{
  "best_metric": 0.1807943880558014,
  "best_model_checkpoint": "whisper-tiny-full-data-v4/checkpoint-60000",
  "epoch": 20.0,
  "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": 4.2192,
      "step": 100
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.999999999999999e-06,
      "loss": 1.8058,
      "step": 200
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.999999999999999e-06,
      "loss": 1.2545,
      "step": 300
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 0.79,
      "step": 400
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 0.6743,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 0.6269,
      "step": 600
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.1e-05,
      "loss": 0.5921,
      "step": 700
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 0.5595,
      "step": 800
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 0.5426,
      "step": 900
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 0.5188,
      "step": 1000
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 0.5034,
      "step": 1100
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 0.4887,
      "step": 1200
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.9e-05,
      "loss": 0.4765,
      "step": 1300
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2e-05,
      "loss": 0.4706,
      "step": 1400
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 0.459,
      "step": 1500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 0.4504,
      "step": 1600
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.1e-05,
      "loss": 0.4458,
      "step": 1700
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.399999999999999e-05,
      "loss": 0.4422,
      "step": 1800
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.4348,
      "step": 1900
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.4318,
      "step": 2000
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.299999999999999e-05,
      "loss": 0.4208,
      "step": 2100
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.599999999999999e-05,
      "loss": 0.4189,
      "step": 2200
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.9e-05,
      "loss": 0.4157,
      "step": 2300
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.199999999999999e-05,
      "loss": 0.4235,
      "step": 2400
    },
    {
      "epoch": 0.78,
      "learning_rate": 7.5e-05,
      "loss": 0.4129,
      "step": 2500
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.8e-05,
      "loss": 0.4167,
      "step": 2600
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.1e-05,
      "loss": 0.4099,
      "step": 2700
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.4e-05,
      "loss": 0.411,
      "step": 2800
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.699999999999999e-05,
      "loss": 0.4075,
      "step": 2900
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.999999999999999e-05,
      "loss": 0.4012,
      "step": 3000
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.3e-05,
      "loss": 0.4032,
      "step": 3100
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.599999999999999e-05,
      "loss": 0.403,
      "step": 3200
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.9e-05,
      "loss": 0.3961,
      "step": 3300
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.000102,
      "loss": 0.3941,
      "step": 3400
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00010499999999999999,
      "loss": 0.3897,
      "step": 3500
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00010799999999999998,
      "loss": 0.3927,
      "step": 3600
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00011099999999999999,
      "loss": 0.3916,
      "step": 3700
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00011399999999999999,
      "loss": 0.3923,
      "step": 3800
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.000117,
      "loss": 0.3957,
      "step": 3900
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.3903,
      "step": 4000
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012299999999999998,
      "loss": 0.3957,
      "step": 4100
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012599999999999997,
      "loss": 0.3909,
      "step": 4200
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.000129,
      "loss": 0.3895,
      "step": 4300
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00013199999999999998,
      "loss": 0.3915,
      "step": 4400
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.000135,
      "loss": 0.3943,
      "step": 4500
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.000138,
      "loss": 0.3964,
      "step": 4600
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00014099999999999998,
      "loss": 0.3988,
      "step": 4700
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00014399999999999998,
      "loss": 0.3945,
      "step": 4800
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.000147,
      "loss": 0.3923,
      "step": 4900
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00015,
      "loss": 0.3932,
      "step": 5000
    },
    {
      "epoch": 1.57,
      "eval_loss": 0.5166156888008118,
      "eval_runtime": 161.0137,
      "eval_samples_per_second": 99.004,
      "eval_steps_per_second": 1.553,
      "step": 5000
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0001497450713800136,
      "loss": 0.3942,
      "step": 5100
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00014949014276002717,
      "loss": 0.3903,
      "step": 5200
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00014923521414004077,
      "loss": 0.3858,
      "step": 5300
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00014898028552005438,
      "loss": 0.3844,
      "step": 5400
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00014872535690006795,
      "loss": 0.3782,
      "step": 5500
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00014847042828008156,
      "loss": 0.3754,
      "step": 5600
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00014821549966009516,
      "loss": 0.3702,
      "step": 5700
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00014796057104010877,
      "loss": 0.3668,
      "step": 5800
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00014770564242012234,
      "loss": 0.3668,
      "step": 5900
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00014745071380013595,
      "loss": 0.3606,
      "step": 6000
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00014719578518014955,
      "loss": 0.3617,
      "step": 6100
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00014694085656016313,
      "loss": 0.3578,
      "step": 6200
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00014668592794017673,
      "loss": 0.3503,
      "step": 6300
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00014643099932019034,
      "loss": 0.3512,
      "step": 6400
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00014617607070020392,
      "loss": 0.3401,
      "step": 6500
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00014592114208021755,
      "loss": 0.3339,
      "step": 6600
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00014566621346023112,
      "loss": 0.3358,
      "step": 6700
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00014541128484024473,
      "loss": 0.3331,
      "step": 6800
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001451563562202583,
      "loss": 0.3297,
      "step": 6900
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001449014276002719,
      "loss": 0.3306,
      "step": 7000
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00014464649898028551,
      "loss": 0.3225,
      "step": 7100
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0001443915703602991,
      "loss": 0.3272,
      "step": 7200
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0001441366417403127,
      "loss": 0.3209,
      "step": 7300
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0001438817131203263,
      "loss": 0.32,
      "step": 7400
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00014362678450033988,
      "loss": 0.3135,
      "step": 7500
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00014337185588035348,
      "loss": 0.3181,
      "step": 7600
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00014311692726036708,
      "loss": 0.3166,
      "step": 7700
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001428619986403807,
      "loss": 0.3141,
      "step": 7800
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0001426070700203943,
      "loss": 0.3102,
      "step": 7900
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00014235214140040787,
      "loss": 0.3109,
      "step": 8000
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00014209721278042147,
      "loss": 0.3102,
      "step": 8100
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00014184228416043508,
      "loss": 0.3056,
      "step": 8200
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00014158735554044866,
      "loss": 0.3018,
      "step": 8300
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00014133242692046226,
      "loss": 0.3044,
      "step": 8400
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00014107749830047584,
      "loss": 0.2998,
      "step": 8500
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00014082256968048944,
      "loss": 0.2998,
      "step": 8600
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00014056764106050305,
      "loss": 0.2947,
      "step": 8700
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00014031271244051665,
      "loss": 0.296,
      "step": 8800
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00014005778382053025,
      "loss": 0.2939,
      "step": 8900
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00013980285520054383,
