{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.3474183255149363,
  "eval_steps": 500,
  "global_step": 390000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 173.0,
      "learning_rate": 4.4539461963299484e-07,
      "loss": 2.153,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 40.5,
      "learning_rate": 8.907892392659897e-07,
      "loss": 2.398,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 223.0,
      "learning_rate": 1.3361838588989846e-06,
      "loss": 1.8912,
      "step": 300
    },
    {
      "epoch": 0.0,
      "grad_norm": 95.5,
      "learning_rate": 1.7815784785319793e-06,
      "loss": 1.6403,
      "step": 400
    },
    {
      "epoch": 0.0,
      "grad_norm": 600.0,
      "learning_rate": 2.226973098164974e-06,
      "loss": 1.4652,
      "step": 500
    },
    {
      "epoch": 0.0,
      "grad_norm": 51.75,
      "learning_rate": 2.672367717797969e-06,
      "loss": 1.2489,
      "step": 600
    },
    {
      "epoch": 0.0,
      "grad_norm": 290.0,
      "learning_rate": 3.1177623374309637e-06,
      "loss": 1.2421,
      "step": 700
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.75,
      "learning_rate": 3.5631569570639587e-06,
      "loss": 1.2188,
      "step": 800
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.0,
      "learning_rate": 4.008551576696954e-06,
      "loss": 1.123,
      "step": 900
    },
    {
      "epoch": 0.0,
      "grad_norm": 39.75,
      "learning_rate": 4.453946196329948e-06,
      "loss": 1.0388,
      "step": 1000
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.375,
      "learning_rate": 4.899340815962943e-06,
      "loss": 1.3034,
      "step": 1100
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.75,
      "learning_rate": 5.344735435595938e-06,
      "loss": 1.2272,
      "step": 1200
    },
    {
      "epoch": 0.0,
      "grad_norm": 71.5,
      "learning_rate": 5.790130055228933e-06,
      "loss": 1.1825,
      "step": 1300
    },
    {
      "epoch": 0.0,
      "grad_norm": 54.5,
      "learning_rate": 6.235524674861927e-06,
      "loss": 1.3466,
      "step": 1400
    },
    {
      "epoch": 0.0,
      "grad_norm": 169.0,
      "learning_rate": 6.680919294494922e-06,
      "loss": 1.0543,
      "step": 1500
    },
    {
      "epoch": 0.0,
      "grad_norm": 47.75,
      "learning_rate": 7.126313914127917e-06,
      "loss": 1.1117,
      "step": 1600
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.5,
      "learning_rate": 7.571708533760913e-06,
      "loss": 1.1458,
      "step": 1700
    },
    {
      "epoch": 0.0,
      "grad_norm": 119.0,
      "learning_rate": 8.017103153393907e-06,
      "loss": 1.1823,
      "step": 1800
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.0,
      "learning_rate": 8.462497773026902e-06,
      "loss": 1.2171,
      "step": 1900
    },
    {
      "epoch": 0.0,
      "grad_norm": 57.25,
      "learning_rate": 8.907892392659896e-06,
      "loss": 1.16,
      "step": 2000
    },
    {
      "epoch": 0.0,
      "grad_norm": 77.0,
      "learning_rate": 9.353287012292893e-06,
      "loss": 1.1243,
      "step": 2100
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.5,
      "learning_rate": 9.798681631925886e-06,
      "loss": 1.1218,
      "step": 2200
    },
    {
      "epoch": 0.0,
      "grad_norm": 98.0,
      "learning_rate": 1.0244076251558882e-05,
      "loss": 1.1527,
      "step": 2300
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.25,
      "learning_rate": 1.0689470871191876e-05,
      "loss": 1.1042,
      "step": 2400
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.0,
      "learning_rate": 1.1134865490824871e-05,
      "loss": 1.0131,
      "step": 2500
    },
    {
      "epoch": 0.0,
      "grad_norm": 50.0,
      "learning_rate": 1.1580260110457866e-05,
      "loss": 1.2544,
      "step": 2600
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.625,
      "learning_rate": 1.2025654730090862e-05,
      "loss": 1.1475,
      "step": 2700
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.375,
      "learning_rate": 1.2471049349723855e-05,
      "loss": 1.2345,
      "step": 2800
    },
    {
      "epoch": 0.0,
      "grad_norm": 39.5,
      "learning_rate": 1.2916443969356851e-05,
      "loss": 1.269,
      "step": 2900
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.5,
      "learning_rate": 1.3361838588989844e-05,
      "loss": 1.1995,
      "step": 3000
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.875,
      "learning_rate": 1.3807233208622842e-05,
      "loss": 1.1367,
      "step": 3100
    },
    {
      "epoch": 0.0,
      "grad_norm": 255.0,
      "learning_rate": 1.4252627828255835e-05,
      "loss": 1.1993,
      "step": 3200
    },
    {
      "epoch": 0.0,
      "grad_norm": 73.5,
      "learning_rate": 1.469802244788883e-05,
      "loss": 1.039,
      "step": 3300
    },
    {
      "epoch": 0.0,
      "grad_norm": 106.5,
      "learning_rate": 1.5143417067521826e-05,
      "loss": 1.2538,
      "step": 3400
    },
    {
      "epoch": 0.0,
      "grad_norm": 153.0,
      "learning_rate": 1.558881168715482e-05,
      "loss": 1.156,
      "step": 3500
    },
    {
      "epoch": 0.0,
      "grad_norm": 111.0,
      "learning_rate": 1.6034206306787815e-05,
      "loss": 1.2418,
      "step": 3600
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.25,
      "learning_rate": 1.647960092642081e-05,
      "loss": 1.0018,
      "step": 3700
    },
    {
      "epoch": 0.0,
      "grad_norm": 68.0,
      "learning_rate": 1.6924995546053804e-05,
      "loss": 1.101,
      "step": 3800
    },
    {
      "epoch": 0.0,
      "grad_norm": 79.5,
      "learning_rate": 1.7370390165686802e-05,
      "loss": 1.1538,
      "step": 3900
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.5,
      "learning_rate": 1.7815784785319793e-05,
      "loss": 1.176,
      "step": 4000
    },
    {
      "epoch": 0.0,
      "grad_norm": 96.5,
      "learning_rate": 1.8261179404952788e-05,
      "loss": 0.9909,
      "step": 4100
    },
    {
      "epoch": 0.0,
      "grad_norm": 45.75,
      "learning_rate": 1.8706574024585786e-05,
      "loss": 1.1521,
      "step": 4200
    },
    {
      "epoch": 0.0,
      "grad_norm": 42.0,
      "learning_rate": 1.915196864421878e-05,
      "loss": 1.1438,
      "step": 4300
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.125,
      "learning_rate": 1.959736326385177e-05,
      "loss": 1.1887,
      "step": 4400
    },
    {
      "epoch": 0.0,
      "grad_norm": 38.75,
      "learning_rate": 2.004275788348477e-05,
      "loss": 1.2946,
      "step": 4500
    },
    {
      "epoch": 0.0,
      "grad_norm": 99.5,
      "learning_rate": 2.0488152503117764e-05,
      "loss": 1.2574,
      "step": 4600
    },
    {
      "epoch": 0.0,
      "grad_norm": 65.5,
      "learning_rate": 2.093354712275076e-05,
      "loss": 1.1517,
      "step": 4700
    },
    {
      "epoch": 0.0,
      "grad_norm": 79.5,
      "learning_rate": 2.1378941742383753e-05,
      "loss": 1.187,
      "step": 4800
    },
    {
      "epoch": 0.0,
      "grad_norm": 162.0,
      "learning_rate": 2.1824336362016748e-05,
      "loss": 0.9974,
      "step": 4900
    },
    {
      "epoch": 0.0,
      "grad_norm": 48.5,
      "learning_rate": 2.2269730981649742e-05,
      "loss": 1.2278,
      "step": 5000
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.09423828125,
      "learning_rate": 2.271512560128274e-05,
      "loss": 1.2128,
      "step": 5100
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.75,
      "learning_rate": 2.316052022091573e-05,
      "loss": 1.1712,
      "step": 5200
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.01483154296875,
      "learning_rate": 2.3605914840548726e-05,
      "loss": 0.9784,
      "step": 5300
    },
    {
      "epoch": 0.0,
      "grad_norm": 89.0,
      "learning_rate": 2.4051309460181724e-05,
      "loss": 1.1382,
      "step": 5400
    },
    {
      "epoch": 0.0,
      "grad_norm": 59.5,
      "learning_rate": 2.449670407981472e-05,
      "loss": 1.291,
      "step": 5500
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.5,
      "learning_rate": 2.494209869944771e-05,
      "loss": 1.1317,
      "step": 5600
    },
    {
      "epoch": 0.01,
      "grad_norm": 24.125,
      "learning_rate": 2.5387493319080707e-05,
      "loss": 1.1689,
      "step": 5700
    },
    {
      "epoch": 0.01,
      "grad_norm": 34.0,
      "learning_rate": 2.5832887938713702e-05,
      "loss": 1.2427,
      "step": 5800
    },
    {
      "epoch": 0.01,
      "grad_norm": 34.5,
      "learning_rate": 2.6278282558346697e-05,
      "loss": 1.353,
      "step": 5900
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.25,
      "learning_rate": 2.6723677177979688e-05,
      "loss": 1.1593,
      "step": 6000
    },
    {
      "epoch": 0.01,
      "grad_norm": 33.75,
      "learning_rate": 2.716907179761269e-05,
      "loss": 1.1391,
      "step": 6100
    },
    {
      "epoch": 0.01,
      "grad_norm": 82.0,
      "learning_rate": 2.7614466417245684e-05,
      "loss": 1.1999,
      "step": 6200
    },
    {
      "epoch": 0.01,
      "grad_norm": 25.375,
      "learning_rate": 2.8059861036878675e-05,
      "loss": 1.0865,
      "step": 6300
    },
    {
      "epoch": 0.01,
      "grad_norm": 34.25,
      "learning_rate": 2.850525565651167e-05,
      "loss": 1.0761,
      "step": 6400
    },
    {
      "epoch": 0.01,
      "grad_norm": 44.0,
      "learning_rate": 2.8950650276144664e-05,
      "loss": 1.1187,
      "step": 6500
    },
    {
      "epoch": 0.01,
      "grad_norm": 38.25,
      "learning_rate": 2.939604489577766e-05,
      "loss": 1.0824,
      "step": 6600
    },
    {
      "epoch": 0.01,
      "grad_norm": 49.75,
      "learning_rate": 2.9841439515410657e-05,
      "loss": 1.0826,
      "step": 6700
    },
    {
      "epoch": 0.01,
      "grad_norm": 47.0,
      "learning_rate": 3.028683413504365e-05,
      "loss": 1.1387,
      "step": 6800
    },
    {
      "epoch": 0.01,
      "grad_norm": 61.0,
      "learning_rate": 3.073222875467664e-05,
      "loss": 1.1661,
      "step": 6900
    },
    {
      "epoch": 0.01,
      "grad_norm": 31.625,
      "learning_rate": 3.117762337430964e-05,
      "loss": 1.2022,
      "step": 7000
    },
    {
      "epoch": 0.01,
      "grad_norm": 38.75,
      "learning_rate": 3.162301799394263e-05,
      "loss": 1.2461,
      "step": 7100
    },
    {
      "epoch": 0.01,
      "grad_norm": 28.5,
      "learning_rate": 3.206841261357563e-05,
      "loss": 1.2583,
      "step": 7200
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.0,
      "learning_rate": 3.251380723320863e-05,
      "loss": 1.1094,
      "step": 7300
    },
    {
      "epoch": 0.01,
      "grad_norm": 68.0,
      "learning_rate": 3.295920185284162e-05,
      "loss": 1.2244,
      "step": 7400
    },
    {
      "epoch": 0.01,
      "grad_norm": 42.0,
      "learning_rate": 3.3404596472474617e-05,
      "loss": 1.1486,
      "step": 7500
    },
    {
      "epoch": 0.01,
      "grad_norm": 344.0,
      "learning_rate": 3.384999109210761e-05,
      "loss": 1.2049,
      "step": 7600
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.6875,
      "learning_rate": 3.42953857117406e-05,
      "loss": 1.0226,
      "step": 7700
    },
    {
      "epoch": 0.01,
      "grad_norm": 43.75,
      "learning_rate": 3.4740780331373604e-05,
      "loss": 1.2662,
      "step": 7800
    },
    {
      "epoch": 0.01,
      "grad_norm": 51.5,
      "learning_rate": 3.5186174951006595e-05,
      "loss": 1.1029,
      "step": 7900
    },
    {
      "epoch": 0.01,
      "grad_norm": 36.25,
      "learning_rate": 3.5631569570639586e-05,
      "loss": 1.1623,
      "step": 8000
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.75,
      "learning_rate": 3.6076964190272584e-05,
      "loss": 1.0682,
      "step": 8100
    },
    {
      "epoch": 0.01,
      "grad_norm": 57.75,
      "learning_rate": 3.6522358809905575e-05,
      "loss": 1.1232,
      "step": 8200
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.125,
      "learning_rate": 3.696775342953857e-05,
      "loss": 1.1474,
      "step": 8300
    },
    {
      "epoch": 0.01,
      "grad_norm": 93.0,
      "learning_rate": 3.741314804917157e-05,
      "loss": 1.0526,
      "step": 8400
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.125,
      "learning_rate": 3.785854266880456e-05,
      "loss": 1.2,
      "step": 8500
    },
    {
      "epoch": 0.01,
      "grad_norm": 46.75,
      "learning_rate": 3.830393728843756e-05,
      "loss": 1.1716,
      "step": 8600
    },
    {
      "epoch": 0.01,
      "grad_norm": 54.75,
      "learning_rate": 3.874933190807055e-05,
      "loss": 1.042,
      "step": 8700
    },
    {
      "epoch": 0.01,
      "grad_norm": 51.0,
      "learning_rate": 3.919472652770354e-05,
      "loss": 1.1756,
      "step": 8800
    },
    {
      "epoch": 0.01,
      "grad_norm": 48.25,
      "learning_rate": 3.964012114733654e-05,
      "loss": 1.2597,
      "step": 8900
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.75,
      "learning_rate": 4.008551576696954e-05,
      "loss": 1.0449,
      "step": 9000
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.25,
      "learning_rate": 4.0530910386602536e-05,
      "loss": 1.2163,
      "step": 9100
    },
    {
      "epoch": 0.01,
      "grad_norm": 80.0,
      "learning_rate": 4.097630500623553e-05,
      "loss": 1.2302,
      "step": 9200
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.0,
      "learning_rate": 4.142169962586852e-05,
      "loss": 1.1843,
      "step": 9300
    },
    {
      "epoch": 0.01,
      "grad_norm": 44.25,
      "learning_rate": 4.186709424550152e-05,
      "loss": 1.0384,
      "step": 9400
    },
    {
      "epoch": 0.01,
      "grad_norm": 66.5,
      "learning_rate": 4.231248886513451e-05,
      "loss": 1.0315,
      "step": 9500
    },
    {
      "epoch": 0.01,
      "grad_norm": 72.5,
      "learning_rate": 4.2757883484767506e-05,
      "loss": 1.1746,
      "step": 9600
    },
    {
      "epoch": 0.01,
      "grad_norm": 37.5,
      "learning_rate": 4.3203278104400504e-05,
      "loss": 1.2093,
      "step": 9700
    },
    {
      "epoch": 0.01,
      "grad_norm": 81.5,
      "learning_rate": 4.3648672724033495e-05,
      "loss": 1.3342,
      "step": 9800
    },
    {
      "epoch": 0.01,
      "grad_norm": 88.5,
      "learning_rate": 4.409406734366649e-05,
      "loss": 1.1162,
      "step": 9900
    },
    {
      "epoch": 0.01,
      "grad_norm": 356.0,
      "learning_rate": 4.4539461963299484e-05,
      "loss": 0.8431,
      "step": 10000
    },
    {
      "epoch": 0.01,
      "grad_norm": 75.5,
      "learning_rate": 4.498485658293248e-05,
      "loss": 1.15,
      "step": 10100
    },
    {
      "epoch": 0.01,
      "grad_norm": 51.0,
      "learning_rate": 4.543025120256548e-05,
      "loss": 1.1635,
      "step": 10200
    },
    {
      "epoch": 0.01,
      "grad_norm": 136.0,
      "learning_rate": 4.587564582219847e-05,
      "loss": 1.1175,
      "step": 10300
    },
    {
      "epoch": 0.01,
      "grad_norm": 180.0,
      "learning_rate": 4.632104044183146e-05,
      "loss": 1.0931,
      "step": 10400
    },
    {
      "epoch": 0.01,
      "grad_norm": 107.5,
      "learning_rate": 4.676643506146446e-05,
      "loss": 1.3118,
      "step": 10500
    },
    {
      "epoch": 0.01,
      "grad_norm": 34.0,
      "learning_rate": 4.721182968109745e-05,
      "loss": 1.0301,
      "step": 10600
    },
    {
      "epoch": 0.01,
      "grad_norm": 33.75,
      "learning_rate": 4.765722430073045e-05,
      "loss": 0.9912,
      "step": 10700
    },
    {
      "epoch": 0.01,
      "grad_norm": 48.5,
      "learning_rate": 4.810261892036345e-05,
      "loss": 1.1649,
      "step": 10800
    },
    {
      "epoch": 0.01,
      "grad_norm": 88.5,
      "learning_rate": 4.854801353999644e-05,
      "loss": 1.4056,
      "step": 10900
    },
    {
      "epoch": 0.01,
      "grad_norm": 143.0,
      "learning_rate": 4.899340815962944e-05,
      "loss": 1.1315,
      "step": 11000
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.5,
      "learning_rate": 4.943880277926243e-05,
      "loss": 1.231,
      "step": 11100
    },
    {
      "epoch": 0.01,
      "grad_norm": 47.75,
      "learning_rate": 4.988419739889542e-05,
      "loss": 1.2519,
      "step": 11200
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5625,
      "learning_rate": 4.999667068583872e-05,
      "loss": 1.1036,
      "step": 11300
    },
    {
      "epoch": 0.01,
      "grad_norm": 72.5,
      "learning_rate": 4.99921716126478e-05,
      "loss": 1.0886,
      "step": 11400
    },
    {
      "epoch": 0.01,
      "grad_norm": 102.5,
      "learning_rate": 4.998767253945688e-05,
      "loss": 1.1742,
      "step": 11500
    },
    {
      "epoch": 0.01,
      "grad_norm": 127.0,
      "learning_rate": 4.9983173466265954e-05,
      "loss": 1.2533,
      "step": 11600
    },
    {
      "epoch": 0.01,
      "grad_norm": 221.0,
      "learning_rate": 4.997867439307503e-05,
      "loss": 1.1356,
      "step": 11700
    },
    {
      "epoch": 0.01,
      "grad_norm": 132.0,
      "learning_rate": 4.997417531988411e-05,
      "loss": 1.0534,
      "step": 11800
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.625,
      "learning_rate": 4.996967624669318e-05,
      "loss": 0.936,
      "step": 11900
    },
    {
      "epoch": 0.01,
      "grad_norm": 56.5,
      "learning_rate": 4.9965177173502256e-05,
      "loss": 1.0134,
      "step": 12000
    },
    {
      "epoch": 0.01,
      "grad_norm": 37.75,
      "learning_rate": 4.996067810031134e-05,
      "loss": 1.1623,
      "step": 12100
    },
    {
      "epoch": 0.01,
      "grad_norm": 175.0,
      "learning_rate": 4.995617902712041e-05,
      "loss": 1.0928,
      "step": 12200
    },
    {
      "epoch": 0.01,
      "grad_norm": 44.5,
      "learning_rate": 4.9951679953929495e-05,
      "loss": 1.2636,
      "step": 12300
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.625,
      "learning_rate": 4.994718088073857e-05,
      "loss": 1.07,
      "step": 12400
    },
    {
      "epoch": 0.01,
      "grad_norm": 28.75,
      "learning_rate": 4.9942681807547646e-05,
      "loss": 0.9899,
      "step": 12500
    },
    {
      "epoch": 0.01,
      "grad_norm": 132.0,
      "learning_rate": 4.993818273435673e-05,
      "loss": 1.1851,
      "step": 12600
    },
    {
      "epoch": 0.01,
      "grad_norm": 133.0,
      "learning_rate": 4.9933683661165803e-05,
      "loss": 1.0886,
      "step": 12700
    },
    {
      "epoch": 0.01,
      "grad_norm": 84.0,
      "learning_rate": 4.992918458797488e-05,
      "loss": 1.0992,
      "step": 12800
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.48046875,
      "learning_rate": 4.992468551478396e-05,
      "loss": 1.1081,
      "step": 12900
    },
    {
      "epoch": 0.01,
      "grad_norm": 33.0,
      "learning_rate": 4.9920186441593036e-05,
      "loss": 1.3126,
      "step": 13000
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.5,
      "learning_rate": 4.991568736840211e-05,
      "loss": 1.1685,
      "step": 13100
    },
    {
      "epoch": 0.01,
      "grad_norm": 53.25,
      "learning_rate": 4.991118829521119e-05,
      "loss": 1.118,
      "step": 13200
    },
    {
      "epoch": 0.01,
      "grad_norm": 410.0,
      "learning_rate": 4.990668922202026e-05,
      "loss": 1.0968,
      "step": 13300
    },
    {
      "epoch": 0.01,
      "grad_norm": 56.25,
      "learning_rate": 4.9902190148829344e-05,
      "loss": 1.1793,
      "step": 13400
    },
    {
      "epoch": 0.01,
      "grad_norm": 25.75,
      "learning_rate": 4.989769107563842e-05,
      "loss": 1.1671,
      "step": 13500
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.25,
      "learning_rate": 4.9893192002447495e-05,
      "loss": 1.1825,
      "step": 13600
    },
    {
      "epoch": 0.01,
      "grad_norm": 105.5,
      "learning_rate": 4.988869292925658e-05,
      "loss": 1.1509,
      "step": 13700
    },
    {
      "epoch": 0.01,
      "grad_norm": 79.5,
      "learning_rate": 4.988419385606565e-05,
      "loss": 1.1337,
      "step": 13800
    },
    {
      "epoch": 0.01,
      "grad_norm": 36.75,
      "learning_rate": 4.987969478287473e-05,
      "loss": 1.0261,
      "step": 13900
    },
    {
      "epoch": 0.01,
      "grad_norm": 48.75,
      "learning_rate": 4.987519570968381e-05,
      "loss": 1.2461,
      "step": 14000
    },
    {
      "epoch": 0.01,
      "grad_norm": 52.75,
      "learning_rate": 4.9870696636492885e-05,
      "loss": 1.1742,
      "step": 14100
    },
    {
      "epoch": 0.01,
      "grad_norm": 236.0,
      "learning_rate": 4.986619756330197e-05,
      "loss": 0.9742,
      "step": 14200
    },
    {
      "epoch": 0.01,
      "grad_norm": 402.0,
      "learning_rate": 4.986169849011104e-05,
      "loss": 1.2615,
      "step": 14300
    },
    {
      "epoch": 0.01,
      "grad_norm": 120.0,
      "learning_rate": 4.985719941692012e-05,
      "loss": 1.3339,
      "step": 14400
    },
    {
      "epoch": 0.01,
      "grad_norm": 66.0,
      "learning_rate": 4.9852700343729193e-05,
      "loss": 1.2984,
      "step": 14500
    },
    {
      "epoch": 0.01,
      "grad_norm": 37.5,
      "learning_rate": 4.984820127053827e-05,
      "loss": 0.9931,
      "step": 14600
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.625,
      "learning_rate": 4.9843702197347344e-05,
      "loss": 1.1177,
      "step": 14700
    },
    {
      "epoch": 0.01,
      "grad_norm": 143.0,
      "learning_rate": 4.9839203124156426e-05,
      "loss": 1.1625,
      "step": 14800
    },
    {
      "epoch": 0.01,
      "grad_norm": 58.75,
      "learning_rate": 4.98347040509655e-05,
      "loss": 1.2321,
      "step": 14900
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.625,
      "learning_rate": 4.9830204977774584e-05,
      "loss": 1.2545,
      "step": 15000
    },
    {
      "epoch": 0.01,
      "grad_norm": 40.75,
      "learning_rate": 4.982570590458366e-05,
      "loss": 1.1757,
      "step": 15100
    },
    {
      "epoch": 0.01,
      "grad_norm": 44.5,
      "learning_rate": 4.9821206831392734e-05,
      "loss": 1.1684,
      "step": 15200
    },
    {
      "epoch": 0.01,
      "grad_norm": 31.0,
      "learning_rate": 4.9816707758201816e-05,
      "loss": 1.2152,
      "step": 15300
    },
    {
      "epoch": 0.01,
      "grad_norm": 56.5,
      "learning_rate": 4.981220868501089e-05,
      "loss": 1.2536,
      "step": 15400
    },
    {
      "epoch": 0.01,
      "grad_norm": 93.0,
      "learning_rate": 4.980770961181997e-05,
      "loss": 1.2398,
      "step": 15500
    },
    {
      "epoch": 0.01,
      "grad_norm": 63.5,
      "learning_rate": 4.980321053862905e-05,
      "loss": 1.0495,
      "step": 15600
    },
    {
      "epoch": 0.01,
      "grad_norm": 294.0,
      "learning_rate": 4.9798711465438125e-05,
      "loss": 1.2303,
      "step": 15700
    },
    {
      "epoch": 0.01,
      "grad_norm": 97.0,
      "learning_rate": 4.97942123922472e-05,
      "loss": 1.0398,
      "step": 15800
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.125,
      "learning_rate": 4.9789713319056275e-05,
      "loss": 1.2357,
      "step": 15900
    },
    {
      "epoch": 0.01,
      "grad_norm": 64.0,
      "learning_rate": 4.978521424586535e-05,
      "loss": 1.1963,
      "step": 16000
    },
    {
      "epoch": 0.01,
      "grad_norm": 69.0,
      "learning_rate": 4.978071517267443e-05,
      "loss": 1.1217,
      "step": 16100
    },
    {
      "epoch": 0.01,
      "grad_norm": 73.0,
      "learning_rate": 4.977621609948351e-05,
      "loss": 1.3172,
      "step": 16200
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.4375,
      "learning_rate": 4.977171702629258e-05,
      "loss": 1.0252,
      "step": 16300
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.5625,
      "learning_rate": 4.9767217953101666e-05,
      "loss": 1.0254,
      "step": 16400
    },
    {
      "epoch": 0.01,
      "grad_norm": 28.375,
      "learning_rate": 4.976271887991074e-05,
      "loss": 1.116,
      "step": 16500
    },
    {
      "epoch": 0.01,
      "grad_norm": 664.0,
      "learning_rate": 4.9758219806719816e-05,
      "loss": 1.1861,
      "step": 16600
    },
    {
      "epoch": 0.01,
      "grad_norm": 157.0,
      "learning_rate": 4.97537207335289e-05,
      "loss": 1.1762,
      "step": 16700
    },
    {
      "epoch": 0.01,
      "grad_norm": 42.75,
      "learning_rate": 4.9749221660337974e-05,
      "loss": 1.2372,
      "step": 16800
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.578125,
      "learning_rate": 4.974472258714705e-05,
      "loss": 1.1629,
      "step": 16900
    },
    {
      "epoch": 0.02,
      "grad_norm": 196.0,
      "learning_rate": 4.974022351395613e-05,
      "loss": 1.1746,
      "step": 17000
    },
    {
      "epoch": 0.02,
      "grad_norm": 43.5,
      "learning_rate": 4.97357244407652e-05,
      "loss": 1.1731,
      "step": 17100
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.375,
      "learning_rate": 4.973122536757428e-05,
      "loss": 1.0003,
      "step": 17200
    },
    {
      "epoch": 0.02,
      "grad_norm": 44.75,
      "learning_rate": 4.972672629438336e-05,
      "loss": 1.0736,
      "step": 17300
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5625,
      "learning_rate": 4.972222722119243e-05,
      "loss": 1.2785,
      "step": 17400
    },
    {
      "epoch": 0.02,
      "grad_norm": 31.5,
      "learning_rate": 4.9717728148001515e-05,
      "loss": 1.0953,
      "step": 17500
    },
    {
      "epoch": 0.02,
      "grad_norm": 83.5,
      "learning_rate": 4.971322907481059e-05,
      "loss": 1.1549,
      "step": 17600
    },
    {
      "epoch": 0.02,
      "grad_norm": 54.0,
      "learning_rate": 4.970873000161967e-05,
      "loss": 1.3649,
      "step": 17700
    },
    {
      "epoch": 0.02,
      "grad_norm": 182.0,
      "learning_rate": 4.970423092842875e-05,
      "loss": 1.109,
      "step": 17800
    },
    {
      "epoch": 0.02,
      "grad_norm": 233.0,
      "learning_rate": 4.969973185523782e-05,
      "loss": 1.1443,
      "step": 17900
    },
    {
      "epoch": 0.02,
      "grad_norm": 65.5,
      "learning_rate": 4.9695232782046905e-05,
      "loss": 1.1159,
      "step": 18000
    },
    {
      "epoch": 0.02,
      "grad_norm": 494.0,
      "learning_rate": 4.969073370885598e-05,
      "loss": 1.1072,
      "step": 18100
    },
    {
      "epoch": 0.02,
      "grad_norm": 105.0,
      "learning_rate": 4.9686234635665055e-05,
      "loss": 1.0828,
      "step": 18200
    },
    {
      "epoch": 0.02,
      "grad_norm": 32.0,
      "learning_rate": 4.968173556247414e-05,
      "loss": 1.092,
      "step": 18300
    },
    {
      "epoch": 0.02,
      "grad_norm": 40.5,
      "learning_rate": 4.9677236489283206e-05,
      "loss": 1.093,
      "step": 18400
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.0,
      "learning_rate": 4.967273741609229e-05,
      "loss": 1.283,
      "step": 18500
    },
    {
      "epoch": 0.02,
      "grad_norm": 69.0,
      "learning_rate": 4.9668238342901364e-05,
      "loss": 1.2988,
      "step": 18600
    },
    {
      "epoch": 0.02,
      "grad_norm": 102.5,
      "learning_rate": 4.966373926971044e-05,
      "loss": 1.2539,
      "step": 18700
    },
    {
      "epoch": 0.02,
      "grad_norm": 54.5,
      "learning_rate": 4.965924019651952e-05,
      "loss": 0.9519,
      "step": 18800
    },
    {
      "epoch": 0.02,
      "grad_norm": 52.0,
      "learning_rate": 4.9654741123328596e-05,
      "loss": 1.0343,
      "step": 18900
    },
    {
      "epoch": 0.02,
      "grad_norm": 91.0,
      "learning_rate": 4.965024205013767e-05,
      "loss": 1.0141,
      "step": 19000
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.5,
      "learning_rate": 4.9645742976946754e-05,
      "loss": 1.0664,
      "step": 19100
    },
    {
      "epoch": 0.02,
      "grad_norm": 82.0,
      "learning_rate": 4.964124390375583e-05,
      "loss": 1.1607,
      "step": 19200
    },
    {
      "epoch": 0.02,
      "grad_norm": 133.0,
      "learning_rate": 4.9636744830564905e-05,
      "loss": 1.0688,
      "step": 19300
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.5,
      "learning_rate": 4.963224575737399e-05,
      "loss": 1.006,
      "step": 19400
    },
    {
      "epoch": 0.02,
      "grad_norm": 29.125,
      "learning_rate": 4.962774668418306e-05,
      "loss": 1.2201,
      "step": 19500
    },
    {
      "epoch": 0.02,
      "grad_norm": 214.0,
      "learning_rate": 4.962324761099214e-05,
      "loss": 1.1192,
      "step": 19600
    },
    {
      "epoch": 0.02,
      "grad_norm": 34.75,
      "learning_rate": 4.961874853780121e-05,
      "loss": 1.1362,
      "step": 19700
    },
    {
      "epoch": 0.02,
      "grad_norm": 156.0,
      "learning_rate": 4.961424946461029e-05,
      "loss": 1.1204,
      "step": 19800
    },
    {
      "epoch": 0.02,
      "grad_norm": 157.0,
      "learning_rate": 4.960975039141937e-05,
      "loss": 1.0584,
      "step": 19900
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.5,
      "learning_rate": 4.9605251318228445e-05,
      "loss": 1.2305,
      "step": 20000
    },
    {
      "epoch": 0.02,
      "grad_norm": 127.0,
      "learning_rate": 4.960075224503752e-05,
      "loss": 1.0515,
      "step": 20100
    },
    {
      "epoch": 0.02,
      "grad_norm": 126.5,
      "learning_rate": 4.95962531718466e-05,
      "loss": 1.1291,
      "step": 20200
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.75,
      "learning_rate": 4.959175409865568e-05,
      "loss": 1.2646,
      "step": 20300
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.25,
      "learning_rate": 4.958725502546476e-05,
      "loss": 1.0123,
      "step": 20400
    },
    {
      "epoch": 0.02,
      "grad_norm": 39.25,
      "learning_rate": 4.9582755952273836e-05,
      "loss": 1.2387,
      "step": 20500
    },
    {
      "epoch": 0.02,
      "grad_norm": 103.5,
      "learning_rate": 4.957825687908291e-05,
      "loss": 1.097,
      "step": 20600
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.25,
      "learning_rate": 4.957375780589199e-05,
      "loss": 0.9842,
      "step": 20700
    },
    {
      "epoch": 0.02,
      "grad_norm": 41.75,
      "learning_rate": 4.956925873270107e-05,
      "loss": 1.018,
      "step": 20800
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.5,
      "learning_rate": 4.9564759659510144e-05,
      "loss": 1.0691,
      "step": 20900
    },
    {
      "epoch": 0.02,
      "grad_norm": 72.0,
      "learning_rate": 4.956026058631922e-05,
      "loss": 1.1096,
      "step": 21000
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.0006103515625,
      "learning_rate": 4.9555761513128295e-05,
      "loss": 1.2363,
      "step": 21100
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.75,
      "learning_rate": 4.955126243993738e-05,
      "loss": 1.1209,
      "step": 21200
    },
    {
      "epoch": 0.02,
      "grad_norm": 43.25,
      "learning_rate": 4.954676336674645e-05,
      "loss": 1.3024,
      "step": 21300
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.125,
      "learning_rate": 4.954226429355553e-05,
      "loss": 1.1737,
      "step": 21400
    },
    {
      "epoch": 0.02,
      "grad_norm": 203.0,
      "learning_rate": 4.953776522036461e-05,
      "loss": 1.0662,
      "step": 21500
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.0,
      "learning_rate": 4.9533266147173685e-05,
      "loss": 1.0818,
      "step": 21600
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.3125,
      "learning_rate": 4.952876707398276e-05,
      "loss": 1.1323,
      "step": 21700
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.59375,
      "learning_rate": 4.952426800079184e-05,
      "loss": 0.9627,
      "step": 21800
    },
    {
      "epoch": 0.02,
      "grad_norm": 145.0,
      "learning_rate": 4.951976892760092e-05,
      "loss": 1.1693,
      "step": 21900
    },
    {
      "epoch": 0.02,
      "grad_norm": 65.5,
      "learning_rate": 4.951526985440999e-05,
      "loss": 1.2508,
      "step": 22000
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.875,
      "learning_rate": 4.9510770781219075e-05,
      "loss": 0.9449,
      "step": 22100
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.625,
      "learning_rate": 4.950627170802815e-05,
      "loss": 0.9817,
      "step": 22200
    },
    {
      "epoch": 0.02,
      "grad_norm": 83.0,
      "learning_rate": 4.9501772634837226e-05,
      "loss": 1.1505,
      "step": 22300
    },
    {
      "epoch": 0.02,
      "grad_norm": 33.0,
      "learning_rate": 4.94972735616463e-05,
      "loss": 1.0348,
      "step": 22400
    },
    {
      "epoch": 0.02,
      "grad_norm": 77.5,
      "learning_rate": 4.9492774488455376e-05,
      "loss": 1.1222,
      "step": 22500
    },
    {
      "epoch": 0.02,
      "grad_norm": 40.25,
      "learning_rate": 4.948827541526446e-05,
      "loss": 1.2024,
      "step": 22600
    },
    {
      "epoch": 0.02,
      "grad_norm": 103.5,
      "learning_rate": 4.9483776342073534e-05,
      "loss": 1.1287,
      "step": 22700
    },
    {
      "epoch": 0.02,
      "grad_norm": 45.25,
      "learning_rate": 4.947927726888261e-05,
      "loss": 1.1186,
      "step": 22800
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.625,
      "learning_rate": 4.947477819569169e-05,
      "loss": 1.0359,
      "step": 22900
    },
    {
      "epoch": 0.02,
      "grad_norm": 36.25,
      "learning_rate": 4.947027912250077e-05,
      "loss": 0.9546,
      "step": 23000
    },
    {
      "epoch": 0.02,
      "grad_norm": 73.5,
      "learning_rate": 4.946578004930985e-05,
      "loss": 1.2416,
      "step": 23100
    },
    {
      "epoch": 0.02,
      "grad_norm": 33.0,
      "learning_rate": 4.9461280976118924e-05,
      "loss": 1.0936,
      "step": 23200
    },
    {
      "epoch": 0.02,
      "grad_norm": 74.5,
      "learning_rate": 4.9456781902928e-05,
      "loss": 1.1805,
      "step": 23300
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.90625,
      "learning_rate": 4.945228282973708e-05,
      "loss": 1.1044,
      "step": 23400
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.375,
      "learning_rate": 4.944778375654616e-05,
      "loss": 1.1009,
      "step": 23500
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.125,
      "learning_rate": 4.9443284683355225e-05,
      "loss": 1.1798,
      "step": 23600
    },
    {
      "epoch": 0.02,
      "grad_norm": 72.0,
      "learning_rate": 4.943878561016431e-05,
      "loss": 1.0636,
      "step": 23700
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.5625,
      "learning_rate": 4.943428653697338e-05,
      "loss": 1.2526,
      "step": 23800
    },
    {
      "epoch": 0.02,
      "grad_norm": 61.75,
      "learning_rate": 4.9429787463782465e-05,
      "loss": 1.2469,
      "step": 23900
    },
    {
      "epoch": 0.02,
      "grad_norm": 48.75,
      "learning_rate": 4.942528839059154e-05,
      "loss": 1.1861,
      "step": 24000
    },
    {
      "epoch": 0.02,
      "grad_norm": 216.0,
      "learning_rate": 4.9420789317400616e-05,
      "loss": 1.1375,
      "step": 24100
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.625,
      "learning_rate": 4.94162902442097e-05,
      "loss": 1.1822,
      "step": 24200
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.875,
      "learning_rate": 4.941179117101877e-05,
      "loss": 1.1196,
      "step": 24300
    },
    {
      "epoch": 0.02,
      "grad_norm": 39.0,
      "learning_rate": 4.940729209782785e-05,
      "loss": 1.1318,
      "step": 24400
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.625,
      "learning_rate": 4.940279302463693e-05,
      "loss": 1.0661,
      "step": 24500
    },
    {
      "epoch": 0.02,
      "grad_norm": 52.75,
      "learning_rate": 4.9398293951446006e-05,
      "loss": 1.0103,
      "step": 24600
    },
    {
      "epoch": 0.02,
      "grad_norm": 124.5,
      "learning_rate": 4.939379487825508e-05,
      "loss": 1.1614,
      "step": 24700
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.1875,
      "learning_rate": 4.9389295805064163e-05,
      "loss": 1.0209,
      "step": 24800
    },
    {
      "epoch": 0.02,
      "grad_norm": 47.0,
      "learning_rate": 4.938479673187323e-05,
      "loss": 1.1871,
      "step": 24900
    },
    {
      "epoch": 0.02,
      "grad_norm": 278.0,
      "learning_rate": 4.9380297658682314e-05,
      "loss": 1.0667,
      "step": 25000
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.359375,
      "learning_rate": 4.937579858549139e-05,
      "loss": 1.0638,
      "step": 25100
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.75,
      "learning_rate": 4.9371299512300465e-05,
      "loss": 0.8903,
      "step": 25200
    },
    {
      "epoch": 0.02,
      "grad_norm": 464.0,
      "learning_rate": 4.936680043910955e-05,
      "loss": 1.0502,
      "step": 25300
    },
    {
      "epoch": 0.02,
      "grad_norm": 38.5,
      "learning_rate": 4.936230136591862e-05,
      "loss": 1.1984,
      "step": 25400
    },
    {
      "epoch": 0.02,
      "grad_norm": 32.5,
      "learning_rate": 4.93578022927277e-05,
      "loss": 1.2607,
      "step": 25500
    },
    {
      "epoch": 0.02,
      "grad_norm": 48.5,
      "learning_rate": 4.935330321953678e-05,
      "loss": 1.058,
      "step": 25600
    },
    {
      "epoch": 0.02,
      "grad_norm": 59.5,
      "learning_rate": 4.9348804146345855e-05,
      "loss": 1.0378,
      "step": 25700
    },
    {
      "epoch": 0.02,
      "grad_norm": 512.0,
      "learning_rate": 4.934430507315494e-05,
      "loss": 1.0418,
      "step": 25800
    },
    {
      "epoch": 0.02,
      "grad_norm": 47.25,
      "learning_rate": 4.933980599996401e-05,
      "loss": 1.1618,
      "step": 25900
    },
    {
      "epoch": 0.02,
      "grad_norm": 44.75,
      "learning_rate": 4.933530692677309e-05,
      "loss": 1.0471,
      "step": 26000
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.125,
      "learning_rate": 4.933080785358217e-05,
      "loss": 1.0217,
      "step": 26100
    },
    {
      "epoch": 0.02,
      "grad_norm": 112.0,
      "learning_rate": 4.932630878039124e-05,
      "loss": 1.0314,
      "step": 26200
    },
    {
      "epoch": 0.02,
      "grad_norm": 36.25,
      "learning_rate": 4.9321809707200314e-05,
      "loss": 1.1879,
      "step": 26300
    },
    {
      "epoch": 0.02,
      "grad_norm": 36.25,
      "learning_rate": 4.9317310634009396e-05,
      "loss": 1.0381,
      "step": 26400
    },
    {
      "epoch": 0.02,
      "grad_norm": 29.625,
      "learning_rate": 4.931281156081847e-05,
      "loss": 1.1444,
      "step": 26500
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.25,
      "learning_rate": 4.9308312487627553e-05,
      "loss": 0.9723,
      "step": 26600
    },
    {
      "epoch": 0.02,
      "grad_norm": 65.0,
      "learning_rate": 4.930381341443663e-05,
      "loss": 1.13,
      "step": 26700
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.875,
      "learning_rate": 4.9299314341245704e-05,
      "loss": 0.9686,
      "step": 26800
    },
    {
      "epoch": 0.02,
      "grad_norm": 68.0,
      "learning_rate": 4.9294815268054786e-05,
      "loss": 1.0283,
      "step": 26900
    },
    {
      "epoch": 0.02,
      "grad_norm": 42.25,
      "learning_rate": 4.929031619486386e-05,
      "loss": 1.1077,
      "step": 27000
    },
    {
      "epoch": 0.02,
      "grad_norm": 38.25,
      "learning_rate": 4.928581712167294e-05,
      "loss": 1.1153,
      "step": 27100
    },
    {
      "epoch": 0.02,
      "grad_norm": 28.5,
      "learning_rate": 4.928131804848202e-05,
      "loss": 1.1458,
      "step": 27200
    },
    {
      "epoch": 0.02,
      "grad_norm": 352.0,
      "learning_rate": 4.9276818975291094e-05,
      "loss": 1.0234,
      "step": 27300
    },
    {
      "epoch": 0.02,
      "grad_norm": 58.25,
      "learning_rate": 4.927231990210017e-05,
      "loss": 1.1267,
      "step": 27400
    },
    {
      "epoch": 0.02,
      "grad_norm": 326.0,
      "learning_rate": 4.9267820828909245e-05,
      "loss": 1.0168,
      "step": 27500
    },
    {
      "epoch": 0.02,
      "grad_norm": 42.25,
      "learning_rate": 4.926332175571832e-05,
      "loss": 1.0986,
      "step": 27600
    },
    {
      "epoch": 0.02,
      "grad_norm": 81.0,
      "learning_rate": 4.92588226825274e-05,
      "loss": 0.9895,
      "step": 27700
    },
    {
      "epoch": 0.02,
      "grad_norm": 160.0,
      "learning_rate": 4.925432360933648e-05,
      "loss": 1.1471,
      "step": 27800
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.578125,
      "learning_rate": 4.924982453614555e-05,
      "loss": 1.0556,
      "step": 27900
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.421875,
      "learning_rate": 4.9245325462954635e-05,
      "loss": 1.0177,
      "step": 28000
    },
    {
      "epoch": 0.03,
      "grad_norm": 82.5,
      "learning_rate": 4.924082638976371e-05,
      "loss": 1.1341,
      "step": 28100
    },
    {
      "epoch": 0.03,
      "grad_norm": 98.5,
      "learning_rate": 4.9236327316572786e-05,
      "loss": 1.2145,
      "step": 28200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.05224609375,
      "learning_rate": 4.923182824338187e-05,
      "loss": 1.0562,
      "step": 28300
    },
    {
      "epoch": 0.03,
      "grad_norm": 43.0,
      "learning_rate": 4.9227329170190943e-05,
      "loss": 1.2268,
      "step": 28400
    },
    {
      "epoch": 0.03,
      "grad_norm": 59.25,
      "learning_rate": 4.9222830097000026e-05,
      "loss": 1.1499,
      "step": 28500
    },
    {
      "epoch": 0.03,
      "grad_norm": 592.0,
      "learning_rate": 4.92183310238091e-05,
      "loss": 1.215,
      "step": 28600
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.75,
      "learning_rate": 4.9213831950618176e-05,
      "loss": 1.0979,
      "step": 28700
    },
    {
      "epoch": 0.03,
      "grad_norm": 38.25,
      "learning_rate": 4.920933287742725e-05,
      "loss": 1.128,
      "step": 28800
    },
    {
      "epoch": 0.03,
      "grad_norm": 62.5,
      "learning_rate": 4.920483380423633e-05,
      "loss": 1.046,
      "step": 28900
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.703125,
      "learning_rate": 4.92003347310454e-05,
      "loss": 1.063,
      "step": 29000
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.875,
      "learning_rate": 4.9195835657854484e-05,
      "loss": 1.0012,
      "step": 29100
    },
    {
      "epoch": 0.03,
      "grad_norm": 19.25,
      "learning_rate": 4.919133658466356e-05,
      "loss": 1.0919,
      "step": 29200
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.6875,
      "learning_rate": 4.918683751147264e-05,
      "loss": 0.9316,
      "step": 29300
    },
    {
      "epoch": 0.03,
      "grad_norm": 50.25,
      "learning_rate": 4.918233843828172e-05,
      "loss": 1.0796,
      "step": 29400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.01422119140625,
      "learning_rate": 4.917783936509079e-05,
      "loss": 1.0421,
      "step": 29500
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.75,
      "learning_rate": 4.9173340291899875e-05,
      "loss": 1.1928,
      "step": 29600
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.25,
      "learning_rate": 4.916884121870895e-05,
      "loss": 1.1847,
      "step": 29700
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.9375,
      "learning_rate": 4.9164342145518025e-05,
      "loss": 1.1563,
      "step": 29800
    },
    {
      "epoch": 0.03,
      "grad_norm": 32.5,
      "learning_rate": 4.915984307232711e-05,
      "loss": 1.179,
      "step": 29900
    },
    {
      "epoch": 0.03,
      "grad_norm": 44.0,
      "learning_rate": 4.915534399913618e-05,
      "loss": 1.1422,
      "step": 30000
    },
    {
      "epoch": 0.03,
      "grad_norm": 46.75,
      "learning_rate": 4.915084492594526e-05,
      "loss": 1.0485,
      "step": 30100
    },
    {
      "epoch": 0.03,
      "grad_norm": 88.0,
      "learning_rate": 4.9146345852754333e-05,
      "loss": 1.0639,
      "step": 30200
    },
    {
      "epoch": 0.03,
      "grad_norm": 31.5,
      "learning_rate": 4.914184677956341e-05,
      "loss": 1.1275,
      "step": 30300
    },
    {
      "epoch": 0.03,
      "grad_norm": 65.5,
      "learning_rate": 4.913734770637249e-05,
      "loss": 1.1221,
      "step": 30400
    },
    {
      "epoch": 0.03,
      "grad_norm": 84.0,
      "learning_rate": 4.9132848633181566e-05,
      "loss": 1.1461,
      "step": 30500
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.09326171875,
      "learning_rate": 4.912834955999064e-05,
      "loss": 1.0183,
      "step": 30600
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7578125,
      "learning_rate": 4.9123850486799724e-05,
      "loss": 1.1438,
      "step": 30700
    },
    {
      "epoch": 0.03,
      "grad_norm": 43.25,
      "learning_rate": 4.91193514136088e-05,
      "loss": 0.9956,
      "step": 30800
    },
    {
      "epoch": 0.03,
      "grad_norm": 120.0,
      "learning_rate": 4.9114852340417874e-05,
      "loss": 1.1497,
      "step": 30900
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.625,
      "learning_rate": 4.9110353267226956e-05,
      "loss": 1.1515,
      "step": 31000
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.5625,
      "learning_rate": 4.910585419403603e-05,
      "loss": 1.0692,
      "step": 31100
    },
    {
      "epoch": 0.03,
      "grad_norm": 113.0,
      "learning_rate": 4.9101355120845114e-05,
      "loss": 1.1573,
      "step": 31200
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.125,
      "learning_rate": 4.909685604765419e-05,
      "loss": 1.0614,
      "step": 31300
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.359375,
      "learning_rate": 4.909235697446326e-05,
      "loss": 1.1084,
      "step": 31400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.66796875,
      "learning_rate": 4.908785790127234e-05,
      "loss": 0.9684,
      "step": 31500
    },
    {
      "epoch": 0.03,
      "grad_norm": 36.0,
      "learning_rate": 4.9083358828081415e-05,
      "loss": 0.9898,
      "step": 31600
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.625,
      "learning_rate": 4.907885975489049e-05,
      "loss": 1.0407,
      "step": 31700
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.3125,
      "learning_rate": 4.907436068169957e-05,
      "loss": 1.1401,
      "step": 31800
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.171875,
      "learning_rate": 4.906986160850865e-05,
      "loss": 0.9899,
      "step": 31900
    },
    {
      "epoch": 0.03,
      "grad_norm": 63.25,
      "learning_rate": 4.906536253531773e-05,
      "loss": 1.1188,
      "step": 32000
    },
    {
      "epoch": 0.03,
      "grad_norm": 116.5,
      "learning_rate": 4.9060863462126806e-05,
      "loss": 1.1581,
      "step": 32100
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.1875,
      "learning_rate": 4.905636438893588e-05,
      "loss": 1.1116,
      "step": 32200
    },
    {
      "epoch": 0.03,
      "grad_norm": 71.0,
      "learning_rate": 4.905186531574496e-05,
      "loss": 1.0821,
      "step": 32300
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.875,
      "learning_rate": 4.904736624255404e-05,
      "loss": 1.0241,
      "step": 32400
    },
    {
      "epoch": 0.03,
      "grad_norm": 71.0,
      "learning_rate": 4.9042867169363114e-05,
      "loss": 0.9884,
      "step": 32500
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.25,
      "learning_rate": 4.9038368096172196e-05,
      "loss": 1.0877,
      "step": 32600
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.0625,
      "learning_rate": 4.9033869022981264e-05,
      "loss": 1.1312,
      "step": 32700
    },
    {
      "epoch": 0.03,
      "grad_norm": 28.25,
      "learning_rate": 4.9029369949790346e-05,
      "loss": 1.1194,
      "step": 32800
    },
    {
      "epoch": 0.03,
      "grad_norm": 72.5,
      "learning_rate": 4.902487087659942e-05,
      "loss": 1.2167,
      "step": 32900
    },
    {
      "epoch": 0.03,
      "grad_norm": 60.5,
      "learning_rate": 4.90203718034085e-05,
      "loss": 1.0633,
      "step": 33000
    },
    {
      "epoch": 0.03,
      "grad_norm": 62.25,
      "learning_rate": 4.901587273021758e-05,
      "loss": 1.1538,
      "step": 33100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.9011373657026655e-05,
      "loss": 0.9982,
      "step": 33200
    },
    {
      "epoch": 0.03,
      "grad_norm": 45.25,
      "learning_rate": 4.900687458383573e-05,
      "loss": 1.1433,
      "step": 33300
    },
    {
      "epoch": 0.03,
      "grad_norm": 548.0,
      "learning_rate": 4.900237551064481e-05,
      "loss": 1.1228,
      "step": 33400
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.125,
      "learning_rate": 4.899787643745389e-05,
      "loss": 1.0949,
      "step": 33500
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.625,
      "learning_rate": 4.899337736426296e-05,
      "loss": 0.9741,
      "step": 33600
    },
    {
      "epoch": 0.03,
      "grad_norm": 68.5,
      "learning_rate": 4.8988878291072045e-05,
      "loss": 1.1313,
      "step": 33700
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.125,
      "learning_rate": 4.898437921788112e-05,
      "loss": 0.9327,
      "step": 33800
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.625,
      "learning_rate": 4.8979880144690196e-05,
      "loss": 1.0312,
      "step": 33900
    },
    {
      "epoch": 0.03,
      "grad_norm": 28.0,
      "learning_rate": 4.897538107149927e-05,
      "loss": 1.0795,
      "step": 34000
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23828125,
      "learning_rate": 4.8970881998308346e-05,
      "loss": 1.028,
      "step": 34100
    },
    {
      "epoch": 0.03,
      "grad_norm": 336.0,
      "learning_rate": 4.896638292511743e-05,
      "loss": 0.9252,
      "step": 34200
    },
    {
      "epoch": 0.03,
      "grad_norm": 59.5,
      "learning_rate": 4.8961883851926504e-05,
      "loss": 1.0244,
      "step": 34300
    },
    {
      "epoch": 0.03,
      "grad_norm": 91.5,
      "learning_rate": 4.895738477873558e-05,
      "loss": 1.2062,
      "step": 34400
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.25,
      "learning_rate": 4.895288570554466e-05,
      "loss": 1.016,
      "step": 34500
    },
    {
      "epoch": 0.03,
      "grad_norm": 326.0,
      "learning_rate": 4.8948386632353736e-05,
      "loss": 0.946,
      "step": 34600
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.75,
      "learning_rate": 4.894388755916282e-05,
      "loss": 1.0384,
      "step": 34700
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.361328125,
      "learning_rate": 4.8939388485971894e-05,
      "loss": 1.0267,
      "step": 34800
    },
    {
      "epoch": 0.03,
      "grad_norm": 26.75,
      "learning_rate": 4.893488941278097e-05,
      "loss": 1.0807,
      "step": 34900
    },
    {
      "epoch": 0.03,
      "grad_norm": 72.0,
      "learning_rate": 4.893039033959005e-05,
      "loss": 1.0735,
      "step": 35000
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.875,
      "learning_rate": 4.892589126639913e-05,
      "loss": 1.2151,
      "step": 35100
    },
    {
      "epoch": 0.03,
      "grad_norm": 170.0,
      "learning_rate": 4.89213921932082e-05,
      "loss": 1.189,
      "step": 35200
    },
    {
      "epoch": 0.03,
      "grad_norm": 56.75,
      "learning_rate": 4.891689312001728e-05,
      "loss": 1.1221,
      "step": 35300
    },
    {
      "epoch": 0.03,
      "grad_norm": 63.75,
      "learning_rate": 4.891239404682635e-05,
      "loss": 1.1452,
      "step": 35400
    },
    {
      "epoch": 0.03,
      "grad_norm": 39.75,
      "learning_rate": 4.8907894973635435e-05,
      "loss": 1.2071,
      "step": 35500
    },
    {
      "epoch": 0.03,
      "grad_norm": 48.0,
      "learning_rate": 4.890339590044451e-05,
      "loss": 1.1314,
      "step": 35600
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.5,
      "learning_rate": 4.8898896827253586e-05,
      "loss": 1.1872,
      "step": 35700
    },
    {
      "epoch": 0.03,
      "grad_norm": 222.0,
      "learning_rate": 4.889439775406267e-05,
      "loss": 1.1534,
      "step": 35800
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.25,
      "learning_rate": 4.888989868087174e-05,
      "loss": 1.1041,
      "step": 35900
    },
    {
      "epoch": 0.03,
      "grad_norm": 79.0,
      "learning_rate": 4.888539960768082e-05,
      "loss": 1.0419,
      "step": 36000
    },
    {
      "epoch": 0.03,
      "grad_norm": 35.75,
      "learning_rate": 4.88809005344899e-05,
      "loss": 1.1561,
      "step": 36100
    },
    {
      "epoch": 0.03,
      "grad_norm": 19.0,
      "learning_rate": 4.8876401461298976e-05,
      "loss": 1.0917,
      "step": 36200
    },
    {
      "epoch": 0.03,
      "grad_norm": 45.5,
      "learning_rate": 4.887190238810805e-05,
      "loss": 1.1914,
      "step": 36300
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.0,
      "learning_rate": 4.886740331491713e-05,
      "loss": 1.0026,
      "step": 36400
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.125,
      "learning_rate": 4.886290424172621e-05,
      "loss": 1.1217,
      "step": 36500
    },
    {
      "epoch": 0.03,
      "grad_norm": 31.125,
      "learning_rate": 4.8858405168535284e-05,
      "loss": 1.0603,
      "step": 36600
    },
    {
      "epoch": 0.03,
      "grad_norm": 42.5,
      "learning_rate": 4.885390609534436e-05,
      "loss": 1.1187,
      "step": 36700
    },
    {
      "epoch": 0.03,
      "grad_norm": 368.0,
      "learning_rate": 4.8849407022153435e-05,
      "loss": 1.0626,
      "step": 36800
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.25,
      "learning_rate": 4.884490794896252e-05,
      "loss": 1.1262,
      "step": 36900
    },
    {
      "epoch": 0.03,
      "grad_norm": 62.0,
      "learning_rate": 4.884040887577159e-05,
      "loss": 1.1231,
      "step": 37000
    },
    {
      "epoch": 0.03,
      "grad_norm": 52.25,
      "learning_rate": 4.883590980258067e-05,
      "loss": 1.0812,
      "step": 37100
    },
    {
      "epoch": 0.03,
      "grad_norm": 114.0,
      "learning_rate": 4.883141072938975e-05,
      "loss": 0.9969,
      "step": 37200
    },
    {
      "epoch": 0.03,
      "grad_norm": 45.0,
      "learning_rate": 4.8826911656198825e-05,
      "loss": 1.1014,
      "step": 37300
    },
    {
      "epoch": 0.03,
      "grad_norm": 102.0,
      "learning_rate": 4.882241258300791e-05,
      "loss": 1.1065,
      "step": 37400
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.375,
      "learning_rate": 4.881791350981698e-05,
      "loss": 1.0295,
      "step": 37500
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.24609375,
      "learning_rate": 4.881341443662606e-05,
      "loss": 1.0879,
      "step": 37600
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.0,
      "learning_rate": 4.880891536343514e-05,
      "loss": 1.2481,
      "step": 37700
    },
    {
      "epoch": 0.03,
      "grad_norm": 127.0,
      "learning_rate": 4.8804416290244215e-05,
      "loss": 1.1123,
      "step": 37800
    },
    {
      "epoch": 0.03,
      "grad_norm": 46.25,
      "learning_rate": 4.8799917217053284e-05,
      "loss": 1.1555,
      "step": 37900
    },
    {
      "epoch": 0.03,
      "grad_norm": 234.0,
      "learning_rate": 4.8795418143862366e-05,
      "loss": 1.0754,
      "step": 38000
    },
    {
      "epoch": 0.03,
      "grad_norm": 31.875,
      "learning_rate": 4.879091907067144e-05,
      "loss": 1.029,
      "step": 38100
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.34375,
      "learning_rate": 4.878641999748052e-05,
      "loss": 1.1966,
      "step": 38200
    },
    {
      "epoch": 0.03,
      "grad_norm": 96.0,
      "learning_rate": 4.87819209242896e-05,
      "loss": 1.0488,
      "step": 38300
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.8125,
      "learning_rate": 4.8777421851098674e-05,
      "loss": 1.075,
      "step": 38400
    },
    {
      "epoch": 0.03,
      "grad_norm": 35.75,
      "learning_rate": 4.8772922777907756e-05,
      "loss": 1.0888,
      "step": 38500
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.3125,
      "learning_rate": 4.876842370471683e-05,
      "loss": 1.0776,
      "step": 38600
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.0,
      "learning_rate": 4.876392463152591e-05,
      "loss": 1.1729,
      "step": 38700
    },
    {
      "epoch": 0.03,
      "grad_norm": 48.25,
      "learning_rate": 4.875942555833499e-05,
      "loss": 1.0547,
      "step": 38800
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.125,
      "learning_rate": 4.8754926485144064e-05,
      "loss": 1.033,
      "step": 38900
    },
    {
      "epoch": 0.03,
      "grad_norm": 22.625,
      "learning_rate": 4.875042741195314e-05,
      "loss": 1.0052,
      "step": 39000
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.5625,
      "learning_rate": 4.874592833876222e-05,
      "loss": 1.0835,
      "step": 39100
    },
    {
      "epoch": 0.03,
      "grad_norm": 38.0,
      "learning_rate": 4.874142926557129e-05,
      "loss": 0.9219,
      "step": 39200
    },
    {
      "epoch": 0.04,
      "grad_norm": 121.5,
      "learning_rate": 4.873693019238037e-05,
      "loss": 1.0016,
      "step": 39300
    },
    {
      "epoch": 0.04,
      "grad_norm": 49.25,
      "learning_rate": 4.873243111918945e-05,
      "loss": 0.9643,
      "step": 39400
    },
    {
      "epoch": 0.04,
      "grad_norm": 40.0,
      "learning_rate": 4.872793204599852e-05,
      "loss": 1.1834,
      "step": 39500
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.125,
      "learning_rate": 4.8723432972807605e-05,
      "loss": 1.1207,
      "step": 39600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.01385498046875,
      "learning_rate": 4.871893389961668e-05,
      "loss": 1.1989,
      "step": 39700
    },
    {
      "epoch": 0.04,
      "grad_norm": 101.0,
      "learning_rate": 4.8714434826425756e-05,
      "loss": 1.0312,
      "step": 39800
    },
    {
      "epoch": 0.04,
      "grad_norm": 25.625,
      "learning_rate": 4.870993575323484e-05,
      "loss": 1.0033,
      "step": 39900
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.375,
      "learning_rate": 4.870543668004391e-05,
      "loss": 1.1085,
      "step": 40000
    },
    {
      "epoch": 0.04,
      "grad_norm": 71.0,
      "learning_rate": 4.8700937606852995e-05,
      "loss": 1.0862,
      "step": 40100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.00616455078125,
      "learning_rate": 4.869643853366207e-05,
      "loss": 1.0734,
      "step": 40200
    },
    {
      "epoch": 0.04,
      "grad_norm": 162.0,
      "learning_rate": 4.8691939460471146e-05,
      "loss": 1.1716,
      "step": 40300
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.6875,
      "learning_rate": 4.868744038728022e-05,
      "loss": 1.0932,
      "step": 40400
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.0,
      "learning_rate": 4.86829413140893e-05,
      "loss": 1.2164,
      "step": 40500
    },
    {
      "epoch": 0.04,
      "grad_norm": 46.5,
      "learning_rate": 4.867844224089837e-05,
      "loss": 1.1104,
      "step": 40600
    },
    {
      "epoch": 0.04,
      "grad_norm": 262.0,
      "learning_rate": 4.8673943167707454e-05,
      "loss": 1.0848,
      "step": 40700
    },
    {
      "epoch": 0.04,
      "grad_norm": 42.75,
      "learning_rate": 4.866944409451653e-05,
      "loss": 1.0706,
      "step": 40800
    },
    {
      "epoch": 0.04,
      "grad_norm": 78.5,
      "learning_rate": 4.866494502132561e-05,
      "loss": 1.2088,
      "step": 40900
    },
    {
      "epoch": 0.04,
      "grad_norm": 49.0,
      "learning_rate": 4.866044594813469e-05,
      "loss": 1.0286,
      "step": 41000
    },
    {
      "epoch": 0.04,
      "grad_norm": 77.0,
      "learning_rate": 4.865594687494376e-05,
      "loss": 1.0746,
      "step": 41100
    },
    {
      "epoch": 0.04,
      "grad_norm": 89.0,
      "learning_rate": 4.8651447801752844e-05,
      "loss": 1.0762,
      "step": 41200
    },
    {
      "epoch": 0.04,
      "grad_norm": 81.5,
      "learning_rate": 4.864694872856192e-05,
      "loss": 1.0354,
      "step": 41300
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.0,
      "learning_rate": 4.8642449655370995e-05,
      "loss": 1.1843,
      "step": 41400
    },
    {
      "epoch": 0.04,
      "grad_norm": 33.0,
      "learning_rate": 4.863795058218008e-05,
      "loss": 1.1052,
      "step": 41500
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.625,
      "learning_rate": 4.863345150898915e-05,
      "loss": 1.0887,
      "step": 41600
    },
    {
      "epoch": 0.04,
      "grad_norm": 98.5,
      "learning_rate": 4.862895243579823e-05,
      "loss": 1.1573,
      "step": 41700
    },
    {
      "epoch": 0.04,
      "grad_norm": 47.75,
      "learning_rate": 4.86244533626073e-05,
      "loss": 1.2246,
      "step": 41800
    },
    {
      "epoch": 0.04,
      "grad_norm": 132.0,
      "learning_rate": 4.861995428941638e-05,
      "loss": 1.1189,
      "step": 41900
    },
    {
      "epoch": 0.04,
      "grad_norm": 28.375,
      "learning_rate": 4.861545521622546e-05,
      "loss": 1.0678,
      "step": 42000
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.1875,
      "learning_rate": 4.8610956143034536e-05,
      "loss": 1.1277,
      "step": 42100
    },
    {
      "epoch": 0.04,
      "grad_norm": 107.0,
      "learning_rate": 4.860645706984361e-05,
      "loss": 0.9764,
      "step": 42200
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.5,
      "learning_rate": 4.8601957996652693e-05,
      "loss": 1.0772,
      "step": 42300
    },
    {
      "epoch": 0.04,
      "grad_norm": 61.0,
      "learning_rate": 4.859745892346177e-05,
      "loss": 0.8331,
      "step": 42400
    },
    {
      "epoch": 0.04,
      "grad_norm": 43.5,
      "learning_rate": 4.8592959850270844e-05,
      "loss": 1.1747,
      "step": 42500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.0189208984375,
      "learning_rate": 4.8588460777079926e-05,
      "loss": 1.2166,
      "step": 42600
    },
    {
      "epoch": 0.04,
      "grad_norm": 44.75,
      "learning_rate": 4.8583961703889e-05,
      "loss": 1.0518,
      "step": 42700
    },
    {
      "epoch": 0.04,
      "grad_norm": 147.0,
      "learning_rate": 4.8579462630698084e-05,
      "loss": 1.1158,
      "step": 42800
    },
    {
      "epoch": 0.04,
      "grad_norm": 99.0,
      "learning_rate": 4.857496355750716e-05,
      "loss": 0.9201,
      "step": 42900
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.0,
      "learning_rate": 4.857046448431623e-05,
      "loss": 1.1372,
      "step": 43000
    },
    {
      "epoch": 0.04,
      "grad_norm": 69.0,
      "learning_rate": 4.856596541112531e-05,
      "loss": 0.9705,
      "step": 43100
    },
    {
      "epoch": 0.04,
      "grad_norm": 25.75,
      "learning_rate": 4.8561466337934385e-05,
      "loss": 0.9867,
      "step": 43200
    },
    {
      "epoch": 0.04,
      "grad_norm": 226.0,
      "learning_rate": 4.855696726474346e-05,
      "loss": 1.119,
      "step": 43300
    },
    {
      "epoch": 0.04,
      "grad_norm": 36.5,
      "learning_rate": 4.855246819155254e-05,
      "loss": 1.1015,
      "step": 43400
    },
    {
      "epoch": 0.04,
      "grad_norm": 32.75,
      "learning_rate": 4.854796911836162e-05,
      "loss": 1.1207,
      "step": 43500
    },
    {
      "epoch": 0.04,
      "grad_norm": 77.0,
      "learning_rate": 4.85434700451707e-05,
      "loss": 1.1423,
      "step": 43600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.59765625,
      "learning_rate": 4.8538970971979775e-05,
      "loss": 1.1564,
      "step": 43700
    },
    {
      "epoch": 0.04,
      "grad_norm": 48.0,
      "learning_rate": 4.853447189878885e-05,
      "loss": 1.237,
      "step": 43800
    },
    {
      "epoch": 0.04,
      "grad_norm": 28.125,
      "learning_rate": 4.852997282559793e-05,
      "loss": 1.1437,
      "step": 43900
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.5,
      "learning_rate": 4.852547375240701e-05,
      "loss": 0.9318,
      "step": 44000
    },
    {
      "epoch": 0.04,
      "grad_norm": 1408.0,
      "learning_rate": 4.8520974679216083e-05,
      "loss": 0.9653,
      "step": 44100
    },
    {
      "epoch": 0.04,
      "grad_norm": 24.125,
      "learning_rate": 4.8516475606025166e-05,
      "loss": 0.9739,
      "step": 44200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.035888671875,
      "learning_rate": 4.8511976532834234e-05,
      "loss": 1.0071,
      "step": 44300
    },
    {
      "epoch": 0.04,
      "grad_norm": 33.5,
      "learning_rate": 4.8507477459643316e-05,
      "loss": 1.103,
      "step": 44400
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.375,
      "learning_rate": 4.850297838645239e-05,
      "loss": 1.0733,
      "step": 44500
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.0,
      "learning_rate": 4.849847931326147e-05,
      "loss": 1.1211,
      "step": 44600
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.8125,
      "learning_rate": 4.849398024007055e-05,
      "loss": 0.9089,
      "step": 44700
    },
    {
      "epoch": 0.04,
      "grad_norm": 38.0,
      "learning_rate": 4.8489481166879624e-05,
      "loss": 1.069,
      "step": 44800
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.474609375,
      "learning_rate": 4.84849820936887e-05,
      "loss": 1.0493,
      "step": 44900
    },
    {
      "epoch": 0.04,
      "grad_norm": 31.125,
      "learning_rate": 4.848048302049778e-05,
      "loss": 0.9652,
      "step": 45000
    },
    {
      "epoch": 0.04,
      "grad_norm": 41.25,
      "learning_rate": 4.847598394730686e-05,
      "loss": 1.2541,
      "step": 45100
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.21875,
      "learning_rate": 4.847148487411593e-05,
      "loss": 1.0102,
      "step": 45200
    },
    {
      "epoch": 0.04,
      "grad_norm": 52.5,
      "learning_rate": 4.8466985800925015e-05,
      "loss": 1.0949,
      "step": 45300
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.75,
      "learning_rate": 4.846248672773409e-05,
      "loss": 1.0092,
      "step": 45400
    },
    {
      "epoch": 0.04,
      "grad_norm": 36.5,
      "learning_rate": 4.845798765454317e-05,
      "loss": 1.0239,
      "step": 45500
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.375,
      "learning_rate": 4.845348858135224e-05,
      "loss": 0.9919,
      "step": 45600
    },
    {
      "epoch": 0.04,
      "grad_norm": 48.0,
      "learning_rate": 4.8448989508161316e-05,
      "loss": 1.0217,
      "step": 45700
    },
    {
      "epoch": 0.04,
      "grad_norm": 35.5,
      "learning_rate": 4.84444904349704e-05,
      "loss": 1.038,
      "step": 45800
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.34375,
      "learning_rate": 4.8439991361779473e-05,
      "loss": 1.0612,
      "step": 45900
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.8125,
      "learning_rate": 4.843549228858855e-05,
      "loss": 1.1405,
      "step": 46000
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.125,
      "learning_rate": 4.843099321539763e-05,
      "loss": 1.0643,
      "step": 46100
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.125,
      "learning_rate": 4.8426494142206706e-05,
      "loss": 1.0632,
      "step": 46200
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.875,
      "learning_rate": 4.842199506901579e-05,
      "loss": 1.2547,
      "step": 46300
    },
    {
      "epoch": 0.04,
      "grad_norm": 143.0,
      "learning_rate": 4.8417495995824864e-05,
      "loss": 1.0667,
      "step": 46400
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.75,
      "learning_rate": 4.841299692263394e-05,
      "loss": 1.0599,
      "step": 46500
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.0,
      "learning_rate": 4.840849784944302e-05,
      "loss": 1.0818,
      "step": 46600
    },
    {
      "epoch": 0.04,
      "grad_norm": 852.0,
      "learning_rate": 4.8403998776252097e-05,
      "loss": 1.1998,
      "step": 46700
    },
    {
      "epoch": 0.04,
      "grad_norm": 48.5,
      "learning_rate": 4.839949970306117e-05,
      "loss": 1.1431,
      "step": 46800
    },
    {
      "epoch": 0.04,
      "grad_norm": 33.0,
      "learning_rate": 4.839500062987025e-05,
      "loss": 1.1179,
      "step": 46900
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.625,
      "learning_rate": 4.839050155667932e-05,
      "loss": 0.9942,
      "step": 47000
    },
    {
      "epoch": 0.04,
      "grad_norm": 51.5,
      "learning_rate": 4.8386002483488405e-05,
      "loss": 1.0463,
      "step": 47100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.045654296875,
      "learning_rate": 4.838150341029748e-05,
      "loss": 1.1198,
      "step": 47200
    },
    {
      "epoch": 0.04,
      "grad_norm": 50.5,
      "learning_rate": 4.8377004337106555e-05,
      "loss": 1.0793,
      "step": 47300
    },
    {
      "epoch": 0.04,
      "grad_norm": 25.125,
      "learning_rate": 4.837250526391564e-05,
      "loss": 0.9773,
      "step": 47400
    },
    {
      "epoch": 0.04,
      "grad_norm": 72.0,
      "learning_rate": 4.836800619072471e-05,
      "loss": 1.1212,
      "step": 47500
    },
    {
      "epoch": 0.04,
      "grad_norm": 119.0,
      "learning_rate": 4.836350711753379e-05,
      "loss": 1.077,
      "step": 47600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.0037384033203125,
      "learning_rate": 4.835900804434287e-05,
      "loss": 0.9038,
      "step": 47700
    },
    {
      "epoch": 0.04,
      "grad_norm": 154.0,
      "learning_rate": 4.8354508971151946e-05,
      "loss": 1.0858,
      "step": 47800
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8359375,
      "learning_rate": 4.835000989796102e-05,
      "loss": 0.9363,
      "step": 47900
    },
    {
      "epoch": 0.04,
      "grad_norm": 46.25,
      "learning_rate": 4.83455108247701e-05,
      "loss": 1.1046,
      "step": 48000
    },
    {
      "epoch": 0.04,
      "grad_norm": 28.125,
      "learning_rate": 4.834101175157918e-05,
      "loss": 1.0324,
      "step": 48100
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.5,
      "learning_rate": 4.8336512678388254e-05,
      "loss": 1.139,
      "step": 48200
    },
    {
      "epoch": 0.04,
      "grad_norm": 81.5,
      "learning_rate": 4.833201360519733e-05,
      "loss": 1.0533,
      "step": 48300
    },
    {
      "epoch": 0.04,
      "grad_norm": 124.5,
      "learning_rate": 4.8327514532006404e-05,
      "loss": 0.9259,
      "step": 48400
    },
    {
      "epoch": 0.04,
      "grad_norm": 97.0,
      "learning_rate": 4.8323015458815487e-05,
      "loss": 0.9167,
      "step": 48500
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.625,
      "learning_rate": 4.831851638562456e-05,
      "loss": 1.1529,
      "step": 48600
    },
    {
      "epoch": 0.04,
      "grad_norm": 72.0,
      "learning_rate": 4.831401731243364e-05,
      "loss": 1.1205,
      "step": 48700
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.003814697265625,
      "learning_rate": 4.830951823924272e-05,
      "loss": 1.0103,
      "step": 48800
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.15625,
      "learning_rate": 4.8305019166051795e-05,
      "loss": 1.1474,
      "step": 48900
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.875,
      "learning_rate": 4.830052009286088e-05,
      "loss": 1.0274,
      "step": 49000
    },
    {
      "epoch": 0.04,
      "grad_norm": 28.125,
      "learning_rate": 4.829602101966995e-05,
      "loss": 1.0594,
      "step": 49100
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.84375,
