{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "grad_norm": 0.3057188391685486,
      "learning_rate": 1.8e-06,
      "loss": 1.1212,
      "step": 10
    },
    {
      "grad_norm": 0.1444840282201767,
      "learning_rate": 3.8e-06,
      "loss": 1.1317,
      "step": 20
    },
    {
      "grad_norm": 0.3467788100242615,
      "learning_rate": 5.8e-06,
      "loss": 1.1308,
      "step": 30
    },
    {
      "grad_norm": 0.17815707623958588,
      "learning_rate": 7.8e-06,
      "loss": 1.1381,
      "step": 40
    },
    {
      "grad_norm": 0.15318413078784943,
      "learning_rate": 9.800000000000001e-06,
      "loss": 1.1142,
      "step": 50
    },
    {
      "grad_norm": 0.13940805196762085,
      "learning_rate": 1.18e-05,
      "loss": 1.136,
      "step": 60
    },
    {
      "grad_norm": 0.1543205827474594,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 1.1136,
      "step": 70
    },
    {
      "grad_norm": 0.5840712189674377,
      "learning_rate": 1.58e-05,
      "loss": 1.1304,
      "step": 80
    },
    {
      "grad_norm": 3.530726671218872,
      "learning_rate": 1.78e-05,
      "loss": 1.0871,
      "step": 90
    },
    {
      "grad_norm": 3.2765843868255615,
      "learning_rate": 1.9800000000000004e-05,
      "loss": 0.9507,
      "step": 100
    },
    {
      "grad_norm": 3.01340651512146,
      "learning_rate": 2.18e-05,
      "loss": 0.7881,
      "step": 110
    },
    {
      "grad_norm": 2.948817253112793,
      "learning_rate": 2.38e-05,
      "loss": 0.6435,
      "step": 120
    },
    {
      "grad_norm": 3.288931369781494,
      "learning_rate": 2.58e-05,
      "loss": 0.5627,
      "step": 130
    },
    {
      "grad_norm": 2.4159297943115234,
      "learning_rate": 2.7800000000000005e-05,
      "loss": 0.5449,
      "step": 140
    },
    {
      "grad_norm": 2.800011396408081,
      "learning_rate": 2.98e-05,
      "loss": 0.5317,
      "step": 150
    },
    {
      "grad_norm": 3.540977716445923,
      "learning_rate": 3.18e-05,
      "loss": 0.5521,
      "step": 160
    },
    {
      "grad_norm": 3.3192760944366455,
      "learning_rate": 3.38e-05,
      "loss": 0.4971,
      "step": 170
    },
    {
      "grad_norm": 2.114555597305298,
      "learning_rate": 3.58e-05,
      "loss": 0.5149,
      "step": 180
    },
    {
      "grad_norm": 2.2151782512664795,
      "learning_rate": 3.7800000000000004e-05,
      "loss": 0.4706,
      "step": 190
    },
    {
      "grad_norm": 1.8451696634292603,
      "learning_rate": 3.9800000000000005e-05,
      "loss": 0.4748,
      "step": 200
    },
    {
      "grad_norm": 1.6983399391174316,
      "learning_rate": 4.18e-05,
      "loss": 0.4774,
      "step": 210
    },
    {
      "grad_norm": 1.5557819604873657,
      "learning_rate": 4.38e-05,
      "loss": 0.5016,
      "step": 220
    },
    {
      "grad_norm": 1.8644224405288696,
      "learning_rate": 4.58e-05,
      "loss": 0.5003,
      "step": 230
    },
    {
      "grad_norm": 1.5129348039627075,
      "learning_rate": 4.78e-05,
      "loss": 0.4621,
      "step": 240
    },
    {
      "grad_norm": 1.366440773010254,
      "learning_rate": 4.9800000000000004e-05,
      "loss": 0.4807,
      "step": 250
    },
    {
      "grad_norm": 1.8761663436889648,
      "learning_rate": 5.1800000000000005e-05,
      "loss": 0.4791,
      "step": 260
    },
    {
      "grad_norm": 1.301880955696106,
