077250719e-05,
          "validation_metric": 1.0001738701248541e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.264446999999564
        },
        {
          "epoch": 387,
          "updates": 6192,
          "samples_seen": 396288,
          "online_loss": 0.00011509702835610369,
          "train_loss": 0.00011331788846291602,
          "validation_loss": 9.951237734640017e-05,
          "train_metric": 1.1412962521717418e-05,
          "validation_metric": 1.0022522474173456e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.274939984999946
        },
        {
          "epoch": 388,
          "updates": 6208,
          "samples_seen": 397312,
          "online_loss": 0.00011720858537955792,
          "train_loss": 0.00011447772703832015,
          "validation_loss": 9.859549027169123e-05,
          "train_metric": 1.1529778021213133e-05,
          "validation_metric": 9.930176929628942e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.285787931999948
        },
        {
          "epoch": 389,
          "updates": 6224,
          "samples_seen": 398336,
          "online_loss": 0.00011925807211810024,
          "train_loss": 0.0001140281674452126,
          "validation_loss": 0.0001008929539239034,
          "train_metric": 1.1484498827485368e-05,
          "validation_metric": 1.0161567843169905e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.296332927999174
        },
        {
          "epoch": 390,
          "updates": 6240,
          "samples_seen": 399360,
          "online_loss": 0.00011492352587083587,
          "train_loss": 0.00011381888180039823,
          "validation_loss": 9.95094160316512e-05,
          "train_metric": 1.1463422197266482e-05,
          "validation_metric": 1.0022223250416573e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.30719068200051
        },
        {
          "epoch": 391,
          "updates": 6256,
          "samples_seen": 400384,
          "online_loss": 0.00011630894096015254,
          "train_loss": 0.00011318628094159067,
          "validation_loss": 9.853530355030671e-05,
          "train_metric": 1.1399706636439078e-05,
          "validation_metric": 9.924115147441626e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.318534418000127
        },
        {
          "epoch": 392,
          "updates": 6272,
          "samples_seen": 401408,
          "online_loss": 0.00011702520669132355,
          "train_loss": 0.00011440122761996463,
          "validation_loss": 0.00010210680193267763,
          "train_metric": 1.1522072782099713e-05,
          "validation_metric": 1.0283823030476924e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.329015281000466
        },
        {
          "epoch": 393,
          "updates": 6288,
          "samples_seen": 402432,
          "online_loss": 0.00011478357100713765,
          "train_loss": 0.00011426592391217127,
          "validation_loss": 9.887678606901318e-05,
          "train_metric": 1.1508445822983049e-05,
          "validation_metric": 9.958507689589169e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.339575337000497
        },
        {
          "epoch": 394,
          "updates": 6304,
          "samples_seen": 403456,
          "online_loss": 0.00011755983450711938,
          "train_loss": 0.00011319336044834927,
          "validation_loss": 9.86841696430929e-05,
          "train_metric": 1.1400422408769373e-05,
          "validation_metric": 9.93910725810565e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.350345307999305
        },
        {
          "epoch": 395,
          "updates": 6320,
          "samples_seen": 404480,
          "online_loss": 0.00011484301012387732,
          "train_loss": 0.00011236728460062295,
          "validation_loss": 9.946926729753613e-05,
          "train_metric": 1.1317220923956484e-05,
          "validation_metric": 1.001817963697249e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.361090439999316
        },
        {
          "epoch": 396,
          "updates": 6336,
          "samples_seen": 405504,
          "online_loss": 0.00011860741642522044,
          "train_loss": 0.00011269073729636148,
          "validation_loss": 0.00010062866931548342,
          "train_metric": 1.1349798114679288e-05,
          "validation_metric": 1.0134950571227819e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.371497509000619
        },
        {
          "epoch": 397,
          "updates": 6352,
          "samples_seen": 406528,
          "online_loss": 0.00011840965862575104,
          "train_loss": 0.00011283194908173755,
          "validation_loss": 9.758459782460704e-05,
          "train_metric": 1.1364020792825613e-05,
          "validation_metric": 9.828363545238972e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.3825793500000145
        },
        {
          "epoch": 398,
          "updates": 6368,
          "samples_seen": 407552,
          "online_loss": 0.00011580969157876098,
          "train_loss": 0.00011248410737607628,
          "validation_loss": 9.812042844714597e-05,
          "train_metric": 1.132898705691332e-05,
          "validation_metric": 9.88232932286337e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.393689767999604
        },
        {
          "epoch": 399,
          "updates": 6384,
          "samples_seen": 408576,
          "online_loss": 0.00011415096469136188,
          "train_loss": 0.0001119306543841958,
          "validation_loss": 9.945653437171131e-05,
          "train_metric": 1.1273246855125763e-05,
          "validation_metric": 1.0016899068432394e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.404317810000066
        },
        {
          "epoch": 400,
          "updates": 6400,
          "samples_seen": 409600,
          "online_loss": 0.00011343197638780111,
          "train_loss": 0.00011194698163308203,
          "validation_loss": 9.871529618976638e-05,
          "train_metric": 1.1274889402557164e-05,
          "validation_metric": 9.942243195837364e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.415183713999795
        },
        {
          "epoch": 401,
          "updates": 6416,
          "samples_seen": 410624,
          "online_loss": 0.00011318441420371528,
          "train_loss": 0.00011199665459571406,
          "validation_loss": 9.779026004252955e-05,
          "train_metric": 1.1279895261395723e-05,
          "validation_metric": 9.849078196566552e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.426112529999955
        },
        {
          "epoch": 402,
          "updates": 6432,
          "samples_seen": 411648,
          "online_loss": 0.00011569881689865724,
          "train_loss": 0.00011213816469535232,
          "validation_loss": 9.794446668820456e-05,
          "train_metric": 1.1294143405393697e-05,
          "validation_metric": 9.864607818599325e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.436655643999984
        },
        {
          "epoch": 403,
          "updates": 6448,
          "samples_seen": 412672,
          "online_loss": 0.00012107031852792716,
          "train_loss": 0.00011174041719641536,
          "validation_loss": 0.00010023728827945888,
          "train_metric": 1.125408562074881e-05,
          "validation_metric": 1.0095532161358278e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.509380565999891
        },
        {
          "epoch": 404,
          "updates": 6464,
          "samples_seen": 413696,
          "online_loss": 0.0001129152824432822,
          "train_loss": 0.00011141198046971112,
          "validation_loss": 9.889063221635297e-05,
          "train_metric": 1.1221006388950627e-05,
          "validation_metric": 9.95990376395639e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.520894522000162
        },
        {
          "epoch": 405,
          "updates": 6480,
          "samples_seen": 414720,
          "online_loss": 0.00011398750712032779,
          "train_loss": 0.00011435010674176738,
          "validation_loss": 9.831484203459695e-05,
          "train_metric": 1.1516924132592976e-05,
          "validation_metric": 9.90190983429784e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.531772669999555
        },
        {
          "epoch": 406,
          "updates": 6496,
          "samples_seen": 415744,
          "online_loss": 0.00011870577327499632,
          "train_loss": 0.00012070733646396548,
          "validation_loss": 0.00011247756629018113,
          "train_metric": 1.2157202036178205e-05,
          "validation_metric": 1.1328328582749236e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.54294144800042
        },
        {
          "epoch": 407,
          "updates": 6512,
          "samples_seen": 416768,
          "online_loss": 0.00011444150095485384,
          "train_loss": 0.0001147189614130184,
          "validation_loss": 9.958698501577601e-05,
          "train_metric": 1.1554073353181593e-05,
          "validation_metric": 1.0030036719399504e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.553507382999669
        },
        {
          "epoch": 408,
          "updates": 6528,
          "samples_seen": 417792,
          "online_loss": 0.00012598485227499623,
          "train_loss": 0.00012151057308074087,
          "validation_loss": 0.00010953928722301498,
          "train_metric": 1.2238101589900907e-05,
          "validation_metric": 1.103239628719166e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.564625230999809
        },
        {
          "epoch": 409,
          "updates": 6544,
          "samples_seen": 418816,
          "online_loss": 0.00011841857212857576,
          "train_loss": 0.00011103863653261214,
          "validation_loss": 9.671389852883294e-05,
          "train_metric": 1.1183406058989931e-05,
          "validation_metric": 9.740669156599324e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.575945976999719
        },
        {
          "epoch": 410,
          "updates": 6560,
          "samples_seen": 419840,
          "online_loss": 0.00011348322914273012,
          "train_loss": 0.00011089001782238483,
          "validation_loss": 9.839069389272481e-05,
          "train_metric": 1.116843668569345e-05,
          "validation_metric": 9.909550499287434e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.586728947000665
        },
        {
          "epoch": 411,
          "updates": 6576,
          "samples_seen": 420864,
          "online_loss": 0.00011534524810485891,
          "train_loss": 0.00011280905891908333,
          "validation_loss": 9.805863373912871e-05,
          "train_metric": 1.1361717042746022e-05,
          "validation_metric": 9.876107469608542e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.597336532999179
        },
        {
          "epoch": 412,
          "updates": 6592,
          "samples_seen": 421888,
          "online_loss": 0.00011416446591283602,
          "train_loss": 0.00011266983347013593,
          "validation_loss": 0.00010249268962070346,
          "train_metric": 1.1347691724949982e-05,
          "validation_metric": 1.0322687558073085e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.607910882999931
        },
        {
          "epoch": 413,
          "updates": 6608,
          "samples_seen": 422912,
          "online_loss": 0.0001121482869166357,
          "train_loss": 0.00011129979975521564,
          "validation_loss": 9.722316463012248e-05,
          "train_metric": 1.1209708645765204e-05,
          "validation_metric": 9.791960110305808e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.6185571549995075
        },
        {
          "epoch": 414,
          "updates": 6624,
          "samples_seen": 423936,
          "online_loss": 0.00011568094760150416,
          "train_loss": 0.00011517491657286882,
          "validation_loss": 0.00010274887608829886,
          "train_metric": 1.1599995559663512e-05,
          "validation_metric": 1.0348490832257085e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.629049701999975
        },
        {
          "epoch": 415,
          "updates": 6640,
          "samples_seen": 424960,
          "online_loss": 0.00011426559467508923,
          "train_loss": 0.00011434390762588009,
          "validation_loss": 9.710701124276966e-05,
          "train_metric": 1.1516299309732858e-05,
          "validation_metric": 9.780263098946307e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.639338812999995
        },
        {
          "epoch": 416,
          "updates": 6656,
          "samples_seen": 425984,
          "online_loss": 0.00011350387558195507,
          "train_loss": 0.00011114742665085942,
          "validation_loss": 9.960289025912061e-05,
          "train_metric": 1.1194361832167488e-05,
          "validation_metric": 1.0031638339569326e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.649761438000496
        },
        {
          "epoch": 417,
          "updates": 6672,
          "samples_seen": 427008,
          "online_loss": 0.00011340971605022787,
          "train_loss": 0.00011420306691434234,
          "validation_loss": 9.891821537166834e-05,
          "train_metric": 1.1502114830364008e-05,
          "validation_metric": 9.962681360775605e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.660157753000021
        },
        {
          "epoch": 418,
          "updates": 6688,
          "samples_seen": 428032,
          "online_loss": 0.0001132703032453719,
          "train_loss": 0.00011244932102272287,
          "validation_loss": 0.00010412128176540136,
          "train_metric": 1.1325484592816792e-05,
          "validation_metric": 1.0486713108548429e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.670567421999294
        },
        {
          "epoch": 419,
          "updates": 6704,
          "samples_seen": 429056,
          "online_loss": 0.00011945090682274895,
          "train_loss": 0.00011217664723517373,
          "validation_loss": 9.695017797639593e-05,
          "train_metric": 1.1298021490802057e-05,
          "validation_metric": 9.764468813955318e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.680795222000597
        },
        {
          "epoch": 420,
          "updates": 6720,
          "samples_seen": 430080,
          "online_loss": 0.00011579058627830818,
          "train_loss": 0.00011240438470849767,
          "validation_loss": 0.00010128672874998301,
          "train_metric": 1.1320958037686069e-05,
          "validation_metric": 1.0201229088124819e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.691253437000341
        },
        {
          "epoch": 421,
          "updates": 6736,
          "samples_seen": 431104,
          "online_loss": 0.00012314212563069304,
          "train_loss": 0.00011083731078542769,
          "validation_loss": 9.91106717265211e-05,
          "train_metric": 1.1163128874613903e-05,
          "validation_metric": 9.98206451185979e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.701608256000327
        },
        {
          "epoch": 422,
          "updates": 6752,
          "samples_seen": 432128,
          "online_loss": 0.00011646719940472394,
          "train_loss": 0.00011044798884540796,
          "validation_loss": 9.650251740822569e-05,
          "train_metric": 1.1123916010546964e-05,
          "validation_metric": 9.719381523609627e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.712726063000446
        },
        {
          "epoch": 423,
          "updates": 6768,
          "samples_seen": 433152,
          "online_loss": 0.00011163794783897174,
          "train_loss": 0.00011005810665665194,
          "validation_loss": 9.97727329377085e-05,
          "train_metric": 1.108464948629262e-05,
          "validation_metric": 1.0048744115920272e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.723162087000674
        },
        {
          "epoch": 424,
          "updates": 6784,
          "samples_seen": 434176,
          "online_loss": 0.00011685320760079776,
          "train_loss": 0.00011419077054597437,
          "validation_loss": 9.773339843377471e-05,
          "train_metric": 1.1500877008074895e-05,
          "validation_metric": 9.843349289440084e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.733608747999824
        },
        {
          "epoch": 425,
          "updates": 6800,
          "samples_seen": 435200,
          "online_loss": 0.00011604075484683563,
          "train_loss": 0.00011044455459341407,
          "validation_loss": 9.92924251477234e-05,
          "train_metric": 1.1123571312054992e-05,
          "validation_metric": 1.0000367183238268e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.744201987999986
        },
        {
          "epoch": 426,
          "updates": 6816,
          "samples_seen": 436224,
          "online_loss": 0.00011131335304526147,
          "train_loss": 0.00011078029638156295,
          "validation_loss": 9.656239853939041e-05,
          "train_metric": 1.115738632506691e-05,
          "validation_metric": 9.72541056398768e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.754565240999
        },
        {
          "epoch": 427,
          "updates": 6832,
          "samples_seen": 437248,
          "online_loss": 0.00011450921283540083,
          "train_loss": 0.00011128067126264796,
          "validation_loss": 0.00010173735790885985,
          "train_metric": 1.1207781426492147e-05,
          "validation_metric": 1.0246614692732692e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.764779613000428
        },
        {
          "epoch": 428,
          "updates": 6848,
          "samples_seen": 438272,
          "online_loss": 0.00011446103280832176,
          "train_loss": 0.00010904434020631015,
          "validation_loss": 9.810055780690163e-05,
          "train_metric": 1.0982546882587485e-05,
          "validation_metric": 9.88032843451947e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.775463829999353
        },
        {
          "epoch": 429,
          "updates": 6864,
          "samples_seen": 439296,
          "online_loss": 0.00011349654050718527,
          "train_loss": 0.00010991173621732742,
          "validation_loss": 9.614020382286981e-05,
          "train_metric": 1.1069908396166284e-05,
          "validation_metric": 9.682889867690392e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.7862726589992235
        },
        {
          "epoch": 430,
          "updates": 6880,
          "samples_seen": 440320,
          "online_loss": 0.00011277276644250378,
          "train_loss": 0.00010866894444916397,
          "validation_loss": 9.686005068942904e-05,
          "train_metric": 1.094473736884538e-05,
          "validation_metric": 9.755389328347519e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.796743188000619
        },
        {
          "epoch": 431,
          "updates": 6896,
          "samples_seen": 441344,
          "online_loss": 0.0001119897367516387,
          "train_loss": 0.0001095066181733273,
          "validation_loss": 9.86789382295683e-05,
          "train_metric": 1.1029104825865943e-05,
          "validation_metric": 9.938581570168026e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.80718581399924
        },
        {
          "epoch": 432,
          "updates": 6912,
          "samples_seen": 442368,
          "online_loss": 0.00011497466903165332,
          "train_loss": 0.00010893161379499361,
          "validation_loss": 9.861210128292441e-05,
          "train_metric": 1.0971193660225254e-05,
          "validation_metric": 9.931849490385503e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.817416945999867
        },
        {
          "epoch": 433,
          "updates": 6928,
          "samples_seen": 443392,
          "online_loss": 0.00011217560995646636,
          "train_loss": 0.00010848032252397388,
          "validation_loss": 9.60082616074942e-05,
          "train_metric": 1.092574075300945e-05,
          "validation_metric": 9.669601240602788e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.828380023000136
        },
        {
          "epoch": 434,
          "updates": 6944,
          "samples_seen": 444416,
          "online_loss": 0.00011485256982268766,
          "train_loss": 0.00011089146573795006,
          "validation_loss": 0.0001011273852782324,
          "train_metric": 1.1168582204845734e-05,
          "validation_metric": 1.0185179235122632e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.839178576999984
        },
        {
          "epoch": 435,
          "updates": 6960,
          "samples_seen": 445440,
          "online_loss": 0.0001142889288985316,
          "train_loss": 0.00010815662972163409,
          "validation_loss": 9.666327241575345e-05,
          "train_metric": 1.0893139915424399e-05,
          "validation_metric": 9.735570529301185e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.849667958998907
        },
        {
          "epoch": 436,
          "updates": 6976,
          "samples_seen": 446464,
          "online_loss": 0.00011847047335322713,
          "train_loss": 0.00011232848919462413,
          "validation_loss": 9.769415919436142e-05,
          "train_metric": 1.1313313734717667e-05,
          "validation_metric": 9.839397534960881e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.860275847000594
        },
        {
          "epoch": 437,
          "updates": 6992,
          "samples_seen": 447488,
          "online_loss": 0.00012190973802717053,
          "train_loss": 0.0001115727354772389,
          "validation_loss": 0.0001017674439935945,
          "train_metric": 1.123719812312629e-05,
          "validation_metric": 1.0249645129079e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.870734645999619
        },
        {
          "epoch": 438,
          "updates": 7008,
          "samples_seen": 448512,
          "online_loss": 0.00011214901360290241,
          "train_loss": 0.00010809367813635617,
          "validation_loss": 9.773574129212648e-05,
          "train_metric": 1.088679982785834e-05,
          "validation_metric": 9.843584848567843e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.881348664999678
        },
        {
          "epoch": 439,
          "updates": 7024,
          "samples_seen": 449536,
          "online_loss": 0.0001135424931817397,
          "train_loss": 0.00011022409307770431,
          "validation_loss": 0.00010118460340891033,
          "train_metric": 1.1101365998911206e-05,
          "validation_metric": 1.0190943612542469e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.956335970999135
        },
        {
          "epoch": 440,
          "updates": 7040,
          "samples_seen": 450560,
          "online_loss": 0.00011019824978575343,
          "train_loss": 0.00010999116784660146,
          "validation_loss": 9.584355575498194e-05,
          "train_metric": 1.1077906492573675e-05,
          "validation_metric": 9.653011147747748e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.968044478000593
        },
        {
          "epoch": 441,
          "updates": 7056,
          "samples_seen": 451584,
          "online_loss": 0.00011505716588544601,
          "train_loss": 0.00010752258822321892,
          "validation_loss": 9.488720388617367e-05,
          "train_metric": 1.0829281563928816e-05,
          "validation_metric": 9.556691111356486e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.979165102000479
        },
        {
          "epoch": 442,
          "updates": 7072,
          "samples_seen": 452608,
          "online_loss": 0.00010987433461195906,
          "train_loss": 0.00010785248741740361,
          "validation_loss": 9.777276136446744e-05,
          "train_metric": 1.0862508133868687e-05,
          "validation_metric": 9.847313776845112e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 4.9897976869997365
        },
        {
          "epoch": 443,
          "updates": 7088,
          "samples_seen": 453632,
          "online_loss": 0.00010911763774856809,
          "train_loss": 0.00010933881276287138,
          "validation_loss": 9.692188177723438e-05,
          "train_metric": 1.10122055048123e-05,
          "validation_metric": 9.761618457559962e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.000083664999693
        },
        {
          "epoch": 444,
          "updates": 7104,
          "samples_seen": 454656,
          "online_loss": 0.00011001697293977486,
          "train_loss": 0.00010765816841740161,
          "validation_loss": 9.765353024704382e-05,
          "train_metric": 1.0842936717381235e-05,
          "validation_metric": 9.835305718297604e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.010600648000036
        },
        {
          "epoch": 445,
          "updates": 7120,
          "samples_seen": 455680,
          "online_loss": 0.00011395893670851365,
          "train_loss": 0.00010933442536043003,
          "validation_loss": 9.495588165009394e-05,
          "train_metric": 1.1011761671397835e-05,
          "validation_metric": 9.563609637552872e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.020884383000521
        },
        {
          "epoch": 446,
          "updates": 7136,
          "samples_seen": 456704,
          "online_loss": 0.00011168536138939089,
          "train_loss": 0.00010741773439804092,
          "validation_loss": 9.464520553592592e-05,
          "train_metric": 1.081872051145183e-05,
          "validation_metric": 9.532318472338375e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.031956053000613
        },
        {
          "epoch": 447,
          "updates": 7152,
          "samples_seen": 457728,
          "online_loss": 0.0001162800899692229,
          "train_loss": 0.00010690790077205747,
          "validation_loss": 9.639984637033194e-05,
          "train_metric": 1.0767371350084431e-05,
          "validation_metric": 9.709038749861065e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.042518795000433
        },
        {
          "epoch": 448,
          "updates": 7168,
          "samples_seen": 458752,
          "online_loss": 0.0001166065817415074,
          "train_loss": 0.00011706599616445601,
          "validation_loss": 0.00010398840822745115,
          "train_metric": 1.1790459211624693e-05,
          "validation_metric": 1.0473330803506542e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.053025795999929
        },
        {
          "epoch": 449,
          "updates": 7184,
          "samples_seen": 459776,
          "online_loss": 0.00011629417758740601,
          "train_loss": 0.00011912670015590265,
          "validation_loss": 9.934431000147015e-05,
          "train_metric": 1.1998004993074574e-05,
          "validation_metric": 1.0005594958784059e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.06361270600064
        },
        {
          "epoch": 450,
          "updates": 7200,
          "samples_seen": 460800,
          "online_loss": 0.00011389928477001376,
          "train_loss": 0.00011998607806162909,
          "validation_loss": 0.00010936687613138929,
          "train_metric": 1.2084558875358198e-05,
          "validation_metric": 1.1015030395356007e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.074045056999239
        },
        {
          "epoch": 451,
          "updates": 7216,
          "samples_seen": 461824,
          "online_loss": 0.00011606053021750995,
          "train_loss": 0.00010630693577695638,
          "validation_loss": 9.562243212712929e-05,
          "train_metric": 1.0706844477681443e-05,
          "validation_metric": 9.630742169974837e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.084622505999505
        },
        {
          "epoch": 452,
          "updates": 7232,
          "samples_seen": 462848,
          "online_loss": 0.0001100759845940047,
          "train_loss": 0.00010715381358750165,
          "validation_loss": 9.490686352364719e-05,
          "train_metric": 1.0792139619297814e-05,
          "validation_metric": 9.558672900311649e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.095016330000362
        },
        {
          "epoch": 453,
          "updates": 7248,
          "samples_seen": 463872,
          "online_loss": 0.00011169486651851912,
          "train_loss": 0.00011168907803948969,
          "validation_loss": 9.927481733029708e-05,
          "train_metric": 1.124891423387453e-05,
          "validation_metric": 9.998595487559214e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.105345264999414
        },
        {
          "epoch": 454,
          "updates": 7264,
          "samples_seen": 464896,
          "online_loss": 0.00011043948461519904,
          "train_loss": 0.00010928334813797846,
          "validation_loss": 9.696312918094918e-05,
          "train_metric": 1.1006619388354011e-05,
          "validation_metric": 9.765771210368257e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.115799244998925
        },
        {
          "epoch": 455,
          "updates": 7280,
          "samples_seen": 465920,
          "online_loss": 0.00010814691950145061,
          "train_loss": 0.00010782759636640549,
          "validation_loss": 9.829366172198206e-05,
          "train_metric": 1.08600015664706e-05,
          "validation_metric": 9.899778888211586e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.126290498999879
        },
        {
          "epoch": 456,
          "updates": 7296,
          "samples_seen": 466944,
          "online_loss": 0.00010885192500609264,
          "train_loss": 0.00010834258864633739,
          "validation_loss": 9.594037692295387e-05,
          "train_metric": 1.0911868230323307e-05,
          "validation_metric": 9.66276456892956e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.13671654800055
        },
        {
          "epoch": 457,
          "updates": 7312,
          "samples_seen": 467968,
          "online_loss": 0.00011384774870748515,
          "train_loss": 0.0001255360257346183,
          "validation_loss": 0.00011609296052483842,
          "train_metric": 1.2643527952604927e-05,
          "validation_metric": 1.1692457519529853e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.14724818600007
        },
        {
          "epoch": 458,
          "updates": 7328,
          "samples_seen": 468992,
          "online_loss": 0.00012361062545096502,
          "train_loss": 0.00010783446487039328,
          "validation_loss": 9.47879088926129e-05,
          "train_metric": 1.0860690963454545e-05,
          "validation_metric": 9.546691217110492e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.157748678999269
        },
        {
          "epoch": 459,
          "updates": 7344,
          "samples_seen": 470016,
          "online_loss": 0.00011014131905540125,
          "train_loss": 0.00010620432294672355,
          "validation_loss": 9.399710688740015e-05,
          "train_metric": 1.0696510798879899e-05,
          "validation_metric": 9.467044947086833e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.168719372999476
        },
        {
          "epoch": 460,
          "updates": 7360,
          "samples_seen": 471040,
          "online_loss": 0.00010856383914870094,
          "train_loss": 0.00010967445996357128,
          "validation_loss": 0.00010151420428883284,
          "train_metric": 1.1046009603887796e-05,
          "validation_metric": 1.0224138350167777e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.179275491000226
        },
        {
          "epoch": 461,
          "updates": 7376,
          "samples_seen": 472064,
          "online_loss": 0.00012098080151190516,
          "train_loss": 0.00010769236541818827,
          "validation_loss": 9.421185677638277e-05,
          "train_metric": 1.0846378245332744e-05,
          "validation_metric": 9.488671821600292e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.18963001100019
        },
        {
          "epoch": 462,
          "updates": 7392,
          "samples_seen": 473088,
          "online_loss": 0.00010642884490152937,
          "train_loss": 0.00010750491492217407,
          "validation_loss": 9.687171404948458e-05,
          "train_metric": 1.0827501682797447e-05,
          "validation_metric": 9.756563486007508e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.199947441000404
        },
        {
          "epoch": 463,
          "updates": 7408,
          "samples_seen": 474112,
          "online_loss": 0.00010628622658259701,
          "train_loss": 0.00010840101458597928,
          "validation_loss": 9.530904208077118e-05,
          "train_metric": 1.091775357053848e-05,
          "validation_metric": 9.599177246855106e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.210531075999825
        },
        {
          "epoch": 464,
          "updates": 7424,
          "samples_seen": 475136,
          "online_loss": 0.00011012024560841382,
          "train_loss": 0.00010934702004306018,
          "validation_loss": 0.00010297170956619084,
          "train_metric": 1.1013029507012106e-05,
          "validation_metric": 1.0370932614023332e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.22084902899951
        },
        {
          "epoch": 465,
          "updates": 7440,
          "samples_seen": 476160,
          "online_loss": 0.00010987410132656805,
          "train_loss": 0.00010962299711536616,
          "validation_loss": 9.450079232919961e-05,
          "train_metric": 1.1040826393582392e-05,
          "validation_metric": 9.51777292357292e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.231088020000243
        },
        {
          "epoch": 466,
          "updates": 7456,
          "samples_seen": 477184,
          "online_loss": 0.00010710380456657731,
          "train_loss": 0.00010534487955737859,
          "validation_loss": 9.5357172540389e-05,
          "train_metric": 1.0609950550133362e-05,
          "validation_metric": 9.604025763110258e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.241558731000623
        },
        {
          "epoch": 467,
          "updates": 7472,
          "samples_seen": 478208,
          "online_loss": 0.00010804630755956168,
          "train_loss": 0.00010699193808250129,
          "validation_loss": 9.503951878286898e-05,
          "train_metric": 1.0775836926768534e-05,
          "validation_metric": 9.57203155849129e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.2517182100000355
        },
        {
          "epoch": 468,
          "updates": 7488,
          "samples_seen": 479232,
          "online_loss": 0.00011241403581152554,
          "train_loss": 0.00010717156692408025,
          "validation_loss": 9.895603579934686e-05,
          "train_metric": 1.0793928595376201e-05,
          "validation_metric": 9.96649032458663e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.261994226999377
        },
        {
          "epoch": 469,
          "updates": 7504,
          "samples_seen": 480256,
          "online_loss": 0.0001136335972660163,
          "train_loss": 0.00010440966434543952,
          "validation_loss": 9.490806405665353e-05,
          "train_metric": 1.0515759640838951e-05,
          "validation_metric": 9.558793863106985e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.272595180998906
        },
        {
          "epoch": 470,
          "updates": 7520,
          "samples_seen": 481280,
          "online_loss": 0.00011332608619341045,
          "train_loss": 0.00011183268361492082,
          "validation_loss": 9.615484304958954e-05,
          "train_metric": 1.1263379747106228e-05,
          "validation_metric": 9.684362339612562e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.282898584000577
        },
        {
          "epoch": 471,
          "updates": 7536,
          "samples_seen": 482304,
          "online_loss": 0.00011111687763332156,
          "train_loss": 0.00010432514682179317,
          "validation_loss": 9.497194696450606e-05,
          "train_metric": 1.0507246770430356e-05,
          "validation_metric": 9.565226719132625e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.293095824999909
        },
        {
          "epoch": 472,
          "updates": 7552,
          "samples_seen": 483328,
          "online_loss": 0.00010664694764273008,
          "train_loss": 0.00010489229316590354,
          "validation_loss": 9.523126209387556e-05,
          "train_metric": 1.0564368494669907e-05,
          "validation_metric": 9.591342859494034e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.30345435900017
