{
  "best_metric": 0.1957029402256012,
  "best_model_checkpoint": "whisper-base-full-data-language-v2-20ep/checkpoint-60000",
  "epoch": 20.003133322888925,
  "global_step": 63840,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 1.6223,
      "step": 100
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.999999999999999e-06,
      "loss": 0.6381,
      "step": 200
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.999999999999999e-06,
      "loss": 0.5142,
      "step": 300
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 0.4612,
      "step": 400
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 0.4227,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 0.4016,
      "step": 600
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.1e-05,
      "loss": 0.3848,
      "step": 700
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 0.3705,
      "step": 800
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 0.3638,
      "step": 900
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 0.3519,
      "step": 1000
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 0.3435,
      "step": 1100
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 0.3356,
      "step": 1200
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.9e-05,
      "loss": 0.331,
      "step": 1300
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2e-05,
      "loss": 0.3284,
      "step": 1400
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 0.3212,
      "step": 1500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 0.3194,
      "step": 1600
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.1e-05,
      "loss": 0.3167,
      "step": 1700
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.399999999999999e-05,
      "loss": 0.3167,
      "step": 1800
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.311,
      "step": 1900
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.3116,
      "step": 2000
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.299999999999999e-05,
      "loss": 0.3051,
      "step": 2100
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.599999999999999e-05,
      "loss": 0.3057,
      "step": 2200
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.9e-05,
      "loss": 0.304,
      "step": 2300
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.199999999999999e-05,
      "loss": 0.3119,
      "step": 2400
    },
    {
      "epoch": 0.78,
      "learning_rate": 7.5e-05,
      "loss": 0.3044,
      "step": 2500
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.8e-05,
      "loss": 0.3095,
      "step": 2600
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.1e-05,
      "loss": 0.306,
      "step": 2700
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.4e-05,
      "loss": 0.3081,
      "step": 2800
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.699999999999999e-05,
      "loss": 0.3062,
      "step": 2900
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.999999999999999e-05,
      "loss": 0.3022,
      "step": 3000
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.3e-05,
      "loss": 0.3037,
      "step": 3100
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.599999999999999e-05,
      "loss": 0.306,
      "step": 3200
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.9e-05,
      "loss": 0.2986,
      "step": 3300
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.000102,
      "loss": 0.2994,
      "step": 3400
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00010499999999999999,
      "loss": 0.2966,
      "step": 3500
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00010799999999999998,
      "loss": 0.299,
      "step": 3600
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00011099999999999999,
      "loss": 0.2994,
      "step": 3700
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00011399999999999999,
      "loss": 0.3021,
      "step": 3800
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.000117,
      "loss": 0.3037,
      "step": 3900
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.3035,
      "step": 4000
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012299999999999998,
      "loss": 0.307,
      "step": 4100
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012599999999999997,
      "loss": 0.3035,
      "step": 4200
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.000129,
      "loss": 0.3038,
      "step": 4300
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00013199999999999998,
      "loss": 0.3063,
      "step": 4400
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.000135,
      "loss": 0.3072,
      "step": 4500
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.000138,
      "loss": 0.3093,
      "step": 4600
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00014099999999999998,
      "loss": 0.3142,
      "step": 4700
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00014399999999999998,
      "loss": 0.3097,
      "step": 4800
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.000147,
      "loss": 0.309,
      "step": 4900
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00015,
      "loss": 0.3116,
      "step": 5000
    },
    {
      "epoch": 1.57,
      "eval_loss": 0.5301366448402405,
      "eval_runtime": 103.8143,
      "eval_samples_per_second": 86.125,
      "eval_steps_per_second": 1.349,
      "step": 5000
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0001497450713800136,
      "loss": 0.3111,
      "step": 5100
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00014949014276002717,
      "loss": 0.3085,
      "step": 5200
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00014923521414004077,
      "loss": 0.306,
      "step": 5300
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00014898028552005438,
      "loss": 0.3046,
      "step": 5400
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00014872535690006795,
      "loss": 0.3016,
      "step": 5500
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00014847042828008156,
      "loss": 0.2954,
      "step": 5600
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00014821549966009516,
      "loss": 0.2919,
      "step": 5700
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00014796057104010877,
      "loss": 0.293,
      "step": 5800
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00014770564242012234,
      "loss": 0.2893,
      "step": 5900
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00014745071380013595,
      "loss": 0.2824,
      "step": 6000
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00014719578518014955,
      "loss": 0.2868,
      "step": 6100
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00014694085656016313,
      "loss": 0.2816,
      "step": 6200
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00014668592794017673,
      "loss": 0.2765,
      "step": 6300
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00014643099932019034,
      "loss": 0.2782,
      "step": 6400
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00014617607070020392,
      "loss": 0.2669,
      "step": 6500
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00014592114208021755,
      "loss": 0.2622,
      "step": 6600
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00014566621346023112,
      "loss": 0.2651,
      "step": 6700
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00014541128484024473,
      "loss": 0.2624,
      "step": 6800
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001451563562202583,
      "loss": 0.2581,
      "step": 6900
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001449014276002719,
      "loss": 0.2604,
      "step": 7000
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00014464649898028551,
      "loss": 0.2531,
      "step": 7100
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0001443915703602991,
      "loss": 0.2584,
      "step": 7200
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0001441366417403127,
      "loss": 0.2528,
      "step": 7300
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0001438817131203263,
      "loss": 0.2522,
      "step": 7400
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00014362678450033988,
      "loss": 0.248,
      "step": 7500
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00014337185588035348,
      "loss": 0.2503,
      "step": 7600
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00014311692726036708,
      "loss": 0.2491,
      "step": 7700
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001428619986403807,
      "loss": 0.2468,
      "step": 7800
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0001426070700203943,
      "loss": 0.2429,
      "step": 7900
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00014235214140040787,
      "loss": 0.2443,
      "step": 8000
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00014209721278042147,
      "loss": 0.2432,
      "step": 8100
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00014184228416043508,
      "loss": 0.2395,
      "step": 8200
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00014158735554044866,
      "loss": 0.2356,
      "step": 8300
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00014133242692046226,
      "loss": 0.2394,
      "step": 8400
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00014107749830047584,
      "loss": 0.2354,
      "step": 8500
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00014082256968048944,
      "loss": 0.2369,
      "step": 8600
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00014056764106050305,
      "loss": 0.2323,
      "step": 8700
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00014031271244051665,
      "loss": 0.233,
      "step": 8800
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00014005778382053025,
      "loss": 0.2324,
      "step": 8900