      "learning_rate": 5.380000000000001e-05,
      "loss": 0.4948,
      "step": 270
    },
    {
      "grad_norm": 1.84480619430542,
      "learning_rate": 5.580000000000001e-05,
      "loss": 0.4679,
      "step": 280
    },
    {
      "grad_norm": 1.4715685844421387,
      "learning_rate": 5.7799999999999995e-05,
      "loss": 0.4906,
      "step": 290
    },
    {
      "grad_norm": 1.0838373899459839,
      "learning_rate": 5.9800000000000003e-05,
      "loss": 0.4657,
      "step": 300
    },
    {
      "grad_norm": 1.1891741752624512,
      "learning_rate": 6.18e-05,
      "loss": 0.5023,
      "step": 310
    },
    {
      "grad_norm": 1.5453864336013794,
      "learning_rate": 6.38e-05,
      "loss": 0.454,
      "step": 320
    },
    {
      "grad_norm": 1.4289326667785645,
      "learning_rate": 6.58e-05,
      "loss": 0.4769,
      "step": 330
    },
    {
      "grad_norm": 1.1819672584533691,
      "learning_rate": 6.780000000000001e-05,
      "loss": 0.459,
      "step": 340
    },
    {
      "grad_norm": 0.919075608253479,
      "learning_rate": 6.98e-05,
      "loss": 0.4863,
      "step": 350
    },
    {
      "grad_norm": 1.5617965459823608,
      "learning_rate": 7.18e-05,
      "loss": 0.4636,
      "step": 360
    },
    {
      "grad_norm": 1.2063156366348267,
      "learning_rate": 7.38e-05,
      "loss": 0.4627,
      "step": 370
    },
    {
      "grad_norm": 1.6134825944900513,
      "learning_rate": 7.58e-05,
      "loss": 0.4669,
      "step": 380
    },
    {
      "grad_norm": 1.0930230617523193,
      "learning_rate": 7.780000000000001e-05,
      "loss": 0.4433,
      "step": 390
    },
    {
      "grad_norm": 0.9772116541862488,
      "learning_rate": 7.98e-05,
      "loss": 0.4597,
      "step": 400
    },
    {
      "grad_norm": 0.8216540813446045,
      "learning_rate": 8.18e-05,
      "loss": 0.471,
      "step": 410
    },
    {
      "grad_norm": 0.9532214999198914,
      "learning_rate": 8.38e-05,
      "loss": 0.4726,
      "step": 420
    },
    {
      "grad_norm": 0.7853254079818726,
      "learning_rate": 8.58e-05,
      "loss": 0.4675,
      "step": 430
    },
    {
      "grad_norm": 1.3092178106307983,
      "learning_rate": 8.78e-05,
      "loss": 0.438,
      "step": 440
    },
    {
      "grad_norm": 1.250414490699768,
      "learning_rate": 8.98e-05,
      "loss": 0.4541,
      "step": 450
    },
    {
      "grad_norm": 0.9576596617698669,
      "learning_rate": 9.180000000000001e-05,
      "loss": 0.4612,
      "step": 460
    },
    {
      "grad_norm": 0.7267900109291077,
      "learning_rate": 9.38e-05,
      "loss": 0.4627,
      "step": 470
    },
    {
      "grad_norm": 0.7806752324104309,
      "learning_rate": 9.58e-05,
      "loss": 0.4498,
      "step": 480
    },
    {
      "grad_norm": 0.627242922782898,
      "learning_rate": 9.78e-05,
      "loss": 0.4516,
      "step": 490
    },
    {
      "grad_norm": 1.1385681629180908,
      "learning_rate": 9.98e-05,
      "loss": 0.4367,
      "step": 500
    },
    {
      "grad_norm": 0.8676849603652954,
      "learning_rate": 9.9999778549206e-05,
      "loss": 0.4595,
      "step": 510
    },
    {
      "grad_norm": 0.7239831686019897,
      "learning_rate": 9.999901304280685e-05,
      "loss": 0.441,
      "step": 520
    },
    {
      "grad_norm": 0.6182336211204529,
      "learning_rate": 9.999770075521164e-05,