      "loss": 0.2968,
      "step": 9000
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00013954792658055744,
      "loss": 0.2945,
      "step": 9100
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00013929299796057104,
      "loss": 0.2917,
      "step": 9200
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00013903806934058462,
      "loss": 0.2896,
      "step": 9300
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00013878314072059822,
      "loss": 0.2863,
      "step": 9400
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00013852821210061183,
      "loss": 0.2822,
      "step": 9500
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0001382732834806254,
      "loss": 0.2843,
      "step": 9600
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.000138018354860639,
      "loss": 0.2754,
      "step": 9700
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001377634262406526,
      "loss": 0.2723,
      "step": 9800
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.00013750849762066622,
      "loss": 0.2741,
      "step": 9900
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0001372535690006798,
      "loss": 0.2683,
      "step": 10000
    },
    {
      "epoch": 3.13,
      "eval_loss": 0.38186827301979065,
      "eval_runtime": 143.9336,
      "eval_samples_per_second": 110.752,
      "eval_steps_per_second": 1.737,
      "step": 10000
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.0001369986403806934,
      "loss": 0.2717,
      "step": 10100
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.000136743711760707,
      "loss": 0.2652,
      "step": 10200
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.00013648878314072058,
      "loss": 0.273,
      "step": 10300
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.00013623385452073418,
      "loss": 0.2701,
      "step": 10400
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0001359789259007478,
      "loss": 0.2669,
      "step": 10500
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00013572399728076136,
      "loss": 0.2679,
      "step": 10600
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00013546906866077497,
      "loss": 0.2696,
      "step": 10700
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.00013521414004078857,
      "loss": 0.2684,
      "step": 10800
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.00013495921142080218,
      "loss": 0.2653,
      "step": 10900
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00013470428280081575,
      "loss": 0.2645,
      "step": 11000
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00013444935418082936,
      "loss": 0.2581,
      "step": 11100
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00013419442556084296,
      "loss": 0.2599,
      "step": 11200
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.00013393949694085654,
      "loss": 0.2563,
      "step": 11300
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.00013368456832087014,
      "loss": 0.2557,
      "step": 11400
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.00013342963970088375,
      "loss": 0.2532,
      "step": 11500
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.00013317471108089733,
      "loss": 0.2568,
      "step": 11600
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.00013291978246091093,
      "loss": 0.2549,
      "step": 11700
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.00013266485384092453,
      "loss": 0.2563,
      "step": 11800
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0001324099252209381,
      "loss": 0.2529,
      "step": 11900
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.00013215499660095172,
      "loss": 0.2531,
      "step": 12000
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.00013190006798096532,
      "loss": 0.253,
      "step": 12100
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00013164513936097892,
      "loss": 0.2533,
      "step": 12200
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.0001313902107409925,
      "loss": 0.2542,
      "step": 12300
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.0001311352821210061,
      "loss": 0.249,
      "step": 12400
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.0001308803535010197,
      "loss": 0.2515,
      "step": 12500
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.0001306254248810333,
      "loss": 0.2469,
      "step": 12600
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.0001303704962610469,
      "loss": 0.2473,
      "step": 12700
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.0001301155676410605,
      "loss": 0.241,
      "step": 12800
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.00012986063902107407,
      "loss": 0.2383,
      "step": 12900
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0001296057104010877,
      "loss": 0.24,
      "step": 13000
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.00012935078178110128,
      "loss": 0.2372,
      "step": 13100
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00012909585316111489,
      "loss": 0.2357,
      "step": 13200
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.0001288409245411285,
      "loss": 0.2354,
      "step": 13300
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.00012858599592114207,
      "loss": 0.2343,
      "step": 13400
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00012833106730115567,
      "loss": 0.2321,
      "step": 13500
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.00012807613868116925,
      "loss": 0.2311,
      "step": 13600
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00012782121006118285,
      "loss": 0.2333,
      "step": 13700
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00012756628144119646,
      "loss": 0.231,
      "step": 13800
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.00012731135282121003,
      "loss": 0.232,
      "step": 13900
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.00012705642420122364,
      "loss": 0.2338,
      "step": 14000
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00012680149558123724,
      "loss": 0.2343,
      "step": 14100
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.00012654656696125085,
      "loss": 0.2305,
      "step": 14200
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.00012629163834126445,
      "loss": 0.2286,
      "step": 14300
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.00012603670972127803,
      "loss": 0.2319,
      "step": 14400
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00012578178110129163,
      "loss": 0.2293,
      "step": 14500
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00012552685248130524,
      "loss": 0.2286,
      "step": 14600
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.00012527192386131881,
      "loss": 0.2279,
      "step": 14700
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.00012501699524133242,
      "loss": 0.2264,
      "step": 14800
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.000124762066621346,
      "loss": 0.2247,
      "step": 14900
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.0001245071380013596,
      "loss": 0.2227,
      "step": 15000
    },
    {
      "epoch": 4.7,
      "eval_loss": 0.3289188742637634,
      "eval_runtime": 160.0592,
      "eval_samples_per_second": 99.594,
      "eval_steps_per_second": 1.562,
      "step": 15000
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.0001242522093813732,
      "loss": 0.2266,
      "step": 15100
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.0001239972807613868,
      "loss": 0.2244,
      "step": 15200
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.0001237423521414004,
      "loss": 0.2254,
      "step": 15300
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.000123487423521414,
      "loss": 0.2194,
      "step": 15400
    },
    {
      "epoch": 4.86,
      "learning_rate": 0.0001232324949014276,
      "loss": 0.2253,
      "step": 15500
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.0001229775662814412,
      "loss": 0.2205,
      "step": 15600
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.00012272263766145478,
      "loss": 0.2237,
      "step": 15700
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.00012246770904146838,
      "loss": 0.218,
      "step": 15800
    },
    {
      "epoch": 4.98,
      "learning_rate": 0.00012221278042148198,
      "loss": 0.2222,
      "step": 15900
    },
    {
      "epoch": 5.01,
      "learning_rate": 0.00012195785180149556,
      "loss": 0.2151,
      "step": 16000
    },
    {
      "epoch": 5.04,
      "learning_rate": 0.00012170292318150917,
      "loss": 0.2072,
      "step": 16100
    },
    {
      "epoch": 5.08,
      "learning_rate": 0.00012144799456152277,
      "loss": 0.211,
      "step": 16200
    },
    {
      "epoch": 5.11,
      "learning_rate": 0.00012119306594153636,
      "loss": 0.2098,
      "step": 16300
    },
    {
      "epoch": 5.14,
      "learning_rate": 0.00012093813732154996,
      "loss": 0.2083,
      "step": 16400
    },
    {
      "epoch": 5.17,
      "learning_rate": 0.00012068320870156356,
      "loss": 0.209,
      "step": 16500
    },
    {
      "epoch": 5.2,
      "learning_rate": 0.00012042828008157715,
      "loss": 0.2092,
      "step": 16600
    },
    {
      "epoch": 5.23,
      "learning_rate": 0.00012017335146159075,
      "loss": 0.2077,
      "step": 16700
    },
    {
      "epoch": 5.26,
      "learning_rate": 0.00011991842284160434,
      "loss": 0.2105,
      "step": 16800
    },
    {
      "epoch": 5.29,
      "learning_rate": 0.00011966349422161793,
      "loss": 0.2098,
      "step": 16900
    },
    {
      "epoch": 5.33,
      "learning_rate": 0.00011940856560163152,
      "loss": 0.2124,
      "step": 17000
    },
    {
      "epoch": 5.36,
