{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.7711654268508679,
  "eval_steps": 500,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.017711654268508677,
      "grad_norm": 1475.1734619140625,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 183.6145,
      "step": 100
    },
    {
      "epoch": 0.035423308537017355,
      "grad_norm": 3148.26611328125,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 278.8031,
      "step": 200
    },
    {
      "epoch": 0.053134962805526036,
      "grad_norm": 802.7176513671875,
      "learning_rate": 3e-06,
      "loss": 284.8548,
      "step": 300
    },
    {
      "epoch": 0.07084661707403471,
      "grad_norm": 374.4765625,
      "learning_rate": 4.000000000000001e-06,
      "loss": 342.4145,
      "step": 400
    },
    {
      "epoch": 0.08855827134254339,
      "grad_norm": 203.01939392089844,
      "learning_rate": 5e-06,
      "loss": 145.8225,
      "step": 500
    },
    {
      "epoch": 0.10626992561105207,
      "grad_norm": 701.1649780273438,
      "learning_rate": 6e-06,
      "loss": 160.1669,
      "step": 600
    },
    {
      "epoch": 0.12398157987956075,
      "grad_norm": 835.7929077148438,
      "learning_rate": 7e-06,
      "loss": 186.2168,
      "step": 700
    },
    {
      "epoch": 0.14169323414806942,
      "grad_norm": 2253.17529296875,
      "learning_rate": 8.000000000000001e-06,
      "loss": 216.2165,
      "step": 800
    },
    {
      "epoch": 0.1594048884165781,
      "grad_norm": 461.9115295410156,
      "learning_rate": 9e-06,
      "loss": 180.0223,
      "step": 900
    },
    {
      "epoch": 0.17711654268508678,
      "grad_norm": 765.1061401367188,
      "learning_rate": 1e-05,
      "loss": 83.2702,
      "step": 1000
    },
    {
      "epoch": 0.19482819695359546,
      "grad_norm": 584.3681030273438,
      "learning_rate": 9.99695413509548e-06,
      "loss": 205.8748,
      "step": 1100
    },
    {
      "epoch": 0.21253985122210414,
      "grad_norm": 294.4954528808594,
      "learning_rate": 9.987820251299121e-06,
      "loss": 183.6346,
      "step": 1200
    },
    {
      "epoch": 0.23025150549061282,
      "grad_norm": 1621.862548828125,
      "learning_rate": 9.972609476841368e-06,
      "loss": 186.5347,
      "step": 1300
    },
    {
      "epoch": 0.2479631597591215,
      "grad_norm": 297.78009033203125,
      "learning_rate": 9.951340343707852e-06,
      "loss": 177.5844,
      "step": 1400
    },
    {
      "epoch": 0.26567481402763016,
      "grad_norm": 5771.119140625,
      "learning_rate": 9.924038765061042e-06,
      "loss": 149.4232,
      "step": 1500
    },
    {
      "epoch": 0.28338646829613884,
      "grad_norm": 330.3929138183594,
      "learning_rate": 9.890738003669029e-06,
      "loss": 138.1254,
      "step": 1600
    },
    {
      "epoch": 0.3010981225646475,
      "grad_norm": 188.89022827148438,
      "learning_rate": 9.851478631379982e-06,
      "loss": 110.1628,
      "step": 1700
    },
    {
      "epoch": 0.3188097768331562,
      "grad_norm": 570.4500122070312,
      "learning_rate": 9.806308479691595e-06,
      "loss": 86.4707,
      "step": 1800
    },
    {
      "epoch": 0.3365214311016649,
      "grad_norm": 373.85772705078125,
      "learning_rate": 9.755282581475769e-06,
      "loss": 196.61,
      "step": 1900
    },
    {
      "epoch": 0.35423308537017356,
      "grad_norm": 453.0614929199219,
      "learning_rate": 9.698463103929542e-06,
      "loss": 90.848,
      "step": 2000
    },
    {
      "epoch": 0.37194473963868224,