      "learning_rate": 4.829152194647903e-05,
      "loss": 1.076,
      "step": 49200
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.625,
      "learning_rate": 4.828702287328811e-05,
      "loss": 0.9598,
      "step": 49300
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.96875,
      "learning_rate": 4.8282523800097185e-05,
      "loss": 1.1013,
      "step": 49400
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.625,
      "learning_rate": 4.827802472690626e-05,
      "loss": 1.0096,
      "step": 49500
    },
    {
      "epoch": 0.04,
      "grad_norm": 45.25,
      "learning_rate": 4.8273525653715336e-05,
      "loss": 1.1038,
      "step": 49600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8203125,
      "learning_rate": 4.826902658052441e-05,
      "loss": 1.1955,
      "step": 49700
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.75,
      "learning_rate": 4.826452750733349e-05,
      "loss": 0.9651,
      "step": 49800
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.0,
      "learning_rate": 4.826002843414257e-05,
      "loss": 1.0677,
      "step": 49900
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.0140380859375,
      "learning_rate": 4.8255529360951644e-05,
      "loss": 0.9213,
      "step": 50000
    },
    {
      "epoch": 0.04,
      "grad_norm": 27.75,
      "learning_rate": 4.8251030287760726e-05,
      "loss": 1.1253,
      "step": 50100
    },
    {
      "epoch": 0.04,
      "grad_norm": 1048.0,
      "learning_rate": 4.82465312145698e-05,
      "loss": 1.0198,
      "step": 50200
    },
    {
      "epoch": 0.04,
      "grad_norm": 80.5,
      "learning_rate": 4.8242032141378877e-05,
      "loss": 1.0778,
      "step": 50300
    },
    {
      "epoch": 0.04,
      "grad_norm": 120.0,
      "learning_rate": 4.823753306818796e-05,
      "loss": 1.0936,
      "step": 50400
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.0,
      "learning_rate": 4.8233033994997034e-05,
      "loss": 1.145,
      "step": 50500
    },
    {
      "epoch": 0.05,
      "grad_norm": 82.0,
      "learning_rate": 4.822853492180611e-05,
      "loss": 1.0195,
      "step": 50600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.07568359375,
      "learning_rate": 4.822403584861519e-05,
      "loss": 1.1293,
      "step": 50700
    },
    {
      "epoch": 0.05,
      "grad_norm": 592.0,
      "learning_rate": 4.821953677542426e-05,
      "loss": 1.0629,
      "step": 50800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.06787109375,
      "learning_rate": 4.821503770223334e-05,
      "loss": 1.0955,
      "step": 50900
    },
    {
      "epoch": 0.05,
      "grad_norm": 31.875,
      "learning_rate": 4.821053862904242e-05,
      "loss": 1.0129,
      "step": 51000
    },
    {
      "epoch": 0.05,
      "grad_norm": 53.25,
      "learning_rate": 4.820603955585149e-05,
      "loss": 1.0709,
      "step": 51100
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.9375,
      "learning_rate": 4.8201540482660575e-05,
      "loss": 1.0282,
      "step": 51200
    },
    {
      "epoch": 0.05,
      "grad_norm": 44.5,
      "learning_rate": 4.819704140946965e-05,
      "loss": 1.1032,
      "step": 51300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.004547119140625,
      "learning_rate": 4.8192542336278726e-05,
      "loss": 1.1192,
      "step": 51400
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.5,
      "learning_rate": 4.818804326308781e-05,
      "loss": 0.9293,
      "step": 51500
    },
    {
      "epoch": 0.05,
      "grad_norm": 34.0,
      "learning_rate": 4.818354418989688e-05,
      "loss": 0.9689,
      "step": 51600
    },
    {
      "epoch": 0.05,
      "grad_norm": 280.0,
      "learning_rate": 4.8179045116705965e-05,
      "loss": 1.0233,
      "step": 51700
    },
    {
      "epoch": 0.05,
      "grad_norm": 72.0,
      "learning_rate": 4.817454604351504e-05,
      "loss": 1.0125,
      "step": 51800
    },
    {
      "epoch": 0.05,
      "grad_norm": 33.5,
      "learning_rate": 4.8170046970324116e-05,
      "loss": 1.0061,
      "step": 51900
    },
    {
      "epoch": 0.05,
      "grad_norm": 39.75,
      "learning_rate": 4.81655478971332e-05,
      "loss": 1.0524,
      "step": 52000
    },
    {
      "epoch": 0.05,
      "grad_norm": 28.0,
      "learning_rate": 4.8161048823942266e-05,
      "loss": 1.1626,
      "step": 52100
    },
    {
      "epoch": 0.05,
      "grad_norm": 47.75,
      "learning_rate": 4.815654975075135e-05,
      "loss": 1.0667,
      "step": 52200
    },
    {
      "epoch": 0.05,
      "grad_norm": 135.0,
      "learning_rate": 4.8152050677560424e-05,
      "loss": 1.0158,
      "step": 52300
    },
    {
      "epoch": 0.05,
      "grad_norm": 41.25,
      "learning_rate": 4.81475516043695e-05,
      "loss": 1.0858,
      "step": 52400
    },
    {
      "epoch": 0.05,
      "grad_norm": 524.0,
      "learning_rate": 4.814305253117858e-05,
      "loss": 1.0831,
      "step": 52500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2470703125,
      "learning_rate": 4.813855345798766e-05,
      "loss": 1.0296,
      "step": 52600
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.875,
      "learning_rate": 4.813405438479673e-05,
      "loss": 1.0289,
      "step": 52700
    },
    {
      "epoch": 0.05,
      "grad_norm": 24.875,
      "learning_rate": 4.8129555311605814e-05,
      "loss": 1.115,
      "step": 52800
    },
    {
      "epoch": 0.05,
      "grad_norm": 203.0,
      "learning_rate": 4.812505623841489e-05,
      "loss": 0.9661,
      "step": 52900
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.40625,
      "learning_rate": 4.8120557165223965e-05,
      "loss": 0.9604,
      "step": 53000
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.25,
      "learning_rate": 4.811605809203305e-05,
      "loss": 1.0973,
      "step": 53100
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.0625,
      "learning_rate": 4.811155901884212e-05,
      "loss": 1.1134,
      "step": 53200
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.875,
      "learning_rate": 4.81070599456512e-05,
      "loss": 0.9593,
      "step": 53300
    },
    {
      "epoch": 0.05,
      "grad_norm": 1616.0,
      "learning_rate": 4.810256087246027e-05,
      "loss": 0.925,
      "step": 53400
    },
    {
      "epoch": 0.05,
      "grad_norm": 122.5,
      "learning_rate": 4.809806179926935e-05,
      "loss": 1.021,
      "step": 53500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.061767578125,
      "learning_rate": 4.809356272607843e-05,
      "loss": 1.1105,
      "step": 53600
    },
    {
      "epoch": 0.05,
      "grad_norm": 788.0,
      "learning_rate": 4.8089063652887506e-05,
      "loss": 1.2342,
      "step": 53700
    },
    {
      "epoch": 0.05,
      "grad_norm": 25.375,
      "learning_rate": 4.808456457969658e-05,
      "loss": 1.0668,
      "step": 53800
    },
    {
      "epoch": 0.05,
      "grad_norm": 113.0,
      "learning_rate": 4.808006550650566e-05,
      "loss": 1.129,
      "step": 53900
    },
    {
      "epoch": 0.05,
      "grad_norm": 31.5,
      "learning_rate": 4.807556643331474e-05,
      "loss": 1.1183,
      "step": 54000
    },
    {
      "epoch": 0.05,
      "grad_norm": 79.5,
      "learning_rate": 4.8071067360123814e-05,
      "loss": 1.0775,
      "step": 54100
    },
    {
      "epoch": 0.05,
      "grad_norm": 296.0,
      "learning_rate": 4.8066568286932896e-05,
      "loss": 1.1357,
      "step": 54200
    },
    {
      "epoch": 0.05,
      "grad_norm": 34.0,
      "learning_rate": 4.806206921374197e-05,
      "loss": 1.1471,
      "step": 54300
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.9375,
      "learning_rate": 4.8057570140551054e-05,
      "loss": 1.0694,
      "step": 54400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.0159912109375,
      "learning_rate": 4.805307106736013e-05,
      "loss": 1.1395,
      "step": 54500
    },
    {
      "epoch": 0.05,
      "grad_norm": 37.75,
      "learning_rate": 4.8048571994169204e-05,
      "loss": 1.1149,
      "step": 54600
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.75,
      "learning_rate": 4.804407292097828e-05,
      "loss": 1.2619,
      "step": 54700
    },
    {
      "epoch": 0.05,
      "grad_norm": 30.625,
      "learning_rate": 4.8039573847787355e-05,
      "loss": 1.0838,
      "step": 54800
    },
    {
      "epoch": 0.05,
      "grad_norm": 51.25,
      "learning_rate": 4.803507477459643e-05,
      "loss": 1.0213,
      "step": 54900
    },
    {
      "epoch": 0.05,
      "grad_norm": 161.0,
      "learning_rate": 4.803057570140551e-05,
      "loss": 0.9585,
      "step": 55000
    },
    {
      "epoch": 0.05,
      "grad_norm": 103.0,
      "learning_rate": 4.802607662821459e-05,
      "loss": 1.0219,
      "step": 55100
    },
    {
      "epoch": 0.05,
      "grad_norm": 41.0,
      "learning_rate": 4.802157755502367e-05,
      "loss": 1.0754,
      "step": 55200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.267578125,
      "learning_rate": 4.8017078481832745e-05,
      "loss": 1.058,
      "step": 55300
    },
    {
      "epoch": 0.05,
      "grad_norm": 23.875,
      "learning_rate": 4.801257940864182e-05,
      "loss": 1.0358,
      "step": 55400
    },
    {
      "epoch": 0.05,
      "grad_norm": 38.5,
      "learning_rate": 4.80080803354509e-05,
      "loss": 1.255,
      "step": 55500
    },
    {
      "epoch": 0.05,
      "grad_norm": 89.5,
      "learning_rate": 4.800358126225998e-05,
      "loss": 1.1077,
      "step": 55600
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.625,
      "learning_rate": 4.799908218906905e-05,
      "loss": 1.19,
      "step": 55700
    },
    {
      "epoch": 0.05,
      "grad_norm": 141.0,
      "learning_rate": 4.7994583115878135e-05,
      "loss": 1.046,
      "step": 55800
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.375,
      "learning_rate": 4.799008404268721e-05,
      "loss": 1.0007,
      "step": 55900
    },
    {
      "epoch": 0.05,
      "grad_norm": 89.0,
      "learning_rate": 4.7985584969496286e-05,
      "loss": 0.957,
      "step": 56000
    },
    {
      "epoch": 0.05,
      "grad_norm": 94.0,
      "learning_rate": 4.798108589630536e-05,
      "loss": 1.0887,
      "step": 56100
    },
    {
      "epoch": 0.05,
      "grad_norm": 25.625,
      "learning_rate": 4.797658682311444e-05,
      "loss": 1.1572,
      "step": 56200
    },
    {
      "epoch": 0.05,
      "grad_norm": 52.25,
      "learning_rate": 4.797208774992352e-05,
      "loss": 1.096,
      "step": 56300
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.5,
      "learning_rate": 4.7967588676732594e-05,
      "loss": 1.1291,
      "step": 56400
    },
    {
      "epoch": 0.05,
      "grad_norm": 46.5,
      "learning_rate": 4.796308960354167e-05,
      "loss": 1.0168,
      "step": 56500
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.625,
      "learning_rate": 4.795859053035075e-05,
      "loss": 0.9168,
      "step": 56600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.01611328125,
      "learning_rate": 4.795409145715983e-05,
      "loss": 0.9272,
      "step": 56700
    },
    {
      "epoch": 0.05,
      "grad_norm": 60.5,
      "learning_rate": 4.79495923839689e-05,
      "loss": 1.0723,
      "step": 56800
    },
    {
      "epoch": 0.05,
      "grad_norm": 54.5,
      "learning_rate": 4.7945093310777984e-05,
      "loss": 1.1016,
      "step": 56900
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.359375,
      "learning_rate": 4.794059423758706e-05,
      "loss": 1.0385,
      "step": 57000
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.625,
      "learning_rate": 4.793609516439614e-05,
      "loss": 1.0596,
      "step": 57100
    },
    {
      "epoch": 0.05,
      "grad_norm": 36.0,
      "learning_rate": 4.793159609120522e-05,
      "loss": 1.0204,
      "step": 57200
    },
    {
      "epoch": 0.05,
      "grad_norm": 26.25,
      "learning_rate": 4.7927097018014286e-05,
      "loss": 1.0025,
      "step": 57300
    },
    {
      "epoch": 0.05,
      "grad_norm": 27.0,
      "learning_rate": 4.792259794482337e-05,
      "loss": 1.0586,
      "step": 57400
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.125,
      "learning_rate": 4.791809887163244e-05,
      "loss": 1.0268,
      "step": 57500
    },
    {
      "epoch": 0.05,
      "grad_norm": 39.25,
      "learning_rate": 4.791359979844152e-05,
      "loss": 1.1759,
      "step": 57600
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.125,
      "learning_rate": 4.79091007252506e-05,
      "loss": 1.0709,
      "step": 57700
    },
    {
      "epoch": 0.05,
      "grad_norm": 105.0,
      "learning_rate": 4.7904601652059676e-05,
      "loss": 1.0584,
      "step": 57800
    },
    {
      "epoch": 0.05,
      "grad_norm": 80.0,
      "learning_rate": 4.790010257886876e-05,
      "loss": 1.0015,
      "step": 57900
    },
    {
      "epoch": 0.05,
      "grad_norm": 31.5,
      "learning_rate": 4.7895603505677834e-05,
      "loss": 1.0309,
      "step": 58000
    },
    {
      "epoch": 0.05,
      "grad_norm": 28.625,
      "learning_rate": 4.789110443248691e-05,
      "loss": 1.0922,
      "step": 58100
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.75,
      "learning_rate": 4.788660535929599e-05,
      "loss": 1.1791,
      "step": 58200
    },
    {
      "epoch": 0.05,
      "grad_norm": 3120.0,
      "learning_rate": 4.7882106286105066e-05,
      "loss": 1.2718,
      "step": 58300
    },
    {
      "epoch": 0.05,
      "grad_norm": 38.75,
      "learning_rate": 4.787760721291414e-05,
      "loss": 0.9101,
      "step": 58400
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.75,
      "learning_rate": 4.7873108139723224e-05,
      "loss": 0.9628,
      "step": 58500
    },
    {
      "epoch": 0.05,
      "grad_norm": 26.875,
      "learning_rate": 4.786860906653229e-05,
      "loss": 1.0926,
      "step": 58600
    },
    {
      "epoch": 0.05,
      "grad_norm": 29.0,
      "learning_rate": 4.7864109993341374e-05,
      "loss": 1.0893,
      "step": 58700
    },
    {
      "epoch": 0.05,
      "grad_norm": 32.0,
      "learning_rate": 4.785961092015045e-05,
      "loss": 1.0435,
      "step": 58800
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.625,
      "learning_rate": 4.7855111846959525e-05,
      "loss": 1.1382,
      "step": 58900
    },
    {
      "epoch": 0.05,
      "grad_norm": 165.0,
      "learning_rate": 4.785061277376861e-05,
      "loss": 0.9335,
      "step": 59000
    },
    {
      "epoch": 0.05,
      "grad_norm": 23.875,
      "learning_rate": 4.784611370057768e-05,
      "loss": 0.9897,
      "step": 59100
    },
    {
      "epoch": 0.05,
      "grad_norm": 99.0,
      "learning_rate": 4.784161462738676e-05,
      "loss": 1.077,
      "step": 59200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.0322265625,
      "learning_rate": 4.783711555419584e-05,
      "loss": 0.927,
      "step": 59300
    },
    {
      "epoch": 0.05,
      "grad_norm": 134.0,
      "learning_rate": 4.7832616481004915e-05,
      "loss": 0.9779,
      "step": 59400
    },
    {
      "epoch": 0.05,
      "grad_norm": 33.25,
      "learning_rate": 4.782811740781399e-05,
      "loss": 1.114,
      "step": 59500
    },
    {
      "epoch": 0.05,
      "grad_norm": 32.75,
      "learning_rate": 4.782361833462307e-05,
      "loss": 1.1215,
      "step": 59600
    },
    {
      "epoch": 0.05,
      "grad_norm": 32.25,
      "learning_rate": 4.781911926143215e-05,
      "loss": 1.0068,
      "step": 59700
    },
    {
      "epoch": 0.05,
      "grad_norm": 87.0,
      "learning_rate": 4.781462018824123e-05,
      "loss": 1.0691,
      "step": 59800
    },
    {
      "epoch": 0.05,
      "grad_norm": 24.75,
      "learning_rate": 4.78101211150503e-05,
      "loss": 0.9855,
      "step": 59900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.0074462890625,
      "learning_rate": 4.7805622041859374e-05,
      "loss": 1.0395,
      "step": 60000
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1435546875,
      "learning_rate": 4.7801122968668456e-05,
      "loss": 1.0282,
      "step": 60100
    },
    {
      "epoch": 0.05,
      "grad_norm": 62.25,
      "learning_rate": 4.779662389547753e-05,
      "loss": 1.0955,
      "step": 60200
    },
    {
      "epoch": 0.05,
      "grad_norm": 95.0,
      "learning_rate": 4.779212482228661e-05,
      "loss": 0.9779,
      "step": 60300
    },
    {
      "epoch": 0.05,
      "grad_norm": 76.5,
      "learning_rate": 4.778762574909569e-05,
      "loss": 1.1875,
      "step": 60400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.00750732421875,
      "learning_rate": 4.7783126675904764e-05,
      "loss": 1.144,
      "step": 60500
    },
    {
      "epoch": 0.05,
      "grad_norm": 123.0,
      "learning_rate": 4.7778627602713847e-05,
      "loss": 1.0817,
      "step": 60600
    },
    {
      "epoch": 0.05,
      "grad_norm": 31.5,
      "learning_rate": 4.777412852952292e-05,
      "loss": 1.0511,
      "step": 60700
    },
    {
      "epoch": 0.05,
      "grad_norm": 38.5,
      "learning_rate": 4.7769629456332e-05,
      "loss": 1.1687,
      "step": 60800
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.4375,
      "learning_rate": 4.776513038314108e-05,
      "loss": 0.9554,
      "step": 60900
    },
    {
      "epoch": 0.05,
      "grad_norm": 51.0,
      "learning_rate": 4.7760631309950155e-05,
      "loss": 1.1859,
      "step": 61000
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.75,
      "learning_rate": 4.775613223675923e-05,
      "loss": 1.1318,
      "step": 61100
    },
    {
      "epoch": 0.05,
      "grad_norm": 23.0,
      "learning_rate": 4.7751633163568305e-05,
      "loss": 0.9677,
      "step": 61200
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.46875,
      "learning_rate": 4.774713409037738e-05,
      "loss": 1.0597,
      "step": 61300
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.75,
      "learning_rate": 4.774263501718646e-05,
      "loss": 1.0387,
      "step": 61400
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.5,
      "learning_rate": 4.773813594399554e-05,
      "loss": 1.1281,
      "step": 61500
    },
    {
      "epoch": 0.05,
      "grad_norm": 47.75,
      "learning_rate": 4.7733636870804613e-05,
      "loss": 1.1926,
      "step": 61600
    },
    {
      "epoch": 0.05,
      "grad_norm": 32.25,
      "learning_rate": 4.7729137797613696e-05,
      "loss": 1.0866,
      "step": 61700
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.003997802734375,
      "learning_rate": 4.772463872442277e-05,
      "loss": 1.0067,
      "step": 61800
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.00518798828125,
      "learning_rate": 4.7720139651231846e-05,
      "loss": 0.9618,
      "step": 61900
    },
    {
      "epoch": 0.06,
      "grad_norm": 38.25,
      "learning_rate": 4.771564057804093e-05,
      "loss": 0.9947,
      "step": 62000
    },
    {
      "epoch": 0.06,
      "grad_norm": 35.25,
      "learning_rate": 4.7711141504850004e-05,
      "loss": 0.9221,
      "step": 62100
    },
    {
      "epoch": 0.06,
      "grad_norm": 17.375,
      "learning_rate": 4.770664243165908e-05,
      "loss": 1.0307,
      "step": 62200
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.146484375,
      "learning_rate": 4.770214335846816e-05,
      "loss": 1.1203,
      "step": 62300
    },
    {
      "epoch": 0.06,
      "grad_norm": 120.5,
      "learning_rate": 4.7697644285277237e-05,
      "loss": 1.1088,
      "step": 62400
    },
    {
      "epoch": 0.06,
      "grad_norm": 78.0,
      "learning_rate": 4.769314521208631e-05,
      "loss": 0.9565,
      "step": 62500
    },
    {
      "epoch": 0.06,
      "grad_norm": 25.625,
      "learning_rate": 4.768864613889539e-05,
      "loss": 1.096,
      "step": 62600
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.25,
      "learning_rate": 4.768414706570446e-05,
      "loss": 1.0787,
      "step": 62700
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.5,
      "learning_rate": 4.7679647992513545e-05,
      "loss": 1.0096,
      "step": 62800
    },
    {
      "epoch": 0.06,
      "grad_norm": 50.25,
      "learning_rate": 4.767514891932262e-05,
      "loss": 1.0513,
      "step": 62900
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.00445556640625,
      "learning_rate": 4.7670649846131695e-05,
      "loss": 0.9897,
      "step": 63000
    },
    {
      "epoch": 0.06,
      "grad_norm": 400.0,
      "learning_rate": 4.766615077294078e-05,
      "loss": 1.044,
      "step": 63100
    },
    {
      "epoch": 0.06,
      "grad_norm": 89.0,
      "learning_rate": 4.766165169974985e-05,
      "loss": 1.0449,
      "step": 63200
    },
    {
      "epoch": 0.06,
      "grad_norm": 58.0,
      "learning_rate": 4.7657152626558935e-05,
      "loss": 1.0454,
      "step": 63300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.048828125,
      "learning_rate": 4.765265355336801e-05,
      "loss": 0.9011,
      "step": 63400
    },
    {
      "epoch": 0.06,
      "grad_norm": 38.75,
      "learning_rate": 4.7648154480177086e-05,
      "loss": 1.1728,
      "step": 63500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.055419921875,
      "learning_rate": 4.764365540698617e-05,
      "loss": 1.0142,
      "step": 63600
    },
    {
      "epoch": 0.06,
      "grad_norm": 102.0,
      "learning_rate": 4.763915633379524e-05,
      "loss": 1.0692,
      "step": 63700
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.625,
      "learning_rate": 4.763465726060432e-05,
      "loss": 0.8402,
      "step": 63800
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.25,
      "learning_rate": 4.7630158187413394e-05,
      "loss": 1.135,
      "step": 63900
    },
    {
      "epoch": 0.06,
      "grad_norm": 48.0,
      "learning_rate": 4.762565911422247e-05,
      "loss": 0.9995,
      "step": 64000
    },
    {
      "epoch": 0.06,
      "grad_norm": 45.5,
      "learning_rate": 4.762116004103155e-05,
      "loss": 1.085,
      "step": 64100
    },
    {
      "epoch": 0.06,
      "grad_norm": 65.5,
      "learning_rate": 4.7616660967840627e-05,
      "loss": 1.0364,
      "step": 64200
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.25,
      "learning_rate": 4.76121618946497e-05,
      "loss": 1.1136,
      "step": 64300
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.5625,
      "learning_rate": 4.7607662821458784e-05,
      "loss": 1.0889,
      "step": 64400
    },
    {
      "epoch": 0.06,
      "grad_norm": 103.0,
      "learning_rate": 4.760316374826786e-05,
      "loss": 1.1256,
      "step": 64500
    },
    {
      "epoch": 0.06,
      "grad_norm": 34.25,
      "learning_rate": 4.7598664675076935e-05,
      "loss": 0.9965,
      "step": 64600
    },
    {
      "epoch": 0.06,
      "grad_norm": 25.5,
      "learning_rate": 4.759416560188602e-05,
      "loss": 1.0614,
      "step": 64700
    },
    {
      "epoch": 0.06,
      "grad_norm": 80.5,
      "learning_rate": 4.758966652869509e-05,
      "loss": 1.0382,
      "step": 64800
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.5625,
      "learning_rate": 4.758516745550417e-05,
      "loss": 0.8942,
      "step": 64900
    },
    {
      "epoch": 0.06,
      "grad_norm": 378.0,
      "learning_rate": 4.758066838231325e-05,
      "loss": 1.1136,
      "step": 65000
    },
    {
      "epoch": 0.06,
      "grad_norm": 62.5,
      "learning_rate": 4.757616930912232e-05,
      "loss": 1.1306,
      "step": 65100
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.212890625,
      "learning_rate": 4.75716702359314e-05,
      "loss": 0.9875,
      "step": 65200
    },
    {
      "epoch": 0.06,
      "grad_norm": 28.5,
      "learning_rate": 4.7567171162740476e-05,
      "loss": 1.0619,
      "step": 65300
    },
    {
      "epoch": 0.06,
      "grad_norm": 47.25,
      "learning_rate": 4.756267208954955e-05,
      "loss": 0.9667,
      "step": 65400
    },
    {
      "epoch": 0.06,
      "grad_norm": 79.5,
      "learning_rate": 4.755817301635863e-05,
      "loss": 1.0267,
      "step": 65500
    },
    {
      "epoch": 0.06,
      "grad_norm": 68.5,
      "learning_rate": 4.755367394316771e-05,
      "loss": 1.1107,
      "step": 65600
    },
    {
      "epoch": 0.06,
      "grad_norm": 35.5,
      "learning_rate": 4.7549174869976784e-05,
      "loss": 0.9501,
      "step": 65700
    },
    {
      "epoch": 0.06,
      "grad_norm": 74.5,
      "learning_rate": 4.7544675796785866e-05,
      "loss": 1.0483,
      "step": 65800
    },
    {
      "epoch": 0.06,
      "grad_norm": 54.5,
      "learning_rate": 4.754017672359494e-05,
      "loss": 1.1021,
      "step": 65900
    },
    {
      "epoch": 0.06,
      "grad_norm": 25.125,
      "learning_rate": 4.753567765040402e-05,
      "loss": 1.006,
      "step": 66000
    },
    {
      "epoch": 0.06,
      "grad_norm": 55.25,
      "learning_rate": 4.75311785772131e-05,
      "loss": 1.1606,
      "step": 66100
    },
    {
      "epoch": 0.06,
      "grad_norm": 112.0,
      "learning_rate": 4.7526679504022174e-05,
      "loss": 0.9369,
      "step": 66200
    },
    {
      "epoch": 0.06,
      "grad_norm": 57.0,
      "learning_rate": 4.7522180430831256e-05,
      "loss": 1.0697,
      "step": 66300
    },
    {
      "epoch": 0.06,
      "grad_norm": 62.0,
      "learning_rate": 4.7517681357640325e-05,
      "loss": 1.0148,
      "step": 66400
    },
    {
      "epoch": 0.06,
      "grad_norm": 51.5,
      "learning_rate": 4.751318228444941e-05,
      "loss": 1.0458,
      "step": 66500
    },
    {
      "epoch": 0.06,
      "grad_norm": 64.0,
      "learning_rate": 4.750868321125848e-05,
      "loss": 1.0097,
      "step": 66600
    },
    {
      "epoch": 0.06,
      "grad_norm": 29.375,
      "learning_rate": 4.750418413806756e-05,
      "loss": 1.0261,
      "step": 66700
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.125,
      "learning_rate": 4.749968506487664e-05,
      "loss": 1.1002,
      "step": 66800
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.625,
      "learning_rate": 4.7495185991685715e-05,
      "loss": 1.1513,
      "step": 66900
    },
    {
      "epoch": 0.06,
      "grad_norm": 28.0,
      "learning_rate": 4.749068691849479e-05,
      "loss": 1.0036,
      "step": 67000
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5703125,
      "learning_rate": 4.748618784530387e-05,
      "loss": 0.9781,
      "step": 67100
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0390625,
      "learning_rate": 4.748168877211295e-05,
      "loss": 1.1338,
      "step": 67200
    },
    {
      "epoch": 0.06,
      "grad_norm": 160.0,
      "learning_rate": 4.747718969892202e-05,
      "loss": 1.104,
      "step": 67300
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.5,
      "learning_rate": 4.7472690625731105e-05,
      "loss": 1.0585,
      "step": 67400
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.125,
      "learning_rate": 4.746819155254018e-05,
      "loss": 0.9737,
      "step": 67500
    },
    {
      "epoch": 0.06,
      "grad_norm": 29.5,
      "learning_rate": 4.7463692479349256e-05,
      "loss": 1.0911,
      "step": 67600
    },
    {
      "epoch": 0.06,
      "grad_norm": 48.25,
      "learning_rate": 4.745919340615833e-05,
      "loss": 1.0497,
      "step": 67700
    },
    {
      "epoch": 0.06,
      "grad_norm": 139.0,
      "learning_rate": 4.7454694332967407e-05,
      "loss": 1.0999,
      "step": 67800
    },
    {
      "epoch": 0.06,
      "grad_norm": 106.0,
      "learning_rate": 4.745019525977649e-05,
      "loss": 0.952,
      "step": 67900
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.75,
      "learning_rate": 4.7445696186585564e-05,
      "loss": 1.0808,
      "step": 68000
    },
    {
      "epoch": 0.06,
      "grad_norm": 100.0,
      "learning_rate": 4.744119711339464e-05,
      "loss": 1.129,
      "step": 68100
    },
    {
      "epoch": 0.06,
      "grad_norm": 45.75,
      "learning_rate": 4.743669804020372e-05,
      "loss": 1.1297,
      "step": 68200
    },
    {
      "epoch": 0.06,
      "grad_norm": 52.5,
      "learning_rate": 4.74321989670128e-05,
      "loss": 0.9746,
      "step": 68300
    },
    {
      "epoch": 0.06,
      "grad_norm": 41.0,
      "learning_rate": 4.742769989382187e-05,
      "loss": 1.0569,
      "step": 68400
    },
    {
      "epoch": 0.06,
      "grad_norm": 36.0,
      "learning_rate": 4.7423200820630954e-05,
      "loss": 1.0026,
      "step": 68500
    },
    {
      "epoch": 0.06,
      "grad_norm": 43.0,
      "learning_rate": 4.741870174744003e-05,
      "loss": 0.8584,
      "step": 68600
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.625,
      "learning_rate": 4.741420267424911e-05,
      "loss": 1.0649,
      "step": 68700
    },
    {
      "epoch": 0.06,
      "grad_norm": 65.5,
      "learning_rate": 4.740970360105819e-05,
      "loss": 1.1102,
      "step": 68800
    },
    {
      "epoch": 0.06,
      "grad_norm": 28.875,
      "learning_rate": 4.740520452786726e-05,
      "loss": 1.0797,
      "step": 68900
    },
    {
      "epoch": 0.06,
      "grad_norm": 80.0,
      "learning_rate": 4.740070545467634e-05,
      "loss": 1.0115,
      "step": 69000
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.3125,
      "learning_rate": 4.739620638148541e-05,
      "loss": 1.0605,
      "step": 69100
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.375,
      "learning_rate": 4.7391707308294495e-05,
      "loss": 1.1096,
      "step": 69200
    },
    {
      "epoch": 0.06,
      "grad_norm": 69.5,
      "learning_rate": 4.738720823510357e-05,
      "loss": 1.1468,
      "step": 69300
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.375,
      "learning_rate": 4.7382709161912646e-05,
      "loss": 1.1837,
      "step": 69400
    },
    {
      "epoch": 0.06,
      "grad_norm": 98.0,
      "learning_rate": 4.737821008872173e-05,
      "loss": 1.0676,
      "step": 69500
    },
    {
      "epoch": 0.06,
      "grad_norm": 35.75,
      "learning_rate": 4.73737110155308e-05,
      "loss": 1.112,
      "step": 69600
    },
    {
      "epoch": 0.06,
      "grad_norm": 74.0,
      "learning_rate": 4.736921194233988e-05,
      "loss": 0.983,
      "step": 69700
    },
    {
      "epoch": 0.06,
      "grad_norm": 60.0,
      "learning_rate": 4.736471286914896e-05,
      "loss": 1.0985,
      "step": 69800
    },
    {
      "epoch": 0.06,
      "grad_norm": 66.0,
      "learning_rate": 4.7360213795958036e-05,
      "loss": 1.1919,
      "step": 69900
    },
    {
      "epoch": 0.06,
      "grad_norm": 40.25,
      "learning_rate": 4.735571472276711e-05,
      "loss": 1.1042,
      "step": 70000
    },
    {
      "epoch": 0.06,
      "grad_norm": 46.0,
      "learning_rate": 4.7351215649576194e-05,
      "loss": 1.0638,
      "step": 70100
    },
    {
      "epoch": 0.06,
      "grad_norm": 176.0,
      "learning_rate": 4.734671657638527e-05,
      "loss": 0.9685,
      "step": 70200
    },
    {
      "epoch": 0.06,
      "grad_norm": 57.5,
      "learning_rate": 4.7342217503194344e-05,
      "loss": 1.0771,
      "step": 70300
    },
    {
      "epoch": 0.06,
      "grad_norm": 37.25,
      "learning_rate": 4.733771843000342e-05,
      "loss": 1.0881,
      "step": 70400
    },
    {
      "epoch": 0.06,
      "grad_norm": 41.0,
      "learning_rate": 4.7333219356812495e-05,
      "loss": 0.9475,
      "step": 70500
    },
    {
      "epoch": 0.06,
      "grad_norm": 83.5,
      "learning_rate": 4.732872028362158e-05,
      "loss": 1.1723,
      "step": 70600
    },
    {
      "epoch": 0.06,
      "grad_norm": 64.5,
      "learning_rate": 4.732422121043065e-05,
      "loss": 1.072,
      "step": 70700
    },
    {
      "epoch": 0.06,
      "grad_norm": 72.5,
      "learning_rate": 4.731972213723973e-05,
      "loss": 0.9838,
      "step": 70800
    },
    {
      "epoch": 0.06,
      "grad_norm": 41.5,
      "learning_rate": 4.731522306404881e-05,
      "loss": 1.078,
      "step": 70900
    },
    {
      "epoch": 0.06,
      "grad_norm": 115.5,
      "learning_rate": 4.7310723990857885e-05,
      "loss": 1.0592,
      "step": 71000
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.095703125,
      "learning_rate": 4.730622491766696e-05,
      "loss": 0.9734,
      "step": 71100
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.375,
      "learning_rate": 4.730172584447604e-05,
      "loss": 0.9183,
      "step": 71200
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.75,
      "learning_rate": 4.729722677128512e-05,
      "loss": 0.9231,
      "step": 71300
    },
    {
      "epoch": 0.06,
      "grad_norm": 80.0,
      "learning_rate": 4.72927276980942e-05,
      "loss": 1.1361,
      "step": 71400
    },
    {
      "epoch": 0.06,
      "grad_norm": 90.0,
      "learning_rate": 4.7288228624903275e-05,
      "loss": 1.0136,
      "step": 71500
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.9375,
      "learning_rate": 4.7283729551712344e-05,
      "loss": 1.0182,
      "step": 71600
    },
    {
      "epoch": 0.06,
      "grad_norm": 59.5,
      "learning_rate": 4.7279230478521426e-05,
      "loss": 0.9439,
      "step": 71700
    },
    {
      "epoch": 0.06,
      "grad_norm": 54.5,
      "learning_rate": 4.72747314053305e-05,
      "loss": 1.0811,
      "step": 71800
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.25,
      "learning_rate": 4.727023233213958e-05,
      "loss": 1.0841,
      "step": 71900
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.8125,
      "learning_rate": 4.726573325894866e-05,
      "loss": 0.988,
      "step": 72000
    },
    {
      "epoch": 0.06,
      "grad_norm": 73.5,
      "learning_rate": 4.7261234185757734e-05,
      "loss": 1.1638,
      "step": 72100
    },
    {
      "epoch": 0.06,
      "grad_norm": 69.5,
      "learning_rate": 4.7256735112566816e-05,
      "loss": 1.111,
      "step": 72200
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.125,
      "learning_rate": 4.725223603937589e-05,
      "loss": 0.9247,
      "step": 72300
    },
    {
      "epoch": 0.06,
      "grad_norm": 36.75,
      "learning_rate": 4.724773696618497e-05,
      "loss": 1.029,
      "step": 72400
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.0,
      "learning_rate": 4.724323789299405e-05,
      "loss": 1.1028,
      "step": 72500
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.6875,
      "learning_rate": 4.7238738819803124e-05,
      "loss": 1.1013,
      "step": 72600
    },
    {
      "epoch": 0.06,
      "grad_norm": 42.75,
      "learning_rate": 4.72342397466122e-05,
      "loss": 1.2265,
      "step": 72700
    },
    {
      "epoch": 0.06,
      "grad_norm": 30.625,
      "learning_rate": 4.722974067342128e-05,
      "loss": 0.9438,
      "step": 72800
    },
    {
      "epoch": 0.06,
      "grad_norm": 37.75,
      "learning_rate": 4.722524160023035e-05,
      "loss": 0.9759,
      "step": 72900
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2578125,
      "learning_rate": 4.722074252703943e-05,
      "loss": 0.9565,
      "step": 73000
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.721624345384851e-05,
      "loss": 0.9849,
      "step": 73100
    },
    {
      "epoch": 0.07,
      "grad_norm": 74.0,
      "learning_rate": 4.721174438065758e-05,
      "loss": 1.0244,
      "step": 73200
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.375,
      "learning_rate": 4.7207245307466665e-05,
      "loss": 1.1064,
      "step": 73300
    },
    {
      "epoch": 0.07,
      "grad_norm": 84.0,
      "learning_rate": 4.720274623427574e-05,
      "loss": 1.0538,
      "step": 73400
    },
    {
      "epoch": 0.07,
      "grad_norm": 27.875,
      "learning_rate": 4.7198247161084816e-05,
      "loss": 1.057,
      "step": 73500
    },
    {
      "epoch": 0.07,
      "grad_norm": 185.0,
      "learning_rate": 4.71937480878939e-05,
      "loss": 1.0565,
      "step": 73600
    },
    {
      "epoch": 0.07,
      "grad_norm": 90.5,
      "learning_rate": 4.7189249014702974e-05,
      "loss": 1.11,
      "step": 73700
    },
    {
      "epoch": 0.07,
      "grad_norm": 808.0,
      "learning_rate": 4.718474994151205e-05,
      "loss": 1.1171,
      "step": 73800
    },
    {
      "epoch": 0.07,
      "grad_norm": 191.0,
      "learning_rate": 4.718025086832113e-05,
      "loss": 1.0666,
      "step": 73900
    },
    {
      "epoch": 0.07,
      "grad_norm": 50.25,
      "learning_rate": 4.7175751795130206e-05,
      "loss": 1.0441,
      "step": 74000
    },
    {
      "epoch": 0.07,
      "grad_norm": 44.75,
      "learning_rate": 4.717125272193929e-05,
      "loss": 0.9349,
      "step": 74100
    },
    {
      "epoch": 0.07,
      "grad_norm": 54.5,
      "learning_rate": 4.716675364874836e-05,
      "loss": 1.0472,
      "step": 74200
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.25,
      "learning_rate": 4.716225457555743e-05,
      "loss": 0.9954,
      "step": 74300
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.75,
      "learning_rate": 4.7157755502366514e-05,
      "loss": 0.9964,
      "step": 74400
    },
    {
      "epoch": 0.07,
      "grad_norm": 38.0,
      "learning_rate": 4.715325642917559e-05,
      "loss": 1.0264,
      "step": 74500
    },
    {
      "epoch": 0.07,
      "grad_norm": 19.0,
      "learning_rate": 4.7148757355984665e-05,
      "loss": 1.1607,
      "step": 74600
    },
    {
      "epoch": 0.07,
      "grad_norm": 23.125,
      "learning_rate": 4.714425828279375e-05,
      "loss": 1.0605,
      "step": 74700
    },
    {
      "epoch": 0.07,
      "grad_norm": 78.5,
      "learning_rate": 4.713975920960282e-05,
      "loss": 0.9948,
      "step": 74800
    },
    {
      "epoch": 0.07,
      "grad_norm": 48.75,
      "learning_rate": 4.7135260136411905e-05,
      "loss": 0.9176,
      "step": 74900
    },
    {
      "epoch": 0.07,
      "grad_norm": 45.5,
      "learning_rate": 4.713076106322098e-05,
      "loss": 0.9902,
      "step": 75000
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.00396728515625,
      "learning_rate": 4.7126261990030055e-05,
      "loss": 1.1327,
      "step": 75100
    },
    {
      "epoch": 0.07,
      "grad_norm": 34.75,
      "learning_rate": 4.712176291683914e-05,
      "loss": 1.054,
      "step": 75200
    },
    {
      "epoch": 0.07,
      "grad_norm": 38.5,
      "learning_rate": 4.711726384364821e-05,
      "loss": 0.9967,
      "step": 75300
    },
    {
      "epoch": 0.07,
      "grad_norm": 35.0,
      "learning_rate": 4.711276477045729e-05,
      "loss": 1.0065,
      "step": 75400
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.875,
      "learning_rate": 4.7108265697266364e-05,
      "loss": 0.9735,
      "step": 75500
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.039794921875,
      "learning_rate": 4.710376662407544e-05,
      "loss": 0.9869,
      "step": 75600
    },
    {
      "epoch": 0.07,
      "grad_norm": 75.0,
      "learning_rate": 4.709926755088452e-05,
      "loss": 1.1004,
      "step": 75700
    },
    {
      "epoch": 0.07,
      "grad_norm": 27.75,
      "learning_rate": 4.7094768477693596e-05,
      "loss": 1.0081,
      "step": 75800
    },
    {
      "epoch": 0.07,
      "grad_norm": 144.0,
      "learning_rate": 4.709026940450267e-05,
      "loss": 1.1244,
      "step": 75900
    },
    {
      "epoch": 0.07,
      "grad_norm": 196.0,
      "learning_rate": 4.7085770331311754e-05,
      "loss": 1.0563,
      "step": 76000
    },
    {
      "epoch": 0.07,
      "grad_norm": 180.0,
      "learning_rate": 4.708127125812083e-05,
      "loss": 1.0529,
      "step": 76100
    },
    {
      "epoch": 0.07,
      "grad_norm": 38.0,
      "learning_rate": 4.7076772184929904e-05,
      "loss": 0.9047,
      "step": 76200
    },
    {
      "epoch": 0.07,
      "grad_norm": 82.0,
      "learning_rate": 4.7072273111738987e-05,
      "loss": 1.048,
      "step": 76300
    },
    {
      "epoch": 0.07,
      "grad_norm": 28.5,
      "learning_rate": 4.706777403854806e-05,
      "loss": 1.146,
      "step": 76400
    },
    {
      "epoch": 0.07,
      "grad_norm": 74.5,
      "learning_rate": 4.706327496535714e-05,
      "loss": 1.0799,
      "step": 76500
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.75,
      "learning_rate": 4.705877589216622e-05,
      "loss": 1.1352,
      "step": 76600
    },
    {
      "epoch": 0.07,
      "grad_norm": 46.25,
      "learning_rate": 4.7054276818975295e-05,
      "loss": 1.0056,
      "step": 76700
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.96875,
      "learning_rate": 4.704977774578437e-05,
      "loss": 1.1245,
      "step": 76800
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.25,
      "learning_rate": 4.7045278672593445e-05,
      "loss": 1.352,
      "step": 76900
    },
    {
      "epoch": 0.07,
      "grad_norm": 21.5,
      "learning_rate": 4.704077959940252e-05,
      "loss": 1.0695,
      "step": 77000
    },
    {
      "epoch": 0.07,
      "grad_norm": 25.0,
      "learning_rate": 4.70362805262116e-05,
      "loss": 1.0156,
      "step": 77100
    },
    {
      "epoch": 0.07,
      "grad_norm": 40.25,
      "learning_rate": 4.703178145302068e-05,
      "loss": 0.9797,
      "step": 77200
    },
    {
      "epoch": 0.07,
      "grad_norm": 27.875,
      "learning_rate": 4.7027282379829754e-05,
      "loss": 1.0771,
      "step": 77300
    },
    {
      "epoch": 0.07,
      "grad_norm": 80.5,
      "learning_rate": 4.7022783306638836e-05,
      "loss": 0.9401,
      "step": 77400
    },
    {
      "epoch": 0.07,
      "grad_norm": 51.5,
      "learning_rate": 4.701828423344791e-05,
      "loss": 0.9542,
      "step": 77500
    },
    {
      "epoch": 0.07,
      "grad_norm": 42.25,
      "learning_rate": 4.701378516025699e-05,
      "loss": 1.1057,
      "step": 77600
    },
    {
      "epoch": 0.07,
      "grad_norm": 127.5,
      "learning_rate": 4.700928608706607e-05,
      "loss": 1.1478,
      "step": 77700
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.375,
      "learning_rate": 4.7004787013875144e-05,
      "loss": 0.964,
      "step": 77800
    },
    {
      "epoch": 0.07,
      "grad_norm": 213.0,
      "learning_rate": 4.7000287940684226e-05,
      "loss": 0.9888,
      "step": 77900
    },
    {
      "epoch": 0.07,
      "grad_norm": 41.5,
      "learning_rate": 4.69957888674933e-05,
      "loss": 1.0407,
      "step": 78000
    },
    {
      "epoch": 0.07,
      "grad_norm": 124.0,
      "learning_rate": 4.6991289794302377e-05,
      "loss": 0.8326,
      "step": 78100
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.0,
      "learning_rate": 4.698679072111145e-05,
      "loss": 1.1145,
      "step": 78200
    },
    {
      "epoch": 0.07,
      "grad_norm": 42.5,
      "learning_rate": 4.698229164792053e-05,
      "loss": 1.0665,
      "step": 78300
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.875,
      "learning_rate": 4.697779257472961e-05,
      "loss": 1.0049,
      "step": 78400
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.25,
      "learning_rate": 4.6973293501538685e-05,
      "loss": 1.2033,
      "step": 78500
    },
    {
      "epoch": 0.07,
      "grad_norm": 46.25,
      "learning_rate": 4.696879442834776e-05,
      "loss": 1.0287,
      "step": 78600
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.875,
      "learning_rate": 4.696429535515684e-05,
      "loss": 1.0317,
      "step": 78700
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.162109375,
      "learning_rate": 4.695979628196592e-05,
      "loss": 1.1977,
      "step": 78800
    },
    {
      "epoch": 0.07,
      "grad_norm": 36.0,
      "learning_rate": 4.695529720877499e-05,
      "loss": 1.0422,
      "step": 78900
    },
    {
      "epoch": 0.07,
      "grad_norm": 44.5,
      "learning_rate": 4.6950798135584075e-05,
      "loss": 0.9911,
      "step": 79000
    },
    {
      "epoch": 0.07,
      "grad_norm": 25.0,
      "learning_rate": 4.694629906239315e-05,
      "loss": 1.2047,
      "step": 79100
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.75,
      "learning_rate": 4.6941799989202226e-05,
      "loss": 1.0781,
      "step": 79200
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.875,
      "learning_rate": 4.693730091601131e-05,
      "loss": 0.9866,
      "step": 79300
    },
    {
      "epoch": 0.07,
      "grad_norm": 32.25,
      "learning_rate": 4.6932801842820376e-05,
      "loss": 1.089,
      "step": 79400
    },
    {
      "epoch": 0.07,
      "grad_norm": 29.125,
      "learning_rate": 4.692830276962946e-05,
      "loss": 0.9768,
      "step": 79500
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.4375,
      "learning_rate": 4.6923803696438534e-05,
      "loss": 0.9505,
      "step": 79600
    },
    {
      "epoch": 0.07,
      "grad_norm": 61.0,
      "learning_rate": 4.691930462324761e-05,
      "loss": 1.0787,
      "step": 79700
    },
    {
      "epoch": 0.07,
      "grad_norm": 46.0,
      "learning_rate": 4.691480555005669e-05,
      "loss": 1.0963,
      "step": 79800
    },
    {
      "epoch": 0.07,
      "grad_norm": 32.75,
      "learning_rate": 4.6910306476865767e-05,
      "loss": 1.1343,
      "step": 79900
    },
    {
      "epoch": 0.07,
      "grad_norm": 29.5,
      "learning_rate": 4.690580740367484e-05,
      "loss": 1.0554,
      "step": 80000
    },
    {
      "epoch": 0.07,
      "grad_norm": 87.0,
      "learning_rate": 4.6901308330483924e-05,
      "loss": 0.9815,
      "step": 80100
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.625,
      "learning_rate": 4.6896809257293e-05,
      "loss": 1.0009,
      "step": 80200
    },
    {
      "epoch": 0.07,
      "grad_norm": 87.5,
      "learning_rate": 4.689231018410208e-05,
      "loss": 1.0204,
      "step": 80300
    },
    {
      "epoch": 0.07,
      "grad_norm": 63.75,
      "learning_rate": 4.688781111091116e-05,
      "loss": 1.0705,
      "step": 80400
    },
    {
      "epoch": 0.07,
      "grad_norm": 30.875,
      "learning_rate": 4.688331203772023e-05,
      "loss": 1.0482,
      "step": 80500
    },
    {
      "epoch": 0.07,
      "grad_norm": 19.0,
      "learning_rate": 4.6878812964529314e-05,
      "loss": 1.0991,
      "step": 80600
    },
    {
      "epoch": 0.07,
      "grad_norm": 141.0,
      "learning_rate": 4.687431389133838e-05,
      "loss": 1.1017,
      "step": 80700
    },
    {
      "epoch": 0.07,
      "grad_norm": 240.0,
      "learning_rate": 4.6869814818147465e-05,
      "loss": 1.045,
      "step": 80800
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.875,
      "learning_rate": 4.686531574495654e-05,
      "loss": 1.1315,
      "step": 80900
    },
    {
      "epoch": 0.07,
      "grad_norm": 39.5,
      "learning_rate": 4.6860816671765616e-05,
      "loss": 1.0082,
      "step": 81000
    },
    {
      "epoch": 0.07,
      "grad_norm": 68.5,
      "learning_rate": 4.68563175985747e-05,
      "loss": 1.0634,
      "step": 81100
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.625,
      "learning_rate": 4.685181852538377e-05,
      "loss": 1.0875,
      "step": 81200
    },
    {
      "epoch": 0.07,
      "grad_norm": 77.5,
      "learning_rate": 4.684731945219285e-05,
      "loss": 0.9573,
      "step": 81300
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1123046875,
      "learning_rate": 4.684282037900193e-05,
      "loss": 0.8969,
      "step": 81400
    },
    {
      "epoch": 0.07,
      "grad_norm": 86.5,
      "learning_rate": 4.6838321305811006e-05,
      "loss": 1.0937,
      "step": 81500
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.625,
      "learning_rate": 4.683382223262008e-05,
      "loss": 0.9974,
      "step": 81600
    },
    {
      "epoch": 0.07,
      "grad_norm": 180.0,
      "learning_rate": 4.682932315942916e-05,
      "loss": 0.9794,
      "step": 81700
    },
    {
      "epoch": 0.07,
      "grad_norm": 64.5,
      "learning_rate": 4.682482408623824e-05,
      "loss": 1.0085,
      "step": 81800
    },
    {
      "epoch": 0.07,
      "grad_norm": 42.75,
      "learning_rate": 4.6820325013047314e-05,
      "loss": 1.104,
      "step": 81900
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.0,
      "learning_rate": 4.681582593985639e-05,
      "loss": 0.9597,
      "step": 82000
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.3125,
      "learning_rate": 4.6811326866665465e-05,
      "loss": 1.0694,
      "step": 82100
    },
    {
      "epoch": 0.07,
      "grad_norm": 19.25,
      "learning_rate": 4.680682779347455e-05,
      "loss": 1.1842,
      "step": 82200
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.5,
      "learning_rate": 4.680232872028362e-05,
      "loss": 1.1325,
      "step": 82300
    },
    {
      "epoch": 0.07,
      "grad_norm": 19.5,
      "learning_rate": 4.67978296470927e-05,
      "loss": 1.1123,
      "step": 82400
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.75,
      "learning_rate": 4.679333057390178e-05,
      "loss": 1.1663,
      "step": 82500
    },
    {
      "epoch": 0.07,
      "grad_norm": 39.25,
      "learning_rate": 4.6788831500710855e-05,
      "loss": 1.1589,
      "step": 82600
    },
    {
      "epoch": 0.07,
      "grad_norm": 79.5,
      "learning_rate": 4.678433242751993e-05,
      "loss": 1.0633,
      "step": 82700
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.5,
      "learning_rate": 4.677983335432901e-05,
      "loss": 1.0339,
      "step": 82800
    },
    {
      "epoch": 0.07,
      "grad_norm": 46.0,
      "learning_rate": 4.677533428113809e-05,
      "loss": 1.1827,
      "step": 82900
    },
    {
      "epoch": 0.07,
      "grad_norm": 124.0,
      "learning_rate": 4.677083520794717e-05,
      "loss": 1.0051,
      "step": 83000
    },
    {
      "epoch": 0.07,
      "grad_norm": 35.75,
      "learning_rate": 4.6766336134756245e-05,
      "loss": 1.028,
      "step": 83100
    },
    {
      "epoch": 0.07,
      "grad_norm": 75.0,
      "learning_rate": 4.676183706156532e-05,
      "loss": 0.923,
      "step": 83200
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.875,
      "learning_rate": 4.6757337988374396e-05,
      "loss": 0.9642,
      "step": 83300
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.875,
      "learning_rate": 4.675283891518347e-05,
      "loss": 1.138,
      "step": 83400
    },
    {
      "epoch": 0.07,
      "grad_norm": 42.5,
      "learning_rate": 4.674833984199255e-05,
      "loss": 1.2567,
      "step": 83500
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.625,
      "learning_rate": 4.674384076880163e-05,
      "loss": 1.0483,
      "step": 83600
    },
    {
      "epoch": 0.07,
      "grad_norm": 86.5,
      "learning_rate": 4.6739341695610704e-05,
      "loss": 1.2013,
      "step": 83700
    },
    {
      "epoch": 0.07,
      "grad_norm": 596.0,
      "learning_rate": 4.6734842622419786e-05,
      "loss": 1.0766,
      "step": 83800
    },
    {
      "epoch": 0.07,
      "grad_norm": 35.75,
      "learning_rate": 4.673034354922886e-05,
      "loss": 0.9455,
      "step": 83900
    },
    {
      "epoch": 0.07,
      "grad_norm": 54.0,
      "learning_rate": 4.672584447603794e-05,
      "loss": 1.0611,
      "step": 84000
    },
    {
      "epoch": 0.07,
      "grad_norm": 157.0,
      "learning_rate": 4.672134540284702e-05,
      "loss": 1.1654,
      "step": 84100
    },
    {
      "epoch": 0.08,
      "grad_norm": 59.25,
      "learning_rate": 4.6716846329656094e-05,
      "loss": 1.039,
      "step": 84200
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.375,
      "learning_rate": 4.671234725646517e-05,
      "loss": 0.9963,
      "step": 84300
    },
    {
      "epoch": 0.08,
      "grad_norm": 66.5,
      "learning_rate": 4.670784818327425e-05,
      "loss": 1.0757,
      "step": 84400
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.875,
      "learning_rate": 4.670334911008333e-05,
      "loss": 0.9207,
      "step": 84500
    },
    {
      "epoch": 0.08,
      "grad_norm": 37.75,
      "learning_rate": 4.66988500368924e-05,
      "loss": 1.0711,
      "step": 84600
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.875,
      "learning_rate": 4.669435096370148e-05,
      "loss": 0.9419,
      "step": 84700
    },
    {
      "epoch": 0.08,
      "grad_norm": 108.5,
      "learning_rate": 4.668985189051055e-05,
      "loss": 1.0834,
      "step": 84800
    },
    {
      "epoch": 0.08,
      "grad_norm": 288.0,
      "learning_rate": 4.6685352817319635e-05,
      "loss": 1.0528,
      "step": 84900
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.625,
      "learning_rate": 4.668085374412871e-05,
      "loss": 0.8459,
      "step": 85000
    },
    {
      "epoch": 0.08,
      "grad_norm": 110.5,
      "learning_rate": 4.6676354670937786e-05,
      "loss": 0.8702,
      "step": 85100
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.4375,
      "learning_rate": 4.667185559774687e-05,
      "loss": 0.9928,
      "step": 85200
    },
    {
      "epoch": 0.08,
      "grad_norm": 35.0,
      "learning_rate": 4.666735652455594e-05,
      "loss": 0.8058,
      "step": 85300
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.875,
      "learning_rate": 4.666285745136502e-05,
      "loss": 1.072,
      "step": 85400
    },
    {
      "epoch": 0.08,
      "grad_norm": 47.5,
      "learning_rate": 4.66583583781741e-05,
      "loss": 0.9577,
      "step": 85500
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.125,
      "learning_rate": 4.6653859304983176e-05,
      "loss": 1.1584,
      "step": 85600
    },
    {
      "epoch": 0.08,
      "grad_norm": 23.25,
      "learning_rate": 4.664936023179226e-05,
      "loss": 1.003,
      "step": 85700
    },
    {
      "epoch": 0.08,
      "grad_norm": 29.625,
      "learning_rate": 4.6644861158601334e-05,
      "loss": 1.0796,
      "step": 85800
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.00189971923828125,
      "learning_rate": 4.66403620854104e-05,
      "loss": 1.0614,
      "step": 85900
    },
    {
      "epoch": 0.08,
      "grad_norm": 126.0,
      "learning_rate": 4.6635863012219484e-05,
      "loss": 1.0196,
      "step": 86000
    },
    {
      "epoch": 0.08,
      "grad_norm": 70.0,
      "learning_rate": 4.663136393902856e-05,
      "loss": 0.992,
      "step": 86100
    },
    {
      "epoch": 0.08,
      "grad_norm": 26.5,
      "learning_rate": 4.662686486583764e-05,
      "loss": 1.0376,
      "step": 86200
    },
    {
      "epoch": 0.08,
      "grad_norm": 26.125,
      "learning_rate": 4.662236579264672e-05,
      "loss": 1.046,
      "step": 86300
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.125,
      "learning_rate": 4.661786671945579e-05,
      "loss": 0.9676,
      "step": 86400
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.15625,
      "learning_rate": 4.6613367646264875e-05,
      "loss": 0.9605,
      "step": 86500
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.0625,
      "learning_rate": 4.660886857307395e-05,
      "loss": 1.1733,
      "step": 86600
    },
    {
      "epoch": 0.08,
      "grad_norm": 51.0,
      "learning_rate": 4.6604369499883025e-05,
      "loss": 1.0272,
      "step": 86700
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.375,
      "learning_rate": 4.659987042669211e-05,
      "loss": 0.9423,
      "step": 86800
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.6875,
      "learning_rate": 4.659537135350118e-05,
      "loss": 1.0338,
      "step": 86900
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.4375,
      "learning_rate": 4.659087228031026e-05,
      "loss": 1.0813,
      "step": 87000
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.9375,
      "learning_rate": 4.658637320711934e-05,
      "loss": 1.0381,
      "step": 87100
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.75,
      "learning_rate": 4.658187413392841e-05,
      "loss": 0.9795,
      "step": 87200
    },
    {
      "epoch": 0.08,
      "grad_norm": 44.0,
      "learning_rate": 4.657737506073749e-05,
      "loss": 0.9689,
      "step": 87300
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.5,
      "learning_rate": 4.6572875987546566e-05,
      "loss": 0.9152,
      "step": 87400
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.875,
      "learning_rate": 4.656837691435564e-05,
      "loss": 1.0113,
      "step": 87500
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.75,
      "learning_rate": 4.6563877841164724e-05,
      "loss": 1.0188,
      "step": 87600
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.076171875,
      "learning_rate": 4.65593787679738e-05,
      "loss": 1.0329,
      "step": 87700
    },
    {
      "epoch": 0.08,
      "grad_norm": 105.5,
      "learning_rate": 4.6554879694782874e-05,
      "loss": 1.0492,
      "step": 87800
    },
    {
      "epoch": 0.08,
      "grad_norm": 69.0,
      "learning_rate": 4.6550380621591956e-05,
      "loss": 0.9479,
      "step": 87900
    },
    {
      "epoch": 0.08,
      "grad_norm": 245.0,
      "learning_rate": 4.654588154840103e-05,
      "loss": 0.9938,
      "step": 88000
    },
    {
      "epoch": 0.08,
      "grad_norm": 676.0,
      "learning_rate": 4.654138247521011e-05,
      "loss": 0.9638,
      "step": 88100
    },
    {
      "epoch": 0.08,
      "grad_norm": 39.5,
      "learning_rate": 4.653688340201919e-05,
      "loss": 1.0982,
      "step": 88200
    },
    {
      "epoch": 0.08,
      "grad_norm": 29.625,
      "learning_rate": 4.6532384328828265e-05,
      "loss": 1.0785,
      "step": 88300
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.875,
      "learning_rate": 4.652788525563735e-05,
      "loss": 1.0102,
      "step": 88400
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.034423828125,
      "learning_rate": 4.6523386182446415e-05,
      "loss": 1.088,
      "step": 88500
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.33203125,
      "learning_rate": 4.651888710925549e-05,
      "loss": 0.946,
      "step": 88600
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.125,
      "learning_rate": 4.651438803606457e-05,
      "loss": 1.0201,
      "step": 88700
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.3125,
      "learning_rate": 4.650988896287365e-05,
      "loss": 0.9087,
      "step": 88800
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.01336669921875,
      "learning_rate": 4.650538988968272e-05,
      "loss": 1.0322,
      "step": 88900
    },
    {
      "epoch": 0.08,
      "grad_norm": 19.875,
      "learning_rate": 4.6500890816491805e-05,
      "loss": 1.099,
      "step": 89000
    },
    {
      "epoch": 0.08,
      "grad_norm": 26.125,
      "learning_rate": 4.649639174330088e-05,
      "loss": 1.0903,
      "step": 89100
    },
    {
      "epoch": 0.08,
      "grad_norm": 62.0,
      "learning_rate": 4.649189267010996e-05,
      "loss": 1.1445,
      "step": 89200
    },
    {
      "epoch": 0.08,
      "grad_norm": 165.0,
      "learning_rate": 4.648739359691904e-05,
      "loss": 1.0607,
      "step": 89300
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.75,
      "learning_rate": 4.6482894523728114e-05,
      "loss": 1.1787,
      "step": 89400
    },
    {
      "epoch": 0.08,
      "grad_norm": 38.5,
      "learning_rate": 4.6478395450537196e-05,
      "loss": 0.973,
      "step": 89500
    },
    {
      "epoch": 0.08,
      "grad_norm": 53.25,
      "learning_rate": 4.647389637734627e-05,
      "loss": 0.9323,
      "step": 89600
    },
    {
      "epoch": 0.08,
      "grad_norm": 28.25,
      "learning_rate": 4.6469397304155346e-05,
      "loss": 1.1569,
      "step": 89700
    },
    {
      "epoch": 0.08,
      "grad_norm": 34.25,
      "learning_rate": 4.646489823096442e-05,
      "loss": 1.1027,
      "step": 89800
    },
    {
      "epoch": 0.08,
      "grad_norm": 49.0,
      "learning_rate": 4.64603991577735e-05,
      "loss": 1.2724,
      "step": 89900
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.75,
      "learning_rate": 4.645590008458258e-05,
      "loss": 1.0498,
      "step": 90000
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3203125,
      "learning_rate": 4.6451401011391655e-05,
      "loss": 0.9929,
      "step": 90100
    },
    {
      "epoch": 0.08,
      "grad_norm": 65.0,
      "learning_rate": 4.644690193820073e-05,
      "loss": 1.1277,
      "step": 90200
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.125,
      "learning_rate": 4.644240286500981e-05,
      "loss": 1.1095,
      "step": 90300
    },
    {
      "epoch": 0.08,
      "grad_norm": 28.375,
      "learning_rate": 4.643790379181889e-05,
      "loss": 1.1632,
      "step": 90400
    },
    {
      "epoch": 0.08,
      "grad_norm": 322.0,
      "learning_rate": 4.643340471862796e-05,
      "loss": 1.005,
      "step": 90500
    },
    {
      "epoch": 0.08,
      "grad_norm": 68.0,
      "learning_rate": 4.6428905645437045e-05,
      "loss": 1.0186,
      "step": 90600
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.625,
      "learning_rate": 4.642440657224612e-05,
      "loss": 1.1503,
      "step": 90700
    },
    {
      "epoch": 0.08,
      "grad_norm": 38.5,
      "learning_rate": 4.6419907499055195e-05,
      "loss": 1.1201,
      "step": 90800
    },
    {
      "epoch": 0.08,
      "grad_norm": 43.25,
      "learning_rate": 4.641540842586428e-05,
      "loss": 1.0718,
      "step": 90900
    },
    {
      "epoch": 0.08,
      "grad_norm": 206.0,
      "learning_rate": 4.641090935267335e-05,
      "loss": 1.1328,
      "step": 91000
    },
    {
      "epoch": 0.08,
      "grad_norm": 31.125,
      "learning_rate": 4.640641027948243e-05,
      "loss": 1.0125,
      "step": 91100
    },
    {
      "epoch": 0.08,
      "grad_norm": 59.25,
      "learning_rate": 4.6401911206291504e-05,
      "loss": 0.9647,
      "step": 91200
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.125,
      "learning_rate": 4.639741213310058e-05,
      "loss": 0.9768,
      "step": 91300
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.5,
      "learning_rate": 4.639291305990966e-05,
      "loss": 0.9173,
      "step": 91400
    },
    {
      "epoch": 0.08,
      "grad_norm": 37.25,
      "learning_rate": 4.6388413986718736e-05,
      "loss": 0.9748,
      "step": 91500
    },
    {
      "epoch": 0.08,
      "grad_norm": 70.0,
      "learning_rate": 4.638391491352781e-05,
      "loss": 0.9763,
      "step": 91600
    },
    {
      "epoch": 0.08,
      "grad_norm": 34.0,
      "learning_rate": 4.6379415840336894e-05,
      "loss": 1.0998,
      "step": 91700
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.8125,
      "learning_rate": 4.637491676714597e-05,
      "loss": 0.9437,
      "step": 91800
    },
    {
      "epoch": 0.08,
      "grad_norm": 740.0,
      "learning_rate": 4.637041769395505e-05,
      "loss": 1.094,
      "step": 91900
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.328125,
      "learning_rate": 4.636591862076413e-05,
      "loss": 1.0091,
      "step": 92000
    },
    {
      "epoch": 0.08,
      "grad_norm": 34.25,
      "learning_rate": 4.63614195475732e-05,
      "loss": 1.1075,
      "step": 92100
    },
    {
      "epoch": 0.08,
      "grad_norm": 163.0,
      "learning_rate": 4.6356920474382284e-05,
      "loss": 1.0266,
      "step": 92200
    },
    {
      "epoch": 0.08,
      "grad_norm": 55.5,
      "learning_rate": 4.635242140119136e-05,
      "loss": 0.9571,
      "step": 92300
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.875,
      "learning_rate": 4.6347922328000435e-05,
      "loss": 1.095,
      "step": 92400
    },
    {
      "epoch": 0.08,
      "grad_norm": 46.25,
      "learning_rate": 4.634342325480951e-05,
      "loss": 1.0834,
      "step": 92500
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.1875,
      "learning_rate": 4.6338924181618585e-05,
      "loss": 0.9692,
      "step": 92600
    },
    {
      "epoch": 0.08,
      "grad_norm": 38.0,
      "learning_rate": 4.633442510842767e-05,
      "loss": 1.1476,
      "step": 92700
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5625,
      "learning_rate": 4.632992603523674e-05,
      "loss": 0.973,
      "step": 92800
    },
    {
      "epoch": 0.08,
      "grad_norm": 164.0,
      "learning_rate": 4.632542696204582e-05,
      "loss": 0.9548,
      "step": 92900
    },
    {
      "epoch": 0.08,
      "grad_norm": 51.5,
      "learning_rate": 4.63209278888549e-05,
      "loss": 1.0253,
      "step": 93000
    },
    {
      "epoch": 0.08,
      "grad_norm": 19.75,
      "learning_rate": 4.6316428815663976e-05,
      "loss": 1.0976,
      "step": 93100
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.875,
      "learning_rate": 4.631192974247305e-05,
      "loss": 0.9796,
      "step": 93200
    },
    {
      "epoch": 0.08,
      "grad_norm": 872.0,
      "learning_rate": 4.630743066928213e-05,
      "loss": 0.8708,
      "step": 93300
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.375,
      "learning_rate": 4.630293159609121e-05,
      "loss": 1.0497,
      "step": 93400
    },
    {
      "epoch": 0.08,
      "grad_norm": 34.5,
      "learning_rate": 4.6298432522900284e-05,
      "loss": 1.1437,
      "step": 93500
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.625,
      "learning_rate": 4.6293933449709366e-05,
      "loss": 0.8803,
      "step": 93600
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2265625,
      "learning_rate": 4.6289434376518435e-05,
      "loss": 0.9964,
      "step": 93700
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.75,
      "learning_rate": 4.628493530332752e-05,
      "loss": 0.9963,
      "step": 93800
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.0164794921875,
      "learning_rate": 4.628043623013659e-05,
      "loss": 1.0988,
      "step": 93900
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.875,
      "learning_rate": 4.627593715694567e-05,
      "loss": 1.0421,
      "step": 94000
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.875,
      "learning_rate": 4.627143808375475e-05,
      "loss": 0.9977,
      "step": 94100
    },
    {
      "epoch": 0.08,
      "grad_norm": 32.0,
      "learning_rate": 4.6266939010563825e-05,
      "loss": 0.9836,
      "step": 94200
    },
    {
      "epoch": 0.08,
      "grad_norm": 169.0,
      "learning_rate": 4.62624399373729e-05,
      "loss": 1.0128,
      "step": 94300
    },
    {
      "epoch": 0.08,
      "grad_norm": 86.5,
      "learning_rate": 4.625794086418198e-05,
      "loss": 0.952,
      "step": 94400
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.62109375,
      "learning_rate": 4.625344179099106e-05,
      "loss": 1.1159,
      "step": 94500
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.126953125,
      "learning_rate": 4.624894271780014e-05,
      "loss": 1.0818,
      "step": 94600
    },
    {
      "epoch": 0.08,
      "grad_norm": 78.0,
      "learning_rate": 4.6244443644609215e-05,
      "loss": 1.0271,
      "step": 94700
    },
    {
      "epoch": 0.08,
      "grad_norm": 66.5,
      "learning_rate": 4.623994457141829e-05,
      "loss": 1.0018,
      "step": 94800
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.75,
      "learning_rate": 4.623544549822737e-05,
      "loss": 0.9924,
      "step": 94900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.0023193359375,
      "learning_rate": 4.623094642503644e-05,
      "loss": 0.9408,
      "step": 95000
    },
    {
      "epoch": 0.08,
      "grad_norm": 35.5,
      "learning_rate": 4.622644735184552e-05,
      "loss": 1.2318,
      "step": 95100
    },
    {
      "epoch": 0.08,
      "grad_norm": 53.0,
      "learning_rate": 4.62219482786546e-05,
      "loss": 0.9541,
      "step": 95200
    },
    {
      "epoch": 0.08,
      "grad_norm": 108.5,
      "learning_rate": 4.6217449205463674e-05,
      "loss": 1.0628,
      "step": 95300
    },
    {
      "epoch": 0.08,
      "grad_norm": 159.0,
      "learning_rate": 4.6212950132272756e-05,
      "loss": 1.0599,
      "step": 95400
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.5,
      "learning_rate": 4.620845105908183e-05,
      "loss": 1.1268,
      "step": 95500
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.375,
      "learning_rate": 4.620395198589091e-05,
      "loss": 1.0284,
      "step": 95600
    },
    {
      "epoch": 0.09,
      "grad_norm": 35.25,
      "learning_rate": 4.619945291269999e-05,
      "loss": 1.0674,
      "step": 95700
    },
    {
      "epoch": 0.09,
      "grad_norm": 51.75,
      "learning_rate": 4.6194953839509064e-05,
      "loss": 1.1138,
      "step": 95800
    },
    {
      "epoch": 0.09,
      "grad_norm": 41.5,
      "learning_rate": 4.619045476631814e-05,
      "loss": 1.1292,
      "step": 95900
    },
    {
      "epoch": 0.09,
      "grad_norm": 40.5,
      "learning_rate": 4.618595569312722e-05,
      "loss": 1.0113,
      "step": 96000
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.5,
      "learning_rate": 4.61814566199363e-05,
      "loss": 1.1243,
      "step": 96100
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.1875,
      "learning_rate": 4.617695754674537e-05,
      "loss": 0.9991,
      "step": 96200
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.25,
      "learning_rate": 4.617245847355445e-05,
      "loss": 1.0575,
      "step": 96300
    },
    {
      "epoch": 0.09,
      "grad_norm": 330.0,
      "learning_rate": 4.616795940036352e-05,
      "loss": 1.0142,
      "step": 96400
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.75,
      "learning_rate": 4.6163460327172605e-05,
      "loss": 1.1727,
      "step": 96500
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.75,
      "learning_rate": 4.615896125398168e-05,
      "loss": 1.1062,
      "step": 96600
    },
    {
      "epoch": 0.09,
      "grad_norm": 117.0,
      "learning_rate": 4.6154462180790756e-05,
      "loss": 0.9946,
      "step": 96700
    },
    {
      "epoch": 0.09,
      "grad_norm": 41.5,
      "learning_rate": 4.614996310759984e-05,
      "loss": 1.0037,
      "step": 96800
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.294921875,
      "learning_rate": 4.614546403440891e-05,
      "loss": 0.964,
      "step": 96900
    },
    {
      "epoch": 0.09,
      "grad_norm": 660.0,
      "learning_rate": 4.614096496121799e-05,
      "loss": 1.0423,
      "step": 97000
    },
    {
      "epoch": 0.09,
      "grad_norm": 29.5,
      "learning_rate": 4.613646588802707e-05,
      "loss": 1.0066,
      "step": 97100
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.0,
      "learning_rate": 4.6131966814836146e-05,
      "loss": 0.9157,
      "step": 97200
    },
    {
      "epoch": 0.09,
      "grad_norm": 94.5,
      "learning_rate": 4.612746774164523e-05,
      "loss": 1.0699,
      "step": 97300
    },
    {
      "epoch": 0.09,
      "grad_norm": 103.5,
      "learning_rate": 4.61229686684543e-05,
      "loss": 1.1165,
      "step": 97400
    },
    {
      "epoch": 0.09,
      "grad_norm": 31.375,
      "learning_rate": 4.611846959526338e-05,
      "loss": 1.0617,
      "step": 97500
    },
    {
      "epoch": 0.09,
      "grad_norm": 159.0,
      "learning_rate": 4.6113970522072454e-05,
      "loss": 1.0553,
      "step": 97600
    },
    {
      "epoch": 0.09,
      "grad_norm": 137.0,
      "learning_rate": 4.610947144888153e-05,
      "loss": 1.0892,
      "step": 97700
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.875,
      "learning_rate": 4.610497237569061e-05,
      "loss": 0.8908,
      "step": 97800
    },
    {
      "epoch": 0.09,
      "grad_norm": 66.0,
      "learning_rate": 4.610047330249969e-05,
      "loss": 0.9143,
      "step": 97900
    },
    {
      "epoch": 0.09,
      "grad_norm": 79.5,
      "learning_rate": 4.609597422930876e-05,
      "loss": 1.019,
      "step": 98000
    },
    {
      "epoch": 0.09,
      "grad_norm": 23.75,
      "learning_rate": 4.6091475156117844e-05,
      "loss": 1.1294,
      "step": 98100
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.921875,
      "learning_rate": 4.608697608292692e-05,
