{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.9890764647467725,
  "global_step": 2510,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 9.960238568588469e-05,
      "loss": 0.1445,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.924453280318093e-05,
      "loss": 0.3715,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.884691848906562e-05,
      "loss": 0.5201,
      "step": 30
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.84493041749503e-05,
      "loss": 0.355,
      "step": 40
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.805168986083499e-05,
      "loss": 0.4739,
      "step": 50
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.765407554671967e-05,
      "loss": 0.3795,
      "step": 60
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.725646123260437e-05,
      "loss": 0.2449,
      "step": 70
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.685884691848906e-05,
      "loss": 0.2834,
      "step": 80
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.646123260437376e-05,
      "loss": 0.4803,
      "step": 90
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.606361829025845e-05,
      "loss": 0.4239,
      "step": 100
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.566600397614315e-05,
      "loss": 0.4822,
      "step": 110
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.530815109343937e-05,
      "loss": 0.6423,
      "step": 120
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.491053677932406e-05,
      "loss": 0.4146,
      "step": 130
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.451292246520876e-05,
      "loss": 0.3988,
      "step": 140
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.411530815109344e-05,
      "loss": 0.576,
      "step": 150
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.371769383697814e-05,
      "loss": 0.3698,
      "step": 160
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.332007952286283e-05,
      "loss": 0.4959,
      "step": 170
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.292246520874752e-05,
      "loss": 0.353,
      "step": 180
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.25248508946322e-05,
      "loss": 0.3502,
      "step": 190
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.21272365805169e-05,
      "loss": 0.2586,
      "step": 200
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.172962226640159e-05,
      "loss": 0.3487,
      "step": 210
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.133200795228629e-05,
      "loss": 0.2338,
      "step": 220
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.093439363817098e-05,
      "loss": 0.3604,
      "step": 230
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.053677932405566e-05,
      "loss": 0.1628,
      "step": 240
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.013916500994036e-05,
      "loss": 0.3106,
      "step": 250
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.974155069582505e-05,
      "loss": 0.3068,
      "step": 260
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.934393638170975e-05,
      "loss": 0.273,
      "step": 270
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.894632206759443e-05,
      "loss": 0.1321,
      "step": 280
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.854870775347913e-05,
      "loss": 0.4806,
      "step": 290
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.815109343936382e-05,
      "loss": 0.1793,
      "step": 300
    },
    {
      "epoch": 0.62,
      "learning_rate": 8.775347912524852e-05,
      "loss": 0.2027,
      "step": 310
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.735586481113321e-05,
      "loss": 0.3706,
      "step": 320
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.695825049701791e-05,
      "loss": 0.4457,
      "step": 330
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.65606361829026e-05,
      "loss": 0.3953,
      "step": 340
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.616302186878728e-05,
      "loss": 0.132,
      "step": 350
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.576540755467197e-05,
      "loss": 0.3439,
      "step": 360
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.536779324055667e-05,
      "loss": 0.6228,
      "step": 370
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.497017892644135e-05,
      "loss": 0.3251,
