{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 4649,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00021510002151000216,
      "grad_norm": 0.0,
      "learning_rate": 1.4285714285714287e-07,
      "loss": 1.7358,
      "step": 1
    },
    {
      "epoch": 0.00043020004302000433,
      "grad_norm": 0.0,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 1.7069,
      "step": 2
    },
    {
      "epoch": 0.0006453000645300065,
      "grad_norm": 0.0,
      "learning_rate": 4.285714285714286e-07,
      "loss": 1.7119,
      "step": 3
    },
    {
      "epoch": 0.0008604000860400087,
      "grad_norm": 0.0,
      "learning_rate": 5.714285714285715e-07,
      "loss": 1.6695,
      "step": 4
    },
    {
      "epoch": 0.0010755001075500108,
      "grad_norm": 0.0,
      "learning_rate": 7.142857142857143e-07,
      "loss": 1.6631,
      "step": 5
    },
    {
      "epoch": 0.001290600129060013,
      "grad_norm": 0.0,
      "learning_rate": 8.571428571428572e-07,
      "loss": 1.7013,
      "step": 6
    },
    {
      "epoch": 0.001505700150570015,
      "grad_norm": 0.0,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.649,
      "step": 7
    },
    {
      "epoch": 0.0017208001720800173,
      "grad_norm": 0.0,
      "learning_rate": 1.142857142857143e-06,
      "loss": 1.431,
      "step": 8
    },
    {
      "epoch": 0.0019359001935900194,
      "grad_norm": 0.0,
      "learning_rate": 1.2857142857142856e-06,
      "loss": 1.4435,
      "step": 9
    },
    {
      "epoch": 0.0021510002151000217,
      "grad_norm": 0.0,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 1.3671,
      "step": 10
    },
    {
      "epoch": 0.0023661002366100236,
      "grad_norm": 0.0,
      "learning_rate": 1.5714285714285714e-06,
      "loss": 1.3399,
      "step": 11
    },
    {
      "epoch": 0.002581200258120026,
      "grad_norm": 0.0,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 1.3014,
      "step": 12
    },
    {
      "epoch": 0.002796300279630028,
      "grad_norm": 0.0,
      "learning_rate": 1.8571428571428573e-06,
      "loss": 1.3198,
      "step": 13
    },
    {
      "epoch": 0.00301140030114003,
      "grad_norm": 0.0,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.3075,
      "step": 14
    },
    {
      "epoch": 0.0032265003226500323,
      "grad_norm": 0.0,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 1.3019,
      "step": 15
    },
    {
      "epoch": 0.0034416003441600346,
      "grad_norm": 0.0,
      "learning_rate": 2.285714285714286e-06,
      "loss": 1.3052,
      "step": 16
    },
    {
      "epoch": 0.0036567003656700365,
      "grad_norm": 0.0,
      "learning_rate": 2.428571428571429e-06,
      "loss": 1.1353,
      "step": 17
    },
    {
      "epoch": 0.003871800387180039,
      "grad_norm": 0.0,
      "learning_rate": 2.571428571428571e-06,
      "loss": 1.1694,
      "step": 18
    },
    {
      "epoch": 0.004086900408690041,
      "grad_norm": 0.0,
      "learning_rate": 2.7142857142857144e-06,
      "loss": 1.2111,
      "step": 19
    },
    {
      "epoch": 0.004302000430200043,
      "grad_norm": 0.0,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 1.1543,
      "step": 20
    },
    {
      "epoch": 0.004517100451710045,
      "grad_norm": 0.0,
      "learning_rate": 3e-06,
      "loss": 1.1976,
      "step": 21
    },
    {
      "epoch": 0.004732200473220047,
      "grad_norm": 0.0,
      "learning_rate": 3.142857142857143e-06,
      "loss": 1.1319,
      "step": 22
    },
    {
      "epoch": 0.004947300494730049,
      "grad_norm": 0.0,
      "learning_rate": 3.285714285714286e-06,
      "loss": 1.1672,
      "step": 23
    },
    {
      "epoch": 0.005162400516240052,
      "grad_norm": 0.0,
      "learning_rate": 3.428571428571429e-06,
      "loss": 1.1673,
      "step": 24
    },
    {
      "epoch": 0.005377500537750054,
      "grad_norm": 0.0,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 1.1228,
      "step": 25
    },
    {
      "epoch": 0.005592600559260056,
      "grad_norm": 0.0,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 1.1347,
      "step": 26
    },
    {
      "epoch": 0.005807700580770058,
      "grad_norm": 0.0,
      "learning_rate": 3.857142857142858e-06,
      "loss": 1.1153,
      "step": 27
    },
    {
      "epoch": 0.00602280060228006,
      "grad_norm": 0.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.1935,
      "step": 28
    },
    {
      "epoch": 0.006237900623790062,
      "grad_norm": 0.0,
      "learning_rate": 4.1428571428571435e-06,
      "loss": 1.1182,
      "step": 29
    },
    {
      "epoch": 0.006453000645300065,
      "grad_norm": 0.0,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 1.1337,
      "step": 30
    },
    {
      "epoch": 0.006668100666810067,
      "grad_norm": 0.0,
      "learning_rate": 4.428571428571429e-06,
      "loss": 0.9977,
      "step": 31
    },
    {
      "epoch": 0.006883200688320069,
      "grad_norm": 0.0,
      "learning_rate": 4.571428571428572e-06,
      "loss": 1.0817,
      "step": 32
    },
    {
      "epoch": 0.007098300709830071,
      "grad_norm": 0.0,
      "learning_rate": 4.714285714285715e-06,
      "loss": 1.0712,
      "step": 33
    },
    {
      "epoch": 0.007313400731340073,
      "grad_norm": 0.0,
      "learning_rate": 4.857142857142858e-06,
      "loss": 1.0085,
      "step": 34
    },
    {
      "epoch": 0.007528500752850075,
      "grad_norm": 0.0,
      "learning_rate": 5e-06,
      "loss": 1.1077,
      "step": 35
    },
    {
      "epoch": 0.007743600774360078,
      "grad_norm": 0.0,
      "learning_rate": 5.142857142857142e-06,
      "loss": 1.0453,
      "step": 36
    },
    {
      "epoch": 0.007958700795870079,
      "grad_norm": 0.0,
      "learning_rate": 5.285714285714286e-06,
      "loss": 1.101,
      "step": 37
    },
    {
      "epoch": 0.008173800817380082,
      "grad_norm": 0.0,
      "learning_rate": 5.428571428571429e-06,
      "loss": 1.0156,
      "step": 38
    },
    {
      "epoch": 0.008388900838890084,
      "grad_norm": 0.0,
      "learning_rate": 5.571428571428572e-06,
      "loss": 1.0343,
      "step": 39
    },
    {
      "epoch": 0.008604000860400087,
      "grad_norm": 0.0,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.0585,
      "step": 40
    },
    {
      "epoch": 0.008819100881910088,
      "grad_norm": 0.0,
      "learning_rate": 5.857142857142858e-06,
      "loss": 1.0678,
      "step": 41
    },
    {
      "epoch": 0.00903420090342009,
      "grad_norm": 0.0,
      "learning_rate": 6e-06,
      "loss": 1.0106,
      "step": 42
    },
    {
      "epoch": 0.009249300924930093,
      "grad_norm": 0.0,
      "learning_rate": 6.142857142857144e-06,
      "loss": 1.1114,
      "step": 43
    },
    {
      "epoch": 0.009464400946440094,
      "grad_norm": 0.0,
      "learning_rate": 6.285714285714286e-06,
      "loss": 1.0184,
      "step": 44
    },
    {
      "epoch": 0.009679500967950097,
      "grad_norm": 0.0,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 1.0557,
      "step": 45
    },
    {
      "epoch": 0.009894600989460099,
      "grad_norm": 0.0,
      "learning_rate": 6.571428571428572e-06,
      "loss": 1.0479,
      "step": 46
    },
    {
      "epoch": 0.0101097010109701,
      "grad_norm": 0.0,
      "learning_rate": 6.714285714285714e-06,
      "loss": 1.048,
      "step": 47
    },
    {
      "epoch": 0.010324801032480103,
      "grad_norm": 0.0,
      "learning_rate": 6.857142857142858e-06,
      "loss": 1.0314,
      "step": 48
    },
    {
      "epoch": 0.010539901053990105,
      "grad_norm": 0.0,
      "learning_rate": 7e-06,
      "loss": 1.0827,
      "step": 49
    },
    {
      "epoch": 0.010755001075500108,
      "grad_norm": 0.0,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 1.0486,
      "step": 50
    },
    {
      "epoch": 0.01097010109701011,
      "grad_norm": 0.0,
      "learning_rate": 7.285714285714286e-06,
      "loss": 1.0477,
      "step": 51
    },
    {
      "epoch": 0.011185201118520113,
      "grad_norm": 0.0,
      "learning_rate": 7.428571428571429e-06,
      "loss": 1.0078,
      "step": 52
    },
    {
      "epoch": 0.011400301140030114,
      "grad_norm": 0.0,
      "learning_rate": 7.571428571428572e-06,
      "loss": 1.0178,
      "step": 53
    },
    {
      "epoch": 0.011615401161540116,
      "grad_norm": 0.0,
      "learning_rate": 7.714285714285716e-06,
      "loss": 0.9819,
      "step": 54
    },
    {
      "epoch": 0.011830501183050119,
      "grad_norm": 0.0,
      "learning_rate": 7.857142857142858e-06,
      "loss": 1.0593,
      "step": 55
    },
    {
      "epoch": 0.01204560120456012,
      "grad_norm": 0.0,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.0313,
      "step": 56
    },
    {
      "epoch": 0.012260701226070123,
      "grad_norm": 0.0,
      "learning_rate": 8.142857142857143e-06,
      "loss": 1.0131,
      "step": 57
    },
    {
      "epoch": 0.012475801247580125,
      "grad_norm": 0.0,
      "learning_rate": 8.285714285714287e-06,
      "loss": 0.9946,
      "step": 58
    },
    {
      "epoch": 0.012690901269090126,
      "grad_norm": 0.0,
      "learning_rate": 8.428571428571429e-06,
      "loss": 1.0208,
      "step": 59
    },
    {
      "epoch": 0.01290600129060013,
      "grad_norm": 0.0,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.0307,
      "step": 60
    },
    {
      "epoch": 0.01312110131211013,
      "grad_norm": 0.0,
      "learning_rate": 8.714285714285715e-06,
      "loss": 0.9715,
      "step": 61
    },
    {
      "epoch": 0.013336201333620134,
      "grad_norm": 0.0,
      "learning_rate": 8.857142857142858e-06,
      "loss": 0.9103,
      "step": 62
    },
    {
      "epoch": 0.013551301355130135,
      "grad_norm": 0.0,
      "learning_rate": 9e-06,
      "loss": 1.093,
      "step": 63
    },
    {
      "epoch": 0.013766401376640139,
      "grad_norm": 0.0,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.9808,
      "step": 64
    },
    {
      "epoch": 0.01398150139815014,
      "grad_norm": 0.0,
      "learning_rate": 9.285714285714288e-06,
      "loss": 1.0582,
      "step": 65
    },
    {
      "epoch": 0.014196601419660141,
      "grad_norm": 0.0,
      "learning_rate": 9.42857142857143e-06,
      "loss": 0.975,
      "step": 66
    },
    {
      "epoch": 0.014411701441170145,
      "grad_norm": 0.0,
      "learning_rate": 9.571428571428573e-06,
      "loss": 0.97,
      "step": 67
    },
    {
      "epoch": 0.014626801462680146,
      "grad_norm": 0.0,
      "learning_rate": 9.714285714285715e-06,
      "loss": 1.0694,
      "step": 68
    },
    {
      "epoch": 0.01484190148419015,
      "grad_norm": 0.0,
      "learning_rate": 9.857142857142859e-06,
      "loss": 1.0238,
      "step": 69
    },
    {
      "epoch": 0.01505700150570015,
      "grad_norm": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.0397,
      "step": 70
    },
    {
      "epoch": 0.015272101527210152,
      "grad_norm": 0.0,
      "learning_rate": 1.0142857142857143e-05,
      "loss": 1.0694,
      "step": 71
    },
    {
      "epoch": 0.015487201548720155,
      "grad_norm": 0.0,
      "learning_rate": 1.0285714285714285e-05,
      "loss": 0.9487,
      "step": 72
    },
    {
      "epoch": 0.015702301570230157,
      "grad_norm": 0.0,
      "learning_rate": 1.042857142857143e-05,
      "loss": 0.9332,
      "step": 73
    },
    {
      "epoch": 0.015917401591740158,
      "grad_norm": 0.0,
      "learning_rate": 1.0571428571428572e-05,
      "loss": 1.0271,
      "step": 74
    },
    {
      "epoch": 0.016132501613250163,
      "grad_norm": 0.0,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 1.0254,
      "step": 75
    },
    {
      "epoch": 0.016347601634760164,
      "grad_norm": 0.0,
      "learning_rate": 1.0857142857142858e-05,
      "loss": 0.9678,
      "step": 76
    },
    {
      "epoch": 0.016562701656270166,
      "grad_norm": 0.0,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 1.0253,
      "step": 77
    },
    {
      "epoch": 0.016777801677780167,
      "grad_norm": 0.0,
      "learning_rate": 1.1142857142857143e-05,
      "loss": 1.0322,
      "step": 78
    },
    {
      "epoch": 0.01699290169929017,
      "grad_norm": 0.0,
      "learning_rate": 1.1285714285714287e-05,
      "loss": 0.9864,
      "step": 79
    },
    {
      "epoch": 0.017208001720800174,
      "grad_norm": 0.0,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 1.017,
      "step": 80
    },
    {
      "epoch": 0.017423101742310175,
      "grad_norm": 0.0,
      "learning_rate": 1.1571428571428573e-05,
      "loss": 1.048,
      "step": 81
    },
    {
      "epoch": 0.017638201763820176,
      "grad_norm": 0.0,
      "learning_rate": 1.1714285714285716e-05,
      "loss": 1.0558,
      "step": 82
    },
    {
      "epoch": 0.017853301785330178,
      "grad_norm": 0.0,
      "learning_rate": 1.1857142857142858e-05,
      "loss": 1.0355,
      "step": 83
    },
    {
      "epoch": 0.01806840180684018,
      "grad_norm": 0.0,
      "learning_rate": 1.2e-05,
      "loss": 0.9978,
      "step": 84
    },
    {
      "epoch": 0.018283501828350184,
      "grad_norm": 0.0,
      "learning_rate": 1.2142857142857142e-05,
      "loss": 0.959,
      "step": 85
    },
    {
      "epoch": 0.018498601849860186,
      "grad_norm": 0.0,
      "learning_rate": 1.2285714285714288e-05,
      "loss": 1.0093,
      "step": 86
    },
    {
      "epoch": 0.018713701871370187,
      "grad_norm": 0.0,
      "learning_rate": 1.242857142857143e-05,
      "loss": 1.0048,
      "step": 87
    },
    {
      "epoch": 0.01892880189288019,
      "grad_norm": 0.0,
      "learning_rate": 1.2571428571428572e-05,
      "loss": 1.0063,
      "step": 88
    },
    {
      "epoch": 0.01914390191439019,
      "grad_norm": 0.0,
      "learning_rate": 1.2714285714285715e-05,
      "loss": 1.0857,
      "step": 89
    },
    {
      "epoch": 0.019359001935900195,
      "grad_norm": 0.0,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 1.0478,
      "step": 90
    },
    {
      "epoch": 0.019574101957410196,
      "grad_norm": 0.0,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.9996,
      "step": 91
    },
    {
      "epoch": 0.019789201978920198,
      "grad_norm": 0.0,
      "learning_rate": 1.3142857142857145e-05,
      "loss": 0.9492,
      "step": 92
    },
    {
      "epoch": 0.0200043020004302,
      "grad_norm": 0.0,
      "learning_rate": 1.3285714285714287e-05,
      "loss": 0.9939,
      "step": 93
    },
    {
      "epoch": 0.0202194020219402,
      "grad_norm": 0.0,
      "learning_rate": 1.3428571428571429e-05,
      "loss": 1.0402,
      "step": 94
    },
    {
      "epoch": 0.020434502043450205,
      "grad_norm": 0.0,
      "learning_rate": 1.3571428571428574e-05,
      "loss": 0.9867,
      "step": 95
    },
    {
      "epoch": 0.020649602064960207,
      "grad_norm": 0.0,
      "learning_rate": 1.3714285714285716e-05,
      "loss": 0.9301,
      "step": 96
    },
    {
      "epoch": 0.02086470208647021,
      "grad_norm": 0.0,
      "learning_rate": 1.3857142857142858e-05,
      "loss": 1.0195,
      "step": 97
    },
    {
      "epoch": 0.02107980210798021,
      "grad_norm": 0.0,
      "learning_rate": 1.4e-05,
      "loss": 0.9919,
      "step": 98
    },
    {
      "epoch": 0.021294902129490215,
      "grad_norm": 0.0,
      "learning_rate": 1.4142857142857145e-05,
      "loss": 1.043,
      "step": 99
    },
    {
      "epoch": 0.021510002151000216,
      "grad_norm": 0.0,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.9567,
      "step": 100
    },
    {
      "epoch": 0.021725102172510218,
      "grad_norm": 0.0,
      "learning_rate": 1.4428571428571429e-05,
      "loss": 1.0453,
      "step": 101
    },
    {
      "epoch": 0.02194020219402022,
      "grad_norm": 0.0,
      "learning_rate": 1.4571428571428573e-05,
      "loss": 0.9903,
      "step": 102
    },
    {
      "epoch": 0.02215530221553022,
      "grad_norm": 0.0,
      "learning_rate": 1.4714285714285716e-05,
      "loss": 0.9636,
      "step": 103
    },
    {
      "epoch": 0.022370402237040225,
      "grad_norm": 0.0,
      "learning_rate": 1.4857142857142858e-05,
      "loss": 0.9519,
      "step": 104
    },
    {
      "epoch": 0.022585502258550227,
      "grad_norm": 0.0,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.9829,
      "step": 105
    },
    {
      "epoch": 0.022800602280060228,
      "grad_norm": 0.0,
      "learning_rate": 1.5142857142857144e-05,
      "loss": 0.9673,
      "step": 106
    },
    {
      "epoch": 0.02301570230157023,
      "grad_norm": 0.0,
      "learning_rate": 1.5285714285714286e-05,
      "loss": 1.0144,
      "step": 107
    },
    {
      "epoch": 0.02323080232308023,
      "grad_norm": 0.0,
      "learning_rate": 1.542857142857143e-05,
      "loss": 1.0068,
      "step": 108
    },
    {
      "epoch": 0.023445902344590236,
      "grad_norm": 0.0,
      "learning_rate": 1.5571428571428573e-05,
      "loss": 1.062,
      "step": 109
    },
    {
      "epoch": 0.023661002366100237,
      "grad_norm": 0.0,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.9843,
      "step": 110
    },
    {
      "epoch": 0.02387610238761024,
      "grad_norm": 0.0,
      "learning_rate": 1.5857142857142857e-05,
      "loss": 1.0297,
      "step": 111
    },
    {
      "epoch": 0.02409120240912024,
      "grad_norm": 0.0,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.9894,
      "step": 112
    },
    {
      "epoch": 0.02430630243063024,
      "grad_norm": 0.0,
      "learning_rate": 1.6142857142857145e-05,
      "loss": 1.0086,
      "step": 113
    },
    {
      "epoch": 0.024521402452140247,
      "grad_norm": 0.0,
      "learning_rate": 1.6285714285714287e-05,
      "loss": 0.965,
      "step": 114
    },
    {
      "epoch": 0.024736502473650248,
      "grad_norm": 0.0,
      "learning_rate": 1.642857142857143e-05,
      "loss": 1.0184,
      "step": 115
    },
    {
      "epoch": 0.02495160249516025,
      "grad_norm": 0.0,
      "learning_rate": 1.6571428571428574e-05,
      "loss": 0.9504,
      "step": 116
    },
    {
      "epoch": 0.02516670251667025,
      "grad_norm": 0.0,
      "learning_rate": 1.6714285714285716e-05,
      "loss": 0.9846,
      "step": 117
    },
    {
      "epoch": 0.025381802538180252,
      "grad_norm": 0.0,
      "learning_rate": 1.6857142857142858e-05,
      "loss": 1.0332,
      "step": 118
    },
    {
      "epoch": 0.025596902559690257,
      "grad_norm": 0.0,
      "learning_rate": 1.7e-05,
      "loss": 1.0284,
      "step": 119
    },
    {
      "epoch": 0.02581200258120026,
      "grad_norm": 0.0,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.9632,
      "step": 120
    },
    {
      "epoch": 0.02602710260271026,
      "grad_norm": 0.0,
      "learning_rate": 1.7285714285714287e-05,
      "loss": 0.9937,
      "step": 121
    },
    {
      "epoch": 0.02624220262422026,
      "grad_norm": 0.0,
      "learning_rate": 1.742857142857143e-05,
      "loss": 1.0101,
      "step": 122
    },
    {
      "epoch": 0.026457302645730263,
      "grad_norm": 0.0,
      "learning_rate": 1.757142857142857e-05,
      "loss": 1.02,
      "step": 123
    },
    {
      "epoch": 0.026672402667240268,
      "grad_norm": 0.0,
      "learning_rate": 1.7714285714285717e-05,
      "loss": 0.9759,
      "step": 124
    },
    {
      "epoch": 0.02688750268875027,
      "grad_norm": 0.0,
      "learning_rate": 1.785714285714286e-05,
      "loss": 1.0175,
      "step": 125
    },
    {
      "epoch": 0.02710260271026027,
      "grad_norm": 0.0,
      "learning_rate": 1.8e-05,
      "loss": 0.9886,
      "step": 126
    },
    {
      "epoch": 0.027317702731770272,
      "grad_norm": 0.0,
      "learning_rate": 1.8142857142857146e-05,
      "loss": 1.0129,
      "step": 127
    },
    {
      "epoch": 0.027532802753280277,
      "grad_norm": 0.0,
      "learning_rate": 1.8285714285714288e-05,
      "loss": 1.0234,
      "step": 128
    },
    {
      "epoch": 0.02774790277479028,
      "grad_norm": 0.0,
      "learning_rate": 1.842857142857143e-05,
      "loss": 1.0325,
      "step": 129
    },
    {
      "epoch": 0.02796300279630028,
      "grad_norm": 0.0,
      "learning_rate": 1.8571428571428575e-05,
      "loss": 1.0275,
      "step": 130
    },
    {
      "epoch": 0.02817810281781028,
      "grad_norm": 0.0,
      "learning_rate": 1.8714285714285717e-05,
      "loss": 1.025,
      "step": 131
    },
    {
      "epoch": 0.028393202839320283,
      "grad_norm": 0.0,
      "learning_rate": 1.885714285714286e-05,
      "loss": 1.0253,
      "step": 132
    },
    {
      "epoch": 0.028608302860830288,
      "grad_norm": 0.0,
      "learning_rate": 1.9e-05,
      "loss": 0.9791,
      "step": 133
    },
    {
      "epoch": 0.02882340288234029,
      "grad_norm": 0.0,
      "learning_rate": 1.9142857142857146e-05,
      "loss": 1.0389,
      "step": 134
    },
    {
      "epoch": 0.02903850290385029,
      "grad_norm": 0.0,
      "learning_rate": 1.928571428571429e-05,
      "loss": 1.0425,
      "step": 135
    },
    {
      "epoch": 0.029253602925360292,
      "grad_norm": 0.0,
      "learning_rate": 1.942857142857143e-05,
      "loss": 1.0423,
      "step": 136
    },
    {
      "epoch": 0.029468702946870293,
      "grad_norm": 0.0,
      "learning_rate": 1.9571428571428572e-05,
      "loss": 1.0444,
      "step": 137
    },
    {
      "epoch": 0.0296838029683803,
      "grad_norm": 0.0,
      "learning_rate": 1.9714285714285718e-05,
      "loss": 0.9717,
      "step": 138
    },
    {
      "epoch": 0.0298989029898903,
      "grad_norm": 0.0,
      "learning_rate": 1.985714285714286e-05,
      "loss": 1.0227,
      "step": 139
    },
    {
      "epoch": 0.0301140030114003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0157,
      "step": 140
    },
    {
      "epoch": 0.030329103032910303,
      "grad_norm": 0.0,
      "learning_rate": 1.9999997572779332e-05,
      "loss": 1.0007,
      "step": 141
    },
    {
      "epoch": 0.030544203054420304,
      "grad_norm": 0.0,
      "learning_rate": 1.9999990291118506e-05,
      "loss": 1.055,
      "step": 142
    },
    {
      "epoch": 0.03075930307593031,
      "grad_norm": 0.0,
      "learning_rate": 1.9999978155021058e-05,
      "loss": 1.0181,
      "step": 143
    },
    {
      "epoch": 0.03097440309744031,
      "grad_norm": 0.0,
      "learning_rate": 1.9999961164492875e-05,
      "loss": 1.0719,
      "step": 144
    },
    {
      "epoch": 0.031189503118950312,
      "grad_norm": 0.0,
      "learning_rate": 1.9999939319542212e-05,
      "loss": 0.9398,
      "step": 145
    },
    {
      "epoch": 0.03140460314046031,
      "grad_norm": 0.0,
      "learning_rate": 1.9999912620179666e-05,
      "loss": 1.0339,
      "step": 146
    },
    {
      "epoch": 0.03161970316197032,
      "grad_norm": 0.0,
      "learning_rate": 1.9999881066418204e-05,
      "loss": 1.0131,
      "step": 147
    },
    {
      "epoch": 0.031834803183480316,
      "grad_norm": 0.0,
      "learning_rate": 1.999984465827314e-05,
      "loss": 0.9945,
      "step": 148
    },
    {
      "epoch": 0.03204990320499032,
      "grad_norm": 0.0,
      "learning_rate": 1.999980339576215e-05,
      "loss": 1.0186,
      "step": 149
    },
    {
      "epoch": 0.032265003226500326,
      "grad_norm": 0.0,
      "learning_rate": 1.9999757278905266e-05,
      "loss": 1.031,
      "step": 150
    },
    {
      "epoch": 0.032480103248010324,
      "grad_norm": 0.0,
      "learning_rate": 1.999970630772487e-05,
      "loss": 1.0405,
      "step": 151
    },
    {
      "epoch": 0.03269520326952033,
      "grad_norm": 0.0,
      "learning_rate": 1.999965048224571e-05,
      "loss": 1.0178,
      "step": 152
    },
    {
      "epoch": 0.03291030329103033,
      "grad_norm": 0.0,
      "learning_rate": 1.9999589802494888e-05,
      "loss": 1.0204,
      "step": 153
    },
    {
      "epoch": 0.03312540331254033,
      "grad_norm": 0.0,
      "learning_rate": 1.9999524268501854e-05,
      "loss": 1.0537,
      "step": 154
    },
    {
      "epoch": 0.033340503334050337,
      "grad_norm": 0.0,
      "learning_rate": 1.9999453880298423e-05,
      "loss": 1.0239,
      "step": 155
    },
    {
      "epoch": 0.033555603355560334,
      "grad_norm": 0.0,
      "learning_rate": 1.999937863791877e-05,
      "loss": 1.0326,
      "step": 156
    },
    {
      "epoch": 0.03377070337707034,
      "grad_norm": 0.0,
      "learning_rate": 1.999929854139942e-05,
      "loss": 1.0052,
      "step": 157
    },
    {
      "epoch": 0.03398580339858034,
      "grad_norm": 0.0,
      "learning_rate": 1.9999213590779245e-05,
      "loss": 1.0182,
      "step": 158
    },
    {
      "epoch": 0.03420090342009034,
      "grad_norm": 0.0,
      "learning_rate": 1.9999123786099497e-05,
      "loss": 1.0151,
      "step": 159
    },
    {
      "epoch": 0.03441600344160035,
      "grad_norm": 0.0,
      "learning_rate": 1.9999029127403763e-05,
      "loss": 0.9961,
      "step": 160
    },
    {
      "epoch": 0.034631103463110345,
      "grad_norm": 0.0,
      "learning_rate": 1.9998929614738e-05,
      "loss": 1.0555,
      "step": 161
    },
    {
      "epoch": 0.03484620348462035,
      "grad_norm": 0.0,
      "learning_rate": 1.999882524815051e-05,
      "loss": 1.0047,
      "step": 162
    },
    {
      "epoch": 0.03506130350613035,
      "grad_norm": 0.0,
      "learning_rate": 1.999871602769196e-05,
      "loss": 0.9963,
      "step": 163
    },
    {
      "epoch": 0.03527640352764035,
      "grad_norm": 0.0,
      "learning_rate": 1.9998601953415374e-05,
      "loss": 0.9714,
      "step": 164
    },
    {
      "epoch": 0.03549150354915036,
      "grad_norm": 0.0,
      "learning_rate": 1.999848302537612e-05,
      "loss": 1.054,
      "step": 165
    },
    {
      "epoch": 0.035706603570660356,
      "grad_norm": 0.0,
      "learning_rate": 1.999835924363194e-05,
      "loss": 1.0144,
      "step": 166
    },
    {
      "epoch": 0.03592170359217036,
      "grad_norm": 0.0,
      "learning_rate": 1.9998230608242917e-05,
      "loss": 1.0421,
      "step": 167
    },
    {
      "epoch": 0.03613680361368036,
      "grad_norm": 0.0,
      "learning_rate": 1.99980971192715e-05,
      "loss": 0.9899,
      "step": 168
    },
    {
      "epoch": 0.036351903635190363,
      "grad_norm": 0.0,
      "learning_rate": 1.999795877678249e-05,
      "loss": 1.0686,
      "step": 169
    },
    {
      "epoch": 0.03656700365670037,
      "grad_norm": 0.0,
      "learning_rate": 1.9997815580843043e-05,
      "loss": 0.9609,
      "step": 170
    },
    {
      "epoch": 0.036782103678210366,
      "grad_norm": 0.0,
      "learning_rate": 1.999766753152267e-05,
      "loss": 1.0215,
      "step": 171
    },
    {
      "epoch": 0.03699720369972037,
      "grad_norm": 0.0,
      "learning_rate": 1.9997514628893247e-05,
      "loss": 0.9995,
      "step": 172
    },
    {
      "epoch": 0.03721230372123037,
      "grad_norm": 0.0,
      "learning_rate": 1.9997356873028994e-05,
      "loss": 1.0368,
      "step": 173
    },
    {
      "epoch": 0.037427403742740374,
      "grad_norm": 0.0,
      "learning_rate": 1.9997194264006498e-05,
      "loss": 0.9855,
      "step": 174
    },
    {
      "epoch": 0.03764250376425038,
      "grad_norm": 0.0,
      "learning_rate": 1.999702680190469e-05,
      "loss": 0.9558,
      "step": 175
    },
    {
      "epoch": 0.03785760378576038,
      "grad_norm": 0.0,
      "learning_rate": 1.9996854486804874e-05,
      "loss": 0.9937,
      "step": 176
    },
    {
      "epoch": 0.03807270380727038,
      "grad_norm": 0.0,
      "learning_rate": 1.9996677318790687e-05,
      "loss": 0.9762,
      "step": 177
    },
    {
      "epoch": 0.03828780382878038,
      "grad_norm": 0.0,
      "learning_rate": 1.9996495297948143e-05,
      "loss": 1.0705,
      "step": 178
    },
    {
      "epoch": 0.038502903850290385,
      "grad_norm": 0.0,
      "learning_rate": 1.9996308424365596e-05,
      "loss": 0.9845,
      "step": 179
    },
    {
      "epoch": 0.03871800387180039,
      "grad_norm": 0.0,
      "learning_rate": 1.999611669813377e-05,
      "loss": 1.0645,
      "step": 180
    },
    {
      "epoch": 0.03893310389331039,
      "grad_norm": 0.0,
      "learning_rate": 1.999592011934573e-05,
      "loss": 1.0481,
      "step": 181
    },
    {
      "epoch": 0.03914820391482039,
      "grad_norm": 0.0,
      "learning_rate": 1.999571868809691e-05,
      "loss": 1.0405,
      "step": 182
    },
    {
      "epoch": 0.03936330393633039,
      "grad_norm": 0.0,
      "learning_rate": 1.999551240448509e-05,
      "loss": 1.0632,
      "step": 183
    },
    {
      "epoch": 0.039578403957840395,
      "grad_norm": 0.0,
      "learning_rate": 1.9995301268610415e-05,
      "loss": 1.0103,
      "step": 184
    },
    {
      "epoch": 0.0397935039793504,
      "grad_norm": 0.0,
      "learning_rate": 1.999508528057537e-05,
      "loss": 1.0134,
      "step": 185
    },
    {
      "epoch": 0.0400086040008604,
      "grad_norm": 0.0,
      "learning_rate": 1.9994864440484812e-05,
      "loss": 0.9762,
      "step": 186
    },
    {
      "epoch": 0.0402237040223704,
      "grad_norm": 0.0,
      "learning_rate": 1.999463874844595e-05,
      "loss": 1.0603,
      "step": 187
    },
    {
      "epoch": 0.0404388040438804,
      "grad_norm": 0.0,
      "learning_rate": 1.9994408204568335e-05,
      "loss": 0.9945,
      "step": 188
    },
    {
      "epoch": 0.040653904065390406,
      "grad_norm": 0.0,
      "learning_rate": 1.9994172808963887e-05,
      "loss": 1.0014,
      "step": 189
    },
    {
      "epoch": 0.04086900408690041,
      "grad_norm": 0.0,
      "learning_rate": 1.999393256174688e-05,
      "loss": 1.0286,
      "step": 190
    },
    {
      "epoch": 0.04108410410841041,
      "grad_norm": 0.0,
      "learning_rate": 1.999368746303394e-05,
      "loss": 1.0704,
      "step": 191
    },
    {
      "epoch": 0.041299204129920414,
      "grad_norm": 0.0,
      "learning_rate": 1.9993437512944046e-05,
      "loss": 0.9998,
      "step": 192
    },
    {
      "epoch": 0.04151430415143041,
      "grad_norm": 0.0,
      "learning_rate": 1.9993182711598536e-05,
      "loss": 1.0206,
      "step": 193
    },
    {
      "epoch": 0.04172940417294042,
      "grad_norm": 0.0,
      "learning_rate": 1.9992923059121107e-05,
      "loss": 0.9683,
      "step": 194
    },
    {
      "epoch": 0.04194450419445042,
      "grad_norm": 0.0,
      "learning_rate": 1.9992658555637795e-05,
      "loss": 0.943,
      "step": 195
    },
    {
      "epoch": 0.04215960421596042,
      "grad_norm": 0.0,
      "learning_rate": 1.999238920127701e-05,
      "loss": 1.0395,
      "step": 196
    },
    {
      "epoch": 0.042374704237470424,
      "grad_norm": 0.0,
      "learning_rate": 1.9992114996169506e-05,
      "loss": 0.9962,
      "step": 197
    },
    {
      "epoch": 0.04258980425898043,
      "grad_norm": 0.0,
      "learning_rate": 1.9991835940448396e-05,
      "loss": 1.0622,
      "step": 198
    },
    {
      "epoch": 0.04280490428049043,
      "grad_norm": 0.0,
      "learning_rate": 1.9991552034249146e-05,
      "loss": 1.0278,
      "step": 199
    },
    {
      "epoch": 0.04302000430200043,
      "grad_norm": 0.0,
      "learning_rate": 1.9991263277709572e-05,
      "loss": 1.0942,
      "step": 200
    },
    {
      "epoch": 0.04323510432351043,
      "grad_norm": 0.0,
      "learning_rate": 1.9990969670969852e-05,
      "loss": 1.0118,
      "step": 201
    },
    {
      "epoch": 0.043450204345020435,
      "grad_norm": 0.0,
      "learning_rate": 1.9990671214172517e-05,
      "loss": 1.0182,
      "step": 202
    },
    {
      "epoch": 0.04366530436653044,
      "grad_norm": 0.0,
      "learning_rate": 1.999036790746245e-05,
      "loss": 1.0803,
      "step": 203
    },
    {
      "epoch": 0.04388040438804044,
      "grad_norm": 0.0,
      "learning_rate": 1.999005975098689e-05,
      "loss": 1.0718,
      "step": 204
    },
    {
      "epoch": 0.04409550440955044,
      "grad_norm": 0.0,
      "learning_rate": 1.998974674489543e-05,
      "loss": 1.0441,
      "step": 205
    },
    {
      "epoch": 0.04431060443106044,
      "grad_norm": 0.0,
      "learning_rate": 1.9989428889340016e-05,
      "loss": 1.0395,
      "step": 206
    },
    {
      "epoch": 0.044525704452570446,
      "grad_norm": 0.0,
      "learning_rate": 1.9989106184474948e-05,
      "loss": 0.9808,
      "step": 207
    },
    {
      "epoch": 0.04474080447408045,
      "grad_norm": 0.0,
      "learning_rate": 1.9988778630456883e-05,
      "loss": 1.0357,
      "step": 208
    },
    {
      "epoch": 0.04495590449559045,
      "grad_norm": 0.0,
      "learning_rate": 1.998844622744483e-05,
      "loss": 0.9994,
      "step": 209
    },
    {
      "epoch": 0.04517100451710045,
      "grad_norm": 0.0,
      "learning_rate": 1.9988108975600153e-05,
      "loss": 0.9807,
      "step": 210
    },
    {
      "epoch": 0.04538610453861045,
      "grad_norm": 0.0,
      "learning_rate": 1.9987766875086562e-05,
      "loss": 1.009,
      "step": 211
    },
    {
      "epoch": 0.045601204560120456,
      "grad_norm": 0.0,
      "learning_rate": 1.998741992607014e-05,
      "loss": 1.0176,
      "step": 212
    },
    {
      "epoch": 0.04581630458163046,
      "grad_norm": 0.0,
      "learning_rate": 1.9987068128719304e-05,
      "loss": 0.9829,
      "step": 213
    },
    {
      "epoch": 0.04603140460314046,
      "grad_norm": 0.0,
      "learning_rate": 1.998671148320483e-05,
      "loss": 1.02,
      "step": 214
    },
    {
      "epoch": 0.046246504624650464,
      "grad_norm": 0.0,
      "learning_rate": 1.998634998969985e-05,
      "loss": 0.9454,
      "step": 215
    },
    {
      "epoch": 0.04646160464616046,
      "grad_norm": 0.0,
      "learning_rate": 1.9985983648379857e-05,
      "loss": 0.9612,
      "step": 216
    },
    {
      "epoch": 0.04667670466767047,
      "grad_norm": 0.0,
      "learning_rate": 1.998561245942268e-05,
      "loss": 0.9566,
      "step": 217
    },
    {
      "epoch": 0.04689180468918047,
      "grad_norm": 0.0,
      "learning_rate": 1.9985236423008513e-05,
      "loss": 0.9382,
      "step": 218
    },
    {
      "epoch": 0.04710690471069047,
      "grad_norm": 0.0,
      "learning_rate": 1.9984855539319904e-05,
      "loss": 0.99,
      "step": 219
    },
    {
      "epoch": 0.047322004732200475,
      "grad_norm": 0.0,
      "learning_rate": 1.9984469808541747e-05,
      "loss": 1.0389,
      "step": 220
    },
    {
      "epoch": 0.04753710475371047,
      "grad_norm": 0.0,
      "learning_rate": 1.9984079230861296e-05,
      "loss": 0.9997,
      "step": 221
    },
    {
      "epoch": 0.04775220477522048,
      "grad_norm": 0.0,
      "learning_rate": 1.998368380646815e-05,
      "loss": 1.0078,
      "step": 222
    },
    {
      "epoch": 0.04796730479673048,
      "grad_norm": 0.0,
      "learning_rate": 1.998328353555427e-05,
      "loss": 0.9862,
      "step": 223
    },
    {
      "epoch": 0.04818240481824048,
      "grad_norm": 0.0,
      "learning_rate": 1.998287841831396e-05,
      "loss": 1.0607,
      "step": 224
    },
    {
      "epoch": 0.048397504839750485,
      "grad_norm": 0.0,
      "learning_rate": 1.998246845494389e-05,
      "loss": 0.9533,
      "step": 225
    },
    {
      "epoch": 0.04861260486126048,
      "grad_norm": 0.0,
      "learning_rate": 1.998205364564307e-05,
      "loss": 1.036,
      "step": 226
    },
    {
      "epoch": 0.04882770488277049,
      "grad_norm": 0.0,
      "learning_rate": 1.998163399061286e-05,
      "loss": 1.0615,
      "step": 227
    },
    {
      "epoch": 0.04904280490428049,
      "grad_norm": 0.0,
      "learning_rate": 1.998120949005699e-05,
      "loss": 1.019,
      "step": 228
    },
    {
      "epoch": 0.04925790492579049,
      "grad_norm": 0.0,
      "learning_rate": 1.9980780144181524e-05,
      "loss": 1.044,
      "step": 229
    },
    {
      "epoch": 0.049473004947300496,
      "grad_norm": 0.0,
      "learning_rate": 1.9980345953194892e-05,
      "loss": 1.0828,
      "step": 230
    },
    {
      "epoch": 0.049688104968810494,
      "grad_norm": 0.0,
      "learning_rate": 1.9979906917307857e-05,
      "loss": 1.0152,
      "step": 231
    },
    {
      "epoch": 0.0499032049903205,
      "grad_norm": 0.0,
      "learning_rate": 1.9979463036733562e-05,
      "loss": 1.0352,
      "step": 232
    },
    {
      "epoch": 0.050118305011830504,
      "grad_norm": 0.0,
      "learning_rate": 1.9979014311687475e-05,
      "loss": 1.0282,
      "step": 233
    },
    {
      "epoch": 0.0503334050333405,
      "grad_norm": 0.0,
      "learning_rate": 1.9978560742387437e-05,
      "loss": 1.0557,
      "step": 234
    },
    {
      "epoch": 0.05054850505485051,
      "grad_norm": 0.0,
      "learning_rate": 1.9978102329053618e-05,
      "loss": 0.9244,
      "step": 235
    },
    {
      "epoch": 0.050763605076360505,
      "grad_norm": 0.0,
      "learning_rate": 1.9977639071908564e-05,
      "loss": 1.0218,
      "step": 236
    },
    {
      "epoch": 0.05097870509787051,
      "grad_norm": 0.0,
      "learning_rate": 1.9977170971177152e-05,
      "loss": 1.0706,
      "step": 237
    },
    {
      "epoch": 0.051193805119380514,
      "grad_norm": 0.0,
      "learning_rate": 1.9976698027086622e-05,
      "loss": 1.0053,
      "step": 238
    },
    {
      "epoch": 0.05140890514089051,
      "grad_norm": 0.0,
      "learning_rate": 1.9976220239866563e-05,
      "loss": 0.9486,
      "step": 239
    },
    {
      "epoch": 0.05162400516240052,
      "grad_norm": 0.0,
      "learning_rate": 1.997573760974891e-05,
      "loss": 0.9735,
      "step": 240
    },
    {
      "epoch": 0.051839105183910515,
      "grad_norm": 0.0,
      "learning_rate": 1.9975250136967957e-05,
      "loss": 0.999,
      "step": 241
    },
    {
      "epoch": 0.05205420520542052,
      "grad_norm": 0.0,
      "learning_rate": 1.9974757821760343e-05,
      "loss": 1.0376,
      "step": 242
    },
    {
      "epoch": 0.052269305226930525,
      "grad_norm": 0.0,
      "learning_rate": 1.997426066436506e-05,
      "loss": 1.0271,
      "step": 243
    },
    {
      "epoch": 0.05248440524844052,
      "grad_norm": 0.0,
      "learning_rate": 1.997375866502345e-05,
      "loss": 1.0412,
      "step": 244
    },
    {
      "epoch": 0.05269950526995053,
      "grad_norm": 0.0,
      "learning_rate": 1.997325182397921e-05,
      "loss": 1.034,
      "step": 245
    },
    {
      "epoch": 0.052914605291460526,
      "grad_norm": 0.0,
      "learning_rate": 1.9972740141478374e-05,
      "loss": 0.904,
      "step": 246
    },
    {
      "epoch": 0.05312970531297053,
      "grad_norm": 0.0,
      "learning_rate": 1.9972223617769343e-05,
      "loss": 1.0233,
      "step": 247
    },
    {
      "epoch": 0.053344805334480536,
      "grad_norm": 0.0,
      "learning_rate": 1.9971702253102856e-05,
      "loss": 1.0107,
      "step": 248
    },
    {
      "epoch": 0.053559905355990534,
      "grad_norm": 0.0,
      "learning_rate": 1.9971176047732008e-05,
      "loss": 1.0228,
      "step": 249
    },
    {
      "epoch": 0.05377500537750054,
      "grad_norm": 0.0,
      "learning_rate": 1.9970645001912243e-05,
      "loss": 1.0251,
      "step": 250
    },
    {
      "epoch": 0.053990105399010536,
      "grad_norm": 0.0,
      "learning_rate": 1.997010911590135e-05,
      "loss": 0.952,
      "step": 251
    },
    {
      "epoch": 0.05420520542052054,
      "grad_norm": 0.0,
      "learning_rate": 1.996956838995948e-05,
      "loss": 1.0073,
      "step": 252
    },
    {
      "epoch": 0.054420305442030546,
      "grad_norm": 0.0,
      "learning_rate": 1.9969022824349115e-05,
      "loss": 0.9844,
      "step": 253
    },
    {
      "epoch": 0.054635405463540544,
      "grad_norm": 0.0,
      "learning_rate": 1.9968472419335106e-05,
      "loss": 1.0232,
      "step": 254
    },
    {
      "epoch": 0.05485050548505055,
      "grad_norm": 0.0,
      "learning_rate": 1.996791717518464e-05,
      "loss": 1.0286,
      "step": 255
    },
    {
      "epoch": 0.055065605506560554,
      "grad_norm": 0.0,
      "learning_rate": 1.9967357092167258e-05,
      "loss": 1.0014,
      "step": 256
    },
    {
      "epoch": 0.05528070552807055,
      "grad_norm": 0.0,
      "learning_rate": 1.9966792170554842e-05,
      "loss": 0.9896,
      "step": 257
    },
    {
      "epoch": 0.05549580554958056,
      "grad_norm": 0.0,
      "learning_rate": 1.996622241062164e-05,
      "loss": 0.9869,
      "step": 258
    },
    {
      "epoch": 0.055710905571090555,
      "grad_norm": 0.0,
      "learning_rate": 1.9965647812644232e-05,
      "loss": 1.0099,
      "step": 259
    },
    {
      "epoch": 0.05592600559260056,
      "grad_norm": 0.0,
      "learning_rate": 1.9965068376901556e-05,
      "loss": 1.0089,
      "step": 260
    },
    {
      "epoch": 0.056141105614110565,
      "grad_norm": 0.0,
      "learning_rate": 1.9964484103674896e-05,
      "loss": 0.9594,
      "step": 261
    },
    {
      "epoch": 0.05635620563562056,
      "grad_norm": 0.0,
      "learning_rate": 1.9963894993247884e-05,
      "loss": 0.9938,
      "step": 262
    },
    {
      "epoch": 0.05657130565713057,
      "grad_norm": 0.0,
      "learning_rate": 1.9963301045906497e-05,
      "loss": 1.0276,
      "step": 263
    },
    {
      "epoch": 0.056786405678640565,
      "grad_norm": 0.0,
      "learning_rate": 1.9962702261939067e-05,
      "loss": 1.0705,
      "step": 264
    },
    {
      "epoch": 0.05700150570015057,
      "grad_norm": 0.0,
      "learning_rate": 1.9962098641636268e-05,
      "loss": 1.0053,
      "step": 265
    },
    {
      "epoch": 0.057216605721660575,
      "grad_norm": 0.0,
      "learning_rate": 1.9961490185291128e-05,
      "loss": 1.0102,
      "step": 266
    },
    {
      "epoch": 0.05743170574317057,
      "grad_norm": 0.0,
      "learning_rate": 1.9960876893199016e-05,
      "loss": 0.9833,
      "step": 267
    },
    {
      "epoch": 0.05764680576468058,
      "grad_norm": 0.0,
      "learning_rate": 1.9960258765657646e-05,
      "loss": 0.9396,
      "step": 268
    },
    {
      "epoch": 0.057861905786190576,
      "grad_norm": 0.0,
      "learning_rate": 1.995963580296709e-05,
      "loss": 0.9708,
      "step": 269
    },
    {
      "epoch": 0.05807700580770058,
      "grad_norm": 0.0,
      "learning_rate": 1.9959008005429757e-05,
      "loss": 1.0087,
      "step": 270
    },
    {
      "epoch": 0.058292105829210586,
      "grad_norm": 0.0,
      "learning_rate": 1.995837537335042e-05,
      "loss": 0.9785,
      "step": 271
    },
    {
      "epoch": 0.058507205850720584,
      "grad_norm": 0.0,
      "learning_rate": 1.995773790703617e-05,
      "loss": 0.9179,
      "step": 272
    },
    {
      "epoch": 0.05872230587223059,
      "grad_norm": 0.0,
      "learning_rate": 1.9957095606796472e-05,
      "loss": 0.8997,
      "step": 273
    },
    {
      "epoch": 0.05893740589374059,
      "grad_norm": 0.0,
      "learning_rate": 1.9956448472943122e-05,
      "loss": 0.918,
      "step": 274
    },
    {
      "epoch": 0.05915250591525059,
      "grad_norm": 0.0,
      "learning_rate": 1.9955796505790266e-05,
      "loss": 1.0035,
      "step": 275
    },
    {
      "epoch": 0.0593676059367606,
      "grad_norm": 0.0,
      "learning_rate": 1.9955139705654405e-05,
      "loss": 0.9604,
      "step": 276
    },
    {
      "epoch": 0.059582705958270595,
      "grad_norm": 0.0,
      "learning_rate": 1.9954478072854374e-05,
      "loss": 0.9775,
      "step": 277
    },
    {
      "epoch": 0.0597978059797806,
      "grad_norm": 0.0,
      "learning_rate": 1.995381160771136e-05,
      "loss": 1.05,
      "step": 278
    },
    {
      "epoch": 0.0600129060012906,
      "grad_norm": 0.0,
      "learning_rate": 1.9953140310548892e-05,
      "loss": 0.9708,
      "step": 279
    },
    {
      "epoch": 0.0602280060228006,
      "grad_norm": 0.0,
      "learning_rate": 1.995246418169285e-05,
      "loss": 1.0018,
      "step": 280
    },
    {
      "epoch": 0.06044310604431061,
      "grad_norm": 0.0,
      "learning_rate": 1.9951783221471456e-05,
      "loss": 1.0487,
      "step": 281
    },
    {
      "epoch": 0.060658206065820605,
      "grad_norm": 0.0,
      "learning_rate": 1.995109743021528e-05,
      "loss": 0.9987,
      "step": 282
    },
    {
      "epoch": 0.06087330608733061,
      "grad_norm": 0.0,
      "learning_rate": 1.9950406808257227e-05,
      "loss": 0.9853,
      "step": 283
    },
    {
      "epoch": 0.06108840610884061,
      "grad_norm": 0.0,
      "learning_rate": 1.9949711355932565e-05,
      "loss": 1.0024,
      "step": 284
    },
    {
      "epoch": 0.06130350613035061,
      "grad_norm": 0.0,
      "learning_rate": 1.9949011073578894e-05,
      "loss": 0.9622,
      "step": 285
    },
    {
      "epoch": 0.06151860615186062,
      "grad_norm": 0.0,
      "learning_rate": 1.9948305961536163e-05,
      "loss": 0.9528,
      "step": 286
    },
    {
      "epoch": 0.061733706173370616,
      "grad_norm": 0.0,
      "learning_rate": 1.9947596020146666e-05,
      "loss": 1.0037,
      "step": 287
    },
    {
      "epoch": 0.06194880619488062,
      "grad_norm": 0.0,
      "learning_rate": 1.994688124975503e-05,
      "loss": 1.0771,
      "step": 288
    },
    {
      "epoch": 0.06216390621639062,
      "grad_norm": 0.0,
      "learning_rate": 1.994616165070825e-05,
      "loss": 1.0046,
      "step": 289
    },
    {
      "epoch": 0.062379006237900624,
      "grad_norm": 0.0,
      "learning_rate": 1.994543722335564e-05,
      "loss": 0.9943,
      "step": 290
    },
    {
      "epoch": 0.06259410625941063,
      "grad_norm": 0.0,
      "learning_rate": 1.9944707968048875e-05,
      "loss": 1.0105,
      "step": 291
    },
    {
      "epoch": 0.06280920628092063,
      "grad_norm": 0.0,
      "learning_rate": 1.9943973885141966e-05,
      "loss": 0.9987,
      "step": 292
    },
    {
      "epoch": 0.06302430630243062,
      "grad_norm": 0.0,
      "learning_rate": 1.994323497499127e-05,
      "loss": 1.0077,
      "step": 293
    },
    {
      "epoch": 0.06323940632394064,
      "grad_norm": 0.0,
      "learning_rate": 1.9942491237955485e-05,
      "loss": 1.0056,
      "step": 294
    },
    {
      "epoch": 0.06345450634545063,
      "grad_norm": 0.0,
      "learning_rate": 1.994174267439566e-05,
      "loss": 0.9977,
      "step": 295
    },
    {
      "epoch": 0.06366960636696063,
      "grad_norm": 0.0,
      "learning_rate": 1.994098928467517e-05,
      "loss": 0.9779,
      "step": 296
    },
    {
      "epoch": 0.06388470638847064,
      "grad_norm": 0.0,
      "learning_rate": 1.9940231069159753e-05,
      "loss": 1.0025,
      "step": 297
    },
    {
      "epoch": 0.06409980640998064,
      "grad_norm": 0.0,
      "learning_rate": 1.9939468028217473e-05,
      "loss": 0.9805,
      "step": 298
    },
    {
      "epoch": 0.06431490643149064,
      "grad_norm": 0.0,
      "learning_rate": 1.993870016221875e-05,
      "loss": 0.9834,
      "step": 299
    },
    {
      "epoch": 0.06453000645300065,
      "grad_norm": 0.0,
      "learning_rate": 1.9937927471536335e-05,
      "loss": 0.9441,
      "step": 300
    },
    {
      "epoch": 0.06474510647451065,
      "grad_norm": 0.0,
      "learning_rate": 1.993714995654533e-05,
      "loss": 1.017,
      "step": 301
    },
    {
      "epoch": 0.06496020649602065,
      "grad_norm": 0.0,
      "learning_rate": 1.993636761762317e-05,
      "loss": 0.9782,
      "step": 302
    },
    {
      "epoch": 0.06517530651753065,
      "grad_norm": 0.0,
      "learning_rate": 1.9935580455149645e-05,
      "loss": 1.0041,
      "step": 303
    },
    {
      "epoch": 0.06539040653904066,
      "grad_norm": 0.0,
      "learning_rate": 1.993478846950687e-05,
      "loss": 1.0593,
      "step": 304
    },
    {
      "epoch": 0.06560550656055066,
      "grad_norm": 0.0,
      "learning_rate": 1.9933991661079316e-05,
      "loss": 0.9384,
      "step": 305
    },
    {
      "epoch": 0.06582060658206065,
      "grad_norm": 0.0,
      "learning_rate": 1.9933190030253788e-05,
      "loss": 1.05,
      "step": 306
    },
    {
      "epoch": 0.06603570660357067,
      "grad_norm": 0.0,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 0.944,
      "step": 307
    },
    {
      "epoch": 0.06625080662508066,
      "grad_norm": 0.0,
      "learning_rate": 1.9931572302967735e-05,
      "loss": 1.0194,
      "step": 308
    },
    {
      "epoch": 0.06646590664659066,
      "grad_norm": 0.0,
      "learning_rate": 1.9930756207292522e-05,
      "loss": 0.9951,
      "step": 309
    },
    {
      "epoch": 0.06668100666810067,
      "grad_norm": 0.0,
      "learning_rate": 1.9929935290789968e-05,
      "loss": 0.9343,
      "step": 310
    },
    {
      "epoch": 0.06689610668961067,
      "grad_norm": 0.0,
      "learning_rate": 1.9929109553858578e-05,
      "loss": 0.9807,
      "step": 311
    },
    {
      "epoch": 0.06711120671112067,
      "grad_norm": 0.0,
      "learning_rate": 1.9928278996899205e-05,
      "loss": 0.9783,
      "step": 312
    },
    {
      "epoch": 0.06732630673263067,
      "grad_norm": 0.0,
      "learning_rate": 1.9927443620315037e-05,
      "loss": 0.9828,
      "step": 313
    },
    {
      "epoch": 0.06754140675414068,
      "grad_norm": 0.0,
      "learning_rate": 1.99266034245116e-05,
      "loss": 0.9795,
      "step": 314
    },
    {
      "epoch": 0.06775650677565068,
      "grad_norm": 0.0,
      "learning_rate": 1.9925758409896763e-05,
      "loss": 1.0048,
      "step": 315
    },
    {
      "epoch": 0.06797160679716067,
      "grad_norm": 0.0,
      "learning_rate": 1.9924908576880738e-05,
      "loss": 1.0114,
      "step": 316
    },
    {
      "epoch": 0.06818670681867069,
      "grad_norm": 0.0,
      "learning_rate": 1.9924053925876064e-05,
      "loss": 0.9853,
      "step": 317
    },
    {
      "epoch": 0.06840180684018068,
      "grad_norm": 0.0,
      "learning_rate": 1.9923194457297628e-05,
      "loss": 0.9964,
      "step": 318
    },
    {
      "epoch": 0.06861690686169068,
      "grad_norm": 0.0,
      "learning_rate": 1.9922330171562657e-05,
      "loss": 0.9402,
      "step": 319
    },
    {
      "epoch": 0.0688320068832007,
      "grad_norm": 0.0,
      "learning_rate": 1.9921461069090715e-05,
      "loss": 0.9307,
      "step": 320
    },
    {
      "epoch": 0.06904710690471069,
      "grad_norm": 0.0,
      "learning_rate": 1.9920587150303697e-05,
      "loss": 1.0541,
      "step": 321
    },
    {
      "epoch": 0.06926220692622069,
      "grad_norm": 0.0,
      "learning_rate": 1.9919708415625847e-05,
      "loss": 0.9857,
      "step": 322
    },
    {
      "epoch": 0.06947730694773069,
      "grad_norm": 0.0,
      "learning_rate": 1.991882486548374e-05,
      "loss": 0.9511,
      "step": 323
    },
    {
      "epoch": 0.0696924069692407,
      "grad_norm": 0.0,
      "learning_rate": 1.9917936500306285e-05,
      "loss": 0.9404,
      "step": 324
    },
    {
      "epoch": 0.0699075069907507,
      "grad_norm": 0.0,
      "learning_rate": 1.991704332052474e-05,
      "loss": 0.9678,
      "step": 325
    },
    {
      "epoch": 0.0701226070122607,
      "grad_norm": 0.0,
      "learning_rate": 1.9916145326572696e-05,
      "loss": 0.9417,
      "step": 326
    },
    {
      "epoch": 0.07033770703377071,
      "grad_norm": 0.0,
      "learning_rate": 1.991524251888607e-05,
      "loss": 1.0437,
      "step": 327
    },
    {
      "epoch": 0.0705528070552807,
      "grad_norm": 0.0,
      "learning_rate": 1.9914334897903134e-05,
      "loss": 0.9753,
      "step": 328
    },
    {
      "epoch": 0.0707679070767907,
      "grad_norm": 0.0,
      "learning_rate": 1.991342246406448e-05,
      "loss": 0.9854,
      "step": 329
    },
    {
      "epoch": 0.07098300709830072,
      "grad_norm": 0.0,
      "learning_rate": 1.991250521781305e-05,
      "loss": 1.0302,
      "step": 330
    },
    {
      "epoch": 0.07119810711981071,
      "grad_norm": 0.0,
      "learning_rate": 1.9911583159594107e-05,
      "loss": 0.9938,
      "step": 331
    },
    {
      "epoch": 0.07141320714132071,
      "grad_norm": 0.0,
      "learning_rate": 1.991065628985527e-05,
      "loss": 0.9583,
      "step": 332
    },
    {
      "epoch": 0.07162830716283071,
      "grad_norm": 0.0,
      "learning_rate": 1.9909724609046473e-05,
      "loss": 0.9288,
      "step": 333
    },
    {
      "epoch": 0.07184340718434072,
      "grad_norm": 0.0,
      "learning_rate": 1.9908788117619997e-05,
      "loss": 1.0728,
      "step": 334
    },
    {
      "epoch": 0.07205850720585072,
      "grad_norm": 0.0,
      "learning_rate": 1.990784681603046e-05,
      "loss": 1.0048,
      "step": 335
    },
    {
      "epoch": 0.07227360722736072,
      "grad_norm": 0.0,
      "learning_rate": 1.9906900704734808e-05,
      "loss": 0.9559,
      "step": 336
    },
    {
      "epoch": 0.07248870724887073,
      "grad_norm": 0.0,
      "learning_rate": 1.990594978419233e-05,
      "loss": 0.9848,
      "step": 337
    },
    {
      "epoch": 0.07270380727038073,
      "grad_norm": 0.0,
      "learning_rate": 1.990499405486464e-05,
      "loss": 1.048,
      "step": 338
    },
    {
      "epoch": 0.07291890729189072,
      "grad_norm": 0.0,
      "learning_rate": 1.990403351721569e-05,
      "loss": 0.9511,
      "step": 339
    },
    {
      "epoch": 0.07313400731340074,
      "grad_norm": 0.0,
      "learning_rate": 1.990306817171177e-05,
      "loss": 1.0538,
      "step": 340
    },
    {
      "epoch": 0.07334910733491073,
      "grad_norm": 0.0,
      "learning_rate": 1.9902098018821504e-05,
      "loss": 1.0083,
      "step": 341
    },
    {
      "epoch": 0.07356420735642073,
      "grad_norm": 0.0,
      "learning_rate": 1.9901123059015843e-05,
      "loss": 0.963,
      "step": 342
    },
    {
      "epoch": 0.07377930737793074,
      "grad_norm": 0.0,
      "learning_rate": 1.9900143292768075e-05,
      "loss": 1.0131,
      "step": 343
    },
    {
      "epoch": 0.07399440739944074,
      "grad_norm": 0.0,
      "learning_rate": 1.9899158720553828e-05,
      "loss": 1.0032,
      "step": 344
    },
    {
      "epoch": 0.07420950742095074,
      "grad_norm": 0.0,
      "learning_rate": 1.9898169342851046e-05,
      "loss": 1.0037,
      "step": 345
    },
    {
      "epoch": 0.07442460744246074,
      "grad_norm": 0.0,
      "learning_rate": 1.9897175160140027e-05,
      "loss": 0.935,
      "step": 346
    },
    {
      "epoch": 0.07463970746397075,
      "grad_norm": 0.0,
      "learning_rate": 1.9896176172903384e-05,
      "loss": 1.0063,
      "step": 347
    },
    {
      "epoch": 0.07485480748548075,
      "grad_norm": 0.0,
      "learning_rate": 1.9895172381626072e-05,
      "loss": 0.9272,
      "step": 348
    },
    {
      "epoch": 0.07506990750699075,
      "grad_norm": 0.0,
      "learning_rate": 1.9894163786795378e-05,
      "loss": 1.0057,
      "step": 349
    },
    {
      "epoch": 0.07528500752850076,
      "grad_norm": 0.0,
      "learning_rate": 1.9893150388900918e-05,
      "loss": 0.9658,
      "step": 350
    },
    {
      "epoch": 0.07550010755001076,
      "grad_norm": 0.0,
      "learning_rate": 1.989213218843464e-05,
      "loss": 0.9152,
      "step": 351
    },
    {
      "epoch": 0.07571520757152075,
      "grad_norm": 0.0,
      "learning_rate": 1.9891109185890814e-05,
      "loss": 0.9858,
      "step": 352
    },
    {
      "epoch": 0.07593030759303077,
      "grad_norm": 0.0,
      "learning_rate": 1.9890081381766066e-05,
      "loss": 0.9606,
      "step": 353
    },
    {
      "epoch": 0.07614540761454076,
      "grad_norm": 0.0,
      "learning_rate": 1.988904877655933e-05,
      "loss": 1.0109,
      "step": 354
    },
    {
      "epoch": 0.07636050763605076,
      "grad_norm": 0.0,
      "learning_rate": 1.9888011370771874e-05,
      "loss": 0.9712,
      "step": 355
    },
    {
      "epoch": 0.07657560765756076,
      "grad_norm": 0.0,
      "learning_rate": 1.9886969164907306e-05,
      "loss": 0.9352,
      "step": 356
    },
    {
      "epoch": 0.07679070767907077,
      "grad_norm": 0.0,
      "learning_rate": 1.988592215947156e-05,
      "loss": 1.0107,
      "step": 357
    },
    {
      "epoch": 0.07700580770058077,
      "grad_norm": 0.0,
      "learning_rate": 1.9884870354972895e-05,
      "loss": 0.9775,
      "step": 358
    },
    {
      "epoch": 0.07722090772209077,
      "grad_norm": 0.0,
      "learning_rate": 1.9883813751921904e-05,
      "loss": 1.0029,
      "step": 359
    },
    {
      "epoch": 0.07743600774360078,
      "grad_norm": 0.0,
      "learning_rate": 1.988275235083151e-05,
      "loss": 0.9722,
      "step": 360
    },
    {
      "epoch": 0.07765110776511078,
      "grad_norm": 0.0,
      "learning_rate": 1.9881686152216962e-05,
      "loss": 1.0135,
      "step": 361
    },
    {
      "epoch": 0.07786620778662078,
      "grad_norm": 0.0,
      "learning_rate": 1.988061515659584e-05,
      "loss": 0.9172,
      "step": 362
    },
    {
      "epoch": 0.07808130780813079,
      "grad_norm": 0.0,
      "learning_rate": 1.9879539364488057e-05,
      "loss": 0.9994,
      "step": 363
    },
    {
      "epoch": 0.07829640782964079,
      "grad_norm": 0.0,
      "learning_rate": 1.9878458776415842e-05,
      "loss": 0.9632,
      "step": 364
    },
    {
      "epoch": 0.07851150785115078,
      "grad_norm": 0.0,
      "learning_rate": 1.9877373392903768e-05,
      "loss": 1.0173,
      "step": 365
    },
    {
      "epoch": 0.07872660787266078,
      "grad_norm": 0.0,
      "learning_rate": 1.9876283214478725e-05,
      "loss": 1.035,
      "step": 366
    },
    {
      "epoch": 0.07894170789417079,
      "grad_norm": 0.0,
      "learning_rate": 1.987518824166993e-05,
      "loss": 1.0533,
      "step": 367
    },
    {
      "epoch": 0.07915680791568079,
      "grad_norm": 0.0,
      "learning_rate": 1.987408847500894e-05,
      "loss": 0.9066,
      "step": 368
    },
    {
      "epoch": 0.07937190793719079,
      "grad_norm": 0.0,
      "learning_rate": 1.9872983915029623e-05,
      "loss": 1.0099,
      "step": 369
    },
    {
      "epoch": 0.0795870079587008,
      "grad_norm": 0.0,
      "learning_rate": 1.987187456226818e-05,
      "loss": 0.9315,
      "step": 370
    },
    {
      "epoch": 0.0798021079802108,
      "grad_norm": 0.0,
      "learning_rate": 1.9870760417263145e-05,
      "loss": 0.986,
      "step": 371
    },
    {
      "epoch": 0.0800172080017208,
      "grad_norm": 0.0,
      "learning_rate": 1.986964148055537e-05,
      "loss": 0.9271,
      "step": 372
    },
    {
      "epoch": 0.08023230802323081,
      "grad_norm": 0.0,
      "learning_rate": 1.9868517752688037e-05,
      "loss": 0.9642,
      "step": 373
    },
    {
      "epoch": 0.0804474080447408,
      "grad_norm": 0.0,
      "learning_rate": 1.9867389234206655e-05,
      "loss": 0.9539,
      "step": 374
    },
    {
      "epoch": 0.0806625080662508,
      "grad_norm": 0.0,
      "learning_rate": 1.9866255925659055e-05,
      "loss": 1.0291,
      "step": 375
    },
    {
      "epoch": 0.0808776080877608,
      "grad_norm": 0.0,
      "learning_rate": 1.986511782759539e-05,
      "loss": 0.9975,
      "step": 376
    },
    {
      "epoch": 0.08109270810927081,
      "grad_norm": 0.0,
      "learning_rate": 1.9863974940568157e-05,
      "loss": 0.9298,
      "step": 377
    },
    {
      "epoch": 0.08130780813078081,
      "grad_norm": 0.0,
      "learning_rate": 1.9862827265132147e-05,
      "loss": 0.9902,
      "step": 378
    },
    {
      "epoch": 0.08152290815229081,
      "grad_norm": 0.0,
      "learning_rate": 1.9861674801844507e-05,
      "loss": 1.0535,
      "step": 379
    },
    {
      "epoch": 0.08173800817380082,
      "grad_norm": 0.0,
      "learning_rate": 1.986051755126468e-05,
      "loss": 0.9997,
      "step": 380
    },
    {
      "epoch": 0.08195310819531082,
      "grad_norm": 0.0,
      "learning_rate": 1.9859355513954458e-05,
      "loss": 0.9858,
      "step": 381
    },
    {
      "epoch": 0.08216820821682082,
      "grad_norm": 0.0,
      "learning_rate": 1.9858188690477938e-05,
      "loss": 1.0153,
      "step": 382
    },
    {
      "epoch": 0.08238330823833083,
      "grad_norm": 0.0,
      "learning_rate": 1.985701708140155e-05,
      "loss": 1.0178,
      "step": 383
    },
    {
      "epoch": 0.08259840825984083,
      "grad_norm": 0.0,
      "learning_rate": 1.9855840687294043e-05,
      "loss": 0.9706,
      "step": 384
    },
    {
      "epoch": 0.08281350828135083,
      "grad_norm": 0.0,
      "learning_rate": 1.9854659508726497e-05,
      "loss": 1.0434,
      "step": 385
    },
    {
      "epoch": 0.08302860830286082,
      "grad_norm": 0.0,
      "learning_rate": 1.98534735462723e-05,
      "loss": 1.0067,
      "step": 386
    },
    {
      "epoch": 0.08324370832437084,
      "grad_norm": 0.0,
      "learning_rate": 1.9852282800507176e-05,
      "loss": 0.9681,
      "step": 387
    },
    {
      "epoch": 0.08345880834588083,
      "grad_norm": 0.0,
      "learning_rate": 1.9851087272009162e-05,
      "loss": 1.0024,
      "step": 388
    },
    {
      "epoch": 0.08367390836739083,
      "grad_norm": 0.0,
      "learning_rate": 1.9849886961358624e-05,
      "loss": 0.9706,
      "step": 389
    },
    {
      "epoch": 0.08388900838890084,
      "grad_norm": 0.0,
      "learning_rate": 1.9848681869138244e-05,
      "loss": 0.9997,
      "step": 390
    },
    {
      "epoch": 0.08410410841041084,
      "grad_norm": 0.0,
      "learning_rate": 1.984747199593302e-05,
      "loss": 0.985,
      "step": 391
    },
    {
      "epoch": 0.08431920843192084,
      "grad_norm": 0.0,
      "learning_rate": 1.984625734233029e-05,
      "loss": 0.9471,
      "step": 392
    },
    {
      "epoch": 0.08453430845343085,
      "grad_norm": 0.0,
      "learning_rate": 1.9845037908919697e-05,
      "loss": 0.9406,
      "step": 393
    },
    {
      "epoch": 0.08474940847494085,
      "grad_norm": 0.0,
      "learning_rate": 1.9843813696293202e-05,
      "loss": 0.9958,
      "step": 394
    },
    {
      "epoch": 0.08496450849645085,
      "grad_norm": 0.0,
      "learning_rate": 1.9842584705045097e-05,
      "loss": 0.9581,
      "step": 395
    },
    {
      "epoch": 0.08517960851796086,
      "grad_norm": 0.0,
      "learning_rate": 1.9841350935771987e-05,
      "loss": 0.9743,
      "step": 396
    },
    {
      "epoch": 0.08539470853947086,
      "grad_norm": 0.0,
      "learning_rate": 1.9840112389072796e-05,
      "loss": 0.9861,
      "step": 397
    },
    {
      "epoch": 0.08560980856098085,
      "grad_norm": 0.0,
      "learning_rate": 1.9838869065548772e-05,
      "loss": 0.9129,
      "step": 398
    },
    {
      "epoch": 0.08582490858249085,
      "grad_norm": 0.0,
      "learning_rate": 1.9837620965803483e-05,
      "loss": 1.0068,
      "step": 399
    },
    {
      "epoch": 0.08604000860400086,
      "grad_norm": 0.0,
      "learning_rate": 1.9836368090442805e-05,
      "loss": 0.9607,
      "step": 400
    },
    {
      "epoch": 0.08625510862551086,
      "grad_norm": 0.0,
      "learning_rate": 1.9835110440074938e-05,
      "loss": 0.9707,
      "step": 401
    },
    {
      "epoch": 0.08647020864702086,
      "grad_norm": 0.0,
      "learning_rate": 1.983384801531041e-05,
      "loss": 1.0328,
      "step": 402
    },
    {
      "epoch": 0.08668530866853087,
      "grad_norm": 0.0,
      "learning_rate": 1.983258081676205e-05,
      "loss": 1.0515,
      "step": 403
    },
    {
      "epoch": 0.08690040869004087,
      "grad_norm": 0.0,
      "learning_rate": 1.983130884504501e-05,
      "loss": 1.0055,
      "step": 404
    },
    {
      "epoch": 0.08711550871155087,
      "grad_norm": 0.0,
      "learning_rate": 1.9830032100776767e-05,
      "loss": 0.9657,
      "step": 405
    },
    {
      "epoch": 0.08733060873306088,
      "grad_norm": 0.0,
      "learning_rate": 1.9828750584577107e-05,
      "loss": 0.996,
      "step": 406
    },
    {
      "epoch": 0.08754570875457088,
      "grad_norm": 0.0,
      "learning_rate": 1.9827464297068137e-05,
      "loss": 1.04,
      "step": 407
    },
    {
      "epoch": 0.08776080877608088,
      "grad_norm": 0.0,
      "learning_rate": 1.9826173238874275e-05,
      "loss": 1.0959,
      "step": 408
    },
    {
      "epoch": 0.08797590879759087,
      "grad_norm": 0.0,
