{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 25527,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001175226231049477,
      "grad_norm": 9.590228080749512,
      "learning_rate": 4.5e-06,
      "loss": 4.5746,
      "step": 10
    },
    {
      "epoch": 0.002350452462098954,
      "grad_norm": 11.64316463470459,
      "learning_rate": 9.5e-06,
      "loss": 4.4941,
      "step": 20
    },
    {
      "epoch": 0.003525678693148431,
      "grad_norm": 9.773070335388184,
      "learning_rate": 1.45e-05,
      "loss": 4.3607,
      "step": 30
    },
    {
      "epoch": 0.004700904924197908,
      "grad_norm": 9.516993522644043,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 4.0584,
      "step": 40
    },
    {
      "epoch": 0.005876131155247385,
      "grad_norm": 9.81268310546875,
      "learning_rate": 2.45e-05,
      "loss": 3.8522,
      "step": 50
    },
    {
      "epoch": 0.007051357386296862,
      "grad_norm": 8.22573184967041,
      "learning_rate": 2.95e-05,
      "loss": 3.4889,
      "step": 60
    },
    {
      "epoch": 0.00822658361734634,
      "grad_norm": 9.205648422241211,
      "learning_rate": 3.45e-05,
      "loss": 3.7265,
      "step": 70
    },
    {
      "epoch": 0.009401809848395816,
      "grad_norm": 10.11063289642334,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 3.1641,
      "step": 80
    },
    {
      "epoch": 0.010577036079445294,
      "grad_norm": 9.774422645568848,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 2.9166,
      "step": 90
    },
    {
      "epoch": 0.01175226231049477,
      "grad_norm": 10.686172485351562,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 3.1807,
      "step": 100
    },
    {
      "epoch": 0.012927488541544248,
      "grad_norm": 9.6532621383667,
      "learning_rate": 4.9982302277107015e-05,
      "loss": 2.7838,
      "step": 110
    },
    {
      "epoch": 0.014102714772593724,
      "grad_norm": 9.985005378723145,
      "learning_rate": 4.996263814055925e-05,
      "loss": 3.216,
      "step": 120
    },
    {
      "epoch": 0.015277941003643202,
      "grad_norm": 9.0400972366333,
      "learning_rate": 4.994297400401148e-05,
      "loss": 2.8816,
      "step": 130
    },
    {
      "epoch": 0.01645316723469268,
      "grad_norm": 12.565206527709961,
      "learning_rate": 4.992330986746372e-05,
      "loss": 2.8579,
      "step": 140
    },
    {
      "epoch": 0.017628393465742156,
      "grad_norm": 7.827625751495361,
      "learning_rate": 4.990364573091596e-05,
      "loss": 3.4208,
      "step": 150
    },
    {
      "epoch": 0.018803619696791632,
      "grad_norm": 7.236791610717773,
      "learning_rate": 4.9883981594368196e-05,
      "loss": 2.5219,
      "step": 160
    },
    {
      "epoch": 0.019978845927841108,
      "grad_norm": 11.720057487487793,
      "learning_rate": 4.986431745782043e-05,
      "loss": 2.6236,
      "step": 170
    },
    {
      "epoch": 0.021154072158890588,
      "grad_norm": 6.441433429718018,
      "learning_rate": 4.9844653321272664e-05,
      "loss": 3.0318,
      "step": 180
    },
    {
      "epoch": 0.022329298389940064,
      "grad_norm": 6.328097343444824,
      "learning_rate": 4.9824989184724904e-05,
      "loss": 2.9817,
      "step": 190
    },
    {
      "epoch": 0.02350452462098954,
      "grad_norm": 7.321470260620117,
      "learning_rate": 4.980532504817714e-05,
      "loss": 3.1743,
      "step": 200
    },
    {
      "epoch": 0.024679750852039016,
      "grad_norm": 7.99439811706543,
      "learning_rate": 4.978566091162937e-05,
      "loss": 2.7966,
      "step": 210
    },
    {
      "epoch": 0.025854977083088496,
      "grad_norm": 6.283413410186768,
      "learning_rate": 4.9765996775081605e-05,
      "loss": 2.7353,
      "step": 220
    },
    {
      "epoch": 0.027030203314137972,
      "grad_norm": 9.933158874511719,
      "learning_rate": 4.9746332638533845e-05,
      "loss": 2.5699,
      "step": 230
    },
    {
      "epoch": 0.028205429545187448,
      "grad_norm": 10.819753646850586,
      "learning_rate": 4.972666850198608e-05,
      "loss": 3.169,
      "step": 240
    },
    {
      "epoch": 0.029380655776236924,
      "grad_norm": 10.243263244628906,
      "learning_rate": 4.970700436543831e-05,
      "loss": 2.8204,
      "step": 250
    },
    {
      "epoch": 0.030555882007286404,
      "grad_norm": 9.26700496673584,
      "learning_rate": 4.968734022889055e-05,
      "loss": 3.1053,
      "step": 260
    },
    {
      "epoch": 0.03173110823833588,
      "grad_norm": 9.63497543334961,
      "learning_rate": 4.9667676092342786e-05,
      "loss": 3.4401,
      "step": 270
    },
    {
      "epoch": 0.03290633446938536,
      "grad_norm": 14.208921432495117,
      "learning_rate": 4.964801195579502e-05,
      "loss": 2.5517,
      "step": 280
    },
    {
      "epoch": 0.03408156070043483,
      "grad_norm": 8.876089096069336,
      "learning_rate": 4.962834781924726e-05,
      "loss": 2.4533,
      "step": 290
    },
    {
      "epoch": 0.03525678693148431,
      "grad_norm": 10.713753700256348,
      "learning_rate": 4.9608683682699494e-05,
      "loss": 2.8526,
      "step": 300
    },
    {
      "epoch": 0.03643201316253379,
      "grad_norm": 7.973067760467529,
      "learning_rate": 4.9589019546151734e-05,
      "loss": 2.4754,
      "step": 310
    },
    {
      "epoch": 0.037607239393583264,
      "grad_norm": 10.162976264953613,
      "learning_rate": 4.956935540960397e-05,
      "loss": 2.8464,
      "step": 320
    },
    {
      "epoch": 0.03878246562463274,
      "grad_norm": 9.717177391052246,
      "learning_rate": 4.95496912730562e-05,
      "loss": 2.5793,
      "step": 330
    },
    {
      "epoch": 0.039957691855682216,
      "grad_norm": 8.374249458312988,
      "learning_rate": 4.953002713650844e-05,
      "loss": 2.3008,
      "step": 340
    },
    {
      "epoch": 0.041132918086731696,
      "grad_norm": 9.05173110961914,
      "learning_rate": 4.9510362999960676e-05,
      "loss": 2.1993,
      "step": 350
    },
    {
      "epoch": 0.042308144317781175,
      "grad_norm": 10.086522102355957,
      "learning_rate": 4.949069886341291e-05,
      "loss": 3.0577,
      "step": 360
    },
    {
      "epoch": 0.04348337054883065,
      "grad_norm": 9.100815773010254,
      "learning_rate": 4.947103472686514e-05,
      "loss": 2.6398,
      "step": 370
    },
    {
      "epoch": 0.04465859677988013,
      "grad_norm": 15.208490371704102,
      "learning_rate": 4.945137059031738e-05,
      "loss": 2.2727,
      "step": 380
    },
    {
      "epoch": 0.04583382301092961,
      "grad_norm": 12.444480895996094,
      "learning_rate": 4.943170645376962e-05,
      "loss": 2.9964,
      "step": 390
    },
    {
      "epoch": 0.04700904924197908,
      "grad_norm": 18.707473754882812,
      "learning_rate": 4.941204231722185e-05,
      "loss": 2.6308,
      "step": 400
    },
    {
      "epoch": 0.04818427547302856,
      "grad_norm": 9.994440078735352,
      "learning_rate": 4.9392378180674084e-05,
      "loss": 2.4073,
      "step": 410
    },
    {
      "epoch": 0.04935950170407803,
      "grad_norm": 14.023969650268555,
      "learning_rate": 4.9372714044126324e-05,
      "loss": 2.4941,
      "step": 420
    },
    {
      "epoch": 0.05053472793512751,
      "grad_norm": 11.976531028747559,
      "learning_rate": 4.935304990757856e-05,
      "loss": 2.3695,
      "step": 430
    },
    {
      "epoch": 0.05170995416617699,
      "grad_norm": 10.255765914916992,
      "learning_rate": 4.93333857710308e-05,
      "loss": 2.914,
      "step": 440
    },
    {
      "epoch": 0.052885180397226464,
      "grad_norm": 10.70988941192627,
      "learning_rate": 4.931372163448303e-05,
      "loss": 2.5756,
      "step": 450
    },
    {
      "epoch": 0.054060406628275944,
      "grad_norm": 9.749191284179688,
      "learning_rate": 4.929405749793527e-05,
      "loss": 3.3529,
      "step": 460
    },
    {
      "epoch": 0.05523563285932542,
      "grad_norm": 18.65987205505371,
      "learning_rate": 4.9274393361387506e-05,
      "loss": 2.4748,
      "step": 470
    },
    {
      "epoch": 0.056410859090374896,
      "grad_norm": 12.78252124786377,
      "learning_rate": 4.925472922483974e-05,
      "loss": 2.8525,
      "step": 480
    },
    {
      "epoch": 0.057586085321424375,
      "grad_norm": 10.084146499633789,
      "learning_rate": 4.923506508829197e-05,
      "loss": 2.4496,
      "step": 490
    },
    {
      "epoch": 0.05876131155247385,
      "grad_norm": 6.39194393157959,
      "learning_rate": 4.9215400951744214e-05,
      "loss": 2.2924,
      "step": 500
    },
    {
      "epoch": 0.05993653778352333,
      "grad_norm": 8.746315956115723,
      "learning_rate": 4.919573681519645e-05,
      "loss": 2.0585,
      "step": 510
    },
    {
      "epoch": 0.06111176401457281,
      "grad_norm": 10.933807373046875,
      "learning_rate": 4.917607267864868e-05,
      "loss": 2.6111,
      "step": 520
    },
    {
      "epoch": 0.06228699024562228,
      "grad_norm": 11.978423118591309,
      "learning_rate": 4.915640854210092e-05,
      "loss": 3.1402,
      "step": 530
    },
    {
      "epoch": 0.06346221647667176,
      "grad_norm": 15.995674133300781,
      "learning_rate": 4.9136744405553155e-05,
      "loss": 2.4219,
      "step": 540
    },
    {
      "epoch": 0.06463744270772123,
      "grad_norm": 10.61567211151123,
      "learning_rate": 4.911708026900539e-05,
      "loss": 2.0817,
      "step": 550
    },
    {
      "epoch": 0.06581266893877072,
      "grad_norm": 11.215569496154785,
      "learning_rate": 4.909741613245762e-05,
      "loss": 2.804,
      "step": 560
    },
    {
      "epoch": 0.06698789516982019,
      "grad_norm": 9.939019203186035,
      "learning_rate": 4.907775199590986e-05,
      "loss": 2.5731,
      "step": 570
    },
    {
      "epoch": 0.06816312140086966,
      "grad_norm": 11.719276428222656,
      "learning_rate": 4.9058087859362096e-05,
      "loss": 2.0223,
      "step": 580
    },
    {
      "epoch": 0.06933834763191915,
      "grad_norm": 19.698078155517578,
      "learning_rate": 4.9038423722814337e-05,
      "loss": 2.9708,
      "step": 590
    },
    {
      "epoch": 0.07051357386296862,
      "grad_norm": 10.88604736328125,
      "learning_rate": 4.901875958626657e-05,
      "loss": 2.5684,
      "step": 600
    },
    {
      "epoch": 0.0716888000940181,
      "grad_norm": 4.218529224395752,
      "learning_rate": 4.8999095449718804e-05,
      "loss": 2.4295,
      "step": 610
    },
    {
      "epoch": 0.07286402632506758,
      "grad_norm": 18.491594314575195,
      "learning_rate": 4.8979431313171044e-05,
      "loss": 2.2901,
      "step": 620
    },
    {
      "epoch": 0.07403925255611706,
      "grad_norm": 7.509721755981445,
      "learning_rate": 4.895976717662328e-05,
      "loss": 2.6479,
      "step": 630
    },
    {
      "epoch": 0.07521447878716653,
      "grad_norm": 10.260770797729492,
      "learning_rate": 4.894010304007551e-05,
      "loss": 2.6717,
      "step": 640
    },
    {
      "epoch": 0.076389705018216,
      "grad_norm": 11.057743072509766,
      "learning_rate": 4.892043890352775e-05,
      "loss": 2.8309,
      "step": 650
    },
    {
      "epoch": 0.07756493124926549,
      "grad_norm": 11.912470817565918,
      "learning_rate": 4.8900774766979985e-05,
      "loss": 2.4458,
      "step": 660
    },
    {
      "epoch": 0.07874015748031496,
      "grad_norm": 13.406147956848145,
      "learning_rate": 4.888111063043222e-05,
      "loss": 2.7975,
      "step": 670
    },
    {
      "epoch": 0.07991538371136443,
      "grad_norm": 10.94129753112793,
      "learning_rate": 4.886144649388445e-05,
      "loss": 2.2264,
      "step": 680
    },
    {
      "epoch": 0.08109060994241392,
      "grad_norm": 11.463362693786621,
      "learning_rate": 4.884178235733669e-05,
      "loss": 2.1515,
      "step": 690
    },
    {
      "epoch": 0.08226583617346339,
      "grad_norm": 11.720123291015625,
      "learning_rate": 4.8822118220788927e-05,
      "loss": 2.6853,
      "step": 700
    },
    {
      "epoch": 0.08344106240451286,
      "grad_norm": 18.693950653076172,
      "learning_rate": 4.880245408424116e-05,
      "loss": 2.7192,
      "step": 710
    },
    {
      "epoch": 0.08461628863556235,
      "grad_norm": 15.79572868347168,
      "learning_rate": 4.8782789947693394e-05,
      "loss": 2.1117,
      "step": 720
    },
    {
      "epoch": 0.08579151486661182,
      "grad_norm": 10.005640029907227,
      "learning_rate": 4.8763125811145634e-05,
      "loss": 2.227,
      "step": 730
    },
    {
      "epoch": 0.0869667410976613,
      "grad_norm": 9.486937522888184,
      "learning_rate": 4.8743461674597875e-05,
      "loss": 2.3384,
      "step": 740
    },
    {
      "epoch": 0.08814196732871078,
      "grad_norm": 27.901615142822266,
      "learning_rate": 4.872379753805011e-05,
      "loss": 2.395,
      "step": 750
    },
    {
      "epoch": 0.08931719355976026,
      "grad_norm": 13.90008544921875,
      "learning_rate": 4.870413340150234e-05,
      "loss": 2.4109,
      "step": 760
    },
    {
      "epoch": 0.09049241979080973,
      "grad_norm": 10.572047233581543,
      "learning_rate": 4.868446926495458e-05,
      "loss": 1.9994,
      "step": 770
    },
    {
      "epoch": 0.09166764602185921,
      "grad_norm": 12.800186157226562,
      "learning_rate": 4.8664805128406816e-05,
      "loss": 3.2189,
      "step": 780
    },
    {
      "epoch": 0.09284287225290869,
      "grad_norm": 17.034366607666016,
      "learning_rate": 4.864514099185905e-05,
      "loss": 2.4533,
      "step": 790
    },
    {
      "epoch": 0.09401809848395816,
      "grad_norm": 8.7072172164917,
      "learning_rate": 4.862547685531128e-05,
      "loss": 2.171,
      "step": 800
    },
    {
      "epoch": 0.09519332471500763,
      "grad_norm": 7.957915306091309,
      "learning_rate": 4.8605812718763523e-05,
      "loss": 2.0721,
      "step": 810
    },
    {
      "epoch": 0.09636855094605712,
      "grad_norm": 6.7745890617370605,
      "learning_rate": 4.858614858221576e-05,
      "loss": 2.3687,
      "step": 820
    },
    {
      "epoch": 0.09754377717710659,
      "grad_norm": 15.410567283630371,
      "learning_rate": 4.856648444566799e-05,
      "loss": 2.7417,
      "step": 830
    },
    {
      "epoch": 0.09871900340815606,
      "grad_norm": 9.778885841369629,
      "learning_rate": 4.854682030912023e-05,
      "loss": 2.8093,
      "step": 840
    },
    {
      "epoch": 0.09989422963920555,
      "grad_norm": 8.895602226257324,
      "learning_rate": 4.8527156172572465e-05,
      "loss": 2.2858,
      "step": 850
    },
    {
      "epoch": 0.10106945587025502,
      "grad_norm": 10.767144203186035,
      "learning_rate": 4.85074920360247e-05,
      "loss": 2.3874,
      "step": 860
    },
    {
      "epoch": 0.1022446821013045,
      "grad_norm": 17.409183502197266,
      "learning_rate": 4.848782789947693e-05,
      "loss": 2.9872,
      "step": 870
    },
    {
      "epoch": 0.10341990833235398,
      "grad_norm": 10.862594604492188,
      "learning_rate": 4.846816376292917e-05,
      "loss": 2.51,
      "step": 880
    },
    {
      "epoch": 0.10459513456340346,
      "grad_norm": 19.80902862548828,
      "learning_rate": 4.844849962638141e-05,
      "loss": 2.2342,
      "step": 890
    },
    {
      "epoch": 0.10577036079445293,
      "grad_norm": 11.094535827636719,
      "learning_rate": 4.8428835489833646e-05,
      "loss": 1.8185,
      "step": 900
    },
    {
      "epoch": 0.10694558702550241,
      "grad_norm": 10.327164649963379,
      "learning_rate": 4.840917135328588e-05,
      "loss": 1.8309,
      "step": 910
    },
    {
      "epoch": 0.10812081325655189,
      "grad_norm": 20.061925888061523,
      "learning_rate": 4.838950721673812e-05,
      "loss": 1.8068,
      "step": 920
    },
    {
      "epoch": 0.10929603948760136,
      "grad_norm": 12.094876289367676,
      "learning_rate": 4.8369843080190354e-05,
      "loss": 2.6411,
      "step": 930
    },
    {
      "epoch": 0.11047126571865085,
      "grad_norm": 10.030497550964355,
      "learning_rate": 4.835017894364259e-05,
      "loss": 2.6722,
      "step": 940
    },
    {
      "epoch": 0.11164649194970032,
      "grad_norm": 26.68846321105957,
      "learning_rate": 4.833051480709482e-05,
      "loss": 2.7148,
      "step": 950
    },
    {
      "epoch": 0.11282171818074979,
      "grad_norm": 6.905248641967773,
      "learning_rate": 4.831085067054706e-05,
      "loss": 2.0387,
      "step": 960
    },
    {
      "epoch": 0.11399694441179928,
      "grad_norm": 12.575652122497559,
      "learning_rate": 4.8291186533999295e-05,
      "loss": 2.5451,
      "step": 970
    },
    {
      "epoch": 0.11517217064284875,
      "grad_norm": 30.023967742919922,
      "learning_rate": 4.827152239745153e-05,
      "loss": 2.5136,
      "step": 980
    },
    {
      "epoch": 0.11634739687389822,
      "grad_norm": 11.657227516174316,
      "learning_rate": 4.825185826090376e-05,
      "loss": 2.49,
      "step": 990
    },
    {
      "epoch": 0.1175226231049477,
      "grad_norm": 12.139093399047852,
      "learning_rate": 4.8232194124356e-05,
      "loss": 2.2525,
      "step": 1000
    },
    {
      "epoch": 0.11869784933599718,
      "grad_norm": 9.760957717895508,
      "learning_rate": 4.8212529987808236e-05,
      "loss": 2.1936,
      "step": 1010
    },
    {
      "epoch": 0.11987307556704666,
      "grad_norm": 19.141986846923828,
      "learning_rate": 4.819286585126047e-05,
      "loss": 2.2632,
      "step": 1020
    },
    {
      "epoch": 0.12104830179809613,
      "grad_norm": 11.168855667114258,
      "learning_rate": 4.8173201714712704e-05,
      "loss": 2.3453,
      "step": 1030
    },
    {
      "epoch": 0.12222352802914561,
      "grad_norm": 13.198949813842773,
      "learning_rate": 4.815353757816495e-05,
      "loss": 2.2629,
      "step": 1040
    },
    {
      "epoch": 0.12339875426019509,
      "grad_norm": 11.708571434020996,
      "learning_rate": 4.8133873441617184e-05,
      "loss": 2.5435,
      "step": 1050
    },
    {
      "epoch": 0.12457398049124456,
      "grad_norm": 13.318507194519043,
      "learning_rate": 4.811420930506942e-05,
      "loss": 2.2351,
      "step": 1060
    },
    {
      "epoch": 0.12574920672229403,
      "grad_norm": 15.794631958007812,
      "learning_rate": 4.809454516852165e-05,
      "loss": 2.7351,
      "step": 1070
    },
    {
      "epoch": 0.12692443295334352,
      "grad_norm": 24.130250930786133,
      "learning_rate": 4.807488103197389e-05,
      "loss": 2.5857,
      "step": 1080
    },
    {
      "epoch": 0.128099659184393,
      "grad_norm": 11.430598258972168,
      "learning_rate": 4.8055216895426126e-05,
      "loss": 2.1483,
      "step": 1090
    },
    {
      "epoch": 0.12927488541544246,
      "grad_norm": 7.576192378997803,
      "learning_rate": 4.803555275887836e-05,
      "loss": 2.197,
      "step": 1100
    },
    {
      "epoch": 0.13045011164649195,
      "grad_norm": 10.378724098205566,
      "learning_rate": 4.801588862233059e-05,
      "loss": 2.4228,
      "step": 1110
    },
    {
      "epoch": 0.13162533787754144,
      "grad_norm": 9.494956970214844,
      "learning_rate": 4.799622448578283e-05,
      "loss": 2.0244,
      "step": 1120
    },
    {
      "epoch": 0.1328005641085909,
      "grad_norm": 12.84150505065918,
      "learning_rate": 4.797656034923507e-05,
      "loss": 2.3792,
      "step": 1130
    },
    {
      "epoch": 0.13397579033964038,
      "grad_norm": 11.509659767150879,
      "learning_rate": 4.79568962126873e-05,
      "loss": 2.4997,
      "step": 1140
    },
    {
      "epoch": 0.13515101657068987,
      "grad_norm": 16.23176383972168,
      "learning_rate": 4.793723207613954e-05,
      "loss": 2.2318,
      "step": 1150
    },
    {
      "epoch": 0.13632624280173933,
      "grad_norm": 11.870035171508789,
      "learning_rate": 4.7917567939591774e-05,
      "loss": 2.3831,
      "step": 1160
    },
    {
      "epoch": 0.13750146903278881,
      "grad_norm": 21.441043853759766,
      "learning_rate": 4.789790380304401e-05,
      "loss": 1.9827,
      "step": 1170
    },
    {
      "epoch": 0.1386766952638383,
      "grad_norm": 14.712178230285645,
      "learning_rate": 4.787823966649624e-05,
      "loss": 2.4623,
      "step": 1180
    },
    {
      "epoch": 0.13985192149488776,
      "grad_norm": 13.701732635498047,
      "learning_rate": 4.785857552994848e-05,
      "loss": 2.0949,
      "step": 1190
    },
    {
      "epoch": 0.14102714772593725,
      "grad_norm": 21.27647590637207,
      "learning_rate": 4.783891139340072e-05,
      "loss": 2.519,
      "step": 1200
    },
    {
      "epoch": 0.14220237395698673,
      "grad_norm": 7.600924968719482,
      "learning_rate": 4.7819247256852956e-05,
      "loss": 2.365,
      "step": 1210
    },
    {
      "epoch": 0.1433776001880362,
      "grad_norm": 7.005496025085449,
      "learning_rate": 4.779958312030519e-05,
      "loss": 2.0264,
      "step": 1220
    },
    {
      "epoch": 0.14455282641908568,
      "grad_norm": 14.817516326904297,
      "learning_rate": 4.777991898375743e-05,
      "loss": 2.0793,
      "step": 1230
    },
    {
      "epoch": 0.14572805265013516,
      "grad_norm": 8.185234069824219,
      "learning_rate": 4.7760254847209664e-05,
      "loss": 1.987,
      "step": 1240
    },
    {
      "epoch": 0.14690327888118462,
      "grad_norm": 10.493677139282227,
      "learning_rate": 4.77405907106619e-05,
      "loss": 2.5908,
      "step": 1250
    },
    {
      "epoch": 0.1480785051122341,
      "grad_norm": 8.343379974365234,
      "learning_rate": 4.772092657411413e-05,
      "loss": 1.9069,
      "step": 1260
    },
    {
      "epoch": 0.14925373134328357,
      "grad_norm": 7.902554035186768,
      "learning_rate": 4.770126243756637e-05,
      "loss": 2.0049,
      "step": 1270
    },
    {
      "epoch": 0.15042895757433306,
      "grad_norm": 16.56104278564453,
      "learning_rate": 4.7681598301018605e-05,
      "loss": 2.7209,
      "step": 1280
    },
    {
      "epoch": 0.15160418380538254,
      "grad_norm": 18.097352981567383,
      "learning_rate": 4.766193416447084e-05,
      "loss": 2.3022,
      "step": 1290
    },
    {
      "epoch": 0.152779410036432,
      "grad_norm": 11.292770385742188,
      "learning_rate": 4.764227002792307e-05,
      "loss": 2.284,
      "step": 1300
    },
    {
      "epoch": 0.1539546362674815,
      "grad_norm": 12.159402847290039,
      "learning_rate": 4.762260589137531e-05,
      "loss": 2.2108,
      "step": 1310
    },
    {
      "epoch": 0.15512986249853097,
      "grad_norm": 14.429909706115723,
      "learning_rate": 4.7602941754827546e-05,
      "loss": 2.2802,
      "step": 1320
    },
    {
      "epoch": 0.15630508872958043,
      "grad_norm": 11.598194122314453,
      "learning_rate": 4.758327761827978e-05,
      "loss": 2.6336,
      "step": 1330
    },
    {
      "epoch": 0.15748031496062992,
      "grad_norm": 14.566303253173828,
      "learning_rate": 4.756361348173201e-05,
      "loss": 2.6178,
      "step": 1340
    },
    {
      "epoch": 0.1586555411916794,
      "grad_norm": 15.97060775756836,
      "learning_rate": 4.754394934518426e-05,
      "loss": 2.4354,
      "step": 1350
    },
    {
      "epoch": 0.15983076742272886,
      "grad_norm": 18.308185577392578,
      "learning_rate": 4.7524285208636494e-05,
      "loss": 2.0192,
      "step": 1360
    },
    {
      "epoch": 0.16100599365377835,
      "grad_norm": 12.767841339111328,
      "learning_rate": 4.750462107208873e-05,
      "loss": 1.7823,
      "step": 1370
    },
    {
      "epoch": 0.16218121988482784,
      "grad_norm": 10.590620040893555,
      "learning_rate": 4.748495693554096e-05,
      "loss": 1.5233,
      "step": 1380
    },
    {
      "epoch": 0.1633564461158773,
      "grad_norm": 11.266690254211426,
      "learning_rate": 4.74652927989932e-05,
      "loss": 2.3078,
      "step": 1390
    },
    {
      "epoch": 0.16453167234692678,
      "grad_norm": 8.440206527709961,
      "learning_rate": 4.7445628662445435e-05,
      "loss": 2.5174,
      "step": 1400
    },
    {
      "epoch": 0.16570689857797627,
      "grad_norm": 21.115808486938477,
      "learning_rate": 4.742596452589767e-05,
      "loss": 2.6034,
      "step": 1410
    },
    {
      "epoch": 0.16688212480902573,
      "grad_norm": 7.1569013595581055,
      "learning_rate": 4.74063003893499e-05,
      "loss": 2.3546,
      "step": 1420
    },
    {
      "epoch": 0.16805735104007521,
      "grad_norm": 11.650368690490723,
      "learning_rate": 4.738663625280214e-05,
      "loss": 2.4372,
      "step": 1430
    },
    {
      "epoch": 0.1692325772711247,
      "grad_norm": 10.576516151428223,
      "learning_rate": 4.7366972116254377e-05,
      "loss": 2.0734,
      "step": 1440
    },
    {
      "epoch": 0.17040780350217416,
      "grad_norm": 20.367582321166992,
      "learning_rate": 4.734730797970661e-05,
      "loss": 2.7651,
      "step": 1450
    },
    {
      "epoch": 0.17158302973322365,
      "grad_norm": 14.366838455200195,
      "learning_rate": 4.732764384315885e-05,
      "loss": 2.6322,
      "step": 1460
    },
    {
      "epoch": 0.17275825596427313,
      "grad_norm": 17.339632034301758,
      "learning_rate": 4.7307979706611084e-05,
      "loss": 1.987,
      "step": 1470
    },
    {
      "epoch": 0.1739334821953226,
      "grad_norm": 11.452990531921387,
      "learning_rate": 4.728831557006332e-05,
      "loss": 2.1176,
      "step": 1480
    },
    {
      "epoch": 0.17510870842637208,
      "grad_norm": 10.269814491271973,
      "learning_rate": 4.726865143351555e-05,
      "loss": 2.2955,
      "step": 1490
    },
    {
      "epoch": 0.17628393465742156,
      "grad_norm": 12.44279670715332,
      "learning_rate": 4.724898729696779e-05,
      "loss": 2.1097,
      "step": 1500
    },
    {
      "epoch": 0.17745916088847102,
      "grad_norm": 16.941822052001953,
      "learning_rate": 4.722932316042003e-05,
      "loss": 2.0696,
      "step": 1510
    },
    {
      "epoch": 0.1786343871195205,
      "grad_norm": 12.989263534545898,
      "learning_rate": 4.7209659023872266e-05,
      "loss": 2.4416,
      "step": 1520
    },
    {
      "epoch": 0.17980961335057,
      "grad_norm": 13.40894889831543,
      "learning_rate": 4.71899948873245e-05,
      "loss": 1.9779,
      "step": 1530
    },
    {
      "epoch": 0.18098483958161946,
      "grad_norm": 18.52610969543457,
      "learning_rate": 4.717033075077674e-05,
      "loss": 2.2469,
      "step": 1540
    },
    {
      "epoch": 0.18216006581266894,
      "grad_norm": 7.203619956970215,
      "learning_rate": 4.7150666614228973e-05,
      "loss": 2.0088,
      "step": 1550
    },
    {
      "epoch": 0.18333529204371843,
      "grad_norm": 21.765207290649414,
      "learning_rate": 4.713100247768121e-05,
      "loss": 2.6393,
      "step": 1560
    },
    {
      "epoch": 0.1845105182747679,
      "grad_norm": 7.197035789489746,
      "learning_rate": 4.711133834113344e-05,
      "loss": 2.3519,
      "step": 1570
    },
    {
      "epoch": 0.18568574450581737,
      "grad_norm": 8.563071250915527,
      "learning_rate": 4.709167420458568e-05,
      "loss": 2.4106,
      "step": 1580
    },
    {
      "epoch": 0.18686097073686686,
      "grad_norm": 7.1824235916137695,
      "learning_rate": 4.7072010068037915e-05,
      "loss": 1.9087,
      "step": 1590
    },
    {
      "epoch": 0.18803619696791632,
      "grad_norm": 8.8600492477417,
      "learning_rate": 4.705234593149015e-05,
      "loss": 2.3407,
      "step": 1600
    },
    {
      "epoch": 0.1892114231989658,
      "grad_norm": 8.733967781066895,
      "learning_rate": 4.703268179494238e-05,
      "loss": 2.1223,
      "step": 1610
    },
    {
      "epoch": 0.19038664943001526,
      "grad_norm": 9.305221557617188,
      "learning_rate": 4.701301765839462e-05,
      "loss": 2.2005,
      "step": 1620
    },
    {
      "epoch": 0.19156187566106475,
      "grad_norm": 16.524259567260742,
      "learning_rate": 4.6993353521846856e-05,
      "loss": 1.8482,
      "step": 1630
    },
    {
      "epoch": 0.19273710189211424,
      "grad_norm": 13.297372817993164,
      "learning_rate": 4.697368938529909e-05,
      "loss": 2.8593,
      "step": 1640
    },
    {
      "epoch": 0.1939123281231637,
      "grad_norm": 12.369548797607422,
      "learning_rate": 4.695402524875133e-05,
      "loss": 2.452,
      "step": 1650
    },
    {
      "epoch": 0.19508755435421318,
      "grad_norm": 15.604462623596191,
      "learning_rate": 4.693436111220357e-05,
      "loss": 2.7148,
      "step": 1660
    },
    {
      "epoch": 0.19626278058526267,
      "grad_norm": 21.474924087524414,
      "learning_rate": 4.6914696975655804e-05,
      "loss": 2.1127,
      "step": 1670
    },
    {
      "epoch": 0.19743800681631213,
      "grad_norm": 8.564940452575684,
      "learning_rate": 4.689503283910804e-05,
      "loss": 1.86,
      "step": 1680
    },
    {
      "epoch": 0.19861323304736161,
      "grad_norm": 7.952285289764404,
      "learning_rate": 4.687536870256027e-05,
      "loss": 2.3261,
      "step": 1690
    },
    {
      "epoch": 0.1997884592784111,
      "grad_norm": 8.620429039001465,
      "learning_rate": 4.685570456601251e-05,
      "loss": 1.9218,
      "step": 1700
    },
    {
      "epoch": 0.20096368550946056,
      "grad_norm": 10.83399486541748,
      "learning_rate": 4.6836040429464745e-05,
      "loss": 2.0312,
      "step": 1710
    },
    {
      "epoch": 0.20213891174051005,
      "grad_norm": 12.055916786193848,
      "learning_rate": 4.681637629291698e-05,
      "loss": 2.0827,
      "step": 1720
    },
    {
      "epoch": 0.20331413797155953,
      "grad_norm": 18.215389251708984,
      "learning_rate": 4.679671215636921e-05,
      "loss": 1.8414,
      "step": 1730
    },
    {
      "epoch": 0.204489364202609,
      "grad_norm": 12.821799278259277,
      "learning_rate": 4.677704801982145e-05,
      "loss": 2.4041,
      "step": 1740
    },
    {
      "epoch": 0.20566459043365848,
      "grad_norm": 12.869539260864258,
      "learning_rate": 4.6757383883273686e-05,
      "loss": 2.3793,
      "step": 1750
    },
    {
      "epoch": 0.20683981666470797,
      "grad_norm": 14.654969215393066,
      "learning_rate": 4.673771974672592e-05,
      "loss": 2.5125,
      "step": 1760
    },
    {
      "epoch": 0.20801504289575742,
      "grad_norm": 7.43568754196167,
      "learning_rate": 4.671805561017816e-05,
      "loss": 1.9297,
      "step": 1770
    },
    {
      "epoch": 0.2091902691268069,
      "grad_norm": 10.371169090270996,
      "learning_rate": 4.6698391473630394e-05,
      "loss": 2.3237,
      "step": 1780
    },
    {
      "epoch": 0.2103654953578564,
      "grad_norm": 13.4314546585083,
      "learning_rate": 4.667872733708263e-05,
      "loss": 2.3069,
      "step": 1790
    },
    {
      "epoch": 0.21154072158890586,
      "grad_norm": 8.68255615234375,
      "learning_rate": 4.665906320053487e-05,
      "loss": 1.8557,
      "step": 1800
    },
    {
      "epoch": 0.21271594781995534,
      "grad_norm": 9.930658340454102,
      "learning_rate": 4.66393990639871e-05,
      "loss": 2.5925,
      "step": 1810
    },
    {
      "epoch": 0.21389117405100483,
      "grad_norm": 10.779340744018555,
      "learning_rate": 4.661973492743934e-05,
      "loss": 2.2831,
      "step": 1820
    },
    {
      "epoch": 0.2150664002820543,
      "grad_norm": 13.439559936523438,
      "learning_rate": 4.6600070790891576e-05,
      "loss": 2.562,
      "step": 1830
    },
    {
      "epoch": 0.21624162651310377,
      "grad_norm": 9.177550315856934,
      "learning_rate": 4.658040665434381e-05,
      "loss": 2.1731,
      "step": 1840
    },
    {
      "epoch": 0.21741685274415326,
      "grad_norm": 15.04172420501709,
      "learning_rate": 4.656074251779605e-05,
      "loss": 2.4359,
      "step": 1850
    },
    {
      "epoch": 0.21859207897520272,
      "grad_norm": 13.249862670898438,
      "learning_rate": 4.654107838124828e-05,
      "loss": 2.1302,
      "step": 1860
    },
    {
      "epoch": 0.2197673052062522,
      "grad_norm": 18.084012985229492,
      "learning_rate": 4.652141424470052e-05,
      "loss": 2.324,
      "step": 1870
    },
    {
      "epoch": 0.2209425314373017,
      "grad_norm": 12.341134071350098,
      "learning_rate": 4.650175010815275e-05,
      "loss": 2.4515,
      "step": 1880
    },
    {
      "epoch": 0.22211775766835115,
      "grad_norm": 13.925853729248047,
      "learning_rate": 4.648208597160499e-05,
      "loss": 1.7898,
      "step": 1890
    },
    {
      "epoch": 0.22329298389940064,
      "grad_norm": 13.357644081115723,
      "learning_rate": 4.6462421835057224e-05,
      "loss": 2.0782,
      "step": 1900
    },
    {
      "epoch": 0.22446821013045012,
      "grad_norm": 15.981082916259766,
      "learning_rate": 4.644275769850946e-05,
      "loss": 2.4408,
      "step": 1910
    },
    {
      "epoch": 0.22564343636149958,
      "grad_norm": 12.21264934539795,
      "learning_rate": 4.642309356196169e-05,
      "loss": 2.6853,
      "step": 1920
    },
    {
      "epoch": 0.22681866259254907,
      "grad_norm": 14.55837345123291,
      "learning_rate": 4.640342942541393e-05,
      "loss": 2.4508,
      "step": 1930
    },
    {
      "epoch": 0.22799388882359856,
      "grad_norm": 18.830799102783203,
      "learning_rate": 4.6383765288866166e-05,
      "loss": 2.0275,
      "step": 1940
    },
    {
      "epoch": 0.22916911505464801,
      "grad_norm": 9.989890098571777,
      "learning_rate": 4.6364101152318406e-05,
      "loss": 2.1036,
      "step": 1950
    },
    {
      "epoch": 0.2303443412856975,
      "grad_norm": 12.995017051696777,
      "learning_rate": 4.634443701577064e-05,
      "loss": 2.3391,
      "step": 1960
    },
    {
      "epoch": 0.23151956751674696,
      "grad_norm": 5.836287021636963,
      "learning_rate": 4.632477287922288e-05,
      "loss": 2.0074,
      "step": 1970
    },
    {
      "epoch": 0.23269479374779645,
      "grad_norm": 12.490884780883789,
      "learning_rate": 4.6305108742675114e-05,
      "loss": 2.3604,
      "step": 1980
    },
    {
      "epoch": 0.23387001997884593,
      "grad_norm": 13.466679573059082,
      "learning_rate": 4.628544460612735e-05,
      "loss": 2.3894,
      "step": 1990
    },
    {
      "epoch": 0.2350452462098954,
      "grad_norm": 12.20577335357666,
      "learning_rate": 4.626578046957958e-05,
      "loss": 1.6581,
      "step": 2000
    },
    {
      "epoch": 0.23622047244094488,
      "grad_norm": 13.342066764831543,
      "learning_rate": 4.624611633303182e-05,
      "loss": 2.5037,
      "step": 2010
    },
    {
      "epoch": 0.23739569867199437,
      "grad_norm": 11.58968734741211,
      "learning_rate": 4.6226452196484055e-05,
      "loss": 2.5089,
      "step": 2020
    },
    {
      "epoch": 0.23857092490304382,
      "grad_norm": 10.412965774536133,
      "learning_rate": 4.620678805993629e-05,
      "loss": 1.8504,
      "step": 2030
    },
    {
      "epoch": 0.2397461511340933,
      "grad_norm": 12.951904296875,
      "learning_rate": 4.618712392338853e-05,
      "loss": 2.1271,
      "step": 2040
    },
    {
      "epoch": 0.2409213773651428,
      "grad_norm": 16.987974166870117,
      "learning_rate": 4.616745978684076e-05,
      "loss": 2.3715,
      "step": 2050
    },
    {
      "epoch": 0.24209660359619226,
      "grad_norm": 9.956963539123535,
      "learning_rate": 4.6147795650292996e-05,
      "loss": 2.2986,
      "step": 2060
    },
    {
      "epoch": 0.24327182982724174,
      "grad_norm": 15.570144653320312,
      "learning_rate": 4.612813151374523e-05,
      "loss": 1.9594,
      "step": 2070
    },
    {
      "epoch": 0.24444705605829123,
      "grad_norm": 10.939438819885254,
      "learning_rate": 4.610846737719747e-05,
      "loss": 2.2034,
      "step": 2080
    },
    {
      "epoch": 0.2456222822893407,
      "grad_norm": 11.594959259033203,
      "learning_rate": 4.6088803240649704e-05,
      "loss": 1.8871,
      "step": 2090
    },
    {
      "epoch": 0.24679750852039017,
      "grad_norm": 10.001277923583984,
      "learning_rate": 4.6069139104101944e-05,
      "loss": 2.272,
      "step": 2100
    },
    {
      "epoch": 0.24797273475143966,
      "grad_norm": 12.721929550170898,
      "learning_rate": 4.604947496755418e-05,
      "loss": 2.6568,
      "step": 2110
    },
    {
      "epoch": 0.24914796098248912,
      "grad_norm": 9.350130081176758,
      "learning_rate": 4.602981083100641e-05,
      "loss": 2.1066,
      "step": 2120
    },
    {
      "epoch": 0.25032318721353863,
      "grad_norm": 14.498701095581055,
      "learning_rate": 4.601014669445865e-05,
      "loss": 2.6651,
      "step": 2130
    },
    {
      "epoch": 0.25149841344458806,
      "grad_norm": 20.394874572753906,
      "learning_rate": 4.5990482557910885e-05,
      "loss": 1.7377,
      "step": 2140
    },
    {
      "epoch": 0.25267363967563755,
      "grad_norm": 11.835417747497559,
      "learning_rate": 4.597081842136312e-05,
      "loss": 1.8461,
      "step": 2150
    },
    {
      "epoch": 0.25384886590668704,
      "grad_norm": 28.0660457611084,
      "learning_rate": 4.595115428481536e-05,
      "loss": 2.3513,
      "step": 2160
    },
    {
      "epoch": 0.2550240921377365,
      "grad_norm": 9.940584182739258,
      "learning_rate": 4.593149014826759e-05,
      "loss": 2.0292,
      "step": 2170
    },
    {
      "epoch": 0.256199318368786,
      "grad_norm": 11.010923385620117,
      "learning_rate": 4.5911826011719827e-05,
      "loss": 2.0836,
      "step": 2180
    },
    {
      "epoch": 0.25737454459983544,
      "grad_norm": 8.637474060058594,
      "learning_rate": 4.589216187517206e-05,
      "loss": 2.07,
      "step": 2190
    },
    {
      "epoch": 0.25854977083088493,
      "grad_norm": 7.732728481292725,
      "learning_rate": 4.58724977386243e-05,
      "loss": 2.2884,
      "step": 2200
    },
    {
      "epoch": 0.2597249970619344,
      "grad_norm": 7.300166130065918,
      "learning_rate": 4.5852833602076534e-05,
      "loss": 2.3463,
      "step": 2210
    },
    {
      "epoch": 0.2609002232929839,
      "grad_norm": 10.405685424804688,
      "learning_rate": 4.583316946552877e-05,
      "loss": 2.1359,
      "step": 2220
    },
    {
      "epoch": 0.2620754495240334,
      "grad_norm": 10.574128150939941,
      "learning_rate": 4.5813505328981e-05,
      "loss": 2.2953,
      "step": 2230
    },
    {
      "epoch": 0.2632506757550829,
      "grad_norm": 14.738314628601074,
      "learning_rate": 4.579384119243324e-05,
      "loss": 2.0022,
      "step": 2240
    },
    {
      "epoch": 0.2644259019861323,
      "grad_norm": 13.995366096496582,
      "learning_rate": 4.577417705588548e-05,
      "loss": 1.9343,
      "step": 2250
    },
    {
      "epoch": 0.2656011282171818,
      "grad_norm": 18.527624130249023,
      "learning_rate": 4.5754512919337716e-05,
      "loss": 2.2848,
      "step": 2260
    },
    {
      "epoch": 0.2667763544482313,
      "grad_norm": 7.673961162567139,
      "learning_rate": 4.573484878278995e-05,
      "loss": 1.6406,
      "step": 2270
    },
    {
      "epoch": 0.26795158067928077,
      "grad_norm": 14.271183013916016,
      "learning_rate": 4.571518464624219e-05,
      "loss": 1.8892,
      "step": 2280
    },
    {
      "epoch": 0.26912680691033025,
      "grad_norm": 6.555535316467285,
      "learning_rate": 4.5695520509694423e-05,
      "loss": 1.5719,
      "step": 2290
    },
    {
      "epoch": 0.27030203314137974,
      "grad_norm": 11.82761001586914,
      "learning_rate": 4.567585637314666e-05,
      "loss": 2.0124,
      "step": 2300
    },
    {
      "epoch": 0.27147725937242917,
      "grad_norm": 10.424272537231445,
      "learning_rate": 4.565619223659889e-05,
      "loss": 2.3354,
      "step": 2310
    },
    {
      "epoch": 0.27265248560347866,
      "grad_norm": 12.277241706848145,
      "learning_rate": 4.563652810005113e-05,
      "loss": 2.1821,
      "step": 2320
    },
    {
      "epoch": 0.27382771183452814,
      "grad_norm": 24.27671241760254,
      "learning_rate": 4.5616863963503365e-05,
      "loss": 2.0108,
      "step": 2330
    },
    {
      "epoch": 0.27500293806557763,
      "grad_norm": 29.57710838317871,
      "learning_rate": 4.55971998269556e-05,
      "loss": 2.1742,
      "step": 2340
    },
    {
      "epoch": 0.2761781642966271,
      "grad_norm": 18.82907485961914,
      "learning_rate": 4.557753569040784e-05,
      "loss": 2.2859,
      "step": 2350
    },
    {
      "epoch": 0.2773533905276766,
      "grad_norm": 8.655754089355469,
      "learning_rate": 4.555787155386007e-05,
      "loss": 2.2726,
      "step": 2360
    },
    {
      "epoch": 0.27852861675872603,
      "grad_norm": 13.961230278015137,
      "learning_rate": 4.5538207417312306e-05,
      "loss": 1.9066,
      "step": 2370
    },
    {
      "epoch": 0.2797038429897755,
      "grad_norm": 15.827817916870117,
      "learning_rate": 4.551854328076454e-05,
      "loss": 2.2725,
      "step": 2380
    },
    {
      "epoch": 0.280879069220825,
      "grad_norm": 9.861345291137695,
      "learning_rate": 4.549887914421678e-05,
      "loss": 2.1223,
      "step": 2390
    },
    {
      "epoch": 0.2820542954518745,
      "grad_norm": 6.375272274017334,
      "learning_rate": 4.5479215007669014e-05,
      "loss": 1.6095,
      "step": 2400
    },
    {
      "epoch": 0.283229521682924,
      "grad_norm": 12.90380573272705,
      "learning_rate": 4.5459550871121254e-05,
      "loss": 2.6513,
      "step": 2410
    },
    {
      "epoch": 0.28440474791397347,
      "grad_norm": 14.938035011291504,
      "learning_rate": 4.543988673457349e-05,
      "loss": 2.4532,
      "step": 2420
    },
    {
      "epoch": 0.2855799741450229,
      "grad_norm": 13.562979698181152,
      "learning_rate": 4.542022259802573e-05,
      "loss": 1.7937,
      "step": 2430
    },
    {
      "epoch": 0.2867552003760724,
      "grad_norm": 14.37785530090332,
      "learning_rate": 4.540055846147796e-05,
      "loss": 2.0081,
      "step": 2440
    },
    {
      "epoch": 0.28793042660712187,
      "grad_norm": 8.769142150878906,
      "learning_rate": 4.5380894324930195e-05,
      "loss": 1.9866,
      "step": 2450
    },
    {
      "epoch": 0.28910565283817136,
      "grad_norm": 16.50182342529297,
      "learning_rate": 4.536123018838243e-05,
      "loss": 2.0775,
      "step": 2460
    },
    {
      "epoch": 0.29028087906922084,
      "grad_norm": 10.107303619384766,
      "learning_rate": 4.534156605183467e-05,
      "loss": 1.8174,
      "step": 2470
    },
    {
      "epoch": 0.29145610530027033,
      "grad_norm": 14.553534507751465,
      "learning_rate": 4.53219019152869e-05,
      "loss": 1.986,
      "step": 2480
    },
    {
      "epoch": 0.29263133153131976,
      "grad_norm": 16.569091796875,
      "learning_rate": 4.5302237778739136e-05,
      "loss": 1.5776,
      "step": 2490
    },
    {
      "epoch": 0.29380655776236925,
      "grad_norm": 11.1905517578125,
      "learning_rate": 4.528257364219137e-05,
      "loss": 2.1366,
      "step": 2500
    },
    {
      "epoch": 0.29498178399341873,
      "grad_norm": 15.324417114257812,
      "learning_rate": 4.526290950564361e-05,
      "loss": 2.4093,
      "step": 2510
    },
    {
      "epoch": 0.2961570102244682,
      "grad_norm": 10.297943115234375,
      "learning_rate": 4.5243245369095844e-05,
      "loss": 2.7478,
      "step": 2520
    },
    {
      "epoch": 0.2973322364555177,
      "grad_norm": 13.996519088745117,
      "learning_rate": 4.522358123254808e-05,
      "loss": 2.0905,
      "step": 2530
    },
    {
      "epoch": 0.29850746268656714,
      "grad_norm": 22.422443389892578,
      "learning_rate": 4.520391709600031e-05,
      "loss": 2.0118,
      "step": 2540
    },
    {
      "epoch": 0.2996826889176166,
      "grad_norm": 12.758740425109863,
      "learning_rate": 4.518425295945255e-05,
      "loss": 2.4585,
      "step": 2550
    },
    {
      "epoch": 0.3008579151486661,
      "grad_norm": 10.88376235961914,
      "learning_rate": 4.516458882290479e-05,
      "loss": 2.1597,
      "step": 2560
    },
    {
      "epoch": 0.3020331413797156,
      "grad_norm": 18.371503829956055,
      "learning_rate": 4.5144924686357026e-05,
      "loss": 1.8498,
      "step": 2570
    },
    {
      "epoch": 0.3032083676107651,
      "grad_norm": 10.518600463867188,
      "learning_rate": 4.512526054980926e-05,
      "loss": 1.9983,
      "step": 2580
    },
    {
      "epoch": 0.30438359384181457,
      "grad_norm": 10.76690673828125,
      "learning_rate": 4.51055964132615e-05,
      "loss": 2.0534,
      "step": 2590
    },
    {
      "epoch": 0.305558820072864,
      "grad_norm": 14.284499168395996,
      "learning_rate": 4.508593227671373e-05,
      "loss": 2.0009,
      "step": 2600
    },
    {
      "epoch": 0.3067340463039135,
      "grad_norm": 17.21082878112793,
      "learning_rate": 4.506626814016597e-05,
      "loss": 2.0008,
      "step": 2610
    },
    {
      "epoch": 0.307909272534963,
      "grad_norm": 8.41026782989502,
      "learning_rate": 4.50466040036182e-05,
      "loss": 1.8474,
      "step": 2620
    },
    {
      "epoch": 0.30908449876601246,
      "grad_norm": 11.582876205444336,
      "learning_rate": 4.502693986707044e-05,
      "loss": 2.2205,
      "step": 2630
    },
    {
      "epoch": 0.31025972499706195,
      "grad_norm": 12.89295482635498,
      "learning_rate": 4.5007275730522674e-05,
      "loss": 2.1339,
      "step": 2640
    },
    {
      "epoch": 0.31143495122811143,
      "grad_norm": 10.507551193237305,
      "learning_rate": 4.498761159397491e-05,
      "loss": 2.1003,
      "step": 2650
    },
    {
      "epoch": 0.31261017745916087,
      "grad_norm": 14.874002456665039,
      "learning_rate": 4.496794745742715e-05,
      "loss": 1.6884,
      "step": 2660
    },
    {
      "epoch": 0.31378540369021035,
      "grad_norm": 15.139491081237793,
      "learning_rate": 4.494828332087938e-05,
      "loss": 2.0319,
      "step": 2670
    },
    {
      "epoch": 0.31496062992125984,
      "grad_norm": 11.448942184448242,
      "learning_rate": 4.4928619184331616e-05,
      "loss": 1.7911,
      "step": 2680
    },
    {
      "epoch": 0.3161358561523093,
      "grad_norm": 16.786968231201172,
      "learning_rate": 4.490895504778385e-05,
      "loss": 2.3414,
      "step": 2690
    },
    {
      "epoch": 0.3173110823833588,
      "grad_norm": 12.124197959899902,
      "learning_rate": 4.488929091123609e-05,
      "loss": 2.0604,
      "step": 2700
    },
    {
      "epoch": 0.3184863086144083,
      "grad_norm": 7.912026882171631,
      "learning_rate": 4.486962677468833e-05,
      "loss": 1.5075,
      "step": 2710
    },
    {
      "epoch": 0.31966153484545773,
      "grad_norm": 10.787160873413086,
      "learning_rate": 4.4849962638140564e-05,
      "loss": 2.3545,
      "step": 2720
    },
    {
      "epoch": 0.3208367610765072,
      "grad_norm": 14.095837593078613,
      "learning_rate": 4.48302985015928e-05,
      "loss": 2.5108,
      "step": 2730
    },
    {
      "epoch": 0.3220119873075567,
      "grad_norm": 12.881146430969238,
      "learning_rate": 4.481063436504504e-05,
      "loss": 2.3034,
      "step": 2740
    },
    {
      "epoch": 0.3231872135386062,
      "grad_norm": 11.281492233276367,
      "learning_rate": 4.479097022849727e-05,
      "loss": 2.5467,
      "step": 2750
    },
    {
      "epoch": 0.3243624397696557,
      "grad_norm": 10.781302452087402,
      "learning_rate": 4.4771306091949505e-05,
      "loss": 2.1866,
      "step": 2760
    },
    {
      "epoch": 0.32553766600070516,
      "grad_norm": 20.17291259765625,
      "learning_rate": 4.475164195540174e-05,
      "loss": 1.6799,
      "step": 2770
    },
    {
      "epoch": 0.3267128922317546,
      "grad_norm": 13.338380813598633,
      "learning_rate": 4.473197781885398e-05,
      "loss": 2.0871,
      "step": 2780
    },
    {
      "epoch": 0.3278881184628041,
      "grad_norm": 13.675209045410156,
      "learning_rate": 4.471231368230621e-05,
      "loss": 1.4583,
      "step": 2790
    },
    {
      "epoch": 0.32906334469385357,
      "grad_norm": 9.432843208312988,
      "learning_rate": 4.4692649545758446e-05,
      "loss": 2.0212,
      "step": 2800
    },
    {
      "epoch": 0.33023857092490305,
      "grad_norm": 11.289777755737305,
      "learning_rate": 4.467298540921068e-05,
      "loss": 2.0874,
      "step": 2810
    },
    {
      "epoch": 0.33141379715595254,
      "grad_norm": 18.68067169189453,
      "learning_rate": 4.465332127266292e-05,
      "loss": 2.0698,
      "step": 2820
    },
    {
      "epoch": 0.332589023387002,
      "grad_norm": 11.250036239624023,
      "learning_rate": 4.4633657136115154e-05,
      "loss": 2.0967,
      "step": 2830
    },
    {
      "epoch": 0.33376424961805146,
      "grad_norm": 16.076000213623047,
      "learning_rate": 4.461399299956739e-05,
      "loss": 1.9948,
      "step": 2840
    },
    {
      "epoch": 0.33493947584910094,
      "grad_norm": 9.77022933959961,
      "learning_rate": 4.459432886301962e-05,
      "loss": 2.2048,
      "step": 2850
    },
    {
      "epoch": 0.33611470208015043,
      "grad_norm": 17.92767333984375,
      "learning_rate": 4.457466472647187e-05,
      "loss": 2.2382,
      "step": 2860
    },
    {
      "epoch": 0.3372899283111999,
      "grad_norm": 14.65797233581543,
      "learning_rate": 4.45550005899241e-05,
      "loss": 1.5517,
      "step": 2870
    },
    {
      "epoch": 0.3384651545422494,
      "grad_norm": 12.269695281982422,
      "learning_rate": 4.4535336453376335e-05,
      "loss": 2.0276,
      "step": 2880
    },
    {
      "epoch": 0.33964038077329883,
      "grad_norm": 18.651845932006836,
      "learning_rate": 4.451567231682857e-05,
      "loss": 2.0256,
      "step": 2890
    },
    {
      "epoch": 0.3408156070043483,
      "grad_norm": 14.476424217224121,
      "learning_rate": 4.449600818028081e-05,
      "loss": 2.3591,
      "step": 2900
    },
    {
      "epoch": 0.3419908332353978,
      "grad_norm": 14.62807846069336,
      "learning_rate": 4.447634404373304e-05,
      "loss": 2.0881,
      "step": 2910
    },
    {
      "epoch": 0.3431660594664473,
      "grad_norm": 16.06639862060547,
      "learning_rate": 4.4456679907185277e-05,
      "loss": 2.039,
      "step": 2920
    },
    {
      "epoch": 0.3443412856974968,
      "grad_norm": 9.91940975189209,
      "learning_rate": 4.443701577063751e-05,
      "loss": 2.2571,
      "step": 2930
    },
    {
      "epoch": 0.34551651192854627,
      "grad_norm": 17.437978744506836,
      "learning_rate": 4.441735163408975e-05,
      "loss": 2.4414,
      "step": 2940
    },
    {
      "epoch": 0.3466917381595957,
      "grad_norm": 11.554434776306152,
      "learning_rate": 4.4397687497541984e-05,
      "loss": 1.8624,
      "step": 2950
    },
    {
      "epoch": 0.3478669643906452,
      "grad_norm": 19.4925537109375,
      "learning_rate": 4.437802336099422e-05,
      "loss": 2.3618,
      "step": 2960
    },
    {
      "epoch": 0.34904219062169467,
      "grad_norm": 13.750736236572266,
      "learning_rate": 4.435835922444646e-05,
      "loss": 1.8597,
      "step": 2970
    },
    {
      "epoch": 0.35021741685274416,
      "grad_norm": 11.494909286499023,
      "learning_rate": 4.433869508789869e-05,
      "loss": 1.8383,
      "step": 2980
    },
    {
      "epoch": 0.35139264308379364,
      "grad_norm": 14.424116134643555,
      "learning_rate": 4.4319030951350925e-05,
      "loss": 1.8661,
      "step": 2990
    },
    {
      "epoch": 0.35256786931484313,
      "grad_norm": 9.345441818237305,
      "learning_rate": 4.429936681480316e-05,
      "loss": 2.6657,
      "step": 3000
    },
    {
      "epoch": 0.35374309554589256,
      "grad_norm": 10.68867301940918,
      "learning_rate": 4.42797026782554e-05,
      "loss": 2.2455,
      "step": 3010
    },
    {
      "epoch": 0.35491832177694205,
      "grad_norm": 19.620765686035156,
      "learning_rate": 4.426003854170764e-05,
      "loss": 2.6182,
      "step": 3020
    },
    {
      "epoch": 0.35609354800799153,
      "grad_norm": 17.44060516357422,
      "learning_rate": 4.4240374405159873e-05,
      "loss": 2.0774,
      "step": 3030
    },
    {
      "epoch": 0.357268774239041,
      "grad_norm": 14.881667137145996,
      "learning_rate": 4.422071026861211e-05,
      "loss": 2.4547,
      "step": 3040
    },
    {
      "epoch": 0.3584440004700905,
      "grad_norm": 7.451622486114502,
      "learning_rate": 4.420104613206435e-05,
      "loss": 2.4684,
      "step": 3050
    },
    {
      "epoch": 0.35961922670114,
      "grad_norm": 15.097880363464355,
      "learning_rate": 4.418138199551658e-05,
      "loss": 1.8218,
      "step": 3060
    },
    {
      "epoch": 0.3607944529321894,
      "grad_norm": 10.629120826721191,
      "learning_rate": 4.4161717858968815e-05,
      "loss": 2.0352,
      "step": 3070
    },
    {
      "epoch": 0.3619696791632389,
      "grad_norm": 4.64254093170166,
      "learning_rate": 4.414205372242105e-05,
      "loss": 2.2133,
      "step": 3080
    },
    {
      "epoch": 0.3631449053942884,
      "grad_norm": 9.514970779418945,
      "learning_rate": 4.412238958587329e-05,
      "loss": 2.1009,
      "step": 3090
    },
    {
      "epoch": 0.3643201316253379,
      "grad_norm": 30.130203247070312,
      "learning_rate": 4.410272544932552e-05,
      "loss": 2.8949,
      "step": 3100
    },
    {
      "epoch": 0.36549535785638737,
      "grad_norm": 10.335441589355469,
      "learning_rate": 4.4083061312777756e-05,
      "loss": 2.0667,
      "step": 3110
    },
    {
      "epoch": 0.36667058408743686,
      "grad_norm": 14.504185676574707,
      "learning_rate": 4.406339717622999e-05,
      "loss": 2.3184,
      "step": 3120
    },
    {
      "epoch": 0.3678458103184863,
      "grad_norm": 15.151309967041016,
      "learning_rate": 4.404373303968223e-05,
      "loss": 2.2048,
      "step": 3130
    },
    {
      "epoch": 0.3690210365495358,
      "grad_norm": 6.774374961853027,
      "learning_rate": 4.4024068903134464e-05,
      "loss": 2.261,
      "step": 3140
    },
    {
      "epoch": 0.37019626278058526,
      "grad_norm": 14.650129318237305,
      "learning_rate": 4.40044047665867e-05,
      "loss": 2.0387,
      "step": 3150
    },
    {
      "epoch": 0.37137148901163475,
      "grad_norm": 15.160332679748535,
      "learning_rate": 4.398474063003894e-05,
      "loss": 2.5395,
      "step": 3160
    },
    {
      "epoch": 0.37254671524268423,
      "grad_norm": 14.537505149841309,
      "learning_rate": 4.396507649349118e-05,
      "loss": 2.1946,
      "step": 3170
    },
    {
      "epoch": 0.3737219414737337,
      "grad_norm": 10.083491325378418,
      "learning_rate": 4.394541235694341e-05,
      "loss": 2.2304,
      "step": 3180
    },
    {
      "epoch": 0.37489716770478315,
      "grad_norm": 10.047405242919922,
      "learning_rate": 4.3925748220395645e-05,
      "loss": 2.2654,
      "step": 3190
    },
    {
      "epoch": 0.37607239393583264,
      "grad_norm": 11.017961502075195,
