{
  "best_metric": 0.2249804437160492,
  "best_model_checkpoint": "/opt/ml/model/checkpoint-3060",
  "epoch": 9.0,
  "global_step": 3060,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 1.4705882352941178e-07,
      "loss": 4.4596,
      "step": 10
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.9411764705882356e-07,
      "loss": 4.2962,
      "step": 20
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.4117647058823536e-07,
      "loss": 4.4439,
      "step": 30
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.882352941176471e-07,
      "loss": 4.3771,
      "step": 40
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.352941176470589e-07,
      "loss": 4.3934,
      "step": 50
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.823529411764707e-07,
      "loss": 4.3209,
      "step": 60
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.0294117647058825e-06,
      "loss": 4.2707,
      "step": 70
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 4.2453,
      "step": 80
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.323529411764706e-06,
      "loss": 4.0978,
      "step": 90
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4705882352941177e-06,
      "loss": 4.0954,
      "step": 100
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6176470588235297e-06,
      "loss": 3.8289,
      "step": 110
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 4.1465,
      "step": 120
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9117647058823528e-06,
      "loss": 4.1067,
      "step": 130
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.058823529411765e-06,
      "loss": 4.0671,
      "step": 140
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.2058823529411767e-06,
      "loss": 3.7774,
      "step": 150
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 3.8341,
      "step": 160
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.5e-06,
      "loss": 3.5665,
      "step": 170
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.647058823529412e-06,
      "loss": 3.5655,
      "step": 180
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.7941176470588237e-06,
      "loss": 3.505,
      "step": 190
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 3.6714,
      "step": 200
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.0882352941176476e-06,
      "loss": 3.4006,
      "step": 210
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.2352941176470594e-06,
      "loss": 3.1562,
      "step": 220
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.382352941176471e-06,
      "loss": 3.322,
      "step": 230
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.529411764705883e-06,
      "loss": 3.1499,
      "step": 240
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.6764705882352946e-06,
      "loss": 3.0769,
      "step": 250
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.8235294117647055e-06,
      "loss": 2.9092,
      "step": 260
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.970588235294118e-06,
      "loss": 2.9317,
      "step": 270
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.11764705882353e-06,
      "loss": 2.8851,
      "step": 280
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.264705882352942e-06,
      "loss": 2.9637,
      "step": 290
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.411764705882353e-06,
      "loss": 2.5192,
      "step": 300
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.558823529411765e-06,
      "loss": 2.6035,
      "step": 310
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.705882352941177e-06,
      "loss": 2.4931,
      "step": 320
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.852941176470589e-06,
      "loss": 2.6514,
      "step": 330
    },
    {
      "epoch": 1.0,
      "learning_rate": 5e-06,
      "loss": 2.4065,
      "step": 340
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5992141453831041,
      "eval_f1": 0.5463480113288633,
      "eval_loss": 2.2797043323516846,
      "eval_precision": 0.6294796834703609,
      "eval_recall": 0.580237956829516,
      "eval_runtime": 161.1898,
      "eval_samples_per_second": 6.316,
      "eval_steps_per_second": 0.099,
      "step": 340
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.983660130718955e-06,
      "loss": 2.1525,
      "step": 350
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.967320261437909e-06,
      "loss": 1.8859,
      "step": 360
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9509803921568634e-06,
      "loss": 1.9121,
      "step": 370
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.934640522875817e-06,
      "loss": 1.6673,
      "step": 380
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.918300653594771e-06,
      "loss": 1.996,
      "step": 390
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.901960784313726e-06,