        },
        {
          "epoch": 473,
          "updates": 7568,
          "samples_seen": 484352,
          "online_loss": 0.00010695965738705127,
          "train_loss": 0.00010535180626902729,
          "validation_loss": 9.679129288997501e-05,
          "train_metric": 1.0610648132569622e-05,
          "validation_metric": 9.74846443568822e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.31377172700013
        },
        {
          "epoch": 474,
          "updates": 7584,
          "samples_seen": 485376,
          "online_loss": 0.00010597625805530697,
          "train_loss": 0.00010396406287327409,
          "validation_loss": 9.440379653824493e-05,
          "train_metric": 1.0470879715285264e-05,
          "validation_metric": 9.50800585997058e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.32431144399925
        },
        {
          "epoch": 475,
          "updates": 7600,
          "samples_seen": 486400,
          "online_loss": 0.0001069185068445222,
          "train_loss": 0.00010411054245196283,
          "validation_loss": 9.453867824049667e-05,
          "train_metric": 1.0485632628842723e-05,
          "validation_metric": 9.52158916334156e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.394621454999651
        },
        {
          "epoch": 476,
          "updates": 7616,
          "samples_seen": 487424,
          "online_loss": 0.00010494034995645052,
          "train_loss": 0.00010372794349677861,
          "validation_loss": 9.450950165046379e-05,
          "train_metric": 1.0447099157318007e-05,
          "validation_metric": 9.518650585960131e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.405502694000461
        },
        {
          "epoch": 477,
          "updates": 7632,
          "samples_seen": 488448,
          "online_loss": 0.0001066803006324335,
          "train_loss": 0.0001078283748938702,
          "validation_loss": 9.338502422906458e-05,
          "train_metric": 1.0860079783014953e-05,
          "validation_metric": 9.40539757721126e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.416950513999836
        },
        {
          "epoch": 478,
          "updates": 7648,
          "samples_seen": 489472,
          "online_loss": 0.00010944953055513906,
          "train_loss": 0.00011472990445327014,
          "validation_loss": 0.00010494954767636955,
          "train_metric": 1.1555175660760142e-05,
          "validation_metric": 1.0570134691079147e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.427472444000159
        },
        {
          "epoch": 479,
          "updates": 7664,
          "samples_seen": 490496,
          "online_loss": 0.00011288184805380297,
          "train_loss": 0.00010389093949925154,
          "validation_loss": 9.421988215763122e-05,
          "train_metric": 1.0463514627190307e-05,
          "validation_metric": 9.489482181379572e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.438044005999473
        },
        {
          "epoch": 480,
          "updates": 7680,
          "samples_seen": 491520,
          "online_loss": 0.00010791632121254224,
          "train_loss": 0.00010392324475105852,
          "validation_loss": 9.484781912760809e-05,
          "train_metric": 1.046676879923325e-05,
          "validation_metric": 9.552724804962054e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.448582802000601
        },
        {
          "epoch": 481,
          "updates": 7696,
          "samples_seen": 492544,
          "online_loss": 0.00010905315548370709,
          "train_loss": 0.00010404109343653545,
          "validation_loss": 9.323405538452789e-05,
          "train_metric": 1.0478637705091387e-05,
          "validation_metric": 9.39019173529232e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.459497542000463
        },
        {
          "epoch": 482,
          "updates": 7712,
          "samples_seen": 493568,
          "online_loss": 0.00010508045602364291,
          "train_loss": 0.0001036516041494906,
          "validation_loss": 9.310798486694694e-05,
          "train_metric": 1.0439410289109219e-05,
          "validation_metric": 9.377495189255569e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.470766239999648
        },
        {
          "epoch": 483,
          "updates": 7728,
          "samples_seen": 494592,
          "online_loss": 0.00011074986286985222,
          "train_loss": 0.00011201926099602133,
          "validation_loss": 0.0001041077557601966,
          "train_metric": 1.1282168998150155e-05,
          "validation_metric": 1.0485350685485173e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.4812825939989125
        },
        {
          "epoch": 484,
          "updates": 7744,
          "samples_seen": 495616,
          "online_loss": 0.00011380027081031585,
          "train_loss": 0.00010336559353163466,
          "validation_loss": 9.422965376870707e-05,
          "train_metric": 1.0410604772914667e-05,
          "validation_metric": 9.49046625464689e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.491712581000684
        },
        {
          "epoch": 485,
          "updates": 7760,
          "samples_seen": 496640,
          "online_loss": 0.00010725685433499166,
          "train_loss": 0.0001055042739608325,
          "validation_loss": 9.314039198216051e-05,
          "train_metric": 1.0626003131619655e-05,
          "validation_metric": 9.380759365740232e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.501971432999198
        },
        {
          "epoch": 486,
          "updates": 7776,
          "samples_seen": 497664,
          "online_loss": 0.00010887214421018143,
          "train_loss": 0.00010512520384509116,
          "validation_loss": 9.848560875980183e-05,
          "train_metric": 1.058782436302863e-05,
          "validation_metric": 9.919109288603067e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.512697250000201
        },
        {
          "epoch": 487,
          "updates": 7792,
          "samples_seen": 498688,
          "online_loss": 0.00010653244680725038,
          "train_loss": 0.00010272075451212004,
          "validation_loss": 9.231166040990502e-05,
          "train_metric": 1.0345657756261062e-05,
          "validation_metric": 9.297292308474425e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.523617201999514
        },
        {
          "epoch": 488,
          "updates": 7808,
          "samples_seen": 499712,
          "online_loss": 0.00010644353642419446,
          "train_loss": 0.00010294251114828512,
          "validation_loss": 9.227760892827064e-05,
          "train_metric": 1.03679922176525e-05,
          "validation_metric": 9.29386260395404e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.534862960999817
        },
        {
          "epoch": 489,
          "updates": 7824,
          "samples_seen": 500736,
          "online_loss": 0.00010563337991698063,
          "train_loss": 0.00010244866280118003,
          "validation_loss": 9.410978964297101e-05,
          "train_metric": 1.0318253771401942e-05,
          "validation_metric": 9.478393621975556e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.545448544000465
        },
        {
          "epoch": 490,
          "updates": 7840,
          "samples_seen": 501760,
          "online_loss": 0.00010804066732816864,
          "train_loss": 0.00010374842531746253,
          "validation_loss": 9.218131162924692e-05,
          "train_metric": 1.0449160072312225e-05,
          "validation_metric": 9.284163752454333e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.5564821120005945
        },
        {
          "epoch": 491,
          "updates": 7856,
          "samples_seen": 502784,
          "online_loss": 0.00010698561754907132,
          "train_loss": 0.0001047200639732182,
          "validation_loss": 9.678597416495904e-05,
          "train_metric": 1.054702079272829e-05,
          "validation_metric": 9.747928743308876e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.567205253999418
        },
        {
          "epoch": 492,
          "updates": 7872,
          "samples_seen": 503808,
          "online_loss": 0.00010809798823174788,
          "train_loss": 0.00010331910016248003,
          "validation_loss": 9.56089497776702e-05,
          "train_metric": 1.0405921784695238e-05,
          "validation_metric": 9.629382475395687e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.577826999999161
        },
        {
          "epoch": 493,
          "updates": 7888,
          "samples_seen": 504832,
          "online_loss": 0.00010803436111928022,
          "train_loss": 0.00012274495384190232,
          "validation_loss": 0.00010142864630324766,
          "train_metric": 1.2362424058665056e-05,
          "validation_metric": 1.0215519068879075e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.588329156998952
        },
        {
          "epoch": 494,
          "updates": 7904,
          "samples_seen": 505856,
          "online_loss": 0.00011516721679072361,
          "train_loss": 0.00010425388609291986,
          "validation_loss": 9.879092976916581e-05,
          "train_metric": 1.0500069947738666e-05,
          "validation_metric": 9.949861123459414e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.598562473000129
        },
        {
          "epoch": 495,
          "updates": 7920,
          "samples_seen": 506880,
          "online_loss": 0.00010308935179637047,
          "train_loss": 0.00010522016964387149,
          "validation_loss": 9.188994590658695e-05,
          "train_metric": 1.0597389518807177e-05,
          "validation_metric": 9.254817996406928e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.609715693999533
        },
        {
          "epoch": 496,
          "updates": 7936,
          "samples_seen": 507904,
          "online_loss": 0.00010597099321785208,
          "train_loss": 0.00010179495438933372,
          "validation_loss": 9.324699931312352e-05,
          "train_metric": 1.02524154499406e-05,
          "validation_metric": 9.391495950694662e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.620893367999088
        },
        {
          "epoch": 497,
          "updates": 7952,
          "samples_seen": 508928,
          "online_loss": 0.00010410056961518421,
          "train_loss": 0.00010196486255154014,
          "validation_loss": 9.476338163949549e-05,
          "train_metric": 1.0269527592754457e-05,
          "validation_metric": 9.544221029500477e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.631319640000584
        },
        {
          "epoch": 498,
          "updates": 7968,
          "samples_seen": 509952,
          "online_loss": 0.00010681848107196856,
          "train_loss": 0.00010150634625460953,
          "validation_loss": 9.268999565392733e-05,
          "train_metric": 1.0223347089777235e-05,
          "validation_metric": 9.335397407994606e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.642087719999836
        },
        {
          "epoch": 499,
          "updates": 7984,
          "samples_seen": 510976,
          "online_loss": 0.00010704803344196989,
          "train_loss": 0.0001072776794899255,
          "validation_loss": 9.352789493277669e-05,
          "train_metric": 1.0804616977111436e-05,
          "validation_metric": 9.41978669288801e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.652282843000648
        },
        {
          "epoch": 500,
          "updates": 8000,
          "samples_seen": 512000,
          "online_loss": 0.00011447632778072148,
          "train_loss": 0.0001108586584450677,
          "validation_loss": 0.00010584377741906792,
          "train_metric": 1.1165277101099491e-05,
          "validation_metric": 1.0660198313416913e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.662881295000261
        },
        {
          "epoch": 501,
          "updates": 8016,
          "samples_seen": 513024,
          "online_loss": 0.00011612909474933986,
          "train_loss": 0.00010233028297079727,
          "validation_loss": 9.374590445077047e-05,
          "train_metric": 1.03063312053564e-05,
          "validation_metric": 9.44174280448351e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.673244370000248
        },
        {
          "epoch": 502,
          "updates": 8032,
          "samples_seen": 514048,
          "online_loss": 0.00010693050126064918,
          "train_loss": 0.00010397715959697962,
          "validation_loss": 9.382157440995798e-05,
          "train_metric": 1.0472199392097536e-05,
          "validation_metric": 9.44936527957907e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.683585455999491
        },
        {
          "epoch": 503,
          "updates": 8048,
          "samples_seen": 515072,
          "online_loss": 0.0001109381200876669,
          "train_loss": 0.00010894123988691717,
          "validation_loss": 0.00010214520443696529,
          "train_metric": 1.0972162272082642e-05,
          "validation_metric": 1.0287691111443564e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.693943749000027
        },
        {
          "epoch": 504,
          "updates": 8064,
          "samples_seen": 516096,
          "online_loss": 0.00011025294406863395,
          "train_loss": 0.00010293425293639302,
          "validation_loss": 9.57328375079669e-05,
          "train_metric": 1.036716093949508e-05,
          "validation_metric": 9.641861652198713e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.704338768000525
        },
        {
          "epoch": 505,
          "updates": 8080,
          "samples_seen": 517120,
          "online_loss": 0.00010245089288218878,
          "train_loss": 0.00010135486809303984,
          "validation_loss": 9.097545989789069e-05,
          "train_metric": 1.0208092135144398e-05,
          "validation_metric": 9.162716196442489e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.715463815000476
        },
        {
          "epoch": 506,
          "updates": 8096,
          "samples_seen": 518144,
          "online_loss": 0.00010720392219809582,
          "train_loss": 0.00010868359822779894,
          "validation_loss": 0.00010171614121645689,
          "train_metric": 1.0946213478746358e-05,
          "validation_metric": 1.0244477380183525e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.726114912000412
        },
        {
          "epoch": 507,
          "updates": 8112,
          "samples_seen": 519168,
          "online_loss": 0.0001059835715295776,
          "train_loss": 0.00010088373528560624,
          "validation_loss": 9.120201139012352e-05,
          "train_metric": 1.0160641068068799e-05,
          "validation_metric": 9.185532690025866e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.736541046000639
        },
        {
          "epoch": 508,
          "updates": 8128,
          "samples_seen": 520192,
          "online_loss": 0.00011057600568165071,
          "train_loss": 0.00010142941755475476,
          "validation_loss": 9.255680197384208e-05,
          "train_metric": 1.0215599104412831e-05,
          "validation_metric": 9.321981451648753e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.747192941000321
        },
        {
          "epoch": 509,
          "updates": 8144,
          "samples_seen": 521216,
          "online_loss": 0.00010347196757720667,
          "train_loss": 0.00010040045890491456,
          "validation_loss": 9.12891628104262e-05,
          "train_metric": 1.0111965821124613e-05,
          "validation_metric": 9.194311132887378e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.757521018000261
        },
        {
          "epoch": 510,
          "updates": 8160,
          "samples_seen": 522240,
          "online_loss": 0.00010164140758206486,
          "train_loss": 0.00010119763464899734,
          "validation_loss": 9.418631088919938e-05,
          "train_metric": 1.0192254194407724e-05,
          "validation_metric": 9.48610068007838e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.767751326999132
        },
        {
          "epoch": 511,
          "updates": 8176,
          "samples_seen": 523264,
          "online_loss": 0.00010621185492709628,
          "train_loss": 0.00010438223398523405,
          "validation_loss": 9.742302790982649e-05,
          "train_metric": 1.0512996595934965e-05,
          "validation_metric": 9.81209086603485e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.837231297000471
        },
        {
          "epoch": 512,
          "updates": 8192,
          "samples_seen": 524288,
          "online_loss": 0.00010405734110463527,
          "train_loss": 0.00010125331027666107,
          "validation_loss": 9.099159797187895e-05,
          "train_metric": 1.019786395772826e-05,
          "validation_metric": 9.164340553979855e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.848131349999676
        },
        {
          "epoch": 513,
          "updates": 8208,
          "samples_seen": 525312,
          "online_loss": 0.00010459895338499337,
          "train_loss": 0.00010711253707995638,
          "validation_loss": 9.886400948744267e-05,
          "train_metric": 1.078798231901601e-05,
          "validation_metric": 9.957222573575564e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.858606088999295
        },
        {
          "epoch": 514,
          "updates": 8224,
          "samples_seen": 526336,
          "online_loss": 0.00010622151148709236,
          "train_loss": 0.00010016241139965132,
          "validation_loss": 9.100131137529388e-05,
          "train_metric": 1.008799245028058e-05,
          "validation_metric": 9.165319170278963e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.868946982000125
        },
        {
          "epoch": 515,
          "updates": 8240,
          "samples_seen": 527360,
          "online_loss": 0.00010198088557444862,
          "train_loss": 0.00010010071855504066,
          "validation_loss": 9.163623326458037e-05,
          "train_metric": 1.0081776963488664e-05,
          "validation_metric": 9.229266652255319e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.879440774999239
        },
        {
          "epoch": 516,
          "updates": 8256,
          "samples_seen": 528384,
          "online_loss": 0.00010388677901573828,
          "train_loss": 0.00010047670366475359,
          "validation_loss": 9.348355524707586e-05,
          "train_metric": 1.0119645594386384e-05,
          "validation_metric": 9.415321983397007e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.889894395999363
        },
        {
          "epoch": 517,
          "updates": 8272,
          "samples_seen": 529408,
          "online_loss": 0.00010573886584097636,
          "train_loss": 0.00010348354408051819,
          "validation_loss": 9.25205895327963e-05,
          "train_metric": 1.0422482773719821e-05,
          "validation_metric": 9.318335287389345e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.9001030340004945
        },
        {
          "epoch": 518,
          "updates": 8288,
          "samples_seen": 530432,
          "online_loss": 0.0001069394111254951,
          "train_loss": 9.96118615148589e-05,
          "validation_loss": 9.019606659421697e-05,
          "train_metric": 1.0032540558313485e-05,
          "validation_metric": 9.084216799237765e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.911562295999829
        },
        {
          "epoch": 519,
          "updates": 8304,
          "samples_seen": 531456,
          "online_loss": 0.00010514601626709918,
          "train_loss": 0.00010128999565495178,
          "validation_loss": 9.080387826543301e-05,
          "train_metric": 1.0201558325206861e-05,
          "validation_metric": 9.145434887614101e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.9220798240003205
        },
        {
          "epoch": 520,
          "updates": 8320,
          "samples_seen": 532480,
          "online_loss": 0.0001045983290168806,
          "train_loss": 0.00010028993710875511,
          "validation_loss": 9.331786714028567e-05,
          "train_metric": 1.0100835424964316e-05,
          "validation_metric": 9.398634574608877e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.932470960999126
        },
        {
          "epoch": 521,
          "updates": 8336,
          "samples_seen": 533504,
          "online_loss": 0.00010150901744054863,
          "train_loss": 9.985746146412566e-05,
          "validation_loss": 9.036287519847974e-05,
          "train_metric": 1.0057276995212305e-05,
          "validation_metric": 9.101017894863617e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.943134851999275
        },
        {
          "epoch": 522,
          "updates": 8352,
          "samples_seen": 534528,
          "online_loss": 0.00010141603979718639,
          "train_loss": 0.00010194123024120927,
          "validation_loss": 9.464554750593379e-05,
          "train_metric": 1.0267147445119917e-05,
          "validation_metric": 9.532353033137042e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.953425985999274
        },
        {
          "epoch": 523,
          "updates": 8368,
          "samples_seen": 535552,
          "online_loss": 0.00010057012468678295,
          "train_loss": 0.00010270498023601249,
          "validation_loss": 8.980598795460537e-05,
          "train_metric": 1.0344069778511766e-05,
          "validation_metric": 9.044930266099982e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.964153356999304
        },
        {
          "epoch": 524,
          "updates": 8384,
          "samples_seen": 536576,
          "online_loss": 0.00010359580096519494,
          "train_loss": 9.899325959850103e-05,
          "validation_loss": 9.127054363489151e-05,
          "train_metric": 9.970238352252636e-06,
          "validation_metric": 9.192433935822919e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.974820371999158
        },
        {
          "epoch": 525,
          "updates": 8400,
          "samples_seen": 537600,
          "online_loss": 0.00010383131848357152,
          "train_loss": 0.0001061859802575782,
          "validation_loss": 9.840492566581815e-05,
          "train_metric": 1.0694663615140598e-05,
          "validation_metric": 9.91098477243213e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.985031670999888
        },
        {
          "epoch": 526,
          "updates": 8416,
          "samples_seen": 538624,
          "online_loss": 0.0001009628704196075,
          "train_loss": 0.00010064274101750925,
          "validation_loss": 9.028003842104226e-05,
          "train_metric": 1.0136367563973181e-05,
          "validation_metric": 9.092675099964254e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 5.995322578999549
        },
        {
          "epoch": 527,
          "updates": 8432,
          "samples_seen": 539648,
          "online_loss": 0.00010221190996162477,
          "train_loss": 9.867782500805333e-05,
          "validation_loss": 9.018101991387084e-05,
          "train_metric": 9.938468792825006e-06,
          "validation_metric": 9.082701581064612e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.005613094999717
        },
        {
          "epoch": 528,
          "updates": 8448,
          "samples_seen": 540672,
          "online_loss": 0.00010040198458227678,
          "train_loss": 9.864993626251817e-05,
          "validation_loss": 9.060287266038358e-05,
          "train_metric": 9.93565936369123e-06,
          "validation_metric": 9.125189535552636e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.0159641390000616
        },
        {
          "epoch": 529,
          "updates": 8464,
          "samples_seen": 541696,
          "online_loss": 0.00010067894208987127,
          "train_loss": 9.897271229419857e-05,
          "validation_loss": 9.127683733822778e-05,
          "train_metric": 9.968169251806103e-06,
          "validation_metric": 9.193068763124757e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.026196769998933
        },
        {
          "epoch": 530,
          "updates": 8480,
          "samples_seen": 542720,
          "online_loss": 0.00010170910582019133,
          "train_loss": 9.870893700281158e-05,
          "validation_loss": 9.112090629059821e-05,
          "train_metric": 9.941602002072614e-06,
          "validation_metric": 9.177363608614542e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.036682863999886
        },
        {
          "epoch": 531,
          "updates": 8496,
          "samples_seen": 543744,
          "online_loss": 9.920869206325733e-05,
          "train_loss": 9.832634532358497e-05,
          "validation_loss": 8.931777847465128e-05,
          "train_metric": 9.903067621053196e-06,
          "validation_metric": 8.995760254038032e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.047451465999984
        },
        {
          "epoch": 532,
          "updates": 8512,
          "samples_seen": 544768,
          "online_loss": 9.995947993957088e-05,
          "train_loss": 9.817755926633254e-05,
          "validation_loss": 8.99770820979029e-05,
          "train_metric": 9.88808460533619e-06,
          "validation_metric": 9.062163371709175e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.057926385999963
        },
        {
          "epoch": 533,
          "updates": 8528,
          "samples_seen": 545792,
          "online_loss": 0.00010023098161582311,
          "train_loss": 9.799536201171577e-05,
          "validation_loss": 9.017012780532241e-05,
          "train_metric": 9.869734640233219e-06,
          "validation_metric": 9.081605639948975e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.0683791499996005
        },
        {
          "epoch": 534,
          "updates": 8544,
          "samples_seen": 546816,
          "online_loss": 0.00010022064816439524,
          "train_loss": 9.80255936156027e-05,
          "validation_loss": 8.986119064502418e-05,
          "train_metric": 9.872778719000053e-06,
          "validation_metric": 9.050490916706622e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.078870470999391
        },
        {
          "epoch": 535,
          "updates": 8560,
          "samples_seen": 547840,
          "online_loss": 9.858073826762848e-05,
          "train_loss": 0.00010239967377856374,
          "validation_loss": 9.685788245406002e-05,
          "train_metric": 1.0313320672139525e-05,
          "validation_metric": 9.755171049619094e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.0890091080000275
        },
        {
          "epoch": 536,
          "updates": 8576,
          "samples_seen": 548864,
          "online_loss": 0.0001013692885862838,
          "train_loss": 9.794013749342412e-05,
          "validation_loss": 8.892365440260619e-05,
          "train_metric": 9.864172170637175e-06,
          "validation_metric": 8.956065357779153e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.09988771099961
        },
        {
          "epoch": 537,
          "updates": 8592,
          "samples_seen": 549888,
          "online_loss": 0.00010401666531834053,
          "train_loss": 0.00010080044012283906,
          "validation_loss": 9.552272967994213e-05,
          "train_metric": 1.0152251888939645e-05,
          "validation_metric": 9.62069952947786e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.110577335999551
        },
        {
          "epoch": 538,
          "updates": 8608,
          "samples_seen": 550912,
          "online_loss": 0.00010535839533076796,
          "train_loss": 0.00010950622527161613,
          "validation_loss": 9.793914068723097e-05,
          "train_metric": 1.1029065717593767e-05,
          "validation_metric": 9.86407212621998e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.12096119600028
        },
        {
          "epoch": 539,
          "updates": 8624,
          "samples_seen": 551936,
          "online_loss": 0.00010134273770745494,
          "train_loss": 9.971038525691256e-05,
          "validation_loss": 8.928304305300117e-05,
          "train_metric": 1.0042464964499231e-05,
          "validation_metric": 8.992261427920312e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.131253891000597
        },
        {
          "epoch": 540,
          "updates": 8640,
          "samples_seen": 552960,
          "online_loss": 0.00010281663981004385,
          "train_loss": 9.835743549047038e-05,
          "validation_loss": 8.919847459765151e-05,
          "train_metric": 9.906200830300804e-06,
          "validation_metric": 8.983744010038208e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.141737926000133
        },
        {
          "epoch": 541,
          "updates": 8656,
          "samples_seen": 553984,
          "online_loss": 9.940374434336263e-05,
          "train_loss": 9.844476880971342e-05,
          "validation_loss": 9.100844181375578e-05,
          "train_metric": 9.91499655356165e-06,
          "validation_metric": 9.166036761598662e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.1520393600003445
        },
        {
          "epoch": 542,
          "updates": 8672,
          "samples_seen": 555008,
          "online_loss": 0.00010150862044611131,
          "train_loss": 9.736217907629907e-05,
          "validation_loss": 9.055961709236726e-05,
          "train_metric": 9.805962690734304e-06,
          "validation_metric": 9.120833055931143e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.162337805999414
        },
        {
          "epoch": 543,
          "updates": 8688,
          "samples_seen": 556032,
          "online_loss": 0.00010181917105001048,
          "train_loss": 9.866042819339782e-05,
          "validation_loss": 9.171372221317142e-05,
          "train_metric": 9.936718015524093e-06,
          "validation_metric": 9.237069207301829e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.172637838999435
        },
        {
          "epoch": 544,
          "updates": 8704,
          "samples_seen": 557056,
          "online_loss": 0.00010808131173689617,
          "train_loss": 0.00011369206185918301,
          "validation_loss": 0.00010787132487166673,
          "train_metric": 1.1450648344180081e-05,
          "validation_metric": 1.0864404430321883e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.183124483999563
        },
        {
          "epoch": 545,
          "updates": 8720,
          "samples_seen": 558080,
          "online_loss": 0.00010030527846538462,
          "train_loss": 0.00011505576549097896,
          "validation_loss": 0.00010002008639276028,
          "train_metric": 1.1587994777073618e-05,
          "validation_metric": 1.0073657904285938e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.193338314000357
        },
        {
          "epoch": 546,
          "updates": 8736,
          "samples_seen": 559104,
          "online_loss": 0.00010792993634822778,
          "train_loss": 0.00010298000415787101,
          "validation_loss": 9.95450172922574e-05,
          "train_metric": 1.0371767530159559e-05,
          "validation_metric": 1.0025809388025664e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.203741966999587
        },
        {
          "epoch": 547,
          "updates": 8752,
          "samples_seen": 560128,
          "online_loss": 0.00010175683246416156,
          "train_loss": 9.684282122179866e-05,
          "validation_loss": 9.069561929209158e-05,
          "train_metric": 9.753654012456536e-06,
          "validation_metric": 9.134530955634546e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.273401551999996
        },
        {
          "epoch": 548,
          "updates": 8768,
          "samples_seen": 561152,
          "online_loss": 0.00010144647603738122,
          "train_loss": 9.752507321536541e-05,
          "validation_loss": 8.814580360194668e-05,
          "train_metric": 9.822368156164885e-06,
          "validation_metric": 8.877721484168433e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.284783057999448
        },
        {
          "epoch": 549,
          "updates": 8784,
          "samples_seen": 562176,
          "online_loss": 0.00010067629477816809,
          "train_loss": 0.00010195942741120234,
          "validation_loss": 9.827107714954764e-05,
          "train_metric": 1.026898007694399e-05,
          "validation_metric": 9.89750333246775e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.295221332999063
        },
        {
          "epoch": 550,
          "updates": 8800,
          "samples_seen": 563200,
          "online_loss": 0.00010539352251726086,
          "train_loss": 9.781964763533324e-05,
          "validation_loss": 8.81667438079603e-05,
          "train_metric": 9.852035873336717e-06,
          "validation_metric": 8.879832421371248e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.305889526000101
        },
        {
          "epoch": 551,
          "updates": 8816,
          "samples_seen": 564224,
          "online_loss": 9.688933346296835e-05,
          "train_loss": 9.945352940121666e-05,
          "validation_loss": 9.241479710908607e-05,
          "train_metric": 1.0016596206696704e-05,
          "validation_metric": 9.307679647463374e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.316786338999009
        },
        {
          "epoch": 552,
          "updates": 8832,
          "samples_seen": 565248,
          "online_loss": 0.00010295770425727824,
          "train_loss": 9.806764137465507e-05,
          "validation_loss": 9.107633377425373e-05,
          "train_metric": 9.877013326331507e-06,
          "validation_metric": 9.172876161755994e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.32733549600016
        },
        {
          "epoch": 553,
          "updates": 8848,
          "samples_seen": 566272,
          "online_loss": 9.899064934870694e-05,
          "train_loss": 9.61022888077423e-05,
          "validation_loss": 9.018786658998579e-05,
          "train_metric": 9.679069989942946e-06,
          "validation_metric": 9.083390978048556e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.337634354998954
        },
        {
          "epoch": 554,
          "updates": 8864,
          "samples_seen": 567296,
          "online_loss": 9.814962368182023e-05,
          "train_loss": 9.598011820344254e-05,
          "validation_loss": 8.916029764804989e-05,
          "train_metric": 9.666766345617361e-06,
          "validation_metric": 8.979898666439112e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.348038070000257
        },
        {
          "epoch": 555,
          "updates": 8880,
          "samples_seen": 568320,
          "online_loss": 0.0001034854901718063,
          "train_loss": 0.00010107785055879503,
          "validation_loss": 9.405497985426337e-05,
          "train_metric": 1.0180190656683408e-05,
          "validation_metric": 9.472872989135794e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.358325691999198
        },
        {
          "epoch": 556,
          "updates": 8896,
          "samples_seen": 569344,
          "online_loss": 9.805879335544887e-05,
          "train_loss": 9.764580318005756e-05,
          "validation_loss": 8.832932508084923e-05,
          "train_metric": 9.834529919316992e-06,
          "validation_metric": 8.896206963981967e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.368909083999824
        },
        {
          "epoch": 557,
          "updates": 8912,
          "samples_seen": 570368,
          "online_loss": 0.00010220864737675583,
          "train_loss": 0.00010313940583728254,
          "validation_loss": 8.923090354073793e-05,
          "train_metric": 1.0387823749624658e-05,
          "validation_metric": 8.987010005512275e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.379340138999396
        },
        {
          "epoch": 558,
          "updates": 8928,
          "samples_seen": 571392,
          "online_loss": 0.0001029570323680673,
          "train_loss": 0.0001105895935324952,
          "validation_loss": 0.0001058492052834481,
          "train_metric": 1.1138179615954868e-05,
          "validation_metric": 1.0660744919732679e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.389735165999809
        },
        {
          "epoch": 559,
          "updates": 8944,
          "samples_seen": 572416,
          "online_loss": 0.0001069367453965242,