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00013980285520054383,
      "loss": 0.2335,
      "step": 9000
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00013954792658055744,
      "loss": 0.2311,
      "step": 9100
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00013929299796057104,
      "loss": 0.2302,
      "step": 9200
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00013903806934058462,
      "loss": 0.2264,
      "step": 9300
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00013878314072059822,
      "loss": 0.226,
      "step": 9400
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00013852821210061183,
      "loss": 0.2214,
      "step": 9500
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0001382732834806254,
      "loss": 0.2218,
      "step": 9600
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.000138018354860639,
      "loss": 0.2141,
      "step": 9700
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001377634262406526,
      "loss": 0.2132,
      "step": 9800
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.00013750849762066622,
      "loss": 0.2131,
      "step": 9900
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0001372535690006798,
      "loss": 0.2104,
      "step": 10000
    },
    {
      "epoch": 3.13,
      "eval_loss": 0.40657177567481995,
      "eval_runtime": 67.9841,
      "eval_samples_per_second": 131.516,
      "eval_steps_per_second": 2.059,
      "step": 10000
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.0001369986403806934,
      "loss": 0.2127,
      "step": 10100
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.000136743711760707,
      "loss": 0.2071,
      "step": 10200
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.00013648878314072058,
      "loss": 0.2125,
      "step": 10300
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.00013623385452073418,
      "loss": 0.2089,
      "step": 10400
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0001359789259007478,
      "loss": 0.209,
      "step": 10500
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00013572399728076136,
      "loss": 0.2109,
      "step": 10600
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00013546906866077497,
      "loss": 0.2105,
      "step": 10700
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.00013521414004078857,
      "loss": 0.2092,
      "step": 10800
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.00013495921142080218,
      "loss": 0.2093,
      "step": 10900
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00013470428280081575,
      "loss": 0.206,
      "step": 11000
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00013444935418082936,
      "loss": 0.2032,
      "step": 11100
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00013419442556084296,
      "loss": 0.204,
      "step": 11200
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.00013393949694085654,
      "loss": 0.2005,
      "step": 11300
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.00013368456832087014,
      "loss": 0.1988,
      "step": 11400
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.00013342963970088375,
      "loss": 0.1981,
      "step": 11500
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.00013317471108089733,
      "loss": 0.2017,
      "step": 11600
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.00013291978246091093,
      "loss": 0.1989,
      "step": 11700
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.00013266485384092453,
      "loss": 0.2011,
      "step": 11800
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0001324099252209381,
      "loss": 0.197,
      "step": 11900
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.00013215499660095172,
      "loss": 0.1975,
      "step": 12000
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.00013190006798096532,
      "loss": 0.1982,
      "step": 12100
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00013164513936097892,
      "loss": 0.1983,
      "step": 12200
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.0001313902107409925,
      "loss": 0.1992,
      "step": 12300
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.0001311352821210061,
      "loss": 0.1943,
      "step": 12400
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.0001308803535010197,
      "loss": 0.1958,
      "step": 12500
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.0001306254248810333,
      "loss": 0.1927,
      "step": 12600
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.0001303704962610469,
      "loss": 0.1927,
      "step": 12700
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.0001301155676410605,
      "loss": 0.1882,
      "step": 12800
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.00012986063902107407,
      "loss": 0.1854,
      "step": 12900
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0001296057104010877,
      "loss": 0.1842,
      "step": 13000
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.00012935078178110128,
      "loss": 0.1846,
      "step": 13100
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00012909585316111489,
      "loss": 0.1834,
      "step": 13200
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.0001288409245411285,
      "loss": 0.1833,
      "step": 13300
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.00012858599592114207,
      "loss": 0.1826,
      "step": 13400
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00012833106730115567,
      "loss": 0.1807,
      "step": 13500
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.00012807613868116925,
      "loss": 0.1805,
      "step": 13600
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00012782121006118285,
      "loss": 0.1808,
      "step": 13700
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00012756628144119646,
      "loss": 0.1791,
      "step": 13800
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.00012731135282121003,
      "loss": 0.181,
      "step": 13900
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.00012705642420122364,
      "loss": 0.1816,
      "step": 14000
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00012680149558123724,
      "loss": 0.1805,
      "step": 14100
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.00012654656696125085,
      "loss": 0.1784,
      "step": 14200
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.00012629163834126445,
      "loss": 0.1767,
      "step": 14300
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.00012603670972127803,
      "loss": 0.18,
      "step": 14400
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00012578178110129163,
      "loss": 0.1773,
      "step": 14500
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00012552685248130524,
      "loss": 0.1794,
      "step": 14600
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.00012527192386131881,
      "loss": 0.1776,
      "step": 14700
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.00012501699524133242,
      "loss": 0.1749,
      "step": 14800
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.000124762066621346,
      "loss": 0.1746,
      "step": 14900
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.0001245071380013596,
      "loss": 0.1729,
      "step": 15000
    },
    {
      "epoch": 4.7,
      "eval_loss": 0.3554973900318146,
      "eval_runtime": 74.4919,
      "eval_samples_per_second": 120.026,
      "eval_steps_per_second": 1.879,
      "step": 15000
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.0001242522093813732,
      "loss": 0.1768,
      "step": 15100
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.0001239972807613868,
      "loss": 0.1727,
      "step": 15200
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.0001237423521414004,
      "loss": 0.1742,
      "step": 15300
    },
    {
      "epoch": 4.83,
      "learning_rate": 0.000123487423521414,
      "loss": 0.1697,
      "step": 15400
    },
    {
      "epoch": 4.86,
      "learning_rate": 0.0001232324949014276,
      "loss": 0.1755,
      "step": 15500
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.0001229775662814412,
      "loss": 0.1709,
      "step": 15600
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.00012272263766145478,
      "loss": 0.1735,
      "step": 15700
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.00012246770904146838,
      "loss": 0.1692,
      "step": 15800
    },
    {
      "epoch": 4.98,
      "learning_rate": 0.00012221278042148198,
      "loss": 0.1706,
      "step": 15900
    },
    {
      "epoch": 5.01,
      "learning_rate": 0.00012195785180149556,
      "loss": 0.1665,
      "step": 16000
    },
    {
      "epoch": 5.04,
      "learning_rate": 0.00012170292318150917,
      "loss": 0.1601,
      "step": 16100
    },
    {
      "epoch": 5.08,
      "learning_rate": 0.00012144799456152277,
      "loss": 0.1619,
      "step": 16200
    },
    {
      "epoch": 5.11,
      "learning_rate": 0.00012119306594153636,
      "loss": 0.1615,
      "step": 16300
    },
    {
      "epoch": 5.14,
      "learning_rate": 0.00012093813732154996,
      "loss": 0.1613,
      "step": 16400
    },
    {
      "epoch": 5.17,
      "learning_rate": 0.00012068320870156356,
      "loss": 0.1603,
      "step": 16500
    },
    {
      "epoch": 5.2,
      "learning_rate": 0.00012042828008157715,
      "loss": 0.1611,
      "step": 16600
    },
    {
      "epoch": 5.23,
      "learning_rate": 0.00012017335146159075,
      "loss": 0.16,
      "step": 16700
    },
    {
      "epoch": 5.26,
      "learning_rate": 0.00011991842284160434,
      "loss": 0.1616,
      "step": 16800
    },
    {
      "epoch": 5.3,
      "learning_rate": 0.00011966349422161793,
      "loss": 0.1628,
      "step": 16900
    },
    {
      "epoch": 5.33,
      "learning_rate": 0.00011940856560163152,
      "loss": 0.1652,
      "step": 17000
    },
    {