      "learning_rate": 1.9824877410622258e-05,
      "loss": 1.0074,
      "step": 409
    },
    {
      "epoch": 0.08819100881910089,
      "grad_norm": 0.0,
      "learning_rate": 1.982357681294114e-05,
      "loss": 0.981,
      "step": 410
    },
    {
      "epoch": 0.08840610884061088,
      "grad_norm": 0.0,
      "learning_rate": 1.9822271446462283e-05,
      "loss": 1.0075,
      "step": 411
    },
    {
      "epoch": 0.08862120886212088,
      "grad_norm": 0.0,
      "learning_rate": 1.9820961311819376e-05,
      "loss": 0.9987,
      "step": 412
    },
    {
      "epoch": 0.0888363088836309,
      "grad_norm": 0.0,
      "learning_rate": 1.9819646409648415e-05,
      "loss": 0.9566,
      "step": 413
    },
    {
      "epoch": 0.08905140890514089,
      "grad_norm": 0.0,
      "learning_rate": 1.9818326740587707e-05,
      "loss": 1.0033,
      "step": 414
    },
    {
      "epoch": 0.08926650892665089,
      "grad_norm": 0.0,
      "learning_rate": 1.9817002305277883e-05,
      "loss": 1.0157,
      "step": 415
    },
    {
      "epoch": 0.0894816089481609,
      "grad_norm": 0.0,
      "learning_rate": 1.9815673104361876e-05,
      "loss": 1.0215,
      "step": 416
    },
    {
      "epoch": 0.0896967089696709,
      "grad_norm": 0.0,
      "learning_rate": 1.9814339138484948e-05,
      "loss": 0.9552,
      "step": 417
    },
    {
      "epoch": 0.0899118089911809,
      "grad_norm": 0.0,
      "learning_rate": 1.9813000408294656e-05,
      "loss": 0.9753,
      "step": 418
    },
    {
      "epoch": 0.0901269090126909,
      "grad_norm": 0.0,
      "learning_rate": 1.9811656914440885e-05,
      "loss": 0.9532,
      "step": 419
    },
    {
      "epoch": 0.0903420090342009,
      "grad_norm": 0.0,
      "learning_rate": 1.9810308657575822e-05,
      "loss": 0.997,
      "step": 420
    },
    {
      "epoch": 0.0905571090557109,
      "grad_norm": 0.0,
      "learning_rate": 1.9808955638353973e-05,
      "loss": 1.0101,
      "step": 421
    },
    {
      "epoch": 0.0907722090772209,
      "grad_norm": 0.0,
      "learning_rate": 1.9807597857432152e-05,
      "loss": 0.9645,
      "step": 422
    },
    {
      "epoch": 0.09098730909873091,
      "grad_norm": 0.0,
      "learning_rate": 1.9806235315469485e-05,
      "loss": 0.9993,
      "step": 423
    },
    {
      "epoch": 0.09120240912024091,
      "grad_norm": 0.0,
      "learning_rate": 1.980486801312741e-05,
      "loss": 0.9147,
      "step": 424
    },
    {
      "epoch": 0.09141750914175091,
      "grad_norm": 0.0,
      "learning_rate": 1.980349595106968e-05,
      "loss": 0.9377,
      "step": 425
    },
    {
      "epoch": 0.09163260916326092,
      "grad_norm": 0.0,
      "learning_rate": 1.9802119129962352e-05,
      "loss": 0.8853,
      "step": 426
    },
    {
      "epoch": 0.09184770918477092,
      "grad_norm": 0.0,
      "learning_rate": 1.9800737550473793e-05,
      "loss": 1.0028,
      "step": 427
    },
    {
      "epoch": 0.09206280920628092,
      "grad_norm": 0.0,
      "learning_rate": 1.9799351213274688e-05,
      "loss": 0.9541,
      "step": 428
    },
    {
      "epoch": 0.09227790922779092,
      "grad_norm": 0.0,
      "learning_rate": 1.979796011903802e-05,
      "loss": 1.003,
      "step": 429
    },
    {
      "epoch": 0.09249300924930093,
      "grad_norm": 0.0,
      "learning_rate": 1.979656426843909e-05,
      "loss": 0.9735,
      "step": 430
    },
    {
      "epoch": 0.09270810927081093,
      "grad_norm": 0.0,
      "learning_rate": 1.979516366215551e-05,
      "loss": 1.0213,
      "step": 431
    },
    {
      "epoch": 0.09292320929232092,
      "grad_norm": 0.0,
      "learning_rate": 1.979375830086719e-05,
      "loss": 0.989,
      "step": 432
    },
    {
      "epoch": 0.09313830931383094,
      "grad_norm": 0.0,
      "learning_rate": 1.9792348185256356e-05,
      "loss": 1.0623,
      "step": 433
    },
    {
      "epoch": 0.09335340933534093,
      "grad_norm": 0.0,
      "learning_rate": 1.9790933316007543e-05,
      "loss": 0.9875,
      "step": 434
    },
    {
      "epoch": 0.09356850935685093,
      "grad_norm": 0.0,
      "learning_rate": 1.9789513693807584e-05,
      "loss": 0.9279,
      "step": 435
    },
    {
      "epoch": 0.09378360937836094,
      "grad_norm": 0.0,
      "learning_rate": 1.978808931934564e-05,
      "loss": 0.9897,
      "step": 436
    },
    {
      "epoch": 0.09399870939987094,
      "grad_norm": 0.0,
      "learning_rate": 1.978666019331315e-05,
      "loss": 0.9412,
      "step": 437
    },
    {
      "epoch": 0.09421380942138094,
      "grad_norm": 0.0,
      "learning_rate": 1.978522631640388e-05,
      "loss": 1.0034,
      "step": 438
    },
    {
      "epoch": 0.09442890944289094,
      "grad_norm": 0.0,
      "learning_rate": 1.97837876893139e-05,
      "loss": 1.0261,
      "step": 439
    },
    {
      "epoch": 0.09464400946440095,
      "grad_norm": 0.0,
      "learning_rate": 1.9782344312741585e-05,
      "loss": 1.0006,
      "step": 440
    },
    {
      "epoch": 0.09485910948591095,
      "grad_norm": 0.0,
      "learning_rate": 1.9780896187387606e-05,
      "loss": 0.9482,
      "step": 441
    },
    {
      "epoch": 0.09507420950742095,
      "grad_norm": 0.0,
      "learning_rate": 1.9779443313954957e-05,
      "loss": 0.9839,
      "step": 442
    },
    {
      "epoch": 0.09528930952893096,
      "grad_norm": 0.0,
      "learning_rate": 1.9777985693148912e-05,
      "loss": 0.9559,
      "step": 443
    },
    {
      "epoch": 0.09550440955044096,
      "grad_norm": 0.0,
      "learning_rate": 1.9776523325677083e-05,
      "loss": 0.8739,
      "step": 444
    },
    {
      "epoch": 0.09571950957195095,
      "grad_norm": 0.0,
      "learning_rate": 1.977505621224935e-05,
      "loss": 0.9316,
      "step": 445
    },
    {
      "epoch": 0.09593460959346096,
      "grad_norm": 0.0,
      "learning_rate": 1.9773584353577928e-05,
      "loss": 0.9773,
      "step": 446
    },
    {
      "epoch": 0.09614970961497096,
      "grad_norm": 0.0,
      "learning_rate": 1.9772107750377315e-05,
      "loss": 1.0232,
      "step": 447
    },
    {
      "epoch": 0.09636480963648096,
      "grad_norm": 0.0,
      "learning_rate": 1.977062640336432e-05,
      "loss": 0.9707,
      "step": 448
    },
    {
      "epoch": 0.09657990965799097,
      "grad_norm": 0.0,
      "learning_rate": 1.976914031325806e-05,
      "loss": 0.9906,
      "step": 449
    },
    {
      "epoch": 0.09679500967950097,
      "grad_norm": 0.0,
      "learning_rate": 1.9767649480779944e-05,
      "loss": 0.9911,
      "step": 450
    },
    {
      "epoch": 0.09701010970101097,
      "grad_norm": 0.0,
      "learning_rate": 1.9766153906653684e-05,
      "loss": 0.9398,
      "step": 451
    },
    {
      "epoch": 0.09722520972252097,
      "grad_norm": 0.0,
      "learning_rate": 1.9764653591605306e-05,
      "loss": 1.0622,
      "step": 452
    },
    {
      "epoch": 0.09744030974403098,
      "grad_norm": 0.0,
      "learning_rate": 1.976314853636312e-05,
      "loss": 0.9937,
      "step": 453
    },
    {
      "epoch": 0.09765540976554098,
      "grad_norm": 0.0,
      "learning_rate": 1.9761638741657757e-05,
      "loss": 0.874,
      "step": 454
    },
    {
      "epoch": 0.09787050978705097,
      "grad_norm": 0.0,
      "learning_rate": 1.9760124208222128e-05,
      "loss": 0.9683,
      "step": 455
    },
    {
      "epoch": 0.09808560980856099,
      "grad_norm": 0.0,
      "learning_rate": 1.9758604936791462e-05,
      "loss": 1.0063,
      "step": 456
    },
    {
      "epoch": 0.09830070983007098,
      "grad_norm": 0.0,
      "learning_rate": 1.9757080928103273e-05,
      "loss": 1.0373,
      "step": 457
    },
    {
      "epoch": 0.09851580985158098,
      "grad_norm": 0.0,
      "learning_rate": 1.9755552182897388e-05,
      "loss": 1.0465,
      "step": 458
    },
    {
      "epoch": 0.098730909873091,
      "grad_norm": 0.0,
      "learning_rate": 1.9754018701915926e-05,
      "loss": 0.9567,
      "step": 459
    },
    {
      "epoch": 0.09894600989460099,
      "grad_norm": 0.0,
      "learning_rate": 1.9752480485903306e-05,
      "loss": 0.9935,
      "step": 460
    },
    {
      "epoch": 0.09916110991611099,
      "grad_norm": 0.0,
      "learning_rate": 1.9750937535606246e-05,
      "loss": 0.9636,
      "step": 461
    },
    {
      "epoch": 0.09937620993762099,
      "grad_norm": 0.0,
      "learning_rate": 1.974938985177376e-05,
      "loss": 0.9897,
      "step": 462
    },
    {
      "epoch": 0.099591309959131,
      "grad_norm": 0.0,
      "learning_rate": 1.9747837435157168e-05,
      "loss": 0.9958,
      "step": 463
    },
    {
      "epoch": 0.099806409980641,
      "grad_norm": 0.0,
      "learning_rate": 1.974628028651007e-05,
      "loss": 0.9621,
      "step": 464
    },
    {
      "epoch": 0.100021510002151,
      "grad_norm": 0.0,
      "learning_rate": 1.974471840658839e-05,
      "loss": 0.9367,
      "step": 465
    },
    {
      "epoch": 0.10023661002366101,
      "grad_norm": 0.0,
      "learning_rate": 1.9743151796150323e-05,
      "loss": 1.0421,
      "step": 466
    },
    {
      "epoch": 0.100451710045171,
      "grad_norm": 0.0,
      "learning_rate": 1.974158045595637e-05,
      "loss": 0.9802,
      "step": 467
    },
    {
      "epoch": 0.100666810066681,
      "grad_norm": 0.0,
      "learning_rate": 1.9740004386769337e-05,
      "loss": 0.9193,
      "step": 468
    },
    {
      "epoch": 0.10088191008819102,
      "grad_norm": 0.0,
      "learning_rate": 1.9738423589354314e-05,
      "loss": 0.9592,
      "step": 469
    },
    {
      "epoch": 0.10109701010970101,
      "grad_norm": 0.0,
      "learning_rate": 1.9736838064478683e-05,
      "loss": 0.9638,
      "step": 470
    },
    {
      "epoch": 0.10131211013121101,
      "grad_norm": 0.0,
      "learning_rate": 1.9735247812912138e-05,
      "loss": 0.9706,
      "step": 471
    },
    {
      "epoch": 0.10152721015272101,
      "grad_norm": 0.0,
      "learning_rate": 1.973365283542665e-05,
      "loss": 0.9822,
      "step": 472
    },
    {
      "epoch": 0.10174231017423102,
      "grad_norm": 0.0,
      "learning_rate": 1.9732053132796496e-05,
      "loss": 0.9993,
      "step": 473
    },
    {
      "epoch": 0.10195741019574102,
      "grad_norm": 0.0,
      "learning_rate": 1.973044870579824e-05,
      "loss": 0.9414,
      "step": 474
    },
    {
      "epoch": 0.10217251021725102,
      "grad_norm": 0.0,
      "learning_rate": 1.9728839555210744e-05,
      "loss": 0.9384,
      "step": 475
    },
    {
      "epoch": 0.10238761023876103,
      "grad_norm": 0.0,
      "learning_rate": 1.9727225681815154e-05,
      "loss": 0.9645,
      "step": 476
    },
    {
      "epoch": 0.10260271026027103,
      "grad_norm": 0.0,
      "learning_rate": 1.9725607086394923e-05,
      "loss": 0.9927,
      "step": 477
    },
    {
      "epoch": 0.10281781028178102,
      "grad_norm": 0.0,
      "learning_rate": 1.972398376973579e-05,
      "loss": 0.9913,
      "step": 478
    },
    {
      "epoch": 0.10303291030329104,
      "grad_norm": 0.0,
      "learning_rate": 1.9722355732625775e-05,
      "loss": 0.974,
      "step": 479
    },
    {
      "epoch": 0.10324801032480103,
      "grad_norm": 0.0,
      "learning_rate": 1.9720722975855206e-05,
      "loss": 0.9496,
      "step": 480
    },
    {
      "epoch": 0.10346311034631103,
      "grad_norm": 0.0,
      "learning_rate": 1.9719085500216694e-05,
      "loss": 1.0039,
      "step": 481
    },
    {
      "epoch": 0.10367821036782103,
      "grad_norm": 0.0,
      "learning_rate": 1.9717443306505143e-05,
      "loss": 1.0526,
      "step": 482
    },
    {
      "epoch": 0.10389331038933104,
      "grad_norm": 0.0,
      "learning_rate": 1.971579639551774e-05,
      "loss": 0.9164,
      "step": 483
    },
    {
      "epoch": 0.10410841041084104,
      "grad_norm": 0.0,
      "learning_rate": 1.9714144768053978e-05,
      "loss": 0.9875,
      "step": 484
    },
    {
      "epoch": 0.10432351043235104,
      "grad_norm": 0.0,
      "learning_rate": 1.9712488424915625e-05,
      "loss": 0.9767,
      "step": 485
    },
    {
      "epoch": 0.10453861045386105,
      "grad_norm": 0.0,
      "learning_rate": 1.9710827366906743e-05,
      "loss": 0.9997,
      "step": 486
    },
    {
      "epoch": 0.10475371047537105,
      "grad_norm": 0.0,
      "learning_rate": 1.9709161594833682e-05,
      "loss": 0.9569,
      "step": 487
    },
    {
      "epoch": 0.10496881049688105,
      "grad_norm": 0.0,
      "learning_rate": 1.9707491109505083e-05,
      "loss": 0.9012,
      "step": 488
    },
    {
      "epoch": 0.10518391051839106,
      "grad_norm": 0.0,
      "learning_rate": 1.9705815911731873e-05,
      "loss": 0.9706,
      "step": 489
    },
    {
      "epoch": 0.10539901053990106,
      "grad_norm": 0.0,
      "learning_rate": 1.9704136002327262e-05,
      "loss": 0.9529,
      "step": 490
    },
    {
      "epoch": 0.10561411056141105,
      "grad_norm": 0.0,
      "learning_rate": 1.9702451382106763e-05,
      "loss": 1.0382,
      "step": 491
    },
    {
      "epoch": 0.10582921058292105,
      "grad_norm": 0.0,
      "learning_rate": 1.9700762051888153e-05,
      "loss": 1.0273,
      "step": 492
    },
    {
      "epoch": 0.10604431060443106,
      "grad_norm": 0.0,
      "learning_rate": 1.9699068012491515e-05,
      "loss": 0.9511,
      "step": 493
    },
    {
      "epoch": 0.10625941062594106,
      "grad_norm": 0.0,
      "learning_rate": 1.9697369264739213e-05,
      "loss": 0.9633,
      "step": 494
    },
    {
      "epoch": 0.10647451064745106,
      "grad_norm": 0.0,
      "learning_rate": 1.9695665809455885e-05,
      "loss": 1.0158,
      "step": 495
    },
    {
      "epoch": 0.10668961066896107,
      "grad_norm": 0.0,
      "learning_rate": 1.969395764746847e-05,
      "loss": 0.9885,
      "step": 496
    },
    {
      "epoch": 0.10690471069047107,
      "grad_norm": 0.0,
      "learning_rate": 1.9692244779606185e-05,
      "loss": 0.9199,
      "step": 497
    },
    {
      "epoch": 0.10711981071198107,
      "grad_norm": 0.0,
      "learning_rate": 1.9690527206700526e-05,
      "loss": 1.0063,
      "step": 498
    },
    {
      "epoch": 0.10733491073349108,
      "grad_norm": 0.0,
      "learning_rate": 1.9688804929585284e-05,
      "loss": 0.9782,
      "step": 499
    },
    {
      "epoch": 0.10755001075500108,
      "grad_norm": 0.0,
      "learning_rate": 1.9687077949096528e-05,
      "loss": 0.9492,
      "step": 500
    },
    {
      "epoch": 0.10776511077651107,
      "grad_norm": 0.0,
      "learning_rate": 1.968534626607261e-05,
      "loss": 0.9589,
      "step": 501
    },
    {
      "epoch": 0.10798021079802107,
      "grad_norm": 0.0,
      "learning_rate": 1.9683609881354164e-05,
      "loss": 1.066,
      "step": 502
    },
    {
      "epoch": 0.10819531081953108,
      "grad_norm": 0.0,
      "learning_rate": 1.9681868795784108e-05,
      "loss": 0.9504,
      "step": 503
    },
    {
      "epoch": 0.10841041084104108,
      "grad_norm": 0.0,
      "learning_rate": 1.968012301020765e-05,
      "loss": 0.9708,
      "step": 504
    },
    {
      "epoch": 0.10862551086255108,
      "grad_norm": 0.0,
      "learning_rate": 1.9678372525472252e-05,
      "loss": 0.9475,
      "step": 505
    },
    {
      "epoch": 0.10884061088406109,
      "grad_norm": 0.0,
      "learning_rate": 1.9676617342427696e-05,
      "loss": 0.9626,
      "step": 506
    },
    {
      "epoch": 0.10905571090557109,
      "grad_norm": 0.0,
      "learning_rate": 1.9674857461926015e-05,
      "loss": 1.0265,
      "step": 507
    },
    {
      "epoch": 0.10927081092708109,
      "grad_norm": 0.0,
      "learning_rate": 1.9673092884821538e-05,
      "loss": 0.9892,
      "step": 508
    },
    {
      "epoch": 0.1094859109485911,
      "grad_norm": 0.0,
      "learning_rate": 1.9671323611970864e-05,
      "loss": 0.9873,
      "step": 509
    },
    {
      "epoch": 0.1097010109701011,
      "grad_norm": 0.0,
      "learning_rate": 1.9669549644232876e-05,
      "loss": 0.9576,
      "step": 510
    },
    {
      "epoch": 0.1099161109916111,
      "grad_norm": 0.0,
      "learning_rate": 1.966777098246874e-05,
      "loss": 0.9919,
      "step": 511
    },
    {
      "epoch": 0.11013121101312111,
      "grad_norm": 0.0,
      "learning_rate": 1.9665987627541895e-05,
      "loss": 0.9491,
      "step": 512
    },
    {
      "epoch": 0.1103463110346311,
      "grad_norm": 0.0,
      "learning_rate": 1.9664199580318062e-05,
      "loss": 1.0039,
      "step": 513
    },
    {
      "epoch": 0.1105614110561411,
      "grad_norm": 0.0,
      "learning_rate": 1.9662406841665235e-05,
      "loss": 1.0083,
      "step": 514
    },
    {
      "epoch": 0.1107765110776511,
      "grad_norm": 0.0,
      "learning_rate": 1.9660609412453692e-05,
      "loss": 0.9387,
      "step": 515
    },
    {
      "epoch": 0.11099161109916111,
      "grad_norm": 0.0,
      "learning_rate": 1.9658807293555983e-05,
      "loss": 0.9786,
      "step": 516
    },
    {
      "epoch": 0.11120671112067111,
      "grad_norm": 0.0,
      "learning_rate": 1.9657000485846933e-05,
      "loss": 1.0162,
      "step": 517
    },
    {
      "epoch": 0.11142181114218111,
      "grad_norm": 0.0,
      "learning_rate": 1.9655188990203648e-05,
      "loss": 0.9827,
      "step": 518
    },
    {
      "epoch": 0.11163691116369112,
      "grad_norm": 0.0,
      "learning_rate": 1.965337280750551e-05,
      "loss": 0.9913,
      "step": 519
    },
    {
      "epoch": 0.11185201118520112,
      "grad_norm": 0.0,
      "learning_rate": 1.9651551938634173e-05,
      "loss": 0.9308,
      "step": 520
    },
    {
      "epoch": 0.11206711120671112,
      "grad_norm": 0.0,
      "learning_rate": 1.9649726384473568e-05,
      "loss": 0.94,
      "step": 521
    },
    {
      "epoch": 0.11228221122822113,
      "grad_norm": 0.0,
      "learning_rate": 1.9647896145909893e-05,
      "loss": 0.9687,
      "step": 522
    },
    {
      "epoch": 0.11249731124973113,
      "grad_norm": 0.0,
      "learning_rate": 1.964606122383164e-05,
      "loss": 0.9768,
      "step": 523
    },
    {
      "epoch": 0.11271241127124113,
      "grad_norm": 0.0,
      "learning_rate": 1.964422161912955e-05,
      "loss": 1.0694,
      "step": 524
    },
    {
      "epoch": 0.11292751129275112,
      "grad_norm": 0.0,
      "learning_rate": 1.964237733269665e-05,
      "loss": 0.9917,
      "step": 525
    },
    {
      "epoch": 0.11314261131426114,
      "grad_norm": 0.0,
      "learning_rate": 1.964052836542824e-05,
      "loss": 0.978,
      "step": 526
    },
    {
      "epoch": 0.11335771133577113,
      "grad_norm": 0.0,
      "learning_rate": 1.963867471822189e-05,
      "loss": 0.9658,
      "step": 527
    },
    {
      "epoch": 0.11357281135728113,
      "grad_norm": 0.0,
      "learning_rate": 1.9636816391977444e-05,
      "loss": 0.985,
      "step": 528
    },
    {
      "epoch": 0.11378791137879114,
      "grad_norm": 0.0,
      "learning_rate": 1.9634953387597014e-05,
      "loss": 0.9812,
      "step": 529
    },
    {
      "epoch": 0.11400301140030114,
      "grad_norm": 0.0,
      "learning_rate": 1.963308570598498e-05,
      "loss": 0.9539,
      "step": 530
    },
    {
      "epoch": 0.11421811142181114,
      "grad_norm": 0.0,
      "learning_rate": 1.9631213348048005e-05,
      "loss": 1.0291,
      "step": 531
    },
    {
      "epoch": 0.11443321144332115,
      "grad_norm": 0.0,
      "learning_rate": 1.962933631469501e-05,
      "loss": 0.9588,
      "step": 532
    },
    {
      "epoch": 0.11464831146483115,
      "grad_norm": 0.0,
      "learning_rate": 1.9627454606837187e-05,
      "loss": 0.9728,
      "step": 533
    },
    {
      "epoch": 0.11486341148634115,
      "grad_norm": 0.0,
      "learning_rate": 1.9625568225388006e-05,
      "loss": 0.9007,
      "step": 534
    },
    {
      "epoch": 0.11507851150785114,
      "grad_norm": 0.0,
      "learning_rate": 1.9623677171263198e-05,
      "loss": 0.9612,
      "step": 535
    },
    {
      "epoch": 0.11529361152936116,
      "grad_norm": 0.0,
      "learning_rate": 1.962178144538076e-05,
      "loss": 0.9766,
      "step": 536
    },
    {
      "epoch": 0.11550871155087115,
      "grad_norm": 0.0,
      "learning_rate": 1.961988104866097e-05,
      "loss": 0.9679,
      "step": 537
    },
    {
      "epoch": 0.11572381157238115,
      "grad_norm": 0.0,
      "learning_rate": 1.9617975982026352e-05,
      "loss": 0.987,
      "step": 538
    },
    {
      "epoch": 0.11593891159389116,
      "grad_norm": 0.0,
      "learning_rate": 1.961606624640172e-05,
      "loss": 0.9963,
      "step": 539
    },
    {
      "epoch": 0.11615401161540116,
      "grad_norm": 0.0,
      "learning_rate": 1.961415184271414e-05,
      "loss": 0.9571,
      "step": 540
    },
    {
      "epoch": 0.11636911163691116,
      "grad_norm": 0.0,
      "learning_rate": 1.961223277189295e-05,
      "loss": 0.9364,
      "step": 541
    },
    {
      "epoch": 0.11658421165842117,
      "grad_norm": 0.0,
      "learning_rate": 1.9610309034869743e-05,
      "loss": 1.0093,
      "step": 542
    },
    {
      "epoch": 0.11679931167993117,
      "grad_norm": 0.0,
      "learning_rate": 1.96083806325784e-05,
      "loss": 0.9627,
      "step": 543
    },
    {
      "epoch": 0.11701441170144117,
      "grad_norm": 0.0,
      "learning_rate": 1.960644756595504e-05,
      "loss": 0.9874,
      "step": 544
    },
    {
      "epoch": 0.11722951172295117,
      "grad_norm": 0.0,
      "learning_rate": 1.9604509835938066e-05,
      "loss": 0.9323,
      "step": 545
    },
    {
      "epoch": 0.11744461174446118,
      "grad_norm": 0.0,
      "learning_rate": 1.9602567443468136e-05,
      "loss": 0.9896,
      "step": 546
    },
    {
      "epoch": 0.11765971176597118,
      "grad_norm": 0.0,
      "learning_rate": 1.9600620389488173e-05,
      "loss": 0.924,
      "step": 547
    },
    {
      "epoch": 0.11787481178748117,
      "grad_norm": 0.0,
      "learning_rate": 1.9598668674943362e-05,
      "loss": 0.9835,
      "step": 548
    },
    {
      "epoch": 0.11808991180899119,
      "grad_norm": 0.0,
      "learning_rate": 1.959671230078115e-05,
      "loss": 0.9345,
      "step": 549
    },
    {
      "epoch": 0.11830501183050118,
      "grad_norm": 0.0,
      "learning_rate": 1.9594751267951256e-05,
      "loss": 1.0156,
      "step": 550
    },
    {
      "epoch": 0.11852011185201118,
      "grad_norm": 0.0,
      "learning_rate": 1.959278557740564e-05,
      "loss": 0.9889,
      "step": 551
    },
    {
      "epoch": 0.1187352118735212,
      "grad_norm": 0.0,
      "learning_rate": 1.959081523009854e-05,
      "loss": 0.9342,
      "step": 552
    },
    {
      "epoch": 0.11895031189503119,
      "grad_norm": 0.0,
      "learning_rate": 1.9588840226986453e-05,
      "loss": 0.9374,
      "step": 553
    },
    {
      "epoch": 0.11916541191654119,
      "grad_norm": 0.0,
      "learning_rate": 1.9586860569028126e-05,
      "loss": 0.9053,
      "step": 554
    },
    {
      "epoch": 0.11938051193805119,
      "grad_norm": 0.0,
      "learning_rate": 1.958487625718458e-05,
      "loss": 1.0175,
      "step": 555
    },
    {
      "epoch": 0.1195956119595612,
      "grad_norm": 0.0,
      "learning_rate": 1.958288729241908e-05,
      "loss": 0.9451,
      "step": 556
    },
    {
      "epoch": 0.1198107119810712,
      "grad_norm": 0.0,
      "learning_rate": 1.958089367569716e-05,
      "loss": 0.9003,
      "step": 557
    },
    {
      "epoch": 0.1200258120025812,
      "grad_norm": 0.0,
      "learning_rate": 1.9578895407986612e-05,
      "loss": 0.9626,
      "step": 558
    },
    {
      "epoch": 0.1202409120240912,
      "grad_norm": 0.0,
      "learning_rate": 1.9576892490257483e-05,
      "loss": 0.9562,
      "step": 559
    },
    {
      "epoch": 0.1204560120456012,
      "grad_norm": 0.0,
      "learning_rate": 1.9574884923482073e-05,
      "loss": 1.0095,
      "step": 560
    },
    {
      "epoch": 0.1206711120671112,
      "grad_norm": 0.0,
      "learning_rate": 1.957287270863495e-05,
      "loss": 0.96,
      "step": 561
    },
    {
      "epoch": 0.12088621208862121,
      "grad_norm": 0.0,
      "learning_rate": 1.9570855846692924e-05,
      "loss": 0.9843,
      "step": 562
    },
    {
      "epoch": 0.12110131211013121,
      "grad_norm": 0.0,
      "learning_rate": 1.956883433863508e-05,
      "loss": 0.9568,
      "step": 563
    },
    {
      "epoch": 0.12131641213164121,
      "grad_norm": 0.0,
      "learning_rate": 1.9566808185442735e-05,
      "loss": 0.9515,
      "step": 564
    },
    {
      "epoch": 0.12153151215315122,
      "grad_norm": 0.0,
      "learning_rate": 1.956477738809948e-05,
      "loss": 0.961,
      "step": 565
    },
    {
      "epoch": 0.12174661217466122,
      "grad_norm": 0.0,
      "learning_rate": 1.9562741947591154e-05,
      "loss": 0.897,
      "step": 566
    },
    {
      "epoch": 0.12196171219617122,
      "grad_norm": 0.0,
      "learning_rate": 1.9560701864905844e-05,
      "loss": 0.9255,
      "step": 567
    },
    {
      "epoch": 0.12217681221768122,
      "grad_norm": 0.0,
      "learning_rate": 1.9558657141033905e-05,
      "loss": 0.9643,
      "step": 568
    },
    {
      "epoch": 0.12239191223919123,
      "grad_norm": 0.0,
      "learning_rate": 1.955660777696793e-05,
      "loss": 0.9389,
      "step": 569
    },
    {
      "epoch": 0.12260701226070123,
      "grad_norm": 0.0,
      "learning_rate": 1.9554553773702773e-05,
      "loss": 0.9991,
      "step": 570
    },
    {
      "epoch": 0.12282211228221122,
      "grad_norm": 0.0,
      "learning_rate": 1.9552495132235537e-05,
      "loss": 0.9408,
      "step": 571
    },
    {
      "epoch": 0.12303721230372124,
      "grad_norm": 0.0,
      "learning_rate": 1.9550431853565577e-05,
      "loss": 1.0082,
      "step": 572
    },
    {
      "epoch": 0.12325231232523123,
      "grad_norm": 0.0,
      "learning_rate": 1.95483639386945e-05,
      "loss": 1.0204,
      "step": 573
    },
    {
      "epoch": 0.12346741234674123,
      "grad_norm": 0.0,
      "learning_rate": 1.954629138862616e-05,
      "loss": 1.0123,
      "step": 574
    },
    {
      "epoch": 0.12368251236825124,
      "grad_norm": 0.0,
      "learning_rate": 1.9544214204366675e-05,
      "loss": 0.9291,
      "step": 575
    },
    {
      "epoch": 0.12389761238976124,
      "grad_norm": 0.0,
      "learning_rate": 1.9542132386924387e-05,
      "loss": 0.9825,
      "step": 576
    },
    {
      "epoch": 0.12411271241127124,
      "grad_norm": 0.0,
      "learning_rate": 1.9540045937309914e-05,
      "loss": 1.0535,
      "step": 577
    },
    {
      "epoch": 0.12432781243278124,
      "grad_norm": 0.0,
      "learning_rate": 1.9537954856536104e-05,
      "loss": 0.9593,
      "step": 578
    },
    {
      "epoch": 0.12454291245429125,
      "grad_norm": 0.0,
      "learning_rate": 1.953585914561806e-05,
      "loss": 0.917,
      "step": 579
    },
    {
      "epoch": 0.12475801247580125,
      "grad_norm": 0.0,
      "learning_rate": 1.953375880557314e-05,
      "loss": 0.9554,
      "step": 580
    },
    {
      "epoch": 0.12497311249731125,
      "grad_norm": 0.0,
      "learning_rate": 1.9531653837420932e-05,
      "loss": 0.956,
      "step": 581
    },
    {
      "epoch": 0.12518821251882126,
      "grad_norm": 0.0,
      "learning_rate": 1.9529544242183288e-05,
      "loss": 0.956,
      "step": 582
    },
    {
      "epoch": 0.12540331254033124,
      "grad_norm": 0.0,
      "learning_rate": 1.9527430020884293e-05,
      "loss": 0.9564,
      "step": 583
    },
    {
      "epoch": 0.12561841256184125,
      "grad_norm": 0.0,
      "learning_rate": 1.9525311174550286e-05,
      "loss": 1.0534,
      "step": 584
    },
    {
      "epoch": 0.12583351258335126,
      "grad_norm": 0.0,
      "learning_rate": 1.952318770420985e-05,
      "loss": 0.9951,
      "step": 585
    },
    {
      "epoch": 0.12604861260486125,
      "grad_norm": 0.0,
      "learning_rate": 1.9521059610893805e-05,
      "loss": 0.9304,
      "step": 586
    },
    {
      "epoch": 0.12626371262637126,
      "grad_norm": 0.0,
      "learning_rate": 1.9518926895635232e-05,
      "loss": 0.9684,
      "step": 587
    },
    {
      "epoch": 0.12647881264788127,
      "grad_norm": 0.0,
      "learning_rate": 1.951678955946944e-05,
      "loss": 1.0583,
      "step": 588
    },
    {
      "epoch": 0.12669391266939126,
      "grad_norm": 0.0,
      "learning_rate": 1.951464760343398e-05,
      "loss": 0.9435,
      "step": 589
    },
    {
      "epoch": 0.12690901269090127,
      "grad_norm": 0.0,
      "learning_rate": 1.951250102856866e-05,
      "loss": 1.0754,
      "step": 590
    },
    {
      "epoch": 0.12712411271241128,
      "grad_norm": 0.0,
      "learning_rate": 1.9510349835915517e-05,
      "loss": 0.9779,
      "step": 591
    },
    {
      "epoch": 0.12733921273392126,
      "grad_norm": 0.0,
      "learning_rate": 1.9508194026518837e-05,
      "loss": 0.9579,
      "step": 592
    },
    {
      "epoch": 0.12755431275543128,
      "grad_norm": 0.0,
      "learning_rate": 1.9506033601425146e-05,
      "loss": 1.0214,
      "step": 593
    },
    {
      "epoch": 0.1277694127769413,
      "grad_norm": 0.0,
      "learning_rate": 1.950386856168321e-05,
      "loss": 0.9662,
      "step": 594
    },
    {
      "epoch": 0.12798451279845127,
      "grad_norm": 0.0,
      "learning_rate": 1.9501698908344036e-05,
      "loss": 0.9056,
      "step": 595
    },
    {
      "epoch": 0.12819961281996128,
      "grad_norm": 0.0,
      "learning_rate": 1.9499524642460868e-05,
      "loss": 0.9119,
      "step": 596
    },
    {
      "epoch": 0.1284147128414713,
      "grad_norm": 0.0,
      "learning_rate": 1.9497345765089188e-05,
      "loss": 0.9728,
      "step": 597
    },
    {
      "epoch": 0.12862981286298128,
      "grad_norm": 0.0,
      "learning_rate": 1.949516227728672e-05,
      "loss": 1.044,
      "step": 598
    },
    {
      "epoch": 0.1288449128844913,
      "grad_norm": 0.0,
      "learning_rate": 1.9492974180113425e-05,
      "loss": 0.9659,
      "step": 599
    },
    {
      "epoch": 0.1290600129060013,
      "grad_norm": 0.0,
      "learning_rate": 1.9490781474631508e-05,
      "loss": 1.0224,
      "step": 600
    },
    {
      "epoch": 0.1292751129275113,
      "grad_norm": 0.0,
      "learning_rate": 1.94885841619054e-05,
      "loss": 0.9803,
      "step": 601
    },
    {
      "epoch": 0.1294902129490213,
      "grad_norm": 0.0,
      "learning_rate": 1.9486382243001773e-05,
      "loss": 1.012,
      "step": 602
    },
    {
      "epoch": 0.12970531297053128,
      "grad_norm": 0.0,
      "learning_rate": 1.948417571898954e-05,
      "loss": 0.9761,
      "step": 603
    },
    {
      "epoch": 0.1299204129920413,
      "grad_norm": 0.0,
      "learning_rate": 1.9481964590939836e-05,
      "loss": 0.9613,
      "step": 604
    },
    {
      "epoch": 0.1301355130135513,
      "grad_norm": 0.0,
      "learning_rate": 1.947974885992605e-05,
      "loss": 0.9733,
      "step": 605
    },
    {
      "epoch": 0.1303506130350613,
      "grad_norm": 0.0,
      "learning_rate": 1.947752852702379e-05,
      "loss": 0.9728,
      "step": 606
    },
    {
      "epoch": 0.1305657130565713,
      "grad_norm": 0.0,
      "learning_rate": 1.9475303593310904e-05,
      "loss": 0.9834,
      "step": 607
    },
    {
      "epoch": 0.13078081307808131,
      "grad_norm": 0.0,
      "learning_rate": 1.9473074059867476e-05,
      "loss": 0.9025,
      "step": 608
    },
    {
      "epoch": 0.1309959130995913,
      "grad_norm": 0.0,
      "learning_rate": 1.9470839927775817e-05,
      "loss": 0.9469,
      "step": 609
    },
    {
      "epoch": 0.1312110131211013,
      "grad_norm": 0.0,
      "learning_rate": 1.9468601198120476e-05,
      "loss": 0.9362,
      "step": 610
    },
    {
      "epoch": 0.13142611314261132,
      "grad_norm": 0.0,
      "learning_rate": 1.946635787198823e-05,
      "loss": 1.0223,
      "step": 611
    },
    {
      "epoch": 0.1316412131641213,
      "grad_norm": 0.0,
      "learning_rate": 1.9464109950468088e-05,
      "loss": 1.0169,
      "step": 612
    },
    {
      "epoch": 0.13185631318563132,
      "grad_norm": 0.0,
      "learning_rate": 1.946185743465129e-05,
      "loss": 1.0391,
      "step": 613
    },
    {
      "epoch": 0.13207141320714133,
      "grad_norm": 0.0,
      "learning_rate": 1.9459600325631306e-05,
      "loss": 1.0028,
      "step": 614
    },
    {
      "epoch": 0.13228651322865131,
      "grad_norm": 0.0,
      "learning_rate": 1.9457338624503838e-05,
      "loss": 1.0014,
      "step": 615
    },
    {
      "epoch": 0.13250161325016133,
      "grad_norm": 0.0,
      "learning_rate": 1.9455072332366817e-05,
      "loss": 0.9304,
      "step": 616
    },
    {
      "epoch": 0.13271671327167134,
      "grad_norm": 0.0,
      "learning_rate": 1.9452801450320398e-05,
      "loss": 1.0144,
      "step": 617
    },
    {
      "epoch": 0.13293181329318132,
      "grad_norm": 0.0,
      "learning_rate": 1.9450525979466967e-05,
      "loss": 0.9322,
      "step": 618
    },
    {
      "epoch": 0.13314691331469133,
      "grad_norm": 0.0,
      "learning_rate": 1.944824592091114e-05,
      "loss": 0.96,
      "step": 619
    },
    {
      "epoch": 0.13336201333620135,
      "grad_norm": 0.0,
      "learning_rate": 1.9445961275759758e-05,
      "loss": 0.9633,
      "step": 620
    },
    {
      "epoch": 0.13357711335771133,
      "grad_norm": 0.0,
      "learning_rate": 1.9443672045121887e-05,
      "loss": 0.8974,
      "step": 621
    },
    {
      "epoch": 0.13379221337922134,
      "grad_norm": 0.0,
      "learning_rate": 1.9441378230108826e-05,
      "loss": 0.9706,
      "step": 622
    },
    {
      "epoch": 0.13400731340073135,
      "grad_norm": 0.0,
      "learning_rate": 1.9439079831834087e-05,
      "loss": 0.9223,
      "step": 623
    },
    {
      "epoch": 0.13422241342224134,
      "grad_norm": 0.0,
      "learning_rate": 1.9436776851413416e-05,
      "loss": 0.9665,
      "step": 624
    },
    {
      "epoch": 0.13443751344375135,
      "grad_norm": 0.0,
      "learning_rate": 1.9434469289964782e-05,
      "loss": 0.9267,
      "step": 625
    },
    {
      "epoch": 0.13465261346526133,
      "grad_norm": 0.0,
      "learning_rate": 1.943215714860838e-05,
      "loss": 0.9456,
      "step": 626
    },
    {
      "epoch": 0.13486771348677135,
      "grad_norm": 0.0,
      "learning_rate": 1.942984042846662e-05,
      "loss": 0.9737,
      "step": 627
    },
    {
      "epoch": 0.13508281350828136,
      "grad_norm": 0.0,
      "learning_rate": 1.9427519130664145e-05,
      "loss": 0.9565,
      "step": 628
    },
    {
      "epoch": 0.13529791352979134,
      "grad_norm": 0.0,
      "learning_rate": 1.9425193256327812e-05,
      "loss": 0.9381,
      "step": 629
    },
    {
      "epoch": 0.13551301355130135,
      "grad_norm": 0.0,
      "learning_rate": 1.9422862806586703e-05,
      "loss": 1.0323,
      "step": 630
    },
    {
      "epoch": 0.13572811357281137,
      "grad_norm": 0.0,
      "learning_rate": 1.9420527782572126e-05,
      "loss": 0.995,
      "step": 631
    },
    {
      "epoch": 0.13594321359432135,
      "grad_norm": 0.0,
      "learning_rate": 1.94181881854176e-05,
      "loss": 0.9846,
      "step": 632
    },
    {
      "epoch": 0.13615831361583136,
      "grad_norm": 0.0,
      "learning_rate": 1.9415844016258867e-05,
      "loss": 0.8896,
      "step": 633
    },
    {
      "epoch": 0.13637341363734137,
      "grad_norm": 0.0,
      "learning_rate": 1.9413495276233896e-05,
      "loss": 1.008,
      "step": 634
    },
    {
      "epoch": 0.13658851365885136,
      "grad_norm": 0.0,
      "learning_rate": 1.9411141966482865e-05,
      "loss": 0.919,
      "step": 635
    },
    {
      "epoch": 0.13680361368036137,
      "grad_norm": 0.0,
      "learning_rate": 1.9408784088148177e-05,
      "loss": 0.9941,
      "step": 636
    },
    {
      "epoch": 0.13701871370187138,
      "grad_norm": 0.0,
      "learning_rate": 1.9406421642374448e-05,
      "loss": 0.9555,
      "step": 637
    },
    {
      "epoch": 0.13723381372338136,
      "grad_norm": 0.0,
      "learning_rate": 1.9404054630308513e-05,
      "loss": 1.0118,
      "step": 638
    },
    {
      "epoch": 0.13744891374489138,
      "grad_norm": 0.0,
      "learning_rate": 1.9401683053099426e-05,
      "loss": 0.9635,
      "step": 639
    },
    {
      "epoch": 0.1376640137664014,
      "grad_norm": 0.0,
      "learning_rate": 1.9399306911898458e-05,
      "loss": 0.9059,
      "step": 640
    },
    {
      "epoch": 0.13787911378791137,
      "grad_norm": 0.0,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.9824,
      "step": 641
    },
    {
      "epoch": 0.13809421380942138,
      "grad_norm": 0.0,
      "learning_rate": 1.9394540942137012e-05,
      "loss": 0.8592,
      "step": 642
    },
    {
      "epoch": 0.1383093138309314,
      "grad_norm": 0.0,
      "learning_rate": 1.939215111589015e-05,
      "loss": 0.9135,
      "step": 643
    },
    {
      "epoch": 0.13852441385244138,
      "grad_norm": 0.0,
      "learning_rate": 1.9389756730278625e-05,
      "loss": 1.0415,
      "step": 644
    },
    {
      "epoch": 0.1387395138739514,
      "grad_norm": 0.0,
      "learning_rate": 1.9387357786464785e-05,
      "loss": 0.9565,
      "step": 645
    },
    {
      "epoch": 0.13895461389546138,
      "grad_norm": 0.0,
      "learning_rate": 1.9384954285613173e-05,
      "loss": 0.9817,
      "step": 646
    },
    {
      "epoch": 0.1391697139169714,
      "grad_norm": 0.0,
      "learning_rate": 1.9382546228890558e-05,
      "loss": 0.991,
      "step": 647
    },
    {
      "epoch": 0.1393848139384814,
      "grad_norm": 0.0,
      "learning_rate": 1.938013361746592e-05,
      "loss": 0.9469,
      "step": 648
    },
    {
      "epoch": 0.13959991395999138,
      "grad_norm": 0.0,
      "learning_rate": 1.937771645251044e-05,
      "loss": 1.0116,
      "step": 649
    },
    {
      "epoch": 0.1398150139815014,
      "grad_norm": 0.0,
      "learning_rate": 1.9375294735197525e-05,
      "loss": 1.0716,
      "step": 650
    },
    {
      "epoch": 0.1400301140030114,
      "grad_norm": 0.0,
      "learning_rate": 1.9372868466702782e-05,
      "loss": 1.0193,
      "step": 651
    },
    {
      "epoch": 0.1402452140245214,
      "grad_norm": 0.0,
      "learning_rate": 1.9370437648204023e-05,
      "loss": 0.9365,
      "step": 652
    },
    {
      "epoch": 0.1404603140460314,
      "grad_norm": 0.0,
      "learning_rate": 1.936800228088128e-05,
      "loss": 0.9969,
      "step": 653
    },
    {
      "epoch": 0.14067541406754142,
      "grad_norm": 0.0,
      "learning_rate": 1.9365562365916782e-05,
      "loss": 0.9321,
      "step": 654
    },
    {
      "epoch": 0.1408905140890514,
      "grad_norm": 0.0,
      "learning_rate": 1.936311790449498e-05,
      "loss": 0.9712,
      "step": 655
    },
    {
      "epoch": 0.1411056141105614,
      "grad_norm": 0.0,
      "learning_rate": 1.9360668897802518e-05,
      "loss": 0.9019,
      "step": 656
    },
    {
      "epoch": 0.14132071413207142,
      "grad_norm": 0.0,
      "learning_rate": 1.935821534702825e-05,
      "loss": 0.9961,
      "step": 657
    },
    {
      "epoch": 0.1415358141535814,
      "grad_norm": 0.0,
      "learning_rate": 1.9355757253363243e-05,
      "loss": 1.0064,
      "step": 658
    },
    {
      "epoch": 0.14175091417509142,
      "grad_norm": 0.0,
      "learning_rate": 1.935329461800076e-05,
      "loss": 0.9784,
      "step": 659
    },
    {
      "epoch": 0.14196601419660143,
      "grad_norm": 0.0,
      "learning_rate": 1.935082744213628e-05,
      "loss": 0.9261,
      "step": 660
    },
    {
      "epoch": 0.14218111421811142,
      "grad_norm": 0.0,
      "learning_rate": 1.9348355726967466e-05,
      "loss": 0.9926,
      "step": 661
    },
    {
      "epoch": 0.14239621423962143,
      "grad_norm": 0.0,
      "learning_rate": 1.9345879473694207e-05,
      "loss": 0.9174,
      "step": 662
    },
    {
      "epoch": 0.14261131426113144,
      "grad_norm": 0.0,
      "learning_rate": 1.9343398683518586e-05,
      "loss": 0.9817,
      "step": 663
    },
    {
      "epoch": 0.14282641428264142,
      "grad_norm": 0.0,
      "learning_rate": 1.9340913357644885e-05,
      "loss": 0.9449,
      "step": 664
    },
    {
      "epoch": 0.14304151430415143,
      "grad_norm": 0.0,
      "learning_rate": 1.933842349727959e-05,
      "loss": 0.9495,
      "step": 665
    },
    {
      "epoch": 0.14325661432566142,
      "grad_norm": 0.0,
      "learning_rate": 1.9335929103631396e-05,
      "loss": 0.9728,
      "step": 666
    },
    {
      "epoch": 0.14347171434717143,
      "grad_norm": 0.0,
      "learning_rate": 1.9333430177911182e-05,
      "loss": 0.9882,
      "step": 667
    },
    {
      "epoch": 0.14368681436868144,
      "grad_norm": 0.0,
      "learning_rate": 1.933092672133204e-05,
      "loss": 0.9224,
      "step": 668
    },
    {
      "epoch": 0.14390191439019143,
      "grad_norm": 0.0,
      "learning_rate": 1.9328418735109268e-05,
      "loss": 0.9605,
      "step": 669
    },
    {
      "epoch": 0.14411701441170144,
      "grad_norm": 0.0,
      "learning_rate": 1.932590622046034e-05,
      "loss": 0.987,
      "step": 670
    },
    {
      "epoch": 0.14433211443321145,
      "grad_norm": 0.0,
      "learning_rate": 1.9323389178604942e-05,
      "loss": 1.011,
      "step": 671
    },
    {
      "epoch": 0.14454721445472143,
      "grad_norm": 0.0,
      "learning_rate": 1.9320867610764965e-05,
      "loss": 0.9035,
      "step": 672
    },
    {
      "epoch": 0.14476231447623145,
      "grad_norm": 0.0,
      "learning_rate": 1.9318341518164487e-05,
      "loss": 0.9882,
      "step": 673
    },
    {
      "epoch": 0.14497741449774146,
      "grad_norm": 0.0,
      "learning_rate": 1.9315810902029786e-05,
      "loss": 1.0032,
      "step": 674
    },
    {
      "epoch": 0.14519251451925144,
      "grad_norm": 0.0,
      "learning_rate": 1.931327576358933e-05,
      "loss": 0.9605,
      "step": 675
    },
    {
      "epoch": 0.14540761454076145,
      "grad_norm": 0.0,
      "learning_rate": 1.931073610407379e-05,
      "loss": 0.9775,
      "step": 676
    },
    {
      "epoch": 0.14562271456227147,
      "grad_norm": 0.0,
      "learning_rate": 1.9308191924716027e-05,
      "loss": 1.0099,
      "step": 677
    },
    {
      "epoch": 0.14583781458378145,
      "grad_norm": 0.0,
      "learning_rate": 1.9305643226751104e-05,
      "loss": 0.9651,
      "step": 678
    },
    {
      "epoch": 0.14605291460529146,
      "grad_norm": 0.0,
      "learning_rate": 1.930309001141626e-05,
      "loss": 0.9723,
      "step": 679
    },
    {
      "epoch": 0.14626801462680147,
      "grad_norm": 0.0,
      "learning_rate": 1.930053227995095e-05,
      "loss": 0.929,
      "step": 680
    },
    {
      "epoch": 0.14648311464831146,
      "grad_norm": 0.0,
      "learning_rate": 1.9297970033596805e-05,
      "loss": 0.9722,
      "step": 681
    },
    {
      "epoch": 0.14669821466982147,
      "grad_norm": 0.0,
      "learning_rate": 1.929540327359765e-05,
      "loss": 0.9512,
      "step": 682
    },
    {
      "epoch": 0.14691331469133148,
      "grad_norm": 0.0,
      "learning_rate": 1.9292832001199507e-05,
      "loss": 0.9743,
      "step": 683
    },
    {
      "epoch": 0.14712841471284147,
      "grad_norm": 0.0,
      "learning_rate": 1.9290256217650588e-05,
      "loss": 0.9437,
      "step": 684
    },
    {
      "epoch": 0.14734351473435148,
      "grad_norm": 0.0,
      "learning_rate": 1.9287675924201284e-05,
      "loss": 0.9697,
      "step": 685
    },
    {
      "epoch": 0.1475586147558615,
      "grad_norm": 0.0,
      "learning_rate": 1.928509112210419e-05,
      "loss": 0.9757,
      "step": 686
    },
    {
      "epoch": 0.14777371477737147,
      "grad_norm": 0.0,
      "learning_rate": 1.928250181261408e-05,
      "loss": 1.0089,
      "step": 687
    },
    {
      "epoch": 0.14798881479888149,
      "grad_norm": 0.0,
      "learning_rate": 1.9279907996987917e-05,
      "loss": 0.9284,
      "step": 688
    },
    {
      "epoch": 0.14820391482039147,
      "grad_norm": 0.0,
      "learning_rate": 1.927730967648486e-05,
      "loss": 0.9764,
      "step": 689
    },
    {
      "epoch": 0.14841901484190148,
      "grad_norm": 0.0,
      "learning_rate": 1.927470685236624e-05,
      "loss": 0.947,
      "step": 690
    },
    {
      "epoch": 0.1486341148634115,
      "grad_norm": 0.0,
      "learning_rate": 1.927209952589559e-05,
      "loss": 0.9526,
      "step": 691
    },
    {
      "epoch": 0.14884921488492148,
      "grad_norm": 0.0,
      "learning_rate": 1.926948769833862e-05,
      "loss": 0.9571,
      "step": 692
    },
    {
      "epoch": 0.1490643149064315,
      "grad_norm": 0.0,
      "learning_rate": 1.9266871370963223e-05,
      "loss": 0.9534,
      "step": 693
    },
    {
      "epoch": 0.1492794149279415,
      "grad_norm": 0.0,
      "learning_rate": 1.9264250545039485e-05,
      "loss": 0.9444,
      "step": 694
    },
    {
      "epoch": 0.14949451494945148,
      "grad_norm": 0.0,
      "learning_rate": 1.9261625221839664e-05,
      "loss": 0.9247,
      "step": 695
    },
    {
      "epoch": 0.1497096149709615,
      "grad_norm": 0.0,
      "learning_rate": 1.9258995402638212e-05,
      "loss": 1.0066,
      "step": 696
    },
    {
      "epoch": 0.1499247149924715,
      "grad_norm": 0.0,
      "learning_rate": 1.925636108871176e-05,
      "loss": 0.9466,
      "step": 697
    },
    {
      "epoch": 0.1501398150139815,
      "grad_norm": 0.0,
      "learning_rate": 1.925372228133912e-05,
      "loss": 0.9663,
      "step": 698
    },
    {
      "epoch": 0.1503549150354915,
      "grad_norm": 0.0,
      "learning_rate": 1.925107898180128e-05,
      "loss": 0.9395,
      "step": 699
    },
    {
      "epoch": 0.15057001505700152,
      "grad_norm": 0.0,
      "learning_rate": 1.9248431191381418e-05,
      "loss": 0.9515,
      "step": 700
    },
    {
      "epoch": 0.1507851150785115,
      "grad_norm": 0.0,
      "learning_rate": 1.9245778911364894e-05,
      "loss": 0.9927,
      "step": 701
    },
    {
      "epoch": 0.1510002151000215,
      "grad_norm": 0.0,
      "learning_rate": 1.9243122143039232e-05,
      "loss": 1.0028,
      "step": 702
    },
    {
      "epoch": 0.15121531512153152,
      "grad_norm": 0.0,
      "learning_rate": 1.924046088769415e-05,
      "loss": 1.0198,
      "step": 703
    },
    {
      "epoch": 0.1514304151430415,
      "grad_norm": 0.0,
      "learning_rate": 1.923779514662154e-05,
      "loss": 0.9357,
      "step": 704
    },
    {
      "epoch": 0.15164551516455152,
      "grad_norm": 0.0,
      "learning_rate": 1.923512492111547e-05,
      "loss": 0.9594,
      "step": 705
    },
    {
      "epoch": 0.15186061518606153,
      "grad_norm": 0.0,
      "learning_rate": 1.923245021247218e-05,
      "loss": 0.8893,
      "step": 706
    },
    {
      "epoch": 0.15207571520757152,
      "grad_norm": 0.0,
      "learning_rate": 1.9229771021990094e-05,
      "loss": 0.9322,
      "step": 707
    },
    {
      "epoch": 0.15229081522908153,
      "grad_norm": 0.0,
      "learning_rate": 1.9227087350969815e-05,
      "loss": 0.9436,
      "step": 708
    },
    {
      "epoch": 0.1525059152505915,
      "grad_norm": 0.0,
      "learning_rate": 1.922439920071411e-05,
      "loss": 0.9351,
      "step": 709
    },
    {
      "epoch": 0.15272101527210152,
      "grad_norm": 0.0,
      "learning_rate": 1.9221706572527925e-05,
      "loss": 1.0172,
      "step": 710
    },
    {
      "epoch": 0.15293611529361154,
      "grad_norm": 0.0,
      "learning_rate": 1.9219009467718385e-05,
      "loss": 0.984,
      "step": 711
    },
    {
      "epoch": 0.15315121531512152,
      "grad_norm": 0.0,
      "learning_rate": 1.9216307887594777e-05,
      "loss": 0.9751,
      "step": 712
    },
    {
      "epoch": 0.15336631533663153,
      "grad_norm": 0.0,
      "learning_rate": 1.9213601833468573e-05,
      "loss": 0.9742,
      "step": 713
    },
    {
      "epoch": 0.15358141535814154,
      "grad_norm": 0.0,
      "learning_rate": 1.9210891306653413e-05,
      "loss": 0.9819,
      "step": 714
    },
    {
      "epoch": 0.15379651537965153,
      "grad_norm": 0.0,
      "learning_rate": 1.9208176308465098e-05,
      "loss": 0.862,
      "step": 715
    },
    {
      "epoch": 0.15401161540116154,
      "grad_norm": 0.0,
      "learning_rate": 1.9205456840221614e-05,
      "loss": 0.9897,
      "step": 716
    },
    {
      "epoch": 0.15422671542267155,
      "grad_norm": 0.0,
      "learning_rate": 1.920273290324311e-05,
      "loss": 0.9263,
      "step": 717
    },
    {
      "epoch": 0.15444181544418153,
      "grad_norm": 0.0,
      "learning_rate": 1.9200004498851903e-05,
      "loss": 0.9596,
      "step": 718
    },
    {
      "epoch": 0.15465691546569155,
      "grad_norm": 0.0,
      "learning_rate": 1.9197271628372484e-05,
      "loss": 1.0044,
      "step": 719
    },
    {
      "epoch": 0.15487201548720156,
      "grad_norm": 0.0,
      "learning_rate": 1.9194534293131507e-05,
      "loss": 0.9987,
      "step": 720
    },
    {
      "epoch": 0.15508711550871154,
      "grad_norm": 0.0,
      "learning_rate": 1.9191792494457793e-05,
      "loss": 0.9305,
      "step": 721
    },
    {
      "epoch": 0.15530221553022155,
      "grad_norm": 0.0,
      "learning_rate": 1.918904623368234e-05,
      "loss": 0.9187,
      "step": 722
    },
    {
      "epoch": 0.15551731555173157,
      "grad_norm": 0.0,
      "learning_rate": 1.91862955121383e-05,
      "loss": 1.0113,
      "step": 723
    },
    {
      "epoch": 0.15573241557324155,
      "grad_norm": 0.0,
      "learning_rate": 1.918354033116099e-05,
      "loss": 1.007,
      "step": 724
    },
    {
      "epoch": 0.15594751559475156,
      "grad_norm": 0.0,
      "learning_rate": 1.9180780692087903e-05,
      "loss": 0.9315,
      "step": 725
    },
    {
      "epoch": 0.15616261561626157,
      "grad_norm": 0.0,
      "learning_rate": 1.9178016596258688e-05,
      "loss": 0.9444,
      "step": 726
    },
    {
      "epoch": 0.15637771563777156,
      "grad_norm": 0.0,
      "learning_rate": 1.9175248045015153e-05,
      "loss": 0.9399,
      "step": 727
    },
    {
      "epoch": 0.15659281565928157,
      "grad_norm": 0.0,
      "learning_rate": 1.9172475039701283e-05,
      "loss": 0.9602,
      "step": 728
    },
    {
      "epoch": 0.15680791568079158,
      "grad_norm": 0.0,
      "learning_rate": 1.9169697581663212e-05,
      "loss": 0.9955,
      "step": 729
    },
    {
      "epoch": 0.15702301570230157,
      "grad_norm": 0.0,
      "learning_rate": 1.9166915672249246e-05,
      "loss": 0.9357,
      "step": 730
    },
    {
      "epoch": 0.15723811572381158,
      "grad_norm": 0.0,
      "learning_rate": 1.916412931280984e-05,
      "loss": 0.9672,
      "step": 731
    },
    {
      "epoch": 0.15745321574532156,
      "grad_norm": 0.0,
      "learning_rate": 1.9161338504697622e-05,
      "loss": 0.8958,
      "step": 732
    },
    {
      "epoch": 0.15766831576683157,
      "grad_norm": 0.0,
      "learning_rate": 1.915854324926737e-05,
      "loss": 0.9375,
      "step": 733
    },
    {
      "epoch": 0.15788341578834159,
      "grad_norm": 0.0,
      "learning_rate": 1.9155743547876026e-05,
      "loss": 0.9313,
      "step": 734
    },
    {
      "epoch": 0.15809851580985157,
      "grad_norm": 0.0,
      "learning_rate": 1.9152939401882684e-05,
      "loss": 1.0264,
      "step": 735
    },
    {
      "epoch": 0.15831361583136158,
      "grad_norm": 0.0,
      "learning_rate": 1.9150130812648607e-05,
      "loss": 0.9898,
      "step": 736
    },
    {
      "epoch": 0.1585287158528716,
      "grad_norm": 0.0,
      "learning_rate": 1.9147317781537204e-05,
      "loss": 1.0068,
      "step": 737
    },
    {
      "epoch": 0.15874381587438158,
      "grad_norm": 0.0,
      "learning_rate": 1.9144500309914047e-05,
      "loss": 0.9449,
      "step": 738
    },
    {
      "epoch": 0.1589589158958916,
      "grad_norm": 0.0,
      "learning_rate": 1.9141678399146858e-05,
      "loss": 0.9143,
      "step": 739
    },
    {
      "epoch": 0.1591740159174016,
      "grad_norm": 0.0,
      "learning_rate": 1.913885205060552e-05,
      "loss": 1.0497,
      "step": 740
    },
    {
      "epoch": 0.15938911593891159,
      "grad_norm": 0.0,
      "learning_rate": 1.9136021265662067e-05,
      "loss": 0.9597,
      "step": 741
    },
    {
      "epoch": 0.1596042159604216,
      "grad_norm": 0.0,
      "learning_rate": 1.9133186045690682e-05,
      "loss": 0.9295,
      "step": 742
    },
    {
      "epoch": 0.1598193159819316,
      "grad_norm": 0.0,
      "learning_rate": 1.913034639206771e-05,
      "loss": 0.9438,
      "step": 743
    },
    {
      "epoch": 0.1600344160034416,
      "grad_norm": 0.0,
      "learning_rate": 1.9127502306171644e-05,
      "loss": 1.0477,
      "step": 744
    },
    {
      "epoch": 0.1602495160249516,
      "grad_norm": 0.0,
      "learning_rate": 1.9124653789383134e-05,
      "loss": 0.9176,
      "step": 745
    },
    {
      "epoch": 0.16046461604646162,
      "grad_norm": 0.0,
      "learning_rate": 1.9121800843084964e-05,
      "loss": 0.9352,
      "step": 746
    },
    {
      "epoch": 0.1606797160679716,
      "grad_norm": 0.0,
      "learning_rate": 1.911894346866209e-05,
      "loss": 0.9934,
      "step": 747
    },
    {
      "epoch": 0.1608948160894816,
      "grad_norm": 0.0,
      "learning_rate": 1.9116081667501603e-05,
      "loss": 0.9692,
      "step": 748
    },
    {
      "epoch": 0.16110991611099162,
      "grad_norm": 0.0,
      "learning_rate": 1.9113215440992752e-05,
      "loss": 0.9515,
      "step": 749
    },
    {
      "epoch": 0.1613250161325016,
      "grad_norm": 0.0,
      "learning_rate": 1.911034479052692e-05,
      "loss": 1.0278,
      "step": 750
    },
    {
      "epoch": 0.16154011615401162,
      "grad_norm": 0.0,
      "learning_rate": 1.910746971749766e-05,
      "loss": 0.9481,
      "step": 751
    },
    {
      "epoch": 0.1617552161755216,
      "grad_norm": 0.0,
      "learning_rate": 1.9104590223300656e-05,
      "loss": 1.0185,
      "step": 752
    },
    {
      "epoch": 0.16197031619703162,
      "grad_norm": 0.0,
      "learning_rate": 1.9101706309333732e-05,
      "loss": 0.9761,
      "step": 753
    },
    {
      "epoch": 0.16218541621854163,
      "grad_norm": 0.0,
      "learning_rate": 1.909881797699688e-05,
      "loss": 0.9853,
      "step": 754
    },
    {
      "epoch": 0.1624005162400516,
      "grad_norm": 0.0,
      "learning_rate": 1.909592522769221e-05,
      "loss": 0.9937,
      "step": 755
    },
    {
      "epoch": 0.16261561626156162,
      "grad_norm": 0.0,
      "learning_rate": 1.9093028062824005e-05,
      "loss": 0.9799,
      "step": 756
    },
    {
      "epoch": 0.16283071628307164,
      "grad_norm": 0.0,
      "learning_rate": 1.9090126483798667e-05,
      "loss": 0.994,
      "step": 757
    },
    {
      "epoch": 0.16304581630458162,
      "grad_norm": 0.0,
      "learning_rate": 1.9087220492024756e-05,
      "loss": 0.9675,
      "step": 758
    },
    {
      "epoch": 0.16326091632609163,
      "grad_norm": 0.0,
      "learning_rate": 1.9084310088912964e-05,
      "loss": 0.985,
      "step": 759
    },
    {
      "epoch": 0.16347601634760164,
      "grad_norm": 0.0,
      "learning_rate": 1.908139527587613e-05,
      "loss": 0.9446,
      "step": 760
    },
    {
      "epoch": 0.16369111636911163,
      "grad_norm": 0.0,
      "learning_rate": 1.9078476054329237e-05,
      "loss": 0.9707,
      "step": 761
    },
    {
      "epoch": 0.16390621639062164,
      "grad_norm": 0.0,
      "learning_rate": 1.90755524256894e-05,
      "loss": 1.0313,
      "step": 762
    },
    {
      "epoch": 0.16412131641213165,
      "grad_norm": 0.0,
      "learning_rate": 1.907262439137588e-05,
      "loss": 0.984,
      "step": 763
    },
    {
      "epoch": 0.16433641643364164,
      "grad_norm": 0.0,
      "learning_rate": 1.9069691952810073e-05,
      "loss": 1.0376,
      "step": 764
    },
    {
      "epoch": 0.16455151645515165,
      "grad_norm": 0.0,
      "learning_rate": 1.906675511141551e-05,
      "loss": 0.9401,
      "step": 765
    },
    {
      "epoch": 0.16476661647666166,
      "grad_norm": 0.0,
      "learning_rate": 1.906381386861787e-05,
      "loss": 0.9058,
      "step": 766
    },
    {
      "epoch": 0.16498171649817164,
      "grad_norm": 0.0,
      "learning_rate": 1.9060868225844955e-05,
      "loss": 0.9999,
      "step": 767
    },
    {
      "epoch": 0.16519681651968166,
      "grad_norm": 0.0,
      "learning_rate": 1.905791818452672e-05,
      "loss": 0.9403,
      "step": 768
    },
    {
      "epoch": 0.16541191654119167,
      "grad_norm": 0.0,
      "learning_rate": 1.9054963746095235e-05,
      "loss": 0.9356,
      "step": 769
    },
    {
      "epoch": 0.16562701656270165,
      "grad_norm": 0.0,
      "learning_rate": 1.9052004911984722e-05,
      "loss": 0.9319,
      "step": 770
    },
    {
      "epoch": 0.16584211658421166,
      "grad_norm": 0.0,
      "learning_rate": 1.9049041683631526e-05,
      "loss": 0.9677,
      "step": 771
    },
    {
      "epoch": 0.16605721660572165,
      "grad_norm": 0.0,
      "learning_rate": 1.9046074062474127e-05,
      "loss": 0.8562,
      "step": 772
    },
    {
      "epoch": 0.16627231662723166,
      "grad_norm": 0.0,
      "learning_rate": 1.9043102049953148e-05,
      "loss": 0.9725,
      "step": 773
    },
    {
      "epoch": 0.16648741664874167,
      "grad_norm": 0.0,
      "learning_rate": 1.9040125647511324e-05,
      "loss": 1.0131,
      "step": 774
    },
    {
      "epoch": 0.16670251667025165,
      "grad_norm": 0.0,
      "learning_rate": 1.903714485659354e-05,
      "loss": 0.9786,
      "step": 775
    },
    {
      "epoch": 0.16691761669176167,
      "grad_norm": 0.0,
      "learning_rate": 1.9034159678646798e-05,
      "loss": 0.9412,
      "step": 776
    },
    {
      "epoch": 0.16713271671327168,
      "grad_norm": 0.0,
      "learning_rate": 1.9031170115120237e-05,
      "loss": 1.0132,
      "step": 777
    },
    {
      "epoch": 0.16734781673478166,
      "grad_norm": 0.0,
      "learning_rate": 1.9028176167465125e-05,
      "loss": 0.9767,
      "step": 778
    },
    {
      "epoch": 0.16756291675629167,
      "grad_norm": 0.0,
      "learning_rate": 1.9025177837134858e-05,
      "loss": 0.9359,
      "step": 779
    },
    {
      "epoch": 0.1677780167778017,
      "grad_norm": 0.0,
      "learning_rate": 1.902217512558495e-05,
      "loss": 0.9634,
      "step": 780
    },
    {
      "epoch": 0.16799311679931167,
      "grad_norm": 0.0,
      "learning_rate": 1.9019168034273062e-05,
      "loss": 0.9774,
      "step": 781
    },
    {
      "epoch": 0.16820821682082168,
      "grad_norm": 0.0,
      "learning_rate": 1.9016156564658955e-05,
      "loss": 0.9266,
      "step": 782
    },
    {
      "epoch": 0.1684233168423317,
      "grad_norm": 0.0,
      "learning_rate": 1.901314071820454e-05,
      "loss": 0.904,
      "step": 783
    },
    {
      "epoch": 0.16863841686384168,
      "grad_norm": 0.0,
      "learning_rate": 1.9010120496373834e-05,
      "loss": 0.9772,
      "step": 784
    },
    {
      "epoch": 0.1688535168853517,
      "grad_norm": 0.0,
      "learning_rate": 1.900709590063299e-05,
      "loss": 0.9358,
      "step": 785
    },
    {
      "epoch": 0.1690686169068617,
      "grad_norm": 0.0,
      "learning_rate": 1.900406693245028e-05,
      "loss": 0.9484,
      "step": 786
    },
    {
      "epoch": 0.16928371692837169,
      "grad_norm": 0.0,
      "learning_rate": 1.9001033593296103e-05,
      "loss": 0.9448,
      "step": 787
    },
    {
      "epoch": 0.1694988169498817,
      "grad_norm": 0.0,
      "learning_rate": 1.8997995884642968e-05,
      "loss": 0.9405,
      "step": 788
    },
    {
      "epoch": 0.1697139169713917,
      "grad_norm": 0.0,
      "learning_rate": 1.8994953807965516e-05,
      "loss": 0.9512,
      "step": 789
    },
    {
      "epoch": 0.1699290169929017,
      "grad_norm": 0.0,
      "learning_rate": 1.8991907364740508e-05,
      "loss": 0.907,
      "step": 790
    },
    {
      "epoch": 0.1701441170144117,
      "grad_norm": 0.0,
      "learning_rate": 1.898885655644682e-05,
      "loss": 0.9527,
      "step": 791
    },
    {
      "epoch": 0.17035921703592172,
      "grad_norm": 0.0,
      "learning_rate": 1.8985801384565447e-05,
      "loss": 0.8886,
      "step": 792
    },
    {
      "epoch": 0.1705743170574317,
      "grad_norm": 0.0,
      "learning_rate": 1.8982741850579504e-05,
      "loss": 0.9625,
      "step": 793
    },
    {
      "epoch": 0.1707894170789417,
      "grad_norm": 0.0,
      "learning_rate": 1.8979677955974226e-05,
      "loss": 0.9867,
      "step": 794
    },
    {
      "epoch": 0.1710045171004517,
      "grad_norm": 0.0,
      "learning_rate": 1.8976609702236963e-05,
      "loss": 0.9973,
      "step": 795
    },
    {
      "epoch": 0.1712196171219617,
      "grad_norm": 0.0,
      "learning_rate": 1.897353709085718e-05,
      "loss": 0.9262,
      "step": 796
    },
    {
      "epoch": 0.17143471714347172,
      "grad_norm": 0.0,
      "learning_rate": 1.897046012332646e-05,
      "loss": 0.9404,
      "step": 797
    },
    {
      "epoch": 0.1716498171649817,
      "grad_norm": 0.0,
      "learning_rate": 1.8967378801138497e-05,
      "loss": 0.9322,
      "step": 798
    },
    {
      "epoch": 0.17186491718649172,
      "grad_norm": 0.0,
      "learning_rate": 1.89642931257891e-05,
      "loss": 0.9075,
      "step": 799
    },
    {
      "epoch": 0.17208001720800173,
      "grad_norm": 0.0,
      "learning_rate": 1.8961203098776194e-05,
      "loss": 0.9317,
      "step": 800
    },
    {
      "epoch": 0.1722951172295117,
      "grad_norm": 0.0,
      "learning_rate": 1.895810872159981e-05,
      "loss": 0.9474,
      "step": 801
    },
    {
      "epoch": 0.17251021725102172,
      "grad_norm": 0.0,
      "learning_rate": 1.8955009995762106e-05,
      "loss": 0.9365,
      "step": 802
    },
    {
      "epoch": 0.17272531727253174,
      "grad_norm": 0.0,
      "learning_rate": 1.895190692276733e-05,
      "loss": 1.0494,
      "step": 803
    },
    {
      "epoch": 0.17294041729404172,
      "grad_norm": 0.0,
      "learning_rate": 1.894879950412185e-05,
      "loss": 0.917,
      "step": 804
    },
    {
      "epoch": 0.17315551731555173,
      "grad_norm": 0.0,
      "learning_rate": 1.894568774133415e-05,
      "loss": 0.9321,
      "step": 805
    },
    {
      "epoch": 0.17337061733706174,
      "grad_norm": 0.0,
      "learning_rate": 1.8942571635914814e-05,
      "loss": 1.0108,
      "step": 806
    },
    {
      "epoch": 0.17358571735857173,
      "grad_norm": 0.0,
      "learning_rate": 1.8939451189376536e-05,
      "loss": 1.0044,
      "step": 807
    },
    {
      "epoch": 0.17380081738008174,
      "grad_norm": 0.0,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 0.9226,
      "step": 808
    },
    {
      "epoch": 0.17401591740159175,
      "grad_norm": 0.0,
      "learning_rate": 1.893319727900448e-05,
      "loss": 0.9332,
      "step": 809
    },
    {
      "epoch": 0.17423101742310174,
      "grad_norm": 0.0,
      "learning_rate": 1.8930063818206626e-05,
      "loss": 0.9785,
      "step": 810
    },
    {
      "epoch": 0.17444611744461175,
      "grad_norm": 0.0,
      "learning_rate": 1.8926926022361674e-05,
      "loss": 0.9927,
      "step": 811
    },
    {
      "epoch": 0.17466121746612176,
      "grad_norm": 0.0,
      "learning_rate": 1.8923783892992857e-05,
      "loss": 1.0406,
      "step": 812
    },
    {
      "epoch": 0.17487631748763174,
      "grad_norm": 0.0,
      "learning_rate": 1.8920637431625497e-05,
      "loss": 0.9408,
      "step": 813
    },
    {
      "epoch": 0.17509141750914176,
      "grad_norm": 0.0,
      "learning_rate": 1.891748663978703e-05,
      "loss": 0.8625,
      "step": 814
    },
    {
      "epoch": 0.17530651753065174,
      "grad_norm": 0.0,
      "learning_rate": 1.8914331519006985e-05,
      "loss": 0.9882,
      "step": 815
    },
    {
      "epoch": 0.17552161755216175,
      "grad_norm": 0.0,