      "loss": 1.0638,
      "step": 98200
    },
    {
      "epoch": 0.09,
      "grad_norm": 40.0,
      "learning_rate": 4.6082477009735995e-05,
      "loss": 0.9934,
      "step": 98300
    },
    {
      "epoch": 0.09,
      "grad_norm": 153.0,
      "learning_rate": 4.607797793654508e-05,
      "loss": 1.064,
      "step": 98400
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.1875,
      "learning_rate": 4.607347886335415e-05,
      "loss": 1.2169,
      "step": 98500
    },
    {
      "epoch": 0.09,
      "grad_norm": 126.5,
      "learning_rate": 4.606897979016323e-05,
      "loss": 0.9162,
      "step": 98600
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.875,
      "learning_rate": 4.606448071697231e-05,
      "loss": 1.0007,
      "step": 98700
    },
    {
      "epoch": 0.09,
      "grad_norm": 36.25,
      "learning_rate": 4.6059981643781385e-05,
      "loss": 0.9393,
      "step": 98800
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.5,
      "learning_rate": 4.605548257059046e-05,
      "loss": 1.1701,
      "step": 98900
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.0,
      "learning_rate": 4.6050983497399536e-05,
      "loss": 1.1328,
      "step": 99000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.0191650390625,
      "learning_rate": 4.604648442420861e-05,
      "loss": 1.0642,
      "step": 99100
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.375,
      "learning_rate": 4.604198535101769e-05,
      "loss": 1.0403,
      "step": 99200
    },
    {
      "epoch": 0.09,
      "grad_norm": 46.5,
      "learning_rate": 4.603748627782677e-05,
      "loss": 1.1214,
      "step": 99300
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.375,
      "learning_rate": 4.6032987204635844e-05,
      "loss": 0.9587,
      "step": 99400
    },
    {
      "epoch": 0.09,
      "grad_norm": 113.5,
      "learning_rate": 4.6028488131444926e-05,
      "loss": 1.1008,
      "step": 99500
    },
    {
      "epoch": 0.09,
      "grad_norm": 122.0,
      "learning_rate": 4.6023989058254e-05,
      "loss": 0.9344,
      "step": 99600
    },
    {
      "epoch": 0.09,
      "grad_norm": 56.25,
      "learning_rate": 4.601948998506308e-05,
      "loss": 1.0311,
      "step": 99700
    },
    {
      "epoch": 0.09,
      "grad_norm": 1536.0,
      "learning_rate": 4.601499091187216e-05,
      "loss": 1.0442,
      "step": 99800
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.875,
      "learning_rate": 4.6010491838681234e-05,
      "loss": 1.1676,
      "step": 99900
    },
    {
      "epoch": 0.09,
      "grad_norm": 56.75,
      "learning_rate": 4.6005992765490316e-05,
      "loss": 1.1096,
      "step": 100000
    },
    {
      "epoch": 0.09,
      "grad_norm": 107.5,
      "learning_rate": 4.600149369229939e-05,
      "loss": 0.947,
      "step": 100100
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.25,
      "learning_rate": 4.599699461910846e-05,
      "loss": 1.0848,
      "step": 100200
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.8125,
      "learning_rate": 4.599249554591754e-05,
      "loss": 1.0569,
      "step": 100300
    },
    {
      "epoch": 0.09,
      "grad_norm": 48.75,
      "learning_rate": 4.598799647272662e-05,
      "loss": 1.0007,
      "step": 100400
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.5,
      "learning_rate": 4.59834973995357e-05,
      "loss": 1.1294,
      "step": 100500
    },
    {
      "epoch": 0.09,
      "grad_norm": 196.0,
      "learning_rate": 4.5978998326344775e-05,
      "loss": 1.1246,
      "step": 100600
    },
    {
      "epoch": 0.09,
      "grad_norm": 28.0,
      "learning_rate": 4.597449925315385e-05,
      "loss": 1.1526,
      "step": 100700
    },
    {
      "epoch": 0.09,
      "grad_norm": 50.5,
      "learning_rate": 4.597000017996293e-05,
      "loss": 1.0116,
      "step": 100800
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.71875,
      "learning_rate": 4.596550110677201e-05,
      "loss": 1.1103,
      "step": 100900
    },
    {
      "epoch": 0.09,
      "grad_norm": 29.75,
      "learning_rate": 4.596100203358108e-05,
      "loss": 1.0276,
      "step": 101000
    },
    {
      "epoch": 0.09,
      "grad_norm": 32.25,
      "learning_rate": 4.5956502960390165e-05,
      "loss": 1.1312,
      "step": 101100
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.25,
      "learning_rate": 4.595200388719924e-05,
      "loss": 0.8968,
      "step": 101200
    },
    {
      "epoch": 0.09,
      "grad_norm": 45.5,
      "learning_rate": 4.5947504814008316e-05,
      "loss": 0.9783,
      "step": 101300
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.625,
      "learning_rate": 4.59430057408174e-05,
      "loss": 1.1372,
      "step": 101400
    },
    {
      "epoch": 0.09,
      "grad_norm": 29.0,
      "learning_rate": 4.593850666762647e-05,
      "loss": 1.1434,
      "step": 101500
    },
    {
      "epoch": 0.09,
      "grad_norm": 31.0,
      "learning_rate": 4.593400759443555e-05,
      "loss": 0.9434,
      "step": 101600
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.5,
      "learning_rate": 4.5929508521244624e-05,
      "loss": 0.9486,
      "step": 101700
    },
    {
      "epoch": 0.09,
      "grad_norm": 116.5,
      "learning_rate": 4.59250094480537e-05,
      "loss": 1.0677,
      "step": 101800
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.25,
      "learning_rate": 4.592051037486278e-05,
      "loss": 0.8866,
      "step": 101900
    },
    {
      "epoch": 0.09,
      "grad_norm": 82.5,
      "learning_rate": 4.591601130167186e-05,
      "loss": 1.0288,
      "step": 102000
    },
    {
      "epoch": 0.09,
      "grad_norm": 39.75,
      "learning_rate": 4.591151222848093e-05,
      "loss": 1.1237,
      "step": 102100
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.375,
      "learning_rate": 4.5907013155290015e-05,
      "loss": 1.0126,
      "step": 102200
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.75,
      "learning_rate": 4.590251408209909e-05,
      "loss": 1.006,
      "step": 102300
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.482421875,
      "learning_rate": 4.5898015008908165e-05,
      "loss": 1.1302,
      "step": 102400
    },
    {
      "epoch": 0.09,
      "grad_norm": 49.75,
      "learning_rate": 4.589351593571725e-05,
      "loss": 1.0853,
      "step": 102500
    },
    {
      "epoch": 0.09,
      "grad_norm": 144.0,
      "learning_rate": 4.588901686252632e-05,
      "loss": 1.0041,
      "step": 102600
    },
    {
      "epoch": 0.09,
      "grad_norm": 376.0,
      "learning_rate": 4.5884517789335405e-05,
      "loss": 1.0391,
      "step": 102700
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.625,
      "learning_rate": 4.588001871614447e-05,
      "loss": 1.0192,
      "step": 102800
    },
    {
      "epoch": 0.09,
      "grad_norm": 31.375,
      "learning_rate": 4.587551964295355e-05,
      "loss": 1.0514,
      "step": 102900
    },
    {
      "epoch": 0.09,
      "grad_norm": 53.5,
      "learning_rate": 4.587102056976263e-05,
      "loss": 1.0099,
      "step": 103000
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0390625,
      "learning_rate": 4.5866521496571706e-05,
      "loss": 0.9165,
      "step": 103100
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.1875,
      "learning_rate": 4.586202242338079e-05,
      "loss": 0.9151,
      "step": 103200
    },
    {
      "epoch": 0.09,
      "grad_norm": 36.25,
      "learning_rate": 4.5857523350189864e-05,
      "loss": 1.1059,
      "step": 103300
    },
    {
      "epoch": 0.09,
      "grad_norm": 162.0,
      "learning_rate": 4.585302427699894e-05,
      "loss": 1.1004,
      "step": 103400
    },
    {
      "epoch": 0.09,
      "grad_norm": 22.75,
      "learning_rate": 4.584852520380802e-05,
      "loss": 0.9395,
      "step": 103500
    },
    {
      "epoch": 0.09,
      "grad_norm": 84.5,
      "learning_rate": 4.5844026130617096e-05,
      "loss": 1.159,
      "step": 103600
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.345703125,
      "learning_rate": 4.583952705742617e-05,
      "loss": 0.9854,
      "step": 103700
    },
    {
      "epoch": 0.09,
      "grad_norm": 43.0,
      "learning_rate": 4.5835027984235254e-05,
      "loss": 0.9976,
      "step": 103800
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.08544921875,
      "learning_rate": 4.583052891104433e-05,
      "loss": 1.0317,
      "step": 103900
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.000629425048828125,
      "learning_rate": 4.5826029837853405e-05,
      "loss": 1.125,
      "step": 104000
    },
    {
      "epoch": 0.09,
      "grad_norm": 101.0,
      "learning_rate": 4.582153076466248e-05,
      "loss": 1.0035,
      "step": 104100
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.5,
      "learning_rate": 4.5817031691471555e-05,
      "loss": 1.0678,
      "step": 104200
    },
    {
      "epoch": 0.09,
      "grad_norm": 146.0,
      "learning_rate": 4.581253261828064e-05,
      "loss": 1.1163,
      "step": 104300
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.125,
      "learning_rate": 4.580803354508971e-05,
      "loss": 0.9413,
      "step": 104400
    },
    {
      "epoch": 0.09,
      "grad_norm": 38.0,
      "learning_rate": 4.580353447189879e-05,
      "loss": 0.9658,
      "step": 104500
    },
    {
      "epoch": 0.09,
      "grad_norm": 82.5,
      "learning_rate": 4.579903539870787e-05,
      "loss": 1.1039,
      "step": 104600
    },
    {
      "epoch": 0.09,
      "grad_norm": 30.25,
      "learning_rate": 4.5794536325516945e-05,
      "loss": 1.0439,
      "step": 104700
    },
    {
      "epoch": 0.09,
      "grad_norm": 29.25,
      "learning_rate": 4.579003725232602e-05,
      "loss": 0.8397,
      "step": 104800
    },
    {
      "epoch": 0.09,
      "grad_norm": 58.5,
      "learning_rate": 4.57855381791351e-05,
      "loss": 1.0219,
      "step": 104900
    },
    {
      "epoch": 0.09,
      "grad_norm": 48.25,
      "learning_rate": 4.578103910594418e-05,
      "loss": 1.0695,
      "step": 105000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.462890625,
      "learning_rate": 4.5776540032753254e-05,
      "loss": 0.9501,
      "step": 105100
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.375,
      "learning_rate": 4.5772040959562336e-05,
      "loss": 1.044,
      "step": 105200
    },
    {
      "epoch": 0.09,
      "grad_norm": 41.25,
      "learning_rate": 4.576754188637141e-05,
      "loss": 1.0914,
      "step": 105300
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.005096435546875,
      "learning_rate": 4.5763042813180486e-05,
      "loss": 1.1638,
      "step": 105400
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.3125,
      "learning_rate": 4.575854373998956e-05,
      "loss": 1.0377,
      "step": 105500
    },
    {
      "epoch": 0.09,
      "grad_norm": 59.25,
      "learning_rate": 4.575404466679864e-05,
      "loss": 1.0802,
      "step": 105600
    },
    {
      "epoch": 0.09,
      "grad_norm": 41.75,
      "learning_rate": 4.574954559360772e-05,
      "loss": 1.099,
      "step": 105700
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.004180908203125,
      "learning_rate": 4.5745046520416795e-05,
      "loss": 1.1757,
      "step": 105800
    },
    {
      "epoch": 0.09,
      "grad_norm": 33.75,
      "learning_rate": 4.574054744722587e-05,
      "loss": 1.1173,
      "step": 105900
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.0,
      "learning_rate": 4.573604837403495e-05,
      "loss": 1.0574,
      "step": 106000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.004486083984375,
      "learning_rate": 4.573154930084403e-05,
      "loss": 1.1521,
      "step": 106100
    },
    {
      "epoch": 0.09,
      "grad_norm": 63.75,
      "learning_rate": 4.572705022765311e-05,
      "loss": 1.1711,
      "step": 106200
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.5,
      "learning_rate": 4.5722551154462185e-05,
      "loss": 1.042,
      "step": 106300
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.375,
      "learning_rate": 4.571805208127126e-05,
      "loss": 1.0231,
      "step": 106400
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.625,
      "learning_rate": 4.571355300808034e-05,
      "loss": 0.9212,
      "step": 106500
    },
    {
      "epoch": 0.09,
      "grad_norm": 39.5,
      "learning_rate": 4.570905393488942e-05,
      "loss": 0.9912,
      "step": 106600
    },
    {
      "epoch": 0.1,
      "grad_norm": 82.5,
      "learning_rate": 4.570455486169849e-05,
      "loss": 1.2595,
      "step": 106700
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.06787109375,
      "learning_rate": 4.570005578850757e-05,
      "loss": 0.9667,
      "step": 106800
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.75,
      "learning_rate": 4.5695556715316644e-05,
      "loss": 1.0125,
      "step": 106900
    },
    {
      "epoch": 0.1,
      "grad_norm": 276.0,
      "learning_rate": 4.5691057642125726e-05,
      "loss": 1.0792,
      "step": 107000
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.875,
      "learning_rate": 4.56865585689348e-05,
      "loss": 0.8907,
      "step": 107100
    },
    {
      "epoch": 0.1,
      "grad_norm": 33.5,
      "learning_rate": 4.5682059495743876e-05,
      "loss": 1.0903,
      "step": 107200
    },
    {
      "epoch": 0.1,
      "grad_norm": 50.5,
      "learning_rate": 4.567756042255296e-05,
      "loss": 1.1117,
      "step": 107300
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.5,
      "learning_rate": 4.5673061349362034e-05,
      "loss": 1.0159,
      "step": 107400
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.0,
      "learning_rate": 4.566856227617111e-05,
      "loss": 0.9291,
      "step": 107500
    },
    {
      "epoch": 0.1,
      "grad_norm": 58.25,
      "learning_rate": 4.566406320298019e-05,
      "loss": 1.0275,
      "step": 107600
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.25,
      "learning_rate": 4.565956412978927e-05,
      "loss": 1.0228,
      "step": 107700
    },
    {
      "epoch": 0.1,
      "grad_norm": 264.0,
      "learning_rate": 4.565506505659834e-05,
      "loss": 0.9558,
      "step": 107800
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.0,
      "learning_rate": 4.5650565983407424e-05,
      "loss": 1.0886,
      "step": 107900
    },
    {
      "epoch": 0.1,
      "grad_norm": 29.5,
      "learning_rate": 4.564606691021649e-05,
      "loss": 1.0162,
      "step": 108000
    },
    {
      "epoch": 0.1,
      "grad_norm": 141.0,
      "learning_rate": 4.5641567837025575e-05,
      "loss": 1.1004,
      "step": 108100
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.0,
      "learning_rate": 4.563706876383465e-05,
      "loss": 1.0725,
      "step": 108200
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.5,
      "learning_rate": 4.5632569690643725e-05,
      "loss": 1.0277,
      "step": 108300
    },
    {
      "epoch": 0.1,
      "grad_norm": 37.25,
      "learning_rate": 4.562807061745281e-05,
      "loss": 1.0413,
      "step": 108400
    },
    {
      "epoch": 0.1,
      "grad_norm": 92.5,
      "learning_rate": 4.562357154426188e-05,
      "loss": 1.1285,
      "step": 108500
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.83203125,
      "learning_rate": 4.561907247107096e-05,
      "loss": 0.9656,
      "step": 108600
    },
    {
      "epoch": 0.1,
      "grad_norm": 572.0,
      "learning_rate": 4.561457339788004e-05,
      "loss": 0.9537,
      "step": 108700
    },
    {
      "epoch": 0.1,
      "grad_norm": 78.0,
      "learning_rate": 4.5610074324689116e-05,
      "loss": 1.024,
      "step": 108800
    },
    {
      "epoch": 0.1,
      "grad_norm": 31.0,
      "learning_rate": 4.56055752514982e-05,
      "loss": 1.0087,
      "step": 108900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.0269775390625,
      "learning_rate": 4.560107617830727e-05,
      "loss": 1.0142,
      "step": 109000
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.125,
      "learning_rate": 4.559657710511635e-05,
      "loss": 1.0721,
      "step": 109100
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9609375,
      "learning_rate": 4.559207803192543e-05,
      "loss": 1.1147,
      "step": 109200
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.375,
      "learning_rate": 4.55875789587345e-05,
      "loss": 0.9822,
      "step": 109300
    },
    {
      "epoch": 0.1,
      "grad_norm": 46.75,
      "learning_rate": 4.558307988554358e-05,
      "loss": 1.0803,
      "step": 109400
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.94921875,
      "learning_rate": 4.557858081235266e-05,
      "loss": 1.0385,
      "step": 109500
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.125,
      "learning_rate": 4.557408173916173e-05,
      "loss": 1.0949,
      "step": 109600
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.75,
      "learning_rate": 4.5569582665970814e-05,
      "loss": 0.9582,
      "step": 109700
    },
    {
      "epoch": 0.1,
      "grad_norm": 30.25,
      "learning_rate": 4.556508359277989e-05,
      "loss": 1.0487,
      "step": 109800
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.625,
      "learning_rate": 4.5560584519588965e-05,
      "loss": 1.0011,
      "step": 109900
    },
    {
      "epoch": 0.1,
      "grad_norm": 66.0,
      "learning_rate": 4.555608544639805e-05,
      "loss": 1.0864,
      "step": 110000
    },
    {
      "epoch": 0.1,
      "grad_norm": 84.0,
      "learning_rate": 4.555158637320712e-05,
      "loss": 1.065,
      "step": 110100
    },
    {
      "epoch": 0.1,
      "grad_norm": 35.0,
      "learning_rate": 4.55470873000162e-05,
      "loss": 1.0923,
      "step": 110200
    },
    {
      "epoch": 0.1,
      "grad_norm": 482.0,
      "learning_rate": 4.554258822682528e-05,
      "loss": 1.094,
      "step": 110300
    },
    {
      "epoch": 0.1,
      "grad_norm": 25.0,
      "learning_rate": 4.5538089153634355e-05,
      "loss": 1.043,
      "step": 110400
    },
    {
      "epoch": 0.1,
      "grad_norm": 28.25,
      "learning_rate": 4.553359008044343e-05,
      "loss": 1.016,
      "step": 110500
    },
    {
      "epoch": 0.1,
      "grad_norm": 41.5,
      "learning_rate": 4.5529091007252506e-05,
      "loss": 0.9853,
      "step": 110600
    },
    {
      "epoch": 0.1,
      "grad_norm": 51.5,
      "learning_rate": 4.552459193406158e-05,
      "loss": 1.0946,
      "step": 110700
    },
    {
      "epoch": 0.1,
      "grad_norm": 78.5,
      "learning_rate": 4.552009286087066e-05,
      "loss": 0.9516,
      "step": 110800
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.5,
      "learning_rate": 4.551559378767974e-05,
      "loss": 1.0524,
      "step": 110900
    },
    {
      "epoch": 0.1,
      "grad_norm": 510.0,
      "learning_rate": 4.5511094714488814e-05,
      "loss": 1.0463,
      "step": 111000
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.25,
      "learning_rate": 4.5506595641297896e-05,
      "loss": 1.1404,
      "step": 111100
    },
    {
      "epoch": 0.1,
      "grad_norm": 92.0,
      "learning_rate": 4.550209656810697e-05,
      "loss": 1.035,
      "step": 111200
    },
    {
      "epoch": 0.1,
      "grad_norm": 95.5,
      "learning_rate": 4.549759749491605e-05,
      "loss": 1.0988,
      "step": 111300
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.8125,
      "learning_rate": 4.549309842172513e-05,
      "loss": 1.1427,
      "step": 111400
    },
    {
      "epoch": 0.1,
      "grad_norm": 37.25,
      "learning_rate": 4.5488599348534204e-05,
      "loss": 1.1043,
      "step": 111500
    },
    {
      "epoch": 0.1,
      "grad_norm": 83.5,
      "learning_rate": 4.5484100275343286e-05,
      "loss": 0.9516,
      "step": 111600
    },
    {
      "epoch": 0.1,
      "grad_norm": 42.25,
      "learning_rate": 4.547960120215236e-05,
      "loss": 1.0208,
      "step": 111700
    },
    {
      "epoch": 0.1,
      "grad_norm": 47.75,
      "learning_rate": 4.547510212896144e-05,
      "loss": 0.9251,
      "step": 111800
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.4375,
      "learning_rate": 4.547060305577051e-05,
      "loss": 1.0534,
      "step": 111900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.09033203125,
      "learning_rate": 4.546610398257959e-05,
      "loss": 1.1408,
      "step": 112000
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.25,
      "learning_rate": 4.546160490938867e-05,
      "loss": 1.0512,
      "step": 112100
    },
    {
      "epoch": 0.1,
      "grad_norm": 88.0,
      "learning_rate": 4.5457105836197745e-05,
      "loss": 0.9643,
      "step": 112200
    },
    {
      "epoch": 0.1,
      "grad_norm": 139.0,
      "learning_rate": 4.545260676300682e-05,
      "loss": 1.024,
      "step": 112300
    },
    {
      "epoch": 0.1,
      "grad_norm": 62.25,
      "learning_rate": 4.54481076898159e-05,
      "loss": 1.0729,
      "step": 112400
    },
    {
      "epoch": 0.1,
      "grad_norm": 63.75,
      "learning_rate": 4.544360861662498e-05,
      "loss": 1.0241,
      "step": 112500
    },
    {
      "epoch": 0.1,
      "grad_norm": 608.0,
      "learning_rate": 4.543910954343405e-05,
      "loss": 0.9974,
      "step": 112600
    },
    {
      "epoch": 0.1,
      "grad_norm": 28.625,
      "learning_rate": 4.5434610470243135e-05,
      "loss": 1.1885,
      "step": 112700
    },
    {
      "epoch": 0.1,
      "grad_norm": 47.75,
      "learning_rate": 4.543011139705221e-05,
      "loss": 1.0307,
      "step": 112800
    },
    {
      "epoch": 0.1,
      "grad_norm": 30.5,
      "learning_rate": 4.5425612323861286e-05,
      "loss": 1.0306,
      "step": 112900
    },
    {
      "epoch": 0.1,
      "grad_norm": 79.0,
      "learning_rate": 4.542111325067037e-05,
      "loss": 1.0288,
      "step": 113000
    },
    {
      "epoch": 0.1,
      "grad_norm": 77.5,
      "learning_rate": 4.5416614177479443e-05,
      "loss": 0.9431,
      "step": 113100
    },
    {
      "epoch": 0.1,
      "grad_norm": 142.0,
      "learning_rate": 4.541211510428852e-05,
      "loss": 1.0245,
      "step": 113200
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.0,
      "learning_rate": 4.5407616031097594e-05,
      "loss": 0.9562,
      "step": 113300
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.4375,
      "learning_rate": 4.540311695790667e-05,
      "loss": 0.8952,
      "step": 113400
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.34375,
      "learning_rate": 4.539861788471575e-05,
      "loss": 0.9978,
      "step": 113500
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.12060546875,
      "learning_rate": 4.539411881152483e-05,
      "loss": 1.0347,
      "step": 113600
    },
    {
      "epoch": 0.1,
      "grad_norm": 152.0,
      "learning_rate": 4.53896197383339e-05,
      "loss": 1.1466,
      "step": 113700
    },
    {
      "epoch": 0.1,
      "grad_norm": 23.0,
      "learning_rate": 4.5385120665142984e-05,
      "loss": 1.0597,
      "step": 113800
    },
    {
      "epoch": 0.1,
      "grad_norm": 61.25,
      "learning_rate": 4.538062159195206e-05,
      "loss": 1.0811,
      "step": 113900
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.0,
      "learning_rate": 4.5376122518761135e-05,
      "loss": 1.0961,
      "step": 114000
    },
    {
      "epoch": 0.1,
      "grad_norm": 67.0,
      "learning_rate": 4.537162344557022e-05,
      "loss": 1.1227,
      "step": 114100
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.189453125,
      "learning_rate": 4.536712437237929e-05,
      "loss": 1.038,
      "step": 114200
    },
    {
      "epoch": 0.1,
      "grad_norm": 26.625,
      "learning_rate": 4.5362625299188375e-05,
      "loss": 0.9473,
      "step": 114300
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.875,
      "learning_rate": 4.535812622599745e-05,
      "loss": 0.8994,
      "step": 114400
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.25,
      "learning_rate": 4.535362715280652e-05,
      "loss": 1.044,
      "step": 114500
    },
    {
      "epoch": 0.1,
      "grad_norm": 46.0,
      "learning_rate": 4.53491280796156e-05,
      "loss": 0.9068,
      "step": 114600
    },
    {
      "epoch": 0.1,
      "grad_norm": 94.5,
      "learning_rate": 4.5344629006424676e-05,
      "loss": 1.1091,
      "step": 114700
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.875,
      "learning_rate": 4.534012993323376e-05,
      "loss": 0.8964,
      "step": 114800
    },
    {
      "epoch": 0.1,
      "grad_norm": 31.75,
      "learning_rate": 4.5335630860042833e-05,
      "loss": 1.1737,
      "step": 114900
    },
    {
      "epoch": 0.1,
      "grad_norm": 98.5,
      "learning_rate": 4.533113178685191e-05,
      "loss": 1.0359,
      "step": 115000
    },
    {
      "epoch": 0.1,
      "grad_norm": 47.75,
      "learning_rate": 4.532663271366099e-05,
      "loss": 0.9983,
      "step": 115100
    },
    {
      "epoch": 0.1,
      "grad_norm": 40.5,
      "learning_rate": 4.5322133640470066e-05,
      "loss": 0.9794,
      "step": 115200
    },
    {
      "epoch": 0.1,
      "grad_norm": 28.0,
      "learning_rate": 4.531763456727914e-05,
      "loss": 1.087,
      "step": 115300
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.337890625,
      "learning_rate": 4.5313135494088224e-05,
      "loss": 0.9429,
      "step": 115400
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.1875,
      "learning_rate": 4.53086364208973e-05,
      "loss": 0.9066,
      "step": 115500
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.0751953125,
      "learning_rate": 4.5304137347706374e-05,
      "loss": 1.0203,
      "step": 115600
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.0,
      "learning_rate": 4.5299638274515456e-05,
      "loss": 1.0212,
      "step": 115700
    },
    {
      "epoch": 0.1,
      "grad_norm": 218.0,
      "learning_rate": 4.5295139201324525e-05,
      "loss": 0.9827,
      "step": 115800
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25,
      "learning_rate": 4.529064012813361e-05,
      "loss": 1.0948,
      "step": 115900
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.125,
      "learning_rate": 4.528614105494268e-05,
      "loss": 0.9621,
      "step": 116000
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.125,
      "learning_rate": 4.528164198175176e-05,
      "loss": 1.038,
      "step": 116100
    },
    {
      "epoch": 0.1,
      "grad_norm": 39.0,
      "learning_rate": 4.527714290856084e-05,
      "loss": 1.0215,
      "step": 116200
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.025634765625,
      "learning_rate": 4.5272643835369915e-05,
      "loss": 0.993,
      "step": 116300
    },
    {
      "epoch": 0.1,
      "grad_norm": 161.0,
      "learning_rate": 4.526814476217899e-05,
      "loss": 1.095,
      "step": 116400
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.875,
      "learning_rate": 4.526364568898807e-05,
      "loss": 1.049,
      "step": 116500
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.0068359375,
      "learning_rate": 4.525914661579715e-05,
      "loss": 0.9871,
      "step": 116600
    },
    {
      "epoch": 0.1,
      "grad_norm": 47.5,
      "learning_rate": 4.5254647542606223e-05,
      "loss": 0.9609,
      "step": 116700
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.75,
      "learning_rate": 4.5250148469415306e-05,
      "loss": 0.9806,
      "step": 116800
    },
    {
      "epoch": 0.1,
      "grad_norm": 24.5,
      "learning_rate": 4.524564939622438e-05,
      "loss": 0.9371,
      "step": 116900
    },
    {
      "epoch": 0.1,
      "grad_norm": 55.25,
      "learning_rate": 4.524115032303346e-05,
      "loss": 0.9856,
      "step": 117000
    },
    {
      "epoch": 0.1,
      "grad_norm": 89.5,
      "learning_rate": 4.523665124984253e-05,
      "loss": 1.149,
      "step": 117100
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.46875,
      "learning_rate": 4.523215217665161e-05,
      "loss": 1.1008,
      "step": 117200
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.15625,
      "learning_rate": 4.522765310346069e-05,
      "loss": 1.0506,
      "step": 117300
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.25,
      "learning_rate": 4.5223154030269764e-05,
      "loss": 1.0076,
      "step": 117400
    },
    {
      "epoch": 0.1,
      "grad_norm": 23.375,
      "learning_rate": 4.5218654957078846e-05,
      "loss": 1.0853,
      "step": 117500
    },
    {
      "epoch": 0.1,
      "grad_norm": 83.0,
      "learning_rate": 4.521415588388792e-05,
      "loss": 0.9788,
      "step": 117600
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.625,
      "learning_rate": 4.5209656810697e-05,
      "loss": 1.0742,
      "step": 117700
    },
    {
      "epoch": 0.1,
      "grad_norm": 46.0,
      "learning_rate": 4.520515773750608e-05,
      "loss": 1.0977,
      "step": 117800
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.306640625,
      "learning_rate": 4.5200658664315155e-05,
      "loss": 1.0214,
      "step": 117900
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.25,
      "learning_rate": 4.519615959112423e-05,
      "loss": 1.0425,
      "step": 118000
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.75,
      "learning_rate": 4.519166051793331e-05,
      "loss": 1.0495,
      "step": 118100
    },
    {
      "epoch": 0.11,
      "grad_norm": 28.375,
      "learning_rate": 4.518716144474239e-05,
      "loss": 1.0409,
      "step": 118200
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.375,
      "learning_rate": 4.518266237155146e-05,
      "loss": 1.0505,
      "step": 118300
    },
    {
      "epoch": 0.11,
      "grad_norm": 90.0,
      "learning_rate": 4.517816329836054e-05,
      "loss": 0.9198,
      "step": 118400
    },
    {
      "epoch": 0.11,
      "grad_norm": 35.75,
      "learning_rate": 4.5173664225169613e-05,
      "loss": 0.9992,
      "step": 118500
    },
    {
      "epoch": 0.11,
      "grad_norm": 243.0,
      "learning_rate": 4.5169165151978696e-05,
      "loss": 1.1532,
      "step": 118600
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.25,
      "learning_rate": 4.516466607878777e-05,
      "loss": 0.9649,
      "step": 118700
    },
    {
      "epoch": 0.11,
      "grad_norm": 101.0,
      "learning_rate": 4.5160167005596846e-05,
      "loss": 1.2006,
      "step": 118800
    },
    {
      "epoch": 0.11,
      "grad_norm": 75.5,
      "learning_rate": 4.515566793240593e-05,
      "loss": 0.9985,
      "step": 118900
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.64453125,
      "learning_rate": 4.5151168859215004e-05,
      "loss": 1.1021,
      "step": 119000
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.5625,
      "learning_rate": 4.514666978602408e-05,
      "loss": 0.9532,
      "step": 119100
    },
    {
      "epoch": 0.11,
      "grad_norm": 189.0,
      "learning_rate": 4.514217071283316e-05,
      "loss": 1.0827,
      "step": 119200
    },
    {
      "epoch": 0.11,
      "grad_norm": 109.0,
      "learning_rate": 4.5137671639642236e-05,
      "loss": 1.0307,
      "step": 119300
    },
    {
      "epoch": 0.11,
      "grad_norm": 31.625,
      "learning_rate": 4.513317256645131e-05,
      "loss": 1.1052,
      "step": 119400
    },
    {
      "epoch": 0.11,
      "grad_norm": 110.0,
      "learning_rate": 4.5128673493260394e-05,
      "loss": 1.0393,
      "step": 119500
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.5,
      "learning_rate": 4.512417442006947e-05,
      "loss": 0.981,
      "step": 119600
    },
    {
      "epoch": 0.11,
      "grad_norm": 37.75,
      "learning_rate": 4.5119675346878545e-05,
      "loss": 0.9903,
      "step": 119700
    },
    {
      "epoch": 0.11,
      "grad_norm": 149.0,
      "learning_rate": 4.511517627368762e-05,
      "loss": 1.0324,
      "step": 119800
    },
    {
      "epoch": 0.11,
      "grad_norm": 119.0,
      "learning_rate": 4.5110677200496695e-05,
      "loss": 1.0228,
      "step": 119900
    },
    {
      "epoch": 0.11,
      "grad_norm": 58.75,
      "learning_rate": 4.510617812730578e-05,
      "loss": 1.1095,
      "step": 120000
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.875,
      "learning_rate": 4.510167905411485e-05,
      "loss": 1.1055,
      "step": 120100
    },
    {
      "epoch": 0.11,
      "grad_norm": 36.75,
      "learning_rate": 4.5097179980923935e-05,
      "loss": 1.1821,
      "step": 120200
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.25,
      "learning_rate": 4.509268090773301e-05,
      "loss": 0.9721,
      "step": 120300
    },
    {
      "epoch": 0.11,
      "grad_norm": 22.375,
      "learning_rate": 4.5088181834542086e-05,
      "loss": 1.1529,
      "step": 120400
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.4375,
      "learning_rate": 4.508368276135117e-05,
      "loss": 1.1538,
      "step": 120500
    },
    {
      "epoch": 0.11,
      "grad_norm": 70.0,
      "learning_rate": 4.507918368816024e-05,
      "loss": 0.9579,
      "step": 120600
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.0625,
      "learning_rate": 4.507468461496932e-05,
      "loss": 0.8607,
      "step": 120700
    },
    {
      "epoch": 0.11,
      "grad_norm": 41.0,
      "learning_rate": 4.50701855417784e-05,
      "loss": 1.0996,
      "step": 120800
    },
    {
      "epoch": 0.11,
      "grad_norm": 93.0,
      "learning_rate": 4.5065686468587476e-05,
      "loss": 0.9022,
      "step": 120900
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.75,
      "learning_rate": 4.506118739539655e-05,
      "loss": 1.0507,
      "step": 121000
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.25,
      "learning_rate": 4.5056688322205626e-05,
      "loss": 1.0672,
      "step": 121100
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.125,
      "learning_rate": 4.50521892490147e-05,
      "loss": 1.0311,
      "step": 121200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.0078125,
      "learning_rate": 4.5047690175823784e-05,
      "loss": 1.0323,
      "step": 121300
    },
    {
      "epoch": 0.11,
      "grad_norm": 69.0,
      "learning_rate": 4.504319110263286e-05,
      "loss": 0.9827,
      "step": 121400
    },
    {
      "epoch": 0.11,
      "grad_norm": 117.5,
      "learning_rate": 4.5038692029441935e-05,
      "loss": 1.0411,
      "step": 121500
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.59375,
      "learning_rate": 4.503419295625102e-05,
      "loss": 1.1013,
      "step": 121600
    },
    {
      "epoch": 0.11,
      "grad_norm": 418.0,
      "learning_rate": 4.502969388306009e-05,
      "loss": 0.9148,
      "step": 121700
    },
    {
      "epoch": 0.11,
      "grad_norm": 40.0,
      "learning_rate": 4.502519480986917e-05,
      "loss": 1.0555,
      "step": 121800
    },
    {
      "epoch": 0.11,
      "grad_norm": 40.5,
      "learning_rate": 4.502069573667825e-05,
      "loss": 0.9873,
      "step": 121900
    },
    {
      "epoch": 0.11,
      "grad_norm": 75.0,
      "learning_rate": 4.5016196663487325e-05,
      "loss": 1.2124,
      "step": 122000
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.625,
      "learning_rate": 4.50116975902964e-05,
      "loss": 1.0172,
      "step": 122100
    },
    {
      "epoch": 0.11,
      "grad_norm": 47.0,
      "learning_rate": 4.500719851710548e-05,
      "loss": 1.0121,
      "step": 122200
    },
    {
      "epoch": 0.11,
      "grad_norm": 26.125,
      "learning_rate": 4.500269944391455e-05,
      "loss": 1.0633,
      "step": 122300
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.625,
      "learning_rate": 4.499820037072363e-05,
      "loss": 1.0916,
      "step": 122400
    },
    {
      "epoch": 0.11,
      "grad_norm": 90.5,
      "learning_rate": 4.499370129753271e-05,
      "loss": 0.9899,
      "step": 122500
    },
    {
      "epoch": 0.11,
      "grad_norm": 26.875,
      "learning_rate": 4.4989202224341784e-05,
      "loss": 0.9726,
      "step": 122600
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3671875,
      "learning_rate": 4.4984703151150866e-05,
      "loss": 0.9431,
      "step": 122700
    },
    {
      "epoch": 0.11,
      "grad_norm": 76.0,
      "learning_rate": 4.498020407795994e-05,
      "loss": 0.9709,
      "step": 122800
    },
    {
      "epoch": 0.11,
      "grad_norm": 37.5,
      "learning_rate": 4.4975705004769016e-05,
      "loss": 0.9988,
      "step": 122900
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.875,
      "learning_rate": 4.49712059315781e-05,
      "loss": 1.089,
      "step": 123000
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.75,
      "learning_rate": 4.4966706858387174e-05,
      "loss": 0.9941,
      "step": 123100
    },
    {
      "epoch": 0.11,
      "grad_norm": 62.5,
      "learning_rate": 4.4962207785196256e-05,
      "loss": 0.9371,
      "step": 123200
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.375,
      "learning_rate": 4.495770871200533e-05,
      "loss": 1.1702,
      "step": 123300
    },
    {
      "epoch": 0.11,
      "grad_norm": 106.5,
      "learning_rate": 4.495320963881441e-05,
      "loss": 0.9995,
      "step": 123400
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.375,
      "learning_rate": 4.494871056562349e-05,
      "loss": 1.1057,
      "step": 123500
    },
    {
      "epoch": 0.11,
      "grad_norm": 145.0,
      "learning_rate": 4.494421149243256e-05,
      "loss": 1.0401,
      "step": 123600
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.375,
      "learning_rate": 4.493971241924164e-05,
      "loss": 1.0329,
      "step": 123700
    },
    {
      "epoch": 0.11,
      "grad_norm": 42.25,
      "learning_rate": 4.4935213346050715e-05,
      "loss": 1.1075,
      "step": 123800
    },
    {
      "epoch": 0.11,
      "grad_norm": 155.0,
      "learning_rate": 4.493071427285979e-05,
      "loss": 1.1021,
      "step": 123900
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.875,
      "learning_rate": 4.492621519966887e-05,
      "loss": 1.048,
      "step": 124000
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6953125,
      "learning_rate": 4.492171612647795e-05,
      "loss": 0.9331,
      "step": 124100
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.625,
      "learning_rate": 4.491721705328702e-05,
      "loss": 1.1235,
      "step": 124200
    },
    {
      "epoch": 0.11,
      "grad_norm": 47.5,
      "learning_rate": 4.4912717980096105e-05,
      "loss": 1.2219,
      "step": 124300
    },
    {
      "epoch": 0.11,
      "grad_norm": 69.5,
      "learning_rate": 4.490821890690518e-05,
      "loss": 1.0295,
      "step": 124400
    },
    {
      "epoch": 0.11,
      "grad_norm": 38.25,
      "learning_rate": 4.4903719833714256e-05,
      "loss": 1.2938,
      "step": 124500
    },
    {
      "epoch": 0.11,
      "grad_norm": 35.5,
      "learning_rate": 4.489922076052334e-05,
      "loss": 1.0069,
      "step": 124600
    },
    {
      "epoch": 0.11,
      "grad_norm": 38.0,
      "learning_rate": 4.489472168733241e-05,
      "loss": 1.0296,
      "step": 124700
    },
    {
      "epoch": 0.11,
      "grad_norm": 72.5,
      "learning_rate": 4.489022261414149e-05,
      "loss": 0.9872,
      "step": 124800
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.09375,
      "learning_rate": 4.4885723540950564e-05,
      "loss": 1.1223,
      "step": 124900
    },
    {
      "epoch": 0.11,
      "grad_norm": 51.75,
      "learning_rate": 4.488122446775964e-05,
      "loss": 1.046,
      "step": 125000
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.875,
      "learning_rate": 4.487672539456872e-05,
      "loss": 1.1126,
      "step": 125100
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.0,
      "learning_rate": 4.48722263213778e-05,
      "loss": 1.1329,
      "step": 125200
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.125,
      "learning_rate": 4.486772724818687e-05,
      "loss": 1.0184,
      "step": 125300
    },
    {
      "epoch": 0.11,
      "grad_norm": 41.5,
      "learning_rate": 4.4863228174995954e-05,
      "loss": 1.0952,
      "step": 125400
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.009521484375,
      "learning_rate": 4.485872910180503e-05,
      "loss": 0.9959,
      "step": 125500
    },
    {
      "epoch": 0.11,
      "grad_norm": 39.25,
      "learning_rate": 4.4854230028614105e-05,
      "loss": 1.0238,
      "step": 125600
    },
    {
      "epoch": 0.11,
      "grad_norm": 40.25,
      "learning_rate": 4.484973095542319e-05,
      "loss": 0.8617,
      "step": 125700
    },
    {
      "epoch": 0.11,
      "grad_norm": 56.25,
      "learning_rate": 4.484523188223226e-05,
      "loss": 1.1847,
      "step": 125800
    },
    {
      "epoch": 0.11,
      "grad_norm": 120.0,
      "learning_rate": 4.4840732809041344e-05,
      "loss": 0.9826,
      "step": 125900
    },
    {
      "epoch": 0.11,
      "grad_norm": 40.0,
      "learning_rate": 4.483623373585042e-05,
      "loss": 1.1456,
      "step": 126000
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.005889892578125,
      "learning_rate": 4.4831734662659495e-05,
      "loss": 1.136,
      "step": 126100
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.024658203125,
      "learning_rate": 4.482723558946857e-05,
      "loss": 1.0936,
      "step": 126200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.036376953125,
      "learning_rate": 4.4822736516277646e-05,
      "loss": 0.9123,
      "step": 126300
    },
    {
      "epoch": 0.11,
      "grad_norm": 26.75,
      "learning_rate": 4.481823744308673e-05,
      "loss": 0.9552,
      "step": 126400
    },
    {
      "epoch": 0.11,
      "grad_norm": 34.25,
      "learning_rate": 4.48137383698958e-05,
      "loss": 1.0772,
      "step": 126500
    },
    {
      "epoch": 0.11,
      "grad_norm": 249.0,
      "learning_rate": 4.480923929670488e-05,
      "loss": 1.0825,
      "step": 126600
    },
    {
      "epoch": 0.11,
      "grad_norm": 44.0,
      "learning_rate": 4.480474022351396e-05,
      "loss": 0.9959,
      "step": 126700
    },
    {
      "epoch": 0.11,
      "grad_norm": 49.0,
      "learning_rate": 4.4800241150323036e-05,
      "loss": 0.9977,
      "step": 126800
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.375,
      "learning_rate": 4.479574207713211e-05,
      "loss": 1.0124,
      "step": 126900
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.375,
      "learning_rate": 4.4791243003941193e-05,
      "loss": 1.1152,
      "step": 127000
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.0201416015625,
      "learning_rate": 4.478674393075027e-05,
      "loss": 1.1607,
      "step": 127100
    },
    {
      "epoch": 0.11,
      "grad_norm": 39.0,
      "learning_rate": 4.4782244857559344e-05,
      "loss": 0.9655,
      "step": 127200
    },
    {
      "epoch": 0.11,
      "grad_norm": 34.25,
      "learning_rate": 4.4777745784368426e-05,
      "loss": 0.9306,
      "step": 127300
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.25,
      "learning_rate": 4.47732467111775e-05,
      "loss": 1.1897,
      "step": 127400
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.25,
      "learning_rate": 4.476874763798658e-05,
      "loss": 1.022,
      "step": 127500
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.5,
      "learning_rate": 4.476424856479565e-05,
      "loss": 1.1027,
      "step": 127600
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.125,
      "learning_rate": 4.475974949160473e-05,
      "loss": 0.9885,
      "step": 127700
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.625,
      "learning_rate": 4.475525041841381e-05,
      "loss": 0.9968,
      "step": 127800
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.77734375,
      "learning_rate": 4.4750751345222885e-05,
      "loss": 0.9143,
      "step": 127900
    },
    {
      "epoch": 0.11,
      "grad_norm": 36.0,
      "learning_rate": 4.474625227203196e-05,
      "loss": 0.9545,
      "step": 128000
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.875,
      "learning_rate": 4.474175319884104e-05,
      "loss": 0.9204,
      "step": 128100
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.5,
      "learning_rate": 4.473725412565012e-05,
      "loss": 1.2275,
      "step": 128200
    },
    {
      "epoch": 0.11,
      "grad_norm": 28.75,
      "learning_rate": 4.473275505245919e-05,
      "loss": 0.8557,
      "step": 128300
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.625,
      "learning_rate": 4.4728255979268275e-05,
      "loss": 1.0364,
      "step": 128400
    },
    {
      "epoch": 0.11,
      "grad_norm": 80.5,
      "learning_rate": 4.472375690607735e-05,
      "loss": 1.0122,
      "step": 128500
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.19921875,
      "learning_rate": 4.471925783288643e-05,
      "loss": 1.0095,
      "step": 128600
    },
    {
      "epoch": 0.11,
      "grad_norm": 93.0,
      "learning_rate": 4.471475875969551e-05,
      "loss": 1.1243,
      "step": 128700
    },
    {
      "epoch": 0.11,
      "grad_norm": 57.75,
      "learning_rate": 4.471025968650458e-05,
      "loss": 0.9377,
      "step": 128800
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.875,
      "learning_rate": 4.470576061331366e-05,
      "loss": 1.1993,
      "step": 128900
    },
    {
      "epoch": 0.11,
      "grad_norm": 30.5,
      "learning_rate": 4.4701261540122734e-05,
      "loss": 1.1627,
      "step": 129000
    },
    {
      "epoch": 0.12,
      "grad_norm": 28.625,
      "learning_rate": 4.4696762466931816e-05,
      "loss": 1.0471,
      "step": 129100
    },
    {
      "epoch": 0.12,
      "grad_norm": 37.75,
      "learning_rate": 4.469226339374089e-05,
      "loss": 1.0627,
      "step": 129200
    },
    {
      "epoch": 0.12,
      "grad_norm": 20.375,
      "learning_rate": 4.468776432054997e-05,
      "loss": 1.0888,
      "step": 129300
    },
    {
      "epoch": 0.12,
      "grad_norm": 88.0,
      "learning_rate": 4.468326524735905e-05,
      "loss": 1.1228,
      "step": 129400
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.875,
      "learning_rate": 4.4678766174168124e-05,
      "loss": 0.9295,
      "step": 129500
    },
    {
      "epoch": 0.12,
      "grad_norm": 35.75,
      "learning_rate": 4.46742671009772e-05,
      "loss": 0.9467,
      "step": 129600
    },
    {
      "epoch": 0.12,
      "grad_norm": 33.5,
      "learning_rate": 4.466976802778628e-05,
      "loss": 0.9815,
      "step": 129700
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.375,
      "learning_rate": 4.466526895459536e-05,
      "loss": 1.0445,
      "step": 129800
    },
    {
      "epoch": 0.12,
      "grad_norm": 50.25,
      "learning_rate": 4.466076988140443e-05,
      "loss": 1.0879,
      "step": 129900
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.0,
      "learning_rate": 4.4656270808213515e-05,
      "loss": 1.2399,
      "step": 130000
    },
    {
      "epoch": 0.12,
      "grad_norm": 34.25,
      "learning_rate": 4.465177173502258e-05,
      "loss": 0.9757,
      "step": 130100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.251953125,
      "learning_rate": 4.4647272661831665e-05,
      "loss": 1.097,
      "step": 130200
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.0625,
      "learning_rate": 4.464277358864074e-05,
      "loss": 1.0073,
      "step": 130300
    },
    {
      "epoch": 0.12,
      "grad_norm": 113.0,
      "learning_rate": 4.4638274515449816e-05,
      "loss": 0.9838,
      "step": 130400
    },
    {
      "epoch": 0.12,
      "grad_norm": 476.0,
      "learning_rate": 4.46337754422589e-05,
      "loss": 1.0932,
      "step": 130500
    },
    {
      "epoch": 0.12,
      "grad_norm": 65.5,
      "learning_rate": 4.4629276369067973e-05,
      "loss": 1.1927,
      "step": 130600
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6484375,
      "learning_rate": 4.462477729587705e-05,
      "loss": 0.9592,
      "step": 130700
    },
    {
      "epoch": 0.12,
      "grad_norm": 37.75,
      "learning_rate": 4.462027822268613e-05,
      "loss": 1.1063,
      "step": 130800
    },
    {
      "epoch": 0.12,
      "grad_norm": 53.75,
      "learning_rate": 4.4615779149495206e-05,
      "loss": 1.0479,
      "step": 130900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.125,
      "learning_rate": 4.461128007630428e-05,
      "loss": 0.9726,
      "step": 131000
    },
    {
      "epoch": 0.12,
      "grad_norm": 67.0,
      "learning_rate": 4.4606781003113364e-05,
      "loss": 0.9642,
      "step": 131100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.404296875,
      "learning_rate": 4.460228192992244e-05,
      "loss": 0.9941,
      "step": 131200
    },
    {
      "epoch": 0.12,
      "grad_norm": 46.0,
      "learning_rate": 4.459778285673152e-05,
      "loss": 1.0464,
      "step": 131300
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.5625,
      "learning_rate": 4.459328378354059e-05,
      "loss": 0.8999,
      "step": 131400
    },
    {
      "epoch": 0.12,
      "grad_norm": 28.875,
      "learning_rate": 4.4588784710349665e-05,
      "loss": 1.0474,
      "step": 131500
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.625,
      "learning_rate": 4.458428563715875e-05,
      "loss": 0.9166,
      "step": 131600
    },
    {
      "epoch": 0.12,
      "grad_norm": 94.5,
      "learning_rate": 4.457978656396782e-05,
      "loss": 1.0516,
      "step": 131700
    },
    {
      "epoch": 0.12,
      "grad_norm": 35.0,
      "learning_rate": 4.4575287490776905e-05,
      "loss": 1.0711,
      "step": 131800
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.0625,
      "learning_rate": 4.457078841758598e-05,
      "loss": 0.9893,
      "step": 131900
    },
    {
      "epoch": 0.12,
      "grad_norm": 39.5,
      "learning_rate": 4.4566289344395055e-05,
      "loss": 1.043,
      "step": 132000
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.00970458984375,
      "learning_rate": 4.456179027120414e-05,
      "loss": 1.0306,
      "step": 132100
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.8125,
      "learning_rate": 4.455729119801321e-05,
      "loss": 1.0423,
      "step": 132200
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.75,
      "learning_rate": 4.455279212482229e-05,
      "loss": 1.0196,
      "step": 132300
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.3125,
      "learning_rate": 4.454829305163137e-05,
      "loss": 0.9924,
      "step": 132400
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.25,
      "learning_rate": 4.4543793978440446e-05,
      "loss": 0.9755,
      "step": 132500
    },
    {
      "epoch": 0.12,
      "grad_norm": 35.5,
      "learning_rate": 4.453929490524952e-05,
      "loss": 1.0687,
      "step": 132600
    },
    {
      "epoch": 0.12,
      "grad_norm": 41.25,
      "learning_rate": 4.4534795832058596e-05,
      "loss": 1.056,
      "step": 132700
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.028076171875,
      "learning_rate": 4.453029675886767e-05,
      "loss": 1.0784,
      "step": 132800
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.375,
      "learning_rate": 4.4525797685676754e-05,
      "loss": 1.1376,
      "step": 132900
    },
    {
      "epoch": 0.12,
      "grad_norm": 81.5,
      "learning_rate": 4.452129861248583e-05,
      "loss": 1.0206,
      "step": 133000
    },
    {
      "epoch": 0.12,
      "grad_norm": 36.0,
      "learning_rate": 4.4516799539294904e-05,
      "loss": 0.9588,
      "step": 133100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.026123046875,
      "learning_rate": 4.4512300466103986e-05,
      "loss": 1.0697,
      "step": 133200
    },
    {
      "epoch": 0.12,
      "grad_norm": 80.0,
      "learning_rate": 4.450780139291306e-05,
      "loss": 1.0369,
      "step": 133300
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.125,
      "learning_rate": 4.450330231972214e-05,
      "loss": 1.0461,
      "step": 133400
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.6875,
      "learning_rate": 4.449880324653122e-05,
      "loss": 1.0091,
      "step": 133500
    },
    {
      "epoch": 0.12,
      "grad_norm": 38.25,
      "learning_rate": 4.4494304173340295e-05,
      "loss": 1.0225,
      "step": 133600
    },
    {
      "epoch": 0.12,
      "grad_norm": 70.5,
      "learning_rate": 4.448980510014937e-05,
      "loss": 1.0442,
      "step": 133700
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.578125,
      "learning_rate": 4.448530602695845e-05,
      "loss": 1.0823,
      "step": 133800
    },
    {
      "epoch": 0.12,
      "grad_norm": 42.75,
      "learning_rate": 4.448080695376753e-05,
      "loss": 1.0175,
      "step": 133900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.032470703125,
      "learning_rate": 4.44763078805766e-05,
      "loss": 1.0915,
      "step": 134000
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3671875,
      "learning_rate": 4.447180880738568e-05,
      "loss": 0.8835,
      "step": 134100
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.375,
      "learning_rate": 4.4467309734194753e-05,
      "loss": 0.9454,
      "step": 134200
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.3125,
      "learning_rate": 4.4462810661003836e-05,
      "loss": 1.2124,
      "step": 134300
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.25,
      "learning_rate": 4.445831158781291e-05,
      "loss": 0.9435,
      "step": 134400
    },
    {
      "epoch": 0.12,
      "grad_norm": 194.0,
      "learning_rate": 4.445381251462199e-05,
      "loss": 1.0023,
      "step": 134500
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.9375,
      "learning_rate": 4.444931344143107e-05,
      "loss": 1.1119,
      "step": 134600
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.125,
      "learning_rate": 4.4444814368240144e-05,
      "loss": 1.042,
      "step": 134700
    },
    {
      "epoch": 0.12,
      "grad_norm": 17.25,
      "learning_rate": 4.4440315295049226e-05,
      "loss": 0.9875,
      "step": 134800
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.13671875,
      "learning_rate": 4.44358162218583e-05,
      "loss": 1.0372,
      "step": 134900
    },
    {
      "epoch": 0.12,
      "grad_norm": 37.5,
      "learning_rate": 4.4431317148667376e-05,
      "loss": 0.8741,
      "step": 135000
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.375,
      "learning_rate": 4.442681807547646e-05,
      "loss": 0.9989,
      "step": 135100
    },
    {
      "epoch": 0.12,
      "grad_norm": 62.0,
      "learning_rate": 4.4422319002285534e-05,
      "loss": 1.0272,
      "step": 135200
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.625,
      "learning_rate": 4.441781992909461e-05,
      "loss": 0.9341,
      "step": 135300
    },
    {
      "epoch": 0.12,
      "grad_norm": 32.25,
      "learning_rate": 4.4413320855903685e-05,
      "loss": 0.9291,
      "step": 135400
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.25,
      "learning_rate": 4.440882178271276e-05,
      "loss": 0.9806,
      "step": 135500
    },
    {
      "epoch": 0.12,
      "grad_norm": 27.875,
      "learning_rate": 4.440432270952184e-05,
      "loss": 1.0683,
      "step": 135600
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.0166015625,
      "learning_rate": 4.439982363633092e-05,
      "loss": 0.9878,
      "step": 135700
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.25,
      "learning_rate": 4.439532456313999e-05,
      "loss": 1.116,
      "step": 135800
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.125,
      "learning_rate": 4.4390825489949075e-05,
      "loss": 1.0541,
      "step": 135900
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.0,
      "learning_rate": 4.438632641675815e-05,
      "loss": 1.0558,
      "step": 136000
    },
    {
      "epoch": 0.12,
      "grad_norm": 49.25,
      "learning_rate": 4.4381827343567226e-05,
      "loss": 0.8569,
      "step": 136100
    },
    {
      "epoch": 0.12,
      "grad_norm": 20.25,
      "learning_rate": 4.437732827037631e-05,
      "loss": 0.9822,
      "step": 136200
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.875,
      "learning_rate": 4.437282919718538e-05,
      "loss": 1.1,
      "step": 136300
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.3125,
      "learning_rate": 4.436833012399446e-05,
      "loss": 0.9849,
      "step": 136400
    },
    {
      "epoch": 0.12,
      "grad_norm": 43.0,
      "learning_rate": 4.436383105080354e-05,
      "loss": 0.9913,
      "step": 136500
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.0,
      "learning_rate": 4.435933197761261e-05,
      "loss": 0.9118,
      "step": 136600
    },
    {
      "epoch": 0.12,
      "grad_norm": 62.0,
      "learning_rate": 4.435483290442169e-05,
      "loss": 1.1478,
      "step": 136700
    },
    {
      "epoch": 0.12,
      "grad_norm": 52.25,
      "learning_rate": 4.4350333831230766e-05,
      "loss": 1.0215,
      "step": 136800
    },
    {
      "epoch": 0.12,
      "grad_norm": 33.5,
      "learning_rate": 4.434583475803984e-05,
      "loss": 1.0647,
      "step": 136900
    },
    {
      "epoch": 0.12,
      "grad_norm": 29.0,
      "learning_rate": 4.4341335684848924e-05,
      "loss": 0.9871,
      "step": 137000
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.000946044921875,
      "learning_rate": 4.4336836611658e-05,
      "loss": 1.0437,
      "step": 137100
    },
    {
      "epoch": 0.12,
      "grad_norm": 62.25,
      "learning_rate": 4.433233753846708e-05,
      "loss": 1.1933,
      "step": 137200
    },
    {
      "epoch": 0.12,
      "grad_norm": 36.5,
      "learning_rate": 4.432783846527616e-05,
      "loss": 1.081,
      "step": 137300
    },
    {
      "epoch": 0.12,
      "grad_norm": 49.5,
      "learning_rate": 4.432333939208523e-05,
      "loss": 1.0511,
      "step": 137400
    },
    {
      "epoch": 0.12,
      "grad_norm": 37.5,
      "learning_rate": 4.4318840318894314e-05,
      "loss": 1.0126,
      "step": 137500
    },
    {
      "epoch": 0.12,
      "grad_norm": 29.5,
      "learning_rate": 4.431434124570339e-05,
      "loss": 0.9972,
      "step": 137600
    },
    {
      "epoch": 0.12,
      "grad_norm": 32.25,
      "learning_rate": 4.4309842172512465e-05,
      "loss": 0.9747,
      "step": 137700
    },
    {
      "epoch": 0.12,
      "grad_norm": 100.0,
      "learning_rate": 4.430534309932155e-05,
      "loss": 0.9739,
      "step": 137800
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.5625,
      "learning_rate": 4.4300844026130616e-05,
      "loss": 0.9225,
      "step": 137900
    },
    {
      "epoch": 0.12,
      "grad_norm": 43.75,
      "learning_rate": 4.42963449529397e-05,
      "loss": 1.0158,
      "step": 138000
    },
    {
      "epoch": 0.12,
      "grad_norm": 17.375,
      "learning_rate": 4.429184587974877e-05,
      "loss": 0.9643,
      "step": 138100
    },
    {
      "epoch": 0.12,
      "grad_norm": 53.75,
      "learning_rate": 4.428734680655785e-05,
      "loss": 1.0333,
      "step": 138200
    },
    {
      "epoch": 0.12,
      "grad_norm": 916.0,
      "learning_rate": 4.428284773336693e-05,
      "loss": 0.8435,
      "step": 138300
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.9375,
      "learning_rate": 4.4278348660176006e-05,
      "loss": 1.1194,
      "step": 138400
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.625,
      "learning_rate": 4.427384958698508e-05,
      "loss": 1.0688,
      "step": 138500
    },
    {
      "epoch": 0.12,
      "grad_norm": 101.5,
      "learning_rate": 4.426935051379416e-05,
      "loss": 1.0035,
      "step": 138600
    },
    {
      "epoch": 0.12,
      "grad_norm": 34.75,
      "learning_rate": 4.426485144060324e-05,
      "loss": 1.2824,
      "step": 138700
    },
    {
      "epoch": 0.12,
      "grad_norm": 61.5,
      "learning_rate": 4.4260352367412314e-05,
      "loss": 1.0244,
      "step": 138800
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5,
      "learning_rate": 4.4255853294221396e-05,
      "loss": 1.0817,
      "step": 138900
    },
    {
      "epoch": 0.12,
      "grad_norm": 31.25,
      "learning_rate": 4.425135422103047e-05,
      "loss": 0.9969,
      "step": 139000
    },
    {
      "epoch": 0.12,
      "grad_norm": 32.5,
      "learning_rate": 4.424685514783955e-05,
      "loss": 0.9774,
      "step": 139100
    },
    {
      "epoch": 0.12,
      "grad_norm": 36.75,
      "learning_rate": 4.424235607464862e-05,
      "loss": 0.8757,
      "step": 139200
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.375,
      "learning_rate": 4.42378570014577e-05,
      "loss": 0.926,
      "step": 139300
    },
    {
      "epoch": 0.12,
      "grad_norm": 109.5,
      "learning_rate": 4.423335792826678e-05,
      "loss": 0.9832,
      "step": 139400
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.828125,
      "learning_rate": 4.4228858855075855e-05,
      "loss": 0.91,
      "step": 139500
    },
    {
      "epoch": 0.12,
      "grad_norm": 26.625,
      "learning_rate": 4.422435978188493e-05,
      "loss": 0.9774,
      "step": 139600
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.0057373046875,
      "learning_rate": 4.421986070869401e-05,
      "loss": 0.9223,
      "step": 139700
    },
    {
      "epoch": 0.12,
      "grad_norm": 95.5,
      "learning_rate": 4.421536163550309e-05,
      "loss": 0.9779,
      "step": 139800
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.010009765625,
      "learning_rate": 4.421086256231216e-05,
      "loss": 1.0977,
      "step": 139900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2373046875,
      "learning_rate": 4.4206363489121245e-05,
      "loss": 1.0138,
      "step": 140000
    },
    {
      "epoch": 0.12,
      "grad_norm": 47.5,
      "learning_rate": 4.420186441593032e-05,
      "loss": 1.1582,
      "step": 140100
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.25,
      "learning_rate": 4.41973653427394e-05,
      "loss": 0.9933,
      "step": 140200
    },
    {
      "epoch": 0.12,
      "grad_norm": 46.5,
      "learning_rate": 4.419286626954848e-05,
      "loss": 1.0703,
      "step": 140300
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.5,
      "learning_rate": 4.418836719635755e-05,
      "loss": 0.984,
      "step": 140400
    },
    {
      "epoch": 0.13,
      "grad_norm": 63.0,
      "learning_rate": 4.418386812316663e-05,
      "loss": 1.1195,
      "step": 140500
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.0,
      "learning_rate": 4.4179369049975704e-05,
      "loss": 1.0109,
      "step": 140600
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.4375,
      "learning_rate": 4.4174869976784786e-05,
      "loss": 1.2296,
      "step": 140700
    },
    {
      "epoch": 0.13,
      "grad_norm": 35.5,
      "learning_rate": 4.417037090359386e-05,
      "loss": 0.7742,
      "step": 140800
    },
    {
      "epoch": 0.13,
      "grad_norm": 55.75,
      "learning_rate": 4.416587183040294e-05,
      "loss": 0.9351,
      "step": 140900
    },
    {
      "epoch": 0.13,
      "grad_norm": 50.75,
      "learning_rate": 4.416137275721202e-05,
      "loss": 0.9272,
      "step": 141000
    },
    {
      "epoch": 0.13,
      "grad_norm": 32.0,
      "learning_rate": 4.4156873684021094e-05,
      "loss": 1.0042,
      "step": 141100
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.375,
      "learning_rate": 4.415237461083017e-05,
      "loss": 1.0413,
      "step": 141200
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.0,
      "learning_rate": 4.414787553763925e-05,
      "loss": 1.0101,
      "step": 141300
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.5,
      "learning_rate": 4.414337646444833e-05,
      "loss": 1.1085,
      "step": 141400
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.75,
      "learning_rate": 4.41388773912574e-05,
      "loss": 0.9453,
      "step": 141500
    },
    {
      "epoch": 0.13,
      "grad_norm": 117.5,
      "learning_rate": 4.4134378318066484e-05,
      "loss": 1.013,
      "step": 141600
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.5,
      "learning_rate": 4.412987924487556e-05,
      "loss": 1.1468,
      "step": 141700
    },
    {
      "epoch": 0.13,
      "grad_norm": 38.75,
      "learning_rate": 4.4125380171684635e-05,
      "loss": 0.9927,
      "step": 141800
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.8125,
      "learning_rate": 4.412088109849371e-05,
      "loss": 0.9451,
      "step": 141900
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.6875,
      "learning_rate": 4.4116382025302786e-05,
      "loss": 0.9307,
      "step": 142000
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.0625,
      "learning_rate": 4.411188295211187e-05,
      "loss": 1.0743,
      "step": 142100
    },
    {
      "epoch": 0.13,
      "grad_norm": 72.5,
      "learning_rate": 4.410738387892094e-05,
      "loss": 1.1525,
      "step": 142200
    },
    {
      "epoch": 0.13,
      "grad_norm": 21.75,
      "learning_rate": 4.410288480573002e-05,
      "loss": 1.0244,
      "step": 142300
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.25,
      "learning_rate": 4.40983857325391e-05,
      "loss": 1.0243,
      "step": 142400
    },
    {
      "epoch": 0.13,
      "grad_norm": 56.5,
      "learning_rate": 4.4093886659348176e-05,
      "loss": 1.0664,
      "step": 142500
    },
    {
      "epoch": 0.13,
      "grad_norm": 28.125,
      "learning_rate": 4.408938758615725e-05,
      "loss": 0.9965,
      "step": 142600
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.125,
      "learning_rate": 4.4084888512966333e-05,
      "loss": 1.2094,
      "step": 142700
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.109375,
      "learning_rate": 4.408038943977541e-05,
      "loss": 0.9811,
      "step": 142800
    },
    {
      "epoch": 0.13,
      "grad_norm": 32.75,
      "learning_rate": 4.407589036658449e-05,
      "loss": 0.9128,
      "step": 142900
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.25,
      "learning_rate": 4.4071391293393566e-05,
      "loss": 0.9645,
      "step": 143000
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.875,
      "learning_rate": 4.4066892220202635e-05,
      "loss": 0.9836,
      "step": 143100
    },
    {
      "epoch": 0.13,
      "grad_norm": 21.5,
      "learning_rate": 4.406239314701172e-05,
      "loss": 0.9767,
      "step": 143200
    },
    {
      "epoch": 0.13,
      "grad_norm": 48.5,
      "learning_rate": 4.405789407382079e-05,
      "loss": 1.2398,
      "step": 143300
    },
    {
      "epoch": 0.13,
      "grad_norm": 203.0,
      "learning_rate": 4.4053395000629874e-05,
      "loss": 0.9407,
      "step": 143400
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.130859375,
      "learning_rate": 4.404889592743895e-05,
      "loss": 1.0356,
      "step": 143500
    },
    {
      "epoch": 0.13,
      "grad_norm": 48.25,
      "learning_rate": 4.4044396854248025e-05,
      "loss": 0.7877,
      "step": 143600
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.625,
      "learning_rate": 4.403989778105711e-05,
      "loss": 1.0302,
      "step": 143700
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.125,
      "learning_rate": 4.403539870786618e-05,
      "loss": 1.1916,
      "step": 143800
    },
    {
      "epoch": 0.13,
      "grad_norm": 46.5,
      "learning_rate": 4.403089963467526e-05,
      "loss": 0.9749,
      "step": 143900
    },
    {
      "epoch": 0.13,
      "grad_norm": 40.25,
      "learning_rate": 4.402640056148434e-05,
      "loss": 1.0308,
      "step": 144000
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.109375,
      "learning_rate": 4.4021901488293415e-05,
      "loss": 0.9644,
      "step": 144100
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.875,
      "learning_rate": 4.401740241510249e-05,
      "loss": 1.2236,
      "step": 144200
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.189453125,
      "learning_rate": 4.401290334191157e-05,
      "loss": 0.9333,
      "step": 144300
    },
    {
      "epoch": 0.13,
      "grad_norm": 80.5,
      "learning_rate": 4.400840426872064e-05,
      "loss": 1.1109,
      "step": 144400
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.5,
      "learning_rate": 4.4003905195529723e-05,
      "loss": 0.9701,
      "step": 144500
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1181640625,
      "learning_rate": 4.39994061223388e-05,
      "loss": 1.0283,
      "step": 144600
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.5625,
      "learning_rate": 4.3994907049147874e-05,
      "loss": 1.0156,
      "step": 144700
    },
    {
      "epoch": 0.13,
      "grad_norm": 41.25,
      "learning_rate": 4.3990407975956956e-05,
      "loss": 1.0755,
      "step": 144800
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.052490234375,
      "learning_rate": 4.398590890276603e-05,
      "loss": 0.9679,
      "step": 144900
    },
    {
      "epoch": 0.13,
      "grad_norm": 38.25,
      "learning_rate": 4.398140982957511e-05,
      "loss": 0.9356,
      "step": 145000
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.9375,
      "learning_rate": 4.397691075638419e-05,
      "loss": 0.9515,
      "step": 145100
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.4375,
      "learning_rate": 4.3972411683193264e-05,
      "loss": 1.0842,
      "step": 145200
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.75,
      "learning_rate": 4.396791261000234e-05,
      "loss": 1.0994,
      "step": 145300
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.25,
      "learning_rate": 4.396341353681142e-05,
      "loss": 0.987,
      "step": 145400
    },
    {
      "epoch": 0.13,
      "grad_norm": 45.0,
      "learning_rate": 4.39589144636205e-05,
      "loss": 1.0725,
      "step": 145500
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.3125,
      "learning_rate": 4.395441539042958e-05,
      "loss": 1.1202,
      "step": 145600
    },
    {
      "epoch": 0.13,
      "grad_norm": 33.0,
      "learning_rate": 4.394991631723865e-05,
      "loss": 1.137,
      "step": 145700
    },
    {
      "epoch": 0.13,
      "grad_norm": 33.75,
      "learning_rate": 4.394541724404772e-05,
      "loss": 1.2172,
      "step": 145800
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.25,
      "learning_rate": 4.3940918170856805e-05,
      "loss": 0.9738,
      "step": 145900
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.6875,
      "learning_rate": 4.393641909766588e-05,
      "loss": 0.9192,
      "step": 146000
    },
    {
      "epoch": 0.13,
      "grad_norm": 55.75,
      "learning_rate": 4.393192002447496e-05,
      "loss": 0.8443,
      "step": 146100
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.5,
      "learning_rate": 4.392742095128404e-05,
      "loss": 0.9508,
      "step": 146200
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.0625,
      "learning_rate": 4.3922921878093113e-05,
      "loss": 0.9376,
      "step": 146300
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.0106201171875,
      "learning_rate": 4.3918422804902196e-05,
      "loss": 0.9247,
      "step": 146400
    },
    {
      "epoch": 0.13,
      "grad_norm": 50.75,
      "learning_rate": 4.391392373171127e-05,
      "loss": 1.0626,
      "step": 146500
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.0625,
      "learning_rate": 4.3909424658520346e-05,
      "loss": 1.0409,
      "step": 146600
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.07861328125,
      "learning_rate": 4.390492558532943e-05,
      "loss": 0.898,
      "step": 146700
    },
    {
      "epoch": 0.13,
      "grad_norm": 71.5,
      "learning_rate": 4.3900426512138504e-05,
      "loss": 1.086,
      "step": 146800
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.04833984375,
