{
  "best_metric": 0.98815,
  "best_model_checkpoint": "/media/hongss/ssd/hatespeech_modelbackup/new_weights/1216/KoElectra_123/stage1/20241217T17-29-02/checkpoint-275000",
  "epoch": 27.5,
  "eval_steps": 5000,
  "global_step": 275000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 0.8010932207107544,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.6941,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8222021460533142,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.6895,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.551395297050476,
      "learning_rate": 1.5e-06,
      "loss": 0.6726,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8527400493621826,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.5988,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.536347389221191,
      "learning_rate": 2.5e-06,
      "loss": 0.4735,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1166355609893799,
      "learning_rate": 3e-06,
      "loss": 0.3591,
      "step": 600
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7112144231796265,
      "learning_rate": 3.5e-06,
      "loss": 0.2803,
      "step": 700
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.188007354736328,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.2336,
      "step": 800
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2752089500427246,
      "learning_rate": 4.5e-06,
      "loss": 0.1956,
      "step": 900
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.095466613769531,
      "learning_rate": 5e-06,
      "loss": 0.1955,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8912861347198486,
      "learning_rate": 4.998327759197325e-06,
      "loss": 0.1589,
      "step": 1100
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.762054443359375,
      "learning_rate": 4.996655518394649e-06,
      "loss": 0.1708,
      "step": 1200
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42863136529922485,
      "learning_rate": 4.994983277591973e-06,
      "loss": 0.1266,
      "step": 1300
    },
    {
      "epoch": 0.14,
      "grad_norm": 27.861000061035156,
      "learning_rate": 4.993311036789298e-06,
      "loss": 0.1403,
      "step": 1400
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.980772018432617,
      "learning_rate": 4.991638795986622e-06,
      "loss": 0.1378,
      "step": 1500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17961889505386353,
      "learning_rate": 4.989966555183947e-06,
      "loss": 0.1168,
      "step": 1600
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3810911178588867,
      "learning_rate": 4.988294314381271e-06,
      "loss": 0.1129,
      "step": 1700
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.552303314208984,
      "learning_rate": 4.986622073578596e-06,
      "loss": 0.1584,
      "step": 1800
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4360831081867218,
      "learning_rate": 4.98494983277592e-06,
      "loss": 0.1267,
      "step": 1900
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.449009656906128,
      "learning_rate": 4.983277591973244e-06,
      "loss": 0.1199,
      "step": 2000
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0524966716766357,
      "learning_rate": 4.981605351170569e-06,
      "loss": 0.1293,
      "step": 2100
    },
    {
      "epoch": 0.22,
      "grad_norm": 25.5817813873291,
      "learning_rate": 4.979933110367893e-06,
      "loss": 0.0851,
      "step": 2200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.16489553451538086,
      "learning_rate": 4.978260869565218e-06,
      "loss": 0.1246,
      "step": 2300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1531640738248825,
      "learning_rate": 4.976588628762542e-06,
      "loss": 0.1167,
      "step": 2400
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.05634117126465,
      "learning_rate": 4.974916387959867e-06,
      "loss": 0.1099,
      "step": 2500
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.16108916699886322,
      "learning_rate": 4.973244147157191e-06,
      "loss": 0.1278,
      "step": 2600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17259421944618225,
      "learning_rate": 4.971571906354515e-06,
      "loss": 0.0692,
      "step": 2700
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.081533432006836,
      "learning_rate": 4.96989966555184e-06,
      "loss": 0.1063,
      "step": 2800
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4863479137420654,
      "learning_rate": 4.968227424749164e-06,
      "loss": 0.1163,
      "step": 2900
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11070814728736877,
      "learning_rate": 4.966555183946489e-06,
      "loss": 0.0959,
      "step": 3000
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8638683557510376,
      "learning_rate": 4.964882943143814e-06,
      "loss": 0.084,
      "step": 3100
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3781122863292694,
      "learning_rate": 4.963210702341138e-06,
      "loss": 0.086,
      "step": 3200
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.08686799556016922,
      "learning_rate": 4.961538461538462e-06,
      "loss": 0.0905,
      "step": 3300
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.08633287996053696,
      "learning_rate": 4.959866220735786e-06,
      "loss": 0.0843,
      "step": 3400
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.28560230135917664,
      "learning_rate": 4.958193979933111e-06,
      "loss": 0.0929,
      "step": 3500
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.07190747559070587,
      "learning_rate": 4.9565217391304355e-06,
      "loss": 0.0889,
      "step": 3600
    },
    {
      "epoch": 0.37,
      "grad_norm": 25.537450790405273,
      "learning_rate": 4.9548494983277596e-06,
      "loss": 0.0946,
      "step": 3700
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.443319320678711,
      "learning_rate": 4.9531772575250845e-06,
      "loss": 0.1029,
      "step": 3800
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.08515575528144836,
      "learning_rate": 4.9515050167224085e-06,
      "loss": 0.0964,
      "step": 3900
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.10333725810050964,
      "learning_rate": 4.9498327759197325e-06,
      "loss": 0.0977,
      "step": 4000
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.041944719851017,
      "learning_rate": 4.948160535117057e-06,
      "loss": 0.0922,
      "step": 4100
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.2756776809692383,
      "learning_rate": 4.9464882943143815e-06,
      "loss": 0.0675,
      "step": 4200
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.24823522567749,
      "learning_rate": 4.944816053511706e-06,
      "loss": 0.0869,
      "step": 4300
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.09910300374031067,
      "learning_rate": 4.94314381270903e-06,
      "loss": 0.1091,
      "step": 4400
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7329888939857483,
      "learning_rate": 4.941471571906355e-06,
      "loss": 0.1076,
      "step": 4500
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.031429223716259,
      "learning_rate": 4.939799331103679e-06,
      "loss": 0.1021,
      "step": 4600
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.47189056873321533,
      "learning_rate": 4.938127090301003e-06,
      "loss": 0.0815,
      "step": 4700
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.491260051727295,
      "learning_rate": 4.936454849498328e-06,
      "loss": 0.1312,
      "step": 4800
    },
    {
      "epoch": 0.49,
      "grad_norm": 42.855648040771484,
      "learning_rate": 4.934782608695652e-06,
      "loss": 0.0784,
      "step": 4900
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.520151138305664,
      "learning_rate": 4.933110367892977e-06,
      "loss": 0.0815,
      "step": 5000
    },
    {
      "epoch": 0.5,
      "eval_accuracy": 0.979275,
      "eval_f1": 0.979275,
      "eval_loss": 0.07820390909910202,
      "eval_runtime": 134.6565,
      "eval_samples_per_second": 297.052,
      "eval_steps_per_second": 297.052,
      "step": 5000
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.115912437438965,
      "learning_rate": 4.931438127090302e-06,
      "loss": 0.1083,
      "step": 5100
    },
    {
      "epoch": 0.52,
      "grad_norm": 18.950531005859375,
      "learning_rate": 4.929765886287626e-06,
      "loss": 0.0754,
      "step": 5200
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.024242494255304337,
      "learning_rate": 4.92809364548495e-06,
      "loss": 0.0787,
      "step": 5300
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.753853797912598,
      "learning_rate": 4.926421404682274e-06,
      "loss": 0.0824,
      "step": 5400
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.40188217163086,
      "learning_rate": 4.924749163879599e-06,
      "loss": 0.0802,
      "step": 5500
    },
    {
      "epoch": 0.56,
      "grad_norm": 17.343854904174805,
      "learning_rate": 4.923076923076924e-06,
      "loss": 0.0784,
      "step": 5600
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23710471391677856,
      "learning_rate": 4.921404682274248e-06,
      "loss": 0.0989,
      "step": 5700
    },
    {
      "epoch": 0.58,
      "grad_norm": 108.25338745117188,
      "learning_rate": 4.919732441471573e-06,
      "loss": 0.0836,
      "step": 5800
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.040300995111465454,
      "learning_rate": 4.918060200668897e-06,
      "loss": 0.0674,
      "step": 5900
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.05142529681324959,
      "learning_rate": 4.916387959866221e-06,
      "loss": 0.0854,
      "step": 6000
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2920592427253723,
      "learning_rate": 4.914715719063545e-06,
      "loss": 0.0603,
      "step": 6100
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.042372226715088,
      "learning_rate": 4.91304347826087e-06,
      "loss": 0.0918,
      "step": 6200
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.157938241958618,
      "learning_rate": 4.911371237458195e-06,
      "loss": 0.0704,
      "step": 6300
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40491366386413574,
      "learning_rate": 4.909698996655519e-06,
      "loss": 0.0679,
      "step": 6400
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.10786664485931396,
      "learning_rate": 4.908026755852844e-06,
      "loss": 0.0915,
      "step": 6500
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.927928924560547,
      "learning_rate": 4.906354515050168e-06,
      "loss": 0.0622,
      "step": 6600
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.019695168361067772,
      "learning_rate": 4.904682274247492e-06,
      "loss": 0.0755,
      "step": 6700
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.06486581265926361,
      "learning_rate": 4.9030100334448166e-06,
      "loss": 0.0558,
      "step": 6800
    },
    {
      "epoch": 0.69,
      "grad_norm": 16.396251678466797,
      "learning_rate": 4.901337792642141e-06,
      "loss": 0.0648,
      "step": 6900
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3079834580421448,
      "learning_rate": 4.8996655518394655e-06,
      "loss": 0.0842,
      "step": 7000
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2089134454727173,
      "learning_rate": 4.8979933110367895e-06,
      "loss": 0.0754,
      "step": 7100
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.028676798567175865,
      "learning_rate": 4.896321070234114e-06,
      "loss": 0.0802,
      "step": 7200
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.04169798642396927,
      "learning_rate": 4.8946488294314385e-06,
      "loss": 0.0654,
      "step": 7300
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.176326751708984,
      "learning_rate": 4.8929765886287625e-06,
      "loss": 0.0747,
      "step": 7400
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2586785852909088,
      "learning_rate": 4.891304347826087e-06,
      "loss": 0.0716,
      "step": 7500
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4150737226009369,
      "learning_rate": 4.889632107023411e-06,
      "loss": 0.0481,
      "step": 7600
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.011774115264415741,
      "learning_rate": 4.887959866220736e-06,
      "loss": 0.0552,
      "step": 7700
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.0569203644990921,
      "learning_rate": 4.886287625418061e-06,
      "loss": 0.0598,
      "step": 7800
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.26318359375,
      "learning_rate": 4.884615384615385e-06,
      "loss": 0.0769,
      "step": 7900
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.13142751157283783,
      "learning_rate": 4.882943143812709e-06,
      "loss": 0.0833,
      "step": 8000
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16441769897937775,
      "learning_rate": 4.881270903010033e-06,
      "loss": 0.0882,
      "step": 8100
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3407476544380188,
      "learning_rate": 4.879598662207358e-06,
      "loss": 0.0675,
      "step": 8200
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1758737415075302,
      "learning_rate": 4.877926421404683e-06,
      "loss": 0.0703,
      "step": 8300
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.0441744327545166,
      "learning_rate": 4.876254180602007e-06,
      "loss": 0.0617,
      "step": 8400
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.028383277356624603,
      "learning_rate": 4.874581939799332e-06,
      "loss": 0.0633,
      "step": 8500
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.03873937577009201,
      "learning_rate": 4.872909698996656e-06,
      "loss": 0.0737,
      "step": 8600
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.0403941310942173,
      "learning_rate": 4.87123745819398e-06,
      "loss": 0.0633,
      "step": 8700
    },
    {
      "epoch": 0.88,
      "grad_norm": 35.07930374145508,
      "learning_rate": 4.869565217391305e-06,
      "loss": 0.0547,
      "step": 8800
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.07603854686021805,
      "learning_rate": 4.867892976588629e-06,
      "loss": 0.066,
      "step": 8900
    },
    {
      "epoch": 0.9,
      "grad_norm": 19.292327880859375,
      "learning_rate": 4.866220735785954e-06,
      "loss": 0.0535,
      "step": 9000
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.07600151747465134,
      "learning_rate": 4.864548494983278e-06,
      "loss": 0.0829,
      "step": 9100
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5552844405174255,
      "learning_rate": 4.862876254180603e-06,
      "loss": 0.067,
      "step": 9200
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.020827770233154297,
      "learning_rate": 4.861204013377927e-06,
      "loss": 0.0563,
      "step": 9300
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.10981284081935883,
      "learning_rate": 4.859531772575251e-06,
      "loss": 0.0773,
      "step": 9400
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.03421054407954216,
      "learning_rate": 4.857859531772576e-06,
      "loss": 0.0646,
      "step": 9500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17162667214870453,
      "learning_rate": 4.8561872909699e-06,
      "loss": 0.0518,
      "step": 9600
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.10754731297492981,
      "learning_rate": 4.854515050167225e-06,
      "loss": 0.0709,
      "step": 9700
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.033437736332416534,
      "learning_rate": 4.852842809364549e-06,
      "loss": 0.0629,
      "step": 9800
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.021083764731884003,
      "learning_rate": 4.851170568561874e-06,
      "loss": 0.0757,
      "step": 9900
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.52938985824585,
      "learning_rate": 4.849498327759198e-06,
      "loss": 0.0585,
      "step": 10000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.983525,
      "eval_f1": 0.983525,
      "eval_loss": 0.07067874819040298,
      "eval_runtime": 133.7009,
      "eval_samples_per_second": 299.175,
      "eval_steps_per_second": 299.175,
      "step": 10000
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.08288461714982986,
      "learning_rate": 4.847826086956522e-06,
      "loss": 0.0384,
      "step": 10100
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.030511723831295967,
      "learning_rate": 4.8461538461538465e-06,
      "loss": 0.0564,
      "step": 10200
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.006339346989989281,
      "learning_rate": 4.844481605351171e-06,
      "loss": 0.049,
      "step": 10300
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.46108341217041,
      "learning_rate": 4.8428093645484955e-06,
      "loss": 0.0649,
      "step": 10400
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3494547605514526,
      "learning_rate": 4.8411371237458195e-06,
      "loss": 0.0512,
      "step": 10500
    },
    {
      "epoch": 1.06,
      "grad_norm": 3.752967357635498,
      "learning_rate": 4.839464882943144e-06,
      "loss": 0.0527,
      "step": 10600
    },
    {
      "epoch": 1.07,
      "grad_norm": 20.542871475219727,
      "learning_rate": 4.837792642140468e-06,
      "loss": 0.0497,
      "step": 10700
    },
    {
      "epoch": 1.08,
      "grad_norm": 24.39754867553711,
      "learning_rate": 4.836120401337793e-06,
      "loss": 0.0653,
      "step": 10800
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9044870138168335,
      "learning_rate": 4.834448160535117e-06,
      "loss": 0.0375,
      "step": 10900
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.592361450195312,
      "learning_rate": 4.832775919732442e-06,
      "loss": 0.0429,
      "step": 11000
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.018932342529296875,
      "learning_rate": 4.831103678929766e-06,
      "loss": 0.0323,
      "step": 11100
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.08527997881174088,
      "learning_rate": 4.82943143812709e-06,
      "loss": 0.0803,
      "step": 11200
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.010842899791896343,
      "learning_rate": 4.827759197324415e-06,
      "loss": 0.0389,
      "step": 11300
    },
    {
      "epoch": 1.1400000000000001,
      "grad_norm": 0.025464480742812157,
      "learning_rate": 4.826086956521739e-06,
      "loss": 0.0477,
      "step": 11400
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.29770731925964355,
      "learning_rate": 4.824414715719064e-06,
      "loss": 0.0651,
      "step": 11500
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.16732554137706757,
      "learning_rate": 4.822742474916388e-06,
      "loss": 0.0363,
      "step": 11600
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.036259859800338745,
      "learning_rate": 4.821070234113713e-06,
      "loss": 0.0531,
      "step": 11700
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.02447950281202793,
      "learning_rate": 4.819397993311037e-06,
      "loss": 0.0451,
      "step": 11800
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.15884707868099213,
      "learning_rate": 4.817725752508361e-06,
      "loss": 0.0597,
      "step": 11900
    },
    {
      "epoch": 1.2,
      "grad_norm": 12.805810928344727,
      "learning_rate": 4.816053511705686e-06,
      "loss": 0.0455,
      "step": 12000
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.025491496548056602,
      "learning_rate": 4.81438127090301e-06,
      "loss": 0.0615,
      "step": 12100
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6481452584266663,
      "learning_rate": 4.812709030100335e-06,
      "loss": 0.0606,
      "step": 12200
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.02471865341067314,
      "learning_rate": 4.81103678929766e-06,
      "loss": 0.0505,
      "step": 12300
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.005726488307118416,
      "learning_rate": 4.809364548494984e-06,
      "loss": 0.0484,
      "step": 12400
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.004293044563382864,
      "learning_rate": 4.807692307692308e-06,
      "loss": 0.0581,
      "step": 12500
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.12100236862897873,
      "learning_rate": 4.806020066889632e-06,
      "loss": 0.0616,
      "step": 12600
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.4037141501903534,
      "learning_rate": 4.804347826086957e-06,
      "loss": 0.0445,
      "step": 12700
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.907632827758789,
      "learning_rate": 4.802675585284282e-06,
      "loss": 0.0482,
      "step": 12800
    },
    {
      "epoch": 1.29,
      "grad_norm": 11.875676155090332,
      "learning_rate": 4.801003344481606e-06,
      "loss": 0.0406,
      "step": 12900
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1897820234298706,
      "learning_rate": 4.799331103678931e-06,
      "loss": 0.0603,
      "step": 13000
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.04131534695625305,
      "learning_rate": 4.797658862876255e-06,
      "loss": 0.0395,
      "step": 13100
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.11377372592687607,
      "learning_rate": 4.795986622073579e-06,
      "loss": 0.059,
      "step": 13200
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.017639851197600365,
      "learning_rate": 4.7943143812709035e-06,
      "loss": 0.0611,
      "step": 13300
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.03218785300850868,
      "learning_rate": 4.792642140468228e-06,
      "loss": 0.0502,
      "step": 13400
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.049068257212638855,
      "learning_rate": 4.7909698996655525e-06,
      "loss": 0.0645,
      "step": 13500
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.014554962515830994,
      "learning_rate": 4.7892976588628765e-06,
      "loss": 0.0446,
      "step": 13600
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.10987705737352371,
      "learning_rate": 4.787625418060201e-06,
      "loss": 0.0512,
      "step": 13700
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.821701526641846,
      "learning_rate": 4.785953177257525e-06,
      "loss": 0.0552,
      "step": 13800
    },
    {
      "epoch": 1.3900000000000001,
      "grad_norm": 0.0972401350736618,
      "learning_rate": 4.7842809364548495e-06,
      "loss": 0.0482,
      "step": 13900
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.009446914307773113,
      "learning_rate": 4.782608695652174e-06,
      "loss": 0.0332,
      "step": 14000
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3108714818954468,
      "learning_rate": 4.780936454849498e-06,
      "loss": 0.0899,
      "step": 14100
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.04676111787557602,
      "learning_rate": 4.779264214046823e-06,
      "loss": 0.0494,
      "step": 14200
    },
    {
      "epoch": 1.43,
      "grad_norm": 15.421358108520508,
      "learning_rate": 4.777591973244148e-06,
      "loss": 0.0521,
      "step": 14300
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.015584264881908894,
      "learning_rate": 4.775919732441472e-06,
      "loss": 0.0453,
      "step": 14400
    },
    {
      "epoch": 1.45,
      "grad_norm": 3.9511654376983643,
      "learning_rate": 4.774247491638796e-06,
      "loss": 0.045,
      "step": 14500
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.09137571603059769,
      "learning_rate": 4.77257525083612e-06,
      "loss": 0.0607,
      "step": 14600
    },
    {
      "epoch": 1.47,
      "grad_norm": 19.993337631225586,
      "learning_rate": 4.770903010033445e-06,
      "loss": 0.043,
      "step": 14700
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.03419216349720955,
      "learning_rate": 4.76923076923077e-06,
      "loss": 0.0561,
      "step": 14800
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.017296940088272095,
      "learning_rate": 4.767558528428094e-06,
      "loss": 0.0296,
      "step": 14900
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.5799407958984375,
      "learning_rate": 4.765886287625419e-06,
      "loss": 0.044,
      "step": 15000
    },
    {
      "epoch": 1.5,
      "eval_accuracy": 0.983825,
      "eval_f1": 0.983825,
      "eval_loss": 0.07533564418554306,
      "eval_runtime": 133.0916,
      "eval_samples_per_second": 300.545,
      "eval_steps_per_second": 300.545,
      "step": 15000
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.01647213287651539,
      "learning_rate": 4.764214046822743e-06,
      "loss": 0.0514,
      "step": 15100
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.01267435122281313,
      "learning_rate": 4.762541806020067e-06,
      "loss": 0.0448,
      "step": 15200
    },
    {
      "epoch": 1.53,
      "grad_norm": 60.150901794433594,
      "learning_rate": 4.760869565217392e-06,
      "loss": 0.0625,
      "step": 15300
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.058160994201898575,
      "learning_rate": 4.759197324414716e-06,
      "loss": 0.0493,
      "step": 15400
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.02377147413790226,
      "learning_rate": 4.757525083612041e-06,
      "loss": 0.053,
      "step": 15500
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.02492053434252739,
      "learning_rate": 4.755852842809365e-06,
      "loss": 0.0499,
      "step": 15600
    },
    {
      "epoch": 1.5699999999999998,
      "grad_norm": 0.069520004093647,
      "learning_rate": 4.75418060200669e-06,
      "loss": 0.0475,
      "step": 15700
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.18645954132080078,
      "learning_rate": 4.752508361204014e-06,
      "loss": 0.0528,
      "step": 15800
    },
    {
      "epoch": 1.5899999999999999,
      "grad_norm": 0.7005010843276978,
      "learning_rate": 4.750836120401338e-06,
      "loss": 0.0554,
      "step": 15900
    },
    {
      "epoch": 1.6,
      "grad_norm": 11.44532299041748,
      "learning_rate": 4.749163879598663e-06,
      "loss": 0.0453,
      "step": 16000
    },
    {
      "epoch": 1.6099999999999999,
      "grad_norm": 0.028075769543647766,
      "learning_rate": 4.747491638795987e-06,
      "loss": 0.0454,
      "step": 16100
    },
    {
      "epoch": 1.62,
      "grad_norm": 11.789948463439941,
      "learning_rate": 4.745819397993312e-06,
      "loss": 0.0456,
      "step": 16200
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.11324401199817657,
      "learning_rate": 4.744147157190636e-06,
      "loss": 0.0549,
      "step": 16300
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 49.08498001098633,
      "learning_rate": 4.7424749163879605e-06,
      "loss": 0.0346,
      "step": 16400
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.05506042763590813,
      "learning_rate": 4.740802675585285e-06,
      "loss": 0.0629,
      "step": 16500
    },
    {
      "epoch": 1.6600000000000001,
      "grad_norm": 0.012054430320858955,
      "learning_rate": 4.739130434782609e-06,
      "loss": 0.0246,
      "step": 16600
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.14489546418190002,
      "learning_rate": 4.7374581939799335e-06,
      "loss": 0.0803,
      "step": 16700
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 12.88791275024414,
      "learning_rate": 4.7357859531772575e-06,
      "loss": 0.0484,
      "step": 16800
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.472039222717285,
      "learning_rate": 4.7341137123745824e-06,
      "loss": 0.0505,
      "step": 16900
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.09024617075920105,
      "learning_rate": 4.732441471571907e-06,
      "loss": 0.0498,
      "step": 17000
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.6520803570747375,
      "learning_rate": 4.730769230769231e-06,
      "loss": 0.0553,
      "step": 17100
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.04779316112399101,
      "learning_rate": 4.729096989966555e-06,
      "loss": 0.0677,
      "step": 17200
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.020316898822784424,
      "learning_rate": 4.727424749163879e-06,
      "loss": 0.0414,
      "step": 17300
    },
    {
      "epoch": 1.74,
      "grad_norm": 16.33078384399414,
      "learning_rate": 4.725752508361204e-06,
      "loss": 0.0506,
      "step": 17400
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.002986336126923561,
      "learning_rate": 4.724080267558529e-06,
      "loss": 0.0328,
      "step": 17500
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.395185470581055,
      "learning_rate": 4.722408026755853e-06,
      "loss": 0.0483,
      "step": 17600
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.003912128042429686,
      "learning_rate": 4.720735785953178e-06,
      "loss": 0.0541,
      "step": 17700
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.186915397644043,
      "learning_rate": 4.719063545150502e-06,
      "loss": 0.0455,
      "step": 17800
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.02249542437493801,
      "learning_rate": 4.717391304347826e-06,
      "loss": 0.0303,
      "step": 17900
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6102803945541382,
      "learning_rate": 4.715719063545151e-06,
      "loss": 0.049,
      "step": 18000
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.005787149537354708,
      "learning_rate": 4.714046822742475e-06,
      "loss": 0.0495,
      "step": 18100
    },
    {
      "epoch": 1.8199999999999998,
      "grad_norm": 0.01614544540643692,
      "learning_rate": 4.7123745819398e-06,
      "loss": 0.0506,
      "step": 18200
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.06813820451498032,
      "learning_rate": 4.710702341137124e-06,
      "loss": 0.0303,
      "step": 18300
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.004097160417586565,
      "learning_rate": 4.709030100334449e-06,
      "loss": 0.0447,
      "step": 18400
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.0063947769813239574,
      "learning_rate": 4.707357859531773e-06,
      "loss": 0.0497,
      "step": 18500
    },
    {
      "epoch": 1.8599999999999999,
      "grad_norm": 0.026598848402500153,
      "learning_rate": 4.705685618729097e-06,
      "loss": 0.066,
      "step": 18600
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.014818851836025715,
      "learning_rate": 4.704013377926422e-06,
      "loss": 0.0358,
      "step": 18700
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.0058324155397713184,
      "learning_rate": 4.702341137123746e-06,
      "loss": 0.0582,
      "step": 18800
    },
    {
      "epoch": 1.8900000000000001,
      "grad_norm": 13.650464057922363,
      "learning_rate": 4.700668896321071e-06,
      "loss": 0.0548,
      "step": 18900
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.02169709838926792,
      "learning_rate": 4.698996655518395e-06,
      "loss": 0.0441,
      "step": 19000
    },
    {
      "epoch": 1.9100000000000001,
      "grad_norm": 17.301109313964844,
      "learning_rate": 4.69732441471572e-06,
      "loss": 0.0583,
      "step": 19100
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.07251980900764465,
      "learning_rate": 4.695652173913044e-06,
      "loss": 0.0434,
      "step": 19200
    },
    {
      "epoch": 1.9300000000000002,
      "grad_norm": 2.6538381576538086,
      "learning_rate": 4.693979933110368e-06,
      "loss": 0.052,
      "step": 19300
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.03304781764745712,
      "learning_rate": 4.692307692307693e-06,
      "loss": 0.0305,
      "step": 19400
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.0017205484909936786,
      "learning_rate": 4.6906354515050175e-06,
      "loss": 0.0263,
      "step": 19500
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.002642621286213398,
      "learning_rate": 4.688963210702342e-06,
      "loss": 0.0445,
      "step": 19600
    },
    {
      "epoch": 1.97,
      "grad_norm": 3.4412660598754883,
      "learning_rate": 4.687290969899666e-06,
      "loss": 0.043,
      "step": 19700
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.002504587173462,
      "learning_rate": 4.6856187290969905e-06,
      "loss": 0.0745,
      "step": 19800
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.029486706480383873,
      "learning_rate": 4.6839464882943145e-06,
      "loss": 0.0444,
      "step": 19900
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.4088901877403259,
      "learning_rate": 4.6822742474916394e-06,
      "loss": 0.0299,
      "step": 20000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.984325,
      "eval_f1": 0.984325,
      "eval_loss": 0.07127311825752258,
      "eval_runtime": 133.5214,
      "eval_samples_per_second": 299.577,
      "eval_steps_per_second": 299.577,
      "step": 20000
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.011883600614964962,
      "learning_rate": 4.6806020066889635e-06,
      "loss": 0.0421,
      "step": 20100
    },
    {
      "epoch": 2.02,
      "grad_norm": 9.133939743041992,
      "learning_rate": 4.678929765886288e-06,
      "loss": 0.0409,
      "step": 20200
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.11910539120435715,
      "learning_rate": 4.677257525083612e-06,
      "loss": 0.0288,
      "step": 20300
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.025438936427235603,
      "learning_rate": 4.675585284280936e-06,
      "loss": 0.0236,
      "step": 20400
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.008798012509942055,
      "learning_rate": 4.673913043478261e-06,
      "loss": 0.0292,
      "step": 20500
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.0025494080036878586,
      "learning_rate": 4.672240802675585e-06,
      "loss": 0.0193,
      "step": 20600
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.011349349282681942,
      "learning_rate": 4.67056856187291e-06,
      "loss": 0.0249,
      "step": 20700
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.03200814127922058,
      "learning_rate": 4.668896321070234e-06,
      "loss": 0.03,
      "step": 20800
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.013645422644913197,
      "learning_rate": 4.667224080267559e-06,
      "loss": 0.0245,
      "step": 20900
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.006069705821573734,
      "learning_rate": 4.665551839464883e-06,
      "loss": 0.0494,
      "step": 21000
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9513150453567505,
      "learning_rate": 4.663879598662207e-06,
      "loss": 0.0274,
      "step": 21100
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.021194346249103546,
      "learning_rate": 4.662207357859532e-06,
      "loss": 0.0186,
      "step": 21200
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.010533172637224197,
      "learning_rate": 4.660535117056856e-06,
      "loss": 0.0268,
      "step": 21300
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.013652130030095577,
      "learning_rate": 4.658862876254181e-06,
      "loss": 0.0216,
      "step": 21400
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2660781145095825,
      "learning_rate": 4.657190635451506e-06,
      "loss": 0.0289,
      "step": 21500
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.008349668234586716,
      "learning_rate": 4.65551839464883e-06,
      "loss": 0.0359,
      "step": 21600
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1475175619125366,
      "learning_rate": 4.653846153846154e-06,
      "loss": 0.018,
      "step": 21700
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.06436525285243988,
      "learning_rate": 4.652173913043478e-06,
      "loss": 0.0274,
      "step": 21800
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.004730283282697201,
      "learning_rate": 4.650501672240803e-06,
      "loss": 0.0321,
      "step": 21900
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.03142133727669716,
      "learning_rate": 4.648829431438128e-06,
      "loss": 0.0309,
      "step": 22000
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.07798143476247787,
      "learning_rate": 4.647157190635452e-06,
      "loss": 0.0197,
      "step": 22100
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.0035459815990179777,
      "learning_rate": 4.645484949832777e-06,
      "loss": 0.0293,
      "step": 22200
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.04352254420518875,
      "learning_rate": 4.643812709030101e-06,
      "loss": 0.0308,
      "step": 22300
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.0022269452456384897,
      "learning_rate": 4.642140468227425e-06,
      "loss": 0.0367,
      "step": 22400
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.0067388033494353294,
      "learning_rate": 4.64046822742475e-06,
      "loss": 0.0309,
      "step": 22500
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.0021720202639698982,
      "learning_rate": 4.638795986622074e-06,
      "loss": 0.0338,
      "step": 22600
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.01130695827305317,
      "learning_rate": 4.637123745819399e-06,
      "loss": 0.0326,
      "step": 22700
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.027412299066781998,
      "learning_rate": 4.635451505016723e-06,
      "loss": 0.0288,
      "step": 22800
    },
    {
      "epoch": 2.29,
      "grad_norm": 23.797739028930664,
      "learning_rate": 4.6337792642140475e-06,
      "loss": 0.0237,
      "step": 22900
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.016235172748565674,
      "learning_rate": 4.6321070234113715e-06,
      "loss": 0.037,
      "step": 23000
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.0032159725669771433,
      "learning_rate": 4.630434782608696e-06,
      "loss": 0.0204,
      "step": 23100
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.0026251012459397316,
      "learning_rate": 4.6287625418060205e-06,
      "loss": 0.0346,
      "step": 23200
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.056268028914928436,
      "learning_rate": 4.6270903010033445e-06,
      "loss": 0.0349,
      "step": 23300
    },
    {
      "epoch": 2.34,
      "grad_norm": 32.879783630371094,
      "learning_rate": 4.625418060200669e-06,
      "loss": 0.0355,
      "step": 23400
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.05592450127005577,
      "learning_rate": 4.623745819397994e-06,
      "loss": 0.0255,
      "step": 23500
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.021603301167488098,
      "learning_rate": 4.622073578595318e-06,
      "loss": 0.0351,
      "step": 23600
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.0038026655092835426,
      "learning_rate": 4.620401337792642e-06,
      "loss": 0.0212,
      "step": 23700
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.002035396406427026,
      "learning_rate": 4.618729096989966e-06,
      "loss": 0.032,
      "step": 23800
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.054328680038452,
      "learning_rate": 4.617056856187291e-06,
      "loss": 0.0295,
      "step": 23900
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.004924594424664974,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.0157,
      "step": 24000
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.0015326552093029022,
      "learning_rate": 4.61371237458194e-06,
      "loss": 0.0257,
      "step": 24100
    },
    {
      "epoch": 2.42,
      "grad_norm": 45.68626403808594,
      "learning_rate": 4.612040133779265e-06,
      "loss": 0.0402,
      "step": 24200
    },
    {
      "epoch": 2.43,
      "grad_norm": 8.742071151733398,
      "learning_rate": 4.610367892976589e-06,
      "loss": 0.0602,
      "step": 24300
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.11714211106300354,
      "learning_rate": 4.608695652173913e-06,
      "loss": 0.0245,
      "step": 24400
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.02803937904536724,
      "learning_rate": 4.607023411371238e-06,
      "loss": 0.0347,
      "step": 24500
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.0011470599565654993,
      "learning_rate": 4.605351170568562e-06,
      "loss": 0.0214,
      "step": 24600
    },
    {
      "epoch": 2.4699999999999998,
      "grad_norm": 0.004855659324675798,
      "learning_rate": 4.603678929765887e-06,
      "loss": 0.025,
      "step": 24700
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.12526443600654602,
      "learning_rate": 4.602006688963211e-06,
      "loss": 0.0208,
      "step": 24800
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.06537730991840363,
      "learning_rate": 4.600334448160536e-06,
      "loss": 0.0489,
      "step": 24900
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.023756371811032295,
      "learning_rate": 4.59866220735786e-06,
      "loss": 0.027,
      "step": 25000
    },
    {
      "epoch": 2.5,
      "eval_accuracy": 0.98445,
      "eval_f1": 0.98445,
      "eval_loss": 0.08253531157970428,
      "eval_runtime": 133.5816,
      "eval_samples_per_second": 299.442,
      "eval_steps_per_second": 299.442,
      "step": 25000
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.04911258444190025,
      "learning_rate": 4.596989966555184e-06,
      "loss": 0.0412,
      "step": 25100
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.2437671422958374,
      "learning_rate": 4.595317725752509e-06,
      "loss": 0.0226,
      "step": 25200
    },
    {
      "epoch": 2.5300000000000002,
      "grad_norm": 0.01876792125403881,
      "learning_rate": 4.593645484949833e-06,
      "loss": 0.032,
      "step": 25300
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.05922747775912285,
      "learning_rate": 4.591973244147158e-06,
      "loss": 0.0256,
      "step": 25400
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.008321844041347504,
      "learning_rate": 4.590301003344483e-06,
      "loss": 0.0189,
      "step": 25500
    },
    {
      "epoch": 2.56,
      "grad_norm": 37.782745361328125,
      "learning_rate": 4.588628762541807e-06,
      "loss": 0.0366,
      "step": 25600
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.005182887427508831,
      "learning_rate": 4.586956521739131e-06,
      "loss": 0.0404,
      "step": 25700
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.02100360207259655,
      "learning_rate": 4.585284280936455e-06,
      "loss": 0.0269,
      "step": 25800
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.017707619816064835,
      "learning_rate": 4.58361204013378e-06,
      "loss": 0.0185,
      "step": 25900
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.12977050244808197,
      "learning_rate": 4.581939799331104e-06,
      "loss": 0.0207,
      "step": 26000
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.0024964429903775454,
      "learning_rate": 4.5802675585284286e-06,
      "loss": 0.0541,
      "step": 26100
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7643945217132568,
      "learning_rate": 4.5785953177257534e-06,
      "loss": 0.0255,
      "step": 26200
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.13935905694961548,
      "learning_rate": 4.5769230769230775e-06,
      "loss": 0.0331,
      "step": 26300
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.010719064623117447,
      "learning_rate": 4.5752508361204015e-06,
      "loss": 0.025,
      "step": 26400
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.039186276495456696,
      "learning_rate": 4.5735785953177255e-06,
      "loss": 0.0459,
      "step": 26500
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.06738638877868652,
      "learning_rate": 4.5719063545150504e-06,
      "loss": 0.0178,
      "step": 26600
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.014554286375641823,
      "learning_rate": 4.570234113712375e-06,
      "loss": 0.041,
      "step": 26700
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.032161399722099304,
      "learning_rate": 4.568561872909699e-06,
      "loss": 0.0415,
      "step": 26800
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.006425254512578249,
      "learning_rate": 4.566889632107024e-06,
      "loss": 0.0166,
      "step": 26900
    },
    {
      "epoch": 2.7,
      "grad_norm": 120.80119323730469,
      "learning_rate": 4.565217391304348e-06,
      "loss": 0.0515,
      "step": 27000
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.09112800657749176,
      "learning_rate": 4.563545150501672e-06,
      "loss": 0.038,
      "step": 27100
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.24947082996368408,
      "learning_rate": 4.561872909698997e-06,
      "loss": 0.0302,
      "step": 27200
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.03329513594508171,
      "learning_rate": 4.560200668896321e-06,
      "loss": 0.0277,
      "step": 27300
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.010453999973833561,
      "learning_rate": 4.558528428093646e-06,
      "loss": 0.0205,
      "step": 27400
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.18513476848602295,
      "learning_rate": 4.55685618729097e-06,
      "loss": 0.0176,
      "step": 27500
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.016682449728250504,
      "learning_rate": 4.555183946488295e-06,
      "loss": 0.0419,
      "step": 27600
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.0067235734313726425,
      "learning_rate": 4.553511705685619e-06,
      "loss": 0.0413,
      "step": 27700
    },
    {
      "epoch": 2.7800000000000002,
      "grad_norm": 0.008218110539019108,
      "learning_rate": 4.551839464882943e-06,
      "loss": 0.0258,
      "step": 27800
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.006087880581617355,
      "learning_rate": 4.550167224080268e-06,
      "loss": 0.0102,
      "step": 27900
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.0023886614944785833,
      "learning_rate": 4.548494983277592e-06,
      "loss": 0.024,
      "step": 28000
    },
    {
      "epoch": 2.81,
      "grad_norm": 74.37184143066406,
      "learning_rate": 4.546822742474917e-06,
      "loss": 0.0469,
      "step": 28100
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.5758330821990967,
      "learning_rate": 4.545150501672241e-06,
      "loss": 0.0288,
      "step": 28200
    },
    {
      "epoch": 2.83,
      "grad_norm": 10.103140830993652,
      "learning_rate": 4.543478260869566e-06,
      "loss": 0.0369,
      "step": 28300
    },
    {
      "epoch": 2.84,
      "grad_norm": 36.231136322021484,
      "learning_rate": 4.54180602006689e-06,
      "loss": 0.0202,
      "step": 28400
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.0008801018702797592,
      "learning_rate": 4.540133779264214e-06,
      "loss": 0.0331,
      "step": 28500
    },
    {
      "epoch": 2.86,
      "grad_norm": 5.780725955963135,
      "learning_rate": 4.538461538461539e-06,
      "loss": 0.0299,
      "step": 28600
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.0014527951134368777,
      "learning_rate": 4.536789297658864e-06,
      "loss": 0.028,
      "step": 28700
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.0038458879571408033,
      "learning_rate": 4.535117056856188e-06,
      "loss": 0.0192,
      "step": 28800
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.04886145889759064,
      "learning_rate": 4.533444816053512e-06,
      "loss": 0.026,
      "step": 28900
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.034207429736852646,
      "learning_rate": 4.531772575250837e-06,
      "loss": 0.0287,
      "step": 29000
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.1304664611816406,
      "learning_rate": 4.530100334448161e-06,
      "loss": 0.0283,
      "step": 29100
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.0010741673177108169,
      "learning_rate": 4.5284280936454856e-06,
      "loss": 0.0284,
      "step": 29200
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.025598613545298576,
      "learning_rate": 4.52675585284281e-06,
      "loss": 0.0251,
      "step": 29300
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.1095525324344635,
      "learning_rate": 4.5250836120401345e-06,
      "loss": 0.0306,
      "step": 29400
    },
    {
      "epoch": 2.95,
      "grad_norm": 6.888746738433838,
      "learning_rate": 4.5234113712374585e-06,
      "loss": 0.0336,
      "step": 29500
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.03379179537296295,
      "learning_rate": 4.5217391304347826e-06,
      "loss": 0.0231,
      "step": 29600
    },
    {
      "epoch": 2.9699999999999998,
      "grad_norm": 0.00660817651078105,
      "learning_rate": 4.5200668896321074e-06,
      "loss": 0.0258,
      "step": 29700
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.0639776960015297,
      "learning_rate": 4.5183946488294315e-06,
      "loss": 0.0459,
      "step": 29800
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.025130316615104675,
      "learning_rate": 4.516722408026756e-06,
      "loss": 0.0371,
      "step": 29900
    },
    {
      "epoch": 3.0,
      "grad_norm": 12.60319709777832,
      "learning_rate": 4.51505016722408e-06,
      "loss": 0.0262,
      "step": 30000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.985425,
      "eval_f1": 0.985425,
      "eval_loss": 0.07783249020576477,
      "eval_runtime": 134.4077,
      "eval_samples_per_second": 297.602,
      "eval_steps_per_second": 297.602,
      "step": 30000
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.3146221935749054,
      "learning_rate": 4.513377926421405e-06,
      "loss": 0.007,
      "step": 30100
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.02126391790807247,
      "learning_rate": 4.511705685618729e-06,
      "loss": 0.0097,
      "step": 30200
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.0014831286389380693,
      "learning_rate": 4.510033444816053e-06,
      "loss": 0.0172,
      "step": 30300
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.011143388226628304,
      "learning_rate": 4.508361204013378e-06,
      "loss": 0.0233,
      "step": 30400
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.005537919234484434,
      "learning_rate": 4.506688963210702e-06,
      "loss": 0.0375,
      "step": 30500
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.06545744836330414,
      "learning_rate": 4.505016722408027e-06,
      "loss": 0.0218,
      "step": 30600
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.01571493409574032,
      "learning_rate": 4.503344481605352e-06,
      "loss": 0.0149,
      "step": 30700
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.0030460015404969454,
      "learning_rate": 4.501672240802676e-06,
      "loss": 0.0233,
      "step": 30800
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.0006564293871633708,
      "learning_rate": 4.5e-06,
      "loss": 0.0068,
      "step": 30900
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.009112296625971794,
      "learning_rate": 4.498327759197324e-06,
      "loss": 0.0067,
      "step": 31000
    },
    {
      "epoch": 3.11,
      "grad_norm": 20.75179100036621,
      "learning_rate": 4.496655518394649e-06,
      "loss": 0.0252,
      "step": 31100
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.1207340955734253,
      "learning_rate": 4.494983277591974e-06,
      "loss": 0.0118,
      "step": 31200
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.016456255689263344,
      "learning_rate": 4.493311036789298e-06,
      "loss": 0.035,
      "step": 31300
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.0019782930612564087,
      "learning_rate": 4.491638795986623e-06,
      "loss": 0.0167,
      "step": 31400
    },
    {
      "epoch": 3.15,
      "grad_norm": 13.507879257202148,
      "learning_rate": 4.489966555183947e-06,
      "loss": 0.0295,
      "step": 31500
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.37393996119499207,
      "learning_rate": 4.488294314381271e-06,
      "loss": 0.0119,
      "step": 31600
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.001511622336693108,
      "learning_rate": 4.486622073578596e-06,
      "loss": 0.0075,
      "step": 31700
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.005662756972014904,
      "learning_rate": 4.48494983277592e-06,
      "loss": 0.0215,
      "step": 31800
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.042217742651700974,
      "learning_rate": 4.483277591973245e-06,
      "loss": 0.0124,
      "step": 31900
    },
    {
      "epoch": 3.2,
      "grad_norm": 7.794501781463623,
      "learning_rate": 4.481605351170569e-06,
      "loss": 0.0256,
      "step": 32000
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.004403349477797747,
      "learning_rate": 4.479933110367894e-06,
      "loss": 0.0202,
      "step": 32100
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.0071293930523097515,
      "learning_rate": 4.478260869565218e-06,
      "loss": 0.0138,
      "step": 32200
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.020876120775938034,
      "learning_rate": 4.476588628762542e-06,
      "loss": 0.0137,
      "step": 32300
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.18674761056900024,
      "learning_rate": 4.474916387959867e-06,
      "loss": 0.0183,
      "step": 32400
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.09638151526451111,
      "learning_rate": 4.473244147157191e-06,
      "loss": 0.0308,
      "step": 32500
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.0009971513645723462,
      "learning_rate": 4.4715719063545155e-06,
      "loss": 0.0204,
      "step": 32600
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.044390495866537094,
      "learning_rate": 4.46989966555184e-06,
      "loss": 0.0177,
      "step": 32700
    },
    {
      "epoch": 3.2800000000000002,
      "grad_norm": 0.050020378082990646,
      "learning_rate": 4.4682274247491644e-06,
      "loss": 0.0114,
      "step": 32800
    },
    {
      "epoch": 3.29,
      "grad_norm": 18.936073303222656,
      "learning_rate": 4.4665551839464885e-06,
      "loss": 0.0184,
      "step": 32900
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.008622376248240471,
      "learning_rate": 4.4648829431438125e-06,
      "loss": 0.0135,
      "step": 33000
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.005060569848865271,
      "learning_rate": 4.463210702341137e-06,
      "loss": 0.0182,
      "step": 33100
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.004180543590337038,
      "learning_rate": 4.461538461538462e-06,
      "loss": 0.0174,
      "step": 33200
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.0003817367251031101,
      "learning_rate": 4.459866220735786e-06,
      "loss": 0.0163,
      "step": 33300
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.02846093289554119,
      "learning_rate": 4.458193979933111e-06,
      "loss": 0.0343,
      "step": 33400
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.05180152505636215,
      "learning_rate": 4.456521739130435e-06,
      "loss": 0.0273,
      "step": 33500
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.02195872738957405,
      "learning_rate": 4.454849498327759e-06,
      "loss": 0.0139,
      "step": 33600
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.07758418470621109,
      "learning_rate": 4.453177257525084e-06,
      "loss": 0.0185,
      "step": 33700
    },
    {
      "epoch": 3.38,
      "grad_norm": 38.56128692626953,
      "learning_rate": 4.451505016722408e-06,
      "loss": 0.0118,
      "step": 33800
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.009722617454826832,
      "learning_rate": 4.449832775919733e-06,
      "loss": 0.0059,
      "step": 33900
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.0015089679509401321,
      "learning_rate": 4.448160535117057e-06,
      "loss": 0.0358,
      "step": 34000
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.01531163975596428,
      "learning_rate": 4.446488294314382e-06,
      "loss": 0.011,
      "step": 34100
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.009290250949561596,
      "learning_rate": 4.444816053511706e-06,
      "loss": 0.0156,
      "step": 34200
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.09569079428911209,
      "learning_rate": 4.44314381270903e-06,
      "loss": 0.0264,
      "step": 34300
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.0036293207667768,
      "learning_rate": 4.441471571906355e-06,
      "loss": 0.0204,
      "step": 34400
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.002437431598082185,
      "learning_rate": 4.439799331103679e-06,
      "loss": 0.0208,
      "step": 34500
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.0011440415401011705,
      "learning_rate": 4.438127090301004e-06,
      "loss": 0.017,
      "step": 34600
    },
    {
      "epoch": 3.4699999999999998,
      "grad_norm": 0.03422205522656441,
      "learning_rate": 4.436454849498329e-06,
      "loss": 0.0068,
      "step": 34700
    },
    {
      "epoch": 3.48,
      "grad_norm": 11.362318992614746,
      "learning_rate": 4.434782608695653e-06,
      "loss": 0.0208,
      "step": 34800
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.0038512239698320627,
      "learning_rate": 4.433110367892977e-06,
      "loss": 0.011,
      "step": 34900
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.016066549345850945,
      "learning_rate": 4.431438127090301e-06,
      "loss": 0.0181,
      "step": 35000
    },
    {
      "epoch": 3.5,
      "eval_accuracy": 0.9857,
      "eval_f1": 0.9857,
      "eval_loss": 0.09534145146608353,
      "eval_runtime": 134.3932,
      "eval_samples_per_second": 297.634,