      "learning_rate": 1.8911172070817e-05,
      "loss": 0.9626,
      "step": 816
    },
    {
      "epoch": 0.17573671757367176,
      "grad_norm": 0.0,
      "learning_rate": 1.8908008296750808e-05,
      "loss": 0.9714,
      "step": 817
    },
    {
      "epoch": 0.17595181759518175,
      "grad_norm": 0.0,
      "learning_rate": 1.890484019834425e-05,
      "loss": 0.9162,
      "step": 818
    },
    {
      "epoch": 0.17616691761669176,
      "grad_norm": 0.0,
      "learning_rate": 1.8901667777135252e-05,
      "loss": 0.9357,
      "step": 819
    },
    {
      "epoch": 0.17638201763820177,
      "grad_norm": 0.0,
      "learning_rate": 1.8898491034663852e-05,
      "loss": 0.9579,
      "step": 820
    },
    {
      "epoch": 0.17659711765971176,
      "grad_norm": 0.0,
      "learning_rate": 1.8895309972472184e-05,
      "loss": 0.9939,
      "step": 821
    },
    {
      "epoch": 0.17681221768122177,
      "grad_norm": 0.0,
      "learning_rate": 1.8892124592104474e-05,
      "loss": 0.9175,
      "step": 822
    },
    {
      "epoch": 0.17702731770273178,
      "grad_norm": 0.0,
      "learning_rate": 1.8888934895107042e-05,
      "loss": 1.0075,
      "step": 823
    },
    {
      "epoch": 0.17724241772424176,
      "grad_norm": 0.0,
      "learning_rate": 1.888574088302831e-05,
      "loss": 0.898,
      "step": 824
    },
    {
      "epoch": 0.17745751774575177,
      "grad_norm": 0.0,
      "learning_rate": 1.8882542557418795e-05,
      "loss": 0.9271,
      "step": 825
    },
    {
      "epoch": 0.1776726177672618,
      "grad_norm": 0.0,
      "learning_rate": 1.8879339919831103e-05,
      "loss": 0.9169,
      "step": 826
    },
    {
      "epoch": 0.17788771778877177,
      "grad_norm": 0.0,
      "learning_rate": 1.887613297181994e-05,
      "loss": 0.8951,
      "step": 827
    },
    {
      "epoch": 0.17810281781028178,
      "grad_norm": 0.0,
      "learning_rate": 1.887292171494209e-05,
      "loss": 0.9792,
      "step": 828
    },
    {
      "epoch": 0.1783179178317918,
      "grad_norm": 0.0,
      "learning_rate": 1.886970615075645e-05,
      "loss": 0.9336,
      "step": 829
    },
    {
      "epoch": 0.17853301785330178,
      "grad_norm": 0.0,
      "learning_rate": 1.8866486280823994e-05,
      "loss": 0.9176,
      "step": 830
    },
    {
      "epoch": 0.1787481178748118,
      "grad_norm": 0.0,
      "learning_rate": 1.8863262106707783e-05,
      "loss": 0.9817,
      "step": 831
    },
    {
      "epoch": 0.1789632178963218,
      "grad_norm": 0.0,
      "learning_rate": 1.886003362997298e-05,
      "loss": 0.9486,
      "step": 832
    },
    {
      "epoch": 0.1791783179178318,
      "grad_norm": 0.0,
      "learning_rate": 1.8856800852186826e-05,
      "loss": 0.9598,
      "step": 833
    },
    {
      "epoch": 0.1793934179393418,
      "grad_norm": 0.0,
      "learning_rate": 1.8853563774918657e-05,
      "loss": 0.9337,
      "step": 834
    },
    {
      "epoch": 0.17960851796085178,
      "grad_norm": 0.0,
      "learning_rate": 1.885032239973989e-05,
      "loss": 0.944,
      "step": 835
    },
    {
      "epoch": 0.1798236179823618,
      "grad_norm": 0.0,
      "learning_rate": 1.8847076728224036e-05,
      "loss": 0.9886,
      "step": 836
    },
    {
      "epoch": 0.1800387180038718,
      "grad_norm": 0.0,
      "learning_rate": 1.884382676194668e-05,
      "loss": 0.8831,
      "step": 837
    },
    {
      "epoch": 0.1802538180253818,
      "grad_norm": 0.0,
      "learning_rate": 1.884057250248551e-05,
      "loss": 0.9573,
      "step": 838
    },
    {
      "epoch": 0.1804689180468918,
      "grad_norm": 0.0,
      "learning_rate": 1.8837313951420274e-05,
      "loss": 0.9467,
      "step": 839
    },
    {
      "epoch": 0.1806840180684018,
      "grad_norm": 0.0,
      "learning_rate": 1.8834051110332824e-05,
      "loss": 1.0003,
      "step": 840
    },
    {
      "epoch": 0.1808991180899118,
      "grad_norm": 0.0,
      "learning_rate": 1.883078398080709e-05,
      "loss": 0.922,
      "step": 841
    },
    {
      "epoch": 0.1811142181114218,
      "grad_norm": 0.0,
      "learning_rate": 1.8827512564429076e-05,
      "loss": 0.8729,
      "step": 842
    },
    {
      "epoch": 0.18132931813293182,
      "grad_norm": 0.0,
      "learning_rate": 1.8824236862786873e-05,
      "loss": 0.9041,
      "step": 843
    },
    {
      "epoch": 0.1815444181544418,
      "grad_norm": 0.0,
      "learning_rate": 1.8820956877470647e-05,
      "loss": 0.9577,
      "step": 844
    },
    {
      "epoch": 0.18175951817595182,
      "grad_norm": 0.0,
      "learning_rate": 1.8817672610072656e-05,
      "loss": 0.9796,
      "step": 845
    },
    {
      "epoch": 0.18197461819746183,
      "grad_norm": 0.0,
      "learning_rate": 1.8814384062187225e-05,
      "loss": 0.8799,
      "step": 846
    },
    {
      "epoch": 0.1821897182189718,
      "grad_norm": 0.0,
      "learning_rate": 1.881109123541076e-05,
      "loss": 0.9472,
      "step": 847
    },
    {
      "epoch": 0.18240481824048183,
      "grad_norm": 0.0,
      "learning_rate": 1.8807794131341742e-05,
      "loss": 0.9695,
      "step": 848
    },
    {
      "epoch": 0.18261991826199184,
      "grad_norm": 0.0,
      "learning_rate": 1.880449275158073e-05,
      "loss": 0.9385,
      "step": 849
    },
    {
      "epoch": 0.18283501828350182,
      "grad_norm": 0.0,
      "learning_rate": 1.8801187097730367e-05,
      "loss": 0.9742,
      "step": 850
    },
    {
      "epoch": 0.18305011830501183,
      "grad_norm": 0.0,
      "learning_rate": 1.8797877171395357e-05,
      "loss": 0.9294,
      "step": 851
    },
    {
      "epoch": 0.18326521832652184,
      "grad_norm": 0.0,
      "learning_rate": 1.8794562974182483e-05,
      "loss": 0.9989,
      "step": 852
    },
    {
      "epoch": 0.18348031834803183,
      "grad_norm": 0.0,
      "learning_rate": 1.879124450770061e-05,
      "loss": 0.9422,
      "step": 853
    },
    {
      "epoch": 0.18369541836954184,
      "grad_norm": 0.0,
      "learning_rate": 1.8787921773560656e-05,
      "loss": 0.9025,
      "step": 854
    },
    {
      "epoch": 0.18391051839105185,
      "grad_norm": 0.0,
      "learning_rate": 1.8784594773375636e-05,
      "loss": 0.9048,
      "step": 855
    },
    {
      "epoch": 0.18412561841256184,
      "grad_norm": 0.0,
      "learning_rate": 1.8781263508760617e-05,
      "loss": 1.0172,
      "step": 856
    },
    {
      "epoch": 0.18434071843407185,
      "grad_norm": 0.0,
      "learning_rate": 1.877792798133274e-05,
      "loss": 0.8975,
      "step": 857
    },
    {
      "epoch": 0.18455581845558183,
      "grad_norm": 0.0,
      "learning_rate": 1.877458819271122e-05,
      "loss": 0.9543,
      "step": 858
    },
    {
      "epoch": 0.18477091847709184,
      "grad_norm": 0.0,
      "learning_rate": 1.8771244144517337e-05,
      "loss": 0.9786,
      "step": 859
    },
    {
      "epoch": 0.18498601849860186,
      "grad_norm": 0.0,
      "learning_rate": 1.8767895838374438e-05,
      "loss": 0.9653,
      "step": 860
    },
    {
      "epoch": 0.18520111852011184,
      "grad_norm": 0.0,
      "learning_rate": 1.8764543275907946e-05,
      "loss": 0.9196,
      "step": 861
    },
    {
      "epoch": 0.18541621854162185,
      "grad_norm": 0.0,
      "learning_rate": 1.876118645874533e-05,
      "loss": 0.9493,
      "step": 862
    },
    {
      "epoch": 0.18563131856313186,
      "grad_norm": 0.0,
      "learning_rate": 1.8757825388516148e-05,
      "loss": 0.9833,
      "step": 863
    },
    {
      "epoch": 0.18584641858464185,
      "grad_norm": 0.0,
      "learning_rate": 1.8754460066852003e-05,
      "loss": 0.8927,
      "step": 864
    },
    {
      "epoch": 0.18606151860615186,
      "grad_norm": 0.0,
      "learning_rate": 1.875109049538658e-05,
      "loss": 0.9621,
      "step": 865
    },
    {
      "epoch": 0.18627661862766187,
      "grad_norm": 0.0,
      "learning_rate": 1.8747716675755608e-05,
      "loss": 0.924,
      "step": 866
    },
    {
      "epoch": 0.18649171864917186,
      "grad_norm": 0.0,
      "learning_rate": 1.8744338609596897e-05,
      "loss": 0.9956,
      "step": 867
    },
    {
      "epoch": 0.18670681867068187,
      "grad_norm": 0.0,
      "learning_rate": 1.8740956298550307e-05,
      "loss": 0.9217,
      "step": 868
    },
    {
      "epoch": 0.18692191869219188,
      "grad_norm": 0.0,
      "learning_rate": 1.8737569744257754e-05,
      "loss": 0.9127,
      "step": 869
    },
    {
      "epoch": 0.18713701871370186,
      "grad_norm": 0.0,
      "learning_rate": 1.873417894836323e-05,
      "loss": 1.0116,
      "step": 870
    },
    {
      "epoch": 0.18735211873521188,
      "grad_norm": 0.0,
      "learning_rate": 1.8730783912512772e-05,
      "loss": 0.8967,
      "step": 871
    },
    {
      "epoch": 0.1875672187567219,
      "grad_norm": 0.0,
      "learning_rate": 1.8727384638354484e-05,
      "loss": 0.9531,
      "step": 872
    },
    {
      "epoch": 0.18778231877823187,
      "grad_norm": 0.0,
      "learning_rate": 1.872398112753852e-05,
      "loss": 0.951,
      "step": 873
    },
    {
      "epoch": 0.18799741879974188,
      "grad_norm": 0.0,
      "learning_rate": 1.8720573381717096e-05,
      "loss": 0.9948,
      "step": 874
    },
    {
      "epoch": 0.1882125188212519,
      "grad_norm": 0.0,
      "learning_rate": 1.871716140254448e-05,
      "loss": 0.9679,
      "step": 875
    },
    {
      "epoch": 0.18842761884276188,
      "grad_norm": 0.0,
      "learning_rate": 1.8713745191677003e-05,
      "loss": 0.9771,
      "step": 876
    },
    {
      "epoch": 0.1886427188642719,
      "grad_norm": 0.0,
      "learning_rate": 1.8710324750773038e-05,
      "loss": 0.9381,
      "step": 877
    },
    {
      "epoch": 0.18885781888578188,
      "grad_norm": 0.0,
      "learning_rate": 1.870690008149302e-05,
      "loss": 0.9599,
      "step": 878
    },
    {
      "epoch": 0.1890729189072919,
      "grad_norm": 0.0,
      "learning_rate": 1.8703471185499438e-05,
      "loss": 0.9848,
      "step": 879
    },
    {
      "epoch": 0.1892880189288019,
      "grad_norm": 0.0,
      "learning_rate": 1.8700038064456823e-05,
      "loss": 0.9242,
      "step": 880
    },
    {
      "epoch": 0.18950311895031188,
      "grad_norm": 0.0,
      "learning_rate": 1.869660072003177e-05,
      "loss": 0.9521,
      "step": 881
    },
    {
      "epoch": 0.1897182189718219,
      "grad_norm": 0.0,
      "learning_rate": 1.8693159153892917e-05,
      "loss": 0.9977,
      "step": 882
    },
    {
      "epoch": 0.1899333189933319,
      "grad_norm": 0.0,
      "learning_rate": 1.868971336771095e-05,
      "loss": 0.9653,
      "step": 883
    },
    {
      "epoch": 0.1901484190148419,
      "grad_norm": 0.0,
      "learning_rate": 1.86862633631586e-05,
      "loss": 0.9261,
      "step": 884
    },
    {
      "epoch": 0.1903635190363519,
      "grad_norm": 0.0,
      "learning_rate": 1.868280914191066e-05,
      "loss": 0.9821,
      "step": 885
    },
    {
      "epoch": 0.19057861905786191,
      "grad_norm": 0.0,
      "learning_rate": 1.867935070564396e-05,
      "loss": 0.9399,
      "step": 886
    },
    {
      "epoch": 0.1907937190793719,
      "grad_norm": 0.0,
      "learning_rate": 1.8675888056037378e-05,
      "loss": 0.9537,
      "step": 887
    },
    {
      "epoch": 0.1910088191008819,
      "grad_norm": 0.0,
      "learning_rate": 1.867242119477183e-05,
      "loss": 0.9473,
      "step": 888
    },
    {
      "epoch": 0.19122391912239192,
      "grad_norm": 0.0,
      "learning_rate": 1.8668950123530293e-05,
      "loss": 0.9841,
      "step": 889
    },
    {
      "epoch": 0.1914390191439019,
      "grad_norm": 0.0,
      "learning_rate": 1.866547484399777e-05,
      "loss": 0.9584,
      "step": 890
    },
    {
      "epoch": 0.19165411916541192,
      "grad_norm": 0.0,
      "learning_rate": 1.8661995357861325e-05,
      "loss": 0.9258,
      "step": 891
    },
    {
      "epoch": 0.19186921918692193,
      "grad_norm": 0.0,
      "learning_rate": 1.8658511666810043e-05,
      "loss": 0.9893,
      "step": 892
    },
    {
      "epoch": 0.19208431920843191,
      "grad_norm": 0.0,
      "learning_rate": 1.8655023772535064e-05,
      "loss": 0.9038,
      "step": 893
    },
    {
      "epoch": 0.19229941922994193,
      "grad_norm": 0.0,
      "learning_rate": 1.8651531676729574e-05,
      "loss": 0.9736,
      "step": 894
    },
    {
      "epoch": 0.19251451925145194,
      "grad_norm": 0.0,
      "learning_rate": 1.8648035381088784e-05,
      "loss": 0.914,
      "step": 895
    },
    {
      "epoch": 0.19272961927296192,
      "grad_norm": 0.0,
      "learning_rate": 1.8644534887309947e-05,
      "loss": 0.9967,
      "step": 896
    },
    {
      "epoch": 0.19294471929447193,
      "grad_norm": 0.0,
      "learning_rate": 1.864103019709236e-05,
      "loss": 0.9352,
      "step": 897
    },
    {
      "epoch": 0.19315981931598195,
      "grad_norm": 0.0,
      "learning_rate": 1.863752131213736e-05,
      "loss": 0.8926,
      "step": 898
    },
    {
      "epoch": 0.19337491933749193,
      "grad_norm": 0.0,
      "learning_rate": 1.863400823414831e-05,
      "loss": 0.9282,
      "step": 899
    },
    {
      "epoch": 0.19359001935900194,
      "grad_norm": 0.0,
      "learning_rate": 1.863049096483061e-05,
      "loss": 0.9704,
      "step": 900
    },
    {
      "epoch": 0.19380511938051193,
      "grad_norm": 0.0,
      "learning_rate": 1.86269695058917e-05,
      "loss": 0.9286,
      "step": 901
    },
    {
      "epoch": 0.19402021940202194,
      "grad_norm": 0.0,
      "learning_rate": 1.8623443859041055e-05,
      "loss": 1.0141,
      "step": 902
    },
    {
      "epoch": 0.19423531942353195,
      "grad_norm": 0.0,
      "learning_rate": 1.8619914025990175e-05,
      "loss": 0.9915,
      "step": 903
    },
    {
      "epoch": 0.19445041944504193,
      "grad_norm": 0.0,
      "learning_rate": 1.86163800084526e-05,
      "loss": 0.8908,
      "step": 904
    },
    {
      "epoch": 0.19466551946655195,
      "grad_norm": 0.0,
      "learning_rate": 1.8612841808143897e-05,
      "loss": 0.9852,
      "step": 905
    },
    {
      "epoch": 0.19488061948806196,
      "grad_norm": 0.0,
      "learning_rate": 1.8609299426781662e-05,
      "loss": 0.9748,
      "step": 906
    },
    {
      "epoch": 0.19509571950957194,
      "grad_norm": 0.0,
      "learning_rate": 1.860575286608553e-05,
      "loss": 0.926,
      "step": 907
    },
    {
      "epoch": 0.19531081953108195,
      "grad_norm": 0.0,
      "learning_rate": 1.860220212777715e-05,
      "loss": 0.9451,
      "step": 908
    },
    {
      "epoch": 0.19552591955259196,
      "grad_norm": 0.0,
      "learning_rate": 1.8598647213580215e-05,
      "loss": 0.9022,
      "step": 909
    },
    {
      "epoch": 0.19574101957410195,
      "grad_norm": 0.0,
      "learning_rate": 1.859508812522043e-05,
      "loss": 0.9398,
      "step": 910
    },
    {
      "epoch": 0.19595611959561196,
      "grad_norm": 0.0,
      "learning_rate": 1.8591524864425534e-05,
      "loss": 0.9221,
      "step": 911
    },
    {
      "epoch": 0.19617121961712197,
      "grad_norm": 0.0,
      "learning_rate": 1.8587957432925297e-05,
      "loss": 0.838,
      "step": 912
    },
    {
      "epoch": 0.19638631963863196,
      "grad_norm": 0.0,
      "learning_rate": 1.8584385832451502e-05,
      "loss": 0.9448,
      "step": 913
    },
    {
      "epoch": 0.19660141966014197,
      "grad_norm": 0.0,
      "learning_rate": 1.8580810064737967e-05,
      "loss": 0.9553,
      "step": 914
    },
    {
      "epoch": 0.19681651968165198,
      "grad_norm": 0.0,
      "learning_rate": 1.857723013152052e-05,
      "loss": 0.9885,
      "step": 915
    },
    {
      "epoch": 0.19703161970316196,
      "grad_norm": 0.0,
      "learning_rate": 1.8573646034537027e-05,
      "loss": 0.9617,
      "step": 916
    },
    {
      "epoch": 0.19724671972467198,
      "grad_norm": 0.0,
      "learning_rate": 1.857005777552736e-05,
      "loss": 0.9704,
      "step": 917
    },
    {
      "epoch": 0.197461819746182,
      "grad_norm": 0.0,
      "learning_rate": 1.8566465356233425e-05,
      "loss": 0.9196,
      "step": 918
    },
    {
      "epoch": 0.19767691976769197,
      "grad_norm": 0.0,
      "learning_rate": 1.8562868778399132e-05,
      "loss": 0.9519,
      "step": 919
    },
    {
      "epoch": 0.19789201978920198,
      "grad_norm": 0.0,
      "learning_rate": 1.8559268043770427e-05,
      "loss": 0.9744,
      "step": 920
    },
    {
      "epoch": 0.19810711981071197,
      "grad_norm": 0.0,
      "learning_rate": 1.8555663154095257e-05,
      "loss": 0.9771,
      "step": 921
    },
    {
      "epoch": 0.19832221983222198,
      "grad_norm": 0.0,
      "learning_rate": 1.8552054111123607e-05,
      "loss": 0.8906,
      "step": 922
    },
    {
      "epoch": 0.198537319853732,
      "grad_norm": 0.0,
      "learning_rate": 1.8548440916607454e-05,
      "loss": 0.9273,
      "step": 923
    },
    {
      "epoch": 0.19875241987524198,
      "grad_norm": 0.0,
      "learning_rate": 1.8544823572300805e-05,
      "loss": 1.0007,
      "step": 924
    },
    {
      "epoch": 0.198967519896752,
      "grad_norm": 0.0,
      "learning_rate": 1.854120207995968e-05,
      "loss": 0.9524,
      "step": 925
    },
    {
      "epoch": 0.199182619918262,
      "grad_norm": 0.0,
      "learning_rate": 1.8537576441342114e-05,
      "loss": 0.9305,
      "step": 926
    },
    {
      "epoch": 0.19939771993977198,
      "grad_norm": 0.0,
      "learning_rate": 1.853394665820815e-05,
      "loss": 0.9384,
      "step": 927
    },
    {
      "epoch": 0.199612819961282,
      "grad_norm": 0.0,
      "learning_rate": 1.8530312732319838e-05,
      "loss": 0.9571,
      "step": 928
    },
    {
      "epoch": 0.199827919982792,
      "grad_norm": 0.0,
      "learning_rate": 1.852667466544126e-05,
      "loss": 0.9302,
      "step": 929
    },
    {
      "epoch": 0.200043020004302,
      "grad_norm": 0.0,
      "learning_rate": 1.8523032459338482e-05,
      "loss": 0.9765,
      "step": 930
    },
    {
      "epoch": 0.200258120025812,
      "grad_norm": 0.0,
      "learning_rate": 1.8519386115779595e-05,
      "loss": 0.9197,
      "step": 931
    },
    {
      "epoch": 0.20047322004732201,
      "grad_norm": 0.0,
      "learning_rate": 1.8515735636534694e-05,
      "loss": 0.9014,
      "step": 932
    },
    {
      "epoch": 0.200688320068832,
      "grad_norm": 0.0,
      "learning_rate": 1.851208102337589e-05,
      "loss": 1.0676,
      "step": 933
    },
    {
      "epoch": 0.200903420090342,
      "grad_norm": 0.0,
      "learning_rate": 1.8508422278077285e-05,
      "loss": 0.9074,
      "step": 934
    },
    {
      "epoch": 0.20111852011185202,
      "grad_norm": 0.0,
      "learning_rate": 1.8504759402414996e-05,
      "loss": 0.9133,
      "step": 935
    },
    {
      "epoch": 0.201333620133362,
      "grad_norm": 0.0,
      "learning_rate": 1.8501092398167152e-05,
      "loss": 1.0261,
      "step": 936
    },
    {
      "epoch": 0.20154872015487202,
      "grad_norm": 0.0,
      "learning_rate": 1.8497421267113874e-05,
      "loss": 0.9333,
      "step": 937
    },
    {
      "epoch": 0.20176382017638203,
      "grad_norm": 0.0,
      "learning_rate": 1.8493746011037288e-05,
      "loss": 1.0048,
      "step": 938
    },
    {
      "epoch": 0.20197892019789201,
      "grad_norm": 0.0,
      "learning_rate": 1.849006663172153e-05,
      "loss": 0.9366,
      "step": 939
    },
    {
      "epoch": 0.20219402021940203,
      "grad_norm": 0.0,
      "learning_rate": 1.8486383130952733e-05,
      "loss": 0.8922,
      "step": 940
    },
    {
      "epoch": 0.202409120240912,
      "grad_norm": 0.0,
      "learning_rate": 1.8482695510519027e-05,
      "loss": 0.8667,
      "step": 941
    },
    {
      "epoch": 0.20262422026242202,
      "grad_norm": 0.0,
      "learning_rate": 1.8479003772210548e-05,
      "loss": 0.916,
      "step": 942
    },
    {
      "epoch": 0.20283932028393203,
      "grad_norm": 0.0,
      "learning_rate": 1.847530791781943e-05,
      "loss": 0.9862,
      "step": 943
    },
    {
      "epoch": 0.20305442030544202,
      "grad_norm": 0.0,
      "learning_rate": 1.84716079491398e-05,
      "loss": 0.9666,
      "step": 944
    },
    {
      "epoch": 0.20326952032695203,
      "grad_norm": 0.0,
      "learning_rate": 1.846790386796779e-05,
      "loss": 0.9509,
      "step": 945
    },
    {
      "epoch": 0.20348462034846204,
      "grad_norm": 0.0,
      "learning_rate": 1.8464195676101527e-05,
      "loss": 0.9678,
      "step": 946
    },
    {
      "epoch": 0.20369972036997203,
      "grad_norm": 0.0,
      "learning_rate": 1.8460483375341123e-05,
      "loss": 0.9917,
      "step": 947
    },
    {
      "epoch": 0.20391482039148204,
      "grad_norm": 0.0,
      "learning_rate": 1.84567669674887e-05,
      "loss": 0.926,
      "step": 948
    },
    {
      "epoch": 0.20412992041299205,
      "grad_norm": 0.0,
      "learning_rate": 1.845304645434836e-05,
      "loss": 0.9559,
      "step": 949
    },
    {
      "epoch": 0.20434502043450203,
      "grad_norm": 0.0,
      "learning_rate": 1.8449321837726207e-05,
      "loss": 0.9758,
      "step": 950
    },
    {
      "epoch": 0.20456012045601205,
      "grad_norm": 0.0,
      "learning_rate": 1.8445593119430337e-05,
      "loss": 0.9428,
      "step": 951
    },
    {
      "epoch": 0.20477522047752206,
      "grad_norm": 0.0,
      "learning_rate": 1.844186030127083e-05,
      "loss": 0.9387,
      "step": 952
    },
    {
      "epoch": 0.20499032049903204,
      "grad_norm": 0.0,
      "learning_rate": 1.8438123385059767e-05,
      "loss": 0.9684,
      "step": 953
    },
    {
      "epoch": 0.20520542052054205,
      "grad_norm": 0.0,
      "learning_rate": 1.84343823726112e-05,
      "loss": 1.0092,
      "step": 954
    },
    {
      "epoch": 0.20542052054205207,
      "grad_norm": 0.0,
      "learning_rate": 1.8430637265741196e-05,
      "loss": 0.9063,
      "step": 955
    },
    {
      "epoch": 0.20563562056356205,
      "grad_norm": 0.0,
      "learning_rate": 1.8426888066267787e-05,
      "loss": 0.9236,
      "step": 956
    },
    {
      "epoch": 0.20585072058507206,
      "grad_norm": 0.0,
      "learning_rate": 1.8423134776011e-05,
      "loss": 0.8681,
      "step": 957
    },
    {
      "epoch": 0.20606582060658207,
      "grad_norm": 0.0,
      "learning_rate": 1.841937739679285e-05,
      "loss": 0.9779,
      "step": 958
    },
    {
      "epoch": 0.20628092062809206,
      "grad_norm": 0.0,
      "learning_rate": 1.8415615930437335e-05,
      "loss": 0.8851,
      "step": 959
    },
    {
      "epoch": 0.20649602064960207,
      "grad_norm": 0.0,
      "learning_rate": 1.8411850378770436e-05,
      "loss": 0.9396,
      "step": 960
    },
    {
      "epoch": 0.20671112067111208,
      "grad_norm": 0.0,
      "learning_rate": 1.8408080743620117e-05,
      "loss": 0.9375,
      "step": 961
    },
    {
      "epoch": 0.20692622069262206,
      "grad_norm": 0.0,
      "learning_rate": 1.8404307026816326e-05,
      "loss": 0.8708,
      "step": 962
    },
    {
      "epoch": 0.20714132071413208,
      "grad_norm": 0.0,
      "learning_rate": 1.8400529230190992e-05,
      "loss": 0.9968,
      "step": 963
    },
    {
      "epoch": 0.20735642073564206,
      "grad_norm": 0.0,
      "learning_rate": 1.839674735557802e-05,
      "loss": 0.9683,
      "step": 964
    },
    {
      "epoch": 0.20757152075715207,
      "grad_norm": 0.0,
      "learning_rate": 1.8392961404813307e-05,
      "loss": 0.9492,
      "step": 965
    },
    {
      "epoch": 0.20778662077866208,
      "grad_norm": 0.0,
      "learning_rate": 1.8389171379734714e-05,
      "loss": 0.9667,
      "step": 966
    },
    {
      "epoch": 0.20800172080017207,
      "grad_norm": 0.0,
      "learning_rate": 1.838537728218209e-05,
      "loss": 1.0232,
      "step": 967
    },
    {
      "epoch": 0.20821682082168208,
      "grad_norm": 0.0,
      "learning_rate": 1.8381579113997255e-05,
      "loss": 0.9801,
      "step": 968
    },
    {
      "epoch": 0.2084319208431921,
      "grad_norm": 0.0,
      "learning_rate": 1.837777687702401e-05,
      "loss": 0.8881,
      "step": 969
    },
    {
      "epoch": 0.20864702086470208,
      "grad_norm": 0.0,
      "learning_rate": 1.8373970573108125e-05,
      "loss": 0.9856,
      "step": 970
    },
    {
      "epoch": 0.2088621208862121,
      "grad_norm": 0.0,
      "learning_rate": 1.8370160204097352e-05,
      "loss": 0.9458,
      "step": 971
    },
    {
      "epoch": 0.2090772209077221,
      "grad_norm": 0.0,
      "learning_rate": 1.836634577184141e-05,
      "loss": 0.9579,
      "step": 972
    },
    {
      "epoch": 0.20929232092923208,
      "grad_norm": 0.0,
      "learning_rate": 1.8362527278191993e-05,
      "loss": 0.9369,
      "step": 973
    },
    {
      "epoch": 0.2095074209507421,
      "grad_norm": 0.0,
      "learning_rate": 1.8358704725002766e-05,
      "loss": 0.9378,
      "step": 974
    },
    {
      "epoch": 0.2097225209722521,
      "grad_norm": 0.0,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 0.9235,
      "step": 975
    },
    {
      "epoch": 0.2099376209937621,
      "grad_norm": 0.0,
      "learning_rate": 1.83510474474294e-05,
      "loss": 0.9467,
      "step": 976
    },
    {
      "epoch": 0.2101527210152721,
      "grad_norm": 0.0,
      "learning_rate": 1.8347212726762436e-05,
      "loss": 0.969,
      "step": 977
    },
    {
      "epoch": 0.21036782103678212,
      "grad_norm": 0.0,
      "learning_rate": 1.8343373953990028e-05,
      "loss": 0.9659,
      "step": 978
    },
    {
      "epoch": 0.2105829210582921,
      "grad_norm": 0.0,
      "learning_rate": 1.8339531130975676e-05,
      "loss": 0.9662,
      "step": 979
    },
    {
      "epoch": 0.2107980210798021,
      "grad_norm": 0.0,
      "learning_rate": 1.8335684259584855e-05,
      "loss": 0.9583,
      "step": 980
    },
    {
      "epoch": 0.21101312110131212,
      "grad_norm": 0.0,
      "learning_rate": 1.8331833341685017e-05,
      "loss": 0.9931,
      "step": 981
    },
    {
      "epoch": 0.2112282211228221,
      "grad_norm": 0.0,
      "learning_rate": 1.832797837914556e-05,
      "loss": 0.8541,
      "step": 982
    },
    {
      "epoch": 0.21144332114433212,
      "grad_norm": 0.0,
      "learning_rate": 1.832411937383785e-05,
      "loss": 0.9403,
      "step": 983
    },
    {
      "epoch": 0.2116584211658421,
      "grad_norm": 0.0,
      "learning_rate": 1.8320256327635225e-05,
      "loss": 0.9599,
      "step": 984
    },
    {
      "epoch": 0.21187352118735212,
      "grad_norm": 0.0,
      "learning_rate": 1.831638924241298e-05,
      "loss": 0.9212,
      "step": 985
    },
    {
      "epoch": 0.21208862120886213,
      "grad_norm": 0.0,
      "learning_rate": 1.8312518120048354e-05,
      "loss": 0.9633,
      "step": 986
    },
    {
      "epoch": 0.2123037212303721,
      "grad_norm": 0.0,
      "learning_rate": 1.8308642962420577e-05,
      "loss": 0.9901,
      "step": 987
    },
    {
      "epoch": 0.21251882125188212,
      "grad_norm": 0.0,
      "learning_rate": 1.8304763771410815e-05,
      "loss": 0.9193,
      "step": 988
    },
    {
      "epoch": 0.21273392127339213,
      "grad_norm": 0.0,
      "learning_rate": 1.83008805489022e-05,
      "loss": 0.9436,
      "step": 989
    },
    {
      "epoch": 0.21294902129490212,
      "grad_norm": 0.0,
      "learning_rate": 1.8296993296779818e-05,
      "loss": 0.9382,
      "step": 990
    },
    {
      "epoch": 0.21316412131641213,
      "grad_norm": 0.0,
      "learning_rate": 1.8293102016930713e-05,
      "loss": 0.8616,
      "step": 991
    },
    {
      "epoch": 0.21337922133792214,
      "grad_norm": 0.0,
      "learning_rate": 1.8289206711243883e-05,
      "loss": 0.8718,
      "step": 992
    },
    {
      "epoch": 0.21359432135943213,
      "grad_norm": 0.0,
      "learning_rate": 1.8285307381610283e-05,
      "loss": 0.913,
      "step": 993
    },
    {
      "epoch": 0.21380942138094214,
      "grad_norm": 0.0,
      "learning_rate": 1.828140402992282e-05,
      "loss": 0.9259,
      "step": 994
    },
    {
      "epoch": 0.21402452140245215,
      "grad_norm": 0.0,
      "learning_rate": 1.827749665807635e-05,
      "loss": 0.9289,
      "step": 995
    },
    {
      "epoch": 0.21423962142396213,
      "grad_norm": 0.0,
      "learning_rate": 1.8273585267967692e-05,
      "loss": 0.956,
      "step": 996
    },
    {
      "epoch": 0.21445472144547215,
      "grad_norm": 0.0,
      "learning_rate": 1.8269669861495595e-05,
      "loss": 0.9649,
      "step": 997
    },
    {
      "epoch": 0.21466982146698216,
      "grad_norm": 0.0,
      "learning_rate": 1.8265750440560783e-05,
      "loss": 0.9886,
      "step": 998
    },
    {
      "epoch": 0.21488492148849214,
      "grad_norm": 0.0,
      "learning_rate": 1.8261827007065904e-05,
      "loss": 0.9226,
      "step": 999
    },
    {
      "epoch": 0.21510002151000215,
      "grad_norm": 0.0,
      "learning_rate": 1.8257899562915577e-05,
      "loss": 0.9202,
      "step": 1000
    },
    {
      "epoch": 0.21531512153151217,
      "grad_norm": 0.0,
      "learning_rate": 1.8253968110016343e-05,
      "loss": 0.9085,
      "step": 1001
    },
    {
      "epoch": 0.21553022155302215,
      "grad_norm": 0.0,
      "learning_rate": 1.8250032650276718e-05,
      "loss": 1.0105,
      "step": 1002
    },
    {
      "epoch": 0.21574532157453216,
      "grad_norm": 0.0,
      "learning_rate": 1.8246093185607137e-05,
      "loss": 0.9581,
      "step": 1003
    },
    {
      "epoch": 0.21596042159604215,
      "grad_norm": 0.0,
      "learning_rate": 1.8242149717919993e-05,
      "loss": 0.9517,
      "step": 1004
    },
    {
      "epoch": 0.21617552161755216,
      "grad_norm": 0.0,
      "learning_rate": 1.8238202249129625e-05,
      "loss": 0.9074,
      "step": 1005
    },
    {
      "epoch": 0.21639062163906217,
      "grad_norm": 0.0,
      "learning_rate": 1.82342507811523e-05,
      "loss": 0.9313,
      "step": 1006
    },
    {
      "epoch": 0.21660572166057215,
      "grad_norm": 0.0,
      "learning_rate": 1.8230295315906237e-05,
      "loss": 0.8858,
      "step": 1007
    },
    {
      "epoch": 0.21682082168208217,
      "grad_norm": 0.0,
      "learning_rate": 1.8226335855311598e-05,
      "loss": 0.9052,
      "step": 1008
    },
    {
      "epoch": 0.21703592170359218,
      "grad_norm": 0.0,
      "learning_rate": 1.8222372401290475e-05,
      "loss": 0.9322,
      "step": 1009
    },
    {
      "epoch": 0.21725102172510216,
      "grad_norm": 0.0,
      "learning_rate": 1.8218404955766905e-05,
      "loss": 0.9991,
      "step": 1010
    },
    {
      "epoch": 0.21746612174661217,
      "grad_norm": 0.0,
      "learning_rate": 1.821443352066686e-05,
      "loss": 0.9553,
      "step": 1011
    },
    {
      "epoch": 0.21768122176812219,
      "grad_norm": 0.0,
      "learning_rate": 1.8210458097918255e-05,
      "loss": 0.9855,
      "step": 1012
    },
    {
      "epoch": 0.21789632178963217,
      "grad_norm": 0.0,
      "learning_rate": 1.820647868945093e-05,
      "loss": 0.9742,
      "step": 1013
    },
    {
      "epoch": 0.21811142181114218,
      "grad_norm": 0.0,
      "learning_rate": 1.820249529719667e-05,
      "loss": 0.9182,
      "step": 1014
    },
    {
      "epoch": 0.2183265218326522,
      "grad_norm": 0.0,
      "learning_rate": 1.8198507923089185e-05,
      "loss": 0.9222,
      "step": 1015
    },
    {
      "epoch": 0.21854162185416218,
      "grad_norm": 0.0,
      "learning_rate": 1.819451656906412e-05,
      "loss": 0.9579,
      "step": 1016
    },
    {
      "epoch": 0.2187567218756722,
      "grad_norm": 0.0,
      "learning_rate": 1.8190521237059066e-05,
      "loss": 0.9256,
      "step": 1017
    },
    {
      "epoch": 0.2189718218971822,
      "grad_norm": 0.0,
      "learning_rate": 1.8186521929013523e-05,
      "loss": 0.9561,
      "step": 1018
    },
    {
      "epoch": 0.21918692191869218,
      "grad_norm": 0.0,
      "learning_rate": 1.818251864686893e-05,
      "loss": 0.9457,
      "step": 1019
    },
    {
      "epoch": 0.2194020219402022,
      "grad_norm": 0.0,
      "learning_rate": 1.8178511392568667e-05,
      "loss": 0.9749,
      "step": 1020
    },
    {
      "epoch": 0.2196171219617122,
      "grad_norm": 0.0,
      "learning_rate": 1.817450016805803e-05,
      "loss": 1.0254,
      "step": 1021
    },
    {
      "epoch": 0.2198322219832222,
      "grad_norm": 0.0,
      "learning_rate": 1.8170484975284233e-05,
      "loss": 0.9185,
      "step": 1022
    },
    {
      "epoch": 0.2200473220047322,
      "grad_norm": 0.0,
      "learning_rate": 1.8166465816196436e-05,
      "loss": 0.9991,
      "step": 1023
    },
    {
      "epoch": 0.22026242202624222,
      "grad_norm": 0.0,
      "learning_rate": 1.8162442692745722e-05,
      "loss": 0.9264,
      "step": 1024
    },
    {
      "epoch": 0.2204775220477522,
      "grad_norm": 0.0,
      "learning_rate": 1.8158415606885083e-05,
      "loss": 0.9932,
      "step": 1025
    },
    {
      "epoch": 0.2206926220692622,
      "grad_norm": 0.0,
      "learning_rate": 1.815438456056945e-05,
      "loss": 0.9195,
      "step": 1026
    },
    {
      "epoch": 0.2209077220907722,
      "grad_norm": 0.0,
      "learning_rate": 1.8150349555755666e-05,
      "loss": 0.9347,
      "step": 1027
    },
    {
      "epoch": 0.2211228221122822,
      "grad_norm": 0.0,
      "learning_rate": 1.8146310594402504e-05,
      "loss": 0.914,
      "step": 1028
    },
    {
      "epoch": 0.22133792213379222,
      "grad_norm": 0.0,
      "learning_rate": 1.8142267678470654e-05,
      "loss": 0.9449,
      "step": 1029
    },
    {
      "epoch": 0.2215530221553022,
      "grad_norm": 0.0,
      "learning_rate": 1.8138220809922726e-05,
      "loss": 1.0143,
      "step": 1030
    },
    {
      "epoch": 0.22176812217681222,
      "grad_norm": 0.0,
      "learning_rate": 1.8134169990723246e-05,
      "loss": 0.9446,
      "step": 1031
    },
    {
      "epoch": 0.22198322219832223,
      "grad_norm": 0.0,
      "learning_rate": 1.8130115222838663e-05,
      "loss": 0.8984,
      "step": 1032
    },
    {
      "epoch": 0.2221983222198322,
      "grad_norm": 0.0,
      "learning_rate": 1.8126056508237338e-05,
      "loss": 0.9128,
      "step": 1033
    },
    {
      "epoch": 0.22241342224134222,
      "grad_norm": 0.0,
      "learning_rate": 1.8121993848889553e-05,
      "loss": 0.9323,
      "step": 1034
    },
    {
      "epoch": 0.22262852226285224,
      "grad_norm": 0.0,
      "learning_rate": 1.81179272467675e-05,
      "loss": 0.9356,
      "step": 1035
    },
    {
      "epoch": 0.22284362228436222,
      "grad_norm": 0.0,
      "learning_rate": 1.811385670384529e-05,
      "loss": 0.9263,
      "step": 1036
    },
    {
      "epoch": 0.22305872230587223,
      "grad_norm": 0.0,
      "learning_rate": 1.810978222209894e-05,
      "loss": 0.9717,
      "step": 1037
    },
    {
      "epoch": 0.22327382232738224,
      "grad_norm": 0.0,
      "learning_rate": 1.8105703803506384e-05,
      "loss": 0.939,
      "step": 1038
    },
    {
      "epoch": 0.22348892234889223,
      "grad_norm": 0.0,
      "learning_rate": 1.8101621450047467e-05,
      "loss": 0.9302,
      "step": 1039
    },
    {
      "epoch": 0.22370402237040224,
      "grad_norm": 0.0,
      "learning_rate": 1.8097535163703947e-05,
      "loss": 0.9615,
      "step": 1040
    },
    {
      "epoch": 0.22391912239191225,
      "grad_norm": 0.0,
      "learning_rate": 1.8093444946459483e-05,
      "loss": 0.9158,
      "step": 1041
    },
    {
      "epoch": 0.22413422241342223,
      "grad_norm": 0.0,
      "learning_rate": 1.808935080029965e-05,
      "loss": 0.9593,
      "step": 1042
    },
    {
      "epoch": 0.22434932243493225,
      "grad_norm": 0.0,
      "learning_rate": 1.8085252727211927e-05,
      "loss": 0.8494,
      "step": 1043
    },
    {
      "epoch": 0.22456442245644226,
      "grad_norm": 0.0,
      "learning_rate": 1.8081150729185696e-05,
      "loss": 0.894,
      "step": 1044
    },
    {
      "epoch": 0.22477952247795224,
      "grad_norm": 0.0,
      "learning_rate": 1.8077044808212255e-05,
      "loss": 0.941,
      "step": 1045
    },
    {
      "epoch": 0.22499462249946225,
      "grad_norm": 0.0,
      "learning_rate": 1.807293496628479e-05,
      "loss": 0.9897,
      "step": 1046
    },
    {
      "epoch": 0.22520972252097224,
      "grad_norm": 0.0,
      "learning_rate": 1.8068821205398408e-05,
      "loss": 0.9227,
      "step": 1047
    },
    {
      "epoch": 0.22542482254248225,
      "grad_norm": 0.0,
      "learning_rate": 1.8064703527550102e-05,
      "loss": 0.9508,
      "step": 1048
    },
    {
      "epoch": 0.22563992256399226,
      "grad_norm": 0.0,
      "learning_rate": 1.8060581934738783e-05,
      "loss": 0.941,
      "step": 1049
    },
    {
      "epoch": 0.22585502258550225,
      "grad_norm": 0.0,
      "learning_rate": 1.805645642896525e-05,
      "loss": 0.9313,
      "step": 1050
    },
    {
      "epoch": 0.22607012260701226,
      "grad_norm": 0.0,
      "learning_rate": 1.8052327012232203e-05,
      "loss": 0.9778,
      "step": 1051
    },
    {
      "epoch": 0.22628522262852227,
      "grad_norm": 0.0,
      "learning_rate": 1.8048193686544248e-05,
      "loss": 0.9311,
      "step": 1052
    },
    {
      "epoch": 0.22650032265003225,
      "grad_norm": 0.0,
      "learning_rate": 1.8044056453907882e-05,
      "loss": 0.9217,
      "step": 1053
    },
    {
      "epoch": 0.22671542267154227,
      "grad_norm": 0.0,
      "learning_rate": 1.80399153163315e-05,
      "loss": 0.939,
      "step": 1054
    },
    {
      "epoch": 0.22693052269305228,
      "grad_norm": 0.0,
      "learning_rate": 1.803577027582539e-05,
      "loss": 0.9667,
      "step": 1055
    },
    {
      "epoch": 0.22714562271456226,
      "grad_norm": 0.0,
      "learning_rate": 1.8031621334401743e-05,
      "loss": 1.0064,
      "step": 1056
    },
    {
      "epoch": 0.22736072273607227,
      "grad_norm": 0.0,
      "learning_rate": 1.8027468494074632e-05,
      "loss": 0.9452,
      "step": 1057
    },
    {
      "epoch": 0.22757582275758229,
      "grad_norm": 0.0,
      "learning_rate": 1.8023311756860036e-05,
      "loss": 0.9721,
      "step": 1058
    },
    {
      "epoch": 0.22779092277909227,
      "grad_norm": 0.0,
      "learning_rate": 1.8019151124775815e-05,
      "loss": 0.9272,
      "step": 1059
    },
    {
      "epoch": 0.22800602280060228,
      "grad_norm": 0.0,
      "learning_rate": 1.8014986599841724e-05,
      "loss": 0.8523,
      "step": 1060
    },
    {
      "epoch": 0.2282211228221123,
      "grad_norm": 0.0,
      "learning_rate": 1.8010818184079405e-05,
      "loss": 0.861,
      "step": 1061
    },
    {
      "epoch": 0.22843622284362228,
      "grad_norm": 0.0,
      "learning_rate": 1.8006645879512396e-05,
      "loss": 0.9106,
      "step": 1062
    },
    {
      "epoch": 0.2286513228651323,
      "grad_norm": 0.0,
      "learning_rate": 1.8002469688166115e-05,
      "loss": 0.8722,
      "step": 1063
    },
    {
      "epoch": 0.2288664228866423,
      "grad_norm": 0.0,
      "learning_rate": 1.7998289612067864e-05,
      "loss": 0.9003,
      "step": 1064
    },
    {
      "epoch": 0.22908152290815229,
      "grad_norm": 0.0,
      "learning_rate": 1.7994105653246847e-05,
      "loss": 0.946,
      "step": 1065
    },
    {
      "epoch": 0.2292966229296623,
      "grad_norm": 0.0,
      "learning_rate": 1.7989917813734136e-05,
      "loss": 0.9115,
      "step": 1066
    },
    {
      "epoch": 0.2295117229511723,
      "grad_norm": 0.0,
      "learning_rate": 1.798572609556269e-05,
      "loss": 0.939,
      "step": 1067
    },
    {
      "epoch": 0.2297268229726823,
      "grad_norm": 0.0,
      "learning_rate": 1.7981530500767365e-05,
      "loss": 0.9573,
      "step": 1068
    },
    {
      "epoch": 0.2299419229941923,
      "grad_norm": 0.0,
      "learning_rate": 1.7977331031384878e-05,
      "loss": 0.9662,
      "step": 1069
    },
    {
      "epoch": 0.2301570230157023,
      "grad_norm": 0.0,
      "learning_rate": 1.7973127689453838e-05,
      "loss": 0.9753,
      "step": 1070
    },
    {
      "epoch": 0.2303721230372123,
      "grad_norm": 0.0,
      "learning_rate": 1.7968920477014737e-05,
      "loss": 0.997,
      "step": 1071
    },
    {
      "epoch": 0.2305872230587223,
      "grad_norm": 0.0,
      "learning_rate": 1.796470939610994e-05,
      "loss": 0.9278,
      "step": 1072
    },
    {
      "epoch": 0.2308023230802323,
      "grad_norm": 0.0,
      "learning_rate": 1.7960494448783685e-05,
      "loss": 0.8825,
      "step": 1073
    },
    {
      "epoch": 0.2310174231017423,
      "grad_norm": 0.0,
      "learning_rate": 1.79562756370821e-05,
      "loss": 0.8712,
      "step": 1074
    },
    {
      "epoch": 0.23123252312325232,
      "grad_norm": 0.0,
      "learning_rate": 1.7952052963053187e-05,
      "loss": 0.933,
      "step": 1075
    },
    {
      "epoch": 0.2314476231447623,
      "grad_norm": 0.0,
      "learning_rate": 1.794782642874681e-05,
      "loss": 0.9392,
      "step": 1076
    },
    {
      "epoch": 0.23166272316627232,
      "grad_norm": 0.0,
      "learning_rate": 1.7943596036214715e-05,
      "loss": 0.8855,
      "step": 1077
    },
    {
      "epoch": 0.23187782318778233,
      "grad_norm": 0.0,
      "learning_rate": 1.793936178751053e-05,
      "loss": 0.9453,
      "step": 1078
    },
    {
      "epoch": 0.2320929232092923,
      "grad_norm": 0.0,
      "learning_rate": 1.7935123684689735e-05,
      "loss": 0.9418,
      "step": 1079
    },
    {
      "epoch": 0.23230802323080232,
      "grad_norm": 0.0,
      "learning_rate": 1.7930881729809702e-05,
      "loss": 0.8568,
      "step": 1080
    },
    {
      "epoch": 0.23252312325231234,
      "grad_norm": 0.0,
      "learning_rate": 1.7926635924929656e-05,
      "loss": 0.9345,
      "step": 1081
    },
    {
      "epoch": 0.23273822327382232,
      "grad_norm": 0.0,
      "learning_rate": 1.79223862721107e-05,
      "loss": 0.9046,
      "step": 1082
    },
    {
      "epoch": 0.23295332329533233,
      "grad_norm": 0.0,
      "learning_rate": 1.7918132773415807e-05,
      "loss": 0.9445,
      "step": 1083
    },
    {
      "epoch": 0.23316842331684234,
      "grad_norm": 0.0,
      "learning_rate": 1.7913875430909807e-05,
      "loss": 0.867,
      "step": 1084
    },
    {
      "epoch": 0.23338352333835233,
      "grad_norm": 0.0,
      "learning_rate": 1.7909614246659406e-05,
      "loss": 0.9717,
      "step": 1085
    },
    {
      "epoch": 0.23359862335986234,
      "grad_norm": 0.0,
      "learning_rate": 1.7905349222733172e-05,
      "loss": 0.951,
      "step": 1086
    },
    {
      "epoch": 0.23381372338137235,
      "grad_norm": 0.0,
      "learning_rate": 1.790108036120153e-05,
      "loss": 0.9438,
      "step": 1087
    },
    {
      "epoch": 0.23402882340288234,
      "grad_norm": 0.0,
      "learning_rate": 1.789680766413678e-05,
      "loss": 0.945,
      "step": 1088
    },
    {
      "epoch": 0.23424392342439235,
      "grad_norm": 0.0,
      "learning_rate": 1.7892531133613074e-05,
      "loss": 0.9401,
      "step": 1089
    },
    {
      "epoch": 0.23445902344590233,
      "grad_norm": 0.0,
      "learning_rate": 1.788825077170643e-05,
      "loss": 0.9067,
      "step": 1090
    },
    {
      "epoch": 0.23467412346741234,
      "grad_norm": 0.0,
      "learning_rate": 1.7883966580494722e-05,
      "loss": 0.9348,
      "step": 1091
    },
    {
      "epoch": 0.23488922348892236,
      "grad_norm": 0.0,
      "learning_rate": 1.7879678562057688e-05,
      "loss": 0.9277,
      "step": 1092
    },
    {
      "epoch": 0.23510432351043234,
      "grad_norm": 0.0,
      "learning_rate": 1.7875386718476924e-05,
      "loss": 0.9813,
      "step": 1093
    },
    {
      "epoch": 0.23531942353194235,
      "grad_norm": 0.0,
      "learning_rate": 1.7871091051835873e-05,
      "loss": 0.8442,
      "step": 1094
    },
    {
      "epoch": 0.23553452355345236,
      "grad_norm": 0.0,
      "learning_rate": 1.7866791564219853e-05,
      "loss": 0.8753,
      "step": 1095
    },
    {
      "epoch": 0.23574962357496235,
      "grad_norm": 0.0,
      "learning_rate": 1.786248825771601e-05,
      "loss": 0.8609,
      "step": 1096
    },
    {
      "epoch": 0.23596472359647236,
      "grad_norm": 0.0,
      "learning_rate": 1.785818113441337e-05,
      "loss": 0.9574,
      "step": 1097
    },
    {
      "epoch": 0.23617982361798237,
      "grad_norm": 0.0,
      "learning_rate": 1.78538701964028e-05,
      "loss": 0.9111,
      "step": 1098
    },
    {
      "epoch": 0.23639492363949235,
      "grad_norm": 0.0,
      "learning_rate": 1.784955544577701e-05,
      "loss": 0.9431,
      "step": 1099
    },
    {
      "epoch": 0.23661002366100237,
      "grad_norm": 0.0,
      "learning_rate": 1.7845236884630583e-05,
      "loss": 0.9375,
      "step": 1100
    },
    {
      "epoch": 0.23682512368251238,
      "grad_norm": 0.0,
      "learning_rate": 1.7840914515059934e-05,
      "loss": 0.9123,
      "step": 1101
    },
    {
      "epoch": 0.23704022370402236,
      "grad_norm": 0.0,
      "learning_rate": 1.7836588339163324e-05,
      "loss": 0.9345,
      "step": 1102
    },
    {
      "epoch": 0.23725532372553237,
      "grad_norm": 0.0,
      "learning_rate": 1.7832258359040885e-05,
      "loss": 0.9212,
      "step": 1103
    },
    {
      "epoch": 0.2374704237470424,
      "grad_norm": 0.0,
      "learning_rate": 1.7827924576794568e-05,
      "loss": 0.811,
      "step": 1104
    },
    {
      "epoch": 0.23768552376855237,
      "grad_norm": 0.0,
      "learning_rate": 1.782358699452819e-05,
      "loss": 0.9544,
      "step": 1105
    },
    {
      "epoch": 0.23790062379006238,
      "grad_norm": 0.0,
      "learning_rate": 1.78192456143474e-05,
      "loss": 0.9565,
      "step": 1106
    },
    {
      "epoch": 0.2381157238115724,
      "grad_norm": 0.0,
      "learning_rate": 1.7814900438359694e-05,
      "loss": 0.9163,
      "step": 1107
    },
    {
      "epoch": 0.23833082383308238,
      "grad_norm": 0.0,
      "learning_rate": 1.781055146867442e-05,
      "loss": 0.9068,
      "step": 1108
    },
    {
      "epoch": 0.2385459238545924,
      "grad_norm": 0.0,
      "learning_rate": 1.7806198707402754e-05,
      "loss": 0.8795,
      "step": 1109
    },
    {
      "epoch": 0.23876102387610237,
      "grad_norm": 0.0,
      "learning_rate": 1.7801842156657718e-05,
      "loss": 0.9681,
      "step": 1110
    },
    {
      "epoch": 0.23897612389761239,
      "grad_norm": 0.0,
      "learning_rate": 1.7797481818554177e-05,
      "loss": 0.9561,
      "step": 1111
    },
    {
      "epoch": 0.2391912239191224,
      "grad_norm": 0.0,
      "learning_rate": 1.7793117695208828e-05,
      "loss": 0.9202,
      "step": 1112
    },
    {
      "epoch": 0.23940632394063238,
      "grad_norm": 0.0,
      "learning_rate": 1.7788749788740213e-05,
      "loss": 0.897,
      "step": 1113
    },
    {
      "epoch": 0.2396214239621424,
      "grad_norm": 0.0,
      "learning_rate": 1.7784378101268703e-05,
      "loss": 0.9249,
      "step": 1114
    },
    {
      "epoch": 0.2398365239836524,
      "grad_norm": 0.0,
      "learning_rate": 1.7780002634916512e-05,
      "loss": 0.9146,
      "step": 1115
    },
    {
      "epoch": 0.2400516240051624,
      "grad_norm": 0.0,
      "learning_rate": 1.7775623391807683e-05,
      "loss": 0.9302,
      "step": 1116
    },
    {
      "epoch": 0.2402667240266724,
      "grad_norm": 0.0,
      "learning_rate": 1.777124037406809e-05,
      "loss": 0.9865,
      "step": 1117
    },
    {
      "epoch": 0.2404818240481824,
      "grad_norm": 0.0,
      "learning_rate": 1.7766853583825447e-05,
      "loss": 0.9255,
      "step": 1118
    },
    {
      "epoch": 0.2406969240696924,
      "grad_norm": 0.0,
      "learning_rate": 1.7762463023209296e-05,
      "loss": 0.9343,
      "step": 1119
    },
    {
      "epoch": 0.2409120240912024,
      "grad_norm": 0.0,
      "learning_rate": 1.7758068694351007e-05,
      "loss": 0.957,
      "step": 1120
    },
    {
      "epoch": 0.24112712411271242,
      "grad_norm": 0.0,
      "learning_rate": 1.7753670599383784e-05,
      "loss": 0.9204,
      "step": 1121
    },
    {
      "epoch": 0.2413422241342224,
      "grad_norm": 0.0,
      "learning_rate": 1.7749268740442652e-05,
      "loss": 0.947,
      "step": 1122
    },
    {
      "epoch": 0.24155732415573242,
      "grad_norm": 0.0,
      "learning_rate": 1.774486311966447e-05,
      "loss": 0.8804,
      "step": 1123
    },
    {
      "epoch": 0.24177242417724243,
      "grad_norm": 0.0,
      "learning_rate": 1.7740453739187922e-05,
      "loss": 0.9007,
      "step": 1124
    },
    {
      "epoch": 0.2419875241987524,
      "grad_norm": 0.0,
      "learning_rate": 1.7736040601153517e-05,
      "loss": 0.949,
      "step": 1125
    },
    {
      "epoch": 0.24220262422026242,
      "grad_norm": 0.0,
      "learning_rate": 1.7731623707703584e-05,
      "loss": 0.9155,
      "step": 1126
    },
    {
      "epoch": 0.24241772424177244,
      "grad_norm": 0.0,
      "learning_rate": 1.7727203060982273e-05,
      "loss": 0.9138,
      "step": 1127
    },
    {
      "epoch": 0.24263282426328242,
      "grad_norm": 0.0,
      "learning_rate": 1.7722778663135568e-05,
      "loss": 0.8795,
      "step": 1128
    },
    {
      "epoch": 0.24284792428479243,
      "grad_norm": 0.0,
      "learning_rate": 1.771835051631127e-05,
      "loss": 0.9158,
      "step": 1129
    },
    {
      "epoch": 0.24306302430630244,
      "grad_norm": 0.0,
      "learning_rate": 1.771391862265899e-05,
      "loss": 0.8936,
      "step": 1130
    },
    {
      "epoch": 0.24327812432781243,
      "grad_norm": 0.0,
      "learning_rate": 1.770948298433017e-05,
      "loss": 1.0101,
      "step": 1131
    },
    {
      "epoch": 0.24349322434932244,
      "grad_norm": 0.0,
      "learning_rate": 1.7705043603478058e-05,
      "loss": 0.943,
      "step": 1132
    },
    {
      "epoch": 0.24370832437083242,
      "grad_norm": 0.0,
      "learning_rate": 1.7700600482257732e-05,
      "loss": 0.8986,
      "step": 1133
    },
    {
      "epoch": 0.24392342439234244,
      "grad_norm": 0.0,
      "learning_rate": 1.7696153622826077e-05,
      "loss": 0.9182,
      "step": 1134
    },
    {
      "epoch": 0.24413852441385245,
      "grad_norm": 0.0,
      "learning_rate": 1.769170302734179e-05,
      "loss": 0.8997,
      "step": 1135
    },
    {
      "epoch": 0.24435362443536243,
      "grad_norm": 0.0,
      "learning_rate": 1.7687248697965395e-05,
      "loss": 0.9676,
      "step": 1136
    },
    {
      "epoch": 0.24456872445687244,
      "grad_norm": 0.0,
      "learning_rate": 1.7682790636859217e-05,
      "loss": 0.9192,
      "step": 1137
    },
    {
      "epoch": 0.24478382447838246,
      "grad_norm": 0.0,
      "learning_rate": 1.767832884618739e-05,
      "loss": 0.9487,
      "step": 1138
    },
    {
      "epoch": 0.24499892449989244,
      "grad_norm": 0.0,
      "learning_rate": 1.767386332811587e-05,
      "loss": 0.8918,
      "step": 1139
    },
    {
      "epoch": 0.24521402452140245,
      "grad_norm": 0.0,
      "learning_rate": 1.7669394084812417e-05,
      "loss": 0.9278,
      "step": 1140
    },
    {
      "epoch": 0.24542912454291246,
      "grad_norm": 0.0,
      "learning_rate": 1.7664921118446595e-05,
      "loss": 0.9632,
      "step": 1141
    },
    {
      "epoch": 0.24564422456442245,
      "grad_norm": 0.0,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.8185,
      "step": 1142
    },
    {
      "epoch": 0.24585932458593246,
      "grad_norm": 0.0,
      "learning_rate": 1.7655964025215158e-05,
      "loss": 0.8948,
      "step": 1143
    },
    {
      "epoch": 0.24607442460744247,
      "grad_norm": 0.0,
      "learning_rate": 1.765147990269771e-05,
      "loss": 0.8928,
      "step": 1144
    },
    {
      "epoch": 0.24628952462895246,
      "grad_norm": 0.0,
      "learning_rate": 1.7646992065814233e-05,
      "loss": 0.8884,
      "step": 1145
    },
    {
      "epoch": 0.24650462465046247,
      "grad_norm": 0.0,
      "learning_rate": 1.7642500516743318e-05,
      "loss": 0.9392,
      "step": 1146
    },
    {
      "epoch": 0.24671972467197248,
      "grad_norm": 0.0,
      "learning_rate": 1.7638005257665356e-05,
      "loss": 0.9711,
      "step": 1147
    },
    {
      "epoch": 0.24693482469348246,
      "grad_norm": 0.0,
      "learning_rate": 1.763350629076255e-05,
      "loss": 0.9221,
      "step": 1148
    },
    {
      "epoch": 0.24714992471499247,
      "grad_norm": 0.0,
      "learning_rate": 1.76290036182189e-05,
      "loss": 0.9367,
      "step": 1149
    },
    {
      "epoch": 0.2473650247365025,
      "grad_norm": 0.0,
      "learning_rate": 1.7624497242220197e-05,
      "loss": 0.9145,
      "step": 1150
    },
    {
      "epoch": 0.24758012475801247,
      "grad_norm": 0.0,
      "learning_rate": 1.7619987164954034e-05,
      "loss": 0.9248,
      "step": 1151
    },
    {
      "epoch": 0.24779522477952248,
      "grad_norm": 0.0,
      "learning_rate": 1.7615473388609806e-05,
      "loss": 0.9956,
      "step": 1152
    },
    {
      "epoch": 0.24801032480103247,
      "grad_norm": 0.0,
      "learning_rate": 1.7610955915378692e-05,
      "loss": 0.9594,
      "step": 1153
    },
    {
      "epoch": 0.24822542482254248,
      "grad_norm": 0.0,
      "learning_rate": 1.760643474745368e-05,
      "loss": 0.9701,
      "step": 1154
    },
    {
      "epoch": 0.2484405248440525,
      "grad_norm": 0.0,
      "learning_rate": 1.7601909887029544e-05,
      "loss": 0.9441,
      "step": 1155
    },
    {
      "epoch": 0.24865562486556247,
      "grad_norm": 0.0,
      "learning_rate": 1.7597381336302848e-05,
      "loss": 0.9425,
      "step": 1156
    },
    {
      "epoch": 0.2488707248870725,
      "grad_norm": 0.0,
      "learning_rate": 1.759284909747195e-05,
      "loss": 0.8754,
      "step": 1157
    },
    {
      "epoch": 0.2490858249085825,
      "grad_norm": 0.0,
      "learning_rate": 1.7588313172737002e-05,
      "loss": 0.936,
      "step": 1158
    },
    {
      "epoch": 0.24930092493009248,
      "grad_norm": 0.0,
      "learning_rate": 1.7583773564299938e-05,
      "loss": 0.9462,
      "step": 1159
    },
    {
      "epoch": 0.2495160249516025,
      "grad_norm": 0.0,
      "learning_rate": 1.7579230274364486e-05,
      "loss": 0.8976,
      "step": 1160
    },
    {
      "epoch": 0.2497311249731125,
      "grad_norm": 0.0,
      "learning_rate": 1.7574683305136162e-05,
      "loss": 0.9256,
      "step": 1161
    },
    {
      "epoch": 0.2499462249946225,
      "grad_norm": 0.0,
      "learning_rate": 1.7570132658822264e-05,
      "loss": 0.9317,
      "step": 1162
    },
    {
      "epoch": 0.2501613250161325,
      "grad_norm": 0.0,
      "learning_rate": 1.7565578337631876e-05,
      "loss": 0.8749,
      "step": 1163
    },
    {
      "epoch": 0.2503764250376425,
      "grad_norm": 0.0,
      "learning_rate": 1.7561020343775862e-05,
      "loss": 0.8859,
      "step": 1164
    },
    {
      "epoch": 0.2505915250591525,
      "grad_norm": 0.0,
      "learning_rate": 1.755645867946688e-05,
      "loss": 0.9065,
      "step": 1165
    },
    {
      "epoch": 0.2508066250806625,
      "grad_norm": 0.0,
      "learning_rate": 1.7551893346919363e-05,
      "loss": 0.9135,
      "step": 1166
    },
    {
      "epoch": 0.2510217251021725,
      "grad_norm": 0.0,
      "learning_rate": 1.7547324348349523e-05,
      "loss": 0.957,
      "step": 1167
    },
    {
      "epoch": 0.2512368251236825,
      "grad_norm": 0.0,
      "learning_rate": 1.754275168597535e-05,
      "loss": 0.9988,
      "step": 1168
    },
    {
      "epoch": 0.2514519251451925,
      "grad_norm": 0.0,
      "learning_rate": 1.7538175362016622e-05,
      "loss": 0.9433,
      "step": 1169
    },
    {
      "epoch": 0.25166702516670253,
      "grad_norm": 0.0,
      "learning_rate": 1.753359537869489e-05,
      "loss": 1.026,
      "step": 1170
    },
    {
      "epoch": 0.2518821251882125,
      "grad_norm": 0.0,
      "learning_rate": 1.752901173823347e-05,
      "loss": 0.9297,
      "step": 1171
    },
    {
      "epoch": 0.2520972252097225,
      "grad_norm": 0.0,
      "learning_rate": 1.752442444285747e-05,
      "loss": 0.9677,
      "step": 1172
    },
    {
      "epoch": 0.25231232523123254,
      "grad_norm": 0.0,
      "learning_rate": 1.751983349479377e-05,
      "loss": 0.8354,
      "step": 1173
    },
    {
      "epoch": 0.2525274252527425,
      "grad_norm": 0.0,
      "learning_rate": 1.751523889627101e-05,
      "loss": 0.9229,
      "step": 1174
    },
    {
      "epoch": 0.2527425252742525,
      "grad_norm": 0.0,
      "learning_rate": 1.7510640649519617e-05,
      "loss": 0.9658,
      "step": 1175
    },
    {
      "epoch": 0.25295762529576254,
      "grad_norm": 0.0,
      "learning_rate": 1.750603875677178e-05,
      "loss": 0.9644,
      "step": 1176
    },
    {
      "epoch": 0.25317272531727253,
      "grad_norm": 0.0,
      "learning_rate": 1.7501433220261463e-05,
      "loss": 0.954,
      "step": 1177
    },
    {
      "epoch": 0.2533878253387825,
      "grad_norm": 0.0,
      "learning_rate": 1.7496824042224393e-05,
      "loss": 0.8957,
      "step": 1178
    },
    {
      "epoch": 0.25360292536029255,
      "grad_norm": 0.0,
      "learning_rate": 1.749221122489807e-05,
      "loss": 0.9355,
      "step": 1179
    },
    {
      "epoch": 0.25381802538180254,
      "grad_norm": 0.0,
      "learning_rate": 1.7487594770521764e-05,
      "loss": 0.8626,
      "step": 1180
    },
    {
      "epoch": 0.2540331254033125,
      "grad_norm": 0.0,
      "learning_rate": 1.7482974681336502e-05,
      "loss": 0.9312,
      "step": 1181
    },
    {
      "epoch": 0.25424822542482256,
      "grad_norm": 0.0,
      "learning_rate": 1.7478350959585074e-05,
      "loss": 0.9908,
      "step": 1182
    },
    {
      "epoch": 0.25446332544633254,
      "grad_norm": 0.0,
      "learning_rate": 1.7473723607512048e-05,
      "loss": 0.9033,
      "step": 1183
    },
    {
      "epoch": 0.25467842546784253,
      "grad_norm": 0.0,
      "learning_rate": 1.7469092627363738e-05,
      "loss": 0.8185,
      "step": 1184
    },
    {
      "epoch": 0.25489352548935257,
      "grad_norm": 0.0,
      "learning_rate": 1.746445802138823e-05,
      "loss": 0.9368,
      "step": 1185
    },
    {
      "epoch": 0.25510862551086255,
      "grad_norm": 0.0,
      "learning_rate": 1.745981979183536e-05,
      "loss": 0.9934,
      "step": 1186
    },
    {
      "epoch": 0.25532372553237254,
      "grad_norm": 0.0,
      "learning_rate": 1.7455177940956743e-05,
      "loss": 0.9105,
      "step": 1187
    },
    {
      "epoch": 0.2555388255538826,
      "grad_norm": 0.0,
      "learning_rate": 1.745053247100572e-05,
      "loss": 0.9465,
      "step": 1188
    },
    {
      "epoch": 0.25575392557539256,
      "grad_norm": 0.0,
      "learning_rate": 1.7445883384237423e-05,
      "loss": 0.9586,
      "step": 1189
    },
    {
      "epoch": 0.25596902559690254,
      "grad_norm": 0.0,
      "learning_rate": 1.7441230682908713e-05,
      "loss": 0.9355,
      "step": 1190
    },
    {
      "epoch": 0.2561841256184126,
      "grad_norm": 0.0,
      "learning_rate": 1.7436574369278224e-05,
      "loss": 0.8793,
      "step": 1191
    },
    {
      "epoch": 0.25639922563992257,
      "grad_norm": 0.0,
      "learning_rate": 1.7431914445606334e-05,
      "loss": 0.9604,
      "step": 1192
    },
    {
      "epoch": 0.25661432566143255,
      "grad_norm": 0.0,
      "learning_rate": 1.742725091415517e-05,
      "loss": 0.8647,
      "step": 1193
    },
    {
      "epoch": 0.2568294256829426,
      "grad_norm": 0.0,
      "learning_rate": 1.7422583777188625e-05,
      "loss": 0.9573,
      "step": 1194
    },
    {
      "epoch": 0.2570445257044526,
      "grad_norm": 0.0,
      "learning_rate": 1.7417913036972328e-05,
      "loss": 0.9553,
      "step": 1195
    },
    {
      "epoch": 0.25725962572596256,
      "grad_norm": 0.0,
      "learning_rate": 1.741323869577366e-05,
      "loss": 0.911,
      "step": 1196
    },
    {
      "epoch": 0.2574747257474726,
      "grad_norm": 0.0,
      "learning_rate": 1.7408560755861762e-05,
      "loss": 0.9172,
      "step": 1197
    },
    {
      "epoch": 0.2576898257689826,
      "grad_norm": 0.0,
      "learning_rate": 1.7403879219507504e-05,
      "loss": 0.8859,
      "step": 1198
    },
    {
      "epoch": 0.25790492579049257,
      "grad_norm": 0.0,
      "learning_rate": 1.7399194088983513e-05,
      "loss": 0.9185,
      "step": 1199
    },
    {
      "epoch": 0.2581200258120026,
      "grad_norm": 0.0,
      "learning_rate": 1.7394505366564155e-05,
      "loss": 0.8845,
      "step": 1200
    },
    {
      "epoch": 0.2583351258335126,
      "grad_norm": 0.0,
      "learning_rate": 1.738981305452555e-05,
      "loss": 0.9572,
      "step": 1201
    },
    {
      "epoch": 0.2585502258550226,
      "grad_norm": 0.0,
      "learning_rate": 1.7385117155145547e-05,
      "loss": 0.8483,
      "step": 1202
    },
    {
      "epoch": 0.2587653258765326,
      "grad_norm": 0.0,
      "learning_rate": 1.7380417670703744e-05,
      "loss": 0.928,
      "step": 1203
    },
    {
      "epoch": 0.2589804258980426,
      "grad_norm": 0.0,
      "learning_rate": 1.7375714603481483e-05,
      "loss": 0.9026,
      "step": 1204
    },
    {
      "epoch": 0.2591955259195526,
      "grad_norm": 0.0,
      "learning_rate": 1.7371007955761833e-05,
      "loss": 0.9375,
      "step": 1205
    },
    {
      "epoch": 0.25941062594106257,
      "grad_norm": 0.0,
      "learning_rate": 1.736629772982961e-05,
      "loss": 0.9088,
      "step": 1206
    },
    {
      "epoch": 0.2596257259625726,
      "grad_norm": 0.0,
      "learning_rate": 1.7361583927971367e-05,
      "loss": 0.9633,
      "step": 1207
    },
    {
      "epoch": 0.2598408259840826,
      "grad_norm": 0.0,
      "learning_rate": 1.7356866552475397e-05,
      "loss": 0.9222,
      "step": 1208
    },
    {
      "epoch": 0.2600559260055926,
      "grad_norm": 0.0,
      "learning_rate": 1.7352145605631714e-05,
      "loss": 0.945,
      "step": 1209
    },
    {
      "epoch": 0.2602710260271026,
      "grad_norm": 0.0,
      "learning_rate": 1.734742108973208e-05,
      "loss": 0.899,
      "step": 1210
    },
    {
      "epoch": 0.2604861260486126,
      "grad_norm": 0.0,
      "learning_rate": 1.7342693007069974e-05,
      "loss": 0.928,
      "step": 1211
    },
    {
      "epoch": 0.2607012260701226,
      "grad_norm": 0.0,
      "learning_rate": 1.7337961359940633e-05,
      "loss": 0.9018,
      "step": 1212
    },
    {
      "epoch": 0.2609163260916326,
      "grad_norm": 0.0,
      "learning_rate": 1.7333226150640988e-05,
      "loss": 0.8354,
      "step": 1213
    },
    {
      "epoch": 0.2611314261131426,
      "grad_norm": 0.0,
      "learning_rate": 1.732848738146973e-05,
      "loss": 0.908,
      "step": 1214
    },
    {
      "epoch": 0.2613465261346526,
      "grad_norm": 0.0,
      "learning_rate": 1.7323745054727264e-05,
      "loss": 0.9702,
      "step": 1215
    },
    {
      "epoch": 0.26156162615616263,
      "grad_norm": 0.0,
      "learning_rate": 1.731899917271573e-05,
      "loss": 0.922,
      "step": 1216
    },
    {
      "epoch": 0.2617767261776726,
      "grad_norm": 0.0,
      "learning_rate": 1.7314249737738977e-05,
      "loss": 0.9257,
      "step": 1217
    },
    {
      "epoch": 0.2619918261991826,
      "grad_norm": 0.0,
      "learning_rate": 1.73094967521026e-05,
      "loss": 0.8946,
      "step": 1218
    },
    {
      "epoch": 0.26220692622069264,
      "grad_norm": 0.0,
      "learning_rate": 1.73047402181139e-05,
      "loss": 1.0352,
      "step": 1219
    },
    {
      "epoch": 0.2624220262422026,
      "grad_norm": 0.0,
      "learning_rate": 1.729998013808192e-05,
      "loss": 0.906,
      "step": 1220
    },
    {
      "epoch": 0.2626371262637126,
      "grad_norm": 0.0,
      "learning_rate": 1.7295216514317407e-05,
      "loss": 0.8958,
      "step": 1221
    },
    {
      "epoch": 0.26285222628522265,
      "grad_norm": 0.0,
