{
  "best_metric": 0.9186971783638,
  "best_model_checkpoint": "./Finetuned_Classification_model_200k/checkpoint-500",
  "epoch": 0.96,
  "eval_steps": 500,
  "global_step": 6000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.037,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 2e-05,
      "loss": 0.907,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 3e-05,
      "loss": 0.7159,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 4e-05,
      "loss": 0.5758,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-05,
      "loss": 0.3454,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.997326203208557e-05,
      "loss": 0.1851,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.994652406417113e-05,
      "loss": 0.2821,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9919786096256686e-05,
      "loss": 0.1741,
      "step": 80
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9893048128342245e-05,
      "loss": 0.1749,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.986631016042781e-05,
      "loss": 0.2407,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.983957219251337e-05,
      "loss": 0.2612,
      "step": 110
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.981283422459893e-05,
      "loss": 0.1677,
      "step": 120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9786096256684495e-05,
      "loss": 0.1123,
      "step": 130
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.975935828877006e-05,
      "loss": 0.1368,
      "step": 140
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.973262032085561e-05,
      "loss": 0.1449,
      "step": 150
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.970588235294118e-05,
      "loss": 0.152,
      "step": 160
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.967914438502674e-05,
      "loss": 0.2504,
      "step": 170
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9652406417112304e-05,
      "loss": 0.2578,
      "step": 180
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.962566844919786e-05,
      "loss": 0.1289,
      "step": 190
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.959893048128342e-05,
      "loss": 0.1263,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.957219251336899e-05,
      "loss": 0.1031,
      "step": 210
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9545454545454553e-05,
      "loss": 0.2453,
      "step": 220
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9518716577540106e-05,
      "loss": 0.0675,
      "step": 230
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.949197860962567e-05,
      "loss": 0.1807,
      "step": 240
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.946524064171123e-05,
      "loss": 0.1493,
      "step": 250
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9438502673796796e-05,
      "loss": 0.109,
      "step": 260
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9411764705882355e-05,
      "loss": 0.2148,
      "step": 270
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9385026737967914e-05,
      "loss": 0.2574,
      "step": 280
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.935828877005348e-05,
      "loss": 0.2198,
      "step": 290
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.933155080213904e-05,
      "loss": 0.1574,
      "step": 300
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.93048128342246e-05,
      "loss": 0.0866,
      "step": 310
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9278074866310164e-05,
      "loss": 0.1371,
      "step": 320
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.925133689839572e-05,
      "loss": 0.2074,
      "step": 330
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.922459893048129e-05,
      "loss": 0.3628,
      "step": 340
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.919786096256685e-05,
      "loss": 0.1888,
      "step": 350
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.917112299465241e-05,
      "loss": 0.0934,
      "step": 360
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.914438502673797e-05,
      "loss": 0.1593,
      "step": 370
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.911764705882353e-05,
      "loss": 0.0721,
      "step": 380
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.909090909090909e-05,
      "loss": 0.1139,
      "step": 390
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.906417112299466e-05,
      "loss": 0.1118,
      "step": 400
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9037433155080216e-05,
      "loss": 0.218,
      "step": 410
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.901069518716578e-05,
      "loss": 0.126,
      "step": 420
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.898395721925134e-05,
      "loss": 0.1057,
      "step": 430
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.89572192513369e-05,
      "loss": 0.1825,
      "step": 440
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8930481283422465e-05,
      "loss": 0.1379,
      "step": 450
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8903743315508024e-05,
      "loss": 0.1311,
      "step": 460
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.8877005347593584e-05,
      "loss": 0.0722,
      "step": 470
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.885026737967915e-05,
      "loss": 0.1286,
      "step": 480
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.882352941176471e-05,
      "loss": 0.1534,
      "step": 490
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.879679144385027e-05,
      "loss": 0.1332,
      "step": 500
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.8395015105740181,
      "eval_f1": 0.8361086180158485,
      "eval_loss": 0.9186971783638,
      "eval_precision": 0.8701532825604937,
      "eval_recall": 0.8395015105740181,
      "eval_runtime": 10.5139,
      "eval_samples_per_second": 251.857,
      "eval_steps_per_second": 3.995,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.8770053475935826e-05,
      "loss": 0.1317,
      "step": 510
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.874331550802139e-05,
      "loss": 0.0683,
      "step": 520
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.871657754010696e-05,
      "loss": 0.0474,
      "step": 530
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.868983957219252e-05,
      "loss": 0.2302,
      "step": 540
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8663101604278076e-05,
      "loss": 0.1788,
      "step": 550
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.863636363636364e-05,
      "loss": 0.0917,
      "step": 560
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.86096256684492e-05,
      "loss": 0.2001,
      "step": 570
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.858288770053476e-05,
      "loss": 0.3409,
      "step": 580
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.855614973262032e-05,
      "loss": 0.2044,
      "step": 590
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8529411764705885e-05,
      "loss": 0.1292,
      "step": 600
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.850267379679145e-05,
      "loss": 0.1422,
      "step": 610
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8475935828877e-05,
      "loss": 0.1165,
      "step": 620
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.844919786096257e-05,
      "loss": 0.1115,
      "step": 630
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8422459893048135e-05,
      "loss": 0.0957,
      "step": 640
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8395721925133694e-05,
      "loss": 0.1791,
      "step": 650
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.836898395721925e-05,
      "loss": 0.0691,
      "step": 660
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.834224598930481e-05,
      "loss": 0.2829,
      "step": 670
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.831550802139038e-05,
      "loss": 0.2635,
      "step": 680
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.828877005347594e-05,
      "loss": 0.1722,
      "step": 690
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8262032085561496e-05,
      "loss": 0.0974,
      "step": 700
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.823529411764706e-05,
      "loss": 0.0888,
      "step": 710
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820855614973262e-05,
      "loss": 0.3148,
      "step": 720
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8181818181818186e-05,
      "loss": 0.3556,
      "step": 730
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8155080213903745e-05,
      "loss": 0.1788,
      "step": 740
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8128342245989304e-05,
      "loss": 0.4376,
      "step": 750
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.810160427807487e-05,
      "loss": 0.2413,
      "step": 760
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.807486631016043e-05,
      "loss": 0.1802,
      "step": 770
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.804812834224599e-05,
      "loss": 0.1364,
      "step": 780