      "epoch": 5.36,
      "learning_rate": 0.00011915363698164513,
      "loss": 0.1593,
      "step": 17100
    },
    {
      "epoch": 5.39,
      "learning_rate": 0.00011889870836165872,
      "loss": 0.1589,
      "step": 17200
    },
    {
      "epoch": 5.42,
      "learning_rate": 0.00011864377974167233,
      "loss": 0.1593,
      "step": 17300
    },
    {
      "epoch": 5.45,
      "learning_rate": 0.00011838885112168593,
      "loss": 0.159,
      "step": 17400
    },
    {
      "epoch": 5.48,
      "learning_rate": 0.00011813392250169952,
      "loss": 0.1599,
      "step": 17500
    },
    {
      "epoch": 5.51,
      "learning_rate": 0.00011787899388171311,
      "loss": 0.1603,
      "step": 17600
    },
    {
      "epoch": 5.55,
      "learning_rate": 0.00011762406526172671,
      "loss": 0.1549,
      "step": 17700
    },
    {
      "epoch": 5.58,
      "learning_rate": 0.0001173691366417403,
      "loss": 0.1574,
      "step": 17800
    },
    {
      "epoch": 5.61,
      "learning_rate": 0.00011711420802175389,
      "loss": 0.1575,
      "step": 17900
    },
    {
      "epoch": 5.64,
      "learning_rate": 0.0001168592794017675,
      "loss": 0.1583,
      "step": 18000
    },
    {
      "epoch": 5.67,
      "learning_rate": 0.00011660435078178109,
      "loss": 0.1542,
      "step": 18100
    },
    {
      "epoch": 5.7,
      "learning_rate": 0.00011634942216179468,
      "loss": 0.1559,
      "step": 18200
    },
    {
      "epoch": 5.73,
      "learning_rate": 0.00011609449354180827,
      "loss": 0.157,
      "step": 18300
    },
    {
      "epoch": 5.77,
      "learning_rate": 0.00011583956492182189,
      "loss": 0.1568,
      "step": 18400
    },
    {
      "epoch": 5.8,
      "learning_rate": 0.00011558463630183548,
      "loss": 0.155,
      "step": 18500
    },
    {
      "epoch": 5.83,
      "learning_rate": 0.00011532970768184908,
      "loss": 0.1571,
      "step": 18600
    },
    {
      "epoch": 5.86,
      "learning_rate": 0.00011507477906186267,
      "loss": 0.1526,
      "step": 18700
    },
    {
      "epoch": 5.89,
      "learning_rate": 0.00011481985044187626,
      "loss": 0.1513,
      "step": 18800
    },
    {
      "epoch": 5.92,
      "learning_rate": 0.00011456492182188985,
      "loss": 0.1556,
      "step": 18900
    },
    {
      "epoch": 5.95,
      "learning_rate": 0.00011430999320190346,
      "loss": 0.1542,
      "step": 19000
    },
    {
      "epoch": 5.98,
      "learning_rate": 0.00011405506458191705,
      "loss": 0.1519,
      "step": 19100
    },
    {
      "epoch": 6.02,
      "learning_rate": 0.00011380013596193064,
      "loss": 0.1488,
      "step": 19200
    },
    {
      "epoch": 6.05,
      "learning_rate": 0.00011354520734194424,
      "loss": 0.1453,
      "step": 19300
    },
    {
      "epoch": 6.08,
      "learning_rate": 0.00011329027872195785,
      "loss": 0.1458,
      "step": 19400
    },
    {
      "epoch": 6.11,
      "learning_rate": 0.00011303535010197144,
      "loss": 0.1431,
      "step": 19500
    },
    {
      "epoch": 6.14,
      "learning_rate": 0.00011278042148198504,
      "loss": 0.1453,
      "step": 19600
    },
    {
      "epoch": 6.17,
      "learning_rate": 0.00011252549286199863,
      "loss": 0.1416,
      "step": 19700
    },
    {
      "epoch": 6.2,
      "learning_rate": 0.00011227056424201222,
      "loss": 0.1465,
      "step": 19800
    },
    {
      "epoch": 6.24,
      "learning_rate": 0.00011201563562202583,
      "loss": 0.1451,
      "step": 19900
    },
    {
      "epoch": 6.27,
      "learning_rate": 0.00011176070700203942,
      "loss": 0.1472,
      "step": 20000
    },
    {
      "epoch": 6.27,
      "eval_loss": 0.3207816481590271,
      "eval_runtime": 71.6607,
      "eval_samples_per_second": 124.769,
      "eval_steps_per_second": 1.954,
      "step": 20000
    },
    {
      "epoch": 6.3,
      "learning_rate": 0.00011150577838205301,
      "loss": 0.143,
      "step": 20100
    },
    {
      "epoch": 6.33,
      "learning_rate": 0.0001112508497620666,
      "loss": 0.1418,
      "step": 20200
    },
    {
      "epoch": 6.36,
      "learning_rate": 0.0001109959211420802,
      "loss": 0.1461,
      "step": 20300
    },
    {
      "epoch": 6.39,
      "learning_rate": 0.0001107409925220938,
      "loss": 0.1414,
      "step": 20400
    },
    {
      "epoch": 6.42,
      "learning_rate": 0.00011048606390210741,
      "loss": 0.1418,
      "step": 20500
    },
    {
      "epoch": 6.45,
      "learning_rate": 0.000110231135282121,
      "loss": 0.144,
      "step": 20600
    },
    {
      "epoch": 6.49,
      "learning_rate": 0.0001099762066621346,
      "loss": 0.1427,
      "step": 20700
    },
    {
      "epoch": 6.52,
      "learning_rate": 0.00010972127804214819,
      "loss": 0.1434,
      "step": 20800
    },
    {
      "epoch": 6.55,
      "learning_rate": 0.00010946634942216179,
      "loss": 0.1437,
      "step": 20900
    },
    {
      "epoch": 6.58,
      "learning_rate": 0.00010921142080217538,
      "loss": 0.1426,
      "step": 21000
    },
    {
      "epoch": 6.61,
      "learning_rate": 0.00010895649218218897,
      "loss": 0.1403,
      "step": 21100
    },
    {
      "epoch": 6.64,
      "learning_rate": 0.00010870156356220258,
      "loss": 0.1399,
      "step": 21200
    },
    {
      "epoch": 6.67,
      "learning_rate": 0.00010844663494221617,
      "loss": 0.1454,
      "step": 21300
    },
    {
      "epoch": 6.71,
      "learning_rate": 0.00010819170632222976,
      "loss": 0.1402,
      "step": 21400
    },
    {
      "epoch": 6.74,
      "learning_rate": 0.00010793677770224335,
      "loss": 0.1403,
      "step": 21500
    },
    {
      "epoch": 6.77,
      "learning_rate": 0.00010768184908225697,
      "loss": 0.1401,
      "step": 21600
    },
    {
      "epoch": 6.8,
      "learning_rate": 0.00010742692046227056,
      "loss": 0.1408,
      "step": 21700
    },
    {
      "epoch": 6.83,
      "learning_rate": 0.00010717199184228416,
      "loss": 0.1388,
      "step": 21800
    },
    {
      "epoch": 6.86,
      "learning_rate": 0.00010691706322229775,
      "loss": 0.1431,
      "step": 21900
    },
    {
      "epoch": 6.89,
      "learning_rate": 0.00010666213460231134,
      "loss": 0.1384,
      "step": 22000
    },
    {
      "epoch": 6.92,
      "learning_rate": 0.00010640720598232493,
      "loss": 0.1381,
      "step": 22100
    },
    {
      "epoch": 6.96,
      "learning_rate": 0.00010615227736233854,
      "loss": 0.1367,
      "step": 22200
    },
    {
      "epoch": 6.99,
      "learning_rate": 0.00010589734874235213,
      "loss": 0.1424,
      "step": 22300
    },
    {
      "epoch": 7.02,
      "learning_rate": 0.00010564242012236572,
      "loss": 0.1347,
      "step": 22400
    },
    {
      "epoch": 7.05,
      "learning_rate": 0.00010538749150237932,
      "loss": 0.1297,
      "step": 22500
    },
    {
      "epoch": 7.08,
      "learning_rate": 0.00010513256288239293,
      "loss": 0.1323,
      "step": 22600
    },
    {
      "epoch": 7.11,
      "learning_rate": 0.00010487763426240652,
      "loss": 0.1292,
      "step": 22700
    },
    {
      "epoch": 7.14,
      "learning_rate": 0.00010462270564242012,
      "loss": 0.1311,
      "step": 22800
    },
    {
      "epoch": 7.18,
      "learning_rate": 0.00010436777702243371,
      "loss": 0.1328,
      "step": 22900
    },
    {
      "epoch": 7.21,
      "learning_rate": 0.0001041128484024473,
      "loss": 0.1277,
      "step": 23000
    },
    {
      "epoch": 7.24,
      "learning_rate": 0.00010385791978246091,
      "loss": 0.129,
      "step": 23100
    },
    {
      "epoch": 7.27,
      "learning_rate": 0.0001036029911624745,
      "loss": 0.1304,
      "step": 23200
    },
    {
      "epoch": 7.3,
      "learning_rate": 0.00010334806254248809,
      "loss": 0.1308,
      "step": 23300
    },
    {
      "epoch": 7.33,
      "learning_rate": 0.00010309313392250168,
      "loss": 0.1285,
      "step": 23400
    },
    {
      "epoch": 7.36,
      "learning_rate": 0.00010283820530251528,
      "loss": 0.1289,
      "step": 23500
    },
    {
      "epoch": 7.39,
      "learning_rate": 0.00010258327668252887,
      "loss": 0.1318,
      "step": 23600
    },
    {
      "epoch": 7.43,
      "learning_rate": 0.00010232834806254249,
      "loss": 0.1291,
      "step": 23700
    },
    {
      "epoch": 7.46,
      "learning_rate": 0.00010207341944255608,
      "loss": 0.1298,
      "step": 23800
    },
    {
      "epoch": 7.49,
      "learning_rate": 0.00010181849082256967,
      "loss": 0.1287,
      "step": 23900
    },
    {
      "epoch": 7.52,
      "learning_rate": 0.00010156356220258326,
      "loss": 0.1293,
      "step": 24000
    },
    {
      "epoch": 7.55,
      "learning_rate": 0.00010130863358259687,
      "loss": 0.1277,
      "step": 24100
    },
    {
      "epoch": 7.58,
      "learning_rate": 0.00010105370496261046,
      "loss": 0.1295,
      "step": 24200
    },
    {
      "epoch": 7.61,
      "learning_rate": 0.00010079877634262405,
      "loss": 0.1289,
      "step": 24300
    },
    {
      "epoch": 7.65,
      "learning_rate": 0.00010054384772263765,
      "loss": 0.1298,
      "step": 24400
    },
    {
      "epoch": 7.68,
      "learning_rate": 0.00010028891910265125,
      "loss": 0.1287,
      "step": 24500
    },
    {
      "epoch": 7.71,
      "learning_rate": 0.00010003399048266484,
      "loss": 0.1286,
      "step": 24600
    },
    {
      "epoch": 7.74,
      "learning_rate": 9.977906186267843e-05,
      "loss": 0.1294,
      "step": 24700
    },
    {
      "epoch": 7.77,
      "learning_rate": 9.952413324269204e-05,
      "loss": 0.1287,
      "step": 24800
    },
    {
      "epoch": 7.8,
      "learning_rate": 9.926920462270564e-05,
      "loss": 0.1295,
      "step": 24900
    },
    {
      "epoch": 7.83,
      "learning_rate": 9.901427600271924e-05,