          "train_loss": 0.00010156636562896892,
          "validation_loss": 9.190574928652495e-05,
          "train_metric": 1.022939250105992e-05,
          "validation_metric": 9.256411431124434e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.4002944650001155
        },
        {
          "epoch": 560,
          "updates": 8960,
          "samples_seen": 573440,
          "online_loss": 0.00010258043857902521,
          "train_loss": 9.647091792430729e-05,
          "validation_loss": 8.761198114370927e-05,
          "train_metric": 9.71619738265872e-06,
          "validation_metric": 8.823957614367828e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.411408971000128
        },
        {
          "epoch": 561,
          "updates": 8976,
          "samples_seen": 574464,
          "online_loss": 9.921843957272358e-05,
          "train_loss": 0.00010446626401972026,
          "validation_loss": 9.578294702805579e-05,
          "train_metric": 1.0521460353629664e-05,
          "validation_metric": 9.64690752880415e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.421858538000379
        },
        {
          "epoch": 562,
          "updates": 8992,
          "samples_seen": 575488,
          "online_loss": 0.0001023905206238851,
          "train_loss": 9.572540875524282e-05,
          "validation_loss": 8.828820864437148e-05,
          "train_metric": 9.641113138059154e-06,
          "validation_metric": 8.892065125110094e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.432174162000592
        },
        {
          "epoch": 563,
          "updates": 9008,
          "samples_seen": 576512,
          "online_loss": 9.883723123493837e-05,
          "train_loss": 9.722713730297983e-05,
          "validation_loss": 8.89406364876777e-05,
          "train_metric": 9.792360287974589e-06,
          "validation_metric": 8.957775207818486e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.44324378499914
        },
        {
          "epoch": 564,
          "updates": 9024,
          "samples_seen": 577536,
          "online_loss": 0.00010004652085626731,
          "train_loss": 0.0001047647456289269,
          "validation_loss": 9.627699182601646e-05,
          "train_metric": 1.0551522791502066e-05,
          "validation_metric": 9.696665074443445e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.453898858000684
        },
        {
          "epoch": 565,
          "updates": 9040,
          "samples_seen": 578560,
          "online_loss": 9.855516964307753e-05,
          "train_loss": 0.00011121675197500736,
          "validation_loss": 9.649179992265999e-05,
          "train_metric": 1.1201344022992998e-05,
          "validation_metric": 9.71830104390392e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.464353797000513
        },
        {
          "epoch": 566,
          "updates": 9056,
          "samples_seen": 579584,
          "online_loss": 0.00010216158443654422,
          "train_loss": 0.00010590526653686538,
          "validation_loss": 0.00010387977818027139,
          "train_metric": 1.0666390153346583e-05,
          "validation_metric": 1.0462390491738915e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.474783350000507
        },
        {
          "epoch": 567,
          "updates": 9072,
          "samples_seen": 580608,
          "online_loss": 0.00010066624054161366,
          "train_loss": 9.61901678238064e-05,
          "validation_loss": 8.794481982477009e-05,
          "train_metric": 9.6879211923806e-06,
          "validation_metric": 8.857481589075178e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.485066619999998
        },
        {
          "epoch": 568,
          "updates": 9088,
          "samples_seen": 581632,
          "online_loss": 9.778992580322665e-05,
          "train_loss": 9.861097350949422e-05,
          "validation_loss": 9.515253623249009e-05,
          "train_metric": 9.931735803547781e-06,
          "validation_metric": 9.58341479417868e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.495366428000125
        },
        {
          "epoch": 569,
          "updates": 9104,
          "samples_seen": 582656,
          "online_loss": 9.607474339645705e-05,
          "train_loss": 0.00010666450543794781,
          "validation_loss": 9.375894296681508e-05,
          "train_metric": 1.074285955837695e-05,
          "validation_metric": 9.443057024327572e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.50580296499902
        },
        {
          "epoch": 570,
          "updates": 9120,
          "samples_seen": 583680,
          "online_loss": 0.00010701429755499703,
          "train_loss": 9.527727524982765e-05,
          "validation_loss": 8.938656537793577e-05,
          "train_metric": 9.59597855398897e-06,
          "validation_metric": 9.002687875181437e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.516171981000298
        },
        {
          "epoch": 571,
          "updates": 9136,
          "samples_seen": 584704,
          "online_loss": 9.79914627805556e-05,
          "train_loss": 9.501020394964144e-05,
          "validation_loss": 8.896159852156416e-05,
          "train_metric": 9.569081157678738e-06,
          "validation_metric": 8.9598861450213e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.52644669700021
        },
        {
          "epoch": 572,
          "updates": 9152,
          "samples_seen": 585728,
          "online_loss": 0.00010367515164944052,
          "train_loss": 9.85555088846013e-05,
          "validation_loss": 8.887617150321603e-05,
          "train_metric": 9.926149687089492e-06,
          "validation_metric": 8.95128323463723e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.536708432000523
        },
        {
          "epoch": 573,
          "updates": 9168,
          "samples_seen": 586752,
          "online_loss": 9.866493746812921e-05,
          "train_loss": 9.69899192568846e-05,
          "validation_loss": 8.637731662020087e-05,
          "train_metric": 9.768469681148417e-06,
          "validation_metric": 8.699606951267924e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.547595263999028
        },
        {
          "epoch": 574,
          "updates": 9184,
          "samples_seen": 587776,
          "online_loss": 9.735904268382001e-05,
          "train_loss": 9.45683495956473e-05,
          "validation_loss": 8.877043728716671e-05,
          "train_metric": 9.524577762931585e-06,
          "validation_metric": 8.94063305167947e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.558434886999748
        },
        {
          "epoch": 575,
          "updates": 9200,
          "samples_seen": 588800,
          "online_loss": 9.625386223888199e-05,
          "train_loss": 9.422909351997077e-05,
          "validation_loss": 8.665712812216952e-05,
          "train_metric": 9.490408956480678e-06,
          "validation_metric": 8.727787644602358e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.5689617349999025
        },
        {
          "epoch": 576,
          "updates": 9216,
          "samples_seen": 589824,
          "online_loss": 9.950365119948401e-05,
          "train_loss": 9.499929001322016e-05,
          "validation_loss": 8.839608199195936e-05,
          "train_metric": 9.567980669089593e-06,
          "validation_metric": 8.902930858312175e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.5794591729991225
        },
        {
          "epoch": 577,
          "updates": 9232,
          "samples_seen": 590848,
          "online_loss": 9.572347698849626e-05,
          "train_loss": 9.621112258173525e-05,
          "validation_loss": 8.799929491942748e-05,
          "train_metric": 9.690032129583415e-06,
          "validation_metric": 8.86296675162157e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.58970245799901
        },
        {
          "epoch": 578,
          "updates": 9248,
          "samples_seen": 591872,
          "online_loss": 9.778113371794461e-05,
          "train_loss": 9.530009992886335e-05,
          "validation_loss": 9.078234143089503e-05,
          "train_metric": 9.598278666089755e-06,
          "validation_metric": 9.143266652245075e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.599943613000505
        },
        {
          "epoch": 579,
          "updates": 9264,
          "samples_seen": 592896,
          "online_loss": 0.00010134133981409832,
          "train_loss": 9.90337721304968e-05,
          "validation_loss": 8.804292156128213e-05,
          "train_metric": 9.97431834548479e-06,
          "validation_metric": 8.867360520525835e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.610230860000229
        },
        {
          "epoch": 580,
          "updates": 9280,
          "samples_seen": 593920,
          "online_loss": 0.00010766606919787591,
          "train_loss": 0.00010308697528671473,
          "validation_loss": 9.098228474613279e-05,
          "train_metric": 1.0382542313891463e-05,
          "validation_metric": 9.163401955447625e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.620523466999657
        },
        {
          "epoch": 581,
          "updates": 9296,
          "samples_seen": 594944,
          "online_loss": 9.666137088970572e-05,
          "train_loss": 0.00010277461842633784,
          "validation_loss": 9.64072096394375e-05,
          "train_metric": 1.035108380165184e-05,
          "validation_metric": 9.709781807032414e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.6306879879994085
        },
        {
          "epoch": 582,
          "updates": 9312,
          "samples_seen": 595968,
          "online_loss": 0.00011126636945846258,
          "train_loss": 9.574947034707293e-05,
          "validation_loss": 8.893033373169601e-05,
          "train_metric": 9.643535122449975e-06,
          "validation_metric": 8.956737474363763e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.641189064999708
        },
        {
          "epoch": 583,
          "updates": 9328,
          "samples_seen": 596992,
          "online_loss": 0.0001192574113701994,
          "train_loss": 0.00010847657540580258,
          "validation_loss": 9.419993148185313e-05,
          "train_metric": 1.0925363312708214e-05,
          "validation_metric": 9.487472198088653e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.7114124429990625
        },
        {
          "epoch": 584,
          "updates": 9344,
          "samples_seen": 598016,
          "online_loss": 0.00010653466574694903,
          "train_loss": 9.506334754405543e-05,
          "validation_loss": 8.89756774995476e-05,
          "train_metric": 9.574431715009268e-06,
          "validation_metric": 8.961304047261365e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.722240174000035
        },
        {
          "epoch": 585,
          "updates": 9360,
          "samples_seen": 599040,
          "online_loss": 9.588805414750823e-05,
          "train_loss": 9.371625492349267e-05,
          "validation_loss": 8.56920814840123e-05,
          "train_metric": 9.438757842872292e-06,
          "validation_metric": 8.630593583802693e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.7330477479990805
        },
        {
          "epoch": 586,
          "updates": 9376,
          "samples_seen": 600064,
          "online_loss": 9.74383885932184e-05,
          "train_loss": 9.748473530635238e-05,
          "validation_loss": 8.83886605151929e-05,
          "train_metric": 9.818304533837363e-06,
          "validation_metric": 8.902182344172616e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.743624537999494
        },
        {
          "epoch": 587,
          "updates": 9392,
          "samples_seen": 601088,
          "online_loss": 9.616335864848224e-05,
          "train_loss": 9.912134555634111e-05,
          "validation_loss": 9.225002577295527e-05,
          "train_metric": 9.983138625102583e-06,
          "validation_metric": 9.291085007134825e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.753915770999811
        },
        {
          "epoch": 588,
          "updates": 9408,
          "samples_seen": 602112,
          "online_loss": 9.755278438206005e-05,
          "train_loss": 9.315356146544218e-05,
          "validation_loss": 8.599454304203391e-05,
          "train_metric": 9.382084499520715e-06,
          "validation_metric": 8.661055289849173e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.764190017000146
        },
        {
          "epoch": 589,
          "updates": 9424,
          "samples_seen": 603136,
          "online_loss": 9.600600742487586e-05,
          "train_loss": 9.29896195884794e-05,
          "validation_loss": 8.553377847420052e-05,
          "train_metric": 9.36557353270473e-06,
          "validation_metric": 8.61464832269121e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.775088828999287
        },
        {
          "epoch": 590,
          "updates": 9440,
          "samples_seen": 604160,
          "online_loss": 9.688453747003223e-05,
          "train_loss": 9.401902934769168e-05,
          "validation_loss": 8.908303425414488e-05,
          "train_metric": 9.469253200222738e-06,
          "validation_metric": 8.972116120276041e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.785494689000188
        },
        {
          "epoch": 591,
          "updates": 9456,
          "samples_seen": 605184,
          "online_loss": 9.5112835879263e-05,
          "train_loss": 9.319496166426688e-05,
          "validation_loss": 8.574346429668367e-05,
          "train_metric": 9.386255442223046e-06,
          "validation_metric": 8.635767699161079e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.795747082000162
        },
        {
          "epoch": 592,
          "updates": 9472,
          "samples_seen": 606208,
          "online_loss": 0.00010179061268900114,
          "train_loss": 9.433962986804545e-05,
          "validation_loss": 9.015479008667171e-05,
          "train_metric": 9.501541171630379e-06,
          "validation_metric": 9.080059498955961e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.806125204000637
        },
        {
          "epoch": 593,
          "updates": 9488,
          "samples_seen": 607232,
          "online_loss": 0.00010606208797980798,
          "train_loss": 9.490944648860022e-05,
          "validation_loss": 8.595399413025007e-05,
          "train_metric": 9.558931196806952e-06,
          "validation_metric": 8.656970749143511e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.816535003999888
        },
        {
          "epoch": 594,
          "updates": 9504,
          "samples_seen": 608256,
          "online_loss": 9.567920915287687e-05,
          "train_loss": 9.642603254178539e-05,
          "validation_loss": 9.15667405934073e-05,
          "train_metric": 9.711677193990909e-06,
          "validation_metric": 9.222267181030475e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.82682800099974
        },
        {
          "epoch": 595,
          "updates": 9520,
          "samples_seen": 609280,
          "online_loss": 9.499190946371527e-05,
          "train_loss": 9.67451196629554e-05,
          "validation_loss": 8.688584784977138e-05,
          "train_metric": 9.743814189278055e-06,
          "validation_metric": 8.750824235903565e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.837108182999145
        },
        {
          "epoch": 596,
          "updates": 9536,
          "samples_seen": 610304,
          "online_loss": 9.779412562238576e-05,
          "train_loss": 9.238478378392756e-05,
          "validation_loss": 8.610863005742431e-05,
          "train_metric": 9.304655577579979e-06,
          "validation_metric": 8.672546755406074e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.847691073999158
        },
        {
          "epoch": 597,
          "updates": 9552,
          "samples_seen": 611328,
          "online_loss": 9.792150012799539e-05,
          "train_loss": 9.647085244068876e-05,
          "validation_loss": 9.047833736985922e-05,
          "train_metric": 9.716190106701106e-06,
          "validation_metric": 9.112645784625784e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.857926401000441
        },
        {
          "epoch": 598,
          "updates": 9568,
          "samples_seen": 612352,
          "online_loss": 9.832724276748195e-05,
          "train_loss": 9.64123901212588e-05,
          "validation_loss": 8.61419175635092e-05,
          "train_metric": 9.710303856991231e-06,
          "validation_metric": 8.675898243382107e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.86850357000003
        },
        {
          "epoch": 599,
          "updates": 9584,
          "samples_seen": 613376,
          "online_loss": 0.00010107871139553026,
          "train_loss": 9.340104588773102e-05,
          "validation_loss": 8.474675996694714e-05,
          "train_metric": 9.407011930306908e-06,
          "validation_metric": 8.535384040442295e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.8795814080003765
        },
        {
          "epoch": 600,
          "updates": 9600,
          "samples_seen": 614400,
          "online_loss": 9.835233800004062e-05,
          "train_loss": 9.327505540568382e-05,
          "validation_loss": 8.941457053879276e-05,
          "train_metric": 9.39432175073307e-06,
          "validation_metric": 9.005507308756933e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.890133356999286
        },
        {
          "epoch": 601,
          "updates": 9616,
          "samples_seen": 615424,
          "online_loss": 9.491272658124217e-05,
          "train_loss": 9.536056313663721e-05,
          "validation_loss": 8.857890497893095e-05,
          "train_metric": 9.60436591412872e-06,
          "validation_metric": 8.921343578549568e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.900442584999837
        },
        {
          "epoch": 602,
          "updates": 9632,
          "samples_seen": 616448,
          "online_loss": 9.947880243998952e-05,
          "train_loss": 9.197786130243912e-05,
          "validation_loss": 8.525687735527754e-05,
          "train_metric": 9.26367374631809e-06,
          "validation_metric": 8.586759577156045e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.911025728999448
        },
        {
          "epoch": 603,
          "updates": 9648,
          "samples_seen": 617472,
          "online_loss": 9.698446774564218e-05,
          "train_loss": 0.0001021199714159593,
          "validation_loss": 8.94930271897465e-05,
          "train_metric": 1.028514998324681e-05,
          "validation_metric": 9.013409908220638e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.92129635099991
        },
        {
          "epoch": 604,
          "updates": 9664,
          "samples_seen": 618496,
          "online_loss": 0.00010083912684422103,
          "train_loss": 9.345503349322826e-05,
          "validation_loss": 8.754737791605294e-05,
          "train_metric": 9.412448889634106e-06,
          "validation_metric": 8.817451089271344e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.931661131999135
        },
        {
          "epoch": 605,
          "updates": 9680,
          "samples_seen": 619520,
          "online_loss": 9.865141328191385e-05,
          "train_loss": 9.831535862758756e-05,
          "validation_loss": 9.282438986701891e-05,
          "train_metric": 9.901962584990542e-06,
          "validation_metric": 9.34893250814639e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.942045223000605
        },
        {
          "epoch": 606,
          "updates": 9696,
          "samples_seen": 620544,
          "online_loss": 9.679715640231734e-05,
          "train_loss": 9.178449545288458e-05,
          "validation_loss": 8.472849003737792e-05,
          "train_metric": 9.244198736269027e-06,
          "validation_metric": 8.533544132660609e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.952957450999747
        },
        {
          "epoch": 607,
          "updates": 9712,
          "samples_seen": 621568,
          "online_loss": 9.304542027166463e-05,
          "train_loss": 9.19165977393277e-05,
          "validation_loss": 8.493797940900549e-05,
          "train_metric": 9.257503734261263e-06,
          "validation_metric": 8.554642590752337e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.963285334999455
        },
        {
          "epoch": 608,
          "updates": 9728,
          "samples_seen": 622592,
          "online_loss": 9.829126747717964e-05,
          "train_loss": 0.00010627792653394863,
          "validation_loss": 9.748193406267092e-05,
          "train_metric": 1.070392409019405e-05,
          "validation_metric": 9.818024409469217e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.973983361000137
        },
        {
          "epoch": 609,
          "updates": 9744,
          "samples_seen": 623616,
          "online_loss": 0.00010472213875800662,
          "train_loss": 9.2691981990356e-05,
          "validation_loss": 8.609172800788656e-05,
          "train_metric": 9.335597496828996e-06,
          "validation_metric": 8.670844181324355e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.984246099998927
        },
        {
          "epoch": 610,
          "updates": 9760,
          "samples_seen": 624640,
          "online_loss": 9.525661789666628e-05,
          "train_loss": 9.142945054918528e-05,
          "validation_loss": 8.490645996062085e-05,
          "train_metric": 9.208438314090017e-06,
          "validation_metric": 8.551467544748448e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 6.994587372999376
        },
        {
          "epoch": 611,
          "updates": 9776,
          "samples_seen": 625664,
          "online_loss": 9.769492567102134e-05,
          "train_loss": 9.184117516269907e-05,
          "validation_loss": 8.547024481231347e-05,
          "train_metric": 9.249906725017354e-06,
          "validation_metric": 8.608250027464237e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.0047981230000005
        },
        {
          "epoch": 612,
          "updates": 9792,
          "samples_seen": 626688,
          "online_loss": 0.00010475537783349864,
          "train_loss": 0.00010191909677814692,
          "validation_loss": 9.708752622827888e-05,
          "train_metric": 1.0264918273605872e-05,
          "validation_metric": 9.778300409379881e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.015251917999194
        },
        {
          "epoch": 613,
          "updates": 9808,
          "samples_seen": 627712,
          "online_loss": 9.842957615546766e-05,
          "train_loss": 9.15390919544734e-05,
          "validation_loss": 8.584016904933378e-05,
          "train_metric": 9.219482308253646e-06,
          "validation_metric": 8.645507477922365e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.025485427999229
        },
        {
          "epoch": 614,
          "updates": 9824,
          "samples_seen": 628736,
          "online_loss": 9.28509598452365e-05,
          "train_loss": 9.202200453728437e-05,
          "validation_loss": 8.501389675075188e-05,
          "train_metric": 9.268118446925655e-06,
          "validation_metric": 8.56228780321544e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.0357393459999
        },
        {
          "epoch": 615,
          "updates": 9840,
          "samples_seen": 629760,
          "online_loss": 9.557660541759105e-05,
          "train_loss": 9.264135587727651e-05,
          "validation_loss": 8.492800407111645e-05,
          "train_metric": 9.330498869530857e-06,
          "validation_metric": 8.55363850860158e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.046177504998923
        },
        {
          "epoch": 616,
          "updates": 9856,
          "samples_seen": 630784,
          "online_loss": 9.732134162732109e-05,
          "train_loss": 9.239280188921839e-05,
          "validation_loss": 8.386473928112537e-05,
          "train_metric": 9.305465027864557e-06,
          "validation_metric": 8.446548235951923e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.057133598998917
        },
        {
          "epoch": 617,
          "updates": 9872,
          "samples_seen": 631808,
          "online_loss": 9.29248133161309e-05,
          "train_loss": 9.623158257454634e-05,
          "validation_loss": 9.28459339775145e-05,
          "train_metric": 9.69209213508293e-06,
          "validation_metric": 9.351101653010119e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.06745683599911
        },
        {
          "epoch": 618,
          "updates": 9888,
          "samples_seen": 632832,
          "online_loss": 0.00010387615293439012,
          "train_loss": 9.084452176466584e-05,
          "validation_loss": 8.47439223434776e-05,
          "train_metric": 9.14952761377208e-06,
          "validation_metric": 8.535097549611237e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.078082041000016
        },
        {
          "epoch": 619,
          "updates": 9904,
          "samples_seen": 633856,
          "online_loss": 9.593064805812901e-05,
          "train_loss": 9.132148261414841e-05,
          "validation_loss": 8.539074042346328e-05,
          "train_metric": 9.19756439543562e-06,
          "validation_metric": 8.600241926615126e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.148337630000242
        },
        {
          "epoch": 620,
          "updates": 9920,
          "samples_seen": 634880,
          "online_loss": 0.0001003459465209744,
          "train_loss": 9.082298493012786e-05,
          "validation_loss": 8.347171387868002e-05,
          "train_metric": 9.147357559413649e-06,
          "validation_metric": 8.406966117036063e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.159738127998935
        },
        {
          "epoch": 621,
          "updates": 9936,
          "samples_seen": 635904,
          "online_loss": 0.0001060484621575597,
          "train_loss": 0.00011520633415784687,
          "validation_loss": 9.763296111486852e-05,
          "train_metric": 1.1603160601225682e-05,
          "validation_metric": 9.83323570835637e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.170178418000432
        },
        {
          "epoch": 622,
          "updates": 9952,
          "samples_seen": 636928,
          "online_loss": 0.00010503287694518804,
          "train_loss": 9.159358160104603e-05,
          "validation_loss": 8.662539767101407e-05,
          "train_metric": 9.224970199284144e-06,
          "validation_metric": 8.72459258971503e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.182738543000596
        },
        {
          "epoch": 623,
          "updates": 9968,
          "samples_seen": 637952,
          "online_loss": 9.598111000741483e-05,
          "train_loss": 9.220594802172855e-05,
          "validation_loss": 8.690357935847715e-05,
          "train_metric": 9.28664485400077e-06,
          "validation_metric": 8.752609574003145e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.19550172199888
        },
        {
          "epoch": 624,
          "updates": 9984,
          "samples_seen": 638976,
          "online_loss": 9.265241851608153e-05,
          "train_loss": 9.148743993137032e-05,
          "validation_loss": 8.37792395032011e-05,
          "train_metric": 9.214279998559505e-06,
          "validation_metric": 8.437938049610239e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.208276745999683
        },
        {
          "epoch": 625,
          "updates": 10000,
          "samples_seen": 640000,
          "online_loss": 9.21362375265744e-05,
          "train_loss": 9.213927114615217e-05,
          "validation_loss": 8.779520430834964e-05,
          "train_metric": 9.27993005461758e-06,
          "validation_metric": 8.842413080856204e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.220911290000004
        },
        {
          "epoch": 626,
          "updates": 10016,
          "samples_seen": 641024,
          "online_loss": 9.598590304449317e-05,
          "train_loss": 9.468831558478996e-05,
          "validation_loss": 8.990128117147833e-05,
          "train_metric": 9.536660400044639e-06,
          "validation_metric": 9.05452725419309e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.233406600998933
        },
        {
          "epoch": 627,
          "updates": 10032,
          "samples_seen": 642048,
          "online_loss": 9.500761871095165e-05,
          "train_loss": 9.067591599887237e-05,
          "validation_loss": 8.343879017047584e-05,
          "train_metric": 9.132544619205873e-06,
          "validation_metric": 8.403649189858697e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.246558756000013
        },
        {
          "epoch": 628,
          "updates": 10048,
          "samples_seen": 643072,
          "online_loss": 9.276938931179757e-05,
          "train_loss": 9.028392378240824e-05,
          "validation_loss": 8.411895396420732e-05,
          "train_metric": 9.093065273191314e-06,
          "validation_metric": 8.472153240290936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.258504441999321
        },
        {
          "epoch": 629,
          "updates": 10064,
          "samples_seen": 644096,
          "online_loss": 9.513819759376929e-05,
          "train_loss": 9.095091081690043e-05,
          "validation_loss": 8.495133806718513e-05,
          "train_metric": 9.160243280348368e-06,
          "validation_metric": 8.55598773341626e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.270514370999081
        },
        {
          "epoch": 630,
          "updates": 10080,
          "samples_seen": 645120,
          "online_loss": 9.106466177399852e-05,
          "train_loss": 9.025618783198297e-05,
          "validation_loss": 8.306361269205809e-05,
          "train_metric": 9.09027221496217e-06,
          "validation_metric": 8.36586332297884e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.282521339000596
        },
        {
          "epoch": 631,
          "updates": 10096,
          "samples_seen": 646144,
          "online_loss": 9.377212586514361e-05,
          "train_loss": 9.244350803783163e-05,
          "validation_loss": 8.43986272229813e-05,
          "train_metric": 9.31057093112031e-06,
          "validation_metric": 8.500320291204844e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.29319663699971
        },
        {
          "epoch": 632,
          "updates": 10112,
          "samples_seen": 647168,
          "online_loss": 9.448344849261048e-05,
          "train_loss": 9.383731230627745e-05,
          "validation_loss": 8.917199011193588e-05,
          "train_metric": 9.45095052884426e-06,
          "validation_metric": 8.981076462077908e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.303711773000032
        },
        {
          "epoch": 633,
          "updates": 10128,
          "samples_seen": 648192,
          "online_loss": 9.39396491048683e-05,
          "train_loss": 8.974465890787542e-05,
          "validation_loss": 8.353075827471912e-05,
          "train_metric": 9.038753887580242e-06,
          "validation_metric": 8.412911483901553e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.314040508999824
        },
        {
          "epoch": 634,
          "updates": 10144,
          "samples_seen": 649216,
          "online_loss": 9.433012201043312e-05,
          "train_loss": 9.048132051248103e-05,
          "validation_loss": 8.287292439490557e-05,
          "train_metric": 9.11294682737207e-06,
          "validation_metric": 8.346656613866799e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.3251439279993065
        },
        {
          "epoch": 635,
          "updates": 10160,
          "samples_seen": 650240,
          "online_loss": 9.184561577058048e-05,
          "train_loss": 8.955182420322672e-05,
          "validation_loss": 8.256673027062789e-05,
          "train_metric": 9.019332537718583e-06,
          "validation_metric": 8.315818377013784e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.336085384000398
        },
        {
          "epoch": 636,
          "updates": 10176,
          "samples_seen": 651264,
          "online_loss": 9.131728120337357e-05,
          "train_loss": 9.08159781829454e-05,
          "validation_loss": 8.377437188755721e-05,
          "train_metric": 9.146652701019775e-06,
          "validation_metric": 8.437447831965983e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.346544632000587
        },
        {
          "epoch": 637,
          "updates": 10192,
          "samples_seen": 652288,
          "online_loss": 9.093355561162753e-05,
          "train_loss": 8.929951582103968e-05,
          "validation_loss": 8.415041520493105e-05,
          "train_metric": 8.993921255751047e-06,
          "validation_metric": 8.475321919831913e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.357136556000114
        },
        {
          "epoch": 638,
          "updates": 10208,
          "samples_seen": 653312,
          "online_loss": 9.050169933288998e-05,
          "train_loss": 8.934116340242326e-05,
          "validation_loss": 8.325427188538015e-05,
          "train_metric": 8.998114935820922e-06,
          "validation_metric": 8.385065484617371e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.3674521560005815
        },
        {
          "epoch": 639,
          "updates": 10224,
          "samples_seen": 654336,
          "online_loss": 9.143164220404287e-05,
          "train_loss": 8.949950279202312e-05,
          "validation_loss": 8.3981612988282e-05,
          "train_metric": 9.014062015921809e-06,
          "validation_metric": 8.458320735371672e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.377768721000393
        },
        {
          "epoch": 640,
          "updates": 10240,
          "samples_seen": 655360,
          "online_loss": 8.955492171480728e-05,
          "train_loss": 9.368876635562629e-05,
          "validation_loss": 8.480676478939131e-05,
          "train_metric": 9.435990250494797e-06,
          "validation_metric": 8.541426723240875e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.388127587999406
        },
        {
          "epoch": 641,
          "updates": 10256,
          "samples_seen": 656384,
          "online_loss": 9.974327826967055e-05,
          "train_loss": 9.184452937915921e-05,
          "validation_loss": 8.512374188285321e-05,
          "train_metric": 9.250244147551712e-06,
          "validation_metric": 8.573350896767806e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.398365983999611
        },
        {
          "epoch": 642,
          "updates": 10272,
          "samples_seen": 657408,
          "online_loss": 8.911417694434931e-05,
          "train_loss": 8.981437713373452e-05,
          "validation_loss": 8.259509922936559e-05,
          "train_metric": 9.045775186677929e-06,
          "validation_metric": 8.318676009366754e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.40873659799945
        },
        {
          "epoch": 643,
          "updates": 10288,
          "samples_seen": 658432,
          "online_loss": 9.282664154852682e-05,
          "train_loss": 8.929918840294704e-05,
          "validation_loss": 8.332105062436312e-05,
          "train_metric": 8.99388669495238e-06,
          "validation_metric": 8.391791197936982e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.418984216999888
        },
        {
          "epoch": 644,
          "updates": 10304,
          "samples_seen": 659456,
          "online_loss": 9.268430630982039e-05,
          "train_loss": 9.706924902275205e-05,
          "validation_loss": 9.119239257415757e-05,
          "train_metric": 9.776459592103492e-06,
          "validation_metric": 9.184564078168478e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.429199096999582
        },
        {
          "epoch": 645,
          "updates": 10320,
          "samples_seen": 660480,
          "online_loss": 0.00010163820093112008,
          "train_loss": 9.059150033863261e-05,
          "validation_loss": 8.316957973875105e-05,
          "train_metric": 9.124044481723104e-06,
          "validation_metric": 8.376535333809443e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.439401171999634
        },
        {
          "epoch": 646,
          "updates": 10336,
          "samples_seen": 661504,