      "step": 380
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.457256461232604e-05,
      "loss": 0.3342,
      "step": 390
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.417495029821074e-05,
      "loss": 0.316,
      "step": 400
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.377733598409543e-05,
      "loss": 0.2536,
      "step": 410
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.337972166998013e-05,
      "loss": 0.2428,
      "step": 420
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.298210735586481e-05,
      "loss": 0.1508,
      "step": 430
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.258449304174951e-05,
      "loss": 0.2234,
      "step": 440
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.21868787276342e-05,
      "loss": 0.3659,
      "step": 450
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.17892644135189e-05,
      "loss": 0.2216,
      "step": 460
    },
    {
      "epoch": 0.93,
      "learning_rate": 8.139165009940358e-05,
      "loss": 0.3196,
      "step": 470
    },
    {
      "epoch": 0.95,
      "learning_rate": 8.099403578528827e-05,
      "loss": 0.3921,
      "step": 480
    },
    {
      "epoch": 0.97,
      "learning_rate": 8.059642147117297e-05,
      "loss": 0.3935,
      "step": 490
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.019880715705766e-05,
      "loss": 0.1749,
      "step": 500
    },
    {
      "epoch": 0.99,
      "eval_accuracy": 0.9107142686843872,
      "eval_loss": 0.2552814781665802,
      "eval_runtime": 19.16,
      "eval_samples_per_second": 26.305,
      "eval_steps_per_second": 6.576,
      "step": 500
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.980119284294236e-05,
      "loss": 0.3414,
      "step": 510
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.940357852882704e-05,
      "loss": 0.2059,
      "step": 520
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.900596421471174e-05,
      "loss": 0.1762,
      "step": 530
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.860834990059643e-05,
      "loss": 0.118,
      "step": 540
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.821073558648112e-05,
      "loss": 0.1524,
      "step": 550
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.78131212723658e-05,
      "loss": 0.1576,
      "step": 560
    },
    {
      "epoch": 1.13,
      "learning_rate": 7.74155069582505e-05,
      "loss": 0.2975,
      "step": 570
    },
    {
      "epoch": 1.15,
      "learning_rate": 7.701789264413519e-05,
      "loss": 0.2985,
      "step": 580
    },
    {
      "epoch": 1.17,
      "learning_rate": 7.662027833001987e-05,
      "loss": 0.1437,
      "step": 590
    },
    {
      "epoch": 1.19,
      "learning_rate": 7.622266401590457e-05,
      "loss": 0.2702,
      "step": 600
    },
    {
      "epoch": 1.21,
      "learning_rate": 7.582504970178926e-05,
      "loss": 0.2913,
      "step": 610
    },
    {
      "epoch": 1.23,
      "learning_rate": 7.542743538767396e-05,
      "loss": 0.1741,
      "step": 620
    },
    {
      "epoch": 1.25,
      "learning_rate": 7.502982107355865e-05,
      "loss": 0.1987,
      "step": 630
    },
    {
      "epoch": 1.27,
      "learning_rate": 7.463220675944335e-05,
      "loss": 0.2568,
      "step": 640
    },
    {
      "epoch": 1.29,
      "learning_rate": 7.423459244532803e-05,
      "loss": 0.4236,
      "step": 650
    },
    {
      "epoch": 1.31,
      "learning_rate": 7.383697813121273e-05,
      "loss": 0.1788,
      "step": 660
    },
    {
      "epoch": 1.33,
      "learning_rate": 7.343936381709742e-05,
      "loss": 0.2973,
      "step": 670
    },
    {
      "epoch": 1.35,
      "learning_rate": 7.304174950298212e-05,
      "loss": 0.3244,
      "step": 680
    },
    {
      "epoch": 1.37,
      "learning_rate": 7.26441351888668e-05,
      "loss": 0.2661,
      "step": 690
    },
    {
      "epoch": 1.39,
      "learning_rate": 7.22465208747515e-05,
      "loss": 0.3832,
      "step": 700
    },
    {
      "epoch": 1.41,
      "learning_rate": 7.184890656063619e-05,
      "loss": 0.1453,
      "step": 710
    },
    {
      "epoch": 1.43,
      "learning_rate": 7.145129224652088e-05,
      "loss": 0.1034,
      "step": 720
    },
    {
      "epoch": 1.45,
      "learning_rate": 7.105367793240556e-05,
      "loss": 0.2906,
      "step": 730
    },
    {
      "epoch": 1.47,