      "loss": 0.128,
      "step": 25000
    },
    {
      "epoch": 7.83,
      "eval_loss": 0.29230204224586487,
      "eval_runtime": 77.9206,
      "eval_samples_per_second": 114.745,
      "eval_steps_per_second": 1.797,
      "step": 25000
    },
    {
      "epoch": 7.86,
      "learning_rate": 9.875934738273283e-05,
      "loss": 0.1305,
      "step": 25100
    },
    {
      "epoch": 7.9,
      "learning_rate": 9.850441876274642e-05,
      "loss": 0.1284,
      "step": 25200
    },
    {
      "epoch": 7.93,
      "learning_rate": 9.824949014276001e-05,
      "loss": 0.1275,
      "step": 25300
    },
    {
      "epoch": 7.96,
      "learning_rate": 9.799456152277362e-05,
      "loss": 0.1277,
      "step": 25400
    },
    {
      "epoch": 7.99,
      "learning_rate": 9.77396329027872e-05,
      "loss": 0.1252,
      "step": 25500
    },
    {
      "epoch": 8.02,
      "learning_rate": 9.74847042828008e-05,
      "loss": 0.1195,
      "step": 25600
    },
    {
      "epoch": 8.05,
      "learning_rate": 9.72297756628144e-05,
      "loss": 0.1195,
      "step": 25700
    },
    {
      "epoch": 8.08,
      "learning_rate": 9.6974847042828e-05,
      "loss": 0.119,
      "step": 25800
    },
    {
      "epoch": 8.12,
      "learning_rate": 9.67199184228416e-05,
      "loss": 0.1205,
      "step": 25900
    },
    {
      "epoch": 8.15,
      "learning_rate": 9.64649898028552e-05,
      "loss": 0.1173,
      "step": 26000
    },
    {
      "epoch": 8.18,
      "learning_rate": 9.621006118286879e-05,
      "loss": 0.1175,
      "step": 26100
    },
    {
      "epoch": 8.21,
      "learning_rate": 9.595513256288238e-05,
      "loss": 0.1188,
      "step": 26200
    },
    {
      "epoch": 8.24,
      "learning_rate": 9.570020394289599e-05,
      "loss": 0.1177,
      "step": 26300
    },
    {
      "epoch": 8.27,
      "learning_rate": 9.544527532290958e-05,
      "loss": 0.1185,
      "step": 26400
    },
    {
      "epoch": 8.3,
      "learning_rate": 9.519034670292317e-05,
      "loss": 0.1199,
      "step": 26500
    },
    {
      "epoch": 8.33,
      "learning_rate": 9.493541808293676e-05,
      "loss": 0.1206,
      "step": 26600
    },
    {
      "epoch": 8.37,
      "learning_rate": 9.468048946295036e-05,
      "loss": 0.1173,
      "step": 26700
    },
    {
      "epoch": 8.4,
      "learning_rate": 9.442556084296395e-05,
      "loss": 0.1172,
      "step": 26800
    },
    {
      "epoch": 8.43,
      "learning_rate": 9.417063222297757e-05,
      "loss": 0.1173,
      "step": 26900
    },
    {
      "epoch": 8.46,
      "learning_rate": 9.391570360299116e-05,
      "loss": 0.121,
      "step": 27000
    },
    {
      "epoch": 8.49,
      "learning_rate": 9.366077498300475e-05,
      "loss": 0.1173,
      "step": 27100
    },
    {
      "epoch": 8.52,
      "learning_rate": 9.340584636301834e-05,
      "loss": 0.1184,
      "step": 27200
    },
    {
      "epoch": 8.55,
      "learning_rate": 9.315091774303195e-05,
      "loss": 0.1172,
      "step": 27300
    },
    {
      "epoch": 8.59,
      "learning_rate": 9.289598912304554e-05,
      "loss": 0.118,
      "step": 27400
    },
    {
      "epoch": 8.62,
      "learning_rate": 9.264106050305913e-05,
      "loss": 0.1179,
      "step": 27500
    },
    {
      "epoch": 8.65,
      "learning_rate": 9.238613188307273e-05,
      "loss": 0.115,
      "step": 27600
    },
    {
      "epoch": 8.68,
      "learning_rate": 9.213120326308632e-05,
      "loss": 0.1159,
      "step": 27700
    },
    {
      "epoch": 8.71,
      "learning_rate": 9.187627464309991e-05,
      "loss": 0.1152,
      "step": 27800
    },
    {
      "epoch": 8.74,
      "learning_rate": 9.162134602311352e-05,
      "loss": 0.1166,
      "step": 27900
    },
    {
      "epoch": 8.77,
      "learning_rate": 9.136641740312712e-05,
      "loss": 0.1184,
      "step": 28000
    },
    {
      "epoch": 8.8,
      "learning_rate": 9.111148878314071e-05,
      "loss": 0.1155,
      "step": 28100
    },
    {
      "epoch": 8.84,
      "learning_rate": 9.085656016315432e-05,
      "loss": 0.1136,
      "step": 28200
    },
    {
      "epoch": 8.87,
      "learning_rate": 9.060163154316791e-05,
      "loss": 0.117,
      "step": 28300
    },
    {
      "epoch": 8.9,
      "learning_rate": 9.03467029231815e-05,
      "loss": 0.1142,
      "step": 28400
    },
    {
      "epoch": 8.93,
      "learning_rate": 9.009177430319509e-05,
      "loss": 0.1168,
      "step": 28500
    },
    {
      "epoch": 8.96,
      "learning_rate": 8.98368456832087e-05,
      "loss": 0.115,
      "step": 28600
    },
    {
      "epoch": 8.99,
      "learning_rate": 8.958191706322229e-05,
      "loss": 0.1158,
      "step": 28700
    },
    {
      "epoch": 9.02,
      "learning_rate": 8.932698844323588e-05,
      "loss": 0.1103,
      "step": 28800
    },
    {
      "epoch": 9.06,
      "learning_rate": 8.907205982324948e-05,
      "loss": 0.1067,
      "step": 28900
    },
    {
      "epoch": 9.09,
      "learning_rate": 8.881713120326308e-05,
      "loss": 0.1091,
      "step": 29000
    },
    {
      "epoch": 9.12,
      "learning_rate": 8.856220258327668e-05,
      "loss": 0.108,
      "step": 29100
    },
    {
      "epoch": 9.15,
      "learning_rate": 8.830727396329028e-05,
      "loss": 0.1082,
      "step": 29200
    },
    {
      "epoch": 9.18,
      "learning_rate": 8.805234534330387e-05,
      "loss": 0.108,
      "step": 29300
    },
    {
      "epoch": 9.21,
      "learning_rate": 8.779741672331746e-05,
      "loss": 0.1069,
      "step": 29400
    },
    {
      "epoch": 9.24,
      "learning_rate": 8.754248810333107e-05,
      "loss": 0.1111,
      "step": 29500
    },
    {
      "epoch": 9.27,
      "learning_rate": 8.728755948334466e-05,
      "loss": 0.11,
      "step": 29600
    },
    {
      "epoch": 9.31,
      "learning_rate": 8.703263086335825e-05,
      "loss": 0.1085,
      "step": 29700
    },
    {
      "epoch": 9.34,
      "learning_rate": 8.677770224337184e-05,
      "loss": 0.1094,
      "step": 29800
    },
    {
      "epoch": 9.37,
      "learning_rate": 8.652277362338544e-05,
      "loss": 0.1111,
      "step": 29900
    },
    {
      "epoch": 9.4,
      "learning_rate": 8.626784500339903e-05,
      "loss": 0.1065,
      "step": 30000
    },
    {
      "epoch": 9.4,
      "eval_loss": 0.2719452977180481,
      "eval_runtime": 70.8391,
      "eval_samples_per_second": 126.216,
      "eval_steps_per_second": 1.976,
      "step": 30000
    },
    {
      "epoch": 9.43,
      "learning_rate": 8.601291638341265e-05,
      "loss": 0.1072,
      "step": 30100
    },
    {
      "epoch": 9.46,
      "learning_rate": 8.575798776342624e-05,
      "loss": 0.1075,
      "step": 30200
    },
    {
      "epoch": 9.49,
      "learning_rate": 8.550305914343983e-05,
      "loss": 0.1076,
      "step": 30300
    },
    {
      "epoch": 9.53,
      "learning_rate": 8.524813052345342e-05,
      "loss": 0.1056,
      "step": 30400
    },
    {
      "epoch": 9.56,
      "learning_rate": 8.499320190346703e-05,
      "loss": 0.1079,
      "step": 30500
    },
    {
      "epoch": 9.59,
      "learning_rate": 8.473827328348062e-05,
      "loss": 0.106,
      "step": 30600
    },
    {
      "epoch": 9.62,
      "learning_rate": 8.448334466349421e-05,
      "loss": 0.1058,
      "step": 30700
    },
    {
      "epoch": 9.65,
      "learning_rate": 8.422841604350781e-05,
      "loss": 0.1063,
      "step": 30800
    },
    {
      "epoch": 9.68,
      "learning_rate": 8.39734874235214e-05,
      "loss": 0.1102,
      "step": 30900
    },
    {
      "epoch": 9.71,
      "learning_rate": 8.3718558803535e-05,
      "loss": 0.1085,
      "step": 31000
    },
    {
      "epoch": 9.74,
      "learning_rate": 8.34636301835486e-05,
      "loss": 0.1049,
      "step": 31100
    },
    {
      "epoch": 9.78,
      "learning_rate": 8.32087015635622e-05,
      "loss": 0.1047,
      "step": 31200
    },
    {
      "epoch": 9.81,
      "learning_rate": 8.295377294357579e-05,
      "loss": 0.1054,
      "step": 31300
    },
    {
      "epoch": 9.84,
      "learning_rate": 8.26988443235894e-05,
      "loss": 0.1058,
      "step": 31400
    },
    {
      "epoch": 9.87,
      "learning_rate": 8.244391570360299e-05,
      "loss": 0.1065,
      "step": 31500
    },
    {
      "epoch": 9.9,
      "learning_rate": 8.218898708361658e-05,
      "loss": 0.1053,
      "step": 31600
    },
    {
      "epoch": 9.93,
      "learning_rate": 8.193405846363017e-05,
      "loss": 0.1055,
      "step": 31700
    },
    {
      "epoch": 9.96,
      "learning_rate": 8.167912984364377e-05,
      "loss": 0.1048,
      "step": 31800
    },
    {
      "epoch": 10.0,
      "learning_rate": 8.142420122365736e-05,
      "loss": 0.1041,
      "step": 31900
    },
    {
      "epoch": 10.03,
      "learning_rate": 8.116927260367095e-05,
      "loss": 0.0979,
      "step": 32000
    },
    {
      "epoch": 10.06,
      "learning_rate": 8.091434398368456e-05,
      "loss": 0.0987,
      "step": 32100
    },
    {
      "epoch": 10.09,
      "learning_rate": 8.065941536369816e-05,
      "loss": 0.1002,
      "step": 32200
    },
    {
      "epoch": 10.12,
      "learning_rate": 8.040448674371175e-05,
      "loss": 0.0962,
      "step": 32300
    },
    {
      "epoch": 10.15,
      "learning_rate": 8.014955812372536e-05,
      "loss": 0.0982,
      "step": 32400
    },
    {
      "epoch": 10.18,
      "learning_rate": 7.989462950373895e-05,
      "loss": 0.0999,
      "step": 32500
    },
    {
      "epoch": 10.21,
      "learning_rate": 7.963970088375254e-05,
      "loss": 0.0971,
      "step": 32600
    },
    {
      "epoch": 10.25,
      "learning_rate": 7.938477226376614e-05,
      "loss": 0.0986,
      "step": 32700
    },
    {
      "epoch": 10.28,
      "learning_rate": 7.912984364377973e-05,
      "loss": 0.0977,
      "step": 32800
