{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.4362863380443525,
  "global_step": 12500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.968083748244606e-05,
      "loss": 2.9096,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.936167496489212e-05,
      "loss": 2.8631,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.904251244733819e-05,
      "loss": 2.7364,
      "step": 300
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.8723349929784253e-05,
      "loss": 2.8226,
      "step": 400
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.840418741223031e-05,
      "loss": 2.7184,
      "step": 500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.808502489467637e-05,
      "loss": 2.6843,
      "step": 600
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.776586237712243e-05,
      "loss": 2.7507,
      "step": 700
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.744669985956849e-05,
      "loss": 2.7261,
      "step": 800
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.712753734201456e-05,
      "loss": 2.7343,
      "step": 900
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.6808374824460616e-05,
      "loss": 2.7005,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.6489212306906676e-05,
      "loss": 2.6462,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.617004978935274e-05,
      "loss": 2.6276,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.58508872717988e-05,
      "loss": 2.6493,
      "step": 1300
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.553172475424487e-05,
      "loss": 2.7024,
      "step": 1400
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.521256223669093e-05,
      "loss": 2.6616,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.4893399719136986e-05,
      "loss": 2.7136,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.4574237201583046e-05,
      "loss": 2.6842,
      "step": 1700
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.4255074684029105e-05,
      "loss": 2.5911,
      "step": 1800
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.393591216647517e-05,
      "loss": 2.7281,
      "step": 1900
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.361674964892124e-05,
      "loss": 2.6697,
      "step": 2000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.32975871313673e-05,
      "loss": 2.6282,
      "step": 2100
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.2978424613813356e-05,
      "loss": 2.564,
      "step": 2200
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.2659262096259415e-05,
      "loss": 2.4882,
      "step": 2300
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.2340099578705475e-05,
      "loss": 2.6347,
      "step": 2400
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.202093706115154e-05,
      "loss": 2.5994,
      "step": 2500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.17017745435976e-05,
      "loss": 2.6479,
      "step": 2600
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.1382612026043666e-05,
      "loss": 2.7357,
      "step": 2700
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.1063449508489726e-05,
      "loss": 2.6819,
      "step": 2800
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.0744286990935785e-05,
      "loss": 2.5451,
      "step": 2900
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.0425124473381845e-05,
      "loss": 2.6343,
      "step": 3000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.010596195582791e-05,
      "loss": 2.5168,
      "step": 3100
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.978679943827397e-05,
      "loss": 2.5881,
      "step": 3200
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.946763692072003e-05,
      "loss": 2.6411,
      "step": 3300
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.9148474403166096e-05,
      "loss": 2.5901,
      "step": 3400
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.8829311885612155e-05,
      "loss": 2.5842,
      "step": 3500
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.851014936805822e-05,
      "loss": 2.5436,
      "step": 3600
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.819098685050428e-05,
      "loss": 2.6021,
      "step": 3700
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.787182433295034e-05,
      "loss": 2.6307,
      "step": 3800
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.75526618153964e-05,
      "loss": 2.5632,
      "step": 3900
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.723349929784246e-05,
      "loss": 2.5933,
      "step": 4000
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.6914336780288525e-05,
      "loss": 2.5458,
      "step": 4100
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.659517426273459e-05,
      "loss": 2.5212,
      "step": 4200
    },
    {
      "epoch": 0.49,
      "learning_rate": 3.627601174518065e-05,
      "loss": 2.4532,
      "step": 4300
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.595684922762671e-05,
      "loss": 2.5134,
      "step": 4400
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.563768671007277e-05,
      "loss": 2.5368,
      "step": 4500
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.531852419251883e-05,
      "loss": 2.5547,
      "step": 4600
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.4999361674964895e-05,
      "loss": 2.4786,
      "step": 4700
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.4680199157410954e-05,
      "loss": 2.5377,
      "step": 4800
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.436103663985702e-05,
      "loss": 2.5462,
      "step": 4900
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.404187412230308e-05,
      "loss": 2.5644,
      "step": 5000
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.372271160474914e-05,
      "loss": 2.5206,
      "step": 5100
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.3403549087195205e-05,
      "loss": 2.5305,
      "step": 5200
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.3084386569641264e-05,
      "loss": 2.462,
      "step": 5300
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.2765224052087324e-05,
      "loss": 2.4704,
      "step": 5400
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.244606153453338e-05,
      "loss": 2.5133,
      "step": 5500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.212689901697944e-05,
      "loss": 2.4805,
      "step": 5600
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.180773649942551e-05,
      "loss": 2.4572,
      "step": 5700
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.1488573981871575e-05,
      "loss": 2.4128,
      "step": 5800
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.1169411464317634e-05,
      "loss": 2.4577,
      "step": 5900
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.0850248946763694e-05,
      "loss": 2.5539,
      "step": 6000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.053108642920975e-05,
      "loss": 2.472,
      "step": 6100
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.0211923911655816e-05,
      "loss": 2.4468,
      "step": 6200
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.9892761394101875e-05,
      "loss": 2.4831,
      "step": 6300
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.957359887654794e-05,