      "learning_rate": 7.065606361829026e-05,
      "loss": 0.2115,
      "step": 740
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.025844930417495e-05,
      "loss": 0.3433,
      "step": 750
    },
    {
      "epoch": 1.51,
      "learning_rate": 6.986083499005964e-05,
      "loss": 0.1599,
      "step": 760
    },
    {
      "epoch": 1.53,
      "learning_rate": 6.946322067594434e-05,
      "loss": 0.1725,
      "step": 770
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.910536779324056e-05,
      "loss": 0.428,
      "step": 780
    },
    {
      "epoch": 1.57,
      "learning_rate": 6.870775347912525e-05,
      "loss": 0.1033,
      "step": 790
    },
    {
      "epoch": 1.59,
      "learning_rate": 6.831013916500995e-05,
      "loss": 0.314,
      "step": 800
    },
    {
      "epoch": 1.61,
      "learning_rate": 6.791252485089463e-05,
      "loss": 0.146,
      "step": 810
    },
    {
      "epoch": 1.63,
      "learning_rate": 6.751491053677933e-05,
      "loss": 0.181,
      "step": 820
    },
    {
      "epoch": 1.65,
      "learning_rate": 6.711729622266402e-05,
      "loss": 0.2174,
      "step": 830
    },
    {
      "epoch": 1.67,
      "learning_rate": 6.671968190854872e-05,
      "loss": 0.1458,
      "step": 840
    },
    {
      "epoch": 1.69,
      "learning_rate": 6.632206759443341e-05,
      "loss": 0.1477,
      "step": 850
    },
    {
      "epoch": 1.71,
      "learning_rate": 6.59244532803181e-05,
      "loss": 0.3046,
      "step": 860
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.552683896620278e-05,
      "loss": 0.2796,
      "step": 870
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.512922465208748e-05,
      "loss": 0.1284,
      "step": 880
    },
    {
      "epoch": 1.77,
      "learning_rate": 6.473161033797217e-05,
      "loss": 0.3014,
      "step": 890
    },
    {
      "epoch": 1.79,
      "learning_rate": 6.433399602385685e-05,
      "loss": 0.2031,
      "step": 900
    },
    {
      "epoch": 1.81,
      "learning_rate": 6.393638170974155e-05,
      "loss": 0.134,
      "step": 910
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.353876739562624e-05,
      "loss": 0.2085,
      "step": 920
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.314115308151094e-05,
      "loss": 0.2496,
      "step": 930
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.274353876739562e-05,
      "loss": 0.1984,
      "step": 940
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.234592445328032e-05,
      "loss": 0.2887,
      "step": 950
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.194831013916501e-05,
      "loss": 0.1923,
      "step": 960
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.155069582504971e-05,
      "loss": 0.1373,
      "step": 970
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.11530815109344e-05,
      "loss": 0.2089,
      "step": 980
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.075546719681909e-05,
      "loss": 0.2066,
      "step": 990
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.035785288270378e-05,
      "loss": 0.1106,
      "step": 1000
    },
    {
      "epoch": 1.99,
      "eval_accuracy": 0.932539701461792,
      "eval_loss": 0.25381073355674744,
      "eval_runtime": 20.2105,
      "eval_samples_per_second": 24.938,
      "eval_steps_per_second": 6.234,
      "step": 1000
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.996023856858847e-05,
      "loss": 0.125,
      "step": 1010
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.956262425447316e-05,
      "loss": 0.1804,
      "step": 1020
    },
    {
      "epoch": 2.05,
      "learning_rate": 5.916500994035785e-05,
      "loss": 0.0419,
      "step": 1030
    },
    {
      "epoch": 2.07,
      "learning_rate": 5.876739562624255e-05,
      "loss": 0.1823,
      "step": 1040
    },
    {
      "epoch": 2.09,
      "learning_rate": 5.8369781312127236e-05,
      "loss": 0.085,
      "step": 1050
    },
    {
      "epoch": 2.11,
      "learning_rate": 5.7972166998011936e-05,
      "loss": 0.5209,
      "step": 1060
    },
    {
      "epoch": 2.13,
      "learning_rate": 5.757455268389662e-05,
      "loss": 0.1744,
      "step": 1070
    },
    {
      "epoch": 2.15,
      "learning_rate": 5.717693836978132e-05,
      "loss": 0.1288,
      "step": 1080
    },
    {
      "epoch": 2.17,
      "learning_rate": 5.677932405566601e-05,