      "learning_rate": 0.00011915363698164513,
      "loss": 0.2059,
      "step": 17100
    },
    {
      "epoch": 5.39,
      "learning_rate": 0.00011889870836165872,
      "loss": 0.2072,
      "step": 17200
    },
    {
      "epoch": 5.42,
      "learning_rate": 0.00011864377974167233,
      "loss": 0.2065,
      "step": 17300
    },
    {
      "epoch": 5.45,
      "learning_rate": 0.00011838885112168593,
      "loss": 0.2056,
      "step": 17400
    },
    {
      "epoch": 5.48,
      "learning_rate": 0.00011813392250169952,
      "loss": 0.2048,
      "step": 17500
    },
    {
      "epoch": 5.51,
      "learning_rate": 0.00011787899388171311,
      "loss": 0.2058,
      "step": 17600
    },
    {
      "epoch": 5.55,
      "learning_rate": 0.00011762406526172671,
      "loss": 0.2033,
      "step": 17700
    },
    {
      "epoch": 5.58,
      "learning_rate": 0.0001173691366417403,
      "loss": 0.2039,
      "step": 17800
    },
    {
      "epoch": 5.61,
      "learning_rate": 0.00011711420802175389,
      "loss": 0.2047,
      "step": 17900
    },
    {
      "epoch": 5.64,
      "learning_rate": 0.0001168592794017675,
      "loss": 0.2056,
      "step": 18000
    },
    {
      "epoch": 5.67,
      "learning_rate": 0.00011660435078178109,
      "loss": 0.2016,
      "step": 18100
    },
    {
      "epoch": 5.7,
      "learning_rate": 0.00011634942216179468,
      "loss": 0.2036,
      "step": 18200
    },
    {
      "epoch": 5.73,
      "learning_rate": 0.00011609449354180827,
      "loss": 0.2043,
      "step": 18300
    },
    {
      "epoch": 5.76,
      "learning_rate": 0.00011583956492182189,
      "loss": 0.2025,
      "step": 18400
    },
    {
      "epoch": 5.8,
      "learning_rate": 0.00011558463630183548,
      "loss": 0.2021,
      "step": 18500
    },
    {
      "epoch": 5.83,
      "learning_rate": 0.00011532970768184908,
      "loss": 0.2049,
      "step": 18600
    },
    {
      "epoch": 5.86,
      "learning_rate": 0.00011507477906186267,
      "loss": 0.2003,
      "step": 18700
    },
    {
      "epoch": 5.89,
      "learning_rate": 0.00011481985044187626,
      "loss": 0.1972,
      "step": 18800
    },
    {
      "epoch": 5.92,
      "learning_rate": 0.00011456492182188985,
      "loss": 0.2025,
      "step": 18900
    },
    {
      "epoch": 5.95,
      "learning_rate": 0.00011430999320190346,
      "loss": 0.2015,
      "step": 19000
    },
    {
      "epoch": 5.98,
      "learning_rate": 0.00011405506458191705,
      "loss": 0.1993,
      "step": 19100
    },
    {
      "epoch": 6.02,
      "learning_rate": 0.00011380013596193064,
      "loss": 0.1937,
      "step": 19200
    },
    {
      "epoch": 6.05,
      "learning_rate": 0.00011354520734194424,
      "loss": 0.1901,
      "step": 19300
    },
    {
      "epoch": 6.08,
      "learning_rate": 0.00011329027872195785,
      "loss": 0.191,
      "step": 19400
    },
    {
      "epoch": 6.11,
      "learning_rate": 0.00011303535010197144,
      "loss": 0.1861,
      "step": 19500
    },
    {
      "epoch": 6.14,
      "learning_rate": 0.00011278042148198504,
      "loss": 0.1891,
      "step": 19600
    },
    {
      "epoch": 6.17,
      "learning_rate": 0.00011252549286199863,
      "loss": 0.1859,
      "step": 19700
    },
    {
      "epoch": 6.2,
      "learning_rate": 0.00011227056424201222,
      "loss": 0.191,
      "step": 19800
    },
    {
      "epoch": 6.23,
      "learning_rate": 0.00011201563562202583,
      "loss": 0.1903,
      "step": 19900
    },
    {
      "epoch": 6.27,
      "learning_rate": 0.00011176070700203942,
      "loss": 0.1905,
      "step": 20000
    },
    {
      "epoch": 6.27,
      "eval_loss": 0.29407182335853577,
      "eval_runtime": 168.3433,
      "eval_samples_per_second": 94.693,
      "eval_steps_per_second": 1.485,
      "step": 20000
    },
    {
      "epoch": 6.3,
      "learning_rate": 0.00011150577838205301,
      "loss": 0.1866,
      "step": 20100
    },
    {
      "epoch": 6.33,
      "learning_rate": 0.0001112508497620666,
      "loss": 0.1861,
      "step": 20200
    },
    {
      "epoch": 6.36,
      "learning_rate": 0.0001109959211420802,
      "loss": 0.1916,
      "step": 20300
    },
    {
      "epoch": 6.39,
      "learning_rate": 0.0001107409925220938,
      "loss": 0.1877,
      "step": 20400
    },
    {
      "epoch": 6.42,
      "learning_rate": 0.00011048606390210741,
      "loss": 0.1859,
      "step": 20500
    },
    {
      "epoch": 6.45,
      "learning_rate": 0.000110231135282121,
      "loss": 0.1887,
      "step": 20600
    },
    {
      "epoch": 6.48,
      "learning_rate": 0.0001099762066621346,
      "loss": 0.1866,
      "step": 20700
    },
    {
      "epoch": 6.52,
      "learning_rate": 0.00010972127804214819,
      "loss": 0.188,
      "step": 20800
    },
    {
      "epoch": 6.55,
      "learning_rate": 0.00010946634942216179,
      "loss": 0.1887,
      "step": 20900
    },
    {
      "epoch": 6.58,
      "learning_rate": 0.00010921142080217538,
      "loss": 0.1864,
      "step": 21000
    },
    {
      "epoch": 6.61,
      "learning_rate": 0.00010895649218218897,
      "loss": 0.1851,
      "step": 21100
    },
    {
      "epoch": 6.64,
      "learning_rate": 0.00010870156356220258,
      "loss": 0.1839,
      "step": 21200
    },
    {
      "epoch": 6.67,
      "learning_rate": 0.00010844663494221617,
      "loss": 0.1886,
      "step": 21300
    },
    {
      "epoch": 6.7,
      "learning_rate": 0.00010819170632222976,
      "loss": 0.1843,
      "step": 21400
    },
    {
      "epoch": 6.74,
      "learning_rate": 0.00010793677770224335,
      "loss": 0.1833,
      "step": 21500
    },
    {
      "epoch": 6.77,
      "learning_rate": 0.00010768184908225697,
      "loss": 0.1843,
      "step": 21600
    },
    {
      "epoch": 6.8,
      "learning_rate": 0.00010742692046227056,
      "loss": 0.1825,
      "step": 21700
    },
    {
      "epoch": 6.83,
      "learning_rate": 0.00010717199184228416,
      "loss": 0.1833,
      "step": 21800
    },
    {
      "epoch": 6.86,
      "learning_rate": 0.00010691706322229775,
      "loss": 0.1868,
      "step": 21900
    },
    {
      "epoch": 6.89,
      "learning_rate": 0.00010666213460231134,
      "loss": 0.1803,
      "step": 22000
    },
    {
      "epoch": 6.92,
      "learning_rate": 0.00010640720598232493,
      "loss": 0.1818,
      "step": 22100
    },
    {
      "epoch": 6.95,
      "learning_rate": 0.00010615227736233854,
      "loss": 0.1819,
      "step": 22200
    },
    {
      "epoch": 6.99,
      "learning_rate": 0.00010589734874235213,
      "loss": 0.1854,
      "step": 22300
    },
    {
      "epoch": 7.02,
      "learning_rate": 0.00010564242012236572,
      "loss": 0.1779,
      "step": 22400
    },
    {
      "epoch": 7.05,
      "learning_rate": 0.00010538749150237932,
      "loss": 0.1717,
      "step": 22500
    },
    {
      "epoch": 7.08,
      "learning_rate": 0.00010513256288239293,
      "loss": 0.1756,
      "step": 22600
    },
    {
      "epoch": 7.11,
      "learning_rate": 0.00010487763426240652,
      "loss": 0.1695,
      "step": 22700
    },
    {
      "epoch": 7.14,
      "learning_rate": 0.00010462270564242012,
      "loss": 0.1729,
      "step": 22800
    },
    {
      "epoch": 7.17,
      "learning_rate": 0.00010436777702243371,
      "loss": 0.1743,
      "step": 22900
    },
    {
      "epoch": 7.21,
      "learning_rate": 0.0001041128484024473,
      "loss": 0.1698,
      "step": 23000
    },
    {
      "epoch": 7.24,
      "learning_rate": 0.00010385791978246091,
      "loss": 0.1707,
      "step": 23100
    },
    {
      "epoch": 7.27,
      "learning_rate": 0.0001036029911624745,
      "loss": 0.1738,
      "step": 23200
    },
    {
      "epoch": 7.3,
      "learning_rate": 0.00010334806254248809,
      "loss": 0.1729,
      "step": 23300
    },
    {
      "epoch": 7.33,
      "learning_rate": 0.00010309313392250168,
      "loss": 0.1689,
      "step": 23400
    },
    {
      "epoch": 7.36,
      "learning_rate": 0.00010283820530251528,
      "loss": 0.1698,
      "step": 23500
    },
    {
      "epoch": 7.39,
      "learning_rate": 0.00010258327668252887,
      "loss": 0.1735,
      "step": 23600
    },
    {
      "epoch": 7.42,
      "learning_rate": 0.00010232834806254249,
      "loss": 0.1706,
      "step": 23700
    },
    {
      "epoch": 7.46,
      "learning_rate": 0.00010207341944255608,
      "loss": 0.1709,
      "step": 23800
    },
    {
      "epoch": 7.49,
      "learning_rate": 0.00010181849082256967,
      "loss": 0.1708,
      "step": 23900
    },
    {
      "epoch": 7.52,
      "learning_rate": 0.00010156356220258326,
      "loss": 0.17,
      "step": 24000
    },
    {
      "epoch": 7.55,
      "learning_rate": 0.00010130863358259687,
      "loss": 0.1702,
      "step": 24100
    },
    {
      "epoch": 7.58,
      "learning_rate": 0.00010105370496261046,
      "loss": 0.1702,
      "step": 24200
    },
    {
      "epoch": 7.61,
      "learning_rate": 0.00010079877634262405,
      "loss": 0.1704,
      "step": 24300
    },
    {
      "epoch": 7.64,
      "learning_rate": 0.00010054384772263765,
      "loss": 0.1719,
      "step": 24400
    },
    {
      "epoch": 7.68,
      "learning_rate": 0.00010028891910265125,
      "loss": 0.1691,
      "step": 24500
    },
    {
      "epoch": 7.71,
      "learning_rate": 0.00010003399048266484,
      "loss": 0.1711,
      "step": 24600
    },
    {
      "epoch": 7.74,
      "learning_rate": 9.977906186267843e-05,
      "loss": 0.1705,
      "step": 24700
    },
    {
      "epoch": 7.77,
      "learning_rate": 9.952413324269204e-05,
      "loss": 0.1712,
      "step": 24800
    },
    {
      "epoch": 7.8,
      "learning_rate": 9.926920462270564e-05,
      "loss": 0.1705,
      "step": 24900
    },
    {
      "epoch": 7.83,
      "learning_rate": 9.901427600271924e-05,
      "loss": 0.1691,