      "learning_rate": 1.729044934913283e-05,
      "loss": 0.9234,
      "step": 1222
    },
    {
      "epoch": 0.26306732630673263,
      "grad_norm": 0.0,
      "learning_rate": 1.7285678644842384e-05,
      "loss": 0.9542,
      "step": 1223
    },
    {
      "epoch": 0.2632824263282426,
      "grad_norm": 0.0,
      "learning_rate": 1.728090440376198e-05,
      "loss": 0.9554,
      "step": 1224
    },
    {
      "epoch": 0.26349752634975265,
      "grad_norm": 0.0,
      "learning_rate": 1.727612662820925e-05,
      "loss": 0.9607,
      "step": 1225
    },
    {
      "epoch": 0.26371262637126264,
      "grad_norm": 0.0,
      "learning_rate": 1.7271345320503533e-05,
      "loss": 0.9878,
      "step": 1226
    },
    {
      "epoch": 0.2639277263927726,
      "grad_norm": 0.0,
      "learning_rate": 1.7266560482965883e-05,
      "loss": 0.9319,
      "step": 1227
    },
    {
      "epoch": 0.26414282641428266,
      "grad_norm": 0.0,
      "learning_rate": 1.7261772117919077e-05,
      "loss": 0.963,
      "step": 1228
    },
    {
      "epoch": 0.26435792643579265,
      "grad_norm": 0.0,
      "learning_rate": 1.7256980227687596e-05,
      "loss": 0.9772,
      "step": 1229
    },
    {
      "epoch": 0.26457302645730263,
      "grad_norm": 0.0,
      "learning_rate": 1.7252184814597633e-05,
      "loss": 0.9487,
      "step": 1230
    },
    {
      "epoch": 0.26478812647881267,
      "grad_norm": 0.0,
      "learning_rate": 1.7247385880977097e-05,
      "loss": 0.8952,
      "step": 1231
    },
    {
      "epoch": 0.26500322650032265,
      "grad_norm": 0.0,
      "learning_rate": 1.7242583429155602e-05,
      "loss": 0.9095,
      "step": 1232
    },
    {
      "epoch": 0.26521832652183264,
      "grad_norm": 0.0,
      "learning_rate": 1.7237777461464468e-05,
      "loss": 0.8984,
      "step": 1233
    },
    {
      "epoch": 0.2654334265433427,
      "grad_norm": 0.0,
      "learning_rate": 1.7232967980236724e-05,
      "loss": 0.9582,
      "step": 1234
    },
    {
      "epoch": 0.26564852656485266,
      "grad_norm": 0.0,
      "learning_rate": 1.722815498780711e-05,
      "loss": 0.8788,
      "step": 1235
    },
    {
      "epoch": 0.26586362658636264,
      "grad_norm": 0.0,
      "learning_rate": 1.7223338486512054e-05,
      "loss": 0.862,
      "step": 1236
    },
    {
      "epoch": 0.2660787266078727,
      "grad_norm": 0.0,
      "learning_rate": 1.721851847868971e-05,
      "loss": 0.9502,
      "step": 1237
    },
    {
      "epoch": 0.26629382662938267,
      "grad_norm": 0.0,
      "learning_rate": 1.7213694966679913e-05,
      "loss": 0.934,
      "step": 1238
    },
    {
      "epoch": 0.26650892665089265,
      "grad_norm": 0.0,
      "learning_rate": 1.7208867952824215e-05,
      "loss": 0.8429,
      "step": 1239
    },
    {
      "epoch": 0.2667240266724027,
      "grad_norm": 0.0,
      "learning_rate": 1.7204037439465858e-05,
      "loss": 0.9647,
      "step": 1240
    },
    {
      "epoch": 0.2669391266939127,
      "grad_norm": 0.0,
      "learning_rate": 1.7199203428949788e-05,
      "loss": 0.9493,
      "step": 1241
    },
    {
      "epoch": 0.26715422671542266,
      "grad_norm": 0.0,
      "learning_rate": 1.7194365923622646e-05,
      "loss": 0.9742,
      "step": 1242
    },
    {
      "epoch": 0.2673693267369327,
      "grad_norm": 0.0,
      "learning_rate": 1.7189524925832772e-05,
      "loss": 1.0073,
      "step": 1243
    },
    {
      "epoch": 0.2675844267584427,
      "grad_norm": 0.0,
      "learning_rate": 1.71846804379302e-05,
      "loss": 0.908,
      "step": 1244
    },
    {
      "epoch": 0.26779952677995267,
      "grad_norm": 0.0,
      "learning_rate": 1.7179832462266654e-05,
      "loss": 0.8983,
      "step": 1245
    },
    {
      "epoch": 0.2680146268014627,
      "grad_norm": 0.0,
      "learning_rate": 1.7174981001195562e-05,
      "loss": 0.8735,
      "step": 1246
    },
    {
      "epoch": 0.2682297268229727,
      "grad_norm": 0.0,
      "learning_rate": 1.717012605707203e-05,
      "loss": 0.9209,
      "step": 1247
    },
    {
      "epoch": 0.2684448268444827,
      "grad_norm": 0.0,
      "learning_rate": 1.7165267632252872e-05,
      "loss": 0.9847,
      "step": 1248
    },
    {
      "epoch": 0.26865992686599266,
      "grad_norm": 0.0,
      "learning_rate": 1.7160405729096572e-05,
      "loss": 0.9389,
      "step": 1249
    },
    {
      "epoch": 0.2688750268875027,
      "grad_norm": 0.0,
      "learning_rate": 1.715554034996332e-05,
      "loss": 0.9288,
      "step": 1250
    },
    {
      "epoch": 0.2690901269090127,
      "grad_norm": 0.0,
      "learning_rate": 1.7150671497214976e-05,
      "loss": 0.9833,
      "step": 1251
    },
    {
      "epoch": 0.26930522693052267,
      "grad_norm": 0.0,
      "learning_rate": 1.714579917321511e-05,
      "loss": 0.9838,
      "step": 1252
    },
    {
      "epoch": 0.2695203269520327,
      "grad_norm": 0.0,
      "learning_rate": 1.7140923380328947e-05,
      "loss": 0.9452,
      "step": 1253
    },
    {
      "epoch": 0.2697354269735427,
      "grad_norm": 0.0,
      "learning_rate": 1.7136044120923427e-05,
      "loss": 0.9357,
      "step": 1254
    },
    {
      "epoch": 0.2699505269950527,
      "grad_norm": 0.0,
      "learning_rate": 1.7131161397367146e-05,
      "loss": 0.8808,
      "step": 1255
    },
    {
      "epoch": 0.2701656270165627,
      "grad_norm": 0.0,
      "learning_rate": 1.7126275212030403e-05,
      "loss": 0.947,
      "step": 1256
    },
    {
      "epoch": 0.2703807270380727,
      "grad_norm": 0.0,
      "learning_rate": 1.7121385567285162e-05,
      "loss": 0.9193,
      "step": 1257
    },
    {
      "epoch": 0.2705958270595827,
      "grad_norm": 0.0,
      "learning_rate": 1.7116492465505072e-05,
      "loss": 0.9182,
      "step": 1258
    },
    {
      "epoch": 0.2708109270810927,
      "grad_norm": 0.0,
      "learning_rate": 1.7111595909065466e-05,
      "loss": 1.0001,
      "step": 1259
    },
    {
      "epoch": 0.2710260271026027,
      "grad_norm": 0.0,
      "learning_rate": 1.7106695900343344e-05,
      "loss": 0.8793,
      "step": 1260
    },
    {
      "epoch": 0.2712411271241127,
      "grad_norm": 0.0,
      "learning_rate": 1.7101792441717386e-05,
      "loss": 0.95,
      "step": 1261
    },
    {
      "epoch": 0.27145622714562273,
      "grad_norm": 0.0,
      "learning_rate": 1.709688553556795e-05,
      "loss": 0.8618,
      "step": 1262
    },
    {
      "epoch": 0.2716713271671327,
      "grad_norm": 0.0,
      "learning_rate": 1.7091975184277066e-05,
      "loss": 0.8944,
      "step": 1263
    },
    {
      "epoch": 0.2718864271886427,
      "grad_norm": 0.0,
      "learning_rate": 1.708706139022843e-05,
      "loss": 0.9969,
      "step": 1264
    },
    {
      "epoch": 0.27210152721015274,
      "grad_norm": 0.0,
      "learning_rate": 1.708214415580742e-05,
      "loss": 0.9293,
      "step": 1265
    },
    {
      "epoch": 0.2723166272316627,
      "grad_norm": 0.0,
      "learning_rate": 1.7077223483401072e-05,
      "loss": 0.9909,
      "step": 1266
    },
    {
      "epoch": 0.2725317272531727,
      "grad_norm": 0.0,
      "learning_rate": 1.7072299375398107e-05,
      "loss": 0.9401,
      "step": 1267
    },
    {
      "epoch": 0.27274682727468275,
      "grad_norm": 0.0,
      "learning_rate": 1.7067371834188895e-05,
      "loss": 0.8919,
      "step": 1268
    },
    {
      "epoch": 0.27296192729619273,
      "grad_norm": 0.0,
      "learning_rate": 1.7062440862165486e-05,
      "loss": 0.9286,
      "step": 1269
    },
    {
      "epoch": 0.2731770273177027,
      "grad_norm": 0.0,
      "learning_rate": 1.705750646172159e-05,
      "loss": 0.9373,
      "step": 1270
    },
    {
      "epoch": 0.27339212733921275,
      "grad_norm": 0.0,
      "learning_rate": 1.705256863525259e-05,
      "loss": 0.8596,
      "step": 1271
    },
    {
      "epoch": 0.27360722736072274,
      "grad_norm": 0.0,
      "learning_rate": 1.7047627385155513e-05,
      "loss": 1.0048,
      "step": 1272
    },
    {
      "epoch": 0.2738223273822327,
      "grad_norm": 0.0,
      "learning_rate": 1.7042682713829067e-05,
      "loss": 0.85,
      "step": 1273
    },
    {
      "epoch": 0.27403742740374276,
      "grad_norm": 0.0,
      "learning_rate": 1.7037734623673616e-05,
      "loss": 0.9022,
      "step": 1274
    },
    {
      "epoch": 0.27425252742525275,
      "grad_norm": 0.0,
      "learning_rate": 1.7032783117091172e-05,
      "loss": 0.8225,
      "step": 1275
    },
    {
      "epoch": 0.27446762744676273,
      "grad_norm": 0.0,
      "learning_rate": 1.7027828196485426e-05,
      "loss": 0.9309,
      "step": 1276
    },
    {
      "epoch": 0.27468272746827277,
      "grad_norm": 0.0,
      "learning_rate": 1.702286986426171e-05,
      "loss": 0.9522,
      "step": 1277
    },
    {
      "epoch": 0.27489782748978275,
      "grad_norm": 0.0,
      "learning_rate": 1.7017908122827015e-05,
      "loss": 0.9399,
      "step": 1278
    },
    {
      "epoch": 0.27511292751129274,
      "grad_norm": 0.0,
      "learning_rate": 1.7012942974589993e-05,
      "loss": 0.8938,
      "step": 1279
    },
    {
      "epoch": 0.2753280275328028,
      "grad_norm": 0.0,
      "learning_rate": 1.700797442196094e-05,
      "loss": 0.9334,
      "step": 1280
    },
    {
      "epoch": 0.27554312755431276,
      "grad_norm": 0.0,
      "learning_rate": 1.700300246735182e-05,
      "loss": 0.9104,
      "step": 1281
    },
    {
      "epoch": 0.27575822757582275,
      "grad_norm": 0.0,
      "learning_rate": 1.6998027113176232e-05,
      "loss": 0.9384,
      "step": 1282
    },
    {
      "epoch": 0.2759733275973328,
      "grad_norm": 0.0,
      "learning_rate": 1.699304836184944e-05,
      "loss": 0.9585,
      "step": 1283
    },
    {
      "epoch": 0.27618842761884277,
      "grad_norm": 0.0,
      "learning_rate": 1.6988066215788336e-05,
      "loss": 0.8913,
      "step": 1284
    },
    {
      "epoch": 0.27640352764035275,
      "grad_norm": 0.0,
      "learning_rate": 1.6983080677411488e-05,
      "loss": 0.9649,
      "step": 1285
    },
    {
      "epoch": 0.2766186276618628,
      "grad_norm": 0.0,
      "learning_rate": 1.6978091749139088e-05,
      "loss": 0.9119,
      "step": 1286
    },
    {
      "epoch": 0.2768337276833728,
      "grad_norm": 0.0,
      "learning_rate": 1.6973099433392985e-05,
      "loss": 0.9106,
      "step": 1287
    },
    {
      "epoch": 0.27704882770488276,
      "grad_norm": 0.0,
      "learning_rate": 1.696810373259667e-05,
      "loss": 0.9041,
      "step": 1288
    },
    {
      "epoch": 0.2772639277263928,
      "grad_norm": 0.0,
      "learning_rate": 1.696310464917527e-05,
      "loss": 0.9965,
      "step": 1289
    },
    {
      "epoch": 0.2774790277479028,
      "grad_norm": 0.0,
      "learning_rate": 1.6958102185555572e-05,
      "loss": 0.9375,
      "step": 1290
    },
    {
      "epoch": 0.27769412776941277,
      "grad_norm": 0.0,
      "learning_rate": 1.6953096344165988e-05,
      "loss": 0.9413,
      "step": 1291
    },
    {
      "epoch": 0.27790922779092275,
      "grad_norm": 0.0,
      "learning_rate": 1.6948087127436573e-05,
      "loss": 0.9131,
      "step": 1292
    },
    {
      "epoch": 0.2781243278124328,
      "grad_norm": 0.0,
      "learning_rate": 1.6943074537799015e-05,
      "loss": 0.9294,
      "step": 1293
    },
    {
      "epoch": 0.2783394278339428,
      "grad_norm": 0.0,
      "learning_rate": 1.693805857768666e-05,
      "loss": 0.8807,
      "step": 1294
    },
    {
      "epoch": 0.27855452785545276,
      "grad_norm": 0.0,
      "learning_rate": 1.693303924953447e-05,
      "loss": 0.9086,
      "step": 1295
    },
    {
      "epoch": 0.2787696278769628,
      "grad_norm": 0.0,
      "learning_rate": 1.6928016555779047e-05,
      "loss": 0.9306,
      "step": 1296
    },
    {
      "epoch": 0.2789847278984728,
      "grad_norm": 0.0,
      "learning_rate": 1.6922990498858624e-05,
      "loss": 0.9291,
      "step": 1297
    },
    {
      "epoch": 0.27919982791998277,
      "grad_norm": 0.0,
      "learning_rate": 1.691796108121308e-05,
      "loss": 0.9442,
      "step": 1298
    },
    {
      "epoch": 0.2794149279414928,
      "grad_norm": 0.0,
      "learning_rate": 1.6912928305283914e-05,
      "loss": 0.8944,
      "step": 1299
    },
    {
      "epoch": 0.2796300279630028,
      "grad_norm": 0.0,
      "learning_rate": 1.6907892173514255e-05,
      "loss": 0.948,
      "step": 1300
    },
    {
      "epoch": 0.2798451279845128,
      "grad_norm": 0.0,
      "learning_rate": 1.6902852688348866e-05,
      "loss": 0.9667,
      "step": 1301
    },
    {
      "epoch": 0.2800602280060228,
      "grad_norm": 0.0,
      "learning_rate": 1.689780985223413e-05,
      "loss": 0.9982,
      "step": 1302
    },
    {
      "epoch": 0.2802753280275328,
      "grad_norm": 0.0,
      "learning_rate": 1.6892763667618068e-05,
      "loss": 0.8957,
      "step": 1303
    },
    {
      "epoch": 0.2804904280490428,
      "grad_norm": 0.0,
      "learning_rate": 1.6887714136950322e-05,
      "loss": 0.9446,
      "step": 1304
    },
    {
      "epoch": 0.2807055280705528,
      "grad_norm": 0.0,
      "learning_rate": 1.6882661262682153e-05,
      "loss": 0.9011,
      "step": 1305
    },
    {
      "epoch": 0.2809206280920628,
      "grad_norm": 0.0,
      "learning_rate": 1.6877605047266447e-05,
      "loss": 0.9229,
      "step": 1306
    },
    {
      "epoch": 0.2811357281135728,
      "grad_norm": 0.0,
      "learning_rate": 1.6872545493157718e-05,
      "loss": 0.9351,
      "step": 1307
    },
    {
      "epoch": 0.28135082813508283,
      "grad_norm": 0.0,
      "learning_rate": 1.68674826028121e-05,
      "loss": 0.9743,
      "step": 1308
    },
    {
      "epoch": 0.2815659281565928,
      "grad_norm": 0.0,
      "learning_rate": 1.686241637868734e-05,
      "loss": 0.8686,
      "step": 1309
    },
    {
      "epoch": 0.2817810281781028,
      "grad_norm": 0.0,
      "learning_rate": 1.68573468232428e-05,
      "loss": 0.9688,
      "step": 1310
    },
    {
      "epoch": 0.28199612819961284,
      "grad_norm": 0.0,
      "learning_rate": 1.6852273938939475e-05,
      "loss": 1.0057,
      "step": 1311
    },
    {
      "epoch": 0.2822112282211228,
      "grad_norm": 0.0,
      "learning_rate": 1.6847197728239967e-05,
      "loss": 0.9126,
      "step": 1312
    },
    {
      "epoch": 0.2824263282426328,
      "grad_norm": 0.0,
      "learning_rate": 1.684211819360849e-05,
      "loss": 0.981,
      "step": 1313
    },
    {
      "epoch": 0.28264142826414285,
      "grad_norm": 0.0,
      "learning_rate": 1.6837035337510873e-05,
      "loss": 0.8926,
      "step": 1314
    },
    {
      "epoch": 0.28285652828565283,
      "grad_norm": 0.0,
      "learning_rate": 1.6831949162414557e-05,
      "loss": 0.8593,
      "step": 1315
    },
    {
      "epoch": 0.2830716283071628,
      "grad_norm": 0.0,
      "learning_rate": 1.6826859670788604e-05,
      "loss": 0.857,
      "step": 1316
    },
    {
      "epoch": 0.28328672832867285,
      "grad_norm": 0.0,
      "learning_rate": 1.682176686510367e-05,
      "loss": 0.9312,
      "step": 1317
    },
    {
      "epoch": 0.28350182835018284,
      "grad_norm": 0.0,
      "learning_rate": 1.6816670747832032e-05,
      "loss": 0.9467,
      "step": 1318
    },
    {
      "epoch": 0.2837169283716928,
      "grad_norm": 0.0,
      "learning_rate": 1.6811571321447568e-05,
      "loss": 0.9489,
      "step": 1319
    },
    {
      "epoch": 0.28393202839320286,
      "grad_norm": 0.0,
      "learning_rate": 1.680646858842576e-05,
      "loss": 0.9092,
      "step": 1320
    },
    {
      "epoch": 0.28414712841471285,
      "grad_norm": 0.0,
      "learning_rate": 1.6801362551243713e-05,
      "loss": 0.8818,
      "step": 1321
    },
    {
      "epoch": 0.28436222843622283,
      "grad_norm": 0.0,
      "learning_rate": 1.679625321238011e-05,
      "loss": 0.9222,
      "step": 1322
    },
    {
      "epoch": 0.28457732845773287,
      "grad_norm": 0.0,
      "learning_rate": 1.6791140574315256e-05,
      "loss": 0.912,
      "step": 1323
    },
    {
      "epoch": 0.28479242847924285,
      "grad_norm": 0.0,
      "learning_rate": 1.6786024639531048e-05,
      "loss": 0.9131,
      "step": 1324
    },
    {
      "epoch": 0.28500752850075284,
      "grad_norm": 0.0,
      "learning_rate": 1.6780905410510992e-05,
      "loss": 0.8294,
      "step": 1325
    },
    {
      "epoch": 0.2852226285222629,
      "grad_norm": 0.0,
      "learning_rate": 1.677578288974018e-05,
      "loss": 0.9187,
      "step": 1326
    },
    {
      "epoch": 0.28543772854377286,
      "grad_norm": 0.0,
      "learning_rate": 1.6770657079705313e-05,
      "loss": 0.8773,
      "step": 1327
    },
    {
      "epoch": 0.28565282856528285,
      "grad_norm": 0.0,
      "learning_rate": 1.676552798289469e-05,
      "loss": 0.9423,
      "step": 1328
    },
    {
      "epoch": 0.2858679285867929,
      "grad_norm": 0.0,
      "learning_rate": 1.6760395601798192e-05,
      "loss": 0.9262,
      "step": 1329
    },
    {
      "epoch": 0.28608302860830287,
      "grad_norm": 0.0,
      "learning_rate": 1.675525993890731e-05,
      "loss": 0.8916,
      "step": 1330
    },
    {
      "epoch": 0.28629812862981285,
      "grad_norm": 0.0,
      "learning_rate": 1.6750120996715123e-05,
      "loss": 0.9801,
      "step": 1331
    },
    {
      "epoch": 0.28651322865132284,
      "grad_norm": 0.0,
      "learning_rate": 1.6744978777716294e-05,
      "loss": 0.8652,
      "step": 1332
    },
    {
      "epoch": 0.2867283286728329,
      "grad_norm": 0.0,
      "learning_rate": 1.6739833284407083e-05,
      "loss": 0.9483,
      "step": 1333
    },
    {
      "epoch": 0.28694342869434286,
      "grad_norm": 0.0,
      "learning_rate": 1.6734684519285346e-05,
      "loss": 0.8792,
      "step": 1334
    },
    {
      "epoch": 0.28715852871585285,
      "grad_norm": 0.0,
      "learning_rate": 1.6729532484850516e-05,
      "loss": 0.9063,
      "step": 1335
    },
    {
      "epoch": 0.2873736287373629,
      "grad_norm": 0.0,
      "learning_rate": 1.672437718360362e-05,
      "loss": 0.8741,
      "step": 1336
    },
    {
      "epoch": 0.28758872875887287,
      "grad_norm": 0.0,
      "learning_rate": 1.6719218618047263e-05,
      "loss": 0.9216,
      "step": 1337
    },
    {
      "epoch": 0.28780382878038285,
      "grad_norm": 0.0,
      "learning_rate": 1.6714056790685652e-05,
      "loss": 0.9989,
      "step": 1338
    },
    {
      "epoch": 0.2880189288018929,
      "grad_norm": 0.0,
      "learning_rate": 1.6708891704024557e-05,
      "loss": 0.9175,
      "step": 1339
    },
    {
      "epoch": 0.2882340288234029,
      "grad_norm": 0.0,
      "learning_rate": 1.6703723360571338e-05,
      "loss": 0.9802,
      "step": 1340
    },
    {
      "epoch": 0.28844912884491286,
      "grad_norm": 0.0,
      "learning_rate": 1.6698551762834944e-05,
      "loss": 0.919,
      "step": 1341
    },
    {
      "epoch": 0.2886642288664229,
      "grad_norm": 0.0,
      "learning_rate": 1.6693376913325892e-05,
      "loss": 0.8843,
      "step": 1342
    },
    {
      "epoch": 0.2888793288879329,
      "grad_norm": 0.0,
      "learning_rate": 1.6688198814556283e-05,
      "loss": 0.9027,
      "step": 1343
    },
    {
      "epoch": 0.28909442890944287,
      "grad_norm": 0.0,
      "learning_rate": 1.6683017469039794e-05,
      "loss": 0.9616,
      "step": 1344
    },
    {
      "epoch": 0.2893095289309529,
      "grad_norm": 0.0,
      "learning_rate": 1.667783287929168e-05,
      "loss": 0.9386,
      "step": 1345
    },
    {
      "epoch": 0.2895246289524629,
      "grad_norm": 0.0,
      "learning_rate": 1.667264504782877e-05,
      "loss": 0.9134,
      "step": 1346
    },
    {
      "epoch": 0.2897397289739729,
      "grad_norm": 0.0,
      "learning_rate": 1.6667453977169467e-05,
      "loss": 0.895,
      "step": 1347
    },
    {
      "epoch": 0.2899548289954829,
      "grad_norm": 0.0,
      "learning_rate": 1.6662259669833742e-05,
      "loss": 0.9209,
      "step": 1348
    },
    {
      "epoch": 0.2901699290169929,
      "grad_norm": 0.0,
      "learning_rate": 1.6657062128343143e-05,
      "loss": 0.8852,
      "step": 1349
    },
    {
      "epoch": 0.2903850290385029,
      "grad_norm": 0.0,
      "learning_rate": 1.665186135522079e-05,
      "loss": 0.8701,
      "step": 1350
    },
    {
      "epoch": 0.2906001290600129,
      "grad_norm": 0.0,
      "learning_rate": 1.6646657352991362e-05,
      "loss": 0.9703,
      "step": 1351
    },
    {
      "epoch": 0.2908152290815229,
      "grad_norm": 0.0,
      "learning_rate": 1.6641450124181115e-05,
      "loss": 0.8965,
      "step": 1352
    },
    {
      "epoch": 0.2910303291030329,
      "grad_norm": 0.0,
      "learning_rate": 1.6636239671317865e-05,
      "loss": 0.9075,
      "step": 1353
    },
    {
      "epoch": 0.29124542912454293,
      "grad_norm": 0.0,
      "learning_rate": 1.6631025996930998e-05,
      "loss": 0.9386,
      "step": 1354
    },
    {
      "epoch": 0.2914605291460529,
      "grad_norm": 0.0,
      "learning_rate": 1.6625809103551462e-05,
      "loss": 1.0201,
      "step": 1355
    },
    {
      "epoch": 0.2916756291675629,
      "grad_norm": 0.0,
      "learning_rate": 1.6620588993711764e-05,
      "loss": 0.9316,
      "step": 1356
    },
    {
      "epoch": 0.29189072918907294,
      "grad_norm": 0.0,
      "learning_rate": 1.661536566994598e-05,
      "loss": 0.9296,
      "step": 1357
    },
    {
      "epoch": 0.2921058292105829,
      "grad_norm": 0.0,
      "learning_rate": 1.661013913478974e-05,
      "loss": 0.9103,
      "step": 1358
    },
    {
      "epoch": 0.2923209292320929,
      "grad_norm": 0.0,
      "learning_rate": 1.660490939078023e-05,
      "loss": 0.8853,
      "step": 1359
    },
    {
      "epoch": 0.29253602925360295,
      "grad_norm": 0.0,
      "learning_rate": 1.6599676440456208e-05,
      "loss": 0.8812,
      "step": 1360
    },
    {
      "epoch": 0.29275112927511293,
      "grad_norm": 0.0,
      "learning_rate": 1.659444028635797e-05,
      "loss": 0.9172,
      "step": 1361
    },
    {
      "epoch": 0.2929662292966229,
      "grad_norm": 0.0,
      "learning_rate": 1.6589200931027384e-05,
      "loss": 0.9163,
      "step": 1362
    },
    {
      "epoch": 0.29318132931813295,
      "grad_norm": 0.0,
      "learning_rate": 1.6583958377007857e-05,
      "loss": 0.9041,
      "step": 1363
    },
    {
      "epoch": 0.29339642933964294,
      "grad_norm": 0.0,
      "learning_rate": 1.6578712626844365e-05,
      "loss": 0.8541,
      "step": 1364
    },
    {
      "epoch": 0.2936115293611529,
      "grad_norm": 0.0,
      "learning_rate": 1.6573463683083417e-05,
      "loss": 0.8958,
      "step": 1365
    },
    {
      "epoch": 0.29382662938266296,
      "grad_norm": 0.0,
      "learning_rate": 1.656821154827309e-05,
      "loss": 0.9218,
      "step": 1366
    },
    {
      "epoch": 0.29404172940417295,
      "grad_norm": 0.0,
      "learning_rate": 1.6562956224962996e-05,
      "loss": 0.8825,
      "step": 1367
    },
    {
      "epoch": 0.29425682942568293,
      "grad_norm": 0.0,
      "learning_rate": 1.65576977157043e-05,
      "loss": 0.9556,
      "step": 1368
    },
    {
      "epoch": 0.29447192944719297,
      "grad_norm": 0.0,
      "learning_rate": 1.6552436023049724e-05,
      "loss": 0.9329,
      "step": 1369
    },
    {
      "epoch": 0.29468702946870295,
      "grad_norm": 0.0,
      "learning_rate": 1.6547171149553517e-05,
      "loss": 0.9513,
      "step": 1370
    },
    {
      "epoch": 0.29490212949021294,
      "grad_norm": 0.0,
      "learning_rate": 1.6541903097771485e-05,
      "loss": 0.9487,
      "step": 1371
    },
    {
      "epoch": 0.295117229511723,
      "grad_norm": 0.0,
      "learning_rate": 1.6536631870260967e-05,
      "loss": 0.9583,
      "step": 1372
    },
    {
      "epoch": 0.29533232953323296,
      "grad_norm": 0.0,
      "learning_rate": 1.653135746958086e-05,
      "loss": 0.8895,
      "step": 1373
    },
    {
      "epoch": 0.29554742955474295,
      "grad_norm": 0.0,
      "learning_rate": 1.652607989829158e-05,
      "loss": 0.9609,
      "step": 1374
    },
    {
      "epoch": 0.29576252957625293,
      "grad_norm": 0.0,
      "learning_rate": 1.65207991589551e-05,
      "loss": 0.8839,
      "step": 1375
    },
    {
      "epoch": 0.29597762959776297,
      "grad_norm": 0.0,
      "learning_rate": 1.6515515254134918e-05,
      "loss": 0.9636,
      "step": 1376
    },
    {
      "epoch": 0.29619272961927295,
      "grad_norm": 0.0,
      "learning_rate": 1.6510228186396085e-05,
      "loss": 0.9409,
      "step": 1377
    },
    {
      "epoch": 0.29640782964078294,
      "grad_norm": 0.0,
      "learning_rate": 1.6504937958305168e-05,
      "loss": 0.9473,
      "step": 1378
    },
    {
      "epoch": 0.296622929662293,
      "grad_norm": 0.0,
      "learning_rate": 1.6499644572430277e-05,
      "loss": 0.8595,
      "step": 1379
    },
    {
      "epoch": 0.29683802968380296,
      "grad_norm": 0.0,
      "learning_rate": 1.649434803134106e-05,
      "loss": 0.9355,
      "step": 1380
    },
    {
      "epoch": 0.29705312970531295,
      "grad_norm": 0.0,
      "learning_rate": 1.648904833760869e-05,
      "loss": 0.8648,
      "step": 1381
    },
    {
      "epoch": 0.297268229726823,
      "grad_norm": 0.0,
      "learning_rate": 1.648374549380587e-05,
      "loss": 0.9624,
      "step": 1382
    },
    {
      "epoch": 0.29748332974833297,
      "grad_norm": 0.0,
      "learning_rate": 1.6478439502506834e-05,
      "loss": 0.8119,
      "step": 1383
    },
    {
      "epoch": 0.29769842976984295,
      "grad_norm": 0.0,
      "learning_rate": 1.647313036628735e-05,
      "loss": 0.877,
      "step": 1384
    },
    {
      "epoch": 0.297913529791353,
      "grad_norm": 0.0,
      "learning_rate": 1.64678180877247e-05,
      "loss": 0.9126,
      "step": 1385
    },
    {
      "epoch": 0.298128629812863,
      "grad_norm": 0.0,
      "learning_rate": 1.6462502669397703e-05,
      "loss": 0.9155,
      "step": 1386
    },
    {
      "epoch": 0.29834372983437296,
      "grad_norm": 0.0,
      "learning_rate": 1.64571841138867e-05,
      "loss": 0.9081,
      "step": 1387
    },
    {
      "epoch": 0.298558829855883,
      "grad_norm": 0.0,
      "learning_rate": 1.6451862423773545e-05,
      "loss": 0.8428,
      "step": 1388
    },
    {
      "epoch": 0.298773929877393,
      "grad_norm": 0.0,
      "learning_rate": 1.6446537601641628e-05,
      "loss": 0.9321,
      "step": 1389
    },
    {
      "epoch": 0.29898902989890297,
      "grad_norm": 0.0,
      "learning_rate": 1.6441209650075848e-05,
      "loss": 0.971,
      "step": 1390
    },
    {
      "epoch": 0.299204129920413,
      "grad_norm": 0.0,
      "learning_rate": 1.643587857166263e-05,
      "loss": 0.9492,
      "step": 1391
    },
    {
      "epoch": 0.299419229941923,
      "grad_norm": 0.0,
      "learning_rate": 1.6430544368989915e-05,
      "loss": 0.9742,
      "step": 1392
    },
    {
      "epoch": 0.299634329963433,
      "grad_norm": 0.0,
      "learning_rate": 1.6425207044647163e-05,
      "loss": 0.9534,
      "step": 1393
    },
    {
      "epoch": 0.299849429984943,
      "grad_norm": 0.0,
      "learning_rate": 1.641986660122534e-05,
      "loss": 0.9522,
      "step": 1394
    },
    {
      "epoch": 0.300064530006453,
      "grad_norm": 0.0,
      "learning_rate": 1.641452304131694e-05,
      "loss": 0.954,
      "step": 1395
    },
    {
      "epoch": 0.300279630027963,
      "grad_norm": 0.0,
      "learning_rate": 1.6409176367515957e-05,
      "loss": 0.9263,
      "step": 1396
    },
    {
      "epoch": 0.300494730049473,
      "grad_norm": 0.0,
      "learning_rate": 1.6403826582417908e-05,
      "loss": 0.916,
      "step": 1397
    },
    {
      "epoch": 0.300709830070983,
      "grad_norm": 0.0,
      "learning_rate": 1.6398473688619807e-05,
      "loss": 0.8708,
      "step": 1398
    },
    {
      "epoch": 0.300924930092493,
      "grad_norm": 0.0,
      "learning_rate": 1.6393117688720193e-05,
      "loss": 0.908,
      "step": 1399
    },
    {
      "epoch": 0.30114003011400303,
      "grad_norm": 0.0,
      "learning_rate": 1.63877585853191e-05,
      "loss": 0.9462,
      "step": 1400
    },
    {
      "epoch": 0.301355130135513,
      "grad_norm": 0.0,
      "learning_rate": 1.6382396381018076e-05,
      "loss": 0.8776,
      "step": 1401
    },
    {
      "epoch": 0.301570230157023,
      "grad_norm": 0.0,
      "learning_rate": 1.6377031078420168e-05,
      "loss": 0.8947,
      "step": 1402
    },
    {
      "epoch": 0.30178533017853304,
      "grad_norm": 0.0,
      "learning_rate": 1.6371662680129934e-05,
      "loss": 0.9212,
      "step": 1403
    },
    {
      "epoch": 0.302000430200043,
      "grad_norm": 0.0,
      "learning_rate": 1.636629118875343e-05,
      "loss": 0.9199,
      "step": 1404
    },
    {
      "epoch": 0.302215530221553,
      "grad_norm": 0.0,
      "learning_rate": 1.6360916606898217e-05,
      "loss": 0.8549,
      "step": 1405
    },
    {
      "epoch": 0.30243063024306305,
      "grad_norm": 0.0,
      "learning_rate": 1.6355538937173353e-05,
      "loss": 0.9042,
      "step": 1406
    },
    {
      "epoch": 0.30264573026457303,
      "grad_norm": 0.0,
      "learning_rate": 1.6350158182189394e-05,
      "loss": 0.898,
      "step": 1407
    },
    {
      "epoch": 0.302860830286083,
      "grad_norm": 0.0,
      "learning_rate": 1.6344774344558402e-05,
      "loss": 0.9481,
      "step": 1408
    },
    {
      "epoch": 0.30307593030759306,
      "grad_norm": 0.0,
      "learning_rate": 1.633938742689392e-05,
      "loss": 0.8741,
      "step": 1409
    },
    {
      "epoch": 0.30329103032910304,
      "grad_norm": 0.0,
      "learning_rate": 1.6333997431811005e-05,
      "loss": 0.9448,
      "step": 1410
    },
    {
      "epoch": 0.303506130350613,
      "grad_norm": 0.0,
      "learning_rate": 1.632860436192619e-05,
      "loss": 0.929,
      "step": 1411
    },
    {
      "epoch": 0.30372123037212306,
      "grad_norm": 0.0,
      "learning_rate": 1.6323208219857517e-05,
      "loss": 0.9088,
      "step": 1412
    },
    {
      "epoch": 0.30393633039363305,
      "grad_norm": 0.0,
      "learning_rate": 1.6317809008224506e-05,
      "loss": 0.9112,
      "step": 1413
    },
    {
      "epoch": 0.30415143041514303,
      "grad_norm": 0.0,
      "learning_rate": 1.6312406729648177e-05,
      "loss": 0.8788,
      "step": 1414
    },
    {
      "epoch": 0.30436653043665307,
      "grad_norm": 0.0,
      "learning_rate": 1.630700138675103e-05,
      "loss": 0.9504,
      "step": 1415
    },
    {
      "epoch": 0.30458163045816306,
      "grad_norm": 0.0,
      "learning_rate": 1.6301592982157055e-05,
      "loss": 0.8992,
      "step": 1416
    },
    {
      "epoch": 0.30479673047967304,
      "grad_norm": 0.0,
      "learning_rate": 1.629618151849174e-05,
      "loss": 0.8909,
      "step": 1417
    },
    {
      "epoch": 0.305011830501183,
      "grad_norm": 0.0,
      "learning_rate": 1.629076699838204e-05,
      "loss": 0.9028,
      "step": 1418
    },
    {
      "epoch": 0.30522693052269306,
      "grad_norm": 0.0,
      "learning_rate": 1.6285349424456408e-05,
      "loss": 0.894,
      "step": 1419
    },
    {
      "epoch": 0.30544203054420305,
      "grad_norm": 0.0,
      "learning_rate": 1.6279928799344765e-05,
      "loss": 0.9512,
      "step": 1420
    },
    {
      "epoch": 0.30565713056571303,
      "grad_norm": 0.0,
      "learning_rate": 1.627450512567853e-05,
      "loss": 0.892,
      "step": 1421
    },
    {
      "epoch": 0.30587223058722307,
      "grad_norm": 0.0,
      "learning_rate": 1.626907840609059e-05,
      "loss": 1.0079,
      "step": 1422
    },
    {
      "epoch": 0.30608733060873305,
      "grad_norm": 0.0,
      "learning_rate": 1.6263648643215315e-05,
      "loss": 0.8812,
      "step": 1423
    },
    {
      "epoch": 0.30630243063024304,
      "grad_norm": 0.0,
      "learning_rate": 1.625821583968855e-05,
      "loss": 0.8621,
      "step": 1424
    },
    {
      "epoch": 0.3065175306517531,
      "grad_norm": 0.0,
      "learning_rate": 1.625277999814762e-05,
      "loss": 0.8752,
      "step": 1425
    },
    {
      "epoch": 0.30673263067326306,
      "grad_norm": 0.0,
      "learning_rate": 1.6247341121231318e-05,
      "loss": 0.9189,
      "step": 1426
    },
    {
      "epoch": 0.30694773069477305,
      "grad_norm": 0.0,
      "learning_rate": 1.624189921157992e-05,
      "loss": 0.8413,
      "step": 1427
    },
    {
      "epoch": 0.3071628307162831,
      "grad_norm": 0.0,
      "learning_rate": 1.6236454271835167e-05,
      "loss": 0.8928,
      "step": 1428
    },
    {
      "epoch": 0.30737793073779307,
      "grad_norm": 0.0,
      "learning_rate": 1.6231006304640272e-05,
      "loss": 0.9025,
      "step": 1429
    },
    {
      "epoch": 0.30759303075930305,
      "grad_norm": 0.0,
      "learning_rate": 1.622555531263992e-05,
      "loss": 0.969,
      "step": 1430
    },
    {
      "epoch": 0.3078081307808131,
      "grad_norm": 0.0,
      "learning_rate": 1.6220101298480265e-05,
      "loss": 0.9554,
      "step": 1431
    },
    {
      "epoch": 0.3080232308023231,
      "grad_norm": 0.0,
      "learning_rate": 1.621464426480892e-05,
      "loss": 0.8342,
      "step": 1432
    },
    {
      "epoch": 0.30823833082383306,
      "grad_norm": 0.0,
      "learning_rate": 1.620918421427498e-05,
      "loss": 0.9493,
      "step": 1433
    },
    {
      "epoch": 0.3084534308453431,
      "grad_norm": 0.0,
      "learning_rate": 1.6203721149528982e-05,
      "loss": 0.9256,
      "step": 1434
    },
    {
      "epoch": 0.3086685308668531,
      "grad_norm": 0.0,
      "learning_rate": 1.619825507322295e-05,
      "loss": 0.8972,
      "step": 1435
    },
    {
      "epoch": 0.30888363088836307,
      "grad_norm": 0.0,
      "learning_rate": 1.6192785988010354e-05,
      "loss": 0.9969,
      "step": 1436
    },
    {
      "epoch": 0.3090987309098731,
      "grad_norm": 0.0,
      "learning_rate": 1.6187313896546125e-05,
      "loss": 1.0032,
      "step": 1437
    },
    {
      "epoch": 0.3093138309313831,
      "grad_norm": 0.0,
      "learning_rate": 1.6181838801486663e-05,
      "loss": 0.9535,
      "step": 1438
    },
    {
      "epoch": 0.3095289309528931,
      "grad_norm": 0.0,
      "learning_rate": 1.6176360705489826e-05,
      "loss": 0.9267,
      "step": 1439
    },
    {
      "epoch": 0.3097440309744031,
      "grad_norm": 0.0,
      "learning_rate": 1.6170879611214913e-05,
      "loss": 0.8917,
      "step": 1440
    },
    {
      "epoch": 0.3099591309959131,
      "grad_norm": 0.0,
      "learning_rate": 1.616539552132269e-05,
      "loss": 0.9056,
      "step": 1441
    },
    {
      "epoch": 0.3101742310174231,
      "grad_norm": 0.0,
      "learning_rate": 1.6159908438475383e-05,
      "loss": 0.9469,
      "step": 1442
    },
    {
      "epoch": 0.3103893310389331,
      "grad_norm": 0.0,
      "learning_rate": 1.6154418365336662e-05,
      "loss": 0.8631,
      "step": 1443
    },
    {
      "epoch": 0.3106044310604431,
      "grad_norm": 0.0,
      "learning_rate": 1.6148925304571647e-05,
      "loss": 1.0097,
      "step": 1444
    },
    {
      "epoch": 0.3108195310819531,
      "grad_norm": 0.0,
      "learning_rate": 1.6143429258846917e-05,
      "loss": 0.9314,
      "step": 1445
    },
    {
      "epoch": 0.31103463110346313,
      "grad_norm": 0.0,
      "learning_rate": 1.613793023083049e-05,
      "loss": 0.9324,
      "step": 1446
    },
    {
      "epoch": 0.3112497311249731,
      "grad_norm": 0.0,
      "learning_rate": 1.613242822319184e-05,
      "loss": 0.9114,
      "step": 1447
    },
    {
      "epoch": 0.3114648311464831,
      "grad_norm": 0.0,
      "learning_rate": 1.612692323860189e-05,
      "loss": 0.8678,
      "step": 1448
    },
    {
      "epoch": 0.31167993116799314,
      "grad_norm": 0.0,
      "learning_rate": 1.6121415279732992e-05,
      "loss": 0.9051,
      "step": 1449
    },
    {
      "epoch": 0.3118950311895031,
      "grad_norm": 0.0,
      "learning_rate": 1.611590434925896e-05,
      "loss": 0.9497,
      "step": 1450
    },
    {
      "epoch": 0.3121101312110131,
      "grad_norm": 0.0,
      "learning_rate": 1.6110390449855034e-05,
      "loss": 0.9334,
      "step": 1451
    },
    {
      "epoch": 0.31232523123252315,
      "grad_norm": 0.0,
      "learning_rate": 1.6104873584197914e-05,
      "loss": 0.9045,
      "step": 1452
    },
    {
      "epoch": 0.31254033125403313,
      "grad_norm": 0.0,
      "learning_rate": 1.6099353754965723e-05,
      "loss": 0.8545,
      "step": 1453
    },
    {
      "epoch": 0.3127554312755431,
      "grad_norm": 0.0,
      "learning_rate": 1.6093830964838036e-05,
      "loss": 0.9174,
      "step": 1454
    },
    {
      "epoch": 0.31297053129705316,
      "grad_norm": 0.0,
      "learning_rate": 1.6088305216495855e-05,
      "loss": 0.9474,
      "step": 1455
    },
    {
      "epoch": 0.31318563131856314,
      "grad_norm": 0.0,
      "learning_rate": 1.608277651262162e-05,
      "loss": 0.9851,
      "step": 1456
    },
    {
      "epoch": 0.3134007313400731,
      "grad_norm": 0.0,
      "learning_rate": 1.6077244855899213e-05,
      "loss": 0.8797,
      "step": 1457
    },
    {
      "epoch": 0.31361583136158316,
      "grad_norm": 0.0,
      "learning_rate": 1.6071710249013946e-05,
      "loss": 0.9422,
      "step": 1458
    },
    {
      "epoch": 0.31383093138309315,
      "grad_norm": 0.0,
      "learning_rate": 1.6066172694652552e-05,
      "loss": 0.9564,
      "step": 1459
    },
    {
      "epoch": 0.31404603140460313,
      "grad_norm": 0.0,
      "learning_rate": 1.606063219550321e-05,
      "loss": 0.8899,
      "step": 1460
    },
    {
      "epoch": 0.3142611314261131,
      "grad_norm": 0.0,
      "learning_rate": 1.6055088754255526e-05,
      "loss": 0.9071,
      "step": 1461
    },
    {
      "epoch": 0.31447623144762316,
      "grad_norm": 0.0,
      "learning_rate": 1.6049542373600525e-05,
      "loss": 0.9252,
      "step": 1462
    },
    {
      "epoch": 0.31469133146913314,
      "grad_norm": 0.0,
      "learning_rate": 1.604399305623067e-05,
      "loss": 0.9716,
      "step": 1463
    },
    {
      "epoch": 0.3149064314906431,
      "grad_norm": 0.0,
      "learning_rate": 1.6038440804839842e-05,
      "loss": 0.8837,
      "step": 1464
    },
    {
      "epoch": 0.31512153151215316,
      "grad_norm": 0.0,
      "learning_rate": 1.6032885622123347e-05,
      "loss": 0.8724,
      "step": 1465
    },
    {
      "epoch": 0.31533663153366315,
      "grad_norm": 0.0,
      "learning_rate": 1.602732751077792e-05,
      "loss": 0.9012,
      "step": 1466
    },
    {
      "epoch": 0.31555173155517313,
      "grad_norm": 0.0,
      "learning_rate": 1.6021766473501707e-05,
      "loss": 0.9034,
      "step": 1467
    },
    {
      "epoch": 0.31576683157668317,
      "grad_norm": 0.0,
      "learning_rate": 1.601620251299429e-05,
      "loss": 0.9144,
      "step": 1468
    },
    {
      "epoch": 0.31598193159819316,
      "grad_norm": 0.0,
      "learning_rate": 1.6010635631956653e-05,
      "loss": 0.9067,
      "step": 1469
    },
    {
      "epoch": 0.31619703161970314,
      "grad_norm": 0.0,
      "learning_rate": 1.600506583309121e-05,
      "loss": 0.9066,
      "step": 1470
    },
    {
      "epoch": 0.3164121316412132,
      "grad_norm": 0.0,
      "learning_rate": 1.5999493119101786e-05,
      "loss": 0.9147,
      "step": 1471
    },
    {
      "epoch": 0.31662723166272316,
      "grad_norm": 0.0,
      "learning_rate": 1.5993917492693623e-05,
      "loss": 0.8579,
      "step": 1472
    },
    {
      "epoch": 0.31684233168423315,
      "grad_norm": 0.0,
      "learning_rate": 1.5988338956573377e-05,
      "loss": 0.9639,
      "step": 1473
    },
    {
      "epoch": 0.3170574317057432,
      "grad_norm": 0.0,
      "learning_rate": 1.5982757513449114e-05,
      "loss": 0.8842,
      "step": 1474
    },
    {
      "epoch": 0.31727253172725317,
      "grad_norm": 0.0,
      "learning_rate": 1.5977173166030312e-05,
      "loss": 0.9394,
      "step": 1475
    },
    {
      "epoch": 0.31748763174876315,
      "grad_norm": 0.0,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 0.9168,
      "step": 1476
    },
    {
      "epoch": 0.3177027317702732,
      "grad_norm": 0.0,
      "learning_rate": 1.596599576915406e-05,
      "loss": 0.8517,
      "step": 1477
    },
    {
      "epoch": 0.3179178317917832,
      "grad_norm": 0.0,
      "learning_rate": 1.596040272512261e-05,
      "loss": 0.9097,
      "step": 1478
    },
    {
      "epoch": 0.31813293181329316,
      "grad_norm": 0.0,
      "learning_rate": 1.5954806787648624e-05,
      "loss": 0.9095,
      "step": 1479
    },
    {
      "epoch": 0.3183480318348032,
      "grad_norm": 0.0,
      "learning_rate": 1.5949207959448614e-05,
      "loss": 0.841,
      "step": 1480
    },
    {
      "epoch": 0.3185631318563132,
      "grad_norm": 0.0,
      "learning_rate": 1.5943606243240504e-05,
      "loss": 0.8788,
      "step": 1481
    },
    {
      "epoch": 0.31877823187782317,
      "grad_norm": 0.0,
      "learning_rate": 1.5938001641743608e-05,
      "loss": 0.9123,
      "step": 1482
    },
    {
      "epoch": 0.3189933318993332,
      "grad_norm": 0.0,
      "learning_rate": 1.5932394157678646e-05,
      "loss": 0.9195,
      "step": 1483
    },
    {
      "epoch": 0.3192084319208432,
      "grad_norm": 0.0,
      "learning_rate": 1.592678379376775e-05,
      "loss": 0.8798,
      "step": 1484
    },
    {
      "epoch": 0.3194235319423532,
      "grad_norm": 0.0,
      "learning_rate": 1.5921170552734423e-05,
      "loss": 0.9812,
      "step": 1485
    },
    {
      "epoch": 0.3196386319638632,
      "grad_norm": 0.0,
      "learning_rate": 1.591555443730359e-05,
      "loss": 0.943,
      "step": 1486
    },
    {
      "epoch": 0.3198537319853732,
      "grad_norm": 0.0,
      "learning_rate": 1.590993545020156e-05,
      "loss": 0.9324,
      "step": 1487
    },
    {
      "epoch": 0.3200688320068832,
      "grad_norm": 0.0,
      "learning_rate": 1.590431359415603e-05,
      "loss": 0.9272,
      "step": 1488
    },
    {
      "epoch": 0.3202839320283932,
      "grad_norm": 0.0,
      "learning_rate": 1.589868887189611e-05,
      "loss": 0.8967,
      "step": 1489
    },
    {
      "epoch": 0.3204990320499032,
      "grad_norm": 0.0,
      "learning_rate": 1.5893061286152275e-05,
      "loss": 0.876,
      "step": 1490
    },
    {
      "epoch": 0.3207141320714132,
      "grad_norm": 0.0,
      "learning_rate": 1.588743083965641e-05,
      "loss": 0.8617,
      "step": 1491
    },
    {
      "epoch": 0.32092923209292323,
      "grad_norm": 0.0,
      "learning_rate": 1.5881797535141785e-05,
      "loss": 0.9914,
      "step": 1492
    },
    {
      "epoch": 0.3211443321144332,
      "grad_norm": 0.0,
      "learning_rate": 1.587616137534305e-05,
      "loss": 0.93,
      "step": 1493
    },
    {
      "epoch": 0.3213594321359432,
      "grad_norm": 0.0,
      "learning_rate": 1.5870522362996244e-05,
      "loss": 0.9264,
      "step": 1494
    },
    {
      "epoch": 0.32157453215745324,
      "grad_norm": 0.0,
      "learning_rate": 1.58648805008388e-05,
      "loss": 0.9505,
      "step": 1495
    },
    {
      "epoch": 0.3217896321789632,
      "grad_norm": 0.0,
      "learning_rate": 1.585923579160952e-05,
      "loss": 0.9704,
      "step": 1496
    },
    {
      "epoch": 0.3220047322004732,
      "grad_norm": 0.0,
      "learning_rate": 1.58535882380486e-05,
      "loss": 0.9238,
      "step": 1497
    },
    {
      "epoch": 0.32221983222198325,
      "grad_norm": 0.0,
      "learning_rate": 1.5847937842897606e-05,
      "loss": 0.9232,
      "step": 1498
    },
    {
      "epoch": 0.32243493224349323,
      "grad_norm": 0.0,
      "learning_rate": 1.584228460889949e-05,
      "loss": 0.8184,
      "step": 1499
    },
    {
      "epoch": 0.3226500322650032,
      "grad_norm": 0.0,
      "learning_rate": 1.583662853879859e-05,
      "loss": 0.9114,
      "step": 1500
    },
    {
      "epoch": 0.3228651322865132,
      "grad_norm": 0.0,
      "learning_rate": 1.58309696353406e-05,
      "loss": 0.8585,
      "step": 1501
    },
    {
      "epoch": 0.32308023230802324,
      "grad_norm": 0.0,
      "learning_rate": 1.5825307901272612e-05,
      "loss": 0.8938,
      "step": 1502
    },
    {
      "epoch": 0.3232953323295332,
      "grad_norm": 0.0,
      "learning_rate": 1.5819643339343077e-05,
      "loss": 0.9272,
      "step": 1503
    },
    {
      "epoch": 0.3235104323510432,
      "grad_norm": 0.0,
      "learning_rate": 1.5813975952301815e-05,
      "loss": 0.9421,
      "step": 1504
    },
    {
      "epoch": 0.32372553237255325,
      "grad_norm": 0.0,
      "learning_rate": 1.5808305742900044e-05,
      "loss": 0.9263,
      "step": 1505
    },
    {
      "epoch": 0.32394063239406323,
      "grad_norm": 0.0,
      "learning_rate": 1.580263271389032e-05,
      "loss": 0.9315,
      "step": 1506
    },
    {
      "epoch": 0.3241557324155732,
      "grad_norm": 0.0,
      "learning_rate": 1.5796956868026582e-05,
      "loss": 0.9453,
      "step": 1507
    },
    {
      "epoch": 0.32437083243708326,
      "grad_norm": 0.0,
      "learning_rate": 1.5791278208064142e-05,
      "loss": 0.9371,
      "step": 1508
    },
    {
      "epoch": 0.32458593245859324,
      "grad_norm": 0.0,
      "learning_rate": 1.578559673675967e-05,
      "loss": 0.9713,
      "step": 1509
    },
    {
      "epoch": 0.3248010324801032,
      "grad_norm": 0.0,
      "learning_rate": 1.5779912456871202e-05,
      "loss": 0.9865,
      "step": 1510
    },
    {
      "epoch": 0.32501613250161326,
      "grad_norm": 0.0,
      "learning_rate": 1.5774225371158142e-05,
      "loss": 0.8617,
      "step": 1511
    },
    {
      "epoch": 0.32523123252312325,
      "grad_norm": 0.0,
      "learning_rate": 1.5768535482381247e-05,
      "loss": 0.9286,
      "step": 1512
    },
    {
      "epoch": 0.32544633254463323,
      "grad_norm": 0.0,
      "learning_rate": 1.5762842793302638e-05,
      "loss": 0.9245,
      "step": 1513
    },
    {
      "epoch": 0.32566143256614327,
      "grad_norm": 0.0,
      "learning_rate": 1.5757147306685806e-05,
      "loss": 0.9326,
      "step": 1514
    },
    {
      "epoch": 0.32587653258765326,
      "grad_norm": 0.0,
      "learning_rate": 1.575144902529559e-05,
      "loss": 0.9703,
      "step": 1515
    },
    {
      "epoch": 0.32609163260916324,
      "grad_norm": 0.0,
      "learning_rate": 1.5745747951898183e-05,
      "loss": 0.8982,
      "step": 1516
    },
    {
      "epoch": 0.3263067326306733,
      "grad_norm": 0.0,
      "learning_rate": 1.5740044089261137e-05,
      "loss": 0.9359,
      "step": 1517
    },
    {
      "epoch": 0.32652183265218326,
      "grad_norm": 0.0,
      "learning_rate": 1.5734337440153364e-05,
      "loss": 0.9328,
      "step": 1518
    },
    {
      "epoch": 0.32673693267369325,
      "grad_norm": 0.0,
      "learning_rate": 1.5728628007345118e-05,
      "loss": 0.9153,
      "step": 1519
    },
    {
      "epoch": 0.3269520326952033,
      "grad_norm": 0.0,
      "learning_rate": 1.5722915793608017e-05,
      "loss": 0.8988,
      "step": 1520
    },
    {
      "epoch": 0.32716713271671327,
      "grad_norm": 0.0,
      "learning_rate": 1.571720080171501e-05,
      "loss": 0.9051,
      "step": 1521
    },
    {
      "epoch": 0.32738223273822326,
      "grad_norm": 0.0,
      "learning_rate": 1.5711483034440414e-05,
      "loss": 0.9722,
      "step": 1522
    },
    {
      "epoch": 0.3275973327597333,
      "grad_norm": 0.0,
      "learning_rate": 1.570576249455989e-05,
      "loss": 0.8677,
      "step": 1523
    },
    {
      "epoch": 0.3278124327812433,
      "grad_norm": 0.0,
      "learning_rate": 1.570003918485043e-05,
      "loss": 0.9314,
      "step": 1524
    },
    {
      "epoch": 0.32802753280275326,
      "grad_norm": 0.0,
      "learning_rate": 1.5694313108090388e-05,
      "loss": 0.9354,
      "step": 1525
    },
    {
      "epoch": 0.3282426328242633,
      "grad_norm": 0.0,
      "learning_rate": 1.5688584267059453e-05,
      "loss": 0.8223,
      "step": 1526
    },
    {
      "epoch": 0.3284577328457733,
      "grad_norm": 0.0,
      "learning_rate": 1.5682852664538656e-05,
      "loss": 0.947,
      "step": 1527
    },
    {
      "epoch": 0.32867283286728327,
      "grad_norm": 0.0,
      "learning_rate": 1.567711830331037e-05,
      "loss": 0.9461,
      "step": 1528
    },
    {
      "epoch": 0.3288879328887933,
      "grad_norm": 0.0,
      "learning_rate": 1.567138118615831e-05,
      "loss": 0.8489,
      "step": 1529
    },
    {
      "epoch": 0.3291030329103033,
      "grad_norm": 0.0,
      "learning_rate": 1.5665641315867524e-05,
      "loss": 0.9581,
      "step": 1530
    },
    {
      "epoch": 0.3293181329318133,
      "grad_norm": 0.0,
      "learning_rate": 1.5659898695224396e-05,
      "loss": 0.909,
      "step": 1531
    },
    {
      "epoch": 0.3295332329533233,
      "grad_norm": 0.0,
      "learning_rate": 1.565415332701665e-05,
      "loss": 0.8668,
      "step": 1532
    },
    {
      "epoch": 0.3297483329748333,
      "grad_norm": 0.0,
      "learning_rate": 1.564840521403334e-05,
      "loss": 0.8681,
      "step": 1533
    },
    {
      "epoch": 0.3299634329963433,
      "grad_norm": 0.0,
      "learning_rate": 1.5642654359064855e-05,
      "loss": 0.8771,
      "step": 1534
    },
    {
      "epoch": 0.3301785330178533,
      "grad_norm": 0.0,
      "learning_rate": 1.5636900764902914e-05,
      "loss": 0.9462,
      "step": 1535
    },
    {
      "epoch": 0.3303936330393633,
      "grad_norm": 0.0,
      "learning_rate": 1.5631144434340566e-05,
      "loss": 0.9157,
      "step": 1536
    },
    {
      "epoch": 0.3306087330608733,
      "grad_norm": 0.0,
      "learning_rate": 1.5625385370172185e-05,
      "loss": 0.8628,
      "step": 1537
    },
    {
      "epoch": 0.33082383308238333,
      "grad_norm": 0.0,
      "learning_rate": 1.561962357519348e-05,
      "loss": 0.9435,
      "step": 1538
    },
    {
      "epoch": 0.3310389331038933,
      "grad_norm": 0.0,
      "learning_rate": 1.561385905220147e-05,
      "loss": 0.8705,
      "step": 1539
    },
    {
      "epoch": 0.3312540331254033,
      "grad_norm": 0.0,
      "learning_rate": 1.560809180399452e-05,
      "loss": 0.8969,
      "step": 1540
    },
    {
      "epoch": 0.33146913314691334,
      "grad_norm": 0.0,
      "learning_rate": 1.5602321833372304e-05,
      "loss": 0.8876,
      "step": 1541
    },
    {
      "epoch": 0.3316842331684233,
      "grad_norm": 0.0,
      "learning_rate": 1.5596549143135818e-05,
      "loss": 0.9106,
      "step": 1542
    },
    {
      "epoch": 0.3318993331899333,
      "grad_norm": 0.0,
      "learning_rate": 1.5590773736087382e-05,
      "loss": 0.9114,
      "step": 1543
    },
    {
      "epoch": 0.3321144332114433,
      "grad_norm": 0.0,
      "learning_rate": 1.5584995615030634e-05,
      "loss": 0.9097,
      "step": 1544
    },
    {
      "epoch": 0.33232953323295333,
      "grad_norm": 0.0,
      "learning_rate": 1.5579214782770526e-05,
      "loss": 0.9352,
      "step": 1545
    },
    {
      "epoch": 0.3325446332544633,
      "grad_norm": 0.0,
      "learning_rate": 1.5573431242113333e-05,
      "loss": 0.8571,
      "step": 1546
    },
    {
      "epoch": 0.3327597332759733,
      "grad_norm": 0.0,
      "learning_rate": 1.556764499586664e-05,
      "loss": 0.8615,
      "step": 1547
    },
    {
      "epoch": 0.33297483329748334,
      "grad_norm": 0.0,
      "learning_rate": 1.5561856046839347e-05,
      "loss": 0.8344,
      "step": 1548
    },
    {
      "epoch": 0.3331899333189933,
      "grad_norm": 0.0,
      "learning_rate": 1.555606439784166e-05,
      "loss": 0.9101,
      "step": 1549
    },
    {
      "epoch": 0.3334050333405033,
      "grad_norm": 0.0,
      "learning_rate": 1.555027005168511e-05,
      "loss": 0.9512,
      "step": 1550
    },
    {
      "epoch": 0.33362013336201335,
      "grad_norm": 0.0,
      "learning_rate": 1.5544473011182518e-05,
      "loss": 0.8925,
      "step": 1551
    },
    {
      "epoch": 0.33383523338352333,
      "grad_norm": 0.0,
      "learning_rate": 1.5538673279148033e-05,
      "loss": 0.8816,
      "step": 1552
    },
    {
      "epoch": 0.3340503334050333,
      "grad_norm": 0.0,
      "learning_rate": 1.5532870858397093e-05,
      "loss": 0.8818,
      "step": 1553
    },
    {
      "epoch": 0.33426543342654336,
      "grad_norm": 0.0,
      "learning_rate": 1.5527065751746454e-05,
      "loss": 0.9292,
      "step": 1554
    },
    {
      "epoch": 0.33448053344805334,
      "grad_norm": 0.0,
      "learning_rate": 1.5521257962014172e-05,
      "loss": 0.8579,
      "step": 1555
    },
    {
      "epoch": 0.3346956334695633,
      "grad_norm": 0.0,
      "learning_rate": 1.5515447492019598e-05,
      "loss": 0.9385,
      "step": 1556
    },
    {
      "epoch": 0.33491073349107336,
      "grad_norm": 0.0,
      "learning_rate": 1.5509634344583395e-05,
      "loss": 0.8887,
      "step": 1557
    },
    {
      "epoch": 0.33512583351258335,
      "grad_norm": 0.0,
      "learning_rate": 1.5503818522527525e-05,
      "loss": 0.8728,
      "step": 1558
    },
    {
      "epoch": 0.33534093353409333,
      "grad_norm": 0.0,
      "learning_rate": 1.549800002867524e-05,
      "loss": 0.8752,
      "step": 1559
    },
    {
      "epoch": 0.3355560335556034,
      "grad_norm": 0.0,
      "learning_rate": 1.5492178865851092e-05,
      "loss": 0.9537,
      "step": 1560
    },
    {
      "epoch": 0.33577113357711336,
      "grad_norm": 0.0,
      "learning_rate": 1.548635503688093e-05,
      "loss": 0.9951,
      "step": 1561
    },
    {
      "epoch": 0.33598623359862334,
      "grad_norm": 0.0,
      "learning_rate": 1.5480528544591907e-05,
      "loss": 0.897,
      "step": 1562
    },
    {
      "epoch": 0.3362013336201334,
      "grad_norm": 0.0,
      "learning_rate": 1.5474699391812447e-05,
      "loss": 0.8981,
      "step": 1563
    },
    {
      "epoch": 0.33641643364164336,
      "grad_norm": 0.0,
      "learning_rate": 1.5468867581372288e-05,
      "loss": 0.9459,
      "step": 1564
    },
    {
      "epoch": 0.33663153366315335,
      "grad_norm": 0.0,
      "learning_rate": 1.546303311610244e-05,
      "loss": 0.9658,
      "step": 1565
    },
    {
      "epoch": 0.3368466336846634,
      "grad_norm": 0.0,
      "learning_rate": 1.545719599883522e-05,
      "loss": 0.897,
      "step": 1566
    },
    {
      "epoch": 0.33706173370617337,
      "grad_norm": 0.0,
      "learning_rate": 1.545135623240421e-05,
      "loss": 0.9418,
      "step": 1567
    },
    {
      "epoch": 0.33727683372768336,
      "grad_norm": 0.0,
      "learning_rate": 1.54455138196443e-05,
      "loss": 0.8786,
      "step": 1568
    },
    {
      "epoch": 0.3374919337491934,
      "grad_norm": 0.0,
      "learning_rate": 1.5439668763391652e-05,
      "loss": 0.9096,
      "step": 1569
    },
    {
      "epoch": 0.3377070337707034,
      "grad_norm": 0.0,
      "learning_rate": 1.5433821066483713e-05,
      "loss": 0.9589,
      "step": 1570
    },
    {
      "epoch": 0.33792213379221336,
      "grad_norm": 0.0,
      "learning_rate": 1.5427970731759214e-05,
      "loss": 0.9007,
      "step": 1571
    },
    {
      "epoch": 0.3381372338137234,
      "grad_norm": 0.0,
      "learning_rate": 1.5422117762058167e-05,
      "loss": 0.915,
      "step": 1572
    },
    {
      "epoch": 0.3383523338352334,
      "grad_norm": 0.0,
      "learning_rate": 1.541626216022186e-05,
      "loss": 0.8077,
      "step": 1573
    },
    {
      "epoch": 0.33856743385674337,
      "grad_norm": 0.0,
      "learning_rate": 1.5410403929092858e-05,
      "loss": 0.8856,
      "step": 1574
    },
    {
      "epoch": 0.3387825338782534,
      "grad_norm": 0.0,
      "learning_rate": 1.5404543071515008e-05,
      "loss": 0.891,
      "step": 1575
    },
    {
      "epoch": 0.3389976338997634,
      "grad_norm": 0.0,
      "learning_rate": 1.5398679590333434e-05,
      "loss": 0.9325,
      "step": 1576
    },
    {
      "epoch": 0.3392127339212734,
      "grad_norm": 0.0,
      "learning_rate": 1.539281348839452e-05,
      "loss": 0.9244,
      "step": 1577
    },
    {
      "epoch": 0.3394278339427834,
      "grad_norm": 0.0,
      "learning_rate": 1.5386944768545933e-05,
      "loss": 0.9226,
      "step": 1578
    },
    {
      "epoch": 0.3396429339642934,
      "grad_norm": 0.0,
      "learning_rate": 1.5381073433636614e-05,
      "loss": 0.853,
      "step": 1579
    },
    {
      "epoch": 0.3398580339858034,
      "grad_norm": 0.0,
      "learning_rate": 1.537519948651676e-05,
      "loss": 0.9198,
      "step": 1580
    },
    {
      "epoch": 0.3400731340073134,
      "grad_norm": 0.0,
      "learning_rate": 1.5369322930037852e-05,
      "loss": 0.8991,
      "step": 1581
    },
    {
      "epoch": 0.3402882340288234,
      "grad_norm": 0.0,
      "learning_rate": 1.5363443767052624e-05,
      "loss": 0.8874,
      "step": 1582
    },
    {
      "epoch": 0.3405033340503334,
      "grad_norm": 0.0,
      "learning_rate": 1.5357562000415086e-05,
      "loss": 0.8475,
      "step": 1583
    },
    {
      "epoch": 0.34071843407184343,
      "grad_norm": 0.0,
      "learning_rate": 1.5351677632980503e-05,
      "loss": 0.8716,
      "step": 1584
    },
    {
      "epoch": 0.3409335340933534,
      "grad_norm": 0.0,
      "learning_rate": 1.5345790667605403e-05,
      "loss": 0.8565,
      "step": 1585
    },
    {
      "epoch": 0.3411486341148634,
      "grad_norm": 0.0,
      "learning_rate": 1.533990110714759e-05,
      "loss": 0.9228,
      "step": 1586
    },
    {
      "epoch": 0.3413637341363734,
      "grad_norm": 0.0,
      "learning_rate": 1.5334008954466108e-05,
      "loss": 0.8969,
      "step": 1587
    },
    {
      "epoch": 0.3415788341578834,
      "grad_norm": 0.0,
      "learning_rate": 1.532811421242127e-05,
      "loss": 0.9425,
      "step": 1588
    },
    {
      "epoch": 0.3417939341793934,
      "grad_norm": 0.0,
      "learning_rate": 1.5322216883874643e-05,
      "loss": 0.9148,
      "step": 1589
    },
    {
      "epoch": 0.3420090342009034,
      "grad_norm": 0.0,
      "learning_rate": 1.5316316971689053e-05,
      "loss": 0.8639,
      "step": 1590
    },
    {
      "epoch": 0.34222413422241343,
      "grad_norm": 0.0,
      "learning_rate": 1.5310414478728578e-05,
      "loss": 0.8555,
      "step": 1591
    },
    {
      "epoch": 0.3424392342439234,
      "grad_norm": 0.0,
      "learning_rate": 1.5304509407858545e-05,
      "loss": 0.9333,
      "step": 1592
    },
    {
      "epoch": 0.3426543342654334,
      "grad_norm": 0.0,
      "learning_rate": 1.529860176194554e-05,
      "loss": 0.841,
      "step": 1593
    },
    {
      "epoch": 0.34286943428694344,
      "grad_norm": 0.0,
      "learning_rate": 1.529269154385739e-05,
      "loss": 0.9344,
      "step": 1594
    },
    {
      "epoch": 0.3430845343084534,
      "grad_norm": 0.0,
      "learning_rate": 1.5286778756463184e-05,
      "loss": 0.8864,
      "step": 1595
    },
    {
      "epoch": 0.3432996343299634,
      "grad_norm": 0.0,
      "learning_rate": 1.5280863402633242e-05,
      "loss": 0.908,
      "step": 1596
    },
    {
      "epoch": 0.34351473435147345,
      "grad_norm": 0.0,
      "learning_rate": 1.5274945485239143e-05,
      "loss": 0.8776,
      "step": 1597
    },
    {
      "epoch": 0.34372983437298343,
      "grad_norm": 0.0,
      "learning_rate": 1.52690250071537e-05,
      "loss": 0.8394,
      "step": 1598
    },
    {
      "epoch": 0.3439449343944934,
      "grad_norm": 0.0,
      "learning_rate": 1.526310197125098e-05,
      "loss": 0.9653,
      "step": 1599
    },
    {
      "epoch": 0.34416003441600346,
      "grad_norm": 0.0,
      "learning_rate": 1.5257176380406277e-05,
      "loss": 0.9606,
      "step": 1600
    },
    {
      "epoch": 0.34437513443751344,
      "grad_norm": 0.0,
      "learning_rate": 1.5251248237496148e-05,
      "loss": 0.9177,
      "step": 1601
    },
    {
      "epoch": 0.3445902344590234,
      "grad_norm": 0.0,
      "learning_rate": 1.5245317545398363e-05,
      "loss": 0.9433,
      "step": 1602
    },
    {
      "epoch": 0.34480533448053347,
      "grad_norm": 0.0,
      "learning_rate": 1.5239384306991948e-05,
      "loss": 0.8683,
      "step": 1603
    },
    {
      "epoch": 0.34502043450204345,
      "grad_norm": 0.0,
      "learning_rate": 1.5233448525157161e-05,
      "loss": 0.9103,
      "step": 1604
    },
    {
      "epoch": 0.34523553452355343,
      "grad_norm": 0.0,
      "learning_rate": 1.5227510202775486e-05,
      "loss": 0.8916,
      "step": 1605
    },
    {
      "epoch": 0.3454506345450635,
      "grad_norm": 0.0,
      "learning_rate": 1.5221569342729648e-05,
      "loss": 0.9314,
      "step": 1606
    },
    {
      "epoch": 0.34566573456657346,
      "grad_norm": 0.0,
      "learning_rate": 1.5215625947903604e-05,
      "loss": 0.88,
      "step": 1607
    },
    {
      "epoch": 0.34588083458808344,
      "grad_norm": 0.0,
      "learning_rate": 1.5209680021182545e-05,
      "loss": 0.8724,
      "step": 1608
    },
    {
      "epoch": 0.3460959346095935,
      "grad_norm": 0.0,
      "learning_rate": 1.5203731565452878e-05,
      "loss": 0.8724,
      "step": 1609
    },
    {
      "epoch": 0.34631103463110346,
      "grad_norm": 0.0,
      "learning_rate": 1.5197780583602249e-05,
      "loss": 0.9571,
      "step": 1610
    },
    {
      "epoch": 0.34652613465261345,
      "grad_norm": 0.0,
      "learning_rate": 1.519182707851953e-05,
      "loss": 0.9214,
      "step": 1611
    },
    {
      "epoch": 0.3467412346741235,
      "grad_norm": 0.0,
      "learning_rate": 1.5185871053094814e-05,
      "loss": 0.844,
      "step": 1612
    },
    {
      "epoch": 0.3469563346956335,
      "grad_norm": 0.0,
      "learning_rate": 1.5179912510219413e-05,
      "loss": 0.931,
      "step": 1613
    },
    {
      "epoch": 0.34717143471714346,
      "grad_norm": 0.0,
      "learning_rate": 1.5173951452785873e-05,
      "loss": 0.906,
      "step": 1614
    },
    {
      "epoch": 0.3473865347386535,
      "grad_norm": 0.0,
      "learning_rate": 1.5167987883687953e-05,
      "loss": 0.8774,
      "step": 1615
    },
    {
      "epoch": 0.3476016347601635,
      "grad_norm": 0.0,
      "learning_rate": 1.5162021805820633e-05,
      "loss": 0.8961,
      "step": 1616
    },
    {
      "epoch": 0.34781673478167346,
      "grad_norm": 0.0,
      "learning_rate": 1.515605322208011e-05,
      "loss": 0.9146,
      "step": 1617
    },
    {
      "epoch": 0.3480318348031835,
      "grad_norm": 0.0,
      "learning_rate": 1.5150082135363797e-05,
      "loss": 0.8896,
      "step": 1618
    },
    {
      "epoch": 0.3482469348246935,
      "grad_norm": 0.0,
      "learning_rate": 1.5144108548570322e-05,
      "loss": 0.8743,
      "step": 1619
    },
    {
      "epoch": 0.34846203484620347,
      "grad_norm": 0.0,
      "learning_rate": 1.513813246459953e-05,
      "loss": 0.9181,
      "step": 1620
    },
    {
      "epoch": 0.3486771348677135,
      "grad_norm": 0.0,
      "learning_rate": 1.5132153886352478e-05,
      "loss": 0.913,
      "step": 1621
    },
    {
      "epoch": 0.3488922348892235,
      "grad_norm": 0.0,
      "learning_rate": 1.5126172816731424e-05,
      "loss": 0.91,
      "step": 1622
    },
    {
      "epoch": 0.3491073349107335,
      "grad_norm": 0.0,
      "learning_rate": 1.512018925863985e-05,
      "loss": 0.9378,
      "step": 1623
    },
    {
      "epoch": 0.3493224349322435,
      "grad_norm": 0.0,
      "learning_rate": 1.5114203214982439e-05,
      "loss": 0.9625,
      "step": 1624
    },
    {
      "epoch": 0.3495375349537535,
      "grad_norm": 0.0,
      "learning_rate": 1.510821468866508e-05,
      "loss": 0.976,
      "step": 1625
    },
    {
      "epoch": 0.3497526349752635,
      "grad_norm": 0.0,
      "learning_rate": 1.5102223682594867e-05,
      "loss": 0.9279,
      "step": 1626
    },
    {
      "epoch": 0.3499677349967735,
      "grad_norm": 0.0,
      "learning_rate": 1.5096230199680096e-05,
      "loss": 0.8795,