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8021390374331554e-05,
      "loss": 0.0667,
      "step": 790
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799465240641711e-05,
      "loss": 0.1409,
      "step": 800
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.796791443850268e-05,
      "loss": 0.0897,
      "step": 810
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794117647058824e-05,
      "loss": 0.0871,
      "step": 820
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.79144385026738e-05,
      "loss": 0.0552,
      "step": 830
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788770053475936e-05,
      "loss": 0.2569,
      "step": 840
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786096256684492e-05,
      "loss": 0.2877,
      "step": 850
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.783422459893048e-05,
      "loss": 0.1365,
      "step": 860
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7807486631016047e-05,
      "loss": 0.1604,
      "step": 870
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7780748663101606e-05,
      "loss": 0.2179,
      "step": 880
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7754010695187165e-05,
      "loss": 0.0969,
      "step": 890
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772727272727273e-05,
      "loss": 0.0953,
      "step": 900
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.770053475935829e-05,
      "loss": 0.0803,
      "step": 910
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7673796791443855e-05,
      "loss": 0.1573,
      "step": 920
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7647058823529414e-05,
      "loss": 0.1388,
      "step": 930
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7620320855614973e-05,
      "loss": 0.1559,
      "step": 940
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.759358288770054e-05,
      "loss": 0.2411,
      "step": 950
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.75668449197861e-05,
      "loss": 0.1119,
      "step": 960
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.754010695187166e-05,
      "loss": 0.1416,
      "step": 970
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.751336898395722e-05,
      "loss": 0.1073,
      "step": 980
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.748663101604278e-05,
      "loss": 0.1799,
      "step": 990
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.745989304812835e-05,
      "loss": 0.127,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.847809667673716,
      "eval_f1": 0.8460260531471735,
      "eval_loss": 1.017073392868042,
      "eval_precision": 0.8647086198841812,
      "eval_recall": 0.847809667673716,
      "eval_runtime": 10.5379,
      "eval_samples_per_second": 251.284,
      "eval_steps_per_second": 3.986,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.74331550802139e-05,
      "loss": 0.1774,
      "step": 1010
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.7406417112299466e-05,
      "loss": 0.0869,
      "step": 1020
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.737967914438503e-05,
      "loss": 0.0686,
      "step": 1030
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.735294117647059e-05,
      "loss": 0.124,
      "step": 1040
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.732620320855615e-05,
      "loss": 0.0993,
      "step": 1050
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7299465240641716e-05,
      "loss": 0.2451,
      "step": 1060
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7272727272727275e-05,
      "loss": 0.1607,
      "step": 1070
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.724598930481284e-05,
      "loss": 0.1289,
      "step": 1080
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.721925133689839e-05,
      "loss": 0.1813,
      "step": 1090
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.719251336898396e-05,
      "loss": 0.1628,
      "step": 1100
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7165775401069524e-05,
      "loss": 0.3152,
      "step": 1110
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7139037433155083e-05,
      "loss": 0.3835,
      "step": 1120
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.711229946524064e-05,
      "loss": 0.1713,
      "step": 1130
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.70855614973262e-05,
      "loss": 0.1812,
      "step": 1140
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.705882352941177e-05,
      "loss": 0.1928,
      "step": 1150
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.703208556149733e-05,
      "loss": 0.2963,
      "step": 1160
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.7005347593582885e-05,
      "loss": 0.2073,
      "step": 1170
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.697860962566845e-05,
      "loss": 0.2299,
      "step": 1180
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.695187165775402e-05,
      "loss": 0.1832,
      "step": 1190
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6925133689839576e-05,
      "loss": 0.1844,
      "step": 1200
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6898395721925135e-05,
      "loss": 0.4959,
      "step": 1210
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6871657754010694e-05,
      "loss": 0.2516,
      "step": 1220
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.684491978609626e-05,
      "loss": 0.1754,
      "step": 1230
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.681818181818182e-05,
      "loss": 0.2554,
      "step": 1240
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.679144385026738e-05,
      "loss": 0.1976,
      "step": 1250
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6764705882352944e-05,
      "loss": 0.1906,
      "step": 1260
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.673796791443851e-05,
      "loss": 0.0806,
      "step": 1270
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.671122994652407e-05,
      "loss": 0.1229,
      "step": 1280
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.668449197860963e-05,
      "loss": 0.1222,
      "step": 1290
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.665775401069519e-05,
      "loss": 0.1049,
      "step": 1300
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.663101604278075e-05,
      "loss": 0.1548,
      "step": 1310
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.660427807486631e-05,
      "loss": 0.1077,
      "step": 1320
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.657754010695187e-05,
      "loss": 0.1949,
      "step": 1330
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6550802139037436e-05,
      "loss": 0.1349,
      "step": 1340
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6524064171123e-05,
      "loss": 0.223,
      "step": 1350
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6497326203208555e-05,
      "loss": 0.1132,
      "step": 1360
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.647058823529412e-05,
      "loss": 0.123,
      "step": 1370
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.644385026737968e-05,
      "loss": 0.1741,
      "step": 1380
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6417112299465245e-05,
      "loss": 0.0913,
      "step": 1390
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6390374331550804e-05,
      "loss": 0.1428,
      "step": 1400
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.636363636363636e-05,
      "loss": 0.1171,
      "step": 1410
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.633689839572193e-05,
      "loss": 0.1773,
      "step": 1420
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.631016042780749e-05,
      "loss": 0.1372,
      "step": 1430
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.628342245989305e-05,
      "loss": 0.2797,
      "step": 1440
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.625668449197861e-05,
      "loss": 0.2813,
      "step": 1450
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.622994652406417e-05,
      "loss": 0.2861,
      "step": 1460
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.620320855614974e-05,
      "loss": 0.0677,
      "step": 1470
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.61764705882353e-05,
      "loss": 0.104,
      "step": 1480
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.6149732620320856e-05,
      "loss": 0.1533,
      "step": 1490
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.612299465240642e-05,
      "loss": 0.1186,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.8376132930513596,
      "eval_f1": 0.8363352339205664,
      "eval_loss": 1.0860812664031982,
      "eval_precision": 0.8484990166173995,
      "eval_recall": 0.8376132930513596,
      "eval_runtime": 10.5226,
      "eval_samples_per_second": 251.648,
      "eval_steps_per_second": 3.991,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.609625668449198e-05,
      "loss": 0.0683,
      "step": 1510
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.606951871657754e-05,
      "loss": 0.1767,
      "step": 1520