          "online_loss": 9.439223549634335e-05,
          "train_loss": 8.935808000387624e-05,
          "validation_loss": 8.368950511794537e-05,
          "train_metric": 8.999817509902641e-06,
          "validation_metric": 8.428900400758721e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.449772099000256
        },
        {
          "epoch": 647,
          "updates": 10352,
          "samples_seen": 662528,
          "online_loss": 9.356356986245373e-05,
          "train_loss": 0.00010528779966989532,
          "validation_loss": 8.945714216679335e-05,
          "train_metric": 1.0604201634123456e-05,
          "validation_metric": 9.009795576275792e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.460135504999926
        },
        {
          "epoch": 648,
          "updates": 10368,
          "samples_seen": 663552,
          "online_loss": 9.167222469841363e-05,
          "train_loss": 9.921831951942295e-05,
          "validation_loss": 9.491920354776084e-05,
          "train_metric": 9.992905688704923e-06,
          "validation_metric": 9.559914360579569e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.470515826000337
        },
        {
          "epoch": 649,
          "updates": 10384,
          "samples_seen": 664576,
          "online_loss": 9.247915932064643e-05,
          "train_loss": 9.997245797421783e-05,
          "validation_loss": 8.876279753167182e-05,
          "train_metric": 1.0068861229228787e-05,
          "validation_metric": 8.93986361916177e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.4809872539990465
        },
        {
          "epoch": 650,
          "updates": 10400,
          "samples_seen": 665600,
          "online_loss": 9.220287029165775e-05,
          "train_loss": 9.273290925193578e-05,
          "validation_loss": 8.767937106313184e-05,
          "train_metric": 9.339718417322729e-06,
          "validation_metric": 8.830745173327159e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.4912727720002295
        },
        {
          "epoch": 651,
          "updates": 10416,
          "samples_seen": 666624,
          "online_loss": 9.037049221660709e-05,
          "train_loss": 9.166129893856123e-05,
          "validation_loss": 8.291945414384827e-05,
          "train_metric": 9.23179140954744e-06,
          "validation_metric": 8.351344149559736e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.501586489999681
        },
        {
          "epoch": 652,
          "updates": 10432,
          "samples_seen": 667648,
          "online_loss": 9.314343810729042e-05,
          "train_loss": 8.826602424960583e-05,
          "validation_loss": 8.303522918140516e-05,
          "train_metric": 8.889829587133136e-06,
          "validation_metric": 8.363003871636465e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.511937907000174
        },
        {
          "epoch": 653,
          "updates": 10448,
          "samples_seen": 668672,
          "online_loss": 9.22398858165252e-05,
          "train_loss": 8.814203465590253e-05,
          "validation_loss": 8.251352119259536e-05,
          "train_metric": 8.877343134372495e-06,
          "validation_metric": 8.31046054372564e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.522669632000543
        },
        {
          "epoch": 654,
          "updates": 10464,
          "samples_seen": 669696,
          "online_loss": 9.075060734176077e-05,
          "train_loss": 8.841433736961335e-05,
          "validation_loss": 8.314505248563364e-05,
          "train_metric": 8.904767128115054e-06,
          "validation_metric": 8.374064236704726e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.53306761299973
        },
        {
          "epoch": 655,
          "updates": 10480,
          "samples_seen": 670720,
          "online_loss": 9.435702850169037e-05,
          "train_loss": 9.547974332235754e-05,
          "validation_loss": 8.773837907938287e-05,
          "train_metric": 9.61636942520272e-06,
          "validation_metric": 8.836688721203245e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.597897884999838
        },
        {
          "epoch": 656,
          "updates": 10496,
          "samples_seen": 671744,
          "online_loss": 9.60775150815607e-05,
          "train_loss": 8.7947744759731e-05,
          "validation_loss": 8.271469414466992e-05,
          "train_metric": 8.85777535586385e-06,
          "validation_metric": 8.330721357197035e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.608595783000055
        },
        {
          "epoch": 657,
          "updates": 10512,
          "samples_seen": 672768,
          "online_loss": 9.615272233531869e-05,
          "train_loss": 0.00011187454219907522,
          "validation_loss": 9.129160025622696e-05,
          "train_metric": 1.1267593436059542e-05,
          "validation_metric": 9.194555786962155e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.619209122000029
        },
        {
          "epoch": 658,
          "updates": 10528,
          "samples_seen": 673792,
          "online_loss": 9.472495048612473e-05,
          "train_loss": 0.00010970918810926378,
          "validation_loss": 9.802280692383647e-05,
          "train_metric": 1.1049507520510815e-05,
          "validation_metric": 9.872497685137205e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.629763843000546
        },
        {
          "epoch": 659,
          "updates": 10544,
          "samples_seen": 674816,
          "online_loss": 0.00011124670072604204,
          "train_loss": 9.061532909981906e-05,
          "validation_loss": 8.777682523941621e-05,
          "train_metric": 9.12644372874638e-06,
          "validation_metric": 8.840560440148693e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.639983920000304
        },
        {
          "epoch": 660,
          "updates": 10560,
          "samples_seen": 675840,
          "online_loss": 0.00010511147888792038,
          "train_loss": 0.00011839841317851096,
          "validation_loss": 9.687137935543433e-05,
          "train_metric": 1.1924653335881885e-05,
          "validation_metric": 9.756530744198244e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.650226334999388
        },
        {
          "epoch": 661,
          "updates": 10576,
          "samples_seen": 676864,
          "online_loss": 0.00010064541811516392,
          "train_loss": 8.787215483607724e-05,
          "validation_loss": 8.293631981359795e-05,
          "train_metric": 8.850161066220608e-06,
          "validation_metric": 8.353042176167946e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.660433547000139
        },
        {
          "epoch": 662,
          "updates": 10592,
          "samples_seen": 677888,
          "online_loss": 8.831934883346548e-05,
          "train_loss": 9.010919166030362e-05,
          "validation_loss": 8.159155549947172e-05,
          "train_metric": 9.075466550712008e-06,
          "validation_metric": 8.217602044169325e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.6712997979993816
        },
        {
          "epoch": 663,
          "updates": 10608,
          "samples_seen": 678912,
          "online_loss": 9.322255755250808e-05,
          "train_loss": 8.891800098353997e-05,
          "validation_loss": 8.256251021521166e-05,
          "train_metric": 8.955495104601141e-06,
          "validation_metric": 8.315393642988056e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.681704623999394
        },
        {
          "epoch": 664,
          "updates": 10624,
          "samples_seen": 679936,
          "online_loss": 9.10724497771298e-05,
          "train_loss": 8.82812455529347e-05,
          "validation_loss": 8.125976455630735e-05,
          "train_metric": 8.891363904695027e-06,
          "validation_metric": 8.184186299331486e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.692701546000535
        },
        {
          "epoch": 665,
          "updates": 10640,
          "samples_seen": 680960,
          "online_loss": 9.070818759937538e-05,
          "train_loss": 8.751063433010131e-05,
          "validation_loss": 8.145412721205503e-05,
          "train_metric": 8.81375035532983e-06,
          "validation_metric": 8.203760444303043e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.703040845999567
        },
        {
          "epoch": 666,
          "updates": 10656,
          "samples_seen": 681984,
          "online_loss": 9.284199904868728e-05,
          "train_loss": 8.773017179919407e-05,
          "validation_loss": 8.352148142876104e-05,
          "train_metric": 8.835861990519334e-06,
          "validation_metric": 8.411977432842832e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.713382296999043
        },
        {
          "epoch": 667,
          "updates": 10672,
          "samples_seen": 683008,
          "online_loss": 9.583367909726803e-05,
          "train_loss": 9.034860704559833e-05,
          "validation_loss": 8.531586354365572e-05,
          "train_metric": 9.099580893234815e-06,
          "validation_metric": 8.592702215537429e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.723642921000646
        },
        {
          "epoch": 668,
          "updates": 10688,
          "samples_seen": 684032,
          "online_loss": 8.846222044667229e-05,
          "train_loss": 8.74344987096265e-05,
          "validation_loss": 8.088055619737133e-05,
          "train_metric": 8.806081496004481e-06,
          "validation_metric": 8.145993888319936e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.734534983999765
        },
        {
          "epoch": 669,
          "updates": 10704,
          "samples_seen": 685056,
          "online_loss": 8.909316397875955e-05,
          "train_loss": 8.700485341250896e-05,
          "validation_loss": 8.21338762762025e-05,
          "train_metric": 8.762809557083528e-06,
          "validation_metric": 8.272223567473702e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.744908016000409
        },
        {
          "epoch": 670,
          "updates": 10720,
          "samples_seen": 686080,
          "online_loss": 9.084487464861013e-05,
          "train_loss": 8.822303789202124e-05,
          "validation_loss": 8.214730769395828e-05,
          "train_metric": 8.885500392352697e-06,
          "validation_metric": 8.273575076600537e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.755181144999369
        },
        {
          "epoch": 671,
          "updates": 10736,
          "samples_seen": 687104,
          "online_loss": 9.27401792978344e-05,
          "train_loss": 8.752068970352411e-05,
          "validation_loss": 8.195092959795147e-05,
          "train_metric": 8.814762622932903e-06,
          "validation_metric": 8.253798114310484e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.765396303999296
        },
        {
          "epoch": 672,
          "updates": 10752,
          "samples_seen": 688128,
          "online_loss": 8.918718663153413e-05,
          "train_loss": 8.850210724631324e-05,
          "validation_loss": 8.369125134777278e-05,
          "train_metric": 8.913608326110989e-06,
          "validation_metric": 8.429076842730865e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.775664541999504
        },
        {
          "epoch": 673,
          "updates": 10768,
          "samples_seen": 689152,
          "online_loss": 9.764552214619471e-05,
          "train_loss": 8.754323062021285e-05,
          "validation_loss": 8.343253284692764e-05,
          "train_metric": 8.81703363120323e-06,
          "validation_metric": 8.403018910030369e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.785847155999363
        },
        {
          "epoch": 674,
          "updates": 10784,
          "samples_seen": 690176,
          "online_loss": 9.624481435821508e-05,
          "train_loss": 0.00010502408986212686,
          "validation_loss": 8.612513192929327e-05,
          "train_metric": 1.0577640750852879e-05,
          "validation_metric": 8.674208402226213e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.7961598279998725
        },
        {
          "epoch": 675,
          "updates": 10800,
          "samples_seen": 691200,
          "online_loss": 9.200182103086263e-05,
          "train_loss": 8.807354606688023e-05,
          "validation_loss": 8.358243940165266e-05,
          "train_metric": 8.87044552655425e-06,
          "validation_metric": 8.418117431574501e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.806399940000119
        },
        {
          "epoch": 676,
          "updates": 10816,
          "samples_seen": 692224,
          "online_loss": 8.70519236286782e-05,
          "train_loss": 8.825938130030409e-05,
          "validation_loss": 7.987244316609576e-05,
          "train_metric": 8.889161108527333e-06,
          "validation_metric": 8.044458809308708e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.817233448000479
        },
        {
          "epoch": 677,
          "updates": 10832,
          "samples_seen": 693248,
          "online_loss": 9.038551365847525e-05,
          "train_loss": 8.66426489665173e-05,
          "validation_loss": 8.220847666962072e-05,
          "train_metric": 8.726329724595416e-06,
          "validation_metric": 8.279736903205048e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.82762030899903
        },
        {
          "epoch": 678,
          "updates": 10848,
          "samples_seen": 694272,
          "online_loss": 8.75181583523954e-05,
          "train_loss": 8.699354657437652e-05,
          "validation_loss": 8.095391967799515e-05,
          "train_metric": 8.76167177921161e-06,
          "validation_metric": 8.153382623277139e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.837939283999731
        },
        {
          "epoch": 679,
          "updates": 10864,
          "samples_seen": 695296,
          "online_loss": 9.366966514789965e-05,
          "train_loss": 8.703989442437887e-05,
          "validation_loss": 8.266299846582115e-05,
          "train_metric": 8.76633930602111e-06,
          "validation_metric": 8.325514500029385e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.848169933999088
        },
        {
          "epoch": 680,
          "updates": 10880,
          "samples_seen": 696320,
          "online_loss": 8.835306039145507e-05,
          "train_loss": 8.665317000122741e-05,
          "validation_loss": 8.272618288174272e-05,
          "train_metric": 8.727390195417684e-06,
          "validation_metric": 8.33187823445769e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.85844042700046
        },
        {
          "epoch": 681,
          "updates": 10896,
          "samples_seen": 697344,
          "online_loss": 8.759196794017043e-05,
          "train_loss": 8.821907249512151e-05,
          "validation_loss": 8.095057273749262e-05,
          "train_metric": 8.885101124178618e-06,
          "validation_metric": 8.15304429124808e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.868649112000639
        },
        {
          "epoch": 682,
          "updates": 10912,
          "samples_seen": 698368,
          "online_loss": 9.082283986572293e-05,
          "train_loss": 8.812054875306785e-05,
          "validation_loss": 8.044898277148604e-05,
          "train_metric": 8.875180355971679e-06,
          "validation_metric": 8.102526408038102e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.879289444999813
        },
        {
          "epoch": 683,
          "updates": 10928,
          "samples_seen": 699392,
          "online_loss": 8.937787197282887e-05,
          "train_loss": 8.77278289408423e-05,
          "validation_loss": 8.334772428497672e-05,
          "train_metric": 8.835624612402171e-06,
          "validation_metric": 8.394477845286019e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.8897401340000215
        },
        {
          "epoch": 684,
          "updates": 10944,
          "samples_seen": 700416,
          "online_loss": 9.20856300581363e-05,
          "train_loss": 8.803829405223951e-05,
          "validation_loss": 8.462645200779662e-05,
          "train_metric": 8.866894859238528e-06,
          "validation_metric": 8.523265933035873e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.900177940000503
        },
        {
          "epoch": 685,
          "updates": 10960,
          "samples_seen": 701440,
          "online_loss": 9.423330448043998e-05,
          "train_loss": 9.419304842595011e-05,
          "validation_loss": 8.244461787398905e-05,
          "train_metric": 9.486779163125902e-06,
          "validation_metric": 8.30352018965641e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.910433641000054
        },
        {
          "epoch": 686,
          "updates": 10976,
          "samples_seen": 702464,
          "online_loss": 8.987826140582911e-05,
          "train_loss": 8.58597268234007e-05,
          "validation_loss": 8.037852239795029e-05,
          "train_metric": 8.647478352941107e-06,
          "validation_metric": 8.095431439869571e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.920717192000666
        },
        {
          "epoch": 687,
          "updates": 10992,
          "samples_seen": 703488,
          "online_loss": 8.719643460608495e-05,
          "train_loss": 8.624389010947198e-05,
          "validation_loss": 8.089091716101393e-05,
          "train_metric": 8.686168257554527e-06,
          "validation_metric": 8.147036169248167e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.9309692370006815
        },
        {
          "epoch": 688,
          "updates": 11008,
          "samples_seen": 704512,
          "online_loss": 8.842809506859339e-05,
          "train_loss": 8.56191836646758e-05,
          "validation_loss": 8.045601134654135e-05,
          "train_metric": 8.623249414085876e-06,
          "validation_metric": 8.103234904410783e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.941283345999182
        },
        {
          "epoch": 689,
          "updates": 11024,
          "samples_seen": 705536,
          "online_loss": 8.644842637295369e-05,
          "train_loss": 8.607427298557013e-05,
          "validation_loss": 7.919689232949167e-05,
          "train_metric": 8.669085218571126e-06,
          "validation_metric": 7.976420420163777e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.952087235000363
        },
        {
          "epoch": 690,
          "updates": 11040,
          "samples_seen": 706560,
          "online_loss": 8.769441069489403e-05,
          "train_loss": 8.556963439332321e-05,
          "validation_loss": 8.047684241319075e-05,
          "train_metric": 8.61825992615195e-06,
          "validation_metric": 8.105333108687773e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 7.962701127999026
        },
        {
          "epoch": 691,
          "updates": 11056,
          "samples_seen": 707584,
          "online_loss": 8.698252713656984e-05,
          "train_loss": 8.66701448103413e-05,
          "validation_loss": 8.090768824331462e-05,
          "train_metric": 8.729100045457017e-06,
          "validation_metric": 8.148726919898763e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.032513930998903
        },
        {
          "epoch": 692,
          "updates": 11072,
          "samples_seen": 708608,
          "online_loss": 9.086113595913048e-05,
          "train_loss": 9.258713544113562e-05,
          "validation_loss": 8.235286077251658e-05,
          "train_metric": 9.32503735384671e-06,
          "validation_metric": 8.294278813991696e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.04356399199969
        },
        {
          "epoch": 693,
          "updates": 11088,
          "samples_seen": 709632,
          "online_loss": 8.908967265597312e-05,
          "train_loss": 8.676753350300714e-05,
          "validation_loss": 8.275949949165806e-05,
          "train_metric": 8.738908945815638e-06,
          "validation_metric": 8.33523336041253e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.053821772999072
        },
        {
          "epoch": 694,
          "updates": 11104,
          "samples_seen": 710656,
          "online_loss": 8.83923557921662e-05,
          "train_loss": 9.264589607482776e-05,
          "validation_loss": 8.250570681411773e-05,
          "train_metric": 9.330955435871147e-06,
          "validation_metric": 8.309672921313904e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.064160353998886
        },
        {
          "epoch": 695,
          "updates": 11120,
          "samples_seen": 711680,
          "online_loss": 8.833821289044863e-05,
          "train_loss": 8.612912643002346e-05,
          "validation_loss": 8.059130777837709e-05,
          "train_metric": 8.674611308379099e-06,
          "validation_metric": 8.116861863527447e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.074386968999534
        },
        {
          "epoch": 696,
          "updates": 11136,
          "samples_seen": 712704,
          "online_loss": 8.935979803936789e-05,
          "train_loss": 8.511156920576468e-05,
          "validation_loss": 8.021533722057939e-05,
          "train_metric": 8.572125807404518e-06,
          "validation_metric": 8.07899414212443e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.0846620049997
        },
        {
          "epoch": 697,
          "updates": 11152,
          "samples_seen": 713728,
          "online_loss": 8.869080829754239e-05,
          "train_loss": 8.734624861972407e-05,
          "validation_loss": 7.978753274073824e-05,
          "train_metric": 8.797193913778756e-06,
          "validation_metric": 8.03590864961734e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.094811466000465
        },
        {
          "epoch": 698,
          "updates": 11168,
          "samples_seen": 714752,
          "online_loss": 9.298680265601433e-05,
          "train_loss": 9.770324686542153e-05,
          "validation_loss": 9.375240188091993e-05,
          "train_metric": 9.840314305620268e-06,
          "validation_metric": 9.44239945965819e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.105152182999518
        },
        {
          "epoch": 699,
          "updates": 11184,
          "samples_seen": 715776,
          "online_loss": 9.642909844842507e-05,
          "train_loss": 8.745383092900738e-05,
          "validation_loss": 7.997975626494735e-05,
          "train_metric": 8.808028724160977e-06,
          "validation_metric": 8.055268153839279e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.115409433999957
        },
        {
          "epoch": 700,
          "updates": 11200,
          "samples_seen": 716800,
          "online_loss": 8.836695769787184e-05,
          "train_loss": 8.564123709220439e-05,
          "validation_loss": 7.991612801561132e-05,
          "train_metric": 8.62547221913701e-06,
          "validation_metric": 8.048858944675885e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.12575685000047
        },
        {
          "epoch": 701,
          "updates": 11216,
          "samples_seen": 717824,
          "online_loss": 8.622654240753036e-05,
          "train_loss": 9.220013453159481e-05,
          "validation_loss": 8.474628702970222e-05,
          "train_metric": 9.286060958402231e-06,
          "validation_metric": 8.535335837223101e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.135908737000136
        },
        {
          "epoch": 702,
          "updates": 11232,
          "samples_seen": 718848,
          "online_loss": 9.479398113398929e-05,
          "train_loss": 8.888717275112867e-05,
          "validation_loss": 8.564350719098002e-05,
          "train_metric": 8.95239099918399e-06,
          "validation_metric": 8.625700502307154e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.146302352999555
        },
        {
          "epoch": 703,
          "updates": 11248,
          "samples_seen": 719872,
          "online_loss": 8.912625162338372e-05,
          "train_loss": 8.496904774801806e-05,
          "validation_loss": 7.886145613156259e-05,
          "train_metric": 8.557772162021138e-06,
          "validation_metric": 7.94263632997172e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.157120276999194
        },
        {
          "epoch": 704,
          "updates": 11264,
          "samples_seen": 720896,
          "online_loss": 8.775652736403572e-05,
          "train_loss": 8.500367403030396e-05,
          "validation_loss": 8.014705963432789e-05,
          "train_metric": 8.561258255213033e-06,
          "validation_metric": 8.072118362179026e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.167588244999934
        },
        {
          "epoch": 705,
          "updates": 11280,
          "samples_seen": 721920,
          "online_loss": 8.726384567125933e-05,
          "train_loss": 8.559572597732767e-05,
          "validation_loss": 8.031920879147947e-05,
          "train_metric": 8.620887456345372e-06,
          "validation_metric": 8.089456969173625e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.178255941998941
        },
        {
          "epoch": 706,
          "updates": 11296,
          "samples_seen": 722944,
          "online_loss": 8.810298277239781e-05,
          "train_loss": 8.58151470310986e-05,
          "validation_loss": 7.902513607405126e-05,
          "train_metric": 8.642987268103752e-06,
          "validation_metric": 7.959122740430757e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.18871258999934
        },
        {
          "epoch": 707,
          "updates": 11312,
          "samples_seen": 723968,
          "online_loss": 8.615895785624161e-05,
          "train_loss": 9.085877536563203e-05,
          "validation_loss": 8.560991409467533e-05,
          "train_metric": 9.150961886916775e-06,
          "validation_metric": 8.622317182016559e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.19913316900056
        },
        {
          "epoch": 708,
          "updates": 11328,
          "samples_seen": 724992,
          "online_loss": 9.442346072319197e-05,
          "train_loss": 8.439536031801254e-05,
          "validation_loss": 7.882368663558736e-05,
          "train_metric": 8.499991054122802e-06,
          "validation_metric": 7.938832823128905e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.210079425000004
        },
        {
          "epoch": 709,
          "updates": 11344,
          "samples_seen": 726016,
          "online_loss": 8.769435407884885e-05,
          "train_loss": 8.495125075569376e-05,
          "validation_loss": 7.844503852538764e-05,
          "train_metric": 8.555979547963943e-06,
          "validation_metric": 7.900697710283566e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.2209594750002
        },
        {
          "epoch": 710,
          "updates": 11360,
          "samples_seen": 727040,
          "online_loss": 8.635019980829384e-05,
          "train_loss": 8.421968959737569e-05,
          "validation_loss": 7.847500091884285e-05,
          "train_metric": 8.482300472678617e-06,
          "validation_metric": 7.903714504209347e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.231429227998888
        },
        {
          "epoch": 711,
          "updates": 11376,
          "samples_seen": 728064,
          "online_loss": 8.635658309685823e-05,
          "train_loss": 8.583210728829727e-05,
          "validation_loss": 7.852995622670278e-05,
          "train_metric": 8.644696208648384e-06,
          "validation_metric": 7.909248779469635e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.241999415000464
        },
        {
          "epoch": 712,
          "updates": 11392,
          "samples_seen": 729088,
          "online_loss": 8.619807977083838e-05,
          "train_loss": 9.25882050069049e-05,
          "validation_loss": 8.433497714577243e-05,
          "train_metric": 9.325143764726818e-06,
          "validation_metric": 8.49391108204145e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.25225286299974
        },
        {
          "epoch": 713,
          "updates": 11408,
          "samples_seen": 730112,
          "online_loss": 8.837687073537381e-05,
          "train_loss": 8.56662227306515e-05,
          "validation_loss": 7.79593683546409e-05,
          "train_metric": 8.627988790976815e-06,
          "validation_metric": 7.851782356738113e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.263092028999381
        },
        {
          "epoch": 714,
          "updates": 11424,
          "samples_seen": 731136,
          "online_loss": 8.804569142739638e-05,
          "train_loss": 8.416025957558304e-05,
          "validation_loss": 7.827812805771828e-05,
          "train_metric": 8.476312359562144e-06,
          "validation_metric": 7.883886610215995e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.27390953900067
        },
        {
          "epoch": 715,
          "updates": 11440,
          "samples_seen": 732160,
          "online_loss": 8.718021535969456e-05,
          "train_loss": 8.544985757907853e-05,
          "validation_loss": 7.736705447314307e-05,
          "train_metric": 8.606196388427634e-06,
          "validation_metric": 7.792126780259423e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.284728844999336
        },
        {
          "epoch": 716,
          "updates": 11456,
          "samples_seen": 733184,
          "online_loss": 8.939740155256004e-05,
          "train_loss": 8.43101879581809e-05,
          "validation_loss": 7.927088881842792e-05,
          "train_metric": 8.491413609590381e-06,
          "validation_metric": 7.983873729244806e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.295222343000205
        },
        {
          "epoch": 717,
          "updates": 11472,
          "samples_seen": 734208,
          "online_loss": 8.820197035674937e-05,
          "train_loss": 8.409583097090945e-05,
          "validation_loss": 7.78370740590617e-05,
          "train_metric": 8.469824933854397e-06,
          "validation_metric": 7.839465979486704e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.30554909700004
        },
        {
          "epoch": 718,
          "updates": 11488,
          "samples_seen": 735232,
          "online_loss": 9.016161197905603e-05,
          "train_loss": 9.533768752589822e-05,
          "validation_loss": 8.306110248668119e-05,
          "train_metric": 9.602061254554428e-06,
          "validation_metric": 8.365609573957045e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.316143359999842
        },
        {
          "epoch": 719,
          "updates": 11504,
          "samples_seen": 736256,
          "online_loss": 8.998656858238974e-05,
          "train_loss": 8.419880759902298e-05,
          "validation_loss": 8.007208816707134e-05,
          "train_metric": 8.480195901938714e-06,
          "validation_metric": 8.064567737164907e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.326365716999135
        },
        {
          "epoch": 720,
          "updates": 11520,
          "samples_seen": 737280,
          "online_loss": 8.900023317437444e-05,
          "train_loss": 8.460774552077055e-05,
          "validation_loss": 7.767968054395169e-05,
          "train_metric": 8.521382369508501e-06,
          "validation_metric": 7.823613486834802e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.336916111000392
        },
        {
          "epoch": 721,
          "updates": 11536,
          "samples_seen": 738304,
          "online_loss": 8.765005941313575e-05,
          "train_loss": 8.506880840286613e-05,
          "validation_loss": 7.905578968347982e-05,
          "train_metric": 8.567818440496922e-06,
          "validation_metric": 7.962209565448575e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.34737389700058
        },
        {
          "epoch": 722,
          "updates": 11552,
          "samples_seen": 739328,
          "online_loss": 8.58366549891798e-05,
          "train_loss": 8.634959522169083e-05,
          "validation_loss": 7.880998600739986e-05,
          "train_metric": 8.696815712028183e-06,
          "validation_metric": 7.937453119666316e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.357690839000497
        },
        {
          "epoch": 723,
          "updates": 11568,
          "samples_seen": 740352,
          "online_loss": 8.520103210685193e-05,
          "train_loss": 8.36359613458626e-05,
          "validation_loss": 7.766624912619591e-05,
          "train_metric": 8.42350800667191e-06,
          "validation_metric": 7.822261068213265e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.367883503000485
        },
        {
          "epoch": 724,
          "updates": 11584,
          "samples_seen": 741376,
          "online_loss": 8.724258395886864e-05,
          "train_loss": 8.774283196544275e-05,
          "validation_loss": 8.319887274410576e-05,
          "train_metric": 8.83713710209122e-06,
          "validation_metric": 8.379486644116696e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.378211891998944
        },
        {
          "epoch": 725,
          "updates": 11600,
          "samples_seen": 742400,
          "online_loss": 8.804139974927239e-05,
          "train_loss": 0.00010771504457807168,
          "validation_loss": 9.136881999438629e-05,
          "train_metric": 1.0848664715013001e-05,
          "validation_metric": 9.202332876157016e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.388454874000672
        },
        {
          "epoch": 726,
          "updates": 11616,
          "samples_seen": 743424,
          "online_loss": 9.375961030855251e-05,
          "train_loss": 8.350498683284968e-05,
          "validation_loss": 7.71948107285425e-05,
          "train_metric": 8.410315786022693e-06,
          "validation_metric": 7.774778168823104e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.399291581999933
        },
        {
          "epoch": 727,
          "updates": 11632,
          "samples_seen": 744448,
          "online_loss": 8.736713175494515e-05,
          "train_loss": 9.106764628086239e-05,
          "validation_loss": 8.47489427542314e-05,
          "train_metric": 9.172000318358187e-06,
          "validation_metric": 8.535603228665423e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.462004499999239
        },
        {
          "epoch": 728,
          "updates": 11648,
          "samples_seen": 745472,
          "online_loss": 8.747757874516537e-05,
          "train_loss": 8.546331082470715e-05,
          "validation_loss": 7.822398038115352e-05,
          "train_metric": 8.607552445027977e-06,
          "validation_metric": 7.878432370489463e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.472973648000334
        },
        {
          "epoch": 729,
          "updates": 11664,
          "samples_seen": 746496,
          "online_loss": 8.57447760154173e-05,
          "train_loss": 9.07156863831915e-05,
          "validation_loss": 8.577340486226603e-05,
          "train_metric": 9.136551852861885e-06,
          "validation_metric": 8.638782674097456e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.48319110599914
        },
        {
          "epoch": 730,
          "updates": 11680,
          "samples_seen": 747520,
          "online_loss": 8.788593618191953e-05,
          "train_loss": 9.297463111579418e-05,
          "validation_loss": 7.962338713696226e-05,
          "train_metric": 9.364063771499787e-06,
          "validation_metric": 8.019375854928512e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.493535129999145
        },
        {
          "epoch": 731,
          "updates": 11696,
          "samples_seen": 748544,
          "online_loss": 8.68155368607404e-05,
          "train_loss": 8.33834710647352e-05,
          "validation_loss": 7.894389273133129e-05,
          "train_metric": 8.398076715820935e-06,
          "validation_metric": 7.950940016598906e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.503733335999641
        },
        {
          "epoch": 732,
          "updates": 11712,
          "samples_seen": 749568,
          "online_loss": 8.769111036599497e-05,
          "train_loss": 8.35393657325767e-05,
          "validation_loss": 7.808185182511806e-05,
          "train_metric": 8.413779141847044e-06,
          "validation_metric": 7.864118742872961e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.514112712999122
