{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9997613053189132,
  "eval_steps": 500,
  "global_step": 10996,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 0.0004977264459803565,
      "loss": 5.874,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0004954528919607129,
      "loss": 4.0765,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0004931793379410695,
      "loss": 3.7752,
      "step": 150
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000490905783921426,
      "loss": 3.6357,
      "step": 200
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0004886322299017824,
      "loss": 3.5374,
      "step": 250
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00048635867588213893,
      "loss": 3.4875,
      "step": 300
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0004840851218624955,
      "loss": 3.411,
      "step": 350
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00048181156784285196,
      "loss": 3.3791,
      "step": 400
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0004795380138232084,
      "loss": 3.3377,
      "step": 450
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00047726445980356494,
      "loss": 3.292,
      "step": 500
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00047499090578392145,
      "loss": 3.2765,
      "step": 550
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0004727173517642779,
      "loss": 3.2584,
      "step": 600
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0004704437977446344,
      "loss": 3.2319,
      "step": 650
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0004681702437249909,
      "loss": 3.2021,
      "step": 700
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00046589668970534745,
      "loss": 3.1816,
      "step": 750
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0004636231356857039,
      "loss": 3.165,
      "step": 800
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0004613495816660604,
      "loss": 3.1436,
      "step": 850
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0004590760276464169,
      "loss": 3.1491,
      "step": 900
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0004568024736267734,
      "loss": 3.1343,
      "step": 950
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00045452891960712986,
      "loss": 3.1208,
      "step": 1000
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0004522553655874864,
      "loss": 3.0954,
      "step": 1050
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00044998181156784284,
      "loss": 3.0963,
      "step": 1100
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00044770825754819935,
      "loss": 3.098,
      "step": 1150
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00044543470352855586,
      "loss": 3.0695,
      "step": 1200
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0004431611495089123,
      "loss": 3.0785,
      "step": 1250
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00044088759548926884,
      "loss": 3.0524,
      "step": 1300
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0004386140414696253,
      "loss": 3.0275,
      "step": 1350
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0004363404874499818,
      "loss": 3.0368,
      "step": 1400
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00043406693343033833,
      "loss": 3.0177,
      "step": 1450
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0004317933794106948,
      "loss": 3.0117,
      "step": 1500
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0004295198253910513,
      "loss": 3.0023,
      "step": 1550
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0004272462713714078,
      "loss": 2.9814,
      "step": 1600
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0004249727173517643,
      "loss": 2.9867,
      "step": 1650
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0004226991633321208,
      "loss": 2.991,
      "step": 1700
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00042042560931247725,
      "loss": 2.9764,
      "step": 1750
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00041815205529283376,
      "loss": 2.9587,
      "step": 1800
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0004158785012731903,
      "loss": 2.9651,
      "step": 1850
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00041360494725354674,
      "loss": 2.9606,
      "step": 1900
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00041133139323390325,
      "loss": 2.9448,
      "step": 1950
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00040905783921425977,
      "loss": 2.9585,
      "step": 2000
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00040678428519461623,
      "loss": 2.9607,
      "step": 2050
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00040451073117497274,
      "loss": 2.9199,
      "step": 2100
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0004022371771553292,
      "loss": 2.9303,
      "step": 2150
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0003999636231356857,
      "loss": 2.9353,
      "step": 2200
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00039769006911604223,
      "loss": 2.9123,
      "step": 2250
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0003954165150963987,
      "loss": 2.9161,
      "step": 2300
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0003931429610767552,
      "loss": 2.9146,
      "step": 2350
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00039086940705711167,
      "loss": 2.9086,
      "step": 2400
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0003885958530374682,
      "loss": 2.9094,
      "step": 2450
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0003863222990178247,
      "loss": 2.8968,
      "step": 2500
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00038404874499818115,
      "loss": 2.9008,
      "step": 2550
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0003817751909785376,
      "loss": 2.8836,
      "step": 2600
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0003795016369588942,
      "loss": 2.8773,
      "step": 2650
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00037722808293925064,
      "loss": 2.8648,
      "step": 2700
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00037495452891960716,
      "loss": 2.8699,
      "step": 2750
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0003726809748999636,