      "loss": 0.0729,
      "step": 1090
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.63817097415507e-05,
      "loss": 0.1138,
      "step": 1100
    },
    {
      "epoch": 2.21,
      "learning_rate": 5.598409542743539e-05,
      "loss": 0.2507,
      "step": 1110
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.5586481113320074e-05,
      "loss": 0.2038,
      "step": 1120
    },
    {
      "epoch": 2.25,
      "learning_rate": 5.5188866799204774e-05,
      "loss": 0.3174,
      "step": 1130
    },
    {
      "epoch": 2.27,
      "learning_rate": 5.479125248508946e-05,
      "loss": 0.2045,
      "step": 1140
    },
    {
      "epoch": 2.29,
      "learning_rate": 5.439363817097416e-05,
      "loss": 0.1042,
      "step": 1150
    },
    {
      "epoch": 2.31,
      "learning_rate": 5.399602385685885e-05,
      "loss": 0.0531,
      "step": 1160
    },
    {
      "epoch": 2.33,
      "learning_rate": 5.359840954274355e-05,
      "loss": 0.1353,
      "step": 1170
    },
    {
      "epoch": 2.35,
      "learning_rate": 5.320079522862823e-05,
      "loss": 0.3447,
      "step": 1180
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.2803180914512926e-05,
      "loss": 0.1704,
      "step": 1190
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.240556660039761e-05,
      "loss": 0.2148,
      "step": 1200
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.200795228628231e-05,
      "loss": 0.222,
      "step": 1210
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.1610337972167e-05,
      "loss": 0.0645,
      "step": 1220
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.12127236580517e-05,
      "loss": 0.1284,
      "step": 1230
    },
    {
      "epoch": 2.46,
      "learning_rate": 5.0815109343936385e-05,
      "loss": 0.1021,
      "step": 1240
    },
    {
      "epoch": 2.48,
      "learning_rate": 5.041749502982107e-05,
      "loss": 0.1881,
      "step": 1250
    },
    {
      "epoch": 2.5,
      "learning_rate": 5.001988071570577e-05,
      "loss": 0.1158,
      "step": 1260
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.9622266401590465e-05,
      "loss": 0.2495,
      "step": 1270
    },
    {
      "epoch": 2.54,
      "learning_rate": 4.922465208747515e-05,
      "loss": 0.0767,
      "step": 1280
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.882703777335984e-05,
      "loss": 0.1401,
      "step": 1290
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.842942345924453e-05,
      "loss": 0.1357,
      "step": 1300
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.8031809145129224e-05,
      "loss": 0.1962,
      "step": 1310
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.763419483101392e-05,
      "loss": 0.1744,
      "step": 1320
    },
    {
      "epoch": 2.64,
      "learning_rate": 4.723658051689861e-05,
      "loss": 0.1815,
      "step": 1330
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.68389662027833e-05,
      "loss": 0.275,
      "step": 1340
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.6441351888667996e-05,
      "loss": 0.1725,
      "step": 1350
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.604373757455269e-05,
      "loss": 0.1071,
      "step": 1360
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.5646123260437376e-05,
      "loss": 0.0908,
      "step": 1370
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.524850894632207e-05,
      "loss": 0.0793,
      "step": 1380
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.485089463220676e-05,
      "loss": 0.114,
      "step": 1390
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.4453280318091455e-05,
      "loss": 0.2036,
      "step": 1400
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.405566600397614e-05,
      "loss": 0.0941,
      "step": 1410
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.3658051689860835e-05,
      "loss": 0.1597,
      "step": 1420
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.326043737574553e-05,
      "loss": 0.1476,
      "step": 1430
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.286282306163022e-05,
      "loss": 0.1634,
      "step": 1440
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.2465208747514914e-05,
      "loss": 0.3117,
      "step": 1450
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.206759443339961e-05,
      "loss": 0.1822,
      "step": 1460
