{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.883248730964468,
  "eval_steps": 500,
  "global_step": 3500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025380710659898477,
      "grad_norm": 1.7737329006195068,
      "learning_rate": 4.9873096446700515e-05,
      "loss": 0.4425,
      "step": 10
    },
    {
      "epoch": 0.050761421319796954,
      "grad_norm": 1.6950019598007202,
      "learning_rate": 4.9746192893401014e-05,
      "loss": 0.5594,
      "step": 20
    },
    {
      "epoch": 0.07614213197969544,
      "grad_norm": 3.7862815856933594,
      "learning_rate": 4.961928934010153e-05,
      "loss": 0.5435,
      "step": 30
    },
    {
      "epoch": 0.10152284263959391,
      "grad_norm": 2.450183868408203,
      "learning_rate": 4.949238578680203e-05,
      "loss": 0.5429,
      "step": 40
    },
    {
      "epoch": 0.12690355329949238,
      "grad_norm": 1.6361172199249268,
      "learning_rate": 4.936548223350254e-05,
      "loss": 0.6185,
      "step": 50
    },
    {
      "epoch": 0.15228426395939088,
      "grad_norm": 1.3755431175231934,
      "learning_rate": 4.9238578680203045e-05,
      "loss": 0.5633,
      "step": 60
    },
    {
      "epoch": 0.17766497461928935,
      "grad_norm": 2.1979947090148926,
      "learning_rate": 4.911167512690356e-05,
      "loss": 0.5031,
      "step": 70
    },
    {
      "epoch": 0.20304568527918782,
      "grad_norm": 1.493660569190979,
      "learning_rate": 4.8984771573604064e-05,
      "loss": 0.5942,
      "step": 80
    },
    {
      "epoch": 0.22842639593908629,
      "grad_norm": 1.642001748085022,
      "learning_rate": 4.885786802030457e-05,
      "loss": 0.5513,
      "step": 90
    },
    {
      "epoch": 0.25380710659898476,
      "grad_norm": 4.11527681350708,
      "learning_rate": 4.873096446700508e-05,
      "loss": 0.5544,
      "step": 100
    },
    {
      "epoch": 0.27918781725888325,
      "grad_norm": 1.966141700744629,
      "learning_rate": 4.860406091370558e-05,
      "loss": 0.5311,
      "step": 110
    },
    {
      "epoch": 0.30456852791878175,
      "grad_norm": 1.490822196006775,
      "learning_rate": 4.8477157360406095e-05,
      "loss": 0.4546,
      "step": 120
    },
    {
      "epoch": 0.3299492385786802,
      "grad_norm": 2.575047254562378,
      "learning_rate": 4.83502538071066e-05,
      "loss": 0.4532,
      "step": 130
    },
    {
      "epoch": 0.3553299492385787,
      "grad_norm": 1.9474763870239258,
      "learning_rate": 4.822335025380711e-05,
      "loss": 0.4499,
      "step": 140
    },
    {
      "epoch": 0.38071065989847713,
      "grad_norm": 2.9648075103759766,
      "learning_rate": 4.809644670050762e-05,
      "loss": 0.5425,
      "step": 150
    },
    {
      "epoch": 0.40609137055837563,
      "grad_norm": 1.9473344087600708,
      "learning_rate": 4.7969543147208126e-05,
      "loss": 0.529,
      "step": 160
    },
    {
      "epoch": 0.43147208121827413,
      "grad_norm": 2.3414478302001953,
      "learning_rate": 4.784263959390863e-05,
      "loss": 0.5281,
      "step": 170
    },
    {
      "epoch": 0.45685279187817257,
      "grad_norm": 1.2157188653945923,
      "learning_rate": 4.771573604060914e-05,
      "loss": 0.4963,
      "step": 180
    },
    {
      "epoch": 0.48223350253807107,
      "grad_norm": 3.3916754722595215,
      "learning_rate": 4.758883248730965e-05,
      "loss": 0.4245,
      "step": 190
    },
    {
      "epoch": 0.5076142131979695,
      "grad_norm": 3.030148983001709,
      "learning_rate": 4.746192893401015e-05,
      "loss": 0.3352,
      "step": 200
    },
    {
      "epoch": 0.5329949238578681,
      "grad_norm": 2.1008453369140625,
      "learning_rate": 4.733502538071066e-05,
      "loss": 0.6109,
      "step": 210
    },
    {
      "epoch": 0.5583756345177665,
      "grad_norm": 1.6099292039871216,
      "learning_rate": 4.7208121827411175e-05,
      "loss": 0.5327,
      "step": 220
    },
    {
      "epoch": 0.583756345177665,
      "grad_norm": 1.7559535503387451,
      "learning_rate": 4.7081218274111674e-05,
      "loss": 0.4475,
      "step": 230
    },
    {
      "epoch": 0.6091370558375635,
      "grad_norm": 1.5413782596588135,
      "learning_rate": 4.695431472081219e-05,
      "loss": 0.534,
      "step": 240
    },
    {
      "epoch": 0.6345177664974619,
      "grad_norm": 1.810226321220398,
      "learning_rate": 4.682741116751269e-05,
      "loss": 0.5739,
      "step": 250
    },
    {
      "epoch": 0.6598984771573604,
      "grad_norm": 3.676072120666504,
      "learning_rate": 4.67005076142132e-05,
      "loss": 0.4702,
      "step": 260
    },
    {
      "epoch": 0.6852791878172588,
      "grad_norm": 2.691065788269043,
      "learning_rate": 4.6573604060913705e-05,
      "loss": 0.5858,
      "step": 270
    },
    {
      "epoch": 0.7106598984771574,
      "grad_norm": 1.8071280717849731,
      "learning_rate": 4.644670050761422e-05,
      "loss": 0.5294,
      "step": 280
    },
    {
      "epoch": 0.7360406091370558,
      "grad_norm": 4.747298240661621,
      "learning_rate": 4.631979695431472e-05,
      "loss": 0.4718,
      "step": 290
    },
    {
      "epoch": 0.7614213197969543,
      "grad_norm": 2.4223101139068604,
      "learning_rate": 4.619289340101523e-05,
      "loss": 0.4712,
      "step": 300
    },
    {
      "epoch": 0.7868020304568528,
      "grad_norm": 2.2267863750457764,
      "learning_rate": 4.606598984771574e-05,
      "loss": 0.5829,
      "step": 310
    },
    {
      "epoch": 0.8121827411167513,
      "grad_norm": 2.0561490058898926,
      "learning_rate": 4.593908629441624e-05,
      "loss": 0.4033,
      "step": 320
    },
    {
      "epoch": 0.8375634517766497,
      "grad_norm": 3.0562074184417725,
      "learning_rate": 4.5812182741116755e-05,
      "loss": 0.6535,
      "step": 330
    },
    {
      "epoch": 0.8629441624365483,
      "grad_norm": 3.355173110961914,
      "learning_rate": 4.568527918781726e-05,
      "loss": 0.4635,
      "step": 340
    },
    {
      "epoch": 0.8883248730964467,
      "grad_norm": 4.357395172119141,
      "learning_rate": 4.555837563451777e-05,
      "loss": 0.4995,
      "step": 350
    },
    {
      "epoch": 0.9137055837563451,
      "grad_norm": 1.0308293104171753,
      "learning_rate": 4.543147208121827e-05,
      "loss": 0.5207,
      "step": 360
    },
    {
      "epoch": 0.9390862944162437,
      "grad_norm": 1.9433645009994507,
      "learning_rate": 4.5304568527918786e-05,
      "loss": 0.2991,
      "step": 370
    },
    {
      "epoch": 0.9644670050761421,
      "grad_norm": 3.8325693607330322,
      "learning_rate": 4.517766497461929e-05,
      "loss": 0.6014,
      "step": 380
    },
    {
      "epoch": 0.9898477157360406,
      "grad_norm": 1.0364559888839722,
      "learning_rate": 4.50507614213198e-05,
      "loss": 0.3693,
      "step": 390
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.49558719992637634,
      "eval_runtime": 8.8144,
      "eval_samples_per_second": 89.4,
      "eval_steps_per_second": 11.232,
      "step": 394
    },
    {
      "epoch": 1.015228426395939,
      "grad_norm": 2.267185926437378,
      "learning_rate": 4.492385786802031e-05,
      "loss": 0.5173,
      "step": 400
    },
    {
      "epoch": 1.0406091370558375,
      "grad_norm": 1.7508792877197266,
      "learning_rate": 4.479695431472081e-05,
      "loss": 0.4722,
      "step": 410
    },
    {
      "epoch": 1.0659898477157361,
      "grad_norm": 1.0362215042114258,
      "learning_rate": 4.467005076142132e-05,
      "loss": 0.4053,
      "step": 420
    },
    {
      "epoch": 1.0913705583756346,
      "grad_norm": 1.5984660387039185,
      "learning_rate": 4.454314720812183e-05,
      "loss": 0.6716,
      "step": 430
    },
    {
      "epoch": 1.116751269035533,