      "loss": 1.8359,
      "step": 400
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.88562091503268e-06,
      "loss": 1.5955,
      "step": 410
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.869281045751634e-06,
      "loss": 1.6088,
      "step": 420
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.852941176470589e-06,
      "loss": 1.7104,
      "step": 430
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.836601307189543e-06,
      "loss": 1.7513,
      "step": 440
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.820261437908497e-06,
      "loss": 1.5832,
      "step": 450
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.803921568627452e-06,
      "loss": 1.4003,
      "step": 460
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.7875816993464055e-06,
      "loss": 1.2787,
      "step": 470
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.77124183006536e-06,
      "loss": 1.4729,
      "step": 480
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.754901960784314e-06,
      "loss": 1.4468,
      "step": 490
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.7385620915032685e-06,
      "loss": 1.3275,
      "step": 500
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.722222222222222e-06,
      "loss": 1.4053,
      "step": 510
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.705882352941177e-06,
      "loss": 1.1261,
      "step": 520
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.689542483660131e-06,
      "loss": 1.346,
      "step": 530
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.673202614379085e-06,
      "loss": 1.2904,
      "step": 540
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.65686274509804e-06,
      "loss": 1.1763,
      "step": 550
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.640522875816994e-06,
      "loss": 1.232,
      "step": 560
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.624183006535948e-06,
      "loss": 1.0554,
      "step": 570
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.607843137254902e-06,
      "loss": 1.0278,
      "step": 580
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.591503267973857e-06,
      "loss": 1.0268,
      "step": 590
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.5751633986928105e-06,
      "loss": 0.9385,
      "step": 600
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.558823529411765e-06,
      "loss": 1.1249,
      "step": 610
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.542483660130719e-06,
      "loss": 1.0529,
      "step": 620
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.5261437908496736e-06,
      "loss": 1.0088,
      "step": 630
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.509803921568628e-06,
      "loss": 0.981,
      "step": 640
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.493464052287582e-06,
      "loss": 0.8764,
      "step": 650
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.477124183006537e-06,
      "loss": 0.7669,
      "step": 660
    },
    {
      "epoch": 1.97,
      "learning_rate": 4.460784313725491e-06,
      "loss": 0.8074,
      "step": 670
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.9048,
      "step": 680
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8359528487229863,
      "eval_f1": 0.8330712103623911,
      "eval_loss": 0.940582811832428,
      "eval_precision": 0.8510173659101683,
      "eval_recall": 0.8397313099233475,
      "eval_runtime": 158.421,
      "eval_samples_per_second": 6.426,
      "eval_steps_per_second": 0.101,
      "step": 680
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.428104575163399e-06,
      "loss": 0.7005,
      "step": 690
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.411764705882353e-06,
      "loss": 0.6216,
      "step": 700
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.395424836601307e-06,
      "loss": 0.6758,
      "step": 710
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.379084967320262e-06,
      "loss": 0.6859,
      "step": 720
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.3627450980392164e-06,
      "loss": 0.454,
      "step": 730
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.34640522875817e-06,
      "loss": 0.5674,
      "step": 740
    },
    {
      "epoch": 2.21,
      "learning_rate": 4.330065359477125e-06,
      "loss": 0.6264,
      "step": 750
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.313725490196079e-06,
      "loss": 0.5886,
      "step": 760
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.297385620915033e-06,
      "loss": 0.5141,
      "step": 770
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.281045751633987e-06,
      "loss": 0.4116,
      "step": 780
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.264705882352942e-06,
      "loss": 0.5151,
      "step": 790