      "step": 25000
    },
    {
      "epoch": 7.83,
      "eval_loss": 0.2676524817943573,
      "eval_runtime": 164.633,
      "eval_samples_per_second": 96.827,
      "eval_steps_per_second": 1.519,
      "step": 25000
    },
    {
      "epoch": 7.86,
      "learning_rate": 9.875934738273283e-05,
      "loss": 0.1718,
      "step": 25100
    },
    {
      "epoch": 7.89,
      "learning_rate": 9.850441876274642e-05,
      "loss": 0.1682,
      "step": 25200
    },
    {
      "epoch": 7.93,
      "learning_rate": 9.824949014276001e-05,
      "loss": 0.1698,
      "step": 25300
    },
    {
      "epoch": 7.96,
      "learning_rate": 9.799456152277362e-05,
      "loss": 0.1684,
      "step": 25400
    },
    {
      "epoch": 7.99,
      "learning_rate": 9.77396329027872e-05,
      "loss": 0.1668,
      "step": 25500
    },
    {
      "epoch": 8.02,
      "learning_rate": 9.74847042828008e-05,
      "loss": 0.1602,
      "step": 25600
    },
    {
      "epoch": 8.05,
      "learning_rate": 9.72297756628144e-05,
      "loss": 0.1582,
      "step": 25700
    },
    {
      "epoch": 8.08,
      "learning_rate": 9.6974847042828e-05,
      "loss": 0.1593,
      "step": 25800
    },
    {
      "epoch": 8.11,
      "learning_rate": 9.67199184228416e-05,
      "loss": 0.1613,
      "step": 25900
    },
    {
      "epoch": 8.15,
      "learning_rate": 9.64649898028552e-05,
      "loss": 0.1573,
      "step": 26000
    },
    {
      "epoch": 8.18,
      "learning_rate": 9.621006118286879e-05,
      "loss": 0.157,
      "step": 26100
    },
    {
      "epoch": 8.21,
      "learning_rate": 9.595513256288238e-05,
      "loss": 0.1578,
      "step": 26200
    },
    {
      "epoch": 8.24,
      "learning_rate": 9.570020394289599e-05,
      "loss": 0.1586,
      "step": 26300
    },
    {
      "epoch": 8.27,
      "learning_rate": 9.544527532290958e-05,
      "loss": 0.1578,
      "step": 26400
    },
    {
      "epoch": 8.3,
      "learning_rate": 9.519034670292317e-05,
      "loss": 0.1617,
      "step": 26500
    },
    {
      "epoch": 8.33,
      "learning_rate": 9.493541808293676e-05,
      "loss": 0.1595,
      "step": 26600
    },
    {
      "epoch": 8.36,
      "learning_rate": 9.468048946295036e-05,
      "loss": 0.1568,
      "step": 26700
    },
    {
      "epoch": 8.4,
      "learning_rate": 9.442556084296395e-05,
      "loss": 0.1572,
      "step": 26800
    },
    {
      "epoch": 8.43,
      "learning_rate": 9.417063222297757e-05,
      "loss": 0.1554,
      "step": 26900
    },
    {
      "epoch": 8.46,
      "learning_rate": 9.391570360299116e-05,
      "loss": 0.1588,
      "step": 27000
    },
    {
      "epoch": 8.49,
      "learning_rate": 9.366077498300475e-05,
      "loss": 0.157,
      "step": 27100
    },
    {
      "epoch": 8.52,
      "learning_rate": 9.340584636301834e-05,
      "loss": 0.158,
      "step": 27200
    },
    {
      "epoch": 8.55,
      "learning_rate": 9.315091774303195e-05,
      "loss": 0.1567,
      "step": 27300
    },
    {
      "epoch": 8.58,
      "learning_rate": 9.289598912304554e-05,
      "loss": 0.1574,
      "step": 27400
    },
    {
      "epoch": 8.62,
      "learning_rate": 9.264106050305913e-05,
      "loss": 0.1565,
      "step": 27500
    },
    {
      "epoch": 8.65,
      "learning_rate": 9.238613188307273e-05,
      "loss": 0.153,
      "step": 27600
    },
    {
      "epoch": 8.68,
      "learning_rate": 9.213120326308632e-05,
      "loss": 0.1548,
      "step": 27700
    },
    {
      "epoch": 8.71,
      "learning_rate": 9.187627464309991e-05,
      "loss": 0.1537,
      "step": 27800
    },
    {
      "epoch": 8.74,
      "learning_rate": 9.162134602311352e-05,
      "loss": 0.155,
      "step": 27900
    },
    {
      "epoch": 8.77,
      "learning_rate": 9.136641740312712e-05,
      "loss": 0.1565,
      "step": 28000
    },
    {
      "epoch": 8.8,
      "learning_rate": 9.111148878314071e-05,
      "loss": 0.1553,
      "step": 28100
    },
    {
      "epoch": 8.83,
      "learning_rate": 9.085656016315432e-05,
      "loss": 0.1529,
      "step": 28200
    },
    {
      "epoch": 8.87,
      "learning_rate": 9.060163154316791e-05,
      "loss": 0.1568,
      "step": 28300
    },
    {
      "epoch": 8.9,
      "learning_rate": 9.03467029231815e-05,
      "loss": 0.1547,
      "step": 28400
    },
    {
      "epoch": 8.93,
      "learning_rate": 9.009177430319509e-05,
      "loss": 0.1546,
      "step": 28500
    },
    {
      "epoch": 8.96,
      "learning_rate": 8.98368456832087e-05,
      "loss": 0.1535,
      "step": 28600
    },
    {
      "epoch": 8.99,
      "learning_rate": 8.958191706322229e-05,
      "loss": 0.1558,
      "step": 28700
    },
    {
      "epoch": 9.02,
      "learning_rate": 8.932698844323588e-05,
      "loss": 0.1471,
      "step": 28800
    },
    {
      "epoch": 9.05,
      "learning_rate": 8.907205982324948e-05,
      "loss": 0.1445,
      "step": 28900
    },
    {
      "epoch": 9.09,
      "learning_rate": 8.881713120326308e-05,
      "loss": 0.1469,
      "step": 29000
    },
    {
      "epoch": 9.12,
      "learning_rate": 8.856220258327668e-05,
      "loss": 0.1455,
      "step": 29100
    },
    {
      "epoch": 9.15,
      "learning_rate": 8.830727396329028e-05,
      "loss": 0.1467,
      "step": 29200
    },
    {
      "epoch": 9.18,
      "learning_rate": 8.805234534330387e-05,
      "loss": 0.1453,
      "step": 29300
    },
    {
      "epoch": 9.21,
      "learning_rate": 8.779741672331746e-05,
      "loss": 0.1437,
      "step": 29400
    },
    {
      "epoch": 9.24,
      "learning_rate": 8.754248810333107e-05,
      "loss": 0.1464,
      "step": 29500
    },
    {
      "epoch": 9.27,
      "learning_rate": 8.728755948334466e-05,
      "loss": 0.1487,
      "step": 29600
    },
    {
      "epoch": 9.3,
      "learning_rate": 8.703263086335825e-05,
      "loss": 0.1454,
      "step": 29700
    },
    {
      "epoch": 9.34,
      "learning_rate": 8.677770224337184e-05,
      "loss": 0.1489,
      "step": 29800
    },
    {
      "epoch": 9.37,
      "learning_rate": 8.652277362338544e-05,
      "loss": 0.1491,
      "step": 29900
    },
    {
      "epoch": 9.4,
      "learning_rate": 8.626784500339903e-05,
      "loss": 0.1421,
      "step": 30000
    },
    {
      "epoch": 9.4,
      "eval_loss": 0.24991033971309662,
      "eval_runtime": 168.3866,
      "eval_samples_per_second": 94.669,
      "eval_steps_per_second": 1.485,
      "step": 30000
    },
    {
      "epoch": 9.43,
      "learning_rate": 8.601291638341265e-05,
      "loss": 0.144,
      "step": 30100
    },
    {
      "epoch": 9.46,
      "learning_rate": 8.575798776342624e-05,
      "loss": 0.1448,
      "step": 30200
    },
    {
      "epoch": 9.49,
      "learning_rate": 8.550305914343983e-05,
      "loss": 0.1445,
      "step": 30300
    },
    {
      "epoch": 9.52,
      "learning_rate": 8.524813052345342e-05,
      "loss": 0.1428,
      "step": 30400
    },
    {
      "epoch": 9.56,
      "learning_rate": 8.499320190346703e-05,
      "loss": 0.1457,
      "step": 30500
    },
    {
      "epoch": 9.59,
      "learning_rate": 8.473827328348062e-05,
      "loss": 0.1425,
      "step": 30600
    },
    {
      "epoch": 9.62,
      "learning_rate": 8.448334466349421e-05,
      "loss": 0.1434,
      "step": 30700
    },
    {
      "epoch": 9.65,
      "learning_rate": 8.422841604350781e-05,
      "loss": 0.1431,
      "step": 30800
    },
    {
      "epoch": 9.68,
      "learning_rate": 8.39734874235214e-05,
      "loss": 0.1489,
      "step": 30900
    },
    {
      "epoch": 9.71,
      "learning_rate": 8.3718558803535e-05,
      "loss": 0.1448,
      "step": 31000
    },
    {
      "epoch": 9.74,
      "learning_rate": 8.34636301835486e-05,
      "loss": 0.1427,
      "step": 31100
    },
    {
      "epoch": 9.77,
      "learning_rate": 8.32087015635622e-05,
      "loss": 0.1423,
      "step": 31200
    },
    {
      "epoch": 9.81,
      "learning_rate": 8.295377294357579e-05,
      "loss": 0.1431,
      "step": 31300
    },
    {
      "epoch": 9.84,
      "learning_rate": 8.26988443235894e-05,
      "loss": 0.1419,
      "step": 31400
    },
    {
      "epoch": 9.87,
      "learning_rate": 8.244391570360299e-05,
      "loss": 0.1429,
      "step": 31500
    },
    {
      "epoch": 9.9,
      "learning_rate": 8.218898708361658e-05,
      "loss": 0.1432,
      "step": 31600
    },
    {
      "epoch": 9.93,
      "learning_rate": 8.193405846363017e-05,
      "loss": 0.1427,
      "step": 31700
    },
    {
      "epoch": 9.96,
      "learning_rate": 8.167912984364377e-05,
      "loss": 0.1423,
      "step": 31800
    },
    {
      "epoch": 9.99,
      "learning_rate": 8.142420122365736e-05,
      "loss": 0.1419,
      "step": 31900
    },
    {
      "epoch": 10.03,
      "learning_rate": 8.116927260367095e-05,
      "loss": 0.1331,
      "step": 32000
    },
    {
      "epoch": 10.06,
      "learning_rate": 8.091434398368456e-05,
      "loss": 0.1363,
      "step": 32100
    },
    {
      "epoch": 10.09,
      "learning_rate": 8.065941536369816e-05,
      "loss": 0.1348,
      "step": 32200
    },
    {
      "epoch": 10.12,
      "learning_rate": 8.040448674371175e-05,
      "loss": 0.1324,
      "step": 32300
    },
    {
      "epoch": 10.15,
      "learning_rate": 8.014955812372536e-05,
      "loss": 0.1327,
      "step": 32400
    },
    {
      "epoch": 10.18,
      "learning_rate": 7.989462950373895e-05,
      "loss": 0.1353,
      "step": 32500
    },
    {
      "epoch": 10.21,
      "learning_rate": 7.963970088375254e-05,
      "loss": 0.133,
      "step": 32600
    },
    {
      "epoch": 10.24,
      "learning_rate": 7.938477226376614e-05,
      "loss": 0.1342,
      "step": 32700
    },
    {
      "epoch": 10.28,
      "learning_rate": 7.912984364377973e-05,
      "loss": 0.1349,
      "step": 32800
    },
    {