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.6042780748663106e-05,
      "loss": 0.0559,
      "step": 1530
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.6016042780748665e-05,
      "loss": 0.1056,
      "step": 1540
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.598930481283423e-05,
      "loss": 0.1607,
      "step": 1550
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.596256684491979e-05,
      "loss": 0.053,
      "step": 1560
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.593582887700535e-05,
      "loss": 0.2743,
      "step": 1570
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5909090909090914e-05,
      "loss": 0.1569,
      "step": 1580
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.588235294117647e-05,
      "loss": 0.0648,
      "step": 1590
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.585561497326203e-05,
      "loss": 0.132,
      "step": 1600
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.58288770053476e-05,
      "loss": 0.1115,
      "step": 1610
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.580213903743316e-05,
      "loss": 0.2035,
      "step": 1620
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5775401069518716e-05,
      "loss": 0.1193,
      "step": 1630
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5748663101604275e-05,
      "loss": 0.2867,
      "step": 1640
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.572192513368984e-05,
      "loss": 0.2423,
      "step": 1650
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.569518716577541e-05,
      "loss": 0.0959,
      "step": 1660
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5668449197860966e-05,
      "loss": 0.062,
      "step": 1670
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5641711229946525e-05,
      "loss": 0.3052,
      "step": 1680
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.561497326203209e-05,
      "loss": 0.0814,
      "step": 1690
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.558823529411765e-05,
      "loss": 0.4406,
      "step": 1700
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.556149732620321e-05,
      "loss": 0.4316,
      "step": 1710
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.553475935828877e-05,
      "loss": 0.2551,
      "step": 1720
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5508021390374334e-05,
      "loss": 0.4731,
      "step": 1730
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.54812834224599e-05,
      "loss": 0.3468,
      "step": 1740
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.4397,
      "step": 1750
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.542780748663102e-05,
      "loss": 0.2163,
      "step": 1760
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5401069518716583e-05,
      "loss": 0.1675,
      "step": 1770
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.537433155080214e-05,
      "loss": 0.124,
      "step": 1780
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.53475935828877e-05,
      "loss": 0.192,
      "step": 1790
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.532085561497326e-05,
      "loss": 0.2626,
      "step": 1800
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5294117647058826e-05,
      "loss": 0.233,
      "step": 1810
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.526737967914439e-05,
      "loss": 0.2005,
      "step": 1820
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5240641711229944e-05,
      "loss": 0.1611,
      "step": 1830
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.521390374331551e-05,
      "loss": 0.1215,
      "step": 1840
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.518716577540107e-05,
      "loss": 0.1504,
      "step": 1850
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5160427807486635e-05,
      "loss": 0.1317,
      "step": 1860
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5133689839572194e-05,
      "loss": 0.1624,
      "step": 1870
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.510695187165775e-05,
      "loss": 0.1815,
      "step": 1880
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.508021390374332e-05,
      "loss": 0.2438,
      "step": 1890
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5053475935828885e-05,
      "loss": 0.1908,
      "step": 1900
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.502673796791444e-05,
      "loss": 0.2333,
      "step": 1910
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.5e-05,
      "loss": 0.2385,
      "step": 1920
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.497326203208556e-05,
      "loss": 0.1889,
      "step": 1930
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.494652406417113e-05,
      "loss": 0.1202,
      "step": 1940
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.491978609625669e-05,
      "loss": 0.3926,
      "step": 1950
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4893048128342246e-05,
      "loss": 0.3656,
      "step": 1960
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.486631016042781e-05,
      "loss": 0.1626,
      "step": 1970
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.483957219251337e-05,
      "loss": 0.1318,
      "step": 1980
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.481283422459893e-05,
      "loss": 0.2922,
      "step": 1990
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4786096256684495e-05,
      "loss": 0.3545,
      "step": 2000
    },
    {
      "epoch": 0.32,
      "eval_accuracy": 0.8413897280966768,
      "eval_f1": 0.8394373029347564,
      "eval_loss": 2.3635265827178955,
      "eval_precision": 0.8588437332105303,
      "eval_recall": 0.8413897280966768,
      "eval_runtime": 10.5481,
      "eval_samples_per_second": 251.039,
      "eval_steps_per_second": 3.982,
      "step": 2000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4759358288770054e-05,
      "loss": 0.1883,
      "step": 2010
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.473262032085562e-05,
      "loss": 0.4851,
      "step": 2020
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.470588235294118e-05,
      "loss": 0.4886,
      "step": 2030
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.467914438502674e-05,
      "loss": 0.2792,
      "step": 2040
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4652406417112304e-05,
      "loss": 0.2475,
      "step": 2050
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.462566844919786e-05,
      "loss": 0.1728,
      "step": 2060
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.459893048128342e-05,
      "loss": 0.0873,
      "step": 2070
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.457219251336899e-05,
      "loss": 0.1553,
      "step": 2080
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.454545454545455e-05,
      "loss": 0.3127,
      "step": 2090
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.4518716577540106e-05,
      "loss": 0.4534,
      "step": 2100
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.449197860962567e-05,
      "loss": 0.2118,
      "step": 2110
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.446524064171123e-05,
      "loss": 0.3214,
      "step": 2120
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.44385026737968e-05,
      "loss": 0.1867,
      "step": 2130
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.4411764705882356e-05,
      "loss": 0.3318,
      "step": 2140
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.4385026737967915e-05,
      "loss": 0.164,
      "step": 2150
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.435828877005348e-05,
      "loss": 0.2003,
      "step": 2160
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.433155080213904e-05,
      "loss": 0.1854,
      "step": 2170
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.43048128342246e-05,
      "loss": 0.2871,
      "step": 2180
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4278074866310165e-05,
      "loss": 0.306,
      "step": 2190
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4251336898395724e-05,
      "loss": 0.462,
      "step": 2200
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.422459893048129e-05,
      "loss": 0.3016,
      "step": 2210
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.419786096256684e-05,
      "loss": 0.2322,
      "step": 2220
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.417112299465241e-05,
      "loss": 0.141,
      "step": 2230
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.414438502673797e-05,
      "loss": 0.1501,
      "step": 2240
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.411764705882353e-05,
      "loss": 0.2379,
      "step": 2250
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.409090909090909e-05,
      "loss": 0.2233,
      "step": 2260
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.406417112299465e-05,
      "loss": 0.4372,
      "step": 2270
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.4037433155080216e-05,
      "loss": 0.5062,
      "step": 2280
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.401069518716578e-05,
      "loss": 0.1763,