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.2483660130718954e-06,
      "loss": 0.6208,
      "step": 800
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.23202614379085e-06,
      "loss": 0.4929,
      "step": 810
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.215686274509805e-06,
      "loss": 0.6218,
      "step": 820
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.1993464052287585e-06,
      "loss": 0.5226,
      "step": 830
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.183006535947713e-06,
      "loss": 0.4797,
      "step": 840
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.4065,
      "step": 850
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.1503267973856215e-06,
      "loss": 0.5133,
      "step": 860
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.133986928104575e-06,
      "loss": 0.4397,
      "step": 870
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.11764705882353e-06,
      "loss": 0.4372,
      "step": 880
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.101307189542484e-06,
      "loss": 0.5625,
      "step": 890
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.084967320261438e-06,
      "loss": 0.6021,
      "step": 900
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.068627450980392e-06,
      "loss": 0.5916,
      "step": 910
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.052287581699347e-06,
      "loss": 0.3234,
      "step": 920
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.035947712418301e-06,
      "loss": 0.5071,
      "step": 930
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.019607843137255e-06,
      "loss": 0.6007,
      "step": 940
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.00326797385621e-06,
      "loss": 0.5284,
      "step": 950
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.986928104575164e-06,
      "loss": 0.4562,
      "step": 960
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.970588235294118e-06,
      "loss": 0.2356,
      "step": 970
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.954248366013072e-06,
      "loss": 0.3074,
      "step": 980
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.9379084967320266e-06,
      "loss": 0.5702,
      "step": 990
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.92156862745098e-06,
      "loss": 0.5959,
      "step": 1000
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.905228758169935e-06,
      "loss": 0.5857,
      "step": 1010
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.4807,
      "step": 1020
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8870333988212181,
      "eval_f1": 0.8862169026751461,
      "eval_loss": 0.5658098459243774,
      "eval_precision": 0.8973600463348972,
      "eval_recall": 0.8907758376754111,
      "eval_runtime": 161.7981,
      "eval_samples_per_second": 6.292,
      "eval_steps_per_second": 0.099,
      "step": 1020
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.872549019607843e-06,
      "loss": 0.1907,
      "step": 1030
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.856209150326798e-06,
      "loss": 0.2597,
      "step": 1040
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.839869281045752e-06,
      "loss": 0.3115,
      "step": 1050
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.8235294117647055e-06,
      "loss": 0.2734,
      "step": 1060
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.80718954248366e-06,
      "loss": 0.4247,
      "step": 1070
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.7908496732026144e-06,
      "loss": 0.1969,
      "step": 1080
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.774509803921569e-06,
      "loss": 0.3282,
      "step": 1090
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.758169934640523e-06,
      "loss": 0.2913,
      "step": 1100
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.7418300653594774e-06,
      "loss": 0.1869,
      "step": 1110
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.7254901960784316e-06,
      "loss": 0.3293,
      "step": 1120
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.7091503267973862e-06,
      "loss": 0.2212,
      "step": 1130
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.6928104575163404e-06,
      "loss": 0.1879,
      "step": 1140
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.6764705882352946e-06,
      "loss": 0.2919,
      "step": 1150
    },
    {
      "epoch": 3.41,
      "learning_rate": 3.6601307189542484e-06,
      "loss": 0.2825,
      "step": 1160
    },
    {
      "epoch": 3.44,
      "learning_rate": 3.6437908496732026e-06,
      "loss": 0.2485,
      "step": 1170
    },
    {
      "epoch": 3.47,
      "learning_rate": 3.6274509803921573e-06,
      "loss": 0.1167,
      "step": 1180
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.6111111111111115e-06,
      "loss": 0.2036,