      "step": 1627
    },
    {
      "epoch": 0.3501828350182835,
      "grad_norm": 0.0,
      "learning_rate": 1.5090234242830271e-05,
      "loss": 0.9363,
      "step": 1628
    },
    {
      "epoch": 0.3503979350397935,
      "grad_norm": 0.0,
      "learning_rate": 1.5084235814956097e-05,
      "loss": 0.9092,
      "step": 1629
    },
    {
      "epoch": 0.3506130350613035,
      "grad_norm": 0.0,
      "learning_rate": 1.507823491896947e-05,
      "loss": 0.838,
      "step": 1630
    },
    {
      "epoch": 0.3508281350828135,
      "grad_norm": 0.0,
      "learning_rate": 1.5072231557783492e-05,
      "loss": 0.9305,
      "step": 1631
    },
    {
      "epoch": 0.3510432351043235,
      "grad_norm": 0.0,
      "learning_rate": 1.5066225734312462e-05,
      "loss": 0.8745,
      "step": 1632
    },
    {
      "epoch": 0.3512583351258335,
      "grad_norm": 0.0,
      "learning_rate": 1.506021745147187e-05,
      "loss": 0.8888,
      "step": 1633
    },
    {
      "epoch": 0.3514734351473435,
      "grad_norm": 0.0,
      "learning_rate": 1.50542067121784e-05,
      "loss": 0.9076,
      "step": 1634
    },
    {
      "epoch": 0.3516885351688535,
      "grad_norm": 0.0,
      "learning_rate": 1.5048193519349929e-05,
      "loss": 0.8917,
      "step": 1635
    },
    {
      "epoch": 0.3519036351903635,
      "grad_norm": 0.0,
      "learning_rate": 1.5042177875905532e-05,
      "loss": 0.8241,
      "step": 1636
    },
    {
      "epoch": 0.35211873521187353,
      "grad_norm": 0.0,
      "learning_rate": 1.5036159784765464e-05,
      "loss": 0.9461,
      "step": 1637
    },
    {
      "epoch": 0.3523338352333835,
      "grad_norm": 0.0,
      "learning_rate": 1.5030139248851172e-05,
      "loss": 0.9496,
      "step": 1638
    },
    {
      "epoch": 0.3525489352548935,
      "grad_norm": 0.0,
      "learning_rate": 1.5024116271085291e-05,
      "loss": 0.8511,
      "step": 1639
    },
    {
      "epoch": 0.35276403527640354,
      "grad_norm": 0.0,
      "learning_rate": 1.5018090854391638e-05,
      "loss": 0.9623,
      "step": 1640
    },
    {
      "epoch": 0.3529791352979135,
      "grad_norm": 0.0,
      "learning_rate": 1.5012063001695218e-05,
      "loss": 0.8241,
      "step": 1641
    },
    {
      "epoch": 0.3531942353194235,
      "grad_norm": 0.0,
      "learning_rate": 1.5006032715922216e-05,
      "loss": 0.9055,
      "step": 1642
    },
    {
      "epoch": 0.35340933534093355,
      "grad_norm": 0.0,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.9323,
      "step": 1643
    },
    {
      "epoch": 0.35362443536244353,
      "grad_norm": 0.0,
      "learning_rate": 1.4993964856857117e-05,
      "loss": 0.9202,
      "step": 1644
    },
    {
      "epoch": 0.3538395353839535,
      "grad_norm": 0.0,
      "learning_rate": 1.4987927289423289e-05,
      "loss": 0.9508,
      "step": 1645
    },
    {
      "epoch": 0.35405463540546356,
      "grad_norm": 0.0,
      "learning_rate": 1.498188730062942e-05,
      "loss": 0.9508,
      "step": 1646
    },
    {
      "epoch": 0.35426973542697354,
      "grad_norm": 0.0,
      "learning_rate": 1.497584489340759e-05,
      "loss": 0.881,
      "step": 1647
    },
    {
      "epoch": 0.3544848354484835,
      "grad_norm": 0.0,
      "learning_rate": 1.4969800070691042e-05,
      "loss": 0.8666,
      "step": 1648
    },
    {
      "epoch": 0.35469993546999357,
      "grad_norm": 0.0,
      "learning_rate": 1.4963752835414204e-05,
      "loss": 0.9159,
      "step": 1649
    },
    {
      "epoch": 0.35491503549150355,
      "grad_norm": 0.0,
      "learning_rate": 1.4957703190512674e-05,
      "loss": 0.9165,
      "step": 1650
    },
    {
      "epoch": 0.35513013551301353,
      "grad_norm": 0.0,
      "learning_rate": 1.495165113892321e-05,
      "loss": 0.9315,
      "step": 1651
    },
    {
      "epoch": 0.3553452355345236,
      "grad_norm": 0.0,
      "learning_rate": 1.4945596683583754e-05,
      "loss": 0.8489,
      "step": 1652
    },
    {
      "epoch": 0.35556033555603356,
      "grad_norm": 0.0,
      "learning_rate": 1.4939539827433396e-05,
      "loss": 0.9258,
      "step": 1653
    },
    {
      "epoch": 0.35577543557754354,
      "grad_norm": 0.0,
      "learning_rate": 1.4933480573412412e-05,
      "loss": 0.9549,
      "step": 1654
    },
    {
      "epoch": 0.3559905355990536,
      "grad_norm": 0.0,
      "learning_rate": 1.492741892446222e-05,
      "loss": 0.9642,
      "step": 1655
    },
    {
      "epoch": 0.35620563562056357,
      "grad_norm": 0.0,
      "learning_rate": 1.4921354883525419e-05,
      "loss": 0.9176,
      "step": 1656
    },
    {
      "epoch": 0.35642073564207355,
      "grad_norm": 0.0,
      "learning_rate": 1.4915288453545758e-05,
      "loss": 0.8901,
      "step": 1657
    },
    {
      "epoch": 0.3566358356635836,
      "grad_norm": 0.0,
      "learning_rate": 1.4909219637468153e-05,
      "loss": 0.9048,
      "step": 1658
    },
    {
      "epoch": 0.3568509356850936,
      "grad_norm": 0.0,
      "learning_rate": 1.4903148438238676e-05,
      "loss": 0.9159,
      "step": 1659
    },
    {
      "epoch": 0.35706603570660356,
      "grad_norm": 0.0,
      "learning_rate": 1.4897074858804553e-05,
      "loss": 0.9546,
      "step": 1660
    },
    {
      "epoch": 0.3572811357281136,
      "grad_norm": 0.0,
      "learning_rate": 1.4890998902114166e-05,
      "loss": 0.9624,
      "step": 1661
    },
    {
      "epoch": 0.3574962357496236,
      "grad_norm": 0.0,
      "learning_rate": 1.4884920571117057e-05,
      "loss": 0.9097,
      "step": 1662
    },
    {
      "epoch": 0.35771133577113357,
      "grad_norm": 0.0,
      "learning_rate": 1.4878839868763912e-05,
      "loss": 0.9043,
      "step": 1663
    },
    {
      "epoch": 0.3579264357926436,
      "grad_norm": 0.0,
      "learning_rate": 1.4872756798006576e-05,
      "loss": 0.8842,
      "step": 1664
    },
    {
      "epoch": 0.3581415358141536,
      "grad_norm": 0.0,
      "learning_rate": 1.4866671361798035e-05,
      "loss": 0.9411,
      "step": 1665
    },
    {
      "epoch": 0.3583566358356636,
      "grad_norm": 0.0,
      "learning_rate": 1.486058356309243e-05,
      "loss": 0.8752,
      "step": 1666
    },
    {
      "epoch": 0.3585717358571736,
      "grad_norm": 0.0,
      "learning_rate": 1.4854493404845051e-05,
      "loss": 0.8982,
      "step": 1667
    },
    {
      "epoch": 0.3587868358786836,
      "grad_norm": 0.0,
      "learning_rate": 1.4848400890012327e-05,
      "loss": 0.9417,
      "step": 1668
    },
    {
      "epoch": 0.3590019359001936,
      "grad_norm": 0.0,
      "learning_rate": 1.4842306021551833e-05,
      "loss": 0.9476,
      "step": 1669
    },
    {
      "epoch": 0.35921703592170356,
      "grad_norm": 0.0,
      "learning_rate": 1.4836208802422287e-05,
      "loss": 0.8808,
      "step": 1670
    },
    {
      "epoch": 0.3594321359432136,
      "grad_norm": 0.0,
      "learning_rate": 1.4830109235583548e-05,
      "loss": 0.9058,
      "step": 1671
    },
    {
      "epoch": 0.3596472359647236,
      "grad_norm": 0.0,
      "learning_rate": 1.4824007323996616e-05,
      "loss": 0.9863,
      "step": 1672
    },
    {
      "epoch": 0.3598623359862336,
      "grad_norm": 0.0,
      "learning_rate": 1.481790307062363e-05,
      "loss": 0.8945,
      "step": 1673
    },
    {
      "epoch": 0.3600774360077436,
      "grad_norm": 0.0,
      "learning_rate": 1.481179647842786e-05,
      "loss": 0.908,
      "step": 1674
    },
    {
      "epoch": 0.3602925360292536,
      "grad_norm": 0.0,
      "learning_rate": 1.4805687550373719e-05,
      "loss": 0.9296,
      "step": 1675
    },
    {
      "epoch": 0.3605076360507636,
      "grad_norm": 0.0,
      "learning_rate": 1.4799576289426747e-05,
      "loss": 0.9258,
      "step": 1676
    },
    {
      "epoch": 0.3607227360722736,
      "grad_norm": 0.0,
      "learning_rate": 1.4793462698553622e-05,
      "loss": 0.8166,
      "step": 1677
    },
    {
      "epoch": 0.3609378360937836,
      "grad_norm": 0.0,
      "learning_rate": 1.4787346780722152e-05,
      "loss": 0.9232,
      "step": 1678
    },
    {
      "epoch": 0.3611529361152936,
      "grad_norm": 0.0,
      "learning_rate": 1.4781228538901267e-05,
      "loss": 0.8803,
      "step": 1679
    },
    {
      "epoch": 0.3613680361368036,
      "grad_norm": 0.0,
      "learning_rate": 1.477510797606104e-05,
      "loss": 0.9108,
      "step": 1680
    },
    {
      "epoch": 0.3615831361583136,
      "grad_norm": 0.0,
      "learning_rate": 1.4768985095172657e-05,
      "loss": 0.8896,
      "step": 1681
    },
    {
      "epoch": 0.3617982361798236,
      "grad_norm": 0.0,
      "learning_rate": 1.4762859899208437e-05,
      "loss": 0.9184,
      "step": 1682
    },
    {
      "epoch": 0.36201333620133364,
      "grad_norm": 0.0,
      "learning_rate": 1.4756732391141819e-05,
      "loss": 0.9055,
      "step": 1683
    },
    {
      "epoch": 0.3622284362228436,
      "grad_norm": 0.0,
      "learning_rate": 1.4750602573947368e-05,
      "loss": 0.8782,
      "step": 1684
    },
    {
      "epoch": 0.3624435362443536,
      "grad_norm": 0.0,
      "learning_rate": 1.4744470450600766e-05,
      "loss": 0.9242,
      "step": 1685
    },
    {
      "epoch": 0.36265863626586364,
      "grad_norm": 0.0,
      "learning_rate": 1.4738336024078817e-05,
      "loss": 0.8612,
      "step": 1686
    },
    {
      "epoch": 0.3628737362873736,
      "grad_norm": 0.0,
      "learning_rate": 1.4732199297359439e-05,
      "loss": 0.8717,
      "step": 1687
    },
    {
      "epoch": 0.3630888363088836,
      "grad_norm": 0.0,
      "learning_rate": 1.4726060273421675e-05,
      "loss": 0.9059,
      "step": 1688
    },
    {
      "epoch": 0.36330393633039365,
      "grad_norm": 0.0,
      "learning_rate": 1.4719918955245676e-05,
      "loss": 0.8368,
      "step": 1689
    },
    {
      "epoch": 0.36351903635190363,
      "grad_norm": 0.0,
      "learning_rate": 1.4713775345812706e-05,
      "loss": 0.9615,
      "step": 1690
    },
    {
      "epoch": 0.3637341363734136,
      "grad_norm": 0.0,
      "learning_rate": 1.4707629448105155e-05,
      "loss": 0.908,
      "step": 1691
    },
    {
      "epoch": 0.36394923639492366,
      "grad_norm": 0.0,
      "learning_rate": 1.4701481265106499e-05,
      "loss": 0.8675,
      "step": 1692
    },
    {
      "epoch": 0.36416433641643364,
      "grad_norm": 0.0,
      "learning_rate": 1.4695330799801342e-05,
      "loss": 0.8997,
      "step": 1693
    },
    {
      "epoch": 0.3643794364379436,
      "grad_norm": 0.0,
      "learning_rate": 1.4689178055175396e-05,
      "loss": 0.9233,
      "step": 1694
    },
    {
      "epoch": 0.36459453645945367,
      "grad_norm": 0.0,
      "learning_rate": 1.4683023034215472e-05,
      "loss": 0.9689,
      "step": 1695
    },
    {
      "epoch": 0.36480963648096365,
      "grad_norm": 0.0,
      "learning_rate": 1.4676865739909484e-05,
      "loss": 0.9131,
      "step": 1696
    },
    {
      "epoch": 0.36502473650247363,
      "grad_norm": 0.0,
      "learning_rate": 1.4670706175246463e-05,
      "loss": 0.9643,
      "step": 1697
    },
    {
      "epoch": 0.3652398365239837,
      "grad_norm": 0.0,
      "learning_rate": 1.4664544343216529e-05,
      "loss": 0.8614,
      "step": 1698
    },
    {
      "epoch": 0.36545493654549366,
      "grad_norm": 0.0,
      "learning_rate": 1.4658380246810907e-05,
      "loss": 0.9461,
      "step": 1699
    },
    {
      "epoch": 0.36567003656700364,
      "grad_norm": 0.0,
      "learning_rate": 1.465221388902192e-05,
      "loss": 0.8719,
      "step": 1700
    },
    {
      "epoch": 0.3658851365885137,
      "grad_norm": 0.0,
      "learning_rate": 1.4646045272842995e-05,
      "loss": 0.8892,
      "step": 1701
    },
    {
      "epoch": 0.36610023661002367,
      "grad_norm": 0.0,
      "learning_rate": 1.4639874401268646e-05,
      "loss": 0.9194,
      "step": 1702
    },
    {
      "epoch": 0.36631533663153365,
      "grad_norm": 0.0,
      "learning_rate": 1.4633701277294487e-05,
      "loss": 0.8924,
      "step": 1703
    },
    {
      "epoch": 0.3665304366530437,
      "grad_norm": 0.0,
      "learning_rate": 1.4627525903917227e-05,
      "loss": 0.924,
      "step": 1704
    },
    {
      "epoch": 0.3667455366745537,
      "grad_norm": 0.0,
      "learning_rate": 1.4621348284134666e-05,
      "loss": 0.8815,
      "step": 1705
    },
    {
      "epoch": 0.36696063669606366,
      "grad_norm": 0.0,
      "learning_rate": 1.4615168420945689e-05,
      "loss": 0.9028,
      "step": 1706
    },
    {
      "epoch": 0.3671757367175737,
      "grad_norm": 0.0,
      "learning_rate": 1.4608986317350277e-05,
      "loss": 0.8602,
      "step": 1707
    },
    {
      "epoch": 0.3673908367390837,
      "grad_norm": 0.0,
      "learning_rate": 1.4602801976349497e-05,
      "loss": 0.9523,
      "step": 1708
    },
    {
      "epoch": 0.36760593676059367,
      "grad_norm": 0.0,
      "learning_rate": 1.4596615400945498e-05,
      "loss": 0.9049,
      "step": 1709
    },
    {
      "epoch": 0.3678210367821037,
      "grad_norm": 0.0,
      "learning_rate": 1.4590426594141513e-05,
      "loss": 0.9085,
      "step": 1710
    },
    {
      "epoch": 0.3680361368036137,
      "grad_norm": 0.0,
      "learning_rate": 1.4584235558941872e-05,
      "loss": 0.8336,
      "step": 1711
    },
    {
      "epoch": 0.3682512368251237,
      "grad_norm": 0.0,
      "learning_rate": 1.4578042298351973e-05,
      "loss": 0.8548,
      "step": 1712
    },
    {
      "epoch": 0.36846633684663366,
      "grad_norm": 0.0,
      "learning_rate": 1.4571846815378297e-05,
      "loss": 0.9599,
      "step": 1713
    },
    {
      "epoch": 0.3686814368681437,
      "grad_norm": 0.0,
      "learning_rate": 1.4565649113028399e-05,
      "loss": 0.9374,
      "step": 1714
    },
    {
      "epoch": 0.3688965368896537,
      "grad_norm": 0.0,
      "learning_rate": 1.4559449194310926e-05,
      "loss": 0.9406,
      "step": 1715
    },
    {
      "epoch": 0.36911163691116367,
      "grad_norm": 0.0,
      "learning_rate": 1.455324706223559e-05,
      "loss": 0.9525,
      "step": 1716
    },
    {
      "epoch": 0.3693267369326737,
      "grad_norm": 0.0,
      "learning_rate": 1.4547042719813177e-05,
      "loss": 0.8807,
      "step": 1717
    },
    {
      "epoch": 0.3695418369541837,
      "grad_norm": 0.0,
      "learning_rate": 1.4540836170055552e-05,
      "loss": 0.8947,
      "step": 1718
    },
    {
      "epoch": 0.3697569369756937,
      "grad_norm": 0.0,
      "learning_rate": 1.4534627415975646e-05,
      "loss": 0.8799,
      "step": 1719
    },
    {
      "epoch": 0.3699720369972037,
      "grad_norm": 0.0,
      "learning_rate": 1.4528416460587465e-05,
      "loss": 0.8556,
      "step": 1720
    },
    {
      "epoch": 0.3701871370187137,
      "grad_norm": 0.0,
      "learning_rate": 1.4522203306906076e-05,
      "loss": 1.0023,
      "step": 1721
    },
    {
      "epoch": 0.3704022370402237,
      "grad_norm": 0.0,
      "learning_rate": 1.451598795794762e-05,
      "loss": 0.9314,
      "step": 1722
    },
    {
      "epoch": 0.3706173370617337,
      "grad_norm": 0.0,
      "learning_rate": 1.4509770416729301e-05,
      "loss": 0.9,
      "step": 1723
    },
    {
      "epoch": 0.3708324370832437,
      "grad_norm": 0.0,
      "learning_rate": 1.450355068626939e-05,
      "loss": 0.8956,
      "step": 1724
    },
    {
      "epoch": 0.3710475371047537,
      "grad_norm": 0.0,
      "learning_rate": 1.449732876958722e-05,
      "loss": 0.9443,
      "step": 1725
    },
    {
      "epoch": 0.37126263712626373,
      "grad_norm": 0.0,
      "learning_rate": 1.449110466970318e-05,
      "loss": 0.9341,
      "step": 1726
    },
    {
      "epoch": 0.3714777371477737,
      "grad_norm": 0.0,
      "learning_rate": 1.4484878389638728e-05,
      "loss": 0.9174,
      "step": 1727
    },
    {
      "epoch": 0.3716928371692837,
      "grad_norm": 0.0,
      "learning_rate": 1.4478649932416367e-05,
      "loss": 0.9566,
      "step": 1728
    },
    {
      "epoch": 0.37190793719079374,
      "grad_norm": 0.0,
      "learning_rate": 1.4472419301059673e-05,
      "loss": 0.8323,
      "step": 1729
    },
    {
      "epoch": 0.3721230372123037,
      "grad_norm": 0.0,
      "learning_rate": 1.4466186498593265e-05,
      "loss": 0.9201,
      "step": 1730
    },
    {
      "epoch": 0.3723381372338137,
      "grad_norm": 0.0,
      "learning_rate": 1.4459951528042821e-05,
      "loss": 0.9433,
      "step": 1731
    },
    {
      "epoch": 0.37255323725532374,
      "grad_norm": 0.0,
      "learning_rate": 1.4453714392435069e-05,
      "loss": 0.9156,
      "step": 1732
    },
    {
      "epoch": 0.37276833727683373,
      "grad_norm": 0.0,
      "learning_rate": 1.4447475094797796e-05,
      "loss": 0.8842,
      "step": 1733
    },
    {
      "epoch": 0.3729834372983437,
      "grad_norm": 0.0,
      "learning_rate": 1.444123363815983e-05,
      "loss": 0.9086,
      "step": 1734
    },
    {
      "epoch": 0.37319853731985375,
      "grad_norm": 0.0,
      "learning_rate": 1.4434990025551046e-05,
      "loss": 0.9427,
      "step": 1735
    },
    {
      "epoch": 0.37341363734136374,
      "grad_norm": 0.0,
      "learning_rate": 1.442874426000237e-05,
      "loss": 0.9385,
      "step": 1736
    },
    {
      "epoch": 0.3736287373628737,
      "grad_norm": 0.0,
      "learning_rate": 1.4422496344545776e-05,
      "loss": 0.8722,
      "step": 1737
    },
    {
      "epoch": 0.37384383738438376,
      "grad_norm": 0.0,
      "learning_rate": 1.4416246282214275e-05,
      "loss": 0.8745,
      "step": 1738
    },
    {
      "epoch": 0.37405893740589374,
      "grad_norm": 0.0,
      "learning_rate": 1.4409994076041922e-05,
      "loss": 0.9344,
      "step": 1739
    },
    {
      "epoch": 0.3742740374274037,
      "grad_norm": 0.0,
      "learning_rate": 1.4403739729063817e-05,
      "loss": 0.8542,
      "step": 1740
    },
    {
      "epoch": 0.37448913744891377,
      "grad_norm": 0.0,
      "learning_rate": 1.4397483244316095e-05,
      "loss": 0.9249,
      "step": 1741
    },
    {
      "epoch": 0.37470423747042375,
      "grad_norm": 0.0,
      "learning_rate": 1.4391224624835928e-05,
      "loss": 0.8548,
      "step": 1742
    },
    {
      "epoch": 0.37491933749193374,
      "grad_norm": 0.0,
      "learning_rate": 1.4384963873661528e-05,
      "loss": 0.9514,
      "step": 1743
    },
    {
      "epoch": 0.3751344375134438,
      "grad_norm": 0.0,
      "learning_rate": 1.437870099383214e-05,
      "loss": 0.8714,
      "step": 1744
    },
    {
      "epoch": 0.37534953753495376,
      "grad_norm": 0.0,
      "learning_rate": 1.437243598838804e-05,
      "loss": 0.8158,
      "step": 1745
    },
    {
      "epoch": 0.37556463755646374,
      "grad_norm": 0.0,
      "learning_rate": 1.436616886037054e-05,
      "loss": 0.8196,
      "step": 1746
    },
    {
      "epoch": 0.3757797375779738,
      "grad_norm": 0.0,
      "learning_rate": 1.4359899612821981e-05,
      "loss": 0.9784,
      "step": 1747
    },
    {
      "epoch": 0.37599483759948377,
      "grad_norm": 0.0,
      "learning_rate": 1.435362824878573e-05,
      "loss": 0.9334,
      "step": 1748
    },
    {
      "epoch": 0.37620993762099375,
      "grad_norm": 0.0,
      "learning_rate": 1.4347354771306193e-05,
      "loss": 0.9912,
      "step": 1749
    },
    {
      "epoch": 0.3764250376425038,
      "grad_norm": 0.0,
      "learning_rate": 1.4341079183428776e-05,
      "loss": 0.9196,
      "step": 1750
    },
    {
      "epoch": 0.3766401376640138,
      "grad_norm": 0.0,
      "learning_rate": 1.4334801488199941e-05,
      "loss": 0.9339,
      "step": 1751
    },
    {
      "epoch": 0.37685523768552376,
      "grad_norm": 0.0,
      "learning_rate": 1.4328521688667152e-05,
      "loss": 0.9069,
      "step": 1752
    },
    {
      "epoch": 0.3770703377070338,
      "grad_norm": 0.0,
      "learning_rate": 1.4322239787878902e-05,
      "loss": 1.013,
      "step": 1753
    },
    {
      "epoch": 0.3772854377285438,
      "grad_norm": 0.0,
      "learning_rate": 1.43159557888847e-05,
      "loss": 0.8566,
      "step": 1754
    },
    {
      "epoch": 0.37750053775005377,
      "grad_norm": 0.0,
      "learning_rate": 1.4309669694735085e-05,
      "loss": 0.8762,
      "step": 1755
    },
    {
      "epoch": 0.37771563777156375,
      "grad_norm": 0.0,
      "learning_rate": 1.4303381508481595e-05,
      "loss": 0.8951,
      "step": 1756
    },
    {
      "epoch": 0.3779307377930738,
      "grad_norm": 0.0,
      "learning_rate": 1.4297091233176796e-05,
      "loss": 0.9563,
      "step": 1757
    },
    {
      "epoch": 0.3781458378145838,
      "grad_norm": 0.0,
      "learning_rate": 1.4290798871874267e-05,
      "loss": 0.9086,
      "step": 1758
    },
    {
      "epoch": 0.37836093783609376,
      "grad_norm": 0.0,
      "learning_rate": 1.4284504427628598e-05,
      "loss": 0.8805,
      "step": 1759
    },
    {
      "epoch": 0.3785760378576038,
      "grad_norm": 0.0,
      "learning_rate": 1.4278207903495388e-05,
      "loss": 0.8947,
      "step": 1760
    },
    {
      "epoch": 0.3787911378791138,
      "grad_norm": 0.0,
      "learning_rate": 1.427190930253125e-05,
      "loss": 0.9048,
      "step": 1761
    },
    {
      "epoch": 0.37900623790062377,
      "grad_norm": 0.0,
      "learning_rate": 1.4265608627793803e-05,
      "loss": 0.9219,
      "step": 1762
    },
    {
      "epoch": 0.3792213379221338,
      "grad_norm": 0.0,
      "learning_rate": 1.425930588234167e-05,
      "loss": 0.9345,
      "step": 1763
    },
    {
      "epoch": 0.3794364379436438,
      "grad_norm": 0.0,
      "learning_rate": 1.4253001069234483e-05,
      "loss": 0.8421,
      "step": 1764
    },
    {
      "epoch": 0.3796515379651538,
      "grad_norm": 0.0,
      "learning_rate": 1.4246694191532874e-05,
      "loss": 0.8909,
      "step": 1765
    },
    {
      "epoch": 0.3798666379866638,
      "grad_norm": 0.0,
      "learning_rate": 1.4240385252298485e-05,
      "loss": 0.856,
      "step": 1766
    },
    {
      "epoch": 0.3800817380081738,
      "grad_norm": 0.0,
      "learning_rate": 1.423407425459395e-05,
      "loss": 0.8585,
      "step": 1767
    },
    {
      "epoch": 0.3802968380296838,
      "grad_norm": 0.0,
      "learning_rate": 1.4227761201482908e-05,
      "loss": 0.8645,
      "step": 1768
    },
    {
      "epoch": 0.3805119380511938,
      "grad_norm": 0.0,
      "learning_rate": 1.4221446096029992e-05,
      "loss": 0.9566,
      "step": 1769
    },
    {
      "epoch": 0.3807270380727038,
      "grad_norm": 0.0,
      "learning_rate": 1.4215128941300835e-05,
      "loss": 0.9548,
      "step": 1770
    },
    {
      "epoch": 0.3809421380942138,
      "grad_norm": 0.0,
      "learning_rate": 1.420880974036206e-05,
      "loss": 0.8953,
      "step": 1771
    },
    {
      "epoch": 0.38115723811572383,
      "grad_norm": 0.0,
      "learning_rate": 1.4202488496281283e-05,
      "loss": 0.9062,
      "step": 1772
    },
    {
      "epoch": 0.3813723381372338,
      "grad_norm": 0.0,
      "learning_rate": 1.4196165212127122e-05,
      "loss": 0.8989,
      "step": 1773
    },
    {
      "epoch": 0.3815874381587438,
      "grad_norm": 0.0,
      "learning_rate": 1.4189839890969174e-05,
      "loss": 0.8809,
      "step": 1774
    },
    {
      "epoch": 0.38180253818025384,
      "grad_norm": 0.0,
      "learning_rate": 1.4183512535878032e-05,
      "loss": 0.947,
      "step": 1775
    },
    {
      "epoch": 0.3820176382017638,
      "grad_norm": 0.0,
      "learning_rate": 1.4177183149925266e-05,
      "loss": 0.9095,
      "step": 1776
    },
    {
      "epoch": 0.3822327382232738,
      "grad_norm": 0.0,
      "learning_rate": 1.4170851736183452e-05,
      "loss": 0.8512,
      "step": 1777
    },
    {
      "epoch": 0.38244783824478384,
      "grad_norm": 0.0,
      "learning_rate": 1.4164518297726129e-05,
      "loss": 0.862,
      "step": 1778
    },
    {
      "epoch": 0.38266293826629383,
      "grad_norm": 0.0,
      "learning_rate": 1.4158182837627827e-05,
      "loss": 0.8526,
      "step": 1779
    },
    {
      "epoch": 0.3828780382878038,
      "grad_norm": 0.0,
      "learning_rate": 1.4151845358964058e-05,
      "loss": 0.856,
      "step": 1780
    },
    {
      "epoch": 0.38309313830931385,
      "grad_norm": 0.0,
      "learning_rate": 1.4145505864811319e-05,
      "loss": 0.9,
      "step": 1781
    },
    {
      "epoch": 0.38330823833082384,
      "grad_norm": 0.0,
      "learning_rate": 1.4139164358247074e-05,
      "loss": 0.8542,
      "step": 1782
    },
    {
      "epoch": 0.3835233383523338,
      "grad_norm": 0.0,
      "learning_rate": 1.4132820842349778e-05,
      "loss": 0.882,
      "step": 1783
    },
    {
      "epoch": 0.38373843837384386,
      "grad_norm": 0.0,
      "learning_rate": 1.4126475320198845e-05,
      "loss": 0.8639,
      "step": 1784
    },
    {
      "epoch": 0.38395353839535384,
      "grad_norm": 0.0,
      "learning_rate": 1.4120127794874675e-05,
      "loss": 0.8563,
      "step": 1785
    },
    {
      "epoch": 0.38416863841686383,
      "grad_norm": 0.0,
      "learning_rate": 1.411377826945864e-05,
      "loss": 0.8972,
      "step": 1786
    },
    {
      "epoch": 0.38438373843837387,
      "grad_norm": 0.0,
      "learning_rate": 1.4107426747033075e-05,
      "loss": 0.8981,
      "step": 1787
    },
    {
      "epoch": 0.38459883845988385,
      "grad_norm": 0.0,
      "learning_rate": 1.4101073230681292e-05,
      "loss": 0.8329,
      "step": 1788
    },
    {
      "epoch": 0.38481393848139384,
      "grad_norm": 0.0,
      "learning_rate": 1.4094717723487565e-05,
      "loss": 0.8726,
      "step": 1789
    },
    {
      "epoch": 0.3850290385029039,
      "grad_norm": 0.0,
      "learning_rate": 1.4088360228537144e-05,
      "loss": 0.8814,
      "step": 1790
    },
    {
      "epoch": 0.38524413852441386,
      "grad_norm": 0.0,
      "learning_rate": 1.4082000748916232e-05,
      "loss": 0.8579,
      "step": 1791
    },
    {
      "epoch": 0.38545923854592384,
      "grad_norm": 0.0,
      "learning_rate": 1.4075639287712006e-05,
      "loss": 0.9411,
      "step": 1792
    },
    {
      "epoch": 0.3856743385674339,
      "grad_norm": 0.0,
      "learning_rate": 1.4069275848012596e-05,
      "loss": 0.9025,
      "step": 1793
    },
    {
      "epoch": 0.38588943858894387,
      "grad_norm": 0.0,
      "learning_rate": 1.4062910432907094e-05,
      "loss": 0.9487,
      "step": 1794
    },
    {
      "epoch": 0.38610453861045385,
      "grad_norm": 0.0,
      "learning_rate": 1.4056543045485558e-05,
      "loss": 0.9148,
      "step": 1795
    },
    {
      "epoch": 0.3863196386319639,
      "grad_norm": 0.0,
      "learning_rate": 1.4050173688839e-05,
      "loss": 0.8768,
      "step": 1796
    },
    {
      "epoch": 0.3865347386534739,
      "grad_norm": 0.0,
      "learning_rate": 1.4043802366059383e-05,
      "loss": 0.9105,
      "step": 1797
    },
    {
      "epoch": 0.38674983867498386,
      "grad_norm": 0.0,
      "learning_rate": 1.4037429080239629e-05,
      "loss": 0.9424,
      "step": 1798
    },
    {
      "epoch": 0.38696493869649384,
      "grad_norm": 0.0,
      "learning_rate": 1.4031053834473614e-05,
      "loss": 0.8852,
      "step": 1799
    },
    {
      "epoch": 0.3871800387180039,
      "grad_norm": 0.0,
      "learning_rate": 1.4024676631856162e-05,
      "loss": 0.9939,
      "step": 1800
    },
    {
      "epoch": 0.38739513873951387,
      "grad_norm": 0.0,
      "learning_rate": 1.4018297475483049e-05,
      "loss": 0.871,
      "step": 1801
    },
    {
      "epoch": 0.38761023876102385,
      "grad_norm": 0.0,
      "learning_rate": 1.4011916368451e-05,
      "loss": 0.8778,
      "step": 1802
    },
    {
      "epoch": 0.3878253387825339,
      "grad_norm": 0.0,
      "learning_rate": 1.4005533313857686e-05,
      "loss": 0.906,
      "step": 1803
    },
    {
      "epoch": 0.3880404388040439,
      "grad_norm": 0.0,
      "learning_rate": 1.3999148314801722e-05,
      "loss": 0.9358,
      "step": 1804
    },
    {
      "epoch": 0.38825553882555386,
      "grad_norm": 0.0,
      "learning_rate": 1.399276137438267e-05,
      "loss": 0.8868,
      "step": 1805
    },
    {
      "epoch": 0.3884706388470639,
      "grad_norm": 0.0,
      "learning_rate": 1.398637249570103e-05,
      "loss": 0.8854,
      "step": 1806
    },
    {
      "epoch": 0.3886857388685739,
      "grad_norm": 0.0,
      "learning_rate": 1.397998168185825e-05,
      "loss": 0.8583,
      "step": 1807
    },
    {
      "epoch": 0.38890083889008387,
      "grad_norm": 0.0,
      "learning_rate": 1.397358893595671e-05,
      "loss": 0.9139,
      "step": 1808
    },
    {
      "epoch": 0.3891159389115939,
      "grad_norm": 0.0,
      "learning_rate": 1.396719426109973e-05,
      "loss": 0.9681,
      "step": 1809
    },
    {
      "epoch": 0.3893310389331039,
      "grad_norm": 0.0,
      "learning_rate": 1.396079766039157e-05,
      "loss": 0.9382,
      "step": 1810
    },
    {
      "epoch": 0.3895461389546139,
      "grad_norm": 0.0,
      "learning_rate": 1.395439913693742e-05,
      "loss": 0.9427,
      "step": 1811
    },
    {
      "epoch": 0.3897612389761239,
      "grad_norm": 0.0,
      "learning_rate": 1.3947998693843407e-05,
      "loss": 0.9697,
      "step": 1812
    },
    {
      "epoch": 0.3899763389976339,
      "grad_norm": 0.0,
      "learning_rate": 1.3941596334216592e-05,
      "loss": 0.8143,
      "step": 1813
    },
    {
      "epoch": 0.3901914390191439,
      "grad_norm": 0.0,
      "learning_rate": 1.3935192061164957e-05,
      "loss": 0.819,
      "step": 1814
    },
    {
      "epoch": 0.3904065390406539,
      "grad_norm": 0.0,
      "learning_rate": 1.392878587779742e-05,
      "loss": 0.8667,
      "step": 1815
    },
    {
      "epoch": 0.3906216390621639,
      "grad_norm": 0.0,
      "learning_rate": 1.3922377787223828e-05,
      "loss": 0.8553,
      "step": 1816
    },
    {
      "epoch": 0.3908367390836739,
      "grad_norm": 0.0,
      "learning_rate": 1.3915967792554947e-05,
      "loss": 0.9214,
      "step": 1817
    },
    {
      "epoch": 0.39105183910518393,
      "grad_norm": 0.0,
      "learning_rate": 1.3909555896902478e-05,
      "loss": 0.8689,
      "step": 1818
    },
    {
      "epoch": 0.3912669391266939,
      "grad_norm": 0.0,
      "learning_rate": 1.390314210337903e-05,
      "loss": 0.8627,
      "step": 1819
    },
    {
      "epoch": 0.3914820391482039,
      "grad_norm": 0.0,
      "learning_rate": 1.3896726415098147e-05,
      "loss": 0.9638,
      "step": 1820
    },
    {
      "epoch": 0.39169713916971394,
      "grad_norm": 0.0,
      "learning_rate": 1.3890308835174284e-05,
      "loss": 0.9366,
      "step": 1821
    },
    {
      "epoch": 0.3919122391912239,
      "grad_norm": 0.0,
      "learning_rate": 1.388388936672282e-05,
      "loss": 0.8771,
      "step": 1822
    },
    {
      "epoch": 0.3921273392127339,
      "grad_norm": 0.0,
      "learning_rate": 1.3877468012860047e-05,
      "loss": 0.8946,
      "step": 1823
    },
    {
      "epoch": 0.39234243923424394,
      "grad_norm": 0.0,
      "learning_rate": 1.3871044776703173e-05,
      "loss": 0.8707,
      "step": 1824
    },
    {
      "epoch": 0.39255753925575393,
      "grad_norm": 0.0,
      "learning_rate": 1.3864619661370321e-05,
      "loss": 0.9297,
      "step": 1825
    },
    {
      "epoch": 0.3927726392772639,
      "grad_norm": 0.0,
      "learning_rate": 1.3858192669980529e-05,
      "loss": 0.905,
      "step": 1826
    },
    {
      "epoch": 0.39298773929877395,
      "grad_norm": 0.0,
      "learning_rate": 1.3851763805653735e-05,
      "loss": 0.8928,
      "step": 1827
    },
    {
      "epoch": 0.39320283932028394,
      "grad_norm": 0.0,
      "learning_rate": 1.38453330715108e-05,
      "loss": 0.8528,
      "step": 1828
    },
    {
      "epoch": 0.3934179393417939,
      "grad_norm": 0.0,
      "learning_rate": 1.3838900470673482e-05,
      "loss": 0.9355,
      "step": 1829
    },
    {
      "epoch": 0.39363303936330396,
      "grad_norm": 0.0,
      "learning_rate": 1.3832466006264451e-05,
      "loss": 0.9155,
      "step": 1830
    },
    {
      "epoch": 0.39384813938481394,
      "grad_norm": 0.0,
      "learning_rate": 1.382602968140728e-05,
      "loss": 0.8896,
      "step": 1831
    },
    {
      "epoch": 0.39406323940632393,
      "grad_norm": 0.0,
      "learning_rate": 1.3819591499226446e-05,
      "loss": 0.8921,
      "step": 1832
    },
    {
      "epoch": 0.39427833942783397,
      "grad_norm": 0.0,
      "learning_rate": 1.3813151462847323e-05,
      "loss": 0.8622,
      "step": 1833
    },
    {
      "epoch": 0.39449343944934395,
      "grad_norm": 0.0,
      "learning_rate": 1.3806709575396195e-05,
      "loss": 0.8891,
      "step": 1834
    },
    {
      "epoch": 0.39470853947085394,
      "grad_norm": 0.0,
      "learning_rate": 1.3800265840000236e-05,
      "loss": 0.9194,
      "step": 1835
    },
    {
      "epoch": 0.394923639492364,
      "grad_norm": 0.0,
      "learning_rate": 1.3793820259787519e-05,
      "loss": 0.9322,
      "step": 1836
    },
    {
      "epoch": 0.39513873951387396,
      "grad_norm": 0.0,
      "learning_rate": 1.3787372837887007e-05,
      "loss": 0.9211,
      "step": 1837
    },
    {
      "epoch": 0.39535383953538394,
      "grad_norm": 0.0,
      "learning_rate": 1.3780923577428574e-05,
      "loss": 0.8446,
      "step": 1838
    },
    {
      "epoch": 0.39556893955689393,
      "grad_norm": 0.0,
      "learning_rate": 1.3774472481542969e-05,
      "loss": 0.8529,
      "step": 1839
    },
    {
      "epoch": 0.39578403957840397,
      "grad_norm": 0.0,
      "learning_rate": 1.376801955336184e-05,
      "loss": 0.9241,
      "step": 1840
    },
    {
      "epoch": 0.39599913959991395,
      "grad_norm": 0.0,
      "learning_rate": 1.3761564796017724e-05,
      "loss": 0.905,
      "step": 1841
    },
    {
      "epoch": 0.39621423962142394,
      "grad_norm": 0.0,
      "learning_rate": 1.3755108212644044e-05,
      "loss": 0.9177,
      "step": 1842
    },
    {
      "epoch": 0.396429339642934,
      "grad_norm": 0.0,
      "learning_rate": 1.3748649806375112e-05,
      "loss": 0.9038,
      "step": 1843
    },
    {
      "epoch": 0.39664443966444396,
      "grad_norm": 0.0,
      "learning_rate": 1.3742189580346122e-05,
      "loss": 0.873,
      "step": 1844
    },
    {
      "epoch": 0.39685953968595394,
      "grad_norm": 0.0,
      "learning_rate": 1.3735727537693152e-05,
      "loss": 0.8978,
      "step": 1845
    },
    {
      "epoch": 0.397074639707464,
      "grad_norm": 0.0,
      "learning_rate": 1.3729263681553165e-05,
      "loss": 0.9516,
      "step": 1846
    },
    {
      "epoch": 0.39728973972897397,
      "grad_norm": 0.0,
      "learning_rate": 1.3722798015064003e-05,
      "loss": 0.8865,
      "step": 1847
    },
    {
      "epoch": 0.39750483975048395,
      "grad_norm": 0.0,
      "learning_rate": 1.371633054136438e-05,
      "loss": 0.874,
      "step": 1848
    },
    {
      "epoch": 0.397719939771994,
      "grad_norm": 0.0,
      "learning_rate": 1.37098612635939e-05,
      "loss": 0.8982,
      "step": 1849
    },
    {
      "epoch": 0.397935039793504,
      "grad_norm": 0.0,
      "learning_rate": 1.3703390184893034e-05,
      "loss": 0.8921,
      "step": 1850
    },
    {
      "epoch": 0.39815013981501396,
      "grad_norm": 0.0,
      "learning_rate": 1.3696917308403123e-05,
      "loss": 0.875,
      "step": 1851
    },
    {
      "epoch": 0.398365239836524,
      "grad_norm": 0.0,
      "learning_rate": 1.3690442637266398e-05,
      "loss": 0.9117,
      "step": 1852
    },
    {
      "epoch": 0.398580339858034,
      "grad_norm": 0.0,
      "learning_rate": 1.3683966174625942e-05,
      "loss": 0.8856,
      "step": 1853
    },
    {
      "epoch": 0.39879543987954397,
      "grad_norm": 0.0,
      "learning_rate": 1.3677487923625719e-05,
      "loss": 0.8696,
      "step": 1854
    },
    {
      "epoch": 0.399010539901054,
      "grad_norm": 0.0,
      "learning_rate": 1.3671007887410553e-05,
      "loss": 0.843,
      "step": 1855
    },
    {
      "epoch": 0.399225639922564,
      "grad_norm": 0.0,
      "learning_rate": 1.3664526069126148e-05,
      "loss": 0.8491,
      "step": 1856
    },
    {
      "epoch": 0.399440739944074,
      "grad_norm": 0.0,
      "learning_rate": 1.365804247191906e-05,
      "loss": 0.9062,
      "step": 1857
    },
    {
      "epoch": 0.399655839965584,
      "grad_norm": 0.0,
      "learning_rate": 1.3651557098936712e-05,
      "loss": 0.9104,
      "step": 1858
    },
    {
      "epoch": 0.399870939987094,
      "grad_norm": 0.0,
      "learning_rate": 1.364506995332739e-05,
      "loss": 0.8826,
      "step": 1859
    },
    {
      "epoch": 0.400086040008604,
      "grad_norm": 0.0,
      "learning_rate": 1.3638581038240242e-05,
      "loss": 0.9801,
      "step": 1860
    },
    {
      "epoch": 0.400301140030114,
      "grad_norm": 0.0,
      "learning_rate": 1.3632090356825279e-05,
      "loss": 0.9257,
      "step": 1861
    },
    {
      "epoch": 0.400516240051624,
      "grad_norm": 0.0,
      "learning_rate": 1.3625597912233355e-05,
      "loss": 0.9084,
      "step": 1862
    },
    {
      "epoch": 0.400731340073134,
      "grad_norm": 0.0,
      "learning_rate": 1.3619103707616198e-05,
      "loss": 0.8436,
      "step": 1863
    },
    {
      "epoch": 0.40094644009464403,
      "grad_norm": 0.0,
      "learning_rate": 1.3612607746126375e-05,
      "loss": 0.883,
      "step": 1864
    },
    {
      "epoch": 0.401161540116154,
      "grad_norm": 0.0,
      "learning_rate": 1.3606110030917311e-05,
      "loss": 0.9265,
      "step": 1865
    },
    {
      "epoch": 0.401376640137664,
      "grad_norm": 0.0,
      "learning_rate": 1.3599610565143291e-05,
      "loss": 0.8932,
      "step": 1866
    },
    {
      "epoch": 0.40159174015917404,
      "grad_norm": 0.0,
      "learning_rate": 1.3593109351959439e-05,
      "loss": 0.898,
      "step": 1867
    },
    {
      "epoch": 0.401806840180684,
      "grad_norm": 0.0,
      "learning_rate": 1.3586606394521729e-05,
      "loss": 0.8936,
      "step": 1868
    },
    {
      "epoch": 0.402021940202194,
      "grad_norm": 0.0,
      "learning_rate": 1.3580101695986985e-05,
      "loss": 0.8906,
      "step": 1869
    },
    {
      "epoch": 0.40223704022370405,
      "grad_norm": 0.0,
      "learning_rate": 1.3573595259512877e-05,
      "loss": 0.8548,
      "step": 1870
    },
    {
      "epoch": 0.40245214024521403,
      "grad_norm": 0.0,
      "learning_rate": 1.3567087088257913e-05,
      "loss": 0.9231,
      "step": 1871
    },
    {
      "epoch": 0.402667240266724,
      "grad_norm": 0.0,
      "learning_rate": 1.356057718538145e-05,
      "loss": 0.9209,
      "step": 1872
    },
    {
      "epoch": 0.40288234028823405,
      "grad_norm": 0.0,
      "learning_rate": 1.3554065554043675e-05,
      "loss": 0.9232,
      "step": 1873
    },
    {
      "epoch": 0.40309744030974404,
      "grad_norm": 0.0,
      "learning_rate": 1.3547552197405632e-05,
      "loss": 0.8643,
      "step": 1874
    },
    {
      "epoch": 0.403312540331254,
      "grad_norm": 0.0,
      "learning_rate": 1.3541037118629181e-05,
      "loss": 0.9196,
      "step": 1875
    },
    {
      "epoch": 0.40352764035276406,
      "grad_norm": 0.0,
      "learning_rate": 1.3534520320877036e-05,
      "loss": 0.8398,
      "step": 1876
    },
    {
      "epoch": 0.40374274037427405,
      "grad_norm": 0.0,
      "learning_rate": 1.3528001807312734e-05,
      "loss": 0.8372,
      "step": 1877
    },
    {
      "epoch": 0.40395784039578403,
      "grad_norm": 0.0,
      "learning_rate": 1.3521481581100657e-05,
      "loss": 0.8664,
      "step": 1878
    },
    {
      "epoch": 0.40417294041729407,
      "grad_norm": 0.0,
      "learning_rate": 1.3514959645406e-05,
      "loss": 0.8969,
      "step": 1879
    },
    {
      "epoch": 0.40438804043880405,
      "grad_norm": 0.0,
      "learning_rate": 1.3508436003394804e-05,
      "loss": 0.9254,
      "step": 1880
    },
    {
      "epoch": 0.40460314046031404,
      "grad_norm": 0.0,
      "learning_rate": 1.3501910658233932e-05,
      "loss": 0.9016,
      "step": 1881
    },
    {
      "epoch": 0.404818240481824,
      "grad_norm": 0.0,
      "learning_rate": 1.3495383613091073e-05,
      "loss": 0.9003,
      "step": 1882
    },
    {
      "epoch": 0.40503334050333406,
      "grad_norm": 0.0,
      "learning_rate": 1.3488854871134746e-05,
      "loss": 0.8732,
      "step": 1883
    },
    {
      "epoch": 0.40524844052484404,
      "grad_norm": 0.0,
      "learning_rate": 1.348232443553429e-05,
      "loss": 0.9221,
      "step": 1884
    },
    {
      "epoch": 0.40546354054635403,
      "grad_norm": 0.0,
      "learning_rate": 1.3475792309459862e-05,
      "loss": 0.8695,
      "step": 1885
    },
    {
      "epoch": 0.40567864056786407,
      "grad_norm": 0.0,
      "learning_rate": 1.3469258496082452e-05,
      "loss": 0.9422,
      "step": 1886
    },
    {
      "epoch": 0.40589374058937405,
      "grad_norm": 0.0,
      "learning_rate": 1.3462722998573854e-05,
      "loss": 0.9474,
      "step": 1887
    },
    {
      "epoch": 0.40610884061088404,
      "grad_norm": 0.0,
      "learning_rate": 1.345618582010669e-05,
      "loss": 0.9389,
      "step": 1888
    },
    {
      "epoch": 0.4063239406323941,
      "grad_norm": 0.0,
      "learning_rate": 1.3449646963854396e-05,
      "loss": 0.8372,
      "step": 1889
    },
    {
      "epoch": 0.40653904065390406,
      "grad_norm": 0.0,
      "learning_rate": 1.3443106432991218e-05,
      "loss": 0.9275,
      "step": 1890
    },
    {
      "epoch": 0.40675414067541404,
      "grad_norm": 0.0,
      "learning_rate": 1.3436564230692221e-05,
      "loss": 0.8733,
      "step": 1891
    },
    {
      "epoch": 0.4069692406969241,
      "grad_norm": 0.0,
      "learning_rate": 1.3430020360133283e-05,
      "loss": 0.9065,
      "step": 1892
    },
    {
      "epoch": 0.40718434071843407,
      "grad_norm": 0.0,
      "learning_rate": 1.3423474824491081e-05,
      "loss": 0.904,
      "step": 1893
    },
    {
      "epoch": 0.40739944073994405,
      "grad_norm": 0.0,
      "learning_rate": 1.3416927626943105e-05,
      "loss": 0.894,
      "step": 1894
    },
    {
      "epoch": 0.4076145407614541,
      "grad_norm": 0.0,
      "learning_rate": 1.341037877066766e-05,
      "loss": 0.9006,
      "step": 1895
    },
    {
      "epoch": 0.4078296407829641,
      "grad_norm": 0.0,
      "learning_rate": 1.3403828258843847e-05,
      "loss": 0.8571,
      "step": 1896
    },
    {
      "epoch": 0.40804474080447406,
      "grad_norm": 0.0,
      "learning_rate": 1.3397276094651575e-05,
      "loss": 0.9183,
      "step": 1897
    },
    {
      "epoch": 0.4082598408259841,
      "grad_norm": 0.0,
      "learning_rate": 1.3390722281271553e-05,
      "loss": 0.9314,
      "step": 1898
    },
    {
      "epoch": 0.4084749408474941,
      "grad_norm": 0.0,
      "learning_rate": 1.338416682188529e-05,
      "loss": 0.8895,
      "step": 1899
    },
    {
      "epoch": 0.40869004086900407,
      "grad_norm": 0.0,
      "learning_rate": 1.3377609719675096e-05,
      "loss": 0.8404,
      "step": 1900
    },
    {
      "epoch": 0.4089051408905141,
      "grad_norm": 0.0,
      "learning_rate": 1.3371050977824077e-05,
      "loss": 0.9122,
      "step": 1901
    },
    {
      "epoch": 0.4091202409120241,
      "grad_norm": 0.0,
      "learning_rate": 1.3364490599516135e-05,
      "loss": 0.9219,
      "step": 1902
    },
    {
      "epoch": 0.4093353409335341,
      "grad_norm": 0.0,
      "learning_rate": 1.3357928587935968e-05,
      "loss": 0.9313,
      "step": 1903
    },
    {
      "epoch": 0.4095504409550441,
      "grad_norm": 0.0,
      "learning_rate": 1.3351364946269071e-05,
      "loss": 0.9514,
      "step": 1904
    },
    {
      "epoch": 0.4097655409765541,
      "grad_norm": 0.0,
      "learning_rate": 1.334479967770172e-05,
      "loss": 0.8505,
      "step": 1905
    },
    {
      "epoch": 0.4099806409980641,
      "grad_norm": 0.0,
      "learning_rate": 1.3338232785420988e-05,
      "loss": 0.8356,
      "step": 1906
    },
    {
      "epoch": 0.4101957410195741,
      "grad_norm": 0.0,
      "learning_rate": 1.3331664272614733e-05,
      "loss": 0.8273,
      "step": 1907
    },
    {
      "epoch": 0.4104108410410841,
      "grad_norm": 0.0,
      "learning_rate": 1.3325094142471602e-05,
      "loss": 0.8633,
      "step": 1908
    },
    {
      "epoch": 0.4106259410625941,
      "grad_norm": 0.0,
      "learning_rate": 1.3318522398181023e-05,
      "loss": 0.8923,
      "step": 1909
    },
    {
      "epoch": 0.41084104108410413,
      "grad_norm": 0.0,
      "learning_rate": 1.3311949042933216e-05,
      "loss": 0.9089,
      "step": 1910
    },
    {
      "epoch": 0.4110561411056141,
      "grad_norm": 0.0,
      "learning_rate": 1.3305374079919178e-05,
      "loss": 0.8617,
      "step": 1911
    },
    {
      "epoch": 0.4112712411271241,
      "grad_norm": 0.0,
      "learning_rate": 1.3298797512330677e-05,
      "loss": 0.8599,
      "step": 1912
    },
    {
      "epoch": 0.41148634114863414,
      "grad_norm": 0.0,
      "learning_rate": 1.329221934336028e-05,
      "loss": 0.9471,
      "step": 1913
    },
    {
      "epoch": 0.4117014411701441,
      "grad_norm": 0.0,
      "learning_rate": 1.3285639576201315e-05,
      "loss": 0.8817,
      "step": 1914
    },
    {
      "epoch": 0.4119165411916541,
      "grad_norm": 0.0,
      "learning_rate": 1.3279058214047895e-05,
      "loss": 0.8725,
      "step": 1915
    },
    {
      "epoch": 0.41213164121316415,
      "grad_norm": 0.0,
      "learning_rate": 1.3272475260094897e-05,
      "loss": 0.9,
      "step": 1916
    },
    {
      "epoch": 0.41234674123467413,
      "grad_norm": 0.0,
      "learning_rate": 1.3265890717537983e-05,
      "loss": 0.9205,
      "step": 1917
    },
    {
      "epoch": 0.4125618412561841,
      "grad_norm": 0.0,
      "learning_rate": 1.3259304589573581e-05,
      "loss": 0.9393,
      "step": 1918
    },
    {
      "epoch": 0.41277694127769415,
      "grad_norm": 0.0,
      "learning_rate": 1.3252716879398884e-05,
      "loss": 0.9646,
      "step": 1919
    },
    {
      "epoch": 0.41299204129920414,
      "grad_norm": 0.0,
      "learning_rate": 1.3246127590211859e-05,
      "loss": 0.8889,
      "step": 1920
    },
    {
      "epoch": 0.4132071413207141,
      "grad_norm": 0.0,
      "learning_rate": 1.323953672521124e-05,
      "loss": 0.9257,
      "step": 1921
    },
    {
      "epoch": 0.41342224134222416,
      "grad_norm": 0.0,
      "learning_rate": 1.323294428759652e-05,
      "loss": 0.9172,
      "step": 1922
    },
    {
      "epoch": 0.41363734136373415,
      "grad_norm": 0.0,
      "learning_rate": 1.3226350280567965e-05,
      "loss": 0.8958,
      "step": 1923
    },
    {
      "epoch": 0.41385244138524413,
      "grad_norm": 0.0,
      "learning_rate": 1.3219754707326592e-05,
      "loss": 0.8886,
      "step": 1924
    },
    {
      "epoch": 0.4140675414067541,
      "grad_norm": 0.0,
      "learning_rate": 1.321315757107418e-05,
      "loss": 0.9036,
      "step": 1925
    },
    {
      "epoch": 0.41428264142826415,
      "grad_norm": 0.0,
      "learning_rate": 1.320655887501328e-05,
      "loss": 0.921,
      "step": 1926
    },
    {
      "epoch": 0.41449774144977414,
      "grad_norm": 0.0,
      "learning_rate": 1.3199958622347183e-05,
      "loss": 0.9295,
      "step": 1927
    },
    {
      "epoch": 0.4147128414712841,
      "grad_norm": 0.0,
      "learning_rate": 1.3193356816279945e-05,
      "loss": 0.8753,
      "step": 1928
    },
    {
      "epoch": 0.41492794149279416,
      "grad_norm": 0.0,
      "learning_rate": 1.3186753460016376e-05,
      "loss": 0.8717,
      "step": 1929
    },
    {
      "epoch": 0.41514304151430415,
      "grad_norm": 0.0,
      "learning_rate": 1.3180148556762031e-05,
      "loss": 0.857,
      "step": 1930
    },
    {
      "epoch": 0.41535814153581413,
      "grad_norm": 0.0,
      "learning_rate": 1.3173542109723227e-05,
      "loss": 0.888,
      "step": 1931
    },
    {
      "epoch": 0.41557324155732417,
      "grad_norm": 0.0,
      "learning_rate": 1.3166934122107025e-05,
      "loss": 0.8493,
      "step": 1932
    },
    {
      "epoch": 0.41578834157883415,
      "grad_norm": 0.0,
      "learning_rate": 1.3160324597121228e-05,
      "loss": 0.8767,
      "step": 1933
    },
    {
      "epoch": 0.41600344160034414,
      "grad_norm": 0.0,
      "learning_rate": 1.3153713537974396e-05,
      "loss": 0.9203,
      "step": 1934
    },
    {
      "epoch": 0.4162185416218542,
      "grad_norm": 0.0,
      "learning_rate": 1.3147100947875826e-05,
      "loss": 0.8639,
      "step": 1935
    },
    {
      "epoch": 0.41643364164336416,
      "grad_norm": 0.0,
      "learning_rate": 1.3140486830035568e-05,
      "loss": 0.8801,
      "step": 1936
    },
    {
      "epoch": 0.41664874166487414,
      "grad_norm": 0.0,
      "learning_rate": 1.31338711876644e-05,
      "loss": 0.9866,
      "step": 1937
    },
    {
      "epoch": 0.4168638416863842,
      "grad_norm": 0.0,
      "learning_rate": 1.3127254023973846e-05,
      "loss": 0.9479,
      "step": 1938
    },
    {
      "epoch": 0.41707894170789417,
      "grad_norm": 0.0,
      "learning_rate": 1.3120635342176171e-05,
      "loss": 0.9101,
      "step": 1939
    },
    {
      "epoch": 0.41729404172940415,
      "grad_norm": 0.0,
      "learning_rate": 1.311401514548438e-05,
      "loss": 0.8897,
      "step": 1940
    },
    {
      "epoch": 0.4175091417509142,
      "grad_norm": 0.0,
      "learning_rate": 1.3107393437112201e-05,
      "loss": 0.8602,
      "step": 1941
    },
    {
      "epoch": 0.4177242417724242,
      "grad_norm": 0.0,
      "learning_rate": 1.310077022027411e-05,
      "loss": 0.9395,
      "step": 1942
    },
    {
      "epoch": 0.41793934179393416,
      "grad_norm": 0.0,
      "learning_rate": 1.3094145498185305e-05,
      "loss": 0.8834,
      "step": 1943
    },
    {
      "epoch": 0.4181544418154442,
      "grad_norm": 0.0,
      "learning_rate": 1.3087519274061718e-05,
      "loss": 0.8549,
      "step": 1944
    },
    {
      "epoch": 0.4183695418369542,
      "grad_norm": 0.0,
      "learning_rate": 1.3080891551120014e-05,
      "loss": 0.9326,
      "step": 1945
    },
    {
      "epoch": 0.41858464185846417,
      "grad_norm": 0.0,
      "learning_rate": 1.3074262332577579e-05,
      "loss": 0.8703,
      "step": 1946
    },
    {
      "epoch": 0.4187997418799742,
      "grad_norm": 0.0,
      "learning_rate": 1.3067631621652525e-05,
      "loss": 0.8631,
      "step": 1947
    },
    {
      "epoch": 0.4190148419014842,
      "grad_norm": 0.0,
      "learning_rate": 1.30609994215637e-05,
      "loss": 0.8574,
      "step": 1948
    },
    {
      "epoch": 0.4192299419229942,
      "grad_norm": 0.0,
      "learning_rate": 1.3054365735530666e-05,
      "loss": 0.9482,
      "step": 1949
    },
    {
      "epoch": 0.4194450419445042,
      "grad_norm": 0.0,
      "learning_rate": 1.30477305667737e-05,
      "loss": 0.9064,
      "step": 1950
    },
    {
      "epoch": 0.4196601419660142,
      "grad_norm": 0.0,
      "learning_rate": 1.3041093918513812e-05,
      "loss": 0.9263,
      "step": 1951
    },
    {
      "epoch": 0.4198752419875242,
      "grad_norm": 0.0,
      "learning_rate": 1.3034455793972719e-05,
      "loss": 0.9309,
      "step": 1952
    },
    {
      "epoch": 0.4200903420090342,
      "grad_norm": 0.0,
      "learning_rate": 1.3027816196372862e-05,
      "loss": 0.9047,
      "step": 1953
    },
    {
      "epoch": 0.4203054420305442,
      "grad_norm": 0.0,
      "learning_rate": 1.3021175128937395e-05,
      "loss": 0.9666,
      "step": 1954
    },
    {
      "epoch": 0.4205205420520542,
      "grad_norm": 0.0,
      "learning_rate": 1.3014532594890185e-05,
      "loss": 0.9032,
      "step": 1955
    },
    {
      "epoch": 0.42073564207356423,
      "grad_norm": 0.0,
      "learning_rate": 1.3007888597455811e-05,
      "loss": 0.9053,
      "step": 1956
    },
    {
      "epoch": 0.4209507420950742,
      "grad_norm": 0.0,
      "learning_rate": 1.3001243139859562e-05,
      "loss": 0.8876,
      "step": 1957
    },
    {
      "epoch": 0.4211658421165842,
      "grad_norm": 0.0,
      "learning_rate": 1.2994596225327442e-05,
      "loss": 0.8634,
      "step": 1958
    },
    {
      "epoch": 0.42138094213809424,
      "grad_norm": 0.0,
      "learning_rate": 1.298794785708615e-05,
      "loss": 0.9313,
      "step": 1959
    },
    {
      "epoch": 0.4215960421596042,
      "grad_norm": 0.0,
      "learning_rate": 1.2981298038363095e-05,
      "loss": 0.8807,
      "step": 1960
    },
    {
      "epoch": 0.4218111421811142,
      "grad_norm": 0.0,
      "learning_rate": 1.2974646772386399e-05,
      "loss": 0.8325,
      "step": 1961
    },
    {
      "epoch": 0.42202624220262425,
      "grad_norm": 0.0,
      "learning_rate": 1.2967994062384878e-05,
      "loss": 0.8688,
      "step": 1962
    },
    {
      "epoch": 0.42224134222413423,
      "grad_norm": 0.0,
      "learning_rate": 1.2961339911588051e-05,
      "loss": 0.9461,
      "step": 1963
    },
    {
      "epoch": 0.4224564422456442,
      "grad_norm": 0.0,
      "learning_rate": 1.2954684323226137e-05,
      "loss": 0.9472,
      "step": 1964
    },
    {
      "epoch": 0.42267154226715425,
      "grad_norm": 0.0,
      "learning_rate": 1.294802730053005e-05,
      "loss": 0.9034,
      "step": 1965
    },
    {
      "epoch": 0.42288664228866424,
      "grad_norm": 0.0,
      "learning_rate": 1.2941368846731405e-05,
      "loss": 0.898,
      "step": 1966
    },
    {
      "epoch": 0.4231017423101742,
      "grad_norm": 0.0,
      "learning_rate": 1.2934708965062508e-05,
      "loss": 0.8332,
      "step": 1967
    },
    {
      "epoch": 0.4233168423316842,
      "grad_norm": 0.0,
      "learning_rate": 1.2928047658756362e-05,
      "loss": 0.9275,
      "step": 1968
    },
    {
      "epoch": 0.42353194235319425,
      "grad_norm": 0.0,
      "learning_rate": 1.2921384931046651e-05,
      "loss": 0.8734,
      "step": 1969
    },
    {
      "epoch": 0.42374704237470423,
      "grad_norm": 0.0,
      "learning_rate": 1.291472078516777e-05,
      "loss": 0.8414,
      "step": 1970
    },
    {
      "epoch": 0.4239621423962142,
      "grad_norm": 0.0,
      "learning_rate": 1.2908055224354778e-05,
      "loss": 0.9052,
      "step": 1971
    },
    {
      "epoch": 0.42417724241772425,
      "grad_norm": 0.0,
      "learning_rate": 1.2901388251843441e-05,
      "loss": 0.844,
      "step": 1972
    },
    {
      "epoch": 0.42439234243923424,
      "grad_norm": 0.0,
      "learning_rate": 1.2894719870870197e-05,
      "loss": 0.8645,
      "step": 1973
    },
    {
      "epoch": 0.4246074424607442,
      "grad_norm": 0.0,
      "learning_rate": 1.288805008467217e-05,
      "loss": 0.858,
      "step": 1974
    },
    {
      "epoch": 0.42482254248225426,
      "grad_norm": 0.0,
      "learning_rate": 1.2881378896487174e-05,
      "loss": 0.9439,
      "step": 1975
    },
    {
      "epoch": 0.42503764250376425,
      "grad_norm": 0.0,
      "learning_rate": 1.2874706309553697e-05,
      "loss": 0.9266,
      "step": 1976
    },
    {
      "epoch": 0.42525274252527423,
      "grad_norm": 0.0,
      "learning_rate": 1.2868032327110904e-05,
      "loss": 0.8704,
      "step": 1977
    },
    {
      "epoch": 0.42546784254678427,
      "grad_norm": 0.0,
      "learning_rate": 1.2861356952398643e-05,
      "loss": 0.9228,
      "step": 1978
    },
    {
      "epoch": 0.42568294256829425,
      "grad_norm": 0.0,
      "learning_rate": 1.2854680188657439e-05,
      "loss": 0.895,
      "step": 1979
    },
    {
      "epoch": 0.42589804258980424,
      "grad_norm": 0.0,
      "learning_rate": 1.284800203912848e-05,
      "loss": 0.8705,
      "step": 1980
    },
    {
      "epoch": 0.4261131426113143,
      "grad_norm": 0.0,
      "learning_rate": 1.284132250705364e-05,
      "loss": 0.9459,
      "step": 1981
    },
    {
      "epoch": 0.42632824263282426,
      "grad_norm": 0.0,
      "learning_rate": 1.2834641595675454e-05,
      "loss": 0.8987,
      "step": 1982
    },
    {
      "epoch": 0.42654334265433425,
      "grad_norm": 0.0,
      "learning_rate": 1.2827959308237137e-05,
      "loss": 0.9587,
      "step": 1983
    },
    {
      "epoch": 0.4267584426758443,
      "grad_norm": 0.0,
      "learning_rate": 1.2821275647982563e-05,
      "loss": 0.8533,
      "step": 1984
    },
    {
      "epoch": 0.42697354269735427,
      "grad_norm": 0.0,
      "learning_rate": 1.2814590618156278e-05,
      "loss": 0.9141,
      "step": 1985
    },
    {
      "epoch": 0.42718864271886425,
      "grad_norm": 0.0,
      "learning_rate": 1.2807904222003489e-05,
      "loss": 0.858,
      "step": 1986
    },
    {
      "epoch": 0.4274037427403743,
      "grad_norm": 0.0,
      "learning_rate": 1.2801216462770065e-05,
      "loss": 0.8622,
      "step": 1987
    },
    {
      "epoch": 0.4276188427618843,
      "grad_norm": 0.0,
      "learning_rate": 1.2794527343702546e-05,
      "loss": 0.9009,
      "step": 1988
    },
    {
      "epoch": 0.42783394278339426,
      "grad_norm": 0.0,
      "learning_rate": 1.2787836868048118e-05,
      "loss": 0.8341,
      "step": 1989
    },
    {
      "epoch": 0.4280490428049043,
      "grad_norm": 0.0,
      "learning_rate": 1.278114503905464e-05,
      "loss": 0.8735,
      "step": 1990
    },
    {
      "epoch": 0.4282641428264143,
      "grad_norm": 0.0,
      "learning_rate": 1.2774451859970614e-05,
      "loss": 0.9083,
      "step": 1991
    },
    {
      "epoch": 0.42847924284792427,
      "grad_norm": 0.0,
      "learning_rate": 1.2767757334045209e-05,
      "loss": 0.8446,
      "step": 1992
    },
    {
      "epoch": 0.4286943428694343,
      "grad_norm": 0.0,
      "learning_rate": 1.2761061464528246e-05,
      "loss": 0.9031,
      "step": 1993
    },
    {
      "epoch": 0.4289094428909443,
      "grad_norm": 0.0,
      "learning_rate": 1.2754364254670192e-05,
      "loss": 0.8702,
      "step": 1994
    },
    {
      "epoch": 0.4291245429124543,
      "grad_norm": 0.0,
      "learning_rate": 1.2747665707722168e-05,
      "loss": 0.9025,
      "step": 1995
    },
    {
      "epoch": 0.4293396429339643,
      "grad_norm": 0.0,
      "learning_rate": 1.2740965826935945e-05,
      "loss": 1.0105,
      "step": 1996
    },
    {
      "epoch": 0.4295547429554743,
      "grad_norm": 0.0,
      "learning_rate": 1.2734264615563941e-05,
      "loss": 0.8614,
      "step": 1997
    },
    {
      "epoch": 0.4297698429769843,
      "grad_norm": 0.0,
      "learning_rate": 1.2727562076859218e-05,
      "loss": 0.8942,
      "step": 1998
    },
    {
      "epoch": 0.4299849429984943,
      "grad_norm": 0.0,
      "learning_rate": 1.2720858214075486e-05,
      "loss": 0.8377,
      "step": 1999
    },
    {
      "epoch": 0.4302000430200043,
      "grad_norm": 0.0,
      "learning_rate": 1.2714153030467095e-05,
      "loss": 0.8911,
      "step": 2000
    },
    {
      "epoch": 0.4304151430415143,
      "grad_norm": 0.0,
      "learning_rate": 1.2707446529289043e-05,
      "loss": 0.8217,
      "step": 2001
    },
    {
      "epoch": 0.43063024306302433,
      "grad_norm": 0.0,
      "learning_rate": 1.2700738713796952e-05,
      "loss": 0.8554,
      "step": 2002
    },
    {
      "epoch": 0.4308453430845343,
      "grad_norm": 0.0,
      "learning_rate": 1.2694029587247097e-05,
      "loss": 0.9307,
      "step": 2003
    },
    {
      "epoch": 0.4310604431060443,
      "grad_norm": 0.0,
      "learning_rate": 1.2687319152896377e-05,
      "loss": 0.9314,
      "step": 2004
    },
    {
      "epoch": 0.43127554312755434,
      "grad_norm": 0.0,
      "learning_rate": 1.2680607414002346e-05,
      "loss": 0.8944,
      "step": 2005
    },
    {
      "epoch": 0.4314906431490643,
      "grad_norm": 0.0,
      "learning_rate": 1.2673894373823171e-05,
      "loss": 0.8735,
      "step": 2006
    },
    {
      "epoch": 0.4317057431705743,
      "grad_norm": 0.0,
      "learning_rate": 1.2667180035617657e-05,
      "loss": 0.8985,
      "step": 2007
    },
    {
      "epoch": 0.4319208431920843,
      "grad_norm": 0.0,
      "learning_rate": 1.2660464402645241e-05,
      "loss": 0.9021,
      "step": 2008
    },
    {
      "epoch": 0.43213594321359433,
      "grad_norm": 0.0,
      "learning_rate": 1.2653747478165988e-05,
      "loss": 0.8961,
      "step": 2009
    },
    {
      "epoch": 0.4323510432351043,
      "grad_norm": 0.0,
      "learning_rate": 1.2647029265440592e-05,
      "loss": 0.8643,
      "step": 2010
    },
    {
      "epoch": 0.4325661432566143,
      "grad_norm": 0.0,
      "learning_rate": 1.2640309767730363e-05,
      "loss": 0.878,
      "step": 2011
    },
    {
      "epoch": 0.43278124327812434,
      "grad_norm": 0.0,
      "learning_rate": 1.263358898829725e-05,
      "loss": 0.9013,
      "step": 2012
    },
    {
      "epoch": 0.4329963432996343,
      "grad_norm": 0.0,
      "learning_rate": 1.2626866930403808e-05,
      "loss": 0.887,
      "step": 2013
    },
    {
      "epoch": 0.4332114433211443,
      "grad_norm": 0.0,
      "learning_rate": 1.2620143597313229e-05,
      "loss": 0.9073,
      "step": 2014
    },
    {
      "epoch": 0.43342654334265435,
      "grad_norm": 0.0,
      "learning_rate": 1.2613418992289307e-05,
      "loss": 0.8922,
      "step": 2015
    },
    {
      "epoch": 0.43364164336416433,
      "grad_norm": 0.0,
      "learning_rate": 1.2606693118596474e-05,
      "loss": 0.8977,
      "step": 2016
    },
    {
      "epoch": 0.4338567433856743,
      "grad_norm": 0.0,
      "learning_rate": 1.2599965979499751e-05,
      "loss": 0.8869,
      "step": 2017
    },
    {
      "epoch": 0.43407184340718435,
      "grad_norm": 0.0,
      "learning_rate": 1.2593237578264802e-05,
      "loss": 0.857,
      "step": 2018
    },
    {
      "epoch": 0.43428694342869434,
      "grad_norm": 0.0,
      "learning_rate": 1.2586507918157878e-05,
      "loss": 0.8149,
      "step": 2019
    },
    {
      "epoch": 0.4345020434502043,
      "grad_norm": 0.0,
      "learning_rate": 1.2579777002445862e-05,
      "loss": 0.8296,
      "step": 2020
    },
    {
      "epoch": 0.43471714347171436,
      "grad_norm": 0.0,
      "learning_rate": 1.257304483439623e-05,
      "loss": 0.9026,
      "step": 2021
    },
    {
      "epoch": 0.43493224349322435,
      "grad_norm": 0.0,
      "learning_rate": 1.2566311417277082e-05,
      "loss": 0.8382,
      "step": 2022
    },
    {
      "epoch": 0.43514734351473433,
      "grad_norm": 0.0,
      "learning_rate": 1.2559576754357111e-05,
      "loss": 0.9148,
      "step": 2023
    },
    {
      "epoch": 0.43536244353624437,
      "grad_norm": 0.0,
      "learning_rate": 1.2552840848905619e-05,
      "loss": 0.9289,
      "step": 2024
    },
    {
      "epoch": 0.43557754355775435,
      "grad_norm": 0.0,
      "learning_rate": 1.2546103704192511e-05,
      "loss": 0.8939,
      "step": 2025
    },
    {
      "epoch": 0.43579264357926434,
      "grad_norm": 0.0,
      "learning_rate": 1.2539365323488298e-05,
      "loss": 0.8948,
      "step": 2026
    },
    {
      "epoch": 0.4360077436007744,
      "grad_norm": 0.0,
      "learning_rate": 1.2532625710064084e-05,
      "loss": 0.8556,
      "step": 2027
    },
    {
      "epoch": 0.43622284362228436,
      "grad_norm": 0.0,
      "learning_rate": 1.2525884867191579e-05,
      "loss": 0.8354,
      "step": 2028
    },
    {
      "epoch": 0.43643794364379435,
      "grad_norm": 0.0,
      "learning_rate": 1.251914279814308e-05,
      "loss": 0.8729,
      "step": 2029
    },
    {
      "epoch": 0.4366530436653044,
      "grad_norm": 0.0,
      "learning_rate": 1.251239950619149e-05,
      "loss": 0.8712,
      "step": 2030
    },
    {
      "epoch": 0.43686814368681437,
      "grad_norm": 0.0,
      "learning_rate": 1.2505654994610299e-05,
      "loss": 0.9243,
      "step": 2031
    },
    {
      "epoch": 0.43708324370832435,
      "grad_norm": 0.0,
      "learning_rate": 1.249890926667359e-05,
      "loss": 0.8773,