      "grad_norm": 2.6108462810516357,
      "learning_rate": 4.4416243654822335e-05,
      "loss": 0.4829,
      "step": 440
    },
    {
      "epoch": 1.1421319796954315,
      "grad_norm": 6.843586444854736,
      "learning_rate": 4.428934010152285e-05,
      "loss": 0.4716,
      "step": 450
    },
    {
      "epoch": 1.16751269035533,
      "grad_norm": 5.383293151855469,
      "learning_rate": 4.416243654822335e-05,
      "loss": 0.6196,
      "step": 460
    },
    {
      "epoch": 1.1928934010152283,
      "grad_norm": 1.4819399118423462,
      "learning_rate": 4.403553299492386e-05,
      "loss": 0.4594,
      "step": 470
    },
    {
      "epoch": 1.218274111675127,
      "grad_norm": 3.468229055404663,
      "learning_rate": 4.3908629441624365e-05,
      "loss": 0.4838,
      "step": 480
    },
    {
      "epoch": 1.2436548223350254,
      "grad_norm": 3.6312994956970215,
      "learning_rate": 4.378172588832488e-05,
      "loss": 0.4686,
      "step": 490
    },
    {
      "epoch": 1.2690355329949239,
      "grad_norm": 1.4617408514022827,
      "learning_rate": 4.365482233502538e-05,
      "loss": 0.4094,
      "step": 500
    },
    {
      "epoch": 1.2944162436548223,
      "grad_norm": 0.9887295365333557,
      "learning_rate": 4.352791878172589e-05,
      "loss": 0.5219,
      "step": 510
    },
    {
      "epoch": 1.3197969543147208,
      "grad_norm": 1.6651699542999268,
      "learning_rate": 4.34010152284264e-05,
      "loss": 0.6159,
      "step": 520
    },
    {
      "epoch": 1.3451776649746192,
      "grad_norm": 2.0405914783477783,
      "learning_rate": 4.32741116751269e-05,
      "loss": 0.41,
      "step": 530
    },
    {
      "epoch": 1.3705583756345177,
      "grad_norm": 4.893404483795166,
      "learning_rate": 4.3147208121827415e-05,
      "loss": 0.5679,
      "step": 540
    },
    {
      "epoch": 1.3959390862944163,
      "grad_norm": 2.044116735458374,
      "learning_rate": 4.302030456852792e-05,
      "loss": 0.4693,
      "step": 550
    },
    {
      "epoch": 1.4213197969543148,
      "grad_norm": 2.040766477584839,
      "learning_rate": 4.289340101522843e-05,
      "loss": 0.4736,
      "step": 560
    },
    {
      "epoch": 1.4467005076142132,
      "grad_norm": 1.883665919303894,
      "learning_rate": 4.276649746192893e-05,
      "loss": 0.6475,
      "step": 570
    },
    {
      "epoch": 1.4720812182741116,
      "grad_norm": 1.6452559232711792,
      "learning_rate": 4.2639593908629446e-05,
      "loss": 0.5092,
      "step": 580
    },
    {
      "epoch": 1.49746192893401,
      "grad_norm": 2.059413194656372,
      "learning_rate": 4.251269035532995e-05,
      "loss": 0.5469,
      "step": 590
    },
    {
      "epoch": 1.5228426395939088,
      "grad_norm": 5.789554595947266,
      "learning_rate": 4.238578680203046e-05,
      "loss": 0.4548,
      "step": 600
    },
    {
      "epoch": 1.548223350253807,
      "grad_norm": 1.7683823108673096,
      "learning_rate": 4.225888324873097e-05,
      "loss": 0.3807,
      "step": 610
    },
    {
      "epoch": 1.5736040609137056,
      "grad_norm": 1.6267879009246826,
      "learning_rate": 4.213197969543147e-05,
      "loss": 0.3814,
      "step": 620
    },
    {
      "epoch": 1.598984771573604,
      "grad_norm": 2.024305582046509,
      "learning_rate": 4.200507614213198e-05,
      "loss": 0.5454,
      "step": 630
    },
    {
      "epoch": 1.6243654822335025,
      "grad_norm": 2.527076482772827,
      "learning_rate": 4.187817258883249e-05,
      "loss": 0.5996,
      "step": 640
    },
    {
      "epoch": 1.649746192893401,
      "grad_norm": 2.8863742351531982,
      "learning_rate": 4.1751269035532995e-05,
      "loss": 0.4548,
      "step": 650
    },
    {
      "epoch": 1.6751269035532994,
      "grad_norm": 1.8288835287094116,
      "learning_rate": 4.162436548223351e-05,
      "loss": 0.3666,
      "step": 660
    },
    {
      "epoch": 1.700507614213198,
      "grad_norm": 2.004929304122925,
      "learning_rate": 4.1497461928934013e-05,
      "loss": 0.4643,
      "step": 670
    },
    {
      "epoch": 1.7258883248730963,
      "grad_norm": 2.2006640434265137,
      "learning_rate": 4.137055837563452e-05,
      "loss": 0.5934,
      "step": 680
    },
    {
      "epoch": 1.751269035532995,
      "grad_norm": 1.2337054014205933,
      "learning_rate": 4.1243654822335025e-05,
      "loss": 0.5315,
      "step": 690
    },
    {
      "epoch": 1.7766497461928934,
      "grad_norm": 1.163268804550171,
      "learning_rate": 4.111675126903554e-05,
      "loss": 0.4484,
      "step": 700
    },
    {
      "epoch": 1.8020304568527918,
      "grad_norm": 1.2907366752624512,
      "learning_rate": 4.098984771573604e-05,
      "loss": 0.4027,
      "step": 710
    },
    {
      "epoch": 1.8274111675126905,
      "grad_norm": 1.8477003574371338,
      "learning_rate": 4.086294416243655e-05,
      "loss": 0.503,
      "step": 720
    },
    {
      "epoch": 1.8527918781725887,
      "grad_norm": 2.8408010005950928,
      "learning_rate": 4.073604060913706e-05,
      "loss": 0.5602,
      "step": 730
    },
    {
      "epoch": 1.8781725888324874,
      "grad_norm": 3.5004968643188477,
      "learning_rate": 4.060913705583756e-05,
      "loss": 0.5681,
      "step": 740
    },
    {
      "epoch": 1.9035532994923858,
      "grad_norm": 1.7411813735961914,
      "learning_rate": 4.0482233502538075e-05,
      "loss": 0.4501,
      "step": 750
    },
    {
      "epoch": 1.9289340101522843,
      "grad_norm": 4.301488876342773,
      "learning_rate": 4.035532994923858e-05,
      "loss": 0.5262,
      "step": 760
    },
    {
      "epoch": 1.9543147208121827,
      "grad_norm": 0.8759877681732178,
      "learning_rate": 4.022842639593909e-05,
      "loss": 0.4527,
      "step": 770
    },
    {
      "epoch": 1.9796954314720812,
      "grad_norm": 2.361971855163574,
      "learning_rate": 4.010152284263959e-05,
      "loss": 0.487,
      "step": 780
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4712839722633362,
      "eval_runtime": 8.7867,
      "eval_samples_per_second": 89.681,
      "eval_steps_per_second": 11.267,
      "step": 788
    },
    {
      "epoch": 2.00507614213198,
      "grad_norm": 2.517071485519409,
      "learning_rate": 3.9974619289340106e-05,
      "loss": 0.4159,
      "step": 790
    },
    {
      "epoch": 2.030456852791878,
      "grad_norm": 1.2710137367248535,
      "learning_rate": 3.9847715736040605e-05,
      "loss": 0.4777,
      "step": 800
    },
    {
      "epoch": 2.0558375634517767,
      "grad_norm": 3.463624954223633,
      "learning_rate": 3.972081218274112e-05,
      "loss": 0.4792,
      "step": 810
    },
    {
      "epoch": 2.081218274111675,
      "grad_norm": 1.5746160745620728,
      "learning_rate": 3.959390862944163e-05,
      "loss": 0.4457,
      "step": 820
    },
    {
      "epoch": 2.1065989847715736,
      "grad_norm": 1.3191742897033691,
      "learning_rate": 3.946700507614213e-05,
      "loss": 0.4615,
      "step": 830
    },
    {
      "epoch": 2.1319796954314723,
      "grad_norm": 4.860599040985107,
      "learning_rate": 3.934010152284264e-05,
      "loss": 0.4459,
      "step": 840
    },
    {
      "epoch": 2.1573604060913705,
      "grad_norm": 2.636868715286255,
      "learning_rate": 3.921319796954315e-05,
      "loss": 0.537,
      "step": 850
    },
    {
      "epoch": 2.182741116751269,
      "grad_norm": 1.672642469406128,
      "learning_rate": 3.9086294416243655e-05,
      "loss": 0.4848,
      "step": 860
    },
    {
      "epoch": 2.2081218274111674,
      "grad_norm": 8.49184799194336,
      "learning_rate": 3.895939086294416e-05,
      "loss": 0.4826,
      "step": 870
    },
    {
      "epoch": 2.233502538071066,
      "grad_norm": 1.6219210624694824,