      "epoch": 10.31,
      "learning_rate": 7.887491502379333e-05,
      "loss": 0.1324,
      "step": 32900
    },
    {
      "epoch": 10.34,
      "learning_rate": 7.861998640380692e-05,
      "loss": 0.1369,
      "step": 33000
    },
    {
      "epoch": 10.37,
      "learning_rate": 7.836505778382052e-05,
      "loss": 0.1304,
      "step": 33100
    },
    {
      "epoch": 10.4,
      "learning_rate": 7.811012916383411e-05,
      "loss": 0.1331,
      "step": 33200
    },
    {
      "epoch": 10.43,
      "learning_rate": 7.785520054384773e-05,
      "loss": 0.1354,
      "step": 33300
    },
    {
      "epoch": 10.46,
      "learning_rate": 7.760027192386132e-05,
      "loss": 0.1321,
      "step": 33400
    },
    {
      "epoch": 10.49,
      "learning_rate": 7.734534330387491e-05,
      "loss": 0.1363,
      "step": 33500
    },
    {
      "epoch": 10.53,
      "learning_rate": 7.70904146838885e-05,
      "loss": 0.134,
      "step": 33600
    },
    {
      "epoch": 10.56,
      "learning_rate": 7.68354860639021e-05,
      "loss": 0.1336,
      "step": 33700
    },
    {
      "epoch": 10.59,
      "learning_rate": 7.65805574439157e-05,
      "loss": 0.1304,
      "step": 33800
    },
    {
      "epoch": 10.62,
      "learning_rate": 7.632562882392929e-05,
      "loss": 0.1324,
      "step": 33900
    },
    {
      "epoch": 10.65,
      "learning_rate": 7.607070020394289e-05,
      "loss": 0.1347,
      "step": 34000
    },
    {
      "epoch": 10.68,
      "learning_rate": 7.581577158395648e-05,
      "loss": 0.1319,
      "step": 34100
    },
    {
      "epoch": 10.71,
      "learning_rate": 7.556084296397007e-05,
      "loss": 0.135,
      "step": 34200
    },
    {
      "epoch": 10.75,
      "learning_rate": 7.530591434398368e-05,
      "loss": 0.1322,
      "step": 34300
    },
    {
      "epoch": 10.78,
      "learning_rate": 7.505098572399728e-05,
      "loss": 0.1332,
      "step": 34400
    },
    {
      "epoch": 10.81,
      "learning_rate": 7.479605710401087e-05,
      "loss": 0.1302,
      "step": 34500
    },
    {
      "epoch": 10.84,
      "learning_rate": 7.454112848402448e-05,
      "loss": 0.1331,
      "step": 34600
    },
    {
      "epoch": 10.87,
      "learning_rate": 7.428619986403807e-05,
      "loss": 0.133,
      "step": 34700
    },
    {
      "epoch": 10.9,
      "learning_rate": 7.403127124405166e-05,
      "loss": 0.1334,
      "step": 34800
    },
    {
      "epoch": 10.93,
      "learning_rate": 7.377634262406525e-05,
      "loss": 0.1317,
      "step": 34900
    },
    {
      "epoch": 10.96,
      "learning_rate": 7.352141400407885e-05,
      "loss": 0.1355,
      "step": 35000
    },
    {
      "epoch": 10.96,
      "eval_loss": 0.2304794043302536,
      "eval_runtime": 155.1705,
      "eval_samples_per_second": 102.732,
      "eval_steps_per_second": 1.611,
      "step": 35000
    },
    {
      "epoch": 11.0,
      "learning_rate": 7.326648538409244e-05,
      "loss": 0.1331,
      "step": 35100
    },
    {
      "epoch": 11.03,
      "learning_rate": 7.301155676410605e-05,
      "loss": 0.1252,
      "step": 35200
    },
    {
      "epoch": 11.06,
      "learning_rate": 7.275662814411964e-05,
      "loss": 0.1237,
      "step": 35300
    },
    {
      "epoch": 11.09,
      "learning_rate": 7.250169952413324e-05,
      "loss": 0.1238,
      "step": 35400
    },
    {
      "epoch": 11.12,
      "learning_rate": 7.224677090414683e-05,
      "loss": 0.1235,
      "step": 35500
    },
    {
      "epoch": 11.15,
      "learning_rate": 7.199184228416042e-05,
      "loss": 0.121,
      "step": 35600
    },
    {
      "epoch": 11.18,
      "learning_rate": 7.173691366417403e-05,
      "loss": 0.1246,
      "step": 35700
    },
    {
      "epoch": 11.22,
      "learning_rate": 7.148198504418762e-05,
      "loss": 0.1243,
      "step": 35800
    },
    {
      "epoch": 11.25,
      "learning_rate": 7.122705642420122e-05,
      "loss": 0.1233,
      "step": 35900
    },
    {
      "epoch": 11.28,
      "learning_rate": 7.097212780421481e-05,
      "loss": 0.1219,
      "step": 36000
    },
    {
      "epoch": 11.31,
      "learning_rate": 7.07171991842284e-05,
      "loss": 0.124,
      "step": 36100
    },
    {
      "epoch": 11.34,
      "learning_rate": 7.046227056424201e-05,
      "loss": 0.1239,
      "step": 36200
    },
    {
      "epoch": 11.37,
      "learning_rate": 7.02073419442556e-05,
      "loss": 0.1226,
      "step": 36300
    },
    {
      "epoch": 11.4,
      "learning_rate": 6.99524133242692e-05,
      "loss": 0.126,
      "step": 36400
    },
    {
      "epoch": 11.43,
      "learning_rate": 6.96974847042828e-05,
      "loss": 0.1219,
      "step": 36500
    },
    {
      "epoch": 11.47,
      "learning_rate": 6.944255608429638e-05,
      "loss": 0.1236,
      "step": 36600
    },
    {
      "epoch": 11.5,
      "learning_rate": 6.918762746430999e-05,
      "loss": 0.1232,
      "step": 36700
    },
    {
      "epoch": 11.53,
      "learning_rate": 6.893269884432358e-05,
      "loss": 0.1253,
      "step": 36800
    },
    {
      "epoch": 11.56,
      "learning_rate": 6.867777022433718e-05,
      "loss": 0.124,
      "step": 36900
    },
    {
      "epoch": 11.59,
      "learning_rate": 6.842284160435077e-05,
      "loss": 0.1201,
      "step": 37000
    },
    {
      "epoch": 11.62,
      "learning_rate": 6.816791298436437e-05,
      "loss": 0.1272,
      "step": 37100
    },
    {
      "epoch": 11.65,
      "learning_rate": 6.791298436437797e-05,
      "loss": 0.1224,
      "step": 37200
    },
    {
      "epoch": 11.69,
      "learning_rate": 6.765805574439157e-05,
      "loss": 0.123,
      "step": 37300
    },
    {
      "epoch": 11.72,
      "learning_rate": 6.740312712440516e-05,
      "loss": 0.1216,
      "step": 37400
    },
    {
      "epoch": 11.75,
      "learning_rate": 6.714819850441876e-05,
      "loss": 0.1229,
      "step": 37500
    },
    {
      "epoch": 11.78,
      "learning_rate": 6.689326988443235e-05,
      "loss": 0.1223,
      "step": 37600
    },
    {
      "epoch": 11.81,
      "learning_rate": 6.663834126444595e-05,
      "loss": 0.1222,
      "step": 37700
    },
    {
      "epoch": 11.84,
      "learning_rate": 6.638341264445955e-05,
      "loss": 0.1209,
      "step": 37800
    },
    {
      "epoch": 11.87,
      "learning_rate": 6.612848402447315e-05,
      "loss": 0.1209,
      "step": 37900
    },
    {
      "epoch": 11.9,
      "learning_rate": 6.587355540448674e-05,
      "loss": 0.1222,
      "step": 38000
    },
    {
      "epoch": 11.94,
      "learning_rate": 6.561862678450033e-05,
      "loss": 0.1202,
      "step": 38100
    },
    {
      "epoch": 11.97,
      "learning_rate": 6.536369816451393e-05,
      "loss": 0.1236,
      "step": 38200
    },
    {
      "epoch": 12.0,
      "learning_rate": 6.510876954452752e-05,
      "loss": 0.1215,
      "step": 38300
    },
    {
      "epoch": 12.03,
      "learning_rate": 6.485384092454113e-05,
      "loss": 0.1126,
      "step": 38400
    },
    {
      "epoch": 12.06,
      "learning_rate": 6.459891230455472e-05,
      "loss": 0.1124,
      "step": 38500
    },
    {
      "epoch": 12.09,
      "learning_rate": 6.434398368456832e-05,
      "loss": 0.1151,
      "step": 38600
    },
    {
      "epoch": 12.12,
      "learning_rate": 6.408905506458191e-05,
      "loss": 0.1154,
      "step": 38700
    },
    {
      "epoch": 12.16,
      "learning_rate": 6.38341264445955e-05,
      "loss": 0.116,
      "step": 38800
    },
    {
      "epoch": 12.19,
      "learning_rate": 6.35791978246091e-05,
      "loss": 0.1133,
      "step": 38900
    },
    {
      "epoch": 12.22,
      "learning_rate": 6.33242692046227e-05,
      "loss": 0.116,
      "step": 39000
    },
    {
      "epoch": 12.25,
      "learning_rate": 6.30693405846363e-05,
      "loss": 0.1139,
      "step": 39100
    },
    {
      "epoch": 12.28,
      "learning_rate": 6.281441196464989e-05,
      "loss": 0.1166,
      "step": 39200
    },
    {
      "epoch": 12.31,
      "learning_rate": 6.255948334466348e-05,
      "loss": 0.1157,
      "step": 39300
    },
    {
      "epoch": 12.34,
      "learning_rate": 6.230455472467709e-05,
      "loss": 0.1114,
      "step": 39400
    },
    {
      "epoch": 12.37,
      "learning_rate": 6.204962610469068e-05,
      "loss": 0.1116,
      "step": 39500
    },
    {
      "epoch": 12.41,
      "learning_rate": 6.179469748470428e-05,
      "loss": 0.1143,
      "step": 39600
    },
    {
      "epoch": 12.44,
      "learning_rate": 6.153976886471787e-05,
      "loss": 0.1143,
      "step": 39700
    },
    {
      "epoch": 12.47,
      "learning_rate": 6.128484024473146e-05,
      "loss": 0.1128,
      "step": 39800
    },
    {
      "epoch": 12.5,
      "learning_rate": 6.102991162474506e-05,
      "loss": 0.112,
      "step": 39900
    },
    {
      "epoch": 12.53,
      "learning_rate": 6.0774983004758665e-05,
      "loss": 0.1126,
      "step": 40000
    },
    {
      "epoch": 12.53,
      "eval_loss": 0.21851351857185364,
      "eval_runtime": 159.3876,
      "eval_samples_per_second": 100.014,
      "eval_steps_per_second": 1.569,
      "step": 40000
    },
    {
      "epoch": 12.56,
      "learning_rate": 6.052005438477226e-05,
      "loss": 0.1148,
      "step": 40100
    },
    {
      "epoch": 12.59,
      "learning_rate": 6.026512576478585e-05,
      "loss": 0.1135,
      "step": 40200
    },
    {
      "epoch": 12.63,
      "learning_rate": 6.001019714479945e-05,
      "loss": 0.1119,
      "step": 40300
    },
    {
      "epoch": 12.66,
      "learning_rate": 5.975526852481304e-05,
      "loss": 0.112,
      "step": 40400
    },
    {
      "epoch": 12.69,
      "learning_rate": 5.9500339904826646e-05,
      "loss": 0.1134,
      "step": 40500
    },
    {
      "epoch": 12.72,
      "learning_rate": 5.924541128484024e-05,
      "loss": 0.1099,