      "step": 2032
    },
    {
      "epoch": 0.4372983437298344,
      "grad_norm": 0.0,
      "learning_rate": 1.2492162325656035e-05,
      "loss": 0.893,
      "step": 2033
    },
    {
      "epoch": 0.4375134437513444,
      "grad_norm": 0.0,
      "learning_rate": 1.24854141748329e-05,
      "loss": 0.8242,
      "step": 2034
    },
    {
      "epoch": 0.43772854377285436,
      "grad_norm": 0.0,
      "learning_rate": 1.247866481748003e-05,
      "loss": 0.8699,
      "step": 2035
    },
    {
      "epoch": 0.4379436437943644,
      "grad_norm": 0.0,
      "learning_rate": 1.2471914256873871e-05,
      "loss": 0.9406,
      "step": 2036
    },
    {
      "epoch": 0.4381587438158744,
      "grad_norm": 0.0,
      "learning_rate": 1.2465162496291437e-05,
      "loss": 0.9143,
      "step": 2037
    },
    {
      "epoch": 0.43837384383738437,
      "grad_norm": 0.0,
      "learning_rate": 1.245840953901033e-05,
      "loss": 0.9109,
      "step": 2038
    },
    {
      "epoch": 0.4385889438588944,
      "grad_norm": 0.0,
      "learning_rate": 1.245165538830873e-05,
      "loss": 0.8434,
      "step": 2039
    },
    {
      "epoch": 0.4388040438804044,
      "grad_norm": 0.0,
      "learning_rate": 1.2444900047465409e-05,
      "loss": 0.8804,
      "step": 2040
    },
    {
      "epoch": 0.4390191439019144,
      "grad_norm": 0.0,
      "learning_rate": 1.2438143519759701e-05,
      "loss": 0.8827,
      "step": 2041
    },
    {
      "epoch": 0.4392342439234244,
      "grad_norm": 0.0,
      "learning_rate": 1.2431385808471523e-05,
      "loss": 0.8478,
      "step": 2042
    },
    {
      "epoch": 0.4394493439449344,
      "grad_norm": 0.0,
      "learning_rate": 1.242462691688137e-05,
      "loss": 0.9132,
      "step": 2043
    },
    {
      "epoch": 0.4396644439664444,
      "grad_norm": 0.0,
      "learning_rate": 1.2417866848270304e-05,
      "loss": 0.8799,
      "step": 2044
    },
    {
      "epoch": 0.4398795439879544,
      "grad_norm": 0.0,
      "learning_rate": 1.241110560591996e-05,
      "loss": 0.9045,
      "step": 2045
    },
    {
      "epoch": 0.4400946440094644,
      "grad_norm": 0.0,
      "learning_rate": 1.2404343193112543e-05,
      "loss": 0.9109,
      "step": 2046
    },
    {
      "epoch": 0.4403097440309744,
      "grad_norm": 0.0,
      "learning_rate": 1.2397579613130829e-05,
      "loss": 0.932,
      "step": 2047
    },
    {
      "epoch": 0.44052484405248443,
      "grad_norm": 0.0,
      "learning_rate": 1.2390814869258155e-05,
      "loss": 0.8809,
      "step": 2048
    },
    {
      "epoch": 0.4407399440739944,
      "grad_norm": 0.0,
      "learning_rate": 1.2384048964778428e-05,
      "loss": 0.863,
      "step": 2049
    },
    {
      "epoch": 0.4409550440955044,
      "grad_norm": 0.0,
      "learning_rate": 1.2377281902976116e-05,
      "loss": 0.8786,
      "step": 2050
    },
    {
      "epoch": 0.4411701441170144,
      "grad_norm": 0.0,
      "learning_rate": 1.2370513687136254e-05,
      "loss": 0.839,
      "step": 2051
    },
    {
      "epoch": 0.4413852441385244,
      "grad_norm": 0.0,
      "learning_rate": 1.2363744320544427e-05,
      "loss": 0.8945,
      "step": 2052
    },
    {
      "epoch": 0.4416003441600344,
      "grad_norm": 0.0,
      "learning_rate": 1.2356973806486788e-05,
      "loss": 0.8032,
      "step": 2053
    },
    {
      "epoch": 0.4418154441815444,
      "grad_norm": 0.0,
      "learning_rate": 1.2350202148250039e-05,
      "loss": 0.911,
      "step": 2054
    },
    {
      "epoch": 0.44203054420305443,
      "grad_norm": 0.0,
      "learning_rate": 1.2343429349121445e-05,
      "loss": 0.9078,
      "step": 2055
    },
    {
      "epoch": 0.4422456442245644,
      "grad_norm": 0.0,
      "learning_rate": 1.233665541238882e-05,
      "loss": 0.8117,
      "step": 2056
    },
    {
      "epoch": 0.4424607442460744,
      "grad_norm": 0.0,
      "learning_rate": 1.2329880341340533e-05,
      "loss": 0.9069,
      "step": 2057
    },
    {
      "epoch": 0.44267584426758444,
      "grad_norm": 0.0,
      "learning_rate": 1.2323104139265503e-05,
      "loss": 0.8582,
      "step": 2058
    },
    {
      "epoch": 0.4428909442890944,
      "grad_norm": 0.0,
      "learning_rate": 1.2316326809453201e-05,
      "loss": 0.9267,
      "step": 2059
    },
    {
      "epoch": 0.4431060443106044,
      "grad_norm": 0.0,
      "learning_rate": 1.2309548355193632e-05,
      "loss": 0.9155,
      "step": 2060
    },
    {
      "epoch": 0.44332114433211445,
      "grad_norm": 0.0,
      "learning_rate": 1.2302768779777363e-05,
      "loss": 0.8911,
      "step": 2061
    },
    {
      "epoch": 0.44353624435362443,
      "grad_norm": 0.0,
      "learning_rate": 1.22959880864955e-05,
      "loss": 0.9129,
      "step": 2062
    },
    {
      "epoch": 0.4437513443751344,
      "grad_norm": 0.0,
      "learning_rate": 1.228920627863969e-05,
      "loss": 0.9005,
      "step": 2063
    },
    {
      "epoch": 0.44396644439664446,
      "grad_norm": 0.0,
      "learning_rate": 1.228242335950212e-05,
      "loss": 0.9429,
      "step": 2064
    },
    {
      "epoch": 0.44418154441815444,
      "grad_norm": 0.0,
      "learning_rate": 1.2275639332375522e-05,
      "loss": 0.8289,
      "step": 2065
    },
    {
      "epoch": 0.4443966444396644,
      "grad_norm": 0.0,
      "learning_rate": 1.2268854200553158e-05,
      "loss": 0.8953,
      "step": 2066
    },
    {
      "epoch": 0.44461174446117446,
      "grad_norm": 0.0,
      "learning_rate": 1.2262067967328833e-05,
      "loss": 0.9239,
      "step": 2067
    },
    {
      "epoch": 0.44482684448268445,
      "grad_norm": 0.0,
      "learning_rate": 1.2255280635996882e-05,
      "loss": 0.9185,
      "step": 2068
    },
    {
      "epoch": 0.44504194450419443,
      "grad_norm": 0.0,
      "learning_rate": 1.2248492209852178e-05,
      "loss": 0.9006,
      "step": 2069
    },
    {
      "epoch": 0.44525704452570447,
      "grad_norm": 0.0,
      "learning_rate": 1.2241702692190119e-05,
      "loss": 0.8943,
      "step": 2070
    },
    {
      "epoch": 0.44547214454721445,
      "grad_norm": 0.0,
      "learning_rate": 1.223491208630664e-05,
      "loss": 0.8673,
      "step": 2071
    },
    {
      "epoch": 0.44568724456872444,
      "grad_norm": 0.0,
      "learning_rate": 1.2228120395498201e-05,
      "loss": 0.8943,
      "step": 2072
    },
    {
      "epoch": 0.4459023445902345,
      "grad_norm": 0.0,
      "learning_rate": 1.2221327623061784e-05,
      "loss": 0.8581,
      "step": 2073
    },
    {
      "epoch": 0.44611744461174446,
      "grad_norm": 0.0,
      "learning_rate": 1.2214533772294907e-05,
      "loss": 0.8938,
      "step": 2074
    },
    {
      "epoch": 0.44633254463325445,
      "grad_norm": 0.0,
      "learning_rate": 1.22077388464956e-05,
      "loss": 0.9212,
      "step": 2075
    },
    {
      "epoch": 0.4465476446547645,
      "grad_norm": 0.0,
      "learning_rate": 1.2200942848962422e-05,
      "loss": 0.8535,
      "step": 2076
    },
    {
      "epoch": 0.44676274467627447,
      "grad_norm": 0.0,
      "learning_rate": 1.2194145782994448e-05,
      "loss": 0.8934,
      "step": 2077
    },
    {
      "epoch": 0.44697784469778445,
      "grad_norm": 0.0,
      "learning_rate": 1.2187347651891277e-05,
      "loss": 0.9103,
      "step": 2078
    },
    {
      "epoch": 0.4471929447192945,
      "grad_norm": 0.0,
      "learning_rate": 1.218054845895302e-05,
      "loss": 0.9244,
      "step": 2079
    },
    {
      "epoch": 0.4474080447408045,
      "grad_norm": 0.0,
      "learning_rate": 1.2173748207480305e-05,
      "loss": 0.8724,
      "step": 2080
    },
    {
      "epoch": 0.44762314476231446,
      "grad_norm": 0.0,
      "learning_rate": 1.2166946900774275e-05,
      "loss": 0.934,
      "step": 2081
    },
    {
      "epoch": 0.4478382447838245,
      "grad_norm": 0.0,
      "learning_rate": 1.2160144542136585e-05,
      "loss": 0.8899,
      "step": 2082
    },
    {
      "epoch": 0.4480533448053345,
      "grad_norm": 0.0,
      "learning_rate": 1.2153341134869393e-05,
      "loss": 0.8439,
      "step": 2083
    },
    {
      "epoch": 0.44826844482684447,
      "grad_norm": 0.0,
      "learning_rate": 1.2146536682275388e-05,
      "loss": 0.8432,
      "step": 2084
    },
    {
      "epoch": 0.4484835448483545,
      "grad_norm": 0.0,
      "learning_rate": 1.2139731187657737e-05,
      "loss": 0.8284,
      "step": 2085
    },
    {
      "epoch": 0.4486986448698645,
      "grad_norm": 0.0,
      "learning_rate": 1.2132924654320137e-05,
      "loss": 0.8995,
      "step": 2086
    },
    {
      "epoch": 0.4489137448913745,
      "grad_norm": 0.0,
      "learning_rate": 1.2126117085566776e-05,
      "loss": 0.9141,
      "step": 2087
    },
    {
      "epoch": 0.4491288449128845,
      "grad_norm": 0.0,
      "learning_rate": 1.2119308484702345e-05,
      "loss": 0.8811,
      "step": 2088
    },
    {
      "epoch": 0.4493439449343945,
      "grad_norm": 0.0,
      "learning_rate": 1.2112498855032046e-05,
      "loss": 0.9046,
      "step": 2089
    },
    {
      "epoch": 0.4495590449559045,
      "grad_norm": 0.0,
      "learning_rate": 1.210568819986157e-05,
      "loss": 0.8966,
      "step": 2090
    },
    {
      "epoch": 0.4497741449774145,
      "grad_norm": 0.0,
      "learning_rate": 1.209887652249711e-05,
      "loss": 0.9485,
      "step": 2091
    },
    {
      "epoch": 0.4499892449989245,
      "grad_norm": 0.0,
      "learning_rate": 1.2092063826245352e-05,
      "loss": 0.8954,
      "step": 2092
    },
    {
      "epoch": 0.4502043450204345,
      "grad_norm": 0.0,
      "learning_rate": 1.2085250114413484e-05,
      "loss": 0.9482,
      "step": 2093
    },
    {
      "epoch": 0.4504194450419445,
      "grad_norm": 0.0,
      "learning_rate": 1.2078435390309182e-05,
      "loss": 0.8764,
      "step": 2094
    },
    {
      "epoch": 0.4506345450634545,
      "grad_norm": 0.0,
      "learning_rate": 1.2071619657240614e-05,
      "loss": 0.8702,
      "step": 2095
    },
    {
      "epoch": 0.4508496450849645,
      "grad_norm": 0.0,
      "learning_rate": 1.2064802918516433e-05,
      "loss": 0.9547,
      "step": 2096
    },
    {
      "epoch": 0.4510647451064745,
      "grad_norm": 0.0,
      "learning_rate": 1.2057985177445792e-05,
      "loss": 0.8725,
      "step": 2097
    },
    {
      "epoch": 0.4512798451279845,
      "grad_norm": 0.0,
      "learning_rate": 1.2051166437338318e-05,
      "loss": 0.8611,
      "step": 2098
    },
    {
      "epoch": 0.4514949451494945,
      "grad_norm": 0.0,
      "learning_rate": 1.2044346701504129e-05,
      "loss": 0.8809,
      "step": 2099
    },
    {
      "epoch": 0.4517100451710045,
      "grad_norm": 0.0,
      "learning_rate": 1.2037525973253826e-05,
      "loss": 0.8596,
      "step": 2100
    },
    {
      "epoch": 0.45192514519251453,
      "grad_norm": 0.0,
      "learning_rate": 1.2030704255898492e-05,
      "loss": 0.872,
      "step": 2101
    },
    {
      "epoch": 0.4521402452140245,
      "grad_norm": 0.0,
      "learning_rate": 1.2023881552749693e-05,
      "loss": 0.9014,
      "step": 2102
    },
    {
      "epoch": 0.4523553452355345,
      "grad_norm": 0.0,
      "learning_rate": 1.2017057867119464e-05,
      "loss": 0.8686,
      "step": 2103
    },
    {
      "epoch": 0.45257044525704454,
      "grad_norm": 0.0,
      "learning_rate": 1.2010233202320327e-05,
      "loss": 0.8744,
      "step": 2104
    },
    {
      "epoch": 0.4527855452785545,
      "grad_norm": 0.0,
      "learning_rate": 1.2003407561665272e-05,
      "loss": 0.8556,
      "step": 2105
    },
    {
      "epoch": 0.4530006453000645,
      "grad_norm": 0.0,
      "learning_rate": 1.1996580948467773e-05,
      "loss": 0.8656,
      "step": 2106
    },
    {
      "epoch": 0.45321574532157455,
      "grad_norm": 0.0,
      "learning_rate": 1.1989753366041763e-05,
      "loss": 0.8919,
      "step": 2107
    },
    {
      "epoch": 0.45343084534308453,
      "grad_norm": 0.0,
      "learning_rate": 1.1982924817701656e-05,
      "loss": 0.8446,
      "step": 2108
    },
    {
      "epoch": 0.4536459453645945,
      "grad_norm": 0.0,
      "learning_rate": 1.1976095306762328e-05,
      "loss": 0.9105,
      "step": 2109
    },
    {
      "epoch": 0.45386104538610456,
      "grad_norm": 0.0,
      "learning_rate": 1.1969264836539124e-05,
      "loss": 0.8933,
      "step": 2110
    },
    {
      "epoch": 0.45407614540761454,
      "grad_norm": 0.0,
      "learning_rate": 1.196243341034786e-05,
      "loss": 0.9292,
      "step": 2111
    },
    {
      "epoch": 0.4542912454291245,
      "grad_norm": 0.0,
      "learning_rate": 1.1955601031504808e-05,
      "loss": 0.8886,
      "step": 2112
    },
    {
      "epoch": 0.45450634545063456,
      "grad_norm": 0.0,
      "learning_rate": 1.194876770332671e-05,
      "loss": 0.7866,
      "step": 2113
    },
    {
      "epoch": 0.45472144547214455,
      "grad_norm": 0.0,
      "learning_rate": 1.1941933429130759e-05,
      "loss": 0.8492,
      "step": 2114
    },
    {
      "epoch": 0.45493654549365453,
      "grad_norm": 0.0,
      "learning_rate": 1.193509821223462e-05,
      "loss": 0.9262,
      "step": 2115
    },
    {
      "epoch": 0.45515164551516457,
      "grad_norm": 0.0,
      "learning_rate": 1.1928262055956403e-05,
      "loss": 0.8522,
      "step": 2116
    },
    {
      "epoch": 0.45536674553667456,
      "grad_norm": 0.0,
      "learning_rate": 1.1921424963614686e-05,
      "loss": 0.8842,
      "step": 2117
    },
    {
      "epoch": 0.45558184555818454,
      "grad_norm": 0.0,
      "learning_rate": 1.191458693852849e-05,
      "loss": 0.8609,
      "step": 2118
    },
    {
      "epoch": 0.4557969455796946,
      "grad_norm": 0.0,
      "learning_rate": 1.1907747984017295e-05,
      "loss": 0.9428,
      "step": 2119
    },
    {
      "epoch": 0.45601204560120456,
      "grad_norm": 0.0,
      "learning_rate": 1.1900908103401037e-05,
      "loss": 0.843,
      "step": 2120
    },
    {
      "epoch": 0.45622714562271455,
      "grad_norm": 0.0,
      "learning_rate": 1.1894067300000088e-05,
      "loss": 0.8847,
      "step": 2121
    },
    {
      "epoch": 0.4564422456442246,
      "grad_norm": 0.0,
      "learning_rate": 1.1887225577135279e-05,
      "loss": 0.935,
      "step": 2122
    },
    {
      "epoch": 0.45665734566573457,
      "grad_norm": 0.0,
      "learning_rate": 1.1880382938127886e-05,
      "loss": 0.959,
      "step": 2123
    },
    {
      "epoch": 0.45687244568724455,
      "grad_norm": 0.0,
      "learning_rate": 1.1873539386299629e-05,
      "loss": 0.8504,
      "step": 2124
    },
    {
      "epoch": 0.4570875457087546,
      "grad_norm": 0.0,
      "learning_rate": 1.1866694924972667e-05,
      "loss": 0.8217,
      "step": 2125
    },
    {
      "epoch": 0.4573026457302646,
      "grad_norm": 0.0,
      "learning_rate": 1.1859849557469603e-05,
      "loss": 0.8707,
      "step": 2126
    },
    {
      "epoch": 0.45751774575177456,
      "grad_norm": 0.0,
      "learning_rate": 1.185300328711348e-05,
      "loss": 0.8451,
      "step": 2127
    },
    {
      "epoch": 0.4577328457732846,
      "grad_norm": 0.0,
      "learning_rate": 1.1846156117227783e-05,
      "loss": 0.8761,
      "step": 2128
    },
    {
      "epoch": 0.4579479457947946,
      "grad_norm": 0.0,
      "learning_rate": 1.1839308051136432e-05,
      "loss": 0.8715,
      "step": 2129
    },
    {
      "epoch": 0.45816304581630457,
      "grad_norm": 0.0,
      "learning_rate": 1.1832459092163775e-05,
      "loss": 0.8835,
      "step": 2130
    },
    {
      "epoch": 0.4583781458378146,
      "grad_norm": 0.0,
      "learning_rate": 1.1825609243634601e-05,
      "loss": 0.8821,
      "step": 2131
    },
    {
      "epoch": 0.4585932458593246,
      "grad_norm": 0.0,
      "learning_rate": 1.1818758508874133e-05,
      "loss": 0.8317,
      "step": 2132
    },
    {
      "epoch": 0.4588083458808346,
      "grad_norm": 0.0,
      "learning_rate": 1.1811906891208014e-05,
      "loss": 0.8879,
      "step": 2133
    },
    {
      "epoch": 0.4590234459023446,
      "grad_norm": 0.0,
      "learning_rate": 1.1805054393962323e-05,
      "loss": 0.8923,
      "step": 2134
    },
    {
      "epoch": 0.4592385459238546,
      "grad_norm": 0.0,
      "learning_rate": 1.1798201020463567e-05,
      "loss": 0.8637,
      "step": 2135
    },
    {
      "epoch": 0.4594536459453646,
      "grad_norm": 0.0,
      "learning_rate": 1.1791346774038673e-05,
      "loss": 0.885,
      "step": 2136
    },
    {
      "epoch": 0.45966874596687457,
      "grad_norm": 0.0,
      "learning_rate": 1.1784491658014996e-05,
      "loss": 0.8509,
      "step": 2137
    },
    {
      "epoch": 0.4598838459883846,
      "grad_norm": 0.0,
      "learning_rate": 1.1777635675720313e-05,
      "loss": 0.8679,
      "step": 2138
    },
    {
      "epoch": 0.4600989460098946,
      "grad_norm": 0.0,
      "learning_rate": 1.1770778830482824e-05,
      "loss": 0.9006,
      "step": 2139
    },
    {
      "epoch": 0.4603140460314046,
      "grad_norm": 0.0,
      "learning_rate": 1.1763921125631132e-05,
      "loss": 0.8925,
      "step": 2140
    },
    {
      "epoch": 0.4605291460529146,
      "grad_norm": 0.0,
      "learning_rate": 1.1757062564494282e-05,
      "loss": 0.868,
      "step": 2141
    },
    {
      "epoch": 0.4607442460744246,
      "grad_norm": 0.0,
      "learning_rate": 1.1750203150401718e-05,
      "loss": 0.8293,
      "step": 2142
    },
    {
      "epoch": 0.4609593460959346,
      "grad_norm": 0.0,
      "learning_rate": 1.17433428866833e-05,
      "loss": 0.8904,
      "step": 2143
    },
    {
      "epoch": 0.4611744461174446,
      "grad_norm": 0.0,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.8626,
      "step": 2144
    },
    {
      "epoch": 0.4613895461389546,
      "grad_norm": 0.0,
      "learning_rate": 1.1729619823690419e-05,
      "loss": 0.9423,
      "step": 2145
    },
    {
      "epoch": 0.4616046461604646,
      "grad_norm": 0.0,
      "learning_rate": 1.1722757031077737e-05,
      "loss": 0.8748,
      "step": 2146
    },
    {
      "epoch": 0.46181974618197463,
      "grad_norm": 0.0,
      "learning_rate": 1.171589340216276e-05,
      "loss": 0.8545,
      "step": 2147
    },
    {
      "epoch": 0.4620348462034846,
      "grad_norm": 0.0,
      "learning_rate": 1.1709028940277396e-05,
      "loss": 1.0443,
      "step": 2148
    },
    {
      "epoch": 0.4622499462249946,
      "grad_norm": 0.0,
      "learning_rate": 1.1702163648753957e-05,
      "loss": 0.8402,
      "step": 2149
    },
    {
      "epoch": 0.46246504624650464,
      "grad_norm": 0.0,
      "learning_rate": 1.1695297530925162e-05,
      "loss": 0.8344,
      "step": 2150
    },
    {
      "epoch": 0.4626801462680146,
      "grad_norm": 0.0,
      "learning_rate": 1.1688430590124128e-05,
      "loss": 0.8912,
      "step": 2151
    },
    {
      "epoch": 0.4628952462895246,
      "grad_norm": 0.0,
      "learning_rate": 1.1681562829684365e-05,
      "loss": 0.8248,
      "step": 2152
    },
    {
      "epoch": 0.46311034631103465,
      "grad_norm": 0.0,
      "learning_rate": 1.1674694252939795e-05,
      "loss": 0.9114,
      "step": 2153
    },
    {
      "epoch": 0.46332544633254463,
      "grad_norm": 0.0,
      "learning_rate": 1.1667824863224722e-05,
      "loss": 0.8476,
      "step": 2154
    },
    {
      "epoch": 0.4635405463540546,
      "grad_norm": 0.0,
      "learning_rate": 1.1660954663873856e-05,
      "loss": 0.8927,
      "step": 2155
    },
    {
      "epoch": 0.46375564637556466,
      "grad_norm": 0.0,
      "learning_rate": 1.1654083658222287e-05,
      "loss": 0.8641,
      "step": 2156
    },
    {
      "epoch": 0.46397074639707464,
      "grad_norm": 0.0,
      "learning_rate": 1.1647211849605516e-05,
      "loss": 0.838,
      "step": 2157
    },
    {
      "epoch": 0.4641858464185846,
      "grad_norm": 0.0,
      "learning_rate": 1.1640339241359408e-05,
      "loss": 0.8505,
      "step": 2158
    },
    {
      "epoch": 0.46440094644009466,
      "grad_norm": 0.0,
      "learning_rate": 1.1633465836820243e-05,
      "loss": 0.8776,
      "step": 2159
    },
    {
      "epoch": 0.46461604646160465,
      "grad_norm": 0.0,
      "learning_rate": 1.1626591639324673e-05,
      "loss": 0.8378,
      "step": 2160
    },
    {
      "epoch": 0.46483114648311463,
      "grad_norm": 0.0,
      "learning_rate": 1.1619716652209728e-05,
      "loss": 0.9603,
      "step": 2161
    },
    {
      "epoch": 0.46504624650462467,
      "grad_norm": 0.0,
      "learning_rate": 1.1612840878812839e-05,
      "loss": 0.9032,
      "step": 2162
    },
    {
      "epoch": 0.46526134652613466,
      "grad_norm": 0.0,
      "learning_rate": 1.1605964322471806e-05,
      "loss": 0.8626,
      "step": 2163
    },
    {
      "epoch": 0.46547644654764464,
      "grad_norm": 0.0,
      "learning_rate": 1.1599086986524815e-05,
      "loss": 0.8337,
      "step": 2164
    },
    {
      "epoch": 0.4656915465691547,
      "grad_norm": 0.0,
      "learning_rate": 1.1592208874310426e-05,
      "loss": 0.8288,
      "step": 2165
    },
    {
      "epoch": 0.46590664659066466,
      "grad_norm": 0.0,
      "learning_rate": 1.158532998916758e-05,
      "loss": 0.8519,
      "step": 2166
    },
    {
      "epoch": 0.46612174661217465,
      "grad_norm": 0.0,
      "learning_rate": 1.1578450334435592e-05,
      "loss": 0.9159,
      "step": 2167
    },
    {
      "epoch": 0.4663368466336847,
      "grad_norm": 0.0,
      "learning_rate": 1.1571569913454145e-05,
      "loss": 0.8848,
      "step": 2168
    },
    {
      "epoch": 0.46655194665519467,
      "grad_norm": 0.0,
      "learning_rate": 1.1564688729563306e-05,
      "loss": 0.828,
      "step": 2169
    },
    {
      "epoch": 0.46676704667670466,
      "grad_norm": 0.0,
      "learning_rate": 1.1557806786103502e-05,
      "loss": 0.8688,
      "step": 2170
    },
    {
      "epoch": 0.4669821466982147,
      "grad_norm": 0.0,
      "learning_rate": 1.1550924086415532e-05,
      "loss": 0.8533,
      "step": 2171
    },
    {
      "epoch": 0.4671972467197247,
      "grad_norm": 0.0,
      "learning_rate": 1.1544040633840563e-05,
      "loss": 0.9031,
      "step": 2172
    },
    {
      "epoch": 0.46741234674123466,
      "grad_norm": 0.0,
      "learning_rate": 1.1537156431720125e-05,
      "loss": 0.9393,
      "step": 2173
    },
    {
      "epoch": 0.4676274467627447,
      "grad_norm": 0.0,
      "learning_rate": 1.1530271483396115e-05,
      "loss": 0.8975,
      "step": 2174
    },
    {
      "epoch": 0.4678425467842547,
      "grad_norm": 0.0,
      "learning_rate": 1.1523385792210792e-05,
      "loss": 0.8594,
      "step": 2175
    },
    {
      "epoch": 0.46805764680576467,
      "grad_norm": 0.0,
      "learning_rate": 1.1516499361506772e-05,
      "loss": 0.8952,
      "step": 2176
    },
    {
      "epoch": 0.46827274682727466,
      "grad_norm": 0.0,
      "learning_rate": 1.1509612194627033e-05,
      "loss": 0.8499,
      "step": 2177
    },
    {
      "epoch": 0.4684878468487847,
      "grad_norm": 0.0,
      "learning_rate": 1.150272429491491e-05,
      "loss": 0.8694,
      "step": 2178
    },
    {
      "epoch": 0.4687029468702947,
      "grad_norm": 0.0,
      "learning_rate": 1.1495835665714096e-05,
      "loss": 0.8989,
      "step": 2179
    },
    {
      "epoch": 0.46891804689180466,
      "grad_norm": 0.0,
      "learning_rate": 1.1488946310368629e-05,
      "loss": 0.8317,
      "step": 2180
    },
    {
      "epoch": 0.4691331469133147,
      "grad_norm": 0.0,
      "learning_rate": 1.1482056232222913e-05,
      "loss": 0.9092,
      "step": 2181
    },
    {
      "epoch": 0.4693482469348247,
      "grad_norm": 0.0,
      "learning_rate": 1.1475165434621696e-05,
      "loss": 0.8507,
      "step": 2182
    },
    {
      "epoch": 0.46956334695633467,
      "grad_norm": 0.0,
      "learning_rate": 1.146827392091007e-05,
      "loss": 0.878,
      "step": 2183
    },
    {
      "epoch": 0.4697784469778447,
      "grad_norm": 0.0,
      "learning_rate": 1.146138169443348e-05,
      "loss": 0.9397,
      "step": 2184
    },
    {
      "epoch": 0.4699935469993547,
      "grad_norm": 0.0,
      "learning_rate": 1.145448875853772e-05,
      "loss": 0.8657,
      "step": 2185
    },
    {
      "epoch": 0.4702086470208647,
      "grad_norm": 0.0,
      "learning_rate": 1.1447595116568928e-05,
      "loss": 0.9339,
      "step": 2186
    },
    {
      "epoch": 0.4704237470423747,
      "grad_norm": 0.0,
      "learning_rate": 1.1440700771873576e-05,
      "loss": 0.9026,
      "step": 2187
    },
    {
      "epoch": 0.4706388470638847,
      "grad_norm": 0.0,
      "learning_rate": 1.1433805727798488e-05,
      "loss": 0.918,
      "step": 2188
    },
    {
      "epoch": 0.4708539470853947,
      "grad_norm": 0.0,
      "learning_rate": 1.142690998769082e-05,
      "loss": 0.8799,
      "step": 2189
    },
    {
      "epoch": 0.4710690471069047,
      "grad_norm": 0.0,
      "learning_rate": 1.1420013554898068e-05,
      "loss": 0.8194,
      "step": 2190
    },
    {
      "epoch": 0.4712841471284147,
      "grad_norm": 0.0,
      "learning_rate": 1.1413116432768067e-05,
      "loss": 0.9264,
      "step": 2191
    },
    {
      "epoch": 0.4714992471499247,
      "grad_norm": 0.0,
      "learning_rate": 1.1406218624648986e-05,
      "loss": 0.8475,
      "step": 2192
    },
    {
      "epoch": 0.47171434717143473,
      "grad_norm": 0.0,
      "learning_rate": 1.1399320133889318e-05,
      "loss": 0.8545,
      "step": 2193
    },
    {
      "epoch": 0.4719294471929447,
      "grad_norm": 0.0,
      "learning_rate": 1.1392420963837902e-05,
      "loss": 0.9143,
      "step": 2194
    },
    {
      "epoch": 0.4721445472144547,
      "grad_norm": 0.0,
      "learning_rate": 1.1385521117843898e-05,
      "loss": 0.8584,
      "step": 2195
    },
    {
      "epoch": 0.47235964723596474,
      "grad_norm": 0.0,
      "learning_rate": 1.1378620599256797e-05,
      "loss": 0.8303,
      "step": 2196
    },
    {
      "epoch": 0.4725747472574747,
      "grad_norm": 0.0,
      "learning_rate": 1.1371719411426412e-05,
      "loss": 0.9243,
      "step": 2197
    },
    {
      "epoch": 0.4727898472789847,
      "grad_norm": 0.0,
      "learning_rate": 1.1364817557702886e-05,
      "loss": 0.8806,
      "step": 2198
    },
    {
      "epoch": 0.47300494730049475,
      "grad_norm": 0.0,
      "learning_rate": 1.1357915041436685e-05,
      "loss": 0.8115,
      "step": 2199
    },
    {
      "epoch": 0.47322004732200473,
      "grad_norm": 0.0,
      "learning_rate": 1.135101186597859e-05,
      "loss": 0.8922,
      "step": 2200
    },
    {
      "epoch": 0.4734351473435147,
      "grad_norm": 0.0,
      "learning_rate": 1.134410803467971e-05,
      "loss": 0.8904,
      "step": 2201
    },
    {
      "epoch": 0.47365024736502476,
      "grad_norm": 0.0,
      "learning_rate": 1.133720355089147e-05,
      "loss": 0.8229,
      "step": 2202
    },
    {
      "epoch": 0.47386534738653474,
      "grad_norm": 0.0,
      "learning_rate": 1.1330298417965614e-05,
      "loss": 0.8808,
      "step": 2203
    },
    {
      "epoch": 0.4740804474080447,
      "grad_norm": 0.0,
      "learning_rate": 1.1323392639254194e-05,
      "loss": 0.9052,
      "step": 2204
    },
    {
      "epoch": 0.47429554742955476,
      "grad_norm": 0.0,
      "learning_rate": 1.1316486218109579e-05,
      "loss": 0.9364,
      "step": 2205
    },
    {
      "epoch": 0.47451064745106475,
      "grad_norm": 0.0,
      "learning_rate": 1.1309579157884451e-05,
      "loss": 0.895,
      "step": 2206
    },
    {
      "epoch": 0.47472574747257473,
      "grad_norm": 0.0,
      "learning_rate": 1.1302671461931805e-05,
      "loss": 0.907,
      "step": 2207
    },
    {
      "epoch": 0.4749408474940848,
      "grad_norm": 0.0,
      "learning_rate": 1.129576313360494e-05,
      "loss": 0.8734,
      "step": 2208
    },
    {
      "epoch": 0.47515594751559476,
      "grad_norm": 0.0,
      "learning_rate": 1.1288854176257464e-05,
      "loss": 0.8928,
      "step": 2209
    },
    {
      "epoch": 0.47537104753710474,
      "grad_norm": 0.0,
      "learning_rate": 1.1281944593243289e-05,
      "loss": 0.8854,
      "step": 2210
    },
    {
      "epoch": 0.4755861475586148,
      "grad_norm": 0.0,
      "learning_rate": 1.127503438791663e-05,
      "loss": 0.8162,
      "step": 2211
    },
    {
      "epoch": 0.47580124758012476,
      "grad_norm": 0.0,
      "learning_rate": 1.1268123563632007e-05,
      "loss": 0.9023,
      "step": 2212
    },
    {
      "epoch": 0.47601634760163475,
      "grad_norm": 0.0,
      "learning_rate": 1.126121212374424e-05,
      "loss": 0.945,
      "step": 2213
    },
    {
      "epoch": 0.4762314476231448,
      "grad_norm": 0.0,
      "learning_rate": 1.1254300071608448e-05,
      "loss": 0.8933,
      "step": 2214
    },
    {
      "epoch": 0.47644654764465477,
      "grad_norm": 0.0,
      "learning_rate": 1.1247387410580042e-05,
      "loss": 0.9003,
      "step": 2215
    },
    {
      "epoch": 0.47666164766616476,
      "grad_norm": 0.0,
      "learning_rate": 1.1240474144014735e-05,
      "loss": 0.8463,
      "step": 2216
    },
    {
      "epoch": 0.4768767476876748,
      "grad_norm": 0.0,
      "learning_rate": 1.1233560275268534e-05,
      "loss": 0.8618,
      "step": 2217
    },
    {
      "epoch": 0.4770918477091848,
      "grad_norm": 0.0,
      "learning_rate": 1.1226645807697737e-05,
      "loss": 0.8648,
      "step": 2218
    },
    {
      "epoch": 0.47730694773069476,
      "grad_norm": 0.0,
      "learning_rate": 1.1219730744658921e-05,
      "loss": 0.8221,
      "step": 2219
    },
    {
      "epoch": 0.47752204775220475,
      "grad_norm": 0.0,
      "learning_rate": 1.1212815089508973e-05,
      "loss": 0.8839,
      "step": 2220
    },
    {
      "epoch": 0.4777371477737148,
      "grad_norm": 0.0,
      "learning_rate": 1.1205898845605056e-05,
      "loss": 0.8595,
      "step": 2221
    },
    {
      "epoch": 0.47795224779522477,
      "grad_norm": 0.0,
      "learning_rate": 1.1198982016304618e-05,
      "loss": 0.9479,
      "step": 2222
    },
    {
      "epoch": 0.47816734781673476,
      "grad_norm": 0.0,
      "learning_rate": 1.1192064604965396e-05,
      "loss": 0.8804,
      "step": 2223
    },
    {
      "epoch": 0.4783824478382448,
      "grad_norm": 0.0,
      "learning_rate": 1.11851466149454e-05,
      "loss": 0.8391,
      "step": 2224
    },
    {
      "epoch": 0.4785975478597548,
      "grad_norm": 0.0,
      "learning_rate": 1.1178228049602938e-05,
      "loss": 0.8551,
      "step": 2225
    },
    {
      "epoch": 0.47881264788126476,
      "grad_norm": 0.0,
      "learning_rate": 1.1171308912296576e-05,
      "loss": 0.9221,
      "step": 2226
    },
    {
      "epoch": 0.4790277479027748,
      "grad_norm": 0.0,
      "learning_rate": 1.1164389206385177e-05,
      "loss": 0.8239,
      "step": 2227
    },
    {
      "epoch": 0.4792428479242848,
      "grad_norm": 0.0,
      "learning_rate": 1.1157468935227866e-05,
      "loss": 0.8526,
      "step": 2228
    },
    {
      "epoch": 0.47945794794579477,
      "grad_norm": 0.0,
      "learning_rate": 1.1150548102184051e-05,
      "loss": 0.8904,
      "step": 2229
    },
    {
      "epoch": 0.4796730479673048,
      "grad_norm": 0.0,
      "learning_rate": 1.1143626710613411e-05,
      "loss": 0.8323,
      "step": 2230
    },
    {
      "epoch": 0.4798881479888148,
      "grad_norm": 0.0,
      "learning_rate": 1.1136704763875895e-05,
      "loss": 0.8685,
      "step": 2231
    },
    {
      "epoch": 0.4801032480103248,
      "grad_norm": 0.0,
      "learning_rate": 1.1129782265331716e-05,
      "loss": 0.8913,
      "step": 2232
    },
    {
      "epoch": 0.4803183480318348,
      "grad_norm": 0.0,
      "learning_rate": 1.1122859218341366e-05,
      "loss": 0.8004,
      "step": 2233
    },
    {
      "epoch": 0.4805334480533448,
      "grad_norm": 0.0,
      "learning_rate": 1.1115935626265596e-05,
      "loss": 0.9043,
      "step": 2234
    },
    {
      "epoch": 0.4807485480748548,
      "grad_norm": 0.0,
      "learning_rate": 1.110901149246542e-05,
      "loss": 0.8763,
      "step": 2235
    },
    {
      "epoch": 0.4809636480963648,
      "grad_norm": 0.0,
      "learning_rate": 1.1102086820302122e-05,
      "loss": 0.9019,
      "step": 2236
    },
    {
      "epoch": 0.4811787481178748,
      "grad_norm": 0.0,
      "learning_rate": 1.1095161613137242e-05,
      "loss": 0.8568,
      "step": 2237
    },
    {
      "epoch": 0.4813938481393848,
      "grad_norm": 0.0,
      "learning_rate": 1.1088235874332583e-05,
      "loss": 0.9326,
      "step": 2238
    },
    {
      "epoch": 0.48160894816089483,
      "grad_norm": 0.0,
      "learning_rate": 1.1081309607250202e-05,
      "loss": 0.8604,
      "step": 2239
    },
    {
      "epoch": 0.4818240481824048,
      "grad_norm": 0.0,
      "learning_rate": 1.1074382815252418e-05,
      "loss": 0.8528,
      "step": 2240
    },
    {
      "epoch": 0.4820391482039148,
      "grad_norm": 0.0,
      "learning_rate": 1.1067455501701794e-05,
      "loss": 0.8873,
      "step": 2241
    },
    {
      "epoch": 0.48225424822542484,
      "grad_norm": 0.0,
      "learning_rate": 1.1060527669961163e-05,
      "loss": 0.8507,
      "step": 2242
    },
    {
      "epoch": 0.4824693482469348,
      "grad_norm": 0.0,
      "learning_rate": 1.1053599323393594e-05,
      "loss": 0.8448,
      "step": 2243
    },
    {
      "epoch": 0.4826844482684448,
      "grad_norm": 0.0,
      "learning_rate": 1.1046670465362415e-05,
      "loss": 0.9194,
      "step": 2244
    },
    {
      "epoch": 0.48289954828995485,
      "grad_norm": 0.0,
      "learning_rate": 1.1039741099231198e-05,
      "loss": 0.8994,
      "step": 2245
    },
    {
      "epoch": 0.48311464831146483,
      "grad_norm": 0.0,
      "learning_rate": 1.1032811228363766e-05,
      "loss": 0.9294,
      "step": 2246
    },
    {
      "epoch": 0.4833297483329748,
      "grad_norm": 0.0,
      "learning_rate": 1.102588085612418e-05,
      "loss": 0.8856,
      "step": 2247
    },
    {
      "epoch": 0.48354484835448486,
      "grad_norm": 0.0,
      "learning_rate": 1.1018949985876753e-05,
      "loss": 0.8836,
      "step": 2248
    },
    {
      "epoch": 0.48375994837599484,
      "grad_norm": 0.0,
      "learning_rate": 1.1012018620986028e-05,
      "loss": 0.8849,
      "step": 2249
    },
    {
      "epoch": 0.4839750483975048,
      "grad_norm": 0.0,
      "learning_rate": 1.1005086764816803e-05,
      "loss": 0.9167,
      "step": 2250
    },
    {
      "epoch": 0.48419014841901487,
      "grad_norm": 0.0,
      "learning_rate": 1.0998154420734104e-05,
      "loss": 0.9103,
      "step": 2251
    },
    {
      "epoch": 0.48440524844052485,
      "grad_norm": 0.0,
      "learning_rate": 1.0991221592103198e-05,
      "loss": 0.9301,
      "step": 2252
    },
    {
      "epoch": 0.48462034846203483,
      "grad_norm": 0.0,
      "learning_rate": 1.0984288282289586e-05,
      "loss": 0.8033,
      "step": 2253
    },
    {
      "epoch": 0.4848354484835449,
      "grad_norm": 0.0,
      "learning_rate": 1.0977354494659e-05,
      "loss": 0.8578,
      "step": 2254
    },
    {
      "epoch": 0.48505054850505486,
      "grad_norm": 0.0,
      "learning_rate": 1.0970420232577408e-05,
      "loss": 0.7873,
      "step": 2255
    },
    {
      "epoch": 0.48526564852656484,
      "grad_norm": 0.0,
      "learning_rate": 1.0963485499411007e-05,
      "loss": 0.8104,
      "step": 2256
    },
    {
      "epoch": 0.4854807485480749,
      "grad_norm": 0.0,
      "learning_rate": 1.0956550298526223e-05,
      "loss": 0.8278,
      "step": 2257
    },
    {
      "epoch": 0.48569584856958486,
      "grad_norm": 0.0,
      "learning_rate": 1.0949614633289708e-05,
      "loss": 0.8232,
      "step": 2258
    },
    {
      "epoch": 0.48591094859109485,
      "grad_norm": 0.0,
      "learning_rate": 1.0942678507068338e-05,
      "loss": 0.8406,
      "step": 2259
    },
    {
      "epoch": 0.4861260486126049,
      "grad_norm": 0.0,
      "learning_rate": 1.0935741923229219e-05,
      "loss": 0.893,
      "step": 2260
    },
    {
      "epoch": 0.4863411486341149,
      "grad_norm": 0.0,
      "learning_rate": 1.0928804885139674e-05,
      "loss": 0.8891,
      "step": 2261
    },
    {
      "epoch": 0.48655624865562486,
      "grad_norm": 0.0,
      "learning_rate": 1.0921867396167244e-05,
      "loss": 0.898,
      "step": 2262
    },
    {
      "epoch": 0.48677134867713484,
      "grad_norm": 0.0,
      "learning_rate": 1.0914929459679695e-05,
      "loss": 0.9301,
      "step": 2263
    },
    {
      "epoch": 0.4869864486986449,
      "grad_norm": 0.0,
      "learning_rate": 1.0907991079045006e-05,
      "loss": 0.887,
      "step": 2264
    },
    {
      "epoch": 0.48720154872015486,
      "grad_norm": 0.0,
      "learning_rate": 1.0901052257631376e-05,
      "loss": 0.8608,
      "step": 2265
    },
    {
      "epoch": 0.48741664874166485,
      "grad_norm": 0.0,
      "learning_rate": 1.0894112998807213e-05,
      "loss": 0.875,
      "step": 2266
    },
    {
      "epoch": 0.4876317487631749,
      "grad_norm": 0.0,
      "learning_rate": 1.088717330594114e-05,
      "loss": 0.8911,
      "step": 2267
    },
    {
      "epoch": 0.48784684878468487,
      "grad_norm": 0.0,
      "learning_rate": 1.088023318240199e-05,
      "loss": 0.8593,
      "step": 2268
    },
    {
      "epoch": 0.48806194880619486,
      "grad_norm": 0.0,
      "learning_rate": 1.0873292631558805e-05,
      "loss": 0.8642,
      "step": 2269
    },
    {
      "epoch": 0.4882770488277049,
      "grad_norm": 0.0,
      "learning_rate": 1.0866351656780835e-05,
      "loss": 0.8849,
      "step": 2270
    },
    {
      "epoch": 0.4884921488492149,
      "grad_norm": 0.0,
      "learning_rate": 1.0859410261437538e-05,
      "loss": 0.8878,
      "step": 2271
    },
    {
      "epoch": 0.48870724887072486,
      "grad_norm": 0.0,
      "learning_rate": 1.0852468448898567e-05,
      "loss": 0.8799,
      "step": 2272
    },
    {
      "epoch": 0.4889223488922349,
      "grad_norm": 0.0,
      "learning_rate": 1.0845526222533791e-05,
      "loss": 0.8824,
      "step": 2273
    },
    {
      "epoch": 0.4891374489137449,
      "grad_norm": 0.0,
      "learning_rate": 1.083858358571327e-05,
      "loss": 0.8392,
      "step": 2274
    },
    {
      "epoch": 0.48935254893525487,
      "grad_norm": 0.0,
      "learning_rate": 1.0831640541807267e-05,
      "loss": 0.9158,
      "step": 2275
    },
    {
      "epoch": 0.4895676489567649,
      "grad_norm": 0.0,
      "learning_rate": 1.082469709418624e-05,
      "loss": 0.8844,
      "step": 2276
    },
    {
      "epoch": 0.4897827489782749,
      "grad_norm": 0.0,
      "learning_rate": 1.081775324622085e-05,
      "loss": 0.889,
      "step": 2277
    },
    {
      "epoch": 0.4899978489997849,
      "grad_norm": 0.0,
      "learning_rate": 1.0810809001281942e-05,
      "loss": 0.8793,
      "step": 2278
    },
    {
      "epoch": 0.4902129490212949,
      "grad_norm": 0.0,
      "learning_rate": 1.0803864362740562e-05,
      "loss": 0.8235,
      "step": 2279
    },
    {
      "epoch": 0.4904280490428049,
      "grad_norm": 0.0,
      "learning_rate": 1.079691933396794e-05,
      "loss": 0.9378,
      "step": 2280
    },
    {
      "epoch": 0.4906431490643149,
      "grad_norm": 0.0,
      "learning_rate": 1.0789973918335503e-05,
      "loss": 0.8523,
      "step": 2281
    },
    {
      "epoch": 0.4908582490858249,
      "grad_norm": 0.0,
      "learning_rate": 1.0783028119214863e-05,
      "loss": 0.8547,
      "step": 2282
    },
    {
      "epoch": 0.4910733491073349,
      "grad_norm": 0.0,
      "learning_rate": 1.0776081939977816e-05,
      "loss": 0.8969,
      "step": 2283
    },
    {
      "epoch": 0.4912884491288449,
      "grad_norm": 0.0,
      "learning_rate": 1.0769135383996345e-05,
      "loss": 0.8918,
      "step": 2284
    },
    {
      "epoch": 0.49150354915035493,
      "grad_norm": 0.0,
      "learning_rate": 1.076218845464261e-05,
      "loss": 0.923,
      "step": 2285
    },
    {
      "epoch": 0.4917186491718649,
      "grad_norm": 0.0,
      "learning_rate": 1.0755241155288965e-05,
      "loss": 0.8356,
      "step": 2286
    },
    {
      "epoch": 0.4919337491933749,
      "grad_norm": 0.0,
      "learning_rate": 1.0748293489307931e-05,
      "loss": 0.8584,
      "step": 2287
    },
    {
      "epoch": 0.49214884921488494,
      "grad_norm": 0.0,
      "learning_rate": 1.074134546007221e-05,
      "loss": 0.8978,
      "step": 2288
    },
    {
      "epoch": 0.4923639492363949,
      "grad_norm": 0.0,
      "learning_rate": 1.0734397070954688e-05,
      "loss": 0.8767,
      "step": 2289
    },
    {
      "epoch": 0.4925790492579049,
      "grad_norm": 0.0,
      "learning_rate": 1.0727448325328414e-05,
      "loss": 0.819,
      "step": 2290
    },
    {
      "epoch": 0.49279414927941495,
      "grad_norm": 0.0,
      "learning_rate": 1.0720499226566616e-05,
      "loss": 0.893,
      "step": 2291
    },
    {
      "epoch": 0.49300924930092493,
      "grad_norm": 0.0,
      "learning_rate": 1.0713549778042697e-05,
      "loss": 0.9013,
      "step": 2292
    },
    {
      "epoch": 0.4932243493224349,
      "grad_norm": 0.0,
      "learning_rate": 1.0706599983130226e-05,
      "loss": 0.8755,
      "step": 2293
    },
    {
      "epoch": 0.49343944934394496,
      "grad_norm": 0.0,
      "learning_rate": 1.0699649845202935e-05,
      "loss": 0.9106,
      "step": 2294
    },
    {
      "epoch": 0.49365454936545494,
      "grad_norm": 0.0,
      "learning_rate": 1.0692699367634733e-05,
      "loss": 0.85,
      "step": 2295
    },
    {
      "epoch": 0.4938696493869649,
      "grad_norm": 0.0,
      "learning_rate": 1.0685748553799687e-05,
      "loss": 0.8947,
      "step": 2296
    },
    {
      "epoch": 0.49408474940847497,
      "grad_norm": 0.0,
      "learning_rate": 1.0678797407072027e-05,
      "loss": 0.9221,
      "step": 2297
    },
    {
      "epoch": 0.49429984942998495,
      "grad_norm": 0.0,
      "learning_rate": 1.067184593082615e-05,
      "loss": 0.8267,
      "step": 2298
    },
    {
      "epoch": 0.49451494945149493,
      "grad_norm": 0.0,
      "learning_rate": 1.0664894128436607e-05,
      "loss": 0.9163,
      "step": 2299
    },
    {
      "epoch": 0.494730049473005,
      "grad_norm": 0.0,
      "learning_rate": 1.0657942003278107e-05,
      "loss": 0.8488,
      "step": 2300
    },
    {
      "epoch": 0.49494514949451496,
      "grad_norm": 0.0,
      "learning_rate": 1.0650989558725524e-05,
      "loss": 0.8342,
      "step": 2301
    },
    {
      "epoch": 0.49516024951602494,
      "grad_norm": 0.0,
      "learning_rate": 1.0644036798153878e-05,
      "loss": 0.8952,
      "step": 2302
    },
    {
      "epoch": 0.495375349537535,
      "grad_norm": 0.0,
      "learning_rate": 1.0637083724938344e-05,
      "loss": 0.8305,
      "step": 2303
    },
    {
      "epoch": 0.49559044955904497,
      "grad_norm": 0.0,
      "learning_rate": 1.0630130342454254e-05,
      "loss": 0.8448,
      "step": 2304
    },
    {
      "epoch": 0.49580554958055495,
      "grad_norm": 0.0,
      "learning_rate": 1.0623176654077091e-05,
      "loss": 0.8784,
      "step": 2305
    },
    {
      "epoch": 0.49602064960206493,
      "grad_norm": 0.0,
      "learning_rate": 1.0616222663182474e-05,
      "loss": 0.8459,
      "step": 2306
    },
    {
      "epoch": 0.496235749623575,
      "grad_norm": 0.0,
      "learning_rate": 1.0609268373146178e-05,
      "loss": 0.8805,
      "step": 2307
    },
    {
      "epoch": 0.49645084964508496,
      "grad_norm": 0.0,
      "learning_rate": 1.0602313787344125e-05,
      "loss": 0.8233,
      "step": 2308
    },
    {
      "epoch": 0.49666594966659494,
      "grad_norm": 0.0,
      "learning_rate": 1.0595358909152377e-05,
      "loss": 0.9001,
      "step": 2309
    },
    {
      "epoch": 0.496881049688105,
      "grad_norm": 0.0,
      "learning_rate": 1.0588403741947141e-05,
      "loss": 0.9346,
      "step": 2310
    },
    {
      "epoch": 0.49709614970961497,
      "grad_norm": 0.0,
      "learning_rate": 1.0581448289104759e-05,
      "loss": 0.8904,
      "step": 2311
    },
    {
      "epoch": 0.49731124973112495,
      "grad_norm": 0.0,
      "learning_rate": 1.0574492554001718e-05,
      "loss": 0.8416,
      "step": 2312
    },
    {
      "epoch": 0.497526349752635,
      "grad_norm": 0.0,
      "learning_rate": 1.0567536540014636e-05,
      "loss": 0.8833,
      "step": 2313
    },
    {
      "epoch": 0.497741449774145,
      "grad_norm": 0.0,
      "learning_rate": 1.0560580250520269e-05,
      "loss": 0.8973,
      "step": 2314
    },
    {
      "epoch": 0.49795654979565496,
      "grad_norm": 0.0,
      "learning_rate": 1.0553623688895509e-05,
      "loss": 0.8584,
      "step": 2315
    },
    {
      "epoch": 0.498171649817165,
      "grad_norm": 0.0,
      "learning_rate": 1.0546666858517375e-05,
      "loss": 0.9071,
      "step": 2316
    },
    {
      "epoch": 0.498386749838675,
      "grad_norm": 0.0,
      "learning_rate": 1.0539709762763021e-05,
      "loss": 0.8378,
      "step": 2317
    },
    {
      "epoch": 0.49860184986018496,
      "grad_norm": 0.0,
      "learning_rate": 1.0532752405009733e-05,
      "loss": 0.9124,
      "step": 2318
    },
    {
      "epoch": 0.498816949881695,
      "grad_norm": 0.0,
      "learning_rate": 1.0525794788634912e-05,
      "loss": 0.861,
      "step": 2319
    },
    {
      "epoch": 0.499032049903205,
      "grad_norm": 0.0,
      "learning_rate": 1.0518836917016095e-05,
      "loss": 0.8088,
      "step": 2320
    },
    {
      "epoch": 0.499247149924715,
      "grad_norm": 0.0,
      "learning_rate": 1.0511878793530943e-05,
      "loss": 0.7763,
      "step": 2321
    },
    {
      "epoch": 0.499462249946225,
      "grad_norm": 0.0,
      "learning_rate": 1.0504920421557232e-05,
      "loss": 0.8514,
      "step": 2322
    },
    {
      "epoch": 0.499677349967735,
      "grad_norm": 0.0,
      "learning_rate": 1.0497961804472866e-05,
      "loss": 0.837,
      "step": 2323
    },
    {
      "epoch": 0.499892449989245,
      "grad_norm": 0.0,
      "learning_rate": 1.0491002945655862e-05,
      "loss": 0.8872,
      "step": 2324
    },
    {
      "epoch": 0.500107550010755,
      "grad_norm": 0.0,
      "learning_rate": 1.0484043848484356e-05,
      "loss": 0.8688,
      "step": 2325
    },
    {
      "epoch": 0.500322650032265,
      "grad_norm": 0.0,
      "learning_rate": 1.0477084516336607e-05,
      "loss": 0.8793,
      "step": 2326
    },
    {
      "epoch": 0.500537750053775,
      "grad_norm": 0.0,
      "learning_rate": 1.0470124952590976e-05,
      "loss": 0.84,
      "step": 2327
    },
    {
      "epoch": 0.500752850075285,
      "grad_norm": 0.0,
      "learning_rate": 1.0463165160625947e-05,
      "loss": 0.8786,
      "step": 2328
    },
    {
      "epoch": 0.500967950096795,
      "grad_norm": 0.0,
      "learning_rate": 1.0456205143820105e-05,
      "loss": 0.8552,
      "step": 2329
    },
    {
      "epoch": 0.501183050118305,
      "grad_norm": 0.0,
      "learning_rate": 1.0449244905552153e-05,
      "loss": 0.9268,
      "step": 2330
    },
    {
      "epoch": 0.501398150139815,
      "grad_norm": 0.0,
      "learning_rate": 1.0442284449200899e-05,
      "loss": 0.8553,
      "step": 2331
    },
    {
      "epoch": 0.501613250161325,
      "grad_norm": 0.0,
      "learning_rate": 1.0435323778145253e-05,
      "loss": 0.8732,
      "step": 2332
    },
    {
      "epoch": 0.501828350182835,
      "grad_norm": 0.0,
      "learning_rate": 1.0428362895764231e-05,
      "loss": 0.887,
      "step": 2333
    },
    {
      "epoch": 0.502043450204345,
      "grad_norm": 0.0,
      "learning_rate": 1.0421401805436954e-05,
      "loss": 0.8639,
      "step": 2334
    },
    {
      "epoch": 0.502258550225855,
      "grad_norm": 0.0,
      "learning_rate": 1.0414440510542645e-05,
      "loss": 0.8642,
      "step": 2335
    },
    {
      "epoch": 0.502473650247365,
      "grad_norm": 0.0,
      "learning_rate": 1.0407479014460622e-05,
      "loss": 0.7644,
      "step": 2336
    },
    {
      "epoch": 0.502688750268875,
      "grad_norm": 0.0,
      "learning_rate": 1.04005173205703e-05,
      "loss": 0.8605,
      "step": 2337
    },
    {
      "epoch": 0.502903850290385,
      "grad_norm": 0.0,
      "learning_rate": 1.0393555432251193e-05,
      "loss": 0.8586,
      "step": 2338
    },
    {
      "epoch": 0.503118950311895,
      "grad_norm": 0.0,
      "learning_rate": 1.038659335288291e-05,
      "loss": 0.8196,
      "step": 2339
    },
    {
      "epoch": 0.5033340503334051,
      "grad_norm": 0.0,
      "learning_rate": 1.0379631085845154e-05,
      "loss": 0.9193,
      "step": 2340
    },
    {
      "epoch": 0.503549150354915,
      "grad_norm": 0.0,
      "learning_rate": 1.0372668634517718e-05,
      "loss": 0.8456,
      "step": 2341
    },
    {
      "epoch": 0.503764250376425,
      "grad_norm": 0.0,
      "learning_rate": 1.0365706002280471e-05,
      "loss": 0.8803,
      "step": 2342
    },
    {
      "epoch": 0.5039793503979351,
      "grad_norm": 0.0,
      "learning_rate": 1.0358743192513396e-05,
      "loss": 0.8846,
      "step": 2343
    },
    {
      "epoch": 0.504194450419445,
      "grad_norm": 0.0,
      "learning_rate": 1.0351780208596542e-05,
      "loss": 0.8255,
      "step": 2344
    },
    {
      "epoch": 0.504409550440955,
      "grad_norm": 0.0,
      "learning_rate": 1.034481705391005e-05,
      "loss": 0.8976,
      "step": 2345
    },
    {
      "epoch": 0.5046246504624651,
      "grad_norm": 0.0,
      "learning_rate": 1.033785373183414e-05,
      "loss": 0.8455,
      "step": 2346
    },
    {
      "epoch": 0.504839750483975,
      "grad_norm": 0.0,
      "learning_rate": 1.0330890245749119e-05,
      "loss": 0.8546,
      "step": 2347
    },
    {
      "epoch": 0.505054850505485,
      "grad_norm": 0.0,
      "learning_rate": 1.032392659903537e-05,
      "loss": 0.8477,
      "step": 2348
    },
    {
      "epoch": 0.5052699505269951,
      "grad_norm": 0.0,
      "learning_rate": 1.0316962795073353e-05,
      "loss": 0.8814,
      "step": 2349
    },
    {
      "epoch": 0.505485050548505,
      "grad_norm": 0.0,
      "learning_rate": 1.0309998837243607e-05,
      "loss": 0.9282,
      "step": 2350
    },
    {
      "epoch": 0.505700150570015,
      "grad_norm": 0.0,
      "learning_rate": 1.0303034728926741e-05,
      "loss": 0.8466,
      "step": 2351
    },
    {
      "epoch": 0.5059152505915251,
      "grad_norm": 0.0,
      "learning_rate": 1.0296070473503448e-05,
      "loss": 0.7954,
      "step": 2352
    },
    {
      "epoch": 0.506130350613035,
      "grad_norm": 0.0,
      "learning_rate": 1.0289106074354477e-05,
      "loss": 0.8294,
      "step": 2353
    },
    {
      "epoch": 0.5063454506345451,
      "grad_norm": 0.0,
      "learning_rate": 1.0282141534860662e-05,
      "loss": 0.8601,
      "step": 2354
    },
    {
      "epoch": 0.5065605506560551,
      "grad_norm": 0.0,
      "learning_rate": 1.0275176858402891e-05,
      "loss": 0.818,
      "step": 2355
    },
    {
      "epoch": 0.506775650677565,
      "grad_norm": 0.0,
      "learning_rate": 1.026821204836213e-05,
      "loss": 0.8692,
      "step": 2356
    },
    {
      "epoch": 0.5069907506990751,
      "grad_norm": 0.0,
      "learning_rate": 1.0261247108119403e-05,
      "loss": 0.8231,
      "step": 2357
    },
    {
      "epoch": 0.5072058507205851,
      "grad_norm": 0.0,
      "learning_rate": 1.02542820410558e-05,
      "loss": 0.8108,
      "step": 2358
    },
    {
      "epoch": 0.507420950742095,
      "grad_norm": 0.0,
      "learning_rate": 1.0247316850552474e-05,
      "loss": 0.864,
      "step": 2359
    },
    {
      "epoch": 0.5076360507636051,
      "grad_norm": 0.0,
      "learning_rate": 1.0240351539990628e-05,
      "loss": 0.8383,
      "step": 2360
    },
    {
      "epoch": 0.5078511507851151,
      "grad_norm": 0.0,
      "learning_rate": 1.023338611275154e-05,
      "loss": 0.8019,
      "step": 2361
    },
    {
      "epoch": 0.508066250806625,
      "grad_norm": 0.0,
      "learning_rate": 1.0226420572216534e-05,
      "loss": 0.8269,
      "step": 2362
    },
    {
      "epoch": 0.5082813508281351,
      "grad_norm": 0.0,
      "learning_rate": 1.0219454921766991e-05,
      "loss": 0.8248,
      "step": 2363
    },
    {
      "epoch": 0.5084964508496451,
      "grad_norm": 0.0,
      "learning_rate": 1.0212489164784338e-05,
      "loss": 0.9252,
      "step": 2364
    },
    {
      "epoch": 0.508711550871155,
      "grad_norm": 0.0,
      "learning_rate": 1.020552330465007e-05,
      "loss": 0.7875,
      "step": 2365
    },
    {
      "epoch": 0.5089266508926651,
      "grad_norm": 0.0,
      "learning_rate": 1.0198557344745718e-05,
      "loss": 0.9084,
      "step": 2366
    },
    {
      "epoch": 0.5091417509141751,
      "grad_norm": 0.0,
      "learning_rate": 1.0191591288452872e-05,
      "loss": 0.8841,
      "step": 2367
    },
    {
      "epoch": 0.5093568509356851,
      "grad_norm": 0.0,
      "learning_rate": 1.0184625139153155e-05,
      "loss": 0.9188,
      "step": 2368
    },
    {
      "epoch": 0.5095719509571951,
      "grad_norm": 0.0,
      "learning_rate": 1.0177658900228249e-05,
      "loss": 0.78,
      "step": 2369
    },
    {
      "epoch": 0.5097870509787051,
      "grad_norm": 0.0,
      "learning_rate": 1.017069257505987e-05,
      "loss": 0.8492,
      "step": 2370
    },
    {
      "epoch": 0.5100021510002151,
      "grad_norm": 0.0,
      "learning_rate": 1.0163726167029785e-05,
      "loss": 0.8516,
      "step": 2371
    },
    {
      "epoch": 0.5102172510217251,
      "grad_norm": 0.0,
      "learning_rate": 1.015675967951979e-05,
      "loss": 0.9341,
      "step": 2372
    },
    {
      "epoch": 0.5104323510432351,
      "grad_norm": 0.0,
      "learning_rate": 1.0149793115911725e-05,
      "loss": 0.9071,
      "step": 2373
    },
    {
      "epoch": 0.5106474510647451,
      "grad_norm": 0.0,
      "learning_rate": 1.0142826479587475e-05,
      "loss": 0.88,
      "step": 2374
    },
    {
      "epoch": 0.5108625510862551,
      "grad_norm": 0.0,
      "learning_rate": 1.0135859773928949e-05,
      "loss": 0.8542,
      "step": 2375
    },
    {
      "epoch": 0.5110776511077652,
      "grad_norm": 0.0,
      "learning_rate": 1.0128893002318092e-05,
      "loss": 0.7619,
      "step": 2376
    },
    {
      "epoch": 0.5112927511292751,
      "grad_norm": 0.0,
      "learning_rate": 1.0121926168136884e-05,
      "loss": 0.8506,
      "step": 2377
    },
    {
      "epoch": 0.5115078511507851,
      "grad_norm": 0.0,
      "learning_rate": 1.011495927476733e-05,
      "loss": 0.8479,
      "step": 2378
    },
    {
      "epoch": 0.5117229511722952,
      "grad_norm": 0.0,
      "learning_rate": 1.0107992325591474e-05,
      "loss": 0.9215,
      "step": 2379
    },
    {
      "epoch": 0.5119380511938051,
      "grad_norm": 0.0,
      "learning_rate": 1.0101025323991375e-05,
      "loss": 0.8843,
      "step": 2380
    },
    {
      "epoch": 0.5121531512153151,
      "grad_norm": 0.0,
      "learning_rate": 1.0094058273349125e-05,
      "loss": 0.8613,
      "step": 2381
    },
    {
      "epoch": 0.5123682512368252,
      "grad_norm": 0.0,
      "learning_rate": 1.0087091177046838e-05,
      "loss": 0.8483,
      "step": 2382
    },
    {
      "epoch": 0.5125833512583351,
      "grad_norm": 0.0,
      "learning_rate": 1.008012403846665e-05,
      "loss": 0.8484,
      "step": 2383
    },
    {
      "epoch": 0.5127984512798451,
      "grad_norm": 0.0,
      "learning_rate": 1.007315686099072e-05,
      "loss": 0.8849,
      "step": 2384
    },
    {
      "epoch": 0.5130135513013552,
      "grad_norm": 0.0,
      "learning_rate": 1.006618964800122e-05,
      "loss": 0.7923,
      "step": 2385
    },
    {
      "epoch": 0.5132286513228651,
      "grad_norm": 0.0,
      "learning_rate": 1.0059222402880339e-05,
      "loss": 0.9168,
      "step": 2386
    },
    {
      "epoch": 0.5134437513443751,
      "grad_norm": 0.0,
      "learning_rate": 1.0052255129010293e-05,
      "loss": 0.8649,
      "step": 2387
    },
    {
      "epoch": 0.5136588513658852,
      "grad_norm": 0.0,
      "learning_rate": 1.0045287829773302e-05,
      "loss": 0.8523,
      "step": 2388
    },
    {
      "epoch": 0.5138739513873951,
      "grad_norm": 0.0,
      "learning_rate": 1.00383205085516e-05,
      "loss": 0.7904,
      "step": 2389
    },
    {
      "epoch": 0.5140890514089052,
      "grad_norm": 0.0,
      "learning_rate": 1.0031353168727432e-05,
      "loss": 0.9191,
      "step": 2390
    },
    {
      "epoch": 0.5143041514304152,
      "grad_norm": 0.0,
      "learning_rate": 1.0024385813683054e-05,
      "loss": 0.8608,
      "step": 2391
    },
    {
      "epoch": 0.5145192514519251,
      "grad_norm": 0.0,
      "learning_rate": 1.0017418446800722e-05,
      "loss": 0.9278,
      "step": 2392
    },
    {
      "epoch": 0.5147343514734352,
      "grad_norm": 0.0,
      "learning_rate": 1.0010451071462707e-05,
      "loss": 0.7781,
      "step": 2393
    },
    {
      "epoch": 0.5149494514949452,
      "grad_norm": 0.0,
      "learning_rate": 1.0003483691051283e-05,
      "loss": 0.8558,
      "step": 2394
    },
    {
      "epoch": 0.5151645515164551,
      "grad_norm": 0.0,
      "learning_rate": 9.996516308948722e-06,
      "loss": 0.8254,
      "step": 2395
    },
    {
      "epoch": 0.5153796515379652,
      "grad_norm": 0.0,
      "learning_rate": 9.989548928537294e-06,
      "loss": 0.8226,
      "step": 2396
    },
    {
      "epoch": 0.5155947515594752,
      "grad_norm": 0.0,
      "learning_rate": 9.98258155319928e-06,
      "loss": 0.8539,
      "step": 2397
    },
    {
      "epoch": 0.5158098515809851,
      "grad_norm": 0.0,
      "learning_rate": 9.975614186316952e-06,
      "loss": 0.888,
      "step": 2398
    },
    {
      "epoch": 0.5160249516024952,
      "grad_norm": 0.0,
      "learning_rate": 9.96864683127257e-06,
      "loss": 0.9246,
      "step": 2399
    },
    {
      "epoch": 0.5162400516240052,
      "grad_norm": 0.0,
      "learning_rate": 9.961679491448403e-06,
      "loss": 0.8201,
      "step": 2400
    },
    {
      "epoch": 0.5164551516455151,
      "grad_norm": 0.0,
      "learning_rate": 9.9547121702267e-06,
      "loss": 0.8765,
      "step": 2401
    },
    {
      "epoch": 0.5166702516670252,
      "grad_norm": 0.0,
      "learning_rate": 9.947744870989705e-06,
      "loss": 0.8334,
      "step": 2402
    },
    {
      "epoch": 0.5168853516885352,
      "grad_norm": 0.0,
      "learning_rate": 9.940777597119663e-06,
      "loss": 0.8789,
      "step": 2403
    },
    {
      "epoch": 0.5171004517100452,
      "grad_norm": 0.0,
      "learning_rate": 9.933810351998784e-06,
      "loss": 0.9486,
      "step": 2404
    },
    {
      "epoch": 0.5173155517315552,
      "grad_norm": 0.0,
      "learning_rate": 9.926843139009286e-06,
      "loss": 0.7912,
      "step": 2405
    },
    {
      "epoch": 0.5175306517530652,
      "grad_norm": 0.0,
      "learning_rate": 9.919875961533355e-06,
      "loss": 0.8211,
      "step": 2406
    },
    {
      "epoch": 0.5177457517745752,
      "grad_norm": 0.0,
      "learning_rate": 9.912908822953164e-06,
      "loss": 0.844,
      "step": 2407
    },
    {
      "epoch": 0.5179608517960852,
      "grad_norm": 0.0,
      "learning_rate": 9.90594172665088e-06,
      "loss": 0.8888,
      "step": 2408
    },
    {
      "epoch": 0.5181759518175952,
      "grad_norm": 0.0,
      "learning_rate": 9.898974676008628e-06,
      "loss": 0.9058,
      "step": 2409
    },
    {
      "epoch": 0.5183910518391052,
      "grad_norm": 0.0,
      "learning_rate": 9.892007674408527e-06,
      "loss": 0.8364,
      "step": 2410
    },
    {
      "epoch": 0.5186061518606152,
      "grad_norm": 0.0,
      "learning_rate": 9.885040725232671e-06,
      "loss": 0.9053,
      "step": 2411
    },
    {
      "epoch": 0.5188212518821251,
      "grad_norm": 0.0,
      "learning_rate": 9.87807383186312e-06,
      "loss": 0.8589,
      "step": 2412
    },
    {
      "epoch": 0.5190363519036352,
      "grad_norm": 0.0,
      "learning_rate": 9.87110699768191e-06,
      "loss": 0.8817,
      "step": 2413
    },
    {
      "epoch": 0.5192514519251452,
      "grad_norm": 0.0,
      "learning_rate": 9.864140226071054e-06,
      "loss": 0.8529,
      "step": 2414
    },
    {
      "epoch": 0.5194665519466551,
      "grad_norm": 0.0,
      "learning_rate": 9.857173520412524e-06,
      "loss": 0.7802,
      "step": 2415
    },
    {
      "epoch": 0.5196816519681652,
      "grad_norm": 0.0,
      "learning_rate": 9.850206884088276e-06,
      "loss": 0.8609,
      "step": 2416
    },
    {
      "epoch": 0.5198967519896752,
      "grad_norm": 0.0,
      "learning_rate": 9.843240320480213e-06,
      "loss": 0.8421,
      "step": 2417
    },
    {
      "epoch": 0.5201118520111851,
      "grad_norm": 0.0,
      "learning_rate": 9.83627383297022e-06,
      "loss": 0.8831,
      "step": 2418
    },
    {
      "epoch": 0.5203269520326952,
      "grad_norm": 0.0,
      "learning_rate": 9.829307424940133e-06,
      "loss": 0.8703,
      "step": 2419
    },
    {
      "epoch": 0.5205420520542052,
      "grad_norm": 0.0,
      "learning_rate": 9.822341099771755e-06,
      "loss": 0.9144,
      "step": 2420
    },
    {
      "epoch": 0.5207571520757152,
      "grad_norm": 0.0,
      "learning_rate": 9.81537486084685e-06,
      "loss": 0.8334,
      "step": 2421
    },
    {
      "epoch": 0.5209722520972252,
      "grad_norm": 0.0,
      "learning_rate": 9.808408711547132e-06,
      "loss": 0.886,
      "step": 2422
    },
    {
      "epoch": 0.5211873521187352,
      "grad_norm": 0.0,
      "learning_rate": 9.801442655254285e-06,
      "loss": 0.895,
      "step": 2423
    },
    {
      "epoch": 0.5214024521402452,
      "grad_norm": 0.0,
      "learning_rate": 9.794476695349933e-06,
      "loss": 0.8665,
      "step": 2424
    },
    {
      "epoch": 0.5216175521617552,
      "grad_norm": 0.0,
      "learning_rate": 9.787510835215662e-06,
      "loss": 0.944,
      "step": 2425
    },
    {
      "epoch": 0.5218326521832652,
      "grad_norm": 0.0,
      "learning_rate": 9.780545078233015e-06,
      "loss": 0.8481,
      "step": 2426
    },
    {
      "epoch": 0.5220477522047752,
      "grad_norm": 0.0,
      "learning_rate": 9.77357942778347e-06,
      "loss": 0.7987,
      "step": 2427
    },
    {
      "epoch": 0.5222628522262852,
      "grad_norm": 0.0,
      "learning_rate": 9.76661388724846e-06,
      "loss": 0.8183,
      "step": 2428
    },
    {
      "epoch": 0.5224779522477953,
      "grad_norm": 0.0,
      "learning_rate": 9.759648460009376e-06,
      "loss": 0.8736,
      "step": 2429
    },
    {
      "epoch": 0.5226930522693052,
      "grad_norm": 0.0,
      "learning_rate": 9.752683149447531e-06,
      "loss": 0.9153,
      "step": 2430
    },
    {
      "epoch": 0.5229081522908152,
      "grad_norm": 0.0,
      "learning_rate": 9.745717958944204e-06,
      "loss": 0.9107,
      "step": 2431
    },
    {
      "epoch": 0.5231232523123253,
      "grad_norm": 0.0,
      "learning_rate": 9.7387528918806e-06,
      "loss": 0.8963,
      "step": 2432
    },
    {
      "epoch": 0.5233383523338352,
      "grad_norm": 0.0,
      "learning_rate": 9.73178795163787e-06,
      "loss": 0.8101,
      "step": 2433
    },
    {
      "epoch": 0.5235534523553452,
      "grad_norm": 0.0,
      "learning_rate": 9.72482314159711e-06,
      "loss": 0.8871,
      "step": 2434
    },
    {
      "epoch": 0.5237685523768553,
      "grad_norm": 0.0,
      "learning_rate": 9.717858465139341e-06,
      "loss": 0.8651,
      "step": 2435
    },
    {
      "epoch": 0.5239836523983652,
      "grad_norm": 0.0,
      "learning_rate": 9.710893925645528e-06,
      "loss": 0.8565,
      "step": 2436
    },
    {
      "epoch": 0.5241987524198752,
      "grad_norm": 0.0,
      "learning_rate": 9.703929526496555e-06,
      "loss": 0.8544,
      "step": 2437
    },
    {
      "epoch": 0.5244138524413853,
      "grad_norm": 0.0,