      "learning_rate": 4.389592743894758e-05,
      "loss": 1.2122,
      "step": 146900
    },
    {
      "epoch": 0.13,
      "grad_norm": 30.5,
      "learning_rate": 4.3891428365756654e-05,
      "loss": 1.2413,
      "step": 147000
    },
    {
      "epoch": 0.13,
      "grad_norm": 48.75,
      "learning_rate": 4.388692929256573e-05,
      "loss": 1.0457,
      "step": 147100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.01025390625,
      "learning_rate": 4.388243021937481e-05,
      "loss": 1.0146,
      "step": 147200
    },
    {
      "epoch": 0.13,
      "grad_norm": 48.5,
      "learning_rate": 4.387793114618389e-05,
      "loss": 1.1213,
      "step": 147300
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.375,
      "learning_rate": 4.387343207299296e-05,
      "loss": 0.7852,
      "step": 147400
    },
    {
      "epoch": 0.13,
      "grad_norm": 33.75,
      "learning_rate": 4.3868932999802045e-05,
      "loss": 1.0076,
      "step": 147500
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.875,
      "learning_rate": 4.386443392661112e-05,
      "loss": 1.0222,
      "step": 147600
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.1875,
      "learning_rate": 4.3859934853420195e-05,
      "loss": 1.0606,
      "step": 147700
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3828125,
      "learning_rate": 4.385543578022928e-05,
      "loss": 1.0254,
      "step": 147800
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.75,
      "learning_rate": 4.385093670703835e-05,
      "loss": 1.0288,
      "step": 147900
    },
    {
      "epoch": 0.13,
      "grad_norm": 1904.0,
      "learning_rate": 4.384643763384743e-05,
      "loss": 1.0253,
      "step": 148000
    },
    {
      "epoch": 0.13,
      "grad_norm": 47.5,
      "learning_rate": 4.384193856065651e-05,
      "loss": 0.9882,
      "step": 148100
    },
    {
      "epoch": 0.13,
      "grad_norm": 72.5,
      "learning_rate": 4.3837439487465586e-05,
      "loss": 1.0051,
      "step": 148200
    },
    {
      "epoch": 0.13,
      "grad_norm": 184.0,
      "learning_rate": 4.383294041427466e-05,
      "loss": 1.2004,
      "step": 148300
    },
    {
      "epoch": 0.13,
      "grad_norm": 33.75,
      "learning_rate": 4.3828441341083736e-05,
      "loss": 1.1212,
      "step": 148400
    },
    {
      "epoch": 0.13,
      "grad_norm": 340.0,
      "learning_rate": 4.382394226789281e-05,
      "loss": 1.0302,
      "step": 148500
    },
    {
      "epoch": 0.13,
      "grad_norm": 106.5,
      "learning_rate": 4.3819443194701894e-05,
      "loss": 0.9579,
      "step": 148600
    },
    {
      "epoch": 0.13,
      "grad_norm": 57.0,
      "learning_rate": 4.381494412151097e-05,
      "loss": 0.9339,
      "step": 148700
    },
    {
      "epoch": 0.13,
      "grad_norm": 64.0,
      "learning_rate": 4.381044504832005e-05,
      "loss": 1.0883,
      "step": 148800
    },
    {
      "epoch": 0.13,
      "grad_norm": 47.75,
      "learning_rate": 4.3805945975129127e-05,
      "loss": 0.9696,
      "step": 148900
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.75,
      "learning_rate": 4.38014469019382e-05,
      "loss": 1.0203,
      "step": 149000
    },
    {
      "epoch": 0.13,
      "grad_norm": 78.0,
      "learning_rate": 4.3796947828747284e-05,
      "loss": 1.0517,
      "step": 149100
    },
    {
      "epoch": 0.13,
      "grad_norm": 24.5,
      "learning_rate": 4.379244875555636e-05,
      "loss": 0.8561,
      "step": 149200
    },
    {
      "epoch": 0.13,
      "grad_norm": 87.0,
      "learning_rate": 4.3787949682365435e-05,
      "loss": 1.0067,
      "step": 149300
    },
    {
      "epoch": 0.13,
      "grad_norm": 26.375,
      "learning_rate": 4.378345060917452e-05,
      "loss": 1.0164,
      "step": 149400
    },
    {
      "epoch": 0.13,
      "grad_norm": 106.0,
      "learning_rate": 4.377895153598359e-05,
      "loss": 0.9985,
      "step": 149500
    },
    {
      "epoch": 0.13,
      "grad_norm": 21.25,
      "learning_rate": 4.377445246279267e-05,
      "loss": 1.0696,
      "step": 149600
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.875,
      "learning_rate": 4.376995338960174e-05,
      "loss": 0.9058,
      "step": 149700
    },
    {
      "epoch": 0.13,
      "grad_norm": 34.25,
      "learning_rate": 4.376545431641082e-05,
      "loss": 0.9745,
      "step": 149800
    },
    {
      "epoch": 0.13,
      "grad_norm": 21.625,
      "learning_rate": 4.37609552432199e-05,
      "loss": 1.0585,
      "step": 149900
    },
    {
      "epoch": 0.13,
      "grad_norm": 47.5,
      "learning_rate": 4.3756456170028976e-05,
      "loss": 1.0539,
      "step": 150000
    },
    {
      "epoch": 0.13,
      "grad_norm": 26.25,
      "learning_rate": 4.375195709683805e-05,
      "loss": 1.1171,
      "step": 150100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5703125,
      "learning_rate": 4.374745802364713e-05,
      "loss": 1.1021,
      "step": 150200
    },
    {
      "epoch": 0.13,
      "grad_norm": 30.625,
      "learning_rate": 4.374295895045621e-05,
      "loss": 1.101,
      "step": 150300
    },
    {
      "epoch": 0.13,
      "grad_norm": 53.5,
      "learning_rate": 4.3738459877265284e-05,
      "loss": 0.9998,
      "step": 150400
    },
    {
      "epoch": 0.13,
      "grad_norm": 39.0,
      "learning_rate": 4.3733960804074366e-05,
      "loss": 1.1215,
      "step": 150500
    },
    {
      "epoch": 0.13,
      "grad_norm": 2784.0,
      "learning_rate": 4.372946173088344e-05,
      "loss": 1.097,
      "step": 150600
    },
    {
      "epoch": 0.13,
      "grad_norm": 52.0,
      "learning_rate": 4.3724962657692517e-05,
      "loss": 1.0714,
      "step": 150700
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.625,
      "learning_rate": 4.37204635845016e-05,
      "loss": 1.0957,
      "step": 150800
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.75,
      "learning_rate": 4.371596451131067e-05,
      "loss": 1.0594,
      "step": 150900
    },
    {
      "epoch": 0.13,
      "grad_norm": 66.0,
      "learning_rate": 4.371146543811975e-05,
      "loss": 1.0495,
      "step": 151000
    },
    {
      "epoch": 0.13,
      "grad_norm": 67.0,
      "learning_rate": 4.3706966364928825e-05,
      "loss": 1.0036,
      "step": 151100
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.0,
      "learning_rate": 4.37024672917379e-05,
      "loss": 1.0579,
      "step": 151200
    },
    {
      "epoch": 0.13,
      "grad_norm": 25.75,
      "learning_rate": 4.369796821854698e-05,
      "loss": 1.0552,
      "step": 151300
    },
    {
      "epoch": 0.13,
      "grad_norm": 60.75,
      "learning_rate": 4.369346914535606e-05,
      "loss": 1.1418,
      "step": 151400
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.00098419189453125,
      "learning_rate": 4.368897007216514e-05,
      "loss": 1.0137,
      "step": 151500
    },
    {
      "epoch": 0.14,
      "grad_norm": 128.0,
      "learning_rate": 4.3684470998974215e-05,
      "loss": 1.1362,
      "step": 151600
    },
    {
      "epoch": 0.14,
      "grad_norm": 27.625,
      "learning_rate": 4.367997192578329e-05,
      "loss": 0.8731,
      "step": 151700
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.8125,
      "learning_rate": 4.367547285259237e-05,
      "loss": 1.028,
      "step": 151800
    },
    {
      "epoch": 0.14,
      "grad_norm": 82.5,
      "learning_rate": 4.367097377940145e-05,
      "loss": 1.1782,
      "step": 151900
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.625,
      "learning_rate": 4.366647470621052e-05,
      "loss": 0.9741,
      "step": 152000
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.75,
      "learning_rate": 4.3661975633019605e-05,
      "loss": 1.049,
      "step": 152100
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.8125,
      "learning_rate": 4.3657476559828674e-05,
      "loss": 1.0108,
      "step": 152200
    },
    {
      "epoch": 0.14,
      "grad_norm": 55.0,
      "learning_rate": 4.3652977486637756e-05,
      "loss": 1.05,
      "step": 152300
    },
    {
      "epoch": 0.14,
      "grad_norm": 30.125,
      "learning_rate": 4.364847841344683e-05,
      "loss": 1.127,
      "step": 152400
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.8125,
      "learning_rate": 4.3643979340255907e-05,
      "loss": 0.9838,
      "step": 152500
    },
    {
      "epoch": 0.14,
      "grad_norm": 122.5,
      "learning_rate": 4.363948026706499e-05,
      "loss": 0.9758,
      "step": 152600
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.75,
      "learning_rate": 4.3634981193874064e-05,
      "loss": 1.0675,
      "step": 152700
    },
    {
      "epoch": 0.14,
      "grad_norm": 33.25,
      "learning_rate": 4.363048212068314e-05,
      "loss": 1.0756,
      "step": 152800
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.080078125,
      "learning_rate": 4.362598304749222e-05,
      "loss": 1.1334,
      "step": 152900
    },
    {
      "epoch": 0.14,
      "grad_norm": 160.0,
      "learning_rate": 4.36214839743013e-05,
      "loss": 1.0536,
      "step": 153000
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.58984375,
      "learning_rate": 4.361698490111037e-05,
      "loss": 0.9971,
      "step": 153100
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.25,
      "learning_rate": 4.3612485827919454e-05,
      "loss": 0.9975,
      "step": 153200
    },
    {
      "epoch": 0.14,
      "grad_norm": 56.75,
      "learning_rate": 4.360798675472853e-05,
      "loss": 0.9696,
      "step": 153300
    },
    {
      "epoch": 0.14,
      "grad_norm": 21.125,
      "learning_rate": 4.3603487681537605e-05,
      "loss": 1.0212,
      "step": 153400
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.5,
      "learning_rate": 4.359898860834668e-05,
      "loss": 0.9699,
      "step": 153500
    },
    {
      "epoch": 0.14,
      "grad_norm": 30.25,
      "learning_rate": 4.3594489535155756e-05,
      "loss": 0.9782,
      "step": 153600
    },
    {
      "epoch": 0.14,
      "grad_norm": 340.0,
      "learning_rate": 4.358999046196484e-05,
      "loss": 1.0191,
      "step": 153700
    },
    {
      "epoch": 0.14,
      "grad_norm": 50.75,
      "learning_rate": 4.358549138877391e-05,
      "loss": 0.9227,
      "step": 153800
    },
    {
      "epoch": 0.14,
      "grad_norm": 32.25,
      "learning_rate": 4.358099231558299e-05,
      "loss": 1.1029,
      "step": 153900
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3046875,
      "learning_rate": 4.357649324239207e-05,
      "loss": 1.0705,
      "step": 154000
    },
    {
      "epoch": 0.14,
      "grad_norm": 45.25,
      "learning_rate": 4.3571994169201146e-05,
      "loss": 1.0076,
      "step": 154100
    },
    {
      "epoch": 0.14,
      "grad_norm": 48.75,
      "learning_rate": 4.356749509601023e-05,
      "loss": 1.1447,
      "step": 154200
    },
    {
      "epoch": 0.14,
      "grad_norm": 32.25,
      "learning_rate": 4.35629960228193e-05,
      "loss": 1.1206,
      "step": 154300
    },
    {
      "epoch": 0.14,
      "grad_norm": 44.75,
      "learning_rate": 4.355849694962838e-05,
      "loss": 0.9393,
      "step": 154400
    },
    {
      "epoch": 0.14,
      "grad_norm": 28.75,
      "learning_rate": 4.355399787643746e-05,
      "loss": 1.1693,
      "step": 154500
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.6875,
      "learning_rate": 4.3549498803246536e-05,
      "loss": 0.9107,
      "step": 154600
    },
    {
      "epoch": 0.14,
      "grad_norm": 79.0,
      "learning_rate": 4.354499973005561e-05,
      "loss": 1.1249,
      "step": 154700
    },
    {
      "epoch": 0.14,
      "grad_norm": 22.625,
      "learning_rate": 4.354050065686469e-05,
      "loss": 1.0055,
      "step": 154800
    },
    {
      "epoch": 0.14,
      "grad_norm": 45.5,
      "learning_rate": 4.353600158367376e-05,
      "loss": 1.021,
      "step": 154900
    },
    {
      "epoch": 0.14,
      "grad_norm": 62.5,
      "learning_rate": 4.3531502510482844e-05,
      "loss": 0.862,
      "step": 155000
    },
    {
      "epoch": 0.14,
      "grad_norm": 74.0,
      "learning_rate": 4.352700343729192e-05,
      "loss": 1.0474,
      "step": 155100
    },
    {
      "epoch": 0.14,
      "grad_norm": 260.0,
      "learning_rate": 4.3522504364100995e-05,
      "loss": 1.0278,
      "step": 155200
    },
    {
      "epoch": 0.14,
      "grad_norm": 120.0,
      "learning_rate": 4.351800529091008e-05,
      "loss": 1.0703,
      "step": 155300
    },
    {
      "epoch": 0.14,
      "grad_norm": 30.25,
      "learning_rate": 4.351350621771915e-05,
      "loss": 0.9811,
      "step": 155400
    },
    {
      "epoch": 0.14,
      "grad_norm": 36.5,
      "learning_rate": 4.350900714452823e-05,
      "loss": 1.0293,
      "step": 155500
    },
    {
      "epoch": 0.14,
      "grad_norm": 35.5,
      "learning_rate": 4.350450807133731e-05,
      "loss": 0.9772,
      "step": 155600
    },
    {
      "epoch": 0.14,
      "grad_norm": 118.0,
      "learning_rate": 4.3500008998146385e-05,
      "loss": 1.1033,
      "step": 155700
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.13671875,
      "learning_rate": 4.349550992495546e-05,
      "loss": 1.0257,
      "step": 155800
    },
    {
      "epoch": 0.14,
      "grad_norm": 34.25,
      "learning_rate": 4.349101085176454e-05,
      "loss": 1.0198,
      "step": 155900
    },
    {
      "epoch": 0.14,
      "grad_norm": 51.0,
      "learning_rate": 4.348651177857362e-05,
      "loss": 0.9208,
      "step": 156000
    },
    {
      "epoch": 0.14,
      "grad_norm": 20.125,
      "learning_rate": 4.348201270538269e-05,
      "loss": 1.0556,
      "step": 156100
    },
    {
      "epoch": 0.14,
      "grad_norm": 30.25,
      "learning_rate": 4.347751363219177e-05,
      "loss": 1.076,
      "step": 156200
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.5625,
      "learning_rate": 4.3473014559000844e-05,
      "loss": 1.044,
      "step": 156300
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.25,
      "learning_rate": 4.3468515485809926e-05,
      "loss": 1.0219,
      "step": 156400
    },
    {
      "epoch": 0.14,
      "grad_norm": 23.625,
      "learning_rate": 4.3464016412619e-05,
      "loss": 1.1348,
      "step": 156500
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.90625,
      "learning_rate": 4.345951733942808e-05,
      "loss": 0.789,
      "step": 156600
    },
    {
      "epoch": 0.14,
      "grad_norm": 43.25,
      "learning_rate": 4.345501826623716e-05,
      "loss": 0.9491,
      "step": 156700
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.0015716552734375,
      "learning_rate": 4.3450519193046234e-05,
      "loss": 0.9913,
      "step": 156800
    },
    {
      "epoch": 0.14,
      "grad_norm": 105.0,
      "learning_rate": 4.3446020119855316e-05,
      "loss": 1.0048,
      "step": 156900
    },
    {
      "epoch": 0.14,
      "grad_norm": 33.0,
      "learning_rate": 4.344152104666439e-05,
      "loss": 0.8363,
      "step": 157000
    },
    {
      "epoch": 0.14,
      "grad_norm": 113.0,
      "learning_rate": 4.343702197347347e-05,
      "loss": 0.929,
      "step": 157100
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.1875,
      "learning_rate": 4.343252290028255e-05,
      "loss": 0.9984,
      "step": 157200
    },
    {
      "epoch": 0.14,
      "grad_norm": 102.5,
      "learning_rate": 4.3428023827091624e-05,
      "loss": 1.0548,
      "step": 157300
    },
    {
      "epoch": 0.14,
      "grad_norm": 80.5,
      "learning_rate": 4.342352475390069e-05,
      "loss": 0.9331,
      "step": 157400
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.023193359375,
      "learning_rate": 4.3419025680709775e-05,
      "loss": 0.935,
      "step": 157500
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.75,
      "learning_rate": 4.341452660751885e-05,
      "loss": 0.968,
      "step": 157600
    },
    {
      "epoch": 0.14,
      "grad_norm": 44.25,
      "learning_rate": 4.341002753432793e-05,
      "loss": 1.1513,
      "step": 157700
    },
    {
      "epoch": 0.14,
      "grad_norm": 81.0,
      "learning_rate": 4.340552846113701e-05,
      "loss": 0.9997,
      "step": 157800
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.625,
      "learning_rate": 4.340102938794608e-05,
      "loss": 1.0114,
      "step": 157900
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.099609375,
      "learning_rate": 4.3396530314755165e-05,
      "loss": 0.9502,
      "step": 158000
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.8125,
      "learning_rate": 4.339203124156424e-05,
      "loss": 0.889,
      "step": 158100
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.25,
      "learning_rate": 4.3387532168373316e-05,
      "loss": 1.1494,
      "step": 158200
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.32421875,
      "learning_rate": 4.33830330951824e-05,
      "loss": 0.9385,
      "step": 158300
    },
    {
      "epoch": 0.14,
      "grad_norm": 28.125,
      "learning_rate": 4.3378534021991474e-05,
      "loss": 0.815,
      "step": 158400
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.08251953125,
      "learning_rate": 4.337403494880055e-05,
      "loss": 1.0117,
      "step": 158500
    },
    {
      "epoch": 0.14,
      "grad_norm": 107.5,
      "learning_rate": 4.3369535875609624e-05,
      "loss": 1.0547,
      "step": 158600
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.06005859375,
      "learning_rate": 4.33650368024187e-05,
      "loss": 1.0249,
      "step": 158700
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.336053772922778e-05,
      "loss": 0.9179,
      "step": 158800
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.52734375,
      "learning_rate": 4.335603865603686e-05,
      "loss": 1.0161,
      "step": 158900
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.00946044921875,
      "learning_rate": 4.335153958284593e-05,
      "loss": 0.987,
      "step": 159000
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.3347040509655014e-05,
      "loss": 1.0456,
      "step": 159100
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.334254143646409e-05,
      "loss": 1.0138,
      "step": 159200
    },
    {
      "epoch": 0.14,
      "grad_norm": 20.5,
      "learning_rate": 4.3338042363273165e-05,
      "loss": 1.0523,
      "step": 159300
    },
    {
      "epoch": 0.14,
      "grad_norm": 25.625,
      "learning_rate": 4.333354329008225e-05,
      "loss": 0.9729,
      "step": 159400
    },
    {
      "epoch": 0.14,
      "grad_norm": 79.0,
      "learning_rate": 4.332904421689132e-05,
      "loss": 1.1203,
      "step": 159500
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.4375,
      "learning_rate": 4.33245451437004e-05,
      "loss": 1.063,
      "step": 159600
    },
    {
      "epoch": 0.14,
      "grad_norm": 109.0,
      "learning_rate": 4.332004607050948e-05,
      "loss": 1.0203,
      "step": 159700
    },
    {
      "epoch": 0.14,
      "grad_norm": 75.5,
      "learning_rate": 4.3315546997318555e-05,
      "loss": 1.019,
      "step": 159800
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.125,
      "learning_rate": 4.331104792412763e-05,
      "loss": 0.9972,
      "step": 159900
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.038330078125,
      "learning_rate": 4.3306548850936706e-05,
      "loss": 0.9881,
      "step": 160000
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.125,
      "learning_rate": 4.330204977774578e-05,
      "loss": 0.8157,
      "step": 160100
    },
    {
      "epoch": 0.14,
      "grad_norm": 55.75,
      "learning_rate": 4.3297550704554864e-05,
      "loss": 1.1089,
      "step": 160200
    },
    {
      "epoch": 0.14,
      "grad_norm": 56.75,
      "learning_rate": 4.329305163136394e-05,
      "loss": 0.9508,
      "step": 160300
    },
    {
      "epoch": 0.14,
      "grad_norm": 22.5,
      "learning_rate": 4.328855255817302e-05,
      "loss": 1.0825,
      "step": 160400
    },
    {
      "epoch": 0.14,
      "grad_norm": 27.875,
      "learning_rate": 4.3284053484982096e-05,
      "loss": 0.978,
      "step": 160500
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.75,
      "learning_rate": 4.327955441179117e-05,
      "loss": 0.9958,
      "step": 160600
    },
    {
      "epoch": 0.14,
      "grad_norm": 97.5,
      "learning_rate": 4.3275055338600254e-05,
      "loss": 0.8639,
      "step": 160700
    },
    {
      "epoch": 0.14,
      "grad_norm": 227.0,
      "learning_rate": 4.327055626540933e-05,
      "loss": 1.0966,
      "step": 160800
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.875,
      "learning_rate": 4.3266057192218404e-05,
      "loss": 1.0282,
      "step": 160900
    },
    {
      "epoch": 0.14,
      "grad_norm": 30.5,
      "learning_rate": 4.3261558119027487e-05,
      "loss": 0.9126,
      "step": 161000
    },
    {
      "epoch": 0.14,
      "grad_norm": 290.0,
      "learning_rate": 4.325705904583656e-05,
      "loss": 1.1288,
      "step": 161100
    },
    {
      "epoch": 0.14,
      "grad_norm": 68.5,
      "learning_rate": 4.325255997264564e-05,
      "loss": 1.099,
      "step": 161200
    },
    {
      "epoch": 0.14,
      "grad_norm": 19.25,
      "learning_rate": 4.324806089945471e-05,
      "loss": 1.0288,
      "step": 161300
    },
    {
      "epoch": 0.14,
      "grad_norm": 47.0,
      "learning_rate": 4.324356182626379e-05,
      "loss": 1.0872,
      "step": 161400
    },
    {
      "epoch": 0.14,
      "grad_norm": 149.0,
      "learning_rate": 4.323906275307287e-05,
      "loss": 1.0714,
      "step": 161500
    },
    {
      "epoch": 0.14,
      "grad_norm": 103.0,
      "learning_rate": 4.3234563679881945e-05,
      "loss": 1.1005,
      "step": 161600
    },
    {
      "epoch": 0.14,
      "grad_norm": 45.0,
      "learning_rate": 4.323006460669102e-05,
      "loss": 1.0969,
      "step": 161700
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.625,
      "learning_rate": 4.32255655335001e-05,
      "loss": 0.9946,
      "step": 161800
    },
    {
      "epoch": 0.14,
      "grad_norm": 66.0,
      "learning_rate": 4.322106646030918e-05,
      "loss": 0.9462,
      "step": 161900
    },
    {
      "epoch": 0.14,
      "grad_norm": 23.25,
      "learning_rate": 4.3216567387118254e-05,
      "loss": 0.982,
      "step": 162000
    },
    {
      "epoch": 0.14,
      "grad_norm": 20.875,
      "learning_rate": 4.3212068313927336e-05,
      "loss": 0.9534,
      "step": 162100
    },
    {
      "epoch": 0.14,
      "grad_norm": 20.75,
      "learning_rate": 4.320756924073641e-05,
      "loss": 1.0807,
      "step": 162200
    },
    {
      "epoch": 0.14,
      "grad_norm": 39.5,
      "learning_rate": 4.3203070167545486e-05,
      "loss": 1.0738,
      "step": 162300
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.5,
      "learning_rate": 4.319857109435457e-05,
      "loss": 1.0418,
      "step": 162400
    },
    {
      "epoch": 0.14,
      "grad_norm": 46.75,
      "learning_rate": 4.319407202116364e-05,
      "loss": 1.1008,
      "step": 162500
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.0,
      "learning_rate": 4.318957294797272e-05,
      "loss": 0.9745,
      "step": 162600
    },
    {
      "epoch": 0.14,
      "grad_norm": 39.5,
      "learning_rate": 4.3185073874781794e-05,
      "loss": 1.0785,
      "step": 162700
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.875,
      "learning_rate": 4.318057480159087e-05,
      "loss": 0.9414,
      "step": 162800
    },
    {
      "epoch": 0.15,
      "grad_norm": 61.5,
      "learning_rate": 4.317607572839995e-05,
      "loss": 1.02,
      "step": 162900
    },
    {
      "epoch": 0.15,
      "grad_norm": 53.0,
      "learning_rate": 4.317157665520903e-05,
      "loss": 1.0859,
      "step": 163000
    },
    {
      "epoch": 0.15,
      "grad_norm": 25.375,
      "learning_rate": 4.316707758201811e-05,
      "loss": 1.0585,
      "step": 163100
    },
    {
      "epoch": 0.15,
      "grad_norm": 23.125,
      "learning_rate": 4.3162578508827185e-05,
      "loss": 0.9966,
      "step": 163200
    },
    {
      "epoch": 0.15,
      "grad_norm": 47.0,
      "learning_rate": 4.315807943563626e-05,
      "loss": 0.8642,
      "step": 163300
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.189453125,
      "learning_rate": 4.315358036244534e-05,
      "loss": 0.9856,
      "step": 163400
    },
    {
      "epoch": 0.15,
      "grad_norm": 63.25,
      "learning_rate": 4.314908128925442e-05,
      "loss": 0.9111,
      "step": 163500
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.8125,
      "learning_rate": 4.314458221606349e-05,
      "loss": 1.0371,
      "step": 163600
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.875,
      "learning_rate": 4.3140083142872575e-05,
      "loss": 1.1399,
      "step": 163700
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.96875,
      "learning_rate": 4.3135584069681644e-05,
      "loss": 0.8889,
      "step": 163800
    },
    {
      "epoch": 0.15,
      "grad_norm": 22.875,
      "learning_rate": 4.3131084996490726e-05,
      "loss": 1.0474,
      "step": 163900
    },
    {
      "epoch": 0.15,
      "grad_norm": 47.25,
      "learning_rate": 4.31265859232998e-05,
      "loss": 0.9557,
      "step": 164000
    },
    {
      "epoch": 0.15,
      "grad_norm": 43.75,
      "learning_rate": 4.3122086850108876e-05,
      "loss": 1.0484,
      "step": 164100
    },
    {
      "epoch": 0.15,
      "grad_norm": 65.0,
      "learning_rate": 4.311758777691796e-05,
      "loss": 1.22,
      "step": 164200
    },
    {
      "epoch": 0.15,
      "grad_norm": 44.25,
      "learning_rate": 4.3113088703727034e-05,
      "loss": 0.9491,
      "step": 164300
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.5,
      "learning_rate": 4.310858963053611e-05,
      "loss": 0.9652,
      "step": 164400
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.0,
      "learning_rate": 4.310409055734519e-05,
      "loss": 1.0211,
      "step": 164500
    },
    {
      "epoch": 0.15,
      "grad_norm": 58.75,
      "learning_rate": 4.3099591484154267e-05,
      "loss": 0.9964,
      "step": 164600
    },
    {
      "epoch": 0.15,
      "grad_norm": 59.5,
      "learning_rate": 4.309509241096334e-05,
      "loss": 1.0356,
      "step": 164700
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.4375,
      "learning_rate": 4.3090593337772424e-05,
      "loss": 1.0443,
      "step": 164800
    },
    {
      "epoch": 0.15,
      "grad_norm": 35.25,
      "learning_rate": 4.30860942645815e-05,
      "loss": 1.0688,
      "step": 164900
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.5,
      "learning_rate": 4.3081595191390575e-05,
      "loss": 1.0226,
      "step": 165000
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.875,
      "learning_rate": 4.307709611819965e-05,
      "loss": 0.9612,
      "step": 165100
    },
    {
      "epoch": 0.15,
      "grad_norm": 33.25,
      "learning_rate": 4.3072597045008725e-05,
      "loss": 1.1404,
      "step": 165200
    },
    {
      "epoch": 0.15,
      "grad_norm": 500.0,
      "learning_rate": 4.306809797181781e-05,
      "loss": 0.9332,
      "step": 165300
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.5625,
      "learning_rate": 4.306359889862688e-05,
      "loss": 1.0052,
      "step": 165400
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.25,
      "learning_rate": 4.305909982543596e-05,
      "loss": 1.0863,
      "step": 165500
    },
    {
      "epoch": 0.15,
      "grad_norm": 45.75,
      "learning_rate": 4.305460075224504e-05,
      "loss": 1.0065,
      "step": 165600
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.6875,
      "learning_rate": 4.3050101679054116e-05,
      "loss": 1.1437,
      "step": 165700
    },
    {
      "epoch": 0.15,
      "grad_norm": 41.25,
      "learning_rate": 4.30456026058632e-05,
      "loss": 0.9853,
      "step": 165800
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.609375,
      "learning_rate": 4.304110353267227e-05,
      "loss": 1.0444,
      "step": 165900
    },
    {
      "epoch": 0.15,
      "grad_norm": 1032.0,
      "learning_rate": 4.303660445948135e-05,
      "loss": 1.0349,
      "step": 166000
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.375,
      "learning_rate": 4.303210538629043e-05,
      "loss": 0.9965,
      "step": 166100
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.0537109375,
      "learning_rate": 4.3027606313099506e-05,
      "loss": 1.1163,
      "step": 166200
    },
    {
      "epoch": 0.15,
      "grad_norm": 37.5,
      "learning_rate": 4.302310723990858e-05,
      "loss": 1.0659,
      "step": 166300
    },
    {
      "epoch": 0.15,
      "grad_norm": 32.0,
      "learning_rate": 4.3018608166717657e-05,
      "loss": 0.983,
      "step": 166400
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.125,
      "learning_rate": 4.301410909352673e-05,
      "loss": 0.9871,
      "step": 166500
    },
    {
      "epoch": 0.15,
      "grad_norm": 34.5,
      "learning_rate": 4.3009610020335814e-05,
      "loss": 1.1223,
      "step": 166600
    },
    {
      "epoch": 0.15,
      "grad_norm": 45.5,
      "learning_rate": 4.300511094714489e-05,
      "loss": 1.0868,
      "step": 166700
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.5,
      "learning_rate": 4.3000611873953965e-05,
      "loss": 1.0496,
      "step": 166800
    },
    {
      "epoch": 0.15,
      "grad_norm": 39.25,
      "learning_rate": 4.299611280076305e-05,
      "loss": 1.0291,
      "step": 166900
    },
    {
      "epoch": 0.15,
      "grad_norm": 25.875,
      "learning_rate": 4.299161372757212e-05,
      "loss": 1.013,
      "step": 167000
    },
    {
      "epoch": 0.15,
      "grad_norm": 36.5,
      "learning_rate": 4.29871146543812e-05,
      "loss": 1.0793,
      "step": 167100
    },
    {
      "epoch": 0.15,
      "grad_norm": 134.0,
      "learning_rate": 4.298261558119028e-05,
      "loss": 1.126,
      "step": 167200
    },
    {
      "epoch": 0.15,
      "grad_norm": 44.5,
      "learning_rate": 4.2978116507999355e-05,
      "loss": 1.1957,
      "step": 167300
    },
    {
      "epoch": 0.15,
      "grad_norm": 22.0,
      "learning_rate": 4.297361743480843e-05,
      "loss": 1.147,
      "step": 167400
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.40625,
      "learning_rate": 4.296911836161751e-05,
      "loss": 1.011,
      "step": 167500
    },
    {
      "epoch": 0.15,
      "grad_norm": 266.0,
      "learning_rate": 4.296461928842659e-05,
      "loss": 1.1574,
      "step": 167600
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.625,
      "learning_rate": 4.296012021523566e-05,
      "loss": 1.0342,
      "step": 167700
    },
    {
      "epoch": 0.15,
      "grad_norm": 67.5,
      "learning_rate": 4.295562114204474e-05,
      "loss": 1.087,
      "step": 167800
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.625,
      "learning_rate": 4.2951122068853814e-05,
      "loss": 1.0295,
      "step": 167900
    },
    {
      "epoch": 0.15,
      "grad_norm": 107.5,
      "learning_rate": 4.2946622995662896e-05,
      "loss": 1.0703,
      "step": 168000
    },
    {
      "epoch": 0.15,
      "grad_norm": 31.5,
      "learning_rate": 4.294212392247197e-05,
      "loss": 1.0369,
      "step": 168100
    },
    {
      "epoch": 0.15,
      "grad_norm": 310.0,
      "learning_rate": 4.2937624849281047e-05,
      "loss": 1.127,
      "step": 168200
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.6875,
      "learning_rate": 4.293312577609013e-05,
      "loss": 1.0062,
      "step": 168300
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.4375,
      "learning_rate": 4.2928626702899204e-05,
      "loss": 1.0202,
      "step": 168400
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.0,
      "learning_rate": 4.2924127629708286e-05,
      "loss": 0.9727,
      "step": 168500
    },
    {
      "epoch": 0.15,
      "grad_norm": 208.0,
      "learning_rate": 4.291962855651736e-05,
      "loss": 0.9625,
      "step": 168600
    },
    {
      "epoch": 0.15,
      "grad_norm": 55.25,
      "learning_rate": 4.291512948332644e-05,
      "loss": 1.1664,
      "step": 168700
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.1875,
      "learning_rate": 4.291063041013552e-05,
      "loss": 1.1378,
      "step": 168800
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.25,
      "learning_rate": 4.2906131336944594e-05,
      "loss": 0.9084,
      "step": 168900
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.875,
      "learning_rate": 4.290163226375366e-05,
      "loss": 0.9553,
      "step": 169000
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.0517578125,
      "learning_rate": 4.2897133190562745e-05,
      "loss": 0.9911,
      "step": 169100
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.289263411737182e-05,
      "loss": 1.0727,
      "step": 169200
    },
    {
      "epoch": 0.15,
      "grad_norm": 89.0,
      "learning_rate": 4.28881350441809e-05,
      "loss": 1.1094,
      "step": 169300
    },
    {
      "epoch": 0.15,
      "grad_norm": 47.0,
      "learning_rate": 4.288363597098998e-05,
      "loss": 0.901,
      "step": 169400
    },
    {
      "epoch": 0.15,
      "grad_norm": 32.25,
      "learning_rate": 4.287913689779905e-05,
      "loss": 0.9228,
      "step": 169500
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.125,
      "learning_rate": 4.2874637824608135e-05,
      "loss": 0.9525,
      "step": 169600
    },
    {
      "epoch": 0.15,
      "grad_norm": 284.0,
      "learning_rate": 4.287013875141721e-05,
      "loss": 1.0846,
      "step": 169700
    },
    {
      "epoch": 0.15,
      "grad_norm": 33.25,
      "learning_rate": 4.2865639678226286e-05,
      "loss": 1.1031,
      "step": 169800
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19140625,
      "learning_rate": 4.286114060503537e-05,
      "loss": 0.9617,
      "step": 169900
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.625,
      "learning_rate": 4.285664153184444e-05,
      "loss": 0.9865,
      "step": 170000
    },
    {
      "epoch": 0.15,
      "grad_norm": 44.25,
      "learning_rate": 4.285214245865352e-05,
      "loss": 1.0339,
      "step": 170100
    },
    {
      "epoch": 0.15,
      "grad_norm": 42.0,
      "learning_rate": 4.28476433854626e-05,
      "loss": 1.0454,
      "step": 170200
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.09326171875,
      "learning_rate": 4.284314431227167e-05,
      "loss": 1.1814,
      "step": 170300
    },
    {
      "epoch": 0.15,
      "grad_norm": 22.125,
      "learning_rate": 4.283864523908075e-05,
      "loss": 1.0357,
      "step": 170400
    },
    {
      "epoch": 0.15,
      "grad_norm": 30.125,
      "learning_rate": 4.283414616588983e-05,
      "loss": 0.9267,
      "step": 170500
    },
    {
      "epoch": 0.15,
      "grad_norm": 37.5,
      "learning_rate": 4.28296470926989e-05,
      "loss": 1.1452,
      "step": 170600
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.625,
      "learning_rate": 4.2825148019507984e-05,
      "loss": 1.0902,
      "step": 170700
    },
    {
      "epoch": 0.15,
      "grad_norm": 38.25,
      "learning_rate": 4.282064894631706e-05,
      "loss": 0.9456,
      "step": 170800
    },
    {
      "epoch": 0.15,
      "grad_norm": 27.5,
      "learning_rate": 4.2816149873126135e-05,
      "loss": 0.9448,
      "step": 170900
    },
    {
      "epoch": 0.15,
      "grad_norm": 35.0,
      "learning_rate": 4.281165079993522e-05,
      "loss": 1.0657,
      "step": 171000
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.25,
      "learning_rate": 4.280715172674429e-05,
      "loss": 0.9641,
      "step": 171100
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.375,
      "learning_rate": 4.2802652653553375e-05,
      "loss": 0.9344,
      "step": 171200
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.375,
      "learning_rate": 4.279815358036245e-05,
      "loss": 0.9064,
      "step": 171300
    },
    {
      "epoch": 0.15,
      "grad_norm": 59.5,
      "learning_rate": 4.2793654507171525e-05,
      "loss": 0.9617,
      "step": 171400
    },
    {
      "epoch": 0.15,
      "grad_norm": 18.5,
      "learning_rate": 4.278915543398061e-05,
      "loss": 1.0616,
      "step": 171500
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.5,
      "learning_rate": 4.2784656360789676e-05,
      "loss": 1.013,
      "step": 171600
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.56640625,
      "learning_rate": 4.278015728759875e-05,
      "loss": 0.9883,
      "step": 171700
    },
    {
      "epoch": 0.15,
      "grad_norm": 78.5,
      "learning_rate": 4.277565821440783e-05,
      "loss": 0.9044,
      "step": 171800
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.875,
      "learning_rate": 4.277115914121691e-05,
      "loss": 0.9471,
      "step": 171900
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9921875,
      "learning_rate": 4.276666006802599e-05,
      "loss": 1.041,
      "step": 172000
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.5625,
      "learning_rate": 4.2762160994835066e-05,
      "loss": 1.1452,
      "step": 172100
    },
    {
      "epoch": 0.15,
      "grad_norm": 67.5,
      "learning_rate": 4.275766192164414e-05,
      "loss": 1.1141,
      "step": 172200
    },
    {
      "epoch": 0.15,
      "grad_norm": 39.75,
      "learning_rate": 4.2753162848453224e-05,
      "loss": 0.8427,
      "step": 172300
    },
    {
      "epoch": 0.15,
      "grad_norm": 67.5,
      "learning_rate": 4.27486637752623e-05,
      "loss": 0.9316,
      "step": 172400
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8046875,
      "learning_rate": 4.2744164702071374e-05,
      "loss": 1.1593,
      "step": 172500
    },
    {
      "epoch": 0.15,
      "grad_norm": 127.0,
      "learning_rate": 4.2739665628880456e-05,
      "loss": 1.0972,
      "step": 172600
    },
    {
      "epoch": 0.15,
      "grad_norm": 70.5,
      "learning_rate": 4.273516655568953e-05,
      "loss": 0.8319,
      "step": 172700
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.25,
      "learning_rate": 4.273066748249861e-05,
      "loss": 0.913,
      "step": 172800
    },
    {
      "epoch": 0.15,
      "grad_norm": 35.75,
      "learning_rate": 4.272616840930768e-05,
      "loss": 1.0221,
      "step": 172900
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.375,
      "learning_rate": 4.272166933611676e-05,
      "loss": 0.961,
      "step": 173000
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.9375,
      "learning_rate": 4.271717026292584e-05,
      "loss": 1.1848,
      "step": 173100
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.0203857421875,
      "learning_rate": 4.2712671189734915e-05,
      "loss": 1.1035,
      "step": 173200
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.375,
      "learning_rate": 4.270817211654399e-05,
      "loss": 1.1222,
      "step": 173300
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.375,
      "learning_rate": 4.270367304335307e-05,
      "loss": 1.0297,
      "step": 173400
    },
    {
      "epoch": 0.15,
      "grad_norm": 78.0,
      "learning_rate": 4.269917397016215e-05,
      "loss": 1.0118,
      "step": 173500
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.0,
      "learning_rate": 4.269467489697122e-05,
      "loss": 0.9917,
      "step": 173600
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.97265625,
      "learning_rate": 4.2690175823780305e-05,
      "loss": 0.903,
      "step": 173700
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.75,
      "learning_rate": 4.268567675058938e-05,
      "loss": 0.8788,
      "step": 173800
    },
    {
      "epoch": 0.15,
      "grad_norm": 78.5,
      "learning_rate": 4.268117767739846e-05,
      "loss": 1.0553,
      "step": 173900
    },
    {
      "epoch": 0.16,
      "grad_norm": 29.75,
      "learning_rate": 4.267667860420754e-05,
      "loss": 0.9663,
      "step": 174000
    },
    {
      "epoch": 0.16,
      "grad_norm": 38.5,
      "learning_rate": 4.2672179531016614e-05,
      "loss": 1.0341,
      "step": 174100
    },
    {
      "epoch": 0.16,
      "grad_norm": 53.5,
      "learning_rate": 4.266768045782569e-05,
      "loss": 0.931,
      "step": 174200
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.765625,
      "learning_rate": 4.2663181384634764e-05,
      "loss": 0.9649,
      "step": 174300
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.265868231144384e-05,
      "loss": 0.9726,
      "step": 174400
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.875,
      "learning_rate": 4.265418323825292e-05,
      "loss": 0.8931,
      "step": 174500
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.375,
      "learning_rate": 4.2649684165062e-05,
      "loss": 1.0549,
      "step": 174600
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0703125,
      "learning_rate": 4.264518509187108e-05,
      "loss": 1.0041,
      "step": 174700
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.75,
      "learning_rate": 4.2640686018680155e-05,
      "loss": 0.9753,
      "step": 174800
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.875,
      "learning_rate": 4.263618694548923e-05,
      "loss": 1.0315,
      "step": 174900
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.25,
      "learning_rate": 4.263168787229831e-05,
      "loss": 1.0886,
      "step": 175000
    },
    {
      "epoch": 0.16,
      "grad_norm": 55.0,
      "learning_rate": 4.262718879910739e-05,
      "loss": 0.9983,
      "step": 175100
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.75,
      "learning_rate": 4.262268972591646e-05,
      "loss": 0.9282,
      "step": 175200
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.1875,
      "learning_rate": 4.2618190652725545e-05,
      "loss": 1.1496,
      "step": 175300
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.5,
      "learning_rate": 4.261369157953462e-05,
      "loss": 1.0169,
      "step": 175400
    },
    {
      "epoch": 0.16,
      "grad_norm": 48.75,
      "learning_rate": 4.2609192506343695e-05,
      "loss": 1.0309,
      "step": 175500
    },
    {
      "epoch": 0.16,
      "grad_norm": 48.75,
      "learning_rate": 4.260469343315277e-05,
      "loss": 1.0293,
      "step": 175600
    },
    {
      "epoch": 0.16,
      "grad_norm": 43.25,
      "learning_rate": 4.2600194359961846e-05,
      "loss": 1.0975,
      "step": 175700
    },
    {
      "epoch": 0.16,
      "grad_norm": 24.75,
      "learning_rate": 4.259569528677093e-05,
      "loss": 1.0853,
      "step": 175800
    },
    {
      "epoch": 0.16,
      "grad_norm": 24.25,
      "learning_rate": 4.2591196213580004e-05,
      "loss": 1.0873,
      "step": 175900
    },
    {
      "epoch": 0.16,
      "grad_norm": 50.0,
      "learning_rate": 4.258669714038908e-05,
      "loss": 1.0547,
      "step": 176000
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.375,
      "learning_rate": 4.258219806719816e-05,
      "loss": 1.0271,
      "step": 176100
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.625,
      "learning_rate": 4.2577698994007236e-05,
      "loss": 0.9698,
      "step": 176200
    },
    {
      "epoch": 0.16,
      "grad_norm": 39.25,
      "learning_rate": 4.257319992081631e-05,
      "loss": 1.1282,
      "step": 176300
    },
    {
      "epoch": 0.16,
      "grad_norm": 21.875,
      "learning_rate": 4.2568700847625394e-05,
      "loss": 1.0358,
      "step": 176400
    },
    {
      "epoch": 0.16,
      "grad_norm": 31.75,
      "learning_rate": 4.256420177443447e-05,
      "loss": 1.0171,
      "step": 176500
    },
    {
      "epoch": 0.16,
      "grad_norm": 186.0,
      "learning_rate": 4.2559702701243544e-05,
      "loss": 0.8717,
      "step": 176600
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5625,
      "learning_rate": 4.255520362805263e-05,
      "loss": 1.044,
      "step": 176700
    },
    {
      "epoch": 0.16,
      "grad_norm": 262.0,
      "learning_rate": 4.2550704554861695e-05,
      "loss": 0.9051,
      "step": 176800
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.490234375,
      "learning_rate": 4.254620548167078e-05,
      "loss": 1.1272,
      "step": 176900
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.84375,
      "learning_rate": 4.254170640847985e-05,
      "loss": 1.0549,
      "step": 177000
    },
    {
      "epoch": 0.16,
      "grad_norm": 37.0,
      "learning_rate": 4.253720733528893e-05,
      "loss": 1.0612,
      "step": 177100
    },
    {
      "epoch": 0.16,
      "grad_norm": 223.0,
      "learning_rate": 4.253270826209801e-05,
      "loss": 0.9417,
      "step": 177200
    },
    {
      "epoch": 0.16,
      "grad_norm": 556.0,
      "learning_rate": 4.2528209188907085e-05,
      "loss": 0.8947,
      "step": 177300
    },
    {
      "epoch": 0.16,
      "grad_norm": 40.75,
      "learning_rate": 4.252371011571617e-05,
      "loss": 1.0793,
      "step": 177400
    },
    {
      "epoch": 0.16,
      "grad_norm": 50.5,
      "learning_rate": 4.251921104252524e-05,
      "loss": 1.03,
      "step": 177500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.85546875,
      "learning_rate": 4.251471196933432e-05,
      "loss": 1.0201,
      "step": 177600
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.125,
      "learning_rate": 4.25102128961434e-05,
      "loss": 1.0034,
      "step": 177700
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.00274658203125,
      "learning_rate": 4.2505713822952476e-05,
      "loss": 0.9403,
      "step": 177800
    },
    {
      "epoch": 0.16,
      "grad_norm": 29.375,
      "learning_rate": 4.250121474976155e-05,
      "loss": 0.9524,
      "step": 177900
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.25,
      "learning_rate": 4.249671567657063e-05,
      "loss": 0.9907,
      "step": 178000
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.5625,
      "learning_rate": 4.24922166033797e-05,
      "loss": 0.8962,
      "step": 178100
    },
    {
      "epoch": 0.16,
      "grad_norm": 860.0,
      "learning_rate": 4.2487717530188784e-05,
      "loss": 0.8981,
      "step": 178200
    },
    {
      "epoch": 0.16,
      "grad_norm": 410.0,
      "learning_rate": 4.248321845699786e-05,
      "loss": 1.0501,
      "step": 178300
    },
    {
      "epoch": 0.16,
      "grad_norm": 41.25,
      "learning_rate": 4.2478719383806934e-05,
      "loss": 0.8877,
      "step": 178400
    },
    {
      "epoch": 0.16,
      "grad_norm": 42.75,
      "learning_rate": 4.2474220310616017e-05,
      "loss": 1.0879,
      "step": 178500
    },
    {
      "epoch": 0.16,
      "grad_norm": 56.0,
      "learning_rate": 4.246972123742509e-05,
      "loss": 0.9904,
      "step": 178600
    },
    {
      "epoch": 0.16,
      "grad_norm": 24.0,
      "learning_rate": 4.246522216423417e-05,
      "loss": 1.1506,
      "step": 178700
    },
    {
      "epoch": 0.16,
      "grad_norm": 50.25,
      "learning_rate": 4.246072309104325e-05,
      "loss": 1.1529,
      "step": 178800
    },
    {
      "epoch": 0.16,
      "grad_norm": 58.0,
      "learning_rate": 4.2456224017852325e-05,
      "loss": 1.0478,
      "step": 178900
    },
    {
      "epoch": 0.16,
      "grad_norm": 33.0,
      "learning_rate": 4.24517249446614e-05,
      "loss": 1.0825,
      "step": 179000
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.375,
      "learning_rate": 4.244722587147048e-05,
      "loss": 0.9179,
      "step": 179100
    },
    {
      "epoch": 0.16,
      "grad_norm": 86.0,
      "learning_rate": 4.244272679827956e-05,
      "loss": 1.0197,
      "step": 179200
    },
    {
      "epoch": 0.16,
      "grad_norm": 30.25,
      "learning_rate": 4.243822772508863e-05,
      "loss": 1.1412,
      "step": 179300
    },
    {
      "epoch": 0.16,
      "grad_norm": 88.5,
      "learning_rate": 4.243372865189771e-05,
      "loss": 1.1946,
      "step": 179400
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.232421875,
      "learning_rate": 4.2429229578706784e-05,
      "loss": 0.9457,
      "step": 179500
    },
    {
      "epoch": 0.16,
      "grad_norm": 41.25,
      "learning_rate": 4.2424730505515866e-05,
      "loss": 1.1139,
      "step": 179600
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.0,
      "learning_rate": 4.242023143232494e-05,
      "loss": 1.1794,
      "step": 179700
    },
    {
      "epoch": 0.16,
      "grad_norm": 61.25,
      "learning_rate": 4.2415732359134016e-05,
      "loss": 0.98,
      "step": 179800
    },
    {
      "epoch": 0.16,
      "grad_norm": 39.0,
      "learning_rate": 4.24112332859431e-05,
      "loss": 1.0908,
      "step": 179900
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.0,
      "learning_rate": 4.2406734212752174e-05,
      "loss": 1.149,
      "step": 180000
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.75,
      "learning_rate": 4.2402235139561256e-05,
      "loss": 1.1205,
      "step": 180100
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.007171630859375,
      "learning_rate": 4.239773606637033e-05,
      "loss": 0.8894,
      "step": 180200
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.125,
      "learning_rate": 4.2393236993179407e-05,
      "loss": 0.9269,
      "step": 180300
    },
    {
      "epoch": 0.16,
      "grad_norm": 23.0,
      "learning_rate": 4.238873791998849e-05,
      "loss": 0.9371,
      "step": 180400
    },
    {
      "epoch": 0.16,
      "grad_norm": 70.0,
      "learning_rate": 4.2384238846797564e-05,
      "loss": 0.9446,
      "step": 180500
    },
    {
      "epoch": 0.16,
      "grad_norm": 28.375,
      "learning_rate": 4.237973977360664e-05,
      "loss": 0.988,
      "step": 180600
    },
    {
      "epoch": 0.16,
      "grad_norm": 75.5,
      "learning_rate": 4.2375240700415715e-05,
      "loss": 0.9541,
      "step": 180700
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.36328125,
      "learning_rate": 4.237074162722479e-05,
      "loss": 0.9531,
      "step": 180800
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.4375,
      "learning_rate": 4.236624255403387e-05,
      "loss": 0.9431,
      "step": 180900
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.0,
      "learning_rate": 4.236174348084295e-05,
      "loss": 0.9906,
      "step": 181000
    },
    {
      "epoch": 0.16,
      "grad_norm": 50.75,
      "learning_rate": 4.235724440765202e-05,
      "loss": 1.0392,
      "step": 181100
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.037109375,
      "learning_rate": 4.2352745334461105e-05,
      "loss": 1.0021,
      "step": 181200
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.40625,
      "learning_rate": 4.234824626127018e-05,
      "loss": 1.122,
      "step": 181300
    },
    {
      "epoch": 0.16,
      "grad_norm": 28.0,
      "learning_rate": 4.2343747188079256e-05,
      "loss": 1.1188,
      "step": 181400
    },
    {
      "epoch": 0.16,
      "grad_norm": 100.0,
      "learning_rate": 4.233924811488834e-05,
      "loss": 1.0672,
      "step": 181500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.233474904169741e-05,
      "loss": 1.031,
      "step": 181600
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.00154876708984375,
      "learning_rate": 4.233024996850649e-05,
      "loss": 0.9535,
      "step": 181700
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.162109375,
      "learning_rate": 4.232575089531557e-05,
      "loss": 0.9562,
      "step": 181800
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.3125,
      "learning_rate": 4.2321251822124646e-05,
      "loss": 0.8395,
      "step": 181900
    },
    {
      "epoch": 0.16,
      "grad_norm": 60.0,
      "learning_rate": 4.231675274893372e-05,
      "loss": 0.9971,
      "step": 182000
    },
    {
      "epoch": 0.16,
      "grad_norm": 38.5,
      "learning_rate": 4.2312253675742797e-05,
      "loss": 1.0235,
      "step": 182100
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.0186767578125,
      "learning_rate": 4.230775460255187e-05,
      "loss": 1.1027,
      "step": 182200
    },
    {
      "epoch": 0.16,
      "grad_norm": 78.5,
      "learning_rate": 4.2303255529360954e-05,
      "loss": 1.0614,
      "step": 182300
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.8125,
      "learning_rate": 4.229875645617003e-05,
      "loss": 1.0475,
      "step": 182400
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.125,
      "learning_rate": 4.2294257382979105e-05,
      "loss": 0.991,
      "step": 182500
    },
    {
      "epoch": 0.16,
      "grad_norm": 21.375,
      "learning_rate": 4.228975830978819e-05,
      "loss": 1.0521,
      "step": 182600
    },
    {
      "epoch": 0.16,
      "grad_norm": 23.875,
      "learning_rate": 4.228525923659726e-05,
      "loss": 1.168,
      "step": 182700
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.03515625,
      "learning_rate": 4.2280760163406344e-05,
      "loss": 1.0485,
      "step": 182800
    },
    {
      "epoch": 0.16,
      "grad_norm": 88.5,
      "learning_rate": 4.227626109021542e-05,
      "loss": 0.9773,
      "step": 182900
    },
    {
      "epoch": 0.16,
      "grad_norm": 89.0,
      "learning_rate": 4.2271762017024495e-05,
      "loss": 0.9462,
      "step": 183000
    },
    {
      "epoch": 0.16,
      "grad_norm": 46.25,
      "learning_rate": 4.226726294383358e-05,
      "loss": 1.1444,
      "step": 183100
    },
    {
      "epoch": 0.16,
      "grad_norm": 26.375,
      "learning_rate": 4.226276387064265e-05,
      "loss": 0.8055,
      "step": 183200
    },
    {
      "epoch": 0.16,
      "grad_norm": 85.5,
      "learning_rate": 4.225826479745172e-05,
      "loss": 0.9264,
      "step": 183300
    },
    {
      "epoch": 0.16,
      "grad_norm": 18.375,
      "learning_rate": 4.22537657242608e-05,
      "loss": 0.8415,
      "step": 183400
    },
    {
      "epoch": 0.16,
      "grad_norm": 58.75,
      "learning_rate": 4.224926665106988e-05,
      "loss": 1.0071,
      "step": 183500
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.0,
      "learning_rate": 4.224476757787896e-05,
      "loss": 1.0227,
      "step": 183600
    },
    {
      "epoch": 0.16,
      "grad_norm": 23.625,
      "learning_rate": 4.2240268504688036e-05,
      "loss": 1.2288,
      "step": 183700
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.21875,
      "learning_rate": 4.223576943149711e-05,
      "loss": 0.8231,
      "step": 183800
    },
    {
      "epoch": 0.16,
      "grad_norm": 38.25,
      "learning_rate": 4.223127035830619e-05,
      "loss": 1.1156,
      "step": 183900
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.875,
      "learning_rate": 4.222677128511527e-05,
      "loss": 0.9293,
      "step": 184000
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.375,
      "learning_rate": 4.2222272211924344e-05,
      "loss": 0.9727,
      "step": 184100
    },
    {
      "epoch": 0.16,
      "grad_norm": 278.0,
      "learning_rate": 4.2217773138733426e-05,
      "loss": 1.0683,
      "step": 184200
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.375,
      "learning_rate": 4.22132740655425e-05,
      "loss": 0.9945,
      "step": 184300
    },
    {
      "epoch": 0.16,
      "grad_norm": 26.375,
      "learning_rate": 4.220877499235158e-05,
      "loss": 1.0406,
      "step": 184400
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.8125,
      "learning_rate": 4.220427591916066e-05,
      "loss": 1.1124,
      "step": 184500
    },
    {
      "epoch": 0.16,
      "grad_norm": 40.5,
      "learning_rate": 4.219977684596973e-05,
      "loss": 1.0967,
      "step": 184600
    },
    {
      "epoch": 0.16,
      "grad_norm": 231.0,
      "learning_rate": 4.219527777277881e-05,
      "loss": 1.0927,
      "step": 184700
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.75,
      "learning_rate": 4.2190778699587885e-05,
      "loss": 1.1185,
      "step": 184800
    },
    {
      "epoch": 0.16,
      "grad_norm": 35.0,
      "learning_rate": 4.218627962639696e-05,
      "loss": 1.1544,
      "step": 184900
    },
    {
      "epoch": 0.16,
      "grad_norm": 90.0,
      "learning_rate": 4.218178055320604e-05,
      "loss": 0.9832,
      "step": 185000
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.296875,
      "learning_rate": 4.217728148001512e-05,
      "loss": 1.0514,
      "step": 185100
    },
    {
      "epoch": 0.16,
      "grad_norm": 62.5,
      "learning_rate": 4.217278240682419e-05,
      "loss": 0.9806,
      "step": 185200
    },
    {
      "epoch": 0.17,
      "grad_norm": 58.25,
      "learning_rate": 4.2168283333633275e-05,
      "loss": 1.0324,
      "step": 185300
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.0,
      "learning_rate": 4.216378426044235e-05,
      "loss": 1.0414,
      "step": 185400
    },
    {
      "epoch": 0.17,
      "grad_norm": 29.5,
      "learning_rate": 4.215928518725143e-05,
      "loss": 1.1363,
      "step": 185500
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.625,
      "learning_rate": 4.215478611406051e-05,
      "loss": 1.0124,
      "step": 185600
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.25,
      "learning_rate": 4.215028704086958e-05,
      "loss": 1.1343,
      "step": 185700
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.3125,
      "learning_rate": 4.2145787967678665e-05,
      "loss": 0.9466,
      "step": 185800
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.0986328125,
      "learning_rate": 4.2141288894487734e-05,
      "loss": 0.9751,
      "step": 185900
    },
    {
      "epoch": 0.17,
      "grad_norm": 51.5,
      "learning_rate": 4.213678982129681e-05,
      "loss": 1.2408,
      "step": 186000
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.96875,
      "learning_rate": 4.213229074810589e-05,
      "loss": 1.0506,
      "step": 186100
    },
    {
      "epoch": 0.17,
      "grad_norm": 47.0,
      "learning_rate": 4.212779167491497e-05,
      "loss": 1.0667,
      "step": 186200
    },
    {
      "epoch": 0.17,
      "grad_norm": 124.5,
      "learning_rate": 4.212329260172405e-05,
      "loss": 1.0205,
      "step": 186300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 4.2118793528533124e-05,
      "loss": 1.0697,
      "step": 186400
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3984375,
      "learning_rate": 4.21142944553422e-05,
      "loss": 0.8599,
      "step": 186500
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.0031585693359375,
      "learning_rate": 4.210979538215128e-05,
      "loss": 0.9359,
      "step": 186600
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1455078125,
      "learning_rate": 4.210529630896036e-05,
      "loss": 1.0606,
      "step": 186700
    },
    {
      "epoch": 0.17,
      "grad_norm": 49.25,
      "learning_rate": 4.210079723576943e-05,
      "loss": 1.1038,
      "step": 186800
    },
    {
      "epoch": 0.17,
      "grad_norm": 27.0,
      "learning_rate": 4.2096298162578515e-05,
      "loss": 0.9061,
      "step": 186900
    },
    {
      "epoch": 0.17,
      "grad_norm": 516.0,
      "learning_rate": 4.209179908938759e-05,
      "loss": 0.9282,
      "step": 187000
    },
    {
      "epoch": 0.17,
      "grad_norm": 61.25,
      "learning_rate": 4.2087300016196665e-05,
      "loss": 1.0016,
      "step": 187100
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.208280094300574e-05,
      "loss": 0.9262,
      "step": 187200
    },
    {
      "epoch": 0.17,
      "grad_norm": 44.25,
      "learning_rate": 4.2078301869814816e-05,
      "loss": 0.9576,
      "step": 187300
    },
    {
      "epoch": 0.17,
      "grad_norm": 25.5,
      "learning_rate": 4.20738027966239e-05,
      "loss": 0.9147,
      "step": 187400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7265625,
      "learning_rate": 4.206930372343297e-05,
      "loss": 1.1562,
      "step": 187500
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.625,
      "learning_rate": 4.206480465024205e-05,
      "loss": 0.9978,
      "step": 187600
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.5,
      "learning_rate": 4.206030557705113e-05,
      "loss": 1.1307,
      "step": 187700
    },
    {
      "epoch": 0.17,
      "grad_norm": 52.0,
      "learning_rate": 4.2055806503860206e-05,
      "loss": 0.9417,
      "step": 187800
    },
    {
      "epoch": 0.17,
      "grad_norm": 57.75,
      "learning_rate": 4.205130743066928e-05,
      "loss": 0.8818,
      "step": 187900
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.75,
      "learning_rate": 4.2046808357478364e-05,
      "loss": 1.0595,
      "step": 188000
    },
    {
      "epoch": 0.17,
      "grad_norm": 46.75,
      "learning_rate": 4.204230928428744e-05,
      "loss": 1.063,
      "step": 188100
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2119140625,
      "learning_rate": 4.203781021109652e-05,
      "loss": 1.0472,
      "step": 188200
    },
    {
      "epoch": 0.17,
      "grad_norm": 57.5,
      "learning_rate": 4.2033311137905596e-05,
      "loss": 1.0571,
      "step": 188300
    },
    {
      "epoch": 0.17,
      "grad_norm": 76.0,
      "learning_rate": 4.202881206471467e-05,
      "loss": 1.0313,
      "step": 188400
    },
    {
      "epoch": 0.17,
      "grad_norm": 87.5,
      "learning_rate": 4.202431299152375e-05,
      "loss": 1.0989,
      "step": 188500
    },
    {
      "epoch": 0.17,
      "grad_norm": 79.5,
      "learning_rate": 4.201981391833282e-05,
      "loss": 0.9495,
      "step": 188600
    },
    {
      "epoch": 0.17,
      "grad_norm": 266.0,
      "learning_rate": 4.20153148451419e-05,
      "loss": 1.1051,
      "step": 188700
    },
    {
      "epoch": 0.17,
      "grad_norm": 28.75,
      "learning_rate": 4.201081577195098e-05,
      "loss": 1.0755,
      "step": 188800
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.625,
      "learning_rate": 4.2006316698760055e-05,
      "loss": 0.854,
      "step": 188900
    },
    {
      "epoch": 0.17,
      "grad_norm": 23.875,
      "learning_rate": 4.200181762556914e-05,
      "loss": 0.9885,
      "step": 189000
    },
    {
      "epoch": 0.17,
      "grad_norm": 64.0,
      "learning_rate": 4.199731855237821e-05,
      "loss": 0.9234,
      "step": 189100
    },
    {
      "epoch": 0.17,
      "grad_norm": 596.0,
      "learning_rate": 4.199281947918729e-05,
      "loss": 1.0651,
      "step": 189200
    },
    {
      "epoch": 0.17,
      "grad_norm": 24.5,
      "learning_rate": 4.198832040599637e-05,
      "loss": 1.0574,
      "step": 189300
    },
    {
      "epoch": 0.17,
      "grad_norm": 38.75,
      "learning_rate": 4.1983821332805445e-05,
      "loss": 1.0161,
      "step": 189400
    },
    {
      "epoch": 0.17,
      "grad_norm": 32.5,
      "learning_rate": 4.197932225961452e-05,
      "loss": 1.1001,
      "step": 189500
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.6875,
      "learning_rate": 4.19748231864236e-05,
      "loss": 1.0056,
      "step": 189600
    },
    {
      "epoch": 0.17,
      "grad_norm": 58.5,
      "learning_rate": 4.197032411323268e-05,
      "loss": 0.9923,
      "step": 189700
    },
    {
      "epoch": 0.17,
      "grad_norm": 35.5,
      "learning_rate": 4.1965825040041754e-05,
      "loss": 1.0193,
      "step": 189800
    },
    {
      "epoch": 0.17,
      "grad_norm": 27.625,
      "learning_rate": 4.196132596685083e-05,
      "loss": 1.0113,
      "step": 189900
    },
    {
      "epoch": 0.17,
      "grad_norm": 51.0,
      "learning_rate": 4.1956826893659904e-05,
      "loss": 0.9433,
      "step": 190000
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.267578125,
      "learning_rate": 4.1952327820468986e-05,
      "loss": 1.0347,
      "step": 190100
    },
    {
      "epoch": 0.17,
      "grad_norm": 35.0,
      "learning_rate": 4.194782874727806e-05,
      "loss": 0.9434,
      "step": 190200
    },
    {
      "epoch": 0.17,
      "grad_norm": 56.5,
      "learning_rate": 4.194332967408714e-05,
      "loss": 0.9746,
      "step": 190300
    },
    {
      "epoch": 0.17,
      "grad_norm": 81.5,
      "learning_rate": 4.193883060089622e-05,
      "loss": 1.1085,
      "step": 190400
    },
    {
      "epoch": 0.17,
      "grad_norm": 95.5,
      "learning_rate": 4.1934331527705295e-05,
      "loss": 0.9829,
      "step": 190500
    },
    {
      "epoch": 0.17,
      "grad_norm": 35.25,
      "learning_rate": 4.192983245451437e-05,
      "loss": 0.8869,
      "step": 190600
    },
    {
      "epoch": 0.17,
      "grad_norm": 280.0,
      "learning_rate": 4.192533338132345e-05,
      "loss": 1.0473,
      "step": 190700
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.375,
      "learning_rate": 4.192083430813253e-05,
      "loss": 1.0365,
      "step": 190800
    },
    {
      "epoch": 0.17,
      "grad_norm": 360.0,
      "learning_rate": 4.191633523494161e-05,
      "loss": 0.9568,
      "step": 190900
    },
    {
      "epoch": 0.17,
      "grad_norm": 46.25,
      "learning_rate": 4.1911836161750685e-05,
      "loss": 0.9797,
      "step": 191000
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.625,
      "learning_rate": 4.190733708855975e-05,
      "loss": 1.0198,
      "step": 191100
    },
    {
      "epoch": 0.17,
      "grad_norm": 35.0,
      "learning_rate": 4.1902838015368835e-05,
      "loss": 1.1112,
      "step": 191200
    },
    {
      "epoch": 0.17,
      "grad_norm": 84.0,
      "learning_rate": 4.189833894217791e-05,
      "loss": 1.0277,
      "step": 191300
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.6875,
      "learning_rate": 4.1893839868986986e-05,
      "loss": 1.0063,
      "step": 191400
    },
    {
      "epoch": 0.17,
      "grad_norm": 82.0,
      "learning_rate": 4.188934079579607e-05,
      "loss": 1.0716,
      "step": 191500
    },
    {
      "epoch": 0.17,
      "grad_norm": 24.75,
      "learning_rate": 4.1884841722605144e-05,
      "loss": 1.0852,
      "step": 191600
    },
    {
      "epoch": 0.17,
      "grad_norm": 23.5,
      "learning_rate": 4.1880342649414226e-05,
      "loss": 1.105,
      "step": 191700
    },
    {
      "epoch": 0.17,
      "grad_norm": 24.0,
      "learning_rate": 4.18758435762233e-05,
      "loss": 0.8718,
      "step": 191800
    },
    {
      "epoch": 0.17,
      "grad_norm": 42.0,
      "learning_rate": 4.1871344503032376e-05,
      "loss": 0.9035,
      "step": 191900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.0517578125,
      "learning_rate": 4.186684542984146e-05,
      "loss": 0.9971,
      "step": 192000
    },
    {
      "epoch": 0.17,
      "grad_norm": 68.5,
      "learning_rate": 4.1862346356650534e-05,
      "loss": 1.016,
      "step": 192100
    },
    {
      "epoch": 0.17,
      "grad_norm": 22.5,
      "learning_rate": 4.185784728345961e-05,
      "loss": 0.9734,
      "step": 192200
    },
    {
      "epoch": 0.17,
      "grad_norm": 49.0,
      "learning_rate": 4.185334821026869e-05,
      "loss": 1.0572,
      "step": 192300
    },
    {
      "epoch": 0.17,
      "grad_norm": 59.25,
      "learning_rate": 4.184884913707776e-05,
      "loss": 1.0149,
      "step": 192400
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.5,
      "learning_rate": 4.184435006388684e-05,
      "loss": 0.8981,
      "step": 192500
    },
    {
      "epoch": 0.17,
      "grad_norm": 20.625,
      "learning_rate": 4.183985099069592e-05,
      "loss": 0.9419,
      "step": 192600
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.0625,
      "learning_rate": 4.183535191750499e-05,
      "loss": 1.0664,
      "step": 192700
    },
    {
      "epoch": 0.17,
      "grad_norm": 156.0,
      "learning_rate": 4.1830852844314075e-05,
      "loss": 1.0785,
      "step": 192800
    },
    {
      "epoch": 0.17,
      "grad_norm": 28.625,
      "learning_rate": 4.182635377112315e-05,
      "loss": 0.9858,
      "step": 192900
    },
    {
      "epoch": 0.17,
      "grad_norm": 21.5,
      "learning_rate": 4.1821854697932225e-05,
      "loss": 1.1164,
      "step": 193000
    },
    {
      "epoch": 0.17,
      "grad_norm": 52.25,
      "learning_rate": 4.181735562474131e-05,
      "loss": 1.0492,
      "step": 193100
    },
    {
      "epoch": 0.17,
      "grad_norm": 25.375,
      "learning_rate": 4.181285655155038e-05,
      "loss": 0.8838,
      "step": 193200
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.015625,
      "learning_rate": 4.180835747835946e-05,
      "loss": 0.8675,
      "step": 193300
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.75,
      "learning_rate": 4.180385840516854e-05,
      "loss": 1.0667,
      "step": 193400
    },
    {
      "epoch": 0.17,
      "grad_norm": 21.25,
      "learning_rate": 4.1799359331977616e-05,
      "loss": 0.9365,
      "step": 193500
    },
    {
      "epoch": 0.17,
      "grad_norm": 53.75,
      "learning_rate": 4.179486025878669e-05,
      "loss": 1.1836,
      "step": 193600
    },
    {
      "epoch": 0.17,
      "grad_norm": 26.375,
      "learning_rate": 4.1790361185595766e-05,
      "loss": 1.0216,
      "step": 193700
    },
    {
      "epoch": 0.17,
      "grad_norm": 41.25,
      "learning_rate": 4.178586211240484e-05,
      "loss": 1.0362,
      "step": 193800
    },
    {
      "epoch": 0.17,
      "grad_norm": 70.5,
      "learning_rate": 4.1781363039213924e-05,
      "loss": 0.9231,
      "step": 193900
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.8125,
      "learning_rate": 4.1776863966023e-05,
      "loss": 0.9694,
      "step": 194000
    },
    {
      "epoch": 0.17,
      "grad_norm": 39.5,
      "learning_rate": 4.1772364892832075e-05,
      "loss": 0.9674,
      "step": 194100
    },
    {
      "epoch": 0.17,
      "grad_norm": 77.5,
      "learning_rate": 4.176786581964116e-05,
      "loss": 0.9426,
      "step": 194200
    },
    {
      "epoch": 0.17,
      "grad_norm": 52.25,
      "learning_rate": 4.176336674645023e-05,
      "loss": 0.9666,
      "step": 194300
    },
    {
      "epoch": 0.17,
      "grad_norm": 47.0,
      "learning_rate": 4.1758867673259314e-05,
      "loss": 0.9828,
      "step": 194400
    },
    {
      "epoch": 0.17,
      "grad_norm": 177.0,
      "learning_rate": 4.175436860006839e-05,
      "loss": 1.0319,
      "step": 194500
    },
    {
      "epoch": 0.17,
      "grad_norm": 44.0,
      "learning_rate": 4.1749869526877465e-05,
      "loss": 1.0749,
      "step": 194600
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.6875,
      "learning_rate": 4.174537045368655e-05,
      "loss": 1.0851,
      "step": 194700
    },
    {
      "epoch": 0.17,
      "grad_norm": 33.75,
      "learning_rate": 4.174087138049562e-05,
      "loss": 1.0028,
      "step": 194800
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.75,
      "learning_rate": 4.17363723073047e-05,
      "loss": 1.1624,
      "step": 194900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.703125,
      "learning_rate": 4.173187323411377e-05,
      "loss": 0.8968,
      "step": 195000
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.375,
      "learning_rate": 4.172737416092285e-05,
      "loss": 1.0454,
      "step": 195100
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.625,
      "learning_rate": 4.172287508773193e-05,
      "loss": 1.0142,
      "step": 195200
    },
    {
      "epoch": 0.17,
      "grad_norm": 29.25,
      "learning_rate": 4.1718376014541006e-05,
      "loss": 0.8795,
      "step": 195300
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.0,
      "learning_rate": 4.171387694135008e-05,
      "loss": 0.967,
      "step": 195400
    },
    {
      "epoch": 0.17,
      "grad_norm": 25.375,
      "learning_rate": 4.170937786815916e-05,
      "loss": 0.9382,
      "step": 195500
    },
    {
      "epoch": 0.17,
      "grad_norm": 240.0,
      "learning_rate": 4.170487879496824e-05,
      "loss": 0.9448,
      "step": 195600
    },
    {
      "epoch": 0.17,
      "grad_norm": 46.5,