      "step": 2290
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3983957219251334e-05,
      "loss": 0.1486,
      "step": 2300
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.39572192513369e-05,
      "loss": 0.2072,
      "step": 2310
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3930481283422466e-05,
      "loss": 0.0901,
      "step": 2320
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3903743315508025e-05,
      "loss": 0.351,
      "step": 2330
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3877005347593584e-05,
      "loss": 0.2836,
      "step": 2340
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.385026737967914e-05,
      "loss": 0.3663,
      "step": 2350
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.382352941176471e-05,
      "loss": 0.4,
      "step": 2360
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.3796791443850275e-05,
      "loss": 0.3752,
      "step": 2370
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.377005347593583e-05,
      "loss": 0.22,
      "step": 2380
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.374331550802139e-05,
      "loss": 0.3702,
      "step": 2390
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.371657754010696e-05,
      "loss": 0.1035,
      "step": 2400
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.368983957219252e-05,
      "loss": 0.1581,
      "step": 2410
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3663101604278077e-05,
      "loss": 0.2575,
      "step": 2420
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3636363636363636e-05,
      "loss": 0.1866,
      "step": 2430
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.36096256684492e-05,
      "loss": 0.3816,
      "step": 2440
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.358288770053476e-05,
      "loss": 0.4613,
      "step": 2450
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.355614973262032e-05,
      "loss": 0.1567,
      "step": 2460
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3529411764705885e-05,
      "loss": 0.1287,
      "step": 2470
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3502673796791444e-05,
      "loss": 0.187,
      "step": 2480
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.347593582887701e-05,
      "loss": 0.1354,
      "step": 2490
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.344919786096257e-05,
      "loss": 0.0646,
      "step": 2500
    },
    {
      "epoch": 0.4,
      "eval_accuracy": 0.849320241691843,
      "eval_f1": 0.8472401124107073,
      "eval_loss": 1.3920754194259644,
      "eval_precision": 0.869443045077059,
      "eval_recall": 0.849320241691843,
      "eval_runtime": 10.5377,
      "eval_samples_per_second": 251.287,
      "eval_steps_per_second": 3.986,
      "step": 2500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.342245989304813e-05,
      "loss": 0.1651,
      "step": 2510
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3395721925133694e-05,
      "loss": 0.0982,
      "step": 2520
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.336898395721925e-05,
      "loss": 0.2102,
      "step": 2530
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.334224598930481e-05,
      "loss": 0.2102,
      "step": 2540
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.331550802139038e-05,
      "loss": 0.2738,
      "step": 2550
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.328877005347594e-05,
      "loss": 0.0848,
      "step": 2560
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.3262032085561496e-05,
      "loss": 0.2072,
      "step": 2570
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.323529411764706e-05,
      "loss": 0.6024,
      "step": 2580
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.320855614973262e-05,
      "loss": 0.5888,
      "step": 2590
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.318181818181819e-05,
      "loss": 0.1546,
      "step": 2600
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.3155080213903746e-05,
      "loss": 0.2223,
      "step": 2610
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.3128342245989305e-05,
      "loss": 0.2447,
      "step": 2620
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.310160427807487e-05,
      "loss": 0.1098,
      "step": 2630
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.307486631016043e-05,
      "loss": 0.3385,
      "step": 2640
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.304812834224599e-05,
      "loss": 0.1067,
      "step": 2650
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.3021390374331554e-05,
      "loss": 0.2155,
      "step": 2660
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.2994652406417113e-05,
      "loss": 0.345,
      "step": 2670
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.296791443850268e-05,
      "loss": 0.2231,
      "step": 2680
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.294117647058823e-05,
      "loss": 0.2863,
      "step": 2690
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.29144385026738e-05,
      "loss": 0.211,
      "step": 2700
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.288770053475936e-05,
      "loss": 0.1859,
      "step": 2710
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.286096256684492e-05,
      "loss": 0.4434,
      "step": 2720
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.283422459893048e-05,
      "loss": 0.2067,
      "step": 2730
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.280748663101605e-05,
      "loss": 0.2656,
      "step": 2740
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2780748663101606e-05,
      "loss": 0.3386,
      "step": 2750
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.275401069518717e-05,
      "loss": 0.6852,
      "step": 2760
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2727272727272724e-05,
      "loss": 0.2283,
      "step": 2770
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.270053475935829e-05,
      "loss": 0.2341,
      "step": 2780
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.2673796791443856e-05,
      "loss": 0.268,
      "step": 2790
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.2647058823529415e-05,
      "loss": 0.2749,
      "step": 2800
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.2620320855614974e-05,
      "loss": 0.2868,
      "step": 2810
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.259358288770054e-05,
      "loss": 0.2351,
      "step": 2820
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.25668449197861e-05,
      "loss": 0.156,
      "step": 2830
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.254010695187166e-05,
      "loss": 0.1897,
      "step": 2840
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.251336898395722e-05,
      "loss": 0.2117,
      "step": 2850
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.248663101604278e-05,
      "loss": 0.2102,
      "step": 2860
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.245989304812835e-05,
      "loss": 0.3609,
      "step": 2870
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.243315508021391e-05,
      "loss": 0.2409,
      "step": 2880
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2406417112299466e-05,
      "loss": 0.5209,
      "step": 2890
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2379679144385025e-05,
      "loss": 0.3209,
      "step": 2900
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.235294117647059e-05,
      "loss": 0.3782,
      "step": 2910
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.232620320855615e-05,
      "loss": 0.1297,
      "step": 2920
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.229946524064171e-05,
      "loss": 0.365,
      "step": 2930
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2272727272727275e-05,
      "loss": 0.4153,
      "step": 2940
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.224598930481284e-05,
      "loss": 0.1112,
      "step": 2950
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.221925133689839e-05,
      "loss": 0.1813,
      "step": 2960
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.219251336898396e-05,
      "loss": 0.1488,
      "step": 2970
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.216577540106952e-05,
      "loss": 0.1983,
      "step": 2980
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.2139037433155084e-05,
      "loss": 0.3848,
      "step": 2990
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.211229946524064e-05,
      "loss": 0.3005,
      "step": 3000
    },
    {
      "epoch": 0.48,
      "eval_accuracy": 0.8425226586102719,
      "eval_f1": 0.8406242841944506,
      "eval_loss": 2.313152313232422,
      "eval_precision": 0.8596586910099404,
      "eval_recall": 0.8425226586102719,
      "eval_runtime": 10.5381,
      "eval_samples_per_second": 251.278,
      "eval_steps_per_second": 3.986,
      "step": 3000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.20855614973262e-05,
      "loss": 0.3137,
      "step": 3010
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.205882352941177e-05,
      "loss": 0.3145,
