{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.4193548387096775,
  "eval_steps": 500,
  "global_step": 1200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 1.2860161066055298,
      "learning_rate": 4.999442829949762e-05,
      "loss": 1.1916,
      "step": 10
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6525003910064697,
      "learning_rate": 4.997771568149818e-05,
      "loss": 1.0982,
      "step": 20
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.212003231048584,
      "learning_rate": 4.9949869595417876e-05,
      "loss": 1.021,
      "step": 30
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9332901239395142,
      "learning_rate": 4.9910902453260824e-05,
      "loss": 1.0453,
      "step": 40
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3961759805679321,
      "learning_rate": 4.986083162408669e-05,
      "loss": 0.9987,
      "step": 50
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5509816408157349,
      "learning_rate": 4.979967942626858e-05,
      "loss": 0.9685,
      "step": 60
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.23104727268219,
      "learning_rate": 4.972747311754501e-05,
      "loss": 0.9597,
      "step": 70
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7351208925247192,
      "learning_rate": 4.964424488287009e-05,
      "loss": 0.9614,
      "step": 80
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6495784521102905,
      "learning_rate": 4.955003182006761e-05,
      "loss": 0.9191,
      "step": 90
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1316168308258057,
      "learning_rate": 4.944487592329509e-05,
      "loss": 1.012,
      "step": 100
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.793336033821106,
      "learning_rate": 4.9328824064325566e-05,
      "loss": 0.9564,
      "step": 110
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4563814401626587,
      "learning_rate": 4.920192797165511e-05,
      "loss": 0.9283,
      "step": 120
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6153773069381714,
      "learning_rate": 4.906424420744559e-05,
      "loss": 0.9326,
      "step": 130
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5536580085754395,
      "learning_rate": 4.891583414231287e-05,
      "loss": 0.9359,
      "step": 140
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4636590480804443,
      "learning_rate": 4.875676392797168e-05,
      "loss": 0.9579,
      "step": 150
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6544209718704224,
      "learning_rate": 4.858710446774951e-05,
      "loss": 0.9231,
      "step": 160
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6398296356201172,
      "learning_rate": 4.840693138498231e-05,
      "loss": 0.9009,
      "step": 170
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8886735439300537,
      "learning_rate": 4.821632498930656e-05,
      "loss": 0.9199,
      "step": 180
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6928956508636475,
      "learning_rate": 4.801537024086229e-05,
      "loss": 0.9679,
      "step": 190
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1802239418029785,
      "learning_rate": 4.780415671242334e-05,
      "loss": 0.9218,
      "step": 200
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9001857042312622,
      "learning_rate": 4.7582778549471494e-05,
      "loss": 0.9036,
      "step": 210
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.371631622314453,
      "learning_rate": 4.735133442823252e-05,
      "loss": 0.932,
      "step": 220
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7190943956375122,
      "learning_rate": 4.710992751169252e-05,
      "loss": 0.9329,
      "step": 230
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9145939350128174,
      "learning_rate": 4.685866540361456e-05,
      "loss": 0.9496,
      "step": 240
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0224311351776123,
      "learning_rate": 4.659766010057574e-05,
      "loss": 0.9074,
      "step": 250
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7447651624679565,
      "learning_rate": 4.6327027942046286e-05,
      "loss": 0.875,
      "step": 260
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.789925217628479,
      "learning_rate": 4.604688955853293e-05,
      "loss": 0.8906,
      "step": 270
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6785054206848145,
      "learning_rate": 4.5757369817809415e-05,
      "loss": 0.89,
      "step": 280
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9952877759933472,
      "learning_rate": 4.5458597769258535e-05,
      "loss": 0.9216,
      "step": 290
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9971449375152588,
      "learning_rate": 4.515070658635013e-05,
      "loss": 0.9329,
