{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.3000901233520383,
  "eval_steps": 1890,
  "global_step": 9449,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00027518580775060826,
      "grad_norm": 7.053801536560059,
      "learning_rate": 8.333333333333333e-05,
      "loss": 11.388574600219727,
      "step": 2
    },
    {
      "epoch": 0.0005503716155012165,
      "grad_norm": 7.558135032653809,
      "learning_rate": 0.00025,
      "loss": 11.739964485168457,
      "step": 4
    },
    {
      "epoch": 0.0008255574232518248,
      "grad_norm": 3.583465576171875,
      "learning_rate": 0.0004166666666666667,
      "loss": 10.599815368652344,
      "step": 6
    },
    {
      "epoch": 0.001100743231002433,
      "grad_norm": 25.06006622314453,
      "learning_rate": 0.0005833333333333334,
      "loss": 13.511728286743164,
      "step": 8
    },
    {
      "epoch": 0.0013759290387530413,
      "grad_norm": 3.470775842666626,
      "learning_rate": 0.00075,
      "loss": 9.371379852294922,
      "step": 10
    },
    {
      "epoch": 0.0016511148465036497,
      "grad_norm": 2.4591104984283447,
      "learning_rate": 0.0009166666666666666,
      "loss": 8.809574127197266,
      "step": 12
    },
    {
      "epoch": 0.001926300654254258,
      "grad_norm": 22.531356811523438,
      "learning_rate": 0.0010833333333333333,
      "loss": 9.198809623718262,
      "step": 14
    },
    {
      "epoch": 0.002201486462004866,
      "grad_norm": 8.94678783416748,
      "learning_rate": 0.00125,
      "loss": 9.365514755249023,
      "step": 16
    },
    {
      "epoch": 0.0024766722697554745,
      "grad_norm": 1.9878818988800049,
      "learning_rate": 0.0014166666666666668,
      "loss": 9.109071731567383,
      "step": 18
    },
    {
      "epoch": 0.0027518580775060825,
      "grad_norm": 2.3458945751190186,
      "learning_rate": 0.0015833333333333333,
      "loss": 8.711134910583496,
      "step": 20
    },
    {
      "epoch": 0.003027043885256691,
      "grad_norm": 1.1199277639389038,
      "learning_rate": 0.00175,
      "loss": 8.785715103149414,
      "step": 22
    },
    {
      "epoch": 0.0033022296930072994,
      "grad_norm": 1.8996467590332031,
      "learning_rate": 0.0019166666666666668,
      "loss": 8.655513763427734,
      "step": 24
    },
    {
      "epoch": 0.0035774155007579074,
      "grad_norm": 1.4467793703079224,
      "learning_rate": 0.002,
      "loss": 8.684551239013672,
      "step": 26
    },
    {
      "epoch": 0.003852601308508516,
      "grad_norm": 1.1416981220245361,
      "learning_rate": 0.002,
      "loss": 8.654457092285156,
      "step": 28
    },
    {
      "epoch": 0.004127787116259124,
      "grad_norm": 1.2193622589111328,
      "learning_rate": 0.002,
      "loss": 8.542226791381836,
      "step": 30
    },
    {
      "epoch": 0.004402972924009732,
      "grad_norm": 0.9774078130722046,
      "learning_rate": 0.002,
      "loss": 8.54671859741211,
      "step": 32
    },
    {
      "epoch": 0.004678158731760341,
      "grad_norm": 0.805271327495575,
      "learning_rate": 0.002,
      "loss": 8.447345733642578,
      "step": 34
    },
    {
      "epoch": 0.004953344539510949,
      "grad_norm": 0.6902488470077515,
      "learning_rate": 0.002,
      "loss": 8.501413345336914,
      "step": 36
    },
    {
      "epoch": 0.0052285303472615575,
      "grad_norm": 0.8730037808418274,
      "learning_rate": 0.002,
      "loss": 8.443878173828125,
      "step": 38
    },
    {
      "epoch": 0.005503716155012165,
      "grad_norm": 1.0465078353881836,
      "learning_rate": 0.002,
      "loss": 8.443363189697266,
      "step": 40
    },
    {
      "epoch": 0.0057789019627627735,
      "grad_norm": 1.2899569272994995,
      "learning_rate": 0.002,
      "loss": 8.444768905639648,
      "step": 42
    },
    {
      "epoch": 0.006054087770513382,
      "grad_norm": 1.150313377380371,
      "learning_rate": 0.002,
      "loss": 8.377187728881836,
      "step": 44
    },
    {
      "epoch": 0.00632927357826399,
      "grad_norm": 2.412557601928711,
      "learning_rate": 0.002,
      "loss": 8.395917892456055,
      "step": 46
    },
    {
      "epoch": 0.006604459386014599,
      "grad_norm": 1.0635197162628174,
      "learning_rate": 0.002,
      "loss": 8.345839500427246,
      "step": 48
    },
    {
      "epoch": 0.006879645193765207,
      "grad_norm": 1.4153708219528198,
      "learning_rate": 0.002,
      "loss": 8.219489097595215,
      "step": 50
    },
    {
      "epoch": 0.007154831001515815,
      "grad_norm": 0.9424594640731812,
      "learning_rate": 0.002,
      "loss": 8.182621002197266,
      "step": 52
    },
    {
      "epoch": 0.007430016809266423,
      "grad_norm": 1.4184988737106323,
      "learning_rate": 0.002,
      "loss": 8.145915985107422,
      "step": 54
    },
    {
      "epoch": 0.007705202617017032,
      "grad_norm": 2.1485114097595215,
      "learning_rate": 0.002,
      "loss": 8.145328521728516,
      "step": 56
    },
    {
      "epoch": 0.00798038842476764,
      "grad_norm": 1.0020501613616943,
      "learning_rate": 0.002,
      "loss": 8.076604843139648,
      "step": 58
    },
    {
      "epoch": 0.008255574232518248,
      "grad_norm": 1.2933049201965332,
      "learning_rate": 0.002,
      "loss": 8.005566596984863,
      "step": 60
    },
    {
      "epoch": 0.008530760040268856,
      "grad_norm": 0.9638057947158813,
      "learning_rate": 0.002,
      "loss": 8.102912902832031,
      "step": 62
    },
    {
      "epoch": 0.008805945848019464,
      "grad_norm": 0.736072301864624,
      "learning_rate": 0.002,
      "loss": 8.012557983398438,
      "step": 64
    },
    {
      "epoch": 0.009081131655770073,
      "grad_norm": 1.3946400880813599,
      "learning_rate": 0.002,
      "loss": 7.903233528137207,
      "step": 66
    },
    {
      "epoch": 0.009356317463520681,
      "grad_norm": 0.49676042795181274,
      "learning_rate": 0.002,
      "loss": 7.9667229652404785,
      "step": 68
    },
    {
      "epoch": 0.00963150327127129,
      "grad_norm": 0.6100605130195618,
      "learning_rate": 0.002,
      "loss": 7.994020462036133,
      "step": 70
    },
    {
      "epoch": 0.009906689079021898,
      "grad_norm": 0.4911879301071167,
      "learning_rate": 0.002,
      "loss": 7.875064849853516,
      "step": 72
    },
    {
      "epoch": 0.010181874886772507,
      "grad_norm": 0.4266131818294525,
      "learning_rate": 0.002,
      "loss": 7.929782390594482,
      "step": 74
    },
    {
      "epoch": 0.010457060694523115,
      "grad_norm": 0.44705379009246826,
      "learning_rate": 0.002,
      "loss": 7.812183856964111,
      "step": 76
    },
    {
      "epoch": 0.010732246502273723,
      "grad_norm": 0.5087490677833557,
      "learning_rate": 0.002,
      "loss": 7.8495683670043945,
      "step": 78
    },
    {
      "epoch": 0.01100743231002433,
      "grad_norm": 0.44210085272789,
      "learning_rate": 0.002,
      "loss": 7.780263900756836,
      "step": 80
    },
    {
      "epoch": 0.011282618117774939,
      "grad_norm": 0.46045753359794617,
      "learning_rate": 0.002,
      "loss": 7.804459571838379,
      "step": 82
    },
    {
      "epoch": 0.011557803925525547,
      "grad_norm": 0.46733689308166504,
      "learning_rate": 0.002,
      "loss": 7.755415916442871,
      "step": 84
    },
    {
      "epoch": 0.011832989733276155,
      "grad_norm": 0.29618290066719055,
      "learning_rate": 0.002,
      "loss": 7.636168003082275,
      "step": 86
    },
    {
      "epoch": 0.012108175541026764,
      "grad_norm": 0.26736584305763245,
      "learning_rate": 0.002,
      "loss": 7.642195701599121,
      "step": 88
    },
    {
      "epoch": 0.012383361348777372,
      "grad_norm": 0.5445185303688049,
      "learning_rate": 0.002,
      "loss": 7.688406467437744,
      "step": 90
    },
    {
      "epoch": 0.01265854715652798,
      "grad_norm": 0.6012760400772095,
      "learning_rate": 0.002,
      "loss": 7.6673078536987305,
      "step": 92
    },
    {
      "epoch": 0.01293373296427859,
      "grad_norm": 0.5545694231987,
      "learning_rate": 0.002,
      "loss": 7.617863655090332,
      "step": 94
    },
    {
      "epoch": 0.013208918772029198,
      "grad_norm": 0.39871472120285034,
      "learning_rate": 0.002,
      "loss": 7.67018985748291,
      "step": 96
    },
    {
      "epoch": 0.013484104579779806,
      "grad_norm": 0.9112051129341125,
      "learning_rate": 0.002,
      "loss": 7.5599517822265625,
      "step": 98
    },
    {
      "epoch": 0.013759290387530414,
      "grad_norm": 0.8745017647743225,
      "learning_rate": 0.002,
      "loss": 7.542838096618652,
      "step": 100
    },
    {
      "epoch": 0.014034476195281023,
      "grad_norm": 0.5904999375343323,
      "learning_rate": 0.002,
      "loss": 7.543461799621582,
      "step": 102
    },
    {
      "epoch": 0.01430966200303163,
      "grad_norm": 0.5321375131607056,
      "learning_rate": 0.002,
      "loss": 7.462248802185059,
      "step": 104
    },
    {
      "epoch": 0.014584847810782238,
      "grad_norm": 0.41776517033576965,
      "learning_rate": 0.002,
      "loss": 7.4792985916137695,
      "step": 106
    },
    {
      "epoch": 0.014860033618532846,
      "grad_norm": 0.41203972697257996,
      "learning_rate": 0.002,
      "loss": 7.402154445648193,
      "step": 108
    },
    {
      "epoch": 0.015135219426283455,
      "grad_norm": 0.3548356592655182,
      "learning_rate": 0.002,
      "loss": 7.372166633605957,
      "step": 110
    },
    {
      "epoch": 0.015410405234034063,
      "grad_norm": 0.38654717803001404,
      "learning_rate": 0.002,
      "loss": 7.3752946853637695,
      "step": 112
    },
    {
      "epoch": 0.01568559104178467,
      "grad_norm": 0.4159736931324005,
      "learning_rate": 0.002,
      "loss": 7.357211112976074,
      "step": 114
    },
    {
      "epoch": 0.01596077684953528,
      "grad_norm": 0.3689005970954895,
      "learning_rate": 0.002,
      "loss": 7.3626933097839355,
      "step": 116
    },
    {
      "epoch": 0.01623596265728589,
      "grad_norm": 0.31604018807411194,
      "learning_rate": 0.002,
      "loss": 7.312067031860352,
      "step": 118
    },
    {
      "epoch": 0.016511148465036495,
      "grad_norm": 0.2781250774860382,
      "learning_rate": 0.002,
      "loss": 7.367823123931885,
      "step": 120
    },
    {
      "epoch": 0.016786334272787105,
      "grad_norm": 0.22659680247306824,
      "learning_rate": 0.002,
      "loss": 7.287016868591309,
      "step": 122
    },
    {
      "epoch": 0.017061520080537712,
      "grad_norm": 0.29917341470718384,
      "learning_rate": 0.002,
      "loss": 7.282745361328125,
      "step": 124
    },
    {
      "epoch": 0.017336705888288322,
      "grad_norm": 0.5459050536155701,
      "learning_rate": 0.002,
      "loss": 7.3504438400268555,
      "step": 126
    },
    {
      "epoch": 0.01761189169603893,
      "grad_norm": 0.34762924909591675,
      "learning_rate": 0.002,
      "loss": 7.325478553771973,
      "step": 128
    },
    {
      "epoch": 0.01788707750378954,
      "grad_norm": 0.3289381265640259,
      "learning_rate": 0.002,
      "loss": 7.252686500549316,
      "step": 130
    },
    {
      "epoch": 0.018162263311540146,
      "grad_norm": 0.5142496228218079,
      "learning_rate": 0.002,
      "loss": 7.180061340332031,
      "step": 132
    },
    {
      "epoch": 0.018437449119290756,
      "grad_norm": 0.43205526471138,
      "learning_rate": 0.002,
      "loss": 7.145871162414551,
      "step": 134
    },
    {
      "epoch": 0.018712634927041363,
      "grad_norm": 0.5921562910079956,
      "learning_rate": 0.002,
      "loss": 7.208243370056152,
      "step": 136
    },
    {
      "epoch": 0.01898782073479197,
      "grad_norm": 0.3849842846393585,
      "learning_rate": 0.002,
      "loss": 7.192992687225342,
      "step": 138
    },
    {
      "epoch": 0.01926300654254258,
      "grad_norm": 0.379642516374588,
      "learning_rate": 0.002,
      "loss": 7.159093856811523,
      "step": 140
    },
    {
      "epoch": 0.019538192350293186,
      "grad_norm": 0.4007287919521332,
      "learning_rate": 0.002,
      "loss": 7.158816814422607,
      "step": 142
    },
    {
      "epoch": 0.019813378158043796,
      "grad_norm": 0.3065553605556488,
      "learning_rate": 0.002,
      "loss": 7.162346839904785,
      "step": 144
    },
    {
      "epoch": 0.020088563965794403,
      "grad_norm": 0.19435709714889526,
      "learning_rate": 0.002,
      "loss": 7.152171611785889,
      "step": 146
    },
    {
      "epoch": 0.020363749773545013,
      "grad_norm": 0.4215281307697296,
      "learning_rate": 0.002,
      "loss": 7.094367980957031,
      "step": 148
    },
    {
      "epoch": 0.02063893558129562,
      "grad_norm": 0.3014766573905945,
      "learning_rate": 0.002,
      "loss": 7.069222450256348,
      "step": 150
    },
    {
      "epoch": 0.02091412138904623,
      "grad_norm": 0.27510693669319153,
      "learning_rate": 0.002,
      "loss": 7.082720756530762,
      "step": 152
    },
    {
      "epoch": 0.021189307196796837,
      "grad_norm": 0.22899171710014343,
      "learning_rate": 0.002,
      "loss": 7.036725997924805,
      "step": 154
    },
    {
      "epoch": 0.021464493004547447,
      "grad_norm": 0.23482806980609894,
      "learning_rate": 0.002,
      "loss": 7.076622009277344,
      "step": 156
    },
    {
      "epoch": 0.021739678812298054,
      "grad_norm": 0.24542653560638428,
      "learning_rate": 0.002,
      "loss": 7.052186965942383,
      "step": 158
    },
    {
      "epoch": 0.02201486462004866,
      "grad_norm": 0.20042400062084198,
      "learning_rate": 0.002,
      "loss": 6.988194465637207,
      "step": 160
    },
    {
      "epoch": 0.02229005042779927,
      "grad_norm": 0.3489605188369751,
      "learning_rate": 0.002,
      "loss": 6.957248687744141,
      "step": 162
    },
    {
      "epoch": 0.022565236235549877,
      "grad_norm": 0.22401060163974762,
      "learning_rate": 0.002,
      "loss": 6.9442853927612305,
      "step": 164
    },
    {
      "epoch": 0.022840422043300487,
      "grad_norm": 0.18925704061985016,
      "learning_rate": 0.002,
      "loss": 6.9237141609191895,
      "step": 166
    },
    {
      "epoch": 0.023115607851051094,
      "grad_norm": 0.34174031019210815,
      "learning_rate": 0.002,
      "loss": 6.859572410583496,
      "step": 168
    },
    {
      "epoch": 0.023390793658801704,
      "grad_norm": 0.3179168999195099,
      "learning_rate": 0.002,
      "loss": 6.85260534286499,
      "step": 170
    },
    {
      "epoch": 0.02366597946655231,
      "grad_norm": 0.22768643498420715,
      "learning_rate": 0.002,
      "loss": 6.908085346221924,
      "step": 172
    },
    {
      "epoch": 0.02394116527430292,
      "grad_norm": 0.2918003499507904,
      "learning_rate": 0.002,
      "loss": 6.911751747131348,
      "step": 174
    },
    {
      "epoch": 0.024216351082053528,
      "grad_norm": 0.2749776244163513,
      "learning_rate": 0.002,
      "loss": 6.849135398864746,
      "step": 176
    },
    {
      "epoch": 0.024491536889804138,
      "grad_norm": 0.319549560546875,
      "learning_rate": 0.002,
      "loss": 6.841396808624268,
      "step": 178
    },
    {
      "epoch": 0.024766722697554745,
      "grad_norm": 0.27626776695251465,
      "learning_rate": 0.002,
      "loss": 6.808116436004639,
      "step": 180
    },
    {
      "epoch": 0.025041908505305355,
      "grad_norm": 0.24615061283111572,
      "learning_rate": 0.002,
      "loss": 6.801154136657715,
      "step": 182
    },
    {
      "epoch": 0.02531709431305596,
      "grad_norm": 0.24768510460853577,
      "learning_rate": 0.002,
      "loss": 6.772426605224609,
      "step": 184
    },
    {
      "epoch": 0.025592280120806568,
      "grad_norm": 0.2629677355289459,
      "learning_rate": 0.002,
      "loss": 6.733229637145996,
      "step": 186
    },
    {
      "epoch": 0.02586746592855718,
      "grad_norm": 0.3167332410812378,
      "learning_rate": 0.002,
      "loss": 6.702438831329346,
      "step": 188
    },
    {
      "epoch": 0.026142651736307785,
      "grad_norm": 0.31601452827453613,
      "learning_rate": 0.002,
      "loss": 6.6851806640625,
      "step": 190
    },
    {
      "epoch": 0.026417837544058395,
      "grad_norm": 0.36339420080184937,
      "learning_rate": 0.002,
      "loss": 6.754533767700195,
      "step": 192
    },
    {
      "epoch": 0.026693023351809002,
      "grad_norm": 0.20179785788059235,
      "learning_rate": 0.002,
      "loss": 6.719021797180176,
      "step": 194
    },
    {
      "epoch": 0.026968209159559612,
      "grad_norm": 0.3848166763782501,
      "learning_rate": 0.002,
      "loss": 6.674118995666504,
      "step": 196
    },
    {
      "epoch": 0.02724339496731022,
      "grad_norm": 0.23161272704601288,
      "learning_rate": 0.002,
      "loss": 6.663918495178223,
      "step": 198
    },
    {
      "epoch": 0.02751858077506083,
      "grad_norm": 0.23778744041919708,
      "learning_rate": 0.002,
      "loss": 6.692194938659668,
      "step": 200
    },
    {
      "epoch": 0.027793766582811436,
      "grad_norm": 0.18710927665233612,
      "learning_rate": 0.002,
      "loss": 6.66090202331543,
      "step": 202
    },
    {
      "epoch": 0.028068952390562046,
      "grad_norm": 0.20118694007396698,
      "learning_rate": 0.002,
      "loss": 6.648110389709473,
      "step": 204
    },
    {
      "epoch": 0.028344138198312652,
      "grad_norm": 0.2604389786720276,
      "learning_rate": 0.002,
      "loss": 6.637610912322998,
      "step": 206
    },
    {
      "epoch": 0.02861932400606326,
      "grad_norm": 0.5853762030601501,
      "learning_rate": 0.002,
      "loss": 6.637144565582275,
      "step": 208
    },
    {
      "epoch": 0.02889450981381387,
      "grad_norm": 0.4096166789531708,
      "learning_rate": 0.002,
      "loss": 6.654441833496094,
      "step": 210
    },
    {
      "epoch": 0.029169695621564476,
      "grad_norm": 0.33405011892318726,
      "learning_rate": 0.002,
      "loss": 6.547055244445801,
      "step": 212
    },
    {
      "epoch": 0.029444881429315086,
      "grad_norm": 0.32981884479522705,
      "learning_rate": 0.002,
      "loss": 6.587018966674805,
      "step": 214
    },
    {
      "epoch": 0.029720067237065693,
      "grad_norm": 0.25928136706352234,
      "learning_rate": 0.002,
      "loss": 6.551738739013672,
      "step": 216
    },
    {
      "epoch": 0.029995253044816303,
      "grad_norm": 0.2555151581764221,
      "learning_rate": 0.002,
      "loss": 6.569494724273682,
      "step": 218
    },
    {
      "epoch": 0.03027043885256691,
      "grad_norm": 0.18236006796360016,
      "learning_rate": 0.002,
      "loss": 6.536982536315918,
      "step": 220
    },
    {
      "epoch": 0.03054562466031752,
      "grad_norm": 0.1933513730764389,
      "learning_rate": 0.002,
      "loss": 6.454246997833252,
      "step": 222
    },
    {
      "epoch": 0.030820810468068126,
      "grad_norm": 0.1799364686012268,
      "learning_rate": 0.002,
      "loss": 6.477840423583984,
      "step": 224
    },
    {
      "epoch": 0.031095996275818737,
      "grad_norm": 0.1453200727701187,
      "learning_rate": 0.002,
      "loss": 6.435481071472168,
      "step": 226
    },
    {
      "epoch": 0.03137118208356934,
      "grad_norm": 0.23650521039962769,
      "learning_rate": 0.002,
      "loss": 6.474623680114746,
      "step": 228
    },
    {
      "epoch": 0.03164636789131995,
      "grad_norm": 0.3354651629924774,
      "learning_rate": 0.002,
      "loss": 6.470672130584717,
      "step": 230
    },
    {
      "epoch": 0.03192155369907056,
      "grad_norm": 0.4492896497249603,
      "learning_rate": 0.002,
      "loss": 6.507452964782715,
      "step": 232
    },
    {
      "epoch": 0.03219673950682117,
      "grad_norm": 0.43507370352745056,
      "learning_rate": 0.002,
      "loss": 6.386468887329102,
      "step": 234
    },
    {
      "epoch": 0.03247192531457178,
      "grad_norm": 0.2701570987701416,
      "learning_rate": 0.002,
      "loss": 6.4474778175354,
      "step": 236
    },
    {
      "epoch": 0.032747111122322384,
      "grad_norm": 0.43229812383651733,
      "learning_rate": 0.002,
      "loss": 6.435725212097168,
      "step": 238
    },
    {
      "epoch": 0.03302229693007299,
      "grad_norm": 0.40472277998924255,
      "learning_rate": 0.002,
      "loss": 6.4167938232421875,
      "step": 240
    },
    {
      "epoch": 0.033297482737823604,
      "grad_norm": 0.2184945046901703,
      "learning_rate": 0.002,
      "loss": 6.376438140869141,
      "step": 242
    },
    {
      "epoch": 0.03357266854557421,
      "grad_norm": 0.3910417854785919,
      "learning_rate": 0.002,
      "loss": 6.380894660949707,
      "step": 244
    },
    {
      "epoch": 0.03384785435332482,
      "grad_norm": 0.3155927062034607,
      "learning_rate": 0.002,
      "loss": 6.3454132080078125,
      "step": 246
    },
    {
      "epoch": 0.034123040161075424,
      "grad_norm": 0.2170950025320053,
      "learning_rate": 0.002,
      "loss": 6.360450267791748,
      "step": 248
    },
    {
      "epoch": 0.03439822596882604,
      "grad_norm": 0.18148884177207947,
      "learning_rate": 0.002,
      "loss": 6.339743614196777,
      "step": 250
    },
    {
      "epoch": 0.034673411776576644,
      "grad_norm": 0.18744991719722748,
      "learning_rate": 0.002,
      "loss": 6.276296615600586,
      "step": 252
    },
    {
      "epoch": 0.03494859758432725,
      "grad_norm": 0.23324045538902283,
      "learning_rate": 0.002,
      "loss": 6.269101619720459,
      "step": 254
    },
    {
      "epoch": 0.03522378339207786,
      "grad_norm": 0.213908389210701,
      "learning_rate": 0.002,
      "loss": 6.284084796905518,
      "step": 256
    },
    {
      "epoch": 0.035498969199828465,
      "grad_norm": 0.23218698799610138,
      "learning_rate": 0.002,
      "loss": 6.239173889160156,
      "step": 258
    },
    {
      "epoch": 0.03577415500757908,
      "grad_norm": 0.21279890835285187,
      "learning_rate": 0.002,
      "loss": 6.283649444580078,
      "step": 260
    },
    {
      "epoch": 0.036049340815329685,
      "grad_norm": 0.17391885817050934,
      "learning_rate": 0.002,
      "loss": 6.278305530548096,
      "step": 262
    },
    {
      "epoch": 0.03632452662308029,
      "grad_norm": 0.14646387100219727,
      "learning_rate": 0.002,
      "loss": 6.208422660827637,
      "step": 264
    },
    {
      "epoch": 0.0365997124308309,
      "grad_norm": 0.19890061020851135,
      "learning_rate": 0.002,
      "loss": 6.271259307861328,
      "step": 266
    },
    {
      "epoch": 0.03687489823858151,
      "grad_norm": 0.36767998337745667,
      "learning_rate": 0.002,
      "loss": 6.245121955871582,
      "step": 268
    },
    {
      "epoch": 0.03715008404633212,
      "grad_norm": 0.4364091455936432,
      "learning_rate": 0.002,
      "loss": 6.309082508087158,
      "step": 270
    },
    {
      "epoch": 0.037425269854082725,
      "grad_norm": 0.23240774869918823,
      "learning_rate": 0.002,
      "loss": 6.166597366333008,
      "step": 272
    },
    {
      "epoch": 0.03770045566183333,
      "grad_norm": 0.3186095952987671,
      "learning_rate": 0.002,
      "loss": 6.240274429321289,
      "step": 274
    },
    {
      "epoch": 0.03797564146958394,
      "grad_norm": 0.25660595297813416,
      "learning_rate": 0.002,
      "loss": 6.209404945373535,
      "step": 276
    },
    {
      "epoch": 0.03825082727733455,
      "grad_norm": 0.23214584589004517,
      "learning_rate": 0.002,
      "loss": 6.160274982452393,
      "step": 278
    },
    {
      "epoch": 0.03852601308508516,
      "grad_norm": 0.23841989040374756,
      "learning_rate": 0.002,
      "loss": 6.1383161544799805,
      "step": 280
    },
    {
      "epoch": 0.038801198892835766,
      "grad_norm": 0.19925029575824738,
      "learning_rate": 0.002,
      "loss": 6.174008369445801,
      "step": 282
    },
    {
      "epoch": 0.03907638470058637,
      "grad_norm": 0.28103432059288025,
      "learning_rate": 0.002,
      "loss": 6.148692607879639,
      "step": 284
    },
    {
      "epoch": 0.039351570508336986,
      "grad_norm": 0.18915946781635284,
      "learning_rate": 0.002,
      "loss": 6.1004180908203125,
      "step": 286
    },
    {
      "epoch": 0.03962675631608759,
      "grad_norm": 0.17474333941936493,
      "learning_rate": 0.002,
      "loss": 6.122094631195068,
      "step": 288
    },
    {
      "epoch": 0.0399019421238382,
      "grad_norm": 0.1856822669506073,
      "learning_rate": 0.002,
      "loss": 6.086495876312256,
      "step": 290
    },
    {
      "epoch": 0.040177127931588806,
      "grad_norm": 0.3012590706348419,
      "learning_rate": 0.002,
      "loss": 6.093333721160889,
      "step": 292
    },
    {
      "epoch": 0.04045231373933942,
      "grad_norm": 0.41956236958503723,
      "learning_rate": 0.002,
      "loss": 6.097911834716797,
      "step": 294
    },
    {
      "epoch": 0.040727499547090026,
      "grad_norm": 0.2689352035522461,
      "learning_rate": 0.002,
      "loss": 6.094822883605957,
      "step": 296
    },
    {
      "epoch": 0.04100268535484063,
      "grad_norm": 0.24790361523628235,
      "learning_rate": 0.002,
      "loss": 6.076343059539795,
      "step": 298
    },
    {
      "epoch": 0.04127787116259124,
      "grad_norm": 0.19465188682079315,
      "learning_rate": 0.002,
      "loss": 6.077427864074707,
      "step": 300
    },
    {
      "epoch": 0.041553056970341846,
      "grad_norm": 0.1721596121788025,
      "learning_rate": 0.002,
      "loss": 5.956333637237549,
      "step": 302
    },
    {
      "epoch": 0.04182824277809246,
      "grad_norm": 0.25611165165901184,
      "learning_rate": 0.002,
      "loss": 5.982329845428467,
      "step": 304
    },
    {
      "epoch": 0.04210342858584307,
      "grad_norm": 0.2046833336353302,
      "learning_rate": 0.002,
      "loss": 6.020071029663086,
      "step": 306
    },
    {
      "epoch": 0.042378614393593673,
      "grad_norm": 0.2434958964586258,
      "learning_rate": 0.002,
      "loss": 6.010205268859863,
      "step": 308
    },
    {
      "epoch": 0.04265380020134428,
      "grad_norm": 0.3358174264431,
      "learning_rate": 0.002,
      "loss": 5.963714599609375,
      "step": 310
    },
    {
      "epoch": 0.042928986009094894,
      "grad_norm": 0.4181540906429291,
      "learning_rate": 0.002,
      "loss": 5.9268012046813965,
      "step": 312
    },
    {
      "epoch": 0.0432041718168455,
      "grad_norm": 0.24285808205604553,
      "learning_rate": 0.002,
      "loss": 5.986179351806641,
      "step": 314
    },
    {
      "epoch": 0.04347935762459611,
      "grad_norm": 0.3972429931163788,
      "learning_rate": 0.002,
      "loss": 5.9228105545043945,
      "step": 316
    },
    {
      "epoch": 0.043754543432346714,
      "grad_norm": 0.18683181703090668,
      "learning_rate": 0.002,
      "loss": 5.941620349884033,
      "step": 318
    },
    {
      "epoch": 0.04402972924009732,
      "grad_norm": 0.2502729296684265,
      "learning_rate": 0.002,
      "loss": 5.933855056762695,
      "step": 320
    },
    {
      "epoch": 0.044304915047847934,
      "grad_norm": 0.49798017740249634,
      "learning_rate": 0.002,
      "loss": 5.934805870056152,
      "step": 322
    },
    {
      "epoch": 0.04458010085559854,
      "grad_norm": 0.34854474663734436,
      "learning_rate": 0.002,
      "loss": 5.962124824523926,
      "step": 324
    },
    {
      "epoch": 0.04485528666334915,
      "grad_norm": 0.2614416480064392,
      "learning_rate": 0.002,
      "loss": 5.953027725219727,
      "step": 326
    },
    {
      "epoch": 0.045130472471099754,
      "grad_norm": 0.23937691748142242,
      "learning_rate": 0.002,
      "loss": 5.890893936157227,
      "step": 328
    },
    {
      "epoch": 0.04540565827885037,
      "grad_norm": 0.236050084233284,
      "learning_rate": 0.002,
      "loss": 5.912225723266602,
      "step": 330
    },
    {
      "epoch": 0.045680844086600975,
      "grad_norm": 0.21355098485946655,
      "learning_rate": 0.002,
      "loss": 5.867430686950684,
      "step": 332
    },
    {
      "epoch": 0.04595602989435158,
      "grad_norm": 0.2504339814186096,
      "learning_rate": 0.002,
      "loss": 5.865871906280518,
      "step": 334
    },
    {
      "epoch": 0.04623121570210219,
      "grad_norm": 0.14930841326713562,
      "learning_rate": 0.002,
      "loss": 5.826433181762695,
      "step": 336
    },
    {
      "epoch": 0.0465064015098528,
      "grad_norm": 0.18486200273036957,
      "learning_rate": 0.002,
      "loss": 5.836857795715332,
      "step": 338
    },
    {
      "epoch": 0.04678158731760341,
      "grad_norm": 0.25455281138420105,
      "learning_rate": 0.002,
      "loss": 5.803670406341553,
      "step": 340
    },
    {
      "epoch": 0.047056773125354015,
      "grad_norm": 0.2531868517398834,
      "learning_rate": 0.002,
      "loss": 5.895256996154785,
      "step": 342
    },
    {
      "epoch": 0.04733195893310462,
      "grad_norm": 0.27131620049476624,
      "learning_rate": 0.002,
      "loss": 5.830260276794434,
      "step": 344
    },
    {
      "epoch": 0.04760714474085523,
      "grad_norm": 0.2448132038116455,
      "learning_rate": 0.002,
      "loss": 5.798295021057129,
      "step": 346
    },
    {
      "epoch": 0.04788233054860584,
      "grad_norm": 0.3160443902015686,
      "learning_rate": 0.002,
      "loss": 5.807307243347168,
      "step": 348
    },
    {
      "epoch": 0.04815751635635645,
      "grad_norm": 0.22071008384227753,
      "learning_rate": 0.002,
      "loss": 5.783094882965088,
      "step": 350
    },
    {
      "epoch": 0.048432702164107055,
      "grad_norm": 0.24213366210460663,
      "learning_rate": 0.002,
      "loss": 5.726943016052246,
      "step": 352
    },
    {
      "epoch": 0.04870788797185766,
      "grad_norm": 0.3100344240665436,
      "learning_rate": 0.002,
      "loss": 5.731729507446289,
      "step": 354
    },
    {
      "epoch": 0.048983073779608276,
      "grad_norm": 0.3896303176879883,
      "learning_rate": 0.002,
      "loss": 5.8033881187438965,
      "step": 356
    },
    {
      "epoch": 0.04925825958735888,
      "grad_norm": 0.24003364145755768,
      "learning_rate": 0.002,
      "loss": 5.750821590423584,
      "step": 358
    },
    {
      "epoch": 0.04953344539510949,
      "grad_norm": 0.19001154601573944,
      "learning_rate": 0.002,
      "loss": 5.744946002960205,
      "step": 360
    },
    {
      "epoch": 0.049808631202860096,
      "grad_norm": 0.1581803411245346,
      "learning_rate": 0.002,
      "loss": 5.752904415130615,
      "step": 362
    },
    {
      "epoch": 0.05008381701061071,
      "grad_norm": 0.2261851578950882,
      "learning_rate": 0.002,
      "loss": 5.702903747558594,
      "step": 364
    },
    {
      "epoch": 0.050359002818361316,
      "grad_norm": 0.365060955286026,
      "learning_rate": 0.002,
      "loss": 5.710434436798096,
      "step": 366
    },
    {
      "epoch": 0.05063418862611192,
      "grad_norm": 0.3399103879928589,
      "learning_rate": 0.002,
      "loss": 5.747560024261475,
      "step": 368
    },
    {
      "epoch": 0.05090937443386253,
      "grad_norm": 0.3257260322570801,
      "learning_rate": 0.002,
      "loss": 5.677846431732178,
      "step": 370
    },
    {
      "epoch": 0.051184560241613136,
      "grad_norm": 0.32304447889328003,
      "learning_rate": 0.002,
      "loss": 5.656373023986816,
      "step": 372
    },
    {
      "epoch": 0.05145974604936375,
      "grad_norm": 0.2766451835632324,
      "learning_rate": 0.002,
      "loss": 5.643589496612549,
      "step": 374
    },
    {
      "epoch": 0.05173493185711436,
      "grad_norm": 0.20961201190948486,
      "learning_rate": 0.002,
      "loss": 5.673385143280029,
      "step": 376
    },
    {
      "epoch": 0.05201011766486496,
      "grad_norm": 0.20695270597934723,
      "learning_rate": 0.002,
      "loss": 5.595235347747803,
      "step": 378
    },
    {
      "epoch": 0.05228530347261557,
      "grad_norm": 0.1818195879459381,
      "learning_rate": 0.002,
      "loss": 5.625968933105469,
      "step": 380
    },
    {
      "epoch": 0.052560489280366184,
      "grad_norm": 0.24791789054870605,
      "learning_rate": 0.002,
      "loss": 5.627171516418457,
      "step": 382
    },
    {
      "epoch": 0.05283567508811679,
      "grad_norm": 0.24782583117485046,
      "learning_rate": 0.002,
      "loss": 5.633144855499268,
      "step": 384
    },
    {
      "epoch": 0.0531108608958674,
      "grad_norm": 0.25601255893707275,
      "learning_rate": 0.002,
      "loss": 5.605512619018555,
      "step": 386
    },
    {
      "epoch": 0.053386046703618004,
      "grad_norm": 0.3580726683139801,
      "learning_rate": 0.002,
      "loss": 5.603791236877441,
      "step": 388
    },
    {
      "epoch": 0.05366123251136861,
      "grad_norm": 0.34141427278518677,
      "learning_rate": 0.002,
      "loss": 5.60898494720459,
      "step": 390
    },
    {
      "epoch": 0.053936418319119224,
      "grad_norm": 0.3045906126499176,
      "learning_rate": 0.002,
      "loss": 5.6319475173950195,
      "step": 392
    },
    {
      "epoch": 0.05421160412686983,
      "grad_norm": 0.21149297058582306,
      "learning_rate": 0.002,
      "loss": 5.707817077636719,
      "step": 394
    },
    {
      "epoch": 0.05448678993462044,
      "grad_norm": 0.2465360462665558,
      "learning_rate": 0.002,
      "loss": 5.600747108459473,
      "step": 396
    },
    {
      "epoch": 0.054761975742371044,
      "grad_norm": 0.17950813472270966,
      "learning_rate": 0.002,
      "loss": 5.557934761047363,
      "step": 398
    },
    {
      "epoch": 0.05503716155012166,
      "grad_norm": 0.15238267183303833,
      "learning_rate": 0.002,
      "loss": 5.55340576171875,
      "step": 400
    },
    {
      "epoch": 0.055312347357872264,
      "grad_norm": 0.15906371176242828,
      "learning_rate": 0.002,
      "loss": 5.514875411987305,
      "step": 402
    },
    {
      "epoch": 0.05558753316562287,
      "grad_norm": 0.5140070915222168,
      "learning_rate": 0.002,
      "loss": 5.533358573913574,
      "step": 404
    },
    {
      "epoch": 0.05586271897337348,
      "grad_norm": 0.426023006439209,
      "learning_rate": 0.002,
      "loss": 5.5861687660217285,
      "step": 406
    },
    {
      "epoch": 0.05613790478112409,
      "grad_norm": 0.5879791378974915,
      "learning_rate": 0.002,
      "loss": 5.554373741149902,
      "step": 408
    },
    {
      "epoch": 0.0564130905888747,
      "grad_norm": 0.6436289548873901,
      "learning_rate": 0.002,
      "loss": 5.55023193359375,
      "step": 410
    },
    {
      "epoch": 0.056688276396625305,
      "grad_norm": 0.2639814615249634,
      "learning_rate": 0.002,
      "loss": 5.530845642089844,
      "step": 412
    },
    {
      "epoch": 0.05696346220437591,
      "grad_norm": 0.19631877541542053,
      "learning_rate": 0.002,
      "loss": 5.5167646408081055,
      "step": 414
    },
    {
      "epoch": 0.05723864801212652,
      "grad_norm": 0.21685880422592163,
      "learning_rate": 0.002,
      "loss": 5.45224666595459,
      "step": 416
    },
    {
      "epoch": 0.05751383381987713,
      "grad_norm": 0.16595172882080078,
      "learning_rate": 0.002,
      "loss": 5.505030632019043,
      "step": 418
    },
    {
      "epoch": 0.05778901962762774,
      "grad_norm": 0.1743602603673935,
      "learning_rate": 0.002,
      "loss": 5.458126544952393,
      "step": 420
    },
    {
      "epoch": 0.058064205435378345,
      "grad_norm": 0.14870530366897583,
      "learning_rate": 0.002,
      "loss": 5.469132423400879,
      "step": 422
    },
    {
      "epoch": 0.05833939124312895,
      "grad_norm": 0.188246488571167,
      "learning_rate": 0.002,
      "loss": 5.494062423706055,
      "step": 424
    },
    {
      "epoch": 0.058614577050879565,
      "grad_norm": 0.28493183851242065,
      "learning_rate": 0.002,
      "loss": 5.4519805908203125,
      "step": 426
    },
    {
      "epoch": 0.05888976285863017,
      "grad_norm": 0.2820991277694702,
      "learning_rate": 0.002,
      "loss": 5.484132766723633,
      "step": 428
    },
    {
      "epoch": 0.05916494866638078,
      "grad_norm": 0.4048371911048889,
      "learning_rate": 0.002,
      "loss": 5.468401908874512,
      "step": 430
    },
    {
      "epoch": 0.059440134474131386,
      "grad_norm": 0.15520060062408447,
      "learning_rate": 0.002,
      "loss": 5.400453567504883,
      "step": 432
    },
    {
      "epoch": 0.059715320281882,
      "grad_norm": 0.2782343327999115,
      "learning_rate": 0.002,
      "loss": 5.436036109924316,
      "step": 434
    },
    {
      "epoch": 0.059990506089632606,
      "grad_norm": 0.1888812929391861,
      "learning_rate": 0.002,
      "loss": 5.38045597076416,
      "step": 436
    },
    {
      "epoch": 0.06026569189738321,
      "grad_norm": 0.13723966479301453,
      "learning_rate": 0.002,
      "loss": 5.397899627685547,
      "step": 438
    },
    {
      "epoch": 0.06054087770513382,
      "grad_norm": 0.18245477974414825,
      "learning_rate": 0.002,
      "loss": 5.355703353881836,
      "step": 440
    },
    {
      "epoch": 0.060816063512884426,
      "grad_norm": 0.4200305640697479,
      "learning_rate": 0.002,
      "loss": 5.418264389038086,
      "step": 442
    },
    {
      "epoch": 0.06109124932063504,
      "grad_norm": 0.2881113588809967,
      "learning_rate": 0.002,
      "loss": 5.410123348236084,
      "step": 444
    },
    {
      "epoch": 0.061366435128385646,
      "grad_norm": 0.19265086948871613,
      "learning_rate": 0.002,
      "loss": 5.422796249389648,
      "step": 446
    },
    {
      "epoch": 0.06164162093613625,
      "grad_norm": 0.1680508255958557,
      "learning_rate": 0.002,
      "loss": 5.3624162673950195,
      "step": 448
    },
    {
      "epoch": 0.06191680674388686,
      "grad_norm": 0.17103669047355652,
      "learning_rate": 0.002,
      "loss": 5.401288986206055,
      "step": 450
    },
    {
      "epoch": 0.06219199255163747,
      "grad_norm": 0.16605453193187714,
      "learning_rate": 0.002,
      "loss": 5.3648762702941895,
      "step": 452
    },
    {
      "epoch": 0.06246717835938808,
      "grad_norm": 0.17053014039993286,
      "learning_rate": 0.002,
      "loss": 5.376862525939941,
      "step": 454
    },
    {
      "epoch": 0.06274236416713869,
      "grad_norm": 0.2742466330528259,
      "learning_rate": 0.002,
      "loss": 5.2997589111328125,
      "step": 456
    },
    {
      "epoch": 0.0630175499748893,
      "grad_norm": 0.2918921411037445,
      "learning_rate": 0.002,
      "loss": 5.331727027893066,
      "step": 458
    },
    {
      "epoch": 0.0632927357826399,
      "grad_norm": 0.26032403111457825,
      "learning_rate": 0.002,
      "loss": 5.317394256591797,
      "step": 460
    },
    {
      "epoch": 0.06356792159039051,
      "grad_norm": 0.20663008093833923,
      "learning_rate": 0.002,
      "loss": 5.370279312133789,
      "step": 462
    },
    {
      "epoch": 0.06384310739814111,
      "grad_norm": 0.48780983686447144,
      "learning_rate": 0.002,
      "loss": 5.362845420837402,
      "step": 464
    },
    {
      "epoch": 0.06411829320589173,
      "grad_norm": 0.22591857612133026,
      "learning_rate": 0.002,
      "loss": 5.3127546310424805,
      "step": 466
    },
    {
      "epoch": 0.06439347901364234,
      "grad_norm": 0.2500840127468109,
      "learning_rate": 0.002,
      "loss": 5.30235481262207,
      "step": 468
    },
    {
      "epoch": 0.06466866482139294,
      "grad_norm": 0.2511637210845947,
      "learning_rate": 0.002,
      "loss": 5.274158954620361,
      "step": 470
    },
    {
      "epoch": 0.06494385062914355,
      "grad_norm": 0.21295130252838135,
      "learning_rate": 0.002,
      "loss": 5.267327785491943,
      "step": 472
    },
    {
      "epoch": 0.06521903643689417,
      "grad_norm": 0.18395768105983734,
      "learning_rate": 0.002,
      "loss": 5.307414531707764,
      "step": 474
    },
    {
      "epoch": 0.06549422224464477,
      "grad_norm": 0.20234695076942444,
      "learning_rate": 0.002,
      "loss": 5.240487098693848,
      "step": 476
    },
    {
      "epoch": 0.06576940805239538,
      "grad_norm": 0.1783929467201233,
      "learning_rate": 0.002,
      "loss": 5.216714859008789,
      "step": 478
    },
    {
      "epoch": 0.06604459386014598,
      "grad_norm": 0.2747371196746826,
      "learning_rate": 0.002,
      "loss": 5.250484466552734,
      "step": 480
    },
    {
      "epoch": 0.0663197796678966,
      "grad_norm": 0.2754431664943695,
      "learning_rate": 0.002,
      "loss": 5.2366533279418945,
      "step": 482
    },
    {
      "epoch": 0.06659496547564721,
      "grad_norm": 0.20493867993354797,
      "learning_rate": 0.002,
      "loss": 5.269064426422119,
      "step": 484
    },
    {
      "epoch": 0.06687015128339781,
      "grad_norm": 0.19416649639606476,
      "learning_rate": 0.002,
      "loss": 5.226369857788086,
      "step": 486
    },
    {
      "epoch": 0.06714533709114842,
      "grad_norm": 0.20199145376682281,
      "learning_rate": 0.002,
      "loss": 5.238809108734131,
      "step": 488
    },
    {
      "epoch": 0.06742052289889902,
      "grad_norm": 0.28233054280281067,
      "learning_rate": 0.002,
      "loss": 5.165689468383789,
      "step": 490
    },
    {
      "epoch": 0.06769570870664963,
      "grad_norm": 0.19025918841362,
      "learning_rate": 0.002,
      "loss": 5.223665237426758,
      "step": 492
    },
    {
      "epoch": 0.06797089451440025,
      "grad_norm": 0.3509269058704376,
      "learning_rate": 0.002,
      "loss": 5.2210612297058105,
      "step": 494
    },
    {
      "epoch": 0.06824608032215085,
      "grad_norm": 0.36469268798828125,
      "learning_rate": 0.002,
      "loss": 5.178593635559082,
      "step": 496
    },
    {
      "epoch": 0.06852126612990146,
      "grad_norm": 0.3895169496536255,
      "learning_rate": 0.002,
      "loss": 5.101150989532471,
      "step": 498
    },
    {
      "epoch": 0.06879645193765208,
      "grad_norm": 0.18548665940761566,
      "learning_rate": 0.002,
      "loss": 5.160384178161621,
      "step": 500
    },
    {
      "epoch": 0.06907163774540268,
      "grad_norm": 0.2315627932548523,
      "learning_rate": 0.002,
      "loss": 5.1751909255981445,
      "step": 502
    },
    {
      "epoch": 0.06934682355315329,
      "grad_norm": 0.1968149095773697,
      "learning_rate": 0.002,
      "loss": 5.108625411987305,
      "step": 504
    },
    {
      "epoch": 0.06962200936090389,
      "grad_norm": 0.3903138041496277,
      "learning_rate": 0.002,
      "loss": 5.1992316246032715,
      "step": 506
    },
    {
      "epoch": 0.0698971951686545,
      "grad_norm": 0.22219625115394592,
      "learning_rate": 0.002,
      "loss": 5.190418243408203,
      "step": 508
    },
    {
      "epoch": 0.07017238097640512,
      "grad_norm": 0.16376633942127228,
      "learning_rate": 0.002,
      "loss": 5.195463180541992,
      "step": 510
    },
    {
      "epoch": 0.07044756678415572,
      "grad_norm": 0.2850797772407532,
      "learning_rate": 0.002,
      "loss": 5.092731475830078,
      "step": 512
    },
    {
      "epoch": 0.07072275259190633,
      "grad_norm": 0.4434758126735687,
      "learning_rate": 0.002,
      "loss": 5.143641471862793,
      "step": 514
    },
    {
      "epoch": 0.07099793839965693,
      "grad_norm": 0.21823500096797943,
      "learning_rate": 0.002,
      "loss": 5.127373695373535,
      "step": 516
    },
    {
      "epoch": 0.07127312420740754,
      "grad_norm": 0.32407787442207336,
      "learning_rate": 0.002,
      "loss": 5.190301895141602,
      "step": 518
    },
    {
      "epoch": 0.07154831001515816,
      "grad_norm": 0.13600824773311615,
      "learning_rate": 0.002,
      "loss": 5.173614025115967,
      "step": 520
    },
    {
      "epoch": 0.07182349582290876,
      "grad_norm": 0.14482347667217255,
      "learning_rate": 0.002,
      "loss": 5.082902908325195,
      "step": 522
    },
    {
      "epoch": 0.07209868163065937,
      "grad_norm": 0.15018437802791595,
      "learning_rate": 0.002,
      "loss": 5.101534366607666,
      "step": 524
    },
    {
      "epoch": 0.07237386743840997,
      "grad_norm": 0.20824319124221802,
      "learning_rate": 0.002,
      "loss": 5.099643707275391,
      "step": 526
    },
    {
      "epoch": 0.07264905324616058,
      "grad_norm": 0.1869087666273117,
      "learning_rate": 0.002,
      "loss": 5.025424003601074,
      "step": 528
    },
    {
      "epoch": 0.0729242390539112,
      "grad_norm": 0.4003094434738159,
      "learning_rate": 0.002,
      "loss": 5.094318389892578,
      "step": 530
    },
    {
      "epoch": 0.0731994248616618,
      "grad_norm": 0.2473045438528061,
      "learning_rate": 0.002,
      "loss": 5.070079803466797,
      "step": 532
    },
    {
      "epoch": 0.07347461066941241,
      "grad_norm": 0.23732584714889526,
      "learning_rate": 0.002,
      "loss": 5.086002826690674,
      "step": 534
    },
    {
      "epoch": 0.07374979647716302,
      "grad_norm": 0.3893032968044281,
      "learning_rate": 0.002,
      "loss": 5.108865261077881,
      "step": 536
    },
    {
      "epoch": 0.07402498228491362,
      "grad_norm": 0.2228219211101532,
      "learning_rate": 0.002,
      "loss": 5.084925174713135,
      "step": 538
    },
    {
      "epoch": 0.07430016809266424,
      "grad_norm": 0.2623922824859619,
      "learning_rate": 0.002,
      "loss": 5.002023696899414,
      "step": 540
    },
    {
      "epoch": 0.07457535390041484,
      "grad_norm": 0.17372991144657135,
      "learning_rate": 0.002,
      "loss": 5.020524024963379,
      "step": 542
    },
    {
      "epoch": 0.07485053970816545,
      "grad_norm": 0.11761126667261124,
      "learning_rate": 0.002,
      "loss": 4.986300468444824,
      "step": 544
    },
    {
      "epoch": 0.07512572551591606,
      "grad_norm": 0.3312435448169708,
      "learning_rate": 0.002,
      "loss": 5.074474334716797,
      "step": 546
    },
    {
      "epoch": 0.07540091132366666,
      "grad_norm": 0.17060814797878265,
      "learning_rate": 0.002,
      "loss": 5.008332252502441,
      "step": 548
    },
    {
      "epoch": 0.07567609713141728,
      "grad_norm": 0.3611827790737152,
      "learning_rate": 0.002,
      "loss": 4.996535301208496,
      "step": 550
    },
    {
      "epoch": 0.07595128293916788,
      "grad_norm": 0.19224949181079865,
      "learning_rate": 0.002,
      "loss": 4.994973659515381,
      "step": 552
    },
    {
      "epoch": 0.07622646874691849,
      "grad_norm": 0.39643338322639465,
      "learning_rate": 0.002,
      "loss": 4.967432975769043,
      "step": 554
    },
    {
      "epoch": 0.0765016545546691,
      "grad_norm": 0.18120507895946503,
      "learning_rate": 0.002,
      "loss": 5.000313758850098,
      "step": 556
    },
    {
      "epoch": 0.0767768403624197,
      "grad_norm": 0.17837364971637726,
      "learning_rate": 0.002,
      "loss": 5.027045249938965,
      "step": 558
    },
    {
      "epoch": 0.07705202617017032,
      "grad_norm": 0.16820532083511353,
      "learning_rate": 0.002,
      "loss": 4.974862575531006,
      "step": 560
    },
    {
      "epoch": 0.07732721197792093,
      "grad_norm": 0.18187011778354645,
      "learning_rate": 0.002,
      "loss": 5.01527214050293,
      "step": 562
    },
    {
      "epoch": 0.07760239778567153,
      "grad_norm": 0.15758183598518372,
      "learning_rate": 0.002,
      "loss": 4.922277450561523,
      "step": 564
    },
    {
      "epoch": 0.07787758359342214,
      "grad_norm": 0.22981365025043488,
      "learning_rate": 0.002,
      "loss": 4.946080207824707,
      "step": 566
    },
    {
      "epoch": 0.07815276940117274,
      "grad_norm": 0.20590882003307343,
      "learning_rate": 0.002,
      "loss": 4.954504489898682,
      "step": 568
    },
    {
      "epoch": 0.07842795520892336,
      "grad_norm": 0.28083306550979614,
      "learning_rate": 0.002,
      "loss": 4.9085845947265625,
      "step": 570
    },
    {
      "epoch": 0.07870314101667397,
      "grad_norm": 0.2500035762786865,
      "learning_rate": 0.002,
      "loss": 4.914648532867432,
      "step": 572
    },
    {
      "epoch": 0.07897832682442457,
      "grad_norm": 0.17200511693954468,
      "learning_rate": 0.002,
      "loss": 4.9434967041015625,
      "step": 574
    },
    {
      "epoch": 0.07925351263217519,
      "grad_norm": 0.16137336194515228,
      "learning_rate": 0.002,
      "loss": 4.946732521057129,
      "step": 576
    },
    {
      "epoch": 0.07952869843992579,
      "grad_norm": 0.1545051783323288,
      "learning_rate": 0.002,
      "loss": 4.8626813888549805,
      "step": 578
    },
    {
      "epoch": 0.0798038842476764,
      "grad_norm": 0.20111393928527832,
      "learning_rate": 0.002,
      "loss": 4.875678539276123,
      "step": 580
    },
    {
      "epoch": 0.08007907005542701,
      "grad_norm": 0.17795763909816742,
      "learning_rate": 0.002,
      "loss": 4.9494476318359375,
      "step": 582
    },
    {
      "epoch": 0.08035425586317761,
      "grad_norm": 0.4063122272491455,
      "learning_rate": 0.002,
      "loss": 4.918981552124023,
      "step": 584
    },
    {
      "epoch": 0.08062944167092823,
      "grad_norm": 0.20020106434822083,
      "learning_rate": 0.002,
      "loss": 4.850935935974121,
      "step": 586
    },
    {
      "epoch": 0.08090462747867884,
      "grad_norm": 0.47259241342544556,
      "learning_rate": 0.002,
      "loss": 4.902453422546387,
      "step": 588
    },
    {
      "epoch": 0.08117981328642944,
      "grad_norm": 0.2422272264957428,
      "learning_rate": 0.002,
      "loss": 4.878162860870361,
      "step": 590
    },
    {
      "epoch": 0.08145499909418005,
      "grad_norm": 0.15860146284103394,
      "learning_rate": 0.002,
      "loss": 4.8455810546875,
      "step": 592
    },
    {
      "epoch": 0.08173018490193065,
      "grad_norm": 0.2307584434747696,
      "learning_rate": 0.002,
      "loss": 4.8768205642700195,
      "step": 594
    },
    {
      "epoch": 0.08200537070968127,
      "grad_norm": 0.19694218039512634,
      "learning_rate": 0.002,
      "loss": 4.822709083557129,
      "step": 596
    },
    {
      "epoch": 0.08228055651743188,
      "grad_norm": 0.1305280178785324,
      "learning_rate": 0.002,
      "loss": 4.8699564933776855,
      "step": 598
    },
    {
      "epoch": 0.08255574232518248,
      "grad_norm": 0.18540716171264648,
      "learning_rate": 0.002,
      "loss": 4.860142230987549,
      "step": 600
    },
    {
      "epoch": 0.0828309281329331,
      "grad_norm": 0.24909895658493042,
      "learning_rate": 0.002,
      "loss": 4.864336967468262,
      "step": 602
    },
    {
      "epoch": 0.08310611394068369,
      "grad_norm": 0.18472068011760712,
      "learning_rate": 0.002,
      "loss": 4.790886402130127,
      "step": 604
    },
    {
      "epoch": 0.0833812997484343,
      "grad_norm": 0.21933820843696594,
      "learning_rate": 0.002,
      "loss": 4.853397846221924,
      "step": 606
    },
    {
      "epoch": 0.08365648555618492,
      "grad_norm": 0.22688309848308563,
      "learning_rate": 0.002,
      "loss": 4.805846214294434,
      "step": 608
    },
    {
      "epoch": 0.08393167136393552,
      "grad_norm": 0.15011639893054962,
      "learning_rate": 0.002,
      "loss": 4.852166175842285,
      "step": 610
    },
    {
      "epoch": 0.08420685717168613,
      "grad_norm": 0.19522041082382202,
      "learning_rate": 0.002,
      "loss": 4.83161735534668,
      "step": 612
    },
    {
      "epoch": 0.08448204297943675,
      "grad_norm": 0.1357078105211258,
      "learning_rate": 0.002,
      "loss": 4.7488555908203125,
      "step": 614
    },
    {
      "epoch": 0.08475722878718735,
      "grad_norm": 0.17145530879497528,
      "learning_rate": 0.002,
      "loss": 4.757747650146484,
      "step": 616
    },
    {
      "epoch": 0.08503241459493796,
      "grad_norm": 0.18144947290420532,
      "learning_rate": 0.002,
      "loss": 4.78239107131958,
      "step": 618
    },
    {
      "epoch": 0.08530760040268856,
      "grad_norm": 0.17999231815338135,
      "learning_rate": 0.002,
      "loss": 4.714495658874512,
      "step": 620
    },
    {
      "epoch": 0.08558278621043917,
      "grad_norm": 0.2243984341621399,
      "learning_rate": 0.002,
      "loss": 4.782314300537109,
      "step": 622
    },
    {
      "epoch": 0.08585797201818979,
      "grad_norm": 0.3798968493938446,
      "learning_rate": 0.002,
      "loss": 4.807233810424805,
      "step": 624
    },
    {
      "epoch": 0.08613315782594039,
      "grad_norm": 0.5582059025764465,
      "learning_rate": 0.002,
      "loss": 4.818206787109375,
      "step": 626
    },
    {
      "epoch": 0.086408343633691,
      "grad_norm": 0.21050851047039032,
      "learning_rate": 0.002,
      "loss": 4.757009983062744,
      "step": 628
    },
    {
      "epoch": 0.0866835294414416,
      "grad_norm": 0.3325020968914032,
      "learning_rate": 0.002,
      "loss": 4.767024993896484,
      "step": 630
    },
    {
      "epoch": 0.08695871524919221,
      "grad_norm": 0.19882626831531525,
      "learning_rate": 0.002,
      "loss": 4.717421531677246,
      "step": 632
    },
    {
      "epoch": 0.08723390105694283,
      "grad_norm": 0.16984866559505463,
      "learning_rate": 0.002,
      "loss": 4.705257415771484,
      "step": 634
    },
    {
      "epoch": 0.08750908686469343,
      "grad_norm": 0.13186778128147125,
      "learning_rate": 0.002,
      "loss": 4.702050685882568,
      "step": 636
    },
    {
      "epoch": 0.08778427267244404,
      "grad_norm": 0.18790850043296814,
      "learning_rate": 0.002,
      "loss": 4.69830322265625,
      "step": 638
    },
    {
      "epoch": 0.08805945848019464,
      "grad_norm": 0.18860362470149994,
      "learning_rate": 0.002,
      "loss": 4.674144744873047,
      "step": 640
    },
    {
      "epoch": 0.08833464428794525,
      "grad_norm": 0.2098092883825302,
      "learning_rate": 0.002,
      "loss": 4.667142868041992,
      "step": 642
    },
    {
      "epoch": 0.08860983009569587,
      "grad_norm": 0.3412386178970337,
      "learning_rate": 0.002,
      "loss": 4.688514709472656,
      "step": 644
    },
    {
      "epoch": 0.08888501590344647,
      "grad_norm": 0.23807968199253082,
      "learning_rate": 0.002,
      "loss": 4.640605926513672,
      "step": 646
    },
    {
      "epoch": 0.08916020171119708,
      "grad_norm": 0.2109958380460739,
      "learning_rate": 0.002,
      "loss": 4.659372329711914,
      "step": 648
    },
    {
      "epoch": 0.0894353875189477,
      "grad_norm": 0.27253326773643494,
      "learning_rate": 0.002,
      "loss": 4.699779987335205,
      "step": 650
    },
    {
      "epoch": 0.0897105733266983,
      "grad_norm": 0.1882486343383789,
      "learning_rate": 0.002,
      "loss": 4.700141906738281,
      "step": 652
    },
    {
      "epoch": 0.08998575913444891,
      "grad_norm": 0.14631207287311554,
      "learning_rate": 0.002,
      "loss": 4.695714950561523,
      "step": 654
    },
    {
      "epoch": 0.09026094494219951,
      "grad_norm": 0.18822193145751953,
      "learning_rate": 0.002,
      "loss": 4.6283674240112305,
      "step": 656
    },
    {
      "epoch": 0.09053613074995012,
      "grad_norm": 0.13328978419303894,
      "learning_rate": 0.002,
      "loss": 4.611923694610596,
      "step": 658
    },
    {
      "epoch": 0.09081131655770074,
      "grad_norm": 0.2148582935333252,
      "learning_rate": 0.002,
      "loss": 4.657674312591553,
      "step": 660
    },
    {
      "epoch": 0.09108650236545134,
      "grad_norm": 0.2444244921207428,
      "learning_rate": 0.002,
      "loss": 4.57067346572876,
      "step": 662
    },
    {
      "epoch": 0.09136168817320195,
      "grad_norm": 0.12405724078416824,
      "learning_rate": 0.002,
      "loss": 4.602912902832031,
      "step": 664
    },
    {
      "epoch": 0.09163687398095255,
      "grad_norm": 0.16591112315654755,
      "learning_rate": 0.002,
      "loss": 4.563486576080322,
      "step": 666
    },
    {
      "epoch": 0.09191205978870316,
      "grad_norm": 0.4338766932487488,
      "learning_rate": 0.002,
      "loss": 4.594257354736328,
      "step": 668
    },
    {
      "epoch": 0.09218724559645378,
      "grad_norm": 0.40908506512641907,
      "learning_rate": 0.002,
      "loss": 4.633572578430176,
      "step": 670
    },
    {
      "epoch": 0.09246243140420438,
      "grad_norm": 0.4793647229671478,
      "learning_rate": 0.002,
      "loss": 4.554540157318115,
      "step": 672
    },
    {
      "epoch": 0.09273761721195499,
      "grad_norm": 0.18422242999076843,
      "learning_rate": 0.002,
      "loss": 4.597209930419922,
      "step": 674
    },
    {
      "epoch": 0.0930128030197056,
      "grad_norm": 0.15094241499900818,
      "learning_rate": 0.002,
      "loss": 4.514779567718506,
      "step": 676
    },
    {
      "epoch": 0.0932879888274562,
      "grad_norm": 0.150856614112854,
      "learning_rate": 0.002,
      "loss": 4.510847091674805,
      "step": 678
    },
    {
      "epoch": 0.09356317463520682,
      "grad_norm": 0.16908828914165497,
      "learning_rate": 0.002,
      "loss": 4.523092746734619,
      "step": 680
    },
    {
      "epoch": 0.09383836044295742,
      "grad_norm": 0.16371223330497742,
      "learning_rate": 0.002,
      "loss": 4.513749122619629,
      "step": 682
    },
    {
      "epoch": 0.09411354625070803,
      "grad_norm": 0.17606227099895477,
      "learning_rate": 0.002,
      "loss": 4.492681503295898,
      "step": 684
    },
    {
      "epoch": 0.09438873205845864,
      "grad_norm": 0.27963921427726746,
      "learning_rate": 0.002,
      "loss": 4.536216735839844,
      "step": 686
    },
    {
      "epoch": 0.09466391786620924,
      "grad_norm": 0.21236661076545715,
      "learning_rate": 0.002,
      "loss": 4.4606475830078125,
      "step": 688
    },
    {
      "epoch": 0.09493910367395986,
      "grad_norm": 0.14710013568401337,
      "learning_rate": 0.002,
      "loss": 4.501225471496582,
      "step": 690
    },
    {
      "epoch": 0.09521428948171046,
      "grad_norm": 0.1271935999393463,
      "learning_rate": 0.002,
      "loss": 4.503880023956299,
      "step": 692
    },
    {
      "epoch": 0.09548947528946107,
      "grad_norm": 0.3680391311645508,
      "learning_rate": 0.002,
      "loss": 4.4654388427734375,
      "step": 694
    },
    {
      "epoch": 0.09576466109721168,
      "grad_norm": 0.17871491611003876,
      "learning_rate": 0.002,
      "loss": 4.443416118621826,
      "step": 696
    },
    {
      "epoch": 0.09603984690496228,
      "grad_norm": 0.2777758240699768,
      "learning_rate": 0.002,
      "loss": 4.398466110229492,
      "step": 698
    },
    {
      "epoch": 0.0963150327127129,
      "grad_norm": 0.17833998799324036,
      "learning_rate": 0.002,
      "loss": 4.452940940856934,
      "step": 700
    },
    {
      "epoch": 0.09659021852046351,
      "grad_norm": 0.15595296025276184,
      "learning_rate": 0.002,
      "loss": 4.438858985900879,
      "step": 702
    },
    {
      "epoch": 0.09686540432821411,
      "grad_norm": 0.16619594395160675,
      "learning_rate": 0.002,
      "loss": 4.433330059051514,
      "step": 704
    },
    {
      "epoch": 0.09714059013596472,
      "grad_norm": 0.1222728043794632,
      "learning_rate": 0.002,
      "loss": 4.39434289932251,
      "step": 706
    },
    {
      "epoch": 0.09741577594371532,
      "grad_norm": 0.15771134197711945,
      "learning_rate": 0.002,
      "loss": 4.333987712860107,
      "step": 708
    },
    {
      "epoch": 0.09769096175146594,
      "grad_norm": 0.17772743105888367,
      "learning_rate": 0.002,
      "loss": 4.4423933029174805,
      "step": 710
    },
    {
      "epoch": 0.09796614755921655,
      "grad_norm": 0.15167096257209778,
      "learning_rate": 0.002,
      "loss": 4.440036296844482,
      "step": 712
    },
    {
      "epoch": 0.09824133336696715,
      "grad_norm": 0.17949995398521423,
      "learning_rate": 0.002,
      "loss": 4.398971080780029,
      "step": 714
    },
    {
      "epoch": 0.09851651917471776,
      "grad_norm": 0.21128439903259277,
      "learning_rate": 0.002,
      "loss": 4.401559829711914,
      "step": 716
    },
    {
      "epoch": 0.09879170498246836,
      "grad_norm": 0.2326313853263855,
      "learning_rate": 0.002,
      "loss": 4.393276691436768,
      "step": 718
    },
    {
      "epoch": 0.09906689079021898,
      "grad_norm": 0.1820513904094696,
      "learning_rate": 0.002,
      "loss": 4.378488540649414,
      "step": 720
    },
    {
      "epoch": 0.09934207659796959,
      "grad_norm": 0.14860595762729645,
      "learning_rate": 0.002,
      "loss": 4.357822418212891,
      "step": 722
    },
    {
      "epoch": 0.09961726240572019,
      "grad_norm": 0.19569337368011475,
      "learning_rate": 0.002,
      "loss": 4.310861587524414,
      "step": 724
    },
    {
      "epoch": 0.0998924482134708,
      "grad_norm": 0.26346126198768616,
      "learning_rate": 0.002,
      "loss": 4.3457441329956055,
      "step": 726
    },
    {
      "epoch": 0.10016763402122142,
      "grad_norm": 0.27065011858940125,
      "learning_rate": 0.002,
      "loss": 4.337185859680176,
      "step": 728
    },
    {
      "epoch": 0.10044281982897202,
      "grad_norm": 0.31008613109588623,
      "learning_rate": 0.002,
      "loss": 4.274467468261719,
      "step": 730
    },
    {
      "epoch": 0.10071800563672263,
      "grad_norm": 0.12164389342069626,
      "learning_rate": 0.002,
      "loss": 4.305158615112305,
      "step": 732
    },
    {
      "epoch": 0.10099319144447323,
      "grad_norm": 0.19151917099952698,
      "learning_rate": 0.002,
      "loss": 4.340338230133057,
      "step": 734
    },
    {
      "epoch": 0.10126837725222385,
      "grad_norm": 0.13470330834388733,
      "learning_rate": 0.002,
      "loss": 4.335775375366211,
      "step": 736
    },
    {
      "epoch": 0.10154356305997446,
      "grad_norm": 0.20150071382522583,
      "learning_rate": 0.002,
      "loss": 4.3309173583984375,
      "step": 738
    },
    {
      "epoch": 0.10181874886772506,
      "grad_norm": 0.1891111433506012,
      "learning_rate": 0.002,
      "loss": 4.333064079284668,
      "step": 740
    },
    {
      "epoch": 0.10209393467547567,
      "grad_norm": 0.21104037761688232,
      "learning_rate": 0.002,
      "loss": 4.330865859985352,
      "step": 742
    },
    {
      "epoch": 0.10236912048322627,
      "grad_norm": 0.1712193787097931,
      "learning_rate": 0.002,
      "loss": 4.340246200561523,
      "step": 744
    },
    {
      "epoch": 0.10264430629097689,
      "grad_norm": 0.13044415414333344,
      "learning_rate": 0.002,
      "loss": 4.302634239196777,
      "step": 746
    },
    {
      "epoch": 0.1029194920987275,
      "grad_norm": 0.12096560746431351,
      "learning_rate": 0.002,
      "loss": 4.3311357498168945,
      "step": 748
    },
    {
      "epoch": 0.1031946779064781,
      "grad_norm": 0.20095813274383545,
      "learning_rate": 0.002,
      "loss": 4.235688209533691,
      "step": 750
    },
    {
      "epoch": 0.10346986371422871,
      "grad_norm": 0.2745949923992157,
      "learning_rate": 0.002,
      "loss": 4.289026737213135,
      "step": 752
    },
    {
      "epoch": 0.10374504952197933,
      "grad_norm": 0.333859920501709,
      "learning_rate": 0.002,
      "loss": 4.3064069747924805,
      "step": 754
    },
    {
      "epoch": 0.10402023532972993,
      "grad_norm": 0.22356009483337402,
      "learning_rate": 0.002,
      "loss": 4.273139953613281,
      "step": 756
    },
    {
      "epoch": 0.10429542113748054,
      "grad_norm": 0.2368573099374771,
      "learning_rate": 0.002,
      "loss": 4.236272811889648,
      "step": 758
    },
    {
      "epoch": 0.10457060694523114,
      "grad_norm": 0.2351982295513153,
      "learning_rate": 0.002,
      "loss": 4.235448837280273,
      "step": 760
    },
    {
      "epoch": 0.10484579275298175,
      "grad_norm": 0.14131933450698853,
      "learning_rate": 0.002,
      "loss": 4.225160598754883,
      "step": 762
    },
    {
      "epoch": 0.10512097856073237,
      "grad_norm": 0.19678615033626556,
      "learning_rate": 0.002,
      "loss": 4.279294013977051,
      "step": 764
    },
    {
      "epoch": 0.10539616436848297,
      "grad_norm": 0.13368827104568481,
      "learning_rate": 0.002,
      "loss": 4.214916706085205,
      "step": 766
    },
    {
      "epoch": 0.10567135017623358,
      "grad_norm": 0.15939494967460632,
      "learning_rate": 0.002,
      "loss": 4.1887311935424805,
      "step": 768
    },
    {
      "epoch": 0.10594653598398418,
      "grad_norm": 0.1326276659965515,
      "learning_rate": 0.002,
      "loss": 4.231847763061523,
      "step": 770
    },
    {
      "epoch": 0.1062217217917348,
      "grad_norm": 0.16426512598991394,
      "learning_rate": 0.002,
      "loss": 4.213579177856445,
      "step": 772
    },
    {
      "epoch": 0.10649690759948541,
      "grad_norm": 0.17351055145263672,
      "learning_rate": 0.002,
      "loss": 4.261187553405762,
      "step": 774
    },
    {
      "epoch": 0.10677209340723601,
      "grad_norm": 0.27499088644981384,
      "learning_rate": 0.002,
      "loss": 4.188815116882324,
      "step": 776
    },
    {
      "epoch": 0.10704727921498662,
      "grad_norm": 0.2516055703163147,
      "learning_rate": 0.002,
      "loss": 4.19350004196167,
      "step": 778
    },
    {
      "epoch": 0.10732246502273722,
      "grad_norm": 0.12650951743125916,
      "learning_rate": 0.002,
      "loss": 4.1874589920043945,
      "step": 780
    },
    {
      "epoch": 0.10759765083048783,
      "grad_norm": 0.28820979595184326,
      "learning_rate": 0.002,
      "loss": 4.201515197753906,
      "step": 782
    },
    {
      "epoch": 0.10787283663823845,
      "grad_norm": 0.24676893651485443,
      "learning_rate": 0.002,
      "loss": 4.179391860961914,
      "step": 784
    },
    {
      "epoch": 0.10814802244598905,
      "grad_norm": 0.18064388632774353,
      "learning_rate": 0.002,
      "loss": 4.186478137969971,
      "step": 786
    },
    {
      "epoch": 0.10842320825373966,
      "grad_norm": 0.1668812483549118,
      "learning_rate": 0.002,
      "loss": 4.195071220397949,
      "step": 788
    },
    {
      "epoch": 0.10869839406149027,
      "grad_norm": 0.143462672829628,
      "learning_rate": 0.002,
      "loss": 4.178162097930908,
      "step": 790
    },
    {
      "epoch": 0.10897357986924087,
      "grad_norm": 0.15839596092700958,
      "learning_rate": 0.002,
      "loss": 4.16415548324585,
      "step": 792
    },
    {
      "epoch": 0.10924876567699149,
      "grad_norm": 0.11118791252374649,
      "learning_rate": 0.002,
      "loss": 4.165972709655762,
      "step": 794
    },
    {
      "epoch": 0.10952395148474209,
      "grad_norm": 0.2280082106590271,
      "learning_rate": 0.002,
      "loss": 4.1916093826293945,
      "step": 796
    },
    {
      "epoch": 0.1097991372924927,
      "grad_norm": 0.43693625926971436,
      "learning_rate": 0.002,
      "loss": 4.123095512390137,
      "step": 798
    },
    {
      "epoch": 0.11007432310024332,
      "grad_norm": 0.17305146157741547,
      "learning_rate": 0.002,
      "loss": 4.131656646728516,
      "step": 800
    },
    {
      "epoch": 0.11034950890799392,
      "grad_norm": 0.14902959764003754,
      "learning_rate": 0.002,
      "loss": 4.1973676681518555,
      "step": 802
    },
    {
      "epoch": 0.11062469471574453,
      "grad_norm": 0.1554868370294571,
      "learning_rate": 0.002,
      "loss": 4.1456298828125,
      "step": 804
    },
    {
      "epoch": 0.11089988052349513,
      "grad_norm": 0.17361950874328613,
      "learning_rate": 0.002,
      "loss": 4.100800037384033,
      "step": 806
    },
    {
      "epoch": 0.11117506633124574,
      "grad_norm": 0.22472627460956573,
      "learning_rate": 0.002,
      "loss": 4.104100227355957,
      "step": 808
    },
    {
      "epoch": 0.11145025213899636,
      "grad_norm": 0.18032525479793549,
      "learning_rate": 0.002,
      "loss": 4.034926891326904,
      "step": 810
    },
    {
      "epoch": 0.11172543794674696,
      "grad_norm": 0.10299986600875854,
      "learning_rate": 0.002,
      "loss": 4.079965591430664,
      "step": 812
    },
    {
      "epoch": 0.11200062375449757,
      "grad_norm": 0.14288437366485596,
      "learning_rate": 0.002,
      "loss": 4.105721950531006,
      "step": 814
    },
    {
      "epoch": 0.11227580956224818,
      "grad_norm": 0.1732914000749588,
      "learning_rate": 0.002,
      "loss": 4.101507186889648,
      "step": 816
    },
    {
      "epoch": 0.11255099536999878,
      "grad_norm": 0.13818740844726562,
      "learning_rate": 0.002,
      "loss": 4.079327583312988,
      "step": 818
    },
    {
      "epoch": 0.1128261811777494,
      "grad_norm": 0.2522113621234894,
      "learning_rate": 0.002,
      "loss": 4.066526412963867,
      "step": 820
    },
    {
      "epoch": 0.1131013669855,
      "grad_norm": 0.34140270948410034,
      "learning_rate": 0.002,
      "loss": 4.108159065246582,
      "step": 822
    },
    {
      "epoch": 0.11337655279325061,
      "grad_norm": 0.20059771835803986,
      "learning_rate": 0.002,
      "loss": 4.098658561706543,
      "step": 824
    },
    {
      "epoch": 0.11365173860100122,
      "grad_norm": 0.38151419162750244,
      "learning_rate": 0.002,
      "loss": 4.147982597351074,
      "step": 826
    },
    {
      "epoch": 0.11392692440875182,
      "grad_norm": 0.14646784961223602,
      "learning_rate": 0.002,
      "loss": 4.097867965698242,
      "step": 828
    },
    {
      "epoch": 0.11420211021650244,
      "grad_norm": 0.12609107792377472,
      "learning_rate": 0.002,
      "loss": 4.066325664520264,
      "step": 830
    },
    {
      "epoch": 0.11447729602425304,
      "grad_norm": 0.1117686852812767,
      "learning_rate": 0.002,
      "loss": 4.003030300140381,
      "step": 832
    },
    {
      "epoch": 0.11475248183200365,
      "grad_norm": 0.12490089237689972,
      "learning_rate": 0.002,
      "loss": 4.112677574157715,
      "step": 834
    },
    {
      "epoch": 0.11502766763975426,
      "grad_norm": 0.14242874085903168,
      "learning_rate": 0.002,
      "loss": 4.0326690673828125,
      "step": 836
    },
    {
      "epoch": 0.11530285344750486,
      "grad_norm": 0.14653807878494263,
      "learning_rate": 0.002,
      "loss": 4.025739669799805,
      "step": 838
    },
    {
      "epoch": 0.11557803925525548,
      "grad_norm": 0.1314365714788437,
      "learning_rate": 0.002,
      "loss": 4.008185386657715,
      "step": 840
    },
    {
      "epoch": 0.11585322506300609,
      "grad_norm": 0.16904298961162567,
      "learning_rate": 0.002,
      "loss": 4.093296051025391,
      "step": 842
    },
    {
      "epoch": 0.11612841087075669,
      "grad_norm": 0.18276812136173248,
      "learning_rate": 0.002,
      "loss": 4.027013778686523,
      "step": 844
    },
    {
      "epoch": 0.1164035966785073,
      "grad_norm": 0.25904011726379395,
      "learning_rate": 0.002,
      "loss": 4.082959175109863,
      "step": 846
    },
    {
      "epoch": 0.1166787824862579,
      "grad_norm": 0.23439869284629822,
      "learning_rate": 0.002,
      "loss": 4.0461649894714355,
      "step": 848
    },
    {
      "epoch": 0.11695396829400852,
      "grad_norm": 0.17634738981723785,
      "learning_rate": 0.002,
      "loss": 4.060341835021973,
      "step": 850
    },
    {
      "epoch": 0.11722915410175913,
      "grad_norm": 0.289179265499115,
      "learning_rate": 0.002,
      "loss": 3.9904918670654297,
      "step": 852
    },
    {
      "epoch": 0.11750433990950973,
      "grad_norm": 0.16185270249843597,
      "learning_rate": 0.002,
      "loss": 4.02348518371582,
      "step": 854
    },
    {
      "epoch": 0.11777952571726034,
      "grad_norm": 0.11348506063222885,
      "learning_rate": 0.002,
      "loss": 3.9959726333618164,
      "step": 856
    },
    {
      "epoch": 0.11805471152501094,
      "grad_norm": 0.1196630597114563,
      "learning_rate": 0.002,
      "loss": 4.025449275970459,
      "step": 858
    },
    {
      "epoch": 0.11832989733276156,
      "grad_norm": 0.11272067576646805,
      "learning_rate": 0.002,
      "loss": 3.9549190998077393,
      "step": 860
    },
    {
      "epoch": 0.11860508314051217,
      "grad_norm": 0.153950035572052,
      "learning_rate": 0.002,
      "loss": 3.9449968338012695,
      "step": 862
    },
    {
      "epoch": 0.11888026894826277,
      "grad_norm": 0.35272225737571716,
      "learning_rate": 0.002,
      "loss": 4.047335624694824,
      "step": 864
    },
    {
      "epoch": 0.11915545475601338,
      "grad_norm": 0.11982373148202896,
      "learning_rate": 0.002,
      "loss": 4.024560451507568,
      "step": 866
    },
    {
      "epoch": 0.119430640563764,
      "grad_norm": 0.2983357608318329,
      "learning_rate": 0.002,
      "loss": 4.0367913246154785,
      "step": 868
    },
    {
      "epoch": 0.1197058263715146,
      "grad_norm": 0.1798505187034607,
      "learning_rate": 0.002,
      "loss": 4.046613693237305,
      "step": 870
    },
    {
      "epoch": 0.11998101217926521,
      "grad_norm": 0.31153830885887146,
      "learning_rate": 0.002,
      "loss": 3.996466636657715,
      "step": 872
    },
    {
      "epoch": 0.12025619798701581,
      "grad_norm": 0.1592666059732437,
      "learning_rate": 0.002,
      "loss": 3.962977886199951,
      "step": 874
    },
    {
      "epoch": 0.12053138379476643,
      "grad_norm": 0.09556034952402115,
      "learning_rate": 0.002,
      "loss": 3.977858066558838,
      "step": 876
    },
    {
      "epoch": 0.12080656960251704,
      "grad_norm": 0.1833716779947281,
      "learning_rate": 0.002,
      "loss": 4.001723766326904,
      "step": 878
    },
    {
      "epoch": 0.12108175541026764,
      "grad_norm": 0.4592113792896271,
      "learning_rate": 0.002,
      "loss": 4.004825592041016,
      "step": 880
    },
    {
      "epoch": 0.12135694121801825,
      "grad_norm": 0.18559184670448303,
      "learning_rate": 0.002,
      "loss": 3.9383907318115234,
      "step": 882
    },
    {
      "epoch": 0.12163212702576885,
      "grad_norm": 0.1985873132944107,
      "learning_rate": 0.002,
      "loss": 3.9969868659973145,
      "step": 884
    },
    {
      "epoch": 0.12190731283351947,
      "grad_norm": 0.11674225330352783,
      "learning_rate": 0.002,
      "loss": 4.002057075500488,
      "step": 886
    },
    {
      "epoch": 0.12218249864127008,
      "grad_norm": 0.1745089590549469,
      "learning_rate": 0.002,
      "loss": 3.99122953414917,
      "step": 888
    },
    {
      "epoch": 0.12245768444902068,
      "grad_norm": 0.14514592289924622,
      "learning_rate": 0.002,
      "loss": 3.9463210105895996,
      "step": 890
    },
    {
      "epoch": 0.12273287025677129,
      "grad_norm": 0.20514898002147675,
      "learning_rate": 0.002,
      "loss": 3.9163708686828613,
      "step": 892
    },
    {
      "epoch": 0.12300805606452189,
      "grad_norm": 0.11595834791660309,
      "learning_rate": 0.002,
      "loss": 3.982377052307129,
      "step": 894
    },
    {
      "epoch": 0.1232832418722725,
      "grad_norm": 0.08535938709974289,
      "learning_rate": 0.002,
      "loss": 3.9554548263549805,
      "step": 896
    },
    {
      "epoch": 0.12355842768002312,
      "grad_norm": 0.12063740938901901,
      "learning_rate": 0.002,
      "loss": 3.935635805130005,
      "step": 898
    },
    {
      "epoch": 0.12383361348777372,
      "grad_norm": 0.11287730187177658,
      "learning_rate": 0.002,
      "loss": 3.918991804122925,
      "step": 900
    },
    {
      "epoch": 0.12410879929552433,
      "grad_norm": 0.16277316212654114,
      "learning_rate": 0.002,
      "loss": 3.914597988128662,
      "step": 902
    },
    {
      "epoch": 0.12438398510327495,
      "grad_norm": 0.15358968079090118,
      "learning_rate": 0.002,
      "loss": 3.9022271633148193,
      "step": 904
    },
    {
      "epoch": 0.12465917091102555,
      "grad_norm": 0.10879185050725937,
      "learning_rate": 0.002,
      "loss": 3.857028007507324,
      "step": 906
    },
    {
      "epoch": 0.12493435671877616,
      "grad_norm": 0.1691272258758545,
      "learning_rate": 0.002,
      "loss": 3.9106898307800293,
      "step": 908
    },
    {
      "epoch": 0.12520954252652677,
      "grad_norm": 0.2615005373954773,
      "learning_rate": 0.002,
      "loss": 3.935793399810791,
      "step": 910
    },
    {
      "epoch": 0.12548472833427737,
      "grad_norm": 0.15224875509738922,
      "learning_rate": 0.002,
      "loss": 3.911163806915283,
      "step": 912
    },
    {
      "epoch": 0.12575991414202797,
      "grad_norm": 0.1366247534751892,
      "learning_rate": 0.002,
      "loss": 3.862751007080078,
      "step": 914
    },
    {
      "epoch": 0.1260350999497786,
      "grad_norm": 0.15627583861351013,
      "learning_rate": 0.002,
      "loss": 3.891646385192871,
      "step": 916
    },
    {
      "epoch": 0.1263102857575292,
      "grad_norm": 0.11523263901472092,
      "learning_rate": 0.002,
      "loss": 3.891777515411377,
      "step": 918
    },
    {
      "epoch": 0.1265854715652798,
      "grad_norm": 0.20912784337997437,
      "learning_rate": 0.002,
      "loss": 3.897001266479492,
      "step": 920
    },
    {
      "epoch": 0.12686065737303043,
      "grad_norm": 0.43247708678245544,
      "learning_rate": 0.002,
      "loss": 3.869617223739624,
      "step": 922
    },
    {
      "epoch": 0.12713584318078103,
      "grad_norm": 0.23489487171173096,
      "learning_rate": 0.002,
      "loss": 3.9122962951660156,
      "step": 924
    },
    {
      "epoch": 0.12741102898853163,
      "grad_norm": 0.3303970694541931,
      "learning_rate": 0.002,
      "loss": 3.903201103210449,
      "step": 926
    },
    {
      "epoch": 0.12768621479628223,
      "grad_norm": 0.11706748604774475,
      "learning_rate": 0.002,
      "loss": 3.867835283279419,
      "step": 928
    },
    {
      "epoch": 0.12796140060403285,
      "grad_norm": 0.32393646240234375,
      "learning_rate": 0.002,
      "loss": 3.941718101501465,
      "step": 930
    },
    {
      "epoch": 0.12823658641178345,
      "grad_norm": 0.23881882429122925,
      "learning_rate": 0.002,
      "loss": 3.889450788497925,
      "step": 932
    },
    {
      "epoch": 0.12851177221953405,
      "grad_norm": 0.14376091957092285,
      "learning_rate": 0.002,
      "loss": 3.9146013259887695,
      "step": 934
    },
    {
      "epoch": 0.12878695802728468,
      "grad_norm": 0.1142439916729927,
      "learning_rate": 0.002,
      "loss": 3.8513431549072266,
      "step": 936
    },
    {
      "epoch": 0.12906214383503528,
      "grad_norm": 0.21028535068035126,
      "learning_rate": 0.002,
      "loss": 3.8759570121765137,
      "step": 938
    },
    {
      "epoch": 0.12933732964278588,
      "grad_norm": 0.11377944052219391,
      "learning_rate": 0.002,
      "loss": 3.8657193183898926,
      "step": 940
    },
    {
      "epoch": 0.1296125154505365,
      "grad_norm": 0.20806388556957245,
      "learning_rate": 0.002,
      "loss": 3.8477492332458496,
      "step": 942
    },
    {
      "epoch": 0.1298877012582871,
      "grad_norm": 0.15762116014957428,
      "learning_rate": 0.002,
      "loss": 3.8398070335388184,
      "step": 944
    },
    {
      "epoch": 0.1301628870660377,
      "grad_norm": 0.13926874101161957,
      "learning_rate": 0.002,
      "loss": 3.8425097465515137,
      "step": 946
    },
    {
      "epoch": 0.13043807287378834,
      "grad_norm": 0.19038385152816772,
      "learning_rate": 0.002,
      "loss": 3.824704170227051,
      "step": 948
    },
    {
      "epoch": 0.13071325868153894,
      "grad_norm": 0.10288475453853607,
      "learning_rate": 0.002,
      "loss": 3.7838666439056396,
      "step": 950
    },
    {
      "epoch": 0.13098844448928953,
      "grad_norm": 0.10028061270713806,
      "learning_rate": 0.002,
      "loss": 3.8294644355773926,
      "step": 952
    },
    {
      "epoch": 0.13126363029704013,
      "grad_norm": 0.17030370235443115,
      "learning_rate": 0.002,
      "loss": 3.865086793899536,
      "step": 954
    },
    {
      "epoch": 0.13153881610479076,
      "grad_norm": 0.3969671130180359,
      "learning_rate": 0.002,
      "loss": 3.8152871131896973,
      "step": 956
    },
    {
      "epoch": 0.13181400191254136,
      "grad_norm": 0.1454179286956787,
      "learning_rate": 0.002,
      "loss": 3.852588176727295,
      "step": 958
    },
    {
      "epoch": 0.13208918772029196,
      "grad_norm": 0.1507142186164856,
      "learning_rate": 0.002,
      "loss": 3.8434958457946777,
      "step": 960
    },
    {
      "epoch": 0.1323643735280426,
      "grad_norm": 0.09384779632091522,
      "learning_rate": 0.002,
      "loss": 3.8424158096313477,
      "step": 962
    },
    {
      "epoch": 0.1326395593357932,
      "grad_norm": 0.11985055357217789,
      "learning_rate": 0.002,
      "loss": 3.81976318359375,
      "step": 964
    },
    {
      "epoch": 0.1329147451435438,
      "grad_norm": 0.13601060211658478,
      "learning_rate": 0.002,
      "loss": 3.814598560333252,
      "step": 966
    },
    {
      "epoch": 0.13318993095129442,
      "grad_norm": 0.12052304297685623,
      "learning_rate": 0.002,
      "loss": 3.812844753265381,
      "step": 968
    },
    {
      "epoch": 0.13346511675904502,
      "grad_norm": 0.13505634665489197,
      "learning_rate": 0.002,
      "loss": 3.8131518363952637,
      "step": 970
    },
    {
      "epoch": 0.13374030256679562,
      "grad_norm": 0.1872735321521759,
      "learning_rate": 0.002,
      "loss": 3.8375844955444336,
      "step": 972
    },
    {
      "epoch": 0.13401548837454624,
      "grad_norm": 0.22092807292938232,
      "learning_rate": 0.002,
      "loss": 3.8417468070983887,
      "step": 974
    },
    {
      "epoch": 0.13429067418229684,
      "grad_norm": 0.3179752230644226,
      "learning_rate": 0.002,
      "loss": 3.81636381149292,
      "step": 976
    },
    {
      "epoch": 0.13456585999004744,
      "grad_norm": 0.11317793279886246,
      "learning_rate": 0.002,
      "loss": 3.826150417327881,
      "step": 978
    },
    {
      "epoch": 0.13484104579779804,
      "grad_norm": 0.08831849694252014,
      "learning_rate": 0.002,
      "loss": 3.7797670364379883,
      "step": 980
    },
    {
      "epoch": 0.13511623160554867,
      "grad_norm": 0.09026988595724106,
      "learning_rate": 0.002,
      "loss": 3.7572975158691406,
      "step": 982
    },
    {
      "epoch": 0.13539141741329927,
      "grad_norm": 0.14789772033691406,
      "learning_rate": 0.002,
      "loss": 3.785588264465332,
      "step": 984
    },
    {
      "epoch": 0.13566660322104987,
      "grad_norm": 0.24050858616828918,
      "learning_rate": 0.002,
      "loss": 3.7474160194396973,
      "step": 986
    },
    {
      "epoch": 0.1359417890288005,
      "grad_norm": 0.12483681738376617,
      "learning_rate": 0.002,
      "loss": 3.746908664703369,
      "step": 988
    },
    {
      "epoch": 0.1362169748365511,
      "grad_norm": 0.11405229568481445,
      "learning_rate": 0.002,
      "loss": 3.765876293182373,
      "step": 990
    },
    {
      "epoch": 0.1364921606443017,
      "grad_norm": 0.24101564288139343,
      "learning_rate": 0.002,
      "loss": 3.8102867603302,
      "step": 992
    },
    {
      "epoch": 0.13676734645205232,
      "grad_norm": 0.41024816036224365,
      "learning_rate": 0.002,
      "loss": 3.8046863079071045,
      "step": 994
    },
    {
      "epoch": 0.13704253225980292,
      "grad_norm": 0.10290379077196121,
      "learning_rate": 0.002,
      "loss": 3.7817134857177734,
      "step": 996
    },
    {
      "epoch": 0.13731771806755352,
      "grad_norm": 0.18507619202136993,
      "learning_rate": 0.002,
      "loss": 3.7873215675354004,
      "step": 998
    },
    {
      "epoch": 0.13759290387530415,
      "grad_norm": 0.08045053482055664,
      "learning_rate": 0.002,
      "loss": 3.7475507259368896,
      "step": 1000
    },
    {
      "epoch": 0.13786808968305475,
      "grad_norm": 0.10094771534204483,
      "learning_rate": 0.002,
      "loss": 3.7906365394592285,
      "step": 1002
    },
    {
      "epoch": 0.13814327549080535,
      "grad_norm": 0.093572236597538,
      "learning_rate": 0.002,
      "loss": 3.7191340923309326,
      "step": 1004
    },
    {
      "epoch": 0.13841846129855595,
      "grad_norm": 0.08307474106550217,
      "learning_rate": 0.002,
      "loss": 3.7779312133789062,
      "step": 1006
    },
    {
      "epoch": 0.13869364710630658,
      "grad_norm": 0.10328079760074615,
      "learning_rate": 0.002,
      "loss": 3.7759623527526855,
      "step": 1008
    },
    {
      "epoch": 0.13896883291405718,
      "grad_norm": 0.12152378261089325,
      "learning_rate": 0.002,
      "loss": 3.793700695037842,
      "step": 1010
    },
    {
      "epoch": 0.13924401872180778,
      "grad_norm": 0.17292208969593048,
      "learning_rate": 0.002,
      "loss": 3.765381336212158,
      "step": 1012
    },
    {
      "epoch": 0.1395192045295584,
      "grad_norm": 0.24233701825141907,
      "learning_rate": 0.002,
      "loss": 3.72212553024292,
      "step": 1014
    },
    {
      "epoch": 0.139794390337309,
      "grad_norm": 0.3111379146575928,
      "learning_rate": 0.002,
      "loss": 3.7495572566986084,
      "step": 1016
    },
    {
      "epoch": 0.1400695761450596,
      "grad_norm": 0.2051006257534027,
      "learning_rate": 0.002,
      "loss": 3.784846305847168,
      "step": 1018
    },
    {
      "epoch": 0.14034476195281023,
      "grad_norm": 0.11217872053384781,
      "learning_rate": 0.002,
      "loss": 3.7435367107391357,
      "step": 1020
    },
    {
      "epoch": 0.14061994776056083,
      "grad_norm": 0.09849204868078232,
      "learning_rate": 0.002,
      "loss": 3.7510733604431152,
      "step": 1022
    },
    {
      "epoch": 0.14089513356831143,
      "grad_norm": 0.24824956059455872,
      "learning_rate": 0.002,
      "loss": 3.789506435394287,
      "step": 1024
    },
    {
      "epoch": 0.14117031937606206,
      "grad_norm": 0.23324385285377502,
      "learning_rate": 0.002,
      "loss": 3.6975460052490234,
      "step": 1026
    },
    {
      "epoch": 0.14144550518381266,
      "grad_norm": 0.21733108162879944,
      "learning_rate": 0.002,
      "loss": 3.744752883911133,
      "step": 1028
    },
    {
      "epoch": 0.14172069099156326,
      "grad_norm": 0.13343565165996552,
      "learning_rate": 0.002,
      "loss": 3.707146406173706,
      "step": 1030
    },
    {
      "epoch": 0.14199587679931386,
      "grad_norm": 0.21538479626178741,
      "learning_rate": 0.002,
      "loss": 3.7560341358184814,
      "step": 1032
    },
    {
      "epoch": 0.14227106260706449,
      "grad_norm": 0.18774917721748352,
      "learning_rate": 0.002,
      "loss": 3.6938934326171875,
      "step": 1034
    },
    {
      "epoch": 0.14254624841481509,
      "grad_norm": 0.07940248399972916,
      "learning_rate": 0.002,
      "loss": 3.769143581390381,
      "step": 1036
    },
    {
      "epoch": 0.14282143422256569,
      "grad_norm": 0.12895141541957855,
      "learning_rate": 0.002,
      "loss": 3.7071025371551514,
      "step": 1038
    },
    {
      "epoch": 0.1430966200303163,
      "grad_norm": 0.20427992939949036,
      "learning_rate": 0.002,
      "loss": 3.7568459510803223,
      "step": 1040
    },
    {
      "epoch": 0.1433718058380669,
      "grad_norm": 0.08386005461215973,
      "learning_rate": 0.002,
      "loss": 3.683204412460327,
      "step": 1042
    },
    {
      "epoch": 0.1436469916458175,
      "grad_norm": 0.22142958641052246,
      "learning_rate": 0.002,
      "loss": 3.7107913494110107,
      "step": 1044
    },
    {
      "epoch": 0.14392217745356814,
      "grad_norm": 0.15693138539791107,
      "learning_rate": 0.002,
      "loss": 3.708256244659424,
      "step": 1046
    },
    {
      "epoch": 0.14419736326131874,
      "grad_norm": 0.1918092966079712,
      "learning_rate": 0.002,
      "loss": 3.7109341621398926,
      "step": 1048
    },
    {
      "epoch": 0.14447254906906934,
      "grad_norm": 0.08109885454177856,
      "learning_rate": 0.002,
      "loss": 3.683880090713501,
      "step": 1050
    },
    {
      "epoch": 0.14474773487681994,
      "grad_norm": 0.0937102884054184,
      "learning_rate": 0.002,
      "loss": 3.7475223541259766,
      "step": 1052
    },
    {
      "epoch": 0.14502292068457057,
      "grad_norm": 0.14164066314697266,
      "learning_rate": 0.002,
      "loss": 3.747835159301758,
      "step": 1054
    },
    {
      "epoch": 0.14529810649232117,
      "grad_norm": 0.24326269328594208,
      "learning_rate": 0.002,
      "loss": 3.695523738861084,
      "step": 1056
    },
    {
      "epoch": 0.14557329230007177,
      "grad_norm": 0.320967435836792,
      "learning_rate": 0.002,
      "loss": 3.749316930770874,
      "step": 1058
    },
    {
      "epoch": 0.1458484781078224,
      "grad_norm": 0.42719653248786926,
      "learning_rate": 0.002,
      "loss": 3.7424111366271973,
      "step": 1060
    },
    {
      "epoch": 0.146123663915573,
      "grad_norm": 0.17539623379707336,
      "learning_rate": 0.002,
      "loss": 3.712524890899658,
      "step": 1062
    },
    {
      "epoch": 0.1463988497233236,
      "grad_norm": 0.2129695862531662,
      "learning_rate": 0.002,
      "loss": 3.7443900108337402,
      "step": 1064
    },
    {
      "epoch": 0.14667403553107422,
      "grad_norm": 0.10103823989629745,
      "learning_rate": 0.002,
      "loss": 3.720921277999878,
      "step": 1066
    },
    {
      "epoch": 0.14694922133882482,
      "grad_norm": 0.0752260833978653,
      "learning_rate": 0.002,
      "loss": 3.6444520950317383,
      "step": 1068
    },
    {
      "epoch": 0.14722440714657542,
      "grad_norm": 0.10334290564060211,
      "learning_rate": 0.002,
      "loss": 3.6640994548797607,
      "step": 1070
    },
    {
      "epoch": 0.14749959295432605,
      "grad_norm": 0.07769158482551575,
      "learning_rate": 0.002,
      "loss": 3.6503398418426514,
      "step": 1072
    },
    {
      "epoch": 0.14777477876207665,
      "grad_norm": 0.07988089323043823,
      "learning_rate": 0.002,
      "loss": 3.675619602203369,
      "step": 1074
    },
    {
      "epoch": 0.14804996456982725,
      "grad_norm": 0.14441116154193878,
      "learning_rate": 0.002,
      "loss": 3.6706323623657227,
      "step": 1076
    },
    {
      "epoch": 0.14832515037757785,
      "grad_norm": 0.184129998087883,
      "learning_rate": 0.002,
      "loss": 3.7159409523010254,
      "step": 1078
    },
    {
      "epoch": 0.14860033618532847,
      "grad_norm": 0.07949097454547882,
      "learning_rate": 0.002,
      "loss": 3.655160665512085,
      "step": 1080
    },
    {
      "epoch": 0.14887552199307907,
      "grad_norm": 0.3006451427936554,
      "learning_rate": 0.002,
      "loss": 3.679159164428711,
      "step": 1082
    },
    {
      "epoch": 0.14915070780082967,
      "grad_norm": 0.2364521622657776,
      "learning_rate": 0.002,
      "loss": 3.675116539001465,
      "step": 1084
    },
    {
      "epoch": 0.1494258936085803,
      "grad_norm": 0.08561393618583679,
      "learning_rate": 0.002,
      "loss": 3.6708712577819824,
      "step": 1086
    },
    {
      "epoch": 0.1497010794163309,
      "grad_norm": 0.15465213358402252,
      "learning_rate": 0.002,
      "loss": 3.671647071838379,
      "step": 1088
    },
    {
      "epoch": 0.1499762652240815,
      "grad_norm": 0.2501758635044098,
      "learning_rate": 0.002,
      "loss": 3.694368839263916,
      "step": 1090
    },
    {
      "epoch": 0.15025145103183213,
      "grad_norm": 0.21571341156959534,
      "learning_rate": 0.002,
      "loss": 3.677192211151123,
      "step": 1092
    },
    {
      "epoch": 0.15052663683958273,
      "grad_norm": 0.10029919445514679,
      "learning_rate": 0.002,
      "loss": 3.6831493377685547,
      "step": 1094
    },
    {
      "epoch": 0.15080182264733333,
      "grad_norm": 0.2572176456451416,
      "learning_rate": 0.002,
      "loss": 3.6238512992858887,
      "step": 1096
    },
    {
      "epoch": 0.15107700845508396,
      "grad_norm": 0.477219820022583,
      "learning_rate": 0.002,
      "loss": 3.6763916015625,
      "step": 1098
    },
    {
      "epoch": 0.15135219426283456,
      "grad_norm": 0.11550236493349075,
      "learning_rate": 0.002,
      "loss": 3.667283296585083,
      "step": 1100
    },
    {
      "epoch": 0.15162738007058515,
      "grad_norm": 0.10296933352947235,
      "learning_rate": 0.002,
      "loss": 3.669147491455078,
      "step": 1102
    },
    {
      "epoch": 0.15190256587833575,
      "grad_norm": 0.11152639240026474,
      "learning_rate": 0.002,
      "loss": 3.6782636642456055,
      "step": 1104
    },
    {
      "epoch": 0.15217775168608638,
      "grad_norm": 0.08908011764287949,
      "learning_rate": 0.002,
      "loss": 3.6332015991210938,
      "step": 1106
    },
    {
      "epoch": 0.15245293749383698,
      "grad_norm": 0.10761765390634537,
      "learning_rate": 0.002,
      "loss": 3.6690924167633057,
      "step": 1108
    },
    {
      "epoch": 0.15272812330158758,
      "grad_norm": 0.08407898247241974,
      "learning_rate": 0.002,
      "loss": 3.637307643890381,
      "step": 1110
    },
    {
      "epoch": 0.1530033091093382,
      "grad_norm": 0.08351393789052963,
      "learning_rate": 0.002,
      "loss": 3.6131176948547363,
      "step": 1112
    },
    {
      "epoch": 0.1532784949170888,
      "grad_norm": 0.13890618085861206,
      "learning_rate": 0.002,
      "loss": 3.6572694778442383,
      "step": 1114
    },
    {
      "epoch": 0.1535536807248394,
      "grad_norm": 0.16165763139724731,
      "learning_rate": 0.002,
      "loss": 3.613588571548462,
      "step": 1116
    },
    {
      "epoch": 0.15382886653259004,
      "grad_norm": 0.17129099369049072,
      "learning_rate": 0.002,
      "loss": 3.5668067932128906,
      "step": 1118
    },
    {
      "epoch": 0.15410405234034064,
      "grad_norm": 0.20492419600486755,
      "learning_rate": 0.002,
      "loss": 3.6077756881713867,
      "step": 1120
    },
    {
      "epoch": 0.15437923814809124,
      "grad_norm": 0.1402195692062378,
      "learning_rate": 0.002,
      "loss": 3.6016018390655518,
      "step": 1122
    },
    {
      "epoch": 0.15465442395584186,
      "grad_norm": 0.10957857966423035,
      "learning_rate": 0.002,
      "loss": 3.6003777980804443,
      "step": 1124
    },
    {
      "epoch": 0.15492960976359246,
      "grad_norm": 0.1616055816411972,
      "learning_rate": 0.002,
      "loss": 3.664889097213745,
      "step": 1126
    },
    {
      "epoch": 0.15520479557134306,
      "grad_norm": 0.35922208428382874,
      "learning_rate": 0.002,
      "loss": 3.6619527339935303,
      "step": 1128
    },
    {
      "epoch": 0.15547998137909366,
      "grad_norm": 0.23212897777557373,
      "learning_rate": 0.002,
      "loss": 3.5974416732788086,
      "step": 1130
    },
    {
      "epoch": 0.1557551671868443,
      "grad_norm": 0.1001708135008812,
      "learning_rate": 0.002,
      "loss": 3.61344575881958,
      "step": 1132
    },
    {
      "epoch": 0.1560303529945949,
      "grad_norm": 0.10374768823385239,
      "learning_rate": 0.002,
      "loss": 3.644503355026245,
      "step": 1134
    },
    {
      "epoch": 0.1563055388023455,
      "grad_norm": 0.08039294183254242,
      "learning_rate": 0.002,
      "loss": 3.6058828830718994,
      "step": 1136
    },
    {
      "epoch": 0.15658072461009612,
      "grad_norm": 0.14270003139972687,
      "learning_rate": 0.002,
      "loss": 3.5740880966186523,
      "step": 1138
    },
    {
      "epoch": 0.15685591041784672,
      "grad_norm": 0.2273457646369934,
      "learning_rate": 0.002,
      "loss": 3.632328987121582,
      "step": 1140
    },
    {
      "epoch": 0.15713109622559732,
      "grad_norm": 0.19229896366596222,
      "learning_rate": 0.002,
      "loss": 3.5946192741394043,
      "step": 1142
    },
    {
      "epoch": 0.15740628203334794,
      "grad_norm": 0.27113208174705505,
      "learning_rate": 0.002,
      "loss": 3.623528003692627,
      "step": 1144
    },
    {
      "epoch": 0.15768146784109854,
      "grad_norm": 0.16132675111293793,
      "learning_rate": 0.002,
      "loss": 3.5714335441589355,
      "step": 1146
    },
    {
      "epoch": 0.15795665364884914,
      "grad_norm": 0.14332860708236694,
      "learning_rate": 0.002,
      "loss": 3.652313709259033,
      "step": 1148
    },
    {
      "epoch": 0.15823183945659977,
      "grad_norm": 0.07485025376081467,
      "learning_rate": 0.002,
      "loss": 3.588320255279541,
      "step": 1150
    },
    {
      "epoch": 0.15850702526435037,
      "grad_norm": 0.09231500327587128,
      "learning_rate": 0.002,
      "loss": 3.6211564540863037,
      "step": 1152
    },
    {
      "epoch": 0.15878221107210097,
      "grad_norm": 0.162511944770813,
      "learning_rate": 0.002,
      "loss": 3.6582319736480713,
      "step": 1154
    },
    {
      "epoch": 0.15905739687985157,
      "grad_norm": 0.10635953396558762,
      "learning_rate": 0.002,
      "loss": 3.5401415824890137,
      "step": 1156
    },
    {
      "epoch": 0.1593325826876022,
      "grad_norm": 0.08762052655220032,
      "learning_rate": 0.002,
      "loss": 3.596684455871582,
      "step": 1158
    },
    {
      "epoch": 0.1596077684953528,
      "grad_norm": 0.15650759637355804,
      "learning_rate": 0.002,
      "loss": 3.6147990226745605,
      "step": 1160
    },
    {
      "epoch": 0.1598829543031034,
      "grad_norm": 0.4491199254989624,
      "learning_rate": 0.002,
      "loss": 3.6555614471435547,
      "step": 1162
    },
    {
      "epoch": 0.16015814011085402,
      "grad_norm": 0.3275395929813385,
      "learning_rate": 0.002,
      "loss": 3.6130685806274414,
      "step": 1164
    },
    {
      "epoch": 0.16043332591860462,
      "grad_norm": 0.3229146897792816,
      "learning_rate": 0.002,
      "loss": 3.643585681915283,
      "step": 1166
    },
    {
      "epoch": 0.16070851172635522,
      "grad_norm": 0.11051736772060394,
      "learning_rate": 0.002,
      "loss": 3.5659475326538086,
      "step": 1168
    },
    {
      "epoch": 0.16098369753410585,
      "grad_norm": 0.1079726591706276,
      "learning_rate": 0.002,
      "loss": 3.6179563999176025,
      "step": 1170
    },
    {
      "epoch": 0.16125888334185645,
      "grad_norm": 0.0758347362279892,
      "learning_rate": 0.002,
      "loss": 3.5436019897460938,
      "step": 1172
    },
    {
      "epoch": 0.16153406914960705,
      "grad_norm": 0.08512304723262787,
      "learning_rate": 0.002,
      "loss": 3.5783233642578125,
      "step": 1174
    },
    {
      "epoch": 0.16180925495735768,
      "grad_norm": 0.08840682357549667,
      "learning_rate": 0.002,
      "loss": 3.5537638664245605,
      "step": 1176
    },
    {
      "epoch": 0.16208444076510828,
      "grad_norm": 0.07437105476856232,
      "learning_rate": 0.002,
      "loss": 3.6465210914611816,
      "step": 1178
    },
    {
      "epoch": 0.16235962657285888,
      "grad_norm": 0.06886128336191177,
      "learning_rate": 0.002,
      "loss": 3.531740665435791,
      "step": 1180
    },
    {
      "epoch": 0.16263481238060948,
      "grad_norm": 0.07994213700294495,
      "learning_rate": 0.002,
      "loss": 3.5959548950195312,
      "step": 1182
    },
    {
      "epoch": 0.1629099981883601,
      "grad_norm": 0.13869403302669525,
      "learning_rate": 0.002,
      "loss": 3.579404830932617,
      "step": 1184
    },
    {
      "epoch": 0.1631851839961107,
      "grad_norm": 0.10044139623641968,
      "learning_rate": 0.002,
      "loss": 3.5740442276000977,
      "step": 1186
    },
    {
      "epoch": 0.1634603698038613,
      "grad_norm": 0.14856308698654175,
      "learning_rate": 0.002,
      "loss": 3.5843939781188965,
      "step": 1188
    },
    {
      "epoch": 0.16373555561161193,
      "grad_norm": 0.3230150043964386,
      "learning_rate": 0.002,
      "loss": 3.5989222526550293,
      "step": 1190
    },
    {
      "epoch": 0.16401074141936253,
      "grad_norm": 0.39391180872917175,
      "learning_rate": 0.002,
      "loss": 3.5880048274993896,
      "step": 1192
    },
    {
      "epoch": 0.16428592722711313,
      "grad_norm": 0.10774212330579758,
      "learning_rate": 0.002,
      "loss": 3.5772433280944824,
      "step": 1194
    },
    {
      "epoch": 0.16456111303486376,
      "grad_norm": 0.0729217529296875,
      "learning_rate": 0.002,
      "loss": 3.5448532104492188,
      "step": 1196
    },
    {
      "epoch": 0.16483629884261436,
      "grad_norm": 0.14048808813095093,
      "learning_rate": 0.002,
      "loss": 3.539982318878174,
      "step": 1198
    },
    {
      "epoch": 0.16511148465036496,
      "grad_norm": 0.24935895204544067,
      "learning_rate": 0.002,
      "loss": 3.544029951095581,
      "step": 1200
    },
    {
      "epoch": 0.1653866704581156,
      "grad_norm": 0.06701217591762543,
      "learning_rate": 0.002,
      "loss": 3.5621633529663086,
      "step": 1202
    },
    {
      "epoch": 0.1656618562658662,
      "grad_norm": 0.3608054518699646,
      "learning_rate": 0.002,
      "loss": 3.577733039855957,
      "step": 1204
    },
    {
      "epoch": 0.1659370420736168,
      "grad_norm": 0.14364048838615417,
      "learning_rate": 0.002,
      "loss": 3.525162696838379,
      "step": 1206
    },
    {
      "epoch": 0.16621222788136739,
      "grad_norm": 0.1339109092950821,
      "learning_rate": 0.002,
      "loss": 3.5782172679901123,
      "step": 1208
    },
    {
      "epoch": 0.166487413689118,
      "grad_norm": 0.10771143436431885,
      "learning_rate": 0.002,
      "loss": 3.574492931365967,
      "step": 1210
    },
    {
      "epoch": 0.1667625994968686,
      "grad_norm": 0.20198208093643188,
      "learning_rate": 0.002,
      "loss": 3.5848617553710938,
      "step": 1212
    },
    {
      "epoch": 0.1670377853046192,
      "grad_norm": 0.10644777119159698,
      "learning_rate": 0.002,
      "loss": 3.552183151245117,
      "step": 1214
    },
    {
      "epoch": 0.16731297111236984,
      "grad_norm": 0.08743225783109665,
      "learning_rate": 0.002,
      "loss": 3.5306882858276367,
      "step": 1216
    },
    {
      "epoch": 0.16758815692012044,
      "grad_norm": 0.10974114388227463,
      "learning_rate": 0.002,
      "loss": 3.5330867767333984,
      "step": 1218
    },
    {
      "epoch": 0.16786334272787104,
      "grad_norm": 0.11734060943126678,
      "learning_rate": 0.002,
      "loss": 3.5527584552764893,
      "step": 1220
    },
    {
      "epoch": 0.16813852853562167,
      "grad_norm": 0.06564583629369736,
      "learning_rate": 0.002,
      "loss": 3.542442560195923,
      "step": 1222