      "learning_rate": 4.1700379721777314e-05,
      "loss": 0.9801,
      "step": 195700
    },
    {
      "epoch": 0.17,
      "grad_norm": 25.25,
      "learning_rate": 4.1695880648586396e-05,
      "loss": 0.8991,
      "step": 195800
    },
    {
      "epoch": 0.17,
      "grad_norm": 69.5,
      "learning_rate": 4.169138157539547e-05,
      "loss": 1.0617,
      "step": 195900
    },
    {
      "epoch": 0.17,
      "grad_norm": 94.5,
      "learning_rate": 4.168688250220455e-05,
      "loss": 1.0327,
      "step": 196000
    },
    {
      "epoch": 0.17,
      "grad_norm": 134.0,
      "learning_rate": 4.168238342901363e-05,
      "loss": 1.0698,
      "step": 196100
    },
    {
      "epoch": 0.17,
      "grad_norm": 31.125,
      "learning_rate": 4.1677884355822704e-05,
      "loss": 1.0747,
      "step": 196200
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.01495361328125,
      "learning_rate": 4.167338528263178e-05,
      "loss": 1.0298,
      "step": 196300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1298828125,
      "learning_rate": 4.1668886209440855e-05,
      "loss": 1.1018,
      "step": 196400
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.75,
      "learning_rate": 4.166438713624993e-05,
      "loss": 0.9481,
      "step": 196500
    },
    {
      "epoch": 0.18,
      "grad_norm": 73.0,
      "learning_rate": 4.165988806305901e-05,
      "loss": 1.054,
      "step": 196600
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5546875,
      "learning_rate": 4.165538898986809e-05,
      "loss": 1.0025,
      "step": 196700
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.5,
      "learning_rate": 4.165088991667716e-05,
      "loss": 1.0205,
      "step": 196800
    },
    {
      "epoch": 0.18,
      "grad_norm": 30.25,
      "learning_rate": 4.1646390843486245e-05,
      "loss": 1.108,
      "step": 196900
    },
    {
      "epoch": 0.18,
      "grad_norm": 32.0,
      "learning_rate": 4.164189177029532e-05,
      "loss": 1.0031,
      "step": 197000
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.921875,
      "learning_rate": 4.16373926971044e-05,
      "loss": 0.8931,
      "step": 197100
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.375,
      "learning_rate": 4.163289362391348e-05,
      "loss": 0.937,
      "step": 197200
    },
    {
      "epoch": 0.18,
      "grad_norm": 219.0,
      "learning_rate": 4.162839455072255e-05,
      "loss": 1.0134,
      "step": 197300
    },
    {
      "epoch": 0.18,
      "grad_norm": 61.75,
      "learning_rate": 4.1623895477531635e-05,
      "loss": 1.0368,
      "step": 197400
    },
    {
      "epoch": 0.18,
      "grad_norm": 240.0,
      "learning_rate": 4.161939640434071e-05,
      "loss": 1.1006,
      "step": 197500
    },
    {
      "epoch": 0.18,
      "grad_norm": 42.0,
      "learning_rate": 4.161489733114978e-05,
      "loss": 0.9246,
      "step": 197600
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1796875,
      "learning_rate": 4.161039825795886e-05,
      "loss": 0.9707,
      "step": 197700
    },
    {
      "epoch": 0.18,
      "grad_norm": 25.375,
      "learning_rate": 4.160589918476794e-05,
      "loss": 1.123,
      "step": 197800
    },
    {
      "epoch": 0.18,
      "grad_norm": 68.5,
      "learning_rate": 4.160140011157702e-05,
      "loss": 1.0692,
      "step": 197900
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.0,
      "learning_rate": 4.1596901038386094e-05,
      "loss": 0.9804,
      "step": 198000
    },
    {
      "epoch": 0.18,
      "grad_norm": 41.75,
      "learning_rate": 4.159240196519517e-05,
      "loss": 1.0816,
      "step": 198100
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.91015625,
      "learning_rate": 4.158790289200425e-05,
      "loss": 1.0439,
      "step": 198200
    },
    {
      "epoch": 0.18,
      "grad_norm": 43.5,
      "learning_rate": 4.158340381881333e-05,
      "loss": 0.9404,
      "step": 198300
    },
    {
      "epoch": 0.18,
      "grad_norm": 52.5,
      "learning_rate": 4.15789047456224e-05,
      "loss": 1.0073,
      "step": 198400
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.6875,
      "learning_rate": 4.1574405672431484e-05,
      "loss": 1.1357,
      "step": 198500
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.00360107421875,
      "learning_rate": 4.156990659924056e-05,
      "loss": 1.0117,
      "step": 198600
    },
    {
      "epoch": 0.18,
      "grad_norm": 36.75,
      "learning_rate": 4.1565407526049635e-05,
      "loss": 1.038,
      "step": 198700
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.375,
      "learning_rate": 4.156090845285872e-05,
      "loss": 1.1316,
      "step": 198800
    },
    {
      "epoch": 0.18,
      "grad_norm": 227.0,
      "learning_rate": 4.1556409379667786e-05,
      "loss": 0.9967,
      "step": 198900
    },
    {
      "epoch": 0.18,
      "grad_norm": 27.0,
      "learning_rate": 4.155191030647687e-05,
      "loss": 0.9362,
      "step": 199000
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.75,
      "learning_rate": 4.154741123328594e-05,
      "loss": 1.0924,
      "step": 199100
    },
    {
      "epoch": 0.18,
      "grad_norm": 36.5,
      "learning_rate": 4.154291216009502e-05,
      "loss": 1.1432,
      "step": 199200
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0390625,
      "learning_rate": 4.15384130869041e-05,
      "loss": 0.9493,
      "step": 199300
    },
    {
      "epoch": 0.18,
      "grad_norm": 57.25,
      "learning_rate": 4.1533914013713176e-05,
      "loss": 0.9189,
      "step": 199400
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.5,
      "learning_rate": 4.152941494052225e-05,
      "loss": 0.9244,
      "step": 199500
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.0,
      "learning_rate": 4.1524915867331333e-05,
      "loss": 0.9848,
      "step": 199600
    },
    {
      "epoch": 0.18,
      "grad_norm": 57.75,
      "learning_rate": 4.152041679414041e-05,
      "loss": 0.9719,
      "step": 199700
    },
    {
      "epoch": 0.18,
      "grad_norm": 27.625,
      "learning_rate": 4.151591772094949e-05,
      "loss": 0.9785,
      "step": 199800
    },
    {
      "epoch": 0.18,
      "grad_norm": 136.0,
      "learning_rate": 4.1511418647758566e-05,
      "loss": 1.0904,
      "step": 199900
    },
    {
      "epoch": 0.18,
      "grad_norm": 30.0,
      "learning_rate": 4.150691957456764e-05,
      "loss": 0.952,
      "step": 200000
    },
    {
      "epoch": 0.18,
      "grad_norm": 137.0,
      "learning_rate": 4.1502420501376724e-05,
      "loss": 0.9827,
      "step": 200100
    },
    {
      "epoch": 0.18,
      "grad_norm": 44.5,
      "learning_rate": 4.149792142818579e-05,
      "loss": 1.0779,
      "step": 200200
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.4375,
      "learning_rate": 4.149342235499487e-05,
      "loss": 1.0018,
      "step": 200300
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1005859375,
      "learning_rate": 4.148892328180395e-05,
      "loss": 1.0206,
      "step": 200400
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.25,
      "learning_rate": 4.1484424208613025e-05,
      "loss": 1.2204,
      "step": 200500
    },
    {
      "epoch": 0.18,
      "grad_norm": 44.75,
      "learning_rate": 4.147992513542211e-05,
      "loss": 1.0894,
      "step": 200600
    },
    {
      "epoch": 0.18,
      "grad_norm": 91.0,
      "learning_rate": 4.147542606223118e-05,
      "loss": 1.1154,
      "step": 200700
    },
    {
      "epoch": 0.18,
      "grad_norm": 38.0,
      "learning_rate": 4.147092698904026e-05,
      "loss": 0.9523,
      "step": 200800
    },
    {
      "epoch": 0.18,
      "grad_norm": 43.25,
      "learning_rate": 4.146642791584934e-05,
      "loss": 1.0039,
      "step": 200900
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.0,
      "learning_rate": 4.1461928842658415e-05,
      "loss": 1.0288,
      "step": 201000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.025146484375,
      "learning_rate": 4.145742976946749e-05,
      "loss": 0.9594,
      "step": 201100
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.75,
      "learning_rate": 4.145293069627657e-05,
      "loss": 1.0622,
      "step": 201200
    },
    {
      "epoch": 0.18,
      "grad_norm": 58.0,
      "learning_rate": 4.144843162308565e-05,
      "loss": 1.0108,
      "step": 201300
    },
    {
      "epoch": 0.18,
      "grad_norm": 39.25,
      "learning_rate": 4.1443932549894723e-05,
      "loss": 1.1088,
      "step": 201400
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.25,
      "learning_rate": 4.14394334767038e-05,
      "loss": 1.0485,
      "step": 201500
    },
    {
      "epoch": 0.18,
      "grad_norm": 81.0,
      "learning_rate": 4.1434934403512874e-05,
      "loss": 1.0323,
      "step": 201600
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.375,
      "learning_rate": 4.1430435330321956e-05,
      "loss": 1.0112,
      "step": 201700
    },
    {
      "epoch": 0.18,
      "grad_norm": 25.25,
      "learning_rate": 4.142593625713103e-05,
      "loss": 1.0242,
      "step": 201800
    },
    {
      "epoch": 0.18,
      "grad_norm": 272.0,
      "learning_rate": 4.142143718394011e-05,
      "loss": 1.0801,
      "step": 201900
    },
    {
      "epoch": 0.18,
      "grad_norm": 69.0,
      "learning_rate": 4.141693811074919e-05,
      "loss": 0.8958,
      "step": 202000
    },
    {
      "epoch": 0.18,
      "grad_norm": 34.25,
      "learning_rate": 4.1412439037558264e-05,
      "loss": 1.025,
      "step": 202100
    },
    {
      "epoch": 0.18,
      "grad_norm": 47.0,
      "learning_rate": 4.140793996436734e-05,
      "loss": 1.058,
      "step": 202200
    },
    {
      "epoch": 0.18,
      "grad_norm": 122.5,
      "learning_rate": 4.140344089117642e-05,
      "loss": 0.9138,
      "step": 202300
    },
    {
      "epoch": 0.18,
      "grad_norm": 53.0,
      "learning_rate": 4.13989418179855e-05,
      "loss": 1.0506,
      "step": 202400
    },
    {
      "epoch": 0.18,
      "grad_norm": 50.25,
      "learning_rate": 4.139444274479458e-05,
      "loss": 0.8935,
      "step": 202500
    },
    {
      "epoch": 0.18,
      "grad_norm": 33.5,
      "learning_rate": 4.1389943671603655e-05,
      "loss": 1.0334,
      "step": 202600
    },
    {
      "epoch": 0.18,
      "grad_norm": 40.0,
      "learning_rate": 4.138544459841273e-05,
      "loss": 0.9227,
      "step": 202700
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.25,
      "learning_rate": 4.1380945525221805e-05,
      "loss": 0.9657,
      "step": 202800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.578125,
      "learning_rate": 4.137644645203088e-05,
      "loss": 1.0117,
      "step": 202900
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.00537109375,
      "learning_rate": 4.1371947378839956e-05,
      "loss": 1.1096,
      "step": 203000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.37890625,
      "learning_rate": 4.136744830564904e-05,
      "loss": 0.9909,
      "step": 203100
    },
    {
      "epoch": 0.18,
      "grad_norm": 42.75,
      "learning_rate": 4.1362949232458113e-05,
      "loss": 1.0561,
      "step": 203200
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.25,
      "learning_rate": 4.1358450159267196e-05,
      "loss": 1.0458,
      "step": 203300
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.0,
      "learning_rate": 4.135395108607627e-05,
      "loss": 1.0198,
      "step": 203400
    },
    {
      "epoch": 0.18,
      "grad_norm": 57.5,
      "learning_rate": 4.1349452012885346e-05,
      "loss": 1.0858,
      "step": 203500
    },
    {
      "epoch": 0.18,
      "grad_norm": 43.0,
      "learning_rate": 4.134495293969443e-05,
      "loss": 1.0132,
      "step": 203600
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4375,
      "learning_rate": 4.1340453866503504e-05,
      "loss": 1.0476,
      "step": 203700
    },
    {
      "epoch": 0.18,
      "grad_norm": 31.25,
      "learning_rate": 4.133595479331258e-05,
      "loss": 1.0296,
      "step": 203800
    },
    {
      "epoch": 0.18,
      "grad_norm": 207.0,
      "learning_rate": 4.133145572012166e-05,
      "loss": 1.0641,
      "step": 203900
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.244140625,
      "learning_rate": 4.1326956646930736e-05,
      "loss": 0.934,
      "step": 204000
    },
    {
      "epoch": 0.18,
      "grad_norm": 32.0,
      "learning_rate": 4.132245757373981e-05,
      "loss": 1.0494,
      "step": 204100
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.375,
      "learning_rate": 4.131795850054889e-05,
      "loss": 0.9449,
      "step": 204200
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.375,
      "learning_rate": 4.131345942735796e-05,
      "loss": 1.1892,
      "step": 204300
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.125,
      "learning_rate": 4.1308960354167045e-05,
      "loss": 0.9864,
      "step": 204400
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.0625,
      "learning_rate": 4.130446128097612e-05,
      "loss": 1.0519,
      "step": 204500
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.0,
      "learning_rate": 4.1299962207785195e-05,
      "loss": 0.8999,
      "step": 204600
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.6875,
      "learning_rate": 4.129546313459428e-05,
      "loss": 1.1317,
      "step": 204700
    },
    {
      "epoch": 0.18,
      "grad_norm": 113.5,
      "learning_rate": 4.129096406140335e-05,
      "loss": 1.0446,
      "step": 204800
    },
    {
      "epoch": 0.18,
      "grad_norm": 39.0,
      "learning_rate": 4.128646498821243e-05,
      "loss": 1.0604,
      "step": 204900
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.375,
      "learning_rate": 4.128196591502151e-05,
      "loss": 1.0099,
      "step": 205000
    },
    {
      "epoch": 0.18,
      "grad_norm": 34.0,
      "learning_rate": 4.1277466841830586e-05,
      "loss": 0.9905,
      "step": 205100
    },
    {
      "epoch": 0.18,
      "grad_norm": 45.5,
      "learning_rate": 4.127296776863967e-05,
      "loss": 0.8917,
      "step": 205200
    },
    {
      "epoch": 0.18,
      "grad_norm": 66.0,
      "learning_rate": 4.126846869544874e-05,
      "loss": 1.2461,
      "step": 205300
    },
    {
      "epoch": 0.18,
      "grad_norm": 37.0,
      "learning_rate": 4.126396962225781e-05,
      "loss": 1.0329,
      "step": 205400
    },
    {
      "epoch": 0.18,
      "grad_norm": 38.25,
      "learning_rate": 4.1259470549066894e-05,
      "loss": 1.0766,
      "step": 205500
    },
    {
      "epoch": 0.18,
      "grad_norm": 28.25,
      "learning_rate": 4.125497147587597e-05,
      "loss": 1.0072,
      "step": 205600
    },
    {
      "epoch": 0.18,
      "grad_norm": 92.5,
      "learning_rate": 4.1250472402685044e-05,
      "loss": 1.122,
      "step": 205700
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.25,
      "learning_rate": 4.1245973329494126e-05,
      "loss": 0.9677,
      "step": 205800
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.375,
      "learning_rate": 4.12414742563032e-05,
      "loss": 0.9124,
      "step": 205900
    },
    {
      "epoch": 0.18,
      "grad_norm": 25.0,
      "learning_rate": 4.1236975183112284e-05,
      "loss": 1.0021,
      "step": 206000
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.75,
      "learning_rate": 4.123247610992136e-05,
      "loss": 0.9075,
      "step": 206100
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.25,
      "learning_rate": 4.1227977036730435e-05,
      "loss": 0.9195,
      "step": 206200
    },
    {
      "epoch": 0.18,
      "grad_norm": 31.5,
      "learning_rate": 4.122347796353952e-05,
      "loss": 1.0138,
      "step": 206300
    },
    {
      "epoch": 0.18,
      "grad_norm": 68.5,
      "learning_rate": 4.121897889034859e-05,
      "loss": 1.0014,
      "step": 206400
    },
    {
      "epoch": 0.18,
      "grad_norm": 33.75,
      "learning_rate": 4.121447981715767e-05,
      "loss": 0.9441,
      "step": 206500
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.125,
      "learning_rate": 4.120998074396675e-05,
      "loss": 0.9736,
      "step": 206600
    },
    {
      "epoch": 0.18,
      "grad_norm": 46.25,
      "learning_rate": 4.120548167077582e-05,
      "loss": 0.9764,
      "step": 206700
    },
    {
      "epoch": 0.18,
      "grad_norm": 45.5,
      "learning_rate": 4.12009825975849e-05,
      "loss": 1.0639,
      "step": 206800
    },
    {
      "epoch": 0.18,
      "grad_norm": 83.0,
      "learning_rate": 4.1196483524393976e-05,
      "loss": 1.1208,
      "step": 206900
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.0,
      "learning_rate": 4.119198445120305e-05,
      "loss": 0.9852,
      "step": 207000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.0032501220703125,
      "learning_rate": 4.118748537801213e-05,
      "loss": 0.9806,
      "step": 207100
    },
    {
      "epoch": 0.18,
      "grad_norm": 158.0,
      "learning_rate": 4.118298630482121e-05,
      "loss": 1.0451,
      "step": 207200
    },
    {
      "epoch": 0.18,
      "grad_norm": 27.625,
      "learning_rate": 4.1178487231630284e-05,
      "loss": 0.9113,
      "step": 207300
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.375,
      "learning_rate": 4.1173988158439366e-05,
      "loss": 0.9102,
      "step": 207400
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.116948908524844e-05,
      "loss": 1.0888,
      "step": 207500
    },
    {
      "epoch": 0.18,
      "grad_norm": 45.25,
      "learning_rate": 4.1164990012057516e-05,
      "loss": 0.9614,
      "step": 207600
    },
    {
      "epoch": 0.19,
      "grad_norm": 91.5,
      "learning_rate": 4.11604909388666e-05,
      "loss": 1.0828,
      "step": 207700
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.125,
      "learning_rate": 4.1155991865675674e-05,
      "loss": 0.975,
      "step": 207800
    },
    {
      "epoch": 0.19,
      "grad_norm": 40.25,
      "learning_rate": 4.1151492792484756e-05,
      "loss": 1.1039,
      "step": 207900
    },
    {
      "epoch": 0.19,
      "grad_norm": 41.0,
      "learning_rate": 4.1146993719293825e-05,
      "loss": 1.0925,
      "step": 208000
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.234375,
      "learning_rate": 4.11424946461029e-05,
      "loss": 1.0953,
      "step": 208100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.0021209716796875,
      "learning_rate": 4.113799557291198e-05,
      "loss": 1.0687,
      "step": 208200
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.09375,
      "learning_rate": 4.113349649972106e-05,
      "loss": 0.8731,
      "step": 208300
    },
    {
      "epoch": 0.19,
      "grad_norm": 53.5,
      "learning_rate": 4.112899742653013e-05,
      "loss": 1.0374,
      "step": 208400
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.625,
      "learning_rate": 4.1124498353339215e-05,
      "loss": 1.2792,
      "step": 208500
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.875,
      "learning_rate": 4.111999928014829e-05,
      "loss": 0.9054,
      "step": 208600
    },
    {
      "epoch": 0.19,
      "grad_norm": 33.0,
      "learning_rate": 4.111550020695737e-05,
      "loss": 1.1398,
      "step": 208700
    },
    {
      "epoch": 0.19,
      "grad_norm": 23.0,
      "learning_rate": 4.111100113376645e-05,
      "loss": 1.0491,
      "step": 208800
    },
    {
      "epoch": 0.19,
      "grad_norm": 54.0,
      "learning_rate": 4.110650206057552e-05,
      "loss": 0.9972,
      "step": 208900
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.875,
      "learning_rate": 4.1102002987384605e-05,
      "loss": 0.9967,
      "step": 209000
    },
    {
      "epoch": 0.19,
      "grad_norm": 24.25,
      "learning_rate": 4.109750391419368e-05,
      "loss": 1.0893,
      "step": 209100
    },
    {
      "epoch": 0.19,
      "grad_norm": 117.5,
      "learning_rate": 4.1093004841002756e-05,
      "loss": 1.0223,
      "step": 209200
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.00634765625,
      "learning_rate": 4.108850576781183e-05,
      "loss": 1.0812,
      "step": 209300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.012939453125,
      "learning_rate": 4.1084006694620906e-05,
      "loss": 0.9931,
      "step": 209400
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.625,
      "learning_rate": 4.107950762142999e-05,
      "loss": 0.9572,
      "step": 209500
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.3125,
      "learning_rate": 4.1075008548239064e-05,
      "loss": 1.0195,
      "step": 209600
    },
    {
      "epoch": 0.19,
      "grad_norm": 104.5,
      "learning_rate": 4.107050947504814e-05,
      "loss": 0.9802,
      "step": 209700
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.5,
      "learning_rate": 4.106601040185722e-05,
      "loss": 1.0643,
      "step": 209800
    },
    {
      "epoch": 0.19,
      "grad_norm": 84.0,
      "learning_rate": 4.10615113286663e-05,
      "loss": 0.8721,
      "step": 209900
    },
    {
      "epoch": 0.19,
      "grad_norm": 27.375,
      "learning_rate": 4.105701225547537e-05,
      "loss": 0.9599,
      "step": 210000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.037841796875,
      "learning_rate": 4.1052513182284454e-05,
      "loss": 0.9251,
      "step": 210100
    },
    {
      "epoch": 0.19,
      "grad_norm": 25.25,
      "learning_rate": 4.104801410909353e-05,
      "loss": 1.015,
      "step": 210200
    },
    {
      "epoch": 0.19,
      "grad_norm": 100.5,
      "learning_rate": 4.1043515035902605e-05,
      "loss": 1.0445,
      "step": 210300
    },
    {
      "epoch": 0.19,
      "grad_norm": 75.0,
      "learning_rate": 4.103901596271169e-05,
      "loss": 0.9973,
      "step": 210400
    },
    {
      "epoch": 0.19,
      "grad_norm": 249.0,
      "learning_rate": 4.103451688952076e-05,
      "loss": 0.9951,
      "step": 210500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.201171875,
      "learning_rate": 4.103001781632984e-05,
      "loss": 1.1118,
      "step": 210600
    },
    {
      "epoch": 0.19,
      "grad_norm": 60.75,
      "learning_rate": 4.102551874313891e-05,
      "loss": 0.9056,
      "step": 210700
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.0,
      "learning_rate": 4.102101966994799e-05,
      "loss": 1.0998,
      "step": 210800
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.625,
      "learning_rate": 4.101652059675707e-05,
      "loss": 1.0499,
      "step": 210900
    },
    {
      "epoch": 0.19,
      "grad_norm": 25.5,
      "learning_rate": 4.1012021523566146e-05,
      "loss": 0.9896,
      "step": 211000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.002227783203125,
      "learning_rate": 4.100752245037522e-05,
      "loss": 0.9199,
      "step": 211100
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.6875,
      "learning_rate": 4.10030233771843e-05,
      "loss": 0.9904,
      "step": 211200
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.8125,
      "learning_rate": 4.099852430399338e-05,
      "loss": 1.1601,
      "step": 211300
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.5,
      "learning_rate": 4.099402523080246e-05,
      "loss": 1.0315,
      "step": 211400
    },
    {
      "epoch": 0.19,
      "grad_norm": 43.75,
      "learning_rate": 4.0989526157611536e-05,
      "loss": 0.9983,
      "step": 211500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.36328125,
      "learning_rate": 4.098502708442061e-05,
      "loss": 1.0844,
      "step": 211600
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.25,
      "learning_rate": 4.0980528011229693e-05,
      "loss": 1.0916,
      "step": 211700
    },
    {
      "epoch": 0.19,
      "grad_norm": 29.0,
      "learning_rate": 4.097602893803877e-05,
      "loss": 1.0555,
      "step": 211800
    },
    {
      "epoch": 0.19,
      "grad_norm": 27.875,
      "learning_rate": 4.097152986484784e-05,
      "loss": 1.0497,
      "step": 211900
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.625,
      "learning_rate": 4.096703079165692e-05,
      "loss": 0.8473,
      "step": 212000
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.125,
      "learning_rate": 4.0962531718465995e-05,
      "loss": 1.0515,
      "step": 212100
    },
    {
      "epoch": 0.19,
      "grad_norm": 35.5,
      "learning_rate": 4.095803264527508e-05,
      "loss": 0.965,
      "step": 212200
    },
    {
      "epoch": 0.19,
      "grad_norm": 50.5,
      "learning_rate": 4.095353357208415e-05,
      "loss": 1.0608,
      "step": 212300
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.9375,
      "learning_rate": 4.094903449889323e-05,
      "loss": 1.0581,
      "step": 212400
    },
    {
      "epoch": 0.19,
      "grad_norm": 61.5,
      "learning_rate": 4.094453542570231e-05,
      "loss": 0.973,
      "step": 212500
    },
    {
      "epoch": 0.19,
      "grad_norm": 26.5,
      "learning_rate": 4.0940036352511385e-05,
      "loss": 0.866,
      "step": 212600
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.0,
      "learning_rate": 4.093553727932046e-05,
      "loss": 1.1619,
      "step": 212700
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.5,
      "learning_rate": 4.093103820612954e-05,
      "loss": 1.0849,
      "step": 212800
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.875,
      "learning_rate": 4.092653913293862e-05,
      "loss": 0.8512,
      "step": 212900
    },
    {
      "epoch": 0.19,
      "grad_norm": 984.0,
      "learning_rate": 4.092204005974769e-05,
      "loss": 1.0211,
      "step": 213000
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.875,
      "learning_rate": 4.0917540986556775e-05,
      "loss": 0.996,
      "step": 213100
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.5,
      "learning_rate": 4.0913041913365844e-05,
      "loss": 1.1139,
      "step": 213200
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.125,
      "learning_rate": 4.0908542840174926e-05,
      "loss": 1.0224,
      "step": 213300
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.484375,
      "learning_rate": 4.0904043766984e-05,
      "loss": 1.0536,
      "step": 213400
    },
    {
      "epoch": 0.19,
      "grad_norm": 60.25,
      "learning_rate": 4.089954469379308e-05,
      "loss": 1.1269,
      "step": 213500
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.3125,
      "learning_rate": 4.089504562060216e-05,
      "loss": 1.1374,
      "step": 213600
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.5625,
      "learning_rate": 4.0890546547411234e-05,
      "loss": 1.0655,
      "step": 213700
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.15625,
      "learning_rate": 4.088604747422031e-05,
      "loss": 1.0169,
      "step": 213800
    },
    {
      "epoch": 0.19,
      "grad_norm": 154.0,
      "learning_rate": 4.088154840102939e-05,
      "loss": 0.9249,
      "step": 213900
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12255859375,
      "learning_rate": 4.087704932783847e-05,
      "loss": 0.9403,
      "step": 214000
    },
    {
      "epoch": 0.19,
      "grad_norm": 29.25,
      "learning_rate": 4.087255025464755e-05,
      "loss": 0.9346,
      "step": 214100
    },
    {
      "epoch": 0.19,
      "grad_norm": 175.0,
      "learning_rate": 4.0868051181456624e-05,
      "loss": 0.8497,
      "step": 214200
    },
    {
      "epoch": 0.19,
      "grad_norm": 66.5,
      "learning_rate": 4.08635521082657e-05,
      "loss": 1.0652,
      "step": 214300
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.25,
      "learning_rate": 4.085905303507478e-05,
      "loss": 0.9883,
      "step": 214400
    },
    {
      "epoch": 0.19,
      "grad_norm": 27.75,
      "learning_rate": 4.085455396188385e-05,
      "loss": 0.8639,
      "step": 214500
    },
    {
      "epoch": 0.19,
      "grad_norm": 41.5,
      "learning_rate": 4.0850054888692926e-05,
      "loss": 1.0497,
      "step": 214600
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.25,
      "learning_rate": 4.084555581550201e-05,
      "loss": 1.037,
      "step": 214700
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.25,
      "learning_rate": 4.084105674231108e-05,
      "loss": 0.9879,
      "step": 214800
    },
    {
      "epoch": 0.19,
      "grad_norm": 43.0,
      "learning_rate": 4.0836557669120165e-05,
      "loss": 0.9982,
      "step": 214900
    },
    {
      "epoch": 0.19,
      "grad_norm": 39.75,
      "learning_rate": 4.083205859592924e-05,
      "loss": 1.0842,
      "step": 215000
    },
    {
      "epoch": 0.19,
      "grad_norm": 34.0,
      "learning_rate": 4.0827559522738316e-05,
      "loss": 0.9924,
      "step": 215100
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.75,
      "learning_rate": 4.08230604495474e-05,
      "loss": 1.0327,
      "step": 215200
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.0625,
      "learning_rate": 4.0818561376356473e-05,
      "loss": 1.0646,
      "step": 215300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.02099609375,
      "learning_rate": 4.081406230316555e-05,
      "loss": 0.9434,
      "step": 215400
    },
    {
      "epoch": 0.19,
      "grad_norm": 177.0,
      "learning_rate": 4.080956322997463e-05,
      "loss": 1.0415,
      "step": 215500
    },
    {
      "epoch": 0.19,
      "grad_norm": 94.5,
      "learning_rate": 4.0805064156783706e-05,
      "loss": 0.9127,
      "step": 215600
    },
    {
      "epoch": 0.19,
      "grad_norm": 44.0,
      "learning_rate": 4.080056508359278e-05,
      "loss": 1.0076,
      "step": 215700
    },
    {
      "epoch": 0.19,
      "grad_norm": 34.5,
      "learning_rate": 4.079606601040186e-05,
      "loss": 1.0544,
      "step": 215800
    },
    {
      "epoch": 0.19,
      "grad_norm": 44.75,
      "learning_rate": 4.079156693721093e-05,
      "loss": 0.9754,
      "step": 215900
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.5,
      "learning_rate": 4.0787067864020014e-05,
      "loss": 0.9611,
      "step": 216000
    },
    {
      "epoch": 0.19,
      "grad_norm": 67.0,
      "learning_rate": 4.078256879082909e-05,
      "loss": 1.1344,
      "step": 216100
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.625,
      "learning_rate": 4.0778069717638165e-05,
      "loss": 0.921,
      "step": 216200
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.21875,
      "learning_rate": 4.077357064444725e-05,
      "loss": 1.1117,
      "step": 216300
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.5,
      "learning_rate": 4.076907157125632e-05,
      "loss": 1.0584,
      "step": 216400
    },
    {
      "epoch": 0.19,
      "grad_norm": 28.875,
      "learning_rate": 4.07645724980654e-05,
      "loss": 0.8819,
      "step": 216500
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.3125,
      "learning_rate": 4.076007342487448e-05,
      "loss": 1.1318,
      "step": 216600
    },
    {
      "epoch": 0.19,
      "grad_norm": 106.5,
      "learning_rate": 4.0755574351683555e-05,
      "loss": 1.033,
      "step": 216700
    },
    {
      "epoch": 0.19,
      "grad_norm": 67.0,
      "learning_rate": 4.075107527849264e-05,
      "loss": 0.989,
      "step": 216800
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.5,
      "learning_rate": 4.074657620530171e-05,
      "loss": 0.9674,
      "step": 216900
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.0,
      "learning_rate": 4.074207713211079e-05,
      "loss": 1.0205,
      "step": 217000
    },
    {
      "epoch": 0.19,
      "grad_norm": 56.5,
      "learning_rate": 4.0737578058919863e-05,
      "loss": 1.0161,
      "step": 217100
    },
    {
      "epoch": 0.19,
      "grad_norm": 41.25,
      "learning_rate": 4.073307898572894e-05,
      "loss": 1.0059,
      "step": 217200
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.0,
      "learning_rate": 4.0728579912538014e-05,
      "loss": 1.0873,
      "step": 217300
    },
    {
      "epoch": 0.19,
      "grad_norm": 99.0,
      "learning_rate": 4.0724080839347096e-05,
      "loss": 1.0004,
      "step": 217400
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.0625,
      "learning_rate": 4.071958176615617e-05,
      "loss": 1.08,
      "step": 217500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.388671875,
      "learning_rate": 4.0715082692965254e-05,
      "loss": 0.8686,
      "step": 217600
    },
    {
      "epoch": 0.19,
      "grad_norm": 99.5,
      "learning_rate": 4.071058361977433e-05,
      "loss": 0.9807,
      "step": 217700
    },
    {
      "epoch": 0.19,
      "grad_norm": 76.5,
      "learning_rate": 4.0706084546583404e-05,
      "loss": 1.188,
      "step": 217800
    },
    {
      "epoch": 0.19,
      "grad_norm": 40.0,
      "learning_rate": 4.0701585473392486e-05,
      "loss": 1.0733,
      "step": 217900
    },
    {
      "epoch": 0.19,
      "grad_norm": 29.625,
      "learning_rate": 4.069708640020156e-05,
      "loss": 1.0241,
      "step": 218000
    },
    {
      "epoch": 0.19,
      "grad_norm": 53.75,
      "learning_rate": 4.069258732701064e-05,
      "loss": 1.0858,
      "step": 218100
    },
    {
      "epoch": 0.19,
      "grad_norm": 154.0,
      "learning_rate": 4.068808825381972e-05,
      "loss": 0.8302,
      "step": 218200
    },
    {
      "epoch": 0.19,
      "grad_norm": 270.0,
      "learning_rate": 4.0683589180628795e-05,
      "loss": 1.0766,
      "step": 218300
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.75,
      "learning_rate": 4.067909010743787e-05,
      "loss": 1.0029,
      "step": 218400
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.453125,
      "learning_rate": 4.0674591034246945e-05,
      "loss": 1.0616,
      "step": 218500
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.21875,
      "learning_rate": 4.067009196105602e-05,
      "loss": 1.1286,
      "step": 218600
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.390625,
      "learning_rate": 4.06655928878651e-05,
      "loss": 1.0094,
      "step": 218700
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.0283203125,
      "learning_rate": 4.066109381467418e-05,
      "loss": 1.0558,
      "step": 218800
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.46875,
      "learning_rate": 4.0656594741483253e-05,
      "loss": 0.9468,
      "step": 218900
    },
    {
      "epoch": 0.2,
      "grad_norm": 27.25,
      "learning_rate": 4.0652095668292336e-05,
      "loss": 0.9764,
      "step": 219000
    },
    {
      "epoch": 0.2,
      "grad_norm": 61.25,
      "learning_rate": 4.064759659510141e-05,
      "loss": 1.0133,
      "step": 219100
    },
    {
      "epoch": 0.2,
      "grad_norm": 67.5,
      "learning_rate": 4.0643097521910486e-05,
      "loss": 1.0726,
      "step": 219200
    },
    {
      "epoch": 0.2,
      "grad_norm": 33.75,
      "learning_rate": 4.063859844871957e-05,
      "loss": 1.1053,
      "step": 219300
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.07470703125,
      "learning_rate": 4.0634099375528644e-05,
      "loss": 1.0408,
      "step": 219400
    },
    {
      "epoch": 0.2,
      "grad_norm": 35.0,
      "learning_rate": 4.0629600302337726e-05,
      "loss": 1.1241,
      "step": 219500
    },
    {
      "epoch": 0.2,
      "grad_norm": 174.0,
      "learning_rate": 4.06251012291468e-05,
      "loss": 0.8645,
      "step": 219600
    },
    {
      "epoch": 0.2,
      "grad_norm": 49.5,
      "learning_rate": 4.062060215595587e-05,
      "loss": 0.9771,
      "step": 219700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.061610308276495e-05,
      "loss": 1.0401,
      "step": 219800
    },
    {
      "epoch": 0.2,
      "grad_norm": 101.5,
      "learning_rate": 4.061160400957403e-05,
      "loss": 1.0759,
      "step": 219900
    },
    {
      "epoch": 0.2,
      "grad_norm": 82.0,
      "learning_rate": 4.06071049363831e-05,
      "loss": 0.9021,
      "step": 220000
    },
    {
      "epoch": 0.2,
      "grad_norm": 23.875,
      "learning_rate": 4.0602605863192185e-05,
      "loss": 1.0224,
      "step": 220100
    },
    {
      "epoch": 0.2,
      "grad_norm": 28.875,
      "learning_rate": 4.059810679000126e-05,
      "loss": 1.015,
      "step": 220200
    },
    {
      "epoch": 0.2,
      "grad_norm": 51.75,
      "learning_rate": 4.059360771681034e-05,
      "loss": 0.926,
      "step": 220300
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.25,
      "learning_rate": 4.058910864361942e-05,
      "loss": 0.9364,
      "step": 220400
    },
    {
      "epoch": 0.2,
      "grad_norm": 61.75,
      "learning_rate": 4.058460957042849e-05,
      "loss": 1.0805,
      "step": 220500
    },
    {
      "epoch": 0.2,
      "grad_norm": 17.625,
      "learning_rate": 4.0580110497237575e-05,
      "loss": 0.9326,
      "step": 220600
    },
    {
      "epoch": 0.2,
      "grad_norm": 54.75,
      "learning_rate": 4.057561142404665e-05,
      "loss": 1.1103,
      "step": 220700
    },
    {
      "epoch": 0.2,
      "grad_norm": 42.0,
      "learning_rate": 4.0571112350855726e-05,
      "loss": 1.1509,
      "step": 220800
    },
    {
      "epoch": 0.2,
      "grad_norm": 52.5,
      "learning_rate": 4.056661327766481e-05,
      "loss": 0.9183,
      "step": 220900
    },
    {
      "epoch": 0.2,
      "grad_norm": 66.5,
      "learning_rate": 4.0562114204473876e-05,
      "loss": 1.103,
      "step": 221000
    },
    {
      "epoch": 0.2,
      "grad_norm": 25.75,
      "learning_rate": 4.055761513128296e-05,
      "loss": 1.091,
      "step": 221100
    },
    {
      "epoch": 0.2,
      "grad_norm": 32.5,
      "learning_rate": 4.0553116058092034e-05,
      "loss": 1.1038,
      "step": 221200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.0054931640625,
      "learning_rate": 4.054861698490111e-05,
      "loss": 1.0682,
      "step": 221300
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.4375,
      "learning_rate": 4.054411791171019e-05,
      "loss": 1.1342,
      "step": 221400
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.5,
      "learning_rate": 4.0539618838519266e-05,
      "loss": 0.8835,
      "step": 221500
    },
    {
      "epoch": 0.2,
      "grad_norm": 78.5,
      "learning_rate": 4.053511976532834e-05,
      "loss": 0.9309,
      "step": 221600
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.0625,
      "learning_rate": 4.0530620692137424e-05,
      "loss": 0.9811,
      "step": 221700
    },
    {
      "epoch": 0.2,
      "grad_norm": 40.0,
      "learning_rate": 4.05261216189465e-05,
      "loss": 0.9188,
      "step": 221800
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.875,
      "learning_rate": 4.0521622545755575e-05,
      "loss": 1.0862,
      "step": 221900
    },
    {
      "epoch": 0.2,
      "grad_norm": 103.5,
      "learning_rate": 4.051712347256466e-05,
      "loss": 0.9534,
      "step": 222000
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.671875,
      "learning_rate": 4.051262439937373e-05,
      "loss": 1.0598,
      "step": 222100
    },
    {
      "epoch": 0.2,
      "grad_norm": 32.75,
      "learning_rate": 4.0508125326182814e-05,
      "loss": 0.9286,
      "step": 222200
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.125,
      "learning_rate": 4.050362625299188e-05,
      "loss": 1.0089,
      "step": 222300
    },
    {
      "epoch": 0.2,
      "grad_norm": 30.75,
      "learning_rate": 4.049912717980096e-05,
      "loss": 1.0282,
      "step": 222400
    },
    {
      "epoch": 0.2,
      "grad_norm": 28.125,
      "learning_rate": 4.049462810661004e-05,
      "loss": 1.1468,
      "step": 222500
    },
    {
      "epoch": 0.2,
      "grad_norm": 23.875,
      "learning_rate": 4.0490129033419116e-05,
      "loss": 1.0639,
      "step": 222600
    },
    {
      "epoch": 0.2,
      "grad_norm": 24.875,
      "learning_rate": 4.048562996022819e-05,
      "loss": 1.0247,
      "step": 222700
    },
    {
      "epoch": 0.2,
      "grad_norm": 65.0,
      "learning_rate": 4.048113088703727e-05,
      "loss": 1.0286,
      "step": 222800
    },
    {
      "epoch": 0.2,
      "grad_norm": 27.375,
      "learning_rate": 4.047663181384635e-05,
      "loss": 1.024,
      "step": 222900
    },
    {
      "epoch": 0.2,
      "grad_norm": 77.5,
      "learning_rate": 4.047213274065543e-05,
      "loss": 0.9071,
      "step": 223000
    },
    {
      "epoch": 0.2,
      "grad_norm": 50.75,
      "learning_rate": 4.0467633667464506e-05,
      "loss": 1.0451,
      "step": 223100
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.25,
      "learning_rate": 4.046313459427358e-05,
      "loss": 0.9831,
      "step": 223200
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.4375,
      "learning_rate": 4.045863552108266e-05,
      "loss": 0.955,
      "step": 223300
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.375,
      "learning_rate": 4.045413644789174e-05,
      "loss": 1.0547,
      "step": 223400
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.25,
      "learning_rate": 4.0449637374700814e-05,
      "loss": 1.034,
      "step": 223500
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.625,
      "learning_rate": 4.044513830150989e-05,
      "loss": 1.0968,
      "step": 223600
    },
    {
      "epoch": 0.2,
      "grad_norm": 100.0,
      "learning_rate": 4.0440639228318965e-05,
      "loss": 1.0565,
      "step": 223700
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.375,
      "learning_rate": 4.043614015512805e-05,
      "loss": 1.0507,
      "step": 223800
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.5,
      "learning_rate": 4.043164108193712e-05,
      "loss": 0.9951,
      "step": 223900
    },
    {
      "epoch": 0.2,
      "grad_norm": 524.0,
      "learning_rate": 4.04271420087462e-05,
      "loss": 0.9851,
      "step": 224000
    },
    {
      "epoch": 0.2,
      "grad_norm": 61.25,
      "learning_rate": 4.042264293555528e-05,
      "loss": 1.0471,
      "step": 224100
    },
    {
      "epoch": 0.2,
      "grad_norm": 29.875,
      "learning_rate": 4.0418143862364355e-05,
      "loss": 1.0684,
      "step": 224200
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.625,
      "learning_rate": 4.041364478917343e-05,
      "loss": 0.9652,
      "step": 224300
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.25,
      "learning_rate": 4.040914571598251e-05,
      "loss": 1.1236,
      "step": 224400
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46484375,
      "learning_rate": 4.040464664279159e-05,
      "loss": 1.0768,
      "step": 224500
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.4375,
      "learning_rate": 4.040014756960066e-05,
      "loss": 1.1431,
      "step": 224600
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.0625,
      "learning_rate": 4.0395648496409745e-05,
      "loss": 0.9607,
      "step": 224700
    },
    {
      "epoch": 0.2,
      "grad_norm": 29.125,
      "learning_rate": 4.039114942321882e-05,
      "loss": 0.905,
      "step": 224800
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.125,
      "learning_rate": 4.0386650350027896e-05,
      "loss": 1.0986,
      "step": 224900
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.00518798828125,
      "learning_rate": 4.038215127683697e-05,
      "loss": 0.9506,
      "step": 225000
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.3125,
      "learning_rate": 4.0377652203646046e-05,
      "loss": 1.1942,
      "step": 225100
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.25,
      "learning_rate": 4.037315313045513e-05,
      "loss": 1.0557,
      "step": 225200
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.875,
      "learning_rate": 4.0368654057264204e-05,
      "loss": 1.0554,
      "step": 225300
    },
    {
      "epoch": 0.2,
      "grad_norm": 28.0,
      "learning_rate": 4.036415498407328e-05,
      "loss": 1.0826,
      "step": 225400
    },
    {
      "epoch": 0.2,
      "grad_norm": 161.0,
      "learning_rate": 4.035965591088236e-05,
      "loss": 1.1307,
      "step": 225500
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.047607421875,
      "learning_rate": 4.035515683769144e-05,
      "loss": 1.0308,
      "step": 225600
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.0,
      "learning_rate": 4.035065776450052e-05,
      "loss": 0.9722,
      "step": 225700
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.4375,
      "learning_rate": 4.0346158691309594e-05,
      "loss": 1.0784,
      "step": 225800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.84765625,
      "learning_rate": 4.034165961811867e-05,
      "loss": 1.0371,
      "step": 225900
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.875,
      "learning_rate": 4.033716054492775e-05,
      "loss": 1.1102,
      "step": 226000
    },
    {
      "epoch": 0.2,
      "grad_norm": 22.875,
      "learning_rate": 4.033266147173683e-05,
      "loss": 0.9886,
      "step": 226100
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3671875,
      "learning_rate": 4.03281623985459e-05,
      "loss": 1.126,
      "step": 226200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.043212890625,
      "learning_rate": 4.032366332535498e-05,
      "loss": 1.0707,
      "step": 226300
    },
    {
      "epoch": 0.2,
      "grad_norm": 24.125,
      "learning_rate": 4.031916425216405e-05,
      "loss": 1.0397,
      "step": 226400
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.125,
      "learning_rate": 4.0314665178973135e-05,
      "loss": 1.1157,
      "step": 226500
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10009765625,
      "learning_rate": 4.031016610578221e-05,
      "loss": 1.0381,
      "step": 226600
    },
    {
      "epoch": 0.2,
      "grad_norm": 22.25,
      "learning_rate": 4.0305667032591286e-05,
      "loss": 1.0159,
      "step": 226700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.031982421875,
      "learning_rate": 4.030116795940037e-05,
      "loss": 1.0236,
      "step": 226800
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.53125,
      "learning_rate": 4.029666888620944e-05,
      "loss": 1.0486,
      "step": 226900
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.00921630859375,
      "learning_rate": 4.029216981301852e-05,
      "loss": 1.0281,
      "step": 227000
    },
    {
      "epoch": 0.2,
      "grad_norm": 51.75,
      "learning_rate": 4.02876707398276e-05,
      "loss": 1.1135,
      "step": 227100
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.25,
      "learning_rate": 4.0283171666636676e-05,
      "loss": 1.1518,
      "step": 227200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.0830078125,
      "learning_rate": 4.027867259344575e-05,
      "loss": 0.9774,
      "step": 227300
    },
    {
      "epoch": 0.2,
      "grad_norm": 32.0,
      "learning_rate": 4.0274173520254833e-05,
      "loss": 0.9246,
      "step": 227400
    },
    {
      "epoch": 0.2,
      "grad_norm": 296.0,
      "learning_rate": 4.02696744470639e-05,
      "loss": 1.0522,
      "step": 227500
    },
    {
      "epoch": 0.2,
      "grad_norm": 952.0,
      "learning_rate": 4.0265175373872984e-05,
      "loss": 0.8872,
      "step": 227600
    },
    {
      "epoch": 0.2,
      "grad_norm": 44.25,
      "learning_rate": 4.026067630068206e-05,
      "loss": 0.9439,
      "step": 227700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.01416015625,
      "learning_rate": 4.0256177227491135e-05,
      "loss": 0.9362,
      "step": 227800
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.9375,
      "learning_rate": 4.025167815430022e-05,
      "loss": 1.0321,
      "step": 227900
    },
    {
      "epoch": 0.2,
      "grad_norm": 49.75,
      "learning_rate": 4.024717908110929e-05,
      "loss": 0.8969,
      "step": 228000
    },
    {
      "epoch": 0.2,
      "grad_norm": 34.5,
      "learning_rate": 4.024268000791837e-05,
      "loss": 0.9787,
      "step": 228100
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.25,
      "learning_rate": 4.023818093472745e-05,
      "loss": 1.0048,
      "step": 228200
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.125,
      "learning_rate": 4.0233681861536525e-05,
      "loss": 1.1526,
      "step": 228300
    },
    {
      "epoch": 0.2,
      "grad_norm": 96.0,
      "learning_rate": 4.022918278834561e-05,
      "loss": 1.0252,
      "step": 228400
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.375,
      "learning_rate": 4.022468371515468e-05,
      "loss": 0.9873,
      "step": 228500
    },
    {
      "epoch": 0.2,
      "grad_norm": 28.125,
      "learning_rate": 4.022018464196376e-05,
      "loss": 1.0147,
      "step": 228600
    },
    {
      "epoch": 0.2,
      "grad_norm": 22.375,
      "learning_rate": 4.021568556877284e-05,
      "loss": 0.9519,
      "step": 228700
    },
    {
      "epoch": 0.2,
      "grad_norm": 40.0,
      "learning_rate": 4.021118649558191e-05,
      "loss": 1.1077,
      "step": 228800
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.84375,
      "learning_rate": 4.0206687422390984e-05,
      "loss": 0.9326,
      "step": 228900
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.341796875,
      "learning_rate": 4.0202188349200066e-05,
      "loss": 1.0734,
      "step": 229000
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.9375,
      "learning_rate": 4.019768927600914e-05,
      "loss": 0.9805,
      "step": 229100
    },
    {
      "epoch": 0.2,
      "grad_norm": 48.5,
      "learning_rate": 4.0193190202818223e-05,
      "loss": 0.9345,
      "step": 229200
    },
    {
      "epoch": 0.2,
      "grad_norm": 85.0,
      "learning_rate": 4.01886911296273e-05,
      "loss": 0.9318,
      "step": 229300
    },
    {
      "epoch": 0.2,
      "grad_norm": 56.25,
      "learning_rate": 4.0184192056436374e-05,
      "loss": 1.0994,
      "step": 229400
    },
    {
      "epoch": 0.2,
      "grad_norm": 25.25,
      "learning_rate": 4.0179692983245456e-05,
      "loss": 1.113,
      "step": 229500
    },
    {
      "epoch": 0.2,
      "grad_norm": 37.0,
      "learning_rate": 4.017519391005453e-05,
      "loss": 0.8794,
      "step": 229600
    },
    {
      "epoch": 0.2,
      "grad_norm": 35.5,
      "learning_rate": 4.017069483686361e-05,
      "loss": 1.0169,
      "step": 229700
    },
    {
      "epoch": 0.2,
      "grad_norm": 25.625,
      "learning_rate": 4.016619576367269e-05,
      "loss": 1.022,
      "step": 229800
    },
    {
      "epoch": 0.2,
      "grad_norm": 49.25,
      "learning_rate": 4.0161696690481764e-05,
      "loss": 1.0459,
      "step": 229900
    },
    {
      "epoch": 0.2,
      "grad_norm": 334.0,
      "learning_rate": 4.015719761729084e-05,
      "loss": 0.9685,
      "step": 230000
    },
    {
      "epoch": 0.2,
      "grad_norm": 17.375,
      "learning_rate": 4.0152698544099915e-05,
      "loss": 1.1808,
      "step": 230100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.310546875,
      "learning_rate": 4.014819947090899e-05,
      "loss": 1.0379,
      "step": 230200
    },
    {
      "epoch": 0.21,
      "grad_norm": 89.0,
      "learning_rate": 4.014370039771807e-05,
      "loss": 0.8973,
      "step": 230300
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.28125,
      "learning_rate": 4.013920132452715e-05,
      "loss": 1.0262,
      "step": 230400
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.4375,
      "learning_rate": 4.013470225133622e-05,
      "loss": 0.9842,
      "step": 230500
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.625,
      "learning_rate": 4.0130203178145305e-05,
      "loss": 1.067,
      "step": 230600
    },
    {
      "epoch": 0.21,
      "grad_norm": 284.0,
      "learning_rate": 4.012570410495438e-05,
      "loss": 0.9661,
      "step": 230700
    },
    {
      "epoch": 0.21,
      "grad_norm": 24.125,
      "learning_rate": 4.0121205031763456e-05,
      "loss": 0.9486,
      "step": 230800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.01275634765625,
      "learning_rate": 4.011670595857254e-05,
      "loss": 1.0253,
      "step": 230900
    },
    {
      "epoch": 0.21,
      "grad_norm": 43.75,
      "learning_rate": 4.0112206885381613e-05,
      "loss": 1.0081,
      "step": 231000
    },
    {
      "epoch": 0.21,
      "grad_norm": 66.0,
      "learning_rate": 4.0107707812190696e-05,
      "loss": 0.9609,
      "step": 231100
    },
    {
      "epoch": 0.21,
      "grad_norm": 26.875,
      "learning_rate": 4.010320873899977e-05,
      "loss": 0.9894,
      "step": 231200
    },
    {
      "epoch": 0.21,
      "grad_norm": 19.75,
      "learning_rate": 4.0098709665808846e-05,
      "loss": 1.0742,
      "step": 231300
    },
    {
      "epoch": 0.21,
      "grad_norm": 41.25,
      "learning_rate": 4.009421059261792e-05,
      "loss": 1.0614,
      "step": 231400
    },
    {
      "epoch": 0.21,
      "grad_norm": 49.0,
      "learning_rate": 4.0089711519427e-05,
      "loss": 1.0053,
      "step": 231500
    },
    {
      "epoch": 0.21,
      "grad_norm": 34.5,
      "learning_rate": 4.008521244623607e-05,
      "loss": 0.9822,
      "step": 231600
    },
    {
      "epoch": 0.21,
      "grad_norm": 32.0,
      "learning_rate": 4.0080713373045154e-05,
      "loss": 1.0565,
      "step": 231700
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.8125,
      "learning_rate": 4.007621429985423e-05,
      "loss": 1.0608,
      "step": 231800
    },
    {
      "epoch": 0.21,
      "grad_norm": 31.25,
      "learning_rate": 4.007171522666331e-05,
      "loss": 1.0291,
      "step": 231900
    },
    {
      "epoch": 0.21,
      "grad_norm": 43.5,
      "learning_rate": 4.006721615347239e-05,
      "loss": 1.005,
      "step": 232000
    },
    {
      "epoch": 0.21,
      "grad_norm": 110.5,
      "learning_rate": 4.006271708028146e-05,
      "loss": 0.8585,
      "step": 232100
    },
    {
      "epoch": 0.21,
      "grad_norm": 31.125,
      "learning_rate": 4.0058218007090545e-05,
      "loss": 1.0246,
      "step": 232200
    },
    {
      "epoch": 0.21,
      "grad_norm": 30.125,
      "learning_rate": 4.005371893389962e-05,
      "loss": 1.0897,
      "step": 232300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.032958984375,
      "learning_rate": 4.0049219860708695e-05,
      "loss": 1.1274,
      "step": 232400
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.5625,
      "learning_rate": 4.004472078751778e-05,
      "loss": 0.9087,
      "step": 232500
    },
    {
      "epoch": 0.21,
      "grad_norm": 115.5,
      "learning_rate": 4.004022171432685e-05,
      "loss": 0.9183,
      "step": 232600
    },
    {
      "epoch": 0.21,
      "grad_norm": 43.75,
      "learning_rate": 4.003572264113593e-05,
      "loss": 0.9738,
      "step": 232700
    },
    {
      "epoch": 0.21,
      "grad_norm": 147.0,
      "learning_rate": 4.0031223567945003e-05,
      "loss": 0.9961,
      "step": 232800
    },
    {
      "epoch": 0.21,
      "grad_norm": 19.5,
      "learning_rate": 4.002672449475408e-05,
      "loss": 1.1077,
      "step": 232900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.05859375,
      "learning_rate": 4.002222542156316e-05,
      "loss": 1.0937,
      "step": 233000
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.875,
      "learning_rate": 4.0017726348372236e-05,
      "loss": 0.9967,
      "step": 233100
    },
    {
      "epoch": 0.21,
      "grad_norm": 318.0,
      "learning_rate": 4.001322727518131e-05,
      "loss": 1.0481,
      "step": 233200
    },
    {
      "epoch": 0.21,
      "grad_norm": 28.0,
      "learning_rate": 4.0008728201990394e-05,
      "loss": 1.0507,
      "step": 233300
    },
    {
      "epoch": 0.21,
      "grad_norm": 96.5,
      "learning_rate": 4.000422912879947e-05,
      "loss": 0.933,
      "step": 233400
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.0625,
      "learning_rate": 3.9999730055608544e-05,
      "loss": 0.9831,
      "step": 233500
    },
    {
      "epoch": 0.21,
      "grad_norm": 202.0,
      "learning_rate": 3.9995230982417627e-05,
      "loss": 0.9308,
      "step": 233600
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.75,
      "learning_rate": 3.99907319092267e-05,
      "loss": 1.0269,
      "step": 233700
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.4375,
      "learning_rate": 3.9986232836035784e-05,
      "loss": 0.9417,
      "step": 233800
    },
    {
      "epoch": 0.21,
      "grad_norm": 42.0,
      "learning_rate": 3.998173376284486e-05,
      "loss": 0.9465,
      "step": 233900
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.875,
      "learning_rate": 3.997723468965393e-05,
      "loss": 0.9392,
      "step": 234000
    },
    {
      "epoch": 0.21,
      "grad_norm": 53.25,
      "learning_rate": 3.997273561646301e-05,
      "loss": 0.9887,
      "step": 234100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.005462646484375,
      "learning_rate": 3.9968236543272085e-05,
      "loss": 1.0167,
      "step": 234200
    },
    {
      "epoch": 0.21,
      "grad_norm": 27.25,
      "learning_rate": 3.996373747008116e-05,
      "loss": 1.0032,
      "step": 234300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.328125,
      "learning_rate": 3.995923839689024e-05,
      "loss": 1.0075,
      "step": 234400
    },
    {
      "epoch": 0.21,
      "grad_norm": 229.0,
      "learning_rate": 3.995473932369932e-05,
      "loss": 1.0626,
      "step": 234500
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.025146484375,
      "learning_rate": 3.99502402505084e-05,
      "loss": 1.128,
      "step": 234600
    },
    {
      "epoch": 0.21,
      "grad_norm": 31.75,
      "learning_rate": 3.9945741177317476e-05,
      "loss": 1.1265,
      "step": 234700
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.6875,
      "learning_rate": 3.994124210412655e-05,
      "loss": 1.1406,
      "step": 234800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.0174560546875,
      "learning_rate": 3.993674303093563e-05,
      "loss": 0.9755,
      "step": 234900
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.0,
      "learning_rate": 3.993224395774471e-05,
      "loss": 1.0672,
      "step": 235000
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.375,
      "learning_rate": 3.9927744884553784e-05,
      "loss": 0.9761,
      "step": 235100
    },
    {
      "epoch": 0.21,
      "grad_norm": 57.25,
      "learning_rate": 3.9923245811362866e-05,
      "loss": 0.9685,
      "step": 235200
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.0,
      "learning_rate": 3.9918746738171934e-05,
      "loss": 1.0695,
      "step": 235300
    },
    {
      "epoch": 0.21,
      "grad_norm": 80.5,
      "learning_rate": 3.9914247664981017e-05,
      "loss": 1.0753,
      "step": 235400
    },
    {
      "epoch": 0.21,
      "grad_norm": 24.25,
      "learning_rate": 3.990974859179009e-05,
      "loss": 0.9355,
      "step": 235500
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.036376953125,
      "learning_rate": 3.990524951859917e-05,
      "loss": 0.8267,
      "step": 235600
    },
    {
      "epoch": 0.21,
      "grad_norm": 28.0,
      "learning_rate": 3.990075044540825e-05,
      "loss": 1.0094,
      "step": 235700
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.875,
      "learning_rate": 3.9896251372217325e-05,
      "loss": 0.8892,
      "step": 235800
    },
    {
      "epoch": 0.21,
      "grad_norm": 36.25,
      "learning_rate": 3.98917522990264e-05,
      "loss": 0.9947,
      "step": 235900
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.125,
      "learning_rate": 3.988725322583548e-05,
      "loss": 0.9337,
      "step": 236000
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.4375,
      "learning_rate": 3.988275415264456e-05,
      "loss": 1.2244,
      "step": 236100
    },
    {
      "epoch": 0.21,
      "grad_norm": 29.0,
      "learning_rate": 3.987825507945363e-05,
      "loss": 1.0279,
      "step": 236200
    },
    {
      "epoch": 0.21,
      "grad_norm": 29.0,
      "learning_rate": 3.9873756006262715e-05,
      "loss": 1.1059,
      "step": 236300
    },
    {
      "epoch": 0.21,
      "grad_norm": 27.0,
      "learning_rate": 3.986925693307179e-05,
      "loss": 1.0736,
      "step": 236400
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1298828125,
      "learning_rate": 3.986475785988087e-05,
      "loss": 1.014,
      "step": 236500
    },
    {
      "epoch": 0.21,
      "grad_norm": 40.5,
      "learning_rate": 3.986025878668994e-05,
      "loss": 0.9059,
      "step": 236600
    },
    {
      "epoch": 0.21,
      "grad_norm": 120.5,
      "learning_rate": 3.9855759713499016e-05,
      "loss": 0.9994,
      "step": 236700
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.375,
      "learning_rate": 3.98512606403081e-05,
      "loss": 0.8617,
      "step": 236800
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.6875,
      "learning_rate": 3.9846761567117174e-05,
      "loss": 0.9958,
      "step": 236900
    },
    {
      "epoch": 0.21,
      "grad_norm": 35.0,
      "learning_rate": 3.984226249392625e-05,
      "loss": 0.9294,
      "step": 237000
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2275390625,
      "learning_rate": 3.983776342073533e-05,
      "loss": 0.9823,
      "step": 237100
    },
    {
      "epoch": 0.21,
      "grad_norm": 155.0,
      "learning_rate": 3.9833264347544407e-05,
      "loss": 1.0118,
      "step": 237200
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.004852294921875,
      "learning_rate": 3.982876527435349e-05,
      "loss": 1.0626,
      "step": 237300
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.9375,
      "learning_rate": 3.9824266201162564e-05,
      "loss": 1.0751,
      "step": 237400
    },
    {
      "epoch": 0.21,
      "grad_norm": 39.0,
      "learning_rate": 3.981976712797164e-05,
      "loss": 0.9286,
      "step": 237500
    },
    {
      "epoch": 0.21,
      "grad_norm": 37.75,
      "learning_rate": 3.981526805478072e-05,
      "loss": 1.0099,
      "step": 237600
    },
    {
      "epoch": 0.21,
      "grad_norm": 316.0,
      "learning_rate": 3.98107689815898e-05,
      "loss": 1.0549,
      "step": 237700
    },
    {
      "epoch": 0.21,
      "grad_norm": 44.5,
      "learning_rate": 3.980626990839887e-05,
      "loss": 1.0088,
      "step": 237800
    },
    {
      "epoch": 0.21,
      "grad_norm": 49.5,
      "learning_rate": 3.980177083520795e-05,
      "loss": 0.8736,
      "step": 237900
    },
    {
      "epoch": 0.21,
      "grad_norm": 80.0,
      "learning_rate": 3.979727176201702e-05,
      "loss": 1.1077,
      "step": 238000
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.875,
      "learning_rate": 3.9792772688826105e-05,
      "loss": 0.9999,
      "step": 238100
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.625,
      "learning_rate": 3.978827361563518e-05,
      "loss": 0.9204,
      "step": 238200
    },
    {
      "epoch": 0.21,
      "grad_norm": 39.75,
      "learning_rate": 3.9783774542444256e-05,
      "loss": 0.8332,
      "step": 238300
    },
    {
      "epoch": 0.21,
      "grad_norm": 23.75,
      "learning_rate": 3.977927546925334e-05,
      "loss": 1.014,
      "step": 238400
    },
    {
      "epoch": 0.21,
      "grad_norm": 35.0,
      "learning_rate": 3.977477639606241e-05,
      "loss": 1.0413,
      "step": 238500
    },
    {
      "epoch": 0.21,
      "grad_norm": 46.75,
      "learning_rate": 3.977027732287149e-05,
      "loss": 0.9752,
      "step": 238600
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.375,
      "learning_rate": 3.976577824968057e-05,
      "loss": 1.0089,
      "step": 238700
    },
    {
      "epoch": 0.21,
      "grad_norm": 25.0,
      "learning_rate": 3.9761279176489646e-05,
      "loss": 0.9386,
      "step": 238800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.0140380859375,
      "learning_rate": 3.975678010329872e-05,
      "loss": 1.1574,
      "step": 238900
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.25,
      "learning_rate": 3.97522810301078e-05,
      "loss": 1.1042,
      "step": 239000
    },
    {
      "epoch": 0.21,
      "grad_norm": 20.625,
      "learning_rate": 3.974778195691688e-05,
      "loss": 1.1184,
      "step": 239100
    },
    {
      "epoch": 0.21,
      "grad_norm": 20.0,
      "learning_rate": 3.9743282883725954e-05,
      "loss": 1.0072,
      "step": 239200
    },
    {
      "epoch": 0.21,
      "grad_norm": 62.5,
      "learning_rate": 3.973878381053503e-05,
      "loss": 1.1051,
      "step": 239300
    },
    {
      "epoch": 0.21,
      "grad_norm": 46.0,
      "learning_rate": 3.9734284737344105e-05,
      "loss": 1.024,
      "step": 239400
    },
    {
      "epoch": 0.21,
      "grad_norm": 46.25,
      "learning_rate": 3.972978566415319e-05,
      "loss": 0.9284,
      "step": 239500
    },
    {
      "epoch": 0.21,
      "grad_norm": 20.375,
      "learning_rate": 3.972528659096226e-05,
      "loss": 0.9666,
      "step": 239600
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.28125,
      "learning_rate": 3.972078751777134e-05,
      "loss": 0.9218,
      "step": 239700
    },
    {
      "epoch": 0.21,
      "grad_norm": 41.75,
      "learning_rate": 3.971628844458042e-05,
      "loss": 0.8954,
      "step": 239800
    },
    {
      "epoch": 0.21,
      "grad_norm": 39.75,
      "learning_rate": 3.9711789371389495e-05,
      "loss": 0.9357,
      "step": 239900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.546875,
      "learning_rate": 3.970729029819858e-05,
      "loss": 1.0388,
      "step": 240000
    },
    {
      "epoch": 0.21,
      "grad_norm": 46.5,
      "learning_rate": 3.970279122500765e-05,
      "loss": 0.8919,
      "step": 240100
    },
    {
      "epoch": 0.21,
      "grad_norm": 79.0,
      "learning_rate": 3.969829215181673e-05,
      "loss": 0.9717,
      "step": 240200
    },
    {
      "epoch": 0.21,
      "grad_norm": 35.25,
      "learning_rate": 3.969379307862581e-05,
      "loss": 1.0627,
      "step": 240300
    },
    {
      "epoch": 0.21,
      "grad_norm": 32.75,
      "learning_rate": 3.9689294005434885e-05,
      "loss": 0.8763,
      "step": 240400
    },
    {
      "epoch": 0.21,
      "grad_norm": 65.0,
      "learning_rate": 3.968479493224396e-05,
      "loss": 1.007,
      "step": 240500
    },
    {
      "epoch": 0.21,
      "grad_norm": 45.25,
      "learning_rate": 3.9680295859053036e-05,
      "loss": 0.9996,
      "step": 240600
    },
    {
      "epoch": 0.21,
      "grad_norm": 27.875,
      "learning_rate": 3.967579678586211e-05,
      "loss": 0.9441,
      "step": 240700
    },
    {
      "epoch": 0.21,
      "grad_norm": 38.0,
      "learning_rate": 3.967129771267119e-05,
      "loss": 1.0479,
      "step": 240800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.0262451171875,
      "learning_rate": 3.966679863948027e-05,
      "loss": 1.0767,
      "step": 240900
    },
    {
      "epoch": 0.21,
      "grad_norm": 21.375,
      "learning_rate": 3.9662299566289344e-05,
      "loss": 1.0097,
      "step": 241000
    },
    {
      "epoch": 0.21,
      "grad_norm": 32.25,
      "learning_rate": 3.9657800493098426e-05,
      "loss": 1.0596,
      "step": 241100
    },
    {
      "epoch": 0.21,
      "grad_norm": 42.25,
      "learning_rate": 3.96533014199075e-05,
      "loss": 1.0287,
      "step": 241200
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0,
      "learning_rate": 3.964880234671658e-05,
      "loss": 1.0252,
      "step": 241300
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.3125,
      "learning_rate": 3.964430327352566e-05,
      "loss": 1.0113,
      "step": 241400
    },
    {
      "epoch": 0.22,
      "grad_norm": 141.0,
      "learning_rate": 3.9639804200334734e-05,
      "loss": 0.9814,
      "step": 241500
    },
    {
      "epoch": 0.22,
      "grad_norm": 39.75,
      "learning_rate": 3.963530512714381e-05,
      "loss": 0.8839,
      "step": 241600
    },
    {
      "epoch": 0.22,
      "grad_norm": 32.25,
      "learning_rate": 3.963080605395289e-05,
      "loss": 0.9375,
      "step": 241700
    },
    {
      "epoch": 0.22,
      "grad_norm": 29.125,
      "learning_rate": 3.962630698076196e-05,
      "loss": 0.8722,
      "step": 241800
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.125,
      "learning_rate": 3.962180790757104e-05,
      "loss": 0.9342,
      "step": 241900
    },
    {
      "epoch": 0.22,
      "grad_norm": 91.0,
      "learning_rate": 3.961730883438012e-05,
      "loss": 0.8654,
      "step": 242000
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.25,
      "learning_rate": 3.961280976118919e-05,
      "loss": 1.0902,
      "step": 242100
    },
    {
      "epoch": 0.22,
      "grad_norm": 91.5,
      "learning_rate": 3.9608310687998275e-05,
      "loss": 1.0038,
      "step": 242200
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.53125,
      "learning_rate": 3.960381161480735e-05,
      "loss": 0.9835,
      "step": 242300
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.1875,
      "learning_rate": 3.9599312541616426e-05,
      "loss": 1.0077,
      "step": 242400
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.6875,
      "learning_rate": 3.959481346842551e-05,
      "loss": 1.1076,
      "step": 242500
    },
    {
      "epoch": 0.22,
      "grad_norm": 34.75,
      "learning_rate": 3.959031439523458e-05,
      "loss": 0.9872,
      "step": 242600
    },
    {
      "epoch": 0.22,
      "grad_norm": 95.5,
      "learning_rate": 3.9585815322043665e-05,
      "loss": 1.1495,
      "step": 242700
    },
    {
      "epoch": 0.22,
      "grad_norm": 29.375,
      "learning_rate": 3.958131624885274e-05,
      "loss": 1.1064,
      "step": 242800
    },
    {
      "epoch": 0.22,
      "grad_norm": 27.125,
      "learning_rate": 3.9576817175661816e-05,
      "loss": 1.061,
      "step": 242900
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.84375,
      "learning_rate": 3.95723181024709e-05,
      "loss": 1.024,
      "step": 243000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.412109375,
      "learning_rate": 3.956781902927997e-05,
      "loss": 0.9002,
      "step": 243100
    },
    {
      "epoch": 0.22,
      "grad_norm": 54.75,
      "learning_rate": 3.956331995608905e-05,
      "loss": 1.1241,
      "step": 243200
    },
    {
      "epoch": 0.22,
      "grad_norm": 23.875,
      "learning_rate": 3.9558820882898124e-05,
      "loss": 1.0088,
      "step": 243300
    },
    {
      "epoch": 0.22,
      "grad_norm": 26.125,
      "learning_rate": 3.95543218097072e-05,
      "loss": 0.992,
      "step": 243400
    },
    {
      "epoch": 0.22,
      "grad_norm": 25.875,
      "learning_rate": 3.954982273651628e-05,
      "loss": 1.024,
      "step": 243500
    },
    {
      "epoch": 0.22,
      "grad_norm": 34.75,
      "learning_rate": 3.954532366332536e-05,
      "loss": 0.9248,
      "step": 243600
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.6875,
      "learning_rate": 3.954082459013443e-05,
      "loss": 0.9332,
      "step": 243700
    },
    {
      "epoch": 0.22,
      "grad_norm": 39.75,
      "learning_rate": 3.9536325516943514e-05,
      "loss": 1.0014,
      "step": 243800
    },
    {
      "epoch": 0.22,
      "grad_norm": 23.75,
      "learning_rate": 3.953182644375259e-05,
      "loss": 1.1147,
      "step": 243900
    },
    {
      "epoch": 0.22,
      "grad_norm": 30.375,
      "learning_rate": 3.9527327370561665e-05,
      "loss": 1.0165,
      "step": 244000
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.5,
      "learning_rate": 3.952282829737075e-05,
      "loss": 1.0904,
      "step": 244100
    },
    {
      "epoch": 0.22,
      "grad_norm": 100.5,
      "learning_rate": 3.951832922417982e-05,
      "loss": 1.0877,
      "step": 244200
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.625,
      "learning_rate": 3.95138301509889e-05,
      "loss": 0.8984,
      "step": 244300
    },
    {