      "step": 300
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9367233514785767,
      "learning_rate": 4.4833833507280884e-05,
      "loss": 0.8769,
      "step": 310
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0975091457366943,
      "learning_rate": 4.45081197738023e-05,
      "loss": 0.9395,
      "step": 320
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8488638401031494,
      "learning_rate": 4.417371056826417e-05,
      "loss": 0.9189,
      "step": 330
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9082151651382446,
      "learning_rate": 4.383075494890159e-05,
      "loss": 0.9306,
      "step": 340
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.708274483680725,
      "learning_rate": 4.347940578339428e-05,
      "loss": 0.8993,
      "step": 350
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8434147834777832,
      "learning_rate": 4.3119819680728e-05,
      "loss": 0.9212,
      "step": 360
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0493478775024414,
      "learning_rate": 4.2752156921388264e-05,
      "loss": 0.8769,
      "step": 370
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8880027532577515,
      "learning_rate": 4.2376581385917547e-05,
      "loss": 0.8963,
      "step": 380
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.563737630844116,
      "learning_rate": 4.199326048186782e-05,
      "loss": 0.9929,
      "step": 390
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.850777506828308,
      "learning_rate": 4.160236506918098e-05,
      "loss": 0.917,
      "step": 400
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.723869800567627,
      "learning_rate": 4.1204069384030396e-05,
      "loss": 0.9051,
      "step": 410
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1148617267608643,
      "learning_rate": 4.07985509611576e-05,
      "loss": 0.9369,
      "step": 420
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.888234257698059,
      "learning_rate": 4.038599055473863e-05,
      "loss": 0.9504,
      "step": 430
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6350065469741821,
      "learning_rate": 3.9966572057815373e-05,
      "loss": 0.8794,
      "step": 440
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.113657236099243,
      "learning_rate": 3.9540482420327845e-05,
      "loss": 0.8842,
      "step": 450
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.951561450958252,
      "learning_rate": 3.910791156578382e-05,
      "loss": 0.8631,
      "step": 460
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7464616298675537,
      "learning_rate": 3.866905230660309e-05,
      "loss": 0.8699,
      "step": 470
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.176771640777588,
      "learning_rate": 3.822410025817406e-05,
      "loss": 0.9188,
      "step": 480
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0469777584075928,
      "learning_rate": 3.777325375166088e-05,
      "loss": 0.9286,
      "step": 490
    },
    {
      "epoch": 1.01,
      "grad_norm": 2.167973518371582,
      "learning_rate": 3.731671374560007e-05,
      "loss": 0.909,
      "step": 500
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.1665143966674805,
      "learning_rate": 3.6854683736326125e-05,
      "loss": 0.8855,
      "step": 510
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.2790310382843018,
      "learning_rate": 3.638736966726585e-05,
      "loss": 0.9811,
      "step": 520
    },
    {
      "epoch": 1.07,
      "grad_norm": 2.1944940090179443,
      "learning_rate": 3.59149798371419e-05,
      "loss": 0.8948,
      "step": 530
    },
    {
      "epoch": 1.09,
      "grad_norm": 2.0177252292633057,
      "learning_rate": 3.543772480712658e-05,
      "loss": 0.8636,
      "step": 540
    },
    {
      "epoch": 1.11,
      "grad_norm": 2.1394903659820557,
      "learning_rate": 3.4955817306987124e-05,
      "loss": 0.8467,
      "step": 550
    },
    {
      "epoch": 1.13,
      "grad_norm": 2.131467342376709,
      "learning_rate": 3.44694721402644e-05,
      "loss": 0.894,
      "step": 560
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.935081958770752,
      "learning_rate": 3.397890608852718e-05,
      "loss": 0.868,
      "step": 570
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.752773642539978,
      "learning_rate": 3.348433781474481e-05,
      "loss": 0.8807,
      "step": 580
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7388436794281006,
      "learning_rate": 3.298598776582126e-05,
      "loss": 0.9313,
      "step": 590
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.0265705585479736,
      "learning_rate": 3.2484078074333954e-05,
      "loss": 0.9366,
      "step": 600
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8241621255874634,