      "step": 40600
    },
    {
      "epoch": 12.75,
      "learning_rate": 5.8990482664853834e-05,
      "loss": 0.113,
      "step": 40700
    },
    {
      "epoch": 12.78,
      "learning_rate": 5.873555404486743e-05,
      "loss": 0.1146,
      "step": 40800
    },
    {
      "epoch": 12.81,
      "learning_rate": 5.848062542488102e-05,
      "loss": 0.1136,
      "step": 40900
    },
    {
      "epoch": 12.84,
      "learning_rate": 5.8225696804894626e-05,
      "loss": 0.1103,
      "step": 41000
    },
    {
      "epoch": 12.88,
      "learning_rate": 5.7970768184908224e-05,
      "loss": 0.1122,
      "step": 41100
    },
    {
      "epoch": 12.91,
      "learning_rate": 5.7715839564921815e-05,
      "loss": 0.1127,
      "step": 41200
    },
    {
      "epoch": 12.94,
      "learning_rate": 5.746091094493541e-05,
      "loss": 0.1139,
      "step": 41300
    },
    {
      "epoch": 12.97,
      "learning_rate": 5.720598232494901e-05,
      "loss": 0.1097,
      "step": 41400
    },
    {
      "epoch": 13.0,
      "learning_rate": 5.69510537049626e-05,
      "loss": 0.1139,
      "step": 41500
    },
    {
      "epoch": 13.03,
      "learning_rate": 5.6696125084976205e-05,
      "loss": 0.1045,
      "step": 41600
    },
    {
      "epoch": 13.06,
      "learning_rate": 5.64411964649898e-05,
      "loss": 0.1076,
      "step": 41700
    },
    {
      "epoch": 13.1,
      "learning_rate": 5.618626784500339e-05,
      "loss": 0.1043,
      "step": 41800
    },
    {
      "epoch": 13.13,
      "learning_rate": 5.593133922501699e-05,
      "loss": 0.1051,
      "step": 41900
    },
    {
      "epoch": 13.16,
      "learning_rate": 5.567641060503058e-05,
      "loss": 0.1048,
      "step": 42000
    },
    {
      "epoch": 13.19,
      "learning_rate": 5.5421481985044185e-05,
      "loss": 0.1022,
      "step": 42100
    },
    {
      "epoch": 13.22,
      "learning_rate": 5.516655336505778e-05,
      "loss": 0.1062,
      "step": 42200
    },
    {
      "epoch": 13.25,
      "learning_rate": 5.491162474507137e-05,
      "loss": 0.1059,
      "step": 42300
    },
    {
      "epoch": 13.28,
      "learning_rate": 5.465669612508497e-05,
      "loss": 0.107,
      "step": 42400
    },
    {
      "epoch": 13.31,
      "learning_rate": 5.440176750509856e-05,
      "loss": 0.1045,
      "step": 42500
    },
    {
      "epoch": 13.35,
      "learning_rate": 5.4146838885112166e-05,
      "loss": 0.1085,
      "step": 42600
    },
    {
      "epoch": 13.38,
      "learning_rate": 5.389191026512576e-05,
      "loss": 0.1064,
      "step": 42700
    },
    {
      "epoch": 13.41,
      "learning_rate": 5.3636981645139354e-05,
      "loss": 0.1038,
      "step": 42800
    },
    {
      "epoch": 13.44,
      "learning_rate": 5.338205302515295e-05,
      "loss": 0.1025,
      "step": 42900
    },
    {
      "epoch": 13.47,
      "learning_rate": 5.312712440516655e-05,
      "loss": 0.1024,
      "step": 43000
    },
    {
      "epoch": 13.5,
      "learning_rate": 5.287219578518014e-05,
      "loss": 0.1048,
      "step": 43100
    },
    {
      "epoch": 13.53,
      "learning_rate": 5.2617267165193744e-05,
      "loss": 0.1025,
      "step": 43200
    },
    {
      "epoch": 13.57,
      "learning_rate": 5.236233854520734e-05,
      "loss": 0.1034,
      "step": 43300
    },
    {
      "epoch": 13.6,
      "learning_rate": 5.210740992522093e-05,
      "loss": 0.1069,
      "step": 43400
    },
    {
      "epoch": 13.63,
      "learning_rate": 5.185248130523453e-05,
      "loss": 0.1048,
      "step": 43500
    },
    {
      "epoch": 13.66,
      "learning_rate": 5.159755268524812e-05,
      "loss": 0.1027,
      "step": 43600
    },
    {
      "epoch": 13.69,
      "learning_rate": 5.1342624065261725e-05,
      "loss": 0.1031,
      "step": 43700
    },
    {
      "epoch": 13.72,
      "learning_rate": 5.108769544527532e-05,
      "loss": 0.1052,
      "step": 43800
    },
    {
      "epoch": 13.75,
      "learning_rate": 5.083276682528891e-05,
      "loss": 0.1026,
      "step": 43900
    },
    {
      "epoch": 13.78,
      "learning_rate": 5.057783820530251e-05,
      "loss": 0.1032,
      "step": 44000
    },
    {
      "epoch": 13.82,
      "learning_rate": 5.03229095853161e-05,
      "loss": 0.1023,
      "step": 44100
    },
    {
      "epoch": 13.85,
      "learning_rate": 5.0067980965329705e-05,
      "loss": 0.1029,
      "step": 44200
    },
    {
      "epoch": 13.88,
      "learning_rate": 4.98130523453433e-05,
      "loss": 0.1039,
      "step": 44300
    },
    {
      "epoch": 13.91,
      "learning_rate": 4.9558123725356893e-05,
      "loss": 0.1072,
      "step": 44400
    },
    {
      "epoch": 13.94,
      "learning_rate": 4.930319510537049e-05,
      "loss": 0.1025,
      "step": 44500
    },
    {
      "epoch": 13.97,
      "learning_rate": 4.904826648538409e-05,
      "loss": 0.103,
      "step": 44600
    },
    {
      "epoch": 14.0,
      "learning_rate": 4.879333786539768e-05,
      "loss": 0.1014,
      "step": 44700
    },
    {
      "epoch": 14.04,
      "learning_rate": 4.853840924541128e-05,
      "loss": 0.096,
      "step": 44800
    },
    {
      "epoch": 14.07,
      "learning_rate": 4.828348062542488e-05,
      "loss": 0.0962,
      "step": 44900
    },
    {
      "epoch": 14.1,
      "learning_rate": 4.802855200543847e-05,
      "loss": 0.0954,
      "step": 45000
    },
    {
      "epoch": 14.1,
      "eval_loss": 0.20609818398952484,
      "eval_runtime": 153.2814,
      "eval_samples_per_second": 103.998,
      "eval_steps_per_second": 1.631,
      "step": 45000
    },
    {
      "epoch": 14.13,
      "learning_rate": 4.777362338545207e-05,
      "loss": 0.0963,
      "step": 45100
    },
    {
      "epoch": 14.16,
      "learning_rate": 4.751869476546566e-05,
      "loss": 0.0967,
      "step": 45200
    },
    {
      "epoch": 14.19,
      "learning_rate": 4.7263766145479264e-05,
      "loss": 0.0966,
      "step": 45300
    },
    {
      "epoch": 14.22,
      "learning_rate": 4.700883752549286e-05,
      "loss": 0.0954,
      "step": 45400
    },
    {
      "epoch": 14.25,
      "learning_rate": 4.675390890550645e-05,
      "loss": 0.0947,
      "step": 45500
    },
    {
      "epoch": 14.29,
      "learning_rate": 4.649898028552005e-05,
      "loss": 0.0964,
      "step": 45600
    },
    {
      "epoch": 14.32,
      "learning_rate": 4.624405166553364e-05,
      "loss": 0.0968,
      "step": 45700
    },
    {
      "epoch": 14.35,
      "learning_rate": 4.5989123045547245e-05,
      "loss": 0.0973,
      "step": 45800
    },
    {
      "epoch": 14.38,
      "learning_rate": 4.573419442556084e-05,
      "loss": 0.0944,
      "step": 45900
    },
    {
      "epoch": 14.41,
      "learning_rate": 4.547926580557443e-05,
      "loss": 0.0975,
      "step": 46000
    },
    {
      "epoch": 14.44,
      "learning_rate": 4.522433718558803e-05,
      "loss": 0.098,
      "step": 46100
    },
    {
      "epoch": 14.47,
      "learning_rate": 4.496940856560163e-05,
      "loss": 0.0977,
      "step": 46200
    },
    {
      "epoch": 14.51,
      "learning_rate": 4.471447994561522e-05,
      "loss": 0.0951,
      "step": 46300
    },
    {
      "epoch": 14.54,
      "learning_rate": 4.445955132562882e-05,
      "loss": 0.0971,
      "step": 46400
    },
    {
      "epoch": 14.57,
      "learning_rate": 4.420462270564242e-05,
      "loss": 0.0969,
      "step": 46500
    },
    {
      "epoch": 14.6,
      "learning_rate": 4.394969408565601e-05,
      "loss": 0.0954,
      "step": 46600
    },
    {
      "epoch": 14.63,
      "learning_rate": 4.369476546566961e-05,
      "loss": 0.0929,
      "step": 46700
    },
    {
      "epoch": 14.66,
      "learning_rate": 4.34398368456832e-05,
      "loss": 0.0967,
      "step": 46800
    },
    {
      "epoch": 14.69,
      "learning_rate": 4.31849082256968e-05,
      "loss": 0.0938,
      "step": 46900
    },
    {
      "epoch": 14.72,
      "learning_rate": 4.29299796057104e-05,
      "loss": 0.094,
      "step": 47000
    },
    {
      "epoch": 14.76,
      "learning_rate": 4.267505098572399e-05,
      "loss": 0.0966,
      "step": 47100
    },
    {
      "epoch": 14.79,
      "learning_rate": 4.242012236573759e-05,
      "loss": 0.0951,
      "step": 47200
    },
    {
      "epoch": 14.82,
      "learning_rate": 4.216519374575118e-05,
      "loss": 0.0948,
      "step": 47300
    },
    {
      "epoch": 14.85,
      "learning_rate": 4.1910265125764784e-05,
      "loss": 0.0954,
      "step": 47400
    },
    {
      "epoch": 14.88,
      "learning_rate": 4.165533650577838e-05,
      "loss": 0.0944,
      "step": 47500
    },
    {
      "epoch": 14.91,
      "learning_rate": 4.140040788579197e-05,
      "loss": 0.0953,
      "step": 47600
    },
    {
      "epoch": 14.94,
      "learning_rate": 4.114547926580557e-05,
      "loss": 0.0914,
      "step": 47700
    },
    {
      "epoch": 14.97,
      "learning_rate": 4.089055064581917e-05,
      "loss": 0.0939,
      "step": 47800
    },
    {
      "epoch": 15.01,
      "learning_rate": 4.063562202583276e-05,
      "loss": 0.0921,
      "step": 47900
    },
    {
      "epoch": 15.04,
      "learning_rate": 4.038069340584636e-05,
      "loss": 0.0877,
      "step": 48000
    },
    {
      "epoch": 15.07,
      "learning_rate": 4.012576478585996e-05,
      "loss": 0.087,
      "step": 48100
    },
    {
      "epoch": 15.1,
      "learning_rate": 3.987083616587355e-05,
      "loss": 0.0884,
      "step": 48200
    },
    {
      "epoch": 15.13,
      "learning_rate": 3.961590754588715e-05,
      "loss": 0.0887,
      "step": 48300
    },
    {
      "epoch": 15.16,
      "learning_rate": 3.936097892590074e-05,
      "loss": 0.0891,
      "step": 48400
    },
    {
      "epoch": 15.19,
      "learning_rate": 3.910605030591434e-05,
      "loss": 0.0876,
      "step": 48500