      "loss": 0.434,
      "step": 530
    },
    {
      "grad_norm": 0.8173977732658386,
      "learning_rate": 9.99958417007713e-05,
      "loss": 0.4388,
      "step": 540
    },
    {
      "grad_norm": 1.1430753469467163,
      "learning_rate": 9.999343589981615e-05,
      "loss": 0.4384,
      "step": 550
    },
    {
      "grad_norm": 0.7719456553459167,
      "learning_rate": 9.999048337865568e-05,
      "loss": 0.4289,
      "step": 560
    },
    {
      "grad_norm": 0.9089356660842896,
      "learning_rate": 9.998698416957815e-05,
      "loss": 0.4448,
      "step": 570
    },
    {
      "grad_norm": 0.7337244749069214,
      "learning_rate": 9.998293831085037e-05,
      "loss": 0.4502,
      "step": 580
    },
    {
      "grad_norm": 0.8413501381874084,
      "learning_rate": 9.997834584671719e-05,
      "loss": 0.4807,
      "step": 590
    },
    {
      "grad_norm": 0.8633179068565369,
      "learning_rate": 9.997320682740107e-05,
      "loss": 0.4633,
      "step": 600
    },
    {
      "grad_norm": 0.7958359122276306,
      "learning_rate": 9.996752130910149e-05,
      "loss": 0.4819,
      "step": 610
    },
    {
      "grad_norm": 0.557285726070404,
      "learning_rate": 9.99612893539944e-05,
      "loss": 0.4386,
      "step": 620
    },
    {
      "grad_norm": 0.7920674085617065,
      "learning_rate": 9.995451103023144e-05,
      "loss": 0.4099,
      "step": 630
    },
    {
      "grad_norm": 0.5818178653717041,
      "learning_rate": 9.994718641193928e-05,
      "loss": 0.4182,
      "step": 640
    },
    {
      "grad_norm": 0.6897380352020264,
      "learning_rate": 9.993931557921874e-05,
      "loss": 0.4651,
      "step": 650
    },
    {
      "grad_norm": 0.7744024991989136,
      "learning_rate": 9.993089861814402e-05,
      "loss": 0.4453,
      "step": 660
    },
    {
      "grad_norm": 0.7095301747322083,
      "learning_rate": 9.992193562076166e-05,
      "loss": 0.4626,
      "step": 670
    },
    {
      "grad_norm": 0.620883047580719,
      "learning_rate": 9.991242668508954e-05,
      "loss": 0.4739,
      "step": 680
    },
    {
      "grad_norm": 0.520000696182251,
      "learning_rate": 9.990237191511587e-05,
      "loss": 0.4403,
      "step": 690
    },
    {
      "grad_norm": 0.6390187740325928,
      "learning_rate": 9.989177142079802e-05,
      "loss": 0.4377,
      "step": 700
    },
    {
      "grad_norm": 0.6145787239074707,
      "learning_rate": 9.988062531806126e-05,
      "loss": 0.4408,
      "step": 710
    },
    {
      "grad_norm": 0.61893230676651,
      "learning_rate": 9.986893372879762e-05,
      "loss": 0.4508,
      "step": 720
    },
    {
      "grad_norm": 0.5453311204910278,
      "learning_rate": 9.985669678086443e-05,
      "loss": 0.425,
      "step": 730
    },
    {
      "grad_norm": 0.5797358155250549,
      "learning_rate": 9.984391460808298e-05,
      "loss": 0.3795,
      "step": 740
    },
    {
      "grad_norm": 0.5589283108711243,
      "learning_rate": 9.983058735023709e-05,
      "loss": 0.4258,
      "step": 750
    },
    {
      "grad_norm": 0.6372461915016174,
      "learning_rate": 9.98167151530715e-05,
      "loss": 0.3975,
      "step": 760
    },
    {
      "grad_norm": 0.6028401255607605,
      "learning_rate": 9.980229816829034e-05,
      "loss": 0.4428,
      "step": 770
    },
    {
      "grad_norm": 0.5026216506958008,