    },
    {
      "epoch": 0.16841371434337227,
      "grad_norm": 0.2052120864391327,
      "learning_rate": 0.002,
      "loss": 3.5325684547424316,
      "step": 1224
    },
    {
      "epoch": 0.16868890015112287,
      "grad_norm": 0.31805357336997986,
      "learning_rate": 0.002,
      "loss": 3.547168731689453,
      "step": 1226
    },
    {
      "epoch": 0.1689640859588735,
      "grad_norm": 0.3698894679546356,
      "learning_rate": 0.002,
      "loss": 3.5373759269714355,
      "step": 1228
    },
    {
      "epoch": 0.1692392717666241,
      "grad_norm": 0.10911718755960464,
      "learning_rate": 0.002,
      "loss": 3.522480010986328,
      "step": 1230
    },
    {
      "epoch": 0.1695144575743747,
      "grad_norm": 0.12088444083929062,
      "learning_rate": 0.002,
      "loss": 3.5060200691223145,
      "step": 1232
    },
    {
      "epoch": 0.1697896433821253,
      "grad_norm": 0.10645020753145218,
      "learning_rate": 0.002,
      "loss": 3.5644335746765137,
      "step": 1234
    },
    {
      "epoch": 0.17006482918987592,
      "grad_norm": 0.08090711385011673,
      "learning_rate": 0.002,
      "loss": 3.505399227142334,
      "step": 1236
    },
    {
      "epoch": 0.17034001499762652,
      "grad_norm": 0.13356108963489532,
      "learning_rate": 0.002,
      "loss": 3.560582160949707,
      "step": 1238
    },
    {
      "epoch": 0.17061520080537712,
      "grad_norm": 0.11297784745693207,
      "learning_rate": 0.002,
      "loss": 3.5433597564697266,
      "step": 1240
    },
    {
      "epoch": 0.17089038661312775,
      "grad_norm": 0.11743848770856857,
      "learning_rate": 0.002,
      "loss": 3.512444496154785,
      "step": 1242
    },
    {
      "epoch": 0.17116557242087835,
      "grad_norm": 0.10496261715888977,
      "learning_rate": 0.002,
      "loss": 3.493410110473633,
      "step": 1244
    },
    {
      "epoch": 0.17144075822862895,
      "grad_norm": 0.10611592233181,
      "learning_rate": 0.002,
      "loss": 3.5291147232055664,
      "step": 1246
    },
    {
      "epoch": 0.17171594403637958,
      "grad_norm": 0.14262360334396362,
      "learning_rate": 0.002,
      "loss": 3.4741320610046387,
      "step": 1248
    },
    {
      "epoch": 0.17199112984413018,
      "grad_norm": 0.24626424908638,
      "learning_rate": 0.002,
      "loss": 3.5149269104003906,
      "step": 1250
    },
    {
      "epoch": 0.17226631565188077,
      "grad_norm": 0.1350671350955963,
      "learning_rate": 0.002,
      "loss": 3.551710605621338,
      "step": 1252
    },
    {
      "epoch": 0.1725415014596314,
      "grad_norm": 0.2355857640504837,
      "learning_rate": 0.002,
      "loss": 3.533849000930786,
      "step": 1254
    },
    {
      "epoch": 0.172816687267382,
      "grad_norm": 0.35363736748695374,
      "learning_rate": 0.002,
      "loss": 3.4792885780334473,
      "step": 1256
    },
    {
      "epoch": 0.1730918730751326,
      "grad_norm": 0.10337286442518234,
      "learning_rate": 0.002,
      "loss": 3.5114989280700684,
      "step": 1258
    },
    {
      "epoch": 0.1733670588828832,
      "grad_norm": 0.09399651736021042,
      "learning_rate": 0.002,
      "loss": 3.465043783187866,
      "step": 1260
    },
    {
      "epoch": 0.17364224469063383,
      "grad_norm": 0.08849848061800003,
      "learning_rate": 0.002,
      "loss": 3.4812049865722656,
      "step": 1262
    },
    {
      "epoch": 0.17391743049838443,
      "grad_norm": 0.07701487839221954,
      "learning_rate": 0.002,
      "loss": 3.503012180328369,
      "step": 1264
    },
    {
      "epoch": 0.17419261630613503,
      "grad_norm": 0.10815402865409851,
      "learning_rate": 0.002,
      "loss": 3.5235085487365723,
      "step": 1266
    },
    {
      "epoch": 0.17446780211388566,
      "grad_norm": 0.11272697895765305,
      "learning_rate": 0.002,
      "loss": 3.5226614475250244,
      "step": 1268
    },
    {
      "epoch": 0.17474298792163626,
      "grad_norm": 0.2088182270526886,
      "learning_rate": 0.002,
      "loss": 3.440427780151367,
      "step": 1270
    },
    {
      "epoch": 0.17501817372938686,
      "grad_norm": 0.43902587890625,
      "learning_rate": 0.002,
      "loss": 3.4992246627807617,
      "step": 1272
    },
    {
      "epoch": 0.17529335953713748,
      "grad_norm": 0.3623107373714447,
      "learning_rate": 0.002,
      "loss": 3.520562171936035,
      "step": 1274
    },
    {
      "epoch": 0.17556854534488808,
      "grad_norm": 0.5525084733963013,
      "learning_rate": 0.002,
      "loss": 3.565762996673584,
      "step": 1276
    },
    {
      "epoch": 0.17584373115263868,
      "grad_norm": 0.10574793815612793,
      "learning_rate": 0.002,
      "loss": 3.5101404190063477,
      "step": 1278
    },
    {
      "epoch": 0.17611891696038928,
      "grad_norm": 0.18381735682487488,
      "learning_rate": 0.002,
      "loss": 3.5203022956848145,
      "step": 1280
    },
    {
      "epoch": 0.1763941027681399,
      "grad_norm": 0.10731271654367447,
      "learning_rate": 0.002,
      "loss": 3.4877257347106934,
      "step": 1282
    },
    {
      "epoch": 0.1766692885758905,
      "grad_norm": 0.11104144155979156,
      "learning_rate": 0.002,
      "loss": 3.4803943634033203,
      "step": 1284
    },
    {
      "epoch": 0.1769444743836411,
      "grad_norm": 0.06467290222644806,
      "learning_rate": 0.002,
      "loss": 3.4578824043273926,
      "step": 1286
    },
    {
      "epoch": 0.17721966019139174,
      "grad_norm": 0.07283943146467209,
      "learning_rate": 0.002,
      "loss": 3.4953737258911133,
      "step": 1288
    },
    {
      "epoch": 0.17749484599914234,
      "grad_norm": 0.10624629259109497,
      "learning_rate": 0.002,
      "loss": 3.4654698371887207,
      "step": 1290
    },
    {
      "epoch": 0.17777003180689294,
      "grad_norm": 0.06409531831741333,
      "learning_rate": 0.002,
      "loss": 3.4108595848083496,
      "step": 1292
    },
    {
      "epoch": 0.17804521761464356,
      "grad_norm": 0.063442662358284,
      "learning_rate": 0.002,
      "loss": 3.4603383541107178,
      "step": 1294
    },
    {
      "epoch": 0.17832040342239416,
      "grad_norm": 0.09721831977367401,
      "learning_rate": 0.002,
      "loss": 3.4456787109375,
      "step": 1296
    },
    {
      "epoch": 0.17859558923014476,
      "grad_norm": 0.2665596604347229,
      "learning_rate": 0.002,
      "loss": 3.4719643592834473,
      "step": 1298
    },
    {
      "epoch": 0.1788707750378954,
      "grad_norm": 0.4105665683746338,
      "learning_rate": 0.002,
      "loss": 3.484092950820923,
      "step": 1300
    },
    {
      "epoch": 0.179145960845646,
      "grad_norm": 0.09327073395252228,
      "learning_rate": 0.002,
      "loss": 3.503678321838379,
      "step": 1302
    },
    {
      "epoch": 0.1794211466533966,
      "grad_norm": 0.1324431151151657,
      "learning_rate": 0.002,
      "loss": 3.4784867763519287,
      "step": 1304
    },
    {
      "epoch": 0.1796963324611472,
      "grad_norm": 0.09585768729448318,
      "learning_rate": 0.002,
      "loss": 3.452836036682129,
      "step": 1306
    },
    {
      "epoch": 0.17997151826889782,
      "grad_norm": 0.11056527495384216,
      "learning_rate": 0.002,
      "loss": 3.495598316192627,
      "step": 1308
    },
    {
      "epoch": 0.18024670407664842,
      "grad_norm": 0.11636508256196976,
      "learning_rate": 0.002,
      "loss": 3.467597484588623,
      "step": 1310
    },
    {
      "epoch": 0.18052188988439902,
      "grad_norm": 0.10934772342443466,
      "learning_rate": 0.002,
      "loss": 3.4966201782226562,
      "step": 1312
    },
    {
      "epoch": 0.18079707569214964,
      "grad_norm": 0.12734605371952057,
      "learning_rate": 0.002,
      "loss": 3.4567103385925293,
      "step": 1314
    },
    {
      "epoch": 0.18107226149990024,
      "grad_norm": 0.173899307847023,
      "learning_rate": 0.002,
      "loss": 3.4859063625335693,
      "step": 1316
    },
    {
      "epoch": 0.18134744730765084,
      "grad_norm": 0.18236899375915527,
      "learning_rate": 0.002,
      "loss": 3.4860668182373047,
      "step": 1318
    },
    {
      "epoch": 0.18162263311540147,
      "grad_norm": 0.2738397419452667,
      "learning_rate": 0.002,
      "loss": 3.465953826904297,
      "step": 1320
    },
    {
      "epoch": 0.18189781892315207,
      "grad_norm": 0.23161646723747253,
      "learning_rate": 0.002,
      "loss": 3.4938697814941406,
      "step": 1322
    },
    {
      "epoch": 0.18217300473090267,
      "grad_norm": 0.1281944364309311,
      "learning_rate": 0.002,
      "loss": 3.476853847503662,
      "step": 1324
    },
    {
      "epoch": 0.1824481905386533,
      "grad_norm": 0.14537914097309113,
      "learning_rate": 0.002,
      "loss": 3.443909168243408,
      "step": 1326
    },
    {
      "epoch": 0.1827233763464039,
      "grad_norm": 0.2090064138174057,
      "learning_rate": 0.002,
      "loss": 3.441394805908203,
      "step": 1328
    },
    {
      "epoch": 0.1829985621541545,
      "grad_norm": 0.12064865976572037,
      "learning_rate": 0.002,
      "loss": 3.477768898010254,
      "step": 1330
    },
    {
      "epoch": 0.1832737479619051,
      "grad_norm": 0.1064484715461731,
      "learning_rate": 0.002,
      "loss": 3.4894423484802246,
      "step": 1332
    },
    {
      "epoch": 0.18354893376965573,
      "grad_norm": 0.09534869343042374,
      "learning_rate": 0.002,
      "loss": 3.469210147857666,
      "step": 1334
    },
    {
      "epoch": 0.18382411957740633,
      "grad_norm": 0.07369991391897202,
      "learning_rate": 0.002,
      "loss": 3.455972671508789,
      "step": 1336
    },
    {
      "epoch": 0.18409930538515692,
      "grad_norm": 0.095199815928936,
      "learning_rate": 0.002,
      "loss": 3.467724323272705,
      "step": 1338
    },
    {
      "epoch": 0.18437449119290755,
      "grad_norm": 0.05792967230081558,
      "learning_rate": 0.002,
      "loss": 3.39170241355896,
      "step": 1340
    },
    {
      "epoch": 0.18464967700065815,
      "grad_norm": 0.06182250380516052,
      "learning_rate": 0.002,
      "loss": 3.454998016357422,
      "step": 1342
    },
    {
      "epoch": 0.18492486280840875,
      "grad_norm": 0.091793954372406,
      "learning_rate": 0.002,
      "loss": 3.388129949569702,
      "step": 1344
    },
    {
      "epoch": 0.18520004861615938,
      "grad_norm": 0.136644184589386,
      "learning_rate": 0.002,
      "loss": 3.468214511871338,
      "step": 1346
    },
    {
      "epoch": 0.18547523442390998,
      "grad_norm": 0.32831433415412903,
      "learning_rate": 0.002,
      "loss": 3.4225945472717285,
      "step": 1348
    },
    {
      "epoch": 0.18575042023166058,
      "grad_norm": 0.6453784108161926,
      "learning_rate": 0.002,
      "loss": 3.522280693054199,
      "step": 1350
    },
    {
      "epoch": 0.1860256060394112,
      "grad_norm": 0.1407242864370346,
      "learning_rate": 0.002,
      "loss": 3.447908878326416,
      "step": 1352
    },
    {
      "epoch": 0.1863007918471618,
      "grad_norm": 0.20061640441417694,
      "learning_rate": 0.002,
      "loss": 3.433323621749878,
      "step": 1354
    },
    {
      "epoch": 0.1865759776549124,
      "grad_norm": 0.07665834575891495,
      "learning_rate": 0.002,
      "loss": 3.4330873489379883,
      "step": 1356
    },
    {
      "epoch": 0.186851163462663,
      "grad_norm": 0.057976190000772476,
      "learning_rate": 0.002,
      "loss": 3.4125819206237793,
      "step": 1358
    },
    {
      "epoch": 0.18712634927041363,
      "grad_norm": 0.05429922789335251,
      "learning_rate": 0.002,
      "loss": 3.410552501678467,
      "step": 1360
    },
    {
      "epoch": 0.18740153507816423,
      "grad_norm": 0.06801348924636841,
      "learning_rate": 0.002,
      "loss": 3.414659023284912,
      "step": 1362
    },
    {
      "epoch": 0.18767672088591483,
      "grad_norm": 0.057471733540296555,
      "learning_rate": 0.002,
      "loss": 3.441664695739746,
      "step": 1364
    },
    {
      "epoch": 0.18795190669366546,
      "grad_norm": 0.08019419759511948,
      "learning_rate": 0.002,
      "loss": 3.390740394592285,
      "step": 1366
    },
    {
      "epoch": 0.18822709250141606,
      "grad_norm": 0.06514187902212143,
      "learning_rate": 0.002,
      "loss": 3.4281556606292725,
      "step": 1368
    },
    {
      "epoch": 0.18850227830916666,
      "grad_norm": 0.08074332028627396,
      "learning_rate": 0.002,
      "loss": 3.432084560394287,
      "step": 1370
    },
    {
      "epoch": 0.1887774641169173,
      "grad_norm": 0.0769992396235466,
      "learning_rate": 0.002,
      "loss": 3.416470527648926,
      "step": 1372
    },
    {
      "epoch": 0.1890526499246679,
      "grad_norm": 0.22504931688308716,
      "learning_rate": 0.002,
      "loss": 3.4701597690582275,
      "step": 1374
    },
    {
      "epoch": 0.1893278357324185,
      "grad_norm": 0.3720528483390808,
      "learning_rate": 0.002,
      "loss": 3.4140267372131348,
      "step": 1376
    },
    {
      "epoch": 0.18960302154016911,
      "grad_norm": 0.28588995337486267,
      "learning_rate": 0.002,
      "loss": 3.4101815223693848,
      "step": 1378
    },
    {
      "epoch": 0.18987820734791971,
      "grad_norm": 0.07436482608318329,
      "learning_rate": 0.002,
      "loss": 3.449413776397705,
      "step": 1380
    },
    {
      "epoch": 0.1901533931556703,
      "grad_norm": 0.12414494156837463,
      "learning_rate": 0.002,
      "loss": 3.4246480464935303,
      "step": 1382
    },
    {
      "epoch": 0.1904285789634209,
      "grad_norm": 0.060708995908498764,
      "learning_rate": 0.002,
      "loss": 3.384298801422119,
      "step": 1384
    },
    {
      "epoch": 0.19070376477117154,
      "grad_norm": 0.07749102264642715,
      "learning_rate": 0.002,
      "loss": 3.4186620712280273,
      "step": 1386
    },
    {
      "epoch": 0.19097895057892214,
      "grad_norm": 0.10543335974216461,
      "learning_rate": 0.002,
      "loss": 3.4581758975982666,
      "step": 1388
    },
    {
      "epoch": 0.19125413638667274,
      "grad_norm": 0.12143590301275253,
      "learning_rate": 0.002,
      "loss": 3.464442253112793,
      "step": 1390
    },
    {
      "epoch": 0.19152932219442337,
      "grad_norm": 0.12265697121620178,
      "learning_rate": 0.002,
      "loss": 3.4577956199645996,
      "step": 1392
    },
    {
      "epoch": 0.19180450800217397,
      "grad_norm": 0.11212204396724701,
      "learning_rate": 0.002,
      "loss": 3.423067092895508,
      "step": 1394
    },
    {
      "epoch": 0.19207969380992457,
      "grad_norm": 0.06503834575414658,
      "learning_rate": 0.002,
      "loss": 3.429342269897461,
      "step": 1396
    },
    {
      "epoch": 0.1923548796176752,
      "grad_norm": 0.08931032568216324,
      "learning_rate": 0.002,
      "loss": 3.4308347702026367,
      "step": 1398
    },
    {
      "epoch": 0.1926300654254258,
      "grad_norm": 0.09227101504802704,
      "learning_rate": 0.002,
      "loss": 3.410374164581299,
      "step": 1400
    },
    {
      "epoch": 0.1929052512331764,
      "grad_norm": 0.12366588413715363,
      "learning_rate": 0.002,
      "loss": 3.3796114921569824,
      "step": 1402
    },
    {
      "epoch": 0.19318043704092702,
      "grad_norm": 0.2027212679386139,
      "learning_rate": 0.002,
      "loss": 3.433619499206543,
      "step": 1404
    },
    {
      "epoch": 0.19345562284867762,
      "grad_norm": 0.4945548474788666,
      "learning_rate": 0.002,
      "loss": 3.446545124053955,
      "step": 1406
    },
    {
      "epoch": 0.19373080865642822,
      "grad_norm": 0.14716897904872894,
      "learning_rate": 0.002,
      "loss": 3.420773983001709,
      "step": 1408
    },
    {
      "epoch": 0.19400599446417882,
      "grad_norm": 0.1160804033279419,
      "learning_rate": 0.002,
      "loss": 3.4335079193115234,
      "step": 1410
    },
    {
      "epoch": 0.19428118027192945,
      "grad_norm": 0.07386626303195953,
      "learning_rate": 0.002,
      "loss": 3.412898540496826,
      "step": 1412
    },
    {
      "epoch": 0.19455636607968005,
      "grad_norm": 0.22946570813655853,
      "learning_rate": 0.002,
      "loss": 3.455470561981201,
      "step": 1414
    },
    {
      "epoch": 0.19483155188743065,
      "grad_norm": 0.10113082081079483,
      "learning_rate": 0.002,
      "loss": 3.364804744720459,
      "step": 1416
    },
    {
      "epoch": 0.19510673769518128,
      "grad_norm": 0.05531404912471771,
      "learning_rate": 0.002,
      "loss": 3.4378273487091064,
      "step": 1418
    },
    {
      "epoch": 0.19538192350293188,
      "grad_norm": 0.07064833492040634,
      "learning_rate": 0.002,
      "loss": 3.394867420196533,
      "step": 1420
    },
    {
      "epoch": 0.19565710931068248,
      "grad_norm": 0.07080385833978653,
      "learning_rate": 0.002,
      "loss": 3.3610317707061768,
      "step": 1422
    },
    {
      "epoch": 0.1959322951184331,
      "grad_norm": 0.1578567922115326,
      "learning_rate": 0.002,
      "loss": 3.426797866821289,
      "step": 1424
    },
    {
      "epoch": 0.1962074809261837,
      "grad_norm": 0.18544192612171173,
      "learning_rate": 0.002,
      "loss": 3.3697752952575684,
      "step": 1426
    },
    {
      "epoch": 0.1964826667339343,
      "grad_norm": 0.16461433470249176,
      "learning_rate": 0.002,
      "loss": 3.3718996047973633,
      "step": 1428
    },
    {
      "epoch": 0.19675785254168493,
      "grad_norm": 0.046760838478803635,
      "learning_rate": 0.002,
      "loss": 3.406404733657837,
      "step": 1430
    },
    {
      "epoch": 0.19703303834943553,
      "grad_norm": 0.1031666100025177,
      "learning_rate": 0.002,
      "loss": 3.4025752544403076,
      "step": 1432
    },
    {
      "epoch": 0.19730822415718613,
      "grad_norm": 0.10020844638347626,
      "learning_rate": 0.002,
      "loss": 3.4315381050109863,
      "step": 1434
    },
    {
      "epoch": 0.19758340996493673,
      "grad_norm": 0.2763330042362213,
      "learning_rate": 0.002,
      "loss": 3.389645576477051,
      "step": 1436
    },
    {
      "epoch": 0.19785859577268736,
      "grad_norm": 0.6699429750442505,
      "learning_rate": 0.002,
      "loss": 3.405376434326172,
      "step": 1438
    },
    {
      "epoch": 0.19813378158043796,
      "grad_norm": 0.09036489576101303,
      "learning_rate": 0.002,
      "loss": 3.4152326583862305,
      "step": 1440
    },
    {
      "epoch": 0.19840896738818856,
      "grad_norm": 0.3120998442173004,
      "learning_rate": 0.002,
      "loss": 3.407344102859497,
      "step": 1442
    },
    {
      "epoch": 0.19868415319593918,
      "grad_norm": 0.21600286662578583,
      "learning_rate": 0.002,
      "loss": 3.381465196609497,
      "step": 1444
    },
    {
      "epoch": 0.19895933900368978,
      "grad_norm": 0.09724187105894089,
      "learning_rate": 0.002,
      "loss": 3.4296836853027344,
      "step": 1446
    },
    {
      "epoch": 0.19923452481144038,
      "grad_norm": 0.06237368658185005,
      "learning_rate": 0.002,
      "loss": 3.368934392929077,
      "step": 1448
    },
    {
      "epoch": 0.199509710619191,
      "grad_norm": 0.06262972950935364,
      "learning_rate": 0.002,
      "loss": 3.4324419498443604,
      "step": 1450
    },
    {
      "epoch": 0.1997848964269416,
      "grad_norm": 0.05676635727286339,
      "learning_rate": 0.002,
      "loss": 3.344668388366699,
      "step": 1452
    },
    {
      "epoch": 0.2000600822346922,
      "grad_norm": 0.06960613280534744,
      "learning_rate": 0.002,
      "loss": 3.370175838470459,
      "step": 1454
    },
    {
      "epoch": 0.20033526804244284,
      "grad_norm": 0.14308519661426544,
      "learning_rate": 0.002,
      "loss": 3.420433282852173,
      "step": 1456
    },
    {
      "epoch": 0.20061045385019344,
      "grad_norm": 0.05311449617147446,
      "learning_rate": 0.002,
      "loss": 3.404411792755127,
      "step": 1458
    },
    {
      "epoch": 0.20088563965794404,
      "grad_norm": 0.05297238752245903,
      "learning_rate": 0.002,
      "loss": 3.3750147819519043,
      "step": 1460
    },
    {
      "epoch": 0.20116082546569464,
      "grad_norm": 0.061209164559841156,
      "learning_rate": 0.002,
      "loss": 3.349984645843506,
      "step": 1462
    },
    {
      "epoch": 0.20143601127344526,
      "grad_norm": 0.0846041738986969,
      "learning_rate": 0.002,
      "loss": 3.395935535430908,
      "step": 1464
    },
    {
      "epoch": 0.20171119708119586,
      "grad_norm": 0.1720387488603592,
      "learning_rate": 0.002,
      "loss": 3.363253593444824,
      "step": 1466
    },
    {
      "epoch": 0.20198638288894646,
      "grad_norm": 0.27552738785743713,
      "learning_rate": 0.002,
      "loss": 3.351912021636963,
      "step": 1468
    },
    {
      "epoch": 0.2022615686966971,
      "grad_norm": 0.34960833191871643,
      "learning_rate": 0.002,
      "loss": 3.3971829414367676,
      "step": 1470
    },
    {
      "epoch": 0.2025367545044477,
      "grad_norm": 0.0716126412153244,
      "learning_rate": 0.002,
      "loss": 3.360199213027954,
      "step": 1472
    },
    {
      "epoch": 0.2028119403121983,
      "grad_norm": 0.08068204671144485,
      "learning_rate": 0.002,
      "loss": 3.3938069343566895,
      "step": 1474
    },
    {
      "epoch": 0.20308712611994892,
      "grad_norm": 0.0591827929019928,
      "learning_rate": 0.002,
      "loss": 3.370391368865967,
      "step": 1476
    },
    {
      "epoch": 0.20336231192769952,
      "grad_norm": 0.07285415381193161,
      "learning_rate": 0.002,
      "loss": 3.3845295906066895,
      "step": 1478
    },
    {
      "epoch": 0.20363749773545012,
      "grad_norm": 0.0969148501753807,
      "learning_rate": 0.002,
      "loss": 3.3314270973205566,
      "step": 1480
    },
    {
      "epoch": 0.20391268354320075,
      "grad_norm": 0.15086357295513153,
      "learning_rate": 0.002,
      "loss": 3.3815183639526367,
      "step": 1482
    },
    {
      "epoch": 0.20418786935095135,
      "grad_norm": 0.387690931558609,
      "learning_rate": 0.002,
      "loss": 3.4200282096862793,
      "step": 1484
    },
    {
      "epoch": 0.20446305515870195,
      "grad_norm": 0.17175152897834778,
      "learning_rate": 0.002,
      "loss": 3.3776025772094727,
      "step": 1486
    },
    {
      "epoch": 0.20473824096645254,
      "grad_norm": 0.17976950109004974,
      "learning_rate": 0.002,
      "loss": 3.350940704345703,
      "step": 1488
    },
    {
      "epoch": 0.20501342677420317,
      "grad_norm": 0.11202415078878403,
      "learning_rate": 0.002,
      "loss": 3.3744633197784424,
      "step": 1490
    },
    {
      "epoch": 0.20528861258195377,
      "grad_norm": 0.05705700442194939,
      "learning_rate": 0.002,
      "loss": 3.3581790924072266,
      "step": 1492
    },
    {
      "epoch": 0.20556379838970437,
      "grad_norm": 0.05645012855529785,
      "learning_rate": 0.002,
      "loss": 3.348386287689209,
      "step": 1494
    },
    {
      "epoch": 0.205838984197455,
      "grad_norm": 0.18236154317855835,
      "learning_rate": 0.002,
      "loss": 3.3658204078674316,
      "step": 1496
    },
    {
      "epoch": 0.2061141700052056,
      "grad_norm": 0.23495613038539886,
      "learning_rate": 0.002,
      "loss": 3.3475747108459473,
      "step": 1498
    },
    {
      "epoch": 0.2063893558129562,
      "grad_norm": 0.24208851158618927,
      "learning_rate": 0.002,
      "loss": 3.371781826019287,
      "step": 1500
    },
    {
      "epoch": 0.20666454162070683,
      "grad_norm": 0.24596337974071503,
      "learning_rate": 0.002,
      "loss": 3.3450725078582764,
      "step": 1502
    },
    {
      "epoch": 0.20693972742845743,
      "grad_norm": 0.13268467783927917,
      "learning_rate": 0.002,
      "loss": 3.348442554473877,
      "step": 1504
    },
    {
      "epoch": 0.20721491323620803,
      "grad_norm": 0.1306609958410263,
      "learning_rate": 0.002,
      "loss": 3.3634443283081055,
      "step": 1506
    },
    {
      "epoch": 0.20749009904395865,
      "grad_norm": 0.06283531337976456,
      "learning_rate": 0.002,
      "loss": 3.3036575317382812,
      "step": 1508
    },
    {
      "epoch": 0.20776528485170925,
      "grad_norm": 0.06599945574998856,
      "learning_rate": 0.002,
      "loss": 3.3886468410491943,
      "step": 1510
    },
    {
      "epoch": 0.20804047065945985,
      "grad_norm": 0.05974239110946655,
      "learning_rate": 0.002,
      "loss": 3.3431756496429443,
      "step": 1512
    },
    {
      "epoch": 0.20831565646721045,
      "grad_norm": 0.05453081429004669,
      "learning_rate": 0.002,
      "loss": 3.3646860122680664,
      "step": 1514
    },
    {
      "epoch": 0.20859084227496108,
      "grad_norm": 0.15309691429138184,
      "learning_rate": 0.002,
      "loss": 3.3377013206481934,
      "step": 1516
    },
    {
      "epoch": 0.20886602808271168,
      "grad_norm": 0.16293621063232422,
      "learning_rate": 0.002,
      "loss": 3.3842015266418457,
      "step": 1518
    },
    {
      "epoch": 0.20914121389046228,
      "grad_norm": 0.1433548927307129,
      "learning_rate": 0.002,
      "loss": 3.3673858642578125,
      "step": 1520
    },
    {
      "epoch": 0.2094163996982129,
      "grad_norm": 0.1625138819217682,
      "learning_rate": 0.002,
      "loss": 3.3722901344299316,
      "step": 1522
    },
    {
      "epoch": 0.2096915855059635,
      "grad_norm": 0.22670191526412964,
      "learning_rate": 0.002,
      "loss": 3.3375000953674316,
      "step": 1524
    },
    {
      "epoch": 0.2099667713137141,
      "grad_norm": 0.3131541311740875,
      "learning_rate": 0.002,
      "loss": 3.344764232635498,
      "step": 1526
    },
    {
      "epoch": 0.21024195712146473,
      "grad_norm": 0.3518402576446533,
      "learning_rate": 0.002,
      "loss": 3.3197786808013916,
      "step": 1528
    },
    {
      "epoch": 0.21051714292921533,
      "grad_norm": 0.06559180468320847,
      "learning_rate": 0.002,
      "loss": 3.3635997772216797,
      "step": 1530
    },
    {
      "epoch": 0.21079232873696593,
      "grad_norm": 0.1959219127893448,
      "learning_rate": 0.002,
      "loss": 3.3474135398864746,
      "step": 1532
    },
    {
      "epoch": 0.21106751454471653,
      "grad_norm": 0.07392096519470215,
      "learning_rate": 0.002,
      "loss": 3.307915210723877,
      "step": 1534
    },
    {
      "epoch": 0.21134270035246716,
      "grad_norm": 0.15558989346027374,
      "learning_rate": 0.002,
      "loss": 3.2933831214904785,
      "step": 1536
    },
    {
      "epoch": 0.21161788616021776,
      "grad_norm": 0.11437168717384338,
      "learning_rate": 0.002,
      "loss": 3.3160476684570312,
      "step": 1538
    },
    {
      "epoch": 0.21189307196796836,
      "grad_norm": 0.07336057722568512,
      "learning_rate": 0.002,
      "loss": 3.350870132446289,
      "step": 1540
    },
    {
      "epoch": 0.212168257775719,
      "grad_norm": 0.18846538662910461,
      "learning_rate": 0.002,
      "loss": 3.3757834434509277,
      "step": 1542
    },
    {
      "epoch": 0.2124434435834696,
      "grad_norm": 0.09070298820734024,
      "learning_rate": 0.002,
      "loss": 3.322791576385498,
      "step": 1544
    },
    {
      "epoch": 0.2127186293912202,
      "grad_norm": 0.06144991144537926,
      "learning_rate": 0.002,
      "loss": 3.3818893432617188,
      "step": 1546
    },
    {
      "epoch": 0.21299381519897082,
      "grad_norm": 0.10392835736274719,
      "learning_rate": 0.002,
      "loss": 3.348295211791992,
      "step": 1548
    },
    {
      "epoch": 0.21326900100672141,
      "grad_norm": 0.10746674984693527,
      "learning_rate": 0.002,
      "loss": 3.3614468574523926,
      "step": 1550
    },
    {
      "epoch": 0.21354418681447201,
      "grad_norm": 0.2610447406768799,
      "learning_rate": 0.002,
      "loss": 3.339446544647217,
      "step": 1552
    },
    {
      "epoch": 0.21381937262222264,
      "grad_norm": 0.38080957531929016,
      "learning_rate": 0.002,
      "loss": 3.349900245666504,
      "step": 1554
    },
    {
      "epoch": 0.21409455842997324,
      "grad_norm": 0.10447186231613159,
      "learning_rate": 0.002,
      "loss": 3.3526456356048584,
      "step": 1556
    },
    {
      "epoch": 0.21436974423772384,
      "grad_norm": 0.09259314835071564,
      "learning_rate": 0.002,
      "loss": 3.3124070167541504,
      "step": 1558
    },
    {
      "epoch": 0.21464493004547444,
      "grad_norm": 0.06719797104597092,
      "learning_rate": 0.002,
      "loss": 3.3240182399749756,
      "step": 1560
    },
    {
      "epoch": 0.21492011585322507,
      "grad_norm": 0.07398545742034912,
      "learning_rate": 0.002,
      "loss": 3.29685640335083,
      "step": 1562
    },
    {
      "epoch": 0.21519530166097567,
      "grad_norm": 0.050749052315950394,
      "learning_rate": 0.002,
      "loss": 3.3185858726501465,
      "step": 1564
    },
    {
      "epoch": 0.21547048746872627,
      "grad_norm": 0.07416415214538574,
      "learning_rate": 0.002,
      "loss": 3.314452648162842,
      "step": 1566
    },
    {
      "epoch": 0.2157456732764769,
      "grad_norm": 0.3375581204891205,
      "learning_rate": 0.002,
      "loss": 3.2777695655822754,
      "step": 1568
    },
    {
      "epoch": 0.2160208590842275,
      "grad_norm": 0.5405097007751465,
      "learning_rate": 0.002,
      "loss": 3.305365562438965,
      "step": 1570
    },
    {
      "epoch": 0.2162960448919781,
      "grad_norm": 0.10453715920448303,
      "learning_rate": 0.002,
      "loss": 3.298741102218628,
      "step": 1572
    },
    {
      "epoch": 0.21657123069972872,
      "grad_norm": 0.2142341434955597,
      "learning_rate": 0.002,
      "loss": 3.3640575408935547,
      "step": 1574
    },
    {
      "epoch": 0.21684641650747932,
      "grad_norm": 0.051744330674409866,
      "learning_rate": 0.002,
      "loss": 3.3603360652923584,
      "step": 1576
    },
    {
      "epoch": 0.21712160231522992,
      "grad_norm": 0.1685178428888321,
      "learning_rate": 0.002,
      "loss": 3.2798142433166504,
      "step": 1578
    },
    {
      "epoch": 0.21739678812298055,
      "grad_norm": 0.09637490659952164,
      "learning_rate": 0.002,
      "loss": 3.3434197902679443,
      "step": 1580
    },
    {
      "epoch": 0.21767197393073115,
      "grad_norm": 0.15397463738918304,
      "learning_rate": 0.002,
      "loss": 3.3233299255371094,
      "step": 1582
    },
    {
      "epoch": 0.21794715973848175,
      "grad_norm": 0.056552451103925705,
      "learning_rate": 0.002,
      "loss": 3.3478846549987793,
      "step": 1584
    },
    {
      "epoch": 0.21822234554623235,
      "grad_norm": 0.11164233088493347,
      "learning_rate": 0.002,
      "loss": 3.3255019187927246,
      "step": 1586
    },
    {
      "epoch": 0.21849753135398298,
      "grad_norm": 0.049393102526664734,
      "learning_rate": 0.002,
      "loss": 3.2842812538146973,
      "step": 1588
    },
    {
      "epoch": 0.21877271716173358,
      "grad_norm": 0.08273810893297195,
      "learning_rate": 0.002,
      "loss": 3.3157100677490234,
      "step": 1590
    },
    {
      "epoch": 0.21904790296948418,
      "grad_norm": 0.051240917295217514,
      "learning_rate": 0.002,
      "loss": 3.294919013977051,
      "step": 1592
    },
    {
      "epoch": 0.2193230887772348,
      "grad_norm": 0.14862634241580963,
      "learning_rate": 0.002,
      "loss": 3.286875009536743,
      "step": 1594
    },
    {
      "epoch": 0.2195982745849854,
      "grad_norm": 0.34521734714508057,
      "learning_rate": 0.002,
      "loss": 3.275292158126831,
      "step": 1596
    },
    {
      "epoch": 0.219873460392736,
      "grad_norm": 0.33528777956962585,
      "learning_rate": 0.002,
      "loss": 3.319859504699707,
      "step": 1598
    },
    {
      "epoch": 0.22014864620048663,
      "grad_norm": 0.071530282497406,
      "learning_rate": 0.002,
      "loss": 3.309084415435791,
      "step": 1600
    },
    {
      "epoch": 0.22042383200823723,
      "grad_norm": 0.06073305755853653,
      "learning_rate": 0.002,
      "loss": 3.3559422492980957,
      "step": 1602
    },
    {
      "epoch": 0.22069901781598783,
      "grad_norm": 0.05295354127883911,
      "learning_rate": 0.002,
      "loss": 3.295863389968872,
      "step": 1604
    },
    {
      "epoch": 0.22097420362373846,
      "grad_norm": 0.06194473057985306,
      "learning_rate": 0.002,
      "loss": 3.294739007949829,
      "step": 1606
    },
    {
      "epoch": 0.22124938943148906,
      "grad_norm": 0.053373679518699646,
      "learning_rate": 0.002,
      "loss": 3.2668581008911133,
      "step": 1608
    },
    {
      "epoch": 0.22152457523923966,
      "grad_norm": 0.1422806680202484,
      "learning_rate": 0.002,
      "loss": 3.3027353286743164,
      "step": 1610
    },
    {
      "epoch": 0.22179976104699026,
      "grad_norm": 0.2091410607099533,
      "learning_rate": 0.002,
      "loss": 3.2850239276885986,
      "step": 1612
    },
    {
      "epoch": 0.22207494685474088,
      "grad_norm": 0.27296796441078186,
      "learning_rate": 0.002,
      "loss": 3.3227362632751465,
      "step": 1614
    },
    {
      "epoch": 0.22235013266249148,
      "grad_norm": 0.28295210003852844,
      "learning_rate": 0.002,
      "loss": 3.341137409210205,
      "step": 1616
    },
    {
      "epoch": 0.22262531847024208,
      "grad_norm": 0.055836331099271774,
      "learning_rate": 0.002,
      "loss": 3.274357795715332,
      "step": 1618
    },
    {
      "epoch": 0.2229005042779927,
      "grad_norm": 0.28925955295562744,
      "learning_rate": 0.002,
      "loss": 3.3111562728881836,
      "step": 1620
    },
    {
      "epoch": 0.2231756900857433,
      "grad_norm": 0.15126672387123108,
      "learning_rate": 0.002,
      "loss": 3.310713291168213,
      "step": 1622
    },
    {
      "epoch": 0.2234508758934939,
      "grad_norm": 0.07159405201673508,
      "learning_rate": 0.002,
      "loss": 3.3269970417022705,
      "step": 1624
    },
    {
      "epoch": 0.22372606170124454,
      "grad_norm": 0.0713730901479721,
      "learning_rate": 0.002,
      "loss": 3.255680799484253,
      "step": 1626
    },
    {
      "epoch": 0.22400124750899514,
      "grad_norm": 0.13066396117210388,
      "learning_rate": 0.002,
      "loss": 3.283635139465332,
      "step": 1628
    },
    {
      "epoch": 0.22427643331674574,
      "grad_norm": 0.19248764216899872,
      "learning_rate": 0.002,
      "loss": 3.273108720779419,
      "step": 1630
    },
    {
      "epoch": 0.22455161912449637,
      "grad_norm": 0.108396016061306,
      "learning_rate": 0.002,
      "loss": 3.242439031600952,
      "step": 1632
    },
    {
      "epoch": 0.22482680493224697,
      "grad_norm": 0.1784796416759491,
      "learning_rate": 0.002,
      "loss": 3.298644542694092,
      "step": 1634
    },
    {
      "epoch": 0.22510199073999757,
      "grad_norm": 0.18611541390419006,
      "learning_rate": 0.002,
      "loss": 3.3009541034698486,
      "step": 1636
    },
    {
      "epoch": 0.22537717654774816,
      "grad_norm": 0.06508322060108185,
      "learning_rate": 0.002,
      "loss": 3.322195529937744,
      "step": 1638
    },
    {
      "epoch": 0.2256523623554988,
      "grad_norm": 0.0728175938129425,
      "learning_rate": 0.002,
      "loss": 3.2993335723876953,
      "step": 1640
    },
    {
      "epoch": 0.2259275481632494,
      "grad_norm": 0.21938741207122803,
      "learning_rate": 0.002,
      "loss": 3.312265396118164,
      "step": 1642
    },
    {
      "epoch": 0.226202733971,
      "grad_norm": 0.1462758630514145,
      "learning_rate": 0.002,
      "loss": 3.269096851348877,
      "step": 1644
    },
    {
      "epoch": 0.22647791977875062,
      "grad_norm": 0.15126970410346985,
      "learning_rate": 0.002,
      "loss": 3.3186068534851074,
      "step": 1646
    },
    {
      "epoch": 0.22675310558650122,
      "grad_norm": 0.22091121971607208,
      "learning_rate": 0.002,
      "loss": 3.3230767250061035,
      "step": 1648
    },
    {
      "epoch": 0.22702829139425182,
      "grad_norm": 0.14479000866413116,
      "learning_rate": 0.002,
      "loss": 3.296621322631836,
      "step": 1650
    },
    {
      "epoch": 0.22730347720200245,
      "grad_norm": 0.10686621069908142,
      "learning_rate": 0.002,
      "loss": 3.2553725242614746,
      "step": 1652
    },
    {
      "epoch": 0.22757866300975305,
      "grad_norm": 0.1515531986951828,
      "learning_rate": 0.002,
      "loss": 3.2685210704803467,
      "step": 1654
    },
    {
      "epoch": 0.22785384881750365,
      "grad_norm": 0.097958043217659,
      "learning_rate": 0.002,
      "loss": 3.279702663421631,
      "step": 1656
    },
    {
      "epoch": 0.22812903462525427,
      "grad_norm": 0.21302282810211182,
      "learning_rate": 0.002,
      "loss": 3.2516753673553467,
      "step": 1658
    },
    {
      "epoch": 0.22840422043300487,
      "grad_norm": 0.3977883756160736,
      "learning_rate": 0.002,
      "loss": 3.324972152709961,
      "step": 1660
    },
    {
      "epoch": 0.22867940624075547,
      "grad_norm": 0.33674320578575134,
      "learning_rate": 0.002,
      "loss": 3.2392892837524414,
      "step": 1662
    },
    {
      "epoch": 0.22895459204850607,
      "grad_norm": 0.22063398361206055,
      "learning_rate": 0.002,
      "loss": 3.3028669357299805,
      "step": 1664
    },
    {
      "epoch": 0.2292297778562567,
      "grad_norm": 0.09157482534646988,
      "learning_rate": 0.002,
      "loss": 3.280102491378784,
      "step": 1666
    },
    {
      "epoch": 0.2295049636640073,
      "grad_norm": 0.11160188168287277,
      "learning_rate": 0.002,
      "loss": 3.264538288116455,
      "step": 1668
    },
    {
      "epoch": 0.2297801494717579,
      "grad_norm": 0.0521952286362648,
      "learning_rate": 0.002,
      "loss": 3.237532615661621,
      "step": 1670
    },
    {
      "epoch": 0.23005533527950853,
      "grad_norm": 0.06775745749473572,
      "learning_rate": 0.002,
      "loss": 3.2575111389160156,
      "step": 1672
    },
    {
      "epoch": 0.23033052108725913,
      "grad_norm": 0.09935734421014786,
      "learning_rate": 0.002,
      "loss": 3.275717258453369,
      "step": 1674
    },
    {
      "epoch": 0.23060570689500973,
      "grad_norm": 0.04828578233718872,
      "learning_rate": 0.002,
      "loss": 3.3141427040100098,
      "step": 1676
    },
    {
      "epoch": 0.23088089270276035,
      "grad_norm": 0.053256966173648834,
      "learning_rate": 0.002,
      "loss": 3.2629010677337646,
      "step": 1678
    },
    {
      "epoch": 0.23115607851051095,
      "grad_norm": 0.1195879802107811,
      "learning_rate": 0.002,
      "loss": 3.26652193069458,
      "step": 1680
    },
    {
      "epoch": 0.23143126431826155,
      "grad_norm": 0.3817826509475708,
      "learning_rate": 0.002,
      "loss": 3.2711634635925293,
      "step": 1682
    },
    {
      "epoch": 0.23170645012601218,
      "grad_norm": 0.4392907917499542,
      "learning_rate": 0.002,
      "loss": 3.3092470169067383,
      "step": 1684
    },
    {
      "epoch": 0.23198163593376278,
      "grad_norm": 0.21520806849002838,
      "learning_rate": 0.002,
      "loss": 3.3012747764587402,
      "step": 1686
    },
    {
      "epoch": 0.23225682174151338,
      "grad_norm": 0.053666118532419205,
      "learning_rate": 0.002,
      "loss": 3.2607011795043945,
      "step": 1688
    },
    {
      "epoch": 0.23253200754926398,
      "grad_norm": 0.11952002346515656,
      "learning_rate": 0.002,
      "loss": 3.2398648262023926,
      "step": 1690
    },
    {
      "epoch": 0.2328071933570146,
      "grad_norm": 0.06949695199728012,
      "learning_rate": 0.002,
      "loss": 3.278073310852051,
      "step": 1692
    },
    {
      "epoch": 0.2330823791647652,
      "grad_norm": 0.07635536044836044,
      "learning_rate": 0.002,
      "loss": 3.260685920715332,
      "step": 1694
    },
    {
      "epoch": 0.2333575649725158,
      "grad_norm": 0.05747153237462044,
      "learning_rate": 0.002,
      "loss": 3.257840156555176,
      "step": 1696
    },
    {
      "epoch": 0.23363275078026643,
      "grad_norm": 0.1261298805475235,
      "learning_rate": 0.002,
      "loss": 3.263123035430908,
      "step": 1698
    },
    {
      "epoch": 0.23390793658801703,
      "grad_norm": 0.058672986924648285,
      "learning_rate": 0.002,
      "loss": 3.302393674850464,
      "step": 1700
    },
    {
      "epoch": 0.23418312239576763,
      "grad_norm": 0.2371913194656372,
      "learning_rate": 0.002,
      "loss": 3.302074909210205,
      "step": 1702
    },
    {
      "epoch": 0.23445830820351826,
      "grad_norm": 0.21627677977085114,
      "learning_rate": 0.002,
      "loss": 3.286879062652588,
      "step": 1704
    },
    {
      "epoch": 0.23473349401126886,
      "grad_norm": 0.26456326246261597,
      "learning_rate": 0.002,
      "loss": 3.267287254333496,
      "step": 1706
    },
    {
      "epoch": 0.23500867981901946,
      "grad_norm": 0.08768317103385925,
      "learning_rate": 0.002,
      "loss": 3.272706985473633,
      "step": 1708
    },
    {
      "epoch": 0.2352838656267701,
      "grad_norm": 0.06316641718149185,
      "learning_rate": 0.002,
      "loss": 3.266122341156006,
      "step": 1710
    },
    {
      "epoch": 0.2355590514345207,
      "grad_norm": 0.061427902430295944,
      "learning_rate": 0.002,
      "loss": 3.2061288356781006,
      "step": 1712
    },
    {
      "epoch": 0.2358342372422713,
      "grad_norm": 0.16371293365955353,
      "learning_rate": 0.002,
      "loss": 3.23630428314209,
      "step": 1714
    },
    {
      "epoch": 0.2361094230500219,
      "grad_norm": 0.22355104982852936,
      "learning_rate": 0.002,
      "loss": 3.2487235069274902,
      "step": 1716
    },
    {
      "epoch": 0.23638460885777252,
      "grad_norm": 0.21422825753688812,
      "learning_rate": 0.002,
      "loss": 3.2854630947113037,
      "step": 1718
    },
    {
      "epoch": 0.23665979466552312,
      "grad_norm": 0.19415952265262604,
      "learning_rate": 0.002,
      "loss": 3.250565767288208,
      "step": 1720
    },
    {
      "epoch": 0.23693498047327372,
      "grad_norm": 0.12682609260082245,
      "learning_rate": 0.002,
      "loss": 3.2417941093444824,
      "step": 1722
    },
    {
      "epoch": 0.23721016628102434,
      "grad_norm": 0.05280957743525505,
      "learning_rate": 0.002,
      "loss": 3.242882013320923,
      "step": 1724
    },
    {
      "epoch": 0.23748535208877494,
      "grad_norm": 0.06671430170536041,
      "learning_rate": 0.002,
      "loss": 3.2520108222961426,
      "step": 1726
    },
    {
      "epoch": 0.23776053789652554,
      "grad_norm": 0.06631242483854294,
      "learning_rate": 0.002,
      "loss": 3.2790756225585938,
      "step": 1728
    },
    {
      "epoch": 0.23803572370427617,
      "grad_norm": 0.1627863496541977,
      "learning_rate": 0.002,
      "loss": 3.2341506481170654,
      "step": 1730
    },
    {
      "epoch": 0.23831090951202677,
      "grad_norm": 0.29476040601730347,
      "learning_rate": 0.002,
      "loss": 3.313310384750366,
      "step": 1732
    },
    {
      "epoch": 0.23858609531977737,
      "grad_norm": 0.563186764717102,
      "learning_rate": 0.002,
      "loss": 3.2710602283477783,
      "step": 1734
    },
    {
      "epoch": 0.238861281127528,
      "grad_norm": 0.09720827639102936,
      "learning_rate": 0.002,
      "loss": 3.295616626739502,
      "step": 1736
    },
    {
      "epoch": 0.2391364669352786,
      "grad_norm": 0.3294048309326172,
      "learning_rate": 0.002,
      "loss": 3.2156476974487305,
      "step": 1738
    },
    {
      "epoch": 0.2394116527430292,
      "grad_norm": 0.19160617887973785,
      "learning_rate": 0.002,
      "loss": 3.2438387870788574,
      "step": 1740
    },
    {
      "epoch": 0.2396868385507798,
      "grad_norm": 0.0464240163564682,
      "learning_rate": 0.002,
      "loss": 3.268153190612793,
      "step": 1742
    },
    {
      "epoch": 0.23996202435853042,
      "grad_norm": 0.13557545840740204,
      "learning_rate": 0.002,
      "loss": 3.251685619354248,
      "step": 1744
    },
    {
      "epoch": 0.24023721016628102,
      "grad_norm": 0.08359313011169434,
      "learning_rate": 0.002,
      "loss": 3.24583101272583,
      "step": 1746
    },
    {
      "epoch": 0.24051239597403162,
      "grad_norm": 0.0827624648809433,
      "learning_rate": 0.002,
      "loss": 3.2550930976867676,
      "step": 1748
    },
    {
      "epoch": 0.24078758178178225,
      "grad_norm": 0.11472095549106598,
      "learning_rate": 0.002,
      "loss": 3.1803367137908936,
      "step": 1750
    },
    {
      "epoch": 0.24106276758953285,
      "grad_norm": 0.08166660368442535,
      "learning_rate": 0.002,
      "loss": 3.2406251430511475,
      "step": 1752
    },
    {
      "epoch": 0.24133795339728345,
      "grad_norm": 0.07245507091283798,
      "learning_rate": 0.002,
      "loss": 3.2762298583984375,
      "step": 1754
    },
    {
      "epoch": 0.24161313920503408,
      "grad_norm": 0.13962237536907196,
      "learning_rate": 0.002,
      "loss": 3.2402710914611816,
      "step": 1756
    },
    {
      "epoch": 0.24188832501278468,
      "grad_norm": 0.13799798488616943,
      "learning_rate": 0.002,
      "loss": 3.212770938873291,
      "step": 1758
    },
    {
      "epoch": 0.24216351082053528,
      "grad_norm": 0.24975141882896423,
      "learning_rate": 0.002,
      "loss": 3.227634906768799,
      "step": 1760
    },
    {
      "epoch": 0.24243869662828588,
      "grad_norm": 0.28674325346946716,
      "learning_rate": 0.002,
      "loss": 3.273547649383545,
      "step": 1762
    },
    {
      "epoch": 0.2427138824360365,
      "grad_norm": 0.05146781727671623,
      "learning_rate": 0.002,
      "loss": 3.1918277740478516,
      "step": 1764
    },
    {
      "epoch": 0.2429890682437871,
      "grad_norm": 0.18101565539836884,
      "learning_rate": 0.002,
      "loss": 3.266979932785034,
      "step": 1766
    },
    {
      "epoch": 0.2432642540515377,
      "grad_norm": 0.11454565078020096,
      "learning_rate": 0.002,
      "loss": 3.272458076477051,
      "step": 1768
    },
    {
      "epoch": 0.24353943985928833,
      "grad_norm": 0.06159938499331474,
      "learning_rate": 0.002,
      "loss": 3.2813258171081543,
      "step": 1770
    },
    {
      "epoch": 0.24381462566703893,
      "grad_norm": 0.13241791725158691,
      "learning_rate": 0.002,
      "loss": 3.2499685287475586,
      "step": 1772
    },
    {
      "epoch": 0.24408981147478953,
      "grad_norm": 0.10403008759021759,
      "learning_rate": 0.002,
      "loss": 3.2202677726745605,
      "step": 1774
    },
    {
      "epoch": 0.24436499728254016,
      "grad_norm": 0.09278176724910736,
      "learning_rate": 0.002,
      "loss": 3.2259907722473145,
      "step": 1776
    },
    {
      "epoch": 0.24464018309029076,
      "grad_norm": 0.2774653434753418,
      "learning_rate": 0.002,
      "loss": 3.2141261100769043,
      "step": 1778
    },
    {
      "epoch": 0.24491536889804136,
      "grad_norm": 0.29162752628326416,
      "learning_rate": 0.002,
      "loss": 3.2237372398376465,
      "step": 1780
    },
    {
      "epoch": 0.24519055470579199,
      "grad_norm": 0.056215085089206696,
      "learning_rate": 0.002,
      "loss": 3.252415180206299,
      "step": 1782
    },
    {
      "epoch": 0.24546574051354259,
      "grad_norm": 0.1322833150625229,
      "learning_rate": 0.002,
      "loss": 3.239326000213623,
      "step": 1784
    },
    {
      "epoch": 0.24574092632129318,
      "grad_norm": 0.1144147664308548,
      "learning_rate": 0.002,
      "loss": 3.2506093978881836,
      "step": 1786
    },
    {
      "epoch": 0.24601611212904378,
      "grad_norm": 0.09121379256248474,
      "learning_rate": 0.002,
      "loss": 3.1973371505737305,
      "step": 1788
    },
    {
      "epoch": 0.2462912979367944,
      "grad_norm": 0.07515296339988708,
      "learning_rate": 0.002,
      "loss": 3.2762677669525146,
      "step": 1790
    },
    {
      "epoch": 0.246566483744545,
      "grad_norm": 0.2206030786037445,
      "learning_rate": 0.002,
      "loss": 3.16424560546875,
      "step": 1792
    },
    {
      "epoch": 0.2468416695522956,
      "grad_norm": 0.2986094653606415,
      "learning_rate": 0.002,
      "loss": 3.2351956367492676,
      "step": 1794
    },
    {
      "epoch": 0.24711685536004624,
      "grad_norm": 0.29513049125671387,
      "learning_rate": 0.002,
      "loss": 3.257066249847412,
      "step": 1796
    },
    {
      "epoch": 0.24739204116779684,
      "grad_norm": 0.11348287761211395,
      "learning_rate": 0.002,
      "loss": 3.2058041095733643,
      "step": 1798
    },
    {
      "epoch": 0.24766722697554744,
      "grad_norm": 0.04735762998461723,
      "learning_rate": 0.002,
      "loss": 3.2090582847595215,
      "step": 1800
    },
    {
      "epoch": 0.24794241278329807,
      "grad_norm": 0.09363241493701935,
      "learning_rate": 0.002,
      "loss": 3.2816858291625977,
      "step": 1802
    },
    {
      "epoch": 0.24821759859104867,
      "grad_norm": 0.0729379653930664,
      "learning_rate": 0.002,
      "loss": 3.2436418533325195,
      "step": 1804
    },
    {
      "epoch": 0.24849278439879927,
      "grad_norm": 0.11345548927783966,
      "learning_rate": 0.002,
      "loss": 3.1400861740112305,
      "step": 1806
    },
    {
      "epoch": 0.2487679702065499,
      "grad_norm": 0.20031893253326416,
      "learning_rate": 0.002,
      "loss": 3.2165441513061523,
      "step": 1808
    },
    {
      "epoch": 0.2490431560143005,
      "grad_norm": 0.3339705765247345,
      "learning_rate": 0.002,
      "loss": 3.2066993713378906,
      "step": 1810
    },
    {
      "epoch": 0.2493183418220511,
      "grad_norm": 0.1793355941772461,
      "learning_rate": 0.002,
      "loss": 3.203003168106079,
      "step": 1812
    },
    {
      "epoch": 0.2495935276298017,
      "grad_norm": 0.1530846357345581,
      "learning_rate": 0.002,
      "loss": 3.2218849658966064,
      "step": 1814
    },
    {
      "epoch": 0.24986871343755232,
      "grad_norm": 0.2158678025007248,
      "learning_rate": 0.002,
      "loss": 3.243082046508789,
      "step": 1816
    },
    {
      "epoch": 0.25014389924530295,
      "grad_norm": 0.055402934551239014,
      "learning_rate": 0.002,
      "loss": 3.2087695598602295,
      "step": 1818
    },
    {
      "epoch": 0.25041908505305355,
      "grad_norm": 0.06294067203998566,
      "learning_rate": 0.002,
      "loss": 3.226872444152832,
      "step": 1820
    },
    {
      "epoch": 0.25069427086080415,
      "grad_norm": 0.1095491424202919,
      "learning_rate": 0.002,
      "loss": 3.194913625717163,
      "step": 1822
    },
    {
      "epoch": 0.25096945666855475,
      "grad_norm": 0.11658176779747009,
      "learning_rate": 0.002,
      "loss": 3.181403398513794,
      "step": 1824
    },
    {
      "epoch": 0.25124464247630535,
      "grad_norm": 0.11570718139410019,
      "learning_rate": 0.002,
      "loss": 3.2505412101745605,
      "step": 1826
    },
    {
      "epoch": 0.25151982828405595,
      "grad_norm": 0.1525447517633438,
      "learning_rate": 0.002,
      "loss": 3.2340800762176514,
      "step": 1828
    },
    {
      "epoch": 0.25179501409180655,
      "grad_norm": 0.06167489290237427,
      "learning_rate": 0.002,
      "loss": 3.235105037689209,
      "step": 1830
    },
    {
      "epoch": 0.2520701998995572,
      "grad_norm": 0.4804210662841797,
      "learning_rate": 0.002,
      "loss": 3.2319834232330322,
      "step": 1832
    },
    {
      "epoch": 0.2523453857073078,
      "grad_norm": 0.26886504888534546,
      "learning_rate": 0.002,
      "loss": 3.218724250793457,
      "step": 1834
    },
    {
      "epoch": 0.2526205715150584,
      "grad_norm": 0.31622833013534546,
      "learning_rate": 0.002,
      "loss": 3.1982038021087646,
      "step": 1836
    },
    {
      "epoch": 0.252895757322809,
      "grad_norm": 0.14302672445774078,
      "learning_rate": 0.002,
      "loss": 3.1742148399353027,
      "step": 1838
    },
    {
      "epoch": 0.2531709431305596,
      "grad_norm": 0.19828152656555176,
      "learning_rate": 0.002,
      "loss": 3.237978219985962,
      "step": 1840
    },
    {
      "epoch": 0.2534461289383102,
      "grad_norm": 0.06747227162122726,
      "learning_rate": 0.002,
      "loss": 3.2271041870117188,
      "step": 1842
    },
    {
      "epoch": 0.25372131474606086,
      "grad_norm": 0.16722413897514343,
      "learning_rate": 0.002,
      "loss": 3.2593836784362793,
      "step": 1844
    },
    {
      "epoch": 0.25399650055381146,
      "grad_norm": 0.06578987091779709,
      "learning_rate": 0.002,
      "loss": 3.1929898262023926,
      "step": 1846
    },
    {
      "epoch": 0.25427168636156205,
      "grad_norm": 0.10053300112485886,
      "learning_rate": 0.002,
      "loss": 3.1978673934936523,
      "step": 1848
    },
    {
      "epoch": 0.25454687216931265,
      "grad_norm": 0.04376651719212532,
      "learning_rate": 0.002,
      "loss": 3.213629961013794,
      "step": 1850
    },
    {
      "epoch": 0.25482205797706325,
      "grad_norm": 0.05656079575419426,
      "learning_rate": 0.002,
      "loss": 3.1705570220947266,
      "step": 1852
    },
    {
      "epoch": 0.25509724378481385,
      "grad_norm": 0.04303845763206482,
      "learning_rate": 0.002,
      "loss": 3.167907238006592,
      "step": 1854
    },
    {
      "epoch": 0.25537242959256445,
      "grad_norm": 0.17434543371200562,
      "learning_rate": 0.002,
      "loss": 3.2207417488098145,
      "step": 1856
    },
    {
      "epoch": 0.2556476154003151,
      "grad_norm": 0.15523068606853485,
      "learning_rate": 0.002,
      "loss": 3.2301642894744873,
      "step": 1858
    },
    {
      "epoch": 0.2559228012080657,
      "grad_norm": 0.15388701856136322,
      "learning_rate": 0.002,
      "loss": 3.217487335205078,
      "step": 1860
    },
    {
      "epoch": 0.2561979870158163,
      "grad_norm": 0.16042199730873108,
      "learning_rate": 0.002,
      "loss": 3.1571545600891113,
      "step": 1862
    },
    {
      "epoch": 0.2564731728235669,
      "grad_norm": 0.31065770983695984,
      "learning_rate": 0.002,
      "loss": 3.17991304397583,
      "step": 1864
    },
    {
      "epoch": 0.2567483586313175,
      "grad_norm": 0.3437763452529907,
      "learning_rate": 0.002,
      "loss": 3.1613893508911133,
      "step": 1866
    },
    {
      "epoch": 0.2570235444390681,
      "grad_norm": 0.053971629589796066,
      "learning_rate": 0.002,
      "loss": 3.1566429138183594,
      "step": 1868
    },
    {
      "epoch": 0.25729873024681876,
      "grad_norm": 0.3118561804294586,
      "learning_rate": 0.002,
      "loss": 3.1772756576538086,
      "step": 1870
    },
    {
      "epoch": 0.25757391605456936,
      "grad_norm": 0.13607153296470642,
      "learning_rate": 0.002,
      "loss": 3.233067512512207,
      "step": 1872
    },
    {
      "epoch": 0.25784910186231996,
      "grad_norm": 0.1817374974489212,
      "learning_rate": 0.002,
      "loss": 3.1843295097351074,
      "step": 1874
    },
    {
      "epoch": 0.25812428767007056,
      "grad_norm": 0.05380414426326752,
      "learning_rate": 0.002,
      "loss": 3.1847100257873535,
      "step": 1876
    },
    {
      "epoch": 0.25839947347782116,
      "grad_norm": 0.21412716805934906,
      "learning_rate": 0.002,
      "loss": 3.205430746078491,
      "step": 1878
    },
    {
      "epoch": 0.25867465928557176,
      "grad_norm": 0.1524515002965927,
      "learning_rate": 0.002,
      "loss": 3.2065720558166504,
      "step": 1880
    },
    {
      "epoch": 0.25894984509332236,
      "grad_norm": 0.10627499222755432,
      "learning_rate": 0.002,
      "loss": 3.1970033645629883,
      "step": 1882
    },
    {
      "epoch": 0.259225030901073,
      "grad_norm": 0.23585329949855804,
      "learning_rate": 0.002,
      "loss": 3.258258581161499,
      "step": 1884
    },
    {
      "epoch": 0.2595002167088236,
      "grad_norm": 0.04842812940478325,
      "learning_rate": 0.002,
      "loss": 3.1610026359558105,
      "step": 1886
    },
    {
      "epoch": 0.2597754025165742,
      "grad_norm": 0.15152667462825775,
      "learning_rate": 0.002,
      "loss": 3.214799165725708,
      "step": 1888
    },
    {
      "epoch": 0.2600505883243248,
      "grad_norm": 0.09184227138757706,
      "learning_rate": 0.002,
      "loss": 3.1533164978027344,
      "step": 1890
    },
    {
      "epoch": 0.2600505883243248,
      "eval_loss": 3.176182985305786,
      "eval_runtime": 380.6511,
      "eval_samples_per_second": 92.386,
      "eval_steps_per_second": 11.549,
      "step": 1890
    },
    {
      "epoch": 0.2603257741320754,
      "grad_norm": 0.12588734924793243,
      "learning_rate": 0.002,
      "loss": 3.2103652954101562,
      "step": 1892
    },
    {
      "epoch": 0.260600959939826,
      "grad_norm": 0.052816782146692276,
      "learning_rate": 0.002,
      "loss": 3.236208915710449,
      "step": 1894
    },
    {
      "epoch": 0.26087614574757667,
      "grad_norm": 0.06057298183441162,
      "learning_rate": 0.002,
      "loss": 3.198521137237549,
      "step": 1896
    },
    {
      "epoch": 0.26115133155532727,
      "grad_norm": 0.0537308044731617,
      "learning_rate": 0.002,
      "loss": 3.154092311859131,
      "step": 1898
    },
    {
      "epoch": 0.26142651736307787,
      "grad_norm": 0.09083525091409683,
      "learning_rate": 0.002,
      "loss": 3.1412501335144043,
      "step": 1900
    },
    {
      "epoch": 0.26170170317082847,
      "grad_norm": 0.20277053117752075,
      "learning_rate": 0.002,
      "loss": 3.1704025268554688,
      "step": 1902
    },
    {
      "epoch": 0.26197688897857907,
      "grad_norm": 0.4057079553604126,
      "learning_rate": 0.002,
      "loss": 3.1694445610046387,
      "step": 1904
    },
    {
      "epoch": 0.26225207478632967,
      "grad_norm": 0.21167045831680298,
      "learning_rate": 0.002,
      "loss": 3.2039852142333984,
      "step": 1906
    },
    {
      "epoch": 0.26252726059408027,
      "grad_norm": 0.2329169362783432,
      "learning_rate": 0.002,
      "loss": 3.1561477184295654,
      "step": 1908
    },
    {
      "epoch": 0.2628024464018309,
      "grad_norm": 0.043573543429374695,
      "learning_rate": 0.002,
      "loss": 3.2100045680999756,
      "step": 1910
    },
    {
      "epoch": 0.2630776322095815,
      "grad_norm": 0.13504497706890106,
      "learning_rate": 0.002,
      "loss": 3.1578776836395264,
      "step": 1912
    },
    {
      "epoch": 0.2633528180173321,
      "grad_norm": 0.059264298528432846,
      "learning_rate": 0.002,
      "loss": 3.1560213565826416,
      "step": 1914
    },
    {
      "epoch": 0.2636280038250827,
      "grad_norm": 0.08075001835823059,
      "learning_rate": 0.002,
      "loss": 3.180509328842163,
      "step": 1916
    },
    {
      "epoch": 0.2639031896328333,
      "grad_norm": 0.08264581859111786,
      "learning_rate": 0.002,
      "loss": 3.1637496948242188,
      "step": 1918
    },
    {
      "epoch": 0.2641783754405839,
      "grad_norm": 0.17173288762569427,
      "learning_rate": 0.002,
      "loss": 3.2128164768218994,
      "step": 1920
    },
    {
      "epoch": 0.2644535612483346,
      "grad_norm": 0.2465369552373886,
      "learning_rate": 0.002,
      "loss": 3.2187442779541016,
      "step": 1922
    },
    {
      "epoch": 0.2647287470560852,
      "grad_norm": 0.1616891324520111,
      "learning_rate": 0.002,
      "loss": 3.133366107940674,
      "step": 1924
    },
    {
      "epoch": 0.2650039328638358,
      "grad_norm": 0.05225484445691109,
      "learning_rate": 0.002,
      "loss": 3.142951011657715,
      "step": 1926
    },
    {
      "epoch": 0.2652791186715864,
      "grad_norm": 0.04410627856850624,
      "learning_rate": 0.002,
      "loss": 3.1473348140716553,
      "step": 1928
    },
    {
      "epoch": 0.265554304479337,
      "grad_norm": 0.1339377611875534,
      "learning_rate": 0.002,
      "loss": 3.204653739929199,
      "step": 1930
    },
    {
      "epoch": 0.2658294902870876,
      "grad_norm": 0.3294074833393097,
      "learning_rate": 0.002,
      "loss": 3.183940887451172,
      "step": 1932
    },
    {
      "epoch": 0.2661046760948382,
      "grad_norm": 0.18728768825531006,
      "learning_rate": 0.002,
      "loss": 3.184286117553711,
      "step": 1934
    },
    {
      "epoch": 0.26637986190258883,
      "grad_norm": 0.04516671597957611,
      "learning_rate": 0.002,
      "loss": 3.1691274642944336,
      "step": 1936
    },
    {
      "epoch": 0.26665504771033943,
      "grad_norm": 0.11506596952676773,
      "learning_rate": 0.002,
      "loss": 3.2045416831970215,
      "step": 1938
    },
    {
      "epoch": 0.26693023351809003,
      "grad_norm": 0.20102575421333313,
      "learning_rate": 0.002,
      "loss": 3.182945966720581,
      "step": 1940
    },
    {
      "epoch": 0.26720541932584063,
      "grad_norm": 0.20068085193634033,
      "learning_rate": 0.002,
      "loss": 3.157865047454834,
      "step": 1942
    },
    {
      "epoch": 0.26748060513359123,
      "grad_norm": 0.08790222555398941,
      "learning_rate": 0.002,
      "loss": 3.137359380722046,
      "step": 1944
    },
    {
      "epoch": 0.26775579094134183,
      "grad_norm": 0.25828883051872253,
      "learning_rate": 0.002,
      "loss": 3.21307635307312,
      "step": 1946
    },
    {
      "epoch": 0.2680309767490925,
      "grad_norm": 0.3192070424556732,
      "learning_rate": 0.002,
      "loss": 3.1206483840942383,
      "step": 1948
    },
    {
      "epoch": 0.2683061625568431,
      "grad_norm": 0.11474697291851044,
      "learning_rate": 0.002,
      "loss": 3.160757064819336,
      "step": 1950
    },
    {
      "epoch": 0.2685813483645937,
      "grad_norm": 0.053839366883039474,
      "learning_rate": 0.002,
      "loss": 3.186373710632324,
      "step": 1952
    },
    {
      "epoch": 0.2688565341723443,
      "grad_norm": 0.2871586084365845,
      "learning_rate": 0.002,
      "loss": 3.1257455348968506,
      "step": 1954
    },
    {
      "epoch": 0.2691317199800949,
      "grad_norm": 0.20745187997817993,
      "learning_rate": 0.002,
      "loss": 3.1556687355041504,
      "step": 1956
    },
    {
      "epoch": 0.2694069057878455,
      "grad_norm": 0.31576260924339294,
      "learning_rate": 0.002,
      "loss": 3.1860122680664062,
      "step": 1958
    },
    {
      "epoch": 0.2696820915955961,
      "grad_norm": 0.04596790298819542,
      "learning_rate": 0.002,
      "loss": 3.1565752029418945,
      "step": 1960
    },
    {
      "epoch": 0.26995727740334674,
      "grad_norm": 0.09026051312685013,
      "learning_rate": 0.002,
      "loss": 3.1956284046173096,
      "step": 1962
    },
    {
      "epoch": 0.27023246321109734,
      "grad_norm": 0.17145048081874847,
      "learning_rate": 0.002,
      "loss": 3.1693029403686523,
      "step": 1964
    },
    {
      "epoch": 0.27050764901884794,
      "grad_norm": 0.12311998009681702,
      "learning_rate": 0.002,
      "loss": 3.192366361618042,
      "step": 1966
    },
    {
      "epoch": 0.27078283482659854,
      "grad_norm": 0.039950285106897354,
      "learning_rate": 0.002,
      "loss": 3.162144660949707,
      "step": 1968
    },
    {
      "epoch": 0.27105802063434914,
      "grad_norm": 0.06406160444021225,
      "learning_rate": 0.002,
      "loss": 3.1663503646850586,
      "step": 1970
    },
    {
      "epoch": 0.27133320644209974,
      "grad_norm": 0.07727673649787903,
      "learning_rate": 0.002,
      "loss": 3.211007595062256,
      "step": 1972
    },
    {
      "epoch": 0.2716083922498504,
      "grad_norm": 0.2037239968776703,
      "learning_rate": 0.002,
      "loss": 3.182701587677002,
      "step": 1974
    },
    {
      "epoch": 0.271883578057601,
      "grad_norm": 0.1353788524866104,
      "learning_rate": 0.002,
      "loss": 3.132495880126953,
      "step": 1976
    },
    {
      "epoch": 0.2721587638653516,
      "grad_norm": 0.13208836317062378,
      "learning_rate": 0.002,
      "loss": 3.1755988597869873,
      "step": 1978
    },
    {
      "epoch": 0.2724339496731022,
      "grad_norm": 0.04436550661921501,
      "learning_rate": 0.002,
      "loss": 3.1352038383483887,
      "step": 1980
    },
    {
      "epoch": 0.2727091354808528,
      "grad_norm": 0.05146125704050064,
      "learning_rate": 0.002,
      "loss": 3.181680202484131,
      "step": 1982
    },
    {
      "epoch": 0.2729843212886034,
      "grad_norm": 0.042421821504831314,
      "learning_rate": 0.002,
      "loss": 3.1942763328552246,
      "step": 1984
    },
    {
      "epoch": 0.273259507096354,
      "grad_norm": 0.07853424549102783,
      "learning_rate": 0.002,
      "loss": 3.1372475624084473,
      "step": 1986
    },
    {
      "epoch": 0.27353469290410465,
      "grad_norm": 0.21450339257717133,
      "learning_rate": 0.002,
      "loss": 3.1193549633026123,
      "step": 1988
    },
    {
      "epoch": 0.27380987871185525,
      "grad_norm": 0.5218379497528076,
      "learning_rate": 0.002,
      "loss": 3.184783935546875,
      "step": 1990
    },
    {
      "epoch": 0.27408506451960585,
      "grad_norm": 0.3034469783306122,
      "learning_rate": 0.002,
      "loss": 3.1842570304870605,
      "step": 1992
    },
    {
      "epoch": 0.27436025032735645,
      "grad_norm": 0.28592392802238464,
      "learning_rate": 0.002,
      "loss": 3.13702392578125,
      "step": 1994
    },
    {
      "epoch": 0.27463543613510705,
      "grad_norm": 0.09354047477245331,
      "learning_rate": 0.002,
      "loss": 3.1669704914093018,
      "step": 1996
    },
    {
      "epoch": 0.27491062194285765,
      "grad_norm": 0.1718577891588211,
      "learning_rate": 0.002,
      "loss": 3.1806447505950928,
      "step": 1998
    },
    {
      "epoch": 0.2751858077506083,
      "grad_norm": 0.19726619124412537,
      "learning_rate": 0.002,
      "loss": 3.143404483795166,
      "step": 2000
    },
    {
      "epoch": 0.2754609935583589,
      "grad_norm": 0.08154014497995377,
      "learning_rate": 0.002,
      "loss": 3.184307098388672,
      "step": 2002
    },
    {
      "epoch": 0.2757361793661095,
      "grad_norm": 0.16125193238258362,
      "learning_rate": 0.002,
      "loss": 3.1487622261047363,
      "step": 2004
    },
    {
      "epoch": 0.2760113651738601,
      "grad_norm": 0.045539215207099915,
      "learning_rate": 0.002,
      "loss": 3.177561044692993,
      "step": 2006
    },
    {
      "epoch": 0.2762865509816107,
      "grad_norm": 0.17023375630378723,
      "learning_rate": 0.002,
      "loss": 3.1088109016418457,
      "step": 2008
    },
    {
      "epoch": 0.2765617367893613,
      "grad_norm": 0.14624565839767456,
      "learning_rate": 0.002,
      "loss": 3.204413890838623,
      "step": 2010
    },
    {
      "epoch": 0.2768369225971119,
      "grad_norm": 0.05581739544868469,
      "learning_rate": 0.002,
      "loss": 3.1468448638916016,
      "step": 2012
    },
    {
      "epoch": 0.27711210840486256,
      "grad_norm": 0.14053048193454742,
      "learning_rate": 0.002,
      "loss": 3.1873936653137207,
      "step": 2014
    },
    {
      "epoch": 0.27738729421261316,
      "grad_norm": 0.1398552805185318,
      "learning_rate": 0.002,
      "loss": 3.136190176010132,
      "step": 2016
    },
    {
      "epoch": 0.27766248002036376,
      "grad_norm": 0.17551912367343903,
      "learning_rate": 0.002,
      "loss": 3.0893030166625977,
      "step": 2018
    },
    {
      "epoch": 0.27793766582811436,
      "grad_norm": 0.3550892770290375,
      "learning_rate": 0.002,
      "loss": 3.1598973274230957,
      "step": 2020
    },
    {
      "epoch": 0.27821285163586496,
      "grad_norm": 0.13444504141807556,
      "learning_rate": 0.002,
      "loss": 3.154766082763672,
      "step": 2022
    },
    {
      "epoch": 0.27848803744361555,
      "grad_norm": 0.06247945874929428,
      "learning_rate": 0.002,
      "loss": 3.2010273933410645,
      "step": 2024
    },
    {
      "epoch": 0.2787632232513662,
      "grad_norm": 0.14805282652378082,
      "learning_rate": 0.002,
      "loss": 3.1986286640167236,
      "step": 2026
    },
    {
      "epoch": 0.2790384090591168,
      "grad_norm": 0.11516019701957703,
      "learning_rate": 0.002,
      "loss": 3.141326427459717,
      "step": 2028
    },
    {
      "epoch": 0.2793135948668674,
      "grad_norm": 0.05906543508172035,
      "learning_rate": 0.002,
      "loss": 3.157531261444092,
      "step": 2030
    },
    {
      "epoch": 0.279588780674618,
      "grad_norm": 0.12357135862112045,
      "learning_rate": 0.002,
      "loss": 3.152631998062134,
      "step": 2032
    },
    {
      "epoch": 0.2798639664823686,
      "grad_norm": 0.053285419940948486,
      "learning_rate": 0.002,
      "loss": 3.159782886505127,
      "step": 2034
    },
    {
      "epoch": 0.2801391522901192,
      "grad_norm": 0.1380310356616974,
      "learning_rate": 0.002,
      "loss": 3.1242284774780273,
      "step": 2036
    },
    {
      "epoch": 0.2804143380978698,
      "grad_norm": 0.19364893436431885,
      "learning_rate": 0.002,
      "loss": 3.157935619354248,
      "step": 2038
    },
    {
      "epoch": 0.28068952390562046,
      "grad_norm": 0.11578918248414993,
      "learning_rate": 0.002,
      "loss": 3.123955488204956,
      "step": 2040
    },
    {
      "epoch": 0.28096470971337106,
      "grad_norm": 0.07522974908351898,
      "learning_rate": 0.002,
      "loss": 3.130859375,
      "step": 2042
    },
    {
      "epoch": 0.28123989552112166,
      "grad_norm": 0.10105916857719421,
      "learning_rate": 0.002,
      "loss": 3.1130056381225586,
      "step": 2044
    },
    {
      "epoch": 0.28151508132887226,
      "grad_norm": 0.06926781684160233,
      "learning_rate": 0.002,
      "loss": 3.1527624130249023,
      "step": 2046
    },
    {
      "epoch": 0.28179026713662286,
      "grad_norm": 0.051490239799022675,
      "learning_rate": 0.002,
      "loss": 3.1354658603668213,
      "step": 2048
    },
    {
      "epoch": 0.28206545294437346,
      "grad_norm": 0.10749945044517517,
      "learning_rate": 0.002,
      "loss": 3.1534485816955566,
      "step": 2050
    },
    {
      "epoch": 0.2823406387521241,
      "grad_norm": 0.18131151795387268,
      "learning_rate": 0.002,
      "loss": 3.099430561065674,
      "step": 2052
    },
    {
      "epoch": 0.2826158245598747,
      "grad_norm": 0.2257726937532425,
      "learning_rate": 0.002,
      "loss": 3.1578164100646973,
      "step": 2054
    },
    {
      "epoch": 0.2828910103676253,
      "grad_norm": 0.4138951897621155,
      "learning_rate": 0.002,
      "loss": 3.1219635009765625,
      "step": 2056
    },
    {
      "epoch": 0.2831661961753759,
      "grad_norm": 0.2943064868450165,
      "learning_rate": 0.002,
      "loss": 3.169651508331299,
      "step": 2058
    },
    {
      "epoch": 0.2834413819831265,
      "grad_norm": 0.11193985491991043,
      "learning_rate": 0.002,
      "loss": 3.147505521774292,
      "step": 2060
    },
    {
      "epoch": 0.2837165677908771,
      "grad_norm": 0.13775576651096344,
      "learning_rate": 0.002,
      "loss": 3.129592180252075,
      "step": 2062
    },
    {
      "epoch": 0.2839917535986277,
      "grad_norm": 0.049641042947769165,
      "learning_rate": 0.002,
      "loss": 3.1208460330963135,
      "step": 2064
    },
    {
      "epoch": 0.28426693940637837,
      "grad_norm": 0.084688201546669,
      "learning_rate": 0.002,
      "loss": 3.1484851837158203,
      "step": 2066
    },
    {
      "epoch": 0.28454212521412897,
      "grad_norm": 0.04610047489404678,
      "learning_rate": 0.002,
      "loss": 3.142787218093872,
      "step": 2068
    },
    {
      "epoch": 0.28481731102187957,
      "grad_norm": 0.16162139177322388,
      "learning_rate": 0.002,
      "loss": 3.1120223999023438,
      "step": 2070
    },
    {
      "epoch": 0.28509249682963017,
      "grad_norm": 0.0413517989218235,
      "learning_rate": 0.002,
      "loss": 3.1291098594665527,
      "step": 2072
    },
    {
      "epoch": 0.28536768263738077,
      "grad_norm": 0.13970135152339935,
      "learning_rate": 0.002,
      "loss": 3.1392884254455566,
      "step": 2074
    },
    {
      "epoch": 0.28564286844513137,
      "grad_norm": 0.2135646492242813,
      "learning_rate": 0.002,
      "loss": 3.15743350982666,
      "step": 2076
    },
    {
      "epoch": 0.285918054252882,
      "grad_norm": 0.1294681429862976,
      "learning_rate": 0.002,
      "loss": 3.1447110176086426,
      "step": 2078
    },
    {
      "epoch": 0.2861932400606326,
      "grad_norm": 0.1367528736591339,
      "learning_rate": 0.002,
      "loss": 3.128030300140381,
      "step": 2080
    },
    {
      "epoch": 0.2864684258683832,
      "grad_norm": 0.2769624590873718,
      "learning_rate": 0.002,
      "loss": 3.0457401275634766,
      "step": 2082
    },
    {
      "epoch": 0.2867436116761338,
      "grad_norm": 0.1207069605588913,
      "learning_rate": 0.002,
      "loss": 3.1582589149475098,
      "step": 2084
    },
    {
      "epoch": 0.2870187974838844,
      "grad_norm": 0.03929191827774048,
      "learning_rate": 0.002,
      "loss": 3.111754894256592,
      "step": 2086
    },
    {
      "epoch": 0.287293983291635,
      "grad_norm": 0.0913391187787056,
      "learning_rate": 0.002,
      "loss": 3.151440143585205,
      "step": 2088
    },
    {
      "epoch": 0.2875691690993856,
      "grad_norm": 0.043259814381599426,
      "learning_rate": 0.002,
      "loss": 3.1075949668884277,
      "step": 2090
    },
    {
      "epoch": 0.2878443549071363,
      "grad_norm": 0.0697200745344162,
      "learning_rate": 0.002,
      "loss": 3.1442677974700928,
      "step": 2092
    },
    {
      "epoch": 0.2881195407148869,
      "grad_norm": 0.23518815636634827,
      "learning_rate": 0.002,
      "loss": 3.1675915718078613,
      "step": 2094
    },
    {
      "epoch": 0.2883947265226375,
      "grad_norm": 0.3114938735961914,
      "learning_rate": 0.002,
      "loss": 3.1286916732788086,
      "step": 2096
    },
    {
      "epoch": 0.2886699123303881,
      "grad_norm": 0.08489477634429932,
      "learning_rate": 0.002,
      "loss": 3.082106113433838,
      "step": 2098
    },
    {
      "epoch": 0.2889450981381387,
      "grad_norm": 0.05386471003293991,
      "learning_rate": 0.002,
      "loss": 3.1707983016967773,
      "step": 2100
    },
    {
      "epoch": 0.2892202839458893,
      "grad_norm": 0.09171373397111893,
      "learning_rate": 0.002,
      "loss": 3.115945339202881,
      "step": 2102
    },
    {
      "epoch": 0.2894954697536399,
      "grad_norm": 0.25937142968177795,
      "learning_rate": 0.002,
      "loss": 3.146247386932373,
      "step": 2104
    },
    {
      "epoch": 0.28977065556139053,
      "grad_norm": 0.11488514393568039,
      "learning_rate": 0.002,
      "loss": 3.145846366882324,
      "step": 2106
    },
    {
      "epoch": 0.29004584136914113,
      "grad_norm": 0.08964496105909348,
      "learning_rate": 0.002,
      "loss": 3.134930372238159,
      "step": 2108
    },
    {
      "epoch": 0.29032102717689173,
      "grad_norm": 0.06983549892902374,
      "learning_rate": 0.002,
      "loss": 3.1132028102874756,
      "step": 2110
    },
    {
      "epoch": 0.29059621298464233,
      "grad_norm": 0.21406994760036469,
      "learning_rate": 0.002,
      "loss": 3.1617138385772705,
      "step": 2112
    },
    {
      "epoch": 0.29087139879239293,
      "grad_norm": 0.2794387936592102,
      "learning_rate": 0.002,
      "loss": 3.1141014099121094,
      "step": 2114
    },
    {
      "epoch": 0.29114658460014353,
      "grad_norm": 0.09863928705453873,
      "learning_rate": 0.002,
      "loss": 3.1709113121032715,
      "step": 2116
    },
    {
      "epoch": 0.2914217704078942,
      "grad_norm": 0.12550505995750427,
      "learning_rate": 0.002,
      "loss": 3.1427688598632812,
      "step": 2118
    },
    {
      "epoch": 0.2916969562156448,
      "grad_norm": 0.18036441504955292,
      "learning_rate": 0.002,
      "loss": 3.1222052574157715,
      "step": 2120
    },
    {
      "epoch": 0.2919721420233954,
      "grad_norm": 0.07688433676958084,
      "learning_rate": 0.002,
      "loss": 3.1721293926239014,
      "step": 2122
    },
    {
      "epoch": 0.292247327831146,
      "grad_norm": 0.04071015864610672,
      "learning_rate": 0.002,
      "loss": 3.1151318550109863,
      "step": 2124
    },
    {
      "epoch": 0.2925225136388966,
      "grad_norm": 0.03895954415202141,
      "learning_rate": 0.002,
      "loss": 3.1200215816497803,
      "step": 2126
    },
    {
      "epoch": 0.2927976994466472,
      "grad_norm": 0.1539289653301239,
      "learning_rate": 0.002,
      "loss": 3.104166030883789,
      "step": 2128
    },
    {
      "epoch": 0.2930728852543978,
      "grad_norm": 0.2921256124973297,
      "learning_rate": 0.002,
      "loss": 3.1116366386413574,
      "step": 2130
    },
    {
      "epoch": 0.29334807106214844,
      "grad_norm": 0.44509825110435486,
      "learning_rate": 0.002,
      "loss": 3.141834259033203,
      "step": 2132
    },
    {
      "epoch": 0.29362325686989904,
      "grad_norm": 0.12057450413703918,
      "learning_rate": 0.002,
      "loss": 3.151336193084717,
      "step": 2134
    },
    {
      "epoch": 0.29389844267764964,
      "grad_norm": 0.2805466651916504,
      "learning_rate": 0.002,
      "loss": 3.1097559928894043,
      "step": 2136
    },
    {
      "epoch": 0.29417362848540024,
      "grad_norm": 0.03926708176732063,
      "learning_rate": 0.002,
      "loss": 3.1419034004211426,
      "step": 2138
    },
    {
      "epoch": 0.29444881429315084,
      "grad_norm": 0.09283816069364548,
      "learning_rate": 0.002,
      "loss": 3.1302504539489746,
      "step": 2140
    },
    {
      "epoch": 0.29472400010090144,
      "grad_norm": 0.04012809693813324,
      "learning_rate": 0.002,
      "loss": 3.12261962890625,
      "step": 2142
    },
    {
      "epoch": 0.2949991859086521,
      "grad_norm": 0.07509735971689224,
      "learning_rate": 0.002,
      "loss": 3.0749096870422363,
      "step": 2144
    },
    {
      "epoch": 0.2952743717164027,
      "grad_norm": 0.06528500467538834,
      "learning_rate": 0.002,
      "loss": 3.1013617515563965,
      "step": 2146
    },
    {
      "epoch": 0.2955495575241533,
      "grad_norm": 0.10629057139158249,
      "learning_rate": 0.002,
      "loss": 3.1476340293884277,
      "step": 2148
    },
    {
      "epoch": 0.2958247433319039,
      "grad_norm": 0.04837453365325928,
      "learning_rate": 0.002,
      "loss": 3.1363582611083984,
      "step": 2150
    },
    {
      "epoch": 0.2960999291396545,
      "grad_norm": 0.10822561383247375,
      "learning_rate": 0.002,
      "loss": 3.137174129486084,
      "step": 2152
    },
    {
      "epoch": 0.2963751149474051,
      "grad_norm": 0.07053729891777039,
      "learning_rate": 0.002,
      "loss": 3.0813374519348145,
      "step": 2154
    },
    {
      "epoch": 0.2966503007551557,
      "grad_norm": 0.14629530906677246,
      "learning_rate": 0.002,
      "loss": 3.1103334426879883,
      "step": 2156
    },
    {
      "epoch": 0.29692548656290635,
      "grad_norm": 0.37745723128318787,
      "learning_rate": 0.002,
      "loss": 3.1042428016662598,
      "step": 2158
    },
    {
      "epoch": 0.29720067237065695,
      "grad_norm": 0.3474879860877991,
      "learning_rate": 0.002,
      "loss": 3.135793685913086,
      "step": 2160
    },
    {
      "epoch": 0.29747585817840755,
      "grad_norm": 0.12471060454845428,
      "learning_rate": 0.002,
      "loss": 3.1050662994384766,
      "step": 2162
    },
    {
      "epoch": 0.29775104398615815,
      "grad_norm": 0.1213543713092804,
      "learning_rate": 0.002,
      "loss": 3.0826988220214844,
      "step": 2164
    },
    {
      "epoch": 0.29802622979390875,
      "grad_norm": 0.09878309071063995,
      "learning_rate": 0.002,
      "loss": 3.091637372970581,
      "step": 2166
    },
    {
      "epoch": 0.29830141560165935,
      "grad_norm": 0.039900828152894974,
      "learning_rate": 0.002,
      "loss": 3.067028522491455,
      "step": 2168
    },
    {
      "epoch": 0.29857660140941,
      "grad_norm": 0.08990321308374405,
      "learning_rate": 0.002,
      "loss": 3.057288646697998,
      "step": 2170
    },
    {
      "epoch": 0.2988517872171606,
      "grad_norm": 0.036592572927474976,
      "learning_rate": 0.002,
      "loss": 3.087406635284424,
      "step": 2172
    },
    {
      "epoch": 0.2991269730249112,
      "grad_norm": 0.04363669455051422,
      "learning_rate": 0.002,
      "loss": 3.059144973754883,
      "step": 2174
    },
    {
      "epoch": 0.2994021588326618,
      "grad_norm": 0.05050744488835335,
      "learning_rate": 0.002,
      "loss": 3.099968433380127,
      "step": 2176
    },
    {
      "epoch": 0.2996773446404124,
      "grad_norm": 0.11527672410011292,
      "learning_rate": 0.002,
      "loss": 3.1419358253479004,
      "step": 2178
    },
    {
      "epoch": 0.299952530448163,
      "grad_norm": 0.22617091238498688,
      "learning_rate": 0.002,
      "loss": 3.0339198112487793,
      "step": 2180
    },
    {
      "epoch": 0.3002277162559136,
      "grad_norm": 0.2869737148284912,
      "learning_rate": 0.002,
      "loss": 3.087963104248047,
      "step": 2182
    },
    {
      "epoch": 0.30050290206366426,
      "grad_norm": 0.23840877413749695,
      "learning_rate": 0.002,
      "loss": 3.1176047325134277,
      "step": 2184
    },
    {
      "epoch": 0.30077808787141486,
      "grad_norm": 0.05807257816195488,
      "learning_rate": 0.002,
      "loss": 3.0752809047698975,
      "step": 2186
    },
    {
      "epoch": 0.30105327367916546,
      "grad_norm": 0.14360904693603516,
      "learning_rate": 0.002,
      "loss": 3.1177005767822266,
      "step": 2188
    },
    {
      "epoch": 0.30132845948691606,
      "grad_norm": 0.06884349882602692,
      "learning_rate": 0.002,
      "loss": 3.121006965637207,
      "step": 2190
    },
    {
      "epoch": 0.30160364529466666,
      "grad_norm": 0.04133482277393341,
      "learning_rate": 0.002,
      "loss": 3.116774082183838,
      "step": 2192
    },
    {
      "epoch": 0.30187883110241726,
      "grad_norm": 0.1395515501499176,
      "learning_rate": 0.002,
      "loss": 3.108187675476074,
      "step": 2194
    },
    {
      "epoch": 0.3021540169101679,
      "grad_norm": 0.24527189135551453,
      "learning_rate": 0.002,
      "loss": 3.1005330085754395,
      "step": 2196
    },
    {
      "epoch": 0.3024292027179185,
      "grad_norm": 0.11050906777381897,
      "learning_rate": 0.002,
      "loss": 3.02618408203125,
      "step": 2198
    },
    {
      "epoch": 0.3027043885256691,
      "grad_norm": 0.1085546612739563,
      "learning_rate": 0.002,
      "loss": 3.092054843902588,
      "step": 2200
    },
    {
      "epoch": 0.3029795743334197,
      "grad_norm": 0.041853100061416626,
      "learning_rate": 0.002,
      "loss": 3.1077775955200195,
      "step": 2202
    },
    {
      "epoch": 0.3032547601411703,
      "grad_norm": 0.14910896122455597,
      "learning_rate": 0.002,
      "loss": 3.0972554683685303,
      "step": 2204
    },
    {
      "epoch": 0.3035299459489209,
      "grad_norm": 0.14926546812057495,
      "learning_rate": 0.002,
      "loss": 3.1270344257354736,
      "step": 2206
    },
    {
      "epoch": 0.3038051317566715,
      "grad_norm": 0.17304351925849915,
      "learning_rate": 0.002,
      "loss": 3.1128180027008057,
      "step": 2208
    },
    {
      "epoch": 0.30408031756442216,
      "grad_norm": 0.054847054183483124,
      "learning_rate": 0.002,
      "loss": 3.0929245948791504,
      "step": 2210
    },
    {
      "epoch": 0.30435550337217276,
      "grad_norm": 0.07579727470874786,
      "learning_rate": 0.002,
      "loss": 3.091480255126953,
      "step": 2212
    },
    {
      "epoch": 0.30463068917992336,
      "grad_norm": 0.15581727027893066,
      "learning_rate": 0.002,
      "loss": 3.0815205574035645,
      "step": 2214
    },
    {
      "epoch": 0.30490587498767396,
      "grad_norm": 0.265018105506897,
      "learning_rate": 0.002,
      "loss": 3.0683298110961914,
      "step": 2216
    },
    {
      "epoch": 0.30518106079542456,
      "grad_norm": 0.2057543396949768,
      "learning_rate": 0.002,
      "loss": 3.098215341567993,
      "step": 2218
    },
    {
      "epoch": 0.30545624660317516,
      "grad_norm": 0.0420483835041523,
      "learning_rate": 0.002,
      "loss": 3.095231294631958,
      "step": 2220
    },
    {
      "epoch": 0.3057314324109258,
      "grad_norm": 0.1608852744102478,
      "learning_rate": 0.002,
      "loss": 3.0613784790039062,
      "step": 2222
    },
    {
      "epoch": 0.3060066182186764,
      "grad_norm": 0.07281152904033661,
      "learning_rate": 0.002,
      "loss": 3.0712108612060547,
      "step": 2224
    },
    {
      "epoch": 0.306281804026427,
      "grad_norm": 0.046884872019290924,
      "learning_rate": 0.002,
      "loss": 3.098275661468506,
      "step": 2226
    },
    {
      "epoch": 0.3065569898341776,
      "grad_norm": 0.12097907066345215,
      "learning_rate": 0.002,
      "loss": 3.07134747505188,
      "step": 2228
    },
    {
      "epoch": 0.3068321756419282,
      "grad_norm": 0.36741402745246887,
      "learning_rate": 0.002,
      "loss": 3.0793209075927734,
      "step": 2230
    },
    {
      "epoch": 0.3071073614496788,
      "grad_norm": 0.2881520986557007,
      "learning_rate": 0.002,
      "loss": 3.0948948860168457,
      "step": 2232
    },
    {
      "epoch": 0.3073825472574294,
      "grad_norm": 0.3098124563694,
      "learning_rate": 0.002,
      "loss": 3.084611415863037,
      "step": 2234
    },
    {
      "epoch": 0.3076577330651801,
      "grad_norm": 0.2985008656978607,
      "learning_rate": 0.002,
      "loss": 3.0826873779296875,
      "step": 2236
    },
    {
      "epoch": 0.30793291887293067,
      "grad_norm": 0.19051606953144073,
      "learning_rate": 0.002,
      "loss": 3.1281259059906006,
      "step": 2238
    },
    {
      "epoch": 0.30820810468068127,
      "grad_norm": 0.06678138673305511,
      "learning_rate": 0.002,
      "loss": 3.1181492805480957,
      "step": 2240
    },
    {
      "epoch": 0.30848329048843187,
      "grad_norm": 0.26118794083595276,
      "learning_rate": 0.002,
      "loss": 3.0860462188720703,
      "step": 2242
    },
    {
      "epoch": 0.30875847629618247,
      "grad_norm": 0.03655305504798889,
      "learning_rate": 0.002,
      "loss": 3.0633111000061035,
      "step": 2244
    },
    {
      "epoch": 0.30903366210393307,
      "grad_norm": 0.20692887902259827,
      "learning_rate": 0.002,
      "loss": 3.1030044555664062,
      "step": 2246
    },
    {
      "epoch": 0.3093088479116837,
      "grad_norm": 0.07354625314474106,
      "learning_rate": 0.002,
      "loss": 3.0952086448669434,
      "step": 2248
    },
    {
      "epoch": 0.3095840337194343,
      "grad_norm": 0.1462719440460205,
      "learning_rate": 0.002,
      "loss": 3.0773825645446777,
      "step": 2250
    },
    {
      "epoch": 0.3098592195271849,
      "grad_norm": 0.06153813749551773,
      "learning_rate": 0.002,
      "loss": 3.128737449645996,
      "step": 2252
    },
    {
      "epoch": 0.3101344053349355,
      "grad_norm": 0.1264212280511856,
      "learning_rate": 0.002,
      "loss": 3.1232986450195312,
      "step": 2254
    },
    {
      "epoch": 0.3104095911426861,
      "grad_norm": 0.2110326737165451,
      "learning_rate": 0.002,
      "loss": 3.058558940887451,
      "step": 2256
    },
    {
      "epoch": 0.3106847769504367,
      "grad_norm": 0.052678242325782776,
      "learning_rate": 0.002,
      "loss": 3.0633633136749268,
      "step": 2258
    },
    {
      "epoch": 0.3109599627581873,
      "grad_norm": 0.07671646773815155,
      "learning_rate": 0.002,
      "loss": 3.081782102584839,
      "step": 2260
    },
    {
      "epoch": 0.311235148565938,
      "grad_norm": 0.06693492829799652,
      "learning_rate": 0.002,
      "loss": 3.1317031383514404,
      "step": 2262
    },
    {
      "epoch": 0.3115103343736886,
      "grad_norm": 0.18338151276111603,
      "learning_rate": 0.002,
      "loss": 3.084897994995117,
      "step": 2264
    },
    {
      "epoch": 0.3117855201814392,
      "grad_norm": 0.0921526849269867,
      "learning_rate": 0.002,
      "loss": 3.0824434757232666,
      "step": 2266
    },
    {
      "epoch": 0.3120607059891898,
      "grad_norm": 0.09370721876621246,
      "learning_rate": 0.002,
      "loss": 3.0942506790161133,
      "step": 2268
    },
    {
      "epoch": 0.3123358917969404,
      "grad_norm": 0.07016541808843613,
      "learning_rate": 0.002,
      "loss": 3.063425064086914,
      "step": 2270
    },
    {
      "epoch": 0.312611077604691,
      "grad_norm": 0.07182526588439941,
      "learning_rate": 0.002,
      "loss": 3.0625853538513184,
      "step": 2272
    },
    {
      "epoch": 0.31288626341244163,
      "grad_norm": 0.08050546795129776,
      "learning_rate": 0.002,
      "loss": 3.121581554412842,
      "step": 2274
    },
    {
      "epoch": 0.31316144922019223,
      "grad_norm": 0.19104225933551788,
      "learning_rate": 0.002,
      "loss": 3.082686424255371,
      "step": 2276
    },
    {
      "epoch": 0.31343663502794283,
      "grad_norm": 0.37305599451065063,
      "learning_rate": 0.002,
      "loss": 3.082512378692627,
      "step": 2278
    },
    {
      "epoch": 0.31371182083569343,
      "grad_norm": 0.3879246711730957,
      "learning_rate": 0.002,
      "loss": 3.0655674934387207,
      "step": 2280
    },
    {
      "epoch": 0.31398700664344403,
      "grad_norm": 0.18475867807865143,
      "learning_rate": 0.002,
      "loss": 3.0693931579589844,
      "step": 2282
    },
    {
      "epoch": 0.31426219245119463,
      "grad_norm": 0.2370530515909195,
      "learning_rate": 0.002,
      "loss": 3.0931544303894043,
      "step": 2284
    },
    {
      "epoch": 0.31453737825894523,
      "grad_norm": 0.15024621784687042,
      "learning_rate": 0.002,
      "loss": 3.091038942337036,
      "step": 2286
    },
    {
      "epoch": 0.3148125640666959,
      "grad_norm": 0.11110799014568329,
      "learning_rate": 0.002,
      "loss": 3.1207032203674316,
      "step": 2288
    },
    {
      "epoch": 0.3150877498744465,
      "grad_norm": 0.25639286637306213,
      "learning_rate": 0.002,
      "loss": 3.1151039600372314,
      "step": 2290
    },
    {
      "epoch": 0.3153629356821971,
      "grad_norm": 0.059088557958602905,
      "learning_rate": 0.002,
      "loss": 3.05486798286438,
      "step": 2292
    },
    {
      "epoch": 0.3156381214899477,
      "grad_norm": 0.14709527790546417,
      "learning_rate": 0.002,
      "loss": 3.095165729522705,
      "step": 2294
    },
    {
      "epoch": 0.3159133072976983,
      "grad_norm": 0.10768505930900574,
      "learning_rate": 0.002,
      "loss": 3.0470073223114014,
      "step": 2296
    },
    {
      "epoch": 0.3161884931054489,
      "grad_norm": 0.03858761489391327,
      "learning_rate": 0.002,
      "loss": 3.080543041229248,
      "step": 2298
    },
    {
      "epoch": 0.31646367891319954,
      "grad_norm": 0.11050071567296982,
      "learning_rate": 0.002,
      "loss": 3.129875659942627,
      "step": 2300
    },
    {
      "epoch": 0.31673886472095014,
      "grad_norm": 0.09900562465190887,
      "learning_rate": 0.002,
      "loss": 3.085287094116211,
      "step": 2302
    },
    {
      "epoch": 0.31701405052870074,
      "grad_norm": 0.04179767519235611,
      "learning_rate": 0.002,
      "loss": 3.075934410095215,
      "step": 2304
    },
    {
      "epoch": 0.31728923633645134,
      "grad_norm": 0.03855935484170914,
      "learning_rate": 0.002,
      "loss": 3.0352392196655273,
      "step": 2306
    },
    {
      "epoch": 0.31756442214420194,
      "grad_norm": 0.11970162391662598,
      "learning_rate": 0.002,
      "loss": 3.0833396911621094,
      "step": 2308
    },
    {
      "epoch": 0.31783960795195254,
      "grad_norm": 0.18590915203094482,
      "learning_rate": 0.002,
      "loss": 3.070537805557251,
      "step": 2310
    },
    {
      "epoch": 0.31811479375970314,
      "grad_norm": 0.20122329890727997,
      "learning_rate": 0.002,
      "loss": 3.113758087158203,
      "step": 2312
    },
    {
      "epoch": 0.3183899795674538,
      "grad_norm": 0.23040354251861572,
      "learning_rate": 0.002,
      "loss": 3.066885471343994,
      "step": 2314
    },
    {
      "epoch": 0.3186651653752044,
      "grad_norm": 0.10901755839586258,
      "learning_rate": 0.002,
      "loss": 3.0630416870117188,
      "step": 2316
    },
    {
      "epoch": 0.318940351182955,
      "grad_norm": 0.04745902121067047,
      "learning_rate": 0.002,
      "loss": 3.079105854034424,
      "step": 2318
    },
    {
      "epoch": 0.3192155369907056,
      "grad_norm": 0.12325778603553772,
      "learning_rate": 0.002,
      "loss": 3.0601892471313477,
      "step": 2320
    },
    {
      "epoch": 0.3194907227984562,
      "grad_norm": 0.10685629397630692,
      "learning_rate": 0.002,
      "loss": 3.0745315551757812,
      "step": 2322
    },
    {
      "epoch": 0.3197659086062068,
      "grad_norm": 0.0578596368432045,
      "learning_rate": 0.002,
      "loss": 3.1202964782714844,
      "step": 2324
    },
    {
      "epoch": 0.32004109441395745,
      "grad_norm": 0.05385231971740723,
      "learning_rate": 0.002,
      "loss": 3.044231653213501,
      "step": 2326
    },
    {
      "epoch": 0.32031628022170805,
      "grad_norm": 0.1948811113834381,
      "learning_rate": 0.002,
      "loss": 3.065443277359009,
      "step": 2328
    },
    {
      "epoch": 0.32059146602945865,
      "grad_norm": 0.3364434242248535,
      "learning_rate": 0.002,
      "loss": 3.0428194999694824,
      "step": 2330
    },
    {
      "epoch": 0.32086665183720925,
      "grad_norm": 0.17216694355010986,
      "learning_rate": 0.002,
      "loss": 3.0358800888061523,
      "step": 2332
    },
    {
      "epoch": 0.32114183764495985,
      "grad_norm": 0.06304626911878586,
      "learning_rate": 0.002,
      "loss": 3.1231727600097656,
      "step": 2334
    },
    {
      "epoch": 0.32141702345271045,
      "grad_norm": 0.035074442625045776,
      "learning_rate": 0.002,
      "loss": 3.071943759918213,
      "step": 2336
    },
    {
      "epoch": 0.32169220926046105,
      "grad_norm": 0.047363076359033585,
      "learning_rate": 0.002,
      "loss": 3.0899946689605713,
      "step": 2338
    },
    {
      "epoch": 0.3219673950682117,
      "grad_norm": 0.08741086721420288,
      "learning_rate": 0.002,
      "loss": 3.0665183067321777,
      "step": 2340
    },
    {
      "epoch": 0.3222425808759623,
      "grad_norm": 0.061711981892585754,
      "learning_rate": 0.002,
      "loss": 2.9963219165802,
      "step": 2342
    },
    {
      "epoch": 0.3225177666837129,
      "grad_norm": 0.04030074551701546,
      "learning_rate": 0.002,
      "loss": 3.076340675354004,
      "step": 2344
    },
    {
      "epoch": 0.3227929524914635,
      "grad_norm": 0.05742712318897247,
      "learning_rate": 0.002,
      "loss": 3.074228525161743,
      "step": 2346
    },
    {
      "epoch": 0.3230681382992141,
      "grad_norm": 0.04873465374112129,
      "learning_rate": 0.002,
      "loss": 3.0631768703460693,
      "step": 2348
    },
    {
      "epoch": 0.3233433241069647,
      "grad_norm": 0.0374622717499733,
      "learning_rate": 0.002,
      "loss": 3.0317320823669434,
      "step": 2350
    },
    {
      "epoch": 0.32361850991471536,
      "grad_norm": 0.04317442700266838,
      "learning_rate": 0.002,
      "loss": 3.079056978225708,
      "step": 2352
    },
    {
      "epoch": 0.32389369572246596,
      "grad_norm": 0.0578560009598732,
      "learning_rate": 0.002,
      "loss": 3.026437759399414,
      "step": 2354
    },
    {
      "epoch": 0.32416888153021656,
      "grad_norm": 0.17598989605903625,
      "learning_rate": 0.002,
      "loss": 3.077414035797119,
      "step": 2356
    },
    {
      "epoch": 0.32444406733796716,
      "grad_norm": 0.6458200216293335,
      "learning_rate": 0.002,
      "loss": 3.0779170989990234,
      "step": 2358
    },
    {
      "epoch": 0.32471925314571776,
      "grad_norm": 0.3382525146007538,
      "learning_rate": 0.002,
      "loss": 3.030529022216797,
      "step": 2360
    },
    {
      "epoch": 0.32499443895346836,
      "grad_norm": 0.4159870147705078,
      "learning_rate": 0.002,
      "loss": 3.099264621734619,
      "step": 2362
    },
    {
      "epoch": 0.32526962476121896,
      "grad_norm": 0.2080809473991394,
      "learning_rate": 0.002,
      "loss": 3.0985684394836426,
      "step": 2364
    },
    {
      "epoch": 0.3255448105689696,
      "grad_norm": 0.050616148859262466,
      "learning_rate": 0.002,
      "loss": 3.08852481842041,
      "step": 2366
    },
    {
      "epoch": 0.3258199963767202,
      "grad_norm": 0.10680672526359558,
      "learning_rate": 0.002,
      "loss": 3.0364222526550293,
      "step": 2368
    },
    {
      "epoch": 0.3260951821844708,
      "grad_norm": 0.09642253816127777,
      "learning_rate": 0.002,
      "loss": 3.0556137561798096,
      "step": 2370
    },
    {
      "epoch": 0.3263703679922214,
      "grad_norm": 0.0621255598962307,
      "learning_rate": 0.002,
      "loss": 3.058898448944092,
      "step": 2372
    },
    {
      "epoch": 0.326645553799972,
      "grad_norm": 0.06574704498052597,
      "learning_rate": 0.002,
      "loss": 3.060737371444702,
      "step": 2374
    },
    {
      "epoch": 0.3269207396077226,
      "grad_norm": 0.09796186536550522,
      "learning_rate": 0.002,
      "loss": 3.0165228843688965,
      "step": 2376
    },
    {
      "epoch": 0.32719592541547327,
      "grad_norm": 0.10212180018424988,
      "learning_rate": 0.002,
      "loss": 3.0580620765686035,
      "step": 2378
    },
    {
      "epoch": 0.32747111122322387,
      "grad_norm": 0.0741724893450737,
      "learning_rate": 0.002,
      "loss": 3.0850284099578857,
      "step": 2380
    },
    {
      "epoch": 0.32774629703097446,
      "grad_norm": 0.039644114673137665,
      "learning_rate": 0.002,
      "loss": 3.034531831741333,
      "step": 2382
    },
    {
      "epoch": 0.32802148283872506,
      "grad_norm": 0.1771385669708252,
      "learning_rate": 0.002,
      "loss": 3.059727668762207,
      "step": 2384
    },
    {
      "epoch": 0.32829666864647566,
      "grad_norm": 0.1576443612575531,
      "learning_rate": 0.002,
      "loss": 3.0448787212371826,
      "step": 2386
    },
    {
      "epoch": 0.32857185445422626,
      "grad_norm": 0.21459491550922394,
      "learning_rate": 0.002,
      "loss": 2.9967761039733887,
      "step": 2388
    },
    {
      "epoch": 0.32884704026197686,
      "grad_norm": 0.08634496480226517,
      "learning_rate": 0.002,
      "loss": 3.077462673187256,
      "step": 2390
    },
    {
      "epoch": 0.3291222260697275,
      "grad_norm": 0.1548713743686676,
      "learning_rate": 0.002,
      "loss": 3.0376503467559814,
      "step": 2392
    },
    {
      "epoch": 0.3293974118774781,
      "grad_norm": 0.1018066257238388,
      "learning_rate": 0.002,
      "loss": 3.0434398651123047,
      "step": 2394
    },
    {
      "epoch": 0.3296725976852287,
      "grad_norm": 0.13582971692085266,
      "learning_rate": 0.002,
      "loss": 3.1011390686035156,
      "step": 2396
    },
    {
      "epoch": 0.3299477834929793,
      "grad_norm": 0.2392358034849167,
      "learning_rate": 0.002,
      "loss": 3.0715255737304688,
      "step": 2398
    },
    {
      "epoch": 0.3302229693007299,
      "grad_norm": 0.1419321745634079,
      "learning_rate": 0.002,
      "loss": 3.063403606414795,
      "step": 2400
    },
    {
      "epoch": 0.3304981551084805,
      "grad_norm": 0.08590245991945267,
      "learning_rate": 0.002,
      "loss": 3.0898334980010986,
      "step": 2402
    },
    {
      "epoch": 0.3307733409162312,
      "grad_norm": 0.14661429822444916,
      "learning_rate": 0.002,
      "loss": 3.0678553581237793,
      "step": 2404
    },
    {
      "epoch": 0.3310485267239818,
      "grad_norm": 0.03561217710375786,
      "learning_rate": 0.002,
      "loss": 3.036510467529297,
      "step": 2406
    },
    {
      "epoch": 0.3313237125317324,
      "grad_norm": 0.13649196922779083,
      "learning_rate": 0.002,
      "loss": 3.081310987472534,
      "step": 2408
    },
    {
      "epoch": 0.331598898339483,
      "grad_norm": 0.05882587656378746,
      "learning_rate": 0.002,
      "loss": 3.0460457801818848,
      "step": 2410
    },
    {
      "epoch": 0.3318740841472336,
      "grad_norm": 0.03834550827741623,
      "learning_rate": 0.002,
      "loss": 3.0609326362609863,
      "step": 2412
    },
    {
      "epoch": 0.33214926995498417,
      "grad_norm": 0.05723734200000763,
      "learning_rate": 0.002,
      "loss": 3.0416195392608643,
      "step": 2414
    },
    {
      "epoch": 0.33242445576273477,
      "grad_norm": 0.04220975562930107,
      "learning_rate": 0.002,
      "loss": 3.00174617767334,
      "step": 2416
    },
    {
      "epoch": 0.3326996415704854,
      "grad_norm": 0.2583889961242676,
      "learning_rate": 0.002,
      "loss": 3.029215097427368,
      "step": 2418
    },
    {
      "epoch": 0.332974827378236,
      "grad_norm": 0.3991207778453827,
      "learning_rate": 0.002,
      "loss": 3.0724141597747803,
      "step": 2420
    },
    {
      "epoch": 0.3332500131859866,
      "grad_norm": 0.23447631299495697,
      "learning_rate": 0.002,
      "loss": 3.06632661819458,
      "step": 2422
    },
    {
      "epoch": 0.3335251989937372,
      "grad_norm": 0.2443021833896637,
      "learning_rate": 0.002,
      "loss": 3.0501022338867188,
      "step": 2424
    },
    {
      "epoch": 0.3338003848014878,
      "grad_norm": 0.19970211386680603,
      "learning_rate": 0.002,
      "loss": 3.0812296867370605,
      "step": 2426
    },
    {
      "epoch": 0.3340755706092384,
      "grad_norm": 0.19553296267986298,
      "learning_rate": 0.002,
      "loss": 3.051666259765625,
      "step": 2428
    },
    {
      "epoch": 0.3343507564169891,
      "grad_norm": 0.07868775725364685,
      "learning_rate": 0.002,
      "loss": 3.0837645530700684,
      "step": 2430
    },
    {
      "epoch": 0.3346259422247397,
      "grad_norm": 0.14915350079536438,
      "learning_rate": 0.002,
      "loss": 3.038058280944824,
      "step": 2432
    },
    {
      "epoch": 0.3349011280324903,
      "grad_norm": 0.034936774522066116,
      "learning_rate": 0.002,
      "loss": 3.059438705444336,
      "step": 2434
    },
    {
      "epoch": 0.3351763138402409,
      "grad_norm": 0.08349720388650894,
      "learning_rate": 0.002,
      "loss": 3.0548198223114014,
      "step": 2436
    },
    {
      "epoch": 0.3354514996479915,
      "grad_norm": 0.0578167699277401,
      "learning_rate": 0.002,
      "loss": 3.0192763805389404,
      "step": 2438
    },
    {
