{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.977777777777778,
  "eval_steps": 500,
  "global_step": 560,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.044444444444444446,
      "grad_norm": 68.42689661345901,
      "learning_rate": 4.4642857142857147e-07,
      "loss": 2.5092,
      "step": 5
    },
    {
      "epoch": 0.08888888888888889,
      "grad_norm": 39.852986715695174,
      "learning_rate": 8.928571428571429e-07,
      "loss": 2.2004,
      "step": 10
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 15.077781946338447,
      "learning_rate": 1.3392857142857143e-06,
      "loss": 1.969,
      "step": 15
    },
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 11.852084980849574,
      "learning_rate": 1.7857142857142859e-06,
      "loss": 1.8527,
      "step": 20
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 11.008361320264118,
      "learning_rate": 2.2321428571428573e-06,
      "loss": 1.7176,
      "step": 25
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 9.888735985119096,
      "learning_rate": 2.6785714285714285e-06,
      "loss": 1.7062,
      "step": 30
    },
    {
      "epoch": 0.3111111111111111,
      "grad_norm": 8.982680719711558,
      "learning_rate": 3.125e-06,
      "loss": 1.6965,
      "step": 35
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 9.07367234198108,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 1.6517,
      "step": 40
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.846961617300426,
      "learning_rate": 4.017857142857143e-06,
      "loss": 1.6269,
      "step": 45
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 9.400240316956404,
      "learning_rate": 4.464285714285715e-06,
      "loss": 1.5998,
      "step": 50
    },
    {
      "epoch": 0.4888888888888889,
      "grad_norm": 7.866962818759754,
      "learning_rate": 4.910714285714286e-06,
      "loss": 1.5382,
      "step": 55
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 7.642853494273342,
      "learning_rate": 5.357142857142857e-06,
      "loss": 1.4911,
      "step": 60
    },
    {
      "epoch": 0.5777777777777777,
      "grad_norm": 8.734944006479568,
      "learning_rate": 5.8035714285714295e-06,
      "loss": 1.5042,
      "step": 65
    },
    {
      "epoch": 0.6222222222222222,
      "grad_norm": 8.19033857608755,
      "learning_rate": 6.25e-06,
      "loss": 1.4525,
      "step": 70
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 6.9887239984624365,
      "learning_rate": 6.696428571428571e-06,
      "loss": 1.4513,
      "step": 75
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 7.62197813685645,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 1.5256,
      "step": 80
    },
    {
      "epoch": 0.7555555555555555,
      "grad_norm": 7.251988396487642,
      "learning_rate": 7.589285714285714e-06,
      "loss": 1.4834,
      "step": 85
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.127135113535223,
      "learning_rate": 8.035714285714286e-06,
      "loss": 1.4678,
      "step": 90
    },
    {
      "epoch": 0.8444444444444444,
      "grad_norm": 7.505433979097814,
      "learning_rate": 8.482142857142858e-06,
      "loss": 1.475,
      "step": 95
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 7.279504299445843,
      "learning_rate": 8.92857142857143e-06,
      "loss": 1.4639,
      "step": 100
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 7.199608771033321,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.4675,
      "step": 105
    },
    {
      "epoch": 0.9777777777777777,
      "grad_norm": 7.3489571757156,
      "learning_rate": 9.821428571428573e-06,
      "loss": 1.4208,
      "step": 110
    },
    {
      "epoch": 1.0222222222222221,
      "grad_norm": 6.762935414285647,
      "learning_rate": 9.933035714285715e-06,