      "step": 3020
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.2032085561497334e-05,
      "loss": 0.1838,
      "step": 3030
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.2005347593582886e-05,
      "loss": 0.328,
      "step": 3040
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.197860962566845e-05,
      "loss": 0.2712,
      "step": 3050
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.195187165775401e-05,
      "loss": 0.2833,
      "step": 3060
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.1925133689839577e-05,
      "loss": 0.425,
      "step": 3070
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.1898395721925136e-05,
      "loss": 0.227,
      "step": 3080
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.1871657754010695e-05,
      "loss": 0.299,
      "step": 3090
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.184491978609626e-05,
      "loss": 0.3831,
      "step": 3100
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.181818181818182e-05,
      "loss": 0.1673,
      "step": 3110
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.179144385026738e-05,
      "loss": 0.2316,
      "step": 3120
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1764705882352944e-05,
      "loss": 0.3018,
      "step": 3130
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.17379679144385e-05,
      "loss": 0.1735,
      "step": 3140
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.171122994652407e-05,
      "loss": 0.3896,
      "step": 3150
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.168449197860963e-05,
      "loss": 0.166,
      "step": 3160
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.165775401069519e-05,
      "loss": 0.1794,
      "step": 3170
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.163101604278075e-05,
      "loss": 0.2399,
      "step": 3180
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.160427807486631e-05,
      "loss": 0.331,
      "step": 3190
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.157754010695187e-05,
      "loss": 0.4568,
      "step": 3200
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.155080213903744e-05,
      "loss": 0.4554,
      "step": 3210
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1524064171122996e-05,
      "loss": 0.3125,
      "step": 3220
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.149732620320856e-05,
      "loss": 0.2818,
      "step": 3230
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.147058823529412e-05,
      "loss": 0.4162,
      "step": 3240
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.144385026737968e-05,
      "loss": 0.4248,
      "step": 3250
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1417112299465246e-05,
      "loss": 0.3776,
      "step": 3260
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1390374331550805e-05,
      "loss": 0.3877,
      "step": 3270
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1363636363636364e-05,
      "loss": 0.0185,
      "step": 3280
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.133689839572193e-05,
      "loss": 0.1879,
      "step": 3290
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.131016042780749e-05,
      "loss": 0.1577,
      "step": 3300
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.128342245989305e-05,
      "loss": 0.342,
      "step": 3310
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.125668449197861e-05,
      "loss": 0.2412,
      "step": 3320
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.122994652406417e-05,
      "loss": 0.2567,
      "step": 3330
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.120320855614974e-05,
      "loss": 0.2748,
      "step": 3340
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.11764705882353e-05,
      "loss": 0.1134,
      "step": 3350
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.1149732620320856e-05,
      "loss": 0.3265,
      "step": 3360
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.112299465240642e-05,
      "loss": 0.2037,
      "step": 3370
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.109625668449198e-05,
      "loss": 0.2894,
      "step": 3380
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.106951871657754e-05,
      "loss": 0.3418,
      "step": 3390
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.10427807486631e-05,
      "loss": 0.3356,
      "step": 3400
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.1016042780748665e-05,
      "loss": 0.228,
      "step": 3410
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.098930481283423e-05,
      "loss": 0.3992,
      "step": 3420
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.096256684491978e-05,
      "loss": 0.2551,
      "step": 3430
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.093582887700535e-05,
      "loss": 0.184,
      "step": 3440
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.3881,
      "step": 3450
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0882352941176474e-05,
      "loss": 0.3422,
      "step": 3460
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.085561497326203e-05,
      "loss": 0.2786,
      "step": 3470
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.082887700534759e-05,
      "loss": 0.2854,
      "step": 3480
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.080213903743316e-05,
      "loss": 0.4252,
      "step": 3490
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0775401069518723e-05,
      "loss": 0.1091,
      "step": 3500
    },
    {
      "epoch": 0.56,
      "eval_accuracy": 0.8459214501510574,
      "eval_f1": 0.843780419456653,
      "eval_loss": 2.0897836685180664,
      "eval_precision": 0.8659851431247721,
      "eval_recall": 0.8459214501510574,
      "eval_runtime": 10.5263,
      "eval_samples_per_second": 251.559,
      "eval_steps_per_second": 3.99,
      "step": 3500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0748663101604276e-05,
      "loss": 0.3208,
      "step": 3510
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.072192513368984e-05,
      "loss": 0.0881,
      "step": 3520
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.06951871657754e-05,
      "loss": 0.1677,
      "step": 3530
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0668449197860966e-05,
      "loss": 0.049,
      "step": 3540
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0641711229946525e-05,
      "loss": 0.1629,
      "step": 3550
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0614973262032084e-05,
      "loss": 0.2262,
      "step": 3560
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.058823529411765e-05,
      "loss": 0.2134,
      "step": 3570
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0561497326203216e-05,
      "loss": 0.3984,
      "step": 3580
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.053475935828877e-05,
      "loss": 0.1527,
      "step": 3590
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.0508021390374334e-05,
      "loss": 0.1532,
      "step": 3600
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.048128342245989e-05,
      "loss": 0.1918,
      "step": 3610
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.045454545454546e-05,
      "loss": 0.2594,
      "step": 3620
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.042780748663102e-05,
      "loss": 0.1803,
      "step": 3630
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.040106951871658e-05,
      "loss": 0.2573,
      "step": 3640
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.037433155080214e-05,
      "loss": 0.2865,
      "step": 3650
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.03475935828877e-05,
      "loss": 0.053,
      "step": 3660
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.032085561497326e-05,
      "loss": 0.4603,
      "step": 3670
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.029411764705883e-05,
      "loss": 0.1409,
      "step": 3680
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0267379679144386e-05,
      "loss": 0.2257,
      "step": 3690
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.024064171122995e-05,
      "loss": 0.2827,
      "step": 3700
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.021390374331551e-05,
      "loss": 0.2166,
      "step": 3710
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.018716577540107e-05,
      "loss": 0.2091,
      "step": 3720
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.0160427807486636e-05,
      "loss": 0.4851,
      "step": 3730
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.0133689839572195e-05,
      "loss": 0.3742,
      "step": 3740
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.0106951871657754e-05,
      "loss": 0.492,
      "step": 3750
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.008021390374332e-05,
      "loss": 0.5621,
      "step": 3760
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.005347593582888e-05,
      "loss": 0.2801,
      "step": 3770
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.002673796791444e-05,
      "loss": 0.2604,
      "step": 3780
    },
    {
      "epoch": 0.61,
