{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 2226,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 3.999950204782701e-05,
      "loss": 1.7807,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.999800821610369e-05,
      "loss": 1.781,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.999551857921571e-05,
      "loss": 1.8258,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.999203326113507e-05,
      "loss": 1.5688,
      "step": 40
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.9987552435413944e-05,
      "loss": 1.5824,
      "step": 50
    },
    {
      "epoch": 0.08,
      "learning_rate": 3.9982076325176035e-05,
      "loss": 1.7159,
      "step": 60
    },
    {
      "epoch": 0.09,
      "learning_rate": 3.9975605203105434e-05,
      "loss": 1.7095,
      "step": 70
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.996813939143307e-05,
      "loss": 1.6194,
      "step": 80
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.9959679261920665e-05,
      "loss": 1.6602,
      "step": 90
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.995022523584219e-05,
      "loss": 1.6734,
      "step": 100
    },
    {
      "epoch": 0.15,
      "learning_rate": 3.9939777783962946e-05,
      "loss": 1.715,
      "step": 110
    },
    {
      "epoch": 0.16,
      "learning_rate": 3.992833742651606e-05,
      "loss": 1.7322,
      "step": 120
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.9915904733176614e-05,
      "loss": 1.6645,
      "step": 130
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.9902480323033285e-05,
      "loss": 1.6249,
      "step": 140
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.9888064864557486e-05,
      "loss": 1.6279,
      "step": 150
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.987265907557011e-05,
      "loss": 1.5166,
      "step": 160
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.9856263723205755e-05,
      "loss": 1.4946,
      "step": 170
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.983887962387457e-05,
      "loss": 1.5583,
      "step": 180
    },
    {
      "epoch": 0.26,
      "learning_rate": 3.982050764322154e-05,
      "loss": 1.526,
      "step": 190
    },
    {
      "epoch": 0.27,
      "learning_rate": 3.9801148696083455e-05,
      "loss": 1.6003,
      "step": 200
    },
    {
      "epoch": 0.28,
      "learning_rate": 3.9780803746443284e-05,
      "loss": 1.6403,
      "step": 210
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.9759473807382214e-05,
      "loss": 1.5128,
      "step": 220
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.97371599410292e-05,
      "loss": 1.4739,
      "step": 230
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.9713863258508064e-05,
      "loss": 1.5466,
      "step": 240
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.968958491988216e-05,
      "loss": 1.5307,
      "step": 250
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.966432613409667e-05,
      "loss": 1.4508,
      "step": 260
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.9638088158918285e-05,
      "loss": 1.5163,
      "step": 270
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9610872300872704e-05,
      "loss": 1.5583,
      "step": 280
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.958267991517948e-05,
      "loss": 1.3893,
      "step": 290
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.955351240568459e-05,
      "loss": 1.4718,
      "step": 300
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.9523371224790505e-05,
      "loss": 1.479,
      "step": 310
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.949225787338388e-05,
      "loss": 1.4874,
      "step": 320
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.946017390076081e-05,
      "loss": 1.2723,
      "step": 330
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.942712090454968e-05,
      "loss": 1.3925,
      "step": 340
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.939310053063161e-05,
      "loss": 1.3833,
      "step": 350
    },
    {
      "epoch": 0.49,
      "learning_rate": 3.935811447305853e-05,
      "loss": 1.4989,
      "step": 360
    },
    {
      "epoch": 0.5,
      "learning_rate": 3.9322164473968774e-05,
      "loss": 1.4961,
      "step": 370
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.928525232350035e-05,
      "loss": 1.5295,
      "step": 380
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.924737985970182e-05,
      "loss": 1.4051,
      "step": 390
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.920854896844074e-05,
      "loss": 1.3306,
      "step": 400
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.916876158330979e-05,
      "loss": 1.3522,
      "step": 410
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.912801968553045e-05,
      "loss": 1.4479,
      "step": 420
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.908632530385438e-05,
      "loss": 1.5481,
      "step": 430
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.9043680514462366e-05,
      "loss": 1.5375,
      "step": 440
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.900008744086097e-05,
      "loss": 1.321,
      "step": 450
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.895554825377676e-05,
      "loss": 1.4584,
      "step": 460
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.891006517104823e-05,
      "loss": 1.4188,
      "step": 470
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.886364045751538e-05,
      "loss": 1.2557,
      "step": 480
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.881627642490691e-05,
      "loss": 1.412,
      "step": 490
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.876797543172511e-05,
      "loss": 1.3267,
      "step": 500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.871873988312842e-05,
      "loss": 1.3768,
      "step": 510
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.86685722308117e-05,
      "loss": 1.5065,
      "step": 520
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.861747497288409e-05,
      "loss": 1.382,
      "step": 530
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.856545065374465e-05,
      "loss": 1.3336,
      "step": 540
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.851250186395565e-05,
      "loss": 1.2626,
      "step": 550
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.845863124011361e-05,
      "loss": 1.4174,
      "step": 560