    },
    {
      "epoch": 10.31,
      "learning_rate": 7.887491502379333e-05,
      "loss": 0.0965,
      "step": 32900
    },
    {
      "epoch": 10.34,
      "learning_rate": 7.861998640380692e-05,
      "loss": 0.1012,
      "step": 33000
    },
    {
      "epoch": 10.37,
      "learning_rate": 7.836505778382052e-05,
      "loss": 0.0973,
      "step": 33100
    },
    {
      "epoch": 10.4,
      "learning_rate": 7.811012916383411e-05,
      "loss": 0.0969,
      "step": 33200
    },
    {
      "epoch": 10.43,
      "learning_rate": 7.785520054384773e-05,
      "loss": 0.0983,
      "step": 33300
    },
    {
      "epoch": 10.47,
      "learning_rate": 7.760027192386132e-05,
      "loss": 0.0986,
      "step": 33400
    },
    {
      "epoch": 10.5,
      "learning_rate": 7.734534330387491e-05,
      "loss": 0.1001,
      "step": 33500
    },
    {
      "epoch": 10.53,
      "learning_rate": 7.70904146838885e-05,
      "loss": 0.0981,
      "step": 33600
    },
    {
      "epoch": 10.56,
      "learning_rate": 7.68354860639021e-05,
      "loss": 0.098,
      "step": 33700
    },
    {
      "epoch": 10.59,
      "learning_rate": 7.65805574439157e-05,
      "loss": 0.0963,
      "step": 33800
    },
    {
      "epoch": 10.62,
      "learning_rate": 7.632562882392929e-05,
      "loss": 0.0971,
      "step": 33900
    },
    {
      "epoch": 10.65,
      "learning_rate": 7.607070020394289e-05,
      "loss": 0.0994,
      "step": 34000
    },
    {
      "epoch": 10.68,
      "learning_rate": 7.581577158395648e-05,
      "loss": 0.096,
      "step": 34100
    },
    {
      "epoch": 10.72,
      "learning_rate": 7.556084296397007e-05,
      "loss": 0.0992,
      "step": 34200
    },
    {
      "epoch": 10.75,
      "learning_rate": 7.530591434398368e-05,
      "loss": 0.0968,
      "step": 34300
    },
    {
      "epoch": 10.78,
      "learning_rate": 7.505098572399728e-05,
      "loss": 0.0974,
      "step": 34400
    },
    {
      "epoch": 10.81,
      "learning_rate": 7.479605710401087e-05,
      "loss": 0.0961,
      "step": 34500
    },
    {
      "epoch": 10.84,
      "learning_rate": 7.454112848402448e-05,
      "loss": 0.0973,
      "step": 34600
    },
    {
      "epoch": 10.87,
      "learning_rate": 7.428619986403807e-05,
      "loss": 0.0968,
      "step": 34700
    },
    {
      "epoch": 10.9,
      "learning_rate": 7.403127124405166e-05,
      "loss": 0.0974,
      "step": 34800
    },
    {
      "epoch": 10.94,
      "learning_rate": 7.377634262406525e-05,
      "loss": 0.0959,
      "step": 34900
    },
    {
      "epoch": 10.97,
      "learning_rate": 7.352141400407885e-05,
      "loss": 0.0995,
      "step": 35000
    },
    {
      "epoch": 10.97,
      "eval_loss": 0.2515884339809418,
      "eval_runtime": 88.1892,
      "eval_samples_per_second": 101.384,
      "eval_steps_per_second": 1.587,
      "step": 35000
    },
    {
      "epoch": 11.0,
      "learning_rate": 7.326648538409244e-05,
      "loss": 0.0975,
      "step": 35100
    },
    {
      "epoch": 11.03,
      "learning_rate": 7.301155676410605e-05,
      "loss": 0.0904,
      "step": 35200
    },
    {
      "epoch": 11.06,
      "learning_rate": 7.275662814411964e-05,
      "loss": 0.0888,
      "step": 35300
    },
    {
      "epoch": 11.09,
      "learning_rate": 7.250169952413324e-05,
      "loss": 0.0902,
      "step": 35400
    },
    {
      "epoch": 11.12,
      "learning_rate": 7.224677090414683e-05,
      "loss": 0.0908,
      "step": 35500
    },
    {
      "epoch": 11.15,
      "learning_rate": 7.199184228416042e-05,
      "loss": 0.0885,
      "step": 35600
    },
    {
      "epoch": 11.19,
      "learning_rate": 7.173691366417403e-05,
      "loss": 0.0908,
      "step": 35700
    },
    {
      "epoch": 11.22,
      "learning_rate": 7.148198504418762e-05,
      "loss": 0.0892,
      "step": 35800
    },
    {
      "epoch": 11.25,
      "learning_rate": 7.122705642420122e-05,
      "loss": 0.0902,
      "step": 35900
    },
    {
      "epoch": 11.28,
      "learning_rate": 7.097212780421481e-05,
      "loss": 0.0883,
      "step": 36000
    },
    {
      "epoch": 11.31,
      "learning_rate": 7.07171991842284e-05,
      "loss": 0.0898,
      "step": 36100
    },
    {
      "epoch": 11.34,
      "learning_rate": 7.046227056424201e-05,
      "loss": 0.0887,
      "step": 36200
    },
    {
      "epoch": 11.37,
      "learning_rate": 7.02073419442556e-05,
      "loss": 0.0895,
      "step": 36300
    },
    {
      "epoch": 11.41,
      "learning_rate": 6.99524133242692e-05,
      "loss": 0.0905,
      "step": 36400
    },
    {
      "epoch": 11.44,
      "learning_rate": 6.96974847042828e-05,
      "loss": 0.0891,
      "step": 36500
    },
    {
      "epoch": 11.47,
      "learning_rate": 6.944255608429638e-05,
      "loss": 0.0896,
      "step": 36600
    },
    {
      "epoch": 11.5,
      "learning_rate": 6.918762746430999e-05,
      "loss": 0.0903,
      "step": 36700
    },
    {
      "epoch": 11.53,
      "learning_rate": 6.893269884432358e-05,
      "loss": 0.0901,
      "step": 36800
    },
    {
      "epoch": 11.56,
      "learning_rate": 6.867777022433718e-05,
      "loss": 0.0899,
      "step": 36900
    },
    {
      "epoch": 11.59,
      "learning_rate": 6.842284160435077e-05,
      "loss": 0.0881,
      "step": 37000
    },
    {
      "epoch": 11.62,
      "learning_rate": 6.816791298436437e-05,
      "loss": 0.0921,
      "step": 37100
    },
    {
      "epoch": 11.66,
      "learning_rate": 6.791298436437797e-05,
      "loss": 0.0884,
      "step": 37200
    },
    {
      "epoch": 11.69,
      "learning_rate": 6.765805574439157e-05,
      "loss": 0.0897,
      "step": 37300
    },
    {
      "epoch": 11.72,
      "learning_rate": 6.740312712440516e-05,
      "loss": 0.0877,
      "step": 37400
    },
    {
      "epoch": 11.75,
      "learning_rate": 6.714819850441876e-05,
      "loss": 0.0894,
      "step": 37500
    },
    {
      "epoch": 11.78,
      "learning_rate": 6.689326988443235e-05,
      "loss": 0.0881,
      "step": 37600
    },
    {
      "epoch": 11.81,
      "learning_rate": 6.663834126444595e-05,
      "loss": 0.0891,
      "step": 37700
    },
    {
      "epoch": 11.84,
      "learning_rate": 6.638341264445955e-05,
      "loss": 0.0873,
      "step": 37800
    },
    {
      "epoch": 11.88,
      "learning_rate": 6.612848402447315e-05,
      "loss": 0.0873,
      "step": 37900
    },
    {
      "epoch": 11.91,
      "learning_rate": 6.587355540448674e-05,
      "loss": 0.0887,
      "step": 38000
    },
    {
      "epoch": 11.94,
      "learning_rate": 6.561862678450033e-05,
      "loss": 0.0867,
      "step": 38100
    },
    {
      "epoch": 11.97,
      "learning_rate": 6.536369816451393e-05,
      "loss": 0.0906,
      "step": 38200
    },
    {
      "epoch": 12.0,
      "learning_rate": 6.510876954452752e-05,
      "loss": 0.0871,
      "step": 38300
    },
    {
      "epoch": 12.03,
      "learning_rate": 6.485384092454113e-05,
      "loss": 0.0816,
      "step": 38400
    },
    {
      "epoch": 12.06,
      "learning_rate": 6.459891230455472e-05,
      "loss": 0.0815,
      "step": 38500
    },
    {
      "epoch": 12.09,
      "learning_rate": 6.434398368456832e-05,
      "loss": 0.0817,
      "step": 38600
    },
    {
      "epoch": 12.13,
      "learning_rate": 6.408905506458191e-05,
      "loss": 0.0835,
      "step": 38700
    },
    {
      "epoch": 12.16,
      "learning_rate": 6.38341264445955e-05,
      "loss": 0.0833,
      "step": 38800
    },
    {
      "epoch": 12.19,
      "learning_rate": 6.35791978246091e-05,
      "loss": 0.081,
      "step": 38900
    },
    {
      "epoch": 12.22,
      "learning_rate": 6.33242692046227e-05,
      "loss": 0.0838,
      "step": 39000
    },
    {
      "epoch": 12.25,
      "learning_rate": 6.30693405846363e-05,
      "loss": 0.0829,
      "step": 39100
    },
    {
      "epoch": 12.28,
      "learning_rate": 6.281441196464989e-05,
      "loss": 0.082,
      "step": 39200
    },
    {
      "epoch": 12.31,
      "learning_rate": 6.255948334466348e-05,
      "loss": 0.0826,
      "step": 39300
    },
    {
      "epoch": 12.35,
      "learning_rate": 6.230455472467709e-05,
      "loss": 0.0799,
      "step": 39400
    },
    {
      "epoch": 12.38,
      "learning_rate": 6.204962610469068e-05,
      "loss": 0.0788,
      "step": 39500
    },
    {
      "epoch": 12.41,
      "learning_rate": 6.179469748470428e-05,
      "loss": 0.082,
      "step": 39600
    },
    {
      "epoch": 12.44,
      "learning_rate": 6.153976886471787e-05,
      "loss": 0.0823,
      "step": 39700
    },
    {
      "epoch": 12.47,
      "learning_rate": 6.128484024473146e-05,
      "loss": 0.08,
      "step": 39800
    },
    {
      "epoch": 12.5,
      "learning_rate": 6.102991162474506e-05,
      "loss": 0.0802,
      "step": 39900
    },
    {
      "epoch": 12.53,
      "learning_rate": 6.0774983004758665e-05,
      "loss": 0.0812,
      "step": 40000
    },
    {
      "epoch": 12.53,
      "eval_loss": 0.23678846657276154,
      "eval_runtime": 82.6634,
      "eval_samples_per_second": 108.162,
      "eval_steps_per_second": 1.694,
      "step": 40000
    },
    {
      "epoch": 12.56,
      "learning_rate": 6.052005438477226e-05,
      "loss": 0.0826,
      "step": 40100
    },
    {
      "epoch": 12.6,
      "learning_rate": 6.026512576478585e-05,
      "loss": 0.0814,
      "step": 40200
    },
    {
      "epoch": 12.63,
      "learning_rate": 6.001019714479945e-05,
      "loss": 0.0797,
      "step": 40300
    },
    {
      "epoch": 12.66,
      "learning_rate": 5.975526852481304e-05,
      "loss": 0.0805,
      "step": 40400
    },
    {
      "epoch": 12.69,
      "learning_rate": 5.9500339904826646e-05,
      "loss": 0.0796,
      "step": 40500
    },
    {
      "epoch": 12.72,
      "learning_rate": 5.924541128484024e-05,