      "epoch": 0.3357266854557421,
      "grad_norm": 0.0423780158162117,
      "learning_rate": 0.002,
      "loss": 3.0360069274902344,
      "step": 2440
    },
    {
      "epoch": 0.3360018712634927,
      "grad_norm": 0.09225593507289886,
      "learning_rate": 0.002,
      "loss": 3.0221920013427734,
      "step": 2442
    },
    {
      "epoch": 0.33627705707124333,
      "grad_norm": 0.12429685890674591,
      "learning_rate": 0.002,
      "loss": 3.081803798675537,
      "step": 2444
    },
    {
      "epoch": 0.33655224287899393,
      "grad_norm": 0.11384772509336472,
      "learning_rate": 0.002,
      "loss": 3.0542805194854736,
      "step": 2446
    },
    {
      "epoch": 0.33682742868674453,
      "grad_norm": 0.22250664234161377,
      "learning_rate": 0.002,
      "loss": 3.04050874710083,
      "step": 2448
    },
    {
      "epoch": 0.33710261449449513,
      "grad_norm": 0.22932171821594238,
      "learning_rate": 0.002,
      "loss": 3.048790454864502,
      "step": 2450
    },
    {
      "epoch": 0.33737780030224573,
      "grad_norm": 0.03900391608476639,
      "learning_rate": 0.002,
      "loss": 3.001356363296509,
      "step": 2452
    },
    {
      "epoch": 0.33765298610999633,
      "grad_norm": 0.11754678934812546,
      "learning_rate": 0.002,
      "loss": 3.07786226272583,
      "step": 2454
    },
    {
      "epoch": 0.337928171917747,
      "grad_norm": 0.08663099259138107,
      "learning_rate": 0.002,
      "loss": 3.0448479652404785,
      "step": 2456
    },
    {
      "epoch": 0.3382033577254976,
      "grad_norm": 0.1235249862074852,
      "learning_rate": 0.002,
      "loss": 3.003847122192383,
      "step": 2458
    },
    {
      "epoch": 0.3384785435332482,
      "grad_norm": 0.2654241919517517,
      "learning_rate": 0.002,
      "loss": 3.0156450271606445,
      "step": 2460
    },
    {
      "epoch": 0.3387537293409988,
      "grad_norm": 0.10982787609100342,
      "learning_rate": 0.002,
      "loss": 3.0085716247558594,
      "step": 2462
    },
    {
      "epoch": 0.3390289151487494,
      "grad_norm": 0.10419948399066925,
      "learning_rate": 0.002,
      "loss": 3.0549912452697754,
      "step": 2464
    },
    {
      "epoch": 0.3393041009565,
      "grad_norm": 0.220064178109169,
      "learning_rate": 0.002,
      "loss": 3.0334272384643555,
      "step": 2466
    },
    {
      "epoch": 0.3395792867642506,
      "grad_norm": 0.19624660909175873,
      "learning_rate": 0.002,
      "loss": 3.0371828079223633,
      "step": 2468
    },
    {
      "epoch": 0.33985447257200124,
      "grad_norm": 0.10002633184194565,
      "learning_rate": 0.002,
      "loss": 3.0355942249298096,
      "step": 2470
    },
    {
      "epoch": 0.34012965837975184,
      "grad_norm": 0.2511661648750305,
      "learning_rate": 0.002,
      "loss": 2.9916062355041504,
      "step": 2472
    },
    {
      "epoch": 0.34040484418750244,
      "grad_norm": 0.1163979172706604,
      "learning_rate": 0.002,
      "loss": 3.048316478729248,
      "step": 2474
    },
    {
      "epoch": 0.34068002999525304,
      "grad_norm": 0.1195315346121788,
      "learning_rate": 0.002,
      "loss": 2.989320755004883,
      "step": 2476
    },
    {
      "epoch": 0.34095521580300364,
      "grad_norm": 0.04658740386366844,
      "learning_rate": 0.002,
      "loss": 3.01676082611084,
      "step": 2478
    },
    {
      "epoch": 0.34123040161075424,
      "grad_norm": 0.22428938746452332,
      "learning_rate": 0.002,
      "loss": 3.0243377685546875,
      "step": 2480
    },
    {
      "epoch": 0.3415055874185049,
      "grad_norm": 0.24145203828811646,
      "learning_rate": 0.002,
      "loss": 3.096771240234375,
      "step": 2482
    },
    {
      "epoch": 0.3417807732262555,
      "grad_norm": 0.14435964822769165,
      "learning_rate": 0.002,
      "loss": 3.021070718765259,
      "step": 2484
    },
    {
      "epoch": 0.3420559590340061,
      "grad_norm": 0.035604316741228104,
      "learning_rate": 0.002,
      "loss": 3.031980514526367,
      "step": 2486
    },
    {
      "epoch": 0.3423311448417567,
      "grad_norm": 0.06752578169107437,
      "learning_rate": 0.002,
      "loss": 3.064784049987793,
      "step": 2488
    },
    {
      "epoch": 0.3426063306495073,
      "grad_norm": 0.12427617609500885,
      "learning_rate": 0.002,
      "loss": 3.0630829334259033,
      "step": 2490
    },
    {
      "epoch": 0.3428815164572579,
      "grad_norm": 0.04174149036407471,
      "learning_rate": 0.002,
      "loss": 3.0645651817321777,
      "step": 2492
    },
    {
      "epoch": 0.3431567022650085,
      "grad_norm": 0.09016519039869308,
      "learning_rate": 0.002,
      "loss": 2.9736485481262207,
      "step": 2494
    },
    {
      "epoch": 0.34343188807275915,
      "grad_norm": 0.10312225669622421,
      "learning_rate": 0.002,
      "loss": 3.0419907569885254,
      "step": 2496
    },
    {
      "epoch": 0.34370707388050975,
      "grad_norm": 0.17370367050170898,
      "learning_rate": 0.002,
      "loss": 2.9874637126922607,
      "step": 2498
    },
    {
      "epoch": 0.34398225968826035,
      "grad_norm": 0.16659043729305267,
      "learning_rate": 0.002,
      "loss": 3.033513307571411,
      "step": 2500
    },
    {
      "epoch": 0.34425744549601095,
      "grad_norm": 0.11442777514457703,
      "learning_rate": 0.002,
      "loss": 3.0261311531066895,
      "step": 2502
    },
    {
      "epoch": 0.34453263130376155,
      "grad_norm": 0.09744203835725784,
      "learning_rate": 0.002,
      "loss": 3.0320563316345215,
      "step": 2504
    },
    {
      "epoch": 0.34480781711151215,
      "grad_norm": 0.1310109943151474,
      "learning_rate": 0.002,
      "loss": 3.0707290172576904,
      "step": 2506
    },
    {
      "epoch": 0.3450830029192628,
      "grad_norm": 0.07419542968273163,
      "learning_rate": 0.002,
      "loss": 3.0418169498443604,
      "step": 2508
    },
    {
      "epoch": 0.3453581887270134,
      "grad_norm": 0.10984942317008972,
      "learning_rate": 0.002,
      "loss": 3.0171213150024414,
      "step": 2510
    },
    {
      "epoch": 0.345633374534764,
      "grad_norm": 0.07399982959032059,
      "learning_rate": 0.002,
      "loss": 3.030930519104004,
      "step": 2512
    },
    {
      "epoch": 0.3459085603425146,
      "grad_norm": 0.04942847788333893,
      "learning_rate": 0.002,
      "loss": 3.061676025390625,
      "step": 2514
    },
    {
      "epoch": 0.3461837461502652,
      "grad_norm": 0.1341182142496109,
      "learning_rate": 0.002,
      "loss": 3.0112357139587402,
      "step": 2516
    },
    {
      "epoch": 0.3464589319580158,
      "grad_norm": 0.6057227253913879,
      "learning_rate": 0.002,
      "loss": 3.022840976715088,
      "step": 2518
    },
    {
      "epoch": 0.3467341177657664,
      "grad_norm": 0.11502119898796082,
      "learning_rate": 0.002,
      "loss": 3.059833526611328,
      "step": 2520
    },
    {
      "epoch": 0.34700930357351706,
      "grad_norm": 0.4498050808906555,
      "learning_rate": 0.002,
      "loss": 3.0266733169555664,
      "step": 2522
    },
    {
      "epoch": 0.34728448938126766,
      "grad_norm": 0.5031160116195679,
      "learning_rate": 0.002,
      "loss": 3.010098934173584,
      "step": 2524
    },
    {
      "epoch": 0.34755967518901826,
      "grad_norm": 0.3709568977355957,
      "learning_rate": 0.002,
      "loss": 3.0281789302825928,
      "step": 2526
    },
    {
      "epoch": 0.34783486099676886,
      "grad_norm": 0.23711255192756653,
      "learning_rate": 0.002,
      "loss": 2.9697117805480957,
      "step": 2528
    },
    {
      "epoch": 0.34811004680451946,
      "grad_norm": 0.05329679325222969,
      "learning_rate": 0.002,
      "loss": 3.0220913887023926,
      "step": 2530
    },
    {
      "epoch": 0.34838523261227006,
      "grad_norm": 0.03845962509512901,
      "learning_rate": 0.002,
      "loss": 3.0403921604156494,
      "step": 2532
    },
    {
      "epoch": 0.3486604184200207,
      "grad_norm": 0.06526032090187073,
      "learning_rate": 0.002,
      "loss": 3.041268825531006,
      "step": 2534
    },
    {
      "epoch": 0.3489356042277713,
      "grad_norm": 0.0372769869863987,
      "learning_rate": 0.002,
      "loss": 3.025538921356201,
      "step": 2536
    },
    {
      "epoch": 0.3492107900355219,
      "grad_norm": 0.03430238738656044,
      "learning_rate": 0.002,
      "loss": 3.0445895195007324,
      "step": 2538
    },
    {
      "epoch": 0.3494859758432725,
      "grad_norm": 0.04652726650238037,
      "learning_rate": 0.002,
      "loss": 3.0204741954803467,
      "step": 2540
    },
    {
      "epoch": 0.3497611616510231,
      "grad_norm": 0.03532101586461067,
      "learning_rate": 0.002,
      "loss": 3.044893741607666,
      "step": 2542
    },
    {
      "epoch": 0.3500363474587737,
      "grad_norm": 0.10526811331510544,
      "learning_rate": 0.002,
      "loss": 3.0322937965393066,
      "step": 2544
    },
    {
      "epoch": 0.3503115332665243,
      "grad_norm": 0.06660433113574982,
      "learning_rate": 0.002,
      "loss": 3.0019681453704834,
      "step": 2546
    },
    {
      "epoch": 0.35058671907427497,
      "grad_norm": 0.04463665932416916,
      "learning_rate": 0.002,
      "loss": 3.0018208026885986,
      "step": 2548
    },
    {
      "epoch": 0.35086190488202557,
      "grad_norm": 0.14502575993537903,
      "learning_rate": 0.002,
      "loss": 3.03670334815979,
      "step": 2550
    },
    {
      "epoch": 0.35113709068977617,
      "grad_norm": 0.1591581255197525,
      "learning_rate": 0.002,
      "loss": 3.01576566696167,
      "step": 2552
    },
    {
      "epoch": 0.35141227649752677,
      "grad_norm": 0.041181813925504684,
      "learning_rate": 0.002,
      "loss": 3.0282273292541504,
      "step": 2554
    },
    {
      "epoch": 0.35168746230527737,
      "grad_norm": 0.051816217601299286,
      "learning_rate": 0.002,
      "loss": 3.012969970703125,
      "step": 2556
    },
    {
      "epoch": 0.35196264811302796,
      "grad_norm": 0.1746257096529007,
      "learning_rate": 0.002,
      "loss": 3.037754535675049,
      "step": 2558
    },
    {
      "epoch": 0.35223783392077856,
      "grad_norm": 0.2959519326686859,
      "learning_rate": 0.002,
      "loss": 3.0373597145080566,
      "step": 2560
    },
    {
      "epoch": 0.3525130197285292,
      "grad_norm": 0.08101298660039902,
      "learning_rate": 0.002,
      "loss": 3.005772113800049,
      "step": 2562
    },
    {
      "epoch": 0.3527882055362798,
      "grad_norm": 0.2444789707660675,
      "learning_rate": 0.002,
      "loss": 3.018239974975586,
      "step": 2564
    },
    {
      "epoch": 0.3530633913440304,
      "grad_norm": 0.10789593309164047,
      "learning_rate": 0.002,
      "loss": 3.0262398719787598,
      "step": 2566
    },
    {
      "epoch": 0.353338577151781,
      "grad_norm": 0.06887687742710114,
      "learning_rate": 0.002,
      "loss": 2.9951601028442383,
      "step": 2568
    },
    {
      "epoch": 0.3536137629595316,
      "grad_norm": 0.12360434234142303,
      "learning_rate": 0.002,
      "loss": 2.9975204467773438,
      "step": 2570
    },
    {
      "epoch": 0.3538889487672822,
      "grad_norm": 0.03741439804434776,
      "learning_rate": 0.002,
      "loss": 3.0023083686828613,
      "step": 2572
    },
    {
      "epoch": 0.3541641345750329,
      "grad_norm": 0.20061802864074707,
      "learning_rate": 0.002,
      "loss": 2.9657974243164062,
      "step": 2574
    },
    {
      "epoch": 0.3544393203827835,
      "grad_norm": 0.07832284271717072,
      "learning_rate": 0.002,
      "loss": 3.0171141624450684,
      "step": 2576
    },
    {
      "epoch": 0.3547145061905341,
      "grad_norm": 0.0753929540514946,
      "learning_rate": 0.002,
      "loss": 3.024580478668213,
      "step": 2578
    },
    {
      "epoch": 0.3549896919982847,
      "grad_norm": 0.03504106402397156,
      "learning_rate": 0.002,
      "loss": 3.0042033195495605,
      "step": 2580
    },
    {
      "epoch": 0.3552648778060353,
      "grad_norm": 0.13808442652225494,
      "learning_rate": 0.002,
      "loss": 3.003207206726074,
      "step": 2582
    },
    {
      "epoch": 0.3555400636137859,
      "grad_norm": 0.13467465341091156,
      "learning_rate": 0.002,
      "loss": 3.014716863632202,
      "step": 2584
    },
    {
      "epoch": 0.3558152494215365,
      "grad_norm": 0.04015209898352623,
      "learning_rate": 0.002,
      "loss": 2.990072727203369,
      "step": 2586
    },
    {
      "epoch": 0.35609043522928713,
      "grad_norm": 0.03905940800905228,
      "learning_rate": 0.002,
      "loss": 3.022135019302368,
      "step": 2588
    },
    {
      "epoch": 0.3563656210370377,
      "grad_norm": 0.15236590802669525,
      "learning_rate": 0.002,
      "loss": 3.036802291870117,
      "step": 2590
    },
    {
      "epoch": 0.3566408068447883,
      "grad_norm": 0.26712706685066223,
      "learning_rate": 0.002,
      "loss": 3.0077314376831055,
      "step": 2592
    },
    {
      "epoch": 0.3569159926525389,
      "grad_norm": 0.562771737575531,
      "learning_rate": 0.002,
      "loss": 2.999943733215332,
      "step": 2594
    },
    {
      "epoch": 0.3571911784602895,
      "grad_norm": 0.12422212213277817,
      "learning_rate": 0.002,
      "loss": 3.019263744354248,
      "step": 2596
    },
    {
      "epoch": 0.3574663642680401,
      "grad_norm": 0.18729840219020844,
      "learning_rate": 0.002,
      "loss": 2.967679500579834,
      "step": 2598
    },
    {
      "epoch": 0.3577415500757908,
      "grad_norm": 0.25022780895233154,
      "learning_rate": 0.002,
      "loss": 3.016345262527466,
      "step": 2600
    },
    {
      "epoch": 0.3580167358835414,
      "grad_norm": 0.04599983990192413,
      "learning_rate": 0.002,
      "loss": 3.0075883865356445,
      "step": 2602
    },
    {
      "epoch": 0.358291921691292,
      "grad_norm": 0.08380816876888275,
      "learning_rate": 0.002,
      "loss": 3.0218405723571777,
      "step": 2604
    },
    {
      "epoch": 0.3585671074990426,
      "grad_norm": 0.0648907795548439,
      "learning_rate": 0.002,
      "loss": 3.0194287300109863,
      "step": 2606
    },
    {
      "epoch": 0.3588422933067932,
      "grad_norm": 0.03607875481247902,
      "learning_rate": 0.002,
      "loss": 2.999748945236206,
      "step": 2608
    },
    {
      "epoch": 0.3591174791145438,
      "grad_norm": 0.03503335639834404,
      "learning_rate": 0.002,
      "loss": 3.0219998359680176,
      "step": 2610
    },
    {
      "epoch": 0.3593926649222944,
      "grad_norm": 0.06545288115739822,
      "learning_rate": 0.002,
      "loss": 3.0454554557800293,
      "step": 2612
    },
    {
      "epoch": 0.35966785073004504,
      "grad_norm": 0.17574743926525116,
      "learning_rate": 0.002,
      "loss": 3.0085561275482178,
      "step": 2614
    },
    {
      "epoch": 0.35994303653779564,
      "grad_norm": 0.07439739257097244,
      "learning_rate": 0.002,
      "loss": 2.974316120147705,
      "step": 2616
    },
    {
      "epoch": 0.36021822234554624,
      "grad_norm": 0.05371567979454994,
      "learning_rate": 0.002,
      "loss": 3.0085158348083496,
      "step": 2618
    },
    {
      "epoch": 0.36049340815329683,
      "grad_norm": 0.04352811723947525,
      "learning_rate": 0.002,
      "loss": 3.0518012046813965,
      "step": 2620
    },
    {
      "epoch": 0.36076859396104743,
      "grad_norm": 0.043870143592357635,
      "learning_rate": 0.002,
      "loss": 2.980942964553833,
      "step": 2622
    },
    {
      "epoch": 0.36104377976879803,
      "grad_norm": 0.13445618748664856,
      "learning_rate": 0.002,
      "loss": 3.024482250213623,
      "step": 2624
    },
    {
      "epoch": 0.3613189655765487,
      "grad_norm": 0.29000797867774963,
      "learning_rate": 0.002,
      "loss": 2.999288558959961,
      "step": 2626
    },
    {
      "epoch": 0.3615941513842993,
      "grad_norm": 0.2003462165594101,
      "learning_rate": 0.002,
      "loss": 3.0087242126464844,
      "step": 2628
    },
    {
      "epoch": 0.3618693371920499,
      "grad_norm": 0.09690649062395096,
      "learning_rate": 0.002,
      "loss": 2.9604406356811523,
      "step": 2630
    },
    {
      "epoch": 0.3621445229998005,
      "grad_norm": 0.37842291593551636,
      "learning_rate": 0.002,
      "loss": 3.0122780799865723,
      "step": 2632
    },
    {
      "epoch": 0.3624197088075511,
      "grad_norm": 0.036301251500844955,
      "learning_rate": 0.002,
      "loss": 2.9955267906188965,
      "step": 2634
    },
    {
      "epoch": 0.3626948946153017,
      "grad_norm": 0.3072527348995209,
      "learning_rate": 0.002,
      "loss": 3.004340171813965,
      "step": 2636
    },
    {
      "epoch": 0.3629700804230523,
      "grad_norm": 0.17611417174339294,
      "learning_rate": 0.002,
      "loss": 2.9949259757995605,
      "step": 2638
    },
    {
      "epoch": 0.36324526623080294,
      "grad_norm": 0.13078387081623077,
      "learning_rate": 0.002,
      "loss": 3.024749279022217,
      "step": 2640
    },
    {
      "epoch": 0.36352045203855354,
      "grad_norm": 0.13333213329315186,
      "learning_rate": 0.002,
      "loss": 3.059798002243042,
      "step": 2642
    },
    {
      "epoch": 0.36379563784630414,
      "grad_norm": 0.039148688316345215,
      "learning_rate": 0.002,
      "loss": 3.023632526397705,
      "step": 2644
    },
    {
      "epoch": 0.36407082365405474,
      "grad_norm": 0.08456747233867645,
      "learning_rate": 0.002,
      "loss": 3.0356080532073975,
      "step": 2646
    },
    {
      "epoch": 0.36434600946180534,
      "grad_norm": 0.10805362462997437,
      "learning_rate": 0.002,
      "loss": 3.0241665840148926,
      "step": 2648
    },
    {
      "epoch": 0.36462119526955594,
      "grad_norm": 0.11277935653924942,
      "learning_rate": 0.002,
      "loss": 3.0143065452575684,
      "step": 2650
    },
    {
      "epoch": 0.3648963810773066,
      "grad_norm": 0.05675135552883148,
      "learning_rate": 0.002,
      "loss": 3.0292468070983887,
      "step": 2652
    },
    {
      "epoch": 0.3651715668850572,
      "grad_norm": 0.10213879495859146,
      "learning_rate": 0.002,
      "loss": 2.9992222785949707,
      "step": 2654
    },
    {
      "epoch": 0.3654467526928078,
      "grad_norm": 0.12808886170387268,
      "learning_rate": 0.002,
      "loss": 2.993600845336914,
      "step": 2656
    },
    {
      "epoch": 0.3657219385005584,
      "grad_norm": 0.0355711467564106,
      "learning_rate": 0.002,
      "loss": 2.9781222343444824,
      "step": 2658
    },
    {
      "epoch": 0.365997124308309,
      "grad_norm": 0.16497758030891418,
      "learning_rate": 0.002,
      "loss": 2.9618306159973145,
      "step": 2660
    },
    {
      "epoch": 0.3662723101160596,
      "grad_norm": 0.1986156553030014,
      "learning_rate": 0.002,
      "loss": 2.9986257553100586,
      "step": 2662
    },
    {
      "epoch": 0.3665474959238102,
      "grad_norm": 0.03340606763958931,
      "learning_rate": 0.002,
      "loss": 2.9594359397888184,
      "step": 2664
    },
    {
      "epoch": 0.36682268173156085,
      "grad_norm": 0.097864530980587,
      "learning_rate": 0.002,
      "loss": 2.9924983978271484,
      "step": 2666
    },
    {
      "epoch": 0.36709786753931145,
      "grad_norm": 0.13379201292991638,
      "learning_rate": 0.002,
      "loss": 2.970318555831909,
      "step": 2668
    },
    {
      "epoch": 0.36737305334706205,
      "grad_norm": 0.13183875381946564,
      "learning_rate": 0.002,
      "loss": 3.009033679962158,
      "step": 2670
    },
    {
      "epoch": 0.36764823915481265,
      "grad_norm": 0.07732298225164413,
      "learning_rate": 0.002,
      "loss": 2.9690394401550293,
      "step": 2672
    },
    {
      "epoch": 0.36792342496256325,
      "grad_norm": 0.16573114693164825,
      "learning_rate": 0.002,
      "loss": 2.9833288192749023,
      "step": 2674
    },
    {
      "epoch": 0.36819861077031385,
      "grad_norm": 0.26922324299812317,
      "learning_rate": 0.002,
      "loss": 3.005220890045166,
      "step": 2676
    },
    {
      "epoch": 0.3684737965780645,
      "grad_norm": 0.21578650176525116,
      "learning_rate": 0.002,
      "loss": 2.9611010551452637,
      "step": 2678
    },
    {
      "epoch": 0.3687489823858151,
      "grad_norm": 0.09986850619316101,
      "learning_rate": 0.002,
      "loss": 3.0167717933654785,
      "step": 2680
    },
    {
      "epoch": 0.3690241681935657,
      "grad_norm": 0.28671666979789734,
      "learning_rate": 0.002,
      "loss": 3.0170512199401855,
      "step": 2682
    },
    {
      "epoch": 0.3692993540013163,
      "grad_norm": 0.04645860567688942,
      "learning_rate": 0.002,
      "loss": 3.029151439666748,
      "step": 2684
    },
    {
      "epoch": 0.3695745398090669,
      "grad_norm": 0.20122700929641724,
      "learning_rate": 0.002,
      "loss": 2.990776538848877,
      "step": 2686
    },
    {
      "epoch": 0.3698497256168175,
      "grad_norm": 0.11683423817157745,
      "learning_rate": 0.002,
      "loss": 2.980980396270752,
      "step": 2688
    },
    {
      "epoch": 0.3701249114245681,
      "grad_norm": 0.12612441182136536,
      "learning_rate": 0.002,
      "loss": 2.9415926933288574,
      "step": 2690
    },
    {
      "epoch": 0.37040009723231876,
      "grad_norm": 0.14240092039108276,
      "learning_rate": 0.002,
      "loss": 2.960864543914795,
      "step": 2692
    },
    {
      "epoch": 0.37067528304006936,
      "grad_norm": 0.04127832129597664,
      "learning_rate": 0.002,
      "loss": 2.9806699752807617,
      "step": 2694
    },
    {
      "epoch": 0.37095046884781996,
      "grad_norm": 0.045567695051431656,
      "learning_rate": 0.002,
      "loss": 2.962446689605713,
      "step": 2696
    },
    {
      "epoch": 0.37122565465557056,
      "grad_norm": 0.08326444774866104,
      "learning_rate": 0.002,
      "loss": 2.976284980773926,
      "step": 2698
    },
    {
      "epoch": 0.37150084046332116,
      "grad_norm": 0.06327324360609055,
      "learning_rate": 0.002,
      "loss": 3.0066847801208496,
      "step": 2700
    },
    {
      "epoch": 0.37177602627107176,
      "grad_norm": 0.05907059833407402,
      "learning_rate": 0.002,
      "loss": 2.982354164123535,
      "step": 2702
    },
    {
      "epoch": 0.3720512120788224,
      "grad_norm": 0.12017647176980972,
      "learning_rate": 0.002,
      "loss": 2.962538957595825,
      "step": 2704
    },
    {
      "epoch": 0.372326397886573,
      "grad_norm": 0.21821916103363037,
      "learning_rate": 0.002,
      "loss": 3.010695457458496,
      "step": 2706
    },
    {
      "epoch": 0.3726015836943236,
      "grad_norm": 0.25848957896232605,
      "learning_rate": 0.002,
      "loss": 3.005621910095215,
      "step": 2708
    },
    {
      "epoch": 0.3728767695020742,
      "grad_norm": 0.06917940080165863,
      "learning_rate": 0.002,
      "loss": 2.9957449436187744,
      "step": 2710
    },
    {
      "epoch": 0.3731519553098248,
      "grad_norm": 0.215652734041214,
      "learning_rate": 0.002,
      "loss": 2.9788079261779785,
      "step": 2712
    },
    {
      "epoch": 0.3734271411175754,
      "grad_norm": 0.18186359107494354,
      "learning_rate": 0.002,
      "loss": 2.997440814971924,
      "step": 2714
    },
    {
      "epoch": 0.373702326925326,
      "grad_norm": 0.20062273740768433,
      "learning_rate": 0.002,
      "loss": 3.0261783599853516,
      "step": 2716
    },
    {
      "epoch": 0.37397751273307667,
      "grad_norm": 0.03665821626782417,
      "learning_rate": 0.002,
      "loss": 2.9676337242126465,
      "step": 2718
    },
    {
      "epoch": 0.37425269854082727,
      "grad_norm": 0.13791701197624207,
      "learning_rate": 0.002,
      "loss": 3.020658016204834,
      "step": 2720
    },
    {
      "epoch": 0.37452788434857787,
      "grad_norm": 0.09557662904262543,
      "learning_rate": 0.002,
      "loss": 3.003833770751953,
      "step": 2722
    },
    {
      "epoch": 0.37480307015632847,
      "grad_norm": 0.17084650695323944,
      "learning_rate": 0.002,
      "loss": 3.0020546913146973,
      "step": 2724
    },
    {
      "epoch": 0.37507825596407907,
      "grad_norm": 0.30765339732170105,
      "learning_rate": 0.002,
      "loss": 3.0137476921081543,
      "step": 2726
    },
    {
      "epoch": 0.37535344177182967,
      "grad_norm": 0.04414133355021477,
      "learning_rate": 0.002,
      "loss": 2.9953010082244873,
      "step": 2728
    },
    {
      "epoch": 0.3756286275795803,
      "grad_norm": 0.1827855408191681,
      "learning_rate": 0.002,
      "loss": 3.0283594131469727,
      "step": 2730
    },
    {
      "epoch": 0.3759038133873309,
      "grad_norm": 0.06280441582202911,
      "learning_rate": 0.002,
      "loss": 2.983724594116211,
      "step": 2732
    },
    {
      "epoch": 0.3761789991950815,
      "grad_norm": 0.2127757966518402,
      "learning_rate": 0.002,
      "loss": 3.0224008560180664,
      "step": 2734
    },
    {
      "epoch": 0.3764541850028321,
      "grad_norm": 0.2208549976348877,
      "learning_rate": 0.002,
      "loss": 3.0110771656036377,
      "step": 2736
    },
    {
      "epoch": 0.3767293708105827,
      "grad_norm": 0.05795064568519592,
      "learning_rate": 0.002,
      "loss": 2.96907901763916,
      "step": 2738
    },
    {
      "epoch": 0.3770045566183333,
      "grad_norm": 0.10728857666254044,
      "learning_rate": 0.002,
      "loss": 3.011197566986084,
      "step": 2740
    },
    {
      "epoch": 0.3772797424260839,
      "grad_norm": 0.0371382050216198,
      "learning_rate": 0.002,
      "loss": 2.992786407470703,
      "step": 2742
    },
    {
      "epoch": 0.3775549282338346,
      "grad_norm": 0.03503953292965889,
      "learning_rate": 0.002,
      "loss": 3.0098819732666016,
      "step": 2744
    },
    {
      "epoch": 0.3778301140415852,
      "grad_norm": 0.040539853274822235,
      "learning_rate": 0.002,
      "loss": 2.9428963661193848,
      "step": 2746
    },
    {
      "epoch": 0.3781052998493358,
      "grad_norm": 0.0725243091583252,
      "learning_rate": 0.002,
      "loss": 2.98580265045166,
      "step": 2748
    },
    {
      "epoch": 0.3783804856570864,
      "grad_norm": 0.0698922798037529,
      "learning_rate": 0.002,
      "loss": 3.013659954071045,
      "step": 2750
    },
    {
      "epoch": 0.378655671464837,
      "grad_norm": 0.13633966445922852,
      "learning_rate": 0.002,
      "loss": 2.954822540283203,
      "step": 2752
    },
    {
      "epoch": 0.3789308572725876,
      "grad_norm": 0.24925346672534943,
      "learning_rate": 0.002,
      "loss": 3.0334653854370117,
      "step": 2754
    },
    {
      "epoch": 0.37920604308033823,
      "grad_norm": 0.1526431143283844,
      "learning_rate": 0.002,
      "loss": 2.9761362075805664,
      "step": 2756
    },
    {
      "epoch": 0.37948122888808883,
      "grad_norm": 0.11551204323768616,
      "learning_rate": 0.002,
      "loss": 3.046640396118164,
      "step": 2758
    },
    {
      "epoch": 0.37975641469583943,
      "grad_norm": 0.08599456399679184,
      "learning_rate": 0.002,
      "loss": 2.9551548957824707,
      "step": 2760
    },
    {
      "epoch": 0.38003160050359003,
      "grad_norm": 0.2000586986541748,
      "learning_rate": 0.002,
      "loss": 3.0184812545776367,
      "step": 2762
    },
    {
      "epoch": 0.3803067863113406,
      "grad_norm": 0.23858925700187683,
      "learning_rate": 0.002,
      "loss": 2.9677894115448,
      "step": 2764
    },
    {
      "epoch": 0.3805819721190912,
      "grad_norm": 0.14023327827453613,
      "learning_rate": 0.002,
      "loss": 2.981550455093384,
      "step": 2766
    },
    {
      "epoch": 0.3808571579268418,
      "grad_norm": 0.13828496634960175,
      "learning_rate": 0.002,
      "loss": 2.915719509124756,
      "step": 2768
    },
    {
      "epoch": 0.3811323437345925,
      "grad_norm": 0.18346083164215088,
      "learning_rate": 0.002,
      "loss": 3.010589122772217,
      "step": 2770
    },
    {
      "epoch": 0.3814075295423431,
      "grad_norm": 0.24073930084705353,
      "learning_rate": 0.002,
      "loss": 2.937084913253784,
      "step": 2772
    },
    {
      "epoch": 0.3816827153500937,
      "grad_norm": 0.07312172651290894,
      "learning_rate": 0.002,
      "loss": 3.0034425258636475,
      "step": 2774
    },
    {
      "epoch": 0.3819579011578443,
      "grad_norm": 0.17252613604068756,
      "learning_rate": 0.002,
      "loss": 2.9602904319763184,
      "step": 2776
    },
    {
      "epoch": 0.3822330869655949,
      "grad_norm": 0.051381614059209824,
      "learning_rate": 0.002,
      "loss": 2.9816179275512695,
      "step": 2778
    },
    {
      "epoch": 0.3825082727733455,
      "grad_norm": 0.09209275990724564,
      "learning_rate": 0.002,
      "loss": 2.953120231628418,
      "step": 2780
    },
    {
      "epoch": 0.38278345858109614,
      "grad_norm": 0.0627182275056839,
      "learning_rate": 0.002,
      "loss": 2.9760630130767822,
      "step": 2782
    },
    {
      "epoch": 0.38305864438884674,
      "grad_norm": 0.0552746057510376,
      "learning_rate": 0.002,
      "loss": 2.948075771331787,
      "step": 2784
    },
    {
      "epoch": 0.38333383019659734,
      "grad_norm": 0.21461524069309235,
      "learning_rate": 0.002,
      "loss": 2.992304563522339,
      "step": 2786
    },
    {
      "epoch": 0.38360901600434794,
      "grad_norm": 0.1556852161884308,
      "learning_rate": 0.002,
      "loss": 2.9553134441375732,
      "step": 2788
    },
    {
      "epoch": 0.38388420181209854,
      "grad_norm": 0.08700846135616302,
      "learning_rate": 0.002,
      "loss": 2.9584665298461914,
      "step": 2790
    },
    {
      "epoch": 0.38415938761984914,
      "grad_norm": 0.06548883765935898,
      "learning_rate": 0.002,
      "loss": 2.9867982864379883,
      "step": 2792
    },
    {
      "epoch": 0.38443457342759974,
      "grad_norm": 0.08365391194820404,
      "learning_rate": 0.002,
      "loss": 2.9723215103149414,
      "step": 2794
    },
    {
      "epoch": 0.3847097592353504,
      "grad_norm": 0.045336924493312836,
      "learning_rate": 0.002,
      "loss": 2.9781386852264404,
      "step": 2796
    },
    {
      "epoch": 0.384984945043101,
      "grad_norm": 0.08787307888269424,
      "learning_rate": 0.002,
      "loss": 3.021118640899658,
      "step": 2798
    },
    {
      "epoch": 0.3852601308508516,
      "grad_norm": 0.22257691621780396,
      "learning_rate": 0.002,
      "loss": 2.986105442047119,
      "step": 2800
    },
    {
      "epoch": 0.3855353166586022,
      "grad_norm": 0.3544997274875641,
      "learning_rate": 0.002,
      "loss": 2.9893369674682617,
      "step": 2802
    },
    {
      "epoch": 0.3858105024663528,
      "grad_norm": 0.08036980777978897,
      "learning_rate": 0.002,
      "loss": 2.98343563079834,
      "step": 2804
    },
    {
      "epoch": 0.3860856882741034,
      "grad_norm": 0.1952127367258072,
      "learning_rate": 0.002,
      "loss": 2.9882864952087402,
      "step": 2806
    },
    {
      "epoch": 0.38636087408185404,
      "grad_norm": 0.051052600145339966,
      "learning_rate": 0.002,
      "loss": 2.9532828330993652,
      "step": 2808
    },
    {
      "epoch": 0.38663605988960464,
      "grad_norm": 0.11808133125305176,
      "learning_rate": 0.002,
      "loss": 2.9557690620422363,
      "step": 2810
    },
    {
      "epoch": 0.38691124569735524,
      "grad_norm": 0.1521487832069397,
      "learning_rate": 0.002,
      "loss": 3.0015273094177246,
      "step": 2812
    },
    {
      "epoch": 0.38718643150510584,
      "grad_norm": 0.07155986875295639,
      "learning_rate": 0.002,
      "loss": 2.9899377822875977,
      "step": 2814
    },
    {
      "epoch": 0.38746161731285644,
      "grad_norm": 0.08077829331159592,
      "learning_rate": 0.002,
      "loss": 2.9980616569519043,
      "step": 2816
    },
    {
      "epoch": 0.38773680312060704,
      "grad_norm": 0.12973977625370026,
      "learning_rate": 0.002,
      "loss": 3.006920337677002,
      "step": 2818
    },
    {
      "epoch": 0.38801198892835764,
      "grad_norm": 0.1251581907272339,
      "learning_rate": 0.002,
      "loss": 2.948910713195801,
      "step": 2820
    },
    {
      "epoch": 0.3882871747361083,
      "grad_norm": 0.03932889178395271,
      "learning_rate": 0.002,
      "loss": 2.9767768383026123,
      "step": 2822
    },
    {
      "epoch": 0.3885623605438589,
      "grad_norm": 0.09250680357217789,
      "learning_rate": 0.002,
      "loss": 2.9708597660064697,
      "step": 2824
    },
    {
      "epoch": 0.3888375463516095,
      "grad_norm": 0.1022181585431099,
      "learning_rate": 0.002,
      "loss": 2.984832763671875,
      "step": 2826
    },
    {
      "epoch": 0.3891127321593601,
      "grad_norm": 0.05209087207913399,
      "learning_rate": 0.002,
      "loss": 2.9734244346618652,
      "step": 2828
    },
    {
      "epoch": 0.3893879179671107,
      "grad_norm": 0.0819583311676979,
      "learning_rate": 0.002,
      "loss": 2.9551773071289062,
      "step": 2830
    },
    {
      "epoch": 0.3896631037748613,
      "grad_norm": 0.23944196105003357,
      "learning_rate": 0.002,
      "loss": 2.946262836456299,
      "step": 2832
    },
    {
      "epoch": 0.38993828958261195,
      "grad_norm": 0.36911776661872864,
      "learning_rate": 0.002,
      "loss": 2.948697566986084,
      "step": 2834
    },
    {
      "epoch": 0.39021347539036255,
      "grad_norm": 0.06162373721599579,
      "learning_rate": 0.002,
      "loss": 2.9608192443847656,
      "step": 2836
    },
    {
      "epoch": 0.39048866119811315,
      "grad_norm": 0.3542216122150421,
      "learning_rate": 0.002,
      "loss": 2.9898593425750732,
      "step": 2838
    },
    {
      "epoch": 0.39076384700586375,
      "grad_norm": 0.062095217406749725,
      "learning_rate": 0.002,
      "loss": 2.9450597763061523,
      "step": 2840
    },
    {
      "epoch": 0.39103903281361435,
      "grad_norm": 0.17705148458480835,
      "learning_rate": 0.002,
      "loss": 2.994359254837036,
      "step": 2842
    },
    {
      "epoch": 0.39131421862136495,
      "grad_norm": 0.18082134425640106,
      "learning_rate": 0.002,
      "loss": 2.97509503364563,
      "step": 2844
    },
    {
      "epoch": 0.39158940442911555,
      "grad_norm": 0.04594917222857475,
      "learning_rate": 0.002,
      "loss": 3.002420425415039,
      "step": 2846
    },
    {
      "epoch": 0.3918645902368662,
      "grad_norm": 0.10350244492292404,
      "learning_rate": 0.002,
      "loss": 2.964691162109375,
      "step": 2848
    },
    {
      "epoch": 0.3921397760446168,
      "grad_norm": 0.07899139076471329,
      "learning_rate": 0.002,
      "loss": 2.94370698928833,
      "step": 2850
    },
    {
      "epoch": 0.3924149618523674,
      "grad_norm": 0.03256286680698395,
      "learning_rate": 0.002,
      "loss": 2.971632957458496,
      "step": 2852
    },
    {
      "epoch": 0.392690147660118,
      "grad_norm": 0.041326772421598434,
      "learning_rate": 0.002,
      "loss": 2.973875045776367,
      "step": 2854
    },
    {
      "epoch": 0.3929653334678686,
      "grad_norm": 0.04979491978883743,
      "learning_rate": 0.002,
      "loss": 2.986067533493042,
      "step": 2856
    },
    {
      "epoch": 0.3932405192756192,
      "grad_norm": 0.15646061301231384,
      "learning_rate": 0.002,
      "loss": 2.951663017272949,
      "step": 2858
    },
    {
      "epoch": 0.39351570508336986,
      "grad_norm": 0.2977072298526764,
      "learning_rate": 0.002,
      "loss": 3.0092484951019287,
      "step": 2860
    },
    {
      "epoch": 0.39379089089112046,
      "grad_norm": 0.1390559822320938,
      "learning_rate": 0.002,
      "loss": 2.932865858078003,
      "step": 2862
    },
    {
      "epoch": 0.39406607669887106,
      "grad_norm": 0.11614140123128891,
      "learning_rate": 0.002,
      "loss": 3.0050668716430664,
      "step": 2864
    },
    {
      "epoch": 0.39434126250662166,
      "grad_norm": 0.11546838283538818,
      "learning_rate": 0.002,
      "loss": 2.971442699432373,
      "step": 2866
    },
    {
      "epoch": 0.39461644831437226,
      "grad_norm": 0.05468763783574104,
      "learning_rate": 0.002,
      "loss": 2.9631850719451904,
      "step": 2868
    },
    {
      "epoch": 0.39489163412212286,
      "grad_norm": 0.1292220652103424,
      "learning_rate": 0.002,
      "loss": 2.985287666320801,
      "step": 2870
    },
    {
      "epoch": 0.39516681992987346,
      "grad_norm": 0.21875296533107758,
      "learning_rate": 0.002,
      "loss": 2.967909574508667,
      "step": 2872
    },
    {
      "epoch": 0.3954420057376241,
      "grad_norm": 0.043798383325338364,
      "learning_rate": 0.002,
      "loss": 2.9087395668029785,
      "step": 2874
    },
    {
      "epoch": 0.3957171915453747,
      "grad_norm": 0.1863406002521515,
      "learning_rate": 0.002,
      "loss": 2.9650683403015137,
      "step": 2876
    },
    {
      "epoch": 0.3959923773531253,
      "grad_norm": 0.07814750075340271,
      "learning_rate": 0.002,
      "loss": 3.030445098876953,
      "step": 2878
    },
    {
      "epoch": 0.3962675631608759,
      "grad_norm": 0.31983205676078796,
      "learning_rate": 0.002,
      "loss": 2.941957950592041,
      "step": 2880
    },
    {
      "epoch": 0.3965427489686265,
      "grad_norm": 0.09276070445775986,
      "learning_rate": 0.002,
      "loss": 2.9333698749542236,
      "step": 2882
    },
    {
      "epoch": 0.3968179347763771,
      "grad_norm": 0.1728220283985138,
      "learning_rate": 0.002,
      "loss": 2.986785888671875,
      "step": 2884
    },
    {
      "epoch": 0.39709312058412777,
      "grad_norm": 0.1111244484782219,
      "learning_rate": 0.002,
      "loss": 3.0019967555999756,
      "step": 2886
    },
    {
      "epoch": 0.39736830639187837,
      "grad_norm": 0.14429162442684174,
      "learning_rate": 0.002,
      "loss": 2.968463897705078,
      "step": 2888
    },
    {
      "epoch": 0.39764349219962897,
      "grad_norm": 0.11534049361944199,
      "learning_rate": 0.002,
      "loss": 2.9847400188446045,
      "step": 2890
    },
    {
      "epoch": 0.39791867800737957,
      "grad_norm": 0.20330455899238586,
      "learning_rate": 0.002,
      "loss": 2.9337713718414307,
      "step": 2892
    },
    {
      "epoch": 0.39819386381513017,
      "grad_norm": 0.03470835089683533,
      "learning_rate": 0.002,
      "loss": 2.923461437225342,
      "step": 2894
    },
    {
      "epoch": 0.39846904962288077,
      "grad_norm": 0.1396961361169815,
      "learning_rate": 0.002,
      "loss": 2.97109055519104,
      "step": 2896
    },
    {
      "epoch": 0.39874423543063137,
      "grad_norm": 0.06475773453712463,
      "learning_rate": 0.002,
      "loss": 3.0181167125701904,
      "step": 2898
    },
    {
      "epoch": 0.399019421238382,
      "grad_norm": 0.08129701018333435,
      "learning_rate": 0.002,
      "loss": 2.9546051025390625,
      "step": 2900
    },
    {
      "epoch": 0.3992946070461326,
      "grad_norm": 0.20379124581813812,
      "learning_rate": 0.002,
      "loss": 2.9539923667907715,
      "step": 2902
    },
    {
      "epoch": 0.3995697928538832,
      "grad_norm": 0.13264989852905273,
      "learning_rate": 0.002,
      "loss": 3.0038866996765137,
      "step": 2904
    },
    {
      "epoch": 0.3998449786616338,
      "grad_norm": 0.05123626068234444,
      "learning_rate": 0.002,
      "loss": 2.9384427070617676,
      "step": 2906
    },
    {
      "epoch": 0.4001201644693844,
      "grad_norm": 0.05676788464188576,
      "learning_rate": 0.002,
      "loss": 2.9487414360046387,
      "step": 2908
    },
    {
      "epoch": 0.400395350277135,
      "grad_norm": 0.040738411247730255,
      "learning_rate": 0.002,
      "loss": 2.9781460762023926,
      "step": 2910
    },
    {
      "epoch": 0.4006705360848857,
      "grad_norm": 0.06402325630187988,
      "learning_rate": 0.002,
      "loss": 2.9562106132507324,
      "step": 2912
    },
    {
      "epoch": 0.4009457218926363,
      "grad_norm": 0.14263619482517242,
      "learning_rate": 0.002,
      "loss": 2.9881949424743652,
      "step": 2914
    },
    {
      "epoch": 0.4012209077003869,
      "grad_norm": 0.34028127789497375,
      "learning_rate": 0.002,
      "loss": 2.961427688598633,
      "step": 2916
    },
    {
      "epoch": 0.4014960935081375,
      "grad_norm": 0.2898019254207611,
      "learning_rate": 0.002,
      "loss": 2.943721055984497,
      "step": 2918
    },
    {
      "epoch": 0.4017712793158881,
      "grad_norm": 0.23261834681034088,
      "learning_rate": 0.002,
      "loss": 2.928921699523926,
      "step": 2920
    },
    {
      "epoch": 0.4020464651236387,
      "grad_norm": 0.10179512947797775,
      "learning_rate": 0.002,
      "loss": 2.902559995651245,
      "step": 2922
    },
    {
      "epoch": 0.4023216509313893,
      "grad_norm": 0.14414535462856293,
      "learning_rate": 0.002,
      "loss": 2.9567272663116455,
      "step": 2924
    },
    {
      "epoch": 0.40259683673913993,
      "grad_norm": 0.16425804793834686,
      "learning_rate": 0.002,
      "loss": 2.964517593383789,
      "step": 2926
    },
    {
      "epoch": 0.40287202254689053,
      "grad_norm": 0.049826059490442276,
      "learning_rate": 0.002,
      "loss": 2.977527618408203,
      "step": 2928
    },
    {
      "epoch": 0.40314720835464113,
      "grad_norm": 0.045704737305641174,
      "learning_rate": 0.002,
      "loss": 2.99827241897583,
      "step": 2930
    },
    {
      "epoch": 0.40342239416239173,
      "grad_norm": 0.041943684220314026,
      "learning_rate": 0.002,
      "loss": 2.906802177429199,
      "step": 2932
    },
    {
      "epoch": 0.40369757997014233,
      "grad_norm": 0.09103121608495712,
      "learning_rate": 0.002,
      "loss": 2.951206684112549,
      "step": 2934
    },
    {
      "epoch": 0.40397276577789293,
      "grad_norm": 0.05064282938838005,
      "learning_rate": 0.002,
      "loss": 2.998269557952881,
      "step": 2936
    },
    {
      "epoch": 0.4042479515856436,
      "grad_norm": 0.10912565141916275,
      "learning_rate": 0.002,
      "loss": 2.9207401275634766,
      "step": 2938
    },
    {
      "epoch": 0.4045231373933942,
      "grad_norm": 0.03628617897629738,
      "learning_rate": 0.002,
      "loss": 2.9552464485168457,
      "step": 2940
    },
    {
      "epoch": 0.4047983232011448,
      "grad_norm": 0.052804265171289444,
      "learning_rate": 0.002,
      "loss": 2.9219725131988525,
      "step": 2942
    },
    {
      "epoch": 0.4050735090088954,
      "grad_norm": 0.0699649527668953,
      "learning_rate": 0.002,
      "loss": 2.9549269676208496,
      "step": 2944
    },
    {
      "epoch": 0.405348694816646,
      "grad_norm": 0.1920134425163269,
      "learning_rate": 0.002,
      "loss": 2.92965030670166,
      "step": 2946
    },
    {
      "epoch": 0.4056238806243966,
      "grad_norm": 0.26489344239234924,
      "learning_rate": 0.002,
      "loss": 2.964869499206543,
      "step": 2948
    },
    {
      "epoch": 0.4058990664321472,
      "grad_norm": 0.17100246250629425,
      "learning_rate": 0.002,
      "loss": 2.9693710803985596,
      "step": 2950
    },
    {
      "epoch": 0.40617425223989784,
      "grad_norm": 0.2935944199562073,
      "learning_rate": 0.002,
      "loss": 2.9676661491394043,
      "step": 2952
    },
    {
      "epoch": 0.40644943804764844,
      "grad_norm": 0.2456170618534088,
      "learning_rate": 0.002,
      "loss": 2.9451417922973633,
      "step": 2954
    },
    {
      "epoch": 0.40672462385539904,
      "grad_norm": 0.15335728228092194,
      "learning_rate": 0.002,
      "loss": 2.9033427238464355,
      "step": 2956
    },
    {
      "epoch": 0.40699980966314964,
      "grad_norm": 0.08329708874225616,
      "learning_rate": 0.002,
      "loss": 2.981393814086914,
      "step": 2958
    },
    {
      "epoch": 0.40727499547090024,
      "grad_norm": 0.0366833433508873,
      "learning_rate": 0.002,
      "loss": 2.9482431411743164,
      "step": 2960
    },
    {
      "epoch": 0.40755018127865084,
      "grad_norm": 0.06313035637140274,
      "learning_rate": 0.002,
      "loss": 2.9260916709899902,
      "step": 2962
    },
    {
      "epoch": 0.4078253670864015,
      "grad_norm": 0.08443814516067505,
      "learning_rate": 0.002,
      "loss": 2.9767355918884277,
      "step": 2964
    },
    {
      "epoch": 0.4081005528941521,
      "grad_norm": 0.041709836572408676,
      "learning_rate": 0.002,
      "loss": 2.945765256881714,
      "step": 2966
    },
    {
      "epoch": 0.4083757387019027,
      "grad_norm": 0.038367271423339844,
      "learning_rate": 0.002,
      "loss": 2.925506830215454,
      "step": 2968
    },
    {
      "epoch": 0.4086509245096533,
      "grad_norm": 0.03307550400495529,
      "learning_rate": 0.002,
      "loss": 2.9367709159851074,
      "step": 2970
    },
    {
      "epoch": 0.4089261103174039,
      "grad_norm": 0.04454585537314415,
      "learning_rate": 0.002,
      "loss": 2.926302433013916,
      "step": 2972
    },
    {
      "epoch": 0.4092012961251545,
      "grad_norm": 0.03450579568743706,
      "learning_rate": 0.002,
      "loss": 2.978280544281006,
      "step": 2974
    },
    {
      "epoch": 0.4094764819329051,
      "grad_norm": 0.05455109849572182,
      "learning_rate": 0.002,
      "loss": 2.944030284881592,
      "step": 2976
    },
    {
      "epoch": 0.40975166774065575,
      "grad_norm": 0.1240062415599823,
      "learning_rate": 0.002,
      "loss": 2.9069178104400635,
      "step": 2978
    },
    {
      "epoch": 0.41002685354840634,
      "grad_norm": 0.6237960457801819,
      "learning_rate": 0.002,
      "loss": 2.9206395149230957,
      "step": 2980
    },
    {
      "epoch": 0.41030203935615694,
      "grad_norm": 0.2151561826467514,
      "learning_rate": 0.002,
      "loss": 2.9687557220458984,
      "step": 2982
    },
    {
      "epoch": 0.41057722516390754,
      "grad_norm": 0.2809375524520874,
      "learning_rate": 0.002,
      "loss": 2.9292190074920654,
      "step": 2984
    },
    {
      "epoch": 0.41085241097165814,
      "grad_norm": 0.13896550238132477,
      "learning_rate": 0.002,
      "loss": 2.9815304279327393,
      "step": 2986
    },
    {
      "epoch": 0.41112759677940874,
      "grad_norm": 0.0398767851293087,
      "learning_rate": 0.002,
      "loss": 2.9398818016052246,
      "step": 2988
    },
    {
      "epoch": 0.4114027825871594,
      "grad_norm": 0.0913551077246666,
      "learning_rate": 0.002,
      "loss": 2.9454150199890137,
      "step": 2990
    },
    {
      "epoch": 0.41167796839491,
      "grad_norm": 0.1317954659461975,
      "learning_rate": 0.002,
      "loss": 2.932919979095459,
      "step": 2992
    },
    {
      "epoch": 0.4119531542026606,
      "grad_norm": 0.08762674778699875,
      "learning_rate": 0.002,
      "loss": 2.942802906036377,
      "step": 2994
    },
    {
      "epoch": 0.4122283400104112,
      "grad_norm": 0.06637014448642731,
      "learning_rate": 0.002,
      "loss": 2.992483377456665,
      "step": 2996
    },
    {
      "epoch": 0.4125035258181618,
      "grad_norm": 0.16887512803077698,
      "learning_rate": 0.002,
      "loss": 2.9864561557769775,
      "step": 2998
    },
    {
      "epoch": 0.4127787116259124,
      "grad_norm": 0.0954810157418251,
      "learning_rate": 0.002,
      "loss": 2.9209022521972656,
      "step": 3000
    },
    {
      "epoch": 0.413053897433663,
      "grad_norm": 0.20503056049346924,
      "learning_rate": 0.002,
      "loss": 2.9876396656036377,
      "step": 3002
    },
    {
      "epoch": 0.41332908324141365,
      "grad_norm": 0.03991119936108589,
      "learning_rate": 0.002,
      "loss": 2.97524094581604,
      "step": 3004
    },
    {
      "epoch": 0.41360426904916425,
      "grad_norm": 0.12379512190818787,
      "learning_rate": 0.002,
      "loss": 2.9285471439361572,
      "step": 3006
    },
    {
      "epoch": 0.41387945485691485,
      "grad_norm": 0.08463685214519501,
      "learning_rate": 0.002,
      "loss": 2.9793877601623535,
      "step": 3008
    },
    {
      "epoch": 0.41415464066466545,
      "grad_norm": 0.14142043888568878,
      "learning_rate": 0.002,
      "loss": 2.952693462371826,
      "step": 3010
    },
    {
      "epoch": 0.41442982647241605,
      "grad_norm": 0.07723836600780487,
      "learning_rate": 0.002,
      "loss": 2.914024829864502,
      "step": 3012
    },
    {
      "epoch": 0.41470501228016665,
      "grad_norm": 0.11186918616294861,
      "learning_rate": 0.002,
      "loss": 2.9159512519836426,
      "step": 3014
    },
    {
      "epoch": 0.4149801980879173,
      "grad_norm": 0.11188530176877975,
      "learning_rate": 0.002,
      "loss": 2.9634780883789062,
      "step": 3016
    },
    {
      "epoch": 0.4152553838956679,
      "grad_norm": 0.06988155096769333,
      "learning_rate": 0.002,
      "loss": 2.9495205879211426,
      "step": 3018
    },
    {
      "epoch": 0.4155305697034185,
      "grad_norm": 0.07450466603040695,
      "learning_rate": 0.002,
      "loss": 2.9661684036254883,
      "step": 3020
    },
    {
      "epoch": 0.4158057555111691,
      "grad_norm": 0.07617293298244476,
      "learning_rate": 0.002,
      "loss": 2.9638872146606445,
      "step": 3022
    },
    {
      "epoch": 0.4160809413189197,
      "grad_norm": 0.09884940087795258,
      "learning_rate": 0.002,
      "loss": 2.9662137031555176,
      "step": 3024
    },
    {
      "epoch": 0.4163561271266703,
      "grad_norm": 0.1651366949081421,
      "learning_rate": 0.002,
      "loss": 2.9108190536499023,
      "step": 3026
    },
    {
      "epoch": 0.4166313129344209,
      "grad_norm": 0.18354319036006927,
      "learning_rate": 0.002,
      "loss": 2.8954343795776367,
      "step": 3028
    },
    {
      "epoch": 0.41690649874217156,
      "grad_norm": 0.1527041345834732,
      "learning_rate": 0.002,
      "loss": 2.9281411170959473,
      "step": 3030
    },
    {
      "epoch": 0.41718168454992216,
      "grad_norm": 0.07782766968011856,
      "learning_rate": 0.002,
      "loss": 2.9380273818969727,
      "step": 3032
    },
    {
      "epoch": 0.41745687035767276,
      "grad_norm": 0.33022740483283997,
      "learning_rate": 0.002,
      "loss": 2.9708614349365234,
      "step": 3034
    },
    {
      "epoch": 0.41773205616542336,
      "grad_norm": 0.23935087025165558,
      "learning_rate": 0.002,
      "loss": 2.939937114715576,
      "step": 3036
    },
    {
      "epoch": 0.41800724197317396,
      "grad_norm": 0.20309630036354065,
      "learning_rate": 0.002,
      "loss": 2.97261118888855,
      "step": 3038
    },
    {
      "epoch": 0.41828242778092456,
      "grad_norm": 0.1002189889550209,
      "learning_rate": 0.002,
      "loss": 2.933015823364258,
      "step": 3040
    },
    {
      "epoch": 0.41855761358867516,
      "grad_norm": 0.1963382363319397,
      "learning_rate": 0.002,
      "loss": 2.947089672088623,
      "step": 3042
    },
    {
      "epoch": 0.4188327993964258,
      "grad_norm": 0.05509370192885399,
      "learning_rate": 0.002,
      "loss": 2.939573287963867,
      "step": 3044
    },
    {
      "epoch": 0.4191079852041764,
      "grad_norm": 0.17855304479599,
      "learning_rate": 0.002,
      "loss": 2.964900493621826,
      "step": 3046
    },
    {
      "epoch": 0.419383171011927,
      "grad_norm": 0.1404879242181778,
      "learning_rate": 0.002,
      "loss": 2.954484462738037,
      "step": 3048
    },
    {
      "epoch": 0.4196583568196776,
      "grad_norm": 0.10280466079711914,
      "learning_rate": 0.002,
      "loss": 2.9377341270446777,
      "step": 3050
    },
    {
      "epoch": 0.4199335426274282,
      "grad_norm": 0.20778177678585052,
      "learning_rate": 0.002,
      "loss": 2.914139986038208,
      "step": 3052
    },
    {
      "epoch": 0.4202087284351788,
      "grad_norm": 0.09244099259376526,
      "learning_rate": 0.002,
      "loss": 2.962663173675537,
      "step": 3054
    },
    {
      "epoch": 0.42048391424292947,
      "grad_norm": 0.14360103011131287,
      "learning_rate": 0.002,
      "loss": 2.895705461502075,
      "step": 3056
    },
    {
      "epoch": 0.42075910005068007,
      "grad_norm": 0.03476482257246971,
      "learning_rate": 0.002,
      "loss": 2.9406769275665283,
      "step": 3058
    },
    {
      "epoch": 0.42103428585843067,
      "grad_norm": 0.11318614333868027,
      "learning_rate": 0.002,
      "loss": 2.9353091716766357,
      "step": 3060
    },
    {
      "epoch": 0.42130947166618127,
      "grad_norm": 0.2247876226902008,
      "learning_rate": 0.002,
      "loss": 2.911818504333496,
      "step": 3062
    },
    {
      "epoch": 0.42158465747393187,
      "grad_norm": 0.08577501028776169,
      "learning_rate": 0.002,
      "loss": 2.9390978813171387,
      "step": 3064
    },
    {
      "epoch": 0.42185984328168247,
      "grad_norm": 0.0887366458773613,
      "learning_rate": 0.002,
      "loss": 2.9142065048217773,
      "step": 3066
    },
    {
      "epoch": 0.42213502908943307,
      "grad_norm": 0.09156767278909683,
      "learning_rate": 0.002,
      "loss": 2.9434471130371094,
      "step": 3068
    },
    {
      "epoch": 0.4224102148971837,
      "grad_norm": 0.12679672241210938,
      "learning_rate": 0.002,
      "loss": 2.921217441558838,
      "step": 3070
    },
    {
      "epoch": 0.4226854007049343,
      "grad_norm": 0.04805592820048332,
      "learning_rate": 0.002,
      "loss": 2.927647113800049,
      "step": 3072
    },
    {
      "epoch": 0.4229605865126849,
      "grad_norm": 0.03828797489404678,
      "learning_rate": 0.002,
      "loss": 2.955422878265381,
      "step": 3074
    },
    {
      "epoch": 0.4232357723204355,
      "grad_norm": 0.04117602854967117,
      "learning_rate": 0.002,
      "loss": 2.9551053047180176,
      "step": 3076
    },
    {
      "epoch": 0.4235109581281861,
      "grad_norm": 0.044412512332201004,
      "learning_rate": 0.002,
      "loss": 2.9348084926605225,
      "step": 3078
    },
    {
      "epoch": 0.4237861439359367,
      "grad_norm": 0.09679652750492096,
      "learning_rate": 0.002,
      "loss": 2.953962802886963,
      "step": 3080
    },
    {
      "epoch": 0.4240613297436874,
      "grad_norm": 0.2329535335302353,
      "learning_rate": 0.002,
      "loss": 2.962078809738159,
      "step": 3082
    },
    {
      "epoch": 0.424336515551438,
      "grad_norm": 0.4342271387577057,
      "learning_rate": 0.002,
      "loss": 2.9435651302337646,
      "step": 3084
    },
    {
      "epoch": 0.4246117013591886,
      "grad_norm": 0.04529537633061409,
      "learning_rate": 0.002,
      "loss": 2.9207096099853516,
      "step": 3086
    },
    {
      "epoch": 0.4248868871669392,
      "grad_norm": 0.29550546407699585,
      "learning_rate": 0.002,
      "loss": 2.946715831756592,
      "step": 3088
    },
    {
      "epoch": 0.4251620729746898,
      "grad_norm": 0.1729639768600464,
      "learning_rate": 0.002,
      "loss": 2.9515509605407715,
      "step": 3090
    },
    {
      "epoch": 0.4254372587824404,
      "grad_norm": 0.13553619384765625,
      "learning_rate": 0.002,
      "loss": 2.9657187461853027,
      "step": 3092
    },
    {
      "epoch": 0.425712444590191,
      "grad_norm": 0.10693918168544769,
      "learning_rate": 0.002,
      "loss": 2.868391513824463,
      "step": 3094
    },
    {
      "epoch": 0.42598763039794163,
      "grad_norm": 0.10155228525400162,
      "learning_rate": 0.002,
      "loss": 2.9247148036956787,
      "step": 3096
    },
    {
      "epoch": 0.42626281620569223,
      "grad_norm": 0.09339885413646698,
      "learning_rate": 0.002,
      "loss": 2.9949889183044434,
      "step": 3098
    },
    {
      "epoch": 0.42653800201344283,
      "grad_norm": 0.03929370641708374,
      "learning_rate": 0.002,
      "loss": 2.9114975929260254,
      "step": 3100
    },
    {
      "epoch": 0.42681318782119343,
      "grad_norm": 0.08110008388757706,
      "learning_rate": 0.002,
      "loss": 2.9558193683624268,
      "step": 3102
    },
    {
      "epoch": 0.42708837362894403,
      "grad_norm": 0.142096146941185,
      "learning_rate": 0.002,
      "loss": 2.9379587173461914,
      "step": 3104
    },
    {
      "epoch": 0.42736355943669463,
      "grad_norm": 0.04979509487748146,
      "learning_rate": 0.002,
      "loss": 2.9376931190490723,
      "step": 3106
    },
    {
      "epoch": 0.4276387452444453,
      "grad_norm": 0.08993412554264069,
      "learning_rate": 0.002,
      "loss": 2.9509730339050293,
      "step": 3108
    },
    {
      "epoch": 0.4279139310521959,
      "grad_norm": 0.11123108863830566,
      "learning_rate": 0.002,
      "loss": 2.937145233154297,
      "step": 3110
    },
    {
      "epoch": 0.4281891168599465,
      "grad_norm": 0.07519081979990005,
      "learning_rate": 0.002,
      "loss": 2.9195332527160645,
      "step": 3112
    },
    {
      "epoch": 0.4284643026676971,
      "grad_norm": 0.17412634193897247,
      "learning_rate": 0.002,
      "loss": 2.9310333728790283,
      "step": 3114
    },
    {
      "epoch": 0.4287394884754477,
      "grad_norm": 0.13373993337154388,
      "learning_rate": 0.002,
      "loss": 2.886303424835205,
      "step": 3116
    },
    {
      "epoch": 0.4290146742831983,
      "grad_norm": 0.06487099826335907,
      "learning_rate": 0.002,
      "loss": 2.96706485748291,
      "step": 3118
    },
    {
      "epoch": 0.4292898600909489,
      "grad_norm": 0.1777057945728302,
      "learning_rate": 0.002,
      "loss": 2.911628246307373,
      "step": 3120
    },
    {
      "epoch": 0.42956504589869954,
      "grad_norm": 0.19948731362819672,
      "learning_rate": 0.002,
      "loss": 2.918044090270996,
      "step": 3122
    },
    {
      "epoch": 0.42984023170645014,
      "grad_norm": 0.07067212462425232,
      "learning_rate": 0.002,
      "loss": 2.9429874420166016,
      "step": 3124
    },
    {
      "epoch": 0.43011541751420074,
      "grad_norm": 0.13505196571350098,
      "learning_rate": 0.002,
      "loss": 2.9647860527038574,
      "step": 3126
    },
    {
      "epoch": 0.43039060332195134,
      "grad_norm": 0.33417466282844543,
      "learning_rate": 0.002,
      "loss": 2.898134469985962,
      "step": 3128
    },
    {
      "epoch": 0.43066578912970194,
      "grad_norm": 0.09878938645124435,
      "learning_rate": 0.002,
      "loss": 2.9404358863830566,
      "step": 3130
    },
    {
      "epoch": 0.43094097493745254,
      "grad_norm": 0.2745094895362854,
      "learning_rate": 0.002,
      "loss": 2.893026113510132,
      "step": 3132
    },
    {
      "epoch": 0.4312161607452032,
      "grad_norm": 0.0821603462100029,
      "learning_rate": 0.002,
      "loss": 2.9123077392578125,
      "step": 3134
    },
    {
      "epoch": 0.4314913465529538,
      "grad_norm": 0.22109772264957428,
      "learning_rate": 0.002,
      "loss": 2.9721760749816895,
      "step": 3136
    },
    {
      "epoch": 0.4317665323607044,
      "grad_norm": 0.05892012640833855,
      "learning_rate": 0.002,
      "loss": 2.9370293617248535,
      "step": 3138
    },
    {
      "epoch": 0.432041718168455,
      "grad_norm": 0.11051787436008453,
      "learning_rate": 0.002,
      "loss": 2.920234441757202,
      "step": 3140
    },
    {
      "epoch": 0.4323169039762056,
      "grad_norm": 0.03313830867409706,
      "learning_rate": 0.002,
      "loss": 2.911874294281006,
      "step": 3142
    },
    {
      "epoch": 0.4325920897839562,
      "grad_norm": 0.08854277431964874,
      "learning_rate": 0.002,
      "loss": 2.9211764335632324,
      "step": 3144
    },
    {
      "epoch": 0.4328672755917068,
      "grad_norm": 0.04178934916853905,
      "learning_rate": 0.002,
      "loss": 2.9221503734588623,
      "step": 3146
    },
    {
      "epoch": 0.43314246139945745,
      "grad_norm": 0.036498747766017914,
      "learning_rate": 0.002,
      "loss": 2.9157509803771973,
      "step": 3148
    },
    {
      "epoch": 0.43341764720720805,
      "grad_norm": 0.03198159858584404,
      "learning_rate": 0.002,
      "loss": 2.9171791076660156,
      "step": 3150
    },
    {
      "epoch": 0.43369283301495865,
      "grad_norm": 0.04276842251420021,
      "learning_rate": 0.002,
      "loss": 2.9400081634521484,
      "step": 3152
    },
    {
      "epoch": 0.43396801882270924,
      "grad_norm": 0.07739010453224182,
      "learning_rate": 0.002,
      "loss": 2.932680368423462,
      "step": 3154
    },
    {
      "epoch": 0.43424320463045984,
      "grad_norm": 0.25547119975090027,
      "learning_rate": 0.002,
      "loss": 2.954073667526245,
      "step": 3156
    },
    {
      "epoch": 0.43451839043821044,
      "grad_norm": 0.3077176511287689,
      "learning_rate": 0.002,
      "loss": 2.910017967224121,
      "step": 3158
    },
    {
      "epoch": 0.4347935762459611,
      "grad_norm": 0.03638036176562309,
      "learning_rate": 0.002,
      "loss": 2.930527687072754,
      "step": 3160
    },
    {
      "epoch": 0.4350687620537117,
      "grad_norm": 0.14363734424114227,
      "learning_rate": 0.002,
      "loss": 2.942110538482666,
      "step": 3162
    },
    {
      "epoch": 0.4353439478614623,
      "grad_norm": 0.07568328082561493,
      "learning_rate": 0.002,
      "loss": 2.929562568664551,
      "step": 3164
    },
    {
      "epoch": 0.4356191336692129,
      "grad_norm": 0.03213123232126236,
      "learning_rate": 0.002,
      "loss": 2.8819732666015625,
      "step": 3166
    },
    {
      "epoch": 0.4358943194769635,
      "grad_norm": 0.18457214534282684,
      "learning_rate": 0.002,
      "loss": 2.8697242736816406,
      "step": 3168
    },
    {
      "epoch": 0.4361695052847141,
      "grad_norm": 0.35481005907058716,
      "learning_rate": 0.002,
      "loss": 2.9350202083587646,
      "step": 3170
    },
    {
      "epoch": 0.4364446910924647,
      "grad_norm": 0.09771513193845749,
      "learning_rate": 0.002,
      "loss": 2.899484157562256,
      "step": 3172
    },
    {
      "epoch": 0.43671987690021535,
      "grad_norm": 0.2229761928319931,
      "learning_rate": 0.002,
      "loss": 2.8932347297668457,
      "step": 3174
    },
    {
      "epoch": 0.43699506270796595,
      "grad_norm": 0.3593446612358093,
      "learning_rate": 0.002,
      "loss": 2.9604177474975586,
      "step": 3176
    },
    {
      "epoch": 0.43727024851571655,
      "grad_norm": 0.037735000252723694,
      "learning_rate": 0.002,
      "loss": 2.918600559234619,
      "step": 3178
    },
    {
      "epoch": 0.43754543432346715,
      "grad_norm": 0.21080134809017181,
      "learning_rate": 0.002,
      "loss": 2.9078402519226074,
      "step": 3180
    },
    {
      "epoch": 0.43782062013121775,
      "grad_norm": 0.15902861952781677,
      "learning_rate": 0.002,
      "loss": 2.897308349609375,
      "step": 3182
    },
    {
      "epoch": 0.43809580593896835,
      "grad_norm": 0.046320512890815735,
      "learning_rate": 0.002,
      "loss": 2.9128103256225586,
      "step": 3184
    },
    {
      "epoch": 0.438370991746719,
      "grad_norm": 0.048616182059049606,
      "learning_rate": 0.002,
      "loss": 2.8892860412597656,
      "step": 3186
    },
    {
      "epoch": 0.4386461775544696,
      "grad_norm": 0.05865589156746864,
      "learning_rate": 0.002,
      "loss": 2.9176437854766846,
      "step": 3188
    },
    {
      "epoch": 0.4389213633622202,
      "grad_norm": 0.1027141809463501,
      "learning_rate": 0.002,
      "loss": 2.9005990028381348,
      "step": 3190
    },
    {
      "epoch": 0.4391965491699708,
      "grad_norm": 0.06341557204723358,
      "learning_rate": 0.002,
      "loss": 2.9086689949035645,
      "step": 3192
    },
    {
      "epoch": 0.4394717349777214,
      "grad_norm": 0.041222989559173584,
      "learning_rate": 0.002,
      "loss": 2.9408469200134277,
      "step": 3194
    },
    {
      "epoch": 0.439746920785472,
      "grad_norm": 0.13716019690036774,
      "learning_rate": 0.002,
      "loss": 2.9514923095703125,
      "step": 3196
    },
    {
      "epoch": 0.4400221065932226,
      "grad_norm": 0.15990272164344788,
      "learning_rate": 0.002,
      "loss": 2.8879036903381348,
      "step": 3198
    },
    {
      "epoch": 0.44029729240097326,
      "grad_norm": 0.053280964493751526,
      "learning_rate": 0.002,
      "loss": 2.977242946624756,
      "step": 3200
    },
    {
      "epoch": 0.44057247820872386,
      "grad_norm": 0.03302744776010513,
      "learning_rate": 0.002,
      "loss": 2.914196014404297,
      "step": 3202
    },
    {
      "epoch": 0.44084766401647446,
      "grad_norm": 0.04687171429395676,
      "learning_rate": 0.002,
      "loss": 2.937279462814331,
      "step": 3204
    },
    {
      "epoch": 0.44112284982422506,
      "grad_norm": 0.05665431544184685,
      "learning_rate": 0.002,
      "loss": 2.930196762084961,
      "step": 3206
    },
    {
      "epoch": 0.44139803563197566,
      "grad_norm": 0.06455138325691223,
      "learning_rate": 0.002,
      "loss": 2.896388053894043,
      "step": 3208
    },
    {
      "epoch": 0.44167322143972626,
      "grad_norm": 0.15441447496414185,
      "learning_rate": 0.002,
      "loss": 2.897904872894287,
      "step": 3210
    },
    {
      "epoch": 0.4419484072474769,
      "grad_norm": 0.1625903695821762,
      "learning_rate": 0.002,
      "loss": 2.9102354049682617,
      "step": 3212
    },
    {
      "epoch": 0.4422235930552275,
      "grad_norm": 0.19082333147525787,
      "learning_rate": 0.002,
      "loss": 2.9055590629577637,
      "step": 3214
    },
    {
      "epoch": 0.4424987788629781,
      "grad_norm": 0.2138534039258957,
      "learning_rate": 0.002,
      "loss": 2.9185163974761963,
      "step": 3216
    },
    {
      "epoch": 0.4427739646707287,
      "grad_norm": 0.13898923993110657,
      "learning_rate": 0.002,
      "loss": 2.908402919769287,
      "step": 3218
    },
    {
      "epoch": 0.4430491504784793,
      "grad_norm": 0.17549976706504822,
      "learning_rate": 0.002,
      "loss": 2.908686637878418,
      "step": 3220
    },
    {
      "epoch": 0.4433243362862299,
      "grad_norm": 0.3012905418872833,
      "learning_rate": 0.002,
      "loss": 2.9199585914611816,
      "step": 3222
    },
    {
      "epoch": 0.4435995220939805,
      "grad_norm": 0.09363629668951035,
      "learning_rate": 0.002,
      "loss": 2.9184441566467285,
      "step": 3224
    },
    {
      "epoch": 0.44387470790173117,
      "grad_norm": 0.12915214896202087,
      "learning_rate": 0.002,
      "loss": 2.8866353034973145,
      "step": 3226
    },
    {
      "epoch": 0.44414989370948177,
      "grad_norm": 0.0625842809677124,
      "learning_rate": 0.002,
      "loss": 2.9211525917053223,
      "step": 3228
    },
    {
      "epoch": 0.44442507951723237,
      "grad_norm": 0.044685058295726776,
      "learning_rate": 0.002,
      "loss": 2.946216344833374,
      "step": 3230
    },
    {
      "epoch": 0.44470026532498297,
      "grad_norm": 0.12884896993637085,
      "learning_rate": 0.002,
      "loss": 2.95566987991333,
      "step": 3232
    },
    {
      "epoch": 0.44497545113273357,
      "grad_norm": 0.11259929090738297,
      "learning_rate": 0.002,
      "loss": 2.912548065185547,
      "step": 3234
    },
    {
      "epoch": 0.44525063694048417,
      "grad_norm": 0.04561520740389824,
      "learning_rate": 0.002,
      "loss": 2.8907036781311035,
      "step": 3236
    },
    {
      "epoch": 0.4455258227482348,
      "grad_norm": 0.06416847556829453,
      "learning_rate": 0.002,
      "loss": 2.9202380180358887,
      "step": 3238
    },
    {
      "epoch": 0.4458010085559854,
      "grad_norm": 0.03845023363828659,
      "learning_rate": 0.002,
      "loss": 2.9227147102355957,
      "step": 3240
    },
    {
      "epoch": 0.446076194363736,
      "grad_norm": 0.04061271250247955,
      "learning_rate": 0.002,
      "loss": 2.8862407207489014,
      "step": 3242
    },
    {
      "epoch": 0.4463513801714866,
      "grad_norm": 0.08536335080862045,
      "learning_rate": 0.002,
      "loss": 2.9202284812927246,
      "step": 3244
    },
    {
      "epoch": 0.4466265659792372,
      "grad_norm": 0.18985380232334137,
      "learning_rate": 0.002,
      "loss": 2.93119478225708,
      "step": 3246
    },
    {
      "epoch": 0.4469017517869878,
      "grad_norm": 0.33057162165641785,
      "learning_rate": 0.002,
      "loss": 2.9142677783966064,
      "step": 3248
    },
    {
      "epoch": 0.4471769375947384,
      "grad_norm": 0.09608467668294907,
      "learning_rate": 0.002,
      "loss": 2.9602043628692627,
      "step": 3250
    },
    {
      "epoch": 0.4474521234024891,
      "grad_norm": 0.08477487415075302,
      "learning_rate": 0.002,
      "loss": 2.858527898788452,
      "step": 3252
    },
    {
      "epoch": 0.4477273092102397,
      "grad_norm": 0.03362851217389107,
      "learning_rate": 0.002,
      "loss": 2.9146814346313477,
      "step": 3254
    },
    {
      "epoch": 0.4480024950179903,
      "grad_norm": 0.14240743219852448,
      "learning_rate": 0.002,
      "loss": 2.8809523582458496,
      "step": 3256
    },
    {
      "epoch": 0.4482776808257409,
      "grad_norm": 0.17174075543880463,
      "learning_rate": 0.002,
      "loss": 2.8972184658050537,
      "step": 3258
    },
    {
      "epoch": 0.4485528666334915,
      "grad_norm": 0.07427310943603516,
      "learning_rate": 0.002,
      "loss": 2.9155611991882324,
      "step": 3260
    },
    {
      "epoch": 0.4488280524412421,
      "grad_norm": 0.2373998463153839,
      "learning_rate": 0.002,
      "loss": 2.898158550262451,
      "step": 3262
    },
    {
      "epoch": 0.44910323824899273,
      "grad_norm": 0.06400507688522339,
      "learning_rate": 0.002,
      "loss": 2.893097400665283,
      "step": 3264
    },
    {
      "epoch": 0.44937842405674333,
      "grad_norm": 0.12875191867351532,
      "learning_rate": 0.002,
      "loss": 2.8993659019470215,
      "step": 3266
    },
    {
      "epoch": 0.44965360986449393,
      "grad_norm": 0.13272185623645782,
      "learning_rate": 0.002,
      "loss": 2.8874146938323975,
      "step": 3268
    },
    {
      "epoch": 0.44992879567224453,
      "grad_norm": 0.07263006269931793,
      "learning_rate": 0.002,
      "loss": 2.8859758377075195,
      "step": 3270
    },
    {
      "epoch": 0.45020398147999513,
      "grad_norm": 0.20985989272594452,
      "learning_rate": 0.002,
      "loss": 2.9094367027282715,
      "step": 3272
    },
    {
      "epoch": 0.45047916728774573,
      "grad_norm": 0.13188789784908295,
      "learning_rate": 0.002,
      "loss": 2.915461540222168,
      "step": 3274
    },
    {
      "epoch": 0.45075435309549633,
      "grad_norm": 0.07988876104354858,
      "learning_rate": 0.002,
      "loss": 2.947159767150879,
      "step": 3276
    },
    {
      "epoch": 0.451029538903247,
      "grad_norm": 0.2296564280986786,
      "learning_rate": 0.002,
      "loss": 2.8553614616394043,
      "step": 3278
    },
    {
      "epoch": 0.4513047247109976,
      "grad_norm": 0.3296601176261902,
      "learning_rate": 0.002,
      "loss": 2.9354286193847656,
      "step": 3280
    },
    {
      "epoch": 0.4515799105187482,
      "grad_norm": 0.20095476508140564,
      "learning_rate": 0.002,
      "loss": 2.924221992492676,
      "step": 3282
    },
    {
      "epoch": 0.4518550963264988,
      "grad_norm": 0.2574116885662079,
      "learning_rate": 0.002,
      "loss": 2.912403106689453,
      "step": 3284
    },
    {
      "epoch": 0.4521302821342494,
      "grad_norm": 0.24549685418605804,
      "learning_rate": 0.002,
      "loss": 2.9134764671325684,
      "step": 3286
    },
    {
      "epoch": 0.452405467942,
      "grad_norm": 0.04557913541793823,
      "learning_rate": 0.002,
      "loss": 2.9064319133758545,
      "step": 3288
    },
    {
      "epoch": 0.45268065374975064,
      "grad_norm": 0.07557070255279541,
      "learning_rate": 0.002,
      "loss": 2.922633409500122,
      "step": 3290
    },
    {
      "epoch": 0.45295583955750124,
      "grad_norm": 0.16987895965576172,
      "learning_rate": 0.002,
      "loss": 2.889753580093384,
      "step": 3292
    },
    {
      "epoch": 0.45323102536525184,
      "grad_norm": 0.034233491867780685,
      "learning_rate": 0.002,
      "loss": 2.9426722526550293,
      "step": 3294
    },
    {
      "epoch": 0.45350621117300244,
      "grad_norm": 0.04721177741885185,
      "learning_rate": 0.002,
      "loss": 2.8742103576660156,
      "step": 3296
    },
    {
      "epoch": 0.45378139698075304,
      "grad_norm": 0.03818132355809212,
      "learning_rate": 0.002,
      "loss": 2.896082878112793,
      "step": 3298
    },
    {
      "epoch": 0.45405658278850364,
      "grad_norm": 0.046772461384534836,
      "learning_rate": 0.002,
      "loss": 2.8960697650909424,
      "step": 3300
    },
    {
      "epoch": 0.45433176859625424,
      "grad_norm": 0.03166767954826355,
      "learning_rate": 0.002,
      "loss": 2.861288070678711,
      "step": 3302
    },
    {
      "epoch": 0.4546069544040049,
      "grad_norm": 0.058109741657972336,
      "learning_rate": 0.002,
      "loss": 2.8984270095825195,
      "step": 3304
    },
    {
      "epoch": 0.4548821402117555,
      "grad_norm": 0.18682266771793365,
      "learning_rate": 0.002,
      "loss": 2.897921085357666,
      "step": 3306
    },
    {
      "epoch": 0.4551573260195061,
      "grad_norm": 0.12290655821561813,
      "learning_rate": 0.002,
      "loss": 2.9362633228302,
      "step": 3308
    },
    {
      "epoch": 0.4554325118272567,
      "grad_norm": 0.06684078276157379,
      "learning_rate": 0.002,
      "loss": 2.908350944519043,
      "step": 3310
    },
    {
      "epoch": 0.4557076976350073,
      "grad_norm": 0.08659686148166656,
      "learning_rate": 0.002,
      "loss": 2.8967819213867188,
      "step": 3312
    },
    {
      "epoch": 0.4559828834427579,
      "grad_norm": 0.08263377845287323,
      "learning_rate": 0.002,
      "loss": 2.935258626937866,
      "step": 3314
    },
    {
      "epoch": 0.45625806925050855,
      "grad_norm": 0.07612796872854233,
      "learning_rate": 0.002,
      "loss": 2.8967549800872803,
      "step": 3316
    },
    {
      "epoch": 0.45653325505825915,
      "grad_norm": 0.08585991710424423,
      "learning_rate": 0.002,
      "loss": 2.9096615314483643,
      "step": 3318
    },
    {
      "epoch": 0.45680844086600975,
      "grad_norm": 0.03305236995220184,
      "learning_rate": 0.002,
      "loss": 2.9150781631469727,
      "step": 3320
    },
    {
      "epoch": 0.45708362667376035,
      "grad_norm": 0.16413234174251556,
      "learning_rate": 0.002,
      "loss": 2.950890064239502,
      "step": 3322
    },
    {
      "epoch": 0.45735881248151095,
      "grad_norm": 0.44554755091667175,
      "learning_rate": 0.002,
      "loss": 2.906071662902832,
      "step": 3324
    },
    {
      "epoch": 0.45763399828926155,
      "grad_norm": 0.051399506628513336,
      "learning_rate": 0.002,
      "loss": 2.897965908050537,
      "step": 3326
    },
    {
      "epoch": 0.45790918409701215,
      "grad_norm": 0.21788065135478973,
      "learning_rate": 0.002,
      "loss": 2.9161324501037598,
      "step": 3328
    },
    {
      "epoch": 0.4581843699047628,
      "grad_norm": 0.053506769239902496,
      "learning_rate": 0.002,
      "loss": 2.892305850982666,
      "step": 3330
    },
    {
      "epoch": 0.4584595557125134,
      "grad_norm": 0.14554503560066223,
      "learning_rate": 0.002,
      "loss": 2.9016549587249756,
      "step": 3332
    },
    {
      "epoch": 0.458734741520264,
      "grad_norm": 0.17073556780815125,
      "learning_rate": 0.002,
      "loss": 2.951655387878418,
      "step": 3334
    },
    {
      "epoch": 0.4590099273280146,
      "grad_norm": 0.04099294915795326,
      "learning_rate": 0.002,
      "loss": 2.925168752670288,
      "step": 3336
    },
    {
      "epoch": 0.4592851131357652,
      "grad_norm": 0.06888925284147263,
      "learning_rate": 0.002,
      "loss": 2.8976473808288574,
      "step": 3338
    },
    {
      "epoch": 0.4595602989435158,
      "grad_norm": 0.0726202130317688,
      "learning_rate": 0.002,
      "loss": 2.9145617485046387,
      "step": 3340
    },
    {
      "epoch": 0.45983548475126645,
      "grad_norm": 0.06770753115415573,
      "learning_rate": 0.002,
      "loss": 2.8760557174682617,
      "step": 3342
    },
    {
      "epoch": 0.46011067055901705,
      "grad_norm": 0.07052946090698242,
      "learning_rate": 0.002,
      "loss": 2.8870177268981934,
      "step": 3344
    },
    {
      "epoch": 0.46038585636676765,
      "grad_norm": 0.07814301550388336,
      "learning_rate": 0.002,
      "loss": 2.8795838356018066,
      "step": 3346
    },
    {
      "epoch": 0.46066104217451825,
      "grad_norm": 0.12283441424369812,
      "learning_rate": 0.002,
      "loss": 2.9207382202148438,
      "step": 3348
    },
    {
      "epoch": 0.46093622798226885,
      "grad_norm": 0.07868249714374542,
      "learning_rate": 0.002,
      "loss": 2.8816821575164795,
      "step": 3350
    },
    {
      "epoch": 0.46121141379001945,
      "grad_norm": 0.1920585334300995,
      "learning_rate": 0.002,
      "loss": 2.9070401191711426,
      "step": 3352
    },
    {
      "epoch": 0.46148659959777005,
      "grad_norm": 0.27700749039649963,
      "learning_rate": 0.002,
      "loss": 2.8992040157318115,
      "step": 3354
    },
    {
      "epoch": 0.4617617854055207,
      "grad_norm": 0.12067601084709167,
      "learning_rate": 0.002,
      "loss": 2.909830331802368,
      "step": 3356
    },
    {
      "epoch": 0.4620369712132713,
      "grad_norm": 0.21719743311405182,
      "learning_rate": 0.002,
      "loss": 2.93459415435791,
      "step": 3358
    },
    {
      "epoch": 0.4623121570210219,
      "grad_norm": 0.06505338102579117,
      "learning_rate": 0.002,
      "loss": 2.9020917415618896,
      "step": 3360
    },
    {
      "epoch": 0.4625873428287725,
      "grad_norm": 0.10035115480422974,
      "learning_rate": 0.002,
      "loss": 2.8798842430114746,
      "step": 3362
    },
    {
      "epoch": 0.4628625286365231,
      "grad_norm": 0.03240400180220604,
      "learning_rate": 0.002,
      "loss": 2.8890247344970703,
      "step": 3364
    },
    {
      "epoch": 0.4631377144442737,
      "grad_norm": 0.1484358012676239,
      "learning_rate": 0.002,
      "loss": 2.8816380500793457,
      "step": 3366
    },
    {
      "epoch": 0.46341290025202436,
      "grad_norm": 0.14357781410217285,
      "learning_rate": 0.002,
      "loss": 2.9050216674804688,
      "step": 3368
    },
    {
      "epoch": 0.46368808605977496,
      "grad_norm": 0.17170509696006775,
      "learning_rate": 0.002,
      "loss": 2.881974935531616,
      "step": 3370
    },
    {
      "epoch": 0.46396327186752556,
      "grad_norm": 0.059816718101501465,
      "learning_rate": 0.002,
      "loss": 2.8977460861206055,
      "step": 3372
    },
    {
      "epoch": 0.46423845767527616,
      "grad_norm": 0.0521489754319191,
      "learning_rate": 0.002,
      "loss": 2.8932080268859863,
      "step": 3374
    },
    {
      "epoch": 0.46451364348302676,
      "grad_norm": 0.035166412591934204,
      "learning_rate": 0.002,
      "loss": 2.9134178161621094,
      "step": 3376
    },
    {
      "epoch": 0.46478882929077736,
      "grad_norm": 0.055657509714365005,
      "learning_rate": 0.002,
      "loss": 2.8948721885681152,
      "step": 3378
    },
    {
      "epoch": 0.46506401509852796,
      "grad_norm": 0.15432040393352509,
      "learning_rate": 0.002,
      "loss": 2.890535831451416,
      "step": 3380
    },
    {
      "epoch": 0.4653392009062786,
      "grad_norm": 0.36270543932914734,
      "learning_rate": 0.002,
      "loss": 2.85732364654541,
      "step": 3382
    },
    {
      "epoch": 0.4656143867140292,
      "grad_norm": 0.08466119319200516,
      "learning_rate": 0.002,
      "loss": 2.918816566467285,
      "step": 3384
    },
    {
      "epoch": 0.4658895725217798,
      "grad_norm": 0.29820016026496887,
      "learning_rate": 0.002,
      "loss": 2.9414145946502686,
      "step": 3386
    },
    {
      "epoch": 0.4661647583295304,
      "grad_norm": 0.15121397376060486,
      "learning_rate": 0.002,
      "loss": 2.930600643157959,
      "step": 3388
    },
    {
      "epoch": 0.466439944137281,
      "grad_norm": 0.22017620503902435,
      "learning_rate": 0.002,
      "loss": 2.936206102371216,
      "step": 3390
    },
    {
      "epoch": 0.4667151299450316,
      "grad_norm": 0.05258959159255028,
      "learning_rate": 0.002,
      "loss": 2.8973851203918457,
      "step": 3392
    },
    {
      "epoch": 0.46699031575278227,
      "grad_norm": 0.14651863276958466,
      "learning_rate": 0.002,
      "loss": 2.900902509689331,
      "step": 3394
    },
    {
      "epoch": 0.46726550156053287,
      "grad_norm": 0.036795347929000854,
      "learning_rate": 0.002,
      "loss": 2.9246864318847656,
      "step": 3396
    },
    {
      "epoch": 0.46754068736828347,
      "grad_norm": 0.07122556865215302,
      "learning_rate": 0.002,
      "loss": 2.91802716255188,
      "step": 3398
    },
    {
      "epoch": 0.46781587317603407,
      "grad_norm": 0.15420761704444885,
      "learning_rate": 0.002,
      "loss": 2.8706729412078857,
      "step": 3400
    },
    {
      "epoch": 0.46809105898378467,
      "grad_norm": 0.1442088633775711,
      "learning_rate": 0.002,
      "loss": 2.887788772583008,
      "step": 3402
    },
    {
      "epoch": 0.46836624479153527,
      "grad_norm": 0.042210571467876434,
      "learning_rate": 0.002,
      "loss": 2.8875892162323,
      "step": 3404
    },
    {
      "epoch": 0.46864143059928587,
      "grad_norm": 0.08820483088493347,
      "learning_rate": 0.002,
      "loss": 2.9314932823181152,
      "step": 3406
    },
    {
      "epoch": 0.4689166164070365,
      "grad_norm": 0.036883316934108734,
      "learning_rate": 0.002,
      "loss": 2.911865234375,
      "step": 3408
    },
    {
      "epoch": 0.4691918022147871,
      "grad_norm": 0.031800974160432816,
      "learning_rate": 0.002,
      "loss": 2.8635382652282715,
      "step": 3410
    },
    {
      "epoch": 0.4694669880225377,
      "grad_norm": 0.030138477683067322,
      "learning_rate": 0.002,
      "loss": 2.887218952178955,
      "step": 3412
    },
    {
      "epoch": 0.4697421738302883,
      "grad_norm": 0.038503628224134445,
      "learning_rate": 0.002,
      "loss": 2.9133434295654297,
      "step": 3414
    },
    {
      "epoch": 0.4700173596380389,
      "grad_norm": 0.09955205023288727,
      "learning_rate": 0.002,
      "loss": 2.9200665950775146,
      "step": 3416
    },
    {
      "epoch": 0.4702925454457895,
      "grad_norm": 0.28184831142425537,
      "learning_rate": 0.002,
      "loss": 2.9059553146362305,
      "step": 3418
    },
    {
      "epoch": 0.4705677312535402,
      "grad_norm": 0.4072735607624054,
      "learning_rate": 0.002,
      "loss": 2.8776395320892334,
      "step": 3420
    },
    {
      "epoch": 0.4708429170612908,
      "grad_norm": 0.04514824599027634,
      "learning_rate": 0.002,
      "loss": 2.922563076019287,
      "step": 3422
    },
    {
      "epoch": 0.4711181028690414,
      "grad_norm": 0.3143835961818695,
      "learning_rate": 0.002,
      "loss": 2.8542184829711914,
      "step": 3424
    },
    {
      "epoch": 0.471393288676792,
      "grad_norm": 0.1986970454454422,
      "learning_rate": 0.002,
      "loss": 2.8962783813476562,
      "step": 3426
    },
    {
      "epoch": 0.4716684744845426,
      "grad_norm": 0.08306798338890076,
      "learning_rate": 0.002,
      "loss": 2.90842866897583,
      "step": 3428
    },
    {
      "epoch": 0.4719436602922932,
      "grad_norm": 0.2453182190656662,
      "learning_rate": 0.002,
      "loss": 2.886836290359497,
      "step": 3430
    },
    {
      "epoch": 0.4722188461000438,
      "grad_norm": 0.1380433291196823,
      "learning_rate": 0.002,
      "loss": 2.8885817527770996,
      "step": 3432
    },
    {
      "epoch": 0.47249403190779443,
      "grad_norm": 0.03392796218395233,
      "learning_rate": 0.002,
      "loss": 2.882033348083496,
      "step": 3434
    },
    {
      "epoch": 0.47276921771554503,
      "grad_norm": 0.037055324763059616,
      "learning_rate": 0.002,
      "loss": 2.904109001159668,
      "step": 3436
    },
    {
      "epoch": 0.47304440352329563,
      "grad_norm": 0.11227245628833771,
      "learning_rate": 0.002,
      "loss": 2.917422294616699,
      "step": 3438
    },
    {
      "epoch": 0.47331958933104623,
      "grad_norm": 0.03890475258231163,
      "learning_rate": 0.002,
      "loss": 2.910250663757324,
      "step": 3440
    },
    {
      "epoch": 0.47359477513879683,
      "grad_norm": 0.07831960171461105,
      "learning_rate": 0.002,
      "loss": 2.888249397277832,
      "step": 3442
    },
    {
      "epoch": 0.47386996094654743,
      "grad_norm": 0.058242280036211014,
      "learning_rate": 0.002,
      "loss": 2.8738584518432617,
      "step": 3444
    },
    {
      "epoch": 0.4741451467542981,
      "grad_norm": 0.1544533371925354,
      "learning_rate": 0.002,
      "loss": 2.8419229984283447,
      "step": 3446
    },
    {
      "epoch": 0.4744203325620487,
      "grad_norm": 0.09979221224784851,
      "learning_rate": 0.002,
      "loss": 2.8971242904663086,
      "step": 3448
    },
    {
      "epoch": 0.4746955183697993,
      "grad_norm": 0.032020192593336105,
      "learning_rate": 0.002,
      "loss": 2.897522449493408,
      "step": 3450
    },
    {
      "epoch": 0.4749707041775499,
      "grad_norm": 0.033940333873033524,
      "learning_rate": 0.002,
      "loss": 2.8851494789123535,
      "step": 3452
    },
    {
      "epoch": 0.4752458899853005,
      "grad_norm": 0.055991869419813156,
      "learning_rate": 0.002,
      "loss": 2.853715419769287,
      "step": 3454
    },
    {
      "epoch": 0.4755210757930511,
      "grad_norm": 0.03863435611128807,
      "learning_rate": 0.002,
      "loss": 2.865671157836914,
      "step": 3456
    },
    {
      "epoch": 0.4757962616008017,
      "grad_norm": 0.09965306520462036,
      "learning_rate": 0.002,
      "loss": 2.889899253845215,
      "step": 3458
    },
    {
      "epoch": 0.47607144740855234,
      "grad_norm": 0.07319165766239166,
      "learning_rate": 0.002,
      "loss": 2.9266185760498047,
      "step": 3460
    },
    {
      "epoch": 0.47634663321630294,
      "grad_norm": 0.1297682672739029,
      "learning_rate": 0.002,
      "loss": 2.886537551879883,
      "step": 3462
    },
    {
      "epoch": 0.47662181902405354,
      "grad_norm": 0.3837379515171051,
      "learning_rate": 0.002,
      "loss": 2.9067933559417725,
      "step": 3464
    },
    {
      "epoch": 0.47689700483180414,
      "grad_norm": 0.44138333201408386,
      "learning_rate": 0.002,
      "loss": 2.8686776161193848,
      "step": 3466
    },
    {
      "epoch": 0.47717219063955474,
      "grad_norm": 0.34511303901672363,
      "learning_rate": 0.002,
      "loss": 2.8924145698547363,
      "step": 3468
    },
    {
      "epoch": 0.47744737644730534,
      "grad_norm": 0.03367142006754875,
      "learning_rate": 0.002,
      "loss": 2.872293472290039,
      "step": 3470
    },
    {
      "epoch": 0.477722562255056,
      "grad_norm": 0.2888614535331726,
      "learning_rate": 0.002,
      "loss": 2.8444020748138428,
      "step": 3472
    },
    {
      "epoch": 0.4779977480628066,
      "grad_norm": 0.2841568887233734,
      "learning_rate": 0.002,
      "loss": 2.865377902984619,
      "step": 3474
    },
    {
      "epoch": 0.4782729338705572,
      "grad_norm": 0.09220251441001892,
      "learning_rate": 0.002,
      "loss": 2.8835885524749756,
      "step": 3476
    },
    {
      "epoch": 0.4785481196783078,
      "grad_norm": 0.09025457501411438,
      "learning_rate": 0.002,
      "loss": 2.8705763816833496,
      "step": 3478
    },
    {
      "epoch": 0.4788233054860584,
      "grad_norm": 0.1920749843120575,
      "learning_rate": 0.002,
      "loss": 2.889981746673584,
      "step": 3480
    },
    {
      "epoch": 0.479098491293809,
      "grad_norm": 0.05852344632148743,
      "learning_rate": 0.002,
      "loss": 2.911642074584961,
      "step": 3482
    },
    {
      "epoch": 0.4793736771015596,
      "grad_norm": 0.08367963880300522,
      "learning_rate": 0.002,
      "loss": 2.8793325424194336,
      "step": 3484
    },
    {
      "epoch": 0.47964886290931025,
      "grad_norm": 0.0846090167760849,
      "learning_rate": 0.002,
      "loss": 2.897369861602783,
      "step": 3486
    },
    {
      "epoch": 0.47992404871706085,
      "grad_norm": 0.03285707160830498,
      "learning_rate": 0.002,
      "loss": 2.921435832977295,
      "step": 3488
    },
    {
      "epoch": 0.48019923452481145,
      "grad_norm": 0.07839765399694443,
      "learning_rate": 0.002,
      "loss": 2.88856840133667,
      "step": 3490
    },
    {
      "epoch": 0.48047442033256205,
      "grad_norm": 0.05001337453722954,
      "learning_rate": 0.002,
      "loss": 2.870199680328369,
      "step": 3492
    },
    {
      "epoch": 0.48074960614031265,
      "grad_norm": 0.059446271508932114,
      "learning_rate": 0.002,
      "loss": 2.9052624702453613,
      "step": 3494
    },
    {
      "epoch": 0.48102479194806325,
      "grad_norm": 0.2360471934080124,
      "learning_rate": 0.002,
      "loss": 2.8958966732025146,
      "step": 3496
    },
    {
      "epoch": 0.48129997775581385,
      "grad_norm": 0.042629558593034744,
      "learning_rate": 0.002,
      "loss": 2.8595335483551025,
      "step": 3498
    },
    {
      "epoch": 0.4815751635635645,
      "grad_norm": 0.13908636569976807,
      "learning_rate": 0.002,
      "loss": 2.896684169769287,
      "step": 3500
    },
    {
      "epoch": 0.4818503493713151,
      "grad_norm": 0.0378078855574131,
      "learning_rate": 0.002,
      "loss": 2.9017279148101807,
      "step": 3502
    },
    {
      "epoch": 0.4821255351790657,
      "grad_norm": 0.12253782153129578,
      "learning_rate": 0.002,
      "loss": 2.9328675270080566,
      "step": 3504
    },
    {
      "epoch": 0.4824007209868163,
      "grad_norm": 0.049407172948122025,
      "learning_rate": 0.002,
      "loss": 2.884518623352051,
      "step": 3506
    },
    {
      "epoch": 0.4826759067945669,
      "grad_norm": 0.18762779235839844,
      "learning_rate": 0.002,
      "loss": 2.902233600616455,
      "step": 3508
    },
    {
      "epoch": 0.4829510926023175,
      "grad_norm": 0.34802594780921936,
      "learning_rate": 0.002,
      "loss": 2.8646206855773926,
      "step": 3510
    },
    {
      "epoch": 0.48322627841006816,
      "grad_norm": 0.04095824062824249,
      "learning_rate": 0.002,
      "loss": 2.87204647064209,
      "step": 3512
    },
    {
      "epoch": 0.48350146421781875,
      "grad_norm": 0.23051708936691284,
      "learning_rate": 0.002,
      "loss": 2.8623552322387695,
      "step": 3514
    },
    {
      "epoch": 0.48377665002556935,
      "grad_norm": 0.03461106866598129,
      "learning_rate": 0.002,
      "loss": 2.8455734252929688,
      "step": 3516
    },
    {
      "epoch": 0.48405183583331995,
      "grad_norm": 0.1692645400762558,
      "learning_rate": 0.002,
      "loss": 2.9189047813415527,
      "step": 3518
    },
    {
      "epoch": 0.48432702164107055,
      "grad_norm": 0.09495585411787033,
      "learning_rate": 0.002,
      "loss": 2.879237651824951,
      "step": 3520
    },
    {
      "epoch": 0.48460220744882115,
      "grad_norm": 0.12721601128578186,
      "learning_rate": 0.002,
      "loss": 2.8798906803131104,
      "step": 3522
    },
    {
      "epoch": 0.48487739325657175,
      "grad_norm": 0.03592287749052048,
      "learning_rate": 0.002,
      "loss": 2.9269957542419434,
      "step": 3524
    },
    {
      "epoch": 0.4851525790643224,
      "grad_norm": 0.10701726377010345,
      "learning_rate": 0.002,
      "loss": 2.8821754455566406,
      "step": 3526
    },
    {
      "epoch": 0.485427764872073,
      "grad_norm": 0.05313631892204285,
      "learning_rate": 0.002,
      "loss": 2.8813133239746094,
      "step": 3528
    },
    {
      "epoch": 0.4857029506798236,
      "grad_norm": 0.1527870148420334,
      "learning_rate": 0.002,
      "loss": 2.844172477722168,
      "step": 3530
    },
    {
      "epoch": 0.4859781364875742,
      "grad_norm": 0.26069894433021545,
      "learning_rate": 0.002,
      "loss": 2.880542516708374,
      "step": 3532
    },
    {
      "epoch": 0.4862533222953248,
      "grad_norm": 0.07404647767543793,
      "learning_rate": 0.002,
      "loss": 2.8726556301116943,
      "step": 3534
    },
    {
      "epoch": 0.4865285081030754,
      "grad_norm": 0.23969024419784546,
      "learning_rate": 0.002,
      "loss": 2.85805082321167,
      "step": 3536
    },
    {
      "epoch": 0.48680369391082606,
      "grad_norm": 0.03642550855875015,
      "learning_rate": 0.002,
      "loss": 2.871547222137451,
      "step": 3538
    },
    {
      "epoch": 0.48707887971857666,
      "grad_norm": 0.2857511043548584,
      "learning_rate": 0.002,
      "loss": 2.8881940841674805,
      "step": 3540
    },
    {
      "epoch": 0.48735406552632726,
      "grad_norm": 0.0626654103398323,
      "learning_rate": 0.002,
      "loss": 2.954375982284546,
      "step": 3542
    },
    {
      "epoch": 0.48762925133407786,
      "grad_norm": 0.10125993192195892,
      "learning_rate": 0.002,
      "loss": 2.8828415870666504,
      "step": 3544
    },
    {
      "epoch": 0.48790443714182846,
      "grad_norm": 0.07122769206762314,
      "learning_rate": 0.002,
      "loss": 2.875443458557129,
      "step": 3546
    },
    {
      "epoch": 0.48817962294957906,
      "grad_norm": 0.04973595589399338,
      "learning_rate": 0.002,
      "loss": 2.8641433715820312,
      "step": 3548
    },
    {
      "epoch": 0.48845480875732966,
      "grad_norm": 0.06519800424575806,
      "learning_rate": 0.002,
      "loss": 2.862375259399414,
      "step": 3550
    },
    {
      "epoch": 0.4887299945650803,
      "grad_norm": 0.21083898842334747,
      "learning_rate": 0.002,
      "loss": 2.8443291187286377,
      "step": 3552
    },
    {
      "epoch": 0.4890051803728309,
      "grad_norm": 0.23663032054901123,
      "learning_rate": 0.002,
      "loss": 2.8782427310943604,
      "step": 3554
    },
    {
      "epoch": 0.4892803661805815,
      "grad_norm": 0.10358761250972748,
      "learning_rate": 0.002,
      "loss": 2.8557863235473633,
      "step": 3556
    },
    {
      "epoch": 0.4895555519883321,
      "grad_norm": 0.2601027190685272,
      "learning_rate": 0.002,
      "loss": 2.8923416137695312,
      "step": 3558
    },
    {
      "epoch": 0.4898307377960827,
      "grad_norm": 0.032850541174411774,
      "learning_rate": 0.002,
      "loss": 2.8803188800811768,
      "step": 3560
    },
    {
      "epoch": 0.4901059236038333,
      "grad_norm": 0.18456798791885376,
      "learning_rate": 0.002,
      "loss": 2.918302536010742,
      "step": 3562
    },
    {
      "epoch": 0.49038110941158397,
      "grad_norm": 0.050962626934051514,
      "learning_rate": 0.002,
      "loss": 2.867138385772705,
      "step": 3564
    },
    {
      "epoch": 0.49065629521933457,
      "grad_norm": 0.16916967928409576,
      "learning_rate": 0.002,
      "loss": 2.8634135723114014,
      "step": 3566
    },
    {
      "epoch": 0.49093148102708517,
      "grad_norm": 0.06129293143749237,
      "learning_rate": 0.002,
      "loss": 2.8949689865112305,
      "step": 3568
    },
    {
      "epoch": 0.49120666683483577,
      "grad_norm": 0.03668734431266785,
      "learning_rate": 0.002,
      "loss": 2.863274097442627,
      "step": 3570
    },
    {
      "epoch": 0.49148185264258637,
      "grad_norm": 0.1023898720741272,
      "learning_rate": 0.002,
      "loss": 2.876565933227539,
      "step": 3572
    },
    {
      "epoch": 0.49175703845033697,
      "grad_norm": 0.19755594432353973,
      "learning_rate": 0.002,
      "loss": 2.846771240234375,
      "step": 3574
    },
    {
      "epoch": 0.49203222425808757,
      "grad_norm": 0.05000760033726692,
      "learning_rate": 0.002,
      "loss": 2.8652875423431396,
      "step": 3576
    },
    {
      "epoch": 0.4923074100658382,
      "grad_norm": 0.08502162247896194,
      "learning_rate": 0.002,
      "loss": 2.86635160446167,
      "step": 3578
    },
    {
      "epoch": 0.4925825958735888,
      "grad_norm": 0.1239347979426384,
      "learning_rate": 0.002,
      "loss": 2.9354312419891357,
      "step": 3580
    },
    {
      "epoch": 0.4928577816813394,
      "grad_norm": 0.03188255801796913,
      "learning_rate": 0.002,
      "loss": 2.834789276123047,
      "step": 3582
    },
    {
      "epoch": 0.49313296748909,
      "grad_norm": 0.03175861015915871,
      "learning_rate": 0.002,
      "loss": 2.9243366718292236,
      "step": 3584
    },
    {
      "epoch": 0.4934081532968406,
      "grad_norm": 0.11531311273574829,
      "learning_rate": 0.002,
      "loss": 2.851583957672119,
      "step": 3586
    },
    {
      "epoch": 0.4936833391045912,
      "grad_norm": 0.24321115016937256,
      "learning_rate": 0.002,
      "loss": 2.9081106185913086,
      "step": 3588
    },
    {
      "epoch": 0.4939585249123419,
      "grad_norm": 0.32461896538734436,
      "learning_rate": 0.002,
      "loss": 2.8602356910705566,
      "step": 3590
    },
    {
      "epoch": 0.4942337107200925,
      "grad_norm": 0.042774781584739685,
      "learning_rate": 0.002,
      "loss": 2.863525629043579,
      "step": 3592
    },
    {
      "epoch": 0.4945088965278431,
      "grad_norm": 0.15812090039253235,
      "learning_rate": 0.002,
      "loss": 2.8080430030822754,
      "step": 3594
    },
    {
      "epoch": 0.4947840823355937,
      "grad_norm": 0.13816294074058533,
      "learning_rate": 0.002,
      "loss": 2.8779196739196777,
      "step": 3596
    },
    {
      "epoch": 0.4950592681433443,
      "grad_norm": 0.1469511091709137,
      "learning_rate": 0.002,
      "loss": 2.8496532440185547,
      "step": 3598
    },
    {
      "epoch": 0.4953344539510949,
      "grad_norm": 0.13329192996025085,
      "learning_rate": 0.002,
      "loss": 2.855673313140869,
      "step": 3600
    },
    {
      "epoch": 0.4956096397588455,
      "grad_norm": 0.037979766726493835,
      "learning_rate": 0.002,
      "loss": 2.8765885829925537,
      "step": 3602
    },
    {
      "epoch": 0.49588482556659613,
      "grad_norm": 0.08447406440973282,
      "learning_rate": 0.002,
      "loss": 2.873089551925659,
      "step": 3604
    },
    {
      "epoch": 0.49616001137434673,
      "grad_norm": 0.17347478866577148,
      "learning_rate": 0.002,
      "loss": 2.8551573753356934,
      "step": 3606
    },
    {
      "epoch": 0.49643519718209733,
      "grad_norm": 0.24442653357982635,
      "learning_rate": 0.002,
      "loss": 2.87520170211792,
      "step": 3608
    },
    {
      "epoch": 0.49671038298984793,
      "grad_norm": 0.1338181495666504,
      "learning_rate": 0.002,
      "loss": 2.915757656097412,
      "step": 3610
    },
    {
      "epoch": 0.49698556879759853,
      "grad_norm": 0.1809980273246765,
      "learning_rate": 0.002,
      "loss": 2.8763139247894287,
      "step": 3612
    },
    {
      "epoch": 0.49726075460534913,
      "grad_norm": 0.030742904171347618,
      "learning_rate": 0.002,
      "loss": 2.875527858734131,
      "step": 3614
    },
    {
      "epoch": 0.4975359404130998,
      "grad_norm": 0.1755427122116089,
      "learning_rate": 0.002,
      "loss": 2.902930974960327,
      "step": 3616
    },
    {
      "epoch": 0.4978111262208504,
      "grad_norm": 0.06695840507745743,
      "learning_rate": 0.002,
      "loss": 2.857541084289551,
      "step": 3618
    },
    {
      "epoch": 0.498086312028601,
      "grad_norm": 0.23752418160438538,
      "learning_rate": 0.002,
      "loss": 2.8713605403900146,
      "step": 3620
    },
    {
      "epoch": 0.4983614978363516,
      "grad_norm": 0.10189250111579895,
      "learning_rate": 0.002,
      "loss": 2.8637852668762207,
      "step": 3622
    },
    {
      "epoch": 0.4986366836441022,
      "grad_norm": 0.09209489822387695,
      "learning_rate": 0.002,
      "loss": 2.860783100128174,
      "step": 3624
    },
    {
      "epoch": 0.4989118694518528,
      "grad_norm": 0.1651604175567627,
      "learning_rate": 0.002,
      "loss": 2.883491277694702,
      "step": 3626
    },
    {
      "epoch": 0.4991870552596034,
      "grad_norm": 0.0725279375910759,
      "learning_rate": 0.002,
      "loss": 2.849249839782715,
      "step": 3628
    },
    {
      "epoch": 0.49946224106735404,
      "grad_norm": 0.04515557736158371,
      "learning_rate": 0.002,
      "loss": 2.889005661010742,
      "step": 3630
    },
    {
      "epoch": 0.49973742687510464,
      "grad_norm": 0.09369302541017532,
      "learning_rate": 0.002,
      "loss": 2.843536853790283,
      "step": 3632
    },
    {
      "epoch": 0.5000126126828552,
      "grad_norm": 0.2510358393192291,
      "learning_rate": 0.002,
      "loss": 2.8753433227539062,
      "step": 3634
    },
    {
      "epoch": 0.5002877984906059,
      "grad_norm": 0.26231372356414795,
      "learning_rate": 0.002,
      "loss": 2.8402810096740723,
      "step": 3636
    },
    {
      "epoch": 0.5005629842983564,
      "grad_norm": 0.05821973830461502,
      "learning_rate": 0.002,
      "loss": 2.8329544067382812,
      "step": 3638
    },
    {
      "epoch": 0.5008381701061071,
      "grad_norm": 0.27570515871047974,
      "learning_rate": 0.002,
      "loss": 2.8552088737487793,
      "step": 3640
    },
    {
      "epoch": 0.5011133559138576,
      "grad_norm": 0.22180862724781036,
      "learning_rate": 0.002,
      "loss": 2.8874895572662354,
      "step": 3642
    },
    {
      "epoch": 0.5013885417216083,
      "grad_norm": 0.27367863059043884,
      "learning_rate": 0.002,
      "loss": 2.851119041442871,
      "step": 3644
    },
    {
      "epoch": 0.5016637275293588,
      "grad_norm": 0.03782866522669792,
      "learning_rate": 0.002,
      "loss": 2.8852357864379883,
      "step": 3646
    },
    {
      "epoch": 0.5019389133371095,
      "grad_norm": 0.24154818058013916,
      "learning_rate": 0.002,
      "loss": 2.869330406188965,
      "step": 3648
    },
    {
      "epoch": 0.5022140991448601,
      "grad_norm": 0.07760791480541229,
      "learning_rate": 0.002,
      "loss": 2.84877347946167,
      "step": 3650
    },
    {
      "epoch": 0.5024892849526107,
      "grad_norm": 0.14507927000522614,
      "learning_rate": 0.002,
      "loss": 2.873481273651123,
      "step": 3652
    },
    {
      "epoch": 0.5027644707603613,
      "grad_norm": 0.03409874811768532,
      "learning_rate": 0.002,
      "loss": 2.8801708221435547,
      "step": 3654
    },
    {
      "epoch": 0.5030396565681119,
      "grad_norm": 0.08223024010658264,
      "learning_rate": 0.002,
      "loss": 2.8447749614715576,
      "step": 3656
    },
    {
      "epoch": 0.5033148423758625,
      "grad_norm": 0.030314592644572258,
      "learning_rate": 0.002,
