{
  "best_global_step": 2148,
  "best_metric": 0.06230008,
  "best_model_checkpoint": "/home/mithil/PycharmProjects/HebrewSearch/output/Qwen3-Embedding-/v10-20250927-032234/checkpoint-2148",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 2148,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0009315323707498836,
      "grad_norm": 5.125,
      "learning_rate": 5.999996791349807e-05,
      "loss": 0.22629377245903015,
      "step": 1
    },
    {
      "epoch": 0.004657661853749418,
      "grad_norm": 2.734375,
      "learning_rate": 5.999919784088342e-05,
      "loss": 0.11207550764083862,
      "step": 5
    },
    {
      "epoch": 0.009315323707498836,
      "grad_norm": 2.328125,
      "learning_rate": 5.999679140643096e-05,
      "loss": 0.1241973876953125,
      "step": 10
    },
    {
      "epoch": 0.013972985561248253,
      "grad_norm": 1.0,
      "learning_rate": 5.999278082533218e-05,
      "loss": 0.08484302759170533,
      "step": 15
    },
    {
      "epoch": 0.018630647414997672,
      "grad_norm": 1.1484375,
      "learning_rate": 5.998716631206202e-05,
      "loss": 0.11840656995773316,
      "step": 20
    },
    {
      "epoch": 0.02328830926874709,
      "grad_norm": 1.3125,
      "learning_rate": 5.997994816686935e-05,
      "loss": 0.10533121824264527,
      "step": 25
    },
    {
      "epoch": 0.027945971122496506,
      "grad_norm": 1.96875,
      "learning_rate": 5.997112677576091e-05,
      "loss": 0.07531467080116272,
      "step": 30
    },
    {
      "epoch": 0.032603632976245925,
      "grad_norm": 1.4765625,
      "learning_rate": 5.9960702610480635e-05,
      "loss": 0.08655669689178466,
      "step": 35
    },
    {
      "epoch": 0.037261294829995344,
      "grad_norm": 1.140625,
      "learning_rate": 5.994867622848448e-05,
      "loss": 0.08251949548721313,
      "step": 40
    },
    {
      "epoch": 0.04191895668374476,
      "grad_norm": 1.3515625,
      "learning_rate": 5.993504827291059e-05,
      "loss": 0.07267643213272094,
      "step": 45
    },
    {
      "epoch": 0.04657661853749418,
      "grad_norm": 1.0078125,
      "learning_rate": 5.991981947254487e-05,
      "loss": 0.08904569149017334,
      "step": 50
    },
    {
      "epoch": 0.05123428039124359,
      "grad_norm": 1.4140625,
      "learning_rate": 5.990299064178205e-05,
      "loss": 0.06884298324584961,
      "step": 55
    },
    {
      "epoch": 0.05589194224499301,
      "grad_norm": 1.7734375,
      "learning_rate": 5.9884562680582146e-05,
      "loss": 0.08851982355117798,
      "step": 60
    },
    {
      "epoch": 0.06054960409874243,
      "grad_norm": 4.4375,
      "learning_rate": 5.98645365744223e-05,
      "loss": 0.08651072978973388,
      "step": 65
    },
    {
      "epoch": 0.06520726595249185,
      "grad_norm": 1.0859375,
      "learning_rate": 5.9842913394244063e-05,
      "loss": 0.07889164686203003,
      "step": 70
    },
    {
      "epoch": 0.06986492780624126,
      "grad_norm": 1.34375,
      "learning_rate": 5.9819694296396204e-05,
      "loss": 0.0799069881439209,
      "step": 75
    },
    {
      "epoch": 0.07452258965999069,
      "grad_norm": 1.890625,
      "learning_rate": 5.979488052257277e-05,
      "loss": 0.10144302845001221,
      "step": 80
    },
    {
      "epoch": 0.0791802515137401,
      "grad_norm": 1.1953125,
      "learning_rate": 5.9768473399746766e-05,
      "loss": 0.07646682262420654,
      "step": 85
    },
    {
      "epoch": 0.08383791336748952,
      "grad_norm": 1.6640625,
      "learning_rate": 5.974047434009914e-05,
      "loss": 0.08959404826164245,
      "step": 90
    },
    {
      "epoch": 0.08849557522123894,
      "grad_norm": 1.1796875,
      "learning_rate": 5.9710884840943296e-05,
      "loss": 0.07164722084999084,
      "step": 95
    },
    {
      "epoch": 0.09315323707498836,
      "grad_norm": 1.359375,
      "learning_rate": 5.967970648464499e-05,
      "loss": 0.07039438486099243,
      "step": 100
    },
    {
      "epoch": 0.09781089892873777,
      "grad_norm": 1.1953125,
      "learning_rate": 5.964694093853774e-05,
      "loss": 0.06741951704025269,
      "step": 105
    },
    {
      "epoch": 0.10246856078248719,
      "grad_norm": 1.0625,
      "learning_rate": 5.9612589954833654e-05,
      "loss": 0.059288203716278076,
      "step": 110
    },
    {
      "epoch": 0.10712622263623661,
      "grad_norm": 2.28125,
      "learning_rate": 5.957665537052972e-05,
      "loss": 0.08943802118301392,
      "step": 115
    },
    {
      "epoch": 0.11178388448998602,
      "grad_norm": 1.75,
      "learning_rate": 5.953913910730955e-05,
      "loss": 0.0767773985862732,
      "step": 120
    },
    {
      "epoch": 0.11644154634373545,
      "grad_norm": 1.625,
      "learning_rate": 5.9500043171440655e-05,
      "loss": 0.061524766683578494,
      "step": 125
    },
    {
      "epoch": 0.12109920819748486,
      "grad_norm": 1.859375,
      "learning_rate": 5.9459369653667134e-05,
      "loss": 0.10620454549789429,
      "step": 130
    },
    {
      "epoch": 0.1257568700512343,
      "grad_norm": 1.4140625,
      "learning_rate": 5.9417120729097845e-05,
      "loss": 0.04981007874011993,
      "step": 135
    },
    {
      "epoch": 0.1304145319049837,
      "grad_norm": 2.1875,
      "learning_rate": 5.937329865709012e-05,
      "loss": 0.06490777730941773,
      "step": 140
    },
    {
      "epoch": 0.1350721937587331,
      "grad_norm": 1.375,
      "learning_rate": 5.932790578112896e-05,
      "loss": 0.06334985494613647,
      "step": 145
    },
    {
      "epoch": 0.13972985561248252,
      "grad_norm": 1.078125,
      "learning_rate": 5.928094452870161e-05,
      "loss": 0.07352197170257568,
      "step": 150
    },
    {
      "epoch": 0.14438751746623196,
      "grad_norm": 1.75,
      "learning_rate": 5.923241741116789e-05,
      "loss": 0.07104206681251526,
      "step": 155
    },
    {
      "epoch": 0.14904517931998137,
      "grad_norm": 1.0078125,
      "learning_rate": 5.918232702362575e-05,
      "loss": 0.05300582647323608,
      "step": 160
    },
    {
      "epoch": 0.1537028411737308,
      "grad_norm": 2.078125,
      "learning_rate": 5.913067604477262e-05,
      "loss": 0.13233035802841187,
      "step": 165
    },
    {
      "epoch": 0.1583605030274802,
      "grad_norm": 1.3125,
      "learning_rate": 5.907746723676205e-05,
      "loss": 0.07034454345703126,
      "step": 170
    },
    {
      "epoch": 0.1630181648812296,
      "grad_norm": 1.0,
      "learning_rate": 5.902270344505608e-05,
      "loss": 0.06918042302131652,
      "step": 175
    },
    {
      "epoch": 0.16767582673497905,
      "grad_norm": 0.96484375,
      "learning_rate": 5.896638759827303e-05,
      "loss": 0.061054134368896486,
      "step": 180
    },
    {
      "epoch": 0.17233348858872846,
      "grad_norm": 1.6015625,
      "learning_rate": 5.8908522708030884e-05,
      "loss": 0.10253015756607056,
      "step": 185
    },
    {
      "epoch": 0.17699115044247787,
      "grad_norm": 1.3125,
      "learning_rate": 5.884911186878626e-05,
      "loss": 0.09637802839279175,
      "step": 190
    },
    {
      "epoch": 0.18164881229622729,
      "grad_norm": 1.3125,
      "learning_rate": 5.878815825766892e-05,
      "loss": 0.09289785027503968,
      "step": 195
    },
    {
      "epoch": 0.18630647414997673,
      "grad_norm": 3.484375,
      "learning_rate": 5.872566513431184e-05,
      "loss": 0.08359139561653137,
      "step": 200
    },
    {
      "epoch": 0.19096413600372614,
      "grad_norm": 7.6875,
      "learning_rate": 5.8661635840676944e-05,
      "loss": 0.07835670113563538,
      "step": 205
    },
    {
      "epoch": 0.19562179785747555,
      "grad_norm": 1.1171875,
      "learning_rate": 5.859607380087634e-05,
      "loss": 0.06977214813232421,
      "step": 210
    },
    {
      "epoch": 0.20027945971122496,
      "grad_norm": 1.1953125,
      "learning_rate": 5.852898252098921e-05,
      "loss": 0.0683005690574646,
      "step": 215
    },
    {
      "epoch": 0.20493712156497437,
      "grad_norm": 1.0859375,
      "learning_rate": 5.846036558887435e-05,
      "loss": 0.06400226354598999,
      "step": 220
    },
    {
      "epoch": 0.2095947834187238,
      "grad_norm": 1.0078125,
      "learning_rate": 5.839022667397827e-05,
      "loss": 0.06587435007095337,
      "step": 225
    },
    {
      "epoch": 0.21425244527247322,
      "grad_norm": 1.3125,
      "learning_rate": 5.831856952713896e-05,
      "loss": 0.08793356418609619,
      "step": 230
    },
    {
      "epoch": 0.21891010712622264,
      "grad_norm": 1.53125,
      "learning_rate": 5.8245397980385344e-05,
      "loss": 0.06810299754142761,
      "step": 235
    },
    {
      "epoch": 0.22356776897997205,
      "grad_norm": 2.921875,
      "learning_rate": 5.817071594673229e-05,
      "loss": 0.06869109869003295,
      "step": 240
    },
    {
      "epoch": 0.22822543083372146,
      "grad_norm": 1.5234375,
      "learning_rate": 5.809452741997143e-05,
      "loss": 0.08056961297988892,
      "step": 245
    },
    {
      "epoch": 0.2328830926874709,
      "grad_norm": 1.6171875,
      "learning_rate": 5.801683647445749e-05,
      "loss": 0.08139086961746216,
      "step": 250
    },
    {
      "epoch": 0.2375407545412203,
      "grad_norm": 0.9375,
      "learning_rate": 5.7937647264890514e-05,
      "loss": 0.06604200601577759,
      "step": 255
    },
    {
      "epoch": 0.24219841639496972,
      "grad_norm": 3.53125,
      "learning_rate": 5.7856964026093575e-05,
      "loss": 0.08705702424049377,
      "step": 260
    },
    {
      "epoch": 0.24685607824871914,
      "grad_norm": 0.9765625,
      "learning_rate": 5.777479107278639e-05,
      "loss": 0.08005684614181519,
      "step": 265