      "eval_steps_per_second": 297.634,
      "step": 35000
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.0015177098102867603,
      "learning_rate": 4.429765886287626e-06,
      "loss": 0.0243,
      "step": 35100
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.011164546944200993,
      "learning_rate": 4.428093645484951e-06,
      "loss": 0.0174,
      "step": 35200
    },
    {
      "epoch": 3.5300000000000002,
      "grad_norm": 0.4808778762817383,
      "learning_rate": 4.426421404682275e-06,
      "loss": 0.017,
      "step": 35300
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.07465644180774689,
      "learning_rate": 4.4247491638795996e-06,
      "loss": 0.026,
      "step": 35400
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.05953597277402878,
      "learning_rate": 4.423076923076924e-06,
      "loss": 0.0343,
      "step": 35500
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.2559371590614319,
      "learning_rate": 4.421404682274248e-06,
      "loss": 0.0203,
      "step": 35600
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.000921146129257977,
      "learning_rate": 4.4197324414715725e-06,
      "loss": 0.0174,
      "step": 35700
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.03776590898633003,
      "learning_rate": 4.4180602006688966e-06,
      "loss": 0.0152,
      "step": 35800
    },
    {
      "epoch": 3.59,
      "grad_norm": 9.220237731933594,
      "learning_rate": 4.4163879598662214e-06,
      "loss": 0.0214,
      "step": 35900
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.0010363379260525107,
      "learning_rate": 4.4147157190635455e-06,
      "loss": 0.0268,
      "step": 36000
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.11337871849536896,
      "learning_rate": 4.41304347826087e-06,
      "loss": 0.0289,
      "step": 36100
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.03623759001493454,
      "learning_rate": 4.411371237458194e-06,
      "loss": 0.0165,
      "step": 36200
    },
    {
      "epoch": 3.63,
      "grad_norm": 59.48095703125,
      "learning_rate": 4.4096989966555184e-06,
      "loss": 0.0111,
      "step": 36300
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.0512019619345665,
      "learning_rate": 4.408026755852843e-06,
      "loss": 0.0289,
      "step": 36400
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.004787762649357319,
      "learning_rate": 4.406354515050167e-06,
      "loss": 0.0125,
      "step": 36500
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.2356896698474884,
      "learning_rate": 4.404682274247492e-06,
      "loss": 0.0203,
      "step": 36600
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.07559309899806976,
      "learning_rate": 4.403010033444816e-06,
      "loss": 0.0112,
      "step": 36700
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.0012310033198446035,
      "learning_rate": 4.401337792642141e-06,
      "loss": 0.0167,
      "step": 36800
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.07424982637166977,
      "learning_rate": 4.399665551839465e-06,
      "loss": 0.0151,
      "step": 36900
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.00213808030821383,
      "learning_rate": 4.397993311036789e-06,
      "loss": 0.0087,
      "step": 37000
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.08704973757266998,
      "learning_rate": 4.396321070234114e-06,
      "loss": 0.0081,
      "step": 37100
    },
    {
      "epoch": 3.7199999999999998,
      "grad_norm": 0.0880984216928482,
      "learning_rate": 4.394648829431438e-06,
      "loss": 0.0347,
      "step": 37200
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.0027694138698279858,
      "learning_rate": 4.392976588628763e-06,
      "loss": 0.0319,
      "step": 37300
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.003729386255145073,
      "learning_rate": 4.391304347826087e-06,
      "loss": 0.0114,
      "step": 37400
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.018702877685427666,
      "learning_rate": 4.389632107023412e-06,
      "loss": 0.0229,
      "step": 37500
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.00089402956655249,
      "learning_rate": 4.387959866220736e-06,
      "loss": 0.0144,
      "step": 37600
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.0016536782495677471,
      "learning_rate": 4.38628762541806e-06,
      "loss": 0.0251,
      "step": 37700
    },
    {
      "epoch": 3.7800000000000002,
      "grad_norm": 0.0014757058816030622,
      "learning_rate": 4.384615384615385e-06,
      "loss": 0.0407,
      "step": 37800
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.0005884135607630014,
      "learning_rate": 4.38294314381271e-06,
      "loss": 0.0171,
      "step": 37900
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.0046568759717047215,
      "learning_rate": 4.381270903010034e-06,
      "loss": 0.017,
      "step": 38000
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.03397432714700699,
      "learning_rate": 4.379598662207358e-06,
      "loss": 0.0338,
      "step": 38100
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.0011023873230442405,
      "learning_rate": 4.377926421404683e-06,
      "loss": 0.0124,
      "step": 38200
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.002390155801549554,
      "learning_rate": 4.376254180602007e-06,
      "loss": 0.0255,
      "step": 38300
    },
    {
      "epoch": 3.84,
      "grad_norm": 16.321758270263672,
      "learning_rate": 4.374581939799332e-06,
      "loss": 0.0244,
      "step": 38400
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.0222801323980093,
      "learning_rate": 4.372909698996656e-06,
      "loss": 0.0185,
      "step": 38500
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.002239386085420847,
      "learning_rate": 4.371237458193981e-06,
      "loss": 0.0116,
      "step": 38600
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.00036294609890319407,
      "learning_rate": 4.369565217391305e-06,
      "loss": 0.024,
      "step": 38700
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.0006557508604601026,
      "learning_rate": 4.367892976588629e-06,
      "loss": 0.0166,
      "step": 38800
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.0019109762506559491,
      "learning_rate": 4.3662207357859536e-06,
      "loss": 0.0123,
      "step": 38900
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.0005525159067474306,
      "learning_rate": 4.364548494983278e-06,
      "loss": 0.0111,
      "step": 39000
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.00040455596172250807,
      "learning_rate": 4.3628762541806025e-06,
      "loss": 0.0213,
      "step": 39100
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.0005091234343126416,
      "learning_rate": 4.3612040133779265e-06,
      "loss": 0.0156,
      "step": 39200
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.0005204415647312999,
      "learning_rate": 4.359531772575251e-06,
      "loss": 0.0086,
      "step": 39300
    },
    {
      "epoch": 3.94,
      "grad_norm": 22.845434188842773,
      "learning_rate": 4.3578595317725754e-06,
      "loss": 0.022,
      "step": 39400
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.051367294043302536,
      "learning_rate": 4.3561872909698995e-06,
      "loss": 0.0303,
      "step": 39500
    },
    {
      "epoch": 3.96,
      "grad_norm": 48.237937927246094,
      "learning_rate": 4.354515050167224e-06,
      "loss": 0.0177,
      "step": 39600
    },
    {
      "epoch": 3.9699999999999998,
      "grad_norm": 0.00723648676648736,
      "learning_rate": 4.352842809364548e-06,
      "loss": 0.024,
      "step": 39700
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.07704585045576096,
      "learning_rate": 4.351170568561873e-06,
      "loss": 0.0159,
      "step": 39800
    },
    {
      "epoch": 3.99,
      "grad_norm": 8.135265350341797,
      "learning_rate": 4.349498327759198e-06,
      "loss": 0.0237,
      "step": 39900
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.006071310956031084,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.0291,
      "step": 40000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.985475,
      "eval_f1": 0.985475,
      "eval_loss": 0.07918225228786469,
      "eval_runtime": 137.0078,
      "eval_samples_per_second": 291.954,
      "eval_steps_per_second": 291.954,
      "step": 40000
    },
    {
      "epoch": 4.01,
      "grad_norm": 0.06916201114654541,
      "learning_rate": 4.346153846153846e-06,
      "loss": 0.0175,
      "step": 40100
    },
    {
      "epoch": 4.02,
      "grad_norm": 0.013865008018910885,
      "learning_rate": 4.34448160535117e-06,
      "loss": 0.0096,
      "step": 40200
    },
    {
      "epoch": 4.03,
      "grad_norm": 0.00577643932774663,
      "learning_rate": 4.342809364548495e-06,
      "loss": 0.0143,
      "step": 40300
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.00038059207145124674,
      "learning_rate": 4.34113712374582e-06,
      "loss": 0.0054,
      "step": 40400
    },
    {
      "epoch": 4.05,
      "grad_norm": 0.0015450104838237166,
      "learning_rate": 4.339464882943144e-06,
      "loss": 0.0107,
      "step": 40500
    },
    {
      "epoch": 4.06,
      "grad_norm": 0.00023171660723164678,
      "learning_rate": 4.337792642140469e-06,
      "loss": 0.0055,
      "step": 40600
    },
    {
      "epoch": 4.07,
      "grad_norm": 0.00019739058916456997,
      "learning_rate": 4.336120401337793e-06,
      "loss": 0.0093,
      "step": 40700
    },
    {
      "epoch": 4.08,
      "grad_norm": 0.0005960884736850858,
      "learning_rate": 4.334448160535117e-06,
      "loss": 0.0144,
      "step": 40800
    },
    {
      "epoch": 4.09,
      "grad_norm": 0.0004620914114639163,
      "learning_rate": 4.332775919732442e-06,
      "loss": 0.0192,
      "step": 40900
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.16995316743850708,
      "learning_rate": 4.331103678929766e-06,
      "loss": 0.0061,
      "step": 41000
    },
    {
      "epoch": 4.11,
      "grad_norm": 0.03042794018983841,
      "learning_rate": 4.329431438127091e-06,
      "loss": 0.0102,
      "step": 41100
    },
    {
      "epoch": 4.12,
      "grad_norm": 0.00546935573220253,
      "learning_rate": 4.327759197324415e-06,
      "loss": 0.0164,
      "step": 41200
    },
    {
      "epoch": 4.13,
      "grad_norm": 0.012843598611652851,
      "learning_rate": 4.32608695652174e-06,
      "loss": 0.0089,
      "step": 41300
    },
    {
      "epoch": 4.14,
      "grad_norm": 0.000863418448716402,
      "learning_rate": 4.324414715719064e-06,
      "loss": 0.0182,
      "step": 41400
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.007233878131955862,
      "learning_rate": 4.322742474916388e-06,
      "loss": 0.0094,
      "step": 41500
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.003833547467365861,
      "learning_rate": 4.321070234113713e-06,
      "loss": 0.0095,
      "step": 41600
    },
    {
      "epoch": 4.17,
      "grad_norm": 0.00022056704619899392,
      "learning_rate": 4.319397993311037e-06,
      "loss": 0.0067,
      "step": 41700
    },
    {
      "epoch": 4.18,
      "grad_norm": 0.00015522346075158566,
      "learning_rate": 4.317725752508362e-06,
      "loss": 0.0031,
      "step": 41800
    },
    {
      "epoch": 4.19,
      "grad_norm": 107.11628723144531,
      "learning_rate": 4.3160535117056865e-06,
      "loss": 0.0072,
      "step": 41900
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.013127053156495094,
      "learning_rate": 4.3143812709030106e-06,
      "loss": 0.0164,
      "step": 42000
    },
    {
      "epoch": 4.21,
      "grad_norm": 0.0003005210601259023,
      "learning_rate": 4.312709030100335e-06,
      "loss": 0.0173,
      "step": 42100
    },
    {
      "epoch": 4.22,
      "grad_norm": 0.0003234084288123995,
      "learning_rate": 4.311036789297659e-06,
      "loss": 0.0122,
      "step": 42200
    },
    {
      "epoch": 4.23,
      "grad_norm": 0.1619456708431244,
      "learning_rate": 4.3093645484949835e-06,
      "loss": 0.0099,
      "step": 42300
    },
    {
      "epoch": 4.24,
      "grad_norm": 0.0008889096206985414,
      "learning_rate": 4.307692307692308e-06,
      "loss": 0.0191,
      "step": 42400
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.0017285248031839728,
      "learning_rate": 4.3060200668896324e-06,
      "loss": 0.0106,
      "step": 42500
    },
    {
      "epoch": 4.26,
      "grad_norm": 0.00019850315584335476,
      "learning_rate": 4.304347826086957e-06,
      "loss": 0.0172,
      "step": 42600
    },
    {
      "epoch": 4.27,
      "grad_norm": 0.00038836541352793574,
      "learning_rate": 4.302675585284281e-06,
      "loss": 0.0152,
      "step": 42700
    },
    {
      "epoch": 4.28,
      "grad_norm": 0.00012064904876751825,
      "learning_rate": 4.301003344481605e-06,
      "loss": 0.0062,
      "step": 42800
    },
    {
      "epoch": 4.29,
      "grad_norm": 0.021381106227636337,
      "learning_rate": 4.29933110367893e-06,
      "loss": 0.0141,
      "step": 42900
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.02078850567340851,
      "learning_rate": 4.297658862876254e-06,
      "loss": 0.0144,
      "step": 43000
    },
    {
      "epoch": 4.31,
      "grad_norm": 0.0006510872044600546,
      "learning_rate": 4.295986622073579e-06,
      "loss": 0.0067,
      "step": 43100
    },
    {
      "epoch": 4.32,
      "grad_norm": 0.0023323597852140665,
      "learning_rate": 4.294314381270903e-06,
      "loss": 0.0047,
      "step": 43200
    },
    {
      "epoch": 4.33,
      "grad_norm": 0.0004781058814842254,
      "learning_rate": 4.292642140468228e-06,
      "loss": 0.0063,
      "step": 43300
    },
    {
      "epoch": 4.34,
      "grad_norm": 0.0051570613868534565,
      "learning_rate": 4.290969899665552e-06,
      "loss": 0.0222,
      "step": 43400
    },
    {
      "epoch": 4.35,
      "grad_norm": 0.23171906173229218,
      "learning_rate": 4.289297658862876e-06,
      "loss": 0.0053,
      "step": 43500
    },
    {
      "epoch": 4.36,
      "grad_norm": 0.04821045324206352,
      "learning_rate": 4.287625418060201e-06,
      "loss": 0.0177,
      "step": 43600
    },
    {
      "epoch": 4.37,
      "grad_norm": 0.0022421092726290226,
      "learning_rate": 4.285953177257525e-06,
      "loss": 0.0012,
      "step": 43700
    },
    {
      "epoch": 4.38,
      "grad_norm": 0.0016597072826698422,
      "learning_rate": 4.28428093645485e-06,
      "loss": 0.0063,
      "step": 43800
    },
    {
      "epoch": 4.39,
      "grad_norm": 0.166465625166893,
      "learning_rate": 4.282608695652175e-06,
      "loss": 0.0209,
      "step": 43900
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.05712735652923584,
      "learning_rate": 4.280936454849499e-06,
      "loss": 0.0171,
      "step": 44000
    },
    {
      "epoch": 4.41,
      "grad_norm": 0.0028721943963319063,
      "learning_rate": 4.279264214046823e-06,
      "loss": 0.0002,
      "step": 44100
    },
    {
      "epoch": 4.42,
      "grad_norm": 0.014931490644812584,
      "learning_rate": 4.277591973244147e-06,
      "loss": 0.0155,
      "step": 44200
    },
    {
      "epoch": 4.43,
      "grad_norm": 0.0601530522108078,
      "learning_rate": 4.275919732441472e-06,
      "loss": 0.0048,
      "step": 44300
    },
    {
      "epoch": 4.44,
      "grad_norm": 0.00023735656577628106,
      "learning_rate": 4.274247491638797e-06,
      "loss": 0.0067,
      "step": 44400
    },
    {
      "epoch": 4.45,
      "grad_norm": 0.00011775241728173569,
      "learning_rate": 4.272575250836121e-06,
      "loss": 0.0102,
      "step": 44500
    },
    {
      "epoch": 4.46,
      "grad_norm": 0.0006663117092102766,
      "learning_rate": 4.270903010033446e-06,
      "loss": 0.0159,
      "step": 44600
    },
    {
      "epoch": 4.47,
      "grad_norm": 0.0003513609990477562,
      "learning_rate": 4.26923076923077e-06,
      "loss": 0.0035,
      "step": 44700
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.00022971085854806006,
      "learning_rate": 4.267558528428094e-06,
      "loss": 0.0238,
      "step": 44800
    },
    {
      "epoch": 4.49,
      "grad_norm": 0.0004504798853304237,
      "learning_rate": 4.265886287625419e-06,
      "loss": 0.012,
      "step": 44900
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.00031274466891773045,
      "learning_rate": 4.264214046822743e-06,
      "loss": 0.0194,
      "step": 45000
    },
    {
      "epoch": 4.5,
      "eval_accuracy": 0.98525,
      "eval_f1": 0.98525,
      "eval_loss": 0.10297723114490509,
      "eval_runtime": 138.583,
      "eval_samples_per_second": 288.636,
      "eval_steps_per_second": 288.636,
      "step": 45000
    },
    {
      "epoch": 4.51,
      "grad_norm": 0.0387171134352684,
      "learning_rate": 4.2625418060200676e-06,
      "loss": 0.0134,
      "step": 45100
    },
    {
      "epoch": 4.52,
      "grad_norm": 0.00037836728733964264,
      "learning_rate": 4.260869565217392e-06,
      "loss": 0.0142,
      "step": 45200
    },
    {
      "epoch": 4.53,
      "grad_norm": 0.04093967378139496,
      "learning_rate": 4.2591973244147165e-06,
      "loss": 0.0105,
      "step": 45300
    },
    {
      "epoch": 4.54,
      "grad_norm": 0.0007085870602168143,
      "learning_rate": 4.2575250836120405e-06,
      "loss": 0.0196,
      "step": 45400
    },
    {
      "epoch": 4.55,
      "grad_norm": 0.003028183476999402,
      "learning_rate": 4.2558528428093646e-06,
      "loss": 0.005,
      "step": 45500
    },
    {
      "epoch": 4.5600000000000005,
      "grad_norm": 0.019922267645597458,
      "learning_rate": 4.2541806020066895e-06,
      "loss": 0.0144,
      "step": 45600
    },
    {
      "epoch": 4.57,
      "grad_norm": 0.0007607261650264263,
      "learning_rate": 4.2525083612040135e-06,
      "loss": 0.0121,
      "step": 45700
    },
    {
      "epoch": 4.58,
      "grad_norm": 0.0019423003541305661,
      "learning_rate": 4.250836120401338e-06,
      "loss": 0.0075,
      "step": 45800
    },
    {
      "epoch": 4.59,
      "grad_norm": 0.0002597976417746395,
      "learning_rate": 4.249163879598662e-06,
      "loss": 0.0077,
      "step": 45900
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.0011591215152293444,
      "learning_rate": 4.247491638795987e-06,
      "loss": 0.0206,
      "step": 46000
    },
    {
      "epoch": 4.61,
      "grad_norm": 0.02335204742848873,
      "learning_rate": 4.245819397993311e-06,
      "loss": 0.0175,
      "step": 46100
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.004985714331269264,
      "learning_rate": 4.244147157190635e-06,
      "loss": 0.0134,
      "step": 46200
    },
    {
      "epoch": 4.63,
      "grad_norm": 0.00042676858720369637,
      "learning_rate": 4.24247491638796e-06,
      "loss": 0.0248,
      "step": 46300
    },
    {
      "epoch": 4.64,
      "grad_norm": 0.03829636424779892,
      "learning_rate": 4.240802675585284e-06,
      "loss": 0.0034,
      "step": 46400
    },
    {
      "epoch": 4.65,
      "grad_norm": 0.0008630960946902633,
      "learning_rate": 4.239130434782609e-06,
      "loss": 0.0099,
      "step": 46500
    },
    {
      "epoch": 4.66,
      "grad_norm": 16.075529098510742,
      "learning_rate": 4.237458193979934e-06,
      "loss": 0.0116,
      "step": 46600
    },
    {
      "epoch": 4.67,
      "grad_norm": 0.10533302277326584,
      "learning_rate": 4.235785953177258e-06,
      "loss": 0.0092,
      "step": 46700
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.01659245416522026,
      "learning_rate": 4.234113712374582e-06,
      "loss": 0.0272,
      "step": 46800
    },
    {
      "epoch": 4.6899999999999995,
      "grad_norm": 0.043937236070632935,
      "learning_rate": 4.232441471571906e-06,
      "loss": 0.0188,
      "step": 46900
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.00019682837591972202,
      "learning_rate": 4.230769230769231e-06,
      "loss": 0.0132,
      "step": 47000
    },
    {
      "epoch": 4.71,
      "grad_norm": 0.001616219524294138,
      "learning_rate": 4.229096989966556e-06,
      "loss": 0.0112,
      "step": 47100
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.004383947234600782,
      "learning_rate": 4.22742474916388e-06,
      "loss": 0.0096,
      "step": 47200
    },
    {
      "epoch": 4.73,
      "grad_norm": 0.003946308046579361,
      "learning_rate": 4.225752508361204e-06,
      "loss": 0.0126,
      "step": 47300
    },
    {
      "epoch": 4.74,
      "grad_norm": 0.010722492821514606,
      "learning_rate": 4.224080267558529e-06,
      "loss": 0.0072,
      "step": 47400
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.0002598193532321602,
      "learning_rate": 4.222408026755853e-06,
      "loss": 0.008,
      "step": 47500
    },
    {
      "epoch": 4.76,
      "grad_norm": 0.03757161274552345,
      "learning_rate": 4.220735785953178e-06,
      "loss": 0.0086,
      "step": 47600
    },
    {
      "epoch": 4.77,
      "grad_norm": 0.0003191785654053092,
      "learning_rate": 4.219063545150502e-06,
      "loss": 0.0242,
      "step": 47700
    },
    {
      "epoch": 4.78,
      "grad_norm": 0.006073308642953634,
      "learning_rate": 4.217391304347827e-06,
      "loss": 0.0075,
      "step": 47800
    },
    {
      "epoch": 4.79,
      "grad_norm": 0.00026708669611252844,
      "learning_rate": 4.215719063545151e-06,
      "loss": 0.0121,
      "step": 47900
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.0015478582354262471,
      "learning_rate": 4.214046822742475e-06,
      "loss": 0.0116,
      "step": 48000
    },
    {
      "epoch": 4.8100000000000005,
      "grad_norm": 0.0008126681204885244,
      "learning_rate": 4.2123745819398e-06,
      "loss": 0.0063,
      "step": 48100
    },
    {
      "epoch": 4.82,
      "grad_norm": 0.05607163533568382,
      "learning_rate": 4.210702341137124e-06,
      "loss": 0.0102,
      "step": 48200
    },
    {
      "epoch": 4.83,
      "grad_norm": 0.00012003448500763625,
      "learning_rate": 4.209030100334449e-06,
      "loss": 0.0149,
      "step": 48300
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.0828863382339478,
      "learning_rate": 4.207357859531773e-06,
      "loss": 0.012,
      "step": 48400
    },
    {
      "epoch": 4.85,
      "grad_norm": 0.1325209140777588,
      "learning_rate": 4.2056856187290975e-06,
      "loss": 0.0247,
      "step": 48500
    },
    {
      "epoch": 4.86,
      "grad_norm": 0.01746418885886669,
      "learning_rate": 4.2040133779264216e-06,
      "loss": 0.0053,
      "step": 48600
    },
    {
      "epoch": 4.87,
      "grad_norm": 0.004969676490873098,
      "learning_rate": 4.202341137123746e-06,
      "loss": 0.0059,
      "step": 48700
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.00017602667503524572,
      "learning_rate": 4.2006688963210705e-06,
      "loss": 0.0169,
      "step": 48800
    },
    {
      "epoch": 4.89,
      "grad_norm": 0.0011687574442476034,
      "learning_rate": 4.1989966555183945e-06,
      "loss": 0.0098,
      "step": 48900
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.008183939382433891,
      "learning_rate": 4.197324414715719e-06,
      "loss": 0.01,
      "step": 49000
    },
    {
      "epoch": 4.91,
      "grad_norm": 0.0027384813874959946,
      "learning_rate": 4.195652173913044e-06,
      "loss": 0.0166,
      "step": 49100
    },
    {
      "epoch": 4.92,
      "grad_norm": 0.0357474759221077,
      "learning_rate": 4.193979933110368e-06,
      "loss": 0.0113,
      "step": 49200
    },
    {
      "epoch": 4.93,
      "grad_norm": 0.030190743505954742,
      "learning_rate": 4.192307692307692e-06,
      "loss": 0.0203,
      "step": 49300
    },
    {
      "epoch": 4.9399999999999995,
      "grad_norm": 0.026387318968772888,
      "learning_rate": 4.190635451505016e-06,
      "loss": 0.0165,
      "step": 49400
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.0477469228208065,
      "learning_rate": 4.188963210702341e-06,
      "loss": 0.0038,
      "step": 49500
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.009463177062571049,
      "learning_rate": 4.187290969899666e-06,
      "loss": 0.0194,
      "step": 49600
    },
    {
      "epoch": 4.97,
      "grad_norm": 0.001094246399588883,
      "learning_rate": 4.18561872909699e-06,
      "loss": 0.0016,
      "step": 49700
    },
    {
      "epoch": 4.98,
      "grad_norm": 0.0002311818243470043,
      "learning_rate": 4.183946488294315e-06,
      "loss": 0.0061,
      "step": 49800
    },
    {
      "epoch": 4.99,
      "grad_norm": 0.0019176400965079665,
      "learning_rate": 4.182274247491639e-06,
      "loss": 0.0117,
      "step": 49900
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.00022912977146916091,
      "learning_rate": 4.180602006688963e-06,
      "loss": 0.0109,
      "step": 50000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.985925,
      "eval_f1": 0.985925,
      "eval_loss": 0.10268224030733109,
      "eval_runtime": 137.6245,
      "eval_samples_per_second": 290.646,
      "eval_steps_per_second": 290.646,
      "step": 50000
    },
    {
      "epoch": 5.01,
      "grad_norm": 8.346785034518689e-05,
      "learning_rate": 4.178929765886288e-06,
      "loss": 0.0015,
      "step": 50100
    },
    {
      "epoch": 5.02,
      "grad_norm": 0.000289654330117628,
      "learning_rate": 4.177257525083612e-06,
      "loss": 0.0,
      "step": 50200
    },
    {
      "epoch": 5.03,
      "grad_norm": 0.0004644142754841596,
      "learning_rate": 4.175585284280937e-06,
      "loss": 0.0089,
      "step": 50300
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.0007376486901193857,
      "learning_rate": 4.173913043478261e-06,
      "loss": 0.0098,
      "step": 50400
    },
    {
      "epoch": 5.05,
      "grad_norm": 0.00011336214083712548,
      "learning_rate": 4.172240802675586e-06,
      "loss": 0.0076,
      "step": 50500
    },
    {
      "epoch": 5.06,
      "grad_norm": 0.002228671684861183,
      "learning_rate": 4.17056856187291e-06,
      "loss": 0.0089,
      "step": 50600
    },
    {
      "epoch": 5.07,
      "grad_norm": 0.005318440962582827,
      "learning_rate": 4.168896321070234e-06,
      "loss": 0.0016,
      "step": 50700
    },
    {
      "epoch": 5.08,
      "grad_norm": 0.013958403840661049,
      "learning_rate": 4.167224080267559e-06,
      "loss": 0.0041,
      "step": 50800
    },
    {
      "epoch": 5.09,
      "grad_norm": 0.00047520818770863116,
      "learning_rate": 4.165551839464883e-06,
      "loss": 0.0043,
      "step": 50900
    },
    {
      "epoch": 5.1,
      "grad_norm": 27.530658721923828,
      "learning_rate": 4.163879598662208e-06,
      "loss": 0.0121,
      "step": 51000
    },
    {
      "epoch": 5.11,
      "grad_norm": 0.26363804936408997,
      "learning_rate": 4.162207357859533e-06,
      "loss": 0.0086,
      "step": 51100
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.0017211829544976354,
      "learning_rate": 4.160535117056857e-06,
      "loss": 0.0134,
      "step": 51200
    },
    {
      "epoch": 5.13,
      "grad_norm": 0.0003323334385640919,
      "learning_rate": 4.158862876254181e-06,
      "loss": 0.0161,
      "step": 51300
    },
    {
      "epoch": 5.14,
      "grad_norm": 0.023742739111185074,
      "learning_rate": 4.157190635451505e-06,
      "loss": 0.0076,
      "step": 51400
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.0006456200499087572,
      "learning_rate": 4.15551839464883e-06,
      "loss": 0.0158,
      "step": 51500
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.23926404118537903,
      "learning_rate": 4.1538461538461545e-06,
      "loss": 0.012,
      "step": 51600
    },
    {
      "epoch": 5.17,
      "grad_norm": 0.41261881589889526,
      "learning_rate": 4.1521739130434786e-06,
      "loss": 0.0002,
      "step": 51700
    },
    {
      "epoch": 5.18,
      "grad_norm": 0.022988926619291306,
      "learning_rate": 4.1505016722408035e-06,
      "loss": 0.0173,
      "step": 51800
    },
    {
      "epoch": 5.19,
      "grad_norm": 0.00015973170229699463,
      "learning_rate": 4.1488294314381275e-06,
      "loss": 0.0019,
      "step": 51900
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.00015827955212444067,
      "learning_rate": 4.1471571906354515e-06,
      "loss": 0.014,
      "step": 52000
    },
    {
      "epoch": 5.21,
      "grad_norm": 0.0023194041568785906,
      "learning_rate": 4.145484949832776e-06,
      "loss": 0.004,
      "step": 52100
    },
    {
      "epoch": 5.22,
      "grad_norm": 0.0071893418207764626,
      "learning_rate": 4.1438127090301005e-06,
      "loss": 0.0046,
      "step": 52200
    },
    {
      "epoch": 5.23,
      "grad_norm": 0.0016534804599359632,
      "learning_rate": 4.142140468227425e-06,
      "loss": 0.0043,
      "step": 52300
    },
    {
      "epoch": 5.24,
      "grad_norm": 0.0004097437486052513,
      "learning_rate": 4.140468227424749e-06,
      "loss": 0.0213,
      "step": 52400
    },
    {
      "epoch": 5.25,
      "grad_norm": 0.004914246965199709,
      "learning_rate": 4.138795986622074e-06,
      "loss": 0.0075,
      "step": 52500
    },
    {
      "epoch": 5.26,
      "grad_norm": 0.0005830037407577038,
      "learning_rate": 4.137123745819398e-06,
      "loss": 0.0116,
      "step": 52600
    },
    {
      "epoch": 5.27,
      "grad_norm": 0.0004236988606862724,
      "learning_rate": 4.135451505016722e-06,
      "loss": 0.0054,
      "step": 52700
    },
    {
      "epoch": 5.28,
      "grad_norm": 0.05537344887852669,
      "learning_rate": 4.133779264214047e-06,
      "loss": 0.0069,
      "step": 52800
    },
    {
      "epoch": 5.29,
      "grad_norm": 5.937612513662316e-05,
      "learning_rate": 4.132107023411371e-06,
      "loss": 0.0053,
      "step": 52900
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.04086212441325188,
      "learning_rate": 4.130434782608696e-06,
      "loss": 0.006,
      "step": 53000
    },
    {
      "epoch": 5.31,
      "grad_norm": 6.80720477248542e-05,
      "learning_rate": 4.128762541806021e-06,
      "loss": 0.0087,
      "step": 53100
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.11411894112825394,
      "learning_rate": 4.127090301003345e-06,
      "loss": 0.0055,
      "step": 53200
    },
    {
      "epoch": 5.33,
      "grad_norm": 0.0021275030449032784,
      "learning_rate": 4.125418060200669e-06,
      "loss": 0.0112,
      "step": 53300
    },
    {
      "epoch": 5.34,
      "grad_norm": 0.0030677011236548424,
      "learning_rate": 4.123745819397993e-06,
      "loss": 0.0049,
      "step": 53400
    },
    {
      "epoch": 5.35,
      "grad_norm": 0.0002973228693008423,
      "learning_rate": 4.122073578595318e-06,
      "loss": 0.0083,
      "step": 53500
    },
    {
      "epoch": 5.36,
      "grad_norm": 0.0001635253574931994,
      "learning_rate": 4.120401337792643e-06,
      "loss": 0.0115,
      "step": 53600
    },
    {
      "epoch": 5.37,
      "grad_norm": 96.96515655517578,
      "learning_rate": 4.118729096989967e-06,
      "loss": 0.0046,
      "step": 53700
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.007858989760279655,
      "learning_rate": 4.117056856187292e-06,
      "loss": 0.0127,
      "step": 53800
    },
    {
      "epoch": 5.39,
      "grad_norm": 0.0028765483293682337,
      "learning_rate": 4.115384615384616e-06,
      "loss": 0.0014,
      "step": 53900
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.0008154021925292909,
      "learning_rate": 4.11371237458194e-06,
      "loss": 0.0077,
      "step": 54000
    },
    {
      "epoch": 5.41,
      "grad_norm": 0.001322218799032271,
      "learning_rate": 4.112040133779265e-06,
      "loss": 0.0056,
      "step": 54100
    },
    {
      "epoch": 5.42,
      "grad_norm": 0.023289771750569344,
      "learning_rate": 4.110367892976589e-06,
      "loss": 0.0156,
      "step": 54200
    },
    {
      "epoch": 5.43,
      "grad_norm": 0.005015732254832983,
      "learning_rate": 4.108695652173914e-06,
      "loss": 0.0099,
      "step": 54300
    },
    {
      "epoch": 5.44,
      "grad_norm": 7.15251371730119e-05,
      "learning_rate": 4.107023411371238e-06,
      "loss": 0.0184,
      "step": 54400
    },
    {
      "epoch": 5.45,
      "grad_norm": 0.00010623384878272191,
      "learning_rate": 4.105351170568563e-06,
      "loss": 0.0032,
      "step": 54500
    },
    {
      "epoch": 5.46,
      "grad_norm": 0.0029225721955299377,
      "learning_rate": 4.103678929765887e-06,
      "loss": 0.0112,
      "step": 54600
    },
    {
      "epoch": 5.47,
      "grad_norm": 0.000662469130475074,
      "learning_rate": 4.102006688963211e-06,
      "loss": 0.0089,
      "step": 54700
    },
    {
      "epoch": 5.48,
      "grad_norm": 0.0004447357205208391,
      "learning_rate": 4.100334448160536e-06,
      "loss": 0.0095,
      "step": 54800
    },
    {
      "epoch": 5.49,
      "grad_norm": 0.0009589354740455747,
      "learning_rate": 4.09866220735786e-06,
      "loss": 0.0151,
      "step": 54900
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.002127012936398387,
      "learning_rate": 4.0969899665551845e-06,
      "loss": 0.0075,
      "step": 55000
    },
    {
      "epoch": 5.5,
      "eval_accuracy": 0.98495,
      "eval_f1": 0.98495,
      "eval_loss": 0.11519099026918411,
      "eval_runtime": 135.3873,
      "eval_samples_per_second": 295.449,
      "eval_steps_per_second": 295.449,
      "step": 55000
    },
    {
      "epoch": 5.51,
      "grad_norm": 0.00014911459584254771,
      "learning_rate": 4.095317725752509e-06,
      "loss": 0.0086,
      "step": 55100
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.00037860465818084776,
      "learning_rate": 4.093645484949833e-06,
      "loss": 0.0072,
      "step": 55200
    },
    {
      "epoch": 5.53,
      "grad_norm": 9.189704724121839e-05,
      "learning_rate": 4.0919732441471575e-06,
      "loss": 0.0122,
      "step": 55300
    },
    {
      "epoch": 5.54,
      "grad_norm": 0.0012258023489266634,
      "learning_rate": 4.0903010033444815e-06,
      "loss": 0.0046,
      "step": 55400
    },
    {
      "epoch": 5.55,
      "grad_norm": 0.0069747380912303925,
      "learning_rate": 4.088628762541806e-06,
      "loss": 0.0097,
      "step": 55500
    },
    {
      "epoch": 5.5600000000000005,
      "grad_norm": 0.00017239715089090168,
      "learning_rate": 4.086956521739131e-06,
      "loss": 0.0009,
      "step": 55600
    },
    {
      "epoch": 5.57,
      "grad_norm": 0.0012025295291095972,
      "learning_rate": 4.085284280936455e-06,
      "loss": 0.0111,
      "step": 55700
    },
    {
      "epoch": 5.58,
      "grad_norm": 0.07929231971502304,
      "learning_rate": 4.08361204013378e-06,
      "loss": 0.0067,
      "step": 55800
    },
    {
      "epoch": 5.59,
      "grad_norm": 0.0003187973634339869,
      "learning_rate": 4.081939799331104e-06,
      "loss": 0.0157,
      "step": 55900
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.0013896484160795808,
      "learning_rate": 4.080267558528428e-06,
      "loss": 0.0059,
      "step": 56000
    },
    {
      "epoch": 5.61,
      "grad_norm": 7.541276863776147e-05,
      "learning_rate": 4.078595317725752e-06,
      "loss": 0.0083,
      "step": 56100
    },
    {
      "epoch": 5.62,
      "grad_norm": 0.005246707238256931,
      "learning_rate": 4.076923076923077e-06,
      "loss": 0.0177,
      "step": 56200
    },
    {
      "epoch": 5.63,
      "grad_norm": 0.0028932797722518444,
      "learning_rate": 4.075250836120402e-06,
      "loss": 0.0053,
      "step": 56300
    },
    {
      "epoch": 5.64,
      "grad_norm": 0.0031544279772788286,
      "learning_rate": 4.073578595317726e-06,
      "loss": 0.0008,
      "step": 56400
    },
    {
      "epoch": 5.65,
      "grad_norm": 5.287326712277718e-05,
      "learning_rate": 4.071906354515051e-06,
      "loss": 0.0048,
      "step": 56500
    },
    {
      "epoch": 5.66,
      "grad_norm": 6.959833262953907e-05,
      "learning_rate": 4.070234113712375e-06,
      "loss": 0.0064,
      "step": 56600
    },
    {
      "epoch": 5.67,
      "grad_norm": 0.002718166681006551,
      "learning_rate": 4.068561872909699e-06,
      "loss": 0.0151,
      "step": 56700
    },
    {
      "epoch": 5.68,
      "grad_norm": 3.8481392860412598,
      "learning_rate": 4.066889632107024e-06,
      "loss": 0.0099,
      "step": 56800
    },
    {
      "epoch": 5.6899999999999995,
      "grad_norm": 0.00023723083722870797,
      "learning_rate": 4.065217391304348e-06,
      "loss": 0.0001,
      "step": 56900
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.003258422249928117,
      "learning_rate": 4.063545150501673e-06,
      "loss": 0.0088,
      "step": 57000
    },
    {
      "epoch": 5.71,
      "grad_norm": 8.750943379709497e-05,
      "learning_rate": 4.061872909698997e-06,
      "loss": 0.0059,
      "step": 57100
    },
    {
      "epoch": 5.72,
      "grad_norm": 0.04691299796104431,
      "learning_rate": 4.060200668896322e-06,
      "loss": 0.0164,
      "step": 57200
    },
    {
      "epoch": 5.73,
      "grad_norm": 0.002486943267285824,
      "learning_rate": 4.058528428093646e-06,
      "loss": 0.021,
      "step": 57300
    },
    {
      "epoch": 5.74,
      "grad_norm": 0.00018505608022678643,
      "learning_rate": 4.05685618729097e-06,
      "loss": 0.004,
      "step": 57400
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.0130680613219738,
      "learning_rate": 4.055183946488295e-06,
      "loss": 0.0038,
      "step": 57500
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.002892815973609686,
      "learning_rate": 4.053511705685619e-06,
      "loss": 0.0094,
      "step": 57600
    },
    {
      "epoch": 5.77,
      "grad_norm": 0.0004248152836225927,
      "learning_rate": 4.051839464882944e-06,
      "loss": 0.017,
      "step": 57700
    },
    {
      "epoch": 5.78,
      "grad_norm": 0.024957746267318726,
      "learning_rate": 4.050167224080268e-06,
      "loss": 0.009,
      "step": 57800
    },
    {
      "epoch": 5.79,
      "grad_norm": 5.579432036029175e-05,
      "learning_rate": 4.048494983277593e-06,
      "loss": 0.0113,
      "step": 57900
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.0788150206208229,
      "learning_rate": 4.046822742474917e-06,
      "loss": 0.0161,
      "step": 58000
    },
    {
      "epoch": 5.8100000000000005,
      "grad_norm": 0.00016917276661843061,
      "learning_rate": 4.045150501672241e-06,
      "loss": 0.0093,
      "step": 58100
    },
    {
      "epoch": 5.82,
      "grad_norm": 0.0006015472463332117,
      "learning_rate": 4.0434782608695655e-06,
      "loss": 0.0009,
      "step": 58200
    },
    {
      "epoch": 5.83,
      "grad_norm": 0.0022845177445560694,
      "learning_rate": 4.0418060200668904e-06,
      "loss": 0.0085,
      "step": 58300
    },
    {
      "epoch": 5.84,
      "grad_norm": 0.00025827871286310256,
      "learning_rate": 4.0401337792642145e-06,
      "loss": 0.0075,
      "step": 58400
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.00020007445709779859,
      "learning_rate": 4.0384615384615385e-06,
      "loss": 0.0113,
      "step": 58500
    },
    {
      "epoch": 5.86,
      "grad_norm": 4.889999036095105e-05,
      "learning_rate": 4.036789297658863e-06,
      "loss": 0.0092,
      "step": 58600
    },
    {
      "epoch": 5.87,
      "grad_norm": 13.686702728271484,
      "learning_rate": 4.035117056856187e-06,
      "loss": 0.0077,
      "step": 58700
    },
    {
      "epoch": 5.88,
      "grad_norm": 0.0002892339834943414,
      "learning_rate": 4.033444816053512e-06,
      "loss": 0.003,
      "step": 58800
    },
    {
      "epoch": 5.89,
      "grad_norm": 0.003974799066781998,
      "learning_rate": 4.031772575250836e-06,
      "loss": 0.0088,
      "step": 58900
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.010556302964687347,
      "learning_rate": 4.030100334448161e-06,
      "loss": 0.0171,
      "step": 59000
    },
    {
      "epoch": 5.91,
      "grad_norm": 0.02879416197538376,
      "learning_rate": 4.028428093645485e-06,
      "loss": 0.0168,
      "step": 59100
    },
    {
      "epoch": 5.92,
      "grad_norm": 4.420049299369566e-05,
      "learning_rate": 4.026755852842809e-06,
      "loss": 0.0035,
      "step": 59200
    },
    {
      "epoch": 5.93,
      "grad_norm": 0.0005424703704193234,
      "learning_rate": 4.025083612040134e-06,
      "loss": 0.0105,
      "step": 59300
    },
    {
      "epoch": 5.9399999999999995,
      "grad_norm": 0.07232003659009933,
      "learning_rate": 4.023411371237458e-06,
      "loss": 0.0087,
      "step": 59400
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.00023946212604641914,
      "learning_rate": 4.021739130434783e-06,
      "loss": 0.0001,
      "step": 59500
    },
    {
      "epoch": 5.96,
      "grad_norm": 0.009542220272123814,
      "learning_rate": 4.020066889632107e-06,
      "loss": 0.0151,
      "step": 59600
    },
    {
      "epoch": 5.97,
      "grad_norm": 0.0009311602334491909,
      "learning_rate": 4.018394648829432e-06,
      "loss": 0.0074,
      "step": 59700
    },
    {
      "epoch": 5.98,
      "grad_norm": 0.0011114083463326097,
      "learning_rate": 4.016722408026756e-06,
      "loss": 0.0108,
      "step": 59800
    },
    {
      "epoch": 5.99,
      "grad_norm": 0.0005047828890383244,
      "learning_rate": 4.01505016722408e-06,
      "loss": 0.0031,
      "step": 59900
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.042260460555553436,
      "learning_rate": 4.013377926421405e-06,
      "loss": 0.0278,
      "step": 60000
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.986125,
      "eval_f1": 0.986125,
      "eval_loss": 0.09157590568065643,
      "eval_runtime": 134.4638,
      "eval_samples_per_second": 297.478,
      "eval_steps_per_second": 297.478,
      "step": 60000
    },
    {
      "epoch": 6.01,
      "grad_norm": 20.1020565032959,
      "learning_rate": 4.011705685618729e-06,
      "loss": 0.0068,
      "step": 60100
    },
    {
      "epoch": 6.02,
      "grad_norm": 0.0018263482488691807,
      "learning_rate": 4.010033444816054e-06,
      "loss": 0.0064,
      "step": 60200
    },
    {
      "epoch": 6.03,
      "grad_norm": 0.28091853857040405,
      "learning_rate": 4.008361204013379e-06,
      "loss": 0.006,
      "step": 60300
    },
    {
      "epoch": 6.04,
      "grad_norm": 0.018621103838086128,
      "learning_rate": 4.006688963210703e-06,
      "loss": 0.0027,
      "step": 60400
    },
    {
      "epoch": 6.05,
      "grad_norm": 9.468239295529202e-05,
      "learning_rate": 4.005016722408027e-06,
      "loss": 0.0043,
      "step": 60500
    },
    {
      "epoch": 6.06,
      "grad_norm": 0.005317880306392908,
      "learning_rate": 4.003344481605351e-06,
      "loss": 0.0029,
      "step": 60600
    },
    {
      "epoch": 6.07,
      "grad_norm": 0.0015003898879513144,
      "learning_rate": 4.001672240802676e-06,
      "loss": 0.0107,
      "step": 60700
    },
    {
      "epoch": 6.08,
      "grad_norm": 0.6607133746147156,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0028,
      "step": 60800
    },
    {
      "epoch": 6.09,
      "grad_norm": 0.014779971912503242,
      "learning_rate": 3.998327759197325e-06,
      "loss": 0.0001,
      "step": 60900
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.0003267451247666031,
      "learning_rate": 3.99665551839465e-06,
      "loss": 0.0068,
      "step": 61000
    },
    {
      "epoch": 6.11,
      "grad_norm": 4.1315361158922315e-05,
      "learning_rate": 3.994983277591974e-06,
      "loss": 0.0016,
      "step": 61100
    },
    {
      "epoch": 6.12,
      "grad_norm": 6.84654587530531e-05,
      "learning_rate": 3.993311036789298e-06,
      "loss": 0.0061,
      "step": 61200
    },
    {
      "epoch": 6.13,
      "grad_norm": 0.000273578567430377,
      "learning_rate": 3.9916387959866225e-06,
      "loss": 0.0049,
      "step": 61300
    },
    {
      "epoch": 6.14,
      "grad_norm": 4.0936502045951784e-05,
      "learning_rate": 3.989966555183947e-06,
      "loss": 0.0069,
      "step": 61400
    },
    {
      "epoch": 6.15,
      "grad_norm": 0.0004429342516232282,
      "learning_rate": 3.9882943143812715e-06,
      "loss": 0.0094,
      "step": 61500
    },
    {
      "epoch": 6.16,
      "grad_norm": 0.00038169074105098844,
      "learning_rate": 3.9866220735785955e-06,
      "loss": 0.0048,
      "step": 61600
    },
    {
      "epoch": 6.17,
      "grad_norm": 0.10792536288499832,
      "learning_rate": 3.98494983277592e-06,
      "loss": 0.0049,
      "step": 61700
    },
    {
      "epoch": 6.18,
      "grad_norm": 4.4029828131897375e-05,
      "learning_rate": 3.9832775919732444e-06,
      "loss": 0.0001,
      "step": 61800
    },
    {
      "epoch": 6.19,
      "grad_norm": 4.2638963350327685e-05,
      "learning_rate": 3.9816053511705685e-06,
      "loss": 0.0044,
      "step": 61900
    },
    {
      "epoch": 6.2,
      "grad_norm": 3.456565536907874e-05,
      "learning_rate": 3.979933110367893e-06,
      "loss": 0.0003,
      "step": 62000
    },
    {
      "epoch": 6.21,
      "grad_norm": 0.0010728957131505013,
      "learning_rate": 3.978260869565217e-06,
      "loss": 0.0042,
      "step": 62100
    },
    {
      "epoch": 6.22,
      "grad_norm": 5.354998938855715e-05,
      "learning_rate": 3.976588628762542e-06,
      "loss": 0.0081,
      "step": 62200
    },
    {
      "epoch": 6.23,
      "grad_norm": 35.901885986328125,
      "learning_rate": 3.974916387959867e-06,
      "loss": 0.0096,
      "step": 62300
    },
    {
      "epoch": 6.24,
      "grad_norm": 0.00024061814474407583,
      "learning_rate": 3.973244147157191e-06,
      "loss": 0.0032,
      "step": 62400
    },
    {
      "epoch": 6.25,
      "grad_norm": 3.811920032603666e-05,
      "learning_rate": 3.971571906354515e-06,
      "loss": 0.0049,
      "step": 62500
    },
    {
      "epoch": 6.26,
      "grad_norm": 2.6195273399353027,
      "learning_rate": 3.969899665551839e-06,
      "loss": 0.0029,
      "step": 62600
    },
    {
      "epoch": 6.27,
      "grad_norm": 0.009201622568070889,
      "learning_rate": 3.968227424749164e-06,
      "loss": 0.0035,
      "step": 62700
    },
    {
      "epoch": 6.28,
      "grad_norm": 9.226119436789304e-05,
      "learning_rate": 3.966555183946489e-06,
      "loss": 0.0085,
      "step": 62800
    },
    {
      "epoch": 6.29,
      "grad_norm": 4.603072375175543e-05,
      "learning_rate": 3.964882943143813e-06,
      "loss": 0.0094,
      "step": 62900
    },
    {
      "epoch": 6.3,
      "grad_norm": 2.712052628339734e-05,
      "learning_rate": 3.963210702341138e-06,
      "loss": 0.0,
      "step": 63000
    },
    {
      "epoch": 6.31,
      "grad_norm": 0.00019533267186488956,
      "learning_rate": 3.961538461538462e-06,
      "loss": 0.0105,
      "step": 63100
    },
    {
      "epoch": 6.32,
      "grad_norm": 7.596238719997928e-05,
      "learning_rate": 3.959866220735786e-06,
      "loss": 0.0136,
      "step": 63200
    },
    {
      "epoch": 6.33,
      "grad_norm": 0.0005583142628893256,
      "learning_rate": 3.958193979933111e-06,
      "loss": 0.0083,
      "step": 63300
    },
    {
      "epoch": 6.34,
      "grad_norm": 0.001194102456793189,
      "learning_rate": 3.956521739130435e-06,
      "loss": 0.0098,
      "step": 63400
    },
    {
      "epoch": 6.35,
      "grad_norm": 3.787357491091825e-05,
      "learning_rate": 3.95484949832776e-06,
      "loss": 0.0036,
      "step": 63500
    },
    {
      "epoch": 6.36,
      "grad_norm": 7.270895730471238e-05,
      "learning_rate": 3.953177257525084e-06,
      "loss": 0.0067,
      "step": 63600
    },
    {
      "epoch": 6.37,
      "grad_norm": 0.000623897067271173,
      "learning_rate": 3.951505016722409e-06,
      "loss": 0.0004,
      "step": 63700
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.0003589978732634336,
      "learning_rate": 3.949832775919733e-06,
      "loss": 0.0042,
      "step": 63800
    },
    {
      "epoch": 6.39,
      "grad_norm": 0.0001246546598849818,
      "learning_rate": 3.948160535117057e-06,
      "loss": 0.011,
      "step": 63900
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.008183490484952927,
      "learning_rate": 3.946488294314382e-06,
      "loss": 0.0064,
      "step": 64000
    },
    {
      "epoch": 6.41,
      "grad_norm": 0.05226984620094299,
      "learning_rate": 3.944816053511706e-06,
      "loss": 0.0125,
      "step": 64100
    },
    {
      "epoch": 6.42,
      "grad_norm": 7.40212417440489e-05,
      "learning_rate": 3.943143812709031e-06,
      "loss": 0.0059,
      "step": 64200
    },
    {
      "epoch": 6.43,
      "grad_norm": 0.00734880194067955,
      "learning_rate": 3.9414715719063555e-06,
      "loss": 0.0001,
      "step": 64300
    },
    {
      "epoch": 6.44,
      "grad_norm": 4.586549403029494e-05,
      "learning_rate": 3.9397993311036795e-06,
      "loss": 0.0056,
      "step": 64400
    },
    {
      "epoch": 6.45,
      "grad_norm": 3.2639989512972534e-05,
      "learning_rate": 3.938127090301004e-06,
      "loss": 0.0103,
      "step": 64500
    },
    {
      "epoch": 6.46,
      "grad_norm": 0.004328950308263302,
      "learning_rate": 3.936454849498328e-06,
      "loss": 0.0163,
      "step": 64600
    },
    {
      "epoch": 6.47,
      "grad_norm": 3.590931373764761e-05,
      "learning_rate": 3.9347826086956525e-06,
      "loss": 0.0048,
      "step": 64700
    },
    {
      "epoch": 6.48,
      "grad_norm": 0.0038940191734582186,
      "learning_rate": 3.933110367892977e-06,
      "loss": 0.0091,
      "step": 64800
    },
    {
      "epoch": 6.49,
      "grad_norm": 0.00018863222794607282,
      "learning_rate": 3.9314381270903014e-06,
      "loss": 0.0114,
      "step": 64900
    },
    {
      "epoch": 6.5,
      "grad_norm": 5.312537177815102e-05,
      "learning_rate": 3.929765886287626e-06,
      "loss": 0.0023,
      "step": 65000
    },
    {
      "epoch": 6.5,
      "eval_accuracy": 0.985975,
      "eval_f1": 0.985975,
      "eval_loss": 0.11558345705270767,
      "eval_runtime": 133.3886,
      "eval_samples_per_second": 299.876,
      "eval_steps_per_second": 299.876,
      "step": 65000
    },
    {
      "epoch": 6.51,
      "grad_norm": 0.0002164523903047666,
      "learning_rate": 3.92809364548495e-06,
      "loss": 0.0073,
      "step": 65100
    },
    {
      "epoch": 6.52,
      "grad_norm": 0.00015249662101268768,
      "learning_rate": 3.926421404682274e-06,
      "loss": 0.0034,
      "step": 65200
    },
    {
      "epoch": 6.53,
      "grad_norm": 0.0001434876030543819,
      "learning_rate": 3.924749163879599e-06,
      "loss": 0.0081,
      "step": 65300
    },
    {
      "epoch": 6.54,
      "grad_norm": 2.771373874566052e-05,
      "learning_rate": 3.923076923076923e-06,
      "loss": 0.0087,
      "step": 65400
    },
    {
      "epoch": 6.55,
      "grad_norm": 3.081748218392022e-05,
      "learning_rate": 3.921404682274248e-06,
      "loss": 0.0086,
      "step": 65500
    },
    {
      "epoch": 6.5600000000000005,
      "grad_norm": 3.04588920698734e-05,
      "learning_rate": 3.919732441471572e-06,
      "loss": 0.0065,
      "step": 65600
    },
    {
      "epoch": 6.57,
      "grad_norm": 0.00013289268827065825,
      "learning_rate": 3.918060200668897e-06,
      "loss": 0.0029,
      "step": 65700
    },
    {
      "epoch": 6.58,
      "grad_norm": 0.0002527164469938725,
      "learning_rate": 3.916387959866221e-06,
      "loss": 0.0096,
      "step": 65800
    },
    {
      "epoch": 6.59,
      "grad_norm": 0.006434381008148193,
      "learning_rate": 3.914715719063545e-06,
      "loss": 0.0066,
      "step": 65900
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.001144530950114131,
      "learning_rate": 3.91304347826087e-06,
      "loss": 0.0038,
      "step": 66000
    },
    {
      "epoch": 6.61,
      "grad_norm": 0.030533278360962868,
      "learning_rate": 3.911371237458194e-06,
      "loss": 0.0033,
      "step": 66100
    },
    {
      "epoch": 6.62,
      "grad_norm": 9.253934695152566e-05,
      "learning_rate": 3.909698996655519e-06,
      "loss": 0.0056,
      "step": 66200
    },
    {
      "epoch": 6.63,
      "grad_norm": 0.0017731015104800463,
      "learning_rate": 3.908026755852843e-06,
      "loss": 0.0,
      "step": 66300
    },
    {
      "epoch": 6.64,
      "grad_norm": 6.34994066786021e-05,
      "learning_rate": 3.906354515050168e-06,
      "loss": 0.0115,
      "step": 66400
    },
    {
      "epoch": 6.65,
      "grad_norm": 3.8429308915510774e-05,
      "learning_rate": 3.904682274247492e-06,
      "loss": 0.0097,
      "step": 66500
    },
    {
      "epoch": 6.66,
      "grad_norm": 0.0302902702242136,
      "learning_rate": 3.903010033444816e-06,
      "loss": 0.0115,
      "step": 66600
    },
    {
      "epoch": 6.67,
      "grad_norm": 0.00047553866170346737,
      "learning_rate": 3.901337792642141e-06,
      "loss": 0.0061,
      "step": 66700
    },
    {
      "epoch": 6.68,
      "grad_norm": 3.929653757950291e-05,
      "learning_rate": 3.899665551839465e-06,
      "loss": 0.011,
      "step": 66800
    },
    {
      "epoch": 6.6899999999999995,
      "grad_norm": 0.041665125638246536,
      "learning_rate": 3.89799331103679e-06,
      "loss": 0.0051,
      "step": 66900
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.014791757799685001,
      "learning_rate": 3.896321070234114e-06,
      "loss": 0.0144,
      "step": 67000
    },
    {
      "epoch": 6.71,
      "grad_norm": 0.0002211692917626351,
      "learning_rate": 3.894648829431439e-06,
      "loss": 0.0057,
      "step": 67100
    },
    {
      "epoch": 6.72,
      "grad_norm": 3.8692807720508426e-05,
      "learning_rate": 3.892976588628763e-06,
      "loss": 0.0115,
      "step": 67200
    },
    {
      "epoch": 6.73,
      "grad_norm": 0.005168182775378227,
      "learning_rate": 3.891304347826087e-06,
      "loss": 0.0051,
      "step": 67300
    },
    {
      "epoch": 6.74,
      "grad_norm": 0.00041154478094540536,
      "learning_rate": 3.889632107023412e-06,
      "loss": 0.0001,
      "step": 67400
    },
    {
      "epoch": 6.75,
      "grad_norm": 5.148116542841308e-05,
      "learning_rate": 3.8879598662207366e-06,
      "loss": 0.0069,
      "step": 67500
    },
    {
      "epoch": 6.76,
      "grad_norm": 0.0013762086164206266,
      "learning_rate": 3.886287625418061e-06,
      "loss": 0.0144,
      "step": 67600
    },
    {
      "epoch": 6.77,
      "grad_norm": 4.9009013309841976e-05,
      "learning_rate": 3.884615384615385e-06,
      "loss": 0.0,
      "step": 67700
    },
    {
      "epoch": 6.78,
      "grad_norm": 3.336561348987743e-05,
      "learning_rate": 3.8829431438127095e-06,
      "loss": 0.0047,
      "step": 67800
    },
    {
      "epoch": 6.79,
      "grad_norm": 0.0003458741120994091,
      "learning_rate": 3.8812709030100335e-06,
      "loss": 0.0098,
      "step": 67900
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.001628655707463622,
      "learning_rate": 3.8795986622073584e-06,
      "loss": 0.0073,
      "step": 68000
    },
    {
      "epoch": 6.8100000000000005,
      "grad_norm": 5.698153836419806e-05,
      "learning_rate": 3.8779264214046825e-06,
      "loss": 0.005,
      "step": 68100
    },
    {
      "epoch": 6.82,
      "grad_norm": 0.0005441823159344494,
      "learning_rate": 3.876254180602007e-06,
      "loss": 0.0006,
      "step": 68200
    },
    {
      "epoch": 6.83,
      "grad_norm": 0.007364595774561167,
      "learning_rate": 3.874581939799331e-06,
      "loss": 0.0014,
      "step": 68300
    },
    {
      "epoch": 6.84,
      "grad_norm": 0.004882550798356533,
      "learning_rate": 3.8729096989966554e-06,
      "loss": 0.0284,
      "step": 68400
    },
    {
      "epoch": 6.85,
      "grad_norm": 0.1382811814546585,
      "learning_rate": 3.87123745819398e-06,
      "loss": 0.0095,
      "step": 68500
    },
    {
      "epoch": 6.86,
      "grad_norm": 0.003247621236369014,
      "learning_rate": 3.869565217391304e-06,
      "loss": 0.0126,
      "step": 68600
    },
    {
      "epoch": 6.87,
      "grad_norm": 0.0005879114614799619,
      "learning_rate": 3.867892976588629e-06,
      "loss": 0.0073,
      "step": 68700
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.00033930817153304815,
      "learning_rate": 3.866220735785953e-06,
      "loss": 0.0001,
      "step": 68800
    },
    {
      "epoch": 6.89,
      "grad_norm": 0.0021851826459169388,
      "learning_rate": 3.864548494983278e-06,
      "loss": 0.0097,
      "step": 68900
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.00033705090754665434,
      "learning_rate": 3.862876254180602e-06,
      "loss": 0.0038,
      "step": 69000
    },
    {
      "epoch": 6.91,
      "grad_norm": 3.90254681406077e-05,
      "learning_rate": 3.861204013377926e-06,
      "loss": 0.0047,
      "step": 69100
    },
    {
      "epoch": 6.92,
      "grad_norm": 0.06699187308549881,
      "learning_rate": 3.859531772575251e-06,
      "loss": 0.0001,
      "step": 69200
    },
    {
      "epoch": 6.93,
      "grad_norm": 2.483546813891735e-05,
      "learning_rate": 3.857859531772575e-06,
      "loss": 0.0046,
      "step": 69300
    },
    {
      "epoch": 6.9399999999999995,
      "grad_norm": 6.92076500854455e-05,
      "learning_rate": 3.8561872909699e-06,
      "loss": 0.0189,
      "step": 69400
    },
    {
      "epoch": 6.95,
      "grad_norm": 7.609539898112416e-05,
      "learning_rate": 3.854515050167225e-06,
      "loss": 0.0055,
      "step": 69500
    },
    {
      "epoch": 6.96,
      "grad_norm": 2.9781913326587528e-05,