      "learning_rate": 3.8832487309644673e-05,
      "loss": 0.3403,
      "step": 880
    },
    {
      "epoch": 2.2588832487309647,
      "grad_norm": 1.356152892112732,
      "learning_rate": 3.870558375634518e-05,
      "loss": 0.5415,
      "step": 890
    },
    {
      "epoch": 2.284263959390863,
      "grad_norm": 3.455899238586426,
      "learning_rate": 3.8578680203045685e-05,
      "loss": 0.4958,
      "step": 900
    },
    {
      "epoch": 2.3096446700507616,
      "grad_norm": 3.994067430496216,
      "learning_rate": 3.84517766497462e-05,
      "loss": 0.4324,
      "step": 910
    },
    {
      "epoch": 2.33502538071066,
      "grad_norm": 5.774569034576416,
      "learning_rate": 3.83248730964467e-05,
      "loss": 0.3871,
      "step": 920
    },
    {
      "epoch": 2.3604060913705585,
      "grad_norm": 4.524348258972168,
      "learning_rate": 3.819796954314721e-05,
      "loss": 0.5642,
      "step": 930
    },
    {
      "epoch": 2.3857868020304567,
      "grad_norm": 1.5101829767227173,
      "learning_rate": 3.8071065989847716e-05,
      "loss": 0.5396,
      "step": 940
    },
    {
      "epoch": 2.4111675126903553,
      "grad_norm": 1.4824628829956055,
      "learning_rate": 3.794416243654822e-05,
      "loss": 0.5178,
      "step": 950
    },
    {
      "epoch": 2.436548223350254,
      "grad_norm": 1.3847397565841675,
      "learning_rate": 3.7817258883248735e-05,
      "loss": 0.3747,
      "step": 960
    },
    {
      "epoch": 2.4619289340101522,
      "grad_norm": 2.138828992843628,
      "learning_rate": 3.769035532994924e-05,
      "loss": 0.5756,
      "step": 970
    },
    {
      "epoch": 2.487309644670051,
      "grad_norm": 2.0886220932006836,
      "learning_rate": 3.756345177664975e-05,
      "loss": 0.5496,
      "step": 980
    },
    {
      "epoch": 2.512690355329949,
      "grad_norm": 2.8054535388946533,
      "learning_rate": 3.743654822335025e-05,
      "loss": 0.5984,
      "step": 990
    },
    {
      "epoch": 2.5380710659898478,
      "grad_norm": 3.6035268306732178,
      "learning_rate": 3.7309644670050766e-05,
      "loss": 0.4475,
      "step": 1000
    },
    {
      "epoch": 2.563451776649746,
      "grad_norm": 2.931821823120117,
      "learning_rate": 3.7182741116751265e-05,
      "loss": 0.5179,
      "step": 1010
    },
    {
      "epoch": 2.5888324873096447,
      "grad_norm": 1.8696964979171753,
      "learning_rate": 3.705583756345178e-05,
      "loss": 0.6541,
      "step": 1020
    },
    {
      "epoch": 2.6142131979695433,
      "grad_norm": 1.36885666847229,
      "learning_rate": 3.692893401015229e-05,
      "loss": 0.3136,
      "step": 1030
    },
    {
      "epoch": 2.6395939086294415,
      "grad_norm": 3.520601511001587,
      "learning_rate": 3.680203045685279e-05,
      "loss": 0.4434,
      "step": 1040
    },
    {
      "epoch": 2.66497461928934,
      "grad_norm": 3.95267915725708,
      "learning_rate": 3.66751269035533e-05,
      "loss": 0.5012,
      "step": 1050
    },
    {
      "epoch": 2.6903553299492384,
      "grad_norm": 2.064523935317993,
      "learning_rate": 3.654822335025381e-05,
      "loss": 0.3251,
      "step": 1060
    },
    {
      "epoch": 2.715736040609137,
      "grad_norm": 3.3240392208099365,
      "learning_rate": 3.6421319796954315e-05,
      "loss": 0.7704,
      "step": 1070
    },
    {
      "epoch": 2.7411167512690353,
      "grad_norm": 3.2816076278686523,
      "learning_rate": 3.629441624365482e-05,
      "loss": 0.4753,
      "step": 1080
    },
    {
      "epoch": 2.766497461928934,
      "grad_norm": 2.7639636993408203,
      "learning_rate": 3.6167512690355334e-05,
      "loss": 0.4443,
      "step": 1090
    },
    {
      "epoch": 2.7918781725888326,
      "grad_norm": 2.5042450428009033,
      "learning_rate": 3.604060913705584e-05,
      "loss": 0.4979,
      "step": 1100
    },
    {
      "epoch": 2.817258883248731,
      "grad_norm": 2.695377826690674,
      "learning_rate": 3.5913705583756346e-05,
      "loss": 0.4835,
      "step": 1110
    },
    {
      "epoch": 2.8426395939086295,
      "grad_norm": 2.2236552238464355,
      "learning_rate": 3.578680203045686e-05,
      "loss": 0.5162,
      "step": 1120
    },
    {
      "epoch": 2.868020304568528,
      "grad_norm": 5.009507179260254,
      "learning_rate": 3.565989847715736e-05,
      "loss": 0.4046,
      "step": 1130
    },
    {
      "epoch": 2.8934010152284264,
      "grad_norm": 2.037398338317871,
      "learning_rate": 3.553299492385787e-05,
      "loss": 0.4502,
      "step": 1140
    },
    {
      "epoch": 2.9187817258883246,
      "grad_norm": 5.31472635269165,
      "learning_rate": 3.5406091370558376e-05,
      "loss": 0.6371,
      "step": 1150
    },
    {
      "epoch": 2.9441624365482233,
      "grad_norm": 1.666244387626648,
      "learning_rate": 3.527918781725888e-05,
      "loss": 0.4088,
      "step": 1160
    },
    {
      "epoch": 2.969543147208122,
      "grad_norm": 4.006988525390625,
      "learning_rate": 3.5152284263959395e-05,
      "loss": 0.4483,
      "step": 1170
    },
    {
      "epoch": 2.99492385786802,
      "grad_norm": 2.429619312286377,
      "learning_rate": 3.50253807106599e-05,
      "loss": 0.4617,
      "step": 1180
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.5104432106018066,
      "eval_runtime": 8.8793,
      "eval_samples_per_second": 88.746,
      "eval_steps_per_second": 11.15,
      "step": 1182
    },
    {
      "epoch": 3.020304568527919,
      "grad_norm": 1.9103385210037231,
      "learning_rate": 3.489847715736041e-05,
      "loss": 0.461,
      "step": 1190
    },
    {
      "epoch": 3.045685279187817,
      "grad_norm": 2.838453769683838,
      "learning_rate": 3.477157360406091e-05,
      "loss": 0.4661,
      "step": 1200
    },
    {
      "epoch": 3.0710659898477157,
      "grad_norm": 2.706493616104126,
      "learning_rate": 3.4644670050761426e-05,
      "loss": 0.4335,
      "step": 1210
    },
    {
      "epoch": 3.0964467005076144,
      "grad_norm": 4.179074287414551,
      "learning_rate": 3.451776649746193e-05,
      "loss": 0.3517,
      "step": 1220
    },
    {
      "epoch": 3.1218274111675126,
      "grad_norm": 1.1146340370178223,
      "learning_rate": 3.439086294416244e-05,
      "loss": 0.2261,
      "step": 1230
    },
    {
      "epoch": 3.1472081218274113,
      "grad_norm": 2.5486109256744385,
      "learning_rate": 3.4263959390862944e-05,
      "loss": 0.3664,
      "step": 1240
    },
    {
      "epoch": 3.1725888324873095,
      "grad_norm": 2.886986017227173,
      "learning_rate": 3.413705583756345e-05,
      "loss": 0.3861,
      "step": 1250
    },
    {
      "epoch": 3.197969543147208,
      "grad_norm": 3.5457489490509033,
      "learning_rate": 3.401015228426396e-05,
      "loss": 0.5043,
      "step": 1260
    },
    {
      "epoch": 3.223350253807107,
      "grad_norm": 4.039999485015869,
      "learning_rate": 3.388324873096447e-05,
      "loss": 0.3362,
      "step": 1270
    },
    {
      "epoch": 3.248730964467005,
      "grad_norm": 4.554147720336914,
      "learning_rate": 3.3756345177664975e-05,
      "loss": 0.4709,
      "step": 1280
    },
    {
      "epoch": 3.2741116751269037,
      "grad_norm": 2.4214282035827637,
      "learning_rate": 3.362944162436548e-05,
      "loss": 0.3041,
      "step": 1290
    },
    {
      "epoch": 3.299492385786802,
      "grad_norm": 4.587721824645996,
      "learning_rate": 3.3502538071065994e-05,
      "loss": 0.6426,
      "step": 1300
    },
    {
      "epoch": 3.3248730964467006,
      "grad_norm": 2.0584123134613037,
      "learning_rate": 3.33756345177665e-05,
      "loss": 0.3861,
      "step": 1310
    },
    {
      "epoch": 3.350253807106599,
      "grad_norm": 2.152904510498047,