      "loss": 2.8418405055999756,
      "step": 3658
    },
    {
      "epoch": 0.5035900281836131,
      "grad_norm": 0.0827958732843399,
      "learning_rate": 0.002,
      "loss": 2.8758034706115723,
      "step": 3660
    },
    {
      "epoch": 0.5038652139913637,
      "grad_norm": 0.037257131189107895,
      "learning_rate": 0.002,
      "loss": 2.846348762512207,
      "step": 3662
    },
    {
      "epoch": 0.5041403997991144,
      "grad_norm": 0.03218909353017807,
      "learning_rate": 0.002,
      "loss": 2.858567237854004,
      "step": 3664
    },
    {
      "epoch": 0.504415585606865,
      "grad_norm": 0.03173711895942688,
      "learning_rate": 0.002,
      "loss": 2.8552370071411133,
      "step": 3666
    },
    {
      "epoch": 0.5046907714146156,
      "grad_norm": 0.06593728810548782,
      "learning_rate": 0.002,
      "loss": 2.871736526489258,
      "step": 3668
    },
    {
      "epoch": 0.5049659572223661,
      "grad_norm": 0.03329279646277428,
      "learning_rate": 0.002,
      "loss": 2.890106201171875,
      "step": 3670
    },
    {
      "epoch": 0.5052411430301168,
      "grad_norm": 0.044816821813583374,
      "learning_rate": 0.002,
      "loss": 2.868805408477783,
      "step": 3672
    },
    {
      "epoch": 0.5055163288378675,
      "grad_norm": 0.13528233766555786,
      "learning_rate": 0.002,
      "loss": 2.8916311264038086,
      "step": 3674
    },
    {
      "epoch": 0.505791514645618,
      "grad_norm": 0.19882731139659882,
      "learning_rate": 0.002,
      "loss": 2.8222837448120117,
      "step": 3676
    },
    {
      "epoch": 0.5060667004533687,
      "grad_norm": 0.3622395098209381,
      "learning_rate": 0.002,
      "loss": 2.8596694469451904,
      "step": 3678
    },
    {
      "epoch": 0.5063418862611192,
      "grad_norm": 0.2517480254173279,
      "learning_rate": 0.002,
      "loss": 2.8246188163757324,
      "step": 3680
    },
    {
      "epoch": 0.5066170720688699,
      "grad_norm": 0.28232812881469727,
      "learning_rate": 0.002,
      "loss": 2.848701238632202,
      "step": 3682
    },
    {
      "epoch": 0.5068922578766204,
      "grad_norm": 0.03439105674624443,
      "learning_rate": 0.002,
      "loss": 2.8504080772399902,
      "step": 3684
    },
    {
      "epoch": 0.507167443684371,
      "grad_norm": 0.24129649996757507,
      "learning_rate": 0.002,
      "loss": 2.8843283653259277,
      "step": 3686
    },
    {
      "epoch": 0.5074426294921217,
      "grad_norm": 0.11585385352373123,
      "learning_rate": 0.002,
      "loss": 2.880582332611084,
      "step": 3688
    },
    {
      "epoch": 0.5077178152998723,
      "grad_norm": 0.19121946394443512,
      "learning_rate": 0.002,
      "loss": 2.883746385574341,
      "step": 3690
    },
    {
      "epoch": 0.5079930011076229,
      "grad_norm": 0.057400867342948914,
      "learning_rate": 0.002,
      "loss": 2.888329029083252,
      "step": 3692
    },
    {
      "epoch": 0.5082681869153735,
      "grad_norm": 0.04848761856555939,
      "learning_rate": 0.002,
      "loss": 2.861985206604004,
      "step": 3694
    },
    {
      "epoch": 0.5085433727231241,
      "grad_norm": 0.060251522809267044,
      "learning_rate": 0.002,
      "loss": 2.8060083389282227,
      "step": 3696
    },
    {
      "epoch": 0.5088185585308747,
      "grad_norm": 0.0870944932103157,
      "learning_rate": 0.002,
      "loss": 2.8277969360351562,
      "step": 3698
    },
    {
      "epoch": 0.5090937443386253,
      "grad_norm": 0.05423762649297714,
      "learning_rate": 0.002,
      "loss": 2.827981948852539,
      "step": 3700
    },
    {
      "epoch": 0.509368930146376,
      "grad_norm": 0.08844295144081116,
      "learning_rate": 0.002,
      "loss": 2.8935983180999756,
      "step": 3702
    },
    {
      "epoch": 0.5096441159541265,
      "grad_norm": 0.15851670503616333,
      "learning_rate": 0.002,
      "loss": 2.863015651702881,
      "step": 3704
    },
    {
      "epoch": 0.5099193017618772,
      "grad_norm": 0.066059410572052,
      "learning_rate": 0.002,
      "loss": 2.846766233444214,
      "step": 3706
    },
    {
      "epoch": 0.5101944875696277,
      "grad_norm": 0.23906652629375458,
      "learning_rate": 0.002,
      "loss": 2.8498430252075195,
      "step": 3708
    },
    {
      "epoch": 0.5104696733773784,
      "grad_norm": 0.23994114995002747,
      "learning_rate": 0.002,
      "loss": 2.8752174377441406,
      "step": 3710
    },
    {
      "epoch": 0.5107448591851289,
      "grad_norm": 0.03593659773468971,
      "learning_rate": 0.002,
      "loss": 2.884249448776245,
      "step": 3712
    },
    {
      "epoch": 0.5110200449928796,
      "grad_norm": 0.06301847845315933,
      "learning_rate": 0.002,
      "loss": 2.907148838043213,
      "step": 3714
    },
    {
      "epoch": 0.5112952308006302,
      "grad_norm": 0.09128792583942413,
      "learning_rate": 0.002,
      "loss": 2.8549537658691406,
      "step": 3716
    },
    {
      "epoch": 0.5115704166083808,
      "grad_norm": 0.03113800659775734,
      "learning_rate": 0.002,
      "loss": 2.8623011112213135,
      "step": 3718
    },
    {
      "epoch": 0.5118456024161314,
      "grad_norm": 0.15169399976730347,
      "learning_rate": 0.002,
      "loss": 2.8688840866088867,
      "step": 3720
    },
    {
      "epoch": 0.512120788223882,
      "grad_norm": 0.180873841047287,
      "learning_rate": 0.002,
      "loss": 2.844372034072876,
      "step": 3722
    },
    {
      "epoch": 0.5123959740316326,
      "grad_norm": 0.06396602094173431,
      "learning_rate": 0.002,
      "loss": 2.810659408569336,
      "step": 3724
    },
    {
      "epoch": 0.5126711598393832,
      "grad_norm": 0.12350482493638992,
      "learning_rate": 0.002,
      "loss": 2.8646678924560547,
      "step": 3726
    },
    {
      "epoch": 0.5129463456471338,
      "grad_norm": 0.06736484915018082,
      "learning_rate": 0.002,
      "loss": 2.8740222454071045,
      "step": 3728
    },
    {
      "epoch": 0.5132215314548845,
      "grad_norm": 0.14296570420265198,
      "learning_rate": 0.002,
      "loss": 2.8436384201049805,
      "step": 3730
    },
    {
      "epoch": 0.513496717262635,
      "grad_norm": 0.2502298653125763,
      "learning_rate": 0.002,
      "loss": 2.8826000690460205,
      "step": 3732
    },
    {
      "epoch": 0.5137719030703857,
      "grad_norm": 0.32945385575294495,
      "learning_rate": 0.002,
      "loss": 2.887864112854004,
      "step": 3734
    },
    {
      "epoch": 0.5140470888781362,
      "grad_norm": 0.041152115911245346,
      "learning_rate": 0.002,
      "loss": 2.865701675415039,
      "step": 3736
    },
    {
      "epoch": 0.5143222746858869,
      "grad_norm": 0.19585026800632477,
      "learning_rate": 0.002,
      "loss": 2.858551502227783,
      "step": 3738
    },
    {
      "epoch": 0.5145974604936375,
      "grad_norm": 0.07210588455200195,
      "learning_rate": 0.002,
      "loss": 2.8120768070220947,
      "step": 3740
    },
    {
      "epoch": 0.5148726463013881,
      "grad_norm": 0.04323945567011833,
      "learning_rate": 0.002,
      "loss": 2.8658535480499268,
      "step": 3742
    },
    {
      "epoch": 0.5151478321091387,
      "grad_norm": 0.0780702605843544,
      "learning_rate": 0.002,
      "loss": 2.8748035430908203,
      "step": 3744
    },
    {
      "epoch": 0.5154230179168893,
      "grad_norm": 0.11324166506528854,
      "learning_rate": 0.002,
      "loss": 2.8564133644104004,
      "step": 3746
    },
    {
      "epoch": 0.5156982037246399,
      "grad_norm": 0.12883034348487854,
      "learning_rate": 0.002,
      "loss": 2.8868746757507324,
      "step": 3748
    },
    {
      "epoch": 0.5159733895323905,
      "grad_norm": 0.03954165428876877,
      "learning_rate": 0.002,
      "loss": 2.8251864910125732,
      "step": 3750
    },
    {
      "epoch": 0.5162485753401411,
      "grad_norm": 0.17965736985206604,
      "learning_rate": 0.002,
      "loss": 2.8780198097229004,
      "step": 3752
    },
    {
      "epoch": 0.5165237611478918,
      "grad_norm": 0.09556274861097336,
      "learning_rate": 0.002,
      "loss": 2.867135524749756,
      "step": 3754
    },
    {
      "epoch": 0.5167989469556423,
      "grad_norm": 0.03063773922622204,
      "learning_rate": 0.002,
      "loss": 2.920686721801758,
      "step": 3756
    },
    {
      "epoch": 0.517074132763393,
      "grad_norm": 0.20492003858089447,
      "learning_rate": 0.002,
      "loss": 2.866591453552246,
      "step": 3758
    },
    {
      "epoch": 0.5173493185711435,
      "grad_norm": 0.14068609476089478,
      "learning_rate": 0.002,
      "loss": 2.8988327980041504,
      "step": 3760
    },
    {
      "epoch": 0.5176245043788942,
      "grad_norm": 0.041509080678224564,
      "learning_rate": 0.002,
      "loss": 2.853835105895996,
      "step": 3762
    },
    {
      "epoch": 0.5178996901866447,
      "grad_norm": 0.1437326967716217,
      "learning_rate": 0.002,
      "loss": 2.8114123344421387,
      "step": 3764
    },
    {
      "epoch": 0.5181748759943954,
      "grad_norm": 0.19545981287956238,
      "learning_rate": 0.002,
      "loss": 2.8202059268951416,
      "step": 3766
    },
    {
      "epoch": 0.518450061802146,
      "grad_norm": 0.0639013722538948,
      "learning_rate": 0.002,
      "loss": 2.866403579711914,
      "step": 3768
    },
    {
      "epoch": 0.5187252476098966,
      "grad_norm": 0.18778812885284424,
      "learning_rate": 0.002,
      "loss": 2.8820436000823975,
      "step": 3770
    },
    {
      "epoch": 0.5190004334176472,
      "grad_norm": 0.19220806658267975,
      "learning_rate": 0.002,
      "loss": 2.860421657562256,
      "step": 3772
    },
    {
      "epoch": 0.5192756192253978,
      "grad_norm": 0.040481045842170715,
      "learning_rate": 0.002,
      "loss": 2.84332537651062,
      "step": 3774
    },
    {
      "epoch": 0.5195508050331484,
      "grad_norm": 0.12040466070175171,
      "learning_rate": 0.002,
      "loss": 2.8123741149902344,
      "step": 3776
    },
    {
      "epoch": 0.519825990840899,
      "grad_norm": 0.20838631689548492,
      "learning_rate": 0.002,
      "loss": 2.8861992359161377,
      "step": 3778
    },
    {
      "epoch": 0.5201011766486496,
      "grad_norm": 0.05101212486624718,
      "learning_rate": 0.002,
      "loss": 2.821812152862549,
      "step": 3780
    },
    {
      "epoch": 0.5201011766486496,
      "eval_loss": 2.835872173309326,
      "eval_runtime": 378.7095,
      "eval_samples_per_second": 92.86,
      "eval_steps_per_second": 11.608,
      "step": 3780
    },
    {
      "epoch": 0.5203763624564003,
      "grad_norm": 0.07806325703859329,
      "learning_rate": 0.002,
      "loss": 2.835742950439453,
      "step": 3782
    },
    {
      "epoch": 0.5206515482641508,
      "grad_norm": 0.05323084071278572,
      "learning_rate": 0.002,
      "loss": 2.845754623413086,
      "step": 3784
    },
    {
      "epoch": 0.5209267340719015,
      "grad_norm": 0.08375286310911179,
      "learning_rate": 0.002,
      "loss": 2.849193572998047,
      "step": 3786
    },
    {
      "epoch": 0.521201919879652,
      "grad_norm": 0.15944157540798187,
      "learning_rate": 0.002,
      "loss": 2.8631913661956787,
      "step": 3788
    },
    {
      "epoch": 0.5214771056874027,
      "grad_norm": 0.23642097413539886,
      "learning_rate": 0.002,
      "loss": 2.8156590461730957,
      "step": 3790
    },
    {
      "epoch": 0.5217522914951533,
      "grad_norm": 0.1761690080165863,
      "learning_rate": 0.002,
      "loss": 2.87019944190979,
      "step": 3792
    },
    {
      "epoch": 0.5220274773029039,
      "grad_norm": 0.039802178740501404,
      "learning_rate": 0.002,
      "loss": 2.8437676429748535,
      "step": 3794
    },
    {
      "epoch": 0.5223026631106545,
      "grad_norm": 0.28256696462631226,
      "learning_rate": 0.002,
      "loss": 2.8286890983581543,
      "step": 3796
    },
    {
      "epoch": 0.5225778489184051,
      "grad_norm": 0.22404815256595612,
      "learning_rate": 0.002,
      "loss": 2.8634603023529053,
      "step": 3798
    },
    {
      "epoch": 0.5228530347261557,
      "grad_norm": 0.06155410781502724,
      "learning_rate": 0.002,
      "loss": 2.8552703857421875,
      "step": 3800
    },
    {
      "epoch": 0.5231282205339063,
      "grad_norm": 0.17690517008304596,
      "learning_rate": 0.002,
      "loss": 2.8508100509643555,
      "step": 3802
    },
    {
      "epoch": 0.5234034063416569,
      "grad_norm": 0.14013125002384186,
      "learning_rate": 0.002,
      "loss": 2.880223274230957,
      "step": 3804
    },
    {
      "epoch": 0.5236785921494076,
      "grad_norm": 0.12182744592428207,
      "learning_rate": 0.002,
      "loss": 2.817315101623535,
      "step": 3806
    },
    {
      "epoch": 0.5239537779571581,
      "grad_norm": 0.10308282822370529,
      "learning_rate": 0.002,
      "loss": 2.7925310134887695,
      "step": 3808
    },
    {
      "epoch": 0.5242289637649088,
      "grad_norm": 0.07856100797653198,
      "learning_rate": 0.002,
      "loss": 2.858539342880249,
      "step": 3810
    },
    {
      "epoch": 0.5245041495726593,
      "grad_norm": 0.11467935889959335,
      "learning_rate": 0.002,
      "loss": 2.8492040634155273,
      "step": 3812
    },
    {
      "epoch": 0.52477933538041,
      "grad_norm": 0.15282367169857025,
      "learning_rate": 0.002,
      "loss": 2.815336227416992,
      "step": 3814
    },
    {
      "epoch": 0.5250545211881605,
      "grad_norm": 0.057671431452035904,
      "learning_rate": 0.002,
      "loss": 2.842461347579956,
      "step": 3816
    },
    {
      "epoch": 0.5253297069959112,
      "grad_norm": 0.054565440863370895,
      "learning_rate": 0.002,
      "loss": 2.8327226638793945,
      "step": 3818
    },
    {
      "epoch": 0.5256048928036618,
      "grad_norm": 0.032105352729558945,
      "learning_rate": 0.002,
      "loss": 2.899449586868286,
      "step": 3820
    },
    {
      "epoch": 0.5258800786114124,
      "grad_norm": 0.10346736013889313,
      "learning_rate": 0.002,
      "loss": 2.820878505706787,
      "step": 3822
    },
    {
      "epoch": 0.526155264419163,
      "grad_norm": 0.2777109742164612,
      "learning_rate": 0.002,
      "loss": 2.829479217529297,
      "step": 3824
    },
    {
      "epoch": 0.5264304502269136,
      "grad_norm": 0.21913112699985504,
      "learning_rate": 0.002,
      "loss": 2.827651023864746,
      "step": 3826
    },
    {
      "epoch": 0.5267056360346642,
      "grad_norm": 0.07729192823171616,
      "learning_rate": 0.002,
      "loss": 2.824805498123169,
      "step": 3828
    },
    {
      "epoch": 0.5269808218424148,
      "grad_norm": 0.2098807394504547,
      "learning_rate": 0.002,
      "loss": 2.811887264251709,
      "step": 3830
    },
    {
      "epoch": 0.5272560076501654,
      "grad_norm": 0.1443440467119217,
      "learning_rate": 0.002,
      "loss": 2.844205856323242,
      "step": 3832
    },
    {
      "epoch": 0.5275311934579161,
      "grad_norm": 0.03385039046406746,
      "learning_rate": 0.002,
      "loss": 2.8443140983581543,
      "step": 3834
    },
    {
      "epoch": 0.5278063792656666,
      "grad_norm": 0.036605507135391235,
      "learning_rate": 0.002,
      "loss": 2.865926742553711,
      "step": 3836
    },
    {
      "epoch": 0.5280815650734173,
      "grad_norm": 0.06113201379776001,
      "learning_rate": 0.002,
      "loss": 2.821244239807129,
      "step": 3838
    },
    {
      "epoch": 0.5283567508811678,
      "grad_norm": 0.17430341243743896,
      "learning_rate": 0.002,
      "loss": 2.8580613136291504,
      "step": 3840
    },
    {
      "epoch": 0.5286319366889185,
      "grad_norm": 0.191009521484375,
      "learning_rate": 0.002,
      "loss": 2.876485586166382,
      "step": 3842
    },
    {
      "epoch": 0.5289071224966692,
      "grad_norm": 0.12286780029535294,
      "learning_rate": 0.002,
      "loss": 2.842841863632202,
      "step": 3844
    },
    {
      "epoch": 0.5291823083044197,
      "grad_norm": 0.057806260883808136,
      "learning_rate": 0.002,
      "loss": 2.865339756011963,
      "step": 3846
    },
    {
      "epoch": 0.5294574941121704,
      "grad_norm": 0.03073750250041485,
      "learning_rate": 0.002,
      "loss": 2.85955810546875,
      "step": 3848
    },
    {
      "epoch": 0.5297326799199209,
      "grad_norm": 0.0807613730430603,
      "learning_rate": 0.002,
      "loss": 2.844715118408203,
      "step": 3850
    },
    {
      "epoch": 0.5300078657276716,
      "grad_norm": 0.2143712192773819,
      "learning_rate": 0.002,
      "loss": 2.8504409790039062,
      "step": 3852
    },
    {
      "epoch": 0.5302830515354221,
      "grad_norm": 0.1315469890832901,
      "learning_rate": 0.002,
      "loss": 2.820941209793091,
      "step": 3854
    },
    {
      "epoch": 0.5305582373431728,
      "grad_norm": 0.08871589601039886,
      "learning_rate": 0.002,
      "loss": 2.7800188064575195,
      "step": 3856
    },
    {
      "epoch": 0.5308334231509234,
      "grad_norm": 0.035028185695409775,
      "learning_rate": 0.002,
      "loss": 2.861110210418701,
      "step": 3858
    },
    {
      "epoch": 0.531108608958674,
      "grad_norm": 0.0546867661178112,
      "learning_rate": 0.002,
      "loss": 2.832773208618164,
      "step": 3860
    },
    {
      "epoch": 0.5313837947664246,
      "grad_norm": 0.16550588607788086,
      "learning_rate": 0.002,
      "loss": 2.8272035121917725,
      "step": 3862
    },
    {
      "epoch": 0.5316589805741752,
      "grad_norm": 0.38660070300102234,
      "learning_rate": 0.002,
      "loss": 2.7990877628326416,
      "step": 3864
    },
    {
      "epoch": 0.5319341663819258,
      "grad_norm": 0.1510116308927536,
      "learning_rate": 0.002,
      "loss": 2.827970504760742,
      "step": 3866
    },
    {
      "epoch": 0.5322093521896764,
      "grad_norm": 0.33541423082351685,
      "learning_rate": 0.002,
      "loss": 2.8502285480499268,
      "step": 3868
    },
    {
      "epoch": 0.532484537997427,
      "grad_norm": 0.05173756182193756,
      "learning_rate": 0.002,
      "loss": 2.851665735244751,
      "step": 3870
    },
    {
      "epoch": 0.5327597238051777,
      "grad_norm": 0.20328296720981598,
      "learning_rate": 0.002,
      "loss": 2.8268890380859375,
      "step": 3872
    },
    {
      "epoch": 0.5330349096129282,
      "grad_norm": 0.11952225863933563,
      "learning_rate": 0.002,
      "loss": 2.9049060344696045,
      "step": 3874
    },
    {
      "epoch": 0.5333100954206789,
      "grad_norm": 0.09671736508607864,
      "learning_rate": 0.002,
      "loss": 2.8609185218811035,
      "step": 3876
    },
    {
      "epoch": 0.5335852812284294,
      "grad_norm": 0.18068063259124756,
      "learning_rate": 0.002,
      "loss": 2.824953317642212,
      "step": 3878
    },
    {
      "epoch": 0.5338604670361801,
      "grad_norm": 0.08063757419586182,
      "learning_rate": 0.002,
      "loss": 2.7952094078063965,
      "step": 3880
    },
    {
      "epoch": 0.5341356528439306,
      "grad_norm": 0.07927925884723663,
      "learning_rate": 0.002,
      "loss": 2.85377836227417,
      "step": 3882
    },
    {
      "epoch": 0.5344108386516813,
      "grad_norm": 0.045713432133197784,
      "learning_rate": 0.002,
      "loss": 2.8118653297424316,
      "step": 3884
    },
    {
      "epoch": 0.5346860244594319,
      "grad_norm": 0.02985835075378418,
      "learning_rate": 0.002,
      "loss": 2.8269543647766113,
      "step": 3886
    },
    {
      "epoch": 0.5349612102671825,
      "grad_norm": 0.15203534066677094,
      "learning_rate": 0.002,
      "loss": 2.868365526199341,
      "step": 3888
    },
    {
      "epoch": 0.5352363960749331,
      "grad_norm": 0.032998405396938324,
      "learning_rate": 0.002,
      "loss": 2.826063871383667,
      "step": 3890
    },
    {
      "epoch": 0.5355115818826837,
      "grad_norm": 0.1622799038887024,
      "learning_rate": 0.002,
      "loss": 2.8687474727630615,
      "step": 3892
    },
    {
      "epoch": 0.5357867676904343,
      "grad_norm": 0.09533314406871796,
      "learning_rate": 0.002,
      "loss": 2.8239803314208984,
      "step": 3894
    },
    {
      "epoch": 0.536061953498185,
      "grad_norm": 0.07460594922304153,
      "learning_rate": 0.002,
      "loss": 2.8184685707092285,
      "step": 3896
    },
    {
      "epoch": 0.5363371393059355,
      "grad_norm": 0.11855970323085785,
      "learning_rate": 0.002,
      "loss": 2.813217878341675,
      "step": 3898
    },
    {
      "epoch": 0.5366123251136862,
      "grad_norm": 0.08526967465877533,
      "learning_rate": 0.002,
      "loss": 2.8438570499420166,
      "step": 3900
    },
    {
      "epoch": 0.5368875109214367,
      "grad_norm": 0.16170653700828552,
      "learning_rate": 0.002,
      "loss": 2.816040277481079,
      "step": 3902
    },
    {
      "epoch": 0.5371626967291874,
      "grad_norm": 0.13283267617225647,
      "learning_rate": 0.002,
      "loss": 2.899972438812256,
      "step": 3904
    },
    {
      "epoch": 0.5374378825369379,
      "grad_norm": 0.17718611657619476,
      "learning_rate": 0.002,
      "loss": 2.837820053100586,
      "step": 3906
    },
    {
      "epoch": 0.5377130683446886,
      "grad_norm": 0.14087869226932526,
      "learning_rate": 0.002,
      "loss": 2.8456027507781982,
      "step": 3908
    },
    {
      "epoch": 0.5379882541524392,
      "grad_norm": 0.07945337146520615,
      "learning_rate": 0.002,
      "loss": 2.7900567054748535,
      "step": 3910
    },
    {
      "epoch": 0.5382634399601898,
      "grad_norm": 0.058955118060112,
      "learning_rate": 0.002,
      "loss": 2.8381357192993164,
      "step": 3912
    },
    {
      "epoch": 0.5385386257679404,
      "grad_norm": 0.07453940063714981,
      "learning_rate": 0.002,
      "loss": 2.884763240814209,
      "step": 3914
    },
    {
      "epoch": 0.538813811575691,
      "grad_norm": 0.10496264696121216,
      "learning_rate": 0.002,
      "loss": 2.818918228149414,
      "step": 3916
    },
    {
      "epoch": 0.5390889973834416,
      "grad_norm": 0.07938790321350098,
      "learning_rate": 0.002,
      "loss": 2.8328006267547607,
      "step": 3918
    },
    {
      "epoch": 0.5393641831911922,
      "grad_norm": 0.06053560599684715,
      "learning_rate": 0.002,
      "loss": 2.849944591522217,
      "step": 3920
    },
    {
      "epoch": 0.5396393689989428,
      "grad_norm": 0.3437298536300659,
      "learning_rate": 0.002,
      "loss": 2.832658052444458,
      "step": 3922
    },
    {
      "epoch": 0.5399145548066935,
      "grad_norm": 0.34989356994628906,
      "learning_rate": 0.002,
      "loss": 2.8104867935180664,
      "step": 3924
    },
    {
      "epoch": 0.540189740614444,
      "grad_norm": 0.20392337441444397,
      "learning_rate": 0.002,
      "loss": 2.8329591751098633,
      "step": 3926
    },
    {
      "epoch": 0.5404649264221947,
      "grad_norm": 0.16621147096157074,
      "learning_rate": 0.002,
      "loss": 2.8547587394714355,
      "step": 3928
    },
    {
      "epoch": 0.5407401122299452,
      "grad_norm": 0.3400348424911499,
      "learning_rate": 0.002,
      "loss": 2.8699116706848145,
      "step": 3930
    },
    {
      "epoch": 0.5410152980376959,
      "grad_norm": 0.11474677920341492,
      "learning_rate": 0.002,
      "loss": 2.8200523853302,
      "step": 3932
    },
    {
      "epoch": 0.5412904838454464,
      "grad_norm": 0.10509119927883148,
      "learning_rate": 0.002,
      "loss": 2.853060722351074,
      "step": 3934
    },
    {
      "epoch": 0.5415656696531971,
      "grad_norm": 0.2997501492500305,
      "learning_rate": 0.002,
      "loss": 2.870678424835205,
      "step": 3936
    },
    {
      "epoch": 0.5418408554609477,
      "grad_norm": 0.05695755407214165,
      "learning_rate": 0.002,
      "loss": 2.8072218894958496,
      "step": 3938
    },
    {
      "epoch": 0.5421160412686983,
      "grad_norm": 0.1423744559288025,
      "learning_rate": 0.002,
      "loss": 2.8513987064361572,
      "step": 3940
    },
    {
      "epoch": 0.5423912270764489,
      "grad_norm": 0.14426647126674652,
      "learning_rate": 0.002,
      "loss": 2.8176932334899902,
      "step": 3942
    },
    {
      "epoch": 0.5426664128841995,
      "grad_norm": 0.04046877101063728,
      "learning_rate": 0.002,
      "loss": 2.8550753593444824,
      "step": 3944
    },
    {
      "epoch": 0.5429415986919501,
      "grad_norm": 0.03149569779634476,
      "learning_rate": 0.002,
      "loss": 2.821286201477051,
      "step": 3946
    },
    {
      "epoch": 0.5432167844997008,
      "grad_norm": 0.03449452668428421,
      "learning_rate": 0.002,
      "loss": 2.8306658267974854,
      "step": 3948
    },
    {
      "epoch": 0.5434919703074513,
      "grad_norm": 0.054681114852428436,
      "learning_rate": 0.002,
      "loss": 2.8323307037353516,
      "step": 3950
    },
    {
      "epoch": 0.543767156115202,
      "grad_norm": 0.03204827383160591,
      "learning_rate": 0.002,
      "loss": 2.8482120037078857,
      "step": 3952
    },
    {
      "epoch": 0.5440423419229525,
      "grad_norm": 0.11560880392789841,
      "learning_rate": 0.002,
      "loss": 2.8279459476470947,
      "step": 3954
    },
    {
      "epoch": 0.5443175277307032,
      "grad_norm": 0.04844314977526665,
      "learning_rate": 0.002,
      "loss": 2.8445191383361816,
      "step": 3956
    },
    {
      "epoch": 0.5445927135384537,
      "grad_norm": 0.18341617286205292,
      "learning_rate": 0.002,
      "loss": 2.8512842655181885,
      "step": 3958
    },
    {
      "epoch": 0.5448678993462044,
      "grad_norm": 0.2060878574848175,
      "learning_rate": 0.002,
      "loss": 2.8462514877319336,
      "step": 3960
    },
    {
      "epoch": 0.545143085153955,
      "grad_norm": 0.030807657167315483,
      "learning_rate": 0.002,
      "loss": 2.8753514289855957,
      "step": 3962
    },
    {
      "epoch": 0.5454182709617056,
      "grad_norm": 0.04456183686852455,
      "learning_rate": 0.002,
      "loss": 2.882181167602539,
      "step": 3964
    },
    {
      "epoch": 0.5456934567694562,
      "grad_norm": 0.04193778708577156,
      "learning_rate": 0.002,
      "loss": 2.829832077026367,
      "step": 3966
    },
    {
      "epoch": 0.5459686425772068,
      "grad_norm": 0.03760799765586853,
      "learning_rate": 0.002,
      "loss": 2.852335214614868,
      "step": 3968
    },
    {
      "epoch": 0.5462438283849574,
      "grad_norm": 0.03965524584054947,
      "learning_rate": 0.002,
      "loss": 2.822469711303711,
      "step": 3970
    },
    {
      "epoch": 0.546519014192708,
      "grad_norm": 0.07627121359109879,
      "learning_rate": 0.002,
      "loss": 2.826882839202881,
      "step": 3972
    },
    {
      "epoch": 0.5467942000004586,
      "grad_norm": 0.18852916359901428,
      "learning_rate": 0.002,
      "loss": 2.804994821548462,
      "step": 3974
    },
    {
      "epoch": 0.5470693858082093,
      "grad_norm": 0.25866252183914185,
      "learning_rate": 0.002,
      "loss": 2.8412375450134277,
      "step": 3976
    },
    {
      "epoch": 0.5473445716159598,
      "grad_norm": 0.15695197880268097,
      "learning_rate": 0.002,
      "loss": 2.7869327068328857,
      "step": 3978
    },
    {
      "epoch": 0.5476197574237105,
      "grad_norm": 0.09828200936317444,
      "learning_rate": 0.002,
      "loss": 2.8504321575164795,
      "step": 3980
    },
    {
      "epoch": 0.547894943231461,
      "grad_norm": 0.31818273663520813,
      "learning_rate": 0.002,
      "loss": 2.8472371101379395,
      "step": 3982
    },
    {
      "epoch": 0.5481701290392117,
      "grad_norm": 0.06718368828296661,
      "learning_rate": 0.002,
      "loss": 2.7884747982025146,
      "step": 3984
    },
    {
      "epoch": 0.5484453148469622,
      "grad_norm": 0.3519662916660309,
      "learning_rate": 0.002,
      "loss": 2.826625347137451,
      "step": 3986
    },
    {
      "epoch": 0.5487205006547129,
      "grad_norm": 0.06551429629325867,
      "learning_rate": 0.002,
      "loss": 2.8598122596740723,
      "step": 3988
    },
    {
      "epoch": 0.5489956864624636,
      "grad_norm": 0.21639949083328247,
      "learning_rate": 0.002,
      "loss": 2.83610200881958,
      "step": 3990
    },
    {
      "epoch": 0.5492708722702141,
      "grad_norm": 0.18006151914596558,
      "learning_rate": 0.002,
      "loss": 2.8310506343841553,
      "step": 3992
    },
    {
      "epoch": 0.5495460580779647,
      "grad_norm": 0.09218730032444,
      "learning_rate": 0.002,
      "loss": 2.8514933586120605,
      "step": 3994
    },
    {
      "epoch": 0.5498212438857153,
      "grad_norm": 0.03925531730055809,
      "learning_rate": 0.002,
      "loss": 2.8469655513763428,
      "step": 3996
    },
    {
      "epoch": 0.550096429693466,
      "grad_norm": 0.15459169447422028,
      "learning_rate": 0.002,
      "loss": 2.852483034133911,
      "step": 3998
    },
    {
      "epoch": 0.5503716155012166,
      "grad_norm": 0.061109308153390884,
      "learning_rate": 0.002,
      "loss": 2.804816722869873,
      "step": 4000
    },
    {
      "epoch": 0.5506468013089671,
      "grad_norm": 0.08709947764873505,
      "learning_rate": 0.002,
      "loss": 2.8506832122802734,
      "step": 4002
    },
    {
      "epoch": 0.5509219871167178,
      "grad_norm": 0.0968494787812233,
      "learning_rate": 0.002,
      "loss": 2.785526752471924,
      "step": 4004
    },
    {
      "epoch": 0.5511971729244683,
      "grad_norm": 0.16058965027332306,
      "learning_rate": 0.002,
      "loss": 2.811122417449951,
      "step": 4006
    },
    {
      "epoch": 0.551472358732219,
      "grad_norm": 0.07241848856210709,
      "learning_rate": 0.002,
      "loss": 2.7979979515075684,
      "step": 4008
    },
    {
      "epoch": 0.5517475445399695,
      "grad_norm": 0.1091693714261055,
      "learning_rate": 0.002,
      "loss": 2.8377151489257812,
      "step": 4010
    },
    {
      "epoch": 0.5520227303477202,
      "grad_norm": 0.22626769542694092,
      "learning_rate": 0.002,
      "loss": 2.8437585830688477,
      "step": 4012
    },
    {
      "epoch": 0.5522979161554709,
      "grad_norm": 0.09697632491588593,
      "learning_rate": 0.002,
      "loss": 2.7774040699005127,
      "step": 4014
    },
    {
      "epoch": 0.5525731019632214,
      "grad_norm": 0.13649965822696686,
      "learning_rate": 0.002,
      "loss": 2.801882266998291,
      "step": 4016
    },
    {
      "epoch": 0.5528482877709721,
      "grad_norm": 0.029167158529162407,
      "learning_rate": 0.002,
      "loss": 2.811702013015747,
      "step": 4018
    },
    {
      "epoch": 0.5531234735787226,
      "grad_norm": 0.15237388014793396,
      "learning_rate": 0.002,
      "loss": 2.8302536010742188,
      "step": 4020
    },
    {
      "epoch": 0.5533986593864733,
      "grad_norm": 0.1477382779121399,
      "learning_rate": 0.002,
      "loss": 2.8353829383850098,
      "step": 4022
    },
    {
      "epoch": 0.5536738451942238,
      "grad_norm": 0.05414079874753952,
      "learning_rate": 0.002,
      "loss": 2.8295178413391113,
      "step": 4024
    },
    {
      "epoch": 0.5539490310019745,
      "grad_norm": 0.14717169106006622,
      "learning_rate": 0.002,
      "loss": 2.8252785205841064,
      "step": 4026
    },
    {
      "epoch": 0.5542242168097251,
      "grad_norm": 0.17679397761821747,
      "learning_rate": 0.002,
      "loss": 2.827923059463501,
      "step": 4028
    },
    {
      "epoch": 0.5544994026174757,
      "grad_norm": 0.1438676118850708,
      "learning_rate": 0.002,
      "loss": 2.8060781955718994,
      "step": 4030
    },
    {
      "epoch": 0.5547745884252263,
      "grad_norm": 0.05143098905682564,
      "learning_rate": 0.002,
      "loss": 2.8836941719055176,
      "step": 4032
    },
    {
      "epoch": 0.5550497742329769,
      "grad_norm": 0.03436022996902466,
      "learning_rate": 0.002,
      "loss": 2.838423728942871,
      "step": 4034
    },
    {
      "epoch": 0.5553249600407275,
      "grad_norm": 0.15609559416770935,
      "learning_rate": 0.002,
      "loss": 2.8591647148132324,
      "step": 4036
    },
    {
      "epoch": 0.555600145848478,
      "grad_norm": 0.30525052547454834,
      "learning_rate": 0.002,
      "loss": 2.8476943969726562,
      "step": 4038
    },
    {
      "epoch": 0.5558753316562287,
      "grad_norm": 0.2829124331474304,
      "learning_rate": 0.002,
      "loss": 2.8234691619873047,
      "step": 4040
    },
    {
      "epoch": 0.5561505174639794,
      "grad_norm": 0.20593222975730896,
      "learning_rate": 0.002,
      "loss": 2.796415328979492,
      "step": 4042
    },
    {
      "epoch": 0.5564257032717299,
      "grad_norm": 0.14253316819667816,
      "learning_rate": 0.002,
      "loss": 2.8419508934020996,
      "step": 4044
    },
    {
      "epoch": 0.5567008890794806,
      "grad_norm": 0.18019071221351624,
      "learning_rate": 0.002,
      "loss": 2.797551155090332,
      "step": 4046
    },
    {
      "epoch": 0.5569760748872311,
      "grad_norm": 0.11487998813390732,
      "learning_rate": 0.002,
      "loss": 2.8350372314453125,
      "step": 4048
    },
    {
      "epoch": 0.5572512606949818,
      "grad_norm": 0.1171715185046196,
      "learning_rate": 0.002,
      "loss": 2.8032474517822266,
      "step": 4050
    },
    {
      "epoch": 0.5575264465027324,
      "grad_norm": 0.13443514704704285,
      "learning_rate": 0.002,
      "loss": 2.86014986038208,
      "step": 4052
    },
    {
      "epoch": 0.557801632310483,
      "grad_norm": 0.22648777067661285,
      "learning_rate": 0.002,
      "loss": 2.8295860290527344,
      "step": 4054
    },
    {
      "epoch": 0.5580768181182336,
      "grad_norm": 0.17185074090957642,
      "learning_rate": 0.002,
      "loss": 2.845520496368408,
      "step": 4056
    },
    {
      "epoch": 0.5583520039259842,
      "grad_norm": 0.1518024206161499,
      "learning_rate": 0.002,
      "loss": 2.8141183853149414,
      "step": 4058
    },
    {
      "epoch": 0.5586271897337348,
      "grad_norm": 0.09493538737297058,
      "learning_rate": 0.002,
      "loss": 2.8114466667175293,
      "step": 4060
    },
    {
      "epoch": 0.5589023755414854,
      "grad_norm": 0.05634348466992378,
      "learning_rate": 0.002,
      "loss": 2.8142848014831543,
      "step": 4062
    },
    {
      "epoch": 0.559177561349236,
      "grad_norm": 0.1316581517457962,
      "learning_rate": 0.002,
      "loss": 2.799929618835449,
      "step": 4064
    },
    {
      "epoch": 0.5594527471569867,
      "grad_norm": 0.08365991711616516,
      "learning_rate": 0.002,
      "loss": 2.845879077911377,
      "step": 4066
    },
    {
      "epoch": 0.5597279329647372,
      "grad_norm": 0.09110312908887863,
      "learning_rate": 0.002,
      "loss": 2.82736873626709,
      "step": 4068
    },
    {
      "epoch": 0.5600031187724879,
      "grad_norm": 0.18346436321735382,
      "learning_rate": 0.002,
      "loss": 2.8339052200317383,
      "step": 4070
    },
    {
      "epoch": 0.5602783045802384,
      "grad_norm": 0.1255030483007431,
      "learning_rate": 0.002,
      "loss": 2.7941207885742188,
      "step": 4072
    },
    {
      "epoch": 0.5605534903879891,
      "grad_norm": 0.06710512936115265,
      "learning_rate": 0.002,
      "loss": 2.7911322116851807,
      "step": 4074
    },
    {
      "epoch": 0.5608286761957396,
      "grad_norm": 0.03760185465216637,
      "learning_rate": 0.002,
      "loss": 2.827227830886841,
      "step": 4076
    },
    {
      "epoch": 0.5611038620034903,
      "grad_norm": 0.041978709399700165,
      "learning_rate": 0.002,
      "loss": 2.8247716426849365,
      "step": 4078
    },
    {
      "epoch": 0.5613790478112409,
      "grad_norm": 0.189247727394104,
      "learning_rate": 0.002,
      "loss": 2.8795599937438965,
      "step": 4080
    },
    {
      "epoch": 0.5616542336189915,
      "grad_norm": 0.365403413772583,
      "learning_rate": 0.002,
      "loss": 2.861321449279785,
      "step": 4082
    },
    {
      "epoch": 0.5619294194267421,
      "grad_norm": 0.0816722959280014,
      "learning_rate": 0.002,
      "loss": 2.8467936515808105,
      "step": 4084
    },
    {
      "epoch": 0.5622046052344927,
      "grad_norm": 0.30655914545059204,
      "learning_rate": 0.002,
      "loss": 2.8144631385803223,
      "step": 4086
    },
    {
      "epoch": 0.5624797910422433,
      "grad_norm": 0.05291673541069031,
      "learning_rate": 0.002,
      "loss": 2.797710418701172,
      "step": 4088
    },
    {
      "epoch": 0.5627549768499939,
      "grad_norm": 0.11954104155302048,
      "learning_rate": 0.002,
      "loss": 2.7982165813446045,
      "step": 4090
    },
    {
      "epoch": 0.5630301626577445,
      "grad_norm": 0.03172395005822182,
      "learning_rate": 0.002,
      "loss": 2.834428310394287,
      "step": 4092
    },
    {
      "epoch": 0.5633053484654952,
      "grad_norm": 0.06624466180801392,
      "learning_rate": 0.002,
      "loss": 2.852962017059326,
      "step": 4094
    },
    {
      "epoch": 0.5635805342732457,
      "grad_norm": 0.10029198229312897,
      "learning_rate": 0.002,
      "loss": 2.824956178665161,
      "step": 4096
    },
    {
      "epoch": 0.5638557200809964,
      "grad_norm": 0.031509846448898315,
      "learning_rate": 0.002,
      "loss": 2.7788166999816895,
      "step": 4098
    },
    {
      "epoch": 0.5641309058887469,
      "grad_norm": 0.1421998292207718,
      "learning_rate": 0.002,
      "loss": 2.808432102203369,
      "step": 4100
    },
    {
      "epoch": 0.5644060916964976,
      "grad_norm": 0.21269859373569489,
      "learning_rate": 0.002,
      "loss": 2.8594374656677246,
      "step": 4102
    },
    {
      "epoch": 0.5646812775042482,
      "grad_norm": 0.05762547254562378,
      "learning_rate": 0.002,
      "loss": 2.839628219604492,
      "step": 4104
    },
    {
      "epoch": 0.5649564633119988,
      "grad_norm": 0.1290253847837448,
      "learning_rate": 0.002,
      "loss": 2.848994255065918,
      "step": 4106
    },
    {
      "epoch": 0.5652316491197494,
      "grad_norm": 0.13909083604812622,
      "learning_rate": 0.002,
      "loss": 2.8639895915985107,
      "step": 4108
    },
    {
      "epoch": 0.5655068349275,
      "grad_norm": 0.05075880512595177,
      "learning_rate": 0.002,
      "loss": 2.8310256004333496,
      "step": 4110
    },
    {
      "epoch": 0.5657820207352506,
      "grad_norm": 0.17113536596298218,
      "learning_rate": 0.002,
      "loss": 2.8360390663146973,
      "step": 4112
    },
    {
      "epoch": 0.5660572065430012,
      "grad_norm": 0.09810604900121689,
      "learning_rate": 0.002,
      "loss": 2.8112244606018066,
      "step": 4114
    },
    {
      "epoch": 0.5663323923507518,
      "grad_norm": 0.08843237161636353,
      "learning_rate": 0.002,
      "loss": 2.8066554069519043,
      "step": 4116
    },
    {
      "epoch": 0.5666075781585025,
      "grad_norm": 0.03033740632236004,
      "learning_rate": 0.002,
      "loss": 2.8388845920562744,
      "step": 4118
    },
    {
      "epoch": 0.566882763966253,
      "grad_norm": 0.04484846442937851,
      "learning_rate": 0.002,
      "loss": 2.8178019523620605,
      "step": 4120
    },
    {
      "epoch": 0.5671579497740037,
      "grad_norm": 0.07417047023773193,
      "learning_rate": 0.002,
      "loss": 2.846020221710205,
      "step": 4122
    },
    {
      "epoch": 0.5674331355817542,
      "grad_norm": 0.24591322243213654,
      "learning_rate": 0.002,
      "loss": 2.821303606033325,
      "step": 4124
    },
    {
      "epoch": 0.5677083213895049,
      "grad_norm": 0.43602317571640015,
      "learning_rate": 0.002,
      "loss": 2.824016809463501,
      "step": 4126
    },
    {
      "epoch": 0.5679835071972554,
      "grad_norm": 0.20647937059402466,
      "learning_rate": 0.002,
      "loss": 2.8158111572265625,
      "step": 4128
    },
    {
      "epoch": 0.5682586930050061,
      "grad_norm": 0.13585242629051208,
      "learning_rate": 0.002,
      "loss": 2.8227338790893555,
      "step": 4130
    },
    {
      "epoch": 0.5685338788127567,
      "grad_norm": 0.16605012118816376,
      "learning_rate": 0.002,
      "loss": 2.8151044845581055,
      "step": 4132
    },
    {
      "epoch": 0.5688090646205073,
      "grad_norm": 0.04301491007208824,
      "learning_rate": 0.002,
      "loss": 2.8235561847686768,
      "step": 4134
    },
    {
      "epoch": 0.5690842504282579,
      "grad_norm": 0.06466622650623322,
      "learning_rate": 0.002,
      "loss": 2.8449807167053223,
      "step": 4136
    },
    {
      "epoch": 0.5693594362360085,
      "grad_norm": 0.18077492713928223,
      "learning_rate": 0.002,
      "loss": 2.806946277618408,
      "step": 4138
    },
    {
      "epoch": 0.5696346220437591,
      "grad_norm": 0.04964018240571022,
      "learning_rate": 0.002,
      "loss": 2.8126707077026367,
      "step": 4140
    },
    {
      "epoch": 0.5699098078515097,
      "grad_norm": 0.09212277829647064,
      "learning_rate": 0.002,
      "loss": 2.8429336547851562,
      "step": 4142
    },
    {
      "epoch": 0.5701849936592603,
      "grad_norm": 0.11324101686477661,
      "learning_rate": 0.002,
      "loss": 2.8168253898620605,
      "step": 4144
    },
    {
      "epoch": 0.570460179467011,
      "grad_norm": 0.04046540707349777,
      "learning_rate": 0.002,
      "loss": 2.8126039505004883,
      "step": 4146
    },
    {
      "epoch": 0.5707353652747615,
      "grad_norm": 0.08459272235631943,
      "learning_rate": 0.002,
      "loss": 2.8192906379699707,
      "step": 4148
    },
    {
      "epoch": 0.5710105510825122,
      "grad_norm": 0.1620699018239975,
      "learning_rate": 0.002,
      "loss": 2.7885982990264893,
      "step": 4150
    },
    {
      "epoch": 0.5712857368902627,
      "grad_norm": 0.20151282846927643,
      "learning_rate": 0.002,
      "loss": 2.8081676959991455,
      "step": 4152
    },
    {
      "epoch": 0.5715609226980134,
      "grad_norm": 0.055882420390844345,
      "learning_rate": 0.002,
      "loss": 2.7962074279785156,
      "step": 4154
    },
    {
      "epoch": 0.571836108505764,
      "grad_norm": 0.042874038219451904,
      "learning_rate": 0.002,
      "loss": 2.8296666145324707,
      "step": 4156
    },
    {
      "epoch": 0.5721112943135146,
      "grad_norm": 0.06986548006534576,
      "learning_rate": 0.002,
      "loss": 2.8338544368743896,
      "step": 4158
    },
    {
      "epoch": 0.5723864801212653,
      "grad_norm": 0.048385150730609894,
      "learning_rate": 0.002,
      "loss": 2.8150038719177246,
      "step": 4160
    },
    {
      "epoch": 0.5726616659290158,
      "grad_norm": 0.04279249534010887,
      "learning_rate": 0.002,
      "loss": 2.8402490615844727,
      "step": 4162
    },
    {
      "epoch": 0.5729368517367665,
      "grad_norm": 0.07447534799575806,
      "learning_rate": 0.002,
      "loss": 2.792879581451416,
      "step": 4164
    },
    {
      "epoch": 0.573212037544517,
      "grad_norm": 0.07524089515209198,
      "learning_rate": 0.002,
      "loss": 2.81691837310791,
      "step": 4166
    },
    {
      "epoch": 0.5734872233522676,
      "grad_norm": 0.15807205438613892,
      "learning_rate": 0.002,
      "loss": 2.8363277912139893,
      "step": 4168
    },
    {
      "epoch": 0.5737624091600183,
      "grad_norm": 0.2516591548919678,
      "learning_rate": 0.002,
      "loss": 2.785977363586426,
      "step": 4170
    },
    {
      "epoch": 0.5740375949677688,
      "grad_norm": 0.2783066928386688,
      "learning_rate": 0.002,
      "loss": 2.7693727016448975,
      "step": 4172
    },
    {
      "epoch": 0.5743127807755195,
      "grad_norm": 0.1312628835439682,
      "learning_rate": 0.002,
      "loss": 2.824554204940796,
      "step": 4174
    },
    {
      "epoch": 0.57458796658327,
      "grad_norm": 0.2211017906665802,
      "learning_rate": 0.002,
      "loss": 2.7876687049865723,
      "step": 4176
    },
    {
      "epoch": 0.5748631523910207,
      "grad_norm": 0.06453431397676468,
      "learning_rate": 0.002,
      "loss": 2.817774772644043,
      "step": 4178
    },
    {
      "epoch": 0.5751383381987712,
      "grad_norm": 0.03413352370262146,
      "learning_rate": 0.002,
      "loss": 2.782939910888672,
      "step": 4180
    },
    {
      "epoch": 0.5754135240065219,
      "grad_norm": 0.03697779402136803,
      "learning_rate": 0.002,
      "loss": 2.8099920749664307,
      "step": 4182
    },
    {
      "epoch": 0.5756887098142726,
      "grad_norm": 0.10921536386013031,
      "learning_rate": 0.002,
      "loss": 2.844834089279175,
      "step": 4184
    },
    {
      "epoch": 0.5759638956220231,
      "grad_norm": 0.04804648086428642,
      "learning_rate": 0.002,
      "loss": 2.826334238052368,
      "step": 4186
    },
    {
      "epoch": 0.5762390814297738,
      "grad_norm": 0.11907287687063217,
      "learning_rate": 0.002,
      "loss": 2.8342084884643555,
      "step": 4188
    },
    {
      "epoch": 0.5765142672375243,
      "grad_norm": 0.3271154761314392,
      "learning_rate": 0.002,
      "loss": 2.853644847869873,
      "step": 4190
    },
    {
      "epoch": 0.576789453045275,
      "grad_norm": 0.12375056743621826,
      "learning_rate": 0.002,
      "loss": 2.8042688369750977,
      "step": 4192
    },
    {
      "epoch": 0.5770646388530255,
      "grad_norm": 0.2571203112602234,
      "learning_rate": 0.002,
      "loss": 2.785330057144165,
      "step": 4194
    },
    {
      "epoch": 0.5773398246607762,
      "grad_norm": 0.09425774961709976,
      "learning_rate": 0.002,
      "loss": 2.797410488128662,
      "step": 4196
    },
    {
      "epoch": 0.5776150104685268,
      "grad_norm": 0.3238604664802551,
      "learning_rate": 0.002,
      "loss": 2.7938992977142334,
      "step": 4198
    },
    {
      "epoch": 0.5778901962762774,
      "grad_norm": 0.10715726763010025,
      "learning_rate": 0.002,
      "loss": 2.81929349899292,
      "step": 4200
    },
    {
      "epoch": 0.578165382084028,
      "grad_norm": 0.20491698384284973,
      "learning_rate": 0.002,
      "loss": 2.8172545433044434,
      "step": 4202
    },
    {
      "epoch": 0.5784405678917786,
      "grad_norm": 0.3234810531139374,
      "learning_rate": 0.002,
      "loss": 2.7812488079071045,
      "step": 4204
    },
    {
      "epoch": 0.5787157536995292,
      "grad_norm": 0.04565688595175743,
      "learning_rate": 0.002,
      "loss": 2.820589303970337,
      "step": 4206
    },
    {
      "epoch": 0.5789909395072798,
      "grad_norm": 0.23579943180084229,
      "learning_rate": 0.002,
      "loss": 2.851109266281128,
      "step": 4208
    },
    {
      "epoch": 0.5792661253150304,
      "grad_norm": 0.16554109752178192,
      "learning_rate": 0.002,
      "loss": 2.7987160682678223,
      "step": 4210
    },
    {
      "epoch": 0.5795413111227811,
      "grad_norm": 0.035486772656440735,
      "learning_rate": 0.002,
      "loss": 2.818202018737793,
      "step": 4212
    },
    {
      "epoch": 0.5798164969305316,
      "grad_norm": 0.09854763001203537,
      "learning_rate": 0.002,
      "loss": 2.8016839027404785,
      "step": 4214
    },
    {
      "epoch": 0.5800916827382823,
      "grad_norm": 0.17276576161384583,
      "learning_rate": 0.002,
      "loss": 2.8463518619537354,
      "step": 4216
    },
    {
      "epoch": 0.5803668685460328,
      "grad_norm": 0.045605044811964035,
      "learning_rate": 0.002,
      "loss": 2.783705234527588,
      "step": 4218
    },
    {
      "epoch": 0.5806420543537835,
      "grad_norm": 0.22184215486049652,
      "learning_rate": 0.002,
      "loss": 2.8226139545440674,
      "step": 4220
    },
    {
      "epoch": 0.5809172401615341,
      "grad_norm": 0.14769861102104187,
      "learning_rate": 0.002,
      "loss": 2.806519031524658,
      "step": 4222
    },
    {
      "epoch": 0.5811924259692847,
      "grad_norm": 0.12942250072956085,
      "learning_rate": 0.002,
      "loss": 2.8002941608428955,
      "step": 4224
    },
    {
      "epoch": 0.5814676117770353,
      "grad_norm": 0.0435965321958065,
      "learning_rate": 0.002,
      "loss": 2.835242748260498,
      "step": 4226
    },
    {
      "epoch": 0.5817427975847859,
      "grad_norm": 0.033432431519031525,
      "learning_rate": 0.002,
      "loss": 2.8259809017181396,
      "step": 4228
    },
    {
      "epoch": 0.5820179833925365,
      "grad_norm": 0.05798964202404022,
      "learning_rate": 0.002,
      "loss": 2.777311325073242,
      "step": 4230
    },
    {
      "epoch": 0.5822931692002871,
      "grad_norm": 0.031373485922813416,
      "learning_rate": 0.002,
      "loss": 2.7991433143615723,
      "step": 4232
    },
    {
      "epoch": 0.5825683550080377,
      "grad_norm": 0.06389721482992172,
      "learning_rate": 0.002,
      "loss": 2.811476707458496,
      "step": 4234
    },
    {
      "epoch": 0.5828435408157884,
      "grad_norm": 0.10603220760822296,
      "learning_rate": 0.002,
      "loss": 2.796243190765381,
      "step": 4236
    },
    {
      "epoch": 0.5831187266235389,
      "grad_norm": 0.10496234893798828,
      "learning_rate": 0.002,
      "loss": 2.859818458557129,
      "step": 4238
    },
    {
      "epoch": 0.5833939124312896,
      "grad_norm": 0.08232710510492325,
      "learning_rate": 0.002,
      "loss": 2.821084976196289,
      "step": 4240
    },
    {
      "epoch": 0.5836690982390401,
      "grad_norm": 0.15262793004512787,
      "learning_rate": 0.002,
      "loss": 2.828543186187744,
      "step": 4242
    },
    {
      "epoch": 0.5839442840467908,
      "grad_norm": 0.2946659028530121,
      "learning_rate": 0.002,
      "loss": 2.7904210090637207,
      "step": 4244
    },
    {
      "epoch": 0.5842194698545413,
      "grad_norm": 0.18007950484752655,
      "learning_rate": 0.002,
      "loss": 2.8059353828430176,
      "step": 4246
    },
    {
      "epoch": 0.584494655662292,
      "grad_norm": 0.22771090269088745,
      "learning_rate": 0.002,
      "loss": 2.802110433578491,
      "step": 4248
    },
    {
      "epoch": 0.5847698414700426,
      "grad_norm": 0.1346217691898346,
      "learning_rate": 0.002,
      "loss": 2.7859601974487305,
      "step": 4250
    },
    {
      "epoch": 0.5850450272777932,
      "grad_norm": 0.09865598380565643,
      "learning_rate": 0.002,
      "loss": 2.8112378120422363,
      "step": 4252
    },
    {
      "epoch": 0.5853202130855438,
      "grad_norm": 0.1477927714586258,
      "learning_rate": 0.002,
      "loss": 2.815988540649414,
      "step": 4254
    },
    {
      "epoch": 0.5855953988932944,
      "grad_norm": 0.045640312135219574,
      "learning_rate": 0.002,
      "loss": 2.79360032081604,
      "step": 4256
    },
    {
      "epoch": 0.585870584701045,
      "grad_norm": 0.08416791260242462,
      "learning_rate": 0.002,
      "loss": 2.8284683227539062,
      "step": 4258
    },
    {
      "epoch": 0.5861457705087956,
      "grad_norm": 0.12042475491762161,
      "learning_rate": 0.002,
      "loss": 2.82444167137146,
      "step": 4260
    },
    {
      "epoch": 0.5864209563165462,
      "grad_norm": 0.03216049075126648,
      "learning_rate": 0.002,
      "loss": 2.81406569480896,
      "step": 4262
    },
    {
      "epoch": 0.5866961421242969,
      "grad_norm": 0.1184920072555542,
      "learning_rate": 0.002,
      "loss": 2.8256235122680664,
      "step": 4264
    },
    {
      "epoch": 0.5869713279320474,
      "grad_norm": 0.12069770693778992,
      "learning_rate": 0.002,
      "loss": 2.8202691078186035,
      "step": 4266
    },
    {
      "epoch": 0.5872465137397981,
      "grad_norm": 0.10419037938117981,
      "learning_rate": 0.002,
      "loss": 2.8235535621643066,
      "step": 4268
    },
    {
      "epoch": 0.5875216995475486,
      "grad_norm": 0.040743302553892136,
      "learning_rate": 0.002,
      "loss": 2.8264365196228027,
      "step": 4270
    },
    {
      "epoch": 0.5877968853552993,
      "grad_norm": 0.049732763320207596,
      "learning_rate": 0.002,
      "loss": 2.835073709487915,
      "step": 4272
    },
    {
      "epoch": 0.5880720711630499,
      "grad_norm": 0.13046886026859283,
      "learning_rate": 0.002,
      "loss": 2.8122105598449707,
      "step": 4274
    },
    {
      "epoch": 0.5883472569708005,
      "grad_norm": 0.49406251311302185,
      "learning_rate": 0.002,
      "loss": 2.808103561401367,
      "step": 4276
    },
    {
      "epoch": 0.5886224427785511,
      "grad_norm": 0.12534360587596893,
      "learning_rate": 0.002,
      "loss": 2.813481330871582,
      "step": 4278
    },
    {
      "epoch": 0.5888976285863017,
      "grad_norm": 0.38703593611717224,
      "learning_rate": 0.002,
      "loss": 2.7730600833892822,
      "step": 4280
    },
    {
      "epoch": 0.5891728143940523,
      "grad_norm": 0.3453435003757477,
      "learning_rate": 0.002,
      "loss": 2.8012099266052246,
      "step": 4282
    },
    {
      "epoch": 0.5894480002018029,
      "grad_norm": 0.10793429613113403,
      "learning_rate": 0.002,
      "loss": 2.81010103225708,
      "step": 4284
    },
    {
      "epoch": 0.5897231860095535,
      "grad_norm": 0.09862425178289413,
      "learning_rate": 0.002,
      "loss": 2.8399038314819336,
      "step": 4286
    },
    {
      "epoch": 0.5899983718173042,
      "grad_norm": 0.3023028075695038,
      "learning_rate": 0.002,
      "loss": 2.7525601387023926,
      "step": 4288
    },
    {
      "epoch": 0.5902735576250547,
      "grad_norm": 0.11813991516828537,
      "learning_rate": 0.002,
      "loss": 2.804765224456787,
      "step": 4290
    },
    {
      "epoch": 0.5905487434328054,
      "grad_norm": 0.0303591750562191,
      "learning_rate": 0.002,
      "loss": 2.8026223182678223,
      "step": 4292
    },
    {
      "epoch": 0.5908239292405559,
      "grad_norm": 0.0844598114490509,
      "learning_rate": 0.002,
      "loss": 2.793119192123413,
      "step": 4294
    },
    {
      "epoch": 0.5910991150483066,
      "grad_norm": 0.04297167435288429,
      "learning_rate": 0.002,
      "loss": 2.8022241592407227,
      "step": 4296
    },
    {
      "epoch": 0.5913743008560571,
      "grad_norm": 0.04577907919883728,
      "learning_rate": 0.002,
      "loss": 2.824225425720215,
      "step": 4298
    },
    {
      "epoch": 0.5916494866638078,
      "grad_norm": 0.12799882888793945,
      "learning_rate": 0.002,
      "loss": 2.8478927612304688,
      "step": 4300
    },
    {
      "epoch": 0.5919246724715584,
      "grad_norm": 0.03000240959227085,
      "learning_rate": 0.002,
      "loss": 2.823474645614624,
      "step": 4302
    },
    {
      "epoch": 0.592199858279309,
      "grad_norm": 0.11237344890832901,
      "learning_rate": 0.002,
      "loss": 2.8065404891967773,
      "step": 4304
    },
    {
      "epoch": 0.5924750440870596,
      "grad_norm": 0.06779836118221283,
      "learning_rate": 0.002,
      "loss": 2.800780773162842,
      "step": 4306
    },
    {
      "epoch": 0.5927502298948102,
      "grad_norm": 0.045478455722332,
      "learning_rate": 0.002,
      "loss": 2.822380542755127,
      "step": 4308
    },
    {
      "epoch": 0.5930254157025608,
      "grad_norm": 0.05432621017098427,
      "learning_rate": 0.002,
      "loss": 2.8183650970458984,
      "step": 4310
    },
    {
      "epoch": 0.5933006015103114,
      "grad_norm": 0.03005114011466503,
      "learning_rate": 0.002,
      "loss": 2.841068983078003,
      "step": 4312
    },
    {
      "epoch": 0.593575787318062,
      "grad_norm": 0.03023904748260975,
      "learning_rate": 0.002,
      "loss": 2.743952751159668,
      "step": 4314
    },
    {
      "epoch": 0.5938509731258127,
      "grad_norm": 0.19337908923625946,
      "learning_rate": 0.002,
      "loss": 2.8245577812194824,
      "step": 4316
    },
    {
      "epoch": 0.5941261589335632,
      "grad_norm": 0.3029247522354126,
      "learning_rate": 0.002,
      "loss": 2.783801555633545,
      "step": 4318
    },
    {
      "epoch": 0.5944013447413139,
      "grad_norm": 0.03823729604482651,
      "learning_rate": 0.002,
      "loss": 2.7830686569213867,
      "step": 4320
    },
    {
      "epoch": 0.5946765305490644,
      "grad_norm": 0.26123368740081787,
      "learning_rate": 0.002,
      "loss": 2.7628023624420166,
      "step": 4322
    },
    {
      "epoch": 0.5949517163568151,
      "grad_norm": 0.038104940205812454,
      "learning_rate": 0.002,
      "loss": 2.7941410541534424,
      "step": 4324
    },
    {
      "epoch": 0.5952269021645658,
      "grad_norm": 0.19346947968006134,
      "learning_rate": 0.002,
      "loss": 2.7751026153564453,
      "step": 4326
    },
    {
      "epoch": 0.5955020879723163,
      "grad_norm": 0.03318820893764496,
      "learning_rate": 0.002,
      "loss": 2.8204946517944336,
      "step": 4328
    },
    {
      "epoch": 0.595777273780067,
      "grad_norm": 0.033488623797893524,
      "learning_rate": 0.002,
      "loss": 2.830216884613037,
      "step": 4330
    },
    {
      "epoch": 0.5960524595878175,
      "grad_norm": 0.05938156694173813,
      "learning_rate": 0.002,
      "loss": 2.785964250564575,
      "step": 4332
    },
    {
      "epoch": 0.5963276453955682,
      "grad_norm": 0.03774993494153023,
      "learning_rate": 0.002,
      "loss": 2.784371852874756,
      "step": 4334
    },
    {
      "epoch": 0.5966028312033187,
      "grad_norm": 0.04373997449874878,
      "learning_rate": 0.002,
      "loss": 2.817739486694336,
      "step": 4336
    },
    {
      "epoch": 0.5968780170110694,
      "grad_norm": 0.0629860907793045,
      "learning_rate": 0.002,
      "loss": 2.7867019176483154,
      "step": 4338
    },
    {
      "epoch": 0.59715320281882,
      "grad_norm": 0.1701185554265976,
      "learning_rate": 0.002,
      "loss": 2.832479953765869,
      "step": 4340
    },
    {
      "epoch": 0.5974283886265706,
      "grad_norm": 0.33860406279563904,
      "learning_rate": 0.002,
      "loss": 2.754943370819092,
      "step": 4342
    },
    {
      "epoch": 0.5977035744343212,
      "grad_norm": 0.2724539637565613,
      "learning_rate": 0.002,
      "loss": 2.808568000793457,
      "step": 4344
    },
    {
      "epoch": 0.5979787602420717,
      "grad_norm": 0.10357118397951126,
      "learning_rate": 0.002,
      "loss": 2.8391315937042236,
      "step": 4346
    },
    {
      "epoch": 0.5982539460498224,
      "grad_norm": 0.19817793369293213,
      "learning_rate": 0.002,
      "loss": 2.8145947456359863,
      "step": 4348
    },
    {
      "epoch": 0.598529131857573,
      "grad_norm": 0.18826107680797577,
      "learning_rate": 0.002,
      "loss": 2.8145158290863037,
      "step": 4350
    },
    {
      "epoch": 0.5988043176653236,
      "grad_norm": 0.17902958393096924,
      "learning_rate": 0.002,
      "loss": 2.796504497528076,
      "step": 4352
    },
    {
      "epoch": 0.5990795034730743,
      "grad_norm": 0.1717875450849533,
      "learning_rate": 0.002,
      "loss": 2.799726963043213,
      "step": 4354
    },
    {
      "epoch": 0.5993546892808248,
      "grad_norm": 0.0382474921643734,
      "learning_rate": 0.002,
      "loss": 2.797696828842163,
      "step": 4356
    },
    {
      "epoch": 0.5996298750885755,
      "grad_norm": 0.2006983458995819,
      "learning_rate": 0.002,
      "loss": 2.7692744731903076,
      "step": 4358
    },
    {
      "epoch": 0.599905060896326,
      "grad_norm": 0.03342743590474129,
      "learning_rate": 0.002,
      "loss": 2.8421738147735596,
      "step": 4360
    },
    {
      "epoch": 0.6001802467040767,
      "grad_norm": 0.08266236633062363,
      "learning_rate": 0.002,
      "loss": 2.826603889465332,
      "step": 4362
    },
    {
      "epoch": 0.6004554325118272,
      "grad_norm": 0.2779766619205475,
      "learning_rate": 0.002,
      "loss": 2.8050739765167236,
      "step": 4364
    },
    {
      "epoch": 0.6007306183195779,
      "grad_norm": 0.04702824726700783,
      "learning_rate": 0.002,
      "loss": 2.774299383163452,
      "step": 4366
    },
    {
      "epoch": 0.6010058041273285,
      "grad_norm": 0.27914318442344666,
      "learning_rate": 0.002,
      "loss": 2.8398847579956055,
      "step": 4368
    },
    {
      "epoch": 0.6012809899350791,
      "grad_norm": 0.18428054451942444,
      "learning_rate": 0.002,
      "loss": 2.788926124572754,
      "step": 4370
    },
    {
      "epoch": 0.6015561757428297,
      "grad_norm": 0.20618097484111786,
      "learning_rate": 0.002,
      "loss": 2.8681514263153076,
      "step": 4372
    },
    {
      "epoch": 0.6018313615505803,
      "grad_norm": 0.14500246942043304,
      "learning_rate": 0.002,
      "loss": 2.7726950645446777,
      "step": 4374
    },
    {
      "epoch": 0.6021065473583309,
      "grad_norm": 0.10397118330001831,
      "learning_rate": 0.002,
      "loss": 2.8043339252471924,
      "step": 4376
    },
    {
      "epoch": 0.6023817331660816,
      "grad_norm": 0.053976692259311676,
      "learning_rate": 0.002,
      "loss": 2.768726110458374,
      "step": 4378
    },
    {
      "epoch": 0.6026569189738321,
      "grad_norm": 0.04765314608812332,
      "learning_rate": 0.002,
      "loss": 2.7953944206237793,
      "step": 4380
    },
    {
      "epoch": 0.6029321047815828,
      "grad_norm": 0.11403712630271912,
      "learning_rate": 0.002,
      "loss": 2.8083925247192383,
      "step": 4382
    },
    {
      "epoch": 0.6032072905893333,
      "grad_norm": 0.09890139102935791,
      "learning_rate": 0.002,
      "loss": 2.79784893989563,
      "step": 4384
    },
    {
      "epoch": 0.603482476397084,
      "grad_norm": 0.06135477125644684,
      "learning_rate": 0.002,
      "loss": 2.824429750442505,
      "step": 4386
    },
    {
      "epoch": 0.6037576622048345,
      "grad_norm": 0.04368108883500099,
      "learning_rate": 0.002,
      "loss": 2.833831548690796,
      "step": 4388
    },
    {
      "epoch": 0.6040328480125852,
      "grad_norm": 0.19889599084854126,
      "learning_rate": 0.002,
      "loss": 2.799112319946289,
      "step": 4390
    },
    {
      "epoch": 0.6043080338203358,
      "grad_norm": 0.2551203966140747,
      "learning_rate": 0.002,
      "loss": 2.767484188079834,
      "step": 4392
    },
    {
      "epoch": 0.6045832196280864,
      "grad_norm": 0.08222612738609314,
      "learning_rate": 0.002,
      "loss": 2.7814388275146484,
      "step": 4394
    },
    {
      "epoch": 0.604858405435837,
      "grad_norm": 0.11451450735330582,
      "learning_rate": 0.002,
      "loss": 2.7894506454467773,
      "step": 4396
    },
    {
      "epoch": 0.6051335912435876,
      "grad_norm": 0.16889537870883942,
      "learning_rate": 0.002,
      "loss": 2.808133840560913,
      "step": 4398
    },
    {
      "epoch": 0.6054087770513382,
      "grad_norm": 0.11839750409126282,
      "learning_rate": 0.002,
      "loss": 2.7820568084716797,
      "step": 4400
    },
    {
      "epoch": 0.6056839628590888,
      "grad_norm": 0.14414258301258087,
      "learning_rate": 0.002,
      "loss": 2.823925018310547,
      "step": 4402
    },
    {
      "epoch": 0.6059591486668394,
      "grad_norm": 0.14550930261611938,
      "learning_rate": 0.002,
      "loss": 2.832911252975464,
      "step": 4404
    },
    {
      "epoch": 0.6062343344745901,
      "grad_norm": 0.0426371730864048,
      "learning_rate": 0.002,
      "loss": 2.8028008937835693,
      "step": 4406
    },
    {
      "epoch": 0.6065095202823406,
      "grad_norm": 0.12993325293064117,
      "learning_rate": 0.002,
      "loss": 2.8061795234680176,
      "step": 4408
    },
    {
      "epoch": 0.6067847060900913,
      "grad_norm": 0.02822275273501873,
      "learning_rate": 0.002,
      "loss": 2.7862558364868164,
      "step": 4410
    },
    {
      "epoch": 0.6070598918978418,
      "grad_norm": 0.09823548793792725,
      "learning_rate": 0.002,
      "loss": 2.8047499656677246,
      "step": 4412
    },
    {
      "epoch": 0.6073350777055925,
      "grad_norm": 0.1026366725564003,
      "learning_rate": 0.002,
      "loss": 2.826751232147217,
      "step": 4414
    },
    {
      "epoch": 0.607610263513343,
      "grad_norm": 0.09753008186817169,
      "learning_rate": 0.002,
      "loss": 2.755133628845215,
      "step": 4416
    },
    {
      "epoch": 0.6078854493210937,
      "grad_norm": 0.0641813799738884,
      "learning_rate": 0.002,
      "loss": 2.7916674613952637,
      "step": 4418
    },
    {
      "epoch": 0.6081606351288443,
      "grad_norm": 0.28798967599868774,
      "learning_rate": 0.002,
      "loss": 2.808137893676758,
      "step": 4420
    },
    {
      "epoch": 0.6084358209365949,
      "grad_norm": 0.3802366256713867,
      "learning_rate": 0.002,
      "loss": 2.770019292831421,
      "step": 4422
    },
    {
      "epoch": 0.6087110067443455,
      "grad_norm": 0.2342732697725296,
      "learning_rate": 0.002,
      "loss": 2.8128294944763184,
      "step": 4424
    },
    {
      "epoch": 0.6089861925520961,
      "grad_norm": 0.13003018498420715,
      "learning_rate": 0.002,
      "loss": 2.7771201133728027,
      "step": 4426
    },
    {
      "epoch": 0.6092613783598467,
      "grad_norm": 0.34147384762763977,
      "learning_rate": 0.002,
      "loss": 2.7660889625549316,
      "step": 4428
    },
    {
      "epoch": 0.6095365641675974,
      "grad_norm": 0.17244654893875122,
      "learning_rate": 0.002,
      "loss": 2.7982406616210938,
      "step": 4430
    },
    {
      "epoch": 0.6098117499753479,
      "grad_norm": 0.05124259367585182,
      "learning_rate": 0.002,
      "loss": 2.7892885208129883,
      "step": 4432
    },
    {
      "epoch": 0.6100869357830986,
      "grad_norm": 0.23387812077999115,
      "learning_rate": 0.002,
      "loss": 2.7715108394622803,
      "step": 4434
    },
    {
      "epoch": 0.6103621215908491,
      "grad_norm": 0.14098867774009705,
      "learning_rate": 0.002,
      "loss": 2.7884340286254883,
      "step": 4436
    },
    {
      "epoch": 0.6106373073985998,
      "grad_norm": 0.04456488788127899,
      "learning_rate": 0.002,
      "loss": 2.7813472747802734,
      "step": 4438
    },
    {
      "epoch": 0.6109124932063503,
      "grad_norm": 0.24660301208496094,
      "learning_rate": 0.002,
      "loss": 2.7871389389038086,
      "step": 4440
    },
    {
      "epoch": 0.611187679014101,
      "grad_norm": 0.11902879178524017,
      "learning_rate": 0.002,
      "loss": 2.7697532176971436,
      "step": 4442
    },
    {
      "epoch": 0.6114628648218516,
      "grad_norm": 0.03668056055903435,
      "learning_rate": 0.002,
      "loss": 2.7672181129455566,
      "step": 4444
    },
    {
      "epoch": 0.6117380506296022,
      "grad_norm": 0.1646283119916916,
      "learning_rate": 0.002,
      "loss": 2.763734817504883,
      "step": 4446
    },
    {
      "epoch": 0.6120132364373528,
      "grad_norm": 0.10654648393392563,
      "learning_rate": 0.002,
      "loss": 2.7770118713378906,
      "step": 4448
    },
    {
      "epoch": 0.6122884222451034,
      "grad_norm": 0.08314383029937744,
      "learning_rate": 0.002,
      "loss": 2.7753372192382812,
      "step": 4450
    },
    {
      "epoch": 0.612563608052854,
      "grad_norm": 0.19155454635620117,
      "learning_rate": 0.002,
      "loss": 2.822366237640381,
      "step": 4452
    },
    {
      "epoch": 0.6128387938606046,
      "grad_norm": 0.05785002559423447,
      "learning_rate": 0.002,
      "loss": 2.81850004196167,
      "step": 4454
    },
    {
      "epoch": 0.6131139796683552,
      "grad_norm": 0.057717498391866684,
      "learning_rate": 0.002,
      "loss": 2.8248612880706787,
      "step": 4456
    },
    {
      "epoch": 0.6133891654761059,
      "grad_norm": 0.15112553536891937,
      "learning_rate": 0.002,
      "loss": 2.7821574211120605,
      "step": 4458
    },
    {
      "epoch": 0.6136643512838564,
      "grad_norm": 0.0561191700398922,
      "learning_rate": 0.002,
      "loss": 2.800067663192749,
      "step": 4460
    },
    {
      "epoch": 0.6139395370916071,
      "grad_norm": 0.04588048905134201,
      "learning_rate": 0.002,
      "loss": 2.7463746070861816,
      "step": 4462
    },
    {
      "epoch": 0.6142147228993576,
      "grad_norm": 0.08456812053918839,
      "learning_rate": 0.002,
      "loss": 2.7711944580078125,
      "step": 4464
    },
    {
      "epoch": 0.6144899087071083,
      "grad_norm": 0.17677924036979675,
      "learning_rate": 0.002,
      "loss": 2.772325038909912,
      "step": 4466
    },
    {
      "epoch": 0.6147650945148588,
      "grad_norm": 0.03393809124827385,
      "learning_rate": 0.002,
      "loss": 2.8215532302856445,
      "step": 4468
    },
    {
      "epoch": 0.6150402803226095,
      "grad_norm": 0.02906888909637928,
      "learning_rate": 0.002,
      "loss": 2.8064637184143066,
      "step": 4470
    },
    {
      "epoch": 0.6153154661303601,
      "grad_norm": 0.10922855883836746,
      "learning_rate": 0.002,
      "loss": 2.8238253593444824,
      "step": 4472
    },
    {
      "epoch": 0.6155906519381107,
      "grad_norm": 0.18692541122436523,
      "learning_rate": 0.002,
      "loss": 2.815256118774414,
      "step": 4474
    },
    {
      "epoch": 0.6158658377458613,
      "grad_norm": 0.1807418167591095,
      "learning_rate": 0.002,
      "loss": 2.7664568424224854,
      "step": 4476
    },
    {
      "epoch": 0.6161410235536119,
      "grad_norm": 0.1003740131855011,
      "learning_rate": 0.002,
      "loss": 2.7975988388061523,
      "step": 4478
    },
    {
      "epoch": 0.6164162093613625,
      "grad_norm": 0.20083856582641602,
      "learning_rate": 0.002,
      "loss": 2.7565011978149414,
      "step": 4480
    },
    {
      "epoch": 0.6166913951691132,
      "grad_norm": 0.09010782837867737,
      "learning_rate": 0.002,
      "loss": 2.7929630279541016,
      "step": 4482
    },
    {
      "epoch": 0.6169665809768637,
      "grad_norm": 0.05212217569351196,
      "learning_rate": 0.002,
      "loss": 2.7953996658325195,
      "step": 4484
    },
    {
      "epoch": 0.6172417667846144,
      "grad_norm": 0.155099019408226,
      "learning_rate": 0.002,
      "loss": 2.8100833892822266,
      "step": 4486
    },
    {
      "epoch": 0.6175169525923649,
      "grad_norm": 0.26840680837631226,
      "learning_rate": 0.002,
      "loss": 2.788043975830078,
      "step": 4488
    },
    {
      "epoch": 0.6177921384001156,
      "grad_norm": 0.10257375985383987,
      "learning_rate": 0.002,
      "loss": 2.802314281463623,
      "step": 4490
    },
    {
      "epoch": 0.6180673242078661,
      "grad_norm": 0.25729188323020935,
      "learning_rate": 0.002,
      "loss": 2.8465914726257324,
      "step": 4492
    },
    {
      "epoch": 0.6183425100156168,
      "grad_norm": 0.11800071597099304,
      "learning_rate": 0.002,
      "loss": 2.833535671234131,
      "step": 4494
    },
    {
      "epoch": 0.6186176958233675,
      "grad_norm": 0.22314254939556122,
      "learning_rate": 0.002,
      "loss": 2.750213861465454,
      "step": 4496
    },
    {
      "epoch": 0.618892881631118,
      "grad_norm": 0.11670488119125366,
      "learning_rate": 0.002,
      "loss": 2.794198989868164,
      "step": 4498
    },
    {
      "epoch": 0.6191680674388687,
      "grad_norm": 0.21482525765895844,
      "learning_rate": 0.002,
      "loss": 2.7729721069335938,
      "step": 4500
    },
    {
      "epoch": 0.6194432532466192,
      "grad_norm": 0.14823825657367706,
      "learning_rate": 0.002,
      "loss": 2.840712308883667,
      "step": 4502
    },
    {
      "epoch": 0.6197184390543699,
      "grad_norm": 0.2119729071855545,
      "learning_rate": 0.002,
      "loss": 2.765592575073242,
      "step": 4504
    },
    {
      "epoch": 0.6199936248621204,
      "grad_norm": 0.15675444900989532,
      "learning_rate": 0.002,
      "loss": 2.7862870693206787,
      "step": 4506
    },
    {
      "epoch": 0.620268810669871,
      "grad_norm": 0.10691428929567337,
      "learning_rate": 0.002,
      "loss": 2.7994022369384766,
      "step": 4508
    },
    {
      "epoch": 0.6205439964776217,
      "grad_norm": 0.1523008495569229,
      "learning_rate": 0.002,
      "loss": 2.797682285308838,
      "step": 4510
    },
    {
      "epoch": 0.6208191822853723,
      "grad_norm": 0.10924112051725388,
      "learning_rate": 0.002,
      "loss": 2.833784580230713,
      "step": 4512
    },
    {
      "epoch": 0.6210943680931229,
      "grad_norm": 0.07924924045801163,
      "learning_rate": 0.002,
      "loss": 2.7797398567199707,
      "step": 4514
    },
    {
      "epoch": 0.6213695539008735,
      "grad_norm": 0.15442657470703125,
      "learning_rate": 0.002,
      "loss": 2.7833774089813232,
      "step": 4516
    },
    {
      "epoch": 0.6216447397086241,
      "grad_norm": 0.17616771161556244,
      "learning_rate": 0.002,
      "loss": 2.810415267944336,
      "step": 4518
    },
    {
      "epoch": 0.6219199255163746,
      "grad_norm": 0.051903922110795975,
      "learning_rate": 0.002,
      "loss": 2.779097080230713,
      "step": 4520
    },
    {
      "epoch": 0.6221951113241253,
      "grad_norm": 0.17587412893772125,
      "learning_rate": 0.002,
      "loss": 2.8091044425964355,
      "step": 4522
    },
    {
      "epoch": 0.622470297131876,
      "grad_norm": 0.07675572484731674,
      "learning_rate": 0.002,
      "loss": 2.7753872871398926,
      "step": 4524
    },
    {
      "epoch": 0.6227454829396265,
      "grad_norm": 0.3027505874633789,
      "learning_rate": 0.002,
      "loss": 2.797774314880371,
      "step": 4526
    },
    {
      "epoch": 0.6230206687473772,
      "grad_norm": 0.12339334189891815,
      "learning_rate": 0.002,
      "loss": 2.770112991333008,
      "step": 4528
    },
    {
      "epoch": 0.6232958545551277,
      "grad_norm": 0.2437736690044403,
      "learning_rate": 0.002,
      "loss": 2.7809762954711914,
      "step": 4530
    },
    {
      "epoch": 0.6235710403628784,
      "grad_norm": 0.167412668466568,
      "learning_rate": 0.002,
      "loss": 2.80009126663208,
      "step": 4532
    },
    {
      "epoch": 0.623846226170629,
      "grad_norm": 0.09293320029973984,
      "learning_rate": 0.002,
      "loss": 2.794508457183838,
      "step": 4534
    },
    {
      "epoch": 0.6241214119783796,
      "grad_norm": 0.10628589242696762,
      "learning_rate": 0.002,
      "loss": 2.7925589084625244,
      "step": 4536
    },
    {
      "epoch": 0.6243965977861302,
      "grad_norm": 0.05143354460597038,
      "learning_rate": 0.002,
      "loss": 2.773508071899414,
      "step": 4538
    },
    {
      "epoch": 0.6246717835938808,
      "grad_norm": 0.03253292664885521,
      "learning_rate": 0.002,
      "loss": 2.8028500080108643,
      "step": 4540
    },
    {
      "epoch": 0.6249469694016314,
      "grad_norm": 0.05254313349723816,
      "learning_rate": 0.002,
      "loss": 2.79512357711792,
      "step": 4542
    },
    {
      "epoch": 0.625222155209382,
      "grad_norm": 0.12480263411998749,
      "learning_rate": 0.002,
      "loss": 2.764289379119873,
      "step": 4544
    },
    {
      "epoch": 0.6254973410171326,
      "grad_norm": 0.1218949407339096,
      "learning_rate": 0.002,
      "loss": 2.7858991622924805,
      "step": 4546
    },
    {
      "epoch": 0.6257725268248833,
      "grad_norm": 0.04429067671298981,
      "learning_rate": 0.002,
      "loss": 2.7918286323547363,
      "step": 4548
    },
    {
      "epoch": 0.6260477126326338,
      "grad_norm": 0.15019403398036957,
      "learning_rate": 0.002,
      "loss": 2.808915615081787,
      "step": 4550
    },
    {
      "epoch": 0.6263228984403845,
      "grad_norm": 0.20115694403648376,
      "learning_rate": 0.002,
      "loss": 2.796635627746582,
      "step": 4552
    },
    {
      "epoch": 0.626598084248135,
      "grad_norm": 0.038474712520837784,
      "learning_rate": 0.002,
      "loss": 2.7867789268493652,
      "step": 4554
    },
    {
      "epoch": 0.6268732700558857,
      "grad_norm": 0.04549677297472954,
      "learning_rate": 0.002,
      "loss": 2.7609877586364746,
      "step": 4556
    },
    {
      "epoch": 0.6271484558636362,
      "grad_norm": 0.04143692925572395,
      "learning_rate": 0.002,
      "loss": 2.7960262298583984,
      "step": 4558
    },
    {
      "epoch": 0.6274236416713869,
      "grad_norm": 0.037147536873817444,
      "learning_rate": 0.002,
      "loss": 2.7675180435180664,
      "step": 4560
    },
    {
      "epoch": 0.6276988274791375,
      "grad_norm": 0.18653830885887146,
      "learning_rate": 0.002,
      "loss": 2.787292718887329,
      "step": 4562
    },
    {
      "epoch": 0.6279740132868881,
      "grad_norm": 0.24812553822994232,
      "learning_rate": 0.002,
      "loss": 2.7921719551086426,
      "step": 4564
    },
    {
      "epoch": 0.6282491990946387,
      "grad_norm": 0.146382138133049,
      "learning_rate": 0.002,
      "loss": 2.7917428016662598,
      "step": 4566
    },
    {
      "epoch": 0.6285243849023893,
      "grad_norm": 0.23463280498981476,
      "learning_rate": 0.002,
      "loss": 2.788790225982666,
      "step": 4568
    },
    {
      "epoch": 0.6287995707101399,
      "grad_norm": 0.2305314987897873,
      "learning_rate": 0.002,
      "loss": 2.7360551357269287,
      "step": 4570
    },
    {
      "epoch": 0.6290747565178905,
      "grad_norm": 0.1717698574066162,
      "learning_rate": 0.002,
      "loss": 2.7400565147399902,
      "step": 4572
    },
    {
      "epoch": 0.6293499423256411,
      "grad_norm": 0.2554915249347687,
      "learning_rate": 0.002,
      "loss": 2.803736448287964,
      "step": 4574
    },
    {
      "epoch": 0.6296251281333918,
      "grad_norm": 0.19080203771591187,
      "learning_rate": 0.002,
      "loss": 2.739419460296631,
      "step": 4576
    },
    {
      "epoch": 0.6299003139411423,
      "grad_norm": 0.05915770307183266,
      "learning_rate": 0.002,
      "loss": 2.7544617652893066,
      "step": 4578
    },
    {
      "epoch": 0.630175499748893,
      "grad_norm": 0.24102634191513062,
      "learning_rate": 0.002,
      "loss": 2.812976360321045,
      "step": 4580
    },
    {
      "epoch": 0.6304506855566435,
      "grad_norm": 0.04205991327762604,
      "learning_rate": 0.002,
      "loss": 2.7983150482177734,
      "step": 4582
    },
    {
      "epoch": 0.6307258713643942,
      "grad_norm": 0.14791329205036163,
      "learning_rate": 0.002,
      "loss": 2.7942657470703125,
      "step": 4584
    },
    {
      "epoch": 0.6310010571721448,
      "grad_norm": 0.03345121815800667,
      "learning_rate": 0.002,
      "loss": 2.7450215816497803,
      "step": 4586
    },
    {
      "epoch": 0.6312762429798954,
      "grad_norm": 0.08230923861265182,
      "learning_rate": 0.002,
      "loss": 2.796541690826416,
      "step": 4588
    },
    {
      "epoch": 0.631551428787646,
      "grad_norm": 0.08553677797317505,
      "learning_rate": 0.002,
      "loss": 2.765592575073242,
      "step": 4590
    },
    {
      "epoch": 0.6318266145953966,
      "grad_norm": 0.11855030059814453,
      "learning_rate": 0.002,
      "loss": 2.759077548980713,
      "step": 4592
    },
    {
      "epoch": 0.6321018004031472,
      "grad_norm": 0.03765087574720383,
      "learning_rate": 0.002,
      "loss": 2.7932169437408447,
      "step": 4594
    },
    {
      "epoch": 0.6323769862108978,
      "grad_norm": 0.06253018230199814,
      "learning_rate": 0.002,
      "loss": 2.820244073867798,
      "step": 4596
    },
    {
      "epoch": 0.6326521720186484,
      "grad_norm": 0.07626193761825562,
      "learning_rate": 0.002,
      "loss": 2.800222396850586,
      "step": 4598
    },
    {
      "epoch": 0.6329273578263991,
      "grad_norm": 0.06358031928539276,
      "learning_rate": 0.002,
      "loss": 2.74550199508667,
      "step": 4600
    },
    {
      "epoch": 0.6332025436341496,
      "grad_norm": 0.088703453540802,
      "learning_rate": 0.002,
      "loss": 2.767183542251587,
      "step": 4602
    },
    {
      "epoch": 0.6334777294419003,
      "grad_norm": 0.22587761282920837,
      "learning_rate": 0.002,
      "loss": 2.783795118331909,
      "step": 4604
    },
    {
      "epoch": 0.6337529152496508,
      "grad_norm": 0.3209495544433594,
      "learning_rate": 0.002,
      "loss": 2.8025779724121094,
      "step": 4606
    },
    {
      "epoch": 0.6340281010574015,
      "grad_norm": 0.05320362374186516,
      "learning_rate": 0.002,
      "loss": 2.8060317039489746,
      "step": 4608
    },
    {
      "epoch": 0.634303286865152,
      "grad_norm": 0.24387192726135254,
      "learning_rate": 0.002,
      "loss": 2.7964744567871094,
      "step": 4610
    },
    {
      "epoch": 0.6345784726729027,
      "grad_norm": 0.03901360183954239,
      "learning_rate": 0.002,
      "loss": 2.765023708343506,
      "step": 4612
    },
    {
      "epoch": 0.6348536584806533,
      "grad_norm": 0.20302768051624298,
      "learning_rate": 0.002,
      "loss": 2.804135322570801,
      "step": 4614
    },
    {
      "epoch": 0.6351288442884039,
      "grad_norm": 0.06281812489032745,
      "learning_rate": 0.002,
      "loss": 2.822126865386963,
      "step": 4616
    },
    {
      "epoch": 0.6354040300961545,
      "grad_norm": 0.09621944278478622,
      "learning_rate": 0.002,
      "loss": 2.8281028270721436,
      "step": 4618
    },
    {
      "epoch": 0.6356792159039051,
      "grad_norm": 0.13215160369873047,
      "learning_rate": 0.002,
      "loss": 2.761394500732422,
      "step": 4620
    },
    {
      "epoch": 0.6359544017116557,
      "grad_norm": 0.03548162057995796,
      "learning_rate": 0.002,
      "loss": 2.829702615737915,
      "step": 4622
    },
    {
      "epoch": 0.6362295875194063,
      "grad_norm": 0.10813958942890167,
      "learning_rate": 0.002,
      "loss": 2.777963161468506,
      "step": 4624
    },
    {
      "epoch": 0.6365047733271569,
      "grad_norm": 0.031656138598918915,
      "learning_rate": 0.002,
      "loss": 2.8019752502441406,
      "step": 4626
    },
    {
      "epoch": 0.6367799591349076,
      "grad_norm": 0.0562039315700531,
      "learning_rate": 0.002,
      "loss": 2.781214475631714,
      "step": 4628
    },
    {
      "epoch": 0.6370551449426581,
      "grad_norm": 0.032474469393491745,
      "learning_rate": 0.002,
      "loss": 2.783313274383545,
      "step": 4630
    },
    {
      "epoch": 0.6373303307504088,
      "grad_norm": 0.12802830338478088,
      "learning_rate": 0.002,
      "loss": 2.7871618270874023,
      "step": 4632
    },
    {
      "epoch": 0.6376055165581593,
      "grad_norm": 0.2777152359485626,
      "learning_rate": 0.002,
      "loss": 2.816922187805176,
      "step": 4634
    },
    {
      "epoch": 0.63788070236591,
      "grad_norm": 0.28132811188697815,
      "learning_rate": 0.002,
      "loss": 2.7671754360198975,
      "step": 4636
    },
    {
      "epoch": 0.6381558881736606,
      "grad_norm": 0.07926633208990097,
      "learning_rate": 0.002,
      "loss": 2.7602734565734863,
      "step": 4638
    },
    {
      "epoch": 0.6384310739814112,
      "grad_norm": 0.17038145661354065,
      "learning_rate": 0.002,
      "loss": 2.805208444595337,
      "step": 4640
    },
    {
      "epoch": 0.6387062597891618,
      "grad_norm": 0.09717871993780136,
      "learning_rate": 0.002,
      "loss": 2.7855052947998047,
      "step": 4642
    },
    {
      "epoch": 0.6389814455969124,
      "grad_norm": 0.19458778202533722,
      "learning_rate": 0.002,
      "loss": 2.7919116020202637,
      "step": 4644
    },
    {
      "epoch": 0.639256631404663,
      "grad_norm": 0.09778326004743576,
      "learning_rate": 0.002,
      "loss": 2.780381441116333,
      "step": 4646
    },
    {
      "epoch": 0.6395318172124136,
      "grad_norm": 0.16929592192173004,
      "learning_rate": 0.002,
      "loss": 2.7688121795654297,
      "step": 4648
    },
    {
      "epoch": 0.6398070030201642,
      "grad_norm": 0.0390695258975029,
      "learning_rate": 0.002,
      "loss": 2.770759105682373,
      "step": 4650
    },
    {
      "epoch": 0.6400821888279149,
      "grad_norm": 0.2107284814119339,
      "learning_rate": 0.002,
      "loss": 2.766496181488037,
      "step": 4652
    },
    {
      "epoch": 0.6403573746356654,
      "grad_norm": 0.10515637695789337,
      "learning_rate": 0.002,
      "loss": 2.7950708866119385,
      "step": 4654
    },
    {
      "epoch": 0.6406325604434161,
      "grad_norm": 0.12834572792053223,
      "learning_rate": 0.002,
      "loss": 2.7429473400115967,
      "step": 4656
    },
    {
      "epoch": 0.6409077462511666,
      "grad_norm": 0.15969769656658173,
      "learning_rate": 0.002,
      "loss": 2.74763822555542,
      "step": 4658
    },
    {
      "epoch": 0.6411829320589173,
      "grad_norm": 0.12188316881656647,
      "learning_rate": 0.002,
      "loss": 2.8142917156219482,
      "step": 4660
    },
    {
      "epoch": 0.6414581178666678,
      "grad_norm": 0.28544580936431885,
      "learning_rate": 0.002,
      "loss": 2.806434392929077,
      "step": 4662
    },
    {
      "epoch": 0.6417333036744185,
      "grad_norm": 0.17773008346557617,
      "learning_rate": 0.002,
      "loss": 2.7826738357543945,
      "step": 4664
    },
    {
      "epoch": 0.6420084894821692,
      "grad_norm": 0.16734449565410614,
      "learning_rate": 0.002,
      "loss": 2.7683777809143066,
      "step": 4666
    },
    {
      "epoch": 0.6422836752899197,
      "grad_norm": 0.18399451673030853,
      "learning_rate": 0.002,
      "loss": 2.792196750640869,
      "step": 4668
    },
    {
      "epoch": 0.6425588610976704,
      "grad_norm": 0.10733193904161453,
      "learning_rate": 0.002,
      "loss": 2.780931234359741,
      "step": 4670
    },
    {
      "epoch": 0.6428340469054209,
      "grad_norm": 0.13659271597862244,
      "learning_rate": 0.002,
      "loss": 2.7958731651306152,
      "step": 4672
    },
    {
      "epoch": 0.6431092327131716,
      "grad_norm": 0.1094350665807724,
      "learning_rate": 0.002,
      "loss": 2.75169038772583,
      "step": 4674
    },
    {
      "epoch": 0.6433844185209221,
      "grad_norm": 0.10221274197101593,
      "learning_rate": 0.002,
      "loss": 2.795790195465088,
      "step": 4676
    },
    {
      "epoch": 0.6436596043286728,
      "grad_norm": 0.0794576033949852,
      "learning_rate": 0.002,
      "loss": 2.8189611434936523,
      "step": 4678
    },
    {
      "epoch": 0.6439347901364234,
      "grad_norm": 0.06349688768386841,
      "learning_rate": 0.002,
      "loss": 2.7532525062561035,
      "step": 4680
    },
    {
      "epoch": 0.644209975944174,
      "grad_norm": 0.15700069069862366,
      "learning_rate": 0.002,
      "loss": 2.739464521408081,
      "step": 4682
    },
    {
      "epoch": 0.6444851617519246,
      "grad_norm": 0.12475479394197464,
      "learning_rate": 0.002,
      "loss": 2.7952451705932617,
      "step": 4684
    },
    {
      "epoch": 0.6447603475596752,
      "grad_norm": 0.060632362961769104,
      "learning_rate": 0.002,
      "loss": 2.8051462173461914,
      "step": 4686
    },
    {
      "epoch": 0.6450355333674258,
      "grad_norm": 0.15978257358074188,
      "learning_rate": 0.002,
      "loss": 2.7954492568969727,
      "step": 4688
    },
    {
      "epoch": 0.6453107191751764,
      "grad_norm": 0.1511439085006714,
      "learning_rate": 0.002,
      "loss": 2.7521989345550537,
      "step": 4690
    },
    {
      "epoch": 0.645585904982927,
      "grad_norm": 0.03229740262031555,
      "learning_rate": 0.002,
      "loss": 2.7486205101013184,
      "step": 4692
    },
    {
      "epoch": 0.6458610907906777,
      "grad_norm": 0.1393975019454956,
      "learning_rate": 0.002,
      "loss": 2.7724363803863525,
      "step": 4694
    },
    {
      "epoch": 0.6461362765984282,
      "grad_norm": 0.1982836276292801,
      "learning_rate": 0.002,
      "loss": 2.7308006286621094,
      "step": 4696
    },
    {
      "epoch": 0.6464114624061789,
      "grad_norm": 0.10918018221855164,
      "learning_rate": 0.002,
      "loss": 2.7433321475982666,
      "step": 4698
    },
    {
      "epoch": 0.6466866482139294,
      "grad_norm": 0.03244606778025627,
      "learning_rate": 0.002,
      "loss": 2.780330181121826,
      "step": 4700
    },
    {
      "epoch": 0.6469618340216801,
      "grad_norm": 0.19268669188022614,
      "learning_rate": 0.002,
      "loss": 2.819324254989624,
      "step": 4702
    },
    {
      "epoch": 0.6472370198294307,
      "grad_norm": 0.09058530628681183,
      "learning_rate": 0.002,
      "loss": 2.756251096725464,
      "step": 4704
    },
    {
      "epoch": 0.6475122056371813,
      "grad_norm": 0.19886817038059235,
      "learning_rate": 0.002,
      "loss": 2.7404091358184814,
      "step": 4706
    },
    {
      "epoch": 0.6477873914449319,
      "grad_norm": 0.23764492571353912,
      "learning_rate": 0.002,
      "loss": 2.7574472427368164,
      "step": 4708
    },
    {
      "epoch": 0.6480625772526825,
      "grad_norm": 0.05093980208039284,
      "learning_rate": 0.002,
      "loss": 2.754936695098877,
      "step": 4710
    },
    {
      "epoch": 0.6483377630604331,
      "grad_norm": 0.10006202757358551,
      "learning_rate": 0.002,
      "loss": 2.7626953125,
      "step": 4712
    },
    {
      "epoch": 0.6486129488681837,
      "grad_norm": 0.18028271198272705,
      "learning_rate": 0.002,
      "loss": 2.813612461090088,
      "step": 4714
    },
    {
      "epoch": 0.6488881346759343,
      "grad_norm": 0.037930119782686234,
      "learning_rate": 0.002,
      "loss": 2.7709412574768066,
      "step": 4716
    },
    {
      "epoch": 0.649163320483685,
      "grad_norm": 0.19873249530792236,
      "learning_rate": 0.002,
      "loss": 2.7579030990600586,
      "step": 4718
    },
    {
      "epoch": 0.6494385062914355,
      "grad_norm": 0.1793244630098343,
      "learning_rate": 0.002,
      "loss": 2.775012969970703,
      "step": 4720
    },
    {
      "epoch": 0.6497136920991862,
      "grad_norm": 0.11467408388853073,
      "learning_rate": 0.002,
      "loss": 2.7858643531799316,
      "step": 4722
    },
    {
      "epoch": 0.6499888779069367,
      "grad_norm": 0.2838207483291626,
      "learning_rate": 0.002,
      "loss": 2.77858829498291,
      "step": 4724
    },
    {
      "epoch": 0.6502640637146874,
      "grad_norm": 0.09016822278499603,
      "learning_rate": 0.002,
      "loss": 2.7793688774108887,
      "step": 4726
    },
    {
      "epoch": 0.6505392495224379,
      "grad_norm": 0.30260828137397766,
      "learning_rate": 0.002,
      "loss": 2.7414190769195557,
      "step": 4728
    },
    {
      "epoch": 0.6508144353301886,
      "grad_norm": 0.03622308745980263,
      "learning_rate": 0.002,
      "loss": 2.764153480529785,
      "step": 4730
    },
    {
      "epoch": 0.6510896211379392,
      "grad_norm": 0.2565052807331085,
      "learning_rate": 0.002,
      "loss": 2.797391414642334,
      "step": 4732
    },
    {
      "epoch": 0.6513648069456898,
      "grad_norm": 0.08207742124795914,
      "learning_rate": 0.002,
      "loss": 2.75695538520813,
      "step": 4734
    },
    {
      "epoch": 0.6516399927534404,
      "grad_norm": 0.09151878952980042,
      "learning_rate": 0.002,
      "loss": 2.785304546356201,
      "step": 4736
    },
    {
      "epoch": 0.651915178561191,
      "grad_norm": 0.11183864623308182,
      "learning_rate": 0.002,
      "loss": 2.801602840423584,
      "step": 4738
    },
    {
      "epoch": 0.6521903643689416,
      "grad_norm": 0.17482808232307434,
      "learning_rate": 0.002,
      "loss": 2.804063081741333,
      "step": 4740
    },
    {
      "epoch": 0.6524655501766922,
      "grad_norm": 0.041642289608716965,
      "learning_rate": 0.002,
      "loss": 2.784606456756592,
      "step": 4742
    },
    {
      "epoch": 0.6527407359844428,
      "grad_norm": 0.09446289390325546,
      "learning_rate": 0.002,
      "loss": 2.7673394680023193,
      "step": 4744
    },
    {
      "epoch": 0.6530159217921935,
      "grad_norm": 0.11597919464111328,
      "learning_rate": 0.002,
      "loss": 2.8006420135498047,
      "step": 4746
    },
    {
      "epoch": 0.653291107599944,
      "grad_norm": 0.20845229923725128,
      "learning_rate": 0.002,
      "loss": 2.7381491661071777,
      "step": 4748
    },
    {
      "epoch": 0.6535662934076947,
      "grad_norm": 0.13979794085025787,
      "learning_rate": 0.002,
      "loss": 2.7775936126708984,
      "step": 4750
    },
    {
      "epoch": 0.6538414792154452,
      "grad_norm": 0.15387804806232452,
      "learning_rate": 0.002,
      "loss": 2.766619920730591,
      "step": 4752
    },
    {
      "epoch": 0.6541166650231959,
      "grad_norm": 0.18752141296863556,
      "learning_rate": 0.002,
      "loss": 2.7739500999450684,
      "step": 4754
    },
    {
      "epoch": 0.6543918508309465,
      "grad_norm": 0.16656121611595154,
      "learning_rate": 0.002,
      "loss": 2.7256178855895996,
      "step": 4756
    },
    {
      "epoch": 0.6546670366386971,
      "grad_norm": 0.03116757422685623,
      "learning_rate": 0.002,
      "loss": 2.798841953277588,
      "step": 4758
    },
    {
      "epoch": 0.6549422224464477,
      "grad_norm": 0.15299807488918304,
      "learning_rate": 0.002,
      "loss": 2.800633668899536,
      "step": 4760
    },
    {
      "epoch": 0.6552174082541983,
      "grad_norm": 0.18604402244091034,
      "learning_rate": 0.002,
      "loss": 2.795367479324341,
      "step": 4762
    },
    {
      "epoch": 0.6554925940619489,
      "grad_norm": 0.06993712484836578,
      "learning_rate": 0.002,
      "loss": 2.7793021202087402,
      "step": 4764
    },
    {
      "epoch": 0.6557677798696995,
      "grad_norm": 0.0666222795844078,
      "learning_rate": 0.002,
      "loss": 2.7589473724365234,
      "step": 4766
    },
    {
      "epoch": 0.6560429656774501,
      "grad_norm": 0.11489243805408478,
      "learning_rate": 0.002,
      "loss": 2.7228102684020996,
      "step": 4768
    },
    {
      "epoch": 0.6563181514852008,
      "grad_norm": 0.060897063463926315,
      "learning_rate": 0.002,
      "loss": 2.795868396759033,
      "step": 4770
    },
    {
      "epoch": 0.6565933372929513,
      "grad_norm": 0.16138707101345062,
      "learning_rate": 0.002,
      "loss": 2.8429007530212402,
      "step": 4772
    },
    {
      "epoch": 0.656868523100702,
      "grad_norm": 0.19939859211444855,
      "learning_rate": 0.002,
      "loss": 2.7920730113983154,
      "step": 4774
    },
    {
      "epoch": 0.6571437089084525,
      "grad_norm": 0.09286274015903473,
      "learning_rate": 0.002,
      "loss": 2.7474801540374756,
      "step": 4776
    },
    {
      "epoch": 0.6574188947162032,
      "grad_norm": 0.1861429661512375,
      "learning_rate": 0.002,
      "loss": 2.801693916320801,
      "step": 4778
    },
    {
      "epoch": 0.6576940805239537,
      "grad_norm": 0.10972597450017929,
      "learning_rate": 0.002,
      "loss": 2.751243829727173,
      "step": 4780
    },
    {
      "epoch": 0.6579692663317044,
      "grad_norm": 0.146956667304039,
      "learning_rate": 0.002,
      "loss": 2.7787535190582275,
      "step": 4782
    },
    {
      "epoch": 0.658244452139455,
      "grad_norm": 0.20821505784988403,
      "learning_rate": 0.002,
      "loss": 2.766542911529541,
      "step": 4784
    },
    {
      "epoch": 0.6585196379472056,
      "grad_norm": 0.044310394674539566,
      "learning_rate": 0.002,
      "loss": 2.7613580226898193,
      "step": 4786
    },
    {
      "epoch": 0.6587948237549562,
      "grad_norm": 0.24552291631698608,
      "learning_rate": 0.002,
      "loss": 2.7489254474639893,
      "step": 4788
    },
    {
      "epoch": 0.6590700095627068,
      "grad_norm": 0.08798548579216003,
      "learning_rate": 0.002,
      "loss": 2.7501158714294434,
      "step": 4790
    },
    {
      "epoch": 0.6593451953704574,
      "grad_norm": 0.41396281123161316,
      "learning_rate": 0.002,
      "loss": 2.7694740295410156,
      "step": 4792
    },
    {
      "epoch": 0.659620381178208,
      "grad_norm": 0.07932556420564651,
      "learning_rate": 0.002,
      "loss": 2.7650318145751953,
      "step": 4794
    },
    {
      "epoch": 0.6598955669859586,
      "grad_norm": 0.27110689878463745,
      "learning_rate": 0.002,
      "loss": 2.7458839416503906,
      "step": 4796
    },
    {
      "epoch": 0.6601707527937093,
      "grad_norm": 0.37526601552963257,
      "learning_rate": 0.002,
      "loss": 2.7147629261016846,
      "step": 4798
    },
    {
      "epoch": 0.6604459386014598,
      "grad_norm": 0.10831064730882645,
      "learning_rate": 0.002,
      "loss": 2.7888269424438477,
      "step": 4800
    },
    {
      "epoch": 0.6607211244092105,
      "grad_norm": 0.10339199006557465,
      "learning_rate": 0.002,
      "loss": 2.8254756927490234,
      "step": 4802
    },
    {
      "epoch": 0.660996310216961,
      "grad_norm": 0.17311997711658478,
      "learning_rate": 0.002,
      "loss": 2.77605938911438,
      "step": 4804
    },
    {
      "epoch": 0.6612714960247117,
      "grad_norm": 0.16257180273532867,
      "learning_rate": 0.002,
      "loss": 2.7822651863098145,
      "step": 4806
    },
    {
      "epoch": 0.6615466818324623,
      "grad_norm": 0.09401389211416245,
      "learning_rate": 0.002,
      "loss": 2.746924877166748,
      "step": 4808
    },
    {
      "epoch": 0.6618218676402129,
      "grad_norm": 0.13996100425720215,
      "learning_rate": 0.002,
      "loss": 2.7621560096740723,
      "step": 4810
    },
    {
      "epoch": 0.6620970534479635,
      "grad_norm": 0.10447783768177032,
      "learning_rate": 0.002,
      "loss": 2.801396608352661,
      "step": 4812
    },
    {
      "epoch": 0.6623722392557141,
      "grad_norm": 0.08732876181602478,
      "learning_rate": 0.002,
      "loss": 2.7875423431396484,
      "step": 4814
    },
    {
      "epoch": 0.6626474250634647,
      "grad_norm": 0.08171604573726654,
      "learning_rate": 0.002,
      "loss": 2.8206067085266113,
      "step": 4816
    },
    {
      "epoch": 0.6629226108712153,
      "grad_norm": 0.05519060045480728,
      "learning_rate": 0.002,
      "loss": 2.766233205795288,
      "step": 4818
    },
    {
      "epoch": 0.663197796678966,
      "grad_norm": 0.03227052465081215,
      "learning_rate": 0.002,
      "loss": 2.734121799468994,
      "step": 4820
    },
    {
      "epoch": 0.6634729824867166,
      "grad_norm": 0.1813543140888214,
      "learning_rate": 0.002,
      "loss": 2.77500057220459,
      "step": 4822
    },
    {
      "epoch": 0.6637481682944671,
      "grad_norm": 0.06399334967136383,
      "learning_rate": 0.002,
      "loss": 2.731595516204834,
      "step": 4824
    },
    {
      "epoch": 0.6640233541022178,
      "grad_norm": 0.28849032521247864,
      "learning_rate": 0.002,
      "loss": 2.732293128967285,
      "step": 4826
    },
    {
      "epoch": 0.6642985399099683,
      "grad_norm": 0.03354024142026901,
      "learning_rate": 0.002,
      "loss": 2.774752616882324,
      "step": 4828
    },
    {
      "epoch": 0.664573725717719,
      "grad_norm": 0.09102088212966919,
      "learning_rate": 0.002,
      "loss": 2.73762845993042,
      "step": 4830
    },
    {
      "epoch": 0.6648489115254695,
      "grad_norm": 0.045940302312374115,
      "learning_rate": 0.002,
      "loss": 2.7538938522338867,
      "step": 4832
    },
    {
      "epoch": 0.6651240973332202,
      "grad_norm": 0.14363162219524384,
      "learning_rate": 0.002,
      "loss": 2.736144781112671,
      "step": 4834
    },
    {
      "epoch": 0.6653992831409709,
      "grad_norm": 0.06739766895771027,
      "learning_rate": 0.002,
      "loss": 2.7190585136413574,
      "step": 4836
    },
    {
      "epoch": 0.6656744689487214,
      "grad_norm": 0.1260792762041092,
      "learning_rate": 0.002,
      "loss": 2.786677122116089,
      "step": 4838
    },
    {
      "epoch": 0.665949654756472,
      "grad_norm": 0.042678654193878174,
      "learning_rate": 0.002,
      "loss": 2.7382588386535645,
      "step": 4840
    },
    {
      "epoch": 0.6662248405642226,
      "grad_norm": 0.048617567867040634,
      "learning_rate": 0.002,
      "loss": 2.7799267768859863,
      "step": 4842
    },
    {
      "epoch": 0.6665000263719733,
      "grad_norm": 0.10351022332906723,
      "learning_rate": 0.002,
      "loss": 2.7402777671813965,
      "step": 4844
    },
    {
      "epoch": 0.6667752121797238,
      "grad_norm": 0.03907134011387825,
      "learning_rate": 0.002,
      "loss": 2.748260974884033,
      "step": 4846
    },
    {
      "epoch": 0.6670503979874745,
      "grad_norm": 0.30435043573379517,
      "learning_rate": 0.002,
      "loss": 2.739015579223633,
      "step": 4848
    },
    {
      "epoch": 0.6673255837952251,
      "grad_norm": 0.2642003297805786,
      "learning_rate": 0.002,
      "loss": 2.76741361618042,
      "step": 4850
    },
    {
      "epoch": 0.6676007696029757,
      "grad_norm": 0.1566535383462906,
      "learning_rate": 0.002,
      "loss": 2.7817583084106445,
      "step": 4852
    },
    {
      "epoch": 0.6678759554107263,
      "grad_norm": 0.3720585107803345,
      "learning_rate": 0.002,
      "loss": 2.7799084186553955,
      "step": 4854
    },
    {
      "epoch": 0.6681511412184769,
      "grad_norm": 0.16646818816661835,
      "learning_rate": 0.002,
      "loss": 2.7882566452026367,
      "step": 4856
    },
    {
      "epoch": 0.6684263270262275,
      "grad_norm": 0.14866362512111664,
      "learning_rate": 0.002,
      "loss": 2.7687737941741943,
      "step": 4858
    },
    {
      "epoch": 0.6687015128339782,
      "grad_norm": 0.30698463320732117,
      "learning_rate": 0.002,
      "loss": 2.8247790336608887,
      "step": 4860
    },
    {
      "epoch": 0.6689766986417287,
      "grad_norm": 0.1297796666622162,
      "learning_rate": 0.002,
      "loss": 2.7469425201416016,
      "step": 4862
    },
    {
      "epoch": 0.6692518844494794,
      "grad_norm": 0.083084836602211,
      "learning_rate": 0.002,
      "loss": 2.7432894706726074,
      "step": 4864
    },
    {
      "epoch": 0.6695270702572299,
      "grad_norm": 0.2130274772644043,
      "learning_rate": 0.002,
      "loss": 2.7803890705108643,
      "step": 4866
    },
    {
      "epoch": 0.6698022560649806,
      "grad_norm": 0.040097348392009735,
      "learning_rate": 0.002,
      "loss": 2.7708468437194824,
      "step": 4868
    },
    {
      "epoch": 0.6700774418727311,
      "grad_norm": 0.1906694620847702,
      "learning_rate": 0.002,
      "loss": 2.739084243774414,
      "step": 4870
    },
    {
      "epoch": 0.6703526276804818,
      "grad_norm": 0.1013701856136322,
      "learning_rate": 0.002,
      "loss": 2.7156333923339844,
      "step": 4872
    },
    {
      "epoch": 0.6706278134882324,
      "grad_norm": 0.14024211466312408,
      "learning_rate": 0.002,
      "loss": 2.777454376220703,
      "step": 4874
    },
    {
      "epoch": 0.670902999295983,
      "grad_norm": 0.26761820912361145,
      "learning_rate": 0.002,
      "loss": 2.719508171081543,
      "step": 4876
    },
    {
      "epoch": 0.6711781851037336,
      "grad_norm": 0.04919480159878731,
      "learning_rate": 0.002,
      "loss": 2.783919334411621,