      "learning_rate": 4.390608408384788e-05,
      "loss": 2.2902,
      "step": 3200
    },
    {
      "epoch": 0.3772476201668821,
      "grad_norm": 10.062576293945312,
      "learning_rate": 4.388641994730012e-05,
      "loss": 2.0575,
      "step": 3210
    },
    {
      "epoch": 0.3784228463979316,
      "grad_norm": 10.382721900939941,
      "learning_rate": 4.386675581075235e-05,
      "loss": 1.7652,
      "step": 3220
    },
    {
      "epoch": 0.3795980726289811,
      "grad_norm": 8.329229354858398,
      "learning_rate": 4.3847091674204586e-05,
      "loss": 2.6148,
      "step": 3230
    },
    {
      "epoch": 0.38077329886003053,
      "grad_norm": 12.023482322692871,
      "learning_rate": 4.382742753765682e-05,
      "loss": 2.0443,
      "step": 3240
    },
    {
      "epoch": 0.38194852509108,
      "grad_norm": 8.978275299072266,
      "learning_rate": 4.380776340110906e-05,
      "loss": 1.9293,
      "step": 3250
    },
    {
      "epoch": 0.3831237513221295,
      "grad_norm": 29.615039825439453,
      "learning_rate": 4.3788099264561294e-05,
      "loss": 2.1497,
      "step": 3260
    },
    {
      "epoch": 0.384298977553179,
      "grad_norm": 14.481039047241211,
      "learning_rate": 4.376843512801353e-05,
      "loss": 1.4673,
      "step": 3270
    },
    {
      "epoch": 0.3854742037842285,
      "grad_norm": 11.902708053588867,
      "learning_rate": 4.374877099146577e-05,
      "loss": 1.84,
      "step": 3280
    },
    {
      "epoch": 0.38664943001527796,
      "grad_norm": 8.38027286529541,
      "learning_rate": 4.3729106854918e-05,
      "loss": 2.2113,
      "step": 3290
    },
    {
      "epoch": 0.3878246562463274,
      "grad_norm": 12.43402099609375,
      "learning_rate": 4.3709442718370235e-05,
      "loss": 1.8384,
      "step": 3300
    },
    {
      "epoch": 0.3889998824773769,
      "grad_norm": 18.369665145874023,
      "learning_rate": 4.3689778581822476e-05,
      "loss": 1.8694,
      "step": 3310
    },
    {
      "epoch": 0.39017510870842637,
      "grad_norm": 14.281049728393555,
      "learning_rate": 4.367011444527471e-05,
      "loss": 2.0992,
      "step": 3320
    },
    {
      "epoch": 0.39135033493947585,
      "grad_norm": 13.209550857543945,
      "learning_rate": 4.365045030872695e-05,
      "loss": 1.8129,
      "step": 3330
    },
    {
      "epoch": 0.39252556117052534,
      "grad_norm": 12.920531272888184,
      "learning_rate": 4.363078617217918e-05,
      "loss": 2.2309,
      "step": 3340
    },
    {
      "epoch": 0.3937007874015748,
      "grad_norm": 10.804939270019531,
      "learning_rate": 4.361112203563142e-05,
      "loss": 2.2426,
      "step": 3350
    },
    {
      "epoch": 0.39487601363262426,
      "grad_norm": 19.461454391479492,
      "learning_rate": 4.359145789908366e-05,
      "loss": 2.4012,
      "step": 3360
    },
    {
      "epoch": 0.39605123986367374,
      "grad_norm": 9.866844177246094,
      "learning_rate": 4.357179376253589e-05,
      "loss": 2.3566,
      "step": 3370
    },
    {
      "epoch": 0.39722646609472323,
      "grad_norm": 14.622405052185059,
      "learning_rate": 4.3552129625988124e-05,
      "loss": 1.6249,
      "step": 3380
    },
    {
      "epoch": 0.3984016923257727,
      "grad_norm": 24.4198055267334,
      "learning_rate": 4.353246548944036e-05,
      "loss": 1.8708,
      "step": 3390
    },
    {
      "epoch": 0.3995769185568222,
      "grad_norm": 27.92443084716797,
      "learning_rate": 4.35128013528926e-05,
      "loss": 2.2566,
      "step": 3400
    },
    {
      "epoch": 0.4007521447878717,
      "grad_norm": 14.937458038330078,
      "learning_rate": 4.349313721634483e-05,
      "loss": 1.9429,
      "step": 3410
    },
    {
      "epoch": 0.4019273710189211,
      "grad_norm": 7.759469032287598,
      "learning_rate": 4.3473473079797066e-05,
      "loss": 2.0036,
      "step": 3420
    },
    {
      "epoch": 0.4031025972499706,
      "grad_norm": 10.682683944702148,
      "learning_rate": 4.34538089432493e-05,
      "loss": 2.111,
      "step": 3430
    },
    {
      "epoch": 0.4042778234810201,
      "grad_norm": 22.561649322509766,
      "learning_rate": 4.343414480670154e-05,
      "loss": 1.8978,
      "step": 3440
    },
    {
      "epoch": 0.4054530497120696,
      "grad_norm": 8.976452827453613,
      "learning_rate": 4.341448067015377e-05,
      "loss": 1.8,
      "step": 3450
    },
    {
      "epoch": 0.40662827594311907,
      "grad_norm": 13.460932731628418,
      "learning_rate": 4.3394816533606014e-05,
      "loss": 1.8476,
      "step": 3460
    },
    {
      "epoch": 0.40780350217416855,
      "grad_norm": 7.637677192687988,
      "learning_rate": 4.337515239705825e-05,
      "loss": 2.1489,
      "step": 3470
    },
    {
      "epoch": 0.408978728405218,
      "grad_norm": 8.005024909973145,
      "learning_rate": 4.335548826051049e-05,
      "loss": 1.8835,
      "step": 3480
    },
    {
      "epoch": 0.41015395463626747,
      "grad_norm": 8.152450561523438,
      "learning_rate": 4.333582412396272e-05,
      "loss": 1.5813,
      "step": 3490
    },
    {
      "epoch": 0.41132918086731696,
      "grad_norm": 11.706449508666992,
      "learning_rate": 4.3316159987414955e-05,
      "loss": 2.2055,
      "step": 3500
    },
    {
      "epoch": 0.41250440709836644,
      "grad_norm": 8.143502235412598,
      "learning_rate": 4.329649585086719e-05,
      "loss": 1.946,
      "step": 3510
    },
    {
      "epoch": 0.41367963332941593,
      "grad_norm": 10.333789825439453,
      "learning_rate": 4.327683171431943e-05,
      "loss": 2.3976,
      "step": 3520
    },
    {
      "epoch": 0.4148548595604654,
      "grad_norm": 14.559487342834473,
      "learning_rate": 4.325716757777166e-05,
      "loss": 2.2099,
      "step": 3530
    },
    {
      "epoch": 0.41603008579151485,
      "grad_norm": 21.66111183166504,
      "learning_rate": 4.3237503441223896e-05,
      "loss": 2.7174,
      "step": 3540
    },
    {
      "epoch": 0.41720531202256433,
      "grad_norm": 26.202903747558594,
      "learning_rate": 4.3217839304676137e-05,
      "loss": 2.3694,
      "step": 3550
    },
    {
      "epoch": 0.4183805382536138,
      "grad_norm": 12.280159950256348,
      "learning_rate": 4.319817516812837e-05,
      "loss": 2.018,
      "step": 3560
    },
    {
      "epoch": 0.4195557644846633,
      "grad_norm": 12.876908302307129,
      "learning_rate": 4.3178511031580604e-05,
      "loss": 2.3413,
      "step": 3570
    },
    {
      "epoch": 0.4207309907157128,
      "grad_norm": 15.166952133178711,
      "learning_rate": 4.315884689503284e-05,
      "loss": 1.9394,
      "step": 3580
    },
    {
      "epoch": 0.4219062169467622,
      "grad_norm": 10.816182136535645,
      "learning_rate": 4.313918275848508e-05,
      "loss": 2.6222,
      "step": 3590
    },
    {
      "epoch": 0.4230814431778117,
      "grad_norm": 12.005484580993652,
      "learning_rate": 4.311951862193731e-05,
      "loss": 1.9729,
      "step": 3600
    },
    {
      "epoch": 0.4242566694088612,
      "grad_norm": 9.671234130859375,
      "learning_rate": 4.3099854485389545e-05,
      "loss": 1.3613,
      "step": 3610
    },
    {
      "epoch": 0.4254318956399107,
      "grad_norm": 15.237287521362305,
      "learning_rate": 4.3080190348841785e-05,
      "loss": 1.8509,
      "step": 3620
    },
    {
      "epoch": 0.42660712187096017,
      "grad_norm": 14.498955726623535,
      "learning_rate": 4.306052621229402e-05,
      "loss": 2.175,
      "step": 3630
    },
    {
      "epoch": 0.42778234810200966,
      "grad_norm": 13.743219375610352,
      "learning_rate": 4.304086207574626e-05,
      "loss": 1.497,
      "step": 3640
    },
    {
      "epoch": 0.4289575743330591,
      "grad_norm": 8.793191909790039,
      "learning_rate": 4.302119793919849e-05,
      "loss": 2.456,
      "step": 3650
    },
    {
      "epoch": 0.4301328005641086,
      "grad_norm": 25.491207122802734,
      "learning_rate": 4.3001533802650727e-05,
      "loss": 2.0508,
      "step": 3660
    },
    {
      "epoch": 0.43130802679515806,
      "grad_norm": 10.937836647033691,
      "learning_rate": 4.298186966610297e-05,
      "loss": 1.6672,
      "step": 3670
    },
    {
      "epoch": 0.43248325302620755,
      "grad_norm": 17.144060134887695,
      "learning_rate": 4.29622055295552e-05,
      "loss": 1.9603,
      "step": 3680
    },
    {
      "epoch": 0.43365847925725703,
      "grad_norm": 16.03167724609375,
      "learning_rate": 4.2942541393007434e-05,
      "loss": 1.8016,
      "step": 3690
    },
    {
      "epoch": 0.4348337054883065,
      "grad_norm": 9.631890296936035,
      "learning_rate": 4.292287725645967e-05,
      "loss": 1.6146,
      "step": 3700
    },
    {
      "epoch": 0.43600893171935595,
      "grad_norm": 19.041616439819336,
      "learning_rate": 4.290321311991191e-05,
      "loss": 2.1542,
      "step": 3710
    },
    {
      "epoch": 0.43718415795040544,
      "grad_norm": 11.411686897277832,
      "learning_rate": 4.288354898336414e-05,
      "loss": 2.2195,
      "step": 3720
    },
    {
      "epoch": 0.4383593841814549,
      "grad_norm": 11.324926376342773,
      "learning_rate": 4.2863884846816375e-05,
      "loss": 2.446,
      "step": 3730
    },
    {
      "epoch": 0.4395346104125044,
      "grad_norm": 12.61486530303955,
      "learning_rate": 4.284422071026861e-05,
      "loss": 2.6535,
      "step": 3740
    },
    {
      "epoch": 0.4407098366435539,
      "grad_norm": 7.921072959899902,
      "learning_rate": 4.282455657372085e-05,
      "loss": 2.3209,
      "step": 3750
    },
    {
      "epoch": 0.4418850628746034,
      "grad_norm": 5.424452304840088,
      "learning_rate": 4.280489243717308e-05,
      "loss": 1.8933,
      "step": 3760
    },
    {
      "epoch": 0.4430602891056528,
      "grad_norm": 11.358354568481445,
      "learning_rate": 4.2785228300625323e-05,
      "loss": 1.8086,
      "step": 3770
    },
    {
      "epoch": 0.4442355153367023,
      "grad_norm": 20.183950424194336,
      "learning_rate": 4.276556416407756e-05,
      "loss": 2.5959,
      "step": 3780
    },
    {
      "epoch": 0.4454107415677518,
      "grad_norm": 7.571549415588379,
      "learning_rate": 4.27459000275298e-05,
      "loss": 2.037,
      "step": 3790
    },
    {
      "epoch": 0.4465859677988013,
      "grad_norm": 13.75265121459961,
      "learning_rate": 4.272623589098203e-05,
      "loss": 1.8076,
      "step": 3800
    },
    {
      "epoch": 0.44776119402985076,
      "grad_norm": 12.459274291992188,
      "learning_rate": 4.2706571754434265e-05,
      "loss": 1.9017,
      "step": 3810
    },
    {
      "epoch": 0.44893642026090025,
      "grad_norm": 15.286564826965332,
      "learning_rate": 4.26869076178865e-05,
      "loss": 2.3778,
      "step": 3820
    },
    {
      "epoch": 0.4501116464919497,
      "grad_norm": 9.838086128234863,
      "learning_rate": 4.266724348133874e-05,
      "loss": 1.8336,
      "step": 3830
    },
    {
      "epoch": 0.45128687272299917,
      "grad_norm": 17.423757553100586,
      "learning_rate": 4.264757934479097e-05,
      "loss": 2.3053,
      "step": 3840
    },
    {
      "epoch": 0.45246209895404865,
      "grad_norm": 16.56212043762207,
      "learning_rate": 4.2627915208243206e-05,
      "loss": 2.2212,
      "step": 3850
    },
    {
      "epoch": 0.45363732518509814,
      "grad_norm": 8.529197692871094,
      "learning_rate": 4.2608251071695446e-05,
      "loss": 1.461,
      "step": 3860
    },
    {
      "epoch": 0.4548125514161476,
      "grad_norm": 16.603160858154297,
      "learning_rate": 4.258858693514768e-05,
      "loss": 2.2433,
      "step": 3870
    },
    {
      "epoch": 0.4559877776471971,
      "grad_norm": 23.680496215820312,
      "learning_rate": 4.2568922798599913e-05,
      "loss": 2.1169,
      "step": 3880
    },
    {
      "epoch": 0.45716300387824654,
      "grad_norm": 13.801122665405273,
      "learning_rate": 4.254925866205215e-05,
      "loss": 2.0748,
      "step": 3890
    },
    {
      "epoch": 0.45833823010929603,
      "grad_norm": 12.804221153259277,
      "learning_rate": 4.252959452550439e-05,
      "loss": 1.4953,
      "step": 3900
    },
    {
      "epoch": 0.4595134563403455,
      "grad_norm": 14.113533020019531,
      "learning_rate": 4.250993038895662e-05,
      "loss": 1.9623,
      "step": 3910
    },
    {
      "epoch": 0.460688682571395,
      "grad_norm": 10.386401176452637,
      "learning_rate": 4.249026625240886e-05,
      "loss": 2.133,
      "step": 3920
    },
    {
      "epoch": 0.4618639088024445,
      "grad_norm": 14.808308601379395,
      "learning_rate": 4.2470602115861095e-05,
      "loss": 2.126,
      "step": 3930
    },
    {
      "epoch": 0.4630391350334939,
      "grad_norm": 11.301342964172363,
      "learning_rate": 4.2450937979313336e-05,
      "loss": 2.0977,
      "step": 3940
    },
    {
      "epoch": 0.4642143612645434,
      "grad_norm": 31.93461036682129,
      "learning_rate": 4.243127384276557e-05,
      "loss": 2.8041,
      "step": 3950
    },
    {
      "epoch": 0.4653895874955929,
      "grad_norm": 9.82878303527832,
      "learning_rate": 4.24116097062178e-05,
      "loss": 1.9237,
      "step": 3960
    },
    {
      "epoch": 0.4665648137266424,
      "grad_norm": 22.952302932739258,
      "learning_rate": 4.2391945569670036e-05,
      "loss": 1.6247,
      "step": 3970
    },
    {
      "epoch": 0.46774003995769187,
      "grad_norm": 20.092456817626953,
      "learning_rate": 4.237228143312228e-05,
      "loss": 1.6963,
      "step": 3980
    },
    {
      "epoch": 0.46891526618874135,
      "grad_norm": 13.511238098144531,
      "learning_rate": 4.235261729657451e-05,
      "loss": 2.3019,
      "step": 3990
    },
    {
      "epoch": 0.4700904924197908,
      "grad_norm": 14.6659574508667,
      "learning_rate": 4.2332953160026744e-05,
      "loss": 1.8313,
      "step": 4000
    },
    {
      "epoch": 0.47126571865084027,
      "grad_norm": 29.815410614013672,
      "learning_rate": 4.231328902347898e-05,
      "loss": 2.0803,
      "step": 4010
    },
    {
      "epoch": 0.47244094488188976,
      "grad_norm": 17.60256576538086,
      "learning_rate": 4.229362488693122e-05,
      "loss": 2.0572,
      "step": 4020
    },
    {
      "epoch": 0.47361617111293924,
      "grad_norm": 7.25173807144165,
      "learning_rate": 4.227396075038345e-05,
      "loss": 1.6277,
      "step": 4030
    },
    {
      "epoch": 0.47479139734398873,
      "grad_norm": 9.788707733154297,
      "learning_rate": 4.2254296613835685e-05,
      "loss": 1.9751,
      "step": 4040
    },
    {
      "epoch": 0.4759666235750382,
      "grad_norm": 9.009198188781738,
      "learning_rate": 4.223463247728792e-05,
      "loss": 1.8932,
      "step": 4050
    },
    {
      "epoch": 0.47714184980608765,
      "grad_norm": 11.432634353637695,
      "learning_rate": 4.221496834074016e-05,
      "loss": 2.3385,
      "step": 4060
    },
    {
      "epoch": 0.47831707603713713,
      "grad_norm": 13.13003158569336,
      "learning_rate": 4.21953042041924e-05,
      "loss": 1.5965,
      "step": 4070
    },
    {
      "epoch": 0.4794923022681866,
      "grad_norm": 11.103805541992188,
      "learning_rate": 4.217564006764463e-05,
      "loss": 1.4054,
      "step": 4080
    },
    {
      "epoch": 0.4806675284992361,
      "grad_norm": 7.080469131469727,
      "learning_rate": 4.215597593109687e-05,
      "loss": 1.7903,
      "step": 4090
    },
    {
      "epoch": 0.4818427547302856,
      "grad_norm": 9.128670692443848,
      "learning_rate": 4.213631179454911e-05,
      "loss": 1.6322,
      "step": 4100
    },
    {
      "epoch": 0.4830179809613351,
      "grad_norm": 23.249889373779297,
      "learning_rate": 4.211664765800134e-05,
      "loss": 2.2454,
      "step": 4110
    },
    {
      "epoch": 0.4841932071923845,
      "grad_norm": 18.72980499267578,
      "learning_rate": 4.2096983521453574e-05,
      "loss": 2.7004,
      "step": 4120
    },
    {
      "epoch": 0.485368433423434,
      "grad_norm": 20.79292869567871,
      "learning_rate": 4.207731938490581e-05,
      "loss": 1.9669,
      "step": 4130
    },
    {
      "epoch": 0.4865436596544835,
      "grad_norm": 10.966432571411133,
      "learning_rate": 4.205765524835805e-05,
      "loss": 1.9055,
      "step": 4140
    },
    {
      "epoch": 0.48771888588553297,
      "grad_norm": 12.479768753051758,
      "learning_rate": 4.203799111181028e-05,
      "loss": 2.0681,
      "step": 4150
    },
    {
      "epoch": 0.48889411211658246,
      "grad_norm": 7.451715469360352,
      "learning_rate": 4.2018326975262516e-05,
      "loss": 2.1528,
      "step": 4160
    },
    {
      "epoch": 0.49006933834763194,
      "grad_norm": 36.1834831237793,
      "learning_rate": 4.1998662838714756e-05,
      "loss": 2.3168,
      "step": 4170
    },
    {
      "epoch": 0.4912445645786814,
      "grad_norm": 15.368968963623047,
      "learning_rate": 4.197899870216699e-05,
      "loss": 1.8125,
      "step": 4180
    },
    {
      "epoch": 0.49241979080973086,
      "grad_norm": 11.01513385772705,
      "learning_rate": 4.195933456561922e-05,
      "loss": 1.8633,
      "step": 4190
    },
    {
      "epoch": 0.49359501704078035,
      "grad_norm": 8.861076354980469,
      "learning_rate": 4.193967042907146e-05,
      "loss": 1.9773,
      "step": 4200
    },
    {
      "epoch": 0.49477024327182983,
      "grad_norm": 10.340984344482422,
      "learning_rate": 4.19200062925237e-05,
      "loss": 1.8655,
      "step": 4210
    },
    {
      "epoch": 0.4959454695028793,
      "grad_norm": 23.525793075561523,
      "learning_rate": 4.190034215597594e-05,
      "loss": 1.9222,
      "step": 4220
    },
    {
      "epoch": 0.4971206957339288,
      "grad_norm": 18.960613250732422,
      "learning_rate": 4.188067801942817e-05,
      "loss": 2.8072,
      "step": 4230
    },
    {
      "epoch": 0.49829592196497824,
      "grad_norm": 5.935115337371826,
      "learning_rate": 4.1861013882880405e-05,
      "loss": 1.7495,
      "step": 4240
    },
    {
      "epoch": 0.4994711481960277,
      "grad_norm": 9.175984382629395,
      "learning_rate": 4.1841349746332645e-05,
      "loss": 2.0713,
      "step": 4250
    },
    {
      "epoch": 0.5006463744270773,
      "grad_norm": 14.657028198242188,
      "learning_rate": 4.182168560978488e-05,
      "loss": 2.19,
      "step": 4260
    },
    {
      "epoch": 0.5018216006581266,
      "grad_norm": 18.148820877075195,
      "learning_rate": 4.180202147323711e-05,
      "loss": 1.688,
      "step": 4270
    },
    {
      "epoch": 0.5029968268891761,
      "grad_norm": 9.3037691116333,
      "learning_rate": 4.1782357336689346e-05,
      "loss": 2.0113,
      "step": 4280
    },
    {
      "epoch": 0.5041720531202256,
      "grad_norm": 7.911787033081055,
      "learning_rate": 4.1762693200141587e-05,
      "loss": 1.894,
      "step": 4290
    },
    {
      "epoch": 0.5053472793512751,
      "grad_norm": 19.427867889404297,
      "learning_rate": 4.174302906359382e-05,
      "loss": 1.8809,
      "step": 4300
    },
    {
      "epoch": 0.5065225055823246,
      "grad_norm": 7.948591709136963,
      "learning_rate": 4.1723364927046054e-05,
      "loss": 2.4952,
      "step": 4310
    },
    {
      "epoch": 0.5076977318133741,
      "grad_norm": 19.013072967529297,
      "learning_rate": 4.170370079049829e-05,
      "loss": 1.84,
      "step": 4320
    },
    {
      "epoch": 0.5088729580444236,
      "grad_norm": 10.020354270935059,
      "learning_rate": 4.168403665395053e-05,
      "loss": 2.1785,
      "step": 4330
    },
    {
      "epoch": 0.510048184275473,
      "grad_norm": 16.234521865844727,
      "learning_rate": 4.166437251740276e-05,
      "loss": 2.6848,
      "step": 4340
    },
    {
      "epoch": 0.5112234105065225,
      "grad_norm": 26.92982292175293,
      "learning_rate": 4.1644708380854995e-05,
      "loss": 2.0676,
      "step": 4350
    },
    {
      "epoch": 0.512398636737572,
      "grad_norm": 8.8961820602417,
      "learning_rate": 4.162504424430723e-05,
      "loss": 1.9217,
      "step": 4360
    },
    {
      "epoch": 0.5135738629686215,
      "grad_norm": 19.90564727783203,
      "learning_rate": 4.1605380107759476e-05,
      "loss": 1.8848,
      "step": 4370
    },
    {
      "epoch": 0.5147490891996709,
      "grad_norm": 10.00037956237793,
      "learning_rate": 4.158571597121171e-05,
      "loss": 2.2078,
      "step": 4380
    },
    {
      "epoch": 0.5159243154307204,
      "grad_norm": 16.554664611816406,
      "learning_rate": 4.156605183466394e-05,
      "loss": 1.5676,
      "step": 4390
    },
    {
      "epoch": 0.5170995416617699,
      "grad_norm": 9.677738189697266,
      "learning_rate": 4.1546387698116177e-05,
      "loss": 1.6343,
      "step": 4400
    },
    {
      "epoch": 0.5182747678928193,
      "grad_norm": 14.435365676879883,
      "learning_rate": 4.152672356156842e-05,
      "loss": 1.9617,
      "step": 4410
    },
    {
      "epoch": 0.5194499941238688,
      "grad_norm": 20.678543090820312,
      "learning_rate": 4.150705942502065e-05,
      "loss": 2.0349,
      "step": 4420
    },
    {
      "epoch": 0.5206252203549183,
      "grad_norm": 12.773311614990234,
      "learning_rate": 4.1487395288472884e-05,
      "loss": 2.0174,
      "step": 4430
    },
    {
      "epoch": 0.5218004465859678,
      "grad_norm": 10.031023979187012,
      "learning_rate": 4.146773115192512e-05,
      "loss": 1.5735,
      "step": 4440
    },
    {
      "epoch": 0.5229756728170173,
      "grad_norm": 10.132521629333496,
      "learning_rate": 4.144806701537736e-05,
      "loss": 2.1188,
      "step": 4450
    },
    {
      "epoch": 0.5241508990480668,
      "grad_norm": 15.685379981994629,
      "learning_rate": 4.142840287882959e-05,
      "loss": 1.9602,
      "step": 4460
    },
    {
      "epoch": 0.5253261252791163,
      "grad_norm": 13.38106918334961,
      "learning_rate": 4.1408738742281825e-05,
      "loss": 2.0992,
      "step": 4470
    },
    {
      "epoch": 0.5265013515101657,
      "grad_norm": 21.55803108215332,
      "learning_rate": 4.1389074605734066e-05,
      "loss": 2.2924,
      "step": 4480
    },
    {
      "epoch": 0.5276765777412152,
      "grad_norm": 9.96091365814209,
      "learning_rate": 4.13694104691863e-05,
      "loss": 1.8804,
      "step": 4490
    },
    {
      "epoch": 0.5288518039722646,
      "grad_norm": 16.56869888305664,
      "learning_rate": 4.134974633263853e-05,
      "loss": 2.0584,
      "step": 4500
    },
    {
      "epoch": 0.5300270302033141,
      "grad_norm": 6.626291751861572,
      "learning_rate": 4.133008219609077e-05,
      "loss": 2.035,
      "step": 4510
    },
    {
      "epoch": 0.5312022564343636,
      "grad_norm": 15.444177627563477,
      "learning_rate": 4.131041805954301e-05,
      "loss": 1.8537,
      "step": 4520
    },
    {
      "epoch": 0.5323774826654131,
      "grad_norm": 11.888521194458008,
      "learning_rate": 4.129075392299525e-05,
      "loss": 2.0271,
      "step": 4530
    },
    {
      "epoch": 0.5335527088964626,
      "grad_norm": 16.603174209594727,
      "learning_rate": 4.127108978644748e-05,
      "loss": 1.766,
      "step": 4540
    },
    {
      "epoch": 0.534727935127512,
      "grad_norm": 15.03260612487793,
      "learning_rate": 4.1251425649899715e-05,
      "loss": 1.8355,
      "step": 4550
    },
    {
      "epoch": 0.5359031613585615,
      "grad_norm": 10.825209617614746,
      "learning_rate": 4.1231761513351955e-05,
      "loss": 1.5945,
      "step": 4560
    },
    {
      "epoch": 0.537078387589611,
      "grad_norm": 21.298431396484375,
      "learning_rate": 4.121209737680419e-05,
      "loss": 2.0559,
      "step": 4570
    },
    {
      "epoch": 0.5382536138206605,
      "grad_norm": 16.333276748657227,
      "learning_rate": 4.119243324025642e-05,
      "loss": 1.935,
      "step": 4580
    },
    {
      "epoch": 0.53942884005171,
      "grad_norm": 9.99832534790039,
      "learning_rate": 4.1172769103708656e-05,
      "loss": 2.0367,
      "step": 4590
    },
    {
      "epoch": 0.5406040662827595,
      "grad_norm": 12.822352409362793,
      "learning_rate": 4.1153104967160896e-05,
      "loss": 2.4237,
      "step": 4600
    },
    {
      "epoch": 0.541779292513809,
      "grad_norm": 23.028684616088867,
      "learning_rate": 4.113344083061313e-05,
      "loss": 2.1051,
      "step": 4610
    },
    {
      "epoch": 0.5429545187448583,
      "grad_norm": 27.08348274230957,
      "learning_rate": 4.1113776694065363e-05,
      "loss": 2.3112,
      "step": 4620
    },
    {
      "epoch": 0.5441297449759078,
      "grad_norm": 14.82834243774414,
      "learning_rate": 4.10941125575176e-05,
      "loss": 1.7727,
      "step": 4630
    },
    {
      "epoch": 0.5453049712069573,
      "grad_norm": 14.505446434020996,
      "learning_rate": 4.107444842096984e-05,
      "loss": 2.1626,
      "step": 4640
    },
    {
      "epoch": 0.5464801974380068,
      "grad_norm": 17.283321380615234,
      "learning_rate": 4.105478428442207e-05,
      "loss": 2.0162,
      "step": 4650
    },
    {
      "epoch": 0.5476554236690563,
      "grad_norm": 23.50490379333496,
      "learning_rate": 4.1035120147874305e-05,
      "loss": 1.9643,
      "step": 4660
    },
    {
      "epoch": 0.5488306499001058,
      "grad_norm": 12.335837364196777,
      "learning_rate": 4.1015456011326545e-05,
      "loss": 2.2999,
      "step": 4670
    },
    {
      "epoch": 0.5500058761311553,
      "grad_norm": 10.853656768798828,
      "learning_rate": 4.0995791874778786e-05,
      "loss": 1.7362,
      "step": 4680
    },
    {
      "epoch": 0.5511811023622047,
      "grad_norm": 17.436281204223633,
      "learning_rate": 4.097612773823102e-05,
      "loss": 1.9095,
      "step": 4690
    },
    {
      "epoch": 0.5523563285932542,
      "grad_norm": 12.001564025878906,
      "learning_rate": 4.095646360168325e-05,
      "loss": 2.2028,
      "step": 4700
    },
    {
      "epoch": 0.5535315548243037,
      "grad_norm": 14.096331596374512,
      "learning_rate": 4.0936799465135486e-05,
      "loss": 1.9678,
      "step": 4710
    },
    {
      "epoch": 0.5547067810553532,
      "grad_norm": 16.666255950927734,
      "learning_rate": 4.091713532858773e-05,
      "loss": 2.4611,
      "step": 4720
    },
    {
      "epoch": 0.5558820072864026,
      "grad_norm": 20.95039176940918,
      "learning_rate": 4.089747119203996e-05,
      "loss": 2.27,
      "step": 4730
    },
    {
      "epoch": 0.5570572335174521,
      "grad_norm": 11.924006462097168,
      "learning_rate": 4.0877807055492194e-05,
      "loss": 1.6886,
      "step": 4740
    },
    {
      "epoch": 0.5582324597485016,
      "grad_norm": 10.582413673400879,
      "learning_rate": 4.085814291894443e-05,
      "loss": 2.0723,
      "step": 4750
    },
    {
      "epoch": 0.559407685979551,
      "grad_norm": 9.969587326049805,
      "learning_rate": 4.083847878239667e-05,
      "loss": 1.8284,
      "step": 4760
    },
    {
      "epoch": 0.5605829122106005,
      "grad_norm": 15.493760108947754,
      "learning_rate": 4.08188146458489e-05,
      "loss": 2.3784,
      "step": 4770
    },
    {
      "epoch": 0.56175813844165,
      "grad_norm": 19.177885055541992,
      "learning_rate": 4.0799150509301135e-05,
      "loss": 1.6298,
      "step": 4780
    },
    {
      "epoch": 0.5629333646726995,
      "grad_norm": 20.232471466064453,
      "learning_rate": 4.0779486372753376e-05,
      "loss": 1.797,
      "step": 4790
    },
    {
      "epoch": 0.564108590903749,
      "grad_norm": 10.184670448303223,
      "learning_rate": 4.075982223620561e-05,
      "loss": 1.9593,
      "step": 4800
    },
    {
      "epoch": 0.5652838171347985,
      "grad_norm": 5.950296401977539,
      "learning_rate": 4.074015809965784e-05,
      "loss": 1.369,
      "step": 4810
    },
    {
      "epoch": 0.566459043365848,
      "grad_norm": 12.744242668151855,
      "learning_rate": 4.0720493963110076e-05,
      "loss": 1.7761,
      "step": 4820
    },
    {
      "epoch": 0.5676342695968974,
      "grad_norm": 9.7475004196167,
      "learning_rate": 4.070082982656232e-05,
      "loss": 1.9751,
      "step": 4830
    },
    {
      "epoch": 0.5688094958279469,
      "grad_norm": 15.445152282714844,
      "learning_rate": 4.068116569001456e-05,
      "loss": 1.6615,
      "step": 4840
    },
    {
      "epoch": 0.5699847220589963,
      "grad_norm": 12.188678741455078,
      "learning_rate": 4.066150155346679e-05,
      "loss": 1.6747,
      "step": 4850
    },
    {
      "epoch": 0.5711599482900458,
      "grad_norm": 10.670065879821777,
      "learning_rate": 4.0641837416919024e-05,
      "loss": 2.6432,
      "step": 4860
    },
    {
      "epoch": 0.5723351745210953,
      "grad_norm": 16.704030990600586,
      "learning_rate": 4.0622173280371265e-05,
      "loss": 1.5313,
      "step": 4870
    },
    {
      "epoch": 0.5735104007521448,
      "grad_norm": 8.647698402404785,
      "learning_rate": 4.06025091438235e-05,
      "loss": 1.722,
      "step": 4880
    },
    {
      "epoch": 0.5746856269831943,
      "grad_norm": 13.972620010375977,
      "learning_rate": 4.058284500727573e-05,
      "loss": 1.5692,
      "step": 4890
    },
    {
      "epoch": 0.5758608532142437,
      "grad_norm": 12.39388656616211,
      "learning_rate": 4.0563180870727966e-05,
      "loss": 2.015,
      "step": 4900
    },
    {
      "epoch": 0.5770360794452932,
      "grad_norm": 6.4351115226745605,
      "learning_rate": 4.0543516734180206e-05,
      "loss": 2.2266,
      "step": 4910
    },
    {
      "epoch": 0.5782113056763427,
      "grad_norm": 23.075220108032227,
      "learning_rate": 4.052385259763244e-05,
      "loss": 2.0282,
      "step": 4920
    },
    {
      "epoch": 0.5793865319073922,
      "grad_norm": 13.686119079589844,
      "learning_rate": 4.050418846108467e-05,
      "loss": 2.3258,
      "step": 4930
    },
    {
      "epoch": 0.5805617581384417,
      "grad_norm": 18.06836700439453,
      "learning_rate": 4.048452432453691e-05,
      "loss": 2.1781,
      "step": 4940
    },
    {
      "epoch": 0.5817369843694912,
      "grad_norm": 9.043302536010742,
      "learning_rate": 4.046486018798915e-05,
      "loss": 2.1305,
      "step": 4950
    },
    {
      "epoch": 0.5829122106005407,
      "grad_norm": 15.016231536865234,
      "learning_rate": 4.044519605144138e-05,
      "loss": 2.1385,
      "step": 4960
    },
    {
      "epoch": 0.58408743683159,
      "grad_norm": 9.747109413146973,
      "learning_rate": 4.0425531914893614e-05,
      "loss": 1.9265,
      "step": 4970
    },
    {
      "epoch": 0.5852626630626395,
      "grad_norm": 23.34383773803711,
      "learning_rate": 4.0405867778345855e-05,
      "loss": 2.1382,
      "step": 4980
    },
    {
      "epoch": 0.586437889293689,
      "grad_norm": 20.86282730102539,
      "learning_rate": 4.0386203641798095e-05,
      "loss": 2.3051,
      "step": 4990
    },
    {
      "epoch": 0.5876131155247385,
      "grad_norm": 27.73655891418457,
      "learning_rate": 4.036653950525033e-05,
      "loss": 2.2232,
      "step": 5000
    },
    {
      "epoch": 0.588788341755788,
      "grad_norm": 9.232508659362793,
      "learning_rate": 4.034687536870256e-05,
      "loss": 1.825,
      "step": 5010
    },
    {
      "epoch": 0.5899635679868375,
      "grad_norm": 6.710579872131348,
      "learning_rate": 4.0327211232154796e-05,
      "loss": 2.1866,
      "step": 5020
    },
    {
      "epoch": 0.591138794217887,
      "grad_norm": 10.981929779052734,
      "learning_rate": 4.0307547095607036e-05,
      "loss": 2.3773,
      "step": 5030
    },
    {
      "epoch": 0.5923140204489364,
      "grad_norm": 24.413522720336914,
      "learning_rate": 4.028788295905927e-05,
      "loss": 1.8833,
      "step": 5040
    },
    {
      "epoch": 0.5934892466799859,
      "grad_norm": 14.641810417175293,
      "learning_rate": 4.0268218822511504e-05,
      "loss": 2.1224,
      "step": 5050
    },
    {
      "epoch": 0.5946644729110354,
      "grad_norm": 15.01045036315918,
      "learning_rate": 4.0248554685963744e-05,
      "loss": 1.8218,
      "step": 5060
    },
    {
      "epoch": 0.5958396991420849,
      "grad_norm": 10.174042701721191,
      "learning_rate": 4.022889054941598e-05,
      "loss": 2.573,
      "step": 5070
    },
    {
      "epoch": 0.5970149253731343,
      "grad_norm": 26.6146240234375,
      "learning_rate": 4.020922641286821e-05,
      "loss": 2.4968,
      "step": 5080
    },
    {
      "epoch": 0.5981901516041838,
      "grad_norm": 10.708161354064941,
      "learning_rate": 4.0189562276320445e-05,
      "loss": 1.76,
      "step": 5090
    },
    {
      "epoch": 0.5993653778352332,
      "grad_norm": 14.08447551727295,
      "learning_rate": 4.0169898139772685e-05,
      "loss": 1.9514,
      "step": 5100
    },
    {
      "epoch": 0.6005406040662827,
      "grad_norm": 17.408418655395508,
      "learning_rate": 4.015023400322492e-05,
      "loss": 2.261,
      "step": 5110
    },
    {
      "epoch": 0.6017158302973322,
      "grad_norm": 10.453878402709961,
      "learning_rate": 4.013056986667715e-05,
      "loss": 2.1979,
      "step": 5120
    },
    {
      "epoch": 0.6028910565283817,
      "grad_norm": 22.25663185119629,
      "learning_rate": 4.011090573012939e-05,
      "loss": 2.2191,
      "step": 5130
    },
    {
      "epoch": 0.6040662827594312,
      "grad_norm": 11.508066177368164,
      "learning_rate": 4.0091241593581627e-05,
      "loss": 1.8439,
      "step": 5140
    },
    {
      "epoch": 0.6052415089904807,
      "grad_norm": 11.035123825073242,
      "learning_rate": 4.007157745703387e-05,
      "loss": 2.4275,
      "step": 5150
    },
    {
      "epoch": 0.6064167352215302,
      "grad_norm": 12.286118507385254,
      "learning_rate": 4.00519133204861e-05,
      "loss": 2.0651,
      "step": 5160
    },
    {
      "epoch": 0.6075919614525797,
      "grad_norm": 13.767475128173828,
      "learning_rate": 4.0032249183938334e-05,
      "loss": 1.9408,
      "step": 5170
    },
    {
      "epoch": 0.6087671876836291,
      "grad_norm": 25.4036865234375,
      "learning_rate": 4.0012585047390575e-05,
      "loss": 2.2239,
      "step": 5180
    },
    {
      "epoch": 0.6099424139146786,
      "grad_norm": 10.103066444396973,
      "learning_rate": 3.999292091084281e-05,
      "loss": 1.867,
      "step": 5190
    },
    {
      "epoch": 0.611117640145728,
      "grad_norm": 13.518135070800781,
      "learning_rate": 3.997325677429504e-05,
      "loss": 2.0714,
      "step": 5200
    },
    {
      "epoch": 0.6122928663767775,
      "grad_norm": 22.083480834960938,
      "learning_rate": 3.9953592637747275e-05,
      "loss": 1.9703,
      "step": 5210
    },
    {
      "epoch": 0.613468092607827,
      "grad_norm": 9.441324234008789,
      "learning_rate": 3.9933928501199516e-05,
      "loss": 2.1924,
      "step": 5220
    },
    {
      "epoch": 0.6146433188388765,
      "grad_norm": 22.47870445251465,
      "learning_rate": 3.991426436465175e-05,
      "loss": 2.1398,
      "step": 5230
    },
    {
      "epoch": 0.615818545069926,
      "grad_norm": 19.50821304321289,
      "learning_rate": 3.989460022810398e-05,
      "loss": 2.3753,
      "step": 5240
    },
    {
      "epoch": 0.6169937713009754,
      "grad_norm": 8.568436622619629,
      "learning_rate": 3.987493609155622e-05,
      "loss": 2.4899,
      "step": 5250
    },
    {
      "epoch": 0.6181689975320249,
      "grad_norm": 11.066839218139648,
      "learning_rate": 3.985527195500846e-05,
      "loss": 1.81,
      "step": 5260
    },
    {
      "epoch": 0.6193442237630744,
      "grad_norm": 16.264280319213867,
      "learning_rate": 3.983560781846069e-05,
      "loss": 1.8081,
      "step": 5270
    },
    {
      "epoch": 0.6205194499941239,
      "grad_norm": 12.543164253234863,
      "learning_rate": 3.981594368191293e-05,
      "loss": 1.9799,
      "step": 5280
    },
    {
      "epoch": 0.6216946762251734,
      "grad_norm": 10.995649337768555,
      "learning_rate": 3.9796279545365165e-05,
      "loss": 2.0387,
      "step": 5290
    },
    {
      "epoch": 0.6228699024562229,
      "grad_norm": 29.905086517333984,
      "learning_rate": 3.9776615408817405e-05,
      "loss": 1.5276,
      "step": 5300
    },
    {
      "epoch": 0.6240451286872724,
      "grad_norm": 7.774795055389404,
      "learning_rate": 3.975695127226964e-05,
      "loss": 1.7981,
      "step": 5310
    },
    {
      "epoch": 0.6252203549183217,
      "grad_norm": 24.474409103393555,
      "learning_rate": 3.973728713572187e-05,
      "loss": 2.3126,
      "step": 5320
    },
    {
      "epoch": 0.6263955811493712,
      "grad_norm": 8.131673812866211,
      "learning_rate": 3.9717622999174106e-05,
      "loss": 2.3414,
      "step": 5330
    },
    {
      "epoch": 0.6275708073804207,
      "grad_norm": 9.89977741241455,
      "learning_rate": 3.9697958862626346e-05,
      "loss": 1.6193,
      "step": 5340
    },
    {
      "epoch": 0.6287460336114702,
      "grad_norm": 31.068796157836914,
      "learning_rate": 3.967829472607858e-05,
      "loss": 1.743,
      "step": 5350
    },
    {
      "epoch": 0.6299212598425197,
      "grad_norm": 16.621179580688477,
      "learning_rate": 3.9658630589530813e-05,
      "loss": 2.1564,
      "step": 5360
    },
    {
      "epoch": 0.6310964860735692,
      "grad_norm": 8.795929908752441,
      "learning_rate": 3.9638966452983054e-05,
      "loss": 2.0362,
      "step": 5370
    },
    {
      "epoch": 0.6322717123046186,
      "grad_norm": 4.510135650634766,
      "learning_rate": 3.961930231643529e-05,
      "loss": 1.443,
      "step": 5380
    },
    {
      "epoch": 0.6334469385356681,
      "grad_norm": 23.916004180908203,
      "learning_rate": 3.959963817988752e-05,
      "loss": 2.3416,
      "step": 5390
    },
    {
      "epoch": 0.6346221647667176,
      "grad_norm": 12.815582275390625,
      "learning_rate": 3.9579974043339755e-05,
      "loss": 1.93,
      "step": 5400
    },
    {
      "epoch": 0.6357973909977671,
      "grad_norm": 21.411727905273438,
      "learning_rate": 3.9560309906791995e-05,
      "loss": 2.0291,
      "step": 5410
    },
    {
      "epoch": 0.6369726172288166,
      "grad_norm": 10.167665481567383,
      "learning_rate": 3.954064577024423e-05,
      "loss": 1.9472,
      "step": 5420
    },
    {
      "epoch": 0.638147843459866,
      "grad_norm": 7.823069095611572,
      "learning_rate": 3.952098163369647e-05,
      "loss": 1.9115,
      "step": 5430
    },
    {
      "epoch": 0.6393230696909155,
      "grad_norm": 5.262133598327637,
      "learning_rate": 3.95013174971487e-05,
      "loss": 2.014,
      "step": 5440
    },
    {
      "epoch": 0.6404982959219649,
      "grad_norm": 8.032809257507324,
      "learning_rate": 3.948165336060094e-05,
      "loss": 1.9596,
      "step": 5450
    },
    {
      "epoch": 0.6416735221530144,
      "grad_norm": 8.476487159729004,
      "learning_rate": 3.946198922405318e-05,
      "loss": 1.2427,
      "step": 5460
    },
    {
      "epoch": 0.6428487483840639,
      "grad_norm": 15.777174949645996,
      "learning_rate": 3.944232508750541e-05,
      "loss": 1.9224,
      "step": 5470
    },
    {
      "epoch": 0.6440239746151134,
      "grad_norm": 17.618005752563477,
      "learning_rate": 3.9422660950957644e-05,
      "loss": 1.398,
      "step": 5480
    },
    {
      "epoch": 0.6451992008461629,
      "grad_norm": 21.035030364990234,
      "learning_rate": 3.9402996814409884e-05,
      "loss": 1.8636,
      "step": 5490
    },
    {
      "epoch": 0.6463744270772124,
      "grad_norm": 14.07835578918457,
      "learning_rate": 3.938333267786212e-05,
      "loss": 1.6521,
      "step": 5500
    },
    {
      "epoch": 0.6475496533082619,
      "grad_norm": 12.100227355957031,
      "learning_rate": 3.936366854131435e-05,
      "loss": 2.2591,
      "step": 5510
    },
    {
      "epoch": 0.6487248795393113,
      "grad_norm": 13.00106143951416,
      "learning_rate": 3.9344004404766585e-05,
      "loss": 1.9429,
      "step": 5520
    },
    {
      "epoch": 0.6499001057703608,
      "grad_norm": 73.96709442138672,
      "learning_rate": 3.9324340268218826e-05,
      "loss": 2.1761,
      "step": 5530
    },
    {
      "epoch": 0.6510753320014103,
      "grad_norm": 17.091020584106445,
      "learning_rate": 3.930467613167106e-05,
      "loss": 1.6888,
      "step": 5540
    },
    {
      "epoch": 0.6522505582324597,
      "grad_norm": 22.972827911376953,
      "learning_rate": 3.928501199512329e-05,
      "loss": 2.0907,
      "step": 5550
    },
    {
      "epoch": 0.6534257844635092,
      "grad_norm": 14.848004341125488,
      "learning_rate": 3.9265347858575526e-05,
      "loss": 1.7837,
      "step": 5560
    },
    {
      "epoch": 0.6546010106945587,
      "grad_norm": 12.785064697265625,
      "learning_rate": 3.924568372202777e-05,
      "loss": 1.771,
      "step": 5570
    },
    {
      "epoch": 0.6557762369256082,
      "grad_norm": 13.27304458618164,
      "learning_rate": 3.922601958548001e-05,
      "loss": 1.5672,
      "step": 5580
    },
    {
      "epoch": 0.6569514631566576,
      "grad_norm": 11.053291320800781,
      "learning_rate": 3.920635544893224e-05,
      "loss": 1.9526,
      "step": 5590
    },
    {
      "epoch": 0.6581266893877071,
      "grad_norm": 31.38985252380371,
      "learning_rate": 3.9186691312384474e-05,
      "loss": 1.6125,
      "step": 5600
    },
    {
      "epoch": 0.6593019156187566,
      "grad_norm": 12.434470176696777,
      "learning_rate": 3.9167027175836715e-05,
      "loss": 2.2247,
      "step": 5610
    },
    {
      "epoch": 0.6604771418498061,
      "grad_norm": 13.813992500305176,
      "learning_rate": 3.914736303928895e-05,
      "loss": 2.0007,
      "step": 5620
    },
    {
      "epoch": 0.6616523680808556,
      "grad_norm": 8.093582153320312,
      "learning_rate": 3.912769890274118e-05,
      "loss": 1.6136,
      "step": 5630
    },
    {
      "epoch": 0.6628275943119051,
      "grad_norm": 27.6748104095459,
      "learning_rate": 3.9108034766193416e-05,
      "loss": 1.6736,
      "step": 5640
    },
    {
      "epoch": 0.6640028205429546,
      "grad_norm": 10.874761581420898,
      "learning_rate": 3.9088370629645656e-05,
      "loss": 2.171,
      "step": 5650
    },
    {
      "epoch": 0.665178046774004,
      "grad_norm": 7.07586669921875,
      "learning_rate": 3.906870649309789e-05,
      "loss": 2.4721,
      "step": 5660
    },
    {
      "epoch": 0.6663532730050534,
      "grad_norm": 14.131163597106934,
      "learning_rate": 3.904904235655012e-05,
      "loss": 2.2313,
      "step": 5670
    },
    {
      "epoch": 0.6675284992361029,
      "grad_norm": 18.61333656311035,
      "learning_rate": 3.9029378220002364e-05,
      "loss": 1.8836,
      "step": 5680
    },
    {
      "epoch": 0.6687037254671524,
      "grad_norm": 22.422771453857422,
      "learning_rate": 3.90097140834546e-05,
      "loss": 2.4565,
      "step": 5690
    },
    {
      "epoch": 0.6698789516982019,
      "grad_norm": 11.821118354797363,
      "learning_rate": 3.899004994690683e-05,
      "loss": 2.1077,
      "step": 5700
    },
    {
      "epoch": 0.6710541779292514,
      "grad_norm": 17.069198608398438,
      "learning_rate": 3.8970385810359064e-05,
      "loss": 1.9352,
      "step": 5710
    },
    {
      "epoch": 0.6722294041603009,
      "grad_norm": 17.396081924438477,
      "learning_rate": 3.8950721673811305e-05,
      "loss": 2.3457,
      "step": 5720
    },
    {
      "epoch": 0.6734046303913503,
      "grad_norm": 25.62839126586914,
      "learning_rate": 3.8931057537263545e-05,
      "loss": 1.8997,
      "step": 5730
    },
    {
      "epoch": 0.6745798566223998,
      "grad_norm": 11.360404968261719,
      "learning_rate": 3.891139340071578e-05,
      "loss": 1.6624,
      "step": 5740
    },
    {
      "epoch": 0.6757550828534493,
      "grad_norm": 7.564605236053467,
      "learning_rate": 3.889172926416801e-05,
      "loss": 1.6724,
      "step": 5750
    },
    {
      "epoch": 0.6769303090844988,
      "grad_norm": 9.889302253723145,
      "learning_rate": 3.887206512762025e-05,
      "loss": 1.664,
      "step": 5760
    },
    {
      "epoch": 0.6781055353155483,
      "grad_norm": 12.758750915527344,
      "learning_rate": 3.8852400991072486e-05,
      "loss": 2.2464,
      "step": 5770
    },
    {
      "epoch": 0.6792807615465977,
      "grad_norm": 10.290943145751953,
      "learning_rate": 3.883273685452472e-05,
      "loss": 1.683,
      "step": 5780
    },
    {
      "epoch": 0.6804559877776472,
      "grad_norm": 10.635228157043457,
      "learning_rate": 3.8813072717976954e-05,
      "loss": 1.6601,
      "step": 5790
    },
    {
      "epoch": 0.6816312140086966,
      "grad_norm": 9.073125839233398,
      "learning_rate": 3.8793408581429194e-05,
      "loss": 1.7135,
      "step": 5800
    },
    {
      "epoch": 0.6828064402397461,
      "grad_norm": 9.590621948242188,
      "learning_rate": 3.877374444488143e-05,
      "loss": 1.9232,
      "step": 5810
    },
    {
      "epoch": 0.6839816664707956,
      "grad_norm": 12.242513656616211,
      "learning_rate": 3.875408030833366e-05,
      "loss": 2.4672,
      "step": 5820
    },
    {
      "epoch": 0.6851568927018451,
      "grad_norm": 19.91387367248535,
      "learning_rate": 3.8734416171785895e-05,
      "loss": 1.563,
      "step": 5830
    },
    {
      "epoch": 0.6863321189328946,
      "grad_norm": 9.85755443572998,
      "learning_rate": 3.8714752035238135e-05,
      "loss": 2.1931,
      "step": 5840
    },
    {
      "epoch": 0.6875073451639441,
      "grad_norm": 17.608531951904297,
      "learning_rate": 3.869508789869037e-05,
      "loss": 1.9464,
      "step": 5850
    },
    {
      "epoch": 0.6886825713949936,
      "grad_norm": 14.787233352661133,
      "learning_rate": 3.86754237621426e-05,
      "loss": 1.6162,
      "step": 5860
    },
    {
      "epoch": 0.689857797626043,
      "grad_norm": 16.22344207763672,
      "learning_rate": 3.8655759625594836e-05,
      "loss": 1.6166,
      "step": 5870
    },
    {
      "epoch": 0.6910330238570925,
      "grad_norm": 6.786860466003418,
      "learning_rate": 3.863609548904708e-05,
      "loss": 1.8421,
      "step": 5880
    },
    {
      "epoch": 0.692208250088142,
      "grad_norm": 12.529298782348633,
      "learning_rate": 3.861643135249932e-05,
      "loss": 2.1653,
      "step": 5890
    },
    {
      "epoch": 0.6933834763191914,
      "grad_norm": 12.059136390686035,
      "learning_rate": 3.859676721595155e-05,
      "loss": 1.8071,
      "step": 5900
    },
    {
      "epoch": 0.6945587025502409,
      "grad_norm": 15.251533508300781,
      "learning_rate": 3.8577103079403784e-05,
      "loss": 1.9666,
      "step": 5910
    },
    {
      "epoch": 0.6957339287812904,
      "grad_norm": 13.918059349060059,
      "learning_rate": 3.8557438942856025e-05,
      "loss": 1.8882,
      "step": 5920
    },
    {
      "epoch": 0.6969091550123399,
      "grad_norm": 9.304638862609863,
      "learning_rate": 3.853777480630826e-05,
      "loss": 1.5483,
      "step": 5930
    },
    {
      "epoch": 0.6980843812433893,
      "grad_norm": 11.300518989562988,
      "learning_rate": 3.851811066976049e-05,
      "loss": 1.785,
      "step": 5940
    },
    {
      "epoch": 0.6992596074744388,
      "grad_norm": 18.209569931030273,
      "learning_rate": 3.8498446533212725e-05,
      "loss": 2.2356,
      "step": 5950
    },
    {
      "epoch": 0.7004348337054883,
      "grad_norm": 18.26411247253418,
      "learning_rate": 3.8478782396664966e-05,
      "loss": 1.9583,
      "step": 5960
    },
    {
      "epoch": 0.7016100599365378,
      "grad_norm": 29.290313720703125,
      "learning_rate": 3.84591182601172e-05,
      "loss": 2.3154,
      "step": 5970
    },
    {
      "epoch": 0.7027852861675873,
      "grad_norm": 20.333215713500977,
      "learning_rate": 3.843945412356943e-05,
      "loss": 2.4223,
      "step": 5980
    },
    {
      "epoch": 0.7039605123986368,
      "grad_norm": 28.403518676757812,
      "learning_rate": 3.8419789987021673e-05,
      "loss": 2.3534,
      "step": 5990
    },
    {
      "epoch": 0.7051357386296863,
      "grad_norm": 11.260937690734863,
      "learning_rate": 3.840012585047391e-05,
      "loss": 2.1588,
      "step": 6000
    },
    {
      "epoch": 0.7063109648607357,
      "grad_norm": 18.140644073486328,
      "learning_rate": 3.838046171392614e-05,
      "loss": 2.165,
      "step": 6010
    },
    {
      "epoch": 0.7074861910917851,
      "grad_norm": 11.891374588012695,
      "learning_rate": 3.8360797577378374e-05,
      "loss": 2.1279,
      "step": 6020
    },
    {
      "epoch": 0.7086614173228346,
      "grad_norm": 16.706867218017578,
      "learning_rate": 3.8341133440830615e-05,
      "loss": 1.9422,
      "step": 6030
    },
    {
      "epoch": 0.7098366435538841,
      "grad_norm": 14.689982414245605,
      "learning_rate": 3.8321469304282855e-05,
      "loss": 1.4434,
      "step": 6040
    },
    {
      "epoch": 0.7110118697849336,
      "grad_norm": 13.030900955200195,
      "learning_rate": 3.830180516773509e-05,
      "loss": 1.9848,
      "step": 6050
    },
    {
      "epoch": 0.7121870960159831,
      "grad_norm": 11.093910217285156,
      "learning_rate": 3.828214103118732e-05,
      "loss": 1.9454,
      "step": 6060
    },
    {
      "epoch": 0.7133623222470326,
      "grad_norm": 24.869171142578125,
      "learning_rate": 3.826247689463956e-05,
      "loss": 1.7205,
      "step": 6070
    },
    {
      "epoch": 0.714537548478082,
      "grad_norm": 23.954235076904297,
      "learning_rate": 3.8242812758091796e-05,
      "loss": 2.4208,
      "step": 6080
    },
    {
      "epoch": 0.7157127747091315,
      "grad_norm": 6.028237819671631,
      "learning_rate": 3.822314862154403e-05,
      "loss": 1.7076,
      "step": 6090
    },
    {
      "epoch": 0.716888000940181,
      "grad_norm": 12.844260215759277,
      "learning_rate": 3.8203484484996263e-05,
      "loss": 1.5325,
      "step": 6100
    },
    {
      "epoch": 0.7180632271712305,
      "grad_norm": 11.704940795898438,
      "learning_rate": 3.8183820348448504e-05,
      "loss": 2.0805,
      "step": 6110
    },
    {
      "epoch": 0.71923845340228,
      "grad_norm": 9.7611722946167,
      "learning_rate": 3.816415621190074e-05,
      "loss": 1.8108,
      "step": 6120
    },
    {
      "epoch": 0.7204136796333294,
      "grad_norm": 20.009904861450195,
      "learning_rate": 3.814449207535297e-05,
      "loss": 1.9875,
      "step": 6130
    },
    {
      "epoch": 0.7215889058643788,
      "grad_norm": 9.196527481079102,
      "learning_rate": 3.8124827938805205e-05,
      "loss": 1.9854,
      "step": 6140
    },
    {
      "epoch": 0.7227641320954283,
      "grad_norm": 13.214926719665527,
      "learning_rate": 3.8105163802257445e-05,
      "loss": 1.9409,
      "step": 6150
    },
    {
      "epoch": 0.7239393583264778,
      "grad_norm": 12.050033569335938,
      "learning_rate": 3.808549966570968e-05,
      "loss": 1.7806,
      "step": 6160
    },
    {
      "epoch": 0.7251145845575273,
      "grad_norm": 26.977994918823242,
      "learning_rate": 3.806583552916191e-05,
      "loss": 1.7743,
      "step": 6170
    },
    {
      "epoch": 0.7262898107885768,
      "grad_norm": 11.205341339111328,
      "learning_rate": 3.804617139261415e-05,
      "loss": 2.2143,
      "step": 6180
    },
    {
      "epoch": 0.7274650370196263,
      "grad_norm": 9.094237327575684,
      "learning_rate": 3.802650725606639e-05,
      "loss": 1.8843,
      "step": 6190
    },
    {
      "epoch": 0.7286402632506758,
      "grad_norm": 15.118256568908691,
      "learning_rate": 3.800684311951863e-05,
      "loss": 1.6474,
      "step": 6200
    },
    {
      "epoch": 0.7298154894817253,
      "grad_norm": 9.911669731140137,
      "learning_rate": 3.798717898297086e-05,
      "loss": 1.8026,
      "step": 6210
    },
    {
      "epoch": 0.7309907157127747,
      "grad_norm": 11.229463577270508,
      "learning_rate": 3.7967514846423094e-05,
      "loss": 1.8114,
      "step": 6220
    },
    {
      "epoch": 0.7321659419438242,
      "grad_norm": 27.58434295654297,
      "learning_rate": 3.7947850709875334e-05,
      "loss": 1.976,
      "step": 6230
    },
    {
      "epoch": 0.7333411681748737,
      "grad_norm": 12.51749038696289,
      "learning_rate": 3.792818657332757e-05,
      "loss": 2.4114,
      "step": 6240
    },
    {
      "epoch": 0.7345163944059231,
      "grad_norm": 17.795377731323242,
      "learning_rate": 3.79085224367798e-05,
      "loss": 2.0498,
      "step": 6250
    },
    {
      "epoch": 0.7356916206369726,
      "grad_norm": 10.41335678100586,
      "learning_rate": 3.7888858300232035e-05,
      "loss": 1.5032,
      "step": 6260
    },
    {
      "epoch": 0.7368668468680221,
      "grad_norm": 19.377588272094727,
      "learning_rate": 3.7869194163684276e-05,
      "loss": 1.8558,
      "step": 6270
    },
    {
      "epoch": 0.7380420730990715,
      "grad_norm": 11.052696228027344,
      "learning_rate": 3.784953002713651e-05,
      "loss": 1.8426,
      "step": 6280
    },
    {
      "epoch": 0.739217299330121,
      "grad_norm": 11.877016067504883,
      "learning_rate": 3.782986589058874e-05,
      "loss": 1.7679,
      "step": 6290
    },
    {
      "epoch": 0.7403925255611705,
      "grad_norm": 7.949277877807617,
      "learning_rate": 3.781020175404098e-05,
      "loss": 1.8336,
      "step": 6300
    },
    {
      "epoch": 0.74156775179222,
      "grad_norm": 15.558849334716797,
      "learning_rate": 3.779053761749322e-05,
      "loss": 1.3514,
      "step": 6310
    },
    {
      "epoch": 0.7427429780232695,
      "grad_norm": 11.91271686553955,
      "learning_rate": 3.777087348094545e-05,
      "loss": 1.6977,
      "step": 6320
    },
    {
      "epoch": 0.743918204254319,
      "grad_norm": 17.67454719543457,
      "learning_rate": 3.7751209344397684e-05,
      "loss": 1.9341,
      "step": 6330
    },
    {
      "epoch": 0.7450934304853685,
      "grad_norm": 27.854265213012695,
      "learning_rate": 3.7731545207849924e-05,
      "loss": 1.5597,
      "step": 6340
    },
    {
      "epoch": 0.746268656716418,
      "grad_norm": 14.907734870910645,
      "learning_rate": 3.7711881071302165e-05,
      "loss": 2.0119,
      "step": 6350
    },
    {
      "epoch": 0.7474438829474674,
      "grad_norm": 14.951427459716797,
      "learning_rate": 3.76922169347544e-05,
      "loss": 1.9893,
      "step": 6360
    },
    {
      "epoch": 0.7486191091785168,
      "grad_norm": 17.917308807373047,
      "learning_rate": 3.767255279820663e-05,
      "loss": 1.7075,
      "step": 6370
    },
    {
      "epoch": 0.7497943354095663,
      "grad_norm": 11.919801712036133,
      "learning_rate": 3.765288866165887e-05,
      "loss": 1.8995,
      "step": 6380
    },
    {
      "epoch": 0.7509695616406158,
      "grad_norm": 13.7293062210083,
      "learning_rate": 3.7633224525111106e-05,
      "loss": 2.1212,
      "step": 6390
    },
    {
      "epoch": 0.7521447878716653,
      "grad_norm": 15.8491792678833,
      "learning_rate": 3.761356038856334e-05,