      "learning_rate": 3.852842809364549e-06,
      "loss": 0.0109,
      "step": 69600
    },
    {
      "epoch": 6.97,
      "grad_norm": 9.487575880484655e-05,
      "learning_rate": 3.851170568561873e-06,
      "loss": 0.0001,
      "step": 69700
    },
    {
      "epoch": 6.98,
      "grad_norm": 0.03806305304169655,
      "learning_rate": 3.849498327759197e-06,
      "loss": 0.0115,
      "step": 69800
    },
    {
      "epoch": 6.99,
      "grad_norm": 0.0004042502259835601,
      "learning_rate": 3.847826086956522e-06,
      "loss": 0.0135,
      "step": 69900
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.047401949763298035,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.0115,
      "step": 70000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.986475,
      "eval_f1": 0.986475,
      "eval_loss": 0.09723663330078125,
      "eval_runtime": 133.6436,
      "eval_samples_per_second": 299.304,
      "eval_steps_per_second": 299.304,
      "step": 70000
    },
    {
      "epoch": 7.01,
      "grad_norm": 5.497029997059144e-05,
      "learning_rate": 3.844481605351171e-06,
      "loss": 0.0001,
      "step": 70100
    },
    {
      "epoch": 7.02,
      "grad_norm": 0.04041426256299019,
      "learning_rate": 3.842809364548496e-06,
      "loss": 0.0001,
      "step": 70200
    },
    {
      "epoch": 7.03,
      "grad_norm": 0.000893338059540838,
      "learning_rate": 3.84113712374582e-06,
      "loss": 0.0007,
      "step": 70300
    },
    {
      "epoch": 7.04,
      "grad_norm": 3.694546830956824e-05,
      "learning_rate": 3.839464882943144e-06,
      "loss": 0.0008,
      "step": 70400
    },
    {
      "epoch": 7.05,
      "grad_norm": 0.0002109079505316913,
      "learning_rate": 3.837792642140469e-06,
      "loss": 0.0098,
      "step": 70500
    },
    {
      "epoch": 7.06,
      "grad_norm": 0.000186313918675296,
      "learning_rate": 3.836120401337793e-06,
      "loss": 0.0,
      "step": 70600
    },
    {
      "epoch": 7.07,
      "grad_norm": 0.03715241327881813,
      "learning_rate": 3.834448160535118e-06,
      "loss": 0.009,
      "step": 70700
    },
    {
      "epoch": 7.08,
      "grad_norm": 4.64997865492478e-05,
      "learning_rate": 3.832775919732442e-06,
      "loss": 0.0061,
      "step": 70800
    },
    {
      "epoch": 7.09,
      "grad_norm": 4.012215504189953e-05,
      "learning_rate": 3.8311036789297665e-06,
      "loss": 0.0002,
      "step": 70900
    },
    {
      "epoch": 7.1,
      "grad_norm": 3.099231980741024e-05,
      "learning_rate": 3.8294314381270906e-06,
      "loss": 0.0072,
      "step": 71000
    },
    {
      "epoch": 7.11,
      "grad_norm": 3.7772646464873105e-05,
      "learning_rate": 3.827759197324415e-06,
      "loss": 0.0093,
      "step": 71100
    },
    {
      "epoch": 7.12,
      "grad_norm": 7.007938256720081e-05,
      "learning_rate": 3.8260869565217395e-06,
      "loss": 0.008,
      "step": 71200
    },
    {
      "epoch": 7.13,
      "grad_norm": 8.04358787718229e-05,
      "learning_rate": 3.8244147157190635e-06,
      "loss": 0.0003,
      "step": 71300
    },
    {
      "epoch": 7.14,
      "grad_norm": 0.009740136563777924,
      "learning_rate": 3.822742474916388e-06,
      "loss": 0.0034,
      "step": 71400
    },
    {
      "epoch": 7.15,
      "grad_norm": 4.2379640945000574e-05,
      "learning_rate": 3.821070234113713e-06,
      "loss": 0.003,
      "step": 71500
    },
    {
      "epoch": 7.16,
      "grad_norm": 0.001223787316121161,
      "learning_rate": 3.819397993311037e-06,
      "loss": 0.0,
      "step": 71600
    },
    {
      "epoch": 7.17,
      "grad_norm": 3.370898775756359e-05,
      "learning_rate": 3.817725752508361e-06,
      "loss": 0.0,
      "step": 71700
    },
    {
      "epoch": 7.18,
      "grad_norm": 0.0017037318320944905,
      "learning_rate": 3.816053511705685e-06,
      "loss": 0.012,
      "step": 71800
    },
    {
      "epoch": 7.19,
      "grad_norm": 0.1010851189494133,
      "learning_rate": 3.8143812709030103e-06,
      "loss": 0.0086,
      "step": 71900
    },
    {
      "epoch": 7.2,
      "grad_norm": 3.95359966205433e-05,
      "learning_rate": 3.812709030100335e-06,
      "loss": 0.0053,
      "step": 72000
    },
    {
      "epoch": 7.21,
      "grad_norm": 0.03214002400636673,
      "learning_rate": 3.811036789297659e-06,
      "loss": 0.0,
      "step": 72100
    },
    {
      "epoch": 7.22,
      "grad_norm": 9.369623876409605e-05,
      "learning_rate": 3.8093645484949837e-06,
      "loss": 0.0018,
      "step": 72200
    },
    {
      "epoch": 7.23,
      "grad_norm": 4.141992030781694e-05,
      "learning_rate": 3.8076923076923077e-06,
      "loss": 0.0078,
      "step": 72300
    },
    {
      "epoch": 7.24,
      "grad_norm": 0.006547427736222744,
      "learning_rate": 3.8060200668896326e-06,
      "loss": 0.0043,
      "step": 72400
    },
    {
      "epoch": 7.25,
      "grad_norm": 0.009587573818862438,
      "learning_rate": 3.804347826086957e-06,
      "loss": 0.0,
      "step": 72500
    },
    {
      "epoch": 7.26,
      "grad_norm": 0.0001476012112107128,
      "learning_rate": 3.802675585284281e-06,
      "loss": 0.0156,
      "step": 72600
    },
    {
      "epoch": 7.27,
      "grad_norm": 0.00019884717767126858,
      "learning_rate": 3.801003344481606e-06,
      "loss": 0.0,
      "step": 72700
    },
    {
      "epoch": 7.28,
      "grad_norm": 0.0004751971864607185,
      "learning_rate": 3.79933110367893e-06,
      "loss": 0.0,
      "step": 72800
    },
    {
      "epoch": 7.29,
      "grad_norm": 0.00021417044627014548,
      "learning_rate": 3.7976588628762545e-06,
      "loss": 0.0,
      "step": 72900
    },
    {
      "epoch": 7.3,
      "grad_norm": 2.5816289053182118e-05,
      "learning_rate": 3.7959866220735793e-06,
      "loss": 0.0051,
      "step": 73000
    },
    {
      "epoch": 7.31,
      "grad_norm": 3.802667197305709e-05,
      "learning_rate": 3.7943143812709034e-06,
      "loss": 0.0079,
      "step": 73100
    },
    {
      "epoch": 7.32,
      "grad_norm": 7.326734339585528e-05,
      "learning_rate": 3.792642140468228e-06,
      "loss": 0.0001,
      "step": 73200
    },
    {
      "epoch": 7.33,
      "grad_norm": 2.148441126337275e-05,
      "learning_rate": 3.790969899665552e-06,
      "loss": 0.0097,
      "step": 73300
    },
    {
      "epoch": 7.34,
      "grad_norm": 0.000500816386193037,
      "learning_rate": 3.7892976588628768e-06,
      "loss": 0.0036,
      "step": 73400
    },
    {
      "epoch": 7.35,
      "grad_norm": 2.385185143793933e-05,
      "learning_rate": 3.7876254180602012e-06,
      "loss": 0.0039,
      "step": 73500
    },
    {
      "epoch": 7.36,
      "grad_norm": 2.7678737751557492e-05,
      "learning_rate": 3.7859531772575253e-06,
      "loss": 0.0023,
      "step": 73600
    },
    {
      "epoch": 7.37,
      "grad_norm": 0.01611788012087345,
      "learning_rate": 3.78428093645485e-06,
      "loss": 0.0042,
      "step": 73700
    },
    {
      "epoch": 7.38,
      "grad_norm": 2.1817544620716944e-05,
      "learning_rate": 3.782608695652174e-06,
      "loss": 0.0033,
      "step": 73800
    },
    {
      "epoch": 7.39,
      "grad_norm": 2.7087255148217082e-05,
      "learning_rate": 3.7809364548494986e-06,
      "loss": 0.0071,
      "step": 73900
    },
    {
      "epoch": 7.4,
      "grad_norm": 1.970182165678125e-05,
      "learning_rate": 3.7792642140468235e-06,
      "loss": 0.0025,
      "step": 74000
    },
    {
      "epoch": 7.41,
      "grad_norm": 7.84520962042734e-05,
      "learning_rate": 3.7775919732441476e-06,
      "loss": 0.0008,
      "step": 74100
    },
    {
      "epoch": 7.42,
      "grad_norm": 3.033554276044015e-05,
      "learning_rate": 3.775919732441472e-06,
      "loss": 0.011,
      "step": 74200
    },
    {
      "epoch": 7.43,
      "grad_norm": 0.0001484583190176636,
      "learning_rate": 3.774247491638796e-06,
      "loss": 0.0061,
      "step": 74300
    },
    {
      "epoch": 7.44,
      "grad_norm": 0.00032925844425335526,
      "learning_rate": 3.772575250836121e-06,
      "loss": 0.0,
      "step": 74400
    },
    {
      "epoch": 7.45,
      "grad_norm": 0.0002478963870089501,
      "learning_rate": 3.7709030100334454e-06,
      "loss": 0.0059,
      "step": 74500
    },
    {
      "epoch": 7.46,
      "grad_norm": 0.0003401384165044874,
      "learning_rate": 3.7692307692307694e-06,
      "loss": 0.007,
      "step": 74600
    },
    {
      "epoch": 7.47,
      "grad_norm": 0.0001823725615395233,
      "learning_rate": 3.7675585284280943e-06,
      "loss": 0.006,
      "step": 74700
    },
    {
      "epoch": 7.48,
      "grad_norm": 0.0011837665224447846,
      "learning_rate": 3.7658862876254184e-06,
      "loss": 0.0191,
      "step": 74800
    },
    {
      "epoch": 7.49,
      "grad_norm": 0.003498908132314682,
      "learning_rate": 3.764214046822743e-06,
      "loss": 0.0,
      "step": 74900
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.0009637516341172159,
      "learning_rate": 3.7625418060200673e-06,
      "loss": 0.0,
      "step": 75000
    },
    {
      "epoch": 7.5,
      "eval_accuracy": 0.9865,
      "eval_f1": 0.9865,
      "eval_loss": 0.12122364342212677,
      "eval_runtime": 133.3981,
      "eval_samples_per_second": 299.854,
      "eval_steps_per_second": 299.854,
      "step": 75000
    },
    {
      "epoch": 7.51,
      "grad_norm": 0.0003032281238120049,
      "learning_rate": 3.7608695652173917e-06,
      "loss": 0.0118,
      "step": 75100
    },
    {
      "epoch": 7.52,
      "grad_norm": 4.516799162956886e-05,
      "learning_rate": 3.759197324414716e-06,
      "loss": 0.0128,
      "step": 75200
    },
    {
      "epoch": 7.53,
      "grad_norm": 0.00011762830399675295,
      "learning_rate": 3.7575250836120402e-06,
      "loss": 0.001,
      "step": 75300
    },
    {
      "epoch": 7.54,
      "grad_norm": 0.05131925269961357,
      "learning_rate": 3.755852842809365e-06,
      "loss": 0.0001,
      "step": 75400
    },
    {
      "epoch": 7.55,
      "grad_norm": 1.698526284599211e-05,
      "learning_rate": 3.7541806020066896e-06,
      "loss": 0.0,
      "step": 75500
    },
    {
      "epoch": 7.5600000000000005,
      "grad_norm": 0.03275621682405472,
      "learning_rate": 3.7525083612040136e-06,
      "loss": 0.0095,
      "step": 75600
    },
    {
      "epoch": 7.57,
      "grad_norm": 3.255869887652807e-05,
      "learning_rate": 3.750836120401338e-06,
      "loss": 0.0044,
      "step": 75700
    },
    {
      "epoch": 7.58,
      "grad_norm": 0.10182749480009079,
      "learning_rate": 3.7491638795986625e-06,
      "loss": 0.0028,
      "step": 75800
    },
    {
      "epoch": 7.59,
      "grad_norm": 9.048231731867418e-05,
      "learning_rate": 3.747491638795987e-06,
      "loss": 0.0,
      "step": 75900
    },
    {
      "epoch": 7.6,
      "grad_norm": 1.5391413398901932e-05,
      "learning_rate": 3.745819397993311e-06,
      "loss": 0.0037,
      "step": 76000
    },
    {
      "epoch": 7.61,
      "grad_norm": 2.5761735741980374e-05,
      "learning_rate": 3.7441471571906355e-06,
      "loss": 0.0001,
      "step": 76100
    },
    {
      "epoch": 7.62,
      "grad_norm": 1.6684647562215105e-05,
      "learning_rate": 3.7424749163879604e-06,
      "loss": 0.0067,
      "step": 76200
    },
    {
      "epoch": 7.63,
      "grad_norm": 3.152184217469767e-05,
      "learning_rate": 3.7408026755852844e-06,
      "loss": 0.0096,
      "step": 76300
    },
    {
      "epoch": 7.64,
      "grad_norm": 0.0017304780194535851,
      "learning_rate": 3.739130434782609e-06,
      "loss": 0.0122,
      "step": 76400
    },
    {
      "epoch": 7.65,
      "grad_norm": 0.004649181850254536,
      "learning_rate": 3.7374581939799333e-06,
      "loss": 0.0099,
      "step": 76500
    },
    {
      "epoch": 7.66,
      "grad_norm": 2.4674011001479812e-05,
      "learning_rate": 3.735785953177258e-06,
      "loss": 0.0,
      "step": 76600
    },
    {
      "epoch": 7.67,
      "grad_norm": 4.725448525277898e-05,
      "learning_rate": 3.7341137123745823e-06,
      "loss": 0.0009,
      "step": 76700
    },
    {
      "epoch": 7.68,
      "grad_norm": 2.612950629554689e-05,
      "learning_rate": 3.7324414715719063e-06,
      "loss": 0.0046,
      "step": 76800
    },
    {
      "epoch": 7.6899999999999995,
      "grad_norm": 0.00784962996840477,
      "learning_rate": 3.730769230769231e-06,
      "loss": 0.0043,
      "step": 76900
    },
    {
      "epoch": 7.7,
      "grad_norm": 0.0011779394699260592,
      "learning_rate": 3.7290969899665552e-06,
      "loss": 0.0072,
      "step": 77000
    },
    {
      "epoch": 7.71,
      "grad_norm": 0.004041343927383423,
      "learning_rate": 3.7274247491638797e-06,
      "loss": 0.0083,
      "step": 77100
    },
    {
      "epoch": 7.72,
      "grad_norm": 0.0005305418744683266,
      "learning_rate": 3.7257525083612046e-06,
      "loss": 0.0062,
      "step": 77200
    },
    {
      "epoch": 7.73,
      "grad_norm": 4.281869769329205e-05,
      "learning_rate": 3.7240802675585286e-06,
      "loss": 0.0054,
      "step": 77300
    },
    {
      "epoch": 7.74,
      "grad_norm": 9.92060377029702e-05,
      "learning_rate": 3.722408026755853e-06,
      "loss": 0.0031,
      "step": 77400
    },
    {
      "epoch": 7.75,
      "grad_norm": 7.640924013685435e-05,
      "learning_rate": 3.720735785953177e-06,
      "loss": 0.013,
      "step": 77500
    },
    {
      "epoch": 7.76,
      "grad_norm": 0.0024570985697209835,
      "learning_rate": 3.719063545150502e-06,
      "loss": 0.0027,
      "step": 77600
    },
    {
      "epoch": 7.77,
      "grad_norm": 6.261836097110063e-05,
      "learning_rate": 3.7173913043478264e-06,
      "loss": 0.0045,
      "step": 77700
    },
    {
      "epoch": 7.78,
      "grad_norm": 0.0010023782961070538,
      "learning_rate": 3.7157190635451505e-06,
      "loss": 0.0043,
      "step": 77800
    },
    {
      "epoch": 7.79,
      "grad_norm": 8.210300438804552e-05,
      "learning_rate": 3.7140468227424754e-06,
      "loss": 0.0065,
      "step": 77900
    },
    {
      "epoch": 7.8,
      "grad_norm": 4.371823888504878e-05,
      "learning_rate": 3.7123745819397994e-06,
      "loss": 0.0074,
      "step": 78000
    },
    {
      "epoch": 7.8100000000000005,
      "grad_norm": 0.00750390999019146,
      "learning_rate": 3.710702341137124e-06,
      "loss": 0.0105,
      "step": 78100
    },
    {
      "epoch": 7.82,
      "grad_norm": 5.786586189060472e-05,
      "learning_rate": 3.7090301003344487e-06,
      "loss": 0.008,
      "step": 78200
    },
    {
      "epoch": 7.83,
      "grad_norm": 0.0022236057557165623,
      "learning_rate": 3.7073578595317728e-06,
      "loss": 0.0001,
      "step": 78300
    },
    {
      "epoch": 7.84,
      "grad_norm": 0.001743564149364829,
      "learning_rate": 3.7056856187290972e-06,
      "loss": 0.0195,
      "step": 78400
    },
    {
      "epoch": 7.85,
      "grad_norm": 0.00038661062717437744,
      "learning_rate": 3.7040133779264213e-06,
      "loss": 0.0029,
      "step": 78500
    },
    {
      "epoch": 7.86,
      "grad_norm": 0.0006590617704205215,
      "learning_rate": 3.702341137123746e-06,
      "loss": 0.0109,
      "step": 78600
    },
    {
      "epoch": 7.87,
      "grad_norm": 3.630742139648646e-05,
      "learning_rate": 3.7006688963210706e-06,
      "loss": 0.0,
      "step": 78700
    },
    {
      "epoch": 7.88,
      "grad_norm": 20.627477645874023,
      "learning_rate": 3.6989966555183947e-06,
      "loss": 0.0084,
      "step": 78800
    },
    {
      "epoch": 7.89,
      "grad_norm": 0.0004545499396044761,
      "learning_rate": 3.6973244147157195e-06,
      "loss": 0.0081,
      "step": 78900
    },
    {
      "epoch": 7.9,
      "grad_norm": 4.7894198360154405e-05,
      "learning_rate": 3.6956521739130436e-06,
      "loss": 0.0125,
      "step": 79000
    },
    {
      "epoch": 7.91,
      "grad_norm": 0.00010200806718785316,
      "learning_rate": 3.693979933110368e-06,
      "loss": 0.0033,
      "step": 79100
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.03108019381761551,
      "learning_rate": 3.692307692307693e-06,
      "loss": 0.0056,
      "step": 79200
    },
    {
      "epoch": 7.93,
      "grad_norm": 2.7014459192287177e-05,
      "learning_rate": 3.690635451505017e-06,
      "loss": 0.0001,
      "step": 79300
    },
    {
      "epoch": 7.9399999999999995,
      "grad_norm": 0.001831627101637423,
      "learning_rate": 3.6889632107023414e-06,
      "loss": 0.0087,
      "step": 79400
    },
    {
      "epoch": 7.95,
      "grad_norm": 3.703348556882702e-05,
      "learning_rate": 3.6872909698996655e-06,
      "loss": 0.0008,
      "step": 79500
    },
    {
      "epoch": 7.96,
      "grad_norm": 0.013346439227461815,
      "learning_rate": 3.6856187290969903e-06,
      "loss": 0.0008,
      "step": 79600
    },
    {
      "epoch": 7.97,
      "grad_norm": 6.900202424731106e-05,
      "learning_rate": 3.683946488294315e-06,
      "loss": 0.0054,
      "step": 79700
    },
    {
      "epoch": 7.98,
      "grad_norm": 0.014959572814404964,
      "learning_rate": 3.682274247491639e-06,
      "loss": 0.0126,
      "step": 79800
    },
    {
      "epoch": 7.99,
      "grad_norm": 0.0004625216533895582,
      "learning_rate": 3.6806020066889637e-06,
      "loss": 0.0009,
      "step": 79900
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.00030464696465060115,
      "learning_rate": 3.6789297658862878e-06,
      "loss": 0.003,
      "step": 80000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9853,
      "eval_f1": 0.9853,
      "eval_loss": 0.1378253996372223,
      "eval_runtime": 134.8364,
      "eval_samples_per_second": 296.656,
      "eval_steps_per_second": 296.656,
      "step": 80000
    },
    {
      "epoch": 8.01,
      "grad_norm": 2.883007073251065e-05,
      "learning_rate": 3.6772575250836122e-06,
      "loss": 0.0,
      "step": 80100
    },
    {
      "epoch": 8.02,
      "grad_norm": 0.001146984868682921,
      "learning_rate": 3.675585284280937e-06,
      "loss": 0.0,
      "step": 80200
    },
    {
      "epoch": 8.03,
      "grad_norm": 0.0017075365176424384,
      "learning_rate": 3.673913043478261e-06,
      "loss": 0.0049,
      "step": 80300
    },
    {
      "epoch": 8.04,
      "grad_norm": 1.7525797375128604e-05,
      "learning_rate": 3.6722408026755856e-06,
      "loss": 0.0042,
      "step": 80400
    },
    {
      "epoch": 8.05,
      "grad_norm": 3.318498420412652e-05,
      "learning_rate": 3.6705685618729096e-06,
      "loss": 0.0018,
      "step": 80500
    },
    {
      "epoch": 8.06,
      "grad_norm": 1.8242984879179858e-05,
      "learning_rate": 3.6688963210702345e-06,
      "loss": 0.0114,
      "step": 80600
    },
    {
      "epoch": 8.07,
      "grad_norm": 2.8576640033861622e-05,
      "learning_rate": 3.667224080267559e-06,
      "loss": 0.0061,
      "step": 80700
    },
    {
      "epoch": 8.08,
      "grad_norm": 2.6531528419582173e-05,
      "learning_rate": 3.665551839464883e-06,
      "loss": 0.0,
      "step": 80800
    },
    {
      "epoch": 8.09,
      "grad_norm": 3.302233017166145e-05,
      "learning_rate": 3.663879598662208e-06,
      "loss": 0.0037,
      "step": 80900
    },
    {
      "epoch": 8.1,
      "grad_norm": 3.1501982448389754e-05,
      "learning_rate": 3.662207357859532e-06,
      "loss": 0.0055,
      "step": 81000
    },
    {
      "epoch": 8.11,
      "grad_norm": 0.00011675096902763471,
      "learning_rate": 3.6605351170568564e-06,
      "loss": 0.0051,
      "step": 81100
    },
    {
      "epoch": 8.12,
      "grad_norm": 8.468510350212455e-05,
      "learning_rate": 3.6588628762541813e-06,
      "loss": 0.0,
      "step": 81200
    },
    {
      "epoch": 8.13,
      "grad_norm": 3.6243935028323904e-05,
      "learning_rate": 3.6571906354515053e-06,
      "loss": 0.0092,
      "step": 81300
    },
    {
      "epoch": 8.14,
      "grad_norm": 6.308049341896549e-05,
      "learning_rate": 3.6555183946488298e-06,
      "loss": 0.0,
      "step": 81400
    },
    {
      "epoch": 8.15,
      "grad_norm": 3.311773616587743e-05,
      "learning_rate": 3.653846153846154e-06,
      "loss": 0.0016,
      "step": 81500
    },
    {
      "epoch": 8.16,
      "grad_norm": 0.00015161107876338065,
      "learning_rate": 3.6521739130434787e-06,
      "loss": 0.0155,
      "step": 81600
    },
    {
      "epoch": 8.17,
      "grad_norm": 1.4969894436944742e-05,
      "learning_rate": 3.650501672240803e-06,
      "loss": 0.0071,
      "step": 81700
    },
    {
      "epoch": 8.18,
      "grad_norm": 2.1335055862437002e-05,
      "learning_rate": 3.648829431438127e-06,
      "loss": 0.0029,
      "step": 81800
    },
    {
      "epoch": 8.19,
      "grad_norm": 6.477204442489892e-05,
      "learning_rate": 3.647157190635452e-06,
      "loss": 0.0,
      "step": 81900
    },
    {
      "epoch": 8.2,
      "grad_norm": 0.3152509331703186,
      "learning_rate": 3.645484949832776e-06,
      "loss": 0.0129,
      "step": 82000
    },
    {
      "epoch": 8.21,
      "grad_norm": 2.7150046662427485e-05,
      "learning_rate": 3.6438127090301006e-06,
      "loss": 0.0062,
      "step": 82100
    },
    {
      "epoch": 8.22,
      "grad_norm": 0.0003274416958447546,
      "learning_rate": 3.6421404682274255e-06,
      "loss": 0.0049,
      "step": 82200
    },
    {
      "epoch": 8.23,
      "grad_norm": 0.014805939979851246,
      "learning_rate": 3.6404682274247495e-06,
      "loss": 0.0005,
      "step": 82300
    },
    {
      "epoch": 8.24,
      "grad_norm": 0.07976293563842773,
      "learning_rate": 3.638795986622074e-06,
      "loss": 0.0061,
      "step": 82400
    },
    {
      "epoch": 8.25,
      "grad_norm": 0.000629770162049681,
      "learning_rate": 3.637123745819398e-06,
      "loss": 0.0055,
      "step": 82500
    },
    {
      "epoch": 8.26,
      "grad_norm": 5.8729201555252075e-05,
      "learning_rate": 3.635451505016723e-06,
      "loss": 0.0095,
      "step": 82600
    },
    {
      "epoch": 8.27,
      "grad_norm": 2.821264206431806e-05,
      "learning_rate": 3.6337792642140473e-06,
      "loss": 0.0018,
      "step": 82700
    },
    {
      "epoch": 8.28,
      "grad_norm": 2.697145646379795e-05,
      "learning_rate": 3.6321070234113714e-06,
      "loss": 0.0058,
      "step": 82800
    },
    {
      "epoch": 8.29,
      "grad_norm": 2.547060103097465e-05,
      "learning_rate": 3.6304347826086963e-06,
      "loss": 0.0004,
      "step": 82900
    },
    {
      "epoch": 8.3,
      "grad_norm": 2.24012019316433e-05,
      "learning_rate": 3.6287625418060203e-06,
      "loss": 0.0011,
      "step": 83000
    },
    {
      "epoch": 8.31,
      "grad_norm": 5.2949766541132703e-05,
      "learning_rate": 3.6270903010033448e-06,
      "loss": 0.0033,
      "step": 83100
    },
    {
      "epoch": 8.32,
      "grad_norm": 0.12880505621433258,
      "learning_rate": 3.6254180602006696e-06,
      "loss": 0.0051,
      "step": 83200
    },
    {
      "epoch": 8.33,
      "grad_norm": 5.116287138662301e-05,
      "learning_rate": 3.6237458193979937e-06,
      "loss": 0.0028,
      "step": 83300
    },
    {
      "epoch": 8.34,
      "grad_norm": 2.302383836649824e-05,
      "learning_rate": 3.622073578595318e-06,
      "loss": 0.005,
      "step": 83400
    },
    {
      "epoch": 8.35,
      "grad_norm": 0.002800439950078726,
      "learning_rate": 3.620401337792642e-06,
      "loss": 0.0075,
      "step": 83500
    },
    {
      "epoch": 8.36,
      "grad_norm": 3.3340267691528425e-05,
      "learning_rate": 3.618729096989967e-06,
      "loss": 0.0055,
      "step": 83600
    },
    {
      "epoch": 8.37,
      "grad_norm": 5.368978236219846e-05,
      "learning_rate": 3.6170568561872915e-06,
      "loss": 0.0037,
      "step": 83700
    },
    {
      "epoch": 8.38,
      "grad_norm": 4.120146331842989e-05,
      "learning_rate": 3.6153846153846156e-06,
      "loss": 0.0015,
      "step": 83800
    },
    {
      "epoch": 8.39,
      "grad_norm": 2.2930082195671275e-05,
      "learning_rate": 3.6137123745819404e-06,
      "loss": 0.0104,
      "step": 83900
    },
    {
      "epoch": 8.4,
      "grad_norm": 1.8325974451727234e-05,
      "learning_rate": 3.6120401337792645e-06,
      "loss": 0.0085,
      "step": 84000
    },
    {
      "epoch": 8.41,
      "grad_norm": 1.8396835002931766e-05,
      "learning_rate": 3.610367892976589e-06,
      "loss": 0.0,
      "step": 84100
    },
    {
      "epoch": 8.42,
      "grad_norm": 1.9814324332401156e-05,
      "learning_rate": 3.6086956521739134e-06,
      "loss": 0.0,
      "step": 84200
    },
    {
      "epoch": 8.43,
      "grad_norm": 48.710208892822266,
      "learning_rate": 3.607023411371238e-06,
      "loss": 0.015,
      "step": 84300
    },
    {
      "epoch": 8.44,
      "grad_norm": 2.536492502258625e-05,
      "learning_rate": 3.6053511705685623e-06,
      "loss": 0.0078,
      "step": 84400
    },
    {
      "epoch": 8.45,
      "grad_norm": 0.00015233943122439086,
      "learning_rate": 3.6036789297658864e-06,
      "loss": 0.014,
      "step": 84500
    },
    {
      "epoch": 8.46,
      "grad_norm": 1.858683390310034e-05,
      "learning_rate": 3.6020066889632112e-06,
      "loss": 0.0016,
      "step": 84600
    },
    {
      "epoch": 8.47,
      "grad_norm": 1.7697208022582345e-05,
      "learning_rate": 3.6003344481605357e-06,
      "loss": 0.0026,
      "step": 84700
    },
    {
      "epoch": 8.48,
      "grad_norm": 3.1560120987705886e-05,
      "learning_rate": 3.5986622073578597e-06,
      "loss": 0.0,
      "step": 84800
    },
    {
      "epoch": 8.49,
      "grad_norm": 1.6911069906200282e-05,
      "learning_rate": 3.596989966555184e-06,
      "loss": 0.0038,
      "step": 84900
    },
    {
      "epoch": 8.5,
      "grad_norm": 2.0758106984430924e-05,
      "learning_rate": 3.5953177257525087e-06,
      "loss": 0.0005,
      "step": 85000
    },
    {
      "epoch": 8.5,
      "eval_accuracy": 0.9857,
      "eval_f1": 0.9857,
      "eval_loss": 0.13761389255523682,
      "eval_runtime": 135.2379,
      "eval_samples_per_second": 295.775,
      "eval_steps_per_second": 295.775,
      "step": 85000
    },
    {
      "epoch": 8.51,
      "grad_norm": 6.27836343483068e-05,
      "learning_rate": 3.593645484949833e-06,
      "loss": 0.0204,
      "step": 85100
    },
    {
      "epoch": 8.52,
      "grad_norm": 1.9296541722724214e-05,
      "learning_rate": 3.5919732441471576e-06,
      "loss": 0.0039,
      "step": 85200
    },
    {
      "epoch": 8.53,
      "grad_norm": 0.0005484924768097699,
      "learning_rate": 3.590301003344482e-06,
      "loss": 0.0083,
      "step": 85300
    },
    {
      "epoch": 8.54,
      "grad_norm": 1.317109854426235e-05,
      "learning_rate": 3.5886287625418065e-06,
      "loss": 0.0,
      "step": 85400
    },
    {
      "epoch": 8.55,
      "grad_norm": 1.4201648809830658e-05,
      "learning_rate": 3.5869565217391305e-06,
      "loss": 0.0061,
      "step": 85500
    },
    {
      "epoch": 8.56,
      "grad_norm": 4.045464811497368e-05,
      "learning_rate": 3.585284280936455e-06,
      "loss": 0.0,
      "step": 85600
    },
    {
      "epoch": 8.57,
      "grad_norm": 4.9041085731005296e-05,
      "learning_rate": 3.58361204013378e-06,
      "loss": 0.0001,
      "step": 85700
    },
    {
      "epoch": 8.58,
      "grad_norm": 0.0012458403361961246,
      "learning_rate": 3.581939799331104e-06,
      "loss": 0.0035,
      "step": 85800
    },
    {
      "epoch": 8.59,
      "grad_norm": 1.6935151506913826e-05,
      "learning_rate": 3.5802675585284284e-06,
      "loss": 0.0003,
      "step": 85900
    },
    {
      "epoch": 8.6,
      "grad_norm": 2.6776717277243733e-05,
      "learning_rate": 3.578595317725753e-06,
      "loss": 0.0002,
      "step": 86000
    },
    {
      "epoch": 8.61,
      "grad_norm": 2.0100162146263756e-05,
      "learning_rate": 3.5769230769230773e-06,
      "loss": 0.0102,
      "step": 86100
    },
    {
      "epoch": 8.62,
      "grad_norm": 1.1070689652115107e-05,
      "learning_rate": 3.5752508361204013e-06,
      "loss": 0.0,
      "step": 86200
    },
    {
      "epoch": 8.63,
      "grad_norm": 1.861604869191069e-05,
      "learning_rate": 3.573578595317726e-06,
      "loss": 0.0044,
      "step": 86300
    },
    {
      "epoch": 8.64,
      "grad_norm": 8.287372475024313e-05,
      "learning_rate": 3.5719063545150507e-06,
      "loss": 0.006,
      "step": 86400
    },
    {
      "epoch": 8.65,
      "grad_norm": 1.6306277757394128e-05,
      "learning_rate": 3.5702341137123747e-06,
      "loss": 0.0166,
      "step": 86500
    },
    {
      "epoch": 8.66,
      "grad_norm": 0.00019994894682895392,
      "learning_rate": 3.568561872909699e-06,
      "loss": 0.0032,
      "step": 86600
    },
    {
      "epoch": 8.67,
      "grad_norm": 1.5932404494378716e-05,
      "learning_rate": 3.5668896321070236e-06,
      "loss": 0.0048,
      "step": 86700
    },
    {
      "epoch": 8.68,
      "grad_norm": 4.9370060878572986e-05,
      "learning_rate": 3.565217391304348e-06,
      "loss": 0.0027,
      "step": 86800
    },
    {
      "epoch": 8.69,
      "grad_norm": 1.9548153431969695e-05,
      "learning_rate": 3.5635451505016726e-06,
      "loss": 0.0,
      "step": 86900
    },
    {
      "epoch": 8.7,
      "grad_norm": 1.9069799236604013e-05,
      "learning_rate": 3.5618729096989966e-06,
      "loss": 0.008,
      "step": 87000
    },
    {
      "epoch": 8.71,
      "grad_norm": 0.00030942470766603947,
      "learning_rate": 3.5602006688963215e-06,
      "loss": 0.014,
      "step": 87100
    },
    {
      "epoch": 8.72,
      "grad_norm": 0.00028676740475930274,
      "learning_rate": 3.5585284280936455e-06,
      "loss": 0.0001,
      "step": 87200
    },
    {
      "epoch": 8.73,
      "grad_norm": 0.00011423743126215413,
      "learning_rate": 3.55685618729097e-06,
      "loss": 0.0036,
      "step": 87300
    },
    {
      "epoch": 8.74,
      "grad_norm": 2.1552994439844042e-05,
      "learning_rate": 3.555183946488295e-06,
      "loss": 0.0,
      "step": 87400
    },
    {
      "epoch": 8.75,
      "grad_norm": 1.99559326574672e-05,
      "learning_rate": 3.553511705685619e-06,
      "loss": 0.0018,
      "step": 87500
    },
    {
      "epoch": 8.76,
      "grad_norm": 1.4471507711277809e-05,
      "learning_rate": 3.5518394648829434e-06,
      "loss": 0.0058,
      "step": 87600
    },
    {
      "epoch": 8.77,
      "grad_norm": 0.00047445972450077534,
      "learning_rate": 3.5501672240802674e-06,
      "loss": 0.0,
      "step": 87700
    },
    {
      "epoch": 8.78,
      "grad_norm": 7.757473940728232e-05,
      "learning_rate": 3.5484949832775923e-06,
      "loss": 0.0,
      "step": 87800
    },
    {
      "epoch": 8.79,
      "grad_norm": 2.4758850486250594e-05,
      "learning_rate": 3.5468227424749167e-06,
      "loss": 0.0,
      "step": 87900
    },
    {
      "epoch": 8.8,
      "grad_norm": 3.1655439670430496e-05,
      "learning_rate": 3.5451505016722408e-06,
      "loss": 0.0,
      "step": 88000
    },
    {
      "epoch": 8.81,
      "grad_norm": 0.0323539599776268,
      "learning_rate": 3.5434782608695657e-06,
      "loss": 0.0066,
      "step": 88100
    },
    {
      "epoch": 8.82,
      "grad_norm": 2.997820411110297e-05,
      "learning_rate": 3.5418060200668897e-06,
      "loss": 0.0019,
      "step": 88200
    },
    {
      "epoch": 8.83,
      "grad_norm": 2.695979674172122e-05,
      "learning_rate": 3.540133779264214e-06,
      "loss": 0.0037,
      "step": 88300
    },
    {
      "epoch": 8.84,
      "grad_norm": 0.00042679201578721404,
      "learning_rate": 3.538461538461539e-06,
      "loss": 0.0101,
      "step": 88400
    },
    {
      "epoch": 8.85,
      "grad_norm": 0.0005938719259575009,
      "learning_rate": 3.536789297658863e-06,
      "loss": 0.0,
      "step": 88500
    },
    {
      "epoch": 8.86,
      "grad_norm": 1.996893115574494e-05,
      "learning_rate": 3.5351170568561875e-06,
      "loss": 0.0093,
      "step": 88600
    },
    {
      "epoch": 8.87,
      "grad_norm": 1.4617453416576609e-05,
      "learning_rate": 3.5334448160535116e-06,
      "loss": 0.0053,
      "step": 88700
    },
    {
      "epoch": 8.88,
      "grad_norm": 2.820068402797915e-05,
      "learning_rate": 3.5317725752508365e-06,
      "loss": 0.0102,
      "step": 88800
    },
    {
      "epoch": 8.89,
      "grad_norm": 2.9803282814100385e-05,
      "learning_rate": 3.530100334448161e-06,
      "loss": 0.0025,
      "step": 88900
    },
    {
      "epoch": 8.9,
      "grad_norm": 1.5636564057786018e-05,
      "learning_rate": 3.528428093645485e-06,
      "loss": 0.006,
      "step": 89000
    },
    {
      "epoch": 8.91,
      "grad_norm": 1.9173994587617926e-05,
      "learning_rate": 3.52675585284281e-06,
      "loss": 0.0007,
      "step": 89100
    },
    {
      "epoch": 8.92,
      "grad_norm": 8.140524732880294e-05,
      "learning_rate": 3.525083612040134e-06,
      "loss": 0.0103,
      "step": 89200
    },
    {
      "epoch": 8.93,
      "grad_norm": 2.191733437939547e-05,
      "learning_rate": 3.5234113712374583e-06,
      "loss": 0.0,
      "step": 89300
    },
    {
      "epoch": 8.94,
      "grad_norm": 0.00012123118358431384,
      "learning_rate": 3.5217391304347832e-06,
      "loss": 0.0046,
      "step": 89400
    },
    {
      "epoch": 8.95,
      "grad_norm": 0.00010186954023083672,
      "learning_rate": 3.5200668896321073e-06,
      "loss": 0.0054,
      "step": 89500
    },
    {
      "epoch": 8.96,
      "grad_norm": 0.010097470134496689,
      "learning_rate": 3.5183946488294317e-06,
      "loss": 0.0,
      "step": 89600
    },
    {
      "epoch": 8.97,
      "grad_norm": 0.02606087736785412,
      "learning_rate": 3.5167224080267558e-06,
      "loss": 0.0127,
      "step": 89700
    },
    {
      "epoch": 8.98,
      "grad_norm": 0.005077203270047903,
      "learning_rate": 3.5150501672240807e-06,
      "loss": 0.0029,
      "step": 89800
    },
    {
      "epoch": 8.99,
      "grad_norm": 0.0013905907981097698,
      "learning_rate": 3.513377926421405e-06,
      "loss": 0.0073,
      "step": 89900
    },
    {
      "epoch": 9.0,
      "grad_norm": 6.247458077268675e-05,
      "learning_rate": 3.511705685618729e-06,
      "loss": 0.0103,
      "step": 90000
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.986125,
      "eval_f1": 0.986125,
      "eval_loss": 0.11607175320386887,
      "eval_runtime": 138.0617,
      "eval_samples_per_second": 289.726,
      "eval_steps_per_second": 289.726,
      "step": 90000
    },
    {
      "epoch": 9.01,
      "grad_norm": 0.00237733474932611,
      "learning_rate": 3.510033444816054e-06,
      "loss": 0.003,
      "step": 90100
    },
    {
      "epoch": 9.02,
      "grad_norm": 2.1511707018362358e-05,
      "learning_rate": 3.508361204013378e-06,
      "loss": 0.0,
      "step": 90200
    },
    {
      "epoch": 9.03,
      "grad_norm": 1.744981091178488e-05,
      "learning_rate": 3.5066889632107025e-06,
      "loss": 0.0,
      "step": 90300
    },
    {
      "epoch": 9.04,
      "grad_norm": 3.586035018088296e-05,
      "learning_rate": 3.5050167224080274e-06,
      "loss": 0.0049,
      "step": 90400
    },
    {
      "epoch": 9.05,
      "grad_norm": 0.003070132341235876,
      "learning_rate": 3.5033444816053515e-06,
      "loss": 0.0073,
      "step": 90500
    },
    {
      "epoch": 9.06,
      "grad_norm": 0.0012648507254198194,
      "learning_rate": 3.501672240802676e-06,
      "loss": 0.0024,
      "step": 90600
    },
    {
      "epoch": 9.07,
      "grad_norm": 2.4553286493755877e-05,
      "learning_rate": 3.5e-06,
      "loss": 0.0027,
      "step": 90700
    },
    {
      "epoch": 9.08,
      "grad_norm": 0.03265082463622093,
      "learning_rate": 3.498327759197325e-06,
      "loss": 0.0055,
      "step": 90800
    },
    {
      "epoch": 9.09,
      "grad_norm": 2.972443326143548e-05,
      "learning_rate": 3.4966555183946493e-06,
      "loss": 0.0,
      "step": 90900
    },
    {
      "epoch": 9.1,
      "grad_norm": 8.496625378029421e-05,
      "learning_rate": 3.4949832775919733e-06,
      "loss": 0.0038,
      "step": 91000
    },
    {
      "epoch": 9.11,
      "grad_norm": 3.153005309286527e-05,
      "learning_rate": 3.4933110367892982e-06,
      "loss": 0.0042,
      "step": 91100
    },
    {
      "epoch": 9.12,
      "grad_norm": 4.433723370311782e-05,
      "learning_rate": 3.4916387959866222e-06,
      "loss": 0.0,
      "step": 91200
    },
    {
      "epoch": 9.13,
      "grad_norm": 1.847981366154272e-05,
      "learning_rate": 3.4899665551839467e-06,
      "loss": 0.0015,
      "step": 91300
    },
    {
      "epoch": 9.14,
      "grad_norm": 1.2556378351291642e-05,
      "learning_rate": 3.4882943143812716e-06,
      "loss": 0.0001,
      "step": 91400
    },
    {
      "epoch": 9.15,
      "grad_norm": 29.66756248474121,
      "learning_rate": 3.4866220735785956e-06,
      "loss": 0.0089,
      "step": 91500
    },
    {
      "epoch": 9.16,
      "grad_norm": 0.005458858795464039,
      "learning_rate": 3.48494983277592e-06,
      "loss": 0.0085,
      "step": 91600
    },
    {
      "epoch": 9.17,
      "grad_norm": 3.206008477718569e-05,
      "learning_rate": 3.483277591973244e-06,
      "loss": 0.0072,
      "step": 91700
    },
    {
      "epoch": 9.18,
      "grad_norm": 0.0695156380534172,
      "learning_rate": 3.481605351170569e-06,
      "loss": 0.0003,
      "step": 91800
    },
    {
      "epoch": 9.19,
      "grad_norm": 2.1034235032857396e-05,
      "learning_rate": 3.4799331103678935e-06,
      "loss": 0.0024,
      "step": 91900
    },
    {
      "epoch": 9.2,
      "grad_norm": 0.0007182686822488904,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.0,
      "step": 92000
    },
    {
      "epoch": 9.21,
      "grad_norm": 5.026232611271553e-05,
      "learning_rate": 3.4765886287625424e-06,
      "loss": 0.01,
      "step": 92100
    },
    {
      "epoch": 9.22,
      "grad_norm": 2.3394995878334157e-05,
      "learning_rate": 3.4749163879598664e-06,
      "loss": 0.0142,
      "step": 92200
    },
    {
      "epoch": 9.23,
      "grad_norm": 8.789805724518374e-05,
      "learning_rate": 3.473244147157191e-06,
      "loss": 0.0001,
      "step": 92300
    },
    {
      "epoch": 9.24,
      "grad_norm": 2.1054107492091134e-05,
      "learning_rate": 3.4715719063545158e-06,
      "loss": 0.0011,
      "step": 92400
    },
    {
      "epoch": 9.25,
      "grad_norm": 0.00026566089945845306,
      "learning_rate": 3.46989966555184e-06,
      "loss": 0.0049,
      "step": 92500
    },
    {
      "epoch": 9.26,
      "grad_norm": 4.639206963474862e-05,
      "learning_rate": 3.4682274247491643e-06,
      "loss": 0.0115,
      "step": 92600
    },
    {
      "epoch": 9.27,
      "grad_norm": 3.660786387627013e-05,
      "learning_rate": 3.4665551839464883e-06,
      "loss": 0.0002,
      "step": 92700
    },
    {
      "epoch": 9.28,
      "grad_norm": 3.224415559088811e-05,
      "learning_rate": 3.464882943143813e-06,
      "loss": 0.0148,
      "step": 92800
    },
    {
      "epoch": 9.29,
      "grad_norm": 0.002478808630257845,
      "learning_rate": 3.4632107023411377e-06,
      "loss": 0.0041,
      "step": 92900
    },
    {
      "epoch": 9.3,
      "grad_norm": 5.7758701586863026e-05,
      "learning_rate": 3.4615384615384617e-06,
      "loss": 0.0039,
      "step": 93000
    },
    {
      "epoch": 9.31,
      "grad_norm": 0.00024893361842259765,
      "learning_rate": 3.4598662207357866e-06,
      "loss": 0.0032,
      "step": 93100
    },
    {
      "epoch": 9.32,
      "grad_norm": 6.84029291733168e-05,
      "learning_rate": 3.4581939799331106e-06,
      "loss": 0.0062,
      "step": 93200
    },
    {
      "epoch": 9.33,
      "grad_norm": 0.0036423206329345703,
      "learning_rate": 3.456521739130435e-06,
      "loss": 0.0008,
      "step": 93300
    },
    {
      "epoch": 9.34,
      "grad_norm": 0.029986459761857986,
      "learning_rate": 3.4548494983277595e-06,
      "loss": 0.0,
      "step": 93400
    },
    {
      "epoch": 9.35,
      "grad_norm": 0.00015107858052942902,
      "learning_rate": 3.453177257525084e-06,
      "loss": 0.0,
      "step": 93500
    },
    {
      "epoch": 9.36,
      "grad_norm": 5.476970545714721e-05,
      "learning_rate": 3.4515050167224085e-06,
      "loss": 0.0018,
      "step": 93600
    },
    {
      "epoch": 9.37,
      "grad_norm": 0.00011632416135398671,
      "learning_rate": 3.4498327759197325e-06,
      "loss": 0.0088,
      "step": 93700
    },
    {
      "epoch": 9.38,
      "grad_norm": 0.0003741678374353796,
      "learning_rate": 3.4481605351170574e-06,
      "loss": 0.0045,
      "step": 93800
    },
    {
      "epoch": 9.39,
      "grad_norm": 0.0009761012624949217,
      "learning_rate": 3.446488294314382e-06,
      "loss": 0.0033,
      "step": 93900
    },
    {
      "epoch": 9.4,
      "grad_norm": 3.911781823262572e-05,
      "learning_rate": 3.444816053511706e-06,
      "loss": 0.0019,
      "step": 94000
    },
    {
      "epoch": 9.41,
      "grad_norm": 1.7435120753361844e-05,
      "learning_rate": 3.4431438127090303e-06,
      "loss": 0.0048,
      "step": 94100
    },
    {
      "epoch": 9.42,
      "grad_norm": 3.1139043130679056e-05,
      "learning_rate": 3.441471571906355e-06,
      "loss": 0.0,
      "step": 94200
    },
    {
      "epoch": 9.43,
      "grad_norm": 0.0002475950459484011,
      "learning_rate": 3.4397993311036793e-06,
      "loss": 0.0072,
      "step": 94300
    },
    {
      "epoch": 9.44,
      "grad_norm": 1.8925727999885567e-05,
      "learning_rate": 3.4381270903010037e-06,
      "loss": 0.0018,
      "step": 94400
    },
    {
      "epoch": 9.45,
      "grad_norm": 1.3380113159655593e-05,
      "learning_rate": 3.436454849498328e-06,
      "loss": 0.0,
      "step": 94500
    },
    {
      "epoch": 9.46,
      "grad_norm": 4.596728831529617e-05,
      "learning_rate": 3.4347826086956526e-06,
      "loss": 0.0027,
      "step": 94600
    },
    {
      "epoch": 9.47,
      "grad_norm": 6.345294241327792e-05,
      "learning_rate": 3.4331103678929767e-06,
      "loss": 0.0032,
      "step": 94700
    },
    {
      "epoch": 9.48,
      "grad_norm": 1.4074191312829498e-05,
      "learning_rate": 3.431438127090301e-06,
      "loss": 0.0038,
      "step": 94800
    },
    {
      "epoch": 9.49,
      "grad_norm": 0.0012907675700262189,
      "learning_rate": 3.429765886287626e-06,
      "loss": 0.0,
      "step": 94900
    },
    {
      "epoch": 9.5,
      "grad_norm": 0.010470391251146793,
      "learning_rate": 3.42809364548495e-06,
      "loss": 0.0152,
      "step": 95000
    },
    {
      "epoch": 9.5,
      "eval_accuracy": 0.9842,
      "eval_f1": 0.9842,
      "eval_loss": 0.13753095269203186,
      "eval_runtime": 137.4145,
      "eval_samples_per_second": 291.09,
      "eval_steps_per_second": 291.09,
      "step": 95000
    },
    {
      "epoch": 9.51,
      "grad_norm": 0.004034177400171757,
      "learning_rate": 3.4264214046822745e-06,
      "loss": 0.0143,
      "step": 95100
    },
    {
      "epoch": 9.52,
      "grad_norm": 0.002143730642274022,
      "learning_rate": 3.424749163879599e-06,
      "loss": 0.0018,
      "step": 95200
    },
    {
      "epoch": 9.53,
      "grad_norm": 0.0022789237555116415,
      "learning_rate": 3.4230769230769234e-06,
      "loss": 0.0,
      "step": 95300
    },
    {
      "epoch": 9.54,
      "grad_norm": 9.519604645902291e-05,
      "learning_rate": 3.421404682274248e-06,
      "loss": 0.0159,
      "step": 95400
    },
    {
      "epoch": 9.55,
      "grad_norm": 0.0038595341611653566,
      "learning_rate": 3.419732441471572e-06,
      "loss": 0.004,
      "step": 95500
    },
    {
      "epoch": 9.56,
      "grad_norm": 2.268450589326676e-05,
      "learning_rate": 3.418060200668897e-06,
      "loss": 0.0006,
      "step": 95600
    },
    {
      "epoch": 9.57,
      "grad_norm": 3.149811891489662e-05,
      "learning_rate": 3.416387959866221e-06,
      "loss": 0.0064,
      "step": 95700
    },
    {
      "epoch": 9.58,
      "grad_norm": 2.6584746592561714e-05,
      "learning_rate": 3.4147157190635453e-06,
      "loss": 0.0101,
      "step": 95800
    },
    {
      "epoch": 9.59,
      "grad_norm": 0.020238105207681656,
      "learning_rate": 3.4130434782608698e-06,
      "loss": 0.0022,
      "step": 95900
    },
    {
      "epoch": 9.6,
      "grad_norm": 0.0001373113045701757,
      "learning_rate": 3.4113712374581942e-06,
      "loss": 0.0,
      "step": 96000
    },
    {
      "epoch": 9.61,
      "grad_norm": 0.0002782682713586837,
      "learning_rate": 3.4096989966555187e-06,
      "loss": 0.0,
      "step": 96100
    },
    {
      "epoch": 9.62,
      "grad_norm": 0.00020216793927829713,
      "learning_rate": 3.4080267558528427e-06,
      "loss": 0.0025,
      "step": 96200
    },
    {
      "epoch": 9.63,
      "grad_norm": 0.02557799033820629,
      "learning_rate": 3.4063545150501676e-06,
      "loss": 0.0133,
      "step": 96300
    },
    {
      "epoch": 9.64,
      "grad_norm": 1.481761228205869e-05,
      "learning_rate": 3.4046822742474917e-06,
      "loss": 0.0029,
      "step": 96400
    },
    {
      "epoch": 9.65,
      "grad_norm": 0.007463955320417881,
      "learning_rate": 3.403010033444816e-06,
      "loss": 0.0064,
      "step": 96500
    },
    {
      "epoch": 9.66,
      "grad_norm": 0.00026536182849667966,
      "learning_rate": 3.401337792642141e-06,
      "loss": 0.006,
      "step": 96600
    },
    {
      "epoch": 9.67,
      "grad_norm": 3.175417077727616e-05,
      "learning_rate": 3.399665551839465e-06,
      "loss": 0.0082,
      "step": 96700
    },
    {
      "epoch": 9.68,
      "grad_norm": 2.764562304946594e-05,
      "learning_rate": 3.3979933110367895e-06,
      "loss": 0.0001,
      "step": 96800
    },
    {
      "epoch": 9.69,
      "grad_norm": 2.0898842194583267e-05,
      "learning_rate": 3.3963210702341135e-06,
      "loss": 0.0,
      "step": 96900
    },
    {
      "epoch": 9.7,
      "grad_norm": 0.00020050223974976689,
      "learning_rate": 3.3946488294314384e-06,
      "loss": 0.0012,
      "step": 97000
    },
    {
      "epoch": 9.71,
      "grad_norm": 4.08537634939421e-05,
      "learning_rate": 3.392976588628763e-06,
      "loss": 0.0,
      "step": 97100
    },
    {
      "epoch": 9.72,
      "grad_norm": 1.0601573194435332e-05,
      "learning_rate": 3.391304347826087e-06,
      "loss": 0.0,
      "step": 97200
    },
    {
      "epoch": 9.73,
      "grad_norm": 2.0686477000708692e-05,
      "learning_rate": 3.389632107023412e-06,
      "loss": 0.0,
      "step": 97300
    },
    {
      "epoch": 9.74,
      "grad_norm": 0.034469928592443466,
      "learning_rate": 3.387959866220736e-06,
      "loss": 0.0074,
      "step": 97400
    },
    {
      "epoch": 9.75,
      "grad_norm": 5.762694854638539e-05,
      "learning_rate": 3.3862876254180603e-06,
      "loss": 0.013,
      "step": 97500
    },
    {
      "epoch": 9.76,
      "grad_norm": 8.739017357584089e-05,
      "learning_rate": 3.384615384615385e-06,
      "loss": 0.0065,
      "step": 97600
    },
    {
      "epoch": 9.77,
      "grad_norm": 1.8738399376161397e-05,
      "learning_rate": 3.3829431438127092e-06,
      "loss": 0.0025,
      "step": 97700
    },
    {
      "epoch": 9.78,
      "grad_norm": 1.5836463717278093e-05,
      "learning_rate": 3.3812709030100337e-06,
      "loss": 0.0,
      "step": 97800
    },
    {
      "epoch": 9.79,
      "grad_norm": 33.77321243286133,
      "learning_rate": 3.3795986622073577e-06,
      "loss": 0.0102,
      "step": 97900
    },
    {
      "epoch": 9.8,
      "grad_norm": 2.430802669550758e-05,
      "learning_rate": 3.3779264214046826e-06,
      "loss": 0.0,
      "step": 98000
    },
    {
      "epoch": 9.81,
      "grad_norm": 1.6347927157767117e-05,
      "learning_rate": 3.376254180602007e-06,
      "loss": 0.0,
      "step": 98100
    },
    {
      "epoch": 9.82,
      "grad_norm": 2.01090078917332e-05,
      "learning_rate": 3.374581939799331e-06,
      "loss": 0.008,
      "step": 98200
    },
    {
      "epoch": 9.83,
      "grad_norm": 0.00010819236194947734,
      "learning_rate": 3.372909698996656e-06,
      "loss": 0.0,
      "step": 98300
    },
    {
      "epoch": 9.84,
      "grad_norm": 0.0038400471676141024,
      "learning_rate": 3.37123745819398e-06,
      "loss": 0.0,
      "step": 98400
    },
    {
      "epoch": 9.85,
      "grad_norm": 6.486551865236834e-05,
      "learning_rate": 3.3695652173913045e-06,
      "loss": 0.0001,
      "step": 98500
    },
    {
      "epoch": 9.86,
      "grad_norm": 0.00033912554499693215,
      "learning_rate": 3.3678929765886294e-06,
      "loss": 0.0,
      "step": 98600
    },
    {
      "epoch": 9.87,
      "grad_norm": 2.700369259400759e-05,
      "learning_rate": 3.3662207357859534e-06,
      "loss": 0.0026,
      "step": 98700
    },
    {
      "epoch": 9.88,
      "grad_norm": 1.5537827493972145e-05,
      "learning_rate": 3.364548494983278e-06,
      "loss": 0.0074,
      "step": 98800
    },
    {
      "epoch": 9.89,
      "grad_norm": 2.3932556359795853e-05,
      "learning_rate": 3.362876254180602e-06,
      "loss": 0.0037,
      "step": 98900
    },
    {
      "epoch": 9.9,
      "grad_norm": 1.4799331438553054e-05,
      "learning_rate": 3.3612040133779268e-06,
      "loss": 0.0024,
      "step": 99000
    },
    {
      "epoch": 9.91,
      "grad_norm": 2.9224020181572996e-05,
      "learning_rate": 3.3595317725752512e-06,
      "loss": 0.0041,
      "step": 99100
    },
    {
      "epoch": 9.92,
      "grad_norm": 0.0006482133758254349,
      "learning_rate": 3.3578595317725753e-06,
      "loss": 0.003,
      "step": 99200
    },
    {
      "epoch": 9.93,
      "grad_norm": 2.1120746168890037e-05,
      "learning_rate": 3.3561872909699e-06,
      "loss": 0.01,
      "step": 99300
    },
    {
      "epoch": 9.94,
      "grad_norm": 5.945864904788323e-05,
      "learning_rate": 3.354515050167224e-06,
      "loss": 0.0051,
      "step": 99400
    },
    {
      "epoch": 9.95,
      "grad_norm": 4.6771976485615596e-05,
      "learning_rate": 3.3528428093645487e-06,
      "loss": 0.0,
      "step": 99500
    },
    {
      "epoch": 9.96,
      "grad_norm": 1.595135108800605e-05,
      "learning_rate": 3.3511705685618735e-06,
      "loss": 0.0,
      "step": 99600
    },
    {
      "epoch": 9.97,
      "grad_norm": 46.883026123046875,
      "learning_rate": 3.3494983277591976e-06,
      "loss": 0.0033,
      "step": 99700
    },
    {
      "epoch": 9.98,
      "grad_norm": 0.00022700062254443765,
      "learning_rate": 3.347826086956522e-06,
      "loss": 0.0156,
      "step": 99800
    },
    {
      "epoch": 9.99,
      "grad_norm": 0.000159778879606165,
      "learning_rate": 3.346153846153846e-06,
      "loss": 0.0,
      "step": 99900
    },
    {
      "epoch": 10.0,
      "grad_norm": 8.880961104296148e-05,
      "learning_rate": 3.344481605351171e-06,
      "loss": 0.0001,
      "step": 100000
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.98565,
      "eval_f1": 0.98565,
      "eval_loss": 0.12817147374153137,
      "eval_runtime": 136.2325,
      "eval_samples_per_second": 293.616,
      "eval_steps_per_second": 293.616,
      "step": 100000
    },
    {
      "epoch": 10.01,
      "grad_norm": 3.595802991185337e-05,
      "learning_rate": 3.3428093645484954e-06,
      "loss": 0.0095,
      "step": 100100
    },
    {
      "epoch": 10.02,
      "grad_norm": 0.0004151228931732476,
      "learning_rate": 3.3411371237458195e-06,
      "loss": 0.0075,
      "step": 100200
    },
    {
      "epoch": 10.03,
      "grad_norm": 0.010722924955189228,
      "learning_rate": 3.3394648829431443e-06,
      "loss": 0.0,
      "step": 100300
    },
    {
      "epoch": 10.04,
      "grad_norm": 0.016409115865826607,
      "learning_rate": 3.3377926421404684e-06,
      "loss": 0.0,
      "step": 100400
    },
    {
      "epoch": 10.05,
      "grad_norm": 1.7606289475224912e-05,
      "learning_rate": 3.336120401337793e-06,
      "loss": 0.0075,
      "step": 100500
    },
    {
      "epoch": 10.06,
      "grad_norm": 2.802386188704986e-05,
      "learning_rate": 3.3344481605351177e-06,
      "loss": 0.0082,
      "step": 100600
    },
    {
      "epoch": 10.07,
      "grad_norm": 9.355850488645956e-05,
      "learning_rate": 3.3327759197324418e-06,
      "loss": 0.0,
      "step": 100700
    },
    {
      "epoch": 10.08,
      "grad_norm": 2.488598329364322e-05,
      "learning_rate": 3.3311036789297662e-06,
      "loss": 0.0,
      "step": 100800
    },
    {
      "epoch": 10.09,
      "grad_norm": 1.7552529243403114e-05,
      "learning_rate": 3.3294314381270903e-06,
      "loss": 0.0057,
      "step": 100900
    },
    {
      "epoch": 10.1,
      "grad_norm": 4.751099550048821e-05,
      "learning_rate": 3.327759197324415e-06,
      "loss": 0.0,
      "step": 101000
    },
    {
      "epoch": 10.11,
      "grad_norm": 6.215456232894212e-05,
      "learning_rate": 3.3260869565217396e-06,
      "loss": 0.0,
      "step": 101100
    },
    {
      "epoch": 10.12,
      "grad_norm": 0.012414290569722652,
      "learning_rate": 3.3244147157190636e-06,
      "loss": 0.0056,
      "step": 101200
    },
    {
      "epoch": 10.13,
      "grad_norm": 5.210730159888044e-05,
      "learning_rate": 3.3227424749163885e-06,
      "loss": 0.0,
      "step": 101300
    },
    {
      "epoch": 10.14,
      "grad_norm": 6.133528222562745e-05,
      "learning_rate": 3.3210702341137126e-06,
      "loss": 0.0028,
      "step": 101400
    },
    {
      "epoch": 10.15,
      "grad_norm": 0.003408723743632436,
      "learning_rate": 3.319397993311037e-06,
      "loss": 0.0002,
      "step": 101500
    },
    {
      "epoch": 10.16,
      "grad_norm": 3.5033637686865404e-05,
      "learning_rate": 3.317725752508362e-06,
      "loss": 0.0003,
      "step": 101600
    },
    {
      "epoch": 10.17,
      "grad_norm": 4.893932418781333e-05,
      "learning_rate": 3.316053511705686e-06,
      "loss": 0.0021,
      "step": 101700
    },
    {
      "epoch": 10.18,
      "grad_norm": 3.814970114035532e-05,
      "learning_rate": 3.3143812709030104e-06,
      "loss": 0.0024,
      "step": 101800
    },
    {
      "epoch": 10.19,
      "grad_norm": 1.5137900845729746e-05,
      "learning_rate": 3.3127090301003344e-06,
      "loss": 0.0017,
      "step": 101900
    },
    {
      "epoch": 10.2,
      "grad_norm": 2.083205254166387e-05,
      "learning_rate": 3.3110367892976593e-06,
      "loss": 0.0104,
      "step": 102000
    },
    {
      "epoch": 10.21,
      "grad_norm": 0.003052659798413515,
      "learning_rate": 3.3093645484949838e-06,
      "loss": 0.0001,
      "step": 102100
    },
    {
      "epoch": 10.22,
      "grad_norm": 2.759070230240468e-05,
      "learning_rate": 3.307692307692308e-06,
      "loss": 0.0047,
      "step": 102200
    },
    {
      "epoch": 10.23,
      "grad_norm": 2.606885027489625e-05,
      "learning_rate": 3.3060200668896327e-06,
      "loss": 0.0002,
      "step": 102300
    },
    {
      "epoch": 10.24,
      "grad_norm": 1.9891685951733962e-05,
      "learning_rate": 3.3043478260869567e-06,
      "loss": 0.0,
      "step": 102400
    },
    {
      "epoch": 10.25,
      "grad_norm": 0.0007487214170396328,
      "learning_rate": 3.302675585284281e-06,
      "loss": 0.0,
      "step": 102500
    },
    {
      "epoch": 10.26,
      "grad_norm": 1.7976259186980315e-05,
      "learning_rate": 3.3010033444816057e-06,
      "loss": 0.0057,
      "step": 102600
    },
    {
      "epoch": 10.27,
      "grad_norm": 1.4554972949554212e-05,
      "learning_rate": 3.29933110367893e-06,
      "loss": 0.0,
      "step": 102700
    },
    {
      "epoch": 10.28,
      "grad_norm": 4.100278238183819e-05,
      "learning_rate": 3.2976588628762546e-06,
      "loss": 0.0,
      "step": 102800
    },
    {
      "epoch": 10.29,
      "grad_norm": 1.692805381026119e-05,
      "learning_rate": 3.2959866220735786e-06,
      "loss": 0.0046,
      "step": 102900
    },
    {
      "epoch": 10.3,
      "grad_norm": 5.216592035139911e-05,
      "learning_rate": 3.2943143812709035e-06,
      "loss": 0.0004,
      "step": 103000
    },
    {
      "epoch": 10.31,
      "grad_norm": 1.638688081584405e-05,
      "learning_rate": 3.292642140468228e-06,
      "loss": 0.005,
      "step": 103100
    },
    {
      "epoch": 10.32,
      "grad_norm": 1.8229413399240002e-05,
      "learning_rate": 3.290969899665552e-06,
      "loss": 0.0034,
      "step": 103200
    },
    {
      "epoch": 10.33,
      "grad_norm": 2.5418792574782856e-05,
      "learning_rate": 3.2892976588628765e-06,
      "loss": 0.0,
      "step": 103300
    },
    {
      "epoch": 10.34,
      "grad_norm": 74.45674896240234,
      "learning_rate": 3.287625418060201e-06,
      "loss": 0.0064,
      "step": 103400
    },
    {
      "epoch": 10.35,
      "grad_norm": 1.6789086657809094e-05,
      "learning_rate": 3.2859531772575254e-06,
      "loss": 0.0046,
      "step": 103500
    },
    {
      "epoch": 10.36,
      "grad_norm": 4.680682104662992e-05,
      "learning_rate": 3.28428093645485e-06,
      "loss": 0.0043,
      "step": 103600
    },
    {
      "epoch": 10.37,
      "grad_norm": 2.3295109713217244e-05,
      "learning_rate": 3.2826086956521743e-06,
      "loss": 0.0049,
      "step": 103700
    },
    {
      "epoch": 10.38,
      "grad_norm": 4.589736636262387e-05,
      "learning_rate": 3.2809364548494988e-06,
      "loss": 0.0,
      "step": 103800
    },
    {
      "epoch": 10.39,
      "grad_norm": 2.4069267965387553e-05,
      "learning_rate": 3.279264214046823e-06,
      "loss": 0.0,
      "step": 103900
    },
    {
      "epoch": 10.4,