      "step": 4878
    },
    {
      "epoch": 0.6714533709114842,
      "grad_norm": 0.18746471405029297,
      "learning_rate": 0.002,
      "loss": 2.746492624282837,
      "step": 4880
    },
    {
      "epoch": 0.6717285567192348,
      "grad_norm": 0.11452198773622513,
      "learning_rate": 0.002,
      "loss": 2.701842784881592,
      "step": 4882
    },
    {
      "epoch": 0.6720037425269854,
      "grad_norm": 0.10188298672437668,
      "learning_rate": 0.002,
      "loss": 2.7898545265197754,
      "step": 4884
    },
    {
      "epoch": 0.672278928334736,
      "grad_norm": 0.1383192539215088,
      "learning_rate": 0.002,
      "loss": 2.768636703491211,
      "step": 4886
    },
    {
      "epoch": 0.6725541141424867,
      "grad_norm": 0.13941606879234314,
      "learning_rate": 0.002,
      "loss": 2.7455625534057617,
      "step": 4888
    },
    {
      "epoch": 0.6728292999502372,
      "grad_norm": 0.04237629100680351,
      "learning_rate": 0.002,
      "loss": 2.736356496810913,
      "step": 4890
    },
    {
      "epoch": 0.6731044857579879,
      "grad_norm": 0.041843146085739136,
      "learning_rate": 0.002,
      "loss": 2.7671456336975098,
      "step": 4892
    },
    {
      "epoch": 0.6733796715657384,
      "grad_norm": 0.04753315821290016,
      "learning_rate": 0.002,
      "loss": 2.7744269371032715,
      "step": 4894
    },
    {
      "epoch": 0.6736548573734891,
      "grad_norm": 0.22024521231651306,
      "learning_rate": 0.002,
      "loss": 2.7752885818481445,
      "step": 4896
    },
    {
      "epoch": 0.6739300431812396,
      "grad_norm": 0.1325867772102356,
      "learning_rate": 0.002,
      "loss": 2.7491912841796875,
      "step": 4898
    },
    {
      "epoch": 0.6742052289889903,
      "grad_norm": 0.10824231058359146,
      "learning_rate": 0.002,
      "loss": 2.7361419200897217,
      "step": 4900
    },
    {
      "epoch": 0.6744804147967409,
      "grad_norm": 0.11158780008554459,
      "learning_rate": 0.002,
      "loss": 2.7524478435516357,
      "step": 4902
    },
    {
      "epoch": 0.6747556006044915,
      "grad_norm": 0.11814066767692566,
      "learning_rate": 0.002,
      "loss": 2.79014253616333,
      "step": 4904
    },
    {
      "epoch": 0.6750307864122421,
      "grad_norm": 0.04439884051680565,
      "learning_rate": 0.002,
      "loss": 2.7512412071228027,
      "step": 4906
    },
    {
      "epoch": 0.6753059722199927,
      "grad_norm": 0.10660838335752487,
      "learning_rate": 0.002,
      "loss": 2.734577178955078,
      "step": 4908
    },
    {
      "epoch": 0.6755811580277433,
      "grad_norm": 0.17953601479530334,
      "learning_rate": 0.002,
      "loss": 2.749026298522949,
      "step": 4910
    },
    {
      "epoch": 0.675856343835494,
      "grad_norm": 0.10318600386381149,
      "learning_rate": 0.002,
      "loss": 2.7781264781951904,
      "step": 4912
    },
    {
      "epoch": 0.6761315296432445,
      "grad_norm": 0.04907579720020294,
      "learning_rate": 0.002,
      "loss": 2.7548975944519043,
      "step": 4914
    },
    {
      "epoch": 0.6764067154509952,
      "grad_norm": 0.12304458767175674,
      "learning_rate": 0.002,
      "loss": 2.773611545562744,
      "step": 4916
    },
    {
      "epoch": 0.6766819012587457,
      "grad_norm": 0.14195096492767334,
      "learning_rate": 0.002,
      "loss": 2.7733755111694336,
      "step": 4918
    },
    {
      "epoch": 0.6769570870664964,
      "grad_norm": 0.21418553590774536,
      "learning_rate": 0.002,
      "loss": 2.793851852416992,
      "step": 4920
    },
    {
      "epoch": 0.6772322728742469,
      "grad_norm": 0.12151968479156494,
      "learning_rate": 0.002,
      "loss": 2.706146240234375,
      "step": 4922
    },
    {
      "epoch": 0.6775074586819976,
      "grad_norm": 0.20775587856769562,
      "learning_rate": 0.002,
      "loss": 2.764652967453003,
      "step": 4924
    },
    {
      "epoch": 0.6777826444897482,
      "grad_norm": 0.33756956458091736,
      "learning_rate": 0.002,
      "loss": 2.767137050628662,
      "step": 4926
    },
    {
      "epoch": 0.6780578302974988,
      "grad_norm": 0.08111308515071869,
      "learning_rate": 0.002,
      "loss": 2.8024210929870605,
      "step": 4928
    },
    {
      "epoch": 0.6783330161052494,
      "grad_norm": 0.2510635554790497,
      "learning_rate": 0.002,
      "loss": 2.7731757164001465,
      "step": 4930
    },
    {
      "epoch": 0.678608201913,
      "grad_norm": 0.12190916389226913,
      "learning_rate": 0.002,
      "loss": 2.779721260070801,
      "step": 4932
    },
    {
      "epoch": 0.6788833877207506,
      "grad_norm": 0.12117210775613785,
      "learning_rate": 0.002,
      "loss": 2.746683120727539,
      "step": 4934
    },
    {
      "epoch": 0.6791585735285012,
      "grad_norm": 0.0842788964509964,
      "learning_rate": 0.002,
      "loss": 2.7839393615722656,
      "step": 4936
    },
    {
      "epoch": 0.6794337593362518,
      "grad_norm": 0.13267627358436584,
      "learning_rate": 0.002,
      "loss": 2.7744503021240234,
      "step": 4938
    },
    {
      "epoch": 0.6797089451440025,
      "grad_norm": 0.042954228818416595,
      "learning_rate": 0.002,
      "loss": 2.772307872772217,
      "step": 4940
    },
    {
      "epoch": 0.679984130951753,
      "grad_norm": 0.034787412732839584,
      "learning_rate": 0.002,
      "loss": 2.7838046550750732,
      "step": 4942
    },
    {
      "epoch": 0.6802593167595037,
      "grad_norm": 0.07391159236431122,
      "learning_rate": 0.002,
      "loss": 2.7849373817443848,
      "step": 4944
    },
    {
      "epoch": 0.6805345025672542,
      "grad_norm": 0.12996214628219604,
      "learning_rate": 0.002,
      "loss": 2.710892915725708,
      "step": 4946
    },
    {
      "epoch": 0.6808096883750049,
      "grad_norm": 0.2574785351753235,
      "learning_rate": 0.002,
      "loss": 2.7773070335388184,
      "step": 4948
    },
    {
      "epoch": 0.6810848741827554,
      "grad_norm": 0.09591894596815109,
      "learning_rate": 0.002,
      "loss": 2.773817300796509,
      "step": 4950
    },
    {
      "epoch": 0.6813600599905061,
      "grad_norm": 0.11545904725790024,
      "learning_rate": 0.002,
      "loss": 2.738666534423828,
      "step": 4952
    },
    {
      "epoch": 0.6816352457982567,
      "grad_norm": 0.09856667369604111,
      "learning_rate": 0.002,
      "loss": 2.771507740020752,
      "step": 4954
    },
    {
      "epoch": 0.6819104316060073,
      "grad_norm": 0.059616778045892715,
      "learning_rate": 0.002,
      "loss": 2.7659780979156494,
      "step": 4956
    },
    {
      "epoch": 0.6821856174137579,
      "grad_norm": 0.058501966297626495,
      "learning_rate": 0.002,
      "loss": 2.7293357849121094,
      "step": 4958
    },
    {
      "epoch": 0.6824608032215085,
      "grad_norm": 0.030506644397974014,
      "learning_rate": 0.002,
      "loss": 2.7371163368225098,
      "step": 4960
    },
    {
      "epoch": 0.6827359890292591,
      "grad_norm": 0.03721730411052704,
      "learning_rate": 0.002,
      "loss": 2.7482619285583496,
      "step": 4962
    },
    {
      "epoch": 0.6830111748370098,
      "grad_norm": 0.04072214290499687,
      "learning_rate": 0.002,
      "loss": 2.7675743103027344,
      "step": 4964
    },
    {
      "epoch": 0.6832863606447603,
      "grad_norm": 0.07370176166296005,
      "learning_rate": 0.002,
      "loss": 2.756796360015869,
      "step": 4966
    },
    {
      "epoch": 0.683561546452511,
      "grad_norm": 0.13594961166381836,
      "learning_rate": 0.002,
      "loss": 2.748814582824707,
      "step": 4968
    },
    {
      "epoch": 0.6838367322602615,
      "grad_norm": 0.3188156187534332,
      "learning_rate": 0.002,
      "loss": 2.745969772338867,
      "step": 4970
    },
    {
      "epoch": 0.6841119180680122,
      "grad_norm": 0.4308308959007263,
      "learning_rate": 0.002,
      "loss": 2.790353775024414,
      "step": 4972
    },
    {
      "epoch": 0.6843871038757627,
      "grad_norm": 0.17594464123249054,
      "learning_rate": 0.002,
      "loss": 2.7127156257629395,
      "step": 4974
    },
    {
      "epoch": 0.6846622896835134,
      "grad_norm": 0.24598178267478943,
      "learning_rate": 0.002,
      "loss": 2.770634651184082,
      "step": 4976
    },
    {
      "epoch": 0.684937475491264,
      "grad_norm": 0.3996891379356384,
      "learning_rate": 0.002,
      "loss": 2.732117176055908,
      "step": 4978
    },
    {
      "epoch": 0.6852126612990146,
      "grad_norm": 0.21615208685398102,
      "learning_rate": 0.002,
      "loss": 2.7416200637817383,
      "step": 4980
    },
    {
      "epoch": 0.6854878471067652,
      "grad_norm": 0.03868839144706726,
      "learning_rate": 0.002,
      "loss": 2.6854190826416016,
      "step": 4982
    },
    {
      "epoch": 0.6857630329145158,
      "grad_norm": 0.11136237531900406,
      "learning_rate": 0.002,
      "loss": 2.7948427200317383,
      "step": 4984
    },
    {
      "epoch": 0.6860382187222664,
      "grad_norm": 0.20224246382713318,
      "learning_rate": 0.002,
      "loss": 2.7056682109832764,
      "step": 4986
    },
    {
      "epoch": 0.686313404530017,
      "grad_norm": 0.13259178400039673,
      "learning_rate": 0.002,
      "loss": 2.7689595222473145,
      "step": 4988
    },
    {
      "epoch": 0.6865885903377676,
      "grad_norm": 0.0394219234585762,
      "learning_rate": 0.002,
      "loss": 2.7426466941833496,
      "step": 4990
    },
    {
      "epoch": 0.6868637761455183,
      "grad_norm": 0.19577835500240326,
      "learning_rate": 0.002,
      "loss": 2.738414764404297,
      "step": 4992
    },
    {
      "epoch": 0.6871389619532688,
      "grad_norm": 0.14546853303909302,
      "learning_rate": 0.002,
      "loss": 2.755131721496582,
      "step": 4994
    },
    {
      "epoch": 0.6874141477610195,
      "grad_norm": 0.0336531326174736,
      "learning_rate": 0.002,
      "loss": 2.737781286239624,
      "step": 4996
    },
    {
      "epoch": 0.68768933356877,
      "grad_norm": 0.08386953920125961,
      "learning_rate": 0.002,
      "loss": 2.736933946609497,
      "step": 4998
    },
    {
      "epoch": 0.6879645193765207,
      "grad_norm": 0.09036701917648315,
      "learning_rate": 0.002,
      "loss": 2.7870240211486816,
      "step": 5000
    },
    {
      "epoch": 0.6882397051842712,
      "grad_norm": 0.03252725675702095,
      "learning_rate": 0.002,
      "loss": 2.7699694633483887,
      "step": 5002
    },
    {
      "epoch": 0.6885148909920219,
      "grad_norm": 0.04963240772485733,
      "learning_rate": 0.002,
      "loss": 2.715970993041992,
      "step": 5004
    },
    {
      "epoch": 0.6887900767997726,
      "grad_norm": 0.05736686661839485,
      "learning_rate": 0.002,
      "loss": 2.78068208694458,
      "step": 5006
    },
    {
      "epoch": 0.6890652626075231,
      "grad_norm": 0.04147119075059891,
      "learning_rate": 0.002,
      "loss": 2.7660117149353027,
      "step": 5008
    },
    {
      "epoch": 0.6893404484152738,
      "grad_norm": 0.09883071482181549,
      "learning_rate": 0.002,
      "loss": 2.7223641872406006,
      "step": 5010
    },
    {
      "epoch": 0.6896156342230243,
      "grad_norm": 0.27365341782569885,
      "learning_rate": 0.002,
      "loss": 2.7457103729248047,
      "step": 5012
    },
    {
      "epoch": 0.689890820030775,
      "grad_norm": 0.2590028941631317,
      "learning_rate": 0.002,
      "loss": 2.7431604862213135,
      "step": 5014
    },
    {
      "epoch": 0.6901660058385256,
      "grad_norm": 0.06908375769853592,
      "learning_rate": 0.002,
      "loss": 2.7460594177246094,
      "step": 5016
    },
    {
      "epoch": 0.6904411916462762,
      "grad_norm": 0.13981248438358307,
      "learning_rate": 0.002,
      "loss": 2.7556703090667725,
      "step": 5018
    },
    {
      "epoch": 0.6907163774540268,
      "grad_norm": 0.09859006851911545,
      "learning_rate": 0.002,
      "loss": 2.739158868789673,
      "step": 5020
    },
    {
      "epoch": 0.6909915632617774,
      "grad_norm": 0.1772218644618988,
      "learning_rate": 0.002,
      "loss": 2.770200729370117,
      "step": 5022
    },
    {
      "epoch": 0.691266749069528,
      "grad_norm": 0.08911735564470291,
      "learning_rate": 0.002,
      "loss": 2.742715358734131,
      "step": 5024
    },
    {
      "epoch": 0.6915419348772786,
      "grad_norm": 0.08021596819162369,
      "learning_rate": 0.002,
      "loss": 2.7161808013916016,
      "step": 5026
    },
    {
      "epoch": 0.6918171206850292,
      "grad_norm": 0.14807003736495972,
      "learning_rate": 0.002,
      "loss": 2.724924325942993,
      "step": 5028
    },
    {
      "epoch": 0.6920923064927799,
      "grad_norm": 0.22628124058246613,
      "learning_rate": 0.002,
      "loss": 2.7445833683013916,
      "step": 5030
    },
    {
      "epoch": 0.6923674923005304,
      "grad_norm": 0.05553789809346199,
      "learning_rate": 0.002,
      "loss": 2.763990879058838,
      "step": 5032
    },
    {
      "epoch": 0.6926426781082811,
      "grad_norm": 0.16443507373332977,
      "learning_rate": 0.002,
      "loss": 2.761540174484253,
      "step": 5034
    },
    {
      "epoch": 0.6929178639160316,
      "grad_norm": 0.15191857516765594,
      "learning_rate": 0.002,
      "loss": 2.6987085342407227,
      "step": 5036
    },
    {
      "epoch": 0.6931930497237823,
      "grad_norm": 0.18954932689666748,
      "learning_rate": 0.002,
      "loss": 2.74210262298584,
      "step": 5038
    },
    {
      "epoch": 0.6934682355315328,
      "grad_norm": 0.16355451941490173,
      "learning_rate": 0.002,
      "loss": 2.7266902923583984,
      "step": 5040
    },
    {
      "epoch": 0.6937434213392835,
      "grad_norm": 0.13810615241527557,
      "learning_rate": 0.002,
      "loss": 2.7450640201568604,
      "step": 5042
    },
    {
      "epoch": 0.6940186071470341,
      "grad_norm": 0.12235613167285919,
      "learning_rate": 0.002,
      "loss": 2.7522966861724854,
      "step": 5044
    },
    {
      "epoch": 0.6942937929547847,
      "grad_norm": 0.03459567576646805,
      "learning_rate": 0.002,
      "loss": 2.7404470443725586,
      "step": 5046
    },
    {
      "epoch": 0.6945689787625353,
      "grad_norm": 0.128171905875206,
      "learning_rate": 0.002,
      "loss": 2.740525722503662,
      "step": 5048
    },
    {
      "epoch": 0.6948441645702859,
      "grad_norm": 0.10178013890981674,
      "learning_rate": 0.002,
      "loss": 2.7670071125030518,
      "step": 5050
    },
    {
      "epoch": 0.6951193503780365,
      "grad_norm": 0.0541137233376503,
      "learning_rate": 0.002,
      "loss": 2.757509231567383,
      "step": 5052
    },
    {
      "epoch": 0.6953945361857871,
      "grad_norm": 0.24903662502765656,
      "learning_rate": 0.002,
      "loss": 2.7445645332336426,
      "step": 5054
    },
    {
      "epoch": 0.6956697219935377,
      "grad_norm": 0.06801507622003555,
      "learning_rate": 0.002,
      "loss": 2.724306106567383,
      "step": 5056
    },
    {
      "epoch": 0.6959449078012884,
      "grad_norm": 0.06653958559036255,
      "learning_rate": 0.002,
      "loss": 2.731931209564209,
      "step": 5058
    },
    {
      "epoch": 0.6962200936090389,
      "grad_norm": 0.12251731008291245,
      "learning_rate": 0.002,
      "loss": 2.722886562347412,
      "step": 5060
    },
    {
      "epoch": 0.6964952794167896,
      "grad_norm": 0.04208992049098015,
      "learning_rate": 0.002,
      "loss": 2.7520830631256104,
      "step": 5062
    },
    {
      "epoch": 0.6967704652245401,
      "grad_norm": 0.03353813290596008,
      "learning_rate": 0.002,
      "loss": 2.758491039276123,
      "step": 5064
    },
    {
      "epoch": 0.6970456510322908,
      "grad_norm": 0.13965629041194916,
      "learning_rate": 0.002,
      "loss": 2.766890287399292,
      "step": 5066
    },
    {
      "epoch": 0.6973208368400414,
      "grad_norm": 0.28530746698379517,
      "learning_rate": 0.002,
      "loss": 2.7301883697509766,
      "step": 5068
    },
    {
      "epoch": 0.697596022647792,
      "grad_norm": 0.10992494970560074,
      "learning_rate": 0.002,
      "loss": 2.806942939758301,
      "step": 5070
    },
    {
      "epoch": 0.6978712084555426,
      "grad_norm": 0.07598753273487091,
      "learning_rate": 0.002,
      "loss": 2.7742018699645996,
      "step": 5072
    },
    {
      "epoch": 0.6981463942632932,
      "grad_norm": 0.08700065314769745,
      "learning_rate": 0.002,
      "loss": 2.714505195617676,
      "step": 5074
    },
    {
      "epoch": 0.6984215800710438,
      "grad_norm": 0.2163492888212204,
      "learning_rate": 0.002,
      "loss": 2.7411561012268066,
      "step": 5076
    },
    {
      "epoch": 0.6986967658787944,
      "grad_norm": 0.11316950619220734,
      "learning_rate": 0.002,
      "loss": 2.754544734954834,
      "step": 5078
    },
    {
      "epoch": 0.698971951686545,
      "grad_norm": 0.05309319868683815,
      "learning_rate": 0.002,
      "loss": 2.745251178741455,
      "step": 5080
    },
    {
      "epoch": 0.6992471374942957,
      "grad_norm": 0.19135428965091705,
      "learning_rate": 0.002,
      "loss": 2.742215394973755,
      "step": 5082
    },
    {
      "epoch": 0.6995223233020462,
      "grad_norm": 0.08830466866493225,
      "learning_rate": 0.002,
      "loss": 2.732840061187744,
      "step": 5084
    },
    {
      "epoch": 0.6997975091097969,
      "grad_norm": 0.19623616337776184,
      "learning_rate": 0.002,
      "loss": 2.726914405822754,
      "step": 5086
    },
    {
      "epoch": 0.7000726949175474,
      "grad_norm": 0.1256786435842514,
      "learning_rate": 0.002,
      "loss": 2.771195888519287,
      "step": 5088
    },
    {
      "epoch": 0.7003478807252981,
      "grad_norm": 0.1438838541507721,
      "learning_rate": 0.002,
      "loss": 2.7426939010620117,
      "step": 5090
    },
    {
      "epoch": 0.7006230665330486,
      "grad_norm": 0.3017856478691101,
      "learning_rate": 0.002,
      "loss": 2.7565832138061523,
      "step": 5092
    },
    {
      "epoch": 0.7008982523407993,
      "grad_norm": 0.23628365993499756,
      "learning_rate": 0.002,
      "loss": 2.760361671447754,
      "step": 5094
    },
    {
      "epoch": 0.7011734381485499,
      "grad_norm": 0.21108247339725494,
      "learning_rate": 0.002,
      "loss": 2.7274436950683594,
      "step": 5096
    },
    {
      "epoch": 0.7014486239563005,
      "grad_norm": 0.13628320395946503,
      "learning_rate": 0.002,
      "loss": 2.7246923446655273,
      "step": 5098
    },
    {
      "epoch": 0.7017238097640511,
      "grad_norm": 0.11961481720209122,
      "learning_rate": 0.002,
      "loss": 2.721449375152588,
      "step": 5100
    },
    {
      "epoch": 0.7019989955718017,
      "grad_norm": 0.04153917357325554,
      "learning_rate": 0.002,
      "loss": 2.7460432052612305,
      "step": 5102
    },
    {
      "epoch": 0.7022741813795523,
      "grad_norm": 0.11897610872983932,
      "learning_rate": 0.002,
      "loss": 2.743234634399414,
      "step": 5104
    },
    {
      "epoch": 0.7025493671873029,
      "grad_norm": 0.06572137773036957,
      "learning_rate": 0.002,
      "loss": 2.7100534439086914,
      "step": 5106
    },
    {
      "epoch": 0.7028245529950535,
      "grad_norm": 0.0329618901014328,
      "learning_rate": 0.002,
      "loss": 2.712924003601074,
      "step": 5108
    },
    {
      "epoch": 0.7030997388028042,
      "grad_norm": 0.06054699420928955,
      "learning_rate": 0.002,
      "loss": 2.708054304122925,
      "step": 5110
    },
    {
      "epoch": 0.7033749246105547,
      "grad_norm": 0.042254917323589325,
      "learning_rate": 0.002,
      "loss": 2.8109130859375,
      "step": 5112
    },
    {
      "epoch": 0.7036501104183054,
      "grad_norm": 0.11490693688392639,
      "learning_rate": 0.002,
      "loss": 2.7592997550964355,
      "step": 5114
    },
    {
      "epoch": 0.7039252962260559,
      "grad_norm": 0.20078390836715698,
      "learning_rate": 0.002,
      "loss": 2.7321927547454834,
      "step": 5116
    },
    {
      "epoch": 0.7042004820338066,
      "grad_norm": 0.3113877475261688,
      "learning_rate": 0.002,
      "loss": 2.7593016624450684,
      "step": 5118
    },
    {
      "epoch": 0.7044756678415571,
      "grad_norm": 0.06452342122793198,
      "learning_rate": 0.002,
      "loss": 2.6933770179748535,
      "step": 5120
    },
    {
      "epoch": 0.7047508536493078,
      "grad_norm": 0.265828013420105,
      "learning_rate": 0.002,
      "loss": 2.7194838523864746,
      "step": 5122
    },
    {
      "epoch": 0.7050260394570584,
      "grad_norm": 0.1277655065059662,
      "learning_rate": 0.002,
      "loss": 2.754706859588623,
      "step": 5124
    },
    {
      "epoch": 0.705301225264809,
      "grad_norm": 0.11631792783737183,
      "learning_rate": 0.002,
      "loss": 2.731449842453003,
      "step": 5126
    },
    {
      "epoch": 0.7055764110725596,
      "grad_norm": 0.04764695465564728,
      "learning_rate": 0.002,
      "loss": 2.7655837535858154,
      "step": 5128
    },
    {
      "epoch": 0.7058515968803102,
      "grad_norm": 0.05676012858748436,
      "learning_rate": 0.002,
      "loss": 2.756338596343994,
      "step": 5130
    },
    {
      "epoch": 0.7061267826880608,
      "grad_norm": 0.05571962893009186,
      "learning_rate": 0.002,
      "loss": 2.7516732215881348,
      "step": 5132
    },
    {
      "epoch": 0.7064019684958115,
      "grad_norm": 0.0453939326107502,
      "learning_rate": 0.002,
      "loss": 2.772299289703369,
      "step": 5134
    },
    {
      "epoch": 0.706677154303562,
      "grad_norm": 0.054444942623376846,
      "learning_rate": 0.002,
      "loss": 2.71602725982666,
      "step": 5136
    },
    {
      "epoch": 0.7069523401113127,
      "grad_norm": 0.03041304275393486,
      "learning_rate": 0.002,
      "loss": 2.7244279384613037,
      "step": 5138
    },
    {
      "epoch": 0.7072275259190632,
      "grad_norm": 0.15981118381023407,
      "learning_rate": 0.002,
      "loss": 2.72352933883667,
      "step": 5140
    },
    {
      "epoch": 0.7075027117268139,
      "grad_norm": 0.43889734148979187,
      "learning_rate": 0.002,
      "loss": 2.7495126724243164,
      "step": 5142
    },
    {
      "epoch": 0.7077778975345644,
      "grad_norm": 0.09823177754878998,
      "learning_rate": 0.002,
      "loss": 2.7317919731140137,
      "step": 5144
    },
    {
      "epoch": 0.7080530833423151,
      "grad_norm": 0.32765713334083557,
      "learning_rate": 0.002,
      "loss": 2.744907855987549,
      "step": 5146
    },
    {
      "epoch": 0.7083282691500657,
      "grad_norm": 0.15168263018131256,
      "learning_rate": 0.002,
      "loss": 2.7719736099243164,
      "step": 5148
    },
    {
      "epoch": 0.7086034549578163,
      "grad_norm": 0.04158763960003853,
      "learning_rate": 0.002,
      "loss": 2.7832272052764893,
      "step": 5150
    },
    {
      "epoch": 0.708878640765567,
      "grad_norm": 0.03655025362968445,
      "learning_rate": 0.002,
      "loss": 2.7822365760803223,
      "step": 5152
    },
    {
      "epoch": 0.7091538265733175,
      "grad_norm": 0.03997894003987312,
      "learning_rate": 0.002,
      "loss": 2.7423346042633057,
      "step": 5154
    },
    {
      "epoch": 0.7094290123810681,
      "grad_norm": 0.09835279732942581,
      "learning_rate": 0.002,
      "loss": 2.7490808963775635,
      "step": 5156
    },
    {
      "epoch": 0.7097041981888187,
      "grad_norm": 0.07743558287620544,
      "learning_rate": 0.002,
      "loss": 2.7629287242889404,
      "step": 5158
    },
    {
      "epoch": 0.7099793839965693,
      "grad_norm": 0.0710218995809555,
      "learning_rate": 0.002,
      "loss": 2.7496337890625,
      "step": 5160
    },
    {
      "epoch": 0.71025456980432,
      "grad_norm": 0.04383315145969391,
      "learning_rate": 0.002,
      "loss": 2.738748073577881,
      "step": 5162
    },
    {
      "epoch": 0.7105297556120705,
      "grad_norm": 0.03822679445147514,
      "learning_rate": 0.002,
      "loss": 2.6809616088867188,
      "step": 5164
    },
    {
      "epoch": 0.7108049414198212,
      "grad_norm": 0.04910925403237343,
      "learning_rate": 0.002,
      "loss": 2.7466747760772705,
      "step": 5166
    },
    {
      "epoch": 0.7110801272275717,
      "grad_norm": 0.043189335614442825,
      "learning_rate": 0.002,
      "loss": 2.7878642082214355,
      "step": 5168
    },
    {
      "epoch": 0.7113553130353224,
      "grad_norm": 0.06630773097276688,
      "learning_rate": 0.002,
      "loss": 2.752969264984131,
      "step": 5170
    },
    {
      "epoch": 0.711630498843073,
      "grad_norm": 0.20272696018218994,
      "learning_rate": 0.002,
      "loss": 2.707780361175537,
      "step": 5172
    },
    {
      "epoch": 0.7119056846508236,
      "grad_norm": 0.3515608310699463,
      "learning_rate": 0.002,
      "loss": 2.6925196647644043,
      "step": 5174
    },
    {
      "epoch": 0.7121808704585743,
      "grad_norm": 0.08160601556301117,
      "learning_rate": 0.002,
      "loss": 2.756422519683838,
      "step": 5176
    },
    {
      "epoch": 0.7124560562663248,
      "grad_norm": 0.13465537130832672,
      "learning_rate": 0.002,
      "loss": 2.7144598960876465,
      "step": 5178
    },
    {
      "epoch": 0.7127312420740755,
      "grad_norm": 0.048140786588191986,
      "learning_rate": 0.002,
      "loss": 2.7473247051239014,
      "step": 5180
    },
    {
      "epoch": 0.713006427881826,
      "grad_norm": 0.060167111456394196,
      "learning_rate": 0.002,
      "loss": 2.7308688163757324,
      "step": 5182
    },
    {
      "epoch": 0.7132816136895767,
      "grad_norm": 0.04813157021999359,
      "learning_rate": 0.002,
      "loss": 2.7338833808898926,
      "step": 5184
    },
    {
      "epoch": 0.7135567994973273,
      "grad_norm": 0.20479492843151093,
      "learning_rate": 0.002,
      "loss": 2.759087324142456,
      "step": 5186
    },
    {
      "epoch": 0.7138319853050779,
      "grad_norm": 0.1580643504858017,
      "learning_rate": 0.002,
      "loss": 2.7368340492248535,
      "step": 5188
    },
    {
      "epoch": 0.7141071711128285,
      "grad_norm": 0.09421370178461075,
      "learning_rate": 0.002,
      "loss": 2.719545364379883,
      "step": 5190
    },
    {
      "epoch": 0.714382356920579,
      "grad_norm": 0.2667948603630066,
      "learning_rate": 0.002,
      "loss": 2.720402956008911,
      "step": 5192
    },
    {
      "epoch": 0.7146575427283297,
      "grad_norm": 0.0968773365020752,
      "learning_rate": 0.002,
      "loss": 2.750148296356201,
      "step": 5194
    },
    {
      "epoch": 0.7149327285360803,
      "grad_norm": 0.41851916909217834,
      "learning_rate": 0.002,
      "loss": 2.742311477661133,
      "step": 5196
    },
    {
      "epoch": 0.7152079143438309,
      "grad_norm": 0.2052542269229889,
      "learning_rate": 0.002,
      "loss": 2.744442939758301,
      "step": 5198
    },
    {
      "epoch": 0.7154831001515816,
      "grad_norm": 0.16675549745559692,
      "learning_rate": 0.002,
      "loss": 2.7052230834960938,
      "step": 5200
    },
    {
      "epoch": 0.7157582859593321,
      "grad_norm": 0.34551024436950684,
      "learning_rate": 0.002,
      "loss": 2.7283639907836914,
      "step": 5202
    },
    {
      "epoch": 0.7160334717670828,
      "grad_norm": 0.03716428205370903,
      "learning_rate": 0.002,
      "loss": 2.7348008155822754,
      "step": 5204
    },
    {
      "epoch": 0.7163086575748333,
      "grad_norm": 0.25605109333992004,
      "learning_rate": 0.002,
      "loss": 2.7303709983825684,
      "step": 5206
    },
    {
      "epoch": 0.716583843382584,
      "grad_norm": 0.11429384350776672,
      "learning_rate": 0.002,
      "loss": 2.7193386554718018,
      "step": 5208
    },
    {
      "epoch": 0.7168590291903345,
      "grad_norm": 0.11325014382600784,
      "learning_rate": 0.002,
      "loss": 2.7385270595550537,
      "step": 5210
    },
    {
      "epoch": 0.7171342149980852,
      "grad_norm": 0.06642210483551025,
      "learning_rate": 0.002,
      "loss": 2.7787230014801025,
      "step": 5212
    },
    {
      "epoch": 0.7174094008058358,
      "grad_norm": 0.04675738513469696,
      "learning_rate": 0.002,
      "loss": 2.760915756225586,
      "step": 5214
    },
    {
      "epoch": 0.7176845866135864,
      "grad_norm": 0.08507761359214783,
      "learning_rate": 0.002,
      "loss": 2.7005114555358887,
      "step": 5216
    },
    {
      "epoch": 0.717959772421337,
      "grad_norm": 0.13258042931556702,
      "learning_rate": 0.002,
      "loss": 2.750274181365967,
      "step": 5218
    },
    {
      "epoch": 0.7182349582290876,
      "grad_norm": 0.10810472816228867,
      "learning_rate": 0.002,
      "loss": 2.7409515380859375,
      "step": 5220
    },
    {
      "epoch": 0.7185101440368382,
      "grad_norm": 0.28754526376724243,
      "learning_rate": 0.002,
      "loss": 2.756099224090576,
      "step": 5222
    },
    {
      "epoch": 0.7187853298445888,
      "grad_norm": 0.04905043914914131,
      "learning_rate": 0.002,
      "loss": 2.737917184829712,
      "step": 5224
    },
    {
      "epoch": 0.7190605156523394,
      "grad_norm": 0.2820434272289276,
      "learning_rate": 0.002,
      "loss": 2.7391245365142822,
      "step": 5226
    },
    {
      "epoch": 0.7193357014600901,
      "grad_norm": 0.07548294961452484,
      "learning_rate": 0.002,
      "loss": 2.7501864433288574,
      "step": 5228
    },
    {
      "epoch": 0.7196108872678406,
      "grad_norm": 0.11995084583759308,
      "learning_rate": 0.002,
      "loss": 2.736485004425049,
      "step": 5230
    },
    {
      "epoch": 0.7198860730755913,
      "grad_norm": 0.2262517213821411,
      "learning_rate": 0.002,
      "loss": 2.7338881492614746,
      "step": 5232
    },
    {
      "epoch": 0.7201612588833418,
      "grad_norm": 0.09683408588171005,
      "learning_rate": 0.002,
      "loss": 2.7240676879882812,
      "step": 5234
    },
    {
      "epoch": 0.7204364446910925,
      "grad_norm": 0.1853836327791214,
      "learning_rate": 0.002,
      "loss": 2.744168996810913,
      "step": 5236
    },
    {
      "epoch": 0.7207116304988431,
      "grad_norm": 0.03201982378959656,
      "learning_rate": 0.002,
      "loss": 2.7580976486206055,
      "step": 5238
    },
    {
      "epoch": 0.7209868163065937,
      "grad_norm": 0.20157277584075928,
      "learning_rate": 0.002,
      "loss": 2.7400331497192383,
      "step": 5240
    },
    {
      "epoch": 0.7212620021143443,
      "grad_norm": 0.03003731369972229,
      "learning_rate": 0.002,
      "loss": 2.7073869705200195,
      "step": 5242
    },
    {
      "epoch": 0.7215371879220949,
      "grad_norm": 0.3131496012210846,
      "learning_rate": 0.002,
      "loss": 2.7528653144836426,
      "step": 5244
    },
    {
      "epoch": 0.7218123737298455,
      "grad_norm": 0.1683933585882187,
      "learning_rate": 0.002,
      "loss": 2.742889881134033,
      "step": 5246
    },
    {
      "epoch": 0.7220875595375961,
      "grad_norm": 0.07552363723516464,
      "learning_rate": 0.002,
      "loss": 2.7571980953216553,
      "step": 5248
    },
    {
      "epoch": 0.7223627453453467,
      "grad_norm": 0.17775364220142365,
      "learning_rate": 0.002,
      "loss": 2.7059378623962402,
      "step": 5250
    },
    {
      "epoch": 0.7226379311530974,
      "grad_norm": 0.049823760986328125,
      "learning_rate": 0.002,
      "loss": 2.716686248779297,
      "step": 5252
    },
    {
      "epoch": 0.7229131169608479,
      "grad_norm": 0.0461730994284153,
      "learning_rate": 0.002,
      "loss": 2.6978721618652344,
      "step": 5254
    },
    {
      "epoch": 0.7231883027685986,
      "grad_norm": 0.10014572739601135,
      "learning_rate": 0.002,
      "loss": 2.737150192260742,
      "step": 5256
    },
    {
      "epoch": 0.7234634885763491,
      "grad_norm": 0.057838257402181625,
      "learning_rate": 0.002,
      "loss": 2.7339301109313965,
      "step": 5258
    },
    {
      "epoch": 0.7237386743840998,
      "grad_norm": 0.16916964948177338,
      "learning_rate": 0.002,
      "loss": 2.740330696105957,
      "step": 5260
    },
    {
      "epoch": 0.7240138601918503,
      "grad_norm": 0.043244618922472,
      "learning_rate": 0.002,
      "loss": 2.73663330078125,
      "step": 5262
    },
    {
      "epoch": 0.724289045999601,
      "grad_norm": 0.19956795871257782,
      "learning_rate": 0.002,
      "loss": 2.7380900382995605,
      "step": 5264
    },
    {
      "epoch": 0.7245642318073516,
      "grad_norm": 0.29503822326660156,
      "learning_rate": 0.002,
      "loss": 2.7604308128356934,
      "step": 5266
    },
    {
      "epoch": 0.7248394176151022,
      "grad_norm": 0.09469465166330338,
      "learning_rate": 0.002,
      "loss": 2.734872817993164,
      "step": 5268
    },
    {
      "epoch": 0.7251146034228528,
      "grad_norm": 0.20454619824886322,
      "learning_rate": 0.002,
      "loss": 2.7566394805908203,
      "step": 5270
    },
    {
      "epoch": 0.7253897892306034,
      "grad_norm": 0.04195214807987213,
      "learning_rate": 0.002,
      "loss": 2.7228212356567383,
      "step": 5272
    },
    {
      "epoch": 0.725664975038354,
      "grad_norm": 0.2537505626678467,
      "learning_rate": 0.002,
      "loss": 2.768928050994873,
      "step": 5274
    },
    {
      "epoch": 0.7259401608461046,
      "grad_norm": 0.06991337984800339,
      "learning_rate": 0.002,
      "loss": 2.7569210529327393,
      "step": 5276
    },
    {
      "epoch": 0.7262153466538552,
      "grad_norm": 0.11429796367883682,
      "learning_rate": 0.002,
      "loss": 2.680638551712036,
      "step": 5278
    },
    {
      "epoch": 0.7264905324616059,
      "grad_norm": 0.1281295120716095,
      "learning_rate": 0.002,
      "loss": 2.756094217300415,
      "step": 5280
    },
    {
      "epoch": 0.7267657182693564,
      "grad_norm": 0.2657342851161957,
      "learning_rate": 0.002,
      "loss": 2.730393409729004,
      "step": 5282
    },
    {
      "epoch": 0.7270409040771071,
      "grad_norm": 0.08159232139587402,
      "learning_rate": 0.002,
      "loss": 2.7140097618103027,
      "step": 5284
    },
    {
      "epoch": 0.7273160898848576,
      "grad_norm": 0.1442427635192871,
      "learning_rate": 0.002,
      "loss": 2.7419183254241943,
      "step": 5286
    },
    {
      "epoch": 0.7275912756926083,
      "grad_norm": 0.22151897847652435,
      "learning_rate": 0.002,
      "loss": 2.755460500717163,
      "step": 5288
    },
    {
      "epoch": 0.7278664615003589,
      "grad_norm": 0.1704312115907669,
      "learning_rate": 0.002,
      "loss": 2.736084222793579,
      "step": 5290
    },
    {
      "epoch": 0.7281416473081095,
      "grad_norm": 0.17361141741275787,
      "learning_rate": 0.002,
      "loss": 2.7374629974365234,
      "step": 5292
    },
    {
      "epoch": 0.7284168331158601,
      "grad_norm": 0.20856544375419617,
      "learning_rate": 0.002,
      "loss": 2.731370449066162,
      "step": 5294
    },
    {
      "epoch": 0.7286920189236107,
      "grad_norm": 0.08784572035074234,
      "learning_rate": 0.002,
      "loss": 2.7079524993896484,
      "step": 5296
    },
    {
      "epoch": 0.7289672047313613,
      "grad_norm": 0.2037772834300995,
      "learning_rate": 0.002,
      "loss": 2.7074899673461914,
      "step": 5298
    },
    {
      "epoch": 0.7292423905391119,
      "grad_norm": 0.03497314453125,
      "learning_rate": 0.002,
      "loss": 2.7060954570770264,
      "step": 5300
    },
    {
      "epoch": 0.7295175763468625,
      "grad_norm": 0.1308802366256714,
      "learning_rate": 0.002,
      "loss": 2.677262306213379,
      "step": 5302
    },
    {
      "epoch": 0.7297927621546132,
      "grad_norm": 0.12115684896707535,
      "learning_rate": 0.002,
      "loss": 2.749521255493164,
      "step": 5304
    },
    {
      "epoch": 0.7300679479623637,
      "grad_norm": 0.0896412804722786,
      "learning_rate": 0.002,
      "loss": 2.707195997238159,
      "step": 5306
    },
    {
      "epoch": 0.7303431337701144,
      "grad_norm": 0.07257010787725449,
      "learning_rate": 0.002,
      "loss": 2.717095375061035,
      "step": 5308
    },
    {
      "epoch": 0.7306183195778649,
      "grad_norm": 0.028701404109597206,
      "learning_rate": 0.002,
      "loss": 2.7069735527038574,
      "step": 5310
    },
    {
      "epoch": 0.7308935053856156,
      "grad_norm": 0.04566067457199097,
      "learning_rate": 0.002,
      "loss": 2.7215499877929688,
      "step": 5312
    },
    {
      "epoch": 0.7311686911933661,
      "grad_norm": 0.11486159265041351,
      "learning_rate": 0.002,
      "loss": 2.7438559532165527,
      "step": 5314
    },
    {
      "epoch": 0.7314438770011168,
      "grad_norm": 0.1713673621416092,
      "learning_rate": 0.002,
      "loss": 2.753812074661255,
      "step": 5316
    },
    {
      "epoch": 0.7317190628088674,
      "grad_norm": 0.27080628275871277,
      "learning_rate": 0.002,
      "loss": 2.715667963027954,
      "step": 5318
    },
    {
      "epoch": 0.731994248616618,
      "grad_norm": 0.2336093634366989,
      "learning_rate": 0.002,
      "loss": 2.736769676208496,
      "step": 5320
    },
    {
      "epoch": 0.7322694344243686,
      "grad_norm": 0.07911860942840576,
      "learning_rate": 0.002,
      "loss": 2.742013454437256,
      "step": 5322
    },
    {
      "epoch": 0.7325446202321192,
      "grad_norm": 0.25058916211128235,
      "learning_rate": 0.002,
      "loss": 2.743830680847168,
      "step": 5324
    },
    {
      "epoch": 0.7328198060398698,
      "grad_norm": 0.05936885252594948,
      "learning_rate": 0.002,
      "loss": 2.7647008895874023,
      "step": 5326
    },
    {
      "epoch": 0.7330949918476204,
      "grad_norm": 0.03546294569969177,
      "learning_rate": 0.002,
      "loss": 2.726436138153076,
      "step": 5328
    },
    {
      "epoch": 0.733370177655371,
      "grad_norm": 0.031376659870147705,
      "learning_rate": 0.002,
      "loss": 2.7537221908569336,
      "step": 5330
    },
    {
      "epoch": 0.7336453634631217,
      "grad_norm": 0.051873065531253815,
      "learning_rate": 0.002,
      "loss": 2.7248599529266357,
      "step": 5332
    },
    {
      "epoch": 0.7339205492708722,
      "grad_norm": 0.03323166072368622,
      "learning_rate": 0.002,
      "loss": 2.7007369995117188,
      "step": 5334
    },
    {
      "epoch": 0.7341957350786229,
      "grad_norm": 0.06660645455121994,
      "learning_rate": 0.002,
      "loss": 2.741938591003418,
      "step": 5336
    },
    {
      "epoch": 0.7344709208863734,
      "grad_norm": 0.037418924272060394,
      "learning_rate": 0.002,
      "loss": 2.682450294494629,
      "step": 5338
    },
    {
      "epoch": 0.7347461066941241,
      "grad_norm": 0.1004800796508789,
      "learning_rate": 0.002,
      "loss": 2.716909170150757,
      "step": 5340
    },
    {
      "epoch": 0.7350212925018748,
      "grad_norm": 0.3279283344745636,
      "learning_rate": 0.002,
      "loss": 2.720233917236328,
      "step": 5342
    },
    {
      "epoch": 0.7352964783096253,
      "grad_norm": 0.27650192379951477,
      "learning_rate": 0.002,
      "loss": 2.728104591369629,
      "step": 5344
    },
    {
      "epoch": 0.735571664117376,
      "grad_norm": 0.3145110011100769,
      "learning_rate": 0.002,
      "loss": 2.7015140056610107,
      "step": 5346
    },
    {
      "epoch": 0.7358468499251265,
      "grad_norm": 0.161802276968956,
      "learning_rate": 0.002,
      "loss": 2.728156566619873,
      "step": 5348
    },
    {
      "epoch": 0.7361220357328772,
      "grad_norm": 0.3629784882068634,
      "learning_rate": 0.002,
      "loss": 2.7618818283081055,
      "step": 5350
    },
    {
      "epoch": 0.7363972215406277,
      "grad_norm": 0.20646952092647552,
      "learning_rate": 0.002,
      "loss": 2.74552845954895,
      "step": 5352
    },
    {
      "epoch": 0.7366724073483784,
      "grad_norm": 0.1814870536327362,
      "learning_rate": 0.002,
      "loss": 2.7080039978027344,
      "step": 5354
    },
    {
      "epoch": 0.736947593156129,
      "grad_norm": 0.28050899505615234,
      "learning_rate": 0.002,
      "loss": 2.7104077339172363,
      "step": 5356
    },
    {
      "epoch": 0.7372227789638796,
      "grad_norm": 0.08096028864383698,
      "learning_rate": 0.002,
      "loss": 2.714719772338867,
      "step": 5358
    },
    {
      "epoch": 0.7374979647716302,
      "grad_norm": 0.1051672026515007,
      "learning_rate": 0.002,
      "loss": 2.6961216926574707,
      "step": 5360
    },
    {
      "epoch": 0.7377731505793808,
      "grad_norm": 0.18235480785369873,
      "learning_rate": 0.002,
      "loss": 2.7391762733459473,
      "step": 5362
    },
    {
      "epoch": 0.7380483363871314,
      "grad_norm": 0.08448714762926102,
      "learning_rate": 0.002,
      "loss": 2.6909666061401367,
      "step": 5364
    },
    {
      "epoch": 0.738323522194882,
      "grad_norm": 0.11532846093177795,
      "learning_rate": 0.002,
      "loss": 2.7186360359191895,
      "step": 5366
    },
    {
      "epoch": 0.7385987080026326,
      "grad_norm": 0.3220888078212738,
      "learning_rate": 0.002,
      "loss": 2.733574390411377,
      "step": 5368
    },
    {
      "epoch": 0.7388738938103833,
      "grad_norm": 0.06898949295282364,
      "learning_rate": 0.002,
      "loss": 2.7195487022399902,
      "step": 5370
    },
    {
      "epoch": 0.7391490796181338,
      "grad_norm": 0.2124025821685791,
      "learning_rate": 0.002,
      "loss": 2.762742042541504,
      "step": 5372
    },
    {
      "epoch": 0.7394242654258845,
      "grad_norm": 0.21833188831806183,
      "learning_rate": 0.002,
      "loss": 2.7359461784362793,
      "step": 5374
    },
    {
      "epoch": 0.739699451233635,
      "grad_norm": 0.1346292346715927,
      "learning_rate": 0.002,
      "loss": 2.728116035461426,
      "step": 5376
    },
    {
      "epoch": 0.7399746370413857,
      "grad_norm": 0.14032913744449615,
      "learning_rate": 0.002,
      "loss": 2.7054805755615234,
      "step": 5378
    },
    {
      "epoch": 0.7402498228491362,
      "grad_norm": 0.04280541464686394,
      "learning_rate": 0.002,
      "loss": 2.7217226028442383,
      "step": 5380
    },
    {
      "epoch": 0.7405250086568869,
      "grad_norm": 0.1619100570678711,
      "learning_rate": 0.002,
      "loss": 2.738339424133301,
      "step": 5382
    },
    {
      "epoch": 0.7408001944646375,
      "grad_norm": 0.10214894264936447,
      "learning_rate": 0.002,
      "loss": 2.7753405570983887,
      "step": 5384
    },
    {
      "epoch": 0.7410753802723881,
      "grad_norm": 0.06580190360546112,
      "learning_rate": 0.002,
      "loss": 2.736112117767334,
      "step": 5386
    },
    {
      "epoch": 0.7413505660801387,
      "grad_norm": 0.14319632947444916,
      "learning_rate": 0.002,
      "loss": 2.7211503982543945,
      "step": 5388
    },
    {
      "epoch": 0.7416257518878893,
      "grad_norm": 0.06163400784134865,
      "learning_rate": 0.002,
      "loss": 2.654545783996582,
      "step": 5390
    },
    {
      "epoch": 0.7419009376956399,
      "grad_norm": 0.29435354471206665,
      "learning_rate": 0.002,
      "loss": 2.7408692836761475,
      "step": 5392
    },
    {
      "epoch": 0.7421761235033906,
      "grad_norm": 0.12164409458637238,
      "learning_rate": 0.002,
      "loss": 2.7310190200805664,
      "step": 5394
    },
    {
      "epoch": 0.7424513093111411,
      "grad_norm": 0.16437895596027374,
      "learning_rate": 0.002,
      "loss": 2.7417337894439697,
      "step": 5396
    },
    {
      "epoch": 0.7427264951188918,
      "grad_norm": 0.03179769217967987,
      "learning_rate": 0.002,
      "loss": 2.6838979721069336,
      "step": 5398
    },
    {
      "epoch": 0.7430016809266423,
      "grad_norm": 0.06578903645277023,
      "learning_rate": 0.002,
      "loss": 2.6911919116973877,
      "step": 5400
    },
    {
      "epoch": 0.743276866734393,
      "grad_norm": 0.10786733776330948,
      "learning_rate": 0.002,
      "loss": 2.733405828475952,
      "step": 5402
    },
    {
      "epoch": 0.7435520525421435,
      "grad_norm": 0.058895669877529144,
      "learning_rate": 0.002,
      "loss": 2.71083927154541,
      "step": 5404
    },
    {
      "epoch": 0.7438272383498942,
      "grad_norm": 0.15935808420181274,
      "learning_rate": 0.002,
      "loss": 2.7701754570007324,
      "step": 5406
    },
    {
      "epoch": 0.7441024241576448,
      "grad_norm": 0.1817716360092163,
      "learning_rate": 0.002,
      "loss": 2.726612091064453,
      "step": 5408
    },
    {
      "epoch": 0.7443776099653954,
      "grad_norm": 0.028017381206154823,
      "learning_rate": 0.002,
      "loss": 2.7276463508605957,
      "step": 5410
    },
    {
      "epoch": 0.744652795773146,
      "grad_norm": 0.1031234934926033,
      "learning_rate": 0.002,
      "loss": 2.7451250553131104,
      "step": 5412
    },
    {
      "epoch": 0.7449279815808966,
      "grad_norm": 0.14842116832733154,
      "learning_rate": 0.002,
      "loss": 2.728548765182495,
      "step": 5414
    },
    {
      "epoch": 0.7452031673886472,
      "grad_norm": 0.10160546004772186,
      "learning_rate": 0.002,
      "loss": 2.751969575881958,
      "step": 5416
    },
    {
      "epoch": 0.7454783531963978,
      "grad_norm": 0.05548594892024994,
      "learning_rate": 0.002,
      "loss": 2.7870562076568604,
      "step": 5418
    },
    {
      "epoch": 0.7457535390041484,
      "grad_norm": 0.027089348062872887,
      "learning_rate": 0.002,
      "loss": 2.7700090408325195,
      "step": 5420
    },
    {
      "epoch": 0.7460287248118991,
      "grad_norm": 0.07821185886859894,
      "learning_rate": 0.002,
      "loss": 2.7776498794555664,
      "step": 5422
    },
    {
      "epoch": 0.7463039106196496,
      "grad_norm": 0.05125211924314499,
      "learning_rate": 0.002,
      "loss": 2.740558624267578,
      "step": 5424
    },
    {
      "epoch": 0.7465790964274003,
      "grad_norm": 0.08150587975978851,
      "learning_rate": 0.002,
      "loss": 2.74393892288208,
      "step": 5426
    },
    {
      "epoch": 0.7468542822351508,
      "grad_norm": 0.23151135444641113,
      "learning_rate": 0.002,
      "loss": 2.727950096130371,
      "step": 5428
    },
    {
      "epoch": 0.7471294680429015,
      "grad_norm": 0.3800869286060333,
      "learning_rate": 0.002,
      "loss": 2.7217578887939453,
      "step": 5430
    },
    {
      "epoch": 0.747404653850652,
      "grad_norm": 0.204360693693161,
      "learning_rate": 0.002,
      "loss": 2.7410647869110107,
      "step": 5432
    },
    {
      "epoch": 0.7476798396584027,
      "grad_norm": 0.2200455367565155,
      "learning_rate": 0.002,
      "loss": 2.7536587715148926,
      "step": 5434
    },
    {
      "epoch": 0.7479550254661533,
      "grad_norm": 0.0419032946228981,
      "learning_rate": 0.002,
      "loss": 2.6903927326202393,
      "step": 5436
    },
    {
      "epoch": 0.7482302112739039,
      "grad_norm": 0.048067737370729446,
      "learning_rate": 0.002,
      "loss": 2.772143602371216,
      "step": 5438
    },
    {
      "epoch": 0.7485053970816545,
      "grad_norm": 0.04524460807442665,
      "learning_rate": 0.002,
      "loss": 2.741966962814331,
      "step": 5440
    },
    {
      "epoch": 0.7487805828894051,
      "grad_norm": 0.09834493696689606,
      "learning_rate": 0.002,
      "loss": 2.725099802017212,
      "step": 5442
    },
    {
      "epoch": 0.7490557686971557,
      "grad_norm": 0.05092385411262512,
      "learning_rate": 0.002,
      "loss": 2.7374472618103027,
      "step": 5444
    },
    {
      "epoch": 0.7493309545049064,
      "grad_norm": 0.16609103977680206,
      "learning_rate": 0.002,
      "loss": 2.7272257804870605,
      "step": 5446
    },
    {
      "epoch": 0.7496061403126569,
      "grad_norm": 0.1627616286277771,
      "learning_rate": 0.002,
      "loss": 2.749703884124756,
      "step": 5448
    },
    {
      "epoch": 0.7498813261204076,
      "grad_norm": 0.04973606765270233,
      "learning_rate": 0.002,
      "loss": 2.7090094089508057,
      "step": 5450
    },
    {
      "epoch": 0.7501565119281581,
      "grad_norm": 0.07076851278543472,
      "learning_rate": 0.002,
      "loss": 2.735344886779785,
      "step": 5452
    },
    {
      "epoch": 0.7504316977359088,
      "grad_norm": 0.028438666835427284,
      "learning_rate": 0.002,
      "loss": 2.758849620819092,
      "step": 5454
    },
    {
      "epoch": 0.7507068835436593,
      "grad_norm": 0.18451164662837982,
      "learning_rate": 0.002,
      "loss": 2.7302162647247314,
      "step": 5456
    },
    {
      "epoch": 0.75098206935141,
      "grad_norm": 0.4610634744167328,
      "learning_rate": 0.002,
      "loss": 2.687952995300293,
      "step": 5458
    },
    {
      "epoch": 0.7512572551591606,
      "grad_norm": 0.04800763353705406,
      "learning_rate": 0.002,
      "loss": 2.7274837493896484,
      "step": 5460
    },
    {
      "epoch": 0.7515324409669112,
      "grad_norm": 0.3402794599533081,
      "learning_rate": 0.002,
      "loss": 2.6969707012176514,
      "step": 5462
    },
    {
      "epoch": 0.7518076267746618,
      "grad_norm": 0.16740486025810242,
      "learning_rate": 0.002,
      "loss": 2.7366831302642822,
      "step": 5464
    },
    {
      "epoch": 0.7520828125824124,
      "grad_norm": 0.1253676861524582,
      "learning_rate": 0.002,
      "loss": 2.7386245727539062,
      "step": 5466
    },
    {
      "epoch": 0.752357998390163,
      "grad_norm": 0.23277564346790314,
      "learning_rate": 0.002,
      "loss": 2.703867197036743,
      "step": 5468
    },
    {
      "epoch": 0.7526331841979136,
      "grad_norm": 0.13670942187309265,
      "learning_rate": 0.002,
      "loss": 2.7454919815063477,
      "step": 5470
    },
    {
      "epoch": 0.7529083700056642,
      "grad_norm": 0.17384925484657288,
      "learning_rate": 0.002,
      "loss": 2.7347028255462646,
      "step": 5472
    },
    {
      "epoch": 0.7531835558134149,
      "grad_norm": 0.19846388697624207,
      "learning_rate": 0.002,
      "loss": 2.7580347061157227,
      "step": 5474
    },
    {
      "epoch": 0.7534587416211654,
      "grad_norm": 0.10073129832744598,
      "learning_rate": 0.002,
      "loss": 2.7295961380004883,
      "step": 5476
    },
    {
      "epoch": 0.7537339274289161,
      "grad_norm": 0.10150282829999924,
      "learning_rate": 0.002,
      "loss": 2.7198753356933594,
      "step": 5478
    },
    {
      "epoch": 0.7540091132366666,
      "grad_norm": 0.1624523401260376,
      "learning_rate": 0.002,
      "loss": 2.744389533996582,
      "step": 5480
    },
    {
      "epoch": 0.7542842990444173,
      "grad_norm": 0.05159242823719978,
      "learning_rate": 0.002,
      "loss": 2.7190327644348145,
      "step": 5482
    },
    {
      "epoch": 0.7545594848521678,
      "grad_norm": 0.27621522545814514,
      "learning_rate": 0.002,
      "loss": 2.764462471008301,
      "step": 5484
    },
    {
      "epoch": 0.7548346706599185,
      "grad_norm": 0.0548023022711277,
      "learning_rate": 0.002,
      "loss": 2.726566791534424,
      "step": 5486
    },
    {
      "epoch": 0.7551098564676691,
      "grad_norm": 0.16495871543884277,
      "learning_rate": 0.002,
      "loss": 2.707754373550415,
      "step": 5488
    },
    {
      "epoch": 0.7553850422754197,
      "grad_norm": 0.1307213008403778,
      "learning_rate": 0.002,
      "loss": 2.6936676502227783,
      "step": 5490
    },
    {
      "epoch": 0.7556602280831703,
      "grad_norm": 0.21619193255901337,
      "learning_rate": 0.002,
      "loss": 2.702791929244995,
      "step": 5492
    },
    {
      "epoch": 0.7559354138909209,
      "grad_norm": 0.21340064704418182,
      "learning_rate": 0.002,
      "loss": 2.7021031379699707,
      "step": 5494
    },
    {
      "epoch": 0.7562105996986715,
      "grad_norm": 0.0894065648317337,
      "learning_rate": 0.002,
      "loss": 2.7191290855407715,
      "step": 5496
    },
    {
      "epoch": 0.7564857855064222,
      "grad_norm": 0.15556621551513672,
      "learning_rate": 0.002,
      "loss": 2.742288112640381,
      "step": 5498
    },
    {
      "epoch": 0.7567609713141727,
      "grad_norm": 0.11494958400726318,
      "learning_rate": 0.002,
      "loss": 2.748063564300537,
      "step": 5500
    },
    {
      "epoch": 0.7570361571219234,
      "grad_norm": 0.18676093220710754,
      "learning_rate": 0.002,
      "loss": 2.718191146850586,
      "step": 5502
    },
    {
      "epoch": 0.757311342929674,
      "grad_norm": 0.21073105931282043,
      "learning_rate": 0.002,
      "loss": 2.7056918144226074,
      "step": 5504
    },
    {
      "epoch": 0.7575865287374246,
      "grad_norm": 0.12686996161937714,
      "learning_rate": 0.002,
      "loss": 2.7453012466430664,
      "step": 5506
    },
    {
      "epoch": 0.7578617145451751,
      "grad_norm": 0.18561969697475433,
      "learning_rate": 0.002,
      "loss": 2.7210540771484375,
      "step": 5508
    },
    {
      "epoch": 0.7581369003529258,
      "grad_norm": 0.04741664603352547,
      "learning_rate": 0.002,
      "loss": 2.7243502140045166,
      "step": 5510
    },
    {
      "epoch": 0.7584120861606765,
      "grad_norm": 0.14863817393779755,
      "learning_rate": 0.002,
      "loss": 2.733799457550049,
      "step": 5512
    },
    {
      "epoch": 0.758687271968427,
      "grad_norm": 0.08175837248563766,
      "learning_rate": 0.002,
      "loss": 2.7007460594177246,
      "step": 5514
    },
    {
      "epoch": 0.7589624577761777,
      "grad_norm": 0.11436764895915985,
      "learning_rate": 0.002,
      "loss": 2.74221134185791,
      "step": 5516
    },
    {
      "epoch": 0.7592376435839282,
      "grad_norm": 0.10654667019844055,
      "learning_rate": 0.002,
      "loss": 2.6965668201446533,
      "step": 5518
    },
    {
      "epoch": 0.7595128293916789,
      "grad_norm": 0.2870754599571228,
      "learning_rate": 0.002,
      "loss": 2.7392632961273193,
      "step": 5520
    },
    {
      "epoch": 0.7597880151994294,
      "grad_norm": 0.19685041904449463,
      "learning_rate": 0.002,
      "loss": 2.715751886367798,
      "step": 5522
    },
    {
      "epoch": 0.7600632010071801,
      "grad_norm": 0.0862874686717987,
      "learning_rate": 0.002,
      "loss": 2.7555723190307617,
      "step": 5524
    },
    {
      "epoch": 0.7603383868149307,
      "grad_norm": 0.17575864493846893,
      "learning_rate": 0.002,
      "loss": 2.7459049224853516,
      "step": 5526
    },
    {
      "epoch": 0.7606135726226813,
      "grad_norm": 0.03414633497595787,
      "learning_rate": 0.002,
      "loss": 2.672846555709839,
      "step": 5528
    },
    {
      "epoch": 0.7608887584304319,
      "grad_norm": 0.12212257087230682,
      "learning_rate": 0.002,
      "loss": 2.722029447555542,
      "step": 5530
    },
    {
      "epoch": 0.7611639442381825,
      "grad_norm": 0.04557446017861366,
      "learning_rate": 0.002,
      "loss": 2.692021369934082,
      "step": 5532
    },
    {
      "epoch": 0.7614391300459331,
      "grad_norm": 0.11097995936870575,
      "learning_rate": 0.002,
      "loss": 2.741345167160034,
      "step": 5534
    },
    {
      "epoch": 0.7617143158536837,
      "grad_norm": 0.037585288286209106,
      "learning_rate": 0.002,
      "loss": 2.6967928409576416,
      "step": 5536
    },
    {
      "epoch": 0.7619895016614343,
      "grad_norm": 0.1752534657716751,
      "learning_rate": 0.002,
      "loss": 2.7249157428741455,
      "step": 5538
    },
    {
      "epoch": 0.762264687469185,
      "grad_norm": 0.12276816368103027,
      "learning_rate": 0.002,
      "loss": 2.7336478233337402,
      "step": 5540
    },
    {
      "epoch": 0.7625398732769355,
      "grad_norm": 0.04509880021214485,
      "learning_rate": 0.002,
      "loss": 2.728102207183838,
      "step": 5542
    },
    {
      "epoch": 0.7628150590846862,
      "grad_norm": 0.1317267119884491,
      "learning_rate": 0.002,
      "loss": 2.7524523735046387,
      "step": 5544
    },
    {
      "epoch": 0.7630902448924367,
      "grad_norm": 0.17402943968772888,
      "learning_rate": 0.002,
      "loss": 2.7299423217773438,
      "step": 5546
    },
    {
      "epoch": 0.7633654307001874,
      "grad_norm": 0.17454741895198822,
      "learning_rate": 0.002,
      "loss": 2.7462801933288574,
      "step": 5548
    },
    {
      "epoch": 0.763640616507938,
      "grad_norm": 0.08640190958976746,
      "learning_rate": 0.002,
      "loss": 2.7444686889648438,
      "step": 5550
    },
    {
      "epoch": 0.7639158023156886,
      "grad_norm": 0.27382421493530273,
      "learning_rate": 0.002,
      "loss": 2.70570969581604,
      "step": 5552
    },
    {
      "epoch": 0.7641909881234392,
      "grad_norm": 0.07072192430496216,
      "learning_rate": 0.002,
      "loss": 2.7045302391052246,
      "step": 5554
    },
    {
      "epoch": 0.7644661739311898,
      "grad_norm": 0.28320661187171936,
      "learning_rate": 0.002,
      "loss": 2.7307419776916504,
      "step": 5556
    },
    {
      "epoch": 0.7647413597389404,
      "grad_norm": 0.08107183128595352,
      "learning_rate": 0.002,
      "loss": 2.7272133827209473,
      "step": 5558
    },
    {
      "epoch": 0.765016545546691,
      "grad_norm": 0.1600152999162674,
      "learning_rate": 0.002,
      "loss": 2.743739128112793,
      "step": 5560
    },
    {
      "epoch": 0.7652917313544416,
      "grad_norm": 0.07756832242012024,
      "learning_rate": 0.002,
      "loss": 2.6937639713287354,
      "step": 5562
    },
    {
      "epoch": 0.7655669171621923,
      "grad_norm": 0.19187501072883606,
      "learning_rate": 0.002,
      "loss": 2.707102060317993,
      "step": 5564
    },
    {
      "epoch": 0.7658421029699428,
      "grad_norm": 0.18116042017936707,
      "learning_rate": 0.002,
      "loss": 2.697570323944092,
      "step": 5566
    },
    {
      "epoch": 0.7661172887776935,
      "grad_norm": 0.19166098535060883,
      "learning_rate": 0.002,
      "loss": 2.7013957500457764,
      "step": 5568
    },
    {
      "epoch": 0.766392474585444,
      "grad_norm": 0.11441538482904434,
      "learning_rate": 0.002,
      "loss": 2.718090057373047,
      "step": 5570
    },
    {
      "epoch": 0.7666676603931947,
      "grad_norm": 0.06639756262302399,
      "learning_rate": 0.002,
      "loss": 2.720088005065918,
      "step": 5572
    },
    {
      "epoch": 0.7669428462009452,
      "grad_norm": 0.08775195479393005,
      "learning_rate": 0.002,
      "loss": 2.725778341293335,
      "step": 5574
    },
    {
      "epoch": 0.7672180320086959,
      "grad_norm": 0.047604650259017944,
      "learning_rate": 0.002,
      "loss": 2.69064998626709,
      "step": 5576
    },
    {
      "epoch": 0.7674932178164465,
      "grad_norm": 0.09258148819208145,
      "learning_rate": 0.002,
      "loss": 2.729259490966797,
      "step": 5578
    },
    {
      "epoch": 0.7677684036241971,
      "grad_norm": 0.19014503061771393,
      "learning_rate": 0.002,
      "loss": 2.736876964569092,
      "step": 5580
    },
    {
      "epoch": 0.7680435894319477,
      "grad_norm": 0.10664612799882889,
      "learning_rate": 0.002,
      "loss": 2.705259084701538,
      "step": 5582
    },
    {
      "epoch": 0.7683187752396983,
      "grad_norm": 0.13453032076358795,
      "learning_rate": 0.002,
      "loss": 2.700377941131592,
      "step": 5584
    },
    {
      "epoch": 0.7685939610474489,
      "grad_norm": 0.19300080835819244,
      "learning_rate": 0.002,
      "loss": 2.7104389667510986,
      "step": 5586
    },
    {
      "epoch": 0.7688691468551995,
      "grad_norm": 0.11661265045404434,
      "learning_rate": 0.002,
      "loss": 2.679213523864746,
      "step": 5588
    },
    {
      "epoch": 0.7691443326629501,
      "grad_norm": 0.12858343124389648,
      "learning_rate": 0.002,
      "loss": 2.6852149963378906,
      "step": 5590
    },
    {
      "epoch": 0.7694195184707008,
      "grad_norm": 0.27021917700767517,
      "learning_rate": 0.002,
      "loss": 2.703125,
      "step": 5592
    },
    {
      "epoch": 0.7696947042784513,
      "grad_norm": 0.1325863003730774,
      "learning_rate": 0.002,
      "loss": 2.7241759300231934,
      "step": 5594
    },
    {
      "epoch": 0.769969890086202,
      "grad_norm": 0.17928768694400787,
      "learning_rate": 0.002,
      "loss": 2.7310144901275635,
      "step": 5596
    },
    {
      "epoch": 0.7702450758939525,
      "grad_norm": 0.07317907363176346,
      "learning_rate": 0.002,
      "loss": 2.7029926776885986,
      "step": 5598
    },
    {
      "epoch": 0.7705202617017032,
      "grad_norm": 0.20460347831249237,
      "learning_rate": 0.002,
      "loss": 2.7044992446899414,
      "step": 5600
    },
    {
      "epoch": 0.7707954475094537,
      "grad_norm": 0.10513727366924286,
      "learning_rate": 0.002,
      "loss": 2.7390308380126953,
      "step": 5602
    },
    {
      "epoch": 0.7710706333172044,
      "grad_norm": 0.26105260848999023,
      "learning_rate": 0.002,
      "loss": 2.700479507446289,
      "step": 5604
    },
    {
      "epoch": 0.771345819124955,
      "grad_norm": 0.07503360509872437,
      "learning_rate": 0.002,
      "loss": 2.740171432495117,
      "step": 5606
    },
    {
      "epoch": 0.7716210049327056,
      "grad_norm": 0.19845275580883026,
      "learning_rate": 0.002,
      "loss": 2.6755988597869873,
      "step": 5608
    },
    {
      "epoch": 0.7718961907404562,
      "grad_norm": 0.14130935072898865,
      "learning_rate": 0.002,
      "loss": 2.709566593170166,
      "step": 5610
    },
    {
      "epoch": 0.7721713765482068,
      "grad_norm": 0.2132335603237152,
      "learning_rate": 0.002,
      "loss": 2.7388291358947754,
      "step": 5612
    },
    {
      "epoch": 0.7724465623559574,
      "grad_norm": 0.17650462687015533,
      "learning_rate": 0.002,
      "loss": 2.768336534500122,
      "step": 5614
    },
    {
      "epoch": 0.7727217481637081,
      "grad_norm": 0.19512079656124115,
      "learning_rate": 0.002,
      "loss": 2.7116239070892334,
      "step": 5616
    },
    {
      "epoch": 0.7729969339714586,
      "grad_norm": 0.05258332937955856,
      "learning_rate": 0.002,
      "loss": 2.7216663360595703,
      "step": 5618
    },
    {
      "epoch": 0.7732721197792093,
      "grad_norm": 0.3276945650577545,
      "learning_rate": 0.002,
      "loss": 2.72248911857605,
      "step": 5620
    },
    {
      "epoch": 0.7735473055869598,
      "grad_norm": 0.07767099142074585,
      "learning_rate": 0.002,
      "loss": 2.7237603664398193,
      "step": 5622
    },
    {
      "epoch": 0.7738224913947105,
      "grad_norm": 0.17710773646831512,
      "learning_rate": 0.002,
      "loss": 2.6845455169677734,
      "step": 5624
    },
    {
      "epoch": 0.774097677202461,
      "grad_norm": 0.2734055519104004,
      "learning_rate": 0.002,
      "loss": 2.7148799896240234,
      "step": 5626
    },
    {
      "epoch": 0.7743728630102117,
      "grad_norm": 0.02834184467792511,
      "learning_rate": 0.002,
      "loss": 2.665785312652588,
      "step": 5628
    },
    {
      "epoch": 0.7746480488179623,
      "grad_norm": 0.17327633500099182,
      "learning_rate": 0.002,
      "loss": 2.7355546951293945,
      "step": 5630
    },
    {
      "epoch": 0.7749232346257129,
      "grad_norm": 0.046394042670726776,
      "learning_rate": 0.002,
      "loss": 2.7427377700805664,
      "step": 5632
    },
    {
      "epoch": 0.7751984204334635,
      "grad_norm": 0.036860182881355286,
      "learning_rate": 0.002,
      "loss": 2.730928897857666,
      "step": 5634
    },
    {
      "epoch": 0.7754736062412141,
      "grad_norm": 0.08738909661769867,
      "learning_rate": 0.002,
      "loss": 2.738088607788086,
      "step": 5636
    },
    {
      "epoch": 0.7757487920489647,
      "grad_norm": 0.16459223628044128,
      "learning_rate": 0.002,
      "loss": 2.753356456756592,
      "step": 5638
    },
    {
      "epoch": 0.7760239778567153,
      "grad_norm": 0.15357846021652222,
      "learning_rate": 0.002,
      "loss": 2.696227550506592,
      "step": 5640
    },
    {
      "epoch": 0.7762991636644659,
      "grad_norm": 0.2236928641796112,
      "learning_rate": 0.002,
      "loss": 2.756168842315674,
      "step": 5642
    },
    {
      "epoch": 0.7765743494722166,
      "grad_norm": 0.09823135286569595,
      "learning_rate": 0.002,
      "loss": 2.730968475341797,
      "step": 5644
    },
    {
      "epoch": 0.7768495352799671,
      "grad_norm": 0.1391511857509613,
      "learning_rate": 0.002,
      "loss": 2.6942033767700195,
      "step": 5646
    },
    {
      "epoch": 0.7771247210877178,
      "grad_norm": 0.04678016155958176,
      "learning_rate": 0.002,
      "loss": 2.6742844581604004,
      "step": 5648
    },
    {
      "epoch": 0.7773999068954683,
      "grad_norm": 0.09164540469646454,
      "learning_rate": 0.002,
      "loss": 2.77140736579895,
      "step": 5650
    },
    {
      "epoch": 0.777675092703219,
      "grad_norm": 0.05971049889922142,
      "learning_rate": 0.002,
      "loss": 2.691042900085449,
      "step": 5652
    },
    {
      "epoch": 0.7779502785109695,
      "grad_norm": 0.16434916853904724,
      "learning_rate": 0.002,
      "loss": 2.7083706855773926,
      "step": 5654
    },
    {
      "epoch": 0.7782254643187202,
      "grad_norm": 0.264715313911438,
      "learning_rate": 0.002,
      "loss": 2.744114875793457,
      "step": 5656
    },
    {
      "epoch": 0.7785006501264709,
      "grad_norm": 0.057643260806798935,
      "learning_rate": 0.002,
      "loss": 2.7253026962280273,
      "step": 5658
    },
    {
      "epoch": 0.7787758359342214,
      "grad_norm": 0.23285283148288727,
      "learning_rate": 0.002,
      "loss": 2.7370471954345703,
      "step": 5660
    },
    {
      "epoch": 0.779051021741972,
      "grad_norm": 0.11223302781581879,
      "learning_rate": 0.002,
      "loss": 2.7261741161346436,
      "step": 5662
    },
    {
      "epoch": 0.7793262075497226,
      "grad_norm": 0.20074081420898438,
      "learning_rate": 0.002,
      "loss": 2.674323797225952,
      "step": 5664
    },
    {
      "epoch": 0.7796013933574732,
      "grad_norm": 0.1730562150478363,
      "learning_rate": 0.002,
      "loss": 2.673499345779419,
      "step": 5666
    },
    {
      "epoch": 0.7798765791652239,
      "grad_norm": 0.18366743624210358,
      "learning_rate": 0.002,
      "loss": 2.673468589782715,
      "step": 5668
    },
    {
      "epoch": 0.7801517649729744,
      "grad_norm": 0.07575666904449463,
      "learning_rate": 0.002,
      "loss": 2.7085623741149902,
      "step": 5670
    },
    {
      "epoch": 0.7801517649729744,
      "eval_loss": 2.698002576828003,
      "eval_runtime": 380.0392,
      "eval_samples_per_second": 92.535,
      "eval_steps_per_second": 11.567,
      "step": 5670
    },
    {
      "epoch": 0.7804269507807251,
      "grad_norm": 0.2191534787416458,
      "learning_rate": 0.002,
      "loss": 2.7085471153259277,
      "step": 5672
    },
    {
      "epoch": 0.7807021365884756,
      "grad_norm": 0.14512953162193298,
      "learning_rate": 0.002,
      "loss": 2.6995809078216553,
      "step": 5674
    },
    {
      "epoch": 0.7809773223962263,
      "grad_norm": 0.33041366934776306,
      "learning_rate": 0.002,
      "loss": 2.6803789138793945,
      "step": 5676
    },
    {
      "epoch": 0.7812525082039768,
      "grad_norm": 0.12169376015663147,
      "learning_rate": 0.002,
      "loss": 2.7043867111206055,
      "step": 5678
    },
    {
      "epoch": 0.7815276940117275,
      "grad_norm": 0.2236432582139969,
      "learning_rate": 0.002,
      "loss": 2.7410812377929688,
      "step": 5680
    },
    {
      "epoch": 0.7818028798194782,
      "grad_norm": 0.0330001637339592,
      "learning_rate": 0.002,
      "loss": 2.7053980827331543,
      "step": 5682
    },
    {
      "epoch": 0.7820780656272287,
      "grad_norm": 0.07267382740974426,
      "learning_rate": 0.002,
      "loss": 2.700737953186035,
      "step": 5684
    },
    {
      "epoch": 0.7823532514349794,
      "grad_norm": 0.18797567486763,
      "learning_rate": 0.002,
      "loss": 2.7119550704956055,
      "step": 5686
    },
    {
      "epoch": 0.7826284372427299,
      "grad_norm": 0.072520911693573,
      "learning_rate": 0.002,
      "loss": 2.744596004486084,
      "step": 5688
    },
    {
      "epoch": 0.7829036230504806,
      "grad_norm": 0.11477525532245636,
      "learning_rate": 0.002,
      "loss": 2.7130818367004395,
      "step": 5690
    },
    {
      "epoch": 0.7831788088582311,
      "grad_norm": 0.16979697346687317,
      "learning_rate": 0.002,
      "loss": 2.7225255966186523,
      "step": 5692
    },
    {
      "epoch": 0.7834539946659818,
      "grad_norm": 0.11643542349338531,
      "learning_rate": 0.002,
      "loss": 2.7306649684906006,
      "step": 5694
    },
    {
      "epoch": 0.7837291804737324,
      "grad_norm": 0.04787943884730339,
      "learning_rate": 0.002,
      "loss": 2.7343432903289795,
      "step": 5696
    },
    {
      "epoch": 0.784004366281483,
      "grad_norm": 0.24024781584739685,
      "learning_rate": 0.002,
      "loss": 2.6583304405212402,
      "step": 5698
    },
    {
      "epoch": 0.7842795520892336,
      "grad_norm": 0.16668958961963654,
      "learning_rate": 0.002,
      "loss": 2.748293399810791,
      "step": 5700
    },
    {
      "epoch": 0.7845547378969842,
      "grad_norm": 0.3092277944087982,
      "learning_rate": 0.002,
      "loss": 2.7552309036254883,
      "step": 5702
    },
    {
      "epoch": 0.7848299237047348,
      "grad_norm": 0.1092662364244461,
      "learning_rate": 0.002,
      "loss": 2.695739269256592,
      "step": 5704
    },
    {
      "epoch": 0.7851051095124854,
      "grad_norm": 0.2882915735244751,
      "learning_rate": 0.002,
      "loss": 2.6874396800994873,
      "step": 5706
    },
    {
      "epoch": 0.785380295320236,
      "grad_norm": 0.034257933497428894,
      "learning_rate": 0.002,
      "loss": 2.622493267059326,
      "step": 5708
    },
    {
      "epoch": 0.7856554811279867,
      "grad_norm": 0.1366323083639145,
      "learning_rate": 0.002,
      "loss": 2.7075111865997314,
      "step": 5710
    },
    {
      "epoch": 0.7859306669357372,
      "grad_norm": 0.04368678480386734,
      "learning_rate": 0.002,
      "loss": 2.681715965270996,
      "step": 5712
    },
    {
      "epoch": 0.7862058527434879,
      "grad_norm": 0.06357915699481964,
      "learning_rate": 0.002,
      "loss": 2.711022138595581,
      "step": 5714
    },
    {
      "epoch": 0.7864810385512384,
      "grad_norm": 0.028752321377396584,
      "learning_rate": 0.002,
      "loss": 2.673764228820801,
      "step": 5716
    },
    {
      "epoch": 0.7867562243589891,
      "grad_norm": 0.10737695544958115,
      "learning_rate": 0.002,
      "loss": 2.70995831489563,
      "step": 5718
    },
    {
      "epoch": 0.7870314101667397,
      "grad_norm": 0.0929102748632431,
      "learning_rate": 0.002,
      "loss": 2.6954948902130127,
      "step": 5720
    },
    {
      "epoch": 0.7873065959744903,
      "grad_norm": 0.2844030261039734,
      "learning_rate": 0.002,
      "loss": 2.745307445526123,
      "step": 5722
    },
    {
      "epoch": 0.7875817817822409,
      "grad_norm": 0.13635708391666412,
      "learning_rate": 0.002,
      "loss": 2.7006680965423584,
      "step": 5724
    },
    {
      "epoch": 0.7878569675899915,
      "grad_norm": 0.14232473075389862,
      "learning_rate": 0.002,
      "loss": 2.7103281021118164,
      "step": 5726
    },
    {
      "epoch": 0.7881321533977421,
      "grad_norm": 0.25027984380722046,
      "learning_rate": 0.002,
      "loss": 2.7261550426483154,
      "step": 5728
    },
    {
      "epoch": 0.7884073392054927,
      "grad_norm": 0.03978078439831734,
      "learning_rate": 0.002,
      "loss": 2.682363510131836,
      "step": 5730
    },
    {
      "epoch": 0.7886825250132433,
      "grad_norm": 0.3243900239467621,
      "learning_rate": 0.002,
      "loss": 2.747607707977295,
      "step": 5732
    },
    {
      "epoch": 0.788957710820994,
      "grad_norm": 0.08686048537492752,
      "learning_rate": 0.002,
      "loss": 2.732344388961792,
      "step": 5734
    },
    {
      "epoch": 0.7892328966287445,
      "grad_norm": 0.19950959086418152,
      "learning_rate": 0.002,
      "loss": 2.6976795196533203,
      "step": 5736
    },
    {
      "epoch": 0.7895080824364952,
      "grad_norm": 0.0785156637430191,
      "learning_rate": 0.002,
      "loss": 2.6959738731384277,
      "step": 5738
    },
    {
      "epoch": 0.7897832682442457,
      "grad_norm": 0.06630580872297287,
      "learning_rate": 0.002,
      "loss": 2.687284469604492,
      "step": 5740
    },
    {
      "epoch": 0.7900584540519964,
      "grad_norm": 0.11704078316688538,
      "learning_rate": 0.002,
      "loss": 2.688605308532715,
      "step": 5742
    },
    {
      "epoch": 0.7903336398597469,
      "grad_norm": 0.07545988261699677,
      "learning_rate": 0.002,
      "loss": 2.7152018547058105,
      "step": 5744
    },
    {
      "epoch": 0.7906088256674976,
      "grad_norm": 0.054600317031145096,
      "learning_rate": 0.002,
      "loss": 2.7398509979248047,
      "step": 5746
    },
    {
      "epoch": 0.7908840114752482,
      "grad_norm": 0.039238762110471725,
      "learning_rate": 0.002,
      "loss": 2.6501660346984863,
      "step": 5748
    },
    {
      "epoch": 0.7911591972829988,
      "grad_norm": 0.027897346764802933,
      "learning_rate": 0.002,
      "loss": 2.725125312805176,
      "step": 5750
    },
    {
      "epoch": 0.7914343830907494,
      "grad_norm": 0.06787458807229996,
      "learning_rate": 0.002,
      "loss": 2.722201347351074,
      "step": 5752
    },
    {
      "epoch": 0.7917095688985,
      "grad_norm": 0.09558141976594925,
      "learning_rate": 0.002,
      "loss": 2.718432664871216,
      "step": 5754
    },
    {
      "epoch": 0.7919847547062506,
      "grad_norm": 0.08302492648363113,
      "learning_rate": 0.002,
      "loss": 2.7101523876190186,
      "step": 5756
    },
    {
      "epoch": 0.7922599405140012,
      "grad_norm": 0.1173919066786766,
      "learning_rate": 0.002,
      "loss": 2.694917678833008,
      "step": 5758
    },
    {
      "epoch": 0.7925351263217518,
      "grad_norm": 0.21389082074165344,
      "learning_rate": 0.002,
      "loss": 2.7386391162872314,
      "step": 5760
    },
    {
      "epoch": 0.7928103121295025,
      "grad_norm": 0.24807006120681763,
      "learning_rate": 0.002,
      "loss": 2.719207763671875,
      "step": 5762
    },
    {
      "epoch": 0.793085497937253,
      "grad_norm": 0.14053063094615936,
      "learning_rate": 0.002,
      "loss": 2.7176530361175537,
      "step": 5764
    },
    {
      "epoch": 0.7933606837450037,
      "grad_norm": 0.2467871606349945,
      "learning_rate": 0.002,
      "loss": 2.7157888412475586,
      "step": 5766
    },
    {
      "epoch": 0.7936358695527542,
      "grad_norm": 0.20187383890151978,
      "learning_rate": 0.002,
      "loss": 2.7215769290924072,
      "step": 5768
    },
    {
      "epoch": 0.7939110553605049,
      "grad_norm": 0.26615530252456665,
      "learning_rate": 0.002,
      "loss": 2.7003166675567627,
      "step": 5770
    },
    {
      "epoch": 0.7941862411682555,
      "grad_norm": 0.05804391950368881,
      "learning_rate": 0.002,
      "loss": 2.691084384918213,
      "step": 5772
    },
    {
      "epoch": 0.7944614269760061,
      "grad_norm": 0.2844776213169098,
      "learning_rate": 0.002,
      "loss": 2.700500011444092,
      "step": 5774
    },
    {
      "epoch": 0.7947366127837567,
      "grad_norm": 0.09139382094144821,
      "learning_rate": 0.002,
      "loss": 2.7101588249206543,
      "step": 5776
    },
    {
      "epoch": 0.7950117985915073,
      "grad_norm": 0.12606045603752136,
      "learning_rate": 0.002,
      "loss": 2.728827476501465,
      "step": 5778
    },
    {
      "epoch": 0.7952869843992579,
      "grad_norm": 0.16468952596187592,
      "learning_rate": 0.002,
      "loss": 2.73936128616333,
      "step": 5780
    },
    {
      "epoch": 0.7955621702070085,
      "grad_norm": 0.05048297718167305,
      "learning_rate": 0.002,
      "loss": 2.729562282562256,
      "step": 5782
    },
    {
      "epoch": 0.7958373560147591,
      "grad_norm": 0.14686720073223114,
      "learning_rate": 0.002,
      "loss": 2.7048799991607666,
      "step": 5784
    },
    {
      "epoch": 0.7961125418225098,
      "grad_norm": 0.034180834889411926,
      "learning_rate": 0.002,
      "loss": 2.7205889225006104,
      "step": 5786
    },
    {
      "epoch": 0.7963877276302603,
      "grad_norm": 0.20239350199699402,
      "learning_rate": 0.002,
      "loss": 2.6615138053894043,
      "step": 5788
    },
    {
      "epoch": 0.796662913438011,
      "grad_norm": 0.1745232343673706,
      "learning_rate": 0.002,
      "loss": 2.7098944187164307,
      "step": 5790
    },
    {
      "epoch": 0.7969380992457615,
      "grad_norm": 0.10829172283411026,
      "learning_rate": 0.002,
      "loss": 2.7067363262176514,
      "step": 5792
    },
    {
      "epoch": 0.7972132850535122,
      "grad_norm": 0.11411265283823013,
      "learning_rate": 0.002,
      "loss": 2.6799731254577637,
      "step": 5794
    },
    {
      "epoch": 0.7974884708612627,
      "grad_norm": 0.06804240494966507,
      "learning_rate": 0.002,
      "loss": 2.664473056793213,
      "step": 5796
    },
    {
      "epoch": 0.7977636566690134,
      "grad_norm": 0.10480640828609467,
      "learning_rate": 0.002,
      "loss": 2.6949856281280518,
      "step": 5798
    },
    {
      "epoch": 0.798038842476764,
      "grad_norm": 0.19172558188438416,
      "learning_rate": 0.002,
      "loss": 2.7188844680786133,
      "step": 5800
    },
    {
      "epoch": 0.7983140282845146,
      "grad_norm": 0.0312957763671875,
      "learning_rate": 0.002,
      "loss": 2.725066661834717,
      "step": 5802
    },
    {
      "epoch": 0.7985892140922652,
      "grad_norm": 0.1294817328453064,
      "learning_rate": 0.002,
      "loss": 2.7210845947265625,
      "step": 5804
    },
    {
      "epoch": 0.7988643999000158,
      "grad_norm": 0.29513880610466003,
      "learning_rate": 0.002,
      "loss": 2.6946089267730713,
      "step": 5806
    },
    {
      "epoch": 0.7991395857077664,
      "grad_norm": 0.061328575015068054,
      "learning_rate": 0.002,
      "loss": 2.7116918563842773,
      "step": 5808
    },
    {
      "epoch": 0.799414771515517,
      "grad_norm": 0.227827250957489,
      "learning_rate": 0.002,
      "loss": 2.7353811264038086,
      "step": 5810
    },
    {
      "epoch": 0.7996899573232676,
      "grad_norm": 0.20561587810516357,
      "learning_rate": 0.002,
      "loss": 2.6800289154052734,
      "step": 5812
    },
    {
      "epoch": 0.7999651431310183,
      "grad_norm": 0.11749380081892014,
      "learning_rate": 0.002,
      "loss": 2.6833629608154297,
      "step": 5814
    },
    {
      "epoch": 0.8002403289387688,
      "grad_norm": 0.19437435269355774,
      "learning_rate": 0.002,
      "loss": 2.719148635864258,
      "step": 5816
    },
    {
      "epoch": 0.8005155147465195,
      "grad_norm": 0.08396140486001968,
      "learning_rate": 0.002,
      "loss": 2.713017702102661,
      "step": 5818
    },
    {
      "epoch": 0.80079070055427,
      "grad_norm": 0.1685422956943512,
      "learning_rate": 0.002,
      "loss": 2.7442805767059326,
      "step": 5820
    },
    {
      "epoch": 0.8010658863620207,
      "grad_norm": 0.06974931061267853,
      "learning_rate": 0.002,
      "loss": 2.701531410217285,
      "step": 5822
    },
    {
      "epoch": 0.8013410721697714,
      "grad_norm": 0.1716153919696808,
      "learning_rate": 0.002,
      "loss": 2.695272922515869,
      "step": 5824
    },
    {
      "epoch": 0.8016162579775219,
      "grad_norm": 0.16187764704227448,
      "learning_rate": 0.002,
      "loss": 2.740889549255371,
      "step": 5826
    },
    {
      "epoch": 0.8018914437852726,
      "grad_norm": 0.24644595384597778,
      "learning_rate": 0.002,
      "loss": 2.731771945953369,
      "step": 5828
    },
    {
      "epoch": 0.8021666295930231,
      "grad_norm": 0.10488201677799225,
      "learning_rate": 0.002,
      "loss": 2.7414965629577637,
      "step": 5830
    },
    {
      "epoch": 0.8024418154007738,
      "grad_norm": 0.08634066581726074,
      "learning_rate": 0.002,
      "loss": 2.7146081924438477,
      "step": 5832
    },
    {
      "epoch": 0.8027170012085243,
      "grad_norm": 0.18742242455482483,
      "learning_rate": 0.002,
      "loss": 2.713186740875244,
      "step": 5834
    },
    {
      "epoch": 0.802992187016275,
      "grad_norm": 0.1605006456375122,
      "learning_rate": 0.002,
      "loss": 2.7121753692626953,
      "step": 5836
    },
    {
      "epoch": 0.8032673728240256,
      "grad_norm": 0.03147124871611595,
      "learning_rate": 0.002,
      "loss": 2.691865921020508,
      "step": 5838
    },
    {
      "epoch": 0.8035425586317761,
      "grad_norm": 0.27511224150657654,
      "learning_rate": 0.002,
      "loss": 2.708163261413574,
      "step": 5840
    },
    {
      "epoch": 0.8038177444395268,
      "grad_norm": 0.03452141955494881,
      "learning_rate": 0.002,
      "loss": 2.709047317504883,
      "step": 5842
    },
    {
      "epoch": 0.8040929302472773,
      "grad_norm": 0.33932235836982727,
      "learning_rate": 0.002,
      "loss": 2.670644998550415,
      "step": 5844
    },
    {
      "epoch": 0.804368116055028,
      "grad_norm": 0.0281843189150095,
      "learning_rate": 0.002,
      "loss": 2.691800594329834,
      "step": 5846
    },
    {
      "epoch": 0.8046433018627785,
      "grad_norm": 0.23356053233146667,
      "learning_rate": 0.002,
      "loss": 2.7084991931915283,
      "step": 5848
    },
    {
      "epoch": 0.8049184876705292,
      "grad_norm": 0.27555227279663086,
      "learning_rate": 0.002,
      "loss": 2.6561412811279297,
      "step": 5850
    },
    {
      "epoch": 0.8051936734782799,
      "grad_norm": 0.05604272335767746,
      "learning_rate": 0.002,
      "loss": 2.727952480316162,
      "step": 5852
    },
    {
      "epoch": 0.8054688592860304,
      "grad_norm": 0.16672182083129883,
      "learning_rate": 0.002,
      "loss": 2.69278621673584,
      "step": 5854
    },
    {
      "epoch": 0.8057440450937811,
      "grad_norm": 0.1998693346977234,
      "learning_rate": 0.002,
      "loss": 2.661997079849243,
      "step": 5856
    },
    {
      "epoch": 0.8060192309015316,
      "grad_norm": 0.0903523862361908,
      "learning_rate": 0.002,
      "loss": 2.705576181411743,
      "step": 5858
    },
    {
      "epoch": 0.8062944167092823,
      "grad_norm": 0.19485270977020264,
      "learning_rate": 0.002,
      "loss": 2.6751651763916016,
      "step": 5860
    },
    {
      "epoch": 0.8065696025170328,
      "grad_norm": 0.059701137244701385,
      "learning_rate": 0.002,
      "loss": 2.7261624336242676,
      "step": 5862
    },
    {
      "epoch": 0.8068447883247835,
      "grad_norm": 0.06447116285562515,
      "learning_rate": 0.002,
      "loss": 2.7228598594665527,
      "step": 5864
    },
    {
      "epoch": 0.8071199741325341,
      "grad_norm": 0.1332075595855713,
      "learning_rate": 0.002,
      "loss": 2.7269325256347656,
      "step": 5866
    },
    {
      "epoch": 0.8073951599402847,
      "grad_norm": 0.061131883412599564,
      "learning_rate": 0.002,
      "loss": 2.7227039337158203,
      "step": 5868
    },
    {
      "epoch": 0.8076703457480353,
      "grad_norm": 0.17834427952766418,
      "learning_rate": 0.002,
      "loss": 2.700263500213623,
      "step": 5870
    },
    {
      "epoch": 0.8079455315557859,
      "grad_norm": 0.12339629977941513,
      "learning_rate": 0.002,
      "loss": 2.6912806034088135,
      "step": 5872
    },
    {
      "epoch": 0.8082207173635365,
      "grad_norm": 0.21917201578617096,
      "learning_rate": 0.002,
      "loss": 2.694736957550049,
      "step": 5874
    },
    {
      "epoch": 0.8084959031712872,
      "grad_norm": 0.04035812243819237,
      "learning_rate": 0.002,
      "loss": 2.766824722290039,
      "step": 5876
    },
    {
      "epoch": 0.8087710889790377,
      "grad_norm": 0.08582404255867004,
      "learning_rate": 0.002,
      "loss": 2.686819553375244,
      "step": 5878
    },
    {
      "epoch": 0.8090462747867884,
      "grad_norm": 0.059533555060625076,
      "learning_rate": 0.002,
      "loss": 2.7196059226989746,
      "step": 5880
    },
    {
      "epoch": 0.8093214605945389,
      "grad_norm": 0.03847323730587959,
      "learning_rate": 0.002,
      "loss": 2.7131731510162354,
      "step": 5882
    },
    {
      "epoch": 0.8095966464022896,
      "grad_norm": 0.13680216670036316,
      "learning_rate": 0.002,
      "loss": 2.6923327445983887,
      "step": 5884
    },
    {
      "epoch": 0.8098718322100401,
      "grad_norm": 0.10925120860338211,
      "learning_rate": 0.002,
      "loss": 2.669607162475586,
      "step": 5886
    },
    {
      "epoch": 0.8101470180177908,
      "grad_norm": 0.12080197036266327,
      "learning_rate": 0.002,
      "loss": 2.6842565536499023,
      "step": 5888
    },
    {
      "epoch": 0.8104222038255414,
      "grad_norm": 0.27982354164123535,
      "learning_rate": 0.002,
      "loss": 2.7211968898773193,
      "step": 5890
    },
    {
      "epoch": 0.810697389633292,
      "grad_norm": 0.04116354510188103,
      "learning_rate": 0.002,
      "loss": 2.6950714588165283,
      "step": 5892
    },
    {
      "epoch": 0.8109725754410426,
      "grad_norm": 0.15461285412311554,
      "learning_rate": 0.002,
      "loss": 2.711198568344116,
      "step": 5894
    },
    {
      "epoch": 0.8112477612487932,
      "grad_norm": 0.12845711410045624,
      "learning_rate": 0.002,
      "loss": 2.699906826019287,
      "step": 5896
    },
    {
      "epoch": 0.8115229470565438,
      "grad_norm": 0.15341241657733917,
      "learning_rate": 0.002,
      "loss": 2.7135791778564453,
      "step": 5898
    },
    {
      "epoch": 0.8117981328642944,
      "grad_norm": 0.1940414309501648,
      "learning_rate": 0.002,
      "loss": 2.7263708114624023,
      "step": 5900
    },
    {
      "epoch": 0.812073318672045,
      "grad_norm": 0.25144273042678833,
      "learning_rate": 0.002,
      "loss": 2.6949429512023926,
      "step": 5902
    },
    {
      "epoch": 0.8123485044797957,
      "grad_norm": 0.045706652104854584,
      "learning_rate": 0.002,
      "loss": 2.6662378311157227,
      "step": 5904
    },
    {
      "epoch": 0.8126236902875462,
      "grad_norm": 0.1253756433725357,
      "learning_rate": 0.002,
      "loss": 2.718595027923584,
      "step": 5906
    },
    {
      "epoch": 0.8128988760952969,
      "grad_norm": 0.0370040200650692,
      "learning_rate": 0.002,
      "loss": 2.711021900177002,
      "step": 5908
    },
    {
      "epoch": 0.8131740619030474,
      "grad_norm": 0.1377059668302536,
      "learning_rate": 0.002,
      "loss": 2.6981234550476074,
      "step": 5910
    },
    {
      "epoch": 0.8134492477107981,
      "grad_norm": 0.2209048718214035,
      "learning_rate": 0.002,
      "loss": 2.717163324356079,
      "step": 5912
    },
    {
      "epoch": 0.8137244335185486,
      "grad_norm": 0.05718986690044403,
      "learning_rate": 0.002,
      "loss": 2.7106478214263916,
      "step": 5914
    },
    {
      "epoch": 0.8139996193262993,
      "grad_norm": 0.03907901793718338,
      "learning_rate": 0.002,
      "loss": 2.7004008293151855,
      "step": 5916
    },
    {
      "epoch": 0.8142748051340499,
      "grad_norm": 0.05135295167565346,
      "learning_rate": 0.002,
      "loss": 2.7041797637939453,
      "step": 5918
    },
    {
      "epoch": 0.8145499909418005,
      "grad_norm": 0.09484105557203293,
      "learning_rate": 0.002,
      "loss": 2.7139225006103516,
      "step": 5920
    },
    {
      "epoch": 0.8148251767495511,
      "grad_norm": 0.1124827191233635,
      "learning_rate": 0.002,
      "loss": 2.7077674865722656,
      "step": 5922
    },
    {
      "epoch": 0.8151003625573017,
      "grad_norm": 0.03371020033955574,
      "learning_rate": 0.002,
      "loss": 2.705747604370117,
      "step": 5924
    },
    {
      "epoch": 0.8153755483650523,
      "grad_norm": 0.10779882222414017,
      "learning_rate": 0.002,
      "loss": 2.6739931106567383,
      "step": 5926
    },
    {
      "epoch": 0.815650734172803,
      "grad_norm": 0.30635786056518555,
      "learning_rate": 0.002,
      "loss": 2.713778495788574,
      "step": 5928
    },
    {
      "epoch": 0.8159259199805535,
      "grad_norm": 0.26441797614097595,
      "learning_rate": 0.002,
      "loss": 2.6885886192321777,
      "step": 5930
    },
    {
      "epoch": 0.8162011057883042,
      "grad_norm": 0.048641376197338104,
      "learning_rate": 0.002,
      "loss": 2.6597676277160645,
      "step": 5932
    },
    {
      "epoch": 0.8164762915960547,
      "grad_norm": 0.07775875180959702,
      "learning_rate": 0.002,
      "loss": 2.7202515602111816,
      "step": 5934
    },
    {
      "epoch": 0.8167514774038054,
      "grad_norm": 0.13529184460639954,
      "learning_rate": 0.002,
      "loss": 2.6887149810791016,
      "step": 5936
    },
    {
      "epoch": 0.8170266632115559,
      "grad_norm": 0.03140903636813164,
      "learning_rate": 0.002,
      "loss": 2.7211947441101074,
      "step": 5938
    },
    {
      "epoch": 0.8173018490193066,
      "grad_norm": 0.16028885543346405,
      "learning_rate": 0.002,
      "loss": 2.697690010070801,
      "step": 5940
    },
    {
      "epoch": 0.8175770348270572,
      "grad_norm": 0.16046853363513947,
      "learning_rate": 0.002,
      "loss": 2.728400230407715,
      "step": 5942
    },
    {
      "epoch": 0.8178522206348078,
      "grad_norm": 0.03414692357182503,
      "learning_rate": 0.002,
      "loss": 2.7154884338378906,
      "step": 5944
    },
    {
      "epoch": 0.8181274064425584,
      "grad_norm": 0.18310688436031342,
      "learning_rate": 0.002,
      "loss": 2.673889636993408,
      "step": 5946
    },
    {
      "epoch": 0.818402592250309,
      "grad_norm": 0.2363537698984146,
      "learning_rate": 0.002,
      "loss": 2.7020697593688965,
      "step": 5948
    },
    {
      "epoch": 0.8186777780580596,
      "grad_norm": 0.12773217260837555,
      "learning_rate": 0.002,
      "loss": 2.644676923751831,
      "step": 5950
    },
    {
      "epoch": 0.8189529638658102,
      "grad_norm": 0.22297227382659912,
      "learning_rate": 0.002,
      "loss": 2.704843521118164,
      "step": 5952
    },
    {
      "epoch": 0.8192281496735608,
      "grad_norm": 0.03239684924483299,
      "learning_rate": 0.002,
      "loss": 2.6812896728515625,
      "step": 5954
    },
    {
      "epoch": 0.8195033354813115,
      "grad_norm": 0.14413107931613922,
      "learning_rate": 0.002,
      "loss": 2.6990082263946533,
      "step": 5956
    },
    {
      "epoch": 0.819778521289062,
      "grad_norm": 0.1943587213754654,
      "learning_rate": 0.002,
      "loss": 2.708691358566284,
      "step": 5958
    },
    {
      "epoch": 0.8200537070968127,
      "grad_norm": 0.1438041627407074,
      "learning_rate": 0.002,
      "loss": 2.6812968254089355,
      "step": 5960
    },
    {
      "epoch": 0.8203288929045632,
      "grad_norm": 0.18719911575317383,
      "learning_rate": 0.002,
      "loss": 2.721457004547119,
      "step": 5962
    },
    {
      "epoch": 0.8206040787123139,
      "grad_norm": 0.29690900444984436,
      "learning_rate": 0.002,
      "loss": 2.7218570709228516,
      "step": 5964
    },
    {
      "epoch": 0.8208792645200644,
      "grad_norm": 0.15121249854564667,
      "learning_rate": 0.002,
      "loss": 2.6873416900634766,
      "step": 5966
    },
    {
      "epoch": 0.8211544503278151,
      "grad_norm": 0.21888180077075958,
      "learning_rate": 0.002,
      "loss": 2.7092785835266113,
      "step": 5968
    },
    {
      "epoch": 0.8214296361355657,
      "grad_norm": 0.1181945651769638,
      "learning_rate": 0.002,
      "loss": 2.7235946655273438,
      "step": 5970
    },
    {
      "epoch": 0.8217048219433163,
      "grad_norm": 0.1776047796010971,
      "learning_rate": 0.002,
      "loss": 2.7119874954223633,
      "step": 5972
    },
    {
      "epoch": 0.8219800077510669,
      "grad_norm": 0.06950753927230835,
      "learning_rate": 0.002,
      "loss": 2.7254552841186523,
      "step": 5974
    },
    {
      "epoch": 0.8222551935588175,
      "grad_norm": 0.17793841660022736,
      "learning_rate": 0.002,
      "loss": 2.7123780250549316,
      "step": 5976
    },
    {
      "epoch": 0.8225303793665681,
      "grad_norm": 0.0609467513859272,
      "learning_rate": 0.002,
      "loss": 2.676687240600586,
      "step": 5978
    },
    {
      "epoch": 0.8228055651743188,
      "grad_norm": 0.19659940898418427,
      "learning_rate": 0.002,
      "loss": 2.6932897567749023,
      "step": 5980
    },
    {
      "epoch": 0.8230807509820693,
      "grad_norm": 0.1328379511833191,
      "learning_rate": 0.002,
      "loss": 2.667283535003662,
      "step": 5982
    },
    {
      "epoch": 0.82335593678982,
      "grad_norm": 0.055357564240694046,
      "learning_rate": 0.002,
      "loss": 2.6706576347351074,
      "step": 5984
    },
    {
      "epoch": 0.8236311225975705,
      "grad_norm": 0.17916339635849,
      "learning_rate": 0.002,
      "loss": 2.6674489974975586,
      "step": 5986
    },
    {
      "epoch": 0.8239063084053212,
      "grad_norm": 0.11521446704864502,
      "learning_rate": 0.002,
      "loss": 2.6719298362731934,
      "step": 5988
    },
    {
      "epoch": 0.8241814942130717,
      "grad_norm": 0.031046438962221146,
      "learning_rate": 0.002,
      "loss": 2.6568479537963867,
      "step": 5990
    },
    {
      "epoch": 0.8244566800208224,
      "grad_norm": 0.05478933826088905,
      "learning_rate": 0.002,
      "loss": 2.6542935371398926,
      "step": 5992
    },
    {
      "epoch": 0.824731865828573,
      "grad_norm": 0.035462040454149246,
      "learning_rate": 0.002,
      "loss": 2.690943717956543,
      "step": 5994
    },
    {
      "epoch": 0.8250070516363236,
      "grad_norm": 0.03455584868788719,
      "learning_rate": 0.002,
      "loss": 2.6366260051727295,
      "step": 5996
    },
    {
      "epoch": 0.8252822374440743,
      "grad_norm": 0.10874679684638977,
      "learning_rate": 0.002,
      "loss": 2.677377223968506,
      "step": 5998
    },
    {
      "epoch": 0.8255574232518248,
      "grad_norm": 0.2032281756401062,
      "learning_rate": 0.002,
      "loss": 2.714653968811035,
      "step": 6000
    },
    {
      "epoch": 0.8258326090595755,
      "grad_norm": 0.1472606509923935,
      "learning_rate": 0.002,
      "loss": 2.701991081237793,
      "step": 6002
    },
    {
      "epoch": 0.826107794867326,
      "grad_norm": 0.07047142833471298,
      "learning_rate": 0.002,
      "loss": 2.6824092864990234,
      "step": 6004
    },
    {
      "epoch": 0.8263829806750767,
      "grad_norm": 0.043363023549318314,
      "learning_rate": 0.002,
      "loss": 2.6963658332824707,
      "step": 6006
    },
    {
      "epoch": 0.8266581664828273,
      "grad_norm": 0.04944485053420067,
      "learning_rate": 0.002,
      "loss": 2.7176764011383057,
      "step": 6008
    },
    {
      "epoch": 0.8269333522905778,
      "grad_norm": 0.04051681607961655,
      "learning_rate": 0.002,
      "loss": 2.6939406394958496,
      "step": 6010
    },
    {
      "epoch": 0.8272085380983285,
      "grad_norm": 0.03420990705490112,
      "learning_rate": 0.002,
      "loss": 2.6974639892578125,
      "step": 6012
    },
    {
      "epoch": 0.827483723906079,
      "grad_norm": 0.14933165907859802,
      "learning_rate": 0.002,
      "loss": 2.6853742599487305,
      "step": 6014
    },
    {
      "epoch": 0.8277589097138297,
      "grad_norm": 0.4034498631954193,
      "learning_rate": 0.002,
      "loss": 2.7174768447875977,
      "step": 6016
    },
    {
      "epoch": 0.8280340955215802,
      "grad_norm": 0.2503485381603241,
      "learning_rate": 0.002,
      "loss": 2.692903995513916,
      "step": 6018
    },
    {
      "epoch": 0.8283092813293309,
      "grad_norm": 0.27222028374671936,
      "learning_rate": 0.002,
      "loss": 2.6678717136383057,
      "step": 6020
    },
    {
      "epoch": 0.8285844671370816,
      "grad_norm": 0.11555394530296326,
      "learning_rate": 0.002,
      "loss": 2.664242744445801,
      "step": 6022
    },
    {
      "epoch": 0.8288596529448321,
      "grad_norm": 0.17604348063468933,
      "learning_rate": 0.002,
      "loss": 2.689481258392334,
      "step": 6024
    },
    {
      "epoch": 0.8291348387525828,
      "grad_norm": 0.08902259171009064,
      "learning_rate": 0.002,
      "loss": 2.695474147796631,
      "step": 6026
    },
    {
      "epoch": 0.8294100245603333,
      "grad_norm": 0.10760541260242462,
      "learning_rate": 0.002,
      "loss": 2.7001452445983887,
      "step": 6028
    },
    {
      "epoch": 0.829685210368084,
      "grad_norm": 0.03170352801680565,
      "learning_rate": 0.002,
      "loss": 2.7107300758361816,
      "step": 6030
    },
    {
      "epoch": 0.8299603961758346,
      "grad_norm": 0.06093885377049446,
      "learning_rate": 0.002,
      "loss": 2.6773035526275635,
      "step": 6032
    },
    {
      "epoch": 0.8302355819835852,
      "grad_norm": 0.10364031046628952,
      "learning_rate": 0.002,
      "loss": 2.6669650077819824,
      "step": 6034
    },
    {
      "epoch": 0.8305107677913358,
      "grad_norm": 0.04478621110320091,
      "learning_rate": 0.002,
      "loss": 2.672482490539551,
      "step": 6036
    },
    {
      "epoch": 0.8307859535990864,
      "grad_norm": 0.05495259538292885,
      "learning_rate": 0.002,
      "loss": 2.667281150817871,
      "step": 6038
    },
    {
      "epoch": 0.831061139406837,
      "grad_norm": 0.06848214566707611,
      "learning_rate": 0.002,
      "loss": 2.683990716934204,
      "step": 6040
    },
    {
      "epoch": 0.8313363252145876,
      "grad_norm": 0.05321577191352844,
      "learning_rate": 0.002,
      "loss": 2.6736369132995605,
      "step": 6042
    },
    {
      "epoch": 0.8316115110223382,
      "grad_norm": 0.09231294691562653,
      "learning_rate": 0.002,
      "loss": 2.7143702507019043,
      "step": 6044
    },
    {
      "epoch": 0.8318866968300889,
      "grad_norm": 0.053873009979724884,
      "learning_rate": 0.002,
      "loss": 2.7287912368774414,
      "step": 6046
    },
    {
      "epoch": 0.8321618826378394,
      "grad_norm": 0.038605011999607086,
      "learning_rate": 0.002,
      "loss": 2.6286914348602295,
      "step": 6048
    },
    {
      "epoch": 0.8324370684455901,
      "grad_norm": 0.11158081889152527,
      "learning_rate": 0.002,
      "loss": 2.691187858581543,
      "step": 6050
    },
    {
      "epoch": 0.8327122542533406,
      "grad_norm": 0.27189189195632935,
      "learning_rate": 0.002,
      "loss": 2.679004192352295,
      "step": 6052
    },
    {
      "epoch": 0.8329874400610913,
      "grad_norm": 0.31340745091438293,
      "learning_rate": 0.002,
      "loss": 2.68479061126709,
      "step": 6054
    },
    {
      "epoch": 0.8332626258688418,
      "grad_norm": 0.16664153337478638,
      "learning_rate": 0.002,
      "loss": 2.6850767135620117,
      "step": 6056
    },
    {
      "epoch": 0.8335378116765925,
      "grad_norm": 0.34356653690338135,
      "learning_rate": 0.002,
      "loss": 2.7280118465423584,
      "step": 6058
    },
    {
      "epoch": 0.8338129974843431,
      "grad_norm": 0.29546961188316345,
      "learning_rate": 0.002,
      "loss": 2.7053909301757812,
      "step": 6060
    },
    {
      "epoch": 0.8340881832920937,
      "grad_norm": 0.0633322149515152,
      "learning_rate": 0.002,
      "loss": 2.708700656890869,
      "step": 6062
    },
    {
      "epoch": 0.8343633690998443,
      "grad_norm": 0.37892580032348633,
      "learning_rate": 0.002,
      "loss": 2.683220386505127,
      "step": 6064
    },
    {
      "epoch": 0.8346385549075949,
      "grad_norm": 0.2366165667772293,
      "learning_rate": 0.002,
      "loss": 2.7008934020996094,
      "step": 6066
    },
    {
      "epoch": 0.8349137407153455,
      "grad_norm": 0.06175840273499489,
      "learning_rate": 0.002,
      "loss": 2.692591667175293,
      "step": 6068
    },
    {
      "epoch": 0.8351889265230961,
      "grad_norm": 0.17162653803825378,
      "learning_rate": 0.002,
      "loss": 2.687314033508301,
      "step": 6070
    },
    {
      "epoch": 0.8354641123308467,
      "grad_norm": 0.16992118954658508,
      "learning_rate": 0.002,
      "loss": 2.680476188659668,
      "step": 6072
    },
    {
      "epoch": 0.8357392981385974,
      "grad_norm": 0.03314380347728729,
      "learning_rate": 0.002,
      "loss": 2.723421573638916,
      "step": 6074
    },
    {
      "epoch": 0.8360144839463479,
      "grad_norm": 0.14642317593097687,
      "learning_rate": 0.002,
      "loss": 2.6921873092651367,
      "step": 6076
    },
    {
      "epoch": 0.8362896697540986,
      "grad_norm": 0.08361495286226273,
      "learning_rate": 0.002,
      "loss": 2.746896743774414,
      "step": 6078
    },
    {
      "epoch": 0.8365648555618491,
      "grad_norm": 0.07086802273988724,
      "learning_rate": 0.002,
      "loss": 2.7201032638549805,
      "step": 6080
    },
    {
      "epoch": 0.8368400413695998,
      "grad_norm": 0.11700782924890518,
      "learning_rate": 0.002,
      "loss": 2.6947507858276367,
      "step": 6082
    },
    {
      "epoch": 0.8371152271773503,
      "grad_norm": 0.05032774433493614,
      "learning_rate": 0.002,
      "loss": 2.69551944732666,
      "step": 6084
    },
    {
      "epoch": 0.837390412985101,
      "grad_norm": 0.09806249290704727,
      "learning_rate": 0.002,
      "loss": 2.643718719482422,
      "step": 6086
    },
    {
      "epoch": 0.8376655987928516,
      "grad_norm": 0.08513031899929047,
      "learning_rate": 0.002,
      "loss": 2.699636459350586,
      "step": 6088
    },
    {
      "epoch": 0.8379407846006022,
      "grad_norm": 0.06528325378894806,
      "learning_rate": 0.002,
      "loss": 2.672138214111328,
      "step": 6090
    },
    {
      "epoch": 0.8382159704083528,
      "grad_norm": 0.08392908424139023,
      "learning_rate": 0.002,
      "loss": 2.698448896408081,
      "step": 6092
    },
    {
      "epoch": 0.8384911562161034,
      "grad_norm": 0.0318012498319149,
      "learning_rate": 0.002,
      "loss": 2.6835687160491943,
      "step": 6094
    },
    {
      "epoch": 0.838766342023854,
      "grad_norm": 0.0830468162894249,
      "learning_rate": 0.002,
      "loss": 2.67694091796875,
      "step": 6096
    },
    {
      "epoch": 0.8390415278316047,
      "grad_norm": 0.03197489678859711,
      "learning_rate": 0.002,
      "loss": 2.6367878913879395,
      "step": 6098
    },
    {