      "loss": 1.2787,
      "step": 115
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 8.219383140650084,
      "learning_rate": 9.821428571428573e-06,
      "loss": 1.094,
      "step": 120
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 6.841085798098678,
      "learning_rate": 9.70982142857143e-06,
      "loss": 1.1307,
      "step": 125
    },
    {
      "epoch": 1.1555555555555554,
      "grad_norm": 8.447700779081304,
      "learning_rate": 9.598214285714287e-06,
      "loss": 1.0377,
      "step": 130
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.258686167858292,
      "learning_rate": 9.486607142857144e-06,
      "loss": 1.032,
      "step": 135
    },
    {
      "epoch": 1.2444444444444445,
      "grad_norm": 7.453545107272504,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.093,
      "step": 140
    },
    {
      "epoch": 1.2888888888888888,
      "grad_norm": 7.799606910604757,
      "learning_rate": 9.263392857142858e-06,
      "loss": 1.056,
      "step": 145
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 7.846429832692078,
      "learning_rate": 9.151785714285715e-06,
      "loss": 1.1157,
      "step": 150
    },
    {
      "epoch": 1.3777777777777778,
      "grad_norm": 9.496069639808155,
      "learning_rate": 9.040178571428572e-06,
      "loss": 1.0328,
      "step": 155
    },
    {
      "epoch": 1.4222222222222223,
      "grad_norm": 8.556283686076158,
      "learning_rate": 8.92857142857143e-06,
      "loss": 1.069,
      "step": 160
    },
    {
      "epoch": 1.4666666666666668,
      "grad_norm": 7.975408464194922,
      "learning_rate": 8.816964285714286e-06,
      "loss": 1.1391,
      "step": 165
    },
    {
      "epoch": 1.511111111111111,
      "grad_norm": 7.721363741996411,
      "learning_rate": 8.705357142857143e-06,
      "loss": 1.0796,
      "step": 170
    },
    {
      "epoch": 1.5555555555555556,
      "grad_norm": 8.450482541723963,
      "learning_rate": 8.59375e-06,
      "loss": 1.0616,
      "step": 175
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.741813046749222,
      "learning_rate": 8.482142857142858e-06,
      "loss": 1.1004,
      "step": 180
    },
    {
      "epoch": 1.6444444444444444,
      "grad_norm": 6.913255006160549,
      "learning_rate": 8.370535714285715e-06,
      "loss": 1.0657,
      "step": 185
    },
    {
      "epoch": 1.6888888888888889,
      "grad_norm": 7.6645305589476465,
      "learning_rate": 8.258928571428572e-06,
      "loss": 1.0792,
      "step": 190
    },
    {
      "epoch": 1.7333333333333334,
      "grad_norm": 7.674126321557261,
      "learning_rate": 8.147321428571429e-06,
      "loss": 0.989,
      "step": 195
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 7.458761675629579,
      "learning_rate": 8.035714285714286e-06,
      "loss": 1.1224,
      "step": 200
    },
    {
      "epoch": 1.8222222222222222,
      "grad_norm": 7.455498289317789,
      "learning_rate": 7.924107142857143e-06,
      "loss": 1.1243,
      "step": 205
    },
    {
      "epoch": 1.8666666666666667,
      "grad_norm": 7.352544941891832,
      "learning_rate": 7.8125e-06,
      "loss": 1.0668,
      "step": 210
    },
    {
      "epoch": 1.911111111111111,
      "grad_norm": 6.780778867894781,
      "learning_rate": 7.700892857142857e-06,
      "loss": 1.0972,
      "step": 215
    },
    {
      "epoch": 1.9555555555555557,
      "grad_norm": 6.690390099595921,
      "learning_rate": 7.589285714285714e-06,
      "loss": 1.0639,
      "step": 220
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.284060043318805,
      "learning_rate": 7.4776785714285714e-06,
      "loss": 1.0809,
      "step": 225
    },
    {
      "epoch": 2.0444444444444443,
      "grad_norm": 5.941163451968988,
      "learning_rate": 7.366071428571429e-06,
      "loss": 0.4991,