        },
        {
          "epoch": 733,
          "updates": 11728,
          "samples_seen": 750592,
          "online_loss": 8.902003492039512e-05,
          "train_loss": 8.357624028576538e-05,
          "validation_loss": 7.887641550041735e-05,
          "train_metric": 8.417492608714383e-06,
          "validation_metric": 7.944144272187259e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.524374437000006
        },
        {
          "epoch": 734,
          "updates": 11744,
          "samples_seen": 751616,
          "online_loss": 8.654597900203953e-05,
          "train_loss": 8.66102272993885e-05,
          "validation_loss": 7.980168447829783e-05,
          "train_metric": 8.723065548110753e-06,
          "validation_metric": 8.037332918320317e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.534610017999512
        },
        {
          "epoch": 735,
          "updates": 11760,
          "samples_seen": 752640,
          "online_loss": 8.82192086919531e-05,
          "train_loss": 8.260671165771782e-05,
          "validation_loss": 7.696489046793431e-05,
          "train_metric": 8.319845619553234e-06,
          "validation_metric": 7.751622433715966e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.545466892999684
        },
        {
          "epoch": 736,
          "updates": 11776,
          "samples_seen": 753664,
          "online_loss": 8.545498144485464e-05,
          "train_loss": 8.265198266599327e-05,
          "validation_loss": 7.66225130064413e-05,
          "train_metric": 8.324404916493222e-06,
          "validation_metric": 7.717139851592947e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.55644143400059
        },
        {
          "epoch": 737,
          "updates": 11792,
          "samples_seen": 754688,
          "online_loss": 8.850159269968572e-05,
          "train_loss": 8.490824257023633e-05,
          "validation_loss": 7.746603660052642e-05,
          "train_metric": 8.551646715204697e-06,
          "validation_metric": 7.802095751685556e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.566733233999912
        },
        {
          "epoch": 738,
          "updates": 11808,
          "samples_seen": 755712,
          "online_loss": 8.59212470913917e-05,
          "train_loss": 8.295791485579684e-05,
          "validation_loss": 7.800862658768892e-05,
          "train_metric": 8.355217687494587e-06,
          "validation_metric": 7.856743650336284e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.577032150999003
        },
        {
          "epoch": 739,
          "updates": 11824,
          "samples_seen": 756736,
          "online_loss": 8.53524779813597e-05,
          "train_loss": 8.260272443294525e-05,
          "validation_loss": 7.6647600508295e-05,
          "train_metric": 8.319444532389753e-06,
          "validation_metric": 7.719666427874472e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.587212856000406
        },
        {
          "epoch": 740,
          "updates": 11840,
          "samples_seen": 757760,
          "online_loss": 8.86937395989662e-05,
          "train_loss": 8.758113835938275e-05,
          "validation_loss": 7.891099812695757e-05,
          "train_metric": 8.820850780466571e-06,
          "validation_metric": 7.947625817905646e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.597478375999344
        },
        {
          "epoch": 741,
          "updates": 11856,
          "samples_seen": 758784,
          "online_loss": 8.306512745548389e-05,
          "train_loss": 8.424560655839741e-05,
          "validation_loss": 7.79411566327326e-05,
          "train_metric": 8.4849079939886e-06,
          "validation_metric": 7.849946996429935e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.60773235600027
        },
        {
          "epoch": 742,
          "updates": 11872,
          "samples_seen": 759808,
          "online_loss": 8.408102621615399e-05,
          "train_loss": 8.25128227006644e-05,
          "validation_loss": 7.766797352815047e-05,
          "train_metric": 8.310389603138901e-06,
          "validation_metric": 7.822435691196006e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.618003039999167
        },
        {
          "epoch": 743,
          "updates": 11888,
          "samples_seen": 760832,
          "online_loss": 8.535261122233351e-05,
          "train_loss": 8.422773680649698e-05,
          "validation_loss": 7.91405764175579e-05,
          "train_metric": 8.483108103973791e-06,
          "validation_metric": 7.970749720698223e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.628133038999295
        },
        {
          "epoch": 744,
          "updates": 11904,
          "samples_seen": 761856,
          "online_loss": 8.478669269607053e-05,
          "train_loss": 8.312039426527917e-05,
          "validation_loss": 7.784267654642463e-05,
          "train_metric": 8.371582225663587e-06,
          "validation_metric": 7.840029866201803e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.638338232000024
        },
        {
          "epoch": 745,
          "updates": 11920,
          "samples_seen": 762880,
          "online_loss": 8.6245557440634e-05,
          "train_loss": 8.278946188511327e-05,
          "validation_loss": 7.773733523208648e-05,
          "train_metric": 8.338251973327715e-06,
          "validation_metric": 7.829420610505622e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.648549383999125
        },
        {
          "epoch": 746,
          "updates": 11936,
          "samples_seen": 763904,
          "online_loss": 8.783901284914464e-05,
          "train_loss": 9.008385677589104e-05,
          "validation_loss": 7.883810758357868e-05,
          "train_metric": 9.072916327568237e-06,
          "validation_metric": 7.940285286167637e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.658815216000221
        },
        {
          "epoch": 747,
          "updates": 11952,
          "samples_seen": 764928,
          "online_loss": 8.463049584861437e-05,
          "train_loss": 8.227895887102932e-05,
          "validation_loss": 7.655663648620248e-05,
          "train_metric": 8.286835509352386e-06,
          "validation_metric": 7.71050326875411e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.669609041999138
        },
        {
          "epoch": 748,
          "updates": 11968,
          "samples_seen": 765952,
          "online_loss": 8.432658478341182e-05,
          "train_loss": 8.281043847091496e-05,
          "validation_loss": 7.703057781327516e-05,
          "train_metric": 8.34036291053053e-06,
          "validation_metric": 7.758238098176662e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.680084215000534
        },
        {
          "epoch": 749,
          "updates": 11984,
          "samples_seen": 766976,
          "online_loss": 8.408823441641289e-05,
          "train_loss": 9.295155177824199e-05,
          "validation_loss": 7.981109956745058e-05,
          "train_metric": 9.361740012536757e-06,
          "validation_metric": 8.038281521294266e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.69028182799957
        },
        {
          "epoch": 750,
          "updates": 12000,
          "samples_seen": 768000,
          "online_loss": 9.178107211482711e-05,
          "train_loss": 8.297515887534246e-05,
          "validation_loss": 7.692824874538928e-05,
          "train_metric": 8.356953912880272e-06,
          "validation_metric": 7.747931704216171e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.700555379000434
        },
        {
          "epoch": 751,
          "updates": 12016,
          "samples_seen": 769024,
          "online_loss": 8.433329276158474e-05,
          "train_loss": 8.402831008424982e-05,
          "validation_loss": 7.859321340220049e-05,
          "train_metric": 8.463022822979838e-06,
          "validation_metric": 7.915620699350256e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.710956234999685
        },
        {
          "epoch": 752,
          "updates": 12032,
          "samples_seen": 770048,
          "online_loss": 8.241081741289236e-05,
          "train_loss": 8.268560486612841e-05,
          "validation_loss": 7.65650620451197e-05,
          "train_metric": 8.327792784257326e-06,
          "validation_metric": 7.711353646300267e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.721439687000384
        },
        {
          "epoch": 753,
          "updates": 12048,
          "samples_seen": 771072,
          "online_loss": 8.711944610695355e-05,
          "train_loss": 8.378172788070515e-05,
          "validation_loss": 7.701763388467953e-05,
          "train_metric": 8.43818997964263e-06,
          "validation_metric": 7.756934792269021e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.73195467000005
        },
        {
          "epoch": 754,
          "updates": 12064,
          "samples_seen": 772096,
          "online_loss": 9.468979897064855e-05,
          "train_loss": 8.69331561261788e-05,
          "validation_loss": 8.112465729936957e-05,
          "train_metric": 8.755589988140855e-06,
          "validation_metric": 8.17057843960356e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.742340896000314
        },
        {
          "epoch": 755,
          "updates": 12080,
          "samples_seen": 773120,
          "online_loss": 8.473014531773515e-05,
          "train_loss": 8.18361877463758e-05,
          "validation_loss": 7.587771688122302e-05,
          "train_metric": 8.242240255640354e-06,
          "validation_metric": 7.642125638085417e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.75309958499929
        },
        {
          "epoch": 756,
          "updates": 12096,
          "samples_seen": 774144,
          "online_loss": 8.453390455542831e-05,
          "train_loss": 8.194430847652256e-05,
          "validation_loss": 7.61091650929302e-05,
          "train_metric": 8.253130545199383e-06,
          "validation_metric": 7.665435987291858e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.763501621000614
        },
        {
          "epoch": 757,
          "updates": 12112,
          "samples_seen": 775168,
          "online_loss": 8.846370610626764e-05,
          "train_loss": 9.00182276382111e-05,
          "validation_loss": 8.341771899722517e-05,
          "train_metric": 9.066306120075751e-06,
          "validation_metric": 8.401527338719461e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.773718230999293
        },
        {
          "epoch": 758,
          "updates": 12128,
          "samples_seen": 776192,
          "online_loss": 8.658522256155265e-05,
          "train_loss": 8.194628026103601e-05,
          "validation_loss": 7.567207649117336e-05,
          "train_metric": 8.25332881504437e-06,
          "validation_metric": 7.621414624736644e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.784490271000323
        },
        {
          "epoch": 759,
          "updates": 12144,
          "samples_seen": 777216,
          "online_loss": 8.347891866833379e-05,
          "train_loss": 8.196241833502427e-05,
          "validation_loss": 7.559305231552571e-05,
          "train_metric": 8.254954082076438e-06,
          "validation_metric": 7.613454727106728e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.79535291799948
        },
        {
          "epoch": 760,
          "updates": 12160,
          "samples_seen": 778240,
          "online_loss": 8.35075893519388e-05,
          "train_loss": 8.116647222777829e-05,
          "validation_loss": 7.531270239269361e-05,
          "train_metric": 8.174789400072768e-06,
          "validation_metric": 7.585219464090187e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.80637596700035
        },
        {
          "epoch": 761,
          "updates": 12176,
          "samples_seen": 779264,
          "online_loss": 8.34260954434285e-05,
          "train_loss": 8.173472451744601e-05,
          "validation_loss": 7.667218596907333e-05,
          "train_metric": 8.232022082665935e-06,
          "validation_metric": 7.722142072452698e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.816694088000077
        },
        {
          "epoch": 762,
          "updates": 12192,
          "samples_seen": 780288,
          "online_loss": 8.376340406357485e-05,
          "train_loss": 8.511485793860629e-05,
          "validation_loss": 7.816854486009106e-05,
          "train_metric": 8.572456863475963e-06,
          "validation_metric": 7.872849892009981e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.826982616999885
        },
        {
          "epoch": 763,
          "updates": 12208,
          "samples_seen": 781312,
          "online_loss": 8.529372303200944e-05,
          "train_loss": 8.163026359397918e-05,
          "validation_loss": 7.53008498577401e-05,
          "train_metric": 8.221500138461124e-06,
          "validation_metric": 7.584026207041461e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.891724298000554
        },
        {
          "epoch": 764,
          "updates": 12224,
          "samples_seen": 782336,
          "online_loss": 8.364288032680633e-05,
          "train_loss": 8.183958561858162e-05,
          "validation_loss": 7.501299114665017e-05,
          "train_metric": 8.242583135142922e-06,
          "validation_metric": 7.555034699180396e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.902929449999647
        },
        {
          "epoch": 765,
          "updates": 12240,
          "samples_seen": 783360,
          "online_loss": 8.305394476337824e-05,
          "train_loss": 8.09603079687804e-05,
          "validation_loss": 7.420068141072989e-05,
          "train_metric": 8.15402472653659e-06,
          "validation_metric": 7.4732211032824125e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.913880370999323
        },
        {
          "epoch": 766,
          "updates": 12256,
          "samples_seen": 784384,
          "online_loss": 8.38919736452226e-05,
          "train_loss": 8.113757212413475e-05,
          "validation_loss": 7.529163849540055e-05,
          "train_metric": 8.171880836016499e-06,
          "validation_metric": 7.583097612950951e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.924442198000179
        },
        {
          "epoch": 767,
          "updates": 12272,
          "samples_seen": 785408,
          "online_loss": 8.300734998556436e-05,
          "train_loss": 8.306954259751365e-05,
          "validation_loss": 7.62611161917448e-05,
          "train_metric": 8.3664590420085e-06,
          "validation_metric": 7.680740964133292e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.934691648999433
        },
        {
          "epoch": 768,
          "updates": 12288,
          "samples_seen": 786432,
          "online_loss": 8.285015769615711e-05,
          "train_loss": 8.853442705003545e-05,
          "validation_loss": 7.995991472853348e-05,
          "train_metric": 8.916864317143336e-06,
          "validation_metric": 8.053270903474186e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.945160294999368
        },
        {
          "epoch": 769,
          "updates": 12304,
          "samples_seen": 787456,
          "online_loss": 8.736171866985387e-05,
          "train_loss": 8.232989785028622e-05,
          "validation_loss": 7.46080040698871e-05,
          "train_metric": 8.291966878459789e-06,
          "validation_metric": 7.514245226047933e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.955365550000352
        },
        {
          "epoch": 770,
          "updates": 12320,
          "samples_seen": 788480,
          "online_loss": 9.150616733677452e-05,
          "train_loss": 0.00010168849257752299,
          "validation_loss": 8.935764344641939e-05,
          "train_metric": 1.0241692507406697e-05,
          "validation_metric": 8.999774763651658e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.965697573999933
        },
        {
          "epoch": 771,
          "updates": 12336,
          "samples_seen": 789504,
          "online_loss": 9.283138547289127e-05,
          "train_loss": 8.146819163812324e-05,
          "validation_loss": 7.4505660450086e-05,
          "train_metric": 8.205177437048405e-06,
          "validation_metric": 7.5039379225927405e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.97600608799985
        },
        {
          "epoch": 772,
          "updates": 12352,
          "samples_seen": 790528,
          "online_loss": 8.850848666952515e-05,
          "train_loss": 9.463007882004604e-05,
          "validation_loss": 8.710224210517481e-05,
          "train_metric": 9.530795068712905e-06,
          "validation_metric": 8.772617547947448e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.986322517999724
        },
        {
          "epoch": 773,
          "updates": 12368,
          "samples_seen": 791552,
          "online_loss": 9.156823216471821e-05,
          "train_loss": 8.531285129720345e-05,
          "validation_loss": 7.790954259689897e-05,
          "train_metric": 8.592398444307037e-06,
          "validation_metric": 7.84676376497373e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 8.996556854999653
        },
        {
          "epoch": 774,
          "updates": 12384,
          "samples_seen": 792576,
          "online_loss": 8.370352566089423e-05,
          "train_loss": 8.451344911009073e-05,
          "validation_loss": 7.883763464633375e-05,
          "train_metric": 8.511884516337886e-06,
          "validation_metric": 7.940237082948443e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.006890528999065
        },
        {
          "epoch": 775,
          "updates": 12400,
          "samples_seen": 793600,
          "online_loss": 8.260655295089236e-05,
          "train_loss": 8.270400576293468e-05,
          "validation_loss": 7.55811488488689e-05,
          "train_metric": 8.329644515470136e-06,
          "validation_metric": 7.612256922584493e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.017113985999458
        },
        {
          "epoch": 776,
          "updates": 12416,
          "samples_seen": 794624,
          "online_loss": 8.510844986631128e-05,
          "train_loss": 8.03079383331351e-05,
          "validation_loss": 7.506622205255553e-05,
          "train_metric": 8.08832101029111e-06,
          "validation_metric": 7.560394806205295e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.027443818000393
        },
        {
          "epoch": 777,
          "updates": 12432,
          "samples_seen": 795648,
          "online_loss": 8.20897223547945e-05,
          "train_loss": 8.053195051616058e-05,
          "validation_loss": 7.360260497080162e-05,
          "train_metric": 8.110882845357992e-06,
          "validation_metric": 7.41298435968929e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.038207387000512
        },
        {
          "epoch": 778,
          "updates": 12448,
          "samples_seen": 796672,
          "online_loss": 8.462284131383058e-05,
          "train_loss": 8.136559335980564e-05,
          "validation_loss": 7.553053728770465e-05,
          "train_metric": 8.194844667741563e-06,
          "validation_metric": 7.6071592047810555e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.048720022999987
        },
        {
          "epoch": 779,
          "updates": 12464,
          "samples_seen": 797696,
          "online_loss": 9.021323353408661e-05,
          "train_loss": 9.71076515270397e-05,
          "validation_loss": 8.459194214083254e-05,
          "train_metric": 9.780327673070133e-06,
          "validation_metric": 8.5197916632751e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.058959027999663
        },
        {
          "epoch": 780,
          "updates": 12480,
          "samples_seen": 798720,
          "online_loss": 8.633176457806258e-05,
          "train_loss": 8.158652053680271e-05,
          "validation_loss": 7.333639950957149e-05,
          "train_metric": 8.217095455620438e-06,
          "validation_metric": 7.3861738201230764e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.069966077999197
        },
        {
          "epoch": 781,
          "updates": 12496,
          "samples_seen": 799744,
          "online_loss": 8.343385479747667e-05,
          "train_loss": 8.409778092755005e-05,
          "validation_loss": 7.643733260920271e-05,
          "train_metric": 8.470020475215279e-06,
          "validation_metric": 7.698487934248988e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.080299171999286
        },
        {
          "epoch": 782,
          "updates": 12512,
          "samples_seen": 800768,
          "online_loss": 8.256254454863665e-05,
          "train_loss": 8.366208930965513e-05,
          "validation_loss": 7.631506014149636e-05,
          "train_metric": 8.426138265349437e-06,
          "validation_metric": 7.686174285481684e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.09065047600052
        },
        {
          "epoch": 783,
          "updates": 12528,
          "samples_seen": 801792,
          "online_loss": 8.795362464297796e-05,
          "train_loss": 8.146878826664761e-05,
          "validation_loss": 7.635775546077639e-05,
          "train_metric": 8.205237463698722e-06,
          "validation_metric": 7.690473466936965e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.101188356999046
        },
        {
          "epoch": 784,
          "updates": 12544,
          "samples_seen": 802816,
          "online_loss": 8.271055389741377e-05,
          "train_loss": 0.00010120595106855035,
          "validation_loss": 8.78743885550648e-05,
          "train_metric": 1.019309365801746e-05,
          "validation_metric": 8.850385711411946e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.111528290999559
        },
        {
          "epoch": 785,
          "updates": 12560,
          "samples_seen": 803840,
          "online_loss": 8.918056710172095e-05,
          "train_loss": 8.078588143689558e-05,
          "validation_loss": 7.564515544800088e-05,
          "train_metric": 8.136458745866548e-06,
          "validation_metric": 7.618702966283308e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.12181062200034
        },
        {
          "epoch": 786,
          "updates": 12576,
          "samples_seen": 804864,
          "online_loss": 8.230880166593124e-05,
          "train_loss": 7.972399180289358e-05,
          "validation_loss": 7.362350879702717e-05,
          "train_metric": 8.029508535400964e-06,
          "validation_metric": 7.415089839923894e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.13217071500003
        },
        {
          "epoch": 787,
          "updates": 12592,
          "samples_seen": 805888,
          "online_loss": 8.195052532755653e-05,
          "train_loss": 8.208382496377453e-05,
          "validation_loss": 7.355792331509292e-05,
          "train_metric": 8.267182238341775e-06,
          "validation_metric": 7.40848554414697e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.142488199999207
        },
        {
          "epoch": 788,
          "updates": 12608,
          "samples_seen": 806912,
          "online_loss": 8.891140964806254e-05,
          "train_loss": 9.499439329374582e-05,
          "validation_loss": 8.297607564600185e-05,
          "train_metric": 9.56748681346653e-06,
          "validation_metric": 8.357046681339853e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.152788403000159
        },
        {
          "epoch": 789,
          "updates": 12624,
          "samples_seen": 807936,
          "online_loss": 0.00010118396880898217,
          "train_loss": 8.042581612244248e-05,
          "validation_loss": 7.397946319542825e-05,
          "train_metric": 8.100193554128055e-06,
          "validation_metric": 7.45094075682573e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.16309580000052
        },
        {
          "epoch": 790,
          "updates": 12640,
          "samples_seen": 808960,
          "online_loss": 0.00010097019867316703,
          "train_loss": 0.00011527277092682198,
          "validation_loss": 0.00010199740063399076,
          "train_metric": 1.1609850844251923e-05,
          "validation_metric": 1.0272804502164945e-05,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.173413004000395
        },
        {
          "epoch": 791,
          "updates": 12656,
          "samples_seen": 809984,
          "online_loss": 9.040478312272171e-05,
          "train_loss": 8.420093945460394e-05,
          "validation_loss": 7.390353130176663e-05,
          "train_metric": 8.480410542688332e-06,
          "validation_metric": 7.443292815878522e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.18367298299927
        },
        {
          "epoch": 792,
          "updates": 12672,
          "samples_seen": 811008,
          "online_loss": 8.333551113537396e-05,
          "train_loss": 8.261282346211374e-05,
          "validation_loss": 7.571464084321633e-05,
          "train_metric": 8.320462256961036e-06,
          "validation_metric": 7.625701528013451e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.193981505999545
        },
        {
          "epoch": 793,
          "updates": 12688,
          "samples_seen": 812032,
          "online_loss": 8.238483519562578e-05,
          "train_loss": 7.975423068273813e-05,
          "validation_loss": 7.414801075356081e-05,
          "train_metric": 8.0325535236625e-06,
          "validation_metric": 7.467916475434322e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.204198343999451
        },
        {
          "epoch": 794,
          "updates": 12704,
          "samples_seen": 813056,
          "online_loss": 8.324102964252234e-05,
          "train_loss": 8.494102803524584e-05,
          "validation_loss": 7.809628004906699e-05,
          "train_metric": 8.554950909456238e-06,
          "validation_metric": 7.865571205911692e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.214504903999114
        },
        {
          "epoch": 795,
          "updates": 12720,
          "samples_seen": 814080,
          "online_loss": 8.169693887793983e-05,
          "train_loss": 8.243579941336066e-05,
          "validation_loss": 7.375786663033068e-05,
          "train_metric": 8.30263252282748e-06,
          "validation_metric": 7.428622666338924e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.22477720399911
        },
        {
          "epoch": 796,
          "updates": 12736,
          "samples_seen": 815104,
          "online_loss": 8.319992298311263e-05,
          "train_loss": 7.945051766000688e-05,
          "validation_loss": 7.355812704190612e-05,
          "train_metric": 8.001965397852473e-06,
          "validation_metric": 7.408504643535707e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.235054909999235
        },
        {
          "epoch": 797,
          "updates": 12752,
          "samples_seen": 816128,
          "online_loss": 8.169287298187555e-05,
          "train_loss": 8.148798951879144e-05,
          "validation_loss": 7.45545476092957e-05,
          "train_metric": 8.207172868424095e-06,
          "validation_metric": 7.508861472160788e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.245434479000323
        },
        {
          "epoch": 798,
          "updates": 12768,
          "samples_seen": 817152,
          "online_loss": 8.38896098684927e-05,
          "train_loss": 8.507119491696358e-05,
          "validation_loss": 7.830812683096156e-05,
          "train_metric": 8.568059456592891e-06,
          "validation_metric": 7.886907042120583e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.255802239998957
        },
        {
          "epoch": 799,
          "updates": 12784,
          "samples_seen": 818176,
          "online_loss": 8.561243021176779e-05,
          "train_loss": 7.961092342156917e-05,
          "validation_loss": 7.289869245141745e-05,
          "train_metric": 8.018120752240065e-06,
          "validation_metric": 7.342088792938739e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.322991574999833
        },
        {
          "epoch": 800,
          "updates": 12800,
          "samples_seen": 819200,
          "online_loss": 8.966264545051672e-05,
          "train_loss": 0.00010063419176731259,
          "validation_loss": 8.7690532382112e-05,
          "train_metric": 1.0135507181985304e-05,
          "validation_metric": 8.831868399283849e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.333732208000583
        },
        {
          "epoch": 801,
          "updates": 12816,
          "samples_seen": 820224,
          "online_loss": 8.82688161709666e-05,
          "train_loss": 7.906086830189452e-05,
          "validation_loss": 7.308945350814611e-05,
          "train_metric": 7.962721610965673e-06,
          "validation_metric": 7.361302323261043e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.344393053999738
        },
        {
          "epoch": 802,
          "updates": 12832,
          "samples_seen": 821248,
          "online_loss": 8.388373180423514e-05,
          "train_loss": 8.834994514472783e-05,
          "validation_loss": 7.673347136005759e-05,
          "train_metric": 8.898282430891413e-06,
          "validation_metric": 7.728313903498929e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.354723651998938
        },
        {
          "epoch": 803,
          "updates": 12848,
          "samples_seen": 822272,
          "online_loss": 8.373432797270652e-05,
          "train_loss": 7.884074148023501e-05,
          "validation_loss": 7.287424523383379e-05,
          "train_metric": 7.940549949125852e-06,
          "validation_metric": 7.339627245528391e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.365438958000595
        },
        {
          "epoch": 804,
          "updates": 12864,
          "samples_seen": 823296,
          "online_loss": 8.49537966587377e-05,
          "train_loss": 7.98516339273192e-05,
          "validation_loss": 7.360642484854907e-05,
          "train_metric": 8.042364243010525e-06,
          "validation_metric": 7.4133695306954905e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.376001958000415
        },
        {
          "epoch": 805,
          "updates": 12880,
          "samples_seen": 824320,
          "online_loss": 8.321082736983954e-05,
          "train_loss": 7.913017907412723e-05,
          "validation_loss": 7.297030970221385e-05,
          "train_metric": 7.96970198280178e-06,
          "validation_metric": 7.3493019954185e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.386649021000267
        },
        {
          "epoch": 806,
          "updates": 12896,
          "samples_seen": 825344,
          "online_loss": 8.065425208769739e-05,
          "train_loss": 7.939585339045152e-05,
          "validation_loss": 7.342449680436403e-05,
          "train_metric": 7.996458407433238e-06,
          "validation_metric": 7.395046850433573e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.396938003999821
        },
        {
          "epoch": 807,
          "updates": 12912,
          "samples_seen": 826368,
          "online_loss": 8.42400206693128e-05,
          "train_loss": 9.090953244594857e-05,
          "validation_loss": 7.571651804028079e-05,
          "train_metric": 9.156074156635441e-06,
          "validation_metric": 7.6258920671534725e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.407220421999227
        },
        {
          "epoch": 808,
          "updates": 12928,
          "samples_seen": 827392,
          "online_loss": 8.589588787799585e-05,
          "train_loss": 7.987680146470666e-05,
          "validation_loss": 7.541507511632517e-05,
          "train_metric": 8.044898095249664e-06,
          "validation_metric": 7.595529950776836e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.417648908000047
        },
        {
          "epoch": 809,
          "updates": 12944,
          "samples_seen": 828416,
          "online_loss": 8.897866041479574e-05,
          "train_loss": 9.457320265937597e-05,
          "validation_loss": 8.538347901776433e-05,
          "train_metric": 9.525066161586437e-06,
          "validation_metric": 8.599511602369603e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.427908109999407
        },
        {
          "epoch": 810,
          "updates": 12960,
          "samples_seen": 829440,
          "online_loss": 9.193121172756946e-05,
          "train_loss": 8.390931907342747e-05,
          "validation_loss": 7.969664875417948e-05,
          "train_metric": 8.451039320789278e-06,
          "validation_metric": 8.026754585443996e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.438152251999782
        },
        {
          "epoch": 811,
          "updates": 12976,
          "samples_seen": 830464,
          "online_loss": 9.270500368074863e-05,
          "train_loss": 7.88704928709194e-05,
          "validation_loss": 7.292659574886784e-05,
          "train_metric": 7.943545824673492e-06,
          "validation_metric": 7.3448986768198665e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.44859504699889
        },
        {
          "epoch": 812,
          "updates": 12992,
          "samples_seen": 831488,
          "online_loss": 8.10462186109362e-05,
          "train_loss": 8.028405136428773e-05,
          "validation_loss": 7.295498653547838e-05,
          "train_metric": 8.085916306299623e-06,
          "validation_metric": 7.3477585829095915e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.458994550999705
        },
        {
          "epoch": 813,
          "updates": 13008,
          "samples_seen": 832512,
          "online_loss": 9.039602286975423e-05,
          "train_loss": 8.088028698693961e-05,
          "validation_loss": 7.507205009460449e-05,
          "train_metric": 8.14596569398418e-06,
          "validation_metric": 7.56098233978264e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.469251219999933
        },
        {
          "epoch": 814,
          "updates": 13024,
          "samples_seen": 833536,
          "online_loss": 8.519202037859941e-05,
          "train_loss": 8.199727744795382e-05,
          "validation_loss": 7.547796121798456e-05,
          "train_metric": 8.258464731625281e-06,
          "validation_metric": 7.601864581374684e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.479689498999505
        },
        {
          "epoch": 815,
          "updates": 13040,
          "samples_seen": 834560,
          "online_loss": 8.818481228445307e-05,
          "train_loss": 9.14595220820047e-05,
          "validation_loss": 7.60595576139167e-05,
          "train_metric": 9.211468750436325e-06,
          "validation_metric": 7.66044104238972e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.49004005200004
        },
        {
          "epoch": 816,
          "updates": 13056,
          "samples_seen": 835584,
          "online_loss": 8.514196360920323e-05,
          "train_loss": 7.860126061132178e-05,
          "validation_loss": 7.42431147955358e-05,
          "train_metric": 7.916431059129536e-06,
          "validation_metric": 7.477494364138693e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.500326216999383
        },
        {
          "epoch": 817,
          "updates": 13072,
          "samples_seen": 836608,
          "online_loss": 8.300560489260533e-05,
          "train_loss": 8.097453974187374e-05,
          "validation_loss": 7.530294533353299e-05,
          "train_metric": 8.155458999681287e-06,
          "validation_metric": 7.5842367550649215e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.510760967999886
        },
        {
          "epoch": 818,
          "updates": 13088,
          "samples_seen": 837632,
          "online_loss": 8.026748309930554e-05,
          "train_loss": 8.148046617861837e-05,
          "validation_loss": 7.41418480174616e-05,
          "train_metric": 8.206414349842817e-06,
          "validation_metric": 7.467295290553011e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.521108331000505
        },
        {
          "epoch": 819,
          "updates": 13104,
          "samples_seen": 838656,
          "online_loss": 8.476604648421926e-05,
          "train_loss": 8.004886331036687e-05,
          "validation_loss": 7.344200275838375e-05,
          "train_metric": 8.062228516791947e-06,