      "epoch": 0.22,
      "grad_norm": 70.5,
      "learning_rate": 3.950933107779797e-05,
      "loss": 0.9939,
      "step": 244400
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.8125,
      "learning_rate": 3.950483200460705e-05,
      "loss": 1.1845,
      "step": 244500
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.5625,
      "learning_rate": 3.950033293141613e-05,
      "loss": 0.9884,
      "step": 244600
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.625,
      "learning_rate": 3.9495833858225206e-05,
      "loss": 1.0626,
      "step": 244700
    },
    {
      "epoch": 0.22,
      "grad_norm": 40.75,
      "learning_rate": 3.949133478503428e-05,
      "loss": 0.9637,
      "step": 244800
    },
    {
      "epoch": 0.22,
      "grad_norm": 40.5,
      "learning_rate": 3.9486835711843364e-05,
      "loss": 1.1104,
      "step": 244900
    },
    {
      "epoch": 0.22,
      "grad_norm": 38.75,
      "learning_rate": 3.948233663865244e-05,
      "loss": 0.9862,
      "step": 245000
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.25,
      "learning_rate": 3.9477837565461514e-05,
      "loss": 1.0989,
      "step": 245100
    },
    {
      "epoch": 0.22,
      "grad_norm": 56.75,
      "learning_rate": 3.9473338492270596e-05,
      "loss": 1.0406,
      "step": 245200
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.6875,
      "learning_rate": 3.946883941907967e-05,
      "loss": 1.0767,
      "step": 245300
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.3125,
      "learning_rate": 3.9464340345888754e-05,
      "loss": 0.8896,
      "step": 245400
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.8125,
      "learning_rate": 3.945984127269783e-05,
      "loss": 1.022,
      "step": 245500
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.5,
      "learning_rate": 3.9455342199506904e-05,
      "loss": 1.0143,
      "step": 245600
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.15625,
      "learning_rate": 3.945084312631598e-05,
      "loss": 0.909,
      "step": 245700
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.01409912109375,
      "learning_rate": 3.9446344053125055e-05,
      "loss": 1.0102,
      "step": 245800
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15234375,
      "learning_rate": 3.944184497993413e-05,
      "loss": 0.901,
      "step": 245900
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.5625,
      "learning_rate": 3.943734590674321e-05,
      "loss": 1.1719,
      "step": 246000
    },
    {
      "epoch": 0.22,
      "grad_norm": 35.75,
      "learning_rate": 3.943284683355229e-05,
      "loss": 1.0459,
      "step": 246100
    },
    {
      "epoch": 0.22,
      "grad_norm": 448.0,
      "learning_rate": 3.942834776036137e-05,
      "loss": 1.0996,
      "step": 246200
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.375,
      "learning_rate": 3.9423848687170445e-05,
      "loss": 0.9833,
      "step": 246300
    },
    {
      "epoch": 0.22,
      "grad_norm": 91.5,
      "learning_rate": 3.941934961397952e-05,
      "loss": 1.0997,
      "step": 246400
    },
    {
      "epoch": 0.22,
      "grad_norm": 56.75,
      "learning_rate": 3.94148505407886e-05,
      "loss": 0.8731,
      "step": 246500
    },
    {
      "epoch": 0.22,
      "grad_norm": 620.0,
      "learning_rate": 3.941035146759768e-05,
      "loss": 1.0876,
      "step": 246600
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.875,
      "learning_rate": 3.9405852394406754e-05,
      "loss": 0.9109,
      "step": 246700
    },
    {
      "epoch": 0.22,
      "grad_norm": 97.0,
      "learning_rate": 3.9401353321215836e-05,
      "loss": 1.0517,
      "step": 246800
    },
    {
      "epoch": 0.22,
      "grad_norm": 36.0,
      "learning_rate": 3.939685424802491e-05,
      "loss": 1.0084,
      "step": 246900
    },
    {
      "epoch": 0.22,
      "grad_norm": 63.0,
      "learning_rate": 3.9392355174833986e-05,
      "loss": 0.977,
      "step": 247000
    },
    {
      "epoch": 0.22,
      "grad_norm": 17.0,
      "learning_rate": 3.938785610164306e-05,
      "loss": 1.1385,
      "step": 247100
    },
    {
      "epoch": 0.22,
      "grad_norm": 23.625,
      "learning_rate": 3.938335702845214e-05,
      "loss": 1.1118,
      "step": 247200
    },
    {
      "epoch": 0.22,
      "grad_norm": 25.5,
      "learning_rate": 3.937885795526122e-05,
      "loss": 0.9351,
      "step": 247300
    },
    {
      "epoch": 0.22,
      "grad_norm": 28.25,
      "learning_rate": 3.9374358882070294e-05,
      "loss": 1.1201,
      "step": 247400
    },
    {
      "epoch": 0.22,
      "grad_norm": 29.75,
      "learning_rate": 3.936985980887937e-05,
      "loss": 1.1406,
      "step": 247500
    },
    {
      "epoch": 0.22,
      "grad_norm": 52.5,
      "learning_rate": 3.936536073568845e-05,
      "loss": 1.1221,
      "step": 247600
    },
    {
      "epoch": 0.22,
      "grad_norm": 60.5,
      "learning_rate": 3.936086166249753e-05,
      "loss": 0.9895,
      "step": 247700
    },
    {
      "epoch": 0.22,
      "grad_norm": 1104.0,
      "learning_rate": 3.93563625893066e-05,
      "loss": 0.9288,
      "step": 247800
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.3125,
      "learning_rate": 3.9351863516115685e-05,
      "loss": 1.0274,
      "step": 247900
    },
    {
      "epoch": 0.22,
      "grad_norm": 133.0,
      "learning_rate": 3.934736444292476e-05,
      "loss": 0.98,
      "step": 248000
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.90625,
      "learning_rate": 3.934286536973384e-05,
      "loss": 0.8549,
      "step": 248100
    },
    {
      "epoch": 0.22,
      "grad_norm": 52.25,
      "learning_rate": 3.933836629654292e-05,
      "loss": 0.9437,
      "step": 248200
    },
    {
      "epoch": 0.22,
      "grad_norm": 30.625,
      "learning_rate": 3.9333867223351986e-05,
      "loss": 1.0633,
      "step": 248300
    },
    {
      "epoch": 0.22,
      "grad_norm": 57.0,
      "learning_rate": 3.932936815016107e-05,
      "loss": 0.9639,
      "step": 248400
    },
    {
      "epoch": 0.22,
      "grad_norm": 188.0,
      "learning_rate": 3.9324869076970144e-05,
      "loss": 0.8477,
      "step": 248500
    },
    {
      "epoch": 0.22,
      "grad_norm": 56.75,
      "learning_rate": 3.932037000377922e-05,
      "loss": 1.1589,
      "step": 248600
    },
    {
      "epoch": 0.22,
      "grad_norm": 30.375,
      "learning_rate": 3.93158709305883e-05,
      "loss": 1.0893,
      "step": 248700
    },
    {
      "epoch": 0.22,
      "grad_norm": 128.0,
      "learning_rate": 3.9311371857397376e-05,
      "loss": 1.1633,
      "step": 248800
    },
    {
      "epoch": 0.22,
      "grad_norm": 17.625,
      "learning_rate": 3.930687278420646e-05,
      "loss": 0.9066,
      "step": 248900
    },
    {
      "epoch": 0.22,
      "grad_norm": 59.5,
      "learning_rate": 3.9302373711015534e-05,
      "loss": 1.1544,
      "step": 249000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.037109375,
      "learning_rate": 3.929787463782461e-05,
      "loss": 1.0615,
      "step": 249100
    },
    {
      "epoch": 0.22,
      "grad_norm": 368.0,
      "learning_rate": 3.929337556463369e-05,
      "loss": 1.1389,
      "step": 249200
    },
    {
      "epoch": 0.22,
      "grad_norm": 45.75,
      "learning_rate": 3.9288876491442767e-05,
      "loss": 1.2061,
      "step": 249300
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.375,
      "learning_rate": 3.928437741825184e-05,
      "loss": 1.0681,
      "step": 249400
    },
    {
      "epoch": 0.22,
      "grad_norm": 193.0,
      "learning_rate": 3.9279878345060924e-05,
      "loss": 0.9814,
      "step": 249500
    },
    {
      "epoch": 0.22,
      "grad_norm": 39.75,
      "learning_rate": 3.927537927186999e-05,
      "loss": 0.9327,
      "step": 249600
    },
    {
      "epoch": 0.22,
      "grad_norm": 23.25,
      "learning_rate": 3.9270880198679075e-05,
      "loss": 0.9946,
      "step": 249700
    },
    {
      "epoch": 0.22,
      "grad_norm": 35.5,
      "learning_rate": 3.926638112548815e-05,
      "loss": 1.063,
      "step": 249800
    },
    {
      "epoch": 0.22,
      "grad_norm": 288.0,
      "learning_rate": 3.9261882052297225e-05,
      "loss": 0.9444,
      "step": 249900
    },
    {
      "epoch": 0.22,
      "grad_norm": 51.25,
      "learning_rate": 3.925738297910631e-05,
      "loss": 1.071,
      "step": 250000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1572265625,
      "learning_rate": 3.925288390591538e-05,
      "loss": 0.801,
      "step": 250100
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.032958984375,
      "learning_rate": 3.924838483272446e-05,
      "loss": 0.9229,
      "step": 250200
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.25,
      "learning_rate": 3.924388575953354e-05,
      "loss": 1.0075,
      "step": 250300
    },
    {
      "epoch": 0.22,
      "grad_norm": 73.5,
      "learning_rate": 3.9239386686342616e-05,
      "loss": 1.1117,
      "step": 250400
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.375,
      "learning_rate": 3.923488761315169e-05,
      "loss": 0.8667,
      "step": 250500
    },
    {
      "epoch": 0.22,
      "grad_norm": 104.0,
      "learning_rate": 3.923038853996077e-05,
      "loss": 0.9855,
      "step": 250600
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.060791015625,
      "learning_rate": 3.922588946676985e-05,
      "loss": 1.0965,
      "step": 250700
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.01336669921875,
      "learning_rate": 3.922139039357893e-05,
      "loss": 1.0748,
      "step": 250800
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.375,
      "learning_rate": 3.9216891320388e-05,
      "loss": 1.1585,
      "step": 250900
    },
    {
      "epoch": 0.22,
      "grad_norm": 46.25,
      "learning_rate": 3.9212392247197074e-05,
      "loss": 1.1373,
      "step": 251000
    },
    {
      "epoch": 0.22,
      "grad_norm": 89.5,
      "learning_rate": 3.9207893174006157e-05,
      "loss": 1.0942,
      "step": 251100
    },
    {
      "epoch": 0.22,
      "grad_norm": 49.75,
      "learning_rate": 3.920339410081523e-05,
      "loss": 0.9776,
      "step": 251200
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.6875,
      "learning_rate": 3.919889502762431e-05,
      "loss": 1.0352,
      "step": 251300
    },
    {
      "epoch": 0.22,
      "grad_norm": 47.5,
      "learning_rate": 3.919439595443339e-05,
      "loss": 1.0185,
      "step": 251400
    },
    {
      "epoch": 0.22,
      "grad_norm": 49.75,
      "learning_rate": 3.9189896881242465e-05,
      "loss": 0.9082,
      "step": 251500
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.375,
      "learning_rate": 3.918539780805155e-05,
      "loss": 0.9834,
      "step": 251600
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.375,
      "learning_rate": 3.918089873486062e-05,
      "loss": 0.9561,
      "step": 251700
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.125,
      "learning_rate": 3.91763996616697e-05,
      "loss": 1.0995,
      "step": 251800
    },
    {
      "epoch": 0.22,
      "grad_norm": 33.0,
      "learning_rate": 3.917190058847878e-05,
      "loss": 1.1076,
      "step": 251900
    },
    {
      "epoch": 0.22,
      "grad_norm": 62.0,
      "learning_rate": 3.9167401515287855e-05,
      "loss": 0.8693,
      "step": 252000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2470703125,
      "learning_rate": 3.916290244209693e-05,
      "loss": 1.0181,
      "step": 252100
    },
    {
      "epoch": 0.22,
      "grad_norm": 41.25,
      "learning_rate": 3.9158403368906006e-05,
      "loss": 0.9503,
      "step": 252200
    },
    {
      "epoch": 0.22,
      "grad_norm": 17.5,
      "learning_rate": 3.915390429571508e-05,
      "loss": 1.0744,
      "step": 252300
    },
    {
      "epoch": 0.22,
      "grad_norm": 41.5,
      "learning_rate": 3.914940522252416e-05,
      "loss": 1.003,
      "step": 252400
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.75,
      "learning_rate": 3.914490614933324e-05,
      "loss": 1.0178,
      "step": 252500
    },
    {
      "epoch": 0.23,
      "grad_norm": 23.125,
      "learning_rate": 3.9140407076142314e-05,
      "loss": 1.0478,
      "step": 252600
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.8125,
      "learning_rate": 3.9135908002951396e-05,
      "loss": 1.0237,
      "step": 252700
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.0,
      "learning_rate": 3.913140892976047e-05,
      "loss": 0.9183,
      "step": 252800
    },
    {
      "epoch": 0.23,
      "grad_norm": 19.5,
      "learning_rate": 3.9126909856569547e-05,
      "loss": 1.2169,
      "step": 252900
    },
    {
      "epoch": 0.23,
      "grad_norm": 213.0,
      "learning_rate": 3.912241078337863e-05,
      "loss": 1.0258,
      "step": 253000
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.625,
      "learning_rate": 3.9117911710187704e-05,
      "loss": 0.9357,
      "step": 253100
    },
    {
      "epoch": 0.23,
      "grad_norm": 30.0,
      "learning_rate": 3.911341263699678e-05,
      "loss": 1.0019,
      "step": 253200
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.5,
      "learning_rate": 3.910891356380586e-05,
      "loss": 0.8838,
      "step": 253300
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.71875,
      "learning_rate": 3.910441449061494e-05,
      "loss": 0.9828,
      "step": 253400
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.375,
      "learning_rate": 3.909991541742401e-05,
      "loss": 0.9326,
      "step": 253500
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.8125,
      "learning_rate": 3.909541634423309e-05,
      "loss": 1.0334,
      "step": 253600
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.75,
      "learning_rate": 3.909091727104216e-05,
      "loss": 1.0168,
      "step": 253700
    },
    {
      "epoch": 0.23,
      "grad_norm": 33.0,
      "learning_rate": 3.9086418197851245e-05,
      "loss": 0.9915,
      "step": 253800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.193359375,
      "learning_rate": 3.908191912466032e-05,
      "loss": 1.0294,
      "step": 253900
    },
    {
      "epoch": 0.23,
      "grad_norm": 29.625,
      "learning_rate": 3.9077420051469396e-05,
      "loss": 1.0334,
      "step": 254000
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.3125,
      "learning_rate": 3.907292097827848e-05,
      "loss": 1.1583,
      "step": 254100
    },
    {
      "epoch": 0.23,
      "grad_norm": 33.0,
      "learning_rate": 3.906842190508755e-05,
      "loss": 0.9236,
      "step": 254200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.007110595703125,
      "learning_rate": 3.9063922831896635e-05,
      "loss": 0.9179,
      "step": 254300
    },
    {
      "epoch": 0.23,
      "grad_norm": 43.25,
      "learning_rate": 3.905942375870571e-05,
      "loss": 0.9746,
      "step": 254400
    },
    {
      "epoch": 0.23,
      "grad_norm": 19.125,
      "learning_rate": 3.9054924685514786e-05,
      "loss": 1.0515,
      "step": 254500
    },
    {
      "epoch": 0.23,
      "grad_norm": 153.0,
      "learning_rate": 3.905042561232387e-05,
      "loss": 1.1089,
      "step": 254600
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.375,
      "learning_rate": 3.904592653913294e-05,
      "loss": 0.9498,
      "step": 254700
    },
    {
      "epoch": 0.23,
      "grad_norm": 66.0,
      "learning_rate": 3.904142746594202e-05,
      "loss": 1.1103,
      "step": 254800
    },
    {
      "epoch": 0.23,
      "grad_norm": 86.5,
      "learning_rate": 3.9036928392751094e-05,
      "loss": 1.0989,
      "step": 254900
    },
    {
      "epoch": 0.23,
      "grad_norm": 114.5,
      "learning_rate": 3.903242931956017e-05,
      "loss": 1.1017,
      "step": 255000
    },
    {
      "epoch": 0.23,
      "grad_norm": 27.125,
      "learning_rate": 3.902793024636925e-05,
      "loss": 1.0728,
      "step": 255100
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.5,
      "learning_rate": 3.902343117317833e-05,
      "loss": 0.9304,
      "step": 255200
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.6875,
      "learning_rate": 3.90189320999874e-05,
      "loss": 1.062,
      "step": 255300
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.625,
      "learning_rate": 3.9014433026796484e-05,
      "loss": 1.021,
      "step": 255400
    },
    {
      "epoch": 0.23,
      "grad_norm": 98.0,
      "learning_rate": 3.900993395360556e-05,
      "loss": 0.9793,
      "step": 255500
    },
    {
      "epoch": 0.23,
      "grad_norm": 57.5,
      "learning_rate": 3.9005434880414635e-05,
      "loss": 0.9593,
      "step": 255600
    },
    {
      "epoch": 0.23,
      "grad_norm": 21.875,
      "learning_rate": 3.900093580722372e-05,
      "loss": 0.8503,
      "step": 255700
    },
    {
      "epoch": 0.23,
      "grad_norm": 22.0,
      "learning_rate": 3.899643673403279e-05,
      "loss": 1.0543,
      "step": 255800
    },
    {
      "epoch": 0.23,
      "grad_norm": 56.25,
      "learning_rate": 3.899193766084187e-05,
      "loss": 1.0556,
      "step": 255900
    },
    {
      "epoch": 0.23,
      "grad_norm": 58.75,
      "learning_rate": 3.898743858765095e-05,
      "loss": 1.0963,
      "step": 256000
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.25,
      "learning_rate": 3.898293951446002e-05,
      "loss": 0.9329,
      "step": 256100
    },
    {
      "epoch": 0.23,
      "grad_norm": 39.25,
      "learning_rate": 3.89784404412691e-05,
      "loss": 1.0692,
      "step": 256200
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.125,
      "learning_rate": 3.8973941368078176e-05,
      "loss": 0.9528,
      "step": 256300
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5703125,
      "learning_rate": 3.896944229488725e-05,
      "loss": 0.951,
      "step": 256400
    },
    {
      "epoch": 0.23,
      "grad_norm": 24.5,
      "learning_rate": 3.896494322169633e-05,
      "loss": 0.9127,
      "step": 256500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.166015625,
      "learning_rate": 3.896044414850541e-05,
      "loss": 0.9868,
      "step": 256600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21484375,
      "learning_rate": 3.8955945075314484e-05,
      "loss": 1.1349,
      "step": 256700
    },
    {
      "epoch": 0.23,
      "grad_norm": 37.5,
      "learning_rate": 3.8951446002123566e-05,
      "loss": 0.9999,
      "step": 256800
    },
    {
      "epoch": 0.23,
      "grad_norm": 44.25,
      "learning_rate": 3.894694692893264e-05,
      "loss": 1.0133,
      "step": 256900
    },
    {
      "epoch": 0.23,
      "grad_norm": 628.0,
      "learning_rate": 3.8942447855741724e-05,
      "loss": 1.0403,
      "step": 257000
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.125,
      "learning_rate": 3.89379487825508e-05,
      "loss": 1.0205,
      "step": 257100
    },
    {
      "epoch": 0.23,
      "grad_norm": 39.0,
      "learning_rate": 3.8933449709359874e-05,
      "loss": 0.9979,
      "step": 257200
    },
    {
      "epoch": 0.23,
      "grad_norm": 31.875,
      "learning_rate": 3.8928950636168956e-05,
      "loss": 1.1494,
      "step": 257300
    },
    {
      "epoch": 0.23,
      "grad_norm": 68.5,
      "learning_rate": 3.8924451562978025e-05,
      "loss": 1.0635,
      "step": 257400
    },
    {
      "epoch": 0.23,
      "grad_norm": 62.0,
      "learning_rate": 3.891995248978711e-05,
      "loss": 1.1233,
      "step": 257500
    },
    {
      "epoch": 0.23,
      "grad_norm": 71.0,
      "learning_rate": 3.891545341659618e-05,
      "loss": 1.0237,
      "step": 257600
    },
    {
      "epoch": 0.23,
      "grad_norm": 21.625,
      "learning_rate": 3.891095434340526e-05,
      "loss": 1.0337,
      "step": 257700
    },
    {
      "epoch": 0.23,
      "grad_norm": 21.625,
      "learning_rate": 3.890645527021434e-05,
      "loss": 0.9592,
      "step": 257800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37890625,
      "learning_rate": 3.8901956197023415e-05,
      "loss": 1.0695,
      "step": 257900
    },
    {
      "epoch": 0.23,
      "grad_norm": 21.125,
      "learning_rate": 3.889745712383249e-05,
      "loss": 1.1847,
      "step": 258000
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.125,
      "learning_rate": 3.889295805064157e-05,
      "loss": 0.843,
      "step": 258100
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.375,
      "learning_rate": 3.888845897745065e-05,
      "loss": 1.1496,
      "step": 258200
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.75,
      "learning_rate": 3.888395990425972e-05,
      "loss": 0.9838,
      "step": 258300
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.3125,
      "learning_rate": 3.8879460831068805e-05,
      "loss": 0.9052,
      "step": 258400
    },
    {
      "epoch": 0.23,
      "grad_norm": 51.0,
      "learning_rate": 3.887496175787788e-05,
      "loss": 1.0382,
      "step": 258500
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.125,
      "learning_rate": 3.8870462684686956e-05,
      "loss": 1.0682,
      "step": 258600
    },
    {
      "epoch": 0.23,
      "grad_norm": 23.625,
      "learning_rate": 3.886596361149603e-05,
      "loss": 0.959,
      "step": 258700
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.384765625,
      "learning_rate": 3.886146453830511e-05,
      "loss": 1.1241,
      "step": 258800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.32421875,
      "learning_rate": 3.885696546511419e-05,
      "loss": 0.9618,
      "step": 258900
    },
    {
      "epoch": 0.23,
      "grad_norm": 90.5,
      "learning_rate": 3.8852466391923264e-05,
      "loss": 1.1003,
      "step": 259000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.01068115234375,
      "learning_rate": 3.884796731873234e-05,
      "loss": 1.1317,
      "step": 259100
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.75,
      "learning_rate": 3.884346824554142e-05,
      "loss": 1.0843,
      "step": 259200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.031005859375,
      "learning_rate": 3.88389691723505e-05,
      "loss": 1.0241,
      "step": 259300
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.0081787109375,
      "learning_rate": 3.883447009915957e-05,
      "loss": 1.0026,
      "step": 259400
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.0625,
      "learning_rate": 3.8829971025968654e-05,
      "loss": 0.978,
      "step": 259500
    },
    {
      "epoch": 0.23,
      "grad_norm": 50.25,
      "learning_rate": 3.882547195277773e-05,
      "loss": 1.0678,
      "step": 259600
    },
    {
      "epoch": 0.23,
      "grad_norm": 22.625,
      "learning_rate": 3.882097287958681e-05,
      "loss": 0.9762,
      "step": 259700
    },
    {
      "epoch": 0.23,
      "grad_norm": 69.5,
      "learning_rate": 3.881647380639589e-05,
      "loss": 1.0388,
      "step": 259800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2470703125,
      "learning_rate": 3.881197473320496e-05,
      "loss": 0.9259,
      "step": 259900
    },
    {
      "epoch": 0.23,
      "grad_norm": 19.0,
      "learning_rate": 3.880747566001404e-05,
      "loss": 1.1373,
      "step": 260000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.08642578125,
      "learning_rate": 3.880297658682311e-05,
      "loss": 0.9832,
      "step": 260100
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.75,
      "learning_rate": 3.8798477513632195e-05,
      "loss": 1.0725,
      "step": 260200
    },
    {
      "epoch": 0.23,
      "grad_norm": 39.0,
      "learning_rate": 3.879397844044127e-05,
      "loss": 1.0731,
      "step": 260300
    },
    {
      "epoch": 0.23,
      "grad_norm": 49.25,
      "learning_rate": 3.8789479367250346e-05,
      "loss": 1.0931,
      "step": 260400
    },
    {
      "epoch": 0.23,
      "grad_norm": 28.0,
      "learning_rate": 3.878498029405943e-05,
      "loss": 0.9126,
      "step": 260500
    },
    {
      "epoch": 0.23,
      "grad_norm": 96.5,
      "learning_rate": 3.8780481220868504e-05,
      "loss": 0.9054,
      "step": 260600
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.625,
      "learning_rate": 3.877598214767758e-05,
      "loss": 1.0186,
      "step": 260700
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.0224609375,
      "learning_rate": 3.877148307448666e-05,
      "loss": 0.9594,
      "step": 260800
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.125,
      "learning_rate": 3.8766984001295736e-05,
      "loss": 1.0147,
      "step": 260900
    },
    {
      "epoch": 0.23,
      "grad_norm": 39.5,
      "learning_rate": 3.876248492810481e-05,
      "loss": 0.989,
      "step": 261000
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.625,
      "learning_rate": 3.8757985854913894e-05,
      "loss": 0.9604,
      "step": 261100
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2392578125,
      "learning_rate": 3.875348678172297e-05,
      "loss": 1.0045,
      "step": 261200
    },
    {
      "epoch": 0.23,
      "grad_norm": 164.0,
      "learning_rate": 3.8748987708532044e-05,
      "loss": 1.1207,
      "step": 261300
    },
    {
      "epoch": 0.23,
      "grad_norm": 79.5,
      "learning_rate": 3.874448863534112e-05,
      "loss": 1.0898,
      "step": 261400
    },
    {
      "epoch": 0.23,
      "grad_norm": 118.5,
      "learning_rate": 3.8739989562150195e-05,
      "loss": 0.9858,
      "step": 261500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1171875,
      "learning_rate": 3.873549048895928e-05,
      "loss": 1.0268,
      "step": 261600
    },
    {
      "epoch": 0.23,
      "grad_norm": 58.25,
      "learning_rate": 3.873099141576835e-05,
      "loss": 1.0691,
      "step": 261700
    },
    {
      "epoch": 0.23,
      "grad_norm": 808.0,
      "learning_rate": 3.872649234257743e-05,
      "loss": 1.0907,
      "step": 261800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.498046875,
      "learning_rate": 3.872199326938651e-05,
      "loss": 1.0142,
      "step": 261900
    },
    {
      "epoch": 0.23,
      "grad_norm": 25.375,
      "learning_rate": 3.8717494196195585e-05,
      "loss": 1.1959,
      "step": 262000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.55859375,
      "learning_rate": 3.871299512300466e-05,
      "loss": 1.0116,
      "step": 262100
    },
    {
      "epoch": 0.23,
      "grad_norm": 38.75,
      "learning_rate": 3.870849604981374e-05,
      "loss": 0.8472,
      "step": 262200
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.8125,
      "learning_rate": 3.870399697662282e-05,
      "loss": 0.999,
      "step": 262300
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.0625,
      "learning_rate": 3.86994979034319e-05,
      "loss": 1.097,
      "step": 262400
    },
    {
      "epoch": 0.23,
      "grad_norm": 48.0,
      "learning_rate": 3.8694998830240976e-05,
      "loss": 0.9909,
      "step": 262500
    },
    {
      "epoch": 0.23,
      "grad_norm": 49.25,
      "learning_rate": 3.8690499757050044e-05,
      "loss": 0.9547,
      "step": 262600
    },
    {
      "epoch": 0.23,
      "grad_norm": 41.25,
      "learning_rate": 3.8686000683859126e-05,
      "loss": 0.8995,
      "step": 262700
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.007171630859375,
      "learning_rate": 3.86815016106682e-05,
      "loss": 1.1877,
      "step": 262800
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.25,
      "learning_rate": 3.8677002537477284e-05,
      "loss": 1.046,
      "step": 262900
    },
    {
      "epoch": 0.23,
      "grad_norm": 32.25,
      "learning_rate": 3.867250346428636e-05,
      "loss": 0.9654,
      "step": 263000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.00732421875,
      "learning_rate": 3.8668004391095434e-05,
      "loss": 0.9455,
      "step": 263100
    },
    {
      "epoch": 0.23,
      "grad_norm": 24.375,
      "learning_rate": 3.8663505317904517e-05,
      "loss": 1.0026,
      "step": 263200
    },
    {
      "epoch": 0.23,
      "grad_norm": 21.625,
      "learning_rate": 3.865900624471359e-05,
      "loss": 1.0009,
      "step": 263300
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1630859375,
      "learning_rate": 3.865450717152267e-05,
      "loss": 0.9792,
      "step": 263400
    },
    {
      "epoch": 0.23,
      "grad_norm": 66.5,
      "learning_rate": 3.865000809833175e-05,
      "loss": 1.1434,
      "step": 263500
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.375,
      "learning_rate": 3.8645509025140825e-05,
      "loss": 1.0509,
      "step": 263600
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.25,
      "learning_rate": 3.86410099519499e-05,
      "loss": 0.9607,
      "step": 263700
    },
    {
      "epoch": 0.23,
      "grad_norm": 43.75,
      "learning_rate": 3.863651087875898e-05,
      "loss": 1.1413,
      "step": 263800
    },
    {
      "epoch": 0.24,
      "grad_norm": 49.75,
      "learning_rate": 3.863201180556805e-05,
      "loss": 1.0306,
      "step": 263900
    },
    {
      "epoch": 0.24,
      "grad_norm": 46.25,
      "learning_rate": 3.862751273237713e-05,
      "loss": 1.0876,
      "step": 264000
    },
    {
      "epoch": 0.24,
      "grad_norm": 75.5,
      "learning_rate": 3.862301365918621e-05,
      "loss": 1.1211,
      "step": 264100
    },
    {
      "epoch": 0.24,
      "grad_norm": 62.5,
      "learning_rate": 3.8618514585995284e-05,
      "loss": 1.016,
      "step": 264200
    },
    {
      "epoch": 0.24,
      "grad_norm": 38.0,
      "learning_rate": 3.8614015512804366e-05,
      "loss": 0.8601,
      "step": 264300
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.5,
      "learning_rate": 3.860951643961344e-05,
      "loss": 1.0999,
      "step": 264400
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.02978515625,
      "learning_rate": 3.8605017366422516e-05,
      "loss": 1.0766,
      "step": 264500
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.5,
      "learning_rate": 3.86005182932316e-05,
      "loss": 0.9753,
      "step": 264600
    },
    {
      "epoch": 0.24,
      "grad_norm": 78.0,
      "learning_rate": 3.8596019220040674e-05,
      "loss": 0.9668,
      "step": 264700
    },
    {
      "epoch": 0.24,
      "grad_norm": 19.25,
      "learning_rate": 3.859152014684975e-05,
      "loss": 0.8695,
      "step": 264800
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.75,
      "learning_rate": 3.858702107365883e-05,
      "loss": 1.0599,
      "step": 264900
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.125,
      "learning_rate": 3.8582522000467907e-05,
      "loss": 1.028,
      "step": 265000
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.5,
      "learning_rate": 3.857802292727699e-05,
      "loss": 1.1109,
      "step": 265100
    },
    {
      "epoch": 0.24,
      "grad_norm": 37.5,
      "learning_rate": 3.857352385408606e-05,
      "loss": 0.9585,
      "step": 265200
    },
    {
      "epoch": 0.24,
      "grad_norm": 108.0,
      "learning_rate": 3.856902478089513e-05,
      "loss": 0.8214,
      "step": 265300
    },
    {
      "epoch": 0.24,
      "grad_norm": 63.0,
      "learning_rate": 3.8564525707704215e-05,
      "loss": 1.0191,
      "step": 265400
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.05517578125,
      "learning_rate": 3.856002663451329e-05,
      "loss": 1.0374,
      "step": 265500
    },
    {
      "epoch": 0.24,
      "grad_norm": 43.75,
      "learning_rate": 3.8555527561322365e-05,
      "loss": 1.1707,
      "step": 265600
    },
    {
      "epoch": 0.24,
      "grad_norm": 52.25,
      "learning_rate": 3.855102848813145e-05,
      "loss": 1.0462,
      "step": 265700
    },
    {
      "epoch": 0.24,
      "grad_norm": 21.0,
      "learning_rate": 3.854652941494052e-05,
      "loss": 1.0787,
      "step": 265800
    },
    {
      "epoch": 0.24,
      "grad_norm": 21.125,
      "learning_rate": 3.8542030341749605e-05,
      "loss": 1.0676,
      "step": 265900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.0201416015625,
      "learning_rate": 3.853753126855868e-05,
      "loss": 1.0169,
      "step": 266000
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.4375,
      "learning_rate": 3.8533032195367756e-05,
      "loss": 0.9729,
      "step": 266100
    },
    {
      "epoch": 0.24,
      "grad_norm": 78.5,
      "learning_rate": 3.852853312217684e-05,
      "loss": 0.9858,
      "step": 266200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.0216064453125,
      "learning_rate": 3.852403404898591e-05,
      "loss": 1.1963,
      "step": 266300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.040771484375,
      "learning_rate": 3.851953497579499e-05,
      "loss": 0.9544,
      "step": 266400
    },
    {
      "epoch": 0.24,
      "grad_norm": 19.25,
      "learning_rate": 3.8515035902604064e-05,
      "loss": 1.0033,
      "step": 266500
    },
    {
      "epoch": 0.24,
      "grad_norm": 28.0,
      "learning_rate": 3.851053682941314e-05,
      "loss": 1.0625,
      "step": 266600
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.3125,
      "learning_rate": 3.850603775622222e-05,
      "loss": 0.873,
      "step": 266700
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.625,
      "learning_rate": 3.8501538683031297e-05,
      "loss": 1.0363,
      "step": 266800
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.5,
      "learning_rate": 3.849703960984037e-05,
      "loss": 1.035,
      "step": 266900
    },
    {
      "epoch": 0.24,
      "grad_norm": 215.0,
      "learning_rate": 3.8492540536649454e-05,
      "loss": 1.0572,
      "step": 267000
    },
    {
      "epoch": 0.24,
      "grad_norm": 21.875,
      "learning_rate": 3.848804146345853e-05,
      "loss": 1.116,
      "step": 267100
    },
    {
      "epoch": 0.24,
      "grad_norm": 101.0,
      "learning_rate": 3.8483542390267605e-05,
      "loss": 1.0954,
      "step": 267200
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.234375,
      "learning_rate": 3.847904331707669e-05,
      "loss": 1.0449,
      "step": 267300
    },
    {
      "epoch": 0.24,
      "grad_norm": 47.25,
      "learning_rate": 3.847454424388576e-05,
      "loss": 1.055,
      "step": 267400
    },
    {
      "epoch": 0.24,
      "grad_norm": 23.375,
      "learning_rate": 3.847004517069484e-05,
      "loss": 0.9842,
      "step": 267500
    },
    {
      "epoch": 0.24,
      "grad_norm": 20.25,
      "learning_rate": 3.846554609750392e-05,
      "loss": 1.1345,
      "step": 267600
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.189453125,
      "learning_rate": 3.8461047024312995e-05,
      "loss": 1.0896,
      "step": 267700
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.0,
      "learning_rate": 3.845654795112207e-05,
      "loss": 0.9783,
      "step": 267800
    },
    {
      "epoch": 0.24,
      "grad_norm": 23.0,
      "learning_rate": 3.8452048877931146e-05,
      "loss": 1.0952,
      "step": 267900
    },
    {
      "epoch": 0.24,
      "grad_norm": 106.0,
      "learning_rate": 3.844754980474022e-05,
      "loss": 1.0655,
      "step": 268000
    },
    {
      "epoch": 0.24,
      "grad_norm": 38.0,
      "learning_rate": 3.84430507315493e-05,
      "loss": 0.9758,
      "step": 268100
    },
    {
      "epoch": 0.24,
      "grad_norm": 54.5,
      "learning_rate": 3.843855165835838e-05,
      "loss": 0.989,
      "step": 268200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.0062255859375,
      "learning_rate": 3.8434052585167454e-05,
      "loss": 1.0824,
      "step": 268300
    },
    {
      "epoch": 0.24,
      "grad_norm": 47.75,
      "learning_rate": 3.8429553511976536e-05,
      "loss": 0.8245,
      "step": 268400
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12255859375,
      "learning_rate": 3.842505443878561e-05,
      "loss": 0.8582,
      "step": 268500
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.0269775390625,
      "learning_rate": 3.842055536559469e-05,
      "loss": 0.9881,
      "step": 268600
    },
    {
      "epoch": 0.24,
      "grad_norm": 93.0,
      "learning_rate": 3.841605629240377e-05,
      "loss": 0.9859,
      "step": 268700
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.125,
      "learning_rate": 3.8411557219212844e-05,
      "loss": 0.9935,
      "step": 268800
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.158203125,
      "learning_rate": 3.8407058146021926e-05,
      "loss": 0.8388,
      "step": 268900
    },
    {
      "epoch": 0.24,
      "grad_norm": 35.0,
      "learning_rate": 3.8402559072831e-05,
      "loss": 1.0776,
      "step": 269000
    },
    {
      "epoch": 0.24,
      "grad_norm": 53.75,
      "learning_rate": 3.839805999964008e-05,
      "loss": 1.0681,
      "step": 269100
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.75,
      "learning_rate": 3.839356092644915e-05,
      "loss": 0.9243,
      "step": 269200
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.875,
      "learning_rate": 3.838906185325823e-05,
      "loss": 0.9357,
      "step": 269300
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.875,
      "learning_rate": 3.838456278006731e-05,
      "loss": 1.041,
      "step": 269400
    },
    {
      "epoch": 0.24,
      "grad_norm": 35.0,
      "learning_rate": 3.8380063706876385e-05,
      "loss": 1.0491,
      "step": 269500
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.875,
      "learning_rate": 3.837556463368546e-05,
      "loss": 0.9161,
      "step": 269600
    },
    {
      "epoch": 0.24,
      "grad_norm": 41.75,
      "learning_rate": 3.837106556049454e-05,
      "loss": 1.0651,
      "step": 269700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.234375,
      "learning_rate": 3.836656648730362e-05,
      "loss": 1.1058,
      "step": 269800
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.01171875,
      "learning_rate": 3.836206741411269e-05,
      "loss": 0.8572,
      "step": 269900
    },
    {
      "epoch": 0.24,
      "grad_norm": 66.0,
      "learning_rate": 3.8357568340921775e-05,
      "loss": 1.1186,
      "step": 270000
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1591796875,
      "learning_rate": 3.835306926773085e-05,
      "loss": 0.9725,
      "step": 270100
    },
    {
      "epoch": 0.24,
      "grad_norm": 25.75,
      "learning_rate": 3.8348570194539926e-05,
      "loss": 1.0514,
      "step": 270200
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.8125,
      "learning_rate": 3.834407112134901e-05,
      "loss": 1.0911,
      "step": 270300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.005950927734375,
      "learning_rate": 3.8339572048158077e-05,
      "loss": 0.9847,
      "step": 270400
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.6875,
      "learning_rate": 3.833507297496716e-05,
      "loss": 0.9347,
      "step": 270500
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.0,
      "learning_rate": 3.8330573901776234e-05,
      "loss": 1.045,
      "step": 270600
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.189453125,
      "learning_rate": 3.832607482858531e-05,
      "loss": 1.0941,
      "step": 270700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.832157575539439e-05,
      "loss": 1.0154,
      "step": 270800
    },
    {
      "epoch": 0.24,
      "grad_norm": 36.25,
      "learning_rate": 3.831707668220347e-05,
      "loss": 0.888,
      "step": 270900
    },
    {
      "epoch": 0.24,
      "grad_norm": 105.5,
      "learning_rate": 3.831257760901254e-05,
      "loss": 0.9389,
      "step": 271000
    },
    {
      "epoch": 0.24,
      "grad_norm": 178.0,
      "learning_rate": 3.8308078535821624e-05,
      "loss": 0.8784,
      "step": 271100
    },
    {
      "epoch": 0.24,
      "grad_norm": 47.75,
      "learning_rate": 3.83035794626307e-05,
      "loss": 0.894,
      "step": 271200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.546875,
      "learning_rate": 3.829908038943978e-05,
      "loss": 1.119,
      "step": 271300
    },
    {
      "epoch": 0.24,
      "grad_norm": 34.75,
      "learning_rate": 3.829458131624886e-05,
      "loss": 1.0578,
      "step": 271400
    },
    {
      "epoch": 0.24,
      "grad_norm": 96.0,
      "learning_rate": 3.829008224305793e-05,
      "loss": 1.1486,
      "step": 271500
    },
    {
      "epoch": 0.24,
      "grad_norm": 33.0,
      "learning_rate": 3.8285583169867015e-05,
      "loss": 0.8229,
      "step": 271600
    },
    {
      "epoch": 0.24,
      "grad_norm": 32.25,
      "learning_rate": 3.828108409667608e-05,
      "loss": 0.8392,
      "step": 271700
    },
    {
      "epoch": 0.24,
      "grad_norm": 19.625,
      "learning_rate": 3.8276585023485165e-05,
      "loss": 0.9758,
      "step": 271800
    },
    {
      "epoch": 0.24,
      "grad_norm": 24.625,
      "learning_rate": 3.827208595029424e-05,
      "loss": 1.055,
      "step": 271900
    },
    {
      "epoch": 0.24,
      "grad_norm": 91.0,
      "learning_rate": 3.8267586877103316e-05,
      "loss": 0.954,
      "step": 272000
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.640625,
      "learning_rate": 3.82630878039124e-05,
      "loss": 1.0351,
      "step": 272100
    },
    {
      "epoch": 0.24,
      "grad_norm": 39.75,
      "learning_rate": 3.825858873072147e-05,
      "loss": 1.0571,
      "step": 272200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3203125,
      "learning_rate": 3.825408965753055e-05,
      "loss": 1.0111,
      "step": 272300
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.8125,
      "learning_rate": 3.824959058433963e-05,
      "loss": 1.0805,
      "step": 272400
    },
    {
      "epoch": 0.24,
      "grad_norm": 23.375,
      "learning_rate": 3.8245091511148706e-05,
      "loss": 0.957,
      "step": 272500
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.8125,
      "learning_rate": 3.824059243795778e-05,
      "loss": 0.9219,
      "step": 272600
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.625,
      "learning_rate": 3.8236093364766864e-05,
      "loss": 1.0054,
      "step": 272700
    },
    {
      "epoch": 0.24,
      "grad_norm": 166.0,
      "learning_rate": 3.823159429157594e-05,
      "loss": 1.0138,
      "step": 272800
    },
    {
      "epoch": 0.24,
      "grad_norm": 157.0,
      "learning_rate": 3.8227095218385014e-05,
      "loss": 1.0631,
      "step": 272900
    },
    {
      "epoch": 0.24,
      "grad_norm": 38.75,
      "learning_rate": 3.822259614519409e-05,
      "loss": 0.9157,
      "step": 273000
    },
    {
      "epoch": 0.24,
      "grad_norm": 23.625,
      "learning_rate": 3.8218097072003165e-05,
      "loss": 0.9601,
      "step": 273100
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.291015625,
      "learning_rate": 3.821359799881225e-05,
      "loss": 0.8231,
      "step": 273200
    },
    {
      "epoch": 0.24,
      "grad_norm": 29.875,
      "learning_rate": 3.820909892562132e-05,
      "loss": 1.0122,
      "step": 273300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.82045998524304e-05,
      "loss": 1.0578,
      "step": 273400
    },
    {
      "epoch": 0.24,
      "grad_norm": 45.0,
      "learning_rate": 3.820010077923948e-05,
      "loss": 0.9441,
      "step": 273500
    },
    {
      "epoch": 0.24,
      "grad_norm": 139.0,
      "learning_rate": 3.8195601706048555e-05,
      "loss": 0.9708,
      "step": 273600
    },
    {
      "epoch": 0.24,
      "grad_norm": 29.625,
      "learning_rate": 3.819110263285763e-05,
      "loss": 1.0925,
      "step": 273700
    },
    {
      "epoch": 0.24,
      "grad_norm": 30.75,
      "learning_rate": 3.818660355966671e-05,
      "loss": 1.1019,
      "step": 273800
    },
    {
      "epoch": 0.24,
      "grad_norm": 37.0,
      "learning_rate": 3.818210448647579e-05,
      "loss": 0.8965,
      "step": 273900
    },
    {
      "epoch": 0.24,
      "grad_norm": 41.75,
      "learning_rate": 3.817760541328487e-05,
      "loss": 0.8904,
      "step": 274000
    },
    {
      "epoch": 0.24,
      "grad_norm": 108.5,
      "learning_rate": 3.8173106340093945e-05,
      "loss": 0.8578,
      "step": 274100
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.5625,
      "learning_rate": 3.816860726690302e-05,
      "loss": 0.9038,
      "step": 274200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.00384521484375,
      "learning_rate": 3.8164108193712096e-05,
      "loss": 0.9433,
      "step": 274300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12255859375,
      "learning_rate": 3.815960912052117e-05,
      "loss": 0.9756,
      "step": 274400
    },
    {
      "epoch": 0.24,
      "grad_norm": 64.5,
      "learning_rate": 3.8155110047330254e-05,
      "loss": 0.9267,
      "step": 274500
    },
    {
      "epoch": 0.24,
      "grad_norm": 220.0,
      "learning_rate": 3.815061097413933e-05,
      "loss": 1.0132,
      "step": 274600
    },
    {
      "epoch": 0.24,
      "grad_norm": 552.0,
      "learning_rate": 3.8146111900948404e-05,
      "loss": 0.937,
      "step": 274700
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.96875,
      "learning_rate": 3.8141612827757486e-05,
      "loss": 1.0196,
      "step": 274800
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.01904296875,
      "learning_rate": 3.813711375456656e-05,
      "loss": 0.9484,
      "step": 274900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.08154296875,
      "learning_rate": 3.813261468137564e-05,
      "loss": 0.9945,
      "step": 275000
    },
    {
      "epoch": 0.25,
      "grad_norm": 572.0,
      "learning_rate": 3.812811560818472e-05,
      "loss": 1.0876,
      "step": 275100
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1953125,
      "learning_rate": 3.8123616534993795e-05,
      "loss": 0.8657,
      "step": 275200
    },
    {
      "epoch": 0.25,
      "grad_norm": 37.75,
      "learning_rate": 3.811911746180287e-05,
      "loss": 0.9201,
      "step": 275300
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.375,
      "learning_rate": 3.811461838861195e-05,
      "loss": 1.115,
      "step": 275400
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.201171875,
      "learning_rate": 3.811011931542103e-05,
      "loss": 0.9549,
      "step": 275500
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.375,
      "learning_rate": 3.81056202422301e-05,
      "loss": 0.9049,
      "step": 275600
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.46875,
      "learning_rate": 3.810112116903918e-05,
      "loss": 0.9228,
      "step": 275700
    },
    {
      "epoch": 0.25,
      "grad_norm": 186.0,
      "learning_rate": 3.809662209584825e-05,
      "loss": 0.9166,
      "step": 275800
    },
    {
      "epoch": 0.25,
      "grad_norm": 67.5,
      "learning_rate": 3.8092123022657335e-05,
      "loss": 0.9586,
      "step": 275900
    },
    {
      "epoch": 0.25,
      "grad_norm": 41.75,
      "learning_rate": 3.808762394946641e-05,
      "loss": 0.9645,
      "step": 276000
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.375,
      "learning_rate": 3.8083124876275486e-05,
      "loss": 1.1052,
      "step": 276100
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.0,
      "learning_rate": 3.807862580308457e-05,
      "loss": 0.9588,
      "step": 276200
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.875,
      "learning_rate": 3.8074126729893644e-05,
      "loss": 1.0982,
      "step": 276300
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.875,
      "learning_rate": 3.806962765670272e-05,
      "loss": 0.9041,
      "step": 276400
    },
    {
      "epoch": 0.25,
      "grad_norm": 188.0,
      "learning_rate": 3.80651285835118e-05,
      "loss": 1.0159,
      "step": 276500
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.25,
      "learning_rate": 3.8060629510320876e-05,
      "loss": 1.0217,
      "step": 276600
    },
    {
      "epoch": 0.25,
      "grad_norm": 129.0,
      "learning_rate": 3.805613043712996e-05,
      "loss": 1.0292,
      "step": 276700
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.0,
      "learning_rate": 3.805163136393903e-05,
      "loss": 1.0513,
      "step": 276800
    },
    {
      "epoch": 0.25,
      "grad_norm": 48.75,
      "learning_rate": 3.80471322907481e-05,
      "loss": 0.9502,
      "step": 276900
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.625,
      "learning_rate": 3.8042633217557185e-05,
      "loss": 1.0215,
      "step": 277000
    },
    {
      "epoch": 0.25,
      "grad_norm": 30.125,
      "learning_rate": 3.803813414436626e-05,
      "loss": 1.1013,
      "step": 277100
    },
    {
      "epoch": 0.25,
      "grad_norm": 44.5,
      "learning_rate": 3.803363507117534e-05,
      "loss": 1.0524,
      "step": 277200
    },
    {
      "epoch": 0.25,
      "grad_norm": 25.5,
      "learning_rate": 3.802913599798442e-05,
      "loss": 0.8994,
      "step": 277300
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.171875,
      "learning_rate": 3.802463692479349e-05,
      "loss": 0.9864,
      "step": 277400
    },
    {
      "epoch": 0.25,
      "grad_norm": 106.0,
      "learning_rate": 3.8020137851602575e-05,
      "loss": 1.0491,
      "step": 277500
    },
    {
      "epoch": 0.25,
      "grad_norm": 33.75,
      "learning_rate": 3.801563877841165e-05,
      "loss": 1.0561,
      "step": 277600
    },
    {
      "epoch": 0.25,
      "grad_norm": 31.75,
      "learning_rate": 3.8011139705220725e-05,
      "loss": 0.9155,
      "step": 277700
    },
    {
      "epoch": 0.25,
      "grad_norm": 37.75,
      "learning_rate": 3.800664063202981e-05,
      "loss": 0.8815,
      "step": 277800
    },
    {
      "epoch": 0.25,
      "grad_norm": 57.75,
      "learning_rate": 3.800214155883888e-05,
      "loss": 1.0203,
      "step": 277900
    },
    {
      "epoch": 0.25,
      "grad_norm": 41.0,
      "learning_rate": 3.799764248564796e-05,
      "loss": 0.8779,
      "step": 278000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.05908203125,
      "learning_rate": 3.7993143412457034e-05,
      "loss": 0.9544,
      "step": 278100
    },
    {
      "epoch": 0.25,
      "grad_norm": 42.5,
      "learning_rate": 3.798864433926611e-05,
      "loss": 0.8568,
      "step": 278200
    },
    {
      "epoch": 0.25,
      "grad_norm": 32.75,
      "learning_rate": 3.798414526607519e-05,
      "loss": 1.054,
      "step": 278300
    },
    {
      "epoch": 0.25,
      "grad_norm": 53.25,
      "learning_rate": 3.7979646192884266e-05,
      "loss": 1.0164,
      "step": 278400
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.625,
      "learning_rate": 3.797514711969334e-05,
      "loss": 0.9564,
      "step": 278500
    },
    {
      "epoch": 0.25,
      "grad_norm": 59.75,
      "learning_rate": 3.7970648046502424e-05,
      "loss": 1.1013,
      "step": 278600
    },
    {
      "epoch": 0.25,
      "grad_norm": 79.0,
      "learning_rate": 3.79661489733115e-05,
      "loss": 1.0964,
      "step": 278700
    },
    {
      "epoch": 0.25,
      "grad_norm": 63.75,
      "learning_rate": 3.7961649900120575e-05,
      "loss": 1.0402,
      "step": 278800
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.0,
      "learning_rate": 3.795715082692966e-05,
      "loss": 1.015,
      "step": 278900
    },
    {
      "epoch": 0.25,
      "grad_norm": 54.0,
      "learning_rate": 3.795265175373873e-05,
      "loss": 0.9332,
      "step": 279000
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.859375,
      "learning_rate": 3.794815268054781e-05,
      "loss": 1.0376,
      "step": 279100
    },
    {
      "epoch": 0.25,
      "grad_norm": 49.75,
      "learning_rate": 3.794365360735689e-05,
      "loss": 1.0458,
      "step": 279200
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.375,
      "learning_rate": 3.7939154534165965e-05,
      "loss": 1.0153,
      "step": 279300
    },
    {
      "epoch": 0.25,
      "grad_norm": 41.75,
      "learning_rate": 3.793465546097504e-05,
      "loss": 1.0531,
      "step": 279400
    },
    {
      "epoch": 0.25,
      "grad_norm": 40.75,
      "learning_rate": 3.7930156387784115e-05,
      "loss": 0.9474,
      "step": 279500
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.75,
      "learning_rate": 3.792565731459319e-05,
      "loss": 1.1138,
      "step": 279600
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.375,
      "learning_rate": 3.792115824140227e-05,
      "loss": 1.1098,
      "step": 279700
    },
    {
      "epoch": 0.25,
      "grad_norm": 147.0,
      "learning_rate": 3.791665916821135e-05,
      "loss": 1.1804,
      "step": 279800
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.5,
      "learning_rate": 3.791216009502043e-05,
      "loss": 0.9355,
      "step": 279900
    },
    {
      "epoch": 0.25,
      "grad_norm": 38.0,
      "learning_rate": 3.7907661021829506e-05,
      "loss": 1.0924,
      "step": 280000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.310546875,
      "learning_rate": 3.790316194863858e-05,
      "loss": 1.0058,
      "step": 280100
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.75,
      "learning_rate": 3.789866287544766e-05,
      "loss": 1.0419,
      "step": 280200
    },
    {
      "epoch": 0.25,
      "grad_norm": 176.0,
      "learning_rate": 3.789416380225674e-05,
      "loss": 0.8807,
      "step": 280300
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2734375,
      "learning_rate": 3.7889664729065814e-05,
      "loss": 1.2521,
      "step": 280400
    },
    {
      "epoch": 0.25,
      "grad_norm": 39.0,
      "learning_rate": 3.7885165655874896e-05,
      "loss": 1.0638,
      "step": 280500
    },
    {
      "epoch": 0.25,
      "grad_norm": 50.75,
      "learning_rate": 3.788066658268397e-05,
      "loss": 0.9331,
      "step": 280600
    },
    {
      "epoch": 0.25,
      "grad_norm": 43.0,
      "learning_rate": 3.787616750949305e-05,
      "loss": 1.0003,
      "step": 280700
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.125,
      "learning_rate": 3.787166843630212e-05,
      "loss": 1.0782,
      "step": 280800
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.036865234375,
      "learning_rate": 3.78671693631112e-05,
      "loss": 1.0198,
      "step": 280900
    },
    {
      "epoch": 0.25,
      "grad_norm": 20.625,
      "learning_rate": 3.786267028992028e-05,
      "loss": 1.018,
      "step": 281000
    },
    {
      "epoch": 0.25,
      "grad_norm": 36.5,
      "learning_rate": 3.7858171216729355e-05,
      "loss": 0.9739,
      "step": 281100
    },
    {
      "epoch": 0.25,
      "grad_norm": 49.25,
      "learning_rate": 3.785367214353843e-05,
      "loss": 1.1613,
      "step": 281200
    },
    {
      "epoch": 0.25,
      "grad_norm": 25.125,
      "learning_rate": 3.784917307034751e-05,
      "loss": 0.9868,
      "step": 281300
    },
    {
      "epoch": 0.25,
      "grad_norm": 134.0,
      "learning_rate": 3.784467399715659e-05,
      "loss": 0.9122,
      "step": 281400
    },
    {
      "epoch": 0.25,
      "grad_norm": 33.25,
      "learning_rate": 3.784017492396566e-05,
      "loss": 0.8729,
      "step": 281500
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4140625,
      "learning_rate": 3.7835675850774745e-05,
      "loss": 1.0995,
      "step": 281600
    },
    {
      "epoch": 0.25,
      "grad_norm": 25.375,
      "learning_rate": 3.783117677758382e-05,
      "loss": 0.9733,
      "step": 281700
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1455078125,
      "learning_rate": 3.7826677704392896e-05,
      "loss": 1.1197,
      "step": 281800
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.9375,
      "learning_rate": 3.782217863120198e-05,
      "loss": 0.8921,
      "step": 281900
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.25,
      "learning_rate": 3.7817679558011046e-05,
      "loss": 0.9566,
      "step": 282000
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.5,
      "learning_rate": 3.781318048482013e-05,
      "loss": 1.1232,
      "step": 282100
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.625,
      "learning_rate": 3.7808681411629204e-05,
      "loss": 1.1001,
      "step": 282200
    },
    {
      "epoch": 0.25,
      "grad_norm": 38.0,
      "learning_rate": 3.780418233843828e-05,
      "loss": 1.0222,
      "step": 282300
    },
    {
      "epoch": 0.25,
      "grad_norm": 32.25,
      "learning_rate": 3.779968326524736e-05,
      "loss": 0.9127,
      "step": 282400
    },
    {
      "epoch": 0.25,
      "grad_norm": 127.5,
      "learning_rate": 3.779518419205644e-05,
      "loss": 1.1838,
      "step": 282500
    },
    {
      "epoch": 0.25,
      "grad_norm": 184.0,
      "learning_rate": 3.779068511886551e-05,
      "loss": 0.9834,
      "step": 282600
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.75,
      "learning_rate": 3.7786186045674594e-05,
      "loss": 1.097,
      "step": 282700
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.03662109375,
      "learning_rate": 3.778168697248367e-05,
      "loss": 0.9354,
      "step": 282800
    },
    {
      "epoch": 0.25,
      "grad_norm": 42.75,
      "learning_rate": 3.777718789929275e-05,
      "loss": 1.0269,
      "step": 282900
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.5625,
      "learning_rate": 3.777268882610183e-05,
      "loss": 1.0666,
      "step": 283000
    },
    {
      "epoch": 0.25,
      "grad_norm": 57.0,
      "learning_rate": 3.77681897529109e-05,
      "loss": 1.0767,
      "step": 283100
    },
    {
      "epoch": 0.25,
      "grad_norm": 113.0,
      "learning_rate": 3.7763690679719984e-05,
      "loss": 0.9977,
      "step": 283200
    },
    {
      "epoch": 0.25,
      "grad_norm": 109.0,
      "learning_rate": 3.775919160652905e-05,
      "loss": 0.9438,
      "step": 283300
    },
    {
      "epoch": 0.25,
      "grad_norm": 51.25,
      "learning_rate": 3.7754692533338135e-05,
      "loss": 1.0746,
      "step": 283400
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.07080078125,
      "learning_rate": 3.775019346014721e-05,
      "loss": 0.9062,
      "step": 283500
    },
    {
      "epoch": 0.25,
      "grad_norm": 74.0,
      "learning_rate": 3.7745694386956286e-05,
      "loss": 0.9957,
      "step": 283600
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.375,
      "learning_rate": 3.774119531376537e-05,
      "loss": 1.0649,
      "step": 283700
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.875,
      "learning_rate": 3.773669624057444e-05,
      "loss": 0.8601,
      "step": 283800
    },
    {
      "epoch": 0.25,
      "grad_norm": 60.25,
      "learning_rate": 3.773219716738352e-05,
      "loss": 0.9882,
      "step": 283900
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.125,
      "learning_rate": 3.77276980941926e-05,
      "loss": 0.9612,
      "step": 284000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.7723199021001676e-05,
      "loss": 0.8524,
      "step": 284100
    },
    {
      "epoch": 0.25,
      "grad_norm": 69.5,
      "learning_rate": 3.771869994781075e-05,
      "loss": 0.9708,
      "step": 284200
    },
    {
      "epoch": 0.25,
      "grad_norm": 17.25,
      "learning_rate": 3.7714200874619833e-05,
      "loss": 1.0929,
      "step": 284300
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.875,
      "learning_rate": 3.770970180142891e-05,
      "loss": 1.1401,
      "step": 284400
    },
    {
      "epoch": 0.25,
      "grad_norm": 118.5,
      "learning_rate": 3.7705202728237984e-05,
      "loss": 1.1539,
      "step": 284500
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.28125,
      "learning_rate": 3.770070365504706e-05,
      "loss": 0.988,
      "step": 284600
    },
    {
      "epoch": 0.25,
      "grad_norm": 42.25,
      "learning_rate": 3.7696204581856135e-05,
      "loss": 1.018,
      "step": 284700
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.5,
      "learning_rate": 3.769170550866522e-05,
      "loss": 0.9587,
      "step": 284800
    },
    {
      "epoch": 0.25,
      "grad_norm": 72.5,
      "learning_rate": 3.768720643547429e-05,
      "loss": 0.9842,
      "step": 284900
    },
    {
      "epoch": 0.25,
      "grad_norm": 43.5,
      "learning_rate": 3.768270736228337e-05,
      "loss": 1.055,
      "step": 285000
    },
    {
      "epoch": 0.25,
      "grad_norm": 55.0,
      "learning_rate": 3.767820828909245e-05,
      "loss": 0.8903,
      "step": 285100
    },
    {
      "epoch": 0.25,
      "grad_norm": 60.5,
      "learning_rate": 3.7673709215901525e-05,
      "loss": 0.9298,
      "step": 285200
    },
    {
      "epoch": 0.25,
      "grad_norm": 20.125,
      "learning_rate": 3.76692101427106e-05,
      "loss": 0.9636,
      "step": 285300
    },
    {
      "epoch": 0.25,
      "grad_norm": 18.625,
      "learning_rate": 3.766471106951968e-05,
      "loss": 0.9638,
      "step": 285400
    },
    {
      "epoch": 0.25,
      "grad_norm": 110.0,
      "learning_rate": 3.766021199632876e-05,
      "loss": 1.0231,
      "step": 285500
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.625,
      "learning_rate": 3.765571292313784e-05,
      "loss": 1.0215,
      "step": 285600
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.625,
      "learning_rate": 3.7651213849946915e-05,
      "loss": 0.9749,
      "step": 285700
    },
    {
      "epoch": 0.25,
      "grad_norm": 41.75,
      "learning_rate": 3.764671477675599e-05,
      "loss": 1.0126,
      "step": 285800
    },
    {
      "epoch": 0.25,
      "grad_norm": 17.625,
      "learning_rate": 3.7642215703565066e-05,
      "loss": 1.0122,
      "step": 285900
    },
    {
      "epoch": 0.25,
      "grad_norm": 33.25,
      "learning_rate": 3.763771663037414e-05,
      "loss": 0.9669,
      "step": 286000
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.125,
      "learning_rate": 3.7633217557183223e-05,
      "loss": 0.9164,
      "step": 286100
    },
    {
      "epoch": 0.25,
      "grad_norm": 18.5,
      "learning_rate": 3.76287184839923e-05,
      "loss": 1.0143,
      "step": 286200
    },
    {
      "epoch": 0.26,
      "grad_norm": 18.5,
      "learning_rate": 3.7624219410801374e-05,
      "loss": 0.9201,
      "step": 286300
    },
    {
      "epoch": 0.26,
      "grad_norm": 119.0,
      "learning_rate": 3.7619720337610456e-05,
      "loss": 1.0825,
      "step": 286400
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.1875,
      "learning_rate": 3.761522126441953e-05,
      "loss": 1.0808,
      "step": 286500
    },
    {
      "epoch": 0.26,
      "grad_norm": 70.5,
      "learning_rate": 3.761072219122861e-05,
      "loss": 0.8891,
      "step": 286600
    },
    {
      "epoch": 0.26,
      "grad_norm": 134.0,
      "learning_rate": 3.760622311803769e-05,
      "loss": 0.9578,
      "step": 286700
    },
    {
      "epoch": 0.26,
      "grad_norm": 21.875,
      "learning_rate": 3.7601724044846764e-05,
      "loss": 1.0709,
      "step": 286800
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.020263671875,
      "learning_rate": 3.759722497165584e-05,
      "loss": 1.1112,
      "step": 286900
    },
    {
      "epoch": 0.26,
      "grad_norm": 37.25,
      "learning_rate": 3.759272589846492e-05,
      "loss": 1.0595,
      "step": 287000
    },
    {
      "epoch": 0.26,
      "grad_norm": 42.25,
      "learning_rate": 3.7588226825274e-05,
      "loss": 1.0129,
      "step": 287100
    },
    {
      "epoch": 0.26,
      "grad_norm": 28.5,
      "learning_rate": 3.758372775208307e-05,
      "loss": 1.0426,
      "step": 287200
    },
    {
      "epoch": 0.26,
      "grad_norm": 75.5,
      "learning_rate": 3.757922867889215e-05,
      "loss": 0.9792,
      "step": 287300
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.02392578125,
      "learning_rate": 3.757472960570122e-05,
      "loss": 1.0452,
      "step": 287400
    },
    {
      "epoch": 0.26,
      "grad_norm": 28.0,
      "learning_rate": 3.7570230532510305e-05,
      "loss": 1.0643,
      "step": 287500
    },
    {
      "epoch": 0.26,
      "grad_norm": 30.5,
      "learning_rate": 3.756573145931938e-05,
      "loss": 1.0373,
      "step": 287600
    },
    {
      "epoch": 0.26,
      "grad_norm": 27.875,
      "learning_rate": 3.7561232386128456e-05,
      "loss": 0.9409,
      "step": 287700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1416015625,
      "learning_rate": 3.755673331293754e-05,
      "loss": 0.9208,
      "step": 287800
    },
    {
      "epoch": 0.26,
      "grad_norm": 74.0,
      "learning_rate": 3.755223423974661e-05,
      "loss": 0.8953,
      "step": 287900
    },
    {
      "epoch": 0.26,
      "grad_norm": 28.375,
      "learning_rate": 3.754773516655569e-05,
      "loss": 0.906,
      "step": 288000
    },
    {
      "epoch": 0.26,
      "grad_norm": 62.75,
      "learning_rate": 3.754323609336477e-05,
      "loss": 1.039,
      "step": 288100
    },
    {
      "epoch": 0.26,
      "grad_norm": 22.75,
      "learning_rate": 3.7538737020173846e-05,
      "loss": 1.0885,
      "step": 288200
    },
    {
      "epoch": 0.26,
      "grad_norm": 30.0,
      "learning_rate": 3.753423794698293e-05,
      "loss": 1.0548,
      "step": 288300
    },
    {
      "epoch": 0.26,
      "grad_norm": 27.375,
      "learning_rate": 3.7529738873792004e-05,
      "loss": 1.1686,
      "step": 288400
    },
    {
      "epoch": 0.26,
      "grad_norm": 21.125,
      "learning_rate": 3.752523980060107e-05,
      "loss": 1.013,
      "step": 288500
    },
    {
      "epoch": 0.26,
      "grad_norm": 47.25,
      "learning_rate": 3.7520740727410154e-05,
      "loss": 0.9353,
      "step": 288600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.751624165421923e-05,
      "loss": 0.9683,
      "step": 288700
    },
    {
      "epoch": 0.26,
      "grad_norm": 69.0,
      "learning_rate": 3.751174258102831e-05,
      "loss": 0.9861,
      "step": 288800
    },
    {
      "epoch": 0.26,
      "grad_norm": 268.0,
      "learning_rate": 3.750724350783739e-05,
      "loss": 0.9531,
      "step": 288900
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.9375,
      "learning_rate": 3.750274443464646e-05,
      "loss": 1.1006,
      "step": 289000
    },
    {
      "epoch": 0.26,
      "grad_norm": 21.5,
      "learning_rate": 3.7498245361455545e-05,
      "loss": 0.999,
      "step": 289100
    },
    {
      "epoch": 0.26,
      "grad_norm": 52.25,
      "learning_rate": 3.749374628826462e-05,
      "loss": 0.8388,
      "step": 289200
    },
    {
      "epoch": 0.26,
      "grad_norm": 45.0,
      "learning_rate": 3.7489247215073695e-05,
      "loss": 1.055,
      "step": 289300
    },
    {
      "epoch": 0.26,
      "grad_norm": 42.25,
      "learning_rate": 3.748474814188278e-05,
      "loss": 1.066,
      "step": 289400
    },
    {
      "epoch": 0.26,
      "grad_norm": 51.5,
      "learning_rate": 3.748024906869185e-05,
      "loss": 0.9666,
      "step": 289500
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.3125,
      "learning_rate": 3.747574999550093e-05,
      "loss": 1.1302,
      "step": 289600
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.25,
      "learning_rate": 3.747125092231001e-05,
      "loss": 1.0545,
      "step": 289700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.006591796875,
      "learning_rate": 3.746675184911908e-05,
      "loss": 0.965,
      "step": 289800
    },
    {
      "epoch": 0.26,
      "grad_norm": 44.0,
      "learning_rate": 3.746225277592816e-05,
      "loss": 0.9575,
      "step": 289900
    },
    {
      "epoch": 0.26,
      "grad_norm": 23.375,
      "learning_rate": 3.7457753702737236e-05,
      "loss": 1.1131,
      "step": 290000
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.5,
      "learning_rate": 3.745325462954631e-05,
      "loss": 0.9814,
      "step": 290100
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.201171875,
      "learning_rate": 3.7448755556355394e-05,
      "loss": 0.9384,
      "step": 290200
    },
    {
      "epoch": 0.26,
      "grad_norm": 25.5,
      "learning_rate": 3.744425648316447e-05,
      "loss": 0.9451,
      "step": 290300
    },
    {
      "epoch": 0.26,
      "grad_norm": 17.25,
      "learning_rate": 3.7439757409973544e-05,
      "loss": 1.0421,
      "step": 290400
    },
    {
      "epoch": 0.26,
      "grad_norm": 48.25,
      "learning_rate": 3.7435258336782626e-05,
      "loss": 0.9799,
      "step": 290500
    },
    {
      "epoch": 0.26,
      "grad_norm": 57.25,
      "learning_rate": 3.74307592635917e-05,
      "loss": 0.9903,
      "step": 290600
    },
    {
      "epoch": 0.26,
      "grad_norm": 22.375,
      "learning_rate": 3.742626019040078e-05,
      "loss": 0.9827,
      "step": 290700
    },
    {
      "epoch": 0.26,
      "grad_norm": 18.5,
      "learning_rate": 3.742176111720986e-05,
      "loss": 1.0665,
      "step": 290800
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.00347900390625,
      "learning_rate": 3.7417262044018935e-05,
      "loss": 0.9944,
      "step": 290900
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.75,
      "learning_rate": 3.741276297082802e-05,
      "loss": 0.9932,
      "step": 291000
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.375,
      "learning_rate": 3.7408263897637085e-05,
      "loss": 1.1128,
      "step": 291100
    },
    {
      "epoch": 0.26,
      "grad_norm": 41.5,
      "learning_rate": 3.740376482444616e-05,
      "loss": 1.1565,
      "step": 291200
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.75,
      "learning_rate": 3.739926575125524e-05,
      "loss": 0.9661,
      "step": 291300
    },
    {
      "epoch": 0.26,
      "grad_norm": 66.5,
      "learning_rate": 3.739476667806432e-05,
      "loss": 1.0081,
      "step": 291400
    },
    {
      "epoch": 0.26,
      "grad_norm": 69.5,
      "learning_rate": 3.73902676048734e-05,
      "loss": 1.0014,
      "step": 291500
    },
    {
      "epoch": 0.26,
      "grad_norm": 70.0,
      "learning_rate": 3.7385768531682475e-05,
      "loss": 1.0749,
      "step": 291600
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9140625,
      "learning_rate": 3.738126945849155e-05,
      "loss": 0.9294,
      "step": 291700
    },
    {
      "epoch": 0.26,
      "grad_norm": 24.625,
      "learning_rate": 3.737677038530063e-05,
      "loss": 1.1287,
      "step": 291800
    },
    {
      "epoch": 0.26,
      "grad_norm": 22.0,
      "learning_rate": 3.737227131210971e-05,
      "loss": 0.9587,
      "step": 291900
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.62890625,
      "learning_rate": 3.7367772238918784e-05,
      "loss": 0.9536,
      "step": 292000
    },
    {
      "epoch": 0.26,
      "grad_norm": 25.25,
      "learning_rate": 3.7363273165727866e-05,
      "loss": 0.9381,
      "step": 292100
    },
    {
      "epoch": 0.26,
      "grad_norm": 19.375,
      "learning_rate": 3.735877409253694e-05,
      "loss": 0.9211,
      "step": 292200
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.0,
      "learning_rate": 3.7354275019346016e-05,
      "loss": 1.0364,
      "step": 292300
    },
    {
      "epoch": 0.26,
      "grad_norm": 65.0,
      "learning_rate": 3.734977594615509e-05,
      "loss": 0.9292,
      "step": 292400
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.4375,
      "learning_rate": 3.734527687296417e-05,
      "loss": 1.1398,
      "step": 292500
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.5625,
      "learning_rate": 3.734077779977325e-05,
      "loss": 1.0559,
      "step": 292600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.0068359375,
      "learning_rate": 3.7336278726582325e-05,
      "loss": 1.0671,
      "step": 292700
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.3125,
      "learning_rate": 3.73317796533914e-05,
      "loss": 0.9718,
      "step": 292800
    },
    {
      "epoch": 0.26,
      "grad_norm": 76.0,
      "learning_rate": 3.732728058020048e-05,
      "loss": 1.1268,
      "step": 292900