      "loss": 0.0789,
      "step": 40600
    },
    {
      "epoch": 12.75,
      "learning_rate": 5.8990482664853834e-05,
      "loss": 0.0808,
      "step": 40700
    },
    {
      "epoch": 12.78,
      "learning_rate": 5.873555404486743e-05,
      "loss": 0.0819,
      "step": 40800
    },
    {
      "epoch": 12.82,
      "learning_rate": 5.848062542488102e-05,
      "loss": 0.0814,
      "step": 40900
    },
    {
      "epoch": 12.85,
      "learning_rate": 5.8225696804894626e-05,
      "loss": 0.0788,
      "step": 41000
    },
    {
      "epoch": 12.88,
      "learning_rate": 5.7970768184908224e-05,
      "loss": 0.0799,
      "step": 41100
    },
    {
      "epoch": 12.91,
      "learning_rate": 5.7715839564921815e-05,
      "loss": 0.0798,
      "step": 41200
    },
    {
      "epoch": 12.94,
      "learning_rate": 5.746091094493541e-05,
      "loss": 0.081,
      "step": 41300
    },
    {
      "epoch": 12.97,
      "learning_rate": 5.720598232494901e-05,
      "loss": 0.079,
      "step": 41400
    },
    {
      "epoch": 13.0,
      "learning_rate": 5.69510537049626e-05,
      "loss": 0.0804,
      "step": 41500
    },
    {
      "epoch": 13.03,
      "learning_rate": 5.6696125084976205e-05,
      "loss": 0.0745,
      "step": 41600
    },
    {
      "epoch": 13.07,
      "learning_rate": 5.64411964649898e-05,
      "loss": 0.0754,
      "step": 41700
    },
    {
      "epoch": 13.1,
      "learning_rate": 5.618626784500339e-05,
      "loss": 0.0735,
      "step": 41800
    },
    {
      "epoch": 13.13,
      "learning_rate": 5.593133922501699e-05,
      "loss": 0.0739,
      "step": 41900
    },
    {
      "epoch": 13.16,
      "learning_rate": 5.567641060503058e-05,
      "loss": 0.0734,
      "step": 42000
    },
    {
      "epoch": 13.19,
      "learning_rate": 5.5421481985044185e-05,
      "loss": 0.0726,
      "step": 42100
    },
    {
      "epoch": 13.22,
      "learning_rate": 5.516655336505778e-05,
      "loss": 0.0753,
      "step": 42200
    },
    {
      "epoch": 13.25,
      "learning_rate": 5.491162474507137e-05,
      "loss": 0.0743,
      "step": 42300
    },
    {
      "epoch": 13.29,
      "learning_rate": 5.465669612508497e-05,
      "loss": 0.0755,
      "step": 42400
    },
    {
      "epoch": 13.32,
      "learning_rate": 5.440176750509856e-05,
      "loss": 0.0726,
      "step": 42500
    },
    {
      "epoch": 13.35,
      "learning_rate": 5.4146838885112166e-05,
      "loss": 0.077,
      "step": 42600
    },
    {
      "epoch": 13.38,
      "learning_rate": 5.389191026512576e-05,
      "loss": 0.0753,
      "step": 42700
    },
    {
      "epoch": 13.41,
      "learning_rate": 5.3636981645139354e-05,
      "loss": 0.0729,
      "step": 42800
    },
    {
      "epoch": 13.44,
      "learning_rate": 5.338205302515295e-05,
      "loss": 0.0731,
      "step": 42900
    },
    {
      "epoch": 13.47,
      "learning_rate": 5.312712440516655e-05,
      "loss": 0.0723,
      "step": 43000
    },
    {
      "epoch": 13.5,
      "learning_rate": 5.287219578518014e-05,
      "loss": 0.0741,
      "step": 43100
    },
    {
      "epoch": 13.54,
      "learning_rate": 5.2617267165193744e-05,
      "loss": 0.0721,
      "step": 43200
    },
    {
      "epoch": 13.57,
      "learning_rate": 5.236233854520734e-05,
      "loss": 0.0726,
      "step": 43300
    },
    {
      "epoch": 13.6,
      "learning_rate": 5.210740992522093e-05,
      "loss": 0.0754,
      "step": 43400
    },
    {
      "epoch": 13.63,
      "learning_rate": 5.185248130523453e-05,
      "loss": 0.0741,
      "step": 43500
    },
    {
      "epoch": 13.66,
      "learning_rate": 5.159755268524812e-05,
      "loss": 0.0736,
      "step": 43600
    },
    {
      "epoch": 13.69,
      "learning_rate": 5.1342624065261725e-05,
      "loss": 0.0724,
      "step": 43700
    },
    {
      "epoch": 13.72,
      "learning_rate": 5.108769544527532e-05,
      "loss": 0.0739,
      "step": 43800
    },
    {
      "epoch": 13.76,
      "learning_rate": 5.083276682528891e-05,
      "loss": 0.072,
      "step": 43900
    },
    {
      "epoch": 13.79,
      "learning_rate": 5.057783820530251e-05,
      "loss": 0.0734,
      "step": 44000
    },
    {
      "epoch": 13.82,
      "learning_rate": 5.03229095853161e-05,
      "loss": 0.0734,
      "step": 44100
    },
    {
      "epoch": 13.85,
      "learning_rate": 5.0067980965329705e-05,
      "loss": 0.0718,
      "step": 44200
    },
    {
      "epoch": 13.88,
      "learning_rate": 4.98130523453433e-05,
      "loss": 0.0732,
      "step": 44300
    },
    {
      "epoch": 13.91,
      "learning_rate": 4.9558123725356893e-05,
      "loss": 0.0759,
      "step": 44400
    },
    {
      "epoch": 13.94,
      "learning_rate": 4.930319510537049e-05,
      "loss": 0.0715,
      "step": 44500
    },
    {
      "epoch": 13.97,
      "learning_rate": 4.904826648538409e-05,
      "loss": 0.0729,
      "step": 44600
    },
    {
      "epoch": 14.01,
      "learning_rate": 4.879333786539768e-05,
      "loss": 0.0712,
      "step": 44700
    },
    {
      "epoch": 14.04,
      "learning_rate": 4.853840924541128e-05,
      "loss": 0.0668,
      "step": 44800
    },
    {
      "epoch": 14.07,
      "learning_rate": 4.828348062542488e-05,
      "loss": 0.0677,
      "step": 44900
    },
    {
      "epoch": 14.1,
      "learning_rate": 4.802855200543847e-05,
      "loss": 0.066,
      "step": 45000
    },
    {
      "epoch": 14.1,
      "eval_loss": 0.22296179831027985,
      "eval_runtime": 91.9595,
      "eval_samples_per_second": 97.228,
      "eval_steps_per_second": 1.522,
      "step": 45000
    },
    {
      "epoch": 14.13,
      "learning_rate": 4.777362338545207e-05,
      "loss": 0.0669,
      "step": 45100
    },
    {
      "epoch": 14.16,
      "learning_rate": 4.751869476546566e-05,
      "loss": 0.0668,
      "step": 45200
    },
    {
      "epoch": 14.19,
      "learning_rate": 4.7263766145479264e-05,
      "loss": 0.0664,
      "step": 45300
    },
    {
      "epoch": 14.23,
      "learning_rate": 4.700883752549286e-05,
      "loss": 0.0661,
      "step": 45400
    },
    {
      "epoch": 14.26,
      "learning_rate": 4.675390890550645e-05,
      "loss": 0.0672,
      "step": 45500
    },
    {
      "epoch": 14.29,
      "learning_rate": 4.649898028552005e-05,
      "loss": 0.0665,
      "step": 45600
    },
    {
      "epoch": 14.32,
      "learning_rate": 4.624405166553364e-05,
      "loss": 0.0677,
      "step": 45700
    },
    {
      "epoch": 14.35,
      "learning_rate": 4.5989123045547245e-05,
      "loss": 0.0676,
      "step": 45800
    },
    {
      "epoch": 14.38,
      "learning_rate": 4.573419442556084e-05,
      "loss": 0.0665,
      "step": 45900
    },
    {
      "epoch": 14.41,
      "learning_rate": 4.547926580557443e-05,
      "loss": 0.0681,
      "step": 46000
    },
    {
      "epoch": 14.44,
      "learning_rate": 4.522433718558803e-05,
      "loss": 0.0673,
      "step": 46100
    },
    {
      "epoch": 14.48,
      "learning_rate": 4.496940856560163e-05,
      "loss": 0.0669,
      "step": 46200
    },
    {
      "epoch": 14.51,
      "learning_rate": 4.471447994561522e-05,
      "loss": 0.067,
      "step": 46300
    },
    {
      "epoch": 14.54,
      "learning_rate": 4.445955132562882e-05,
      "loss": 0.0665,
      "step": 46400
    },
    {
      "epoch": 14.57,
      "learning_rate": 4.420462270564242e-05,
      "loss": 0.0675,
      "step": 46500
    },
    {
      "epoch": 14.6,
      "learning_rate": 4.394969408565601e-05,
      "loss": 0.0658,
      "step": 46600
    },
    {
      "epoch": 14.63,
      "learning_rate": 4.369476546566961e-05,
      "loss": 0.064,
      "step": 46700
    },
    {
      "epoch": 14.66,
      "learning_rate": 4.34398368456832e-05,
      "loss": 0.0679,
      "step": 46800
    },
    {
      "epoch": 14.7,
      "learning_rate": 4.31849082256968e-05,
      "loss": 0.0652,
      "step": 46900
    },
    {
      "epoch": 14.73,
      "learning_rate": 4.29299796057104e-05,
      "loss": 0.0654,
      "step": 47000
    },
    {
      "epoch": 14.76,
      "learning_rate": 4.267505098572399e-05,
      "loss": 0.0672,
      "step": 47100
    },
    {
      "epoch": 14.79,
      "learning_rate": 4.242012236573759e-05,
      "loss": 0.0656,
      "step": 47200
    },
    {
      "epoch": 14.82,
      "learning_rate": 4.216519374575118e-05,
      "loss": 0.066,
      "step": 47300
    },
    {
      "epoch": 14.85,
      "learning_rate": 4.1910265125764784e-05,
      "loss": 0.065,
      "step": 47400
    },
    {
      "epoch": 14.88,
      "learning_rate": 4.165533650577838e-05,
      "loss": 0.0649,
      "step": 47500
    },
    {
      "epoch": 14.91,
      "learning_rate": 4.140040788579197e-05,
      "loss": 0.0665,
      "step": 47600
    },
    {
      "epoch": 14.95,
      "learning_rate": 4.114547926580557e-05,
      "loss": 0.0631,
      "step": 47700
    },
    {
      "epoch": 14.98,
      "learning_rate": 4.089055064581917e-05,
      "loss": 0.0658,
      "step": 47800
    },
    {
      "epoch": 15.01,
      "learning_rate": 4.063562202583276e-05,
      "loss": 0.0646,
      "step": 47900
    },
    {
      "epoch": 15.04,
      "learning_rate": 4.038069340584636e-05,
      "loss": 0.0594,
      "step": 48000
    },
    {
      "epoch": 15.07,
      "learning_rate": 4.012576478585996e-05,
      "loss": 0.0594,
      "step": 48100
    },
    {
      "epoch": 15.1,
      "learning_rate": 3.987083616587355e-05,
      "loss": 0.0598,
      "step": 48200
    },
    {
      "epoch": 15.13,
      "learning_rate": 3.961590754588715e-05,
      "loss": 0.0603,
      "step": 48300
    },
    {
      "epoch": 15.17,
      "learning_rate": 3.936097892590074e-05,
      "loss": 0.0616,
      "step": 48400
    },
    {
      "epoch": 15.2,
      "learning_rate": 3.910605030591434e-05,
      "loss": 0.0595,
      "step": 48500