      "learning_rate": 3.3248730964467006e-05,
      "loss": 0.434,
      "step": 1320
    },
    {
      "epoch": 3.3756345177664975,
      "grad_norm": 3.3086836338043213,
      "learning_rate": 3.312182741116752e-05,
      "loss": 0.406,
      "step": 1330
    },
    {
      "epoch": 3.401015228426396,
      "grad_norm": 2.4695167541503906,
      "learning_rate": 3.299492385786802e-05,
      "loss": 0.4226,
      "step": 1340
    },
    {
      "epoch": 3.4263959390862944,
      "grad_norm": 2.482093334197998,
      "learning_rate": 3.286802030456853e-05,
      "loss": 0.335,
      "step": 1350
    },
    {
      "epoch": 3.451776649746193,
      "grad_norm": 6.9000959396362305,
      "learning_rate": 3.2741116751269036e-05,
      "loss": 0.3506,
      "step": 1360
    },
    {
      "epoch": 3.4771573604060912,
      "grad_norm": 2.6196579933166504,
      "learning_rate": 3.261421319796954e-05,
      "loss": 0.3377,
      "step": 1370
    },
    {
      "epoch": 3.50253807106599,
      "grad_norm": 1.8537606000900269,
      "learning_rate": 3.248730964467005e-05,
      "loss": 0.399,
      "step": 1380
    },
    {
      "epoch": 3.527918781725888,
      "grad_norm": 7.022123336791992,
      "learning_rate": 3.236040609137056e-05,
      "loss": 0.4255,
      "step": 1390
    },
    {
      "epoch": 3.553299492385787,
      "grad_norm": 3.112264394760132,
      "learning_rate": 3.223350253807107e-05,
      "loss": 0.399,
      "step": 1400
    },
    {
      "epoch": 3.5786802030456855,
      "grad_norm": 4.884408473968506,
      "learning_rate": 3.210659898477157e-05,
      "loss": 0.3533,
      "step": 1410
    },
    {
      "epoch": 3.6040609137055837,
      "grad_norm": 3.7339515686035156,
      "learning_rate": 3.1979695431472086e-05,
      "loss": 0.4329,
      "step": 1420
    },
    {
      "epoch": 3.6294416243654823,
      "grad_norm": 11.797500610351562,
      "learning_rate": 3.185279187817259e-05,
      "loss": 0.4496,
      "step": 1430
    },
    {
      "epoch": 3.6548223350253806,
      "grad_norm": 2.509971857070923,
      "learning_rate": 3.17258883248731e-05,
      "loss": 0.3758,
      "step": 1440
    },
    {
      "epoch": 3.6802030456852792,
      "grad_norm": 2.554452419281006,
      "learning_rate": 3.1598984771573604e-05,
      "loss": 0.2789,
      "step": 1450
    },
    {
      "epoch": 3.7055837563451774,
      "grad_norm": 13.046540260314941,
      "learning_rate": 3.147208121827411e-05,
      "loss": 0.52,
      "step": 1460
    },
    {
      "epoch": 3.730964467005076,
      "grad_norm": 6.821984767913818,
      "learning_rate": 3.134517766497462e-05,
      "loss": 0.4285,
      "step": 1470
    },
    {
      "epoch": 3.7563451776649748,
      "grad_norm": 2.692779064178467,
      "learning_rate": 3.121827411167513e-05,
      "loss": 0.2977,
      "step": 1480
    },
    {
      "epoch": 3.781725888324873,
      "grad_norm": 3.510523796081543,
      "learning_rate": 3.1091370558375635e-05,
      "loss": 0.4887,
      "step": 1490
    },
    {
      "epoch": 3.8071065989847717,
      "grad_norm": 4.374626159667969,
      "learning_rate": 3.096446700507614e-05,
      "loss": 0.2093,
      "step": 1500
    },
    {
      "epoch": 3.8324873096446703,
      "grad_norm": 0.6560042500495911,
      "learning_rate": 3.0837563451776654e-05,
      "loss": 0.613,
      "step": 1510
    },
    {
      "epoch": 3.8578680203045685,
      "grad_norm": 4.9201483726501465,
      "learning_rate": 3.071065989847716e-05,
      "loss": 0.4483,
      "step": 1520
    },
    {
      "epoch": 3.8832487309644668,
      "grad_norm": 2.497208833694458,
      "learning_rate": 3.0583756345177666e-05,
      "loss": 0.4434,
      "step": 1530
    },
    {
      "epoch": 3.9086294416243654,
      "grad_norm": 5.912675380706787,
      "learning_rate": 3.0456852791878175e-05,
      "loss": 0.4259,
      "step": 1540
    },
    {
      "epoch": 3.934010152284264,
      "grad_norm": 1.8950650691986084,
      "learning_rate": 3.032994923857868e-05,
      "loss": 0.4649,
      "step": 1550
    },
    {
      "epoch": 3.9593908629441623,
      "grad_norm": 15.265602111816406,
      "learning_rate": 3.020304568527919e-05,
      "loss": 0.4176,
      "step": 1560
    },
    {
      "epoch": 3.984771573604061,
      "grad_norm": 6.062721252441406,
      "learning_rate": 3.0076142131979696e-05,
      "loss": 0.3854,
      "step": 1570
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.5223032832145691,
      "eval_runtime": 8.7536,
      "eval_samples_per_second": 90.02,
      "eval_steps_per_second": 11.31,
      "step": 1576
    },
    {
      "epoch": 4.01015228426396,
      "grad_norm": 4.975412368774414,
      "learning_rate": 2.9949238578680206e-05,
      "loss": 0.4798,
      "step": 1580
    },
    {
      "epoch": 4.035532994923858,
      "grad_norm": 5.022281646728516,
      "learning_rate": 2.982233502538071e-05,
      "loss": 0.3322,
      "step": 1590
    },
    {
      "epoch": 4.060913705583756,
      "grad_norm": 5.229846000671387,
      "learning_rate": 2.969543147208122e-05,
      "loss": 0.2691,
      "step": 1600
    },
    {
      "epoch": 4.086294416243655,
      "grad_norm": 21.07980728149414,
      "learning_rate": 2.956852791878173e-05,
      "loss": 0.285,
      "step": 1610
    },
    {
      "epoch": 4.111675126903553,
      "grad_norm": 28.23366355895996,
      "learning_rate": 2.9441624365482233e-05,
      "loss": 0.1946,
      "step": 1620
    },
    {
      "epoch": 4.137055837563452,
      "grad_norm": 0.49203088879585266,
      "learning_rate": 2.9314720812182743e-05,
      "loss": 0.2176,
      "step": 1630
    },
    {
      "epoch": 4.16243654822335,
      "grad_norm": 6.196091175079346,
      "learning_rate": 2.918781725888325e-05,
      "loss": 0.1592,
      "step": 1640
    },
    {
      "epoch": 4.187817258883249,
      "grad_norm": 0.43765848875045776,
      "learning_rate": 2.9060913705583758e-05,
      "loss": 0.3764,
      "step": 1650
    },
    {
      "epoch": 4.213197969543147,
      "grad_norm": 4.543421745300293,
      "learning_rate": 2.8934010152284264e-05,
      "loss": 0.3371,
      "step": 1660
    },
    {
      "epoch": 4.238578680203045,
      "grad_norm": 2.611579418182373,
      "learning_rate": 2.8807106598984774e-05,
      "loss": 0.3052,
      "step": 1670
    },
    {
      "epoch": 4.2639593908629445,
      "grad_norm": 3.3673095703125,
      "learning_rate": 2.8680203045685283e-05,
      "loss": 0.1776,
      "step": 1680
    },
    {
      "epoch": 4.289340101522843,
      "grad_norm": 0.7387822270393372,
      "learning_rate": 2.855329949238579e-05,
      "loss": 0.3827,
      "step": 1690
    },
    {
      "epoch": 4.314720812182741,
      "grad_norm": 16.637300491333008,
      "learning_rate": 2.84263959390863e-05,
      "loss": 0.3993,
      "step": 1700
    },
    {
      "epoch": 4.340101522842639,
      "grad_norm": 6.28633975982666,
      "learning_rate": 2.82994923857868e-05,
      "loss": 0.3733,
      "step": 1710
    },
    {
      "epoch": 4.365482233502538,
      "grad_norm": 3.3295745849609375,
      "learning_rate": 2.8172588832487314e-05,
      "loss": 0.1166,
      "step": 1720
    },
    {
      "epoch": 4.3908629441624365,
      "grad_norm": 14.652837753295898,
      "learning_rate": 2.8045685279187816e-05,
      "loss": 0.3326,
      "step": 1730
    },
    {
      "epoch": 4.416243654822335,
      "grad_norm": 31.232574462890625,
      "learning_rate": 2.7918781725888326e-05,
      "loss": 0.2127,
      "step": 1740
    },
    {
      "epoch": 4.441624365482234,
      "grad_norm": 12.528191566467285,
      "learning_rate": 2.7791878172588832e-05,
      "loss": 0.4061,
      "step": 1750
    },
    {
      "epoch": 4.467005076142132,
      "grad_norm": 1.7689141035079956,