      "grad_norm": 1280.0504150390625,
      "learning_rate": 9.635919272833938e-06,
      "loss": 212.6628,
      "step": 2100
    },
    {
      "epoch": 0.3896563939071909,
      "grad_norm": 5861.28125,
      "learning_rate": 9.567727288213005e-06,
      "loss": 130.1471,
      "step": 2200
    },
    {
      "epoch": 0.4073680481756996,
      "grad_norm": 1938.48486328125,
      "learning_rate": 9.493970231495836e-06,
      "loss": 185.4775,
      "step": 2300
    },
    {
      "epoch": 0.4250797024442083,
      "grad_norm": 699.9638061523438,
      "learning_rate": 9.414737964294636e-06,
      "loss": 132.9917,
      "step": 2400
    },
    {
      "epoch": 0.44279135671271697,
      "grad_norm": 868.6508178710938,
      "learning_rate": 9.330127018922195e-06,
      "loss": 143.6872,
      "step": 2500
    },
    {
      "epoch": 0.46050301098122565,
      "grad_norm": 147.92196655273438,
      "learning_rate": 9.24024048078213e-06,
      "loss": 128.0953,
      "step": 2600
    },
    {
      "epoch": 0.4782146652497343,
      "grad_norm": 189.57456970214844,
      "learning_rate": 9.145187862775208e-06,
      "loss": 135.5999,
      "step": 2700
    },
    {
      "epoch": 0.495926319518243,
      "grad_norm": 361.0294189453125,
      "learning_rate": 9.045084971874738e-06,
      "loss": 197.3174,
      "step": 2800
    },
    {
      "epoch": 0.5136379737867517,
      "grad_norm": 555.4054565429688,
      "learning_rate": 8.94005376803361e-06,
      "loss": 125.2118,
      "step": 2900
    },
    {
      "epoch": 0.5313496280552603,
      "grad_norm": 1461.2225341796875,
      "learning_rate": 8.83022221559489e-06,
      "loss": 148.4647,
      "step": 3000
    },
    {
      "epoch": 0.549061282323769,
      "grad_norm": 244.3065948486328,
      "learning_rate": 8.715724127386971e-06,
      "loss": 89.2689,
      "step": 3100
    },
    {
      "epoch": 0.5667729365922777,
      "grad_norm": 1078.7288818359375,
      "learning_rate": 8.596699001693257e-06,
      "loss": 157.0174,
      "step": 3200
    },
    {
      "epoch": 0.5844845908607864,
      "grad_norm": 274.42144775390625,
      "learning_rate": 8.473291852294986e-06,
      "loss": 141.7283,
      "step": 3300
    },
    {
      "epoch": 0.602196245129295,
      "grad_norm": 377.5411376953125,
      "learning_rate": 8.345653031794292e-06,
      "loss": 65.1926,
      "step": 3400
    },
    {
      "epoch": 0.6199078993978038,
      "grad_norm": 553.4656982421875,
      "learning_rate": 8.213938048432697e-06,
      "loss": 99.7509,
      "step": 3500
    },
    {
      "epoch": 0.6376195536663124,
      "grad_norm": 679.3318481445312,
      "learning_rate": 8.078307376628292e-06,
      "loss": 156.2424,
      "step": 3600
    },
    {
      "epoch": 0.6553312079348211,
      "grad_norm": 1766.850830078125,
      "learning_rate": 7.938926261462366e-06,
      "loss": 153.1544,
      "step": 3700
    },
    {
      "epoch": 0.6730428622033298,
      "grad_norm": 1371.9398193359375,
      "learning_rate": 7.795964517353734e-06,
      "loss": 72.5237,
      "step": 3800
    },
    {
      "epoch": 0.6907545164718385,
      "grad_norm": 780.5692138671875,
      "learning_rate": 7.649596321166024e-06,
      "loss": 138.9397,
      "step": 3900
    },
    {
      "epoch": 0.7084661707403471,
      "grad_norm": 1025.0152587890625,
      "learning_rate": 7.500000000000001e-06,
      "loss": 148.5116,
      "step": 4000
    },
    {
      "epoch": 0.7261778250088559,
      "grad_norm": 589.7218627929688,