      "epoch": 0.8393167136393552,
      "grad_norm": 0.10361988842487335,
      "learning_rate": 0.002,
      "loss": 2.770209789276123,
      "step": 6100
    },
    {
      "epoch": 0.8395918994471059,
      "grad_norm": 0.25714150071144104,
      "learning_rate": 0.002,
      "loss": 2.67527437210083,
      "step": 6102
    },
    {
      "epoch": 0.8398670852548564,
      "grad_norm": 0.363869845867157,
      "learning_rate": 0.002,
      "loss": 2.7123172283172607,
      "step": 6104
    },
    {
      "epoch": 0.8401422710626071,
      "grad_norm": 0.12423325330018997,
      "learning_rate": 0.002,
      "loss": 2.702737808227539,
      "step": 6106
    },
    {
      "epoch": 0.8404174568703576,
      "grad_norm": 0.33450356125831604,
      "learning_rate": 0.002,
      "loss": 2.684049129486084,
      "step": 6108
    },
    {
      "epoch": 0.8406926426781083,
      "grad_norm": 0.3997102975845337,
      "learning_rate": 0.002,
      "loss": 2.693419933319092,
      "step": 6110
    },
    {
      "epoch": 0.8409678284858589,
      "grad_norm": 0.09867778420448303,
      "learning_rate": 0.002,
      "loss": 2.660097599029541,
      "step": 6112
    },
    {
      "epoch": 0.8412430142936095,
      "grad_norm": 0.25921839475631714,
      "learning_rate": 0.002,
      "loss": 2.682150363922119,
      "step": 6114
    },
    {
      "epoch": 0.8415182001013601,
      "grad_norm": 0.33328166604042053,
      "learning_rate": 0.002,
      "loss": 2.6571638584136963,
      "step": 6116
    },
    {
      "epoch": 0.8417933859091107,
      "grad_norm": 0.2501026391983032,
      "learning_rate": 0.002,
      "loss": 2.6624577045440674,
      "step": 6118
    },
    {
      "epoch": 0.8420685717168613,
      "grad_norm": 0.0840572938323021,
      "learning_rate": 0.002,
      "loss": 2.6686911582946777,
      "step": 6120
    },
    {
      "epoch": 0.8423437575246119,
      "grad_norm": 0.06582627445459366,
      "learning_rate": 0.002,
      "loss": 2.6946678161621094,
      "step": 6122
    },
    {
      "epoch": 0.8426189433323625,
      "grad_norm": 0.1920076310634613,
      "learning_rate": 0.002,
      "loss": 2.6843862533569336,
      "step": 6124
    },
    {
      "epoch": 0.8428941291401132,
      "grad_norm": 0.13305668532848358,
      "learning_rate": 0.002,
      "loss": 2.70115327835083,
      "step": 6126
    },
    {
      "epoch": 0.8431693149478637,
      "grad_norm": 0.04962474852800369,
      "learning_rate": 0.002,
      "loss": 2.6565070152282715,
      "step": 6128
    },
    {
      "epoch": 0.8434445007556144,
      "grad_norm": 0.08673220127820969,
      "learning_rate": 0.002,
      "loss": 2.709651231765747,
      "step": 6130
    },
    {
      "epoch": 0.8437196865633649,
      "grad_norm": 0.15846677124500275,
      "learning_rate": 0.002,
      "loss": 2.69659423828125,
      "step": 6132
    },
    {
      "epoch": 0.8439948723711156,
      "grad_norm": 0.047172028571367264,
      "learning_rate": 0.002,
      "loss": 2.717039108276367,
      "step": 6134
    },
    {
      "epoch": 0.8442700581788661,
      "grad_norm": 0.13614699244499207,
      "learning_rate": 0.002,
      "loss": 2.67547345161438,
      "step": 6136
    },
    {
      "epoch": 0.8445452439866168,
      "grad_norm": 0.23029880225658417,
      "learning_rate": 0.002,
      "loss": 2.657956123352051,
      "step": 6138
    },
    {
      "epoch": 0.8448204297943674,
      "grad_norm": 0.10441999137401581,
      "learning_rate": 0.002,
      "loss": 2.6816763877868652,
      "step": 6140
    },
    {
      "epoch": 0.845095615602118,
      "grad_norm": 0.21205653250217438,
      "learning_rate": 0.002,
      "loss": 2.6787056922912598,
      "step": 6142
    },
    {
      "epoch": 0.8453708014098686,
      "grad_norm": 0.13849422335624695,
      "learning_rate": 0.002,
      "loss": 2.667748212814331,
      "step": 6144
    },
    {
      "epoch": 0.8456459872176192,
      "grad_norm": 0.14836642146110535,
      "learning_rate": 0.002,
      "loss": 2.656999111175537,
      "step": 6146
    },
    {
      "epoch": 0.8459211730253698,
      "grad_norm": 0.13607719540596008,
      "learning_rate": 0.002,
      "loss": 2.682551383972168,
      "step": 6148
    },
    {
      "epoch": 0.8461963588331205,
      "grad_norm": 0.058331504464149475,
      "learning_rate": 0.002,
      "loss": 2.6935572624206543,
      "step": 6150
    },
    {
      "epoch": 0.846471544640871,
      "grad_norm": 0.10345762968063354,
      "learning_rate": 0.002,
      "loss": 2.7305376529693604,
      "step": 6152
    },
    {
      "epoch": 0.8467467304486217,
      "grad_norm": 0.08815161883831024,
      "learning_rate": 0.002,
      "loss": 2.6842730045318604,
      "step": 6154
    },
    {
      "epoch": 0.8470219162563722,
      "grad_norm": 0.11787927150726318,
      "learning_rate": 0.002,
      "loss": 2.661803722381592,
      "step": 6156
    },
    {
      "epoch": 0.8472971020641229,
      "grad_norm": 0.03924933075904846,
      "learning_rate": 0.002,
      "loss": 2.7214558124542236,
      "step": 6158
    },
    {
      "epoch": 0.8475722878718734,
      "grad_norm": 0.155321404337883,
      "learning_rate": 0.002,
      "loss": 2.6769204139709473,
      "step": 6160
    },
    {
      "epoch": 0.8478474736796241,
      "grad_norm": 0.07687405496835709,
      "learning_rate": 0.002,
      "loss": 2.6761913299560547,
      "step": 6162
    },
    {
      "epoch": 0.8481226594873748,
      "grad_norm": 0.07075394690036774,
      "learning_rate": 0.002,
      "loss": 2.668893814086914,
      "step": 6164
    },
    {
      "epoch": 0.8483978452951253,
      "grad_norm": 0.052823152393102646,
      "learning_rate": 0.002,
      "loss": 2.7147488594055176,
      "step": 6166
    },
    {
      "epoch": 0.848673031102876,
      "grad_norm": 0.06578262150287628,
      "learning_rate": 0.001999390801096558,
      "loss": 2.7132983207702637,
      "step": 6168
    },
    {
      "epoch": 0.8489482169106265,
      "grad_norm": 0.3355269730091095,
      "learning_rate": 0.001998172403289674,
      "loss": 2.6825923919677734,
      "step": 6170
    },
    {
      "epoch": 0.8492234027183772,
      "grad_norm": 0.45907288789749146,
      "learning_rate": 0.00199695400548279,
      "loss": 2.727832078933716,
      "step": 6172
    },
    {
      "epoch": 0.8494985885261277,
      "grad_norm": 0.3693128526210785,
      "learning_rate": 0.001995735607675906,
      "loss": 2.736424446105957,
      "step": 6174
    },
    {
      "epoch": 0.8497737743338784,
      "grad_norm": 0.07882816344499588,
      "learning_rate": 0.0019945172098690222,
      "loss": 2.6654183864593506,
      "step": 6176
    },
    {
      "epoch": 0.850048960141629,
      "grad_norm": 0.2020597755908966,
      "learning_rate": 0.001993298812062138,
      "loss": 2.7129979133605957,
      "step": 6178
    },
    {
      "epoch": 0.8503241459493796,
      "grad_norm": 0.279024213552475,
      "learning_rate": 0.0019920804142552545,
      "loss": 2.658773899078369,
      "step": 6180
    },
    {
      "epoch": 0.8505993317571302,
      "grad_norm": 0.1903461366891861,
      "learning_rate": 0.0019908620164483704,
      "loss": 2.7173867225646973,
      "step": 6182
    },
    {
      "epoch": 0.8508745175648808,
      "grad_norm": 0.08083893358707428,
      "learning_rate": 0.0019896436186414867,
      "loss": 2.666494607925415,
      "step": 6184
    },
    {
      "epoch": 0.8511497033726314,
      "grad_norm": 0.06447666138410568,
      "learning_rate": 0.0019884252208346026,
      "loss": 2.6862034797668457,
      "step": 6186
    },
    {
      "epoch": 0.851424889180382,
      "grad_norm": 0.08685064315795898,
      "learning_rate": 0.0019872068230277185,
      "loss": 2.695232391357422,
      "step": 6188
    },
    {
      "epoch": 0.8517000749881326,
      "grad_norm": 0.19905884563922882,
      "learning_rate": 0.001985988425220835,
      "loss": 2.714505672454834,
      "step": 6190
    },
    {
      "epoch": 0.8519752607958833,
      "grad_norm": 0.049870528280735016,
      "learning_rate": 0.0019847700274139507,
      "loss": 2.6758625507354736,
      "step": 6192
    },
    {
      "epoch": 0.8522504466036338,
      "grad_norm": 0.07728881388902664,
      "learning_rate": 0.0019835516296070666,
      "loss": 2.638861656188965,
      "step": 6194
    },
    {
      "epoch": 0.8525256324113845,
      "grad_norm": 0.03559962660074234,
      "learning_rate": 0.001982333231800183,
      "loss": 2.6684398651123047,
      "step": 6196
    },
    {
      "epoch": 0.852800818219135,
      "grad_norm": 0.05494149774312973,
      "learning_rate": 0.001981114833993299,
      "loss": 2.6188886165618896,
      "step": 6198
    },
    {
      "epoch": 0.8530760040268857,
      "grad_norm": 0.0377986803650856,
      "learning_rate": 0.0019798964361864148,
      "loss": 2.675886631011963,
      "step": 6200
    },
    {
      "epoch": 0.8533511898346363,
      "grad_norm": 0.06497540324926376,
      "learning_rate": 0.0019786780383795307,
      "loss": 2.676851749420166,
      "step": 6202
    },
    {
      "epoch": 0.8536263756423869,
      "grad_norm": 0.07806174457073212,
      "learning_rate": 0.001977459640572647,
      "loss": 2.721743583679199,
      "step": 6204
    },
    {
      "epoch": 0.8539015614501375,
      "grad_norm": 0.11683250218629837,
      "learning_rate": 0.001976241242765763,
      "loss": 2.6650822162628174,
      "step": 6206
    },
    {
      "epoch": 0.8541767472578881,
      "grad_norm": 0.12592020630836487,
      "learning_rate": 0.0019750228449588792,
      "loss": 2.5946249961853027,
      "step": 6208
    },
    {
      "epoch": 0.8544519330656387,
      "grad_norm": 0.03788188099861145,
      "learning_rate": 0.001973804447151995,
      "loss": 2.683979034423828,
      "step": 6210
    },
    {
      "epoch": 0.8547271188733893,
      "grad_norm": 0.14525063335895538,
      "learning_rate": 0.0019725860493451115,
      "loss": 2.6716105937957764,
      "step": 6212
    },
    {
      "epoch": 0.8550023046811399,
      "grad_norm": 0.22297875583171844,
      "learning_rate": 0.0019713676515382274,
      "loss": 2.6807637214660645,
      "step": 6214
    },
    {
      "epoch": 0.8552774904888906,
      "grad_norm": 0.06666562706232071,
      "learning_rate": 0.0019701492537313433,
      "loss": 2.698269844055176,
      "step": 6216
    },
    {
      "epoch": 0.8555526762966411,
      "grad_norm": 0.29852399230003357,
      "learning_rate": 0.0019689308559244596,
      "loss": 2.6543960571289062,
      "step": 6218
    },
    {
      "epoch": 0.8558278621043918,
      "grad_norm": 0.24781003594398499,
      "learning_rate": 0.0019677124581175755,
      "loss": 2.6717305183410645,
      "step": 6220
    },
    {
      "epoch": 0.8561030479121423,
      "grad_norm": 0.18305350840091705,
      "learning_rate": 0.0019664940603106914,
      "loss": 2.6567306518554688,
      "step": 6222
    },
    {
      "epoch": 0.856378233719893,
      "grad_norm": 0.04452558606863022,
      "learning_rate": 0.0019652756625038077,
      "loss": 2.6752138137817383,
      "step": 6224
    },
    {
      "epoch": 0.8566534195276435,
      "grad_norm": 0.12201040238142014,
      "learning_rate": 0.0019640572646969236,
      "loss": 2.7037229537963867,
      "step": 6226
    },
    {
      "epoch": 0.8569286053353942,
      "grad_norm": 0.16192921996116638,
      "learning_rate": 0.0019628388668900395,
      "loss": 2.6962547302246094,
      "step": 6228
    },
    {
      "epoch": 0.8572037911431448,
      "grad_norm": 0.13636067509651184,
      "learning_rate": 0.0019616204690831554,
      "loss": 2.6417131423950195,
      "step": 6230
    },
    {
      "epoch": 0.8574789769508954,
      "grad_norm": 0.1677650511264801,
      "learning_rate": 0.0019604020712762718,
      "loss": 2.684954881668091,
      "step": 6232
    },
    {
      "epoch": 0.857754162758646,
      "grad_norm": 0.08813727647066116,
      "learning_rate": 0.0019591836734693877,
      "loss": 2.6521003246307373,
      "step": 6234
    },
    {
      "epoch": 0.8580293485663966,
      "grad_norm": 0.17891471087932587,
      "learning_rate": 0.001957965275662504,
      "loss": 2.689203977584839,
      "step": 6236
    },
    {
      "epoch": 0.8583045343741472,
      "grad_norm": 0.04739949852228165,
      "learning_rate": 0.00195674687785562,
      "loss": 2.6478476524353027,
      "step": 6238
    },
    {
      "epoch": 0.8585797201818978,
      "grad_norm": 0.3151126205921173,
      "learning_rate": 0.001955528480048736,
      "loss": 2.705981492996216,
      "step": 6240
    },
    {
      "epoch": 0.8588549059896484,
      "grad_norm": 0.035952530801296234,
      "learning_rate": 0.001954310082241852,
      "loss": 2.6728358268737793,
      "step": 6242
    },
    {
      "epoch": 0.8591300917973991,
      "grad_norm": 0.4416637122631073,
      "learning_rate": 0.001953091684434968,
      "loss": 2.731609344482422,
      "step": 6244
    },
    {
      "epoch": 0.8594052776051496,
      "grad_norm": 0.12793651223182678,
      "learning_rate": 0.0019518732866280841,
      "loss": 2.7094051837921143,
      "step": 6246
    },
    {
      "epoch": 0.8596804634129003,
      "grad_norm": 0.2153903841972351,
      "learning_rate": 0.0019506548888212,
      "loss": 2.6653599739074707,
      "step": 6248
    },
    {
      "epoch": 0.8599556492206508,
      "grad_norm": 0.23665837943553925,
      "learning_rate": 0.0019494364910143164,
      "loss": 2.654186725616455,
      "step": 6250
    },
    {
      "epoch": 0.8602308350284015,
      "grad_norm": 0.16340656578540802,
      "learning_rate": 0.0019482180932074323,
      "loss": 2.6662511825561523,
      "step": 6252
    },
    {
      "epoch": 0.8605060208361521,
      "grad_norm": 0.04437893256545067,
      "learning_rate": 0.0019469996954005482,
      "loss": 2.620131015777588,
      "step": 6254
    },
    {
      "epoch": 0.8607812066439027,
      "grad_norm": 0.21817588806152344,
      "learning_rate": 0.0019457812975936645,
      "loss": 2.663961887359619,
      "step": 6256
    },
    {
      "epoch": 0.8610563924516533,
      "grad_norm": 0.1275026649236679,
      "learning_rate": 0.0019445628997867804,
      "loss": 2.670654296875,
      "step": 6258
    },
    {
      "epoch": 0.8613315782594039,
      "grad_norm": 0.04352701082825661,
      "learning_rate": 0.0019433445019798965,
      "loss": 2.6798901557922363,
      "step": 6260
    },
    {
      "epoch": 0.8616067640671545,
      "grad_norm": 0.14268480241298676,
      "learning_rate": 0.0019421261041730124,
      "loss": 2.7059245109558105,
      "step": 6262
    },
    {
      "epoch": 0.8618819498749051,
      "grad_norm": 0.07823630422353745,
      "learning_rate": 0.0019409077063661287,
      "loss": 2.7150142192840576,
      "step": 6264
    },
    {
      "epoch": 0.8621571356826557,
      "grad_norm": 0.04234723001718521,
      "learning_rate": 0.0019396893085592446,
      "loss": 2.6812000274658203,
      "step": 6266
    },
    {
      "epoch": 0.8624323214904064,
      "grad_norm": 0.07816974073648453,
      "learning_rate": 0.0019384709107523605,
      "loss": 2.6933693885803223,
      "step": 6268
    },
    {
      "epoch": 0.8627075072981569,
      "grad_norm": 0.04563206061720848,
      "learning_rate": 0.0019372525129454769,
      "loss": 2.6729166507720947,
      "step": 6270
    },
    {
      "epoch": 0.8629826931059076,
      "grad_norm": 0.08831585198640823,
      "learning_rate": 0.0019360341151385928,
      "loss": 2.647723913192749,
      "step": 6272
    },
    {
      "epoch": 0.8632578789136581,
      "grad_norm": 0.10238515585660934,
      "learning_rate": 0.001934815717331709,
      "loss": 2.664639949798584,
      "step": 6274
    },
    {
      "epoch": 0.8635330647214088,
      "grad_norm": 0.1360102742910385,
      "learning_rate": 0.0019335973195248248,
      "loss": 2.678079605102539,
      "step": 6276
    },
    {
      "epoch": 0.8638082505291593,
      "grad_norm": 0.05301332473754883,
      "learning_rate": 0.0019323789217179411,
      "loss": 2.6890687942504883,
      "step": 6278
    },
    {
      "epoch": 0.86408343633691,
      "grad_norm": 0.02968369424343109,
      "learning_rate": 0.001931160523911057,
      "loss": 2.6941680908203125,
      "step": 6280
    },
    {
      "epoch": 0.8643586221446606,
      "grad_norm": 0.0897030383348465,
      "learning_rate": 0.001929942126104173,
      "loss": 2.6323986053466797,
      "step": 6282
    },
    {
      "epoch": 0.8646338079524112,
      "grad_norm": 0.03523460403084755,
      "learning_rate": 0.0019287237282972893,
      "loss": 2.682234287261963,
      "step": 6284
    },
    {
      "epoch": 0.8649089937601618,
      "grad_norm": 0.08551566302776337,
      "learning_rate": 0.0019275053304904052,
      "loss": 2.6977927684783936,
      "step": 6286
    },
    {
      "epoch": 0.8651841795679124,
      "grad_norm": 0.06068364158272743,
      "learning_rate": 0.0019262869326835213,
      "loss": 2.697399377822876,
      "step": 6288
    },
    {
      "epoch": 0.865459365375663,
      "grad_norm": 0.03945232555270195,
      "learning_rate": 0.0019250685348766372,
      "loss": 2.636732578277588,
      "step": 6290
    },
    {
      "epoch": 0.8657345511834136,
      "grad_norm": 0.11203355342149734,
      "learning_rate": 0.0019238501370697535,
      "loss": 2.6915628910064697,
      "step": 6292
    },
    {
      "epoch": 0.8660097369911642,
      "grad_norm": 0.33748552203178406,
      "learning_rate": 0.0019226317392628694,
      "loss": 2.6894822120666504,
      "step": 6294
    },
    {
      "epoch": 0.8662849227989149,
      "grad_norm": 0.37277528643608093,
      "learning_rate": 0.0019214133414559853,
      "loss": 2.682466983795166,
      "step": 6296
    },
    {
      "epoch": 0.8665601086066654,
      "grad_norm": 0.3913414180278778,
      "learning_rate": 0.0019201949436491014,
      "loss": 2.6657872200012207,
      "step": 6298
    },
    {
      "epoch": 0.8668352944144161,
      "grad_norm": 0.08478415757417679,
      "learning_rate": 0.0019189765458422175,
      "loss": 2.655668258666992,
      "step": 6300
    },
    {
      "epoch": 0.8671104802221666,
      "grad_norm": 0.17448468506336212,
      "learning_rate": 0.0019177581480353337,
      "loss": 2.702772617340088,
      "step": 6302
    },
    {
      "epoch": 0.8673856660299173,
      "grad_norm": 0.31880149245262146,
      "learning_rate": 0.0019165397502284495,
      "loss": 2.691488742828369,
      "step": 6304
    },
    {
      "epoch": 0.867660851837668,
      "grad_norm": 0.2468518763780594,
      "learning_rate": 0.0019153213524215657,
      "loss": 2.6324572563171387,
      "step": 6306
    },
    {
      "epoch": 0.8679360376454185,
      "grad_norm": 0.12547725439071655,
      "learning_rate": 0.0019141029546146818,
      "loss": 2.7025651931762695,
      "step": 6308
    },
    {
      "epoch": 0.8682112234531691,
      "grad_norm": 0.11658823490142822,
      "learning_rate": 0.0019128845568077977,
      "loss": 2.671213150024414,
      "step": 6310
    },
    {
      "epoch": 0.8684864092609197,
      "grad_norm": 0.20040738582611084,
      "learning_rate": 0.0019116661590009138,
      "loss": 2.6584420204162598,
      "step": 6312
    },
    {
      "epoch": 0.8687615950686703,
      "grad_norm": 0.07385959476232529,
      "learning_rate": 0.00191044776119403,
      "loss": 2.6671228408813477,
      "step": 6314
    },
    {
      "epoch": 0.8690367808764209,
      "grad_norm": 0.06840259581804276,
      "learning_rate": 0.001909229363387146,
      "loss": 2.6868319511413574,
      "step": 6316
    },
    {
      "epoch": 0.8693119666841715,
      "grad_norm": 0.10243239253759384,
      "learning_rate": 0.001908010965580262,
      "loss": 2.6799612045288086,
      "step": 6318
    },
    {
      "epoch": 0.8695871524919222,
      "grad_norm": 0.035122424364089966,
      "learning_rate": 0.001906792567773378,
      "loss": 2.7112603187561035,
      "step": 6320
    },
    {
      "epoch": 0.8698623382996727,
      "grad_norm": 0.052054133266210556,
      "learning_rate": 0.0019055741699664942,
      "loss": 2.6832644939422607,
      "step": 6322
    },
    {
      "epoch": 0.8701375241074234,
      "grad_norm": 0.043832167983055115,
      "learning_rate": 0.00190435577215961,
      "loss": 2.6688485145568848,
      "step": 6324
    },
    {
      "epoch": 0.8704127099151739,
      "grad_norm": 0.08721809089183807,
      "learning_rate": 0.0019031373743527262,
      "loss": 2.6680479049682617,
      "step": 6326
    },
    {
      "epoch": 0.8706878957229246,
      "grad_norm": 0.15823763608932495,
      "learning_rate": 0.0019019189765458423,
      "loss": 2.718752384185791,
      "step": 6328
    },
    {
      "epoch": 0.8709630815306751,
      "grad_norm": 0.177388995885849,
      "learning_rate": 0.0019007005787389584,
      "loss": 2.656538963317871,
      "step": 6330
    },
    {
      "epoch": 0.8712382673384258,
      "grad_norm": 0.12642055749893188,
      "learning_rate": 0.0018994821809320743,
      "loss": 2.699207067489624,
      "step": 6332
    },
    {
      "epoch": 0.8715134531461765,
      "grad_norm": 0.058639876544475555,
      "learning_rate": 0.0018982637831251904,
      "loss": 2.6593575477600098,
      "step": 6334
    },
    {
      "epoch": 0.871788638953927,
      "grad_norm": 0.037440430372953415,
      "learning_rate": 0.0018970453853183065,
      "loss": 2.7151832580566406,
      "step": 6336
    },
    {
      "epoch": 0.8720638247616777,
      "grad_norm": 0.11376090347766876,
      "learning_rate": 0.0018958269875114224,
      "loss": 2.7072668075561523,
      "step": 6338
    },
    {
      "epoch": 0.8723390105694282,
      "grad_norm": 0.1928931623697281,
      "learning_rate": 0.0018946085897045386,
      "loss": 2.6280341148376465,
      "step": 6340
    },
    {
      "epoch": 0.8726141963771789,
      "grad_norm": 0.06338932365179062,
      "learning_rate": 0.0018933901918976547,
      "loss": 2.676791191101074,
      "step": 6342
    },
    {
      "epoch": 0.8728893821849294,
      "grad_norm": 0.11820343136787415,
      "learning_rate": 0.0018921717940907708,
      "loss": 2.666189193725586,
      "step": 6344
    },
    {
      "epoch": 0.87316456799268,
      "grad_norm": 0.03063373826444149,
      "learning_rate": 0.0018909533962838867,
      "loss": 2.64520525932312,
      "step": 6346
    },
    {
      "epoch": 0.8734397538004307,
      "grad_norm": 0.1308855563402176,
      "learning_rate": 0.0018897349984770028,
      "loss": 2.6729958057403564,
      "step": 6348
    },
    {
      "epoch": 0.8737149396081813,
      "grad_norm": 0.14029629528522491,
      "learning_rate": 0.001888516600670119,
      "loss": 2.6529040336608887,
      "step": 6350
    },
    {
      "epoch": 0.8739901254159319,
      "grad_norm": 0.060828305780887604,
      "learning_rate": 0.0018872982028632348,
      "loss": 2.6775565147399902,
      "step": 6352
    },
    {
      "epoch": 0.8742653112236825,
      "grad_norm": 0.1053302064538002,
      "learning_rate": 0.001886079805056351,
      "loss": 2.6621172428131104,
      "step": 6354
    },
    {
      "epoch": 0.8745404970314331,
      "grad_norm": 0.27102044224739075,
      "learning_rate": 0.001884861407249467,
      "loss": 2.699777603149414,
      "step": 6356
    },
    {
      "epoch": 0.8748156828391838,
      "grad_norm": 0.03783746436238289,
      "learning_rate": 0.001883643009442583,
      "loss": 2.7082977294921875,
      "step": 6358
    },
    {
      "epoch": 0.8750908686469343,
      "grad_norm": 0.2077716439962387,
      "learning_rate": 0.001882424611635699,
      "loss": 2.714493751525879,
      "step": 6360
    },
    {
      "epoch": 0.875366054454685,
      "grad_norm": 0.08349249511957169,
      "learning_rate": 0.001881206213828815,
      "loss": 2.633291721343994,
      "step": 6362
    },
    {
      "epoch": 0.8756412402624355,
      "grad_norm": 0.05389676243066788,
      "learning_rate": 0.0018799878160219313,
      "loss": 2.662141799926758,
      "step": 6364
    },
    {
      "epoch": 0.8759164260701862,
      "grad_norm": 0.166610985994339,
      "learning_rate": 0.0018787694182150472,
      "loss": 2.6688005924224854,
      "step": 6366
    },
    {
      "epoch": 0.8761916118779367,
      "grad_norm": 0.09682833403348923,
      "learning_rate": 0.0018775510204081633,
      "loss": 2.677356719970703,
      "step": 6368
    },
    {
      "epoch": 0.8764667976856874,
      "grad_norm": 0.13510112464427948,
      "learning_rate": 0.0018763326226012794,
      "loss": 2.733781337738037,
      "step": 6370
    },
    {
      "epoch": 0.876741983493438,
      "grad_norm": 0.1677357405424118,
      "learning_rate": 0.0018751142247943953,
      "loss": 2.6569151878356934,
      "step": 6372
    },
    {
      "epoch": 0.8770171693011886,
      "grad_norm": 0.03949865698814392,
      "learning_rate": 0.0018738958269875114,
      "loss": 2.6427042484283447,
      "step": 6374
    },
    {
      "epoch": 0.8772923551089392,
      "grad_norm": 0.03280359506607056,
      "learning_rate": 0.0018726774291806273,
      "loss": 2.656831741333008,
      "step": 6376
    },
    {
      "epoch": 0.8775675409166898,
      "grad_norm": 0.07030162215232849,
      "learning_rate": 0.0018714590313737437,
      "loss": 2.7145938873291016,
      "step": 6378
    },
    {
      "epoch": 0.8778427267244404,
      "grad_norm": 0.03449106216430664,
      "learning_rate": 0.0018702406335668596,
      "loss": 2.666872501373291,
      "step": 6380
    },
    {
      "epoch": 0.878117912532191,
      "grad_norm": 0.036083463579416275,
      "learning_rate": 0.0018690222357599757,
      "loss": 2.7047321796417236,
      "step": 6382
    },
    {
      "epoch": 0.8783930983399416,
      "grad_norm": 0.12308032065629959,
      "learning_rate": 0.0018678038379530918,
      "loss": 2.6665725708007812,
      "step": 6384
    },
    {
      "epoch": 0.8786682841476923,
      "grad_norm": 0.2935299575328827,
      "learning_rate": 0.0018665854401462077,
      "loss": 2.6770458221435547,
      "step": 6386
    },
    {
      "epoch": 0.8789434699554428,
      "grad_norm": 0.20172326266765594,
      "learning_rate": 0.0018653670423393238,
      "loss": 2.6745798587799072,
      "step": 6388
    },
    {
      "epoch": 0.8792186557631935,
      "grad_norm": 0.39448606967926025,
      "learning_rate": 0.0018641486445324397,
      "loss": 2.664073944091797,
      "step": 6390
    },
    {
      "epoch": 0.879493841570944,
      "grad_norm": 0.05031166598200798,
      "learning_rate": 0.001862930246725556,
      "loss": 2.649858236312866,
      "step": 6392
    },
    {
      "epoch": 0.8797690273786947,
      "grad_norm": 0.3303467929363251,
      "learning_rate": 0.001861711848918672,
      "loss": 2.659008026123047,
      "step": 6394
    },
    {
      "epoch": 0.8800442131864452,
      "grad_norm": 0.3170781433582306,
      "learning_rate": 0.001860493451111788,
      "loss": 2.634525775909424,
      "step": 6396
    },
    {
      "epoch": 0.8803193989941959,
      "grad_norm": 0.15637919306755066,
      "learning_rate": 0.0018592750533049042,
      "loss": 2.6979336738586426,
      "step": 6398
    },
    {
      "epoch": 0.8805945848019465,
      "grad_norm": 0.04421396180987358,
      "learning_rate": 0.00185805665549802,
      "loss": 2.671766519546509,
      "step": 6400
    },
    {
      "epoch": 0.8808697706096971,
      "grad_norm": 0.04129449278116226,
      "learning_rate": 0.0018568382576911362,
      "loss": 2.6784780025482178,
      "step": 6402
    },
    {
      "epoch": 0.8811449564174477,
      "grad_norm": 0.03343198075890541,
      "learning_rate": 0.001855619859884252,
      "loss": 2.6804282665252686,
      "step": 6404
    },
    {
      "epoch": 0.8814201422251983,
      "grad_norm": 0.04486339911818504,
      "learning_rate": 0.0018544014620773684,
      "loss": 2.6910619735717773,
      "step": 6406
    },
    {
      "epoch": 0.8816953280329489,
      "grad_norm": 0.05377988889813423,
      "learning_rate": 0.0018531830642704843,
      "loss": 2.6706953048706055,
      "step": 6408
    },
    {
      "epoch": 0.8819705138406996,
      "grad_norm": 0.1418788731098175,
      "learning_rate": 0.0018519646664636005,
      "loss": 2.6836178302764893,
      "step": 6410
    },
    {
      "epoch": 0.8822456996484501,
      "grad_norm": 0.07759148627519608,
      "learning_rate": 0.0018507462686567166,
      "loss": 2.7115578651428223,
      "step": 6412
    },
    {
      "epoch": 0.8825208854562008,
      "grad_norm": 0.15016190707683563,
      "learning_rate": 0.0018495278708498325,
      "loss": 2.611220359802246,
      "step": 6414
    },
    {
      "epoch": 0.8827960712639513,
      "grad_norm": 0.13292814791202545,
      "learning_rate": 0.0018483094730429486,
      "loss": 2.660609245300293,
      "step": 6416
    },
    {
      "epoch": 0.883071257071702,
      "grad_norm": 0.14978140592575073,
      "learning_rate": 0.0018470910752360645,
      "loss": 2.682150363922119,
      "step": 6418
    },
    {
      "epoch": 0.8833464428794525,
      "grad_norm": 0.12220805138349533,
      "learning_rate": 0.0018458726774291808,
      "loss": 2.652751922607422,
      "step": 6420
    },
    {
      "epoch": 0.8836216286872032,
      "grad_norm": 0.03884109854698181,
      "learning_rate": 0.0018446542796222967,
      "loss": 2.6904711723327637,
      "step": 6422
    },
    {
      "epoch": 0.8838968144949538,
      "grad_norm": 0.05433305352926254,
      "learning_rate": 0.0018434358818154126,
      "loss": 2.6462764739990234,
      "step": 6424
    },
    {
      "epoch": 0.8841720003027044,
      "grad_norm": 0.04095993936061859,
      "learning_rate": 0.001842217484008529,
      "loss": 2.6777420043945312,
      "step": 6426
    },
    {
      "epoch": 0.884447186110455,
      "grad_norm": 0.10603372752666473,
      "learning_rate": 0.0018409990862016448,
      "loss": 2.684753894805908,
      "step": 6428
    },
    {
      "epoch": 0.8847223719182056,
      "grad_norm": 0.04841538146138191,
      "learning_rate": 0.001839780688394761,
      "loss": 2.654852867126465,
      "step": 6430
    },
    {
      "epoch": 0.8849975577259562,
      "grad_norm": 0.08339881151914597,
      "learning_rate": 0.0018385622905878769,
      "loss": 2.687847852706909,
      "step": 6432
    },
    {
      "epoch": 0.8852727435337068,
      "grad_norm": 0.08087374269962311,
      "learning_rate": 0.0018373438927809932,
      "loss": 2.659666061401367,
      "step": 6434
    },
    {
      "epoch": 0.8855479293414574,
      "grad_norm": 0.03579192981123924,
      "learning_rate": 0.001836125494974109,
      "loss": 2.7126452922821045,
      "step": 6436
    },
    {
      "epoch": 0.8858231151492081,
      "grad_norm": 0.06090547889471054,
      "learning_rate": 0.001834907097167225,
      "loss": 2.643338203430176,
      "step": 6438
    },
    {
      "epoch": 0.8860983009569586,
      "grad_norm": 0.1411380171775818,
      "learning_rate": 0.0018336886993603413,
      "loss": 2.6231751441955566,
      "step": 6440
    },
    {
      "epoch": 0.8863734867647093,
      "grad_norm": 0.29628682136535645,
      "learning_rate": 0.0018324703015534572,
      "loss": 2.657780885696411,
      "step": 6442
    },
    {
      "epoch": 0.8866486725724598,
      "grad_norm": 0.062197741121053696,
      "learning_rate": 0.0018312519037465733,
      "loss": 2.671140670776367,
      "step": 6444
    },
    {
      "epoch": 0.8869238583802105,
      "grad_norm": 0.2706228494644165,
      "learning_rate": 0.0018300335059396892,
      "loss": 2.720339775085449,
      "step": 6446
    },
    {
      "epoch": 0.887199044187961,
      "grad_norm": 0.12770967185497284,
      "learning_rate": 0.0018288151081328056,
      "loss": 2.656576633453369,
      "step": 6448
    },
    {
      "epoch": 0.8874742299957117,
      "grad_norm": 0.13932259380817413,
      "learning_rate": 0.0018275967103259215,
      "loss": 2.6934618949890137,
      "step": 6450
    },
    {
      "epoch": 0.8877494158034623,
      "grad_norm": 0.033222541213035583,
      "learning_rate": 0.0018263783125190374,
      "loss": 2.6558353900909424,
      "step": 6452
    },
    {
      "epoch": 0.8880246016112129,
      "grad_norm": 0.06763698905706406,
      "learning_rate": 0.0018251599147121535,
      "loss": 2.6644906997680664,
      "step": 6454
    },
    {
      "epoch": 0.8882997874189635,
      "grad_norm": 0.07842137664556503,
      "learning_rate": 0.0018239415169052696,
      "loss": 2.673234224319458,
      "step": 6456
    },
    {
      "epoch": 0.8885749732267141,
      "grad_norm": 0.035320866852998734,
      "learning_rate": 0.0018227231190983857,
      "loss": 2.6790428161621094,
      "step": 6458
    },
    {
      "epoch": 0.8888501590344647,
      "grad_norm": 0.0356484092772007,
      "learning_rate": 0.0018215047212915016,
      "loss": 2.651068687438965,
      "step": 6460
    },
    {
      "epoch": 0.8891253448422154,
      "grad_norm": 0.1054459661245346,
      "learning_rate": 0.001820286323484618,
      "loss": 2.698331117630005,
      "step": 6462
    },
    {
      "epoch": 0.8894005306499659,
      "grad_norm": 0.15835152566432953,
      "learning_rate": 0.0018190679256777339,
      "loss": 2.675522804260254,
      "step": 6464
    },
    {
      "epoch": 0.8896757164577166,
      "grad_norm": 0.18506404757499695,
      "learning_rate": 0.0018178495278708497,
      "loss": 2.6409335136413574,
      "step": 6466
    },
    {
      "epoch": 0.8899509022654671,
      "grad_norm": 0.03545254468917847,
      "learning_rate": 0.0018166311300639659,
      "loss": 2.645949363708496,
      "step": 6468
    },
    {
      "epoch": 0.8902260880732178,
      "grad_norm": 0.17536504566669464,
      "learning_rate": 0.001815412732257082,
      "loss": 2.6348302364349365,
      "step": 6470
    },
    {
      "epoch": 0.8905012738809683,
      "grad_norm": 0.08377684652805328,
      "learning_rate": 0.001814194334450198,
      "loss": 2.6293301582336426,
      "step": 6472
    },
    {
      "epoch": 0.890776459688719,
      "grad_norm": 0.10235488414764404,
      "learning_rate": 0.001812975936643314,
      "loss": 2.621713638305664,
      "step": 6474
    },
    {
      "epoch": 0.8910516454964696,
      "grad_norm": 0.19961048662662506,
      "learning_rate": 0.0018117575388364303,
      "loss": 2.70703387260437,
      "step": 6476
    },
    {
      "epoch": 0.8913268313042202,
      "grad_norm": 0.04732269048690796,
      "learning_rate": 0.0018105391410295462,
      "loss": 2.6608550548553467,
      "step": 6478
    },
    {
      "epoch": 0.8916020171119708,
      "grad_norm": 0.1933644860982895,
      "learning_rate": 0.0018093207432226621,
      "loss": 2.6837172508239746,
      "step": 6480
    },
    {
      "epoch": 0.8918772029197214,
      "grad_norm": 0.0798564925789833,
      "learning_rate": 0.0018081023454157782,
      "loss": 2.648900032043457,
      "step": 6482
    },
    {
      "epoch": 0.892152388727472,
      "grad_norm": 0.16445311903953552,
      "learning_rate": 0.0018068839476088944,
      "loss": 2.6845293045043945,
      "step": 6484
    },
    {
      "epoch": 0.8924275745352226,
      "grad_norm": 0.036795541644096375,
      "learning_rate": 0.0018056655498020105,
      "loss": 2.62105131149292,
      "step": 6486
    },
    {
      "epoch": 0.8927027603429732,
      "grad_norm": 0.19904400408267975,
      "learning_rate": 0.0018044471519951264,
      "loss": 2.6819605827331543,
      "step": 6488
    },
    {
      "epoch": 0.8929779461507239,
      "grad_norm": 0.06248465180397034,
      "learning_rate": 0.0018032287541882425,
      "loss": 2.6753478050231934,
      "step": 6490
    },
    {
      "epoch": 0.8932531319584744,
      "grad_norm": 0.20511819422245026,
      "learning_rate": 0.0018020103563813586,
      "loss": 2.6594078540802,
      "step": 6492
    },
    {
      "epoch": 0.8935283177662251,
      "grad_norm": 0.11986098438501358,
      "learning_rate": 0.0018007919585744745,
      "loss": 2.663445472717285,
      "step": 6494
    },
    {
      "epoch": 0.8938035035739756,
      "grad_norm": 0.042932942509651184,
      "learning_rate": 0.0017995735607675906,
      "loss": 2.6260714530944824,
      "step": 6496
    },
    {
      "epoch": 0.8940786893817263,
      "grad_norm": 0.04417005553841591,
      "learning_rate": 0.0017983551629607067,
      "loss": 2.6661980152130127,
      "step": 6498
    },
    {
      "epoch": 0.8943538751894768,
      "grad_norm": 0.07507164776325226,
      "learning_rate": 0.0017971367651538229,
      "loss": 2.70220947265625,
      "step": 6500
    },
    {
      "epoch": 0.8946290609972275,
      "grad_norm": 0.1353432834148407,
      "learning_rate": 0.0017959183673469388,
      "loss": 2.658055305480957,
      "step": 6502
    },
    {
      "epoch": 0.8949042468049782,
      "grad_norm": 0.1265808790922165,
      "learning_rate": 0.0017946999695400549,
      "loss": 2.6945528984069824,
      "step": 6504
    },
    {
      "epoch": 0.8951794326127287,
      "grad_norm": 0.04533235356211662,
      "learning_rate": 0.001793481571733171,
      "loss": 2.6554694175720215,
      "step": 6506
    },
    {
      "epoch": 0.8954546184204794,
      "grad_norm": 0.19613540172576904,
      "learning_rate": 0.0017922631739262869,
      "loss": 2.667327880859375,
      "step": 6508
    },
    {
      "epoch": 0.8957298042282299,
      "grad_norm": 0.10317420214414597,
      "learning_rate": 0.001791044776119403,
      "loss": 2.647465705871582,
      "step": 6510
    },
    {
      "epoch": 0.8960049900359806,
      "grad_norm": 0.2781607210636139,
      "learning_rate": 0.0017898263783125191,
      "loss": 2.670867681503296,
      "step": 6512
    },
    {
      "epoch": 0.8962801758437311,
      "grad_norm": 0.12497787922620773,
      "learning_rate": 0.0017886079805056352,
      "loss": 2.652137517929077,
      "step": 6514
    },
    {
      "epoch": 0.8965553616514818,
      "grad_norm": 0.260028600692749,
      "learning_rate": 0.0017873895826987511,
      "loss": 2.644087791442871,
      "step": 6516
    },
    {
      "epoch": 0.8968305474592324,
      "grad_norm": 0.04682578891515732,
      "learning_rate": 0.001786171184891867,
      "loss": 2.6933131217956543,
      "step": 6518
    },
    {
      "epoch": 0.897105733266983,
      "grad_norm": 0.1894054412841797,
      "learning_rate": 0.0017849527870849834,
      "loss": 2.668869972229004,
      "step": 6520
    },
    {
      "epoch": 0.8973809190747336,
      "grad_norm": 0.09442190080881119,
      "learning_rate": 0.0017837343892780993,
      "loss": 2.6908416748046875,
      "step": 6522
    },
    {
      "epoch": 0.8976561048824842,
      "grad_norm": 0.06643743813037872,
      "learning_rate": 0.0017825159914712154,
      "loss": 2.606438159942627,
      "step": 6524
    },
    {
      "epoch": 0.8979312906902348,
      "grad_norm": 0.06564697623252869,
      "learning_rate": 0.0017812975936643315,
      "loss": 2.61641788482666,
      "step": 6526
    },
    {
      "epoch": 0.8982064764979855,
      "grad_norm": 0.03843184933066368,
      "learning_rate": 0.0017800791958574476,
      "loss": 2.646162509918213,
      "step": 6528
    },
    {
      "epoch": 0.898481662305736,
      "grad_norm": 0.04457039386034012,
      "learning_rate": 0.0017788607980505635,
      "loss": 2.632981300354004,
      "step": 6530
    },
    {
      "epoch": 0.8987568481134867,
      "grad_norm": 0.1310417801141739,
      "learning_rate": 0.0017776424002436794,
      "loss": 2.6423935890197754,
      "step": 6532
    },
    {
      "epoch": 0.8990320339212372,
      "grad_norm": 0.02896060049533844,
      "learning_rate": 0.0017764240024367957,
      "loss": 2.642648935317993,
      "step": 6534
    },
    {
      "epoch": 0.8993072197289879,
      "grad_norm": 0.18990756571292877,
      "learning_rate": 0.0017752056046299116,
      "loss": 2.6497602462768555,
      "step": 6536
    },
    {
      "epoch": 0.8995824055367384,
      "grad_norm": 0.03734392300248146,
      "learning_rate": 0.0017739872068230278,
      "loss": 2.643141269683838,
      "step": 6538
    },
    {
      "epoch": 0.8998575913444891,
      "grad_norm": 0.2395292967557907,
      "learning_rate": 0.0017727688090161439,
      "loss": 2.667360305786133,
      "step": 6540
    },
    {
      "epoch": 0.9001327771522397,
      "grad_norm": 0.07680654525756836,
      "learning_rate": 0.00177155041120926,
      "loss": 2.658520221710205,
      "step": 6542
    },
    {
      "epoch": 0.9004079629599903,
      "grad_norm": 0.1795373409986496,
      "learning_rate": 0.001770332013402376,
      "loss": 2.6830413341522217,
      "step": 6544
    },
    {
      "epoch": 0.9006831487677409,
      "grad_norm": 0.07581128925085068,
      "learning_rate": 0.0017691136155954918,
      "loss": 2.703702926635742,
      "step": 6546
    },
    {
      "epoch": 0.9009583345754915,
      "grad_norm": 0.1496330201625824,
      "learning_rate": 0.0017678952177886081,
      "loss": 2.647918701171875,
      "step": 6548
    },
    {
      "epoch": 0.9012335203832421,
      "grad_norm": 0.050308603793382645,
      "learning_rate": 0.001766676819981724,
      "loss": 2.706932783126831,
      "step": 6550
    },
    {
      "epoch": 0.9015087061909927,
      "grad_norm": 0.220636248588562,
      "learning_rate": 0.0017654584221748401,
      "loss": 2.6865949630737305,
      "step": 6552
    },
    {
      "epoch": 0.9017838919987433,
      "grad_norm": 0.06944242119789124,
      "learning_rate": 0.0017642400243679563,
      "loss": 2.6494059562683105,
      "step": 6554
    },
    {
      "epoch": 0.902059077806494,
      "grad_norm": 0.18680645525455475,
      "learning_rate": 0.0017630216265610722,
      "loss": 2.6472110748291016,
      "step": 6556
    },
    {
      "epoch": 0.9023342636142445,
      "grad_norm": 0.032905321568250656,
      "learning_rate": 0.0017618032287541883,
      "loss": 2.6532888412475586,
      "step": 6558
    },
    {
      "epoch": 0.9026094494219952,
      "grad_norm": 0.16976074874401093,
      "learning_rate": 0.0017605848309473042,
      "loss": 2.6369435787200928,
      "step": 6560
    },
    {
      "epoch": 0.9028846352297457,
      "grad_norm": 0.03373270854353905,
      "learning_rate": 0.0017593664331404205,
      "loss": 2.668546676635742,
      "step": 6562
    },
    {
      "epoch": 0.9031598210374964,
      "grad_norm": 0.03433596342802048,
      "learning_rate": 0.0017581480353335364,
      "loss": 2.7157182693481445,
      "step": 6564
    },
    {
      "epoch": 0.9034350068452469,
      "grad_norm": 0.15871189534664154,
      "learning_rate": 0.0017569296375266525,
      "loss": 2.6356911659240723,
      "step": 6566
    },
    {
      "epoch": 0.9037101926529976,
      "grad_norm": 0.17083221673965454,
      "learning_rate": 0.0017557112397197686,
      "loss": 2.657536029815674,
      "step": 6568
    },
    {
      "epoch": 0.9039853784607482,
      "grad_norm": 0.1934972107410431,
      "learning_rate": 0.0017544928419128845,
      "loss": 2.6635589599609375,
      "step": 6570
    },
    {
      "epoch": 0.9042605642684988,
      "grad_norm": 0.1409066617488861,
      "learning_rate": 0.0017532744441060007,
      "loss": 2.642958641052246,
      "step": 6572
    },
    {
      "epoch": 0.9045357500762494,
      "grad_norm": 0.11927130073308945,
      "learning_rate": 0.0017520560462991165,
      "loss": 2.6283228397369385,
      "step": 6574
    },
    {
      "epoch": 0.904810935884,
      "grad_norm": 0.20614635944366455,
      "learning_rate": 0.0017508376484922329,
      "loss": 2.6250486373901367,
      "step": 6576
    },
    {
      "epoch": 0.9050861216917506,
      "grad_norm": 0.10790194571018219,
      "learning_rate": 0.0017496192506853488,
      "loss": 2.6805667877197266,
      "step": 6578
    },
    {
      "epoch": 0.9053613074995013,
      "grad_norm": 0.1454182118177414,
      "learning_rate": 0.001748400852878465,
      "loss": 2.6362338066101074,
      "step": 6580
    },
    {
      "epoch": 0.9056364933072518,
      "grad_norm": 0.291746586561203,
      "learning_rate": 0.001747182455071581,
      "loss": 2.6358280181884766,
      "step": 6582
    },
    {
      "epoch": 0.9059116791150025,
      "grad_norm": 0.03922594338655472,
      "learning_rate": 0.001745964057264697,
      "loss": 2.6795201301574707,
      "step": 6584
    },
    {
      "epoch": 0.906186864922753,
      "grad_norm": 0.25765714049339294,
      "learning_rate": 0.001744745659457813,
      "loss": 2.6835856437683105,
      "step": 6586
    },
    {
      "epoch": 0.9064620507305037,
      "grad_norm": 0.1267739087343216,
      "learning_rate": 0.001743527261650929,
      "loss": 2.675701141357422,
      "step": 6588
    },
    {
      "epoch": 0.9067372365382542,
      "grad_norm": 0.036614272743463516,
      "learning_rate": 0.0017423088638440453,
      "loss": 2.703756809234619,
      "step": 6590
    },
    {
      "epoch": 0.9070124223460049,
      "grad_norm": 0.12097303569316864,
      "learning_rate": 0.0017410904660371612,
      "loss": 2.631957530975342,
      "step": 6592
    },
    {
      "epoch": 0.9072876081537555,
      "grad_norm": 0.12054899334907532,
      "learning_rate": 0.0017398720682302773,
      "loss": 2.682358980178833,
      "step": 6594
    },
    {
      "epoch": 0.9075627939615061,
      "grad_norm": 0.037118714302778244,
      "learning_rate": 0.0017386536704233934,
      "loss": 2.688812255859375,
      "step": 6596
    },
    {
      "epoch": 0.9078379797692567,
      "grad_norm": 0.11533883213996887,
      "learning_rate": 0.0017374352726165093,
      "loss": 2.622849941253662,
      "step": 6598
    },
    {
      "epoch": 0.9081131655770073,
      "grad_norm": 0.06812402606010437,
      "learning_rate": 0.0017362168748096254,
      "loss": 2.6519365310668945,
      "step": 6600
    },
    {
      "epoch": 0.9083883513847579,
      "grad_norm": 0.1060088500380516,
      "learning_rate": 0.0017349984770027413,
      "loss": 2.6804933547973633,
      "step": 6602
    },
    {
      "epoch": 0.9086635371925085,
      "grad_norm": 0.07915689051151276,
      "learning_rate": 0.0017337800791958576,
      "loss": 2.5967812538146973,
      "step": 6604
    },
    {
      "epoch": 0.9089387230002591,
      "grad_norm": 0.1505192071199417,
      "learning_rate": 0.0017325616813889735,
      "loss": 2.6988584995269775,
      "step": 6606
    },
    {
      "epoch": 0.9092139088080098,
      "grad_norm": 0.05840275064110756,
      "learning_rate": 0.0017313432835820894,
      "loss": 2.61954402923584,
      "step": 6608
    },
    {
      "epoch": 0.9094890946157603,
      "grad_norm": 0.10516121983528137,
      "learning_rate": 0.0017301248857752056,
      "loss": 2.6730024814605713,
      "step": 6610
    },
    {
      "epoch": 0.909764280423511,
      "grad_norm": 0.03566355258226395,
      "learning_rate": 0.0017289064879683217,
      "loss": 2.6905176639556885,
      "step": 6612
    },
    {
      "epoch": 0.9100394662312615,
      "grad_norm": 0.1506408154964447,
      "learning_rate": 0.0017276880901614378,
      "loss": 2.6844961643218994,
      "step": 6614
    },
    {
      "epoch": 0.9103146520390122,
      "grad_norm": 0.04962019622325897,
      "learning_rate": 0.0017264696923545537,
      "loss": 2.656100034713745,
      "step": 6616
    },
    {
      "epoch": 0.9105898378467627,
      "grad_norm": 0.25605514645576477,
      "learning_rate": 0.00172525129454767,
      "loss": 2.6291871070861816,
      "step": 6618
    },
    {
      "epoch": 0.9108650236545134,
      "grad_norm": 0.03658574819564819,
      "learning_rate": 0.001724032896740786,
      "loss": 2.6315317153930664,
      "step": 6620
    },
    {
      "epoch": 0.911140209462264,
      "grad_norm": 0.13665716350078583,
      "learning_rate": 0.0017228144989339018,
      "loss": 2.6262221336364746,
      "step": 6622
    },
    {
      "epoch": 0.9114153952700146,
      "grad_norm": 0.06857297569513321,
      "learning_rate": 0.001721596101127018,
      "loss": 2.6585636138916016,
      "step": 6624
    },
    {
      "epoch": 0.9116905810777652,
      "grad_norm": 0.08541840314865112,
      "learning_rate": 0.001720377703320134,
      "loss": 2.671811819076538,
      "step": 6626
    },
    {
      "epoch": 0.9119657668855158,
      "grad_norm": 0.0989607721567154,
      "learning_rate": 0.0017191593055132502,
      "loss": 2.6452786922454834,
      "step": 6628
    },
    {
      "epoch": 0.9122409526932664,
      "grad_norm": 0.04168534278869629,
      "learning_rate": 0.001717940907706366,
      "loss": 2.6810245513916016,
      "step": 6630
    },
    {
      "epoch": 0.9125161385010171,
      "grad_norm": 0.07653797417879105,
      "learning_rate": 0.0017167225098994824,
      "loss": 2.654320240020752,
      "step": 6632
    },
    {
      "epoch": 0.9127913243087676,
      "grad_norm": 0.12085521221160889,
      "learning_rate": 0.0017155041120925983,
      "loss": 2.6778974533081055,
      "step": 6634
    },
    {
      "epoch": 0.9130665101165183,
      "grad_norm": 0.3071478307247162,
      "learning_rate": 0.0017142857142857142,
      "loss": 2.664992332458496,
      "step": 6636
    },
    {
      "epoch": 0.9133416959242688,
      "grad_norm": 0.14519698917865753,
      "learning_rate": 0.0017130673164788303,
      "loss": 2.6285195350646973,
      "step": 6638
    },
    {
      "epoch": 0.9136168817320195,
      "grad_norm": 0.27480822801589966,
      "learning_rate": 0.0017118489186719464,
      "loss": 2.614973783493042,
      "step": 6640
    },
    {
      "epoch": 0.91389206753977,
      "grad_norm": 0.2958192229270935,
      "learning_rate": 0.0017106305208650625,
      "loss": 2.6757044792175293,
      "step": 6642
    },
    {
      "epoch": 0.9141672533475207,
      "grad_norm": 0.08222149312496185,
      "learning_rate": 0.0017094121230581784,
      "loss": 2.6096901893615723,
      "step": 6644
    },
    {
      "epoch": 0.9144424391552713,
      "grad_norm": 0.09665880352258682,
      "learning_rate": 0.0017081937252512948,
      "loss": 2.641566514968872,
      "step": 6646
    },
    {
      "epoch": 0.9147176249630219,
      "grad_norm": 0.11861685663461685,
      "learning_rate": 0.0017069753274444107,
      "loss": 2.6713132858276367,
      "step": 6648
    },
    {
      "epoch": 0.9149928107707725,
      "grad_norm": 0.06126582249999046,
      "learning_rate": 0.0017057569296375266,
      "loss": 2.6388401985168457,
      "step": 6650
    },
    {
      "epoch": 0.9152679965785231,
      "grad_norm": 0.03147628530859947,
      "learning_rate": 0.0017045385318306427,
      "loss": 2.655320644378662,
      "step": 6652
    },
    {
      "epoch": 0.9155431823862737,
      "grad_norm": 0.03623277693986893,
      "learning_rate": 0.0017033201340237588,
      "loss": 2.638939619064331,
      "step": 6654
    },
    {
      "epoch": 0.9158183681940243,
      "grad_norm": 0.03249303624033928,
      "learning_rate": 0.001702101736216875,
      "loss": 2.702404737472534,
      "step": 6656
    },
    {
      "epoch": 0.916093554001775,
      "grad_norm": 0.0766565352678299,
      "learning_rate": 0.0017008833384099908,
      "loss": 2.654101848602295,
      "step": 6658
    },
    {
      "epoch": 0.9163687398095256,
      "grad_norm": 0.029498621821403503,
      "learning_rate": 0.0016996649406031072,
      "loss": 2.624237060546875,
      "step": 6660
    },
    {
      "epoch": 0.9166439256172761,
      "grad_norm": 0.11189699918031693,
      "learning_rate": 0.001698446542796223,
      "loss": 2.6424217224121094,
      "step": 6662
    },
    {
      "epoch": 0.9169191114250268,
      "grad_norm": 0.03601978346705437,
      "learning_rate": 0.001697228144989339,
      "loss": 2.6646060943603516,
      "step": 6664
    },
    {
      "epoch": 0.9171942972327773,
      "grad_norm": 0.0982208326458931,
      "learning_rate": 0.001696009747182455,
      "loss": 2.6592917442321777,
      "step": 6666
    },
    {
      "epoch": 0.917469483040528,
      "grad_norm": 0.08821946382522583,
      "learning_rate": 0.0016947913493755712,
      "loss": 2.6457462310791016,
      "step": 6668
    },
    {
      "epoch": 0.9177446688482785,
      "grad_norm": 0.15712201595306396,
      "learning_rate": 0.0016935729515686873,
      "loss": 2.639554262161255,
      "step": 6670
    },
    {
      "epoch": 0.9180198546560292,
      "grad_norm": 0.027122966945171356,
      "learning_rate": 0.0016923545537618032,
      "loss": 2.660430431365967,
      "step": 6672
    },
    {
      "epoch": 0.9182950404637799,
      "grad_norm": 0.09468477964401245,
      "learning_rate": 0.001691136155954919,
      "loss": 2.649712085723877,
      "step": 6674
    },
    {
      "epoch": 0.9185702262715304,
      "grad_norm": 0.1120760515332222,
      "learning_rate": 0.0016899177581480354,
      "loss": 2.624619960784912,
      "step": 6676
    },
    {
      "epoch": 0.918845412079281,
      "grad_norm": 0.24221181869506836,
      "learning_rate": 0.0016886993603411513,
      "loss": 2.6635799407958984,
      "step": 6678
    },
    {
      "epoch": 0.9191205978870316,
      "grad_norm": 0.033889900892972946,
      "learning_rate": 0.0016874809625342675,
      "loss": 2.6858596801757812,
      "step": 6680
    },
    {
      "epoch": 0.9193957836947823,
      "grad_norm": 0.32424429059028625,
      "learning_rate": 0.0016862625647273836,
      "loss": 2.5911643505096436,
      "step": 6682
    },
    {
      "epoch": 0.9196709695025329,
      "grad_norm": 0.12901155650615692,
      "learning_rate": 0.0016850441669204997,
      "loss": 2.6243910789489746,
      "step": 6684
    },
    {
      "epoch": 0.9199461553102835,
      "grad_norm": 0.16346824169158936,
      "learning_rate": 0.0016838257691136156,
      "loss": 2.685324192047119,
      "step": 6686
    },
    {
      "epoch": 0.9202213411180341,
      "grad_norm": 0.16586409509181976,
      "learning_rate": 0.0016826073713067315,
      "loss": 2.615936756134033,
      "step": 6688
    },
    {
      "epoch": 0.9204965269257847,
      "grad_norm": 0.1712968498468399,
      "learning_rate": 0.0016813889734998478,
      "loss": 2.646632194519043,
      "step": 6690
    },
    {
      "epoch": 0.9207717127335353,
      "grad_norm": 0.05835619568824768,
      "learning_rate": 0.0016801705756929637,
      "loss": 2.6540746688842773,
      "step": 6692
    },
    {
      "epoch": 0.9210468985412859,
      "grad_norm": 0.2211555391550064,
      "learning_rate": 0.0016789521778860798,
      "loss": 2.6513261795043945,
      "step": 6694
    },
    {
      "epoch": 0.9213220843490365,
      "grad_norm": 0.06987286359071732,
      "learning_rate": 0.001677733780079196,
      "loss": 2.63887357711792,
      "step": 6696
    },
    {
      "epoch": 0.9215972701567872,
      "grad_norm": 0.08902879804372787,
      "learning_rate": 0.001676515382272312,
      "loss": 2.649014472961426,
      "step": 6698
    },
    {
      "epoch": 0.9218724559645377,
      "grad_norm": 0.15348830819129944,
      "learning_rate": 0.001675296984465428,
      "loss": 2.622760772705078,
      "step": 6700
    },
    {
      "epoch": 0.9221476417722884,
      "grad_norm": 0.04026218503713608,
      "learning_rate": 0.0016740785866585439,
      "loss": 2.6251869201660156,
      "step": 6702
    },
    {
      "epoch": 0.9224228275800389,
      "grad_norm": 0.09073542058467865,
      "learning_rate": 0.0016728601888516602,
      "loss": 2.624483108520508,
      "step": 6704
    },
    {
      "epoch": 0.9226980133877896,
      "grad_norm": 0.19347932934761047,
      "learning_rate": 0.001671641791044776,
      "loss": 2.609808921813965,
      "step": 6706
    },
    {
      "epoch": 0.9229731991955401,
      "grad_norm": 0.031187426298856735,
      "learning_rate": 0.0016704233932378922,
      "loss": 2.6367087364196777,
      "step": 6708
    },
    {
      "epoch": 0.9232483850032908,
      "grad_norm": 0.1109781488776207,
      "learning_rate": 0.0016692049954310083,
      "loss": 2.6506285667419434,
      "step": 6710
    },
    {
      "epoch": 0.9235235708110414,
      "grad_norm": 0.098908931016922,
      "learning_rate": 0.0016679865976241244,
      "loss": 2.6847376823425293,
      "step": 6712
    },
    {
      "epoch": 0.923798756618792,
      "grad_norm": 0.036317888647317886,
      "learning_rate": 0.0016667681998172403,
      "loss": 2.6162540912628174,
      "step": 6714
    },
    {
      "epoch": 0.9240739424265426,
      "grad_norm": 0.04467134550213814,
      "learning_rate": 0.0016655498020103562,
      "loss": 2.6234898567199707,
      "step": 6716
    },
    {
      "epoch": 0.9243491282342932,
      "grad_norm": 0.03880814090371132,
      "learning_rate": 0.0016643314042034726,
      "loss": 2.634986400604248,
      "step": 6718
    },
    {
      "epoch": 0.9246243140420438,
      "grad_norm": 0.0914827287197113,
      "learning_rate": 0.0016631130063965885,
      "loss": 2.6685690879821777,
      "step": 6720
    },
    {
      "epoch": 0.9248994998497944,
      "grad_norm": 0.053054165095090866,
      "learning_rate": 0.0016618946085897046,
      "loss": 2.634718418121338,
      "step": 6722
    },
    {
      "epoch": 0.925174685657545,
      "grad_norm": 0.12835340201854706,
      "learning_rate": 0.0016606762107828207,
      "loss": 2.6429131031036377,
      "step": 6724
    },
    {
      "epoch": 0.9254498714652957,
      "grad_norm": 0.0689706951379776,
      "learning_rate": 0.0016594578129759368,
      "loss": 2.624847412109375,
      "step": 6726
    },
    {
      "epoch": 0.9257250572730462,
      "grad_norm": 0.04198030009865761,
      "learning_rate": 0.0016582394151690527,
      "loss": 2.665191173553467,
      "step": 6728
    },
    {
      "epoch": 0.9260002430807969,
      "grad_norm": 0.079593725502491,
      "learning_rate": 0.0016570210173621686,
      "loss": 2.6162333488464355,
      "step": 6730
    },
    {
      "epoch": 0.9262754288885474,
      "grad_norm": 0.12944401800632477,
      "learning_rate": 0.001655802619555285,
      "loss": 2.649031639099121,
      "step": 6732
    },
    {
      "epoch": 0.9265506146962981,
      "grad_norm": 0.267687052488327,
      "learning_rate": 0.0016545842217484009,
      "loss": 2.6180882453918457,
      "step": 6734
    },
    {
      "epoch": 0.9268258005040487,
      "grad_norm": 0.06829001754522324,
      "learning_rate": 0.001653365823941517,
      "loss": 2.624816417694092,
      "step": 6736
    },
    {
      "epoch": 0.9271009863117993,
      "grad_norm": 0.2451668530702591,
      "learning_rate": 0.001652147426134633,
      "loss": 2.6640334129333496,
      "step": 6738
    },
    {
      "epoch": 0.9273761721195499,
      "grad_norm": 0.17136549949645996,
      "learning_rate": 0.001650929028327749,
      "loss": 2.6492726802825928,
      "step": 6740
    },
    {
      "epoch": 0.9276513579273005,
      "grad_norm": 0.08367171883583069,
      "learning_rate": 0.001649710630520865,
      "loss": 2.6750407218933105,
      "step": 6742
    },
    {
      "epoch": 0.9279265437350511,
      "grad_norm": 0.1716756522655487,
      "learning_rate": 0.001648492232713981,
      "loss": 2.650629997253418,
      "step": 6744
    },
    {
      "epoch": 0.9282017295428017,
      "grad_norm": 0.11342863738536835,
      "learning_rate": 0.0016472738349070973,
      "loss": 2.6375951766967773,
      "step": 6746
    },
    {
      "epoch": 0.9284769153505523,
      "grad_norm": 0.0298096165060997,
      "learning_rate": 0.0016460554371002132,
      "loss": 2.6400461196899414,
      "step": 6748
    },
    {
      "epoch": 0.928752101158303,
      "grad_norm": 0.11896102875471115,
      "learning_rate": 0.0016448370392933293,
      "loss": 2.6542325019836426,
      "step": 6750
    },
    {
      "epoch": 0.9290272869660535,
      "grad_norm": 0.060931216925382614,
      "learning_rate": 0.0016436186414864455,
      "loss": 2.649873733520508,
      "step": 6752
    },
    {
      "epoch": 0.9293024727738042,
      "grad_norm": 0.031189749017357826,
      "learning_rate": 0.0016424002436795614,
      "loss": 2.598505973815918,
      "step": 6754
    },
    {
      "epoch": 0.9295776585815547,
      "grad_norm": 0.03487801179289818,
      "learning_rate": 0.0016411818458726775,
      "loss": 2.661215305328369,
      "step": 6756
    },
    {
      "epoch": 0.9298528443893054,
      "grad_norm": 0.02732064388692379,
      "learning_rate": 0.0016399634480657934,
      "loss": 2.5778276920318604,
      "step": 6758
    },
    {
      "epoch": 0.9301280301970559,
      "grad_norm": 0.06612678617238998,
      "learning_rate": 0.0016387450502589097,
      "loss": 2.6404900550842285,
      "step": 6760
    },
    {
      "epoch": 0.9304032160048066,
      "grad_norm": 0.04485834762454033,
      "learning_rate": 0.0016375266524520256,
      "loss": 2.6489791870117188,
      "step": 6762
    },
    {
      "epoch": 0.9306784018125572,
      "grad_norm": 0.1352410763502121,
      "learning_rate": 0.0016363082546451417,
      "loss": 2.6437559127807617,
      "step": 6764
    },
    {
      "epoch": 0.9309535876203078,
      "grad_norm": 0.11785218119621277,
      "learning_rate": 0.0016350898568382576,
      "loss": 2.649991035461426,
      "step": 6766
    },
    {
      "epoch": 0.9312287734280584,
      "grad_norm": 0.09513581544160843,
      "learning_rate": 0.0016338714590313737,
      "loss": 2.659546136856079,
      "step": 6768
    },
    {
      "epoch": 0.931503959235809,
      "grad_norm": 0.15663573145866394,
      "learning_rate": 0.0016326530612244899,
      "loss": 2.644207000732422,
      "step": 6770
    },
    {
      "epoch": 0.9317791450435596,
      "grad_norm": 0.10862938314676285,
      "learning_rate": 0.0016314346634176058,
      "loss": 2.629103660583496,
      "step": 6772
    },
    {
      "epoch": 0.9320543308513102,
      "grad_norm": 0.073142409324646,
      "learning_rate": 0.001630216265610722,
      "loss": 2.644327402114868,
      "step": 6774
    },
    {
      "epoch": 0.9323295166590608,
      "grad_norm": 0.03474358096718788,
      "learning_rate": 0.001628997867803838,
      "loss": 2.657970428466797,
      "step": 6776
    },
    {
      "epoch": 0.9326047024668115,
      "grad_norm": 0.060403596609830856,
      "learning_rate": 0.001627779469996954,
      "loss": 2.613652229309082,
      "step": 6778
    },
    {
      "epoch": 0.932879888274562,
      "grad_norm": 0.1330999732017517,
      "learning_rate": 0.00162656107219007,
      "loss": 2.6054043769836426,
      "step": 6780
    },
    {
      "epoch": 0.9331550740823127,
      "grad_norm": 0.0399349108338356,
      "learning_rate": 0.0016253426743831861,
      "loss": 2.646510124206543,
      "step": 6782
    },
    {
      "epoch": 0.9334302598900632,
      "grad_norm": 0.2321109175682068,
      "learning_rate": 0.0016241242765763022,
      "loss": 2.6318113803863525,
      "step": 6784
    },
    {
      "epoch": 0.9337054456978139,
      "grad_norm": 0.0692065954208374,
      "learning_rate": 0.0016229058787694181,
      "loss": 2.66621732711792,
      "step": 6786
    },
    {
      "epoch": 0.9339806315055645,
      "grad_norm": 0.19802933931350708,
      "learning_rate": 0.0016216874809625345,
      "loss": 2.6276750564575195,
      "step": 6788
    },
    {
      "epoch": 0.9342558173133151,
      "grad_norm": 0.063431516289711,
      "learning_rate": 0.0016204690831556504,
      "loss": 2.6322877407073975,
      "step": 6790
    },
    {
      "epoch": 0.9345310031210657,
      "grad_norm": 0.18695349991321564,
      "learning_rate": 0.0016192506853487665,
      "loss": 2.65260910987854,
      "step": 6792
    },
    {
      "epoch": 0.9348061889288163,
      "grad_norm": 0.031321942806243896,
      "learning_rate": 0.0016180322875418824,
      "loss": 2.643073558807373,
      "step": 6794
    },
    {
      "epoch": 0.9350813747365669,
      "grad_norm": 0.056078944355249405,
      "learning_rate": 0.0016168138897349985,
      "loss": 2.6386890411376953,
      "step": 6796
    },
    {
      "epoch": 0.9353565605443175,
      "grad_norm": 0.16480103135108948,
      "learning_rate": 0.0016155954919281146,
      "loss": 2.6243138313293457,
      "step": 6798
    },
    {
      "epoch": 0.9356317463520681,
      "grad_norm": 0.08653315156698227,
      "learning_rate": 0.0016143770941212305,
      "loss": 2.6335296630859375,
      "step": 6800
    },
    {
      "epoch": 0.9359069321598188,
      "grad_norm": 0.17658308148384094,
      "learning_rate": 0.0016131586963143468,
      "loss": 2.64863657951355,
      "step": 6802
    },
    {
      "epoch": 0.9361821179675693,
      "grad_norm": 0.03427954763174057,
      "learning_rate": 0.0016119402985074627,
      "loss": 2.6591415405273438,
      "step": 6804
    },
    {
      "epoch": 0.93645730377532,
      "grad_norm": 0.08648452907800674,
      "learning_rate": 0.0016107219007005786,
      "loss": 2.6294209957122803,
      "step": 6806
    },
    {
      "epoch": 0.9367324895830705,
      "grad_norm": 0.08697447925806046,
      "learning_rate": 0.0016095035028936948,
      "loss": 2.6383605003356934,
      "step": 6808
    },
    {
      "epoch": 0.9370076753908212,
      "grad_norm": 0.03419235721230507,
      "learning_rate": 0.0016082851050868109,
      "loss": 2.6524131298065186,
      "step": 6810
    },
    {
      "epoch": 0.9372828611985717,
      "grad_norm": 0.0999702662229538,
      "learning_rate": 0.001607066707279927,
      "loss": 2.6279449462890625,
      "step": 6812
    },
    {
      "epoch": 0.9375580470063224,
      "grad_norm": 0.14719051122665405,
      "learning_rate": 0.001605848309473043,
      "loss": 2.6493544578552246,
      "step": 6814
    },
    {
      "epoch": 0.937833232814073,
      "grad_norm": 0.11975278705358505,
      "learning_rate": 0.0016046299116661592,
      "loss": 2.6281466484069824,
      "step": 6816
    },
    {
      "epoch": 0.9381084186218236,
      "grad_norm": 0.1915886402130127,
      "learning_rate": 0.0016034115138592751,
      "loss": 2.647474527359009,
      "step": 6818
    },
    {
      "epoch": 0.9383836044295742,
      "grad_norm": 0.1014411449432373,
      "learning_rate": 0.001602193116052391,
      "loss": 2.6216700077056885,
      "step": 6820
    },
    {
      "epoch": 0.9386587902373248,
      "grad_norm": 0.16280868649482727,
      "learning_rate": 0.0016009747182455071,
      "loss": 2.59242844581604,
      "step": 6822
    },
    {
      "epoch": 0.9389339760450754,
      "grad_norm": 0.09852806478738785,
      "learning_rate": 0.0015997563204386233,
      "loss": 2.6655678749084473,
      "step": 6824
    },
    {
      "epoch": 0.939209161852826,
      "grad_norm": 0.08603417128324509,
      "learning_rate": 0.0015985379226317394,
      "loss": 2.6617326736450195,
      "step": 6826
    },
    {
      "epoch": 0.9394843476605766,
      "grad_norm": 0.09832686930894852,
      "learning_rate": 0.0015973195248248553,
      "loss": 2.639188289642334,
      "step": 6828
    },
    {
      "epoch": 0.9397595334683273,
      "grad_norm": 0.07388213276863098,
      "learning_rate": 0.0015961011270179716,
      "loss": 2.6232142448425293,
      "step": 6830
    },
    {
      "epoch": 0.9400347192760778,
      "grad_norm": 0.08974307775497437,
      "learning_rate": 0.0015948827292110875,
      "loss": 2.5991287231445312,
      "step": 6832
    },
    {
      "epoch": 0.9403099050838285,
      "grad_norm": 0.02852836810052395,
      "learning_rate": 0.0015936643314042034,
      "loss": 2.651249408721924,
      "step": 6834
    },
    {
      "epoch": 0.940585090891579,
      "grad_norm": 0.04934679716825485,
      "learning_rate": 0.0015924459335973195,
      "loss": 2.6374642848968506,
      "step": 6836
    },
    {
      "epoch": 0.9408602766993297,
      "grad_norm": 0.09404280036687851,
      "learning_rate": 0.0015912275357904356,
      "loss": 2.653052806854248,
      "step": 6838
    },
    {
      "epoch": 0.9411354625070804,
      "grad_norm": 0.10855881869792938,
      "learning_rate": 0.0015900091379835518,
      "loss": 2.638598680496216,
      "step": 6840
    },
    {
      "epoch": 0.9414106483148309,
      "grad_norm": 0.08627064526081085,
      "learning_rate": 0.0015887907401766677,
      "loss": 2.6295971870422363,
      "step": 6842
    },
    {
      "epoch": 0.9416858341225816,
      "grad_norm": 0.13687622547149658,
      "learning_rate": 0.0015875723423697838,
      "loss": 2.6381707191467285,
      "step": 6844
    },
    {
      "epoch": 0.9419610199303321,
      "grad_norm": 0.07047820836305618,
      "learning_rate": 0.0015863539445628999,
      "loss": 2.621323585510254,
      "step": 6846
    },
    {
      "epoch": 0.9422362057380828,
      "grad_norm": 0.03287198022007942,
      "learning_rate": 0.0015851355467560158,
      "loss": 2.6209282875061035,
      "step": 6848
    },
    {
      "epoch": 0.9425113915458333,
      "grad_norm": 0.03700323402881622,
      "learning_rate": 0.001583917148949132,
      "loss": 2.6369616985321045,
      "step": 6850
    },
    {
      "epoch": 0.942786577353584,
      "grad_norm": 0.04219698905944824,
      "learning_rate": 0.001582698751142248,
      "loss": 2.636697292327881,
      "step": 6852
    },
    {
      "epoch": 0.9430617631613346,
      "grad_norm": 0.04258700832724571,
      "learning_rate": 0.0015814803533353641,
      "loss": 2.6562390327453613,
      "step": 6854
    },
    {
      "epoch": 0.9433369489690852,
      "grad_norm": 0.16148357093334198,
      "learning_rate": 0.00158026195552848,
      "loss": 2.6445980072021484,
      "step": 6856
    },
    {
      "epoch": 0.9436121347768358,
      "grad_norm": 0.28962039947509766,
      "learning_rate": 0.001579043557721596,
      "loss": 2.6122801303863525,
      "step": 6858
    },
    {
      "epoch": 0.9438873205845864,
      "grad_norm": 0.08587105572223663,
      "learning_rate": 0.0015778251599147123,
      "loss": 2.6037654876708984,
      "step": 6860
    },
    {
      "epoch": 0.944162506392337,
      "grad_norm": 0.1699834167957306,
      "learning_rate": 0.0015766067621078282,
      "loss": 2.66497802734375,
      "step": 6862
    },
    {
      "epoch": 0.9444376922000876,
      "grad_norm": 0.2618996202945709,
      "learning_rate": 0.0015753883643009443,
      "loss": 2.6021485328674316,
      "step": 6864
    },
    {
      "epoch": 0.9447128780078382,
      "grad_norm": 0.25282445549964905,
      "learning_rate": 0.0015741699664940604,
      "loss": 2.6442925930023193,
      "step": 6866
    },
    {
      "epoch": 0.9449880638155889,
      "grad_norm": 0.028723299503326416,
      "learning_rate": 0.0015729515686871765,
      "loss": 2.644653558731079,
      "step": 6868
    },
    {
      "epoch": 0.9452632496233394,
      "grad_norm": 0.03542667627334595,
      "learning_rate": 0.0015717331708802924,
      "loss": 2.6178126335144043,
      "step": 6870
    },
    {
      "epoch": 0.9455384354310901,
      "grad_norm": 0.12110951542854309,
      "learning_rate": 0.0015705147730734083,
      "loss": 2.640578508377075,
      "step": 6872
    },
    {
      "epoch": 0.9458136212388406,
      "grad_norm": 0.09158312529325485,
      "learning_rate": 0.0015692963752665246,
      "loss": 2.6219465732574463,
      "step": 6874
    },
    {
      "epoch": 0.9460888070465913,
      "grad_norm": 0.027310488745570183,
      "learning_rate": 0.0015680779774596405,
      "loss": 2.6402463912963867,
      "step": 6876
    },
    {
      "epoch": 0.9463639928543418,
      "grad_norm": 0.04234203323721886,
      "learning_rate": 0.0015668595796527567,
      "loss": 2.625041961669922,
      "step": 6878
    },
    {
      "epoch": 0.9466391786620925,
      "grad_norm": 0.050247255712747574,
      "learning_rate": 0.0015656411818458728,
      "loss": 2.623518705368042,
      "step": 6880
    },
    {
      "epoch": 0.9469143644698431,
      "grad_norm": 0.04597342759370804,
      "learning_rate": 0.0015644227840389889,
      "loss": 2.6286439895629883,
      "step": 6882
    },
    {
      "epoch": 0.9471895502775937,
      "grad_norm": 0.09568143635988235,
      "learning_rate": 0.0015632043862321048,
      "loss": 2.660066604614258,
      "step": 6884
    },
    {
      "epoch": 0.9474647360853443,
      "grad_norm": 0.07130748778581619,
      "learning_rate": 0.0015619859884252207,
      "loss": 2.602458953857422,
      "step": 6886
    },
    {
      "epoch": 0.9477399218930949,
      "grad_norm": 0.19874243438243866,
      "learning_rate": 0.001560767590618337,
      "loss": 2.629405975341797,
      "step": 6888
    },
    {
      "epoch": 0.9480151077008455,
      "grad_norm": 0.08318581432104111,
      "learning_rate": 0.001559549192811453,
      "loss": 2.6268184185028076,
      "step": 6890
    },
    {
      "epoch": 0.9482902935085962,
      "grad_norm": 0.04603923484683037,
      "learning_rate": 0.001558330795004569,
      "loss": 2.6374335289001465,
      "step": 6892
    },
    {
      "epoch": 0.9485654793163467,
      "grad_norm": 0.04889914020895958,
      "learning_rate": 0.0015571123971976852,
      "loss": 2.626732110977173,
      "step": 6894
    },
    {
      "epoch": 0.9488406651240974,
      "grad_norm": 0.03131958469748497,
      "learning_rate": 0.0015558939993908013,
      "loss": 2.605618953704834,
      "step": 6896
    },
    {
      "epoch": 0.9491158509318479,
      "grad_norm": 0.08033682405948639,
      "learning_rate": 0.0015546756015839172,
      "loss": 2.6494603157043457,
      "step": 6898
    },
    {
      "epoch": 0.9493910367395986,
      "grad_norm": 0.14173734188079834,
      "learning_rate": 0.001553457203777033,
      "loss": 2.6359286308288574,
      "step": 6900
    },
    {
      "epoch": 0.9496662225473491,
      "grad_norm": 0.0822206661105156,
      "learning_rate": 0.0015522388059701494,
      "loss": 2.622056484222412,
      "step": 6902
    },
    {
      "epoch": 0.9499414083550998,
      "grad_norm": 0.22719734907150269,
      "learning_rate": 0.0015510204081632653,
      "loss": 2.6402955055236816,
      "step": 6904
    },
    {
      "epoch": 0.9502165941628504,
      "grad_norm": 0.042555756866931915,
      "learning_rate": 0.0015498020103563814,
      "loss": 2.6364307403564453,
      "step": 6906
    },
    {
      "epoch": 0.950491779970601,
      "grad_norm": 0.23308701813220978,
      "learning_rate": 0.0015485836125494975,
      "loss": 2.63128399848938,
      "step": 6908
    },
    {
      "epoch": 0.9507669657783516,
      "grad_norm": 0.11610971391201019,
      "learning_rate": 0.0015473652147426136,
      "loss": 2.5928432941436768,
      "step": 6910
    },
    {
      "epoch": 0.9510421515861022,
      "grad_norm": 0.09966244548559189,
      "learning_rate": 0.0015461468169357295,
      "loss": 2.601412296295166,
      "step": 6912
    },
    {
      "epoch": 0.9513173373938528,
      "grad_norm": 0.27853140234947205,
      "learning_rate": 0.0015449284191288454,
      "loss": 2.618870258331299,
      "step": 6914
    },
    {
      "epoch": 0.9515925232016034,
      "grad_norm": 0.21208207309246063,
      "learning_rate": 0.0015437100213219618,
      "loss": 2.662959575653076,
      "step": 6916
    },
    {
      "epoch": 0.951867709009354,
      "grad_norm": 0.05526116490364075,
      "learning_rate": 0.0015424916235150777,
      "loss": 2.5995254516601562,
      "step": 6918
    },
    {
      "epoch": 0.9521428948171047,
      "grad_norm": 0.094764843583107,
      "learning_rate": 0.0015412732257081938,
      "loss": 2.6667304039001465,
      "step": 6920
    },
    {
      "epoch": 0.9524180806248552,
      "grad_norm": 0.2119300812482834,
      "learning_rate": 0.0015400548279013097,
      "loss": 2.6424670219421387,
      "step": 6922
    },
    {
      "epoch": 0.9526932664326059,
      "grad_norm": 0.19154301285743713,
      "learning_rate": 0.0015388364300944258,
      "loss": 2.5850815773010254,
      "step": 6924
    },
    {
      "epoch": 0.9529684522403564,
      "grad_norm": 0.06523962318897247,
      "learning_rate": 0.001537618032287542,
      "loss": 2.6077451705932617,
      "step": 6926
    },
    {
      "epoch": 0.9532436380481071,
      "grad_norm": 0.03557727858424187,
      "learning_rate": 0.0015363996344806578,
      "loss": 2.632152557373047,
      "step": 6928
    },
    {
      "epoch": 0.9535188238558576,
      "grad_norm": 0.11289624124765396,
      "learning_rate": 0.0015351812366737742,
      "loss": 2.6474528312683105,
      "step": 6930
    },
    {
      "epoch": 0.9537940096636083,
      "grad_norm": 0.16980788111686707,
      "learning_rate": 0.00153396283886689,
      "loss": 2.6014270782470703,
      "step": 6932
    },
    {
      "epoch": 0.9540691954713589,
      "grad_norm": 0.028226541355252266,
      "learning_rate": 0.0015327444410600062,
      "loss": 2.614572048187256,
      "step": 6934
    },
    {
      "epoch": 0.9543443812791095,
      "grad_norm": 0.10361653566360474,
      "learning_rate": 0.001531526043253122,
      "loss": 2.6549103260040283,
      "step": 6936
    },
    {
      "epoch": 0.9546195670868601,
      "grad_norm": 0.13858453929424286,
      "learning_rate": 0.0015303076454462382,
      "loss": 2.5831122398376465,
      "step": 6938
    },
    {
      "epoch": 0.9548947528946107,
      "grad_norm": 0.03722090274095535,
      "learning_rate": 0.0015290892476393543,
      "loss": 2.635702610015869,
      "step": 6940
    },
    {
      "epoch": 0.9551699387023613,
      "grad_norm": 0.10827601701021194,
      "learning_rate": 0.0015278708498324702,
      "loss": 2.6258866786956787,
      "step": 6942
    },
    {
      "epoch": 0.955445124510112,
      "grad_norm": 0.11399087309837341,
      "learning_rate": 0.0015266524520255865,
      "loss": 2.608290910720825,
      "step": 6944
    },
    {
      "epoch": 0.9557203103178625,
      "grad_norm": 0.02917821705341339,
      "learning_rate": 0.0015254340542187024,
      "loss": 2.595740556716919,
      "step": 6946
    },
    {
      "epoch": 0.9559954961256132,
      "grad_norm": 0.029876478016376495,
      "learning_rate": 0.0015242156564118186,
      "loss": 2.663768768310547,
      "step": 6948
    },
    {
      "epoch": 0.9562706819333637,
      "grad_norm": 0.034029796719551086,
      "learning_rate": 0.0015229972586049345,
      "loss": 2.6058483123779297,
      "step": 6950
    },
    {
      "epoch": 0.9565458677411144,
      "grad_norm": 0.06314925849437714,
      "learning_rate": 0.0015217788607980506,
      "loss": 2.599595069885254,
      "step": 6952
    },
    {
      "epoch": 0.9568210535488649,
      "grad_norm": 0.04208005964756012,
      "learning_rate": 0.0015205604629911667,
      "loss": 2.6161515712738037,
      "step": 6954
    },
    {
      "epoch": 0.9570962393566156,
      "grad_norm": 0.08897571265697479,
      "learning_rate": 0.0015193420651842826,
      "loss": 2.596496105194092,
      "step": 6956
    },
    {
      "epoch": 0.9573714251643662,
      "grad_norm": 0.07518979161977768,
      "learning_rate": 0.001518123667377399,
      "loss": 2.634838104248047,
      "step": 6958
    },
    {
      "epoch": 0.9576466109721168,
      "grad_norm": 0.21973976492881775,
      "learning_rate": 0.0015169052695705148,
      "loss": 2.602720260620117,
      "step": 6960
    },
    {
      "epoch": 0.9579217967798674,
      "grad_norm": 0.08242259919643402,
      "learning_rate": 0.001515686871763631,
      "loss": 2.669229030609131,
      "step": 6962
    },
    {
      "epoch": 0.958196982587618,
      "grad_norm": 0.16068433225154877,
      "learning_rate": 0.0015144684739567468,
      "loss": 2.666870594024658,
      "step": 6964
    },
    {
      "epoch": 0.9584721683953686,
      "grad_norm": 0.10688690096139908,
      "learning_rate": 0.001513250076149863,
      "loss": 2.6340694427490234,
      "step": 6966
    },
    {
      "epoch": 0.9587473542031192,
      "grad_norm": 0.11711674183607101,
      "learning_rate": 0.001512031678342979,
      "loss": 2.625410795211792,
      "step": 6968
    },
    {
      "epoch": 0.9590225400108698,
      "grad_norm": 0.030287599191069603,
      "learning_rate": 0.001510813280536095,
      "loss": 2.6161231994628906,
      "step": 6970
    },
    {
      "epoch": 0.9592977258186205,
      "grad_norm": 0.03578935191035271,
      "learning_rate": 0.0015095948827292113,
      "loss": 2.592850923538208,
      "step": 6972
    },
    {
      "epoch": 0.959572911626371,
      "grad_norm": 0.032859690487384796,
      "learning_rate": 0.0015083764849223272,
      "loss": 2.6566882133483887,
      "step": 6974
    },
    {
      "epoch": 0.9598480974341217,
      "grad_norm": 0.041365139186382294,
      "learning_rate": 0.0015071580871154433,
      "loss": 2.6094627380371094,
      "step": 6976
    },
    {
      "epoch": 0.9601232832418722,
      "grad_norm": 0.09515240043401718,
      "learning_rate": 0.0015059396893085592,
      "loss": 2.6094348430633545,
      "step": 6978
    },
    {
      "epoch": 0.9603984690496229,
      "grad_norm": 0.054570555686950684,
      "learning_rate": 0.0015047212915016753,
      "loss": 2.6167118549346924,
      "step": 6980
    },
    {
      "epoch": 0.9606736548573734,
      "grad_norm": 0.2784203886985779,
      "learning_rate": 0.0015035028936947914,
      "loss": 2.601808786392212,
      "step": 6982
    },
    {
      "epoch": 0.9609488406651241,
      "grad_norm": 0.04968578368425369,
      "learning_rate": 0.0015022844958879073,
      "loss": 2.6363208293914795,
      "step": 6984
    },
    {
      "epoch": 0.9612240264728747,
      "grad_norm": 0.06631121039390564,
      "learning_rate": 0.0015010660980810237,
      "loss": 2.6046624183654785,
      "step": 6986
    },
    {
      "epoch": 0.9614992122806253,
      "grad_norm": 0.19595643877983093,
      "learning_rate": 0.0014998477002741396,
      "loss": 2.641249418258667,
      "step": 6988
    },
    {
      "epoch": 0.961774398088376,
      "grad_norm": 0.03487733379006386,
      "learning_rate": 0.0014986293024672555,
      "loss": 2.6592965126037598,
      "step": 6990
    },
    {
      "epoch": 0.9620495838961265,
      "grad_norm": 0.06796946376562119,
      "learning_rate": 0.0014974109046603716,
      "loss": 2.5704665184020996,
      "step": 6992
    },
    {
      "epoch": 0.9623247697038771,
      "grad_norm": 0.03008492849767208,
      "learning_rate": 0.0014961925068534877,
      "loss": 2.630899429321289,
      "step": 6994
    },
    {
      "epoch": 0.9625999555116277,
      "grad_norm": 0.0866246297955513,
      "learning_rate": 0.0014949741090466038,
      "loss": 2.6207938194274902,
      "step": 6996
    },
    {
      "epoch": 0.9628751413193783,
      "grad_norm": 0.09433000534772873,
      "learning_rate": 0.0014937557112397197,
      "loss": 2.5968174934387207,
      "step": 6998
    },
    {
      "epoch": 0.963150327127129,
      "grad_norm": 0.11157339066267014,
      "learning_rate": 0.0014925373134328358,
      "loss": 2.6359028816223145,
      "step": 7000
    },
    {
      "epoch": 0.9634255129348795,
      "grad_norm": 0.12479127198457718,
      "learning_rate": 0.001491318915625952,
      "loss": 2.6118667125701904,
      "step": 7002
    },
    {
      "epoch": 0.9637006987426302,
      "grad_norm": 0.04592638090252876,
      "learning_rate": 0.0014901005178190679,
      "loss": 2.638784170150757,
      "step": 7004
    },
    {
      "epoch": 0.9639758845503807,
      "grad_norm": 0.055425383150577545,
      "learning_rate": 0.001488882120012184,
      "loss": 2.578211545944214,
      "step": 7006
    },
    {
      "epoch": 0.9642510703581314,
      "grad_norm": 0.04801039770245552,
      "learning_rate": 0.0014876637222053,
      "loss": 2.6195664405822754,
      "step": 7008
    },
    {
      "epoch": 0.9645262561658821,
      "grad_norm": 0.1129191666841507,
      "learning_rate": 0.0014864453243984162,
      "loss": 2.621368408203125,
      "step": 7010
    },
    {
      "epoch": 0.9648014419736326,
      "grad_norm": 0.055026255548000336,
      "learning_rate": 0.001485226926591532,
      "loss": 2.5949203968048096,
      "step": 7012
    },
    {
      "epoch": 0.9650766277813833,
      "grad_norm": 0.14235220849514008,
      "learning_rate": 0.0014840085287846482,
      "loss": 2.6573033332824707,
      "step": 7014
    },
    {
      "epoch": 0.9653518135891338,
      "grad_norm": 0.14840658009052277,
      "learning_rate": 0.0014827901309777643,
      "loss": 2.6353917121887207,
      "step": 7016
    },
    {
      "epoch": 0.9656269993968845,
      "grad_norm": 0.03149121254682541,
      "learning_rate": 0.0014815717331708802,
      "loss": 2.631965160369873,
      "step": 7018
    },
    {
      "epoch": 0.965902185204635,
      "grad_norm": 0.06674329936504364,
      "learning_rate": 0.0014803533353639963,
      "loss": 2.6206064224243164,
      "step": 7020
    },
    {
      "epoch": 0.9661773710123857,
      "grad_norm": 0.12707090377807617,
      "learning_rate": 0.0014791349375571125,
      "loss": 2.5978713035583496,
      "step": 7022
    },
    {
      "epoch": 0.9664525568201363,
      "grad_norm": 0.09269757568836212,
      "learning_rate": 0.0014779165397502286,
      "loss": 2.614438772201538,
      "step": 7024
    },
    {
      "epoch": 0.9667277426278869,
      "grad_norm": 0.19472762942314148,
      "learning_rate": 0.0014766981419433445,
      "loss": 2.605867624282837,
      "step": 7026
    },
    {
      "epoch": 0.9670029284356375,
      "grad_norm": 0.0403275229036808,
      "learning_rate": 0.0014754797441364606,
      "loss": 2.6674585342407227,
      "step": 7028
    },
    {
      "epoch": 0.967278114243388,
      "grad_norm": 0.13726505637168884,
      "learning_rate": 0.0014742613463295767,
      "loss": 2.6289854049682617,
      "step": 7030
    },
    {
      "epoch": 0.9675533000511387,
      "grad_norm": 0.10542922466993332,
      "learning_rate": 0.0014730429485226926,
      "loss": 2.6230740547180176,
      "step": 7032
    },
    {
      "epoch": 0.9678284858588893,
      "grad_norm": 0.1843274086713791,
      "learning_rate": 0.0014718245507158087,
      "loss": 2.651343584060669,
      "step": 7034
    },
    {
      "epoch": 0.9681036716666399,
      "grad_norm": 0.23465408384799957,
      "learning_rate": 0.0014706061529089248,
      "loss": 2.6065526008605957,
      "step": 7036
    },
    {
      "epoch": 0.9683788574743906,
      "grad_norm": 0.13100427389144897,
      "learning_rate": 0.001469387755102041,
      "loss": 2.588197708129883,
      "step": 7038
    },
    {
      "epoch": 0.9686540432821411,
      "grad_norm": 0.04979600012302399,
      "learning_rate": 0.0014681693572951569,
      "loss": 2.613913059234619,
      "step": 7040
    },
    {
      "epoch": 0.9689292290898918,
      "grad_norm": 0.1161687821149826,
      "learning_rate": 0.001466950959488273,
      "loss": 2.60681414604187,
      "step": 7042
    },
    {
      "epoch": 0.9692044148976423,
      "grad_norm": 0.16791462898254395,
      "learning_rate": 0.001465732561681389,
      "loss": 2.6085784435272217,
      "step": 7044
    },
    {
      "epoch": 0.969479600705393,
      "grad_norm": 0.13845928013324738,
      "learning_rate": 0.001464514163874505,
      "loss": 2.6523685455322266,
      "step": 7046
    },
    {
      "epoch": 0.9697547865131435,
      "grad_norm": 0.05048372969031334,
      "learning_rate": 0.001463295766067621,
      "loss": 2.6199750900268555,
      "step": 7048
    },
    {
      "epoch": 0.9700299723208942,
      "grad_norm": 0.14254651963710785,
      "learning_rate": 0.0014620773682607372,
      "loss": 2.612870931625366,
      "step": 7050
    },
    {
      "epoch": 0.9703051581286448,
      "grad_norm": 0.04047882556915283,
      "learning_rate": 0.0014608589704538533,
      "loss": 2.5669991970062256,
      "step": 7052
    },
    {
      "epoch": 0.9705803439363954,
      "grad_norm": 0.03226237744092941,
      "learning_rate": 0.0014596405726469692,
      "loss": 2.6067066192626953,
      "step": 7054
    },
    {
      "epoch": 0.970855529744146,
      "grad_norm": 0.036921482533216476,
      "learning_rate": 0.0014584221748400851,
      "loss": 2.630040407180786,
      "step": 7056
    },
    {
      "epoch": 0.9711307155518966,
      "grad_norm": 0.03097798489034176,
      "learning_rate": 0.0014572037770332015,
      "loss": 2.629802703857422,
      "step": 7058
    },
    {
      "epoch": 0.9714059013596472,
      "grad_norm": 0.09362146258354187,
      "learning_rate": 0.0014559853792263174,
      "loss": 2.5965027809143066,
      "step": 7060
    },
    {
      "epoch": 0.9716810871673979,
      "grad_norm": 0.0999496653676033,
      "learning_rate": 0.0014547669814194335,
      "loss": 2.6085968017578125,
      "step": 7062
    },
    {
      "epoch": 0.9719562729751484,
      "grad_norm": 0.048956066370010376,
      "learning_rate": 0.0014535485836125494,
      "loss": 2.5857131481170654,
      "step": 7064
    },
    {
      "epoch": 0.9722314587828991,
      "grad_norm": 0.10262764990329742,
      "learning_rate": 0.0014523301858056657,
      "loss": 2.5985498428344727,
      "step": 7066
    },
    {
      "epoch": 0.9725066445906496,
      "grad_norm": 0.10495644807815552,
      "learning_rate": 0.0014511117879987816,
      "loss": 2.5862650871276855,
      "step": 7068
    },
    {
      "epoch": 0.9727818303984003,
      "grad_norm": 0.027166450396180153,
      "learning_rate": 0.0014498933901918975,
      "loss": 2.631927967071533,
      "step": 7070
    },
    {
      "epoch": 0.9730570162061508,
      "grad_norm": 0.03090870939195156,
      "learning_rate": 0.0014486749923850138,
      "loss": 2.6505885124206543,
      "step": 7072
    },
    {
      "epoch": 0.9733322020139015,
      "grad_norm": 0.13827557861804962,
      "learning_rate": 0.0014474565945781297,
      "loss": 2.6287615299224854,
      "step": 7074
    },
    {
      "epoch": 0.9736073878216521,
      "grad_norm": 0.09324035793542862,
      "learning_rate": 0.0014462381967712459,
      "loss": 2.613952398300171,
      "step": 7076
    },
    {
      "epoch": 0.9738825736294027,
      "grad_norm": 0.18332228064537048,
      "learning_rate": 0.0014450197989643618,
      "loss": 2.598008632659912,
      "step": 7078
    },
    {
      "epoch": 0.9741577594371533,
      "grad_norm": 0.08416702598333359,
      "learning_rate": 0.001443801401157478,
      "loss": 2.60019588470459,
      "step": 7080
    },
    {
      "epoch": 0.9744329452449039,
      "grad_norm": 0.11122936755418777,
      "learning_rate": 0.001442583003350594,
      "loss": 2.6324896812438965,
      "step": 7082
    },
    {
      "epoch": 0.9747081310526545,
      "grad_norm": 0.07525543868541718,
      "learning_rate": 0.00144136460554371,
      "loss": 2.5988688468933105,
      "step": 7084
    },
    {
      "epoch": 0.9749833168604051,
      "grad_norm": 0.07738874852657318,
      "learning_rate": 0.0014401462077368262,
      "loss": 2.607788562774658,
      "step": 7086
    },
    {
      "epoch": 0.9752585026681557,
      "grad_norm": 0.04645910859107971,
      "learning_rate": 0.0014389278099299421,
      "loss": 2.6401193141937256,
      "step": 7088
    },
    {
      "epoch": 0.9755336884759064,
      "grad_norm": 0.12853878736495972,
      "learning_rate": 0.0014377094121230582,
      "loss": 2.609285831451416,
      "step": 7090
    },
    {
      "epoch": 0.9758088742836569,
      "grad_norm": 0.09326208382844925,
      "learning_rate": 0.0014364910143161741,
      "loss": 2.622150421142578,
      "step": 7092
    },
    {
      "epoch": 0.9760840600914076,
      "grad_norm": 0.1150539293885231,
      "learning_rate": 0.0014352726165092905,
      "loss": 2.602013111114502,
      "step": 7094
    },
    {
      "epoch": 0.9763592458991581,
      "grad_norm": 0.07641231268644333,
      "learning_rate": 0.0014340542187024064,
      "loss": 2.6002540588378906,
      "step": 7096
    },
    {
      "epoch": 0.9766344317069088,
      "grad_norm": 0.06720434874296188,
      "learning_rate": 0.0014328358208955223,
      "loss": 2.5756149291992188,
      "step": 7098
    },
    {
      "epoch": 0.9769096175146593,
      "grad_norm": 0.08952721208333969,
      "learning_rate": 0.0014316174230886386,
      "loss": 2.6484010219573975,
      "step": 7100
    },
    {
      "epoch": 0.97718480332241,
      "grad_norm": 0.11324629932641983,
      "learning_rate": 0.0014303990252817545,
      "loss": 2.5725150108337402,
      "step": 7102
    },
    {
      "epoch": 0.9774599891301606,
      "grad_norm": 0.02888636663556099,
      "learning_rate": 0.0014291806274748706,
      "loss": 2.6388869285583496,
      "step": 7104
    },
    {
      "epoch": 0.9777351749379112,
      "grad_norm": 0.07674583792686462,
      "learning_rate": 0.0014279622296679865,
      "loss": 2.594529390335083,
      "step": 7106
    },
    {
      "epoch": 0.9780103607456618,
      "grad_norm": 0.11461985111236572,
      "learning_rate": 0.0014267438318611026,
      "loss": 2.5902867317199707,
      "step": 7108
    },
    {
      "epoch": 0.9782855465534124,
      "grad_norm": 0.053811632096767426,
      "learning_rate": 0.0014255254340542188,
      "loss": 2.5817813873291016,
      "step": 7110
    },
    {
      "epoch": 0.978560732361163,
      "grad_norm": 0.16511663794517517,
      "learning_rate": 0.0014243070362473347,
      "loss": 2.627225875854492,
      "step": 7112
    },
    {
      "epoch": 0.9788359181689137,
      "grad_norm": 0.030564231798052788,
      "learning_rate": 0.001423088638440451,
      "loss": 2.598721504211426,
      "step": 7114
    },
    {
      "epoch": 0.9791111039766642,
      "grad_norm": 0.09532318264245987,
      "learning_rate": 0.0014218702406335669,
      "loss": 2.6029863357543945,
      "step": 7116
    },
    {
      "epoch": 0.9793862897844149,
      "grad_norm": 0.03263423219323158,
      "learning_rate": 0.001420651842826683,
      "loss": 2.638331890106201,
      "step": 7118
    },
    {
      "epoch": 0.9796614755921654,
      "grad_norm": 0.09949354082345963,
      "learning_rate": 0.001419433445019799,
      "loss": 2.634852409362793,
      "step": 7120
    },
    {
      "epoch": 0.9799366613999161,
      "grad_norm": 0.0854094848036766,
      "learning_rate": 0.001418215047212915,
      "loss": 2.5831212997436523,
      "step": 7122
    },
    {
      "epoch": 0.9802118472076666,
      "grad_norm": 0.06251838803291321,
      "learning_rate": 0.0014169966494060311,
      "loss": 2.6101603507995605,
      "step": 7124
    },
    {
      "epoch": 0.9804870330154173,
      "grad_norm": 0.0483749620616436,
      "learning_rate": 0.001415778251599147,
      "loss": 2.6305623054504395,
      "step": 7126
    },
    {
      "epoch": 0.9807622188231679,
      "grad_norm": 0.049629852175712585,
      "learning_rate": 0.0014145598537922634,
      "loss": 2.5976314544677734,
      "step": 7128
    },
    {
      "epoch": 0.9810374046309185,
      "grad_norm": 0.21855410933494568,
      "learning_rate": 0.0014133414559853793,
      "loss": 2.6375479698181152,
      "step": 7130
    },
    {
      "epoch": 0.9813125904386691,
      "grad_norm": 0.028211627155542374,
      "learning_rate": 0.0014121230581784954,
      "loss": 2.6289947032928467,
      "step": 7132
    },
    {
      "epoch": 0.9815877762464197,
      "grad_norm": 0.13348977267742157,
      "learning_rate": 0.0014109046603716113,
      "loss": 2.58951473236084,
      "step": 7134
    },
    {
      "epoch": 0.9818629620541703,
      "grad_norm": 0.24338848888874054,
      "learning_rate": 0.0014096862625647274,
      "loss": 2.5978903770446777,
      "step": 7136
    },
    {
      "epoch": 0.9821381478619209,
      "grad_norm": 0.0360863097012043,
      "learning_rate": 0.0014084678647578435,
      "loss": 2.5872037410736084,
      "step": 7138
    },
    {
      "epoch": 0.9824133336696715,
      "grad_norm": 0.14417293667793274,
      "learning_rate": 0.0014072494669509594,
      "loss": 2.616417407989502,
      "step": 7140
    },
    {
      "epoch": 0.9826885194774222,
      "grad_norm": 0.16633212566375732,
      "learning_rate": 0.0014060310691440757,
      "loss": 2.613107204437256,
      "step": 7142
    },
    {
      "epoch": 0.9829637052851727,
      "grad_norm": 0.093917615711689,
      "learning_rate": 0.0014048126713371916,
      "loss": 2.593668222427368,
      "step": 7144
    },
    {
      "epoch": 0.9832388910929234,
      "grad_norm": 0.08667364716529846,
      "learning_rate": 0.0014035942735303078,
      "loss": 2.6506075859069824,
      "step": 7146
    },
    {
      "epoch": 0.9835140769006739,
      "grad_norm": 0.12633101642131805,
      "learning_rate": 0.0014023758757234237,
      "loss": 2.5944557189941406,
      "step": 7148
    },
    {
      "epoch": 0.9837892627084246,
      "grad_norm": 0.08554825186729431,
      "learning_rate": 0.0014011574779165398,
      "loss": 2.6587371826171875,
      "step": 7150
    },
    {
      "epoch": 0.9840644485161751,
      "grad_norm": 0.03563966974616051,
      "learning_rate": 0.0013999390801096559,
      "loss": 2.605513572692871,
      "step": 7152
    },
    {
      "epoch": 0.9843396343239258,
      "grad_norm": 0.05749896913766861,
      "learning_rate": 0.0013987206823027718,
      "loss": 2.624922752380371,
      "step": 7154
    },
    {
      "epoch": 0.9846148201316764,
      "grad_norm": 0.03532284125685692,
      "learning_rate": 0.001397502284495888,
      "loss": 2.6154422760009766,
      "step": 7156
    },
    {
      "epoch": 0.984890005939427,
      "grad_norm": 0.04042253643274307,
      "learning_rate": 0.001396283886689004,
      "loss": 2.601255416870117,
      "step": 7158
    },
    {
      "epoch": 0.9851651917471776,
      "grad_norm": 0.03725714981555939,
      "learning_rate": 0.0013950654888821201,
      "loss": 2.522522211074829,
      "step": 7160
    },
    {
      "epoch": 0.9854403775549282,
      "grad_norm": 0.1677236109972,
      "learning_rate": 0.001393847091075236,
      "loss": 2.6120524406433105,
      "step": 7162
    },
    {
      "epoch": 0.9857155633626788,
      "grad_norm": 0.0656951367855072,
      "learning_rate": 0.0013926286932683522,
      "loss": 2.5779426097869873,
      "step": 7164
    },
    {
      "epoch": 0.9859907491704295,
      "grad_norm": 0.10681840032339096,
      "learning_rate": 0.0013914102954614683,
      "loss": 2.6031033992767334,
      "step": 7166
    },
    {
      "epoch": 0.98626593497818,
      "grad_norm": 0.03773680329322815,
      "learning_rate": 0.0013901918976545842,
      "loss": 2.556699752807617,
      "step": 7168
    },
    {
      "epoch": 0.9865411207859307,
      "grad_norm": 0.035285234451293945,
      "learning_rate": 0.0013889734998477003,
      "loss": 2.612478733062744,
      "step": 7170
    },
    {
      "epoch": 0.9868163065936812,
      "grad_norm": 0.17708763480186462,
      "learning_rate": 0.0013877551020408164,
      "loss": 2.595939874649048,
      "step": 7172
    },
    {
      "epoch": 0.9870914924014319,
      "grad_norm": 0.05071648582816124,
      "learning_rate": 0.0013865367042339323,
      "loss": 2.5626845359802246,
      "step": 7174
    },
    {
      "epoch": 0.9873666782091824,
      "grad_norm": 0.17783378064632416,
      "learning_rate": 0.0013853183064270484,
      "loss": 2.5938262939453125,
      "step": 7176
    },
    {
      "epoch": 0.9876418640169331,
      "grad_norm": 0.05050039291381836,
      "learning_rate": 0.0013840999086201645,
      "loss": 2.5875253677368164,
      "step": 7178
    },
    {
      "epoch": 0.9879170498246838,
      "grad_norm": 0.15902365744113922,
      "learning_rate": 0.0013828815108132806,
      "loss": 2.595764636993408,
      "step": 7180
    },
    {
      "epoch": 0.9881922356324343,
      "grad_norm": 0.1719236522912979,
      "learning_rate": 0.0013816631130063965,
      "loss": 2.6196889877319336,
      "step": 7182
    },
    {
      "epoch": 0.988467421440185,
      "grad_norm": 0.0939425677061081,
      "learning_rate": 0.0013804447151995127,
      "loss": 2.6313581466674805,
      "step": 7184
    },
    {
      "epoch": 0.9887426072479355,
      "grad_norm": 0.06870869547128677,
      "learning_rate": 0.0013792263173926288,
      "loss": 2.638449192047119,
      "step": 7186
    },
    {
      "epoch": 0.9890177930556862,
      "grad_norm": 0.07364911586046219,
      "learning_rate": 0.0013780079195857447,
      "loss": 2.6070408821105957,
      "step": 7188
    },
    {
      "epoch": 0.9892929788634367,
      "grad_norm": 0.10110954195261002,
      "learning_rate": 0.0013767895217788608,
      "loss": 2.606717586517334,
      "step": 7190
    },
    {
      "epoch": 0.9895681646711874,
      "grad_norm": 0.061566855758428574,
      "learning_rate": 0.001375571123971977,
      "loss": 2.602482795715332,
      "step": 7192
    },
    {
      "epoch": 0.989843350478938,
      "grad_norm": 0.09795261919498444,
      "learning_rate": 0.001374352726165093,
      "loss": 2.5812501907348633,
      "step": 7194
    },
    {
      "epoch": 0.9901185362866886,
      "grad_norm": 0.1440325379371643,
      "learning_rate": 0.001373134328358209,
      "loss": 2.635242462158203,
      "step": 7196
    },
    {
      "epoch": 0.9903937220944392,
      "grad_norm": 0.03979695215821266,
      "learning_rate": 0.001371915930551325,
      "loss": 2.5826644897460938,
      "step": 7198
    },
    {
      "epoch": 0.9906689079021898,
      "grad_norm": 0.06836917996406555,
      "learning_rate": 0.0013706975327444412,
      "loss": 2.6327290534973145,
      "step": 7200
    },
    {
      "epoch": 0.9909440937099404,
      "grad_norm": 0.12362658977508545,
      "learning_rate": 0.001369479134937557,
      "loss": 2.6601767539978027,
      "step": 7202
    },
    {
      "epoch": 0.991219279517691,
      "grad_norm": 0.050241049379110336,
      "learning_rate": 0.0013682607371306732,
      "loss": 2.614861011505127,
      "step": 7204
    },
    {
      "epoch": 0.9914944653254416,
      "grad_norm": 0.21664193272590637,
      "learning_rate": 0.0013670423393237893,
      "loss": 2.5992166996002197,
      "step": 7206
    },
    {
      "epoch": 0.9917696511331923,
      "grad_norm": 0.04214854910969734,
      "learning_rate": 0.0013658239415169054,
      "loss": 2.6156716346740723,
      "step": 7208
    },
    {
      "epoch": 0.9920448369409428,
      "grad_norm": 0.11806903779506683,
      "learning_rate": 0.0013646055437100213,
      "loss": 2.624514579772949,
      "step": 7210
    },
    {
      "epoch": 0.9923200227486935,
      "grad_norm": 0.03990588337182999,
      "learning_rate": 0.0013633871459031374,
      "loss": 2.623495101928711,
      "step": 7212
    },
    {
      "epoch": 0.992595208556444,
      "grad_norm": 0.05693225935101509,
      "learning_rate": 0.0013621687480962535,
      "loss": 2.6017167568206787,
      "step": 7214
    },
    {
      "epoch": 0.9928703943641947,
      "grad_norm": 0.027448557317256927,
      "learning_rate": 0.0013609503502893694,
      "loss": 2.6129040718078613,
      "step": 7216
    },
    {
      "epoch": 0.9931455801719453,
      "grad_norm": 0.10675537586212158,
      "learning_rate": 0.0013597319524824856,
      "loss": 2.5859456062316895,
      "step": 7218
    },
    {
      "epoch": 0.9934207659796959,
      "grad_norm": 0.06502936035394669,
      "learning_rate": 0.0013585135546756015,
      "loss": 2.6127963066101074,
      "step": 7220
    },
    {
      "epoch": 0.9936959517874465,
      "grad_norm": 0.1681014597415924,
      "learning_rate": 0.0013572951568687178,
      "loss": 2.607268810272217,
      "step": 7222
    },
    {
      "epoch": 0.9939711375951971,
      "grad_norm": 0.02842380478978157,
      "learning_rate": 0.0013560767590618337,