          "validation_metric": 7.396809451165609e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.531347829999504
        },
        {
          "epoch": 820,
          "updates": 13120,
          "samples_seen": 839680,
          "online_loss": 8.548886603421124e-05,
          "train_loss": 7.90514750406146e-05,
          "validation_loss": 7.271140930242836e-05,
          "train_metric": 7.961775736475829e-06,
          "validation_metric": 7.323227237066021e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.542339044999608
        },
        {
          "epoch": 821,
          "updates": 13136,
          "samples_seen": 840704,
          "online_loss": 8.389293952859589e-05,
          "train_loss": 8.19257547846064e-05,
          "validation_loss": 7.521720544900745e-05,
          "train_metric": 8.25126153358724e-06,
          "validation_metric": 7.575601557618938e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.552690776999953
        },
        {
          "epoch": 822,
          "updates": 13152,
          "samples_seen": 841728,
          "online_loss": 8.244013861258281e-05,
          "train_loss": 7.999360968824476e-05,
          "validation_loss": 7.31382897356525e-05,
          "train_metric": 8.0566642282065e-06,
          "validation_metric": 7.366220870608231e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.563171255998896
        },
        {
          "epoch": 823,
          "updates": 13168,
          "samples_seen": 842752,
          "online_loss": 8.420904396189144e-05,
          "train_loss": 8.795186295174062e-05,
          "validation_loss": 8.207316568586975e-05,
          "train_metric": 8.858188266458455e-06,
          "validation_metric": 8.266109944088385e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.57353458199941
        },
        {
          "epoch": 824,
          "updates": 13184,
          "samples_seen": 843776,
          "online_loss": 8.31073025437945e-05,
          "train_loss": 7.796620047884062e-05,
          "validation_loss": 7.222330168588087e-05,
          "train_metric": 7.852470844227355e-06,
          "validation_metric": 7.274066319951089e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.584751673999563
        },
        {
          "epoch": 825,
          "updates": 13200,
          "samples_seen": 844800,
          "online_loss": 8.035482164814312e-05,
          "train_loss": 8.192182576749474e-05,
          "validation_loss": 7.452072168234736e-05,
          "train_metric": 8.25086681288667e-06,
          "validation_metric": 7.505454050260596e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.595174939000572
        },
        {
          "epoch": 826,
          "updates": 13216,
          "samples_seen": 845824,
          "online_loss": 8.518427353010338e-05,
          "train_loss": 8.075746154645458e-05,
          "validation_loss": 7.292843656614423e-05,
          "train_metric": 8.133594747050665e-06,
          "validation_metric": 7.345084668486379e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.605565351999758
        },
        {
          "epoch": 827,
          "updates": 13232,
          "samples_seen": 846848,
          "online_loss": 8.3518233623181e-05,
          "train_loss": 8.702075137989596e-05,
          "validation_loss": 7.478702173102647e-05,
          "train_metric": 8.76441117725335e-06,
          "validation_metric": 7.532275958510581e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.615912465000292
        },
        {
          "epoch": 828,
          "updates": 13248,
          "samples_seen": 847872,
          "online_loss": 8.92193465915625e-05,
          "train_loss": 8.636971324449405e-05,
          "validation_loss": 7.700275455135852e-05,
          "train_metric": 8.698841156729031e-06,
          "validation_metric": 7.755435035505798e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.627049107999483
        },
        {
          "epoch": 829,
          "updates": 13264,
          "samples_seen": 848896,
          "online_loss": 7.993206008904963e-05,
          "train_loss": 8.161834557540715e-05,
          "validation_loss": 7.22568147466518e-05,
          "train_metric": 8.220300514949486e-06,
          "validation_metric": 7.277441909536719e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.637266529000044
        },
        {
          "epoch": 830,
          "updates": 13280,
          "samples_seen": 849920,
          "online_loss": 8.282880480692256e-05,
          "train_loss": 7.946636469569057e-05,
          "validation_loss": 7.296716648852453e-05,
          "train_metric": 8.003561561054084e-06,
          "validation_metric": 7.348985036514932e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.647715994000464
        },
        {
          "epoch": 831,
          "updates": 13296,
          "samples_seen": 850944,
          "online_loss": 7.933017923278385e-05,
          "train_loss": 7.785404159221798e-05,
          "validation_loss": 7.164743874454871e-05,
          "train_metric": 7.841174010536633e-06,
          "validation_metric": 7.216066933324328e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.65858647999994
        },
        {
          "epoch": 832,
          "updates": 13312,
          "samples_seen": 851968,
          "online_loss": 8.0215076877721e-05,
          "train_loss": 7.739213469903916e-05,
          "validation_loss": 7.10099411662668e-05,
          "train_metric": 7.794652447046246e-06,
          "validation_metric": 7.151860700105317e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.669616068000323
        },
        {
          "epoch": 833,
          "updates": 13328,
          "samples_seen": 852992,
          "online_loss": 8.180066993190849e-05,
          "train_loss": 8.586804324295372e-05,
          "validation_loss": 7.32118496671319e-05,
          "train_metric": 8.648314178572036e-06,
          "validation_metric": 7.3736291597015224e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.680232681999769
        },
        {
          "epoch": 834,
          "updates": 13344,
          "samples_seen": 854016,
          "online_loss": 9.238574421033263e-05,
          "train_loss": 8.198490832000971e-05,
          "validation_loss": 7.33511260477826e-05,
          "train_metric": 8.257221452367958e-06,
          "validation_metric": 7.387656751234317e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.69064984400029
        },
        {
          "epoch": 835,
          "updates": 13360,
          "samples_seen": 855040,
          "online_loss": 8.281126656584092e-05,
          "train_loss": 8.20331260911189e-05,
          "validation_loss": 7.85844968049787e-05,
          "train_metric": 8.262076335086022e-06,
          "validation_metric": 7.914743036963046e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.755629211000269
        },
        {
          "epoch": 836,
          "updates": 13376,
          "samples_seen": 856064,
          "online_loss": 8.088354275059828e-05,
          "train_loss": 7.739481225144118e-05,
          "validation_loss": 7.065106183290482e-05,
          "train_metric": 7.79492074798327e-06,
          "validation_metric": 7.11571692590951e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.767090068999096
        },
        {
          "epoch": 837,
          "updates": 13392,
          "samples_seen": 857088,
          "online_loss": 8.037507086555706e-05,
          "train_loss": 8.135466487146914e-05,
          "validation_loss": 7.366665522567928e-05,
          "train_metric": 8.193743269657716e-06,
          "validation_metric": 7.419435405608965e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.777416791999713
        },
        {
          "epoch": 838,
          "updates": 13408,
          "samples_seen": 858112,
          "online_loss": 7.937006671454583e-05,
          "train_loss": 8.072212222032249e-05,
          "validation_loss": 7.164894486777484e-05,
          "train_metric": 8.130035894282628e-06,
          "validation_metric": 7.216219273686875e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.787828115999218
        },
        {
          "epoch": 839,
          "updates": 13424,
          "samples_seen": 859136,
          "online_loss": 8.062262031671708e-05,
          "train_loss": 7.735004328424111e-05,
          "validation_loss": 7.145394192775711e-05,
          "train_metric": 7.790413292241283e-06,
          "validation_metric": 7.19657964509679e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.798099970999829
        },
        {
          "epoch": 840,
          "updates": 13440,
          "samples_seen": 860160,
          "online_loss": 7.885631566750817e-05,
          "train_loss": 7.708172779530287e-05,
          "validation_loss": 7.01391400070861e-05,
          "train_metric": 7.763389476167504e-06,
          "validation_metric": 7.064157671266003e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.809073456999613
        },
        {
          "epoch": 841,
          "updates": 13456,
          "samples_seen": 861184,
          "online_loss": 8.017671302695817e-05,
          "train_loss": 7.722109148744494e-05,
          "validation_loss": 7.215880032163113e-05,
          "train_metric": 7.777426617394667e-06,
          "validation_metric": 7.267570254043676e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.81952498300052
        },
        {
          "epoch": 842,
          "updates": 13472,
          "samples_seen": 862208,
          "online_loss": 7.896785882621771e-05,
          "train_loss": 7.748022471787408e-05,
          "validation_loss": 6.990525434957817e-05,
          "train_metric": 7.803524567862041e-06,
          "validation_metric": 7.040600848995382e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.83044063899979
        },
        {
          "epoch": 843,
          "updates": 13488,
          "samples_seen": 863232,
          "online_loss": 7.973545916684088e-05,
          "train_loss": 8.188439096556976e-05,
          "validation_loss": 7.518669735873118e-05,
          "train_metric": 8.24709604785312e-06,
          "validation_metric": 7.5725292845163494e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.84098540299965
        },
        {
          "epoch": 844,
          "updates": 13504,
          "samples_seen": 864256,
          "online_loss": 8.198653631552588e-05,
          "train_loss": 7.797821308486164e-05,
          "validation_loss": 7.042171637294814e-05,
          "train_metric": 7.853679562686011e-06,
          "validation_metric": 7.09261712472653e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.851515685999402
        },
        {
          "epoch": 845,
          "updates": 13520,
          "samples_seen": 865280,
          "online_loss": 8.060248296715145e-05,
          "train_loss": 8.128127228701487e-05,
          "validation_loss": 7.271073991432786e-05,
          "train_metric": 8.186351806216408e-06,
          "validation_metric": 7.323159024963388e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.861786440000287
        },
        {
          "epoch": 846,
          "updates": 13536,
          "samples_seen": 866304,
          "online_loss": 8.216931587412546e-05,
          "train_loss": 7.77378081693314e-05,
          "validation_loss": 7.180780085036531e-05,
          "train_metric": 7.829467904230114e-06,
          "validation_metric": 7.232219104480464e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.87242661199889
        },
        {
          "epoch": 847,
          "updates": 13552,
          "samples_seen": 867328,
          "online_loss": 8.397038664043066e-05,
          "train_loss": 7.672331412322819e-05,
          "validation_loss": 6.973979179747403e-05,
          "train_metric": 7.727291631454136e-06,
          "validation_metric": 7.0239366323221475e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.883492699998897
        },
        {
          "epoch": 848,
          "updates": 13568,
          "samples_seen": 868352,
          "online_loss": 7.802875643392326e-05,
          "train_loss": 7.816844299668446e-05,
          "validation_loss": 7.116422784747556e-05,
          "train_metric": 7.872839887568261e-06,
          "validation_metric": 7.167400326579809e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.893919782000012
        },
        {
          "epoch": 849,
          "updates": 13584,
          "samples_seen": 869376,
          "online_loss": 7.903386017460434e-05,
          "train_loss": 7.682674913667142e-05,
          "validation_loss": 7.022613863227889e-05,
          "train_metric": 7.737708074273542e-06,
          "validation_metric": 7.07291837898083e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.90424284800065
        },
        {
          "epoch": 850,
          "updates": 13600,
          "samples_seen": 870400,
          "online_loss": 7.946486562104838e-05,
          "train_loss": 7.647809979971498e-05,
          "validation_loss": 6.974768621148542e-05,
          "train_metric": 7.702592483838089e-06,
          "validation_metric": 7.024731530691497e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.914544392000607
        },
        {
          "epoch": 851,
          "updates": 13616,
          "samples_seen": 871424,
          "online_loss": 7.854015871089359e-05,
          "train_loss": 7.66812008805573e-05,
          "validation_loss": 6.993734132265672e-05,
          "train_metric": 7.723049748165067e-06,
          "validation_metric": 7.043832738418132e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.924865875998876
        },
        {
          "epoch": 852,
          "updates": 13632,
          "samples_seen": 872448,
          "online_loss": 8.089639072750288e-05,
          "train_loss": 8.682703628437594e-05,
          "validation_loss": 7.717634434811771e-05,
          "train_metric": 8.744900696910918e-06,
          "validation_metric": 7.77291916165268e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.935155753999425
        },
        {
          "epoch": 853,
          "updates": 13648,
          "samples_seen": 873472,
          "online_loss": 8.442584271506348e-05,
          "train_loss": 7.701330468989909e-05,
          "validation_loss": 7.023328362265602e-05,
          "train_metric": 7.75649823481217e-06,
          "validation_metric": 7.073639608279336e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.945365903999118
        },
        {
          "epoch": 854,
          "updates": 13664,
          "samples_seen": 874496,
          "online_loss": 9.52709508510452e-05,
          "train_loss": 7.975332118803635e-05,
          "validation_loss": 7.049980922602117e-05,
          "train_metric": 8.032462574192323e-06,
          "validation_metric": 7.100483344402164e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.955663040000218
        },
        {
          "epoch": 855,
          "updates": 13680,
          "samples_seen": 875520,
          "online_loss": 7.866129499234376e-05,
          "train_loss": 7.963836833368987e-05,
          "validation_loss": 7.011130946921185e-05,
          "train_metric": 8.020884706638753e-06,
          "validation_metric": 7.061355063342489e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.965823295000519
        },
        {
          "epoch": 856,
          "updates": 13696,
          "samples_seen": 876544,
          "online_loss": 8.249340521615522e-05,
          "train_loss": 8.39804342831485e-05,
          "validation_loss": 7.706711767241359e-05,
          "train_metric": 8.45820159156574e-06,
          "validation_metric": 7.761917004245333e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.976116157999058
        },
        {
          "epoch": 857,
          "updates": 13712,
          "samples_seen": 877568,
          "online_loss": 8.509281428814575e-05,
          "train_loss": 7.708053453825414e-05,
          "validation_loss": 6.94086411385797e-05,
          "train_metric": 7.76326942286687e-06,
          "validation_metric": 6.9905845521134324e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.986855204000676
        },
        {
          "epoch": 858,
          "updates": 13728,
          "samples_seen": 878592,
          "online_loss": 8.329565525855287e-05,
          "train_loss": 9.446862532058731e-05,
          "validation_loss": 7.780543819535524e-05,
          "train_metric": 9.514534212939907e-06,
          "validation_metric": 7.83627820055699e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 9.997232403000453
        },
        {
          "epoch": 859,
          "updates": 13744,
          "samples_seen": 879616,
          "online_loss": 8.951140080171172e-05,
          "train_loss": 7.811047544237226e-05,
          "validation_loss": 7.271231879713014e-05,
          "train_metric": 7.867001841077581e-06,
          "validation_metric": 7.3233181865361985e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.007499850999011
        },
        {
          "epoch": 860,
          "updates": 13760,
          "samples_seen": 880640,
          "online_loss": 7.866906798881246e-05,
          "train_loss": 7.616651419084519e-05,
          "validation_loss": 6.872743688290939e-05,
          "train_metric": 7.671212188142817e-06,
          "validation_metric": 6.92197590979049e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.018362139999226
        },
        {
          "epoch": 861,
          "updates": 13776,
          "samples_seen": 881664,
          "online_loss": 7.790820927766617e-05,
          "train_loss": 7.642351556569338e-05,
          "validation_loss": 7.043880032142624e-05,
          "train_metric": 7.697097316849977e-06,
          "validation_metric": 7.094338798196986e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.028989708000154
        },
        {
          "epoch": 862,
          "updates": 13792,
          "samples_seen": 882688,
          "online_loss": 7.802426762282266e-05,
          "train_loss": 7.732850644970313e-05,
          "validation_loss": 6.990537804085761e-05,
          "train_metric": 7.788245056872256e-06,
          "validation_metric": 7.040613581921207e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.039280061000682
        },
        {
          "epoch": 863,
          "updates": 13808,
          "samples_seen": 883712,
          "online_loss": 7.814285436325008e-05,
          "train_loss": 7.616534276166931e-05,
          "validation_loss": 6.94409100105986e-05,
          "train_metric": 7.671093953831587e-06,
          "validation_metric": 6.993833721935516e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.049457691999123
        },
        {
          "epoch": 864,
          "updates": 13824,
          "samples_seen": 884736,
          "online_loss": 7.838220039957378e-05,
          "train_loss": 7.648042810615152e-05,
          "validation_loss": 6.913924153195694e-05,
          "train_metric": 7.70282895246055e-06,
          "validation_metric": 6.96345114192809e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.059744506999778
        },
        {
          "epoch": 865,
          "updates": 13840,
          "samples_seen": 885760,
          "online_loss": 7.80032269176445e-05,
          "train_loss": 7.607714360347018e-05,
          "validation_loss": 6.998263415880501e-05,
          "train_metric": 7.66221091907937e-06,
          "validation_metric": 7.048393854347523e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.069879233999018
        },
        {
          "epoch": 866,
          "updates": 13856,
          "samples_seen": 886784,
          "online_loss": 7.87574124387902e-05,
          "train_loss": 7.662175630684942e-05,
          "validation_loss": 7.006956730037928e-05,
          "train_metric": 7.717061635048594e-06,
          "validation_metric": 7.057150014588842e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.080160375999185
        },
        {
          "epoch": 867,
          "updates": 13872,
          "samples_seen": 887808,
          "online_loss": 7.799630043336947e-05,
          "train_loss": 7.875139999669045e-05,
          "validation_loss": 7.007279782555997e-05,
          "train_metric": 7.931553227535915e-06,
          "validation_metric": 7.057476068439428e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.090288146999228
        },
        {
          "epoch": 868,
          "updates": 13888,
          "samples_seen": 888832,
          "online_loss": 8.13288916106103e-05,
          "train_loss": 7.766321505187079e-05,
          "validation_loss": 7.139843364711851e-05,
          "train_metric": 7.82195547799347e-06,
          "validation_metric": 7.190988526417641e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.100530671999877
        },
        {
          "epoch": 869,
          "updates": 13904,
          "samples_seen": 889856,
          "online_loss": 8.20966047285765e-05,
          "train_loss": 7.678941619815305e-05,
          "validation_loss": 6.924360059201717e-05,
          "train_metric": 7.733948223176412e-06,
          "validation_metric": 6.973961717449129e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.110745762000079
        },
        {
          "epoch": 870,
          "updates": 13920,
          "samples_seen": 890880,
          "online_loss": 7.834456323507766e-05,
          "train_loss": 7.78718531364575e-05,
          "validation_loss": 6.895918340887874e-05,
          "train_metric": 7.84296753408853e-06,
          "validation_metric": 6.945315817574738e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.120997301999523
        },
        {
          "epoch": 871,
          "updates": 13936,
          "samples_seen": 891904,
          "online_loss": 7.712835395068396e-05,
          "train_loss": 7.809427188476548e-05,
          "validation_loss": 7.200706750154495e-05,
          "train_metric": 7.865368388593197e-06,
          "validation_metric": 7.252288469317136e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.18546726400018
        },
        {
          "epoch": 872,
          "updates": 13952,
          "samples_seen": 892928,
          "online_loss": 7.903348159743473e-05,
          "train_loss": 7.602118421345949e-05,
          "validation_loss": 6.804027361795306e-05,
          "train_metric": 7.656575689907186e-06,
          "validation_metric": 6.852767910459079e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.196688530999381
        },
        {
          "epoch": 873,
          "updates": 13968,
          "samples_seen": 893952,
          "online_loss": 7.890024039625132e-05,
          "train_loss": 9.37407894525677e-05,
          "validation_loss": 7.672476203879341e-05,
          "train_metric": 9.441228939977009e-06,
          "validation_metric": 7.727436241111718e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.207134885000414
        },
        {
          "epoch": 874,
          "updates": 13984,
          "samples_seen": 894976,
          "online_loss": 8.893205767890322e-05,
          "train_loss": 8.094773511402309e-05,
          "validation_loss": 7.419777102768421e-05,
          "train_metric": 8.152760528901126e-06,
          "validation_metric": 7.47292733649374e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.217473278000398
        },
        {
          "epoch": 875,
          "updates": 14000,
          "samples_seen": 896000,
          "online_loss": 7.785655748193676e-05,
          "train_loss": 7.714729144936427e-05,
          "validation_loss": 6.927052891114727e-05,
          "train_metric": 7.769992407702375e-06,
          "validation_metric": 6.976674285397166e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.227727628000139
        },
        {
          "epoch": 876,
          "updates": 14016,
          "samples_seen": 897024,
          "online_loss": 7.811404748281348e-05,
          "train_loss": 7.556916534667835e-05,
          "validation_loss": 6.935639976290986e-05,
          "train_metric": 7.611049113620538e-06,
          "validation_metric": 6.985322215768974e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.238057569000375
        },
        {
          "epoch": 877,
          "updates": 14032,
          "samples_seen": 898048,
          "online_loss": 7.808059285707714e-05,
          "train_loss": 7.682987779844552e-05,
          "validation_loss": 6.942213076399639e-05,
          "train_metric": 7.738023668935057e-06,
          "validation_metric": 6.99194288245053e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.248353375000079
        },
        {
          "epoch": 878,
          "updates": 14048,
          "samples_seen": 899072,
          "online_loss": 7.760968424008752e-05,
          "train_loss": 7.572174945380539e-05,
          "validation_loss": 6.776469672331586e-05,
          "train_metric": 7.626417755091097e-06,
          "validation_metric": 6.8250124058977235e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.259260300999813
        },
        {
          "epoch": 879,
          "updates": 14064,
          "samples_seen": 900096,
          "online_loss": 7.919844279058452e-05,
          "train_loss": 8.119340054690838e-05,
          "validation_loss": 7.304294558707625e-05,
          "train_metric": 8.177503332262859e-06,
          "validation_metric": 7.356617970799562e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.26979974599999
        },
        {
          "epoch": 880,
          "updates": 14080,
          "samples_seen": 901120,
          "online_loss": 7.764203496662958e-05,
          "train_loss": 7.593875488964841e-05,
          "validation_loss": 6.862333975732327e-05,
          "train_metric": 7.6482729127747e-06,
          "validation_metric": 6.911492164363153e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.280179556999428
        },
        {
          "epoch": 881,
          "updates": 14096,
          "samples_seen": 902144,
          "online_loss": 7.631636322003033e-05,
          "train_loss": 7.956607441883534e-05,
          "validation_loss": 7.024320075288415e-05,
          "train_metric": 8.013605111045763e-06,
          "validation_metric": 7.074637778714532e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.29041432700069
        },
        {
          "epoch": 882,
          "updates": 14112,
          "samples_seen": 903168,
          "online_loss": 8.016390825105191e-05,
          "train_loss": 7.562251266790554e-05,
          "validation_loss": 6.967528315726668e-05,
          "train_metric": 7.616422408318613e-06,
          "validation_metric": 7.017439202172682e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.300691193999228
        },
        {
          "epoch": 883,
          "updates": 14128,
          "samples_seen": 904192,
          "online_loss": 7.799489912940771e-05,
          "train_loss": 8.331783465109766e-05,
          "validation_loss": 7.020108751021326e-05,
          "train_metric": 8.39146741782315e-06,
          "validation_metric": 7.070396350172814e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.311344949999693
        },
        {
          "epoch": 884,
          "updates": 14144,
          "samples_seen": 905216,
          "online_loss": 8.158860282492242e-05,
          "train_loss": 7.59329559514299e-05,
          "validation_loss": 6.855362153146416e-05,
          "train_metric": 7.647689017176162e-06,
          "validation_metric": 6.9044699557707645e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.321742520000043
        },
        {
          "epoch": 885,
          "updates": 14160,
          "samples_seen": 906240,
          "online_loss": 8.05564156962646e-05,
          "train_loss": 7.484614616259933e-05,
          "validation_loss": 6.804148142691702e-05,
          "train_metric": 7.538229510828387e-06,
          "validation_metric": 6.8528893280017655e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.33196011399923
        },
        {
          "epoch": 886,
          "updates": 14176,
          "samples_seen": 907264,
          "online_loss": 7.735804456387996e-05,
          "train_loss": 7.486507820431143e-05,
          "validation_loss": 6.766629667254165e-05,
          "train_metric": 7.540137175965356e-06,
          "validation_metric": 6.815101187385153e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.34297202199923
        },
        {
          "epoch": 887,
          "updates": 14192,
          "samples_seen": 908288,
          "online_loss": 7.703677988502022e-05,
          "train_loss": 7.48992242733948e-05,
          "validation_loss": 6.829742778791115e-05,
          "train_metric": 7.543575520685408e-06,
          "validation_metric": 6.878667136334116e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.353330347999872
        },
        {
          "epoch": 888,
          "updates": 14208,
          "samples_seen": 909312,
          "online_loss": 7.682144632781274e-05,
          "train_loss": 7.602204277645797e-05,
          "validation_loss": 6.807766476413235e-05,
          "train_metric": 7.656662091903854e-06,
          "validation_metric": 6.856532763777068e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.363651162000679
        },
        {
          "epoch": 889,
          "updates": 14224,
          "samples_seen": 910336,
          "online_loss": 7.958682135722484e-05,
          "train_loss": 7.548795838374645e-05,
          "validation_loss": 6.828780169598758e-05,
          "train_metric": 7.602870937262196e-06,
          "validation_metric": 6.877697160234675e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.374070403999212
        },
        {
          "epoch": 890,
          "updates": 14240,
          "samples_seen": 911360,
          "online_loss": 7.978766484484368e-05,
          "train_loss": 8.721580525161698e-05,
          "validation_loss": 7.307942723855376e-05,
          "train_metric": 8.784056262811646e-06,
          "validation_metric": 7.360291874647373e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.384362360000523
        },
        {
          "epoch": 891,
          "updates": 14256,
          "samples_seen": 912384,
          "online_loss": 8.638450890430249e-05,
          "train_loss": 7.557681237813085e-05,
          "validation_loss": 7.010902481852099e-05,
          "train_metric": 7.611819910380291e-06,
          "validation_metric": 7.0611258706776425e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.3945177530004
        },
        {
          "epoch": 892,
          "updates": 14272,
          "samples_seen": 913408,
          "online_loss": 8.660712410346605e-05,
          "train_loss": 7.458144682459533e-05,
          "validation_loss": 6.756059156032279e-05,
          "train_metric": 7.511569947382668e-06,
          "validation_metric": 6.804456006648252e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.40532174299915
        },
        {
          "epoch": 893,
          "updates": 14288,
          "samples_seen": 914432,
          "online_loss": 7.946599180286285e-05,
          "train_loss": 8.027313742786646e-05,
          "validation_loss": 7.097094203345478e-05,
          "train_metric": 8.084817636699881e-06,
          "validation_metric": 7.1479335019830614e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.415644298000188
        },
        {
          "epoch": 894,
          "updates": 14304,
          "samples_seen": 915456,
          "online_loss": 7.833591416783747e-05,
          "train_loss": 7.50013132346794e-05,
          "validation_loss": 6.807548925280571e-05,
          "train_metric": 7.553857358288951e-06,
          "validation_metric": 6.856313575553941e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.425914137998916
        },
        {
          "epoch": 895,
          "updates": 14320,
          "samples_seen": 916480,
          "online_loss": 7.710656609560829e-05,
          "train_loss": 7.485641981475055e-05,
          "validation_loss": 6.687876157229766e-05,
          "train_metric": 7.539263606304303e-06,
          "validation_metric": 6.735783699696185e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.436682932999247
        },
        {
          "epoch": 896,
          "updates": 14336,
          "samples_seen": 917504,
          "online_loss": 7.905723532530828e-05,
          "train_loss": 8.44476671773009e-05,
          "validation_loss": 7.189887401182204e-05,
          "train_metric": 8.50525884743547e-06,
          "validation_metric": 7.241391358547844e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.447198599998956
        },
        {
          "epoch": 897,
          "updates": 14352,
          "samples_seen": 918528,
          "online_loss": 8.314989190694178e-05,
          "train_loss": 7.545914559159428e-05,
          "validation_loss": 6.83791731717065e-05,
          "train_metric": 7.599968739668839e-06,
          "validation_metric": 6.886899882374564e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.457411383000363
        },
        {
          "epoch": 898,
          "updates": 14368,
          "samples_seen": 919552,
          "online_loss": 7.988274410308804e-05,
          "train_loss": 7.903661753516644e-05,
          "validation_loss": 6.866523472126573e-05,
          "train_metric": 7.960277798702009e-06,
          "validation_metric": 6.915710855537327e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.468042266998964
        },
        {
          "epoch": 899,
          "updates": 14384,
          "samples_seen": 920576,
          "online_loss": 8.38766461583873e-05,
          "train_loss": 7.483419176423922e-05,
          "validation_loss": 6.906698399689049e-05,
          "train_metric": 7.5370257945905905e-06,
          "validation_metric": 6.956173820071854e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.47836624899901
        },
        {
          "epoch": 900,
          "updates": 14400,
          "samples_seen": 921600,
          "online_loss": 7.6316872764437e-05,
          "train_loss": 7.578881923109293e-05,
          "validation_loss": 6.71141897328198e-05,
          "train_metric": 7.633171662746463e-06,
          "validation_metric": 6.759495590813458e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.488754800999232
        },
        {
          "epoch": 901,
          "updates": 14416,
          "samples_seen": 922624,
          "online_loss": 7.868247894293745e-05,
          "train_loss": 7.608329178765416e-05,
          "validation_loss": 6.946952635189518e-05,
          "train_metric": 7.66283119446598e-06,
          "validation_metric": 6.9967163653927855e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.49895124700015
        },
        {
          "epoch": 902,
          "updates": 14432,
          "samples_seen": 923648,
          "online_loss": 7.972112553034094e-05,
          "train_loss": 7.569777517346665e-05,
          "validation_loss": 6.770183972548693e-05,
          "train_metric": 7.624003501405241e-06,
          "validation_metric": 6.818681868026033e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.509275819000322
        },
        {
          "epoch": 903,
          "updates": 14448,
          "samples_seen": 924672,
          "online_loss": 7.577008045700495e-05,
          "train_loss": 7.550806913059205e-05,
          "validation_loss": 6.783744902350008e-05,
          "train_metric": 7.604896381963044e-06,
          "validation_metric": 6.832338840467855e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.519568095000068
        },
        {
          "epoch": 904,
          "updates": 14464,
          "samples_seen": 925696,
          "online_loss": 8.365972871615668e-05,
          "train_loss": 7.498334161937237e-05,
          "validation_loss": 6.741298420820385e-05,
          "train_metric": 7.552046554337721e-06,
          "validation_metric": 6.78958849675837e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.529853553000066
        },
        {
          "epoch": 905,
          "updates": 14480,
          "samples_seen": 926720,
          "online_loss": 8.298644979731762e-05,
          "train_loss": 7.883589569246396e-05,
          "validation_loss": 6.849588680779561e-05,
          "train_metric": 7.940063369460404e-06,
          "validation_metric": 6.8986541919002775e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.540181488999224
        },
        {
          "epoch": 906,
          "updates": 14496,
          "samples_seen": 927744,
          "online_loss": 7.887912397563923e-05,
          "train_loss": 7.499031198676676e-05,
          "validation_loss": 6.87405263306573e-05,
