{
  "best_metric": 0.4315283000469208,
  "best_model_checkpoint": "xblock-large-patch2-224/checkpoint-5181",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 5181,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 19.092004776000977,
      "learning_rate": 2.2157996146435453e-06,
      "loss": 2.5421,
      "step": 25
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.274575233459473,
      "learning_rate": 4.624277456647399e-06,
      "loss": 1.9399,
      "step": 50
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.260701179504395,
      "learning_rate": 7.032755298651253e-06,
      "loss": 1.5011,
      "step": 75
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.719381332397461,
      "learning_rate": 9.441233140655107e-06,
      "loss": 1.3912,
      "step": 100
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.234914779663086,
      "learning_rate": 1.184971098265896e-05,
      "loss": 1.2166,
      "step": 125
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.722230911254883,
      "learning_rate": 1.4258188824662813e-05,
      "loss": 1.1751,
      "step": 150
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.890995025634766,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.1862,
      "step": 175
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.72986125946045,
      "learning_rate": 1.907514450867052e-05,
      "loss": 0.8828,
      "step": 200
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.903642654418945,
      "learning_rate": 2.1483622350674377e-05,
      "loss": 0.9541,
      "step": 225
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.937955856323242,
      "learning_rate": 2.3892100192678228e-05,
      "loss": 0.9032,
      "step": 250
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.743949890136719,
      "learning_rate": 2.6300578034682083e-05,
      "loss": 1.0891,
      "step": 275
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.766569137573242,
      "learning_rate": 2.8709055876685937e-05,
      "loss": 0.8951,
      "step": 300
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.973127365112305,
      "learning_rate": 3.111753371868979e-05,
      "loss": 1.0709,
      "step": 325
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.805391311645508,
      "learning_rate": 3.352601156069364e-05,
      "loss": 1.1141,
      "step": 350
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.355842113494873,
      "learning_rate": 3.59344894026975e-05,
      "loss": 1.0031,
      "step": 375
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.688377380371094,
      "learning_rate": 3.834296724470135e-05,
      "loss": 1.0344,
      "step": 400
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.896137237548828,
      "learning_rate": 4.07514450867052e-05,
      "loss": 1.0991,
      "step": 425
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.027874946594238,
      "learning_rate": 4.3159922928709055e-05,
      "loss": 1.0221,
      "step": 450
    },
    {
      "epoch": 0.28,
      "grad_norm": 27.792613983154297,
      "learning_rate": 4.556840077071291e-05,
      "loss": 0.8256,
      "step": 475
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.005478858947754,
      "learning_rate": 4.7976878612716764e-05,
      "loss": 0.9862,
      "step": 500
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.962843894958496,
      "learning_rate": 4.995709995709996e-05,
      "loss": 1.0462,
      "step": 525
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.288177013397217,
      "learning_rate": 4.9688974688974696e-05,
      "loss": 0.9999,
      "step": 550
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.769800901412964,
      "learning_rate": 4.9420849420849425e-05,
      "loss": 1.1262,
      "step": 575
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.510008811950684,
      "learning_rate": 4.9152724152724154e-05,
      "loss": 0.944,
      "step": 600
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.264263153076172,
      "learning_rate": 4.888459888459889e-05,
      "loss": 0.9652,
      "step": 625
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.604584693908691,
      "learning_rate": 4.861647361647362e-05,
      "loss": 0.9246,
      "step": 650
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.483927249908447,
      "learning_rate": 4.834834834834835e-05,
      "loss": 1.0557,