      "loss": 2.8693,
      "step": 2800
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00037040742088032013,
      "loss": 2.8619,
      "step": 2850
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00036813386686067665,
      "loss": 2.8591,
      "step": 2900
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0003658603128410331,
      "loss": 2.8614,
      "step": 2950
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00036358675882138957,
      "loss": 2.8605,
      "step": 3000
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00036131320480174613,
      "loss": 2.8357,
      "step": 3050
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0003590396507821026,
      "loss": 2.847,
      "step": 3100
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00035676609676245905,
      "loss": 2.8503,
      "step": 3150
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00035449254274281557,
      "loss": 2.8334,
      "step": 3200
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0003522189887231721,
      "loss": 2.8424,
      "step": 3250
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0003499454347035286,
      "loss": 2.8295,
      "step": 3300
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00034767188068388506,
      "loss": 2.8359,
      "step": 3350
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0003453983266642415,
      "loss": 2.8252,
      "step": 3400
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00034312477264459803,
      "loss": 2.82,
      "step": 3450
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00034085121862495455,
      "loss": 2.8134,
      "step": 3500
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.000338577664605311,
      "loss": 2.8168,
      "step": 3550
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0003363041105856675,
      "loss": 2.8248,
      "step": 3600
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000334030556566024,
      "loss": 2.8096,
      "step": 3650
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00033175700254638055,
      "loss": 2.8154,
      "step": 3700
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.000329483448526737,
      "loss": 2.8015,
      "step": 3750
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00032720989450709347,
      "loss": 2.8024,
      "step": 3800
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00032493634048745,
      "loss": 2.7998,
      "step": 3850
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0003226627864678065,
      "loss": 2.8011,
      "step": 3900
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00032038923244816296,
      "loss": 2.7963,
      "step": 3950
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00031811567842851947,
      "loss": 2.7863,
      "step": 4000
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00031584212440887593,
      "loss": 2.7973,
      "step": 4050
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0003135685703892325,
      "loss": 2.788,
      "step": 4100
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00031129501636958896,
      "loss": 2.8026,
      "step": 4150
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0003090214623499454,
      "loss": 2.7902,
      "step": 4200
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00030674790833030193,
      "loss": 2.7684,
      "step": 4250
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00030447435431065845,
      "loss": 2.7846,
      "step": 4300
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0003022008002910149,
      "loss": 2.7703,
      "step": 4350
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0002999272462713714,
      "loss": 2.7746,
      "step": 4400
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0002976536922517279,
      "loss": 2.7674,
      "step": 4450
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00029538013823208445,
      "loss": 2.7686,
      "step": 4500
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0002931065842124409,
      "loss": 2.7773,
      "step": 4550
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00029083303019279737,
      "loss": 2.7475,
      "step": 4600
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0002885594761731539,
      "loss": 2.7616,
      "step": 4650
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00028628592215351035,
      "loss": 2.756,
      "step": 4700
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00028401236813386686,
      "loss": 2.7548,
      "step": 4750
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0002817388141142234,
      "loss": 2.7499,
      "step": 4800
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00027946526009457983,
      "loss": 2.7637,
      "step": 4850
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00027719170607493635,
      "loss": 2.7349,
      "step": 4900
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00027491815205529286,
      "loss": 2.7346,
      "step": 4950
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002726445980356493,
      "loss": 2.7384,
      "step": 5000
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00027037104401600584,
      "loss": 2.724,
      "step": 5050
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002680974899963623,
      "loss": 2.7508,
      "step": 5100
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002658239359767188,
      "loss": 2.7523,
      "step": 5150
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002635503819570753,
      "loss": 2.7298,
      "step": 5200
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0002612768279374318,
      "loss": 2.7281,
      "step": 5250
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002590032739177883,
      "loss": 2.7215,
      "step": 5300
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002567297198981448,
      "loss": 2.7318,
      "step": 5350
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002544561658785013,
      "loss": 2.7282,
      "step": 5400
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002521826118588578,
      "loss": 2.7313,
      "step": 5450
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002499090578392143,
      "loss": 2.7133,
      "step": 5500
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00024763550381957076,
      "loss": 2.6344,
      "step": 5550
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002453619497999273,