      "step": 1190
    },
    {
      "epoch": 3.53,
      "learning_rate": 3.5947712418300657e-06,
      "loss": 0.2564,
      "step": 1200
    },
    {
      "epoch": 3.56,
      "learning_rate": 3.57843137254902e-06,
      "loss": 0.2122,
      "step": 1210
    },
    {
      "epoch": 3.59,
      "learning_rate": 3.562091503267974e-06,
      "loss": 0.2749,
      "step": 1220
    },
    {
      "epoch": 3.62,
      "learning_rate": 3.5457516339869287e-06,
      "loss": 0.4118,
      "step": 1230
    },
    {
      "epoch": 3.65,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.2769,
      "step": 1240
    },
    {
      "epoch": 3.68,
      "learning_rate": 3.5130718954248367e-06,
      "loss": 0.2003,
      "step": 1250
    },
    {
      "epoch": 3.71,
      "learning_rate": 3.496732026143791e-06,
      "loss": 0.279,
      "step": 1260
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.480392156862745e-06,
      "loss": 0.1174,
      "step": 1270
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.4640522875816997e-06,
      "loss": 0.2338,
      "step": 1280
    },
    {
      "epoch": 3.79,
      "learning_rate": 3.447712418300654e-06,
      "loss": 0.2891,
      "step": 1290
    },
    {
      "epoch": 3.82,
      "learning_rate": 3.431372549019608e-06,
      "loss": 0.2443,
      "step": 1300
    },
    {
      "epoch": 3.85,
      "learning_rate": 3.4150326797385623e-06,
      "loss": 0.2581,
      "step": 1310
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.398692810457517e-06,
      "loss": 0.2144,
      "step": 1320
    },
    {
      "epoch": 3.91,
      "learning_rate": 3.382352941176471e-06,
      "loss": 0.1662,
      "step": 1330
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.3660130718954253e-06,
      "loss": 0.2396,
      "step": 1340
    },
    {
      "epoch": 3.97,
      "learning_rate": 3.349673202614379e-06,
      "loss": 0.3285,
      "step": 1350
    },
    {
      "epoch": 4.0,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.1307,
      "step": 1360
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9115913555992141,
      "eval_f1": 0.9103864126710947,
      "eval_loss": 0.41891810297966003,
      "eval_precision": 0.9193377642626458,
      "eval_recall": 0.9140736604791629,
      "eval_runtime": 158.7694,
      "eval_samples_per_second": 6.412,
      "eval_steps_per_second": 0.101,
      "step": 1360
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.3169934640522875e-06,
      "loss": 0.1055,
      "step": 1370
    },
    {
      "epoch": 4.06,
      "learning_rate": 3.300653594771242e-06,
      "loss": 0.1316,
      "step": 1380
    },
    {
      "epoch": 4.09,
      "learning_rate": 3.2843137254901964e-06,
      "loss": 0.1568,
      "step": 1390
    },
    {
      "epoch": 4.12,
      "learning_rate": 3.2679738562091506e-06,
      "loss": 0.1172,
      "step": 1400
    },
    {
      "epoch": 4.15,
      "learning_rate": 3.2516339869281048e-06,
      "loss": 0.1561,
      "step": 1410
    },
    {
      "epoch": 4.18,
      "learning_rate": 3.2352941176470594e-06,
      "loss": 0.1353,
      "step": 1420
    },
    {
      "epoch": 4.21,
      "learning_rate": 3.2189542483660136e-06,
      "loss": 0.1428,
      "step": 1430
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.2026143790849674e-06,
      "loss": 0.1704,
      "step": 1440
    },
    {
      "epoch": 4.26,
      "learning_rate": 3.1862745098039216e-06,
      "loss": 0.1453,
      "step": 1450
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.1699346405228758e-06,
      "loss": 0.135,
      "step": 1460
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.1535947712418304e-06,
      "loss": 0.1236,
      "step": 1470
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.1372549019607846e-06,
      "loss": 0.2047,
      "step": 1480
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.120915032679739e-06,
      "loss": 0.1028,
      "step": 1490
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.104575163398693e-06,
      "loss": 0.1737,
      "step": 1500
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.0882352941176476e-06,
      "loss": 0.1892,
      "step": 1510
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.071895424836602e-06,
      "loss": 0.1097,
      "step": 1520
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.055555555555556e-06,
      "loss": 0.1168,
      "step": 1530
    },
    {
      "epoch": 4.53,
      "learning_rate": 3.03921568627451e-06,
      "loss": 0.1019,
      "step": 1540
    },
    {
      "epoch": 4.56,
      "learning_rate": 3.022875816993464e-06,
      "loss": 0.0835,
      "step": 1550
    },
    {
      "epoch": 4.59,
      "learning_rate": 3.0065359477124182e-06,
      "loss": 0.1283,
      "step": 1560
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.990196078431373e-06,
      "loss": 0.1213,
      "step": 1570