      "learning_rate": 2.766497461928934e-05,
      "loss": 0.3366,
      "step": 1760
    },
    {
      "epoch": 4.49238578680203,
      "grad_norm": 4.724018096923828,
      "learning_rate": 2.753807106598985e-05,
      "loss": 0.2359,
      "step": 1770
    },
    {
      "epoch": 4.517766497461929,
      "grad_norm": 1.561919093132019,
      "learning_rate": 2.7411167512690357e-05,
      "loss": 0.3106,
      "step": 1780
    },
    {
      "epoch": 4.543147208121828,
      "grad_norm": 5.754579067230225,
      "learning_rate": 2.7284263959390866e-05,
      "loss": 0.2982,
      "step": 1790
    },
    {
      "epoch": 4.568527918781726,
      "grad_norm": 1.181515097618103,
      "learning_rate": 2.715736040609137e-05,
      "loss": 0.2384,
      "step": 1800
    },
    {
      "epoch": 4.593908629441624,
      "grad_norm": 7.012800693511963,
      "learning_rate": 2.703045685279188e-05,
      "loss": 0.394,
      "step": 1810
    },
    {
      "epoch": 4.619289340101523,
      "grad_norm": 0.49557140469551086,
      "learning_rate": 2.6903553299492384e-05,
      "loss": 0.2869,
      "step": 1820
    },
    {
      "epoch": 4.644670050761421,
      "grad_norm": 3.6110830307006836,
      "learning_rate": 2.6776649746192893e-05,
      "loss": 0.3724,
      "step": 1830
    },
    {
      "epoch": 4.67005076142132,
      "grad_norm": 3.3351943492889404,
      "learning_rate": 2.6649746192893406e-05,
      "loss": 0.2966,
      "step": 1840
    },
    {
      "epoch": 4.695431472081218,
      "grad_norm": 1.342678427696228,
      "learning_rate": 2.652284263959391e-05,
      "loss": 0.3826,
      "step": 1850
    },
    {
      "epoch": 4.720812182741117,
      "grad_norm": 0.4344848096370697,
      "learning_rate": 2.6395939086294418e-05,
      "loss": 0.1962,
      "step": 1860
    },
    {
      "epoch": 4.746192893401015,
      "grad_norm": 56.233985900878906,
      "learning_rate": 2.6269035532994924e-05,
      "loss": 0.3758,
      "step": 1870
    },
    {
      "epoch": 4.771573604060913,
      "grad_norm": 10.647058486938477,
      "learning_rate": 2.6142131979695434e-05,
      "loss": 0.356,
      "step": 1880
    },
    {
      "epoch": 4.7969543147208125,
      "grad_norm": 0.9422564506530762,
      "learning_rate": 2.6015228426395936e-05,
      "loss": 0.2271,
      "step": 1890
    },
    {
      "epoch": 4.822335025380711,
      "grad_norm": 0.22354674339294434,
      "learning_rate": 2.588832487309645e-05,
      "loss": 0.1374,
      "step": 1900
    },
    {
      "epoch": 4.847715736040609,
      "grad_norm": 8.779372215270996,
      "learning_rate": 2.576142131979696e-05,
      "loss": 0.329,
      "step": 1910
    },
    {
      "epoch": 4.873096446700508,
      "grad_norm": 41.76162338256836,
      "learning_rate": 2.563451776649746e-05,
      "loss": 0.5914,
      "step": 1920
    },
    {
      "epoch": 4.898477157360406,
      "grad_norm": 26.878578186035156,
      "learning_rate": 2.5507614213197974e-05,
      "loss": 0.5029,
      "step": 1930
    },
    {
      "epoch": 4.9238578680203045,
      "grad_norm": 0.41534122824668884,
      "learning_rate": 2.5380710659898476e-05,
      "loss": 0.1997,
      "step": 1940
    },
    {
      "epoch": 4.949238578680203,
      "grad_norm": 4.708491325378418,
      "learning_rate": 2.5253807106598986e-05,
      "loss": 0.3183,
      "step": 1950
    },
    {
      "epoch": 4.974619289340102,
      "grad_norm": 0.31694722175598145,
      "learning_rate": 2.5126903553299492e-05,
      "loss": 0.3627,
      "step": 1960
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.7903828620910645,
      "learning_rate": 2.5e-05,
      "loss": 0.454,
      "step": 1970
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.6968420743942261,
      "eval_runtime": 8.7659,
      "eval_samples_per_second": 89.894,
      "eval_steps_per_second": 11.294,
      "step": 1970
    },
    {
      "epoch": 5.025380710659898,
      "grad_norm": 0.33573099970817566,
      "learning_rate": 2.4873096446700507e-05,
      "loss": 0.0922,
      "step": 1980
    },
    {
      "epoch": 5.050761421319797,
      "grad_norm": 0.38098305463790894,
      "learning_rate": 2.4746192893401017e-05,
      "loss": 0.1296,
      "step": 1990
    },
    {
      "epoch": 5.0761421319796955,
      "grad_norm": 6.538486480712891,
      "learning_rate": 2.4619289340101523e-05,
      "loss": 0.2388,
      "step": 2000
    },
    {
      "epoch": 5.101522842639594,
      "grad_norm": 0.3389257788658142,
      "learning_rate": 2.4492385786802032e-05,
      "loss": 0.1525,
      "step": 2010
    },
    {
      "epoch": 5.126903553299492,
      "grad_norm": 29.134950637817383,
      "learning_rate": 2.436548223350254e-05,
      "loss": 0.2023,
      "step": 2020
    },
    {
      "epoch": 5.152284263959391,
      "grad_norm": 2.5168628692626953,
      "learning_rate": 2.4238578680203047e-05,
      "loss": 0.1126,
      "step": 2030
    },
    {
      "epoch": 5.177664974619289,
      "grad_norm": 0.4162174463272095,
      "learning_rate": 2.4111675126903553e-05,
      "loss": 0.1466,
      "step": 2040
    },
    {
      "epoch": 5.2030456852791875,
      "grad_norm": 0.08261796087026596,
      "learning_rate": 2.3984771573604063e-05,
      "loss": 0.1032,
      "step": 2050
    },
    {
      "epoch": 5.228426395939087,
      "grad_norm": 0.13249030709266663,
      "learning_rate": 2.385786802030457e-05,
      "loss": 0.0117,
      "step": 2060
    },
    {
      "epoch": 5.253807106598985,
      "grad_norm": 54.72965621948242,
      "learning_rate": 2.3730964467005075e-05,
      "loss": 0.1383,
      "step": 2070
    },
    {
      "epoch": 5.279187817258883,
      "grad_norm": 10.03510856628418,
      "learning_rate": 2.3604060913705588e-05,
      "loss": 0.3034,
      "step": 2080
    },
    {
      "epoch": 5.304568527918782,
      "grad_norm": 3.790813446044922,
      "learning_rate": 2.3477157360406094e-05,
      "loss": 0.0487,
      "step": 2090
    },
    {
      "epoch": 5.32994923857868,
      "grad_norm": 10.79232120513916,
      "learning_rate": 2.33502538071066e-05,
      "loss": 0.1897,
      "step": 2100
    },
    {
      "epoch": 5.355329949238579,
      "grad_norm": 3.8837833404541016,
      "learning_rate": 2.322335025380711e-05,
      "loss": 0.5641,
      "step": 2110
    },
    {
      "epoch": 5.380710659898477,
      "grad_norm": 0.5432278513908386,
      "learning_rate": 2.3096446700507615e-05,
      "loss": 0.3232,
      "step": 2120
    },
    {
      "epoch": 5.406091370558376,
      "grad_norm": 2.8115179538726807,
      "learning_rate": 2.296954314720812e-05,
      "loss": 0.2467,
      "step": 2130
    },
    {
      "epoch": 5.431472081218274,
      "grad_norm": 0.23173533380031586,
      "learning_rate": 2.284263959390863e-05,
      "loss": 0.0366,
      "step": 2140
    },
    {
      "epoch": 5.456852791878172,
      "grad_norm": 0.22747744619846344,
      "learning_rate": 2.2715736040609136e-05,
      "loss": 0.2833,
      "step": 2150
    },
    {
      "epoch": 5.482233502538071,
      "grad_norm": 0.18748463690280914,
      "learning_rate": 2.2588832487309646e-05,
      "loss": 0.2338,
      "step": 2160
    },
    {
      "epoch": 5.50761421319797,
      "grad_norm": 0.2156940996646881,
      "learning_rate": 2.2461928934010155e-05,
      "loss": 0.0311,
      "step": 2170
    },
    {
      "epoch": 5.532994923857868,
      "grad_norm": 26.930912017822266,
      "learning_rate": 2.233502538071066e-05,
      "loss": 0.2466,
      "step": 2180
    },
    {
      "epoch": 5.558375634517766,
      "grad_norm": 23.53707504272461,
      "learning_rate": 2.2208121827411167e-05,
      "loss": 0.2277,
      "step": 2190
    },
    {
      "epoch": 5.583756345177665,
      "grad_norm": 6.032989025115967,