      "loss": 1.7686,
      "step": 6400
    },
    {
      "epoch": 0.7533200141027148,
      "grad_norm": 17.1448974609375,
      "learning_rate": 3.759389625201557e-05,
      "loss": 1.9182,
      "step": 6410
    },
    {
      "epoch": 0.7544952403337642,
      "grad_norm": 18.209184646606445,
      "learning_rate": 3.7574232115467814e-05,
      "loss": 1.8353,
      "step": 6420
    },
    {
      "epoch": 0.7556704665648137,
      "grad_norm": 6.70125675201416,
      "learning_rate": 3.755456797892005e-05,
      "loss": 2.2013,
      "step": 6430
    },
    {
      "epoch": 0.7568456927958632,
      "grad_norm": 9.912574768066406,
      "learning_rate": 3.753490384237228e-05,
      "loss": 2.0342,
      "step": 6440
    },
    {
      "epoch": 0.7580209190269127,
      "grad_norm": 10.793564796447754,
      "learning_rate": 3.7515239705824514e-05,
      "loss": 1.9426,
      "step": 6450
    },
    {
      "epoch": 0.7591961452579622,
      "grad_norm": 28.88125228881836,
      "learning_rate": 3.7495575569276755e-05,
      "loss": 2.0267,
      "step": 6460
    },
    {
      "epoch": 0.7603713714890117,
      "grad_norm": 17.2239933013916,
      "learning_rate": 3.747591143272899e-05,
      "loss": 1.8522,
      "step": 6470
    },
    {
      "epoch": 0.7615465977200611,
      "grad_norm": 9.050847053527832,
      "learning_rate": 3.745624729618122e-05,
      "loss": 1.4325,
      "step": 6480
    },
    {
      "epoch": 0.7627218239511105,
      "grad_norm": 11.521880149841309,
      "learning_rate": 3.743658315963346e-05,
      "loss": 1.9194,
      "step": 6490
    },
    {
      "epoch": 0.76389705018216,
      "grad_norm": 21.49733543395996,
      "learning_rate": 3.74169190230857e-05,
      "loss": 1.4848,
      "step": 6500
    },
    {
      "epoch": 0.7650722764132095,
      "grad_norm": 27.21480369567871,
      "learning_rate": 3.7397254886537936e-05,
      "loss": 2.2529,
      "step": 6510
    },
    {
      "epoch": 0.766247502644259,
      "grad_norm": 8.720370292663574,
      "learning_rate": 3.737759074999017e-05,
      "loss": 1.5259,
      "step": 6520
    },
    {
      "epoch": 0.7674227288753085,
      "grad_norm": 7.001040458679199,
      "learning_rate": 3.7357926613442404e-05,
      "loss": 2.0116,
      "step": 6530
    },
    {
      "epoch": 0.768597955106358,
      "grad_norm": 13.576958656311035,
      "learning_rate": 3.7338262476894644e-05,
      "loss": 2.0402,
      "step": 6540
    },
    {
      "epoch": 0.7697731813374075,
      "grad_norm": 15.76539134979248,
      "learning_rate": 3.731859834034688e-05,
      "loss": 2.1422,
      "step": 6550
    },
    {
      "epoch": 0.770948407568457,
      "grad_norm": 15.315954208374023,
      "learning_rate": 3.729893420379911e-05,
      "loss": 1.6079,
      "step": 6560
    },
    {
      "epoch": 0.7721236337995064,
      "grad_norm": 20.747562408447266,
      "learning_rate": 3.727927006725135e-05,
      "loss": 1.9393,
      "step": 6570
    },
    {
      "epoch": 0.7732988600305559,
      "grad_norm": 12.648963928222656,
      "learning_rate": 3.7259605930703585e-05,
      "loss": 1.9412,
      "step": 6580
    },
    {
      "epoch": 0.7744740862616054,
      "grad_norm": 27.10748291015625,
      "learning_rate": 3.723994179415582e-05,
      "loss": 1.6691,
      "step": 6590
    },
    {
      "epoch": 0.7756493124926548,
      "grad_norm": 13.652043342590332,
      "learning_rate": 3.722027765760805e-05,
      "loss": 1.7174,
      "step": 6600
    },
    {
      "epoch": 0.7768245387237043,
      "grad_norm": 26.083919525146484,
      "learning_rate": 3.720061352106029e-05,
      "loss": 2.5388,
      "step": 6610
    },
    {
      "epoch": 0.7779997649547538,
      "grad_norm": 10.638964653015137,
      "learning_rate": 3.7180949384512527e-05,
      "loss": 2.2715,
      "step": 6620
    },
    {
      "epoch": 0.7791749911858032,
      "grad_norm": 17.88481330871582,
      "learning_rate": 3.716128524796476e-05,
      "loss": 2.2667,
      "step": 6630
    },
    {
      "epoch": 0.7803502174168527,
      "grad_norm": 15.97347354888916,
      "learning_rate": 3.7141621111417e-05,
      "loss": 2.0879,
      "step": 6640
    },
    {
      "epoch": 0.7815254436479022,
      "grad_norm": 13.433609962463379,
      "learning_rate": 3.7121956974869234e-05,
      "loss": 2.5867,
      "step": 6650
    },
    {
      "epoch": 0.7827006698789517,
      "grad_norm": 21.584537506103516,
      "learning_rate": 3.7102292838321475e-05,
      "loss": 1.5483,
      "step": 6660
    },
    {
      "epoch": 0.7838758961100012,
      "grad_norm": 13.01290225982666,
      "learning_rate": 3.708262870177371e-05,
      "loss": 1.9895,
      "step": 6670
    },
    {
      "epoch": 0.7850511223410507,
      "grad_norm": 15.897327423095703,
      "learning_rate": 3.706296456522594e-05,
      "loss": 1.9484,
      "step": 6680
    },
    {
      "epoch": 0.7862263485721002,
      "grad_norm": 10.022132873535156,
      "learning_rate": 3.704330042867818e-05,
      "loss": 1.6422,
      "step": 6690
    },
    {
      "epoch": 0.7874015748031497,
      "grad_norm": 11.054361343383789,
      "learning_rate": 3.7023636292130416e-05,
      "loss": 2.1063,
      "step": 6700
    },
    {
      "epoch": 0.7885768010341991,
      "grad_norm": 30.680805206298828,
      "learning_rate": 3.700397215558265e-05,
      "loss": 2.3198,
      "step": 6710
    },
    {
      "epoch": 0.7897520272652485,
      "grad_norm": 16.484920501708984,
      "learning_rate": 3.698430801903488e-05,
      "loss": 1.7849,
      "step": 6720
    },
    {
      "epoch": 0.790927253496298,
      "grad_norm": 16.158966064453125,
      "learning_rate": 3.6964643882487123e-05,
      "loss": 2.3246,
      "step": 6730
    },
    {
      "epoch": 0.7921024797273475,
      "grad_norm": 12.623407363891602,
      "learning_rate": 3.694497974593936e-05,
      "loss": 2.2631,
      "step": 6740
    },
    {
      "epoch": 0.793277705958397,
      "grad_norm": 11.289498329162598,
      "learning_rate": 3.692531560939159e-05,
      "loss": 2.6904,
      "step": 6750
    },
    {
      "epoch": 0.7944529321894465,
      "grad_norm": 33.807228088378906,
      "learning_rate": 3.6905651472843824e-05,
      "loss": 2.1143,
      "step": 6760
    },
    {
      "epoch": 0.795628158420496,
      "grad_norm": 16.463993072509766,
      "learning_rate": 3.6885987336296065e-05,
      "loss": 1.4052,
      "step": 6770
    },
    {
      "epoch": 0.7968033846515454,
      "grad_norm": 8.036648750305176,
      "learning_rate": 3.68663231997483e-05,
      "loss": 1.5512,
      "step": 6780
    },
    {
      "epoch": 0.7979786108825949,
      "grad_norm": 16.678142547607422,
      "learning_rate": 3.684665906320054e-05,
      "loss": 1.955,
      "step": 6790
    },
    {
      "epoch": 0.7991538371136444,
      "grad_norm": 26.22309684753418,
      "learning_rate": 3.682699492665277e-05,
      "loss": 2.4884,
      "step": 6800
    },
    {
      "epoch": 0.8003290633446939,
      "grad_norm": 18.768375396728516,
      "learning_rate": 3.680733079010501e-05,
      "loss": 1.9718,
      "step": 6810
    },
    {
      "epoch": 0.8015042895757434,
      "grad_norm": 9.46971321105957,
      "learning_rate": 3.6787666653557246e-05,
      "loss": 1.8164,
      "step": 6820
    },
    {
      "epoch": 0.8026795158067928,
      "grad_norm": 31.08321189880371,
      "learning_rate": 3.676800251700948e-05,
      "loss": 1.8393,
      "step": 6830
    },
    {
      "epoch": 0.8038547420378422,
      "grad_norm": 13.442791938781738,
      "learning_rate": 3.6748338380461713e-05,
      "loss": 1.8165,
      "step": 6840
    },
    {
      "epoch": 0.8050299682688917,
      "grad_norm": 12.324529647827148,
      "learning_rate": 3.6728674243913954e-05,
      "loss": 1.7166,
      "step": 6850
    },
    {
      "epoch": 0.8062051944999412,
      "grad_norm": 12.485682487487793,
      "learning_rate": 3.670901010736619e-05,
      "loss": 1.902,
      "step": 6860
    },
    {
      "epoch": 0.8073804207309907,
      "grad_norm": 11.34929084777832,
      "learning_rate": 3.668934597081842e-05,
      "loss": 1.7721,
      "step": 6870
    },
    {
      "epoch": 0.8085556469620402,
      "grad_norm": 13.027608871459961,
      "learning_rate": 3.666968183427066e-05,
      "loss": 2.1366,
      "step": 6880
    },
    {
      "epoch": 0.8097308731930897,
      "grad_norm": 12.046297073364258,
      "learning_rate": 3.6650017697722895e-05,
      "loss": 2.1751,
      "step": 6890
    },
    {
      "epoch": 0.8109060994241392,
      "grad_norm": 16.194910049438477,
      "learning_rate": 3.663035356117513e-05,
      "loss": 1.9359,
      "step": 6900
    },
    {
      "epoch": 0.8120813256551886,
      "grad_norm": 14.410651206970215,
      "learning_rate": 3.661068942462736e-05,
      "loss": 2.2546,
      "step": 6910
    },
    {
      "epoch": 0.8132565518862381,
      "grad_norm": 12.128220558166504,
      "learning_rate": 3.65910252880796e-05,
      "loss": 1.648,
      "step": 6920
    },
    {
      "epoch": 0.8144317781172876,
      "grad_norm": 4.600808143615723,
      "learning_rate": 3.6571361151531836e-05,
      "loss": 2.0743,
      "step": 6930
    },
    {
      "epoch": 0.8156070043483371,
      "grad_norm": 6.781635761260986,
      "learning_rate": 3.655169701498408e-05,
      "loss": 1.4468,
      "step": 6940
    },
    {
      "epoch": 0.8167822305793865,
      "grad_norm": 14.008431434631348,
      "learning_rate": 3.653203287843631e-05,
      "loss": 1.5395,
      "step": 6950
    },
    {
      "epoch": 0.817957456810436,
      "grad_norm": 15.067609786987305,
      "learning_rate": 3.651236874188855e-05,
      "loss": 2.1563,
      "step": 6960
    },
    {
      "epoch": 0.8191326830414855,
      "grad_norm": 15.527144432067871,
      "learning_rate": 3.6492704605340784e-05,
      "loss": 2.1833,
      "step": 6970
    },
    {
      "epoch": 0.8203079092725349,
      "grad_norm": 15.496681213378906,
      "learning_rate": 3.647304046879302e-05,
      "loss": 2.3473,
      "step": 6980
    },
    {
      "epoch": 0.8214831355035844,
      "grad_norm": 21.34918212890625,
      "learning_rate": 3.645337633224525e-05,
      "loss": 2.3862,
      "step": 6990
    },
    {
      "epoch": 0.8226583617346339,
      "grad_norm": 10.288919448852539,
      "learning_rate": 3.643371219569749e-05,
      "loss": 1.4424,
      "step": 7000
    },
    {
      "epoch": 0.8238335879656834,
      "grad_norm": 16.947193145751953,
      "learning_rate": 3.6414048059149726e-05,
      "loss": 1.8364,
      "step": 7010
    },
    {
      "epoch": 0.8250088141967329,
      "grad_norm": 10.199929237365723,
      "learning_rate": 3.639438392260196e-05,
      "loss": 2.0274,
      "step": 7020
    },
    {
      "epoch": 0.8261840404277824,
      "grad_norm": 23.926572799682617,
      "learning_rate": 3.637471978605419e-05,
      "loss": 2.0012,
      "step": 7030
    },
    {
      "epoch": 0.8273592666588319,
      "grad_norm": 16.770668029785156,
      "learning_rate": 3.635505564950643e-05,
      "loss": 2.0553,
      "step": 7040
    },
    {
      "epoch": 0.8285344928898813,
      "grad_norm": 13.710966110229492,
      "learning_rate": 3.633539151295867e-05,
      "loss": 1.9647,
      "step": 7050
    },
    {
      "epoch": 0.8297097191209308,
      "grad_norm": 13.782581329345703,
      "learning_rate": 3.63157273764109e-05,
      "loss": 2.2359,
      "step": 7060
    },
    {
      "epoch": 0.8308849453519802,
      "grad_norm": 19.12066078186035,
      "learning_rate": 3.6296063239863134e-05,
      "loss": 1.9038,
      "step": 7070
    },
    {
      "epoch": 0.8320601715830297,
      "grad_norm": 16.10237693786621,
      "learning_rate": 3.6276399103315374e-05,
      "loss": 2.3141,
      "step": 7080
    },
    {
      "epoch": 0.8332353978140792,
      "grad_norm": 13.086423873901367,
      "learning_rate": 3.6256734966767615e-05,
      "loss": 1.8449,
      "step": 7090
    },
    {
      "epoch": 0.8344106240451287,
      "grad_norm": 15.729324340820312,
      "learning_rate": 3.623707083021985e-05,
      "loss": 1.599,
      "step": 7100
    },
    {
      "epoch": 0.8355858502761782,
      "grad_norm": 14.840970993041992,
      "learning_rate": 3.621740669367208e-05,
      "loss": 1.6146,
      "step": 7110
    },
    {
      "epoch": 0.8367610765072276,
      "grad_norm": 13.48663330078125,
      "learning_rate": 3.619774255712432e-05,
      "loss": 1.0748,
      "step": 7120
    },
    {
      "epoch": 0.8379363027382771,
      "grad_norm": 20.701025009155273,
      "learning_rate": 3.6178078420576556e-05,
      "loss": 1.4644,
      "step": 7130
    },
    {
      "epoch": 0.8391115289693266,
      "grad_norm": 12.552802085876465,
      "learning_rate": 3.615841428402879e-05,
      "loss": 1.6454,
      "step": 7140
    },
    {
      "epoch": 0.8402867552003761,
      "grad_norm": 14.847660064697266,
      "learning_rate": 3.613875014748102e-05,
      "loss": 1.6019,
      "step": 7150
    },
    {
      "epoch": 0.8414619814314256,
      "grad_norm": 9.980250358581543,
      "learning_rate": 3.6119086010933264e-05,
      "loss": 2.1237,
      "step": 7160
    },
    {
      "epoch": 0.8426372076624751,
      "grad_norm": 14.057222366333008,
      "learning_rate": 3.60994218743855e-05,
      "loss": 2.1068,
      "step": 7170
    },
    {
      "epoch": 0.8438124338935244,
      "grad_norm": 13.82053279876709,
      "learning_rate": 3.607975773783773e-05,
      "loss": 2.4626,
      "step": 7180
    },
    {
      "epoch": 0.8449876601245739,
      "grad_norm": 24.27387809753418,
      "learning_rate": 3.606009360128997e-05,
      "loss": 2.1761,
      "step": 7190
    },
    {
      "epoch": 0.8461628863556234,
      "grad_norm": 11.664134979248047,
      "learning_rate": 3.6040429464742205e-05,
      "loss": 1.4874,
      "step": 7200
    },
    {
      "epoch": 0.8473381125866729,
      "grad_norm": 6.294652938842773,
      "learning_rate": 3.602076532819444e-05,
      "loss": 1.7561,
      "step": 7210
    },
    {
      "epoch": 0.8485133388177224,
      "grad_norm": 11.683948516845703,
      "learning_rate": 3.600110119164667e-05,
      "loss": 1.6405,
      "step": 7220
    },
    {
      "epoch": 0.8496885650487719,
      "grad_norm": 8.298276901245117,
      "learning_rate": 3.598143705509891e-05,
      "loss": 1.791,
      "step": 7230
    },
    {
      "epoch": 0.8508637912798214,
      "grad_norm": 16.4104061126709,
      "learning_rate": 3.5961772918551146e-05,
      "loss": 1.2812,
      "step": 7240
    },
    {
      "epoch": 0.8520390175108709,
      "grad_norm": 17.22831916809082,
      "learning_rate": 3.5942108782003386e-05,
      "loss": 1.7686,
      "step": 7250
    },
    {
      "epoch": 0.8532142437419203,
      "grad_norm": 16.144149780273438,
      "learning_rate": 3.592244464545562e-05,
      "loss": 1.4378,
      "step": 7260
    },
    {
      "epoch": 0.8543894699729698,
      "grad_norm": 13.728645324707031,
      "learning_rate": 3.590278050890786e-05,
      "loss": 1.5646,
      "step": 7270
    },
    {
      "epoch": 0.8555646962040193,
      "grad_norm": 6.314034461975098,
      "learning_rate": 3.5883116372360094e-05,
      "loss": 1.644,
      "step": 7280
    },
    {
      "epoch": 0.8567399224350688,
      "grad_norm": 11.229823112487793,
      "learning_rate": 3.586345223581233e-05,
      "loss": 1.5466,
      "step": 7290
    },
    {
      "epoch": 0.8579151486661182,
      "grad_norm": 12.772022247314453,
      "learning_rate": 3.584378809926456e-05,
      "loss": 1.4922,
      "step": 7300
    },
    {
      "epoch": 0.8590903748971677,
      "grad_norm": 25.962677001953125,
      "learning_rate": 3.58241239627168e-05,
      "loss": 1.8528,
      "step": 7310
    },
    {
      "epoch": 0.8602656011282171,
      "grad_norm": 4.849168300628662,
      "learning_rate": 3.5804459826169035e-05,
      "loss": 1.4968,
      "step": 7320
    },
    {
      "epoch": 0.8614408273592666,
      "grad_norm": 11.967452049255371,
      "learning_rate": 3.578479568962127e-05,
      "loss": 1.8573,
      "step": 7330
    },
    {
      "epoch": 0.8626160535903161,
      "grad_norm": 20.550186157226562,
      "learning_rate": 3.57651315530735e-05,
      "loss": 1.7965,
      "step": 7340
    },
    {
      "epoch": 0.8637912798213656,
      "grad_norm": 11.86355209350586,
      "learning_rate": 3.574546741652574e-05,
      "loss": 2.1767,
      "step": 7350
    },
    {
      "epoch": 0.8649665060524151,
      "grad_norm": 8.180103302001953,
      "learning_rate": 3.5725803279977977e-05,
      "loss": 2.0307,
      "step": 7360
    },
    {
      "epoch": 0.8661417322834646,
      "grad_norm": 10.04309368133545,
      "learning_rate": 3.570613914343021e-05,
      "loss": 2.1121,
      "step": 7370
    },
    {
      "epoch": 0.8673169585145141,
      "grad_norm": 11.695072174072266,
      "learning_rate": 3.5686475006882444e-05,
      "loss": 1.6365,
      "step": 7380
    },
    {
      "epoch": 0.8684921847455636,
      "grad_norm": 17.763805389404297,
      "learning_rate": 3.5666810870334684e-05,
      "loss": 1.6303,
      "step": 7390
    },
    {
      "epoch": 0.869667410976613,
      "grad_norm": 13.722623825073242,
      "learning_rate": 3.5647146733786925e-05,
      "loss": 1.7497,
      "step": 7400
    },
    {
      "epoch": 0.8708426372076625,
      "grad_norm": 13.455446243286133,
      "learning_rate": 3.562748259723916e-05,
      "loss": 1.8146,
      "step": 7410
    },
    {
      "epoch": 0.8720178634387119,
      "grad_norm": 9.228466033935547,
      "learning_rate": 3.560781846069139e-05,
      "loss": 1.4965,
      "step": 7420
    },
    {
      "epoch": 0.8731930896697614,
      "grad_norm": 15.176321983337402,
      "learning_rate": 3.558815432414363e-05,
      "loss": 2.0165,
      "step": 7430
    },
    {
      "epoch": 0.8743683159008109,
      "grad_norm": 10.301900863647461,
      "learning_rate": 3.5568490187595866e-05,
      "loss": 1.863,
      "step": 7440
    },
    {
      "epoch": 0.8755435421318604,
      "grad_norm": 14.082913398742676,
      "learning_rate": 3.55488260510481e-05,
      "loss": 1.8335,
      "step": 7450
    },
    {
      "epoch": 0.8767187683629099,
      "grad_norm": 9.901995658874512,
      "learning_rate": 3.552916191450033e-05,
      "loss": 1.7402,
      "step": 7460
    },
    {
      "epoch": 0.8778939945939593,
      "grad_norm": 13.065939903259277,
      "learning_rate": 3.5509497777952573e-05,
      "loss": 1.6718,
      "step": 7470
    },
    {
      "epoch": 0.8790692208250088,
      "grad_norm": 16.097402572631836,
      "learning_rate": 3.548983364140481e-05,
      "loss": 1.2003,
      "step": 7480
    },
    {
      "epoch": 0.8802444470560583,
      "grad_norm": 7.4037370681762695,
      "learning_rate": 3.547016950485704e-05,
      "loss": 1.8445,
      "step": 7490
    },
    {
      "epoch": 0.8814196732871078,
      "grad_norm": 12.112308502197266,
      "learning_rate": 3.545050536830928e-05,
      "loss": 2.1769,
      "step": 7500
    },
    {
      "epoch": 0.8825948995181573,
      "grad_norm": 12.137428283691406,
      "learning_rate": 3.5430841231761515e-05,
      "loss": 1.71,
      "step": 7510
    },
    {
      "epoch": 0.8837701257492068,
      "grad_norm": 14.16809368133545,
      "learning_rate": 3.541117709521375e-05,
      "loss": 1.8386,
      "step": 7520
    },
    {
      "epoch": 0.8849453519802561,
      "grad_norm": 12.808544158935547,
      "learning_rate": 3.539151295866598e-05,
      "loss": 1.4149,
      "step": 7530
    },
    {
      "epoch": 0.8861205782113056,
      "grad_norm": 16.46702766418457,
      "learning_rate": 3.537184882211822e-05,
      "loss": 2.4025,
      "step": 7540
    },
    {
      "epoch": 0.8872958044423551,
      "grad_norm": 15.187529563903809,
      "learning_rate": 3.535218468557046e-05,
      "loss": 2.1638,
      "step": 7550
    },
    {
      "epoch": 0.8884710306734046,
      "grad_norm": 8.340374946594238,
      "learning_rate": 3.5332520549022696e-05,
      "loss": 1.916,
      "step": 7560
    },
    {
      "epoch": 0.8896462569044541,
      "grad_norm": 27.85256004333496,
      "learning_rate": 3.531285641247493e-05,
      "loss": 1.9299,
      "step": 7570
    },
    {
      "epoch": 0.8908214831355036,
      "grad_norm": 29.517995834350586,
      "learning_rate": 3.529319227592717e-05,
      "loss": 1.9393,
      "step": 7580
    },
    {
      "epoch": 0.8919967093665531,
      "grad_norm": 15.084198951721191,
      "learning_rate": 3.5273528139379404e-05,
      "loss": 2.2398,
      "step": 7590
    },
    {
      "epoch": 0.8931719355976026,
      "grad_norm": 13.708508491516113,
      "learning_rate": 3.525386400283164e-05,
      "loss": 1.6697,
      "step": 7600
    },
    {
      "epoch": 0.894347161828652,
      "grad_norm": 21.984336853027344,
      "learning_rate": 3.523419986628387e-05,
      "loss": 1.5681,
      "step": 7610
    },
    {
      "epoch": 0.8955223880597015,
      "grad_norm": 17.745025634765625,
      "learning_rate": 3.521453572973611e-05,
      "loss": 2.1074,
      "step": 7620
    },
    {
      "epoch": 0.896697614290751,
      "grad_norm": 4.169613838195801,
      "learning_rate": 3.5194871593188345e-05,
      "loss": 2.071,
      "step": 7630
    },
    {
      "epoch": 0.8978728405218005,
      "grad_norm": 14.069290161132812,
      "learning_rate": 3.517520745664058e-05,
      "loss": 2.0623,
      "step": 7640
    },
    {
      "epoch": 0.8990480667528499,
      "grad_norm": 12.117961883544922,
      "learning_rate": 3.515554332009281e-05,
      "loss": 1.7799,
      "step": 7650
    },
    {
      "epoch": 0.9002232929838994,
      "grad_norm": 11.780905723571777,
      "learning_rate": 3.513587918354505e-05,
      "loss": 2.249,
      "step": 7660
    },
    {
      "epoch": 0.9013985192149488,
      "grad_norm": 18.406211853027344,
      "learning_rate": 3.5116215046997286e-05,
      "loss": 1.5808,
      "step": 7670
    },
    {
      "epoch": 0.9025737454459983,
      "grad_norm": 10.338708877563477,
      "learning_rate": 3.509655091044952e-05,
      "loss": 1.6302,
      "step": 7680
    },
    {
      "epoch": 0.9037489716770478,
      "grad_norm": 17.15137481689453,
      "learning_rate": 3.507688677390176e-05,
      "loss": 1.4361,
      "step": 7690
    },
    {
      "epoch": 0.9049241979080973,
      "grad_norm": 10.510708808898926,
      "learning_rate": 3.5057222637354e-05,
      "loss": 1.5944,
      "step": 7700
    },
    {
      "epoch": 0.9060994241391468,
      "grad_norm": 8.057929992675781,
      "learning_rate": 3.5037558500806234e-05,
      "loss": 1.746,
      "step": 7710
    },
    {
      "epoch": 0.9072746503701963,
      "grad_norm": 13.22546100616455,
      "learning_rate": 3.501789436425847e-05,
      "loss": 2.131,
      "step": 7720
    },
    {
      "epoch": 0.9084498766012458,
      "grad_norm": 11.387399673461914,
      "learning_rate": 3.49982302277107e-05,
      "loss": 2.0215,
      "step": 7730
    },
    {
      "epoch": 0.9096251028322953,
      "grad_norm": 8.647712707519531,
      "learning_rate": 3.497856609116294e-05,
      "loss": 1.7065,
      "step": 7740
    },
    {
      "epoch": 0.9108003290633447,
      "grad_norm": 11.75634479522705,
      "learning_rate": 3.4958901954615176e-05,
      "loss": 1.4477,
      "step": 7750
    },
    {
      "epoch": 0.9119755552943942,
      "grad_norm": 8.608869552612305,
      "learning_rate": 3.493923781806741e-05,
      "loss": 1.8211,
      "step": 7760
    },
    {
      "epoch": 0.9131507815254436,
      "grad_norm": 13.830964088439941,
      "learning_rate": 3.491957368151964e-05,
      "loss": 2.2674,
      "step": 7770
    },
    {
      "epoch": 0.9143260077564931,
      "grad_norm": 23.174726486206055,
      "learning_rate": 3.489990954497188e-05,
      "loss": 1.9623,
      "step": 7780
    },
    {
      "epoch": 0.9155012339875426,
      "grad_norm": 13.012772560119629,
      "learning_rate": 3.488024540842412e-05,
      "loss": 2.1506,
      "step": 7790
    },
    {
      "epoch": 0.9166764602185921,
      "grad_norm": 10.499960899353027,
      "learning_rate": 3.486058127187635e-05,
      "loss": 2.315,
      "step": 7800
    },
    {
      "epoch": 0.9178516864496415,
      "grad_norm": 13.583904266357422,
      "learning_rate": 3.484091713532859e-05,
      "loss": 1.9548,
      "step": 7810
    },
    {
      "epoch": 0.919026912680691,
      "grad_norm": 7.479710102081299,
      "learning_rate": 3.4821252998780824e-05,
      "loss": 1.7226,
      "step": 7820
    },
    {
      "epoch": 0.9202021389117405,
      "grad_norm": 4.831395626068115,
      "learning_rate": 3.480158886223306e-05,
      "loss": 1.8122,
      "step": 7830
    },
    {
      "epoch": 0.92137736514279,
      "grad_norm": 20.887252807617188,
      "learning_rate": 3.478192472568529e-05,
      "loss": 1.71,
      "step": 7840
    },
    {
      "epoch": 0.9225525913738395,
      "grad_norm": 10.533137321472168,
      "learning_rate": 3.476226058913753e-05,
      "loss": 1.9617,
      "step": 7850
    },
    {
      "epoch": 0.923727817604889,
      "grad_norm": 25.093826293945312,
      "learning_rate": 3.474259645258977e-05,
      "loss": 2.1241,
      "step": 7860
    },
    {
      "epoch": 0.9249030438359385,
      "grad_norm": 7.615812301635742,
      "learning_rate": 3.4722932316042006e-05,
      "loss": 1.7333,
      "step": 7870
    },
    {
      "epoch": 0.9260782700669878,
      "grad_norm": 22.462026596069336,
      "learning_rate": 3.470326817949424e-05,
      "loss": 1.689,
      "step": 7880
    },
    {
      "epoch": 0.9272534962980373,
      "grad_norm": 6.806323528289795,
      "learning_rate": 3.468360404294648e-05,
      "loss": 1.4039,
      "step": 7890
    },
    {
      "epoch": 0.9284287225290868,
      "grad_norm": 10.227314949035645,
      "learning_rate": 3.4663939906398714e-05,
      "loss": 1.5925,
      "step": 7900
    },
    {
      "epoch": 0.9296039487601363,
      "grad_norm": 18.27678871154785,
      "learning_rate": 3.464427576985095e-05,
      "loss": 1.4871,
      "step": 7910
    },
    {
      "epoch": 0.9307791749911858,
      "grad_norm": 14.01465129852295,
      "learning_rate": 3.462461163330318e-05,
      "loss": 2.0159,
      "step": 7920
    },
    {
      "epoch": 0.9319544012222353,
      "grad_norm": 13.822399139404297,
      "learning_rate": 3.460494749675542e-05,
      "loss": 1.4571,
      "step": 7930
    },
    {
      "epoch": 0.9331296274532848,
      "grad_norm": 15.277059555053711,
      "learning_rate": 3.4585283360207655e-05,
      "loss": 2.0774,
      "step": 7940
    },
    {
      "epoch": 0.9343048536843342,
      "grad_norm": 26.121625900268555,
      "learning_rate": 3.456561922365989e-05,
      "loss": 1.6506,
      "step": 7950
    },
    {
      "epoch": 0.9354800799153837,
      "grad_norm": 12.667616844177246,
      "learning_rate": 3.454595508711212e-05,
      "loss": 1.752,
      "step": 7960
    },
    {
      "epoch": 0.9366553061464332,
      "grad_norm": 15.098954200744629,
      "learning_rate": 3.452629095056436e-05,
      "loss": 1.7956,
      "step": 7970
    },
    {
      "epoch": 0.9378305323774827,
      "grad_norm": 7.752238750457764,
      "learning_rate": 3.4506626814016596e-05,
      "loss": 1.7341,
      "step": 7980
    },
    {
      "epoch": 0.9390057586085322,
      "grad_norm": 16.75479507446289,
      "learning_rate": 3.448696267746883e-05,
      "loss": 2.0786,
      "step": 7990
    },
    {
      "epoch": 0.9401809848395816,
      "grad_norm": 2.718301773071289,
      "learning_rate": 3.446729854092107e-05,
      "loss": 1.0756,
      "step": 8000
    },
    {
      "epoch": 0.941356211070631,
      "grad_norm": 6.539437294006348,
      "learning_rate": 3.444763440437331e-05,
      "loss": 2.251,
      "step": 8010
    },
    {
      "epoch": 0.9425314373016805,
      "grad_norm": 9.995023727416992,
      "learning_rate": 3.4427970267825544e-05,
      "loss": 1.8088,
      "step": 8020
    },
    {
      "epoch": 0.94370666353273,
      "grad_norm": 11.2057466506958,
      "learning_rate": 3.440830613127778e-05,
      "loss": 1.8641,
      "step": 8030
    },
    {
      "epoch": 0.9448818897637795,
      "grad_norm": 12.352863311767578,
      "learning_rate": 3.438864199473001e-05,
      "loss": 2.3051,
      "step": 8040
    },
    {
      "epoch": 0.946057115994829,
      "grad_norm": 21.35056495666504,
      "learning_rate": 3.436897785818225e-05,
      "loss": 1.9841,
      "step": 8050
    },
    {
      "epoch": 0.9472323422258785,
      "grad_norm": 20.196569442749023,
      "learning_rate": 3.4349313721634485e-05,
      "loss": 1.6266,
      "step": 8060
    },
    {
      "epoch": 0.948407568456928,
      "grad_norm": 11.761301040649414,
      "learning_rate": 3.432964958508672e-05,
      "loss": 1.806,
      "step": 8070
    },
    {
      "epoch": 0.9495827946879775,
      "grad_norm": 10.820778846740723,
      "learning_rate": 3.430998544853896e-05,
      "loss": 1.3514,
      "step": 8080
    },
    {
      "epoch": 0.950758020919027,
      "grad_norm": 9.011521339416504,
      "learning_rate": 3.429032131199119e-05,
      "loss": 1.7115,
      "step": 8090
    },
    {
      "epoch": 0.9519332471500764,
      "grad_norm": 14.391156196594238,
      "learning_rate": 3.4270657175443427e-05,
      "loss": 1.6256,
      "step": 8100
    },
    {
      "epoch": 0.9531084733811259,
      "grad_norm": 12.321035385131836,
      "learning_rate": 3.425099303889566e-05,
      "loss": 1.7963,
      "step": 8110
    },
    {
      "epoch": 0.9542836996121753,
      "grad_norm": 49.88715362548828,
      "learning_rate": 3.42313289023479e-05,
      "loss": 1.1451,
      "step": 8120
    },
    {
      "epoch": 0.9554589258432248,
      "grad_norm": 11.835763931274414,
      "learning_rate": 3.4211664765800134e-05,
      "loss": 2.1452,
      "step": 8130
    },
    {
      "epoch": 0.9566341520742743,
      "grad_norm": 26.56298065185547,
      "learning_rate": 3.419200062925237e-05,
      "loss": 1.7477,
      "step": 8140
    },
    {
      "epoch": 0.9578093783053238,
      "grad_norm": 13.018784523010254,
      "learning_rate": 3.417233649270461e-05,
      "loss": 1.7797,
      "step": 8150
    },
    {
      "epoch": 0.9589846045363732,
      "grad_norm": 10.046178817749023,
      "learning_rate": 3.415267235615684e-05,
      "loss": 1.7362,
      "step": 8160
    },
    {
      "epoch": 0.9601598307674227,
      "grad_norm": 21.945579528808594,
      "learning_rate": 3.413300821960908e-05,
      "loss": 1.7206,
      "step": 8170
    },
    {
      "epoch": 0.9613350569984722,
      "grad_norm": 13.408747673034668,
      "learning_rate": 3.4113344083061316e-05,
      "loss": 2.0201,
      "step": 8180
    },
    {
      "epoch": 0.9625102832295217,
      "grad_norm": 11.05411434173584,
      "learning_rate": 3.409367994651355e-05,
      "loss": 1.815,
      "step": 8190
    },
    {
      "epoch": 0.9636855094605712,
      "grad_norm": 7.720512866973877,
      "learning_rate": 3.407401580996579e-05,
      "loss": 1.5342,
      "step": 8200
    },
    {
      "epoch": 0.9648607356916207,
      "grad_norm": 7.947444438934326,
      "learning_rate": 3.4054351673418023e-05,
      "loss": 2.4741,
      "step": 8210
    },
    {
      "epoch": 0.9660359619226702,
      "grad_norm": 16.455705642700195,
      "learning_rate": 3.403468753687026e-05,
      "loss": 1.7176,
      "step": 8220
    },
    {
      "epoch": 0.9672111881537195,
      "grad_norm": 14.396491050720215,
      "learning_rate": 3.401502340032249e-05,
      "loss": 1.624,
      "step": 8230
    },
    {
      "epoch": 0.968386414384769,
      "grad_norm": 23.316326141357422,
      "learning_rate": 3.399535926377473e-05,
      "loss": 1.9179,
      "step": 8240
    },
    {
      "epoch": 0.9695616406158185,
      "grad_norm": 14.161599159240723,
      "learning_rate": 3.3975695127226965e-05,
      "loss": 1.9474,
      "step": 8250
    },
    {
      "epoch": 0.970736866846868,
      "grad_norm": 6.330300331115723,
      "learning_rate": 3.39560309906792e-05,
      "loss": 2.2934,
      "step": 8260
    },
    {
      "epoch": 0.9719120930779175,
      "grad_norm": 9.201351165771484,
      "learning_rate": 3.393636685413143e-05,
      "loss": 1.6233,
      "step": 8270
    },
    {
      "epoch": 0.973087319308967,
      "grad_norm": 13.16748046875,
      "learning_rate": 3.391670271758367e-05,
      "loss": 2.2625,
      "step": 8280
    },
    {
      "epoch": 0.9742625455400165,
      "grad_norm": 19.137828826904297,
      "learning_rate": 3.3897038581035906e-05,
      "loss": 1.8438,
      "step": 8290
    },
    {
      "epoch": 0.9754377717710659,
      "grad_norm": 14.49482250213623,
      "learning_rate": 3.3877374444488146e-05,
      "loss": 2.0851,
      "step": 8300
    },
    {
      "epoch": 0.9766129980021154,
      "grad_norm": 11.3511962890625,
      "learning_rate": 3.385771030794038e-05,
      "loss": 1.4165,
      "step": 8310
    },
    {
      "epoch": 0.9777882242331649,
      "grad_norm": 10.130046844482422,
      "learning_rate": 3.383804617139262e-05,
      "loss": 1.3352,
      "step": 8320
    },
    {
      "epoch": 0.9789634504642144,
      "grad_norm": 16.292701721191406,
      "learning_rate": 3.3818382034844854e-05,
      "loss": 1.0627,
      "step": 8330
    },
    {
      "epoch": 0.9801386766952639,
      "grad_norm": 21.106096267700195,
      "learning_rate": 3.379871789829709e-05,
      "loss": 2.1299,
      "step": 8340
    },
    {
      "epoch": 0.9813139029263133,
      "grad_norm": 18.578838348388672,
      "learning_rate": 3.377905376174932e-05,
      "loss": 2.2149,
      "step": 8350
    },
    {
      "epoch": 0.9824891291573628,
      "grad_norm": 10.297470092773438,
      "learning_rate": 3.375938962520156e-05,
      "loss": 1.8903,
      "step": 8360
    },
    {
      "epoch": 0.9836643553884122,
      "grad_norm": 7.811626434326172,
      "learning_rate": 3.3739725488653795e-05,
      "loss": 2.2534,
      "step": 8370
    },
    {
      "epoch": 0.9848395816194617,
      "grad_norm": 11.995902061462402,
      "learning_rate": 3.372006135210603e-05,
      "loss": 2.1206,
      "step": 8380
    },
    {
      "epoch": 0.9860148078505112,
      "grad_norm": 12.385534286499023,
      "learning_rate": 3.370039721555827e-05,
      "loss": 1.6271,
      "step": 8390
    },
    {
      "epoch": 0.9871900340815607,
      "grad_norm": 18.43326759338379,
      "learning_rate": 3.36807330790105e-05,
      "loss": 1.39,
      "step": 8400
    },
    {
      "epoch": 0.9883652603126102,
      "grad_norm": 8.971661567687988,
      "learning_rate": 3.3661068942462736e-05,
      "loss": 1.8636,
      "step": 8410
    },
    {
      "epoch": 0.9895404865436597,
      "grad_norm": 6.41738748550415,
      "learning_rate": 3.364140480591497e-05,
      "loss": 1.6004,
      "step": 8420
    },
    {
      "epoch": 0.9907157127747092,
      "grad_norm": 9.628888130187988,
      "learning_rate": 3.362174066936721e-05,
      "loss": 1.4881,
      "step": 8430
    },
    {
      "epoch": 0.9918909390057586,
      "grad_norm": 22.59670639038086,
      "learning_rate": 3.3602076532819444e-05,
      "loss": 2.1146,
      "step": 8440
    },
    {
      "epoch": 0.9930661652368081,
      "grad_norm": 21.891006469726562,
      "learning_rate": 3.358241239627168e-05,
      "loss": 2.0502,
      "step": 8450
    },
    {
      "epoch": 0.9942413914678576,
      "grad_norm": 19.647855758666992,
      "learning_rate": 3.356274825972392e-05,
      "loss": 1.7112,
      "step": 8460
    },
    {
      "epoch": 0.995416617698907,
      "grad_norm": 11.652597427368164,
      "learning_rate": 3.354308412317616e-05,
      "loss": 1.7854,
      "step": 8470
    },
    {
      "epoch": 0.9965918439299565,
      "grad_norm": 10.335673332214355,
      "learning_rate": 3.352341998662839e-05,
      "loss": 1.7222,
      "step": 8480
    },
    {
      "epoch": 0.997767070161006,
      "grad_norm": 24.548702239990234,
      "learning_rate": 3.3503755850080626e-05,
      "loss": 2.5009,
      "step": 8490
    },
    {
      "epoch": 0.9989422963920555,
      "grad_norm": 19.78380584716797,
      "learning_rate": 3.348409171353286e-05,
      "loss": 1.6356,
      "step": 8500
    },
    {
      "epoch": 1.000117522623105,
      "grad_norm": 7.633251190185547,
      "learning_rate": 3.34644275769851e-05,
      "loss": 1.4291,
      "step": 8510
    },
    {
      "epoch": 1.0012927488541545,
      "grad_norm": 16.673067092895508,
      "learning_rate": 3.344476344043733e-05,
      "loss": 1.8333,
      "step": 8520
    },
    {
      "epoch": 1.002467975085204,
      "grad_norm": 5.911933898925781,
      "learning_rate": 3.342509930388957e-05,
      "loss": 1.566,
      "step": 8530
    },
    {
      "epoch": 1.0036432013162533,
      "grad_norm": 13.988290786743164,
      "learning_rate": 3.34054351673418e-05,
      "loss": 1.8583,
      "step": 8540
    },
    {
      "epoch": 1.0048184275473029,
      "grad_norm": 13.316267967224121,
      "learning_rate": 3.338577103079404e-05,
      "loss": 1.9232,
      "step": 8550
    },
    {
      "epoch": 1.0059936537783523,
      "grad_norm": 28.31297492980957,
      "learning_rate": 3.3366106894246274e-05,
      "loss": 1.4981,
      "step": 8560
    },
    {
      "epoch": 1.0071688800094019,
      "grad_norm": 13.835394859313965,
      "learning_rate": 3.334644275769851e-05,
      "loss": 1.9763,
      "step": 8570
    },
    {
      "epoch": 1.0083441062404512,
      "grad_norm": 17.103633880615234,
      "learning_rate": 3.332677862115074e-05,
      "loss": 1.9026,
      "step": 8580
    },
    {
      "epoch": 1.0095193324715008,
      "grad_norm": 15.315032005310059,
      "learning_rate": 3.330711448460298e-05,
      "loss": 1.9677,
      "step": 8590
    },
    {
      "epoch": 1.0106945587025502,
      "grad_norm": 13.291214942932129,
      "learning_rate": 3.3287450348055216e-05,
      "loss": 1.3737,
      "step": 8600
    },
    {
      "epoch": 1.0118697849335998,
      "grad_norm": 14.781129837036133,
      "learning_rate": 3.3267786211507456e-05,
      "loss": 2.0993,
      "step": 8610
    },
    {
      "epoch": 1.0130450111646492,
      "grad_norm": 15.991440773010254,
      "learning_rate": 3.324812207495969e-05,
      "loss": 1.578,
      "step": 8620
    },
    {
      "epoch": 1.0142202373956988,
      "grad_norm": 35.000789642333984,
      "learning_rate": 3.322845793841193e-05,
      "loss": 1.4983,
      "step": 8630
    },
    {
      "epoch": 1.0153954636267482,
      "grad_norm": 8.103045463562012,
      "learning_rate": 3.3208793801864164e-05,
      "loss": 1.6347,
      "step": 8640
    },
    {
      "epoch": 1.0165706898577975,
      "grad_norm": 32.42625427246094,
      "learning_rate": 3.31891296653164e-05,
      "loss": 1.5202,
      "step": 8650
    },
    {
      "epoch": 1.0177459160888471,
      "grad_norm": 12.806891441345215,
      "learning_rate": 3.316946552876863e-05,
      "loss": 1.5205,
      "step": 8660
    },
    {
      "epoch": 1.0189211423198965,
      "grad_norm": 6.447507381439209,
      "learning_rate": 3.314980139222087e-05,
      "loss": 1.9935,
      "step": 8670
    },
    {
      "epoch": 1.020096368550946,
      "grad_norm": 10.268720626831055,
      "learning_rate": 3.3130137255673105e-05,
      "loss": 2.1116,
      "step": 8680
    },
    {
      "epoch": 1.0212715947819955,
      "grad_norm": 17.98395347595215,
      "learning_rate": 3.311047311912534e-05,
      "loss": 1.6187,
      "step": 8690
    },
    {
      "epoch": 1.022446821013045,
      "grad_norm": 7.717573642730713,
      "learning_rate": 3.309080898257758e-05,
      "loss": 1.3157,
      "step": 8700
    },
    {
      "epoch": 1.0236220472440944,
      "grad_norm": 11.056267738342285,
      "learning_rate": 3.307114484602981e-05,
      "loss": 1.339,
      "step": 8710
    },
    {
      "epoch": 1.024797273475144,
      "grad_norm": 7.4201788902282715,
      "learning_rate": 3.3051480709482046e-05,
      "loss": 2.2181,
      "step": 8720
    },
    {
      "epoch": 1.0259724997061934,
      "grad_norm": 10.011685371398926,
      "learning_rate": 3.303181657293428e-05,
      "loss": 1.6639,
      "step": 8730
    },
    {
      "epoch": 1.027147725937243,
      "grad_norm": 2.8415615558624268,
      "learning_rate": 3.301215243638652e-05,
      "loss": 1.8901,
      "step": 8740
    },
    {
      "epoch": 1.0283229521682924,
      "grad_norm": 16.809452056884766,
      "learning_rate": 3.2992488299838754e-05,
      "loss": 2.3991,
      "step": 8750
    },
    {
      "epoch": 1.0294981783993418,
      "grad_norm": 12.08823299407959,
      "learning_rate": 3.2972824163290994e-05,
      "loss": 1.7442,
      "step": 8760
    },
    {
      "epoch": 1.0306734046303914,
      "grad_norm": 11.400609016418457,
      "learning_rate": 3.295316002674323e-05,
      "loss": 1.2666,
      "step": 8770
    },
    {
      "epoch": 1.0318486308614407,
      "grad_norm": 16.443422317504883,
      "learning_rate": 3.293349589019547e-05,
      "loss": 1.7925,
      "step": 8780
    },
    {
      "epoch": 1.0330238570924903,
      "grad_norm": 20.24348258972168,
      "learning_rate": 3.29138317536477e-05,
      "loss": 1.9208,
      "step": 8790
    },
    {
      "epoch": 1.0341990833235397,
      "grad_norm": 22.907724380493164,
      "learning_rate": 3.2894167617099935e-05,
      "loss": 1.1833,
      "step": 8800
    },
    {
      "epoch": 1.0353743095545893,
      "grad_norm": 13.387781143188477,
      "learning_rate": 3.287450348055217e-05,
      "loss": 1.7993,
      "step": 8810
    },
    {
      "epoch": 1.0365495357856387,
      "grad_norm": 14.153489112854004,
      "learning_rate": 3.285483934400441e-05,
      "loss": 1.6204,
      "step": 8820
    },
    {
      "epoch": 1.0377247620166883,
      "grad_norm": 14.52529525756836,
      "learning_rate": 3.283517520745664e-05,
      "loss": 1.8451,
      "step": 8830
    },
    {
      "epoch": 1.0388999882477377,
      "grad_norm": 7.879727363586426,
      "learning_rate": 3.2815511070908877e-05,
      "loss": 1.2935,
      "step": 8840
    },
    {
      "epoch": 1.0400752144787873,
      "grad_norm": 15.405386924743652,
      "learning_rate": 3.279584693436111e-05,
      "loss": 1.5509,
      "step": 8850
    },
    {
      "epoch": 1.0412504407098366,
      "grad_norm": 13.574617385864258,
      "learning_rate": 3.277618279781335e-05,
      "loss": 1.1711,
      "step": 8860
    },
    {
      "epoch": 1.0424256669408862,
      "grad_norm": 8.476486206054688,
      "learning_rate": 3.2756518661265584e-05,
      "loss": 1.421,
      "step": 8870
    },
    {
      "epoch": 1.0436008931719356,
      "grad_norm": 19.405458450317383,
      "learning_rate": 3.273685452471782e-05,
      "loss": 1.4389,
      "step": 8880
    },
    {
      "epoch": 1.044776119402985,
      "grad_norm": 8.917885780334473,
      "learning_rate": 3.271719038817006e-05,
      "loss": 1.9867,
      "step": 8890
    },
    {
      "epoch": 1.0459513456340346,
      "grad_norm": 13.277791976928711,
      "learning_rate": 3.269752625162229e-05,
      "loss": 1.7656,
      "step": 8900
    },
    {
      "epoch": 1.047126571865084,
      "grad_norm": 15.581172943115234,
      "learning_rate": 3.267786211507453e-05,
      "loss": 1.5195,
      "step": 8910
    },
    {
      "epoch": 1.0483017980961336,
      "grad_norm": 8.242918014526367,
      "learning_rate": 3.2658197978526766e-05,
      "loss": 2.026,
      "step": 8920
    },
    {
      "epoch": 1.049477024327183,
      "grad_norm": 16.05451774597168,
      "learning_rate": 3.2638533841979e-05,
      "loss": 1.8604,
      "step": 8930
    },
    {
      "epoch": 1.0506522505582325,
      "grad_norm": 11.462225914001465,
      "learning_rate": 3.261886970543124e-05,
      "loss": 1.7959,
      "step": 8940
    },
    {
      "epoch": 1.051827476789282,
      "grad_norm": 24.69618034362793,
      "learning_rate": 3.2599205568883473e-05,
      "loss": 1.5907,
      "step": 8950
    },
    {
      "epoch": 1.0530027030203315,
      "grad_norm": 15.439364433288574,
      "learning_rate": 3.257954143233571e-05,
      "loss": 1.954,
      "step": 8960
    },
    {
      "epoch": 1.0541779292513809,
      "grad_norm": 9.437888145446777,
      "learning_rate": 3.255987729578794e-05,
      "loss": 1.8053,
      "step": 8970
    },
    {
      "epoch": 1.0553531554824305,
      "grad_norm": 15.613516807556152,
      "learning_rate": 3.254021315924018e-05,
      "loss": 1.5221,
      "step": 8980
    },
    {
      "epoch": 1.0565283817134798,
      "grad_norm": 7.121568202972412,
      "learning_rate": 3.2520549022692415e-05,
      "loss": 1.4874,
      "step": 8990
    },
    {
      "epoch": 1.0577036079445292,
      "grad_norm": 10.999530792236328,
      "learning_rate": 3.250088488614465e-05,
      "loss": 1.9325,
      "step": 9000
    },
    {
      "epoch": 1.0588788341755788,
      "grad_norm": 15.860424041748047,
      "learning_rate": 3.248122074959689e-05,
      "loss": 1.4912,
      "step": 9010
    },
    {
      "epoch": 1.0600540604066282,
      "grad_norm": 9.438472747802734,
      "learning_rate": 3.246155661304912e-05,
      "loss": 1.3883,
      "step": 9020
    },
    {
      "epoch": 1.0612292866376778,
      "grad_norm": 22.20499610900879,
      "learning_rate": 3.2441892476501356e-05,
      "loss": 2.3362,
      "step": 9030
    },
    {
      "epoch": 1.0624045128687272,
      "grad_norm": 9.049735069274902,
      "learning_rate": 3.242222833995359e-05,
      "loss": 2.0932,
      "step": 9040
    },
    {
      "epoch": 1.0635797390997768,
      "grad_norm": 17.562368392944336,
      "learning_rate": 3.240256420340583e-05,
      "loss": 1.6162,
      "step": 9050
    },
    {
      "epoch": 1.0647549653308261,
      "grad_norm": 16.84604263305664,
      "learning_rate": 3.238290006685807e-05,
      "loss": 1.982,
      "step": 9060
    },
    {
      "epoch": 1.0659301915618757,
      "grad_norm": 16.083879470825195,
      "learning_rate": 3.2363235930310304e-05,
      "loss": 1.5802,
      "step": 9070
    },
    {
      "epoch": 1.0671054177929251,
      "grad_norm": 10.831442832946777,
      "learning_rate": 3.234357179376254e-05,
      "loss": 1.807,
      "step": 9080
    },
    {
      "epoch": 1.0682806440239747,
      "grad_norm": 4.196460723876953,
      "learning_rate": 3.232390765721478e-05,
      "loss": 1.3687,
      "step": 9090
    },
    {
      "epoch": 1.069455870255024,
      "grad_norm": 18.51616859436035,
      "learning_rate": 3.230424352066701e-05,
      "loss": 1.7225,
      "step": 9100
    },
    {
      "epoch": 1.0706310964860735,
      "grad_norm": 11.079517364501953,
      "learning_rate": 3.2284579384119245e-05,
      "loss": 1.5762,
      "step": 9110
    },
    {
      "epoch": 1.071806322717123,
      "grad_norm": 12.071563720703125,
      "learning_rate": 3.226491524757148e-05,
      "loss": 1.5915,
      "step": 9120
    },
    {
      "epoch": 1.0729815489481724,
      "grad_norm": 12.658705711364746,
      "learning_rate": 3.224525111102372e-05,
      "loss": 1.524,
      "step": 9130
    },
    {
      "epoch": 1.074156775179222,
      "grad_norm": 10.971895217895508,
      "learning_rate": 3.222558697447595e-05,
      "loss": 1.9198,
      "step": 9140
    },
    {
      "epoch": 1.0753320014102714,
      "grad_norm": 27.991199493408203,
      "learning_rate": 3.2205922837928186e-05,
      "loss": 1.5523,
      "step": 9150
    },
    {
      "epoch": 1.076507227641321,
      "grad_norm": 23.932861328125,
      "learning_rate": 3.218625870138042e-05,
      "loss": 1.8826,
      "step": 9160
    },
    {
      "epoch": 1.0776824538723704,
      "grad_norm": 20.17858123779297,
      "learning_rate": 3.216659456483266e-05,
      "loss": 1.4709,
      "step": 9170
    },
    {
      "epoch": 1.07885768010342,
      "grad_norm": 11.156436920166016,
      "learning_rate": 3.2146930428284894e-05,
      "loss": 1.9763,
      "step": 9180
    },
    {
      "epoch": 1.0800329063344694,
      "grad_norm": 11.870014190673828,
      "learning_rate": 3.212726629173713e-05,
      "loss": 1.5361,
      "step": 9190
    },
    {
      "epoch": 1.081208132565519,
      "grad_norm": 15.334145545959473,
      "learning_rate": 3.210760215518937e-05,
      "loss": 1.0656,
      "step": 9200
    },
    {
      "epoch": 1.0823833587965683,
      "grad_norm": 8.23559284210205,
      "learning_rate": 3.208793801864161e-05,
      "loss": 1.0918,
      "step": 9210
    },
    {
      "epoch": 1.083558585027618,
      "grad_norm": 15.648090362548828,
      "learning_rate": 3.206827388209384e-05,
      "loss": 1.8592,
      "step": 9220
    },
    {
      "epoch": 1.0847338112586673,
      "grad_norm": 12.1079683303833,
      "learning_rate": 3.2048609745546076e-05,
      "loss": 1.4451,
      "step": 9230
    },
    {
      "epoch": 1.0859090374897167,
      "grad_norm": 7.538222312927246,
      "learning_rate": 3.202894560899831e-05,
      "loss": 1.8093,
      "step": 9240
    },
    {
      "epoch": 1.0870842637207663,
      "grad_norm": 12.342171669006348,
      "learning_rate": 3.200928147245055e-05,
      "loss": 1.2932,
      "step": 9250
    },
    {
      "epoch": 1.0882594899518157,
      "grad_norm": 24.150184631347656,
      "learning_rate": 3.198961733590278e-05,
      "loss": 2.7266,
      "step": 9260
    },
    {
      "epoch": 1.0894347161828652,
      "grad_norm": 21.519460678100586,
      "learning_rate": 3.196995319935502e-05,
      "loss": 2.3006,
      "step": 9270
    },
    {
      "epoch": 1.0906099424139146,
      "grad_norm": 16.667022705078125,
      "learning_rate": 3.195028906280725e-05,
      "loss": 1.7944,
      "step": 9280
    },
    {
      "epoch": 1.0917851686449642,
      "grad_norm": 17.039274215698242,
      "learning_rate": 3.193062492625949e-05,
      "loss": 1.659,
      "step": 9290
    },
    {
      "epoch": 1.0929603948760136,
      "grad_norm": 7.543942451477051,
      "learning_rate": 3.1910960789711724e-05,
      "loss": 1.7059,
      "step": 9300
    },
    {
      "epoch": 1.0941356211070632,
      "grad_norm": 19.46615982055664,
      "learning_rate": 3.189129665316396e-05,
      "loss": 1.3762,
      "step": 9310
    },
    {
      "epoch": 1.0953108473381126,
      "grad_norm": 19.95938491821289,
      "learning_rate": 3.18716325166162e-05,
      "loss": 1.207,
      "step": 9320
    },
    {
      "epoch": 1.0964860735691622,
      "grad_norm": 12.89750862121582,
      "learning_rate": 3.185196838006843e-05,
      "loss": 1.1446,
      "step": 9330
    },
    {
      "epoch": 1.0976612998002115,
      "grad_norm": 17.641231536865234,
      "learning_rate": 3.1832304243520666e-05,
      "loss": 2.2663,
      "step": 9340
    },
    {
      "epoch": 1.098836526031261,
      "grad_norm": 8.434355735778809,
      "learning_rate": 3.18126401069729e-05,
      "loss": 1.5361,
      "step": 9350
    },
    {
      "epoch": 1.1000117522623105,
      "grad_norm": 7.725466728210449,
      "learning_rate": 3.179297597042514e-05,
      "loss": 1.5998,
      "step": 9360
    },
    {
      "epoch": 1.10118697849336,
      "grad_norm": 16.729501724243164,
      "learning_rate": 3.177331183387738e-05,
      "loss": 1.456,
      "step": 9370
    },
    {
      "epoch": 1.1023622047244095,
      "grad_norm": 17.492849349975586,
      "learning_rate": 3.1753647697329614e-05,
      "loss": 1.9339,
      "step": 9380
    },
    {
      "epoch": 1.1035374309554589,
      "grad_norm": 17.314899444580078,
      "learning_rate": 3.173398356078185e-05,
      "loss": 1.2945,
      "step": 9390
    },
    {
      "epoch": 1.1047126571865085,
      "grad_norm": 14.540234565734863,
      "learning_rate": 3.171431942423409e-05,
      "loss": 1.5871,
      "step": 9400
    },
    {
      "epoch": 1.1058878834175578,
      "grad_norm": 10.265088081359863,
      "learning_rate": 3.169465528768632e-05,
      "loss": 1.5463,
      "step": 9410
    },
    {
      "epoch": 1.1070631096486074,
      "grad_norm": 14.604482650756836,
      "learning_rate": 3.1674991151138555e-05,
      "loss": 1.7485,
      "step": 9420
    },
    {
      "epoch": 1.1082383358796568,
      "grad_norm": 33.51980209350586,
      "learning_rate": 3.165532701459079e-05,
      "loss": 2.0108,
      "step": 9430
    },
    {
      "epoch": 1.1094135621107064,
      "grad_norm": 10.578944206237793,
      "learning_rate": 3.163566287804303e-05,
      "loss": 1.2337,
      "step": 9440
    },
    {
      "epoch": 1.1105887883417558,
      "grad_norm": 32.85029983520508,
      "learning_rate": 3.161599874149526e-05,
      "loss": 1.6906,
      "step": 9450
    },
    {
      "epoch": 1.1117640145728052,
      "grad_norm": 11.604002952575684,
      "learning_rate": 3.1596334604947496e-05,
      "loss": 1.9124,
      "step": 9460
    },
    {
      "epoch": 1.1129392408038548,
      "grad_norm": 11.175519943237305,
      "learning_rate": 3.157667046839973e-05,
      "loss": 1.459,
      "step": 9470
    },
    {
      "epoch": 1.1141144670349041,
      "grad_norm": 10.370555877685547,
      "learning_rate": 3.155700633185197e-05,
      "loss": 1.6281,
      "step": 9480
    },
    {
      "epoch": 1.1152896932659537,
      "grad_norm": 14.46310806274414,
      "learning_rate": 3.1537342195304204e-05,
      "loss": 1.6966,
      "step": 9490
    },
    {
      "epoch": 1.116464919497003,
      "grad_norm": 20.346025466918945,
      "learning_rate": 3.151767805875644e-05,
      "loss": 1.6716,
      "step": 9500
    },
    {
      "epoch": 1.1176401457280527,
      "grad_norm": 20.49203109741211,
      "learning_rate": 3.149801392220868e-05,
      "loss": 1.2062,
      "step": 9510
    },
    {
      "epoch": 1.118815371959102,
      "grad_norm": 35.252315521240234,
      "learning_rate": 3.147834978566092e-05,
      "loss": 1.6774,
      "step": 9520
    },
    {
      "epoch": 1.1199905981901517,
      "grad_norm": 9.530450820922852,
      "learning_rate": 3.145868564911315e-05,
      "loss": 1.652,
      "step": 9530
    },
    {
      "epoch": 1.121165824421201,
      "grad_norm": 23.44917869567871,
      "learning_rate": 3.1439021512565385e-05,
      "loss": 1.7731,
      "step": 9540
    },
    {
      "epoch": 1.1223410506522506,
      "grad_norm": 12.394129753112793,
      "learning_rate": 3.141935737601762e-05,
      "loss": 1.4452,
      "step": 9550
    },
    {
      "epoch": 1.1235162768833,
      "grad_norm": 4.232680320739746,
      "learning_rate": 3.139969323946986e-05,
      "loss": 1.6413,
      "step": 9560
    },
    {
      "epoch": 1.1246915031143496,
      "grad_norm": 28.501487731933594,
      "learning_rate": 3.138002910292209e-05,
      "loss": 1.5199,
      "step": 9570
    },
    {
      "epoch": 1.125866729345399,
      "grad_norm": 12.053088188171387,
      "learning_rate": 3.1360364966374327e-05,
      "loss": 1.5042,
      "step": 9580
    },
    {
      "epoch": 1.1270419555764484,
      "grad_norm": 11.8872709274292,
      "learning_rate": 3.134070082982657e-05,
      "loss": 2.1124,
      "step": 9590
    },
    {
      "epoch": 1.128217181807498,
      "grad_norm": 26.899749755859375,
      "learning_rate": 3.13210366932788e-05,
      "loss": 1.3828,
      "step": 9600
    },
    {