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.840384146471792e-05,
      "loss": 1.3371,
      "step": 570
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.8348135266037364e-05,
      "loss": 1.3496,
      "step": 580
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.829151541797421e-05,
      "loss": 1.2245,
      "step": 590
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.82339847399261e-05,
      "loss": 1.2381,
      "step": 600
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.817554609664564e-05,
      "loss": 1.2805,
      "step": 610
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.811620239809778e-05,
      "loss": 1.2055,
      "step": 620
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.805595659931487e-05,
      "loss": 1.3493,
      "step": 630
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.799481170024957e-05,
      "loss": 1.4545,
      "step": 640
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.7932770745625406e-05,
      "loss": 1.3633,
      "step": 650
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.786983682478519e-05,
      "loss": 1.2677,
      "step": 660
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.78060130715372e-05,
      "loss": 1.2218,
      "step": 670
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.7741302663999085e-05,
      "loss": 1.4738,
      "step": 680
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.7675708824439656e-05,
      "loss": 1.191,
      "step": 690
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.76092348191184e-05,
      "loss": 1.2747,
      "step": 700
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.7541883958122864e-05,
      "loss": 1.2833,
      "step": 710
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.7473659595203806e-05,
      "loss": 1.3725,
      "step": 720
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.74045651276082e-05,
      "loss": 1.31,
      "step": 730
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.7334603995910075e-05,
      "loss": 1.2406,
      "step": 740
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.726377968383919e-05,
      "loss": 1.1966,
      "step": 750
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.719209571810755e-05,
      "loss": 1.2898,
      "step": 760
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.711955566823381e-05,
      "loss": 1.3516,
      "step": 770
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.704616314636551e-05,
      "loss": 1.4107,
      "step": 780
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.697192180709923e-05,
      "loss": 1.3388,
      "step": 790
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.6896835347298583e-05,
      "loss": 1.1733,
      "step": 800
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.682090750591016e-05,
      "loss": 1.3502,
      "step": 810
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.674414206377732e-05,
      "loss": 1.3909,
      "step": 820
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.6666542843451936e-05,
      "loss": 1.2328,
      "step": 830
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.658811370900404e-05,
      "loss": 1.1578,
      "step": 840
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.650885856582943e-05,
      "loss": 1.3599,
      "step": 850
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.6428781360455176e-05,
      "loss": 1.1904,
      "step": 860
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.6347886080343135e-05,
      "loss": 1.2668,
      "step": 870
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.626617675369135e-05,
      "loss": 1.215,
      "step": 880
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.6183657449233484e-05,
      "loss": 1.2978,
      "step": 890
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.610033227603626e-05,
      "loss": 1.3659,
      "step": 900
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.601620538329476e-05,
      "loss": 1.2659,
      "step": 910
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.593128096012589e-05,
      "loss": 1.2425,
      "step": 920
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.584556323535978e-05,
      "loss": 1.362,
      "step": 930
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.5759056477329135e-05,
      "loss": 1.1444,
      "step": 940
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.5671764993656784e-05,
      "loss": 1.3044,
      "step": 950
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.558369313104112e-05,
      "loss": 1.288,
      "step": 960
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.5494845275039676e-05,
      "loss": 1.338,
      "step": 970
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.5405225849850754e-05,
      "loss": 1.4229,
      "step": 980
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.531483931809311e-05,
      "loss": 1.2542,
      "step": 990
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.5223690180583717e-05,
      "loss": 1.3793,
      "step": 1000
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.513178297611369e-05,
      "loss": 1.2778,
      "step": 1010
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.503912228122226e-05,
      "loss": 1.384,
      "step": 1020
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.494571270996885e-05,
      "loss": 1.1319,
      "step": 1030
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.4851558913703367e-05,
      "loss": 1.312,
      "step": 1040
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.475666558083455e-05,
      "loss": 1.1727,
      "step": 1050
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.4661037436596526e-05,
      "loss": 1.1568,
      "step": 1060
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.456467924281353e-05,
      "loss": 1.2801,
      "step": 1070
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.446759579766275e-05,
      "loss": 1.2211,
      "step": 1080
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.436979193543543e-05,
      "loss": 1.2206,
      "step": 1090
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.427127252629616e-05,
      "loss": 1.2377,
      "step": 1100
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.417204247604031e-05,
      "loss": 1.2548,
      "step": 1110
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.407210672584979e-05,
      "loss": 1.1323,
      "step": 1120