      "learning_rate": 2.2081218274111677e-05,
      "loss": 0.1784,
      "step": 2200
    },
    {
      "epoch": 5.6091370558375635,
      "grad_norm": 15.71194076538086,
      "learning_rate": 2.1954314720812183e-05,
      "loss": 0.2382,
      "step": 2210
    },
    {
      "epoch": 5.634517766497462,
      "grad_norm": 6.757187366485596,
      "learning_rate": 2.182741116751269e-05,
      "loss": 0.2233,
      "step": 2220
    },
    {
      "epoch": 5.659898477157361,
      "grad_norm": 6.697112560272217,
      "learning_rate": 2.17005076142132e-05,
      "loss": 0.2094,
      "step": 2230
    },
    {
      "epoch": 5.685279187817259,
      "grad_norm": 12.990015983581543,
      "learning_rate": 2.1573604060913707e-05,
      "loss": 0.167,
      "step": 2240
    },
    {
      "epoch": 5.710659898477157,
      "grad_norm": 0.234835684299469,
      "learning_rate": 2.1446700507614213e-05,
      "loss": 0.1446,
      "step": 2250
    },
    {
      "epoch": 5.7360406091370555,
      "grad_norm": 6.382123947143555,
      "learning_rate": 2.1319796954314723e-05,
      "loss": 0.2736,
      "step": 2260
    },
    {
      "epoch": 5.761421319796955,
      "grad_norm": 0.26427528262138367,
      "learning_rate": 2.119289340101523e-05,
      "loss": 0.1443,
      "step": 2270
    },
    {
      "epoch": 5.786802030456853,
      "grad_norm": 0.2004345953464508,
      "learning_rate": 2.1065989847715735e-05,
      "loss": 0.1685,
      "step": 2280
    },
    {
      "epoch": 5.812182741116751,
      "grad_norm": 0.21756359934806824,
      "learning_rate": 2.0939086294416244e-05,
      "loss": 0.385,
      "step": 2290
    },
    {
      "epoch": 5.837563451776649,
      "grad_norm": 5.085987567901611,
      "learning_rate": 2.0812182741116754e-05,
      "loss": 0.233,
      "step": 2300
    },
    {
      "epoch": 5.862944162436548,
      "grad_norm": 0.2375006377696991,
      "learning_rate": 2.068527918781726e-05,
      "loss": 0.2106,
      "step": 2310
    },
    {
      "epoch": 5.888324873096447,
      "grad_norm": 11.780427932739258,
      "learning_rate": 2.055837563451777e-05,
      "loss": 0.2142,
      "step": 2320
    },
    {
      "epoch": 5.913705583756345,
      "grad_norm": 1.2173970937728882,
      "learning_rate": 2.0431472081218275e-05,
      "loss": 0.0754,
      "step": 2330
    },
    {
      "epoch": 5.939086294416244,
      "grad_norm": 6.028761386871338,
      "learning_rate": 2.030456852791878e-05,
      "loss": 0.4157,
      "step": 2340
    },
    {
      "epoch": 5.964467005076142,
      "grad_norm": 19.581605911254883,
      "learning_rate": 2.017766497461929e-05,
      "loss": 0.1866,
      "step": 2350
    },
    {
      "epoch": 5.98984771573604,
      "grad_norm": 0.22685207426548004,
      "learning_rate": 2.0050761421319797e-05,
      "loss": 0.2823,
      "step": 2360
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.9028552770614624,
      "eval_runtime": 8.7725,
      "eval_samples_per_second": 89.826,
      "eval_steps_per_second": 11.285,
      "step": 2364
    },
    {
      "epoch": 6.0152284263959395,
      "grad_norm": 0.280078649520874,
      "learning_rate": 1.9923857868020303e-05,
      "loss": 0.1616,
      "step": 2370
    },
    {
      "epoch": 6.040609137055838,
      "grad_norm": 0.17733518779277802,
      "learning_rate": 1.9796954314720815e-05,
      "loss": 0.0129,
      "step": 2380
    },
    {
      "epoch": 6.065989847715736,
      "grad_norm": 4.652949333190918,
      "learning_rate": 1.967005076142132e-05,
      "loss": 0.2607,
      "step": 2390
    },
    {
      "epoch": 6.091370558375634,
      "grad_norm": 17.740243911743164,
      "learning_rate": 1.9543147208121827e-05,
      "loss": 0.2297,
      "step": 2400
    },
    {
      "epoch": 6.116751269035533,
      "grad_norm": 59.82321548461914,
      "learning_rate": 1.9416243654822337e-05,
      "loss": 0.2063,
      "step": 2410
    },
    {
      "epoch": 6.1421319796954315,
      "grad_norm": 0.1367008090019226,
      "learning_rate": 1.9289340101522843e-05,
      "loss": 0.1211,
      "step": 2420
    },
    {
      "epoch": 6.16751269035533,
      "grad_norm": 0.14475467801094055,
      "learning_rate": 1.916243654822335e-05,
      "loss": 0.0078,
      "step": 2430
    },
    {
      "epoch": 6.192893401015229,
      "grad_norm": 14.927183151245117,
      "learning_rate": 1.9035532994923858e-05,
      "loss": 0.1697,
      "step": 2440
    },
    {
      "epoch": 6.218274111675127,
      "grad_norm": 0.11621695011854172,
      "learning_rate": 1.8908629441624368e-05,
      "loss": 0.0255,
      "step": 2450
    },
    {
      "epoch": 6.243654822335025,
      "grad_norm": 0.12786021828651428,
      "learning_rate": 1.8781725888324874e-05,
      "loss": 0.008,
      "step": 2460
    },
    {
      "epoch": 6.269035532994923,
      "grad_norm": 0.07750383019447327,
      "learning_rate": 1.8654822335025383e-05,
      "loss": 0.21,
      "step": 2470
    },
    {
      "epoch": 6.2944162436548226,
      "grad_norm": 0.11915043741464615,
      "learning_rate": 1.852791878172589e-05,
      "loss": 0.0661,
      "step": 2480
    },
    {
      "epoch": 6.319796954314721,
      "grad_norm": 0.09122921526432037,
      "learning_rate": 1.8401015228426395e-05,
      "loss": 0.0079,
      "step": 2490
    },
    {
      "epoch": 6.345177664974619,
      "grad_norm": 0.08648554980754852,
      "learning_rate": 1.8274111675126904e-05,
      "loss": 0.121,
      "step": 2500
    },
    {
      "epoch": 6.370558375634518,
      "grad_norm": 0.13753150403499603,
      "learning_rate": 1.814720812182741e-05,
      "loss": 0.0717,
      "step": 2510
    },
    {
      "epoch": 6.395939086294416,
      "grad_norm": 20.555654525756836,
      "learning_rate": 1.802030456852792e-05,
      "loss": 0.0759,
      "step": 2520
    },
    {
      "epoch": 6.4213197969543145,
      "grad_norm": 0.07365628331899643,
      "learning_rate": 1.789340101522843e-05,
      "loss": 0.0801,
      "step": 2530
    },
    {
      "epoch": 6.446700507614214,
      "grad_norm": 0.05513671413064003,
      "learning_rate": 1.7766497461928935e-05,
      "loss": 0.0839,
      "step": 2540
    },
    {
      "epoch": 6.472081218274112,
      "grad_norm": 1.4962226152420044,
      "learning_rate": 1.763959390862944e-05,
      "loss": 0.1288,
      "step": 2550
    },
    {
      "epoch": 6.49746192893401,
      "grad_norm": 10.59736442565918,
      "learning_rate": 1.751269035532995e-05,
      "loss": 0.1318,
      "step": 2560
    },
    {
      "epoch": 6.522842639593908,
      "grad_norm": 0.06739044189453125,
      "learning_rate": 1.7385786802030457e-05,
      "loss": 0.1288,
      "step": 2570
    },
    {
      "epoch": 6.548223350253807,
      "grad_norm": 3.2351067066192627,
      "learning_rate": 1.7258883248730966e-05,
      "loss": 0.0801,
      "step": 2580
    },
    {
      "epoch": 6.573604060913706,
      "grad_norm": 0.10322804003953934,
      "learning_rate": 1.7131979695431472e-05,
      "loss": 0.0687,
      "step": 2590
    },
    {
      "epoch": 6.598984771573604,
      "grad_norm": 1.834286093711853,
      "learning_rate": 1.700507614213198e-05,
      "loss": 0.1842,
      "step": 2600
    },
    {
      "epoch": 6.624365482233502,
      "grad_norm": 0.10404256731271744,
      "learning_rate": 1.6878172588832487e-05,
      "loss": 0.2148,
      "step": 2610
    },
    {
      "epoch": 6.649746192893401,
      "grad_norm": 0.11388051509857178,
      "learning_rate": 1.6751269035532997e-05,
      "loss": 0.2577,
      "step": 2620
    },
    {
      "epoch": 6.675126903553299,
      "grad_norm": 0.1220388114452362,
      "learning_rate": 1.6624365482233503e-05,
      "loss": 0.0425,
      "step": 2630
    },
    {
      "epoch": 6.700507614213198,
      "grad_norm": 18.549720764160156,