    },
    {
      "epoch": 0.2515137401024686,
      "grad_norm": 1.734375,
      "learning_rate": 5.769113279935452e-05,
      "loss": 0.06209390759468079,
      "step": 270
    },
    {
      "epoch": 0.25617140195621796,
      "grad_norm": 1.9609375,
      "learning_rate": 5.760599367961442e-05,
      "loss": 0.10012856721878052,
      "step": 275
    },
    {
      "epoch": 0.2608290638099674,
      "grad_norm": 1.0234375,
      "learning_rate": 5.751937826657417e-05,
      "loss": 0.09347758889198303,
      "step": 280
    },
    {
      "epoch": 0.26548672566371684,
      "grad_norm": 1.3203125,
      "learning_rate": 5.743129119218996e-05,
      "loss": 0.08649051785469056,
      "step": 285
    },
    {
      "epoch": 0.2701443875174662,
      "grad_norm": 1.640625,
      "learning_rate": 5.734173716711847e-05,
      "loss": 0.06381095051765442,
      "step": 290
    },
    {
      "epoch": 0.27480204937121566,
      "grad_norm": 1.59375,
      "learning_rate": 5.725072098046486e-05,
      "loss": 0.06688931584358215,
      "step": 295
    },
    {
      "epoch": 0.27945971122496505,
      "grad_norm": 1.921875,
      "learning_rate": 5.715824749952672e-05,
      "loss": 0.07226124405860901,
      "step": 300
    },
    {
      "epoch": 0.2841173730787145,
      "grad_norm": 2.125,
      "learning_rate": 5.706432166953378e-05,
      "loss": 0.07504152059555054,
      "step": 305
    },
    {
      "epoch": 0.2887750349324639,
      "grad_norm": 0.95703125,
      "learning_rate": 5.696894851338341e-05,
      "loss": 0.049904930591583255,
      "step": 310
    },
    {
      "epoch": 0.2934326967862133,
      "grad_norm": 1.5390625,
      "learning_rate": 5.687213313137208e-05,
      "loss": 0.07915764451026916,
      "step": 315
    },
    {
      "epoch": 0.29809035863996275,
      "grad_norm": 1.9296875,
      "learning_rate": 5.677388070092252e-05,
      "loss": 0.06789354085922242,
      "step": 320
    },
    {
      "epoch": 0.30274802049371213,
      "grad_norm": 1.6015625,
      "learning_rate": 5.667419647630693e-05,
      "loss": 0.07174413204193116,
      "step": 325
    },
    {
      "epoch": 0.3074056823474616,
      "grad_norm": 1.3671875,
      "learning_rate": 5.657308578836594e-05,
      "loss": 0.0827239990234375,
      "step": 330
    },
    {
      "epoch": 0.312063344201211,
      "grad_norm": 1.640625,
      "learning_rate": 5.647055404422357e-05,
      "loss": 0.06073644757270813,
      "step": 335
    },
    {
      "epoch": 0.3167210060549604,
      "grad_norm": 1.328125,
      "learning_rate": 5.636660672699803e-05,
      "loss": 0.050183039903640744,
      "step": 340
    },
    {
      "epoch": 0.32137866790870984,
      "grad_norm": 2.875,
      "learning_rate": 5.626124939550853e-05,
      "loss": 0.08814806938171386,
      "step": 345
    },
    {
      "epoch": 0.3260363297624592,
      "grad_norm": 1.921875,
      "learning_rate": 5.6154487683978e-05,
      "loss": 0.07444313168525696,
      "step": 350
    },
    {
      "epoch": 0.33069399161620866,
      "grad_norm": 1.7734375,
      "learning_rate": 5.60463273017318e-05,
      "loss": 0.05980492830276489,
      "step": 355
    },
    {
      "epoch": 0.3353516534699581,
      "grad_norm": 0.80078125,
      "learning_rate": 5.593677403289235e-05,
      "loss": 0.05909588932991028,
      "step": 360
    },
    {
      "epoch": 0.3400093153237075,
      "grad_norm": 1.296875,
      "learning_rate": 5.5825833736069904e-05,
      "loss": 0.08176165819168091,
      "step": 365
    },
    {
      "epoch": 0.3446669771774569,
      "grad_norm": 1.640625,
      "learning_rate": 5.571351234404913e-05,
      "loss": 0.07512840628623962,
      "step": 370
    },
    {
      "epoch": 0.3493246390312063,
      "grad_norm": 2.0,
      "learning_rate": 5.559981586347196e-05,
      "loss": 0.07859928011894227,
      "step": 375
    },
    {
      "epoch": 0.35398230088495575,
      "grad_norm": 1.9296875,
      "learning_rate": 5.548475037451627e-05,
      "loss": 0.07391362190246582,
      "step": 380
    },
    {
      "epoch": 0.3586399627387052,
      "grad_norm": 2.5625,
      "learning_rate": 5.536832203057081e-05,
      "loss": 0.08393895626068115,
      "step": 385
    },
    {
      "epoch": 0.36329762459245457,
      "grad_norm": 2.9375,
      "learning_rate": 5.525053705790606e-05,
      "loss": 0.06296343803405761,
      "step": 390
    },
    {
      "epoch": 0.367955286446204,
      "grad_norm": 1.2734375,
      "learning_rate": 5.5131401755341354e-05,
      "loss": 0.08504929542541503,
      "step": 395
    },
    {
      "epoch": 0.37261294829995345,
      "grad_norm": 1.1953125,
      "learning_rate": 5.501092249390793e-05,
      "loss": 0.08101065158843994,
      "step": 400
    },
    {
      "epoch": 0.37727061015370283,
      "grad_norm": 7.5,
      "learning_rate": 5.488910571650835e-05,
      "loss": 0.07101285457611084,
      "step": 405
    },
    {
      "epoch": 0.3819282720074523,
      "grad_norm": 1.90625,
      "learning_rate": 5.4765957937571824e-05,
      "loss": 0.08065653443336487,
      "step": 410
    },
    {
      "epoch": 0.38658593386120166,
      "grad_norm": 1.4921875,
      "learning_rate": 5.464148574270594e-05,
      "loss": 0.07403583526611328,
      "step": 415
    },
    {
      "epoch": 0.3912435957149511,
      "grad_norm": 1.46875,
      "learning_rate": 5.451569578834441e-05,
      "loss": 0.06334596872329712,
      "step": 420
    },
    {
      "epoch": 0.39590125756870054,
      "grad_norm": 1.6875,
      "learning_rate": 5.4388594801391146e-05,
      "loss": 0.06494083404541015,
      "step": 425
    },
    {
      "epoch": 0.4005589194224499,
      "grad_norm": 3.25,
      "learning_rate": 5.426018957886052e-05,
      "loss": 0.06925193071365357,
      "step": 430
    },
    {
      "epoch": 0.40521658127619936,
      "grad_norm": 2.84375,
      "learning_rate": 5.413048698751386e-05,
      "loss": 0.08767553567886352,
      "step": 435
    },
    {
      "epoch": 0.40987424312994875,
      "grad_norm": 1.6796875,
      "learning_rate": 5.399949396349221e-05,
      "loss": 0.07317562699317932,
      "step": 440
    },
    {
      "epoch": 0.4145319049836982,
      "grad_norm": 0.984375,
      "learning_rate": 5.386721751194549e-05,
      "loss": 0.07249165177345276,
      "step": 445
    },
    {
      "epoch": 0.4191895668374476,
      "grad_norm": 1.3359375,
      "learning_rate": 5.37336647066578e-05,
      "loss": 0.08126922845840454,
      "step": 450
    },
    {
      "epoch": 0.423847228691197,
      "grad_norm": 1.3359375,
      "learning_rate": 5.3598842689669155e-05,
      "loss": 0.06756317615509033,
      "step": 455
    },
    {
      "epoch": 0.42850489054494645,
      "grad_norm": 1.671875,
      "learning_rate": 5.346275867089355e-05,
      "loss": 0.07897292971611022,
      "step": 460
    },
    {
      "epoch": 0.43316255239869583,
      "grad_norm": 0.8046875,
      "learning_rate": 5.332541992773342e-05,
      "loss": 0.049942368268966676,
      "step": 465
    },
    {
      "epoch": 0.43782021425244527,
      "grad_norm": 1.0859375,
      "learning_rate": 5.3186833804690415e-05,
      "loss": 0.05567145347595215,
      "step": 470
    },
    {
      "epoch": 0.4424778761061947,
      "grad_norm": 1.1796875,
      "learning_rate": 5.304700771297267e-05,
      "loss": 0.06544985771179199,
      "step": 475
    },
    {
      "epoch": 0.4471355379599441,
      "grad_norm": 1.671875,
      "learning_rate": 5.2905949130098465e-05,
      "loss": 0.07286288738250732,
      "step": 480
    },
    {
      "epoch": 0.45179319981369354,
      "grad_norm": 1.6328125,
      "learning_rate": 5.276366559949635e-05,
      "loss": 0.08477526903152466,
      "step": 485
    },
    {
      "epoch": 0.4564508616674429,
      "grad_norm": 1.3125,
      "learning_rate": 5.2620164730101746e-05,
      "loss": 0.07429101467132568,
      "step": 490
    },
    {
      "epoch": 0.46110852352119236,
      "grad_norm": 1.078125,
      "learning_rate": 5.2475454195950024e-05,
      "loss": 0.0752301573753357,
      "step": 495
    },
    {
      "epoch": 0.4657661853749418,
      "grad_norm": 1.171875,
      "learning_rate": 5.2329541735766116e-05,
      "loss": 0.06806610822677613,
      "step": 500
    },
    {
      "epoch": 0.4704238472286912,
      "grad_norm": 3.5,
      "learning_rate": 5.218243515255072e-05,
      "loss": 0.07269450426101684,
      "step": 505
    },
    {
      "epoch": 0.4750815090824406,
      "grad_norm": 1.6640625,
      "learning_rate": 5.203414231316295e-05,
      "loss": 0.0846606969833374,
      "step": 510
    },
    {
      "epoch": 0.47973917093619,
      "grad_norm": 1.125,
      "learning_rate": 5.1884671147899686e-05,
      "loss": 0.05865778923034668,
      "step": 515
    },
    {
      "epoch": 0.48439683278993945,
      "grad_norm": 1.5234375,
      "learning_rate": 5.1734029650071437e-05,
      "loss": 0.06631482243537903,
      "step": 520
    },
    {
      "epoch": 0.4890544946436889,
      "grad_norm": 1.8359375,
      "learning_rate": 5.158222587557493e-05,
      "loss": 0.09194896817207336,
      "step": 525
    },
    {
      "epoch": 0.49371215649743827,
      "grad_norm": 1.3671875,
      "learning_rate": 5.142926794246228e-05,
      "loss": 0.06618914604187012,
      "step": 530
    },
    {
      "epoch": 0.4983698183511877,
      "grad_norm": 1.7734375,
      "learning_rate": 5.1275164030506855e-05,
      "loss": 0.05849605202674866,
      "step": 535
    },
    {
      "epoch": 0.5030274802049371,
      "grad_norm": 1.71875,
      "learning_rate": 5.111992238076583e-05,
      "loss": 0.06805847287178039,
      "step": 540
    },
    {
      "epoch": 0.5076851420586865,
      "grad_norm": 1.328125,