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.8125,
      "learning_rate": 3.732278150700956e-05,
      "loss": 1.0402,
      "step": 293000
    },
    {
      "epoch": 0.26,
      "grad_norm": 184.0,
      "learning_rate": 3.731828243381863e-05,
      "loss": 1.0138,
      "step": 293100
    },
    {
      "epoch": 0.26,
      "grad_norm": 25.25,
      "learning_rate": 3.7313783360627715e-05,
      "loss": 1.0434,
      "step": 293200
    },
    {
      "epoch": 0.26,
      "grad_norm": 23.375,
      "learning_rate": 3.730928428743679e-05,
      "loss": 0.9349,
      "step": 293300
    },
    {
      "epoch": 0.26,
      "grad_norm": 42.75,
      "learning_rate": 3.7304785214245865e-05,
      "loss": 1.0041,
      "step": 293400
    },
    {
      "epoch": 0.26,
      "grad_norm": 848.0,
      "learning_rate": 3.730028614105495e-05,
      "loss": 0.9763,
      "step": 293500
    },
    {
      "epoch": 0.26,
      "grad_norm": 59.25,
      "learning_rate": 3.729578706786402e-05,
      "loss": 1.0238,
      "step": 293600
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.46875,
      "learning_rate": 3.72912879946731e-05,
      "loss": 1.011,
      "step": 293700
    },
    {
      "epoch": 0.26,
      "grad_norm": 47.75,
      "learning_rate": 3.7286788921482174e-05,
      "loss": 0.996,
      "step": 293800
    },
    {
      "epoch": 0.26,
      "grad_norm": 27.625,
      "learning_rate": 3.728228984829125e-05,
      "loss": 1.0576,
      "step": 293900
    },
    {
      "epoch": 0.26,
      "grad_norm": 86.0,
      "learning_rate": 3.727779077510033e-05,
      "loss": 0.9976,
      "step": 294000
    },
    {
      "epoch": 0.26,
      "grad_norm": 98.0,
      "learning_rate": 3.7273291701909406e-05,
      "loss": 0.9828,
      "step": 294100
    },
    {
      "epoch": 0.26,
      "grad_norm": 40.5,
      "learning_rate": 3.726879262871849e-05,
      "loss": 0.9299,
      "step": 294200
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.15625,
      "learning_rate": 3.7264293555527564e-05,
      "loss": 0.9068,
      "step": 294300
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.1875,
      "learning_rate": 3.725979448233664e-05,
      "loss": 1.1035,
      "step": 294400
    },
    {
      "epoch": 0.26,
      "grad_norm": 159.0,
      "learning_rate": 3.725529540914572e-05,
      "loss": 1.0225,
      "step": 294500
    },
    {
      "epoch": 0.26,
      "grad_norm": 296.0,
      "learning_rate": 3.72507963359548e-05,
      "loss": 1.0048,
      "step": 294600
    },
    {
      "epoch": 0.26,
      "grad_norm": 33.5,
      "learning_rate": 3.724629726276387e-05,
      "loss": 1.0528,
      "step": 294700
    },
    {
      "epoch": 0.26,
      "grad_norm": 74.0,
      "learning_rate": 3.7241798189572954e-05,
      "loss": 1.0722,
      "step": 294800
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.443359375,
      "learning_rate": 3.723729911638203e-05,
      "loss": 0.9493,
      "step": 294900
    },
    {
      "epoch": 0.26,
      "grad_norm": 33.25,
      "learning_rate": 3.7232800043191105e-05,
      "loss": 0.9789,
      "step": 295000
    },
    {
      "epoch": 0.26,
      "grad_norm": 114.5,
      "learning_rate": 3.722830097000018e-05,
      "loss": 0.944,
      "step": 295100
    },
    {
      "epoch": 0.26,
      "grad_norm": 104.0,
      "learning_rate": 3.7223801896809255e-05,
      "loss": 1.1147,
      "step": 295200
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2578125,
      "learning_rate": 3.721930282361834e-05,
      "loss": 1.0557,
      "step": 295300
    },
    {
      "epoch": 0.26,
      "grad_norm": 24.25,
      "learning_rate": 3.721480375042741e-05,
      "loss": 0.9894,
      "step": 295400
    },
    {
      "epoch": 0.26,
      "grad_norm": 340.0,
      "learning_rate": 3.721030467723649e-05,
      "loss": 0.9389,
      "step": 295500
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.0791015625,
      "learning_rate": 3.720580560404557e-05,
      "loss": 1.0486,
      "step": 295600
    },
    {
      "epoch": 0.26,
      "grad_norm": 26.5,
      "learning_rate": 3.7201306530854646e-05,
      "loss": 1.0906,
      "step": 295700
    },
    {
      "epoch": 0.26,
      "grad_norm": 34.0,
      "learning_rate": 3.719680745766372e-05,
      "loss": 1.0111,
      "step": 295800
    },
    {
      "epoch": 0.26,
      "grad_norm": 18.5,
      "learning_rate": 3.71923083844728e-05,
      "loss": 1.0953,
      "step": 295900
    },
    {
      "epoch": 0.26,
      "grad_norm": 19.625,
      "learning_rate": 3.718780931128188e-05,
      "loss": 0.9447,
      "step": 296000
    },
    {
      "epoch": 0.26,
      "grad_norm": 18.25,
      "learning_rate": 3.7183310238090954e-05,
      "loss": 0.9725,
      "step": 296100
    },
    {
      "epoch": 0.26,
      "grad_norm": 23.875,
      "learning_rate": 3.7178811164900036e-05,
      "loss": 0.9777,
      "step": 296200
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.4375,
      "learning_rate": 3.7174312091709105e-05,
      "loss": 0.8518,
      "step": 296300
    },
    {
      "epoch": 0.26,
      "grad_norm": 1472.0,
      "learning_rate": 3.716981301851819e-05,
      "loss": 1.0447,
      "step": 296400
    },
    {
      "epoch": 0.26,
      "grad_norm": 37.5,
      "learning_rate": 3.716531394532726e-05,
      "loss": 0.9507,
      "step": 296500
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.0,
      "learning_rate": 3.716081487213634e-05,
      "loss": 1.1126,
      "step": 296600
    },
    {
      "epoch": 0.26,
      "grad_norm": 21.0,
      "learning_rate": 3.715631579894542e-05,
      "loss": 1.0372,
      "step": 296700
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.6875,
      "learning_rate": 3.7151816725754495e-05,
      "loss": 0.9917,
      "step": 296800
    },
    {
      "epoch": 0.26,
      "grad_norm": 59.25,
      "learning_rate": 3.714731765256358e-05,
      "loss": 0.9659,
      "step": 296900
    },
    {
      "epoch": 0.26,
      "grad_norm": 125.5,
      "learning_rate": 3.714281857937265e-05,
      "loss": 1.0493,
      "step": 297000
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.498046875,
      "learning_rate": 3.713831950618173e-05,
      "loss": 1.0106,
      "step": 297100
    },
    {
      "epoch": 0.26,
      "grad_norm": 26.25,
      "learning_rate": 3.713382043299081e-05,
      "loss": 0.9621,
      "step": 297200
    },
    {
      "epoch": 0.26,
      "grad_norm": 51.0,
      "learning_rate": 3.7129321359799885e-05,
      "loss": 0.9449,
      "step": 297300
    },
    {
      "epoch": 0.26,
      "grad_norm": 124.0,
      "learning_rate": 3.712482228660896e-05,
      "loss": 1.129,
      "step": 297400
    },
    {
      "epoch": 0.27,
      "grad_norm": 46.5,
      "learning_rate": 3.712032321341804e-05,
      "loss": 0.9938,
      "step": 297500
    },
    {
      "epoch": 0.27,
      "grad_norm": 29.625,
      "learning_rate": 3.711582414022711e-05,
      "loss": 1.0287,
      "step": 297600
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.375,
      "learning_rate": 3.711132506703619e-05,
      "loss": 1.153,
      "step": 297700
    },
    {
      "epoch": 0.27,
      "grad_norm": 99.5,
      "learning_rate": 3.710682599384527e-05,
      "loss": 1.0182,
      "step": 297800
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.75,
      "learning_rate": 3.7102326920654344e-05,
      "loss": 0.9651,
      "step": 297900
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.1875,
      "learning_rate": 3.7097827847463426e-05,
      "loss": 0.8857,
      "step": 298000
    },
    {
      "epoch": 0.27,
      "grad_norm": 26.375,
      "learning_rate": 3.70933287742725e-05,
      "loss": 1.1251,
      "step": 298100
    },
    {
      "epoch": 0.27,
      "grad_norm": 21.625,
      "learning_rate": 3.708882970108158e-05,
      "loss": 0.9032,
      "step": 298200
    },
    {
      "epoch": 0.27,
      "grad_norm": 87.0,
      "learning_rate": 3.708433062789066e-05,
      "loss": 1.0085,
      "step": 298300
    },
    {
      "epoch": 0.27,
      "grad_norm": 37.5,
      "learning_rate": 3.7079831554699734e-05,
      "loss": 0.9449,
      "step": 298400
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.9375,
      "learning_rate": 3.707533248150881e-05,
      "loss": 0.9516,
      "step": 298500
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.875,
      "learning_rate": 3.707083340831789e-05,
      "loss": 1.0265,
      "step": 298600
    },
    {
      "epoch": 0.27,
      "grad_norm": 372.0,
      "learning_rate": 3.706633433512697e-05,
      "loss": 0.9236,
      "step": 298700
    },
    {
      "epoch": 0.27,
      "grad_norm": 63.0,
      "learning_rate": 3.706183526193604e-05,
      "loss": 0.9214,
      "step": 298800
    },
    {
      "epoch": 0.27,
      "grad_norm": 94.5,
      "learning_rate": 3.705733618874512e-05,
      "loss": 1.0358,
      "step": 298900
    },
    {
      "epoch": 0.27,
      "grad_norm": 62.75,
      "learning_rate": 3.705283711555419e-05,
      "loss": 0.9155,
      "step": 299000
    },
    {
      "epoch": 0.27,
      "grad_norm": 73.5,
      "learning_rate": 3.7048338042363275e-05,
      "loss": 1.0036,
      "step": 299100
    },
    {
      "epoch": 0.27,
      "grad_norm": 26.375,
      "learning_rate": 3.704383896917235e-05,
      "loss": 1.0941,
      "step": 299200
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.8125,
      "learning_rate": 3.7039339895981426e-05,
      "loss": 1.0197,
      "step": 299300
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.00860595703125,
      "learning_rate": 3.703484082279051e-05,
      "loss": 0.9353,
      "step": 299400
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.375,
      "learning_rate": 3.703034174959958e-05,
      "loss": 0.9616,
      "step": 299500
    },
    {
      "epoch": 0.27,
      "grad_norm": 22.0,
      "learning_rate": 3.702584267640866e-05,
      "loss": 1.1224,
      "step": 299600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.251953125,
      "learning_rate": 3.702134360321774e-05,
      "loss": 0.9859,
      "step": 299700
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.875,
      "learning_rate": 3.7016844530026816e-05,
      "loss": 1.0369,
      "step": 299800
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1796875,
      "learning_rate": 3.70123454568359e-05,
      "loss": 0.9392,
      "step": 299900
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.734375,
      "learning_rate": 3.7007846383644973e-05,
      "loss": 1.1231,
      "step": 300000
    },
    {
      "epoch": 0.27,
      "grad_norm": 73.0,
      "learning_rate": 3.700334731045405e-05,
      "loss": 1.1582,
      "step": 300100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.6998848237263124e-05,
      "loss": 0.9658,
      "step": 300200
    },
    {
      "epoch": 0.27,
      "grad_norm": 27.25,
      "learning_rate": 3.69943491640722e-05,
      "loss": 0.8286,
      "step": 300300
    },
    {
      "epoch": 0.27,
      "grad_norm": 96.0,
      "learning_rate": 3.698985009088128e-05,
      "loss": 0.8997,
      "step": 300400
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1650390625,
      "learning_rate": 3.698535101769036e-05,
      "loss": 1.0215,
      "step": 300500
    },
    {
      "epoch": 0.27,
      "grad_norm": 31.5,
      "learning_rate": 3.698085194449943e-05,
      "loss": 0.9601,
      "step": 300600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.09033203125,
      "learning_rate": 3.6976352871308514e-05,
      "loss": 1.0356,
      "step": 300700
    },
    {
      "epoch": 0.27,
      "grad_norm": 156.0,
      "learning_rate": 3.697185379811759e-05,
      "loss": 1.0337,
      "step": 300800
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.0,
      "learning_rate": 3.6967354724926665e-05,
      "loss": 1.0733,
      "step": 300900
    },
    {
      "epoch": 0.27,
      "grad_norm": 90.0,
      "learning_rate": 3.696285565173575e-05,
      "loss": 0.9862,
      "step": 301000
    },
    {
      "epoch": 0.27,
      "grad_norm": 33.75,
      "learning_rate": 3.695835657854482e-05,
      "loss": 1.0186,
      "step": 301100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.08349609375,
      "learning_rate": 3.69538575053539e-05,
      "loss": 1.0127,
      "step": 301200
    },
    {
      "epoch": 0.27,
      "grad_norm": 122.0,
      "learning_rate": 3.694935843216298e-05,
      "loss": 1.01,
      "step": 301300
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.5625,
      "learning_rate": 3.6944859358972055e-05,
      "loss": 0.96,
      "step": 301400
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.46875,
      "learning_rate": 3.694036028578113e-05,
      "loss": 1.0052,
      "step": 301500
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6796875,
      "learning_rate": 3.6935861212590206e-05,
      "loss": 1.0166,
      "step": 301600
    },
    {
      "epoch": 0.27,
      "grad_norm": 150.0,
      "learning_rate": 3.693136213939928e-05,
      "loss": 0.9057,
      "step": 301700
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.3125,
      "learning_rate": 3.6926863066208363e-05,
      "loss": 1.0006,
      "step": 301800
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.625,
      "learning_rate": 3.692236399301744e-05,
      "loss": 1.0435,
      "step": 301900
    },
    {
      "epoch": 0.27,
      "grad_norm": 64.5,
      "learning_rate": 3.6917864919826514e-05,
      "loss": 1.0164,
      "step": 302000
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.3125,
      "learning_rate": 3.6913365846635596e-05,
      "loss": 0.9632,
      "step": 302100
    },
    {
      "epoch": 0.27,
      "grad_norm": 28.125,
      "learning_rate": 3.690886677344467e-05,
      "loss": 1.0699,
      "step": 302200
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.0,
      "learning_rate": 3.690436770025375e-05,
      "loss": 1.0983,
      "step": 302300
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.625,
      "learning_rate": 3.689986862706283e-05,
      "loss": 0.9954,
      "step": 302400
    },
    {
      "epoch": 0.27,
      "grad_norm": 50.25,
      "learning_rate": 3.6895369553871904e-05,
      "loss": 1.1556,
      "step": 302500
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.037109375,
      "learning_rate": 3.6890870480680986e-05,
      "loss": 0.867,
      "step": 302600
    },
    {
      "epoch": 0.27,
      "grad_norm": 218.0,
      "learning_rate": 3.688637140749006e-05,
      "loss": 1.0612,
      "step": 302700
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.080078125,
      "learning_rate": 3.688187233429913e-05,
      "loss": 0.9617,
      "step": 302800
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.625,
      "learning_rate": 3.687737326110821e-05,
      "loss": 0.9486,
      "step": 302900
    },
    {
      "epoch": 0.27,
      "grad_norm": 46.25,
      "learning_rate": 3.687287418791729e-05,
      "loss": 0.9948,
      "step": 303000
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.5625,
      "learning_rate": 3.686837511472637e-05,
      "loss": 1.0681,
      "step": 303100
    },
    {
      "epoch": 0.27,
      "grad_norm": 21.375,
      "learning_rate": 3.6863876041535445e-05,
      "loss": 1.0904,
      "step": 303200
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.046142578125,
      "learning_rate": 3.685937696834452e-05,
      "loss": 1.1284,
      "step": 303300
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.5,
      "learning_rate": 3.68548778951536e-05,
      "loss": 1.0434,
      "step": 303400
    },
    {
      "epoch": 0.27,
      "grad_norm": 50.0,
      "learning_rate": 3.685037882196268e-05,
      "loss": 1.1216,
      "step": 303500
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.875,
      "learning_rate": 3.6845879748771753e-05,
      "loss": 1.0383,
      "step": 303600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19921875,
      "learning_rate": 3.6841380675580836e-05,
      "loss": 0.8577,
      "step": 303700
    },
    {
      "epoch": 0.27,
      "grad_norm": 284.0,
      "learning_rate": 3.683688160238991e-05,
      "loss": 1.0803,
      "step": 303800
    },
    {
      "epoch": 0.27,
      "grad_norm": 52.75,
      "learning_rate": 3.6832382529198986e-05,
      "loss": 0.8806,
      "step": 303900
    },
    {
      "epoch": 0.27,
      "grad_norm": 27.125,
      "learning_rate": 3.682788345600807e-05,
      "loss": 0.9574,
      "step": 304000
    },
    {
      "epoch": 0.27,
      "grad_norm": 482.0,
      "learning_rate": 3.682338438281714e-05,
      "loss": 1.1519,
      "step": 304100
    },
    {
      "epoch": 0.27,
      "grad_norm": 49.5,
      "learning_rate": 3.681888530962622e-05,
      "loss": 0.9249,
      "step": 304200
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.25,
      "learning_rate": 3.6814386236435294e-05,
      "loss": 1.0271,
      "step": 304300
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.0390625,
      "learning_rate": 3.680988716324437e-05,
      "loss": 1.0063,
      "step": 304400
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.1875,
      "learning_rate": 3.680538809005345e-05,
      "loss": 1.0237,
      "step": 304500
    },
    {
      "epoch": 0.27,
      "grad_norm": 35.5,
      "learning_rate": 3.680088901686253e-05,
      "loss": 1.0015,
      "step": 304600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.01397705078125,
      "learning_rate": 3.67963899436716e-05,
      "loss": 1.0098,
      "step": 304700
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.25,
      "learning_rate": 3.6791890870480685e-05,
      "loss": 1.1084,
      "step": 304800
    },
    {
      "epoch": 0.27,
      "grad_norm": 42.25,
      "learning_rate": 3.678739179728976e-05,
      "loss": 1.0012,
      "step": 304900
    },
    {
      "epoch": 0.27,
      "grad_norm": 37.25,
      "learning_rate": 3.6782892724098835e-05,
      "loss": 0.9744,
      "step": 305000
    },
    {
      "epoch": 0.27,
      "grad_norm": 34.5,
      "learning_rate": 3.677839365090792e-05,
      "loss": 1.079,
      "step": 305100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.08203125,
      "learning_rate": 3.677389457771699e-05,
      "loss": 1.0196,
      "step": 305200
    },
    {
      "epoch": 0.27,
      "grad_norm": 940.0,
      "learning_rate": 3.6769395504526075e-05,
      "loss": 1.1222,
      "step": 305300
    },
    {
      "epoch": 0.27,
      "grad_norm": 38.25,
      "learning_rate": 3.6764896431335143e-05,
      "loss": 0.8667,
      "step": 305400
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.875,
      "learning_rate": 3.676039735814422e-05,
      "loss": 0.9969,
      "step": 305500
    },
    {
      "epoch": 0.27,
      "grad_norm": 59.0,
      "learning_rate": 3.67558982849533e-05,
      "loss": 0.9233,
      "step": 305600
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2578125,
      "learning_rate": 3.6751399211762376e-05,
      "loss": 1.0392,
      "step": 305700
    },
    {
      "epoch": 0.27,
      "grad_norm": 48.75,
      "learning_rate": 3.674690013857146e-05,
      "loss": 1.0305,
      "step": 305800
    },
    {
      "epoch": 0.27,
      "grad_norm": 21.625,
      "learning_rate": 3.6742401065380534e-05,
      "loss": 0.9824,
      "step": 305900
    },
    {
      "epoch": 0.27,
      "grad_norm": 26.75,
      "learning_rate": 3.673790199218961e-05,
      "loss": 1.1893,
      "step": 306000
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.0,
      "learning_rate": 3.673340291899869e-05,
      "loss": 0.8237,
      "step": 306100
    },
    {
      "epoch": 0.27,
      "grad_norm": 43.75,
      "learning_rate": 3.6728903845807766e-05,
      "loss": 0.9386,
      "step": 306200
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.5,
      "learning_rate": 3.672440477261684e-05,
      "loss": 0.9905,
      "step": 306300
    },
    {
      "epoch": 0.27,
      "grad_norm": 30.5,
      "learning_rate": 3.6719905699425924e-05,
      "loss": 0.9236,
      "step": 306400
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.0120849609375,
      "learning_rate": 3.6715406626235e-05,
      "loss": 0.9911,
      "step": 306500
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.8125,
      "learning_rate": 3.6710907553044075e-05,
      "loss": 1.0119,
      "step": 306600
    },
    {
      "epoch": 0.27,
      "grad_norm": 33.75,
      "learning_rate": 3.670640847985315e-05,
      "loss": 1.0005,
      "step": 306700
    },
    {
      "epoch": 0.27,
      "grad_norm": 173.0,
      "learning_rate": 3.6701909406662225e-05,
      "loss": 1.0375,
      "step": 306800
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.9375,
      "learning_rate": 3.669741033347131e-05,
      "loss": 0.9426,
      "step": 306900
    },
    {
      "epoch": 0.27,
      "grad_norm": 23.875,
      "learning_rate": 3.669291126028038e-05,
      "loss": 1.0287,
      "step": 307000
    },
    {
      "epoch": 0.27,
      "grad_norm": 21.75,
      "learning_rate": 3.668841218708946e-05,
      "loss": 1.0525,
      "step": 307100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.0849609375,
      "learning_rate": 3.668391311389854e-05,
      "loss": 0.9764,
      "step": 307200
    },
    {
      "epoch": 0.27,
      "grad_norm": 108.0,
      "learning_rate": 3.6679414040707616e-05,
      "loss": 1.0604,
      "step": 307300
    },
    {
      "epoch": 0.27,
      "grad_norm": 140.0,
      "learning_rate": 3.667491496751669e-05,
      "loss": 1.0525,
      "step": 307400
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.75,
      "learning_rate": 3.667041589432577e-05,
      "loss": 1.0677,
      "step": 307500
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48046875,
      "learning_rate": 3.666591682113485e-05,
      "loss": 0.9733,
      "step": 307600
    },
    {
      "epoch": 0.27,
      "grad_norm": 31.875,
      "learning_rate": 3.6661417747943924e-05,
      "loss": 0.9332,
      "step": 307700
    },
    {
      "epoch": 0.27,
      "grad_norm": 26.625,
      "learning_rate": 3.6656918674753006e-05,
      "loss": 1.0315,
      "step": 307800
    },
    {
      "epoch": 0.27,
      "grad_norm": 31.25,
      "learning_rate": 3.665241960156208e-05,
      "loss": 0.9553,
      "step": 307900
    },
    {
      "epoch": 0.27,
      "grad_norm": 506.0,
      "learning_rate": 3.6647920528371156e-05,
      "loss": 1.0836,
      "step": 308000
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.0,
      "learning_rate": 3.664342145518023e-05,
      "loss": 0.9197,
      "step": 308100
    },
    {
      "epoch": 0.27,
      "grad_norm": 25.875,
      "learning_rate": 3.663892238198931e-05,
      "loss": 1.0806,
      "step": 308200
    },
    {
      "epoch": 0.27,
      "grad_norm": 33.0,
      "learning_rate": 3.663442330879839e-05,
      "loss": 1.0173,
      "step": 308300
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.080078125,
      "learning_rate": 3.6629924235607465e-05,
      "loss": 0.9192,
      "step": 308400
    },
    {
      "epoch": 0.27,
      "grad_norm": 110.5,
      "learning_rate": 3.662542516241655e-05,
      "loss": 0.9926,
      "step": 308500
    },
    {
      "epoch": 0.27,
      "grad_norm": 50.75,
      "learning_rate": 3.662092608922562e-05,
      "loss": 1.0538,
      "step": 308600
    },
    {
      "epoch": 0.27,
      "grad_norm": 25.75,
      "learning_rate": 3.66164270160347e-05,
      "loss": 1.0892,
      "step": 308700
    },
    {
      "epoch": 0.28,
      "grad_norm": 116.5,
      "learning_rate": 3.661192794284378e-05,
      "loss": 1.0033,
      "step": 308800
    },
    {
      "epoch": 0.28,
      "grad_norm": 46.75,
      "learning_rate": 3.6607428869652855e-05,
      "loss": 1.0094,
      "step": 308900
    },
    {
      "epoch": 0.28,
      "grad_norm": 87.0,
      "learning_rate": 3.660292979646193e-05,
      "loss": 0.9548,
      "step": 309000
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1630859375,
      "learning_rate": 3.659843072327101e-05,
      "loss": 1.1314,
      "step": 309100
    },
    {
      "epoch": 0.28,
      "grad_norm": 43.25,
      "learning_rate": 3.659393165008009e-05,
      "loss": 0.9092,
      "step": 309200
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.375,
      "learning_rate": 3.658943257688916e-05,
      "loss": 0.9552,
      "step": 309300
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.25,
      "learning_rate": 3.658493350369824e-05,
      "loss": 1.1365,
      "step": 309400
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.625,
      "learning_rate": 3.6580434430507314e-05,
      "loss": 0.9769,
      "step": 309500
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.515625,
      "learning_rate": 3.6575935357316396e-05,
      "loss": 1.0819,
      "step": 309600
    },
    {
      "epoch": 0.28,
      "grad_norm": 26.625,
      "learning_rate": 3.657143628412547e-05,
      "loss": 0.9747,
      "step": 309700
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.25,
      "learning_rate": 3.6566937210934546e-05,
      "loss": 1.068,
      "step": 309800
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.28125,
      "learning_rate": 3.656243813774363e-05,
      "loss": 0.9789,
      "step": 309900
    },
    {
      "epoch": 0.28,
      "grad_norm": 87.0,
      "learning_rate": 3.6557939064552704e-05,
      "loss": 1.0288,
      "step": 310000
    },
    {
      "epoch": 0.28,
      "grad_norm": 65.5,
      "learning_rate": 3.655343999136178e-05,
      "loss": 0.9174,
      "step": 310100
    },
    {
      "epoch": 0.28,
      "grad_norm": 41.0,
      "learning_rate": 3.654894091817086e-05,
      "loss": 1.0209,
      "step": 310200
    },
    {
      "epoch": 0.28,
      "grad_norm": 31.875,
      "learning_rate": 3.654444184497994e-05,
      "loss": 0.975,
      "step": 310300
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.0174560546875,
      "learning_rate": 3.653994277178901e-05,
      "loss": 0.9641,
      "step": 310400
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.0,
      "learning_rate": 3.6535443698598094e-05,
      "loss": 1.0333,
      "step": 310500
    },
    {
      "epoch": 0.28,
      "grad_norm": 32.25,
      "learning_rate": 3.653094462540716e-05,
      "loss": 0.9952,
      "step": 310600
    },
    {
      "epoch": 0.28,
      "grad_norm": 26.875,
      "learning_rate": 3.6526445552216245e-05,
      "loss": 1.0131,
      "step": 310700
    },
    {
      "epoch": 0.28,
      "grad_norm": 71.5,
      "learning_rate": 3.652194647902532e-05,
      "loss": 1.0748,
      "step": 310800
    },
    {
      "epoch": 0.28,
      "grad_norm": 34.75,
      "learning_rate": 3.6517447405834396e-05,
      "loss": 0.9656,
      "step": 310900
    },
    {
      "epoch": 0.28,
      "grad_norm": 49.0,
      "learning_rate": 3.651294833264348e-05,
      "loss": 1.0691,
      "step": 311000
    },
    {
      "epoch": 0.28,
      "grad_norm": 23.25,
      "learning_rate": 3.650844925945255e-05,
      "loss": 1.0165,
      "step": 311100
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.625,
      "learning_rate": 3.6503950186261635e-05,
      "loss": 1.1299,
      "step": 311200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.0126953125,
      "learning_rate": 3.649945111307071e-05,
      "loss": 0.9627,
      "step": 311300
    },
    {
      "epoch": 0.28,
      "grad_norm": 107.0,
      "learning_rate": 3.6494952039879786e-05,
      "loss": 1.0488,
      "step": 311400
    },
    {
      "epoch": 0.28,
      "grad_norm": 77.0,
      "learning_rate": 3.649045296668887e-05,
      "loss": 1.0094,
      "step": 311500
    },
    {
      "epoch": 0.28,
      "grad_norm": 80.5,
      "learning_rate": 3.648595389349794e-05,
      "loss": 1.0298,
      "step": 311600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44921875,
      "learning_rate": 3.648145482030702e-05,
      "loss": 0.9935,
      "step": 311700
    },
    {
      "epoch": 0.28,
      "grad_norm": 39.5,
      "learning_rate": 3.64769557471161e-05,
      "loss": 1.0381,
      "step": 311800
    },
    {
      "epoch": 0.28,
      "grad_norm": 29.625,
      "learning_rate": 3.647245667392517e-05,
      "loss": 0.9743,
      "step": 311900
    },
    {
      "epoch": 0.28,
      "grad_norm": 34.0,
      "learning_rate": 3.646795760073425e-05,
      "loss": 1.0455,
      "step": 312000
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.875,
      "learning_rate": 3.646345852754333e-05,
      "loss": 1.0269,
      "step": 312100
    },
    {
      "epoch": 0.28,
      "grad_norm": 58.25,
      "learning_rate": 3.64589594543524e-05,
      "loss": 1.0804,
      "step": 312200
    },
    {
      "epoch": 0.28,
      "grad_norm": 90.0,
      "learning_rate": 3.6454460381161484e-05,
      "loss": 0.8518,
      "step": 312300
    },
    {
      "epoch": 0.28,
      "grad_norm": 36.0,
      "learning_rate": 3.644996130797056e-05,
      "loss": 1.1012,
      "step": 312400
    },
    {
      "epoch": 0.28,
      "grad_norm": 31.5,
      "learning_rate": 3.6445462234779635e-05,
      "loss": 1.0819,
      "step": 312500
    },
    {
      "epoch": 0.28,
      "grad_norm": 87.0,
      "learning_rate": 3.644096316158872e-05,
      "loss": 1.0311,
      "step": 312600
    },
    {
      "epoch": 0.28,
      "grad_norm": 110.5,
      "learning_rate": 3.643646408839779e-05,
      "loss": 0.984,
      "step": 312700
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.021240234375,
      "learning_rate": 3.643196501520687e-05,
      "loss": 0.9403,
      "step": 312800
    },
    {
      "epoch": 0.28,
      "grad_norm": 109.5,
      "learning_rate": 3.642746594201595e-05,
      "loss": 0.8858,
      "step": 312900
    },
    {
      "epoch": 0.28,
      "grad_norm": 44.25,
      "learning_rate": 3.6422966868825025e-05,
      "loss": 1.057,
      "step": 313000
    },
    {
      "epoch": 0.28,
      "grad_norm": 45.0,
      "learning_rate": 3.64184677956341e-05,
      "loss": 0.9359,
      "step": 313100
    },
    {
      "epoch": 0.28,
      "grad_norm": 32.0,
      "learning_rate": 3.6413968722443176e-05,
      "loss": 1.0052,
      "step": 313200
    },
    {
      "epoch": 0.28,
      "grad_norm": 27.25,
      "learning_rate": 3.640946964925225e-05,
      "loss": 1.1168,
      "step": 313300
    },
    {
      "epoch": 0.28,
      "grad_norm": 103.5,
      "learning_rate": 3.640497057606133e-05,
      "loss": 1.0036,
      "step": 313400
    },
    {
      "epoch": 0.28,
      "grad_norm": 27.625,
      "learning_rate": 3.640047150287041e-05,
      "loss": 1.0273,
      "step": 313500
    },
    {
      "epoch": 0.28,
      "grad_norm": 53.5,
      "learning_rate": 3.6395972429679484e-05,
      "loss": 0.9393,
      "step": 313600
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.0625,
      "learning_rate": 3.6391473356488566e-05,
      "loss": 0.9945,
      "step": 313700
    },
    {
      "epoch": 0.28,
      "grad_norm": 41.0,
      "learning_rate": 3.638697428329764e-05,
      "loss": 0.9459,
      "step": 313800
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.029296875,
      "learning_rate": 3.6382475210106723e-05,
      "loss": 0.8752,
      "step": 313900
    },
    {
      "epoch": 0.28,
      "grad_norm": 42.75,
      "learning_rate": 3.63779761369158e-05,
      "loss": 1.0804,
      "step": 314000
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1484375,
      "learning_rate": 3.6373477063724874e-05,
      "loss": 1.2319,
      "step": 314100
    },
    {
      "epoch": 0.28,
      "grad_norm": 23.25,
      "learning_rate": 3.6368977990533956e-05,
      "loss": 0.9201,
      "step": 314200
    },
    {
      "epoch": 0.28,
      "grad_norm": 48.5,
      "learning_rate": 3.636447891734303e-05,
      "loss": 0.9143,
      "step": 314300
    },
    {
      "epoch": 0.28,
      "grad_norm": 37.75,
      "learning_rate": 3.635997984415211e-05,
      "loss": 0.8795,
      "step": 314400
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.875,
      "learning_rate": 3.635548077096118e-05,
      "loss": 0.8999,
      "step": 314500
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.5625,
      "learning_rate": 3.635098169777026e-05,
      "loss": 1.011,
      "step": 314600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36328125,
      "learning_rate": 3.634648262457934e-05,
      "loss": 1.0106,
      "step": 314700
    },
    {
      "epoch": 0.28,
      "grad_norm": 23.0,
      "learning_rate": 3.6341983551388415e-05,
      "loss": 0.8728,
      "step": 314800
    },
    {
      "epoch": 0.28,
      "grad_norm": 90.0,
      "learning_rate": 3.633748447819749e-05,
      "loss": 1.1142,
      "step": 314900
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.04150390625,
      "learning_rate": 3.633298540500657e-05,
      "loss": 0.9321,
      "step": 315000
    },
    {
      "epoch": 0.28,
      "grad_norm": 46.0,
      "learning_rate": 3.632848633181565e-05,
      "loss": 0.9976,
      "step": 315100
    },
    {
      "epoch": 0.28,
      "grad_norm": 22.625,
      "learning_rate": 3.632398725862472e-05,
      "loss": 0.9879,
      "step": 315200
    },
    {
      "epoch": 0.28,
      "grad_norm": 24.0,
      "learning_rate": 3.6319488185433805e-05,
      "loss": 0.9861,
      "step": 315300
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.1875,
      "learning_rate": 3.631498911224288e-05,
      "loss": 0.9907,
      "step": 315400
    },
    {
      "epoch": 0.28,
      "grad_norm": 79.0,
      "learning_rate": 3.6310490039051956e-05,
      "loss": 1.0425,
      "step": 315500
    },
    {
      "epoch": 0.28,
      "grad_norm": 34.5,
      "learning_rate": 3.630599096586104e-05,
      "loss": 1.135,
      "step": 315600
    },
    {
      "epoch": 0.28,
      "grad_norm": 211.0,
      "learning_rate": 3.6301491892670113e-05,
      "loss": 1.045,
      "step": 315700
    },
    {
      "epoch": 0.28,
      "grad_norm": 87.0,
      "learning_rate": 3.629699281947919e-05,
      "loss": 0.9732,
      "step": 315800
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.125,
      "learning_rate": 3.6292493746288264e-05,
      "loss": 1.1904,
      "step": 315900
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.54296875,
      "learning_rate": 3.628799467309734e-05,
      "loss": 0.9783,
      "step": 316000
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6015625,
      "learning_rate": 3.628349559990642e-05,
      "loss": 0.9245,
      "step": 316100
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.0,
      "learning_rate": 3.62789965267155e-05,
      "loss": 1.0189,
      "step": 316200
    },
    {
      "epoch": 0.28,
      "grad_norm": 44.25,
      "learning_rate": 3.627449745352457e-05,
      "loss": 0.975,
      "step": 316300
    },
    {
      "epoch": 0.28,
      "grad_norm": 67.0,
      "learning_rate": 3.6269998380333654e-05,
      "loss": 0.871,
      "step": 316400
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.25,
      "learning_rate": 3.626549930714273e-05,
      "loss": 0.9881,
      "step": 316500
    },
    {
      "epoch": 0.28,
      "grad_norm": 150.0,
      "learning_rate": 3.6261000233951805e-05,
      "loss": 1.1318,
      "step": 316600
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.75,
      "learning_rate": 3.625650116076089e-05,
      "loss": 1.1347,
      "step": 316700
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.4375,
      "learning_rate": 3.625200208756996e-05,
      "loss": 0.9635,
      "step": 316800
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.875,
      "learning_rate": 3.6247503014379045e-05,
      "loss": 1.0946,
      "step": 316900
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.408203125,
      "learning_rate": 3.624300394118812e-05,
      "loss": 0.9249,
      "step": 317000
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.5625,
      "learning_rate": 3.623850486799719e-05,
      "loss": 0.8626,
      "step": 317100
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.0,
      "learning_rate": 3.623400579480627e-05,
      "loss": 0.9188,
      "step": 317200
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.4375,
      "learning_rate": 3.6229506721615346e-05,
      "loss": 1.048,
      "step": 317300
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.625,
      "learning_rate": 3.622500764842443e-05,
      "loss": 1.1118,
      "step": 317400
    },
    {
      "epoch": 0.28,
      "grad_norm": 72.0,
      "learning_rate": 3.6220508575233503e-05,
      "loss": 0.9467,
      "step": 317500
    },
    {
      "epoch": 0.28,
      "grad_norm": 46.75,
      "learning_rate": 3.621600950204258e-05,
      "loss": 1.1107,
      "step": 317600
    },
    {
      "epoch": 0.28,
      "grad_norm": 23.75,
      "learning_rate": 3.621151042885166e-05,
      "loss": 1.0043,
      "step": 317700
    },
    {
      "epoch": 0.28,
      "grad_norm": 28.375,
      "learning_rate": 3.6207011355660736e-05,
      "loss": 0.9025,
      "step": 317800
    },
    {
      "epoch": 0.28,
      "grad_norm": 36.5,
      "learning_rate": 3.620251228246981e-05,
      "loss": 0.8554,
      "step": 317900
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.375,
      "learning_rate": 3.6198013209278894e-05,
      "loss": 1.026,
      "step": 318000
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.3125,
      "learning_rate": 3.619351413608797e-05,
      "loss": 1.1124,
      "step": 318100
    },
    {
      "epoch": 0.28,
      "grad_norm": 22.75,
      "learning_rate": 3.6189015062897044e-05,
      "loss": 1.1842,
      "step": 318200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.004486083984375,
      "learning_rate": 3.6184515989706127e-05,
      "loss": 0.9869,
      "step": 318300
    },
    {
      "epoch": 0.28,
      "grad_norm": 183.0,
      "learning_rate": 3.6180016916515195e-05,
      "loss": 1.2416,
      "step": 318400
    },
    {
      "epoch": 0.28,
      "grad_norm": 49.25,
      "learning_rate": 3.617551784332428e-05,
      "loss": 1.0201,
      "step": 318500
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.625,
      "learning_rate": 3.617101877013335e-05,
      "loss": 1.0957,
      "step": 318600
    },
    {
      "epoch": 0.28,
      "grad_norm": 54.5,
      "learning_rate": 3.616651969694243e-05,
      "loss": 0.9934,
      "step": 318700
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.375,
      "learning_rate": 3.616202062375151e-05,
      "loss": 0.9565,
      "step": 318800
    },
    {
      "epoch": 0.28,
      "grad_norm": 62.5,
      "learning_rate": 3.6157521550560585e-05,
      "loss": 1.0589,
      "step": 318900
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.125,
      "learning_rate": 3.615302247736966e-05,
      "loss": 0.9754,
      "step": 319000
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.375,
      "learning_rate": 3.614852340417874e-05,
      "loss": 0.9177,
      "step": 319100
    },
    {
      "epoch": 0.28,
      "grad_norm": 117.0,
      "learning_rate": 3.614402433098782e-05,
      "loss": 0.9928,
      "step": 319200
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.375,
      "learning_rate": 3.6139525257796893e-05,
      "loss": 0.9426,
      "step": 319300
    },
    {
      "epoch": 0.28,
      "grad_norm": 136.0,
      "learning_rate": 3.6135026184605976e-05,
      "loss": 0.9318,
      "step": 319400
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.375,
      "learning_rate": 3.613052711141505e-05,
      "loss": 0.9935,
      "step": 319500
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.875,
      "learning_rate": 3.612602803822413e-05,
      "loss": 1.0611,
      "step": 319600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.036376953125,
      "learning_rate": 3.61215289650332e-05,
      "loss": 0.9981,
      "step": 319700
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.611702989184228e-05,
      "loss": 1.1049,
      "step": 319800
    },
    {
      "epoch": 0.28,
      "grad_norm": 63.25,
      "learning_rate": 3.611253081865136e-05,
      "loss": 1.0493,
      "step": 319900
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.625,
      "learning_rate": 3.6108031745460434e-05,
      "loss": 0.9991,
      "step": 320000
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.25,
      "learning_rate": 3.6103532672269517e-05,
      "loss": 0.9046,
      "step": 320100
    },
    {
      "epoch": 0.29,
      "grad_norm": 51.25,
      "learning_rate": 3.609903359907859e-05,
      "loss": 0.9621,
      "step": 320200
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.12890625,
      "learning_rate": 3.609453452588767e-05,
      "loss": 1.0321,
      "step": 320300
    },
    {
      "epoch": 0.29,
      "grad_norm": 21.75,
      "learning_rate": 3.609003545269675e-05,
      "loss": 0.9761,
      "step": 320400
    },
    {
      "epoch": 0.29,
      "grad_norm": 1600.0,
      "learning_rate": 3.6085536379505825e-05,
      "loss": 1.0054,
      "step": 320500
    },
    {
      "epoch": 0.29,
      "grad_norm": 20.875,
      "learning_rate": 3.60810373063149e-05,
      "loss": 0.895,
      "step": 320600
    },
    {
      "epoch": 0.29,
      "grad_norm": 33.75,
      "learning_rate": 3.607653823312398e-05,
      "loss": 1.0764,
      "step": 320700
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.5625,
      "learning_rate": 3.607203915993306e-05,
      "loss": 0.9927,
      "step": 320800
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.5,
      "learning_rate": 3.606754008674213e-05,
      "loss": 0.9858,
      "step": 320900
    },
    {
      "epoch": 0.29,
      "grad_norm": 20.125,
      "learning_rate": 3.606304101355121e-05,
      "loss": 1.142,
      "step": 321000
    },
    {
      "epoch": 0.29,
      "grad_norm": 92.0,
      "learning_rate": 3.6058541940360283e-05,
      "loss": 1.0967,
      "step": 321100
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.5625,
      "learning_rate": 3.6054042867169366e-05,
      "loss": 0.9635,
      "step": 321200
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.0,
      "learning_rate": 3.604954379397844e-05,
      "loss": 0.9771,
      "step": 321300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.03076171875,
      "learning_rate": 3.6045044720787516e-05,
      "loss": 0.9096,
      "step": 321400
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.5,
      "learning_rate": 3.60405456475966e-05,
      "loss": 0.9777,
      "step": 321500
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.005950927734375,
      "learning_rate": 3.6036046574405674e-05,
      "loss": 1.0101,
      "step": 321600
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.359375,
      "learning_rate": 3.603154750121475e-05,
      "loss": 1.0459,
      "step": 321700
    },
    {
      "epoch": 0.29,
      "grad_norm": 38.0,
      "learning_rate": 3.602704842802383e-05,
      "loss": 1.1353,
      "step": 321800
    },
    {
      "epoch": 0.29,
      "grad_norm": 94.0,
      "learning_rate": 3.6022549354832907e-05,
      "loss": 0.9828,
      "step": 321900
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.03857421875,
      "learning_rate": 3.601805028164198e-05,
      "loss": 0.9718,
      "step": 322000
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.375,
      "learning_rate": 3.6013551208451064e-05,
      "loss": 0.9502,
      "step": 322100
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.28125,
      "learning_rate": 3.600905213526014e-05,
      "loss": 0.923,
      "step": 322200
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.375,
      "learning_rate": 3.6004553062069215e-05,
      "loss": 0.7945,
      "step": 322300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.600005398887829e-05,
      "loss": 1.0133,
      "step": 322400
    },
    {
      "epoch": 0.29,
      "grad_norm": 27.375,
      "learning_rate": 3.5995554915687365e-05,
      "loss": 1.022,
      "step": 322500
    },
    {
      "epoch": 0.29,
      "grad_norm": 21.25,
      "learning_rate": 3.599105584249645e-05,
      "loss": 1.1105,
      "step": 322600
    },
    {
      "epoch": 0.29,
      "grad_norm": 98.0,
      "learning_rate": 3.598655676930552e-05,
      "loss": 1.184,
      "step": 322700
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.25,
      "learning_rate": 3.5982057696114605e-05,
      "loss": 1.0759,
      "step": 322800
    },
    {
      "epoch": 0.29,
      "grad_norm": 31.0,
      "learning_rate": 3.597755862292368e-05,
      "loss": 0.966,
      "step": 322900
    },
    {
      "epoch": 0.29,
      "grad_norm": 33.25,
      "learning_rate": 3.5973059549732756e-05,
      "loss": 1.1839,
      "step": 323000
    },
    {
      "epoch": 0.29,
      "grad_norm": 37.75,
      "learning_rate": 3.596856047654184e-05,
      "loss": 1.0618,
      "step": 323100
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.125,
      "learning_rate": 3.596406140335091e-05,
      "loss": 0.9178,
      "step": 323200
    },
    {
      "epoch": 0.29,
      "grad_norm": 88.5,
      "learning_rate": 3.595956233015999e-05,
      "loss": 1.0137,
      "step": 323300
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.25,
      "learning_rate": 3.595506325696907e-05,
      "loss": 0.9731,
      "step": 323400
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.5,
      "learning_rate": 3.5950564183778146e-05,
      "loss": 0.8936,
      "step": 323500
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.01519775390625,
      "learning_rate": 3.594606511058722e-05,
      "loss": 1.0664,
      "step": 323600
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.75,
      "learning_rate": 3.5941566037396296e-05,
      "loss": 0.9372,
      "step": 323700
    },
    {
      "epoch": 0.29,
      "grad_norm": 41.25,
      "learning_rate": 3.593706696420537e-05,
      "loss": 1.1182,
      "step": 323800
    },
    {
      "epoch": 0.29,
      "grad_norm": 25.5,
      "learning_rate": 3.5932567891014454e-05,
      "loss": 1.0484,
      "step": 323900
    },
    {
      "epoch": 0.29,
      "grad_norm": 780.0,
      "learning_rate": 3.592806881782353e-05,
      "loss": 0.9203,
      "step": 324000
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.5,
      "learning_rate": 3.5923569744632605e-05,
      "loss": 0.8622,
      "step": 324100
    },
    {
      "epoch": 0.29,
      "grad_norm": 22.25,
      "learning_rate": 3.591907067144169e-05,
      "loss": 1.1688,
      "step": 324200
    },
    {
      "epoch": 0.29,
      "grad_norm": 14.125,
      "learning_rate": 3.591457159825076e-05,
      "loss": 0.9461,
      "step": 324300
    },
    {
      "epoch": 0.29,
      "grad_norm": 25.25,
      "learning_rate": 3.591007252505984e-05,
      "loss": 0.9347,
      "step": 324400
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.25,
      "learning_rate": 3.590557345186892e-05,
      "loss": 1.1219,
      "step": 324500
    },
    {
      "epoch": 0.29,
      "grad_norm": 34.75,
      "learning_rate": 3.5901074378677995e-05,
      "loss": 1.0158,
      "step": 324600
    },
    {
      "epoch": 0.29,
      "grad_norm": 34.75,
      "learning_rate": 3.589657530548707e-05,
      "loss": 1.1649,
      "step": 324700
    },
    {
      "epoch": 0.29,
      "grad_norm": 63.0,
      "learning_rate": 3.589207623229615e-05,
      "loss": 0.9947,
      "step": 324800
    },
    {
      "epoch": 0.29,
      "grad_norm": 57.5,
      "learning_rate": 3.588757715910522e-05,
      "loss": 0.9695,
      "step": 324900
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.375,
      "learning_rate": 3.58830780859143e-05,
      "loss": 1.104,
      "step": 325000
    },
    {
      "epoch": 0.29,
      "grad_norm": 33.75,
      "learning_rate": 3.587857901272338e-05,
      "loss": 1.0179,
      "step": 325100
    },
    {
      "epoch": 0.29,
      "grad_norm": 29.0,
      "learning_rate": 3.5874079939532454e-05,
      "loss": 1.0921,
      "step": 325200
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.125,
      "learning_rate": 3.5869580866341536e-05,
      "loss": 0.9883,
      "step": 325300
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.625,
      "learning_rate": 3.586508179315061e-05,
      "loss": 0.9916,
      "step": 325400
    },
    {
      "epoch": 0.29,
      "grad_norm": 32.25,
      "learning_rate": 3.586058271995969e-05,
      "loss": 1.0047,
      "step": 325500
    },
    {
      "epoch": 0.29,
      "grad_norm": 48.0,
      "learning_rate": 3.585608364676877e-05,
      "loss": 1.0746,
      "step": 325600
    },
    {
      "epoch": 0.29,
      "grad_norm": 24.0,
      "learning_rate": 3.5851584573577844e-05,
      "loss": 1.2072,
      "step": 325700
    },
    {
      "epoch": 0.29,
      "grad_norm": 54.0,
      "learning_rate": 3.5847085500386926e-05,
      "loss": 1.2004,
      "step": 325800
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.0,
      "learning_rate": 3.5842586427196e-05,
      "loss": 0.9823,
      "step": 325900
    },
    {
      "epoch": 0.29,
      "grad_norm": 48.0,
      "learning_rate": 3.583808735400508e-05,
      "loss": 1.0313,
      "step": 326000
    },
    {
      "epoch": 0.29,
      "grad_norm": 192.0,
      "learning_rate": 3.583358828081416e-05,
      "loss": 1.0086,
      "step": 326100
    },
    {
      "epoch": 0.29,
      "grad_norm": 36.75,
      "learning_rate": 3.582908920762323e-05,
      "loss": 0.9467,
      "step": 326200
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.125,
      "learning_rate": 3.582459013443231e-05,
      "loss": 1.0403,
      "step": 326300
    },
    {
      "epoch": 0.29,
      "grad_norm": 117.0,
      "learning_rate": 3.5820091061241385e-05,
      "loss": 0.9163,
      "step": 326400
    },
    {
      "epoch": 0.29,
      "grad_norm": 33.75,
      "learning_rate": 3.581559198805046e-05,
      "loss": 1.1902,
      "step": 326500
    },
    {
      "epoch": 0.29,
      "grad_norm": 28.375,
      "learning_rate": 3.581109291485954e-05,
      "loss": 1.0054,
      "step": 326600
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5859375,
      "learning_rate": 3.580659384166862e-05,
      "loss": 0.9536,
      "step": 326700
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.042236328125,
      "learning_rate": 3.580209476847769e-05,
      "loss": 1.115,
      "step": 326800
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.375,
      "learning_rate": 3.5797595695286775e-05,
      "loss": 0.9789,
      "step": 326900
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.5,
      "learning_rate": 3.579309662209585e-05,
      "loss": 1.0123,
      "step": 327000
    },
    {
      "epoch": 0.29,
      "grad_norm": 38.5,
      "learning_rate": 3.5788597548904926e-05,
      "loss": 0.9962,
      "step": 327100
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.5,
      "learning_rate": 3.578409847571401e-05,
      "loss": 1.0599,
      "step": 327200
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.75,
      "learning_rate": 3.577959940252308e-05,
      "loss": 0.9727,
      "step": 327300
    },
    {
      "epoch": 0.29,
      "grad_norm": 21.5,
      "learning_rate": 3.577510032933216e-05,
      "loss": 0.9051,
      "step": 327400
    },
    {
      "epoch": 0.29,
      "grad_norm": 85.5,
      "learning_rate": 3.5770601256141234e-05,
      "loss": 0.9214,
      "step": 327500
    },
    {
      "epoch": 0.29,
      "grad_norm": 43.75,
      "learning_rate": 3.576610218295031e-05,
      "loss": 0.8978,
      "step": 327600
    },
    {
      "epoch": 0.29,
      "grad_norm": 54.0,
      "learning_rate": 3.576160310975939e-05,
      "loss": 0.9393,
      "step": 327700
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.3125,
      "learning_rate": 3.575710403656847e-05,
      "loss": 0.9979,
      "step": 327800
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.002655029296875,
      "learning_rate": 3.575260496337754e-05,
      "loss": 0.9615,
      "step": 327900
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.138671875,
      "learning_rate": 3.5748105890186624e-05,
      "loss": 1.0525,
      "step": 328000
    },
    {
      "epoch": 0.29,
      "grad_norm": 672.0,
      "learning_rate": 3.57436068169957e-05,
      "loss": 0.8802,
      "step": 328100
    },
    {
      "epoch": 0.29,
      "grad_norm": 45.0,
      "learning_rate": 3.573910774380478e-05,
      "loss": 0.9863,
      "step": 328200
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.0625,
      "learning_rate": 3.573460867061386e-05,
      "loss": 1.0176,
      "step": 328300
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.03125,
      "learning_rate": 3.573010959742293e-05,
      "loss": 0.9135,
      "step": 328400
    },
    {
      "epoch": 0.29,
      "grad_norm": 117.0,
      "learning_rate": 3.5725610524232014e-05,
      "loss": 1.0044,
      "step": 328500
    },
    {
      "epoch": 0.29,
      "grad_norm": 143.0,
      "learning_rate": 3.572111145104109e-05,
      "loss": 1.1217,
      "step": 328600
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.039794921875,
      "learning_rate": 3.5716612377850165e-05,
      "loss": 0.9841,
      "step": 328700
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.375,
      "learning_rate": 3.571211330465924e-05,
      "loss": 0.9709,
      "step": 328800
    },
    {
      "epoch": 0.29,
      "grad_norm": 36.25,
      "learning_rate": 3.5707614231468316e-05,
      "loss": 1.0775,
      "step": 328900
    },
    {
      "epoch": 0.29,
      "grad_norm": 67.0,
      "learning_rate": 3.57031151582774e-05,
      "loss": 0.9732,
      "step": 329000
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.25,
      "learning_rate": 3.569861608508647e-05,
      "loss": 1.049,
      "step": 329100
    },
    {
      "epoch": 0.29,
      "grad_norm": 27.875,
      "learning_rate": 3.569411701189555e-05,
      "loss": 0.9743,
      "step": 329200
    },
    {
      "epoch": 0.29,
      "grad_norm": 26.25,
      "learning_rate": 3.568961793870463e-05,
      "loss": 0.8575,
      "step": 329300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.055908203125,
      "learning_rate": 3.5685118865513706e-05,
      "loss": 0.9061,
      "step": 329400
    },
    {
      "epoch": 0.29,
      "grad_norm": 71.0,
      "learning_rate": 3.568061979232278e-05,
      "loss": 1.0331,
      "step": 329500
    },
    {
      "epoch": 0.29,
      "grad_norm": 22.375,
      "learning_rate": 3.5676120719131864e-05,
      "loss": 0.8514,
      "step": 329600
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.125,
      "learning_rate": 3.567162164594094e-05,
      "loss": 1.1511,
      "step": 329700
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.006500244140625,
      "learning_rate": 3.5667122572750014e-05,
      "loss": 0.8774,
      "step": 329800
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.875,
      "learning_rate": 3.5662623499559096e-05,
      "loss": 0.9567,
      "step": 329900
    },
    {
      "epoch": 0.29,
      "grad_norm": 91.0,
      "learning_rate": 3.565812442636817e-05,
      "loss": 0.9579,
      "step": 330000
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1708984375,
      "learning_rate": 3.565362535317725e-05,
      "loss": 1.027,
      "step": 330100
    },
    {
      "epoch": 0.29,
      "grad_norm": 23.375,
      "learning_rate": 3.564912627998632e-05,
      "loss": 0.9305,
      "step": 330200
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.1875,
      "learning_rate": 3.56446272067954e-05,
      "loss": 1.0634,
      "step": 330300
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.75,
      "learning_rate": 3.564012813360448e-05,
      "loss": 0.9462,
      "step": 330400
    },
    {
      "epoch": 0.29,
      "grad_norm": 28.75,
      "learning_rate": 3.5635629060413555e-05,
      "loss": 0.9551,
      "step": 330500
    },
    {
      "epoch": 0.29,
      "grad_norm": 139.0,
      "learning_rate": 3.563112998722263e-05,
      "loss": 1.1876,
      "step": 330600
    },
    {
      "epoch": 0.29,
      "grad_norm": 22.75,
      "learning_rate": 3.562663091403171e-05,
      "loss": 0.9847,
      "step": 330700
    },
    {
      "epoch": 0.29,
      "grad_norm": 23.875,
      "learning_rate": 3.562213184084079e-05,
      "loss": 0.9629,
      "step": 330800
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.15234375,
      "learning_rate": 3.561763276764987e-05,
      "loss": 0.9678,
      "step": 330900
    },
    {
      "epoch": 0.29,
      "grad_norm": 115.0,
      "learning_rate": 3.5613133694458945e-05,
      "loss": 1.0061,
      "step": 331000
    },
    {
      "epoch": 0.29,
      "grad_norm": 32.75,
      "learning_rate": 3.560863462126802e-05,
      "loss": 1.0344,
      "step": 331100
    },
    {
      "epoch": 0.3,
      "grad_norm": 65.0,
      "learning_rate": 3.56041355480771e-05,
      "loss": 0.9472,
      "step": 331200
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0498046875,
      "learning_rate": 3.559963647488618e-05,
      "loss": 1.0289,
      "step": 331300
    },
    {
      "epoch": 0.3,
      "grad_norm": 65.0,
      "learning_rate": 3.559513740169525e-05,
      "loss": 0.9999,
      "step": 331400
    },
    {
      "epoch": 0.3,
      "grad_norm": 38.5,
      "learning_rate": 3.559063832850433e-05,
      "loss": 1.0686,
      "step": 331500
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.3125,
      "learning_rate": 3.5586139255313404e-05,
      "loss": 0.9857,
      "step": 331600
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.984375,
      "learning_rate": 3.5581640182122486e-05,
      "loss": 0.9316,
      "step": 331700
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.828125,
      "learning_rate": 3.557714110893156e-05,
      "loss": 1.0798,
      "step": 331800
    },
    {
      "epoch": 0.3,
      "grad_norm": 135.0,
      "learning_rate": 3.557264203574064e-05,
      "loss": 0.9377,
      "step": 331900
    },
    {
      "epoch": 0.3,
      "grad_norm": 20.625,
      "learning_rate": 3.556814296254972e-05,
      "loss": 0.9441,
      "step": 332000
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8125,
      "learning_rate": 3.5563643889358794e-05,
      "loss": 0.9284,
      "step": 332100
    },
    {
      "epoch": 0.3,
      "grad_norm": 50.25,
      "learning_rate": 3.555914481616787e-05,
      "loss": 1.0711,
      "step": 332200
    },
    {
      "epoch": 0.3,
      "grad_norm": 35.75,
      "learning_rate": 3.555464574297695e-05,
      "loss": 1.0742,
      "step": 332300
    },
    {
      "epoch": 0.3,
      "grad_norm": 29.125,
      "learning_rate": 3.555014666978603e-05,
      "loss": 0.8731,
      "step": 332400
    },
    {
      "epoch": 0.3,
      "grad_norm": 36.25,
      "learning_rate": 3.55456475965951e-05,
      "loss": 0.9239,
      "step": 332500
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.421875,
      "learning_rate": 3.5541148523404185e-05,
      "loss": 1.078,
      "step": 332600
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.625,
      "learning_rate": 3.553664945021325e-05,
      "loss": 0.9942,
      "step": 332700
    },
    {
      "epoch": 0.3,
      "grad_norm": 101.5,
      "learning_rate": 3.5532150377022335e-05,
      "loss": 1.0795,
      "step": 332800
    },
    {
      "epoch": 0.3,
      "grad_norm": 74.0,
      "learning_rate": 3.552765130383141e-05,
      "loss": 1.0356,
      "step": 332900
    },
    {
      "epoch": 0.3,
      "grad_norm": 57.25,
      "learning_rate": 3.5523152230640486e-05,
      "loss": 1.1318,
      "step": 333000
    },
    {
      "epoch": 0.3,
      "grad_norm": 118.5,
      "learning_rate": 3.551865315744957e-05,
      "loss": 0.9528,
      "step": 333100
    },
    {
      "epoch": 0.3,
      "grad_norm": 25.0,
      "learning_rate": 3.5514154084258643e-05,
      "loss": 0.9894,
      "step": 333200
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0018310546875,
      "learning_rate": 3.550965501106772e-05,
      "loss": 0.9249,
      "step": 333300
    },
    {
      "epoch": 0.3,
      "grad_norm": 56.0,
      "learning_rate": 3.55051559378768e-05,
      "loss": 1.0157,
      "step": 333400
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0771484375,
      "learning_rate": 3.5500656864685876e-05,
      "loss": 1.1021,
      "step": 333500
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.84375,
      "learning_rate": 3.549615779149495e-05,
      "loss": 1.0148,
      "step": 333600
    },
    {
      "epoch": 0.3,
      "grad_norm": 49.25,
      "learning_rate": 3.5491658718304034e-05,
      "loss": 1.0459,
      "step": 333700
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.96875,
      "learning_rate": 3.548715964511311e-05,
      "loss": 1.0988,
      "step": 333800
    },
    {
      "epoch": 0.3,
      "grad_norm": 36.5,
      "learning_rate": 3.548266057192219e-05,
      "loss": 1.0738,
      "step": 333900
    },
    {
      "epoch": 0.3,
      "grad_norm": 35.0,
      "learning_rate": 3.547816149873126e-05,
      "loss": 1.0493,
      "step": 334000
    },
    {
      "epoch": 0.3,
      "grad_norm": 304.0,
      "learning_rate": 3.5473662425540335e-05,
      "loss": 1.0974,
      "step": 334100
    },
    {
      "epoch": 0.3,
      "grad_norm": 56.75,
      "learning_rate": 3.546916335234942e-05,
      "loss": 0.9967,
      "step": 334200
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.0625,
      "learning_rate": 3.546466427915849e-05,
      "loss": 0.9931,
      "step": 334300
    },
    {
      "epoch": 0.3,
      "grad_norm": 22.375,
      "learning_rate": 3.5460165205967575e-05,
      "loss": 0.9772,
      "step": 334400
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.53125,
      "learning_rate": 3.545566613277665e-05,
      "loss": 1.0019,
      "step": 334500
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.640625,
      "learning_rate": 3.5451167059585725e-05,
      "loss": 1.0308,
      "step": 334600
    },
    {
      "epoch": 0.3,
      "grad_norm": 42.25,
      "learning_rate": 3.544666798639481e-05,
      "loss": 1.032,
      "step": 334700
    },
    {
      "epoch": 0.3,
      "grad_norm": 26.25,
      "learning_rate": 3.544216891320388e-05,
      "loss": 0.9726,
      "step": 334800
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.054443359375,
      "learning_rate": 3.543766984001296e-05,
      "loss": 1.074,
      "step": 334900
    },
    {
      "epoch": 0.3,
      "grad_norm": 29.0,
      "learning_rate": 3.543317076682204e-05,
      "loss": 0.9047,
      "step": 335000
    },
    {
      "epoch": 0.3,
      "grad_norm": 60.5,
      "learning_rate": 3.5428671693631116e-05,
      "loss": 0.9681,
      "step": 335100
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.3125,
      "learning_rate": 3.542417262044019e-05,
      "loss": 0.9256,
      "step": 335200
    },
    {
      "epoch": 0.3,
      "grad_norm": 39.0,
      "learning_rate": 3.5419673547249266e-05,
      "loss": 0.9005,
      "step": 335300
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0260009765625,
      "learning_rate": 3.541517447405834e-05,
      "loss": 1.0256,
      "step": 335400
    },
    {
      "epoch": 0.3,
      "grad_norm": 117.5,
      "learning_rate": 3.5410675400867424e-05,
      "loss": 1.0543,
      "step": 335500
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.8125,
      "learning_rate": 3.54061763276765e-05,
      "loss": 1.0191,
      "step": 335600
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.4375,
      "learning_rate": 3.5401677254485574e-05,
      "loss": 1.0416,
      "step": 335700
    },
    {
      "epoch": 0.3,
      "grad_norm": 31.25,
      "learning_rate": 3.5397178181294657e-05,
      "loss": 1.0673,
      "step": 335800
    },
    {
      "epoch": 0.3,
      "grad_norm": 20.875,
      "learning_rate": 3.539267910810373e-05,
      "loss": 0.9754,
      "step": 335900
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.140625,
      "learning_rate": 3.538818003491281e-05,
      "loss": 1.0473,
      "step": 336000
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.75,
      "learning_rate": 3.538368096172189e-05,
      "loss": 0.9843,
      "step": 336100
    },
    {
      "epoch": 0.3,
      "grad_norm": 27.75,
      "learning_rate": 3.5379181888530965e-05,
      "loss": 1.0561,
      "step": 336200
    },
    {
      "epoch": 0.3,
      "grad_norm": 16.875,
      "learning_rate": 3.537468281534004e-05,
      "loss": 0.809,
      "step": 336300
    },
    {
      "epoch": 0.3,
      "grad_norm": 43.25,
      "learning_rate": 3.537018374214912e-05,
      "loss": 1.0007,
      "step": 336400
    },
    {
      "epoch": 0.3,
      "grad_norm": 30.625,
      "learning_rate": 3.53656846689582e-05,
      "loss": 0.9604,
      "step": 336500
    },
    {
      "epoch": 0.3,
      "grad_norm": 66.5,
      "learning_rate": 3.536118559576727e-05,
      "loss": 1.0873,
      "step": 336600
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.5,
      "learning_rate": 3.535668652257635e-05,
      "loss": 1.2278,
      "step": 336700
    },
    {
      "epoch": 0.3,
      "grad_norm": 28.25,
      "learning_rate": 3.5352187449385423e-05,
      "loss": 1.078,
      "step": 336800
    },
    {
      "epoch": 0.3,
      "grad_norm": 103.5,
      "learning_rate": 3.5347688376194506e-05,
      "loss": 0.9048,
      "step": 336900
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.5,
      "learning_rate": 3.534318930300358e-05,
      "loss": 1.0114,
      "step": 337000
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.003631591796875,
      "learning_rate": 3.533869022981266e-05,
      "loss": 1.0559,
      "step": 337100
    },
    {
      "epoch": 0.3,
      "grad_norm": 100.5,
      "learning_rate": 3.533419115662174e-05,
      "loss": 0.9829,
      "step": 337200
    },
    {
      "epoch": 0.3,
      "grad_norm": 44.75,
      "learning_rate": 3.5329692083430814e-05,
      "loss": 1.0918,
      "step": 337300
    },
    {
      "epoch": 0.3,
      "grad_norm": 27.875,
      "learning_rate": 3.5325193010239896e-05,
      "loss": 0.8966,
      "step": 337400
    },
    {
      "epoch": 0.3,
      "grad_norm": 27.875,
      "learning_rate": 3.532069393704897e-05,
      "loss": 0.9762,
      "step": 337500
    },
    {
      "epoch": 0.3,
      "grad_norm": 22.625,
      "learning_rate": 3.5316194863858047e-05,
      "loss": 0.975,
      "step": 337600
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.75,
      "learning_rate": 3.531169579066713e-05,
      "loss": 0.9939,
      "step": 337700
    },
    {
      "epoch": 0.3,
      "grad_norm": 17.875,
      "learning_rate": 3.5307196717476204e-05,
      "loss": 0.9822,
      "step": 337800
    },
    {
      "epoch": 0.3,
      "grad_norm": 136.0,
      "learning_rate": 3.530269764428528e-05,
      "loss": 0.9633,
      "step": 337900
    },
    {
      "epoch": 0.3,
      "grad_norm": 68.5,
      "learning_rate": 3.5298198571094355e-05,
      "loss": 0.9347,
      "step": 338000
    },
    {
      "epoch": 0.3,
      "grad_norm": 38.0,
      "learning_rate": 3.529369949790343e-05,
      "loss": 0.9838,
      "step": 338100
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.25,
      "learning_rate": 3.528920042471251e-05,
      "loss": 0.8962,
      "step": 338200
    },
    {
      "epoch": 0.3,
      "grad_norm": 21.375,
      "learning_rate": 3.528470135152159e-05,
      "loss": 0.9742,
      "step": 338300
    },
    {
      "epoch": 0.3,
      "grad_norm": 310.0,
      "learning_rate": 3.528020227833066e-05,
      "loss": 1.0379,
      "step": 338400
    },
    {
      "epoch": 0.3,
      "grad_norm": 40.75,
      "learning_rate": 3.5275703205139745e-05,
      "loss": 0.9511,
      "step": 338500
    },
    {
      "epoch": 0.3,
      "grad_norm": 19.375,
      "learning_rate": 3.527120413194882e-05,
      "loss": 1.1389,
      "step": 338600
    },
    {
      "epoch": 0.3,
      "grad_norm": 95.0,
      "learning_rate": 3.5266705058757896e-05,
      "loss": 0.9534,
      "step": 338700
    },
    {
      "epoch": 0.3,
      "grad_norm": 32.25,
      "learning_rate": 3.526220598556698e-05,
      "loss": 0.8412,
      "step": 338800
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.90625,
      "learning_rate": 3.525770691237605e-05,
      "loss": 0.8912,
      "step": 338900
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.138671875,
      "learning_rate": 3.525320783918513e-05,
      "loss": 0.9501,
      "step": 339000
    },
    {
      "epoch": 0.3,
      "grad_norm": 22.5,
      "learning_rate": 3.524870876599421e-05,
      "loss": 0.9967,
      "step": 339100
    },
    {
      "epoch": 0.3,
      "grad_norm": 23.125,
      "learning_rate": 3.524420969280328e-05,
      "loss": 0.9,
      "step": 339200
    },
    {
      "epoch": 0.3,
      "grad_norm": 28.25,
      "learning_rate": 3.523971061961236e-05,
      "loss": 1.0166,
      "step": 339300
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.0,
      "learning_rate": 3.5235211546421437e-05,
      "loss": 0.9221,
      "step": 339400
    },
    {
      "epoch": 0.3,
      "grad_norm": 39.0,
      "learning_rate": 3.523071247323051e-05,
      "loss": 0.9019,
      "step": 339500
    },
    {
      "epoch": 0.3,
      "grad_norm": 39.5,
      "learning_rate": 3.5226213400039594e-05,
      "loss": 0.927,
      "step": 339600
    },
    {
      "epoch": 0.3,
      "grad_norm": 33.25,
      "learning_rate": 3.522171432684867e-05,
      "loss": 1.0473,
      "step": 339700
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.625,
      "learning_rate": 3.521721525365775e-05,
      "loss": 0.9213,
      "step": 339800
    },
    {
      "epoch": 0.3,
      "grad_norm": 31.125,
      "learning_rate": 3.521271618046683e-05,
      "loss": 1.0288,
      "step": 339900
    },
    {
      "epoch": 0.3,
      "grad_norm": 54.75,
      "learning_rate": 3.52082171072759e-05,
      "loss": 0.8142,
      "step": 340000
    },
    {
      "epoch": 0.3,
      "grad_norm": 58.0,
      "learning_rate": 3.5203718034084984e-05,
      "loss": 1.0138,
      "step": 340100
    },
    {
      "epoch": 0.3,
      "grad_norm": 61.25,
      "learning_rate": 3.519921896089406e-05,
      "loss": 0.8596,
      "step": 340200
    },
    {
      "epoch": 0.3,
      "grad_norm": 16.875,
      "learning_rate": 3.5194719887703135e-05,
      "loss": 0.8493,
      "step": 340300
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4140625,
      "learning_rate": 3.519022081451222e-05,
      "loss": 1.0531,
      "step": 340400
    },
    {
      "epoch": 0.3,
      "grad_norm": 25.625,
      "learning_rate": 3.5185721741321286e-05,
      "loss": 0.9826,
      "step": 340500
    },
    {
      "epoch": 0.3,
      "grad_norm": 68.0,
      "learning_rate": 3.518122266813037e-05,
      "loss": 0.9105,
      "step": 340600
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0,
      "learning_rate": 3.517672359493944e-05,
      "loss": 0.9738,
      "step": 340700
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8515625,
      "learning_rate": 3.517222452174852e-05,
      "loss": 0.8979,
      "step": 340800
    },
    {
      "epoch": 0.3,
      "grad_norm": 86.5,
      "learning_rate": 3.51677254485576e-05,
      "loss": 1.028,
      "step": 340900
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.060302734375,
      "learning_rate": 3.5163226375366676e-05,
      "loss": 0.9634,
      "step": 341000
    },
    {
      "epoch": 0.3,
      "grad_norm": 17.25,
      "learning_rate": 3.515872730217575e-05,
      "loss": 1.0268,
      "step": 341100
    },
    {
      "epoch": 0.3,
      "grad_norm": 32.75,
      "learning_rate": 3.515422822898483e-05,
      "loss": 0.9466,
      "step": 341200
    },
    {
      "epoch": 0.3,
      "grad_norm": 48.5,
      "learning_rate": 3.514972915579391e-05,
      "loss": 0.956,
      "step": 341300
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.0,
      "learning_rate": 3.5145230082602984e-05,
      "loss": 1.0007,
      "step": 341400
    },
    {
      "epoch": 0.3,
      "grad_norm": 62.25,
      "learning_rate": 3.5140731009412066e-05,
      "loss": 0.8858,
      "step": 341500
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.016845703125,