      "step": 230
    },
    {
      "epoch": 2.088888888888889,
      "grad_norm": 8.149920395070575,
      "learning_rate": 7.2544642857142865e-06,
      "loss": 0.4372,
      "step": 235
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 7.728965175949651,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.4173,
      "step": 240
    },
    {
      "epoch": 2.1777777777777776,
      "grad_norm": 9.376621531683403,
      "learning_rate": 7.031250000000001e-06,
      "loss": 0.4196,
      "step": 245
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 7.7606102118489,
      "learning_rate": 6.919642857142858e-06,
      "loss": 0.4215,
      "step": 250
    },
    {
      "epoch": 2.2666666666666666,
      "grad_norm": 6.453067601785949,
      "learning_rate": 6.808035714285715e-06,
      "loss": 0.4119,
      "step": 255
    },
    {
      "epoch": 2.311111111111111,
      "grad_norm": 7.569136345322084,
      "learning_rate": 6.696428571428571e-06,
      "loss": 0.4505,
      "step": 260
    },
    {
      "epoch": 2.3555555555555556,
      "grad_norm": 6.7543830296249965,
      "learning_rate": 6.58482142857143e-06,
      "loss": 0.458,
      "step": 265
    },
    {
      "epoch": 2.4,
      "grad_norm": 6.716260935521558,
      "learning_rate": 6.473214285714287e-06,
      "loss": 0.4483,
      "step": 270
    },
    {
      "epoch": 2.4444444444444446,
      "grad_norm": 7.389437845033128,
      "learning_rate": 6.361607142857143e-06,
      "loss": 0.4432,
      "step": 275
    },
    {
      "epoch": 2.488888888888889,
      "grad_norm": 8.50887044840266,
      "learning_rate": 6.25e-06,
      "loss": 0.4649,
      "step": 280
    },
    {
      "epoch": 2.533333333333333,
      "grad_norm": 7.22261594860323,
      "learning_rate": 6.138392857142857e-06,
      "loss": 0.4275,
      "step": 285
    },
    {
      "epoch": 2.5777777777777775,
      "grad_norm": 7.051771827978584,
      "learning_rate": 6.0267857142857145e-06,
      "loss": 0.4661,
      "step": 290
    },
    {
      "epoch": 2.6222222222222222,
      "grad_norm": 7.497859635927052,
      "learning_rate": 5.9151785714285716e-06,
      "loss": 0.4452,
      "step": 295
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 7.235064313385639,
      "learning_rate": 5.8035714285714295e-06,
      "loss": 0.4331,
      "step": 300
    },
    {
      "epoch": 2.7111111111111112,
      "grad_norm": 6.6195886161559185,
      "learning_rate": 5.691964285714287e-06,
      "loss": 0.4273,
      "step": 305
    },
    {
      "epoch": 2.7555555555555555,
      "grad_norm": 7.070431454924444,
      "learning_rate": 5.580357142857144e-06,
      "loss": 0.4364,
      "step": 310
    },
    {
      "epoch": 2.8,
      "grad_norm": 7.812072711402095,
      "learning_rate": 5.468750000000001e-06,
      "loss": 0.4403,
      "step": 315
    },
    {
      "epoch": 2.8444444444444446,
      "grad_norm": 6.759441784972788,
      "learning_rate": 5.357142857142857e-06,
      "loss": 0.4536,
      "step": 320
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 7.870197506940289,
      "learning_rate": 5.245535714285714e-06,
      "loss": 0.4195,
      "step": 325
    },
    {
      "epoch": 2.9333333333333336,
      "grad_norm": 7.541654513991995,
      "learning_rate": 5.133928571428571e-06,
      "loss": 0.4648,
      "step": 330
    },
    {
      "epoch": 2.977777777777778,
      "grad_norm": 7.926146834244662,
      "learning_rate": 5.022321428571429e-06,
      "loss": 0.4582,
      "step": 335
    },
    {
      "epoch": 3.022222222222222,
      "grad_norm": 4.453965632553679,
      "learning_rate": 4.910714285714286e-06,
      "loss": 0.289,
      "step": 340
    },
    {
      "epoch": 3.066666666666667,
      "grad_norm": 6.784041239449323,
      "learning_rate": 4.799107142857143e-06,
      "loss": 0.1349,