      "step": 675
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.221104621887207,
      "learning_rate": 4.808022308022308e-05,
      "loss": 0.7906,
      "step": 700
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.725340843200684,
      "learning_rate": 4.781209781209782e-05,
      "loss": 0.9,
      "step": 725
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.057939052581787,
      "learning_rate": 4.754397254397255e-05,
      "loss": 0.8744,
      "step": 750
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.354517936706543,
      "learning_rate": 4.727584727584728e-05,
      "loss": 1.0611,
      "step": 775
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.2830491065979,
      "learning_rate": 4.700772200772201e-05,
      "loss": 0.8894,
      "step": 800
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.702705383300781,
      "learning_rate": 4.673959673959674e-05,
      "loss": 1.0246,
      "step": 825
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.7863359451293945,
      "learning_rate": 4.647147147147147e-05,
      "loss": 0.7967,
      "step": 850
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.841278553009033,
      "learning_rate": 4.62033462033462e-05,
      "loss": 0.8616,
      "step": 875
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.256266117095947,
      "learning_rate": 4.593522093522094e-05,
      "loss": 0.9735,
      "step": 900
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.177362442016602,
      "learning_rate": 4.566709566709567e-05,
      "loss": 0.8287,
      "step": 925
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.448288917541504,
      "learning_rate": 4.53989703989704e-05,
      "loss": 1.0062,
      "step": 950
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.912018299102783,
      "learning_rate": 4.513084513084513e-05,
      "loss": 1.0174,
      "step": 975
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.055561065673828,
      "learning_rate": 4.486271986271987e-05,
      "loss": 0.9616,
      "step": 1000
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.433628559112549,
      "learning_rate": 4.4594594594594596e-05,
      "loss": 0.9309,
      "step": 1025
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.5334954261779785,
      "learning_rate": 4.4326469326469325e-05,
      "loss": 0.807,
      "step": 1050
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.620259761810303,
      "learning_rate": 4.405834405834406e-05,
      "loss": 0.8042,
      "step": 1075
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.67726469039917,
      "learning_rate": 4.379021879021879e-05,
      "loss": 0.7394,
      "step": 1100
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.430630683898926,
      "learning_rate": 4.3522093522093526e-05,
      "loss": 0.7895,
      "step": 1125
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.241034507751465,
      "learning_rate": 4.3253968253968256e-05,
      "loss": 0.8032,
      "step": 1150
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.471988201141357,
      "learning_rate": 4.298584298584299e-05,
      "loss": 0.7669,
      "step": 1175
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.161353588104248,
      "learning_rate": 4.271771771771772e-05,
      "loss": 0.8795,
      "step": 1200
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.6813201904296875,
      "learning_rate": 4.244959244959245e-05,
      "loss": 0.819,
      "step": 1225
    },
    {
      "epoch": 0.72,
      "grad_norm": 16.08786392211914,
      "learning_rate": 4.2181467181467186e-05,
      "loss": 0.8779,
      "step": 1250
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.753849506378174,
      "learning_rate": 4.1913341913341915e-05,
      "loss": 0.9255,
      "step": 1275
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.4661431312561035,
      "learning_rate": 4.1645216645216644e-05,
      "loss": 0.8028,
      "step": 1300
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.559650897979736,
      "learning_rate": 4.137709137709138e-05,
      "loss": 0.7556,
      "step": 1325
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.7179341316223145,
      "learning_rate": 4.1108966108966116e-05,
      "loss": 0.7076,
      "step": 1350
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.687734603881836,
      "learning_rate": 4.0840840840840845e-05,