      "learning_rate": 9.978733655355544e-05,
      "loss": 0.4238,
      "step": 780
    },
    {
      "grad_norm": 0.6490947604179382,
      "learning_rate": 9.977183047248464e-05,
      "loss": 0.4475,
      "step": 790
    },
    {
      "grad_norm": 0.7175934910774231,
      "learning_rate": 9.975578009464992e-05,
      "loss": 0.4349,
      "step": 800
    },
    {
      "grad_norm": 0.6480410099029541,
      "learning_rate": 9.97391855955757e-05,
      "loss": 0.4471,
      "step": 810
    },
    {
      "grad_norm": 0.5713026523590088,
      "learning_rate": 9.972204715673669e-05,
      "loss": 0.4306,
      "step": 820
    },
    {
      "grad_norm": 0.7045426368713379,
      "learning_rate": 9.970436496555617e-05,
      "loss": 0.4282,
      "step": 830
    },
    {
      "grad_norm": 0.7591593265533447,
      "learning_rate": 9.968613921540373e-05,
      "loss": 0.4318,
      "step": 840
    },
    {
      "grad_norm": 0.5254964232444763,
      "learning_rate": 9.966737010559326e-05,
      "loss": 0.4542,
      "step": 850
    },
    {
      "grad_norm": 0.553036630153656,
      "learning_rate": 9.964805784138072e-05,
      "loss": 0.4544,
      "step": 860
    },
    {
      "grad_norm": 0.5461755990982056,
      "learning_rate": 9.962820263396195e-05,
      "loss": 0.43,
      "step": 870
    },
    {
      "grad_norm": 0.4797649383544922,
      "learning_rate": 9.960780470047033e-05,
      "loss": 0.422,
      "step": 880
    },
    {
      "grad_norm": 0.5344374775886536,
      "learning_rate": 9.958686426397437e-05,
      "loss": 0.4031,
      "step": 890
    },
    {
      "grad_norm": 0.5567972660064697,
      "learning_rate": 9.956538155347534e-05,
      "loss": 0.4369,
      "step": 900
    },
    {
      "grad_norm": 0.6837541460990906,
      "learning_rate": 9.95433568039047e-05,
      "loss": 0.4006,
      "step": 910
    },
    {
      "grad_norm": 0.60051029920578,
      "learning_rate": 9.952079025612162e-05,
      "loss": 0.4399,
      "step": 920
    },
    {
      "grad_norm": 0.5134548544883728,
      "learning_rate": 9.949768215691022e-05,
      "loss": 0.429,
      "step": 930
    },
    {
      "grad_norm": 0.892963171005249,
      "learning_rate": 9.9474032758977e-05,
      "loss": 0.4265,
      "step": 940
    },
    {
      "grad_norm": 0.5317549109458923,
      "learning_rate": 9.944984232094794e-05,
      "loss": 0.4507,
      "step": 950
    },
    {
      "grad_norm": 0.5821920037269592,
      "learning_rate": 9.942511110736584e-05,
      "loss": 0.4063,
      "step": 960
    },
    {
      "grad_norm": 0.5321699380874634,
      "learning_rate": 9.939983938868726e-05,
      "loss": 0.416,
      "step": 970
    },
    {
      "grad_norm": 0.5910799503326416,
      "learning_rate": 9.93740274412797e-05,
      "loss": 0.4583,
      "step": 980
    },
    {
      "grad_norm": 0.6233433485031128,
      "learning_rate": 9.934767554741846e-05,
      "loss": 0.4216,
      "step": 990
    },
    {
      "grad_norm": 0.8056014180183411,
      "learning_rate": 9.932078399528361e-05,
      "loss": 0.4222,
      "step": 1000
    },
    {
      "grad_norm": 0.5304357409477234,
      "learning_rate": 9.929335307895689e-05,
      "loss": 0.4,
      "step": 1010
    },
    {
      "grad_norm": 0.6021572351455688,
      "learning_rate": 9.926538309841839e-05,
      "loss": 0.434,
      "step": 1020
    },
    {