      "loss": 2.6027050018310547,
      "step": 7224
    },
    {
      "epoch": 0.9942463234029477,
      "grad_norm": 0.09612075239419937,
      "learning_rate": 0.0013548583612549498,
      "loss": 2.599005699157715,
      "step": 7226
    },
    {
      "epoch": 0.9945215092106983,
      "grad_norm": 0.09715006500482559,
      "learning_rate": 0.001353639963448066,
      "loss": 2.615445613861084,
      "step": 7228
    },
    {
      "epoch": 0.9947966950184489,
      "grad_norm": 0.06270590424537659,
      "learning_rate": 0.0013524215656411818,
      "loss": 2.602156162261963,
      "step": 7230
    },
    {
      "epoch": 0.9950718808261996,
      "grad_norm": 0.03509427607059479,
      "learning_rate": 0.001351203167834298,
      "loss": 2.5973849296569824,
      "step": 7232
    },
    {
      "epoch": 0.9953470666339501,
      "grad_norm": 0.04345272108912468,
      "learning_rate": 0.0013499847700274138,
      "loss": 2.5679993629455566,
      "step": 7234
    },
    {
      "epoch": 0.9956222524417008,
      "grad_norm": 0.09352730959653854,
      "learning_rate": 0.0013487663722205302,
      "loss": 2.587869644165039,
      "step": 7236
    },
    {
      "epoch": 0.9958974382494513,
      "grad_norm": 0.04277810454368591,
      "learning_rate": 0.001347547974413646,
      "loss": 2.607363700866699,
      "step": 7238
    },
    {
      "epoch": 0.996172624057202,
      "grad_norm": 0.15138466656208038,
      "learning_rate": 0.001346329576606762,
      "loss": 2.615346908569336,
      "step": 7240
    },
    {
      "epoch": 0.9964478098649525,
      "grad_norm": 0.11059655994176865,
      "learning_rate": 0.0013451111787998783,
      "loss": 2.5744292736053467,
      "step": 7242
    },
    {
      "epoch": 0.9967229956727032,
      "grad_norm": 0.0746651217341423,
      "learning_rate": 0.0013438927809929942,
      "loss": 2.563072919845581,
      "step": 7244
    },
    {
      "epoch": 0.9969981814804538,
      "grad_norm": 0.06269986927509308,
      "learning_rate": 0.0013426743831861103,
      "loss": 2.596311569213867,
      "step": 7246
    },
    {
      "epoch": 0.9972733672882044,
      "grad_norm": 0.07413709163665771,
      "learning_rate": 0.0013414559853792262,
      "loss": 2.611720561981201,
      "step": 7248
    },
    {
      "epoch": 0.997548553095955,
      "grad_norm": 0.12419672310352325,
      "learning_rate": 0.0013402375875723425,
      "loss": 2.5722453594207764,
      "step": 7250
    },
    {
      "epoch": 0.9978237389037056,
      "grad_norm": 0.034327346831560135,
      "learning_rate": 0.0013390191897654584,
      "loss": 2.608720302581787,
      "step": 7252
    },
    {
      "epoch": 0.9980989247114562,
      "grad_norm": 0.15102960169315338,
      "learning_rate": 0.0013378007919585743,
      "loss": 2.601410388946533,
      "step": 7254
    },
    {
      "epoch": 0.9983741105192068,
      "grad_norm": 0.03604521602392197,
      "learning_rate": 0.0013365823941516907,
      "loss": 2.6087732315063477,
      "step": 7256
    },
    {
      "epoch": 0.9986492963269574,
      "grad_norm": 0.14073209464550018,
      "learning_rate": 0.0013353639963448066,
      "loss": 2.60091495513916,
      "step": 7258
    },
    {
      "epoch": 0.9989244821347081,
      "grad_norm": 0.037743423134088516,
      "learning_rate": 0.0013341455985379227,
      "loss": 2.5925745964050293,
      "step": 7260
    },
    {
      "epoch": 0.9991996679424586,
      "grad_norm": 0.029798906296491623,
      "learning_rate": 0.0013329272007310386,
      "loss": 2.578889846801758,
      "step": 7262
    },
    {
      "epoch": 0.9994748537502093,
      "grad_norm": 0.10881885886192322,
      "learning_rate": 0.001331708802924155,
      "loss": 2.579087734222412,
      "step": 7264
    },
    {
      "epoch": 0.9997500395579598,
      "grad_norm": 0.13760168850421906,
      "learning_rate": 0.0013304904051172708,
      "loss": 2.5410289764404297,
      "step": 7266
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.10063134878873825,
      "learning_rate": 0.0013292720073103867,
      "loss": 2.5668082237243652,
      "step": 7268
    },
    {
      "epoch": 1.0002751858077505,
      "grad_norm": 0.04625566303730011,
      "learning_rate": 0.001328053609503503,
      "loss": 2.552952527999878,
      "step": 7270
    },
    {
      "epoch": 1.0005503716155013,
      "grad_norm": 0.04463019222021103,
      "learning_rate": 0.001326835211696619,
      "loss": 2.533996105194092,
      "step": 7272
    },
    {
      "epoch": 1.0008255574232519,
      "grad_norm": 0.03408603370189667,
      "learning_rate": 0.001325616813889735,
      "loss": 2.5665760040283203,
      "step": 7274
    },
    {
      "epoch": 1.0011007432310024,
      "grad_norm": 0.038304463028907776,
      "learning_rate": 0.001324398416082851,
      "loss": 2.5427024364471436,
      "step": 7276
    },
    {
      "epoch": 1.001375929038753,
      "grad_norm": 0.04716838151216507,
      "learning_rate": 0.0013231800182759673,
      "loss": 2.5005345344543457,
      "step": 7278
    },
    {
      "epoch": 1.0016511148465037,
      "grad_norm": 0.06163451075553894,
      "learning_rate": 0.0013219616204690832,
      "loss": 2.5458438396453857,
      "step": 7280
    },
    {
      "epoch": 1.0019263006542543,
      "grad_norm": 0.06221252679824829,
      "learning_rate": 0.001320743222662199,
      "loss": 2.511622428894043,
      "step": 7282
    },
    {
      "epoch": 1.0022014864620048,
      "grad_norm": 0.23386874794960022,
      "learning_rate": 0.0013195248248553154,
      "loss": 2.5752475261688232,
      "step": 7284
    },
    {
      "epoch": 1.0024766722697556,
      "grad_norm": 0.12307140231132507,
      "learning_rate": 0.0013183064270484313,
      "loss": 2.525418281555176,
      "step": 7286
    },
    {
      "epoch": 1.002751858077506,
      "grad_norm": 0.03130524605512619,
      "learning_rate": 0.0013170880292415474,
      "loss": 2.532424211502075,
      "step": 7288
    },
    {
      "epoch": 1.0030270438852567,
      "grad_norm": 0.1556885540485382,
      "learning_rate": 0.0013158696314346633,
      "loss": 2.520106792449951,
      "step": 7290
    },
    {
      "epoch": 1.0033022296930072,
      "grad_norm": 0.1946473866701126,
      "learning_rate": 0.0013146512336277797,
      "loss": 2.527153491973877,
      "step": 7292
    },
    {
      "epoch": 1.003577415500758,
      "grad_norm": 0.13888736069202423,
      "learning_rate": 0.0013134328358208956,
      "loss": 2.539949417114258,
      "step": 7294
    },
    {
      "epoch": 1.0038526013085085,
      "grad_norm": 0.03648655116558075,
      "learning_rate": 0.0013122144380140115,
      "loss": 2.5360639095306396,
      "step": 7296
    },
    {
      "epoch": 1.004127787116259,
      "grad_norm": 0.0800069198012352,
      "learning_rate": 0.0013109960402071278,
      "loss": 2.550882339477539,
      "step": 7298
    },
    {
      "epoch": 1.0044029729240098,
      "grad_norm": 0.10367537289857864,
      "learning_rate": 0.0013097776424002437,
      "loss": 2.5033721923828125,
      "step": 7300
    },
    {
      "epoch": 1.0046781587317604,
      "grad_norm": 0.12649576365947723,
      "learning_rate": 0.0013085592445933598,
      "loss": 2.524786949157715,
      "step": 7302
    },
    {
      "epoch": 1.004953344539511,
      "grad_norm": 0.03658008202910423,
      "learning_rate": 0.0013073408467864757,
      "loss": 2.5596885681152344,
      "step": 7304
    },
    {
      "epoch": 1.0052285303472615,
      "grad_norm": 0.04926189035177231,
      "learning_rate": 0.0013061224489795918,
      "loss": 2.541348457336426,
      "step": 7306
    },
    {
      "epoch": 1.0055037161550122,
      "grad_norm": 0.1131928563117981,
      "learning_rate": 0.001304904051172708,
      "loss": 2.5194668769836426,
      "step": 7308
    },
    {
      "epoch": 1.0057789019627628,
      "grad_norm": 0.0570996031165123,
      "learning_rate": 0.0013036856533658239,
      "loss": 2.5254480838775635,
      "step": 7310
    },
    {
      "epoch": 1.0060540877705133,
      "grad_norm": 0.047130245715379715,
      "learning_rate": 0.00130246725555894,
      "loss": 2.5181126594543457,
      "step": 7312
    },
    {
      "epoch": 1.006329273578264,
      "grad_norm": 0.026856228709220886,
      "learning_rate": 0.001301248857752056,
      "loss": 2.5486838817596436,
      "step": 7314
    },
    {
      "epoch": 1.0066044593860146,
      "grad_norm": 0.05324045941233635,
      "learning_rate": 0.0013000304599451722,
      "loss": 2.5459134578704834,
      "step": 7316
    },
    {
      "epoch": 1.0068796451937652,
      "grad_norm": 0.035978756844997406,
      "learning_rate": 0.001298812062138288,
      "loss": 2.559265613555908,
      "step": 7318
    },
    {
      "epoch": 1.0071548310015157,
      "grad_norm": 0.0584685392677784,
      "learning_rate": 0.0012975936643314042,
      "loss": 2.5697739124298096,
      "step": 7320
    },
    {
      "epoch": 1.0074300168092665,
      "grad_norm": 0.14377565681934357,
      "learning_rate": 0.0012963752665245203,
      "loss": 2.514680862426758,
      "step": 7322
    },
    {
      "epoch": 1.007705202617017,
      "grad_norm": 0.035338521003723145,
      "learning_rate": 0.0012951568687176362,
      "loss": 2.5182104110717773,
      "step": 7324
    },
    {
      "epoch": 1.0079803884247676,
      "grad_norm": 0.03420895338058472,
      "learning_rate": 0.0012939384709107524,
      "loss": 2.554428815841675,
      "step": 7326
    },
    {
      "epoch": 1.0082555742325183,
      "grad_norm": 0.11316990852355957,
      "learning_rate": 0.0012927200731038685,
      "loss": 2.552398920059204,
      "step": 7328
    },
    {
      "epoch": 1.0085307600402689,
      "grad_norm": 0.08224035799503326,
      "learning_rate": 0.0012915016752969846,
      "loss": 2.526082992553711,
      "step": 7330
    },
    {
      "epoch": 1.0088059458480194,
      "grad_norm": 0.027018990367650986,
      "learning_rate": 0.0012902832774901005,
      "loss": 2.525087356567383,
      "step": 7332
    },
    {
      "epoch": 1.0090811316557702,
      "grad_norm": 0.06339249759912491,
      "learning_rate": 0.0012890648796832166,
      "loss": 2.5534377098083496,
      "step": 7334
    },
    {
      "epoch": 1.0093563174635207,
      "grad_norm": 0.07821891456842422,
      "learning_rate": 0.0012878464818763327,
      "loss": 2.5104455947875977,
      "step": 7336
    },
    {
      "epoch": 1.0096315032712713,
      "grad_norm": 0.04677979275584221,
      "learning_rate": 0.0012866280840694486,
      "loss": 2.511169195175171,
      "step": 7338
    },
    {
      "epoch": 1.0099066890790218,
      "grad_norm": 0.06959310173988342,
      "learning_rate": 0.0012854096862625647,
      "loss": 2.5334649085998535,
      "step": 7340
    },
    {
      "epoch": 1.0101818748867726,
      "grad_norm": 0.2235443890094757,
      "learning_rate": 0.0012841912884556808,
      "loss": 2.5327024459838867,
      "step": 7342
    },
    {
      "epoch": 1.0104570606945231,
      "grad_norm": 0.0632055252790451,
      "learning_rate": 0.001282972890648797,
      "loss": 2.5341806411743164,
      "step": 7344
    },
    {
      "epoch": 1.0107322465022737,
      "grad_norm": 0.16633357107639313,
      "learning_rate": 0.0012817544928419129,
      "loss": 2.563237190246582,
      "step": 7346
    },
    {
      "epoch": 1.0110074323100244,
      "grad_norm": 0.08545129746198654,
      "learning_rate": 0.001280536095035029,
      "loss": 2.525334596633911,
      "step": 7348
    },
    {
      "epoch": 1.011282618117775,
      "grad_norm": 0.09141939878463745,
      "learning_rate": 0.001279317697228145,
      "loss": 2.549257278442383,
      "step": 7350
    },
    {
      "epoch": 1.0115578039255255,
      "grad_norm": 0.03213021531701088,
      "learning_rate": 0.001278099299421261,
      "loss": 2.545215606689453,
      "step": 7352
    },
    {
      "epoch": 1.011832989733276,
      "grad_norm": 0.0548652783036232,
      "learning_rate": 0.0012768809016143771,
      "loss": 2.5519280433654785,
      "step": 7354
    },
    {
      "epoch": 1.0121081755410268,
      "grad_norm": 0.11324753612279892,
      "learning_rate": 0.0012756625038074932,
      "loss": 2.51102352142334,
      "step": 7356
    },
    {
      "epoch": 1.0123833613487774,
      "grad_norm": 0.05383477360010147,
      "learning_rate": 0.0012744441060006091,
      "loss": 2.5310020446777344,
      "step": 7358
    },
    {
      "epoch": 1.012658547156528,
      "grad_norm": 0.11715984344482422,
      "learning_rate": 0.0012732257081937252,
      "loss": 2.496591091156006,
      "step": 7360
    },
    {
      "epoch": 1.0129337329642787,
      "grad_norm": 0.028134232386946678,
      "learning_rate": 0.0012720073103868414,
      "loss": 2.548525810241699,
      "step": 7362
    },
    {
      "epoch": 1.0132089187720292,
      "grad_norm": 0.11946558207273483,
      "learning_rate": 0.0012707889125799575,
      "loss": 2.4989757537841797,
      "step": 7364
    },
    {
      "epoch": 1.0134841045797798,
      "grad_norm": 0.027796030044555664,
      "learning_rate": 0.0012695705147730734,
      "loss": 2.551696300506592,
      "step": 7366
    },
    {
      "epoch": 1.0137592903875303,
      "grad_norm": 0.03427908569574356,
      "learning_rate": 0.0012683521169661895,
      "loss": 2.5063328742980957,
      "step": 7368
    },
    {
      "epoch": 1.014034476195281,
      "grad_norm": 0.058834053575992584,
      "learning_rate": 0.0012671337191593056,
      "loss": 2.588261842727661,
      "step": 7370
    },
    {
      "epoch": 1.0143096620030316,
      "grad_norm": 0.07158815115690231,
      "learning_rate": 0.0012659153213524215,
      "loss": 2.507800579071045,
      "step": 7372
    },
    {
      "epoch": 1.0145848478107822,
      "grad_norm": 0.03288167715072632,
      "learning_rate": 0.0012646969235455376,
      "loss": 2.535499095916748,
      "step": 7374
    },
    {
      "epoch": 1.014860033618533,
      "grad_norm": 0.047482363879680634,
      "learning_rate": 0.0012634785257386535,
      "loss": 2.5156099796295166,
      "step": 7376
    },
    {
      "epoch": 1.0151352194262835,
      "grad_norm": 0.06012905761599541,
      "learning_rate": 0.0012622601279317699,
      "loss": 2.5407848358154297,
      "step": 7378
    },
    {
      "epoch": 1.015410405234034,
      "grad_norm": 0.04497440159320831,
      "learning_rate": 0.0012610417301248858,
      "loss": 2.5145702362060547,
      "step": 7380
    },
    {
      "epoch": 1.0156855910417846,
      "grad_norm": 0.14027981460094452,
      "learning_rate": 0.0012598233323180019,
      "loss": 2.5546727180480957,
      "step": 7382
    },
    {
      "epoch": 1.0159607768495353,
      "grad_norm": 0.07937947660684586,
      "learning_rate": 0.001258604934511118,
      "loss": 2.515777111053467,
      "step": 7384
    },
    {
      "epoch": 1.0162359626572859,
      "grad_norm": 0.09138157963752747,
      "learning_rate": 0.0012573865367042339,
      "loss": 2.5385406017303467,
      "step": 7386
    },
    {
      "epoch": 1.0165111484650364,
      "grad_norm": 0.030231133103370667,
      "learning_rate": 0.00125616813889735,
      "loss": 2.5355143547058105,
      "step": 7388
    },
    {
      "epoch": 1.0167863342727872,
      "grad_norm": 0.18135114014148712,
      "learning_rate": 0.001254949741090466,
      "loss": 2.522650718688965,
      "step": 7390
    },
    {
      "epoch": 1.0170615200805377,
      "grad_norm": 0.09808991104364395,
      "learning_rate": 0.0012537313432835822,
      "loss": 2.5420565605163574,
      "step": 7392
    },
    {
      "epoch": 1.0173367058882883,
      "grad_norm": 0.05837186053395271,
      "learning_rate": 0.0012525129454766981,
      "loss": 2.529428243637085,
      "step": 7394
    },
    {
      "epoch": 1.0176118916960388,
      "grad_norm": 0.06904146075248718,
      "learning_rate": 0.0012512945476698142,
      "loss": 2.5234546661376953,
      "step": 7396
    },
    {
      "epoch": 1.0178870775037896,
      "grad_norm": 0.13694161176681519,
      "learning_rate": 0.0012500761498629304,
      "loss": 2.5407462120056152,
      "step": 7398
    },
    {
      "epoch": 1.0181622633115401,
      "grad_norm": 0.03177842125296593,
      "learning_rate": 0.0012488577520560463,
      "loss": 2.5401251316070557,
      "step": 7400
    },
    {
      "epoch": 1.0184374491192907,
      "grad_norm": 0.039721325039863586,
      "learning_rate": 0.0012476393542491624,
      "loss": 2.4926910400390625,
      "step": 7402
    },
    {
      "epoch": 1.0187126349270414,
      "grad_norm": 0.05168136581778526,
      "learning_rate": 0.0012464209564422783,
      "loss": 2.538036823272705,
      "step": 7404
    },
    {
      "epoch": 1.018987820734792,
      "grad_norm": 0.03072730265557766,
      "learning_rate": 0.0012452025586353946,
      "loss": 2.522143840789795,
      "step": 7406
    },
    {
      "epoch": 1.0192630065425425,
      "grad_norm": 0.13840961456298828,
      "learning_rate": 0.0012439841608285105,
      "loss": 2.531928300857544,
      "step": 7408
    },
    {
      "epoch": 1.019538192350293,
      "grad_norm": 0.07334087044000626,
      "learning_rate": 0.0012427657630216266,
      "loss": 2.557924270629883,
      "step": 7410
    },
    {
      "epoch": 1.0198133781580438,
      "grad_norm": 0.2250737100839615,
      "learning_rate": 0.0012415473652147427,
      "loss": 2.530808925628662,
      "step": 7412
    },
    {
      "epoch": 1.0200885639657944,
      "grad_norm": 0.20991329848766327,
      "learning_rate": 0.0012403289674078586,
      "loss": 2.522402763366699,
      "step": 7414
    },
    {
      "epoch": 1.020363749773545,
      "grad_norm": 0.09630527347326279,
      "learning_rate": 0.0012391105696009748,
      "loss": 2.511505126953125,
      "step": 7416
    },
    {
      "epoch": 1.0206389355812957,
      "grad_norm": 0.02937893755733967,
      "learning_rate": 0.0012378921717940907,
      "loss": 2.575373649597168,
      "step": 7418
    },
    {
      "epoch": 1.0209141213890462,
      "grad_norm": 0.0663914829492569,
      "learning_rate": 0.001236673773987207,
      "loss": 2.5159308910369873,
      "step": 7420
    },
    {
      "epoch": 1.0211893071967968,
      "grad_norm": 0.028897559270262718,
      "learning_rate": 0.0012354553761803229,
      "loss": 2.4998247623443604,
      "step": 7422
    },
    {
      "epoch": 1.0214644930045473,
      "grad_norm": 0.10869979858398438,
      "learning_rate": 0.0012342369783734388,
      "loss": 2.5266571044921875,
      "step": 7424
    },
    {
      "epoch": 1.021739678812298,
      "grad_norm": 0.06494247168302536,
      "learning_rate": 0.0012330185805665551,
      "loss": 2.541714668273926,
      "step": 7426
    },
    {
      "epoch": 1.0220148646200486,
      "grad_norm": 0.0357535220682621,
      "learning_rate": 0.001231800182759671,
      "loss": 2.5229992866516113,
      "step": 7428
    },
    {
      "epoch": 1.0222900504277992,
      "grad_norm": 0.03860190138220787,
      "learning_rate": 0.0012305817849527871,
      "loss": 2.5574865341186523,
      "step": 7430
    },
    {
      "epoch": 1.02256523623555,
      "grad_norm": 0.0941491648554802,
      "learning_rate": 0.001229363387145903,
      "loss": 2.508007764816284,
      "step": 7432
    },
    {
      "epoch": 1.0228404220433005,
      "grad_norm": 0.03193780779838562,
      "learning_rate": 0.0012281449893390194,
      "loss": 2.5525832176208496,
      "step": 7434
    },
    {
      "epoch": 1.023115607851051,
      "grad_norm": 0.09915158152580261,
      "learning_rate": 0.0012269265915321353,
      "loss": 2.5441651344299316,
      "step": 7436
    },
    {
      "epoch": 1.0233907936588018,
      "grad_norm": 0.02887677028775215,
      "learning_rate": 0.0012257081937252512,
      "loss": 2.5137362480163574,
      "step": 7438
    },
    {
      "epoch": 1.0236659794665524,
      "grad_norm": 0.028164036571979523,
      "learning_rate": 0.0012244897959183675,
      "loss": 2.5606296062469482,
      "step": 7440
    },
    {
      "epoch": 1.023941165274303,
      "grad_norm": 0.08026479929685593,
      "learning_rate": 0.0012232713981114834,
      "loss": 2.518911361694336,
      "step": 7442
    },
    {
      "epoch": 1.0242163510820534,
      "grad_norm": 0.0534331351518631,
      "learning_rate": 0.0012220530003045995,
      "loss": 2.5605833530426025,
      "step": 7444
    },
    {
      "epoch": 1.0244915368898042,
      "grad_norm": 0.04805612936615944,
      "learning_rate": 0.0012208346024977154,
      "loss": 2.558666229248047,
      "step": 7446
    },
    {
      "epoch": 1.0247667226975548,
      "grad_norm": 0.10011253505945206,
      "learning_rate": 0.0012196162046908317,
      "loss": 2.523916482925415,
      "step": 7448
    },
    {
      "epoch": 1.0250419085053053,
      "grad_norm": 0.03302185609936714,
      "learning_rate": 0.0012183978068839476,
      "loss": 2.5757739543914795,
      "step": 7450
    },
    {
      "epoch": 1.025317094313056,
      "grad_norm": 0.17415738105773926,
      "learning_rate": 0.0012171794090770635,
      "loss": 2.557015895843506,
      "step": 7452
    },
    {
      "epoch": 1.0255922801208066,
      "grad_norm": 0.03359447792172432,
      "learning_rate": 0.0012159610112701799,
      "loss": 2.5430946350097656,
      "step": 7454
    },
    {
      "epoch": 1.0258674659285572,
      "grad_norm": 0.09561925381422043,
      "learning_rate": 0.0012147426134632958,
      "loss": 2.539726734161377,
      "step": 7456
    },
    {
      "epoch": 1.0261426517363077,
      "grad_norm": 0.12434910982847214,
      "learning_rate": 0.001213524215656412,
      "loss": 2.482408046722412,
      "step": 7458
    },
    {
      "epoch": 1.0264178375440585,
      "grad_norm": 0.03159698098897934,
      "learning_rate": 0.0012123058178495278,
      "loss": 2.5482029914855957,
      "step": 7460
    },
    {
      "epoch": 1.026693023351809,
      "grad_norm": 0.07228115200996399,
      "learning_rate": 0.0012110874200426441,
      "loss": 2.5195844173431396,
      "step": 7462
    },
    {
      "epoch": 1.0269682091595596,
      "grad_norm": 0.16264796257019043,
      "learning_rate": 0.00120986902223576,
      "loss": 2.5321953296661377,
      "step": 7464
    },
    {
      "epoch": 1.0272433949673103,
      "grad_norm": 0.13098353147506714,
      "learning_rate": 0.001208650624428876,
      "loss": 2.487807273864746,
      "step": 7466
    },
    {
      "epoch": 1.0275185807750609,
      "grad_norm": 0.052263256162405014,
      "learning_rate": 0.001207432226621992,
      "loss": 2.5411477088928223,
      "step": 7468
    },
    {
      "epoch": 1.0277937665828114,
      "grad_norm": 0.21254105865955353,
      "learning_rate": 0.0012062138288151082,
      "loss": 2.5724129676818848,
      "step": 7470
    },
    {
      "epoch": 1.028068952390562,
      "grad_norm": 0.2855127155780792,
      "learning_rate": 0.0012049954310082243,
      "loss": 2.541253089904785,
      "step": 7472
    },
    {
      "epoch": 1.0283441381983127,
      "grad_norm": 0.25111064314842224,
      "learning_rate": 0.0012037770332013402,
      "loss": 2.4996137619018555,
      "step": 7474
    },
    {
      "epoch": 1.0286193240060633,
      "grad_norm": 0.2702436149120331,
      "learning_rate": 0.0012025586353944565,
      "loss": 2.5505778789520264,
      "step": 7476
    },
    {
      "epoch": 1.0288945098138138,
      "grad_norm": 0.19982025027275085,
      "learning_rate": 0.0012013402375875724,
      "loss": 2.518765449523926,
      "step": 7478
    },
    {
      "epoch": 1.0291696956215646,
      "grad_norm": 0.17976313829421997,
      "learning_rate": 0.0012001218397806883,
      "loss": 2.5274267196655273,
      "step": 7480
    },
    {
      "epoch": 1.0294448814293151,
      "grad_norm": 0.1306782215833664,
      "learning_rate": 0.0011989034419738044,
      "loss": 2.5047402381896973,
      "step": 7482
    },
    {
      "epoch": 1.0297200672370657,
      "grad_norm": 0.08974335342645645,
      "learning_rate": 0.0011976850441669205,
      "loss": 2.555525779724121,
      "step": 7484
    },
    {
      "epoch": 1.0299952530448162,
      "grad_norm": 0.11422887444496155,
      "learning_rate": 0.0011964666463600367,
      "loss": 2.5281882286071777,
      "step": 7486
    },
    {
      "epoch": 1.030270438852567,
      "grad_norm": 0.12416139990091324,
      "learning_rate": 0.0011952482485531526,
      "loss": 2.5537619590759277,
      "step": 7488
    },
    {
      "epoch": 1.0305456246603175,
      "grad_norm": 0.08074823021888733,
      "learning_rate": 0.0011940298507462687,
      "loss": 2.564663887023926,
      "step": 7490
    },
    {
      "epoch": 1.030820810468068,
      "grad_norm": 0.04514481872320175,
      "learning_rate": 0.0011928114529393848,
      "loss": 2.513550043106079,
      "step": 7492
    },
    {
      "epoch": 1.0310959962758188,
      "grad_norm": 0.032352667301893234,
      "learning_rate": 0.0011915930551325007,
      "loss": 2.5028958320617676,
      "step": 7494
    },
    {
      "epoch": 1.0313711820835694,
      "grad_norm": 0.028461338952183723,
      "learning_rate": 0.0011903746573256168,
      "loss": 2.52158522605896,
      "step": 7496
    },
    {
      "epoch": 1.03164636789132,
      "grad_norm": 0.03582771122455597,
      "learning_rate": 0.001189156259518733,
      "loss": 2.5377438068389893,
      "step": 7498
    },
    {
      "epoch": 1.0319215536990705,
      "grad_norm": 0.035361990332603455,
      "learning_rate": 0.001187937861711849,
      "loss": 2.498704433441162,
      "step": 7500
    },
    {
      "epoch": 1.0321967395068212,
      "grad_norm": 0.029158547520637512,
      "learning_rate": 0.001186719463904965,
      "loss": 2.5212507247924805,
      "step": 7502
    },
    {
      "epoch": 1.0324719253145718,
      "grad_norm": 0.028149114921689034,
      "learning_rate": 0.001185501066098081,
      "loss": 2.5177206993103027,
      "step": 7504
    },
    {
      "epoch": 1.0327471111223223,
      "grad_norm": 0.027300765737891197,
      "learning_rate": 0.0011842826682911972,
      "loss": 2.541079044342041,
      "step": 7506
    },
    {
      "epoch": 1.033022296930073,
      "grad_norm": 0.05785655230283737,
      "learning_rate": 0.001183064270484313,
      "loss": 2.5435757637023926,
      "step": 7508
    },
    {
      "epoch": 1.0332974827378236,
      "grad_norm": 0.08600085973739624,
      "learning_rate": 0.0011818458726774292,
      "loss": 2.5311989784240723,
      "step": 7510
    },
    {
      "epoch": 1.0335726685455742,
      "grad_norm": 0.04378209635615349,
      "learning_rate": 0.0011806274748705453,
      "loss": 2.5133886337280273,
      "step": 7512
    },
    {
      "epoch": 1.0338478543533247,
      "grad_norm": 0.04048123210668564,
      "learning_rate": 0.0011794090770636614,
      "loss": 2.481541872024536,
      "step": 7514
    },
    {
      "epoch": 1.0341230401610755,
      "grad_norm": 0.06147770583629608,
      "learning_rate": 0.0011781906792567773,
      "loss": 2.5424647331237793,
      "step": 7516
    },
    {
      "epoch": 1.034398225968826,
      "grad_norm": 0.0632287934422493,
      "learning_rate": 0.0011769722814498934,
      "loss": 2.4919755458831787,
      "step": 7518
    },
    {
      "epoch": 1.0346734117765766,
      "grad_norm": 0.0748218297958374,
      "learning_rate": 0.0011757538836430095,
      "loss": 2.527064085006714,
      "step": 7520
    },
    {
      "epoch": 1.0349485975843273,
      "grad_norm": 0.02942631021142006,
      "learning_rate": 0.0011745354858361254,
      "loss": 2.5197575092315674,
      "step": 7522
    },
    {
      "epoch": 1.0352237833920779,
      "grad_norm": 0.09942524135112762,
      "learning_rate": 0.0011733170880292416,
      "loss": 2.5358071327209473,
      "step": 7524
    },
    {
      "epoch": 1.0354989691998284,
      "grad_norm": 0.0316058024764061,
      "learning_rate": 0.0011720986902223577,
      "loss": 2.5061702728271484,
      "step": 7526
    },
    {
      "epoch": 1.0357741550075792,
      "grad_norm": 0.15198813378810883,
      "learning_rate": 0.0011708802924154738,
      "loss": 2.4858479499816895,
      "step": 7528
    },
    {
      "epoch": 1.0360493408153297,
      "grad_norm": 0.1783953756093979,
      "learning_rate": 0.0011696618946085897,
      "loss": 2.505160331726074,
      "step": 7530
    },
    {
      "epoch": 1.0363245266230803,
      "grad_norm": 0.04489297419786453,
      "learning_rate": 0.0011684434968017056,
      "loss": 2.5444142818450928,
      "step": 7532
    },
    {
      "epoch": 1.0365997124308308,
      "grad_norm": 0.06433285027742386,
      "learning_rate": 0.001167225098994822,
      "loss": 2.539031505584717,
      "step": 7534
    },
    {
      "epoch": 1.0368748982385816,
      "grad_norm": 0.0654807761311531,
      "learning_rate": 0.0011660067011879378,
      "loss": 2.5092992782592773,
      "step": 7536
    },
    {
      "epoch": 1.0371500840463321,
      "grad_norm": 0.0758916363120079,
      "learning_rate": 0.001164788303381054,
      "loss": 2.5354621410369873,
      "step": 7538
    },
    {
      "epoch": 1.0374252698540827,
      "grad_norm": 0.12544947862625122,
      "learning_rate": 0.00116356990557417,
      "loss": 2.5135955810546875,
      "step": 7540
    },
    {
      "epoch": 1.0377004556618332,
      "grad_norm": 0.04105224087834358,
      "learning_rate": 0.0011623515077672862,
      "loss": 2.5379486083984375,
      "step": 7542
    },
    {
      "epoch": 1.037975641469584,
      "grad_norm": 0.1303362250328064,
      "learning_rate": 0.001161133109960402,
      "loss": 2.5418901443481445,
      "step": 7544
    },
    {
      "epoch": 1.0382508272773345,
      "grad_norm": 0.19782871007919312,
      "learning_rate": 0.001159914712153518,
      "loss": 2.528488874435425,
      "step": 7546
    },
    {
      "epoch": 1.038526013085085,
      "grad_norm": 0.08752742409706116,
      "learning_rate": 0.0011586963143466343,
      "loss": 2.5408596992492676,
      "step": 7548
    },
    {
      "epoch": 1.0388011988928358,
      "grad_norm": 0.04428254812955856,
      "learning_rate": 0.0011574779165397502,
      "loss": 2.500795841217041,
      "step": 7550
    },
    {
      "epoch": 1.0390763847005864,
      "grad_norm": 0.06997068226337433,
      "learning_rate": 0.0011562595187328663,
      "loss": 2.5021228790283203,
      "step": 7552
    },
    {
      "epoch": 1.039351570508337,
      "grad_norm": 0.08516158163547516,
      "learning_rate": 0.0011550411209259824,
      "loss": 2.494102954864502,
      "step": 7554
    },
    {
      "epoch": 1.0396267563160877,
      "grad_norm": 0.15047799050807953,
      "learning_rate": 0.0011538227231190983,
      "loss": 2.5169739723205566,
      "step": 7556
    },
    {
      "epoch": 1.0399019421238382,
      "grad_norm": 0.10774809122085571,
      "learning_rate": 0.0011526043253122144,
      "loss": 2.5353922843933105,
      "step": 7558
    },
    {
      "epoch": 1.0401771279315888,
      "grad_norm": 0.15467660129070282,
      "learning_rate": 0.0011513859275053303,
      "loss": 2.5258984565734863,
      "step": 7560
    },
    {
      "epoch": 1.0401771279315888,
      "eval_loss": 2.5668392181396484,
      "eval_runtime": 379.2637,
      "eval_samples_per_second": 92.724,
      "eval_steps_per_second": 11.591,
      "step": 7560
    },
    {
      "epoch": 1.0404523137393393,
      "grad_norm": 0.04463931918144226,
      "learning_rate": 0.0011501675296984467,
      "loss": 2.4916763305664062,
      "step": 7562
    },
    {
      "epoch": 1.04072749954709,
      "grad_norm": 0.055113792419433594,
      "learning_rate": 0.0011489491318915626,
      "loss": 2.5606608390808105,
      "step": 7564
    },
    {
      "epoch": 1.0410026853548406,
      "grad_norm": 0.04213327541947365,
      "learning_rate": 0.0011477307340846787,
      "loss": 2.5128026008605957,
      "step": 7566
    },
    {
      "epoch": 1.0412778711625912,
      "grad_norm": 0.02955934777855873,
      "learning_rate": 0.0011465123362777948,
      "loss": 2.5285396575927734,
      "step": 7568
    },
    {
      "epoch": 1.041553056970342,
      "grad_norm": 0.04015043005347252,
      "learning_rate": 0.0011452939384709107,
      "loss": 2.532573699951172,
      "step": 7570
    },
    {
      "epoch": 1.0418282427780925,
      "grad_norm": 0.10096640884876251,
      "learning_rate": 0.0011440755406640268,
      "loss": 2.51096248626709,
      "step": 7572
    },
    {
      "epoch": 1.042103428585843,
      "grad_norm": 0.029273811727762222,
      "learning_rate": 0.0011428571428571427,
      "loss": 2.5088725090026855,
      "step": 7574
    },
    {
      "epoch": 1.0423786143935936,
      "grad_norm": 0.062396466732025146,
      "learning_rate": 0.001141638745050259,
      "loss": 2.542689323425293,
      "step": 7576
    },
    {
      "epoch": 1.0426538002013443,
      "grad_norm": 0.10129246115684509,
      "learning_rate": 0.001140420347243375,
      "loss": 2.520679473876953,
      "step": 7578
    },
    {
      "epoch": 1.042928986009095,
      "grad_norm": 0.030465004965662956,
      "learning_rate": 0.001139201949436491,
      "loss": 2.5268502235412598,
      "step": 7580
    },
    {
      "epoch": 1.0432041718168454,
      "grad_norm": 0.0607750341296196,
      "learning_rate": 0.0011379835516296072,
      "loss": 2.542598247528076,
      "step": 7582
    },
    {
      "epoch": 1.0434793576245962,
      "grad_norm": 0.10560689121484756,
      "learning_rate": 0.001136765153822723,
      "loss": 2.595358371734619,
      "step": 7584
    },
    {
      "epoch": 1.0437545434323467,
      "grad_norm": 0.044016752392053604,
      "learning_rate": 0.0011355467560158392,
      "loss": 2.5235791206359863,
      "step": 7586
    },
    {
      "epoch": 1.0440297292400973,
      "grad_norm": 0.12500011920928955,
      "learning_rate": 0.001134328358208955,
      "loss": 2.5341014862060547,
      "step": 7588
    },
    {
      "epoch": 1.0443049150478478,
      "grad_norm": 0.19439207017421722,
      "learning_rate": 0.0011331099604020714,
      "loss": 2.488471031188965,
      "step": 7590
    },
    {
      "epoch": 1.0445801008555986,
      "grad_norm": 0.11615044623613358,
      "learning_rate": 0.0011318915625951873,
      "loss": 2.516500949859619,
      "step": 7592
    },
    {
      "epoch": 1.0448552866633491,
      "grad_norm": 0.029947154223918915,
      "learning_rate": 0.0011306731647883035,
      "loss": 2.4900379180908203,
      "step": 7594
    },
    {
      "epoch": 1.0451304724710997,
      "grad_norm": 0.09343954920768738,
      "learning_rate": 0.0011294547669814196,
      "loss": 2.5024590492248535,
      "step": 7596
    },
    {
      "epoch": 1.0454056582788505,
      "grad_norm": 0.10305511206388474,
      "learning_rate": 0.0011282363691745355,
      "loss": 2.531059741973877,
      "step": 7598
    },
    {
      "epoch": 1.045680844086601,
      "grad_norm": 0.030482089146971703,
      "learning_rate": 0.0011270179713676516,
      "loss": 2.5159943103790283,
      "step": 7600
    },
    {
      "epoch": 1.0459560298943515,
      "grad_norm": 0.05789951607584953,
      "learning_rate": 0.0011257995735607675,
      "loss": 2.5297482013702393,
      "step": 7602
    },
    {
      "epoch": 1.046231215702102,
      "grad_norm": 0.05591101571917534,
      "learning_rate": 0.0011245811757538838,
      "loss": 2.530040740966797,
      "step": 7604
    },
    {
      "epoch": 1.0465064015098529,
      "grad_norm": 0.036175552755594254,
      "learning_rate": 0.0011233627779469997,
      "loss": 2.5357680320739746,
      "step": 7606
    },
    {
      "epoch": 1.0467815873176034,
      "grad_norm": 0.04643429443240166,
      "learning_rate": 0.0011221443801401156,
      "loss": 2.5502233505249023,
      "step": 7608
    },
    {
      "epoch": 1.047056773125354,
      "grad_norm": 0.07070915400981903,
      "learning_rate": 0.001120925982333232,
      "loss": 2.475275754928589,
      "step": 7610
    },
    {
      "epoch": 1.0473319589331047,
      "grad_norm": 0.06161219999194145,
      "learning_rate": 0.0011197075845263478,
      "loss": 2.525135040283203,
      "step": 7612
    },
    {
      "epoch": 1.0476071447408553,
      "grad_norm": 0.053855061531066895,
      "learning_rate": 0.001118489186719464,
      "loss": 2.5650601387023926,
      "step": 7614
    },
    {
      "epoch": 1.0478823305486058,
      "grad_norm": 0.03149452432990074,
      "learning_rate": 0.0011172707889125799,
      "loss": 2.5311837196350098,
      "step": 7616
    },
    {
      "epoch": 1.0481575163563563,
      "grad_norm": 0.0424458272755146,
      "learning_rate": 0.0011160523911056962,
      "loss": 2.5258917808532715,
      "step": 7618
    },
    {
      "epoch": 1.048432702164107,
      "grad_norm": 0.05533222481608391,
      "learning_rate": 0.001114833993298812,
      "loss": 2.4927263259887695,
      "step": 7620
    },
    {
      "epoch": 1.0487078879718577,
      "grad_norm": 0.11531311273574829,
      "learning_rate": 0.001113615595491928,
      "loss": 2.525630474090576,
      "step": 7622
    },
    {
      "epoch": 1.0489830737796082,
      "grad_norm": 0.04452578350901604,
      "learning_rate": 0.0011123971976850441,
      "loss": 2.4892609119415283,
      "step": 7624
    },
    {
      "epoch": 1.049258259587359,
      "grad_norm": 0.11699841171503067,
      "learning_rate": 0.0011111787998781602,
      "loss": 2.5158042907714844,
      "step": 7626
    },
    {
      "epoch": 1.0495334453951095,
      "grad_norm": 0.1365104764699936,
      "learning_rate": 0.0011099604020712763,
      "loss": 2.5120627880096436,
      "step": 7628
    },
    {
      "epoch": 1.04980863120286,
      "grad_norm": 0.02865118347108364,
      "learning_rate": 0.0011087420042643922,
      "loss": 2.4968624114990234,
      "step": 7630
    },
    {
      "epoch": 1.0500838170106106,
      "grad_norm": 0.10870101302862167,
      "learning_rate": 0.0011075236064575086,
      "loss": 2.5440869331359863,
      "step": 7632
    },
    {
      "epoch": 1.0503590028183614,
      "grad_norm": 0.133027583360672,
      "learning_rate": 0.0011063052086506245,
      "loss": 2.5158097743988037,
      "step": 7634
    },
    {
      "epoch": 1.050634188626112,
      "grad_norm": 0.05654548481106758,
      "learning_rate": 0.0011050868108437404,
      "loss": 2.5303473472595215,
      "step": 7636
    },
    {
      "epoch": 1.0509093744338625,
      "grad_norm": 0.03577905520796776,
      "learning_rate": 0.0011038684130368565,
      "loss": 2.513824462890625,
      "step": 7638
    },
    {
      "epoch": 1.0511845602416132,
      "grad_norm": 0.05204233527183533,
      "learning_rate": 0.0011026500152299726,
      "loss": 2.5200281143188477,
      "step": 7640
    },
    {
      "epoch": 1.0514597460493638,
      "grad_norm": 0.12694001197814941,
      "learning_rate": 0.0011014316174230887,
      "loss": 2.4975500106811523,
      "step": 7642
    },
    {
      "epoch": 1.0517349318571143,
      "grad_norm": 0.13062381744384766,
      "learning_rate": 0.0011002132196162046,
      "loss": 2.508416175842285,
      "step": 7644
    },
    {
      "epoch": 1.052010117664865,
      "grad_norm": 0.027690060436725616,
      "learning_rate": 0.001098994821809321,
      "loss": 2.5193538665771484,
      "step": 7646
    },
    {
      "epoch": 1.0522853034726156,
      "grad_norm": 0.08209002017974854,
      "learning_rate": 0.0010977764240024369,
      "loss": 2.536393404006958,
      "step": 7648
    },
    {
      "epoch": 1.0525604892803662,
      "grad_norm": 0.07161477208137512,
      "learning_rate": 0.0010965580261955528,
      "loss": 2.5656871795654297,
      "step": 7650
    },
    {
      "epoch": 1.0528356750881167,
      "grad_norm": 0.05971178784966469,
      "learning_rate": 0.0010953396283886689,
      "loss": 2.5210447311401367,
      "step": 7652
    },
    {
      "epoch": 1.0531108608958675,
      "grad_norm": 0.030835706740617752,
      "learning_rate": 0.001094121230581785,
      "loss": 2.537951946258545,
      "step": 7654
    },
    {
      "epoch": 1.053386046703618,
      "grad_norm": 0.04550750181078911,
      "learning_rate": 0.001092902832774901,
      "loss": 2.5317420959472656,
      "step": 7656
    },
    {
      "epoch": 1.0536612325113686,
      "grad_norm": 0.04168342798948288,
      "learning_rate": 0.001091684434968017,
      "loss": 2.525984764099121,
      "step": 7658
    },
    {
      "epoch": 1.0539364183191193,
      "grad_norm": 0.04987914860248566,
      "learning_rate": 0.0010904660371611333,
      "loss": 2.520115375518799,
      "step": 7660
    },
    {
      "epoch": 1.0542116041268699,
      "grad_norm": 0.057998016476631165,
      "learning_rate": 0.0010892476393542492,
      "loss": 2.5027456283569336,
      "step": 7662
    },
    {
      "epoch": 1.0544867899346204,
      "grad_norm": 0.09140031039714813,
      "learning_rate": 0.0010880292415473651,
      "loss": 2.5020036697387695,
      "step": 7664
    },
    {
      "epoch": 1.054761975742371,
      "grad_norm": 0.04557941108942032,
      "learning_rate": 0.0010868108437404812,
      "loss": 2.4876301288604736,
      "step": 7666
    },
    {
      "epoch": 1.0550371615501217,
      "grad_norm": 0.03929101303219795,
      "learning_rate": 0.0010855924459335974,
      "loss": 2.48909592628479,
      "step": 7668
    },
    {
      "epoch": 1.0553123473578723,
      "grad_norm": 0.1752747744321823,
      "learning_rate": 0.0010843740481267135,
      "loss": 2.524186611175537,
      "step": 7670
    },
    {
      "epoch": 1.0555875331656228,
      "grad_norm": 0.06664783507585526,
      "learning_rate": 0.0010831556503198294,
      "loss": 2.5481297969818115,
      "step": 7672
    },
    {
      "epoch": 1.0558627189733736,
      "grad_norm": 0.03367410972714424,
      "learning_rate": 0.0010819372525129455,
      "loss": 2.5129051208496094,
      "step": 7674
    },
    {
      "epoch": 1.0561379047811241,
      "grad_norm": 0.04581203684210777,
      "learning_rate": 0.0010807188547060616,
      "loss": 2.543344020843506,
      "step": 7676
    },
    {
      "epoch": 1.0564130905888747,
      "grad_norm": 0.05413931980729103,
      "learning_rate": 0.0010795004568991775,
      "loss": 2.5084519386291504,
      "step": 7678
    },
    {
      "epoch": 1.0566882763966252,
      "grad_norm": 0.06425213813781738,
      "learning_rate": 0.0010782820590922936,
      "loss": 2.543674945831299,
      "step": 7680
    },
    {
      "epoch": 1.056963462204376,
      "grad_norm": 0.0364309661090374,
      "learning_rate": 0.0010770636612854097,
      "loss": 2.528069019317627,
      "step": 7682
    },
    {
      "epoch": 1.0572386480121265,
      "grad_norm": 0.0473799966275692,
      "learning_rate": 0.0010758452634785259,
      "loss": 2.5534329414367676,
      "step": 7684
    },
    {
      "epoch": 1.057513833819877,
      "grad_norm": 0.056108392775058746,
      "learning_rate": 0.0010746268656716418,
      "loss": 2.501600980758667,
      "step": 7686
    },
    {
      "epoch": 1.0577890196276278,
      "grad_norm": 0.06938701868057251,
      "learning_rate": 0.0010734084678647577,
      "loss": 2.5005316734313965,
      "step": 7688
    },
    {
      "epoch": 1.0580642054353784,
      "grad_norm": 0.08221978694200516,
      "learning_rate": 0.001072190070057874,
      "loss": 2.4905500411987305,
      "step": 7690
    },
    {
      "epoch": 1.058339391243129,
      "grad_norm": 0.040956053882837296,
      "learning_rate": 0.0010709716722509899,
      "loss": 2.5187783241271973,
      "step": 7692
    },
    {
      "epoch": 1.0586145770508795,
      "grad_norm": 0.16014236211776733,
      "learning_rate": 0.001069753274444106,
      "loss": 2.530179262161255,
      "step": 7694
    },
    {
      "epoch": 1.0588897628586302,
      "grad_norm": 0.11360663175582886,
      "learning_rate": 0.0010685348766372221,
      "loss": 2.5051422119140625,
      "step": 7696
    },
    {
      "epoch": 1.0591649486663808,
      "grad_norm": 0.03151527792215347,
      "learning_rate": 0.0010673164788303382,
      "loss": 2.4977338314056396,
      "step": 7698
    },
    {
      "epoch": 1.0594401344741313,
      "grad_norm": 0.12496738135814667,
      "learning_rate": 0.0010660980810234541,
      "loss": 2.4871954917907715,
      "step": 7700
    },
    {
      "epoch": 1.059715320281882,
      "grad_norm": 0.06725958734750748,
      "learning_rate": 0.00106487968321657,
      "loss": 2.4964702129364014,
      "step": 7702
    },
    {
      "epoch": 1.0599905060896326,
      "grad_norm": 0.03978082910180092,
      "learning_rate": 0.0010636612854096864,
      "loss": 2.5259342193603516,
      "step": 7704
    },
    {
      "epoch": 1.0602656918973832,
      "grad_norm": 0.05468893423676491,
      "learning_rate": 0.0010624428876028023,
      "loss": 2.5476503372192383,
      "step": 7706
    },
    {
      "epoch": 1.0605408777051337,
      "grad_norm": 0.09280503541231155,
      "learning_rate": 0.0010612244897959184,
      "loss": 2.5711989402770996,
      "step": 7708
    },
    {
      "epoch": 1.0608160635128845,
      "grad_norm": 0.04298907145857811,
      "learning_rate": 0.0010600060919890345,
      "loss": 2.543311357498169,
      "step": 7710
    },
    {
      "epoch": 1.061091249320635,
      "grad_norm": 0.06027235835790634,
      "learning_rate": 0.0010587876941821506,
      "loss": 2.492586374282837,
      "step": 7712
    },
    {
      "epoch": 1.0613664351283856,
      "grad_norm": 0.08572300523519516,
      "learning_rate": 0.0010575692963752665,
      "loss": 2.542229175567627,
      "step": 7714
    },
    {
      "epoch": 1.0616416209361363,
      "grad_norm": 0.10255923122167587,
      "learning_rate": 0.0010563508985683824,
      "loss": 2.528646469116211,
      "step": 7716
    },
    {
      "epoch": 1.0619168067438869,
      "grad_norm": 0.08163150399923325,
      "learning_rate": 0.0010551325007614987,
      "loss": 2.5419416427612305,
      "step": 7718
    },
    {
      "epoch": 1.0621919925516374,
      "grad_norm": 0.0860946998000145,
      "learning_rate": 0.0010539141029546146,
      "loss": 2.5227370262145996,
      "step": 7720
    },
    {
      "epoch": 1.062467178359388,
      "grad_norm": 0.197193905711174,
      "learning_rate": 0.0010526957051477308,
      "loss": 2.5323939323425293,
      "step": 7722
    },
    {
      "epoch": 1.0627423641671387,
      "grad_norm": 0.11709121614694595,
      "learning_rate": 0.0010514773073408469,
      "loss": 2.519534111022949,
      "step": 7724
    },
    {
      "epoch": 1.0630175499748893,
      "grad_norm": 0.07198797166347504,
      "learning_rate": 0.001050258909533963,
      "loss": 2.527099132537842,
      "step": 7726
    },
    {
      "epoch": 1.0632927357826398,
      "grad_norm": 0.12351524829864502,
      "learning_rate": 0.001049040511727079,
      "loss": 2.538506507873535,
      "step": 7728
    },
    {
      "epoch": 1.0635679215903906,
      "grad_norm": 0.06939375400543213,
      "learning_rate": 0.0010478221139201948,
      "loss": 2.517972469329834,
      "step": 7730
    },
    {
      "epoch": 1.0638431073981411,
      "grad_norm": 0.027988210320472717,
      "learning_rate": 0.0010466037161133111,
      "loss": 2.517502784729004,
      "step": 7732
    },
    {
      "epoch": 1.0641182932058917,
      "grad_norm": 0.12223651260137558,
      "learning_rate": 0.001045385318306427,
      "loss": 2.4716997146606445,
      "step": 7734
    },
    {
      "epoch": 1.0643934790136425,
      "grad_norm": 0.09874579310417175,
      "learning_rate": 0.0010441669204995431,
      "loss": 2.5141549110412598,
      "step": 7736
    },
    {
      "epoch": 1.064668664821393,
      "grad_norm": 0.11859770864248276,
      "learning_rate": 0.0010429485226926593,
      "loss": 2.5123958587646484,
      "step": 7738
    },
    {
      "epoch": 1.0649438506291435,
      "grad_norm": 0.09037738293409348,
      "learning_rate": 0.0010417301248857752,
      "loss": 2.5182082653045654,
      "step": 7740
    },
    {
      "epoch": 1.065219036436894,
      "grad_norm": 0.03350543975830078,
      "learning_rate": 0.0010405117270788913,
      "loss": 2.5511903762817383,
      "step": 7742
    },
    {
      "epoch": 1.0654942222446449,
      "grad_norm": 0.03748774528503418,
      "learning_rate": 0.0010392933292720072,
      "loss": 2.516677141189575,
      "step": 7744
    },
    {
      "epoch": 1.0657694080523954,
      "grad_norm": 0.05980457365512848,
      "learning_rate": 0.0010380749314651235,
      "loss": 2.5415403842926025,
      "step": 7746
    },
    {
      "epoch": 1.066044593860146,
      "grad_norm": 0.12324986606836319,
      "learning_rate": 0.0010368565336582394,
      "loss": 2.5227127075195312,
      "step": 7748
    },
    {
      "epoch": 1.0663197796678965,
      "grad_norm": 0.12558403611183167,
      "learning_rate": 0.0010356381358513555,
      "loss": 2.515758752822876,
      "step": 7750
    },
    {
      "epoch": 1.0665949654756472,
      "grad_norm": 0.10076937824487686,
      "learning_rate": 0.0010344197380444716,
      "loss": 2.554219961166382,
      "step": 7752
    },
    {
      "epoch": 1.0668701512833978,
      "grad_norm": 0.030532481148838997,
      "learning_rate": 0.0010332013402375875,
      "loss": 2.520082950592041,
      "step": 7754
    },
    {
      "epoch": 1.0671453370911483,
      "grad_norm": 0.053134020417928696,
      "learning_rate": 0.0010319829424307037,
      "loss": 2.5101146697998047,
      "step": 7756
    },
    {
      "epoch": 1.067420522898899,
      "grad_norm": 0.10220511257648468,
      "learning_rate": 0.0010307645446238196,
      "loss": 2.557291269302368,
      "step": 7758
    },
    {
      "epoch": 1.0676957087066496,
      "grad_norm": 0.09713035821914673,
      "learning_rate": 0.0010295461468169359,
      "loss": 2.538177490234375,
      "step": 7760
    },
    {
      "epoch": 1.0679708945144002,
      "grad_norm": 0.11858706176280975,
      "learning_rate": 0.0010283277490100518,
      "loss": 2.5124032497406006,
      "step": 7762
    },
    {
      "epoch": 1.068246080322151,
      "grad_norm": 0.1018323078751564,
      "learning_rate": 0.001027109351203168,
      "loss": 2.517148971557617,
      "step": 7764
    },
    {
      "epoch": 1.0685212661299015,
      "grad_norm": 0.06299692392349243,
      "learning_rate": 0.001025890953396284,
      "loss": 2.479764938354492,
      "step": 7766
    },
    {
      "epoch": 1.068796451937652,
      "grad_norm": 0.1011025533080101,
      "learning_rate": 0.0010246725555894,
      "loss": 2.4730544090270996,
      "step": 7768
    },
    {
      "epoch": 1.0690716377454026,
      "grad_norm": 0.06987209618091583,
      "learning_rate": 0.001023454157782516,
      "loss": 2.491506576538086,
      "step": 7770
    },
    {
      "epoch": 1.0693468235531534,
      "grad_norm": 0.08819492906332016,
      "learning_rate": 0.001022235759975632,
      "loss": 2.5398166179656982,
      "step": 7772
    },
    {
      "epoch": 1.069622009360904,
      "grad_norm": 0.030576912686228752,
      "learning_rate": 0.0010210173621687483,
      "loss": 2.5506985187530518,
      "step": 7774
    },
    {
      "epoch": 1.0698971951686544,
      "grad_norm": 0.028919868171215057,
      "learning_rate": 0.0010197989643618642,
      "loss": 2.487348794937134,
      "step": 7776
    },
    {
      "epoch": 1.0701723809764052,
      "grad_norm": 0.044925518333911896,
      "learning_rate": 0.0010185805665549803,
      "loss": 2.509521007537842,
      "step": 7778
    },
    {
      "epoch": 1.0704475667841558,
      "grad_norm": 0.04470658674836159,
      "learning_rate": 0.0010173621687480962,
      "loss": 2.472842216491699,
      "step": 7780
    },
    {
      "epoch": 1.0707227525919063,
      "grad_norm": 0.04095330461859703,
      "learning_rate": 0.0010161437709412123,
      "loss": 2.4547696113586426,
      "step": 7782
    },
    {
      "epoch": 1.0709979383996568,
      "grad_norm": 0.12514528632164001,
      "learning_rate": 0.0010149253731343284,
      "loss": 2.509768009185791,
      "step": 7784
    },
    {
      "epoch": 1.0712731242074076,
      "grad_norm": 0.05614142119884491,
      "learning_rate": 0.0010137069753274443,
      "loss": 2.4867658615112305,
      "step": 7786
    },
    {
      "epoch": 1.0715483100151582,
      "grad_norm": 0.10032720863819122,
      "learning_rate": 0.0010124885775205606,
      "loss": 2.5076093673706055,
      "step": 7788
    },
    {
      "epoch": 1.0718234958229087,
      "grad_norm": 0.07548489421606064,
      "learning_rate": 0.0010112701797136765,
      "loss": 2.4617762565612793,
      "step": 7790
    },
    {
      "epoch": 1.0720986816306595,
      "grad_norm": 0.03101092390716076,
      "learning_rate": 0.0010100517819067927,
      "loss": 2.5115230083465576,
      "step": 7792
    },
    {
      "epoch": 1.07237386743841,
      "grad_norm": 0.08552227169275284,
      "learning_rate": 0.0010088333840999086,
      "loss": 2.5371346473693848,
      "step": 7794
    },
    {
      "epoch": 1.0726490532461606,
      "grad_norm": 0.09113933891057968,
      "learning_rate": 0.0010076149862930247,
      "loss": 2.5177924633026123,
      "step": 7796
    },
    {
      "epoch": 1.072924239053911,
      "grad_norm": 0.03900548815727234,
      "learning_rate": 0.0010063965884861408,
      "loss": 2.513719081878662,
      "step": 7798
    },
    {
      "epoch": 1.0731994248616619,
      "grad_norm": 0.07121025770902634,
      "learning_rate": 0.0010051781906792567,
      "loss": 2.531322956085205,
      "step": 7800
    },
    {
      "epoch": 1.0734746106694124,
      "grad_norm": 0.07578196376562119,
      "learning_rate": 0.001003959792872373,
      "loss": 2.520160675048828,
      "step": 7802
    },
    {
      "epoch": 1.073749796477163,
      "grad_norm": 0.09089653193950653,
      "learning_rate": 0.001002741395065489,
      "loss": 2.5254006385803223,
      "step": 7804
    },
    {
      "epoch": 1.0740249822849137,
      "grad_norm": 0.05442091077566147,
      "learning_rate": 0.0010015229972586048,
      "loss": 2.5380349159240723,
      "step": 7806
    },
    {
      "epoch": 1.0743001680926643,
      "grad_norm": 0.07179011404514313,
      "learning_rate": 0.001000304599451721,
      "loss": 2.494250774383545,
      "step": 7808
    },
    {
      "epoch": 1.0745753539004148,
      "grad_norm": 0.06284698098897934,
      "learning_rate": 0.000999086201644837,
      "loss": 2.550926923751831,
      "step": 7810
    },
    {
      "epoch": 1.0748505397081654,
      "grad_norm": 0.029302721843123436,
      "learning_rate": 0.000997867803837953,
      "loss": 2.5502538681030273,
      "step": 7812
    },
    {
      "epoch": 1.0751257255159161,
      "grad_norm": 0.059922829270362854,
      "learning_rate": 0.000996649406031069,
      "loss": 2.520132064819336,
      "step": 7814
    },
    {
      "epoch": 1.0754009113236667,
      "grad_norm": 0.0714743435382843,
      "learning_rate": 0.0009954310082241852,
      "loss": 2.517899513244629,
      "step": 7816
    },
    {
      "epoch": 1.0756760971314172,
      "grad_norm": 0.043015312403440475,
      "learning_rate": 0.0009942126104173013,
      "loss": 2.541677713394165,
      "step": 7818
    },
    {
      "epoch": 1.075951282939168,
      "grad_norm": 0.051190853118896484,
      "learning_rate": 0.0009929942126104174,
      "loss": 2.5175364017486572,
      "step": 7820
    },
    {
      "epoch": 1.0762264687469185,
      "grad_norm": 0.03210562840104103,
      "learning_rate": 0.0009917758148035333,
      "loss": 2.514192581176758,
      "step": 7822
    },
    {
      "epoch": 1.076501654554669,
      "grad_norm": 0.029547307640314102,
      "learning_rate": 0.0009905574169966494,
      "loss": 2.5463223457336426,
      "step": 7824
    },
    {
      "epoch": 1.0767768403624196,
      "grad_norm": 0.05703118070960045,
      "learning_rate": 0.0009893390191897653,
      "loss": 2.499013900756836,
      "step": 7826
    },
    {
      "epoch": 1.0770520261701704,
      "grad_norm": 0.046748556196689606,
      "learning_rate": 0.0009881206213828814,
      "loss": 2.5282273292541504,
      "step": 7828
    },
    {
      "epoch": 1.077327211977921,
      "grad_norm": 0.049288567155599594,
      "learning_rate": 0.0009869022235759976,
      "loss": 2.528913974761963,
      "step": 7830
    },
    {
      "epoch": 1.0776023977856715,
      "grad_norm": 0.0861998200416565,
      "learning_rate": 0.0009856838257691137,
      "loss": 2.5199244022369385,
      "step": 7832
    },
    {
      "epoch": 1.0778775835934222,
      "grad_norm": 0.03964720293879509,
      "learning_rate": 0.0009844654279622298,
      "loss": 2.502685070037842,
      "step": 7834
    },
    {
      "epoch": 1.0781527694011728,
      "grad_norm": 0.03722946345806122,
      "learning_rate": 0.0009832470301553457,
      "loss": 2.4863009452819824,
      "step": 7836
    },
    {
      "epoch": 1.0784279552089233,
      "grad_norm": 0.044494617730379105,
      "learning_rate": 0.0009820286323484618,
      "loss": 2.5061445236206055,
      "step": 7838
    },
    {
      "epoch": 1.0787031410166739,
      "grad_norm": 0.07911401242017746,
      "learning_rate": 0.0009808102345415777,
      "loss": 2.516676187515259,
      "step": 7840
    },
    {
      "epoch": 1.0789783268244246,
      "grad_norm": 0.040548231452703476,
      "learning_rate": 0.0009795918367346938,
      "loss": 2.5281434059143066,
      "step": 7842
    },
    {
      "epoch": 1.0792535126321752,
      "grad_norm": 0.047396060079336166,
      "learning_rate": 0.00097837343892781,
      "loss": 2.5466556549072266,
      "step": 7844
    },
    {
      "epoch": 1.0795286984399257,
      "grad_norm": 0.06880377978086472,
      "learning_rate": 0.000977155041120926,
      "loss": 2.523861885070801,
      "step": 7846
    },
    {
      "epoch": 1.0798038842476765,
      "grad_norm": 0.09871715307235718,
      "learning_rate": 0.0009759366433140421,
      "loss": 2.5132174491882324,
      "step": 7848
    },
    {
      "epoch": 1.080079070055427,
      "grad_norm": 0.10644891113042831,
      "learning_rate": 0.0009747182455071582,
      "loss": 2.4907140731811523,
      "step": 7850
    },
    {
      "epoch": 1.0803542558631776,
      "grad_norm": 0.02886623702943325,
      "learning_rate": 0.0009734998477002741,
      "loss": 2.5220909118652344,
      "step": 7852
    },
    {
      "epoch": 1.0806294416709283,
      "grad_norm": 0.12060966342687607,
      "learning_rate": 0.0009722814498933902,
      "loss": 2.50007700920105,
      "step": 7854
    },
    {
      "epoch": 1.0809046274786789,
      "grad_norm": 0.1237286627292633,
      "learning_rate": 0.0009710630520865062,
      "loss": 2.508582353591919,
      "step": 7856
    },
    {
      "epoch": 1.0811798132864294,
      "grad_norm": 0.029192527756094933,
      "learning_rate": 0.0009698446542796223,
      "loss": 2.53432035446167,
      "step": 7858
    },
    {
      "epoch": 1.08145499909418,
      "grad_norm": 0.026801563799381256,
      "learning_rate": 0.0009686262564727384,
      "loss": 2.5256576538085938,
      "step": 7860
    },
    {
      "epoch": 1.0817301849019307,
      "grad_norm": 0.09236399829387665,
      "learning_rate": 0.0009674078586658544,
      "loss": 2.4989218711853027,
      "step": 7862
    },
    {
      "epoch": 1.0820053707096813,
      "grad_norm": 0.13578101992607117,
      "learning_rate": 0.0009661894608589706,
      "loss": 2.5255696773529053,
      "step": 7864
    },
    {
      "epoch": 1.0822805565174318,
      "grad_norm": 0.10815603286027908,
      "learning_rate": 0.0009649710630520865,
      "loss": 2.50211238861084,
      "step": 7866
    },
    {
      "epoch": 1.0825557423251824,
      "grad_norm": 0.11884628981351852,
      "learning_rate": 0.0009637526652452026,
      "loss": 2.559054374694824,
      "step": 7868
    },
    {
      "epoch": 1.0828309281329331,
      "grad_norm": 0.14341236650943756,
      "learning_rate": 0.0009625342674383186,
      "loss": 2.5160670280456543,
      "step": 7870
    },
    {
      "epoch": 1.0831061139406837,
      "grad_norm": 0.07547520846128464,
      "learning_rate": 0.0009613158696314347,
      "loss": 2.485997200012207,
      "step": 7872
    },
    {
      "epoch": 1.0833812997484342,
      "grad_norm": 0.04206234961748123,
      "learning_rate": 0.0009600974718245507,
      "loss": 2.5211124420166016,
      "step": 7874
    },
    {
      "epoch": 1.083656485556185,
      "grad_norm": 0.027857838198542595,
      "learning_rate": 0.0009588790740176668,
      "loss": 2.5488996505737305,
      "step": 7876
    },
    {
      "epoch": 1.0839316713639355,
      "grad_norm": 0.03061770647764206,
      "learning_rate": 0.0009576606762107828,
      "loss": 2.482276439666748,
      "step": 7878
    },
    {
      "epoch": 1.084206857171686,
      "grad_norm": 0.034743938595056534,
      "learning_rate": 0.0009564422784038988,
      "loss": 2.527097702026367,
      "step": 7880
    },
    {
      "epoch": 1.0844820429794368,
      "grad_norm": 0.0954989492893219,
      "learning_rate": 0.000955223880597015,
      "loss": 2.541818141937256,
      "step": 7882
    },
    {
      "epoch": 1.0847572287871874,
      "grad_norm": 0.13192163407802582,
      "learning_rate": 0.000954005482790131,
      "loss": 2.459080934524536,
      "step": 7884
    },
    {
      "epoch": 1.085032414594938,
      "grad_norm": 0.08750738203525543,
      "learning_rate": 0.0009527870849832471,
      "loss": 2.488492488861084,
      "step": 7886
    },
    {
      "epoch": 1.0853076004026885,
      "grad_norm": 0.05986379086971283,
      "learning_rate": 0.0009515686871763631,
      "loss": 2.508082151412964,
      "step": 7888
    },
    {
      "epoch": 1.0855827862104392,
      "grad_norm": 0.04717819020152092,
      "learning_rate": 0.0009503502893694792,
      "loss": 2.499436855316162,
      "step": 7890
    },
    {
      "epoch": 1.0858579720181898,
      "grad_norm": 0.043836306780576706,
      "learning_rate": 0.0009491318915625952,
      "loss": 2.5440592765808105,
      "step": 7892
    },
    {
      "epoch": 1.0861331578259403,
      "grad_norm": 0.037467990070581436,
      "learning_rate": 0.0009479134937557112,
      "loss": 2.5103745460510254,
      "step": 7894
    },
    {
      "epoch": 1.086408343633691,
      "grad_norm": 0.0912962257862091,
      "learning_rate": 0.0009466950959488273,
      "loss": 2.538893699645996,
      "step": 7896
    },
    {
      "epoch": 1.0866835294414416,
      "grad_norm": 0.14282764494419098,
      "learning_rate": 0.0009454766981419433,
      "loss": 2.5439963340759277,
      "step": 7898
    },
    {
      "epoch": 1.0869587152491922,
      "grad_norm": 0.11134784668684006,
      "learning_rate": 0.0009442583003350595,
      "loss": 2.508615493774414,
      "step": 7900
    },
    {
      "epoch": 1.0872339010569427,
      "grad_norm": 0.11268293857574463,
      "learning_rate": 0.0009430399025281755,
      "loss": 2.5164334774017334,
      "step": 7902
    },
    {
      "epoch": 1.0875090868646935,
      "grad_norm": 0.14991158246994019,
      "learning_rate": 0.0009418215047212915,
      "loss": 2.546978712081909,
      "step": 7904
    },
    {
      "epoch": 1.087784272672444,
      "grad_norm": 0.10124582052230835,
      "learning_rate": 0.0009406031069144075,
      "loss": 2.48608660697937,
      "step": 7906
    },
    {
      "epoch": 1.0880594584801946,
      "grad_norm": 0.07817636430263519,
      "learning_rate": 0.0009393847091075236,
      "loss": 2.546621799468994,
      "step": 7908
    },
    {
      "epoch": 1.0883346442879454,
      "grad_norm": 0.06455856561660767,
      "learning_rate": 0.0009381663113006397,
      "loss": 2.5090723037719727,
      "step": 7910
    },
    {
      "epoch": 1.088609830095696,
      "grad_norm": 0.04907487332820892,
      "learning_rate": 0.0009369479134937557,
      "loss": 2.549452781677246,
      "step": 7912
    },
    {
      "epoch": 1.0888850159034464,
      "grad_norm": 0.040508072823286057,
      "learning_rate": 0.0009357295156868718,
      "loss": 2.511445999145508,
      "step": 7914
    },
    {
      "epoch": 1.089160201711197,
      "grad_norm": 0.05233365297317505,
      "learning_rate": 0.0009345111178799878,
      "loss": 2.528559684753418,
      "step": 7916
    },
    {
      "epoch": 1.0894353875189478,
      "grad_norm": 0.02800733596086502,
      "learning_rate": 0.0009332927200731039,
      "loss": 2.5004055500030518,
      "step": 7918
    },
    {
      "epoch": 1.0897105733266983,
      "grad_norm": 0.058708641678094864,
      "learning_rate": 0.0009320743222662199,
      "loss": 2.5374960899353027,
      "step": 7920
    },
    {
      "epoch": 1.0899857591344488,
      "grad_norm": 0.06864805519580841,
      "learning_rate": 0.000930855924459336,
      "loss": 2.482239246368408,
      "step": 7922
    },
    {
      "epoch": 1.0902609449421996,
      "grad_norm": 0.03645868971943855,
      "learning_rate": 0.0009296375266524521,
      "loss": 2.460970401763916,
      "step": 7924
    },
    {
      "epoch": 1.0905361307499501,
      "grad_norm": 0.03661692142486572,
      "learning_rate": 0.0009284191288455681,
      "loss": 2.5057411193847656,
      "step": 7926
    },
    {
      "epoch": 1.0908113165577007,
      "grad_norm": 0.02982218563556671,
      "learning_rate": 0.0009272007310386842,
      "loss": 2.518641233444214,
      "step": 7928
    },
    {
      "epoch": 1.0910865023654512,
      "grad_norm": 0.08623287081718445,
      "learning_rate": 0.0009259823332318002,
      "loss": 2.506864547729492,
      "step": 7930
    },
    {
      "epoch": 1.091361688173202,
      "grad_norm": 0.03514585271477699,
      "learning_rate": 0.0009247639354249162,
      "loss": 2.4904401302337646,
      "step": 7932
    },
    {
      "epoch": 1.0916368739809525,
      "grad_norm": 0.04592243582010269,
      "learning_rate": 0.0009235455376180322,
      "loss": 2.5033435821533203,
      "step": 7934
    },
    {
      "epoch": 1.091912059788703,
      "grad_norm": 0.0292506143450737,
      "learning_rate": 0.0009223271398111484,
      "loss": 2.4804205894470215,
      "step": 7936
    },
    {
      "epoch": 1.0921872455964539,
      "grad_norm": 0.03731818497180939,
      "learning_rate": 0.0009211087420042645,
      "loss": 2.533651828765869,
      "step": 7938
    },
    {
      "epoch": 1.0924624314042044,
      "grad_norm": 0.04809374362230301,
      "learning_rate": 0.0009198903441973805,
      "loss": 2.476861000061035,
      "step": 7940
    },
    {
      "epoch": 1.092737617211955,
      "grad_norm": 0.03450511768460274,
      "learning_rate": 0.0009186719463904966,
      "loss": 2.4993386268615723,
      "step": 7942
    },
    {
      "epoch": 1.0930128030197057,
      "grad_norm": 0.06729047000408173,
      "learning_rate": 0.0009174535485836125,
      "loss": 2.515273332595825,
      "step": 7944
    },
    {
      "epoch": 1.0932879888274563,
      "grad_norm": 0.10010652244091034,
      "learning_rate": 0.0009162351507767286,
      "loss": 2.530777931213379,
      "step": 7946
    },
    {
      "epoch": 1.0935631746352068,
      "grad_norm": 0.06394756585359573,
      "learning_rate": 0.0009150167529698446,
      "loss": 2.4871115684509277,
      "step": 7948
    },
    {
      "epoch": 1.0938383604429573,
      "grad_norm": 0.02923850156366825,
      "learning_rate": 0.0009137983551629607,
      "loss": 2.5391335487365723,
      "step": 7950
    },
    {
      "epoch": 1.0941135462507081,
      "grad_norm": 0.044644638895988464,
      "learning_rate": 0.0009125799573560767,
      "loss": 2.494527816772461,
      "step": 7952
    },
    {
      "epoch": 1.0943887320584587,
      "grad_norm": 0.07423017919063568,
      "learning_rate": 0.0009113615595491929,
      "loss": 2.495924234390259,
      "step": 7954
    },
    {
      "epoch": 1.0946639178662092,
      "grad_norm": 0.03010951168835163,
      "learning_rate": 0.000910143161742309,
      "loss": 2.4628689289093018,
      "step": 7956
    },
    {
      "epoch": 1.0949391036739597,
      "grad_norm": 0.04727386310696602,
      "learning_rate": 0.0009089247639354249,
      "loss": 2.5240964889526367,
      "step": 7958
    },
    {
      "epoch": 1.0952142894817105,
      "grad_norm": 0.029722118750214577,
      "learning_rate": 0.000907706366128541,
      "loss": 2.492046356201172,
      "step": 7960
    },
    {
      "epoch": 1.095489475289461,
      "grad_norm": 0.11079192906618118,
      "learning_rate": 0.000906487968321657,
      "loss": 2.521291494369507,
      "step": 7962
    },
    {
      "epoch": 1.0957646610972116,
      "grad_norm": 0.05191962048411369,
      "learning_rate": 0.0009052695705147731,
      "loss": 2.4922842979431152,
      "step": 7964
    },
    {
      "epoch": 1.0960398469049624,
      "grad_norm": 0.03646034747362137,
      "learning_rate": 0.0009040511727078891,
      "loss": 2.5606775283813477,
      "step": 7966
    },
    {
      "epoch": 1.096315032712713,
      "grad_norm": 0.19813863933086395,
      "learning_rate": 0.0009028327749010052,
      "loss": 2.4885940551757812,
      "step": 7968
    },
    {
      "epoch": 1.0965902185204635,
      "grad_norm": 0.14941556751728058,
      "learning_rate": 0.0009016143770941212,
      "loss": 2.510404586791992,
      "step": 7970
    },
    {
      "epoch": 1.0968654043282142,
      "grad_norm": 0.207256019115448,
      "learning_rate": 0.0009003959792872373,
      "loss": 2.5373244285583496,
      "step": 7972
    },
    {
      "epoch": 1.0971405901359648,
      "grad_norm": 0.1546003371477127,
      "learning_rate": 0.0008991775814803534,
      "loss": 2.46706485748291,
      "step": 7974
    },
    {
      "epoch": 1.0974157759437153,
      "grad_norm": 0.10693752020597458,
      "learning_rate": 0.0008979591836734694,
      "loss": 2.4959282875061035,
      "step": 7976
    },
    {
      "epoch": 1.0976909617514659,
      "grad_norm": 0.11567588895559311,
      "learning_rate": 0.0008967407858665855,
      "loss": 2.5319314002990723,
      "step": 7978
    },
    {
      "epoch": 1.0979661475592166,
      "grad_norm": 0.20719224214553833,
      "learning_rate": 0.0008955223880597015,
      "loss": 2.5106053352355957,
      "step": 7980
    },
    {
      "epoch": 1.0982413333669672,
      "grad_norm": 0.09368588775396347,
      "learning_rate": 0.0008943039902528176,
      "loss": 2.510941743850708,
      "step": 7982
    },
    {
      "epoch": 1.0985165191747177,
      "grad_norm": 0.08868434280157089,
      "learning_rate": 0.0008930855924459335,
      "loss": 2.4999136924743652,
      "step": 7984
    },
    {
      "epoch": 1.0987917049824683,
      "grad_norm": 0.13472546637058258,
      "learning_rate": 0.0008918671946390496,
      "loss": 2.5137691497802734,
      "step": 7986
    },
    {
      "epoch": 1.099066890790219,
      "grad_norm": 0.16968581080436707,
      "learning_rate": 0.0008906487968321657,
      "loss": 2.5066165924072266,
      "step": 7988
    },
    {
      "epoch": 1.0993420765979696,
      "grad_norm": 0.1595127433538437,
      "learning_rate": 0.0008894303990252818,
      "loss": 2.5118894577026367,
      "step": 7990
    },
    {
      "epoch": 1.09961726240572,
      "grad_norm": 0.1142064705491066,
      "learning_rate": 0.0008882120012183979,
      "loss": 2.4969706535339355,
      "step": 7992
    },
    {
      "epoch": 1.0998924482134709,
      "grad_norm": 0.11280138790607452,
      "learning_rate": 0.0008869936034115139,
      "loss": 2.48567533493042,
      "step": 7994
    },
    {
      "epoch": 1.1001676340212214,
      "grad_norm": 0.08252853900194168,
      "learning_rate": 0.00088577520560463,
      "loss": 2.5135467052459717,
      "step": 7996
    },
    {
      "epoch": 1.100442819828972,
      "grad_norm": 0.0979347676038742,
      "learning_rate": 0.0008845568077977459,
      "loss": 2.5136969089508057,
      "step": 7998
    },
    {
      "epoch": 1.1007180056367227,
      "grad_norm": 0.08176714181900024,
      "learning_rate": 0.000883338409990862,
      "loss": 2.5221965312957764,
      "step": 8000
    },
    {
      "epoch": 1.1009931914444733,
      "grad_norm": 0.0951780378818512,
      "learning_rate": 0.0008821200121839781,
      "loss": 2.5410332679748535,
      "step": 8002
    },
    {
      "epoch": 1.1012683772522238,
      "grad_norm": 0.10111148655414581,
      "learning_rate": 0.0008809016143770941,
      "loss": 2.486765146255493,
      "step": 8004
    },
    {
      "epoch": 1.1015435630599744,
      "grad_norm": 0.07750803977251053,
      "learning_rate": 0.0008796832165702103,
      "loss": 2.535219430923462,
      "step": 8006
    },
    {
      "epoch": 1.1018187488677251,
      "grad_norm": 0.0422697551548481,
      "learning_rate": 0.0008784648187633263,
      "loss": 2.554810047149658,
      "step": 8008
    },
    {
      "epoch": 1.1020939346754757,
      "grad_norm": 0.06252557039260864,
      "learning_rate": 0.0008772464209564423,
      "loss": 2.5456225872039795,
      "step": 8010
    },
    {
      "epoch": 1.1023691204832262,
      "grad_norm": 0.06441215425729752,
      "learning_rate": 0.0008760280231495583,
      "loss": 2.5403499603271484,
      "step": 8012
    },
    {
      "epoch": 1.102644306290977,
      "grad_norm": 0.039301469922065735,
      "learning_rate": 0.0008748096253426744,
      "loss": 2.5025291442871094,
      "step": 8014
    },
    {
      "epoch": 1.1029194920987275,
      "grad_norm": 0.03559224680066109,
      "learning_rate": 0.0008735912275357905,
      "loss": 2.492455005645752,
      "step": 8016
    },
    {
      "epoch": 1.103194677906478,
      "grad_norm": 0.049108851701021194,
      "learning_rate": 0.0008723728297289065,
      "loss": 2.4538230895996094,
      "step": 8018
    },
    {
      "epoch": 1.1034698637142286,
      "grad_norm": 0.028865614905953407,
      "learning_rate": 0.0008711544319220226,
      "loss": 2.5101499557495117,
      "step": 8020
    },
    {
      "epoch": 1.1037450495219794,
      "grad_norm": 0.02729044295847416,
      "learning_rate": 0.0008699360341151386,
      "loss": 2.477839946746826,
      "step": 8022
    },
    {
      "epoch": 1.10402023532973,
      "grad_norm": 0.026379376649856567,
      "learning_rate": 0.0008687176363082546,
      "loss": 2.51776123046875,
      "step": 8024
    },
    {
      "epoch": 1.1042954211374805,
      "grad_norm": 0.030416812747716904,
      "learning_rate": 0.0008674992385013707,
      "loss": 2.4904286861419678,
      "step": 8026
    },
    {
      "epoch": 1.1045706069452312,
      "grad_norm": 0.035191379487514496,
      "learning_rate": 0.0008662808406944868,
      "loss": 2.5181756019592285,
      "step": 8028
    },
    {
      "epoch": 1.1048457927529818,
      "grad_norm": 0.06433981657028198,
      "learning_rate": 0.0008650624428876028,
      "loss": 2.504537582397461,
      "step": 8030
    },
    {
      "epoch": 1.1051209785607323,
      "grad_norm": 0.03194567561149597,
      "learning_rate": 0.0008638440450807189,
      "loss": 2.515869140625,
      "step": 8032
    },
    {
      "epoch": 1.1053961643684829,
      "grad_norm": 0.027658242732286453,
      "learning_rate": 0.000862625647273835,
      "loss": 2.4814302921295166,
      "step": 8034
    },
    {
      "epoch": 1.1056713501762336,
      "grad_norm": 0.10651957243680954,
      "learning_rate": 0.0008614072494669509,
      "loss": 2.534369945526123,
      "step": 8036
    },
    {
      "epoch": 1.1059465359839842,
      "grad_norm": 0.105066679418087,
      "learning_rate": 0.000860188851660067,
      "loss": 2.5213093757629395,
      "step": 8038
    },
    {
      "epoch": 1.1062217217917347,
      "grad_norm": 0.025580372661352158,
      "learning_rate": 0.000858970453853183,
      "loss": 2.5287418365478516,
      "step": 8040
    },
    {
      "epoch": 1.1064969075994855,
      "grad_norm": 0.02788848616182804,
      "learning_rate": 0.0008577520560462991,
      "loss": 2.472411632537842,
      "step": 8042
    },
    {
      "epoch": 1.106772093407236,
      "grad_norm": 0.08894327282905579,
      "learning_rate": 0.0008565336582394152,
      "loss": 2.4592862129211426,
      "step": 8044
    },
    {
      "epoch": 1.1070472792149866,
      "grad_norm": 0.07027870416641235,
      "learning_rate": 0.0008553152604325313,
      "loss": 2.5064549446105957,
      "step": 8046
    },
    {
      "epoch": 1.1073224650227371,
      "grad_norm": 0.057376254349946976,
      "learning_rate": 0.0008540968626256474,
      "loss": 2.5144171714782715,
      "step": 8048
    },
    {
      "epoch": 1.107597650830488,
      "grad_norm": 0.03551328927278519,
      "learning_rate": 0.0008528784648187633,
      "loss": 2.5029871463775635,
      "step": 8050
    },
    {
      "epoch": 1.1078728366382384,
      "grad_norm": 0.03395308554172516,
      "learning_rate": 0.0008516600670118794,
      "loss": 2.457850456237793,
      "step": 8052
    },
    {
      "epoch": 1.108148022445989,
      "grad_norm": 0.040509141981601715,
      "learning_rate": 0.0008504416692049954,
      "loss": 2.5392866134643555,
      "step": 8054
    },
    {
      "epoch": 1.1084232082537397,
      "grad_norm": 0.06089415028691292,
      "learning_rate": 0.0008492232713981115,
      "loss": 2.508061170578003,
      "step": 8056
    },
    {
      "epoch": 1.1086983940614903,
      "grad_norm": 0.06605853140354156,
      "learning_rate": 0.0008480048735912275,
      "loss": 2.492952346801758,
      "step": 8058
    },
    {
      "epoch": 1.1089735798692408,
      "grad_norm": 0.02792150154709816,
      "learning_rate": 0.0008467864757843437,
      "loss": 2.5332298278808594,
      "step": 8060
    },
    {
      "epoch": 1.1092487656769916,
      "grad_norm": 0.06177910789847374,
      "learning_rate": 0.0008455680779774596,
      "loss": 2.493088722229004,
      "step": 8062
    },
    {
      "epoch": 1.1095239514847421,
      "grad_norm": 0.02801436185836792,
      "learning_rate": 0.0008443496801705757,
      "loss": 2.518051862716675,
      "step": 8064
    },
    {
      "epoch": 1.1097991372924927,
      "grad_norm": 0.02798466570675373,
      "learning_rate": 0.0008431312823636918,
      "loss": 2.4874768257141113,
      "step": 8066
    },
    {
      "epoch": 1.1100743231002432,
      "grad_norm": 0.03595089912414551,
      "learning_rate": 0.0008419128845568078,
      "loss": 2.5114223957061768,
      "step": 8068
    },
    {
      "epoch": 1.110349508907994,
      "grad_norm": 0.02873462438583374,
      "learning_rate": 0.0008406944867499239,
      "loss": 2.4973549842834473,
      "step": 8070
    },
    {
      "epoch": 1.1106246947157445,
      "grad_norm": 0.05736279487609863,
      "learning_rate": 0.0008394760889430399,
      "loss": 2.455564498901367,
      "step": 8072
    },
    {
      "epoch": 1.110899880523495,
      "grad_norm": 0.06297782063484192,
      "learning_rate": 0.000838257691136156,
      "loss": 2.5204877853393555,
      "step": 8074
    },
    {
      "epoch": 1.1111750663312456,
      "grad_norm": 0.11854040622711182,
      "learning_rate": 0.0008370392933292719,
      "loss": 2.5210869312286377,
      "step": 8076
    },
    {
      "epoch": 1.1114502521389964,
      "grad_norm": 0.0771612748503685,
      "learning_rate": 0.000835820895522388,
      "loss": 2.5159192085266113,
      "step": 8078
    },
    {
      "epoch": 1.111725437946747,
      "grad_norm": 0.02993665635585785,
      "learning_rate": 0.0008346024977155042,
      "loss": 2.4899306297302246,
      "step": 8080
    },
    {
      "epoch": 1.1120006237544975,
      "grad_norm": 0.030085550621151924,
      "learning_rate": 0.0008333840999086202,
      "loss": 2.5461416244506836,
      "step": 8082
    },
    {
      "epoch": 1.1122758095622483,
      "grad_norm": 0.0894346833229065,
      "learning_rate": 0.0008321657021017363,
      "loss": 2.5017104148864746,
      "step": 8084
    },
    {
      "epoch": 1.1125509953699988,
      "grad_norm": 0.0807172954082489,
      "learning_rate": 0.0008309473042948523,
      "loss": 2.4782395362854004,
      "step": 8086
    },
    {
      "epoch": 1.1128261811777493,
      "grad_norm": 0.08307875692844391,
      "learning_rate": 0.0008297289064879684,
      "loss": 2.511974334716797,
      "step": 8088
    },
    {
      "epoch": 1.1131013669855,
      "grad_norm": 0.03435284644365311,
      "learning_rate": 0.0008285105086810843,
      "loss": 2.491790294647217,
      "step": 8090
    },
    {
      "epoch": 1.1133765527932507,
      "grad_norm": 0.03048594295978546,
      "learning_rate": 0.0008272921108742004,
      "loss": 2.4856910705566406,
      "step": 8092
    },
    {
      "epoch": 1.1136517386010012,
      "grad_norm": 0.032086003571748734,
      "learning_rate": 0.0008260737130673165,
      "loss": 2.515960216522217,
      "step": 8094
    },
    {
      "epoch": 1.1139269244087517,
      "grad_norm": 0.03864971548318863,
      "learning_rate": 0.0008248553152604325,
      "loss": 2.4681692123413086,
      "step": 8096
    },
    {
      "epoch": 1.1142021102165025,
      "grad_norm": 0.05428634583950043,
      "learning_rate": 0.0008236369174535487,
      "loss": 2.4831433296203613,
      "step": 8098
    },
    {
      "epoch": 1.114477296024253,
      "grad_norm": 0.027756793424487114,
      "learning_rate": 0.0008224185196466647,
      "loss": 2.5006155967712402,
      "step": 8100
    },
    {
      "epoch": 1.1147524818320036,
      "grad_norm": 0.035419151186943054,
      "learning_rate": 0.0008212001218397807,
      "loss": 2.523946523666382,
      "step": 8102
    },
    {
      "epoch": 1.1150276676397544,
      "grad_norm": 0.0684950202703476,
      "learning_rate": 0.0008199817240328967,
      "loss": 2.4850754737854004,
      "step": 8104
    },
    {
      "epoch": 1.115302853447505,
      "grad_norm": 0.080427385866642,
      "learning_rate": 0.0008187633262260128,
      "loss": 2.4624204635620117,
      "step": 8106
    },
    {
      "epoch": 1.1155780392552554,
      "grad_norm": 0.031792052090168,
      "learning_rate": 0.0008175449284191288,
      "loss": 2.4849801063537598,
      "step": 8108
    },
    {
      "epoch": 1.115853225063006,
      "grad_norm": 0.05647988244891167,
      "learning_rate": 0.0008163265306122449,
      "loss": 2.470083236694336,
      "step": 8110
    },
    {
      "epoch": 1.1161284108707568,
      "grad_norm": 0.02750456891953945,
      "learning_rate": 0.000815108132805361,
      "loss": 2.5314221382141113,
      "step": 8112
    },
    {
      "epoch": 1.1164035966785073,
      "grad_norm": 0.028133001178503036,
      "learning_rate": 0.000813889734998477,
      "loss": 2.4613449573516846,
      "step": 8114
    },
    {
      "epoch": 1.1166787824862578,
      "grad_norm": 0.04105991870164871,
      "learning_rate": 0.0008126713371915931,
      "loss": 2.4803004264831543,
      "step": 8116
    },
    {
      "epoch": 1.1169539682940086,
      "grad_norm": 0.03123358264565468,
      "learning_rate": 0.0008114529393847091,
      "loss": 2.4884729385375977,
      "step": 8118
    },
    {
      "epoch": 1.1172291541017592,
      "grad_norm": 0.05271749198436737,
      "learning_rate": 0.0008102345415778252,
      "loss": 2.505048990249634,
      "step": 8120
    },
    {
      "epoch": 1.1175043399095097,
      "grad_norm": 0.07851790636777878,
      "learning_rate": 0.0008090161437709412,
      "loss": 2.5329036712646484,
      "step": 8122
    },
    {
      "epoch": 1.1177795257172602,
      "grad_norm": 0.06988466531038284,
      "learning_rate": 0.0008077977459640573,
      "loss": 2.4547786712646484,
      "step": 8124
    },
    {
      "epoch": 1.118054711525011,
      "grad_norm": 0.03436512127518654,
      "learning_rate": 0.0008065793481571734,
      "loss": 2.4998316764831543,
      "step": 8126
    },
    {
      "epoch": 1.1183298973327616,
      "grad_norm": 0.06801183521747589,
      "learning_rate": 0.0008053609503502893,
      "loss": 2.4947516918182373,
      "step": 8128
    },
    {
      "epoch": 1.118605083140512,
      "grad_norm": 0.10921048372983932,
      "learning_rate": 0.0008041425525434054,
      "loss": 2.4949162006378174,
      "step": 8130
    },
    {
      "epoch": 1.1188802689482629,
      "grad_norm": 0.05695357173681259,
      "learning_rate": 0.0008029241547365214,
      "loss": 2.474834442138672,
      "step": 8132
    },
    {
      "epoch": 1.1191554547560134,
      "grad_norm": 0.032397642731666565,
      "learning_rate": 0.0008017057569296376,
      "loss": 2.4987034797668457,
      "step": 8134
    },
    {
      "epoch": 1.119430640563764,
      "grad_norm": 0.052857741713523865,
      "learning_rate": 0.0008004873591227536,
      "loss": 2.477116107940674,
      "step": 8136
    },
    {
      "epoch": 1.1197058263715145,
      "grad_norm": 0.0675104409456253,
      "learning_rate": 0.0007992689613158697,
      "loss": 2.454792022705078,
      "step": 8138
    },
    {
      "epoch": 1.1199810121792653,
      "grad_norm": 0.06749003380537033,
      "learning_rate": 0.0007980505635089858,
      "loss": 2.512417793273926,
      "step": 8140
    },
    {
      "epoch": 1.1202561979870158,
      "grad_norm": 0.04485924914479256,
      "learning_rate": 0.0007968321657021017,
      "loss": 2.5245180130004883,
      "step": 8142
    },
    {
      "epoch": 1.1205313837947664,
      "grad_norm": 0.027643881738185883,
      "learning_rate": 0.0007956137678952178,
      "loss": 2.508788585662842,
      "step": 8144
    },
    {
      "epoch": 1.1208065696025171,
      "grad_norm": 0.06234624609351158,
      "learning_rate": 0.0007943953700883338,
      "loss": 2.4899189472198486,
      "step": 8146
    },
    {
      "epoch": 1.1210817554102677,
      "grad_norm": 0.04552260786294937,
      "learning_rate": 0.0007931769722814499,
      "loss": 2.4509782791137695,
      "step": 8148
    },
    {
      "epoch": 1.1213569412180182,
      "grad_norm": 0.09311044961214066,
      "learning_rate": 0.000791958574474566,
      "loss": 2.509885311126709,
      "step": 8150
    },
    {
      "epoch": 1.1216321270257688,
      "grad_norm": 0.05285623297095299,
      "learning_rate": 0.0007907401766676821,
      "loss": 2.509187698364258,
      "step": 8152
    },
    {
      "epoch": 1.1219073128335195,
      "grad_norm": 0.07026425004005432,
      "learning_rate": 0.000789521778860798,
      "loss": 2.4754271507263184,
      "step": 8154
    },
    {
      "epoch": 1.12218249864127,
      "grad_norm": 0.1001950278878212,
      "learning_rate": 0.0007883033810539141,
      "loss": 2.467043399810791,
      "step": 8156
    },
    {
      "epoch": 1.1224576844490206,
      "grad_norm": 0.054363913834095,
      "learning_rate": 0.0007870849832470302,
      "loss": 2.4755337238311768,
      "step": 8158
    },
    {
      "epoch": 1.1227328702567714,
      "grad_norm": 0.10551432520151138,
      "learning_rate": 0.0007858665854401462,
      "loss": 2.485689640045166,
      "step": 8160
    },
    {
      "epoch": 1.123008056064522,
      "grad_norm": 0.1108849048614502,
      "learning_rate": 0.0007846481876332623,
      "loss": 2.4621834754943848,
      "step": 8162
    },
    {
      "epoch": 1.1232832418722725,
      "grad_norm": 0.043662045150995255,
      "learning_rate": 0.0007834297898263783,
      "loss": 2.4796156883239746,
      "step": 8164
    },
    {
      "epoch": 1.123558427680023,
      "grad_norm": 0.030653811991214752,
      "learning_rate": 0.0007822113920194944,
      "loss": 2.5440704822540283,
      "step": 8166
    },
    {
      "epoch": 1.1238336134877738,
      "grad_norm": 0.03429726883769035,
      "learning_rate": 0.0007809929942126103,
      "loss": 2.548555850982666,
      "step": 8168
    },
    {
      "epoch": 1.1241087992955243,
      "grad_norm": 0.052573107182979584,
      "learning_rate": 0.0007797745964057265,
      "loss": 2.5408506393432617,
      "step": 8170
    },
    {
      "epoch": 1.1243839851032749,
      "grad_norm": 0.03100688010454178,
      "learning_rate": 0.0007785561985988426,
      "loss": 2.520521640777588,
      "step": 8172
    },
    {
      "epoch": 1.1246591709110256,
      "grad_norm": 0.02650621347129345,
      "learning_rate": 0.0007773378007919586,
      "loss": 2.487828493118286,
      "step": 8174
    },
    {
      "epoch": 1.1249343567187762,
      "grad_norm": 0.03089853748679161,
      "learning_rate": 0.0007761194029850747,
      "loss": 2.4745655059814453,
      "step": 8176
    },
    {
      "epoch": 1.1252095425265267,
      "grad_norm": 0.035853419452905655,
      "learning_rate": 0.0007749010051781907,
      "loss": 2.486945152282715,
      "step": 8178
    },
    {
      "epoch": 1.1254847283342775,
      "grad_norm": 0.03177864849567413,
      "learning_rate": 0.0007736826073713068,
      "loss": 2.481142282485962,
      "step": 8180
    },
    {
      "epoch": 1.125759914142028,
      "grad_norm": 0.04017817601561546,
      "learning_rate": 0.0007724642095644227,
      "loss": 2.5350351333618164,
      "step": 8182
    },
    {
      "epoch": 1.1260350999497786,
      "grad_norm": 0.07513938844203949,
      "learning_rate": 0.0007712458117575388,
      "loss": 2.4916200637817383,
      "step": 8184
    },
    {
      "epoch": 1.1263102857575291,
      "grad_norm": 0.059088945388793945,
      "learning_rate": 0.0007700274139506548,
      "loss": 2.498750686645508,
      "step": 8186
    },
    {
      "epoch": 1.1265854715652799,
      "grad_norm": 0.0549178384244442,
      "learning_rate": 0.000768809016143771,
      "loss": 2.491720676422119,
      "step": 8188
    },
    {
      "epoch": 1.1268606573730304,
      "grad_norm": 0.031138254329562187,
      "learning_rate": 0.0007675906183368871,
      "loss": 2.453031539916992,
      "step": 8190
    },
    {
      "epoch": 1.127135843180781,
      "grad_norm": 0.047318343073129654,
      "learning_rate": 0.0007663722205300031,
      "loss": 2.470125675201416,
      "step": 8192
    },
    {
      "epoch": 1.1274110289885315,
      "grad_norm": 0.026889173313975334,
      "learning_rate": 0.0007651538227231191,
      "loss": 2.512949228286743,
      "step": 8194
    },
    {
      "epoch": 1.1276862147962823,
      "grad_norm": 0.07464047521352768,
      "learning_rate": 0.0007639354249162351,
      "loss": 2.4899415969848633,
      "step": 8196
    },
    {
      "epoch": 1.1279614006040328,
      "grad_norm": 0.07030083239078522,
      "learning_rate": 0.0007627170271093512,
      "loss": 2.4590301513671875,
      "step": 8198
    },
    {
      "epoch": 1.1282365864117834,
      "grad_norm": 0.06151893734931946,
      "learning_rate": 0.0007614986293024672,
      "loss": 2.5165305137634277,
      "step": 8200
    },
    {
      "epoch": 1.1285117722195341,
      "grad_norm": 0.04853873327374458,
      "learning_rate": 0.0007602802314955833,
      "loss": 2.471571922302246,
      "step": 8202
    },
    {
      "epoch": 1.1287869580272847,
      "grad_norm": 0.10920491069555283,
      "learning_rate": 0.0007590618336886995,
      "loss": 2.4728779792785645,
      "step": 8204
    },
    {
      "epoch": 1.1290621438350352,
      "grad_norm": 0.054910484701395035,
      "learning_rate": 0.0007578434358818155,
      "loss": 2.480358839035034,
      "step": 8206
    },
    {
      "epoch": 1.129337329642786,
      "grad_norm": 0.094294473528862,
      "learning_rate": 0.0007566250380749315,
      "loss": 2.5130088329315186,
      "step": 8208
    },
    {
      "epoch": 1.1296125154505365,
      "grad_norm": 0.0647880807518959,
      "learning_rate": 0.0007554066402680475,
      "loss": 2.49481201171875,
      "step": 8210
    },
    {
      "epoch": 1.129887701258287,
      "grad_norm": 0.029621142894029617,
      "learning_rate": 0.0007541882424611636,
      "loss": 2.505192756652832,
      "step": 8212
    },
    {
      "epoch": 1.1301628870660376,
      "grad_norm": 0.02781938575208187,
      "learning_rate": 0.0007529698446542796,
      "loss": 2.523852825164795,
      "step": 8214
    },
    {
      "epoch": 1.1304380728737884,
      "grad_norm": 0.058949828147888184,
      "learning_rate": 0.0007517514468473957,
      "loss": 2.5338096618652344,
      "step": 8216
    },
    {
      "epoch": 1.130713258681539,
      "grad_norm": 0.04405047744512558,
      "learning_rate": 0.0007505330490405118,
      "loss": 2.473074197769165,
      "step": 8218
    },
    {
      "epoch": 1.1309884444892895,
      "grad_norm": 0.03366885334253311,
      "learning_rate": 0.0007493146512336277,
      "loss": 2.460676908493042,
      "step": 8220
    },
    {
      "epoch": 1.13126363029704,
      "grad_norm": 0.035217106342315674,
      "learning_rate": 0.0007480962534267439,
      "loss": 2.5050177574157715,
      "step": 8222
    },
    {
      "epoch": 1.1315388161047908,
      "grad_norm": 0.03094514273107052,
      "learning_rate": 0.0007468778556198599,
      "loss": 2.5082902908325195,
      "step": 8224
    },
    {
      "epoch": 1.1318140019125413,
      "grad_norm": 0.032832954078912735,
      "learning_rate": 0.000745659457812976,
      "loss": 2.4461588859558105,
      "step": 8226
    },
    {
      "epoch": 1.1320891877202919,
      "grad_norm": 0.038462333381175995,
      "learning_rate": 0.000744441060006092,
      "loss": 2.529290199279785,
      "step": 8228
    },
    {
      "epoch": 1.1323643735280426,
      "grad_norm": 0.04357236996293068,
      "learning_rate": 0.0007432226621992081,
      "loss": 2.5270419120788574,
      "step": 8230
    },
    {
      "epoch": 1.1326395593357932,
      "grad_norm": 0.052701227366924286,
      "learning_rate": 0.0007420042643923241,
      "loss": 2.5076708793640137,
      "step": 8232
    },
    {
      "epoch": 1.1329147451435437,
      "grad_norm": 0.10910900682210922,
      "learning_rate": 0.0007407858665854401,
      "loss": 2.4836714267730713,
      "step": 8234
    },
    {
      "epoch": 1.1331899309512945,
      "grad_norm": 0.046847615391016006,
      "learning_rate": 0.0007395674687785562,
      "loss": 2.4809250831604004,
      "step": 8236
    },
    {
      "epoch": 1.133465116759045,
      "grad_norm": 0.0328650176525116,
      "learning_rate": 0.0007383490709716722,
      "loss": 2.492110252380371,
      "step": 8238
    },
    {
      "epoch": 1.1337403025667956,
      "grad_norm": 0.04414726793766022,
      "learning_rate": 0.0007371306731647884,
      "loss": 2.465463638305664,
      "step": 8240
    },
    {
      "epoch": 1.1340154883745464,
      "grad_norm": 0.08822086453437805,
      "learning_rate": 0.0007359122753579044,
      "loss": 2.5344767570495605,
      "step": 8242
    },
    {
      "epoch": 1.134290674182297,
      "grad_norm": 0.04081292077898979,
      "learning_rate": 0.0007346938775510205,
      "loss": 2.5025687217712402,
      "step": 8244
    },
    {
      "epoch": 1.1345658599900474,
      "grad_norm": 0.029918299987912178,
      "learning_rate": 0.0007334754797441365,
      "loss": 2.496809959411621,
      "step": 8246
    },
    {
      "epoch": 1.134841045797798,
      "grad_norm": 0.029106931760907173,
      "learning_rate": 0.0007322570819372525,
      "loss": 2.4918127059936523,
      "step": 8248
    },
    {
      "epoch": 1.1351162316055488,
      "grad_norm": 0.046049460768699646,
      "learning_rate": 0.0007310386841303686,
      "loss": 2.4991350173950195,
      "step": 8250
    },
    {
      "epoch": 1.1353914174132993,
      "grad_norm": 0.028959576040506363,
      "learning_rate": 0.0007298202863234846,
      "loss": 2.474726676940918,
      "step": 8252
    },
    {
      "epoch": 1.1356666032210498,
      "grad_norm": 0.03503545746207237,
      "learning_rate": 0.0007286018885166007,
      "loss": 2.4735524654388428,
      "step": 8254
    },
    {
      "epoch": 1.1359417890288004,
      "grad_norm": 0.11716299504041672,
      "learning_rate": 0.0007273834907097167,
      "loss": 2.4778270721435547,
      "step": 8256
    },
    {
      "epoch": 1.1362169748365512,
      "grad_norm": 0.08702637255191803,
      "learning_rate": 0.0007261650929028329,
      "loss": 2.537877082824707,
      "step": 8258
    },
    {
      "epoch": 1.1364921606443017,
      "grad_norm": 0.06464211642742157,
      "learning_rate": 0.0007249466950959488,
      "loss": 2.5124645233154297,
      "step": 8260
    },
    {
      "epoch": 1.1367673464520522,
      "grad_norm": 0.04695899039506912,
      "learning_rate": 0.0007237282972890649,
      "loss": 2.487623453140259,
      "step": 8262
    },
    {
      "epoch": 1.137042532259803,
      "grad_norm": 0.05112483724951744,
      "learning_rate": 0.0007225098994821809,
      "loss": 2.498110771179199,
      "step": 8264
    },
    {
      "epoch": 1.1373177180675536,
      "grad_norm": 0.027894550934433937,
      "learning_rate": 0.000721291501675297,
      "loss": 2.4596383571624756,
      "step": 8266
    },
    {
      "epoch": 1.137592903875304,
      "grad_norm": 0.05257590487599373,
      "learning_rate": 0.0007200731038684131,
      "loss": 2.494110584259033,
      "step": 8268
    },
    {
      "epoch": 1.1378680896830549,
      "grad_norm": 0.04530532658100128,
      "learning_rate": 0.0007188547060615291,
      "loss": 2.4836878776550293,
      "step": 8270
    },
    {
      "epoch": 1.1381432754908054,
      "grad_norm": 0.02969382517039776,
      "learning_rate": 0.0007176363082546452,
      "loss": 2.4822118282318115,
      "step": 8272
    },
    {
      "epoch": 1.138418461298556,
      "grad_norm": 0.047046124935150146,
      "learning_rate": 0.0007164179104477611,
      "loss": 2.500729560852051,
      "step": 8274
    },
    {
      "epoch": 1.1386936471063065,
      "grad_norm": 0.027907751500606537,
      "learning_rate": 0.0007151995126408773,
      "loss": 2.4788801670074463,
      "step": 8276
    },
    {
      "epoch": 1.1389688329140573,
      "grad_norm": 0.04255083575844765,
      "learning_rate": 0.0007139811148339933,
      "loss": 2.4846506118774414,
      "step": 8278
    },
    {
      "epoch": 1.1392440187218078,
      "grad_norm": 0.07538124918937683,
      "learning_rate": 0.0007127627170271094,
      "loss": 2.4828875064849854,
      "step": 8280
    },
    {
      "epoch": 1.1395192045295583,
      "grad_norm": 0.05636218190193176,
      "learning_rate": 0.0007115443192202255,
      "loss": 2.4374053478240967,
      "step": 8282
    },
    {
      "epoch": 1.139794390337309,
      "grad_norm": 0.06932029128074646,
      "learning_rate": 0.0007103259214133415,
      "loss": 2.481215000152588,
      "step": 8284
    },
    {
      "epoch": 1.1400695761450597,
      "grad_norm": 0.05284881219267845,
      "learning_rate": 0.0007091075236064575,
      "loss": 2.4882709980010986,
      "step": 8286
    },
    {
      "epoch": 1.1403447619528102,
      "grad_norm": 0.05426977574825287,
      "learning_rate": 0.0007078891257995735,
      "loss": 2.4817116260528564,
      "step": 8288
    },
    {
      "epoch": 1.1406199477605607,
      "grad_norm": 0.026990100741386414,
      "learning_rate": 0.0007066707279926896,
      "loss": 2.493539333343506,
      "step": 8290
    },
    {
      "epoch": 1.1408951335683115,
      "grad_norm": 0.03242146223783493,
      "learning_rate": 0.0007054523301858056,
      "loss": 2.452486515045166,
      "step": 8292
    },
    {
      "epoch": 1.141170319376062,
      "grad_norm": 0.06333518773317337,
      "learning_rate": 0.0007042339323789218,
      "loss": 2.519176959991455,
      "step": 8294
    },
    {
      "epoch": 1.1414455051838126,
      "grad_norm": 0.05462523177266121,
      "learning_rate": 0.0007030155345720379,
      "loss": 2.4799816608428955,
      "step": 8296
    },
    {
      "epoch": 1.1417206909915634,
      "grad_norm": 0.031124169006943703,
      "learning_rate": 0.0007017971367651539,
      "loss": 2.463773727416992,
      "step": 8298
    },
    {
      "epoch": 1.141995876799314,
      "grad_norm": 0.033469852060079575,
      "learning_rate": 0.0007005787389582699,
      "loss": 2.521340847015381,
      "step": 8300
    },
    {
      "epoch": 1.1422710626070645,
      "grad_norm": 0.0479627400636673,
      "learning_rate": 0.0006993603411513859,
      "loss": 2.50653076171875,
      "step": 8302
    },
    {
      "epoch": 1.142546248414815,
      "grad_norm": 0.05013016238808632,
      "learning_rate": 0.000698141943344502,
      "loss": 2.451716184616089,
      "step": 8304
    },
    {
      "epoch": 1.1428214342225658,
      "grad_norm": 0.06460792571306229,
      "learning_rate": 0.000696923545537618,
      "loss": 2.524547576904297,
      "step": 8306
    },
    {
      "epoch": 1.1430966200303163,
      "grad_norm": 0.05582503601908684,
      "learning_rate": 0.0006957051477307341,
      "loss": 2.521920919418335,
      "step": 8308
    },
    {
      "epoch": 1.1433718058380669,
      "grad_norm": 0.048682160675525665,
      "learning_rate": 0.0006944867499238501,
      "loss": 2.4811789989471436,
      "step": 8310
    },
    {
      "epoch": 1.1436469916458174,
      "grad_norm": 0.08005623519420624,
      "learning_rate": 0.0006932683521169661,
      "loss": 2.4617390632629395,
      "step": 8312
    },
    {
      "epoch": 1.1439221774535682,
      "grad_norm": 0.03682665899395943,
      "learning_rate": 0.0006920499543100823,
      "loss": 2.4946703910827637,
      "step": 8314
    },
    {
      "epoch": 1.1441973632613187,
      "grad_norm": 0.02654961682856083,
      "learning_rate": 0.0006908315565031983,
      "loss": 2.4604194164276123,
      "step": 8316
    },
    {
      "epoch": 1.1444725490690693,
      "grad_norm": 0.03845710679888725,
      "learning_rate": 0.0006896131586963144,
      "loss": 2.4772047996520996,
      "step": 8318
    },
    {
      "epoch": 1.14474773487682,
      "grad_norm": 0.04822942242026329,
      "learning_rate": 0.0006883947608894304,
      "loss": 2.483963966369629,
      "step": 8320
    },
    {
      "epoch": 1.1450229206845706,
      "grad_norm": 0.0591675229370594,
      "learning_rate": 0.0006871763630825465,
      "loss": 2.434091329574585,
      "step": 8322
    },
    {
      "epoch": 1.145298106492321,
      "grad_norm": 0.08893642574548721,
      "learning_rate": 0.0006859579652756625,
      "loss": 2.4822421073913574,
      "step": 8324
    },
    {
      "epoch": 1.1455732923000719,
      "grad_norm": 0.10550463944673538,
      "learning_rate": 0.0006847395674687785,
      "loss": 2.493900775909424,
      "step": 8326
    },
    {
      "epoch": 1.1458484781078224,
      "grad_norm": 0.08169044554233551,
      "learning_rate": 0.0006835211696618946,
      "loss": 2.475841999053955,
      "step": 8328
    },
    {
      "epoch": 1.146123663915573,
      "grad_norm": 0.05678975209593773,
      "learning_rate": 0.0006823027718550107,
      "loss": 2.48543643951416,
      "step": 8330
    },
    {
      "epoch": 1.1463988497233235,
      "grad_norm": 0.07580310851335526,
      "learning_rate": 0.0006810843740481268,
      "loss": 2.500166416168213,
      "step": 8332
    },
    {
      "epoch": 1.1466740355310743,
      "grad_norm": 0.0513673797249794,
      "learning_rate": 0.0006798659762412428,
      "loss": 2.4727134704589844,
      "step": 8334
    },
    {
      "epoch": 1.1469492213388248,
      "grad_norm": 0.06574403494596481,
      "learning_rate": 0.0006786475784343589,
      "loss": 2.455137252807617,
      "step": 8336
    },
    {
      "epoch": 1.1472244071465754,
      "grad_norm": 0.0827818512916565,
      "learning_rate": 0.0006774291806274749,
      "loss": 2.4350414276123047,
      "step": 8338
    },
    {
      "epoch": 1.1474995929543261,
      "grad_norm": 0.05623174458742142,
      "learning_rate": 0.0006762107828205909,