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.973856209150327e-06,
      "loss": 0.1162,
      "step": 1580
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.9575163398692813e-06,
      "loss": 0.0795,
      "step": 1590
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.1179,
      "step": 1600
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.92483660130719e-06,
      "loss": 0.0731,
      "step": 1610
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.9084967320261443e-06,
      "loss": 0.2145,
      "step": 1620
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.8921568627450985e-06,
      "loss": 0.1445,
      "step": 1630
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.8758169934640523e-06,
      "loss": 0.0867,
      "step": 1640
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.8594771241830065e-06,
      "loss": 0.0668,
      "step": 1650
    },
    {
      "epoch": 4.88,
      "learning_rate": 2.843137254901961e-06,
      "loss": 0.1466,
      "step": 1660
    },
    {
      "epoch": 4.91,
      "learning_rate": 2.8267973856209153e-06,
      "loss": 0.1207,
      "step": 1670
    },
    {
      "epoch": 4.94,
      "learning_rate": 2.8104575163398695e-06,
      "loss": 0.0815,
      "step": 1680
    },
    {
      "epoch": 4.97,
      "learning_rate": 2.7941176470588237e-06,
      "loss": 0.0874,
      "step": 1690
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.1169,
      "step": 1700
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.925343811394892,
      "eval_f1": 0.9263233067835679,
      "eval_loss": 0.3335131108760834,
      "eval_precision": 0.9302145251196247,
      "eval_recall": 0.9311557041318327,
      "eval_runtime": 160.0601,
      "eval_samples_per_second": 6.36,
      "eval_steps_per_second": 0.1,
      "step": 1700
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.7614379084967325e-06,
      "loss": 0.0505,
      "step": 1710
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.7450980392156867e-06,
      "loss": 0.0733,
      "step": 1720
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.7287581699346405e-06,
      "loss": 0.0911,
      "step": 1730
    },
    {
      "epoch": 5.12,
      "learning_rate": 2.7124183006535947e-06,
      "loss": 0.0365,
      "step": 1740
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.696078431372549e-06,
      "loss": 0.0712,
      "step": 1750
    },
    {
      "epoch": 5.18,
      "learning_rate": 2.6797385620915036e-06,
      "loss": 0.0747,
      "step": 1760
    },
    {
      "epoch": 5.21,
      "learning_rate": 2.6633986928104578e-06,
      "loss": 0.0877,
      "step": 1770
    },
    {
      "epoch": 5.24,
      "learning_rate": 2.647058823529412e-06,
      "loss": 0.0526,
      "step": 1780
    },
    {
      "epoch": 5.26,
      "learning_rate": 2.630718954248366e-06,
      "loss": 0.0697,
      "step": 1790
    },
    {
      "epoch": 5.29,
      "learning_rate": 2.6143790849673208e-06,
      "loss": 0.0822,
      "step": 1800
    },
    {
      "epoch": 5.32,
      "learning_rate": 2.598039215686275e-06,
      "loss": 0.1042,
      "step": 1810
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.581699346405229e-06,
      "loss": 0.0703,
      "step": 1820
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.565359477124183e-06,
      "loss": 0.046,
      "step": 1830
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.549019607843137e-06,
      "loss": 0.0904,
      "step": 1840
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.532679738562092e-06,
      "loss": 0.0404,
      "step": 1850
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.516339869281046e-06,
      "loss": 0.042,
      "step": 1860
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.5e-06,
      "loss": 0.0745,
      "step": 1870
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.4836601307189544e-06,
      "loss": 0.0605,
      "step": 1880
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.4673202614379086e-06,
      "loss": 0.0431,
      "step": 1890
    },
    {
      "epoch": 5.59,
      "learning_rate": 2.450980392156863e-06,
      "loss": 0.0383,
      "step": 1900
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.434640522875817e-06,
      "loss": 0.0365,
      "step": 1910
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.4183006535947716e-06,
      "loss": 0.078,
      "step": 1920
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.401960784313726e-06,
      "loss": 0.0478,
      "step": 1930
    },
    {
      "epoch": 5.71,
      "learning_rate": 2.38562091503268e-06,
      "loss": 0.1199,
      "step": 1940
    },
    {
      "epoch": 5.74,
      "learning_rate": 2.3692810457516342e-06,
      "loss": 0.0487,
      "step": 1950
    },
    {
      "epoch": 5.76,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 0.0704,
      "step": 1960