      "grad_norm": 0.7962700128555298,
      "learning_rate": 9.923687435954334e-05,
      "loss": 0.419,
      "step": 1030
    },
    {
      "grad_norm": 0.5842007994651794,
      "learning_rate": 9.920782717409873e-05,
      "loss": 0.43,
      "step": 1040
    },
    {
      "grad_norm": 0.6993943452835083,
      "learning_rate": 9.917824185973994e-05,
      "loss": 0.3964,
      "step": 1050
    },
    {
      "grad_norm": 0.7126774787902832,
      "learning_rate": 9.914811874000723e-05,
      "loss": 0.4565,
      "step": 1060
    },
    {
      "grad_norm": 0.5427873134613037,
      "learning_rate": 9.911745814432218e-05,
      "loss": 0.4255,
      "step": 1070
    },
    {
      "grad_norm": 0.7471935153007507,
      "learning_rate": 9.90862604079842e-05,
      "loss": 0.4169,
      "step": 1080
    },
    {
      "grad_norm": 0.508635401725769,
      "learning_rate": 9.90545258721667e-05,
      "loss": 0.4018,
      "step": 1090
    },
    {
      "grad_norm": 0.8062880635261536,
      "learning_rate": 9.90222548839135e-05,
      "loss": 0.4116,
      "step": 1100
    },
    {
      "grad_norm": 0.5755869746208191,
      "learning_rate": 9.898944779613495e-05,
      "loss": 0.4251,
      "step": 1110
    },
    {
      "grad_norm": 0.4646329879760742,
      "learning_rate": 9.89561049676041e-05,
      "loss": 0.4059,
      "step": 1120
    },
    {
      "grad_norm": 0.5535596013069153,
      "learning_rate": 9.89222267629528e-05,
      "loss": 0.4327,
      "step": 1130
    },
    {
      "grad_norm": 0.706025242805481,
      "learning_rate": 9.888781355266763e-05,
      "loss": 0.4263,
      "step": 1140
    },
    {
      "grad_norm": 0.5745698809623718,
      "learning_rate": 9.885286571308598e-05,
      "loss": 0.4297,
      "step": 1150
    },
    {
      "grad_norm": 0.4360511898994446,
      "learning_rate": 9.881738362639182e-05,
      "loss": 0.414,
      "step": 1160
    },
    {
      "grad_norm": 0.4219804108142853,
      "learning_rate": 9.878136768061154e-05,
      "loss": 0.4411,
      "step": 1170
    },
    {
      "grad_norm": 0.49969974160194397,
      "learning_rate": 9.874481826960979e-05,
      "loss": 0.4102,
      "step": 1180
    },
    {
      "grad_norm": 0.5531954765319824,
      "learning_rate": 9.870773579308503e-05,
      "loss": 0.4161,
      "step": 1190
    },
    {
      "grad_norm": 0.456541508436203,
      "learning_rate": 9.867012065656533e-05,
      "loss": 0.4269,
      "step": 1200
    },
    {
      "grad_norm": 0.55974942445755,
      "learning_rate": 9.863197327140376e-05,
      "loss": 0.4074,
      "step": 1210
    },
    {
      "grad_norm": 0.39756521582603455,
      "learning_rate": 9.859329405477403e-05,
      "loss": 0.404,
      "step": 1220
    },
    {
      "grad_norm": 0.6980628371238708,
      "learning_rate": 9.855408342966585e-05,
      "loss": 0.4228,
      "step": 1230
    },
    {
      "grad_norm": 0.44752705097198486,
      "learning_rate": 9.851434182488033e-05,
      "loss": 0.4186,
      "step": 1240
    },
    {
      "grad_norm": 0.46628278493881226,
      "learning_rate": 9.84740696750253e-05,
      "loss": 0.4386,
      "step": 1250
    },
    {
      "grad_norm": 0.6598864793777466,
      "learning_rate": 9.843326742051055e-05,
      "loss": 0.4225,
      "step": 1260
    },
    {
      "grad_norm": 0.6485694050788879,
      "learning_rate": 9.839193550754297e-05,