      "learning_rate": 3.197883245952131e-05,
      "loss": 0.8784,
      "step": 610
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.0975425243377686,
      "learning_rate": 3.147047612756302e-05,
      "loss": 0.8345,
      "step": 620
    },
    {
      "epoch": 1.27,
      "grad_norm": 2.1606051921844482,
      "learning_rate": 3.095923567119748e-05,
      "loss": 0.8452,
      "step": 630
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8565340042114258,
      "learning_rate": 3.0445338968721287e-05,
      "loss": 0.8517,
      "step": 640
    },
    {
      "epoch": 1.31,
      "grad_norm": 2.326995372772217,
      "learning_rate": 2.992901508241569e-05,
      "loss": 0.8721,
      "step": 650
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.2724409103393555,
      "learning_rate": 2.9410494156445216e-05,
      "loss": 0.864,
      "step": 660
    },
    {
      "epoch": 1.35,
      "grad_norm": 2.4907262325286865,
      "learning_rate": 2.889000731427416e-05,
      "loss": 0.8825,
      "step": 670
    },
    {
      "epoch": 1.37,
      "grad_norm": 2.073660135269165,
      "learning_rate": 2.836778655564653e-05,
      "loss": 0.8617,
      "step": 680
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.923628807067871,
      "learning_rate": 2.7844064653175378e-05,
      "loss": 0.8619,
      "step": 690
    },
    {
      "epoch": 1.41,
      "grad_norm": 2.1420040130615234,
      "learning_rate": 2.7319075048587666e-05,
      "loss": 0.8525,
      "step": 700
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.0220742225646973,
      "learning_rate": 2.6793051748670896e-05,
      "loss": 0.86,
      "step": 710
    },
    {
      "epoch": 1.45,
      "grad_norm": 2.0513052940368652,
      "learning_rate": 2.6266229220967818e-05,
      "loss": 0.9214,
      "step": 720
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.164041042327881,
      "learning_rate": 2.5738842289265814e-05,
      "loss": 0.888,
      "step": 730
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8874469995498657,
      "learning_rate": 2.5211126028927464e-05,
      "loss": 0.9022,
      "step": 740
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6967768669128418,
      "learning_rate": 2.4683315662109e-05,
      "loss": 0.8912,
      "step": 750
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8921899795532227,
      "learning_rate": 2.4155646452913296e-05,
      "loss": 0.8499,
      "step": 760
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.830534815788269,
      "learning_rate": 2.36283536025242e-05,
      "loss": 0.8306,
      "step": 770
    },
    {
      "epoch": 1.57,
      "grad_norm": 2.6946194171905518,
      "learning_rate": 2.310167214436885e-05,
      "loss": 0.8753,
      "step": 780
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9234410524368286,
      "learning_rate": 2.2575836839354848e-05,
      "loss": 0.8758,
      "step": 790
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.620464563369751,
      "learning_rate": 2.2051082071228854e-05,
      "loss": 0.9053,
      "step": 800
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.201439619064331,
      "learning_rate": 2.152764174210328e-05,
      "loss": 0.8763,
      "step": 810
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.2076492309570312,
      "learning_rate": 2.1005749168197696e-05,
      "loss": 0.859,
      "step": 820
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.1520049571990967,
      "learning_rate": 2.0485636975841415e-05,
      "loss": 0.8849,
      "step": 830
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9997478723526,
      "learning_rate": 1.9967536997783494e-05,
      "loss": 0.8669,
      "step": 840
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.3713417053222656,
      "learning_rate": 1.9451680169856602e-05,
      "loss": 0.9124,
      "step": 850
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.459920883178711,
      "learning_rate": 1.89382964280405e-05,
      "loss": 0.879,
      "step": 860
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.571826219558716,
      "learning_rate": 1.842761460597138e-05,
      "loss": 0.8446,
      "step": 870
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.335891008377075,
      "learning_rate": 1.79198623329424e-05,
      "loss": 0.855,
      "step": 880
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8321956396102905,
      "learning_rate": 1.7415265932441136e-05,
      "loss": 0.868,
      "step": 890
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7981292009353638,
      "learning_rate": 1.6914050321269047e-05,
      "loss": 0.8442,
      "step": 900
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.0625789165496826,
      "learning_rate": 1.641643890928794e-05,
      "loss": 0.8335,