      "loss": 2.6231,
      "step": 5600
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00024308839578028374,
      "loss": 2.609,
      "step": 5650
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00024081484176064022,
      "loss": 2.6154,
      "step": 5700
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00023854128774099674,
      "loss": 2.6371,
      "step": 5750
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0002362677337213532,
      "loss": 2.6082,
      "step": 5800
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002339941797017097,
      "loss": 2.6218,
      "step": 5850
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0002317206256820662,
      "loss": 2.6284,
      "step": 5900
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00022944707166242271,
      "loss": 2.6191,
      "step": 5950
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00022717351764277917,
      "loss": 2.6252,
      "step": 6000
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0002248999636231357,
      "loss": 2.6266,
      "step": 6050
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00022262640960349218,
      "loss": 2.6426,
      "step": 6100
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0002203528555838487,
      "loss": 2.6072,
      "step": 6150
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00021807930156420515,
      "loss": 2.614,
      "step": 6200
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00021580574754456166,
      "loss": 2.5959,
      "step": 6250
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00021353219352491815,
      "loss": 2.6065,
      "step": 6300
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00021125863950527467,
      "loss": 2.6163,
      "step": 6350
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00020898508548563113,
      "loss": 2.6083,
      "step": 6400
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00020671153146598764,
      "loss": 2.6119,
      "step": 6450
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00020443797744634413,
      "loss": 2.6023,
      "step": 6500
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00020216442342670064,
      "loss": 2.6086,
      "step": 6550
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001998908694070571,
      "loss": 2.5988,
      "step": 6600
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019761731538741362,
      "loss": 2.6046,
      "step": 6650
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001953437613677701,
      "loss": 2.5929,
      "step": 6700
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019307020734812662,
      "loss": 2.5935,
      "step": 6750
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00019079665332848308,
      "loss": 2.5875,
      "step": 6800
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00018852309930883956,
      "loss": 2.6001,
      "step": 6850
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00018624954528919608,
      "loss": 2.6021,
      "step": 6900
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00018397599126955257,
      "loss": 2.5887,
      "step": 6950
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00018170243724990905,
      "loss": 2.6022,
      "step": 7000
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00017942888323026554,
      "loss": 2.5965,
      "step": 7050
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00017715532921062205,
      "loss": 2.5879,
      "step": 7100
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00017488177519097854,
      "loss": 2.5935,
      "step": 7150
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00017260822117133503,
      "loss": 2.6041,
      "step": 7200
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00017033466715169152,
      "loss": 2.6041,
      "step": 7250
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00016806111313204803,
      "loss": 2.6024,
      "step": 7300
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00016578755911240452,
      "loss": 2.5995,
      "step": 7350
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.000163514005092761,
      "loss": 2.5959,
      "step": 7400
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001612404510731175,
      "loss": 2.579,
      "step": 7450
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.000158966897053474,
      "loss": 2.5807,
      "step": 7500
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001566933430338305,
      "loss": 2.5922,
      "step": 7550
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00015441978901418698,
      "loss": 2.5729,
      "step": 7600
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00015214623499454347,
      "loss": 2.5677,
      "step": 7650
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00014987268097489998,
      "loss": 2.5841,
      "step": 7700
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00014759912695525647,
      "loss": 2.5908,
      "step": 7750
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00014532557293561296,
      "loss": 2.5857,
      "step": 7800
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00014305201891596944,
      "loss": 2.588,
      "step": 7850
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00014077846489632596,
      "loss": 2.5843,
      "step": 7900
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00013850491087668244,
      "loss": 2.5802,
      "step": 7950
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001362313568570389,
      "loss": 2.5631,
      "step": 8000
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00013395780283739542,
      "loss": 2.5767,
      "step": 8050
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001316842488177519,
      "loss": 2.563,
      "step": 8100
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00012941069479810842,
      "loss": 2.5844,
      "step": 8150
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00012713714077846488,
      "loss": 2.5721,
      "step": 8200
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0001248635867588214,
      "loss": 2.5871,
      "step": 8250
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00012259003273917788,
      "loss": 2.5781,
      "step": 8300
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00012031647871953438,
      "loss": 2.5646,
      "step": 8350
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00011804292469989088,