    },
    {
      "epoch": 15.23,
      "learning_rate": 3.885112168592794e-05,
      "loss": 0.0884,
      "step": 48600
    },
    {
      "epoch": 15.26,
      "learning_rate": 3.859619306594153e-05,
      "loss": 0.09,
      "step": 48700
    },
    {
      "epoch": 15.29,
      "learning_rate": 3.834126444595513e-05,
      "loss": 0.0868,
      "step": 48800
    },
    {
      "epoch": 15.32,
      "learning_rate": 3.808633582596872e-05,
      "loss": 0.0903,
      "step": 48900
    },
    {
      "epoch": 15.35,
      "learning_rate": 3.783140720598232e-05,
      "loss": 0.0883,
      "step": 49000
    },
    {
      "epoch": 15.38,
      "learning_rate": 3.757647858599592e-05,
      "loss": 0.0863,
      "step": 49100
    },
    {
      "epoch": 15.41,
      "learning_rate": 3.732154996600951e-05,
      "loss": 0.0875,
      "step": 49200
    },
    {
      "epoch": 15.44,
      "learning_rate": 3.706662134602311e-05,
      "loss": 0.0871,
      "step": 49300
    },
    {
      "epoch": 15.48,
      "learning_rate": 3.6811692726036707e-05,
      "loss": 0.0896,
      "step": 49400
    },
    {
      "epoch": 15.51,
      "learning_rate": 3.6556764106050304e-05,
      "loss": 0.0891,
      "step": 49500
    },
    {
      "epoch": 15.54,
      "learning_rate": 3.6301835486063895e-05,
      "loss": 0.0872,
      "step": 49600
    },
    {
      "epoch": 15.57,
      "learning_rate": 3.60469068660775e-05,
      "loss": 0.0869,
      "step": 49700
    },
    {
      "epoch": 15.6,
      "learning_rate": 3.579197824609109e-05,
      "loss": 0.0849,
      "step": 49800
    },
    {
      "epoch": 15.63,
      "learning_rate": 3.553704962610469e-05,
      "loss": 0.0874,
      "step": 49900
    },
    {
      "epoch": 15.66,
      "learning_rate": 3.5282121006118285e-05,
      "loss": 0.0858,
      "step": 50000
    },
    {
      "epoch": 15.66,
      "eval_loss": 0.19506359100341797,
      "eval_runtime": 164.1483,
      "eval_samples_per_second": 97.113,
      "eval_steps_per_second": 1.523,
      "step": 50000
    },
    {
      "epoch": 15.7,
      "learning_rate": 3.502719238613188e-05,
      "loss": 0.0865,
      "step": 50100
    },
    {
      "epoch": 15.73,
      "learning_rate": 3.477226376614548e-05,
      "loss": 0.0869,
      "step": 50200
    },
    {
      "epoch": 15.76,
      "learning_rate": 3.451733514615907e-05,
      "loss": 0.0848,
      "step": 50300
    },
    {
      "epoch": 15.79,
      "learning_rate": 3.426240652617267e-05,
      "loss": 0.0883,
      "step": 50400
    },
    {
      "epoch": 15.82,
      "learning_rate": 3.4007477906186265e-05,
      "loss": 0.0873,
      "step": 50500
    },
    {
      "epoch": 15.85,
      "learning_rate": 3.375254928619986e-05,
      "loss": 0.0856,
      "step": 50600
    },
    {
      "epoch": 15.88,
      "learning_rate": 3.3497620666213453e-05,
      "loss": 0.0859,
      "step": 50700
    },
    {
      "epoch": 15.91,
      "learning_rate": 3.324269204622705e-05,
      "loss": 0.0853,
      "step": 50800
    },
    {
      "epoch": 15.95,
      "learning_rate": 3.298776342624065e-05,
      "loss": 0.0874,
      "step": 50900
    },
    {
      "epoch": 15.98,
      "learning_rate": 3.2732834806254246e-05,
      "loss": 0.0862,
      "step": 51000
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.2477906186267843e-05,
      "loss": 0.0855,
      "step": 51100
    },
    {
      "epoch": 16.04,
      "learning_rate": 3.2222977566281434e-05,
      "loss": 0.0828,
      "step": 51200
    },
    {
      "epoch": 16.07,
      "learning_rate": 3.196804894629504e-05,
      "loss": 0.0807,
      "step": 51300
    },
    {
      "epoch": 16.1,
      "learning_rate": 3.171312032630863e-05,
      "loss": 0.0816,
      "step": 51400
    },
    {
      "epoch": 16.13,
      "learning_rate": 3.1458191706322227e-05,
      "loss": 0.0806,
      "step": 51500
    },
    {
      "epoch": 16.17,
      "learning_rate": 3.1203263086335824e-05,
      "loss": 0.0809,
      "step": 51600
    },
    {
      "epoch": 16.2,
      "learning_rate": 3.094833446634942e-05,
      "loss": 0.0791,
      "step": 51700
    },
    {
      "epoch": 16.23,
      "learning_rate": 3.069340584636302e-05,
      "loss": 0.0787,
      "step": 51800
    },
    {
      "epoch": 16.26,
      "learning_rate": 3.0438477226376613e-05,
      "loss": 0.0811,
      "step": 51900
    },
    {
      "epoch": 16.29,
      "learning_rate": 3.0183548606390207e-05,
      "loss": 0.0798,
      "step": 52000
    },
    {
      "epoch": 16.32,
      "learning_rate": 2.9928619986403805e-05,
      "loss": 0.0776,
      "step": 52100
    },
    {
      "epoch": 16.35,
      "learning_rate": 2.96736913664174e-05,
      "loss": 0.0817,
      "step": 52200
    },
    {
      "epoch": 16.38,
      "learning_rate": 2.9418762746430996e-05,
      "loss": 0.0806,
      "step": 52300
    },
    {
      "epoch": 16.42,
      "learning_rate": 2.9163834126444594e-05,
      "loss": 0.0806,
      "step": 52400
    },
    {
      "epoch": 16.45,
      "learning_rate": 2.8908905506458188e-05,
      "loss": 0.0775,
      "step": 52500
    },
    {
      "epoch": 16.48,
      "learning_rate": 2.8653976886471785e-05,
      "loss": 0.0799,
      "step": 52600
    },
    {
      "epoch": 16.51,
      "learning_rate": 2.8399048266485383e-05,
      "loss": 0.0814,
      "step": 52700
    },
    {
      "epoch": 16.54,
      "learning_rate": 2.8144119646498977e-05,
      "loss": 0.0828,
      "step": 52800
    },
    {
      "epoch": 16.57,
      "learning_rate": 2.7889191026512574e-05,
      "loss": 0.0792,
      "step": 52900
    },
    {
      "epoch": 16.6,
      "learning_rate": 2.763426240652617e-05,
      "loss": 0.081,
      "step": 53000
    },
    {
      "epoch": 16.64,
      "learning_rate": 2.7379333786539766e-05,
      "loss": 0.08,
      "step": 53100
    },
    {
      "epoch": 16.67,
      "learning_rate": 2.7124405166553363e-05,
      "loss": 0.0793,
      "step": 53200
    },
    {
      "epoch": 16.7,
      "learning_rate": 2.6869476546566958e-05,
      "loss": 0.0793,
      "step": 53300
    },
    {
      "epoch": 16.73,
      "learning_rate": 2.6614547926580555e-05,
      "loss": 0.0793,
      "step": 53400
    },
    {
      "epoch": 16.76,
      "learning_rate": 2.6359619306594152e-05,
      "loss": 0.0793,
      "step": 53500
    },
    {
      "epoch": 16.79,
      "learning_rate": 2.6104690686607747e-05,
      "loss": 0.0776,
      "step": 53600
    },
    {
      "epoch": 16.82,
      "learning_rate": 2.5849762066621344e-05,
      "loss": 0.078,
      "step": 53700
    },
    {
      "epoch": 16.85,
      "learning_rate": 2.5594833446634938e-05,
      "loss": 0.0778,
      "step": 53800
    },
    {
      "epoch": 16.89,
      "learning_rate": 2.5339904826648536e-05,
      "loss": 0.079,
      "step": 53900
    },
    {
      "epoch": 16.92,
      "learning_rate": 2.5084976206662133e-05,
      "loss": 0.079,
      "step": 54000
    },
    {
      "epoch": 16.95,
      "learning_rate": 2.4830047586675727e-05,
      "loss": 0.0776,
      "step": 54100
    },
    {
      "epoch": 16.98,
      "learning_rate": 2.4575118966689328e-05,
      "loss": 0.0798,
      "step": 54200
    },
    {
      "epoch": 17.01,
      "learning_rate": 2.4320190346702922e-05,
      "loss": 0.0782,
      "step": 54300
    },
    {
      "epoch": 17.04,
      "learning_rate": 2.4065261726716516e-05,
      "loss": 0.0768,
      "step": 54400
    },
    {
      "epoch": 17.07,
      "learning_rate": 2.3810333106730114e-05,
      "loss": 0.0723,
      "step": 54500
    },
    {
      "epoch": 17.11,
      "learning_rate": 2.3555404486743708e-05,
      "loss": 0.073,
      "step": 54600
    },
    {
      "epoch": 17.14,
      "learning_rate": 2.330047586675731e-05,
      "loss": 0.0751,
      "step": 54700
    },
    {
      "epoch": 17.17,
      "learning_rate": 2.3045547246770903e-05,
      "loss": 0.0726,
      "step": 54800
    },
    {
      "epoch": 17.2,
      "learning_rate": 2.2790618626784497e-05,
      "loss": 0.0726,
      "step": 54900
    },
    {
      "epoch": 17.23,
      "learning_rate": 2.2535690006798098e-05,
      "loss": 0.0715,
      "step": 55000
    },
    {
      "epoch": 17.23,
      "eval_loss": 0.1879633367061615,
      "eval_runtime": 174.3222,
      "eval_samples_per_second": 91.446,
      "eval_steps_per_second": 1.434,
      "step": 55000
    },
    {
      "epoch": 17.26,
      "learning_rate": 2.2280761386811692e-05,
      "loss": 0.0744,
      "step": 55100
    },
    {
      "epoch": 17.29,
      "learning_rate": 2.2025832766825286e-05,
      "loss": 0.0744,
      "step": 55200
    },
    {
      "epoch": 17.32,
      "learning_rate": 2.1770904146838883e-05,
      "loss": 0.0752,
      "step": 55300
    },
    {
      "epoch": 17.36,
      "learning_rate": 2.1515975526852478e-05,
      "loss": 0.0748,
      "step": 55400
    },
    {
      "epoch": 17.39,
      "learning_rate": 2.126104690686608e-05,
      "loss": 0.0732,
      "step": 55500
    },
    {
      "epoch": 17.42,
      "learning_rate": 2.1006118286879672e-05,
      "loss": 0.0709,
      "step": 55600
    },
    {
      "epoch": 17.45,
      "learning_rate": 2.0751189666893267e-05,
      "loss": 0.0762,
      "step": 55700
    },
    {
      "epoch": 17.48,
      "learning_rate": 2.0496261046906867e-05,
      "loss": 0.0724,
      "step": 55800
    },
    {
      "epoch": 17.51,
      "learning_rate": 2.024133242692046e-05,
      "loss": 0.0722,
      "step": 55900
    },
    {
      "epoch": 17.54,
      "learning_rate": 1.9986403806934056e-05,
      "loss": 0.0718,
      "step": 56000
    },
    {
      "epoch": 17.58,
      "learning_rate": 1.9731475186947653e-05,
      "loss": 0.0712,
      "step": 56100
    },
    {
      "epoch": 17.61,
      "learning_rate": 1.9476546566961247e-05,
      "loss": 0.0722,
      "step": 56200
    },
    {
      "epoch": 17.64,