      "learning_rate": 1.649746192893401e-05,
      "loss": 0.2702,
      "step": 2640
    },
    {
      "epoch": 6.725888324873097,
      "grad_norm": 0.12684045732021332,
      "learning_rate": 1.6370558375634518e-05,
      "loss": 0.334,
      "step": 2650
    },
    {
      "epoch": 6.751269035532995,
      "grad_norm": 0.27108675241470337,
      "learning_rate": 1.6243654822335024e-05,
      "loss": 0.1592,
      "step": 2660
    },
    {
      "epoch": 6.776649746192893,
      "grad_norm": 0.17042884230613708,
      "learning_rate": 1.6116751269035534e-05,
      "loss": 0.0726,
      "step": 2670
    },
    {
      "epoch": 6.802030456852792,
      "grad_norm": 0.11051168292760849,
      "learning_rate": 1.5989847715736043e-05,
      "loss": 0.0702,
      "step": 2680
    },
    {
      "epoch": 6.8274111675126905,
      "grad_norm": 0.11623559892177582,
      "learning_rate": 1.586294416243655e-05,
      "loss": 0.1979,
      "step": 2690
    },
    {
      "epoch": 6.852791878172589,
      "grad_norm": 0.16744251549243927,
      "learning_rate": 1.5736040609137055e-05,
      "loss": 0.1624,
      "step": 2700
    },
    {
      "epoch": 6.878172588832487,
      "grad_norm": 62.56240463256836,
      "learning_rate": 1.5609137055837564e-05,
      "loss": 0.1029,
      "step": 2710
    },
    {
      "epoch": 6.903553299492386,
      "grad_norm": 0.11622487008571625,
      "learning_rate": 1.548223350253807e-05,
      "loss": 0.0093,
      "step": 2720
    },
    {
      "epoch": 6.928934010152284,
      "grad_norm": 0.09292230755090714,
      "learning_rate": 1.535532994923858e-05,
      "loss": 0.1308,
      "step": 2730
    },
    {
      "epoch": 6.9543147208121825,
      "grad_norm": 0.20956498384475708,
      "learning_rate": 1.5228426395939088e-05,
      "loss": 0.0052,
      "step": 2740
    },
    {
      "epoch": 6.979695431472082,
      "grad_norm": 44.860416412353516,
      "learning_rate": 1.5101522842639595e-05,
      "loss": 0.2451,
      "step": 2750
    },
    {
      "epoch": 7.0,
      "eval_loss": 1.0761854648590088,
      "eval_runtime": 8.8065,
      "eval_samples_per_second": 89.479,
      "eval_steps_per_second": 11.242,
      "step": 2758
    },
    {
      "epoch": 7.00507614213198,
      "grad_norm": 0.09350736439228058,
      "learning_rate": 1.4974619289340103e-05,
      "loss": 0.0638,
      "step": 2760
    },
    {
      "epoch": 7.030456852791878,
      "grad_norm": 3.844224214553833,
      "learning_rate": 1.484771573604061e-05,
      "loss": 0.0735,
      "step": 2770
    },
    {
      "epoch": 7.055837563451776,
      "grad_norm": 0.09096001088619232,
      "learning_rate": 1.4720812182741117e-05,
      "loss": 0.0736,
      "step": 2780
    },
    {
      "epoch": 7.081218274111675,
      "grad_norm": 0.06552541255950928,
      "learning_rate": 1.4593908629441624e-05,
      "loss": 0.0525,
      "step": 2790
    },
    {
      "epoch": 7.106598984771574,
      "grad_norm": 0.10838861763477325,
      "learning_rate": 1.4467005076142132e-05,
      "loss": 0.4002,
      "step": 2800
    },
    {
      "epoch": 7.131979695431472,
      "grad_norm": 0.12170158326625824,
      "learning_rate": 1.4340101522842641e-05,
      "loss": 0.1041,
      "step": 2810
    },
    {
      "epoch": 7.157360406091371,
      "grad_norm": 0.15080027282238007,
      "learning_rate": 1.421319796954315e-05,
      "loss": 0.085,
      "step": 2820
    },
    {
      "epoch": 7.182741116751269,
      "grad_norm": 0.13942450284957886,
      "learning_rate": 1.4086294416243657e-05,
      "loss": 0.007,
      "step": 2830
    },
    {
      "epoch": 7.208121827411167,
      "grad_norm": 0.09515988826751709,
      "learning_rate": 1.3959390862944163e-05,
      "loss": 0.1357,
      "step": 2840
    },
    {
      "epoch": 7.233502538071066,
      "grad_norm": 0.09349621832370758,
      "learning_rate": 1.383248730964467e-05,
      "loss": 0.0644,
      "step": 2850
    },
    {
      "epoch": 7.258883248730965,
      "grad_norm": 0.25502246618270874,
      "learning_rate": 1.3705583756345178e-05,
      "loss": 0.0059,
      "step": 2860
    },
    {
      "epoch": 7.284263959390863,
      "grad_norm": 5.170173168182373,
      "learning_rate": 1.3578680203045684e-05,
      "loss": 0.072,
      "step": 2870
    },
    {
      "epoch": 7.309644670050761,
      "grad_norm": 0.09840302914381027,
      "learning_rate": 1.3451776649746192e-05,
      "loss": 0.1308,
      "step": 2880
    },
    {
      "epoch": 7.33502538071066,
      "grad_norm": 0.07067205011844635,
      "learning_rate": 1.3324873096446703e-05,
      "loss": 0.1064,
      "step": 2890
    },
    {
      "epoch": 7.3604060913705585,
      "grad_norm": 2.9675955772399902,
      "learning_rate": 1.3197969543147209e-05,
      "loss": 0.1342,
      "step": 2900
    },
    {
      "epoch": 7.385786802030457,
      "grad_norm": 0.07539036124944687,
      "learning_rate": 1.3071065989847717e-05,
      "loss": 0.0642,
      "step": 2910
    },
    {
      "epoch": 7.411167512690355,
      "grad_norm": 0.19838696718215942,
      "learning_rate": 1.2944162436548224e-05,
      "loss": 0.2019,
      "step": 2920
    },
    {
      "epoch": 7.436548223350254,
      "grad_norm": 0.08878432959318161,
      "learning_rate": 1.281725888324873e-05,
      "loss": 0.104,
      "step": 2930
    },
    {
      "epoch": 7.461928934010152,
      "grad_norm": 0.09144201874732971,
      "learning_rate": 1.2690355329949238e-05,
      "loss": 0.0046,
      "step": 2940
    },
    {
      "epoch": 7.4873096446700504,
      "grad_norm": 0.0494062639772892,
      "learning_rate": 1.2563451776649746e-05,
      "loss": 0.0705,
      "step": 2950
    },
    {
      "epoch": 7.5126903553299496,
      "grad_norm": 0.1255367249250412,
      "learning_rate": 1.2436548223350254e-05,
      "loss": 0.0382,
      "step": 2960
    },
    {
      "epoch": 7.538071065989848,
      "grad_norm": 0.06985324621200562,
      "learning_rate": 1.2309644670050761e-05,
      "loss": 0.0863,
      "step": 2970
    },
    {
      "epoch": 7.563451776649746,
      "grad_norm": 0.08622120320796967,
      "learning_rate": 1.218274111675127e-05,
      "loss": 0.0729,
      "step": 2980
    },
    {
      "epoch": 7.588832487309645,
      "grad_norm": 0.07268553227186203,
      "learning_rate": 1.2055837563451777e-05,
      "loss": 0.1981,
      "step": 2990
    },
    {
      "epoch": 7.614213197969543,
      "grad_norm": 5.232954025268555,
      "learning_rate": 1.1928934010152284e-05,
      "loss": 0.0885,
      "step": 3000
    },
    {
      "epoch": 7.6395939086294415,
      "grad_norm": 0.07025603950023651,
      "learning_rate": 1.1802030456852794e-05,
      "loss": 0.1437,
      "step": 3010
    },
    {
      "epoch": 7.66497461928934,
      "grad_norm": 3.0778110027313232,
      "learning_rate": 1.16751269035533e-05,
      "loss": 0.2441,
      "step": 3020
    },
    {
      "epoch": 7.690355329949239,
      "grad_norm": 0.1191609650850296,
      "learning_rate": 1.1548223350253808e-05,
      "loss": 0.0717,
      "step": 3030
    },
    {
      "epoch": 7.715736040609137,
      "grad_norm": 0.07436536252498627,
      "learning_rate": 1.1421319796954315e-05,
      "loss": 0.0636,
      "step": 3040
    },
    {
      "epoch": 7.741116751269035,
      "grad_norm": 0.08062940835952759,
      "learning_rate": 1.1294416243654823e-05,
      "loss": 0.0918,
      "step": 3050
    },
    {
      "epoch": 7.7664974619289335,
      "grad_norm": 0.06816007196903229,
      "learning_rate": 1.116751269035533e-05,
      "loss": 0.1376,
      "step": 3060
    },
    {
      "epoch": 7.791878172588833,
      "grad_norm": 2.068460464477539,
      "learning_rate": 1.1040609137055838e-05,
      "loss": 0.0642,
      "step": 3070
    },
    {