      "step": 345
    },
    {
      "epoch": 3.111111111111111,
      "grad_norm": 7.647875580621476,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.1359,
      "step": 350
    },
    {
      "epoch": 3.1555555555555554,
      "grad_norm": 5.5089833457735375,
      "learning_rate": 4.5758928571428575e-06,
      "loss": 0.1085,
      "step": 355
    },
    {
      "epoch": 3.2,
      "grad_norm": 6.4339922144692165,
      "learning_rate": 4.464285714285715e-06,
      "loss": 0.104,
      "step": 360
    },
    {
      "epoch": 3.2444444444444445,
      "grad_norm": 8.100735915125973,
      "learning_rate": 4.352678571428572e-06,
      "loss": 0.1101,
      "step": 365
    },
    {
      "epoch": 3.2888888888888888,
      "grad_norm": 4.454541477160946,
      "learning_rate": 4.241071428571429e-06,
      "loss": 0.1072,
      "step": 370
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 6.187319521825981,
      "learning_rate": 4.129464285714286e-06,
      "loss": 0.1014,
      "step": 375
    },
    {
      "epoch": 3.3777777777777778,
      "grad_norm": 6.0751996137011925,
      "learning_rate": 4.017857142857143e-06,
      "loss": 0.1015,
      "step": 380
    },
    {
      "epoch": 3.422222222222222,
      "grad_norm": 6.6360058797828465,
      "learning_rate": 3.90625e-06,
      "loss": 0.1077,
      "step": 385
    },
    {
      "epoch": 3.466666666666667,
      "grad_norm": 4.293118386066451,
      "learning_rate": 3.794642857142857e-06,
      "loss": 0.114,
      "step": 390
    },
    {
      "epoch": 3.511111111111111,
      "grad_norm": 4.5572671703263445,
      "learning_rate": 3.6830357142857147e-06,
      "loss": 0.095,
      "step": 395
    },
    {
      "epoch": 3.5555555555555554,
      "grad_norm": 5.50613372953253,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.1091,
      "step": 400
    },
    {
      "epoch": 3.6,
      "grad_norm": 5.7181058894629295,
      "learning_rate": 3.459821428571429e-06,
      "loss": 0.1181,
      "step": 405
    },
    {
      "epoch": 3.6444444444444444,
      "grad_norm": 5.694946111698654,
      "learning_rate": 3.3482142857142855e-06,
      "loss": 0.1158,
      "step": 410
    },
    {
      "epoch": 3.688888888888889,
      "grad_norm": 5.026393936447245,
      "learning_rate": 3.2366071428571435e-06,
      "loss": 0.1063,
      "step": 415
    },
    {
      "epoch": 3.7333333333333334,
      "grad_norm": 7.019156177995795,
      "learning_rate": 3.125e-06,
      "loss": 0.1174,
      "step": 420
    },
    {
      "epoch": 3.7777777777777777,
      "grad_norm": 5.016747035658572,
      "learning_rate": 3.0133928571428572e-06,
      "loss": 0.1055,
      "step": 425
    },
    {
      "epoch": 3.822222222222222,
      "grad_norm": 5.132052583298932,
      "learning_rate": 2.9017857142857148e-06,
      "loss": 0.0933,
      "step": 430
    },
    {
      "epoch": 3.8666666666666667,
      "grad_norm": 5.372165796897467,
      "learning_rate": 2.790178571428572e-06,
      "loss": 0.1248,
      "step": 435
    },
    {
      "epoch": 3.911111111111111,
      "grad_norm": 4.715117973605432,
      "learning_rate": 2.6785714285714285e-06,
      "loss": 0.1102,
      "step": 440
    },
    {
      "epoch": 3.9555555555555557,
      "grad_norm": 4.916282393060736,
      "learning_rate": 2.5669642857142856e-06,
      "loss": 0.104,
      "step": 445
    },
    {
      "epoch": 4.0,
      "grad_norm": 6.123917973410412,
      "learning_rate": 2.455357142857143e-06,
      "loss": 0.1047,
      "step": 450
    },
    {
      "epoch": 4.044444444444444,
      "grad_norm": 2.5957913806483313,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 0.0304,
      "step": 455
    },
    {
      "epoch": 4.088888888888889,
      "grad_norm": 2.2154980762706606,
      "learning_rate": 2.2321428571428573e-06,