      "learning_rate": 7.347357813929455e-06,
      "loss": 76.0393,
      "step": 4100
    },
    {
      "epoch": 0.7438894792773645,
      "grad_norm": 27607.578125,
      "learning_rate": 7.191855733945388e-06,
      "loss": 104.6024,
      "step": 4200
    },
    {
      "epoch": 0.7616011335458732,
      "grad_norm": 7136.6484375,
      "learning_rate": 7.033683215379002e-06,
      "loss": 98.471,
      "step": 4300
    },
    {
      "epoch": 0.7793127878143818,
      "grad_norm": 1065.7740478515625,
      "learning_rate": 6.873032967079562e-06,
      "loss": 91.2521,
      "step": 4400
    },
    {
      "epoch": 0.7970244420828906,
      "grad_norm": 123.02791595458984,
      "learning_rate": 6.710100716628345e-06,
      "loss": 131.9392,
      "step": 4500
    },
    {
      "epoch": 0.8147360963513992,
      "grad_norm": 3401.07275390625,
      "learning_rate": 6.545084971874738e-06,
      "loss": 156.1362,
      "step": 4600
    },
    {
      "epoch": 0.832447750619908,
      "grad_norm": 678.4093627929688,
      "learning_rate": 6.378186779084996e-06,
      "loss": 131.1671,
      "step": 4700
    },
    {
      "epoch": 0.8501594048884166,
      "grad_norm": 482.71087646484375,
      "learning_rate": 6.209609477998339e-06,
      "loss": 75.2061,
      "step": 4800
    },
    {
      "epoch": 0.8678710591569253,
      "grad_norm": 1085.9967041015625,
      "learning_rate": 6.039558454088796e-06,
      "loss": 83.2878,
      "step": 4900
    },
    {
      "epoch": 0.8855827134254339,
      "grad_norm": 1806.950927734375,
      "learning_rate": 5.8682408883346535e-06,
      "loss": 89.1659,
      "step": 5000
    },
    {
      "epoch": 0.9032943676939427,
      "grad_norm": 192.0183868408203,
      "learning_rate": 5.695865504800328e-06,
      "loss": 145.4525,
      "step": 5100
    },
    {
      "epoch": 0.9210060219624513,
      "grad_norm": 360.1806945800781,
      "learning_rate": 5.522642316338268e-06,
      "loss": 160.2776,
      "step": 5200
    },
    {
      "epoch": 0.9387176762309599,
      "grad_norm": 963.9231567382812,
      "learning_rate": 5.348782368720627e-06,
      "loss": 117.9117,
      "step": 5300
    },
    {
      "epoch": 0.9564293304994687,
      "grad_norm": 1065.7203369140625,
      "learning_rate": 5.174497483512506e-06,
      "loss": 195.1825,
      "step": 5400
    },
    {
      "epoch": 0.9741409847679773,
      "grad_norm": 148.5713348388672,
      "learning_rate": 5e-06,
      "loss": 113.0304,
      "step": 5500
    },
    {
      "epoch": 0.991852639036486,
      "grad_norm": 324.159912109375,
      "learning_rate": 4.825502516487497e-06,
      "loss": 131.4805,
      "step": 5600
    },
    {
      "epoch": 1.0,
      "eval_loss": 267.4697265625,
      "eval_runtime": 32.2662,
      "eval_samples_per_second": 77.79,
      "eval_steps_per_second": 19.463,
      "step": 5646
    },
    {
      "epoch": 1.0095642933049946,
      "grad_norm": 310.0011901855469,
      "learning_rate": 4.651217631279374e-06,
      "loss": 129.1599,
      "step": 5700
    },
    {
      "epoch": 1.0272759475735034,
      "grad_norm": 307.11077880859375,
      "learning_rate": 4.477357683661734e-06,
      "loss": 95.957,
      "step": 5800
    },
    {
      "epoch": 1.0449876018420121,
      "grad_norm": 568.346923828125,
      "learning_rate": 4.304134495199675e-06,
      "loss": 78.8842,
      "step": 5900
    },
    {
      "epoch": 1.0626992561105206,
      "grad_norm": 383.0439453125,
      "learning_rate": 4.131759111665349e-06,