      "learning_rate": 4e-05,
      "loss": 0.3608,
      "step": 3790
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.997326203208556e-05,
      "loss": 0.167,
      "step": 3800
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.994652406417113e-05,
      "loss": 0.4472,
      "step": 3810
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.991978609625669e-05,
      "loss": 0.3022,
      "step": 3820
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9893048128342246e-05,
      "loss": 0.3179,
      "step": 3830
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.986631016042781e-05,
      "loss": 0.2939,
      "step": 3840
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.983957219251337e-05,
      "loss": 0.013,
      "step": 3850
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.981283422459893e-05,
      "loss": 0.1864,
      "step": 3860
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.9786096256684496e-05,
      "loss": 0.3368,
      "step": 3870
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.9759358288770055e-05,
      "loss": 0.1275,
      "step": 3880
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.973262032085562e-05,
      "loss": 0.1441,
      "step": 3890
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.970588235294117e-05,
      "loss": 0.358,
      "step": 3900
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.967914438502674e-05,
      "loss": 0.1918,
      "step": 3910
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9652406417112305e-05,
      "loss": 0.2604,
      "step": 3920
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9625668449197864e-05,
      "loss": 0.0787,
      "step": 3930
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.959893048128342e-05,
      "loss": 0.2841,
      "step": 3940
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.957219251336899e-05,
      "loss": 0.4061,
      "step": 3950
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.954545454545455e-05,
      "loss": 0.3767,
      "step": 3960
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.951871657754011e-05,
      "loss": 0.1036,
      "step": 3970
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9491978609625666e-05,
      "loss": 0.38,
      "step": 3980
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.946524064171123e-05,
      "loss": 0.1721,
      "step": 3990
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.94385026737968e-05,
      "loss": 0.2176,
      "step": 4000
    },
    {
      "epoch": 0.64,
      "eval_accuracy": 0.847809667673716,
      "eval_f1": 0.8461258728150335,
      "eval_loss": 2.1692090034484863,
      "eval_precision": 0.8637304048866307,
      "eval_recall": 0.847809667673716,
      "eval_runtime": 10.5307,
      "eval_samples_per_second": 251.455,
      "eval_steps_per_second": 3.988,
      "step": 4000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9411764705882356e-05,
      "loss": 0.1731,
      "step": 4010
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9385026737967915e-05,
      "loss": 0.0994,
      "step": 4020
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9358288770053474e-05,
      "loss": 0.3668,
      "step": 4030
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.933155080213904e-05,
      "loss": 0.6184,
      "step": 4040
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.93048128342246e-05,
      "loss": 0.1739,
      "step": 4050
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.927807486631016e-05,
      "loss": 0.3273,
      "step": 4060
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.9251336898395724e-05,
      "loss": 0.2285,
      "step": 4070
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.922459893048129e-05,
      "loss": 0.2541,
      "step": 4080
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.919786096256685e-05,
      "loss": 0.3116,
      "step": 4090
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.917112299465241e-05,
      "loss": 0.4674,
      "step": 4100
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.914438502673797e-05,
      "loss": 0.2002,
      "step": 4110
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.911764705882353e-05,
      "loss": 0.3206,
      "step": 4120
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.909090909090909e-05,
      "loss": 0.4748,
      "step": 4130
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.906417112299465e-05,
      "loss": 0.1869,
      "step": 4140
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.903743315508022e-05,
      "loss": 0.1935,
      "step": 4150
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.901069518716578e-05,
      "loss": 0.2528,
      "step": 4160
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8983957219251335e-05,
      "loss": 0.275,
      "step": 4170
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.89572192513369e-05,
      "loss": 0.2676,
      "step": 4180
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.893048128342246e-05,
      "loss": 0.3452,
      "step": 4190
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8903743315508025e-05,
      "loss": 0.2249,
      "step": 4200
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8877005347593584e-05,
      "loss": 0.3211,
      "step": 4210
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.8850267379679143e-05,
      "loss": 0.2125,
      "step": 4220
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.882352941176471e-05,
      "loss": 0.2449,
      "step": 4230
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.879679144385027e-05,
      "loss": 0.1108,
      "step": 4240
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.877005347593583e-05,
      "loss": 0.3127,
      "step": 4250
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.874331550802139e-05,
      "loss": 0.1238,
      "step": 4260
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.871657754010695e-05,
      "loss": 0.2596,
      "step": 4270
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.868983957219252e-05,
      "loss": 0.36,
      "step": 4280
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.866310160427808e-05,
      "loss": 0.2548,
      "step": 4290
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 0.1942,
      "step": 4300
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.86096256684492e-05,
      "loss": 0.1626,
      "step": 4310
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.858288770053476e-05,
      "loss": 0.1887,
      "step": 4320
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.855614973262032e-05,
      "loss": 0.3538,
      "step": 4330
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.8529411764705886e-05,
      "loss": 0.2309,
      "step": 4340
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8502673796791445e-05,
      "loss": 0.147,
      "step": 4350
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.847593582887701e-05,
      "loss": 0.1799,
      "step": 4360
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.844919786096257e-05,
      "loss": 0.2291,
      "step": 4370
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.842245989304813e-05,
      "loss": 0.39,
      "step": 4380
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8395721925133694e-05,
      "loss": 0.0407,
      "step": 4390
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8368983957219254e-05,
      "loss": 0.2575,
      "step": 4400
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.834224598930481e-05,
      "loss": 0.1994,
      "step": 4410
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.831550802139038e-05,
      "loss": 0.1931,
      "step": 4420
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.828877005347594e-05,
      "loss": 0.4772,
      "step": 4430
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.82620320855615e-05,
      "loss": 0.406,
      "step": 4440
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8235294117647055e-05,
      "loss": 0.3865,
      "step": 4450
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.820855614973262e-05,
      "loss": 0.2322,
      "step": 4460
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.818181818181819e-05,
      "loss": 0.3505,
      "step": 4470
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.8155080213903746e-05,
      "loss": 0.2413,
      "step": 4480
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.8128342245989305e-05,
      "loss": 0.1657,
      "step": 4490
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.810160427807487e-05,
      "loss": 0.1567,
      "step": 4500
    },
    {
      "epoch": 0.72,
      "eval_accuracy": 0.8462990936555891,
      "eval_f1": 0.8442050087782637,
      "eval_loss": 2.184410333633423,
      "eval_precision": 0.8659758356860412,
      "eval_recall": 0.8462990936555891,
      "eval_runtime": 10.5461,
      "eval_samples_per_second": 251.088,
      "eval_steps_per_second": 3.983,
      "step": 4500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.807486631016043e-05,
      "loss": 0.2413,
      "step": 4510
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.804812834224599e-05,
      "loss": 0.5433,