          "train_metric": 7.552749593742192e-06,
          "validation_metric": 6.9232946771080606e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.550496690999353
        },
        {
          "epoch": 907,
          "updates": 14512,
          "samples_seen": 928768,
          "online_loss": 7.81033243129059e-05,
          "train_loss": 7.655797526240349e-05,
          "validation_loss": 6.897559796925634e-05,
          "train_metric": 7.710638783464674e-06,
          "validation_metric": 6.946969278942561e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.617672316000608
        },
        {
          "epoch": 908,
          "updates": 14528,
          "samples_seen": 929792,
          "online_loss": 8.391470555579872e-05,
          "train_loss": 8.157337288139388e-05,
          "validation_loss": 7.141159585444257e-05,
          "train_metric": 8.215771231334656e-06,
          "validation_metric": 7.192315024440177e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.62840600600066
        },
        {
          "epoch": 909,
          "updates": 14544,
          "samples_seen": 930816,
          "online_loss": 7.855718013161095e-05,
          "train_loss": 7.44152202969417e-05,
          "validation_loss": 6.663252133876085e-05,
          "train_metric": 7.494827968912432e-06,
          "validation_metric": 6.710984507662943e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.639353706999827
        },
        {
          "epoch": 910,
          "updates": 14560,
          "samples_seen": 931840,
          "online_loss": 7.491553333238699e-05,
          "train_loss": 7.404677307931706e-05,
          "validation_loss": 6.853811646578833e-05,
          "train_metric": 7.457718766090693e-06,
          "validation_metric": 6.902907443873119e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.649897949999286
        },
        {
          "epoch": 911,
          "updates": 14576,
          "samples_seen": 932864,
          "online_loss": 7.807135716575431e-05,
          "train_loss": 7.801606261637062e-05,
          "validation_loss": 6.838418630650267e-05,
          "train_metric": 7.857491254981142e-06,
          "validation_metric": 6.88740556142875e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.660274969999591
        },
        {
          "epoch": 912,
          "updates": 14592,
          "samples_seen": 933888,
          "online_loss": 7.91486734215141e-05,
          "train_loss": 7.665793964406475e-05,
          "validation_loss": 7.002209167694673e-05,
          "train_metric": 7.720705980318598e-06,
          "validation_metric": 7.052368346194271e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.670616335999512
        },
        {
          "epoch": 913,
          "updates": 14608,
          "samples_seen": 934912,
          "online_loss": 7.584680997752002e-05,
          "train_loss": 7.394366548396647e-05,
          "validation_loss": 6.738794036209583e-05,
          "train_metric": 7.4473346103332005e-06,
          "validation_metric": 6.787067377445055e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.680875689999084
        },
        {
          "epoch": 914,
          "updates": 14624,
          "samples_seen": 935936,
          "online_loss": 7.630797040292236e-05,
          "train_loss": 7.630141044501215e-05,
          "validation_loss": 6.825086165918037e-05,
          "train_metric": 7.684799129492603e-06,
          "validation_metric": 6.873976872157073e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.691174653999042
        },
        {
          "epoch": 915,
          "updates": 14640,
          "samples_seen": 936960,
          "online_loss": 7.709838791924994e-05,
          "train_loss": 7.368143269559368e-05,
          "validation_loss": 6.781260162824765e-05,
          "train_metric": 7.420923338941066e-06,
          "validation_metric": 6.829836820543278e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.701369414999135
        },
        {
          "epoch": 916,
          "updates": 14656,
          "samples_seen": 937984,
          "online_loss": 7.603214862683672e-05,
          "train_loss": 7.583193655591458e-05,
          "validation_loss": 6.804633449064568e-05,
          "train_metric": 7.63751540944213e-06,
          "validation_metric": 6.853377271909267e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.71168914000009
        },
        {
          "epoch": 917,
          "updates": 14672,
          "samples_seen": 939008,
          "online_loss": 7.557997287221951e-05,
          "train_loss": 7.427684613503516e-05,
          "validation_loss": 6.711936293868348e-05,
          "train_metric": 7.4808913268498145e-06,
          "validation_metric": 6.760016731277574e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.721922949000145
        },
        {
          "epoch": 918,
          "updates": 14688,
          "samples_seen": 940032,
          "online_loss": 7.590975587845605e-05,
          "train_loss": 7.357255526585504e-05,
          "validation_loss": 6.744381971657276e-05,
          "train_metric": 7.409958925563842e-06,
          "validation_metric": 6.792694875912275e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.732278483999835
        },
        {
          "epoch": 919,
          "updates": 14704,
          "samples_seen": 941056,
          "online_loss": 7.886761818554078e-05,
          "train_loss": 8.031206380110234e-05,
          "validation_loss": 6.925357592990622e-05,
          "train_metric": 8.088737558864523e-06,
          "validation_metric": 6.974967163841939e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.742566085998988
        },
        {
          "epoch": 920,
          "updates": 14720,
          "samples_seen": 942080,
          "online_loss": 7.413187631755136e-05,
          "train_loss": 7.52358027966693e-05,
          "validation_loss": 6.674430915154517e-05,
          "train_metric": 7.577474207209889e-06,
          "validation_metric": 6.722242233081488e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.752856683000573
        },
        {
          "epoch": 921,
          "updates": 14736,
          "samples_seen": 943104,
          "online_loss": 7.643493654541089e-05,
          "train_loss": 7.348106009885669e-05,
          "validation_loss": 6.704655243083835e-05,
          "train_metric": 7.400742106256075e-06,
          "validation_metric": 6.752683020749828e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.76307059499959
        },
        {
          "epoch": 922,
          "updates": 14752,
          "samples_seen": 944128,
          "online_loss": 7.507019904551271e-05,
          "train_loss": 7.462671055691317e-05,
          "validation_loss": 6.733382178936154e-05,
          "train_metric": 7.516128334827954e-06,
          "validation_metric": 6.781615866202628e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.773376545999781
        },
        {
          "epoch": 923,
          "updates": 14768,
          "samples_seen": 945152,
          "online_loss": 7.632687334080401e-05,
          "train_loss": 7.373264088528231e-05,
          "validation_loss": 6.673977622995153e-05,
          "train_metric": 7.426081538142171e-06,
          "validation_metric": 6.721786121488549e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.783629671999734
        },
        {
          "epoch": 924,
          "updates": 14784,
          "samples_seen": 946176,
          "online_loss": 7.502436528739054e-05,
          "train_loss": 7.459339394699782e-05,
          "validation_loss": 6.625893001910299e-05,
          "train_metric": 7.512772754125763e-06,
          "validation_metric": 6.673357347608544e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.794485205000456
        },
        {
          "epoch": 925,
          "updates": 14800,
          "samples_seen": 947200,
          "online_loss": 7.47644107832457e-05,
          "train_loss": 8.371468720724806e-05,
          "validation_loss": 7.549013389507309e-05,
          "train_metric": 8.431436071987264e-06,
          "validation_metric": 7.603090125485323e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.805034749999322
        },
        {
          "epoch": 926,
          "updates": 14816,
          "samples_seen": 948224,
          "online_loss": 8.032846790229087e-05,
          "train_loss": 7.403580821119249e-05,
          "validation_loss": 6.694145849905908e-05,
          "train_metric": 7.4566150942700915e-06,
          "validation_metric": 6.742098776157945e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.81536375199903
        },
        {
          "epoch": 927,
          "updates": 14832,
          "samples_seen": 949248,
          "online_loss": 7.768175328237703e-05,
          "train_loss": 7.482319051632658e-05,
          "validation_loss": 6.67552012600936e-05,
          "train_metric": 7.53591757529648e-06,
          "validation_metric": 6.723339083691826e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.825518280000324
        },
        {
          "epoch": 928,
          "updates": 14848,
          "samples_seen": 950272,
          "online_loss": 8.06948744411784e-05,
          "train_loss": 7.348313374677673e-05,
          "validation_loss": 6.682109233224764e-05,
          "train_metric": 7.400952199532185e-06,
          "validation_metric": 6.729976121278014e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.835793448999539
        },
        {
          "epoch": 929,
          "updates": 14864,
          "samples_seen": 951296,
          "online_loss": 9.549626702209935e-05,
          "train_loss": 8.773765875957906e-05,
          "validation_loss": 7.885511877248064e-05,
          "train_metric": 8.836615961627103e-06,
          "validation_metric": 7.941998774185777e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.845932496000387
        },
        {
          "epoch": 930,
          "updates": 14880,
          "samples_seen": 952320,
          "online_loss": 8.562208927287429e-05,
          "train_loss": 9.522712207399309e-05,
          "validation_loss": 8.16634637885727e-05,
          "train_metric": 9.590926310920622e-06,
          "validation_metric": 8.224844350479543e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.856140847999995
        },
        {
          "epoch": 931,
          "updates": 14896,
          "samples_seen": 953344,
          "online_loss": 8.801331819086045e-05,
          "train_loss": 7.853929128032178e-05,
          "validation_loss": 6.717569340253249e-05,
          "train_metric": 7.910191015980672e-06,
          "validation_metric": 6.7656897044798825e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.86632105299941
        },
        {
          "epoch": 932,
          "updates": 14912,
          "samples_seen": 954368,
          "online_loss": 8.252149791587726e-05,
          "train_loss": 7.810744864400476e-05,
          "validation_loss": 6.849365308880806e-05,
          "train_metric": 7.866696250857785e-06,
          "validation_metric": 6.8984300014562905e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.876676382000369
        },
        {
          "epoch": 933,
          "updates": 14928,
          "samples_seen": 955392,
          "online_loss": 7.585563071188517e-05,
          "train_loss": 7.370347157120705e-05,
          "validation_loss": 6.655860488535836e-05,
          "train_metric": 7.423144325002795e-06,
          "validation_metric": 6.7035393840342294e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.886849770000481
        },
        {
          "epoch": 934,
          "updates": 14944,
          "samples_seen": 956416,
          "online_loss": 7.519880296058545e-05,
          "train_loss": 7.339696458075196e-05,
          "validation_loss": 6.630998541368172e-05,
          "train_metric": 7.392272436845815e-06,
          "validation_metric": 6.678499175905017e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.897050645999116
        },
        {
          "epoch": 935,
          "updates": 14960,
          "samples_seen": 957440,
          "online_loss": 7.644910238013836e-05,
          "train_loss": 7.950379222165793e-05,
          "validation_loss": 7.050979911582544e-05,
          "train_metric": 8.007330507098231e-06,
          "validation_metric": 7.101487881300272e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.907356869000068
        },
        {
          "epoch": 936,
          "updates": 14976,
          "samples_seen": 958464,
          "online_loss": 8.021515441214433e-05,
          "train_loss": 8.068903116509318e-05,
          "validation_loss": 7.202917913673446e-05,
          "train_metric": 8.126703505695332e-06,
          "validation_metric": 7.254515367094427e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.917593152000336
        },
        {
          "epoch": 937,
          "updates": 14992,
          "samples_seen": 959488,
          "online_loss": 8.211207614294835e-05,
          "train_loss": 7.383637421298772e-05,
          "validation_loss": 6.600109918508679e-05,
          "train_metric": 7.436529358528787e-06,
          "validation_metric": 6.647389000136172e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.928377932999865
        },
        {
          "epoch": 938,
          "updates": 15008,
          "samples_seen": 960512,
          "online_loss": 7.621697272952588e-05,
          "train_loss": 7.257124525494874e-05,
          "validation_loss": 6.647714326390997e-05,
          "train_metric": 7.309110515052453e-06,
          "validation_metric": 6.695334832329536e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.938865862999592
        },
        {
          "epoch": 939,
          "updates": 15024,
          "samples_seen": 961536,
          "online_loss": 7.49739979255537e-05,
          "train_loss": 7.55083019612357e-05,
          "validation_loss": 6.788170139770955e-05,
          "train_metric": 7.6049200288252905e-06,
          "validation_metric": 6.8367962740012445e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.949172008999085
        },
        {
          "epoch": 940,
          "updates": 15040,
          "samples_seen": 962560,
          "online_loss": 7.880946213845164e-05,
          "train_loss": 7.691467908443883e-05,
          "validation_loss": 6.820485577918589e-05,
          "train_metric": 7.746563824184705e-06,
          "validation_metric": 6.869342541904189e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.959445137999865
        },
        {
          "epoch": 941,
          "updates": 15056,
          "samples_seen": 963584,
          "online_loss": 7.795090391482518e-05,
          "train_loss": 7.807528891135007e-05,
          "validation_loss": 6.826756725786254e-05,
          "train_metric": 7.863455721235368e-06,
          "validation_metric": 6.875658982608002e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.969675235999603
        },
        {
          "epoch": 942,
          "updates": 15072,
          "samples_seen": 964608,
          "online_loss": 7.815009826117603e-05,
          "train_loss": 7.314634422073141e-05,
          "validation_loss": 6.590581324417144e-05,
          "train_metric": 7.367031685134862e-06,
          "validation_metric": 6.637792466790415e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 10.980577841999548
        },
        {
          "epoch": 943,
          "updates": 15088,
          "samples_seen": 965632,
          "online_loss": 7.505272128582874e-05,
          "train_loss": 7.969651778694242e-05,
          "validation_loss": 6.852555088698864e-05,
          "train_metric": 8.02674094302347e-06,
          "validation_metric": 6.901642791490303e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.045077194999976
        },
        {
          "epoch": 944,
          "updates": 15104,
          "samples_seen": 966656,
          "online_loss": 8.057838977038045e-05,
          "train_loss": 7.248108886415139e-05,
          "validation_loss": 6.566809315700084e-05,
          "train_metric": 7.30002921045525e-06,
          "validation_metric": 6.613849109271541e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.05637612999999
        },
        {
          "epoch": 945,
          "updates": 15120,
          "samples_seen": 967680,
          "online_loss": 7.479003352273139e-05,
          "train_loss": 7.261030259542167e-05,
          "validation_loss": 6.464557372964919e-05,
          "train_metric": 7.313043170142919e-06,
          "validation_metric": 6.51086565994774e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.067264160999912
        },
        {
          "epoch": 946,
          "updates": 15136,
          "samples_seen": 968704,
          "online_loss": 7.429560923810641e-05,
          "train_loss": 7.213807111838832e-05,
          "validation_loss": 6.505501369247213e-05,
          "train_metric": 7.265482963703107e-06,
          "validation_metric": 6.552102604473475e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.078172199000619
        },
        {
          "epoch": 947,
          "updates": 15152,
          "samples_seen": 969728,
          "online_loss": 7.807336032783496e-05,
          "train_loss": 7.735074905212969e-05,
          "validation_loss": 6.90148226567544e-05,
          "train_metric": 7.790484232828021e-06,
          "validation_metric": 6.9509196691797115e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.088784264000424
        },
        {
          "epoch": 948,
          "updates": 15168,
          "samples_seen": 970752,
          "online_loss": 7.406426789202669e-05,
          "train_loss": 7.365013880189508e-05,
          "validation_loss": 6.663265230599791e-05,
          "train_metric": 7.417771939799422e-06,
          "validation_metric": 6.7109967858414166e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.099071306000042
        },
        {
          "epoch": 949,
          "updates": 15184,
          "samples_seen": 971776,
          "online_loss": 7.622292696396471e-05,
          "train_loss": 7.256893150042742e-05,
          "validation_loss": 6.456347182393074e-05,
          "train_metric": 7.308876774914097e-06,
          "validation_metric": 6.5025960793718696e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.110331476000283
        },
        {
          "epoch": 950,
          "updates": 15200,
          "samples_seen": 972800,
          "online_loss": 7.561646179965464e-05,
          "train_loss": 7.352106331381947e-05,
          "validation_loss": 6.474009569501504e-05,
          "train_metric": 7.404772532026982e-06,
          "validation_metric": 6.520385340991197e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.121055944000545
        },
        {
          "epoch": 951,
          "updates": 15216,
          "samples_seen": 973824,
          "online_loss": 7.411763385789527e-05,
          "train_loss": 7.201034168247133e-05,
          "validation_loss": 6.504344491986558e-05,
          "train_metric": 7.252617251651827e-06,
          "validation_metric": 6.550937541760504e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.131362476000504
        },
        {
          "epoch": 952,
          "updates": 15232,
          "samples_seen": 974848,
          "online_loss": 7.625129751431814e-05,
          "train_loss": 8.524011354893446e-05,
          "validation_loss": 7.554506009910256e-05,
          "train_metric": 8.585071554989554e-06,
          "validation_metric": 7.608621672261506e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.141818223999508
        },
        {
          "epoch": 953,
          "updates": 15248,
          "samples_seen": 975872,
          "online_loss": 8.252579823420092e-05,
          "train_loss": 7.208828901639208e-05,
          "validation_loss": 6.425452011171728e-05,
          "train_metric": 7.260468009917531e-06,
          "validation_metric": 6.471479991887463e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.15264910299993
        },
        {
          "epoch": 954,
          "updates": 15264,
          "samples_seen": 976896,
          "online_loss": 7.39394745323807e-05,
          "train_loss": 7.406533404719085e-05,
          "validation_loss": 6.76322597428225e-05,
          "train_metric": 7.459589141944889e-06,
          "validation_metric": 6.811673301854171e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.163099555998997
        },
        {
          "epoch": 955,
          "updates": 15280,
          "samples_seen": 977920,
          "online_loss": 7.443728850375919e-05,
          "train_loss": 7.271739013958722e-05,
          "validation_loss": 6.420678255381063e-05,
          "train_metric": 7.323829322558595e-06,
          "validation_metric": 6.466671948146541e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.17395429499993
        },
        {
          "epoch": 956,
          "updates": 15296,
          "samples_seen": 978944,
          "online_loss": 7.36725132810534e-05,
          "train_loss": 7.631837070221081e-05,
          "validation_loss": 6.99614392942749e-05,
          "train_metric": 7.686507160542533e-06,
          "validation_metric": 7.046260634524515e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.184734254999057
        },
        {
          "epoch": 957,
          "updates": 15312,
          "samples_seen": 979968,
          "online_loss": 7.825741568012745e-05,
          "train_loss": 7.325585465878248e-05,
          "validation_loss": 6.647332338616252e-05,
          "train_metric": 7.378062491625315e-06,
          "validation_metric": 6.694948751828633e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.195131189999302
        },
        {
          "epoch": 958,
          "updates": 15328,
          "samples_seen": 980992,
          "online_loss": 7.487558491447999e-05,
          "train_loss": 7.476187602151185e-05,
          "validation_loss": 6.569845572812483e-05,
          "train_metric": 7.529743015766144e-06,
          "validation_metric": 6.616907739953604e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.205525199999101
        },
        {
          "epoch": 959,
          "updates": 15344,
          "samples_seen": 982016,
          "online_loss": 7.673702748434152e-05,
          "train_loss": 7.170742173912004e-05,
          "validation_loss": 6.479687726823613e-05,
          "train_metric": 7.222108706628205e-06,
          "validation_metric": 6.526104243675945e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.216000095999334
        },
        {
          "epoch": 960,
          "updates": 15360,
          "samples_seen": 983040,
          "online_loss": 7.289011864486383e-05,
          "train_loss": 7.275516691152006e-05,
          "validation_loss": 6.483597098849714e-05,
          "train_metric": 7.327634193643462e-06,
          "validation_metric": 6.5300409914925694e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.226663155999631
        },
        {
          "epoch": 961,
          "updates": 15376,
          "samples_seen": 984064,
          "online_loss": 7.320623285522743e-05,
          "train_loss": 7.180019747465849e-05,
          "validation_loss": 6.566898082382977e-05,
          "train_metric": 7.2314524004468694e-06,
          "validation_metric": 6.613939149247017e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.237205686999005
        },
        {
          "epoch": 962,
          "updates": 15392,
          "samples_seen": 985088,
          "online_loss": 7.374791789516166e-05,
          "train_loss": 7.504333916585892e-05,
          "validation_loss": 6.499982555396855e-05,
          "train_metric": 7.558091056125704e-06,
          "validation_metric": 6.546544682350941e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.247436343999652
        },
        {
          "epoch": 963,
          "updates": 15408,
          "samples_seen": 986112,
          "online_loss": 7.461622658411216e-05,
          "train_loss": 7.19793897587806e-05,
          "validation_loss": 6.451442459365353e-05,
          "train_metric": 7.249500868056202e-06,
          "validation_metric": 6.49765615889919e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.257703232000495
        },
        {
          "epoch": 964,
          "updates": 15424,
          "samples_seen": 987136,
          "online_loss": 7.299826074813609e-05,
          "train_loss": 7.537884084740654e-05,
          "validation_loss": 6.609073898289353e-05,
          "train_metric": 7.591880603285972e-06,
          "validation_metric": 6.656417554040672e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.268234691999169
        },
        {
          "epoch": 965,
          "updates": 15440,
          "samples_seen": 988160,
          "online_loss": 7.609740919178876e-05,
          "train_loss": 7.153792103054002e-05,
          "validation_loss": 6.369633774738759e-05,
          "train_metric": 7.205037491075927e-06,
          "validation_metric": 6.415261395886773e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.27902005299984
        },
        {
          "epoch": 966,
          "updates": 15456,
          "samples_seen": 989184,
          "online_loss": 7.429165066241694e-05,
          "train_loss": 7.214192737592384e-05,
          "validation_loss": 6.443646998377517e-05,
          "train_metric": 7.265870408446062e-06,
          "validation_metric": 6.489806310128188e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.289479541999754
        },
        {
          "epoch": 967,
          "updates": 15472,
          "samples_seen": 990208,
          "online_loss": 7.635766269231681e-05,
          "train_loss": 7.178800296969712e-05,
          "validation_loss": 6.553630373673514e-05,
          "train_metric": 7.230225492094178e-06,
          "validation_metric": 6.60057730783592e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.299735688999135
        },
        {
          "epoch": 968,
          "updates": 15488,
          "samples_seen": 991232,
          "online_loss": 7.367639250333013e-05,
          "train_loss": 7.252051000250503e-05,
          "validation_loss": 6.432524241972715e-05,
          "train_metric": 7.3040005190705415e-06,
          "validation_metric": 6.478602699644398e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.310232132000237
        },
        {
          "epoch": 969,
          "updates": 15504,
          "samples_seen": 992256,
          "online_loss": 7.463084284609067e-05,
          "train_loss": 7.225361332530156e-05,
          "validation_loss": 6.489238876383752e-05,
          "train_metric": 7.27711903891759e-06,
          "validation_metric": 6.535723514389247e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.320451855999636
        },
        {
          "epoch": 970,
          "updates": 15520,
          "samples_seen": 993280,
          "online_loss": 8.131516051435028e-05,
          "train_loss": 7.168070442276075e-05,
          "validation_loss": 6.458804273279384e-05,
          "train_metric": 7.219418421300361e-06,
          "validation_metric": 6.505070814455394e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.330956350000633
        },
        {
          "epoch": 971,
          "updates": 15536,
          "samples_seen": 994304,
          "online_loss": 7.88363915944501e-05,
          "train_loss": 7.451217243215069e-05,
          "validation_loss": 6.510784442070872e-05,
          "train_metric": 7.504592758778017e-06,
          "validation_metric": 6.557423603226198e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.34122824899896
        },
        {
          "epoch": 972,
          "updates": 15552,
          "samples_seen": 995328,
          "online_loss": 7.529287745455804e-05,
          "train_loss": 7.390579412458465e-05,
          "validation_loss": 6.67002605041489e-05,
          "train_metric": 7.443521099048667e-06,
          "validation_metric": 6.71780571792624e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.35149939800067
        },
        {
          "epoch": 973,
          "updates": 15568,
          "samples_seen": 996352,
          "online_loss": 7.594543058075942e-05,
          "train_loss": 7.193987403297797e-05,
          "validation_loss": 6.446263432735577e-05,
          "train_metric": 7.245520009746542e-06,
          "validation_metric": 6.492441571026575e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.361672987999555
        },
        {
          "epoch": 974,
          "updates": 15584,
          "samples_seen": 997376,
          "online_loss": 7.96661111053254e-05,
          "train_loss": 8.153545059030876e-05,
          "validation_loss": 6.929683149792254e-05,
          "train_metric": 8.211952263081912e-06,
          "validation_metric": 6.97932318871608e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.3720526470006
        },
        {
          "epoch": 975,
          "updates": 15600,
          "samples_seen": 998400,
          "online_loss": 7.41900296361564e-05,
          "train_loss": 7.181871478678659e-05,
          "validation_loss": 6.357592064887285e-05,
          "train_metric": 7.233317774080206e-06,
          "validation_metric": 6.403133738785982e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.382803393998984
        },
        {
          "epoch": 976,
          "updates": 15616,
          "samples_seen": 999424,
          "online_loss": 7.352719239861472e-05,
          "train_loss": 7.214242941699922e-05,
          "validation_loss": 6.601717177545652e-05,
          "train_metric": 7.265921340149362e-06,
          "validation_metric": 6.649007445957977e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.39324103800027
        },
        {
          "epoch": 977,
          "updates": 15632,
          "samples_seen": 1000448,
          "online_loss": 7.347539758484345e-05,
          "train_loss": 7.161616667872295e-05,
          "validation_loss": 6.440120341721922e-05,
          "train_metric": 7.212918717414141e-06,
          "validation_metric": 6.486253369075712e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.403529286999401
        },
        {
          "epoch": 978,
          "updates": 15648,
          "samples_seen": 1001472,
          "online_loss": 7.391380563603889e-05,
          "train_loss": 7.095399632817134e-05,
          "validation_loss": 6.39697682345286e-05,
          "train_metric": 7.146226380427834e-06,
          "validation_metric": 6.442801804951159e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.413873997000337
        },
        {
          "epoch": 979,
          "updates": 15664,
          "samples_seen": 1002496,
          "online_loss": 7.39123213406856e-05,
          "train_loss": 7.128973084036261e-05,
          "validation_loss": 6.341303378576413e-05,
          "train_metric": 7.1800400291976985e-06,
          "validation_metric": 6.386728273355402e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.479722590000165
        },
        {
          "epoch": 980,
          "updates": 15680,
          "samples_seen": 1003520,
          "online_loss": 7.335663667618064e-05,
          "train_loss": 7.087783887982368e-05,
          "validation_loss": 6.360342376865447e-05,
          "train_metric": 7.138555702113081e-06,
          "validation_metric": 6.405904514394933e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.490446148000046
        },
        {
          "epoch": 981,
          "updates": 15696,
          "samples_seen": 1004544,
          "online_loss": 7.221044188554515e-05,
          "train_loss": 7.100559741957113e-05,
          "validation_loss": 6.382929132087156e-05,
          "train_metric": 7.151423687901115e-06,
          "validation_metric": 6.428652795875678e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.500737974000003
        },
        {
          "epoch": 982,
          "updates": 15712,
          "samples_seen": 1005568,
          "online_loss": 7.321527209569467e-05,
          "train_loss": 7.150106830522418e-05,
          "validation_loss": 6.328281597234309e-05,
          "train_metric": 7.201327207440045e-06,
          "validation_metric": 6.3736133597558364e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.511648960000457
        },
        {
          "epoch": 983,
          "updates": 15728,
          "samples_seen": 1006592,
          "online_loss": 7.337860051848111e-05,
          "train_loss": 7.151893805712461e-05,
          "validation_loss": 6.396735261660069e-05,
          "train_metric": 7.20312527846545e-06,
          "validation_metric": 6.442558060371084e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.522053204998883
        },
        {
          "epoch": 984,
          "updates": 15744,
          "samples_seen": 1007616,
          "online_loss": 7.700974833824148e-05,
          "train_loss": 7.727096090093255e-05,
          "validation_loss": 6.552525155711919e-05,
          "train_metric": 7.782447937643155e-06,
          "validation_metric": 6.599462722078897e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.532356571000491
        },
        {
          "epoch": 985,
          "updates": 15760,
          "samples_seen": 1008640,
          "online_loss": 7.969668968144106e-05,
          "train_loss": 8.325047383550555e-05,
          "validation_loss": 6.84065162204206e-05,
          "train_metric": 8.384683496842626e-06,
          "validation_metric": 6.889653832331533e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.542649070999687
        },
        {
          "epoch": 986,
          "updates": 15776,
          "samples_seen": 1009664,
          "online_loss": 8.092490998024005e-05,
          "train_loss": 7.146721327444538e-05,
          "validation_loss": 6.489137740572914e-05,
          "train_metric": 7.1979156928136945e-06,
          "validation_metric": 6.53562256047735e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.552910469999915
        },
        {
          "epoch": 987,
          "updates": 15792,
          "samples_seen": 1010688,
          "online_loss": 7.758831293358526e-05,
          "train_loss": 7.090430881362408e-05,
          "validation_loss": 6.358732207445428e-05,
          "train_metric": 7.141222340578679e-06,
          "validation_metric": 6.4042819758469705e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.563224383999113
        },
        {
          "epoch": 988,
          "updates": 15808,
          "samples_seen": 1011712,
          "online_loss": 7.49366661239037e-05,
          "train_loss": 7.247780013130978e-05,
          "validation_loss": 6.38486526440829e-05,
          "train_metric": 7.2996986091311555e-06,
          "validation_metric": 6.430602297768928e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.573787183000604
        },
        {
          "epoch": 989,
          "updates": 15824,
          "samples_seen": 1012736,
          "online_loss": 7.240625245685806e-05,
          "train_loss": 7.060223288135603e-05,
          "validation_loss": 6.297021172940731e-05,
          "train_metric": 7.110799288057024e-06,
          "validation_metric": 6.34212801742251e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.58459475699965
        },
        {
          "epoch": 990,
          "updates": 15840,
          "samples_seen": 1013760,
          "online_loss": 7.44119231512741e-05,
          "train_loss": 7.084776734700426e-05,
          "validation_loss": 6.298161315498874e-05,
          "train_metric": 7.135527539503528e-06,
          "validation_metric": 6.3432771639782e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.595108326999252
        },
        {
          "epoch": 991,
          "updates": 15856,
          "samples_seen": 1014784,
          "online_loss": 7.242325978040753e-05,
          "train_loss": 7.622358680237085e-05,
          "validation_loss": 6.709180161124095e-05,
          "train_metric": 7.676961104152724e-06,
          "validation_metric": 6.757240043953061e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.605447625999659
        },
        {
          "epoch": 992,
          "updates": 15872,
          "samples_seen": 1015808,
          "online_loss": 7.179805515988846e-05,
          "train_loss": 8.600780711276457e-05,
          "validation_loss": 7.101234223227948e-05,
          "train_metric": 8.662391337566078e-06,
          "validation_metric": 7.1521026256959885e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.615736108999045
        },
        {
          "epoch": 993,
          "updates": 15888,