      "learning_rate": 5.096355129513954e-05,
      "loss": 0.0659984827041626,
      "step": 545
    },
    {
      "epoch": 0.5123428039124359,
      "grad_norm": 2.421875,
      "learning_rate": 5.0806059135927425e-05,
      "loss": 0.08051547408103943,
      "step": 550
    },
    {
      "epoch": 0.5170004657661854,
      "grad_norm": 1.0234375,
      "learning_rate": 5.06474543253809e-05,
      "loss": 0.05917655229568482,
      "step": 555
    },
    {
      "epoch": 0.5216581276199348,
      "grad_norm": 1.4765625,
      "learning_rate": 5.048774534525296e-05,
      "loss": 0.07989615201950073,
      "step": 560
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 1.53125,
      "learning_rate": 5.032694073634456e-05,
      "loss": 0.0757764756679535,
      "step": 565
    },
    {
      "epoch": 0.5309734513274337,
      "grad_norm": 1.296875,
      "learning_rate": 5.0165049098047905e-05,
      "loss": 0.07939087748527526,
      "step": 570
    },
    {
      "epoch": 0.5356311131811831,
      "grad_norm": 1.0703125,
      "learning_rate": 5.000207908788657e-05,
      "loss": 0.06915702223777771,
      "step": 575
    },
    {
      "epoch": 0.5402887750349324,
      "grad_norm": 1.2890625,
      "learning_rate": 4.9838039421052495e-05,
      "loss": 0.05958174467086792,
      "step": 580
    },
    {
      "epoch": 0.5449464368886818,
      "grad_norm": 0.765625,
      "learning_rate": 4.9672938869939985e-05,
      "loss": 0.053989958763122556,
      "step": 585
    },
    {
      "epoch": 0.5496040987424313,
      "grad_norm": 0.96484375,
      "learning_rate": 4.95067862636765e-05,
      "loss": 0.07627325057983399,
      "step": 590
    },
    {
      "epoch": 0.5542617605961807,
      "grad_norm": 0.86328125,
      "learning_rate": 4.933959048765059e-05,
      "loss": 0.05423608422279358,
      "step": 595
    },
    {
      "epoch": 0.5589194224499301,
      "grad_norm": 1.3515625,
      "learning_rate": 4.9171360483036624e-05,
      "loss": 0.06786616444587708,
      "step": 600
    },
    {
      "epoch": 0.5635770843036796,
      "grad_norm": 0.80078125,
      "learning_rate": 4.900210524631676e-05,
      "loss": 0.06529110670089722,
      "step": 605
    },
    {
      "epoch": 0.568234746157429,
      "grad_norm": 1.3359375,
      "learning_rate": 4.883183382879971e-05,
      "loss": 0.06635728478431702,
      "step": 610
    },
    {
      "epoch": 0.5728924080111784,
      "grad_norm": 1.2578125,
      "learning_rate": 4.8660555336136816e-05,
      "loss": 0.07374628186225891,
      "step": 615
    },
    {
      "epoch": 0.5775500698649279,
      "grad_norm": 1.109375,
      "learning_rate": 4.848827892783503e-05,
      "loss": 0.058982133865356445,
      "step": 620
    },
    {
      "epoch": 0.5822077317186772,
      "grad_norm": 1.328125,
      "learning_rate": 4.831501381676712e-05,
      "loss": 0.08937416672706604,
      "step": 625
    },
    {
      "epoch": 0.5868653935724266,
      "grad_norm": 1.0703125,
      "learning_rate": 4.814076926867899e-05,
      "loss": 0.06815096139907836,
      "step": 630
    },
    {
      "epoch": 0.5915230554261761,
      "grad_norm": 0.77734375,
      "learning_rate": 4.796555460169415e-05,
      "loss": 0.06007786989212036,
      "step": 635
    },
    {
      "epoch": 0.5961807172799255,
      "grad_norm": 1.1796875,
      "learning_rate": 4.778937918581542e-05,
      "loss": 0.06676498651504517,
      "step": 640
    },
    {
      "epoch": 0.6008383791336749,
      "grad_norm": 0.8828125,
      "learning_rate": 4.761225244242389e-05,
      "loss": 0.06312382817268372,
      "step": 645
    },
    {
      "epoch": 0.6054960409874243,
      "grad_norm": 1.875,
      "learning_rate": 4.7434183843775e-05,
      "loss": 0.07485262155532837,
      "step": 650
    },
    {
      "epoch": 0.6101537028411738,
      "grad_norm": 1.2890625,
      "learning_rate": 4.7255182912492086e-05,
      "loss": 0.05705668330192566,
      "step": 655
    },
    {
      "epoch": 0.6148113646949231,
      "grad_norm": 1.3359375,
      "learning_rate": 4.707525922105708e-05,
      "loss": 0.08206185102462768,
      "step": 660
    },
    {
      "epoch": 0.6194690265486725,
      "grad_norm": 1.1875,
      "learning_rate": 4.689442239129858e-05,
      "loss": 0.06421056389808655,
      "step": 665
    },
    {
      "epoch": 0.624126688402422,
      "grad_norm": 0.71484375,
      "learning_rate": 4.6712682093877395e-05,
      "loss": 0.0625627338886261,
      "step": 670
    },
    {
      "epoch": 0.6287843502561714,
      "grad_norm": 1.2265625,
      "learning_rate": 4.653004804776925e-05,
      "loss": 0.08293673396110535,
      "step": 675
    },
    {
      "epoch": 0.6334420121099208,
      "grad_norm": 1.6015625,
      "learning_rate": 4.634653001974519e-05,
      "loss": 0.07818044424057007,
      "step": 680
    },
    {
      "epoch": 0.6380996739636703,
      "grad_norm": 1.1953125,
      "learning_rate": 4.6162137823849125e-05,
      "loss": 0.06400628089904785,
      "step": 685
    },
    {
      "epoch": 0.6427573358174197,
      "grad_norm": 1.296875,
      "learning_rate": 4.597688132087314e-05,
      "loss": 0.05182795524597168,
      "step": 690
    },
    {
      "epoch": 0.6474149976711691,
      "grad_norm": 1.4609375,
      "learning_rate": 4.5790770417830084e-05,
      "loss": 0.08495147228240967,
      "step": 695
    },
    {
      "epoch": 0.6520726595249184,
      "grad_norm": 1.3515625,
      "learning_rate": 4.56038150674238e-05,
      "loss": 0.061500102281570435,
      "step": 700
    },
    {
      "epoch": 0.6567303213786679,
      "grad_norm": 3.25,
      "learning_rate": 4.5416025267516866e-05,
      "loss": 0.069045090675354,
      "step": 705
    },
    {
      "epoch": 0.6613879832324173,
      "grad_norm": 1.328125,
      "learning_rate": 4.522741106059595e-05,
      "loss": 0.059973156452178954,
      "step": 710
    },
    {
      "epoch": 0.6660456450861667,
      "grad_norm": 1.171875,
      "learning_rate": 4.5037982533234734e-05,
      "loss": 0.07730792760848999,
      "step": 715
    },
    {
      "epoch": 0.6707033069399162,
      "grad_norm": 2.25,
      "learning_rate": 4.48477498155546e-05,
      "loss": 0.06633560657501221,
      "step": 720
    },
    {
      "epoch": 0.6753609687936656,
      "grad_norm": 1.5546875,
      "learning_rate": 4.465672308068278e-05,
      "loss": 0.05775619745254516,
      "step": 725
    },
    {
      "epoch": 0.680018630647415,
      "grad_norm": 1.5703125,
      "learning_rate": 4.446491254420841e-05,
      "loss": 0.07142780423164367,
      "step": 730
    },
    {
      "epoch": 0.6846762925011645,
      "grad_norm": 2.15625,
      "learning_rate": 4.427232846363617e-05,
      "loss": 0.06803997159004212,
      "step": 735
    },
    {
      "epoch": 0.6893339543549138,
      "grad_norm": 1.1015625,
      "learning_rate": 4.4078981137837804e-05,
      "loss": 0.06891953945159912,
      "step": 740
    },
    {
      "epoch": 0.6939916162086632,
      "grad_norm": 1.1328125,
      "learning_rate": 4.388488090650132e-05,
      "loss": 0.07929319143295288,
      "step": 745
    },
    {
      "epoch": 0.6986492780624126,
      "grad_norm": 1.3125,
      "learning_rate": 4.3690038149578044e-05,
      "loss": 0.06713913083076477,
      "step": 750
    },
    {
      "epoch": 0.7033069399161621,
      "grad_norm": 2.171875,
      "learning_rate": 4.349446328672756e-05,
      "loss": 0.0739738404750824,
      "step": 755
    },
    {
      "epoch": 0.7079646017699115,
      "grad_norm": 1.1328125,
      "learning_rate": 4.329816677676049e-05,
      "loss": 0.058427393436431885,
      "step": 760
    },
    {
      "epoch": 0.7126222636236609,
      "grad_norm": 1.3046875,
      "learning_rate": 4.310115911707918e-05,
      "loss": 0.06763650178909301,
      "step": 765
    },
    {
      "epoch": 0.7172799254774104,
      "grad_norm": 1.984375,
      "learning_rate": 4.2903450843116284e-05,
      "loss": 0.07517719864845276,
      "step": 770
    },
    {
      "epoch": 0.7219375873311598,
      "grad_norm": 1.4765625,
      "learning_rate": 4.2705052527771444e-05,
      "loss": 0.08248428702354431,
      "step": 775
    },
    {
      "epoch": 0.7265952491849091,
      "grad_norm": 0.93359375,
      "learning_rate": 4.250597478084583e-05,
      "loss": 0.05707501769065857,
      "step": 780
    },
    {
      "epoch": 0.7312529110386586,
      "grad_norm": 1.8671875,
      "learning_rate": 4.230622824847474e-05,
      "loss": 0.07538543343544006,
      "step": 785
    },
    {
      "epoch": 0.735910572892408,
      "grad_norm": 1.4453125,
      "learning_rate": 4.2105823612558285e-05,
      "loss": 0.054673463106155396,
      "step": 790
    },
    {
      "epoch": 0.7405682347461574,
      "grad_norm": 1.8984375,
      "learning_rate": 4.19047715901902e-05,
      "loss": 0.07854291796684265,
      "step": 795
    },
    {
      "epoch": 0.7452258965999069,
      "grad_norm": 1.453125,
      "learning_rate": 4.1703082933084666e-05,
      "loss": 0.05171064734458923,
      "step": 800
    },
    {
      "epoch": 0.7498835584536563,
      "grad_norm": 1.5546875,
      "learning_rate": 4.1500768427001337e-05,
      "loss": 0.05756812691688538,
      "step": 805
    },
    {
      "epoch": 0.7545412203074057,
      "grad_norm": 1.4453125,
      "learning_rate": 4.1297838891168575e-05,
      "loss": 0.07100930213928222,
      "step": 810
    },
    {
      "epoch": 0.759198882161155,
      "grad_norm": 1.203125,
      "learning_rate": 4.109430517770487e-05,
      "loss": 0.06920987367630005,
      "step": 815
    },
    {
      "epoch": 0.7638565440149045,
      "grad_norm": 0.92578125,
      "learning_rate": 4.089017817103847e-05,
      "loss": 0.05224462747573853,
      "step": 820
    },
    {