      "loss": 0.0254,
      "step": 460
    },
    {
      "epoch": 4.133333333333334,
      "grad_norm": 2.5096141897285063,
      "learning_rate": 2.1205357142857144e-06,
      "loss": 0.0248,
      "step": 465
    },
    {
      "epoch": 4.177777777777778,
      "grad_norm": 3.007491230873463,
      "learning_rate": 2.0089285714285715e-06,
      "loss": 0.0199,
      "step": 470
    },
    {
      "epoch": 4.222222222222222,
      "grad_norm": 2.8021254930599095,
      "learning_rate": 1.8973214285714286e-06,
      "loss": 0.0161,
      "step": 475
    },
    {
      "epoch": 4.266666666666667,
      "grad_norm": 1.8835347940753981,
      "learning_rate": 1.7857142857142859e-06,
      "loss": 0.0191,
      "step": 480
    },
    {
      "epoch": 4.311111111111111,
      "grad_norm": 1.9254641935429138,
      "learning_rate": 1.6741071428571428e-06,
      "loss": 0.0213,
      "step": 485
    },
    {
      "epoch": 4.355555555555555,
      "grad_norm": 1.7141458028949819,
      "learning_rate": 1.5625e-06,
      "loss": 0.0163,
      "step": 490
    },
    {
      "epoch": 4.4,
      "grad_norm": 2.524460084340419,
      "learning_rate": 1.4508928571428574e-06,
      "loss": 0.0207,
      "step": 495
    },
    {
      "epoch": 4.444444444444445,
      "grad_norm": 2.1648304929853537,
      "learning_rate": 1.3392857142857143e-06,
      "loss": 0.0207,
      "step": 500
    },
    {
      "epoch": 4.488888888888889,
      "grad_norm": 2.5253785393748323,
      "learning_rate": 1.2276785714285716e-06,
      "loss": 0.0162,
      "step": 505
    },
    {
      "epoch": 4.533333333333333,
      "grad_norm": 2.7358236942844627,
      "learning_rate": 1.1160714285714287e-06,
      "loss": 0.0234,
      "step": 510
    },
    {
      "epoch": 4.5777777777777775,
      "grad_norm": 3.554265924420303,
      "learning_rate": 1.0044642857142857e-06,
      "loss": 0.0272,
      "step": 515
    },
    {
      "epoch": 4.622222222222222,
      "grad_norm": 3.2376281928270196,
      "learning_rate": 8.928571428571429e-07,
      "loss": 0.0188,
      "step": 520
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 3.2437020818641007,
      "learning_rate": 7.8125e-07,
      "loss": 0.0193,
      "step": 525
    },
    {
      "epoch": 4.711111111111111,
      "grad_norm": 2.783913184597399,
      "learning_rate": 6.696428571428571e-07,
      "loss": 0.0182,
      "step": 530
    },
    {
      "epoch": 4.7555555555555555,
      "grad_norm": 3.4123689520524048,
      "learning_rate": 5.580357142857143e-07,
      "loss": 0.0161,
      "step": 535
    },
    {
      "epoch": 4.8,
      "grad_norm": 2.93458629220715,
      "learning_rate": 4.4642857142857147e-07,
      "loss": 0.0159,
      "step": 540
    },
    {
      "epoch": 4.844444444444444,
      "grad_norm": 4.51902248806245,
      "learning_rate": 3.3482142857142856e-07,
      "loss": 0.0154,
      "step": 545
    },
    {
      "epoch": 4.888888888888889,
      "grad_norm": 1.8223723169737671,
      "learning_rate": 2.2321428571428574e-07,
      "loss": 0.0204,
      "step": 550
    },
    {
      "epoch": 4.933333333333334,
      "grad_norm": 4.048861560252665,
      "learning_rate": 1.1160714285714287e-07,
      "loss": 0.0199,
      "step": 555
    },
    {
      "epoch": 4.977777777777778,
      "grad_norm": 2.322780533534595,
      "learning_rate": 0.0,
      "loss": 0.0182,
      "step": 560
    },
    {
      "epoch": 4.977777777777778,
      "step": 560,
      "total_flos": 3034126417920.0,
      "train_loss": 0.6623806708891476,
      "train_runtime": 7462.8051,
      "train_samples_per_second": 2.412,
      "train_steps_per_second": 0.075
    }
  ],
  "logging_steps": 5,
  "max_steps": 560,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3034126417920.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}