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.1669980119284293e-05,
      "loss": 0.2559,
      "step": 1470
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.1272365805168987e-05,
      "loss": 0.0924,
      "step": 1480
    },
    {
      "epoch": 2.96,
      "learning_rate": 4.087475149105368e-05,
      "loss": 0.095,
      "step": 1490
    },
    {
      "epoch": 2.98,
      "learning_rate": 4.047713717693837e-05,
      "loss": 0.0947,
      "step": 1500
    },
    {
      "epoch": 2.98,
      "eval_accuracy": 0.9305555820465088,
      "eval_loss": 0.23964740335941315,
      "eval_runtime": 18.9345,
      "eval_samples_per_second": 26.618,
      "eval_steps_per_second": 6.655,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.0079522862823066e-05,
      "loss": 0.1292,
      "step": 1510
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.968190854870776e-05,
      "loss": 0.0505,
      "step": 1520
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.9284294234592445e-05,
      "loss": 0.0629,
      "step": 1530
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.888667992047714e-05,
      "loss": 0.2428,
      "step": 1540
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.848906560636183e-05,
      "loss": 0.0792,
      "step": 1550
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.8091451292246525e-05,
      "loss": 0.1127,
      "step": 1560
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.769383697813121e-05,
      "loss": 0.0928,
      "step": 1570
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.7296222664015904e-05,
      "loss": 0.1148,
      "step": 1580
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.68986083499006e-05,
      "loss": 0.1969,
      "step": 1590
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.650099403578529e-05,
      "loss": 0.0891,
      "step": 1600
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.6103379721669984e-05,
      "loss": 0.1074,
      "step": 1610
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.570576540755468e-05,
      "loss": 0.1484,
      "step": 1620
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.530815109343937e-05,
      "loss": 0.2352,
      "step": 1630
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.491053677932406e-05,
      "loss": 0.0469,
      "step": 1640
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.451292246520875e-05,
      "loss": 0.1484,
      "step": 1650
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.4115308151093436e-05,
      "loss": 0.1446,
      "step": 1660
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.371769383697813e-05,
      "loss": 0.0165,
      "step": 1670
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.332007952286282e-05,
      "loss": 0.0797,
      "step": 1680
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.2922465208747515e-05,
      "loss": 0.1391,
      "step": 1690
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.252485089463221e-05,
      "loss": 0.0918,
      "step": 1700
    },
    {
      "epoch": 3.4,
      "learning_rate": 3.21272365805169e-05,
      "loss": 0.1694,
      "step": 1710
    },
    {
      "epoch": 3.42,
      "learning_rate": 3.1729622266401595e-05,
      "loss": 0.1379,
      "step": 1720
    },
    {
      "epoch": 3.44,
      "learning_rate": 3.133200795228629e-05,
      "loss": 0.1016,
      "step": 1730
    },
    {
      "epoch": 3.46,
      "learning_rate": 3.0934393638170974e-05,
      "loss": 0.0449,
      "step": 1740
    },
    {
      "epoch": 3.48,
      "learning_rate": 3.053677932405567e-05,
      "loss": 0.0124,
      "step": 1750
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.0139165009940357e-05,
      "loss": 0.0186,
      "step": 1760
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.974155069582505e-05,
      "loss": 0.1103,
      "step": 1770
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.934393638170974e-05,
      "loss": 0.063,
      "step": 1780
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.8946322067594433e-05,
      "loss": 0.061,
      "step": 1790
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.8548707753479126e-05,
      "loss": 0.1044,
      "step": 1800
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.815109343936382e-05,
      "loss": 0.1516,
      "step": 1810
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.775347912524851e-05,