    },
    {
      "epoch": 15.23,
      "learning_rate": 3.885112168592794e-05,
      "loss": 0.0598,
      "step": 48600
    },
    {
      "epoch": 15.26,
      "learning_rate": 3.859619306594153e-05,
      "loss": 0.0614,
      "step": 48700
    },
    {
      "epoch": 15.29,
      "learning_rate": 3.834126444595513e-05,
      "loss": 0.06,
      "step": 48800
    },
    {
      "epoch": 15.32,
      "learning_rate": 3.808633582596872e-05,
      "loss": 0.0611,
      "step": 48900
    },
    {
      "epoch": 15.35,
      "learning_rate": 3.783140720598232e-05,
      "loss": 0.0598,
      "step": 49000
    },
    {
      "epoch": 15.38,
      "learning_rate": 3.757647858599592e-05,
      "loss": 0.059,
      "step": 49100
    },
    {
      "epoch": 15.42,
      "learning_rate": 3.732154996600951e-05,
      "loss": 0.0596,
      "step": 49200
    },
    {
      "epoch": 15.45,
      "learning_rate": 3.706662134602311e-05,
      "loss": 0.0589,
      "step": 49300
    },
    {
      "epoch": 15.48,
      "learning_rate": 3.6811692726036707e-05,
      "loss": 0.0606,
      "step": 49400
    },
    {
      "epoch": 15.51,
      "learning_rate": 3.6556764106050304e-05,
      "loss": 0.0598,
      "step": 49500
    },
    {
      "epoch": 15.54,
      "learning_rate": 3.6301835486063895e-05,
      "loss": 0.0606,
      "step": 49600
    },
    {
      "epoch": 15.57,
      "learning_rate": 3.60469068660775e-05,
      "loss": 0.0588,
      "step": 49700
    },
    {
      "epoch": 15.6,
      "learning_rate": 3.579197824609109e-05,
      "loss": 0.0575,
      "step": 49800
    },
    {
      "epoch": 15.64,
      "learning_rate": 3.553704962610469e-05,
      "loss": 0.0596,
      "step": 49900
    },
    {
      "epoch": 15.67,
      "learning_rate": 3.5282121006118285e-05,
      "loss": 0.0574,
      "step": 50000
    },
    {
      "epoch": 15.67,
      "eval_loss": 0.2118692249059677,
      "eval_runtime": 95.2746,
      "eval_samples_per_second": 93.845,
      "eval_steps_per_second": 1.469,
      "step": 50000
    },
    {
      "epoch": 15.7,
      "learning_rate": 3.502719238613188e-05,
      "loss": 0.0592,
      "step": 50100
    },
    {
      "epoch": 15.73,
      "learning_rate": 3.477226376614548e-05,
      "loss": 0.0585,
      "step": 50200
    },
    {
      "epoch": 15.76,
      "learning_rate": 3.451733514615907e-05,
      "loss": 0.0572,
      "step": 50300
    },
    {
      "epoch": 15.79,
      "learning_rate": 3.426240652617267e-05,
      "loss": 0.0603,
      "step": 50400
    },
    {
      "epoch": 15.82,
      "learning_rate": 3.4007477906186265e-05,
      "loss": 0.0589,
      "step": 50500
    },
    {
      "epoch": 15.85,
      "learning_rate": 3.375254928619986e-05,
      "loss": 0.0573,
      "step": 50600
    },
    {
      "epoch": 15.89,
      "learning_rate": 3.3497620666213453e-05,
      "loss": 0.0582,
      "step": 50700
    },
    {
      "epoch": 15.92,
      "learning_rate": 3.324269204622705e-05,
      "loss": 0.0582,
      "step": 50800
    },
    {
      "epoch": 15.95,
      "learning_rate": 3.298776342624065e-05,
      "loss": 0.0598,
      "step": 50900
    },
    {
      "epoch": 15.98,
      "learning_rate": 3.2732834806254246e-05,
      "loss": 0.059,
      "step": 51000
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.2477906186267843e-05,
      "loss": 0.0567,
      "step": 51100
    },
    {
      "epoch": 16.04,
      "learning_rate": 3.2222977566281434e-05,
      "loss": 0.0555,
      "step": 51200
    },
    {
      "epoch": 16.07,
      "learning_rate": 3.196804894629504e-05,
      "loss": 0.0545,
      "step": 51300
    },
    {
      "epoch": 16.11,
      "learning_rate": 3.171312032630863e-05,
      "loss": 0.0536,
      "step": 51400
    },
    {
      "epoch": 16.14,
      "learning_rate": 3.1458191706322227e-05,
      "loss": 0.0542,
      "step": 51500
    },
    {
      "epoch": 16.17,
      "learning_rate": 3.1203263086335824e-05,
      "loss": 0.0536,
      "step": 51600
    },
    {
      "epoch": 16.2,
      "learning_rate": 3.094833446634942e-05,
      "loss": 0.0529,
      "step": 51700
    },
    {
      "epoch": 16.23,
      "learning_rate": 3.069340584636302e-05,
      "loss": 0.0529,
      "step": 51800
    },
    {
      "epoch": 16.26,
      "learning_rate": 3.0438477226376613e-05,
      "loss": 0.055,
      "step": 51900
    },
    {
      "epoch": 16.29,
      "learning_rate": 3.0183548606390207e-05,
      "loss": 0.0529,
      "step": 52000
    },
    {
      "epoch": 16.32,
      "learning_rate": 2.9928619986403805e-05,
      "loss": 0.0519,
      "step": 52100
    },
    {
      "epoch": 16.36,
      "learning_rate": 2.96736913664174e-05,
      "loss": 0.0545,
      "step": 52200
    },
    {
      "epoch": 16.39,
      "learning_rate": 2.9418762746430996e-05,
      "loss": 0.0538,
      "step": 52300
    },
    {
      "epoch": 16.42,
      "learning_rate": 2.9163834126444594e-05,
      "loss": 0.0534,
      "step": 52400
    },
    {
      "epoch": 16.45,
      "learning_rate": 2.8908905506458188e-05,
      "loss": 0.0521,
      "step": 52500
    },
    {
      "epoch": 16.48,
      "learning_rate": 2.8653976886471785e-05,
      "loss": 0.0532,
      "step": 52600
    },
    {
      "epoch": 16.51,
      "learning_rate": 2.8399048266485383e-05,
      "loss": 0.0547,
      "step": 52700
    },
    {
      "epoch": 16.54,
      "learning_rate": 2.8144119646498977e-05,
      "loss": 0.0545,
      "step": 52800
    },
    {
      "epoch": 16.58,
      "learning_rate": 2.7889191026512574e-05,
      "loss": 0.0519,
      "step": 52900
    },
    {
      "epoch": 16.61,
      "learning_rate": 2.763426240652617e-05,
      "loss": 0.054,
      "step": 53000
    },
    {
      "epoch": 16.64,
      "learning_rate": 2.7379333786539766e-05,
      "loss": 0.0532,
      "step": 53100
    },
    {
      "epoch": 16.67,
      "learning_rate": 2.7124405166553363e-05,
      "loss": 0.053,
      "step": 53200
    },
    {
      "epoch": 16.7,
      "learning_rate": 2.6869476546566958e-05,
      "loss": 0.0524,
      "step": 53300
    },
    {
      "epoch": 16.73,
      "learning_rate": 2.6614547926580555e-05,
      "loss": 0.0531,
      "step": 53400
    },
    {
      "epoch": 16.76,
      "learning_rate": 2.6359619306594152e-05,
      "loss": 0.0519,
      "step": 53500
    },
    {
      "epoch": 16.79,
      "learning_rate": 2.6104690686607747e-05,
      "loss": 0.0526,
      "step": 53600
    },
    {
      "epoch": 16.83,
      "learning_rate": 2.5849762066621344e-05,
      "loss": 0.0513,
      "step": 53700
    },
    {
      "epoch": 16.86,
      "learning_rate": 2.5594833446634938e-05,
      "loss": 0.0514,
      "step": 53800
    },
    {
      "epoch": 16.89,
      "learning_rate": 2.5339904826648536e-05,
      "loss": 0.0525,
      "step": 53900
    },
    {
      "epoch": 16.92,
      "learning_rate": 2.5084976206662133e-05,
      "loss": 0.0514,
      "step": 54000
    },
    {
      "epoch": 16.95,
      "learning_rate": 2.4830047586675727e-05,
      "loss": 0.0521,
      "step": 54100
    },
    {
      "epoch": 16.98,
      "learning_rate": 2.4575118966689328e-05,
      "loss": 0.0538,
      "step": 54200
    },
    {
      "epoch": 17.01,
      "learning_rate": 2.4320190346702922e-05,
      "loss": 0.0512,
      "step": 54300
    },
    {
      "epoch": 17.05,
      "learning_rate": 2.4065261726716516e-05,
      "loss": 0.0507,
      "step": 54400
    },
    {
      "epoch": 17.08,
      "learning_rate": 2.3810333106730114e-05,
      "loss": 0.0478,
      "step": 54500
    },
    {
      "epoch": 17.11,
      "learning_rate": 2.3555404486743708e-05,
      "loss": 0.0478,
      "step": 54600
    },
    {
      "epoch": 17.14,
      "learning_rate": 2.330047586675731e-05,
      "loss": 0.0485,
      "step": 54700
    },
    {
      "epoch": 17.17,
      "learning_rate": 2.3045547246770903e-05,
      "loss": 0.047,
      "step": 54800
    },
    {
      "epoch": 17.2,
      "learning_rate": 2.2790618626784497e-05,
      "loss": 0.048,
      "step": 54900
    },
    {
      "epoch": 17.23,
      "learning_rate": 2.2535690006798098e-05,
      "loss": 0.0463,
      "step": 55000
    },
    {
      "epoch": 17.23,
      "eval_loss": 0.20283755660057068,
      "eval_runtime": 69.2702,
      "eval_samples_per_second": 129.074,
      "eval_steps_per_second": 2.021,
      "step": 55000
    },
    {
      "epoch": 17.26,
      "learning_rate": 2.2280761386811692e-05,
      "loss": 0.0484,
      "step": 55100
    },
    {
      "epoch": 17.3,
      "learning_rate": 2.2025832766825286e-05,
      "loss": 0.0484,
      "step": 55200
    },
    {
      "epoch": 17.33,
      "learning_rate": 2.1770904146838883e-05,
      "loss": 0.0494,
      "step": 55300
    },
    {
      "epoch": 17.36,
      "learning_rate": 2.1515975526852478e-05,
      "loss": 0.0482,
      "step": 55400
    },
    {
      "epoch": 17.39,
      "learning_rate": 2.126104690686608e-05,
      "loss": 0.0474,
      "step": 55500
    },
    {
      "epoch": 17.42,
      "learning_rate": 2.1006118286879672e-05,
      "loss": 0.0457,
      "step": 55600
    },
    {
      "epoch": 17.45,
      "learning_rate": 2.0751189666893267e-05,
      "loss": 0.0497,
      "step": 55700
    },
    {
      "epoch": 17.48,
      "learning_rate": 2.0496261046906867e-05,
      "loss": 0.0466,
      "step": 55800
    },
    {
      "epoch": 17.52,
      "learning_rate": 2.024133242692046e-05,
      "loss": 0.0468,
      "step": 55900
    },
    {
      "epoch": 17.55,
      "learning_rate": 1.9986403806934056e-05,
      "loss": 0.0477,
      "step": 56000
    },
    {
      "epoch": 17.58,
      "learning_rate": 1.9731475186947653e-05,
      "loss": 0.0468,
      "step": 56100
    },
    {
      "epoch": 17.61,
      "learning_rate": 1.9476546566961247e-05,
      "loss": 0.0466,
      "step": 56200
    },
    {
      "epoch": 17.64,