      "loss": 0.8583,
      "step": 1375
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.1677398681640625,
      "learning_rate": 4.0572715572715574e-05,
      "loss": 0.8944,
      "step": 1400
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.766656875610352,
      "learning_rate": 4.03045903045903e-05,
      "loss": 0.5638,
      "step": 1425
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.12522554397583,
      "learning_rate": 4.003646503646504e-05,
      "loss": 0.6883,
      "step": 1450
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.586186408996582,
      "learning_rate": 3.976833976833977e-05,
      "loss": 0.8784,
      "step": 1475
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.075995445251465,
      "learning_rate": 3.95002145002145e-05,
      "loss": 0.7596,
      "step": 1500
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.722098350524902,
      "learning_rate": 3.923208923208923e-05,
      "loss": 0.6316,
      "step": 1525
    },
    {
      "epoch": 0.9,
      "grad_norm": 12.86962890625,
      "learning_rate": 3.896396396396397e-05,
      "loss": 0.7721,
      "step": 1550
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.640520095825195,
      "learning_rate": 3.86958386958387e-05,
      "loss": 0.711,
      "step": 1575
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.080173015594482,
      "learning_rate": 3.842771342771343e-05,
      "loss": 0.8145,
      "step": 1600
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.814232587814331,
      "learning_rate": 3.815958815958816e-05,
      "loss": 0.8509,
      "step": 1625
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.209416389465332,
      "learning_rate": 3.789146289146289e-05,
      "loss": 0.8037,
      "step": 1650
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.5646421909332275,
      "learning_rate": 3.762333762333762e-05,
      "loss": 0.8347,
      "step": 1675
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.00243854522705,
      "learning_rate": 3.735521235521236e-05,
      "loss": 0.7534,
      "step": 1700
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.955112934112549,
      "learning_rate": 3.708708708708709e-05,
      "loss": 0.9062,
      "step": 1725
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7815393518518519,
      "eval_f1_macro": 0.36763187534206965,
      "eval_f1_micro": 0.7815393518518519,
      "eval_f1_weighted": 0.7649698649865229,
      "eval_loss": 0.6848556399345398,
      "eval_precision_macro": 0.4527657259795588,
      "eval_precision_micro": 0.7815393518518519,
      "eval_precision_weighted": 0.7691072277887989,
      "eval_recall_macro": 0.3815255183458625,
      "eval_recall_micro": 0.7815393518518519,
      "eval_recall_weighted": 0.7815393518518519,
      "eval_runtime": 3330.8434,
      "eval_samples_per_second": 1.038,
      "eval_steps_per_second": 0.065,
      "step": 1727
    },
    {
      "epoch": 1.01,
      "grad_norm": 3.4409847259521484,
      "learning_rate": 3.681896181896182e-05,
      "loss": 0.8201,
      "step": 1750
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.564268589019775,
      "learning_rate": 3.655083655083655e-05,
      "loss": 0.548,
      "step": 1775
    },
    {
      "epoch": 1.04,
      "grad_norm": 10.243828773498535,
      "learning_rate": 3.628271128271129e-05,
      "loss": 0.6348,
      "step": 1800
    },
    {
      "epoch": 1.06,
      "grad_norm": 14.57152271270752,
      "learning_rate": 3.6014586014586017e-05,
      "loss": 0.66,
      "step": 1825
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.4091973304748535,
      "learning_rate": 3.5746460746460746e-05,
      "loss": 0.8169,
      "step": 1850
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.423861503601074,
      "learning_rate": 3.547833547833548e-05,
      "loss": 0.7353,
      "step": 1875
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.510718822479248,
      "learning_rate": 3.521021021021021e-05,
      "loss": 0.7008,
      "step": 1900
    },
    {
      "epoch": 1.11,
      "grad_norm": 2.5035436153411865,
      "learning_rate": 3.4942084942084947e-05,
      "loss": 0.5975,
      "step": 1925
    },
    {
      "epoch": 1.13,
      "grad_norm": 15.364286422729492,
      "learning_rate": 3.4673959673959676e-05,
      "loss": 0.8198,
      "step": 1950
    },
    {
      "epoch": 1.14,