      "loss": 2.5614,
      "step": 8400
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00011576937068024737,
      "loss": 2.568,
      "step": 8450
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00011349581666060387,
      "loss": 2.5592,
      "step": 8500
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00011122226264096034,
      "loss": 2.5672,
      "step": 8550
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00010894870862131685,
      "loss": 2.5649,
      "step": 8600
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00010667515460167333,
      "loss": 2.5686,
      "step": 8650
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00010440160058202983,
      "loss": 2.5758,
      "step": 8700
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00010212804656238632,
      "loss": 2.565,
      "step": 8750
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.985449254274281e-05,
      "loss": 2.5904,
      "step": 8800
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.758093852309931e-05,
      "loss": 2.5668,
      "step": 8850
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.53073845034558e-05,
      "loss": 2.5714,
      "step": 8900
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.30338304838123e-05,
      "loss": 2.5511,
      "step": 8950
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.076027646416878e-05,
      "loss": 2.5756,
      "step": 9000
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.848672244452528e-05,
      "loss": 2.5688,
      "step": 9050
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.621316842488177e-05,
      "loss": 2.5567,
      "step": 9100
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.393961440523827e-05,
      "loss": 2.561,
      "step": 9150
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.166606038559476e-05,
      "loss": 2.5525,
      "step": 9200
    },
    {
      "epoch": 1.68,
      "learning_rate": 7.939250636595126e-05,
      "loss": 2.5665,
      "step": 9250
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.711895234630775e-05,
      "loss": 2.5686,
      "step": 9300
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.484539832666425e-05,
      "loss": 2.5627,
      "step": 9350
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.257184430702073e-05,
      "loss": 2.5454,
      "step": 9400
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.029829028737724e-05,
      "loss": 2.5677,
      "step": 9450
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.802473626773372e-05,
      "loss": 2.572,
      "step": 9500
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.575118224809022e-05,
      "loss": 2.5468,
      "step": 9550
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.347762822844671e-05,
      "loss": 2.5627,
      "step": 9600
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.12040742088032e-05,
      "loss": 2.5523,
      "step": 9650
    },
    {
      "epoch": 1.76,
      "learning_rate": 5.89305201891597e-05,
      "loss": 2.559,
      "step": 9700
    },
    {
      "epoch": 1.77,
      "learning_rate": 5.665696616951619e-05,
      "loss": 2.5542,
      "step": 9750
    },
    {
      "epoch": 1.78,
      "learning_rate": 5.4383412149872686e-05,
      "loss": 2.5632,
      "step": 9800
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.210985813022918e-05,
      "loss": 2.5646,
      "step": 9850
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.9836304110585674e-05,
      "loss": 2.545,
      "step": 9900
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.756275009094216e-05,
      "loss": 2.5525,
      "step": 9950
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.5289196071298655e-05,
      "loss": 2.5823,
      "step": 10000
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.301564205165515e-05,
      "loss": 2.5547,
      "step": 10050
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.074208803201164e-05,
      "loss": 2.5645,
      "step": 10100
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.846853401236814e-05,
      "loss": 2.5493,
      "step": 10150
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.619497999272463e-05,
      "loss": 2.5531,
      "step": 10200
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.392142597308112e-05,
      "loss": 2.5537,
      "step": 10250
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.164787195343761e-05,
      "loss": 2.5428,
      "step": 10300
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.937431793379411e-05,
      "loss": 2.5534,
      "step": 10350
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.71007639141506e-05,
      "loss": 2.5462,
      "step": 10400
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.4827209894507094e-05,
      "loss": 2.5555,
      "step": 10450
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.2553655874863588e-05,
      "loss": 2.5622,
      "step": 10500
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.028010185522008e-05,
      "loss": 2.5599,
      "step": 10550
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.8006547835576576e-05,
      "loss": 2.5406,
      "step": 10600
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.5732993815933066e-05,
      "loss": 2.5642,
      "step": 10650
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.345943979628956e-05,
      "loss": 2.5589,
      "step": 10700
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.1185885776646052e-05,
      "loss": 2.5585,
      "step": 10750
    },
    {
      "epoch": 1.96,
      "learning_rate": 8.912331757002546e-06,
      "loss": 2.555,
      "step": 10800
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.638777737359039e-06,
      "loss": 2.5504,
      "step": 10850
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.3652237177155325e-06,
      "loss": 2.5553,
      "step": 10900
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.0916696980720264e-06,
      "loss": 2.5582,
      "step": 10950
    },
    {
      "epoch": 2.0,
      "step": 10996,
      "total_flos": 7.20906520790827e+17,
      "train_loss": 2.7760973889075786,
      "train_runtime": 9463.0789,
      "train_samples_per_second": 37.188,
      "train_steps_per_second": 1.162
    }
  ],
  "logging_steps": 50,
  "max_steps": 10996,
  "num_train_epochs": 2,
  "save_steps": 2500,
  "total_flos": 7.20906520790827e+17,
  "trial_name": null,
  "trial_params": null
}