      "learning_rate": 1.9221617946974848e-05,
      "loss": 0.0484,
      "step": 56300
    },
    {
      "epoch": 17.67,
      "learning_rate": 1.8966689326988442e-05,
      "loss": 0.0441,
      "step": 56400
    },
    {
      "epoch": 17.7,
      "learning_rate": 1.871176070700204e-05,
      "loss": 0.0486,
      "step": 56500
    },
    {
      "epoch": 17.73,
      "learning_rate": 1.8456832087015634e-05,
      "loss": 0.0477,
      "step": 56600
    },
    {
      "epoch": 17.77,
      "learning_rate": 1.820190346702923e-05,
      "loss": 0.0471,
      "step": 56700
    },
    {
      "epoch": 17.8,
      "learning_rate": 1.7946974847042825e-05,
      "loss": 0.0454,
      "step": 56800
    },
    {
      "epoch": 17.83,
      "learning_rate": 1.7692046227056423e-05,
      "loss": 0.0456,
      "step": 56900
    },
    {
      "epoch": 17.86,
      "learning_rate": 1.7437117607070017e-05,
      "loss": 0.0467,
      "step": 57000
    },
    {
      "epoch": 17.89,
      "learning_rate": 1.7182188987083614e-05,
      "loss": 0.0466,
      "step": 57100
    },
    {
      "epoch": 17.92,
      "learning_rate": 1.6927260367097212e-05,
      "loss": 0.046,
      "step": 57200
    },
    {
      "epoch": 17.95,
      "learning_rate": 1.667233174711081e-05,
      "loss": 0.0475,
      "step": 57300
    },
    {
      "epoch": 17.99,
      "learning_rate": 1.6417403127124403e-05,
      "loss": 0.0455,
      "step": 57400
    },
    {
      "epoch": 18.02,
      "learning_rate": 1.6162474507138e-05,
      "loss": 0.0447,
      "step": 57500
    },
    {
      "epoch": 18.05,
      "learning_rate": 1.5907545887151595e-05,
      "loss": 0.0422,
      "step": 57600
    },
    {
      "epoch": 18.08,
      "learning_rate": 1.5652617267165193e-05,
      "loss": 0.0432,
      "step": 57700
    },
    {
      "epoch": 18.11,
      "learning_rate": 1.5397688647178787e-05,
      "loss": 0.0423,
      "step": 57800
    },
    {
      "epoch": 18.14,
      "learning_rate": 1.5142760027192384e-05,
      "loss": 0.0427,
      "step": 57900
    },
    {
      "epoch": 18.17,
      "learning_rate": 1.4887831407205982e-05,
      "loss": 0.0445,
      "step": 58000
    },
    {
      "epoch": 18.2,
      "learning_rate": 1.4632902787219579e-05,
      "loss": 0.0437,
      "step": 58100
    },
    {
      "epoch": 18.24,
      "learning_rate": 1.4377974167233173e-05,
      "loss": 0.0433,
      "step": 58200
    },
    {
      "epoch": 18.27,
      "learning_rate": 1.4123045547246769e-05,
      "loss": 0.0447,
      "step": 58300
    },
    {
      "epoch": 18.3,
      "learning_rate": 1.3868116927260366e-05,
      "loss": 0.0417,
      "step": 58400
    },
    {
      "epoch": 18.33,
      "learning_rate": 1.3613188307273964e-05,
      "loss": 0.0435,
      "step": 58500
    },
    {
      "epoch": 18.36,
      "learning_rate": 1.3358259687287558e-05,
      "loss": 0.0423,
      "step": 58600
    },
    {
      "epoch": 18.39,
      "learning_rate": 1.3103331067301154e-05,
      "loss": 0.0429,
      "step": 58700
    },
    {
      "epoch": 18.42,
      "learning_rate": 1.2848402447314751e-05,
      "loss": 0.0417,
      "step": 58800
    },
    {
      "epoch": 18.46,
      "learning_rate": 1.2593473827328349e-05,
      "loss": 0.0422,
      "step": 58900
    },
    {
      "epoch": 18.49,
      "learning_rate": 1.2338545207341943e-05,
      "loss": 0.0433,
      "step": 59000
    },
    {
      "epoch": 18.52,
      "learning_rate": 1.2083616587355539e-05,
      "loss": 0.0407,
      "step": 59100
    },
    {
      "epoch": 18.55,
      "learning_rate": 1.1828687967369136e-05,
      "loss": 0.0432,
      "step": 59200
    },
    {
      "epoch": 18.58,
      "learning_rate": 1.1573759347382734e-05,
      "loss": 0.0417,
      "step": 59300
    },
    {
      "epoch": 18.61,
      "learning_rate": 1.1318830727396328e-05,
      "loss": 0.0419,
      "step": 59400
    },
    {
      "epoch": 18.64,
      "learning_rate": 1.1063902107409923e-05,
      "loss": 0.0415,
      "step": 59500
    },
    {
      "epoch": 18.67,
      "learning_rate": 1.0808973487423521e-05,
      "loss": 0.0407,
      "step": 59600
    },
    {
      "epoch": 18.71,
      "learning_rate": 1.0554044867437118e-05,
      "loss": 0.0409,
      "step": 59700
    },
    {
      "epoch": 18.74,
      "learning_rate": 1.0299116247450713e-05,
      "loss": 0.0424,
      "step": 59800
    },
    {
      "epoch": 18.77,
      "learning_rate": 1.0044187627464308e-05,
      "loss": 0.0419,
      "step": 59900
    },
    {
      "epoch": 18.8,
      "learning_rate": 9.789259007477906e-06,
      "loss": 0.04,
      "step": 60000
    },
    {
      "epoch": 18.8,
      "eval_loss": 0.1957029402256012,
      "eval_runtime": 84.9296,
      "eval_samples_per_second": 105.275,
      "eval_steps_per_second": 1.648,
      "step": 60000
    },
    {
      "epoch": 18.83,
      "learning_rate": 9.534330387491503e-06,
      "loss": 0.0405,
      "step": 60100
    },
    {
      "epoch": 18.86,
      "learning_rate": 9.279401767505097e-06,
      "loss": 0.0421,
      "step": 60200
    },
    {
      "epoch": 18.89,
      "learning_rate": 9.024473147518693e-06,
      "loss": 0.041,
      "step": 60300
    },
    {
      "epoch": 18.93,
      "learning_rate": 8.76954452753229e-06,
      "loss": 0.0406,
      "step": 60400
    },
    {
      "epoch": 18.96,
      "learning_rate": 8.514615907545886e-06,
      "loss": 0.0407,
      "step": 60500
    },
    {
      "epoch": 18.99,
      "learning_rate": 8.259687287559482e-06,
      "loss": 0.0403,
      "step": 60600
    },
    {
      "epoch": 19.02,
      "learning_rate": 8.004758667573078e-06,
      "loss": 0.0394,
      "step": 60700
    },
    {
      "epoch": 19.05,
      "learning_rate": 7.749830047586676e-06,
      "loss": 0.0392,
      "step": 60800
    },
    {
      "epoch": 19.08,
      "learning_rate": 7.494901427600271e-06,
      "loss": 0.0384,
      "step": 60900
    },
    {
      "epoch": 19.11,
      "learning_rate": 7.239972807613868e-06,
      "loss": 0.0382,
      "step": 61000
    },
    {
      "epoch": 19.14,
      "learning_rate": 6.985044187627464e-06,
      "loss": 0.0387,
      "step": 61100
    },
    {
      "epoch": 19.18,
      "learning_rate": 6.73011556764106e-06,
      "loss": 0.0396,
      "step": 61200
    },
    {
      "epoch": 19.21,
      "learning_rate": 6.475186947654656e-06,
      "loss": 0.0373,
      "step": 61300
    },
    {
      "epoch": 19.24,
      "learning_rate": 6.220258327668253e-06,
      "loss": 0.0367,
      "step": 61400
    },
    {
      "epoch": 19.27,
      "learning_rate": 5.9653297076818486e-06,
      "loss": 0.0403,
      "step": 61500
    },
    {
      "epoch": 19.3,
      "learning_rate": 5.710401087695445e-06,
      "loss": 0.0378,
      "step": 61600
    },
    {
      "epoch": 19.33,
      "learning_rate": 5.455472467709041e-06,
      "loss": 0.0383,
      "step": 61700
    },
    {
      "epoch": 19.36,
      "learning_rate": 5.200543847722638e-06,
      "loss": 0.0382,
      "step": 61800
    },
    {
      "epoch": 19.4,
      "learning_rate": 4.945615227736233e-06,
      "loss": 0.0398,
      "step": 61900
    },
    {
      "epoch": 19.43,
      "learning_rate": 4.69068660774983e-06,
      "loss": 0.0374,
      "step": 62000
    },
    {
      "epoch": 19.46,
      "learning_rate": 4.435757987763426e-06,
      "loss": 0.0383,
      "step": 62100
    },
    {
      "epoch": 19.49,
      "learning_rate": 4.180829367777022e-06,
      "loss": 0.0377,
      "step": 62200
    },
    {
      "epoch": 19.52,
      "learning_rate": 3.925900747790618e-06,
      "loss": 0.0377,
      "step": 62300
    },
    {
      "epoch": 19.55,
      "learning_rate": 3.6709721278042145e-06,
      "loss": 0.04,
      "step": 62400
    },
    {
      "epoch": 19.58,
      "learning_rate": 3.4160435078178107e-06,
      "loss": 0.0379,
      "step": 62500
    },
    {
      "epoch": 19.61,
      "learning_rate": 3.161114887831407e-06,
      "loss": 0.0371,
      "step": 62600
    },
    {
      "epoch": 19.65,
      "learning_rate": 2.906186267845003e-06,
      "loss": 0.0367,
      "step": 62700
    },
    {
      "epoch": 19.68,
      "learning_rate": 2.6512576478585993e-06,
      "loss": 0.0381,
      "step": 62800
    },
    {
      "epoch": 19.71,
      "learning_rate": 2.3963290278721955e-06,
      "loss": 0.0372,
      "step": 62900
    },
    {
      "epoch": 19.74,
      "learning_rate": 2.1414004078857917e-06,
      "loss": 0.0385,
      "step": 63000
    },
    {
      "epoch": 19.77,
      "learning_rate": 1.886471787899388e-06,
      "loss": 0.0377,
      "step": 63100
    },
    {
      "epoch": 19.8,
      "learning_rate": 1.6315431679129842e-06,
      "loss": 0.0377,
      "step": 63200
    },
    {
      "epoch": 19.83,
      "learning_rate": 1.3766145479265804e-06,
      "loss": 0.0366,
      "step": 63300
    },
    {
      "epoch": 19.87,
      "learning_rate": 1.1216859279401766e-06,
      "loss": 0.0382,
      "step": 63400
    },
    {
      "epoch": 19.9,
      "learning_rate": 8.667573079537729e-07,
      "loss": 0.0371,
      "step": 63500
    },
    {
      "epoch": 19.93,
      "learning_rate": 6.118286879673691e-07,
      "loss": 0.037,
      "step": 63600
    },
    {
      "epoch": 19.96,
      "learning_rate": 3.569000679809653e-07,
      "loss": 0.0367,
      "step": 63700
    },
    {
      "epoch": 19.99,
      "learning_rate": 1.0197144799456151e-07,
      "loss": 0.0366,
      "step": 63800
    },
    {
      "epoch": 20.0,
      "step": 63840,
      "total_flos": 1.325011889553408e+20,
      "train_loss": 0.13081293349950235,
      "train_runtime": 214172.6098,
      "train_samples_per_second": 76.308,
      "train_steps_per_second": 0.298
    }
  ],
  "max_steps": 63840,
  "num_train_epochs": 21,
  "total_flos": 1.325011889553408e+20,
  "trial_name": null,
  "trial_params": null
}