      "grad_norm": 1.4721795196237508e-05,
      "learning_rate": 3.2775919732441473e-06,
      "loss": 0.0,
      "step": 104000
    },
    {
      "epoch": 10.41,
      "grad_norm": 1.2814829460694455e-05,
      "learning_rate": 3.275919732441472e-06,
      "loss": 0.0,
      "step": 104100
    },
    {
      "epoch": 10.42,
      "grad_norm": 1.1469286619103514e-05,
      "learning_rate": 3.274247491638796e-06,
      "loss": 0.0066,
      "step": 104200
    },
    {
      "epoch": 10.43,
      "grad_norm": 2.0519388272077776e-05,
      "learning_rate": 3.2725752508361206e-06,
      "loss": 0.0025,
      "step": 104300
    },
    {
      "epoch": 10.44,
      "grad_norm": 0.004769354593008757,
      "learning_rate": 3.270903010033445e-06,
      "loss": 0.0079,
      "step": 104400
    },
    {
      "epoch": 10.45,
      "grad_norm": 2.9404161978163756e-05,
      "learning_rate": 3.2692307692307696e-06,
      "loss": 0.0028,
      "step": 104500
    },
    {
      "epoch": 10.46,
      "grad_norm": 1.2868718840763904e-05,
      "learning_rate": 3.267558528428094e-06,
      "loss": 0.0,
      "step": 104600
    },
    {
      "epoch": 10.47,
      "grad_norm": 2.208415389759466e-05,
      "learning_rate": 3.265886287625418e-06,
      "loss": 0.0033,
      "step": 104700
    },
    {
      "epoch": 10.48,
      "grad_norm": 1.746260022628121e-05,
      "learning_rate": 3.264214046822743e-06,
      "loss": 0.0078,
      "step": 104800
    },
    {
      "epoch": 10.49,
      "grad_norm": 3.7256126233842224e-05,
      "learning_rate": 3.262541806020067e-06,
      "loss": 0.0,
      "step": 104900
    },
    {
      "epoch": 10.5,
      "grad_norm": 1.3822974324284587e-05,
      "learning_rate": 3.2608695652173914e-06,
      "loss": 0.0001,
      "step": 105000
    },
    {
      "epoch": 10.5,
      "eval_accuracy": 0.986725,
      "eval_f1": 0.986725,
      "eval_loss": 0.1294260323047638,
      "eval_runtime": 135.4142,
      "eval_samples_per_second": 295.39,
      "eval_steps_per_second": 295.39,
      "step": 105000
    },
    {
      "epoch": 10.51,
      "grad_norm": 1.273143516300479e-05,
      "learning_rate": 3.2591973244147163e-06,
      "loss": 0.0,
      "step": 105100
    },
    {
      "epoch": 10.52,
      "grad_norm": 1.8719376384979114e-05,
      "learning_rate": 3.2575250836120404e-06,
      "loss": 0.0003,
      "step": 105200
    },
    {
      "epoch": 10.53,
      "grad_norm": 2.3224081814987585e-05,
      "learning_rate": 3.255852842809365e-06,
      "loss": 0.0026,
      "step": 105300
    },
    {
      "epoch": 10.54,
      "grad_norm": 0.00043168303091078997,
      "learning_rate": 3.254180602006689e-06,
      "loss": 0.002,
      "step": 105400
    },
    {
      "epoch": 10.55,
      "grad_norm": 1.3707303878618404e-05,
      "learning_rate": 3.2525083612040137e-06,
      "loss": 0.0034,
      "step": 105500
    },
    {
      "epoch": 10.56,
      "grad_norm": 9.85491533356253e-06,
      "learning_rate": 3.250836120401338e-06,
      "loss": 0.0,
      "step": 105600
    },
    {
      "epoch": 10.57,
      "grad_norm": 0.00011145220196340233,
      "learning_rate": 3.2491638795986622e-06,
      "loss": 0.0059,
      "step": 105700
    },
    {
      "epoch": 10.58,
      "grad_norm": 2.3184100427897647e-05,
      "learning_rate": 3.247491638795987e-06,
      "loss": 0.0065,
      "step": 105800
    },
    {
      "epoch": 10.59,
      "grad_norm": 1.6724783563404344e-05,
      "learning_rate": 3.245819397993311e-06,
      "loss": 0.0037,
      "step": 105900
    },
    {
      "epoch": 10.6,
      "grad_norm": 1.4956675840949174e-05,
      "learning_rate": 3.2441471571906356e-06,
      "loss": 0.0074,
      "step": 106000
    },
    {
      "epoch": 10.61,
      "grad_norm": 3.993123027612455e-05,
      "learning_rate": 3.2424749163879597e-06,
      "loss": 0.0062,
      "step": 106100
    },
    {
      "epoch": 10.62,
      "grad_norm": 4.653423820855096e-05,
      "learning_rate": 3.2408026755852845e-06,
      "loss": 0.0011,
      "step": 106200
    },
    {
      "epoch": 10.63,
      "grad_norm": 0.0006327496375888586,
      "learning_rate": 3.239130434782609e-06,
      "loss": 0.0,
      "step": 106300
    },
    {
      "epoch": 10.64,
      "grad_norm": 3.2069776352727786e-05,
      "learning_rate": 3.237458193979933e-06,
      "loss": 0.0181,
      "step": 106400
    },
    {
      "epoch": 10.65,
      "grad_norm": 0.0016070237616077065,
      "learning_rate": 3.235785953177258e-06,
      "loss": 0.0018,
      "step": 106500
    },
    {
      "epoch": 10.66,
      "grad_norm": 0.00011340404307702556,
      "learning_rate": 3.234113712374582e-06,
      "loss": 0.0,
      "step": 106600
    },
    {
      "epoch": 10.67,
      "grad_norm": 1.2554538443509955e-05,
      "learning_rate": 3.2324414715719064e-06,
      "loss": 0.0,
      "step": 106700
    },
    {
      "epoch": 10.68,
      "grad_norm": 0.0037620030343532562,
      "learning_rate": 3.2307692307692313e-06,
      "loss": 0.0,
      "step": 106800
    },
    {
      "epoch": 10.69,
      "grad_norm": 1.1234938938287087e-05,
      "learning_rate": 3.2290969899665553e-06,
      "loss": 0.0,
      "step": 106900
    },
    {
      "epoch": 10.7,
      "grad_norm": 9.472200872551184e-06,
      "learning_rate": 3.22742474916388e-06,
      "loss": 0.0,
      "step": 107000
    },
    {
      "epoch": 10.71,
      "grad_norm": 1.4786013707634993e-05,
      "learning_rate": 3.225752508361204e-06,
      "loss": 0.0001,
      "step": 107100
    },
    {
      "epoch": 10.72,
      "grad_norm": 0.00032294646371155977,
      "learning_rate": 3.2240802675585287e-06,
      "loss": 0.0033,
      "step": 107200
    },
    {
      "epoch": 10.73,
      "grad_norm": 3.9019418181851506e-05,
      "learning_rate": 3.222408026755853e-06,
      "loss": 0.0047,
      "step": 107300
    },
    {
      "epoch": 10.74,
      "grad_norm": 2.2364813048625365e-05,
      "learning_rate": 3.2207357859531772e-06,
      "loss": 0.005,
      "step": 107400
    },
    {
      "epoch": 10.75,
      "grad_norm": 9.266052984457929e-06,
      "learning_rate": 3.219063545150502e-06,
      "loss": 0.0,
      "step": 107500
    },
    {
      "epoch": 10.76,
      "grad_norm": 2.405466693744529e-05,
      "learning_rate": 3.217391304347826e-06,
      "loss": 0.0011,
      "step": 107600
    },
    {
      "epoch": 10.77,
      "grad_norm": 1.7374895833199844e-05,
      "learning_rate": 3.2157190635451506e-06,
      "loss": 0.0,
      "step": 107700
    },
    {
      "epoch": 10.78,
      "grad_norm": 1.3817982107866555e-05,
      "learning_rate": 3.2140468227424755e-06,
      "loss": 0.0109,
      "step": 107800
    },
    {
      "epoch": 10.79,
      "grad_norm": 1.589250678080134e-05,
      "learning_rate": 3.2123745819397995e-06,
      "loss": 0.0,
      "step": 107900
    },
    {
      "epoch": 10.8,
      "grad_norm": 5.421064997790381e-05,
      "learning_rate": 3.210702341137124e-06,
      "loss": 0.0,
      "step": 108000
    },
    {
      "epoch": 10.81,
      "grad_norm": 0.00010446009400766343,
      "learning_rate": 3.209030100334448e-06,
      "loss": 0.0,
      "step": 108100
    },
    {
      "epoch": 10.82,
      "grad_norm": 1.0952360753435642e-05,
      "learning_rate": 3.207357859531773e-06,
      "loss": 0.0019,
      "step": 108200
    },
    {
      "epoch": 10.83,
      "grad_norm": 0.0003366768651176244,
      "learning_rate": 3.2056856187290974e-06,
      "loss": 0.0082,
      "step": 108300
    },
    {
      "epoch": 10.84,
      "grad_norm": 0.004733449313789606,
      "learning_rate": 3.2040133779264214e-06,
      "loss": 0.0,
      "step": 108400
    },
    {
      "epoch": 10.85,
      "grad_norm": 9.548472007736564e-05,
      "learning_rate": 3.2023411371237463e-06,
      "loss": 0.0,
      "step": 108500
    },
    {
      "epoch": 10.86,
      "grad_norm": 0.00032680193544365466,
      "learning_rate": 3.2006688963210703e-06,
      "loss": 0.0032,
      "step": 108600
    },
    {
      "epoch": 10.87,
      "grad_norm": 2.7095329642179422e-05,
      "learning_rate": 3.1989966555183948e-06,
      "loss": 0.0032,
      "step": 108700
    },
    {
      "epoch": 10.88,
      "grad_norm": 3.111539990641177e-05,
      "learning_rate": 3.1973244147157197e-06,
      "loss": 0.0044,
      "step": 108800
    },
    {
      "epoch": 10.89,
      "grad_norm": 0.0001842112251324579,
      "learning_rate": 3.1956521739130437e-06,
      "loss": 0.0,
      "step": 108900
    },
    {
      "epoch": 10.9,
      "grad_norm": 4.6635905164293945e-05,
      "learning_rate": 3.193979933110368e-06,
      "loss": 0.0046,
      "step": 109000
    },
    {
      "epoch": 10.91,
      "grad_norm": 9.20000093174167e-06,
      "learning_rate": 3.192307692307692e-06,
      "loss": 0.0,
      "step": 109100
    },
    {
      "epoch": 10.92,
      "grad_norm": 1.49307834362844e-05,
      "learning_rate": 3.190635451505017e-06,
      "loss": 0.0068,
      "step": 109200
    },
    {
      "epoch": 10.93,
      "grad_norm": 1.689198506937828e-05,
      "learning_rate": 3.1889632107023415e-06,
      "loss": 0.0048,
      "step": 109300
    },
    {
      "epoch": 10.94,
      "grad_norm": 0.0002182431344408542,
      "learning_rate": 3.1872909698996656e-06,
      "loss": 0.0,
      "step": 109400
    },
    {
      "epoch": 10.95,
      "grad_norm": 1.8919832655228674e-05,
      "learning_rate": 3.1856187290969905e-06,
      "loss": 0.0,
      "step": 109500
    },
    {
      "epoch": 10.96,
      "grad_norm": 0.0008669137023389339,
      "learning_rate": 3.1839464882943145e-06,
      "loss": 0.0029,
      "step": 109600
    },
    {
      "epoch": 10.97,
      "grad_norm": 7.572957838419825e-05,
      "learning_rate": 3.182274247491639e-06,
      "loss": 0.0,
      "step": 109700
    },
    {
      "epoch": 10.98,
      "grad_norm": 0.17629992961883545,
      "learning_rate": 3.180602006688964e-06,
      "loss": 0.0002,
      "step": 109800
    },
    {
      "epoch": 10.99,
      "grad_norm": 1.813526796468068e-05,
      "learning_rate": 3.178929765886288e-06,
      "loss": 0.0052,
      "step": 109900
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.00025834888219833374,
      "learning_rate": 3.1772575250836123e-06,
      "loss": 0.0017,
      "step": 110000
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.98675,
      "eval_f1": 0.98675,
      "eval_loss": 0.13637404143810272,
      "eval_runtime": 133.3484,
      "eval_samples_per_second": 299.966,
      "eval_steps_per_second": 299.966,
      "step": 110000
    },
    {
      "epoch": 11.01,
      "grad_norm": 2.8174052204121836e-05,
      "learning_rate": 3.1755852842809364e-06,
      "loss": 0.0036,
      "step": 110100
    },
    {
      "epoch": 11.02,
      "grad_norm": 6.247372948564589e-05,
      "learning_rate": 3.1739130434782613e-06,
      "loss": 0.0021,
      "step": 110200
    },
    {
      "epoch": 11.03,
      "grad_norm": 1.2338041415205225e-05,
      "learning_rate": 3.1722408026755857e-06,
      "loss": 0.0,
      "step": 110300
    },
    {
      "epoch": 11.04,
      "grad_norm": 3.7150435673538595e-05,
      "learning_rate": 3.1705685618729098e-06,
      "loss": 0.0053,
      "step": 110400
    },
    {
      "epoch": 11.05,
      "grad_norm": 0.01229814998805523,
      "learning_rate": 3.1688963210702347e-06,
      "loss": 0.0,
      "step": 110500
    },
    {
      "epoch": 11.06,
      "grad_norm": 7.667613317607902e-06,
      "learning_rate": 3.1672240802675587e-06,
      "loss": 0.0041,
      "step": 110600
    },
    {
      "epoch": 11.07,
      "grad_norm": 1.0412702977191657e-05,
      "learning_rate": 3.165551839464883e-06,
      "loss": 0.0,
      "step": 110700
    },
    {
      "epoch": 11.08,
      "grad_norm": 9.122279152506962e-06,
      "learning_rate": 3.163879598662208e-06,
      "loss": 0.0,
      "step": 110800
    },
    {
      "epoch": 11.09,
      "grad_norm": 2.4620214389869943e-05,
      "learning_rate": 3.162207357859532e-06,
      "loss": 0.0002,
      "step": 110900
    },
    {
      "epoch": 11.1,
      "grad_norm": 0.002043084939941764,
      "learning_rate": 3.1605351170568565e-06,
      "loss": 0.0028,
      "step": 111000
    },
    {
      "epoch": 11.11,
      "grad_norm": 7.240785180329112e-06,
      "learning_rate": 3.1588628762541806e-06,
      "loss": 0.0,
      "step": 111100
    },
    {
      "epoch": 11.12,
      "grad_norm": 1.0734827810665593e-05,
      "learning_rate": 3.1571906354515055e-06,
      "loss": 0.0,
      "step": 111200
    },
    {
      "epoch": 11.13,
      "grad_norm": 8.010079909581691e-05,
      "learning_rate": 3.15551839464883e-06,
      "loss": 0.0064,
      "step": 111300
    },
    {
      "epoch": 11.14,
      "grad_norm": 7.866760824981611e-06,
      "learning_rate": 3.153846153846154e-06,
      "loss": 0.0022,
      "step": 111400
    },
    {
      "epoch": 11.15,
      "grad_norm": 9.745773240865674e-06,
      "learning_rate": 3.152173913043479e-06,
      "loss": 0.0,
      "step": 111500
    },
    {
      "epoch": 11.16,
      "grad_norm": 2.4606217266409658e-05,
      "learning_rate": 3.150501672240803e-06,
      "loss": 0.0121,
      "step": 111600
    },
    {
      "epoch": 11.17,
      "grad_norm": 0.0005854077753610909,
      "learning_rate": 3.1488294314381273e-06,
      "loss": 0.0036,
      "step": 111700
    },
    {
      "epoch": 11.18,
      "grad_norm": 0.00012307950237300247,
      "learning_rate": 3.147157190635452e-06,
      "loss": 0.0059,
      "step": 111800
    },
    {
      "epoch": 11.19,
      "grad_norm": 0.00010187952284468338,
      "learning_rate": 3.1454849498327763e-06,
      "loss": 0.0,
      "step": 111900
    },
    {
      "epoch": 11.2,
      "grad_norm": 2.967399086628575e-05,
      "learning_rate": 3.1438127090301007e-06,
      "loss": 0.0,
      "step": 112000
    },
    {
      "epoch": 11.21,
      "grad_norm": 8.619118307251483e-05,
      "learning_rate": 3.1421404682274247e-06,
      "loss": 0.0,
      "step": 112100
    },
    {
      "epoch": 11.22,
      "grad_norm": 3.520284371916205e-05,
      "learning_rate": 3.1404682274247496e-06,
      "loss": 0.0002,
      "step": 112200
    },
    {
      "epoch": 11.23,
      "grad_norm": 6.373329961206764e-05,
      "learning_rate": 3.138795986622074e-06,
      "loss": 0.0098,
      "step": 112300
    },
    {
      "epoch": 11.24,
      "grad_norm": 0.0014678030274808407,
      "learning_rate": 3.137123745819398e-06,
      "loss": 0.0078,
      "step": 112400
    },
    {
      "epoch": 11.25,
      "grad_norm": 0.00016898875765036792,
      "learning_rate": 3.1354515050167226e-06,
      "loss": 0.0127,
      "step": 112500
    },
    {
      "epoch": 11.26,
      "grad_norm": 1.7103437130572274e-05,
      "learning_rate": 3.133779264214047e-06,
      "loss": 0.0,
      "step": 112600
    },
    {
      "epoch": 11.27,
      "grad_norm": 3.9083075535018e-05,
      "learning_rate": 3.1321070234113715e-06,
      "loss": 0.0069,
      "step": 112700
    },
    {
      "epoch": 11.28,
      "grad_norm": 0.11483018845319748,
      "learning_rate": 3.130434782608696e-06,
      "loss": 0.0003,
      "step": 112800
    },
    {
      "epoch": 11.29,
      "grad_norm": 0.09915011376142502,
      "learning_rate": 3.1287625418060204e-06,
      "loss": 0.0035,
      "step": 112900
    },
    {
      "epoch": 11.3,
      "grad_norm": 0.0055434671230614185,
      "learning_rate": 3.127090301003345e-06,
      "loss": 0.005,
      "step": 113000
    },
    {
      "epoch": 11.31,
      "grad_norm": 2.727087849052623e-05,
      "learning_rate": 3.125418060200669e-06,
      "loss": 0.0036,
      "step": 113100
    },
    {
      "epoch": 11.32,
      "grad_norm": 4.718463242170401e-05,
      "learning_rate": 3.1237458193979934e-06,
      "loss": 0.0,
      "step": 113200
    },
    {
      "epoch": 11.33,
      "grad_norm": 3.375914820935577e-05,
      "learning_rate": 3.1220735785953183e-06,
      "loss": 0.0031,
      "step": 113300
    },
    {
      "epoch": 11.34,
      "grad_norm": 1.1921174518647604e-05,
      "learning_rate": 3.1204013377926423e-06,
      "loss": 0.0,
      "step": 113400
    },
    {
      "epoch": 11.35,
      "grad_norm": 0.00011794619058491662,
      "learning_rate": 3.1187290969899668e-06,
      "loss": 0.0,
      "step": 113500
    },
    {
      "epoch": 11.36,
      "grad_norm": 0.0018074008403345942,
      "learning_rate": 3.1170568561872912e-06,
      "loss": 0.0038,
      "step": 113600
    },
    {
      "epoch": 11.37,
      "grad_norm": 1.8831220586434938e-05,
      "learning_rate": 3.1153846153846157e-06,
      "loss": 0.0031,
      "step": 113700
    },
    {
      "epoch": 11.38,
      "grad_norm": 0.00034093132126145065,
      "learning_rate": 3.11371237458194e-06,
      "loss": 0.0075,
      "step": 113800
    },
    {
      "epoch": 11.39,
      "grad_norm": 0.0013741998700425029,
      "learning_rate": 3.112040133779264e-06,
      "loss": 0.0067,
      "step": 113900
    },
    {
      "epoch": 11.4,
      "grad_norm": 1.7773139916243963e-05,
      "learning_rate": 3.110367892976589e-06,
      "loss": 0.0066,
      "step": 114000
    },
    {
      "epoch": 11.41,
      "grad_norm": 0.000126648650621064,
      "learning_rate": 3.108695652173913e-06,
      "loss": 0.0066,
      "step": 114100
    },
    {
      "epoch": 11.42,
      "grad_norm": 0.0001225349842570722,
      "learning_rate": 3.1070234113712376e-06,
      "loss": 0.0035,
      "step": 114200
    },
    {
      "epoch": 11.43,
      "grad_norm": 5.7315519370604306e-05,
      "learning_rate": 3.1053511705685625e-06,
      "loss": 0.0034,
      "step": 114300
    },
    {
      "epoch": 11.44,
      "grad_norm": 3.917821231880225e-05,
      "learning_rate": 3.1036789297658865e-06,
      "loss": 0.0043,
      "step": 114400
    },
    {
      "epoch": 11.45,
      "grad_norm": 0.0019230084726586938,
      "learning_rate": 3.102006688963211e-06,
      "loss": 0.0016,
      "step": 114500
    },
    {
      "epoch": 11.46,
      "grad_norm": 2.3124792278395034e-05,
      "learning_rate": 3.100334448160535e-06,
      "loss": 0.0,
      "step": 114600
    },
    {
      "epoch": 11.47,
      "grad_norm": 3.644728712970391e-05,
      "learning_rate": 3.09866220735786e-06,
      "loss": 0.0,
      "step": 114700
    },
    {
      "epoch": 11.48,
      "grad_norm": 0.04555916041135788,
      "learning_rate": 3.0969899665551843e-06,
      "loss": 0.0026,
      "step": 114800
    },
    {
      "epoch": 11.49,
      "grad_norm": 3.428834315855056e-05,
      "learning_rate": 3.0953177257525084e-06,
      "loss": 0.0017,
      "step": 114900
    },
    {
      "epoch": 11.5,
      "grad_norm": 1.4589694728783797e-05,
      "learning_rate": 3.0936454849498333e-06,
      "loss": 0.0033,
      "step": 115000
    },
    {
      "epoch": 11.5,
      "eval_accuracy": 0.987075,
      "eval_f1": 0.987075,
      "eval_loss": 0.12469487637281418,
      "eval_runtime": 134.1209,
      "eval_samples_per_second": 298.238,
      "eval_steps_per_second": 298.238,
      "step": 115000
    },
    {
      "epoch": 11.51,
      "grad_norm": 1.1125358469143976e-05,
      "learning_rate": 3.0919732441471573e-06,
      "loss": 0.0,
      "step": 115100
    },
    {
      "epoch": 11.52,
      "grad_norm": 2.2134607206680812e-05,
      "learning_rate": 3.0903010033444818e-06,
      "loss": 0.0063,
      "step": 115200
    },
    {
      "epoch": 11.53,
      "grad_norm": 1.5470241123693995e-05,
      "learning_rate": 3.0886287625418066e-06,
      "loss": 0.0043,
      "step": 115300
    },
    {
      "epoch": 11.54,
      "grad_norm": 2.767572550510522e-05,
      "learning_rate": 3.0869565217391307e-06,
      "loss": 0.0046,
      "step": 115400
    },
    {
      "epoch": 11.55,
      "grad_norm": 4.574981358018704e-05,
      "learning_rate": 3.085284280936455e-06,
      "loss": 0.0018,
      "step": 115500
    },
    {
      "epoch": 11.56,
      "grad_norm": 0.00040606813854537904,
      "learning_rate": 3.083612040133779e-06,
      "loss": 0.009,
      "step": 115600
    },
    {
      "epoch": 11.57,
      "grad_norm": 8.138959674397483e-05,
      "learning_rate": 3.081939799331104e-06,
      "loss": 0.0,
      "step": 115700
    },
    {
      "epoch": 11.58,
      "grad_norm": 0.0002813829341903329,
      "learning_rate": 3.080267558528428e-06,
      "loss": 0.007,
      "step": 115800
    },
    {
      "epoch": 11.59,
      "grad_norm": 0.00013545318506658077,
      "learning_rate": 3.0785953177257526e-06,
      "loss": 0.0056,
      "step": 115900
    },
    {
      "epoch": 11.6,
      "grad_norm": 8.718832395970821e-05,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.0074,
      "step": 116000
    },
    {
      "epoch": 11.61,
      "grad_norm": 0.049765247851610184,
      "learning_rate": 3.0752508361204015e-06,
      "loss": 0.0013,
      "step": 116100
    },
    {
      "epoch": 11.62,
      "grad_norm": 0.12891064584255219,
      "learning_rate": 3.073578595317726e-06,
      "loss": 0.0038,
      "step": 116200
    },
    {
      "epoch": 11.63,
      "grad_norm": 0.010939703322947025,
      "learning_rate": 3.07190635451505e-06,
      "loss": 0.0016,
      "step": 116300
    },
    {
      "epoch": 11.64,
      "grad_norm": 0.0008908117306418717,
      "learning_rate": 3.070234113712375e-06,
      "loss": 0.0042,
      "step": 116400
    },
    {
      "epoch": 11.65,
      "grad_norm": 4.410389010445215e-05,
      "learning_rate": 3.0685618729096993e-06,
      "loss": 0.0019,
      "step": 116500
    },
    {
      "epoch": 11.66,
      "grad_norm": 3.8732574466848746e-05,
      "learning_rate": 3.0668896321070234e-06,
      "loss": 0.0022,
      "step": 116600
    },
    {
      "epoch": 11.67,
      "grad_norm": 4.904616071144119e-05,
      "learning_rate": 3.0652173913043482e-06,
      "loss": 0.0032,
      "step": 116700
    },
    {
      "epoch": 11.68,
      "grad_norm": 5.659882663167082e-05,
      "learning_rate": 3.0635451505016723e-06,
      "loss": 0.0079,
      "step": 116800
    },
    {
      "epoch": 11.69,
      "grad_norm": 4.263202572474256e-05,
      "learning_rate": 3.0618729096989967e-06,
      "loss": 0.0058,
      "step": 116900
    },
    {
      "epoch": 11.7,
      "grad_norm": 2.6740566681837663e-05,
      "learning_rate": 3.0602006688963216e-06,
      "loss": 0.0,
      "step": 117000
    },
    {
      "epoch": 11.71,
      "grad_norm": 0.0026203403249382973,
      "learning_rate": 3.0585284280936457e-06,
      "loss": 0.007,
      "step": 117100
    },
    {
      "epoch": 11.72,
      "grad_norm": 4.1611296182964e-05,
      "learning_rate": 3.05685618729097e-06,
      "loss": 0.0,
      "step": 117200
    },
    {
      "epoch": 11.73,
      "grad_norm": 0.0006726513965986669,
      "learning_rate": 3.055183946488294e-06,
      "loss": 0.0,
      "step": 117300
    },
    {
      "epoch": 11.74,
      "grad_norm": 0.0001233662769664079,
      "learning_rate": 3.053511705685619e-06,
      "loss": 0.0,
      "step": 117400
    },
    {
      "epoch": 11.75,
      "grad_norm": 1.4310979167930782e-05,
      "learning_rate": 3.0518394648829435e-06,
      "loss": 0.0,
      "step": 117500
    },
    {
      "epoch": 11.76,
      "grad_norm": 7.153803016990423e-05,
      "learning_rate": 3.0501672240802675e-06,
      "loss": 0.0,
      "step": 117600
    },
    {
      "epoch": 11.77,
      "grad_norm": 2.3614837118657306e-05,
      "learning_rate": 3.0484949832775924e-06,
      "loss": 0.0027,
      "step": 117700
    },
    {
      "epoch": 11.78,
      "grad_norm": 0.00029216796974651515,
      "learning_rate": 3.0468227424749165e-06,
      "loss": 0.0041,
      "step": 117800
    },
    {
      "epoch": 11.79,
      "grad_norm": 2.791116821754258e-05,
      "learning_rate": 3.045150501672241e-06,
      "loss": 0.0,
      "step": 117900
    },
    {
      "epoch": 11.8,
      "grad_norm": 4.38535789726302e-05,
      "learning_rate": 3.043478260869566e-06,
      "loss": 0.0037,
      "step": 118000
    },
    {
      "epoch": 11.81,
      "grad_norm": 0.00027548184152692556,
      "learning_rate": 3.04180602006689e-06,
      "loss": 0.0092,
      "step": 118100
    },
    {
      "epoch": 11.82,
      "grad_norm": 7.698433910263702e-05,
      "learning_rate": 3.0401337792642143e-06,
      "loss": 0.0002,
      "step": 118200
    },
    {
      "epoch": 11.83,
      "grad_norm": 2.1024501620559022e-05,
      "learning_rate": 3.0384615384615383e-06,
      "loss": 0.0,
      "step": 118300
    },
    {
      "epoch": 11.84,
      "grad_norm": 0.0005384383839555085,
      "learning_rate": 3.0367892976588632e-06,
      "loss": 0.0,
      "step": 118400
    },
    {
      "epoch": 11.85,
      "grad_norm": 3.6589870433090255e-05,
      "learning_rate": 3.0351170568561877e-06,
      "loss": 0.0,
      "step": 118500
    },
    {
      "epoch": 11.86,
      "grad_norm": 1.6869857063284144e-05,
      "learning_rate": 3.0334448160535117e-06,
      "loss": 0.0,
      "step": 118600
    },
    {
      "epoch": 11.87,
      "grad_norm": 1.1717855159076862e-05,
      "learning_rate": 3.0317725752508366e-06,
      "loss": 0.0,
      "step": 118700
    },
    {
      "epoch": 11.88,
      "grad_norm": 2.092311297019478e-05,
      "learning_rate": 3.0301003344481606e-06,
      "loss": 0.0,
      "step": 118800
    },
    {
      "epoch": 11.89,
      "grad_norm": 0.00024131260579451919,
      "learning_rate": 3.028428093645485e-06,
      "loss": 0.0,
      "step": 118900
    },
    {
      "epoch": 11.9,
      "grad_norm": 5.391419108491391e-05,
      "learning_rate": 3.02675585284281e-06,
      "loss": 0.0038,
      "step": 119000
    },
    {
      "epoch": 11.91,
      "grad_norm": 0.00020960722758900374,
      "learning_rate": 3.025083612040134e-06,
      "loss": 0.0053,
      "step": 119100
    },
    {
      "epoch": 11.92,
      "grad_norm": 0.00011951918713748455,
      "learning_rate": 3.0234113712374585e-06,
      "loss": 0.004,
      "step": 119200
    },
    {
      "epoch": 11.93,
      "grad_norm": 3.663140887510963e-05,
      "learning_rate": 3.0217391304347825e-06,
      "loss": 0.0,
      "step": 119300
    },
    {
      "epoch": 11.94,
      "grad_norm": 4.4150354369776323e-05,
      "learning_rate": 3.0200668896321074e-06,
      "loss": 0.0044,
      "step": 119400
    },
    {
      "epoch": 11.95,
      "grad_norm": 5.2497161959763616e-05,
      "learning_rate": 3.018394648829432e-06,
      "loss": 0.0024,
      "step": 119500
    },
    {
      "epoch": 11.96,
      "grad_norm": 4.427917883731425e-05,
      "learning_rate": 3.016722408026756e-06,
      "loss": 0.0,
      "step": 119600
    },
    {
      "epoch": 11.97,
      "grad_norm": 2.8848799047409557e-05,
      "learning_rate": 3.0150501672240808e-06,
      "loss": 0.0018,
      "step": 119700
    },
    {
      "epoch": 11.98,
      "grad_norm": 1.0785791346279439e-05,
      "learning_rate": 3.013377926421405e-06,
      "loss": 0.0,
      "step": 119800
    },
    {
      "epoch": 11.99,
      "grad_norm": 33.651451110839844,
      "learning_rate": 3.0117056856187293e-06,
      "loss": 0.006,
      "step": 119900
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.00619711447507143,
      "learning_rate": 3.010033444816054e-06,
      "loss": 0.0,
      "step": 120000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9865,
      "eval_f1": 0.9865,
      "eval_loss": 0.1380569487810135,
      "eval_runtime": 133.986,
      "eval_samples_per_second": 298.539,
      "eval_steps_per_second": 298.539,
      "step": 120000
    },
    {
      "epoch": 12.01,
      "grad_norm": 1.0006710908783134e-05,
      "learning_rate": 3.008361204013378e-06,
      "loss": 0.0,
      "step": 120100
    },
    {
      "epoch": 12.02,
      "grad_norm": 0.0002895425714086741,
      "learning_rate": 3.0066889632107027e-06,
      "loss": 0.0073,
      "step": 120200
    },
    {
      "epoch": 12.03,
      "grad_norm": 6.075893543311395e-05,
      "learning_rate": 3.0050167224080267e-06,
      "loss": 0.0046,
      "step": 120300
    },
    {
      "epoch": 12.04,
      "grad_norm": 2.0078807210666128e-05,
      "learning_rate": 3.0033444816053516e-06,
      "loss": 0.0,
      "step": 120400
    },
    {
      "epoch": 12.05,
      "grad_norm": 1.8073727915179916e-05,
      "learning_rate": 3.001672240802676e-06,
      "loss": 0.0,
      "step": 120500
    },
    {
      "epoch": 12.06,
      "grad_norm": 2.0246203348506242e-05,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 120600
    },
    {
      "epoch": 12.07,
      "grad_norm": 0.0022137134801596403,
      "learning_rate": 2.998327759197325e-06,
      "loss": 0.0048,
      "step": 120700
    },
    {
      "epoch": 12.08,
      "grad_norm": 1.632177190913353e-05,
      "learning_rate": 2.996655518394649e-06,
      "loss": 0.0051,
      "step": 120800
    },
    {
      "epoch": 12.09,
      "grad_norm": 0.009438976645469666,
      "learning_rate": 2.9949832775919735e-06,
      "loss": 0.0046,
      "step": 120900
    },
    {
      "epoch": 12.1,
      "grad_norm": 2.1821919290232472e-05,
      "learning_rate": 2.9933110367892983e-06,
      "loss": 0.0003,
      "step": 121000
    },
    {
      "epoch": 12.11,
      "grad_norm": 0.00013410099199973047,
      "learning_rate": 2.9916387959866224e-06,
      "loss": 0.0023,
      "step": 121100
    },
    {
      "epoch": 12.12,
      "grad_norm": 3.4450891689630225e-05,
      "learning_rate": 2.989966555183947e-06,
      "loss": 0.0001,
      "step": 121200
    },
    {
      "epoch": 12.13,
      "grad_norm": 0.010932223871350288,
      "learning_rate": 2.988294314381271e-06,
      "loss": 0.0,
      "step": 121300
    },
    {
      "epoch": 12.14,
      "grad_norm": 4.724752943729982e-05,
      "learning_rate": 2.9866220735785958e-06,
      "loss": 0.0,
      "step": 121400
    },
    {
      "epoch": 12.15,
      "grad_norm": 1.1601387996051926e-05,
      "learning_rate": 2.9849498327759202e-06,
      "loss": 0.0,
      "step": 121500
    },
    {
      "epoch": 12.16,
      "grad_norm": 4.233758227201179e-05,
      "learning_rate": 2.9832775919732443e-06,
      "loss": 0.0,
      "step": 121600
    },
    {
      "epoch": 12.17,
      "grad_norm": 2.8649757950915955e-05,
      "learning_rate": 2.9816053511705687e-06,
      "loss": 0.0,
      "step": 121700
    },
    {
      "epoch": 12.18,
      "grad_norm": 4.887601608061232e-05,
      "learning_rate": 2.979933110367893e-06,
      "loss": 0.0111,
      "step": 121800
    },
    {
      "epoch": 12.19,
      "grad_norm": 0.016880473122000694,
      "learning_rate": 2.9782608695652176e-06,
      "loss": 0.0,
      "step": 121900
    },
    {
      "epoch": 12.2,
      "grad_norm": 1.5798104868736118e-05,
      "learning_rate": 2.976588628762542e-06,
      "loss": 0.0018,
      "step": 122000
    },
    {
      "epoch": 12.21,
      "grad_norm": 2.9448170607793145e-05,
      "learning_rate": 2.9749163879598666e-06,
      "loss": 0.0,
      "step": 122100
    },
    {
      "epoch": 12.22,
      "grad_norm": 2.1456729882629588e-05,
      "learning_rate": 2.973244147157191e-06,
      "loss": 0.0,
      "step": 122200
    },
    {
      "epoch": 12.23,
      "grad_norm": 1.5617501048836857e-05,
      "learning_rate": 2.971571906354515e-06,
      "loss": 0.0,
      "step": 122300
    },
    {
      "epoch": 12.24,
      "grad_norm": 0.00031002156902104616,
      "learning_rate": 2.9698996655518395e-06,
      "loss": 0.0133,
      "step": 122400
    },
    {
      "epoch": 12.25,
      "grad_norm": 1.7442815078538842e-05,
      "learning_rate": 2.9682274247491644e-06,
      "loss": 0.0034,
      "step": 122500
    },
    {
      "epoch": 12.26,
      "grad_norm": 3.2786520023364574e-05,
      "learning_rate": 2.9665551839464884e-06,
      "loss": 0.0035,
      "step": 122600
    },
    {
      "epoch": 12.27,
      "grad_norm": 8.906285074772313e-05,
      "learning_rate": 2.964882943143813e-06,
      "loss": 0.0023,
      "step": 122700
    },
    {
      "epoch": 12.28,
      "grad_norm": 0.0013216121587902308,
      "learning_rate": 2.9632107023411374e-06,
      "loss": 0.0031,
      "step": 122800
    },
    {
      "epoch": 12.29,
      "grad_norm": 2.5754599846550263e-05,
      "learning_rate": 2.961538461538462e-06,
      "loss": 0.0068,
      "step": 122900
    },
    {
      "epoch": 12.3,
      "grad_norm": 0.000346259621437639,
      "learning_rate": 2.9598662207357863e-06,
      "loss": 0.0009,
      "step": 123000
    },
    {
      "epoch": 12.31,
      "grad_norm": 0.0002841234963852912,
      "learning_rate": 2.9581939799331103e-06,
      "loss": 0.0091,
      "step": 123100
    },
    {
      "epoch": 12.32,
      "grad_norm": 0.0008099031983874738,
      "learning_rate": 2.956521739130435e-06,
      "loss": 0.0,
      "step": 123200
    },
    {
      "epoch": 12.33,
      "grad_norm": 2.596194281068165e-05,
      "learning_rate": 2.9548494983277592e-06,
      "loss": 0.0,
      "step": 123300
    },
    {
      "epoch": 12.34,
      "grad_norm": 1.770333619788289e-05,
      "learning_rate": 2.9531772575250837e-06,
      "loss": 0.0,
      "step": 123400
    },
    {
      "epoch": 12.35,
      "grad_norm": 1.7277186998398975e-05,
      "learning_rate": 2.9515050167224086e-06,
      "loss": 0.0,
      "step": 123500
    },
    {
      "epoch": 12.36,
      "grad_norm": 0.0003047818026971072,
      "learning_rate": 2.9498327759197326e-06,
      "loss": 0.0006,
      "step": 123600
    },
    {
      "epoch": 12.37,
      "grad_norm": 4.139426528126933e-05,
      "learning_rate": 2.948160535117057e-06,
      "loss": 0.003,
      "step": 123700
    },
    {
      "epoch": 12.38,
      "grad_norm": 4.857347084907815e-05,
      "learning_rate": 2.946488294314381e-06,
      "loss": 0.0,
      "step": 123800
    },
    {
      "epoch": 12.39,
      "grad_norm": 1.1135462955280673e-05,
      "learning_rate": 2.944816053511706e-06,
      "loss": 0.0,
      "step": 123900
    },
    {
      "epoch": 12.4,
      "grad_norm": 4.5921427954453975e-05,
      "learning_rate": 2.9431438127090305e-06,
      "loss": 0.0,
      "step": 124000
    },
    {
      "epoch": 12.41,
      "grad_norm": 0.0011348113184794784,
      "learning_rate": 2.9414715719063545e-06,
      "loss": 0.0,
      "step": 124100
    },
    {
      "epoch": 12.42,
      "grad_norm": 9.929543011821806e-06,
      "learning_rate": 2.9397993311036794e-06,
      "loss": 0.0,
      "step": 124200
    },
    {
      "epoch": 12.43,
      "grad_norm": 0.0007508114213123918,
      "learning_rate": 2.9381270903010034e-06,
      "loss": 0.0023,
      "step": 124300
    },
    {
      "epoch": 12.44,
      "grad_norm": 0.0012188480468466878,
      "learning_rate": 2.936454849498328e-06,
      "loss": 0.0,
      "step": 124400
    },
    {
      "epoch": 12.45,
      "grad_norm": 1.107308162318077e-05,
      "learning_rate": 2.9347826086956528e-06,
      "loss": 0.0,
      "step": 124500
    },
    {
      "epoch": 12.46,
      "grad_norm": 1.0170252608077135e-05,
      "learning_rate": 2.933110367892977e-06,
      "loss": 0.0,
      "step": 124600
    },
    {
      "epoch": 12.47,
      "grad_norm": 9.492271055933088e-05,
      "learning_rate": 2.9314381270903013e-06,
      "loss": 0.0,
      "step": 124700
    },
    {
      "epoch": 12.48,
      "grad_norm": 0.0013374453410506248,
      "learning_rate": 2.9297658862876253e-06,
      "loss": 0.021,
      "step": 124800
    },
    {
      "epoch": 12.49,
      "grad_norm": 0.000961253943387419,
      "learning_rate": 2.92809364548495e-06,
      "loss": 0.0,
      "step": 124900
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.00014077928790356964,
      "learning_rate": 2.9264214046822746e-06,
      "loss": 0.0002,
      "step": 125000
    },
    {
      "epoch": 12.5,
      "eval_accuracy": 0.98745,
      "eval_f1": 0.98745,
      "eval_loss": 0.11790119856595993,
      "eval_runtime": 133.865,
      "eval_samples_per_second": 298.809,
      "eval_steps_per_second": 298.809,
      "step": 125000
    },
    {
      "epoch": 12.51,
      "grad_norm": 0.00034569227136671543,
      "learning_rate": 2.9247491638795987e-06,
      "loss": 0.0033,
      "step": 125100
    },
    {
      "epoch": 12.52,
      "grad_norm": 2.6383528165752068e-05,
      "learning_rate": 2.9230769230769236e-06,
      "loss": 0.0,
      "step": 125200
    },
    {
      "epoch": 12.53,
      "grad_norm": 0.00013982444943394512,
      "learning_rate": 2.9214046822742476e-06,
      "loss": 0.0,
      "step": 125300
    },
    {
      "epoch": 12.54,
      "grad_norm": 1.3336039955902379e-05,
      "learning_rate": 2.919732441471572e-06,
      "loss": 0.0,
      "step": 125400
    },
    {
      "epoch": 12.55,
      "grad_norm": 0.00027246633544564247,
      "learning_rate": 2.918060200668897e-06,
      "loss": 0.0003,
      "step": 125500
    },
    {
      "epoch": 12.56,
      "grad_norm": 9.627802501199767e-05,
      "learning_rate": 2.916387959866221e-06,
      "loss": 0.0,
      "step": 125600
    },
    {
      "epoch": 12.57,
      "grad_norm": 1.2915442312078085e-05,
      "learning_rate": 2.9147157190635454e-06,
      "loss": 0.0025,
      "step": 125700
    },
    {
      "epoch": 12.58,
      "grad_norm": 1.4243492842069827e-05,
      "learning_rate": 2.9130434782608695e-06,
      "loss": 0.0,
      "step": 125800
    },
    {
      "epoch": 12.59,
      "grad_norm": 0.01041350793093443,
      "learning_rate": 2.9113712374581944e-06,
      "loss": 0.0,
      "step": 125900
    },
    {
      "epoch": 12.6,
      "grad_norm": 1.3331155969353858e-05,
      "learning_rate": 2.9096989966555184e-06,
      "loss": 0.0046,
      "step": 126000
    },
    {
      "epoch": 12.61,
      "grad_norm": 1.940794572874438e-05,
      "learning_rate": 2.908026755852843e-06,
      "loss": 0.0023,
      "step": 126100
    },
    {
      "epoch": 12.62,
      "grad_norm": 0.001803192775696516,
      "learning_rate": 2.9063545150501677e-06,
      "loss": 0.0024,
      "step": 126200
    },
    {
      "epoch": 12.63,
      "grad_norm": 3.6869932955596596e-05,
      "learning_rate": 2.9046822742474918e-06,
      "loss": 0.0047,
      "step": 126300
    },
    {
      "epoch": 12.64,
      "grad_norm": 1.732857163005974e-05,
      "learning_rate": 2.9030100334448162e-06,
      "loss": 0.0,
      "step": 126400
    },
    {
      "epoch": 12.65,
      "grad_norm": 0.00017873154138214886,
      "learning_rate": 2.9013377926421403e-06,
      "loss": 0.0088,
      "step": 126500
    },
    {
      "epoch": 12.66,
      "grad_norm": 2.3783226424711756e-05,
      "learning_rate": 2.899665551839465e-06,
      "loss": 0.0026,
      "step": 126600
    },
    {
      "epoch": 12.67,
      "grad_norm": 0.01853839121758938,
      "learning_rate": 2.8979933110367896e-06,
      "loss": 0.0029,
      "step": 126700
    },
    {
      "epoch": 12.68,
      "grad_norm": 4.210330007481389e-05,
      "learning_rate": 2.8963210702341137e-06,
      "loss": 0.0037,
      "step": 126800
    },
    {
      "epoch": 12.69,
      "grad_norm": 0.00028920164913870394,
      "learning_rate": 2.8946488294314385e-06,
      "loss": 0.0052,
      "step": 126900
    },
    {
      "epoch": 12.7,
      "grad_norm": 7.593008922412992e-05,
      "learning_rate": 2.8929765886287626e-06,
      "loss": 0.0133,
      "step": 127000
    },
    {
      "epoch": 12.71,
      "grad_norm": 20.229549407958984,
      "learning_rate": 2.891304347826087e-06,
      "loss": 0.0194,
      "step": 127100
    },
    {
      "epoch": 12.72,
      "grad_norm": 0.00024867980391718447,
      "learning_rate": 2.889632107023412e-06,
      "loss": 0.0001,
      "step": 127200
    },
    {
      "epoch": 12.73,
      "grad_norm": 0.160929873585701,
      "learning_rate": 2.887959866220736e-06,
      "loss": 0.0018,
      "step": 127300
    },
    {
      "epoch": 12.74,
      "grad_norm": 0.0001910285500343889,
      "learning_rate": 2.8862876254180604e-06,
      "loss": 0.0,
      "step": 127400
    },
    {
      "epoch": 12.75,
      "grad_norm": 0.002502140821889043,
      "learning_rate": 2.8846153846153845e-06,
      "loss": 0.0042,
      "step": 127500
    },
    {
      "epoch": 12.76,
      "grad_norm": 9.12977775442414e-05,
      "learning_rate": 2.8829431438127093e-06,
      "loss": 0.0028,
      "step": 127600
    },
    {
      "epoch": 12.77,
      "grad_norm": 0.0001754213881213218,
      "learning_rate": 2.881270903010034e-06,
      "loss": 0.0,
      "step": 127700
    },
    {
      "epoch": 12.78,
      "grad_norm": 2.844112168531865e-05,
      "learning_rate": 2.879598662207358e-06,
      "loss": 0.0,
      "step": 127800
    },
    {
      "epoch": 12.79,
      "grad_norm": 7.382730109384283e-05,
      "learning_rate": 2.8779264214046827e-06,
      "loss": 0.0036,
      "step": 127900
    },
    {
      "epoch": 12.8,
      "grad_norm": 6.222332740435377e-05,
      "learning_rate": 2.8762541806020068e-06,
      "loss": 0.0,
      "step": 128000
    },
    {
      "epoch": 12.81,
      "grad_norm": 2.280520311614964e-05,
      "learning_rate": 2.8745819397993312e-06,
      "loss": 0.0,
      "step": 128100
    },
    {
      "epoch": 12.82,
      "grad_norm": 1.8901395378634334e-05,
      "learning_rate": 2.872909698996656e-06,
      "loss": 0.0,
      "step": 128200
    },
    {
      "epoch": 12.83,
      "grad_norm": 2.6766725568450056e-05,
      "learning_rate": 2.87123745819398e-06,
      "loss": 0.0038,
      "step": 128300
    },
    {
      "epoch": 12.84,
      "grad_norm": 4.2255389416823164e-05,
      "learning_rate": 2.8695652173913046e-06,
      "loss": 0.0002,
      "step": 128400
    },
    {
      "epoch": 12.85,
      "grad_norm": 1.8130673197447322e-05,
      "learning_rate": 2.8678929765886286e-06,
      "loss": 0.0127,
      "step": 128500
    },
    {
      "epoch": 12.86,
      "grad_norm": 0.00020949130703229457,
      "learning_rate": 2.8662207357859535e-06,
      "loss": 0.0041,
      "step": 128600
    },
    {
      "epoch": 12.87,
      "grad_norm": 8.374531898880377e-05,
      "learning_rate": 2.864548494983278e-06,
      "loss": 0.0,
      "step": 128700
    },
    {
      "epoch": 12.88,
      "grad_norm": 0.005811138078570366,
      "learning_rate": 2.862876254180602e-06,
      "loss": 0.0006,
      "step": 128800
    },
    {
      "epoch": 12.89,
      "grad_norm": 5.1641036407090724e-05,
      "learning_rate": 2.861204013377927e-06,
      "loss": 0.0061,
      "step": 128900
    },
    {
      "epoch": 12.9,
      "grad_norm": 0.00028346985345706344,
      "learning_rate": 2.859531772575251e-06,
      "loss": 0.0019,
      "step": 129000
    },
    {
      "epoch": 12.91,
      "grad_norm": 2.3813428924768232e-05,
      "learning_rate": 2.8578595317725754e-06,
      "loss": 0.0,
      "step": 129100
    },
    {
      "epoch": 12.92,
      "grad_norm": 5.4004725825507194e-05,
      "learning_rate": 2.8561872909699003e-06,
      "loss": 0.0,
      "step": 129200
    },
    {
      "epoch": 12.93,
      "grad_norm": 0.0009715522755868733,
      "learning_rate": 2.8545150501672243e-06,
      "loss": 0.0,
      "step": 129300
    },
    {
      "epoch": 12.94,
      "grad_norm": 6.110258254921064e-05,
      "learning_rate": 2.8528428093645488e-06,
      "loss": 0.0,
      "step": 129400
    },
    {
      "epoch": 12.95,
      "grad_norm": 2.5454342903685756e-05,
      "learning_rate": 2.851170568561873e-06,
      "loss": 0.0075,
      "step": 129500
    },
    {
      "epoch": 12.96,
      "grad_norm": 2.0261213649064302e-05,
      "learning_rate": 2.8494983277591977e-06,
      "loss": 0.0003,
      "step": 129600
    },
    {
      "epoch": 12.97,
      "grad_norm": 0.012711210176348686,
      "learning_rate": 2.847826086956522e-06,
      "loss": 0.0028,
      "step": 129700
    },
    {
      "epoch": 12.98,
      "grad_norm": 0.9762109518051147,
      "learning_rate": 2.846153846153846e-06,
      "loss": 0.0101,
      "step": 129800
    },
    {
      "epoch": 12.99,
      "grad_norm": 0.003656335175037384,
      "learning_rate": 2.844481605351171e-06,
      "loss": 0.0001,
      "step": 129900
    },
    {
      "epoch": 13.0,
      "grad_norm": 1.8448445189278573e-05,
      "learning_rate": 2.842809364548495e-06,
      "loss": 0.0055,
      "step": 130000
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.98705,
      "eval_f1": 0.98705,
      "eval_loss": 0.11996782571077347,
      "eval_runtime": 135.7284,
      "eval_samples_per_second": 294.706,
      "eval_steps_per_second": 294.706,
      "step": 130000
    },
    {
      "epoch": 13.01,
      "grad_norm": 1.713820529403165e-05,
      "learning_rate": 2.8411371237458196e-06,
      "loss": 0.0,
      "step": 130100
    },
    {
      "epoch": 13.02,
      "grad_norm": 4.685423482442275e-05,
      "learning_rate": 2.8394648829431445e-06,
      "loss": 0.0001,
      "step": 130200
    },
    {
      "epoch": 13.03,
      "grad_norm": 0.0014853087486699224,
      "learning_rate": 2.8377926421404685e-06,
      "loss": 0.0042,
      "step": 130300
    },
    {
      "epoch": 13.04,
      "grad_norm": 1.4580614333681297e-05,
      "learning_rate": 2.836120401337793e-06,
      "loss": 0.003,
      "step": 130400
    },
    {
      "epoch": 13.05,
      "grad_norm": 1.3438087080430705e-05,
      "learning_rate": 2.834448160535117e-06,
      "loss": 0.0,
      "step": 130500
    },
    {
      "epoch": 13.06,
      "grad_norm": 7.020106568234041e-05,
      "learning_rate": 2.832775919732442e-06,
      "loss": 0.0,
      "step": 130600
    },
    {
      "epoch": 13.07,
      "grad_norm": 2.4668675905559212e-05,
      "learning_rate": 2.8311036789297663e-06,
      "loss": 0.0004,
      "step": 130700
    },
    {
      "epoch": 13.08,
      "grad_norm": 0.00015571806579828262,
      "learning_rate": 2.8294314381270904e-06,
      "loss": 0.0026,
      "step": 130800
    },
    {
      "epoch": 13.09,
      "grad_norm": 6.426640902645886e-05,
      "learning_rate": 2.8277591973244153e-06,
      "loss": 0.0021,
      "step": 130900
    },
    {
      "epoch": 13.1,
      "grad_norm": 2.0575174858095124e-05,
      "learning_rate": 2.8260869565217393e-06,
      "loss": 0.0035,
      "step": 131000
    },
    {
      "epoch": 13.11,
      "grad_norm": 1.2000875358353369e-05,
      "learning_rate": 2.8244147157190638e-06,
      "loss": 0.0,
      "step": 131100
    },
    {
      "epoch": 13.12,
      "grad_norm": 1.4357594409375452e-05,
      "learning_rate": 2.8227424749163882e-06,
      "loss": 0.0068,
      "step": 131200
    },
    {
      "epoch": 13.13,
      "grad_norm": 0.01679055020213127,
      "learning_rate": 2.8210702341137127e-06,
      "loss": 0.0,
      "step": 131300
    },
    {
      "epoch": 13.14,
      "grad_norm": 3.117432424915023e-05,
      "learning_rate": 2.819397993311037e-06,
      "loss": 0.0001,
      "step": 131400
    },
    {
      "epoch": 13.15,
      "grad_norm": 1.0254871995130088e-05,
      "learning_rate": 2.817725752508361e-06,
      "loss": 0.0013,
      "step": 131500
    },
    {
      "epoch": 13.16,
      "grad_norm": 1.2311931641306728e-05,
      "learning_rate": 2.816053511705686e-06,
      "loss": 0.0024,
      "step": 131600
    },
    {
      "epoch": 13.17,
      "grad_norm": 1.4948631360311992e-05,
      "learning_rate": 2.8143812709030105e-06,
      "loss": 0.0036,
      "step": 131700
    },
    {
      "epoch": 13.18,
      "grad_norm": 3.637359623098746e-05,
      "learning_rate": 2.8127090301003346e-06,
      "loss": 0.0,
      "step": 131800
    },
    {
      "epoch": 13.19,
      "grad_norm": 5.812734889332205e-05,
      "learning_rate": 2.811036789297659e-06,
      "loss": 0.0,
      "step": 131900
    },
    {
      "epoch": 13.2,
      "grad_norm": 2.516805034247227e-05,
      "learning_rate": 2.8093645484949835e-06,
      "loss": 0.0,
      "step": 132000
    },
    {
      "epoch": 13.21,
      "grad_norm": 0.0016383231850340962,
      "learning_rate": 2.807692307692308e-06,
      "loss": 0.0,
      "step": 132100
    },
    {
      "epoch": 13.22,
      "grad_norm": 3.990402183262631e-05,
      "learning_rate": 2.8060200668896324e-06,
      "loss": 0.0,
      "step": 132200
    },
    {
      "epoch": 13.23,
      "grad_norm": 3.431649747653864e-05,
      "learning_rate": 2.804347826086957e-06,
      "loss": 0.0,
      "step": 132300
    },
    {
      "epoch": 13.24,
      "grad_norm": 4.9618389311945066e-05,
      "learning_rate": 2.8026755852842813e-06,
      "loss": 0.0029,
      "step": 132400
    },
    {
      "epoch": 13.25,
      "grad_norm": 1.1744758921850007e-05,
      "learning_rate": 2.8010033444816054e-06,
      "loss": 0.0023,
      "step": 132500
    },
    {
      "epoch": 13.26,
      "grad_norm": 0.0007372087566182017,
      "learning_rate": 2.79933110367893e-06,
      "loss": 0.0039,
      "step": 132600
    },
    {
      "epoch": 13.27,
      "grad_norm": 1.3846334695699625e-05,
      "learning_rate": 2.7976588628762547e-06,
      "loss": 0.0,
      "step": 132700
    },
    {
      "epoch": 13.28,
      "grad_norm": 0.00011616638948908076,
      "learning_rate": 2.7959866220735787e-06,
      "loss": 0.0,
      "step": 132800
    },
    {
      "epoch": 13.29,
      "grad_norm": 1.7173537344206125e-05,
      "learning_rate": 2.794314381270903e-06,
      "loss": 0.0063,
      "step": 132900
    },
    {
      "epoch": 13.3,
      "grad_norm": 7.09253508830443e-05,
      "learning_rate": 2.7926421404682277e-06,
      "loss": 0.0062,
      "step": 133000
    },
    {
      "epoch": 13.31,
      "grad_norm": 0.037380144000053406,
      "learning_rate": 2.790969899665552e-06,
      "loss": 0.0001,
      "step": 133100
    },
    {
      "epoch": 13.32,
      "grad_norm": 1.3415536159300245e-05,
      "learning_rate": 2.7892976588628766e-06,
      "loss": 0.0101,
      "step": 133200
    },
    {
      "epoch": 13.33,
      "grad_norm": 1.4683989320474211e-05,
      "learning_rate": 2.7876254180602006e-06,
      "loss": 0.0032,
      "step": 133300
    },
    {
      "epoch": 13.34,
      "grad_norm": 2.834025872289203e-05,
      "learning_rate": 2.7859531772575255e-06,
      "loss": 0.0,
      "step": 133400
    },
    {
      "epoch": 13.35,
      "grad_norm": 6.283698894549161e-05,
      "learning_rate": 2.7842809364548495e-06,
      "loss": 0.0031,
      "step": 133500
    },
    {
      "epoch": 13.36,
      "grad_norm": 3.11326548398938e-05,
      "learning_rate": 2.782608695652174e-06,
      "loss": 0.0,
      "step": 133600
    },
    {
      "epoch": 13.37,
      "grad_norm": 0.0004122111131437123,
      "learning_rate": 2.780936454849499e-06,
      "loss": 0.0,
      "step": 133700
    },
    {
      "epoch": 13.38,
      "grad_norm": 0.0010729500791057944,
      "learning_rate": 2.779264214046823e-06,
      "loss": 0.0,
      "step": 133800
    },
    {
      "epoch": 13.39,
      "grad_norm": 2.4340455638593994e-05,
      "learning_rate": 2.7775919732441474e-06,
      "loss": 0.0,
      "step": 133900
    },
    {
      "epoch": 13.4,
      "grad_norm": 2.9749480745522305e-05,
      "learning_rate": 2.7759197324414714e-06,
      "loss": 0.0003,
      "step": 134000
    },
    {
      "epoch": 13.41,
      "grad_norm": 0.00011808306589955464,
      "learning_rate": 2.7742474916387963e-06,
      "loss": 0.0,
      "step": 134100
    },
    {
      "epoch": 13.42,
      "grad_norm": 9.720781235955656e-06,
      "learning_rate": 2.7725752508361208e-06,
      "loss": 0.0034,
      "step": 134200
    },
    {
      "epoch": 13.43,
      "grad_norm": 0.08499442040920258,
      "learning_rate": 2.770903010033445e-06,
      "loss": 0.0,
      "step": 134300
    },
    {
      "epoch": 13.44,
      "grad_norm": 6.802148709539324e-05,
      "learning_rate": 2.7692307692307697e-06,
      "loss": 0.0,
      "step": 134400
    },
    {
      "epoch": 13.45,
      "grad_norm": 1.7490610844106413e-05,
      "learning_rate": 2.7675585284280937e-06,
      "loss": 0.0,
      "step": 134500
    },
    {
      "epoch": 13.46,
      "grad_norm": 1.0484201993676834e-05,
      "learning_rate": 2.765886287625418e-06,
      "loss": 0.0,
      "step": 134600
    },
    {
      "epoch": 13.47,
      "grad_norm": 1.108324977394659e-05,
      "learning_rate": 2.764214046822743e-06,
      "loss": 0.0,
      "step": 134700
    },
    {
      "epoch": 13.48,
      "grad_norm": 0.00023472192697227,
      "learning_rate": 2.762541806020067e-06,
      "loss": 0.0028,
      "step": 134800
    },
    {
      "epoch": 13.49,
      "grad_norm": 2.1847616153536364e-05,
      "learning_rate": 2.7608695652173916e-06,
      "loss": 0.0032,
      "step": 134900
    },
    {
      "epoch": 13.5,
      "grad_norm": 1.1873323273903225e-05,
      "learning_rate": 2.7591973244147156e-06,
      "loss": 0.0041,
      "step": 135000
    },
    {
      "epoch": 13.5,
      "eval_accuracy": 0.9864,
      "eval_f1": 0.9864,
      "eval_loss": 0.13605718314647675,
      "eval_runtime": 136.6144,
      "eval_samples_per_second": 292.795,
      "eval_steps_per_second": 292.795,
      "step": 135000
    },
    {
      "epoch": 13.51,
      "grad_norm": 6.469953223131597e-05,
      "learning_rate": 2.7575250836120405e-06,
      "loss": 0.0056,
      "step": 135100
    },
    {
      "epoch": 13.52,
      "grad_norm": 1.1132638064736966e-05,
      "learning_rate": 2.755852842809365e-06,
      "loss": 0.0,
      "step": 135200
    },
    {
      "epoch": 13.53,
      "grad_norm": 1.0743293387349695e-05,
      "learning_rate": 2.754180602006689e-06,
      "loss": 0.0022,
      "step": 135300
    },
    {
      "epoch": 13.54,
      "grad_norm": 1.1827138223452494e-05,
      "learning_rate": 2.752508361204014e-06,
      "loss": 0.0038,
      "step": 135400
    },
    {
      "epoch": 13.55,
      "grad_norm": 1.2733278708765283e-05,
      "learning_rate": 2.750836120401338e-06,
      "loss": 0.0037,
      "step": 135500
    },
    {
      "epoch": 13.56,
      "grad_norm": 1.2386645721562672e-05,
      "learning_rate": 2.7491638795986624e-06,
      "loss": 0.0,
      "step": 135600
    },
    {
      "epoch": 13.57,
      "grad_norm": 8.410525879298802e-06,
      "learning_rate": 2.7474916387959864e-06,
      "loss": 0.0,
      "step": 135700
    },
    {
      "epoch": 13.58,
      "grad_norm": 4.612410339177586e-05,
      "learning_rate": 2.7458193979933113e-06,
      "loss": 0.0,
      "step": 135800
    },
    {
      "epoch": 13.59,
      "grad_norm": 2.6822170184459537e-05,
      "learning_rate": 2.7441471571906358e-06,
      "loss": 0.0,
      "step": 135900
    },
    {
      "epoch": 13.6,
      "grad_norm": 1.6195193893508986e-05,
      "learning_rate": 2.74247491638796e-06,
      "loss": 0.0,
      "step": 136000
    },
    {
      "epoch": 13.61,
      "grad_norm": 1.1398445167287719e-05,
      "learning_rate": 2.7408026755852847e-06,
      "loss": 0.0,
      "step": 136100
    },
    {
      "epoch": 13.62,
      "grad_norm": 7.909869964350946e-06,
      "learning_rate": 2.7391304347826087e-06,
      "loss": 0.0047,
      "step": 136200
    },
    {
      "epoch": 13.63,
      "grad_norm": 8.60071577335475e-06,
      "learning_rate": 2.737458193979933e-06,
      "loss": 0.0002,
      "step": 136300
    },
    {
      "epoch": 13.64,
      "grad_norm": 1.0402597581560258e-05,
      "learning_rate": 2.735785953177258e-06,
      "loss": 0.0036,
      "step": 136400
    },
    {
      "epoch": 13.65,
      "grad_norm": 9.145016520051286e-06,
      "learning_rate": 2.734113712374582e-06,
      "loss": 0.0024,
      "step": 136500
    },
    {
      "epoch": 13.66,
      "grad_norm": 1.3457996828947216e-05,
      "learning_rate": 2.7324414715719066e-06,
      "loss": 0.004,
      "step": 136600
    },
    {
      "epoch": 13.67,
      "grad_norm": 1.3402839613263495e-05,
      "learning_rate": 2.7307692307692306e-06,
      "loss": 0.0,
      "step": 136700
    },
    {
      "epoch": 13.68,
      "grad_norm": 1.228326709679095e-05,
      "learning_rate": 2.7290969899665555e-06,
      "loss": 0.0079,
      "step": 136800
    },
    {
      "epoch": 13.69,
      "grad_norm": 0.00012510253873188049,
      "learning_rate": 2.72742474916388e-06,
      "loss": 0.0026,
      "step": 136900
    },
    {
      "epoch": 13.7,
      "grad_norm": 2.1446710888994858e-05,
      "learning_rate": 2.725752508361204e-06,
      "loss": 0.0,
      "step": 137000
    },
    {
      "epoch": 13.71,
      "grad_norm": 2.1957304852548987e-05,
      "learning_rate": 2.724080267558529e-06,
      "loss": 0.0,
      "step": 137100
    },
    {
      "epoch": 13.72,
      "grad_norm": 1.3254529221740086e-05,
      "learning_rate": 2.722408026755853e-06,
      "loss": 0.0088,
      "step": 137200
    },
    {
      "epoch": 13.73,
      "grad_norm": 2.5398878278792836e-05,
      "learning_rate": 2.7207357859531774e-06,
      "loss": 0.0047,
      "step": 137300
    },
    {
      "epoch": 13.74,
      "grad_norm": 3.166155147482641e-05,
      "learning_rate": 2.7190635451505022e-06,
      "loss": 0.017,
      "step": 137400
    },
    {
      "epoch": 13.75,
      "grad_norm": 0.0029113488271832466,
      "learning_rate": 2.7173913043478263e-06,
      "loss": 0.0,
      "step": 137500
    },
    {
      "epoch": 13.76,
      "grad_norm": 1.447906470275484e-05,
      "learning_rate": 2.7157190635451507e-06,
      "loss": 0.0002,
      "step": 137600
    },
    {
      "epoch": 13.77,
      "grad_norm": 0.00028520452906377614,
      "learning_rate": 2.7140468227424748e-06,
      "loss": 0.0,
      "step": 137700
    },
    {
      "epoch": 13.78,
      "grad_norm": 0.00032373316935263574,
      "learning_rate": 2.7123745819397997e-06,
      "loss": 0.0,
      "step": 137800
    },
    {
      "epoch": 13.79,
      "grad_norm": 1.5336469004978426e-05,
      "learning_rate": 2.710702341137124e-06,
      "loss": 0.0048,
      "step": 137900
    },
    {
      "epoch": 13.8,
      "grad_norm": 4.3872540118172765e-05,
      "learning_rate": 2.709030100334448e-06,
      "loss": 0.0035,
      "step": 138000
    },