      "epoch": 1.1293924080385473,
      "grad_norm": 13.684866905212402,
      "learning_rate": 3.1301372556731034e-05,
      "loss": 2.1172,
      "step": 9610
    },
    {
      "epoch": 1.130567634269597,
      "grad_norm": 30.923133850097656,
      "learning_rate": 3.128170842018327e-05,
      "loss": 1.6613,
      "step": 9620
    },
    {
      "epoch": 1.1317428605006463,
      "grad_norm": 13.553707122802734,
      "learning_rate": 3.126204428363551e-05,
      "loss": 1.6346,
      "step": 9630
    },
    {
      "epoch": 1.132918086731696,
      "grad_norm": 21.608627319335938,
      "learning_rate": 3.124238014708774e-05,
      "loss": 1.6359,
      "step": 9640
    },
    {
      "epoch": 1.1340933129627453,
      "grad_norm": 24.2108211517334,
      "learning_rate": 3.1222716010539975e-05,
      "loss": 1.482,
      "step": 9650
    },
    {
      "epoch": 1.135268539193795,
      "grad_norm": 6.733027458190918,
      "learning_rate": 3.120305187399221e-05,
      "loss": 1.975,
      "step": 9660
    },
    {
      "epoch": 1.1364437654248443,
      "grad_norm": 11.967201232910156,
      "learning_rate": 3.118338773744445e-05,
      "loss": 2.0499,
      "step": 9670
    },
    {
      "epoch": 1.1376189916558936,
      "grad_norm": 21.841163635253906,
      "learning_rate": 3.116372360089669e-05,
      "loss": 1.5706,
      "step": 9680
    },
    {
      "epoch": 1.1387942178869432,
      "grad_norm": 19.665653228759766,
      "learning_rate": 3.114405946434892e-05,
      "loss": 1.9141,
      "step": 9690
    },
    {
      "epoch": 1.1399694441179928,
      "grad_norm": 17.275066375732422,
      "learning_rate": 3.112439532780116e-05,
      "loss": 1.3497,
      "step": 9700
    },
    {
      "epoch": 1.1411446703490422,
      "grad_norm": 6.0438127517700195,
      "learning_rate": 3.11047311912534e-05,
      "loss": 1.517,
      "step": 9710
    },
    {
      "epoch": 1.1423198965800916,
      "grad_norm": 15.279958724975586,
      "learning_rate": 3.108506705470563e-05,
      "loss": 1.2473,
      "step": 9720
    },
    {
      "epoch": 1.1434951228111412,
      "grad_norm": 9.485876083374023,
      "learning_rate": 3.1065402918157865e-05,
      "loss": 1.1942,
      "step": 9730
    },
    {
      "epoch": 1.1446703490421906,
      "grad_norm": 18.725139617919922,
      "learning_rate": 3.10457387816101e-05,
      "loss": 1.2881,
      "step": 9740
    },
    {
      "epoch": 1.1458455752732402,
      "grad_norm": 6.750670909881592,
      "learning_rate": 3.102607464506234e-05,
      "loss": 1.3174,
      "step": 9750
    },
    {
      "epoch": 1.1470208015042895,
      "grad_norm": 11.29848861694336,
      "learning_rate": 3.100641050851457e-05,
      "loss": 1.4555,
      "step": 9760
    },
    {
      "epoch": 1.1481960277353391,
      "grad_norm": 22.285888671875,
      "learning_rate": 3.0986746371966806e-05,
      "loss": 1.5435,
      "step": 9770
    },
    {
      "epoch": 1.1493712539663885,
      "grad_norm": 7.087962627410889,
      "learning_rate": 3.096708223541904e-05,
      "loss": 1.3701,
      "step": 9780
    },
    {
      "epoch": 1.150546480197438,
      "grad_norm": 21.888755798339844,
      "learning_rate": 3.094741809887128e-05,
      "loss": 1.5071,
      "step": 9790
    },
    {
      "epoch": 1.1517217064284875,
      "grad_norm": 21.903913497924805,
      "learning_rate": 3.0927753962323513e-05,
      "loss": 1.8417,
      "step": 9800
    },
    {
      "epoch": 1.1528969326595369,
      "grad_norm": 8.12430191040039,
      "learning_rate": 3.090808982577575e-05,
      "loss": 1.3568,
      "step": 9810
    },
    {
      "epoch": 1.1540721588905865,
      "grad_norm": 11.24582290649414,
      "learning_rate": 3.088842568922799e-05,
      "loss": 1.3931,
      "step": 9820
    },
    {
      "epoch": 1.1552473851216358,
      "grad_norm": 10.200719833374023,
      "learning_rate": 3.086876155268023e-05,
      "loss": 1.5439,
      "step": 9830
    },
    {
      "epoch": 1.1564226113526854,
      "grad_norm": 18.570350646972656,
      "learning_rate": 3.084909741613246e-05,
      "loss": 2.0587,
      "step": 9840
    },
    {
      "epoch": 1.1575978375837348,
      "grad_norm": 10.726289749145508,
      "learning_rate": 3.0829433279584695e-05,
      "loss": 1.765,
      "step": 9850
    },
    {
      "epoch": 1.1587730638147844,
      "grad_norm": 6.840005397796631,
      "learning_rate": 3.080976914303693e-05,
      "loss": 1.6969,
      "step": 9860
    },
    {
      "epoch": 1.1599482900458338,
      "grad_norm": 15.4313325881958,
      "learning_rate": 3.079010500648917e-05,
      "loss": 2.0614,
      "step": 9870
    },
    {
      "epoch": 1.1611235162768834,
      "grad_norm": 20.29742431640625,
      "learning_rate": 3.07704408699414e-05,
      "loss": 1.9913,
      "step": 9880
    },
    {
      "epoch": 1.1622987425079327,
      "grad_norm": 17.89223861694336,
      "learning_rate": 3.0750776733393636e-05,
      "loss": 1.591,
      "step": 9890
    },
    {
      "epoch": 1.1634739687389823,
      "grad_norm": 11.24035930633545,
      "learning_rate": 3.073111259684588e-05,
      "loss": 1.4877,
      "step": 9900
    },
    {
      "epoch": 1.1646491949700317,
      "grad_norm": 7.7438154220581055,
      "learning_rate": 3.071144846029811e-05,
      "loss": 1.6969,
      "step": 9910
    },
    {
      "epoch": 1.1658244212010813,
      "grad_norm": 17.16554832458496,
      "learning_rate": 3.0691784323750344e-05,
      "loss": 1.5224,
      "step": 9920
    },
    {
      "epoch": 1.1669996474321307,
      "grad_norm": 16.246496200561523,
      "learning_rate": 3.067212018720258e-05,
      "loss": 2.0272,
      "step": 9930
    },
    {
      "epoch": 1.16817487366318,
      "grad_norm": 6.815975666046143,
      "learning_rate": 3.065245605065482e-05,
      "loss": 1.5942,
      "step": 9940
    },
    {
      "epoch": 1.1693500998942297,
      "grad_norm": 8.373842239379883,
      "learning_rate": 3.063279191410705e-05,
      "loss": 1.5577,
      "step": 9950
    },
    {
      "epoch": 1.170525326125279,
      "grad_norm": 16.945247650146484,
      "learning_rate": 3.0613127777559285e-05,
      "loss": 2.1771,
      "step": 9960
    },
    {
      "epoch": 1.1717005523563286,
      "grad_norm": 15.885165214538574,
      "learning_rate": 3.0593463641011526e-05,
      "loss": 1.6732,
      "step": 9970
    },
    {
      "epoch": 1.172875778587378,
      "grad_norm": 16.409339904785156,
      "learning_rate": 3.0573799504463766e-05,
      "loss": 1.6044,
      "step": 9980
    },
    {
      "epoch": 1.1740510048184276,
      "grad_norm": 14.912236213684082,
      "learning_rate": 3.0554135367916e-05,
      "loss": 1.8232,
      "step": 9990
    },
    {
      "epoch": 1.175226231049477,
      "grad_norm": 7.842496871948242,
      "learning_rate": 3.053447123136823e-05,
      "loss": 1.378,
      "step": 10000
    },
    {
      "epoch": 1.1764014572805266,
      "grad_norm": 13.350945472717285,
      "learning_rate": 3.051480709482047e-05,
      "loss": 2.1851,
      "step": 10010
    },
    {
      "epoch": 1.177576683511576,
      "grad_norm": 19.09027671813965,
      "learning_rate": 3.0495142958272704e-05,
      "loss": 1.897,
      "step": 10020
    },
    {
      "epoch": 1.1787519097426253,
      "grad_norm": 8.98206901550293,
      "learning_rate": 3.047547882172494e-05,
      "loss": 1.0577,
      "step": 10030
    },
    {
      "epoch": 1.179927135973675,
      "grad_norm": 16.400882720947266,
      "learning_rate": 3.0455814685177174e-05,
      "loss": 1.5016,
      "step": 10040
    },
    {
      "epoch": 1.1811023622047245,
      "grad_norm": 19.694469451904297,
      "learning_rate": 3.043615054862941e-05,
      "loss": 1.9226,
      "step": 10050
    },
    {
      "epoch": 1.182277588435774,
      "grad_norm": 8.965801239013672,
      "learning_rate": 3.0416486412081645e-05,
      "loss": 1.5219,
      "step": 10060
    },
    {
      "epoch": 1.1834528146668233,
      "grad_norm": 10.990447044372559,
      "learning_rate": 3.0396822275533882e-05,
      "loss": 1.5979,
      "step": 10070
    },
    {
      "epoch": 1.1846280408978729,
      "grad_norm": 5.833795547485352,
      "learning_rate": 3.0377158138986116e-05,
      "loss": 1.262,
      "step": 10080
    },
    {
      "epoch": 1.1858032671289223,
      "grad_norm": 17.268335342407227,
      "learning_rate": 3.0357494002438353e-05,
      "loss": 2.0172,
      "step": 10090
    },
    {
      "epoch": 1.1869784933599719,
      "grad_norm": 12.686847686767578,
      "learning_rate": 3.033782986589059e-05,
      "loss": 1.6204,
      "step": 10100
    },
    {
      "epoch": 1.1881537195910212,
      "grad_norm": 10.791407585144043,
      "learning_rate": 3.0318165729342823e-05,
      "loss": 2.0798,
      "step": 10110
    },
    {
      "epoch": 1.1893289458220708,
      "grad_norm": 13.062466621398926,
      "learning_rate": 3.0298501592795064e-05,
      "loss": 1.9742,
      "step": 10120
    },
    {
      "epoch": 1.1905041720531202,
      "grad_norm": 14.931533813476562,
      "learning_rate": 3.02788374562473e-05,
      "loss": 1.9059,
      "step": 10130
    },
    {
      "epoch": 1.1916793982841698,
      "grad_norm": 25.55913734436035,
      "learning_rate": 3.0259173319699534e-05,
      "loss": 1.8614,
      "step": 10140
    },
    {
      "epoch": 1.1928546245152192,
      "grad_norm": 21.991771697998047,
      "learning_rate": 3.023950918315177e-05,
      "loss": 1.1465,
      "step": 10150
    },
    {
      "epoch": 1.1940298507462686,
      "grad_norm": 9.893953323364258,
      "learning_rate": 3.0219845046604005e-05,
      "loss": 1.6788,
      "step": 10160
    },
    {
      "epoch": 1.1952050769773181,
      "grad_norm": 18.44815444946289,
      "learning_rate": 3.0200180910056242e-05,
      "loss": 1.7188,
      "step": 10170
    },
    {
      "epoch": 1.1963803032083675,
      "grad_norm": 21.98198890686035,
      "learning_rate": 3.0180516773508475e-05,
      "loss": 1.6935,
      "step": 10180
    },
    {
      "epoch": 1.1975555294394171,
      "grad_norm": 14.602834701538086,
      "learning_rate": 3.0160852636960712e-05,
      "loss": 1.9277,
      "step": 10190
    },
    {
      "epoch": 1.1987307556704665,
      "grad_norm": 11.346553802490234,
      "learning_rate": 3.014118850041295e-05,
      "loss": 1.6842,
      "step": 10200
    },
    {
      "epoch": 1.199905981901516,
      "grad_norm": 16.093719482421875,
      "learning_rate": 3.0121524363865183e-05,
      "loss": 1.3344,
      "step": 10210
    },
    {
      "epoch": 1.2010812081325655,
      "grad_norm": 12.272793769836426,
      "learning_rate": 3.010186022731742e-05,
      "loss": 1.4712,
      "step": 10220
    },
    {
      "epoch": 1.202256434363615,
      "grad_norm": 47.763710021972656,
      "learning_rate": 3.0082196090769654e-05,
      "loss": 1.8345,
      "step": 10230
    },
    {
      "epoch": 1.2034316605946644,
      "grad_norm": 20.56479835510254,
      "learning_rate": 3.006253195422189e-05,
      "loss": 1.5603,
      "step": 10240
    },
    {
      "epoch": 1.204606886825714,
      "grad_norm": 15.349221229553223,
      "learning_rate": 3.0042867817674124e-05,
      "loss": 1.3459,
      "step": 10250
    },
    {
      "epoch": 1.2057821130567634,
      "grad_norm": 22.957475662231445,
      "learning_rate": 3.002320368112636e-05,
      "loss": 1.8409,
      "step": 10260
    },
    {
      "epoch": 1.206957339287813,
      "grad_norm": 16.750507354736328,
      "learning_rate": 3.00035395445786e-05,
      "loss": 1.966,
      "step": 10270
    },
    {
      "epoch": 1.2081325655188624,
      "grad_norm": 5.620093822479248,
      "learning_rate": 2.998387540803084e-05,
      "loss": 1.6243,
      "step": 10280
    },
    {
      "epoch": 1.2093077917499118,
      "grad_norm": 10.097622871398926,
      "learning_rate": 2.9964211271483072e-05,
      "loss": 1.7129,
      "step": 10290
    },
    {
      "epoch": 1.2104830179809614,
      "grad_norm": 12.2550048828125,
      "learning_rate": 2.994454713493531e-05,
      "loss": 1.6498,
      "step": 10300
    },
    {
      "epoch": 1.2116582442120107,
      "grad_norm": 16.44074821472168,
      "learning_rate": 2.9924882998387543e-05,
      "loss": 1.9741,
      "step": 10310
    },
    {
      "epoch": 1.2128334704430603,
      "grad_norm": 15.880932807922363,
      "learning_rate": 2.990521886183978e-05,
      "loss": 1.9894,
      "step": 10320
    },
    {
      "epoch": 1.2140086966741097,
      "grad_norm": 11.306940078735352,
      "learning_rate": 2.9885554725292014e-05,
      "loss": 1.7162,
      "step": 10330
    },
    {
      "epoch": 1.2151839229051593,
      "grad_norm": 15.715415000915527,
      "learning_rate": 2.986589058874425e-05,
      "loss": 1.3689,
      "step": 10340
    },
    {
      "epoch": 1.2163591491362087,
      "grad_norm": 5.703200817108154,
      "learning_rate": 2.9846226452196484e-05,
      "loss": 1.3531,
      "step": 10350
    },
    {
      "epoch": 1.2175343753672583,
      "grad_norm": 11.76525592803955,
      "learning_rate": 2.982656231564872e-05,
      "loss": 1.8485,
      "step": 10360
    },
    {
      "epoch": 1.2187096015983077,
      "grad_norm": 40.04643249511719,
      "learning_rate": 2.9806898179100955e-05,
      "loss": 1.3974,
      "step": 10370
    },
    {
      "epoch": 1.219884827829357,
      "grad_norm": 21.44671630859375,
      "learning_rate": 2.9787234042553192e-05,
      "loss": 1.4258,
      "step": 10380
    },
    {
      "epoch": 1.2210600540604066,
      "grad_norm": 16.455663681030273,
      "learning_rate": 2.9767569906005425e-05,
      "loss": 1.7827,
      "step": 10390
    },
    {
      "epoch": 1.2222352802914562,
      "grad_norm": 21.763500213623047,
      "learning_rate": 2.9747905769457662e-05,
      "loss": 2.1632,
      "step": 10400
    },
    {
      "epoch": 1.2234105065225056,
      "grad_norm": 13.408509254455566,
      "learning_rate": 2.97282416329099e-05,
      "loss": 1.3295,
      "step": 10410
    },
    {
      "epoch": 1.224585732753555,
      "grad_norm": 7.751506328582764,
      "learning_rate": 2.970857749636214e-05,
      "loss": 1.4951,
      "step": 10420
    },
    {
      "epoch": 1.2257609589846046,
      "grad_norm": 4.732470989227295,
      "learning_rate": 2.9688913359814373e-05,
      "loss": 1.9422,
      "step": 10430
    },
    {
      "epoch": 1.226936185215654,
      "grad_norm": 27.843048095703125,
      "learning_rate": 2.966924922326661e-05,
      "loss": 1.9141,
      "step": 10440
    },
    {
      "epoch": 1.2281114114467035,
      "grad_norm": 15.264243125915527,
      "learning_rate": 2.9649585086718844e-05,
      "loss": 1.5225,
      "step": 10450
    },
    {
      "epoch": 1.229286637677753,
      "grad_norm": 18.483558654785156,
      "learning_rate": 2.962992095017108e-05,
      "loss": 1.6281,
      "step": 10460
    },
    {
      "epoch": 1.2304618639088025,
      "grad_norm": 18.642486572265625,
      "learning_rate": 2.9610256813623315e-05,
      "loss": 1.8881,
      "step": 10470
    },
    {
      "epoch": 1.231637090139852,
      "grad_norm": 21.213855743408203,
      "learning_rate": 2.959059267707555e-05,
      "loss": 1.38,
      "step": 10480
    },
    {
      "epoch": 1.2328123163709015,
      "grad_norm": 19.340030670166016,
      "learning_rate": 2.957092854052779e-05,
      "loss": 1.8643,
      "step": 10490
    },
    {
      "epoch": 1.2339875426019509,
      "grad_norm": 19.339702606201172,
      "learning_rate": 2.9551264403980022e-05,
      "loss": 2.3757,
      "step": 10500
    },
    {
      "epoch": 1.2351627688330002,
      "grad_norm": 24.619678497314453,
      "learning_rate": 2.953160026743226e-05,
      "loss": 2.302,
      "step": 10510
    },
    {
      "epoch": 1.2363379950640498,
      "grad_norm": 40.03661346435547,
      "learning_rate": 2.9511936130884493e-05,
      "loss": 1.4433,
      "step": 10520
    },
    {
      "epoch": 1.2375132212950992,
      "grad_norm": 12.887860298156738,
      "learning_rate": 2.949227199433673e-05,
      "loss": 1.6952,
      "step": 10530
    },
    {
      "epoch": 1.2386884475261488,
      "grad_norm": 13.865290641784668,
      "learning_rate": 2.9472607857788963e-05,
      "loss": 1.1667,
      "step": 10540
    },
    {
      "epoch": 1.2398636737571982,
      "grad_norm": 8.59559154510498,
      "learning_rate": 2.94529437212412e-05,
      "loss": 1.4445,
      "step": 10550
    },
    {
      "epoch": 1.2410388999882478,
      "grad_norm": 13.10847282409668,
      "learning_rate": 2.9433279584693434e-05,
      "loss": 1.3535,
      "step": 10560
    },
    {
      "epoch": 1.2422141262192972,
      "grad_norm": 7.520936489105225,
      "learning_rate": 2.9413615448145674e-05,
      "loss": 1.4389,
      "step": 10570
    },
    {
      "epoch": 1.2433893524503468,
      "grad_norm": 12.720386505126953,
      "learning_rate": 2.939395131159791e-05,
      "loss": 1.6293,
      "step": 10580
    },
    {
      "epoch": 1.2445645786813961,
      "grad_norm": 6.743661880493164,
      "learning_rate": 2.937428717505015e-05,
      "loss": 1.5624,
      "step": 10590
    },
    {
      "epoch": 1.2457398049124457,
      "grad_norm": 12.581682205200195,
      "learning_rate": 2.9354623038502382e-05,
      "loss": 1.8334,
      "step": 10600
    },
    {
      "epoch": 1.2469150311434951,
      "grad_norm": 12.594382286071777,
      "learning_rate": 2.933495890195462e-05,
      "loss": 2.6453,
      "step": 10610
    },
    {
      "epoch": 1.2480902573745447,
      "grad_norm": 10.529570579528809,
      "learning_rate": 2.9315294765406853e-05,
      "loss": 2.2885,
      "step": 10620
    },
    {
      "epoch": 1.249265483605594,
      "grad_norm": 15.088995933532715,
      "learning_rate": 2.929563062885909e-05,
      "loss": 1.8121,
      "step": 10630
    },
    {
      "epoch": 1.2504407098366435,
      "grad_norm": 10.779454231262207,
      "learning_rate": 2.9275966492311323e-05,
      "loss": 1.6448,
      "step": 10640
    },
    {
      "epoch": 1.251615936067693,
      "grad_norm": 8.102433204650879,
      "learning_rate": 2.925630235576356e-05,
      "loss": 1.5787,
      "step": 10650
    },
    {
      "epoch": 1.2527911622987424,
      "grad_norm": 10.337316513061523,
      "learning_rate": 2.9236638219215794e-05,
      "loss": 1.2562,
      "step": 10660
    },
    {
      "epoch": 1.253966388529792,
      "grad_norm": 9.84537124633789,
      "learning_rate": 2.921697408266803e-05,
      "loss": 1.4704,
      "step": 10670
    },
    {
      "epoch": 1.2551416147608414,
      "grad_norm": 6.742969989776611,
      "learning_rate": 2.9197309946120265e-05,
      "loss": 1.3159,
      "step": 10680
    },
    {
      "epoch": 1.256316840991891,
      "grad_norm": 9.811360359191895,
      "learning_rate": 2.91776458095725e-05,
      "loss": 1.5932,
      "step": 10690
    },
    {
      "epoch": 1.2574920672229404,
      "grad_norm": 21.67335319519043,
      "learning_rate": 2.9157981673024735e-05,
      "loss": 1.9533,
      "step": 10700
    },
    {
      "epoch": 1.25866729345399,
      "grad_norm": 14.68276596069336,
      "learning_rate": 2.9138317536476972e-05,
      "loss": 1.6327,
      "step": 10710
    },
    {
      "epoch": 1.2598425196850394,
      "grad_norm": 17.152000427246094,
      "learning_rate": 2.9118653399929213e-05,
      "loss": 1.9298,
      "step": 10720
    },
    {
      "epoch": 1.2610177459160887,
      "grad_norm": 12.899577140808105,
      "learning_rate": 2.909898926338145e-05,
      "loss": 2.2694,
      "step": 10730
    },
    {
      "epoch": 1.2621929721471383,
      "grad_norm": 7.623504638671875,
      "learning_rate": 2.9079325126833683e-05,
      "loss": 1.3075,
      "step": 10740
    },
    {
      "epoch": 1.263368198378188,
      "grad_norm": 20.25750160217285,
      "learning_rate": 2.905966099028592e-05,
      "loss": 1.5144,
      "step": 10750
    },
    {
      "epoch": 1.2645434246092373,
      "grad_norm": 18.027326583862305,
      "learning_rate": 2.9039996853738154e-05,
      "loss": 2.1891,
      "step": 10760
    },
    {
      "epoch": 1.2657186508402867,
      "grad_norm": 12.910726547241211,
      "learning_rate": 2.902033271719039e-05,
      "loss": 1.5096,
      "step": 10770
    },
    {
      "epoch": 1.2668938770713363,
      "grad_norm": 13.475858688354492,
      "learning_rate": 2.9000668580642624e-05,
      "loss": 2.3956,
      "step": 10780
    },
    {
      "epoch": 1.2680691033023856,
      "grad_norm": 18.097627639770508,
      "learning_rate": 2.898100444409486e-05,
      "loss": 1.7843,
      "step": 10790
    },
    {
      "epoch": 1.2692443295334352,
      "grad_norm": 8.213051795959473,
      "learning_rate": 2.89613403075471e-05,
      "loss": 1.4197,
      "step": 10800
    },
    {
      "epoch": 1.2704195557644846,
      "grad_norm": 10.997740745544434,
      "learning_rate": 2.8941676170999332e-05,
      "loss": 2.1979,
      "step": 10810
    },
    {
      "epoch": 1.2715947819955342,
      "grad_norm": 18.352188110351562,
      "learning_rate": 2.892201203445157e-05,
      "loss": 1.571,
      "step": 10820
    },
    {
      "epoch": 1.2727700082265836,
      "grad_norm": 20.82025909423828,
      "learning_rate": 2.8902347897903803e-05,
      "loss": 1.5274,
      "step": 10830
    },
    {
      "epoch": 1.2739452344576332,
      "grad_norm": 4.904001235961914,
      "learning_rate": 2.888268376135604e-05,
      "loss": 1.3398,
      "step": 10840
    },
    {
      "epoch": 1.2751204606886826,
      "grad_norm": 7.401713848114014,
      "learning_rate": 2.8863019624808273e-05,
      "loss": 1.6887,
      "step": 10850
    },
    {
      "epoch": 1.276295686919732,
      "grad_norm": 8.023131370544434,
      "learning_rate": 2.884335548826051e-05,
      "loss": 1.8176,
      "step": 10860
    },
    {
      "epoch": 1.2774709131507815,
      "grad_norm": 19.0070858001709,
      "learning_rate": 2.8823691351712744e-05,
      "loss": 1.6316,
      "step": 10870
    },
    {
      "epoch": 1.2786461393818311,
      "grad_norm": 16.888206481933594,
      "learning_rate": 2.8804027215164988e-05,
      "loss": 1.5952,
      "step": 10880
    },
    {
      "epoch": 1.2798213656128805,
      "grad_norm": 10.436646461486816,
      "learning_rate": 2.878436307861722e-05,
      "loss": 1.1888,
      "step": 10890
    },
    {
      "epoch": 1.2809965918439299,
      "grad_norm": 13.902933120727539,
      "learning_rate": 2.8764698942069458e-05,
      "loss": 1.8916,
      "step": 10900
    },
    {
      "epoch": 1.2821718180749795,
      "grad_norm": 11.719964981079102,
      "learning_rate": 2.8745034805521692e-05,
      "loss": 1.5703,
      "step": 10910
    },
    {
      "epoch": 1.2833470443060289,
      "grad_norm": 7.391235828399658,
      "learning_rate": 2.872537066897393e-05,
      "loss": 1.6301,
      "step": 10920
    },
    {
      "epoch": 1.2845222705370785,
      "grad_norm": 14.870738983154297,
      "learning_rate": 2.8705706532426162e-05,
      "loss": 1.6293,
      "step": 10930
    },
    {
      "epoch": 1.2856974967681278,
      "grad_norm": 14.76219654083252,
      "learning_rate": 2.86860423958784e-05,
      "loss": 1.3957,
      "step": 10940
    },
    {
      "epoch": 1.2868727229991772,
      "grad_norm": 39.41603469848633,
      "learning_rate": 2.8666378259330633e-05,
      "loss": 2.2503,
      "step": 10950
    },
    {
      "epoch": 1.2880479492302268,
      "grad_norm": 15.04417610168457,
      "learning_rate": 2.864671412278287e-05,
      "loss": 1.5057,
      "step": 10960
    },
    {
      "epoch": 1.2892231754612764,
      "grad_norm": 10.144439697265625,
      "learning_rate": 2.8627049986235104e-05,
      "loss": 1.4647,
      "step": 10970
    },
    {
      "epoch": 1.2903984016923258,
      "grad_norm": 15.25263500213623,
      "learning_rate": 2.860738584968734e-05,
      "loss": 1.8707,
      "step": 10980
    },
    {
      "epoch": 1.2915736279233752,
      "grad_norm": 15.393763542175293,
      "learning_rate": 2.8587721713139574e-05,
      "loss": 2.0513,
      "step": 10990
    },
    {
      "epoch": 1.2927488541544248,
      "grad_norm": 44.43406295776367,
      "learning_rate": 2.856805757659181e-05,
      "loss": 2.0493,
      "step": 11000
    },
    {
      "epoch": 1.2939240803854741,
      "grad_norm": 21.40506362915039,
      "learning_rate": 2.8548393440044048e-05,
      "loss": 1.6574,
      "step": 11010
    },
    {
      "epoch": 1.2950993066165237,
      "grad_norm": 16.593446731567383,
      "learning_rate": 2.8528729303496282e-05,
      "loss": 1.396,
      "step": 11020
    },
    {
      "epoch": 1.296274532847573,
      "grad_norm": 23.45464324951172,
      "learning_rate": 2.8509065166948522e-05,
      "loss": 1.827,
      "step": 11030
    },
    {
      "epoch": 1.2974497590786227,
      "grad_norm": 18.41194725036621,
      "learning_rate": 2.848940103040076e-05,
      "loss": 1.6061,
      "step": 11040
    },
    {
      "epoch": 1.298624985309672,
      "grad_norm": 11.302017211914062,
      "learning_rate": 2.8469736893852993e-05,
      "loss": 1.6177,
      "step": 11050
    },
    {
      "epoch": 1.2998002115407217,
      "grad_norm": 7.359001159667969,
      "learning_rate": 2.845007275730523e-05,
      "loss": 1.1775,
      "step": 11060
    },
    {
      "epoch": 1.300975437771771,
      "grad_norm": 34.93105697631836,
      "learning_rate": 2.8430408620757464e-05,
      "loss": 1.9443,
      "step": 11070
    },
    {
      "epoch": 1.3021506640028204,
      "grad_norm": 7.073626518249512,
      "learning_rate": 2.84107444842097e-05,
      "loss": 1.616,
      "step": 11080
    },
    {
      "epoch": 1.30332589023387,
      "grad_norm": 28.067960739135742,
      "learning_rate": 2.8391080347661934e-05,
      "loss": 1.43,
      "step": 11090
    },
    {
      "epoch": 1.3045011164649196,
      "grad_norm": 18.274351119995117,
      "learning_rate": 2.837141621111417e-05,
      "loss": 1.9731,
      "step": 11100
    },
    {
      "epoch": 1.305676342695969,
      "grad_norm": 3.91739559173584,
      "learning_rate": 2.8351752074566408e-05,
      "loss": 1.1769,
      "step": 11110
    },
    {
      "epoch": 1.3068515689270184,
      "grad_norm": 11.410920143127441,
      "learning_rate": 2.8332087938018642e-05,
      "loss": 1.4209,
      "step": 11120
    },
    {
      "epoch": 1.308026795158068,
      "grad_norm": 16.41654396057129,
      "learning_rate": 2.831242380147088e-05,
      "loss": 1.4325,
      "step": 11130
    },
    {
      "epoch": 1.3092020213891173,
      "grad_norm": 22.583492279052734,
      "learning_rate": 2.8292759664923112e-05,
      "loss": 1.4805,
      "step": 11140
    },
    {
      "epoch": 1.310377247620167,
      "grad_norm": 12.792104721069336,
      "learning_rate": 2.827309552837535e-05,
      "loss": 1.5124,
      "step": 11150
    },
    {
      "epoch": 1.3115524738512163,
      "grad_norm": 23.891878128051758,
      "learning_rate": 2.8253431391827583e-05,
      "loss": 1.9824,
      "step": 11160
    },
    {
      "epoch": 1.312727700082266,
      "grad_norm": 13.915809631347656,
      "learning_rate": 2.823376725527982e-05,
      "loss": 1.4199,
      "step": 11170
    },
    {
      "epoch": 1.3139029263133153,
      "grad_norm": 21.967241287231445,
      "learning_rate": 2.821410311873206e-05,
      "loss": 1.8737,
      "step": 11180
    },
    {
      "epoch": 1.3150781525443649,
      "grad_norm": 11.210026741027832,
      "learning_rate": 2.8194438982184297e-05,
      "loss": 1.634,
      "step": 11190
    },
    {
      "epoch": 1.3162533787754143,
      "grad_norm": 13.203349113464355,
      "learning_rate": 2.817477484563653e-05,
      "loss": 1.7694,
      "step": 11200
    },
    {
      "epoch": 1.3174286050064636,
      "grad_norm": 21.821521759033203,
      "learning_rate": 2.8155110709088768e-05,
      "loss": 1.6038,
      "step": 11210
    },
    {
      "epoch": 1.3186038312375132,
      "grad_norm": 14.20639419555664,
      "learning_rate": 2.8135446572541e-05,
      "loss": 1.4799,
      "step": 11220
    },
    {
      "epoch": 1.3197790574685628,
      "grad_norm": 18.438119888305664,
      "learning_rate": 2.811578243599324e-05,
      "loss": 1.4592,
      "step": 11230
    },
    {
      "epoch": 1.3209542836996122,
      "grad_norm": 12.034936904907227,
      "learning_rate": 2.8096118299445472e-05,
      "loss": 1.5322,
      "step": 11240
    },
    {
      "epoch": 1.3221295099306616,
      "grad_norm": 20.794017791748047,
      "learning_rate": 2.807645416289771e-05,
      "loss": 1.5476,
      "step": 11250
    },
    {
      "epoch": 1.3233047361617112,
      "grad_norm": 13.454011917114258,
      "learning_rate": 2.8056790026349943e-05,
      "loss": 1.8171,
      "step": 11260
    },
    {
      "epoch": 1.3244799623927606,
      "grad_norm": 16.644969940185547,
      "learning_rate": 2.803712588980218e-05,
      "loss": 1.614,
      "step": 11270
    },
    {
      "epoch": 1.3256551886238102,
      "grad_norm": 12.984349250793457,
      "learning_rate": 2.8017461753254413e-05,
      "loss": 1.2156,
      "step": 11280
    },
    {
      "epoch": 1.3268304148548595,
      "grad_norm": 13.433834075927734,
      "learning_rate": 2.799779761670665e-05,
      "loss": 1.2416,
      "step": 11290
    },
    {
      "epoch": 1.328005641085909,
      "grad_norm": 11.27765941619873,
      "learning_rate": 2.7978133480158884e-05,
      "loss": 1.9615,
      "step": 11300
    },
    {
      "epoch": 1.3291808673169585,
      "grad_norm": 12.597355842590332,
      "learning_rate": 2.795846934361112e-05,
      "loss": 1.365,
      "step": 11310
    },
    {
      "epoch": 1.330356093548008,
      "grad_norm": 16.65771484375,
      "learning_rate": 2.7938805207063358e-05,
      "loss": 1.574,
      "step": 11320
    },
    {
      "epoch": 1.3315313197790575,
      "grad_norm": 9.136650085449219,
      "learning_rate": 2.79191410705156e-05,
      "loss": 1.5685,
      "step": 11330
    },
    {
      "epoch": 1.3327065460101069,
      "grad_norm": 28.485050201416016,
      "learning_rate": 2.7899476933967832e-05,
      "loss": 2.4021,
      "step": 11340
    },
    {
      "epoch": 1.3338817722411564,
      "grad_norm": 27.05727767944336,
      "learning_rate": 2.787981279742007e-05,
      "loss": 1.3678,
      "step": 11350
    },
    {
      "epoch": 1.3350569984722058,
      "grad_norm": 12.853751182556152,
      "learning_rate": 2.7860148660872303e-05,
      "loss": 1.4137,
      "step": 11360
    },
    {
      "epoch": 1.3362322247032554,
      "grad_norm": 16.161550521850586,
      "learning_rate": 2.784048452432454e-05,
      "loss": 1.516,
      "step": 11370
    },
    {
      "epoch": 1.3374074509343048,
      "grad_norm": 16.6884822845459,
      "learning_rate": 2.7820820387776773e-05,
      "loss": 2.1731,
      "step": 11380
    },
    {
      "epoch": 1.3385826771653544,
      "grad_norm": 17.38615608215332,
      "learning_rate": 2.780115625122901e-05,
      "loss": 1.2421,
      "step": 11390
    },
    {
      "epoch": 1.3397579033964038,
      "grad_norm": 35.974266052246094,
      "learning_rate": 2.7781492114681247e-05,
      "loss": 1.7398,
      "step": 11400
    },
    {
      "epoch": 1.3409331296274534,
      "grad_norm": 35.52827453613281,
      "learning_rate": 2.776182797813348e-05,
      "loss": 1.4708,
      "step": 11410
    },
    {
      "epoch": 1.3421083558585027,
      "grad_norm": 11.246984481811523,
      "learning_rate": 2.7742163841585718e-05,
      "loss": 1.3595,
      "step": 11420
    },
    {
      "epoch": 1.3432835820895521,
      "grad_norm": 4.109337329864502,
      "learning_rate": 2.772249970503795e-05,
      "loss": 1.3878,
      "step": 11430
    },
    {
      "epoch": 1.3444588083206017,
      "grad_norm": 22.100448608398438,
      "learning_rate": 2.770283556849019e-05,
      "loss": 1.1783,
      "step": 11440
    },
    {
      "epoch": 1.3456340345516513,
      "grad_norm": 5.256526470184326,
      "learning_rate": 2.7683171431942422e-05,
      "loss": 1.3884,
      "step": 11450
    },
    {
      "epoch": 1.3468092607827007,
      "grad_norm": 14.826306343078613,
      "learning_rate": 2.766350729539466e-05,
      "loss": 1.4932,
      "step": 11460
    },
    {
      "epoch": 1.34798448701375,
      "grad_norm": 13.343628883361816,
      "learning_rate": 2.7643843158846893e-05,
      "loss": 1.8487,
      "step": 11470
    },
    {
      "epoch": 1.3491597132447997,
      "grad_norm": 15.960487365722656,
      "learning_rate": 2.7624179022299133e-05,
      "loss": 1.8487,
      "step": 11480
    },
    {
      "epoch": 1.350334939475849,
      "grad_norm": 13.073427200317383,
      "learning_rate": 2.760451488575137e-05,
      "loss": 1.1779,
      "step": 11490
    },
    {
      "epoch": 1.3515101657068986,
      "grad_norm": 8.945001602172852,
      "learning_rate": 2.7584850749203607e-05,
      "loss": 1.7049,
      "step": 11500
    },
    {
      "epoch": 1.352685391937948,
      "grad_norm": 3.323681116104126,
      "learning_rate": 2.756518661265584e-05,
      "loss": 1.4807,
      "step": 11510
    },
    {
      "epoch": 1.3538606181689976,
      "grad_norm": 10.061297416687012,
      "learning_rate": 2.7545522476108078e-05,
      "loss": 1.4452,
      "step": 11520
    },
    {
      "epoch": 1.355035844400047,
      "grad_norm": 11.02230453491211,
      "learning_rate": 2.752585833956031e-05,
      "loss": 1.9173,
      "step": 11530
    },
    {
      "epoch": 1.3562110706310966,
      "grad_norm": 23.263782501220703,
      "learning_rate": 2.750619420301255e-05,
      "loss": 1.7209,
      "step": 11540
    },
    {
      "epoch": 1.357386296862146,
      "grad_norm": 3.8383655548095703,
      "learning_rate": 2.7486530066464782e-05,
      "loss": 1.6008,
      "step": 11550
    },
    {
      "epoch": 1.3585615230931953,
      "grad_norm": 17.214651107788086,
      "learning_rate": 2.746686592991702e-05,
      "loss": 1.4539,
      "step": 11560
    },
    {
      "epoch": 1.359736749324245,
      "grad_norm": 13.11422061920166,
      "learning_rate": 2.7447201793369253e-05,
      "loss": 1.8063,
      "step": 11570
    },
    {
      "epoch": 1.3609119755552945,
      "grad_norm": 12.491333961486816,
      "learning_rate": 2.742753765682149e-05,
      "loss": 1.3478,
      "step": 11580
    },
    {
      "epoch": 1.362087201786344,
      "grad_norm": 23.632139205932617,
      "learning_rate": 2.7407873520273723e-05,
      "loss": 1.5068,
      "step": 11590
    },
    {
      "epoch": 1.3632624280173933,
      "grad_norm": 19.87590789794922,
      "learning_rate": 2.738820938372596e-05,
      "loss": 1.4263,
      "step": 11600
    },
    {
      "epoch": 1.3644376542484429,
      "grad_norm": 14.566510200500488,
      "learning_rate": 2.7368545247178197e-05,
      "loss": 1.5126,
      "step": 11610
    },
    {
      "epoch": 1.3656128804794923,
      "grad_norm": 26.74365997314453,
      "learning_rate": 2.734888111063043e-05,
      "loss": 2.1939,
      "step": 11620
    },
    {
      "epoch": 1.3667881067105419,
      "grad_norm": 29.511722564697266,
      "learning_rate": 2.732921697408267e-05,
      "loss": 1.699,
      "step": 11630
    },
    {
      "epoch": 1.3679633329415912,
      "grad_norm": 17.60441780090332,
      "learning_rate": 2.7309552837534908e-05,
      "loss": 1.9884,
      "step": 11640
    },
    {
      "epoch": 1.3691385591726406,
      "grad_norm": 18.107030868530273,
      "learning_rate": 2.7289888700987142e-05,
      "loss": 1.9354,
      "step": 11650
    },
    {
      "epoch": 1.3703137854036902,
      "grad_norm": 20.413179397583008,
      "learning_rate": 2.727022456443938e-05,
      "loss": 1.8564,
      "step": 11660
    },
    {
      "epoch": 1.3714890116347398,
      "grad_norm": 9.7215576171875,
      "learning_rate": 2.7250560427891612e-05,
      "loss": 2.0338,
      "step": 11670
    },
    {
      "epoch": 1.3726642378657892,
      "grad_norm": 18.73110580444336,
      "learning_rate": 2.723089629134385e-05,
      "loss": 2.0567,
      "step": 11680
    },
    {
      "epoch": 1.3738394640968385,
      "grad_norm": 21.407812118530273,
      "learning_rate": 2.7211232154796083e-05,
      "loss": 1.8884,
      "step": 11690
    },
    {
      "epoch": 1.3750146903278881,
      "grad_norm": 12.092113494873047,
      "learning_rate": 2.719156801824832e-05,
      "loss": 1.9791,
      "step": 11700
    },
    {
      "epoch": 1.3761899165589375,
      "grad_norm": 9.511990547180176,
      "learning_rate": 2.7171903881700557e-05,
      "loss": 1.6854,
      "step": 11710
    },
    {
      "epoch": 1.3773651427899871,
      "grad_norm": 19.363487243652344,
      "learning_rate": 2.715223974515279e-05,
      "loss": 1.7347,
      "step": 11720
    },
    {
      "epoch": 1.3785403690210365,
      "grad_norm": 10.41131591796875,
      "learning_rate": 2.7132575608605028e-05,
      "loss": 1.7772,
      "step": 11730
    },
    {
      "epoch": 1.379715595252086,
      "grad_norm": 20.067142486572266,
      "learning_rate": 2.711291147205726e-05,
      "loss": 1.436,
      "step": 11740
    },
    {
      "epoch": 1.3808908214831355,
      "grad_norm": 7.165789604187012,
      "learning_rate": 2.7093247335509498e-05,
      "loss": 1.3496,
      "step": 11750
    },
    {
      "epoch": 1.382066047714185,
      "grad_norm": 15.298054695129395,
      "learning_rate": 2.7073583198961732e-05,
      "loss": 1.8534,
      "step": 11760
    },
    {
      "epoch": 1.3832412739452344,
      "grad_norm": 14.954198837280273,
      "learning_rate": 2.705391906241397e-05,
      "loss": 1.5001,
      "step": 11770
    },
    {
      "epoch": 1.3844165001762838,
      "grad_norm": 15.984830856323242,
      "learning_rate": 2.703425492586621e-05,
      "loss": 2.2389,
      "step": 11780
    },
    {
      "epoch": 1.3855917264073334,
      "grad_norm": 14.65196418762207,
      "learning_rate": 2.7014590789318446e-05,
      "loss": 1.5263,
      "step": 11790
    },
    {
      "epoch": 1.386766952638383,
      "grad_norm": 9.925978660583496,
      "learning_rate": 2.699492665277068e-05,
      "loss": 1.4592,
      "step": 11800
    },
    {
      "epoch": 1.3879421788694324,
      "grad_norm": 18.53803062438965,
      "learning_rate": 2.6975262516222917e-05,
      "loss": 1.7071,
      "step": 11810
    },
    {
      "epoch": 1.3891174051004818,
      "grad_norm": 12.787564277648926,
      "learning_rate": 2.695559837967515e-05,
      "loss": 1.9074,
      "step": 11820
    },
    {
      "epoch": 1.3902926313315314,
      "grad_norm": 6.842711925506592,
      "learning_rate": 2.6935934243127388e-05,
      "loss": 1.5795,
      "step": 11830
    },
    {
      "epoch": 1.3914678575625807,
      "grad_norm": 10.73719310760498,
      "learning_rate": 2.691627010657962e-05,
      "loss": 1.9239,
      "step": 11840
    },
    {
      "epoch": 1.3926430837936303,
      "grad_norm": 14.094656944274902,
      "learning_rate": 2.6896605970031858e-05,
      "loss": 1.1461,
      "step": 11850
    },
    {
      "epoch": 1.3938183100246797,
      "grad_norm": 21.94192886352539,
      "learning_rate": 2.6876941833484092e-05,
      "loss": 1.4102,
      "step": 11860
    },
    {
      "epoch": 1.3949935362557293,
      "grad_norm": 9.719986915588379,
      "learning_rate": 2.685727769693633e-05,
      "loss": 1.3618,
      "step": 11870
    },
    {
      "epoch": 1.3961687624867787,
      "grad_norm": 21.814210891723633,
      "learning_rate": 2.6837613560388562e-05,
      "loss": 1.4127,
      "step": 11880
    },
    {
      "epoch": 1.3973439887178283,
      "grad_norm": 11.344050407409668,
      "learning_rate": 2.68179494238408e-05,
      "loss": 1.6845,
      "step": 11890
    },
    {
      "epoch": 1.3985192149488777,
      "grad_norm": 10.19782543182373,
      "learning_rate": 2.6798285287293033e-05,
      "loss": 1.853,
      "step": 11900
    },
    {
      "epoch": 1.399694441179927,
      "grad_norm": 22.600738525390625,
      "learning_rate": 2.677862115074527e-05,
      "loss": 2.1349,
      "step": 11910
    },
    {
      "epoch": 1.4008696674109766,
      "grad_norm": 8.658480644226074,
      "learning_rate": 2.6758957014197507e-05,
      "loss": 1.1937,
      "step": 11920
    },
    {
      "epoch": 1.4020448936420262,
      "grad_norm": 10.846475601196289,
      "learning_rate": 2.6739292877649747e-05,
      "loss": 1.6806,
      "step": 11930
    },
    {
      "epoch": 1.4032201198730756,
      "grad_norm": 22.422481536865234,
      "learning_rate": 2.671962874110198e-05,
      "loss": 1.2754,
      "step": 11940
    },
    {
      "epoch": 1.404395346104125,
      "grad_norm": 13.107995986938477,
      "learning_rate": 2.6699964604554218e-05,
      "loss": 1.4549,
      "step": 11950
    },
    {
      "epoch": 1.4055705723351746,
      "grad_norm": 9.212376594543457,
      "learning_rate": 2.668030046800645e-05,
      "loss": 1.6583,
      "step": 11960
    },
    {
      "epoch": 1.406745798566224,
      "grad_norm": 20.780282974243164,
      "learning_rate": 2.666063633145869e-05,
      "loss": 1.9136,
      "step": 11970
    },
    {
      "epoch": 1.4079210247972735,
      "grad_norm": 20.260360717773438,
      "learning_rate": 2.6640972194910922e-05,
      "loss": 1.8627,
      "step": 11980
    },
    {
      "epoch": 1.409096251028323,
      "grad_norm": 10.590041160583496,
      "learning_rate": 2.662130805836316e-05,
      "loss": 2.1112,
      "step": 11990
    },
    {
      "epoch": 1.4102714772593723,
      "grad_norm": 24.28062629699707,
      "learning_rate": 2.6601643921815396e-05,
      "loss": 1.5464,
      "step": 12000
    },
    {
      "epoch": 1.411446703490422,
      "grad_norm": 11.526665687561035,
      "learning_rate": 2.658197978526763e-05,
      "loss": 1.407,
      "step": 12010
    },
    {
      "epoch": 1.4126219297214715,
      "grad_norm": 18.705289840698242,
      "learning_rate": 2.6562315648719867e-05,
      "loss": 1.816,
      "step": 12020
    },
    {
      "epoch": 1.4137971559525209,
      "grad_norm": 16.151981353759766,
      "learning_rate": 2.65426515121721e-05,
      "loss": 1.4469,
      "step": 12030
    },
    {
      "epoch": 1.4149723821835702,
      "grad_norm": 12.263604164123535,
      "learning_rate": 2.6522987375624337e-05,
      "loss": 1.4589,
      "step": 12040
    },
    {
      "epoch": 1.4161476084146198,
      "grad_norm": 15.88338565826416,
      "learning_rate": 2.650332323907657e-05,
      "loss": 1.4833,
      "step": 12050
    },
    {
      "epoch": 1.4173228346456692,
      "grad_norm": 9.695067405700684,
      "learning_rate": 2.6483659102528808e-05,
      "loss": 1.5602,
      "step": 12060
    },
    {
      "epoch": 1.4184980608767188,
      "grad_norm": 8.823823928833008,
      "learning_rate": 2.646399496598104e-05,
      "loss": 1.4829,
      "step": 12070
    },
    {
      "epoch": 1.4196732871077682,
      "grad_norm": 5.984291076660156,
      "learning_rate": 2.644433082943328e-05,
      "loss": 1.6807,
      "step": 12080
    },
    {
      "epoch": 1.4208485133388178,
      "grad_norm": 17.793188095092773,
      "learning_rate": 2.642466669288552e-05,
      "loss": 1.7093,
      "step": 12090
    },
    {
      "epoch": 1.4220237395698672,
      "grad_norm": 8.471606254577637,
      "learning_rate": 2.6405002556337756e-05,
      "loss": 2.0336,
      "step": 12100
    },
    {
      "epoch": 1.4231989658009168,
      "grad_norm": 11.574275970458984,
      "learning_rate": 2.638533841978999e-05,
      "loss": 1.3078,
      "step": 12110
    },
    {
      "epoch": 1.4243741920319661,
      "grad_norm": 10.063852310180664,
      "learning_rate": 2.6365674283242227e-05,
      "loss": 1.2002,
      "step": 12120
    },
    {
      "epoch": 1.4255494182630155,
      "grad_norm": 8.697098731994629,
      "learning_rate": 2.634601014669446e-05,
      "loss": 1.791,
      "step": 12130
    },
    {
      "epoch": 1.426724644494065,
      "grad_norm": 15.511042594909668,
      "learning_rate": 2.6326346010146697e-05,
      "loss": 1.811,
      "step": 12140
    },
    {
      "epoch": 1.4278998707251147,
      "grad_norm": 20.964218139648438,
      "learning_rate": 2.630668187359893e-05,
      "loss": 1.9513,
      "step": 12150
    },
    {
      "epoch": 1.429075096956164,
      "grad_norm": 22.208057403564453,
      "learning_rate": 2.6287017737051168e-05,
      "loss": 1.4826,
      "step": 12160
    },
    {
      "epoch": 1.4302503231872135,
      "grad_norm": 10.322023391723633,
      "learning_rate": 2.62673536005034e-05,
      "loss": 1.6293,
      "step": 12170
    },
    {
      "epoch": 1.431425549418263,
      "grad_norm": 7.919378757476807,
      "learning_rate": 2.624768946395564e-05,
      "loss": 1.5432,
      "step": 12180
    },
    {
      "epoch": 1.4326007756493124,
      "grad_norm": 12.551556587219238,
      "learning_rate": 2.6228025327407872e-05,
      "loss": 1.8793,
      "step": 12190
    },
    {
      "epoch": 1.433776001880362,
      "grad_norm": 20.13787841796875,
      "learning_rate": 2.620836119086011e-05,
      "loss": 2.0877,
      "step": 12200
    },
    {
      "epoch": 1.4349512281114114,
      "grad_norm": 26.490598678588867,
      "learning_rate": 2.6188697054312343e-05,
      "loss": 1.9576,
      "step": 12210
    },
    {
      "epoch": 1.436126454342461,
      "grad_norm": 19.447477340698242,
      "learning_rate": 2.616903291776458e-05,
      "loss": 1.2867,
      "step": 12220
    },
    {
      "epoch": 1.4373016805735104,
      "grad_norm": 10.030505180358887,
      "learning_rate": 2.6149368781216817e-05,
      "loss": 1.4494,
      "step": 12230
    },
    {
      "epoch": 1.43847690680456,
      "grad_norm": 12.01306438446045,
      "learning_rate": 2.6129704644669057e-05,
      "loss": 1.7195,
      "step": 12240
    },
    {
      "epoch": 1.4396521330356093,
      "grad_norm": 13.508225440979004,
      "learning_rate": 2.611004050812129e-05,
      "loss": 1.5414,
      "step": 12250
    },
    {
      "epoch": 1.4408273592666587,
      "grad_norm": 18.896499633789062,
      "learning_rate": 2.6090376371573528e-05,
      "loss": 1.8461,
      "step": 12260
    },
    {
      "epoch": 1.4420025854977083,
      "grad_norm": 16.027692794799805,
      "learning_rate": 2.607071223502576e-05,
      "loss": 1.8178,
      "step": 12270
    },
    {
      "epoch": 1.443177811728758,
      "grad_norm": 11.644453048706055,
      "learning_rate": 2.6051048098478e-05,
      "loss": 1.5323,
      "step": 12280
    },
    {
      "epoch": 1.4443530379598073,
      "grad_norm": 18.29025650024414,
      "learning_rate": 2.6031383961930232e-05,
      "loss": 1.7769,
      "step": 12290
    },
    {
      "epoch": 1.4455282641908567,
      "grad_norm": 16.142608642578125,
      "learning_rate": 2.601171982538247e-05,
      "loss": 1.943,
      "step": 12300
    },
    {
      "epoch": 1.4467034904219063,
      "grad_norm": 13.189383506774902,
      "learning_rate": 2.5992055688834706e-05,
      "loss": 1.5323,
      "step": 12310
    },
    {
      "epoch": 1.4478787166529556,
      "grad_norm": 23.666675567626953,
      "learning_rate": 2.597239155228694e-05,
      "loss": 1.7304,
      "step": 12320
    },
    {
      "epoch": 1.4490539428840052,
      "grad_norm": 16.899362564086914,
      "learning_rate": 2.5952727415739177e-05,
      "loss": 1.9551,
      "step": 12330
    },
    {
      "epoch": 1.4502291691150546,
      "grad_norm": 12.799444198608398,
      "learning_rate": 2.593306327919141e-05,
      "loss": 1.3639,
      "step": 12340
    },
    {
      "epoch": 1.451404395346104,
      "grad_norm": 5.0937910079956055,
      "learning_rate": 2.5913399142643647e-05,
      "loss": 1.7204,
      "step": 12350
    },
    {
      "epoch": 1.4525796215771536,
      "grad_norm": 12.78626823425293,
      "learning_rate": 2.589373500609588e-05,
      "loss": 1.6681,
      "step": 12360
    },
    {
      "epoch": 1.4537548478082032,
      "grad_norm": 8.447613716125488,
      "learning_rate": 2.5874070869548118e-05,
      "loss": 1.341,
      "step": 12370
    },
    {
      "epoch": 1.4549300740392526,
      "grad_norm": 13.878530502319336,
      "learning_rate": 2.585440673300035e-05,
      "loss": 1.5001,
      "step": 12380
    },
    {
      "epoch": 1.456105300270302,
      "grad_norm": 24.446868896484375,
      "learning_rate": 2.5834742596452595e-05,
      "loss": 1.7306,
      "step": 12390
    },
    {
      "epoch": 1.4572805265013515,
      "grad_norm": 21.080106735229492,
      "learning_rate": 2.581507845990483e-05,
      "loss": 1.7578,
      "step": 12400
    },
    {
      "epoch": 1.458455752732401,
      "grad_norm": 12.290726661682129,
      "learning_rate": 2.5795414323357066e-05,
      "loss": 1.5363,
      "step": 12410
    },
    {
      "epoch": 1.4596309789634505,
      "grad_norm": 6.670886993408203,
      "learning_rate": 2.57757501868093e-05,
      "loss": 1.4716,
      "step": 12420
    },
    {
      "epoch": 1.4608062051944999,
      "grad_norm": 11.207724571228027,
      "learning_rate": 2.5756086050261536e-05,
      "loss": 1.4057,
      "step": 12430
    },
    {
      "epoch": 1.4619814314255495,
      "grad_norm": 15.84840202331543,
      "learning_rate": 2.573642191371377e-05,
      "loss": 1.2932,
      "step": 12440
    },
    {
      "epoch": 1.4631566576565989,
      "grad_norm": 14.29973316192627,
      "learning_rate": 2.5716757777166007e-05,
      "loss": 1.3691,
      "step": 12450
    },
    {
      "epoch": 1.4643318838876485,
      "grad_norm": 13.607476234436035,
      "learning_rate": 2.569709364061824e-05,
      "loss": 1.3977,
      "step": 12460
    },
    {
      "epoch": 1.4655071101186978,
      "grad_norm": 7.207940578460693,
      "learning_rate": 2.5677429504070478e-05,
      "loss": 1.5189,
      "step": 12470
    },
    {
      "epoch": 1.4666823363497472,
      "grad_norm": 21.874744415283203,
      "learning_rate": 2.565776536752271e-05,
      "loss": 1.4948,
      "step": 12480
    },
    {
      "epoch": 1.4678575625807968,
      "grad_norm": 16.37178611755371,
      "learning_rate": 2.5638101230974948e-05,
      "loss": 1.8882,
      "step": 12490
    },
    {
      "epoch": 1.4690327888118464,
      "grad_norm": 13.950419425964355,
      "learning_rate": 2.5618437094427182e-05,
      "loss": 1.548,
      "step": 12500
    },
    {
      "epoch": 1.4702080150428958,
      "grad_norm": 12.794694900512695,
      "learning_rate": 2.559877295787942e-05,
      "loss": 1.9551,
      "step": 12510
    },
    {
      "epoch": 1.4713832412739452,
      "grad_norm": 11.154829978942871,
      "learning_rate": 2.5579108821331656e-05,
      "loss": 1.4028,
      "step": 12520
    },
    {
      "epoch": 1.4725584675049948,
      "grad_norm": 15.43144416809082,
      "learning_rate": 2.555944468478389e-05,
      "loss": 1.6826,
      "step": 12530
    },
    {
      "epoch": 1.4737336937360441,
      "grad_norm": 9.548750877380371,
      "learning_rate": 2.553978054823613e-05,
      "loss": 1.5637,
      "step": 12540
    },
    {
      "epoch": 1.4749089199670937,
      "grad_norm": 12.563139915466309,
      "learning_rate": 2.5520116411688367e-05,
      "loss": 1.6204,
      "step": 12550
    },
    {
      "epoch": 1.476084146198143,
      "grad_norm": 16.54888153076172,
      "learning_rate": 2.55004522751406e-05,
      "loss": 1.942,
      "step": 12560
    },
    {
      "epoch": 1.4772593724291927,
      "grad_norm": 11.601787567138672,
      "learning_rate": 2.5480788138592838e-05,
      "loss": 1.3411,
      "step": 12570
    },
    {
      "epoch": 1.478434598660242,
      "grad_norm": 6.737152576446533,
      "learning_rate": 2.546112400204507e-05,
      "loss": 1.48,
      "step": 12580
    },
    {
      "epoch": 1.4796098248912917,
      "grad_norm": 31.225566864013672,
      "learning_rate": 2.5441459865497308e-05,
      "loss": 1.8104,
      "step": 12590
    },
    {
      "epoch": 1.480785051122341,
      "grad_norm": 14.594199180603027,
      "learning_rate": 2.5421795728949542e-05,
      "loss": 1.6991,
      "step": 12600
    },
    {
      "epoch": 1.4819602773533904,
      "grad_norm": 12.295329093933105,
      "learning_rate": 2.540213159240178e-05,
      "loss": 1.9118,
      "step": 12610
    },
    {
      "epoch": 1.48313550358444,
      "grad_norm": 8.472497940063477,
      "learning_rate": 2.5382467455854016e-05,
      "loss": 1.2336,
      "step": 12620
    },
    {
      "epoch": 1.4843107298154896,
      "grad_norm": 11.251202583312988,
      "learning_rate": 2.536280331930625e-05,
      "loss": 1.7782,
      "step": 12630
    },
    {
      "epoch": 1.485485956046539,
      "grad_norm": 14.846000671386719,
      "learning_rate": 2.5343139182758486e-05,
      "loss": 1.8617,
      "step": 12640
    },
    {
      "epoch": 1.4866611822775884,
      "grad_norm": 6.011964797973633,
      "learning_rate": 2.532347504621072e-05,
      "loss": 1.2501,
      "step": 12650
    },
    {
      "epoch": 1.487836408508638,
      "grad_norm": 16.281448364257812,
      "learning_rate": 2.5303810909662957e-05,
      "loss": 2.2223,
      "step": 12660
    },
    {
      "epoch": 1.4890116347396873,
      "grad_norm": 17.155363082885742,
      "learning_rate": 2.528414677311519e-05,
      "loss": 1.9128,
      "step": 12670
    },
    {
      "epoch": 1.490186860970737,
      "grad_norm": 29.029436111450195,
      "learning_rate": 2.5264482636567428e-05,
      "loss": 1.9609,
      "step": 12680
    },
    {
      "epoch": 1.4913620872017863,
      "grad_norm": 8.634989738464355,
      "learning_rate": 2.5244818500019668e-05,
      "loss": 1.8355,
      "step": 12690
    },
    {
      "epoch": 1.4925373134328357,
      "grad_norm": 24.213268280029297,
      "learning_rate": 2.5225154363471905e-05,
      "loss": 2.1399,
      "step": 12700
    },
    {
      "epoch": 1.4937125396638853,
      "grad_norm": 7.922143936157227,
      "learning_rate": 2.520549022692414e-05,
      "loss": 1.0539,
      "step": 12710
    },
    {
      "epoch": 1.4948877658949349,
      "grad_norm": 13.985539436340332,
      "learning_rate": 2.5185826090376376e-05,
      "loss": 1.7874,
      "step": 12720
    },
    {
      "epoch": 1.4960629921259843,
      "grad_norm": 17.531444549560547,
      "learning_rate": 2.516616195382861e-05,
      "loss": 1.8924,
      "step": 12730
    },
    {
      "epoch": 1.4972382183570336,
      "grad_norm": 30.035966873168945,
      "learning_rate": 2.5146497817280846e-05,
      "loss": 1.8611,
      "step": 12740
    },
    {
      "epoch": 1.4984134445880832,
      "grad_norm": 20.994159698486328,
      "learning_rate": 2.512683368073308e-05,
      "loss": 1.4244,
      "step": 12750
    },
    {
      "epoch": 1.4995886708191326,
      "grad_norm": 12.328607559204102,
      "learning_rate": 2.5107169544185317e-05,
      "loss": 2.0362,
      "step": 12760
    },
    {
      "epoch": 1.5007638970501822,
      "grad_norm": 17.793134689331055,
      "learning_rate": 2.508750540763755e-05,
      "loss": 1.6352,
      "step": 12770
    },
    {
      "epoch": 1.5019391232812316,
      "grad_norm": 23.22760772705078,
      "learning_rate": 2.5067841271089787e-05,
      "loss": 1.7765,
      "step": 12780
    },
    {
      "epoch": 1.503114349512281,
      "grad_norm": 10.039072036743164,
      "learning_rate": 2.504817713454202e-05,