      "epoch": 7.817258883248731,
      "grad_norm": 0.1358051747083664,
      "learning_rate": 1.0913705583756344e-05,
      "loss": 0.1455,
      "step": 3080
    },
    {
      "epoch": 7.842639593908629,
      "grad_norm": 0.07467884570360184,
      "learning_rate": 1.0786802030456854e-05,
      "loss": 0.0041,
      "step": 3090
    },
    {
      "epoch": 7.868020304568528,
      "grad_norm": 0.08706346899271011,
      "learning_rate": 1.0659898477157361e-05,
      "loss": 0.0838,
      "step": 3100
    },
    {
      "epoch": 7.893401015228426,
      "grad_norm": 0.07726430892944336,
      "learning_rate": 1.0532994923857867e-05,
      "loss": 0.0689,
      "step": 3110
    },
    {
      "epoch": 7.918781725888325,
      "grad_norm": 0.09187603741884232,
      "learning_rate": 1.0406091370558377e-05,
      "loss": 0.0785,
      "step": 3120
    },
    {
      "epoch": 7.944162436548224,
      "grad_norm": 0.1465052366256714,
      "learning_rate": 1.0279187817258885e-05,
      "loss": 0.0778,
      "step": 3130
    },
    {
      "epoch": 7.969543147208122,
      "grad_norm": 0.06538953632116318,
      "learning_rate": 1.015228426395939e-05,
      "loss": 0.0577,
      "step": 3140
    },
    {
      "epoch": 7.99492385786802,
      "grad_norm": 0.17358700931072235,
      "learning_rate": 1.0025380710659898e-05,
      "loss": 0.0041,
      "step": 3150
    },
    {
      "epoch": 8.0,
      "eval_loss": 1.1998591423034668,
      "eval_runtime": 8.6883,
      "eval_samples_per_second": 90.697,
      "eval_steps_per_second": 11.395,
      "step": 3152
    },
    {
      "epoch": 8.02030456852792,
      "grad_norm": 0.05773229897022247,
      "learning_rate": 9.898477157360408e-06,
      "loss": 0.0856,
      "step": 3160
    },
    {
      "epoch": 8.045685279187817,
      "grad_norm": 0.07206518203020096,
      "learning_rate": 9.771573604060914e-06,
      "loss": 0.1201,
      "step": 3170
    },
    {
      "epoch": 8.071065989847716,
      "grad_norm": 0.07798884809017181,
      "learning_rate": 9.644670050761421e-06,
      "loss": 0.069,
      "step": 3180
    },
    {
      "epoch": 8.096446700507615,
      "grad_norm": 0.07664274424314499,
      "learning_rate": 9.517766497461929e-06,
      "loss": 0.0036,
      "step": 3190
    },
    {
      "epoch": 8.121827411167512,
      "grad_norm": 0.050856612622737885,
      "learning_rate": 9.390862944162437e-06,
      "loss": 0.003,
      "step": 3200
    },
    {
      "epoch": 8.147208121827411,
      "grad_norm": 0.0529194138944149,
      "learning_rate": 9.263959390862944e-06,
      "loss": 0.1412,
      "step": 3210
    },
    {
      "epoch": 8.17258883248731,
      "grad_norm": 13.00708293914795,
      "learning_rate": 9.137055837563452e-06,
      "loss": 0.0513,
      "step": 3220
    },
    {
      "epoch": 8.197969543147208,
      "grad_norm": 0.06696105003356934,
      "learning_rate": 9.01015228426396e-06,
      "loss": 0.0704,
      "step": 3230
    },
    {
      "epoch": 8.223350253807107,
      "grad_norm": 0.06790643185377121,
      "learning_rate": 8.883248730964468e-06,
      "loss": 0.1119,
      "step": 3240
    },
    {
      "epoch": 8.248730964467006,
      "grad_norm": 0.06100468337535858,
      "learning_rate": 8.756345177664975e-06,
      "loss": 0.0032,
      "step": 3250
    },
    {
      "epoch": 8.274111675126903,
      "grad_norm": 8.70279312133789,
      "learning_rate": 8.629441624365483e-06,
      "loss": 0.0811,
      "step": 3260
    },
    {
      "epoch": 8.299492385786802,
      "grad_norm": 0.09535681456327438,
      "learning_rate": 8.50253807106599e-06,
      "loss": 0.0795,
      "step": 3270
    },
    {
      "epoch": 8.3248730964467,
      "grad_norm": 0.07176022976636887,
      "learning_rate": 8.375634517766498e-06,
      "loss": 0.0893,
      "step": 3280
    },
    {
      "epoch": 8.350253807106599,
      "grad_norm": 0.051213160157203674,
      "learning_rate": 8.248730964467004e-06,
      "loss": 0.0701,
      "step": 3290
    },
    {
      "epoch": 8.375634517766498,
      "grad_norm": 0.09963097423315048,
      "learning_rate": 8.121827411167512e-06,
      "loss": 0.0745,
      "step": 3300
    },
    {
      "epoch": 8.401015228426395,
      "grad_norm": 0.0691935271024704,
      "learning_rate": 7.994923857868022e-06,
      "loss": 0.0884,
      "step": 3310
    },
    {
      "epoch": 8.426395939086294,
      "grad_norm": 0.16761644184589386,
      "learning_rate": 7.868020304568528e-06,
      "loss": 0.0043,
      "step": 3320
    },
    {
      "epoch": 8.451776649746193,
      "grad_norm": 0.04886673018336296,
      "learning_rate": 7.741116751269035e-06,
      "loss": 0.0037,
      "step": 3330
    },
    {
      "epoch": 8.47715736040609,
      "grad_norm": 0.06593126058578491,
      "learning_rate": 7.614213197969544e-06,
      "loss": 0.0029,
      "step": 3340
    },
    {
      "epoch": 8.50253807106599,
      "grad_norm": 8.315119743347168,
      "learning_rate": 7.4873096446700515e-06,
      "loss": 0.1359,
      "step": 3350
    },
    {
      "epoch": 8.527918781725889,
      "grad_norm": 0.19459432363510132,
      "learning_rate": 7.360406091370558e-06,
      "loss": 0.0771,
      "step": 3360
    },
    {
      "epoch": 8.553299492385786,
      "grad_norm": 0.09421133995056152,
      "learning_rate": 7.233502538071066e-06,
      "loss": 0.003,
      "step": 3370
    },
    {
      "epoch": 8.578680203045685,
      "grad_norm": 0.04559624567627907,
      "learning_rate": 7.106598984771575e-06,
      "loss": 0.0647,
      "step": 3380
    },
    {
      "epoch": 8.604060913705585,
      "grad_norm": 3.285724639892578,
      "learning_rate": 6.9796954314720814e-06,
      "loss": 0.0744,
      "step": 3390
    },
    {
      "epoch": 8.629441624365482,
      "grad_norm": 0.0715055987238884,
      "learning_rate": 6.852791878172589e-06,
      "loss": 0.0696,
      "step": 3400
    },
    {
      "epoch": 8.654822335025381,
      "grad_norm": 0.05040270462632179,
      "learning_rate": 6.725888324873096e-06,
      "loss": 0.0028,
      "step": 3410
    },
    {
      "epoch": 8.680203045685278,
      "grad_norm": 0.05061913654208183,
      "learning_rate": 6.5989847715736045e-06,
      "loss": 0.1397,
      "step": 3420
    },
    {
      "epoch": 8.705583756345177,
      "grad_norm": 0.06146248057484627,
      "learning_rate": 6.472081218274112e-06,
      "loss": 0.0757,
      "step": 3430
    },
    {
      "epoch": 8.730964467005077,
      "grad_norm": 0.0398724228143692,
      "learning_rate": 6.345177664974619e-06,
      "loss": 0.0888,
      "step": 3440
    },
    {
      "epoch": 8.756345177664974,
      "grad_norm": 0.058130159974098206,
      "learning_rate": 6.218274111675127e-06,
      "loss": 0.0714,
      "step": 3450
    },
    {
      "epoch": 8.781725888324873,
      "grad_norm": 0.06121835485100746,
      "learning_rate": 6.091370558375635e-06,
      "loss": 0.0329,
      "step": 3460
    },
    {
      "epoch": 8.807106598984772,
      "grad_norm": 0.048920340836048126,
      "learning_rate": 5.964467005076142e-06,
      "loss": 0.0032,
      "step": 3470
    },
    {
      "epoch": 8.83248730964467,
      "grad_norm": 0.05234299972653389,
      "learning_rate": 5.83756345177665e-06,
      "loss": 0.07,
      "step": 3480
    },
    {
      "epoch": 8.857868020304569,
      "grad_norm": 0.04120004549622536,
      "learning_rate": 5.710659898477158e-06,
      "loss": 0.152,
      "step": 3490
    },
    {
      "epoch": 8.883248730964468,
      "grad_norm": 0.0480622835457325,
      "learning_rate": 5.583756345177665e-06,
      "loss": 0.162,
      "step": 3500
    }
  ],
  "logging_steps": 10,
  "max_steps": 3940,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3709087162368000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}