      "epoch": 0.7685142058686539,
      "grad_norm": 2.0625,
      "learning_rate": 4.0685468787325325e-05,
      "loss": 0.06810380220413208,
      "step": 825
    },
    {
      "epoch": 0.7731718677224033,
      "grad_norm": 2.484375,
      "learning_rate": 4.048018797386533e-05,
      "loss": 0.07529953122138977,
      "step": 830
    },
    {
      "epoch": 0.7778295295761528,
      "grad_norm": 1.3203125,
      "learning_rate": 4.02743467085169e-05,
      "loss": 0.07324286699295043,
      "step": 835
    },
    {
      "epoch": 0.7824871914299022,
      "grad_norm": 1.796875,
      "learning_rate": 4.006795599910986e-05,
      "loss": 0.06480486392974853,
      "step": 840
    },
    {
      "epoch": 0.7871448532836516,
      "grad_norm": 2.140625,
      "learning_rate": 3.986102688285681e-05,
      "loss": 0.07136549353599549,
      "step": 845
    },
    {
      "epoch": 0.7918025151374011,
      "grad_norm": 1.4453125,
      "learning_rate": 3.9653570425762894e-05,
      "loss": 0.0683221161365509,
      "step": 850
    },
    {
      "epoch": 0.7964601769911505,
      "grad_norm": 1.1953125,
      "learning_rate": 3.9445597722034024e-05,
      "loss": 0.07045792937278747,
      "step": 855
    },
    {
      "epoch": 0.8011178388448998,
      "grad_norm": 1.9765625,
      "learning_rate": 3.923711989348352e-05,
      "loss": 0.06577153205871582,
      "step": 860
    },
    {
      "epoch": 0.8057755006986492,
      "grad_norm": 1.9921875,
      "learning_rate": 3.9028148088937454e-05,
      "loss": 0.07191576361656189,
      "step": 865
    },
    {
      "epoch": 0.8104331625523987,
      "grad_norm": 1.4453125,
      "learning_rate": 3.881869348363836e-05,
      "loss": 0.05743700861930847,
      "step": 870
    },
    {
      "epoch": 0.8150908244061481,
      "grad_norm": 1.078125,
      "learning_rate": 3.8608767278647655e-05,
      "loss": 0.06495893597602845,
      "step": 875
    },
    {
      "epoch": 0.8197484862598975,
      "grad_norm": 1.796875,
      "learning_rate": 3.83983807002466e-05,
      "loss": 0.06839098334312439,
      "step": 880
    },
    {
      "epoch": 0.824406148113647,
      "grad_norm": 1.0546875,
      "learning_rate": 3.818754499933601e-05,
      "loss": 0.06613754630088806,
      "step": 885
    },
    {
      "epoch": 0.8290638099673964,
      "grad_norm": 1.2421875,
      "learning_rate": 3.7976271450834504e-05,
      "loss": 0.06127074360847473,
      "step": 890
    },
    {
      "epoch": 0.8337214718211458,
      "grad_norm": 1.3046875,
      "learning_rate": 3.776457135307562e-05,
      "loss": 0.05446965098381042,
      "step": 895
    },
    {
      "epoch": 0.8383791336748952,
      "grad_norm": 1.03125,
      "learning_rate": 3.75524560272036e-05,
      "loss": 0.057818031311035155,
      "step": 900
    },
    {
      "epoch": 0.8430367955286446,
      "grad_norm": 1.1640625,
      "learning_rate": 3.733993681656792e-05,
      "loss": 0.06393643021583557,
      "step": 905
    },
    {
      "epoch": 0.847694457382394,
      "grad_norm": 1.4921875,
      "learning_rate": 3.7127025086116736e-05,
      "loss": 0.06902187466621398,
      "step": 910
    },
    {
      "epoch": 0.8523521192361434,
      "grad_norm": 1.71875,
      "learning_rate": 3.691373222178909e-05,
      "loss": 0.06524651050567627,
      "step": 915
    },
    {
      "epoch": 0.8570097810898929,
      "grad_norm": 1.5234375,
      "learning_rate": 3.670006962990604e-05,
      "loss": 0.06683992147445679,
      "step": 920
    },
    {
      "epoch": 0.8616674429436423,
      "grad_norm": 0.97265625,
      "learning_rate": 3.6486048736560614e-05,
      "loss": 0.07013018727302552,
      "step": 925
    },
    {
      "epoch": 0.8663251047973917,
      "grad_norm": 1.3203125,
      "learning_rate": 3.627168098700688e-05,
      "loss": 0.06274893283843994,
      "step": 930
    },
    {
      "epoch": 0.8709827666511412,
      "grad_norm": 1.4140625,
      "learning_rate": 3.605697784504782e-05,
      "loss": 0.07435029745101929,
      "step": 935
    },
    {
      "epoch": 0.8756404285048905,
      "grad_norm": 2.640625,
      "learning_rate": 3.584195079242227e-05,
      "loss": 0.04883918762207031,
      "step": 940
    },
    {
      "epoch": 0.8802980903586399,
      "grad_norm": 1.328125,
      "learning_rate": 3.562661132819093e-05,
      "loss": 0.0625451922416687,
      "step": 945
    },
    {
      "epoch": 0.8849557522123894,
      "grad_norm": 1.875,
      "learning_rate": 3.541097096812144e-05,
      "loss": 0.0689104974269867,
      "step": 950
    },
    {
      "epoch": 0.8896134140661388,
      "grad_norm": 2.21875,
      "learning_rate": 3.5195041244072506e-05,
      "loss": 0.05478867292404175,
      "step": 955
    },
    {
      "epoch": 0.8942710759198882,
      "grad_norm": 1.6640625,
      "learning_rate": 3.497883370337724e-05,
      "loss": 0.07447841763496399,
      "step": 960
    },
    {
      "epoch": 0.8989287377736377,
      "grad_norm": 1.2734375,
      "learning_rate": 3.476235990822563e-05,
      "loss": 0.07308604717254638,
      "step": 965
    },
    {
      "epoch": 0.9035863996273871,
      "grad_norm": 1.3125,
      "learning_rate": 3.4545631435046233e-05,
      "loss": 0.06406825184822082,
      "step": 970
    },
    {
      "epoch": 0.9082440614811365,
      "grad_norm": 1.2421875,
      "learning_rate": 3.43286598738871e-05,
      "loss": 0.06342694759368897,
      "step": 975
    },
    {
      "epoch": 0.9129017233348858,
      "grad_norm": 1.625,
      "learning_rate": 3.411145682779593e-05,
      "loss": 0.06270037293434143,
      "step": 980
    },
    {
      "epoch": 0.9175593851886353,
      "grad_norm": 4.25,
      "learning_rate": 3.389403391219963e-05,
      "loss": 0.07241315245628357,
      "step": 985
    },
    {
      "epoch": 0.9222170470423847,
      "grad_norm": 0.8515625,
      "learning_rate": 3.3676402754283145e-05,
      "loss": 0.0640365719795227,
      "step": 990
    },
    {
      "epoch": 0.9268747088961341,
      "grad_norm": 1.2265625,
      "learning_rate": 3.345857499236761e-05,
      "loss": 0.07342595458030701,
      "step": 995
    },
    {
      "epoch": 0.9315323707498836,
      "grad_norm": 1.8203125,
      "learning_rate": 3.324056227528805e-05,
      "loss": 0.059070253372192384,
      "step": 1000
    },
    {
      "epoch": 0.936190032603633,
      "grad_norm": 1.046875,
      "learning_rate": 3.302237626177036e-05,
      "loss": 0.06199865341186524,
      "step": 1005
    },
    {
      "epoch": 0.9408476944573824,
      "grad_norm": 1.2421875,
      "learning_rate": 3.2804028619807865e-05,
      "loss": 0.04795067310333252,
      "step": 1010
    },
    {
      "epoch": 0.9455053563111319,
      "grad_norm": 1.1015625,
      "learning_rate": 3.2585531026037325e-05,
      "loss": 0.05880788564682007,
      "step": 1015
    },
    {
      "epoch": 0.9501630181648812,
      "grad_norm": 1.4140625,
      "learning_rate": 3.2366895165114534e-05,
      "loss": 0.06886560916900634,
      "step": 1020
    },
    {
      "epoch": 0.9548206800186306,
      "grad_norm": 0.890625,
      "learning_rate": 3.2148132729089436e-05,
      "loss": 0.05063498020172119,
      "step": 1025
    },
    {
      "epoch": 0.95947834187238,
      "grad_norm": 1.0390625,
      "learning_rate": 3.192925541678085e-05,
      "loss": 0.05080598592758179,
      "step": 1030
    },
    {
      "epoch": 0.9641360037261295,
      "grad_norm": 1.1015625,
      "learning_rate": 3.171027493315089e-05,
      "loss": 0.07384888529777527,
      "step": 1035
    },
    {
      "epoch": 0.9687936655798789,
      "grad_norm": 1.546875,
      "learning_rate": 3.149120298867895e-05,
      "loss": 0.06471606492996215,
      "step": 1040
    },
    {
      "epoch": 0.9734513274336283,
      "grad_norm": 6.40625,
      "learning_rate": 3.127205129873555e-05,
      "loss": 0.08100690245628357,
      "step": 1045
    },
    {
      "epoch": 0.9781089892873778,
      "grad_norm": 1.2109375,
      "learning_rate": 3.105283158295575e-05,
      "loss": 0.07905409932136535,
      "step": 1050
    },
    {
      "epoch": 0.9827666511411272,
      "grad_norm": 1.2734375,
      "learning_rate": 3.083355556461244e-05,
      "loss": 0.05518416166305542,
      "step": 1055
    },
    {
      "epoch": 0.9874243129948765,
      "grad_norm": 0.90625,
      "learning_rate": 3.0614234969989445e-05,
      "loss": 0.07515279650688171,
      "step": 1060
    },
    {
      "epoch": 0.992081974848626,
      "grad_norm": 1.3515625,
      "learning_rate": 3.039488152775439e-05,
      "loss": 0.059355854988098145,
      "step": 1065
    },
    {
      "epoch": 0.9967396367023754,
      "grad_norm": 1.109375,
      "learning_rate": 3.017550696833151e-05,
      "loss": 0.07503612041473388,
      "step": 1070
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.06487495452165604,
      "eval_pearson_cosine": 0.5299683213233948,
      "eval_pearson_dot_product": 0.5299683213233948,
      "eval_pearson_euclidean": 0.536834180355072,
      "eval_pearson_manhattan": 0.534827470779419,
      "eval_runtime": 17.3569,
      "eval_samples_per_second": 117.186,
      "eval_spearman_cosine": 0.5408944697862549,
      "eval_spearman_dot_product": 0.5408955665177739,
      "eval_spearman_euclidean": 0.5408944701719207,
      "eval_spearman_manhattan": 0.5392575148709198,
      "eval_steps_per_second": 19.531,
      "step": 1074
    },
    {
      "epoch": 1.00093153237075,
      "grad_norm": 1.0234375,
      "learning_rate": 2.995612302327431e-05,
      "loss": 0.04589760899543762,
      "step": 1075
    },
    {
      "epoch": 1.0055891942244992,
      "grad_norm": 0.9609375,
      "learning_rate": 2.9736741424638238e-05,
      "loss": 0.049758344888687134,
      "step": 1080