      "grad_norm": 16.03240966796875,
      "learning_rate": 3.440583440583441e-05,
      "loss": 0.7089,
      "step": 1975
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.521039962768555,
      "learning_rate": 3.413770913770914e-05,
      "loss": 0.6514,
      "step": 2000
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.206024169921875,
      "learning_rate": 3.386958386958387e-05,
      "loss": 0.7545,
      "step": 2025
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.756472110748291,
      "learning_rate": 3.36014586014586e-05,
      "loss": 0.8055,
      "step": 2050
    },
    {
      "epoch": 1.2,
      "grad_norm": 17.274944305419922,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.7212,
      "step": 2075
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.303420543670654,
      "learning_rate": 3.3065208065208064e-05,
      "loss": 0.7299,
      "step": 2100
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.07370138168335,
      "learning_rate": 3.27970827970828e-05,
      "loss": 0.5858,
      "step": 2125
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.2755937576293945,
      "learning_rate": 3.252895752895753e-05,
      "loss": 0.7687,
      "step": 2150
    },
    {
      "epoch": 1.26,
      "grad_norm": 12.160276412963867,
      "learning_rate": 3.227155727155727e-05,
      "loss": 0.727,
      "step": 2175
    },
    {
      "epoch": 1.27,
      "grad_norm": 3.9984447956085205,
      "learning_rate": 3.2003432003432e-05,
      "loss": 0.6697,
      "step": 2200
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.756568908691406,
      "learning_rate": 3.173530673530674e-05,
      "loss": 0.7912,
      "step": 2225
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.783411026000977,
      "learning_rate": 3.1467181467181466e-05,
      "loss": 0.7035,
      "step": 2250
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.451704502105713,
      "learning_rate": 3.1199056199056196e-05,
      "loss": 0.7144,
      "step": 2275
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.039503574371338,
      "learning_rate": 3.093093093093093e-05,
      "loss": 0.7893,
      "step": 2300
    },
    {
      "epoch": 1.35,
      "grad_norm": 12.280179023742676,
      "learning_rate": 3.066280566280567e-05,
      "loss": 0.5903,
      "step": 2325
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.999240398406982,
      "learning_rate": 3.0394680394680397e-05,
      "loss": 0.651,
      "step": 2350
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.355953216552734,
      "learning_rate": 3.012655512655513e-05,
      "loss": 0.878,
      "step": 2375
    },
    {
      "epoch": 1.39,
      "grad_norm": 12.737029075622559,
      "learning_rate": 2.9858429858429858e-05,
      "loss": 0.731,
      "step": 2400
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.184784889221191,
      "learning_rate": 2.959030459030459e-05,
      "loss": 0.8775,
      "step": 2425
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.188583374023438,
      "learning_rate": 2.9322179322179323e-05,
      "loss": 0.7259,
      "step": 2450
    },
    {
      "epoch": 1.43,
      "grad_norm": 12.98018741607666,
      "learning_rate": 2.906477906477907e-05,
      "loss": 0.6219,
      "step": 2475
    },
    {
      "epoch": 1.45,
      "grad_norm": 12.084989547729492,
      "learning_rate": 2.87966537966538e-05,
      "loss": 0.5074,
      "step": 2500
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.283312797546387,
      "learning_rate": 2.852852852852853e-05,
      "loss": 0.6606,
      "step": 2525
    },
    {
      "epoch": 1.48,
      "grad_norm": 19.32860565185547,
      "learning_rate": 2.826040326040326e-05,
      "loss": 0.7651,
      "step": 2550
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.0794901847839355,
      "learning_rate": 2.7992277992277993e-05,
      "loss": 0.6737,
      "step": 2575
    },
    {
      "epoch": 1.51,
      "grad_norm": 33.1405029296875,
      "learning_rate": 2.7724152724152726e-05,
      "loss": 0.654,
      "step": 2600
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.921344757080078,
      "learning_rate": 2.7456027456027455e-05,
      "loss": 0.7194,
      "step": 2625
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.453707695007324,
      "learning_rate": 2.7187902187902187e-05,
      "loss": 0.6851,
      "step": 2650