      "loss": 1.9218,
      "step": 12790
    },
    {
      "epoch": 1.5042895757433306,
      "grad_norm": 16.328781127929688,
      "learning_rate": 2.5028512997994258e-05,
      "loss": 1.1876,
      "step": 12800
    },
    {
      "epoch": 1.5054648019743802,
      "grad_norm": 9.655600547790527,
      "learning_rate": 2.500884886144649e-05,
      "loss": 1.9062,
      "step": 12810
    },
    {
      "epoch": 1.5066400282054295,
      "grad_norm": 21.340129852294922,
      "learning_rate": 2.4989184724898732e-05,
      "loss": 1.3482,
      "step": 12820
    },
    {
      "epoch": 1.507815254436479,
      "grad_norm": 30.544523239135742,
      "learning_rate": 2.4969520588350966e-05,
      "loss": 2.2098,
      "step": 12830
    },
    {
      "epoch": 1.5089904806675285,
      "grad_norm": 6.7875871658325195,
      "learning_rate": 2.4949856451803203e-05,
      "loss": 1.3542,
      "step": 12840
    },
    {
      "epoch": 1.510165706898578,
      "grad_norm": 18.240398406982422,
      "learning_rate": 2.4930192315255436e-05,
      "loss": 1.6633,
      "step": 12850
    },
    {
      "epoch": 1.5113409331296275,
      "grad_norm": 20.000505447387695,
      "learning_rate": 2.4910528178707673e-05,
      "loss": 1.9797,
      "step": 12860
    },
    {
      "epoch": 1.5125161593606768,
      "grad_norm": 9.504915237426758,
      "learning_rate": 2.489086404215991e-05,
      "loss": 2.1776,
      "step": 12870
    },
    {
      "epoch": 1.5136913855917264,
      "grad_norm": 7.428323745727539,
      "learning_rate": 2.4871199905612147e-05,
      "loss": 1.5212,
      "step": 12880
    },
    {
      "epoch": 1.514866611822776,
      "grad_norm": 8.118548393249512,
      "learning_rate": 2.485153576906438e-05,
      "loss": 1.749,
      "step": 12890
    },
    {
      "epoch": 1.5160418380538254,
      "grad_norm": 20.74069595336914,
      "learning_rate": 2.4831871632516618e-05,
      "loss": 1.3921,
      "step": 12900
    },
    {
      "epoch": 1.5172170642848748,
      "grad_norm": 14.453350067138672,
      "learning_rate": 2.4812207495968855e-05,
      "loss": 1.7657,
      "step": 12910
    },
    {
      "epoch": 1.5183922905159242,
      "grad_norm": 12.455978393554688,
      "learning_rate": 2.479254335942109e-05,
      "loss": 1.6422,
      "step": 12920
    },
    {
      "epoch": 1.5195675167469738,
      "grad_norm": 15.97265911102295,
      "learning_rate": 2.4772879222873326e-05,
      "loss": 1.6901,
      "step": 12930
    },
    {
      "epoch": 1.5207427429780234,
      "grad_norm": 9.481744766235352,
      "learning_rate": 2.475321508632556e-05,
      "loss": 1.6257,
      "step": 12940
    },
    {
      "epoch": 1.5219179692090727,
      "grad_norm": 12.6028470993042,
      "learning_rate": 2.4733550949777796e-05,
      "loss": 1.7319,
      "step": 12950
    },
    {
      "epoch": 1.5230931954401221,
      "grad_norm": 13.411657333374023,
      "learning_rate": 2.4713886813230033e-05,
      "loss": 1.7858,
      "step": 12960
    },
    {
      "epoch": 1.5242684216711717,
      "grad_norm": 68.91178131103516,
      "learning_rate": 2.469422267668227e-05,
      "loss": 1.8983,
      "step": 12970
    },
    {
      "epoch": 1.5254436479022213,
      "grad_norm": 10.426840782165527,
      "learning_rate": 2.4674558540134504e-05,
      "loss": 1.5062,
      "step": 12980
    },
    {
      "epoch": 1.5266188741332707,
      "grad_norm": 13.138900756835938,
      "learning_rate": 2.465489440358674e-05,
      "loss": 1.9252,
      "step": 12990
    },
    {
      "epoch": 1.52779410036432,
      "grad_norm": 15.167285919189453,
      "learning_rate": 2.4635230267038974e-05,
      "loss": 1.4405,
      "step": 13000
    },
    {
      "epoch": 1.5289693265953697,
      "grad_norm": 27.560468673706055,
      "learning_rate": 2.461556613049121e-05,
      "loss": 1.1561,
      "step": 13010
    },
    {
      "epoch": 1.530144552826419,
      "grad_norm": 13.359074592590332,
      "learning_rate": 2.4595901993943445e-05,
      "loss": 1.0565,
      "step": 13020
    },
    {
      "epoch": 1.5313197790574686,
      "grad_norm": 25.508323669433594,
      "learning_rate": 2.4576237857395685e-05,
      "loss": 1.2674,
      "step": 13030
    },
    {
      "epoch": 1.532495005288518,
      "grad_norm": 12.744986534118652,
      "learning_rate": 2.455657372084792e-05,
      "loss": 1.9768,
      "step": 13040
    },
    {
      "epoch": 1.5336702315195674,
      "grad_norm": 10.413082122802734,
      "learning_rate": 2.4536909584300156e-05,
      "loss": 1.777,
      "step": 13050
    },
    {
      "epoch": 1.534845457750617,
      "grad_norm": 13.364302635192871,
      "learning_rate": 2.451724544775239e-05,
      "loss": 1.1108,
      "step": 13060
    },
    {
      "epoch": 1.5360206839816666,
      "grad_norm": 24.046815872192383,
      "learning_rate": 2.4497581311204627e-05,
      "loss": 1.281,
      "step": 13070
    },
    {
      "epoch": 1.537195910212716,
      "grad_norm": 20.5305233001709,
      "learning_rate": 2.447791717465686e-05,
      "loss": 1.2115,
      "step": 13080
    },
    {
      "epoch": 1.5383711364437653,
      "grad_norm": 11.20641040802002,
      "learning_rate": 2.4458253038109097e-05,
      "loss": 1.8645,
      "step": 13090
    },
    {
      "epoch": 1.539546362674815,
      "grad_norm": 29.87605094909668,
      "learning_rate": 2.443858890156133e-05,
      "loss": 1.4549,
      "step": 13100
    },
    {
      "epoch": 1.5407215889058645,
      "grad_norm": 25.23017120361328,
      "learning_rate": 2.441892476501357e-05,
      "loss": 1.4,
      "step": 13110
    },
    {
      "epoch": 1.541896815136914,
      "grad_norm": 38.69597244262695,
      "learning_rate": 2.4399260628465805e-05,
      "loss": 1.8132,
      "step": 13120
    },
    {
      "epoch": 1.5430720413679633,
      "grad_norm": 15.004799842834473,
      "learning_rate": 2.4379596491918042e-05,
      "loss": 1.576,
      "step": 13130
    },
    {
      "epoch": 1.5442472675990127,
      "grad_norm": 25.99958610534668,
      "learning_rate": 2.4359932355370275e-05,
      "loss": 2.0497,
      "step": 13140
    },
    {
      "epoch": 1.5454224938300622,
      "grad_norm": 19.5682315826416,
      "learning_rate": 2.4340268218822512e-05,
      "loss": 1.8506,
      "step": 13150
    },
    {
      "epoch": 1.5465977200611118,
      "grad_norm": 20.376073837280273,
      "learning_rate": 2.4320604082274746e-05,
      "loss": 2.1418,
      "step": 13160
    },
    {
      "epoch": 1.5477729462921612,
      "grad_norm": 14.121232986450195,
      "learning_rate": 2.4300939945726983e-05,
      "loss": 1.165,
      "step": 13170
    },
    {
      "epoch": 1.5489481725232106,
      "grad_norm": 32.67259979248047,
      "learning_rate": 2.428127580917922e-05,
      "loss": 1.9411,
      "step": 13180
    },
    {
      "epoch": 1.5501233987542602,
      "grad_norm": 14.47479248046875,
      "learning_rate": 2.4261611672631457e-05,
      "loss": 1.226,
      "step": 13190
    },
    {
      "epoch": 1.5512986249853098,
      "grad_norm": 18.62323570251465,
      "learning_rate": 2.424194753608369e-05,
      "loss": 1.3418,
      "step": 13200
    },
    {
      "epoch": 1.5524738512163592,
      "grad_norm": 7.1875810623168945,
      "learning_rate": 2.4222283399535928e-05,
      "loss": 1.3555,
      "step": 13210
    },
    {
      "epoch": 1.5536490774474085,
      "grad_norm": 12.631381034851074,
      "learning_rate": 2.4202619262988165e-05,
      "loss": 2.2608,
      "step": 13220
    },
    {
      "epoch": 1.5548243036784581,
      "grad_norm": 15.894914627075195,
      "learning_rate": 2.4182955126440398e-05,
      "loss": 1.5137,
      "step": 13230
    },
    {
      "epoch": 1.5559995299095077,
      "grad_norm": 19.579395294189453,
      "learning_rate": 2.4163290989892635e-05,
      "loss": 2.143,
      "step": 13240
    },
    {
      "epoch": 1.5571747561405571,
      "grad_norm": 20.4045352935791,
      "learning_rate": 2.414362685334487e-05,
      "loss": 1.5747,
      "step": 13250
    },
    {
      "epoch": 1.5583499823716065,
      "grad_norm": 21.559900283813477,
      "learning_rate": 2.412396271679711e-05,
      "loss": 1.6538,
      "step": 13260
    },
    {
      "epoch": 1.5595252086026559,
      "grad_norm": 16.404823303222656,
      "learning_rate": 2.4104298580249343e-05,
      "loss": 1.3554,
      "step": 13270
    },
    {
      "epoch": 1.5607004348337055,
      "grad_norm": 13.902963638305664,
      "learning_rate": 2.408463444370158e-05,
      "loss": 1.6278,
      "step": 13280
    },
    {
      "epoch": 1.561875661064755,
      "grad_norm": 16.424352645874023,
      "learning_rate": 2.4064970307153814e-05,
      "loss": 1.543,
      "step": 13290
    },
    {
      "epoch": 1.5630508872958044,
      "grad_norm": 23.170391082763672,
      "learning_rate": 2.404530617060605e-05,
      "loss": 1.733,
      "step": 13300
    },
    {
      "epoch": 1.5642261135268538,
      "grad_norm": 9.602981567382812,
      "learning_rate": 2.4025642034058284e-05,
      "loss": 1.4072,
      "step": 13310
    },
    {
      "epoch": 1.5654013397579034,
      "grad_norm": 24.85518455505371,
      "learning_rate": 2.400597789751052e-05,
      "loss": 1.3486,
      "step": 13320
    },
    {
      "epoch": 1.566576565988953,
      "grad_norm": 12.331855773925781,
      "learning_rate": 2.3986313760962758e-05,
      "loss": 1.338,
      "step": 13330
    },
    {
      "epoch": 1.5677517922200024,
      "grad_norm": 27.10284423828125,
      "learning_rate": 2.3966649624414995e-05,
      "loss": 1.4931,
      "step": 13340
    },
    {
      "epoch": 1.5689270184510518,
      "grad_norm": 9.625365257263184,
      "learning_rate": 2.394698548786723e-05,
      "loss": 1.4701,
      "step": 13350
    },
    {
      "epoch": 1.5701022446821014,
      "grad_norm": 37.234012603759766,
      "learning_rate": 2.3927321351319466e-05,
      "loss": 1.0878,
      "step": 13360
    },
    {
      "epoch": 1.5712774709131507,
      "grad_norm": 7.50670051574707,
      "learning_rate": 2.39076572147717e-05,
      "loss": 1.684,
      "step": 13370
    },
    {
      "epoch": 1.5724526971442003,
      "grad_norm": 12.251920700073242,
      "learning_rate": 2.3887993078223936e-05,
      "loss": 1.4608,
      "step": 13380
    },
    {
      "epoch": 1.5736279233752497,
      "grad_norm": 17.520998001098633,
      "learning_rate": 2.386832894167617e-05,
      "loss": 2.5196,
      "step": 13390
    },
    {
      "epoch": 1.574803149606299,
      "grad_norm": 13.084623336791992,
      "learning_rate": 2.3848664805128407e-05,
      "loss": 1.1434,
      "step": 13400
    },
    {
      "epoch": 1.5759783758373487,
      "grad_norm": 14.465731620788574,
      "learning_rate": 2.3829000668580644e-05,
      "loss": 2.1066,
      "step": 13410
    },
    {
      "epoch": 1.5771536020683983,
      "grad_norm": 17.975435256958008,
      "learning_rate": 2.380933653203288e-05,
      "loss": 1.7142,
      "step": 13420
    },
    {
      "epoch": 1.5783288282994477,
      "grad_norm": 8.271294593811035,
      "learning_rate": 2.3789672395485115e-05,
      "loss": 1.6558,
      "step": 13430
    },
    {
      "epoch": 1.579504054530497,
      "grad_norm": 15.479211807250977,
      "learning_rate": 2.377000825893735e-05,
      "loss": 1.5763,
      "step": 13440
    },
    {
      "epoch": 1.5806792807615466,
      "grad_norm": 13.243907928466797,
      "learning_rate": 2.3750344122389585e-05,
      "loss": 1.4872,
      "step": 13450
    },
    {
      "epoch": 1.5818545069925962,
      "grad_norm": 11.824787139892578,
      "learning_rate": 2.3730679985841822e-05,
      "loss": 1.6649,
      "step": 13460
    },
    {
      "epoch": 1.5830297332236456,
      "grad_norm": 11.993182182312012,
      "learning_rate": 2.371101584929406e-05,
      "loss": 1.6277,
      "step": 13470
    },
    {
      "epoch": 1.584204959454695,
      "grad_norm": 22.970918655395508,
      "learning_rate": 2.3691351712746293e-05,
      "loss": 2.0698,
      "step": 13480
    },
    {
      "epoch": 1.5853801856857443,
      "grad_norm": 9.091819763183594,
      "learning_rate": 2.367168757619853e-05,
      "loss": 1.7421,
      "step": 13490
    },
    {
      "epoch": 1.586555411916794,
      "grad_norm": 15.117060661315918,
      "learning_rate": 2.3652023439650767e-05,
      "loss": 1.0451,
      "step": 13500
    },
    {
      "epoch": 1.5877306381478435,
      "grad_norm": 13.008851051330566,
      "learning_rate": 2.3632359303103004e-05,
      "loss": 1.3959,
      "step": 13510
    },
    {
      "epoch": 1.588905864378893,
      "grad_norm": 12.525320053100586,
      "learning_rate": 2.3612695166555237e-05,
      "loss": 1.8474,
      "step": 13520
    },
    {
      "epoch": 1.5900810906099423,
      "grad_norm": 24.26509666442871,
      "learning_rate": 2.3593031030007474e-05,
      "loss": 1.6211,
      "step": 13530
    },
    {
      "epoch": 1.591256316840992,
      "grad_norm": 11.088159561157227,
      "learning_rate": 2.3573366893459708e-05,
      "loss": 2.0404,
      "step": 13540
    },
    {
      "epoch": 1.5924315430720415,
      "grad_norm": 23.95115089416504,
      "learning_rate": 2.3553702756911945e-05,
      "loss": 1.2779,
      "step": 13550
    },
    {
      "epoch": 1.5936067693030909,
      "grad_norm": 6.743244647979736,
      "learning_rate": 2.3534038620364182e-05,
      "loss": 2.0547,
      "step": 13560
    },
    {
      "epoch": 1.5947819955341402,
      "grad_norm": 11.08764362335205,
      "learning_rate": 2.351437448381642e-05,
      "loss": 1.8369,
      "step": 13570
    },
    {
      "epoch": 1.5959572217651898,
      "grad_norm": 23.209733963012695,
      "learning_rate": 2.3494710347268653e-05,
      "loss": 1.6189,
      "step": 13580
    },
    {
      "epoch": 1.5971324479962394,
      "grad_norm": 13.266522407531738,
      "learning_rate": 2.347504621072089e-05,
      "loss": 1.3351,
      "step": 13590
    },
    {
      "epoch": 1.5983076742272888,
      "grad_norm": 8.084033966064453,
      "learning_rate": 2.3455382074173123e-05,
      "loss": 2.0397,
      "step": 13600
    },
    {
      "epoch": 1.5994829004583382,
      "grad_norm": 12.81651496887207,
      "learning_rate": 2.343571793762536e-05,
      "loss": 2.4541,
      "step": 13610
    },
    {
      "epoch": 1.6006581266893876,
      "grad_norm": 12.58523941040039,
      "learning_rate": 2.3416053801077594e-05,
      "loss": 1.4543,
      "step": 13620
    },
    {
      "epoch": 1.6018333529204372,
      "grad_norm": 9.639310836791992,
      "learning_rate": 2.339638966452983e-05,
      "loss": 1.2556,
      "step": 13630
    },
    {
      "epoch": 1.6030085791514868,
      "grad_norm": 15.736313819885254,
      "learning_rate": 2.3376725527982068e-05,
      "loss": 1.684,
      "step": 13640
    },
    {
      "epoch": 1.6041838053825361,
      "grad_norm": 16.98958396911621,
      "learning_rate": 2.3357061391434305e-05,
      "loss": 1.3256,
      "step": 13650
    },
    {
      "epoch": 1.6053590316135855,
      "grad_norm": 23.183887481689453,
      "learning_rate": 2.333739725488654e-05,
      "loss": 1.5978,
      "step": 13660
    },
    {
      "epoch": 1.606534257844635,
      "grad_norm": 11.772687911987305,
      "learning_rate": 2.3317733118338776e-05,
      "loss": 1.2889,
      "step": 13670
    },
    {
      "epoch": 1.6077094840756847,
      "grad_norm": 10.43055248260498,
      "learning_rate": 2.329806898179101e-05,
      "loss": 1.9364,
      "step": 13680
    },
    {
      "epoch": 1.608884710306734,
      "grad_norm": 10.55484390258789,
      "learning_rate": 2.3278404845243246e-05,
      "loss": 0.9454,
      "step": 13690
    },
    {
      "epoch": 1.6100599365377835,
      "grad_norm": 9.031423568725586,
      "learning_rate": 2.325874070869548e-05,
      "loss": 1.0303,
      "step": 13700
    },
    {
      "epoch": 1.611235162768833,
      "grad_norm": 20.539527893066406,
      "learning_rate": 2.323907657214772e-05,
      "loss": 1.7518,
      "step": 13710
    },
    {
      "epoch": 1.6124103889998824,
      "grad_norm": 16.024499893188477,
      "learning_rate": 2.3219412435599954e-05,
      "loss": 1.6738,
      "step": 13720
    },
    {
      "epoch": 1.613585615230932,
      "grad_norm": 9.670287132263184,
      "learning_rate": 2.319974829905219e-05,
      "loss": 2.0154,
      "step": 13730
    },
    {
      "epoch": 1.6147608414619814,
      "grad_norm": 14.615625381469727,
      "learning_rate": 2.3180084162504424e-05,
      "loss": 1.7213,
      "step": 13740
    },
    {
      "epoch": 1.6159360676930308,
      "grad_norm": 24.0756893157959,
      "learning_rate": 2.316042002595666e-05,
      "loss": 1.6149,
      "step": 13750
    },
    {
      "epoch": 1.6171112939240804,
      "grad_norm": 11.231411933898926,
      "learning_rate": 2.3140755889408895e-05,
      "loss": 1.8336,
      "step": 13760
    },
    {
      "epoch": 1.61828652015513,
      "grad_norm": 13.020950317382812,
      "learning_rate": 2.3121091752861132e-05,
      "loss": 1.7166,
      "step": 13770
    },
    {
      "epoch": 1.6194617463861793,
      "grad_norm": 8.341866493225098,
      "learning_rate": 2.310142761631337e-05,
      "loss": 1.7574,
      "step": 13780
    },
    {
      "epoch": 1.6206369726172287,
      "grad_norm": 3.6630146503448486,
      "learning_rate": 2.3081763479765606e-05,
      "loss": 1.6003,
      "step": 13790
    },
    {
      "epoch": 1.6218121988482783,
      "grad_norm": 13.451770782470703,
      "learning_rate": 2.306209934321784e-05,
      "loss": 1.8538,
      "step": 13800
    },
    {
      "epoch": 1.622987425079328,
      "grad_norm": 8.236064910888672,
      "learning_rate": 2.3042435206670077e-05,
      "loss": 1.403,
      "step": 13810
    },
    {
      "epoch": 1.6241626513103773,
      "grad_norm": 21.269493103027344,
      "learning_rate": 2.3022771070122314e-05,
      "loss": 1.5638,
      "step": 13820
    },
    {
      "epoch": 1.6253378775414267,
      "grad_norm": 9.01737117767334,
      "learning_rate": 2.3003106933574547e-05,
      "loss": 1.6201,
      "step": 13830
    },
    {
      "epoch": 1.626513103772476,
      "grad_norm": 10.166827201843262,
      "learning_rate": 2.2983442797026784e-05,
      "loss": 1.3343,
      "step": 13840
    },
    {
      "epoch": 1.6276883300035256,
      "grad_norm": 11.048224449157715,
      "learning_rate": 2.2963778660479018e-05,
      "loss": 1.3495,
      "step": 13850
    },
    {
      "epoch": 1.6288635562345752,
      "grad_norm": 9.533370018005371,
      "learning_rate": 2.2944114523931258e-05,
      "loss": 1.4206,
      "step": 13860
    },
    {
      "epoch": 1.6300387824656246,
      "grad_norm": 5.491836071014404,
      "learning_rate": 2.2924450387383492e-05,
      "loss": 1.5719,
      "step": 13870
    },
    {
      "epoch": 1.631214008696674,
      "grad_norm": 23.731517791748047,
      "learning_rate": 2.290478625083573e-05,
      "loss": 1.4359,
      "step": 13880
    },
    {
      "epoch": 1.6323892349277236,
      "grad_norm": 23.620630264282227,
      "learning_rate": 2.2885122114287962e-05,
      "loss": 1.6474,
      "step": 13890
    },
    {
      "epoch": 1.6335644611587732,
      "grad_norm": 9.7564058303833,
      "learning_rate": 2.28654579777402e-05,
      "loss": 1.4318,
      "step": 13900
    },
    {
      "epoch": 1.6347396873898226,
      "grad_norm": 15.375218391418457,
      "learning_rate": 2.2845793841192433e-05,
      "loss": 1.2751,
      "step": 13910
    },
    {
      "epoch": 1.635914913620872,
      "grad_norm": 4.516286849975586,
      "learning_rate": 2.282612970464467e-05,
      "loss": 1.713,
      "step": 13920
    },
    {
      "epoch": 1.6370901398519215,
      "grad_norm": 24.739105224609375,
      "learning_rate": 2.2806465568096904e-05,
      "loss": 1.5268,
      "step": 13930
    },
    {
      "epoch": 1.6382653660829711,
      "grad_norm": 17.201942443847656,
      "learning_rate": 2.2786801431549144e-05,
      "loss": 1.9446,
      "step": 13940
    },
    {
      "epoch": 1.6394405923140205,
      "grad_norm": 38.92192077636719,
      "learning_rate": 2.2767137295001378e-05,
      "loss": 1.6457,
      "step": 13950
    },
    {
      "epoch": 1.6406158185450699,
      "grad_norm": 15.543179512023926,
      "learning_rate": 2.2747473158453615e-05,
      "loss": 1.8217,
      "step": 13960
    },
    {
      "epoch": 1.6417910447761193,
      "grad_norm": 16.875457763671875,
      "learning_rate": 2.2727809021905848e-05,
      "loss": 2.2456,
      "step": 13970
    },
    {
      "epoch": 1.6429662710071689,
      "grad_norm": 16.5179386138916,
      "learning_rate": 2.2708144885358085e-05,
      "loss": 1.9581,
      "step": 13980
    },
    {
      "epoch": 1.6441414972382185,
      "grad_norm": 13.512782096862793,
      "learning_rate": 2.268848074881032e-05,
      "loss": 1.2551,
      "step": 13990
    },
    {
      "epoch": 1.6453167234692678,
      "grad_norm": 2.64096999168396,
      "learning_rate": 2.2668816612262556e-05,
      "loss": 2.0127,
      "step": 14000
    },
    {
      "epoch": 1.6464919497003172,
      "grad_norm": 5.997403144836426,
      "learning_rate": 2.2649152475714793e-05,
      "loss": 1.1966,
      "step": 14010
    },
    {
      "epoch": 1.6476671759313668,
      "grad_norm": 15.755878448486328,
      "learning_rate": 2.262948833916703e-05,
      "loss": 1.7785,
      "step": 14020
    },
    {
      "epoch": 1.6488424021624164,
      "grad_norm": 11.280073165893555,
      "learning_rate": 2.2609824202619264e-05,
      "loss": 1.3617,
      "step": 14030
    },
    {
      "epoch": 1.6500176283934658,
      "grad_norm": 11.927693367004395,
      "learning_rate": 2.25901600660715e-05,
      "loss": 1.6588,
      "step": 14040
    },
    {
      "epoch": 1.6511928546245151,
      "grad_norm": 13.645112991333008,
      "learning_rate": 2.2570495929523734e-05,
      "loss": 2.074,
      "step": 14050
    },
    {
      "epoch": 1.6523680808555647,
      "grad_norm": 13.581123352050781,
      "learning_rate": 2.255083179297597e-05,
      "loss": 1.6504,
      "step": 14060
    },
    {
      "epoch": 1.6535433070866141,
      "grad_norm": 22.56283187866211,
      "learning_rate": 2.2531167656428208e-05,
      "loss": 1.4168,
      "step": 14070
    },
    {
      "epoch": 1.6547185333176637,
      "grad_norm": 20.056541442871094,
      "learning_rate": 2.2511503519880442e-05,
      "loss": 1.7946,
      "step": 14080
    },
    {
      "epoch": 1.655893759548713,
      "grad_norm": 32.11310577392578,
      "learning_rate": 2.249183938333268e-05,
      "loss": 1.5121,
      "step": 14090
    },
    {
      "epoch": 1.6570689857797625,
      "grad_norm": 12.997039794921875,
      "learning_rate": 2.2472175246784916e-05,
      "loss": 1.5402,
      "step": 14100
    },
    {
      "epoch": 1.658244212010812,
      "grad_norm": 7.409574031829834,
      "learning_rate": 2.245251111023715e-05,
      "loss": 1.4821,
      "step": 14110
    },
    {
      "epoch": 1.6594194382418617,
      "grad_norm": 15.795926094055176,
      "learning_rate": 2.2432846973689386e-05,
      "loss": 1.2498,
      "step": 14120
    },
    {
      "epoch": 1.660594664472911,
      "grad_norm": 7.903818130493164,
      "learning_rate": 2.2413182837141623e-05,
      "loss": 1.5677,
      "step": 14130
    },
    {
      "epoch": 1.6617698907039604,
      "grad_norm": 16.972810745239258,
      "learning_rate": 2.2393518700593857e-05,
      "loss": 1.437,
      "step": 14140
    },
    {
      "epoch": 1.66294511693501,
      "grad_norm": 12.543509483337402,
      "learning_rate": 2.2373854564046094e-05,
      "loss": 1.389,
      "step": 14150
    },
    {
      "epoch": 1.6641203431660596,
      "grad_norm": 24.746238708496094,
      "learning_rate": 2.2354190427498328e-05,
      "loss": 1.4481,
      "step": 14160
    },
    {
      "epoch": 1.665295569397109,
      "grad_norm": 16.795316696166992,
      "learning_rate": 2.2334526290950568e-05,
      "loss": 2.2022,
      "step": 14170
    },
    {
      "epoch": 1.6664707956281584,
      "grad_norm": 25.95477867126465,
      "learning_rate": 2.23148621544028e-05,
      "loss": 1.8232,
      "step": 14180
    },
    {
      "epoch": 1.6676460218592077,
      "grad_norm": 12.485878944396973,
      "learning_rate": 2.229519801785504e-05,
      "loss": 1.3489,
      "step": 14190
    },
    {
      "epoch": 1.6688212480902573,
      "grad_norm": 19.97566032409668,
      "learning_rate": 2.2275533881307272e-05,
      "loss": 1.5134,
      "step": 14200
    },
    {
      "epoch": 1.669996474321307,
      "grad_norm": 24.792984008789062,
      "learning_rate": 2.225586974475951e-05,
      "loss": 1.7997,
      "step": 14210
    },
    {
      "epoch": 1.6711717005523563,
      "grad_norm": 12.283632278442383,
      "learning_rate": 2.2236205608211743e-05,
      "loss": 1.539,
      "step": 14220
    },
    {
      "epoch": 1.6723469267834057,
      "grad_norm": 13.940792083740234,
      "learning_rate": 2.221654147166398e-05,
      "loss": 1.7305,
      "step": 14230
    },
    {
      "epoch": 1.6735221530144553,
      "grad_norm": 16.169710159301758,
      "learning_rate": 2.2196877335116217e-05,
      "loss": 2.1603,
      "step": 14240
    },
    {
      "epoch": 1.6746973792455049,
      "grad_norm": 14.844944953918457,
      "learning_rate": 2.2177213198568454e-05,
      "loss": 1.2753,
      "step": 14250
    },
    {
      "epoch": 1.6758726054765543,
      "grad_norm": 19.826618194580078,
      "learning_rate": 2.2157549062020687e-05,
      "loss": 1.8187,
      "step": 14260
    },
    {
      "epoch": 1.6770478317076036,
      "grad_norm": 7.382688045501709,
      "learning_rate": 2.2137884925472924e-05,
      "loss": 1.5752,
      "step": 14270
    },
    {
      "epoch": 1.6782230579386532,
      "grad_norm": 9.145103454589844,
      "learning_rate": 2.2118220788925158e-05,
      "loss": 1.5878,
      "step": 14280
    },
    {
      "epoch": 1.6793982841697028,
      "grad_norm": 17.487516403198242,
      "learning_rate": 2.2098556652377395e-05,
      "loss": 1.6015,
      "step": 14290
    },
    {
      "epoch": 1.6805735104007522,
      "grad_norm": 22.29290199279785,
      "learning_rate": 2.207889251582963e-05,
      "loss": 1.3597,
      "step": 14300
    },
    {
      "epoch": 1.6817487366318016,
      "grad_norm": 9.829691886901855,
      "learning_rate": 2.2059228379281866e-05,
      "loss": 1.6406,
      "step": 14310
    },
    {
      "epoch": 1.682923962862851,
      "grad_norm": 25.973955154418945,
      "learning_rate": 2.2039564242734103e-05,
      "loss": 1.725,
      "step": 14320
    },
    {
      "epoch": 1.6840991890939006,
      "grad_norm": 17.571422576904297,
      "learning_rate": 2.201990010618634e-05,
      "loss": 1.4696,
      "step": 14330
    },
    {
      "epoch": 1.6852744153249501,
      "grad_norm": 6.229103088378906,
      "learning_rate": 2.2000235969638573e-05,
      "loss": 1.5673,
      "step": 14340
    },
    {
      "epoch": 1.6864496415559995,
      "grad_norm": 15.762721061706543,
      "learning_rate": 2.198057183309081e-05,
      "loss": 1.4129,
      "step": 14350
    },
    {
      "epoch": 1.687624867787049,
      "grad_norm": 17.696561813354492,
      "learning_rate": 2.1960907696543044e-05,
      "loss": 1.6664,
      "step": 14360
    },
    {
      "epoch": 1.6888000940180985,
      "grad_norm": 14.710844993591309,
      "learning_rate": 2.194124355999528e-05,
      "loss": 1.4356,
      "step": 14370
    },
    {
      "epoch": 1.689975320249148,
      "grad_norm": 11.634048461914062,
      "learning_rate": 2.1921579423447518e-05,
      "loss": 1.1632,
      "step": 14380
    },
    {
      "epoch": 1.6911505464801975,
      "grad_norm": 15.194547653198242,
      "learning_rate": 2.1901915286899755e-05,
      "loss": 1.8045,
      "step": 14390
    },
    {
      "epoch": 1.6923257727112468,
      "grad_norm": 5.334444522857666,
      "learning_rate": 2.188225115035199e-05,
      "loss": 1.462,
      "step": 14400
    },
    {
      "epoch": 1.6935009989422964,
      "grad_norm": 19.931560516357422,
      "learning_rate": 2.1862587013804226e-05,
      "loss": 1.8397,
      "step": 14410
    },
    {
      "epoch": 1.6946762251733458,
      "grad_norm": 9.427385330200195,
      "learning_rate": 2.1842922877256463e-05,
      "loss": 1.6331,
      "step": 14420
    },
    {
      "epoch": 1.6958514514043954,
      "grad_norm": 20.44755744934082,
      "learning_rate": 2.1823258740708696e-05,
      "loss": 1.9578,
      "step": 14430
    },
    {
      "epoch": 1.6970266776354448,
      "grad_norm": 4.102517127990723,
      "learning_rate": 2.1803594604160933e-05,
      "loss": 1.2931,
      "step": 14440
    },
    {
      "epoch": 1.6982019038664942,
      "grad_norm": 18.573108673095703,
      "learning_rate": 2.1783930467613167e-05,
      "loss": 1.9478,
      "step": 14450
    },
    {
      "epoch": 1.6993771300975438,
      "grad_norm": 22.589834213256836,
      "learning_rate": 2.1764266331065404e-05,
      "loss": 1.6901,
      "step": 14460
    },
    {
      "epoch": 1.7005523563285934,
      "grad_norm": 12.242244720458984,
      "learning_rate": 2.174460219451764e-05,
      "loss": 1.5778,
      "step": 14470
    },
    {
      "epoch": 1.7017275825596427,
      "grad_norm": 11.019927978515625,
      "learning_rate": 2.1724938057969878e-05,
      "loss": 2.0826,
      "step": 14480
    },
    {
      "epoch": 1.7029028087906921,
      "grad_norm": 4.205226898193359,
      "learning_rate": 2.170527392142211e-05,
      "loss": 1.0502,
      "step": 14490
    },
    {
      "epoch": 1.7040780350217417,
      "grad_norm": 12.436955451965332,
      "learning_rate": 2.168560978487435e-05,
      "loss": 1.9233,
      "step": 14500
    },
    {
      "epoch": 1.7052532612527913,
      "grad_norm": 8.315258979797363,
      "learning_rate": 2.1665945648326582e-05,
      "loss": 0.9952,
      "step": 14510
    },
    {
      "epoch": 1.7064284874838407,
      "grad_norm": 30.017040252685547,
      "learning_rate": 2.164628151177882e-05,
      "loss": 1.2014,
      "step": 14520
    },
    {
      "epoch": 1.70760371371489,
      "grad_norm": 20.600370407104492,
      "learning_rate": 2.1626617375231053e-05,
      "loss": 1.7604,
      "step": 14530
    },
    {
      "epoch": 1.7087789399459394,
      "grad_norm": 25.90689468383789,
      "learning_rate": 2.1606953238683293e-05,
      "loss": 1.9455,
      "step": 14540
    },
    {
      "epoch": 1.709954166176989,
      "grad_norm": 12.067241668701172,
      "learning_rate": 2.1587289102135527e-05,
      "loss": 1.7459,
      "step": 14550
    },
    {
      "epoch": 1.7111293924080386,
      "grad_norm": 27.876375198364258,
      "learning_rate": 2.1567624965587764e-05,
      "loss": 1.7754,
      "step": 14560
    },
    {
      "epoch": 1.712304618639088,
      "grad_norm": 7.695990562438965,
      "learning_rate": 2.1547960829039997e-05,
      "loss": 1.5669,
      "step": 14570
    },
    {
      "epoch": 1.7134798448701374,
      "grad_norm": 19.755817413330078,
      "learning_rate": 2.1528296692492234e-05,
      "loss": 2.7614,
      "step": 14580
    },
    {
      "epoch": 1.714655071101187,
      "grad_norm": 15.366389274597168,
      "learning_rate": 2.1508632555944468e-05,
      "loss": 1.3942,
      "step": 14590
    },
    {
      "epoch": 1.7158302973322366,
      "grad_norm": 12.85650634765625,
      "learning_rate": 2.1488968419396705e-05,
      "loss": 1.6234,
      "step": 14600
    },
    {
      "epoch": 1.717005523563286,
      "grad_norm": 18.98619270324707,
      "learning_rate": 2.146930428284894e-05,
      "loss": 1.6533,
      "step": 14610
    },
    {
      "epoch": 1.7181807497943353,
      "grad_norm": 25.198078155517578,
      "learning_rate": 2.144964014630118e-05,
      "loss": 1.4674,
      "step": 14620
    },
    {
      "epoch": 1.719355976025385,
      "grad_norm": 8.083762168884277,
      "learning_rate": 2.1429976009753412e-05,
      "loss": 1.7511,
      "step": 14630
    },
    {
      "epoch": 1.7205312022564345,
      "grad_norm": 13.305805206298828,
      "learning_rate": 2.141031187320565e-05,
      "loss": 1.5836,
      "step": 14640
    },
    {
      "epoch": 1.721706428487484,
      "grad_norm": 22.670778274536133,
      "learning_rate": 2.1390647736657883e-05,
      "loss": 1.5592,
      "step": 14650
    },
    {
      "epoch": 1.7228816547185333,
      "grad_norm": 14.467066764831543,
      "learning_rate": 2.137098360011012e-05,
      "loss": 2.1421,
      "step": 14660
    },
    {
      "epoch": 1.7240568809495826,
      "grad_norm": 11.111145973205566,
      "learning_rate": 2.1351319463562357e-05,
      "loss": 1.9176,
      "step": 14670
    },
    {
      "epoch": 1.7252321071806322,
      "grad_norm": 9.561894416809082,
      "learning_rate": 2.133165532701459e-05,
      "loss": 1.7773,
      "step": 14680
    },
    {
      "epoch": 1.7264073334116818,
      "grad_norm": 12.908599853515625,
      "learning_rate": 2.1311991190466828e-05,
      "loss": 1.427,
      "step": 14690
    },
    {
      "epoch": 1.7275825596427312,
      "grad_norm": 18.676504135131836,
      "learning_rate": 2.1292327053919065e-05,
      "loss": 1.7821,
      "step": 14700
    },
    {
      "epoch": 1.7287577858737806,
      "grad_norm": 15.169939994812012,
      "learning_rate": 2.1272662917371298e-05,
      "loss": 1.5934,
      "step": 14710
    },
    {
      "epoch": 1.7299330121048302,
      "grad_norm": 11.075714111328125,
      "learning_rate": 2.1252998780823535e-05,
      "loss": 1.7434,
      "step": 14720
    },
    {
      "epoch": 1.7311082383358798,
      "grad_norm": 9.755184173583984,
      "learning_rate": 2.1233334644275772e-05,
      "loss": 1.9055,
      "step": 14730
    },
    {
      "epoch": 1.7322834645669292,
      "grad_norm": 42.8953971862793,
      "learning_rate": 2.1213670507728006e-05,
      "loss": 1.7964,
      "step": 14740
    },
    {
      "epoch": 1.7334586907979785,
      "grad_norm": 19.54481315612793,
      "learning_rate": 2.1194006371180243e-05,
      "loss": 1.7345,
      "step": 14750
    },
    {
      "epoch": 1.7346339170290281,
      "grad_norm": 10.605175018310547,
      "learning_rate": 2.1174342234632477e-05,
      "loss": 1.7256,
      "step": 14760
    },
    {
      "epoch": 1.7358091432600775,
      "grad_norm": 10.987232208251953,
      "learning_rate": 2.1154678098084717e-05,
      "loss": 1.3002,
      "step": 14770
    },
    {
      "epoch": 1.7369843694911271,
      "grad_norm": 8.073573112487793,
      "learning_rate": 2.113501396153695e-05,
      "loss": 2.0063,
      "step": 14780
    },
    {
      "epoch": 1.7381595957221765,
      "grad_norm": 9.713318824768066,
      "learning_rate": 2.1115349824989188e-05,
      "loss": 1.4831,
      "step": 14790
    },
    {
      "epoch": 1.7393348219532259,
      "grad_norm": 5.279054641723633,
      "learning_rate": 2.109568568844142e-05,
      "loss": 1.3633,
      "step": 14800
    },
    {
      "epoch": 1.7405100481842755,
      "grad_norm": 16.06197166442871,
      "learning_rate": 2.1076021551893658e-05,
      "loss": 1.859,
      "step": 14810
    },
    {
      "epoch": 1.741685274415325,
      "grad_norm": 11.998491287231445,
      "learning_rate": 2.1056357415345892e-05,
      "loss": 1.453,
      "step": 14820
    },
    {
      "epoch": 1.7428605006463744,
      "grad_norm": 6.896851539611816,
      "learning_rate": 2.103669327879813e-05,
      "loss": 1.9538,
      "step": 14830
    },
    {
      "epoch": 1.7440357268774238,
      "grad_norm": 13.769227027893066,
      "learning_rate": 2.1017029142250362e-05,
      "loss": 1.8089,
      "step": 14840
    },
    {
      "epoch": 1.7452109531084734,
      "grad_norm": 11.807965278625488,
      "learning_rate": 2.0997365005702603e-05,
      "loss": 1.1886,
      "step": 14850
    },
    {
      "epoch": 1.746386179339523,
      "grad_norm": 15.91023063659668,
      "learning_rate": 2.0977700869154836e-05,
      "loss": 1.9798,
      "step": 14860
    },
    {
      "epoch": 1.7475614055705724,
      "grad_norm": 9.32402515411377,
      "learning_rate": 2.0958036732607073e-05,
      "loss": 2.1306,
      "step": 14870
    },
    {
      "epoch": 1.7487366318016218,
      "grad_norm": 6.0911335945129395,
      "learning_rate": 2.0938372596059307e-05,
      "loss": 1.1028,
      "step": 14880
    },
    {
      "epoch": 1.7499118580326711,
      "grad_norm": 20.981815338134766,
      "learning_rate": 2.0918708459511544e-05,
      "loss": 1.8707,
      "step": 14890
    },
    {
      "epoch": 1.7510870842637207,
      "grad_norm": 28.326730728149414,
      "learning_rate": 2.0899044322963778e-05,
      "loss": 1.958,
      "step": 14900
    },
    {
      "epoch": 1.7522623104947703,
      "grad_norm": 10.203848838806152,
      "learning_rate": 2.0879380186416015e-05,
      "loss": 1.7774,
      "step": 14910
    },
    {
      "epoch": 1.7534375367258197,
      "grad_norm": 8.079524040222168,
      "learning_rate": 2.085971604986825e-05,
      "loss": 1.8357,
      "step": 14920
    },
    {
      "epoch": 1.754612762956869,
      "grad_norm": 7.067212104797363,
      "learning_rate": 2.084005191332049e-05,
      "loss": 1.6046,
      "step": 14930
    },
    {
      "epoch": 1.7557879891879187,
      "grad_norm": 12.718252182006836,
      "learning_rate": 2.0820387776772722e-05,
      "loss": 1.7466,
      "step": 14940
    },
    {
      "epoch": 1.7569632154189683,
      "grad_norm": 16.91217803955078,
      "learning_rate": 2.080072364022496e-05,
      "loss": 1.377,
      "step": 14950
    },
    {
      "epoch": 1.7581384416500176,
      "grad_norm": 7.259335041046143,
      "learning_rate": 2.0781059503677193e-05,
      "loss": 1.5563,
      "step": 14960
    },
    {
      "epoch": 1.759313667881067,
      "grad_norm": 18.012235641479492,
      "learning_rate": 2.076139536712943e-05,
      "loss": 1.2648,
      "step": 14970
    },
    {
      "epoch": 1.7604888941121166,
      "grad_norm": 25.567155838012695,
      "learning_rate": 2.0741731230581667e-05,
      "loss": 2.1488,
      "step": 14980
    },
    {
      "epoch": 1.7616641203431662,
      "grad_norm": 11.246857643127441,
      "learning_rate": 2.07220670940339e-05,
      "loss": 1.7621,
      "step": 14990
    },
    {
      "epoch": 1.7628393465742156,
      "grad_norm": 6.5297112464904785,
      "learning_rate": 2.0702402957486137e-05,
      "loss": 1.4746,
      "step": 15000
    },
    {
      "epoch": 1.764014572805265,
      "grad_norm": 13.333626747131348,
      "learning_rate": 2.0682738820938374e-05,
      "loss": 2.2893,
      "step": 15010
    },
    {
      "epoch": 1.7651897990363143,
      "grad_norm": 4.998834609985352,
      "learning_rate": 2.066307468439061e-05,
      "loss": 1.5355,
      "step": 15020
    },
    {
      "epoch": 1.766365025267364,
      "grad_norm": 22.259754180908203,
      "learning_rate": 2.0643410547842845e-05,
      "loss": 1.7196,
      "step": 15030
    },
    {
      "epoch": 1.7675402514984135,
      "grad_norm": 13.758522987365723,
      "learning_rate": 2.0623746411295082e-05,
      "loss": 1.672,
      "step": 15040
    },
    {
      "epoch": 1.768715477729463,
      "grad_norm": 20.18950080871582,
      "learning_rate": 2.0604082274747316e-05,
      "loss": 1.7418,
      "step": 15050
    },
    {
      "epoch": 1.7698907039605123,
      "grad_norm": 13.950532913208008,
      "learning_rate": 2.0584418138199553e-05,
      "loss": 2.0572,
      "step": 15060
    },
    {
      "epoch": 1.7710659301915619,
      "grad_norm": 42.84744644165039,
      "learning_rate": 2.056475400165179e-05,
      "loss": 1.8849,
      "step": 15070
    },
    {
      "epoch": 1.7722411564226115,
      "grad_norm": 9.433717727661133,
      "learning_rate": 2.0545089865104027e-05,
      "loss": 1.0382,
      "step": 15080
    },
    {
      "epoch": 1.7734163826536609,
      "grad_norm": 5.486362934112549,
      "learning_rate": 2.052542572855626e-05,
      "loss": 1.4285,
      "step": 15090
    },
    {
      "epoch": 1.7745916088847102,
      "grad_norm": 21.206350326538086,
      "learning_rate": 2.0505761592008497e-05,
      "loss": 1.3514,
      "step": 15100
    },
    {
      "epoch": 1.7757668351157598,
      "grad_norm": 15.4902982711792,
      "learning_rate": 2.048609745546073e-05,
      "loss": 1.3046,
      "step": 15110
    },
    {
      "epoch": 1.7769420613468092,
      "grad_norm": 11.523720741271973,
      "learning_rate": 2.0466433318912968e-05,
      "loss": 1.3905,
      "step": 15120
    },
    {
      "epoch": 1.7781172875778588,
      "grad_norm": 17.330299377441406,
      "learning_rate": 2.04467691823652e-05,
      "loss": 1.805,
      "step": 15130
    },
    {
      "epoch": 1.7792925138089082,
      "grad_norm": 12.451208114624023,
      "learning_rate": 2.042710504581744e-05,
      "loss": 1.848,
      "step": 15140
    },
    {
      "epoch": 1.7804677400399576,
      "grad_norm": 11.140957832336426,
      "learning_rate": 2.0407440909269676e-05,
      "loss": 1.6484,
      "step": 15150
    },
    {
      "epoch": 1.7816429662710072,
      "grad_norm": 10.376080513000488,
      "learning_rate": 2.0387776772721913e-05,
      "loss": 1.5915,
      "step": 15160
    },
    {
      "epoch": 1.7828181925020568,
      "grad_norm": 26.204957962036133,
      "learning_rate": 2.0368112636174146e-05,
      "loss": 1.6307,
      "step": 15170
    },
    {
      "epoch": 1.7839934187331061,
      "grad_norm": 12.213476181030273,
      "learning_rate": 2.0348448499626383e-05,
      "loss": 1.6749,
      "step": 15180
    },
    {
      "epoch": 1.7851686449641555,
      "grad_norm": 8.617768287658691,
      "learning_rate": 2.0328784363078617e-05,
      "loss": 1.6233,
      "step": 15190
    },
    {
      "epoch": 1.786343871195205,
      "grad_norm": 12.684225082397461,
      "learning_rate": 2.0309120226530854e-05,
      "loss": 1.0867,
      "step": 15200
    },
    {
      "epoch": 1.7875190974262547,
      "grad_norm": 37.54444885253906,
      "learning_rate": 2.0289456089983087e-05,
      "loss": 1.3772,
      "step": 15210
    },
    {
      "epoch": 1.788694323657304,
      "grad_norm": 14.598477363586426,
      "learning_rate": 2.0269791953435328e-05,
      "loss": 1.5692,
      "step": 15220
    },
    {
      "epoch": 1.7898695498883535,
      "grad_norm": 15.622933387756348,
      "learning_rate": 2.025012781688756e-05,
      "loss": 1.6595,
      "step": 15230
    },
    {
      "epoch": 1.7910447761194028,
      "grad_norm": 17.90993881225586,
      "learning_rate": 2.02304636803398e-05,
      "loss": 1.8906,
      "step": 15240
    },
    {
      "epoch": 1.7922200023504524,
      "grad_norm": 12.037593841552734,
      "learning_rate": 2.0210799543792032e-05,
      "loss": 1.9411,
      "step": 15250
    },
    {
      "epoch": 1.793395228581502,
      "grad_norm": 12.603153228759766,
      "learning_rate": 2.019113540724427e-05,
      "loss": 1.8048,
      "step": 15260
    },
    {
      "epoch": 1.7945704548125514,
      "grad_norm": 22.207059860229492,
      "learning_rate": 2.0171471270696503e-05,
      "loss": 1.5271,
      "step": 15270
    },
    {
      "epoch": 1.7957456810436008,
      "grad_norm": 16.105777740478516,
      "learning_rate": 2.015180713414874e-05,
      "loss": 1.941,
      "step": 15280
    },
    {
      "epoch": 1.7969209072746504,
      "grad_norm": 18.638961791992188,
      "learning_rate": 2.0132142997600977e-05,
      "loss": 2.4668,
      "step": 15290
    },
    {
      "epoch": 1.7980961335057,
      "grad_norm": 30.354568481445312,
      "learning_rate": 2.0112478861053214e-05,
      "loss": 1.8597,
      "step": 15300
    },
    {
      "epoch": 1.7992713597367493,
      "grad_norm": 16.00894546508789,
      "learning_rate": 2.0092814724505447e-05,
      "loss": 1.5038,
      "step": 15310
    },
    {
      "epoch": 1.8004465859677987,
      "grad_norm": 5.624074935913086,
      "learning_rate": 2.0073150587957684e-05,
      "loss": 1.5989,
      "step": 15320
    },
    {
      "epoch": 1.8016218121988483,
      "grad_norm": 12.566372871398926,
      "learning_rate": 2.005348645140992e-05,
      "loss": 1.63,
      "step": 15330
    },
    {
      "epoch": 1.802797038429898,
      "grad_norm": 10.40300178527832,
      "learning_rate": 2.0033822314862155e-05,
      "loss": 1.5011,
      "step": 15340
    },
    {
      "epoch": 1.8039722646609473,
      "grad_norm": 17.811296463012695,
      "learning_rate": 2.0014158178314392e-05,
      "loss": 1.8826,
      "step": 15350
    },
    {
      "epoch": 1.8051474908919967,
      "grad_norm": 12.761034965515137,
      "learning_rate": 1.9994494041766625e-05,
      "loss": 1.627,
      "step": 15360
    },
    {
      "epoch": 1.806322717123046,
      "grad_norm": 16.401540756225586,
      "learning_rate": 1.9974829905218862e-05,
      "loss": 1.5322,
      "step": 15370
    },
    {
      "epoch": 1.8074979433540956,
      "grad_norm": 19.90477752685547,
      "learning_rate": 1.99551657686711e-05,
      "loss": 1.2059,
      "step": 15380
    },
    {
      "epoch": 1.8086731695851452,
      "grad_norm": 16.602996826171875,
      "learning_rate": 1.9935501632123336e-05,
      "loss": 1.9874,
      "step": 15390
    },
    {
      "epoch": 1.8098483958161946,
      "grad_norm": 17.31633186340332,
      "learning_rate": 1.991583749557557e-05,
      "loss": 1.3345,
      "step": 15400
    },
    {
      "epoch": 1.811023622047244,
      "grad_norm": 19.061687469482422,
      "learning_rate": 1.9896173359027807e-05,
      "loss": 1.5221,
      "step": 15410
    },
    {
      "epoch": 1.8121988482782936,
      "grad_norm": 12.664130210876465,
      "learning_rate": 1.987650922248004e-05,
      "loss": 1.4985,
      "step": 15420
    },
    {
      "epoch": 1.8133740745093432,
      "grad_norm": 15.607800483703613,
      "learning_rate": 1.9856845085932278e-05,
      "loss": 1.1826,
      "step": 15430
    },
    {
      "epoch": 1.8145493007403926,
      "grad_norm": 7.8081512451171875,
      "learning_rate": 1.983718094938451e-05,
      "loss": 1.2881,
      "step": 15440
    },
    {
      "epoch": 1.815724526971442,
      "grad_norm": 13.65499210357666,
      "learning_rate": 1.981751681283675e-05,
      "loss": 1.5533,
      "step": 15450
    },
    {
      "epoch": 1.8168997532024915,
      "grad_norm": 16.972270965576172,
      "learning_rate": 1.9797852676288985e-05,
      "loss": 1.6573,
      "step": 15460
    },
    {
      "epoch": 1.818074979433541,
      "grad_norm": 7.3098015785217285,
      "learning_rate": 1.9778188539741222e-05,
      "loss": 1.634,
      "step": 15470
    },
    {
      "epoch": 1.8192502056645905,
      "grad_norm": 15.407628059387207,
      "learning_rate": 1.9758524403193456e-05,
      "loss": 2.2918,
      "step": 15480
    },
    {
      "epoch": 1.8204254318956399,
      "grad_norm": 18.99938201904297,
      "learning_rate": 1.9738860266645693e-05,
      "loss": 1.7189,
      "step": 15490
    },
    {
      "epoch": 1.8216006581266893,
      "grad_norm": 25.40113067626953,
      "learning_rate": 1.9719196130097926e-05,
      "loss": 1.5003,
      "step": 15500
    },
    {
      "epoch": 1.8227758843577389,
      "grad_norm": 8.115812301635742,
      "learning_rate": 1.9699531993550163e-05,
      "loss": 1.0121,
      "step": 15510
    },
    {
      "epoch": 1.8239511105887884,
      "grad_norm": 26.262706756591797,
      "learning_rate": 1.9679867857002397e-05,
      "loss": 1.6084,
      "step": 15520
    },
    {
      "epoch": 1.8251263368198378,
      "grad_norm": 12.809268951416016,
      "learning_rate": 1.9660203720454638e-05,
      "loss": 1.3706,
      "step": 15530
    },
    {
      "epoch": 1.8263015630508872,
      "grad_norm": 11.537069320678711,
      "learning_rate": 1.964053958390687e-05,
      "loss": 1.1964,
      "step": 15540
    },
    {
      "epoch": 1.8274767892819368,
      "grad_norm": 21.79517364501953,
      "learning_rate": 1.9620875447359108e-05,
      "loss": 1.5359,
      "step": 15550
    },
    {
      "epoch": 1.8286520155129864,
      "grad_norm": 14.30504322052002,
      "learning_rate": 1.9601211310811342e-05,
      "loss": 1.8847,
      "step": 15560
    },
    {
      "epoch": 1.8298272417440358,
      "grad_norm": 9.660836219787598,
      "learning_rate": 1.958154717426358e-05,
      "loss": 1.5211,
      "step": 15570
    },
    {
      "epoch": 1.8310024679750851,
      "grad_norm": 10.856986999511719,
      "learning_rate": 1.9561883037715816e-05,
      "loss": 1.6242,
      "step": 15580
    },
    {
      "epoch": 1.8321776942061345,
      "grad_norm": 19.880538940429688,
      "learning_rate": 1.954221890116805e-05,
      "loss": 1.4707,
      "step": 15590
    },
    {
      "epoch": 1.8333529204371841,
      "grad_norm": 18.562408447265625,
      "learning_rate": 1.9522554764620286e-05,
      "loss": 1.5754,
      "step": 15600
    },
    {
      "epoch": 1.8345281466682337,
      "grad_norm": 16.71778678894043,
      "learning_rate": 1.9502890628072523e-05,
      "loss": 1.9576,
      "step": 15610
    },
    {
      "epoch": 1.835703372899283,
      "grad_norm": 11.991899490356445,
      "learning_rate": 1.9483226491524757e-05,
      "loss": 1.8698,
      "step": 15620
    },
    {
      "epoch": 1.8368785991303325,
      "grad_norm": 19.46167755126953,
      "learning_rate": 1.9463562354976994e-05,
      "loss": 2.1434,
      "step": 15630
    },
    {
      "epoch": 1.838053825361382,
      "grad_norm": 18.196678161621094,
      "learning_rate": 1.944389821842923e-05,
      "loss": 1.6767,
      "step": 15640
    },
    {
      "epoch": 1.8392290515924317,
      "grad_norm": 15.599390029907227,
      "learning_rate": 1.9424234081881465e-05,
      "loss": 1.3669,
      "step": 15650
    },
    {
      "epoch": 1.840404277823481,
      "grad_norm": 12.657039642333984,
      "learning_rate": 1.94045699453337e-05,
      "loss": 1.5487,
      "step": 15660
    },
    {
      "epoch": 1.8415795040545304,
      "grad_norm": 12.975388526916504,
      "learning_rate": 1.9384905808785935e-05,
      "loss": 1.535,
      "step": 15670
    },
    {
      "epoch": 1.84275473028558,
      "grad_norm": 6.813320636749268,
      "learning_rate": 1.9365241672238176e-05,
      "loss": 1.5014,
      "step": 15680
    },
    {
      "epoch": 1.8439299565166296,
      "grad_norm": 11.668549537658691,
      "learning_rate": 1.934557753569041e-05,
      "loss": 1.8076,
      "step": 15690
    },
    {
      "epoch": 1.845105182747679,
      "grad_norm": 16.73682975769043,
      "learning_rate": 1.9325913399142646e-05,
      "loss": 1.5948,
      "step": 15700
    },
    {
      "epoch": 1.8462804089787284,
      "grad_norm": 16.59640121459961,
      "learning_rate": 1.930624926259488e-05,
      "loss": 1.7277,
      "step": 15710
    },
    {
      "epoch": 1.8474556352097777,
      "grad_norm": 21.479854583740234,
      "learning_rate": 1.9286585126047117e-05,
      "loss": 1.5802,
      "step": 15720
    },
    {
      "epoch": 1.8486308614408273,
      "grad_norm": 12.45886516571045,
      "learning_rate": 1.926692098949935e-05,
      "loss": 1.5545,
      "step": 15730
    },
    {
      "epoch": 1.849806087671877,
      "grad_norm": 6.674084663391113,
      "learning_rate": 1.9247256852951587e-05,
      "loss": 1.5916,
      "step": 15740
    },
    {
      "epoch": 1.8509813139029263,
      "grad_norm": 23.75096321105957,
      "learning_rate": 1.9227592716403824e-05,
      "loss": 2.1153,
      "step": 15750
    },
    {
      "epoch": 1.8521565401339757,
      "grad_norm": 15.70645809173584,
      "learning_rate": 1.920792857985606e-05,
      "loss": 1.5395,
      "step": 15760
    },
    {
      "epoch": 1.8533317663650253,
      "grad_norm": 13.390483856201172,
      "learning_rate": 1.9188264443308295e-05,
      "loss": 1.862,
      "step": 15770
    },
    {
      "epoch": 1.8545069925960749,
      "grad_norm": 34.62489700317383,
      "learning_rate": 1.9168600306760532e-05,
      "loss": 1.7688,
      "step": 15780
    },
    {
      "epoch": 1.8556822188271243,
      "grad_norm": 13.183849334716797,
      "learning_rate": 1.9148936170212766e-05,
      "loss": 1.0323,
      "step": 15790
    },
    {
      "epoch": 1.8568574450581736,
      "grad_norm": 8.485501289367676,
      "learning_rate": 1.9129272033665003e-05,
      "loss": 1.4933,
      "step": 15800
    },
    {
      "epoch": 1.8580326712892232,
      "grad_norm": 16.310163497924805,
      "learning_rate": 1.9109607897117236e-05,
      "loss": 1.6294,
      "step": 15810
    },
    {
      "epoch": 1.8592078975202726,
      "grad_norm": 7.505554676055908,
      "learning_rate": 1.9089943760569473e-05,
      "loss": 1.4875,
      "step": 15820
    },
    {
      "epoch": 1.8603831237513222,
      "grad_norm": 24.004276275634766,
      "learning_rate": 1.907027962402171e-05,
      "loss": 1.8059,
      "step": 15830
    },
    {
      "epoch": 1.8615583499823716,
      "grad_norm": 9.14604663848877,
      "learning_rate": 1.9050615487473947e-05,
      "loss": 1.4346,
      "step": 15840
    },
    {
      "epoch": 1.862733576213421,
      "grad_norm": 7.27960729598999,
      "learning_rate": 1.903095135092618e-05,
      "loss": 1.2745,
      "step": 15850
    },
    {
      "epoch": 1.8639088024444705,
      "grad_norm": 8.404890060424805,
      "learning_rate": 1.9011287214378418e-05,
      "loss": 1.615,
      "step": 15860
    },
    {
      "epoch": 1.8650840286755201,
      "grad_norm": 28.51494598388672,
      "learning_rate": 1.899162307783065e-05,
      "loss": 1.7927,
      "step": 15870
    },
    {
      "epoch": 1.8662592549065695,
      "grad_norm": 12.503304481506348,
      "learning_rate": 1.897195894128289e-05,
      "loss": 2.0015,
      "step": 15880
    },
    {
      "epoch": 1.867434481137619,
      "grad_norm": 16.425954818725586,
      "learning_rate": 1.8952294804735125e-05,
      "loss": 1.3693,
      "step": 15890
    },
    {
      "epoch": 1.8686097073686685,
      "grad_norm": 11.642315864562988,
      "learning_rate": 1.893263066818736e-05,
      "loss": 1.4905,
      "step": 15900
    },
    {
      "epoch": 1.869784933599718,
      "grad_norm": 3.6006317138671875,
      "learning_rate": 1.8912966531639596e-05,
      "loss": 1.3679,
      "step": 15910
    },
    {
      "epoch": 1.8709601598307675,
      "grad_norm": 10.858203887939453,
      "learning_rate": 1.8893302395091833e-05,
      "loss": 1.5816,
      "step": 15920
    },
    {
      "epoch": 1.8721353860618168,
      "grad_norm": 15.298966407775879,
      "learning_rate": 1.887363825854407e-05,
      "loss": 1.4788,
      "step": 15930
    },
    {
      "epoch": 1.8733106122928662,
      "grad_norm": 17.79996681213379,
      "learning_rate": 1.8853974121996304e-05,
      "loss": 1.9889,
      "step": 15940
    },
    {
      "epoch": 1.8744858385239158,
      "grad_norm": 14.215415000915527,
      "learning_rate": 1.883430998544854e-05,
      "loss": 1.5041,
      "step": 15950
    },
    {
      "epoch": 1.8756610647549654,
      "grad_norm": 10.458852767944336,
      "learning_rate": 1.8814645848900774e-05,
      "loss": 1.5133,
      "step": 15960
    },
    {
      "epoch": 1.8768362909860148,
      "grad_norm": 8.840493202209473,
      "learning_rate": 1.879498171235301e-05,