      "loss": 0.1427,
      "step": 1820
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.7355864811133202e-05,
      "loss": 0.0448,
      "step": 1830
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.6958250497017895e-05,
      "loss": 0.0703,
      "step": 1840
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.656063618290259e-05,
      "loss": 0.0905,
      "step": 1850
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.616302186878728e-05,
      "loss": 0.081,
      "step": 1860
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.576540755467197e-05,
      "loss": 0.1642,
      "step": 1870
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.5367793240556658e-05,
      "loss": 0.1915,
      "step": 1880
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.4970178926441354e-05,
      "loss": 0.0696,
      "step": 1890
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.4572564612326048e-05,
      "loss": 0.0605,
      "step": 1900
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.4174950298210737e-05,
      "loss": 0.1562,
      "step": 1910
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.3777335984095427e-05,
      "loss": 0.1316,
      "step": 1920
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.337972166998012e-05,
      "loss": 0.1364,
      "step": 1930
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2982107355864813e-05,
      "loss": 0.0766,
      "step": 1940
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2584493041749506e-05,
      "loss": 0.2037,
      "step": 1950
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.2186878727634196e-05,
      "loss": 0.029,
      "step": 1960
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.1789264413518886e-05,
      "loss": 0.0443,
      "step": 1970
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.139165009940358e-05,
      "loss": 0.0494,
      "step": 1980
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.0994035785288272e-05,
      "loss": 0.0766,
      "step": 1990
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.0596421471172962e-05,
      "loss": 0.1602,
      "step": 2000
    },
    {
      "epoch": 3.98,
      "eval_accuracy": 0.9444444179534912,
      "eval_loss": 0.26106151938438416,
      "eval_runtime": 19.5055,
      "eval_samples_per_second": 25.839,
      "eval_steps_per_second": 6.46,
      "step": 2000
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.0198807157057655e-05,
      "loss": 0.0202,
      "step": 2010
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.9801192842942348e-05,
      "loss": 0.0856,
      "step": 2020
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.9403578528827038e-05,
      "loss": 0.0999,
      "step": 2030
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.900596421471173e-05,
      "loss": 0.0053,
      "step": 2040
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.860834990059642e-05,
      "loss": 0.1129,
      "step": 2050
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.8210735586481114e-05,
      "loss": 0.0083,
      "step": 2060
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.7813121272365807e-05,
      "loss": 0.0857,
      "step": 2070
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.7415506958250497e-05,
      "loss": 0.0475,
      "step": 2080
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.701789264413519e-05,
      "loss": 0.0267,
      "step": 2090
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.662027833001988e-05,
      "loss": 0.0469,
      "step": 2100
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.6222664015904573e-05,
      "loss": 0.0346,
      "step": 2110
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.5825049701789266e-05,
      "loss": 0.0822,
      "step": 2120
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.542743538767396e-05,
      "loss": 0.0803,
      "step": 2130
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.5029821073558647e-05,
      "loss": 0.0077,
      "step": 2140
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.463220675944334e-05,
      "loss": 0.0347,
      "step": 2150
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.4234592445328032e-05,
      "loss": 0.0665,
      "step": 2160
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.3836978131212725e-05,
      "loss": 0.0779,
      "step": 2170