          "samples_seen": 1016832,
          "online_loss": 7.707497798037366e-05,
          "train_loss": 7.05102866049856e-05,
          "validation_loss": 6.185837992234156e-05,
          "train_metric": 7.101537448761519e-06,
          "validation_metric": 6.230149665498175e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.627231724000012
        },
        {
          "epoch": 994,
          "updates": 15904,
          "samples_seen": 1017856,
          "online_loss": 7.445101937264553e-05,
          "train_loss": 7.07124563632533e-05,
          "validation_loss": 6.276706699281931e-05,
          "train_metric": 7.121899670892162e-06,
          "validation_metric": 6.321668934106128e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.637649087000682
        },
        {
          "epoch": 995,
          "updates": 15920,
          "samples_seen": 1018880,
          "online_loss": 7.294853116945887e-05,
          "train_loss": 7.138500950532034e-05,
          "validation_loss": 6.34502139291726e-05,
          "train_metric": 7.1896365625434555e-06,
          "validation_metric": 6.390473572537303e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.647870660999615
        },
        {
          "epoch": 996,
          "updates": 15936,
          "samples_seen": 1019904,
          "online_loss": 7.171227321123297e-05,
          "train_loss": 7.023786019999534e-05,
          "validation_loss": 6.243402458494529e-05,
          "train_metric": 7.074100722093135e-06,
          "validation_metric": 6.288126314757392e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.658200566000232
        },
        {
          "epoch": 997,
          "updates": 15952,
          "samples_seen": 1020928,
          "online_loss": 7.182856484178046e-05,
          "train_loss": 7.020658085821196e-05,
          "validation_loss": 6.249643774935976e-05,
          "train_metric": 7.070950232446194e-06,
          "validation_metric": 6.294412742136046e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.668352908000088
        },
        {
          "epoch": 998,
          "updates": 15968,
          "samples_seen": 1021952,
          "online_loss": 7.169044511101674e-05,
          "train_loss": 7.049848500173539e-05,
          "validation_loss": 6.267237040447071e-05,
          "train_metric": 7.100349648681004e-06,
          "validation_metric": 6.312131063168636e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.678592913000102
        },
        {
          "epoch": 999,
          "updates": 15984,
          "samples_seen": 1022976,
          "online_loss": 7.430633172589296e-05,
          "train_loss": 7.376378925982863e-05,
          "validation_loss": 6.505042256321758e-05,
          "train_metric": 7.429218385368586e-06,
          "validation_metric": 6.551641035912326e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.688822801999777
        },
        {
          "epoch": 1000,
          "updates": 16000,
          "samples_seen": 1024000,
          "online_loss": 7.284094431270205e-05,
          "train_loss": 7.365958299487829e-05,
          "validation_loss": 6.533785199280828e-05,
          "train_metric": 7.418723726004828e-06,
          "validation_metric": 6.580589342775056e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.699067499999728
        },
        {
          "epoch": 1001,
          "updates": 16016,
          "samples_seen": 1025024,
          "online_loss": 7.18908495400683e-05,
          "train_loss": 7.031767017906532e-05,
          "validation_loss": 6.365219451254234e-05,
          "train_metric": 7.082138836267404e-06,
          "validation_metric": 6.410816240531858e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.709305876000144
        },
        {
          "epoch": 1002,
          "updates": 16032,
          "samples_seen": 1026048,
          "online_loss": 7.231284143927041e-05,
          "train_loss": 7.030711276456714e-05,
          "validation_loss": 6.158594624139369e-05,
          "train_metric": 7.08107472746633e-06,
          "validation_metric": 6.202711119840387e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.720233934000134
        },
        {
          "epoch": 1003,
          "updates": 16048,
          "samples_seen": 1027072,
          "online_loss": 8.067917451626272e-05,
          "train_loss": 8.913745114114136e-05,
          "validation_loss": 7.219273538794369e-05,
          "train_metric": 8.977597644843627e-06,
          "validation_metric": 7.270987680385588e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.730716897000093
        },
        {
          "epoch": 1004,
          "updates": 16064,
          "samples_seen": 1028096,
          "online_loss": 7.147197948143003e-05,
          "train_loss": 7.075941539369524e-05,
          "validation_loss": 6.293587648542598e-05,
          "train_metric": 7.12662813384668e-06,
          "validation_metric": 6.338671028061071e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.741096472000208
        },
        {
          "epoch": 1005,
          "updates": 16080,
          "samples_seen": 1029120,
          "online_loss": 7.459813309651508e-05,
          "train_loss": 7.021337660262361e-05,
          "validation_loss": 6.253947503864765e-05,
          "train_metric": 7.071634627209278e-06,
          "validation_metric": 6.298746939137345e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.751276905999475
        },
        {
          "epoch": 1006,
          "updates": 16096,
          "samples_seen": 1030144,
          "online_loss": 7.284453408828995e-05,
          "train_loss": 7.374978304142132e-05,
          "validation_loss": 6.609928095713258e-05,
          "train_metric": 7.4278086685808375e-06,
          "validation_metric": 6.6572774812811986e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.76162230800037
        },
        {
          "epoch": 1007,
          "updates": 16112,
          "samples_seen": 1031168,
          "online_loss": 7.2220127321998e-05,
          "train_loss": 6.972294067963958e-05,
          "validation_loss": 6.235293403733522e-05,
          "train_metric": 7.022240424703341e-06,
          "validation_metric": 6.279959961830173e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.771861650999199
        },
        {
          "epoch": 1008,
          "updates": 16128,
          "samples_seen": 1032192,
          "online_loss": 7.235808811856259e-05,
          "train_loss": 7.645154255442321e-05,
          "validation_loss": 6.62289239699021e-05,
          "train_metric": 7.699918569414876e-06,
          "validation_metric": 6.670334641967202e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.782118576998982
        },
        {
          "epoch": 1009,
          "updates": 16144,
          "samples_seen": 1033216,
          "online_loss": 7.161234157138097e-05,
          "train_loss": 7.358538277912885e-05,
          "validation_loss": 6.465660408139229e-05,
          "train_metric": 7.411249953293009e-06,
          "validation_metric": 6.511975698231254e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.792348659999334
        },
        {
          "epoch": 1010,
          "updates": 16160,
          "samples_seen": 1034240,
          "online_loss": 7.348210579039005e-05,
          "train_loss": 7.006180385360494e-05,
          "validation_loss": 6.310472235782072e-05,
          "train_metric": 7.05636921338737e-06,
          "validation_metric": 6.355676759994822e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.802584953000405
        },
        {
          "epoch": 1011,
          "updates": 16176,
          "samples_seen": 1035264,
          "online_loss": 7.156118113016419e-05,
          "train_loss": 7.13204572093673e-05,
          "validation_loss": 6.172298162709922e-05,
          "train_metric": 7.183135039667832e-06,
          "validation_metric": 6.216513156687142e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.812916946999394
        },
        {
          "epoch": 1012,
          "updates": 16192,
          "samples_seen": 1036288,
          "online_loss": 7.250421458593337e-05,
          "train_loss": 7.033635483821854e-05,
          "validation_loss": 6.278397631831467e-05,
          "train_metric": 7.0840201260580216e-06,
          "validation_metric": 6.323372417682549e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.823173319999114
        },
        {
          "epoch": 1013,
          "updates": 16208,
          "samples_seen": 1037312,
          "online_loss": 7.133673807402374e-05,
          "train_loss": 7.162626570789143e-05,
          "validation_loss": 6.414310337277129e-05,
          "train_metric": 7.213934622996021e-06,
          "validation_metric": 6.460258191509638e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.833346223000262
        },
        {
          "epoch": 1014,
          "updates": 16224,
          "samples_seen": 1038336,
          "online_loss": 7.143947141230456e-05,
          "train_loss": 7.059094787109643e-05,
          "validation_loss": 6.373005453497171e-05,
          "train_metric": 7.109661964932457e-06,
          "validation_metric": 6.418657449103193e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.843603944998904
        },
        {
          "epoch": 1015,
          "updates": 16240,
          "samples_seen": 1039360,
          "online_loss": 7.801393394402112e-05,
          "train_loss": 8.392266317969188e-05,
          "validation_loss": 7.148047734517604e-05,
          "train_metric": 8.4523844634532e-06,
          "validation_metric": 7.199251285783248e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.909439245000613
        },
        {
          "epoch": 1016,
          "updates": 16256,
          "samples_seen": 1040384,
          "online_loss": 8.002261370165797e-05,
          "train_loss": 7.197511877166107e-05,
          "validation_loss": 6.426410254789516e-05,
          "train_metric": 7.249070677062264e-06,
          "validation_metric": 6.472444965766044e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.920129096999517
        },
        {
          "epoch": 1017,
          "updates": 16272,
          "samples_seen": 1041408,
          "online_loss": 7.280513250407239e-05,
          "train_loss": 7.03971745679155e-05,
          "validation_loss": 6.281134119490162e-05,
          "train_metric": 7.090146937116515e-06,
          "validation_metric": 6.326128641376272e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.930346994999127
        },
        {
          "epoch": 1018,
          "updates": 16288,
          "samples_seen": 1042432,
          "online_loss": 7.483528679586016e-05,
          "train_loss": 7.013519643805921e-05,
          "validation_loss": 6.137190212029964e-05,
          "train_metric": 7.06376113157603e-06,
          "validation_metric": 6.181152912176913e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.941340190000119
        },
        {
          "epoch": 1019,
          "updates": 16304,
          "samples_seen": 1043456,
          "online_loss": 7.410608077407232e-05,
          "train_loss": 7.039818592602387e-05,
          "validation_loss": 6.134113209554926e-05,
          "train_metric": 7.090247891028412e-06,
          "validation_metric": 6.178053808980621e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.952276598000026
        },
        {
          "epoch": 1020,
          "updates": 16320,
          "samples_seen": 1044480,
          "online_loss": 7.081614421622362e-05,
          "train_loss": 6.974448479013517e-05,
          "validation_loss": 6.247397686820477e-05,
          "train_metric": 7.024408660072368e-06,
          "validation_metric": 6.292149919318035e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.96272000099998
        },
        {
          "epoch": 1021,
          "updates": 16336,
          "samples_seen": 1045504,
          "online_loss": 7.076586871335167e-05,
          "train_loss": 7.365513738477603e-05,
          "validation_loss": 6.468909123213962e-05,
          "train_metric": 7.418275345116854e-06,
          "validation_metric": 6.515248060168233e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.972997405999195
        },
        {
          "epoch": 1022,
          "updates": 16352,
          "samples_seen": 1046528,
          "online_loss": 7.231142967611959e-05,
          "train_loss": 6.96308707119897e-05,
          "validation_loss": 6.253670289879665e-05,
          "train_metric": 7.0129663072293624e-06,
          "validation_metric": 6.298466814769199e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.983298589999322
        },
        {
          "epoch": 1023,
          "updates": 16368,
          "samples_seen": 1047552,
          "online_loss": 7.110202932381071e-05,
          "train_loss": 6.963318446651101e-05,
          "validation_loss": 6.159961048979312e-05,
          "train_metric": 7.013199592620367e-06,
          "validation_metric": 6.204086730576819e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 11.993596082000295
        },
        {
          "epoch": 1024,
          "updates": 16384,
          "samples_seen": 1048576,
          "online_loss": 7.295681848518143e-05,
          "train_loss": 7.817666482878849e-05,
          "validation_loss": 6.645719986408949e-05,
          "train_metric": 7.873667527746875e-06,
          "validation_metric": 6.6933253037859686e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.003957929000535
        },
        {
          "epoch": 1025,
          "updates": 16400,
          "samples_seen": 1049600,
          "online_loss": 7.38014703074441e-05,
          "train_loss": 6.9103974965401e-05,
          "validation_loss": 6.094233685871586e-05,
          "train_metric": 6.959899110370316e-06,
          "validation_metric": 6.137888703960925e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.014734827000211
        },
        {
          "epoch": 1026,
          "updates": 16416,
          "samples_seen": 1050624,
          "online_loss": 7.152869306992216e-05,
          "train_loss": 7.683830335736275e-05,
          "validation_loss": 6.593568105017766e-05,
          "train_metric": 7.738872227491811e-06,
          "validation_metric": 6.640800165769178e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.02517589099989
        },
        {
          "epoch": 1027,
          "updates": 16432,
          "samples_seen": 1051648,
          "online_loss": 7.058648657221056e-05,
          "train_loss": 7.899045886006206e-05,
          "validation_loss": 7.070299034239724e-05,
          "train_metric": 7.95563028077595e-06,
          "validation_metric": 7.1209451562026516e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.03540704199986
        },
        {
          "epoch": 1028,
          "updates": 16448,
          "samples_seen": 1052672,
          "online_loss": 7.279959686457005e-05,
          "train_loss": 6.92684916430153e-05,
          "validation_loss": 6.127018423285335e-05,
          "train_metric": 6.976468739594566e-06,
          "validation_metric": 6.170908363856142e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.045733541999653
        },
        {
          "epoch": 1029,
          "updates": 16464,
          "samples_seen": 1053696,
          "online_loss": 7.143096013351169e-05,
          "train_loss": 7.325905608013272e-05,
          "validation_loss": 6.311180914053693e-05,
          "train_metric": 7.378383998002391e-06,
          "validation_metric": 6.356391168083064e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.055986314999245
        },
        {
          "epoch": 1030,
          "updates": 16480,
          "samples_seen": 1054720,
          "online_loss": 8.309656209348759e-05,
          "train_loss": 7.380266470136121e-05,
          "validation_loss": 6.473833491327241e-05,
          "train_metric": 7.433133760059718e-06,
          "validation_metric": 6.5202079895243514e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.066249729999981
        },
        {
          "epoch": 1031,
          "updates": 16496,
          "samples_seen": 1055744,
          "online_loss": 7.560686958640872e-05,
          "train_loss": 7.284867024281994e-05,
          "validation_loss": 6.36928016319871e-05,
          "train_metric": 7.337051101785619e-06,
          "validation_metric": 6.414906238205731e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.076578366999456
        },
        {
          "epoch": 1032,
          "updates": 16512,
          "samples_seen": 1056768,
          "online_loss": 7.347914652200416e-05,
          "train_loss": 7.140763045754284e-05,
          "validation_loss": 6.342351844068617e-05,
          "train_metric": 7.191915301518748e-06,
          "validation_metric": 6.387784651451511e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.086883665000641
        },
        {
          "epoch": 1033,
          "updates": 16528,
          "samples_seen": 1057792,
          "online_loss": 8.245664594141999e-05,
          "train_loss": 7.552999159088358e-05,
          "validation_loss": 6.570073310285807e-05,
          "train_metric": 7.607103725604247e-06,
          "validation_metric": 6.617137387365801e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.097062160999485
        },
        {
          "epoch": 1034,
          "updates": 16544,
          "samples_seen": 1058816,
          "online_loss": 8.040029524636338e-05,
          "train_loss": 7.976326742209494e-05,
          "validation_loss": 6.378209945978597e-05,
          "train_metric": 8.033464837353677e-06,
          "validation_metric": 6.4238997765642125e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.107375495999804
        },
        {
          "epoch": 1035,
          "updates": 16560,
          "samples_seen": 1059840,
          "online_loss": 7.813990396243753e-05,
          "train_loss": 7.214417564682662e-05,
          "validation_loss": 6.235016189748421e-05,
          "train_metric": 7.266097327374155e-06,
          "validation_metric": 6.279679837462027e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.117617808999057
        },
        {
          "epoch": 1036,
          "updates": 16576,
          "samples_seen": 1060864,
          "online_loss": 7.98213020516414e-05,
          "train_loss": 7.709225610597059e-05,
          "validation_loss": 6.587731331819668e-05,
          "train_metric": 7.764449946989771e-06,
          "validation_metric": 6.634921192016918e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.127897344998928
        },
        {
          "epoch": 1037,
          "updates": 16592,
          "samples_seen": 1061888,
          "online_loss": 7.635484507773072e-05,
          "train_loss": 6.93015826982446e-05,
          "validation_loss": 6.193747685756534e-05,
          "train_metric": 6.979802492423914e-06,
          "validation_metric": 6.238115474843653e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.138118279999617
        },
        {
          "epoch": 1038,
          "updates": 16608,
          "samples_seen": 1062912,
          "online_loss": 7.01873345860804e-05,
          "train_loss": 7.464336522389203e-05,
          "validation_loss": 6.652763113379478e-05,
          "train_metric": 7.517806352552725e-06,
          "validation_metric": 6.700419362459797e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.148639017999812
        },
        {
          "epoch": 1039,
          "updates": 16624,
          "samples_seen": 1063936,
          "online_loss": 7.3360549095014e-05,
          "train_loss": 8.99458464118652e-05,
          "validation_loss": 7.58831447456032e-05,
          "train_metric": 9.059016520041041e-06,
          "validation_metric": 7.642673153895885e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.159111249000489
        },
        {
          "epoch": 1040,
          "updates": 16640,
          "samples_seen": 1064960,
          "online_loss": 7.596574960189173e-05,
          "train_loss": 6.901158485561609e-05,
          "validation_loss": 5.9369776863604784e-05,
          "train_metric": 6.950594524823828e-06,
          "validation_metric": 5.979506113362731e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.16990597199947
        },
        {
          "epoch": 1041,
          "updates": 16656,
          "samples_seen": 1065984,
          "online_loss": 7.18043936558388e-05,
          "train_loss": 6.849879719084129e-05,
          "validation_loss": 6.151381239760667e-05,
          "train_metric": 6.898947049194248e-06,
          "validation_metric": 6.195446530909976e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.180456984999182
        },
        {
          "epoch": 1042,
          "updates": 16672,
          "samples_seen": 1067008,
          "online_loss": 6.932961514394265e-05,
          "train_loss": 7.224148430395871e-05,
          "validation_loss": 6.275603664107621e-05,
          "train_metric": 7.275896678038407e-06,
          "validation_metric": 6.3205588958226144e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.190861359000337
        },
        {
          "epoch": 1043,
          "updates": 16688,
          "samples_seen": 1068032,
          "online_loss": 7.386668630715576e-05,
          "train_loss": 6.964399653952569e-05,
          "validation_loss": 6.115133146522567e-05,
          "train_metric": 7.014288257778389e-06,
          "validation_metric": 6.158938504086109e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.201010237000446
        },
        {
          "epoch": 1044,
          "updates": 16704,
          "samples_seen": 1069056,
          "online_loss": 9.216114449372981e-05,
          "train_loss": 8.262797200586647e-05,
          "validation_loss": 7.267564797075465e-05,
          "train_metric": 8.321986570081208e-06,
          "validation_metric": 7.3196251832996495e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.211361978999776
        },
        {
          "epoch": 1045,
          "updates": 16720,
          "samples_seen": 1070080,
          "online_loss": 7.722219788774964e-05,
          "train_loss": 6.935153214726597e-05,
          "validation_loss": 6.0614860558416694e-05,
          "train_metric": 6.9848324528720696e-06,
          "validation_metric": 6.104906788095832e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.221586493998984
        },
        {
          "epoch": 1046,
          "updates": 16736,
          "samples_seen": 1071104,
          "online_loss": 7.666235615033656e-05,
          "train_loss": 7.251630449900404e-05,
          "validation_loss": 6.390884664142504e-05,
          "train_metric": 7.303577149286866e-06,
          "validation_metric": 6.4366649894509465e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.231810442999631
        },
        {
          "epoch": 1047,
          "updates": 16752,
          "samples_seen": 1072128,
          "online_loss": 7.150212127271516e-05,
          "train_loss": 7.145923154894263e-05,
          "validation_loss": 6.198648770805448e-05,
          "train_metric": 7.197112608992029e-06,
          "validation_metric": 6.243051302590175e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.242079454999839
        },
        {
          "epoch": 1048,
          "updates": 16768,
          "samples_seen": 1073152,
          "online_loss": 7.236283227030071e-05,
          "train_loss": 7.068708509905264e-05,
          "validation_loss": 6.456718256231397e-05,
          "train_metric": 7.11934399078018e-06,
          "validation_metric": 6.502970336441649e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.25231738399998
        },
        {
          "epoch": 1049,
          "updates": 16784,
          "samples_seen": 1074176,
          "online_loss": 7.969667876750464e-05,
          "train_loss": 7.107408600859344e-05,
          "validation_loss": 6.257005588850006e-05,
          "train_metric": 7.158321750466712e-06,
          "validation_metric": 6.301826488197548e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.26248913299969
        },
        {
          "epoch": 1050,
          "updates": 16800,
          "samples_seen": 1075200,
          "online_loss": 7.06345613252779e-05,
          "train_loss": 6.990945985307917e-05,
          "validation_loss": 6.0504004068206996e-05,
          "train_metric": 7.04102558302111e-06,
          "validation_metric": 6.093741831136867e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.27300388399999
        },
        {
          "epoch": 1051,
          "updates": 16816,
          "samples_seen": 1076224,
          "online_loss": 7.940628006508632e-05,
          "train_loss": 8.240915485657752e-05,
          "validation_loss": 7.198571256594732e-05,
          "train_metric": 8.299947694467846e-06,
          "validation_metric": 7.250137514347443e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.336505942999793
        },
        {
          "epoch": 1052,
          "updates": 16832,
          "samples_seen": 1077248,
          "online_loss": 7.932067683213972e-05,
          "train_loss": 7.009077671682462e-05,
          "validation_loss": 6.30290960543789e-05,
          "train_metric": 7.059286417643307e-06,
          "validation_metric": 6.3480592871201225e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.347246622999592
        },
        {
          "epoch": 1053,
          "updates": 16848,
          "samples_seen": 1078272,
          "online_loss": 7.303509210032644e-05,
          "train_loss": 7.498572813346982e-05,
          "validation_loss": 6.693640898447484e-05,
          "train_metric": 7.552287570433691e-06,
          "validation_metric": 6.741590368619654e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.357474156000535
        },
        {
          "epoch": 1054,
          "updates": 16864,
          "samples_seen": 1079296,
          "online_loss": 7.469869478882174e-05,
          "train_loss": 6.829061021562666e-05,
          "validation_loss": 5.951795174041763e-05,
          "train_metric": 6.877980922581628e-06,
          "validation_metric": 5.9944295571767725e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.368035941999551
        },
        {
          "epoch": 1055,
          "updates": 16880,
          "samples_seen": 1080320,
          "online_loss": 7.138672026485438e-05,
          "train_loss": 7.279132114490494e-05,
          "validation_loss": 6.23592350166291e-05,
          "train_metric": 7.33127535568201e-06,
          "validation_metric": 6.280594334384659e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.378304974999992
        },
        {
          "epoch": 1056,
          "updates": 16896,
          "samples_seen": 1081344,
          "online_loss": 7.251151669152023e-05,
          "train_loss": 6.801424024160951e-05,
          "validation_loss": 5.934637738391757e-05,
          "train_metric": 6.850144927739166e-06,
          "validation_metric": 5.977149612590438e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.389154855000015
        },
        {
          "epoch": 1057,
          "updates": 16912,
          "samples_seen": 1082368,
          "online_loss": 7.66048142395448e-05,
          "train_loss": 8.028426236705855e-05,
          "validation_loss": 6.934838893357664e-05,
          "train_metric": 8.085937224677764e-06,
          "validation_metric": 6.984515039221151e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.399479698000505
        },
        {
          "epoch": 1058,
          "updates": 16928,
          "samples_seen": 1083392,
          "online_loss": 7.495583099625946e-05,
          "train_loss": 6.981917249504477e-05,
          "validation_loss": 6.222155934665352e-05,
          "train_metric": 7.031932000245433e-06,
          "validation_metric": 6.266728178161429e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.409919243000331
        },
        {
          "epoch": 1059,
          "updates": 16944,
          "samples_seen": 1084416,
          "online_loss": 8.154217107403383e-05,
          "train_loss": 6.98304720572196e-05,
          "validation_loss": 6.020079308655113e-05,
          "train_metric": 7.03306932337e-06,
          "validation_metric": 6.063203272788087e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.420130660999348
        },
        {
          "epoch": 1060,
          "updates": 16960,
          "samples_seen": 1085440,
          "online_loss": 8.10602230103541e-05,
          "train_loss": 7.591229950776324e-05,
          "validation_loss": 6.524616037495434e-05,
          "train_metric": 7.645609002793208e-06,
          "validation_metric": 6.571354788320605e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.430406093999409
        },
        {
          "epoch": 1061,
          "updates": 16976,
          "samples_seen": 1086464,
          "online_loss": 7.444023367497721e-05,
          "train_loss": 6.882959132781252e-05,
          "validation_loss": 6.076466161175631e-05,
          "train_metric": 6.932264113856945e-06,
          "validation_metric": 6.119994395703543e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.440714017999198
        },
        {
          "epoch": 1062,
          "updates": 16992,
          "samples_seen": 1087488,
          "online_loss": 7.149406883399934e-05,
          "train_loss": 7.03284313203767e-05,
          "validation_loss": 6.202299846336246e-05,
          "train_metric": 7.083221589709865e-06,
          "validation_metric": 6.246729753911495e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.451297505000184
        },
        {
          "epoch": 1063,
          "updates": 17008,
          "samples_seen": 1088512,
          "online_loss": 7.879584450165567e-05,
          "train_loss": 7.26285288692452e-05,
          "validation_loss": 6.277067586779594e-05,
          "train_metric": 7.314879439945798e-06,
          "validation_metric": 6.322033641481539e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.461507685999095
        },
        {
          "epoch": 1064,
          "updates": 17024,
          "samples_seen": 1089536,
          "online_loss": 7.263713723659748e-05,
          "train_loss": 6.774938083253801e-05,
          "validation_loss": 5.940779374213889e-05,
          "train_metric": 6.823469902883517e-06,
          "validation_metric": 5.983335995551897e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.472102317000463
        },
        {
          "epoch": 1065,
          "updates": 17040,
          "samples_seen": 1090560,
          "online_loss": 7.082524075485708e-05,
          "train_loss": 7.971606100909412e-05,
          "validation_loss": 6.619557098019868e-05,
          "train_metric": 8.028709089558106e-06,
          "validation_metric": 6.666974968538852e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.482364076999147
        },
        {
          "epoch": 1066,
          "updates": 17056,
          "samples_seen": 1091584,
          "online_loss": 8.093819315035944e-05,
          "train_loss": 8.107697794912383e-05,
          "validation_loss": 6.402095459634438e-05,
          "train_metric": 8.165775398083497e-06,
          "validation_metric": 6.447956366173457e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.492837796000458
        },
        {
          "epoch": 1067,
          "updates": 17072,
          "samples_seen": 1092608,
          "online_loss": 7.142118579395174e-05,
          "train_loss": 6.824378942837939e-05,
          "validation_loss": 6.026302435202524e-05,
          "train_metric": 6.8732647378055844e-06,
          "validation_metric": 6.069471510272706e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.503125318000457
        },
        {
          "epoch": 1068,
          "updates": 17088,
          "samples_seen": 1093632,
          "online_loss": 6.964777435314318e-05,
          "train_loss": 6.849434430478141e-05,
          "validation_loss": 6.071232201065868e-05,
          "train_metric": 6.898498668306274e-06,
          "validation_metric": 6.114722054917365e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.513660464999703
        },
        {
          "epoch": 1069,
          "updates": 17104,
          "samples_seen": 1094656,
          "online_loss": 7.042115157673834e-05,
          "train_loss": 6.87309802742675e-05,
          "validation_loss": 5.98418373556342e-05,
          "train_metric": 6.922332886460936e-06,
          "validation_metric": 6.027050858392613e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.524120770000081
        },
        {
          "epoch": 1070,
          "updates": 17120,
          "samples_seen": 1095680,
          "online_loss": 7.20223927146435e-05,
          "train_loss": 6.819171539973468e-05,
          "validation_loss": 6.0367758123902604e-05,
          "train_metric": 6.868019227113109e-06,
          "validation_metric": 6.080019829823868e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.534422230999553
        },
        {
          "epoch": 1071,
          "updates": 17136,
          "samples_seen": 1096704,
          "online_loss": 7.121827275113901e-05,
          "train_loss": 7.294879469554871e-05,
          "validation_loss": 6.236475746845827e-05,
          "train_metric": 7.34713694328093e-06,
          "validation_metric": 6.281149580900092e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.544895693999933
        },
        {
          "epoch": 1072,
          "updates": 17152,
          "samples_seen": 1097728,
          "online_loss": 6.983157618378755e-05,
          "train_loss": 6.874514656374231e-05,
          "validation_loss": 5.9621896070893854e-05,
          "train_metric": 6.923760338395368e-06,
          "validation_metric": 6.004898750688881e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.555303498000285
        },
        {
          "epoch": 1073,
          "updates": 17168,
          "samples_seen": 1098752,
          "online_loss": 7.014388575043995e-05,
          "train_loss": 6.7899476562161e-05,
          "validation_loss": 6.079831291572191e-05,
          "train_metric": 6.838587523816386e-06,
          "validation_metric": 6.123383172962349e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.565995752000163
        },
        {
          "epoch": 1074,
          "updates": 17184,
          "samples_seen": 1099776,
          "online_loss": 7.108207933015365e-05,
          "train_loss": 7.065743557177484e-05,
          "validation_loss": 6.075187411624938e-05,
          "train_metric": 7.11635811967426e-06,
          "validation_metric": 6.1187065512058325e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.576471857999422
        },
        {
          "epoch": 1075,
          "updates": 17200,
          "samples_seen": 1100800,
          "online_loss": 6.989544817770366e-05,
          "train_loss": 6.831866630818695e-05,
          "validation_loss": 5.955630331300199e-05,
          "train_metric": 6.8808062678726856e-06,
          "validation_metric": 5.998292181175202e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.586667542000214
        },
        {
          "epoch": 1076,
          "updates": 17216,
          "samples_seen": 1101824,
          "online_loss": 6.956317406547896e-05,
          "train_loss": 6.782849231967703e-05,
          "validation_loss": 6.087131259846501e-05,
          "train_metric": 6.831437985965749e-06,
          "validation_metric": 6.130734163889429e-06,
          "lr_first": 1e-05,
          "lr_last": 1e-05,
          "seconds": 12.596906221000609
        }
      ],
      "best": {
        "epoch": 1056,
        "updates": 16896
      },
      "best_validation_metric": 5.977149612590438e-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": 0.0024448209776158332
    }
  },
  "epochs": 50000,
  "patience": 20,
  "seeds": [
    11,
    22,
    33
  ],
  "workers": 3,
  "threads": 1,
  "seconds": 14.055191027000546,
  "selected_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
  },
  "selected_id": "14985df6b90f",
  "test": [
    {
      "seed": 11,
      "rmse_m_s": 0.00459812110996388
    },
    {
      "seed": 22,
      "rmse_m_s": 0.0028561058706251165
    },
    {
      "seed": 33,
      "rmse_m_s": 0.006834206351271969
    }
  ],
  "source_sha256": {
    "training_tools.py": "73f9c91df0bdc2d0681a2025d3d554d11d289e1fbfed038047dec4277c3e6236",
    "training_strategies_experiments.py": "d45307e44b19857ad48314ab8933ed928537b17c540020af6681d6dcc062c9bd",
    "training_final_experiment.py": "10a3c9fcb098288872a0c8e2675871264fb69b47df184edadb79b02667ed4f72"
  }
}