    {
      "epoch": 13.81,
      "grad_norm": 3.1709718314232305e-05,
      "learning_rate": 2.707357859531773e-06,
      "loss": 0.0001,
      "step": 138100
    },
    {
      "epoch": 13.82,
      "grad_norm": 3.154708610964008e-05,
      "learning_rate": 2.705685618729097e-06,
      "loss": 0.0039,
      "step": 138200
    },
    {
      "epoch": 13.83,
      "grad_norm": 0.0019509652629494667,
      "learning_rate": 2.7040133779264215e-06,
      "loss": 0.0101,
      "step": 138300
    },
    {
      "epoch": 13.84,
      "grad_norm": 3.1300398404709995e-05,
      "learning_rate": 2.7023411371237464e-06,
      "loss": 0.0033,
      "step": 138400
    },
    {
      "epoch": 13.85,
      "grad_norm": 0.02786121517419815,
      "learning_rate": 2.7006688963210705e-06,
      "loss": 0.0025,
      "step": 138500
    },
    {
      "epoch": 13.86,
      "grad_norm": 0.01319398358464241,
      "learning_rate": 2.698996655518395e-06,
      "loss": 0.0,
      "step": 138600
    },
    {
      "epoch": 13.87,
      "grad_norm": 3.6790541344089434e-05,
      "learning_rate": 2.697324414715719e-06,
      "loss": 0.0,
      "step": 138700
    },
    {
      "epoch": 13.88,
      "grad_norm": 0.0003256149939261377,
      "learning_rate": 2.695652173913044e-06,
      "loss": 0.0009,
      "step": 138800
    },
    {
      "epoch": 13.89,
      "grad_norm": 0.0008701249025762081,
      "learning_rate": 2.6939799331103683e-06,
      "loss": 0.0057,
      "step": 138900
    },
    {
      "epoch": 13.9,
      "grad_norm": 0.00016028298705350608,
      "learning_rate": 2.6923076923076923e-06,
      "loss": 0.0044,
      "step": 139000
    },
    {
      "epoch": 13.91,
      "grad_norm": 4.8700254410505295e-05,
      "learning_rate": 2.6906354515050172e-06,
      "loss": 0.0037,
      "step": 139100
    },
    {
      "epoch": 13.92,
      "grad_norm": 1.1900821846211329e-05,
      "learning_rate": 2.6889632107023413e-06,
      "loss": 0.0,
      "step": 139200
    },
    {
      "epoch": 13.93,
      "grad_norm": 3.17985650326591e-05,
      "learning_rate": 2.6872909698996657e-06,
      "loss": 0.0,
      "step": 139300
    },
    {
      "epoch": 13.94,
      "grad_norm": 1.7800852219806984e-05,
      "learning_rate": 2.6856187290969906e-06,
      "loss": 0.0,
      "step": 139400
    },
    {
      "epoch": 13.95,
      "grad_norm": 0.00012519603478722274,
      "learning_rate": 2.6839464882943146e-06,
      "loss": 0.0,
      "step": 139500
    },
    {
      "epoch": 13.96,
      "grad_norm": 1.7471717001171783e-05,
      "learning_rate": 2.682274247491639e-06,
      "loss": 0.0,
      "step": 139600
    },
    {
      "epoch": 13.97,
      "grad_norm": 2.409886656096205e-05,
      "learning_rate": 2.680602006688963e-06,
      "loss": 0.0,
      "step": 139700
    },
    {
      "epoch": 13.98,
      "grad_norm": 9.131293336395174e-05,
      "learning_rate": 2.678929765886288e-06,
      "loss": 0.0,
      "step": 139800
    },
    {
      "epoch": 13.99,
      "grad_norm": 0.00039141171146184206,
      "learning_rate": 2.6772575250836125e-06,
      "loss": 0.0017,
      "step": 139900
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.00036528526106849313,
      "learning_rate": 2.6755852842809365e-06,
      "loss": 0.005,
      "step": 140000
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.987025,
      "eval_f1": 0.987025,
      "eval_loss": 0.12981809675693512,
      "eval_runtime": 138.4883,
      "eval_samples_per_second": 288.833,
      "eval_steps_per_second": 288.833,
      "step": 140000
    },
    {
      "epoch": 14.01,
      "grad_norm": 1.970234006876126e-05,
      "learning_rate": 2.6739130434782614e-06,
      "loss": 0.0,
      "step": 140100
    },
    {
      "epoch": 14.02,
      "grad_norm": 4.8702968342695385e-05,
      "learning_rate": 2.6722408026755854e-06,
      "loss": 0.003,
      "step": 140200
    },
    {
      "epoch": 14.03,
      "grad_norm": 1.3415955436357763e-05,
      "learning_rate": 2.67056856187291e-06,
      "loss": 0.0051,
      "step": 140300
    },
    {
      "epoch": 14.04,
      "grad_norm": 9.642380064178724e-06,
      "learning_rate": 2.6688963210702344e-06,
      "loss": 0.0,
      "step": 140400
    },
    {
      "epoch": 14.05,
      "grad_norm": 1.1794251804531086e-05,
      "learning_rate": 2.667224080267559e-06,
      "loss": 0.0,
      "step": 140500
    },
    {
      "epoch": 14.06,
      "grad_norm": 1.1837163583550137e-05,
      "learning_rate": 2.6655518394648833e-06,
      "loss": 0.0,
      "step": 140600
    },
    {
      "epoch": 14.07,
      "grad_norm": 0.00018968567019328475,
      "learning_rate": 2.6638795986622073e-06,
      "loss": 0.0,
      "step": 140700
    },
    {
      "epoch": 14.08,
      "grad_norm": 2.6006642656284384e-05,
      "learning_rate": 2.662207357859532e-06,
      "loss": 0.0,
      "step": 140800
    },
    {
      "epoch": 14.09,
      "grad_norm": 1.6532772860955447e-05,
      "learning_rate": 2.6605351170568567e-06,
      "loss": 0.0,
      "step": 140900
    },
    {
      "epoch": 14.1,
      "grad_norm": 3.34562428179197e-05,
      "learning_rate": 2.6588628762541807e-06,
      "loss": 0.003,
      "step": 141000
    },
    {
      "epoch": 14.11,
      "grad_norm": 0.00025544603704474866,
      "learning_rate": 2.657190635451505e-06,
      "loss": 0.0048,
      "step": 141100
    },
    {
      "epoch": 14.12,
      "grad_norm": 1.0975451914418954e-05,
      "learning_rate": 2.6555183946488296e-06,
      "loss": 0.0,
      "step": 141200
    },
    {
      "epoch": 14.13,
      "grad_norm": 5.2620616770582274e-05,
      "learning_rate": 2.653846153846154e-06,
      "loss": 0.0072,
      "step": 141300
    },
    {
      "epoch": 14.14,
      "grad_norm": 9.236534424417187e-06,
      "learning_rate": 2.6521739130434785e-06,
      "loss": 0.0,
      "step": 141400
    },
    {
      "epoch": 14.15,
      "grad_norm": 1.7573007426108234e-05,
      "learning_rate": 2.650501672240803e-06,
      "loss": 0.0,
      "step": 141500
    },
    {
      "epoch": 14.16,
      "grad_norm": 1.277837418456329e-05,
      "learning_rate": 2.6488294314381275e-06,
      "loss": 0.007,
      "step": 141600
    },
    {
      "epoch": 14.17,
      "grad_norm": 1.1337443538650405e-05,
      "learning_rate": 2.6471571906354515e-06,
      "loss": 0.0,
      "step": 141700
    },
    {
      "epoch": 14.18,
      "grad_norm": 2.1818048480781727e-05,
      "learning_rate": 2.645484949832776e-06,
      "loss": 0.0,
      "step": 141800
    },
    {
      "epoch": 14.19,
      "grad_norm": 2.7288138880976476e-05,
      "learning_rate": 2.643812709030101e-06,
      "loss": 0.0018,
      "step": 141900
    },
    {
      "epoch": 14.2,
      "grad_norm": 1.165820867754519e-05,
      "learning_rate": 2.642140468227425e-06,
      "loss": 0.0028,
      "step": 142000
    },
    {
      "epoch": 14.21,
      "grad_norm": 9.926073289534543e-06,
      "learning_rate": 2.6404682274247493e-06,
      "loss": 0.0047,
      "step": 142100
    },
    {
      "epoch": 14.22,
      "grad_norm": 1.514962968940381e-05,
      "learning_rate": 2.638795986622074e-06,
      "loss": 0.0034,
      "step": 142200
    },
    {
      "epoch": 14.23,
      "grad_norm": 8.196759154088795e-06,
      "learning_rate": 2.6371237458193983e-06,
      "loss": 0.0,
      "step": 142300
    },
    {
      "epoch": 14.24,
      "grad_norm": 2.0415285689523444e-05,
      "learning_rate": 2.6354515050167227e-06,
      "loss": 0.0,
      "step": 142400
    },
    {
      "epoch": 14.25,
      "grad_norm": 0.00016611946921329945,
      "learning_rate": 2.6337792642140468e-06,
      "loss": 0.0005,
      "step": 142500
    },
    {
      "epoch": 14.26,
      "grad_norm": 6.390087946783751e-06,
      "learning_rate": 2.6321070234113716e-06,
      "loss": 0.0001,
      "step": 142600
    },
    {
      "epoch": 14.27,
      "grad_norm": 1.526228697912302e-05,
      "learning_rate": 2.6304347826086957e-06,
      "loss": 0.0105,
      "step": 142700
    },
    {
      "epoch": 14.28,
      "grad_norm": 2.9013786843279377e-05,
      "learning_rate": 2.62876254180602e-06,
      "loss": 0.0,
      "step": 142800
    },
    {
      "epoch": 14.29,
      "grad_norm": 5.3516596381086856e-05,
      "learning_rate": 2.627090301003345e-06,
      "loss": 0.0032,
      "step": 142900
    },
    {
      "epoch": 14.3,
      "grad_norm": 1.080665515473811e-05,
      "learning_rate": 2.625418060200669e-06,
      "loss": 0.0049,
      "step": 143000
    },
    {
      "epoch": 14.31,
      "grad_norm": 1.2666178918152582e-05,
      "learning_rate": 2.6237458193979935e-06,
      "loss": 0.0,
      "step": 143100
    },
    {
      "epoch": 14.32,
      "grad_norm": 3.2697953429305926e-05,
      "learning_rate": 2.6220735785953176e-06,
      "loss": 0.0,
      "step": 143200
    },
    {
      "epoch": 14.33,
      "grad_norm": 3.8729096559109166e-05,
      "learning_rate": 2.6204013377926424e-06,
      "loss": 0.0,
      "step": 143300
    },
    {
      "epoch": 14.34,
      "grad_norm": 7.637468115717638e-06,
      "learning_rate": 2.618729096989967e-06,
      "loss": 0.0016,
      "step": 143400
    },
    {
      "epoch": 14.35,
      "grad_norm": 0.000350749003700912,
      "learning_rate": 2.617056856187291e-06,
      "loss": 0.0,
      "step": 143500
    },
    {
      "epoch": 14.36,
      "grad_norm": 6.134332579676993e-06,
      "learning_rate": 2.615384615384616e-06,
      "loss": 0.0,
      "step": 143600
    },
    {
      "epoch": 14.37,
      "grad_norm": 6.324955393210985e-06,
      "learning_rate": 2.61371237458194e-06,
      "loss": 0.0009,
      "step": 143700
    },
    {
      "epoch": 14.38,
      "grad_norm": 4.8610727390041575e-06,
      "learning_rate": 2.6120401337792643e-06,
      "loss": 0.0,
      "step": 143800
    },
    {
      "epoch": 14.39,
      "grad_norm": 8.644431363791227e-05,
      "learning_rate": 2.610367892976589e-06,
      "loss": 0.0002,
      "step": 143900
    },
    {
      "epoch": 14.4,
      "grad_norm": 7.85917291068472e-06,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.0059,
      "step": 144000
    },
    {
      "epoch": 14.41,
      "grad_norm": 1.2011563740088604e-05,
      "learning_rate": 2.6070234113712377e-06,
      "loss": 0.0042,
      "step": 144100
    },
    {
      "epoch": 14.42,
      "grad_norm": 4.436729068402201e-05,
      "learning_rate": 2.6053511705685617e-06,
      "loss": 0.0017,
      "step": 144200
    },
    {
      "epoch": 14.43,
      "grad_norm": 9.664974641054869e-06,
      "learning_rate": 2.6036789297658866e-06,
      "loss": 0.0,
      "step": 144300
    },
    {
      "epoch": 14.44,
      "grad_norm": 6.7011196733801626e-06,
      "learning_rate": 2.602006688963211e-06,
      "loss": 0.0,
      "step": 144400
    },
    {
      "epoch": 14.45,
      "grad_norm": 3.2653235393809155e-05,
      "learning_rate": 2.600334448160535e-06,
      "loss": 0.0,
      "step": 144500
    },
    {
      "epoch": 14.46,
      "grad_norm": 8.305092705995776e-06,
      "learning_rate": 2.59866220735786e-06,
      "loss": 0.005,
      "step": 144600
    },
    {
      "epoch": 14.47,
      "grad_norm": 4.4582277041627094e-05,
      "learning_rate": 2.596989966555184e-06,
      "loss": 0.0011,
      "step": 144700
    },
    {
      "epoch": 14.48,
      "grad_norm": 9.818021680985112e-06,
      "learning_rate": 2.5953177257525085e-06,
      "loss": 0.0,
      "step": 144800
    },
    {
      "epoch": 14.49,
      "grad_norm": 4.905187597614713e-05,
      "learning_rate": 2.5936454849498334e-06,
      "loss": 0.0,
      "step": 144900
    },
    {
      "epoch": 14.5,
      "grad_norm": 3.9814804040361196e-05,
      "learning_rate": 2.5919732441471574e-06,
      "loss": 0.0,
      "step": 145000
    },
    {
      "epoch": 14.5,
      "eval_accuracy": 0.98695,
      "eval_f1": 0.98695,
      "eval_loss": 0.13318751752376556,
      "eval_runtime": 136.0623,
      "eval_samples_per_second": 293.983,
      "eval_steps_per_second": 293.983,
      "step": 145000
    },
    {
      "epoch": 14.51,
      "grad_norm": 3.010919681400992e-05,
      "learning_rate": 2.590301003344482e-06,
      "loss": 0.0017,
      "step": 145100
    },
    {
      "epoch": 14.52,
      "grad_norm": 1.9201515897293575e-05,
      "learning_rate": 2.588628762541806e-06,
      "loss": 0.0,
      "step": 145200
    },
    {
      "epoch": 14.53,
      "grad_norm": 6.911823220434599e-06,
      "learning_rate": 2.586956521739131e-06,
      "loss": 0.0,
      "step": 145300
    },
    {
      "epoch": 14.54,
      "grad_norm": 0.00012770664761774242,
      "learning_rate": 2.5852842809364553e-06,
      "loss": 0.0,
      "step": 145400
    },
    {
      "epoch": 14.55,
      "grad_norm": 0.004662015475332737,
      "learning_rate": 2.5836120401337793e-06,
      "loss": 0.0,
      "step": 145500
    },
    {
      "epoch": 14.56,
      "grad_norm": 0.0038990043103694916,
      "learning_rate": 2.581939799331104e-06,
      "loss": 0.0046,
      "step": 145600
    },
    {
      "epoch": 14.57,
      "grad_norm": 1.2492128007579595e-05,
      "learning_rate": 2.5802675585284282e-06,
      "loss": 0.0,
      "step": 145700
    },
    {
      "epoch": 14.58,
      "grad_norm": 6.336121259664651e-06,
      "learning_rate": 2.5785953177257527e-06,
      "loss": 0.0,
      "step": 145800
    },
    {
      "epoch": 14.59,
      "grad_norm": 7.761910637782421e-06,
      "learning_rate": 2.5769230769230767e-06,
      "loss": 0.0,
      "step": 145900
    },
    {
      "epoch": 14.6,
      "grad_norm": 0.0003560628683771938,
      "learning_rate": 2.5752508361204016e-06,
      "loss": 0.0,
      "step": 146000
    },
    {
      "epoch": 14.61,
      "grad_norm": 9.447974662180059e-06,
      "learning_rate": 2.573578595317726e-06,
      "loss": 0.0,
      "step": 146100
    },
    {
      "epoch": 14.62,
      "grad_norm": 6.381172715919092e-05,
      "learning_rate": 2.57190635451505e-06,
      "loss": 0.0039,
      "step": 146200
    },
    {
      "epoch": 14.63,
      "grad_norm": 1.7973427020478994e-05,
      "learning_rate": 2.570234113712375e-06,
      "loss": 0.0053,
      "step": 146300
    },
    {
      "epoch": 14.64,
      "grad_norm": 8.584712304582354e-06,
      "learning_rate": 2.568561872909699e-06,
      "loss": 0.0101,
      "step": 146400
    },
    {
      "epoch": 14.65,
      "grad_norm": 1.8466884284862317e-05,
      "learning_rate": 2.5668896321070235e-06,
      "loss": 0.0,
      "step": 146500
    },
    {
      "epoch": 14.66,
      "grad_norm": 7.304724022105802e-06,
      "learning_rate": 2.5652173913043484e-06,
      "loss": 0.0,
      "step": 146600
    },
    {
      "epoch": 14.67,
      "grad_norm": 6.799628863518592e-06,
      "learning_rate": 2.5635451505016724e-06,
      "loss": 0.0,
      "step": 146700
    },
    {
      "epoch": 14.68,
      "grad_norm": 9.02667215996189e-06,
      "learning_rate": 2.561872909698997e-06,
      "loss": 0.0,
      "step": 146800
    },
    {
      "epoch": 14.69,
      "grad_norm": 4.482056829147041e-05,
      "learning_rate": 2.560200668896321e-06,
      "loss": 0.0,
      "step": 146900
    },
    {
      "epoch": 14.7,
      "grad_norm": 8.890767276170664e-06,
      "learning_rate": 2.5585284280936458e-06,
      "loss": 0.0033,
      "step": 147000
    },
    {
      "epoch": 14.71,
      "grad_norm": 1.5810957847861573e-05,
      "learning_rate": 2.5568561872909702e-06,
      "loss": 0.0039,
      "step": 147100
    },
    {
      "epoch": 14.72,
      "grad_norm": 8.200189040508121e-05,
      "learning_rate": 2.5551839464882943e-06,
      "loss": 0.0032,
      "step": 147200
    },
    {
      "epoch": 14.73,
      "grad_norm": 1.996455466724001e-05,
      "learning_rate": 2.553511705685619e-06,
      "loss": 0.0,
      "step": 147300
    },
    {
      "epoch": 14.74,
      "grad_norm": 1.1589927453314885e-05,
      "learning_rate": 2.551839464882943e-06,
      "loss": 0.0,
      "step": 147400
    },
    {
      "epoch": 14.75,
      "grad_norm": 8.36480176076293e-05,
      "learning_rate": 2.5501672240802677e-06,
      "loss": 0.0,
      "step": 147500
    },
    {
      "epoch": 14.76,
      "grad_norm": 1.1121761417598464e-05,
      "learning_rate": 2.5484949832775925e-06,
      "loss": 0.0,
      "step": 147600
    },
    {
      "epoch": 14.77,
      "grad_norm": 9.952851542038843e-06,
      "learning_rate": 2.5468227424749166e-06,
      "loss": 0.0005,
      "step": 147700
    },
    {
      "epoch": 14.78,
      "grad_norm": 0.0005833844188600779,
      "learning_rate": 2.545150501672241e-06,
      "loss": 0.0036,
      "step": 147800
    },
    {
      "epoch": 14.79,
      "grad_norm": 0.000579743180423975,
      "learning_rate": 2.543478260869565e-06,
      "loss": 0.0,
      "step": 147900
    },
    {
      "epoch": 14.8,
      "grad_norm": 0.007290335837751627,
      "learning_rate": 2.54180602006689e-06,
      "loss": 0.0,
      "step": 148000
    },
    {
      "epoch": 14.81,
      "grad_norm": 7.68768040870782e-06,
      "learning_rate": 2.5401337792642144e-06,
      "loss": 0.0032,
      "step": 148100
    },
    {
      "epoch": 14.82,
      "grad_norm": 1.0321867193852086e-05,
      "learning_rate": 2.5384615384615385e-06,
      "loss": 0.0,
      "step": 148200
    },
    {
      "epoch": 14.83,
      "grad_norm": 5.90362378716236e-06,
      "learning_rate": 2.5367892976588633e-06,
      "loss": 0.0,
      "step": 148300
    },
    {
      "epoch": 14.84,
      "grad_norm": 6.799234597565373e-06,
      "learning_rate": 2.5351170568561874e-06,
      "loss": 0.0,
      "step": 148400
    },
    {
      "epoch": 14.85,
      "grad_norm": 7.403478321066359e-06,
      "learning_rate": 2.533444816053512e-06,
      "loss": 0.0045,
      "step": 148500
    },
    {
      "epoch": 14.86,
      "grad_norm": 1.4559896953869611e-05,
      "learning_rate": 2.5317725752508367e-06,
      "loss": 0.0018,
      "step": 148600
    },
    {
      "epoch": 14.87,
      "grad_norm": 6.7038836277788505e-06,
      "learning_rate": 2.5301003344481608e-06,
      "loss": 0.0,
      "step": 148700
    },
    {
      "epoch": 14.88,
      "grad_norm": 6.24095082457643e-06,
      "learning_rate": 2.5284280936454852e-06,
      "loss": 0.0,
      "step": 148800
    },
    {
      "epoch": 14.89,
      "grad_norm": 6.512796971946955e-06,
      "learning_rate": 2.5267558528428093e-06,
      "loss": 0.0079,
      "step": 148900
    },
    {
      "epoch": 14.9,
      "grad_norm": 0.0056665875017642975,
      "learning_rate": 2.525083612040134e-06,
      "loss": 0.0065,
      "step": 149000
    },
    {
      "epoch": 14.91,
      "grad_norm": 9.204671187035274e-06,
      "learning_rate": 2.5234113712374586e-06,
      "loss": 0.0062,
      "step": 149100
    },
    {
      "epoch": 14.92,
      "grad_norm": 1.0256294444843661e-05,
      "learning_rate": 2.5217391304347826e-06,
      "loss": 0.0003,
      "step": 149200
    },
    {
      "epoch": 14.93,
      "grad_norm": 0.0003551799163687974,
      "learning_rate": 2.5200668896321075e-06,
      "loss": 0.0032,
      "step": 149300
    },
    {
      "epoch": 14.94,
      "grad_norm": 0.00010754630056908354,
      "learning_rate": 2.5183946488294316e-06,
      "loss": 0.0001,
      "step": 149400
    },
    {
      "epoch": 14.95,
      "grad_norm": 8.264900679932907e-05,
      "learning_rate": 2.516722408026756e-06,
      "loss": 0.0,
      "step": 149500
    },
    {
      "epoch": 14.96,
      "grad_norm": 1.0027822099800687e-05,
      "learning_rate": 2.5150501672240805e-06,
      "loss": 0.0011,
      "step": 149600
    },
    {
      "epoch": 14.97,
      "grad_norm": 0.0418342724442482,
      "learning_rate": 2.513377926421405e-06,
      "loss": 0.0,
      "step": 149700
    },
    {
      "epoch": 14.98,
      "grad_norm": 8.36113395052962e-06,
      "learning_rate": 2.5117056856187294e-06,
      "loss": 0.0059,
      "step": 149800
    },
    {
      "epoch": 14.99,
      "grad_norm": 1.0185762221226469e-05,
      "learning_rate": 2.5100334448160534e-06,
      "loss": 0.004,
      "step": 149900
    },
    {
      "epoch": 15.0,
      "grad_norm": 4.18316267314367e-05,
      "learning_rate": 2.5083612040133783e-06,
      "loss": 0.0,
      "step": 150000
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9869,
      "eval_f1": 0.9869,
      "eval_loss": 0.13588353991508484,
      "eval_runtime": 135.008,
      "eval_samples_per_second": 296.279,
      "eval_steps_per_second": 296.279,
      "step": 150000
    },
    {
      "epoch": 15.01,
      "grad_norm": 2.9284803531481884e-05,
      "learning_rate": 2.5066889632107028e-06,
      "loss": 0.0014,
      "step": 150100
    },
    {
      "epoch": 15.02,
      "grad_norm": 2.252110789413564e-05,
      "learning_rate": 2.505016722408027e-06,
      "loss": 0.0037,
      "step": 150200
    },
    {
      "epoch": 15.03,
      "grad_norm": 7.923205885163043e-06,
      "learning_rate": 2.5033444816053513e-06,
      "loss": 0.0006,
      "step": 150300
    },
    {
      "epoch": 15.04,
      "grad_norm": 6.102809038566193e-06,
      "learning_rate": 2.5016722408026757e-06,
      "loss": 0.0064,
      "step": 150400
    },
    {
      "epoch": 15.05,
      "grad_norm": 6.924793979123933e-06,
      "learning_rate": 2.5e-06,
      "loss": 0.0,
      "step": 150500
    },
    {
      "epoch": 15.06,
      "grad_norm": 0.00029880605870857835,
      "learning_rate": 2.4983277591973247e-06,
      "loss": 0.0,
      "step": 150600
    },
    {
      "epoch": 15.07,
      "grad_norm": 5.71871669308166e-06,
      "learning_rate": 2.496655518394649e-06,
      "loss": 0.003,
      "step": 150700
    },
    {
      "epoch": 15.08,
      "grad_norm": 7.58796022637398e-06,
      "learning_rate": 2.4949832775919736e-06,
      "loss": 0.0,
      "step": 150800
    },
    {
      "epoch": 15.09,
      "grad_norm": 0.00011463899136288092,
      "learning_rate": 2.493311036789298e-06,
      "loss": 0.0011,
      "step": 150900
    },
    {
      "epoch": 15.1,
      "grad_norm": 9.345097168989014e-06,
      "learning_rate": 2.491638795986622e-06,
      "loss": 0.0,
      "step": 151000
    },
    {
      "epoch": 15.11,
      "grad_norm": 0.2711752653121948,
      "learning_rate": 2.4899665551839465e-06,
      "loss": 0.0,
      "step": 151100
    },
    {
      "epoch": 15.12,
      "grad_norm": 7.29593557480257e-06,
      "learning_rate": 2.488294314381271e-06,
      "loss": 0.0,
      "step": 151200
    },
    {
      "epoch": 15.13,
      "grad_norm": 9.790243893803563e-06,
      "learning_rate": 2.4866220735785955e-06,
      "loss": 0.0,
      "step": 151300
    },
    {
      "epoch": 15.14,
      "grad_norm": 2.9339826141949743e-05,
      "learning_rate": 2.48494983277592e-06,
      "loss": 0.0,
      "step": 151400
    },
    {
      "epoch": 15.15,
      "grad_norm": 1.5411867934744805e-05,
      "learning_rate": 2.4832775919732444e-06,
      "loss": 0.0002,
      "step": 151500
    },
    {
      "epoch": 15.16,
      "grad_norm": 5.220775165071245e-06,
      "learning_rate": 2.481605351170569e-06,
      "loss": 0.0,
      "step": 151600
    },
    {
      "epoch": 15.17,
      "grad_norm": 4.955260919814464e-06,
      "learning_rate": 2.479933110367893e-06,
      "loss": 0.0,
      "step": 151700
    },
    {
      "epoch": 15.18,
      "grad_norm": 4.50190918854787e-06,
      "learning_rate": 2.4782608695652178e-06,
      "loss": 0.0023,
      "step": 151800
    },
    {
      "epoch": 15.19,
      "grad_norm": 6.239830327103846e-06,
      "learning_rate": 2.4765886287625422e-06,
      "loss": 0.0,
      "step": 151900
    },
    {
      "epoch": 15.2,
      "grad_norm": 0.0002775177708826959,
      "learning_rate": 2.4749163879598663e-06,
      "loss": 0.0035,
      "step": 152000
    },
    {
      "epoch": 15.21,
      "grad_norm": 0.0001772949326550588,
      "learning_rate": 2.4732441471571907e-06,
      "loss": 0.0001,
      "step": 152100
    },
    {
      "epoch": 15.22,
      "grad_norm": 0.0001343058975180611,
      "learning_rate": 2.471571906354515e-06,
      "loss": 0.0035,
      "step": 152200
    },
    {
      "epoch": 15.23,
      "grad_norm": 4.652582447306486e-06,
      "learning_rate": 2.4698996655518396e-06,
      "loss": 0.0,
      "step": 152300
    },
    {
      "epoch": 15.24,
      "grad_norm": 4.857491603615927e-06,
      "learning_rate": 2.468227424749164e-06,
      "loss": 0.0,
      "step": 152400
    },
    {
      "epoch": 15.25,
      "grad_norm": 5.226095254329266e-06,
      "learning_rate": 2.4665551839464886e-06,
      "loss": 0.0,
      "step": 152500
    },
    {
      "epoch": 15.26,
      "grad_norm": 6.556086646014592e-06,
      "learning_rate": 2.464882943143813e-06,
      "loss": 0.0029,
      "step": 152600
    },
    {
      "epoch": 15.27,
      "grad_norm": 7.299048320419388e-06,
      "learning_rate": 2.463210702341137e-06,
      "loss": 0.0,
      "step": 152700
    },
    {
      "epoch": 15.28,
      "grad_norm": 6.213837878021877e-06,
      "learning_rate": 2.461538461538462e-06,
      "loss": 0.0,
      "step": 152800
    },
    {
      "epoch": 15.29,
      "grad_norm": 0.0003703912952914834,
      "learning_rate": 2.4598662207357864e-06,
      "loss": 0.0,
      "step": 152900
    },
    {
      "epoch": 15.3,
      "grad_norm": 3.938354893762153e-06,
      "learning_rate": 2.4581939799331104e-06,
      "loss": 0.0,
      "step": 153000
    },
    {
      "epoch": 15.31,
      "grad_norm": 1.6065227100625634e-05,
      "learning_rate": 2.456521739130435e-06,
      "loss": 0.0,
      "step": 153100
    },
    {
      "epoch": 15.32,
      "grad_norm": 1.110832454287447e-05,
      "learning_rate": 2.4548494983277594e-06,
      "loss": 0.0054,
      "step": 153200
    },
    {
      "epoch": 15.33,
      "grad_norm": 7.1401191235054284e-06,
      "learning_rate": 2.453177257525084e-06,
      "loss": 0.0032,
      "step": 153300
    },
    {
      "epoch": 15.34,
      "grad_norm": 6.627889433730161e-06,
      "learning_rate": 2.4515050167224083e-06,
      "loss": 0.0044,
      "step": 153400
    },
    {
      "epoch": 15.35,
      "grad_norm": 0.05635225027799606,
      "learning_rate": 2.4498327759197327e-06,
      "loss": 0.0051,
      "step": 153500
    },
    {
      "epoch": 15.36,
      "grad_norm": 4.67062136522145e-06,
      "learning_rate": 2.448160535117057e-06,
      "loss": 0.0009,
      "step": 153600
    },
    {
      "epoch": 15.37,
      "grad_norm": 0.0002716428425628692,
      "learning_rate": 2.4464882943143812e-06,
      "loss": 0.0068,
      "step": 153700
    },
    {
      "epoch": 15.38,
      "grad_norm": 3.66365748050157e-05,
      "learning_rate": 2.4448160535117057e-06,
      "loss": 0.0042,
      "step": 153800
    },
    {
      "epoch": 15.39,
      "grad_norm": 1.4453570656769443e-05,
      "learning_rate": 2.4431438127090306e-06,
      "loss": 0.0025,
      "step": 153900
    },
    {
      "epoch": 15.4,
      "grad_norm": 0.0001801799953682348,
      "learning_rate": 2.4414715719063546e-06,
      "loss": 0.0,
      "step": 154000
    },
    {
      "epoch": 15.41,
      "grad_norm": 7.897392606537323e-06,
      "learning_rate": 2.439799331103679e-06,
      "loss": 0.0082,
      "step": 154100
    },
    {
      "epoch": 15.42,
      "grad_norm": 0.0011347578838467598,
      "learning_rate": 2.4381270903010035e-06,
      "loss": 0.0012,
      "step": 154200
    },
    {
      "epoch": 15.43,
      "grad_norm": 8.59985357237747e-06,
      "learning_rate": 2.436454849498328e-06,
      "loss": 0.0,
      "step": 154300
    },
    {
      "epoch": 15.44,
      "grad_norm": 1.318647809966933e-05,
      "learning_rate": 2.4347826086956525e-06,
      "loss": 0.0,
      "step": 154400
    },
    {
      "epoch": 15.45,
      "grad_norm": 1.7026190107571892e-05,
      "learning_rate": 2.433110367892977e-06,
      "loss": 0.0008,
      "step": 154500
    },
    {
      "epoch": 15.46,
      "grad_norm": 5.7994143389805686e-06,
      "learning_rate": 2.4314381270903014e-06,
      "loss": 0.0039,
      "step": 154600
    },
    {
      "epoch": 15.47,
      "grad_norm": 5.057373982708668e-06,
      "learning_rate": 2.4297658862876254e-06,
      "loss": 0.0,
      "step": 154700
    },
    {
      "epoch": 15.48,
      "grad_norm": 5.878926458535716e-06,
      "learning_rate": 2.42809364548495e-06,
      "loss": 0.0,
      "step": 154800
    },
    {
      "epoch": 15.49,
      "grad_norm": 5.919266641285503e-06,
      "learning_rate": 2.4264214046822743e-06,
      "loss": 0.0,
      "step": 154900
    },
    {
      "epoch": 15.5,
      "grad_norm": 1.6888472600840032e-05,
      "learning_rate": 2.424749163879599e-06,
      "loss": 0.0,
      "step": 155000
    },
    {
      "epoch": 15.5,
      "eval_accuracy": 0.9874,
      "eval_f1": 0.9874,
      "eval_loss": 0.13571089506149292,
      "eval_runtime": 133.2814,
      "eval_samples_per_second": 300.117,
      "eval_steps_per_second": 300.117,
      "step": 155000
    },
    {
      "epoch": 15.51,
      "grad_norm": 0.0019413414411246777,
      "learning_rate": 2.4230769230769233e-06,
      "loss": 0.0043,
      "step": 155100
    },
    {
      "epoch": 15.52,
      "grad_norm": 2.8746006137225777e-05,
      "learning_rate": 2.4214046822742477e-06,
      "loss": 0.0,
      "step": 155200
    },
    {
      "epoch": 15.53,
      "grad_norm": 6.713657057844102e-05,
      "learning_rate": 2.419732441471572e-06,
      "loss": 0.0,
      "step": 155300
    },
    {
      "epoch": 15.54,
      "grad_norm": 0.01398865319788456,
      "learning_rate": 2.4180602006688967e-06,
      "loss": 0.0,
      "step": 155400
    },
    {
      "epoch": 15.55,
      "grad_norm": 8.839925430947915e-05,
      "learning_rate": 2.416387959866221e-06,
      "loss": 0.0,
      "step": 155500
    },
    {
      "epoch": 15.56,
      "grad_norm": 4.814656676899176e-06,
      "learning_rate": 2.414715719063545e-06,
      "loss": 0.0005,
      "step": 155600
    },
    {
      "epoch": 15.57,
      "grad_norm": 2.2645757780992426e-05,
      "learning_rate": 2.4130434782608696e-06,
      "loss": 0.0,
      "step": 155700
    },
    {
      "epoch": 15.58,
      "grad_norm": 6.3812894950388e-06,
      "learning_rate": 2.411371237458194e-06,
      "loss": 0.0018,
      "step": 155800
    },
    {
      "epoch": 15.59,
      "grad_norm": 9.400127964909188e-06,
      "learning_rate": 2.4096989966555185e-06,
      "loss": 0.0075,
      "step": 155900
    },
    {
      "epoch": 15.6,
      "grad_norm": 0.000328985188389197,
      "learning_rate": 2.408026755852843e-06,
      "loss": 0.0105,
      "step": 156000
    },
    {
      "epoch": 15.61,
      "grad_norm": 0.0001703462185105309,
      "learning_rate": 2.4063545150501674e-06,
      "loss": 0.0,
      "step": 156100
    },
    {
      "epoch": 15.62,
      "grad_norm": 7.5086682045366615e-06,
      "learning_rate": 2.404682274247492e-06,
      "loss": 0.0,
      "step": 156200
    },
    {
      "epoch": 15.63,
      "grad_norm": 4.592197001329623e-06,
      "learning_rate": 2.403010033444816e-06,
      "loss": 0.0,
      "step": 156300
    },
    {
      "epoch": 15.64,
      "grad_norm": 0.0002146685728803277,
      "learning_rate": 2.401337792642141e-06,
      "loss": 0.0,
      "step": 156400
    },
    {
      "epoch": 15.65,
      "grad_norm": 1.4054951861908194e-05,
      "learning_rate": 2.3996655518394653e-06,
      "loss": 0.0018,
      "step": 156500
    },
    {
      "epoch": 15.66,
      "grad_norm": 8.72565942700021e-06,
      "learning_rate": 2.3979933110367893e-06,
      "loss": 0.0,
      "step": 156600
    },
    {
      "epoch": 15.67,
      "grad_norm": 7.5215334618405905e-06,
      "learning_rate": 2.396321070234114e-06,
      "loss": 0.0,
      "step": 156700
    },
    {
      "epoch": 15.68,
      "grad_norm": 1.3494937775249127e-05,
      "learning_rate": 2.3946488294314382e-06,
      "loss": 0.0043,
      "step": 156800
    },
    {
      "epoch": 15.69,
      "grad_norm": 0.00142685417085886,
      "learning_rate": 2.3929765886287627e-06,
      "loss": 0.0,
      "step": 156900
    },
    {
      "epoch": 15.7,
      "grad_norm": 1.0469170774740633e-05,
      "learning_rate": 2.391304347826087e-06,
      "loss": 0.0025,
      "step": 157000
    },
    {
      "epoch": 15.71,
      "grad_norm": 6.9538637035293505e-06,
      "learning_rate": 2.3896321070234116e-06,
      "loss": 0.0,
      "step": 157100
    },
    {
      "epoch": 15.72,
      "grad_norm": 0.00019743008306249976,
      "learning_rate": 2.387959866220736e-06,
      "loss": 0.0,
      "step": 157200
    },
    {
      "epoch": 15.73,
      "grad_norm": 5.979543402645504e-06,
      "learning_rate": 2.38628762541806e-06,
      "loss": 0.0,
      "step": 157300
    },
    {
      "epoch": 15.74,
      "grad_norm": 9.36076685320586e-06,
      "learning_rate": 2.384615384615385e-06,
      "loss": 0.0,
      "step": 157400
    },
    {
      "epoch": 15.75,
      "grad_norm": 4.156654085818445e-06,
      "learning_rate": 2.3829431438127095e-06,
      "loss": 0.0,
      "step": 157500
    },
    {
      "epoch": 15.76,
      "grad_norm": 0.0005622885655611753,
      "learning_rate": 2.3812709030100335e-06,
      "loss": 0.0045,
      "step": 157600
    },
    {
      "epoch": 15.77,
      "grad_norm": 1.6783771570771933e-05,
      "learning_rate": 2.379598662207358e-06,
      "loss": 0.0,
      "step": 157700
    },
    {
      "epoch": 15.78,
      "grad_norm": 2.4302729798364453e-05,
      "learning_rate": 2.3779264214046824e-06,
      "loss": 0.0,
      "step": 157800
    },
    {
      "epoch": 15.79,
      "grad_norm": 4.918417289445642e-06,
      "learning_rate": 2.376254180602007e-06,
      "loss": 0.0065,
      "step": 157900
    },
    {
      "epoch": 15.8,
      "grad_norm": 0.00043107534293085337,
      "learning_rate": 2.3745819397993314e-06,
      "loss": 0.002,
      "step": 158000
    },
    {
      "epoch": 15.81,
      "grad_norm": 6.5696453930286225e-06,
      "learning_rate": 2.372909698996656e-06,
      "loss": 0.0036,
      "step": 158100
    },
    {
      "epoch": 15.82,
      "grad_norm": 6.5576014094403945e-06,
      "learning_rate": 2.3712374581939803e-06,
      "loss": 0.0,
      "step": 158200
    },
    {
      "epoch": 15.83,
      "grad_norm": 6.10633696851437e-06,
      "learning_rate": 2.3695652173913043e-06,
      "loss": 0.0,
      "step": 158300
    },
    {
      "epoch": 15.84,
      "grad_norm": 2.4489367206115276e-05,
      "learning_rate": 2.3678929765886288e-06,
      "loss": 0.0,
      "step": 158400
    },
    {
      "epoch": 15.85,
      "grad_norm": 9.810268238652498e-06,
      "learning_rate": 2.3662207357859537e-06,
      "loss": 0.0,
      "step": 158500
    },
    {
      "epoch": 15.86,
      "grad_norm": 5.690500529453857e-06,
      "learning_rate": 2.3645484949832777e-06,
      "loss": 0.0,
      "step": 158600
    },
    {
      "epoch": 15.87,
      "grad_norm": 5.2947007134207524e-06,
      "learning_rate": 2.362876254180602e-06,
      "loss": 0.0,
      "step": 158700
    },
    {
      "epoch": 15.88,
      "grad_norm": 1.6920499547268264e-05,
      "learning_rate": 2.3612040133779266e-06,
      "loss": 0.0,
      "step": 158800
    },
    {
      "epoch": 15.89,
      "grad_norm": 6.31178954790812e-06,
      "learning_rate": 2.359531772575251e-06,
      "loss": 0.0,
      "step": 158900
    },
    {
      "epoch": 15.9,
      "grad_norm": 1.0661011401680298e-05,
      "learning_rate": 2.3578595317725755e-06,
      "loss": 0.0,
      "step": 159000
    },
    {
      "epoch": 15.91,
      "grad_norm": 2.5542381990817375e-05,
      "learning_rate": 2.3561872909699e-06,
      "loss": 0.0115,
      "step": 159100
    },
    {
      "epoch": 15.92,
      "grad_norm": 6.1712162278126925e-06,
      "learning_rate": 2.3545150501672245e-06,
      "loss": 0.0,
      "step": 159200
    },
    {
      "epoch": 15.93,
      "grad_norm": 0.0012592824641615152,
      "learning_rate": 2.3528428093645485e-06,
      "loss": 0.0,
      "step": 159300
    },
    {
      "epoch": 15.94,
      "grad_norm": 1.3988282262289431e-05,
      "learning_rate": 2.351170568561873e-06,
      "loss": 0.0,
      "step": 159400
    },
    {
      "epoch": 15.95,
      "grad_norm": 0.0015318294754251838,
      "learning_rate": 2.3494983277591974e-06,
      "loss": 0.0,
      "step": 159500
    },
    {
      "epoch": 15.96,
      "grad_norm": 1.7320944607490674e-05,
      "learning_rate": 2.347826086956522e-06,
      "loss": 0.0,
      "step": 159600
    },
    {
      "epoch": 15.97,
      "grad_norm": 6.387051598721882e-06,
      "learning_rate": 2.3461538461538463e-06,
      "loss": 0.0,
      "step": 159700
    },
    {
      "epoch": 15.98,
      "grad_norm": 0.0009833506774157286,
      "learning_rate": 2.344481605351171e-06,
      "loss": 0.0,
      "step": 159800
    },
    {
      "epoch": 15.99,
      "grad_norm": 0.005238015204668045,
      "learning_rate": 2.3428093645484953e-06,
      "loss": 0.0036,
      "step": 159900
    },
    {
      "epoch": 16.0,
      "grad_norm": 8.068822353379801e-06,
      "learning_rate": 2.3411371237458197e-06,
      "loss": 0.0018,
      "step": 160000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.986525,
      "eval_f1": 0.986525,
      "eval_loss": 0.1413874477148056,
      "eval_runtime": 134.4403,
      "eval_samples_per_second": 297.53,
      "eval_steps_per_second": 297.53,
      "step": 160000
    },
    {
      "epoch": 16.01,
      "grad_norm": 0.000655919371638447,
      "learning_rate": 2.339464882943144e-06,
      "loss": 0.0037,
      "step": 160100
    },
    {
      "epoch": 16.02,
      "grad_norm": 5.729572876589373e-06,
      "learning_rate": 2.337792642140468e-06,
      "loss": 0.0,
      "step": 160200
    },
    {
      "epoch": 16.03,
      "grad_norm": 3.707243740791455e-05,
      "learning_rate": 2.3361204013377927e-06,
      "loss": 0.0018,
      "step": 160300
    },
    {
      "epoch": 16.04,
      "grad_norm": 8.37804691400379e-05,
      "learning_rate": 2.334448160535117e-06,
      "loss": 0.0056,
      "step": 160400
    },
    {
      "epoch": 16.05,
      "grad_norm": 0.00018116727005690336,
      "learning_rate": 2.3327759197324416e-06,
      "loss": 0.0,
      "step": 160500
    },
    {
      "epoch": 16.06,
      "grad_norm": 5.471560143632814e-06,
      "learning_rate": 2.331103678929766e-06,
      "loss": 0.0,
      "step": 160600
    },
    {
      "epoch": 16.07,
      "grad_norm": 0.002342527499422431,
      "learning_rate": 2.3294314381270905e-06,
      "loss": 0.0,
      "step": 160700
    },
    {
      "epoch": 16.08,
      "grad_norm": 5.308319941832451e-06,
      "learning_rate": 2.327759197324415e-06,
      "loss": 0.0021,
      "step": 160800
    },
    {
      "epoch": 16.09,
      "grad_norm": 1.8882026779465377e-05,
      "learning_rate": 2.326086956521739e-06,
      "loss": 0.0,
      "step": 160900
    },
    {
      "epoch": 16.1,
      "grad_norm": 5.862941179657355e-06,
      "learning_rate": 2.324414715719064e-06,
      "loss": 0.0026,
      "step": 161000
    },
    {
      "epoch": 16.11,
      "grad_norm": 9.476354534854181e-06,
      "learning_rate": 2.3227424749163884e-06,
      "loss": 0.0,
      "step": 161100
    },
    {
      "epoch": 16.12,
      "grad_norm": 1.4225512131815776e-05,
      "learning_rate": 2.3210702341137124e-06,
      "loss": 0.0018,
      "step": 161200
    },
    {
      "epoch": 16.13,
      "grad_norm": 3.2810748962219805e-05,
      "learning_rate": 2.319397993311037e-06,
      "loss": 0.0024,
      "step": 161300
    },
    {
      "epoch": 16.14,
      "grad_norm": 0.006205264013260603,
      "learning_rate": 2.3177257525083613e-06,
      "loss": 0.0048,
      "step": 161400
    },
    {
      "epoch": 16.15,
      "grad_norm": 2.7496673283167183e-05,
      "learning_rate": 2.3160535117056858e-06,
      "loss": 0.0001,
      "step": 161500
    },
    {
      "epoch": 16.16,
      "grad_norm": 2.4235663659055717e-05,
      "learning_rate": 2.3143812709030102e-06,
      "loss": 0.003,
      "step": 161600
    },
    {
      "epoch": 16.17,
      "grad_norm": 4.839419489144348e-06,
      "learning_rate": 2.3127090301003347e-06,
      "loss": 0.0,
      "step": 161700
    },
    {
      "epoch": 16.18,
      "grad_norm": 6.52590169920586e-05,
      "learning_rate": 2.311036789297659e-06,
      "loss": 0.0019,
      "step": 161800
    },
    {
      "epoch": 16.19,
      "grad_norm": 5.044812041887781e-06,
      "learning_rate": 2.309364548494983e-06,
      "loss": 0.0,
      "step": 161900
    },
    {
      "epoch": 16.2,
      "grad_norm": 7.806534085830208e-06,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.0,
      "step": 162000
    },
    {
      "epoch": 16.21,
      "grad_norm": 0.0001349828380625695,
      "learning_rate": 2.3060200668896325e-06,
      "loss": 0.0,
      "step": 162100
    },
    {
      "epoch": 16.22,
      "grad_norm": 5.507646164915059e-06,
      "learning_rate": 2.3043478260869566e-06,
      "loss": 0.0,
      "step": 162200
    },
    {
      "epoch": 16.23,
      "grad_norm": 5.797621724923374e-06,
      "learning_rate": 2.302675585284281e-06,
      "loss": 0.0,
      "step": 162300
    },
    {
      "epoch": 16.24,
      "grad_norm": 7.901490607764572e-05,
      "learning_rate": 2.3010033444816055e-06,
      "loss": 0.0,
      "step": 162400
    },
    {
      "epoch": 16.25,
      "grad_norm": 5.0614430620044e-06,
      "learning_rate": 2.29933110367893e-06,
      "loss": 0.0,
      "step": 162500
    },
    {
      "epoch": 16.26,
      "grad_norm": 9.34005220187828e-06,
      "learning_rate": 2.2976588628762544e-06,
      "loss": 0.0,
      "step": 162600
    },
    {
      "epoch": 16.27,
      "grad_norm": 6.163528450997546e-05,
      "learning_rate": 2.295986622073579e-06,
      "loss": 0.0,
      "step": 162700
    },
    {
      "epoch": 16.28,
      "grad_norm": 4.1326575228595175e-06,
      "learning_rate": 2.2943143812709033e-06,
      "loss": 0.0003,
      "step": 162800
    },
    {
      "epoch": 16.29,
      "grad_norm": 4.5336778384807985e-06,
      "learning_rate": 2.2926421404682274e-06,
      "loss": 0.0,
      "step": 162900
    },
    {
      "epoch": 16.3,
      "grad_norm": 5.2979930842411704e-06,
      "learning_rate": 2.290969899665552e-06,
      "loss": 0.003,
      "step": 163000
    },
    {
      "epoch": 16.31,
      "grad_norm": 0.0012195265153422952,
      "learning_rate": 2.2892976588628767e-06,
      "loss": 0.0,
      "step": 163100
    },
    {
      "epoch": 16.32,
      "grad_norm": 0.003278920892626047,
      "learning_rate": 2.2876254180602008e-06,
      "loss": 0.0,
      "step": 163200
    },
    {
      "epoch": 16.33,
      "grad_norm": 6.400030542863533e-05,
      "learning_rate": 2.2859531772575252e-06,
      "loss": 0.0035,
      "step": 163300
    },
    {
      "epoch": 16.34,
      "grad_norm": 1.2621836503967643e-05,
      "learning_rate": 2.2842809364548497e-06,
      "loss": 0.0011,
      "step": 163400
    },
    {
      "epoch": 16.35,
      "grad_norm": 0.0002729544066824019,
      "learning_rate": 2.282608695652174e-06,
      "loss": 0.0,
      "step": 163500
    },
    {
      "epoch": 16.36,
      "grad_norm": 0.00021667926921509206,
      "learning_rate": 2.2809364548494986e-06,
      "loss": 0.0,
      "step": 163600
    },
    {
      "epoch": 16.37,
      "grad_norm": 4.100193837075494e-05,
      "learning_rate": 2.279264214046823e-06,
      "loss": 0.0,
      "step": 163700
    },
    {
      "epoch": 16.38,
      "grad_norm": 4.1616758608142845e-06,
      "learning_rate": 2.2775919732441475e-06,
      "loss": 0.0,
      "step": 163800
    },
    {
      "epoch": 16.39,
      "grad_norm": 443.37982177734375,
      "learning_rate": 2.2759197324414716e-06,
      "loss": 0.0012,
      "step": 163900
    },
    {
      "epoch": 16.4,
      "grad_norm": 5.086914370622253e-06,
      "learning_rate": 2.274247491638796e-06,
      "loss": 0.0035,
      "step": 164000
    },
    {
      "epoch": 16.41,
      "grad_norm": 6.053800007066457e-06,
      "learning_rate": 2.2725752508361205e-06,
      "loss": 0.0,
      "step": 164100
    },
    {
      "epoch": 16.42,
      "grad_norm": 3.32295780935965e-06,
      "learning_rate": 2.270903010033445e-06,
      "loss": 0.0071,
      "step": 164200
    },
    {
      "epoch": 16.43,
      "grad_norm": 1.0716218639572617e-05,
      "learning_rate": 2.2692307692307694e-06,
      "loss": 0.0057,
      "step": 164300
    },
    {
      "epoch": 16.44,
      "grad_norm": 4.743649242300307e-06,
      "learning_rate": 2.267558528428094e-06,
      "loss": 0.0,
      "step": 164400
    },
    {
      "epoch": 16.45,
      "grad_norm": 0.00011757192260120064,
      "learning_rate": 2.2658862876254183e-06,
      "loss": 0.0065,
      "step": 164500
    },
    {
      "epoch": 16.46,
      "grad_norm": 5.40465880476404e-05,
      "learning_rate": 2.2642140468227428e-06,
      "loss": 0.0,
      "step": 164600
    },
    {
      "epoch": 16.47,
      "grad_norm": 6.664827378699556e-05,
      "learning_rate": 2.2625418060200672e-06,
      "loss": 0.0006,
      "step": 164700
    },
    {
      "epoch": 16.48,
      "grad_norm": 4.517704383033561e-06,
      "learning_rate": 2.2608695652173913e-06,
      "loss": 0.0025,
      "step": 164800
    },
    {
      "epoch": 16.49,
      "grad_norm": 3.920742983609671e-06,
      "learning_rate": 2.2591973244147157e-06,
      "loss": 0.0,
      "step": 164900
    },
    {
      "epoch": 16.5,
      "grad_norm": 7.613572961417958e-06,
      "learning_rate": 2.25752508361204e-06,
      "loss": 0.0,
      "step": 165000
    },
    {
      "epoch": 16.5,
      "eval_accuracy": 0.98765,
      "eval_f1": 0.98765,
      "eval_loss": 0.1337709277868271,
      "eval_runtime": 134.8045,
      "eval_samples_per_second": 296.726,
      "eval_steps_per_second": 296.726,
      "step": 165000
    },
    {
      "epoch": 16.51,
      "grad_norm": 1.6788189896033145e-05,
      "learning_rate": 2.2558528428093647e-06,
      "loss": 0.0,
      "step": 165100
    },
    {
      "epoch": 16.52,
      "grad_norm": 5.329384748620214e-06,
      "learning_rate": 2.254180602006689e-06,
      "loss": 0.0,
      "step": 165200
    },
    {
      "epoch": 16.53,
      "grad_norm": 6.522398052766221e-06,
      "learning_rate": 2.2525083612040136e-06,
      "loss": 0.0,
      "step": 165300
    },
    {
      "epoch": 16.54,
      "grad_norm": 8.143794730131049e-06,
      "learning_rate": 2.250836120401338e-06,
      "loss": 0.0,
      "step": 165400
    },
    {
      "epoch": 16.55,
      "grad_norm": 4.95312488055788e-05,
      "learning_rate": 2.249163879598662e-06,
      "loss": 0.0001,
      "step": 165500
    },
    {
      "epoch": 16.56,
      "grad_norm": 5.945916655036854e-06,
      "learning_rate": 2.247491638795987e-06,
      "loss": 0.0037,
      "step": 165600
    },
    {
      "epoch": 16.57,
      "grad_norm": 0.0021674486342817545,
      "learning_rate": 2.2458193979933114e-06,
      "loss": 0.0,
      "step": 165700
    },
    {
      "epoch": 16.58,
      "grad_norm": 3.8080481317592785e-05,
      "learning_rate": 2.2441471571906355e-06,
      "loss": 0.0071,
      "step": 165800
    },
    {
      "epoch": 16.59,
      "grad_norm": 2.418838812445756e-05,
      "learning_rate": 2.24247491638796e-06,
      "loss": 0.0029,
      "step": 165900
    },
    {
      "epoch": 16.6,
      "grad_norm": 4.79162554256618e-06,
      "learning_rate": 2.2408026755852844e-06,
      "loss": 0.0001,
      "step": 166000
    },
    {
      "epoch": 16.61,
      "grad_norm": 5.906681963097071e-06,
      "learning_rate": 2.239130434782609e-06,
      "loss": 0.0082,
      "step": 166100
    },
    {
      "epoch": 16.62,
      "grad_norm": 7.5729653872258496e-06,
      "learning_rate": 2.2374581939799333e-06,
      "loss": 0.0027,
      "step": 166200
    },
    {
      "epoch": 16.63,
      "grad_norm": 0.00019007065566256642,
      "learning_rate": 2.2357859531772578e-06,
      "loss": 0.0,
      "step": 166300
    },
    {
      "epoch": 16.64,
      "grad_norm": 3.341139017720707e-05,
      "learning_rate": 2.2341137123745822e-06,
      "loss": 0.0012,
      "step": 166400
    },
    {
      "epoch": 16.65,
      "grad_norm": 5.296933977660956e-06,
      "learning_rate": 2.2324414715719063e-06,
      "loss": 0.0019,
      "step": 166500
    },
    {
      "epoch": 16.66,
      "grad_norm": 0.0005588891799561679,
      "learning_rate": 2.230769230769231e-06,
      "loss": 0.0035,
      "step": 166600
    },
    {
      "epoch": 16.67,
      "grad_norm": 0.0007428290555253625,
      "learning_rate": 2.2290969899665556e-06,
      "loss": 0.0014,
      "step": 166700
    },
    {
      "epoch": 16.68,
      "grad_norm": 1.5256994629453402e-05,
      "learning_rate": 2.2274247491638796e-06,
      "loss": 0.0,
      "step": 166800
    },
    {
      "epoch": 16.69,
      "grad_norm": 3.1883606425253674e-05,
      "learning_rate": 2.225752508361204e-06,
      "loss": 0.0,
      "step": 166900
    },
    {
      "epoch": 16.7,
      "grad_norm": 7.867512067605276e-06,
      "learning_rate": 2.2240802675585286e-06,
      "loss": 0.0,
      "step": 167000
    },
    {
      "epoch": 16.71,
      "grad_norm": 6.410631158360047e-06,
      "learning_rate": 2.222408026755853e-06,
      "loss": 0.0,
      "step": 167100
    },
    {
      "epoch": 16.72,
      "grad_norm": 9.276688615500461e-06,
      "learning_rate": 2.2207357859531775e-06,
      "loss": 0.0,
      "step": 167200
    },
    {
      "epoch": 16.73,
      "grad_norm": 6.474059318861691e-06,
      "learning_rate": 2.219063545150502e-06,
      "loss": 0.0,
      "step": 167300
    },
    {
      "epoch": 16.74,
      "grad_norm": 6.038425908627687e-06,
      "learning_rate": 2.2173913043478264e-06,
      "loss": 0.0042,
      "step": 167400
    },
    {
      "epoch": 16.75,
      "grad_norm": 4.583041118166875e-06,
      "learning_rate": 2.2157190635451504e-06,
      "loss": 0.0,
      "step": 167500
    },
    {
      "epoch": 16.76,
      "grad_norm": 5.181601954973303e-05,
      "learning_rate": 2.2140468227424753e-06,
      "loss": 0.0,
      "step": 167600
    },
    {
      "epoch": 16.77,
      "grad_norm": 3.553269925760105e-05,
      "learning_rate": 2.2123745819397998e-06,
      "loss": 0.0,
      "step": 167700
    },
    {
      "epoch": 16.78,
      "grad_norm": 0.00026847337721847,
      "learning_rate": 2.210702341137124e-06,
      "loss": 0.0,
      "step": 167800
    },
    {
      "epoch": 16.79,
      "grad_norm": 0.00020925169519614428,
      "learning_rate": 2.2090301003344483e-06,
      "loss": 0.0,
      "step": 167900
    },
    {
      "epoch": 16.8,
      "grad_norm": 0.00029760925099253654,
      "learning_rate": 2.2073578595317727e-06,
      "loss": 0.0,
      "step": 168000
    },
    {
      "epoch": 16.81,
      "grad_norm": 8.02914200903615e-06,
      "learning_rate": 2.205685618729097e-06,
      "loss": 0.0,
      "step": 168100
    },
    {
      "epoch": 16.82,
      "grad_norm": 4.700968929682858e-06,
      "learning_rate": 2.2040133779264217e-06,
      "loss": 0.0,
      "step": 168200
    },
    {
      "epoch": 16.83,
      "grad_norm": 67.33990478515625,
      "learning_rate": 2.202341137123746e-06,
      "loss": 0.0033,
      "step": 168300
    },
    {
      "epoch": 16.84,
      "grad_norm": 0.00011456763604655862,
      "learning_rate": 2.2006688963210706e-06,
      "loss": 0.0,
      "step": 168400
    },
    {
      "epoch": 16.85,
      "grad_norm": 0.0007877860916778445,
      "learning_rate": 2.1989966555183946e-06,
      "loss": 0.0,
      "step": 168500
    },
    {
      "epoch": 16.86,
      "grad_norm": 2.5768347313714912e-06,
      "learning_rate": 2.197324414715719e-06,
      "loss": 0.0,
      "step": 168600
    },
    {
      "epoch": 16.87,
      "grad_norm": 3.3434796932851896e-05,
      "learning_rate": 2.1956521739130435e-06,
      "loss": 0.0,
      "step": 168700
    },
    {
      "epoch": 16.88,
      "grad_norm": 4.778345555678243e-06,
      "learning_rate": 2.193979933110368e-06,
      "loss": 0.0,
      "step": 168800
    },
    {
      "epoch": 16.89,
      "grad_norm": 4.4025196075381245e-06,
      "learning_rate": 2.1923076923076925e-06,
      "loss": 0.0,
      "step": 168900
    },
    {
      "epoch": 16.9,
      "grad_norm": 3.4835945825761883e-06,
      "learning_rate": 2.190635451505017e-06,
      "loss": 0.0,
      "step": 169000
    },
    {
      "epoch": 16.91,
      "grad_norm": 4.608343715517549e-06,
      "learning_rate": 2.1889632107023414e-06,
      "loss": 0.0,
      "step": 169100
    },
    {
      "epoch": 16.92,
      "grad_norm": 2.3206870537251234e-05,
      "learning_rate": 2.187290969899666e-06,
      "loss": 0.0,
      "step": 169200
    },
    {
      "epoch": 16.93,
      "grad_norm": 1.2776320545526687e-05,
      "learning_rate": 2.1856187290969903e-06,
      "loss": 0.0046,
      "step": 169300
    },
    {
      "epoch": 16.94,
      "grad_norm": 1.215246993524488e-05,
      "learning_rate": 2.1839464882943143e-06,
      "loss": 0.0,
      "step": 169400
    },
    {
      "epoch": 16.95,
      "grad_norm": 4.593052381096641e-06,
      "learning_rate": 2.182274247491639e-06,
      "loss": 0.0,
      "step": 169500
    },
    {
      "epoch": 16.96,
      "grad_norm": 3.894136898452416e-05,
      "learning_rate": 2.1806020066889633e-06,
      "loss": 0.0047,
      "step": 169600
    },
    {
      "epoch": 16.97,
      "grad_norm": 1.1413544598326553e-05,
      "learning_rate": 2.1789297658862877e-06,
      "loss": 0.0108,
      "step": 169700
    },
    {
      "epoch": 16.98,
      "grad_norm": 9.005904757941607e-06,
      "learning_rate": 2.177257525083612e-06,
      "loss": 0.0032,
      "step": 169800
    },
    {
      "epoch": 16.99,
      "grad_norm": 9.968599624698982e-05,
      "learning_rate": 2.1755852842809366e-06,
      "loss": 0.0086,
      "step": 169900
    },
    {
      "epoch": 17.0,
      "grad_norm": 0.00240413099527359,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.003,
      "step": 170000
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.986175,
      "eval_f1": 0.986175,
      "eval_loss": 0.14242902398109436,
      "eval_runtime": 134.4229,
      "eval_samples_per_second": 297.568,
      "eval_steps_per_second": 297.568,
      "step": 170000
    },
    {
      "epoch": 17.01,
      "grad_norm": 1.1759045264625456e-05,
      "learning_rate": 2.172240802675585e-06,
      "loss": 0.0109,
      "step": 170100
    },
    {
      "epoch": 17.02,
      "grad_norm": 6.563326678588055e-06,
      "learning_rate": 2.17056856187291e-06,
      "loss": 0.0,
      "step": 170200
    },
    {
      "epoch": 17.03,
      "grad_norm": 0.054090823978185654,
      "learning_rate": 2.1688963210702345e-06,
      "loss": 0.0,
      "step": 170300
    },
    {
      "epoch": 17.04,
      "grad_norm": 9.796851372811943e-05,
      "learning_rate": 2.1672240802675585e-06,
      "loss": 0.0,
      "step": 170400
    },
    {
      "epoch": 17.05,
      "grad_norm": 4.6282475523184985e-06,
      "learning_rate": 2.165551839464883e-06,
      "loss": 0.0,
      "step": 170500
    },
    {
      "epoch": 17.06,
      "grad_norm": 4.962850198353408e-06,
      "learning_rate": 2.1638795986622074e-06,
      "loss": 0.0,
      "step": 170600
    },
    {
      "epoch": 17.07,
      "grad_norm": 0.00048357073683291674,
      "learning_rate": 2.162207357859532e-06,
      "loss": 0.0,
      "step": 170700
    },
    {
      "epoch": 17.08,
      "grad_norm": 3.0319617508212104e-05,
      "learning_rate": 2.1605351170568564e-06,
      "loss": 0.0,
      "step": 170800
    },
    {
      "epoch": 17.09,
      "grad_norm": 0.0008551038918085396,
      "learning_rate": 2.158862876254181e-06,
      "loss": 0.0,
      "step": 170900
    },
    {
      "epoch": 17.1,
      "grad_norm": 6.037145340087591e-06,
      "learning_rate": 2.1571906354515053e-06,
      "loss": 0.0,
      "step": 171000
    },
    {
      "epoch": 17.11,
      "grad_norm": 6.677675173705211e-06,
      "learning_rate": 2.1555183946488293e-06,
      "loss": 0.0001,
      "step": 171100
    },
    {
      "epoch": 17.12,
      "grad_norm": 5.688702731276862e-05,
      "learning_rate": 2.153846153846154e-06,
      "loss": 0.0,
      "step": 171200
    },
    {
      "epoch": 17.13,
      "grad_norm": 5.400210739026079e-06,
      "learning_rate": 2.1521739130434787e-06,
      "loss": 0.003,
      "step": 171300
    },
    {
      "epoch": 17.14,
      "grad_norm": 7.61369210522389e-06,
      "learning_rate": 2.1505016722408027e-06,
      "loss": 0.0001,
      "step": 171400
    },
    {
      "epoch": 17.15,
      "grad_norm": 4.224335953040281e-06,
      "learning_rate": 2.148829431438127e-06,
      "loss": 0.0,
      "step": 171500
    },
    {
      "epoch": 17.16,
      "grad_norm": 0.0001257136173080653,
      "learning_rate": 2.1471571906354516e-06,
      "loss": 0.0,
      "step": 171600
    },
    {
      "epoch": 17.17,
      "grad_norm": 2.4170622054953128e-05,
      "learning_rate": 2.145484949832776e-06,
      "loss": 0.0,
      "step": 171700
    },
    {
      "epoch": 17.18,
      "grad_norm": 4.489980256039416e-06,
      "learning_rate": 2.1438127090301005e-06,
      "loss": 0.0,
      "step": 171800
    },
    {
      "epoch": 17.19,
      "grad_norm": 3.3881981380545767e-06,
      "learning_rate": 2.142140468227425e-06,
      "loss": 0.0,
      "step": 171900
    },
    {
      "epoch": 17.2,
      "grad_norm": 1.6857806258485653e-05,
      "learning_rate": 2.1404682274247495e-06,
      "loss": 0.0,
      "step": 172000
    },
    {
      "epoch": 17.21,
      "grad_norm": 3.3245811209781095e-05,
      "learning_rate": 2.1387959866220735e-06,
      "loss": 0.0,
      "step": 172100
    },
    {
      "epoch": 17.22,
      "grad_norm": 9.031922672875226e-06,
      "learning_rate": 2.1371237458193984e-06,
      "loss": 0.0,
      "step": 172200
    },
    {
      "epoch": 17.23,