    },
    {
      "epoch": 1.0102468560782487,
      "grad_norm": 1.34375,
      "learning_rate": 2.951737390435325e-05,
      "loss": 0.0830036759376526,
      "step": 1085
    },
    {
      "epoch": 1.0149045179319982,
      "grad_norm": 1.1484375,
      "learning_rate": 2.9298032193596438e-05,
      "loss": 0.07352387309074401,
      "step": 1090
    },
    {
      "epoch": 1.0195621797857475,
      "grad_norm": 1.71875,
      "learning_rate": 2.9078728022164646e-05,
      "loss": 0.050916272401809695,
      "step": 1095
    },
    {
      "epoch": 1.024219841639497,
      "grad_norm": 1.15625,
      "learning_rate": 2.8859473117847253e-05,
      "loss": 0.05987505912780762,
      "step": 1100
    },
    {
      "epoch": 1.0288775034932465,
      "grad_norm": 1.7109375,
      "learning_rate": 2.864027920579893e-05,
      "loss": 0.05799069404602051,
      "step": 1105
    },
    {
      "epoch": 1.0335351653469957,
      "grad_norm": 1.484375,
      "learning_rate": 2.8421158007912687e-05,
      "loss": 0.06869199872016907,
      "step": 1110
    },
    {
      "epoch": 1.0381928272007452,
      "grad_norm": 1.09375,
      "learning_rate": 2.8202121242192954e-05,
      "loss": 0.05415593981742859,
      "step": 1115
    },
    {
      "epoch": 1.0428504890544947,
      "grad_norm": 1.078125,
      "learning_rate": 2.798318062212895e-05,
      "loss": 0.06130121946334839,
      "step": 1120
    },
    {
      "epoch": 1.047508150908244,
      "grad_norm": 0.80859375,
      "learning_rate": 2.7764347856068322e-05,
      "loss": 0.05010392665863037,
      "step": 1125
    },
    {
      "epoch": 1.0521658127619935,
      "grad_norm": 1.3671875,
      "learning_rate": 2.7545634646590937e-05,
      "loss": 0.06199369430541992,
      "step": 1130
    },
    {
      "epoch": 1.056823474615743,
      "grad_norm": 1.78125,
      "learning_rate": 2.7327052689883128e-05,
      "loss": 0.05268788933753967,
      "step": 1135
    },
    {
      "epoch": 1.0614811364694923,
      "grad_norm": 1.328125,
      "learning_rate": 2.7108613675112173e-05,
      "loss": 0.059926819801330564,
      "step": 1140
    },
    {
      "epoch": 1.0661387983232418,
      "grad_norm": 1.109375,
      "learning_rate": 2.6890329283801237e-05,
      "loss": 0.06648289561271667,
      "step": 1145
    },
    {
      "epoch": 1.0707964601769913,
      "grad_norm": 1.328125,
      "learning_rate": 2.6672211189204604e-05,
      "loss": 0.0454281896352768,
      "step": 1150
    },
    {
      "epoch": 1.0754541220307405,
      "grad_norm": 1.1796875,
      "learning_rate": 2.6454271055683478e-05,
      "loss": 0.06076472997665405,
      "step": 1155
    },
    {
      "epoch": 1.08011178388449,
      "grad_norm": 0.921875,
      "learning_rate": 2.6236520538082193e-05,
      "loss": 0.0678758978843689,
      "step": 1160
    },
    {
      "epoch": 1.0847694457382393,
      "grad_norm": 1.53125,
      "learning_rate": 2.601897128110494e-05,
      "loss": 0.05920932292938232,
      "step": 1165
    },
    {
      "epoch": 1.0894271075919888,
      "grad_norm": 1.0703125,
      "learning_rate": 2.5801634918693032e-05,
      "loss": 0.07065654397010804,
      "step": 1170
    },
    {
      "epoch": 1.0940847694457383,
      "grad_norm": 0.94921875,
      "learning_rate": 2.558452307340276e-05,
      "loss": 0.05684482455253601,
      "step": 1175
    },
    {
      "epoch": 1.0987424312994876,
      "grad_norm": 1.0390625,
      "learning_rate": 2.5367647355783883e-05,
      "loss": 0.04538664221763611,
      "step": 1180
    },
    {
      "epoch": 1.103400093153237,
      "grad_norm": 1.3828125,
      "learning_rate": 2.5151019363758657e-05,
      "loss": 0.04917815029621124,
      "step": 1185
    },
    {
      "epoch": 1.1080577550069866,
      "grad_norm": 1.390625,
      "learning_rate": 2.493465068200167e-05,
      "loss": 0.0480605810880661,
      "step": 1190
    },
    {
      "epoch": 1.1127154168607358,
      "grad_norm": 7.3125,
      "learning_rate": 2.4718552881320275e-05,
      "loss": 0.07286378145217895,
      "step": 1195
    },
    {
      "epoch": 1.1173730787144853,
      "grad_norm": 1.0,
      "learning_rate": 2.450273751803589e-05,
      "loss": 0.05467050075531006,
      "step": 1200
    },
    {
      "epoch": 1.1220307405682348,
      "grad_norm": 0.96875,
      "learning_rate": 2.4287216133365914e-05,
      "loss": 0.046616628766059875,
      "step": 1205
    },
    {
      "epoch": 1.126688402421984,
      "grad_norm": 1.078125,
      "learning_rate": 2.407200025280657e-05,
      "loss": 0.04863499999046326,
      "step": 1210
    },
    {
      "epoch": 1.1313460642757336,
      "grad_norm": 1.5,
      "learning_rate": 2.3857101385516585e-05,
      "loss": 0.07006229758262635,
      "step": 1215
    },
    {
      "epoch": 1.136003726129483,
      "grad_norm": 1.1953125,
      "learning_rate": 2.3642531023701663e-05,
      "loss": 0.054609501361846925,
      "step": 1220
    },
    {
      "epoch": 1.1406613879832324,
      "grad_norm": 1.5390625,
      "learning_rate": 2.342830064199992e-05,
      "loss": 0.05989115834236145,
      "step": 1225
    },
    {
      "epoch": 1.1453190498369819,
      "grad_norm": 1.0625,
      "learning_rate": 2.3214421696868273e-05,
      "loss": 0.059736084938049314,
      "step": 1230
    },
    {
      "epoch": 1.1499767116907313,
      "grad_norm": 1.71875,
      "learning_rate": 2.3000905625969768e-05,
      "loss": 0.06554105281829833,
      "step": 1235
    },
    {
      "epoch": 1.1546343735444806,
      "grad_norm": 1.0546875,
      "learning_rate": 2.278776384756193e-05,
      "loss": 0.057540059089660645,
      "step": 1240
    },
    {
      "epoch": 1.1592920353982301,
      "grad_norm": 1.390625,
      "learning_rate": 2.257500775988612e-05,
      "loss": 0.050140655040740965,
      "step": 1245
    },
    {
      "epoch": 1.1639496972519794,
      "grad_norm": 1.1796875,
      "learning_rate": 2.2362648740558042e-05,
      "loss": 0.056704151630401614,
      "step": 1250
    },
    {
      "epoch": 1.1686073591057289,
      "grad_norm": 1.0546875,
      "learning_rate": 2.215069814595926e-05,
      "loss": 0.06525537371635437,
      "step": 1255
    },
    {
      "epoch": 1.1732650209594784,
      "grad_norm": 1.375,
      "learning_rate": 2.193916731062988e-05,
      "loss": 0.05826699137687683,
      "step": 1260
    },
    {
      "epoch": 1.1779226828132279,
      "grad_norm": 1.171875,
      "learning_rate": 2.1728067546662422e-05,
      "loss": 0.046400585770606996,
      "step": 1265
    },
    {
      "epoch": 1.1825803446669771,
      "grad_norm": 1.0546875,
      "learning_rate": 2.1517410143096922e-05,
      "loss": 0.0535780668258667,
      "step": 1270
    },
    {
      "epoch": 1.1872380065207266,
      "grad_norm": 1.09375,
      "learning_rate": 2.1307206365317133e-05,
      "loss": 0.046080824732780454,
      "step": 1275
    },
    {
      "epoch": 1.191895668374476,
      "grad_norm": 1.53125,
      "learning_rate": 2.10974674544482e-05,
      "loss": 0.07175366878509522,
      "step": 1280
    },
    {
      "epoch": 1.1965533302282254,
      "grad_norm": 1.0234375,
      "learning_rate": 2.0888204626755408e-05,
      "loss": 0.06331425905227661,
      "step": 1285
    },
    {
      "epoch": 1.201210992081975,
      "grad_norm": 1.203125,
      "learning_rate": 2.0679429073044408e-05,
      "loss": 0.0636473536491394,
      "step": 1290
    },
    {
      "epoch": 1.2058686539357242,
      "grad_norm": 1.015625,
      "learning_rate": 2.0471151958062806e-05,
      "loss": 0.03870426416397095,
      "step": 1295
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 0.99609375,
      "learning_rate": 2.0263384419903027e-05,
      "loss": 0.03741786777973175,
      "step": 1300
    },
    {
      "epoch": 1.2151839776432232,
      "grad_norm": 1.3359375,
      "learning_rate": 2.0056137569406722e-05,
      "loss": 0.05014714598655701,
      "step": 1305
    },
    {
      "epoch": 1.2198416394969724,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9849422489570593e-05,
      "loss": 0.042108118534088135,
      "step": 1310
    },
    {
      "epoch": 1.224499301350722,
      "grad_norm": 1.5703125,
      "learning_rate": 1.9643250234953714e-05,
      "loss": 0.04169891178607941,
      "step": 1315
    },
    {
      "epoch": 1.2291569632044714,
      "grad_norm": 1.1484375,
      "learning_rate": 1.943763183108631e-05,
      "loss": 0.06241189241409302,
      "step": 1320
    },
    {
      "epoch": 1.2338146250582207,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9232578273880195e-05,
      "loss": 0.04898407459259033,
      "step": 1325
    },
    {
      "epoch": 1.2384722869119702,
      "grad_norm": 1.296875,
      "learning_rate": 1.9028100529040728e-05,
      "loss": 0.05555582642555237,
      "step": 1330
    },
    {
      "epoch": 1.2431299487657197,
      "grad_norm": 1.25,
      "learning_rate": 1.8824209531480384e-05,
      "loss": 0.05224438309669495,
      "step": 1335
    },
    {
      "epoch": 1.247787610619469,
      "grad_norm": 1.40625,
      "learning_rate": 1.8620916184733993e-05,
      "loss": 0.04730784296989441,
      "step": 1340
    },
    {
      "epoch": 1.2524452724732185,
      "grad_norm": 1.2578125,
      "learning_rate": 1.8418231360375647e-05,
      "loss": 0.045959821343421935,
      "step": 1345
    },
    {
      "epoch": 1.257102934326968,
      "grad_norm": 0.921875,
      "learning_rate": 1.8216165897437335e-05,
      "loss": 0.046253901720046994,
      "step": 1350
    },
    {
      "epoch": 1.2617605961807172,
      "grad_norm": 1.1484375,
      "learning_rate": 1.8014730601829263e-05,
      "loss": 0.05002530813217163,
      "step": 1355
    },
    {
      "epoch": 1.2664182580344667,