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.397147025204701e-05,
      "loss": 1.1272,
      "step": 1130
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.387013806584705e-05,
      "loss": 1.1798,
      "step": 1140
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.376811521310814e-05,
      "loss": 1.4287,
      "step": 1150
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.36654067740804e-05,
      "loss": 1.1023,
      "step": 1160
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.3562017863152867e-05,
      "loss": 1.3098,
      "step": 1170
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.3457953628598826e-05,
      "loss": 1.3028,
      "step": 1180
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.335321925231946e-05,
      "loss": 1.1454,
      "step": 1190
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.3247819949585776e-05,
      "loss": 1.1599,
      "step": 1200
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.314176096877898e-05,
      "loss": 1.2164,
      "step": 1210
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.3035047591129054e-05,
      "loss": 1.266,
      "step": 1220
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.292768513045183e-05,
      "loss": 1.2517,
      "step": 1230
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.281967893288436e-05,
      "loss": 1.1866,
      "step": 1240
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.271103437661873e-05,
      "loss": 1.1079,
      "step": 1250
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.260175687163423e-05,
      "loss": 1.2,
      "step": 1260
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.249185185942795e-05,
      "loss": 1.1048,
      "step": 1270
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.2381324812743875e-05,
      "loss": 1.3801,
      "step": 1280
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.22701812353003e-05,
      "loss": 1.1595,
      "step": 1290
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.215842666151582e-05,
      "loss": 0.9625,
      "step": 1300
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.2046066656233726e-05,
      "loss": 1.0588,
      "step": 1310
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.193310681444488e-05,
      "loss": 1.1553,
      "step": 1320
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.181955276100917e-05,
      "loss": 1.0569,
      "step": 1330
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.170541015037535e-05,
      "loss": 1.1522,
      "step": 1340
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.159068466629951e-05,
      "loss": 1.0199,
      "step": 1350
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.147538202156208e-05,
      "loss": 1.2719,
      "step": 1360
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.135950795768331e-05,
      "loss": 1.1117,
      "step": 1370
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.1243068244637364e-05,
      "loss": 1.0545,
      "step": 1380
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.112606868056508e-05,
      "loss": 1.1852,
      "step": 1390
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.100851509148517e-05,
      "loss": 1.0035,
      "step": 1400
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.089041333100414e-05,
      "loss": 1.2372,
      "step": 1410
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.077176928002482e-05,
      "loss": 1.2852,
      "step": 1420
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.065258884645351e-05,
      "loss": 1.1826,
      "step": 1430
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.05328779649058e-05,
      "loss": 1.2395,
      "step": 1440
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.041264259641104e-05,
      "loss": 1.3701,
      "step": 1450
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.029188872811554e-05,
      "loss": 1.0938,
      "step": 1460
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.017062237298441e-05,
      "loss": 1.2473,
      "step": 1470
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.0048849569502158e-05,
      "loss": 1.2034,
      "step": 1480
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9926576381371987e-05,
      "loss": 1.2143,
      "step": 1490
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.980380889721388e-05,
      "loss": 1.152,
      "step": 1500
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.9680553230261373e-05,
      "loss": 1.0943,
      "step": 1510
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.9556815518057198e-05,
      "loss": 1.4116,
      "step": 1520
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.9432601922147615e-05,
      "loss": 1.1894,
      "step": 1530
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.9307918627775627e-05,
      "loss": 1.0567,
      "step": 1540
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.918277184357297e-05,
      "loss": 1.0172,
      "step": 1550
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.9057167801250962e-05,
      "loss": 1.1905,
      "step": 1560
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.8931112755290173e-05,
      "loss": 1.3694,
      "step": 1570
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.8804612982629015e-05,
      "loss": 1.1596,
      "step": 1580
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.8677674782351164e-05,
      "loss": 1.3288,
      "step": 1590
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.8550304475371876e-05,
      "loss": 1.0669,
      "step": 1600
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.8422508404123264e-05,
      "loss": 1.1202,
      "step": 1610
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.829429293223847e-05,
      "loss": 1.1461,
      "step": 1620
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.8165664444234776e-05,
      "loss": 1.1503,
      "step": 1630
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.8036629345195695e-05,
      "loss": 1.1463,
      "step": 1640
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.7907194060452013e-05,
      "loss": 1.1446,
      "step": 1650
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.777736503526188e-05,
      "loss": 1.1918,
      "step": 1660
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.7647148734489797e-05,
      "loss": 1.2787,
      "step": 1670
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.7516551642284765e-05,
      "loss": 1.1412,
      "step": 1680