      "loss": 2.4093,
      "step": 6400
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9254436358994004e-05,
      "loss": 2.4683,
      "step": 6500
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8935273841440063e-05,
      "loss": 2.4388,
      "step": 6600
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.8616111323886126e-05,
      "loss": 2.3558,
      "step": 6700
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.8296948806332185e-05,
      "loss": 2.4788,
      "step": 6800
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.7977786288778245e-05,
      "loss": 2.4523,
      "step": 6900
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.7658623771224308e-05,
      "loss": 2.4759,
      "step": 7000
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.7339461253670367e-05,
      "loss": 2.3574,
      "step": 7100
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.7020298736116433e-05,
      "loss": 2.4569,
      "step": 7200
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.6701136218562496e-05,
      "loss": 2.4078,
      "step": 7300
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.6381973701008555e-05,
      "loss": 2.3948,
      "step": 7400
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.6062811183454615e-05,
      "loss": 2.5496,
      "step": 7500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.5743648665900677e-05,
      "loss": 2.4191,
      "step": 7600
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.5424486148346737e-05,
      "loss": 2.4895,
      "step": 7700
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.51053236307928e-05,
      "loss": 2.4726,
      "step": 7800
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.4786161113238862e-05,
      "loss": 2.3287,
      "step": 7900
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.446699859568492e-05,
      "loss": 2.4278,
      "step": 8000
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.4147836078130988e-05,
      "loss": 2.4548,
      "step": 8100
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.3828673560577047e-05,
      "loss": 2.5173,
      "step": 8200
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.3509511043023107e-05,
      "loss": 2.4883,
      "step": 8300
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.319034852546917e-05,
      "loss": 2.4435,
      "step": 8400
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.2871186007915232e-05,
      "loss": 2.4825,
      "step": 8500
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.2552023490361295e-05,
      "loss": 2.4539,
      "step": 8600
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.2232860972807354e-05,
      "loss": 2.4115,
      "step": 8700
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.1913698455253417e-05,
      "loss": 2.3452,
      "step": 8800
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.159453593769948e-05,
      "loss": 2.3799,
      "step": 8900
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.127537342014554e-05,
      "loss": 2.4019,
      "step": 9000
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.09562109025916e-05,
      "loss": 2.3678,
      "step": 9100
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.0637048385037665e-05,
      "loss": 2.3711,
      "step": 9200
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.0317885867483724e-05,
      "loss": 2.2767,
      "step": 9300
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.9998723349929783e-05,
      "loss": 2.3761,
      "step": 9400
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.9679560832375846e-05,
      "loss": 2.3392,
      "step": 9500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.936039831482191e-05,
      "loss": 2.3232,
      "step": 9600
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.904123579726797e-05,
      "loss": 2.355,
      "step": 9700
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.872207327971403e-05,
      "loss": 2.3613,
      "step": 9800
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.840291076216009e-05,
      "loss": 2.3475,
      "step": 9900
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.8083748244606157e-05,
      "loss": 2.3654,
      "step": 10000
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.7764585727052216e-05,
      "loss": 2.3825,
      "step": 10100
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.7445423209498275e-05,
      "loss": 2.3245,
      "step": 10200
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.712626069194434e-05,
      "loss": 2.2869,
      "step": 10300
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.68070981743904e-05,
      "loss": 2.3207,
      "step": 10400
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.648793565683646e-05,
      "loss": 2.3041,
      "step": 10500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.6168773139282523e-05,
      "loss": 2.3543,
      "step": 10600
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.5849610621728586e-05,
      "loss": 2.3026,
      "step": 10700
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.553044810417465e-05,
      "loss": 2.2919,
      "step": 10800
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.5211285586620708e-05,
      "loss": 2.3805,
      "step": 10900
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.4892123069066769e-05,
      "loss": 2.3523,
      "step": 11000
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.4572960551512832e-05,
      "loss": 2.2931,
      "step": 11100
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.4253798033958893e-05,
      "loss": 2.2623,
      "step": 11200
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.3934635516404954e-05,
      "loss": 2.28,
      "step": 11300
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.3615472998851015e-05,
      "loss": 2.3121,
      "step": 11400
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.3296310481297078e-05,
      "loss": 2.3111,
      "step": 11500
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2977147963743139e-05,
      "loss": 2.332,
      "step": 11600
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.26579854461892e-05,
      "loss": 2.3495,
      "step": 11700
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.233882292863526e-05,
      "loss": 2.2864,
      "step": 11800
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2019660411081324e-05,
      "loss": 2.3482,
      "step": 11900
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1700497893527385e-05,
      "loss": 2.2843,
      "step": 12000
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1381335375973447e-05,
      "loss": 2.3075,
      "step": 12100
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1062172858419507e-05,
      "loss": 2.3899,
      "step": 12200
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.074301034086557e-05,
      "loss": 2.3496,
      "step": 12300
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.042384782331163e-05,
      "loss": 2.3482,
      "step": 12400
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0104685305757693e-05,
      "loss": 2.2579,
      "step": 12500
    }
  ],
  "max_steps": 15666,
  "num_train_epochs": 2,
  "total_flos": 4.940417519751168e+16,
  "trial_name": null,
  "trial_params": null
}