      "loss": 2.4972567558288574,
      "step": 8340
    },
    {
      "epoch": 1.1477747787620767,
      "grad_norm": 0.04753813147544861,
      "learning_rate": 0.0006749923850137069,
      "loss": 2.543185234069824,
      "step": 8342
    },
    {
      "epoch": 1.1480499645698272,
      "grad_norm": 0.02679951675236225,
      "learning_rate": 0.000673773987206823,
      "loss": 2.486830711364746,
      "step": 8344
    },
    {
      "epoch": 1.1483251503775778,
      "grad_norm": 0.03292723745107651,
      "learning_rate": 0.0006725555893999391,
      "loss": 2.4960556030273438,
      "step": 8346
    },
    {
      "epoch": 1.1486003361853285,
      "grad_norm": 0.025663265958428383,
      "learning_rate": 0.0006713371915930552,
      "loss": 2.4608139991760254,
      "step": 8348
    },
    {
      "epoch": 1.148875521993079,
      "grad_norm": 0.06797395646572113,
      "learning_rate": 0.0006701187937861713,
      "loss": 2.4937620162963867,
      "step": 8350
    },
    {
      "epoch": 1.1491507078008296,
      "grad_norm": 0.056331880390644073,
      "learning_rate": 0.0006689003959792872,
      "loss": 2.502739906311035,
      "step": 8352
    },
    {
      "epoch": 1.1494258936085804,
      "grad_norm": 0.05028936639428139,
      "learning_rate": 0.0006676819981724033,
      "loss": 2.5279712677001953,
      "step": 8354
    },
    {
      "epoch": 1.149701079416331,
      "grad_norm": 0.025509636849164963,
      "learning_rate": 0.0006664636003655193,
      "loss": 2.5007221698760986,
      "step": 8356
    },
    {
      "epoch": 1.1499762652240815,
      "grad_norm": 0.037656884640455246,
      "learning_rate": 0.0006652452025586354,
      "loss": 2.4417781829833984,
      "step": 8358
    },
    {
      "epoch": 1.1502514510318322,
      "grad_norm": 0.07605823129415512,
      "learning_rate": 0.0006640268047517515,
      "loss": 2.4731063842773438,
      "step": 8360
    },
    {
      "epoch": 1.1505266368395828,
      "grad_norm": 0.050488315522670746,
      "learning_rate": 0.0006628084069448675,
      "loss": 2.492469310760498,
      "step": 8362
    },
    {
      "epoch": 1.1508018226473333,
      "grad_norm": 0.029347190633416176,
      "learning_rate": 0.0006615900091379837,
      "loss": 2.5048346519470215,
      "step": 8364
    },
    {
      "epoch": 1.1510770084550839,
      "grad_norm": 0.040915969759225845,
      "learning_rate": 0.0006603716113310995,
      "loss": 2.4528021812438965,
      "step": 8366
    },
    {
      "epoch": 1.1513521942628346,
      "grad_norm": 0.02759440802037716,
      "learning_rate": 0.0006591532135242157,
      "loss": 2.468259334564209,
      "step": 8368
    },
    {
      "epoch": 1.1516273800705852,
      "grad_norm": 0.05474226549267769,
      "learning_rate": 0.0006579348157173317,
      "loss": 2.4972081184387207,
      "step": 8370
    },
    {
      "epoch": 1.1519025658783357,
      "grad_norm": 0.04437842592597008,
      "learning_rate": 0.0006567164179104478,
      "loss": 2.5022640228271484,
      "step": 8372
    },
    {
      "epoch": 1.1521777516860863,
      "grad_norm": 0.0414668433368206,
      "learning_rate": 0.0006554980201035639,
      "loss": 2.4649951457977295,
      "step": 8374
    },
    {
      "epoch": 1.152452937493837,
      "grad_norm": 0.03041396476328373,
      "learning_rate": 0.0006542796222966799,
      "loss": 2.4718120098114014,
      "step": 8376
    },
    {
      "epoch": 1.1527281233015876,
      "grad_norm": 0.04704215005040169,
      "learning_rate": 0.0006530612244897959,
      "loss": 2.4727742671966553,
      "step": 8378
    },
    {
      "epoch": 1.1530033091093381,
      "grad_norm": 0.027179118245840073,
      "learning_rate": 0.0006518428266829119,
      "loss": 2.461089611053467,
      "step": 8380
    },
    {
      "epoch": 1.153278494917089,
      "grad_norm": 0.06935402005910873,
      "learning_rate": 0.000650624428876028,
      "loss": 2.5046067237854004,
      "step": 8382
    },
    {
      "epoch": 1.1535536807248394,
      "grad_norm": 0.04518992453813553,
      "learning_rate": 0.000649406031069144,
      "loss": 2.4762587547302246,
      "step": 8384
    },
    {
      "epoch": 1.15382886653259,
      "grad_norm": 0.052437350153923035,
      "learning_rate": 0.0006481876332622602,
      "loss": 2.480316638946533,
      "step": 8386
    },
    {
      "epoch": 1.1541040523403407,
      "grad_norm": 0.030001549050211906,
      "learning_rate": 0.0006469692354553762,
      "loss": 2.4818224906921387,
      "step": 8388
    },
    {
      "epoch": 1.1543792381480913,
      "grad_norm": 0.05656140670180321,
      "learning_rate": 0.0006457508376484923,
      "loss": 2.528728485107422,
      "step": 8390
    },
    {
      "epoch": 1.1546544239558418,
      "grad_norm": 0.04058216139674187,
      "learning_rate": 0.0006445324398416083,
      "loss": 2.4803194999694824,
      "step": 8392
    },
    {
      "epoch": 1.1549296097635924,
      "grad_norm": 0.04114169627428055,
      "learning_rate": 0.0006433140420347243,
      "loss": 2.48237943649292,
      "step": 8394
    },
    {
      "epoch": 1.1552047955713431,
      "grad_norm": 0.038148943334817886,
      "learning_rate": 0.0006420956442278404,
      "loss": 2.5001273155212402,
      "step": 8396
    },
    {
      "epoch": 1.1554799813790937,
      "grad_norm": 0.055114440619945526,
      "learning_rate": 0.0006408772464209564,
      "loss": 2.4728264808654785,
      "step": 8398
    },
    {
      "epoch": 1.1557551671868442,
      "grad_norm": 0.03029511868953705,
      "learning_rate": 0.0006396588486140725,
      "loss": 2.444441318511963,
      "step": 8400
    },
    {
      "epoch": 1.1560303529945948,
      "grad_norm": 0.030824868008494377,
      "learning_rate": 0.0006384404508071886,
      "loss": 2.452749252319336,
      "step": 8402
    },
    {
      "epoch": 1.1563055388023455,
      "grad_norm": 0.04031574726104736,
      "learning_rate": 0.0006372220530003046,
      "loss": 2.4891574382781982,
      "step": 8404
    },
    {
      "epoch": 1.156580724610096,
      "grad_norm": 0.0839531272649765,
      "learning_rate": 0.0006360036551934207,
      "loss": 2.495685577392578,
      "step": 8406
    },
    {
      "epoch": 1.1568559104178466,
      "grad_norm": 0.07072802633047104,
      "learning_rate": 0.0006347852573865367,
      "loss": 2.446265697479248,
      "step": 8408
    },
    {
      "epoch": 1.1571310962255974,
      "grad_norm": 0.08657628297805786,
      "learning_rate": 0.0006335668595796528,
      "loss": 2.4500489234924316,
      "step": 8410
    },
    {
      "epoch": 1.157406282033348,
      "grad_norm": 0.06394266337156296,
      "learning_rate": 0.0006323484617727688,
      "loss": 2.5003409385681152,
      "step": 8412
    },
    {
      "epoch": 1.1576814678410985,
      "grad_norm": 0.05390682816505432,
      "learning_rate": 0.0006311300639658849,
      "loss": 2.445234775543213,
      "step": 8414
    },
    {
      "epoch": 1.1579566536488493,
      "grad_norm": 0.0663897842168808,
      "learning_rate": 0.0006299116661590009,
      "loss": 2.469597101211548,
      "step": 8416
    },
    {
      "epoch": 1.1582318394565998,
      "grad_norm": 0.0683055892586708,
      "learning_rate": 0.0006286932683521169,
      "loss": 2.4626452922821045,
      "step": 8418
    },
    {
      "epoch": 1.1585070252643503,
      "grad_norm": 0.035672787576913834,
      "learning_rate": 0.000627474870545233,
      "loss": 2.488774299621582,
      "step": 8420
    },
    {
      "epoch": 1.1587822110721009,
      "grad_norm": 0.034180622547864914,
      "learning_rate": 0.0006262564727383491,
      "loss": 2.503910541534424,
      "step": 8422
    },
    {
      "epoch": 1.1590573968798517,
      "grad_norm": 0.04656386747956276,
      "learning_rate": 0.0006250380749314652,
      "loss": 2.433910846710205,
      "step": 8424
    },
    {
      "epoch": 1.1593325826876022,
      "grad_norm": 0.029353365302085876,
      "learning_rate": 0.0006238196771245812,
      "loss": 2.485086441040039,
      "step": 8426
    },
    {
      "epoch": 1.1596077684953527,
      "grad_norm": 0.03350144624710083,
      "learning_rate": 0.0006226012793176973,
      "loss": 2.476926326751709,
      "step": 8428
    },
    {
      "epoch": 1.1598829543031033,
      "grad_norm": 0.03126758337020874,
      "learning_rate": 0.0006213828815108133,
      "loss": 2.500495195388794,
      "step": 8430
    },
    {
      "epoch": 1.160158140110854,
      "grad_norm": 0.033315740525722504,
      "learning_rate": 0.0006201644837039293,
      "loss": 2.5079026222229004,
      "step": 8432
    },
    {
      "epoch": 1.1604333259186046,
      "grad_norm": 0.03746699541807175,
      "learning_rate": 0.0006189460858970453,
      "loss": 2.463106632232666,
      "step": 8434
    },
    {
      "epoch": 1.1607085117263551,
      "grad_norm": 0.05905997380614281,
      "learning_rate": 0.0006177276880901614,
      "loss": 2.4473252296447754,
      "step": 8436
    },
    {
      "epoch": 1.160983697534106,
      "grad_norm": 0.0868668407201767,
      "learning_rate": 0.0006165092902832776,
      "loss": 2.511007785797119,
      "step": 8438
    },
    {
      "epoch": 1.1612588833418565,
      "grad_norm": 0.08727698028087616,
      "learning_rate": 0.0006152908924763936,
      "loss": 2.488154411315918,
      "step": 8440
    },
    {
      "epoch": 1.161534069149607,
      "grad_norm": 0.04096714034676552,
      "learning_rate": 0.0006140724946695097,
      "loss": 2.487762451171875,
      "step": 8442
    },
    {
      "epoch": 1.1618092549573578,
      "grad_norm": 0.09843451529741287,
      "learning_rate": 0.0006128540968626256,
      "loss": 2.4966626167297363,
      "step": 8444
    },
    {
      "epoch": 1.1620844407651083,
      "grad_norm": 0.0932060182094574,
      "learning_rate": 0.0006116356990557417,
      "loss": 2.451753616333008,
      "step": 8446
    },
    {
      "epoch": 1.1623596265728589,
      "grad_norm": 0.11571594327688217,
      "learning_rate": 0.0006104173012488577,
      "loss": 2.505841016769409,
      "step": 8448
    },
    {
      "epoch": 1.1626348123806094,
      "grad_norm": 0.08610532432794571,
      "learning_rate": 0.0006091989034419738,
      "loss": 2.4761962890625,
      "step": 8450
    },
    {
      "epoch": 1.1629099981883602,
      "grad_norm": 0.04853058606386185,
      "learning_rate": 0.0006079805056350899,
      "loss": 2.480099678039551,
      "step": 8452
    },
    {
      "epoch": 1.1631851839961107,
      "grad_norm": 0.05170641839504242,
      "learning_rate": 0.000606762107828206,
      "loss": 2.5146734714508057,
      "step": 8454
    },
    {
      "epoch": 1.1634603698038612,
      "grad_norm": 0.02605094574391842,
      "learning_rate": 0.0006055437100213221,
      "loss": 2.477264165878296,
      "step": 8456
    },
    {
      "epoch": 1.163735555611612,
      "grad_norm": 0.02663533389568329,
      "learning_rate": 0.000604325312214438,
      "loss": 2.4800047874450684,
      "step": 8458
    },
    {
      "epoch": 1.1640107414193626,
      "grad_norm": 0.031657084822654724,
      "learning_rate": 0.0006031069144075541,
      "loss": 2.484431266784668,
      "step": 8460
    },
    {
      "epoch": 1.164285927227113,
      "grad_norm": 0.026923896744847298,
      "learning_rate": 0.0006018885166006701,
      "loss": 2.4724714756011963,
      "step": 8462
    },
    {
      "epoch": 1.1645611130348636,
      "grad_norm": 0.024942195042967796,
      "learning_rate": 0.0006006701187937862,
      "loss": 2.457301139831543,
      "step": 8464
    },
    {
      "epoch": 1.1648362988426144,
      "grad_norm": 0.032196298241615295,
      "learning_rate": 0.0005994517209869022,
      "loss": 2.5219509601593018,
      "step": 8466
    },
    {
      "epoch": 1.165111484650365,
      "grad_norm": 0.04201463982462883,
      "learning_rate": 0.0005982333231800183,
      "loss": 2.447697162628174,
      "step": 8468
    },
    {
      "epoch": 1.1653866704581155,
      "grad_norm": 0.0402870699763298,
      "learning_rate": 0.0005970149253731343,
      "loss": 2.475677251815796,
      "step": 8470
    },
    {
      "epoch": 1.1656618562658663,
      "grad_norm": 0.030494455248117447,
      "learning_rate": 0.0005957965275662503,
      "loss": 2.474106788635254,
      "step": 8472
    },
    {
      "epoch": 1.1659370420736168,
      "grad_norm": 0.031888674944639206,
      "learning_rate": 0.0005945781297593665,
      "loss": 2.478917121887207,
      "step": 8474
    },
    {
      "epoch": 1.1662122278813674,
      "grad_norm": 0.04059184342622757,
      "learning_rate": 0.0005933597319524825,
      "loss": 2.457406997680664,
      "step": 8476
    },
    {
      "epoch": 1.1664874136891181,
      "grad_norm": 0.028478534892201424,
      "learning_rate": 0.0005921413341455986,
      "loss": 2.5085067749023438,
      "step": 8478
    },
    {
      "epoch": 1.1667625994968687,
      "grad_norm": 0.028812674805521965,
      "learning_rate": 0.0005909229363387146,
      "loss": 2.513969898223877,
      "step": 8480
    },
    {
      "epoch": 1.1670377853046192,
      "grad_norm": 0.029816119000315666,
      "learning_rate": 0.0005897045385318307,
      "loss": 2.5162370204925537,
      "step": 8482
    },
    {
      "epoch": 1.1673129711123698,
      "grad_norm": 0.02930161915719509,
      "learning_rate": 0.0005884861407249467,
      "loss": 2.4821486473083496,
      "step": 8484
    },
    {
      "epoch": 1.1675881569201205,
      "grad_norm": 0.048478156328201294,
      "learning_rate": 0.0005872677429180627,
      "loss": 2.50784969329834,
      "step": 8486
    },
    {
      "epoch": 1.167863342727871,
      "grad_norm": 0.031294241547584534,
      "learning_rate": 0.0005860493451111788,
      "loss": 2.470640182495117,
      "step": 8488
    },
    {
      "epoch": 1.1681385285356216,
      "grad_norm": 0.027492785826325417,
      "learning_rate": 0.0005848309473042948,
      "loss": 2.538137674331665,
      "step": 8490
    },
    {
      "epoch": 1.1684137143433722,
      "grad_norm": 0.07919273525476456,
      "learning_rate": 0.000583612549497411,
      "loss": 2.501034736633301,
      "step": 8492
    },
    {
      "epoch": 1.168688900151123,
      "grad_norm": 0.07351185381412506,
      "learning_rate": 0.000582394151690527,
      "loss": 2.4820187091827393,
      "step": 8494
    },
    {
      "epoch": 1.1689640859588735,
      "grad_norm": 0.07372671365737915,
      "learning_rate": 0.0005811757538836431,
      "loss": 2.4468655586242676,
      "step": 8496
    },
    {
      "epoch": 1.169239271766624,
      "grad_norm": 0.07030981034040451,
      "learning_rate": 0.000579957356076759,
      "loss": 2.4817562103271484,
      "step": 8498
    },
    {
      "epoch": 1.1695144575743748,
      "grad_norm": 0.12885154783725739,
      "learning_rate": 0.0005787389582698751,
      "loss": 2.4847638607025146,
      "step": 8500
    },
    {
      "epoch": 1.1697896433821253,
      "grad_norm": 0.07669729739427567,
      "learning_rate": 0.0005775205604629912,
      "loss": 2.4743399620056152,
      "step": 8502
    },
    {
      "epoch": 1.1700648291898759,
      "grad_norm": 0.06208887696266174,
      "learning_rate": 0.0005763021626561072,
      "loss": 2.4784698486328125,
      "step": 8504
    },
    {
      "epoch": 1.1703400149976266,
      "grad_norm": 0.050933148711919785,
      "learning_rate": 0.0005750837648492233,
      "loss": 2.4722843170166016,
      "step": 8506
    },
    {
      "epoch": 1.1706152008053772,
      "grad_norm": 0.09202922880649567,
      "learning_rate": 0.0005738653670423393,
      "loss": 2.474773645401001,
      "step": 8508
    },
    {
      "epoch": 1.1708903866131277,
      "grad_norm": 0.062253188341856,
      "learning_rate": 0.0005726469692354554,
      "loss": 2.4313125610351562,
      "step": 8510
    },
    {
      "epoch": 1.1711655724208783,
      "grad_norm": 0.055153340101242065,
      "learning_rate": 0.0005714285714285714,
      "loss": 2.431378126144409,
      "step": 8512
    },
    {
      "epoch": 1.171440758228629,
      "grad_norm": 0.042116113007068634,
      "learning_rate": 0.0005702101736216875,
      "loss": 2.451446533203125,
      "step": 8514
    },
    {
      "epoch": 1.1717159440363796,
      "grad_norm": 0.02820545621216297,
      "learning_rate": 0.0005689917758148036,
      "loss": 2.5025627613067627,
      "step": 8516
    },
    {
      "epoch": 1.1719911298441301,
      "grad_norm": 0.026804449036717415,
      "learning_rate": 0.0005677733780079196,
      "loss": 2.5057525634765625,
      "step": 8518
    },
    {
      "epoch": 1.1722663156518807,
      "grad_norm": 0.04188213124871254,
      "learning_rate": 0.0005665549802010357,
      "loss": 2.4838624000549316,
      "step": 8520
    },
    {
      "epoch": 1.1725415014596314,
      "grad_norm": 0.026573574170470238,
      "learning_rate": 0.0005653365823941517,
      "loss": 2.4746599197387695,
      "step": 8522
    },
    {
      "epoch": 1.172816687267382,
      "grad_norm": 0.03222622349858284,
      "learning_rate": 0.0005641181845872677,
      "loss": 2.4761481285095215,
      "step": 8524
    },
    {
      "epoch": 1.1730918730751325,
      "grad_norm": 0.05096002295613289,
      "learning_rate": 0.0005628997867803837,
      "loss": 2.4424214363098145,
      "step": 8526
    },
    {
      "epoch": 1.1733670588828833,
      "grad_norm": 0.027245238423347473,
      "learning_rate": 0.0005616813889734999,
      "loss": 2.47282075881958,
      "step": 8528
    },
    {
      "epoch": 1.1736422446906338,
      "grad_norm": 0.02836611308157444,
      "learning_rate": 0.000560462991166616,
      "loss": 2.4825472831726074,
      "step": 8530
    },
    {
      "epoch": 1.1739174304983844,
      "grad_norm": 0.0289299339056015,
      "learning_rate": 0.000559244593359732,
      "loss": 2.4726157188415527,
      "step": 8532
    },
    {
      "epoch": 1.1741926163061351,
      "grad_norm": 0.026455475017428398,
      "learning_rate": 0.0005580261955528481,
      "loss": 2.4694809913635254,
      "step": 8534
    },
    {
      "epoch": 1.1744678021138857,
      "grad_norm": 0.053854621946811676,
      "learning_rate": 0.000556807797745964,
      "loss": 2.4754741191864014,
      "step": 8536
    },
    {
      "epoch": 1.1747429879216362,
      "grad_norm": 0.057867348194122314,
      "learning_rate": 0.0005555893999390801,
      "loss": 2.463995933532715,
      "step": 8538
    },
    {
      "epoch": 1.1750181737293868,
      "grad_norm": 0.049233559519052505,
      "learning_rate": 0.0005543710021321961,
      "loss": 2.439931631088257,
      "step": 8540
    },
    {
      "epoch": 1.1752933595371375,
      "grad_norm": 0.04286424070596695,
      "learning_rate": 0.0005531526043253122,
      "loss": 2.488924264907837,
      "step": 8542
    },
    {
      "epoch": 1.175568545344888,
      "grad_norm": 0.03908887133002281,
      "learning_rate": 0.0005519342065184282,
      "loss": 2.4679527282714844,
      "step": 8544
    },
    {
      "epoch": 1.1758437311526386,
      "grad_norm": 0.026763249188661575,
      "learning_rate": 0.0005507158087115444,
      "loss": 2.4716696739196777,
      "step": 8546
    },
    {
      "epoch": 1.1761189169603892,
      "grad_norm": 0.057844407856464386,
      "learning_rate": 0.0005494974109046605,
      "loss": 2.450366973876953,
      "step": 8548
    },
    {
      "epoch": 1.17639410276814,
      "grad_norm": 0.046296827495098114,
      "learning_rate": 0.0005482790130977764,
      "loss": 2.506364345550537,
      "step": 8550
    },
    {
      "epoch": 1.1766692885758905,
      "grad_norm": 0.030109576880931854,
      "learning_rate": 0.0005470606152908925,
      "loss": 2.397833824157715,
      "step": 8552
    },
    {
      "epoch": 1.176944474383641,
      "grad_norm": 0.03319597616791725,
      "learning_rate": 0.0005458422174840085,
      "loss": 2.4731743335723877,
      "step": 8554
    },
    {
      "epoch": 1.1772196601913918,
      "grad_norm": 0.06661998480558395,
      "learning_rate": 0.0005446238196771246,
      "loss": 2.4965131282806396,
      "step": 8556
    },
    {
      "epoch": 1.1774948459991423,
      "grad_norm": 0.03787920996546745,
      "learning_rate": 0.0005434054218702406,
      "loss": 2.436227798461914,
      "step": 8558
    },
    {
      "epoch": 1.1777700318068929,
      "grad_norm": 0.04407722130417824,
      "learning_rate": 0.0005421870240633567,
      "loss": 2.5293312072753906,
      "step": 8560
    },
    {
      "epoch": 1.1780452176146436,
      "grad_norm": 0.09424177557229996,
      "learning_rate": 0.0005409686262564727,
      "loss": 2.5078659057617188,
      "step": 8562
    },
    {
      "epoch": 1.1783204034223942,
      "grad_norm": 0.0566101036965847,
      "learning_rate": 0.0005397502284495888,
      "loss": 2.46651029586792,
      "step": 8564
    },
    {
      "epoch": 1.1785955892301447,
      "grad_norm": 0.03349711745977402,
      "learning_rate": 0.0005385318306427049,
      "loss": 2.462888240814209,
      "step": 8566
    },
    {
      "epoch": 1.1788707750378955,
      "grad_norm": 0.055710434913635254,
      "learning_rate": 0.0005373134328358209,
      "loss": 2.485239028930664,
      "step": 8568
    },
    {
      "epoch": 1.179145960845646,
      "grad_norm": 0.047695938497781754,
      "learning_rate": 0.000536095035028937,
      "loss": 2.4568920135498047,
      "step": 8570
    },
    {
      "epoch": 1.1794211466533966,
      "grad_norm": 0.04080001637339592,
      "learning_rate": 0.000534876637222053,
      "loss": 2.442403793334961,
      "step": 8572
    },
    {
      "epoch": 1.1796963324611471,
      "grad_norm": 0.07023676484823227,
      "learning_rate": 0.0005336582394151691,
      "loss": 2.455559253692627,
      "step": 8574
    },
    {
      "epoch": 1.179971518268898,
      "grad_norm": 0.10817483067512512,
      "learning_rate": 0.000532439841608285,
      "loss": 2.430891990661621,
      "step": 8576
    },
    {
      "epoch": 1.1802467040766484,
      "grad_norm": 0.11201369762420654,
      "learning_rate": 0.0005312214438014011,
      "loss": 2.4346795082092285,
      "step": 8578
    },
    {
      "epoch": 1.180521889884399,
      "grad_norm": 0.04322679340839386,
      "learning_rate": 0.0005300030459945173,
      "loss": 2.483079433441162,
      "step": 8580
    },
    {
      "epoch": 1.1807970756921495,
      "grad_norm": 0.05259304121136665,
      "learning_rate": 0.0005287846481876333,
      "loss": 2.5023889541625977,
      "step": 8582
    },
    {
      "epoch": 1.1810722614999003,
      "grad_norm": 0.07048050314188004,
      "learning_rate": 0.0005275662503807494,
      "loss": 2.497157573699951,
      "step": 8584
    },
    {
      "epoch": 1.1813474473076508,
      "grad_norm": 0.08057713508605957,
      "learning_rate": 0.0005263478525738654,
      "loss": 2.4843716621398926,
      "step": 8586
    },
    {
      "epoch": 1.1816226331154014,
      "grad_norm": 0.04945176839828491,
      "learning_rate": 0.0005251294547669815,
      "loss": 2.4751336574554443,
      "step": 8588
    },
    {
      "epoch": 1.1818978189231522,
      "grad_norm": 0.03627081215381622,
      "learning_rate": 0.0005239110569600974,
      "loss": 2.4722142219543457,
      "step": 8590
    },
    {
      "epoch": 1.1821730047309027,
      "grad_norm": 0.026835620403289795,
      "learning_rate": 0.0005226926591532135,
      "loss": 2.4823660850524902,
      "step": 8592
    },
    {
      "epoch": 1.1824481905386532,
      "grad_norm": 0.028032803907990456,
      "learning_rate": 0.0005214742613463296,
      "loss": 2.440993070602417,
      "step": 8594
    },
    {
      "epoch": 1.182723376346404,
      "grad_norm": 0.02541583962738514,
      "learning_rate": 0.0005202558635394456,
      "loss": 2.5193333625793457,
      "step": 8596
    },
    {
      "epoch": 1.1829985621541546,
      "grad_norm": 0.02785034477710724,
      "learning_rate": 0.0005190374657325618,
      "loss": 2.490316390991211,
      "step": 8598
    },
    {
      "epoch": 1.183273747961905,
      "grad_norm": 0.02643394097685814,
      "learning_rate": 0.0005178190679256778,
      "loss": 2.4868125915527344,
      "step": 8600
    },
    {
      "epoch": 1.1835489337696556,
      "grad_norm": 0.028258899226784706,
      "learning_rate": 0.0005166006701187938,
      "loss": 2.4677414894104004,
      "step": 8602
    },
    {
      "epoch": 1.1838241195774064,
      "grad_norm": 0.04276259243488312,
      "learning_rate": 0.0005153822723119098,
      "loss": 2.4856784343719482,
      "step": 8604
    },
    {
      "epoch": 1.184099305385157,
      "grad_norm": 0.029236052185297012,
      "learning_rate": 0.0005141638745050259,
      "loss": 2.4903740882873535,
      "step": 8606
    },
    {
      "epoch": 1.1843744911929075,
      "grad_norm": 0.04992251098155975,
      "learning_rate": 0.000512945476698142,
      "loss": 2.4691739082336426,
      "step": 8608
    },
    {
      "epoch": 1.184649677000658,
      "grad_norm": 0.026921182870864868,
      "learning_rate": 0.000511727078891258,
      "loss": 2.5242867469787598,
      "step": 8610
    },
    {
      "epoch": 1.1849248628084088,
      "grad_norm": 0.03775021433830261,
      "learning_rate": 0.0005105086810843741,
      "loss": 2.4671568870544434,
      "step": 8612
    },
    {
      "epoch": 1.1852000486161594,
      "grad_norm": 0.05926031619310379,
      "learning_rate": 0.0005092902832774901,
      "loss": 2.4535930156707764,
      "step": 8614
    },
    {
      "epoch": 1.18547523442391,
      "grad_norm": 0.05345671996474266,
      "learning_rate": 0.0005080718854706061,
      "loss": 2.47186279296875,
      "step": 8616
    },
    {
      "epoch": 1.1857504202316607,
      "grad_norm": 0.07402487844228745,
      "learning_rate": 0.0005068534876637222,
      "loss": 2.475618839263916,
      "step": 8618
    },
    {
      "epoch": 1.1860256060394112,
      "grad_norm": 0.031279727816581726,
      "learning_rate": 0.0005056350898568383,
      "loss": 2.4523935317993164,
      "step": 8620
    },
    {
      "epoch": 1.1863007918471618,
      "grad_norm": 0.029318010434508324,
      "learning_rate": 0.0005044166920499543,
      "loss": 2.4826366901397705,
      "step": 8622
    },
    {
      "epoch": 1.1865759776549125,
      "grad_norm": 0.026097308844327927,
      "learning_rate": 0.0005031982942430704,
      "loss": 2.4601247310638428,
      "step": 8624
    },
    {
      "epoch": 1.186851163462663,
      "grad_norm": 0.030194276943802834,
      "learning_rate": 0.0005019798964361865,
      "loss": 2.4754672050476074,
      "step": 8626
    },
    {
      "epoch": 1.1871263492704136,
      "grad_norm": 0.03776143863797188,
      "learning_rate": 0.0005007614986293024,
      "loss": 2.4421730041503906,
      "step": 8628
    },
    {
      "epoch": 1.1874015350781641,
      "grad_norm": 0.0535174161195755,
      "learning_rate": 0.0004995431008224185,
      "loss": 2.4875621795654297,
      "step": 8630
    },
    {
      "epoch": 1.187676720885915,
      "grad_norm": 0.02867351658642292,
      "learning_rate": 0.0004983247030155345,
      "loss": 2.4918253421783447,
      "step": 8632
    },
    {
      "epoch": 1.1879519066936655,
      "grad_norm": 0.025506816804409027,
      "learning_rate": 0.0004971063052086507,
      "loss": 2.4863393306732178,
      "step": 8634
    },
    {
      "epoch": 1.188227092501416,
      "grad_norm": 0.04484231397509575,
      "learning_rate": 0.0004958879074017667,
      "loss": 2.493821382522583,
      "step": 8636
    },
    {
      "epoch": 1.1885022783091665,
      "grad_norm": 0.030159184709191322,
      "learning_rate": 0.0004946695095948827,
      "loss": 2.4827747344970703,
      "step": 8638
    },
    {
      "epoch": 1.1887774641169173,
      "grad_norm": 0.02866705134510994,
      "learning_rate": 0.0004934511117879988,
      "loss": 2.4644267559051514,
      "step": 8640
    },
    {
      "epoch": 1.1890526499246679,
      "grad_norm": 0.05815074220299721,
      "learning_rate": 0.0004922327139811149,
      "loss": 2.4497568607330322,
      "step": 8642
    },
    {
      "epoch": 1.1893278357324184,
      "grad_norm": 0.0282639991492033,
      "learning_rate": 0.0004910143161742309,
      "loss": 2.494298219680786,
      "step": 8644
    },
    {
      "epoch": 1.1896030215401692,
      "grad_norm": 0.036480583250522614,
      "learning_rate": 0.0004897959183673469,
      "loss": 2.456871747970581,
      "step": 8646
    },
    {
      "epoch": 1.1898782073479197,
      "grad_norm": 0.025883959606289864,
      "learning_rate": 0.000488577520560463,
      "loss": 2.5268614292144775,
      "step": 8648
    },
    {
      "epoch": 1.1901533931556703,
      "grad_norm": 0.026932192966341972,
      "learning_rate": 0.0004873591227535791,
      "loss": 2.5119948387145996,
      "step": 8650
    },
    {
      "epoch": 1.190428578963421,
      "grad_norm": 0.026563217863440514,
      "learning_rate": 0.0004861407249466951,
      "loss": 2.5146727561950684,
      "step": 8652
    },
    {
      "epoch": 1.1907037647711716,
      "grad_norm": 0.053433675318956375,
      "learning_rate": 0.00048492232713981116,
      "loss": 2.4428560733795166,
      "step": 8654
    },
    {
      "epoch": 1.1909789505789221,
      "grad_norm": 0.025516720488667488,
      "learning_rate": 0.0004837039293329272,
      "loss": 2.509239912033081,
      "step": 8656
    },
    {
      "epoch": 1.1912541363866727,
      "grad_norm": 0.04371418058872223,
      "learning_rate": 0.00048248553152604323,
      "loss": 2.4663257598876953,
      "step": 8658
    },
    {
      "epoch": 1.1915293221944234,
      "grad_norm": 0.030810056254267693,
      "learning_rate": 0.0004812671337191593,
      "loss": 2.4689602851867676,
      "step": 8660
    },
    {
      "epoch": 1.191804508002174,
      "grad_norm": 0.03950062021613121,
      "learning_rate": 0.00048004873591227535,
      "loss": 2.4969935417175293,
      "step": 8662
    },
    {
      "epoch": 1.1920796938099245,
      "grad_norm": 0.03437010198831558,
      "learning_rate": 0.0004788303381053914,
      "loss": 2.460188388824463,
      "step": 8664
    },
    {
      "epoch": 1.1923548796176753,
      "grad_norm": 0.030644772574305534,
      "learning_rate": 0.0004776119402985075,
      "loss": 2.42600417137146,
      "step": 8666
    },
    {
      "epoch": 1.1926300654254258,
      "grad_norm": 0.06939103454351425,
      "learning_rate": 0.00047639354249162354,
      "loss": 2.48158597946167,
      "step": 8668
    },
    {
      "epoch": 1.1929052512331764,
      "grad_norm": 0.04514940455555916,
      "learning_rate": 0.0004751751446847396,
      "loss": 2.4459056854248047,
      "step": 8670
    },
    {
      "epoch": 1.193180437040927,
      "grad_norm": 0.02769850753247738,
      "learning_rate": 0.0004739567468778556,
      "loss": 2.4718425273895264,
      "step": 8672
    },
    {
      "epoch": 1.1934556228486777,
      "grad_norm": 0.025819890201091766,
      "learning_rate": 0.00047273834907097167,
      "loss": 2.4717519283294678,
      "step": 8674
    },
    {
      "epoch": 1.1937308086564282,
      "grad_norm": 0.027808645740151405,
      "learning_rate": 0.00047151995126408773,
      "loss": 2.502664804458618,
      "step": 8676
    },
    {
      "epoch": 1.1940059944641788,
      "grad_norm": 0.026463108137249947,
      "learning_rate": 0.00047030155345720374,
      "loss": 2.459801197052002,
      "step": 8678
    },
    {
      "epoch": 1.1942811802719295,
      "grad_norm": 0.025830909609794617,
      "learning_rate": 0.00046908315565031986,
      "loss": 2.4705429077148438,
      "step": 8680
    },
    {
      "epoch": 1.19455636607968,
      "grad_norm": 0.04405985400080681,
      "learning_rate": 0.0004678647578434359,
      "loss": 2.477264881134033,
      "step": 8682
    },
    {
      "epoch": 1.1948315518874306,
      "grad_norm": 0.026802146807312965,
      "learning_rate": 0.0004666463600365519,
      "loss": 2.439289093017578,
      "step": 8684
    },
    {
      "epoch": 1.1951067376951814,
      "grad_norm": 0.05216054245829582,
      "learning_rate": 0.000465427962229668,
      "loss": 2.4781992435455322,
      "step": 8686
    },
    {
      "epoch": 1.195381923502932,
      "grad_norm": 0.031276822090148926,
      "learning_rate": 0.00046420956442278405,
      "loss": 2.469414472579956,
      "step": 8688
    },
    {
      "epoch": 1.1956571093106825,
      "grad_norm": 0.036080725491046906,
      "learning_rate": 0.0004629911666159001,
      "loss": 2.489471435546875,
      "step": 8690
    },
    {
      "epoch": 1.195932295118433,
      "grad_norm": 0.050610702484846115,
      "learning_rate": 0.0004617727688090161,
      "loss": 2.468400239944458,
      "step": 8692
    },
    {
      "epoch": 1.1962074809261838,
      "grad_norm": 0.029053451493382454,
      "learning_rate": 0.00046055437100213224,
      "loss": 2.4778554439544678,
      "step": 8694
    },
    {
      "epoch": 1.1964826667339343,
      "grad_norm": 0.026429925113916397,
      "learning_rate": 0.0004593359731952483,
      "loss": 2.4543070793151855,
      "step": 8696
    },
    {
      "epoch": 1.1967578525416849,
      "grad_norm": 0.02676769159734249,
      "learning_rate": 0.0004581175753883643,
      "loss": 2.4431557655334473,
      "step": 8698
    },
    {
      "epoch": 1.1970330383494354,
      "grad_norm": 0.027419958263635635,
      "learning_rate": 0.00045689917758148037,
      "loss": 2.4383792877197266,
      "step": 8700
    },
    {
      "epoch": 1.1973082241571862,
      "grad_norm": 0.03429315984249115,
      "learning_rate": 0.00045568077977459643,
      "loss": 2.507478713989258,
      "step": 8702
    },
    {
      "epoch": 1.1975834099649367,
      "grad_norm": 0.04442322626709938,
      "learning_rate": 0.00045446238196771244,
      "loss": 2.504960060119629,
      "step": 8704
    },
    {
      "epoch": 1.1978585957726873,
      "grad_norm": 0.05759274214506149,
      "learning_rate": 0.0004532439841608285,
      "loss": 2.504065990447998,
      "step": 8706
    },
    {
      "epoch": 1.198133781580438,
      "grad_norm": 0.034870464354753494,
      "learning_rate": 0.00045202558635394456,
      "loss": 2.481231927871704,
      "step": 8708
    },
    {
      "epoch": 1.1984089673881886,
      "grad_norm": 0.056764017790555954,
      "learning_rate": 0.0004508071885470606,
      "loss": 2.4213624000549316,
      "step": 8710
    },
    {
      "epoch": 1.1986841531959391,
      "grad_norm": 0.06754005700349808,
      "learning_rate": 0.0004495887907401767,
      "loss": 2.489760637283325,
      "step": 8712
    },
    {
      "epoch": 1.19895933900369,
      "grad_norm": 0.06115381047129631,
      "learning_rate": 0.00044837039293329275,
      "loss": 2.458911418914795,
      "step": 8714
    },
    {
      "epoch": 1.1992345248114404,
      "grad_norm": 0.054524607956409454,
      "learning_rate": 0.0004471519951264088,
      "loss": 2.442030191421509,
      "step": 8716
    },
    {
      "epoch": 1.199509710619191,
      "grad_norm": 0.02639007940888405,
      "learning_rate": 0.0004459335973195248,
      "loss": 2.504216432571411,
      "step": 8718
    },
    {
      "epoch": 1.1997848964269415,
      "grad_norm": 0.06522145122289658,
      "learning_rate": 0.0004447151995126409,
      "loss": 2.47813081741333,
      "step": 8720
    },
    {
      "epoch": 1.2000600822346923,
      "grad_norm": 0.05460658296942711,
      "learning_rate": 0.00044349680170575694,
      "loss": 2.4694604873657227,
      "step": 8722
    },
    {
      "epoch": 1.2003352680424428,
      "grad_norm": 0.04274258762598038,
      "learning_rate": 0.00044227840389887295,
      "loss": 2.504851818084717,
      "step": 8724
    },
    {
      "epoch": 1.2006104538501934,
      "grad_norm": 0.02569100260734558,
      "learning_rate": 0.00044106000609198906,
      "loss": 2.478699207305908,
      "step": 8726
    },
    {
      "epoch": 1.200885639657944,
      "grad_norm": 0.02615550346672535,
      "learning_rate": 0.0004398416082851051,
      "loss": 2.4744057655334473,
      "step": 8728
    },
    {
      "epoch": 1.2011608254656947,
      "grad_norm": 0.02948834002017975,
      "learning_rate": 0.00043862321047822113,
      "loss": 2.459836483001709,
      "step": 8730
    },
    {
      "epoch": 1.2014360112734452,
      "grad_norm": 0.026878049597144127,
      "learning_rate": 0.0004374048126713372,
      "loss": 2.470933198928833,
      "step": 8732
    },
    {
      "epoch": 1.2017111970811958,
      "grad_norm": 0.030868232250213623,
      "learning_rate": 0.00043618641486445326,
      "loss": 2.453636646270752,
      "step": 8734
    },
    {
      "epoch": 1.2019863828889465,
      "grad_norm": 0.04337636008858681,
      "learning_rate": 0.0004349680170575693,
      "loss": 2.4556937217712402,
      "step": 8736
    },
    {
      "epoch": 1.202261568696697,
      "grad_norm": 0.030545899644494057,
      "learning_rate": 0.0004337496192506853,
      "loss": 2.4677176475524902,
      "step": 8738
    },
    {
      "epoch": 1.2025367545044476,
      "grad_norm": 0.03017844632267952,
      "learning_rate": 0.0004325312214438014,
      "loss": 2.465578079223633,
      "step": 8740
    },
    {
      "epoch": 1.2028119403121984,
      "grad_norm": 0.024908188730478287,
      "learning_rate": 0.0004313128236369175,
      "loss": 2.473461389541626,
      "step": 8742
    },
    {
      "epoch": 1.203087126119949,
      "grad_norm": 0.0398617647588253,
      "learning_rate": 0.0004300944258300335,
      "loss": 2.4231667518615723,
      "step": 8744
    },
    {
      "epoch": 1.2033623119276995,
      "grad_norm": 0.027448071166872978,
      "learning_rate": 0.0004288760280231496,
      "loss": 2.4859297275543213,
      "step": 8746
    },
    {
      "epoch": 1.20363749773545,
      "grad_norm": 0.04996515065431595,
      "learning_rate": 0.00042765763021626564,
      "loss": 2.4736390113830566,
      "step": 8748
    },
    {
      "epoch": 1.2039126835432008,
      "grad_norm": 0.1139926165342331,
      "learning_rate": 0.00042643923240938164,
      "loss": 2.4843201637268066,
      "step": 8750
    },
    {
      "epoch": 1.2041878693509513,
      "grad_norm": 0.06015070527791977,
      "learning_rate": 0.0004252208346024977,
      "loss": 2.456665515899658,
      "step": 8752
    },
    {
      "epoch": 1.204463055158702,
      "grad_norm": 0.04736275225877762,
      "learning_rate": 0.00042400243679561377,
      "loss": 2.463442325592041,
      "step": 8754
    },
    {
      "epoch": 1.2047382409664524,
      "grad_norm": 0.042718224227428436,
      "learning_rate": 0.0004227840389887298,
      "loss": 2.456305503845215,
      "step": 8756
    },
    {
      "epoch": 1.2050134267742032,
      "grad_norm": 0.03286021947860718,
      "learning_rate": 0.0004215656411818459,
      "loss": 2.4624977111816406,
      "step": 8758
    },
    {
      "epoch": 1.2052886125819537,
      "grad_norm": 0.07304903864860535,
      "learning_rate": 0.00042034724337496195,
      "loss": 2.4338793754577637,
      "step": 8760
    },
    {
      "epoch": 1.2055637983897043,
      "grad_norm": 0.06499272584915161,
      "learning_rate": 0.000419128845568078,
      "loss": 2.4429211616516113,
      "step": 8762
    },
    {
      "epoch": 1.205838984197455,
      "grad_norm": 0.07357119023799896,
      "learning_rate": 0.000417910447761194,
      "loss": 2.4988369941711426,
      "step": 8764
    },
    {
      "epoch": 1.2061141700052056,
      "grad_norm": 0.043625105172395706,
      "learning_rate": 0.0004166920499543101,
      "loss": 2.4689550399780273,
      "step": 8766
    },
    {
      "epoch": 1.2063893558129561,
      "grad_norm": 0.04893765598535538,
      "learning_rate": 0.00041547365214742615,
      "loss": 2.456655979156494,
      "step": 8768
    },
    {
      "epoch": 1.206664541620707,
      "grad_norm": 0.04677034541964531,
      "learning_rate": 0.00041425525434054215,
      "loss": 2.486523151397705,
      "step": 8770
    },
    {
      "epoch": 1.2069397274284575,
      "grad_norm": 0.043685182929039,
      "learning_rate": 0.00041303685653365827,
      "loss": 2.4847545623779297,
      "step": 8772
    },
    {
      "epoch": 1.207214913236208,
      "grad_norm": 0.03671583905816078,
      "learning_rate": 0.00041181845872677433,
      "loss": 2.491459369659424,
      "step": 8774
    },
    {
      "epoch": 1.2074900990439588,
      "grad_norm": 0.042950164526700974,
      "learning_rate": 0.00041060006091989034,
      "loss": 2.488377571105957,
      "step": 8776
    },
    {
      "epoch": 1.2077652848517093,
      "grad_norm": 0.05177051201462746,
      "learning_rate": 0.0004093816631130064,
      "loss": 2.4498181343078613,
      "step": 8778
    },
    {
      "epoch": 1.2080404706594599,
      "grad_norm": 0.07465200126171112,
      "learning_rate": 0.00040816326530612246,
      "loss": 2.4598259925842285,
      "step": 8780
    },
    {
      "epoch": 1.2083156564672104,
      "grad_norm": 0.025465643033385277,
      "learning_rate": 0.0004069448674992385,
      "loss": 2.4219954013824463,
      "step": 8782
    },
    {
      "epoch": 1.2085908422749612,
      "grad_norm": 0.027311531826853752,
      "learning_rate": 0.00040572646969235453,
      "loss": 2.4905481338500977,
      "step": 8784
    },
    {
      "epoch": 1.2088660280827117,
      "grad_norm": 0.029975229874253273,
      "learning_rate": 0.0004045080718854706,
      "loss": 2.4464774131774902,
      "step": 8786
    },
    {
      "epoch": 1.2091412138904623,
      "grad_norm": 0.03452488034963608,
      "learning_rate": 0.0004032896740785867,
      "loss": 2.491997718811035,
      "step": 8788
    },
    {
      "epoch": 1.2094163996982128,
      "grad_norm": 0.03516984358429909,
      "learning_rate": 0.0004020712762717027,
      "loss": 2.4754719734191895,
      "step": 8790
    },
    {
      "epoch": 1.2096915855059636,
      "grad_norm": 0.039657458662986755,
      "learning_rate": 0.0004008528784648188,
      "loss": 2.452157974243164,
      "step": 8792
    },
    {
      "epoch": 1.209966771313714,
      "grad_norm": 0.04248063638806343,
      "learning_rate": 0.00039963448065793484,
      "loss": 2.450876235961914,
      "step": 8794
    },
    {
      "epoch": 1.2102419571214647,
      "grad_norm": 0.035585444420576096,
      "learning_rate": 0.00039841608285105085,
      "loss": 2.470597267150879,
      "step": 8796
    },
    {
      "epoch": 1.2105171429292154,
      "grad_norm": 0.04110002517700195,
      "learning_rate": 0.0003971976850441669,
      "loss": 2.4914891719818115,
      "step": 8798
    },
    {
      "epoch": 1.210792328736966,
      "grad_norm": 0.032566286623477936,
      "learning_rate": 0.000395979287237283,
      "loss": 2.4725921154022217,
      "step": 8800
    },
    {
      "epoch": 1.2110675145447165,
      "grad_norm": 0.026849210262298584,
      "learning_rate": 0.000394760889430399,
      "loss": 2.4793097972869873,
      "step": 8802
    },
    {
      "epoch": 1.2113427003524673,
      "grad_norm": 0.028937242925167084,
      "learning_rate": 0.0003935424916235151,
      "loss": 2.462602138519287,
      "step": 8804
    },
    {
      "epoch": 1.2116178861602178,
      "grad_norm": 0.026845673099160194,
      "learning_rate": 0.00039232409381663116,
      "loss": 2.4440174102783203,
      "step": 8806
    },
    {
      "epoch": 1.2118930719679684,
      "grad_norm": 0.041239842772483826,
      "learning_rate": 0.0003911056960097472,
      "loss": 2.4532017707824707,
      "step": 8808
    },
    {
      "epoch": 1.212168257775719,
      "grad_norm": 0.027225656434893608,
      "learning_rate": 0.00038988729820286323,
      "loss": 2.4820194244384766,
      "step": 8810
    },
    {
      "epoch": 1.2124434435834697,
      "grad_norm": 0.029069174081087112,
      "learning_rate": 0.0003886689003959793,
      "loss": 2.4859776496887207,
      "step": 8812
    },
    {
      "epoch": 1.2127186293912202,
      "grad_norm": 0.024674084037542343,
      "learning_rate": 0.00038745050258909535,
      "loss": 2.4597177505493164,
      "step": 8814
    },
    {
      "epoch": 1.2129938151989708,
      "grad_norm": 0.0267214123159647,
      "learning_rate": 0.00038623210478221136,
      "loss": 2.4306435585021973,
      "step": 8816
    },
    {
      "epoch": 1.2132690010067213,
      "grad_norm": 0.0279080793261528,
      "learning_rate": 0.0003850137069753274,
      "loss": 2.4768738746643066,
      "step": 8818
    },
    {
      "epoch": 1.213544186814472,
      "grad_norm": 0.05523066967725754,
      "learning_rate": 0.00038379530916844354,
      "loss": 2.4986910820007324,
      "step": 8820
    },
    {
      "epoch": 1.2138193726222226,
      "grad_norm": 0.10398225486278534,
      "learning_rate": 0.00038257691136155955,
      "loss": 2.453376054763794,
      "step": 8822
    },
    {
      "epoch": 1.2140945584299732,
      "grad_norm": 0.081839419901371,
      "learning_rate": 0.0003813585135546756,
      "loss": 2.464280128479004,
      "step": 8824
    },
    {
      "epoch": 1.214369744237724,
      "grad_norm": 0.02556564472615719,
      "learning_rate": 0.00038014011574779167,
      "loss": 2.440875768661499,
      "step": 8826
    },
    {
      "epoch": 1.2146449300454745,
      "grad_norm": 0.09514444321393967,
      "learning_rate": 0.00037892171794090773,
      "loss": 2.4620862007141113,
      "step": 8828
    },
    {
      "epoch": 1.214920115853225,
      "grad_norm": 0.1271274834871292,
      "learning_rate": 0.00037770332013402374,
      "loss": 2.457763671875,
      "step": 8830
    },
    {
      "epoch": 1.2151953016609758,
      "grad_norm": 0.032811474055051804,
      "learning_rate": 0.0003764849223271398,
      "loss": 2.4793639183044434,
      "step": 8832
    },
    {
      "epoch": 1.2154704874687263,
      "grad_norm": 0.11348410695791245,
      "learning_rate": 0.0003752665245202559,
      "loss": 2.4599878787994385,
      "step": 8834
    },
    {
      "epoch": 1.2157456732764769,
      "grad_norm": 0.07678667455911636,
      "learning_rate": 0.0003740481267133719,
      "loss": 2.486891746520996,
      "step": 8836
    },
    {
      "epoch": 1.2160208590842274,
      "grad_norm": 0.04440055787563324,
      "learning_rate": 0.000372829728906488,
      "loss": 2.456447124481201,
      "step": 8838
    },
    {
      "epoch": 1.2162960448919782,
      "grad_norm": 0.09812157601118088,
      "learning_rate": 0.00037161133109960405,
      "loss": 2.5057811737060547,
      "step": 8840
    },
    {
      "epoch": 1.2165712306997287,
      "grad_norm": 0.031434498727321625,
      "learning_rate": 0.00037039293329272006,
      "loss": 2.4847702980041504,
      "step": 8842
    },
    {
      "epoch": 1.2168464165074793,
      "grad_norm": 0.10320954024791718,
      "learning_rate": 0.0003691745354858361,
      "loss": 2.4739620685577393,
      "step": 8844
    },
    {
      "epoch": 1.2171216023152298,
      "grad_norm": 0.11398742347955704,
      "learning_rate": 0.0003679561376789522,
      "loss": 2.4958982467651367,
      "step": 8846
    },
    {
      "epoch": 1.2173967881229806,
      "grad_norm": 0.03367593139410019,
      "learning_rate": 0.00036673773987206824,
      "loss": 2.462498664855957,
      "step": 8848
    },
    {
      "epoch": 1.2176719739307311,
      "grad_norm": 0.06830449402332306,
      "learning_rate": 0.0003655193420651843,
      "loss": 2.4631996154785156,
      "step": 8850
    },
    {
      "epoch": 1.2179471597384817,
      "grad_norm": 0.0444931797683239,
      "learning_rate": 0.00036430094425830037,
      "loss": 2.439612865447998,
      "step": 8852
    },
    {
      "epoch": 1.2182223455462324,
      "grad_norm": 0.03447304293513298,
      "learning_rate": 0.00036308254645141643,
      "loss": 2.5316848754882812,
      "step": 8854
    },
    {
      "epoch": 1.218497531353983,
      "grad_norm": 0.06705034524202347,
      "learning_rate": 0.00036186414864453244,
      "loss": 2.4550440311431885,
      "step": 8856
    },
    {
      "epoch": 1.2187727171617335,
      "grad_norm": 0.03081127442419529,
      "learning_rate": 0.0003606457508376485,
      "loss": 2.4654078483581543,
      "step": 8858
    },
    {
      "epoch": 1.2190479029694843,
      "grad_norm": 0.04607037454843521,
      "learning_rate": 0.00035942735303076456,
      "loss": 2.4596080780029297,
      "step": 8860
    },
    {
      "epoch": 1.2193230887772348,
      "grad_norm": 0.04579803720116615,
      "learning_rate": 0.00035820895522388057,
      "loss": 2.4844906330108643,
      "step": 8862
    },
    {
      "epoch": 1.2195982745849854,
      "grad_norm": 0.026145927608013153,
      "learning_rate": 0.00035699055741699663,
      "loss": 2.4469752311706543,
      "step": 8864
    },
    {
      "epoch": 1.219873460392736,
      "grad_norm": 0.03595560044050217,
      "learning_rate": 0.00035577215961011275,
      "loss": 2.461869239807129,
      "step": 8866
    },
    {
      "epoch": 1.2201486462004867,
      "grad_norm": 0.03903980925679207,
      "learning_rate": 0.00035455376180322875,
      "loss": 2.483745574951172,
      "step": 8868
    },
    {
      "epoch": 1.2204238320082372,
      "grad_norm": 0.026750175282359123,
      "learning_rate": 0.0003533353639963448,
      "loss": 2.455982208251953,
      "step": 8870
    },
    {
      "epoch": 1.2206990178159878,
      "grad_norm": 0.06166665628552437,
      "learning_rate": 0.0003521169661894609,
      "loss": 2.4688618183135986,
      "step": 8872
    },
    {
      "epoch": 1.2209742036237385,
      "grad_norm": 0.030036283656954765,
      "learning_rate": 0.00035089856838257694,
      "loss": 2.4493227005004883,
      "step": 8874
    },
    {
      "epoch": 1.221249389431489,
      "grad_norm": 0.02775690332055092,
      "learning_rate": 0.00034968017057569295,
      "loss": 2.4693961143493652,
      "step": 8876
    },
    {
      "epoch": 1.2215245752392396,
      "grad_norm": 0.029315469786524773,
      "learning_rate": 0.000348461772768809,
      "loss": 2.473182201385498,
      "step": 8878
    },
    {
      "epoch": 1.2217997610469902,
      "grad_norm": 0.04584071785211563,
      "learning_rate": 0.00034724337496192507,
      "loss": 2.4239397048950195,
      "step": 8880
    },
    {
      "epoch": 1.222074946854741,
      "grad_norm": 0.05503488704562187,
      "learning_rate": 0.00034602497715504113,
      "loss": 2.449183940887451,
      "step": 8882
    },
    {
      "epoch": 1.2223501326624915,
      "grad_norm": 0.07854048162698746,
      "learning_rate": 0.0003448065793481572,
      "loss": 2.506598711013794,
      "step": 8884
    },
    {
      "epoch": 1.222625318470242,
      "grad_norm": 0.024540036916732788,
      "learning_rate": 0.00034358818154127326,
      "loss": 2.4966657161712646,
      "step": 8886
    },
    {
      "epoch": 1.2229005042779928,
      "grad_norm": 0.05817464739084244,
      "learning_rate": 0.00034236978373438926,
      "loss": 2.468033790588379,
      "step": 8888
    },
    {
      "epoch": 1.2231756900857433,
      "grad_norm": 0.06302707642316818,
      "learning_rate": 0.0003411513859275053,
      "loss": 2.5060083866119385,
      "step": 8890
    },
    {
      "epoch": 1.2234508758934939,
      "grad_norm": 0.0263026412576437,
      "learning_rate": 0.0003399329881206214,
      "loss": 2.454346179962158,
      "step": 8892
    },
    {
      "epoch": 1.2237260617012446,
      "grad_norm": 0.02917100489139557,
      "learning_rate": 0.00033871459031373745,
      "loss": 2.47318172454834,
      "step": 8894
    },
    {
      "epoch": 1.2240012475089952,
      "grad_norm": 0.041920337826013565,
      "learning_rate": 0.00033749619250685346,
      "loss": 2.4973304271698,
      "step": 8896
    },
    {
      "epoch": 1.2242764333167457,
      "grad_norm": 0.059054918587207794,
      "learning_rate": 0.0003362777946999696,
      "loss": 2.4806647300720215,
      "step": 8898
    },
    {
      "epoch": 1.2245516191244963,
      "grad_norm": 0.03674178570508957,
      "learning_rate": 0.00033505939689308564,
      "loss": 2.461876392364502,
      "step": 8900
    },
    {
      "epoch": 1.224826804932247,
      "grad_norm": 0.0263822004199028,
      "learning_rate": 0.00033384099908620164,
      "loss": 2.496554136276245,
      "step": 8902
    },
    {
      "epoch": 1.2251019907399976,
      "grad_norm": 0.05070227384567261,
      "learning_rate": 0.0003326226012793177,
      "loss": 2.4399871826171875,
      "step": 8904
    },
    {
      "epoch": 1.2253771765477481,
      "grad_norm": 0.04350428655743599,
      "learning_rate": 0.00033140420347243377,
      "loss": 2.4651107788085938,
      "step": 8906
    },
    {
      "epoch": 1.2256523623554987,
      "grad_norm": 0.030962686985731125,
      "learning_rate": 0.0003301858056655498,
      "loss": 2.4717912673950195,
      "step": 8908
    },
    {
      "epoch": 1.2259275481632494,
      "grad_norm": 0.03303976729512215,
      "learning_rate": 0.00032896740785866584,
      "loss": 2.4325356483459473,
      "step": 8910
    },
    {
      "epoch": 1.226202733971,
      "grad_norm": 0.029044678434729576,
      "learning_rate": 0.00032774901005178195,
      "loss": 2.4655752182006836,
      "step": 8912
    },
    {
      "epoch": 1.2264779197787505,
      "grad_norm": 0.026125498116016388,
      "learning_rate": 0.00032653061224489796,
      "loss": 2.4559173583984375,
      "step": 8914
    },
    {
      "epoch": 1.2267531055865013,
      "grad_norm": 0.028719931840896606,
      "learning_rate": 0.000325312214438014,
      "loss": 2.4583353996276855,
      "step": 8916
    },
    {
      "epoch": 1.2270282913942518,
      "grad_norm": 0.0288032665848732,
      "learning_rate": 0.0003240938166311301,
      "loss": 2.45771861076355,
      "step": 8918
    },
    {
      "epoch": 1.2273034772020024,
      "grad_norm": 0.02581830322742462,
      "learning_rate": 0.00032287541882424615,
      "loss": 2.4553146362304688,
      "step": 8920
    },
    {
      "epoch": 1.2275786630097532,
      "grad_norm": 0.025615032762289047,
      "learning_rate": 0.00032165702101736215,
      "loss": 2.5127549171447754,
      "step": 8922
    },
    {
      "epoch": 1.2278538488175037,
      "grad_norm": 0.031048277392983437,
      "learning_rate": 0.0003204386232104782,
      "loss": 2.4910147190093994,
      "step": 8924
    },
    {
      "epoch": 1.2281290346252542,
      "grad_norm": 0.04934019595384598,
      "learning_rate": 0.0003192202254035943,
      "loss": 2.4799563884735107,
      "step": 8926
    },
    {
      "epoch": 1.2284042204330048,
      "grad_norm": 0.06664245575666428,
      "learning_rate": 0.00031800182759671034,
      "loss": 2.4664297103881836,
      "step": 8928
    },
    {
      "epoch": 1.2286794062407556,
      "grad_norm": 0.03732762858271599,
      "learning_rate": 0.0003167834297898264,
      "loss": 2.407378673553467,
      "step": 8930
    },
    {
      "epoch": 1.228954592048506,
      "grad_norm": 0.02626616135239601,
      "learning_rate": 0.00031556503198294246,
      "loss": 2.4968457221984863,
      "step": 8932
    },
    {
      "epoch": 1.2292297778562566,
      "grad_norm": 0.02694445662200451,
      "learning_rate": 0.00031434663417605847,
      "loss": 2.4582109451293945,
      "step": 8934
    },
    {
      "epoch": 1.2295049636640072,
      "grad_norm": 0.02504737116396427,
      "learning_rate": 0.00031312823636917453,
      "loss": 2.4719767570495605,
      "step": 8936
    },
    {
      "epoch": 1.229780149471758,
      "grad_norm": 0.026110438629984856,
      "learning_rate": 0.0003119098385622906,
      "loss": 2.494311571121216,
      "step": 8938
    },
    {
      "epoch": 1.2300553352795085,
      "grad_norm": 0.026535863056778908,
      "learning_rate": 0.00031069144075540666,
      "loss": 2.4506266117095947,
      "step": 8940
    },
    {
      "epoch": 1.230330521087259,
      "grad_norm": 0.027939949184656143,
      "learning_rate": 0.00030947304294852266,
      "loss": 2.4675183296203613,
      "step": 8942
    },
    {
      "epoch": 1.2306057068950098,
      "grad_norm": 0.046726975589990616,
      "learning_rate": 0.0003082546451416388,
      "loss": 2.471871852874756,
      "step": 8944
    },
    {
      "epoch": 1.2308808927027604,
      "grad_norm": 0.025117380544543266,
      "learning_rate": 0.00030703624733475484,
      "loss": 2.445155143737793,
      "step": 8946
    },
    {
      "epoch": 1.231156078510511,
      "grad_norm": 0.0278630368411541,
      "learning_rate": 0.00030581784952787085,
      "loss": 2.4678473472595215,
      "step": 8948
    },
    {
      "epoch": 1.2314312643182617,
      "grad_norm": 0.024387259036302567,
      "learning_rate": 0.0003045994517209869,
      "loss": 2.422560214996338,
      "step": 8950
    },
    {
      "epoch": 1.2317064501260122,
      "grad_norm": 0.05300230532884598,
      "learning_rate": 0.000303381053914103,
      "loss": 2.4682836532592773,
      "step": 8952
    },
    {
      "epoch": 1.2319816359337628,
      "grad_norm": 0.051963262259960175,
      "learning_rate": 0.000302162656107219,
      "loss": 2.4685332775115967,
      "step": 8954
    },
    {
      "epoch": 1.2322568217415133,
      "grad_norm": 0.08862706273794174,
      "learning_rate": 0.00030094425830033504,
      "loss": 2.463040351867676,
      "step": 8956
    },
    {
      "epoch": 1.232532007549264,
      "grad_norm": 0.04717070609331131,
      "learning_rate": 0.0002997258604934511,
      "loss": 2.488790988922119,
      "step": 8958
    },
    {
      "epoch": 1.2328071933570146,
      "grad_norm": 0.02617405168712139,
      "learning_rate": 0.00029850746268656717,
      "loss": 2.4692492485046387,
      "step": 8960
    },
    {
      "epoch": 1.2330823791647652,
      "grad_norm": 0.03451040759682655,
      "learning_rate": 0.00029728906487968323,
      "loss": 2.460096597671509,
      "step": 8962
    },
    {
      "epoch": 1.2333575649725157,
      "grad_norm": 0.08512626588344574,
      "learning_rate": 0.0002960706670727993,
      "loss": 2.431614398956299,
      "step": 8964
    },
    {
      "epoch": 1.2336327507802665,
      "grad_norm": 0.04286133497953415,
      "learning_rate": 0.00029485226926591535,
      "loss": 2.4373245239257812,
      "step": 8966
    },
    {
      "epoch": 1.233907936588017,
      "grad_norm": 0.035752519965171814,
      "learning_rate": 0.00029363387145903136,
      "loss": 2.481111526489258,
      "step": 8968
    },
    {
      "epoch": 1.2341831223957676,
      "grad_norm": 0.09488311409950256,
      "learning_rate": 0.0002924154736521474,
      "loss": 2.484152317047119,
      "step": 8970
    },
    {
      "epoch": 1.2344583082035183,
      "grad_norm": 0.05059340223670006,
      "learning_rate": 0.0002911970758452635,
      "loss": 2.4567389488220215,
      "step": 8972
    },
    {
      "epoch": 1.2347334940112689,
      "grad_norm": 0.031188562512397766,
      "learning_rate": 0.0002899786780383795,
      "loss": 2.473804473876953,
      "step": 8974
    },
    {
      "epoch": 1.2350086798190194,
      "grad_norm": 0.062142908573150635,
      "learning_rate": 0.0002887602802314956,
      "loss": 2.452099084854126,
      "step": 8976
    },
    {
      "epoch": 1.2352838656267702,
      "grad_norm": 0.041879184544086456,
      "learning_rate": 0.00028754188242461167,
      "loss": 2.4886016845703125,
      "step": 8978
    },
    {
      "epoch": 1.2355590514345207,
      "grad_norm": 0.03959394246339798,
      "learning_rate": 0.0002863234846177277,
      "loss": 2.43761944770813,
      "step": 8980
    },
    {
      "epoch": 1.2358342372422713,
      "grad_norm": 0.07003889232873917,
      "learning_rate": 0.00028510508681084374,
      "loss": 2.4809274673461914,
      "step": 8982
    },
    {
      "epoch": 1.2361094230500218,
      "grad_norm": 0.02477721869945526,
      "learning_rate": 0.0002838866890039598,
      "loss": 2.4572484493255615,
      "step": 8984
    },
    {
      "epoch": 1.2363846088577726,
      "grad_norm": 0.03895578160881996,
      "learning_rate": 0.00028266829119707586,
      "loss": 2.492746114730835,
      "step": 8986
    },
    {
      "epoch": 1.2366597946655231,
      "grad_norm": 0.02589019387960434,
      "learning_rate": 0.00028144989339019187,
      "loss": 2.463683605194092,
      "step": 8988
    },
    {
      "epoch": 1.2369349804732737,
      "grad_norm": 0.024336805567145348,
      "learning_rate": 0.000280231495583308,
      "loss": 2.462240219116211,
      "step": 8990
    },
    {
      "epoch": 1.2372101662810244,
      "grad_norm": 0.02807413786649704,
      "learning_rate": 0.00027901309777642405,
      "loss": 2.4488985538482666,
      "step": 8992
    },
    {
      "epoch": 1.237485352088775,
      "grad_norm": 0.045868195593357086,
      "learning_rate": 0.00027779469996954006,
      "loss": 2.4374823570251465,
      "step": 8994
    },
    {
      "epoch": 1.2377605378965255,
      "grad_norm": 0.055717382580041885,
      "learning_rate": 0.0002765763021626561,
      "loss": 2.4435203075408936,
      "step": 8996
    },
    {
      "epoch": 1.238035723704276,
      "grad_norm": 0.0261841993778944,
      "learning_rate": 0.0002753579043557722,
      "loss": 2.413534641265869,
      "step": 8998
    },
    {
      "epoch": 1.2383109095120268,
      "grad_norm": 0.10789460688829422,
      "learning_rate": 0.0002741395065488882,
      "loss": 2.4650251865386963,
      "step": 9000
    },
    {
      "epoch": 1.2385860953197774,
      "grad_norm": 0.06245919689536095,
      "learning_rate": 0.00027292110874200425,
      "loss": 2.457040309906006,
      "step": 9002
    },
    {
      "epoch": 1.238861281127528,
      "grad_norm": 0.05550862103700638,
      "learning_rate": 0.0002717027109351203,
      "loss": 2.4677391052246094,
      "step": 9004
    },
    {
      "epoch": 1.2391364669352787,
      "grad_norm": 0.08197030425071716,
      "learning_rate": 0.0002704843131282364,
      "loss": 2.44098162651062,
      "step": 9006
    },
    {
      "epoch": 1.2394116527430292,
      "grad_norm": 0.04194927215576172,
      "learning_rate": 0.00026926591532135244,
      "loss": 2.474851608276367,
      "step": 9008
    },
    {
      "epoch": 1.2396868385507798,
      "grad_norm": 0.06186150759458542,
      "learning_rate": 0.0002680475175144685,
      "loss": 2.4432456493377686,
      "step": 9010
    },
    {
      "epoch": 1.2399620243585305,
      "grad_norm": 0.03969191387295723,
      "learning_rate": 0.00026682911970758456,
      "loss": 2.454441547393799,
      "step": 9012
    },
    {
      "epoch": 1.240237210166281,
      "grad_norm": 0.05759728327393532,
      "learning_rate": 0.00026561072190070057,
      "loss": 2.4498791694641113,
      "step": 9014
    },
    {
      "epoch": 1.2405123959740316,
      "grad_norm": 0.053908172994852066,
      "learning_rate": 0.00026439232409381663,
      "loss": 2.458129405975342,
      "step": 9016
    },
    {
      "epoch": 1.2407875817817822,
      "grad_norm": 0.02970687486231327,
      "learning_rate": 0.0002631739262869327,
      "loss": 2.470946788787842,
      "step": 9018
    },
    {
      "epoch": 1.241062767589533,
      "grad_norm": 0.0539206899702549,
      "learning_rate": 0.0002619555284800487,
      "loss": 2.4942121505737305,
      "step": 9020
    },
    {
      "epoch": 1.2413379533972835,
      "grad_norm": 0.037933267652988434,
      "learning_rate": 0.0002607371306731648,
      "loss": 2.4189348220825195,
      "step": 9022
    },
    {
      "epoch": 1.241613139205034,
      "grad_norm": 0.057484593242406845,
      "learning_rate": 0.0002595187328662809,
      "loss": 2.4525227546691895,
      "step": 9024
    },
    {
      "epoch": 1.2418883250127846,
      "grad_norm": 0.07437066733837128,
      "learning_rate": 0.0002583003350593969,
      "loss": 2.5115623474121094,
      "step": 9026
    },
    {
      "epoch": 1.2421635108205353,
      "grad_norm": 0.056315451860427856,
      "learning_rate": 0.00025708193725251295,
      "loss": 2.483595848083496,
      "step": 9028
    },
    {
      "epoch": 1.2424386966282859,
      "grad_norm": 0.07366979122161865,
      "learning_rate": 0.000255863539445629,
      "loss": 2.471569538116455,
      "step": 9030
    },
    {
      "epoch": 1.2427138824360364,
      "grad_norm": 0.026633277535438538,
      "learning_rate": 0.00025464514163874507,
      "loss": 2.462831974029541,
      "step": 9032
    },
    {
      "epoch": 1.2429890682437872,
      "grad_norm": 0.09097911417484283,
      "learning_rate": 0.0002534267438318611,
      "loss": 2.491210460662842,
      "step": 9034
    },
    {
      "epoch": 1.2432642540515377,
      "grad_norm": 0.04499977454543114,
      "learning_rate": 0.00025220834602497714,
      "loss": 2.476870059967041,
      "step": 9036
    },
    {
      "epoch": 1.2435394398592883,
      "grad_norm": 0.048204243183135986,
      "learning_rate": 0.00025098994821809326,
      "loss": 2.446970224380493,
      "step": 9038
    },
    {
      "epoch": 1.243814625667039,
      "grad_norm": 0.053671423345804214,
      "learning_rate": 0.00024977155041120926,
      "loss": 2.446554183959961,
      "step": 9040
    },
    {
      "epoch": 1.2440898114747896,
      "grad_norm": 0.028479335829615593,
      "learning_rate": 0.0002485531526043253,
      "loss": 2.4327917098999023,
      "step": 9042
    },
    {
      "epoch": 1.2443649972825401,
      "grad_norm": 0.04893416911363602,
      "learning_rate": 0.00024733475479744133,
      "loss": 2.496900796890259,
      "step": 9044
    },
    {
      "epoch": 1.2446401830902907,
      "grad_norm": 0.025767788290977478,
      "learning_rate": 0.00024611635699055745,
      "loss": 2.4836649894714355,
      "step": 9046
    },
    {
      "epoch": 1.2449153688980414,
      "grad_norm": 0.043032169342041016,
      "learning_rate": 0.00024489795918367346,
      "loss": 2.500868558883667,
      "step": 9048
    },
    {
      "epoch": 1.245190554705792,
      "grad_norm": 0.026030398905277252,
      "learning_rate": 0.00024367956137678955,
      "loss": 2.488234519958496,
      "step": 9050
    },
    {
      "epoch": 1.2454657405135425,
      "grad_norm": 0.03727688267827034,
      "learning_rate": 0.00024246116356990558,
      "loss": 2.4308507442474365,
      "step": 9052
    },
    {
      "epoch": 1.245740926321293,
      "grad_norm": 0.02399192936718464,
      "learning_rate": 0.00024124276576302162,
      "loss": 2.466092109680176,
      "step": 9054
    },
    {
      "epoch": 1.2460161121290438,
      "grad_norm": 0.04762476682662964,
      "learning_rate": 0.00024002436795613768,
      "loss": 2.4530653953552246,
      "step": 9056
    },
    {
      "epoch": 1.2462912979367944,
      "grad_norm": 0.03271552547812462,
      "learning_rate": 0.00023880597014925374,
      "loss": 2.4567389488220215,
      "step": 9058
    },
    {
      "epoch": 1.246566483744545,
      "grad_norm": 0.03467216715216637,
      "learning_rate": 0.0002375875723423698,
      "loss": 2.456214427947998,
      "step": 9060
    },
    {
      "epoch": 1.2468416695522957,
      "grad_norm": 0.030006764456629753,
      "learning_rate": 0.00023636917453548584,
      "loss": 2.4589505195617676,
      "step": 9062
    },
    {
      "epoch": 1.2471168553600462,
      "grad_norm": 0.02347988449037075,
      "learning_rate": 0.00023515077672860187,
      "loss": 2.4620018005371094,
      "step": 9064
    },
    {
      "epoch": 1.2473920411677968,
      "grad_norm": 0.024628883227705956,
      "learning_rate": 0.00023393237892171796,
      "loss": 2.435481548309326,
      "step": 9066
    },
    {
      "epoch": 1.2476672269755475,
      "grad_norm": 0.026944318786263466,
      "learning_rate": 0.000232713981114834,
      "loss": 2.4822263717651367,
      "step": 9068
    },
    {
      "epoch": 1.247942412783298,
      "grad_norm": 0.024742722511291504,
      "learning_rate": 0.00023149558330795006,
      "loss": 2.440927505493164,
      "step": 9070
    },
    {
      "epoch": 1.2482175985910486,
      "grad_norm": 0.025591425597667694,
      "learning_rate": 0.00023027718550106612,
      "loss": 2.4935126304626465,
      "step": 9072
    },
    {
      "epoch": 1.2484927843987992,
      "grad_norm": 0.030644886195659637,
      "learning_rate": 0.00022905878769418215,
      "loss": 2.4689183235168457,
      "step": 9074
    },
    {
      "epoch": 1.24876797020655,
      "grad_norm": 0.04450540617108345,
      "learning_rate": 0.00022784038988729821,
      "loss": 2.4379994869232178,
      "step": 9076
    },
    {
      "epoch": 1.2490431560143005,
      "grad_norm": 0.024654757231473923,
      "learning_rate": 0.00022662199208041425,
      "loss": 2.464127540588379,
      "step": 9078
    },
    {
      "epoch": 1.249318341822051,
      "grad_norm": 0.027313780039548874,
      "learning_rate": 0.0002254035942735303,
      "loss": 2.4595882892608643,
      "step": 9080
    },
    {
      "epoch": 1.2495935276298016,
      "grad_norm": 0.02862522378563881,
      "learning_rate": 0.00022418519646664637,
      "loss": 2.47308087348938,
      "step": 9082
    },
    {
      "epoch": 1.2498687134375523,
      "grad_norm": 0.03277771174907684,
      "learning_rate": 0.0002229667986597624,
      "loss": 2.4726290702819824,
      "step": 9084
    },
    {
      "epoch": 1.250143899245303,
      "grad_norm": 0.06404554843902588,
      "learning_rate": 0.00022174840085287847,
      "loss": 2.444108009338379,
      "step": 9086
    },
    {
      "epoch": 1.2504190850530534,
      "grad_norm": 0.034670136868953705,
      "learning_rate": 0.00022053000304599453,
      "loss": 2.4304516315460205,
      "step": 9088
    },
    {
      "epoch": 1.2506942708608042,
      "grad_norm": 0.03571346402168274,
      "learning_rate": 0.00021931160523911057,
      "loss": 2.4753894805908203,
      "step": 9090
    },
    {
      "epoch": 1.2509694566685547,
      "grad_norm": 0.024204323068261147,
      "learning_rate": 0.00021809320743222663,
      "loss": 2.479771137237549,
      "step": 9092
    },
    {
      "epoch": 1.2512446424763053,
      "grad_norm": 0.03631271794438362,
      "learning_rate": 0.00021687480962534266,
      "loss": 2.416874408721924,
      "step": 9094
    },
    {
      "epoch": 1.251519828284056,
      "grad_norm": 0.025105614215135574,
      "learning_rate": 0.00021565641181845875,
      "loss": 2.46626615524292,
      "step": 9096
    },
    {
      "epoch": 1.2517950140918066,
      "grad_norm": 0.043047938495874405,
      "learning_rate": 0.0002144380140115748,
      "loss": 2.462360382080078,
      "step": 9098
    },
    {
      "epoch": 1.2520701998995571,
      "grad_norm": 0.04422559216618538,
      "learning_rate": 0.00021321961620469082,
      "loss": 2.4511795043945312,
      "step": 9100
    },
    {
      "epoch": 1.252345385707308,
      "grad_norm": 0.02456490322947502,
      "learning_rate": 0.00021200121839780688,
      "loss": 2.463484764099121,
      "step": 9102
    },
    {
      "epoch": 1.2526205715150585,
      "grad_norm": 0.06656299531459808,
      "learning_rate": 0.00021078282059092295,
      "loss": 2.4093174934387207,
      "step": 9104
    },
    {
      "epoch": 1.252895757322809,
      "grad_norm": 0.07564622163772583,
      "learning_rate": 0.000209564422784039,
      "loss": 2.4552369117736816,
      "step": 9106
    },
    {
      "epoch": 1.2531709431305595,
      "grad_norm": 0.030012695118784904,
      "learning_rate": 0.00020834602497715504,
      "loss": 2.485116958618164,
      "step": 9108
    },
    {
      "epoch": 1.25344612893831,
      "grad_norm": 0.08046410232782364,
      "learning_rate": 0.00020712762717027108,
      "loss": 2.4423084259033203,
      "step": 9110
    },
    {
      "epoch": 1.2537213147460609,
      "grad_norm": 0.03732582926750183,
      "learning_rate": 0.00020590922936338717,
      "loss": 2.4297032356262207,
      "step": 9112
    },
    {
      "epoch": 1.2539965005538114,
      "grad_norm": 0.05407774820923805,
      "learning_rate": 0.0002046908315565032,
      "loss": 2.470935344696045,
      "step": 9114
    },
    {
      "epoch": 1.254271686361562,
      "grad_norm": 0.03944677114486694,
      "learning_rate": 0.00020347243374961926,
      "loss": 2.434739351272583,
      "step": 9116
    },
    {
      "epoch": 1.2545468721693127,
      "grad_norm": 0.05136119946837425,
      "learning_rate": 0.0002022540359427353,
      "loss": 2.4616265296936035,
      "step": 9118
    },
    {
      "epoch": 1.2548220579770633,
      "grad_norm": 0.04568967595696449,
      "learning_rate": 0.00020103563813585136,
      "loss": 2.468766689300537,
      "step": 9120
    },
    {
      "epoch": 1.2550972437848138,
      "grad_norm": 0.04253152385354042,
      "learning_rate": 0.00019981724032896742,
      "loss": 2.4550366401672363,
      "step": 9122
    },
    {
      "epoch": 1.2553724295925646,
      "grad_norm": 0.06559917330741882,
      "learning_rate": 0.00019859884252208346,
      "loss": 2.4733376502990723,
      "step": 9124
    },
    {
      "epoch": 1.255647615400315,
      "grad_norm": 0.024422714486718178,
      "learning_rate": 0.0001973804447151995,
      "loss": 2.4262566566467285,
      "step": 9126
    },
    {
      "epoch": 1.2559228012080657,
      "grad_norm": 0.024974917992949486,
      "learning_rate": 0.00019616204690831558,
      "loss": 2.4665987491607666,
      "step": 9128
    },
    {
      "epoch": 1.2561979870158164,
      "grad_norm": 0.037340566515922546,
      "learning_rate": 0.00019494364910143161,
      "loss": 2.4573168754577637,
      "step": 9130
    },
    {
      "epoch": 1.256473172823567,
      "grad_norm": 0.02619260549545288,
      "learning_rate": 0.00019372525129454768,
      "loss": 2.4704086780548096,
      "step": 9132
    },
    {
      "epoch": 1.2567483586313175,
      "grad_norm": 0.045244187116622925,
      "learning_rate": 0.0001925068534876637,
      "loss": 2.4865617752075195,
      "step": 9134
    },
    {
      "epoch": 1.257023544439068,
      "grad_norm": 0.022733744233846664,
      "learning_rate": 0.00019128845568077977,
      "loss": 2.471360921859741,
      "step": 9136
    },
    {
      "epoch": 1.2572987302468188,
      "grad_norm": 0.04515485838055611,
      "learning_rate": 0.00019007005787389584,
      "loss": 2.4244136810302734,
      "step": 9138
    },
    {
      "epoch": 1.2575739160545694,
      "grad_norm": 0.03306630998849869,
      "learning_rate": 0.00018885166006701187,
      "loss": 2.480607748031616,
      "step": 9140
    },
    {
      "epoch": 1.25784910186232,
      "grad_norm": 0.025314409285783768,
      "learning_rate": 0.00018763326226012796,
      "loss": 2.4583816528320312,
      "step": 9142
    },
    {
      "epoch": 1.2581242876700705,
      "grad_norm": 0.03562302887439728,
      "learning_rate": 0.000186414864453244,
      "loss": 2.4455204010009766,
      "step": 9144
    },
    {
      "epoch": 1.2583994734778212,
      "grad_norm": 0.02388559654355049,
      "learning_rate": 0.00018519646664636003,
      "loss": 2.4644341468811035,
      "step": 9146
    },
    {
      "epoch": 1.2586746592855718,
      "grad_norm": 0.02913445234298706,
      "learning_rate": 0.0001839780688394761,
      "loss": 2.4225234985351562,
      "step": 9148
    },
    {
      "epoch": 1.2589498450933223,
      "grad_norm": 0.030540654435753822,
      "learning_rate": 0.00018275967103259215,
      "loss": 2.4465856552124023,
      "step": 9150
    },
    {
      "epoch": 1.259225030901073,
      "grad_norm": 0.06099849194288254,
      "learning_rate": 0.00018154127322570821,
      "loss": 2.486354351043701,
      "step": 9152
    },
    {
      "epoch": 1.2595002167088236,
      "grad_norm": 0.04487714543938637,
      "learning_rate": 0.00018032287541882425,
      "loss": 2.4405717849731445,
      "step": 9154
    },
    {
      "epoch": 1.2597754025165742,
      "grad_norm": 0.026889760047197342,
      "learning_rate": 0.00017910447761194028,
      "loss": 2.429837226867676,
      "step": 9156
    },
    {
      "epoch": 1.260050588324325,
      "grad_norm": 0.023671148344874382,
      "learning_rate": 0.00017788607980505637,
      "loss": 2.514474391937256,
      "step": 9158
    },
    {
      "epoch": 1.2603257741320755,
      "grad_norm": 0.025050818920135498,
      "learning_rate": 0.0001766676819981724,
      "loss": 2.42932391166687,
      "step": 9160
    },
    {
      "epoch": 1.260600959939826,
      "grad_norm": 0.02603321336209774,
      "learning_rate": 0.00017544928419128847,
      "loss": 2.4727368354797363,
      "step": 9162
    },
    {
      "epoch": 1.2608761457475768,
      "grad_norm": 0.024021776393055916,
      "learning_rate": 0.0001742308863844045,
      "loss": 2.498426914215088,
      "step": 9164
    },
    {
      "epoch": 1.2611513315553273,
      "grad_norm": 0.026128318160772324,
      "learning_rate": 0.00017301248857752057,
      "loss": 2.4605154991149902,
      "step": 9166
    },
    {
      "epoch": 1.2614265173630779,
      "grad_norm": 0.03261302784085274,
      "learning_rate": 0.00017179409077063663,
      "loss": 2.476871967315674,
      "step": 9168
    },
    {
      "epoch": 1.2617017031708284,
      "grad_norm": 0.030772998929023743,
      "learning_rate": 0.00017057569296375266,
      "loss": 2.4540953636169434,
      "step": 9170
    },
    {
      "epoch": 1.261976888978579,
      "grad_norm": 0.06958162039518356,
      "learning_rate": 0.00016935729515686872,
      "loss": 2.463876962661743,
      "step": 9172
    },
    {
      "epoch": 1.2622520747863297,
      "grad_norm": 0.02849862352013588,
      "learning_rate": 0.0001681388973499848,
      "loss": 2.458806037902832,
      "step": 9174
    },
    {
      "epoch": 1.2625272605940803,
      "grad_norm": 0.08491968363523483,
      "learning_rate": 0.00016692049954310082,
      "loss": 2.4631829261779785,
      "step": 9176
    },
    {
      "epoch": 1.2628024464018308,
      "grad_norm": 0.03385651856660843,
      "learning_rate": 0.00016570210173621688,
      "loss": 2.426635980606079,
      "step": 9178
    },
    {
      "epoch": 1.2630776322095816,
      "grad_norm": 0.07140480726957321,
      "learning_rate": 0.00016448370392933292,
      "loss": 2.4901235103607178,
      "step": 9180
    },
    {
      "epoch": 1.2633528180173321,
      "grad_norm": 0.047666873782873154,
      "learning_rate": 0.00016326530612244898,
      "loss": 2.446725845336914,
      "step": 9182
    },
    {
      "epoch": 1.2636280038250827,
      "grad_norm": 0.03261895850300789,
      "learning_rate": 0.00016204690831556504,
      "loss": 2.4951424598693848,
      "step": 9184
    },
    {
      "epoch": 1.2639031896328334,
      "grad_norm": 0.030796566978096962,
      "learning_rate": 0.00016082851050868108,
      "loss": 2.44705867767334,
      "step": 9186
    },
    {
      "epoch": 1.264178375440584,
      "grad_norm": 0.0378551259636879,
      "learning_rate": 0.00015961011270179714,
      "loss": 2.4421229362487793,
      "step": 9188
    },
    {
      "epoch": 1.2644535612483345,
      "grad_norm": 0.023401297628879547,
      "learning_rate": 0.0001583917148949132,
      "loss": 2.484905242919922,
      "step": 9190
    },
    {
      "epoch": 1.2647287470560853,
      "grad_norm": 0.045987844467163086,
      "learning_rate": 0.00015717331708802924,
      "loss": 2.46445894241333,
      "step": 9192
    },
    {
      "epoch": 1.2650039328638358,
      "grad_norm": 0.037334635853767395,
      "learning_rate": 0.0001559549192811453,
      "loss": 2.473769187927246,
      "step": 9194
    },
    {
      "epoch": 1.2652791186715864,
      "grad_norm": 0.023572152480483055,
      "learning_rate": 0.00015473652147426133,
      "loss": 2.4687976837158203,
      "step": 9196
    },
    {
      "epoch": 1.265554304479337,
      "grad_norm": 0.025871429592370987,
      "learning_rate": 0.00015351812366737742,
      "loss": 2.4440529346466064,
      "step": 9198
    },
    {
      "epoch": 1.2658294902870875,
      "grad_norm": 0.02478305622935295,
      "learning_rate": 0.00015229972586049346,
      "loss": 2.4661784172058105,
      "step": 9200
    },
    {
      "epoch": 1.2661046760948382,
      "grad_norm": 0.03008429892361164,
      "learning_rate": 0.0001510813280536095,
      "loss": 2.4998819828033447,
      "step": 9202
    },
    {
      "epoch": 1.2663798619025888,
      "grad_norm": 0.02485612966120243,
      "learning_rate": 0.00014986293024672555,
      "loss": 2.4726576805114746,
      "step": 9204
    },
    {
      "epoch": 1.2666550477103393,
      "grad_norm": 0.03864492103457451,
      "learning_rate": 0.00014864453243984161,
      "loss": 2.4271836280822754,
      "step": 9206
    },
    {
      "epoch": 1.26693023351809,
      "grad_norm": 0.04841318726539612,
      "learning_rate": 0.00014742613463295768,
      "loss": 2.4692211151123047,
      "step": 9208
    },
    {
      "epoch": 1.2672054193258406,
      "grad_norm": 0.025785375386476517,
      "learning_rate": 0.0001462077368260737,
      "loss": 2.42588472366333,
      "step": 9210
    },
    {
      "epoch": 1.2674806051335912,
      "grad_norm": 0.026831820607185364,
      "learning_rate": 0.00014498933901918975,
      "loss": 2.4767918586730957,
      "step": 9212
    },
    {
      "epoch": 1.267755790941342,
      "grad_norm": 0.024084411561489105,
      "learning_rate": 0.00014377094121230583,
      "loss": 2.467566967010498,
      "step": 9214
    },
    {
      "epoch": 1.2680309767490925,
      "grad_norm": 0.024515077471733093,
      "learning_rate": 0.00014255254340542187,
      "loss": 2.482212543487549,
      "step": 9216
    },
    {
      "epoch": 1.268306162556843,
      "grad_norm": 0.03350868821144104,
      "learning_rate": 0.00014133414559853793,
      "loss": 2.5129265785217285,
      "step": 9218
    },
    {
      "epoch": 1.2685813483645938,
      "grad_norm": 0.0381435826420784,
      "learning_rate": 0.000140115747791654,
      "loss": 2.484919548034668,
      "step": 9220
    },
    {
      "epoch": 1.2688565341723443,
      "grad_norm": 0.028340229764580727,
      "learning_rate": 0.00013889734998477003,
      "loss": 2.500598430633545,
      "step": 9222
    },
    {
      "epoch": 1.2691317199800949,
      "grad_norm": 0.03274098411202431,
      "learning_rate": 0.0001376789521778861,
      "loss": 2.5282845497131348,
      "step": 9224
    },
    {
      "epoch": 1.2694069057878454,
      "grad_norm": 0.03800109773874283,
      "learning_rate": 0.00013646055437100213,
      "loss": 2.450237274169922,
      "step": 9226
    },
    {
      "epoch": 1.269682091595596,
      "grad_norm": 0.02317206934094429,
      "learning_rate": 0.0001352421565641182,
      "loss": 2.4409570693969727,
      "step": 9228
    },
    {
      "epoch": 1.2699572774033467,
      "grad_norm": 0.04322657361626625,
      "learning_rate": 0.00013402375875723425,
      "loss": 2.4901251792907715,
      "step": 9230
    },
    {
      "epoch": 1.2702324632110973,
      "grad_norm": 0.026108358055353165,
      "learning_rate": 0.00013280536095035028,
      "loss": 2.4575915336608887,
      "step": 9232
    },
    {
      "epoch": 1.2705076490188478,
      "grad_norm": 0.023000361397862434,
      "learning_rate": 0.00013158696314346635,
      "loss": 2.4791340827941895,
      "step": 9234
    },
    {
      "epoch": 1.2707828348265986,
      "grad_norm": 0.024312103167176247,
      "learning_rate": 0.0001303685653365824,
      "loss": 2.484363079071045,
      "step": 9236
    },
    {
      "epoch": 1.2710580206343491,
      "grad_norm": 0.04383526369929314,
      "learning_rate": 0.00012915016752969844,
      "loss": 2.4615278244018555,
      "step": 9238
    },
    {
      "epoch": 1.2713332064420997,
      "grad_norm": 0.05114690586924553,
      "learning_rate": 0.0001279317697228145,
      "loss": 2.4673848152160645,
      "step": 9240
    },
    {
      "epoch": 1.2716083922498504,
      "grad_norm": 0.023235918954014778,
      "learning_rate": 0.00012671337191593054,
      "loss": 2.4590530395507812,
      "step": 9242
    },
    {
      "epoch": 1.271883578057601,
      "grad_norm": 0.0435694120824337,
      "learning_rate": 0.00012549497410904663,
      "loss": 2.466273307800293,
      "step": 9244
    },
    {
      "epoch": 1.2721587638653515,
      "grad_norm": 0.024943910539150238,
      "learning_rate": 0.00012427657630216266,
      "loss": 2.4613115787506104,
      "step": 9246
    },
    {
      "epoch": 1.2724339496731023,
      "grad_norm": 0.032235994935035706,
      "learning_rate": 0.00012305817849527872,
      "loss": 2.4269895553588867,
      "step": 9248
    },
    {
      "epoch": 1.2727091354808528,
      "grad_norm": 0.03350569307804108,
      "learning_rate": 0.00012183978068839477,
      "loss": 2.476099967956543,
      "step": 9250
    },
    {
      "epoch": 1.2729843212886034,
      "grad_norm": 0.032456789165735245,
      "learning_rate": 0.00012062138288151081,
      "loss": 2.4684367179870605,
      "step": 9252
    },
    {
      "epoch": 1.273259507096354,
      "grad_norm": 0.03041711077094078,
      "learning_rate": 0.00011940298507462687,
      "loss": 2.422231912612915,
      "step": 9254
    },
    {
      "epoch": 1.2735346929041047,
      "grad_norm": 0.04057612270116806,
      "learning_rate": 0.00011818458726774292,
      "loss": 2.4237263202667236,
      "step": 9256
    },
    {
      "epoch": 1.2738098787118552,
      "grad_norm": 0.02729097753763199,
      "learning_rate": 0.00011696618946085898,
      "loss": 2.469296932220459,
      "step": 9258
    },
    {
      "epoch": 1.2740850645196058,
      "grad_norm": 0.025067254900932312,
      "learning_rate": 0.00011574779165397503,
      "loss": 2.4658350944519043,
      "step": 9260
    },
    {
      "epoch": 1.2743602503273563,
      "grad_norm": 0.023453684523701668,
      "learning_rate": 0.00011452939384709108,
      "loss": 2.473017692565918,
      "step": 9262
    },
    {
      "epoch": 1.274635436135107,
      "grad_norm": 0.02786240167915821,
      "learning_rate": 0.00011331099604020712,
      "loss": 2.438194751739502,
      "step": 9264
    },
    {
      "epoch": 1.2749106219428576,
      "grad_norm": 0.025899238884449005,
      "learning_rate": 0.00011209259823332319,
      "loss": 2.4736762046813965,
      "step": 9266
    },
    {
      "epoch": 1.2751858077506082,
      "grad_norm": 0.04977986589074135,
      "learning_rate": 0.00011087420042643924,
      "loss": 2.4508111476898193,
      "step": 9268
    },
    {
      "epoch": 1.275460993558359,
      "grad_norm": 0.043027158826589584,
      "learning_rate": 0.00010965580261955528,
      "loss": 2.4447317123413086,
      "step": 9270
    },
    {
      "epoch": 1.2757361793661095,
      "grad_norm": 0.025275692343711853,
      "learning_rate": 0.00010843740481267133,
      "loss": 2.4406440258026123,
      "step": 9272
    },
    {
      "epoch": 1.27601136517386,
      "grad_norm": 0.02418424002826214,
      "learning_rate": 0.0001072190070057874,
      "loss": 2.4242236614227295,
      "step": 9274
    },
    {
      "epoch": 1.2762865509816108,
      "grad_norm": 0.04770015925168991,
      "learning_rate": 0.00010600060919890344,
      "loss": 2.4857683181762695,
      "step": 9276
    },
    {
      "epoch": 1.2765617367893614,
      "grad_norm": 0.026742637157440186,
      "learning_rate": 0.0001047822113920195,
      "loss": 2.469913959503174,
      "step": 9278
    },
    {
      "epoch": 1.276836922597112,
      "grad_norm": 0.041335515677928925,
      "learning_rate": 0.00010356381358513554,
      "loss": 2.449284553527832,
      "step": 9280
    },
    {
      "epoch": 1.2771121084048627,
      "grad_norm": 0.023468248546123505,
      "learning_rate": 0.0001023454157782516,
      "loss": 2.4724225997924805,
      "step": 9282
    },
    {
      "epoch": 1.2773872942126132,
      "grad_norm": 0.04042794927954674,
      "learning_rate": 0.00010112701797136765,
      "loss": 2.5269744396209717,
      "step": 9284
    },
    {
      "epoch": 1.2776624800203638,
      "grad_norm": 0.028787851333618164,
      "learning_rate": 9.990862016448371e-05,
      "loss": 2.4344067573547363,
      "step": 9286
    },
    {
      "epoch": 1.2779376658281143,
      "grad_norm": 0.023643752560019493,
      "learning_rate": 9.869022235759975e-05,
      "loss": 2.445364475250244,
      "step": 9288
    },
    {
      "epoch": 1.2782128516358648,
      "grad_norm": 0.024720333516597748,
      "learning_rate": 9.747182455071581e-05,
      "loss": 2.4627742767333984,
      "step": 9290
    },
    {
      "epoch": 1.2784880374436156,
      "grad_norm": 0.02406024932861328,
      "learning_rate": 9.625342674383186e-05,
      "loss": 2.4599733352661133,
      "step": 9292
    },
    {
      "epoch": 1.2787632232513662,
      "grad_norm": 0.035563528537750244,
      "learning_rate": 9.503502893694792e-05,
      "loss": 2.466968536376953,
      "step": 9294
    },
    {
      "epoch": 1.2790384090591167,
      "grad_norm": 0.0645768791437149,
      "learning_rate": 9.381663113006398e-05,
      "loss": 2.5136094093322754,
      "step": 9296
    },
    {
      "epoch": 1.2793135948668675,
      "grad_norm": 0.02829975076019764,
      "learning_rate": 9.259823332318001e-05,
      "loss": 2.456143856048584,
      "step": 9298
    },
    {
      "epoch": 1.279588780674618,
      "grad_norm": 0.0350259393453598,
      "learning_rate": 9.137983551629608e-05,
      "loss": 2.4724559783935547,
      "step": 9300
    },
    {
      "epoch": 1.2798639664823686,
      "grad_norm": 0.02330256812274456,
      "learning_rate": 9.016143770941212e-05,
      "loss": 2.4723682403564453,
      "step": 9302
    },
    {
      "epoch": 1.2801391522901193,
      "grad_norm": 0.06514086574316025,
      "learning_rate": 8.894303990252819e-05,
      "loss": 2.441443920135498,
      "step": 9304
    },
    {
      "epoch": 1.2804143380978699,
      "grad_norm": 0.02913660928606987,
      "learning_rate": 8.772464209564423e-05,
      "loss": 2.4984564781188965,
      "step": 9306
    },
    {
      "epoch": 1.2806895239056204,
      "grad_norm": 0.04519667103886604,
      "learning_rate": 8.650624428876028e-05,
      "loss": 2.4247121810913086,
      "step": 9308
    },
    {
      "epoch": 1.2809647097133712,
      "grad_norm": 0.030100347474217415,
      "learning_rate": 8.528784648187633e-05,
      "loss": 2.4700722694396973,
      "step": 9310
    },
    {
      "epoch": 1.2812398955211217,
      "grad_norm": 0.04937539994716644,
      "learning_rate": 8.40694486749924e-05,
      "loss": 2.442837715148926,
      "step": 9312
    },
    {
      "epoch": 1.2815150813288723,
      "grad_norm": 0.03726276010274887,
      "learning_rate": 8.285105086810844e-05,
      "loss": 2.4769034385681152,
      "step": 9314
    },
    {
      "epoch": 1.2817902671366228,
      "grad_norm": 0.025645513087511063,
      "learning_rate": 8.163265306122449e-05,
      "loss": 2.41886043548584,
      "step": 9316
    },
    {
      "epoch": 1.2820654529443734,
      "grad_norm": 0.059149716049432755,
      "learning_rate": 8.041425525434054e-05,
      "loss": 2.510690212249756,
      "step": 9318
    },
    {
      "epoch": 1.2823406387521241,
      "grad_norm": 0.024304525926709175,
      "learning_rate": 7.91958574474566e-05,
      "loss": 2.474242687225342,
      "step": 9320
    },
    {
      "epoch": 1.2826158245598747,
      "grad_norm": 0.02650417573750019,
      "learning_rate": 7.797745964057265e-05,
      "loss": 2.4739956855773926,
      "step": 9322
    },
    {
      "epoch": 1.2828910103676252,
      "grad_norm": 0.024143308401107788,
      "learning_rate": 7.675906183368871e-05,
      "loss": 2.464813709259033,
      "step": 9324
    },
    {
      "epoch": 1.283166196175376,
      "grad_norm": 0.02908219024538994,
      "learning_rate": 7.554066402680475e-05,
      "loss": 2.4559426307678223,
      "step": 9326
    },
    {
      "epoch": 1.2834413819831265,
      "grad_norm": 0.028742872178554535,
      "learning_rate": 7.432226621992081e-05,
      "loss": 2.483405828475952,
      "step": 9328
    },
    {
      "epoch": 1.283716567790877,
      "grad_norm": 0.025400971993803978,
      "learning_rate": 7.310386841303686e-05,
      "loss": 2.494812488555908,
      "step": 9330
    },
    {
      "epoch": 1.2839917535986278,
      "grad_norm": 0.03265144303441048,
      "learning_rate": 7.188547060615292e-05,
      "loss": 2.4260623455047607,
      "step": 9332
    },
    {
      "epoch": 1.2842669394063784,
      "grad_norm": 0.02289043366909027,
      "learning_rate": 7.066707279926897e-05,
      "loss": 2.4652228355407715,
      "step": 9334
    },
    {
      "epoch": 1.284542125214129,
      "grad_norm": 0.024191470816731453,
      "learning_rate": 6.944867499238501e-05,
      "loss": 2.534149646759033,
      "step": 9336
    },
    {
      "epoch": 1.2848173110218797,
      "grad_norm": 0.02270405925810337,
      "learning_rate": 6.823027718550106e-05,
      "loss": 2.442457675933838,
      "step": 9338
    },
    {
      "epoch": 1.2850924968296302,
      "grad_norm": 0.02323989011347294,
      "learning_rate": 6.701187937861712e-05,
      "loss": 2.4669671058654785,
      "step": 9340
    },
    {
      "epoch": 1.2853676826373808,
      "grad_norm": 0.026859449222683907,
      "learning_rate": 6.579348157173317e-05,
      "loss": 2.480384349822998,
      "step": 9342
    },
    {
      "epoch": 1.2856428684451313,
      "grad_norm": 0.029326442629098892,
      "learning_rate": 6.457508376484922e-05,
      "loss": 2.460057020187378,
      "step": 9344
    },
    {
      "epoch": 1.285918054252882,
      "grad_norm": 0.022145941853523254,
      "learning_rate": 6.335668595796527e-05,
      "loss": 2.431990146636963,
      "step": 9346
    },
    {
      "epoch": 1.2861932400606326,
      "grad_norm": 0.026378050446510315,
      "learning_rate": 6.213828815108133e-05,
      "loss": 2.4867806434631348,
      "step": 9348
    },
    {
      "epoch": 1.2864684258683832,
      "grad_norm": 0.03834126144647598,
      "learning_rate": 6.0919890344197386e-05,
      "loss": 2.4674935340881348,
      "step": 9350
    },
    {
      "epoch": 1.2867436116761337,
      "grad_norm": 0.027719242498278618,
      "learning_rate": 5.9701492537313435e-05,
      "loss": 2.4642841815948486,
      "step": 9352
    },
    {
      "epoch": 1.2870187974838845,
      "grad_norm": 0.031444255262613297,
      "learning_rate": 5.848309473042949e-05,
      "loss": 2.471680164337158,
      "step": 9354
    },
    {
      "epoch": 1.287293983291635,
      "grad_norm": 0.03111363761126995,
      "learning_rate": 5.726469692354554e-05,
      "loss": 2.4536776542663574,
      "step": 9356
    },
    {
      "epoch": 1.2875691690993856,
      "grad_norm": 0.02274118922650814,
      "learning_rate": 5.604629911666159e-05,
      "loss": 2.437633991241455,
      "step": 9358
    },
    {
      "epoch": 1.2878443549071363,
      "grad_norm": 0.032064806669950485,
      "learning_rate": 5.482790130977764e-05,
      "loss": 2.4505927562713623,
      "step": 9360
    },
    {
      "epoch": 1.2881195407148869,
      "grad_norm": 0.022799208760261536,
      "learning_rate": 5.36095035028937e-05,
      "loss": 2.4661970138549805,
      "step": 9362
    },
    {
      "epoch": 1.2883947265226374,
      "grad_norm": 0.03112195059657097,
      "learning_rate": 5.239110569600975e-05,
      "loss": 2.4658312797546387,
      "step": 9364
    },
    {
      "epoch": 1.2886699123303882,
      "grad_norm": 0.057028234004974365,
      "learning_rate": 5.11727078891258e-05,
      "loss": 2.4279375076293945,
      "step": 9366
    },
    {
      "epoch": 1.2889450981381387,
      "grad_norm": 0.051933858543634415,
      "learning_rate": 4.9954310082241855e-05,
      "loss": 2.464329242706299,
      "step": 9368
    },
    {
      "epoch": 1.2892202839458893,
      "grad_norm": 0.02488640509545803,
      "learning_rate": 4.8735912275357904e-05,
      "loss": 2.4464011192321777,
      "step": 9370
    },
    {
      "epoch": 1.2894954697536398,
      "grad_norm": 0.033568259328603745,
      "learning_rate": 4.751751446847396e-05,
      "loss": 2.4866280555725098,
      "step": 9372
    },
    {
      "epoch": 1.2897706555613906,
      "grad_norm": 0.03608524054288864,
      "learning_rate": 4.629911666159001e-05,
      "loss": 2.47916316986084,
      "step": 9374
    },
    {
      "epoch": 1.2900458413691411,
      "grad_norm": 0.029297254979610443,
      "learning_rate": 4.508071885470606e-05,
      "loss": 2.450876235961914,
      "step": 9376
    },
    {
      "epoch": 1.2903210271768917,
      "grad_norm": 0.02411983162164688,
      "learning_rate": 4.386232104782212e-05,
      "loss": 2.4418458938598633,
      "step": 9378
    },
    {
      "epoch": 1.2905962129846422,
      "grad_norm": 0.028337502852082253,
      "learning_rate": 4.2643923240938166e-05,
      "loss": 2.466611623764038,
      "step": 9380
    },
    {
      "epoch": 1.290871398792393,
      "grad_norm": 0.023473840206861496,
      "learning_rate": 4.142552543405422e-05,
      "loss": 2.4249253273010254,
      "step": 9382
    },
    {
      "epoch": 1.2911465846001435,
      "grad_norm": 0.026947252452373505,
      "learning_rate": 4.020712762717027e-05,
      "loss": 2.5074667930603027,
      "step": 9384
    },
    {
      "epoch": 1.291421770407894,
      "grad_norm": 0.024754062294960022,
      "learning_rate": 3.8988729820286324e-05,
      "loss": 2.4222748279571533,
      "step": 9386
    },
    {
      "epoch": 1.2916969562156448,
      "grad_norm": 0.023064929991960526,
      "learning_rate": 3.777033201340237e-05,
      "loss": 2.486654758453369,
      "step": 9388
    },
    {
      "epoch": 1.2919721420233954,
      "grad_norm": 0.02911621704697609,
      "learning_rate": 3.655193420651843e-05,
      "loss": 2.4243130683898926,
      "step": 9390
    },
    {
      "epoch": 1.292247327831146,
      "grad_norm": 0.02474200911819935,
      "learning_rate": 3.533353639963448e-05,
      "loss": 2.4485673904418945,
      "step": 9392
    },
    {
      "epoch": 1.2925225136388967,
      "grad_norm": 0.02235531620681286,
      "learning_rate": 3.411513859275053e-05,
      "loss": 2.4671630859375,
      "step": 9394
    },
    {
      "epoch": 1.2927976994466472,
      "grad_norm": 0.025915909558534622,
      "learning_rate": 3.2896740785866586e-05,
      "loss": 2.4600918292999268,
      "step": 9396
    },
    {
      "epoch": 1.2930728852543978,
      "grad_norm": 0.024452533572912216,
      "learning_rate": 3.1678342978982635e-05,
      "loss": 2.438499927520752,
      "step": 9398
    },
    {
      "epoch": 1.2933480710621486,
      "grad_norm": 0.022966155782341957,
      "learning_rate": 3.0459945172098693e-05,
      "loss": 2.444155216217041,
      "step": 9400
    },
    {
      "epoch": 1.293623256869899,
      "grad_norm": 0.023288989439606667,
      "learning_rate": 2.9241547365214745e-05,
      "loss": 2.4615085124969482,
      "step": 9402
    },
    {
      "epoch": 1.2938984426776496,
      "grad_norm": 0.02599608525633812,
      "learning_rate": 2.8023149558330797e-05,
      "loss": 2.471071481704712,
      "step": 9404
    },
    {
      "epoch": 1.2941736284854002,
      "grad_norm": 0.023484863340854645,
      "learning_rate": 2.680475175144685e-05,
      "loss": 2.492124557495117,
      "step": 9406
    },
    {
      "epoch": 1.2944488142931507,
      "grad_norm": 0.02862633764743805,
      "learning_rate": 2.55863539445629e-05,
      "loss": 2.4857335090637207,
      "step": 9408
    },
    {
      "epoch": 1.2947240001009015,
      "grad_norm": 0.023178761824965477,
      "learning_rate": 2.4367956137678952e-05,
      "loss": 2.4644484519958496,
      "step": 9410
    },
    {
      "epoch": 1.294999185908652,
      "grad_norm": 0.024128887802362442,
      "learning_rate": 2.3149558330795004e-05,
      "loss": 2.4657139778137207,
      "step": 9412
    },
    {
      "epoch": 1.2952743717164026,
      "grad_norm": 0.02320338971912861,
      "learning_rate": 2.193116052391106e-05,
      "loss": 2.43255877494812,
      "step": 9414
    },
    {
      "epoch": 1.2955495575241534,
      "grad_norm": 0.027649955824017525,
      "learning_rate": 2.071276271702711e-05,
      "loss": 2.4375805854797363,
      "step": 9416
    },
    {
      "epoch": 1.295824743331904,
      "grad_norm": 0.028232116252183914,
      "learning_rate": 1.9494364910143162e-05,
      "loss": 2.473940849304199,
      "step": 9418
    },
    {
      "epoch": 1.2960999291396544,
      "grad_norm": 0.028577309101819992,
      "learning_rate": 1.8275967103259214e-05,
      "loss": 2.453368663787842,
      "step": 9420
    },
    {
      "epoch": 1.2963751149474052,
      "grad_norm": 0.027163874357938766,
      "learning_rate": 1.7057569296375266e-05,
      "loss": 2.521793842315674,
      "step": 9422
    },
    {
      "epoch": 1.2966503007551557,
      "grad_norm": 0.02428358420729637,
      "learning_rate": 1.5839171489491317e-05,
      "loss": 2.457825183868408,
      "step": 9424
    },
    {
      "epoch": 1.2969254865629063,
      "grad_norm": 0.030103236436843872,
      "learning_rate": 1.4620773682607372e-05,
      "loss": 2.4823217391967773,
      "step": 9426
    },
    {
      "epoch": 1.297200672370657,
      "grad_norm": 0.023016395047307014,
      "learning_rate": 1.3402375875723424e-05,
      "loss": 2.475515842437744,
      "step": 9428
    },
    {
      "epoch": 1.2974758581784076,
      "grad_norm": 0.02249881997704506,
      "learning_rate": 1.2183978068839476e-05,
      "loss": 2.429299831390381,
      "step": 9430
    },
    {
      "epoch": 1.2977510439861581,
      "grad_norm": 0.02431398257613182,
      "learning_rate": 1.096558026195553e-05,
      "loss": 2.4118399620056152,
      "step": 9432
    },
    {
      "epoch": 1.2980262297939087,
      "grad_norm": 0.023900052532553673,
      "learning_rate": 9.747182455071581e-06,
      "loss": 2.4317069053649902,
      "step": 9434
    },
    {
      "epoch": 1.2983014156016592,
      "grad_norm": 0.022661179304122925,
      "learning_rate": 8.528784648187633e-06,
      "loss": 2.464376926422119,
      "step": 9436
    },
    {
      "epoch": 1.29857660140941,
      "grad_norm": 0.02387865260243416,
      "learning_rate": 7.310386841303686e-06,
      "loss": 2.4512503147125244,
      "step": 9438
    },
    {
      "epoch": 1.2988517872171605,
      "grad_norm": 0.025546187534928322,
      "learning_rate": 6.091989034419738e-06,
      "loss": 2.4238765239715576,
      "step": 9440
    },
    {
      "epoch": 1.299126973024911,
      "grad_norm": 0.025863472372293472,
      "learning_rate": 4.8735912275357905e-06,
      "loss": 2.452146053314209,
      "step": 9442
    },
    {
      "epoch": 1.2994021588326619,
      "grad_norm": 0.03757040947675705,
      "learning_rate": 3.655193420651843e-06,
      "loss": 2.4905474185943604,
      "step": 9444
    },
    {
      "epoch": 1.2996773446404124,
      "grad_norm": 0.023091528564691544,
      "learning_rate": 2.4367956137678953e-06,
      "loss": 2.4607601165771484,
      "step": 9446
    },
    {
      "epoch": 1.299952530448163,
      "grad_norm": 0.026085583493113518,
      "learning_rate": 1.2183978068839476e-06,
      "loss": 2.480759620666504,
      "step": 9448
    },
    {
      "epoch": 1.3000901233520383,
      "eval_loss": 2.497500419616699,
      "eval_runtime": 379.6263,
      "eval_samples_per_second": 92.636,
      "eval_steps_per_second": 11.58,
      "step": 9449
    }
  ],
  "logging_steps": 2,
  "max_steps": 9449,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 945,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.536902454968425e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}