    },
    {
      "epoch": 5.79,
      "learning_rate": 2.3366013071895427e-06,
      "loss": 0.0686,
      "step": 1970
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.320261437908497e-06,
      "loss": 0.0235,
      "step": 1980
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.303921568627451e-06,
      "loss": 0.0729,
      "step": 1990
    },
    {
      "epoch": 5.88,
      "learning_rate": 2.2875816993464053e-06,
      "loss": 0.0844,
      "step": 2000
    },
    {
      "epoch": 5.91,
      "learning_rate": 2.2712418300653595e-06,
      "loss": 0.0491,
      "step": 2010
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.254901960784314e-06,
      "loss": 0.111,
      "step": 2020
    },
    {
      "epoch": 5.97,
      "learning_rate": 2.2385620915032683e-06,
      "loss": 0.0728,
      "step": 2030
    },
    {
      "epoch": 6.0,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0621,
      "step": 2040
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.93713163064833,
      "eval_f1": 0.9356746512723884,
      "eval_loss": 0.2819439172744751,
      "eval_precision": 0.9410873053457958,
      "eval_recall": 0.9364388610680477,
      "eval_runtime": 157.7429,
      "eval_samples_per_second": 6.454,
      "eval_steps_per_second": 0.101,
      "step": 2040
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.2058823529411767e-06,
      "loss": 0.0192,
      "step": 2050
    },
    {
      "epoch": 6.06,
      "learning_rate": 2.189542483660131e-06,
      "loss": 0.0247,
      "step": 2060
    },
    {
      "epoch": 6.09,
      "learning_rate": 2.173202614379085e-06,
      "loss": 0.1012,
      "step": 2070
    },
    {
      "epoch": 6.12,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 0.0288,
      "step": 2080
    },
    {
      "epoch": 6.15,
      "learning_rate": 2.1405228758169935e-06,
      "loss": 0.0221,
      "step": 2090
    },
    {
      "epoch": 6.18,
      "learning_rate": 2.1241830065359477e-06,
      "loss": 0.0328,
      "step": 2100
    },
    {
      "epoch": 6.21,
      "learning_rate": 2.1078431372549023e-06,
      "loss": 0.0321,
      "step": 2110
    },
    {
      "epoch": 6.24,
      "learning_rate": 2.0915032679738565e-06,
      "loss": 0.0342,
      "step": 2120
    },
    {
      "epoch": 6.26,
      "learning_rate": 2.0751633986928107e-06,
      "loss": 0.0444,
      "step": 2130
    },
    {
      "epoch": 6.29,
      "learning_rate": 2.058823529411765e-06,
      "loss": 0.0257,
      "step": 2140
    },
    {
      "epoch": 6.32,
      "learning_rate": 2.042483660130719e-06,
      "loss": 0.0236,
      "step": 2150
    },
    {
      "epoch": 6.35,
      "learning_rate": 2.0261437908496734e-06,
      "loss": 0.0301,
      "step": 2160
    },
    {
      "epoch": 6.38,
      "learning_rate": 2.0098039215686276e-06,
      "loss": 0.0234,
      "step": 2170
    },
    {
      "epoch": 6.41,
      "learning_rate": 1.993464052287582e-06,
      "loss": 0.0431,
      "step": 2180
    },
    {
      "epoch": 6.44,
      "learning_rate": 1.977124183006536e-06,
      "loss": 0.0333,
      "step": 2190
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.96078431372549e-06,
      "loss": 0.0195,
      "step": 2200
    },
    {
      "epoch": 6.5,
      "learning_rate": 1.944444444444445e-06,
      "loss": 0.0236,
      "step": 2210
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.928104575163399e-06,
      "loss": 0.0633,
      "step": 2220
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.9117647058823528e-06,
      "loss": 0.0239,
      "step": 2230
    },
    {
      "epoch": 6.59,
      "learning_rate": 1.8954248366013072e-06,
      "loss": 0.0257,
      "step": 2240
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.8790849673202616e-06,
      "loss": 0.0247,
      "step": 2250
    },
    {
      "epoch": 6.65,
      "learning_rate": 1.8627450980392158e-06,
      "loss": 0.0224,
      "step": 2260
    },
    {
      "epoch": 6.68,
      "learning_rate": 1.8464052287581702e-06,
      "loss": 0.049,
      "step": 2270
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.8300653594771242e-06,
      "loss": 0.06,
      "step": 2280
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8137254901960786e-06,
      "loss": 0.0273,
      "step": 2290
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.7973856209150328e-06,
      "loss": 0.0399,
      "step": 2300
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.781045751633987e-06,
      "loss": 0.0958,
      "step": 2310
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 0.034,
      "step": 2320
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.7483660130718954e-06,
      "loss": 0.0299,
      "step": 2330
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.7320261437908499e-06,
      "loss": 0.018,
      "step": 2340
    },
    {
      "epoch": 6.91,