      "loss": 0.436,
      "step": 1270
    },
    {
      "grad_norm": 0.384075790643692,
      "learning_rate": 9.835007438812177e-05,
      "loss": 0.4307,
      "step": 1280
    },
    {
      "grad_norm": 0.4718746542930603,
      "learning_rate": 9.830768452003341e-05,
      "loss": 0.4455,
      "step": 1290
    },
    {
      "grad_norm": 0.5701053738594055,
      "learning_rate": 9.826476636684671e-05,
      "loss": 0.4195,
      "step": 1300
    },
    {
      "grad_norm": 0.5587959289550781,
      "learning_rate": 9.822132039790773e-05,
      "loss": 0.4561,
      "step": 1310
    },
    {
      "grad_norm": 0.5194388031959534,
      "learning_rate": 9.817734708833461e-05,
      "loss": 0.4273,
      "step": 1320
    },
    {
      "grad_norm": 0.5819528102874756,
      "learning_rate": 9.813284691901243e-05,
      "loss": 0.3985,
      "step": 1330
    },
    {
      "grad_norm": 0.5790693163871765,
      "learning_rate": 9.808782037658792e-05,
      "loss": 0.4031,
      "step": 1340
    },
    {
      "grad_norm": 0.4841820001602173,
      "learning_rate": 9.804226795346411e-05,
      "loss": 0.3937,
      "step": 1350
    },
    {
      "grad_norm": 0.5018506646156311,
      "learning_rate": 9.799619014779503e-05,
      "loss": 0.4162,
      "step": 1360
    },
    {
      "grad_norm": 0.4540066421031952,
      "learning_rate": 9.794958746348013e-05,
      "loss": 0.3763,
      "step": 1370
    },
    {
      "grad_norm": 0.7710187435150146,
      "learning_rate": 9.790246041015896e-05,
      "loss": 0.4476,
      "step": 1380
    },
    {
      "grad_norm": 0.667389988899231,
      "learning_rate": 9.785480950320538e-05,
      "loss": 0.407,
      "step": 1390
    },
    {
      "grad_norm": 0.4693636894226074,
      "learning_rate": 9.78066352637221e-05,
      "loss": 0.4182,
      "step": 1400
    },
    {
      "grad_norm": 0.5108399391174316,
      "learning_rate": 9.775793821853488e-05,
      "loss": 0.4001,
      "step": 1410
    },
    {
      "grad_norm": 0.3898954689502716,
      "learning_rate": 9.77087189001868e-05,
      "loss": 0.4222,
      "step": 1420
    },
    {
      "grad_norm": 0.8244535326957703,
      "learning_rate": 9.765897784693243e-05,
      "loss": 0.3918,
      "step": 1430
    },
    {
      "grad_norm": 0.5128754377365112,
      "learning_rate": 9.760871560273197e-05,
      "loss": 0.4198,
      "step": 1440
    },
    {
      "grad_norm": 0.33137279748916626,
      "learning_rate": 9.755793271724526e-05,
      "loss": 0.4113,
      "step": 1450
    },
    {
      "grad_norm": 0.7306418418884277,
      "learning_rate": 9.750662974582584e-05,
      "loss": 0.4095,
      "step": 1460
    },
    {
      "grad_norm": 0.5617614984512329,
      "learning_rate": 9.745480724951473e-05,
      "loss": 0.4279,
      "step": 1470
    },
    {
      "grad_norm": 0.4621225595474243,
      "learning_rate": 9.740246579503447e-05,
      "loss": 0.4335,
      "step": 1480
    },
    {
      "grad_norm": 0.4054778814315796,
      "learning_rate": 9.734960595478284e-05,
      "loss": 0.4039,
      "step": 1490
    },
    {
      "grad_norm": 0.4362834393978119,
      "learning_rate": 9.729622830682657e-05,
      "loss": 0.436,
      "step": 1500
    },
    {
      "grad_norm": 0.7111794948577881,
      "learning_rate": 9.724233343489504e-05,
      "loss": 0.4165,
      "step": 1510
    },
    {
      "grad_norm": 0.5596457123756409,