      "learning_rate": 9.696965271073259e-06,
      "loss": 0.9293,
      "step": 2438
    },
    {
      "epoch": 0.5246289524628952,
      "grad_norm": 0.0,
      "learning_rate": 9.690001162756398e-06,
      "loss": 0.8482,
      "step": 2439
    },
    {
      "epoch": 0.5248440524844052,
      "grad_norm": 0.0,
      "learning_rate": 9.683037204926648e-06,
      "loss": 0.948,
      "step": 2440
    },
    {
      "epoch": 0.5250591525059153,
      "grad_norm": 0.0,
      "learning_rate": 9.676073400964631e-06,
      "loss": 0.873,
      "step": 2441
    },
    {
      "epoch": 0.5252742525274252,
      "grad_norm": 0.0,
      "learning_rate": 9.669109754250884e-06,
      "loss": 0.7878,
      "step": 2442
    },
    {
      "epoch": 0.5254893525489353,
      "grad_norm": 0.0,
      "learning_rate": 9.662146268165862e-06,
      "loss": 0.9263,
      "step": 2443
    },
    {
      "epoch": 0.5257044525704453,
      "grad_norm": 0.0,
      "learning_rate": 9.655182946089957e-06,
      "loss": 0.8667,
      "step": 2444
    },
    {
      "epoch": 0.5259195525919552,
      "grad_norm": 0.0,
      "learning_rate": 9.648219791403461e-06,
      "loss": 0.8843,
      "step": 2445
    },
    {
      "epoch": 0.5261346526134653,
      "grad_norm": 0.0,
      "learning_rate": 9.641256807486604e-06,
      "loss": 0.941,
      "step": 2446
    },
    {
      "epoch": 0.5263497526349753,
      "grad_norm": 0.0,
      "learning_rate": 9.63429399771953e-06,
      "loss": 0.829,
      "step": 2447
    },
    {
      "epoch": 0.5265648526564852,
      "grad_norm": 0.0,
      "learning_rate": 9.627331365482286e-06,
      "loss": 0.8584,
      "step": 2448
    },
    {
      "epoch": 0.5267799526779953,
      "grad_norm": 0.0,
      "learning_rate": 9.620368914154848e-06,
      "loss": 0.8885,
      "step": 2449
    },
    {
      "epoch": 0.5269950526995053,
      "grad_norm": 0.0,
      "learning_rate": 9.613406647117092e-06,
      "loss": 0.875,
      "step": 2450
    },
    {
      "epoch": 0.5272101527210152,
      "grad_norm": 0.0,
      "learning_rate": 9.606444567748809e-06,
      "loss": 0.824,
      "step": 2451
    },
    {
      "epoch": 0.5274252527425253,
      "grad_norm": 0.0,
      "learning_rate": 9.599482679429705e-06,
      "loss": 0.8999,
      "step": 2452
    },
    {
      "epoch": 0.5276403527640353,
      "grad_norm": 0.0,
      "learning_rate": 9.592520985539382e-06,
      "loss": 0.9078,
      "step": 2453
    },
    {
      "epoch": 0.5278554527855452,
      "grad_norm": 0.0,
      "learning_rate": 9.585559489457355e-06,
      "loss": 0.9486,
      "step": 2454
    },
    {
      "epoch": 0.5280705528070553,
      "grad_norm": 0.0,
      "learning_rate": 9.578598194563047e-06,
      "loss": 0.8582,
      "step": 2455
    },
    {
      "epoch": 0.5282856528285653,
      "grad_norm": 0.0,
      "learning_rate": 9.57163710423577e-06,
      "loss": 0.9133,
      "step": 2456
    },
    {
      "epoch": 0.5285007528500753,
      "grad_norm": 0.0,
      "learning_rate": 9.56467622185475e-06,
      "loss": 0.8242,
      "step": 2457
    },
    {
      "epoch": 0.5287158528715853,
      "grad_norm": 0.0,
      "learning_rate": 9.557715550799104e-06,
      "loss": 0.9218,
      "step": 2458
    },
    {
      "epoch": 0.5289309528930953,
      "grad_norm": 0.0,
      "learning_rate": 9.550755094447848e-06,
      "loss": 0.9028,
      "step": 2459
    },
    {
      "epoch": 0.5291460529146053,
      "grad_norm": 0.0,
      "learning_rate": 9.543794856179899e-06,
      "loss": 0.8956,
      "step": 2460
    },
    {
      "epoch": 0.5293611529361153,
      "grad_norm": 0.0,
      "learning_rate": 9.536834839374056e-06,
      "loss": 0.9174,
      "step": 2461
    },
    {
      "epoch": 0.5295762529576253,
      "grad_norm": 0.0,
      "learning_rate": 9.529875047409027e-06,
      "loss": 0.9135,
      "step": 2462
    },
    {
      "epoch": 0.5297913529791353,
      "grad_norm": 0.0,
      "learning_rate": 9.522915483663395e-06,
      "loss": 0.8229,
      "step": 2463
    },
    {
      "epoch": 0.5300064530006453,
      "grad_norm": 0.0,
      "learning_rate": 9.515956151515646e-06,
      "loss": 0.9024,
      "step": 2464
    },
    {
      "epoch": 0.5302215530221553,
      "grad_norm": 0.0,
      "learning_rate": 9.508997054344145e-06,
      "loss": 0.8961,
      "step": 2465
    },
    {
      "epoch": 0.5304366530436653,
      "grad_norm": 0.0,
      "learning_rate": 9.502038195527138e-06,
      "loss": 0.9333,
      "step": 2466
    },
    {
      "epoch": 0.5306517530651753,
      "grad_norm": 0.0,
      "learning_rate": 9.495079578442768e-06,
      "loss": 0.8727,
      "step": 2467
    },
    {
      "epoch": 0.5308668530866854,
      "grad_norm": 0.0,
      "learning_rate": 9.48812120646906e-06,
      "loss": 0.9574,
      "step": 2468
    },
    {
      "epoch": 0.5310819531081953,
      "grad_norm": 0.0,
      "learning_rate": 9.481163082983905e-06,
      "loss": 0.7898,
      "step": 2469
    },
    {
      "epoch": 0.5312970531297053,
      "grad_norm": 0.0,
      "learning_rate": 9.474205211365092e-06,
      "loss": 0.8729,
      "step": 2470
    },
    {
      "epoch": 0.5315121531512154,
      "grad_norm": 0.0,
      "learning_rate": 9.467247594990272e-06,
      "loss": 0.8597,
      "step": 2471
    },
    {
      "epoch": 0.5317272531727253,
      "grad_norm": 0.0,
      "learning_rate": 9.460290237236979e-06,
      "loss": 0.8334,
      "step": 2472
    },
    {
      "epoch": 0.5319423531942353,
      "grad_norm": 0.0,
      "learning_rate": 9.453333141482629e-06,
      "loss": 0.8653,
      "step": 2473
    },
    {
      "epoch": 0.5321574532157454,
      "grad_norm": 0.0,
      "learning_rate": 9.446376311104494e-06,
      "loss": 0.8342,
      "step": 2474
    },
    {
      "epoch": 0.5323725532372553,
      "grad_norm": 0.0,
      "learning_rate": 9.439419749479738e-06,
      "loss": 0.8734,
      "step": 2475
    },
    {
      "epoch": 0.5325876532587653,
      "grad_norm": 0.0,
      "learning_rate": 9.432463459985367e-06,
      "loss": 0.8927,
      "step": 2476
    },
    {
      "epoch": 0.5328027532802754,
      "grad_norm": 0.0,
      "learning_rate": 9.425507445998286e-06,
      "loss": 0.9243,
      "step": 2477
    },
    {
      "epoch": 0.5330178533017853,
      "grad_norm": 0.0,
      "learning_rate": 9.418551710895243e-06,
      "loss": 0.8621,
      "step": 2478
    },
    {
      "epoch": 0.5332329533232953,
      "grad_norm": 0.0,
      "learning_rate": 9.411596258052862e-06,
      "loss": 0.8135,
      "step": 2479
    },
    {
      "epoch": 0.5334480533448054,
      "grad_norm": 0.0,
      "learning_rate": 9.404641090847628e-06,
      "loss": 0.8087,
      "step": 2480
    },
    {
      "epoch": 0.5336631533663153,
      "grad_norm": 0.0,
      "learning_rate": 9.397686212655878e-06,
      "loss": 0.9029,
      "step": 2481
    },
    {
      "epoch": 0.5338782533878254,
      "grad_norm": 0.0,
      "learning_rate": 9.390731626853823e-06,
      "loss": 0.8882,
      "step": 2482
    },
    {
      "epoch": 0.5340933534093354,
      "grad_norm": 0.0,
      "learning_rate": 9.383777336817532e-06,
      "loss": 0.9257,
      "step": 2483
    },
    {
      "epoch": 0.5343084534308453,
      "grad_norm": 0.0,
      "learning_rate": 9.376823345922912e-06,
      "loss": 0.8271,
      "step": 2484
    },
    {
      "epoch": 0.5345235534523554,
      "grad_norm": 0.0,
      "learning_rate": 9.369869657545744e-06,
      "loss": 0.8889,
      "step": 2485
    },
    {
      "epoch": 0.5347386534738654,
      "grad_norm": 0.0,
      "learning_rate": 9.36291627506166e-06,
      "loss": 0.8463,
      "step": 2486
    },
    {
      "epoch": 0.5349537534953753,
      "grad_norm": 0.0,
      "learning_rate": 9.355963201846125e-06,
      "loss": 0.8603,
      "step": 2487
    },
    {
      "epoch": 0.5351688535168854,
      "grad_norm": 0.0,
      "learning_rate": 9.349010441274481e-06,
      "loss": 0.8922,
      "step": 2488
    },
    {
      "epoch": 0.5353839535383954,
      "grad_norm": 0.0,
      "learning_rate": 9.342057996721894e-06,
      "loss": 0.8582,
      "step": 2489
    },
    {
      "epoch": 0.5355990535599053,
      "grad_norm": 0.0,
      "learning_rate": 9.335105871563394e-06,
      "loss": 0.8989,
      "step": 2490
    },
    {
      "epoch": 0.5358141535814154,
      "grad_norm": 0.0,
      "learning_rate": 9.328154069173853e-06,
      "loss": 0.8268,
      "step": 2491
    },
    {
      "epoch": 0.5360292536029254,
      "grad_norm": 0.0,
      "learning_rate": 9.321202592927974e-06,
      "loss": 0.8695,
      "step": 2492
    },
    {
      "epoch": 0.5362443536244353,
      "grad_norm": 0.0,
      "learning_rate": 9.314251446200317e-06,
      "loss": 0.828,
      "step": 2493
    },
    {
      "epoch": 0.5364594536459454,
      "grad_norm": 0.0,
      "learning_rate": 9.307300632365269e-06,
      "loss": 0.8644,
      "step": 2494
    },
    {
      "epoch": 0.5366745536674553,
      "grad_norm": 0.0,
      "learning_rate": 9.300350154797065e-06,
      "loss": 0.8526,
      "step": 2495
    },
    {
      "epoch": 0.5368896536889654,
      "grad_norm": 0.0,
      "learning_rate": 9.293400016869779e-06,
      "loss": 0.9347,
      "step": 2496
    },
    {
      "epoch": 0.5371047537104754,
      "grad_norm": 0.0,
      "learning_rate": 9.286450221957304e-06,
      "loss": 0.8549,
      "step": 2497
    },
    {
      "epoch": 0.5373198537319853,
      "grad_norm": 0.0,
      "learning_rate": 9.279500773433385e-06,
      "loss": 0.9085,
      "step": 2498
    },
    {
      "epoch": 0.5375349537534954,
      "grad_norm": 0.0,
      "learning_rate": 9.272551674671591e-06,
      "loss": 0.9082,
      "step": 2499
    },
    {
      "epoch": 0.5377500537750054,
      "grad_norm": 0.0,
      "learning_rate": 9.265602929045317e-06,
      "loss": 0.8932,
      "step": 2500
    },
    {
      "epoch": 0.5379651537965153,
      "grad_norm": 0.0,
      "learning_rate": 9.258654539927795e-06,
      "loss": 0.8949,
      "step": 2501
    },
    {
      "epoch": 0.5381802538180254,
      "grad_norm": 0.0,
      "learning_rate": 9.251706510692074e-06,
      "loss": 0.8988,
      "step": 2502
    },
    {
      "epoch": 0.5383953538395354,
      "grad_norm": 0.0,
      "learning_rate": 9.244758844711035e-06,
      "loss": 0.862,
      "step": 2503
    },
    {
      "epoch": 0.5386104538610453,
      "grad_norm": 0.0,
      "learning_rate": 9.237811545357393e-06,
      "loss": 0.8028,
      "step": 2504
    },
    {
      "epoch": 0.5388255538825554,
      "grad_norm": 0.0,
      "learning_rate": 9.230864616003657e-06,
      "loss": 0.8007,
      "step": 2505
    },
    {
      "epoch": 0.5390406539040654,
      "grad_norm": 0.0,
      "learning_rate": 9.223918060022189e-06,
      "loss": 0.795,
      "step": 2506
    },
    {
      "epoch": 0.5392557539255753,
      "grad_norm": 0.0,
      "learning_rate": 9.21697188078514e-06,
      "loss": 0.8831,
      "step": 2507
    },
    {
      "epoch": 0.5394708539470854,
      "grad_norm": 0.0,
      "learning_rate": 9.210026081664499e-06,
      "loss": 0.8231,
      "step": 2508
    },
    {
      "epoch": 0.5396859539685954,
      "grad_norm": 0.0,
      "learning_rate": 9.203080666032064e-06,
      "loss": 0.7832,
      "step": 2509
    },
    {
      "epoch": 0.5399010539901054,
      "grad_norm": 0.0,
      "learning_rate": 9.196135637259443e-06,
      "loss": 0.8945,
      "step": 2510
    },
    {
      "epoch": 0.5401161540116154,
      "grad_norm": 0.0,
      "learning_rate": 9.189190998718058e-06,
      "loss": 0.8874,
      "step": 2511
    },
    {
      "epoch": 0.5403312540331254,
      "grad_norm": 0.0,
      "learning_rate": 9.182246753779154e-06,
      "loss": 0.9276,
      "step": 2512
    },
    {
      "epoch": 0.5405463540546354,
      "grad_norm": 0.0,
      "learning_rate": 9.175302905813761e-06,
      "loss": 0.8788,
      "step": 2513
    },
    {
      "epoch": 0.5407614540761454,
      "grad_norm": 0.0,
      "learning_rate": 9.168359458192737e-06,
      "loss": 0.8725,
      "step": 2514
    },
    {
      "epoch": 0.5409765540976554,
      "grad_norm": 0.0,
      "learning_rate": 9.161416414286734e-06,
      "loss": 0.9063,
      "step": 2515
    },
    {
      "epoch": 0.5411916541191654,
      "grad_norm": 0.0,
      "learning_rate": 9.154473777466209e-06,
      "loss": 0.9505,
      "step": 2516
    },
    {
      "epoch": 0.5414067541406754,
      "grad_norm": 0.0,
      "learning_rate": 9.147531551101436e-06,
      "loss": 0.8973,
      "step": 2517
    },
    {
      "epoch": 0.5416218541621854,
      "grad_norm": 0.0,
      "learning_rate": 9.140589738562465e-06,
      "loss": 0.9543,
      "step": 2518
    },
    {
      "epoch": 0.5418369541836954,
      "grad_norm": 0.0,
      "learning_rate": 9.133648343219168e-06,
      "loss": 0.8104,
      "step": 2519
    },
    {
      "epoch": 0.5420520542052054,
      "grad_norm": 0.0,
      "learning_rate": 9.126707368441196e-06,
      "loss": 0.899,
      "step": 2520
    },
    {
      "epoch": 0.5422671542267155,
      "grad_norm": 0.0,
      "learning_rate": 9.119766817598013e-06,
      "loss": 0.9535,
      "step": 2521
    },
    {
      "epoch": 0.5424822542482254,
      "grad_norm": 0.0,
      "learning_rate": 9.112826694058864e-06,
      "loss": 0.8499,
      "step": 2522
    },
    {
      "epoch": 0.5426973542697354,
      "grad_norm": 0.0,
      "learning_rate": 9.10588700119279e-06,
      "loss": 0.7899,
      "step": 2523
    },
    {
      "epoch": 0.5429124542912455,
      "grad_norm": 0.0,
      "learning_rate": 9.098947742368629e-06,
      "loss": 0.9355,
      "step": 2524
    },
    {
      "epoch": 0.5431275543127554,
      "grad_norm": 0.0,
      "learning_rate": 9.092008920954997e-06,
      "loss": 0.7915,
      "step": 2525
    },
    {
      "epoch": 0.5433426543342654,
      "grad_norm": 0.0,
      "learning_rate": 9.085070540320306e-06,
      "loss": 0.8155,
      "step": 2526
    },
    {
      "epoch": 0.5435577543557755,
      "grad_norm": 0.0,
      "learning_rate": 9.07813260383276e-06,
      "loss": 0.8327,
      "step": 2527
    },
    {
      "epoch": 0.5437728543772854,
      "grad_norm": 0.0,
      "learning_rate": 9.07119511486033e-06,
      "loss": 0.8158,
      "step": 2528
    },
    {
      "epoch": 0.5439879543987954,
      "grad_norm": 0.0,
      "learning_rate": 9.064258076770783e-06,
      "loss": 0.8309,
      "step": 2529
    },
    {
      "epoch": 0.5442030544203055,
      "grad_norm": 0.0,
      "learning_rate": 9.057321492931665e-06,
      "loss": 0.904,
      "step": 2530
    },
    {
      "epoch": 0.5444181544418154,
      "grad_norm": 0.0,
      "learning_rate": 9.050385366710295e-06,
      "loss": 0.8524,
      "step": 2531
    },
    {
      "epoch": 0.5446332544633254,
      "grad_norm": 0.0,
      "learning_rate": 9.043449701473782e-06,
      "loss": 0.8372,
      "step": 2532
    },
    {
      "epoch": 0.5448483544848355,
      "grad_norm": 0.0,
      "learning_rate": 9.036514500588995e-06,
      "loss": 0.906,
      "step": 2533
    },
    {
      "epoch": 0.5450634545063454,
      "grad_norm": 0.0,
      "learning_rate": 9.029579767422592e-06,
      "loss": 0.9143,
      "step": 2534
    },
    {
      "epoch": 0.5452785545278555,
      "grad_norm": 0.0,
      "learning_rate": 9.022645505341004e-06,
      "loss": 0.903,
      "step": 2535
    },
    {
      "epoch": 0.5454936545493655,
      "grad_norm": 0.0,
      "learning_rate": 9.015711717710418e-06,
      "loss": 0.8228,
      "step": 2536
    },
    {
      "epoch": 0.5457087545708754,
      "grad_norm": 0.0,
      "learning_rate": 9.008778407896807e-06,
      "loss": 0.8041,
      "step": 2537
    },
    {
      "epoch": 0.5459238545923855,
      "grad_norm": 0.0,
      "learning_rate": 9.001845579265897e-06,
      "loss": 0.8577,
      "step": 2538
    },
    {
      "epoch": 0.5461389546138955,
      "grad_norm": 0.0,
      "learning_rate": 8.994913235183197e-06,
      "loss": 0.8023,
      "step": 2539
    },
    {
      "epoch": 0.5463540546354054,
      "grad_norm": 0.0,
      "learning_rate": 8.987981379013975e-06,
      "loss": 0.8464,
      "step": 2540
    },
    {
      "epoch": 0.5465691546569155,
      "grad_norm": 0.0,
      "learning_rate": 8.981050014123252e-06,
      "loss": 0.8056,
      "step": 2541
    },
    {
      "epoch": 0.5467842546784255,
      "grad_norm": 0.0,
      "learning_rate": 8.974119143875822e-06,
      "loss": 0.8353,
      "step": 2542
    },
    {
      "epoch": 0.5469993546999354,
      "grad_norm": 0.0,
      "learning_rate": 8.967188771636237e-06,
      "loss": 0.8248,
      "step": 2543
    },
    {
      "epoch": 0.5472144547214455,
      "grad_norm": 0.0,
      "learning_rate": 8.960258900768804e-06,
      "loss": 0.8765,
      "step": 2544
    },
    {
      "epoch": 0.5474295547429555,
      "grad_norm": 0.0,
      "learning_rate": 8.95332953463759e-06,
      "loss": 0.8365,
      "step": 2545
    },
    {
      "epoch": 0.5476446547644654,
      "grad_norm": 0.0,
      "learning_rate": 8.946400676606408e-06,
      "loss": 0.8718,
      "step": 2546
    },
    {
      "epoch": 0.5478597547859755,
      "grad_norm": 0.0,
      "learning_rate": 8.939472330038839e-06,
      "loss": 0.8499,
      "step": 2547
    },
    {
      "epoch": 0.5480748548074855,
      "grad_norm": 0.0,
      "learning_rate": 8.932544498298207e-06,
      "loss": 0.8396,
      "step": 2548
    },
    {
      "epoch": 0.5482899548289955,
      "grad_norm": 0.0,
      "learning_rate": 8.925617184747586e-06,
      "loss": 0.8084,
      "step": 2549
    },
    {
      "epoch": 0.5485050548505055,
      "grad_norm": 0.0,
      "learning_rate": 8.9186903927498e-06,
      "loss": 0.8743,
      "step": 2550
    },
    {
      "epoch": 0.5487201548720155,
      "grad_norm": 0.0,
      "learning_rate": 8.91176412566742e-06,
      "loss": 0.82,
      "step": 2551
    },
    {
      "epoch": 0.5489352548935255,
      "grad_norm": 0.0,
      "learning_rate": 8.904838386862758e-06,
      "loss": 0.804,
      "step": 2552
    },
    {
      "epoch": 0.5491503549150355,
      "grad_norm": 0.0,
      "learning_rate": 8.897913179697881e-06,
      "loss": 0.8273,
      "step": 2553
    },
    {
      "epoch": 0.5493654549365455,
      "grad_norm": 0.0,
      "learning_rate": 8.890988507534582e-06,
      "loss": 0.8758,
      "step": 2554
    },
    {
      "epoch": 0.5495805549580555,
      "grad_norm": 0.0,
      "learning_rate": 8.884064373734406e-06,
      "loss": 0.8227,
      "step": 2555
    },
    {
      "epoch": 0.5497956549795655,
      "grad_norm": 0.0,
      "learning_rate": 8.877140781658637e-06,
      "loss": 0.8791,
      "step": 2556
    },
    {
      "epoch": 0.5500107550010755,
      "grad_norm": 0.0,
      "learning_rate": 8.870217734668287e-06,
      "loss": 0.8542,
      "step": 2557
    },
    {
      "epoch": 0.5502258550225855,
      "grad_norm": 0.0,
      "learning_rate": 8.86329523612411e-06,
      "loss": 0.7885,
      "step": 2558
    },
    {
      "epoch": 0.5504409550440955,
      "grad_norm": 0.0,
      "learning_rate": 8.85637328938659e-06,
      "loss": 0.8615,
      "step": 2559
    },
    {
      "epoch": 0.5506560550656056,
      "grad_norm": 0.0,
      "learning_rate": 8.849451897815949e-06,
      "loss": 0.8867,
      "step": 2560
    },
    {
      "epoch": 0.5508711550871155,
      "grad_norm": 0.0,
      "learning_rate": 8.842531064772137e-06,
      "loss": 0.8717,
      "step": 2561
    },
    {
      "epoch": 0.5510862551086255,
      "grad_norm": 0.0,
      "learning_rate": 8.835610793614824e-06,
      "loss": 0.8679,
      "step": 2562
    },
    {
      "epoch": 0.5513013551301356,
      "grad_norm": 0.0,
      "learning_rate": 8.828691087703429e-06,
      "loss": 0.9234,
      "step": 2563
    },
    {
      "epoch": 0.5515164551516455,
      "grad_norm": 0.0,
      "learning_rate": 8.821771950397065e-06,
      "loss": 0.8331,
      "step": 2564
    },
    {
      "epoch": 0.5517315551731555,
      "grad_norm": 0.0,
      "learning_rate": 8.814853385054602e-06,
      "loss": 0.863,
      "step": 2565
    },
    {
      "epoch": 0.5519466551946656,
      "grad_norm": 0.0,
      "learning_rate": 8.80793539503461e-06,
      "loss": 0.9109,
      "step": 2566
    },
    {
      "epoch": 0.5521617552161755,
      "grad_norm": 0.0,
      "learning_rate": 8.801017983695383e-06,
      "loss": 0.9305,
      "step": 2567
    },
    {
      "epoch": 0.5523768552376855,
      "grad_norm": 0.0,
      "learning_rate": 8.794101154394943e-06,
      "loss": 0.8877,
      "step": 2568
    },
    {
      "epoch": 0.5525919552591956,
      "grad_norm": 0.0,
      "learning_rate": 8.787184910491028e-06,
      "loss": 0.9097,
      "step": 2569
    },
    {
      "epoch": 0.5528070552807055,
      "grad_norm": 0.0,
      "learning_rate": 8.78026925534108e-06,
      "loss": 0.8391,
      "step": 2570
    },
    {
      "epoch": 0.5530221553022155,
      "grad_norm": 0.0,
      "learning_rate": 8.77335419230227e-06,
      "loss": 0.8835,
      "step": 2571
    },
    {
      "epoch": 0.5532372553237256,
      "grad_norm": 0.0,
      "learning_rate": 8.766439724731469e-06,
      "loss": 0.8623,
      "step": 2572
    },
    {
      "epoch": 0.5534523553452355,
      "grad_norm": 0.0,
      "learning_rate": 8.759525855985265e-06,
      "loss": 0.9333,
      "step": 2573
    },
    {
      "epoch": 0.5536674553667456,
      "grad_norm": 0.0,
      "learning_rate": 8.752612589419963e-06,
      "loss": 0.8806,
      "step": 2574
    },
    {
      "epoch": 0.5538825553882556,
      "grad_norm": 0.0,
      "learning_rate": 8.745699928391555e-06,
      "loss": 0.8479,
      "step": 2575
    },
    {
      "epoch": 0.5540976554097655,
      "grad_norm": 0.0,
      "learning_rate": 8.738787876255763e-06,
      "loss": 0.8844,
      "step": 2576
    },
    {
      "epoch": 0.5543127554312756,
      "grad_norm": 0.0,
      "learning_rate": 8.731876436367994e-06,
      "loss": 0.885,
      "step": 2577
    },
    {
      "epoch": 0.5545278554527856,
      "grad_norm": 0.0,
      "learning_rate": 8.724965612083373e-06,
      "loss": 0.8964,
      "step": 2578
    },
    {
      "epoch": 0.5547429554742955,
      "grad_norm": 0.0,
      "learning_rate": 8.718055406756714e-06,
      "loss": 0.8766,
      "step": 2579
    },
    {
      "epoch": 0.5549580554958056,
      "grad_norm": 0.0,
      "learning_rate": 8.71114582374254e-06,
      "loss": 0.9016,
      "step": 2580
    },
    {
      "epoch": 0.5551731555173155,
      "grad_norm": 0.0,
      "learning_rate": 8.704236866395066e-06,
      "loss": 0.9271,
      "step": 2581
    },
    {
      "epoch": 0.5553882555388255,
      "grad_norm": 0.0,
      "learning_rate": 8.697328538068197e-06,
      "loss": 0.7914,
      "step": 2582
    },
    {
      "epoch": 0.5556033555603356,
      "grad_norm": 0.0,
      "learning_rate": 8.69042084211555e-06,
      "loss": 0.8581,
      "step": 2583
    },
    {
      "epoch": 0.5558184555818455,
      "grad_norm": 0.0,
      "learning_rate": 8.683513781890426e-06,
      "loss": 0.8297,
      "step": 2584
    },
    {
      "epoch": 0.5560335556033555,
      "grad_norm": 0.0,
      "learning_rate": 8.67660736074581e-06,
      "loss": 0.8946,
      "step": 2585
    },
    {
      "epoch": 0.5562486556248656,
      "grad_norm": 0.0,
      "learning_rate": 8.669701582034389e-06,
      "loss": 0.8954,
      "step": 2586
    },
    {
      "epoch": 0.5564637556463755,
      "grad_norm": 0.0,
      "learning_rate": 8.662796449108532e-06,
      "loss": 0.875,
      "step": 2587
    },
    {
      "epoch": 0.5566788556678856,
      "grad_norm": 0.0,
      "learning_rate": 8.655891965320292e-06,
      "loss": 0.8429,
      "step": 2588
    },
    {
      "epoch": 0.5568939556893956,
      "grad_norm": 0.0,
      "learning_rate": 8.648988134021416e-06,
      "loss": 0.8863,
      "step": 2589
    },
    {
      "epoch": 0.5571090557109055,
      "grad_norm": 0.0,
      "learning_rate": 8.64208495856332e-06,
      "loss": 0.8868,
      "step": 2590
    },
    {
      "epoch": 0.5573241557324156,
      "grad_norm": 0.0,
      "learning_rate": 8.635182442297114e-06,
      "loss": 0.8534,
      "step": 2591
    },
    {
      "epoch": 0.5575392557539256,
      "grad_norm": 0.0,
      "learning_rate": 8.62828058857359e-06,
      "loss": 0.888,
      "step": 2592
    },
    {
      "epoch": 0.5577543557754355,
      "grad_norm": 0.0,
      "learning_rate": 8.621379400743206e-06,
      "loss": 0.8966,
      "step": 2593
    },
    {
      "epoch": 0.5579694557969456,
      "grad_norm": 0.0,
      "learning_rate": 8.614478882156104e-06,
      "loss": 0.9128,
      "step": 2594
    },
    {
      "epoch": 0.5581845558184556,
      "grad_norm": 0.0,
      "learning_rate": 8.607579036162101e-06,
      "loss": 0.8651,
      "step": 2595
    },
    {
      "epoch": 0.5583996558399655,
      "grad_norm": 0.0,
      "learning_rate": 8.600679866110682e-06,
      "loss": 0.874,
      "step": 2596
    },
    {
      "epoch": 0.5586147558614756,
      "grad_norm": 0.0,
      "learning_rate": 8.59378137535102e-06,
      "loss": 0.8994,
      "step": 2597
    },
    {
      "epoch": 0.5588298558829856,
      "grad_norm": 0.0,
      "learning_rate": 8.586883567231935e-06,
      "loss": 0.837,
      "step": 2598
    },
    {
      "epoch": 0.5590449559044955,
      "grad_norm": 0.0,
      "learning_rate": 8.579986445101932e-06,
      "loss": 0.8963,
      "step": 2599
    },
    {
      "epoch": 0.5592600559260056,
      "grad_norm": 0.0,
      "learning_rate": 8.573090012309181e-06,
      "loss": 0.9604,
      "step": 2600
    },
    {
      "epoch": 0.5594751559475156,
      "grad_norm": 0.0,
      "learning_rate": 8.566194272201515e-06,
      "loss": 0.8098,
      "step": 2601
    },
    {
      "epoch": 0.5596902559690256,
      "grad_norm": 0.0,
      "learning_rate": 8.559299228126427e-06,
      "loss": 0.8612,
      "step": 2602
    },
    {
      "epoch": 0.5599053559905356,
      "grad_norm": 0.0,
      "learning_rate": 8.552404883431075e-06,
      "loss": 0.7856,
      "step": 2603
    },
    {
      "epoch": 0.5601204560120456,
      "grad_norm": 0.0,
      "learning_rate": 8.54551124146228e-06,
      "loss": 0.9112,
      "step": 2604
    },
    {
      "epoch": 0.5603355560335556,
      "grad_norm": 0.0,
      "learning_rate": 8.538618305566523e-06,
      "loss": 0.8754,
      "step": 2605
    },
    {
      "epoch": 0.5605506560550656,
      "grad_norm": 0.0,
      "learning_rate": 8.531726079089934e-06,
      "loss": 0.8725,
      "step": 2606
    },
    {
      "epoch": 0.5607657560765756,
      "grad_norm": 0.0,
      "learning_rate": 8.524834565378309e-06,
      "loss": 0.8443,
      "step": 2607
    },
    {
      "epoch": 0.5609808560980856,
      "grad_norm": 0.0,
      "learning_rate": 8.517943767777089e-06,
      "loss": 0.8287,
      "step": 2608
    },
    {
      "epoch": 0.5611959561195956,
      "grad_norm": 0.0,
      "learning_rate": 8.511053689631371e-06,
      "loss": 0.87,
      "step": 2609
    },
    {
      "epoch": 0.5614110561411056,
      "grad_norm": 0.0,
      "learning_rate": 8.504164334285909e-06,
      "loss": 0.8951,
      "step": 2610
    },
    {
      "epoch": 0.5616261561626156,
      "grad_norm": 0.0,
      "learning_rate": 8.497275705085091e-06,
      "loss": 0.8959,
      "step": 2611
    },
    {
      "epoch": 0.5618412561841256,
      "grad_norm": 0.0,
      "learning_rate": 8.490387805372965e-06,
      "loss": 0.8058,
      "step": 2612
    },
    {
      "epoch": 0.5620563562056357,
      "grad_norm": 0.0,
      "learning_rate": 8.48350063849323e-06,
      "loss": 0.8447,
      "step": 2613
    },
    {
      "epoch": 0.5622714562271456,
      "grad_norm": 0.0,
      "learning_rate": 8.476614207789208e-06,
      "loss": 0.8771,
      "step": 2614
    },
    {
      "epoch": 0.5624865562486556,
      "grad_norm": 0.0,
      "learning_rate": 8.469728516603886e-06,
      "loss": 0.9087,
      "step": 2615
    },
    {
      "epoch": 0.5627016562701657,
      "grad_norm": 0.0,
      "learning_rate": 8.462843568279877e-06,
      "loss": 0.8484,
      "step": 2616
    },
    {
      "epoch": 0.5629167562916756,
      "grad_norm": 0.0,
      "learning_rate": 8.455959366159439e-06,
      "loss": 0.9005,
      "step": 2617
    },
    {
      "epoch": 0.5631318563131856,
      "grad_norm": 0.0,
      "learning_rate": 8.449075913584472e-06,
      "loss": 0.8341,
      "step": 2618
    },
    {
      "epoch": 0.5633469563346957,
      "grad_norm": 0.0,
      "learning_rate": 8.4421932138965e-06,
      "loss": 0.8462,
      "step": 2619
    },
    {
      "epoch": 0.5635620563562056,
      "grad_norm": 0.0,
      "learning_rate": 8.435311270436697e-06,
      "loss": 0.8988,
      "step": 2620
    },
    {
      "epoch": 0.5637771563777156,
      "grad_norm": 0.0,
      "learning_rate": 8.428430086545857e-06,
      "loss": 0.874,
      "step": 2621
    },
    {
      "epoch": 0.5639922563992257,
      "grad_norm": 0.0,
      "learning_rate": 8.421549665564412e-06,
      "loss": 0.8119,
      "step": 2622
    },
    {
      "epoch": 0.5642073564207356,
      "grad_norm": 0.0,
      "learning_rate": 8.414670010832425e-06,
      "loss": 0.8327,
      "step": 2623
    },
    {
      "epoch": 0.5644224564422456,
      "grad_norm": 0.0,
      "learning_rate": 8.407791125689577e-06,
      "loss": 0.865,
      "step": 2624
    },
    {
      "epoch": 0.5646375564637557,
      "grad_norm": 0.0,
      "learning_rate": 8.400913013475186e-06,
      "loss": 0.8528,
      "step": 2625
    },
    {
      "epoch": 0.5648526564852656,
      "grad_norm": 0.0,
      "learning_rate": 8.394035677528197e-06,
      "loss": 0.8631,
      "step": 2626
    },
    {
      "epoch": 0.5650677565067757,
      "grad_norm": 0.0,
      "learning_rate": 8.387159121187163e-06,
      "loss": 0.8367,
      "step": 2627
    },
    {
      "epoch": 0.5652828565282857,
      "grad_norm": 0.0,
      "learning_rate": 8.380283347790275e-06,
      "loss": 0.8277,
      "step": 2628
    },
    {
      "epoch": 0.5654979565497956,
      "grad_norm": 0.0,
      "learning_rate": 8.37340836067533e-06,
      "loss": 0.8245,
      "step": 2629
    },
    {
      "epoch": 0.5657130565713057,
      "grad_norm": 0.0,
      "learning_rate": 8.366534163179757e-06,
      "loss": 0.9448,
      "step": 2630
    },
    {
      "epoch": 0.5659281565928157,
      "grad_norm": 0.0,
      "learning_rate": 8.359660758640595e-06,
      "loss": 0.923,
      "step": 2631
    },
    {
      "epoch": 0.5661432566143256,
      "grad_norm": 0.0,
      "learning_rate": 8.352788150394489e-06,
      "loss": 0.9228,
      "step": 2632
    },
    {
      "epoch": 0.5663583566358357,
      "grad_norm": 0.0,
      "learning_rate": 8.345916341777716e-06,
      "loss": 0.8602,
      "step": 2633
    },
    {
      "epoch": 0.5665734566573457,
      "grad_norm": 0.0,
      "learning_rate": 8.339045336126149e-06,
      "loss": 0.8616,
      "step": 2634
    },
    {
      "epoch": 0.5667885566788556,
      "grad_norm": 0.0,
      "learning_rate": 8.332175136775278e-06,
      "loss": 0.9217,
      "step": 2635
    },
    {
      "epoch": 0.5670036567003657,
      "grad_norm": 0.0,
      "learning_rate": 8.325305747060208e-06,
      "loss": 0.9061,
      "step": 2636
    },
    {
      "epoch": 0.5672187567218757,
      "grad_norm": 0.0,
      "learning_rate": 8.318437170315636e-06,
      "loss": 0.8975,
      "step": 2637
    },
    {
      "epoch": 0.5674338567433856,
      "grad_norm": 0.0,
      "learning_rate": 8.311569409875876e-06,
      "loss": 0.8693,
      "step": 2638
    },
    {
      "epoch": 0.5676489567648957,
      "grad_norm": 0.0,
      "learning_rate": 8.30470246907484e-06,
      "loss": 0.8535,
      "step": 2639
    },
    {
      "epoch": 0.5678640567864057,
      "grad_norm": 0.0,
      "learning_rate": 8.297836351246044e-06,
      "loss": 0.8347,
      "step": 2640
    },
    {
      "epoch": 0.5680791568079157,
      "grad_norm": 0.0,
      "learning_rate": 8.29097105972261e-06,
      "loss": 0.8176,
      "step": 2641
    },
    {
      "epoch": 0.5682942568294257,
      "grad_norm": 0.0,
      "learning_rate": 8.284106597837244e-06,
      "loss": 0.8421,
      "step": 2642
    },
    {
      "epoch": 0.5685093568509357,
      "grad_norm": 0.0,
      "learning_rate": 8.277242968922267e-06,
      "loss": 0.8442,
      "step": 2643
    },
    {
      "epoch": 0.5687244568724457,
      "grad_norm": 0.0,
      "learning_rate": 8.270380176309584e-06,
      "loss": 0.7927,
      "step": 2644
    },
    {
      "epoch": 0.5689395568939557,
      "grad_norm": 0.0,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.8576,
      "step": 2645
    },
    {
      "epoch": 0.5691546569154657,
      "grad_norm": 0.0,
      "learning_rate": 8.256657113316705e-06,
      "loss": 0.8623,
      "step": 2646
    },
    {
      "epoch": 0.5693697569369757,
      "grad_norm": 0.0,
      "learning_rate": 8.249796849598286e-06,
      "loss": 0.8651,
      "step": 2647
    },
    {
      "epoch": 0.5695848569584857,
      "grad_norm": 0.0,
      "learning_rate": 8.242937435505718e-06,
      "loss": 0.8166,
      "step": 2648
    },
    {
      "epoch": 0.5697999569799957,
      "grad_norm": 0.0,
      "learning_rate": 8.236078874368871e-06,
      "loss": 0.8651,
      "step": 2649
    },
    {
      "epoch": 0.5700150570015057,
      "grad_norm": 0.0,
      "learning_rate": 8.22922116951718e-06,
      "loss": 0.8719,
      "step": 2650
    },
    {
      "epoch": 0.5702301570230157,
      "grad_norm": 0.0,
      "learning_rate": 8.222364324279689e-06,
      "loss": 0.8648,
      "step": 2651
    },
    {
      "epoch": 0.5704452570445258,
      "grad_norm": 0.0,
      "learning_rate": 8.215508341985007e-06,
      "loss": 0.8439,
      "step": 2652
    },
    {
      "epoch": 0.5706603570660357,
      "grad_norm": 0.0,
      "learning_rate": 8.208653225961329e-06,
      "loss": 0.8927,
      "step": 2653
    },
    {
      "epoch": 0.5708754570875457,
      "grad_norm": 0.0,
      "learning_rate": 8.201798979536438e-06,
      "loss": 0.9096,
      "step": 2654
    },
    {
      "epoch": 0.5710905571090558,
      "grad_norm": 0.0,
      "learning_rate": 8.194945606037679e-06,
      "loss": 0.9074,
      "step": 2655
    },
    {
      "epoch": 0.5713056571305657,
      "grad_norm": 0.0,
      "learning_rate": 8.188093108791988e-06,
      "loss": 0.8662,
      "step": 2656
    },
    {
      "epoch": 0.5715207571520757,
      "grad_norm": 0.0,
      "learning_rate": 8.18124149112587e-06,
      "loss": 0.9003,
      "step": 2657
    },
    {
      "epoch": 0.5717358571735858,
      "grad_norm": 0.0,
      "learning_rate": 8.1743907563654e-06,
      "loss": 0.903,
      "step": 2658
    },
    {
      "epoch": 0.5719509571950957,
      "grad_norm": 0.0,
      "learning_rate": 8.167540907836229e-06,
      "loss": 0.8488,
      "step": 2659
    },
    {
      "epoch": 0.5721660572166057,
      "grad_norm": 0.0,
      "learning_rate": 8.160691948863572e-06,
      "loss": 0.7802,
      "step": 2660
    },
    {
      "epoch": 0.5723811572381158,
      "grad_norm": 0.0,
      "learning_rate": 8.153843882772217e-06,
      "loss": 0.865,
      "step": 2661
    },
    {
      "epoch": 0.5725962572596257,
      "grad_norm": 0.0,
      "learning_rate": 8.146996712886524e-06,
      "loss": 0.8329,
      "step": 2662
    },
    {
      "epoch": 0.5728113572811357,
      "grad_norm": 0.0,
      "learning_rate": 8.1401504425304e-06,
      "loss": 0.8109,
      "step": 2663
    },
    {
      "epoch": 0.5730264573026457,
      "grad_norm": 0.0,
      "learning_rate": 8.133305075027338e-06,
      "loss": 0.8447,
      "step": 2664
    },
    {
      "epoch": 0.5732415573241557,
      "grad_norm": 0.0,
      "learning_rate": 8.126460613700373e-06,
      "loss": 0.8525,
      "step": 2665
    },
    {
      "epoch": 0.5734566573456658,
      "grad_norm": 0.0,
      "learning_rate": 8.119617061872115e-06,
      "loss": 0.8316,
      "step": 2666
    },
    {
      "epoch": 0.5736717573671757,
      "grad_norm": 0.0,
      "learning_rate": 8.112774422864724e-06,
      "loss": 0.8237,
      "step": 2667
    },
    {
      "epoch": 0.5738868573886857,
      "grad_norm": 0.0,
      "learning_rate": 8.105932699999916e-06,
      "loss": 0.8759,
      "step": 2668
    },
    {
      "epoch": 0.5741019574101958,
      "grad_norm": 0.0,
      "learning_rate": 8.099091896598964e-06,
      "loss": 0.9352,
      "step": 2669
    },
    {
      "epoch": 0.5743170574317057,
      "grad_norm": 0.0,
      "learning_rate": 8.092252015982706e-06,
      "loss": 0.8651,
      "step": 2670
    },
    {
      "epoch": 0.5745321574532157,
      "grad_norm": 0.0,
      "learning_rate": 8.085413061471511e-06,
      "loss": 0.8765,
      "step": 2671
    },
    {
      "epoch": 0.5747472574747258,
      "grad_norm": 0.0,
      "learning_rate": 8.07857503638532e-06,
      "loss": 0.9265,
      "step": 2672
    },
    {
      "epoch": 0.5749623574962357,
      "grad_norm": 0.0,
      "learning_rate": 8.0717379440436e-06,
      "loss": 0.8242,
      "step": 2673
    },
    {
      "epoch": 0.5751774575177457,
      "grad_norm": 0.0,
      "learning_rate": 8.064901787765382e-06,
      "loss": 0.8316,
      "step": 2674
    },
    {
      "epoch": 0.5753925575392558,
      "grad_norm": 0.0,
      "learning_rate": 8.058066570869244e-06,
      "loss": 0.8467,
      "step": 2675
    },
    {
      "epoch": 0.5756076575607657,
      "grad_norm": 0.0,
      "learning_rate": 8.051232296673292e-06,
      "loss": 0.7885,
      "step": 2676
    },
    {
      "epoch": 0.5758227575822757,
      "grad_norm": 0.0,
      "learning_rate": 8.044398968495195e-06,
      "loss": 0.9196,
      "step": 2677
    },
    {
      "epoch": 0.5760378576037858,
      "grad_norm": 0.0,
      "learning_rate": 8.037566589652141e-06,
      "loss": 0.8533,
      "step": 2678
    },
    {
      "epoch": 0.5762529576252957,
      "grad_norm": 0.0,
      "learning_rate": 8.030735163460877e-06,
      "loss": 0.8657,
      "step": 2679
    },
    {
      "epoch": 0.5764680576468058,
      "grad_norm": 0.0,
      "learning_rate": 8.023904693237675e-06,
      "loss": 0.8595,
      "step": 2680
    },
    {
      "epoch": 0.5766831576683158,
      "grad_norm": 0.0,
      "learning_rate": 8.017075182298348e-06,
      "loss": 0.8275,
      "step": 2681
    },
    {
      "epoch": 0.5768982576898257,
      "grad_norm": 0.0,
      "learning_rate": 8.01024663395824e-06,
      "loss": 0.8751,
      "step": 2682
    },
    {
      "epoch": 0.5771133577113358,
      "grad_norm": 0.0,
      "learning_rate": 8.003419051532232e-06,
      "loss": 0.8648,
      "step": 2683
    },
    {
      "epoch": 0.5773284577328458,
      "grad_norm": 0.0,
      "learning_rate": 7.996592438334728e-06,
      "loss": 0.8202,
      "step": 2684
    },
    {
      "epoch": 0.5775435577543557,
      "grad_norm": 0.0,
      "learning_rate": 7.989766797679678e-06,
      "loss": 0.9222,
      "step": 2685
    },
    {
      "epoch": 0.5777586577758658,
      "grad_norm": 0.0,
      "learning_rate": 7.98294213288054e-06,
      "loss": 0.9228,
      "step": 2686
    },
    {
      "epoch": 0.5779737577973758,
      "grad_norm": 0.0,
      "learning_rate": 7.97611844725031e-06,
      "loss": 0.9316,
      "step": 2687
    },
    {
      "epoch": 0.5781888578188857,
      "grad_norm": 0.0,
      "learning_rate": 7.969295744101512e-06,
      "loss": 0.9193,
      "step": 2688
    },
    {
      "epoch": 0.5784039578403958,
      "grad_norm": 0.0,
      "learning_rate": 7.962474026746176e-06,
      "loss": 0.8412,
      "step": 2689
    },
    {
      "epoch": 0.5786190578619058,
      "grad_norm": 0.0,
      "learning_rate": 7.955653298495876e-06,
      "loss": 0.8426,
      "step": 2690
    },
    {
      "epoch": 0.5788341578834157,
      "grad_norm": 0.0,
      "learning_rate": 7.948833562661686e-06,
      "loss": 0.843,
      "step": 2691
    },
    {
      "epoch": 0.5790492579049258,
      "grad_norm": 0.0,
      "learning_rate": 7.94201482255421e-06,
      "loss": 0.8602,
      "step": 2692
    },
    {
      "epoch": 0.5792643579264358,
      "grad_norm": 0.0,
      "learning_rate": 7.935197081483568e-06,
      "loss": 0.8297,
      "step": 2693
    },
    {
      "epoch": 0.5794794579479458,
      "grad_norm": 0.0,
      "learning_rate": 7.928380342759389e-06,
      "loss": 0.8953,
      "step": 2694
    },
    {
      "epoch": 0.5796945579694558,
      "grad_norm": 0.0,
      "learning_rate": 7.92156460969082e-06,
      "loss": 0.8546,
      "step": 2695
    },
    {
      "epoch": 0.5799096579909658,
      "grad_norm": 0.0,
      "learning_rate": 7.914749885586517e-06,
      "loss": 0.8262,
      "step": 2696
    },
    {
      "epoch": 0.5801247580124758,
      "grad_norm": 0.0,
      "learning_rate": 7.90793617375465e-06,
      "loss": 0.9249,
      "step": 2697
    },
    {
      "epoch": 0.5803398580339858,
      "grad_norm": 0.0,
      "learning_rate": 7.901123477502895e-06,
      "loss": 0.8565,
      "step": 2698
    },
    {
      "epoch": 0.5805549580554958,
      "grad_norm": 0.0,
      "learning_rate": 7.894311800138432e-06,
      "loss": 0.7988,
      "step": 2699
    },
    {
      "epoch": 0.5807700580770058,
      "grad_norm": 0.0,
      "learning_rate": 7.887501144967954e-06,
      "loss": 0.8485,
      "step": 2700
    },
    {
      "epoch": 0.5809851580985158,
      "grad_norm": 0.0,
      "learning_rate": 7.880691515297656e-06,
      "loss": 0.8233,
      "step": 2701
    },
    {
      "epoch": 0.5812002581200258,
      "grad_norm": 0.0,
      "learning_rate": 7.87388291443323e-06,
      "loss": 0.8323,
      "step": 2702
    },
    {
      "epoch": 0.5814153581415358,
      "grad_norm": 0.0,
      "learning_rate": 7.867075345679868e-06,
      "loss": 0.8995,
      "step": 2703
    },
    {
      "epoch": 0.5816304581630458,
      "grad_norm": 0.0,
      "learning_rate": 7.860268812342265e-06,
      "loss": 0.8623,
      "step": 2704
    },
    {
      "epoch": 0.5818455581845559,
      "grad_norm": 0.0,
      "learning_rate": 7.853463317724614e-06,
      "loss": 0.8695,
      "step": 2705
    },
    {
      "epoch": 0.5820606582060658,
      "grad_norm": 0.0,
      "learning_rate": 7.846658865130608e-06,
      "loss": 0.8502,
      "step": 2706
    },
    {
      "epoch": 0.5822757582275758,
      "grad_norm": 0.0,
      "learning_rate": 7.839855457863419e-06,
      "loss": 0.8688,
      "step": 2707
    },
    {
      "epoch": 0.5824908582490859,
      "grad_norm": 0.0,
      "learning_rate": 7.83305309922573e-06,
      "loss": 0.8386,
      "step": 2708
    },
    {
      "epoch": 0.5827059582705958,
      "grad_norm": 0.0,
      "learning_rate": 7.826251792519697e-06,
      "loss": 0.8422,
      "step": 2709
    },
    {
      "epoch": 0.5829210582921058,
      "grad_norm": 0.0,
      "learning_rate": 7.819451541046982e-06,
      "loss": 0.9093,
      "step": 2710
    },
    {
      "epoch": 0.5831361583136159,
      "grad_norm": 0.0,
      "learning_rate": 7.812652348108726e-06,
      "loss": 0.8475,
      "step": 2711
    },
    {
      "epoch": 0.5833512583351258,
      "grad_norm": 0.0,
      "learning_rate": 7.805854217005553e-06,
      "loss": 0.8771,
      "step": 2712
    },
    {
      "epoch": 0.5835663583566358,
      "grad_norm": 0.0,
      "learning_rate": 7.799057151037577e-06,
      "loss": 0.8727,
      "step": 2713
    },
    {
      "epoch": 0.5837814583781459,
      "grad_norm": 0.0,
      "learning_rate": 7.792261153504403e-06,
      "loss": 0.8516,
      "step": 2714
    },
    {
      "epoch": 0.5839965583996558,
      "grad_norm": 0.0,
      "learning_rate": 7.785466227705093e-06,
      "loss": 0.8151,
      "step": 2715
    },
    {
      "epoch": 0.5842116584211658,
      "grad_norm": 0.0,
      "learning_rate": 7.778672376938217e-06,
      "loss": 0.8482,
      "step": 2716
    },
    {
      "epoch": 0.5844267584426759,
      "grad_norm": 0.0,
      "learning_rate": 7.771879604501802e-06,
      "loss": 0.8717,
      "step": 2717
    },
    {
      "epoch": 0.5846418584641858,
      "grad_norm": 0.0,
      "learning_rate": 7.76508791369336e-06,
      "loss": 0.8036,
      "step": 2718
    },
    {
      "epoch": 0.5848569584856959,
      "grad_norm": 0.0,
      "learning_rate": 7.758297307809882e-06,
      "loss": 0.7963,
      "step": 2719
    },
    {
      "epoch": 0.5850720585072059,
      "grad_norm": 0.0,
      "learning_rate": 7.751507790147825e-06,
      "loss": 0.8934,
      "step": 2720
    },
    {
      "epoch": 0.5852871585287158,
      "grad_norm": 0.0,
      "learning_rate": 7.744719364003123e-06,
      "loss": 0.8517,
      "step": 2721
    },
    {
      "epoch": 0.5855022585502259,
      "grad_norm": 0.0,
      "learning_rate": 7.73793203267117e-06,
      "loss": 0.854,
      "step": 2722
    },
    {
      "epoch": 0.5857173585717359,
      "grad_norm": 0.0,
      "learning_rate": 7.731145799446846e-06,
      "loss": 0.883,
      "step": 2723
    },
    {
      "epoch": 0.5859324585932458,
      "grad_norm": 0.0,
      "learning_rate": 7.724360667624482e-06,
      "loss": 0.7992,
      "step": 2724
    },
    {
      "epoch": 0.5861475586147559,
      "grad_norm": 0.0,
      "learning_rate": 7.717576640497883e-06,
      "loss": 0.8688,
      "step": 2725
    },
    {
      "epoch": 0.5863626586362659,
      "grad_norm": 0.0,
      "learning_rate": 7.710793721360312e-06,
      "loss": 0.8981,
      "step": 2726
    },
    {
      "epoch": 0.5865777586577758,
      "grad_norm": 0.0,
      "learning_rate": 7.704011913504503e-06,
      "loss": 0.9244,
      "step": 2727
    },
    {
      "epoch": 0.5867928586792859,
      "grad_norm": 0.0,
      "learning_rate": 7.697231220222638e-06,
      "loss": 0.818,
      "step": 2728
    },
    {
      "epoch": 0.5870079587007959,
      "grad_norm": 0.0,
      "learning_rate": 7.690451644806372e-06,
      "loss": 0.8948,
      "step": 2729
    },
    {
      "epoch": 0.5872230587223058,
      "grad_norm": 0.0,
      "learning_rate": 7.683673190546804e-06,
      "loss": 0.7731,
      "step": 2730
    },
    {
      "epoch": 0.5874381587438159,
      "grad_norm": 0.0,
      "learning_rate": 7.676895860734497e-06,
      "loss": 0.8302,
      "step": 2731
    },
    {
      "epoch": 0.5876532587653259,
      "grad_norm": 0.0,
      "learning_rate": 7.670119658659469e-06,
      "loss": 0.8866,
      "step": 2732
    },
    {
      "epoch": 0.5878683587868359,
      "grad_norm": 0.0,
      "learning_rate": 7.663344587611181e-06,
      "loss": 0.7838,
      "step": 2733
    },
    {
      "epoch": 0.5880834588083459,
      "grad_norm": 0.0,
      "learning_rate": 7.65657065087856e-06,
      "loss": 0.9152,
      "step": 2734
    },
    {
      "epoch": 0.5882985588298559,
      "grad_norm": 0.0,
      "learning_rate": 7.649797851749965e-06,
      "loss": 0.8715,
      "step": 2735
    },
    {
      "epoch": 0.5885136588513659,
      "grad_norm": 0.0,
      "learning_rate": 7.643026193513214e-06,
      "loss": 0.8743,
      "step": 2736
    },
    {
      "epoch": 0.5887287588728759,
      "grad_norm": 0.0,
      "learning_rate": 7.636255679455575e-06,
      "loss": 0.8119,
      "step": 2737
    },
    {
      "epoch": 0.5889438588943859,
      "grad_norm": 0.0,
      "learning_rate": 7.629486312863749e-06,
      "loss": 0.7652,
      "step": 2738
    },
    {
      "epoch": 0.5891589589158959,
      "grad_norm": 0.0,
      "learning_rate": 7.622718097023884e-06,
      "loss": 0.8223,
      "step": 2739
    },
    {
      "epoch": 0.5893740589374059,
      "grad_norm": 0.0,
      "learning_rate": 7.615951035221576e-06,
      "loss": 0.8735,
      "step": 2740
    },
    {
      "epoch": 0.589589158958916,
      "grad_norm": 0.0,
      "learning_rate": 7.6091851307418475e-06,
      "loss": 0.8891,
      "step": 2741
    },
    {
      "epoch": 0.5898042589804259,
      "grad_norm": 0.0,
      "learning_rate": 7.602420386869177e-06,
      "loss": 0.8816,
      "step": 2742
    },
    {
      "epoch": 0.5900193590019359,
      "grad_norm": 0.0,
      "learning_rate": 7.59565680688746e-06,
      "loss": 0.78,
      "step": 2743
    },
    {
      "epoch": 0.590234459023446,
      "grad_norm": 0.0,
      "learning_rate": 7.588894394080044e-06,
      "loss": 0.8248,
      "step": 2744
    },
    {
      "epoch": 0.5904495590449559,
      "grad_norm": 0.0,
      "learning_rate": 7.5821331517297005e-06,
      "loss": 0.8889,
      "step": 2745
    },
    {
      "epoch": 0.5906646590664659,
      "grad_norm": 0.0,
      "learning_rate": 7.575373083118633e-06,
      "loss": 0.9106,
      "step": 2746
    },
    {
      "epoch": 0.590879759087976,
      "grad_norm": 0.0,
      "learning_rate": 7.568614191528482e-06,
      "loss": 0.8522,
      "step": 2747
    },
    {
      "epoch": 0.5910948591094859,
      "grad_norm": 0.0,
      "learning_rate": 7.5618564802403025e-06,
      "loss": 0.8594,
      "step": 2748
    },
    {
      "epoch": 0.5913099591309959,
      "grad_norm": 0.0,
      "learning_rate": 7.555099952534593e-06,
      "loss": 0.8711,
      "step": 2749
    },
    {
      "epoch": 0.5915250591525059,
      "grad_norm": 0.0,
      "learning_rate": 7.548344611691272e-06,
      "loss": 0.8924,
      "step": 2750
    },
    {
      "epoch": 0.5917401591740159,
      "grad_norm": 0.0,
      "learning_rate": 7.541590460989674e-06,
      "loss": 0.8712,
      "step": 2751
    },
    {
      "epoch": 0.5919552591955259,
      "grad_norm": 0.0,
      "learning_rate": 7.534837503708567e-06,
      "loss": 0.8506,
      "step": 2752
    },
    {
      "epoch": 0.5921703592170359,
      "grad_norm": 0.0,
      "learning_rate": 7.528085743126132e-06,
      "loss": 0.8605,
      "step": 2753
    },
    {
      "epoch": 0.5923854592385459,
      "grad_norm": 0.0,
      "learning_rate": 7.521335182519968e-06,
      "loss": 0.9168,
      "step": 2754
    },
    {
      "epoch": 0.592600559260056,
      "grad_norm": 0.0,
      "learning_rate": 7.514585825167106e-06,
      "loss": 0.8068,
      "step": 2755
    },
    {
      "epoch": 0.5928156592815659,
      "grad_norm": 0.0,
      "learning_rate": 7.507837674343969e-06,
      "loss": 0.8453,
      "step": 2756
    },
    {
      "epoch": 0.5930307593030759,
      "grad_norm": 0.0,
      "learning_rate": 7.501090733326413e-06,
      "loss": 0.8881,
      "step": 2757
    },
    {
      "epoch": 0.593245859324586,
      "grad_norm": 0.0,
      "learning_rate": 7.494345005389705e-06,
      "loss": 0.8327,
      "step": 2758
    },
    {
      "epoch": 0.5934609593460959,
      "grad_norm": 0.0,
      "learning_rate": 7.487600493808513e-06,
      "loss": 0.9437,
      "step": 2759
    },
    {
      "epoch": 0.5936760593676059,
      "grad_norm": 0.0,
      "learning_rate": 7.480857201856923e-06,
      "loss": 0.875,
      "step": 2760
    },
    {
      "epoch": 0.593891159389116,
      "grad_norm": 0.0,
      "learning_rate": 7.474115132808425e-06,
      "loss": 0.8221,
      "step": 2761
    },
    {
      "epoch": 0.5941062594106259,
      "grad_norm": 0.0,
      "learning_rate": 7.467374289935917e-06,
      "loss": 0.8822,
      "step": 2762
    },
    {
      "epoch": 0.5943213594321359,
      "grad_norm": 0.0,
      "learning_rate": 7.460634676511706e-06,
      "loss": 0.7984,
      "step": 2763
    },
    {
      "epoch": 0.594536459453646,
      "grad_norm": 0.0,
      "learning_rate": 7.4538962958074915e-06,
      "loss": 0.799,
      "step": 2764
    },
    {
      "epoch": 0.5947515594751559,
      "grad_norm": 0.0,
      "learning_rate": 7.447159151094388e-06,
      "loss": 0.894,
      "step": 2765
    },
    {
      "epoch": 0.5949666594966659,
      "grad_norm": 0.0,
      "learning_rate": 7.440423245642892e-06,
      "loss": 0.8448,
      "step": 2766
    },
    {
      "epoch": 0.595181759518176,
      "grad_norm": 0.0,
      "learning_rate": 7.43368858272292e-06,
      "loss": 0.899,
      "step": 2767
    },
    {
      "epoch": 0.5953968595396859,
      "grad_norm": 0.0,
      "learning_rate": 7.426955165603773e-06,
      "loss": 0.8583,
      "step": 2768
    },
    {
      "epoch": 0.595611959561196,
      "grad_norm": 0.0,
      "learning_rate": 7.420222997554142e-06,
      "loss": 0.8848,
      "step": 2769
    },
    {
      "epoch": 0.595827059582706,
      "grad_norm": 0.0,
      "learning_rate": 7.413492081842122e-06,
      "loss": 0.8996,
      "step": 2770
    },
    {
      "epoch": 0.5960421596042159,
      "grad_norm": 0.0,
      "learning_rate": 7.406762421735203e-06,
      "loss": 0.9123,
      "step": 2771
    },
    {
      "epoch": 0.596257259625726,
      "grad_norm": 0.0,
      "learning_rate": 7.400034020500249e-06,
      "loss": 0.79,
      "step": 2772
    },
    {
      "epoch": 0.596472359647236,
      "grad_norm": 0.0,
      "learning_rate": 7.393306881403531e-06,
      "loss": 0.8646,
      "step": 2773
    },
    {
      "epoch": 0.5966874596687459,
      "grad_norm": 0.0,
      "learning_rate": 7.386581007710694e-06,
      "loss": 0.883,
      "step": 2774
    },
    {
      "epoch": 0.596902559690256,
      "grad_norm": 0.0,
      "learning_rate": 7.379856402686773e-06,
      "loss": 0.8428,
      "step": 2775
    },
    {
      "epoch": 0.597117659711766,
      "grad_norm": 0.0,
      "learning_rate": 7.3731330695961945e-06,
      "loss": 0.8113,
      "step": 2776
    },
    {
      "epoch": 0.5973327597332759,
      "grad_norm": 0.0,
      "learning_rate": 7.366411011702754e-06,
      "loss": 0.8849,
      "step": 2777
    },
    {
      "epoch": 0.597547859754786,
      "grad_norm": 0.0,
      "learning_rate": 7.35969023226964e-06,
      "loss": 0.807,
      "step": 2778
    },
    {
      "epoch": 0.597762959776296,
      "grad_norm": 0.0,
      "learning_rate": 7.352970734559412e-06,
      "loss": 0.905,
      "step": 2779
    },
    {
      "epoch": 0.5979780597978059,
      "grad_norm": 0.0,
      "learning_rate": 7.346252521834012e-06,
      "loss": 0.8298,
      "step": 2780
    },
    {
      "epoch": 0.598193159819316,
      "grad_norm": 0.0,
      "learning_rate": 7.339535597354762e-06,
      "loss": 0.7963,
      "step": 2781
    },
    {
      "epoch": 0.598408259840826,
      "grad_norm": 0.0,
      "learning_rate": 7.332819964382348e-06,
      "loss": 0.9024,
      "step": 2782
    },
    {
      "epoch": 0.598623359862336,
      "grad_norm": 0.0,
      "learning_rate": 7.326105626176835e-06,
      "loss": 0.8406,
      "step": 2783
    },
    {
      "epoch": 0.598838459883846,
      "grad_norm": 0.0,
      "learning_rate": 7.319392585997655e-06,
      "loss": 0.8384,
      "step": 2784
    },
    {
      "epoch": 0.599053559905356,
      "grad_norm": 0.0,
      "learning_rate": 7.312680847103622e-06,
      "loss": 0.8416,
      "step": 2785
    },
    {
      "epoch": 0.599268659926866,
      "grad_norm": 0.0,
      "learning_rate": 7.30597041275291e-06,
      "loss": 0.8361,
      "step": 2786
    },
    {
      "epoch": 0.599483759948376,
      "grad_norm": 0.0,
      "learning_rate": 7.299261286203052e-06,
      "loss": 0.8793,
      "step": 2787
    },
    {
      "epoch": 0.599698859969886,
      "grad_norm": 0.0,
      "learning_rate": 7.292553470710962e-06,
      "loss": 0.811,
      "step": 2788
    },
    {
      "epoch": 0.599913959991396,
      "grad_norm": 0.0,
      "learning_rate": 7.285846969532907e-06,
      "loss": 0.9478,
      "step": 2789
    },
    {
      "epoch": 0.600129060012906,
      "grad_norm": 0.0,
      "learning_rate": 7.279141785924515e-06,
      "loss": 0.844,
      "step": 2790
    },
    {
      "epoch": 0.600344160034416,
      "grad_norm": 0.0,
      "learning_rate": 7.2724379231407875e-06,
      "loss": 0.8343,
      "step": 2791
    },
    {
      "epoch": 0.600559260055926,
      "grad_norm": 0.0,
      "learning_rate": 7.265735384436064e-06,
      "loss": 0.9276,
      "step": 2792
    },
    {
      "epoch": 0.600774360077436,
      "grad_norm": 0.0,
      "learning_rate": 7.2590341730640565e-06,
      "loss": 0.8287,
      "step": 2793
    },
    {
      "epoch": 0.600989460098946,
      "grad_norm": 0.0,
      "learning_rate": 7.2523342922778364e-06,
      "loss": 0.8585,
      "step": 2794
    },
    {
      "epoch": 0.601204560120456,
      "grad_norm": 0.0,
      "learning_rate": 7.24563574532981e-06,
      "loss": 0.8326,
      "step": 2795
    },
    {
      "epoch": 0.601419660141966,
      "grad_norm": 0.0,
      "learning_rate": 7.2389385354717575e-06,
      "loss": 0.8467,
      "step": 2796
    },
    {
      "epoch": 0.6016347601634761,
      "grad_norm": 0.0,
      "learning_rate": 7.2322426659547925e-06,
      "loss": 0.8483,
      "step": 2797
    },
    {
      "epoch": 0.601849860184986,
      "grad_norm": 0.0,
      "learning_rate": 7.225548140029388e-06,
      "loss": 0.9004,
      "step": 2798
    },
    {
      "epoch": 0.602064960206496,
      "grad_norm": 0.0,
      "learning_rate": 7.218854960945366e-06,
      "loss": 0.8664,
      "step": 2799
    },
    {
      "epoch": 0.6022800602280061,
      "grad_norm": 0.0,
      "learning_rate": 7.212163131951885e-06,
      "loss": 0.8856,
      "step": 2800
    },
    {
      "epoch": 0.602495160249516,
      "grad_norm": 0.0,
      "learning_rate": 7.205472656297457e-06,
      "loss": 0.7894,
      "step": 2801
    },
    {
      "epoch": 0.602710260271026,
      "grad_norm": 0.0,
      "learning_rate": 7.198783537229937e-06,
      "loss": 0.8437,
      "step": 2802
    },
    {
      "epoch": 0.6029253602925361,
      "grad_norm": 0.0,
      "learning_rate": 7.192095777996515e-06,
      "loss": 0.8406,
      "step": 2803
    },
    {
      "epoch": 0.603140460314046,
      "grad_norm": 0.0,
      "learning_rate": 7.185409381843727e-06,
      "loss": 0.8543,
      "step": 2804
    },
    {
      "epoch": 0.603355560335556,
      "grad_norm": 0.0,
      "learning_rate": 7.1787243520174385e-06,
      "loss": 0.9414,
      "step": 2805
    },
    {
      "epoch": 0.6035706603570661,
      "grad_norm": 0.0,
      "learning_rate": 7.172040691762864e-06,
      "loss": 0.8442,
      "step": 2806
    },
    {
      "epoch": 0.603785760378576,
      "grad_norm": 0.0,
      "learning_rate": 7.1653584043245495e-06,
      "loss": 0.8459,
      "step": 2807
    },
    {
      "epoch": 0.604000860400086,
      "grad_norm": 0.0,
      "learning_rate": 7.158677492946364e-06,
      "loss": 0.7783,
      "step": 2808
    },
    {
      "epoch": 0.6042159604215961,
      "grad_norm": 0.0,
      "learning_rate": 7.151997960871526e-06,
      "loss": 0.8384,
      "step": 2809
    },
    {
      "epoch": 0.604431060443106,
      "grad_norm": 0.0,
      "learning_rate": 7.1453198113425655e-06,
      "loss": 0.8171,
      "step": 2810
    },
    {
      "epoch": 0.6046461604646161,
      "grad_norm": 0.0,
      "learning_rate": 7.138643047601359e-06,
      "loss": 0.8294,
      "step": 2811
    },
    {
      "epoch": 0.6048612604861261,
      "grad_norm": 0.0,
      "learning_rate": 7.131967672889101e-06,
      "loss": 0.842,
      "step": 2812
    },
    {
      "epoch": 0.605076360507636,
      "grad_norm": 0.0,
      "learning_rate": 7.125293690446307e-06,
      "loss": 0.8082,
      "step": 2813
    },
    {
      "epoch": 0.6052914605291461,
      "grad_norm": 0.0,
      "learning_rate": 7.118621103512826e-06,
      "loss": 0.847,
      "step": 2814
    },
    {
      "epoch": 0.6055065605506561,
      "grad_norm": 0.0,
      "learning_rate": 7.111949915327833e-06,
      "loss": 0.9022,
      "step": 2815
    },
    {
      "epoch": 0.605721660572166,
      "grad_norm": 0.0,
      "learning_rate": 7.105280129129806e-06,
      "loss": 0.7914,
      "step": 2816
    },
    {
      "epoch": 0.6059367605936761,
      "grad_norm": 0.0,
      "learning_rate": 7.098611748156563e-06,
      "loss": 0.8389,
      "step": 2817
    },
    {
      "epoch": 0.6061518606151861,
      "grad_norm": 0.0,
      "learning_rate": 7.091944775645224e-06,
      "loss": 0.8014,
      "step": 2818
    },
    {
      "epoch": 0.606366960636696,
      "grad_norm": 0.0,
      "learning_rate": 7.085279214832233e-06,
      "loss": 0.8874,
      "step": 2819
    },
    {
      "epoch": 0.6065820606582061,
      "grad_norm": 0.0,
      "learning_rate": 7.07861506895335e-06,
      "loss": 0.8365,
      "step": 2820
    },
    {
      "epoch": 0.6067971606797161,
      "grad_norm": 0.0,
      "learning_rate": 7.071952341243642e-06,
      "loss": 0.7908,
      "step": 2821
    },
    {
      "epoch": 0.607012260701226,
      "grad_norm": 0.0,
      "learning_rate": 7.065291034937496e-06,
      "loss": 0.8674,
      "step": 2822
    },
    {
      "epoch": 0.6072273607227361,
      "grad_norm": 0.0,
      "learning_rate": 7.058631153268598e-06,
      "loss": 0.9119,
      "step": 2823
    },
    {
      "epoch": 0.6074424607442461,
      "grad_norm": 0.0,
      "learning_rate": 7.051972699469953e-06,
      "loss": 0.8452,
      "step": 2824
    },
    {
      "epoch": 0.607657560765756,
      "grad_norm": 0.0,
      "learning_rate": 7.045315676773866e-06,
      "loss": 0.8107,
      "step": 2825
    },
    {
      "epoch": 0.6078726607872661,
      "grad_norm": 0.0,
      "learning_rate": 7.038660088411951e-06,
      "loss": 0.9055,
      "step": 2826
    },
    {
      "epoch": 0.6080877608087761,
      "grad_norm": 0.0,
      "learning_rate": 7.0320059376151225e-06,
      "loss": 0.7948,
      "step": 2827
    },
    {
      "epoch": 0.6083028608302861,
      "grad_norm": 0.0,
      "learning_rate": 7.025353227613604e-06,
      "loss": 0.8479,
      "step": 2828
    },
    {
      "epoch": 0.6085179608517961,
      "grad_norm": 0.0,
      "learning_rate": 7.018701961636907e-06,
      "loss": 0.7937,
      "step": 2829
    },
    {
      "epoch": 0.6087330608733061,
      "grad_norm": 0.0,
      "learning_rate": 7.012052142913857e-06,
      "loss": 0.833,
      "step": 2830
    },
    {
      "epoch": 0.6089481608948161,
      "grad_norm": 0.0,
      "learning_rate": 7.00540377467256e-06,
      "loss": 0.8496,
      "step": 2831
    },
    {
      "epoch": 0.6091632609163261,
      "grad_norm": 0.0,
      "learning_rate": 6.998756860140437e-06,
      "loss": 0.8469,
      "step": 2832
    },
    {
      "epoch": 0.609378360937836,
      "grad_norm": 0.0,
      "learning_rate": 6.9921114025441916e-06,
      "loss": 0.8344,
      "step": 2833
    },
    {
      "epoch": 0.6095934609593461,
      "grad_norm": 0.0,
      "learning_rate": 6.985467405109816e-06,
      "loss": 0.8494,
      "step": 2834
    },
    {
      "epoch": 0.6098085609808561,
      "grad_norm": 0.0,
      "learning_rate": 6.978824871062609e-06,
      "loss": 0.8512,
      "step": 2835
    },
    {
      "epoch": 0.610023661002366,
      "grad_norm": 0.0,
      "learning_rate": 6.97218380362714e-06,
      "loss": 0.8474,
      "step": 2836
    },
    {
      "epoch": 0.6102387610238761,
      "grad_norm": 0.0,
      "learning_rate": 6.965544206027283e-06,
      "loss": 0.845,
      "step": 2837
    },
    {
      "epoch": 0.6104538610453861,
      "grad_norm": 0.0,
      "learning_rate": 6.958906081486192e-06,
      "loss": 0.8425,
      "step": 2838
    },
    {
      "epoch": 0.610668961066896,
      "grad_norm": 0.0,
      "learning_rate": 6.952269433226303e-06,
      "loss": 0.8225,
      "step": 2839
    },
    {
      "epoch": 0.6108840610884061,
      "grad_norm": 0.0,
      "learning_rate": 6.945634264469338e-06,
      "loss": 0.809,
      "step": 2840
    },
    {
      "epoch": 0.6110991611099161,
      "grad_norm": 0.0,
      "learning_rate": 6.939000578436301e-06,
      "loss": 0.7901,
      "step": 2841
    },
    {
      "epoch": 0.6113142611314261,
      "grad_norm": 0.0,
      "learning_rate": 6.932368378347473e-06,
      "loss": 0.8921,
      "step": 2842
    },
    {
      "epoch": 0.6115293611529361,
      "grad_norm": 0.0,
      "learning_rate": 6.925737667422427e-06,
      "loss": 0.9269,
      "step": 2843
    },
    {
      "epoch": 0.6117444611744461,
      "grad_norm": 0.0,
      "learning_rate": 6.91910844887999e-06,
      "loss": 0.8344,
      "step": 2844
    },
    {
      "epoch": 0.6119595611959561,
      "grad_norm": 0.0,
      "learning_rate": 6.912480725938284e-06,
      "loss": 0.8528,
      "step": 2845
    },
    {
      "epoch": 0.6121746612174661,
      "grad_norm": 0.0,
      "learning_rate": 6.9058545018146986e-06,
      "loss": 0.8395,
      "step": 2846
    },
    {