      "step": 4520
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.802139037433155e-05,
      "loss": 0.326,
      "step": 4530
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7994652406417114e-05,
      "loss": 0.3755,
      "step": 4540
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.796791443850268e-05,
      "loss": 0.2589,
      "step": 4550
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.794117647058824e-05,
      "loss": 0.2539,
      "step": 4560
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.79144385026738e-05,
      "loss": 0.2227,
      "step": 4570
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7887700534759364e-05,
      "loss": 0.1897,
      "step": 4580
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.786096256684492e-05,
      "loss": 0.2462,
      "step": 4590
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.783422459893048e-05,
      "loss": 0.1308,
      "step": 4600
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.780748663101604e-05,
      "loss": 0.2567,
      "step": 4610
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.7780748663101607e-05,
      "loss": 0.3583,
      "step": 4620
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.775401069518717e-05,
      "loss": 0.3253,
      "step": 4630
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.7727272727272725e-05,
      "loss": 0.2606,
      "step": 4640
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.770053475935829e-05,
      "loss": 0.2173,
      "step": 4650
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.767379679144385e-05,
      "loss": 0.2203,
      "step": 4660
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7647058823529415e-05,
      "loss": 0.059,
      "step": 4670
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7620320855614974e-05,
      "loss": 0.3312,
      "step": 4680
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.759358288770053e-05,
      "loss": 0.1597,
      "step": 4690
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.75668449197861e-05,
      "loss": 0.3424,
      "step": 4700
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7540106951871665e-05,
      "loss": 0.2294,
      "step": 4710
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.751336898395722e-05,
      "loss": 0.1233,
      "step": 4720
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.748663101604278e-05,
      "loss": 0.1573,
      "step": 4730
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.745989304812834e-05,
      "loss": 0.0886,
      "step": 4740
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.743315508021391e-05,
      "loss": 0.1607,
      "step": 4750
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.740641711229947e-05,
      "loss": 0.2734,
      "step": 4760
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7379679144385026e-05,
      "loss": 0.1011,
      "step": 4770
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.735294117647059e-05,
      "loss": 0.395,
      "step": 4780
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.732620320855615e-05,
      "loss": 0.225,
      "step": 4790
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.729946524064171e-05,
      "loss": 0.3241,
      "step": 4800
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7272727272727276e-05,
      "loss": 0.3954,
      "step": 4810
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7245989304812835e-05,
      "loss": 0.1394,
      "step": 4820
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.72192513368984e-05,
      "loss": 0.2615,
      "step": 4830
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.719251336898396e-05,
      "loss": 0.1806,
      "step": 4840
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.716577540106952e-05,
      "loss": 0.2261,
      "step": 4850
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.7139037433155084e-05,
      "loss": 0.3843,
      "step": 4860
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.7112299465240643e-05,
      "loss": 0.3096,
      "step": 4870
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.70855614973262e-05,
      "loss": 0.2883,
      "step": 4880
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.705882352941177e-05,
      "loss": 0.3197,
      "step": 4890
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.703208556149733e-05,
      "loss": 0.5345,
      "step": 4900
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.700534759358289e-05,
      "loss": 0.3752,
      "step": 4910
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.697860962566845e-05,
      "loss": 0.2157,
      "step": 4920
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.695187165775401e-05,
      "loss": 0.1772,
      "step": 4930
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.692513368983958e-05,
      "loss": 0.2575,
      "step": 4940
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.6898395721925136e-05,
      "loss": 0.2138,
      "step": 4950
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.6871657754010695e-05,
      "loss": 0.2703,
      "step": 4960
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.684491978609626e-05,
      "loss": 0.4287,
      "step": 4970
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.681818181818182e-05,
      "loss": 0.1396,
      "step": 4980
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.679144385026738e-05,
      "loss": 0.3636,
      "step": 4990
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6764705882352945e-05,
      "loss": 0.3738,
      "step": 5000
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.8466767371601208,
      "eval_f1": 0.8446831969931973,
      "eval_loss": 2.3107666969299316,
      "eval_precision": 0.8654388277401478,
      "eval_recall": 0.8466767371601208,
      "eval_runtime": 10.5417,
      "eval_samples_per_second": 251.194,
      "eval_steps_per_second": 3.984,
      "step": 5000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6737967914438504e-05,
      "loss": 0.3446,
      "step": 5010
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.671122994652407e-05,
      "loss": 0.4857,
      "step": 5020
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.668449197860963e-05,
      "loss": 0.2666,
      "step": 5030
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.665775401069519e-05,
      "loss": 0.4162,
      "step": 5040
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6631016042780753e-05,
      "loss": 0.1304,
      "step": 5050
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.660427807486631e-05,
      "loss": 0.1575,
      "step": 5060
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.657754010695187e-05,
      "loss": 0.1837,
      "step": 5070
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.655080213903743e-05,
      "loss": 0.4065,
      "step": 5080
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6524064171122996e-05,
      "loss": 0.3935,
      "step": 5090
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.649732620320856e-05,
      "loss": 0.1707,
      "step": 5100
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6470588235294114e-05,
      "loss": 0.0913,
      "step": 5110
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.644385026737968e-05,
      "loss": 0.2487,
      "step": 5120
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6417112299465246e-05,
      "loss": 0.092,
      "step": 5130
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6390374331550805e-05,
      "loss": 0.2974,
      "step": 5140
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.2422,
      "step": 5150
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.633689839572192e-05,
      "loss": 0.2698,
      "step": 5160
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.631016042780749e-05,
      "loss": 0.2058,
      "step": 5170
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6283422459893055e-05,
      "loss": 0.3773,
      "step": 5180
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.625668449197861e-05,
      "loss": 0.3958,
      "step": 5190
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.622994652406417e-05,
      "loss": 0.2811,
      "step": 5200
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.620320855614974e-05,
      "loss": 0.2158,
      "step": 5210
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.61764705882353e-05,
      "loss": 0.2284,
      "step": 5220
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.614973262032086e-05,
      "loss": 0.1471,
      "step": 5230
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.6122994652406416e-05,
      "loss": 0.2005,
      "step": 5240
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.609625668449198e-05,
      "loss": 0.1275,
      "step": 5250
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.606951871657754e-05,
      "loss": 0.2067,
      "step": 5260
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.60427807486631e-05,
      "loss": 0.5021,
      "step": 5270
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.6016042780748665e-05,
      "loss": 0.4771,
      "step": 5280
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5989304812834225e-05,