      "loss": 1.5758,
      "step": 15970
    },
    {
      "epoch": 1.8780115172170642,
      "grad_norm": 6.272080898284912,
      "learning_rate": 1.877531757580525e-05,
      "loss": 1.3326,
      "step": 15980
    },
    {
      "epoch": 1.8791867434481138,
      "grad_norm": 20.08355712890625,
      "learning_rate": 1.8755653439257485e-05,
      "loss": 1.5035,
      "step": 15990
    },
    {
      "epoch": 1.8803619696791634,
      "grad_norm": 9.066788673400879,
      "learning_rate": 1.873598930270972e-05,
      "loss": 1.3789,
      "step": 16000
    },
    {
      "epoch": 1.8815371959102127,
      "grad_norm": 10.354713439941406,
      "learning_rate": 1.8716325166161956e-05,
      "loss": 1.5884,
      "step": 16010
    },
    {
      "epoch": 1.882712422141262,
      "grad_norm": 23.058609008789062,
      "learning_rate": 1.869666102961419e-05,
      "loss": 2.1413,
      "step": 16020
    },
    {
      "epoch": 1.8838876483723117,
      "grad_norm": 12.43994140625,
      "learning_rate": 1.8676996893066427e-05,
      "loss": 1.3063,
      "step": 16030
    },
    {
      "epoch": 1.8850628746033613,
      "grad_norm": 19.348148345947266,
      "learning_rate": 1.865733275651866e-05,
      "loss": 1.161,
      "step": 16040
    },
    {
      "epoch": 1.8862381008344107,
      "grad_norm": 11.229622840881348,
      "learning_rate": 1.8637668619970897e-05,
      "loss": 1.3643,
      "step": 16050
    },
    {
      "epoch": 1.88741332706546,
      "grad_norm": 14.46694278717041,
      "learning_rate": 1.8618004483423134e-05,
      "loss": 2.0819,
      "step": 16060
    },
    {
      "epoch": 1.8885885532965094,
      "grad_norm": 20.036365509033203,
      "learning_rate": 1.859834034687537e-05,
      "loss": 1.4181,
      "step": 16070
    },
    {
      "epoch": 1.889763779527559,
      "grad_norm": 4.250028610229492,
      "learning_rate": 1.8578676210327605e-05,
      "loss": 1.5371,
      "step": 16080
    },
    {
      "epoch": 1.8909390057586086,
      "grad_norm": 25.050291061401367,
      "learning_rate": 1.8559012073779842e-05,
      "loss": 1.8612,
      "step": 16090
    },
    {
      "epoch": 1.892114231989658,
      "grad_norm": 4.280246734619141,
      "learning_rate": 1.8539347937232075e-05,
      "loss": 1.1585,
      "step": 16100
    },
    {
      "epoch": 1.8932894582207074,
      "grad_norm": 16.050931930541992,
      "learning_rate": 1.8519683800684312e-05,
      "loss": 1.2904,
      "step": 16110
    },
    {
      "epoch": 1.894464684451757,
      "grad_norm": 9.689634323120117,
      "learning_rate": 1.8500019664136546e-05,
      "loss": 1.6821,
      "step": 16120
    },
    {
      "epoch": 1.8956399106828066,
      "grad_norm": 17.52216911315918,
      "learning_rate": 1.8480355527588786e-05,
      "loss": 1.4703,
      "step": 16130
    },
    {
      "epoch": 1.896815136913856,
      "grad_norm": 5.659823894500732,
      "learning_rate": 1.846069139104102e-05,
      "loss": 1.0007,
      "step": 16140
    },
    {
      "epoch": 1.8979903631449053,
      "grad_norm": 10.438307762145996,
      "learning_rate": 1.8441027254493257e-05,
      "loss": 1.4715,
      "step": 16150
    },
    {
      "epoch": 1.899165589375955,
      "grad_norm": 17.221725463867188,
      "learning_rate": 1.842136311794549e-05,
      "loss": 2.3016,
      "step": 16160
    },
    {
      "epoch": 1.9003408156070043,
      "grad_norm": 16.70512580871582,
      "learning_rate": 1.8401698981397728e-05,
      "loss": 1.2332,
      "step": 16170
    },
    {
      "epoch": 1.901516041838054,
      "grad_norm": 10.889735221862793,
      "learning_rate": 1.8382034844849965e-05,
      "loss": 1.4831,
      "step": 16180
    },
    {
      "epoch": 1.9026912680691033,
      "grad_norm": 27.12200164794922,
      "learning_rate": 1.8362370708302198e-05,
      "loss": 2.1669,
      "step": 16190
    },
    {
      "epoch": 1.9038664943001526,
      "grad_norm": 13.53518009185791,
      "learning_rate": 1.8342706571754435e-05,
      "loss": 1.705,
      "step": 16200
    },
    {
      "epoch": 1.9050417205312022,
      "grad_norm": 13.851234436035156,
      "learning_rate": 1.8323042435206672e-05,
      "loss": 1.394,
      "step": 16210
    },
    {
      "epoch": 1.9062169467622518,
      "grad_norm": 8.630250930786133,
      "learning_rate": 1.8303378298658906e-05,
      "loss": 1.3291,
      "step": 16220
    },
    {
      "epoch": 1.9073921729933012,
      "grad_norm": 15.144445419311523,
      "learning_rate": 1.8283714162111143e-05,
      "loss": 1.1048,
      "step": 16230
    },
    {
      "epoch": 1.9085673992243506,
      "grad_norm": 4.658485412597656,
      "learning_rate": 1.826405002556338e-05,
      "loss": 1.0828,
      "step": 16240
    },
    {
      "epoch": 1.9097426254554002,
      "grad_norm": 24.802234649658203,
      "learning_rate": 1.8244385889015613e-05,
      "loss": 1.7807,
      "step": 16250
    },
    {
      "epoch": 1.9109178516864498,
      "grad_norm": 11.185454368591309,
      "learning_rate": 1.822472175246785e-05,
      "loss": 1.5562,
      "step": 16260
    },
    {
      "epoch": 1.9120930779174992,
      "grad_norm": 7.890844345092773,
      "learning_rate": 1.8205057615920084e-05,
      "loss": 1.2878,
      "step": 16270
    },
    {
      "epoch": 1.9132683041485485,
      "grad_norm": 7.199370384216309,
      "learning_rate": 1.8185393479372324e-05,
      "loss": 1.4617,
      "step": 16280
    },
    {
      "epoch": 1.914443530379598,
      "grad_norm": 13.615872383117676,
      "learning_rate": 1.8165729342824558e-05,
      "loss": 1.6291,
      "step": 16290
    },
    {
      "epoch": 1.9156187566106475,
      "grad_norm": 8.589333534240723,
      "learning_rate": 1.8146065206276795e-05,
      "loss": 1.7751,
      "step": 16300
    },
    {
      "epoch": 1.916793982841697,
      "grad_norm": 15.872179985046387,
      "learning_rate": 1.812640106972903e-05,
      "loss": 1.7143,
      "step": 16310
    },
    {
      "epoch": 1.9179692090727465,
      "grad_norm": 9.140477180480957,
      "learning_rate": 1.8106736933181266e-05,
      "loss": 1.5163,
      "step": 16320
    },
    {
      "epoch": 1.9191444353037959,
      "grad_norm": 11.773847579956055,
      "learning_rate": 1.80870727966335e-05,
      "loss": 1.2889,
      "step": 16330
    },
    {
      "epoch": 1.9203196615348455,
      "grad_norm": 16.07056427001953,
      "learning_rate": 1.8067408660085736e-05,
      "loss": 2.1825,
      "step": 16340
    },
    {
      "epoch": 1.921494887765895,
      "grad_norm": 20.641902923583984,
      "learning_rate": 1.804774452353797e-05,
      "loss": 2.7248,
      "step": 16350
    },
    {
      "epoch": 1.9226701139969444,
      "grad_norm": 18.364543914794922,
      "learning_rate": 1.802808038699021e-05,
      "loss": 1.9636,
      "step": 16360
    },
    {
      "epoch": 1.9238453402279938,
      "grad_norm": 12.15346908569336,
      "learning_rate": 1.8008416250442444e-05,
      "loss": 1.8313,
      "step": 16370
    },
    {
      "epoch": 1.9250205664590434,
      "grad_norm": 9.450745582580566,
      "learning_rate": 1.798875211389468e-05,
      "loss": 2.1346,
      "step": 16380
    },
    {
      "epoch": 1.926195792690093,
      "grad_norm": 16.05853271484375,
      "learning_rate": 1.7969087977346915e-05,
      "loss": 1.9819,
      "step": 16390
    },
    {
      "epoch": 1.9273710189211424,
      "grad_norm": 18.698957443237305,
      "learning_rate": 1.794942384079915e-05,
      "loss": 1.4216,
      "step": 16400
    },
    {
      "epoch": 1.9285462451521918,
      "grad_norm": 24.120935440063477,
      "learning_rate": 1.7929759704251385e-05,
      "loss": 1.3903,
      "step": 16410
    },
    {
      "epoch": 1.9297214713832411,
      "grad_norm": 26.493396759033203,
      "learning_rate": 1.7910095567703622e-05,
      "loss": 1.3626,
      "step": 16420
    },
    {
      "epoch": 1.9308966976142907,
      "grad_norm": 9.491853713989258,
      "learning_rate": 1.789043143115586e-05,
      "loss": 1.5435,
      "step": 16430
    },
    {
      "epoch": 1.9320719238453403,
      "grad_norm": 11.292830467224121,
      "learning_rate": 1.7870767294608096e-05,
      "loss": 1.3092,
      "step": 16440
    },
    {
      "epoch": 1.9332471500763897,
      "grad_norm": 20.610729217529297,
      "learning_rate": 1.785110315806033e-05,
      "loss": 1.6158,
      "step": 16450
    },
    {
      "epoch": 1.934422376307439,
      "grad_norm": 5.404063701629639,
      "learning_rate": 1.7831439021512567e-05,
      "loss": 1.8311,
      "step": 16460
    },
    {
      "epoch": 1.9355976025384887,
      "grad_norm": 7.994711875915527,
      "learning_rate": 1.78117748849648e-05,
      "loss": 1.9422,
      "step": 16470
    },
    {
      "epoch": 1.9367728287695383,
      "grad_norm": 16.428604125976562,
      "learning_rate": 1.7792110748417037e-05,
      "loss": 1.5912,
      "step": 16480
    },
    {
      "epoch": 1.9379480550005876,
      "grad_norm": 14.051555633544922,
      "learning_rate": 1.7772446611869274e-05,
      "loss": 1.7075,
      "step": 16490
    },
    {
      "epoch": 1.939123281231637,
      "grad_norm": 16.989072799682617,
      "learning_rate": 1.7752782475321508e-05,
      "loss": 1.5478,
      "step": 16500
    },
    {
      "epoch": 1.9402985074626866,
      "grad_norm": 15.333877563476562,
      "learning_rate": 1.7733118338773745e-05,
      "loss": 2.0339,
      "step": 16510
    },
    {
      "epoch": 1.941473733693736,
      "grad_norm": 12.9176025390625,
      "learning_rate": 1.7713454202225982e-05,
      "loss": 1.6628,
      "step": 16520
    },
    {
      "epoch": 1.9426489599247856,
      "grad_norm": 8.703117370605469,
      "learning_rate": 1.769379006567822e-05,
      "loss": 1.726,
      "step": 16530
    },
    {
      "epoch": 1.943824186155835,
      "grad_norm": 18.127628326416016,
      "learning_rate": 1.7674125929130453e-05,
      "loss": 1.6832,
      "step": 16540
    },
    {
      "epoch": 1.9449994123868843,
      "grad_norm": 37.477577209472656,
      "learning_rate": 1.765446179258269e-05,
      "loss": 1.1885,
      "step": 16550
    },
    {
      "epoch": 1.946174638617934,
      "grad_norm": 22.099246978759766,
      "learning_rate": 1.7634797656034923e-05,
      "loss": 1.819,
      "step": 16560
    },
    {
      "epoch": 1.9473498648489835,
      "grad_norm": 10.099214553833008,
      "learning_rate": 1.761513351948716e-05,
      "loss": 1.5194,
      "step": 16570
    },
    {
      "epoch": 1.948525091080033,
      "grad_norm": 17.57452392578125,
      "learning_rate": 1.7595469382939394e-05,
      "loss": 2.0314,
      "step": 16580
    },
    {
      "epoch": 1.9497003173110823,
      "grad_norm": 22.047142028808594,
      "learning_rate": 1.7575805246391634e-05,
      "loss": 1.8866,
      "step": 16590
    },
    {
      "epoch": 1.9508755435421319,
      "grad_norm": 21.920108795166016,
      "learning_rate": 1.7556141109843868e-05,
      "loss": 1.5285,
      "step": 16600
    },
    {
      "epoch": 1.9520507697731815,
      "grad_norm": 8.862841606140137,
      "learning_rate": 1.7536476973296105e-05,
      "loss": 2.0127,
      "step": 16610
    },
    {
      "epoch": 1.9532259960042309,
      "grad_norm": 19.938579559326172,
      "learning_rate": 1.751681283674834e-05,
      "loss": 1.1755,
      "step": 16620
    },
    {
      "epoch": 1.9544012222352802,
      "grad_norm": 21.19074249267578,
      "learning_rate": 1.7497148700200575e-05,
      "loss": 1.3752,
      "step": 16630
    },
    {
      "epoch": 1.9555764484663296,
      "grad_norm": 11.047587394714355,
      "learning_rate": 1.747748456365281e-05,
      "loss": 1.1056,
      "step": 16640
    },
    {
      "epoch": 1.9567516746973792,
      "grad_norm": 12.672566413879395,
      "learning_rate": 1.7457820427105046e-05,
      "loss": 1.5965,
      "step": 16650
    },
    {
      "epoch": 1.9579269009284288,
      "grad_norm": 10.07076358795166,
      "learning_rate": 1.7438156290557283e-05,
      "loss": 1.4221,
      "step": 16660
    },
    {
      "epoch": 1.9591021271594782,
      "grad_norm": 19.5220890045166,
      "learning_rate": 1.741849215400952e-05,
      "loss": 2.0368,
      "step": 16670
    },
    {
      "epoch": 1.9602773533905276,
      "grad_norm": 22.06297492980957,
      "learning_rate": 1.7398828017461754e-05,
      "loss": 1.6705,
      "step": 16680
    },
    {
      "epoch": 1.9614525796215772,
      "grad_norm": 7.427207946777344,
      "learning_rate": 1.737916388091399e-05,
      "loss": 1.2596,
      "step": 16690
    },
    {
      "epoch": 1.9626278058526268,
      "grad_norm": 22.706241607666016,
      "learning_rate": 1.7359499744366224e-05,
      "loss": 1.1892,
      "step": 16700
    },
    {
      "epoch": 1.9638030320836761,
      "grad_norm": 36.674034118652344,
      "learning_rate": 1.733983560781846e-05,
      "loss": 1.7564,
      "step": 16710
    },
    {
      "epoch": 1.9649782583147255,
      "grad_norm": 17.77531623840332,
      "learning_rate": 1.7320171471270695e-05,
      "loss": 1.7788,
      "step": 16720
    },
    {
      "epoch": 1.966153484545775,
      "grad_norm": 16.37868881225586,
      "learning_rate": 1.7300507334722932e-05,
      "loss": 1.8163,
      "step": 16730
    },
    {
      "epoch": 1.9673287107768247,
      "grad_norm": 24.446521759033203,
      "learning_rate": 1.728084319817517e-05,
      "loss": 1.7667,
      "step": 16740
    },
    {
      "epoch": 1.968503937007874,
      "grad_norm": 36.728851318359375,
      "learning_rate": 1.7261179061627406e-05,
      "loss": 1.6875,
      "step": 16750
    },
    {
      "epoch": 1.9696791632389234,
      "grad_norm": 17.801233291625977,
      "learning_rate": 1.724151492507964e-05,
      "loss": 1.3565,
      "step": 16760
    },
    {
      "epoch": 1.9708543894699728,
      "grad_norm": 16.241514205932617,
      "learning_rate": 1.7221850788531877e-05,
      "loss": 1.7821,
      "step": 16770
    },
    {
      "epoch": 1.9720296157010224,
      "grad_norm": 22.021936416625977,
      "learning_rate": 1.720218665198411e-05,
      "loss": 1.3888,
      "step": 16780
    },
    {
      "epoch": 1.973204841932072,
      "grad_norm": 22.679141998291016,
      "learning_rate": 1.7182522515436347e-05,
      "loss": 1.9362,
      "step": 16790
    },
    {
      "epoch": 1.9743800681631214,
      "grad_norm": 8.430717468261719,
      "learning_rate": 1.7162858378888584e-05,
      "loss": 1.474,
      "step": 16800
    },
    {
      "epoch": 1.9755552943941708,
      "grad_norm": 13.787619590759277,
      "learning_rate": 1.714319424234082e-05,
      "loss": 1.4454,
      "step": 16810
    },
    {
      "epoch": 1.9767305206252204,
      "grad_norm": 29.589284896850586,
      "learning_rate": 1.7123530105793055e-05,
      "loss": 1.8239,
      "step": 16820
    },
    {
      "epoch": 1.97790574685627,
      "grad_norm": 8.779520034790039,
      "learning_rate": 1.7103865969245292e-05,
      "loss": 1.2961,
      "step": 16830
    },
    {
      "epoch": 1.9790809730873193,
      "grad_norm": 11.515761375427246,
      "learning_rate": 1.708420183269753e-05,
      "loss": 1.376,
      "step": 16840
    },
    {
      "epoch": 1.9802561993183687,
      "grad_norm": 51.77559280395508,
      "learning_rate": 1.7064537696149762e-05,
      "loss": 1.7826,
      "step": 16850
    },
    {
      "epoch": 1.9814314255494183,
      "grad_norm": 7.802410125732422,
      "learning_rate": 1.7044873559602e-05,
      "loss": 1.6434,
      "step": 16860
    },
    {
      "epoch": 1.9826066517804677,
      "grad_norm": 10.883423805236816,
      "learning_rate": 1.7025209423054233e-05,
      "loss": 1.9114,
      "step": 16870
    },
    {
      "epoch": 1.9837818780115173,
      "grad_norm": 12.361852645874023,
      "learning_rate": 1.700554528650647e-05,
      "loss": 1.9562,
      "step": 16880
    },
    {
      "epoch": 1.9849571042425667,
      "grad_norm": 14.339815139770508,
      "learning_rate": 1.6985881149958707e-05,
      "loss": 1.0982,
      "step": 16890
    },
    {
      "epoch": 1.986132330473616,
      "grad_norm": 19.644376754760742,
      "learning_rate": 1.6966217013410944e-05,
      "loss": 1.8069,
      "step": 16900
    },
    {
      "epoch": 1.9873075567046656,
      "grad_norm": 8.552693367004395,
      "learning_rate": 1.6946552876863178e-05,
      "loss": 1.2847,
      "step": 16910
    },
    {
      "epoch": 1.9884827829357152,
      "grad_norm": 12.200881958007812,
      "learning_rate": 1.6926888740315415e-05,
      "loss": 1.3813,
      "step": 16920
    },
    {
      "epoch": 1.9896580091667646,
      "grad_norm": 30.344301223754883,
      "learning_rate": 1.6907224603767648e-05,
      "loss": 2.0802,
      "step": 16930
    },
    {
      "epoch": 1.990833235397814,
      "grad_norm": 33.733428955078125,
      "learning_rate": 1.6887560467219885e-05,
      "loss": 2.0929,
      "step": 16940
    },
    {
      "epoch": 1.9920084616288636,
      "grad_norm": 16.987903594970703,
      "learning_rate": 1.686789633067212e-05,
      "loss": 1.2033,
      "step": 16950
    },
    {
      "epoch": 1.9931836878599132,
      "grad_norm": 13.205116271972656,
      "learning_rate": 1.684823219412436e-05,
      "loss": 1.0836,
      "step": 16960
    },
    {
      "epoch": 1.9943589140909626,
      "grad_norm": 20.262910842895508,
      "learning_rate": 1.6828568057576593e-05,
      "loss": 1.885,
      "step": 16970
    },
    {
      "epoch": 1.995534140322012,
      "grad_norm": 9.915228843688965,
      "learning_rate": 1.680890392102883e-05,
      "loss": 1.4877,
      "step": 16980
    },
    {
      "epoch": 1.9967093665530613,
      "grad_norm": 22.03049659729004,
      "learning_rate": 1.6789239784481063e-05,
      "loss": 1.9584,
      "step": 16990
    },
    {
      "epoch": 1.997884592784111,
      "grad_norm": 16.00684356689453,
      "learning_rate": 1.67695756479333e-05,
      "loss": 2.1019,
      "step": 17000
    },
    {
      "epoch": 1.9990598190151605,
      "grad_norm": 25.503759384155273,
      "learning_rate": 1.6749911511385534e-05,
      "loss": 1.525,
      "step": 17010
    },
    {
      "epoch": 2.00023504524621,
      "grad_norm": 11.599724769592285,
      "learning_rate": 1.673024737483777e-05,
      "loss": 1.5429,
      "step": 17020
    },
    {
      "epoch": 2.0014102714772593,
      "grad_norm": 12.23965835571289,
      "learning_rate": 1.6710583238290005e-05,
      "loss": 1.6127,
      "step": 17030
    },
    {
      "epoch": 2.002585497708309,
      "grad_norm": 18.073652267456055,
      "learning_rate": 1.6690919101742245e-05,
      "loss": 1.1183,
      "step": 17040
    },
    {
      "epoch": 2.0037607239393584,
      "grad_norm": 16.80194854736328,
      "learning_rate": 1.667125496519448e-05,
      "loss": 1.7537,
      "step": 17050
    },
    {
      "epoch": 2.004935950170408,
      "grad_norm": 7.427468776702881,
      "learning_rate": 1.6651590828646716e-05,
      "loss": 1.616,
      "step": 17060
    },
    {
      "epoch": 2.006111176401457,
      "grad_norm": 35.93980407714844,
      "learning_rate": 1.663192669209895e-05,
      "loss": 1.0241,
      "step": 17070
    },
    {
      "epoch": 2.0072864026325066,
      "grad_norm": 25.335840225219727,
      "learning_rate": 1.6612262555551186e-05,
      "loss": 1.4661,
      "step": 17080
    },
    {
      "epoch": 2.0084616288635564,
      "grad_norm": 27.820341110229492,
      "learning_rate": 1.6592598419003423e-05,
      "loss": 1.9123,
      "step": 17090
    },
    {
      "epoch": 2.0096368550946058,
      "grad_norm": 17.90120506286621,
      "learning_rate": 1.6572934282455657e-05,
      "loss": 1.6075,
      "step": 17100
    },
    {
      "epoch": 2.010812081325655,
      "grad_norm": 25.944026947021484,
      "learning_rate": 1.6553270145907894e-05,
      "loss": 1.3335,
      "step": 17110
    },
    {
      "epoch": 2.0119873075567045,
      "grad_norm": 18.580318450927734,
      "learning_rate": 1.653360600936013e-05,
      "loss": 1.5341,
      "step": 17120
    },
    {
      "epoch": 2.0131625337877543,
      "grad_norm": 10.985544204711914,
      "learning_rate": 1.6513941872812368e-05,
      "loss": 1.6724,
      "step": 17130
    },
    {
      "epoch": 2.0143377600188037,
      "grad_norm": 9.48175048828125,
      "learning_rate": 1.64942777362646e-05,
      "loss": 1.2636,
      "step": 17140
    },
    {
      "epoch": 2.015512986249853,
      "grad_norm": 11.68127727508545,
      "learning_rate": 1.647461359971684e-05,
      "loss": 1.9407,
      "step": 17150
    },
    {
      "epoch": 2.0166882124809025,
      "grad_norm": 16.13878059387207,
      "learning_rate": 1.6454949463169072e-05,
      "loss": 1.6612,
      "step": 17160
    },
    {
      "epoch": 2.017863438711952,
      "grad_norm": 7.447272777557373,
      "learning_rate": 1.643528532662131e-05,
      "loss": 1.1712,
      "step": 17170
    },
    {
      "epoch": 2.0190386649430017,
      "grad_norm": 14.052169799804688,
      "learning_rate": 1.6415621190073543e-05,
      "loss": 1.1732,
      "step": 17180
    },
    {
      "epoch": 2.020213891174051,
      "grad_norm": 27.970298767089844,
      "learning_rate": 1.6395957053525783e-05,
      "loss": 1.4867,
      "step": 17190
    },
    {
      "epoch": 2.0213891174051004,
      "grad_norm": 19.51994514465332,
      "learning_rate": 1.6376292916978017e-05,
      "loss": 1.9091,
      "step": 17200
    },
    {
      "epoch": 2.02256434363615,
      "grad_norm": 27.972782135009766,
      "learning_rate": 1.6356628780430254e-05,
      "loss": 1.9575,
      "step": 17210
    },
    {
      "epoch": 2.0237395698671996,
      "grad_norm": 19.462421417236328,
      "learning_rate": 1.6336964643882487e-05,
      "loss": 1.3287,
      "step": 17220
    },
    {
      "epoch": 2.024914796098249,
      "grad_norm": 26.256561279296875,
      "learning_rate": 1.6317300507334724e-05,
      "loss": 1.2618,
      "step": 17230
    },
    {
      "epoch": 2.0260900223292984,
      "grad_norm": 20.11598014831543,
      "learning_rate": 1.6297636370786958e-05,
      "loss": 1.3255,
      "step": 17240
    },
    {
      "epoch": 2.0272652485603477,
      "grad_norm": 20.219470977783203,
      "learning_rate": 1.6277972234239195e-05,
      "loss": 1.4164,
      "step": 17250
    },
    {
      "epoch": 2.0284404747913976,
      "grad_norm": 10.452157974243164,
      "learning_rate": 1.625830809769143e-05,
      "loss": 1.5148,
      "step": 17260
    },
    {
      "epoch": 2.029615701022447,
      "grad_norm": 12.048491477966309,
      "learning_rate": 1.623864396114367e-05,
      "loss": 1.2992,
      "step": 17270
    },
    {
      "epoch": 2.0307909272534963,
      "grad_norm": 9.305716514587402,
      "learning_rate": 1.6218979824595903e-05,
      "loss": 1.7475,
      "step": 17280
    },
    {
      "epoch": 2.0319661534845457,
      "grad_norm": 22.204727172851562,
      "learning_rate": 1.619931568804814e-05,
      "loss": 1.8842,
      "step": 17290
    },
    {
      "epoch": 2.033141379715595,
      "grad_norm": 8.000293731689453,
      "learning_rate": 1.6179651551500373e-05,
      "loss": 1.286,
      "step": 17300
    },
    {
      "epoch": 2.034316605946645,
      "grad_norm": 48.598506927490234,
      "learning_rate": 1.615998741495261e-05,
      "loss": 1.2896,
      "step": 17310
    },
    {
      "epoch": 2.0354918321776942,
      "grad_norm": 15.290645599365234,
      "learning_rate": 1.6140323278404844e-05,
      "loss": 2.0544,
      "step": 17320
    },
    {
      "epoch": 2.0366670584087436,
      "grad_norm": 15.551870346069336,
      "learning_rate": 1.612065914185708e-05,
      "loss": 1.5997,
      "step": 17330
    },
    {
      "epoch": 2.037842284639793,
      "grad_norm": 25.116485595703125,
      "learning_rate": 1.6100995005309318e-05,
      "loss": 1.275,
      "step": 17340
    },
    {
      "epoch": 2.039017510870843,
      "grad_norm": 5.582879066467285,
      "learning_rate": 1.6081330868761555e-05,
      "loss": 1.0449,
      "step": 17350
    },
    {
      "epoch": 2.040192737101892,
      "grad_norm": 38.477783203125,
      "learning_rate": 1.606166673221379e-05,
      "loss": 1.7207,
      "step": 17360
    },
    {
      "epoch": 2.0413679633329416,
      "grad_norm": 8.777396202087402,
      "learning_rate": 1.6042002595666025e-05,
      "loss": 1.448,
      "step": 17370
    },
    {
      "epoch": 2.042543189563991,
      "grad_norm": 21.032411575317383,
      "learning_rate": 1.602233845911826e-05,
      "loss": 1.4848,
      "step": 17380
    },
    {
      "epoch": 2.0437184157950403,
      "grad_norm": 25.866559982299805,
      "learning_rate": 1.6002674322570496e-05,
      "loss": 1.4438,
      "step": 17390
    },
    {
      "epoch": 2.04489364202609,
      "grad_norm": 41.68614196777344,
      "learning_rate": 1.5983010186022733e-05,
      "loss": 1.3147,
      "step": 17400
    },
    {
      "epoch": 2.0460688682571395,
      "grad_norm": 27.330162048339844,
      "learning_rate": 1.5963346049474967e-05,
      "loss": 1.6596,
      "step": 17410
    },
    {
      "epoch": 2.047244094488189,
      "grad_norm": 12.59191608428955,
      "learning_rate": 1.5943681912927204e-05,
      "loss": 1.5832,
      "step": 17420
    },
    {
      "epoch": 2.0484193207192383,
      "grad_norm": 9.67031478881836,
      "learning_rate": 1.592401777637944e-05,
      "loss": 1.2839,
      "step": 17430
    },
    {
      "epoch": 2.049594546950288,
      "grad_norm": 19.208816528320312,
      "learning_rate": 1.5904353639831678e-05,
      "loss": 1.6237,
      "step": 17440
    },
    {
      "epoch": 2.0507697731813375,
      "grad_norm": 23.503681182861328,
      "learning_rate": 1.588468950328391e-05,
      "loss": 1.6389,
      "step": 17450
    },
    {
      "epoch": 2.051944999412387,
      "grad_norm": 12.132522583007812,
      "learning_rate": 1.586502536673615e-05,
      "loss": 1.4007,
      "step": 17460
    },
    {
      "epoch": 2.053120225643436,
      "grad_norm": 15.82802963256836,
      "learning_rate": 1.5845361230188382e-05,
      "loss": 1.9146,
      "step": 17470
    },
    {
      "epoch": 2.054295451874486,
      "grad_norm": 16.317501068115234,
      "learning_rate": 1.582569709364062e-05,
      "loss": 1.719,
      "step": 17480
    },
    {
      "epoch": 2.0554706781055354,
      "grad_norm": 15.988179206848145,
      "learning_rate": 1.5806032957092856e-05,
      "loss": 1.4345,
      "step": 17490
    },
    {
      "epoch": 2.056645904336585,
      "grad_norm": 11.656445503234863,
      "learning_rate": 1.5786368820545093e-05,
      "loss": 1.483,
      "step": 17500
    },
    {
      "epoch": 2.057821130567634,
      "grad_norm": 8.017309188842773,
      "learning_rate": 1.5766704683997327e-05,
      "loss": 1.4516,
      "step": 17510
    },
    {
      "epoch": 2.0589963567986835,
      "grad_norm": 3.369671106338501,
      "learning_rate": 1.5747040547449564e-05,
      "loss": 1.2032,
      "step": 17520
    },
    {
      "epoch": 2.0601715830297334,
      "grad_norm": 13.179166793823242,
      "learning_rate": 1.5727376410901797e-05,
      "loss": 1.6706,
      "step": 17530
    },
    {
      "epoch": 2.0613468092607827,
      "grad_norm": 7.350793361663818,
      "learning_rate": 1.5707712274354034e-05,
      "loss": 1.4449,
      "step": 17540
    },
    {
      "epoch": 2.062522035491832,
      "grad_norm": 22.081037521362305,
      "learning_rate": 1.5688048137806268e-05,
      "loss": 1.4,
      "step": 17550
    },
    {
      "epoch": 2.0636972617228815,
      "grad_norm": 15.42470932006836,
      "learning_rate": 1.5668384001258505e-05,
      "loss": 0.9583,
      "step": 17560
    },
    {
      "epoch": 2.0648724879539313,
      "grad_norm": 18.785905838012695,
      "learning_rate": 1.5648719864710742e-05,
      "loss": 1.1673,
      "step": 17570
    },
    {
      "epoch": 2.0660477141849807,
      "grad_norm": 10.969202041625977,
      "learning_rate": 1.562905572816298e-05,
      "loss": 0.9643,
      "step": 17580
    },
    {
      "epoch": 2.06722294041603,
      "grad_norm": 20.32277488708496,
      "learning_rate": 1.5609391591615212e-05,
      "loss": 1.1321,
      "step": 17590
    },
    {
      "epoch": 2.0683981666470794,
      "grad_norm": 22.105915069580078,
      "learning_rate": 1.558972745506745e-05,
      "loss": 1.4312,
      "step": 17600
    },
    {
      "epoch": 2.0695733928781292,
      "grad_norm": 9.834653854370117,
      "learning_rate": 1.5570063318519683e-05,
      "loss": 1.663,
      "step": 17610
    },
    {
      "epoch": 2.0707486191091786,
      "grad_norm": 20.082103729248047,
      "learning_rate": 1.555039918197192e-05,
      "loss": 1.4432,
      "step": 17620
    },
    {
      "epoch": 2.071923845340228,
      "grad_norm": 27.670278549194336,
      "learning_rate": 1.5530735045424154e-05,
      "loss": 1.9243,
      "step": 17630
    },
    {
      "epoch": 2.0730990715712774,
      "grad_norm": 9.036933898925781,
      "learning_rate": 1.5511070908876394e-05,
      "loss": 1.3892,
      "step": 17640
    },
    {
      "epoch": 2.0742742978023267,
      "grad_norm": 8.163311958312988,
      "learning_rate": 1.5491406772328628e-05,
      "loss": 0.818,
      "step": 17650
    },
    {
      "epoch": 2.0754495240333766,
      "grad_norm": 13.027979850769043,
      "learning_rate": 1.5471742635780865e-05,
      "loss": 1.0386,
      "step": 17660
    },
    {
      "epoch": 2.076624750264426,
      "grad_norm": 16.670310974121094,
      "learning_rate": 1.5452078499233098e-05,
      "loss": 1.0838,
      "step": 17670
    },
    {
      "epoch": 2.0777999764954753,
      "grad_norm": 17.19547462463379,
      "learning_rate": 1.5432414362685335e-05,
      "loss": 1.2956,
      "step": 17680
    },
    {
      "epoch": 2.0789752027265247,
      "grad_norm": 13.681255340576172,
      "learning_rate": 1.5412750226137572e-05,
      "loss": 1.6706,
      "step": 17690
    },
    {
      "epoch": 2.0801504289575745,
      "grad_norm": 15.56398868560791,
      "learning_rate": 1.5393086089589806e-05,
      "loss": 0.9958,
      "step": 17700
    },
    {
      "epoch": 2.081325655188624,
      "grad_norm": 11.008705139160156,
      "learning_rate": 1.5373421953042043e-05,
      "loss": 1.4124,
      "step": 17710
    },
    {
      "epoch": 2.0825008814196733,
      "grad_norm": 17.812284469604492,
      "learning_rate": 1.535375781649428e-05,
      "loss": 1.2139,
      "step": 17720
    },
    {
      "epoch": 2.0836761076507226,
      "grad_norm": 21.548702239990234,
      "learning_rate": 1.5334093679946513e-05,
      "loss": 1.4625,
      "step": 17730
    },
    {
      "epoch": 2.0848513338817725,
      "grad_norm": 26.484764099121094,
      "learning_rate": 1.531442954339875e-05,
      "loss": 1.8529,
      "step": 17740
    },
    {
      "epoch": 2.086026560112822,
      "grad_norm": 17.959922790527344,
      "learning_rate": 1.5294765406850987e-05,
      "loss": 1.1694,
      "step": 17750
    },
    {
      "epoch": 2.087201786343871,
      "grad_norm": 13.045345306396484,
      "learning_rate": 1.527510127030322e-05,
      "loss": 1.4307,
      "step": 17760
    },
    {
      "epoch": 2.0883770125749206,
      "grad_norm": 20.319324493408203,
      "learning_rate": 1.5255437133755456e-05,
      "loss": 1.6998,
      "step": 17770
    },
    {
      "epoch": 2.08955223880597,
      "grad_norm": 17.567691802978516,
      "learning_rate": 1.5235772997207692e-05,
      "loss": 1.7245,
      "step": 17780
    },
    {
      "epoch": 2.09072746503702,
      "grad_norm": 21.90340805053711,
      "learning_rate": 1.5216108860659927e-05,
      "loss": 1.4089,
      "step": 17790
    },
    {
      "epoch": 2.091902691268069,
      "grad_norm": 11.877678871154785,
      "learning_rate": 1.5196444724112166e-05,
      "loss": 1.4519,
      "step": 17800
    },
    {
      "epoch": 2.0930779174991185,
      "grad_norm": 40.12141418457031,
      "learning_rate": 1.5176780587564401e-05,
      "loss": 1.1061,
      "step": 17810
    },
    {
      "epoch": 2.094253143730168,
      "grad_norm": 12.880402565002441,
      "learning_rate": 1.5157116451016636e-05,
      "loss": 1.0034,
      "step": 17820
    },
    {
      "epoch": 2.0954283699612177,
      "grad_norm": 10.416561126708984,
      "learning_rate": 1.5137452314468872e-05,
      "loss": 1.4283,
      "step": 17830
    },
    {
      "epoch": 2.096603596192267,
      "grad_norm": 10.388070106506348,
      "learning_rate": 1.5117788177921109e-05,
      "loss": 1.61,
      "step": 17840
    },
    {
      "epoch": 2.0977788224233165,
      "grad_norm": 18.760875701904297,
      "learning_rate": 1.5098124041373344e-05,
      "loss": 1.0312,
      "step": 17850
    },
    {
      "epoch": 2.098954048654366,
      "grad_norm": 12.197856903076172,
      "learning_rate": 1.507845990482558e-05,
      "loss": 0.8183,
      "step": 17860
    },
    {
      "epoch": 2.1001292748854152,
      "grad_norm": 11.679024696350098,
      "learning_rate": 1.5058795768277816e-05,
      "loss": 1.2241,
      "step": 17870
    },
    {
      "epoch": 2.101304501116465,
      "grad_norm": 13.615242958068848,
      "learning_rate": 1.5039131631730052e-05,
      "loss": 0.9503,
      "step": 17880
    },
    {
      "epoch": 2.1024797273475144,
      "grad_norm": 16.466005325317383,
      "learning_rate": 1.5019467495182289e-05,
      "loss": 1.5807,
      "step": 17890
    },
    {
      "epoch": 2.103654953578564,
      "grad_norm": 19.574031829833984,
      "learning_rate": 1.4999803358634524e-05,
      "loss": 1.2666,
      "step": 17900
    },
    {
      "epoch": 2.104830179809613,
      "grad_norm": 10.571694374084473,
      "learning_rate": 1.498013922208676e-05,
      "loss": 1.3831,
      "step": 17910
    },
    {
      "epoch": 2.106005406040663,
      "grad_norm": 11.39920425415039,
      "learning_rate": 1.4960475085538994e-05,
      "loss": 1.1833,
      "step": 17920
    },
    {
      "epoch": 2.1071806322717124,
      "grad_norm": 14.532023429870605,
      "learning_rate": 1.494081094899123e-05,
      "loss": 1.1881,
      "step": 17930
    },
    {
      "epoch": 2.1083558585027617,
      "grad_norm": 20.63963508605957,
      "learning_rate": 1.4921146812443465e-05,
      "loss": 1.626,
      "step": 17940
    },
    {
      "epoch": 2.109531084733811,
      "grad_norm": 15.293091773986816,
      "learning_rate": 1.4901482675895704e-05,
      "loss": 1.0959,
      "step": 17950
    },
    {
      "epoch": 2.110706310964861,
      "grad_norm": 8.295255661010742,
      "learning_rate": 1.4881818539347939e-05,
      "loss": 1.0512,
      "step": 17960
    },
    {
      "epoch": 2.1118815371959103,
      "grad_norm": 2.3863024711608887,
      "learning_rate": 1.4862154402800174e-05,
      "loss": 1.4948,
      "step": 17970
    },
    {
      "epoch": 2.1130567634269597,
      "grad_norm": 4.357798099517822,
      "learning_rate": 1.484249026625241e-05,
      "loss": 1.4979,
      "step": 17980
    },
    {
      "epoch": 2.114231989658009,
      "grad_norm": 28.388656616210938,
      "learning_rate": 1.4822826129704645e-05,
      "loss": 1.4334,
      "step": 17990
    },
    {
      "epoch": 2.1154072158890584,
      "grad_norm": 13.784269332885742,
      "learning_rate": 1.480316199315688e-05,
      "loss": 1.2463,
      "step": 18000
    },
    {
      "epoch": 2.1165824421201083,
      "grad_norm": 12.605283737182617,
      "learning_rate": 1.4783497856609116e-05,
      "loss": 1.3916,
      "step": 18010
    },
    {
      "epoch": 2.1177576683511576,
      "grad_norm": 21.748376846313477,
      "learning_rate": 1.4763833720061354e-05,
      "loss": 1.0898,
      "step": 18020
    },
    {
      "epoch": 2.118932894582207,
      "grad_norm": 6.995212554931641,
      "learning_rate": 1.474416958351359e-05,
      "loss": 1.1879,
      "step": 18030
    },
    {
      "epoch": 2.1201081208132564,
      "grad_norm": 11.11214542388916,
      "learning_rate": 1.4724505446965825e-05,
      "loss": 1.8558,
      "step": 18040
    },
    {
      "epoch": 2.121283347044306,
      "grad_norm": 23.45680809020996,
      "learning_rate": 1.470484131041806e-05,
      "loss": 1.1473,
      "step": 18050
    },
    {
      "epoch": 2.1224585732753556,
      "grad_norm": 10.613316535949707,
      "learning_rate": 1.4685177173870296e-05,
      "loss": 1.442,
      "step": 18060
    },
    {
      "epoch": 2.123633799506405,
      "grad_norm": 27.7166805267334,
      "learning_rate": 1.4665513037322531e-05,
      "loss": 1.343,
      "step": 18070
    },
    {
      "epoch": 2.1248090257374543,
      "grad_norm": 15.64281940460205,
      "learning_rate": 1.4645848900774766e-05,
      "loss": 1.5373,
      "step": 18080
    },
    {
      "epoch": 2.1259842519685037,
      "grad_norm": 6.718026638031006,
      "learning_rate": 1.4626184764227001e-05,
      "loss": 1.1259,
      "step": 18090
    },
    {
      "epoch": 2.1271594781995535,
      "grad_norm": 12.821657180786133,
      "learning_rate": 1.460652062767924e-05,
      "loss": 1.3102,
      "step": 18100
    },
    {
      "epoch": 2.128334704430603,
      "grad_norm": 13.555647850036621,
      "learning_rate": 1.4586856491131475e-05,
      "loss": 1.7504,
      "step": 18110
    },
    {
      "epoch": 2.1295099306616523,
      "grad_norm": 17.524127960205078,
      "learning_rate": 1.456719235458371e-05,
      "loss": 1.3844,
      "step": 18120
    },
    {
      "epoch": 2.1306851568927017,
      "grad_norm": 17.268922805786133,
      "learning_rate": 1.4547528218035946e-05,
      "loss": 1.2896,
      "step": 18130
    },
    {
      "epoch": 2.1318603831237515,
      "grad_norm": 14.460833549499512,
      "learning_rate": 1.4527864081488181e-05,
      "loss": 1.3875,
      "step": 18140
    },
    {
      "epoch": 2.133035609354801,
      "grad_norm": 6.274436950683594,
      "learning_rate": 1.4508199944940418e-05,
      "loss": 1.1537,
      "step": 18150
    },
    {
      "epoch": 2.1342108355858502,
      "grad_norm": 22.373613357543945,
      "learning_rate": 1.4488535808392654e-05,
      "loss": 1.4482,
      "step": 18160
    },
    {
      "epoch": 2.1353860618168996,
      "grad_norm": 11.73139476776123,
      "learning_rate": 1.446887167184489e-05,
      "loss": 1.456,
      "step": 18170
    },
    {
      "epoch": 2.1365612880479494,
      "grad_norm": 12.569405555725098,
      "learning_rate": 1.4449207535297126e-05,
      "loss": 1.3338,
      "step": 18180
    },
    {
      "epoch": 2.137736514278999,
      "grad_norm": 24.968690872192383,
      "learning_rate": 1.4429543398749363e-05,
      "loss": 1.2284,
      "step": 18190
    },
    {
      "epoch": 2.138911740510048,
      "grad_norm": 15.605520248413086,
      "learning_rate": 1.4409879262201598e-05,
      "loss": 1.0898,
      "step": 18200
    },
    {
      "epoch": 2.1400869667410976,
      "grad_norm": 17.658483505249023,
      "learning_rate": 1.4390215125653834e-05,
      "loss": 1.2724,
      "step": 18210
    },
    {
      "epoch": 2.141262192972147,
      "grad_norm": 13.793366432189941,
      "learning_rate": 1.4370550989106069e-05,
      "loss": 1.5125,
      "step": 18220
    },
    {
      "epoch": 2.1424374192031967,
      "grad_norm": 30.1632080078125,
      "learning_rate": 1.4350886852558304e-05,
      "loss": 1.5843,
      "step": 18230
    },
    {
      "epoch": 2.143612645434246,
      "grad_norm": 13.37695598602295,
      "learning_rate": 1.433122271601054e-05,
      "loss": 1.2621,
      "step": 18240
    },
    {
      "epoch": 2.1447878716652955,
      "grad_norm": 11.188718795776367,
      "learning_rate": 1.4311558579462778e-05,
      "loss": 1.208,
      "step": 18250
    },
    {
      "epoch": 2.145963097896345,
      "grad_norm": 16.010765075683594,
      "learning_rate": 1.4291894442915014e-05,
      "loss": 1.474,
      "step": 18260
    },
    {
      "epoch": 2.1471383241273947,
      "grad_norm": 18.7544002532959,
      "learning_rate": 1.4272230306367249e-05,
      "loss": 1.2505,
      "step": 18270
    },
    {
      "epoch": 2.148313550358444,
      "grad_norm": 31.94170379638672,
      "learning_rate": 1.4252566169819484e-05,
      "loss": 1.52,
      "step": 18280
    },
    {
      "epoch": 2.1494887765894934,
      "grad_norm": 12.749345779418945,
      "learning_rate": 1.423290203327172e-05,
      "loss": 1.2383,
      "step": 18290
    },
    {
      "epoch": 2.150664002820543,
      "grad_norm": 11.896326065063477,
      "learning_rate": 1.4213237896723955e-05,
      "loss": 1.1728,
      "step": 18300
    },
    {
      "epoch": 2.1518392290515926,
      "grad_norm": 15.09205436706543,
      "learning_rate": 1.419357376017619e-05,
      "loss": 1.4386,
      "step": 18310
    },
    {
      "epoch": 2.153014455282642,
      "grad_norm": 16.920001983642578,
      "learning_rate": 1.4173909623628425e-05,
      "loss": 1.406,
      "step": 18320
    },
    {
      "epoch": 2.1541896815136914,
      "grad_norm": 11.357789993286133,
      "learning_rate": 1.4154245487080664e-05,
      "loss": 1.8646,
      "step": 18330
    },
    {
      "epoch": 2.1553649077447408,
      "grad_norm": 12.144622802734375,
      "learning_rate": 1.41345813505329e-05,
      "loss": 0.9885,
      "step": 18340
    },
    {
      "epoch": 2.15654013397579,
      "grad_norm": 14.790267944335938,
      "learning_rate": 1.4114917213985135e-05,
      "loss": 1.7552,
      "step": 18350
    },
    {
      "epoch": 2.15771536020684,
      "grad_norm": 27.96136474609375,
      "learning_rate": 1.409525307743737e-05,
      "loss": 1.7531,
      "step": 18360
    },
    {
      "epoch": 2.1588905864378893,
      "grad_norm": 9.862187385559082,
      "learning_rate": 1.4075588940889605e-05,
      "loss": 1.7701,
      "step": 18370
    },
    {
      "epoch": 2.1600658126689387,
      "grad_norm": 9.139622688293457,
      "learning_rate": 1.405592480434184e-05,
      "loss": 1.2164,
      "step": 18380
    },
    {
      "epoch": 2.161241038899988,
      "grad_norm": 6.7819600105285645,
      "learning_rate": 1.4036260667794076e-05,
      "loss": 1.4849,
      "step": 18390
    },
    {
      "epoch": 2.162416265131038,
      "grad_norm": 37.382450103759766,
      "learning_rate": 1.4016596531246315e-05,
      "loss": 1.6581,
      "step": 18400
    },
    {
      "epoch": 2.1635914913620873,
      "grad_norm": 24.253767013549805,
      "learning_rate": 1.399693239469855e-05,
      "loss": 1.4364,
      "step": 18410
    },
    {
      "epoch": 2.1647667175931367,
      "grad_norm": 13.516733169555664,
      "learning_rate": 1.3977268258150785e-05,
      "loss": 1.5065,
      "step": 18420
    },
    {
      "epoch": 2.165941943824186,
      "grad_norm": 20.879596710205078,
      "learning_rate": 1.395760412160302e-05,
      "loss": 1.4314,
      "step": 18430
    },
    {
      "epoch": 2.167117170055236,
      "grad_norm": 14.853495597839355,
      "learning_rate": 1.3937939985055256e-05,
      "loss": 1.2544,
      "step": 18440
    },
    {
      "epoch": 2.1682923962862852,
      "grad_norm": 8.816751480102539,
      "learning_rate": 1.3918275848507493e-05,
      "loss": 1.5301,
      "step": 18450
    },
    {
      "epoch": 2.1694676225173346,
      "grad_norm": 9.912042617797852,
      "learning_rate": 1.3898611711959728e-05,
      "loss": 2.0662,
      "step": 18460
    },
    {
      "epoch": 2.170642848748384,
      "grad_norm": 12.658590316772461,
      "learning_rate": 1.3878947575411963e-05,
      "loss": 1.3353,
      "step": 18470
    },
    {
      "epoch": 2.1718180749794334,
      "grad_norm": 11.644648551940918,
      "learning_rate": 1.38592834388642e-05,
      "loss": 1.6629,
      "step": 18480
    },
    {
      "epoch": 2.172993301210483,
      "grad_norm": 15.506614685058594,
      "learning_rate": 1.3839619302316437e-05,
      "loss": 1.6186,
      "step": 18490
    },
    {
      "epoch": 2.1741685274415326,
      "grad_norm": 18.199235916137695,
      "learning_rate": 1.3819955165768673e-05,
      "loss": 1.714,
      "step": 18500
    },
    {
      "epoch": 2.175343753672582,
      "grad_norm": 17.805233001708984,
      "learning_rate": 1.3800291029220908e-05,
      "loss": 1.6715,
      "step": 18510
    },
    {
      "epoch": 2.1765189799036313,
      "grad_norm": 11.455699920654297,
      "learning_rate": 1.3780626892673143e-05,
      "loss": 1.314,
      "step": 18520
    },
    {
      "epoch": 2.177694206134681,
      "grad_norm": 7.4801554679870605,
      "learning_rate": 1.3760962756125379e-05,
      "loss": 1.3165,
      "step": 18530
    },
    {
      "epoch": 2.1788694323657305,
      "grad_norm": 13.177862167358398,
      "learning_rate": 1.3741298619577614e-05,
      "loss": 1.0229,
      "step": 18540
    },
    {
      "epoch": 2.18004465859678,
      "grad_norm": 15.117539405822754,
      "learning_rate": 1.3721634483029853e-05,
      "loss": 1.3405,
      "step": 18550
    },
    {
      "epoch": 2.1812198848278292,
      "grad_norm": 8.36239242553711,
      "learning_rate": 1.3701970346482088e-05,
      "loss": 1.4766,
      "step": 18560
    },
    {
      "epoch": 2.182395111058879,
      "grad_norm": 27.530237197875977,
      "learning_rate": 1.3682306209934323e-05,
      "loss": 1.4931,
      "step": 18570
    },
    {
      "epoch": 2.1835703372899284,
      "grad_norm": 21.917083740234375,
      "learning_rate": 1.3662642073386559e-05,
      "loss": 1.3232,
      "step": 18580
    },
    {
      "epoch": 2.184745563520978,
      "grad_norm": 15.37550163269043,
      "learning_rate": 1.3642977936838794e-05,
      "loss": 0.9772,
      "step": 18590
    },
    {
      "epoch": 2.185920789752027,
      "grad_norm": 6.664554119110107,
      "learning_rate": 1.362331380029103e-05,
      "loss": 1.4719,
      "step": 18600
    },
    {
      "epoch": 2.1870960159830766,
      "grad_norm": 37.240577697753906,
      "learning_rate": 1.3603649663743265e-05,
      "loss": 1.4922,
      "step": 18610
    },
    {
      "epoch": 2.1882712422141264,
      "grad_norm": 12.793740272521973,
      "learning_rate": 1.35839855271955e-05,
      "loss": 1.0533,
      "step": 18620
    },
    {
      "epoch": 2.1894464684451758,
      "grad_norm": 15.435531616210938,
      "learning_rate": 1.3564321390647739e-05,
      "loss": 1.8706,
      "step": 18630
    },
    {
      "epoch": 2.190621694676225,
      "grad_norm": 11.222558975219727,
      "learning_rate": 1.3544657254099974e-05,
      "loss": 1.3761,
      "step": 18640
    },
    {
      "epoch": 2.1917969209072745,
      "grad_norm": 12.114703178405762,
      "learning_rate": 1.352499311755221e-05,
      "loss": 1.3137,
      "step": 18650
    },
    {
      "epoch": 2.1929721471383243,
      "grad_norm": 26.17620849609375,
      "learning_rate": 1.3505328981004444e-05,
      "loss": 1.5793,
      "step": 18660
    },
    {
      "epoch": 2.1941473733693737,
      "grad_norm": 6.337309837341309,
      "learning_rate": 1.348566484445668e-05,
      "loss": 1.3081,
      "step": 18670
    },
    {
      "epoch": 2.195322599600423,
      "grad_norm": 57.28203201293945,
      "learning_rate": 1.3466000707908915e-05,
      "loss": 1.4884,
      "step": 18680
    },
    {
      "epoch": 2.1964978258314725,
      "grad_norm": 12.656798362731934,
      "learning_rate": 1.344633657136115e-05,
      "loss": 1.3417,
      "step": 18690
    },
    {
      "epoch": 2.197673052062522,
      "grad_norm": 21.63292694091797,
      "learning_rate": 1.3426672434813389e-05,
      "loss": 2.1258,
      "step": 18700
    },
    {
      "epoch": 2.1988482782935717,
      "grad_norm": 42.799537658691406,
      "learning_rate": 1.3407008298265624e-05,
      "loss": 1.451,
      "step": 18710
    },
    {
      "epoch": 2.200023504524621,
      "grad_norm": 13.38326644897461,
      "learning_rate": 1.338734416171786e-05,
      "loss": 1.2884,
      "step": 18720
    },
    {
      "epoch": 2.2011987307556704,
      "grad_norm": 20.218374252319336,
      "learning_rate": 1.3367680025170095e-05,
      "loss": 1.2883,
      "step": 18730
    },
    {
      "epoch": 2.20237395698672,
      "grad_norm": 9.044944763183594,
      "learning_rate": 1.334801588862233e-05,
      "loss": 1.4982,
      "step": 18740
    },
    {
      "epoch": 2.2035491832177696,
      "grad_norm": 9.232402801513672,
      "learning_rate": 1.3328351752074567e-05,
      "loss": 1.6644,
      "step": 18750
    },
    {
      "epoch": 2.204724409448819,
      "grad_norm": 16.34507942199707,
      "learning_rate": 1.3308687615526803e-05,
      "loss": 1.4254,
      "step": 18760
    },
    {
      "epoch": 2.2058996356798684,
      "grad_norm": 18.645097732543945,
      "learning_rate": 1.3289023478979038e-05,
      "loss": 1.2911,
      "step": 18770
    },
    {
      "epoch": 2.2070748619109177,
      "grad_norm": 12.344298362731934,
      "learning_rate": 1.3269359342431275e-05,
      "loss": 1.5816,
      "step": 18780
    },
    {
      "epoch": 2.208250088141967,
      "grad_norm": 32.4506721496582,
      "learning_rate": 1.3249695205883512e-05,
      "loss": 1.5644,
      "step": 18790
    },
    {
      "epoch": 2.209425314373017,
      "grad_norm": 22.729196548461914,
      "learning_rate": 1.3230031069335747e-05,
      "loss": 1.8333,
      "step": 18800
    },
    {
      "epoch": 2.2106005406040663,
      "grad_norm": 12.45051383972168,
      "learning_rate": 1.3210366932787983e-05,
      "loss": 1.1288,
      "step": 18810
    },
    {
      "epoch": 2.2117757668351157,
      "grad_norm": 20.483028411865234,
      "learning_rate": 1.3190702796240218e-05,
      "loss": 1.448,
      "step": 18820
    },
    {
      "epoch": 2.212950993066165,
      "grad_norm": 13.39654541015625,
      "learning_rate": 1.3171038659692453e-05,
      "loss": 1.1115,
      "step": 18830
    },
    {
      "epoch": 2.214126219297215,
      "grad_norm": 14.335679054260254,
      "learning_rate": 1.3151374523144688e-05,
      "loss": 1.4012,
      "step": 18840
    },
    {
      "epoch": 2.2153014455282642,
      "grad_norm": 22.277828216552734,
      "learning_rate": 1.3131710386596927e-05,
      "loss": 1.4458,
      "step": 18850
    },
    {
      "epoch": 2.2164766717593136,
      "grad_norm": 6.56709098815918,
      "learning_rate": 1.3112046250049162e-05,
      "loss": 1.4901,
      "step": 18860
    },
    {
      "epoch": 2.217651897990363,
      "grad_norm": 17.252059936523438,
      "learning_rate": 1.3092382113501398e-05,
      "loss": 1.4867,
      "step": 18870
    },
    {
      "epoch": 2.218827124221413,
      "grad_norm": 26.150714874267578,
      "learning_rate": 1.3072717976953633e-05,
      "loss": 1.3688,
      "step": 18880
    },
    {
      "epoch": 2.220002350452462,
      "grad_norm": 12.26977825164795,
      "learning_rate": 1.3053053840405868e-05,
      "loss": 1.2745,
      "step": 18890
    },
    {
      "epoch": 2.2211775766835116,
      "grad_norm": 11.54210090637207,
      "learning_rate": 1.3033389703858104e-05,
      "loss": 1.1064,
      "step": 18900
    },
    {
      "epoch": 2.222352802914561,
      "grad_norm": 18.978580474853516,
      "learning_rate": 1.3013725567310339e-05,
      "loss": 1.2616,
      "step": 18910
    },
    {
      "epoch": 2.2235280291456103,
      "grad_norm": 22.500675201416016,
      "learning_rate": 1.2994061430762574e-05,
      "loss": 1.6954,
      "step": 18920
    },
    {
      "epoch": 2.22470325537666,
      "grad_norm": 8.30715560913086,
      "learning_rate": 1.2974397294214813e-05,
      "loss": 1.5747,
      "step": 18930
    },
    {
      "epoch": 2.2258784816077095,
      "grad_norm": 11.721282005310059,
      "learning_rate": 1.2954733157667048e-05,
      "loss": 1.1045,
      "step": 18940
    },
    {
      "epoch": 2.227053707838759,
      "grad_norm": 11.549650192260742,
      "learning_rate": 1.2935069021119284e-05,
      "loss": 1.2699,
      "step": 18950
    },
    {
      "epoch": 2.2282289340698083,
      "grad_norm": 11.409252166748047,
      "learning_rate": 1.2915404884571519e-05,
      "loss": 1.3285,
      "step": 18960
    },
    {
      "epoch": 2.229404160300858,
      "grad_norm": 33.66545867919922,
      "learning_rate": 1.2895740748023754e-05,
      "loss": 1.4685,
      "step": 18970
    },
    {
      "epoch": 2.2305793865319075,
      "grad_norm": 21.480497360229492,
      "learning_rate": 1.287607661147599e-05,
      "loss": 1.4585,
      "step": 18980
    },
    {
      "epoch": 2.231754612762957,
      "grad_norm": 17.38619613647461,
      "learning_rate": 1.2856412474928225e-05,
      "loss": 1.6221,
      "step": 18990
    },
    {
      "epoch": 2.232929838994006,
      "grad_norm": 19.99650001525879,
      "learning_rate": 1.283674833838046e-05,
      "loss": 1.464,
      "step": 19000
    },
    {
      "epoch": 2.234105065225056,
      "grad_norm": 16.92568588256836,
      "learning_rate": 1.2817084201832699e-05,
      "loss": 0.9943,
      "step": 19010
    },
    {
      "epoch": 2.2352802914561054,
      "grad_norm": 22.663951873779297,
      "learning_rate": 1.2797420065284934e-05,
      "loss": 1.5452,
      "step": 19020
    },
    {
      "epoch": 2.236455517687155,
      "grad_norm": 11.374763488769531,
      "learning_rate": 1.277775592873717e-05,
      "loss": 1.8519,
      "step": 19030
    },
    {
      "epoch": 2.237630743918204,
      "grad_norm": 19.780324935913086,
      "learning_rate": 1.2758091792189405e-05,
      "loss": 1.5232,
      "step": 19040
    },
    {
      "epoch": 2.2388059701492535,
      "grad_norm": 23.51017951965332,
      "learning_rate": 1.2738427655641642e-05,
      "loss": 0.8771,
      "step": 19050
    },
    {
      "epoch": 2.2399811963803034,
      "grad_norm": 14.932689666748047,
      "learning_rate": 1.2718763519093877e-05,
      "loss": 1.5425,
      "step": 19060
    },
    {
      "epoch": 2.2411564226113527,
      "grad_norm": 21.631633758544922,
      "learning_rate": 1.2699099382546112e-05,
      "loss": 1.0974,
      "step": 19070
    },
    {
      "epoch": 2.242331648842402,
      "grad_norm": 17.417865753173828,
      "learning_rate": 1.267943524599835e-05,
      "loss": 1.3276,
      "step": 19080
    },
    {
      "epoch": 2.2435068750734515,
      "grad_norm": 11.229944229125977,
      "learning_rate": 1.2659771109450585e-05,
      "loss": 1.3768,
      "step": 19090
    },
    {
      "epoch": 2.2446821013045013,
      "grad_norm": 33.79527282714844,
      "learning_rate": 1.2640106972902822e-05,
      "loss": 1.3005,
      "step": 19100
    },
    {
      "epoch": 2.2458573275355507,
      "grad_norm": 14.954133033752441,
      "learning_rate": 1.2620442836355057e-05,
      "loss": 1.9977,
      "step": 19110
    },
    {
      "epoch": 2.2470325537666,
      "grad_norm": 18.07782745361328,
      "learning_rate": 1.2600778699807292e-05,
      "loss": 0.7661,
      "step": 19120
    },
    {
      "epoch": 2.2482077799976494,
      "grad_norm": 1.1597627401351929,
      "learning_rate": 1.2581114563259528e-05,
      "loss": 1.3653,
      "step": 19130
    },
    {
      "epoch": 2.2493830062286992,
      "grad_norm": 12.577461242675781,
      "learning_rate": 1.2561450426711763e-05,
      "loss": 1.1989,
      "step": 19140
    },
    {
      "epoch": 2.2505582324597486,
      "grad_norm": 9.333786964416504,
      "learning_rate": 1.2541786290163998e-05,
      "loss": 1.1468,
      "step": 19150