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.766169548034668,
      "learning_rate": 2.6919776919776923e-05,
      "loss": 0.8228,
      "step": 2675
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.957389831542969,
      "learning_rate": 2.6651651651651656e-05,
      "loss": 0.6645,
      "step": 2700
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.715158939361572,
      "learning_rate": 2.6383526383526385e-05,
      "loss": 0.6905,
      "step": 2725
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.306962490081787,
      "learning_rate": 2.6115401115401117e-05,
      "loss": 0.6924,
      "step": 2750
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.548517227172852,
      "learning_rate": 2.5847275847275846e-05,
      "loss": 0.8402,
      "step": 2775
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.5719895362854,
      "learning_rate": 2.557915057915058e-05,
      "loss": 0.7758,
      "step": 2800
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.910326957702637,
      "learning_rate": 2.531102531102531e-05,
      "loss": 0.5645,
      "step": 2825
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.304277420043945,
      "learning_rate": 2.504290004290004e-05,
      "loss": 0.7066,
      "step": 2850
    },
    {
      "epoch": 1.66,
      "grad_norm": 13.99254035949707,
      "learning_rate": 2.4774774774774777e-05,
      "loss": 0.7396,
      "step": 2875
    },
    {
      "epoch": 1.68,
      "grad_norm": 3.806931257247925,
      "learning_rate": 2.4506649506649506e-05,
      "loss": 0.6334,
      "step": 2900
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.550988674163818,
      "learning_rate": 2.423852423852424e-05,
      "loss": 0.9251,
      "step": 2925
    },
    {
      "epoch": 1.71,
      "grad_norm": 3.1442198753356934,
      "learning_rate": 2.397039897039897e-05,
      "loss": 0.546,
      "step": 2950
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.249305248260498,
      "learning_rate": 2.3702273702273703e-05,
      "loss": 0.6419,
      "step": 2975
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.300810813903809,
      "learning_rate": 2.3434148434148436e-05,
      "loss": 0.535,
      "step": 3000
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.079426765441895,
      "learning_rate": 2.3166023166023168e-05,
      "loss": 0.8142,
      "step": 3025
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.737719535827637,
      "learning_rate": 2.28978978978979e-05,
      "loss": 0.5974,
      "step": 3050
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.037626266479492,
      "learning_rate": 2.262977262977263e-05,
      "loss": 0.7068,
      "step": 3075
    },
    {
      "epoch": 1.8,
      "grad_norm": 3.4523470401763916,
      "learning_rate": 2.2361647361647362e-05,
      "loss": 0.5756,
      "step": 3100
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.2966675758361816,
      "learning_rate": 2.2093522093522095e-05,
      "loss": 0.4941,
      "step": 3125
    },
    {
      "epoch": 1.82,
      "grad_norm": 11.497820854187012,
      "learning_rate": 2.1825396825396827e-05,
      "loss": 0.8353,
      "step": 3150
    },
    {
      "epoch": 1.84,
      "grad_norm": 11.813599586486816,
      "learning_rate": 2.1557271557271557e-05,
      "loss": 0.8303,
      "step": 3175
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.504293203353882,
      "learning_rate": 2.128914628914629e-05,
      "loss": 0.5574,
      "step": 3200
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.983193397521973,
      "learning_rate": 2.102102102102102e-05,
      "loss": 0.6033,
      "step": 3225
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.629824161529541,
      "learning_rate": 2.0752895752895754e-05,
      "loss": 0.6305,
      "step": 3250
    },
    {
      "epoch": 1.9,
      "grad_norm": 10.86919116973877,
      "learning_rate": 2.0484770484770487e-05,
      "loss": 0.6045,
      "step": 3275
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.0854387283325195,
      "learning_rate": 2.0216645216645216e-05,
      "loss": 0.6208,
      "step": 3300
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.228011131286621,
      "learning_rate": 1.994851994851995e-05,
      "loss": 0.6249,
      "step": 3325
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.076812744140625,
      "learning_rate": 1.968039468039468e-05,