      "loss": 0.4451,
      "step": 5290
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.596256684491979e-05,
      "loss": 0.1325,
      "step": 5300
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.593582887700535e-05,
      "loss": 0.2466,
      "step": 5310
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.590909090909091e-05,
      "loss": 0.3195,
      "step": 5320
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5882352941176474e-05,
      "loss": 0.2085,
      "step": 5330
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.585561497326203e-05,
      "loss": 0.1185,
      "step": 5340
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.582887700534759e-05,
      "loss": 0.5152,
      "step": 5350
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.580213903743316e-05,
      "loss": 0.186,
      "step": 5360
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.577540106951872e-05,
      "loss": 0.2386,
      "step": 5370
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5748663101604276e-05,
      "loss": 0.4468,
      "step": 5380
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.572192513368984e-05,
      "loss": 0.1996,
      "step": 5390
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.56951871657754e-05,
      "loss": 0.1182,
      "step": 5400
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.566844919786097e-05,
      "loss": 0.157,
      "step": 5410
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5641711229946526e-05,
      "loss": 0.1696,
      "step": 5420
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5614973262032085e-05,
      "loss": 0.2641,
      "step": 5430
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.558823529411765e-05,
      "loss": 0.2224,
      "step": 5440
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.556149732620321e-05,
      "loss": 0.2676,
      "step": 5450
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.553475935828877e-05,
      "loss": 0.213,
      "step": 5460
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5508021390374335e-05,
      "loss": 0.3497,
      "step": 5470
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5481283422459894e-05,
      "loss": 0.3077,
      "step": 5480
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.545454545454546e-05,
      "loss": 0.4294,
      "step": 5490
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.542780748663101e-05,
      "loss": 0.1212,
      "step": 5500
    },
    {
      "epoch": 0.88,
      "eval_accuracy": 0.8444108761329305,
      "eval_f1": 0.8425221036853814,
      "eval_loss": 2.348144054412842,
      "eval_precision": 0.8617668418744113,
      "eval_recall": 0.8444108761329305,
      "eval_runtime": 10.5364,
      "eval_samples_per_second": 251.32,
      "eval_steps_per_second": 3.986,
      "step": 5500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.540106951871658e-05,
      "loss": 0.2518,
      "step": 5510
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.537433155080214e-05,
      "loss": 0.2727,
      "step": 5520
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.53475935828877e-05,
      "loss": 0.2366,
      "step": 5530
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.532085561497326e-05,
      "loss": 0.3752,
      "step": 5540
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.529411764705883e-05,
      "loss": 0.3085,
      "step": 5550
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5267379679144386e-05,
      "loss": 0.0496,
      "step": 5560
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.524064171122995e-05,
      "loss": 0.3345,
      "step": 5570
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5213903743315504e-05,
      "loss": 0.393,
      "step": 5580
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.518716577540107e-05,
      "loss": 0.2174,
      "step": 5590
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5160427807486636e-05,
      "loss": 0.2527,
      "step": 5600
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5133689839572195e-05,
      "loss": 0.1707,
      "step": 5610
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5106951871657754e-05,
      "loss": 0.227,
      "step": 5620
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.508021390374332e-05,
      "loss": 0.2253,
      "step": 5630
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.505347593582888e-05,
      "loss": 0.2014,
      "step": 5640
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5026737967914445e-05,
      "loss": 0.2545,
      "step": 5650
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.5e-05,
      "loss": 0.2213,
      "step": 5660
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.497326203208556e-05,
      "loss": 0.172,
      "step": 5670
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.494652406417113e-05,
      "loss": 0.3737,
      "step": 5680
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.491978609625669e-05,
      "loss": 0.1347,
      "step": 5690
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.489304812834225e-05,
      "loss": 0.2867,
      "step": 5700
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.4866310160427806e-05,
      "loss": 0.2276,
      "step": 5710
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.483957219251337e-05,
      "loss": 0.3448,
      "step": 5720
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.481283422459893e-05,
      "loss": 0.0974,
      "step": 5730
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.478609625668449e-05,
      "loss": 0.1956,
      "step": 5740
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4759358288770055e-05,
      "loss": 0.1278,
      "step": 5750
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.473262032085562e-05,
      "loss": 0.2225,
      "step": 5760
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.470588235294118e-05,
      "loss": 0.3499,
      "step": 5770
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.467914438502674e-05,
      "loss": 0.2516,
      "step": 5780
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.46524064171123e-05,
      "loss": 0.158,
      "step": 5790
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.4625668449197864e-05,
      "loss": 0.2245,
      "step": 5800
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.459893048128342e-05,
      "loss": 0.3029,
      "step": 5810
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.457219251336898e-05,
      "loss": 0.3581,
      "step": 5820
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.454545454545455e-05,
      "loss": 0.2423,
      "step": 5830
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.4518716577540114e-05,
      "loss": 0.3618,
      "step": 5840
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4491978609625666e-05,
      "loss": 0.1969,
      "step": 5850
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.446524064171123e-05,
      "loss": 0.282,
      "step": 5860
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.443850267379679e-05,
      "loss": 0.2575,
      "step": 5870
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.441176470588236e-05,
      "loss": 0.3495,
      "step": 5880
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4385026737967916e-05,
      "loss": 2.8462,
      "step": 5890
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4358288770053475e-05,
      "loss": 3.0621,
      "step": 5900
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.433155080213904e-05,
      "loss": 1.8662,
      "step": 5910
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.4304812834224606e-05,
      "loss": 0.7081,
      "step": 5920
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.427807486631016e-05,
      "loss": 0.7017,
      "step": 5930
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.4251336898395724e-05,
      "loss": 0.6809,
      "step": 5940
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.4224598930481284e-05,
      "loss": 0.5177,
      "step": 5950
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.419786096256685e-05,
      "loss": 0.3095,
      "step": 5960
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.417112299465241e-05,
      "loss": 0.1414,
      "step": 5970
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.414438502673797e-05,
      "loss": 0.0596,
      "step": 5980
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.411764705882353e-05,
      "loss": 0.2019,
      "step": 5990
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.409090909090909e-05,
      "loss": 0.308,
      "step": 6000
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.8444108761329305,
      "eval_f1": 0.8423328431236531,
      "eval_loss": 1.9854190349578857,
      "eval_precision": 0.8635785518138459,
      "eval_recall": 0.8444108761329305,
      "eval_runtime": 10.5454,
      "eval_samples_per_second": 251.106,
      "eval_steps_per_second": 3.983,
      "step": 6000
    }
  ],
  "logging_steps": 10,
  "max_steps": 18750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 5.0517776203776e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}