      "grad_norm": 0.9921875,
      "learning_rate": 1.781393624576203e-05,
      "loss": 0.058111125230789186,
      "step": 1360
    },
    {
      "epoch": 1.271075919888216,
      "grad_norm": 1.2734375,
      "learning_rate": 1.7613793567170495e-05,
      "loss": 0.057514190673828125,
      "step": 1365
    },
    {
      "epoch": 1.2757335817419655,
      "grad_norm": 1.046875,
      "learning_rate": 1.741431326913963e-05,
      "loss": 0.05811060070991516,
      "step": 1370
    },
    {
      "epoch": 1.280391243595715,
      "grad_norm": 1.0078125,
      "learning_rate": 1.7215506019332065e-05,
      "loss": 0.04376150071620941,
      "step": 1375
    },
    {
      "epoch": 1.2850489054494645,
      "grad_norm": 1.2734375,
      "learning_rate": 1.7017382449417665e-05,
      "loss": 0.05309507846832275,
      "step": 1380
    },
    {
      "epoch": 1.2897065673032138,
      "grad_norm": 0.984375,
      "learning_rate": 1.6819953154504952e-05,
      "loss": 0.04199279248714447,
      "step": 1385
    },
    {
      "epoch": 1.2943642291569633,
      "grad_norm": 0.98046875,
      "learning_rate": 1.6623228692574524e-05,
      "loss": 0.044170570373535153,
      "step": 1390
    },
    {
      "epoch": 1.2990218910107125,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6427219583914396e-05,
      "loss": 0.040523186326026917,
      "step": 1395
    },
    {
      "epoch": 1.303679552864462,
      "grad_norm": 1.3671875,
      "learning_rate": 1.6231936310557472e-05,
      "loss": 0.06709733009338378,
      "step": 1400
    },
    {
      "epoch": 1.3083372147182115,
      "grad_norm": 1.4609375,
      "learning_rate": 1.6037389315720967e-05,
      "loss": 0.06956819295883179,
      "step": 1405
    },
    {
      "epoch": 1.312994876571961,
      "grad_norm": 1.5078125,
      "learning_rate": 1.5843589003247902e-05,
      "loss": 0.0630099892616272,
      "step": 1410
    },
    {
      "epoch": 1.3176525384257103,
      "grad_norm": 0.98046875,
      "learning_rate": 1.565054573705076e-05,
      "loss": 0.038538819551467894,
      "step": 1415
    },
    {
      "epoch": 1.3223102002794598,
      "grad_norm": 1.2890625,
      "learning_rate": 1.545826984055731e-05,
      "loss": 0.038901329040527344,
      "step": 1420
    },
    {
      "epoch": 1.326967862133209,
      "grad_norm": 1.140625,
      "learning_rate": 1.52667715961584e-05,
      "loss": 0.07003207802772522,
      "step": 1425
    },
    {
      "epoch": 1.3316255239869585,
      "grad_norm": 1.953125,
      "learning_rate": 1.5076061244658204e-05,
      "loss": 0.049628344178199765,
      "step": 1430
    },
    {
      "epoch": 1.336283185840708,
      "grad_norm": 1.96875,
      "learning_rate": 1.4886148984726551e-05,
      "loss": 0.05139703750610351,
      "step": 1435
    },
    {
      "epoch": 1.3409408476944573,
      "grad_norm": 1.015625,
      "learning_rate": 1.4697044972353454e-05,
      "loss": 0.05404648780822754,
      "step": 1440
    },
    {
      "epoch": 1.3455985095482068,
      "grad_norm": 1.015625,
      "learning_rate": 1.4508759320306097e-05,
      "loss": 0.042486587166786195,
      "step": 1445
    },
    {
      "epoch": 1.350256171401956,
      "grad_norm": 2.515625,
      "learning_rate": 1.4321302097587964e-05,
      "loss": 0.05852026343345642,
      "step": 1450
    },
    {
      "epoch": 1.3549138332557056,
      "grad_norm": 1.53125,
      "learning_rate": 1.4134683328900397e-05,
      "loss": 0.062414920330047606,
      "step": 1455
    },
    {
      "epoch": 1.359571495109455,
      "grad_norm": 0.83984375,
      "learning_rate": 1.394891299410652e-05,
      "loss": 0.0492287278175354,
      "step": 1460
    },
    {
      "epoch": 1.3642291569632046,
      "grad_norm": 0.9765625,
      "learning_rate": 1.3764001027697486e-05,
      "loss": 0.057495498657226564,
      "step": 1465
    },
    {
      "epoch": 1.3688868188169538,
      "grad_norm": 1.734375,
      "learning_rate": 1.357995731826128e-05,
      "loss": 0.04979708790779114,
      "step": 1470
    },
    {
      "epoch": 1.3735444806707033,
      "grad_norm": 1.109375,
      "learning_rate": 1.3396791707953873e-05,
      "loss": 0.05228898525238037,
      "step": 1475
    },
    {
      "epoch": 1.3782021425244526,
      "grad_norm": 1.265625,
      "learning_rate": 1.321451399197284e-05,
      "loss": 0.06390035152435303,
      "step": 1480
    },
    {
      "epoch": 1.382859804378202,
      "grad_norm": 0.91015625,
      "learning_rate": 1.3033133918033669e-05,
      "loss": 0.053880298137664796,
      "step": 1485
    },
    {
      "epoch": 1.3875174662319516,
      "grad_norm": 1.765625,
      "learning_rate": 1.2852661185848319e-05,
      "loss": 0.06999391913414002,
      "step": 1490
    },
    {
      "epoch": 1.392175128085701,
      "grad_norm": 1.53125,
      "learning_rate": 1.2673105446606623e-05,
      "loss": 0.05712270736694336,
      "step": 1495
    },
    {
      "epoch": 1.3968327899394504,
      "grad_norm": 0.85546875,
      "learning_rate": 1.2494476302460142e-05,
      "loss": 0.048511195182800296,
      "step": 1500
    },
    {
      "epoch": 1.4014904517931999,
      "grad_norm": 1.578125,
      "learning_rate": 1.2316783306008618e-05,
      "loss": 0.05195020437240601,
      "step": 1505
    },
    {
      "epoch": 1.4061481136469491,
      "grad_norm": 2.703125,
      "learning_rate": 1.2140035959789195e-05,
      "loss": 0.06756129860877991,
      "step": 1510
    },
    {
      "epoch": 1.4108057755006986,
      "grad_norm": 2.859375,
      "learning_rate": 1.1964243715768218e-05,
      "loss": 0.06034356355667114,
      "step": 1515
    },
    {
      "epoch": 1.4154634373544481,
      "grad_norm": 1.2578125,
      "learning_rate": 1.1789415974835764e-05,
      "loss": 0.0402948796749115,
      "step": 1520
    },
    {
      "epoch": 1.4201210992081974,
      "grad_norm": 1.3125,
      "learning_rate": 1.1615562086302912e-05,
      "loss": 0.06786430478096009,
      "step": 1525
    },
    {
      "epoch": 1.424778761061947,
      "grad_norm": 0.859375,
      "learning_rate": 1.1442691347401783e-05,
      "loss": 0.04251951277256012,
      "step": 1530
    },
    {
      "epoch": 1.4294364229156964,
      "grad_norm": 1.53125,
      "learning_rate": 1.1270813002788303e-05,
      "loss": 0.06105688214302063,
      "step": 1535
    },
    {
      "epoch": 1.4340940847694457,
      "grad_norm": 1.25,
      "learning_rate": 1.1099936244047885e-05,
      "loss": 0.055726665258407596,
      "step": 1540
    },
    {
      "epoch": 1.4387517466231952,
      "grad_norm": 1.4375,
      "learning_rate": 1.0930070209203867e-05,
      "loss": 0.05416175723075867,
      "step": 1545
    },
    {
      "epoch": 1.4434094084769447,
      "grad_norm": 1.1484375,
      "learning_rate": 1.0761223982228775e-05,
      "loss": 0.05864529609680176,
      "step": 1550
    },
    {
      "epoch": 1.448067070330694,
      "grad_norm": 1.984375,
      "learning_rate": 1.0593406592558674e-05,
      "loss": 0.06089619398117065,
      "step": 1555
    },
    {
      "epoch": 1.4527247321844434,
      "grad_norm": 0.9375,
      "learning_rate": 1.042662701461013e-05,
      "loss": 0.047282624244689944,
      "step": 1560
    },
    {
      "epoch": 1.4573823940381927,
      "grad_norm": 0.8828125,
      "learning_rate": 1.0260894167300419e-05,
      "loss": 0.0594075083732605,
      "step": 1565
    },
    {
      "epoch": 1.4620400558919422,
      "grad_norm": 0.828125,
      "learning_rate": 1.0096216913570506e-05,
      "loss": 0.057616645097732545,
      "step": 1570
    },
    {
      "epoch": 1.4666977177456917,
      "grad_norm": 1.125,
      "learning_rate": 9.932604059911067e-06,
      "loss": 0.03644887804985046,
      "step": 1575
    },
    {
      "epoch": 1.4713553795994412,
      "grad_norm": 0.90234375,
      "learning_rate": 9.770064355891585e-06,
      "loss": 0.0486981213092804,
      "step": 1580
    },
    {
      "epoch": 1.4760130414531905,
      "grad_norm": 0.859375,
      "learning_rate": 9.608606493692426e-06,
      "loss": 0.04411686062812805,
      "step": 1585
    },
    {
      "epoch": 1.48067070330694,
      "grad_norm": 1.8671875,
      "learning_rate": 9.448239107639987e-06,
      "loss": 0.04628075659275055,
      "step": 1590
    },
    {
      "epoch": 1.4853283651606892,
      "grad_norm": 1.3046875,
      "learning_rate": 9.288970773744997e-06,
      "loss": 0.05574510097503662,
      "step": 1595
    },
    {
      "epoch": 1.4899860270144387,
      "grad_norm": 1.796875,
      "learning_rate": 9.13081000924384e-06,
      "loss": 0.05554535984992981,
      "step": 1600
    },
    {
      "epoch": 1.4946436888681882,
      "grad_norm": 1.2265625,
      "learning_rate": 8.973765272143129e-06,
      "loss": 0.04478040039539337,
      "step": 1605
    },
    {
      "epoch": 1.4993013507219377,
      "grad_norm": 0.87890625,
      "learning_rate": 8.817844960767386e-06,
      "loss": 0.04596472382545471,
      "step": 1610
    },
    {
      "epoch": 1.503959012575687,
      "grad_norm": 6.875,
      "learning_rate": 8.663057413309858e-06,
      "loss": 0.05729023814201355,
      "step": 1615
    },
    {
      "epoch": 1.5086166744294365,
      "grad_norm": 1.0,
      "learning_rate": 8.509410907386743e-06,
      "loss": 0.04612794816493988,
      "step": 1620
    },
    {
      "epoch": 1.5132743362831858,
      "grad_norm": 1.1171875,
      "learning_rate": 8.356913659594373e-06,
      "loss": 0.04824492633342743,
      "step": 1625
    },
    {
      "epoch": 1.5179319981369352,
      "grad_norm": 4.0,
      "learning_rate": 8.20557382506993e-06,
      "loss": 0.05382657051086426,
      "step": 1630
    },
    {
      "epoch": 1.5225896599906847,
      "grad_norm": 1.078125,