      "loss": 0.6176,
      "step": 3350
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.893978595733643,
      "learning_rate": 1.9412269412269413e-05,
      "loss": 0.7779,
      "step": 3375
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.72683048248291,
      "learning_rate": 1.9144144144144142e-05,
      "loss": 0.669,
      "step": 3400
    },
    {
      "epoch": 1.98,
      "grad_norm": 3.820025682449341,
      "learning_rate": 1.887601887601888e-05,
      "loss": 0.6182,
      "step": 3425
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.642152309417725,
      "learning_rate": 1.8607893607893607e-05,
      "loss": 0.6453,
      "step": 3450
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8454861111111112,
      "eval_f1_macro": 0.5038698715266201,
      "eval_f1_micro": 0.845486111111111,
      "eval_f1_weighted": 0.830657390454042,
      "eval_loss": 0.4813511371612549,
      "eval_precision_macro": 0.7578927995388053,
      "eval_precision_micro": 0.8454861111111112,
      "eval_precision_weighted": 0.8347843050246918,
      "eval_recall_macro": 0.456096431265928,
      "eval_recall_micro": 0.8454861111111112,
      "eval_recall_weighted": 0.8454861111111112,
      "eval_runtime": 3281.605,
      "eval_samples_per_second": 1.053,
      "eval_steps_per_second": 0.066,
      "step": 3454
    },
    {
      "epoch": 2.01,
      "grad_norm": 8.240702629089355,
      "learning_rate": 1.833976833976834e-05,
      "loss": 0.4854,
      "step": 3475
    },
    {
      "epoch": 2.03,
      "grad_norm": 7.087810039520264,
      "learning_rate": 1.8071643071643072e-05,
      "loss": 0.5459,
      "step": 3500
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7334815859794617,
      "learning_rate": 1.7803517803517805e-05,
      "loss": 0.5212,
      "step": 3525
    },
    {
      "epoch": 2.06,
      "grad_norm": 6.095980167388916,
      "learning_rate": 1.7535392535392538e-05,
      "loss": 0.5191,
      "step": 3550
    },
    {
      "epoch": 2.07,
      "grad_norm": 10.246546745300293,
      "learning_rate": 1.7267267267267267e-05,
      "loss": 0.6645,
      "step": 3575
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.6809566020965576,
      "learning_rate": 1.6999141999142e-05,
      "loss": 0.5629,
      "step": 3600
    },
    {
      "epoch": 2.1,
      "grad_norm": 13.609752655029297,
      "learning_rate": 1.673101673101673e-05,
      "loss": 0.6598,
      "step": 3625
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.554472923278809,
      "learning_rate": 1.6462891462891464e-05,
      "loss": 0.4935,
      "step": 3650
    },
    {
      "epoch": 2.13,
      "grad_norm": 16.72881317138672,
      "learning_rate": 1.6205491205491204e-05,
      "loss": 0.379,
      "step": 3675
    },
    {
      "epoch": 2.14,
      "grad_norm": 9.573266983032227,
      "learning_rate": 1.593736593736594e-05,
      "loss": 0.7141,
      "step": 3700
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.245655059814453,
      "learning_rate": 1.566924066924067e-05,
      "loss": 0.7518,
      "step": 3725
    },
    {
      "epoch": 2.17,
      "grad_norm": 13.11945915222168,
      "learning_rate": 1.54011154011154e-05,
      "loss": 0.6588,
      "step": 3750
    },
    {
      "epoch": 2.19,
      "grad_norm": 6.219137191772461,
      "learning_rate": 1.5132990132990132e-05,
      "loss": 0.5884,
      "step": 3775
    },
    {
      "epoch": 2.2,
      "grad_norm": 6.516097545623779,
      "learning_rate": 1.4864864864864867e-05,
      "loss": 0.4572,
      "step": 3800
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.23282527923584,
      "learning_rate": 1.4596739596739597e-05,
      "loss": 0.6028,
      "step": 3825
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.6169815063476562,
      "learning_rate": 1.4328614328614328e-05,
      "loss": 0.6198,
      "step": 3850
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.720090389251709,
      "learning_rate": 1.4060489060489059e-05,
      "loss": 0.556,
      "step": 3875
    },
    {
      "epoch": 2.26,
      "grad_norm": 7.367048263549805,
      "learning_rate": 1.3792363792363793e-05,
      "loss": 0.5812,
      "step": 3900
    },
    {
      "epoch": 2.27,
      "grad_norm": 7.3934173583984375,
      "learning_rate": 1.3524238524238526e-05,
      "loss": 0.6141,
      "step": 3925
    },
    {