      "learning_rate": 1.9221617946974848e-05,
      "loss": 0.0729,
      "step": 56300
    },
    {
      "epoch": 17.67,
      "learning_rate": 1.8966689326988442e-05,
      "loss": 0.0697,
      "step": 56400
    },
    {
      "epoch": 17.7,
      "learning_rate": 1.871176070700204e-05,
      "loss": 0.0748,
      "step": 56500
    },
    {
      "epoch": 17.73,
      "learning_rate": 1.8456832087015634e-05,
      "loss": 0.0723,
      "step": 56600
    },
    {
      "epoch": 17.76,
      "learning_rate": 1.820190346702923e-05,
      "loss": 0.0733,
      "step": 56700
    },
    {
      "epoch": 17.79,
      "learning_rate": 1.7946974847042825e-05,
      "loss": 0.071,
      "step": 56800
    },
    {
      "epoch": 17.83,
      "learning_rate": 1.7692046227056423e-05,
      "loss": 0.0707,
      "step": 56900
    },
    {
      "epoch": 17.86,
      "learning_rate": 1.7437117607070017e-05,
      "loss": 0.0706,
      "step": 57000
    },
    {
      "epoch": 17.89,
      "learning_rate": 1.7182188987083614e-05,
      "loss": 0.0719,
      "step": 57100
    },
    {
      "epoch": 17.92,
      "learning_rate": 1.6927260367097212e-05,
      "loss": 0.0708,
      "step": 57200
    },
    {
      "epoch": 17.95,
      "learning_rate": 1.667233174711081e-05,
      "loss": 0.0726,
      "step": 57300
    },
    {
      "epoch": 17.98,
      "learning_rate": 1.6417403127124403e-05,
      "loss": 0.0698,
      "step": 57400
    },
    {
      "epoch": 18.01,
      "learning_rate": 1.6162474507138e-05,
      "loss": 0.0687,
      "step": 57500
    },
    {
      "epoch": 18.05,
      "learning_rate": 1.5907545887151595e-05,
      "loss": 0.0669,
      "step": 57600
    },
    {
      "epoch": 18.08,
      "learning_rate": 1.5652617267165193e-05,
      "loss": 0.0666,
      "step": 57700
    },
    {
      "epoch": 18.11,
      "learning_rate": 1.5397688647178787e-05,
      "loss": 0.0676,
      "step": 57800
    },
    {
      "epoch": 18.14,
      "learning_rate": 1.5142760027192384e-05,
      "loss": 0.0673,
      "step": 57900
    },
    {
      "epoch": 18.17,
      "learning_rate": 1.4887831407205982e-05,
      "loss": 0.0696,
      "step": 58000
    },
    {
      "epoch": 18.2,
      "learning_rate": 1.4632902787219579e-05,
      "loss": 0.0674,
      "step": 58100
    },
    {
      "epoch": 18.23,
      "learning_rate": 1.4377974167233173e-05,
      "loss": 0.0674,
      "step": 58200
    },
    {
      "epoch": 18.26,
      "learning_rate": 1.4123045547246769e-05,
      "loss": 0.0684,
      "step": 58300
    },
    {
      "epoch": 18.3,
      "learning_rate": 1.3868116927260366e-05,
      "loss": 0.0662,
      "step": 58400
    },
    {
      "epoch": 18.33,
      "learning_rate": 1.3613188307273964e-05,
      "loss": 0.069,
      "step": 58500
    },
    {
      "epoch": 18.36,
      "learning_rate": 1.3358259687287558e-05,
      "loss": 0.0671,
      "step": 58600
    },
    {
      "epoch": 18.39,
      "learning_rate": 1.3103331067301154e-05,
      "loss": 0.0673,
      "step": 58700
    },
    {
      "epoch": 18.42,
      "learning_rate": 1.2848402447314751e-05,
      "loss": 0.0651,
      "step": 58800
    },
    {
      "epoch": 18.45,
      "learning_rate": 1.2593473827328349e-05,
      "loss": 0.0666,
      "step": 58900
    },
    {
      "epoch": 18.48,
      "learning_rate": 1.2338545207341943e-05,
      "loss": 0.067,
      "step": 59000
    },
    {
      "epoch": 18.52,
      "learning_rate": 1.2083616587355539e-05,
      "loss": 0.0668,
      "step": 59100
    },
    {
      "epoch": 18.55,
      "learning_rate": 1.1828687967369136e-05,
      "loss": 0.0672,
      "step": 59200
    },
    {
      "epoch": 18.58,
      "learning_rate": 1.1573759347382734e-05,
      "loss": 0.0674,
      "step": 59300
    },
    {
      "epoch": 18.61,
      "learning_rate": 1.1318830727396328e-05,
      "loss": 0.0655,
      "step": 59400
    },
    {
      "epoch": 18.64,
      "learning_rate": 1.1063902107409923e-05,
      "loss": 0.0653,
      "step": 59500
    },
    {
      "epoch": 18.67,
      "learning_rate": 1.0808973487423521e-05,
      "loss": 0.0646,
      "step": 59600
    },
    {
      "epoch": 18.7,
      "learning_rate": 1.0554044867437118e-05,
      "loss": 0.0649,
      "step": 59700
    },
    {
      "epoch": 18.73,
      "learning_rate": 1.0299116247450713e-05,
      "loss": 0.067,
      "step": 59800
    },
    {
      "epoch": 18.77,
      "learning_rate": 1.0044187627464308e-05,
      "loss": 0.0669,
      "step": 59900
    },
    {
      "epoch": 18.8,
      "learning_rate": 9.789259007477906e-06,
      "loss": 0.0634,
      "step": 60000
    },
    {
      "epoch": 18.8,
      "eval_loss": 0.1807943880558014,
      "eval_runtime": 154.0016,
      "eval_samples_per_second": 103.512,
      "eval_steps_per_second": 1.623,
      "step": 60000
    },
    {
      "epoch": 18.83,
      "learning_rate": 9.534330387491503e-06,
      "loss": 0.064,
      "step": 60100
    },
    {
      "epoch": 18.86,
      "learning_rate": 9.279401767505097e-06,
      "loss": 0.066,
      "step": 60200
    },
    {
      "epoch": 18.89,
      "learning_rate": 9.024473147518693e-06,
      "loss": 0.064,
      "step": 60300
    },
    {
      "epoch": 18.92,
      "learning_rate": 8.76954452753229e-06,
      "loss": 0.0645,
      "step": 60400
    },
    {
      "epoch": 18.95,
      "learning_rate": 8.514615907545886e-06,
      "loss": 0.0656,
      "step": 60500
    },
    {
      "epoch": 18.98,
      "learning_rate": 8.259687287559482e-06,
      "loss": 0.0646,
      "step": 60600
    },
    {
      "epoch": 19.02,
      "learning_rate": 8.004758667573078e-06,
      "loss": 0.0624,
      "step": 60700
    },
    {
      "epoch": 19.05,
      "learning_rate": 7.749830047586676e-06,
      "loss": 0.063,
      "step": 60800
    },
    {
      "epoch": 19.08,
      "learning_rate": 7.494901427600271e-06,
      "loss": 0.0613,
      "step": 60900
    },
    {
      "epoch": 19.11,
      "learning_rate": 7.239972807613868e-06,
      "loss": 0.0618,
      "step": 61000
    },
    {
      "epoch": 19.14,
      "learning_rate": 6.985044187627464e-06,
      "loss": 0.0621,
      "step": 61100
    },
    {
      "epoch": 19.17,
      "learning_rate": 6.73011556764106e-06,
      "loss": 0.0627,
      "step": 61200
    },
    {
      "epoch": 19.2,
      "learning_rate": 6.475186947654656e-06,
      "loss": 0.0608,
      "step": 61300
    },
    {
      "epoch": 19.24,
      "learning_rate": 6.220258327668253e-06,
      "loss": 0.0601,
      "step": 61400
    },
    {
      "epoch": 19.27,
      "learning_rate": 5.9653297076818486e-06,
      "loss": 0.0641,
      "step": 61500
    },
    {
      "epoch": 19.3,
      "learning_rate": 5.710401087695445e-06,
      "loss": 0.0603,
      "step": 61600
    },
    {
      "epoch": 19.33,
      "learning_rate": 5.455472467709041e-06,
      "loss": 0.062,
      "step": 61700
    },
    {
      "epoch": 19.36,
      "learning_rate": 5.200543847722638e-06,
      "loss": 0.0616,
      "step": 61800
    },
    {
      "epoch": 19.39,
      "learning_rate": 4.945615227736233e-06,
      "loss": 0.0629,
      "step": 61900
    },
    {
      "epoch": 19.42,
      "learning_rate": 4.69068660774983e-06,
      "loss": 0.0589,
      "step": 62000
    },
    {
      "epoch": 19.45,
      "learning_rate": 4.435757987763426e-06,
      "loss": 0.0624,
      "step": 62100
    },
    {
      "epoch": 19.49,
      "learning_rate": 4.180829367777022e-06,
      "loss": 0.0604,
      "step": 62200
    },
    {
      "epoch": 19.52,
      "learning_rate": 3.925900747790618e-06,
      "loss": 0.0611,
      "step": 62300
    },
    {
      "epoch": 19.55,
      "learning_rate": 3.6709721278042145e-06,
      "loss": 0.0635,
      "step": 62400
    },
    {
      "epoch": 19.58,
      "learning_rate": 3.4160435078178107e-06,
      "loss": 0.0613,
      "step": 62500
    },
    {
      "epoch": 19.61,
      "learning_rate": 3.161114887831407e-06,
      "loss": 0.0598,
      "step": 62600
    },
    {
      "epoch": 19.64,
      "learning_rate": 2.906186267845003e-06,
      "loss": 0.0612,
      "step": 62700
    },
    {
      "epoch": 19.67,
      "learning_rate": 2.6512576478585993e-06,
      "loss": 0.0603,
      "step": 62800
    },
    {
      "epoch": 19.71,
      "learning_rate": 2.3963290278721955e-06,
      "loss": 0.061,
      "step": 62900
    },
    {
      "epoch": 19.74,
      "learning_rate": 2.1414004078857917e-06,
      "loss": 0.0618,
      "step": 63000
    },
    {
      "epoch": 19.77,
      "learning_rate": 1.886471787899388e-06,
      "loss": 0.061,
      "step": 63100
    },
    {
      "epoch": 19.8,
      "learning_rate": 1.6315431679129842e-06,
      "loss": 0.0609,
      "step": 63200
    },
    {
      "epoch": 19.83,
      "learning_rate": 1.3766145479265804e-06,
      "loss": 0.0587,
      "step": 63300
    },
    {
      "epoch": 19.86,
      "learning_rate": 1.1216859279401766e-06,
      "loss": 0.0626,
      "step": 63400
    },
    {
      "epoch": 19.89,
      "learning_rate": 8.667573079537729e-07,
      "loss": 0.0612,
      "step": 63500
    },
    {
      "epoch": 19.92,
      "learning_rate": 6.118286879673691e-07,
      "loss": 0.0599,
      "step": 63600
    },
    {
      "epoch": 19.96,
      "learning_rate": 3.569000679809653e-07,
      "loss": 0.0598,
      "step": 63700
    },
    {
      "epoch": 19.99,
      "learning_rate": 1.0197144799456151e-07,
      "loss": 0.0609,
      "step": 63800
    },
    {
      "epoch": 20.0,
      "step": 63840,
      "total_flos": 5.02934280339456e+19,
      "train_loss": 0.18226892867109232,
      "train_runtime": 183516.2556,
      "train_samples_per_second": 89.055,
      "train_steps_per_second": 0.348
    }
  ],
  "max_steps": 63840,
  "num_train_epochs": 20,
  "total_flos": 5.02934280339456e+19,
  "trial_name": null,
  "trial_params": null
}