      "learning_rate": 8.05539949705529e-06,
      "loss": 0.042561736702919004,
      "step": 1635
    },
    {
      "epoch": 1.5272473218444342,
      "grad_norm": 1.1171875,
      "learning_rate": 7.90639870646418e-06,
      "loss": 0.04216844141483307,
      "step": 1640
    },
    {
      "epoch": 1.5319049836981835,
      "grad_norm": 0.7890625,
      "learning_rate": 7.758579421452786e-06,
      "loss": 0.06024673581123352,
      "step": 1645
    },
    {
      "epoch": 1.5365626455519328,
      "grad_norm": 0.76953125,
      "learning_rate": 7.611949546993579e-06,
      "loss": 0.03899524211883545,
      "step": 1650
    },
    {
      "epoch": 1.5412203074056823,
      "grad_norm": 0.92578125,
      "learning_rate": 7.466516924452593e-06,
      "loss": 0.05587719082832336,
      "step": 1655
    },
    {
      "epoch": 1.5458779692594318,
      "grad_norm": 1.0078125,
      "learning_rate": 7.322289331170099e-06,
      "loss": 0.06056018471717835,
      "step": 1660
    },
    {
      "epoch": 1.5505356311131813,
      "grad_norm": 1.3046875,
      "learning_rate": 7.179274480044695e-06,
      "loss": 0.06227887272834778,
      "step": 1665
    },
    {
      "epoch": 1.5551932929669308,
      "grad_norm": 1.578125,
      "learning_rate": 7.037480019120803e-06,
      "loss": 0.05833619236946106,
      "step": 1670
    },
    {
      "epoch": 1.55985095482068,
      "grad_norm": 1.3515625,
      "learning_rate": 6.896913531179733e-06,
      "loss": 0.06514707803726197,
      "step": 1675
    },
    {
      "epoch": 1.5645086166744293,
      "grad_norm": 1.3515625,
      "learning_rate": 6.757582533334142e-06,
      "loss": 0.0441954642534256,
      "step": 1680
    },
    {
      "epoch": 1.5691662785281788,
      "grad_norm": 1.296875,
      "learning_rate": 6.619494476626022e-06,
      "loss": 0.04353019595146179,
      "step": 1685
    },
    {
      "epoch": 1.5738239403819283,
      "grad_norm": 1.4453125,
      "learning_rate": 6.4826567456283095e-06,
      "loss": 0.06212781667709351,
      "step": 1690
    },
    {
      "epoch": 1.5784816022356778,
      "grad_norm": 1.25,
      "learning_rate": 6.347076658049872e-06,
      "loss": 0.04992106556892395,
      "step": 1695
    },
    {
      "epoch": 1.583139264089427,
      "grad_norm": 1.4375,
      "learning_rate": 6.212761464344267e-06,
      "loss": 0.05654107928276062,
      "step": 1700
    },
    {
      "epoch": 1.5877969259431766,
      "grad_norm": 1.171875,
      "learning_rate": 6.079718347321983e-06,
      "loss": 0.04213224053382873,
      "step": 1705
    },
    {
      "epoch": 1.5924545877969258,
      "grad_norm": 1.3359375,
      "learning_rate": 5.947954421766284e-06,
      "loss": 0.05712945461273193,
      "step": 1710
    },
    {
      "epoch": 1.5971122496506753,
      "grad_norm": 1.5,
      "learning_rate": 5.81747673405278e-06,
      "loss": 0.06145240068435669,
      "step": 1715
    },
    {
      "epoch": 1.6017699115044248,
      "grad_norm": 2.859375,
      "learning_rate": 5.688292261772593e-06,
      "loss": 0.05003199577331543,
      "step": 1720
    },
    {
      "epoch": 1.6064275733581743,
      "grad_norm": 1.0078125,
      "learning_rate": 5.560407913359198e-06,
      "loss": 0.05077915191650391,
      "step": 1725
    },
    {
      "epoch": 1.6110852352119236,
      "grad_norm": 1.0234375,
      "learning_rate": 5.4338305277189995e-06,
      "loss": 0.05965242385864258,
      "step": 1730
    },
    {
      "epoch": 1.6157428970656729,
      "grad_norm": 0.9921875,
      "learning_rate": 5.308566873865567e-06,
      "loss": 0.05326489806175232,
      "step": 1735
    },
    {
      "epoch": 1.6204005589194224,
      "grad_norm": 0.79296875,
      "learning_rate": 5.184623650557708e-06,
      "loss": 0.04797676503658295,
      "step": 1740
    },
    {
      "epoch": 1.6250582207731719,
      "grad_norm": 1.453125,
      "learning_rate": 5.062007485941196e-06,
      "loss": 0.05377109050750732,
      "step": 1745
    },
    {
      "epoch": 1.6297158826269214,
      "grad_norm": 1.3671875,
      "learning_rate": 4.9407249371943e-06,
      "loss": 0.06788500547409057,
      "step": 1750
    },
    {
      "epoch": 1.6343735444806708,
      "grad_norm": 1.3828125,
      "learning_rate": 4.820782490177166e-06,
      "loss": 0.0571264922618866,
      "step": 1755
    },
    {
      "epoch": 1.6390312063344201,
      "grad_norm": 1.4140625,
      "learning_rate": 4.702186559084972e-06,
      "loss": 0.05916694402694702,
      "step": 1760
    },
    {
      "epoch": 1.6436888681881694,
      "grad_norm": 1.0703125,
      "learning_rate": 4.584943486104851e-06,
      "loss": 0.051477736234664916,
      "step": 1765
    },
    {
      "epoch": 1.648346530041919,
      "grad_norm": 1.9296875,
      "learning_rate": 4.469059541076806e-06,
      "loss": 0.05232837200164795,
      "step": 1770
    },
    {
      "epoch": 1.6530041918956684,
      "grad_norm": 1.0,
      "learning_rate": 4.3545409211583554e-06,
      "loss": 0.047325053811073305,
      "step": 1775
    },
    {
      "epoch": 1.6576618537494179,
      "grad_norm": 1.4140625,
      "learning_rate": 4.24139375049317e-06,
      "loss": 0.052236026525497435,
      "step": 1780
    },
    {
      "epoch": 1.6623195156031674,
      "grad_norm": 1.03125,
      "learning_rate": 4.129624079883549e-06,
      "loss": 0.055283254384994505,
      "step": 1785
    },
    {
      "epoch": 1.6669771774569166,
      "grad_norm": 1.140625,
      "learning_rate": 4.019237886466839e-06,
      "loss": 0.044303908944129944,
      "step": 1790
    },
    {
      "epoch": 1.671634839310666,
      "grad_norm": 1.34375,
      "learning_rate": 3.910241073395805e-06,
      "loss": 0.06972193121910095,
      "step": 1795
    },
    {
      "epoch": 1.6762925011644154,
      "grad_norm": 1.6640625,
      "learning_rate": 3.8026394695229316e-06,
      "loss": 0.05040590763092041,
      "step": 1800
    },
    {
      "epoch": 1.680950163018165,
      "grad_norm": 1.1328125,
      "learning_rate": 3.696438829088713e-06,
      "loss": 0.060897302627563474,
      "step": 1805
    },
    {
      "epoch": 1.6856078248719144,
      "grad_norm": 1.0859375,
      "learning_rate": 3.591644831413946e-06,
      "loss": 0.06547480821609497,
      "step": 1810
    },
    {
      "epoch": 1.6902654867256637,
      "grad_norm": 0.875,
      "learning_rate": 3.48826308059601e-06,
      "loss": 0.040547126531600954,
      "step": 1815
    },
    {
      "epoch": 1.6949231485794132,
      "grad_norm": 1.3125,
      "learning_rate": 3.3862991052091508e-06,
      "loss": 0.04342193007469177,
      "step": 1820
    },
    {
      "epoch": 1.6995808104331624,
      "grad_norm": 1.03125,
      "learning_rate": 3.2857583580088814e-06,
      "loss": 0.040316405892372134,
      "step": 1825
    },
    {
      "epoch": 1.704238472286912,
      "grad_norm": 1.0234375,
      "learning_rate": 3.1866462156403177e-06,
      "loss": 0.049350231885910034,
      "step": 1830
    },
    {
      "epoch": 1.7088961341406614,
      "grad_norm": 1.2578125,
      "learning_rate": 3.088967978350704e-06,
      "loss": 0.06080756783485412,
      "step": 1835
    },
    {
      "epoch": 1.713553795994411,
      "grad_norm": 1.140625,
      "learning_rate": 2.9927288697059475e-06,
      "loss": 0.0580963134765625,
      "step": 1840
    },
    {
      "epoch": 1.7182114578481602,
      "grad_norm": 1.1640625,
      "learning_rate": 2.8979340363112592e-06,
      "loss": 0.05519581437110901,
      "step": 1845
    },
    {
      "epoch": 1.7228691197019095,
      "grad_norm": 1.4453125,
      "learning_rate": 2.804588547535968e-06,
      "loss": 0.04703545570373535,
      "step": 1850
    },
    {
      "epoch": 1.727526781555659,
      "grad_norm": 1.234375,
      "learning_rate": 2.7126973952423927e-06,
      "loss": 0.06626067757606506,
      "step": 1855
    },
    {
      "epoch": 1.7321844434094085,
      "grad_norm": 1.171875,
      "learning_rate": 2.622265493518907e-06,
      "loss": 0.06088282465934754,
      "step": 1860
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 0.953125,
      "learning_rate": 2.533297678417137e-06,
      "loss": 0.04956701695919037,
      "step": 1865
    },
    {
      "epoch": 1.7414997671169075,
      "grad_norm": 1.375,
      "learning_rate": 2.445798707693344e-06,
      "loss": 0.06391797065734864,
      "step": 1870
    },
    {
      "epoch": 1.7461574289706567,
      "grad_norm": 1.421875,
      "learning_rate": 2.3597732605539945e-06,
      "loss": 0.054262596368789676,
      "step": 1875
    },
    {
      "epoch": 1.750815090824406,
      "grad_norm": 1.03125,
      "learning_rate": 2.2752259374055473e-06,
      "loss": 0.04359312951564789,
      "step": 1880
    },
    {
      "epoch": 1.7554727526781555,
      "grad_norm": 0.84375,
      "learning_rate": 2.1921612596083917e-06,
      "loss": 0.053888702392578126,
      "step": 1885
    },
    {
      "epoch": 1.760130414531905,
      "grad_norm": 2.578125,
      "learning_rate": 2.110583669235098e-06,
      "loss": 0.05314147472381592,
      "step": 1890
    },
    {
      "epoch": 1.7647880763856545,
      "grad_norm": 1.171875,
      "learning_rate": 2.030497528832872e-06,
      "loss": 0.05475490689277649,
      "step": 1895
    },
    {
      "epoch": 1.7694457382394038,
      "grad_norm": 1.6875,
      "learning_rate": 1.951907121190202e-06,
      "loss": 0.05292813777923584,
      "step": 1900
    },
    {
      "epoch": 1.7741034000931533,
      "grad_norm": 1.15625,
      "learning_rate": 1.8748166491078978e-06,
      "loss": 0.04046391844749451,
      "step": 1905
    },
    {
      "epoch": 1.7787610619469025,
      "grad_norm": 1.1640625,
      "learning_rate": 1.799230235174294e-06,
      "loss": 0.05409615635871887,