      "learning_rate": 1.715686274509804e-06,
      "loss": 0.0149,
      "step": 2350
    },
    {
      "epoch": 6.94,
      "learning_rate": 1.6993464052287585e-06,
      "loss": 0.0648,
      "step": 2360
    },
    {
      "epoch": 6.97,
      "learning_rate": 1.6830065359477127e-06,
      "loss": 0.0553,
      "step": 2370
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.0287,
      "step": 2380
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9400785854616895,
      "eval_f1": 0.9397297660899341,
      "eval_loss": 0.25110480189323425,
      "eval_precision": 0.9427838093056656,
      "eval_recall": 0.9422498990022914,
      "eval_runtime": 163.249,
      "eval_samples_per_second": 6.236,
      "eval_steps_per_second": 0.098,
      "step": 2380
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.650326797385621e-06,
      "loss": 0.0135,
      "step": 2390
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.6339869281045753e-06,
      "loss": 0.0134,
      "step": 2400
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.6176470588235297e-06,
      "loss": 0.0141,
      "step": 2410
    },
    {
      "epoch": 7.12,
      "learning_rate": 1.6013071895424837e-06,
      "loss": 0.0203,
      "step": 2420
    },
    {
      "epoch": 7.15,
      "learning_rate": 1.5849673202614379e-06,
      "loss": 0.0157,
      "step": 2430
    },
    {
      "epoch": 7.18,
      "learning_rate": 1.5686274509803923e-06,
      "loss": 0.0126,
      "step": 2440
    },
    {
      "epoch": 7.21,
      "learning_rate": 1.5522875816993465e-06,
      "loss": 0.0264,
      "step": 2450
    },
    {
      "epoch": 7.24,
      "learning_rate": 1.535947712418301e-06,
      "loss": 0.0347,
      "step": 2460
    },
    {
      "epoch": 7.26,
      "learning_rate": 1.519607843137255e-06,
      "loss": 0.0143,
      "step": 2470
    },
    {
      "epoch": 7.29,
      "learning_rate": 1.5032679738562091e-06,
      "loss": 0.0108,
      "step": 2480
    },
    {
      "epoch": 7.32,
      "learning_rate": 1.4869281045751635e-06,
      "loss": 0.0214,
      "step": 2490
    },
    {
      "epoch": 7.35,
      "learning_rate": 1.4705882352941177e-06,
      "loss": 0.0145,
      "step": 2500
    },
    {
      "epoch": 7.38,
      "learning_rate": 1.4542483660130721e-06,
      "loss": 0.0115,
      "step": 2510
    },
    {
      "epoch": 7.41,
      "learning_rate": 1.4379084967320261e-06,
      "loss": 0.0143,
      "step": 2520
    },
    {
      "epoch": 7.44,
      "learning_rate": 1.4215686274509805e-06,
      "loss": 0.0191,
      "step": 2530
    },
    {
      "epoch": 7.47,
      "learning_rate": 1.4052287581699348e-06,
      "loss": 0.0413,
      "step": 2540
    },
    {
      "epoch": 7.5,
      "learning_rate": 1.3888888888888892e-06,
      "loss": 0.0133,
      "step": 2550
    },
    {
      "epoch": 7.53,
      "learning_rate": 1.3725490196078434e-06,
      "loss": 0.0102,
      "step": 2560
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.3562091503267974e-06,
      "loss": 0.0381,
      "step": 2570
    },
    {
      "epoch": 7.59,
      "learning_rate": 1.3398692810457518e-06,
      "loss": 0.0157,
      "step": 2580
    },
    {
      "epoch": 7.62,
      "learning_rate": 1.323529411764706e-06,
      "loss": 0.0163,
      "step": 2590
    },
    {
      "epoch": 7.65,
      "learning_rate": 1.3071895424836604e-06,
      "loss": 0.015,
      "step": 2600
    },
    {
      "epoch": 7.68,
      "learning_rate": 1.2908496732026146e-06,
      "loss": 0.0553,
      "step": 2610
    },
    {
      "epoch": 7.71,
      "learning_rate": 1.2745098039215686e-06,
      "loss": 0.018,
      "step": 2620
    },
    {
      "epoch": 7.74,
      "learning_rate": 1.258169934640523e-06,
      "loss": 0.0161,
      "step": 2630
    },
    {
      "epoch": 7.76,
      "learning_rate": 1.2418300653594772e-06,
      "loss": 0.0256,
      "step": 2640
    },
    {
      "epoch": 7.79,
      "learning_rate": 1.2254901960784314e-06,
      "loss": 0.0095,
      "step": 2650
    },
    {
      "epoch": 7.82,
      "learning_rate": 1.2091503267973858e-06,
      "loss": 0.0353,
      "step": 2660
    },
    {
      "epoch": 7.85,
      "learning_rate": 1.19281045751634e-06,
      "loss": 0.065,
      "step": 2670
    },
    {
      "epoch": 7.88,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 0.0264,
      "step": 2680
    },
    {
      "epoch": 7.91,
      "learning_rate": 1.1601307189542484e-06,
      "loss": 0.0355,
      "step": 2690
    },
    {
      "epoch": 7.94,
      "learning_rate": 1.1437908496732026e-06,
      "loss": 0.0498,
      "step": 2700
    },
    {
      "epoch": 7.97,
      "learning_rate": 1.127450980392157e-06,
      "loss": 0.0101,
      "step": 2710
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0188,
      "step": 2720
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9400785854616895,