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.7385580261757368e-05,
      "loss": 1.0178,
      "step": 1690
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.7254241114655965e-05,
      "loss": 1.0412,
      "step": 1700
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7122540741041915e-05,
      "loss": 1.0242,
      "step": 1710
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6990485698963945e-05,
      "loss": 1.1745,
      "step": 1720
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.6858082564131577e-05,
      "loss": 0.9009,
      "step": 1730
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.672533792958768e-05,
      "loss": 1.0899,
      "step": 1740
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.659225840538016e-05,
      "loss": 0.9803,
      "step": 1750
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.645885061823286e-05,
      "loss": 1.0568,
      "step": 1760
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.6325121211215526e-05,
      "loss": 1.1428,
      "step": 1770
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.619107684341304e-05,
      "loss": 1.2611,
      "step": 1780
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.6056724189593817e-05,
      "loss": 1.0649,
      "step": 1790
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.5922069939877464e-05,
      "loss": 1.0907,
      "step": 1800
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.5787120799401595e-05,
      "loss": 1.1944,
      "step": 1810
    },
    {
      "epoch": 2.45,
      "learning_rate": 2.565188348798798e-05,
      "loss": 1.0326,
      "step": 1820
    },
    {
      "epoch": 2.47,
      "learning_rate": 2.5516364739807938e-05,
      "loss": 1.1004,
      "step": 1830
    },
    {
      "epoch": 2.48,
      "learning_rate": 2.538057130304698e-05,
      "loss": 1.253,
      "step": 1840
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.5244509939568786e-05,
      "loss": 1.0609,
      "step": 1850
    },
    {
      "epoch": 2.51,
      "learning_rate": 2.5108187424578533e-05,
      "loss": 0.9903,
      "step": 1860
    },
    {
      "epoch": 2.52,
      "learning_rate": 2.4971610546285474e-05,
      "loss": 1.0218,
      "step": 1870
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.483478610556494e-05,
      "loss": 1.2747,
      "step": 1880
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.469772091561968e-05,
      "loss": 1.0557,
      "step": 1890
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.4560421801640618e-05,
      "loss": 1.1359,
      "step": 1900
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.4422895600466968e-05,
      "loss": 1.1069,
      "step": 1910
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.4285149160245806e-05,
      "loss": 1.2929,
      "step": 1920
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.414718934009105e-05,
      "loss": 0.932,
      "step": 1930
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.4009023009741932e-05,
      "loss": 1.1145,
      "step": 1940
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.3870657049220885e-05,
      "loss": 0.9006,
      "step": 1950
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.373209834849098e-05,
      "loss": 1.0325,
      "step": 1960
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.3593353807112838e-05,
      "loss": 1.023,
      "step": 1970
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.3454430333901037e-05,
      "loss": 0.9664,
      "step": 1980
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.331533484658011e-05,
      "loss": 1.05,
      "step": 1990
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.3176074271440086e-05,
      "loss": 0.8807,
      "step": 2000
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.303665554299156e-05,
      "loss": 0.9423,
      "step": 2010
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2897085603620405e-05,
      "loss": 1.0951,
      "step": 2020
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.2757371403242094e-05,
      "loss": 1.1253,
      "step": 2030
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.2617519898955572e-05,
      "loss": 1.099,
      "step": 2040
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.2477538054696905e-05,
      "loss": 0.9475,
      "step": 2050
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.2337432840892438e-05,
      "loss": 1.0271,
      "step": 2060
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.2197211234111743e-05,
      "loss": 0.9412,
      "step": 2070
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.2056880216720184e-05,
      "loss": 1.1529,
      "step": 2080
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.1916446776531276e-05,
      "loss": 1.0621,
      "step": 2090
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.1775917906458698e-05,
      "loss": 1.1638,
      "step": 2100
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.163530060416806e-05,
      "loss": 1.1005,
      "step": 2110
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.149460187172849e-05,
      "loss": 1.2081,
      "step": 2120
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1353828715263937e-05,
      "loss": 1.0704,
      "step": 2130
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.121298814460433e-05,
      "loss": 1.0406,
      "step": 2140
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.1072087172936473e-05,
      "loss": 0.8944,
      "step": 2150
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.093113281645488e-05,
      "loss": 0.87,
      "step": 2160
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.079013209401236e-05,
      "loss": 1.1228,
      "step": 2170
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.0649092026770517e-05,
      "loss": 1.0616,
      "step": 2180
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.0508019637850164e-05,
      "loss": 1.1175,
      "step": 2190
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.0366921951981546e-05,
      "loss": 1.1289,
      "step": 2200
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.02258059951546e-05,
      "loss": 1.1765,
      "step": 2210
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.0084678794269057e-05,
      "loss": 1.042,
      "step": 2220
    }
  ],
  "logging_steps": 10,
  "max_steps": 4452,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 1.163304048181248e+16,
  "trial_name": null,
  "trial_params": null
}