      "learning_rate": 3.513623193622114e-05,
      "loss": 1.0892,
      "step": 341600
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.4375,
      "learning_rate": 3.513173286303022e-05,
      "loss": 0.9665,
      "step": 341700
    },
    {
      "epoch": 0.3,
      "grad_norm": 72.0,
      "learning_rate": 3.512723378983929e-05,
      "loss": 0.962,
      "step": 341800
    },
    {
      "epoch": 0.3,
      "grad_norm": 41.25,
      "learning_rate": 3.512273471664837e-05,
      "loss": 1.023,
      "step": 341900
    },
    {
      "epoch": 0.3,
      "grad_norm": 53.25,
      "learning_rate": 3.511823564345745e-05,
      "loss": 1.0495,
      "step": 342000
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.019287109375,
      "learning_rate": 3.5113736570266525e-05,
      "loss": 1.119,
      "step": 342100
    },
    {
      "epoch": 0.3,
      "grad_norm": 97.5,
      "learning_rate": 3.51092374970756e-05,
      "loss": 1.1437,
      "step": 342200
    },
    {
      "epoch": 0.3,
      "grad_norm": 19.25,
      "learning_rate": 3.510473842388468e-05,
      "loss": 1.0745,
      "step": 342300
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.125,
      "learning_rate": 3.510023935069376e-05,
      "loss": 0.9692,
      "step": 342400
    },
    {
      "epoch": 0.31,
      "grad_norm": 33.25,
      "learning_rate": 3.509574027750284e-05,
      "loss": 0.8827,
      "step": 342500
    },
    {
      "epoch": 0.31,
      "grad_norm": 26.25,
      "learning_rate": 3.5091241204311915e-05,
      "loss": 0.9616,
      "step": 342600
    },
    {
      "epoch": 0.31,
      "grad_norm": 49.5,
      "learning_rate": 3.508674213112099e-05,
      "loss": 0.9545,
      "step": 342700
    },
    {
      "epoch": 0.31,
      "grad_norm": 19.875,
      "learning_rate": 3.508224305793007e-05,
      "loss": 0.9747,
      "step": 342800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4296875,
      "learning_rate": 3.507774398473915e-05,
      "loss": 1.014,
      "step": 342900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.318359375,
      "learning_rate": 3.507324491154822e-05,
      "loss": 0.8575,
      "step": 343000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.052978515625,
      "learning_rate": 3.50687458383573e-05,
      "loss": 0.9437,
      "step": 343100
    },
    {
      "epoch": 0.31,
      "grad_norm": 39.75,
      "learning_rate": 3.5064246765166374e-05,
      "loss": 1.1519,
      "step": 343200
    },
    {
      "epoch": 0.31,
      "grad_norm": 43.5,
      "learning_rate": 3.5059747691975456e-05,
      "loss": 1.1083,
      "step": 343300
    },
    {
      "epoch": 0.31,
      "grad_norm": 39.75,
      "learning_rate": 3.505524861878453e-05,
      "loss": 0.9813,
      "step": 343400
    },
    {
      "epoch": 0.31,
      "grad_norm": 46.75,
      "learning_rate": 3.505074954559361e-05,
      "loss": 0.9439,
      "step": 343500
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.75,
      "learning_rate": 3.504625047240269e-05,
      "loss": 1.1206,
      "step": 343600
    },
    {
      "epoch": 0.31,
      "grad_norm": 60.75,
      "learning_rate": 3.5041751399211764e-05,
      "loss": 0.9304,
      "step": 343700
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.5,
      "learning_rate": 3.503725232602084e-05,
      "loss": 0.9854,
      "step": 343800
    },
    {
      "epoch": 0.31,
      "grad_norm": 36.25,
      "learning_rate": 3.503275325282992e-05,
      "loss": 1.0822,
      "step": 343900
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.65625,
      "learning_rate": 3.5028254179639e-05,
      "loss": 0.999,
      "step": 344000
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.125,
      "learning_rate": 3.502375510644807e-05,
      "loss": 0.995,
      "step": 344100
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.5,
      "learning_rate": 3.5019256033257154e-05,
      "loss": 0.9917,
      "step": 344200
    },
    {
      "epoch": 0.31,
      "grad_norm": 54.75,
      "learning_rate": 3.501475696006623e-05,
      "loss": 0.9688,
      "step": 344300
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.5625,
      "learning_rate": 3.5010257886875305e-05,
      "loss": 1.1924,
      "step": 344400
    },
    {
      "epoch": 0.31,
      "grad_norm": 98.5,
      "learning_rate": 3.500575881368438e-05,
      "loss": 1.0433,
      "step": 344500
    },
    {
      "epoch": 0.31,
      "grad_norm": 42.25,
      "learning_rate": 3.5001259740493456e-05,
      "loss": 1.0896,
      "step": 344600
    },
    {
      "epoch": 0.31,
      "grad_norm": 70.5,
      "learning_rate": 3.499676066730254e-05,
      "loss": 1.0824,
      "step": 344700
    },
    {
      "epoch": 0.31,
      "grad_norm": 36.25,
      "learning_rate": 3.499226159411161e-05,
      "loss": 0.9353,
      "step": 344800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.859375,
      "learning_rate": 3.498776252092069e-05,
      "loss": 0.9933,
      "step": 344900
    },
    {
      "epoch": 0.31,
      "grad_norm": 250.0,
      "learning_rate": 3.498326344772977e-05,
      "loss": 1.0695,
      "step": 345000
    },
    {
      "epoch": 0.31,
      "grad_norm": 53.25,
      "learning_rate": 3.4978764374538846e-05,
      "loss": 0.9949,
      "step": 345100
    },
    {
      "epoch": 0.31,
      "grad_norm": 38.0,
      "learning_rate": 3.497426530134793e-05,
      "loss": 0.9965,
      "step": 345200
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0257568359375,
      "learning_rate": 3.4969766228157004e-05,
      "loss": 0.9574,
      "step": 345300
    },
    {
      "epoch": 0.31,
      "grad_norm": 65.0,
      "learning_rate": 3.496526715496608e-05,
      "loss": 0.8941,
      "step": 345400
    },
    {
      "epoch": 0.31,
      "grad_norm": 22.375,
      "learning_rate": 3.496076808177516e-05,
      "loss": 0.9229,
      "step": 345500
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.3125,
      "learning_rate": 3.4956269008584236e-05,
      "loss": 1.0172,
      "step": 345600
    },
    {
      "epoch": 0.31,
      "grad_norm": 24.25,
      "learning_rate": 3.4951769935393305e-05,
      "loss": 1.047,
      "step": 345700
    },
    {
      "epoch": 0.31,
      "grad_norm": 26.125,
      "learning_rate": 3.494727086220239e-05,
      "loss": 1.1756,
      "step": 345800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.033935546875,
      "learning_rate": 3.494277178901146e-05,
      "loss": 0.9654,
      "step": 345900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.015625,
      "learning_rate": 3.4938272715820544e-05,
      "loss": 0.9827,
      "step": 346000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0517578125,
      "learning_rate": 3.493377364262962e-05,
      "loss": 0.9261,
      "step": 346100
    },
    {
      "epoch": 0.31,
      "grad_norm": 63.0,
      "learning_rate": 3.4929274569438695e-05,
      "loss": 1.0586,
      "step": 346200
    },
    {
      "epoch": 0.31,
      "grad_norm": 33.75,
      "learning_rate": 3.492477549624778e-05,
      "loss": 0.974,
      "step": 346300
    },
    {
      "epoch": 0.31,
      "grad_norm": 53.5,
      "learning_rate": 3.492027642305685e-05,
      "loss": 0.9945,
      "step": 346400
    },
    {
      "epoch": 0.31,
      "grad_norm": 30.75,
      "learning_rate": 3.491577734986593e-05,
      "loss": 0.9252,
      "step": 346500
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.125,
      "learning_rate": 3.491127827667501e-05,
      "loss": 1.0636,
      "step": 346600
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.625,
      "learning_rate": 3.4906779203484085e-05,
      "loss": 0.9651,
      "step": 346700
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.0625,
      "learning_rate": 3.490228013029316e-05,
      "loss": 0.9804,
      "step": 346800
    },
    {
      "epoch": 0.31,
      "grad_norm": 346.0,
      "learning_rate": 3.489778105710224e-05,
      "loss": 0.8737,
      "step": 346900
    },
    {
      "epoch": 0.31,
      "grad_norm": 41.0,
      "learning_rate": 3.489328198391131e-05,
      "loss": 0.9257,
      "step": 347000
    },
    {
      "epoch": 0.31,
      "grad_norm": 28.375,
      "learning_rate": 3.4888782910720394e-05,
      "loss": 1.0121,
      "step": 347100
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.009033203125,
      "learning_rate": 3.488428383752947e-05,
      "loss": 0.9638,
      "step": 347200
    },
    {
      "epoch": 0.31,
      "grad_norm": 35.75,
      "learning_rate": 3.4879784764338544e-05,
      "loss": 0.9795,
      "step": 347300
    },
    {
      "epoch": 0.31,
      "grad_norm": 39.75,
      "learning_rate": 3.4875285691147626e-05,
      "loss": 0.9371,
      "step": 347400
    },
    {
      "epoch": 0.31,
      "grad_norm": 38.5,
      "learning_rate": 3.48707866179567e-05,
      "loss": 1.0468,
      "step": 347500
    },
    {
      "epoch": 0.31,
      "grad_norm": 158.0,
      "learning_rate": 3.486628754476578e-05,
      "loss": 1.0942,
      "step": 347600
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.84375,
      "learning_rate": 3.486178847157486e-05,
      "loss": 0.9997,
      "step": 347700
    },
    {
      "epoch": 0.31,
      "grad_norm": 111.0,
      "learning_rate": 3.4857289398383934e-05,
      "loss": 1.037,
      "step": 347800
    },
    {
      "epoch": 0.31,
      "grad_norm": 35.5,
      "learning_rate": 3.4852790325193017e-05,
      "loss": 1.0475,
      "step": 347900
    },
    {
      "epoch": 0.31,
      "grad_norm": 21.125,
      "learning_rate": 3.484829125200209e-05,
      "loss": 1.155,
      "step": 348000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.01043701171875,
      "learning_rate": 3.484379217881117e-05,
      "loss": 0.9906,
      "step": 348100
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.044189453125,
      "learning_rate": 3.483929310562025e-05,
      "loss": 0.9364,
      "step": 348200
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2119140625,
      "learning_rate": 3.483479403242932e-05,
      "loss": 1.0356,
      "step": 348300
    },
    {
      "epoch": 0.31,
      "grad_norm": 42.25,
      "learning_rate": 3.483029495923839e-05,
      "loss": 1.0871,
      "step": 348400
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.25,
      "learning_rate": 3.4825795886047475e-05,
      "loss": 1.0965,
      "step": 348500
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0172119140625,
      "learning_rate": 3.482129681285655e-05,
      "loss": 0.9703,
      "step": 348600
    },
    {
      "epoch": 0.31,
      "grad_norm": 31.875,
      "learning_rate": 3.481679773966563e-05,
      "loss": 0.9847,
      "step": 348700
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.375,
      "learning_rate": 3.481229866647471e-05,
      "loss": 0.9492,
      "step": 348800
    },
    {
      "epoch": 0.31,
      "grad_norm": 75.5,
      "learning_rate": 3.4807799593283784e-05,
      "loss": 1.0443,
      "step": 348900
    },
    {
      "epoch": 0.31,
      "grad_norm": 31.625,
      "learning_rate": 3.4803300520092866e-05,
      "loss": 0.8512,
      "step": 349000
    },
    {
      "epoch": 0.31,
      "grad_norm": 352.0,
      "learning_rate": 3.479880144690194e-05,
      "loss": 0.908,
      "step": 349100
    },
    {
      "epoch": 0.31,
      "grad_norm": 24.0,
      "learning_rate": 3.4794302373711016e-05,
      "loss": 0.9246,
      "step": 349200
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.03125,
      "learning_rate": 3.47898033005201e-05,
      "loss": 0.9786,
      "step": 349300
    },
    {
      "epoch": 0.31,
      "grad_norm": 22.0,
      "learning_rate": 3.4785304227329174e-05,
      "loss": 1.125,
      "step": 349400
    },
    {
      "epoch": 0.31,
      "grad_norm": 62.5,
      "learning_rate": 3.478080515413825e-05,
      "loss": 0.8742,
      "step": 349500
    },
    {
      "epoch": 0.31,
      "grad_norm": 164.0,
      "learning_rate": 3.4776306080947324e-05,
      "loss": 1.0632,
      "step": 349600
    },
    {
      "epoch": 0.31,
      "grad_norm": 146.0,
      "learning_rate": 3.47718070077564e-05,
      "loss": 1.0378,
      "step": 349700
    },
    {
      "epoch": 0.31,
      "grad_norm": 99.5,
      "learning_rate": 3.476730793456548e-05,
      "loss": 1.0271,
      "step": 349800
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.25,
      "learning_rate": 3.476280886137456e-05,
      "loss": 0.9971,
      "step": 349900
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.625,
      "learning_rate": 3.475830978818363e-05,
      "loss": 1.0899,
      "step": 350000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.470703125,
      "learning_rate": 3.4753810714992715e-05,
      "loss": 0.9545,
      "step": 350100
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.625,
      "learning_rate": 3.474931164180179e-05,
      "loss": 0.9789,
      "step": 350200
    },
    {
      "epoch": 0.31,
      "grad_norm": 86.5,
      "learning_rate": 3.4744812568610865e-05,
      "loss": 0.9448,
      "step": 350300
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.25,
      "learning_rate": 3.474031349541995e-05,
      "loss": 1.1435,
      "step": 350400
    },
    {
      "epoch": 0.31,
      "grad_norm": 165.0,
      "learning_rate": 3.473581442222902e-05,
      "loss": 1.1196,
      "step": 350500
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.5,
      "learning_rate": 3.4731315349038105e-05,
      "loss": 1.0311,
      "step": 350600
    },
    {
      "epoch": 0.31,
      "grad_norm": 76.5,
      "learning_rate": 3.472681627584718e-05,
      "loss": 1.0222,
      "step": 350700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0205078125,
      "learning_rate": 3.4722317202656256e-05,
      "loss": 1.0105,
      "step": 350800
    },
    {
      "epoch": 0.31,
      "grad_norm": 32.0,
      "learning_rate": 3.471781812946533e-05,
      "loss": 0.8543,
      "step": 350900
    },
    {
      "epoch": 0.31,
      "grad_norm": 43.25,
      "learning_rate": 3.4713319056274406e-05,
      "loss": 1.0365,
      "step": 351000
    },
    {
      "epoch": 0.31,
      "grad_norm": 31.5,
      "learning_rate": 3.470881998308348e-05,
      "loss": 0.9772,
      "step": 351100
    },
    {
      "epoch": 0.31,
      "grad_norm": 41.0,
      "learning_rate": 3.4704320909892564e-05,
      "loss": 0.9123,
      "step": 351200
    },
    {
      "epoch": 0.31,
      "grad_norm": 34.25,
      "learning_rate": 3.469982183670164e-05,
      "loss": 0.8525,
      "step": 351300
    },
    {
      "epoch": 0.31,
      "grad_norm": 374.0,
      "learning_rate": 3.469532276351072e-05,
      "loss": 0.8905,
      "step": 351400
    },
    {
      "epoch": 0.31,
      "grad_norm": 35.75,
      "learning_rate": 3.4690823690319797e-05,
      "loss": 0.9893,
      "step": 351500
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.0,
      "learning_rate": 3.468632461712887e-05,
      "loss": 1.1033,
      "step": 351600
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.75,
      "learning_rate": 3.4681825543937954e-05,
      "loss": 0.9942,
      "step": 351700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0269775390625,
      "learning_rate": 3.467732647074703e-05,
      "loss": 0.8974,
      "step": 351800
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.75,
      "learning_rate": 3.4672827397556105e-05,
      "loss": 1.0212,
      "step": 351900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.455078125,
      "learning_rate": 3.466832832436519e-05,
      "loss": 0.9682,
      "step": 352000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0206298828125,
      "learning_rate": 3.466382925117426e-05,
      "loss": 1.0093,
      "step": 352100
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.875,
      "learning_rate": 3.465933017798334e-05,
      "loss": 0.7791,
      "step": 352200
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.087890625,
      "learning_rate": 3.465483110479241e-05,
      "loss": 1.0043,
      "step": 352300
    },
    {
      "epoch": 0.31,
      "grad_norm": 43.5,
      "learning_rate": 3.465033203160149e-05,
      "loss": 0.9916,
      "step": 352400
    },
    {
      "epoch": 0.31,
      "grad_norm": 1040.0,
      "learning_rate": 3.464583295841057e-05,
      "loss": 1.0605,
      "step": 352500
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.546875,
      "learning_rate": 3.4641333885219646e-05,
      "loss": 1.0786,
      "step": 352600
    },
    {
      "epoch": 0.31,
      "grad_norm": 28.875,
      "learning_rate": 3.463683481202872e-05,
      "loss": 0.8849,
      "step": 352700
    },
    {
      "epoch": 0.31,
      "grad_norm": 68.0,
      "learning_rate": 3.46323357388378e-05,
      "loss": 0.8792,
      "step": 352800
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.6875,
      "learning_rate": 3.462783666564688e-05,
      "loss": 1.131,
      "step": 352900
    },
    {
      "epoch": 0.31,
      "grad_norm": 27.625,
      "learning_rate": 3.4623337592455954e-05,
      "loss": 1.0118,
      "step": 353000
    },
    {
      "epoch": 0.31,
      "grad_norm": 24.5,
      "learning_rate": 3.4618838519265036e-05,
      "loss": 1.0404,
      "step": 353100
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.75,
      "learning_rate": 3.461433944607411e-05,
      "loss": 0.9962,
      "step": 353200
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.75,
      "learning_rate": 3.4609840372883187e-05,
      "loss": 0.8976,
      "step": 353300
    },
    {
      "epoch": 0.31,
      "grad_norm": 36.25,
      "learning_rate": 3.460534129969227e-05,
      "loss": 1.1852,
      "step": 353400
    },
    {
      "epoch": 0.31,
      "grad_norm": 34.5,
      "learning_rate": 3.460084222650134e-05,
      "loss": 0.9735,
      "step": 353500
    },
    {
      "epoch": 0.31,
      "grad_norm": 56.0,
      "learning_rate": 3.459634315331042e-05,
      "loss": 0.9293,
      "step": 353600
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.1875,
      "learning_rate": 3.4591844080119495e-05,
      "loss": 0.9346,
      "step": 353700
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.0966796875,
      "learning_rate": 3.458734500692857e-05,
      "loss": 1.143,
      "step": 353800
    },
    {
      "epoch": 0.32,
      "grad_norm": 46.75,
      "learning_rate": 3.458284593373765e-05,
      "loss": 0.9157,
      "step": 353900
    },
    {
      "epoch": 0.32,
      "grad_norm": 68.0,
      "learning_rate": 3.457834686054673e-05,
      "loss": 0.971,
      "step": 354000
    },
    {
      "epoch": 0.32,
      "grad_norm": 47.25,
      "learning_rate": 3.457384778735581e-05,
      "loss": 1.0524,
      "step": 354100
    },
    {
      "epoch": 0.32,
      "grad_norm": 100.0,
      "learning_rate": 3.4569348714164885e-05,
      "loss": 1.2063,
      "step": 354200
    },
    {
      "epoch": 0.32,
      "grad_norm": 63.75,
      "learning_rate": 3.456484964097396e-05,
      "loss": 0.9138,
      "step": 354300
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.5625,
      "learning_rate": 3.456035056778304e-05,
      "loss": 0.8864,
      "step": 354400
    },
    {
      "epoch": 0.32,
      "grad_norm": 1688.0,
      "learning_rate": 3.455585149459212e-05,
      "loss": 1.1033,
      "step": 354500
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.078125,
      "learning_rate": 3.455135242140119e-05,
      "loss": 0.9511,
      "step": 354600
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.25,
      "learning_rate": 3.4546853348210275e-05,
      "loss": 0.9354,
      "step": 354700
    },
    {
      "epoch": 0.32,
      "grad_norm": 42.75,
      "learning_rate": 3.4542354275019344e-05,
      "loss": 1.0069,
      "step": 354800
    },
    {
      "epoch": 0.32,
      "grad_norm": 162.0,
      "learning_rate": 3.4537855201828426e-05,
      "loss": 1.0169,
      "step": 354900
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.125,
      "learning_rate": 3.45333561286375e-05,
      "loss": 0.911,
      "step": 355000
    },
    {
      "epoch": 0.32,
      "grad_norm": 42.0,
      "learning_rate": 3.4528857055446577e-05,
      "loss": 1.1081,
      "step": 355100
    },
    {
      "epoch": 0.32,
      "grad_norm": 28.25,
      "learning_rate": 3.452435798225566e-05,
      "loss": 0.9224,
      "step": 355200
    },
    {
      "epoch": 0.32,
      "grad_norm": 56.0,
      "learning_rate": 3.4519858909064734e-05,
      "loss": 1.0065,
      "step": 355300
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.72265625,
      "learning_rate": 3.451535983587381e-05,
      "loss": 1.0743,
      "step": 355400
    },
    {
      "epoch": 0.32,
      "grad_norm": 25.875,
      "learning_rate": 3.451086076268289e-05,
      "loss": 1.0195,
      "step": 355500
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.125,
      "learning_rate": 3.450636168949197e-05,
      "loss": 1.161,
      "step": 355600
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.625,
      "learning_rate": 3.450186261630104e-05,
      "loss": 0.913,
      "step": 355700
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.5625,
      "learning_rate": 3.4497363543110124e-05,
      "loss": 0.8156,
      "step": 355800
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.0,
      "learning_rate": 3.44928644699192e-05,
      "loss": 0.9391,
      "step": 355900
    },
    {
      "epoch": 0.32,
      "grad_norm": 40.5,
      "learning_rate": 3.4488365396728275e-05,
      "loss": 0.8788,
      "step": 356000
    },
    {
      "epoch": 0.32,
      "grad_norm": 22.375,
      "learning_rate": 3.448386632353735e-05,
      "loss": 0.9859,
      "step": 356100
    },
    {
      "epoch": 0.32,
      "grad_norm": 23.625,
      "learning_rate": 3.4479367250346426e-05,
      "loss": 1.0059,
      "step": 356200
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.5,
      "learning_rate": 3.447486817715551e-05,
      "loss": 1.115,
      "step": 356300
    },
    {
      "epoch": 0.32,
      "grad_norm": 18.0,
      "learning_rate": 3.447036910396458e-05,
      "loss": 0.9832,
      "step": 356400
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.9375,
      "learning_rate": 3.446587003077366e-05,
      "loss": 0.9601,
      "step": 356500
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.0625,
      "learning_rate": 3.446137095758274e-05,
      "loss": 1.1986,
      "step": 356600
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.21875,
      "learning_rate": 3.4456871884391816e-05,
      "loss": 1.0122,
      "step": 356700
    },
    {
      "epoch": 0.32,
      "grad_norm": 46.25,
      "learning_rate": 3.44523728112009e-05,
      "loss": 0.7792,
      "step": 356800
    },
    {
      "epoch": 0.32,
      "grad_norm": 45.0,
      "learning_rate": 3.444787373800997e-05,
      "loss": 1.0879,
      "step": 356900
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.0,
      "learning_rate": 3.444337466481905e-05,
      "loss": 0.9483,
      "step": 357000
    },
    {
      "epoch": 0.32,
      "grad_norm": 20.125,
      "learning_rate": 3.443887559162813e-05,
      "loss": 0.9914,
      "step": 357100
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.5,
      "learning_rate": 3.4434376518437206e-05,
      "loss": 0.9563,
      "step": 357200
    },
    {
      "epoch": 0.32,
      "grad_norm": 20.875,
      "learning_rate": 3.442987744524628e-05,
      "loss": 1.0086,
      "step": 357300
    },
    {
      "epoch": 0.32,
      "grad_norm": 51.25,
      "learning_rate": 3.442537837205536e-05,
      "loss": 1.0248,
      "step": 357400
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.25,
      "learning_rate": 3.442087929886443e-05,
      "loss": 0.9769,
      "step": 357500
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.25,
      "learning_rate": 3.4416380225673514e-05,
      "loss": 1.0876,
      "step": 357600
    },
    {
      "epoch": 0.32,
      "grad_norm": 47.5,
      "learning_rate": 3.441188115248259e-05,
      "loss": 0.9682,
      "step": 357700
    },
    {
      "epoch": 0.32,
      "grad_norm": 99.5,
      "learning_rate": 3.4407382079291665e-05,
      "loss": 1.031,
      "step": 357800
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.875,
      "learning_rate": 3.440288300610075e-05,
      "loss": 1.0673,
      "step": 357900
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.8125,
      "learning_rate": 3.439838393290982e-05,
      "loss": 1.0278,
      "step": 358000
    },
    {
      "epoch": 0.32,
      "grad_norm": 35.5,
      "learning_rate": 3.43938848597189e-05,
      "loss": 1.0423,
      "step": 358100
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.8125,
      "learning_rate": 3.438938578652798e-05,
      "loss": 0.9551,
      "step": 358200
    },
    {
      "epoch": 0.32,
      "grad_norm": 18.125,
      "learning_rate": 3.4384886713337055e-05,
      "loss": 0.9817,
      "step": 358300
    },
    {
      "epoch": 0.32,
      "grad_norm": 43.0,
      "learning_rate": 3.438038764014613e-05,
      "loss": 1.0417,
      "step": 358400
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.25,
      "learning_rate": 3.437588856695521e-05,
      "loss": 0.8936,
      "step": 358500
    },
    {
      "epoch": 0.32,
      "grad_norm": 43.5,
      "learning_rate": 3.437138949376429e-05,
      "loss": 1.0977,
      "step": 358600
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.1875,
      "learning_rate": 3.436689042057336e-05,
      "loss": 1.1079,
      "step": 358700
    },
    {
      "epoch": 0.32,
      "grad_norm": 113.5,
      "learning_rate": 3.436239134738244e-05,
      "loss": 0.9851,
      "step": 358800
    },
    {
      "epoch": 0.32,
      "grad_norm": 36.25,
      "learning_rate": 3.4357892274191514e-05,
      "loss": 1.0251,
      "step": 358900
    },
    {
      "epoch": 0.32,
      "grad_norm": 73.0,
      "learning_rate": 3.4353393201000596e-05,
      "loss": 0.9387,
      "step": 359000
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.0625,
      "learning_rate": 3.434889412780967e-05,
      "loss": 1.0755,
      "step": 359100
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.625,
      "learning_rate": 3.434439505461875e-05,
      "loss": 1.0748,
      "step": 359200
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.0,
      "learning_rate": 3.433989598142783e-05,
      "loss": 0.9022,
      "step": 359300
    },
    {
      "epoch": 0.32,
      "grad_norm": 103.5,
      "learning_rate": 3.4335396908236904e-05,
      "loss": 1.0759,
      "step": 359400
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.5,
      "learning_rate": 3.4330897835045986e-05,
      "loss": 0.9412,
      "step": 359500
    },
    {
      "epoch": 0.32,
      "grad_norm": 4640.0,
      "learning_rate": 3.432639876185506e-05,
      "loss": 0.7769,
      "step": 359600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.004974365234375,
      "learning_rate": 3.432189968866414e-05,
      "loss": 1.0443,
      "step": 359700
    },
    {
      "epoch": 0.32,
      "grad_norm": 55.5,
      "learning_rate": 3.431740061547322e-05,
      "loss": 1.0175,
      "step": 359800
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.125,
      "learning_rate": 3.4312901542282295e-05,
      "loss": 0.9163,
      "step": 359900
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.265625,
      "learning_rate": 3.430840246909136e-05,
      "loss": 0.8714,
      "step": 360000
    },
    {
      "epoch": 0.32,
      "grad_norm": 27.0,
      "learning_rate": 3.4303903395900445e-05,
      "loss": 0.903,
      "step": 360100
    },
    {
      "epoch": 0.32,
      "grad_norm": 52.25,
      "learning_rate": 3.429940432270952e-05,
      "loss": 1.0778,
      "step": 360200
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.75,
      "learning_rate": 3.42949052495186e-05,
      "loss": 0.7507,
      "step": 360300
    },
    {
      "epoch": 0.32,
      "grad_norm": 44.0,
      "learning_rate": 3.429040617632768e-05,
      "loss": 0.9221,
      "step": 360400
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.25,
      "learning_rate": 3.428590710313675e-05,
      "loss": 1.0659,
      "step": 360500
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.625,
      "learning_rate": 3.4281408029945835e-05,
      "loss": 1.0189,
      "step": 360600
    },
    {
      "epoch": 0.32,
      "grad_norm": 28.0,
      "learning_rate": 3.427690895675491e-05,
      "loss": 0.9743,
      "step": 360700
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.875,
      "learning_rate": 3.4272409883563986e-05,
      "loss": 0.9364,
      "step": 360800
    },
    {
      "epoch": 0.32,
      "grad_norm": 23.5,
      "learning_rate": 3.426791081037307e-05,
      "loss": 1.0681,
      "step": 360900
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.625,
      "learning_rate": 3.4263411737182144e-05,
      "loss": 0.954,
      "step": 361000
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.030517578125,
      "learning_rate": 3.425891266399122e-05,
      "loss": 0.7981,
      "step": 361100
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.353515625,
      "learning_rate": 3.42544135908003e-05,
      "loss": 0.9293,
      "step": 361200
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.25,
      "learning_rate": 3.424991451760937e-05,
      "loss": 0.9397,
      "step": 361300
    },
    {
      "epoch": 0.32,
      "grad_norm": 58.25,
      "learning_rate": 3.424541544441845e-05,
      "loss": 1.0275,
      "step": 361400
    },
    {
      "epoch": 0.32,
      "grad_norm": 46.0,
      "learning_rate": 3.424091637122753e-05,
      "loss": 1.0131,
      "step": 361500
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.875,
      "learning_rate": 3.42364172980366e-05,
      "loss": 0.904,
      "step": 361600
    },
    {
      "epoch": 0.32,
      "grad_norm": 26.25,
      "learning_rate": 3.4231918224845685e-05,
      "loss": 0.9013,
      "step": 361700
    },
    {
      "epoch": 0.32,
      "grad_norm": 33.5,
      "learning_rate": 3.422741915165476e-05,
      "loss": 1.0755,
      "step": 361800
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1416015625,
      "learning_rate": 3.4222920078463835e-05,
      "loss": 1.0032,
      "step": 361900
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.09375,
      "learning_rate": 3.421842100527292e-05,
      "loss": 1.1277,
      "step": 362000
    },
    {
      "epoch": 0.32,
      "grad_norm": 34.25,
      "learning_rate": 3.421392193208199e-05,
      "loss": 0.8355,
      "step": 362100
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6015625,
      "learning_rate": 3.4209422858891075e-05,
      "loss": 0.8673,
      "step": 362200
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.9375,
      "learning_rate": 3.420492378570015e-05,
      "loss": 1.0876,
      "step": 362300
    },
    {
      "epoch": 0.32,
      "grad_norm": 36.5,
      "learning_rate": 3.4200424712509225e-05,
      "loss": 0.9968,
      "step": 362400
    },
    {
      "epoch": 0.32,
      "grad_norm": 58.0,
      "learning_rate": 3.419592563931831e-05,
      "loss": 1.1515,
      "step": 362500
    },
    {
      "epoch": 0.32,
      "grad_norm": 33.5,
      "learning_rate": 3.4191426566127376e-05,
      "loss": 0.9773,
      "step": 362600
    },
    {
      "epoch": 0.32,
      "grad_norm": 130.0,
      "learning_rate": 3.418692749293645e-05,
      "loss": 0.9652,
      "step": 362700
    },
    {
      "epoch": 0.32,
      "grad_norm": 52.0,
      "learning_rate": 3.4182428419745534e-05,
      "loss": 0.8992,
      "step": 362800
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.5,
      "learning_rate": 3.417792934655461e-05,
      "loss": 1.0248,
      "step": 362900
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.375,
      "learning_rate": 3.417343027336369e-05,
      "loss": 0.9995,
      "step": 363000
    },
    {
      "epoch": 0.32,
      "grad_norm": 60.0,
      "learning_rate": 3.4168931200172766e-05,
      "loss": 1.0267,
      "step": 363100
    },
    {
      "epoch": 0.32,
      "grad_norm": 19.625,
      "learning_rate": 3.416443212698184e-05,
      "loss": 1.1082,
      "step": 363200
    },
    {
      "epoch": 0.32,
      "grad_norm": 47.0,
      "learning_rate": 3.4159933053790924e-05,
      "loss": 1.0299,
      "step": 363300
    },
    {
      "epoch": 0.32,
      "grad_norm": 58.75,
      "learning_rate": 3.41554339806e-05,
      "loss": 0.9064,
      "step": 363400
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.4375,
      "learning_rate": 3.4150934907409075e-05,
      "loss": 1.0907,
      "step": 363500
    },
    {
      "epoch": 0.32,
      "grad_norm": 22.875,
      "learning_rate": 3.414643583421816e-05,
      "loss": 0.832,
      "step": 363600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25390625,
      "learning_rate": 3.414193676102723e-05,
      "loss": 1.0826,
      "step": 363700
    },
    {
      "epoch": 0.32,
      "grad_norm": 35.75,
      "learning_rate": 3.413743768783631e-05,
      "loss": 1.0584,
      "step": 363800
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.057373046875,
      "learning_rate": 3.413293861464538e-05,
      "loss": 0.9768,
      "step": 363900
    },
    {
      "epoch": 0.32,
      "grad_norm": 34.5,
      "learning_rate": 3.412843954145446e-05,
      "loss": 1.1146,
      "step": 364000
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.326171875,
      "learning_rate": 3.412394046826354e-05,
      "loss": 1.0217,
      "step": 364100
    },
    {
      "epoch": 0.32,
      "grad_norm": 31.5,
      "learning_rate": 3.4119441395072615e-05,
      "loss": 1.0237,
      "step": 364200
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.00518798828125,
      "learning_rate": 3.411494232188169e-05,
      "loss": 0.8757,
      "step": 364300
    },
    {
      "epoch": 0.32,
      "grad_norm": 44.25,
      "learning_rate": 3.411044324869077e-05,
      "loss": 1.0125,
      "step": 364400
    },
    {
      "epoch": 0.32,
      "grad_norm": 84.0,
      "learning_rate": 3.410594417549985e-05,
      "loss": 1.0171,
      "step": 364500
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.0,
      "learning_rate": 3.4101445102308924e-05,
      "loss": 1.0036,
      "step": 364600
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.3125,
      "learning_rate": 3.4096946029118006e-05,
      "loss": 0.9506,
      "step": 364700
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.625,
      "learning_rate": 3.409244695592708e-05,
      "loss": 0.8712,
      "step": 364800
    },
    {
      "epoch": 0.33,
      "grad_norm": 59.25,
      "learning_rate": 3.408794788273616e-05,
      "loss": 0.9358,
      "step": 364900
    },
    {
      "epoch": 0.33,
      "grad_norm": 21.375,
      "learning_rate": 3.408344880954524e-05,
      "loss": 1.0977,
      "step": 365000
    },
    {
      "epoch": 0.33,
      "grad_norm": 33.75,
      "learning_rate": 3.4078949736354314e-05,
      "loss": 1.0174,
      "step": 365100
    },
    {
      "epoch": 0.33,
      "grad_norm": 24.875,
      "learning_rate": 3.407445066316339e-05,
      "loss": 1.0531,
      "step": 365200
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.14453125,
      "learning_rate": 3.4069951589972465e-05,
      "loss": 0.9368,
      "step": 365300
    },
    {
      "epoch": 0.33,
      "grad_norm": 49.75,
      "learning_rate": 3.406545251678154e-05,
      "loss": 1.1169,
      "step": 365400
    },
    {
      "epoch": 0.33,
      "grad_norm": 33.0,
      "learning_rate": 3.406095344359062e-05,
      "loss": 0.9764,
      "step": 365500
    },
    {
      "epoch": 0.33,
      "grad_norm": 50.0,
      "learning_rate": 3.40564543703997e-05,
      "loss": 1.1753,
      "step": 365600
    },
    {
      "epoch": 0.33,
      "grad_norm": 65.0,
      "learning_rate": 3.405195529720878e-05,
      "loss": 0.8653,
      "step": 365700
    },
    {
      "epoch": 0.33,
      "grad_norm": 78.0,
      "learning_rate": 3.4047456224017855e-05,
      "loss": 0.9461,
      "step": 365800
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.125,
      "learning_rate": 3.404295715082693e-05,
      "loss": 1.1266,
      "step": 365900
    },
    {
      "epoch": 0.33,
      "grad_norm": 96.5,
      "learning_rate": 3.403845807763601e-05,
      "loss": 1.1212,
      "step": 366000
    },
    {
      "epoch": 0.33,
      "grad_norm": 32.75,
      "learning_rate": 3.403395900444509e-05,
      "loss": 1.0391,
      "step": 366100
    },
    {
      "epoch": 0.33,
      "grad_norm": 29.875,
      "learning_rate": 3.402945993125416e-05,
      "loss": 1.0962,
      "step": 366200
    },
    {
      "epoch": 0.33,
      "grad_norm": 41.75,
      "learning_rate": 3.4024960858063245e-05,
      "loss": 0.982,
      "step": 366300
    },
    {
      "epoch": 0.33,
      "grad_norm": 832.0,
      "learning_rate": 3.402046178487232e-05,
      "loss": 0.9597,
      "step": 366400
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.0,
      "learning_rate": 3.4015962711681396e-05,
      "loss": 0.8734,
      "step": 366500
    },
    {
      "epoch": 0.33,
      "grad_norm": 32.5,
      "learning_rate": 3.401146363849047e-05,
      "loss": 1.1441,
      "step": 366600
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.5,
      "learning_rate": 3.4006964565299546e-05,
      "loss": 1.0511,
      "step": 366700
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.625,
      "learning_rate": 3.400246549210863e-05,
      "loss": 1.0366,
      "step": 366800
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.25,
      "learning_rate": 3.3997966418917704e-05,
      "loss": 0.9527,
      "step": 366900
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2734375,
      "learning_rate": 3.399346734572678e-05,
      "loss": 0.9249,
      "step": 367000
    },
    {
      "epoch": 0.33,
      "grad_norm": 29.375,
      "learning_rate": 3.398896827253586e-05,
      "loss": 0.944,
      "step": 367100
    },
    {
      "epoch": 0.33,
      "grad_norm": 83.0,
      "learning_rate": 3.398446919934494e-05,
      "loss": 0.8695,
      "step": 367200
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.25,
      "learning_rate": 3.397997012615401e-05,
      "loss": 0.9981,
      "step": 367300
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.5,
      "learning_rate": 3.3975471052963094e-05,
      "loss": 1.0779,
      "step": 367400
    },
    {
      "epoch": 0.33,
      "grad_norm": 18.25,
      "learning_rate": 3.397097197977217e-05,
      "loss": 0.954,
      "step": 367500
    },
    {
      "epoch": 0.33,
      "grad_norm": 26.0,
      "learning_rate": 3.396647290658125e-05,
      "loss": 1.163,
      "step": 367600
    },
    {
      "epoch": 0.33,
      "grad_norm": 47.5,
      "learning_rate": 3.396197383339033e-05,
      "loss": 0.98,
      "step": 367700
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.3125,
      "learning_rate": 3.3957474760199395e-05,
      "loss": 1.0356,
      "step": 367800
    },
    {
      "epoch": 0.33,
      "grad_norm": 1960.0,
      "learning_rate": 3.395297568700848e-05,
      "loss": 0.9846,
      "step": 367900
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.25,
      "learning_rate": 3.394847661381755e-05,
      "loss": 0.8628,
      "step": 368000
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.07470703125,
      "learning_rate": 3.394397754062663e-05,
      "loss": 1.1052,
      "step": 368100
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.004608154296875,
      "learning_rate": 3.393947846743571e-05,
      "loss": 1.0655,
      "step": 368200
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4375,
      "learning_rate": 3.3934979394244786e-05,
      "loss": 0.9498,
      "step": 368300
    },
    {
      "epoch": 0.33,
      "grad_norm": 112.5,
      "learning_rate": 3.393048032105387e-05,
      "loss": 0.9401,
      "step": 368400
    },
    {
      "epoch": 0.33,
      "grad_norm": 34.75,
      "learning_rate": 3.392598124786294e-05,
      "loss": 1.0478,
      "step": 368500
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1474609375,
      "learning_rate": 3.392148217467202e-05,
      "loss": 1.188,
      "step": 368600
    },
    {
      "epoch": 0.33,
      "grad_norm": 1576.0,
      "learning_rate": 3.39169831014811e-05,
      "loss": 1.0604,
      "step": 368700
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.71875,
      "learning_rate": 3.3912484028290176e-05,
      "loss": 1.0789,
      "step": 368800
    },
    {
      "epoch": 0.33,
      "grad_norm": 40.75,
      "learning_rate": 3.390798495509925e-05,
      "loss": 0.973,
      "step": 368900
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.9375,
      "learning_rate": 3.390348588190833e-05,
      "loss": 0.9703,
      "step": 369000
    },
    {
      "epoch": 0.33,
      "grad_norm": 32.75,
      "learning_rate": 3.38989868087174e-05,
      "loss": 1.1804,
      "step": 369100
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.04052734375,
      "learning_rate": 3.3894487735526484e-05,
      "loss": 0.9602,
      "step": 369200
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.46875,
      "learning_rate": 3.388998866233556e-05,
      "loss": 1.0113,
      "step": 369300
    },
    {
      "epoch": 0.33,
      "grad_norm": 18.375,
      "learning_rate": 3.3885489589144635e-05,
      "loss": 1.0388,
      "step": 369400
    },
    {
      "epoch": 0.33,
      "grad_norm": 21.75,
      "learning_rate": 3.388099051595372e-05,
      "loss": 1.0397,
      "step": 369500
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.359375,
      "learning_rate": 3.387649144276279e-05,
      "loss": 0.982,
      "step": 369600
    },
    {
      "epoch": 0.33,
      "grad_norm": 199.0,
      "learning_rate": 3.387199236957187e-05,
      "loss": 0.9649,
      "step": 369700
    },
    {
      "epoch": 0.33,
      "grad_norm": 56.0,
      "learning_rate": 3.386749329638095e-05,
      "loss": 1.0288,
      "step": 369800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.0673828125,
      "learning_rate": 3.3862994223190025e-05,
      "loss": 1.1247,
      "step": 369900
    },
    {
      "epoch": 0.33,
      "grad_norm": 35.25,
      "learning_rate": 3.38584951499991e-05,
      "loss": 0.9348,
      "step": 370000
    },
    {
      "epoch": 0.33,
      "grad_norm": 20.75,
      "learning_rate": 3.385399607680818e-05,
      "loss": 1.1281,
      "step": 370100
    },
    {
      "epoch": 0.33,
      "grad_norm": 66.0,
      "learning_rate": 3.384949700361726e-05,
      "loss": 1.0813,
      "step": 370200
    },
    {
      "epoch": 0.33,
      "grad_norm": 170.0,
      "learning_rate": 3.384499793042633e-05,
      "loss": 1.1549,
      "step": 370300
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.25,
      "learning_rate": 3.384049885723541e-05,
      "loss": 1.0384,
      "step": 370400
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.5,
      "learning_rate": 3.3835999784044484e-05,
      "loss": 0.9709,
      "step": 370500
    },
    {
      "epoch": 0.33,
      "grad_norm": 44.5,
      "learning_rate": 3.3831500710853566e-05,
      "loss": 1.0332,
      "step": 370600
    },
    {
      "epoch": 0.33,
      "grad_norm": 64.0,
      "learning_rate": 3.382700163766264e-05,
      "loss": 1.0561,
      "step": 370700
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.375,
      "learning_rate": 3.3822502564471717e-05,
      "loss": 1.0035,
      "step": 370800
    },
    {
      "epoch": 0.33,
      "grad_norm": 173.0,
      "learning_rate": 3.38180034912808e-05,
      "loss": 0.9603,
      "step": 370900
    },
    {
      "epoch": 0.33,
      "grad_norm": 33.0,
      "learning_rate": 3.3813504418089874e-05,
      "loss": 1.0326,
      "step": 371000
    },
    {
      "epoch": 0.33,
      "grad_norm": 29.375,
      "learning_rate": 3.3809005344898956e-05,
      "loss": 1.0167,
      "step": 371100
    },
    {
      "epoch": 0.33,
      "grad_norm": 231.0,
      "learning_rate": 3.380450627170803e-05,
      "loss": 0.9685,
      "step": 371200
    },
    {
      "epoch": 0.33,
      "grad_norm": 19.25,
      "learning_rate": 3.380000719851711e-05,
      "loss": 0.9463,
      "step": 371300
    },
    {
      "epoch": 0.33,
      "grad_norm": 32.75,
      "learning_rate": 3.379550812532619e-05,
      "loss": 0.8895,
      "step": 371400
    },
    {
      "epoch": 0.33,
      "grad_norm": 24.375,
      "learning_rate": 3.3791009052135264e-05,
      "loss": 0.8625,
      "step": 371500
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.1875,
      "learning_rate": 3.378650997894434e-05,
      "loss": 1.0003,
      "step": 371600
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.875,
      "learning_rate": 3.3782010905753415e-05,
      "loss": 0.9104,
      "step": 371700
    },
    {
      "epoch": 0.33,
      "grad_norm": 41.75,
      "learning_rate": 3.377751183256249e-05,
      "loss": 1.0575,
      "step": 371800
    },
    {
      "epoch": 0.33,
      "grad_norm": 15.0,
      "learning_rate": 3.377301275937157e-05,
      "loss": 0.9208,
      "step": 371900
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.25,
      "learning_rate": 3.376851368618065e-05,
      "loss": 0.8582,
      "step": 372000
    },
    {
      "epoch": 0.33,
      "grad_norm": 18.875,
      "learning_rate": 3.376401461298972e-05,
      "loss": 0.9082,
      "step": 372100
    },
    {
      "epoch": 0.33,
      "grad_norm": 21.625,
      "learning_rate": 3.3759515539798805e-05,
      "loss": 1.0285,
      "step": 372200
    },
    {
      "epoch": 0.33,
      "grad_norm": 73.5,
      "learning_rate": 3.375501646660788e-05,
      "loss": 1.0365,
      "step": 372300
    },
    {
      "epoch": 0.33,
      "grad_norm": 47.5,
      "learning_rate": 3.3750517393416956e-05,
      "loss": 0.9553,
      "step": 372400
    },
    {
      "epoch": 0.33,
      "grad_norm": 27.125,
      "learning_rate": 3.374601832022604e-05,
      "loss": 1.1219,
      "step": 372500
    },
    {
      "epoch": 0.33,
      "grad_norm": 27.0,
      "learning_rate": 3.374151924703511e-05,
      "loss": 0.997,
      "step": 372600
    },
    {
      "epoch": 0.33,
      "grad_norm": 86.5,
      "learning_rate": 3.373702017384419e-05,
      "loss": 0.9543,
      "step": 372700
    },
    {
      "epoch": 0.33,
      "grad_norm": 25.5,
      "learning_rate": 3.373252110065327e-05,
      "loss": 0.9207,
      "step": 372800
    },
    {
      "epoch": 0.33,
      "grad_norm": 27.125,
      "learning_rate": 3.3728022027462346e-05,
      "loss": 1.085,
      "step": 372900
    },
    {
      "epoch": 0.33,
      "grad_norm": 99.5,
      "learning_rate": 3.372352295427142e-05,
      "loss": 1.0742,
      "step": 373000
    },
    {
      "epoch": 0.33,
      "grad_norm": 32.5,
      "learning_rate": 3.37190238810805e-05,
      "loss": 1.0516,
      "step": 373100
    },
    {
      "epoch": 0.33,
      "grad_norm": 29.5,
      "learning_rate": 3.371452480788957e-05,
      "loss": 1.0393,
      "step": 373200
    },
    {
      "epoch": 0.33,
      "grad_norm": 33.75,
      "learning_rate": 3.3710025734698654e-05,
      "loss": 0.8406,
      "step": 373300
    },
    {
      "epoch": 0.33,
      "grad_norm": 76.5,
      "learning_rate": 3.370552666150773e-05,
      "loss": 1.0643,
      "step": 373400
    },
    {
      "epoch": 0.33,
      "grad_norm": 31.875,
      "learning_rate": 3.3701027588316805e-05,
      "loss": 0.8349,
      "step": 373500
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.125,
      "learning_rate": 3.369652851512589e-05,
      "loss": 0.98,
      "step": 373600
    },
    {
      "epoch": 0.33,
      "grad_norm": 302.0,
      "learning_rate": 3.369202944193496e-05,
      "loss": 0.9419,
      "step": 373700
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.3125,
      "learning_rate": 3.3687530368744045e-05,
      "loss": 0.9734,
      "step": 373800
    },
    {
      "epoch": 0.33,
      "grad_norm": 59.0,
      "learning_rate": 3.368303129555312e-05,
      "loss": 0.9372,
      "step": 373900
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.9375,
      "learning_rate": 3.3678532222362195e-05,
      "loss": 0.9892,
      "step": 374000
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.90625,
      "learning_rate": 3.367403314917128e-05,
      "loss": 1.0475,
      "step": 374100
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.875,
      "learning_rate": 3.366953407598035e-05,
      "loss": 0.8914,
      "step": 374200
    },
    {
      "epoch": 0.33,
      "grad_norm": 75.0,
      "learning_rate": 3.366503500278942e-05,
      "loss": 0.9592,
      "step": 374300
    },
    {
      "epoch": 0.33,
      "grad_norm": 20.375,
      "learning_rate": 3.36605359295985e-05,
      "loss": 1.1901,
      "step": 374400
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.130859375,
      "learning_rate": 3.365603685640758e-05,
      "loss": 0.9409,
      "step": 374500
    },
    {
      "epoch": 0.33,
      "grad_norm": 57.5,
      "learning_rate": 3.365153778321666e-05,
      "loss": 1.0367,
      "step": 374600
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.96875,
      "learning_rate": 3.3647038710025736e-05,
      "loss": 0.9422,
      "step": 374700
    },
    {
      "epoch": 0.33,
      "grad_norm": 30.625,
      "learning_rate": 3.364253963683481e-05,
      "loss": 1.0069,
      "step": 374800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.330078125,
      "learning_rate": 3.3638040563643894e-05,
      "loss": 0.9289,
      "step": 374900
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.125,
      "learning_rate": 3.363354149045297e-05,
      "loss": 1.0581,
      "step": 375000
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.0625,
      "learning_rate": 3.3629042417262044e-05,
      "loss": 1.0849,
      "step": 375100
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.04248046875,
      "learning_rate": 3.3624543344071126e-05,
      "loss": 0.976,
      "step": 375200
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.640625,
      "learning_rate": 3.36200442708802e-05,
      "loss": 1.0362,
      "step": 375300
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.1875,
      "learning_rate": 3.361554519768928e-05,
      "loss": 1.0939,
      "step": 375400
    },
    {
      "epoch": 0.33,
      "grad_norm": 366.0,
      "learning_rate": 3.361104612449836e-05,
      "loss": 1.0964,
      "step": 375500
    },
    {
      "epoch": 0.33,
      "grad_norm": 86.5,
      "learning_rate": 3.360654705130743e-05,
      "loss": 0.9822,
      "step": 375600
    },
    {
      "epoch": 0.33,
      "grad_norm": 100.0,
      "learning_rate": 3.360204797811651e-05,
      "loss": 1.1474,
      "step": 375700
    },
    {
      "epoch": 0.33,
      "grad_norm": 48.0,
      "learning_rate": 3.3597548904925585e-05,
      "loss": 1.045,
      "step": 375800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.03466796875,
      "learning_rate": 3.359304983173466e-05,
      "loss": 1.0746,
      "step": 375900
    },
    {
      "epoch": 0.33,
      "grad_norm": 109.0,
      "learning_rate": 3.358855075854374e-05,
      "loss": 1.0199,
      "step": 376000
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.91796875,
      "learning_rate": 3.358405168535282e-05,
      "loss": 1.0031,
      "step": 376100
    },
    {
      "epoch": 0.34,
      "grad_norm": 28.5,
      "learning_rate": 3.357955261216189e-05,
      "loss": 0.8974,
      "step": 376200
    },
    {
      "epoch": 0.34,
      "grad_norm": 47.0,
      "learning_rate": 3.3575053538970975e-05,
      "loss": 1.1332,
      "step": 376300
    },
    {
      "epoch": 0.34,
      "grad_norm": 62.25,
      "learning_rate": 3.357055446578005e-05,
      "loss": 0.9769,
      "step": 376400
    },
    {
      "epoch": 0.34,
      "grad_norm": 39.0,
      "learning_rate": 3.356605539258913e-05,
      "loss": 1.076,
      "step": 376500
    },
    {
      "epoch": 0.34,
      "grad_norm": 39.5,
      "learning_rate": 3.356155631939821e-05,
      "loss": 1.109,
      "step": 376600
    },
    {
      "epoch": 0.34,
      "grad_norm": 43.5,
      "learning_rate": 3.3557057246207284e-05,
      "loss": 1.0835,
      "step": 376700
    },
    {
      "epoch": 0.34,
      "grad_norm": 141.0,
      "learning_rate": 3.3552558173016366e-05,
      "loss": 0.9525,
      "step": 376800
    },
    {
      "epoch": 0.34,
      "grad_norm": 35.5,
      "learning_rate": 3.3548059099825434e-05,
      "loss": 1.0132,
      "step": 376900
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.625,
      "learning_rate": 3.354356002663451e-05,
      "loss": 0.9435,
      "step": 377000
    },
    {
      "epoch": 0.34,
      "grad_norm": 39.75,
      "learning_rate": 3.353906095344359e-05,
      "loss": 0.9291,
      "step": 377100
    },
    {
      "epoch": 0.34,
      "grad_norm": 31.125,
      "learning_rate": 3.353456188025267e-05,
      "loss": 1.1039,
      "step": 377200
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.0223388671875,
      "learning_rate": 3.353006280706175e-05,
      "loss": 1.0395,
      "step": 377300
    },
    {
      "epoch": 0.34,
      "grad_norm": 29.75,
      "learning_rate": 3.3525563733870825e-05,
      "loss": 1.0247,
      "step": 377400
    },
    {
      "epoch": 0.34,
      "grad_norm": 69.0,
      "learning_rate": 3.35210646606799e-05,
      "loss": 0.9708,
      "step": 377500
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.09375,
      "learning_rate": 3.351656558748898e-05,
      "loss": 0.9882,
      "step": 377600
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.75,
      "learning_rate": 3.351206651429806e-05,
      "loss": 0.9816,
      "step": 377700
    },
    {
      "epoch": 0.34,
      "grad_norm": 26.375,
      "learning_rate": 3.350756744110713e-05,
      "loss": 0.9454,
      "step": 377800
    },
    {
      "epoch": 0.34,
      "grad_norm": 39.5,
      "learning_rate": 3.3503068367916215e-05,
      "loss": 1.0316,
      "step": 377900
    },
    {
      "epoch": 0.34,
      "grad_norm": 48.5,
      "learning_rate": 3.349856929472529e-05,
      "loss": 0.856,
      "step": 378000
    },
    {
      "epoch": 0.34,
      "grad_norm": 1320.0,
      "learning_rate": 3.3494070221534365e-05,
      "loss": 1.0977,
      "step": 378100
    },
    {
      "epoch": 0.34,
      "grad_norm": 119.5,
      "learning_rate": 3.348957114834344e-05,
      "loss": 0.9325,
      "step": 378200
    },
    {
      "epoch": 0.34,
      "grad_norm": 77.5,
      "learning_rate": 3.3485072075152516e-05,
      "loss": 1.28,
      "step": 378300
    },
    {
      "epoch": 0.34,
      "grad_norm": 28.625,
      "learning_rate": 3.34805730019616e-05,
      "loss": 0.9999,
      "step": 378400
    },
    {
      "epoch": 0.34,
      "grad_norm": 28.0,
      "learning_rate": 3.3476073928770674e-05,
      "loss": 0.9379,
      "step": 378500
    },
    {
      "epoch": 0.34,
      "grad_norm": 23.125,
      "learning_rate": 3.347157485557975e-05,
      "loss": 1.0134,
      "step": 378600
    },
    {
      "epoch": 0.34,
      "grad_norm": 252.0,
      "learning_rate": 3.346707578238883e-05,
      "loss": 0.9712,
      "step": 378700
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.375,
      "learning_rate": 3.3462576709197906e-05,
      "loss": 0.9244,
      "step": 378800
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3515625,
      "learning_rate": 3.345807763600698e-05,
      "loss": 0.9613,
      "step": 378900
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.375,
      "learning_rate": 3.3453578562816064e-05,
      "loss": 0.9041,
      "step": 379000
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.5,
      "learning_rate": 3.344907948962514e-05,
      "loss": 0.9524,
      "step": 379100
    },
    {
      "epoch": 0.34,
      "grad_norm": 39.0,
      "learning_rate": 3.344458041643422e-05,
      "loss": 1.017,
      "step": 379200
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.5,
      "learning_rate": 3.34400813432433e-05,
      "loss": 1.0178,
      "step": 379300
    },
    {
      "epoch": 0.34,
      "grad_norm": 84.5,
      "learning_rate": 3.343558227005237e-05,
      "loss": 1.0021,
      "step": 379400
    },
    {
      "epoch": 0.34,
      "grad_norm": 28.25,
      "learning_rate": 3.343108319686145e-05,
      "loss": 0.9867,
      "step": 379500
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8515625,
      "learning_rate": 3.342658412367052e-05,
      "loss": 1.0452,
      "step": 379600
    },
    {
      "epoch": 0.34,
      "grad_norm": 87.5,
      "learning_rate": 3.34220850504796e-05,
      "loss": 1.0257,
      "step": 379700
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.0201416015625,
      "learning_rate": 3.341758597728868e-05,
      "loss": 1.1778,
      "step": 379800
    },
    {
      "epoch": 0.34,
      "grad_norm": 81.0,
      "learning_rate": 3.3413086904097755e-05,
      "loss": 0.9022,
      "step": 379900
    },
    {
      "epoch": 0.34,
      "grad_norm": 22.25,
      "learning_rate": 3.340858783090684e-05,
      "loss": 1.0625,
      "step": 380000
    },
    {
      "epoch": 0.34,
      "grad_norm": 43.0,
      "learning_rate": 3.340408875771591e-05,
      "loss": 1.0564,
      "step": 380100
    },
    {
      "epoch": 0.34,
      "grad_norm": 42.0,
      "learning_rate": 3.339958968452499e-05,
      "loss": 1.0455,
      "step": 380200
    },
    {
      "epoch": 0.34,
      "grad_norm": 34.25,
      "learning_rate": 3.339509061133407e-05,
      "loss": 0.9761,
      "step": 380300
    },
    {
      "epoch": 0.34,
      "grad_norm": 28.25,
      "learning_rate": 3.3390591538143146e-05,
      "loss": 1.1074,
      "step": 380400
    },
    {
      "epoch": 0.34,
      "grad_norm": 49.75,
      "learning_rate": 3.338609246495222e-05,
      "loss": 0.9837,
      "step": 380500
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.0,
      "learning_rate": 3.33815933917613e-05,
      "loss": 0.7835,
      "step": 380600
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.4375,
      "learning_rate": 3.337709431857038e-05,
      "loss": 0.938,
      "step": 380700
    },
    {
      "epoch": 0.34,
      "grad_norm": 20.0,
      "learning_rate": 3.3372595245379454e-05,
      "loss": 0.8742,
      "step": 380800
    },
    {
      "epoch": 0.34,
      "grad_norm": 149.0,
      "learning_rate": 3.336809617218853e-05,
      "loss": 1.0751,
      "step": 380900
    },
    {
      "epoch": 0.34,
      "grad_norm": 34.75,
      "learning_rate": 3.3363597098997605e-05,
      "loss": 1.0227,
      "step": 381000
    },
    {
      "epoch": 0.34,
      "grad_norm": 110.0,
      "learning_rate": 3.335909802580669e-05,
      "loss": 1.0275,
      "step": 381100
    },
    {
      "epoch": 0.34,
      "grad_norm": 72.5,
      "learning_rate": 3.335459895261576e-05,
      "loss": 1.0093,
      "step": 381200
    },
    {
      "epoch": 0.34,
      "grad_norm": 30.25,
      "learning_rate": 3.335009987942484e-05,
      "loss": 1.0325,
      "step": 381300
    },
    {
      "epoch": 0.34,
      "grad_norm": 127.5,
      "learning_rate": 3.334560080623392e-05,
      "loss": 0.9621,
      "step": 381400
    },
    {
      "epoch": 0.34,
      "grad_norm": 36.0,
      "learning_rate": 3.3341101733042995e-05,
      "loss": 1.0984,
      "step": 381500
    },
    {
      "epoch": 0.34,
      "grad_norm": 37.75,
      "learning_rate": 3.333660265985207e-05,
      "loss": 1.1509,
      "step": 381600
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.625,
      "learning_rate": 3.333210358666115e-05,
      "loss": 0.9006,
      "step": 381700
    },
    {
      "epoch": 0.34,
      "grad_norm": 20.75,
      "learning_rate": 3.332760451347023e-05,
      "loss": 0.9851,
      "step": 381800
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.0115966796875,
      "learning_rate": 3.332310544027931e-05,
      "loss": 0.9302,
      "step": 381900
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.8125,
      "learning_rate": 3.3318606367088385e-05,
      "loss": 0.943,
      "step": 382000
    },
    {
      "epoch": 0.34,
      "grad_norm": 31.875,
      "learning_rate": 3.3314107293897454e-05,
      "loss": 1.1274,
      "step": 382100
    },
    {
      "epoch": 0.34,
      "grad_norm": 29.25,
      "learning_rate": 3.3309608220706536e-05,
      "loss": 0.9225,
      "step": 382200
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.0625,
      "learning_rate": 3.330510914751561e-05,
      "loss": 0.9638,
      "step": 382300
    },
    {
      "epoch": 0.34,
      "grad_norm": 22.375,
      "learning_rate": 3.3300610074324686e-05,
      "loss": 0.8293,
      "step": 382400
    },
    {
      "epoch": 0.34,
      "grad_norm": 94.5,
      "learning_rate": 3.329611100113377e-05,
      "loss": 0.8677,
      "step": 382500
    },
    {
      "epoch": 0.34,
      "grad_norm": 44.25,
      "learning_rate": 3.3291611927942844e-05,
      "loss": 1.037,
      "step": 382600
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.875,
      "learning_rate": 3.3287112854751926e-05,
      "loss": 0.9874,
      "step": 382700
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.875,
      "learning_rate": 3.3282613781561e-05,
      "loss": 1.1759,
      "step": 382800
    },
    {
      "epoch": 0.34,
      "grad_norm": 40.5,
      "learning_rate": 3.327811470837008e-05,
      "loss": 1.024,
      "step": 382900
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.1875,
      "learning_rate": 3.327361563517916e-05,
      "loss": 1.0623,
      "step": 383000
    },
    {
      "epoch": 0.34,
      "grad_norm": 29.25,
      "learning_rate": 3.3269116561988234e-05,
      "loss": 0.8774,
      "step": 383100
    },
    {
      "epoch": 0.34,
      "grad_norm": 195.0,
      "learning_rate": 3.326461748879731e-05,
      "loss": 0.983,
      "step": 383200
    },
    {
      "epoch": 0.34,
      "grad_norm": 100.0,
      "learning_rate": 3.326011841560639e-05,
      "loss": 1.0287,
      "step": 383300
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.6875,
      "learning_rate": 3.325561934241546e-05,
      "loss": 0.9744,
      "step": 383400
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.125,
      "learning_rate": 3.325112026922454e-05,
      "loss": 0.9021,
      "step": 383500
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.75,
      "learning_rate": 3.324662119603362e-05,
      "loss": 1.023,
      "step": 383600
    },
    {
      "epoch": 0.34,
      "grad_norm": 24.875,
      "learning_rate": 3.324212212284269e-05,
      "loss": 0.9416,
      "step": 383700
    },
    {
      "epoch": 0.34,
      "grad_norm": 42.75,
      "learning_rate": 3.3237623049651775e-05,
      "loss": 0.9394,
      "step": 383800
    },
    {
      "epoch": 0.34,
      "grad_norm": 22.125,
      "learning_rate": 3.323312397646085e-05,
      "loss": 0.9702,
      "step": 383900
    },
    {
      "epoch": 0.34,
      "grad_norm": 33.0,
      "learning_rate": 3.3228624903269926e-05,
      "loss": 0.8742,
      "step": 384000
    },
    {
      "epoch": 0.34,
      "grad_norm": 26.125,
      "learning_rate": 3.322412583007901e-05,
      "loss": 1.0123,
      "step": 384100
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.0,
      "learning_rate": 3.321962675688808e-05,
      "loss": 0.936,
      "step": 384200
    },
    {
      "epoch": 0.34,
      "grad_norm": 42.25,
      "learning_rate": 3.321512768369716e-05,
      "loss": 0.9998,
      "step": 384300
    },
    {
      "epoch": 0.34,
      "grad_norm": 25.25,
      "learning_rate": 3.321062861050624e-05,
      "loss": 1.0996,
      "step": 384400
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.9375,
      "learning_rate": 3.3206129537315316e-05,
      "loss": 1.1348,
      "step": 384500
    },
    {
      "epoch": 0.34,
      "grad_norm": 23.5,
      "learning_rate": 3.32016304641244e-05,
      "loss": 1.0157,
      "step": 384600
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.375,
      "learning_rate": 3.319713139093347e-05,
      "loss": 1.0609,
      "step": 384700
    },
    {
      "epoch": 0.34,
      "grad_norm": 20.75,
      "learning_rate": 3.319263231774254e-05,
      "loss": 1.0987,
      "step": 384800
    },
    {
      "epoch": 0.34,
      "grad_norm": 40.5,
      "learning_rate": 3.3188133244551624e-05,
      "loss": 0.9281,
      "step": 384900
    },
    {
      "epoch": 0.34,
      "grad_norm": 34.75,
      "learning_rate": 3.31836341713607e-05,
      "loss": 1.0413,
      "step": 385000
    },
    {
      "epoch": 0.34,
      "grad_norm": 36.75,
      "learning_rate": 3.3179135098169775e-05,
      "loss": 0.9547,
      "step": 385100
    },
    {
      "epoch": 0.34,
      "grad_norm": 26.75,
      "learning_rate": 3.317463602497886e-05,
      "loss": 1.0339,
      "step": 385200
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.625,
      "learning_rate": 3.317013695178793e-05,
      "loss": 1.0669,
      "step": 385300
    },
    {
      "epoch": 0.34,
      "grad_norm": 73.5,
      "learning_rate": 3.3165637878597014e-05,
      "loss": 1.0708,
      "step": 385400
    },
    {
      "epoch": 0.34,
      "grad_norm": 66.5,
      "learning_rate": 3.316113880540609e-05,
      "loss": 1.0079,
      "step": 385500
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.080078125,
      "learning_rate": 3.3156639732215165e-05,
      "loss": 0.9695,
      "step": 385600
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.25,
      "learning_rate": 3.315214065902425e-05,
      "loss": 1.1168,
      "step": 385700
    },
    {
      "epoch": 0.34,
      "grad_norm": 24.125,
      "learning_rate": 3.314764158583332e-05,
      "loss": 1.1013,
      "step": 385800
    },
    {
      "epoch": 0.34,
      "grad_norm": 22.875,
      "learning_rate": 3.31431425126424e-05,
      "loss": 1.1192,
      "step": 385900
    },
    {
      "epoch": 0.34,
      "grad_norm": 21.875,
      "learning_rate": 3.313864343945147e-05,
      "loss": 1.0347,
      "step": 386000
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.4375,
      "learning_rate": 3.313414436626055e-05,
      "loss": 1.0131,
      "step": 386100
    },
    {
      "epoch": 0.34,
      "grad_norm": 104.5,
      "learning_rate": 3.312964529306963e-05,
      "loss": 1.032,
      "step": 386200
    },
    {
      "epoch": 0.34,
      "grad_norm": 18.5,
      "learning_rate": 3.3125146219878706e-05,
      "loss": 1.1368,
      "step": 386300
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.0,
      "learning_rate": 3.312064714668778e-05,
      "loss": 0.9886,
      "step": 386400
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1748046875,
      "learning_rate": 3.3116148073496863e-05,
      "loss": 0.8901,
      "step": 386500
    },
    {
      "epoch": 0.34,
      "grad_norm": 23.875,
      "learning_rate": 3.311164900030594e-05,
      "loss": 1.107,
      "step": 386600
    },
    {
      "epoch": 0.34,
      "grad_norm": 24.375,
      "learning_rate": 3.3107149927115014e-05,
      "loss": 1.0496,
      "step": 386700
    },
    {
      "epoch": 0.34,
      "grad_norm": 20.375,
      "learning_rate": 3.3102650853924096e-05,
      "loss": 1.1145,
      "step": 386800
    },
    {
      "epoch": 0.34,
      "grad_norm": 73.5,
      "learning_rate": 3.309815178073317e-05,
      "loss": 0.9504,
      "step": 386900
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.060791015625,
      "learning_rate": 3.309365270754225e-05,
      "loss": 0.8844,
      "step": 387000
    },
    {
      "epoch": 0.34,
      "grad_norm": 124.0,
      "learning_rate": 3.308915363435133e-05,
      "loss": 1.0265,
      "step": 387100
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.5625,
      "learning_rate": 3.3084654561160404e-05,
      "loss": 1.0152,
      "step": 387200
    },
    {
      "epoch": 0.35,
      "grad_norm": 27.25,
      "learning_rate": 3.308015548796948e-05,
      "loss": 1.0995,
      "step": 387300
    },
    {
      "epoch": 0.35,
      "grad_norm": 73.0,
      "learning_rate": 3.3075656414778555e-05,
      "loss": 0.9643,
      "step": 387400
    },
    {
      "epoch": 0.35,
      "grad_norm": 23.625,
      "learning_rate": 3.307115734158763e-05,
      "loss": 0.9915,
      "step": 387500
    },
    {
      "epoch": 0.35,
      "grad_norm": 26.625,
      "learning_rate": 3.306665826839671e-05,
      "loss": 0.8849,
      "step": 387600
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.263671875,
      "learning_rate": 3.306215919520579e-05,
      "loss": 0.9846,
      "step": 387700
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.125,
      "learning_rate": 3.305766012201486e-05,
      "loss": 1.0451,
      "step": 387800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10791015625,
      "learning_rate": 3.3053161048823945e-05,
      "loss": 1.0468,
      "step": 387900
    },
    {
      "epoch": 0.35,
      "grad_norm": 21.0,
      "learning_rate": 3.304866197563302e-05,
      "loss": 1.0574,
      "step": 388000
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.3125,
      "learning_rate": 3.30441629024421e-05,
      "loss": 1.0069,
      "step": 388100
    },
    {
      "epoch": 0.35,
      "grad_norm": 27.0,
      "learning_rate": 3.303966382925118e-05,
      "loss": 1.1069,
      "step": 388200
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.5,
      "learning_rate": 3.3035164756060253e-05,
      "loss": 0.8954,
      "step": 388300
    },
    {
      "epoch": 0.35,
      "grad_norm": 20.125,
      "learning_rate": 3.3030665682869336e-05,
      "loss": 1.0044,
      "step": 388400
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.55078125,
      "learning_rate": 3.302616660967841e-05,
      "loss": 0.9934,
      "step": 388500
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.25,
      "learning_rate": 3.302166753648748e-05,
      "loss": 0.9737,
      "step": 388600
    },
    {
      "epoch": 0.35,
      "grad_norm": 19.375,
      "learning_rate": 3.301716846329656e-05,
      "loss": 0.9581,
      "step": 388700
    },
    {
      "epoch": 0.35,
      "grad_norm": 95.0,
      "learning_rate": 3.301266939010564e-05,
      "loss": 1.0132,
      "step": 388800
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.546875,
      "learning_rate": 3.300817031691472e-05,
      "loss": 0.9096,
      "step": 388900
    },
    {
      "epoch": 0.35,
      "grad_norm": 21.5,
      "learning_rate": 3.3003671243723794e-05,
      "loss": 1.0295,
      "step": 389000
    },
    {
      "epoch": 0.35,
      "grad_norm": 21.5,
      "learning_rate": 3.299917217053287e-05,
      "loss": 0.9456,
      "step": 389100
    },
    {
      "epoch": 0.35,
      "grad_norm": 30.5,
      "learning_rate": 3.299467309734195e-05,
      "loss": 1.064,
      "step": 389200
    },
    {
      "epoch": 0.35,
      "grad_norm": 28.75,
      "learning_rate": 3.299017402415103e-05,
      "loss": 0.775,
      "step": 389300
    },
    {
      "epoch": 0.35,
      "grad_norm": 21.0,
      "learning_rate": 3.29856749509601e-05,
      "loss": 0.9785,
      "step": 389400
    },
    {
      "epoch": 0.35,
      "grad_norm": 22.5,
      "learning_rate": 3.2981175877769185e-05,
      "loss": 1.0503,
      "step": 389500
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.0849609375,
      "learning_rate": 3.297667680457826e-05,
      "loss": 0.915,
      "step": 389600
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.5,
      "learning_rate": 3.2972177731387335e-05,
      "loss": 1.0422,
      "step": 389700
    },
    {
      "epoch": 0.35,
      "grad_norm": 99.5,
      "learning_rate": 3.296767865819642e-05,
      "loss": 0.9847,
      "step": 389800
    },
    {
      "epoch": 0.35,
      "grad_norm": 60.25,
      "learning_rate": 3.2963179585005486e-05,
      "loss": 1.0998,
      "step": 389900
    },
    {
      "epoch": 0.35,
      "grad_norm": 18.375,
      "learning_rate": 3.295868051181457e-05,
      "loss": 0.9966,
      "step": 390000
    }
  ],
  "logging_steps": 100,
  "max_steps": 1122566,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 6000,
  "total_flos": 6.141852751798272e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}