      "grad_norm": 2.7925609629164683e-06,
      "learning_rate": 2.135451505016723e-06,
      "loss": 0.0,
      "step": 172300
    },
    {
      "epoch": 17.24,
      "grad_norm": 6.467849289037986e-06,
      "learning_rate": 2.133779264214047e-06,
      "loss": 0.0,
      "step": 172400
    },
    {
      "epoch": 17.25,
      "grad_norm": 1.820028046495281e-05,
      "learning_rate": 2.1321070234113713e-06,
      "loss": 0.0,
      "step": 172500
    },
    {
      "epoch": 17.26,
      "grad_norm": 5.631500698655145e-06,
      "learning_rate": 2.130434782608696e-06,
      "loss": 0.0003,
      "step": 172600
    },
    {
      "epoch": 17.27,
      "grad_norm": 0.00015545215865131468,
      "learning_rate": 2.1287625418060203e-06,
      "loss": 0.0,
      "step": 172700
    },
    {
      "epoch": 17.28,
      "grad_norm": 0.00011321782949380577,
      "learning_rate": 2.1270903010033447e-06,
      "loss": 0.0,
      "step": 172800
    },
    {
      "epoch": 17.29,
      "grad_norm": 4.420689947437495e-06,
      "learning_rate": 2.125418060200669e-06,
      "loss": 0.0,
      "step": 172900
    },
    {
      "epoch": 17.3,
      "grad_norm": 4.325940608396195e-06,
      "learning_rate": 2.1237458193979936e-06,
      "loss": 0.0,
      "step": 173000
    },
    {
      "epoch": 17.31,
      "grad_norm": 2.6896172130363993e-05,
      "learning_rate": 2.1220735785953177e-06,
      "loss": 0.0,
      "step": 173100
    },
    {
      "epoch": 17.32,
      "grad_norm": 2.51541132456623e-06,
      "learning_rate": 2.120401337792642e-06,
      "loss": 0.0,
      "step": 173200
    },
    {
      "epoch": 17.33,
      "grad_norm": 0.0002507483586668968,
      "learning_rate": 2.118729096989967e-06,
      "loss": 0.0,
      "step": 173300
    },
    {
      "epoch": 17.34,
      "grad_norm": 1.9010116147910594e-06,
      "learning_rate": 2.117056856187291e-06,
      "loss": 0.0,
      "step": 173400
    },
    {
      "epoch": 17.35,
      "grad_norm": 4.968646408087807e-06,
      "learning_rate": 2.1153846153846155e-06,
      "loss": 0.0,
      "step": 173500
    },
    {
      "epoch": 17.36,
      "grad_norm": 2.4034538910200354e-06,
      "learning_rate": 2.11371237458194e-06,
      "loss": 0.0,
      "step": 173600
    },
    {
      "epoch": 17.37,
      "grad_norm": 3.1217680316331098e-06,
      "learning_rate": 2.1120401337792644e-06,
      "loss": 0.0,
      "step": 173700
    },
    {
      "epoch": 17.38,
      "grad_norm": 4.924338099954184e-06,
      "learning_rate": 2.110367892976589e-06,
      "loss": 0.0,
      "step": 173800
    },
    {
      "epoch": 17.39,
      "grad_norm": 3.1763579499966e-06,
      "learning_rate": 2.1086956521739134e-06,
      "loss": 0.015,
      "step": 173900
    },
    {
      "epoch": 17.4,
      "grad_norm": 8.308285032398999e-06,
      "learning_rate": 2.1070234113712374e-06,
      "loss": 0.0,
      "step": 174000
    },
    {
      "epoch": 17.41,
      "grad_norm": 4.124725364818005e-06,
      "learning_rate": 2.105351170568562e-06,
      "loss": 0.0039,
      "step": 174100
    },
    {
      "epoch": 17.42,
      "grad_norm": 1.6264404621324502e-05,
      "learning_rate": 2.1036789297658863e-06,
      "loss": 0.0,
      "step": 174200
    },
    {
      "epoch": 17.43,
      "grad_norm": 1.3464480616676155e-05,
      "learning_rate": 2.1020066889632108e-06,
      "loss": 0.0,
      "step": 174300
    },
    {
      "epoch": 17.44,
      "grad_norm": 3.041270247194916e-06,
      "learning_rate": 2.1003344481605352e-06,
      "loss": 0.0,
      "step": 174400
    },
    {
      "epoch": 17.45,
      "grad_norm": 2.5817935238592327e-06,
      "learning_rate": 2.0986622073578597e-06,
      "loss": 0.0,
      "step": 174500
    },
    {
      "epoch": 17.46,
      "grad_norm": 5.324686935637146e-05,
      "learning_rate": 2.096989966555184e-06,
      "loss": 0.0,
      "step": 174600
    },
    {
      "epoch": 17.47,
      "grad_norm": 8.143258128257003e-06,
      "learning_rate": 2.095317725752508e-06,
      "loss": 0.0,
      "step": 174700
    },
    {
      "epoch": 17.48,
      "grad_norm": 0.00127193215303123,
      "learning_rate": 2.093645484949833e-06,
      "loss": 0.0017,
      "step": 174800
    },
    {
      "epoch": 17.49,
      "grad_norm": 0.003099586581811309,
      "learning_rate": 2.0919732441471575e-06,
      "loss": 0.0,
      "step": 174900
    },
    {
      "epoch": 17.5,
      "grad_norm": 5.266407697490649e-06,
      "learning_rate": 2.0903010033444816e-06,
      "loss": 0.0,
      "step": 175000
    },
    {
      "epoch": 17.5,
      "eval_accuracy": 0.9876,
      "eval_f1": 0.9876,
      "eval_loss": 0.14104844629764557,
      "eval_runtime": 135.6218,
      "eval_samples_per_second": 294.938,
      "eval_steps_per_second": 294.938,
      "step": 175000
    },
    {
      "epoch": 17.51,
      "grad_norm": 5.470873020385625e-06,
      "learning_rate": 2.088628762541806e-06,
      "loss": 0.0027,
      "step": 175100
    },
    {
      "epoch": 17.52,
      "grad_norm": 3.602494871302042e-06,
      "learning_rate": 2.0869565217391305e-06,
      "loss": 0.0,
      "step": 175200
    },
    {
      "epoch": 17.53,
      "grad_norm": 3.7983577385602985e-06,
      "learning_rate": 2.085284280936455e-06,
      "loss": 0.0049,
      "step": 175300
    },
    {
      "epoch": 17.54,
      "grad_norm": 2.3961443730513565e-05,
      "learning_rate": 2.0836120401337794e-06,
      "loss": 0.0,
      "step": 175400
    },
    {
      "epoch": 17.55,
      "grad_norm": 0.0007328629144467413,
      "learning_rate": 2.081939799331104e-06,
      "loss": 0.0043,
      "step": 175500
    },
    {
      "epoch": 17.56,
      "grad_norm": 1.5070129848027136e-05,
      "learning_rate": 2.0802675585284283e-06,
      "loss": 0.0028,
      "step": 175600
    },
    {
      "epoch": 17.57,
      "grad_norm": 7.220142288133502e-05,
      "learning_rate": 2.0785953177257524e-06,
      "loss": 0.004,
      "step": 175700
    },
    {
      "epoch": 17.58,
      "grad_norm": 3.834291419479996e-06,
      "learning_rate": 2.0769230769230773e-06,
      "loss": 0.0,
      "step": 175800
    },
    {
      "epoch": 17.59,
      "grad_norm": 2.9713196454395074e-06,
      "learning_rate": 2.0752508361204017e-06,
      "loss": 0.0,
      "step": 175900
    },
    {
      "epoch": 17.6,
      "grad_norm": 3.5655093597597443e-06,
      "learning_rate": 2.0735785953177258e-06,
      "loss": 0.0048,
      "step": 176000
    },
    {
      "epoch": 17.61,
      "grad_norm": 2.6606401661410928e-05,
      "learning_rate": 2.0719063545150502e-06,
      "loss": 0.0051,
      "step": 176100
    },
    {
      "epoch": 17.62,
      "grad_norm": 8.27614203444682e-05,
      "learning_rate": 2.0702341137123747e-06,
      "loss": 0.0,
      "step": 176200
    },
    {
      "epoch": 17.63,
      "grad_norm": 0.0005782655789516866,
      "learning_rate": 2.068561872909699e-06,
      "loss": 0.0056,
      "step": 176300
    },
    {
      "epoch": 17.64,
      "grad_norm": 4.853478912991704e-06,
      "learning_rate": 2.0668896321070236e-06,
      "loss": 0.0,
      "step": 176400
    },
    {
      "epoch": 17.65,
      "grad_norm": 1.0566937817202415e-05,
      "learning_rate": 2.065217391304348e-06,
      "loss": 0.0,
      "step": 176500
    },
    {
      "epoch": 17.66,
      "grad_norm": 1.4728007045050617e-05,
      "learning_rate": 2.0635451505016725e-06,
      "loss": 0.0,
      "step": 176600
    },
    {
      "epoch": 17.67,
      "grad_norm": 2.444941310386639e-05,
      "learning_rate": 2.0618729096989966e-06,
      "loss": 0.004,
      "step": 176700
    },
    {
      "epoch": 17.68,
      "grad_norm": 0.0010614750208333135,
      "learning_rate": 2.0602006688963215e-06,
      "loss": 0.0,
      "step": 176800
    },
    {
      "epoch": 17.69,
      "grad_norm": 4.313383669796167e-06,
      "learning_rate": 2.058528428093646e-06,
      "loss": 0.0,
      "step": 176900
    },
    {
      "epoch": 17.7,
      "grad_norm": 3.7007937407906866e-06,
      "learning_rate": 2.05685618729097e-06,
      "loss": 0.0,
      "step": 177000
    },
    {
      "epoch": 17.71,
      "grad_norm": 7.082697720761644e-06,
      "learning_rate": 2.0551839464882944e-06,
      "loss": 0.0029,
      "step": 177100
    },
    {
      "epoch": 17.72,
      "grad_norm": 4.934473054163391e-06,
      "learning_rate": 2.053511705685619e-06,
      "loss": 0.0003,
      "step": 177200
    },
    {
      "epoch": 17.73,
      "grad_norm": 1.3510943063010927e-05,
      "learning_rate": 2.0518394648829433e-06,
      "loss": 0.0,
      "step": 177300
    },
    {
      "epoch": 17.74,
      "grad_norm": 4.9949585445574485e-06,
      "learning_rate": 2.050167224080268e-06,
      "loss": 0.0,
      "step": 177400
    },
    {
      "epoch": 17.75,
      "grad_norm": 3.1264180506695993e-06,
      "learning_rate": 2.0484949832775922e-06,
      "loss": 0.0,
      "step": 177500
    },
    {
      "epoch": 17.76,
      "grad_norm": 0.0013070253189653158,
      "learning_rate": 2.0468227424749167e-06,
      "loss": 0.0047,
      "step": 177600
    },
    {
      "epoch": 17.77,
      "grad_norm": 1.3158923138689715e-05,
      "learning_rate": 2.0451505016722407e-06,
      "loss": 0.0001,
      "step": 177700
    },
    {
      "epoch": 17.78,
      "grad_norm": 7.450151315424591e-06,
      "learning_rate": 2.0434782608695656e-06,
      "loss": 0.0,
      "step": 177800
    },
    {
      "epoch": 17.79,
      "grad_norm": 0.00015990401152521372,
      "learning_rate": 2.04180602006689e-06,
      "loss": 0.0,
      "step": 177900
    },
    {
      "epoch": 17.8,
      "grad_norm": 4.039090981677873e-06,
      "learning_rate": 2.040133779264214e-06,
      "loss": 0.0052,
      "step": 178000
    },
    {
      "epoch": 17.81,
      "grad_norm": 1.5695175534347072e-05,
      "learning_rate": 2.0384615384615386e-06,
      "loss": 0.0,
      "step": 178100
    },
    {
      "epoch": 17.82,
      "grad_norm": 0.0006352474447339773,
      "learning_rate": 2.036789297658863e-06,
      "loss": 0.0065,
      "step": 178200
    },
    {
      "epoch": 17.83,
      "grad_norm": 4.0101067497744225e-06,
      "learning_rate": 2.0351170568561875e-06,
      "loss": 0.0052,
      "step": 178300
    },
    {
      "epoch": 17.84,
      "grad_norm": 0.023364724591374397,
      "learning_rate": 2.033444816053512e-06,
      "loss": 0.0057,
      "step": 178400
    },
    {
      "epoch": 17.85,
      "grad_norm": 5.3487925470108166e-05,
      "learning_rate": 2.0317725752508364e-06,
      "loss": 0.0,
      "step": 178500
    },
    {
      "epoch": 17.86,
      "grad_norm": 1.5035124306450598e-05,
      "learning_rate": 2.030100334448161e-06,
      "loss": 0.0,
      "step": 178600
    },
    {
      "epoch": 17.87,
      "grad_norm": 2.6687663194024935e-05,
      "learning_rate": 2.028428093645485e-06,
      "loss": 0.0,
      "step": 178700
    },
    {
      "epoch": 17.88,
      "grad_norm": 1.9049177353736013e-05,
      "learning_rate": 2.0267558528428094e-06,
      "loss": 0.0002,
      "step": 178800
    },
    {
      "epoch": 17.89,
      "grad_norm": 0.0001085298354155384,
      "learning_rate": 2.025083612040134e-06,
      "loss": 0.0,
      "step": 178900
    },
    {
      "epoch": 17.9,
      "grad_norm": 2.8620918328670086e-06,
      "learning_rate": 2.0234113712374583e-06,
      "loss": 0.0022,
      "step": 179000
    },
    {
      "epoch": 17.91,
      "grad_norm": 3.738312079804018e-05,
      "learning_rate": 2.0217391304347828e-06,
      "loss": 0.0,
      "step": 179100
    },
    {
      "epoch": 17.92,
      "grad_norm": 4.336022357165348e-06,
      "learning_rate": 2.0200668896321072e-06,
      "loss": 0.0,
      "step": 179200
    },
    {
      "epoch": 17.93,
      "grad_norm": 3.867320174322231e-06,
      "learning_rate": 2.0183946488294317e-06,
      "loss": 0.0,
      "step": 179300
    },
    {
      "epoch": 17.94,
      "grad_norm": 5.321133357938379e-06,
      "learning_rate": 2.016722408026756e-06,
      "loss": 0.0,
      "step": 179400
    },
    {
      "epoch": 17.95,
      "grad_norm": 0.00027794932248070836,
      "learning_rate": 2.0150501672240806e-06,
      "loss": 0.0062,
      "step": 179500
    },
    {
      "epoch": 17.96,
      "grad_norm": 4.577206254907651e-06,
      "learning_rate": 2.0133779264214046e-06,
      "loss": 0.0,
      "step": 179600
    },
    {
      "epoch": 17.97,
      "grad_norm": 4.166466897004284e-05,
      "learning_rate": 2.011705685618729e-06,
      "loss": 0.0,
      "step": 179700
    },
    {
      "epoch": 17.98,
      "grad_norm": 6.511841547762742e-06,
      "learning_rate": 2.0100334448160536e-06,
      "loss": 0.0,
      "step": 179800
    },
    {
      "epoch": 17.99,
      "grad_norm": 8.989376510726288e-06,
      "learning_rate": 2.008361204013378e-06,
      "loss": 0.0,
      "step": 179900
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.00018079046276398003,
      "learning_rate": 2.0066889632107025e-06,
      "loss": 0.0018,
      "step": 180000
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.9858,
      "eval_f1": 0.9858,
      "eval_loss": 0.1534043401479721,
      "eval_runtime": 137.7635,
      "eval_samples_per_second": 290.353,
      "eval_steps_per_second": 290.353,
      "step": 180000
    },
    {
      "epoch": 18.01,
      "grad_norm": 0.00037978720501996577,
      "learning_rate": 2.005016722408027e-06,
      "loss": 0.0019,
      "step": 180100
    },
    {
      "epoch": 18.02,
      "grad_norm": 3.5032473988394486e-06,
      "learning_rate": 2.0033444816053514e-06,
      "loss": 0.0,
      "step": 180200
    },
    {
      "epoch": 18.03,
      "grad_norm": 2.6313730359106557e-06,
      "learning_rate": 2.0016722408026754e-06,
      "loss": 0.0,
      "step": 180300
    },
    {
      "epoch": 18.04,
      "grad_norm": 1.732081000227481e-05,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0001,
      "step": 180400
    },
    {
      "epoch": 18.05,
      "grad_norm": 7.704897143412381e-06,
      "learning_rate": 1.998327759197325e-06,
      "loss": 0.0,
      "step": 180500
    },
    {
      "epoch": 18.06,
      "grad_norm": 2.443795665385551e-06,
      "learning_rate": 1.996655518394649e-06,
      "loss": 0.0,
      "step": 180600
    },
    {
      "epoch": 18.07,
      "grad_norm": 8.326176612172276e-05,
      "learning_rate": 1.9949832775919733e-06,
      "loss": 0.0013,
      "step": 180700
    },
    {
      "epoch": 18.08,
      "grad_norm": 2.8846391160186613e-06,
      "learning_rate": 1.9933110367892978e-06,
      "loss": 0.0043,
      "step": 180800
    },
    {
      "epoch": 18.09,
      "grad_norm": 3.090189056820236e-05,
      "learning_rate": 1.9916387959866222e-06,
      "loss": 0.0,
      "step": 180900
    },
    {
      "epoch": 18.1,
      "grad_norm": 0.0011116014793515205,
      "learning_rate": 1.9899665551839467e-06,
      "loss": 0.0,
      "step": 181000
    },
    {
      "epoch": 18.11,
      "grad_norm": 1.265348146262113e-05,
      "learning_rate": 1.988294314381271e-06,
      "loss": 0.0042,
      "step": 181100
    },
    {
      "epoch": 18.12,
      "grad_norm": 3.777616257139016e-06,
      "learning_rate": 1.9866220735785956e-06,
      "loss": 0.0024,
      "step": 181200
    },
    {
      "epoch": 18.13,
      "grad_norm": 3.274929622421041e-05,
      "learning_rate": 1.9849498327759196e-06,
      "loss": 0.0,
      "step": 181300
    },
    {
      "epoch": 18.14,
      "grad_norm": 2.9154903131711762e-06,
      "learning_rate": 1.9832775919732445e-06,
      "loss": 0.0,
      "step": 181400
    },
    {
      "epoch": 18.15,
      "grad_norm": 6.4856340031838045e-06,
      "learning_rate": 1.981605351170569e-06,
      "loss": 0.0052,
      "step": 181500
    },
    {
      "epoch": 18.16,
      "grad_norm": 5.131582838657778e-06,
      "learning_rate": 1.979933110367893e-06,
      "loss": 0.0143,
      "step": 181600
    },
    {
      "epoch": 18.17,
      "grad_norm": 3.5505281630321406e-06,
      "learning_rate": 1.9782608695652175e-06,
      "loss": 0.0022,
      "step": 181700
    },
    {
      "epoch": 18.18,
      "grad_norm": 4.868303221883252e-06,
      "learning_rate": 1.976588628762542e-06,
      "loss": 0.0107,
      "step": 181800
    },
    {
      "epoch": 18.19,
      "grad_norm": 3.616964022512548e-05,
      "learning_rate": 1.9749163879598664e-06,
      "loss": 0.0,
      "step": 181900
    },
    {
      "epoch": 18.2,
      "grad_norm": 3.501211540424265e-05,
      "learning_rate": 1.973244147157191e-06,
      "loss": 0.0,
      "step": 182000
    },
    {
      "epoch": 18.21,
      "grad_norm": 4.2304021917516366e-05,
      "learning_rate": 1.9715719063545153e-06,
      "loss": 0.0,
      "step": 182100
    },
    {
      "epoch": 18.22,
      "grad_norm": 5.119220077176578e-06,
      "learning_rate": 1.9698996655518398e-06,
      "loss": 0.0,
      "step": 182200
    },
    {
      "epoch": 18.23,
      "grad_norm": 3.877084054693114e-06,
      "learning_rate": 1.968227424749164e-06,
      "loss": 0.0,
      "step": 182300
    },
    {
      "epoch": 18.24,
      "grad_norm": 4.858573811361566e-05,
      "learning_rate": 1.9665551839464887e-06,
      "loss": 0.0,
      "step": 182400
    },
    {
      "epoch": 18.25,
      "grad_norm": 0.00018243178783450276,
      "learning_rate": 1.964882943143813e-06,
      "loss": 0.0037,
      "step": 182500
    },
    {
      "epoch": 18.26,
      "grad_norm": 8.987120963865891e-06,
      "learning_rate": 1.963210702341137e-06,
      "loss": 0.0,
      "step": 182600
    },
    {
      "epoch": 18.27,
      "grad_norm": 3.947538971260656e-06,
      "learning_rate": 1.9615384615384617e-06,
      "loss": 0.0076,
      "step": 182700
    },
    {
      "epoch": 18.28,
      "grad_norm": 0.016083484515547752,
      "learning_rate": 1.959866220735786e-06,
      "loss": 0.0,
      "step": 182800
    },
    {
      "epoch": 18.29,
      "grad_norm": 2.228962148365099e-05,
      "learning_rate": 1.9581939799331106e-06,
      "loss": 0.0,
      "step": 182900
    },
    {
      "epoch": 18.3,
      "grad_norm": 4.8957945182337426e-06,
      "learning_rate": 1.956521739130435e-06,
      "loss": 0.0,
      "step": 183000
    },
    {
      "epoch": 18.31,
      "grad_norm": 3.4853383112931624e-06,
      "learning_rate": 1.9548494983277595e-06,
      "loss": 0.005,
      "step": 183100
    },
    {
      "epoch": 18.32,
      "grad_norm": 0.008418547920882702,
      "learning_rate": 1.953177257525084e-06,
      "loss": 0.0033,
      "step": 183200
    },
    {
      "epoch": 18.33,
      "grad_norm": 6.046983799024019e-06,
      "learning_rate": 1.951505016722408e-06,
      "loss": 0.0034,
      "step": 183300
    },
    {
      "epoch": 18.34,
      "grad_norm": 0.000668312597554177,
      "learning_rate": 1.9498327759197325e-06,
      "loss": 0.0,
      "step": 183400
    },
    {
      "epoch": 18.35,
      "grad_norm": 8.495872862113174e-06,
      "learning_rate": 1.948160535117057e-06,
      "loss": 0.0019,
      "step": 183500
    },
    {
      "epoch": 18.36,
      "grad_norm": 0.0002771254803519696,
      "learning_rate": 1.9464882943143814e-06,
      "loss": 0.0,
      "step": 183600
    },
    {
      "epoch": 18.37,
      "grad_norm": 5.837029220856493e-06,
      "learning_rate": 1.944816053511706e-06,
      "loss": 0.0,
      "step": 183700
    },
    {
      "epoch": 18.38,
      "grad_norm": 0.00012922013411298394,
      "learning_rate": 1.9431438127090303e-06,
      "loss": 0.0,
      "step": 183800
    },
    {
      "epoch": 18.39,
      "grad_norm": 3.7937059005344054e-06,
      "learning_rate": 1.9414715719063548e-06,
      "loss": 0.0,
      "step": 183900
    },
    {
      "epoch": 18.4,
      "grad_norm": 4.895188794762362e-06,
      "learning_rate": 1.9397993311036792e-06,
      "loss": 0.0,
      "step": 184000
    },
    {
      "epoch": 18.41,
      "grad_norm": 1.4537821698468179e-05,
      "learning_rate": 1.9381270903010037e-06,
      "loss": 0.0,
      "step": 184100
    },
    {
      "epoch": 18.42,
      "grad_norm": 4.6436987759079784e-05,
      "learning_rate": 1.9364548494983277e-06,
      "loss": 0.0,
      "step": 184200
    },
    {
      "epoch": 18.43,
      "grad_norm": 1.5509576769545674e-05,
      "learning_rate": 1.934782608695652e-06,
      "loss": 0.0,
      "step": 184300
    },
    {
      "epoch": 18.44,
      "grad_norm": 3.2652471418259665e-05,
      "learning_rate": 1.9331103678929766e-06,
      "loss": 0.0,
      "step": 184400
    },
    {
      "epoch": 18.45,
      "grad_norm": 3.94304743167595e-06,
      "learning_rate": 1.931438127090301e-06,
      "loss": 0.0,
      "step": 184500
    },
    {
      "epoch": 18.46,
      "grad_norm": 4.365731911093462e-06,
      "learning_rate": 1.9297658862876256e-06,
      "loss": 0.0051,
      "step": 184600
    },
    {
      "epoch": 18.47,
      "grad_norm": 6.958713584026555e-06,
      "learning_rate": 1.92809364548495e-06,
      "loss": 0.0,
      "step": 184700
    },
    {
      "epoch": 18.48,
      "grad_norm": 1.526621781522408e-05,
      "learning_rate": 1.9264214046822745e-06,
      "loss": 0.0,
      "step": 184800
    },
    {
      "epoch": 18.49,
      "grad_norm": 6.0657366702798754e-05,
      "learning_rate": 1.9247491638795985e-06,
      "loss": 0.0,
      "step": 184900
    },
    {
      "epoch": 18.5,
      "grad_norm": 7.313900823646691e-06,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 0.0,
      "step": 185000
    },
    {
      "epoch": 18.5,
      "eval_accuracy": 0.986975,
      "eval_f1": 0.986975,
      "eval_loss": 0.1347718983888626,
      "eval_runtime": 139.3214,
      "eval_samples_per_second": 287.106,
      "eval_steps_per_second": 287.106,
      "step": 185000
    },
    {
      "epoch": 18.51,
      "grad_norm": 8.065755537245423e-06,
      "learning_rate": 1.921404682274248e-06,
      "loss": 0.0,
      "step": 185100
    },
    {
      "epoch": 18.52,
      "grad_norm": 3.502688514345209e-06,
      "learning_rate": 1.919732441471572e-06,
      "loss": 0.0,
      "step": 185200
    },
    {
      "epoch": 18.53,
      "grad_norm": 3.190935558450292e-06,
      "learning_rate": 1.9180602006688964e-06,
      "loss": 0.0,
      "step": 185300
    },
    {
      "epoch": 18.54,
      "grad_norm": 1.8046868717647158e-05,
      "learning_rate": 1.916387959866221e-06,
      "loss": 0.0,
      "step": 185400
    },
    {
      "epoch": 18.55,
      "grad_norm": 4.194449957140023e-06,
      "learning_rate": 1.9147157190635453e-06,
      "loss": 0.0,
      "step": 185500
    },
    {
      "epoch": 18.56,
      "grad_norm": 2.685572098926059e-06,
      "learning_rate": 1.9130434782608697e-06,
      "loss": 0.0,
      "step": 185600
    },
    {
      "epoch": 18.57,
      "grad_norm": 7.828857633285224e-05,
      "learning_rate": 1.911371237458194e-06,
      "loss": 0.0,
      "step": 185700
    },
    {
      "epoch": 18.58,
      "grad_norm": 5.445906936074607e-06,
      "learning_rate": 1.9096989966555187e-06,
      "loss": 0.0046,
      "step": 185800
    },
    {
      "epoch": 18.59,
      "grad_norm": 2.773958840407431e-06,
      "learning_rate": 1.9080267558528427e-06,
      "loss": 0.0002,
      "step": 185900
    },
    {
      "epoch": 18.6,
      "grad_norm": 3.370199465280166e-06,
      "learning_rate": 1.9063545150501676e-06,
      "loss": 0.0,
      "step": 186000
    },
    {
      "epoch": 18.61,
      "grad_norm": 2.8800272957596462e-06,
      "learning_rate": 1.9046822742474918e-06,
      "loss": 0.0,
      "step": 186100
    },
    {
      "epoch": 18.62,
      "grad_norm": 4.435024038684787e-06,
      "learning_rate": 1.9030100334448163e-06,
      "loss": 0.0,
      "step": 186200
    },
    {
      "epoch": 18.63,
      "grad_norm": 9.851543836703058e-06,
      "learning_rate": 1.9013377926421405e-06,
      "loss": 0.0,
      "step": 186300
    },
    {
      "epoch": 18.64,
      "grad_norm": 4.3996242311550304e-06,
      "learning_rate": 1.899665551839465e-06,
      "loss": 0.0,
      "step": 186400
    },
    {
      "epoch": 18.65,
      "grad_norm": 3.0937371775507927e-06,
      "learning_rate": 1.8979933110367897e-06,
      "loss": 0.0,
      "step": 186500
    },
    {
      "epoch": 18.66,
      "grad_norm": 3.958779871027218e-06,
      "learning_rate": 1.896321070234114e-06,
      "loss": 0.0,
      "step": 186600
    },
    {
      "epoch": 18.67,
      "grad_norm": 0.00148894009180367,
      "learning_rate": 1.8946488294314384e-06,
      "loss": 0.0035,
      "step": 186700
    },
    {
      "epoch": 18.68,
      "grad_norm": 1.5799363609403372e-05,
      "learning_rate": 1.8929765886287626e-06,
      "loss": 0.0,
      "step": 186800
    },
    {
      "epoch": 18.69,
      "grad_norm": 5.034713922213996e-06,
      "learning_rate": 1.891304347826087e-06,
      "loss": 0.0,
      "step": 186900
    },
    {
      "epoch": 18.7,
      "grad_norm": 2.59518128586933e-06,
      "learning_rate": 1.8896321070234118e-06,
      "loss": 0.0,
      "step": 187000
    },
    {
      "epoch": 18.71,
      "grad_norm": 3.2641801226418465e-06,
      "learning_rate": 1.887959866220736e-06,
      "loss": 0.0,
      "step": 187100
    },
    {
      "epoch": 18.72,
      "grad_norm": 9.082512406166643e-06,
      "learning_rate": 1.8862876254180605e-06,
      "loss": 0.0,
      "step": 187200
    },
    {
      "epoch": 18.73,
      "grad_norm": 2.4530313567083795e-06,
      "learning_rate": 1.8846153846153847e-06,
      "loss": 0.0,
      "step": 187300
    },
    {
      "epoch": 18.74,
      "grad_norm": 3.7139268442842877e-06,
      "learning_rate": 1.8829431438127092e-06,
      "loss": 0.0,
      "step": 187400
    },
    {
      "epoch": 18.75,
      "grad_norm": 0.00024536409182474017,
      "learning_rate": 1.8812709030100336e-06,
      "loss": 0.0,
      "step": 187500
    },
    {
      "epoch": 18.76,
      "grad_norm": 0.0001586981670698151,
      "learning_rate": 1.879598662207358e-06,
      "loss": 0.0,
      "step": 187600
    },
    {
      "epoch": 18.77,
      "grad_norm": 3.170421041431837e-05,
      "learning_rate": 1.8779264214046826e-06,
      "loss": 0.0,
      "step": 187700
    },
    {
      "epoch": 18.78,
      "grad_norm": 3.257045136706438e-06,
      "learning_rate": 1.8762541806020068e-06,
      "loss": 0.003,
      "step": 187800
    },
    {
      "epoch": 18.79,
      "grad_norm": 2.242783239125856e-06,
      "learning_rate": 1.8745819397993313e-06,
      "loss": 0.0,
      "step": 187900
    },
    {
      "epoch": 18.8,
      "grad_norm": 1.1425007869547699e-05,
      "learning_rate": 1.8729096989966555e-06,
      "loss": 0.01,
      "step": 188000
    },
    {
      "epoch": 18.81,
      "grad_norm": 2.598640776341199e-06,
      "learning_rate": 1.8712374581939802e-06,
      "loss": 0.0,
      "step": 188100
    },
    {
      "epoch": 18.82,
      "grad_norm": 0.0007181576220318675,
      "learning_rate": 1.8695652173913044e-06,
      "loss": 0.0,
      "step": 188200
    },
    {
      "epoch": 18.83,
      "grad_norm": 5.5871128097351175e-06,
      "learning_rate": 1.867892976588629e-06,
      "loss": 0.0,
      "step": 188300
    },
    {
      "epoch": 18.84,
      "grad_norm": 2.6241129944537533e-06,
      "learning_rate": 1.8662207357859531e-06,
      "loss": 0.0055,
      "step": 188400
    },
    {
      "epoch": 18.85,
      "grad_norm": 0.0003280396922491491,
      "learning_rate": 1.8645484949832776e-06,
      "loss": 0.0058,
      "step": 188500
    },
    {
      "epoch": 18.86,
      "grad_norm": 6.639120692852885e-05,
      "learning_rate": 1.8628762541806023e-06,
      "loss": 0.0053,
      "step": 188600
    },
    {
      "epoch": 18.87,
      "grad_norm": 3.4430977393640205e-05,
      "learning_rate": 1.8612040133779265e-06,
      "loss": 0.0,
      "step": 188700
    },
    {
      "epoch": 18.88,
      "grad_norm": 2.713867070269771e-05,
      "learning_rate": 1.859531772575251e-06,
      "loss": 0.0037,
      "step": 188800
    },
    {
      "epoch": 18.89,
      "grad_norm": 0.0003621370124164969,
      "learning_rate": 1.8578595317725752e-06,
      "loss": 0.0,
      "step": 188900
    },
    {
      "epoch": 18.9,
      "grad_norm": 2.7829876216856064e-06,
      "learning_rate": 1.8561872909698997e-06,
      "loss": 0.0004,
      "step": 189000
    },
    {
      "epoch": 18.91,
      "grad_norm": 9.860410500550643e-05,
      "learning_rate": 1.8545150501672244e-06,
      "loss": 0.0043,
      "step": 189100
    },
    {
      "epoch": 18.92,
      "grad_norm": 2.7196410883334465e-06,
      "learning_rate": 1.8528428093645486e-06,
      "loss": 0.0,
      "step": 189200
    },
    {
      "epoch": 18.93,
      "grad_norm": 2.777722556857043e-06,
      "learning_rate": 1.851170568561873e-06,
      "loss": 0.0,
      "step": 189300
    },
    {
      "epoch": 18.94,
      "grad_norm": 3.0415455967158778e-06,
      "learning_rate": 1.8494983277591973e-06,
      "loss": 0.0,
      "step": 189400
    },
    {
      "epoch": 18.95,
      "grad_norm": 7.348385224759113e-06,
      "learning_rate": 1.8478260869565218e-06,
      "loss": 0.0,
      "step": 189500
    },
    {
      "epoch": 18.96,
      "grad_norm": 4.2450628825463355e-06,
      "learning_rate": 1.8461538461538465e-06,
      "loss": 0.0065,
      "step": 189600
    },
    {
      "epoch": 18.97,
      "grad_norm": 1.0027314601757098e-05,
      "learning_rate": 1.8444816053511707e-06,
      "loss": 0.0004,
      "step": 189700
    },
    {
      "epoch": 18.98,
      "grad_norm": 4.824078132514842e-06,
      "learning_rate": 1.8428093645484952e-06,
      "loss": 0.0,
      "step": 189800
    },
    {
      "epoch": 18.99,
      "grad_norm": 4.105546395294368e-06,
      "learning_rate": 1.8411371237458194e-06,
      "loss": 0.0,
      "step": 189900
    },
    {
      "epoch": 19.0,
      "grad_norm": 3.5762561765295686e-06,
      "learning_rate": 1.8394648829431439e-06,
      "loss": 0.0054,
      "step": 190000
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.9856,
      "eval_f1": 0.9856,
      "eval_loss": 0.15510258078575134,
      "eval_runtime": 137.3776,
      "eval_samples_per_second": 291.168,
      "eval_steps_per_second": 291.168,
      "step": 190000
    },
    {
      "epoch": 19.01,
      "grad_norm": 7.700408787059132e-06,
      "learning_rate": 1.8377926421404686e-06,
      "loss": 0.0001,
      "step": 190100
    },
    {
      "epoch": 19.02,
      "grad_norm": 6.728196694893995e-06,
      "learning_rate": 1.8361204013377928e-06,
      "loss": 0.0,
      "step": 190200
    },
    {
      "epoch": 19.03,
      "grad_norm": 1.0402168300061021e-05,
      "learning_rate": 1.8344481605351173e-06,
      "loss": 0.0,
      "step": 190300
    },
    {
      "epoch": 19.04,
      "grad_norm": 8.382570376852527e-05,
      "learning_rate": 1.8327759197324415e-06,
      "loss": 0.0,
      "step": 190400
    },
    {
      "epoch": 19.05,
      "grad_norm": 2.071788185276091e-05,
      "learning_rate": 1.831103678929766e-06,
      "loss": 0.0,
      "step": 190500
    },
    {
      "epoch": 19.06,
      "grad_norm": 6.8766489675908815e-06,
      "learning_rate": 1.8294314381270906e-06,
      "loss": 0.004,
      "step": 190600
    },
    {
      "epoch": 19.07,
      "grad_norm": 0.0001236250827787444,
      "learning_rate": 1.8277591973244149e-06,
      "loss": 0.0,
      "step": 190700
    },
    {
      "epoch": 19.08,
      "grad_norm": 5.6216026678157505e-06,
      "learning_rate": 1.8260869565217394e-06,
      "loss": 0.0056,
      "step": 190800
    },
    {
      "epoch": 19.09,
      "grad_norm": 2.4372426196350716e-05,
      "learning_rate": 1.8244147157190636e-06,
      "loss": 0.0,
      "step": 190900
    },
    {
      "epoch": 19.1,
      "grad_norm": 0.0024588583037257195,
      "learning_rate": 1.822742474916388e-06,
      "loss": 0.0,
      "step": 191000
    },
    {
      "epoch": 19.11,
      "grad_norm": 2.7965081699221628e-06,
      "learning_rate": 1.8210702341137127e-06,
      "loss": 0.0,
      "step": 191100
    },
    {
      "epoch": 19.12,
      "grad_norm": 1.210660320793977e-05,
      "learning_rate": 1.819397993311037e-06,
      "loss": 0.0,
      "step": 191200
    },
    {
      "epoch": 19.13,
      "grad_norm": 4.8296110435330775e-06,
      "learning_rate": 1.8177257525083614e-06,
      "loss": 0.0081,
      "step": 191300
    },
    {
      "epoch": 19.14,
      "grad_norm": 7.996006024768576e-05,
      "learning_rate": 1.8160535117056857e-06,
      "loss": 0.0,
      "step": 191400
    },
    {
      "epoch": 19.15,
      "grad_norm": 0.00013632285117637366,
      "learning_rate": 1.8143812709030102e-06,
      "loss": 0.0,
      "step": 191500
    },
    {
      "epoch": 19.16,
      "grad_norm": 4.816226282855496e-05,
      "learning_rate": 1.8127090301003348e-06,
      "loss": 0.0,
      "step": 191600
    },
    {
      "epoch": 19.17,
      "grad_norm": 0.00023680346203036606,
      "learning_rate": 1.811036789297659e-06,
      "loss": 0.0015,
      "step": 191700
    },
    {
      "epoch": 19.18,
      "grad_norm": 1.9035480363527313e-05,
      "learning_rate": 1.8093645484949835e-06,
      "loss": 0.0,
      "step": 191800
    },
    {
      "epoch": 19.19,
      "grad_norm": 5.586427050729981e-06,
      "learning_rate": 1.8076923076923078e-06,
      "loss": 0.0093,
      "step": 191900
    },
    {
      "epoch": 19.2,
      "grad_norm": 1.2779854841937777e-05,
      "learning_rate": 1.8060200668896322e-06,
      "loss": 0.0082,
      "step": 192000
    },
    {
      "epoch": 19.21,
      "grad_norm": 3.19543346449791e-06,
      "learning_rate": 1.8043478260869567e-06,
      "loss": 0.0,
      "step": 192100
    },
    {
      "epoch": 19.22,
      "grad_norm": 6.000143457640661e-06,
      "learning_rate": 1.8026755852842812e-06,
      "loss": 0.0035,
      "step": 192200
    },
    {
      "epoch": 19.23,
      "grad_norm": 4.63162396044936e-05,
      "learning_rate": 1.8010033444816056e-06,
      "loss": 0.0,
      "step": 192300
    },
    {
      "epoch": 19.24,
      "grad_norm": 0.004847492091357708,
      "learning_rate": 1.7993311036789299e-06,
      "loss": 0.0,
      "step": 192400
    },
    {
      "epoch": 19.25,
      "grad_norm": 1.812317714211531e-05,
      "learning_rate": 1.7976588628762543e-06,
      "loss": 0.0,
      "step": 192500
    },
    {
      "epoch": 19.26,
      "grad_norm": 6.381682851497317e-06,
      "learning_rate": 1.7959866220735788e-06,
      "loss": 0.0,
      "step": 192600
    },
    {
      "epoch": 19.27,
      "grad_norm": 6.096350261941552e-06,
      "learning_rate": 1.7943143812709033e-06,
      "loss": 0.0,
      "step": 192700
    },
    {
      "epoch": 19.28,
      "grad_norm": 6.9785724008397665e-06,
      "learning_rate": 1.7926421404682275e-06,
      "loss": 0.0,
      "step": 192800
    },
    {
      "epoch": 19.29,
      "grad_norm": 0.00010105594992637634,
      "learning_rate": 1.790969899665552e-06,
      "loss": 0.0,
      "step": 192900
    },
    {
      "epoch": 19.3,
      "grad_norm": 2.523460580050596e-06,
      "learning_rate": 1.7892976588628764e-06,
      "loss": 0.0,
      "step": 193000
    },
    {
      "epoch": 19.31,
      "grad_norm": 6.7169371504860464e-06,
      "learning_rate": 1.7876254180602007e-06,
      "loss": 0.0,
      "step": 193100
    },
    {
      "epoch": 19.32,
      "grad_norm": 0.00013431630213744938,
      "learning_rate": 1.7859531772575253e-06,
      "loss": 0.0,
      "step": 193200
    },
    {
      "epoch": 19.33,
      "grad_norm": 2.817968152157846e-06,
      "learning_rate": 1.7842809364548496e-06,
      "loss": 0.0,
      "step": 193300
    },
    {
      "epoch": 19.34,
      "grad_norm": 5.19740569870919e-06,
      "learning_rate": 1.782608695652174e-06,
      "loss": 0.0006,
      "step": 193400
    },
    {
      "epoch": 19.35,
      "grad_norm": 0.014350646175444126,
      "learning_rate": 1.7809364548494983e-06,
      "loss": 0.0015,
      "step": 193500
    },
    {
      "epoch": 19.36,
      "grad_norm": 7.5378752626420464e-06,
      "learning_rate": 1.7792642140468228e-06,
      "loss": 0.0,
      "step": 193600
    },
    {
      "epoch": 19.37,
      "grad_norm": 2.4976093300210778e-06,
      "learning_rate": 1.7775919732441474e-06,
      "loss": 0.0001,
      "step": 193700
    },
    {
      "epoch": 19.38,
      "grad_norm": 3.9604841731488705e-05,
      "learning_rate": 1.7759197324414717e-06,
      "loss": 0.0,
      "step": 193800
    },
    {
      "epoch": 19.39,
      "grad_norm": 1.6136280464706942e-05,
      "learning_rate": 1.7742474916387961e-06,
      "loss": 0.0,
      "step": 193900
    },
    {
      "epoch": 19.4,
      "grad_norm": 2.822498345267377e-06,
      "learning_rate": 1.7725752508361204e-06,
      "loss": 0.0001,
      "step": 194000
    },
    {
      "epoch": 19.41,
      "grad_norm": 4.150161657889839e-06,
      "learning_rate": 1.7709030100334449e-06,
      "loss": 0.0042,
      "step": 194100
    },
    {
      "epoch": 19.42,
      "grad_norm": 1.879728188214358e-05,
      "learning_rate": 1.7692307692307695e-06,
      "loss": 0.0,
      "step": 194200
    },
    {
      "epoch": 19.43,
      "grad_norm": 8.796157089818735e-06,
      "learning_rate": 1.7675585284280938e-06,
      "loss": 0.0,
      "step": 194300
    },
    {
      "epoch": 19.44,
      "grad_norm": 4.645894023269648e-06,
      "learning_rate": 1.7658862876254182e-06,
      "loss": 0.0,
      "step": 194400
    },
    {
      "epoch": 19.45,
      "grad_norm": 9.539145139569882e-06,
      "learning_rate": 1.7642140468227425e-06,
      "loss": 0.0,
      "step": 194500
    },
    {
      "epoch": 19.46,
      "grad_norm": 2.760291636150214e-06,
      "learning_rate": 1.762541806020067e-06,
      "loss": 0.0,
      "step": 194600
    },
    {
      "epoch": 19.47,
      "grad_norm": 7.266039574460592e-06,
      "learning_rate": 1.7608695652173916e-06,
      "loss": 0.0,
      "step": 194700
    },
    {
      "epoch": 19.48,
      "grad_norm": 2.4798812319204444e-06,
      "learning_rate": 1.7591973244147159e-06,
      "loss": 0.0029,
      "step": 194800
    },
    {
      "epoch": 19.49,
      "grad_norm": 2.3052723918226548e-05,
      "learning_rate": 1.7575250836120403e-06,
      "loss": 0.0,
      "step": 194900
    },
    {
      "epoch": 19.5,
      "grad_norm": 5.209206574363634e-05,
      "learning_rate": 1.7558528428093646e-06,
      "loss": 0.0,
      "step": 195000
    },
    {
      "epoch": 19.5,
      "eval_accuracy": 0.987475,
      "eval_f1": 0.987475,
      "eval_loss": 0.1356872022151947,
      "eval_runtime": 135.0112,
      "eval_samples_per_second": 296.272,
      "eval_steps_per_second": 296.272,
      "step": 195000
    },
    {
      "epoch": 19.51,
      "grad_norm": 8.632729986857157e-06,
      "learning_rate": 1.754180602006689e-06,
      "loss": 0.0,
      "step": 195100
    },
    {
      "epoch": 19.52,
      "grad_norm": 5.0513413043518085e-06,
      "learning_rate": 1.7525083612040137e-06,
      "loss": 0.0,
      "step": 195200
    },
    {
      "epoch": 19.53,
      "grad_norm": 3.816026492131641e-06,
      "learning_rate": 1.750836120401338e-06,
      "loss": 0.0,
      "step": 195300
    },
    {
      "epoch": 19.54,
      "grad_norm": 1.888971496555314e-06,
      "learning_rate": 1.7491638795986624e-06,
      "loss": 0.0,
      "step": 195400
    },
    {
      "epoch": 19.55,
      "grad_norm": 0.0008715542498975992,
      "learning_rate": 1.7474916387959867e-06,
      "loss": 0.0,
      "step": 195500
    },
    {
      "epoch": 19.56,
      "grad_norm": 2.565695467637852e-05,
      "learning_rate": 1.7458193979933111e-06,
      "loss": 0.0,
      "step": 195600
    },
    {
      "epoch": 19.57,
      "grad_norm": 7.099295089574298e-06,
      "learning_rate": 1.7441471571906358e-06,
      "loss": 0.0039,
      "step": 195700
    },
    {
      "epoch": 19.58,
      "grad_norm": 3.0402254651562544e-06,
      "learning_rate": 1.74247491638796e-06,
      "loss": 0.0,
      "step": 195800
    },
    {
      "epoch": 19.59,
      "grad_norm": 3.428632680879673e-06,
      "learning_rate": 1.7408026755852845e-06,
      "loss": 0.0,
      "step": 195900
    },
    {
      "epoch": 19.6,
      "grad_norm": 7.307850410143146e-06,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 0.0,
      "step": 196000
    },
    {
      "epoch": 19.61,
      "grad_norm": 2.104581108142156e-05,
      "learning_rate": 1.7374581939799332e-06,
      "loss": 0.0005,
      "step": 196100
    },
    {
      "epoch": 19.62,
      "grad_norm": 7.845720574550796e-06,
      "learning_rate": 1.7357859531772579e-06,
      "loss": 0.0,
      "step": 196200
    },
    {
      "epoch": 19.63,
      "grad_norm": 2.6197665192739805e-06,
      "learning_rate": 1.7341137123745821e-06,
      "loss": 0.0054,
      "step": 196300
    },
    {
      "epoch": 19.64,
      "grad_norm": 0.00823716539889574,
      "learning_rate": 1.7324414715719066e-06,
      "loss": 0.0043,
      "step": 196400
    },
    {
      "epoch": 19.65,
      "grad_norm": 5.974598479951965e-06,
      "learning_rate": 1.7307692307692308e-06,
      "loss": 0.006,
      "step": 196500
    },
    {
      "epoch": 19.66,
      "grad_norm": 6.843105802545324e-05,
      "learning_rate": 1.7290969899665553e-06,
      "loss": 0.0,
      "step": 196600
    },
    {
      "epoch": 19.67,
      "grad_norm": 1.3177969776734244e-05,
      "learning_rate": 1.7274247491638798e-06,
      "loss": 0.0,
      "step": 196700
    },
    {
      "epoch": 19.68,
      "grad_norm": 1.1815620382549241e-05,
      "learning_rate": 1.7257525083612042e-06,
      "loss": 0.0034,
      "step": 196800
    },
    {
      "epoch": 19.69,
      "grad_norm": 1.2062504538334906e-05,
      "learning_rate": 1.7240802675585287e-06,
      "loss": 0.0,
      "step": 196900
    },
    {
      "epoch": 19.7,
      "grad_norm": 1.7363923689117655e-05,
      "learning_rate": 1.722408026755853e-06,
      "loss": 0.0,
      "step": 197000
    },
    {
      "epoch": 19.71,
      "grad_norm": 0.003212525974959135,
      "learning_rate": 1.7207357859531774e-06,
      "loss": 0.0,
      "step": 197100
    },
    {
      "epoch": 19.72,
      "grad_norm": 2.8709598609566456e-06,
      "learning_rate": 1.7190635451505019e-06,
      "loss": 0.0,
      "step": 197200
    },
    {
      "epoch": 19.73,
      "grad_norm": 3.688925971800927e-06,
      "learning_rate": 1.7173913043478263e-06,
      "loss": 0.0,
      "step": 197300
    },
    {
      "epoch": 19.74,
      "grad_norm": 4.4789187086280435e-05,
      "learning_rate": 1.7157190635451506e-06,
      "loss": 0.0,
      "step": 197400
    },
    {
      "epoch": 19.75,
      "grad_norm": 0.010275106877088547,
      "learning_rate": 1.714046822742475e-06,
      "loss": 0.0,
      "step": 197500
    },
    {
      "epoch": 19.76,
      "grad_norm": 3.637941290435265e-06,
      "learning_rate": 1.7123745819397995e-06,
      "loss": 0.0,
      "step": 197600
    },
    {
      "epoch": 19.77,
      "grad_norm": 8.29253167466959e-06,
      "learning_rate": 1.710702341137124e-06,
      "loss": 0.0018,
      "step": 197700
    },
    {
      "epoch": 19.78,
      "grad_norm": 3.868249223160092e-06,
      "learning_rate": 1.7090301003344484e-06,
      "loss": 0.0,
      "step": 197800
    },
    {
      "epoch": 19.79,
      "grad_norm": 3.6814990380662493e-06,
      "learning_rate": 1.7073578595317727e-06,
      "loss": 0.0,
      "step": 197900
    },
    {
      "epoch": 19.8,
      "grad_norm": 1.3625873179989867e-05,
      "learning_rate": 1.7056856187290971e-06,
      "loss": 0.0,
      "step": 198000
    },
    {
      "epoch": 19.81,
      "grad_norm": 0.0058494736440479755,
      "learning_rate": 1.7040133779264214e-06,
      "loss": 0.0063,
      "step": 198100
    },
    {
      "epoch": 19.82,
      "grad_norm": 3.438921112319804e-06,
      "learning_rate": 1.7023411371237458e-06,
      "loss": 0.0,
      "step": 198200
    },
    {
      "epoch": 19.83,
      "grad_norm": 0.0007142440881580114,
      "learning_rate": 1.7006688963210705e-06,
      "loss": 0.0,
      "step": 198300
    },
    {
      "epoch": 19.84,
      "grad_norm": 1.3060711353318766e-05,
      "learning_rate": 1.6989966555183947e-06,
      "loss": 0.0,
      "step": 198400
    },
    {
      "epoch": 19.85,
      "grad_norm": 0.007419825531542301,
      "learning_rate": 1.6973244147157192e-06,
      "loss": 0.0015,
      "step": 198500
    },
    {
      "epoch": 19.86,
      "grad_norm": 2.1179796476644697e-06,
      "learning_rate": 1.6956521739130435e-06,
      "loss": 0.0,
      "step": 198600
    },
    {
      "epoch": 19.87,
      "grad_norm": 4.210245151625713e-06,
      "learning_rate": 1.693979933110368e-06,
      "loss": 0.0,
      "step": 198700
    },
    {
      "epoch": 19.88,
      "grad_norm": 2.1146457584109157e-05,
      "learning_rate": 1.6923076923076926e-06,
      "loss": 0.0,
      "step": 198800
    },
    {
      "epoch": 19.89,
      "grad_norm": 2.6942605018120958e-06,
      "learning_rate": 1.6906354515050168e-06,
      "loss": 0.0041,
      "step": 198900
    },
    {
      "epoch": 19.9,
      "grad_norm": 2.950105226773303e-06,
      "learning_rate": 1.6889632107023413e-06,
      "loss": 0.0,
      "step": 199000
    },
    {
      "epoch": 19.91,
      "grad_norm": 3.4126630907849176e-06,
      "learning_rate": 1.6872909698996655e-06,
      "loss": 0.0,
      "step": 199100
    },
    {
      "epoch": 19.92,
      "grad_norm": 3.211943521819194e-06,
      "learning_rate": 1.68561872909699e-06,
      "loss": 0.0,
      "step": 199200
    },
    {
      "epoch": 19.93,
      "grad_norm": 4.190772961010225e-05,
      "learning_rate": 1.6839464882943147e-06,
      "loss": 0.0,
      "step": 199300
    },
    {
      "epoch": 19.94,
      "grad_norm": 0.0004158316587563604,
      "learning_rate": 1.682274247491639e-06,
      "loss": 0.0,
      "step": 199400
    },
    {
      "epoch": 19.95,
      "grad_norm": 3.3130665997305186e-06,
      "learning_rate": 1.6806020066889634e-06,
      "loss": 0.0,
      "step": 199500
    },
    {
      "epoch": 19.96,
      "grad_norm": 3.0582334602513583e-06,
      "learning_rate": 1.6789297658862876e-06,
      "loss": 0.0,
      "step": 199600
    },
    {
      "epoch": 19.97,
      "grad_norm": 3.249564770158031e-06,
      "learning_rate": 1.677257525083612e-06,
      "loss": 0.0,
      "step": 199700
    },
    {
      "epoch": 19.98,
      "grad_norm": 2.1675789412256563e-06,
      "learning_rate": 1.6755852842809368e-06,
      "loss": 0.0,
      "step": 199800
    },
    {
      "epoch": 19.99,
      "grad_norm": 3.1598647183272988e-06,
      "learning_rate": 1.673913043478261e-06,
      "loss": 0.0,
      "step": 199900
    },
    {
      "epoch": 20.0,
      "grad_norm": 2.6797370082931593e-06,
      "learning_rate": 1.6722408026755855e-06,
      "loss": 0.0,
      "step": 200000
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.98775,
      "eval_f1": 0.98775,
      "eval_loss": 0.13826440274715424,
      "eval_runtime": 133.9725,
      "eval_samples_per_second": 298.569,
      "eval_steps_per_second": 298.569,
      "step": 200000
    },
    {
      "epoch": 20.01,
      "grad_norm": 4.444530532055069e-06,
      "learning_rate": 1.6705685618729097e-06,
      "loss": 0.0,
      "step": 200100
    },
    {
      "epoch": 20.02,
      "grad_norm": 2.957669721581624e-06,
      "learning_rate": 1.6688963210702342e-06,
      "loss": 0.0,
      "step": 200200
    },
    {
      "epoch": 20.03,
      "grad_norm": 3.291775101388339e-06,
      "learning_rate": 1.6672240802675589e-06,
      "loss": 0.0,
      "step": 200300
    },
    {
      "epoch": 20.04,
      "grad_norm": 2.782702722470276e-06,
      "learning_rate": 1.6655518394648831e-06,
      "loss": 0.0031,
      "step": 200400
    },
    {
      "epoch": 20.05,
      "grad_norm": 8.329800039064139e-06,
      "learning_rate": 1.6638795986622076e-06,
      "loss": 0.0031,
      "step": 200500
    },
    {
      "epoch": 20.06,
      "grad_norm": 3.788985168284853e-06,
      "learning_rate": 1.6622073578595318e-06,
      "loss": 0.0,
      "step": 200600
    },
    {
      "epoch": 20.07,
      "grad_norm": 3.6758290207217215e-06,
      "learning_rate": 1.6605351170568563e-06,
      "loss": 0.0,
      "step": 200700
    },
    {
      "epoch": 20.08,
      "grad_norm": 3.0887517823430244e-06,
      "learning_rate": 1.658862876254181e-06,
      "loss": 0.0,
      "step": 200800
    },
    {
      "epoch": 20.09,
      "grad_norm": 6.320357351796702e-05,
      "learning_rate": 1.6571906354515052e-06,
      "loss": 0.0,
      "step": 200900
    },
    {
      "epoch": 20.1,
      "grad_norm": 2.65393919107737e-06,
      "learning_rate": 1.6555183946488297e-06,
      "loss": 0.0,
      "step": 201000
    },
    {
      "epoch": 20.11,
      "grad_norm": 3.1102713364816736e-06,
      "learning_rate": 1.653846153846154e-06,
      "loss": 0.0,
      "step": 201100
    },
    {
      "epoch": 20.12,
      "grad_norm": 2.6238412829115987e-06,
      "learning_rate": 1.6521739130434784e-06,
      "loss": 0.0112,
      "step": 201200
    },
    {
      "epoch": 20.13,
      "grad_norm": 2.09651898330776e-06,
      "learning_rate": 1.6505016722408028e-06,
      "loss": 0.0,
      "step": 201300
    },
    {
      "epoch": 20.14,
      "grad_norm": 2.435858732496854e-06,
      "learning_rate": 1.6488294314381273e-06,
      "loss": 0.0,
      "step": 201400
    },
    {
      "epoch": 20.15,
      "grad_norm": 4.056108537042746e-06,
      "learning_rate": 1.6471571906354518e-06,
      "loss": 0.0063,
      "step": 201500
    },
    {
      "epoch": 20.16,
      "grad_norm": 4.716017429018393e-06,
      "learning_rate": 1.645484949832776e-06,
      "loss": 0.0003,
      "step": 201600
    },
    {
      "epoch": 20.17,
      "grad_norm": 4.022814209747594e-06,
      "learning_rate": 1.6438127090301005e-06,
      "loss": 0.0,
      "step": 201700
    },
    {
      "epoch": 20.18,
      "grad_norm": 1.1780533895944245e-05,
      "learning_rate": 1.642140468227425e-06,
      "loss": 0.011,
      "step": 201800
    },
    {
      "epoch": 20.19,
      "grad_norm": 3.5756809211306972e-06,
      "learning_rate": 1.6404682274247494e-06,
      "loss": 0.0023,
      "step": 201900
    },
    {
      "epoch": 20.2,
      "grad_norm": 3.5133691653754795e-06,
      "learning_rate": 1.6387959866220736e-06,
      "loss": 0.0,
      "step": 202000
    },
    {
      "epoch": 20.21,
      "grad_norm": 2.1817815650138073e-05,
      "learning_rate": 1.637123745819398e-06,
      "loss": 0.0,
      "step": 202100
    },
    {
      "epoch": 20.22,
      "grad_norm": 7.410139369312674e-05,
      "learning_rate": 1.6354515050167226e-06,
      "loss": 0.0059,
      "step": 202200
    },
    {
      "epoch": 20.23,
      "grad_norm": 1.0194399692409206e-05,
      "learning_rate": 1.633779264214047e-06,
      "loss": 0.0,
      "step": 202300
    },
    {
      "epoch": 20.24,
      "grad_norm": 2.9090826956235105e-06,
      "learning_rate": 1.6321070234113715e-06,
      "loss": 0.0,
      "step": 202400
    },
    {
      "epoch": 20.25,
      "grad_norm": 1.4217706848285161e-05,
      "learning_rate": 1.6304347826086957e-06,
      "loss": 0.0,
      "step": 202500
    },
    {
      "epoch": 20.26,
      "grad_norm": 4.9936047616938595e-06,
      "learning_rate": 1.6287625418060202e-06,
      "loss": 0.0,
      "step": 202600
    },
    {
      "epoch": 20.27,
      "grad_norm": 2.9128027563274372e-06,
      "learning_rate": 1.6270903010033444e-06,
      "loss": 0.0,
      "step": 202700
    },
    {
      "epoch": 20.28,
      "grad_norm": 4.234791049384512e-06,
      "learning_rate": 1.625418060200669e-06,
      "loss": 0.0076,
      "step": 202800
    },
    {
      "epoch": 20.29,
      "grad_norm": 2.635060582178994e-06,
      "learning_rate": 1.6237458193979936e-06,
      "loss": 0.0,
      "step": 202900
    },
    {
      "epoch": 20.3,
      "grad_norm": 2.543606751714833e-05,
      "learning_rate": 1.6220735785953178e-06,
      "loss": 0.0,
      "step": 203000
    },
    {
      "epoch": 20.31,
      "grad_norm": 1.7358966942992993e-05,
      "learning_rate": 1.6204013377926423e-06,
      "loss": 0.0002,
      "step": 203100
    },
    {
      "epoch": 20.32,
      "grad_norm": 0.0010591618483886123,
      "learning_rate": 1.6187290969899665e-06,
      "loss": 0.0,
      "step": 203200
    },
    {
      "epoch": 20.33,
      "grad_norm": 7.357494723692071e-06,
      "learning_rate": 1.617056856187291e-06,
      "loss": 0.0,
      "step": 203300
    },
    {
      "epoch": 20.34,
      "grad_norm": 2.521198803151492e-05,
      "learning_rate": 1.6153846153846157e-06,
      "loss": 0.0036,
      "step": 203400
    },
    {
      "epoch": 20.35,
      "grad_norm": 4.66443998448085e-06,
      "learning_rate": 1.61371237458194e-06,
      "loss": 0.0,
      "step": 203500
    },
    {
      "epoch": 20.36,
      "grad_norm": 0.00010037265747087076,
      "learning_rate": 1.6120401337792644e-06,
      "loss": 0.0019,
      "step": 203600
    },
    {
      "epoch": 20.37,
      "grad_norm": 0.0002850138407666236,
      "learning_rate": 1.6103678929765886e-06,
      "loss": 0.0,
      "step": 203700
    },
    {
      "epoch": 20.38,
      "grad_norm": 6.3181232690112665e-06,
      "learning_rate": 1.608695652173913e-06,
      "loss": 0.0,
      "step": 203800
    },
    {
      "epoch": 20.39,
      "grad_norm": 9.268571375287138e-06,
      "learning_rate": 1.6070234113712377e-06,
      "loss": 0.0,
      "step": 203900
    },
    {
      "epoch": 20.4,
      "grad_norm": 3.4747040444926824e-06,
      "learning_rate": 1.605351170568562e-06,
      "loss": 0.0035,
      "step": 204000
    },
    {
      "epoch": 20.41,
      "grad_norm": 2.8883071081509115e-06,
      "learning_rate": 1.6036789297658865e-06,
      "loss": 0.0,
      "step": 204100
    },
    {
      "epoch": 20.42,
      "grad_norm": 6.735133865731768e-06,
      "learning_rate": 1.6020066889632107e-06,
      "loss": 0.0038,
      "step": 204200
    },
    {
      "epoch": 20.43,
      "grad_norm": 2.2144862668938003e-05,
      "learning_rate": 1.6003344481605352e-06,
      "loss": 0.0001,
      "step": 204300
    },
    {
      "epoch": 20.44,
      "grad_norm": 3.1439285521628335e-06,
      "learning_rate": 1.5986622073578598e-06,
      "loss": 0.001,
      "step": 204400
    },
    {
      "epoch": 20.45,
      "grad_norm": 5.965619129710831e-05,
      "learning_rate": 1.596989966555184e-06,
      "loss": 0.0018,
      "step": 204500
    },
    {
      "epoch": 20.46,
      "grad_norm": 2.0531138943624683e-05,
      "learning_rate": 1.5953177257525085e-06,
      "loss": 0.0,
      "step": 204600
    },
    {
      "epoch": 20.47,
      "grad_norm": 1.9690560293383896e-06,
      "learning_rate": 1.5936454849498328e-06,
      "loss": 0.0,
      "step": 204700
    },
    {
      "epoch": 20.48,
      "grad_norm": 1.1894359886355232e-05,
      "learning_rate": 1.5919732441471573e-06,
      "loss": 0.0008,
      "step": 204800
    },
    {
      "epoch": 20.49,
      "grad_norm": 3.1937240692059277e-06,
      "learning_rate": 1.590301003344482e-06,
      "loss": 0.0,
      "step": 204900
    },
    {
      "epoch": 20.5,
      "grad_norm": 3.63977778761182e-05,
      "learning_rate": 1.5886287625418062e-06,
      "loss": 0.0,
      "step": 205000
    },
    {
      "epoch": 20.5,
      "eval_accuracy": 0.987575,
      "eval_f1": 0.987575,
      "eval_loss": 0.13013528287410736,
      "eval_runtime": 133.1301,
      "eval_samples_per_second": 300.458,
      "eval_steps_per_second": 300.458,
      "step": 205000
    },
    {
      "epoch": 20.51,
      "grad_norm": 0.0012186990352347493,
      "learning_rate": 1.5869565217391306e-06,
      "loss": 0.0,
      "step": 205100
    },
    {
      "epoch": 20.52,
      "grad_norm": 3.058609763684217e-06,
      "learning_rate": 1.5852842809364549e-06,
      "loss": 0.0,
      "step": 205200
    },
    {
      "epoch": 20.53,
      "grad_norm": 4.367856035969453e-06,
      "learning_rate": 1.5836120401337793e-06,
      "loss": 0.005,
      "step": 205300
    },
    {
      "epoch": 20.54,
      "grad_norm": 7.178983196354238e-06,
      "learning_rate": 1.581939799331104e-06,
      "loss": 0.0063,
      "step": 205400
    },
    {
      "epoch": 20.55,
      "grad_norm": 2.945566848211456e-06,
      "learning_rate": 1.5802675585284283e-06,
      "loss": 0.0,
      "step": 205500
    },
    {
      "epoch": 20.56,
      "grad_norm": 5.289690761856036e-06,
      "learning_rate": 1.5785953177257527e-06,
      "loss": 0.0,
      "step": 205600
    },
    {
      "epoch": 20.57,
      "grad_norm": 5.60561784368474e-06,
      "learning_rate": 1.576923076923077e-06,
      "loss": 0.0,
      "step": 205700
    },
    {
      "epoch": 20.58,
      "grad_norm": 2.395588580839103e-06,
      "learning_rate": 1.5752508361204014e-06,
      "loss": 0.0,
      "step": 205800
    },
    {
      "epoch": 20.59,
      "grad_norm": 4.692345100920647e-06,
      "learning_rate": 1.573578595317726e-06,
      "loss": 0.0032,
      "step": 205900
    },
    {
      "epoch": 20.6,
      "grad_norm": 5.715948645956814e-06,
      "learning_rate": 1.5719063545150504e-06,
      "loss": 0.0,
      "step": 206000
    },
    {
      "epoch": 20.61,
      "grad_norm": 3.817430297203828e-06,
      "learning_rate": 1.5702341137123748e-06,
      "loss": 0.0,
      "step": 206100
    },
    {
      "epoch": 20.62,
      "grad_norm": 1.3704256161872763e-05,
      "learning_rate": 1.568561872909699e-06,
      "loss": 0.0002,
      "step": 206200
    },
    {
      "epoch": 20.63,
      "grad_norm": 2.3260201487573795e-05,
      "learning_rate": 1.5668896321070235e-06,
      "loss": 0.0,
      "step": 206300
    },
    {
      "epoch": 20.64,
      "grad_norm": 3.091832013524254e-06,
      "learning_rate": 1.565217391304348e-06,
      "loss": 0.0,
      "step": 206400
    },
    {
      "epoch": 20.65,
      "grad_norm": 5.030944521422498e-05,