      "eval_f1": 0.9395279559278015,
      "eval_loss": 0.2341359555721283,
      "eval_precision": 0.9437460647329915,
      "eval_recall": 0.9404723331338164,
      "eval_runtime": 159.7168,
      "eval_samples_per_second": 6.374,
      "eval_steps_per_second": 0.1,
      "step": 2720
    },
    {
      "epoch": 8.03,
      "learning_rate": 1.0947712418300655e-06,
      "loss": 0.0077,
      "step": 2730
    },
    {
      "epoch": 8.06,
      "learning_rate": 1.0784313725490197e-06,
      "loss": 0.0117,
      "step": 2740
    },
    {
      "epoch": 8.09,
      "learning_rate": 1.0620915032679739e-06,
      "loss": 0.0102,
      "step": 2750
    },
    {
      "epoch": 8.12,
      "learning_rate": 1.0457516339869283e-06,
      "loss": 0.0256,
      "step": 2760
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.0294117647058825e-06,
      "loss": 0.0111,
      "step": 2770
    },
    {
      "epoch": 8.18,
      "learning_rate": 1.0130718954248367e-06,
      "loss": 0.0144,
      "step": 2780
    },
    {
      "epoch": 8.21,
      "learning_rate": 9.96732026143791e-07,
      "loss": 0.0084,
      "step": 2790
    },
    {
      "epoch": 8.24,
      "learning_rate": 9.80392156862745e-07,
      "loss": 0.0061,
      "step": 2800
    },
    {
      "epoch": 8.26,
      "learning_rate": 9.640522875816995e-07,
      "loss": 0.054,
      "step": 2810
    },
    {
      "epoch": 8.29,
      "learning_rate": 9.477124183006536e-07,
      "loss": 0.0126,
      "step": 2820
    },
    {
      "epoch": 8.32,
      "learning_rate": 9.313725490196079e-07,
      "loss": 0.0068,
      "step": 2830
    },
    {
      "epoch": 8.35,
      "learning_rate": 9.150326797385621e-07,
      "loss": 0.0108,
      "step": 2840
    },
    {
      "epoch": 8.38,
      "learning_rate": 8.986928104575164e-07,
      "loss": 0.0074,
      "step": 2850
    },
    {
      "epoch": 8.41,
      "learning_rate": 8.823529411764707e-07,
      "loss": 0.0194,
      "step": 2860
    },
    {
      "epoch": 8.44,
      "learning_rate": 8.660130718954249e-07,
      "loss": 0.0112,
      "step": 2870
    },
    {
      "epoch": 8.47,
      "learning_rate": 8.496732026143792e-07,
      "loss": 0.0133,
      "step": 2880
    },
    {
      "epoch": 8.5,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0138,
      "step": 2890
    },
    {
      "epoch": 8.53,
      "learning_rate": 8.169934640522876e-07,
      "loss": 0.0261,
      "step": 2900
    },
    {
      "epoch": 8.56,
      "learning_rate": 8.006535947712418e-07,
      "loss": 0.0151,
      "step": 2910
    },
    {
      "epoch": 8.59,
      "learning_rate": 7.843137254901962e-07,
      "loss": 0.0176,
      "step": 2920
    },
    {
      "epoch": 8.62,
      "learning_rate": 7.679738562091505e-07,
      "loss": 0.008,
      "step": 2930
    },
    {
      "epoch": 8.65,
      "learning_rate": 7.516339869281046e-07,
      "loss": 0.0091,
      "step": 2940
    },
    {
      "epoch": 8.68,
      "learning_rate": 7.352941176470589e-07,
      "loss": 0.0124,
      "step": 2950
    },
    {
      "epoch": 8.71,
      "learning_rate": 7.189542483660131e-07,
      "loss": 0.0313,
      "step": 2960
    },
    {
      "epoch": 8.74,
      "learning_rate": 7.026143790849674e-07,
      "loss": 0.0286,
      "step": 2970
    },
    {
      "epoch": 8.76,
      "learning_rate": 6.862745098039217e-07,
      "loss": 0.0098,
      "step": 2980
    },
    {
      "epoch": 8.79,
      "learning_rate": 6.699346405228759e-07,
      "loss": 0.0067,
      "step": 2990
    },
    {
      "epoch": 8.82,
      "learning_rate": 6.535947712418302e-07,
      "loss": 0.0081,
      "step": 3000
    },
    {
      "epoch": 8.85,
      "learning_rate": 6.372549019607843e-07,
      "loss": 0.0102,
      "step": 3010
    },
    {
      "epoch": 8.88,
      "learning_rate": 6.209150326797386e-07,
      "loss": 0.0256,
      "step": 3020
    },
    {
      "epoch": 8.91,
      "learning_rate": 6.045751633986929e-07,
      "loss": 0.0117,
      "step": 3030
    },
    {
      "epoch": 8.94,
      "learning_rate": 5.882352941176471e-07,
      "loss": 0.0079,
      "step": 3040
    },
    {
      "epoch": 8.97,
      "learning_rate": 5.718954248366013e-07,
      "loss": 0.017,
      "step": 3050
    },
    {
      "epoch": 9.0,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.0217,
      "step": 3060
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9420432220039293,
      "eval_f1": 0.942289956941725,
      "eval_loss": 0.2249804437160492,
      "eval_precision": 0.9452784201539948,
      "eval_recall": 0.9442615812771316,
      "eval_runtime": 162.0726,
      "eval_samples_per_second": 6.281,
      "eval_steps_per_second": 0.099,
      "step": 3060
    }
  ],
  "max_steps": 3400,
  "num_train_epochs": 10,
  "total_flos": 1.449075129002754e+18,
  "trial_name": null,
  "trial_params": null
}