      "learning_rate": 9.718792192837396e-05,
      "loss": 0.4124,
      "step": 1520
    },
    {
      "grad_norm": 0.4505234658718109,
      "learning_rate": 9.713299438229886e-05,
      "loss": 0.4351,
      "step": 1530
    },
    {
      "grad_norm": 0.48055148124694824,
      "learning_rate": 9.707755139734855e-05,
      "loss": 0.4428,
      "step": 1540
    },
    {
      "grad_norm": 0.4456385672092438,
      "learning_rate": 9.702159357983866e-05,
      "loss": 0.4064,
      "step": 1550
    },
    {
      "grad_norm": 0.828874409198761,
      "learning_rate": 9.696512154171492e-05,
      "loss": 0.421,
      "step": 1560
    },
    {
      "grad_norm": 0.46821802854537964,
      "learning_rate": 9.690813590054645e-05,
      "loss": 0.4301,
      "step": 1570
    },
    {
      "grad_norm": 0.3891923129558563,
      "learning_rate": 9.685063727951914e-05,
      "loss": 0.3976,
      "step": 1580
    },
    {
      "grad_norm": 0.5268163084983826,
      "learning_rate": 9.679262630742865e-05,
      "loss": 0.416,
      "step": 1590
    },
    {
      "grad_norm": 0.5543060898780823,
      "learning_rate": 9.673410361867373e-05,
      "loss": 0.4134,
      "step": 1600
    },
    {
      "grad_norm": 0.6950433254241943,
      "learning_rate": 9.667506985324909e-05,
      "loss": 0.3803,
      "step": 1610
    },
    {
      "grad_norm": 0.43334418535232544,
      "learning_rate": 9.661552565673855e-05,
      "loss": 0.383,
      "step": 1620
    },
    {
      "grad_norm": 0.49166813492774963,
      "learning_rate": 9.655547168030789e-05,
      "loss": 0.4226,
      "step": 1630
    },
    {
      "grad_norm": 0.37123486399650574,
      "learning_rate": 9.649490858069777e-05,
      "loss": 0.4153,
      "step": 1640
    },
    {
      "grad_norm": 0.3389699459075928,
      "learning_rate": 9.643383702021658e-05,
      "loss": 0.425,
      "step": 1650
    },
    {
      "grad_norm": 0.4162275791168213,
      "learning_rate": 9.637225766673307e-05,
      "loss": 0.3853,
      "step": 1660
    },
    {
      "grad_norm": 0.5858766436576843,
      "learning_rate": 9.631017119366922e-05,
      "loss": 0.4125,
      "step": 1670
    },
    {
      "grad_norm": 0.46214354038238525,
      "learning_rate": 9.624757827999273e-05,
      "loss": 0.4028,
      "step": 1680
    },
    {
      "grad_norm": 0.6208974719047546,
      "learning_rate": 9.618447961020971e-05,
      "loss": 0.4234,
      "step": 1690
    },
    {
      "grad_norm": 0.5344316363334656,
      "learning_rate": 9.612087587435707e-05,
      "loss": 0.4061,
      "step": 1700
    },
    {
      "grad_norm": 0.5468342900276184,
      "learning_rate": 9.605676776799508e-05,
      "loss": 0.4394,
      "step": 1710
    },
    {
      "grad_norm": 0.44906333088874817,
      "learning_rate": 9.599215599219973e-05,
      "loss": 0.4027,
      "step": 1720
    },
    {
      "grad_norm": 0.615778386592865,
      "learning_rate": 9.592704125355505e-05,
      "loss": 0.4184,
      "step": 1730
    },
    {
      "grad_norm": 0.4976154565811157,
      "learning_rate": 9.586142426414538e-05,
      "loss": 0.4291,
      "step": 1740
    },
    {
      "grad_norm": 0.5187555551528931,
      "learning_rate": 9.57953057415476e-05,
      "loss": 0.4392,
      "step": 1750
    },
    {
      "grad_norm": 0.6631363034248352,
      "learning_rate": 9.572868640882328e-05,