      "epoch": 2.29,
      "grad_norm": 10.528743743896484,
      "learning_rate": 1.3256113256113257e-05,
      "loss": 0.6047,
      "step": 3950
    },
    {
      "epoch": 2.3,
      "grad_norm": 7.282771110534668,
      "learning_rate": 1.2987987987987987e-05,
      "loss": 0.6281,
      "step": 3975
    },
    {
      "epoch": 2.32,
      "grad_norm": 11.970826148986816,
      "learning_rate": 1.2719862719862722e-05,
      "loss": 0.4103,
      "step": 4000
    },
    {
      "epoch": 2.33,
      "grad_norm": 6.620480537414551,
      "learning_rate": 1.2451737451737452e-05,
      "loss": 0.5386,
      "step": 4025
    },
    {
      "epoch": 2.35,
      "grad_norm": 5.699476718902588,
      "learning_rate": 1.2183612183612183e-05,
      "loss": 0.6507,
      "step": 4050
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.477766275405884,
      "learning_rate": 1.1915486915486916e-05,
      "loss": 0.524,
      "step": 4075
    },
    {
      "epoch": 2.37,
      "grad_norm": 6.517852306365967,
      "learning_rate": 1.1647361647361647e-05,
      "loss": 0.6979,
      "step": 4100
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.506691932678223,
      "learning_rate": 1.1379236379236379e-05,
      "loss": 0.4651,
      "step": 4125
    },
    {
      "epoch": 2.4,
      "grad_norm": 6.522432804107666,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.6845,
      "step": 4150
    },
    {
      "epoch": 2.42,
      "grad_norm": 12.015291213989258,
      "learning_rate": 1.0842985842985844e-05,
      "loss": 0.5348,
      "step": 4175
    },
    {
      "epoch": 2.43,
      "grad_norm": 7.297937393188477,
      "learning_rate": 1.0574860574860575e-05,
      "loss": 0.5412,
      "step": 4200
    },
    {
      "epoch": 2.45,
      "grad_norm": 13.665657997131348,
      "learning_rate": 1.0306735306735307e-05,
      "loss": 0.5137,
      "step": 4225
    },
    {
      "epoch": 2.46,
      "grad_norm": 11.743260383605957,
      "learning_rate": 1.0038610038610038e-05,
      "loss": 0.5738,
      "step": 4250
    },
    {
      "epoch": 2.48,
      "grad_norm": 10.24691104888916,
      "learning_rate": 9.77048477048477e-06,
      "loss": 0.5134,
      "step": 4275
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.543239116668701,
      "learning_rate": 9.502359502359502e-06,
      "loss": 0.6055,
      "step": 4300
    },
    {
      "epoch": 2.5,
      "grad_norm": 7.915064811706543,
      "learning_rate": 9.234234234234234e-06,
      "loss": 0.5153,
      "step": 4325
    },
    {
      "epoch": 2.52,
      "grad_norm": 8.37210750579834,
      "learning_rate": 8.966108966108967e-06,
      "loss": 0.4754,
      "step": 4350
    },
    {
      "epoch": 2.53,
      "grad_norm": 7.3417487144470215,
      "learning_rate": 8.6979836979837e-06,
      "loss": 0.6285,
      "step": 4375
    },
    {
      "epoch": 2.55,
      "grad_norm": 9.027023315429688,
      "learning_rate": 8.42985842985843e-06,
      "loss": 0.4925,
      "step": 4400
    },
    {
      "epoch": 2.56,
      "grad_norm": 7.813179016113281,
      "learning_rate": 8.161733161733163e-06,
      "loss": 0.5089,
      "step": 4425
    },
    {
      "epoch": 2.58,
      "grad_norm": 13.45531940460205,
      "learning_rate": 7.893607893607893e-06,
      "loss": 0.4717,
      "step": 4450
    },
    {
      "epoch": 2.59,
      "grad_norm": 7.09887170791626,
      "learning_rate": 7.625482625482626e-06,
      "loss": 0.6506,
      "step": 4475
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.6297383308410645,
      "learning_rate": 7.357357357357357e-06,
      "loss": 0.4828,
      "step": 4500
    },
    {
      "epoch": 2.62,
      "grad_norm": 8.980986595153809,
      "learning_rate": 7.089232089232089e-06,
      "loss": 0.4233,
      "step": 4525
    },
    {
      "epoch": 2.63,
      "grad_norm": 26.76249885559082,
      "learning_rate": 6.821106821106821e-06,
      "loss": 0.5748,
      "step": 4550
    },
    {
      "epoch": 2.65,
      "grad_norm": 8.72842788696289,
      "learning_rate": 6.552981552981553e-06,
      "loss": 0.6565,
      "step": 4575
    },
    {
      "epoch": 2.66,
      "grad_norm": 9.191315650939941,
      "learning_rate": 6.284856284856284e-06,
      "loss": 0.5332,
      "step": 4600
    },
    {
      "epoch": 2.68,
      "grad_norm": 7.631181240081787,
      "learning_rate": 6.016731016731017e-06,