    },
    {
      "epoch": 2.251733458690798,
      "grad_norm": 26.958385467529297,
      "learning_rate": 1.2522122153616237e-05,
      "loss": 1.4214,
      "step": 19160
    },
    {
      "epoch": 2.2529086849218474,
      "grad_norm": 10.890413284301758,
      "learning_rate": 1.2502458017068472e-05,
      "loss": 1.7016,
      "step": 19170
    },
    {
      "epoch": 2.2540839111528967,
      "grad_norm": 25.467241287231445,
      "learning_rate": 1.2482793880520708e-05,
      "loss": 1.3863,
      "step": 19180
    },
    {
      "epoch": 2.2552591373839466,
      "grad_norm": 11.632078170776367,
      "learning_rate": 1.2463129743972943e-05,
      "loss": 1.4109,
      "step": 19190
    },
    {
      "epoch": 2.256434363614996,
      "grad_norm": 19.48820686340332,
      "learning_rate": 1.2443465607425178e-05,
      "loss": 1.6619,
      "step": 19200
    },
    {
      "epoch": 2.2576095898460453,
      "grad_norm": 23.91132164001465,
      "learning_rate": 1.2423801470877415e-05,
      "loss": 0.8595,
      "step": 19210
    },
    {
      "epoch": 2.2587848160770947,
      "grad_norm": 21.431161880493164,
      "learning_rate": 1.240413733432965e-05,
      "loss": 1.3875,
      "step": 19220
    },
    {
      "epoch": 2.259960042308144,
      "grad_norm": 11.41476058959961,
      "learning_rate": 1.2384473197781886e-05,
      "loss": 1.443,
      "step": 19230
    },
    {
      "epoch": 2.261135268539194,
      "grad_norm": 21.858396530151367,
      "learning_rate": 1.2364809061234121e-05,
      "loss": 1.1429,
      "step": 19240
    },
    {
      "epoch": 2.2623104947702433,
      "grad_norm": 8.751277923583984,
      "learning_rate": 1.2345144924686358e-05,
      "loss": 1.2136,
      "step": 19250
    },
    {
      "epoch": 2.2634857210012926,
      "grad_norm": 10.419465065002441,
      "learning_rate": 1.2325480788138593e-05,
      "loss": 1.5631,
      "step": 19260
    },
    {
      "epoch": 2.2646609472323425,
      "grad_norm": 17.330764770507812,
      "learning_rate": 1.2305816651590829e-05,
      "loss": 1.3134,
      "step": 19270
    },
    {
      "epoch": 2.265836173463392,
      "grad_norm": 15.965252876281738,
      "learning_rate": 1.2286152515043064e-05,
      "loss": 1.6695,
      "step": 19280
    },
    {
      "epoch": 2.267011399694441,
      "grad_norm": 7.362546920776367,
      "learning_rate": 1.2266488378495301e-05,
      "loss": 1.519,
      "step": 19290
    },
    {
      "epoch": 2.2681866259254906,
      "grad_norm": 20.009689331054688,
      "learning_rate": 1.2246824241947536e-05,
      "loss": 1.6663,
      "step": 19300
    },
    {
      "epoch": 2.26936185215654,
      "grad_norm": 13.803670883178711,
      "learning_rate": 1.2227160105399772e-05,
      "loss": 1.6986,
      "step": 19310
    },
    {
      "epoch": 2.27053707838759,
      "grad_norm": 14.399126052856445,
      "learning_rate": 1.2207495968852007e-05,
      "loss": 1.6004,
      "step": 19320
    },
    {
      "epoch": 2.271712304618639,
      "grad_norm": 20.44837760925293,
      "learning_rate": 1.2187831832304244e-05,
      "loss": 1.4509,
      "step": 19330
    },
    {
      "epoch": 2.2728875308496885,
      "grad_norm": 10.440638542175293,
      "learning_rate": 1.216816769575648e-05,
      "loss": 1.179,
      "step": 19340
    },
    {
      "epoch": 2.274062757080738,
      "grad_norm": 25.381580352783203,
      "learning_rate": 1.2148503559208716e-05,
      "loss": 1.5237,
      "step": 19350
    },
    {
      "epoch": 2.2752379833117873,
      "grad_norm": 19.780282974243164,
      "learning_rate": 1.2128839422660952e-05,
      "loss": 1.5063,
      "step": 19360
    },
    {
      "epoch": 2.276413209542837,
      "grad_norm": 35.518192291259766,
      "learning_rate": 1.2109175286113187e-05,
      "loss": 1.3854,
      "step": 19370
    },
    {
      "epoch": 2.2775884357738865,
      "grad_norm": 22.01312828063965,
      "learning_rate": 1.2089511149565424e-05,
      "loss": 1.6275,
      "step": 19380
    },
    {
      "epoch": 2.278763662004936,
      "grad_norm": 9.368061065673828,
      "learning_rate": 1.206984701301766e-05,
      "loss": 1.5387,
      "step": 19390
    },
    {
      "epoch": 2.2799388882359857,
      "grad_norm": 16.953004837036133,
      "learning_rate": 1.2050182876469896e-05,
      "loss": 1.3896,
      "step": 19400
    },
    {
      "epoch": 2.281114114467035,
      "grad_norm": 14.76840591430664,
      "learning_rate": 1.2030518739922131e-05,
      "loss": 1.3326,
      "step": 19410
    },
    {
      "epoch": 2.2822893406980844,
      "grad_norm": 12.418757438659668,
      "learning_rate": 1.2010854603374367e-05,
      "loss": 1.4538,
      "step": 19420
    },
    {
      "epoch": 2.283464566929134,
      "grad_norm": 3.1405277252197266,
      "learning_rate": 1.1991190466826602e-05,
      "loss": 1.3071,
      "step": 19430
    },
    {
      "epoch": 2.284639793160183,
      "grad_norm": 19.515790939331055,
      "learning_rate": 1.1971526330278839e-05,
      "loss": 1.5976,
      "step": 19440
    },
    {
      "epoch": 2.285815019391233,
      "grad_norm": 14.391742706298828,
      "learning_rate": 1.1951862193731074e-05,
      "loss": 1.3443,
      "step": 19450
    },
    {
      "epoch": 2.2869902456222824,
      "grad_norm": 14.822488784790039,
      "learning_rate": 1.193219805718331e-05,
      "loss": 1.3165,
      "step": 19460
    },
    {
      "epoch": 2.2881654718533317,
      "grad_norm": 6.283466815948486,
      "learning_rate": 1.1912533920635545e-05,
      "loss": 1.3982,
      "step": 19470
    },
    {
      "epoch": 2.289340698084381,
      "grad_norm": 33.64301300048828,
      "learning_rate": 1.1892869784087782e-05,
      "loss": 1.2032,
      "step": 19480
    },
    {
      "epoch": 2.2905159243154305,
      "grad_norm": 10.57918930053711,
      "learning_rate": 1.1873205647540017e-05,
      "loss": 1.8216,
      "step": 19490
    },
    {
      "epoch": 2.2916911505464803,
      "grad_norm": 14.405914306640625,
      "learning_rate": 1.1853541510992253e-05,
      "loss": 1.2362,
      "step": 19500
    },
    {
      "epoch": 2.2928663767775297,
      "grad_norm": 22.71546173095703,
      "learning_rate": 1.1833877374444488e-05,
      "loss": 1.7729,
      "step": 19510
    },
    {
      "epoch": 2.294041603008579,
      "grad_norm": 21.75313377380371,
      "learning_rate": 1.1814213237896725e-05,
      "loss": 1.561,
      "step": 19520
    },
    {
      "epoch": 2.2952168292396284,
      "grad_norm": 15.636628150939941,
      "learning_rate": 1.179454910134896e-05,
      "loss": 1.3329,
      "step": 19530
    },
    {
      "epoch": 2.2963920554706783,
      "grad_norm": 4.894004821777344,
      "learning_rate": 1.1774884964801196e-05,
      "loss": 1.0135,
      "step": 19540
    },
    {
      "epoch": 2.2975672817017276,
      "grad_norm": 4.22588586807251,
      "learning_rate": 1.1755220828253433e-05,
      "loss": 0.9931,
      "step": 19550
    },
    {
      "epoch": 2.298742507932777,
      "grad_norm": 13.772153854370117,
      "learning_rate": 1.1735556691705668e-05,
      "loss": 1.2053,
      "step": 19560
    },
    {
      "epoch": 2.2999177341638264,
      "grad_norm": 79.41436004638672,
      "learning_rate": 1.1715892555157903e-05,
      "loss": 1.6036,
      "step": 19570
    },
    {
      "epoch": 2.301092960394876,
      "grad_norm": 15.468794822692871,
      "learning_rate": 1.1696228418610138e-05,
      "loss": 1.4451,
      "step": 19580
    },
    {
      "epoch": 2.3022681866259256,
      "grad_norm": 17.036888122558594,
      "learning_rate": 1.1676564282062375e-05,
      "loss": 1.3354,
      "step": 19590
    },
    {
      "epoch": 2.303443412856975,
      "grad_norm": 18.016199111938477,
      "learning_rate": 1.165690014551461e-05,
      "loss": 1.7542,
      "step": 19600
    },
    {
      "epoch": 2.3046186390880243,
      "grad_norm": 9.665064811706543,
      "learning_rate": 1.1637236008966846e-05,
      "loss": 1.222,
      "step": 19610
    },
    {
      "epoch": 2.3057938653190737,
      "grad_norm": 8.820978164672852,
      "learning_rate": 1.1617571872419081e-05,
      "loss": 0.8143,
      "step": 19620
    },
    {
      "epoch": 2.3069690915501235,
      "grad_norm": 25.11627197265625,
      "learning_rate": 1.1597907735871318e-05,
      "loss": 1.4724,
      "step": 19630
    },
    {
      "epoch": 2.308144317781173,
      "grad_norm": 11.407133102416992,
      "learning_rate": 1.1578243599323554e-05,
      "loss": 1.6807,
      "step": 19640
    },
    {
      "epoch": 2.3093195440122223,
      "grad_norm": 31.461389541625977,
      "learning_rate": 1.1558579462775789e-05,
      "loss": 1.1143,
      "step": 19650
    },
    {
      "epoch": 2.3104947702432717,
      "grad_norm": 12.16323184967041,
      "learning_rate": 1.1538915326228026e-05,
      "loss": 1.407,
      "step": 19660
    },
    {
      "epoch": 2.3116699964743215,
      "grad_norm": 15.891446113586426,
      "learning_rate": 1.1519251189680261e-05,
      "loss": 1.5038,
      "step": 19670
    },
    {
      "epoch": 2.312845222705371,
      "grad_norm": 17.37847137451172,
      "learning_rate": 1.1499587053132498e-05,
      "loss": 1.4995,
      "step": 19680
    },
    {
      "epoch": 2.3140204489364202,
      "grad_norm": 2.8647091388702393,
      "learning_rate": 1.1479922916584734e-05,
      "loss": 1.4086,
      "step": 19690
    },
    {
      "epoch": 2.3151956751674696,
      "grad_norm": 18.396337509155273,
      "learning_rate": 1.1460258780036969e-05,
      "loss": 1.4492,
      "step": 19700
    },
    {
      "epoch": 2.3163709013985194,
      "grad_norm": 23.068330764770508,
      "learning_rate": 1.1440594643489206e-05,
      "loss": 1.0107,
      "step": 19710
    },
    {
      "epoch": 2.317546127629569,
      "grad_norm": 16.487548828125,
      "learning_rate": 1.1420930506941441e-05,
      "loss": 1.3649,
      "step": 19720
    },
    {
      "epoch": 2.318721353860618,
      "grad_norm": 23.600990295410156,
      "learning_rate": 1.1401266370393677e-05,
      "loss": 1.1238,
      "step": 19730
    },
    {
      "epoch": 2.3198965800916675,
      "grad_norm": 10.191581726074219,
      "learning_rate": 1.1381602233845914e-05,
      "loss": 1.3895,
      "step": 19740
    },
    {
      "epoch": 2.321071806322717,
      "grad_norm": 33.926490783691406,
      "learning_rate": 1.1361938097298149e-05,
      "loss": 1.6362,
      "step": 19750
    },
    {
      "epoch": 2.3222470325537667,
      "grad_norm": 18.6094913482666,
      "learning_rate": 1.1342273960750384e-05,
      "loss": 2.0323,
      "step": 19760
    },
    {
      "epoch": 2.323422258784816,
      "grad_norm": 16.07272720336914,
      "learning_rate": 1.132260982420262e-05,
      "loss": 1.8531,
      "step": 19770
    },
    {
      "epoch": 2.3245974850158655,
      "grad_norm": 6.232410430908203,
      "learning_rate": 1.1302945687654856e-05,
      "loss": 0.9309,
      "step": 19780
    },
    {
      "epoch": 2.325772711246915,
      "grad_norm": 10.035528182983398,
      "learning_rate": 1.1283281551107092e-05,
      "loss": 1.1506,
      "step": 19790
    },
    {
      "epoch": 2.3269479374779647,
      "grad_norm": 17.86903953552246,
      "learning_rate": 1.1263617414559327e-05,
      "loss": 1.1372,
      "step": 19800
    },
    {
      "epoch": 2.328123163709014,
      "grad_norm": 19.196191787719727,
      "learning_rate": 1.1243953278011562e-05,
      "loss": 1.0491,
      "step": 19810
    },
    {
      "epoch": 2.3292983899400634,
      "grad_norm": 9.507172584533691,
      "learning_rate": 1.12242891414638e-05,
      "loss": 1.2442,
      "step": 19820
    },
    {
      "epoch": 2.330473616171113,
      "grad_norm": 10.45396900177002,
      "learning_rate": 1.1204625004916035e-05,
      "loss": 1.3912,
      "step": 19830
    },
    {
      "epoch": 2.3316488424021626,
      "grad_norm": 12.420618057250977,
      "learning_rate": 1.118496086836827e-05,
      "loss": 1.2475,
      "step": 19840
    },
    {
      "epoch": 2.332824068633212,
      "grad_norm": 20.106943130493164,
      "learning_rate": 1.1165296731820505e-05,
      "loss": 1.7733,
      "step": 19850
    },
    {
      "epoch": 2.3339992948642614,
      "grad_norm": 6.582354545593262,
      "learning_rate": 1.1145632595272742e-05,
      "loss": 1.2004,
      "step": 19860
    },
    {
      "epoch": 2.3351745210953108,
      "grad_norm": 6.962731838226318,
      "learning_rate": 1.1125968458724978e-05,
      "loss": 1.0193,
      "step": 19870
    },
    {
      "epoch": 2.33634974732636,
      "grad_norm": 32.27769470214844,
      "learning_rate": 1.1106304322177213e-05,
      "loss": 1.4598,
      "step": 19880
    },
    {
      "epoch": 2.33752497355741,
      "grad_norm": 5.714953422546387,
      "learning_rate": 1.108664018562945e-05,
      "loss": 1.4742,
      "step": 19890
    },
    {
      "epoch": 2.3387001997884593,
      "grad_norm": 20.189565658569336,
      "learning_rate": 1.1066976049081685e-05,
      "loss": 1.5039,
      "step": 19900
    },
    {
      "epoch": 2.3398754260195087,
      "grad_norm": 15.567703247070312,
      "learning_rate": 1.104731191253392e-05,
      "loss": 1.6737,
      "step": 19910
    },
    {
      "epoch": 2.341050652250558,
      "grad_norm": 36.07309341430664,
      "learning_rate": 1.1027647775986156e-05,
      "loss": 1.0979,
      "step": 19920
    },
    {
      "epoch": 2.342225878481608,
      "grad_norm": 28.47287940979004,
      "learning_rate": 1.1007983639438393e-05,
      "loss": 1.2514,
      "step": 19930
    },
    {
      "epoch": 2.3434011047126573,
      "grad_norm": 14.854732513427734,
      "learning_rate": 1.0988319502890628e-05,
      "loss": 0.9638,
      "step": 19940
    },
    {
      "epoch": 2.3445763309437067,
      "grad_norm": 21.300100326538086,
      "learning_rate": 1.0968655366342863e-05,
      "loss": 1.3691,
      "step": 19950
    },
    {
      "epoch": 2.345751557174756,
      "grad_norm": 10.83236312866211,
      "learning_rate": 1.09489912297951e-05,
      "loss": 1.4006,
      "step": 19960
    },
    {
      "epoch": 2.346926783405806,
      "grad_norm": 18.004365921020508,
      "learning_rate": 1.0929327093247336e-05,
      "loss": 0.7982,
      "step": 19970
    },
    {
      "epoch": 2.3481020096368552,
      "grad_norm": 30.2900390625,
      "learning_rate": 1.0909662956699573e-05,
      "loss": 1.2049,
      "step": 19980
    },
    {
      "epoch": 2.3492772358679046,
      "grad_norm": 15.321619033813477,
      "learning_rate": 1.0889998820151808e-05,
      "loss": 1.4806,
      "step": 19990
    },
    {
      "epoch": 2.350452462098954,
      "grad_norm": 14.130854606628418,
      "learning_rate": 1.0870334683604043e-05,
      "loss": 1.4317,
      "step": 20000
    },
    {
      "epoch": 2.3516276883300034,
      "grad_norm": 8.958813667297363,
      "learning_rate": 1.085067054705628e-05,
      "loss": 1.8176,
      "step": 20010
    },
    {
      "epoch": 2.352802914561053,
      "grad_norm": 55.16764831542969,
      "learning_rate": 1.0831006410508516e-05,
      "loss": 1.7393,
      "step": 20020
    },
    {
      "epoch": 2.3539781407921025,
      "grad_norm": 15.37669849395752,
      "learning_rate": 1.0811342273960751e-05,
      "loss": 1.4556,
      "step": 20030
    },
    {
      "epoch": 2.355153367023152,
      "grad_norm": 14.300369262695312,
      "learning_rate": 1.0791678137412986e-05,
      "loss": 1.6275,
      "step": 20040
    },
    {
      "epoch": 2.3563285932542013,
      "grad_norm": 11.595333099365234,
      "learning_rate": 1.0772014000865223e-05,
      "loss": 1.3045,
      "step": 20050
    },
    {
      "epoch": 2.3575038194852507,
      "grad_norm": 14.826016426086426,
      "learning_rate": 1.0752349864317459e-05,
      "loss": 1.4929,
      "step": 20060
    },
    {
      "epoch": 2.3586790457163005,
      "grad_norm": 17.838829040527344,
      "learning_rate": 1.0732685727769694e-05,
      "loss": 1.1293,
      "step": 20070
    },
    {
      "epoch": 2.35985427194735,
      "grad_norm": 15.323509216308594,
      "learning_rate": 1.0713021591221931e-05,
      "loss": 1.7281,
      "step": 20080
    },
    {
      "epoch": 2.3610294981783992,
      "grad_norm": 13.375737190246582,
      "learning_rate": 1.0693357454674166e-05,
      "loss": 1.3577,
      "step": 20090
    },
    {
      "epoch": 2.362204724409449,
      "grad_norm": 11.967801094055176,
      "learning_rate": 1.0673693318126402e-05,
      "loss": 0.9709,
      "step": 20100
    },
    {
      "epoch": 2.3633799506404984,
      "grad_norm": 35.875572204589844,
      "learning_rate": 1.0654029181578637e-05,
      "loss": 1.3388,
      "step": 20110
    },
    {
      "epoch": 2.364555176871548,
      "grad_norm": 15.008642196655273,
      "learning_rate": 1.0634365045030874e-05,
      "loss": 1.0208,
      "step": 20120
    },
    {
      "epoch": 2.365730403102597,
      "grad_norm": 11.638132095336914,
      "learning_rate": 1.061470090848311e-05,
      "loss": 1.574,
      "step": 20130
    },
    {
      "epoch": 2.3669056293336466,
      "grad_norm": 20.113487243652344,
      "learning_rate": 1.0595036771935344e-05,
      "loss": 1.5363,
      "step": 20140
    },
    {
      "epoch": 2.3680808555646964,
      "grad_norm": 5.7382330894470215,
      "learning_rate": 1.057537263538758e-05,
      "loss": 1.8044,
      "step": 20150
    },
    {
      "epoch": 2.3692560817957458,
      "grad_norm": 21.539209365844727,
      "learning_rate": 1.0555708498839817e-05,
      "loss": 1.1697,
      "step": 20160
    },
    {
      "epoch": 2.370431308026795,
      "grad_norm": 15.60763168334961,
      "learning_rate": 1.0536044362292052e-05,
      "loss": 1.3563,
      "step": 20170
    },
    {
      "epoch": 2.3716065342578445,
      "grad_norm": 10.494027137756348,
      "learning_rate": 1.0516380225744287e-05,
      "loss": 1.2435,
      "step": 20180
    },
    {
      "epoch": 2.372781760488894,
      "grad_norm": 9.902029037475586,
      "learning_rate": 1.0496716089196523e-05,
      "loss": 1.6599,
      "step": 20190
    },
    {
      "epoch": 2.3739569867199437,
      "grad_norm": 23.01406478881836,
      "learning_rate": 1.047705195264876e-05,
      "loss": 1.7358,
      "step": 20200
    },
    {
      "epoch": 2.375132212950993,
      "grad_norm": 19.79083824157715,
      "learning_rate": 1.0457387816100995e-05,
      "loss": 1.153,
      "step": 20210
    },
    {
      "epoch": 2.3763074391820425,
      "grad_norm": 26.264360427856445,
      "learning_rate": 1.043772367955323e-05,
      "loss": 1.2866,
      "step": 20220
    },
    {
      "epoch": 2.377482665413092,
      "grad_norm": 19.450368881225586,
      "learning_rate": 1.0418059543005466e-05,
      "loss": 1.3494,
      "step": 20230
    },
    {
      "epoch": 2.3786578916441417,
      "grad_norm": 7.8080949783325195,
      "learning_rate": 1.0398395406457703e-05,
      "loss": 1.396,
      "step": 20240
    },
    {
      "epoch": 2.379833117875191,
      "grad_norm": 20.779794692993164,
      "learning_rate": 1.0378731269909938e-05,
      "loss": 1.2116,
      "step": 20250
    },
    {
      "epoch": 2.3810083441062404,
      "grad_norm": 25.108009338378906,
      "learning_rate": 1.0359067133362175e-05,
      "loss": 1.2096,
      "step": 20260
    },
    {
      "epoch": 2.38218357033729,
      "grad_norm": 11.456147193908691,
      "learning_rate": 1.033940299681441e-05,
      "loss": 1.1009,
      "step": 20270
    },
    {
      "epoch": 2.3833587965683396,
      "grad_norm": 12.468731880187988,
      "learning_rate": 1.0319738860266647e-05,
      "loss": 1.2023,
      "step": 20280
    },
    {
      "epoch": 2.384534022799389,
      "grad_norm": 16.84491729736328,
      "learning_rate": 1.0300074723718883e-05,
      "loss": 1.4607,
      "step": 20290
    },
    {
      "epoch": 2.3857092490304384,
      "grad_norm": 23.354372024536133,
      "learning_rate": 1.0280410587171118e-05,
      "loss": 1.5668,
      "step": 20300
    },
    {
      "epoch": 2.3868844752614877,
      "grad_norm": 17.796171188354492,
      "learning_rate": 1.0260746450623355e-05,
      "loss": 1.3812,
      "step": 20310
    },
    {
      "epoch": 2.388059701492537,
      "grad_norm": 39.08509063720703,
      "learning_rate": 1.024108231407559e-05,
      "loss": 1.5634,
      "step": 20320
    },
    {
      "epoch": 2.389234927723587,
      "grad_norm": 14.769865989685059,
      "learning_rate": 1.0221418177527825e-05,
      "loss": 0.7426,
      "step": 20330
    },
    {
      "epoch": 2.3904101539546363,
      "grad_norm": 7.785279273986816,
      "learning_rate": 1.020175404098006e-05,
      "loss": 1.5806,
      "step": 20340
    },
    {
      "epoch": 2.3915853801856857,
      "grad_norm": 23.487592697143555,
      "learning_rate": 1.0182089904432298e-05,
      "loss": 1.3236,
      "step": 20350
    },
    {
      "epoch": 2.392760606416735,
      "grad_norm": 15.55695629119873,
      "learning_rate": 1.0162425767884533e-05,
      "loss": 1.123,
      "step": 20360
    },
    {
      "epoch": 2.393935832647785,
      "grad_norm": 21.920507431030273,
      "learning_rate": 1.0142761631336768e-05,
      "loss": 1.3204,
      "step": 20370
    },
    {
      "epoch": 2.3951110588788342,
      "grad_norm": 25.30333709716797,
      "learning_rate": 1.0123097494789004e-05,
      "loss": 1.7729,
      "step": 20380
    },
    {
      "epoch": 2.3962862851098836,
      "grad_norm": 10.35029411315918,
      "learning_rate": 1.010343335824124e-05,
      "loss": 1.4851,
      "step": 20390
    },
    {
      "epoch": 2.397461511340933,
      "grad_norm": 32.06791687011719,
      "learning_rate": 1.0083769221693476e-05,
      "loss": 2.334,
      "step": 20400
    },
    {
      "epoch": 2.398636737571983,
      "grad_norm": 25.23912811279297,
      "learning_rate": 1.0064105085145711e-05,
      "loss": 1.7536,
      "step": 20410
    },
    {
      "epoch": 2.399811963803032,
      "grad_norm": 18.596120834350586,
      "learning_rate": 1.0044440948597948e-05,
      "loss": 1.153,
      "step": 20420
    },
    {
      "epoch": 2.4009871900340816,
      "grad_norm": 10.901917457580566,
      "learning_rate": 1.0024776812050184e-05,
      "loss": 1.4701,
      "step": 20430
    },
    {
      "epoch": 2.402162416265131,
      "grad_norm": 29.15769386291504,
      "learning_rate": 1.0005112675502419e-05,
      "loss": 1.0235,
      "step": 20440
    },
    {
      "epoch": 2.4033376424961803,
      "grad_norm": 17.028850555419922,
      "learning_rate": 9.985448538954654e-06,
      "loss": 1.2459,
      "step": 20450
    },
    {
      "epoch": 2.40451286872723,
      "grad_norm": 25.590755462646484,
      "learning_rate": 9.965784402406891e-06,
      "loss": 1.6879,
      "step": 20460
    },
    {
      "epoch": 2.4056880949582795,
      "grad_norm": 11.522571563720703,
      "learning_rate": 9.946120265859127e-06,
      "loss": 1.3846,
      "step": 20470
    },
    {
      "epoch": 2.406863321189329,
      "grad_norm": 34.66626739501953,
      "learning_rate": 9.926456129311362e-06,
      "loss": 1.6056,
      "step": 20480
    },
    {
      "epoch": 2.4080385474203783,
      "grad_norm": 24.139562606811523,
      "learning_rate": 9.906791992763597e-06,
      "loss": 0.9902,
      "step": 20490
    },
    {
      "epoch": 2.409213773651428,
      "grad_norm": 25.21050453186035,
      "learning_rate": 9.887127856215834e-06,
      "loss": 0.8862,
      "step": 20500
    },
    {
      "epoch": 2.4103889998824775,
      "grad_norm": 5.5049519538879395,
      "learning_rate": 9.86746371966807e-06,
      "loss": 1.0261,
      "step": 20510
    },
    {
      "epoch": 2.411564226113527,
      "grad_norm": 39.7928352355957,
      "learning_rate": 9.847799583120305e-06,
      "loss": 1.2537,
      "step": 20520
    },
    {
      "epoch": 2.412739452344576,
      "grad_norm": 14.349055290222168,
      "learning_rate": 9.82813544657254e-06,
      "loss": 1.0708,
      "step": 20530
    },
    {
      "epoch": 2.413914678575626,
      "grad_norm": 21.86899757385254,
      "learning_rate": 9.808471310024777e-06,
      "loss": 1.376,
      "step": 20540
    },
    {
      "epoch": 2.4150899048066754,
      "grad_norm": 13.68761157989502,
      "learning_rate": 9.788807173477012e-06,
      "loss": 1.6751,
      "step": 20550
    },
    {
      "epoch": 2.416265131037725,
      "grad_norm": 13.345365524291992,
      "learning_rate": 9.76914303692925e-06,
      "loss": 1.2602,
      "step": 20560
    },
    {
      "epoch": 2.417440357268774,
      "grad_norm": 4.627640724182129,
      "learning_rate": 9.749478900381485e-06,
      "loss": 1.2777,
      "step": 20570
    },
    {
      "epoch": 2.4186155834998235,
      "grad_norm": 27.2735652923584,
      "learning_rate": 9.729814763833722e-06,
      "loss": 1.1088,
      "step": 20580
    },
    {
      "epoch": 2.4197908097308733,
      "grad_norm": 6.465244770050049,
      "learning_rate": 9.710150627285957e-06,
      "loss": 1.4223,
      "step": 20590
    },
    {
      "epoch": 2.4209660359619227,
      "grad_norm": 24.079994201660156,
      "learning_rate": 9.690486490738192e-06,
      "loss": 1.4912,
      "step": 20600
    },
    {
      "epoch": 2.422141262192972,
      "grad_norm": 17.26298713684082,
      "learning_rate": 9.67082235419043e-06,
      "loss": 1.5101,
      "step": 20610
    },
    {
      "epoch": 2.4233164884240215,
      "grad_norm": 42.114646911621094,
      "learning_rate": 9.651158217642665e-06,
      "loss": 1.4431,
      "step": 20620
    },
    {
      "epoch": 2.4244917146550713,
      "grad_norm": 25.387407302856445,
      "learning_rate": 9.6314940810949e-06,
      "loss": 1.0676,
      "step": 20630
    },
    {
      "epoch": 2.4256669408861207,
      "grad_norm": 18.163299560546875,
      "learning_rate": 9.611829944547135e-06,
      "loss": 1.68,
      "step": 20640
    },
    {
      "epoch": 2.42684216711717,
      "grad_norm": 5.62760066986084,
      "learning_rate": 9.592165807999372e-06,
      "loss": 1.8209,
      "step": 20650
    },
    {
      "epoch": 2.4280173933482194,
      "grad_norm": 51.95356750488281,
      "learning_rate": 9.572501671451608e-06,
      "loss": 1.2309,
      "step": 20660
    },
    {
      "epoch": 2.4291926195792692,
      "grad_norm": 3.0368101596832275,
      "learning_rate": 9.552837534903843e-06,
      "loss": 1.7671,
      "step": 20670
    },
    {
      "epoch": 2.4303678458103186,
      "grad_norm": 16.647905349731445,
      "learning_rate": 9.533173398356078e-06,
      "loss": 1.6635,
      "step": 20680
    },
    {
      "epoch": 2.431543072041368,
      "grad_norm": 33.735599517822266,
      "learning_rate": 9.513509261808315e-06,
      "loss": 1.231,
      "step": 20690
    },
    {
      "epoch": 2.4327182982724174,
      "grad_norm": 7.061263084411621,
      "learning_rate": 9.49384512526055e-06,
      "loss": 1.1542,
      "step": 20700
    },
    {
      "epoch": 2.4338935245034667,
      "grad_norm": 28.27484130859375,
      "learning_rate": 9.474180988712786e-06,
      "loss": 1.195,
      "step": 20710
    },
    {
      "epoch": 2.4350687507345166,
      "grad_norm": 10.150604248046875,
      "learning_rate": 9.454516852165021e-06,
      "loss": 1.1993,
      "step": 20720
    },
    {
      "epoch": 2.436243976965566,
      "grad_norm": 17.42117691040039,
      "learning_rate": 9.434852715617258e-06,
      "loss": 1.3652,
      "step": 20730
    },
    {
      "epoch": 2.4374192031966153,
      "grad_norm": 17.694774627685547,
      "learning_rate": 9.415188579069493e-06,
      "loss": 1.6737,
      "step": 20740
    },
    {
      "epoch": 2.4385944294276647,
      "grad_norm": 44.343326568603516,
      "learning_rate": 9.395524442521729e-06,
      "loss": 1.126,
      "step": 20750
    },
    {
      "epoch": 2.439769655658714,
      "grad_norm": 34.99625778198242,
      "learning_rate": 9.375860305973966e-06,
      "loss": 1.3068,
      "step": 20760
    },
    {
      "epoch": 2.440944881889764,
      "grad_norm": 16.315471649169922,
      "learning_rate": 9.356196169426201e-06,
      "loss": 1.8647,
      "step": 20770
    },
    {
      "epoch": 2.4421201081208133,
      "grad_norm": 48.66804504394531,
      "learning_rate": 9.336532032878436e-06,
      "loss": 1.3526,
      "step": 20780
    },
    {
      "epoch": 2.4432953343518626,
      "grad_norm": 25.155141830444336,
      "learning_rate": 9.316867896330672e-06,
      "loss": 1.4143,
      "step": 20790
    },
    {
      "epoch": 2.4444705605829125,
      "grad_norm": 21.226987838745117,
      "learning_rate": 9.297203759782909e-06,
      "loss": 1.3198,
      "step": 20800
    },
    {
      "epoch": 2.445645786813962,
      "grad_norm": 24.906707763671875,
      "learning_rate": 9.277539623235144e-06,
      "loss": 1.2842,
      "step": 20810
    },
    {
      "epoch": 2.446821013045011,
      "grad_norm": 22.06084632873535,
      "learning_rate": 9.25787548668738e-06,
      "loss": 1.3389,
      "step": 20820
    },
    {
      "epoch": 2.4479962392760606,
      "grad_norm": 18.58282470703125,
      "learning_rate": 9.238211350139615e-06,
      "loss": 1.4436,
      "step": 20830
    },
    {
      "epoch": 2.44917146550711,
      "grad_norm": 13.776021003723145,
      "learning_rate": 9.218547213591852e-06,
      "loss": 1.5583,
      "step": 20840
    },
    {
      "epoch": 2.4503466917381598,
      "grad_norm": 24.99762725830078,
      "learning_rate": 9.198883077044087e-06,
      "loss": 1.7917,
      "step": 20850
    },
    {
      "epoch": 2.451521917969209,
      "grad_norm": 33.801177978515625,
      "learning_rate": 9.179218940496324e-06,
      "loss": 1.4201,
      "step": 20860
    },
    {
      "epoch": 2.4526971442002585,
      "grad_norm": 17.97370719909668,
      "learning_rate": 9.15955480394856e-06,
      "loss": 1.4276,
      "step": 20870
    },
    {
      "epoch": 2.453872370431308,
      "grad_norm": 37.57416534423828,
      "learning_rate": 9.139890667400794e-06,
      "loss": 1.4752,
      "step": 20880
    },
    {
      "epoch": 2.4550475966623573,
      "grad_norm": 10.652751922607422,
      "learning_rate": 9.120226530853031e-06,
      "loss": 1.4236,
      "step": 20890
    },
    {
      "epoch": 2.456222822893407,
      "grad_norm": 16.896032333374023,
      "learning_rate": 9.100562394305267e-06,
      "loss": 1.4632,
      "step": 20900
    },
    {
      "epoch": 2.4573980491244565,
      "grad_norm": 12.206268310546875,
      "learning_rate": 9.080898257757502e-06,
      "loss": 1.2269,
      "step": 20910
    },
    {
      "epoch": 2.458573275355506,
      "grad_norm": 27.2872371673584,
      "learning_rate": 9.061234121209739e-06,
      "loss": 1.1907,
      "step": 20920
    },
    {
      "epoch": 2.4597485015865552,
      "grad_norm": 11.037638664245605,
      "learning_rate": 9.041569984661974e-06,
      "loss": 1.2926,
      "step": 20930
    },
    {
      "epoch": 2.460923727817605,
      "grad_norm": 15.507025718688965,
      "learning_rate": 9.02190584811421e-06,
      "loss": 1.7584,
      "step": 20940
    },
    {
      "epoch": 2.4620989540486544,
      "grad_norm": 10.164241790771484,
      "learning_rate": 9.002241711566447e-06,
      "loss": 1.5942,
      "step": 20950
    },
    {
      "epoch": 2.463274180279704,
      "grad_norm": 10.636491775512695,
      "learning_rate": 8.982577575018682e-06,
      "loss": 1.5023,
      "step": 20960
    },
    {
      "epoch": 2.464449406510753,
      "grad_norm": 23.195629119873047,
      "learning_rate": 8.962913438470917e-06,
      "loss": 1.3532,
      "step": 20970
    },
    {
      "epoch": 2.465624632741803,
      "grad_norm": 23.35316276550293,
      "learning_rate": 8.943249301923153e-06,
      "loss": 1.2714,
      "step": 20980
    },
    {
      "epoch": 2.4667998589728524,
      "grad_norm": 12.670083045959473,
      "learning_rate": 8.92358516537539e-06,
      "loss": 1.11,
      "step": 20990
    },
    {
      "epoch": 2.4679750852039017,
      "grad_norm": 8.718060493469238,
      "learning_rate": 8.903921028827625e-06,
      "loss": 1.5115,
      "step": 21000
    },
    {
      "epoch": 2.469150311434951,
      "grad_norm": 7.867190837860107,
      "learning_rate": 8.88425689227986e-06,
      "loss": 1.3439,
      "step": 21010
    },
    {
      "epoch": 2.4703255376660005,
      "grad_norm": 15.300798416137695,
      "learning_rate": 8.864592755732096e-06,
      "loss": 1.4793,
      "step": 21020
    },
    {
      "epoch": 2.4715007638970503,
      "grad_norm": 16.79490089416504,
      "learning_rate": 8.844928619184333e-06,
      "loss": 1.4853,
      "step": 21030
    },
    {
      "epoch": 2.4726759901280997,
      "grad_norm": 19.125534057617188,
      "learning_rate": 8.825264482636568e-06,
      "loss": 1.6526,
      "step": 21040
    },
    {
      "epoch": 2.473851216359149,
      "grad_norm": 21.117168426513672,
      "learning_rate": 8.805600346088803e-06,
      "loss": 1.4868,
      "step": 21050
    },
    {
      "epoch": 2.4750264425901984,
      "grad_norm": 7.764671802520752,
      "learning_rate": 8.785936209541038e-06,
      "loss": 1.2605,
      "step": 21060
    },
    {
      "epoch": 2.4762016688212483,
      "grad_norm": 20.72799301147461,
      "learning_rate": 8.766272072993275e-06,
      "loss": 1.3219,
      "step": 21070
    },
    {
      "epoch": 2.4773768950522976,
      "grad_norm": 9.180767059326172,
      "learning_rate": 8.74660793644551e-06,
      "loss": 1.5791,
      "step": 21080
    },
    {
      "epoch": 2.478552121283347,
      "grad_norm": 22.976375579833984,
      "learning_rate": 8.726943799897746e-06,
      "loss": 1.2349,
      "step": 21090
    },
    {
      "epoch": 2.4797273475143964,
      "grad_norm": 6.9001898765563965,
      "learning_rate": 8.707279663349983e-06,
      "loss": 1.1671,
      "step": 21100
    },
    {
      "epoch": 2.480902573745446,
      "grad_norm": 13.605376243591309,
      "learning_rate": 8.687615526802218e-06,
      "loss": 1.5345,
      "step": 21110
    },
    {
      "epoch": 2.4820777999764956,
      "grad_norm": 25.081809997558594,
      "learning_rate": 8.667951390254454e-06,
      "loss": 1.3851,
      "step": 21120
    },
    {
      "epoch": 2.483253026207545,
      "grad_norm": 19.87027931213379,
      "learning_rate": 8.648287253706689e-06,
      "loss": 1.1119,
      "step": 21130
    },
    {
      "epoch": 2.4844282524385943,
      "grad_norm": 9.800013542175293,
      "learning_rate": 8.628623117158926e-06,
      "loss": 1.4739,
      "step": 21140
    },
    {
      "epoch": 2.4856034786696437,
      "grad_norm": 19.78736686706543,
      "learning_rate": 8.608958980611161e-06,
      "loss": 1.5542,
      "step": 21150
    },
    {
      "epoch": 2.4867787049006935,
      "grad_norm": 16.40639877319336,
      "learning_rate": 8.589294844063398e-06,
      "loss": 1.1229,
      "step": 21160
    },
    {
      "epoch": 2.487953931131743,
      "grad_norm": 17.730443954467773,
      "learning_rate": 8.569630707515634e-06,
      "loss": 1.2057,
      "step": 21170
    },
    {
      "epoch": 2.4891291573627923,
      "grad_norm": 10.755648612976074,
      "learning_rate": 8.549966570967869e-06,
      "loss": 1.1936,
      "step": 21180
    },
    {
      "epoch": 2.4903043835938417,
      "grad_norm": 16.931129455566406,
      "learning_rate": 8.530302434420106e-06,
      "loss": 0.8988,
      "step": 21190
    },
    {
      "epoch": 2.4914796098248915,
      "grad_norm": 16.9395751953125,
      "learning_rate": 8.510638297872341e-06,
      "loss": 1.4634,
      "step": 21200
    },
    {
      "epoch": 2.492654836055941,
      "grad_norm": 19.950428009033203,
      "learning_rate": 8.490974161324577e-06,
      "loss": 1.0216,
      "step": 21210
    },
    {
      "epoch": 2.4938300622869902,
      "grad_norm": 12.50057315826416,
      "learning_rate": 8.471310024776814e-06,
      "loss": 1.5448,
      "step": 21220
    },
    {
      "epoch": 2.4950052885180396,
      "grad_norm": 13.657111167907715,
      "learning_rate": 8.451645888229049e-06,
      "loss": 1.4186,
      "step": 21230
    },
    {
      "epoch": 2.4961805147490894,
      "grad_norm": 14.280421257019043,
      "learning_rate": 8.431981751681284e-06,
      "loss": 2.1692,
      "step": 21240
    },
    {
      "epoch": 2.497355740980139,
      "grad_norm": 13.427152633666992,
      "learning_rate": 8.41231761513352e-06,
      "loss": 1.2037,
      "step": 21250
    },
    {
      "epoch": 2.498530967211188,
      "grad_norm": 8.415739059448242,
      "learning_rate": 8.392653478585756e-06,
      "loss": 1.4312,
      "step": 21260
    },
    {
      "epoch": 2.4997061934422375,
      "grad_norm": 12.526494979858398,
      "learning_rate": 8.372989342037992e-06,
      "loss": 1.4266,
      "step": 21270
    },
    {
      "epoch": 2.500881419673287,
      "grad_norm": 5.9845428466796875,
      "learning_rate": 8.353325205490227e-06,
      "loss": 0.9588,
      "step": 21280
    },
    {
      "epoch": 2.5020566459043367,
      "grad_norm": 11.534880638122559,
      "learning_rate": 8.333661068942464e-06,
      "loss": 1.7498,
      "step": 21290
    },
    {
      "epoch": 2.503231872135386,
      "grad_norm": 17.40300750732422,
      "learning_rate": 8.3139969323947e-06,
      "loss": 1.1364,
      "step": 21300
    },
    {
      "epoch": 2.5044070983664355,
      "grad_norm": 14.290718078613281,
      "learning_rate": 8.294332795846935e-06,
      "loss": 1.3438,
      "step": 21310
    },
    {
      "epoch": 2.505582324597485,
      "grad_norm": 13.005010604858398,
      "learning_rate": 8.27466865929917e-06,
      "loss": 0.9516,
      "step": 21320
    },
    {
      "epoch": 2.5067575508285342,
      "grad_norm": 16.558002471923828,
      "learning_rate": 8.255004522751407e-06,
      "loss": 1.5388,
      "step": 21330
    },
    {
      "epoch": 2.507932777059584,
      "grad_norm": 31.19087791442871,
      "learning_rate": 8.235340386203642e-06,
      "loss": 1.3969,
      "step": 21340
    },
    {
      "epoch": 2.5091080032906334,
      "grad_norm": 36.715389251708984,
      "learning_rate": 8.215676249655878e-06,
      "loss": 1.4076,
      "step": 21350
    },
    {
      "epoch": 2.510283229521683,
      "grad_norm": 14.870002746582031,
      "learning_rate": 8.196012113108113e-06,
      "loss": 1.2177,
      "step": 21360
    },
    {
      "epoch": 2.5114584557527326,
      "grad_norm": 13.938783645629883,
      "learning_rate": 8.17634797656035e-06,
      "loss": 1.664,
      "step": 21370
    },
    {
      "epoch": 2.512633681983782,
      "grad_norm": 23.696346282958984,
      "learning_rate": 8.156683840012585e-06,
      "loss": 1.2752,
      "step": 21380
    },
    {
      "epoch": 2.5138089082148314,
      "grad_norm": 25.399351119995117,
      "learning_rate": 8.13701970346482e-06,
      "loss": 1.8688,
      "step": 21390
    },
    {
      "epoch": 2.5149841344458808,
      "grad_norm": 46.8927116394043,
      "learning_rate": 8.117355566917056e-06,
      "loss": 1.4778,
      "step": 21400
    },
    {
      "epoch": 2.51615936067693,
      "grad_norm": 17.068187713623047,
      "learning_rate": 8.097691430369293e-06,
      "loss": 1.4482,
      "step": 21410
    },
    {
      "epoch": 2.51733458690798,
      "grad_norm": 21.05521583557129,
      "learning_rate": 8.078027293821528e-06,
      "loss": 0.9652,
      "step": 21420
    },
    {
      "epoch": 2.5185098131390293,
      "grad_norm": 28.892616271972656,
      "learning_rate": 8.058363157273763e-06,
      "loss": 1.826,
      "step": 21430
    },
    {
      "epoch": 2.5196850393700787,
      "grad_norm": 12.751649856567383,
      "learning_rate": 8.038699020726e-06,
      "loss": 1.1848,
      "step": 21440
    },
    {
      "epoch": 2.520860265601128,
      "grad_norm": 12.873831748962402,
      "learning_rate": 8.019034884178236e-06,
      "loss": 1.3924,
      "step": 21450
    },
    {
      "epoch": 2.5220354918321775,
      "grad_norm": 27.829378128051758,
      "learning_rate": 7.999370747630471e-06,
      "loss": 1.3067,
      "step": 21460
    },
    {
      "epoch": 2.5232107180632273,
      "grad_norm": 16.4314022064209,
      "learning_rate": 7.979706611082708e-06,
      "loss": 1.2705,
      "step": 21470
    },
    {
      "epoch": 2.5243859442942767,
      "grad_norm": 21.670255661010742,
      "learning_rate": 7.960042474534943e-06,
      "loss": 1.6441,
      "step": 21480
    },
    {
      "epoch": 2.525561170525326,
      "grad_norm": 11.575331687927246,
      "learning_rate": 7.94037833798718e-06,
      "loss": 1.5843,
      "step": 21490
    },
    {
      "epoch": 2.526736396756376,
      "grad_norm": 39.91982650756836,
      "learning_rate": 7.920714201439416e-06,
      "loss": 1.6569,
      "step": 21500
    },
    {
      "epoch": 2.527911622987425,
      "grad_norm": 6.002248764038086,
      "learning_rate": 7.901050064891651e-06,
      "loss": 1.5463,
      "step": 21510
    },
    {
      "epoch": 2.5290868492184746,
      "grad_norm": 17.31133460998535,
      "learning_rate": 7.881385928343888e-06,
      "loss": 0.9006,
      "step": 21520
    },
    {
      "epoch": 2.530262075449524,
      "grad_norm": 34.208675384521484,
      "learning_rate": 7.861721791796123e-06,
      "loss": 1.8371,
      "step": 21530
    },
    {
      "epoch": 2.5314373016805733,
      "grad_norm": 54.78329086303711,
      "learning_rate": 7.842057655248359e-06,
      "loss": 1.075,
      "step": 21540
    },
    {
      "epoch": 2.532612527911623,
      "grad_norm": 18.33517837524414,
      "learning_rate": 7.822393518700594e-06,
      "loss": 1.2885,
      "step": 21550
    },
    {
      "epoch": 2.5337877541426725,
      "grad_norm": 17.683942794799805,
      "learning_rate": 7.802729382152831e-06,
      "loss": 1.2914,
      "step": 21560
    },
    {
      "epoch": 2.534962980373722,
      "grad_norm": 24.086383819580078,
      "learning_rate": 7.783065245605066e-06,
      "loss": 1.2258,
      "step": 21570
    },
    {
      "epoch": 2.5361382066047713,
      "grad_norm": 11.160987854003906,
      "learning_rate": 7.763401109057302e-06,
      "loss": 1.4309,
      "step": 21580
    },
    {
      "epoch": 2.5373134328358207,
      "grad_norm": 11.989442825317383,
      "learning_rate": 7.743736972509537e-06,
      "loss": 1.1724,
      "step": 21590
    },
    {
      "epoch": 2.5384886590668705,
      "grad_norm": 8.299538612365723,
      "learning_rate": 7.724072835961774e-06,
      "loss": 1.0354,
      "step": 21600
    },
    {
      "epoch": 2.53966388529792,
      "grad_norm": 9.012978553771973,
      "learning_rate": 7.704408699414009e-06,
      "loss": 1.0389,
      "step": 21610
    },
    {
      "epoch": 2.5408391115289692,
      "grad_norm": 18.4547176361084,
      "learning_rate": 7.684744562866244e-06,
      "loss": 2.0083,
      "step": 21620
    },
    {
      "epoch": 2.542014337760019,
      "grad_norm": 51.364192962646484,
      "learning_rate": 7.665080426318481e-06,
      "loss": 1.5726,
      "step": 21630
    },
    {
      "epoch": 2.5431895639910684,
      "grad_norm": 20.42410659790039,
      "learning_rate": 7.645416289770717e-06,
      "loss": 1.4143,
      "step": 21640
    },
    {
      "epoch": 2.544364790222118,
      "grad_norm": 23.10498046875,
      "learning_rate": 7.625752153222952e-06,
      "loss": 1.6879,
      "step": 21650
    },
    {
      "epoch": 2.545540016453167,
      "grad_norm": 9.038946151733398,
      "learning_rate": 7.606088016675187e-06,
      "loss": 1.4907,
      "step": 21660
    },
    {
      "epoch": 2.5467152426842166,
      "grad_norm": 23.840518951416016,
      "learning_rate": 7.586423880127424e-06,
      "loss": 1.5096,
      "step": 21670
    },
    {
      "epoch": 2.5478904689152664,
      "grad_norm": 9.873739242553711,
      "learning_rate": 7.56675974357966e-06,
      "loss": 1.7324,
      "step": 21680
    },
    {
      "epoch": 2.5490656951463158,
      "grad_norm": 12.985697746276855,
      "learning_rate": 7.547095607031896e-06,
      "loss": 1.444,
      "step": 21690
    },
    {
      "epoch": 2.550240921377365,
      "grad_norm": 16.502452850341797,
      "learning_rate": 7.527431470484131e-06,
      "loss": 1.537,
      "step": 21700
    },
    {
      "epoch": 2.5514161476084145,
      "grad_norm": 45.040889739990234,
      "learning_rate": 7.507767333936367e-06,
      "loss": 1.8252,
      "step": 21710
    },
    {
      "epoch": 2.552591373839464,
      "grad_norm": 23.055320739746094,
      "learning_rate": 7.4881031973886035e-06,
      "loss": 0.9711,
      "step": 21720
    },
    {
      "epoch": 2.5537666000705137,
      "grad_norm": 23.025142669677734,
      "learning_rate": 7.468439060840839e-06,
      "loss": 1.4391,
      "step": 21730
    },
    {
      "epoch": 2.554941826301563,
      "grad_norm": 6.857464790344238,
      "learning_rate": 7.448774924293074e-06,
      "loss": 1.0715,
      "step": 21740
    },
    {
      "epoch": 2.5561170525326125,
      "grad_norm": 25.59360694885254,
      "learning_rate": 7.429110787745311e-06,
      "loss": 1.2531,
      "step": 21750
    },
    {
      "epoch": 2.5572922787636623,
      "grad_norm": 17.986114501953125,
      "learning_rate": 7.409446651197546e-06,
      "loss": 1.2391,
      "step": 21760
    },
    {
      "epoch": 2.558467504994711,
      "grad_norm": 13.39638900756836,
      "learning_rate": 7.389782514649782e-06,
      "loss": 1.5784,
      "step": 21770
    },
    {
      "epoch": 2.559642731225761,
      "grad_norm": 16.624588012695312,
      "learning_rate": 7.370118378102017e-06,
      "loss": 1.305,
      "step": 21780
    },
    {
      "epoch": 2.5608179574568104,
      "grad_norm": 16.15389060974121,
      "learning_rate": 7.350454241554254e-06,
      "loss": 0.9185,
      "step": 21790
    },
    {
      "epoch": 2.5619931836878598,
      "grad_norm": 11.077719688415527,
      "learning_rate": 7.330790105006489e-06,
      "loss": 1.2586,
      "step": 21800
    },
    {
      "epoch": 2.5631684099189096,
      "grad_norm": 12.195098876953125,
      "learning_rate": 7.311125968458725e-06,
      "loss": 1.2258,
      "step": 21810
    },
    {
      "epoch": 2.564343636149959,
      "grad_norm": 17.53858184814453,
      "learning_rate": 7.291461831910962e-06,
      "loss": 1.736,
      "step": 21820
    },
    {
      "epoch": 2.5655188623810083,
      "grad_norm": 23.413183212280273,
      "learning_rate": 7.271797695363197e-06,
      "loss": 1.5215,
      "step": 21830
    },
    {
      "epoch": 2.5666940886120577,
      "grad_norm": 15.700861930847168,
      "learning_rate": 7.252133558815433e-06,
      "loss": 2.0439,
      "step": 21840
    },
    {
      "epoch": 2.567869314843107,
      "grad_norm": 6.983325004577637,
      "learning_rate": 7.232469422267668e-06,
      "loss": 1.3492,
      "step": 21850
    },
    {
      "epoch": 2.569044541074157,
      "grad_norm": 14.11115837097168,
      "learning_rate": 7.2128052857199045e-06,
      "loss": 1.0097,
      "step": 21860
    },
    {
      "epoch": 2.5702197673052063,
      "grad_norm": 22.942768096923828,
      "learning_rate": 7.193141149172141e-06,
      "loss": 0.8818,
      "step": 21870
    },
    {
      "epoch": 2.5713949935362557,
      "grad_norm": 26.3778018951416,
      "learning_rate": 7.173477012624376e-06,
      "loss": 1.2605,
      "step": 21880
    },
    {
      "epoch": 2.572570219767305,
      "grad_norm": 16.66275978088379,
      "learning_rate": 7.153812876076611e-06,
      "loss": 1.8209,
      "step": 21890
    },
    {
      "epoch": 2.5737454459983544,
      "grad_norm": 33.63488006591797,
      "learning_rate": 7.134148739528848e-06,
      "loss": 1.487,
      "step": 21900
    },
    {
      "epoch": 2.5749206722294042,
      "grad_norm": 19.194496154785156,
      "learning_rate": 7.114484602981084e-06,
      "loss": 1.4581,
      "step": 21910
    },
    {
      "epoch": 2.5760958984604536,
      "grad_norm": 27.809389114379883,
      "learning_rate": 7.094820466433319e-06,
      "loss": 1.3843,
      "step": 21920
    },
    {
      "epoch": 2.577271124691503,
      "grad_norm": 19.007766723632812,
      "learning_rate": 7.075156329885554e-06,
      "loss": 1.0519,
      "step": 21930
    },
    {
      "epoch": 2.578446350922553,
      "grad_norm": 14.892168045043945,
      "learning_rate": 7.055492193337791e-06,
      "loss": 1.4243,
      "step": 21940
    },
    {
      "epoch": 2.579621577153602,
      "grad_norm": 19.57586097717285,
      "learning_rate": 7.0358280567900265e-06,
      "loss": 2.1762,
      "step": 21950
    },
    {
      "epoch": 2.5807968033846516,
      "grad_norm": 41.72001266479492,
      "learning_rate": 7.016163920242262e-06,
      "loss": 1.3801,
      "step": 21960
    },
    {
      "epoch": 2.581972029615701,
      "grad_norm": 27.66671371459961,
      "learning_rate": 6.996499783694499e-06,
      "loss": 1.3792,
      "step": 21970
    },
    {
      "epoch": 2.5831472558467503,
      "grad_norm": 15.526238441467285,
      "learning_rate": 6.976835647146734e-06,
      "loss": 1.4413,
      "step": 21980
    },
    {
      "epoch": 2.5843224820778,
      "grad_norm": 25.757434844970703,
      "learning_rate": 6.9571715105989695e-06,
      "loss": 1.2176,
      "step": 21990
    },
    {
      "epoch": 2.5854977083088495,
      "grad_norm": 10.105486869812012,
      "learning_rate": 6.937507374051206e-06,
      "loss": 1.5914,
      "step": 22000
    },
    {
      "epoch": 2.586672934539899,
      "grad_norm": 17.347667694091797,
      "learning_rate": 6.917843237503442e-06,
      "loss": 0.8333,
      "step": 22010
    },
    {
      "epoch": 2.5878481607709483,
      "grad_norm": 36.442161560058594,
      "learning_rate": 6.898179100955678e-06,
      "loss": 1.6015,
      "step": 22020
    },
    {
      "epoch": 2.5890233870019976,
      "grad_norm": 22.932044982910156,
      "learning_rate": 6.878514964407913e-06,
      "loss": 1.353,
      "step": 22030
    },
    {
      "epoch": 2.5901986132330475,
      "grad_norm": 22.181549072265625,
      "learning_rate": 6.8588508278601485e-06,
      "loss": 2.1035,
      "step": 22040
    },
    {
      "epoch": 2.591373839464097,
      "grad_norm": 26.964025497436523,
      "learning_rate": 6.8391866913123855e-06,
      "loss": 1.1231,
      "step": 22050
    },
    {
      "epoch": 2.592549065695146,
      "grad_norm": 33.24808883666992,
      "learning_rate": 6.819522554764621e-06,
      "loss": 1.132,
      "step": 22060
    },
    {
      "epoch": 2.593724291926196,
      "grad_norm": 10.183686256408691,
      "learning_rate": 6.799858418216856e-06,
      "loss": 1.8753,
      "step": 22070
    },
    {
      "epoch": 2.5948995181572454,
      "grad_norm": 13.495781898498535,
      "learning_rate": 6.7801942816690915e-06,
      "loss": 1.1293,
      "step": 22080
    },
    {
      "epoch": 2.5960747443882948,
      "grad_norm": 19.272367477416992,
      "learning_rate": 6.7605301451213285e-06,
      "loss": 0.9462,
      "step": 22090
    },
    {
      "epoch": 2.597249970619344,
      "grad_norm": 21.824886322021484,
      "learning_rate": 6.740866008573564e-06,
      "loss": 1.3469,
      "step": 22100
    },
    {
      "epoch": 2.5984251968503935,
      "grad_norm": 32.56025695800781,
      "learning_rate": 6.721201872025799e-06,
      "loss": 0.8011,
      "step": 22110
    },
    {
      "epoch": 2.5996004230814433,
      "grad_norm": 11.244007110595703,
      "learning_rate": 6.701537735478035e-06,
      "loss": 1.0813,
      "step": 22120
    },
    {
      "epoch": 2.6007756493124927,
      "grad_norm": 20.41330337524414,
      "learning_rate": 6.681873598930271e-06,
      "loss": 1.4681,
      "step": 22130
    },
    {
      "epoch": 2.601950875543542,
      "grad_norm": 15.54897689819336,
      "learning_rate": 6.662209462382507e-06,
      "loss": 1.2821,
      "step": 22140
    },
    {
      "epoch": 2.6031261017745915,
      "grad_norm": 23.565309524536133,
      "learning_rate": 6.642545325834743e-06,
      "loss": 1.4995,
      "step": 22150
    },
    {
      "epoch": 2.604301328005641,
      "grad_norm": 4.5979814529418945,
      "learning_rate": 6.622881189286979e-06,
      "loss": 1.0414,
      "step": 22160
    },
    {
      "epoch": 2.6054765542366907,
      "grad_norm": 10.381558418273926,
      "learning_rate": 6.603217052739215e-06,
      "loss": 0.8592,
      "step": 22170
    },
    {
      "epoch": 2.60665178046774,
      "grad_norm": 11.596261024475098,
      "learning_rate": 6.5835529161914505e-06,
      "loss": 1.7588,
      "step": 22180
    },
    {
      "epoch": 2.6078270066987894,
      "grad_norm": 17.706356048583984,
      "learning_rate": 6.563888779643686e-06,
      "loss": 1.6118,
      "step": 22190
    },
    {
      "epoch": 2.6090022329298392,
      "grad_norm": 8.03243350982666,
      "learning_rate": 6.544224643095923e-06,
      "loss": 1.6062,
      "step": 22200
    },
    {
      "epoch": 2.6101774591608886,
      "grad_norm": 26.46589469909668,
      "learning_rate": 6.524560506548158e-06,
      "loss": 1.8453,
      "step": 22210
    },
    {
      "epoch": 2.611352685391938,
      "grad_norm": 17.87671661376953,
      "learning_rate": 6.504896370000393e-06,
      "loss": 1.3316,
      "step": 22220
    },
    {
      "epoch": 2.6125279116229874,
      "grad_norm": 14.33586311340332,
      "learning_rate": 6.485232233452629e-06,
      "loss": 1.229,
      "step": 22230
    },
    {
      "epoch": 2.6137031378540367,
      "grad_norm": 18.783714294433594,
      "learning_rate": 6.465568096904866e-06,
      "loss": 2.054,
      "step": 22240
    },
    {
      "epoch": 2.6148783640850866,
      "grad_norm": 15.180062294006348,
      "learning_rate": 6.445903960357101e-06,
      "loss": 1.6917,
      "step": 22250
    },
    {
      "epoch": 2.616053590316136,
      "grad_norm": 34.66228485107422,
      "learning_rate": 6.426239823809336e-06,
      "loss": 1.4319,
      "step": 22260
    },
    {
      "epoch": 2.6172288165471853,
      "grad_norm": 25.01453399658203,
      "learning_rate": 6.4065756872615725e-06,
      "loss": 1.5886,
      "step": 22270
    },
    {
      "epoch": 2.6184040427782347,
      "grad_norm": 11.981661796569824,
      "learning_rate": 6.386911550713809e-06,
      "loss": 1.2848,
      "step": 22280
    },
    {
      "epoch": 2.619579269009284,
      "grad_norm": 19.809602737426758,
      "learning_rate": 6.367247414166044e-06,
      "loss": 1.5062,
      "step": 22290
    },
    {
      "epoch": 2.620754495240334,
      "grad_norm": 19.55670928955078,
      "learning_rate": 6.34758327761828e-06,
      "loss": 1.0508,
      "step": 22300
    },
    {
      "epoch": 2.6219297214713833,
      "grad_norm": 12.409029960632324,
      "learning_rate": 6.327919141070516e-06,
      "loss": 1.2968,
      "step": 22310
    },
    {
      "epoch": 2.6231049477024326,
      "grad_norm": 25.700225830078125,
      "learning_rate": 6.308255004522752e-06,
      "loss": 1.6162,
      "step": 22320
    },
    {
      "epoch": 2.6242801739334825,
      "grad_norm": 31.188671112060547,
      "learning_rate": 6.288590867974988e-06,
      "loss": 1.3425,
      "step": 22330
    },
    {
      "epoch": 2.625455400164532,
      "grad_norm": 12.754523277282715,
      "learning_rate": 6.268926731427223e-06,
      "loss": 1.7452,
      "step": 22340
    },