      "loss": 0.4207,
      "step": 1760
    },
    {
      "grad_norm": 0.533089280128479,
      "learning_rate": 9.56615669945108e-05,
      "loss": 0.4361,
      "step": 1770
    },
    {
      "grad_norm": 0.5720014572143555,
      "learning_rate": 9.55939482326173e-05,
      "loss": 0.4126,
      "step": 1780
    },
    {
      "grad_norm": 0.33309683203697205,
      "learning_rate": 9.552583086261069e-05,
      "loss": 0.412,
      "step": 1790
    },
    {
      "grad_norm": 0.5297278165817261,
      "learning_rate": 9.545721562941168e-05,
      "loss": 0.4345,
      "step": 1800
    },
    {
      "grad_norm": 0.47707921266555786,
      "learning_rate": 9.538810328338543e-05,
      "loss": 0.4349,
      "step": 1810
    },
    {
      "grad_norm": 0.4352557063102722,
      "learning_rate": 9.531849458033349e-05,
      "loss": 0.3924,
      "step": 1820
    },
    {
      "grad_norm": 0.5403316020965576,
      "learning_rate": 9.524839028148547e-05,
      "loss": 0.4081,
      "step": 1830
    },
    {
      "grad_norm": 0.33877307176589966,
      "learning_rate": 9.517779115349077e-05,
      "loss": 0.4198,
      "step": 1840
    },
    {
      "grad_norm": 0.4647667407989502,
      "learning_rate": 9.510669796841014e-05,
      "loss": 0.4141,
      "step": 1850
    },
    {
      "grad_norm": 0.46779754757881165,
      "learning_rate": 9.503511150370727e-05,
      "loss": 0.4595,
      "step": 1860
    },
    {
      "grad_norm": 0.4666247069835663,
      "learning_rate": 9.496303254224024e-05,
      "loss": 0.4094,
      "step": 1870
    },
    {
      "grad_norm": 0.5954828858375549,
      "learning_rate": 9.489046187225306e-05,
      "loss": 0.3938,
      "step": 1880
    },
    {
      "grad_norm": 0.9290278553962708,
      "learning_rate": 9.481740028736692e-05,
      "loss": 0.374,
      "step": 1890
    },
    {
      "grad_norm": 0.5421234965324402,
      "learning_rate": 9.474384858657164e-05,
      "loss": 0.4326,
      "step": 1900
    },
    {
      "grad_norm": 0.4756077229976654,
      "learning_rate": 9.466980757421679e-05,
      "loss": 0.4291,
      "step": 1910
    },
    {
      "grad_norm": 0.6380815505981445,
      "learning_rate": 9.459527806000305e-05,
      "loss": 0.4251,
      "step": 1920
    },
    {
      "grad_norm": 0.465554416179657,
      "learning_rate": 9.452026085897325e-05,
      "loss": 0.4198,
      "step": 1930
    },
    {
      "grad_norm": 0.6021956205368042,
      "learning_rate": 9.444475679150348e-05,
      "loss": 0.4351,
      "step": 1940
    },
    {
      "grad_norm": 0.4163987934589386,
      "learning_rate": 9.436876668329411e-05,
      "loss": 0.4097,
      "step": 1950
    },
    {
      "grad_norm": 0.6961418390274048,
      "learning_rate": 9.429229136536079e-05,
      "loss": 0.3843,
      "step": 1960
    },
    {
      "grad_norm": 0.5080003142356873,
      "learning_rate": 9.421533167402534e-05,
      "loss": 0.4392,
      "step": 1970
    },
    {
      "grad_norm": 0.4982507526874542,
      "learning_rate": 9.413788845090666e-05,
      "loss": 0.4028,
      "step": 1980
    },
    {
      "grad_norm": 0.3845439553260803,
      "learning_rate": 9.405996254291136e-05,
      "loss": 0.415,
      "step": 1990
    },
    {
      "grad_norm": 0.4884818494319916,
      "learning_rate": 9.398155480222474e-05,
      "loss": 0.412,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}