      "learning_rate": 1.5635451505016724e-06,
      "loss": 0.0087,
      "step": 206500
    },
    {
      "epoch": 20.66,
      "grad_norm": 4.412408088683151e-05,
      "learning_rate": 1.5618729096989967e-06,
      "loss": 0.0,
      "step": 206600
    },
    {
      "epoch": 20.67,
      "grad_norm": 6.452126854128437e-06,
      "learning_rate": 1.5602006688963212e-06,
      "loss": 0.0,
      "step": 206700
    },
    {
      "epoch": 20.68,
      "grad_norm": 0.0005374864558689296,
      "learning_rate": 1.5585284280936456e-06,
      "loss": 0.0046,
      "step": 206800
    },
    {
      "epoch": 20.69,
      "grad_norm": 8.672196418046951e-05,
      "learning_rate": 1.55685618729097e-06,
      "loss": 0.0,
      "step": 206900
    },
    {
      "epoch": 20.7,
      "grad_norm": 3.4972249522979837e-06,
      "learning_rate": 1.5551839464882945e-06,
      "loss": 0.0,
      "step": 207000
    },
    {
      "epoch": 20.71,
      "grad_norm": 6.360653515002923e-06,
      "learning_rate": 1.5535117056856188e-06,
      "loss": 0.0,
      "step": 207100
    },
    {
      "epoch": 20.72,
      "grad_norm": 9.319689525000285e-06,
      "learning_rate": 1.5518394648829432e-06,
      "loss": 0.0,
      "step": 207200
    },
    {
      "epoch": 20.73,
      "grad_norm": 3.967582870245678e-06,
      "learning_rate": 1.5501672240802675e-06,
      "loss": 0.0,
      "step": 207300
    },
    {
      "epoch": 20.74,
      "grad_norm": 5.464647074404638e-06,
      "learning_rate": 1.5484949832775922e-06,
      "loss": 0.0,
      "step": 207400
    },
    {
      "epoch": 20.75,
      "grad_norm": 1.3912532267568167e-05,
      "learning_rate": 1.5468227424749166e-06,
      "loss": 0.0,
      "step": 207500
    },
    {
      "epoch": 20.76,
      "grad_norm": 1.1850606824737042e-05,
      "learning_rate": 1.5451505016722409e-06,
      "loss": 0.0,
      "step": 207600
    },
    {
      "epoch": 20.77,
      "grad_norm": 4.910432380711427e-06,
      "learning_rate": 1.5434782608695653e-06,
      "loss": 0.0,
      "step": 207700
    },
    {
      "epoch": 20.78,
      "grad_norm": 3.082827106482e-06,
      "learning_rate": 1.5418060200668896e-06,
      "loss": 0.0,
      "step": 207800
    },
    {
      "epoch": 20.79,
      "grad_norm": 5.340507868822897e-06,
      "learning_rate": 1.540133779264214e-06,
      "loss": 0.0,
      "step": 207900
    },
    {
      "epoch": 20.8,
      "grad_norm": 2.699596734601073e-06,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.0,
      "step": 208000
    },
    {
      "epoch": 20.81,
      "grad_norm": 7.638249371666461e-06,
      "learning_rate": 1.536789297658863e-06,
      "loss": 0.0105,
      "step": 208100
    },
    {
      "epoch": 20.82,
      "grad_norm": 2.4105136617436074e-05,
      "learning_rate": 1.5351170568561874e-06,
      "loss": 0.0017,
      "step": 208200
    },
    {
      "epoch": 20.83,
      "grad_norm": 7.558439392596483e-06,
      "learning_rate": 1.5334448160535117e-06,
      "loss": 0.0,
      "step": 208300
    },
    {
      "epoch": 20.84,
      "grad_norm": 5.754342055297457e-06,
      "learning_rate": 1.5317725752508361e-06,
      "loss": 0.0,
      "step": 208400
    },
    {
      "epoch": 20.85,
      "grad_norm": 4.127535430598073e-05,
      "learning_rate": 1.5301003344481608e-06,
      "loss": 0.0,
      "step": 208500
    },
    {
      "epoch": 20.86,
      "grad_norm": 2.812850652844645e-06,
      "learning_rate": 1.528428093645485e-06,
      "loss": 0.0003,
      "step": 208600
    },
    {
      "epoch": 20.87,
      "grad_norm": 5.573003363679163e-06,
      "learning_rate": 1.5267558528428095e-06,
      "loss": 0.0,
      "step": 208700
    },
    {
      "epoch": 20.88,
      "grad_norm": 9.813478754949756e-06,
      "learning_rate": 1.5250836120401338e-06,
      "loss": 0.0,
      "step": 208800
    },
    {
      "epoch": 20.89,
      "grad_norm": 1.7837548512034118e-05,
      "learning_rate": 1.5234113712374582e-06,
      "loss": 0.0,
      "step": 208900
    },
    {
      "epoch": 20.9,
      "grad_norm": 2.806435077218339e-06,
      "learning_rate": 1.521739130434783e-06,
      "loss": 0.0,
      "step": 209000
    },
    {
      "epoch": 20.91,
      "grad_norm": 2.905662768171169e-06,
      "learning_rate": 1.5200668896321071e-06,
      "loss": 0.0,
      "step": 209100
    },
    {
      "epoch": 20.92,
      "grad_norm": 0.005608703941106796,
      "learning_rate": 1.5183946488294316e-06,
      "loss": 0.0057,
      "step": 209200
    },
    {
      "epoch": 20.93,
      "grad_norm": 4.323443590692477e-06,
      "learning_rate": 1.5167224080267559e-06,
      "loss": 0.0,
      "step": 209300
    },
    {
      "epoch": 20.94,
      "grad_norm": 2.9272045139805414e-05,
      "learning_rate": 1.5150501672240803e-06,
      "loss": 0.0033,
      "step": 209400
    },
    {
      "epoch": 20.95,
      "grad_norm": 1.3083412341075018e-05,
      "learning_rate": 1.513377926421405e-06,
      "loss": 0.0058,
      "step": 209500
    },
    {
      "epoch": 20.96,
      "grad_norm": 5.5376938689732924e-05,
      "learning_rate": 1.5117056856187292e-06,
      "loss": 0.0068,
      "step": 209600
    },
    {
      "epoch": 20.97,
      "grad_norm": 1.1804182577179745e-05,
      "learning_rate": 1.5100334448160537e-06,
      "loss": 0.0,
      "step": 209700
    },
    {
      "epoch": 20.98,
      "grad_norm": 0.0001499229110777378,
      "learning_rate": 1.508361204013378e-06,
      "loss": 0.0,
      "step": 209800
    },
    {
      "epoch": 20.99,
      "grad_norm": 3.3819417240010807e-06,
      "learning_rate": 1.5066889632107024e-06,
      "loss": 0.0,
      "step": 209900
    },
    {
      "epoch": 21.0,
      "grad_norm": 0.0023802791256457567,
      "learning_rate": 1.505016722408027e-06,
      "loss": 0.0,
      "step": 210000
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.987625,
      "eval_f1": 0.987625,
      "eval_loss": 0.12249647080898285,
      "eval_runtime": 135.0345,
      "eval_samples_per_second": 296.221,
      "eval_steps_per_second": 296.221,
      "step": 210000
    },
    {
      "epoch": 21.01,
      "grad_norm": 6.9270877247618046e-06,
      "learning_rate": 1.5033444816053513e-06,
      "loss": 0.0,
      "step": 210100
    },
    {
      "epoch": 21.02,
      "grad_norm": 3.770094917854294e-05,
      "learning_rate": 1.5016722408026758e-06,
      "loss": 0.0,
      "step": 210200
    },
    {
      "epoch": 21.03,
      "grad_norm": 1.953655737452209e-05,
      "learning_rate": 1.5e-06,
      "loss": 0.0,
      "step": 210300
    },
    {
      "epoch": 21.04,
      "grad_norm": 0.0002708936808630824,
      "learning_rate": 1.4983277591973245e-06,
      "loss": 0.0,
      "step": 210400
    },
    {
      "epoch": 21.05,
      "grad_norm": 6.020537512085866e-06,
      "learning_rate": 1.4966555183946492e-06,
      "loss": 0.0,
      "step": 210500
    },
    {
      "epoch": 21.06,
      "grad_norm": 0.03410932794213295,
      "learning_rate": 1.4949832775919734e-06,
      "loss": 0.0,
      "step": 210600
    },
    {
      "epoch": 21.07,
      "grad_norm": 1.6557878552703187e-05,
      "learning_rate": 1.4933110367892979e-06,
      "loss": 0.0,
      "step": 210700
    },
    {
      "epoch": 21.08,
      "grad_norm": 3.707931682583876e-06,
      "learning_rate": 1.4916387959866221e-06,
      "loss": 0.0,
      "step": 210800
    },
    {
      "epoch": 21.09,
      "grad_norm": 4.064024324179627e-06,
      "learning_rate": 1.4899665551839466e-06,
      "loss": 0.0024,
      "step": 210900
    },
    {
      "epoch": 21.1,
      "grad_norm": 3.917386493412778e-06,
      "learning_rate": 1.488294314381271e-06,
      "loss": 0.0,
      "step": 211000
    },
    {
      "epoch": 21.11,
      "grad_norm": 5.051691459811991e-06,
      "learning_rate": 1.4866220735785955e-06,
      "loss": 0.0,
      "step": 211100
    },
    {
      "epoch": 21.12,
      "grad_norm": 1.3631856745632831e-05,
      "learning_rate": 1.4849498327759198e-06,
      "loss": 0.0,
      "step": 211200
    },
    {
      "epoch": 21.13,
      "grad_norm": 4.376207925815834e-06,
      "learning_rate": 1.4832775919732442e-06,
      "loss": 0.0,
      "step": 211300
    },
    {
      "epoch": 21.14,
      "grad_norm": 0.0007653874927200377,
      "learning_rate": 1.4816053511705687e-06,
      "loss": 0.0,
      "step": 211400
    },
    {
      "epoch": 21.15,
      "grad_norm": 0.0003364417643751949,
      "learning_rate": 1.4799331103678931e-06,
      "loss": 0.0055,
      "step": 211500
    },
    {
      "epoch": 21.16,
      "grad_norm": 3.0988155685918173e-06,
      "learning_rate": 1.4782608695652176e-06,
      "loss": 0.0015,
      "step": 211600
    },
    {
      "epoch": 21.17,
      "grad_norm": 3.886037575284718e-06,
      "learning_rate": 1.4765886287625418e-06,
      "loss": 0.0,
      "step": 211700
    },
    {
      "epoch": 21.18,
      "grad_norm": 9.195034181175288e-06,
      "learning_rate": 1.4749163879598663e-06,
      "loss": 0.0041,
      "step": 211800
    },
    {
      "epoch": 21.19,
      "grad_norm": 5.210130439081695e-06,
      "learning_rate": 1.4732441471571906e-06,
      "loss": 0.0027,
      "step": 211900
    },
    {
      "epoch": 21.2,
      "grad_norm": 8.08697677712189e-06,
      "learning_rate": 1.4715719063545152e-06,
      "loss": 0.0,
      "step": 212000
    },
    {
      "epoch": 21.21,
      "grad_norm": 3.7482986954273656e-06,
      "learning_rate": 1.4698996655518397e-06,
      "loss": 0.0001,
      "step": 212100
    },
    {
      "epoch": 21.22,
      "grad_norm": 1.588550367159769e-05,
      "learning_rate": 1.468227424749164e-06,
      "loss": 0.0,
      "step": 212200
    },
    {
      "epoch": 21.23,
      "grad_norm": 0.00480658421292901,
      "learning_rate": 1.4665551839464884e-06,
      "loss": 0.0,
      "step": 212300
    },
    {
      "epoch": 21.24,
      "grad_norm": 7.852279850339983e-06,
      "learning_rate": 1.4648829431438126e-06,
      "loss": 0.0,
      "step": 212400
    },
    {
      "epoch": 21.25,
      "grad_norm": 6.503975600935519e-05,
      "learning_rate": 1.4632107023411373e-06,
      "loss": 0.0047,
      "step": 212500
    },
    {
      "epoch": 21.26,
      "grad_norm": 3.9408669181284495e-06,
      "learning_rate": 1.4615384615384618e-06,
      "loss": 0.0,
      "step": 212600
    },
    {
      "epoch": 21.27,
      "grad_norm": 0.00014495612413156778,
      "learning_rate": 1.459866220735786e-06,
      "loss": 0.0,
      "step": 212700
    },
    {
      "epoch": 21.28,
      "grad_norm": 2.9439470381475985e-06,
      "learning_rate": 1.4581939799331105e-06,
      "loss": 0.0,
      "step": 212800
    },
    {
      "epoch": 21.29,
      "grad_norm": 0.00031986075919121504,
      "learning_rate": 1.4565217391304347e-06,
      "loss": 0.0,
      "step": 212900
    },
    {
      "epoch": 21.3,
      "grad_norm": 0.00012544303899630904,
      "learning_rate": 1.4548494983277592e-06,
      "loss": 0.0079,
      "step": 213000
    },
    {
      "epoch": 21.31,
      "grad_norm": 9.7873898994294e-06,
      "learning_rate": 1.4531772575250839e-06,
      "loss": 0.0,
      "step": 213100
    },
    {
      "epoch": 21.32,
      "grad_norm": 0.029412349686026573,
      "learning_rate": 1.4515050167224081e-06,
      "loss": 0.0,
      "step": 213200
    },
    {
      "epoch": 21.33,
      "grad_norm": 6.39492500340566e-05,
      "learning_rate": 1.4498327759197326e-06,
      "loss": 0.0,
      "step": 213300
    },
    {
      "epoch": 21.34,
      "grad_norm": 7.848659151932225e-06,
      "learning_rate": 1.4481605351170568e-06,
      "loss": 0.0,
      "step": 213400
    },
    {
      "epoch": 21.35,
      "grad_norm": 1.3351653251447715e-05,
      "learning_rate": 1.4464882943143813e-06,
      "loss": 0.0,
      "step": 213500
    },
    {
      "epoch": 21.36,
      "grad_norm": 0.0017086967127397656,
      "learning_rate": 1.444816053511706e-06,
      "loss": 0.0002,
      "step": 213600
    },
    {
      "epoch": 21.37,
      "grad_norm": 0.004947548732161522,
      "learning_rate": 1.4431438127090302e-06,
      "loss": 0.0,
      "step": 213700
    },
    {
      "epoch": 21.38,
      "grad_norm": 2.3903007786429953e-06,
      "learning_rate": 1.4414715719063547e-06,
      "loss": 0.0,
      "step": 213800
    },
    {
      "epoch": 21.39,
      "grad_norm": 0.00041224126471206546,
      "learning_rate": 1.439799331103679e-06,
      "loss": 0.0,
      "step": 213900
    },
    {
      "epoch": 21.4,
      "grad_norm": 0.0007366500794887543,
      "learning_rate": 1.4381270903010034e-06,
      "loss": 0.0028,
      "step": 214000
    },
    {
      "epoch": 21.41,
      "grad_norm": 2.2902284399606287e-05,
      "learning_rate": 1.436454849498328e-06,
      "loss": 0.0,
      "step": 214100
    },
    {
      "epoch": 21.42,
      "grad_norm": 1.2737849829136394e-05,
      "learning_rate": 1.4347826086956523e-06,
      "loss": 0.0039,
      "step": 214200
    },
    {
      "epoch": 21.43,
      "grad_norm": 0.001080359797924757,
      "learning_rate": 1.4331103678929768e-06,
      "loss": 0.0,
      "step": 214300
    },
    {
      "epoch": 21.44,
      "grad_norm": 4.485631507122889e-06,
      "learning_rate": 1.431438127090301e-06,
      "loss": 0.0,
      "step": 214400
    },
    {
      "epoch": 21.45,
      "grad_norm": 0.0007343352190218866,
      "learning_rate": 1.4297658862876255e-06,
      "loss": 0.0,
      "step": 214500
    },
    {
      "epoch": 21.46,
      "grad_norm": 9.769354073796421e-05,
      "learning_rate": 1.4280936454849501e-06,
      "loss": 0.0,
      "step": 214600
    },
    {
      "epoch": 21.47,
      "grad_norm": 0.001233136747032404,
      "learning_rate": 1.4264214046822744e-06,
      "loss": 0.0,
      "step": 214700
    },
    {
      "epoch": 21.48,
      "grad_norm": 4.587949206324993e-06,
      "learning_rate": 1.4247491638795989e-06,
      "loss": 0.0,
      "step": 214800
    },
    {
      "epoch": 21.49,
      "grad_norm": 3.4589143069752026e-06,
      "learning_rate": 1.423076923076923e-06,
      "loss": 0.0,
      "step": 214900
    },
    {
      "epoch": 21.5,
      "grad_norm": 6.62579077470582e-06,
      "learning_rate": 1.4214046822742476e-06,
      "loss": 0.0,
      "step": 215000
    },
    {
      "epoch": 21.5,
      "eval_accuracy": 0.987975,
      "eval_f1": 0.987975,
      "eval_loss": 0.12572382390499115,
      "eval_runtime": 133.9498,
      "eval_samples_per_second": 298.619,
      "eval_steps_per_second": 298.619,
      "step": 215000
    },
    {
      "epoch": 21.51,
      "grad_norm": 3.987763193435967e-06,
      "learning_rate": 1.4197324414715722e-06,
      "loss": 0.0,
      "step": 215100
    },
    {
      "epoch": 21.52,
      "grad_norm": 0.012753859162330627,
      "learning_rate": 1.4180602006688965e-06,
      "loss": 0.0,
      "step": 215200
    },
    {
      "epoch": 21.53,
      "grad_norm": 1.578287992742844e-05,
      "learning_rate": 1.416387959866221e-06,
      "loss": 0.0,
      "step": 215300
    },
    {
      "epoch": 21.54,
      "grad_norm": 2.9020106921961997e-06,
      "learning_rate": 1.4147157190635452e-06,
      "loss": 0.0,
      "step": 215400
    },
    {
      "epoch": 21.55,
      "grad_norm": 7.922858458186965e-06,
      "learning_rate": 1.4130434782608697e-06,
      "loss": 0.0,
      "step": 215500
    },
    {
      "epoch": 21.56,
      "grad_norm": 0.020123714581131935,
      "learning_rate": 1.4113712374581941e-06,
      "loss": 0.0023,
      "step": 215600
    },
    {
      "epoch": 21.57,
      "grad_norm": 6.435708201024681e-05,
      "learning_rate": 1.4096989966555186e-06,
      "loss": 0.0,
      "step": 215700
    },
    {
      "epoch": 21.58,
      "grad_norm": 3.7706938655901467e-06,
      "learning_rate": 1.408026755852843e-06,
      "loss": 0.0055,
      "step": 215800
    },
    {
      "epoch": 21.59,
      "grad_norm": 3.829626621154603e-06,
      "learning_rate": 1.4063545150501673e-06,
      "loss": 0.0,
      "step": 215900
    },
    {
      "epoch": 21.6,
      "grad_norm": 2.566429202488507e-06,
      "learning_rate": 1.4046822742474917e-06,
      "loss": 0.0,
      "step": 216000
    },
    {
      "epoch": 21.61,
      "grad_norm": 8.226523277699016e-06,
      "learning_rate": 1.4030100334448162e-06,
      "loss": 0.0,
      "step": 216100
    },
    {
      "epoch": 21.62,
      "grad_norm": 3.956181444664253e-06,
      "learning_rate": 1.4013377926421407e-06,
      "loss": 0.0,
      "step": 216200
    },
    {
      "epoch": 21.63,
      "grad_norm": 4.045846708322642e-06,
      "learning_rate": 1.399665551839465e-06,
      "loss": 0.0004,
      "step": 216300
    },
    {
      "epoch": 21.64,
      "grad_norm": 1.9764947865041904e-05,
      "learning_rate": 1.3979933110367894e-06,
      "loss": 0.0,
      "step": 216400
    },
    {
      "epoch": 21.65,
      "grad_norm": 2.6922571123577654e-06,
      "learning_rate": 1.3963210702341138e-06,
      "loss": 0.0,
      "step": 216500
    },
    {
      "epoch": 21.66,
      "grad_norm": 4.089896265213611e-06,
      "learning_rate": 1.3946488294314383e-06,
      "loss": 0.0,
      "step": 216600
    },
    {
      "epoch": 21.67,
      "grad_norm": 8.798033377388492e-05,
      "learning_rate": 1.3929765886287628e-06,
      "loss": 0.0,
      "step": 216700
    },
    {
      "epoch": 21.68,
      "grad_norm": 0.0005297737661749125,
      "learning_rate": 1.391304347826087e-06,
      "loss": 0.0,
      "step": 216800
    },
    {
      "epoch": 21.69,
      "grad_norm": 5.265828804112971e-06,
      "learning_rate": 1.3896321070234115e-06,
      "loss": 0.0,
      "step": 216900
    },
    {
      "epoch": 21.7,
      "grad_norm": 3.205472694389755e-06,
      "learning_rate": 1.3879598662207357e-06,
      "loss": 0.0028,
      "step": 217000
    },
    {
      "epoch": 21.71,
      "grad_norm": 3.2379621188738383e-06,
      "learning_rate": 1.3862876254180604e-06,
      "loss": 0.0,
      "step": 217100
    },
    {
      "epoch": 21.72,
      "grad_norm": 3.78774734599574e-06,
      "learning_rate": 1.3846153846153848e-06,
      "loss": 0.0025,
      "step": 217200
    },
    {
      "epoch": 21.73,
      "grad_norm": 4.6226919039327186e-06,
      "learning_rate": 1.382943143812709e-06,
      "loss": 0.0,
      "step": 217300
    },
    {
      "epoch": 21.74,
      "grad_norm": 4.582865221891552e-05,
      "learning_rate": 1.3812709030100336e-06,
      "loss": 0.0,
      "step": 217400
    },
    {
      "epoch": 21.75,
      "grad_norm": 3.6365586311148945e-06,
      "learning_rate": 1.3795986622073578e-06,
      "loss": 0.0,
      "step": 217500
    },
    {
      "epoch": 21.76,
      "grad_norm": 3.7290149066393496e-06,
      "learning_rate": 1.3779264214046825e-06,
      "loss": 0.0,
      "step": 217600
    },
    {
      "epoch": 21.77,
      "grad_norm": 3.5055440093856305e-05,
      "learning_rate": 1.376254180602007e-06,
      "loss": 0.0,
      "step": 217700
    },
    {
      "epoch": 21.78,
      "grad_norm": 0.0001472236035624519,
      "learning_rate": 1.3745819397993312e-06,
      "loss": 0.0,
      "step": 217800
    },
    {
      "epoch": 21.79,
      "grad_norm": 2.937935096269939e-05,
      "learning_rate": 1.3729096989966556e-06,
      "loss": 0.0,
      "step": 217900
    },
    {
      "epoch": 21.8,
      "grad_norm": 1.0437177479616366e-05,
      "learning_rate": 1.37123745819398e-06,
      "loss": 0.0,
      "step": 218000
    },
    {
      "epoch": 21.81,
      "grad_norm": 2.2088892364990897e-05,
      "learning_rate": 1.3695652173913044e-06,
      "loss": 0.0,
      "step": 218100
    },
    {
      "epoch": 21.82,
      "grad_norm": 3.875371112371795e-05,
      "learning_rate": 1.367892976588629e-06,
      "loss": 0.0,
      "step": 218200
    },
    {
      "epoch": 21.83,
      "grad_norm": 4.937228368362412e-06,
      "learning_rate": 1.3662207357859533e-06,
      "loss": 0.0,
      "step": 218300
    },
    {
      "epoch": 21.84,
      "grad_norm": 0.0001001003329292871,
      "learning_rate": 1.3645484949832777e-06,
      "loss": 0.0,
      "step": 218400
    },
    {
      "epoch": 21.85,
      "grad_norm": 1.2318441804382019e-05,
      "learning_rate": 1.362876254180602e-06,
      "loss": 0.0,
      "step": 218500
    },
    {
      "epoch": 21.86,
      "grad_norm": 2.1146149720152607e-06,
      "learning_rate": 1.3612040133779264e-06,
      "loss": 0.0,
      "step": 218600
    },
    {
      "epoch": 21.87,
      "grad_norm": 7.362048199865967e-05,
      "learning_rate": 1.3595317725752511e-06,
      "loss": 0.0,
      "step": 218700
    },
    {
      "epoch": 21.88,
      "grad_norm": 2.4402959297731286e-06,
      "learning_rate": 1.3578595317725754e-06,
      "loss": 0.0,
      "step": 218800
    },
    {
      "epoch": 21.89,
      "grad_norm": 2.172135054934188e-06,
      "learning_rate": 1.3561872909698998e-06,
      "loss": 0.0,
      "step": 218900
    },
    {
      "epoch": 21.9,
      "grad_norm": 7.556840955658117e-06,
      "learning_rate": 1.354515050167224e-06,
      "loss": 0.0039,
      "step": 219000
    },
    {
      "epoch": 21.91,
      "grad_norm": 2.4002990812732605e-06,
      "learning_rate": 1.3528428093645485e-06,
      "loss": 0.0,
      "step": 219100
    },
    {
      "epoch": 21.92,
      "grad_norm": 7.363054464804009e-05,
      "learning_rate": 1.3511705685618732e-06,
      "loss": 0.0,
      "step": 219200
    },
    {
      "epoch": 21.93,
      "grad_norm": 1.6851773807502468e-06,
      "learning_rate": 1.3494983277591975e-06,
      "loss": 0.0,
      "step": 219300
    },
    {
      "epoch": 21.94,
      "grad_norm": 4.769390216097236e-05,
      "learning_rate": 1.347826086956522e-06,
      "loss": 0.0,
      "step": 219400
    },
    {
      "epoch": 21.95,
      "grad_norm": 2.6172328944085166e-06,
      "learning_rate": 1.3461538461538462e-06,
      "loss": 0.0,
      "step": 219500
    },
    {
      "epoch": 21.96,
      "grad_norm": 2.637080115164281e-06,
      "learning_rate": 1.3444816053511706e-06,
      "loss": 0.0,
      "step": 219600
    },
    {
      "epoch": 21.97,
      "grad_norm": 2.824373495968757e-06,
      "learning_rate": 1.3428093645484953e-06,
      "loss": 0.0,
      "step": 219700
    },
    {
      "epoch": 21.98,
      "grad_norm": 3.1197271255223313e-06,
      "learning_rate": 1.3411371237458195e-06,
      "loss": 0.0,
      "step": 219800
    },
    {
      "epoch": 21.99,
      "grad_norm": 3.6116550745646236e-06,
      "learning_rate": 1.339464882943144e-06,
      "loss": 0.0,
      "step": 219900
    },
    {
      "epoch": 22.0,
      "grad_norm": 2.685248318812228e-06,
      "learning_rate": 1.3377926421404683e-06,
      "loss": 0.0,
      "step": 220000
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.987325,
      "eval_f1": 0.987325,
      "eval_loss": 0.1481519639492035,
      "eval_runtime": 135.5655,
      "eval_samples_per_second": 295.06,
      "eval_steps_per_second": 295.06,
      "step": 220000
    },
    {
      "epoch": 22.01,
      "grad_norm": 6.039373602106934e-06,
      "learning_rate": 1.3361204013377927e-06,
      "loss": 0.0,
      "step": 220100
    },
    {
      "epoch": 22.02,
      "grad_norm": 4.358920341474004e-05,
      "learning_rate": 1.3344481605351172e-06,
      "loss": 0.0,
      "step": 220200
    },
    {
      "epoch": 22.03,
      "grad_norm": 2.7751359539252007e-06,
      "learning_rate": 1.3327759197324416e-06,
      "loss": 0.0,
      "step": 220300
    },
    {
      "epoch": 22.04,
      "grad_norm": 2.0891520762233995e-06,
      "learning_rate": 1.331103678929766e-06,
      "loss": 0.0,
      "step": 220400
    },
    {
      "epoch": 22.05,
      "grad_norm": 2.5228462163795484e-06,
      "learning_rate": 1.3294314381270903e-06,
      "loss": 0.0,
      "step": 220500
    },
    {
      "epoch": 22.06,
      "grad_norm": 0.00011139993148390204,
      "learning_rate": 1.3277591973244148e-06,
      "loss": 0.0053,
      "step": 220600
    },
    {
      "epoch": 22.07,
      "grad_norm": 6.763996680092532e-06,
      "learning_rate": 1.3260869565217393e-06,
      "loss": 0.0,
      "step": 220700
    },
    {
      "epoch": 22.08,
      "grad_norm": 0.0006400056299753487,
      "learning_rate": 1.3244147157190637e-06,
      "loss": 0.0,
      "step": 220800
    },
    {
      "epoch": 22.09,
      "grad_norm": 7.192987595772138e-06,
      "learning_rate": 1.322742474916388e-06,
      "loss": 0.0,
      "step": 220900
    },
    {
      "epoch": 22.1,
      "grad_norm": 1.966576974155032e-06,
      "learning_rate": 1.3210702341137124e-06,
      "loss": 0.0,
      "step": 221000
    },
    {
      "epoch": 22.11,
      "grad_norm": 8.61862918100087e-06,
      "learning_rate": 1.319397993311037e-06,
      "loss": 0.0,
      "step": 221100
    },
    {
      "epoch": 22.12,
      "grad_norm": 2.2725034796167165e-06,
      "learning_rate": 1.3177257525083614e-06,
      "loss": 0.0,
      "step": 221200
    },
    {
      "epoch": 22.13,
      "grad_norm": 0.0003781169361900538,
      "learning_rate": 1.3160535117056858e-06,
      "loss": 0.0,
      "step": 221300
    },
    {
      "epoch": 22.14,
      "grad_norm": 1.1108208127552643e-05,
      "learning_rate": 1.31438127090301e-06,
      "loss": 0.0,
      "step": 221400
    },
    {
      "epoch": 22.15,
      "grad_norm": 2.4818687052174937e-06,
      "learning_rate": 1.3127090301003345e-06,
      "loss": 0.0,
      "step": 221500
    },
    {
      "epoch": 22.16,
      "grad_norm": 2.578741487013758e-06,
      "learning_rate": 1.3110367892976588e-06,
      "loss": 0.0,
      "step": 221600
    },
    {
      "epoch": 22.17,
      "grad_norm": 1.6207087583097746e-06,
      "learning_rate": 1.3093645484949834e-06,
      "loss": 0.0,
      "step": 221700
    },
    {
      "epoch": 22.18,
      "grad_norm": 5.453099674923578e-06,
      "learning_rate": 1.307692307692308e-06,
      "loss": 0.0006,
      "step": 221800
    },
    {
      "epoch": 22.19,
      "grad_norm": 2.2827971406513825e-06,
      "learning_rate": 1.3060200668896322e-06,
      "loss": 0.0036,
      "step": 221900
    },
    {
      "epoch": 22.2,
      "grad_norm": 2.2465158053819323e-06,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 0.0,
      "step": 222000
    },
    {
      "epoch": 22.21,
      "grad_norm": 2.7297505766910035e-06,
      "learning_rate": 1.3026755852842809e-06,
      "loss": 0.0,
      "step": 222100
    },
    {
      "epoch": 22.22,
      "grad_norm": 0.006246623583137989,
      "learning_rate": 1.3010033444816055e-06,
      "loss": 0.0,
      "step": 222200
    },
    {
      "epoch": 22.23,
      "grad_norm": 2.5728577384143136e-06,
      "learning_rate": 1.29933110367893e-06,
      "loss": 0.0,
      "step": 222300
    },
    {
      "epoch": 22.24,
      "grad_norm": 2.3373438580165384e-06,
      "learning_rate": 1.2976588628762542e-06,
      "loss": 0.0038,
      "step": 222400
    },
    {
      "epoch": 22.25,
      "grad_norm": 2.6464913389645517e-06,
      "learning_rate": 1.2959866220735787e-06,
      "loss": 0.0,
      "step": 222500
    },
    {
      "epoch": 22.26,
      "grad_norm": 5.262133709038608e-05,
      "learning_rate": 1.294314381270903e-06,
      "loss": 0.0,
      "step": 222600
    },
    {
      "epoch": 22.27,
      "grad_norm": 0.0036346414126455784,
      "learning_rate": 1.2926421404682276e-06,
      "loss": 0.0,
      "step": 222700
    },
    {
      "epoch": 22.28,
      "grad_norm": 1.1790768439823296e-05,
      "learning_rate": 1.290969899665552e-06,
      "loss": 0.0,
      "step": 222800
    },
    {
      "epoch": 22.29,
      "grad_norm": 1.9646349755930714e-06,
      "learning_rate": 1.2892976588628763e-06,
      "loss": 0.0,
      "step": 222900
    },
    {
      "epoch": 22.3,
      "grad_norm": 2.03367494577833e-06,
      "learning_rate": 1.2876254180602008e-06,
      "loss": 0.0,
      "step": 223000
    },
    {
      "epoch": 22.31,
      "grad_norm": 4.9398859118809924e-05,
      "learning_rate": 1.285953177257525e-06,
      "loss": 0.0,
      "step": 223100
    },
    {
      "epoch": 22.32,
      "grad_norm": 2.9453542538249167e-06,
      "learning_rate": 1.2842809364548495e-06,
      "loss": 0.0035,
      "step": 223200
    },
    {
      "epoch": 22.33,
      "grad_norm": 7.152600119297858e-06,
      "learning_rate": 1.2826086956521742e-06,
      "loss": 0.0,
      "step": 223300
    },
    {
      "epoch": 22.34,
      "grad_norm": 2.245857467642054e-05,
      "learning_rate": 1.2809364548494984e-06,
      "loss": 0.0,
      "step": 223400
    },
    {
      "epoch": 22.35,
      "grad_norm": 1.8094841607307899e-06,
      "learning_rate": 1.2792642140468229e-06,
      "loss": 0.0,
      "step": 223500
    },
    {
      "epoch": 22.36,
      "grad_norm": 2.0084596599190263e-06,
      "learning_rate": 1.2775919732441471e-06,
      "loss": 0.0,
      "step": 223600
    },
    {
      "epoch": 22.37,
      "grad_norm": 3.4688196137722116e-06,
      "learning_rate": 1.2759197324414716e-06,
      "loss": 0.0,
      "step": 223700
    },
    {
      "epoch": 22.38,
      "grad_norm": 123.31082916259766,
      "learning_rate": 1.2742474916387963e-06,
      "loss": 0.002,
      "step": 223800
    },
    {
      "epoch": 22.39,
      "grad_norm": 2.896066916946438e-06,
      "learning_rate": 1.2725752508361205e-06,
      "loss": 0.0043,
      "step": 223900
    },
    {
      "epoch": 22.4,
      "grad_norm": 0.04940846189856529,
      "learning_rate": 1.270903010033445e-06,
      "loss": 0.0,
      "step": 224000
    },
    {
      "epoch": 22.41,
      "grad_norm": 1.4823303899902385e-05,
      "learning_rate": 1.2692307692307692e-06,
      "loss": 0.0,
      "step": 224100
    },
    {
      "epoch": 22.42,
      "grad_norm": 2.5912493128998904e-06,
      "learning_rate": 1.2675585284280937e-06,
      "loss": 0.0,
      "step": 224200
    },
    {
      "epoch": 22.43,
      "grad_norm": 2.450393822073238e-06,
      "learning_rate": 1.2658862876254184e-06,
      "loss": 0.0,
      "step": 224300
    },
    {
      "epoch": 22.44,
      "grad_norm": 1.862587168943719e-06,
      "learning_rate": 1.2642140468227426e-06,
      "loss": 0.0,
      "step": 224400
    },
    {
      "epoch": 22.45,
      "grad_norm": 0.00016126820992212743,
      "learning_rate": 1.262541806020067e-06,
      "loss": 0.0,
      "step": 224500
    },
    {
      "epoch": 22.46,
      "grad_norm": 7.110831120371586e-06,
      "learning_rate": 1.2608695652173913e-06,
      "loss": 0.0,
      "step": 224600
    },
    {
      "epoch": 22.47,
      "grad_norm": 2.7295325253362535e-06,
      "learning_rate": 1.2591973244147158e-06,
      "loss": 0.0,
      "step": 224700
    },
    {
      "epoch": 22.48,
      "grad_norm": 1.2589208608915214e-06,
      "learning_rate": 1.2575250836120402e-06,
      "loss": 0.0033,
      "step": 224800
    },
    {
      "epoch": 22.49,
      "grad_norm": 5.134460934641538e-06,
      "learning_rate": 1.2558528428093647e-06,
      "loss": 0.0,
      "step": 224900
    },
    {
      "epoch": 22.5,
      "grad_norm": 3.236037809983827e-05,
      "learning_rate": 1.2541806020066892e-06,
      "loss": 0.0,
      "step": 225000
    },
    {
      "epoch": 22.5,
      "eval_accuracy": 0.98765,
      "eval_f1": 0.98765,
      "eval_loss": 0.14892929792404175,
      "eval_runtime": 136.0909,
      "eval_samples_per_second": 293.921,
      "eval_steps_per_second": 293.921,
      "step": 225000
    },
    {
      "epoch": 22.51,
      "grad_norm": 2.1141204342711717e-06,
      "learning_rate": 1.2525083612040134e-06,
      "loss": 0.0078,
      "step": 225100
    },
    {
      "epoch": 22.52,
      "grad_norm": 4.8540355237491895e-06,
      "learning_rate": 1.2508361204013379e-06,
      "loss": 0.0034,
      "step": 225200
    },
    {
      "epoch": 22.53,
      "grad_norm": 0.00014950527111068368,
      "learning_rate": 1.2491638795986623e-06,
      "loss": 0.0,
      "step": 225300
    },
    {
      "epoch": 22.54,
      "grad_norm": 2.4088958525680937e-06,
      "learning_rate": 1.2474916387959868e-06,
      "loss": 0.0,
      "step": 225400
    },
    {
      "epoch": 22.55,
      "grad_norm": 0.00022813983378000557,
      "learning_rate": 1.245819397993311e-06,
      "loss": 0.0,
      "step": 225500
    },
    {
      "epoch": 22.56,
      "grad_norm": 2.390265308349626e-06,
      "learning_rate": 1.2441471571906355e-06,
      "loss": 0.0,
      "step": 225600
    },
    {
      "epoch": 22.57,
      "grad_norm": 2.549254304540227e-06,
      "learning_rate": 1.24247491638796e-06,
      "loss": 0.0,
      "step": 225700
    },
    {
      "epoch": 22.58,
      "grad_norm": 6.035674687154824e-06,
      "learning_rate": 1.2408026755852844e-06,
      "loss": 0.0,
      "step": 225800
    },
    {
      "epoch": 22.59,
      "grad_norm": 1.8804727233145968e-06,
      "learning_rate": 1.2391304347826089e-06,
      "loss": 0.0,
      "step": 225900
    },
    {
      "epoch": 22.6,
      "grad_norm": 9.251437404600438e-06,
      "learning_rate": 1.2374581939799331e-06,
      "loss": 0.0,
      "step": 226000
    },
    {
      "epoch": 22.61,
      "grad_norm": 3.065869577767444e-06,
      "learning_rate": 1.2357859531772576e-06,
      "loss": 0.0,
      "step": 226100
    },
    {
      "epoch": 22.62,
      "grad_norm": 2.609942384879105e-06,
      "learning_rate": 1.234113712374582e-06,
      "loss": 0.0,
      "step": 226200
    },
    {
      "epoch": 22.63,
      "grad_norm": 1.9762251213251147e-06,
      "learning_rate": 1.2324414715719065e-06,
      "loss": 0.0,
      "step": 226300
    },
    {
      "epoch": 22.64,
      "grad_norm": 0.0001107776042772457,
      "learning_rate": 1.230769230769231e-06,
      "loss": 0.0,
      "step": 226400
    },
    {
      "epoch": 22.65,
      "grad_norm": 1.7592420817891252e-06,
      "learning_rate": 1.2290969899665552e-06,
      "loss": 0.0,
      "step": 226500
    },
    {
      "epoch": 22.66,
      "grad_norm": 4.1662101466499735e-06,
      "learning_rate": 1.2274247491638797e-06,
      "loss": 0.0,
      "step": 226600
    },
    {
      "epoch": 22.67,
      "grad_norm": 4.5451852201949805e-06,
      "learning_rate": 1.2257525083612041e-06,
      "loss": 0.0,
      "step": 226700
    },
    {
      "epoch": 22.68,
      "grad_norm": 6.946735811652616e-05,
      "learning_rate": 1.2240802675585286e-06,
      "loss": 0.0,
      "step": 226800
    },
    {
      "epoch": 22.69,
      "grad_norm": 1.3782610039925203e-06,
      "learning_rate": 1.2224080267558529e-06,
      "loss": 0.006,
      "step": 226900
    },
    {
      "epoch": 22.7,
      "grad_norm": 2.4639061848574784e-06,
      "learning_rate": 1.2207357859531773e-06,
      "loss": 0.0104,
      "step": 227000
    },
    {
      "epoch": 22.71,
      "grad_norm": 2.496316710676183e-06,
      "learning_rate": 1.2190635451505018e-06,
      "loss": 0.0,
      "step": 227100
    },
    {
      "epoch": 22.72,
      "grad_norm": 2.4078549358819146e-06,
      "learning_rate": 1.2173913043478262e-06,
      "loss": 0.0007,
      "step": 227200
    },
    {
      "epoch": 22.73,
      "grad_norm": 3.610181011026725e-06,
      "learning_rate": 1.2157190635451507e-06,
      "loss": 0.0047,
      "step": 227300
    },
    {
      "epoch": 22.74,
      "grad_norm": 2.6597438136377605e-06,
      "learning_rate": 1.214046822742475e-06,
      "loss": 0.0,
      "step": 227400
    },
    {
      "epoch": 22.75,
      "grad_norm": 2.876392727557686e-06,
      "learning_rate": 1.2123745819397994e-06,
      "loss": 0.0,
      "step": 227500
    },
    {
      "epoch": 22.76,
      "grad_norm": 6.276714429986896e-06,
      "learning_rate": 1.2107023411371239e-06,
      "loss": 0.0,
      "step": 227600
    },
    {
      "epoch": 22.77,
      "grad_norm": 1.3213903912401292e-05,
      "learning_rate": 1.2090301003344483e-06,
      "loss": 0.0,
      "step": 227700
    },
    {
      "epoch": 22.78,
      "grad_norm": 0.0025315505918115377,
      "learning_rate": 1.2073578595317726e-06,
      "loss": 0.0048,
      "step": 227800
    },
    {
      "epoch": 22.79,
      "grad_norm": 1.0337781532143708e-05,
      "learning_rate": 1.205685618729097e-06,
      "loss": 0.0,
      "step": 227900
    },
    {
      "epoch": 22.8,
      "grad_norm": 2.1591881704807747e-06,
      "learning_rate": 1.2040133779264215e-06,
      "loss": 0.0,
      "step": 228000
    },
    {
      "epoch": 22.81,
      "grad_norm": 4.338608050602488e-06,
      "learning_rate": 1.202341137123746e-06,
      "loss": 0.0,
      "step": 228100
    },
    {
      "epoch": 22.82,
      "grad_norm": 1.976111889234744e-06,
      "learning_rate": 1.2006688963210704e-06,
      "loss": 0.0,
      "step": 228200
    },
    {
      "epoch": 22.83,
      "grad_norm": 1.1604581231949851e-05,
      "learning_rate": 1.1989966555183947e-06,
      "loss": 0.0051,
      "step": 228300
    },
    {
      "epoch": 22.84,
      "grad_norm": 3.1730382943351287e-06,
      "learning_rate": 1.1973244147157191e-06,
      "loss": 0.0,
      "step": 228400
    },
    {
      "epoch": 22.85,
      "grad_norm": 2.0415036487975158e-06,
      "learning_rate": 1.1956521739130436e-06,
      "loss": 0.0006,
      "step": 228500
    },
    {
      "epoch": 22.86,
      "grad_norm": 7.62998479331145e-06,
      "learning_rate": 1.193979933110368e-06,
      "loss": 0.0,
      "step": 228600
    },
    {
      "epoch": 22.87,
      "grad_norm": 2.8338245101622306e-06,
      "learning_rate": 1.1923076923076925e-06,
      "loss": 0.0,
      "step": 228700
    },
    {
      "epoch": 22.88,
      "grad_norm": 6.511565970868105e-06,
      "learning_rate": 1.1906354515050168e-06,
      "loss": 0.0,
      "step": 228800
    },
    {
      "epoch": 22.89,
      "grad_norm": 1.0614480743242893e-05,
      "learning_rate": 1.1889632107023412e-06,
      "loss": 0.0,
      "step": 228900
    },
    {
      "epoch": 22.9,
      "grad_norm": 8.845283446135e-06,
      "learning_rate": 1.1872909698996657e-06,
      "loss": 0.0031,
      "step": 229000
    },
    {
      "epoch": 22.91,
      "grad_norm": 8.616744162281975e-05,
      "learning_rate": 1.1856187290969901e-06,
      "loss": 0.0,
      "step": 229100
    },
    {
      "epoch": 22.92,
      "grad_norm": 1.3663597201230004e-05,
      "learning_rate": 1.1839464882943144e-06,
      "loss": 0.0,
      "step": 229200
    },
    {
      "epoch": 22.93,
      "grad_norm": 5.31476234755246e-06,
      "learning_rate": 1.1822742474916388e-06,
      "loss": 0.0,
      "step": 229300
    },
    {
      "epoch": 22.94,
      "grad_norm": 1.5794164937688038e-05,
      "learning_rate": 1.1806020066889633e-06,
      "loss": 0.0,
      "step": 229400
    },
    {
      "epoch": 22.95,
      "grad_norm": 3.345578534208471e-06,
      "learning_rate": 1.1789297658862878e-06,
      "loss": 0.0,
      "step": 229500
    },
    {
      "epoch": 22.96,
      "grad_norm": 1.899161907203961e-05,
      "learning_rate": 1.1772575250836122e-06,
      "loss": 0.0,
      "step": 229600
    },
    {
      "epoch": 22.97,
      "grad_norm": 1.202254952659132e-05,
      "learning_rate": 1.1755852842809365e-06,
      "loss": 0.0,
      "step": 229700
    },
    {
      "epoch": 22.98,
      "grad_norm": 3.4707027225522324e-05,
      "learning_rate": 1.173913043478261e-06,
      "loss": 0.0,
      "step": 229800
    },
    {
      "epoch": 22.99,
      "grad_norm": 2.837147349055158e-06,
      "learning_rate": 1.1722408026755854e-06,
      "loss": 0.0,
      "step": 229900
    },
    {
      "epoch": 23.0,
      "grad_norm": 1.9231772512284806e-06,
      "learning_rate": 1.1705685618729099e-06,
      "loss": 0.0,
      "step": 230000
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.988,
      "eval_f1": 0.988,
      "eval_loss": 0.13970816135406494,
      "eval_runtime": 138.5084,
      "eval_samples_per_second": 288.791,
      "eval_steps_per_second": 288.791,
      "step": 230000
    },
    {
      "epoch": 23.01,
      "grad_norm": 0.00043960308539681137,
      "learning_rate": 1.168896321070234e-06,
      "loss": 0.0,
      "step": 230100
    },
    {
      "epoch": 23.02,
      "grad_norm": 5.0748644753184635e-06,
      "learning_rate": 1.1672240802675586e-06,
      "loss": 0.0,
      "step": 230200
    },
    {
      "epoch": 23.03,
      "grad_norm": 6.698410288663581e-05,
      "learning_rate": 1.165551839464883e-06,
      "loss": 0.0,
      "step": 230300
    },
    {
      "epoch": 23.04,
      "grad_norm": 1.7563420442456845e-06,
      "learning_rate": 1.1638795986622075e-06,
      "loss": 0.0,
      "step": 230400
    },
    {
      "epoch": 23.05,
      "grad_norm": 2.1650689632224385e-06,
      "learning_rate": 1.162207357859532e-06,
      "loss": 0.0,
      "step": 230500
    },
    {
      "epoch": 23.06,
      "grad_norm": 2.9938364605186507e-06,
      "learning_rate": 1.1605351170568562e-06,
      "loss": 0.0,
      "step": 230600
    },
    {
      "epoch": 23.07,
      "grad_norm": 2.1382991235441295e-06,
      "learning_rate": 1.1588628762541807e-06,
      "loss": 0.0011,
      "step": 230700
    },
    {
      "epoch": 23.08,
      "grad_norm": 4.4728392822435126e-05,
      "learning_rate": 1.1571906354515051e-06,
      "loss": 0.0,
      "step": 230800
    },
    {
      "epoch": 23.09,
      "grad_norm": 1.2432237781467848e-05,
      "learning_rate": 1.1555183946488296e-06,
      "loss": 0.0,
      "step": 230900
    },
    {
      "epoch": 23.1,
      "grad_norm": 1.834149588830769e-05,
      "learning_rate": 1.153846153846154e-06,
      "loss": 0.0,
      "step": 231000
    },
    {
      "epoch": 23.11,
      "grad_norm": 7.334803740377538e-06,
      "learning_rate": 1.1521739130434783e-06,
      "loss": 0.0,
      "step": 231100
    },
    {
      "epoch": 23.12,
      "grad_norm": 3.3072183214244433e-06,
      "learning_rate": 1.1505016722408027e-06,
      "loss": 0.0001,
      "step": 231200
    },
    {
      "epoch": 23.13,
      "grad_norm": 0.0029784536454826593,
      "learning_rate": 1.1488294314381272e-06,
      "loss": 0.0049,
      "step": 231300
    },
    {
      "epoch": 23.14,
      "grad_norm": 0.00029570041806437075,
      "learning_rate": 1.1471571906354517e-06,
      "loss": 0.0,
      "step": 231400
    },
    {
      "epoch": 23.15,
      "grad_norm": 0.00011625168554019183,
      "learning_rate": 1.145484949832776e-06,
      "loss": 0.0,
      "step": 231500
    },
    {
      "epoch": 23.16,
      "grad_norm": 0.0046653710305690765,
      "learning_rate": 1.1438127090301004e-06,
      "loss": 0.0,
      "step": 231600
    },
    {
      "epoch": 23.17,
      "grad_norm": 6.324274636426708e-06,
      "learning_rate": 1.1421404682274248e-06,
      "loss": 0.0,
      "step": 231700
    },
    {
      "epoch": 23.18,
      "grad_norm": 5.813838924950687e-06,
      "learning_rate": 1.1404682274247493e-06,
      "loss": 0.0,
      "step": 231800
    },
    {
      "epoch": 23.19,
      "grad_norm": 1.855208552115073e-06,
      "learning_rate": 1.1387959866220738e-06,
      "loss": 0.0015,
      "step": 231900
    },
    {
      "epoch": 23.2,
      "grad_norm": 1.5933362647047034e-06,
      "learning_rate": 1.137123745819398e-06,
      "loss": 0.0,
      "step": 232000
    },
    {
      "epoch": 23.21,
      "grad_norm": 2.5436947908019647e-06,
      "learning_rate": 1.1354515050167225e-06,
      "loss": 0.0,
      "step": 232100
    },
    {
      "epoch": 23.22,
      "grad_norm": 1.4847254306005198e-06,
      "learning_rate": 1.133779264214047e-06,
      "loss": 0.0,
      "step": 232200
    },
    {
      "epoch": 23.23,
      "grad_norm": 2.0350166778371204e-06,
      "learning_rate": 1.1321070234113714e-06,
      "loss": 0.0,
      "step": 232300
    },
    {
      "epoch": 23.24,
      "grad_norm": 2.920028009611997e-06,
      "learning_rate": 1.1304347826086956e-06,
      "loss": 0.0054,
      "step": 232400
    },
    {
      "epoch": 23.25,
      "grad_norm": 2.4189600935642375e-06,
      "learning_rate": 1.12876254180602e-06,
      "loss": 0.0,
      "step": 232500
    },
    {
      "epoch": 23.26,
      "grad_norm": 4.0135386370820925e-05,
      "learning_rate": 1.1270903010033446e-06,
      "loss": 0.0,
      "step": 232600
    },
    {
      "epoch": 23.27,
      "grad_norm": 8.483904821332544e-06,
      "learning_rate": 1.125418060200669e-06,
      "loss": 0.0,
      "step": 232700
    },
    {
      "epoch": 23.28,
      "grad_norm": 1.1192724741704296e-05,
      "learning_rate": 1.1237458193979935e-06,
      "loss": 0.0,
      "step": 232800
    },
    {
      "epoch": 23.29,
      "grad_norm": 3.040320734726265e-06,
      "learning_rate": 1.1220735785953177e-06,
      "loss": 0.0,
      "step": 232900
    },
    {
      "epoch": 23.3,
      "grad_norm": 2.234099838460679e-06,
      "learning_rate": 1.1204013377926422e-06,
      "loss": 0.0,
      "step": 233000
    },
    {
      "epoch": 23.31,
      "grad_norm": 1.7016451465678983e-06,
      "learning_rate": 1.1187290969899666e-06,
      "loss": 0.0,
      "step": 233100
    },
    {
      "epoch": 23.32,
      "grad_norm": 2.1192927306401543e-06,
      "learning_rate": 1.1170568561872911e-06,
      "loss": 0.0013,
      "step": 233200
    },
    {
      "epoch": 23.33,
      "grad_norm": 3.3578889997443184e-06,
      "learning_rate": 1.1153846153846156e-06,
      "loss": 0.0,
      "step": 233300
    },
    {
      "epoch": 23.34,
      "grad_norm": 3.584066689654719e-06,
      "learning_rate": 1.1137123745819398e-06,
      "loss": 0.0,
      "step": 233400
    },
    {
      "epoch": 23.35,
      "grad_norm": 1.8127288967662025e-06,
      "learning_rate": 1.1120401337792643e-06,
      "loss": 0.0,
      "step": 233500
    },
    {
      "epoch": 23.36,
      "grad_norm": 1.8148879235013737e-06,
      "learning_rate": 1.1103678929765887e-06,
      "loss": 0.0,
      "step": 233600
    },
    {
      "epoch": 23.37,
      "grad_norm": 1.7151642168755643e-05,
      "learning_rate": 1.1086956521739132e-06,
      "loss": 0.0,
      "step": 233700
    },
    {
      "epoch": 23.38,
      "grad_norm": 2.9488589916581986e-06,
      "learning_rate": 1.1070234113712377e-06,
      "loss": 0.0,
      "step": 233800
    },
    {
      "epoch": 23.39,
      "grad_norm": 3.188793925801292e-06,
      "learning_rate": 1.105351170568562e-06,
      "loss": 0.0,
      "step": 233900
    },
    {
      "epoch": 23.4,
      "grad_norm": 3.0367280487553217e-05,
      "learning_rate": 1.1036789297658864e-06,
      "loss": 0.0,
      "step": 234000
    },
    {
      "epoch": 23.41,
      "grad_norm": 0.0007931198342703283,
      "learning_rate": 1.1020066889632108e-06,
      "loss": 0.0,
      "step": 234100
    },
    {
      "epoch": 23.42,
      "grad_norm": 0.0004764864861499518,
      "learning_rate": 1.1003344481605353e-06,
      "loss": 0.0,
      "step": 234200
    },
    {
      "epoch": 23.43,
      "grad_norm": 2.6186989998677745e-06,
      "learning_rate": 1.0986622073578595e-06,
      "loss": 0.0068,
      "step": 234300
    },
    {
      "epoch": 23.44,
      "grad_norm": 1.6648232303850818e-06,
      "learning_rate": 1.096989966555184e-06,
      "loss": 0.0066,
      "step": 234400
    },
    {
      "epoch": 23.45,
      "grad_norm": 2.701480980249471e-06,
      "learning_rate": 1.0953177257525085e-06,
      "loss": 0.0,
      "step": 234500
    },
    {
      "epoch": 23.46,
      "grad_norm": 2.0431327811820665e-06,
      "learning_rate": 1.093645484949833e-06,
      "loss": 0.0,
      "step": 234600
    },
    {
      "epoch": 23.47,
      "grad_norm": 4.373225237941369e-06,
      "learning_rate": 1.0919732441471572e-06,
      "loss": 0.0,
      "step": 234700
    },
    {
      "epoch": 23.48,
      "grad_norm": 1.955356310645584e-06,
      "learning_rate": 1.0903010033444816e-06,
      "loss": 0.0,
      "step": 234800
    },
    {
      "epoch": 23.49,
      "grad_norm": 3.090561449425877e-06,
      "learning_rate": 1.088628762541806e-06,
      "loss": 0.0,
      "step": 234900
    },
    {
      "epoch": 23.5,
      "grad_norm": 2.478899114066735e-05,
      "learning_rate": 1.0869565217391306e-06,
      "loss": 0.0,
      "step": 235000
    },
    {
      "epoch": 23.5,
      "eval_accuracy": 0.9877,
      "eval_f1": 0.9877,
      "eval_loss": 0.14368464052677155,
      "eval_runtime": 137.2609,
      "eval_samples_per_second": 291.416,
      "eval_steps_per_second": 291.416,
      "step": 235000
    },
    {
      "epoch": 23.51,
      "grad_norm": 2.0134564238105668e-06,
      "learning_rate": 1.085284280936455e-06,
      "loss": 0.0,
      "step": 235100
    },
    {
      "epoch": 23.52,
      "grad_norm": 0.00012607380631379783,
      "learning_rate": 1.0836120401337793e-06,
      "loss": 0.0,
      "step": 235200
    },
    {
      "epoch": 23.53,
      "grad_norm": 3.1202098398352973e-06,
      "learning_rate": 1.0819397993311037e-06,
      "loss": 0.0,
      "step": 235300
    },
    {
      "epoch": 23.54,
      "grad_norm": 3.076041821259423e-06,
      "learning_rate": 1.0802675585284282e-06,
      "loss": 0.0,
      "step": 235400
    },
    {
      "epoch": 23.55,
      "grad_norm": 2.8738932087435387e-06,
      "learning_rate": 1.0785953177257526e-06,
      "loss": 0.0036,
      "step": 235500
    },
    {
      "epoch": 23.56,
      "grad_norm": 1.94904441741528e-06,
      "learning_rate": 1.076923076923077e-06,
      "loss": 0.0,
      "step": 235600
    },
    {
      "epoch": 23.57,
      "grad_norm": 8.461877769150306e-06,
      "learning_rate": 1.0752508361204014e-06,
      "loss": 0.0,
      "step": 235700
    },
    {
      "epoch": 23.58,
      "grad_norm": 8.348670235136524e-06,
      "learning_rate": 1.0735785953177258e-06,
      "loss": 0.0,
      "step": 235800
    },
    {
      "epoch": 23.59,
      "grad_norm": 3.347354549987358e-06,
      "learning_rate": 1.0719063545150503e-06,
      "loss": 0.0,
      "step": 235900
    },
    {
      "epoch": 23.6,
      "grad_norm": 2.0461826352402568e-05,
      "learning_rate": 1.0702341137123747e-06,
      "loss": 0.0,
      "step": 236000
    },
    {
      "epoch": 23.61,
      "grad_norm": 4.463665391085669e-06,
      "learning_rate": 1.0685618729096992e-06,
      "loss": 0.0,
      "step": 236100
    },
    {
      "epoch": 23.62,
      "grad_norm": 2.9481900583050447e-06,
      "learning_rate": 1.0668896321070234e-06,
      "loss": 0.0,
      "step": 236200
    },
    {
      "epoch": 23.63,
      "grad_norm": 1.6733443771954626e-05,
      "learning_rate": 1.065217391304348e-06,
      "loss": 0.0,
      "step": 236300
    },
    {
      "epoch": 23.64,
      "grad_norm": 2.97983660857426e-06,
      "learning_rate": 1.0635451505016724e-06,
      "loss": 0.0,
      "step": 236400
    },
    {
      "epoch": 23.65,
      "grad_norm": 4.343262844486162e-06,
      "learning_rate": 1.0618729096989968e-06,
      "loss": 0.0,
      "step": 236500
    },
    {
      "epoch": 23.66,
      "grad_norm": 1.3636974927067058e-06,
      "learning_rate": 1.060200668896321e-06,
      "loss": 0.0005,
      "step": 236600
    },
    {
      "epoch": 23.67,
      "grad_norm": 1.4470730320681469e-06,
      "learning_rate": 1.0585284280936455e-06,
      "loss": 0.0,
      "step": 236700
    },
    {
      "epoch": 23.68,
      "grad_norm": 0.0001465627137804404,
      "learning_rate": 1.05685618729097e-06,
      "loss": 0.0,
      "step": 236800
    },
    {
      "epoch": 23.69,
      "grad_norm": 5.631826934404671e-05,
      "learning_rate": 1.0551839464882945e-06,
      "loss": 0.0,
      "step": 236900
    },
    {
      "epoch": 23.7,
      "grad_norm": 5.150936885911506e-06,
      "learning_rate": 1.0535117056856187e-06,
      "loss": 0.0,
      "step": 237000
    },
    {
      "epoch": 23.71,
      "grad_norm": 1.5397962442875723e-06,
      "learning_rate": 1.0518394648829432e-06,
      "loss": 0.0,
      "step": 237100
    },
    {
      "epoch": 23.72,
      "grad_norm": 1.8038625739791314e-06,
      "learning_rate": 1.0501672240802676e-06,
      "loss": 0.0031,
      "step": 237200
    },
    {
      "epoch": 23.73,
      "grad_norm": 1.82100211532088e-05,
      "learning_rate": 1.048494983277592e-06,
      "loss": 0.0,
      "step": 237300
    },
    {
      "epoch": 23.74,
      "grad_norm": 1.540220296192274e-06,
      "learning_rate": 1.0468227424749165e-06,
      "loss": 0.0,
      "step": 237400
    },
    {
      "epoch": 23.75,
      "grad_norm": 3.610687826949288e-06,
      "learning_rate": 1.0451505016722408e-06,
      "loss": 0.0,
      "step": 237500
    },
    {
      "epoch": 23.76,
      "grad_norm": 2.127660991391167e-06,
      "learning_rate": 1.0434782608695653e-06,
      "loss": 0.0,
      "step": 237600
    },
    {
      "epoch": 23.77,
      "grad_norm": 1.2974899163964437e-06,
      "learning_rate": 1.0418060200668897e-06,
      "loss": 0.0009,
      "step": 237700
    },
    {
      "epoch": 23.78,
      "grad_norm": 2.5898298190440983e-06,
      "learning_rate": 1.0401337792642142e-06,
      "loss": 0.0,
      "step": 237800
    },
    {
      "epoch": 23.79,
      "grad_norm": 3.946627657569479e-06,
      "learning_rate": 1.0384615384615386e-06,
      "loss": 0.0,
      "step": 237900
    },
    {
      "epoch": 23.8,
      "grad_norm": 8.825607801554725e-05,
      "learning_rate": 1.0367892976588629e-06,
      "loss": 0.0,
      "step": 238000
    },
    {
      "epoch": 23.81,
      "grad_norm": 1.747809960761515e-06,
      "learning_rate": 1.0351170568561873e-06,
      "loss": 0.0,
      "step": 238100
    },
    {
      "epoch": 23.82,
      "grad_norm": 7.072359949233942e-06,
      "learning_rate": 1.0334448160535118e-06,
      "loss": 0.0004,
      "step": 238200
    },
    {
      "epoch": 23.83,
      "grad_norm": 4.897616690868745e-06,
      "learning_rate": 1.0317725752508363e-06,
      "loss": 0.0,
      "step": 238300
    },
    {
      "epoch": 23.84,
      "grad_norm": 2.8052606921846746e-06,
      "learning_rate": 1.0301003344481607e-06,
      "loss": 0.0,
      "step": 238400
    },
    {
      "epoch": 23.85,
      "grad_norm": 1.638777007428871e-06,
      "learning_rate": 1.028428093645485e-06,
      "loss": 0.0,
      "step": 238500
    },
    {
      "epoch": 23.86,
      "grad_norm": 3.3246594739466673e-06,
      "learning_rate": 1.0267558528428094e-06,
      "loss": 0.0,
      "step": 238600
    },
    {
      "epoch": 23.87,
      "grad_norm": 8.469135536870454e-06,
      "learning_rate": 1.025083612040134e-06,
      "loss": 0.0,
      "step": 238700
    },
    {
      "epoch": 23.88,
      "grad_norm": 1.0242590633424697e-06,
      "learning_rate": 1.0234113712374584e-06,
      "loss": 0.0,
      "step": 238800
    },
    {
      "epoch": 23.89,
      "grad_norm": 2.301003405591473e-06,
      "learning_rate": 1.0217391304347828e-06,
      "loss": 0.0,
      "step": 238900
    },
    {
      "epoch": 23.9,
      "grad_norm": 1.0797813274621149e-06,
      "learning_rate": 1.020066889632107e-06,
      "loss": 0.0,
      "step": 239000
    },
    {
      "epoch": 23.91,
      "grad_norm": 9.672876331023872e-06,
      "learning_rate": 1.0183946488294315e-06,
      "loss": 0.0,
      "step": 239100
    },
    {
      "epoch": 23.92,
      "grad_norm": 2.800751190079609e-06,
      "learning_rate": 1.016722408026756e-06,
      "loss": 0.0,
      "step": 239200
    },
    {
      "epoch": 23.93,
      "grad_norm": 9.375202694172913e-07,
      "learning_rate": 1.0150501672240804e-06,
      "loss": 0.0,
      "step": 239300
    },
    {
      "epoch": 23.94,
      "grad_norm": 0.005259317811578512,
      "learning_rate": 1.0133779264214047e-06,
      "loss": 0.0,
      "step": 239400
    },
    {
      "epoch": 23.95,
      "grad_norm": 4.873456418863498e-06,
      "learning_rate": 1.0117056856187292e-06,
      "loss": 0.0,
      "step": 239500
    },
    {
      "epoch": 23.96,
      "grad_norm": 1.0279156867909478e-06,
      "learning_rate": 1.0100334448160536e-06,
      "loss": 0.0,
      "step": 239600
    },
    {
      "epoch": 23.97,
      "grad_norm": 1.5694160993007245e-06,
      "learning_rate": 1.008361204013378e-06,
      "loss": 0.0,
      "step": 239700
    },
    {
      "epoch": 23.98,
      "grad_norm": 5.572123882302549e-06,
      "learning_rate": 1.0066889632107023e-06,
      "loss": 0.0051,
      "step": 239800
    },
    {
      "epoch": 23.99,
      "grad_norm": 8.108318979793694e-06,
      "learning_rate": 1.0050167224080268e-06,
      "loss": 0.0037,
      "step": 239900
    },
    {
      "epoch": 24.0,
      "grad_norm": 1.9160975170962047e-06,
      "learning_rate": 1.0033444816053512e-06,
      "loss": 0.0,
      "step": 240000
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.9875,
      "eval_f1": 0.9875,
      "eval_loss": 0.15013839304447174,
      "eval_runtime": 135.7089,
      "eval_samples_per_second": 294.749,
      "eval_steps_per_second": 294.749,
      "step": 240000
    },
    {
      "epoch": 24.01,
      "grad_norm": 2.217173687313334e-06,
      "learning_rate": 1.0016722408026757e-06,
      "loss": 0.0,
      "step": 240100
    },
    {
      "epoch": 24.02,
      "grad_norm": 8.573334343964234e-05,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0,
      "step": 240200
    },
    {
      "epoch": 24.03,
      "grad_norm": 2.2610554424318252e-06,
      "learning_rate": 9.983277591973244e-07,
      "loss": 0.0,
      "step": 240300
    },
    {
      "epoch": 24.04,
      "grad_norm": 3.525610736687668e-05,
      "learning_rate": 9.966555183946489e-07,
      "loss": 0.002,
      "step": 240400
    },
    {
      "epoch": 24.05,
      "grad_norm": 1.1360249345671036e-06,
      "learning_rate": 9.949832775919733e-07,
      "loss": 0.0,
      "step": 240500
    },
    {
      "epoch": 24.06,
      "grad_norm": 1.7239157159565366e-06,
      "learning_rate": 9.933110367892978e-07,
      "loss": 0.0028,
      "step": 240600
    },
    {
      "epoch": 24.07,
      "grad_norm": 3.1304966796597e-06,
      "learning_rate": 9.916387959866223e-07,
      "loss": 0.0074,
      "step": 240700
    },
    {