      "loss": 68.9493,
      "step": 6000
    },
    {
      "epoch": 1.0804109103790294,
      "grad_norm": 723.8114624023438,
      "learning_rate": 3.960441545911205e-06,
      "loss": 89.5963,
      "step": 6100
    },
    {
      "epoch": 1.098122564647538,
      "grad_norm": 494.0089111328125,
      "learning_rate": 3.790390522001662e-06,
      "loss": 73.7578,
      "step": 6200
    },
    {
      "epoch": 1.1158342189160468,
      "grad_norm": 847.5864868164062,
      "learning_rate": 3.6218132209150047e-06,
      "loss": 94.2538,
      "step": 6300
    },
    {
      "epoch": 1.1335458731845554,
      "grad_norm": 617.9949340820312,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 136.4913,
      "step": 6400
    },
    {
      "epoch": 1.151257527453064,
      "grad_norm": 14180.2412109375,
      "learning_rate": 3.289899283371657e-06,
      "loss": 117.7802,
      "step": 6500
    },
    {
      "epoch": 1.1689691817215728,
      "grad_norm": 636.8546752929688,
      "learning_rate": 3.12696703292044e-06,
      "loss": 96.9991,
      "step": 6600
    },
    {
      "epoch": 1.1866808359900816,
      "grad_norm": 500.31427001953125,
      "learning_rate": 2.966316784621e-06,
      "loss": 72.8566,
      "step": 6700
    },
    {
      "epoch": 1.20439249025859,
      "grad_norm": 222.04693603515625,
      "learning_rate": 2.8081442660546126e-06,
      "loss": 124.0254,
      "step": 6800
    },
    {
      "epoch": 1.2221041445270988,
      "grad_norm": 467.826904296875,
      "learning_rate": 2.6526421860705474e-06,
      "loss": 77.8931,
      "step": 6900
    },
    {
      "epoch": 1.2398157987956075,
      "grad_norm": 240.4927520751953,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 148.6505,
      "step": 7000
    },
    {
      "epoch": 1.257527453064116,
      "grad_norm": 523.0436401367188,
      "learning_rate": 2.3504036788339763e-06,
      "loss": 78.4411,
      "step": 7100
    },
    {
      "epoch": 1.2752391073326248,
      "grad_norm": 614.3518676757812,
      "learning_rate": 2.204035482646267e-06,
      "loss": 97.6218,
      "step": 7200
    },
    {
      "epoch": 1.2929507616011335,
      "grad_norm": 846.6361694335938,
      "learning_rate": 2.061073738537635e-06,
      "loss": 106.9313,
      "step": 7300
    },
    {
      "epoch": 1.3106624158696423,
      "grad_norm": 363.7395324707031,
      "learning_rate": 1.9216926233717087e-06,
      "loss": 84.9967,
      "step": 7400
    },
    {
      "epoch": 1.328374070138151,
      "grad_norm": 1934.39501953125,
      "learning_rate": 1.7860619515673034e-06,
      "loss": 159.7926,
      "step": 7500
    },
    {
      "epoch": 1.3460857244066595,
      "grad_norm": 822.3735961914062,
      "learning_rate": 1.6543469682057105e-06,
      "loss": 72.5312,
      "step": 7600
    },
    {
      "epoch": 1.3637973786751683,
      "grad_norm": 146.9804229736328,
      "learning_rate": 1.5267081477050132e-06,
      "loss": 90.417,
      "step": 7700
    },
    {
      "epoch": 1.381509032943677,
      "grad_norm": 571.5712890625,
      "learning_rate": 1.4033009983067454e-06,
      "loss": 90.5052,
      "step": 7800
    },
    {
      "epoch": 1.3992206872121855,
      "grad_norm": 167.052001953125,
      "learning_rate": 1.2842758726130283e-06,
      "loss": 90.8252,
      "step": 7900
    },
    {
      "epoch": 1.4169323414806942,
      "grad_norm": 1918.76953125,
      "learning_rate": 1.1697777844051105e-06,
      "loss": 132.3367,
      "step": 8000
    },
    {