      "step": 910
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.1844565868377686,
      "learning_rate": 1.5922653499838137e-05,
      "loss": 0.8314,
      "step": 920
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.1349494457244873,
      "learning_rate": 1.5432914190872757e-05,
      "loss": 0.8059,
      "step": 930
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.5662639141082764,
      "learning_rate": 1.4947439276852104e-05,
      "loss": 0.8678,
      "step": 940
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.8356733322143555,
      "learning_rate": 1.4466445151441965e-05,
      "loss": 0.8859,
      "step": 950
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.7699570655822754,
      "learning_rate": 1.399014621105914e-05,
      "loss": 0.8691,
      "step": 960
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.013852834701538,
      "learning_rate": 1.3518754759307289e-05,
      "loss": 0.8456,
      "step": 970
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.38482403755188,
      "learning_rate": 1.3052480912345482e-05,
      "loss": 0.8193,
      "step": 980
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.1879611015319824,
      "learning_rate": 1.2591532505231906e-05,
      "loss": 0.8785,
      "step": 990
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.2595455646514893,
      "learning_rate": 1.2136114999284288e-05,
      "loss": 0.8564,
      "step": 1000
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.1374199390411377,
      "learning_rate": 1.1686431390498354e-05,
      "loss": 0.8316,
      "step": 1010
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.3616902828216553,
      "learning_rate": 1.1242682119065218e-05,
      "loss": 0.8698,
      "step": 1020
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.4855775833129883,
      "learning_rate": 1.0805064980027973e-05,
      "loss": 0.8445,
      "step": 1030
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0951740741729736,
      "learning_rate": 1.0373775035117305e-05,
      "loss": 0.8423,
      "step": 1040
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.2897915840148926,
      "learning_rate": 9.949004525805423e-06,
      "loss": 0.8492,
      "step": 1050
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.129755973815918,
      "learning_rate": 9.530942787617137e-06,
      "loss": 0.8388,
      "step": 1060
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.337714195251465,
      "learning_rate": 9.11977616573618e-06,
      "loss": 0.8437,
      "step": 1070
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.128257989883423,
      "learning_rate": 8.715687931944449e-06,
      "loss": 0.8095,
      "step": 1080
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.5019583702087402,
      "learning_rate": 8.318858202931153e-06,
      "loss": 0.8328,
      "step": 1090
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.477393388748169,
      "learning_rate": 7.929463860008355e-06,
      "loss": 0.8561,
      "step": 1100
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.189155340194702,
      "learning_rate": 7.547678470268526e-06,
      "loss": 0.8551,
      "step": 1110
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.133364200592041,
      "learning_rate": 7.173672209219495e-06,
      "loss": 0.85,
      "step": 1120
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.8646955490112305,
      "learning_rate": 6.807611784931067e-06,
      "loss": 0.8625,
      "step": 1130
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1333656311035156,
      "learning_rate": 6.449660363727236e-06,
      "loss": 0.8542,
      "step": 1140
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.6336278915405273,
      "learning_rate": 6.099977497457062e-06,
      "loss": 0.8474,
      "step": 1150
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3681249618530273,
      "learning_rate": 5.758719052376693e-06,
      "loss": 0.8534,
      "step": 1160
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.100318193435669,
      "learning_rate": 5.4260371396741175e-06,
      "loss": 0.8259,
      "step": 1170
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.3494303226470947,
      "learning_rate": 5.10208004766774e-06,
      "loss": 0.8064,
      "step": 1180
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2627310752868652,
      "learning_rate": 4.786992175708949e-06,
      "loss": 0.8188,
      "step": 1190
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1743149757385254,
      "learning_rate": 4.480913969818098e-06,
      "loss": 0.8474,
      "step": 1200
    }
  ],
  "logging_steps": 10,
  "max_steps": 1488,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "total_flos": 1.1912432826109133e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}