      "epoch": 24.08,
      "grad_norm": 1.5380784361695987e-06,
      "learning_rate": 9.899665551839465e-07,
      "loss": 0.001,
      "step": 240800
    },
    {
      "epoch": 24.09,
      "grad_norm": 5.637221875076648e-06,
      "learning_rate": 9.88294314381271e-07,
      "loss": 0.0,
      "step": 240900
    },
    {
      "epoch": 24.1,
      "grad_norm": 1.9916706150979735e-06,
      "learning_rate": 9.866220735785954e-07,
      "loss": 0.0,
      "step": 241000
    },
    {
      "epoch": 24.11,
      "grad_norm": 1.8978020079885027e-06,
      "learning_rate": 9.849498327759199e-07,
      "loss": 0.0,
      "step": 241100
    },
    {
      "epoch": 24.12,
      "grad_norm": 2.0716247490781825e-06,
      "learning_rate": 9.832775919732443e-07,
      "loss": 0.0,
      "step": 241200
    },
    {
      "epoch": 24.13,
      "grad_norm": 1.9153335415467154e-06,
      "learning_rate": 9.816053511705686e-07,
      "loss": 0.0,
      "step": 241300
    },
    {
      "epoch": 24.14,
      "grad_norm": 1.0115123814102844e-06,
      "learning_rate": 9.79933110367893e-07,
      "loss": 0.0,
      "step": 241400
    },
    {
      "epoch": 24.15,
      "grad_norm": 1.018683747133764e-06,
      "learning_rate": 9.782608695652175e-07,
      "loss": 0.0018,
      "step": 241500
    },
    {
      "epoch": 24.16,
      "grad_norm": 2.7036819574277615e-06,
      "learning_rate": 9.76588628762542e-07,
      "loss": 0.0,
      "step": 241600
    },
    {
      "epoch": 24.17,
      "grad_norm": 6.6156862885691226e-06,
      "learning_rate": 9.749163879598662e-07,
      "loss": 0.0,
      "step": 241700
    },
    {
      "epoch": 24.18,
      "grad_norm": 2.5971964987547835e-06,
      "learning_rate": 9.732441471571907e-07,
      "loss": 0.0,
      "step": 241800
    },
    {
      "epoch": 24.19,
      "grad_norm": 1.5181810795183992e-06,
      "learning_rate": 9.715719063545151e-07,
      "loss": 0.0,
      "step": 241900
    },
    {
      "epoch": 24.2,
      "grad_norm": 1.7683860278339125e-06,
      "learning_rate": 9.698996655518396e-07,
      "loss": 0.0,
      "step": 242000
    },
    {
      "epoch": 24.21,
      "grad_norm": 3.6412154713616474e-06,
      "learning_rate": 9.682274247491639e-07,
      "loss": 0.0004,
      "step": 242100
    },
    {
      "epoch": 24.22,
      "grad_norm": 1.359268935630098e-06,
      "learning_rate": 9.665551839464883e-07,
      "loss": 0.0,
      "step": 242200
    },
    {
      "epoch": 24.23,
      "grad_norm": 3.5012128591915825e-06,
      "learning_rate": 9.648829431438128e-07,
      "loss": 0.0,
      "step": 242300
    },
    {
      "epoch": 24.24,
      "grad_norm": 3.7246281863190234e-06,
      "learning_rate": 9.632107023411372e-07,
      "loss": 0.0,
      "step": 242400
    },
    {
      "epoch": 24.25,
      "grad_norm": 5.6625767683726735e-06,
      "learning_rate": 9.615384615384617e-07,
      "loss": 0.0,
      "step": 242500
    },
    {
      "epoch": 24.26,
      "grad_norm": 1.992144916584948e-06,
      "learning_rate": 9.59866220735786e-07,
      "loss": 0.0,
      "step": 242600
    },
    {
      "epoch": 24.27,
      "grad_norm": 2.249276576549164e-06,
      "learning_rate": 9.581939799331104e-07,
      "loss": 0.0062,
      "step": 242700
    },
    {
      "epoch": 24.28,
      "grad_norm": 1.3135835388311534e-06,
      "learning_rate": 9.565217391304349e-07,
      "loss": 0.0,
      "step": 242800
    },
    {
      "epoch": 24.29,
      "grad_norm": 1.8054714701065677e-06,
      "learning_rate": 9.548494983277593e-07,
      "loss": 0.0,
      "step": 242900
    },
    {
      "epoch": 24.3,
      "grad_norm": 3.5502798709785566e-06,
      "learning_rate": 9.531772575250838e-07,
      "loss": 0.0,
      "step": 243000
    },
    {
      "epoch": 24.31,
      "grad_norm": 2.3826137294236105e-06,
      "learning_rate": 9.515050167224081e-07,
      "loss": 0.0,
      "step": 243100
    },
    {
      "epoch": 24.32,
      "grad_norm": 1.0196755283686798e-05,
      "learning_rate": 9.498327759197325e-07,
      "loss": 0.0,
      "step": 243200
    },
    {
      "epoch": 24.33,
      "grad_norm": 1.386860390084621e-06,
      "learning_rate": 9.48160535117057e-07,
      "loss": 0.0,
      "step": 243300
    },
    {
      "epoch": 24.34,
      "grad_norm": 1.7781057977117598e-06,
      "learning_rate": 9.464882943143813e-07,
      "loss": 0.0,
      "step": 243400
    },
    {
      "epoch": 24.35,
      "grad_norm": 1.5628019127689186e-06,
      "learning_rate": 9.448160535117059e-07,
      "loss": 0.0,
      "step": 243500
    },
    {
      "epoch": 24.36,
      "grad_norm": 9.654933819547296e-06,
      "learning_rate": 9.431438127090302e-07,
      "loss": 0.0,
      "step": 243600
    },
    {
      "epoch": 24.37,
      "grad_norm": 2.1423722955660196e-06,
      "learning_rate": 9.414715719063546e-07,
      "loss": 0.0,
      "step": 243700
    },
    {
      "epoch": 24.38,
      "grad_norm": 2.166335207220982e-06,
      "learning_rate": 9.39799331103679e-07,
      "loss": 0.0,
      "step": 243800
    },
    {
      "epoch": 24.39,
      "grad_norm": 1.555784479023714e-06,
      "learning_rate": 9.381270903010034e-07,
      "loss": 0.0011,
      "step": 243900
    },
    {
      "epoch": 24.4,
      "grad_norm": 2.7639248401101213e-06,
      "learning_rate": 9.364548494983278e-07,
      "loss": 0.0056,
      "step": 244000
    },
    {
      "epoch": 24.41,
      "grad_norm": 1.255690335710824e-06,
      "learning_rate": 9.347826086956522e-07,
      "loss": 0.0,
      "step": 244100
    },
    {
      "epoch": 24.42,
      "grad_norm": 1.6317754898409476e-06,
      "learning_rate": 9.331103678929766e-07,
      "loss": 0.0,
      "step": 244200
    },
    {
      "epoch": 24.43,
      "grad_norm": 5.905946818529628e-05,
      "learning_rate": 9.314381270903011e-07,
      "loss": 0.0,
      "step": 244300
    },
    {
      "epoch": 24.44,
      "grad_norm": 6.9295992943807505e-06,
      "learning_rate": 9.297658862876255e-07,
      "loss": 0.0,
      "step": 244400
    },
    {
      "epoch": 24.45,
      "grad_norm": 6.23083451500861e-06,
      "learning_rate": 9.280936454849498e-07,
      "loss": 0.009,
      "step": 244500
    },
    {
      "epoch": 24.46,
      "grad_norm": 2.21546838474751e-06,
      "learning_rate": 9.264214046822743e-07,
      "loss": 0.0,
      "step": 244600
    },
    {
      "epoch": 24.47,
      "grad_norm": 1.5623327271896414e-06,
      "learning_rate": 9.247491638795987e-07,
      "loss": 0.0,
      "step": 244700
    },
    {
      "epoch": 24.48,
      "grad_norm": 1.841994708229322e-05,
      "learning_rate": 9.230769230769232e-07,
      "loss": 0.0,
      "step": 244800
    },
    {
      "epoch": 24.49,
      "grad_norm": 1.967902562682866e-06,
      "learning_rate": 9.214046822742476e-07,
      "loss": 0.0,
      "step": 244900
    },
    {
      "epoch": 24.5,
      "grad_norm": 7.149414159357548e-05,
      "learning_rate": 9.197324414715719e-07,
      "loss": 0.0,
      "step": 245000
    },
    {
      "epoch": 24.5,
      "eval_accuracy": 0.987675,
      "eval_f1": 0.987675,
      "eval_loss": 0.14463376998901367,
      "eval_runtime": 133.7499,
      "eval_samples_per_second": 299.066,
      "eval_steps_per_second": 299.066,
      "step": 245000
    },
    {
      "epoch": 24.51,
      "grad_norm": 2.040028903138591e-06,
      "learning_rate": 9.180602006688964e-07,
      "loss": 0.0,
      "step": 245100
    },
    {
      "epoch": 24.52,
      "grad_norm": 8.409876500081737e-06,
      "learning_rate": 9.163879598662208e-07,
      "loss": 0.0,
      "step": 245200
    },
    {
      "epoch": 24.53,
      "grad_norm": 0.0001772494870238006,
      "learning_rate": 9.147157190635453e-07,
      "loss": 0.0,
      "step": 245300
    },
    {
      "epoch": 24.54,
      "grad_norm": 1.979961780307349e-06,
      "learning_rate": 9.130434782608697e-07,
      "loss": 0.0,
      "step": 245400
    },
    {
      "epoch": 24.55,
      "grad_norm": 1.6618955669400748e-06,
      "learning_rate": 9.11371237458194e-07,
      "loss": 0.0,
      "step": 245500
    },
    {
      "epoch": 24.56,
      "grad_norm": 9.364392212773964e-07,
      "learning_rate": 9.096989966555185e-07,
      "loss": 0.0,
      "step": 245600
    },
    {
      "epoch": 24.57,
      "grad_norm": 1.2160958249296527e-06,
      "learning_rate": 9.080267558528428e-07,
      "loss": 0.0007,
      "step": 245700
    },
    {
      "epoch": 24.58,
      "grad_norm": 7.665908015042078e-06,
      "learning_rate": 9.063545150501674e-07,
      "loss": 0.0,
      "step": 245800
    },
    {
      "epoch": 24.59,
      "grad_norm": 8.277664892375469e-05,
      "learning_rate": 9.046822742474918e-07,
      "loss": 0.0,
      "step": 245900
    },
    {
      "epoch": 24.6,
      "grad_norm": 3.3383425943611655e-06,
      "learning_rate": 9.030100334448161e-07,
      "loss": 0.0,
      "step": 246000
    },
    {
      "epoch": 24.61,
      "grad_norm": 1.6881747342267772e-06,
      "learning_rate": 9.013377926421406e-07,
      "loss": 0.0,
      "step": 246100
    },
    {
      "epoch": 24.62,
      "grad_norm": 1.5223195077851415e-06,
      "learning_rate": 8.996655518394649e-07,
      "loss": 0.0,
      "step": 246200
    },
    {
      "epoch": 24.63,
      "grad_norm": 1.902065378089901e-06,
      "learning_rate": 8.979933110367894e-07,
      "loss": 0.0047,
      "step": 246300
    },
    {
      "epoch": 24.64,
      "grad_norm": 2.0098498225706862e-06,
      "learning_rate": 8.963210702341138e-07,
      "loss": 0.0,
      "step": 246400
    },
    {
      "epoch": 24.65,
      "grad_norm": 1.1443703442637343e-05,
      "learning_rate": 8.946488294314382e-07,
      "loss": 0.0,
      "step": 246500
    },
    {
      "epoch": 24.66,
      "grad_norm": 4.432857167557813e-06,
      "learning_rate": 8.929765886287627e-07,
      "loss": 0.0,
      "step": 246600
    },
    {
      "epoch": 24.67,
      "grad_norm": 1.3005646906094626e-06,
      "learning_rate": 8.91304347826087e-07,
      "loss": 0.0,
      "step": 246700
    },
    {
      "epoch": 24.68,
      "grad_norm": 2.724513251450844e-06,
      "learning_rate": 8.896321070234114e-07,
      "loss": 0.0,
      "step": 246800
    },
    {
      "epoch": 24.69,
      "grad_norm": 1.3002398191019893e-05,
      "learning_rate": 8.879598662207358e-07,
      "loss": 0.0,
      "step": 246900
    },
    {
      "epoch": 24.7,
      "grad_norm": 3.660285074147396e-06,
      "learning_rate": 8.862876254180602e-07,
      "loss": 0.0,
      "step": 247000
    },
    {
      "epoch": 24.71,
      "grad_norm": 1.4119721527094953e-05,
      "learning_rate": 8.846153846153848e-07,
      "loss": 0.0,
      "step": 247100
    },
    {
      "epoch": 24.72,
      "grad_norm": 1.8244367083752877e-06,
      "learning_rate": 8.829431438127091e-07,
      "loss": 0.0,
      "step": 247200
    },
    {
      "epoch": 24.73,
      "grad_norm": 2.2953408915782347e-06,
      "learning_rate": 8.812709030100335e-07,
      "loss": 0.0,
      "step": 247300
    },
    {
      "epoch": 24.74,
      "grad_norm": 1.3426415534922853e-06,
      "learning_rate": 8.795986622073579e-07,
      "loss": 0.0045,
      "step": 247400
    },
    {
      "epoch": 24.75,
      "grad_norm": 2.1255984393064864e-05,
      "learning_rate": 8.779264214046823e-07,
      "loss": 0.0,
      "step": 247500
    },
    {
      "epoch": 24.76,
      "grad_norm": 1.6822939414851135e-06,
      "learning_rate": 8.762541806020069e-07,
      "loss": 0.0,
      "step": 247600
    },
    {
      "epoch": 24.77,
      "grad_norm": 2.1174896573938895e-06,
      "learning_rate": 8.745819397993312e-07,
      "loss": 0.0,
      "step": 247700
    },
    {
      "epoch": 24.78,
      "grad_norm": 1.8592134438222274e-05,
      "learning_rate": 8.729096989966556e-07,
      "loss": 0.0,
      "step": 247800
    },
    {
      "epoch": 24.79,
      "grad_norm": 2.290886186528951e-06,
      "learning_rate": 8.7123745819398e-07,
      "loss": 0.0,
      "step": 247900
    },
    {
      "epoch": 24.8,
      "grad_norm": 1.563448222441366e-06,
      "learning_rate": 8.695652173913044e-07,
      "loss": 0.0,
      "step": 248000
    },
    {
      "epoch": 24.81,
      "grad_norm": 1.2164559848315548e-06,
      "learning_rate": 8.678929765886289e-07,
      "loss": 0.0028,
      "step": 248100
    },
    {
      "epoch": 24.82,
      "grad_norm": 1.7395184386259643e-06,
      "learning_rate": 8.662207357859533e-07,
      "loss": 0.0001,
      "step": 248200
    },
    {
      "epoch": 24.83,
      "grad_norm": 1.5021948911453364e-06,
      "learning_rate": 8.645484949832777e-07,
      "loss": 0.0,
      "step": 248300
    },
    {
      "epoch": 24.84,
      "grad_norm": 1.945095027622301e-05,
      "learning_rate": 8.628762541806021e-07,
      "loss": 0.0,
      "step": 248400
    },
    {
      "epoch": 24.85,
      "grad_norm": 1.636931528992136e-06,
      "learning_rate": 8.612040133779265e-07,
      "loss": 0.0,
      "step": 248500
    },
    {
      "epoch": 24.86,
      "grad_norm": 2.2785177407058654e-06,
      "learning_rate": 8.595317725752509e-07,
      "loss": 0.0,
      "step": 248600
    },
    {
      "epoch": 24.87,
      "grad_norm": 1.2720621498374385e-06,
      "learning_rate": 8.578595317725753e-07,
      "loss": 0.0,
      "step": 248700
    },
    {
      "epoch": 24.88,
      "grad_norm": 0.0009492366807535291,
      "learning_rate": 8.561872909698997e-07,
      "loss": 0.0,
      "step": 248800
    },
    {
      "epoch": 24.89,
      "grad_norm": 1.7387096704624128e-06,
      "learning_rate": 8.545150501672242e-07,
      "loss": 0.0,
      "step": 248900
    },
    {
      "epoch": 24.9,
      "grad_norm": 3.0261933261499507e-06,
      "learning_rate": 8.528428093645486e-07,
      "loss": 0.0,
      "step": 249000
    },
    {
      "epoch": 24.91,
      "grad_norm": 1.878150214906782e-06,
      "learning_rate": 8.511705685618729e-07,
      "loss": 0.0,
      "step": 249100
    },
    {
      "epoch": 24.92,
      "grad_norm": 4.968315806763712e-06,
      "learning_rate": 8.494983277591974e-07,
      "loss": 0.0,
      "step": 249200
    },
    {
      "epoch": 24.93,
      "grad_norm": 1.8417338196741184e-06,
      "learning_rate": 8.478260869565217e-07,
      "loss": 0.0,
      "step": 249300
    },
    {
      "epoch": 24.94,
      "grad_norm": 1.265710011466581e-06,
      "learning_rate": 8.461538461538463e-07,
      "loss": 0.0,
      "step": 249400
    },
    {
      "epoch": 24.95,
      "grad_norm": 1.504531383034191e-06,
      "learning_rate": 8.444816053511706e-07,
      "loss": 0.0,
      "step": 249500
    },
    {
      "epoch": 24.96,
      "grad_norm": 8.036552753765136e-05,
      "learning_rate": 8.42809364548495e-07,
      "loss": 0.0,
      "step": 249600
    },
    {
      "epoch": 24.97,
      "grad_norm": 1.7299427099715103e-06,
      "learning_rate": 8.411371237458195e-07,
      "loss": 0.0,
      "step": 249700
    },
    {
      "epoch": 24.98,
      "grad_norm": 9.004021194414236e-06,
      "learning_rate": 8.394648829431438e-07,
      "loss": 0.0,
      "step": 249800
    },
    {
      "epoch": 24.99,
      "grad_norm": 1.3094919495415525e-06,
      "learning_rate": 8.377926421404684e-07,
      "loss": 0.0009,
      "step": 249900
    },
    {
      "epoch": 25.0,
      "grad_norm": 3.1047757147462107e-06,
      "learning_rate": 8.361204013377927e-07,
      "loss": 0.0,
      "step": 250000
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.9874,
      "eval_f1": 0.9874,
      "eval_loss": 0.15316404402256012,
      "eval_runtime": 134.4883,
      "eval_samples_per_second": 297.424,
      "eval_steps_per_second": 297.424,
      "step": 250000
    },
    {
      "epoch": 25.01,
      "grad_norm": 1.1634468819465837e-06,
      "learning_rate": 8.344481605351171e-07,
      "loss": 0.0,
      "step": 250100
    },
    {
      "epoch": 25.02,
      "grad_norm": 2.2065494249545736e-06,
      "learning_rate": 8.327759197324416e-07,
      "loss": 0.0,
      "step": 250200
    },
    {
      "epoch": 25.03,
      "grad_norm": 5.784725544799585e-06,
      "learning_rate": 8.311036789297659e-07,
      "loss": 0.0,
      "step": 250300
    },
    {
      "epoch": 25.04,
      "grad_norm": 0.0002647130750119686,
      "learning_rate": 8.294314381270905e-07,
      "loss": 0.0002,
      "step": 250400
    },
    {
      "epoch": 25.05,
      "grad_norm": 6.513052994705504e-07,
      "learning_rate": 8.277591973244148e-07,
      "loss": 0.0,
      "step": 250500
    },
    {
      "epoch": 25.06,
      "grad_norm": 3.345231334606069e-06,
      "learning_rate": 8.260869565217392e-07,
      "loss": 0.0,
      "step": 250600
    },
    {
      "epoch": 25.07,
      "grad_norm": 7.839264981157612e-06,
      "learning_rate": 8.244147157190636e-07,
      "loss": 0.0,
      "step": 250700
    },
    {
      "epoch": 25.08,
      "grad_norm": 5.135179435455939e-06,
      "learning_rate": 8.22742474916388e-07,
      "loss": 0.0,
      "step": 250800
    },
    {
      "epoch": 25.09,
      "grad_norm": 1.4390598153113388e-05,
      "learning_rate": 8.210702341137125e-07,
      "loss": 0.0,
      "step": 250900
    },
    {
      "epoch": 25.1,
      "grad_norm": 8.922448273551709e-07,
      "learning_rate": 8.193979933110368e-07,
      "loss": 0.0,
      "step": 251000
    },
    {
      "epoch": 25.11,
      "grad_norm": 2.3406830678140977e-06,
      "learning_rate": 8.177257525083613e-07,
      "loss": 0.0052,
      "step": 251100
    },
    {
      "epoch": 25.12,
      "grad_norm": 6.374452254931384e-07,
      "learning_rate": 8.160535117056857e-07,
      "loss": 0.0,
      "step": 251200
    },
    {
      "epoch": 25.13,
      "grad_norm": 3.375598680577241e-06,
      "learning_rate": 8.143812709030101e-07,
      "loss": 0.0,
      "step": 251300
    },
    {
      "epoch": 25.14,
      "grad_norm": 1.8560887156127137e-06,
      "learning_rate": 8.127090301003346e-07,
      "loss": 0.0,
      "step": 251400
    },
    {
      "epoch": 25.15,
      "grad_norm": 1.5140713003347628e-06,
      "learning_rate": 8.110367892976589e-07,
      "loss": 0.0,
      "step": 251500
    },
    {
      "epoch": 25.16,
      "grad_norm": 3.200523497071117e-05,
      "learning_rate": 8.093645484949833e-07,
      "loss": 0.0094,
      "step": 251600
    },
    {
      "epoch": 25.17,
      "grad_norm": 1.8809379980666563e-05,
      "learning_rate": 8.076923076923078e-07,
      "loss": 0.0,
      "step": 251700
    },
    {
      "epoch": 25.18,
      "grad_norm": 2.0571696950355545e-06,
      "learning_rate": 8.060200668896322e-07,
      "loss": 0.0,
      "step": 251800
    },
    {
      "epoch": 25.19,
      "grad_norm": 1.0208295861957595e-06,
      "learning_rate": 8.043478260869565e-07,
      "loss": 0.0,
      "step": 251900
    },
    {
      "epoch": 25.2,
      "grad_norm": 2.6802188131114235e-06,
      "learning_rate": 8.02675585284281e-07,
      "loss": 0.0024,
      "step": 252000
    },
    {
      "epoch": 25.21,
      "grad_norm": 2.5915858259395463e-06,
      "learning_rate": 8.010033444816054e-07,
      "loss": 0.0065,
      "step": 252100
    },
    {
      "epoch": 25.22,
      "grad_norm": 1.4809822914685355e-06,
      "learning_rate": 7.993311036789299e-07,
      "loss": 0.0,
      "step": 252200
    },
    {
      "epoch": 25.23,
      "grad_norm": 1.2989927427042858e-06,
      "learning_rate": 7.976588628762543e-07,
      "loss": 0.0,
      "step": 252300
    },
    {
      "epoch": 25.24,
      "grad_norm": 1.4638616221418488e-06,
      "learning_rate": 7.959866220735786e-07,
      "loss": 0.0,
      "step": 252400
    },
    {
      "epoch": 25.25,
      "grad_norm": 3.649939117167378e-06,
      "learning_rate": 7.943143812709031e-07,
      "loss": 0.0023,
      "step": 252500
    },
    {
      "epoch": 25.26,
      "grad_norm": 1.3916521766077494e-06,
      "learning_rate": 7.926421404682274e-07,
      "loss": 0.0,
      "step": 252600
    },
    {
      "epoch": 25.27,
      "grad_norm": 1.5946108078424004e-06,
      "learning_rate": 7.90969899665552e-07,
      "loss": 0.0,
      "step": 252700
    },
    {
      "epoch": 25.28,
      "grad_norm": 1.121036348195048e-05,
      "learning_rate": 7.892976588628764e-07,
      "loss": 0.0,
      "step": 252800
    },
    {
      "epoch": 25.29,
      "grad_norm": 1.0152185723200091e-06,
      "learning_rate": 7.876254180602007e-07,
      "loss": 0.0,
      "step": 252900
    },
    {
      "epoch": 25.3,
      "grad_norm": 3.492725227260962e-05,
      "learning_rate": 7.859531772575252e-07,
      "loss": 0.0,
      "step": 253000
    },
    {
      "epoch": 25.31,
      "grad_norm": 2.6968546080752276e-06,
      "learning_rate": 7.842809364548495e-07,
      "loss": 0.0,
      "step": 253100
    },
    {
      "epoch": 25.32,
      "grad_norm": 1.7197486386066885e-06,
      "learning_rate": 7.82608695652174e-07,
      "loss": 0.0,
      "step": 253200
    },
    {
      "epoch": 25.33,
      "grad_norm": 1.6494540204803343e-06,
      "learning_rate": 7.809364548494983e-07,
      "loss": 0.0,
      "step": 253300
    },
    {
      "epoch": 25.34,
      "grad_norm": 9.803585498957545e-07,
      "learning_rate": 7.792642140468228e-07,
      "loss": 0.0,
      "step": 253400
    },
    {
      "epoch": 25.35,
      "grad_norm": 8.82857307260565e-07,
      "learning_rate": 7.775919732441473e-07,
      "loss": 0.0,
      "step": 253500
    },
    {
      "epoch": 25.36,
      "grad_norm": 2.8013680548610864e-06,
      "learning_rate": 7.759197324414716e-07,
      "loss": 0.0061,
      "step": 253600
    },
    {
      "epoch": 25.37,
      "grad_norm": 2.4795892841211753e-06,
      "learning_rate": 7.742474916387961e-07,
      "loss": 0.0,
      "step": 253700
    },
    {
      "epoch": 25.38,
      "grad_norm": 3.668871158879483e-06,
      "learning_rate": 7.725752508361204e-07,
      "loss": 0.0,
      "step": 253800
    },
    {
      "epoch": 25.39,
      "grad_norm": 4.860013632423943e-06,
      "learning_rate": 7.709030100334448e-07,
      "loss": 0.0,
      "step": 253900
    },
    {
      "epoch": 25.4,
      "grad_norm": 1.904245323203213e-06,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.0,
      "step": 254000
    },
    {
      "epoch": 25.41,
      "grad_norm": 2.4638909508212237e-06,
      "learning_rate": 7.675585284280937e-07,
      "loss": 0.0,
      "step": 254100
    },
    {
      "epoch": 25.42,
      "grad_norm": 2.0516115455393447e-06,
      "learning_rate": 7.658862876254181e-07,
      "loss": 0.0,
      "step": 254200
    },
    {
      "epoch": 25.43,
      "grad_norm": 1.3270642966745072e-06,
      "learning_rate": 7.642140468227425e-07,
      "loss": 0.0,
      "step": 254300
    },
    {
      "epoch": 25.44,
      "grad_norm": 1.8419576690575923e-06,
      "learning_rate": 7.625418060200669e-07,
      "loss": 0.0,
      "step": 254400
    },
    {
      "epoch": 25.45,
      "grad_norm": 1.4690833268105052e-05,
      "learning_rate": 7.608695652173914e-07,
      "loss": 0.0,
      "step": 254500
    },
    {
      "epoch": 25.46,
      "grad_norm": 0.00014639816072303802,
      "learning_rate": 7.591973244147158e-07,
      "loss": 0.0,
      "step": 254600
    },
    {
      "epoch": 25.47,
      "grad_norm": 3.1858828606345924e-06,
      "learning_rate": 7.575250836120402e-07,
      "loss": 0.0038,
      "step": 254700
    },
    {
      "epoch": 25.48,
      "grad_norm": 3.5060836580669275e-06,
      "learning_rate": 7.558528428093646e-07,
      "loss": 0.0,
      "step": 254800
    },
    {
      "epoch": 25.49,
      "grad_norm": 4.6844779717503116e-06,
      "learning_rate": 7.54180602006689e-07,
      "loss": 0.005,
      "step": 254900
    },
    {
      "epoch": 25.5,
      "grad_norm": 6.809616479586111e-07,
      "learning_rate": 7.525083612040135e-07,
      "loss": 0.0,
      "step": 255000
    },
    {
      "epoch": 25.5,
      "eval_accuracy": 0.987225,
      "eval_f1": 0.987225,
      "eval_loss": 0.15019439160823822,
      "eval_runtime": 133.4308,
      "eval_samples_per_second": 299.781,
      "eval_steps_per_second": 299.781,
      "step": 255000
    },
    {
      "epoch": 25.51,
      "grad_norm": 3.1307606604968896e-06,
      "learning_rate": 7.508361204013379e-07,
      "loss": 0.0,
      "step": 255100
    },
    {
      "epoch": 25.52,
      "grad_norm": 1.4498136806651019e-06,
      "learning_rate": 7.491638795986622e-07,
      "loss": 0.003,
      "step": 255200
    },
    {
      "epoch": 25.53,
      "grad_norm": 6.193661101860926e-05,
      "learning_rate": 7.474916387959867e-07,
      "loss": 0.0,
      "step": 255300
    },
    {
      "epoch": 25.54,
      "grad_norm": 1.8212824670627015e-06,
      "learning_rate": 7.458193979933111e-07,
      "loss": 0.0,
      "step": 255400
    },
    {
      "epoch": 25.55,
      "grad_norm": 2.457471509842435e-06,
      "learning_rate": 7.441471571906355e-07,
      "loss": 0.0,
      "step": 255500
    },
    {
      "epoch": 25.56,
      "grad_norm": 0.00020605024474207312,
      "learning_rate": 7.424749163879599e-07,
      "loss": 0.0,
      "step": 255600
    },
    {
      "epoch": 25.57,
      "grad_norm": 7.919943527667783e-06,
      "learning_rate": 7.408026755852843e-07,
      "loss": 0.0,
      "step": 255700
    },
    {
      "epoch": 25.58,
      "grad_norm": 9.080182508114376e-07,
      "learning_rate": 7.391304347826088e-07,
      "loss": 0.0,
      "step": 255800
    },
    {
      "epoch": 25.59,
      "grad_norm": 2.7886048883374315e-06,
      "learning_rate": 7.374581939799332e-07,
      "loss": 0.0,
      "step": 255900
    },
    {
      "epoch": 25.6,
      "grad_norm": 1.3184147746869712e-06,
      "learning_rate": 7.357859531772576e-07,
      "loss": 0.0,
      "step": 256000
    },
    {
      "epoch": 25.61,
      "grad_norm": 1.5250210481099202e-06,
      "learning_rate": 7.34113712374582e-07,
      "loss": 0.0,
      "step": 256100
    },
    {
      "epoch": 25.62,
      "grad_norm": 2.6570999125397066e-06,
      "learning_rate": 7.324414715719063e-07,
      "loss": 0.0,
      "step": 256200
    },
    {
      "epoch": 25.63,
      "grad_norm": 1.266855520043464e-06,
      "learning_rate": 7.307692307692309e-07,
      "loss": 0.0,
      "step": 256300
    },
    {
      "epoch": 25.64,
      "grad_norm": 1.032599357131403e-06,
      "learning_rate": 7.290969899665552e-07,
      "loss": 0.0,
      "step": 256400
    },
    {
      "epoch": 25.65,
      "grad_norm": 1.1953579814871773e-06,
      "learning_rate": 7.274247491638796e-07,
      "loss": 0.0,
      "step": 256500
    },
    {
      "epoch": 25.66,
      "grad_norm": 0.000963278638664633,
      "learning_rate": 7.257525083612041e-07,
      "loss": 0.0032,
      "step": 256600
    },
    {
      "epoch": 25.67,
      "grad_norm": 3.835300958598964e-05,
      "learning_rate": 7.240802675585284e-07,
      "loss": 0.0,
      "step": 256700
    },
    {
      "epoch": 25.68,
      "grad_norm": 1.6685755781509215e-06,
      "learning_rate": 7.22408026755853e-07,
      "loss": 0.0,
      "step": 256800
    },
    {
      "epoch": 25.69,
      "grad_norm": 2.4940084131230833e-06,
      "learning_rate": 7.207357859531773e-07,
      "loss": 0.0,
      "step": 256900
    },
    {
      "epoch": 25.7,
      "grad_norm": 1.5732797464806936e-06,
      "learning_rate": 7.190635451505017e-07,
      "loss": 0.0,
      "step": 257000
    },
    {
      "epoch": 25.71,
      "grad_norm": 9.188497642753646e-07,
      "learning_rate": 7.173913043478262e-07,
      "loss": 0.0,
      "step": 257100
    },
    {
      "epoch": 25.72,
      "grad_norm": 0.00019525196694303304,
      "learning_rate": 7.157190635451505e-07,
      "loss": 0.0,
      "step": 257200
    },
    {
      "epoch": 25.73,
      "grad_norm": 9.72332486526284e-07,
      "learning_rate": 7.140468227424751e-07,
      "loss": 0.0,
      "step": 257300
    },
    {
      "epoch": 25.74,
      "grad_norm": 2.251427076771506e-06,
      "learning_rate": 7.123745819397994e-07,
      "loss": 0.0,
      "step": 257400
    },
    {
      "epoch": 25.75,
      "grad_norm": 4.722788162325742e-06,
      "learning_rate": 7.107023411371238e-07,
      "loss": 0.0,
      "step": 257500
    },
    {
      "epoch": 25.76,
      "grad_norm": 4.204017841402674e-06,
      "learning_rate": 7.090301003344482e-07,
      "loss": 0.0,
      "step": 257600
    },
    {
      "epoch": 25.77,
      "grad_norm": 1.0709164826039341e-06,
      "learning_rate": 7.073578595317726e-07,
      "loss": 0.0,
      "step": 257700
    },
    {
      "epoch": 25.78,
      "grad_norm": 1.1540734021764365e-06,
      "learning_rate": 7.056856187290971e-07,
      "loss": 0.0,
      "step": 257800
    },
    {
      "epoch": 25.79,
      "grad_norm": 9.432333172298968e-07,
      "learning_rate": 7.040133779264215e-07,
      "loss": 0.0,
      "step": 257900
    },
    {
      "epoch": 25.8,
      "grad_norm": 8.946111051955086e-07,
      "learning_rate": 7.023411371237459e-07,
      "loss": 0.0,
      "step": 258000
    },
    {
      "epoch": 25.81,
      "grad_norm": 2.140501010217122e-06,
      "learning_rate": 7.006688963210703e-07,
      "loss": 0.0,
      "step": 258100
    },
    {
      "epoch": 25.82,
      "grad_norm": 9.419418915967981e-07,
      "learning_rate": 6.989966555183947e-07,
      "loss": 0.0,
      "step": 258200
    },
    {
      "epoch": 25.83,
      "grad_norm": 2.9447837732732296e-05,
      "learning_rate": 6.973244147157191e-07,
      "loss": 0.0,
      "step": 258300
    },
    {
      "epoch": 25.84,
      "grad_norm": 2.420624014121131e-06,
      "learning_rate": 6.956521739130435e-07,
      "loss": 0.0,
      "step": 258400
    },
    {
      "epoch": 25.85,
      "grad_norm": 0.011331389658153057,
      "learning_rate": 6.939799331103679e-07,
      "loss": 0.0026,
      "step": 258500
    },
    {
      "epoch": 25.86,
      "grad_norm": 5.007507934351452e-05,
      "learning_rate": 6.923076923076924e-07,
      "loss": 0.0,
      "step": 258600
    },
    {
      "epoch": 25.87,
      "grad_norm": 5.7870420278050005e-06,
      "learning_rate": 6.906354515050168e-07,
      "loss": 0.0042,
      "step": 258700
    },
    {
      "epoch": 25.88,
      "grad_norm": 0.0006274498300626874,
      "learning_rate": 6.889632107023412e-07,
      "loss": 0.0,
      "step": 258800
    },
    {
      "epoch": 25.89,
      "grad_norm": 1.82863800546329e-06,
      "learning_rate": 6.872909698996656e-07,
      "loss": 0.0,
      "step": 258900
    },
    {
      "epoch": 25.9,
      "grad_norm": 1.6336121007043403e-06,
      "learning_rate": 6.8561872909699e-07,
      "loss": 0.0076,
      "step": 259000
    },
    {
      "epoch": 25.91,
      "grad_norm": 4.9206792027689517e-05,
      "learning_rate": 6.839464882943145e-07,
      "loss": 0.0,
      "step": 259100
    },
    {
      "epoch": 25.92,
      "grad_norm": 1.827681217037025e-06,
      "learning_rate": 6.822742474916389e-07,
      "loss": 0.0,
      "step": 259200
    },
    {
      "epoch": 25.93,
      "grad_norm": 1.8825265897248755e-06,
      "learning_rate": 6.806020066889632e-07,
      "loss": 0.0,
      "step": 259300
    },
    {
      "epoch": 25.94,
      "grad_norm": 1.943802317327936e-06,
      "learning_rate": 6.789297658862877e-07,
      "loss": 0.0,
      "step": 259400
    },
    {
      "epoch": 25.95,
      "grad_norm": 2.9297807486727834e-05,
      "learning_rate": 6.77257525083612e-07,
      "loss": 0.0,
      "step": 259500
    },
    {
      "epoch": 25.96,
      "grad_norm": 6.956246943445876e-05,
      "learning_rate": 6.755852842809366e-07,
      "loss": 0.0,
      "step": 259600
    },
    {
      "epoch": 25.97,
      "grad_norm": 5.574890565185342e-06,
      "learning_rate": 6.73913043478261e-07,
      "loss": 0.0,
      "step": 259700
    },
    {
      "epoch": 25.98,
      "grad_norm": 1.5360980114564882e-06,
      "learning_rate": 6.722408026755853e-07,
      "loss": 0.0,
      "step": 259800
    },
    {
      "epoch": 25.99,
      "grad_norm": 0.00014320595073513687,
      "learning_rate": 6.705685618729098e-07,
      "loss": 0.0,
      "step": 259900
    },
    {
      "epoch": 26.0,
      "grad_norm": 2.329884182472597e-06,
      "learning_rate": 6.688963210702341e-07,
      "loss": 0.0,
      "step": 260000
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.987775,
      "eval_f1": 0.987775,
      "eval_loss": 0.14540669322013855,
      "eval_runtime": 128.3816,
      "eval_samples_per_second": 311.571,
      "eval_steps_per_second": 311.571,
      "step": 260000
    },
    {
      "epoch": 26.01,
      "grad_norm": 3.0397443424590165e-06,
      "learning_rate": 6.672240802675586e-07,
      "loss": 0.0,
      "step": 260100
    },
    {
      "epoch": 26.02,
      "grad_norm": 2.670638423296623e-06,
      "learning_rate": 6.65551839464883e-07,
      "loss": 0.0,
      "step": 260200
    },
    {
      "epoch": 26.03,
      "grad_norm": 1.2526317050287616e-06,
      "learning_rate": 6.638795986622074e-07,
      "loss": 0.0,
      "step": 260300
    },
    {
      "epoch": 26.04,
      "grad_norm": 1.997796061914414e-06,
      "learning_rate": 6.622073578595319e-07,
      "loss": 0.0,
      "step": 260400
    },
    {
      "epoch": 26.05,
      "grad_norm": 6.544435677824367e-07,
      "learning_rate": 6.605351170568562e-07,
      "loss": 0.0043,
      "step": 260500
    },
    {
      "epoch": 26.06,
      "grad_norm": 1.0374384373790235e-06,
      "learning_rate": 6.588628762541807e-07,
      "loss": 0.0,
      "step": 260600
    },
    {
      "epoch": 26.07,
      "grad_norm": 1.676651550042152e-06,
      "learning_rate": 6.57190635451505e-07,
      "loss": 0.0,
      "step": 260700
    },
    {
      "epoch": 26.08,
      "grad_norm": 6.39155450699036e-06,
      "learning_rate": 6.555183946488294e-07,
      "loss": 0.0,
      "step": 260800
    },
    {
      "epoch": 26.09,
      "grad_norm": 2.87171360469074e-06,
      "learning_rate": 6.53846153846154e-07,
      "loss": 0.0,
      "step": 260900
    },
    {
      "epoch": 26.1,
      "grad_norm": 1.6348642475350061e-06,
      "learning_rate": 6.521739130434783e-07,
      "loss": 0.0,
      "step": 261000
    },
    {
      "epoch": 26.11,
      "grad_norm": 4.175280992058106e-06,
      "learning_rate": 6.505016722408028e-07,
      "loss": 0.0,
      "step": 261100
    },
    {
      "epoch": 26.12,
      "grad_norm": 0.014059566892683506,
      "learning_rate": 6.488294314381271e-07,
      "loss": 0.0,
      "step": 261200
    },
    {
      "epoch": 26.13,
      "grad_norm": 2.457397158650565e-06,
      "learning_rate": 6.471571906354515e-07,
      "loss": 0.0,
      "step": 261300
    },
    {
      "epoch": 26.14,
      "grad_norm": 2.0873145331279375e-05,
      "learning_rate": 6.45484949832776e-07,
      "loss": 0.0,
      "step": 261400
    },
    {
      "epoch": 26.15,
      "grad_norm": 2.946242375401198e-06,
      "learning_rate": 6.438127090301004e-07,
      "loss": 0.0,
      "step": 261500
    },
    {
      "epoch": 26.16,
      "grad_norm": 5.695550953532802e-06,
      "learning_rate": 6.421404682274248e-07,
      "loss": 0.0,
      "step": 261600
    },
    {
      "epoch": 26.17,
      "grad_norm": 1.6086561345218797e-06,
      "learning_rate": 6.404682274247492e-07,
      "loss": 0.0,
      "step": 261700
    },
    {
      "epoch": 26.18,
      "grad_norm": 3.3722619718901115e-06,
      "learning_rate": 6.387959866220736e-07,
      "loss": 0.0,
      "step": 261800
    },
    {
      "epoch": 26.19,
      "grad_norm": 1.4849182434772956e-06,
      "learning_rate": 6.371237458193981e-07,
      "loss": 0.0,
      "step": 261900
    },
    {
      "epoch": 26.2,
      "grad_norm": 3.5905916320189135e-06,
      "learning_rate": 6.354515050167225e-07,
      "loss": 0.0,
      "step": 262000
    },
    {
      "epoch": 26.21,
      "grad_norm": 1.8669687733563478e-06,
      "learning_rate": 6.337792642140468e-07,
      "loss": 0.0,
      "step": 262100
    },
    {
      "epoch": 26.22,
      "grad_norm": 9.240634426532779e-07,
      "learning_rate": 6.321070234113713e-07,
      "loss": 0.0,
      "step": 262200
    },
    {
      "epoch": 26.23,
      "grad_norm": 8.699660156707978e-07,
      "learning_rate": 6.304347826086957e-07,
      "loss": 0.0,
      "step": 262300
    },
    {
      "epoch": 26.24,
      "grad_norm": 2.213256948380149e-06,
      "learning_rate": 6.287625418060201e-07,
      "loss": 0.0,
      "step": 262400
    },
    {
      "epoch": 26.25,
      "grad_norm": 1.3434569154924247e-06,
      "learning_rate": 6.270903010033446e-07,
      "loss": 0.0,
      "step": 262500
    },
    {
      "epoch": 26.26,
      "grad_norm": 8.861171068019758e-07,
      "learning_rate": 6.254180602006689e-07,
      "loss": 0.0,
      "step": 262600
    },
    {
      "epoch": 26.27,
      "grad_norm": 1.178672050627938e-06,
      "learning_rate": 6.237458193979934e-07,
      "loss": 0.0,
      "step": 262700
    },
    {
      "epoch": 26.28,
      "grad_norm": 1.0737633147073211e-06,
      "learning_rate": 6.220735785953178e-07,
      "loss": 0.0,
      "step": 262800
    },
    {
      "epoch": 26.29,
      "grad_norm": 1.4926870335330022e-06,
      "learning_rate": 6.204013377926422e-07,
      "loss": 0.0,
      "step": 262900
    },
    {
      "epoch": 26.3,
      "grad_norm": 1.8841998326024623e-06,
      "learning_rate": 6.187290969899666e-07,
      "loss": 0.0,
      "step": 263000
    },
    {
      "epoch": 26.31,
      "grad_norm": 2.844284608727321e-06,
      "learning_rate": 6.17056856187291e-07,
      "loss": 0.0,
      "step": 263100
    },
    {
      "epoch": 26.32,
      "grad_norm": 4.620580511982553e-06,
      "learning_rate": 6.153846153846155e-07,
      "loss": 0.0,
      "step": 263200
    },
    {
      "epoch": 26.33,
      "grad_norm": 2.383153059781762e-06,
      "learning_rate": 6.137123745819398e-07,
      "loss": 0.0,
      "step": 263300
    },
    {
      "epoch": 26.34,
      "grad_norm": 1.2054698572683265e-06,
      "learning_rate": 6.120401337792643e-07,
      "loss": 0.0,
      "step": 263400
    },
    {
      "epoch": 26.35,
      "grad_norm": 1.1026223774024402e-06,
      "learning_rate": 6.103678929765887e-07,
      "loss": 0.0,
      "step": 263500
    },
    {
      "epoch": 26.36,
      "grad_norm": 2.2108922621555394e-06,
      "learning_rate": 6.086956521739131e-07,
      "loss": 0.0,
      "step": 263600
    },
    {
      "epoch": 26.37,
      "grad_norm": 9.278629136133532e-07,
      "learning_rate": 6.070234113712375e-07,
      "loss": 0.0,
      "step": 263700
    },
    {
      "epoch": 26.38,
      "grad_norm": 1.3845400417267228e-06,
      "learning_rate": 6.053511705685619e-07,
      "loss": 0.0,
      "step": 263800
    },
    {
      "epoch": 26.39,
      "grad_norm": 6.267321623454336e-06,
      "learning_rate": 6.036789297658863e-07,
      "loss": 0.0,
      "step": 263900
    },
    {
      "epoch": 26.4,
      "grad_norm": 1.303371732319647e-06,
      "learning_rate": 6.020066889632107e-07,
      "loss": 0.0,
      "step": 264000
    },
    {
      "epoch": 26.41,
      "grad_norm": 1.8336708080823882e-06,
      "learning_rate": 6.003344481605352e-07,
      "loss": 0.0,
      "step": 264100
    },
    {
      "epoch": 26.42,
      "grad_norm": 6.330028554657474e-05,
      "learning_rate": 5.986622073578596e-07,
      "loss": 0.0,
      "step": 264200
    },
    {
      "epoch": 26.43,
      "grad_norm": 1.0136543551197974e-06,
      "learning_rate": 5.96989966555184e-07,
      "loss": 0.0,
      "step": 264300
    },
    {
      "epoch": 26.44,
      "grad_norm": 3.786742354350281e-06,
      "learning_rate": 5.953177257525084e-07,
      "loss": 0.0045,
      "step": 264400
    },
    {
      "epoch": 26.45,
      "grad_norm": 1.3409010080067674e-06,
      "learning_rate": 5.936454849498328e-07,
      "loss": 0.0,
      "step": 264500
    },
    {
      "epoch": 26.46,
      "grad_norm": 1.185390374303097e-06,
      "learning_rate": 5.919732441471572e-07,
      "loss": 0.0,
      "step": 264600
    },
    {
      "epoch": 26.47,
      "grad_norm": 1.233201146533247e-05,
      "learning_rate": 5.903010033444817e-07,
      "loss": 0.0036,
      "step": 264700
    },
    {
      "epoch": 26.48,
      "grad_norm": 1.4137769994704286e-06,
      "learning_rate": 5.886287625418061e-07,
      "loss": 0.0,
      "step": 264800
    },
    {
      "epoch": 26.49,
      "grad_norm": 7.992609880602686e-07,
      "learning_rate": 5.869565217391305e-07,
      "loss": 0.0,
      "step": 264900
    },
    {
      "epoch": 26.5,
      "grad_norm": 1.5214644690786372e-06,
      "learning_rate": 5.852842809364549e-07,
      "loss": 0.0,
      "step": 265000
    },
    {
      "epoch": 26.5,
      "eval_accuracy": 0.987925,
      "eval_f1": 0.987925,
      "eval_loss": 0.14723196625709534,
      "eval_runtime": 127.175,
      "eval_samples_per_second": 314.527,
      "eval_steps_per_second": 314.527,
      "step": 265000
    },
    {
      "epoch": 26.51,
      "grad_norm": 8.622855602880009e-06,
      "learning_rate": 5.836120401337793e-07,
      "loss": 0.0,
      "step": 265100
    },
    {
      "epoch": 26.52,
      "grad_norm": 3.604415906011127e-05,
      "learning_rate": 5.819397993311037e-07,
      "loss": 0.0,
      "step": 265200
    },
    {
      "epoch": 26.53,
      "grad_norm": 1.473955080655287e-06,
      "learning_rate": 5.802675585284281e-07,
      "loss": 0.0,
      "step": 265300
    },
    {
      "epoch": 26.54,
      "grad_norm": 2.554420461819973e-06,
      "learning_rate": 5.785953177257526e-07,
      "loss": 0.0,
      "step": 265400
    },
    {
      "epoch": 26.55,
      "grad_norm": 2.2791825813328614e-06,
      "learning_rate": 5.76923076923077e-07,
      "loss": 0.0,
      "step": 265500
    },
    {
      "epoch": 26.56,
      "grad_norm": 1.3585033684648806e-06,
      "learning_rate": 5.752508361204014e-07,
      "loss": 0.0,
      "step": 265600
    },
    {
      "epoch": 26.57,
      "grad_norm": 1.1605886811594246e-06,
      "learning_rate": 5.735785953177258e-07,
      "loss": 0.0,
      "step": 265700
    },
    {
      "epoch": 26.58,
      "grad_norm": 1.446388523618225e-06,
      "learning_rate": 5.719063545150502e-07,
      "loss": 0.0,
      "step": 265800
    },
    {
      "epoch": 26.59,
      "grad_norm": 1.4642024552813382e-06,
      "learning_rate": 5.702341137123746e-07,
      "loss": 0.0,
      "step": 265900
    },
    {
      "epoch": 26.6,
      "grad_norm": 1.8093830931320554e-06,
      "learning_rate": 5.68561872909699e-07,
      "loss": 0.0,
      "step": 266000
    },
    {
      "epoch": 26.61,
      "grad_norm": 2.076187001875951e-06,
      "learning_rate": 5.668896321070235e-07,
      "loss": 0.0,
      "step": 266100
    },
    {
      "epoch": 26.62,
      "grad_norm": 5.8339724091638345e-06,
      "learning_rate": 5.652173913043478e-07,
      "loss": 0.0,
      "step": 266200
    },
    {
      "epoch": 26.63,
      "grad_norm": 1.15121656563133e-06,
      "learning_rate": 5.635451505016723e-07,
      "loss": 0.0,
      "step": 266300
    },
    {
      "epoch": 26.64,
      "grad_norm": 9.865797210295568e-07,
      "learning_rate": 5.618729096989967e-07,
      "loss": 0.0,
      "step": 266400
    },
    {
      "epoch": 26.65,
      "grad_norm": 3.917589765478624e-06,
      "learning_rate": 5.602006688963211e-07,
      "loss": 0.0,
      "step": 266500
    },
    {
      "epoch": 26.66,
      "grad_norm": 5.671467988577206e-07,
      "learning_rate": 5.585284280936456e-07,
      "loss": 0.0,
      "step": 266600
    },
    {
      "epoch": 26.67,
      "grad_norm": 3.5159577237209305e-05,
      "learning_rate": 5.568561872909699e-07,
      "loss": 0.0,
      "step": 266700
    },
    {
      "epoch": 26.68,
      "grad_norm": 9.57158249548229e-07,
      "learning_rate": 5.551839464882944e-07,
      "loss": 0.0054,
      "step": 266800
    },
    {
      "epoch": 26.69,
      "grad_norm": 9.69288748819963e-07,
      "learning_rate": 5.535117056856188e-07,
      "loss": 0.0,
      "step": 266900
    },
    {
      "epoch": 26.7,
      "grad_norm": 6.826439403084805e-06,
      "learning_rate": 5.518394648829432e-07,
      "loss": 0.0,
      "step": 267000
    },
    {
      "epoch": 26.71,
      "grad_norm": 1.5509156128246104e-06,
      "learning_rate": 5.501672240802676e-07,
      "loss": 0.0,
      "step": 267100
    },
    {
      "epoch": 26.72,
      "grad_norm": 1.6871094885573257e-06,
      "learning_rate": 5.48494983277592e-07,
      "loss": 0.0,
      "step": 267200
    },
    {
      "epoch": 26.73,
      "grad_norm": 7.855343255869229e-07,
      "learning_rate": 5.468227424749165e-07,
      "loss": 0.0,
      "step": 267300
    },
    {
      "epoch": 26.74,
      "grad_norm": 7.116402116480458e-07,
      "learning_rate": 5.451505016722408e-07,
      "loss": 0.0047,
      "step": 267400
    },
    {
      "epoch": 26.75,
      "grad_norm": 1.349264834971109e-06,
      "learning_rate": 5.434782608695653e-07,
      "loss": 0.0,
      "step": 267500
    },
    {
      "epoch": 26.76,
      "grad_norm": 1.5048664181449567e-06,
      "learning_rate": 5.418060200668896e-07,
      "loss": 0.0,
      "step": 267600
    },
    {
      "epoch": 26.77,
      "grad_norm": 1.0677230420697015e-05,
      "learning_rate": 5.401337792642141e-07,
      "loss": 0.0,
      "step": 267700
    },
    {
      "epoch": 26.78,
      "grad_norm": 1.0198332347499672e-06,
      "learning_rate": 5.384615384615386e-07,
      "loss": 0.0,
      "step": 267800
    },
    {
      "epoch": 26.79,
      "grad_norm": 1.7846756463768543e-06,
      "learning_rate": 5.367892976588629e-07,
      "loss": 0.0,
      "step": 267900
    },
    {
      "epoch": 26.8,
      "grad_norm": 9.937119784808601e-07,
      "learning_rate": 5.351170568561874e-07,
      "loss": 0.0,
      "step": 268000
    },
    {
      "epoch": 26.81,
      "grad_norm": 1.4857420183034264e-06,
      "learning_rate": 5.334448160535117e-07,
      "loss": 0.0,
      "step": 268100
    },
    {
      "epoch": 26.82,
      "grad_norm": 1.405791294928349e-06,
      "learning_rate": 5.317725752508362e-07,
      "loss": 0.0,
      "step": 268200
    },
    {
      "epoch": 26.83,
      "grad_norm": 1.7296881651418516e-06,
      "learning_rate": 5.301003344481605e-07,
      "loss": 0.0,
      "step": 268300
    },
    {
      "epoch": 26.84,
      "grad_norm": 1.4348109289130662e-05,
      "learning_rate": 5.28428093645485e-07,
      "loss": 0.0,
      "step": 268400
    },
    {
      "epoch": 26.85,
      "grad_norm": 1.8275501361131319e-06,
      "learning_rate": 5.267558528428094e-07,
      "loss": 0.0,
      "step": 268500
    },
    {
      "epoch": 26.86,
      "grad_norm": 1.2868830481238547e-06,
      "learning_rate": 5.250836120401338e-07,
      "loss": 0.0,
      "step": 268600
    },
    {
      "epoch": 26.87,
      "grad_norm": 1.6959703543761862e-06,
      "learning_rate": 5.234113712374583e-07,
      "loss": 0.0,
      "step": 268700
    },
    {
      "epoch": 26.88,
      "grad_norm": 1.57480280904565e-06,
      "learning_rate": 5.217391304347826e-07,
      "loss": 0.0,
      "step": 268800
    },
    {
      "epoch": 26.89,
      "grad_norm": 1.4843112694507e-06,
      "learning_rate": 5.200668896321071e-07,
      "loss": 0.0015,
      "step": 268900
    },
    {
      "epoch": 26.9,
      "grad_norm": 1.2535028872662224e-06,
      "learning_rate": 5.183946488294314e-07,
      "loss": 0.0,
      "step": 269000
    },
    {
      "epoch": 26.91,
      "grad_norm": 0.00023764085199218243,
      "learning_rate": 5.167224080267559e-07,
      "loss": 0.0,
      "step": 269100
    },
    {
      "epoch": 26.92,
      "grad_norm": 0.00029709271620959044,
      "learning_rate": 5.150501672240804e-07,
      "loss": 0.0,
      "step": 269200
    },
    {
      "epoch": 26.93,
      "grad_norm": 7.498289846807893e-07,
      "learning_rate": 5.133779264214047e-07,
      "loss": 0.0,
      "step": 269300
    },
    {
      "epoch": 26.94,
      "grad_norm": 1.443891278540832e-06,
      "learning_rate": 5.117056856187292e-07,
      "loss": 0.0,
      "step": 269400
    },
    {
      "epoch": 26.95,
      "grad_norm": 9.474859689362347e-05,
      "learning_rate": 5.100334448160535e-07,
      "loss": 0.0,
      "step": 269500
    },
    {
      "epoch": 26.96,
      "grad_norm": 1.0035858394985553e-05,
      "learning_rate": 5.08361204013378e-07,
      "loss": 0.0,
      "step": 269600
    },
    {
      "epoch": 26.97,
      "grad_norm": 2.1819903395225992e-06,
      "learning_rate": 5.066889632107023e-07,
      "loss": 0.0034,
      "step": 269700
    },
    {
      "epoch": 26.98,
      "grad_norm": 3.6087269108975306e-05,
      "learning_rate": 5.050167224080268e-07,
      "loss": 0.0,
      "step": 269800
    },
    {
      "epoch": 26.99,
      "grad_norm": 9.546632782075903e-07,
      "learning_rate": 5.033444816053512e-07,
      "loss": 0.0,
      "step": 269900
    },
    {
      "epoch": 27.0,
      "grad_norm": 3.2282516713166842e-06,
      "learning_rate": 5.016722408026756e-07,
      "loss": 0.0,
      "step": 270000
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.988125,
      "eval_f1": 0.988125,
      "eval_loss": 0.1494733989238739,
      "eval_runtime": 128.4859,
      "eval_samples_per_second": 311.318,
      "eval_steps_per_second": 311.318,
      "step": 270000
    },
    {
      "epoch": 27.01,
      "grad_norm": 6.7214832597528584e-06,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0,
      "step": 270100
    },
    {
      "epoch": 27.02,
      "grad_norm": 1.6825376860651886e-06,
      "learning_rate": 4.983277591973244e-07,
      "loss": 0.0,
      "step": 270200
    },
    {
      "epoch": 27.03,
      "grad_norm": 3.110191073574242e-06,
      "learning_rate": 4.966555183946489e-07,
      "loss": 0.0,
      "step": 270300
    },
    {
      "epoch": 27.04,
      "grad_norm": 2.0460465748328716e-06,
      "learning_rate": 4.949832775919733e-07,
      "loss": 0.0,
      "step": 270400
    },
    {
      "epoch": 27.05,
      "grad_norm": 1.6207150110858493e-06,
      "learning_rate": 4.933110367892977e-07,
      "loss": 0.0,
      "step": 270500
    },
    {
      "epoch": 27.06,
      "grad_norm": 1.8051798633678118e-06,
      "learning_rate": 4.916387959866222e-07,
      "loss": 0.0,
      "step": 270600
    },
    {
      "epoch": 27.07,
      "grad_norm": 4.891130629403051e-06,
      "learning_rate": 4.899665551839465e-07,
      "loss": 0.0,
      "step": 270700
    },
    {
      "epoch": 27.08,
      "grad_norm": 1.253480604646029e-06,
      "learning_rate": 4.88294314381271e-07,
      "loss": 0.0,
      "step": 270800
    },
    {
      "epoch": 27.09,
      "grad_norm": 1.683352252257464e-06,
      "learning_rate": 4.866220735785953e-07,
      "loss": 0.0,
      "step": 270900
    },
    {
      "epoch": 27.1,
      "grad_norm": 5.776287434855476e-05,
      "learning_rate": 4.849498327759198e-07,
      "loss": 0.0,
      "step": 271000
    },
    {
      "epoch": 27.11,
      "grad_norm": 5.219275863055373e-06,
      "learning_rate": 4.832775919732442e-07,
      "loss": 0.0,
      "step": 271100
    },
    {
      "epoch": 27.12,
      "grad_norm": 1.6335097825503908e-05,
      "learning_rate": 4.816053511705686e-07,
      "loss": 0.0,
      "step": 271200
    },
    {
      "epoch": 27.13,
      "grad_norm": 1.0641881544870557e-06,
      "learning_rate": 4.79933110367893e-07,
      "loss": 0.0,
      "step": 271300
    },
    {
      "epoch": 27.14,
      "grad_norm": 9.760987040863256e-07,
      "learning_rate": 4.782608695652174e-07,
      "loss": 0.0,
      "step": 271400
    },
    {
      "epoch": 27.15,
      "grad_norm": 1.0345853524995618e-06,
      "learning_rate": 4.765886287625419e-07,
      "loss": 0.0,
      "step": 271500
    },
    {
      "epoch": 27.16,
      "grad_norm": 8.688991783856181e-07,
      "learning_rate": 4.7491638795986625e-07,
      "loss": 0.0,
      "step": 271600
    },
    {
      "epoch": 27.17,
      "grad_norm": 1.5890523172856774e-06,
      "learning_rate": 4.7324414715719066e-07,
      "loss": 0.0,
      "step": 271700
    },
    {
      "epoch": 27.18,
      "grad_norm": 0.00010488680709386244,
      "learning_rate": 4.715719063545151e-07,
      "loss": 0.0,
      "step": 271800
    },
    {
      "epoch": 27.19,
      "grad_norm": 2.86882800537569e-06,
      "learning_rate": 4.698996655518395e-07,
      "loss": 0.0,
      "step": 271900
    },
    {
      "epoch": 27.2,
      "grad_norm": 1.1794188139901962e-05,
      "learning_rate": 4.682274247491639e-07,
      "loss": 0.0,
      "step": 272000
    },
    {
      "epoch": 27.21,
      "grad_norm": 1.7052036582754226e-06,
      "learning_rate": 4.665551839464883e-07,
      "loss": 0.0,
      "step": 272100
    },
    {
      "epoch": 27.22,
      "grad_norm": 8.556891657462984e-07,
      "learning_rate": 4.6488294314381275e-07,
      "loss": 0.0,
      "step": 272200
    },
    {
      "epoch": 27.23,
      "grad_norm": 1.2632311836568988e-06,
      "learning_rate": 4.6321070234113715e-07,
      "loss": 0.0,
      "step": 272300
    },
    {
      "epoch": 27.24,
      "grad_norm": 1.4065376490179915e-06,
      "learning_rate": 4.615384615384616e-07,
      "loss": 0.0,
      "step": 272400
    },
    {
      "epoch": 27.25,
      "grad_norm": 3.707805080921389e-05,
      "learning_rate": 4.5986622073578597e-07,
      "loss": 0.0,
      "step": 272500
    },
    {
      "epoch": 27.26,
      "grad_norm": 6.321478394966107e-06,
      "learning_rate": 4.581939799331104e-07,
      "loss": 0.0,
      "step": 272600
    },
    {
      "epoch": 27.27,
      "grad_norm": 1.3622416190628428e-06,
      "learning_rate": 4.5652173913043484e-07,
      "loss": 0.0,
      "step": 272700
    },
    {
      "epoch": 27.28,
      "grad_norm": 1.52013853949029e-06,
      "learning_rate": 4.5484949832775925e-07,
      "loss": 0.0,
      "step": 272800
    },
    {
      "epoch": 27.29,
      "grad_norm": 6.5507720137247816e-06,
      "learning_rate": 4.531772575250837e-07,
      "loss": 0.0,
      "step": 272900
    },
    {
      "epoch": 27.3,
      "grad_norm": 9.356578630104195e-06,
      "learning_rate": 4.5150501672240806e-07,
      "loss": 0.0,
      "step": 273000
    },
    {
      "epoch": 27.31,
      "grad_norm": 1.1886962738572038e-06,
      "learning_rate": 4.4983277591973247e-07,
      "loss": 0.0,
      "step": 273100
    },
    {
      "epoch": 27.32,
      "grad_norm": 6.281496098381467e-06,
      "learning_rate": 4.481605351170569e-07,
      "loss": 0.0,
      "step": 273200
    },
    {
      "epoch": 27.33,
      "grad_norm": 9.167351890937425e-07,
      "learning_rate": 4.4648829431438134e-07,
      "loss": 0.0,
      "step": 273300
    },
    {
      "epoch": 27.34,
      "grad_norm": 6.569893571395369e-07,
      "learning_rate": 4.448160535117057e-07,
      "loss": 0.0,
      "step": 273400
    },
    {
      "epoch": 27.35,
      "grad_norm": 0.00026023961254395545,
      "learning_rate": 4.431438127090301e-07,
      "loss": 0.0,
      "step": 273500
    },
    {
      "epoch": 27.36,
      "grad_norm": 1.3729372767556924e-06,
      "learning_rate": 4.4147157190635456e-07,
      "loss": 0.0,
      "step": 273600
    },
    {
      "epoch": 27.37,
      "grad_norm": 2.290072643518215e-06,
      "learning_rate": 4.3979933110367897e-07,
      "loss": 0.0,
      "step": 273700
    },
    {
      "epoch": 27.38,
      "grad_norm": 1.0271631936120684e-06,
      "learning_rate": 4.3812709030100343e-07,
      "loss": 0.0,
      "step": 273800
    },
    {
      "epoch": 27.39,
      "grad_norm": 7.632092433595972e-07,
      "learning_rate": 4.364548494983278e-07,
      "loss": 0.0,
      "step": 273900
    },
    {
      "epoch": 27.4,
      "grad_norm": 2.6731247544375947e-06,
      "learning_rate": 4.347826086956522e-07,
      "loss": 0.0,
      "step": 274000
    },
    {
      "epoch": 27.41,
      "grad_norm": 1.3995519111631438e-05,
      "learning_rate": 4.3311036789297665e-07,
      "loss": 0.0,
      "step": 274100
    },
    {
      "epoch": 27.42,
      "grad_norm": 1.2089567462680861e-06,
      "learning_rate": 4.3143812709030106e-07,
      "loss": 0.0,
      "step": 274200
    },
    {
      "epoch": 27.43,
      "grad_norm": 1.994718104469939e-06,
      "learning_rate": 4.2976588628762546e-07,
      "loss": 0.0038,
      "step": 274300
    },
    {
      "epoch": 27.44,
      "grad_norm": 1.106251716009865e-06,
      "learning_rate": 4.2809364548494987e-07,
      "loss": 0.0,
      "step": 274400
    },
    {
      "epoch": 27.45,
      "grad_norm": 0.00026735287974588573,
      "learning_rate": 4.264214046822743e-07,
      "loss": 0.0,
      "step": 274500
    },
    {
      "epoch": 27.46,
      "grad_norm": 2.463541704855743e-06,
      "learning_rate": 4.247491638795987e-07,
      "loss": 0.0,
      "step": 274600
    },
    {
      "epoch": 27.47,
      "grad_norm": 4.47788579549524e-06,
      "learning_rate": 4.2307692307692315e-07,
      "loss": 0.0031,
      "step": 274700
    },
    {
      "epoch": 27.48,
      "grad_norm": 5.374798206503328e-07,
      "learning_rate": 4.214046822742475e-07,
      "loss": 0.0,
      "step": 274800
    },
    {
      "epoch": 27.49,
      "grad_norm": 1.548745672153018e-06,
      "learning_rate": 4.197324414715719e-07,
      "loss": 0.0,
      "step": 274900
    },
    {
      "epoch": 27.5,
      "grad_norm": 6.070744120734162e-07,
      "learning_rate": 4.1806020066889637e-07,
      "loss": 0.0004,
      "step": 275000
    },
    {
      "epoch": 27.5,
      "eval_accuracy": 0.98815,
      "eval_f1": 0.98815,
      "eval_loss": 0.15360614657402039,
      "eval_runtime": 127.9472,
      "eval_samples_per_second": 312.629,
      "eval_steps_per_second": 312.629,
      "step": 275000
    }
  ],
  "logging_steps": 100,
  "max_steps": 300000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 30,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.865769415769616e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}