      "epoch": 1.434643995749203,
      "grad_norm": 1841.577392578125,
      "learning_rate": 1.0599462319663906e-06,
      "loss": 111.2179,
      "step": 8100
    },
    {
      "epoch": 1.4523556500177117,
      "grad_norm": 525.4321899414062,
      "learning_rate": 9.549150281252633e-07,
      "loss": 96.6257,
      "step": 8200
    },
    {
      "epoch": 1.4700673042862205,
      "grad_norm": 330.7159729003906,
      "learning_rate": 8.54812137224792e-07,
      "loss": 93.4043,
      "step": 8300
    },
    {
      "epoch": 1.487778958554729,
      "grad_norm": 1182.404541015625,
      "learning_rate": 7.597595192178702e-07,
      "loss": 107.5777,
      "step": 8400
    },
    {
      "epoch": 1.5054906128232377,
      "grad_norm": 405.40997314453125,
      "learning_rate": 6.698729810778065e-07,
      "loss": 60.6317,
      "step": 8500
    },
    {
      "epoch": 1.5232022670917464,
      "grad_norm": 759.5929565429688,
      "learning_rate": 5.852620357053651e-07,
      "loss": 56.3915,
      "step": 8600
    },
    {
      "epoch": 1.540913921360255,
      "grad_norm": 856.4190673828125,
      "learning_rate": 5.06029768504166e-07,
      "loss": 50.7948,
      "step": 8700
    },
    {
      "epoch": 1.5586255756287637,
      "grad_norm": 387.5693054199219,
      "learning_rate": 4.322727117869951e-07,
      "loss": 77.5953,
      "step": 8800
    },
    {
      "epoch": 1.5763372298972724,
      "grad_norm": 475.8940124511719,
      "learning_rate": 3.6408072716606346e-07,
      "loss": 113.5439,
      "step": 8900
    },
    {
      "epoch": 1.594048884165781,
      "grad_norm": 338.134765625,
      "learning_rate": 3.015368960704584e-07,
      "loss": 97.4924,
      "step": 9000
    },
    {
      "epoch": 1.61176053843429,
      "grad_norm": 334.3011169433594,
      "learning_rate": 2.447174185242324e-07,
      "loss": 103.1555,
      "step": 9100
    },
    {
      "epoch": 1.6294721927027984,
      "grad_norm": 262.0125427246094,
      "learning_rate": 1.9369152030840553e-07,
      "loss": 116.0878,
      "step": 9200
    },
    {
      "epoch": 1.6471838469713072,
      "grad_norm": 267.3951721191406,
      "learning_rate": 1.4852136862001766e-07,
      "loss": 118.6488,
      "step": 9300
    },
    {
      "epoch": 1.664895501239816,
      "grad_norm": 86.43798065185547,
      "learning_rate": 1.0926199633097156e-07,
      "loss": 82.3905,
      "step": 9400
    },
    {
      "epoch": 1.6826071555083244,
      "grad_norm": 334.0291748046875,
      "learning_rate": 7.59612349389599e-08,
      "loss": 127.891,
      "step": 9500
    },
    {
      "epoch": 1.7003188097768331,
      "grad_norm": 3675.962646484375,
      "learning_rate": 4.865965629214819e-08,
      "loss": 125.8603,
      "step": 9600
    },
    {
      "epoch": 1.7180304640453419,
      "grad_norm": 107.04967498779297,
      "learning_rate": 2.7390523158633552e-08,
      "loss": 85.8442,
      "step": 9700
    },
    {
      "epoch": 1.7357421183138504,
      "grad_norm": 296.780029296875,
      "learning_rate": 1.2179748700879013e-08,
      "loss": 93.238,
      "step": 9800
    },
    {
      "epoch": 1.7534537725823593,
      "grad_norm": 577.6653442382812,
      "learning_rate": 3.0458649045211897e-09,
      "loss": 78.3574,
      "step": 9900
    },
    {
      "epoch": 1.7711654268508679,
      "grad_norm": 604.1082763671875,
      "learning_rate": 0.0,
      "loss": 159.3366,
      "step": 10000
    }
  ],
  "logging_steps": 100,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}