      "step": 1910
    },
    {
      "epoch": 1.783418723800652,
      "grad_norm": 1.3203125,
      "learning_rate": 1.7251519215447809e-06,
      "loss": 0.05164237022399902,
      "step": 1915
    },
    {
      "epoch": 1.7880763856544015,
      "grad_norm": 1.359375,
      "learning_rate": 1.6525856697256748e-06,
      "loss": 0.04969433844089508,
      "step": 1920
    },
    {
      "epoch": 1.792734047508151,
      "grad_norm": 1.1484375,
      "learning_rate": 1.581535360362334e-06,
      "loss": 0.05390878915786743,
      "step": 1925
    },
    {
      "epoch": 1.7973917093619003,
      "grad_norm": 1.421875,
      "learning_rate": 1.512004793031655e-06,
      "loss": 0.04839572012424469,
      "step": 1930
    },
    {
      "epoch": 1.8020493712156498,
      "grad_norm": 1.4375,
      "learning_rate": 1.4439976860388715e-06,
      "loss": 0.052365976572036746,
      "step": 1935
    },
    {
      "epoch": 1.806707033069399,
      "grad_norm": 0.8671875,
      "learning_rate": 1.3775176762187036e-06,
      "loss": 0.05246726870536804,
      "step": 1940
    },
    {
      "epoch": 1.8113646949231486,
      "grad_norm": 1.6015625,
      "learning_rate": 1.3125683187408798e-06,
      "loss": 0.06147383451461792,
      "step": 1945
    },
    {
      "epoch": 1.816022356776898,
      "grad_norm": 2.609375,
      "learning_rate": 1.2491530869200196e-06,
      "loss": 0.0531234622001648,
      "step": 1950
    },
    {
      "epoch": 1.8206800186306475,
      "grad_norm": 1.0859375,
      "learning_rate": 1.187275372029868e-06,
      "loss": 0.04805976152420044,
      "step": 1955
    },
    {
      "epoch": 1.8253376804843968,
      "grad_norm": 1.0703125,
      "learning_rate": 1.1269384831219843e-06,
      "loss": 0.056858569383621216,
      "step": 1960
    },
    {
      "epoch": 1.829995342338146,
      "grad_norm": 1.1171875,
      "learning_rate": 1.0681456468487172e-06,
      "loss": 0.058970791101455686,
      "step": 1965
    },
    {
      "epoch": 1.8346530041918956,
      "grad_norm": 1.453125,
      "learning_rate": 1.010900007290716e-06,
      "loss": 0.0543209969997406,
      "step": 1970
    },
    {
      "epoch": 1.839310666045645,
      "grad_norm": 2.59375,
      "learning_rate": 9.552046257887614e-07,
      "loss": 0.047796967625617984,
      "step": 1975
    },
    {
      "epoch": 1.8439683278993946,
      "grad_norm": 1.0546875,
      "learning_rate": 9.010624807800471e-07,
      "loss": 0.05391690135002136,
      "step": 1980
    },
    {
      "epoch": 1.848625989753144,
      "grad_norm": 1.359375,
      "learning_rate": 8.484764676389201e-07,
      "loss": 0.07577978372573853,
      "step": 1985
    },
    {
      "epoch": 1.8532836516068933,
      "grad_norm": 1.140625,
      "learning_rate": 7.97449398522041e-07,
      "loss": 0.06300070285797119,
      "step": 1990
    },
    {
      "epoch": 1.8579413134606426,
      "grad_norm": 1.5703125,
      "learning_rate": 7.479840022179884e-07,
      "loss": 0.07114969491958618,
      "step": 1995
    },
    {
      "epoch": 1.8625989753143921,
      "grad_norm": 1.453125,
      "learning_rate": 7.000829240013384e-07,
      "loss": 0.07656757831573487,
      "step": 2000
    },
    {
      "epoch": 1.8672566371681416,
      "grad_norm": 0.9921875,
      "learning_rate": 6.537487254911911e-07,
      "loss": 0.05285425186157226,
      "step": 2005
    },
    {
      "epoch": 1.871914299021891,
      "grad_norm": 1.6171875,
      "learning_rate": 6.089838845141971e-07,
      "loss": 0.05697638392448425,
      "step": 2010
    },
    {
      "epoch": 1.8765719608756404,
      "grad_norm": 1.3203125,
      "learning_rate": 5.657907949720464e-07,
      "loss": 0.05292525291442871,
      "step": 2015
    },
    {
      "epoch": 1.8812296227293899,
      "grad_norm": 1.359375,
      "learning_rate": 5.241717667134382e-07,
      "loss": 0.040792858600616454,
      "step": 2020
    },
    {
      "epoch": 1.8858872845831391,
      "grad_norm": 9.125,
      "learning_rate": 4.84129025410569e-07,
      "loss": 0.07155822515487671,
      "step": 2025
    },
    {
      "epoch": 1.8905449464368886,
      "grad_norm": 1.3984375,
      "learning_rate": 4.456647124401081e-07,
      "loss": 0.059972846508026124,
      "step": 2030
    },
    {
      "epoch": 1.8952026082906381,
      "grad_norm": 1.2890625,
      "learning_rate": 4.0878088476867246e-07,
      "loss": 0.05164983868598938,
      "step": 2035
    },
    {
      "epoch": 1.8998602701443876,
      "grad_norm": 1.171875,
      "learning_rate": 3.7347951484283825e-07,
      "loss": 0.061475759744644164,
      "step": 2040
    },
    {
      "epoch": 1.904517931998137,
      "grad_norm": 3.75,
      "learning_rate": 3.397624904836549e-07,
      "loss": 0.06030047535896301,
      "step": 2045
    },
    {
      "epoch": 1.9091755938518864,
      "grad_norm": 0.984375,
      "learning_rate": 3.076316147856828e-07,
      "loss": 0.06555002927780151,
      "step": 2050
    },
    {
      "epoch": 1.9138332557056357,
      "grad_norm": 1.71875,
      "learning_rate": 2.770886060205835e-07,
      "loss": 0.03764947056770325,
      "step": 2055
    },
    {
      "epoch": 1.9184909175593852,
      "grad_norm": 1.296875,
      "learning_rate": 2.481350975452168e-07,
      "loss": 0.04990624189376831,
      "step": 2060
    },
    {
      "epoch": 1.9231485794131347,
      "grad_norm": 1.390625,
      "learning_rate": 2.2077263771430378e-07,
      "loss": 0.06039892435073853,
      "step": 2065
    },
    {
      "epoch": 1.9278062412668842,
      "grad_norm": 1.234375,
      "learning_rate": 1.9500268979761982e-07,
      "loss": 0.06764861345291137,
      "step": 2070
    },
    {
      "epoch": 1.9324639031206334,
      "grad_norm": 1.5234375,
      "learning_rate": 1.7082663190173376e-07,
      "loss": 0.05397939085960388,
      "step": 2075
    },
    {
      "epoch": 1.9371215649743827,
      "grad_norm": 1.4453125,
      "learning_rate": 1.482457568963369e-07,
      "loss": 0.058798480033874514,
      "step": 2080
    },
    {
      "epoch": 1.9417792268281322,
      "grad_norm": 1.4140625,
      "learning_rate": 1.27261272345075e-07,
      "loss": 0.036405691504478456,
      "step": 2085
    },
    {
      "epoch": 1.9464368886818817,
      "grad_norm": 1.1640625,
      "learning_rate": 1.0787430044098656e-07,
      "loss": 0.05242518186569214,
      "step": 2090
    },
    {
      "epoch": 1.9510945505356312,
      "grad_norm": 1.265625,
      "learning_rate": 9.008587794649081e-08,
      "loss": 0.06891694664955139,
      "step": 2095
    },
    {
      "epoch": 1.9557522123893807,
      "grad_norm": 1.6796875,
      "learning_rate": 7.389695613793879e-08,
      "loss": 0.060939884185791014,
      "step": 2100
    },
    {
      "epoch": 1.96040987424313,
      "grad_norm": 1.90625,
      "learning_rate": 5.9308400754747304e-08,
      "loss": 0.056053650379180905,
      "step": 2105
    },
    {
      "epoch": 1.9650675360968792,
      "grad_norm": 1.1484375,
      "learning_rate": 4.63209919530927e-08,
      "loss": 0.04752773642539978,
      "step": 2110
    },
    {
      "epoch": 1.9697251979506287,
      "grad_norm": 2.0625,
      "learning_rate": 3.493542426419749e-08,
      "loss": 0.05915160179138183,
      "step": 2115
    },
    {
      "epoch": 1.9743828598043782,
      "grad_norm": 1.28125,
      "learning_rate": 2.5152306557196802e-08,
      "loss": 0.052164262533187865,
      "step": 2120
    },
    {
      "epoch": 1.9790405216581277,
      "grad_norm": 1.15625,
      "learning_rate": 1.697216200655438e-08,
      "loss": 0.05432420372962952,
      "step": 2125
    },
    {
      "epoch": 1.983698183511877,
      "grad_norm": 1.4453125,
      "learning_rate": 1.0395428064108315e-08,
      "loss": 0.05430280566215515,
      "step": 2130
    },
    {
      "epoch": 1.9883558453656265,
      "grad_norm": 1.203125,
      "learning_rate": 5.422456435666412e-09,
      "loss": 0.04979383051395416,
      "step": 2135
    },
    {
      "epoch": 1.9930135072193758,
      "grad_norm": 1.0390625,
      "learning_rate": 2.05351306219459e-09,
      "loss": 0.05388739109039307,
      "step": 2140
    },
    {
      "epoch": 1.9976711690731253,
      "grad_norm": 1.0625,
      "learning_rate": 2.887781056015726e-10,
      "loss": 0.06473076939582825,
      "step": 2145
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.06230008229613304,
      "eval_pearson_cosine": 0.5569300055503845,
      "eval_pearson_dot_product": 0.5569300651550293,
      "eval_pearson_euclidean": 0.5638938546180725,
      "eval_pearson_manhattan": 0.5624173879623413,
      "eval_runtime": 17.219,
      "eval_samples_per_second": 118.126,
      "eval_spearman_cosine": 0.5637662854873432,
      "eval_spearman_dot_product": 0.5637662852863563,
      "eval_spearman_euclidean": 0.5637662854873432,
      "eval_spearman_manhattan": 0.5633536421411317,
      "eval_steps_per_second": 19.688,
      "step": 2148
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.06230008229613304,
      "eval_pearson_cosine": 0.5569300055503845,
      "eval_pearson_dot_product": 0.5569300651550293,
      "eval_pearson_euclidean": 0.5638938546180725,
      "eval_pearson_manhattan": 0.5624173879623413,
      "eval_runtime": 17.4857,
      "eval_samples_per_second": 116.324,
      "eval_spearman_cosine": 0.5637662854873432,
      "eval_spearman_dot_product": 0.5637662852863563,
      "eval_spearman_euclidean": 0.5637662854873432,
      "eval_spearman_manhattan": 0.5633536421411317,
      "eval_steps_per_second": 19.387,
      "step": 2148
    }
  ],
  "logging_steps": 5,
  "max_steps": 2148,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5428717251657728e+17,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}