      "loss": 0.4692,
      "step": 4625
    },
    {
      "epoch": 2.69,
      "grad_norm": 7.811351299285889,
      "learning_rate": 5.748605748605749e-06,
      "loss": 0.6485,
      "step": 4650
    },
    {
      "epoch": 2.71,
      "grad_norm": 8.997116088867188,
      "learning_rate": 5.480480480480481e-06,
      "loss": 0.4207,
      "step": 4675
    },
    {
      "epoch": 2.72,
      "grad_norm": 9.758033752441406,
      "learning_rate": 5.212355212355213e-06,
      "loss": 0.5205,
      "step": 4700
    },
    {
      "epoch": 2.74,
      "grad_norm": 5.98590612411499,
      "learning_rate": 4.944229944229944e-06,
      "loss": 0.6115,
      "step": 4725
    },
    {
      "epoch": 2.75,
      "grad_norm": 11.506319046020508,
      "learning_rate": 4.676104676104676e-06,
      "loss": 0.4449,
      "step": 4750
    },
    {
      "epoch": 2.76,
      "grad_norm": 7.969517230987549,
      "learning_rate": 4.4079794079794084e-06,
      "loss": 0.5384,
      "step": 4775
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8463253974914551,
      "learning_rate": 4.13985413985414e-06,
      "loss": 0.4981,
      "step": 4800
    },
    {
      "epoch": 2.79,
      "grad_norm": 12.771890640258789,
      "learning_rate": 3.871728871728872e-06,
      "loss": 0.4786,
      "step": 4825
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.6047688126564026,
      "learning_rate": 3.603603603603604e-06,
      "loss": 0.4913,
      "step": 4850
    },
    {
      "epoch": 2.82,
      "grad_norm": 7.117040157318115,
      "learning_rate": 3.3354783354783355e-06,
      "loss": 0.5386,
      "step": 4875
    },
    {
      "epoch": 2.84,
      "grad_norm": 5.262890338897705,
      "learning_rate": 3.0673530673530676e-06,
      "loss": 0.6115,
      "step": 4900
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.5500399470329285,
      "learning_rate": 2.7992277992277993e-06,
      "loss": 0.5285,
      "step": 4925
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.5653154850006104,
      "learning_rate": 2.531102531102531e-06,
      "loss": 0.3621,
      "step": 4950
    },
    {
      "epoch": 2.88,
      "grad_norm": 5.71751594543457,
      "learning_rate": 2.262977262977263e-06,
      "loss": 0.6007,
      "step": 4975
    },
    {
      "epoch": 2.9,
      "grad_norm": 12.138904571533203,
      "learning_rate": 1.9948519948519947e-06,
      "loss": 0.5797,
      "step": 5000
    },
    {
      "epoch": 2.91,
      "grad_norm": 8.795024871826172,
      "learning_rate": 1.7267267267267268e-06,
      "loss": 0.5585,
      "step": 5025
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.7619569301605225,
      "learning_rate": 1.4586014586014587e-06,
      "loss": 0.3951,
      "step": 5050
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.763073205947876,
      "learning_rate": 1.1904761904761904e-06,
      "loss": 0.287,
      "step": 5075
    },
    {
      "epoch": 2.95,
      "grad_norm": 7.337412357330322,
      "learning_rate": 9.223509223509224e-07,
      "loss": 0.4416,
      "step": 5100
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.266438961029053,
      "learning_rate": 6.542256542256542e-07,
      "loss": 0.5275,
      "step": 5125
    },
    {
      "epoch": 2.98,
      "grad_norm": 8.12879753112793,
      "learning_rate": 3.8610038610038613e-07,
      "loss": 0.4297,
      "step": 5150
    },
    {
      "epoch": 3.0,
      "grad_norm": 6.199108123779297,
      "learning_rate": 1.1797511797511798e-07,
      "loss": 0.4389,
      "step": 5175
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8602430555555556,
      "eval_f1_macro": 0.6149830093941424,
      "eval_f1_micro": 0.8602430555555556,
      "eval_f1_weighted": 0.8515059109185544,
      "eval_loss": 0.4315283000469208,
      "eval_precision_macro": 0.7610988679415244,
      "eval_precision_micro": 0.8602430555555556,
      "eval_precision_weighted": 0.8532444856848228,
      "eval_recall_macro": 0.5527145295483504,
      "eval_recall_micro": 0.8602430555555556,
      "eval_recall_weighted": 0.8602430555555556,
      "eval_runtime": 3373.7409,
      "eval_samples_per_second": 1.024,
      "eval_steps_per_second": 0.064,
      "step": 5181
    }
  ],
  "logging_steps": 25,
  "max_steps": 5181,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.135272556528692e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}