    {
      "epoch": 2.626630626395581,
      "grad_norm": 21.975717544555664,
      "learning_rate": 6.249262594879459e-06,
      "loss": 1.435,
      "step": 22350
    },
    {
      "epoch": 2.6278058526266306,
      "grad_norm": 11.26756477355957,
      "learning_rate": 6.229598458331695e-06,
      "loss": 1.8447,
      "step": 22360
    },
    {
      "epoch": 2.62898107885768,
      "grad_norm": 17.425365447998047,
      "learning_rate": 6.209934321783931e-06,
      "loss": 1.0014,
      "step": 22370
    },
    {
      "epoch": 2.6301563050887298,
      "grad_norm": 56.19034194946289,
      "learning_rate": 6.190270185236167e-06,
      "loss": 1.7482,
      "step": 22380
    },
    {
      "epoch": 2.631331531319779,
      "grad_norm": 16.3543758392334,
      "learning_rate": 6.170606048688402e-06,
      "loss": 1.3954,
      "step": 22390
    },
    {
      "epoch": 2.6325067575508285,
      "grad_norm": 11.160866737365723,
      "learning_rate": 6.150941912140638e-06,
      "loss": 1.0131,
      "step": 22400
    },
    {
      "epoch": 2.633681983781878,
      "grad_norm": 11.705482482910156,
      "learning_rate": 6.1312777755928735e-06,
      "loss": 1.4835,
      "step": 22410
    },
    {
      "epoch": 2.6348572100129273,
      "grad_norm": 23.058252334594727,
      "learning_rate": 6.11161363904511e-06,
      "loss": 1.4482,
      "step": 22420
    },
    {
      "epoch": 2.636032436243977,
      "grad_norm": 9.271536827087402,
      "learning_rate": 6.091949502497345e-06,
      "loss": 1.1097,
      "step": 22430
    },
    {
      "epoch": 2.6372076624750265,
      "grad_norm": 9.999615669250488,
      "learning_rate": 6.072285365949581e-06,
      "loss": 0.8826,
      "step": 22440
    },
    {
      "epoch": 2.638382888706076,
      "grad_norm": 21.02605438232422,
      "learning_rate": 6.052621229401817e-06,
      "loss": 1.4406,
      "step": 22450
    },
    {
      "epoch": 2.6395581149371257,
      "grad_norm": 4.609375,
      "learning_rate": 6.0329570928540535e-06,
      "loss": 0.8789,
      "step": 22460
    },
    {
      "epoch": 2.6407333411681746,
      "grad_norm": 18.492353439331055,
      "learning_rate": 6.013292956306289e-06,
      "loss": 1.2565,
      "step": 22470
    },
    {
      "epoch": 2.6419085673992244,
      "grad_norm": 37.422733306884766,
      "learning_rate": 5.993628819758525e-06,
      "loss": 1.0464,
      "step": 22480
    },
    {
      "epoch": 2.643083793630274,
      "grad_norm": 25.919992446899414,
      "learning_rate": 5.973964683210761e-06,
      "loss": 1.6037,
      "step": 22490
    },
    {
      "epoch": 2.644259019861323,
      "grad_norm": 24.1092586517334,
      "learning_rate": 5.954300546662996e-06,
      "loss": 1.2726,
      "step": 22500
    },
    {
      "epoch": 2.645434246092373,
      "grad_norm": 24.989826202392578,
      "learning_rate": 5.9346364101152325e-06,
      "loss": 2.0884,
      "step": 22510
    },
    {
      "epoch": 2.6466094723234224,
      "grad_norm": 6.826374053955078,
      "learning_rate": 5.914972273567468e-06,
      "loss": 1.2063,
      "step": 22520
    },
    {
      "epoch": 2.6477846985544717,
      "grad_norm": 15.188252449035645,
      "learning_rate": 5.895308137019704e-06,
      "loss": 1.3625,
      "step": 22530
    },
    {
      "epoch": 2.648959924785521,
      "grad_norm": 7.010193824768066,
      "learning_rate": 5.875644000471939e-06,
      "loss": 1.4425,
      "step": 22540
    },
    {
      "epoch": 2.6501351510165705,
      "grad_norm": 18.788055419921875,
      "learning_rate": 5.8559798639241755e-06,
      "loss": 1.8337,
      "step": 22550
    },
    {
      "epoch": 2.6513103772476203,
      "grad_norm": 15.025808334350586,
      "learning_rate": 5.836315727376411e-06,
      "loss": 1.157,
      "step": 22560
    },
    {
      "epoch": 2.6524856034786697,
      "grad_norm": 12.041687965393066,
      "learning_rate": 5.816651590828647e-06,
      "loss": 1.1535,
      "step": 22570
    },
    {
      "epoch": 2.653660829709719,
      "grad_norm": 11.466375350952148,
      "learning_rate": 5.796987454280882e-06,
      "loss": 1.05,
      "step": 22580
    },
    {
      "epoch": 2.6548360559407684,
      "grad_norm": 15.566766738891602,
      "learning_rate": 5.777323317733118e-06,
      "loss": 1.4739,
      "step": 22590
    },
    {
      "epoch": 2.656011282171818,
      "grad_norm": 24.761259078979492,
      "learning_rate": 5.7576591811853545e-06,
      "loss": 1.3564,
      "step": 22600
    },
    {
      "epoch": 2.6571865084028676,
      "grad_norm": 25.0321044921875,
      "learning_rate": 5.737995044637591e-06,
      "loss": 1.4952,
      "step": 22610
    },
    {
      "epoch": 2.658361734633917,
      "grad_norm": 20.8633975982666,
      "learning_rate": 5.718330908089826e-06,
      "loss": 1.6013,
      "step": 22620
    },
    {
      "epoch": 2.6595369608649664,
      "grad_norm": 26.871906280517578,
      "learning_rate": 5.698666771542062e-06,
      "loss": 1.2028,
      "step": 22630
    },
    {
      "epoch": 2.660712187096016,
      "grad_norm": 30.826711654663086,
      "learning_rate": 5.6790026349942975e-06,
      "loss": 1.2864,
      "step": 22640
    },
    {
      "epoch": 2.6618874133270656,
      "grad_norm": 11.813037872314453,
      "learning_rate": 5.659338498446534e-06,
      "loss": 1.0441,
      "step": 22650
    },
    {
      "epoch": 2.663062639558115,
      "grad_norm": 38.587738037109375,
      "learning_rate": 5.63967436189877e-06,
      "loss": 1.3193,
      "step": 22660
    },
    {
      "epoch": 2.6642378657891643,
      "grad_norm": 22.87686538696289,
      "learning_rate": 5.620010225351005e-06,
      "loss": 1.4412,
      "step": 22670
    },
    {
      "epoch": 2.6654130920202137,
      "grad_norm": 17.256914138793945,
      "learning_rate": 5.600346088803241e-06,
      "loss": 2.1424,
      "step": 22680
    },
    {
      "epoch": 2.6665883182512635,
      "grad_norm": 18.627302169799805,
      "learning_rate": 5.5806819522554765e-06,
      "loss": 1.2855,
      "step": 22690
    },
    {
      "epoch": 2.667763544482313,
      "grad_norm": 17.876827239990234,
      "learning_rate": 5.561017815707713e-06,
      "loss": 1.5593,
      "step": 22700
    },
    {
      "epoch": 2.6689387707133623,
      "grad_norm": 19.181137084960938,
      "learning_rate": 5.541353679159948e-06,
      "loss": 1.9758,
      "step": 22710
    },
    {
      "epoch": 2.6701139969444116,
      "grad_norm": 19.139467239379883,
      "learning_rate": 5.521689542612184e-06,
      "loss": 1.7199,
      "step": 22720
    },
    {
      "epoch": 2.671289223175461,
      "grad_norm": 46.88801574707031,
      "learning_rate": 5.5020254060644194e-06,
      "loss": 1.1184,
      "step": 22730
    },
    {
      "epoch": 2.672464449406511,
      "grad_norm": 11.922622680664062,
      "learning_rate": 5.482361269516656e-06,
      "loss": 1.3519,
      "step": 22740
    },
    {
      "epoch": 2.67363967563756,
      "grad_norm": 11.012360572814941,
      "learning_rate": 5.462697132968892e-06,
      "loss": 1.2217,
      "step": 22750
    },
    {
      "epoch": 2.6748149018686096,
      "grad_norm": 16.620941162109375,
      "learning_rate": 5.443032996421128e-06,
      "loss": 1.6631,
      "step": 22760
    },
    {
      "epoch": 2.6759901280996594,
      "grad_norm": 46.564239501953125,
      "learning_rate": 5.423368859873363e-06,
      "loss": 1.2546,
      "step": 22770
    },
    {
      "epoch": 2.677165354330709,
      "grad_norm": 14.2692289352417,
      "learning_rate": 5.403704723325599e-06,
      "loss": 0.9415,
      "step": 22780
    },
    {
      "epoch": 2.678340580561758,
      "grad_norm": 13.929142951965332,
      "learning_rate": 5.384040586777835e-06,
      "loss": 1.5547,
      "step": 22790
    },
    {
      "epoch": 2.6795158067928075,
      "grad_norm": 18.44194793701172,
      "learning_rate": 5.364376450230071e-06,
      "loss": 1.1629,
      "step": 22800
    },
    {
      "epoch": 2.680691033023857,
      "grad_norm": 24.993778228759766,
      "learning_rate": 5.344712313682306e-06,
      "loss": 1.4465,
      "step": 22810
    },
    {
      "epoch": 2.6818662592549067,
      "grad_norm": 14.625763893127441,
      "learning_rate": 5.325048177134542e-06,
      "loss": 1.2818,
      "step": 22820
    },
    {
      "epoch": 2.683041485485956,
      "grad_norm": 6.405436992645264,
      "learning_rate": 5.3053840405867784e-06,
      "loss": 1.5317,
      "step": 22830
    },
    {
      "epoch": 2.6842167117170055,
      "grad_norm": 25.21360969543457,
      "learning_rate": 5.285719904039014e-06,
      "loss": 1.0112,
      "step": 22840
    },
    {
      "epoch": 2.685391937948055,
      "grad_norm": 12.700654983520508,
      "learning_rate": 5.26605576749125e-06,
      "loss": 1.727,
      "step": 22850
    },
    {
      "epoch": 2.6865671641791042,
      "grad_norm": 8.307047843933105,
      "learning_rate": 5.246391630943485e-06,
      "loss": 1.1332,
      "step": 22860
    },
    {
      "epoch": 2.687742390410154,
      "grad_norm": 14.68301773071289,
      "learning_rate": 5.226727494395721e-06,
      "loss": 1.5359,
      "step": 22870
    },
    {
      "epoch": 2.6889176166412034,
      "grad_norm": 20.793806076049805,
      "learning_rate": 5.207063357847957e-06,
      "loss": 1.3287,
      "step": 22880
    },
    {
      "epoch": 2.690092842872253,
      "grad_norm": 23.047325134277344,
      "learning_rate": 5.187399221300193e-06,
      "loss": 1.4209,
      "step": 22890
    },
    {
      "epoch": 2.6912680691033026,
      "grad_norm": 21.457778930664062,
      "learning_rate": 5.167735084752429e-06,
      "loss": 1.2816,
      "step": 22900
    },
    {
      "epoch": 2.692443295334352,
      "grad_norm": 36.38193893432617,
      "learning_rate": 5.148070948204665e-06,
      "loss": 1.4495,
      "step": 22910
    },
    {
      "epoch": 2.6936185215654014,
      "grad_norm": 26.608449935913086,
      "learning_rate": 5.1284068116569004e-06,
      "loss": 1.4332,
      "step": 22920
    },
    {
      "epoch": 2.6947937477964508,
      "grad_norm": 14.749052047729492,
      "learning_rate": 5.108742675109137e-06,
      "loss": 1.9602,
      "step": 22930
    },
    {
      "epoch": 2.6959689740275,
      "grad_norm": 7.070486545562744,
      "learning_rate": 5.089078538561372e-06,
      "loss": 1.633,
      "step": 22940
    },
    {
      "epoch": 2.69714420025855,
      "grad_norm": 17.16889190673828,
      "learning_rate": 5.069414402013608e-06,
      "loss": 1.3701,
      "step": 22950
    },
    {
      "epoch": 2.6983194264895993,
      "grad_norm": 13.496822357177734,
      "learning_rate": 5.049750265465843e-06,
      "loss": 1.8481,
      "step": 22960
    },
    {
      "epoch": 2.6994946527206487,
      "grad_norm": 109.63990020751953,
      "learning_rate": 5.0300861289180795e-06,
      "loss": 1.4753,
      "step": 22970
    },
    {
      "epoch": 2.700669878951698,
      "grad_norm": 22.99593162536621,
      "learning_rate": 5.010421992370315e-06,
      "loss": 0.7877,
      "step": 22980
    },
    {
      "epoch": 2.7018451051827475,
      "grad_norm": 18.079181671142578,
      "learning_rate": 4.990757855822551e-06,
      "loss": 1.4226,
      "step": 22990
    },
    {
      "epoch": 2.7030203314137973,
      "grad_norm": 19.202375411987305,
      "learning_rate": 4.971093719274786e-06,
      "loss": 1.3592,
      "step": 23000
    },
    {
      "epoch": 2.7041955576448466,
      "grad_norm": 17.447731018066406,
      "learning_rate": 4.9514295827270224e-06,
      "loss": 1.292,
      "step": 23010
    },
    {
      "epoch": 2.705370783875896,
      "grad_norm": 8.854750633239746,
      "learning_rate": 4.931765446179259e-06,
      "loss": 1.4736,
      "step": 23020
    },
    {
      "epoch": 2.706546010106946,
      "grad_norm": 33.852500915527344,
      "learning_rate": 4.912101309631494e-06,
      "loss": 1.0986,
      "step": 23030
    },
    {
      "epoch": 2.707721236337995,
      "grad_norm": 11.862406730651855,
      "learning_rate": 4.89243717308373e-06,
      "loss": 1.2289,
      "step": 23040
    },
    {
      "epoch": 2.7088964625690446,
      "grad_norm": 17.070377349853516,
      "learning_rate": 4.872773036535966e-06,
      "loss": 1.4981,
      "step": 23050
    },
    {
      "epoch": 2.710071688800094,
      "grad_norm": 12.903287887573242,
      "learning_rate": 4.853108899988202e-06,
      "loss": 1.3393,
      "step": 23060
    },
    {
      "epoch": 2.7112469150311433,
      "grad_norm": 11.078631401062012,
      "learning_rate": 4.833444763440438e-06,
      "loss": 1.2031,
      "step": 23070
    },
    {
      "epoch": 2.712422141262193,
      "grad_norm": 23.50149917602539,
      "learning_rate": 4.813780626892674e-06,
      "loss": 1.2112,
      "step": 23080
    },
    {
      "epoch": 2.7135973674932425,
      "grad_norm": 15.879312515258789,
      "learning_rate": 4.794116490344909e-06,
      "loss": 1.1836,
      "step": 23090
    },
    {
      "epoch": 2.714772593724292,
      "grad_norm": 16.18121337890625,
      "learning_rate": 4.774452353797145e-06,
      "loss": 1.3094,
      "step": 23100
    },
    {
      "epoch": 2.7159478199553413,
      "grad_norm": 19.564016342163086,
      "learning_rate": 4.754788217249381e-06,
      "loss": 1.576,
      "step": 23110
    },
    {
      "epoch": 2.7171230461863907,
      "grad_norm": 3.698786735534668,
      "learning_rate": 4.735124080701617e-06,
      "loss": 1.4114,
      "step": 23120
    },
    {
      "epoch": 2.7182982724174405,
      "grad_norm": 16.395263671875,
      "learning_rate": 4.715459944153852e-06,
      "loss": 1.1209,
      "step": 23130
    },
    {
      "epoch": 2.71947349864849,
      "grad_norm": 10.32923412322998,
      "learning_rate": 4.695795807606088e-06,
      "loss": 1.998,
      "step": 23140
    },
    {
      "epoch": 2.7206487248795392,
      "grad_norm": 14.747773170471191,
      "learning_rate": 4.6761316710583235e-06,
      "loss": 1.0805,
      "step": 23150
    },
    {
      "epoch": 2.721823951110589,
      "grad_norm": 15.696168899536133,
      "learning_rate": 4.65646753451056e-06,
      "loss": 1.6019,
      "step": 23160
    },
    {
      "epoch": 2.722999177341638,
      "grad_norm": 33.90687561035156,
      "learning_rate": 4.636803397962795e-06,
      "loss": 1.3346,
      "step": 23170
    },
    {
      "epoch": 2.724174403572688,
      "grad_norm": 10.141483306884766,
      "learning_rate": 4.617139261415031e-06,
      "loss": 1.3958,
      "step": 23180
    },
    {
      "epoch": 2.725349629803737,
      "grad_norm": 5.943274974822998,
      "learning_rate": 4.597475124867267e-06,
      "loss": 1.0777,
      "step": 23190
    },
    {
      "epoch": 2.7265248560347866,
      "grad_norm": 20.705005645751953,
      "learning_rate": 4.5778109883195034e-06,
      "loss": 1.3241,
      "step": 23200
    },
    {
      "epoch": 2.7277000822658364,
      "grad_norm": 22.640522003173828,
      "learning_rate": 4.55814685177174e-06,
      "loss": 1.5118,
      "step": 23210
    },
    {
      "epoch": 2.7288753084968858,
      "grad_norm": 7.2748212814331055,
      "learning_rate": 4.538482715223975e-06,
      "loss": 1.2678,
      "step": 23220
    },
    {
      "epoch": 2.730050534727935,
      "grad_norm": 15.97684383392334,
      "learning_rate": 4.518818578676211e-06,
      "loss": 1.2399,
      "step": 23230
    },
    {
      "epoch": 2.7312257609589845,
      "grad_norm": 8.448159217834473,
      "learning_rate": 4.499154442128446e-06,
      "loss": 1.4712,
      "step": 23240
    },
    {
      "epoch": 2.732400987190034,
      "grad_norm": 20.777236938476562,
      "learning_rate": 4.4794903055806825e-06,
      "loss": 1.3789,
      "step": 23250
    },
    {
      "epoch": 2.7335762134210837,
      "grad_norm": 27.2095947265625,
      "learning_rate": 4.459826169032918e-06,
      "loss": 1.4052,
      "step": 23260
    },
    {
      "epoch": 2.734751439652133,
      "grad_norm": 19.372892379760742,
      "learning_rate": 4.440162032485154e-06,
      "loss": 2.0984,
      "step": 23270
    },
    {
      "epoch": 2.7359266658831825,
      "grad_norm": 15.043946266174316,
      "learning_rate": 4.420497895937389e-06,
      "loss": 1.2857,
      "step": 23280
    },
    {
      "epoch": 2.737101892114232,
      "grad_norm": 19.4161319732666,
      "learning_rate": 4.4008337593896254e-06,
      "loss": 1.1967,
      "step": 23290
    },
    {
      "epoch": 2.738277118345281,
      "grad_norm": 5.450069904327393,
      "learning_rate": 4.381169622841861e-06,
      "loss": 0.99,
      "step": 23300
    },
    {
      "epoch": 2.739452344576331,
      "grad_norm": 19.209009170532227,
      "learning_rate": 4.361505486294097e-06,
      "loss": 1.6128,
      "step": 23310
    },
    {
      "epoch": 2.7406275708073804,
      "grad_norm": 10.122377395629883,
      "learning_rate": 4.341841349746332e-06,
      "loss": 1.1391,
      "step": 23320
    },
    {
      "epoch": 2.7418027970384298,
      "grad_norm": 12.0887451171875,
      "learning_rate": 4.322177213198568e-06,
      "loss": 1.0689,
      "step": 23330
    },
    {
      "epoch": 2.7429780232694796,
      "grad_norm": 19.62712860107422,
      "learning_rate": 4.3025130766508045e-06,
      "loss": 1.2213,
      "step": 23340
    },
    {
      "epoch": 2.744153249500529,
      "grad_norm": 19.864606857299805,
      "learning_rate": 4.282848940103041e-06,
      "loss": 1.0953,
      "step": 23350
    },
    {
      "epoch": 2.7453284757315783,
      "grad_norm": 19.32521629333496,
      "learning_rate": 4.263184803555277e-06,
      "loss": 1.3985,
      "step": 23360
    },
    {
      "epoch": 2.7465037019626277,
      "grad_norm": 37.089725494384766,
      "learning_rate": 4.243520667007512e-06,
      "loss": 1.1502,
      "step": 23370
    },
    {
      "epoch": 2.747678928193677,
      "grad_norm": 30.255666732788086,
      "learning_rate": 4.223856530459748e-06,
      "loss": 1.2747,
      "step": 23380
    },
    {
      "epoch": 2.748854154424727,
      "grad_norm": 8.700368881225586,
      "learning_rate": 4.204192393911984e-06,
      "loss": 1.311,
      "step": 23390
    },
    {
      "epoch": 2.7500293806557763,
      "grad_norm": 13.652261734008789,
      "learning_rate": 4.18452825736422e-06,
      "loss": 1.2175,
      "step": 23400
    },
    {
      "epoch": 2.7512046068868257,
      "grad_norm": 32.172950744628906,
      "learning_rate": 4.164864120816455e-06,
      "loss": 1.4464,
      "step": 23410
    },
    {
      "epoch": 2.752379833117875,
      "grad_norm": 11.54848861694336,
      "learning_rate": 4.145199984268691e-06,
      "loss": 1.5226,
      "step": 23420
    },
    {
      "epoch": 2.7535550593489244,
      "grad_norm": 9.29365348815918,
      "learning_rate": 4.1255358477209265e-06,
      "loss": 1.6424,
      "step": 23430
    },
    {
      "epoch": 2.7547302855799742,
      "grad_norm": 16.899948120117188,
      "learning_rate": 4.105871711173163e-06,
      "loss": 1.4744,
      "step": 23440
    },
    {
      "epoch": 2.7559055118110236,
      "grad_norm": 7.628787040710449,
      "learning_rate": 4.086207574625398e-06,
      "loss": 1.2467,
      "step": 23450
    },
    {
      "epoch": 2.757080738042073,
      "grad_norm": 31.29900360107422,
      "learning_rate": 4.066543438077634e-06,
      "loss": 1.8273,
      "step": 23460
    },
    {
      "epoch": 2.758255964273123,
      "grad_norm": 44.98007583618164,
      "learning_rate": 4.0468793015298694e-06,
      "loss": 1.4981,
      "step": 23470
    },
    {
      "epoch": 2.759431190504172,
      "grad_norm": 19.40351676940918,
      "learning_rate": 4.027215164982106e-06,
      "loss": 1.0744,
      "step": 23480
    },
    {
      "epoch": 2.7606064167352216,
      "grad_norm": 17.36796760559082,
      "learning_rate": 4.007551028434342e-06,
      "loss": 1.1839,
      "step": 23490
    },
    {
      "epoch": 2.761781642966271,
      "grad_norm": 13.97750473022461,
      "learning_rate": 3.987886891886578e-06,
      "loss": 1.4704,
      "step": 23500
    },
    {
      "epoch": 2.7629568691973203,
      "grad_norm": 14.367568016052246,
      "learning_rate": 3.968222755338813e-06,
      "loss": 1.3557,
      "step": 23510
    },
    {
      "epoch": 2.76413209542837,
      "grad_norm": 21.048099517822266,
      "learning_rate": 3.948558618791049e-06,
      "loss": 1.4005,
      "step": 23520
    },
    {
      "epoch": 2.7653073216594195,
      "grad_norm": 31.648693084716797,
      "learning_rate": 3.9288944822432855e-06,
      "loss": 1.7625,
      "step": 23530
    },
    {
      "epoch": 2.766482547890469,
      "grad_norm": 18.72760581970215,
      "learning_rate": 3.909230345695521e-06,
      "loss": 1.4771,
      "step": 23540
    },
    {
      "epoch": 2.7676577741215183,
      "grad_norm": 21.667102813720703,
      "learning_rate": 3.889566209147757e-06,
      "loss": 1.1162,
      "step": 23550
    },
    {
      "epoch": 2.7688330003525676,
      "grad_norm": 8.035703659057617,
      "learning_rate": 3.869902072599992e-06,
      "loss": 1.2088,
      "step": 23560
    },
    {
      "epoch": 2.7700082265836174,
      "grad_norm": 31.99890899658203,
      "learning_rate": 3.8502379360522284e-06,
      "loss": 1.6389,
      "step": 23570
    },
    {
      "epoch": 2.771183452814667,
      "grad_norm": 27.33049201965332,
      "learning_rate": 3.830573799504464e-06,
      "loss": 1.1977,
      "step": 23580
    },
    {
      "epoch": 2.772358679045716,
      "grad_norm": 9.074359893798828,
      "learning_rate": 3.8109096629567e-06,
      "loss": 0.9771,
      "step": 23590
    },
    {
      "epoch": 2.773533905276766,
      "grad_norm": 13.425447463989258,
      "learning_rate": 3.7912455264089352e-06,
      "loss": 1.5324,
      "step": 23600
    },
    {
      "epoch": 2.7747091315078154,
      "grad_norm": 14.605291366577148,
      "learning_rate": 3.7715813898611714e-06,
      "loss": 1.3938,
      "step": 23610
    },
    {
      "epoch": 2.7758843577388648,
      "grad_norm": 16.101675033569336,
      "learning_rate": 3.751917253313407e-06,
      "loss": 1.3966,
      "step": 23620
    },
    {
      "epoch": 2.777059583969914,
      "grad_norm": 13.346261024475098,
      "learning_rate": 3.7322531167656433e-06,
      "loss": 0.9184,
      "step": 23630
    },
    {
      "epoch": 2.7782348102009635,
      "grad_norm": 16.364166259765625,
      "learning_rate": 3.7125889802178786e-06,
      "loss": 1.2892,
      "step": 23640
    },
    {
      "epoch": 2.7794100364320133,
      "grad_norm": 12.626721382141113,
      "learning_rate": 3.6929248436701147e-06,
      "loss": 1.2451,
      "step": 23650
    },
    {
      "epoch": 2.7805852626630627,
      "grad_norm": 26.146100997924805,
      "learning_rate": 3.67326070712235e-06,
      "loss": 1.2711,
      "step": 23660
    },
    {
      "epoch": 2.781760488894112,
      "grad_norm": 10.842971801757812,
      "learning_rate": 3.653596570574586e-06,
      "loss": 0.9546,
      "step": 23670
    },
    {
      "epoch": 2.7829357151251615,
      "grad_norm": 26.328899383544922,
      "learning_rate": 3.633932434026822e-06,
      "loss": 1.4624,
      "step": 23680
    },
    {
      "epoch": 2.784110941356211,
      "grad_norm": 14.634683609008789,
      "learning_rate": 3.614268297479058e-06,
      "loss": 1.0851,
      "step": 23690
    },
    {
      "epoch": 2.7852861675872607,
      "grad_norm": 18.1558780670166,
      "learning_rate": 3.5946041609312942e-06,
      "loss": 1.4572,
      "step": 23700
    },
    {
      "epoch": 2.78646139381831,
      "grad_norm": 16.62590217590332,
      "learning_rate": 3.5749400243835295e-06,
      "loss": 2.0014,
      "step": 23710
    },
    {
      "epoch": 2.7876366200493594,
      "grad_norm": 10.819435119628906,
      "learning_rate": 3.5552758878357657e-06,
      "loss": 1.4593,
      "step": 23720
    },
    {
      "epoch": 2.7888118462804092,
      "grad_norm": 35.17469787597656,
      "learning_rate": 3.535611751288001e-06,
      "loss": 1.3254,
      "step": 23730
    },
    {
      "epoch": 2.7899870725114586,
      "grad_norm": 22.90130615234375,
      "learning_rate": 3.515947614740237e-06,
      "loss": 1.5248,
      "step": 23740
    },
    {
      "epoch": 2.791162298742508,
      "grad_norm": 20.544219970703125,
      "learning_rate": 3.4962834781924724e-06,
      "loss": 0.9607,
      "step": 23750
    },
    {
      "epoch": 2.7923375249735574,
      "grad_norm": 10.659529685974121,
      "learning_rate": 3.4766193416447086e-06,
      "loss": 1.2468,
      "step": 23760
    },
    {
      "epoch": 2.7935127512046067,
      "grad_norm": 11.187203407287598,
      "learning_rate": 3.4569552050969443e-06,
      "loss": 1.5681,
      "step": 23770
    },
    {
      "epoch": 2.7946879774356566,
      "grad_norm": 25.691442489624023,
      "learning_rate": 3.4372910685491805e-06,
      "loss": 1.0333,
      "step": 23780
    },
    {
      "epoch": 2.795863203666706,
      "grad_norm": 17.016246795654297,
      "learning_rate": 3.417626932001416e-06,
      "loss": 1.7502,
      "step": 23790
    },
    {
      "epoch": 2.7970384298977553,
      "grad_norm": 28.531660079956055,
      "learning_rate": 3.397962795453652e-06,
      "loss": 1.3474,
      "step": 23800
    },
    {
      "epoch": 2.7982136561288047,
      "grad_norm": 22.761899948120117,
      "learning_rate": 3.3782986589058872e-06,
      "loss": 1.4911,
      "step": 23810
    },
    {
      "epoch": 2.799388882359854,
      "grad_norm": 17.303213119506836,
      "learning_rate": 3.3586345223581234e-06,
      "loss": 1.1879,
      "step": 23820
    },
    {
      "epoch": 2.800564108590904,
      "grad_norm": 10.859908103942871,
      "learning_rate": 3.338970385810359e-06,
      "loss": 1.4378,
      "step": 23830
    },
    {
      "epoch": 2.8017393348219533,
      "grad_norm": 20.369365692138672,
      "learning_rate": 3.3193062492625953e-06,
      "loss": 1.1001,
      "step": 23840
    },
    {
      "epoch": 2.8029145610530026,
      "grad_norm": 37.70854949951172,
      "learning_rate": 3.2996421127148306e-06,
      "loss": 1.2613,
      "step": 23850
    },
    {
      "epoch": 2.8040897872840524,
      "grad_norm": 20.793745040893555,
      "learning_rate": 3.2799779761670667e-06,
      "loss": 1.0603,
      "step": 23860
    },
    {
      "epoch": 2.8052650135151014,
      "grad_norm": 30.015790939331055,
      "learning_rate": 3.260313839619303e-06,
      "loss": 1.5992,
      "step": 23870
    },
    {
      "epoch": 2.806440239746151,
      "grad_norm": 21.205595016479492,
      "learning_rate": 3.240649703071538e-06,
      "loss": 1.3195,
      "step": 23880
    },
    {
      "epoch": 2.8076154659772006,
      "grad_norm": 24.17450523376465,
      "learning_rate": 3.2209855665237744e-06,
      "loss": 1.2252,
      "step": 23890
    },
    {
      "epoch": 2.80879069220825,
      "grad_norm": 4.107827663421631,
      "learning_rate": 3.2013214299760097e-06,
      "loss": 1.0254,
      "step": 23900
    },
    {
      "epoch": 2.8099659184392998,
      "grad_norm": 25.624021530151367,
      "learning_rate": 3.181657293428246e-06,
      "loss": 1.6812,
      "step": 23910
    },
    {
      "epoch": 2.811141144670349,
      "grad_norm": 29.270254135131836,
      "learning_rate": 3.1619931568804816e-06,
      "loss": 1.1394,
      "step": 23920
    },
    {
      "epoch": 2.8123163709013985,
      "grad_norm": 29.76616096496582,
      "learning_rate": 3.1423290203327177e-06,
      "loss": 1.6461,
      "step": 23930
    },
    {
      "epoch": 2.813491597132448,
      "grad_norm": 15.554059028625488,
      "learning_rate": 3.122664883784953e-06,
      "loss": 1.3606,
      "step": 23940
    },
    {
      "epoch": 2.8146668233634973,
      "grad_norm": 11.776524543762207,
      "learning_rate": 3.1030007472371887e-06,
      "loss": 1.9648,
      "step": 23950
    },
    {
      "epoch": 2.815842049594547,
      "grad_norm": 26.021881103515625,
      "learning_rate": 3.083336610689425e-06,
      "loss": 1.1526,
      "step": 23960
    },
    {
      "epoch": 2.8170172758255965,
      "grad_norm": 10.169289588928223,
      "learning_rate": 3.0636724741416606e-06,
      "loss": 1.3976,
      "step": 23970
    },
    {
      "epoch": 2.818192502056646,
      "grad_norm": 30.54312515258789,
      "learning_rate": 3.0440083375938964e-06,
      "loss": 1.5098,
      "step": 23980
    },
    {
      "epoch": 2.819367728287695,
      "grad_norm": 41.130550384521484,
      "learning_rate": 3.0243442010461325e-06,
      "loss": 1.8619,
      "step": 23990
    },
    {
      "epoch": 2.8205429545187446,
      "grad_norm": 14.237061500549316,
      "learning_rate": 3.0046800644983682e-06,
      "loss": 1.1065,
      "step": 24000
    },
    {
      "epoch": 2.8217181807497944,
      "grad_norm": 23.33035659790039,
      "learning_rate": 2.985015927950604e-06,
      "loss": 1.643,
      "step": 24010
    },
    {
      "epoch": 2.822893406980844,
      "grad_norm": 10.485498428344727,
      "learning_rate": 2.9653517914028397e-06,
      "loss": 1.0186,
      "step": 24020
    },
    {
      "epoch": 2.824068633211893,
      "grad_norm": 33.580772399902344,
      "learning_rate": 2.9456876548550754e-06,
      "loss": 1.4325,
      "step": 24030
    },
    {
      "epoch": 2.825243859442943,
      "grad_norm": 8.591379165649414,
      "learning_rate": 2.926023518307311e-06,
      "loss": 1.3265,
      "step": 24040
    },
    {
      "epoch": 2.8264190856739924,
      "grad_norm": 24.55982208251953,
      "learning_rate": 2.906359381759547e-06,
      "loss": 2.0736,
      "step": 24050
    },
    {
      "epoch": 2.8275943119050417,
      "grad_norm": 12.512347221374512,
      "learning_rate": 2.886695245211783e-06,
      "loss": 1.4005,
      "step": 24060
    },
    {
      "epoch": 2.828769538136091,
      "grad_norm": 24.70680046081543,
      "learning_rate": 2.8670311086640188e-06,
      "loss": 1.135,
      "step": 24070
    },
    {
      "epoch": 2.8299447643671405,
      "grad_norm": 12.688603401184082,
      "learning_rate": 2.8473669721162545e-06,
      "loss": 1.3301,
      "step": 24080
    },
    {
      "epoch": 2.8311199905981903,
      "grad_norm": 24.991071701049805,
      "learning_rate": 2.8277028355684902e-06,
      "loss": 1.266,
      "step": 24090
    },
    {
      "epoch": 2.8322952168292397,
      "grad_norm": 18.668519973754883,
      "learning_rate": 2.808038699020726e-06,
      "loss": 1.0194,
      "step": 24100
    },
    {
      "epoch": 2.833470443060289,
      "grad_norm": 16.74786376953125,
      "learning_rate": 2.7883745624729617e-06,
      "loss": 1.3676,
      "step": 24110
    },
    {
      "epoch": 2.8346456692913384,
      "grad_norm": 25.112262725830078,
      "learning_rate": 2.7687104259251974e-06,
      "loss": 1.5282,
      "step": 24120
    },
    {
      "epoch": 2.835820895522388,
      "grad_norm": 21.78006935119629,
      "learning_rate": 2.7490462893774336e-06,
      "loss": 1.3643,
      "step": 24130
    },
    {
      "epoch": 2.8369961217534376,
      "grad_norm": 17.981000900268555,
      "learning_rate": 2.7293821528296693e-06,
      "loss": 1.5836,
      "step": 24140
    },
    {
      "epoch": 2.838171347984487,
      "grad_norm": 5.5786519050598145,
      "learning_rate": 2.7097180162819055e-06,
      "loss": 1.2682,
      "step": 24150
    },
    {
      "epoch": 2.8393465742155364,
      "grad_norm": 12.210711479187012,
      "learning_rate": 2.690053879734141e-06,
      "loss": 1.498,
      "step": 24160
    },
    {
      "epoch": 2.840521800446586,
      "grad_norm": 39.98478317260742,
      "learning_rate": 2.670389743186377e-06,
      "loss": 1.5025,
      "step": 24170
    },
    {
      "epoch": 2.8416970266776356,
      "grad_norm": 13.030213356018066,
      "learning_rate": 2.6507256066386127e-06,
      "loss": 1.7144,
      "step": 24180
    },
    {
      "epoch": 2.842872252908685,
      "grad_norm": 36.93024444580078,
      "learning_rate": 2.6310614700908484e-06,
      "loss": 1.434,
      "step": 24190
    },
    {
      "epoch": 2.8440474791397343,
      "grad_norm": 17.059268951416016,
      "learning_rate": 2.611397333543084e-06,
      "loss": 1.3559,
      "step": 24200
    },
    {
      "epoch": 2.8452227053707837,
      "grad_norm": 33.87356948852539,
      "learning_rate": 2.59173319699532e-06,
      "loss": 1.1942,
      "step": 24210
    },
    {
      "epoch": 2.8463979316018335,
      "grad_norm": 25.449316024780273,
      "learning_rate": 2.572069060447556e-06,
      "loss": 1.534,
      "step": 24220
    },
    {
      "epoch": 2.847573157832883,
      "grad_norm": 15.116425514221191,
      "learning_rate": 2.5524049238997917e-06,
      "loss": 0.9112,
      "step": 24230
    },
    {
      "epoch": 2.8487483840639323,
      "grad_norm": 3.9569170475006104,
      "learning_rate": 2.5327407873520275e-06,
      "loss": 1.2995,
      "step": 24240
    },
    {
      "epoch": 2.8499236102949816,
      "grad_norm": 17.064868927001953,
      "learning_rate": 2.513076650804263e-06,
      "loss": 1.0416,
      "step": 24250
    },
    {
      "epoch": 2.851098836526031,
      "grad_norm": 17.019834518432617,
      "learning_rate": 2.493412514256499e-06,
      "loss": 1.3188,
      "step": 24260
    },
    {
      "epoch": 2.852274062757081,
      "grad_norm": 28.285985946655273,
      "learning_rate": 2.4737483777087347e-06,
      "loss": 1.5585,
      "step": 24270
    },
    {
      "epoch": 2.85344928898813,
      "grad_norm": 8.948596000671387,
      "learning_rate": 2.4540842411609704e-06,
      "loss": 1.582,
      "step": 24280
    },
    {
      "epoch": 2.8546245152191796,
      "grad_norm": 17.273008346557617,
      "learning_rate": 2.4344201046132066e-06,
      "loss": 1.428,
      "step": 24290
    },
    {
      "epoch": 2.8557997414502294,
      "grad_norm": 11.481239318847656,
      "learning_rate": 2.4147559680654427e-06,
      "loss": 1.4091,
      "step": 24300
    },
    {
      "epoch": 2.856974967681279,
      "grad_norm": 36.896305084228516,
      "learning_rate": 2.3950918315176784e-06,
      "loss": 2.0099,
      "step": 24310
    },
    {
      "epoch": 2.858150193912328,
      "grad_norm": 18.650798797607422,
      "learning_rate": 2.375427694969914e-06,
      "loss": 1.5642,
      "step": 24320
    },
    {
      "epoch": 2.8593254201433775,
      "grad_norm": 12.08358097076416,
      "learning_rate": 2.35576355842215e-06,
      "loss": 1.1136,
      "step": 24330
    },
    {
      "epoch": 2.860500646374427,
      "grad_norm": 3.555288076400757,
      "learning_rate": 2.3360994218743856e-06,
      "loss": 1.1243,
      "step": 24340
    },
    {
      "epoch": 2.8616758726054767,
      "grad_norm": 26.335939407348633,
      "learning_rate": 2.3164352853266214e-06,
      "loss": 1.4088,
      "step": 24350
    },
    {
      "epoch": 2.862851098836526,
      "grad_norm": 18.644290924072266,
      "learning_rate": 2.296771148778857e-06,
      "loss": 0.8798,
      "step": 24360
    },
    {
      "epoch": 2.8640263250675755,
      "grad_norm": 23.003786087036133,
      "learning_rate": 2.2771070122310932e-06,
      "loss": 1.5891,
      "step": 24370
    },
    {
      "epoch": 2.865201551298625,
      "grad_norm": 24.916168212890625,
      "learning_rate": 2.257442875683329e-06,
      "loss": 1.6428,
      "step": 24380
    },
    {
      "epoch": 2.8663767775296742,
      "grad_norm": 17.43243980407715,
      "learning_rate": 2.2377787391355647e-06,
      "loss": 1.6664,
      "step": 24390
    },
    {
      "epoch": 2.867552003760724,
      "grad_norm": 19.877260208129883,
      "learning_rate": 2.2181146025878004e-06,
      "loss": 1.4873,
      "step": 24400
    },
    {
      "epoch": 2.8687272299917734,
      "grad_norm": 21.63187599182129,
      "learning_rate": 2.198450466040036e-06,
      "loss": 1.394,
      "step": 24410
    },
    {
      "epoch": 2.869902456222823,
      "grad_norm": 25.870420455932617,
      "learning_rate": 2.178786329492272e-06,
      "loss": 1.6538,
      "step": 24420
    },
    {
      "epoch": 2.8710776824538726,
      "grad_norm": 13.812296867370605,
      "learning_rate": 2.1591221929445076e-06,
      "loss": 1.4308,
      "step": 24430
    },
    {
      "epoch": 2.872252908684922,
      "grad_norm": 12.17618179321289,
      "learning_rate": 2.1394580563967438e-06,
      "loss": 1.2321,
      "step": 24440
    },
    {
      "epoch": 2.8734281349159714,
      "grad_norm": 16.4500732421875,
      "learning_rate": 2.1197939198489795e-06,
      "loss": 1.578,
      "step": 24450
    },
    {
      "epoch": 2.8746033611470208,
      "grad_norm": 16.73348617553711,
      "learning_rate": 2.1001297833012152e-06,
      "loss": 0.9939,
      "step": 24460
    },
    {
      "epoch": 2.87577858737807,
      "grad_norm": 19.606544494628906,
      "learning_rate": 2.080465646753451e-06,
      "loss": 1.4311,
      "step": 24470
    },
    {
      "epoch": 2.87695381360912,
      "grad_norm": 14.202588081359863,
      "learning_rate": 2.060801510205687e-06,
      "loss": 1.422,
      "step": 24480
    },
    {
      "epoch": 2.8781290398401693,
      "grad_norm": 23.62276840209961,
      "learning_rate": 2.041137373657923e-06,
      "loss": 1.7601,
      "step": 24490
    },
    {
      "epoch": 2.8793042660712187,
      "grad_norm": 14.658174514770508,
      "learning_rate": 2.0214732371101586e-06,
      "loss": 1.3367,
      "step": 24500
    },
    {
      "epoch": 2.880479492302268,
      "grad_norm": 13.585407257080078,
      "learning_rate": 2.0018091005623943e-06,
      "loss": 1.085,
      "step": 24510
    },
    {
      "epoch": 2.8816547185333174,
      "grad_norm": 10.474812507629395,
      "learning_rate": 1.9821449640146305e-06,
      "loss": 1.2994,
      "step": 24520
    },
    {
      "epoch": 2.8828299447643673,
      "grad_norm": 8.92325210571289,
      "learning_rate": 1.962480827466866e-06,
      "loss": 1.1297,
      "step": 24530
    },
    {
      "epoch": 2.8840051709954166,
      "grad_norm": 15.499917030334473,
      "learning_rate": 1.942816690919102e-06,
      "loss": 1.2287,
      "step": 24540
    },
    {
      "epoch": 2.885180397226466,
      "grad_norm": 15.374752044677734,
      "learning_rate": 1.9231525543713377e-06,
      "loss": 1.9246,
      "step": 24550
    },
    {
      "epoch": 2.886355623457516,
      "grad_norm": 21.88521957397461,
      "learning_rate": 1.9034884178235734e-06,
      "loss": 1.1655,
      "step": 24560
    },
    {
      "epoch": 2.8875308496885648,
      "grad_norm": 10.124180793762207,
      "learning_rate": 1.8838242812758093e-06,
      "loss": 0.9984,
      "step": 24570
    },
    {
      "epoch": 2.8887060759196146,
      "grad_norm": 32.11320495605469,
      "learning_rate": 1.864160144728045e-06,
      "loss": 1.3556,
      "step": 24580
    },
    {
      "epoch": 2.889881302150664,
      "grad_norm": 12.486140251159668,
      "learning_rate": 1.8444960081802808e-06,
      "loss": 1.7318,
      "step": 24590
    },
    {
      "epoch": 2.8910565283817133,
      "grad_norm": 25.310298919677734,
      "learning_rate": 1.8248318716325165e-06,
      "loss": 1.417,
      "step": 24600
    },
    {
      "epoch": 2.892231754612763,
      "grad_norm": 19.250789642333984,
      "learning_rate": 1.8051677350847525e-06,
      "loss": 1.5663,
      "step": 24610
    },
    {
      "epoch": 2.8934069808438125,
      "grad_norm": 14.648984909057617,
      "learning_rate": 1.7855035985369882e-06,
      "loss": 1.0855,
      "step": 24620
    },
    {
      "epoch": 2.894582207074862,
      "grad_norm": 38.924888610839844,
      "learning_rate": 1.765839461989224e-06,
      "loss": 1.6293,
      "step": 24630
    },
    {
      "epoch": 2.8957574333059113,
      "grad_norm": 19.63701820373535,
      "learning_rate": 1.7461753254414599e-06,
      "loss": 1.372,
      "step": 24640
    },
    {
      "epoch": 2.8969326595369607,
      "grad_norm": 19.482145309448242,
      "learning_rate": 1.7265111888936958e-06,
      "loss": 1.4076,
      "step": 24650
    },
    {
      "epoch": 2.8981078857680105,
      "grad_norm": 18.064640045166016,
      "learning_rate": 1.7068470523459318e-06,
      "loss": 1.4675,
      "step": 24660
    },
    {
      "epoch": 2.89928311199906,
      "grad_norm": 8.220746994018555,
      "learning_rate": 1.6871829157981675e-06,
      "loss": 1.2748,
      "step": 24670
    },
    {
      "epoch": 2.9004583382301092,
      "grad_norm": 24.8601131439209,
      "learning_rate": 1.6675187792504032e-06,
      "loss": 1.4232,
      "step": 24680
    },
    {
      "epoch": 2.9016335644611586,
      "grad_norm": 33.3975944519043,
      "learning_rate": 1.6478546427026392e-06,
      "loss": 1.3688,
      "step": 24690
    },
    {
      "epoch": 2.902808790692208,
      "grad_norm": 15.687925338745117,
      "learning_rate": 1.628190506154875e-06,
      "loss": 1.2377,
      "step": 24700
    },
    {
      "epoch": 2.903984016923258,
      "grad_norm": 28.61330795288086,
      "learning_rate": 1.6085263696071106e-06,
      "loss": 1.3049,
      "step": 24710
    },
    {
      "epoch": 2.905159243154307,
      "grad_norm": 15.653822898864746,
      "learning_rate": 1.5888622330593464e-06,
      "loss": 1.6557,
      "step": 24720
    },
    {
      "epoch": 2.9063344693853566,
      "grad_norm": 31.916793823242188,
      "learning_rate": 1.5691980965115823e-06,
      "loss": 1.3387,
      "step": 24730
    },
    {
      "epoch": 2.9075096956164064,
      "grad_norm": 19.499357223510742,
      "learning_rate": 1.549533959963818e-06,
      "loss": 1.3564,
      "step": 24740
    },
    {
      "epoch": 2.9086849218474558,
      "grad_norm": 6.992166042327881,
      "learning_rate": 1.5298698234160538e-06,
      "loss": 1.7905,
      "step": 24750
    },
    {
      "epoch": 2.909860148078505,
      "grad_norm": 34.100242614746094,
      "learning_rate": 1.5102056868682897e-06,
      "loss": 1.529,
      "step": 24760
    },
    {
      "epoch": 2.9110353743095545,
      "grad_norm": 24.446857452392578,
      "learning_rate": 1.4905415503205254e-06,
      "loss": 1.6335,
      "step": 24770
    },
    {
      "epoch": 2.912210600540604,
      "grad_norm": 4.785913467407227,
      "learning_rate": 1.4708774137727614e-06,
      "loss": 0.9572,
      "step": 24780
    },
    {
      "epoch": 2.9133858267716537,
      "grad_norm": 13.095010757446289,
      "learning_rate": 1.4512132772249971e-06,
      "loss": 1.7541,
      "step": 24790
    },
    {
      "epoch": 2.914561053002703,
      "grad_norm": 7.214615821838379,
      "learning_rate": 1.431549140677233e-06,
      "loss": 1.5426,
      "step": 24800
    },
    {
      "epoch": 2.9157362792337524,
      "grad_norm": 14.999968528747559,
      "learning_rate": 1.4118850041294688e-06,
      "loss": 1.7624,
      "step": 24810
    },
    {
      "epoch": 2.916911505464802,
      "grad_norm": 27.681489944458008,
      "learning_rate": 1.3922208675817045e-06,
      "loss": 1.8431,
      "step": 24820
    },
    {
      "epoch": 2.918086731695851,
      "grad_norm": 12.736488342285156,
      "learning_rate": 1.3725567310339402e-06,
      "loss": 1.5843,
      "step": 24830
    },
    {
      "epoch": 2.919261957926901,
      "grad_norm": 16.52277183532715,
      "learning_rate": 1.3528925944861762e-06,
      "loss": 1.2579,
      "step": 24840
    },
    {
      "epoch": 2.9204371841579504,
      "grad_norm": 12.794143676757812,
      "learning_rate": 1.333228457938412e-06,
      "loss": 1.2922,
      "step": 24850
    },
    {
      "epoch": 2.9216124103889998,
      "grad_norm": 28.93718910217285,
      "learning_rate": 1.3135643213906476e-06,
      "loss": 1.6462,
      "step": 24860
    },
    {
      "epoch": 2.9227876366200496,
      "grad_norm": 5.062346458435059,
      "learning_rate": 1.2939001848428836e-06,
      "loss": 1.0106,
      "step": 24870
    },
    {
      "epoch": 2.923962862851099,
      "grad_norm": 14.353105545043945,
      "learning_rate": 1.2742360482951195e-06,
      "loss": 1.4745,
      "step": 24880
    },
    {
      "epoch": 2.9251380890821483,
      "grad_norm": 9.255575180053711,
      "learning_rate": 1.2545719117473553e-06,
      "loss": 1.4127,
      "step": 24890
    },
    {
      "epoch": 2.9263133153131977,
      "grad_norm": 12.238701820373535,
      "learning_rate": 1.234907775199591e-06,
      "loss": 1.4194,
      "step": 24900
    },
    {
      "epoch": 2.927488541544247,
      "grad_norm": 14.429340362548828,
      "learning_rate": 1.215243638651827e-06,
      "loss": 1.8294,
      "step": 24910
    },
    {
      "epoch": 2.928663767775297,
      "grad_norm": 12.389949798583984,
      "learning_rate": 1.1955795021040627e-06,
      "loss": 1.5481,
      "step": 24920
    },
    {
      "epoch": 2.9298389940063463,
      "grad_norm": 12.511832237243652,
      "learning_rate": 1.1759153655562984e-06,
      "loss": 1.4829,
      "step": 24930
    },
    {
      "epoch": 2.9310142202373957,
      "grad_norm": 24.27159309387207,
      "learning_rate": 1.1562512290085341e-06,
      "loss": 1.1668,
      "step": 24940
    },
    {
      "epoch": 2.932189446468445,
      "grad_norm": 7.752825736999512,
      "learning_rate": 1.1365870924607703e-06,
      "loss": 1.1875,
      "step": 24950
    },
    {
      "epoch": 2.9333646726994944,
      "grad_norm": 22.447643280029297,
      "learning_rate": 1.116922955913006e-06,
      "loss": 1.2485,
      "step": 24960
    },
    {
      "epoch": 2.9345398989305442,
      "grad_norm": 10.348435401916504,
      "learning_rate": 1.0972588193652417e-06,
      "loss": 0.9972,
      "step": 24970
    },
    {
      "epoch": 2.9357151251615936,
      "grad_norm": 16.046140670776367,
      "learning_rate": 1.0775946828174775e-06,
      "loss": 1.3468,
      "step": 24980
    },
    {
      "epoch": 2.936890351392643,
      "grad_norm": 18.307031631469727,
      "learning_rate": 1.0579305462697134e-06,
      "loss": 1.0302,
      "step": 24990
    },
    {
      "epoch": 2.938065577623693,
      "grad_norm": 6.409160137176514,
      "learning_rate": 1.0382664097219491e-06,
      "loss": 1.0101,
      "step": 25000
    },
    {
      "epoch": 2.939240803854742,
      "grad_norm": 19.389976501464844,
      "learning_rate": 1.0186022731741849e-06,
      "loss": 1.4552,
      "step": 25010
    },
    {
      "epoch": 2.9404160300857916,
      "grad_norm": 19.038022994995117,
      "learning_rate": 9.989381366264208e-07,
      "loss": 1.2641,
      "step": 25020
    },
    {
      "epoch": 2.941591256316841,
      "grad_norm": 10.185078620910645,
      "learning_rate": 9.792740000786565e-07,
      "loss": 1.4889,
      "step": 25030
    },
    {
      "epoch": 2.9427664825478903,
      "grad_norm": 16.847023010253906,
      "learning_rate": 9.596098635308925e-07,
      "loss": 1.8244,
      "step": 25040
    },
    {
      "epoch": 2.94394170877894,
      "grad_norm": 20.172414779663086,
      "learning_rate": 9.399457269831282e-07,
      "loss": 1.4358,
      "step": 25050
    },
    {
      "epoch": 2.9451169350099895,
      "grad_norm": 25.609375,
      "learning_rate": 9.202815904353641e-07,
      "loss": 1.4518,
      "step": 25060
    },
    {
      "epoch": 2.946292161241039,
      "grad_norm": 21.025848388671875,
      "learning_rate": 9.006174538875999e-07,
      "loss": 1.2864,
      "step": 25070
    },
    {
      "epoch": 2.9474673874720883,
      "grad_norm": 21.3228759765625,
      "learning_rate": 8.809533173398356e-07,
      "loss": 1.486,
      "step": 25080
    },
    {
      "epoch": 2.9486426137031376,
      "grad_norm": 26.649381637573242,
      "learning_rate": 8.612891807920715e-07,
      "loss": 1.2193,
      "step": 25090
    },
    {
      "epoch": 2.9498178399341874,
      "grad_norm": 19.233654022216797,
      "learning_rate": 8.416250442443072e-07,
      "loss": 1.3321,
      "step": 25100
    },
    {
      "epoch": 2.950993066165237,
      "grad_norm": 16.1801815032959,
      "learning_rate": 8.21960907696543e-07,
      "loss": 1.2809,
      "step": 25110
    },
    {
      "epoch": 2.952168292396286,
      "grad_norm": 22.231468200683594,
      "learning_rate": 8.022967711487788e-07,
      "loss": 1.2844,
      "step": 25120
    },
    {
      "epoch": 2.953343518627336,
      "grad_norm": 24.559680938720703,
      "learning_rate": 7.826326346010148e-07,
      "loss": 1.3875,
      "step": 25130
    },
    {
      "epoch": 2.9545187448583854,
      "grad_norm": 15.959258079528809,
      "learning_rate": 7.629684980532504e-07,
      "loss": 1.4286,
      "step": 25140
    },
    {
      "epoch": 2.9556939710894348,
      "grad_norm": 9.984113693237305,
      "learning_rate": 7.433043615054864e-07,
      "loss": 1.3627,
      "step": 25150
    },
    {
      "epoch": 2.956869197320484,
      "grad_norm": 12.144821166992188,
      "learning_rate": 7.236402249577221e-07,
      "loss": 1.1897,
      "step": 25160
    },
    {
      "epoch": 2.9580444235515335,
      "grad_norm": 23.222610473632812,
      "learning_rate": 7.039760884099579e-07,
      "loss": 0.9743,
      "step": 25170
    },
    {
      "epoch": 2.9592196497825833,
      "grad_norm": 34.364341735839844,
      "learning_rate": 6.843119518621938e-07,
      "loss": 1.5372,
      "step": 25180
    },
    {
      "epoch": 2.9603948760136327,
      "grad_norm": 36.66718673706055,
      "learning_rate": 6.646478153144296e-07,
      "loss": 1.4797,
      "step": 25190
    },
    {
      "epoch": 2.961570102244682,
      "grad_norm": 7.853646278381348,
      "learning_rate": 6.449836787666655e-07,
      "loss": 1.0789,
      "step": 25200
    },
    {
      "epoch": 2.9627453284757315,
      "grad_norm": 13.912107467651367,
      "learning_rate": 6.253195422189012e-07,
      "loss": 1.3635,
      "step": 25210
    },
    {
      "epoch": 2.963920554706781,
      "grad_norm": 26.352758407592773,
      "learning_rate": 6.05655405671137e-07,
      "loss": 1.3468,
      "step": 25220
    },
    {
      "epoch": 2.9650957809378307,
      "grad_norm": 8.999430656433105,
      "learning_rate": 5.859912691233729e-07,
      "loss": 1.6498,
      "step": 25230
    },
    {
      "epoch": 2.96627100716888,
      "grad_norm": 17.786813735961914,
      "learning_rate": 5.663271325756087e-07,
      "loss": 1.2401,
      "step": 25240
    },
    {
      "epoch": 2.9674462333999294,
      "grad_norm": 26.82585334777832,
      "learning_rate": 5.466629960278444e-07,
      "loss": 1.6294,
      "step": 25250
    },
    {
      "epoch": 2.9686214596309792,
      "grad_norm": 13.783058166503906,
      "learning_rate": 5.269988594800803e-07,
      "loss": 1.0497,
      "step": 25260
    },
    {
      "epoch": 2.9697966858620286,
      "grad_norm": 21.83885383605957,
      "learning_rate": 5.07334722932316e-07,
      "loss": 1.3258,
      "step": 25270
    },
    {
      "epoch": 2.970971912093078,
      "grad_norm": 14.437551498413086,
      "learning_rate": 4.876705863845519e-07,
      "loss": 1.2884,
      "step": 25280
    },
    {
      "epoch": 2.9721471383241274,
      "grad_norm": 13.761397361755371,
      "learning_rate": 4.680064498367877e-07,
      "loss": 1.353,
      "step": 25290
    },
    {
      "epoch": 2.9733223645551767,
      "grad_norm": 12.721868515014648,
      "learning_rate": 4.483423132890235e-07,
      "loss": 1.3247,
      "step": 25300
    },
    {
      "epoch": 2.9744975907862266,
      "grad_norm": 30.464662551879883,
      "learning_rate": 4.286781767412593e-07,
      "loss": 1.1058,
      "step": 25310
    },
    {
      "epoch": 2.975672817017276,
      "grad_norm": 34.93251037597656,
      "learning_rate": 4.0901404019349517e-07,
      "loss": 1.1942,
      "step": 25320
    },
    {
      "epoch": 2.9768480432483253,
      "grad_norm": 16.89225959777832,
      "learning_rate": 3.8934990364573096e-07,
      "loss": 1.7646,
      "step": 25330
    },
    {
      "epoch": 2.9780232694793747,
      "grad_norm": 12.915813446044922,
      "learning_rate": 3.6968576709796674e-07,
      "loss": 1.2145,
      "step": 25340
    },
    {
      "epoch": 2.979198495710424,
      "grad_norm": 21.51253318786621,
      "learning_rate": 3.500216305502026e-07,
      "loss": 1.0515,
      "step": 25350
    },
    {
      "epoch": 2.980373721941474,
      "grad_norm": 36.820457458496094,
      "learning_rate": 3.3035749400243836e-07,
      "loss": 1.3798,
      "step": 25360
    },
    {
      "epoch": 2.9815489481725233,
      "grad_norm": 13.416136741638184,
      "learning_rate": 3.1069335745467414e-07,
      "loss": 1.7547,
      "step": 25370
    },
    {
      "epoch": 2.9827241744035726,
      "grad_norm": 14.17188835144043,
      "learning_rate": 2.9102922090691e-07,
      "loss": 1.3652,
      "step": 25380
    },
    {
      "epoch": 2.983899400634622,
      "grad_norm": 10.272555351257324,
      "learning_rate": 2.7136508435914576e-07,
      "loss": 1.1332,
      "step": 25390
    },
    {
      "epoch": 2.9850746268656714,
      "grad_norm": 31.019784927368164,
      "learning_rate": 2.517009478113816e-07,
      "loss": 1.1626,
      "step": 25400
    },
    {
      "epoch": 2.986249853096721,
      "grad_norm": 8.190850257873535,
      "learning_rate": 2.320368112636174e-07,
      "loss": 1.4348,
      "step": 25410
    },
    {
      "epoch": 2.9874250793277706,
      "grad_norm": 23.045246124267578,
      "learning_rate": 2.1237267471585325e-07,
      "loss": 1.7187,
      "step": 25420
    },
    {
      "epoch": 2.98860030555882,
      "grad_norm": 17.408681869506836,
      "learning_rate": 1.9270853816808906e-07,
      "loss": 1.4528,
      "step": 25430
    },
    {
      "epoch": 2.9897755317898698,
      "grad_norm": 23.260929107666016,
      "learning_rate": 1.7304440162032484e-07,
      "loss": 1.3289,
      "step": 25440
    },
    {
      "epoch": 2.990950758020919,
      "grad_norm": 17.157920837402344,
      "learning_rate": 1.5338026507256065e-07,
      "loss": 1.3095,
      "step": 25450
    },
    {
      "epoch": 2.9921259842519685,
      "grad_norm": 16.402284622192383,
      "learning_rate": 1.3371612852479649e-07,
      "loss": 1.2577,
      "step": 25460
    },
    {
      "epoch": 2.993301210483018,
      "grad_norm": 13.327427864074707,
      "learning_rate": 1.1405199197703228e-07,
      "loss": 1.0474,
      "step": 25470
    },
    {
      "epoch": 2.9944764367140673,
      "grad_norm": 30.685400009155273,
      "learning_rate": 9.43878554292681e-08,
      "loss": 1.3913,
      "step": 25480
    },
    {
      "epoch": 2.995651662945117,
      "grad_norm": 10.82269287109375,
      "learning_rate": 7.472371888150392e-08,
      "loss": 1.5469,
      "step": 25490
    },
    {
      "epoch": 2.9968268891761665,
      "grad_norm": 10.259984016418457,
      "learning_rate": 5.5059582333739727e-08,
      "loss": 1.4276,
      "step": 25500
    },
    {
      "epoch": 2.998002115407216,
      "grad_norm": 30.44392967224121,
      "learning_rate": 3.539544578597554e-08,
      "loss": 1.4366,
      "step": 25510
    },
    {
      "epoch": 2.999177341638265,
      "grad_norm": 20.76551628112793,
      "learning_rate": 1.573130923821135e-08,
      "loss": 1.0587,
      "step": 25520
    }
  ],
  "logging_steps": 10,
  "max_steps": 25527,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3470123167288800.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}