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.3439363817097416e-05,
      "loss": 0.056,
      "step": 2180
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.304174950298211e-05,
      "loss": 0.1503,
      "step": 2190
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.26441351888668e-05,
      "loss": 0.0032,
      "step": 2200
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.2246520874751492e-05,
      "loss": 0.0457,
      "step": 2210
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.1848906560636184e-05,
      "loss": 0.0077,
      "step": 2220
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.1451292246520875e-05,
      "loss": 0.1558,
      "step": 2230
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.1053677932405568e-05,
      "loss": 0.0307,
      "step": 2240
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.0656063618290258e-05,
      "loss": 0.1193,
      "step": 2250
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.0258449304174951e-05,
      "loss": 0.0638,
      "step": 2260
    },
    {
      "epoch": 4.51,
      "learning_rate": 9.860834990059643e-06,
      "loss": 0.059,
      "step": 2270
    },
    {
      "epoch": 4.53,
      "learning_rate": 9.463220675944334e-06,
      "loss": 0.0875,
      "step": 2280
    },
    {
      "epoch": 4.55,
      "learning_rate": 9.065606361829027e-06,
      "loss": 0.1511,
      "step": 2290
    },
    {
      "epoch": 4.57,
      "learning_rate": 8.667992047713717e-06,
      "loss": 0.1217,
      "step": 2300
    },
    {
      "epoch": 4.59,
      "learning_rate": 8.27037773359841e-06,
      "loss": 0.2091,
      "step": 2310
    },
    {
      "epoch": 4.61,
      "learning_rate": 7.872763419483102e-06,
      "loss": 0.1723,
      "step": 2320
    },
    {
      "epoch": 4.63,
      "learning_rate": 7.475149105367793e-06,
      "loss": 0.0763,
      "step": 2330
    },
    {
      "epoch": 4.65,
      "learning_rate": 7.077534791252485e-06,
      "loss": 0.0172,
      "step": 2340
    },
    {
      "epoch": 4.67,
      "learning_rate": 6.679920477137178e-06,
      "loss": 0.0352,
      "step": 2350
    },
    {
      "epoch": 4.69,
      "learning_rate": 6.282306163021868e-06,
      "loss": 0.1989,
      "step": 2360
    },
    {
      "epoch": 4.71,
      "learning_rate": 5.8846918489065606e-06,
      "loss": 0.0896,
      "step": 2370
    },
    {
      "epoch": 4.73,
      "learning_rate": 5.487077534791253e-06,
      "loss": 0.0728,
      "step": 2380
    },
    {
      "epoch": 4.75,
      "learning_rate": 5.089463220675945e-06,
      "loss": 0.0482,
      "step": 2390
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.6918489065606366e-06,
      "loss": 0.1288,
      "step": 2400
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.294234592445328e-06,
      "loss": 0.0756,
      "step": 2410
    },
    {
      "epoch": 4.81,
      "learning_rate": 3.89662027833002e-06,
      "loss": 0.051,
      "step": 2420
    },
    {
      "epoch": 4.83,
      "learning_rate": 3.4990059642147117e-06,
      "loss": 0.0874,
      "step": 2430
    },
    {
      "epoch": 4.85,
      "learning_rate": 3.101391650099404e-06,
      "loss": 0.0381,
      "step": 2440
    },
    {
      "epoch": 4.87,
      "learning_rate": 2.7037773359840955e-06,
      "loss": 0.0031,
      "step": 2450
    },
    {
      "epoch": 4.89,
      "learning_rate": 2.3061630218687873e-06,
      "loss": 0.0371,
      "step": 2460
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.9085487077534792e-06,
      "loss": 0.007,
      "step": 2470
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.510934393638171e-06,
      "loss": 0.115,
      "step": 2480
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.113320079522863e-06,
      "loss": 0.0492,
      "step": 2490
    },
    {
      "epoch": 4.97,
      "learning_rate": 7.157057654075547e-07,
      "loss": 0.0104,
      "step": 2500
    },
    {
      "epoch": 4.97,
      "eval_accuracy": 0.9484127163887024,
      "eval_loss": 0.2534567415714264,
      "eval_runtime": 18.9048,
      "eval_samples_per_second": 26.66,
      "eval_steps_per_second": 6.665,
      "step": 2500
    },
    {
      "epoch": 4.99,
      "learning_rate": 3.1809145129224657e-07,
      "loss": 0.0589,
      "step": 2510
    }
  ],
  "max_steps": 2515,
  "num_train_epochs": 5,
  "total_flos": 7.174884566087474e+17,
  "trial_name": null,
  "trial_params": null
}