      "epoch": 0.6123897612389761,
      "grad_norm": 0.0,
      "learning_rate": 6.899229779725893e-06,
      "loss": 0.7104,
      "step": 2847
    },
    {
      "epoch": 0.6126048612604861,
      "grad_norm": 0.0,
      "learning_rate": 6.892606562887802e-06,
      "loss": 0.8638,
      "step": 2848
    },
    {
      "epoch": 0.6128199612819961,
      "grad_norm": 0.0,
      "learning_rate": 6.885984854515624e-06,
      "loss": 0.9188,
      "step": 2849
    },
    {
      "epoch": 0.6130350613035062,
      "grad_norm": 0.0,
      "learning_rate": 6.879364657823828e-06,
      "loss": 0.8767,
      "step": 2850
    },
    {
      "epoch": 0.6132501613250161,
      "grad_norm": 0.0,
      "learning_rate": 6.872745976026158e-06,
      "loss": 0.8117,
      "step": 2851
    },
    {
      "epoch": 0.6134652613465261,
      "grad_norm": 0.0,
      "learning_rate": 6.866128812335603e-06,
      "loss": 0.892,
      "step": 2852
    },
    {
      "epoch": 0.6136803613680362,
      "grad_norm": 0.0,
      "learning_rate": 6.859513169964436e-06,
      "loss": 0.8277,
      "step": 2853
    },
    {
      "epoch": 0.6138954613895461,
      "grad_norm": 0.0,
      "learning_rate": 6.8528990521241755e-06,
      "loss": 0.834,
      "step": 2854
    },
    {
      "epoch": 0.6141105614110561,
      "grad_norm": 0.0,
      "learning_rate": 6.846286462025606e-06,
      "loss": 0.8298,
      "step": 2855
    },
    {
      "epoch": 0.6143256614325662,
      "grad_norm": 0.0,
      "learning_rate": 6.8396754028787775e-06,
      "loss": 0.8591,
      "step": 2856
    },
    {
      "epoch": 0.6145407614540761,
      "grad_norm": 0.0,
      "learning_rate": 6.833065877892979e-06,
      "loss": 0.8147,
      "step": 2857
    },
    {
      "epoch": 0.6147558614755861,
      "grad_norm": 0.0,
      "learning_rate": 6.826457890276773e-06,
      "loss": 0.8383,
      "step": 2858
    },
    {
      "epoch": 0.6149709614970962,
      "grad_norm": 0.0,
      "learning_rate": 6.819851443237971e-06,
      "loss": 0.8019,
      "step": 2859
    },
    {
      "epoch": 0.6151860615186061,
      "grad_norm": 0.0,
      "learning_rate": 6.813246539983629e-06,
      "loss": 0.7938,
      "step": 2860
    },
    {
      "epoch": 0.6154011615401161,
      "grad_norm": 0.0,
      "learning_rate": 6.8066431837200574e-06,
      "loss": 0.802,
      "step": 2861
    },
    {
      "epoch": 0.6156162615616262,
      "grad_norm": 0.0,
      "learning_rate": 6.80004137765282e-06,
      "loss": 0.882,
      "step": 2862
    },
    {
      "epoch": 0.6158313615831361,
      "grad_norm": 0.0,
      "learning_rate": 6.793441124986722e-06,
      "loss": 0.8473,
      "step": 2863
    },
    {
      "epoch": 0.6160464616046462,
      "grad_norm": 0.0,
      "learning_rate": 6.7868424289258216e-06,
      "loss": 0.8213,
      "step": 2864
    },
    {
      "epoch": 0.6162615616261562,
      "grad_norm": 0.0,
      "learning_rate": 6.780245292673411e-06,
      "loss": 0.8285,
      "step": 2865
    },
    {
      "epoch": 0.6164766616476661,
      "grad_norm": 0.0,
      "learning_rate": 6.7736497194320405e-06,
      "loss": 0.8407,
      "step": 2866
    },
    {
      "epoch": 0.6166917616691762,
      "grad_norm": 0.0,
      "learning_rate": 6.767055712403481e-06,
      "loss": 0.8613,
      "step": 2867
    },
    {
      "epoch": 0.6169068616906862,
      "grad_norm": 0.0,
      "learning_rate": 6.7604632747887625e-06,
      "loss": 0.9103,
      "step": 2868
    },
    {
      "epoch": 0.6171219617121961,
      "grad_norm": 0.0,
      "learning_rate": 6.753872409788145e-06,
      "loss": 0.8777,
      "step": 2869
    },
    {
      "epoch": 0.6173370617337062,
      "grad_norm": 0.0,
      "learning_rate": 6.74728312060112e-06,
      "loss": 0.8276,
      "step": 2870
    },
    {
      "epoch": 0.6175521617552162,
      "grad_norm": 0.0,
      "learning_rate": 6.740695410426421e-06,
      "loss": 0.8387,
      "step": 2871
    },
    {
      "epoch": 0.6177672617767261,
      "grad_norm": 0.0,
      "learning_rate": 6.73410928246202e-06,
      "loss": 0.8302,
      "step": 2872
    },
    {
      "epoch": 0.6179823617982362,
      "grad_norm": 0.0,
      "learning_rate": 6.727524739905104e-06,
      "loss": 0.7588,
      "step": 2873
    },
    {
      "epoch": 0.6181974618197462,
      "grad_norm": 0.0,
      "learning_rate": 6.7209417859521105e-06,
      "loss": 0.9008,
      "step": 2874
    },
    {
      "epoch": 0.6184125618412561,
      "grad_norm": 0.0,
      "learning_rate": 6.714360423798688e-06,
      "loss": 0.8848,
      "step": 2875
    },
    {
      "epoch": 0.6186276618627662,
      "grad_norm": 0.0,
      "learning_rate": 6.707780656639722e-06,
      "loss": 0.7437,
      "step": 2876
    },
    {
      "epoch": 0.6188427618842762,
      "grad_norm": 0.0,
      "learning_rate": 6.701202487669325e-06,
      "loss": 0.7813,
      "step": 2877
    },
    {
      "epoch": 0.6190578619057862,
      "grad_norm": 0.0,
      "learning_rate": 6.694625920080826e-06,
      "loss": 0.7958,
      "step": 2878
    },
    {
      "epoch": 0.6192729619272962,
      "grad_norm": 0.0,
      "learning_rate": 6.688050957066787e-06,
      "loss": 0.8511,
      "step": 2879
    },
    {
      "epoch": 0.6194880619488062,
      "grad_norm": 0.0,
      "learning_rate": 6.681477601818977e-06,
      "loss": 0.8998,
      "step": 2880
    },
    {
      "epoch": 0.6197031619703162,
      "grad_norm": 0.0,
      "learning_rate": 6.674905857528398e-06,
      "loss": 0.8493,
      "step": 2881
    },
    {
      "epoch": 0.6199182619918262,
      "grad_norm": 0.0,
      "learning_rate": 6.6683357273852686e-06,
      "loss": 0.8259,
      "step": 2882
    },
    {
      "epoch": 0.6201333620133362,
      "grad_norm": 0.0,
      "learning_rate": 6.661767214579015e-06,
      "loss": 0.8647,
      "step": 2883
    },
    {
      "epoch": 0.6203484620348462,
      "grad_norm": 0.0,
      "learning_rate": 6.655200322298284e-06,
      "loss": 0.7311,
      "step": 2884
    },
    {
      "epoch": 0.6205635620563562,
      "grad_norm": 0.0,
      "learning_rate": 6.648635053730931e-06,
      "loss": 0.8312,
      "step": 2885
    },
    {
      "epoch": 0.6207786620778663,
      "grad_norm": 0.0,
      "learning_rate": 6.642071412064031e-06,
      "loss": 0.8005,
      "step": 2886
    },
    {
      "epoch": 0.6209937620993762,
      "grad_norm": 0.0,
      "learning_rate": 6.635509400483868e-06,
      "loss": 0.8456,
      "step": 2887
    },
    {
      "epoch": 0.6212088621208862,
      "grad_norm": 0.0,
      "learning_rate": 6.628949022175928e-06,
      "loss": 0.8588,
      "step": 2888
    },
    {
      "epoch": 0.6214239621423963,
      "grad_norm": 0.0,
      "learning_rate": 6.622390280324908e-06,
      "loss": 0.8227,
      "step": 2889
    },
    {
      "epoch": 0.6216390621639062,
      "grad_norm": 0.0,
      "learning_rate": 6.615833178114715e-06,
      "loss": 0.8096,
      "step": 2890
    },
    {
      "epoch": 0.6218541621854162,
      "grad_norm": 0.0,
      "learning_rate": 6.60927771872845e-06,
      "loss": 0.857,
      "step": 2891
    },
    {
      "epoch": 0.6220692622069263,
      "grad_norm": 0.0,
      "learning_rate": 6.60272390534843e-06,
      "loss": 0.8402,
      "step": 2892
    },
    {
      "epoch": 0.6222843622284362,
      "grad_norm": 0.0,
      "learning_rate": 6.596171741156154e-06,
      "loss": 0.8407,
      "step": 2893
    },
    {
      "epoch": 0.6224994622499462,
      "grad_norm": 0.0,
      "learning_rate": 6.58962122933234e-06,
      "loss": 0.8005,
      "step": 2894
    },
    {
      "epoch": 0.6227145622714563,
      "grad_norm": 0.0,
      "learning_rate": 6.583072373056898e-06,
      "loss": 0.8362,
      "step": 2895
    },
    {
      "epoch": 0.6229296622929662,
      "grad_norm": 0.0,
      "learning_rate": 6.576525175508922e-06,
      "loss": 0.8709,
      "step": 2896
    },
    {
      "epoch": 0.6231447623144762,
      "grad_norm": 0.0,
      "learning_rate": 6.5699796398667215e-06,
      "loss": 0.9053,
      "step": 2897
    },
    {
      "epoch": 0.6233598623359863,
      "grad_norm": 0.0,
      "learning_rate": 6.56343576930778e-06,
      "loss": 0.8996,
      "step": 2898
    },
    {
      "epoch": 0.6235749623574962,
      "grad_norm": 0.0,
      "learning_rate": 6.556893567008782e-06,
      "loss": 0.8484,
      "step": 2899
    },
    {
      "epoch": 0.6237900623790062,
      "grad_norm": 0.0,
      "learning_rate": 6.5503530361456095e-06,
      "loss": 0.8743,
      "step": 2900
    },
    {
      "epoch": 0.6240051624005163,
      "grad_norm": 0.0,
      "learning_rate": 6.543814179893312e-06,
      "loss": 0.8366,
      "step": 2901
    },
    {
      "epoch": 0.6242202624220262,
      "grad_norm": 0.0,
      "learning_rate": 6.5372770014261465e-06,
      "loss": 0.898,
      "step": 2902
    },
    {
      "epoch": 0.6244353624435363,
      "grad_norm": 0.0,
      "learning_rate": 6.530741503917551e-06,
      "loss": 0.8121,
      "step": 2903
    },
    {
      "epoch": 0.6246504624650463,
      "grad_norm": 0.0,
      "learning_rate": 6.524207690540138e-06,
      "loss": 0.8526,
      "step": 2904
    },
    {
      "epoch": 0.6248655624865562,
      "grad_norm": 0.0,
      "learning_rate": 6.517675564465713e-06,
      "loss": 0.8216,
      "step": 2905
    },
    {
      "epoch": 0.6250806625080663,
      "grad_norm": 0.0,
      "learning_rate": 6.5111451288652565e-06,
      "loss": 0.8103,
      "step": 2906
    },
    {
      "epoch": 0.6252957625295763,
      "grad_norm": 0.0,
      "learning_rate": 6.504616386908926e-06,
      "loss": 0.898,
      "step": 2907
    },
    {
      "epoch": 0.6255108625510862,
      "grad_norm": 0.0,
      "learning_rate": 6.498089341766072e-06,
      "loss": 0.8459,
      "step": 2908
    },
    {
      "epoch": 0.6257259625725963,
      "grad_norm": 0.0,
      "learning_rate": 6.491563996605198e-06,
      "loss": 0.884,
      "step": 2909
    },
    {
      "epoch": 0.6259410625941063,
      "grad_norm": 0.0,
      "learning_rate": 6.485040354594004e-06,
      "loss": 0.8081,
      "step": 2910
    },
    {
      "epoch": 0.6261561626156162,
      "grad_norm": 0.0,
      "learning_rate": 6.478518418899347e-06,
      "loss": 0.8372,
      "step": 2911
    },
    {
      "epoch": 0.6263712626371263,
      "grad_norm": 0.0,
      "learning_rate": 6.471998192687266e-06,
      "loss": 0.8997,
      "step": 2912
    },
    {
      "epoch": 0.6265863626586363,
      "grad_norm": 0.0,
      "learning_rate": 6.465479679122968e-06,
      "loss": 0.8349,
      "step": 2913
    },
    {
      "epoch": 0.6268014626801462,
      "grad_norm": 0.0,
      "learning_rate": 6.4589628813708215e-06,
      "loss": 0.8369,
      "step": 2914
    },
    {
      "epoch": 0.6270165627016563,
      "grad_norm": 0.0,
      "learning_rate": 6.4524478025943704e-06,
      "loss": 0.7884,
      "step": 2915
    },
    {
      "epoch": 0.6272316627231663,
      "grad_norm": 0.0,
      "learning_rate": 6.445934445956328e-06,
      "loss": 0.8948,
      "step": 2916
    },
    {
      "epoch": 0.6274467627446763,
      "grad_norm": 0.0,
      "learning_rate": 6.439422814618553e-06,
      "loss": 0.898,
      "step": 2917
    },
    {
      "epoch": 0.6276618627661863,
      "grad_norm": 0.0,
      "learning_rate": 6.4329129117420904e-06,
      "loss": 0.8429,
      "step": 2918
    },
    {
      "epoch": 0.6278769627876962,
      "grad_norm": 0.0,
      "learning_rate": 6.426404740487127e-06,
      "loss": 0.8854,
      "step": 2919
    },
    {
      "epoch": 0.6280920628092063,
      "grad_norm": 0.0,
      "learning_rate": 6.4198983040130154e-06,
      "loss": 0.8242,
      "step": 2920
    },
    {
      "epoch": 0.6283071628307163,
      "grad_norm": 0.0,
      "learning_rate": 6.413393605478275e-06,
      "loss": 0.8895,
      "step": 2921
    },
    {
      "epoch": 0.6285222628522262,
      "grad_norm": 0.0,
      "learning_rate": 6.406890648040563e-06,
      "loss": 0.8238,
      "step": 2922
    },
    {
      "epoch": 0.6287373628737363,
      "grad_norm": 0.0,
      "learning_rate": 6.400389434856713e-06,
      "loss": 0.8739,
      "step": 2923
    },
    {
      "epoch": 0.6289524628952463,
      "grad_norm": 0.0,
      "learning_rate": 6.39388996908269e-06,
      "loss": 0.8357,
      "step": 2924
    },
    {
      "epoch": 0.6291675629167562,
      "grad_norm": 0.0,
      "learning_rate": 6.3873922538736296e-06,
      "loss": 0.8206,
      "step": 2925
    },
    {
      "epoch": 0.6293826629382663,
      "grad_norm": 0.0,
      "learning_rate": 6.380896292383807e-06,
      "loss": 0.845,
      "step": 2926
    },
    {
      "epoch": 0.6295977629597763,
      "grad_norm": 0.0,
      "learning_rate": 6.374402087766647e-06,
      "loss": 0.7972,
      "step": 2927
    },
    {
      "epoch": 0.6298128629812862,
      "grad_norm": 0.0,
      "learning_rate": 6.367909643174723e-06,
      "loss": 0.7858,
      "step": 2928
    },
    {
      "epoch": 0.6300279630027963,
      "grad_norm": 0.0,
      "learning_rate": 6.36141896175976e-06,
      "loss": 0.8009,
      "step": 2929
    },
    {
      "epoch": 0.6302430630243063,
      "grad_norm": 0.0,
      "learning_rate": 6.354930046672612e-06,
      "loss": 0.9034,
      "step": 2930
    },
    {
      "epoch": 0.6304581630458163,
      "grad_norm": 0.0,
      "learning_rate": 6.348442901063294e-06,
      "loss": 0.91,
      "step": 2931
    },
    {
      "epoch": 0.6306732630673263,
      "grad_norm": 0.0,
      "learning_rate": 6.341957528080944e-06,
      "loss": 0.8292,
      "step": 2932
    },
    {
      "epoch": 0.6308883630888363,
      "grad_norm": 0.0,
      "learning_rate": 6.335473930873854e-06,
      "loss": 0.8421,
      "step": 2933
    },
    {
      "epoch": 0.6311034631103463,
      "grad_norm": 0.0,
      "learning_rate": 6.32899211258945e-06,
      "loss": 0.8132,
      "step": 2934
    },
    {
      "epoch": 0.6313185631318563,
      "grad_norm": 0.0,
      "learning_rate": 6.3225120763742854e-06,
      "loss": 0.8489,
      "step": 2935
    },
    {
      "epoch": 0.6315336631533663,
      "grad_norm": 0.0,
      "learning_rate": 6.316033825374064e-06,
      "loss": 0.8585,
      "step": 2936
    },
    {
      "epoch": 0.6317487631748763,
      "grad_norm": 0.0,
      "learning_rate": 6.309557362733605e-06,
      "loss": 0.8627,
      "step": 2937
    },
    {
      "epoch": 0.6319638631963863,
      "grad_norm": 0.0,
      "learning_rate": 6.303082691596876e-06,
      "loss": 0.8421,
      "step": 2938
    },
    {
      "epoch": 0.6321789632178964,
      "grad_norm": 0.0,
      "learning_rate": 6.296609815106971e-06,
      "loss": 0.8541,
      "step": 2939
    },
    {
      "epoch": 0.6323940632394063,
      "grad_norm": 0.0,
      "learning_rate": 6.290138736406103e-06,
      "loss": 0.8735,
      "step": 2940
    },
    {
      "epoch": 0.6326091632609163,
      "grad_norm": 0.0,
      "learning_rate": 6.283669458635622e-06,
      "loss": 0.8449,
      "step": 2941
    },
    {
      "epoch": 0.6328242632824264,
      "grad_norm": 0.0,
      "learning_rate": 6.277201984936001e-06,
      "loss": 0.897,
      "step": 2942
    },
    {
      "epoch": 0.6330393633039363,
      "grad_norm": 0.0,
      "learning_rate": 6.270736318446835e-06,
      "loss": 0.8033,
      "step": 2943
    },
    {
      "epoch": 0.6332544633254463,
      "grad_norm": 0.0,
      "learning_rate": 6.264272462306851e-06,
      "loss": 0.8205,
      "step": 2944
    },
    {
      "epoch": 0.6334695633469564,
      "grad_norm": 0.0,
      "learning_rate": 6.25781041965388e-06,
      "loss": 0.7984,
      "step": 2945
    },
    {
      "epoch": 0.6336846633684663,
      "grad_norm": 0.0,
      "learning_rate": 6.25135019362489e-06,
      "loss": 0.8499,
      "step": 2946
    },
    {
      "epoch": 0.6338997633899763,
      "grad_norm": 0.0,
      "learning_rate": 6.244891787355958e-06,
      "loss": 0.8285,
      "step": 2947
    },
    {
      "epoch": 0.6341148634114864,
      "grad_norm": 0.0,
      "learning_rate": 6.238435203982278e-06,
      "loss": 0.8323,
      "step": 2948
    },
    {
      "epoch": 0.6343299634329963,
      "grad_norm": 0.0,
      "learning_rate": 6.231980446638165e-06,
      "loss": 0.8112,
      "step": 2949
    },
    {
      "epoch": 0.6345450634545063,
      "grad_norm": 0.0,
      "learning_rate": 6.225527518457035e-06,
      "loss": 0.8998,
      "step": 2950
    },
    {
      "epoch": 0.6347601634760164,
      "grad_norm": 0.0,
      "learning_rate": 6.2190764225714284e-06,
      "loss": 0.8231,
      "step": 2951
    },
    {
      "epoch": 0.6349752634975263,
      "grad_norm": 0.0,
      "learning_rate": 6.212627162112996e-06,
      "loss": 0.8248,
      "step": 2952
    },
    {
      "epoch": 0.6351903635190363,
      "grad_norm": 0.0,
      "learning_rate": 6.206179740212486e-06,
      "loss": 0.8245,
      "step": 2953
    },
    {
      "epoch": 0.6354054635405464,
      "grad_norm": 0.0,
      "learning_rate": 6.199734159999768e-06,
      "loss": 0.8598,
      "step": 2954
    },
    {
      "epoch": 0.6356205635620563,
      "grad_norm": 0.0,
      "learning_rate": 6.193290424603809e-06,
      "loss": 0.9056,
      "step": 2955
    },
    {
      "epoch": 0.6358356635835664,
      "grad_norm": 0.0,
      "learning_rate": 6.186848537152677e-06,
      "loss": 0.7958,
      "step": 2956
    },
    {
      "epoch": 0.6360507636050764,
      "grad_norm": 0.0,
      "learning_rate": 6.180408500773558e-06,
      "loss": 0.8703,
      "step": 2957
    },
    {
      "epoch": 0.6362658636265863,
      "grad_norm": 0.0,
      "learning_rate": 6.173970318592722e-06,
      "loss": 0.8696,
      "step": 2958
    },
    {
      "epoch": 0.6364809636480964,
      "grad_norm": 0.0,
      "learning_rate": 6.1675339937355505e-06,
      "loss": 0.8716,
      "step": 2959
    },
    {
      "epoch": 0.6366960636696064,
      "grad_norm": 0.0,
      "learning_rate": 6.1610995293265215e-06,
      "loss": 0.8182,
      "step": 2960
    },
    {
      "epoch": 0.6369111636911163,
      "grad_norm": 0.0,
      "learning_rate": 6.154666928489201e-06,
      "loss": 0.8568,
      "step": 2961
    },
    {
      "epoch": 0.6371262637126264,
      "grad_norm": 0.0,
      "learning_rate": 6.148236194346268e-06,
      "loss": 0.8905,
      "step": 2962
    },
    {
      "epoch": 0.6373413637341364,
      "grad_norm": 0.0,
      "learning_rate": 6.141807330019476e-06,
      "loss": 0.8988,
      "step": 2963
    },
    {
      "epoch": 0.6375564637556463,
      "grad_norm": 0.0,
      "learning_rate": 6.135380338629679e-06,
      "loss": 0.8335,
      "step": 2964
    },
    {
      "epoch": 0.6377715637771564,
      "grad_norm": 0.0,
      "learning_rate": 6.128955223296831e-06,
      "loss": 0.8511,
      "step": 2965
    },
    {
      "epoch": 0.6379866637986664,
      "grad_norm": 0.0,
      "learning_rate": 6.122531987139955e-06,
      "loss": 0.8815,
      "step": 2966
    },
    {
      "epoch": 0.6382017638201763,
      "grad_norm": 0.0,
      "learning_rate": 6.1161106332771844e-06,
      "loss": 0.8605,
      "step": 2967
    },
    {
      "epoch": 0.6384168638416864,
      "grad_norm": 0.0,
      "learning_rate": 6.109691164825718e-06,
      "loss": 0.8377,
      "step": 2968
    },
    {
      "epoch": 0.6386319638631964,
      "grad_norm": 0.0,
      "learning_rate": 6.103273584901857e-06,
      "loss": 0.8987,
      "step": 2969
    },
    {
      "epoch": 0.6388470638847064,
      "grad_norm": 0.0,
      "learning_rate": 6.096857896620975e-06,
      "loss": 0.8334,
      "step": 2970
    },
    {
      "epoch": 0.6390621639062164,
      "grad_norm": 0.0,
      "learning_rate": 6.090444103097526e-06,
      "loss": 0.7861,
      "step": 2971
    },
    {
      "epoch": 0.6392772639277264,
      "grad_norm": 0.0,
      "learning_rate": 6.084032207445052e-06,
      "loss": 0.8797,
      "step": 2972
    },
    {
      "epoch": 0.6394923639492364,
      "grad_norm": 0.0,
      "learning_rate": 6.077622212776175e-06,
      "loss": 0.8464,
      "step": 2973
    },
    {
      "epoch": 0.6397074639707464,
      "grad_norm": 0.0,
      "learning_rate": 6.071214122202581e-06,
      "loss": 0.8037,
      "step": 2974
    },
    {
      "epoch": 0.6399225639922564,
      "grad_norm": 0.0,
      "learning_rate": 6.064807938835046e-06,
      "loss": 0.8502,
      "step": 2975
    },
    {
      "epoch": 0.6401376640137664,
      "grad_norm": 0.0,
      "learning_rate": 6.05840366578341e-06,
      "loss": 0.8899,
      "step": 2976
    },
    {
      "epoch": 0.6403527640352764,
      "grad_norm": 0.0,
      "learning_rate": 6.052001306156593e-06,
      "loss": 0.8797,
      "step": 2977
    },
    {
      "epoch": 0.6405678640567865,
      "grad_norm": 0.0,
      "learning_rate": 6.045600863062584e-06,
      "loss": 0.8293,
      "step": 2978
    },
    {
      "epoch": 0.6407829640782964,
      "grad_norm": 0.0,
      "learning_rate": 6.039202339608432e-06,
      "loss": 0.8175,
      "step": 2979
    },
    {
      "epoch": 0.6409980640998064,
      "grad_norm": 0.0,
      "learning_rate": 6.0328057389002736e-06,
      "loss": 0.8794,
      "step": 2980
    },
    {
      "epoch": 0.6412131641213165,
      "grad_norm": 0.0,
      "learning_rate": 6.0264110640432935e-06,
      "loss": 0.8181,
      "step": 2981
    },
    {
      "epoch": 0.6414282641428264,
      "grad_norm": 0.0,
      "learning_rate": 6.0200183181417515e-06,
      "loss": 0.8109,
      "step": 2982
    },
    {
      "epoch": 0.6416433641643364,
      "grad_norm": 0.0,
      "learning_rate": 6.0136275042989714e-06,
      "loss": 0.8998,
      "step": 2983
    },
    {
      "epoch": 0.6418584641858465,
      "grad_norm": 0.0,
      "learning_rate": 6.007238625617333e-06,
      "loss": 0.8342,
      "step": 2984
    },
    {
      "epoch": 0.6420735642073564,
      "grad_norm": 0.0,
      "learning_rate": 6.000851685198283e-06,
      "loss": 0.8358,
      "step": 2985
    },
    {
      "epoch": 0.6422886642288664,
      "grad_norm": 0.0,
      "learning_rate": 5.994466686142317e-06,
      "loss": 0.8734,
      "step": 2986
    },
    {
      "epoch": 0.6425037642503765,
      "grad_norm": 0.0,
      "learning_rate": 5.988083631549001e-06,
      "loss": 0.7831,
      "step": 2987
    },
    {
      "epoch": 0.6427188642718864,
      "grad_norm": 0.0,
      "learning_rate": 5.9817025245169545e-06,
      "loss": 0.8622,
      "step": 2988
    },
    {
      "epoch": 0.6429339642933964,
      "grad_norm": 0.0,
      "learning_rate": 5.975323368143841e-06,
      "loss": 0.8557,
      "step": 2989
    },
    {
      "epoch": 0.6431490643149065,
      "grad_norm": 0.0,
      "learning_rate": 5.968946165526389e-06,
      "loss": 0.8482,
      "step": 2990
    },
    {
      "epoch": 0.6433641643364164,
      "grad_norm": 0.0,
      "learning_rate": 5.962570919760374e-06,
      "loss": 0.7818,
      "step": 2991
    },
    {
      "epoch": 0.6435792643579265,
      "grad_norm": 0.0,
      "learning_rate": 5.956197633940622e-06,
      "loss": 0.8405,
      "step": 2992
    },
    {
      "epoch": 0.6437943643794365,
      "grad_norm": 0.0,
      "learning_rate": 5.949826311161007e-06,
      "loss": 0.8622,
      "step": 2993
    },
    {
      "epoch": 0.6440094644009464,
      "grad_norm": 0.0,
      "learning_rate": 5.943456954514445e-06,
      "loss": 0.8347,
      "step": 2994
    },
    {
      "epoch": 0.6442245644224565,
      "grad_norm": 0.0,
      "learning_rate": 5.937089567092909e-06,
      "loss": 0.8929,
      "step": 2995
    },
    {
      "epoch": 0.6444396644439665,
      "grad_norm": 0.0,
      "learning_rate": 5.930724151987411e-06,
      "loss": 0.8756,
      "step": 2996
    },
    {
      "epoch": 0.6446547644654764,
      "grad_norm": 0.0,
      "learning_rate": 5.9243607122879965e-06,
      "loss": 0.8236,
      "step": 2997
    },
    {
      "epoch": 0.6448698644869865,
      "grad_norm": 0.0,
      "learning_rate": 5.9179992510837706e-06,
      "loss": 0.8755,
      "step": 2998
    },
    {
      "epoch": 0.6450849645084965,
      "grad_norm": 0.0,
      "learning_rate": 5.9116397714628584e-06,
      "loss": 0.8934,
      "step": 2999
    },
    {
      "epoch": 0.6453000645300064,
      "grad_norm": 0.0,
      "learning_rate": 5.905282276512434e-06,
      "loss": 0.8067,
      "step": 3000
    },
    {
      "epoch": 0.6455151645515165,
      "grad_norm": 0.0,
      "learning_rate": 5.898926769318711e-06,
      "loss": 0.8725,
      "step": 3001
    },
    {
      "epoch": 0.6457302645730264,
      "grad_norm": 0.0,
      "learning_rate": 5.892573252966926e-06,
      "loss": 0.8711,
      "step": 3002
    },
    {
      "epoch": 0.6459453645945364,
      "grad_norm": 0.0,
      "learning_rate": 5.88622173054136e-06,
      "loss": 0.8776,
      "step": 3003
    },
    {
      "epoch": 0.6461604646160465,
      "grad_norm": 0.0,
      "learning_rate": 5.879872205125325e-06,
      "loss": 0.833,
      "step": 3004
    },
    {
      "epoch": 0.6463755646375564,
      "grad_norm": 0.0,
      "learning_rate": 5.873524679801157e-06,
      "loss": 0.8289,
      "step": 3005
    },
    {
      "epoch": 0.6465906646590664,
      "grad_norm": 0.0,
      "learning_rate": 5.867179157650225e-06,
      "loss": 0.8093,
      "step": 3006
    },
    {
      "epoch": 0.6468057646805765,
      "grad_norm": 0.0,
      "learning_rate": 5.860835641752927e-06,
      "loss": 0.8745,
      "step": 3007
    },
    {
      "epoch": 0.6470208647020864,
      "grad_norm": 0.0,
      "learning_rate": 5.854494135188682e-06,
      "loss": 0.832,
      "step": 3008
    },
    {
      "epoch": 0.6472359647235965,
      "grad_norm": 0.0,
      "learning_rate": 5.848154641035943e-06,
      "loss": 0.7589,
      "step": 3009
    },
    {
      "epoch": 0.6474510647451065,
      "grad_norm": 0.0,
      "learning_rate": 5.841817162372176e-06,
      "loss": 0.8568,
      "step": 3010
    },
    {
      "epoch": 0.6476661647666164,
      "grad_norm": 0.0,
      "learning_rate": 5.835481702273879e-06,
      "loss": 0.7875,
      "step": 3011
    },
    {
      "epoch": 0.6478812647881265,
      "grad_norm": 0.0,
      "learning_rate": 5.829148263816552e-06,
      "loss": 0.8462,
      "step": 3012
    },
    {
      "epoch": 0.6480963648096365,
      "grad_norm": 0.0,
      "learning_rate": 5.822816850074733e-06,
      "loss": 0.7954,
      "step": 3013
    },
    {
      "epoch": 0.6483114648311464,
      "grad_norm": 0.0,
      "learning_rate": 5.816487464121974e-06,
      "loss": 0.8949,
      "step": 3014
    },
    {
      "epoch": 0.6485265648526565,
      "grad_norm": 0.0,
      "learning_rate": 5.810160109030829e-06,
      "loss": 0.8385,
      "step": 3015
    },
    {
      "epoch": 0.6487416648741665,
      "grad_norm": 0.0,
      "learning_rate": 5.80383478787288e-06,
      "loss": 0.8188,
      "step": 3016
    },
    {
      "epoch": 0.6489567648956764,
      "grad_norm": 0.0,
      "learning_rate": 5.79751150371872e-06,
      "loss": 0.8876,
      "step": 3017
    },
    {
      "epoch": 0.6491718649171865,
      "grad_norm": 0.0,
      "learning_rate": 5.791190259637945e-06,
      "loss": 0.8429,
      "step": 3018
    },
    {
      "epoch": 0.6493869649386965,
      "grad_norm": 0.0,
      "learning_rate": 5.784871058699171e-06,
      "loss": 0.8433,
      "step": 3019
    },
    {
      "epoch": 0.6496020649602064,
      "grad_norm": 0.0,
      "learning_rate": 5.77855390397001e-06,
      "loss": 0.899,
      "step": 3020
    },
    {
      "epoch": 0.6498171649817165,
      "grad_norm": 0.0,
      "learning_rate": 5.7722387985170915e-06,
      "loss": 0.9477,
      "step": 3021
    },
    {
      "epoch": 0.6500322650032265,
      "grad_norm": 0.0,
      "learning_rate": 5.765925745406051e-06,
      "loss": 0.8159,
      "step": 3022
    },
    {
      "epoch": 0.6502473650247365,
      "grad_norm": 0.0,
      "learning_rate": 5.759614747701519e-06,
      "loss": 0.8309,
      "step": 3023
    },
    {
      "epoch": 0.6504624650462465,
      "grad_norm": 0.0,
      "learning_rate": 5.753305808467126e-06,
      "loss": 0.8394,
      "step": 3024
    },
    {
      "epoch": 0.6506775650677565,
      "grad_norm": 0.0,
      "learning_rate": 5.7469989307655225e-06,
      "loss": 0.9191,
      "step": 3025
    },
    {
      "epoch": 0.6508926650892665,
      "grad_norm": 0.0,
      "learning_rate": 5.740694117658334e-06,
      "loss": 0.8686,
      "step": 3026
    },
    {
      "epoch": 0.6511077651107765,
      "grad_norm": 0.0,
      "learning_rate": 5.734391372206203e-06,
      "loss": 0.9043,
      "step": 3027
    },
    {
      "epoch": 0.6513228651322865,
      "grad_norm": 0.0,
      "learning_rate": 5.728090697468751e-06,
      "loss": 0.8462,
      "step": 3028
    },
    {
      "epoch": 0.6515379651537965,
      "grad_norm": 0.0,
      "learning_rate": 5.7217920965046115e-06,
      "loss": 0.8432,
      "step": 3029
    },
    {
      "epoch": 0.6517530651753065,
      "grad_norm": 0.0,
      "learning_rate": 5.715495572371405e-06,
      "loss": 0.8181,
      "step": 3030
    },
    {
      "epoch": 0.6519681651968166,
      "grad_norm": 0.0,
      "learning_rate": 5.709201128125733e-06,
      "loss": 0.863,
      "step": 3031
    },
    {
      "epoch": 0.6521832652183265,
      "grad_norm": 0.0,
      "learning_rate": 5.7029087668232075e-06,
      "loss": 0.8946,
      "step": 3032
    },
    {
      "epoch": 0.6523983652398365,
      "grad_norm": 0.0,
      "learning_rate": 5.696618491518409e-06,
      "loss": 0.873,
      "step": 3033
    },
    {
      "epoch": 0.6526134652613466,
      "grad_norm": 0.0,
      "learning_rate": 5.690330305264917e-06,
      "loss": 0.8392,
      "step": 3034
    },
    {
      "epoch": 0.6528285652828565,
      "grad_norm": 0.0,
      "learning_rate": 5.6840442111153e-06,
      "loss": 0.8225,
      "step": 3035
    },
    {
      "epoch": 0.6530436653043665,
      "grad_norm": 0.0,
      "learning_rate": 5.6777602121211e-06,
      "loss": 0.7992,
      "step": 3036
    },
    {
      "epoch": 0.6532587653258766,
      "grad_norm": 0.0,
      "learning_rate": 5.6714783113328505e-06,
      "loss": 0.8237,
      "step": 3037
    },
    {
      "epoch": 0.6534738653473865,
      "grad_norm": 0.0,
      "learning_rate": 5.665198511800063e-06,
      "loss": 0.7702,
      "step": 3038
    },
    {
      "epoch": 0.6536889653688965,
      "grad_norm": 0.0,
      "learning_rate": 5.658920816571225e-06,
      "loss": 0.7917,
      "step": 3039
    },
    {
      "epoch": 0.6539040653904066,
      "grad_norm": 0.0,
      "learning_rate": 5.652645228693814e-06,
      "loss": 0.817,
      "step": 3040
    },
    {
      "epoch": 0.6541191654119165,
      "grad_norm": 0.0,
      "learning_rate": 5.64637175121427e-06,
      "loss": 0.8304,
      "step": 3041
    },
    {
      "epoch": 0.6543342654334265,
      "grad_norm": 0.0,
      "learning_rate": 5.640100387178023e-06,
      "loss": 0.8664,
      "step": 3042
    },
    {
      "epoch": 0.6545493654549366,
      "grad_norm": 0.0,
      "learning_rate": 5.633831139629462e-06,
      "loss": 0.8887,
      "step": 3043
    },
    {
      "epoch": 0.6547644654764465,
      "grad_norm": 0.0,
      "learning_rate": 5.627564011611961e-06,
      "loss": 0.8324,
      "step": 3044
    },
    {
      "epoch": 0.6549795654979566,
      "grad_norm": 0.0,
      "learning_rate": 5.621299006167864e-06,
      "loss": 0.8533,
      "step": 3045
    },
    {
      "epoch": 0.6551946655194666,
      "grad_norm": 0.0,
      "learning_rate": 5.615036126338473e-06,
      "loss": 0.8105,
      "step": 3046
    },
    {
      "epoch": 0.6554097655409765,
      "grad_norm": 0.0,
      "learning_rate": 5.6087753751640725e-06,
      "loss": 0.8294,
      "step": 3047
    },
    {
      "epoch": 0.6556248655624866,
      "grad_norm": 0.0,
      "learning_rate": 5.602516755683908e-06,
      "loss": 0.8281,
      "step": 3048
    },
    {
      "epoch": 0.6558399655839966,
      "grad_norm": 0.0,
      "learning_rate": 5.5962602709361825e-06,
      "loss": 0.8778,
      "step": 3049
    },
    {
      "epoch": 0.6560550656055065,
      "grad_norm": 0.0,
      "learning_rate": 5.590005923958079e-06,
      "loss": 0.849,
      "step": 3050
    },
    {
      "epoch": 0.6562701656270166,
      "grad_norm": 0.0,
      "learning_rate": 5.583753717785727e-06,
      "loss": 0.9144,
      "step": 3051
    },
    {
      "epoch": 0.6564852656485266,
      "grad_norm": 0.0,
      "learning_rate": 5.577503655454228e-06,
      "loss": 0.863,
      "step": 3052
    },
    {
      "epoch": 0.6567003656700365,
      "grad_norm": 0.0,
      "learning_rate": 5.5712557399976345e-06,
      "loss": 0.86,
      "step": 3053
    },
    {
      "epoch": 0.6569154656915466,
      "grad_norm": 0.0,
      "learning_rate": 5.565009974448958e-06,
      "loss": 0.7987,
      "step": 3054
    },
    {
      "epoch": 0.6571305657130566,
      "grad_norm": 0.0,
      "learning_rate": 5.558766361840177e-06,
      "loss": 0.8447,
      "step": 3055
    },
    {
      "epoch": 0.6573456657345665,
      "grad_norm": 0.0,
      "learning_rate": 5.552524905202208e-06,
      "loss": 0.8768,
      "step": 3056
    },
    {
      "epoch": 0.6575607657560766,
      "grad_norm": 0.0,
      "learning_rate": 5.546285607564932e-06,
      "loss": 0.9048,
      "step": 3057
    },
    {
      "epoch": 0.6577758657775866,
      "grad_norm": 0.0,
      "learning_rate": 5.540048471957184e-06,
      "loss": 0.846,
      "step": 3058
    },
    {
      "epoch": 0.6579909657990965,
      "grad_norm": 0.0,
      "learning_rate": 5.533813501406739e-06,
      "loss": 0.8433,
      "step": 3059
    },
    {
      "epoch": 0.6582060658206066,
      "grad_norm": 0.0,
      "learning_rate": 5.527580698940329e-06,
      "loss": 0.7767,
      "step": 3060
    },
    {
      "epoch": 0.6584211658421166,
      "grad_norm": 0.0,
      "learning_rate": 5.521350067583635e-06,
      "loss": 0.8352,
      "step": 3061
    },
    {
      "epoch": 0.6586362658636266,
      "grad_norm": 0.0,
      "learning_rate": 5.515121610361275e-06,
      "loss": 0.8568,
      "step": 3062
    },
    {
      "epoch": 0.6588513658851366,
      "grad_norm": 0.0,
      "learning_rate": 5.508895330296822e-06,
      "loss": 0.862,
      "step": 3063
    },
    {
      "epoch": 0.6590664659066466,
      "grad_norm": 0.0,
      "learning_rate": 5.502671230412782e-06,
      "loss": 0.8574,
      "step": 3064
    },
    {
      "epoch": 0.6592815659281566,
      "grad_norm": 0.0,
      "learning_rate": 5.496449313730608e-06,
      "loss": 0.7818,
      "step": 3065
    },
    {
      "epoch": 0.6594966659496666,
      "grad_norm": 0.0,
      "learning_rate": 5.4902295832707005e-06,
      "loss": 0.8683,
      "step": 3066
    },
    {
      "epoch": 0.6597117659711766,
      "grad_norm": 0.0,
      "learning_rate": 5.484012042052385e-06,
      "loss": 0.826,
      "step": 3067
    },
    {
      "epoch": 0.6599268659926866,
      "grad_norm": 0.0,
      "learning_rate": 5.477796693093933e-06,
      "loss": 0.8082,
      "step": 3068
    },
    {
      "epoch": 0.6601419660141966,
      "grad_norm": 0.0,
      "learning_rate": 5.471583539412541e-06,
      "loss": 0.8154,
      "step": 3069
    },
    {
      "epoch": 0.6603570660357067,
      "grad_norm": 0.0,
      "learning_rate": 5.4653725840243555e-06,
      "loss": 0.8494,
      "step": 3070
    },
    {
      "epoch": 0.6605721660572166,
      "grad_norm": 0.0,
      "learning_rate": 5.459163829944451e-06,
      "loss": 0.8576,
      "step": 3071
    },
    {
      "epoch": 0.6607872660787266,
      "grad_norm": 0.0,
      "learning_rate": 5.452957280186824e-06,
      "loss": 0.8735,
      "step": 3072
    },
    {
      "epoch": 0.6610023661002367,
      "grad_norm": 0.0,
      "learning_rate": 5.44675293776441e-06,
      "loss": 0.8773,
      "step": 3073
    },
    {
      "epoch": 0.6612174661217466,
      "grad_norm": 0.0,
      "learning_rate": 5.440550805689075e-06,
      "loss": 0.8933,
      "step": 3074
    },
    {
      "epoch": 0.6614325661432566,
      "grad_norm": 0.0,
      "learning_rate": 5.434350886971602e-06,
      "loss": 0.8691,
      "step": 3075
    },
    {
      "epoch": 0.6616476661647667,
      "grad_norm": 0.0,
      "learning_rate": 5.42815318462171e-06,
      "loss": 0.806,
      "step": 3076
    },
    {
      "epoch": 0.6618627661862766,
      "grad_norm": 0.0,
      "learning_rate": 5.421957701648028e-06,
      "loss": 0.8656,
      "step": 3077
    },
    {
      "epoch": 0.6620778662077866,
      "grad_norm": 0.0,
      "learning_rate": 5.415764441058127e-06,
      "loss": 0.8227,
      "step": 3078
    },
    {
      "epoch": 0.6622929662292967,
      "grad_norm": 0.0,
      "learning_rate": 5.409573405858487e-06,
      "loss": 0.8533,
      "step": 3079
    },
    {
      "epoch": 0.6625080662508066,
      "grad_norm": 0.0,
      "learning_rate": 5.403384599054504e-06,
      "loss": 0.8145,
      "step": 3080
    },
    {
      "epoch": 0.6627231662723166,
      "grad_norm": 0.0,
      "learning_rate": 5.397198023650508e-06,
      "loss": 0.8138,
      "step": 3081
    },
    {
      "epoch": 0.6629382662938267,
      "grad_norm": 0.0,
      "learning_rate": 5.391013682649726e-06,
      "loss": 0.8171,
      "step": 3082
    },
    {
      "epoch": 0.6631533663153366,
      "grad_norm": 0.0,
      "learning_rate": 5.384831579054313e-06,
      "loss": 0.899,
      "step": 3083
    },
    {
      "epoch": 0.6633684663368467,
      "grad_norm": 0.0,
      "learning_rate": 5.378651715865338e-06,
      "loss": 0.8329,
      "step": 3084
    },
    {
      "epoch": 0.6635835663583567,
      "grad_norm": 0.0,
      "learning_rate": 5.372474096082774e-06,
      "loss": 0.8482,
      "step": 3085
    },
    {
      "epoch": 0.6637986663798666,
      "grad_norm": 0.0,
      "learning_rate": 5.3662987227055115e-06,
      "loss": 0.8869,
      "step": 3086
    },
    {
      "epoch": 0.6640137664013767,
      "grad_norm": 0.0,
      "learning_rate": 5.360125598731358e-06,
      "loss": 0.9081,
      "step": 3087
    },
    {
      "epoch": 0.6642288664228866,
      "grad_norm": 0.0,
      "learning_rate": 5.3539547271570066e-06,
      "loss": 0.8684,
      "step": 3088
    },
    {
      "epoch": 0.6644439664443966,
      "grad_norm": 0.0,
      "learning_rate": 5.347786110978084e-06,
      "loss": 0.8483,
      "step": 3089
    },
    {
      "epoch": 0.6646590664659067,
      "grad_norm": 0.0,
      "learning_rate": 5.341619753189096e-06,
      "loss": 0.8351,
      "step": 3090
    },
    {
      "epoch": 0.6648741664874166,
      "grad_norm": 0.0,
      "learning_rate": 5.3354556567834725e-06,
      "loss": 0.8166,
      "step": 3091
    },
    {
      "epoch": 0.6650892665089266,
      "grad_norm": 0.0,
      "learning_rate": 5.329293824753539e-06,
      "loss": 0.8045,
      "step": 3092
    },
    {
      "epoch": 0.6653043665304367,
      "grad_norm": 0.0,
      "learning_rate": 5.323134260090516e-06,
      "loss": 0.8755,
      "step": 3093
    },
    {
      "epoch": 0.6655194665519466,
      "grad_norm": 0.0,
      "learning_rate": 5.316976965784532e-06,
      "loss": 0.7934,
      "step": 3094
    },
    {
      "epoch": 0.6657345665734566,
      "grad_norm": 0.0,
      "learning_rate": 5.310821944824606e-06,
      "loss": 0.8449,
      "step": 3095
    },
    {
      "epoch": 0.6659496665949667,
      "grad_norm": 0.0,
      "learning_rate": 5.304669200198661e-06,
      "loss": 0.8289,
      "step": 3096
    },
    {
      "epoch": 0.6661647666164766,
      "grad_norm": 0.0,
      "learning_rate": 5.2985187348935085e-06,
      "loss": 0.8233,
      "step": 3097
    },
    {
      "epoch": 0.6663798666379867,
      "grad_norm": 0.0,
      "learning_rate": 5.292370551894852e-06,
      "loss": 0.7926,
      "step": 3098
    },
    {
      "epoch": 0.6665949666594967,
      "grad_norm": 0.0,
      "learning_rate": 5.286224654187298e-06,
      "loss": 0.871,
      "step": 3099
    },
    {
      "epoch": 0.6668100666810066,
      "grad_norm": 0.0,
      "learning_rate": 5.2800810447543285e-06,
      "loss": 0.849,
      "step": 3100
    },
    {
      "epoch": 0.6670251667025167,
      "grad_norm": 0.0,
      "learning_rate": 5.273939726578328e-06,
      "loss": 0.7902,
      "step": 3101
    },
    {
      "epoch": 0.6672402667240267,
      "grad_norm": 0.0,
      "learning_rate": 5.267800702640565e-06,
      "loss": 0.9172,
      "step": 3102
    },
    {
      "epoch": 0.6674553667455366,
      "grad_norm": 0.0,
      "learning_rate": 5.2616639759211875e-06,
      "loss": 0.8348,
      "step": 3103
    },
    {
      "epoch": 0.6676704667670467,
      "grad_norm": 0.0,
      "learning_rate": 5.255529549399234e-06,
      "loss": 0.84,
      "step": 3104
    },
    {
      "epoch": 0.6678855667885567,
      "grad_norm": 0.0,
      "learning_rate": 5.249397426052634e-06,
      "loss": 0.8609,
      "step": 3105
    },
    {
      "epoch": 0.6681006668100666,
      "grad_norm": 0.0,
      "learning_rate": 5.24326760885818e-06,
      "loss": 0.8494,
      "step": 3106
    },
    {
      "epoch": 0.6683157668315767,
      "grad_norm": 0.0,
      "learning_rate": 5.2371401007915646e-06,
      "loss": 0.8432,
      "step": 3107
    },
    {
      "epoch": 0.6685308668530867,
      "grad_norm": 0.0,
      "learning_rate": 5.231014904827343e-06,
      "loss": 0.8151,
      "step": 3108
    },
    {
      "epoch": 0.6687459668745966,
      "grad_norm": 0.0,
      "learning_rate": 5.224892023938963e-06,
      "loss": 0.8092,
      "step": 3109
    },
    {
      "epoch": 0.6689610668961067,
      "grad_norm": 0.0,
      "learning_rate": 5.218771461098733e-06,
      "loss": 0.83,
      "step": 3110
    },
    {
      "epoch": 0.6691761669176167,
      "grad_norm": 0.0,
      "learning_rate": 5.212653219277854e-06,
      "loss": 0.8679,
      "step": 3111
    },
    {
      "epoch": 0.6693912669391267,
      "grad_norm": 0.0,
      "learning_rate": 5.2065373014463835e-06,
      "loss": 0.8994,
      "step": 3112
    },
    {
      "epoch": 0.6696063669606367,
      "grad_norm": 0.0,
      "learning_rate": 5.200423710573257e-06,
      "loss": 0.8915,
      "step": 3113
    },
    {
      "epoch": 0.6698214669821467,
      "grad_norm": 0.0,
      "learning_rate": 5.1943124496262844e-06,
      "loss": 0.823,
      "step": 3114
    },
    {
      "epoch": 0.6700365670036567,
      "grad_norm": 0.0,
      "learning_rate": 5.188203521572144e-06,
      "loss": 0.8799,
      "step": 3115
    },
    {
      "epoch": 0.6702516670251667,
      "grad_norm": 0.0,
      "learning_rate": 5.182096929376372e-06,
      "loss": 0.8864,
      "step": 3116
    },
    {
      "epoch": 0.6704667670466767,
      "grad_norm": 0.0,
      "learning_rate": 5.175992676003383e-06,
      "loss": 0.8487,
      "step": 3117
    },
    {
      "epoch": 0.6706818670681867,
      "grad_norm": 0.0,
      "learning_rate": 5.169890764416454e-06,
      "loss": 0.8256,
      "step": 3118
    },
    {
      "epoch": 0.6708969670896967,
      "grad_norm": 0.0,
      "learning_rate": 5.163791197577714e-06,
      "loss": 0.8291,
      "step": 3119
    },
    {
      "epoch": 0.6711120671112067,
      "grad_norm": 0.0,
      "learning_rate": 5.157693978448171e-06,
      "loss": 0.8803,
      "step": 3120
    },
    {
      "epoch": 0.6713271671327167,
      "grad_norm": 0.0,
      "learning_rate": 5.151599109987675e-06,
      "loss": 0.8441,
      "step": 3121
    },
    {
      "epoch": 0.6715422671542267,
      "grad_norm": 0.0,
      "learning_rate": 5.145506595154947e-06,
      "loss": 0.8248,
      "step": 3122
    },
    {
      "epoch": 0.6717573671757368,
      "grad_norm": 0.0,
      "learning_rate": 5.139416436907571e-06,
      "loss": 0.7916,
      "step": 3123
    },
    {
      "epoch": 0.6719724671972467,
      "grad_norm": 0.0,
      "learning_rate": 5.133328638201969e-06,
      "loss": 0.8352,
      "step": 3124
    },
    {
      "epoch": 0.6721875672187567,
      "grad_norm": 0.0,
      "learning_rate": 5.127243201993428e-06,
      "loss": 0.7451,
      "step": 3125
    },
    {
      "epoch": 0.6724026672402668,
      "grad_norm": 0.0,
      "learning_rate": 5.1211601312360915e-06,
      "loss": 0.8733,
      "step": 3126
    },
    {
      "epoch": 0.6726177672617767,
      "grad_norm": 0.0,
      "learning_rate": 5.115079428882945e-06,
      "loss": 0.8224,
      "step": 3127
    },
    {
      "epoch": 0.6728328672832867,
      "grad_norm": 0.0,
      "learning_rate": 5.109001097885837e-06,
      "loss": 0.8127,
      "step": 3128
    },
    {
      "epoch": 0.6730479673047968,
      "grad_norm": 0.0,
      "learning_rate": 5.10292514119545e-06,
      "loss": 0.8286,
      "step": 3129
    },
    {
      "epoch": 0.6732630673263067,
      "grad_norm": 0.0,
      "learning_rate": 5.0968515617613244e-06,
      "loss": 0.8655,
      "step": 3130
    },
    {
      "epoch": 0.6734781673478167,
      "grad_norm": 0.0,
      "learning_rate": 5.090780362531849e-06,
      "loss": 0.8389,
      "step": 3131
    },
    {
      "epoch": 0.6736932673693268,
      "grad_norm": 0.0,
      "learning_rate": 5.0847115464542425e-06,
      "loss": 0.8908,
      "step": 3132
    },
    {
      "epoch": 0.6739083673908367,
      "grad_norm": 0.0,
      "learning_rate": 5.0786451164745855e-06,
      "loss": 0.8385,
      "step": 3133
    },
    {
      "epoch": 0.6741234674123467,
      "grad_norm": 0.0,
      "learning_rate": 5.0725810755377825e-06,
      "loss": 0.8454,
      "step": 3134
    },
    {
      "epoch": 0.6743385674338568,
      "grad_norm": 0.0,
      "learning_rate": 5.066519426587591e-06,
      "loss": 0.8656,
      "step": 3135
    },
    {
      "epoch": 0.6745536674553667,
      "grad_norm": 0.0,
      "learning_rate": 5.060460172566605e-06,
      "loss": 0.8441,
      "step": 3136
    },
    {
      "epoch": 0.6747687674768768,
      "grad_norm": 0.0,
      "learning_rate": 5.054403316416247e-06,
      "loss": 0.8368,
      "step": 3137
    },
    {
      "epoch": 0.6749838674983868,
      "grad_norm": 0.0,
      "learning_rate": 5.048348861076791e-06,
      "loss": 0.8121,
      "step": 3138
    },
    {
      "epoch": 0.6751989675198967,
      "grad_norm": 0.0,
      "learning_rate": 5.042296809487332e-06,
      "loss": 0.812,
      "step": 3139
    },
    {
      "epoch": 0.6754140675414068,
      "grad_norm": 0.0,
      "learning_rate": 5.036247164585799e-06,
      "loss": 0.864,
      "step": 3140
    },
    {
      "epoch": 0.6756291675629168,
      "grad_norm": 0.0,
      "learning_rate": 5.0301999293089635e-06,
      "loss": 0.8055,
      "step": 3141
    },
    {
      "epoch": 0.6758442675844267,
      "grad_norm": 0.0,
      "learning_rate": 5.024155106592415e-06,
      "loss": 0.8351,
      "step": 3142
    },
    {
      "epoch": 0.6760593676059368,
      "grad_norm": 0.0,
      "learning_rate": 5.018112699370584e-06,
      "loss": 0.82,
      "step": 3143
    },
    {
      "epoch": 0.6762744676274468,
      "grad_norm": 0.0,
      "learning_rate": 5.012072710576714e-06,
      "loss": 0.901,
      "step": 3144
    },
    {
      "epoch": 0.6764895676489567,
      "grad_norm": 0.0,
      "learning_rate": 5.006035143142884e-06,
      "loss": 0.863,
      "step": 3145
    },
    {
      "epoch": 0.6767046676704668,
      "grad_norm": 0.0,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.8716,
      "step": 3146
    },
    {
      "epoch": 0.6769197676919768,
      "grad_norm": 0.0,
      "learning_rate": 4.993967284077785e-06,
      "loss": 0.8804,
      "step": 3147
    },
    {
      "epoch": 0.6771348677134867,
      "grad_norm": 0.0,
      "learning_rate": 4.987936998304782e-06,
      "loss": 0.838,
      "step": 3148
    },
    {
      "epoch": 0.6773499677349968,
      "grad_norm": 0.0,
      "learning_rate": 4.981909145608365e-06,
      "loss": 0.8272,
      "step": 3149
    },
    {
      "epoch": 0.6775650677565068,
      "grad_norm": 0.0,
      "learning_rate": 4.9758837289147095e-06,
      "loss": 0.8702,
      "step": 3150
    },
    {
      "epoch": 0.6777801677780168,
      "grad_norm": 0.0,
      "learning_rate": 4.9698607511488306e-06,
      "loss": 0.8694,
      "step": 3151
    },
    {
      "epoch": 0.6779952677995268,
      "grad_norm": 0.0,
      "learning_rate": 4.963840215234536e-06,
      "loss": 0.8452,
      "step": 3152
    },
    {
      "epoch": 0.6782103678210368,
      "grad_norm": 0.0,
      "learning_rate": 4.95782212409447e-06,
      "loss": 0.8548,
      "step": 3153
    },
    {
      "epoch": 0.6784254678425468,
      "grad_norm": 0.0,
      "learning_rate": 4.9518064806500745e-06,
      "loss": 0.7569,
      "step": 3154
    },
    {
      "epoch": 0.6786405678640568,
      "grad_norm": 0.0,
      "learning_rate": 4.945793287821604e-06,
      "loss": 0.8506,
      "step": 3155
    },
    {
      "epoch": 0.6788556678855668,
      "grad_norm": 0.0,
      "learning_rate": 4.9397825485281355e-06,
      "loss": 0.8379,
      "step": 3156
    },
    {
      "epoch": 0.6790707679070768,
      "grad_norm": 0.0,
      "learning_rate": 4.933774265687541e-06,
      "loss": 0.7818,
      "step": 3157
    },
    {
      "epoch": 0.6792858679285868,
      "grad_norm": 0.0,
      "learning_rate": 4.9277684422165085e-06,
      "loss": 0.8231,
      "step": 3158
    },
    {
      "epoch": 0.6795009679500968,
      "grad_norm": 0.0,
      "learning_rate": 4.921765081030534e-06,
      "loss": 0.8598,
      "step": 3159
    },
    {
      "epoch": 0.6797160679716068,
      "grad_norm": 0.0,
      "learning_rate": 4.915764185043906e-06,
      "loss": 0.8359,
      "step": 3160
    },
    {
      "epoch": 0.6799311679931168,
      "grad_norm": 0.0,
      "learning_rate": 4.909765757169729e-06,
      "loss": 0.8303,
      "step": 3161
    },
    {
      "epoch": 0.6801462680146269,
      "grad_norm": 0.0,
      "learning_rate": 4.9037698003199075e-06,
      "loss": 0.8538,
      "step": 3162
    },
    {
      "epoch": 0.6803613680361368,
      "grad_norm": 0.0,
      "learning_rate": 4.897776317405135e-06,
      "loss": 0.9153,
      "step": 3163
    },
    {
      "epoch": 0.6805764680576468,
      "grad_norm": 0.0,
      "learning_rate": 4.891785311334924e-06,
      "loss": 0.8292,
      "step": 3164
    },
    {
      "epoch": 0.6807915680791569,
      "grad_norm": 0.0,
      "learning_rate": 4.88579678501756e-06,
      "loss": 0.8085,
      "step": 3165
    },
    {
      "epoch": 0.6810066681006668,
      "grad_norm": 0.0,
      "learning_rate": 4.879810741360147e-06,
      "loss": 0.8724,
      "step": 3166
    },
    {
      "epoch": 0.6812217681221768,
      "grad_norm": 0.0,
      "learning_rate": 4.873827183268577e-06,
      "loss": 0.9115,
      "step": 3167
    },
    {
      "epoch": 0.6814368681436869,
      "grad_norm": 0.0,
      "learning_rate": 4.867846113647527e-06,
      "loss": 0.9,
      "step": 3168
    },
    {
      "epoch": 0.6816519681651968,
      "grad_norm": 0.0,
      "learning_rate": 4.861867535400472e-06,
      "loss": 0.8817,
      "step": 3169
    },
    {
      "epoch": 0.6818670681867068,
      "grad_norm": 0.0,
      "learning_rate": 4.855891451429682e-06,
      "loss": 0.8378,
      "step": 3170
    },
    {
      "epoch": 0.6820821682082168,
      "grad_norm": 0.0,
      "learning_rate": 4.849917864636207e-06,
      "loss": 0.8694,
      "step": 3171
    },
    {
      "epoch": 0.6822972682297268,
      "grad_norm": 0.0,
      "learning_rate": 4.843946777919896e-06,
      "loss": 0.8698,
      "step": 3172
    },
    {
      "epoch": 0.6825123682512368,
      "grad_norm": 0.0,
      "learning_rate": 4.83797819417937e-06,
      "loss": 0.8393,
      "step": 3173
    },
    {
      "epoch": 0.6827274682727468,
      "grad_norm": 0.0,
      "learning_rate": 4.832012116312047e-06,
      "loss": 0.8354,
      "step": 3174
    },
    {
      "epoch": 0.6829425682942568,
      "grad_norm": 0.0,
      "learning_rate": 4.826048547214129e-06,
      "loss": 0.8483,
      "step": 3175
    },
    {
      "epoch": 0.6831576683157669,
      "grad_norm": 0.0,
      "learning_rate": 4.820087489780588e-06,
      "loss": 0.8375,
      "step": 3176
    },
    {
      "epoch": 0.6833727683372768,
      "grad_norm": 0.0,
      "learning_rate": 4.814128946905192e-06,
      "loss": 0.8145,
      "step": 3177
    },
    {
      "epoch": 0.6835878683587868,
      "grad_norm": 0.0,
      "learning_rate": 4.808172921480472e-06,
      "loss": 0.7654,
      "step": 3178
    },
    {
      "epoch": 0.6838029683802969,
      "grad_norm": 0.0,
      "learning_rate": 4.802219416397749e-06,
      "loss": 0.8767,
      "step": 3179
    },
    {
      "epoch": 0.6840180684018068,
      "grad_norm": 0.0,
      "learning_rate": 4.796268434547125e-06,
      "loss": 0.8729,
      "step": 3180
    },
    {
      "epoch": 0.6842331684233168,
      "grad_norm": 0.0,
      "learning_rate": 4.790319978817456e-06,
      "loss": 0.8084,
      "step": 3181
    },
    {
      "epoch": 0.6844482684448269,
      "grad_norm": 0.0,
      "learning_rate": 4.784374052096396e-06,
      "loss": 0.8274,
      "step": 3182
    },
    {
      "epoch": 0.6846633684663368,
      "grad_norm": 0.0,
      "learning_rate": 4.778430657270356e-06,
      "loss": 0.8427,
      "step": 3183
    },
    {
      "epoch": 0.6848784684878468,
      "grad_norm": 0.0,
      "learning_rate": 4.772489797224518e-06,
      "loss": 0.853,
      "step": 3184
    },
    {
      "epoch": 0.6850935685093569,
      "grad_norm": 0.0,
      "learning_rate": 4.7665514748428446e-06,
      "loss": 0.9108,
      "step": 3185
    },
    {
      "epoch": 0.6853086685308668,
      "grad_norm": 0.0,
      "learning_rate": 4.760615693008053e-06,
      "loss": 0.8282,
      "step": 3186
    },
    {
      "epoch": 0.6855237685523768,
      "grad_norm": 0.0,
      "learning_rate": 4.754682454601637e-06,
      "loss": 0.8039,
      "step": 3187
    },
    {
      "epoch": 0.6857388685738869,
      "grad_norm": 0.0,
      "learning_rate": 4.748751762503856e-06,
      "loss": 0.8242,
      "step": 3188
    },
    {
      "epoch": 0.6859539685953968,
      "grad_norm": 0.0,
      "learning_rate": 4.742823619593723e-06,
      "loss": 0.8481,
      "step": 3189
    },
    {
      "epoch": 0.6861690686169069,
      "grad_norm": 0.0,
      "learning_rate": 4.7368980287490265e-06,
      "loss": 0.8304,
      "step": 3190
    },
    {
      "epoch": 0.6863841686384169,
      "grad_norm": 0.0,
      "learning_rate": 4.730974992846304e-06,
      "loss": 0.828,
      "step": 3191
    },
    {
      "epoch": 0.6865992686599268,
      "grad_norm": 0.0,
      "learning_rate": 4.72505451476086e-06,
      "loss": 0.858,
      "step": 3192
    },
    {
      "epoch": 0.6868143686814369,
      "grad_norm": 0.0,
      "learning_rate": 4.71913659736676e-06,
      "loss": 0.8828,
      "step": 3193
    },
    {
      "epoch": 0.6870294687029469,
      "grad_norm": 0.0,
      "learning_rate": 4.7132212435368166e-06,
      "loss": 0.8881,
      "step": 3194
    },
    {
      "epoch": 0.6872445687244568,
      "grad_norm": 0.0,
      "learning_rate": 4.70730845614261e-06,
      "loss": 0.7677,
      "step": 3195
    },
    {
      "epoch": 0.6874596687459669,
      "grad_norm": 0.0,
      "learning_rate": 4.701398238054461e-06,
      "loss": 0.8702,
      "step": 3196
    },
    {
      "epoch": 0.6876747687674769,
      "grad_norm": 0.0,
      "learning_rate": 4.695490592141457e-06,
      "loss": 0.8214,
      "step": 3197
    },
    {
      "epoch": 0.6878898687889868,
      "grad_norm": 0.0,
      "learning_rate": 4.689585521271427e-06,
      "loss": 0.8396,
      "step": 3198
    },
    {
      "epoch": 0.6881049688104969,
      "grad_norm": 0.0,
      "learning_rate": 4.683683028310949e-06,
      "loss": 0.8332,
      "step": 3199
    },
    {
      "epoch": 0.6883200688320069,
      "grad_norm": 0.0,
      "learning_rate": 4.677783116125361e-06,
      "loss": 0.8511,
      "step": 3200
    },
    {
      "epoch": 0.6885351688535168,
      "grad_norm": 0.0,
      "learning_rate": 4.671885787578735e-06,
      "loss": 0.8002,
      "step": 3201
    },
    {
      "epoch": 0.6887502688750269,
      "grad_norm": 0.0,
      "learning_rate": 4.665991045533895e-06,
      "loss": 0.85,
      "step": 3202
    },
    {
      "epoch": 0.6889653688965369,
      "grad_norm": 0.0,
      "learning_rate": 4.660098892852416e-06,
      "loss": 0.8104,
      "step": 3203
    },
    {
      "epoch": 0.6891804689180469,
      "grad_norm": 0.0,
      "learning_rate": 4.654209332394598e-06,
      "loss": 0.8423,
      "step": 3204
    },
    {
      "epoch": 0.6893955689395569,
      "grad_norm": 0.0,
      "learning_rate": 4.648322367019501e-06,
      "loss": 0.8331,
      "step": 3205
    },
    {
      "epoch": 0.6896106689610669,
      "grad_norm": 0.0,
      "learning_rate": 4.642437999584919e-06,
      "loss": 0.8494,
      "step": 3206
    },
    {
      "epoch": 0.6898257689825769,
      "grad_norm": 0.0,
      "learning_rate": 4.636556232947377e-06,
      "loss": 0.8511,
      "step": 3207
    },
    {
      "epoch": 0.6900408690040869,
      "grad_norm": 0.0,
      "learning_rate": 4.630677069962151e-06,
      "loss": 0.8809,
      "step": 3208
    },
    {
      "epoch": 0.6902559690255969,
      "grad_norm": 0.0,
      "learning_rate": 4.62480051348324e-06,
      "loss": 0.7776,
      "step": 3209
    },
    {
      "epoch": 0.6904710690471069,
      "grad_norm": 0.0,
      "learning_rate": 4.61892656636339e-06,
      "loss": 0.8266,
      "step": 3210
    },
    {
      "epoch": 0.6906861690686169,
      "grad_norm": 0.0,
      "learning_rate": 4.613055231454066e-06,
      "loss": 0.8553,
      "step": 3211
    },
    {
      "epoch": 0.690901269090127,
      "grad_norm": 0.0,
      "learning_rate": 4.607186511605483e-06,
      "loss": 0.9176,
      "step": 3212
    },
    {
      "epoch": 0.6911163691116369,
      "grad_norm": 0.0,
      "learning_rate": 4.601320409666573e-06,
      "loss": 0.788,
      "step": 3213
    },
    {
      "epoch": 0.6913314691331469,
      "grad_norm": 0.0,
      "learning_rate": 4.595456928484995e-06,
      "loss": 0.7917,
      "step": 3214
    },
    {
      "epoch": 0.691546569154657,
      "grad_norm": 0.0,
      "learning_rate": 4.589596070907146e-06,
      "loss": 0.7714,
      "step": 3215
    },
    {
      "epoch": 0.6917616691761669,
      "grad_norm": 0.0,
      "learning_rate": 4.5837378397781476e-06,
      "loss": 0.8474,
      "step": 3216
    },
    {
      "epoch": 0.6919767691976769,
      "grad_norm": 0.0,
      "learning_rate": 4.577882237941838e-06,
      "loss": 0.8369,
      "step": 3217
    },
    {
      "epoch": 0.692191869219187,
      "grad_norm": 0.0,
      "learning_rate": 4.572029268240787e-06,
      "loss": 0.8387,
      "step": 3218
    },
    {
      "epoch": 0.6924069692406969,
      "grad_norm": 0.0,
      "learning_rate": 4.566178933516291e-06,
      "loss": 0.8343,
      "step": 3219
    },
    {
      "epoch": 0.6926220692622069,
      "grad_norm": 0.0,
      "learning_rate": 4.56033123660835e-06,
      "loss": 0.8054,
      "step": 3220
    },
    {
      "epoch": 0.692837169283717,
      "grad_norm": 0.0,
      "learning_rate": 4.554486180355703e-06,
      "loss": 0.8458,
      "step": 3221
    },
    {
      "epoch": 0.6930522693052269,
      "grad_norm": 0.0,
      "learning_rate": 4.54864376759579e-06,
      "loss": 0.8086,
      "step": 3222
    },
    {
      "epoch": 0.6932673693267369,
      "grad_norm": 0.0,
      "learning_rate": 4.542804001164781e-06,
      "loss": 0.8073,
      "step": 3223
    },
    {
      "epoch": 0.693482469348247,
      "grad_norm": 0.0,
      "learning_rate": 4.53696688389756e-06,
      "loss": 0.8341,
      "step": 3224
    },
    {
      "epoch": 0.6936975693697569,
      "grad_norm": 0.0,
      "learning_rate": 4.531132418627716e-06,
      "loss": 0.8406,
      "step": 3225
    },
    {
      "epoch": 0.693912669391267,
      "grad_norm": 0.0,
      "learning_rate": 4.525300608187554e-06,
      "loss": 0.7983,
      "step": 3226
    },
    {
      "epoch": 0.694127769412777,
      "grad_norm": 0.0,
      "learning_rate": 4.519471455408099e-06,
      "loss": 0.7661,
      "step": 3227
    },
    {
      "epoch": 0.6943428694342869,
      "grad_norm": 0.0,
      "learning_rate": 4.513644963119071e-06,
      "loss": 0.8748,
      "step": 3228
    },
    {
      "epoch": 0.694557969455797,
      "grad_norm": 0.0,
      "learning_rate": 4.507821134148914e-06,
      "loss": 0.8565,
      "step": 3229
    },
    {
      "epoch": 0.694773069477307,
      "grad_norm": 0.0,
      "learning_rate": 4.5019999713247646e-06,
      "loss": 0.8136,
      "step": 3230
    },
    {
      "epoch": 0.6949881694988169,
      "grad_norm": 0.0,
      "learning_rate": 4.4961814774724765e-06,
      "loss": 0.7948,
      "step": 3231
    },
    {
      "epoch": 0.695203269520327,
      "grad_norm": 0.0,
      "learning_rate": 4.490365655416606e-06,
      "loss": 0.8567,
      "step": 3232
    },
    {
      "epoch": 0.695418369541837,
      "grad_norm": 0.0,
      "learning_rate": 4.484552507980403e-06,
      "loss": 0.874,
      "step": 3233
    },
    {
      "epoch": 0.6956334695633469,
      "grad_norm": 0.0,
      "learning_rate": 4.478742037985833e-06,
      "loss": 0.8973,
      "step": 3234
    },
    {
      "epoch": 0.695848569584857,
      "grad_norm": 0.0,
      "learning_rate": 4.472934248253546e-06,
      "loss": 0.8206,
      "step": 3235
    },
    {
      "epoch": 0.696063669606367,
      "grad_norm": 0.0,
      "learning_rate": 4.467129141602906e-06,
      "loss": 0.8304,
      "step": 3236
    },
    {
      "epoch": 0.6962787696278769,
      "grad_norm": 0.0,
      "learning_rate": 4.461326720851971e-06,
      "loss": 0.7878,
      "step": 3237
    },
    {
      "epoch": 0.696493869649387,
      "grad_norm": 0.0,
      "learning_rate": 4.455526988817481e-06,
      "loss": 0.8381,
      "step": 3238
    },
    {
      "epoch": 0.696708969670897,
      "grad_norm": 0.0,
      "learning_rate": 4.4497299483148935e-06,
      "loss": 0.9021,
      "step": 3239
    },
    {
      "epoch": 0.6969240696924069,
      "grad_norm": 0.0,
      "learning_rate": 4.443935602158339e-06,
      "loss": 0.8291,
      "step": 3240
    },
    {
      "epoch": 0.697139169713917,
      "grad_norm": 0.0,
      "learning_rate": 4.438143953160656e-06,
      "loss": 0.8662,
      "step": 3241
    },
    {
      "epoch": 0.697354269735427,
      "grad_norm": 0.0,
      "learning_rate": 4.4323550041333644e-06,
      "loss": 0.7841,
      "step": 3242
    },
    {
      "epoch": 0.697569369756937,
      "grad_norm": 0.0,
      "learning_rate": 4.426568757886669e-06,
      "loss": 0.8215,
      "step": 3243
    },
    {
      "epoch": 0.697784469778447,
      "grad_norm": 0.0,
      "learning_rate": 4.4207852172294775e-06,
      "loss": 0.7716,
      "step": 3244
    },
    {
      "epoch": 0.697999569799957,
      "grad_norm": 0.0,
      "learning_rate": 4.41500438496937e-06,
      "loss": 0.7887,
      "step": 3245
    },
    {
      "epoch": 0.698214669821467,
      "grad_norm": 0.0,
      "learning_rate": 4.40922626391262e-06,
      "loss": 0.7826,
      "step": 3246
    },
    {
      "epoch": 0.698429769842977,
      "grad_norm": 0.0,
      "learning_rate": 4.403450856864185e-06,
      "loss": 0.8297,
      "step": 3247
    },
    {
      "epoch": 0.698644869864487,
      "grad_norm": 0.0,
      "learning_rate": 4.397678166627698e-06,
      "loss": 0.8251,
      "step": 3248
    },
    {
      "epoch": 0.698859969885997,
      "grad_norm": 0.0,
      "learning_rate": 4.39190819600548e-06,
      "loss": 0.8658,
      "step": 3249
    },
    {
      "epoch": 0.699075069907507,
      "grad_norm": 0.0,
      "learning_rate": 4.386140947798532e-06,
      "loss": 0.8179,
      "step": 3250
    },
    {
      "epoch": 0.699290169929017,
      "grad_norm": 0.0,
      "learning_rate": 4.380376424806524e-06,
      "loss": 0.8295,
      "step": 3251
    },
    {
      "epoch": 0.699505269950527,
      "grad_norm": 0.0,
      "learning_rate": 4.374614629827818e-06,
      "loss": 0.8679,
      "step": 3252
    },
    {
      "epoch": 0.699720369972037,
      "grad_norm": 0.0,
      "learning_rate": 4.368855565659434e-06,
      "loss": 0.7626,
      "step": 3253
    },
    {
      "epoch": 0.699935469993547,
      "grad_norm": 0.0,
      "learning_rate": 4.3630992350970866e-06,
      "loss": 0.799,
      "step": 3254
    },
    {
      "epoch": 0.700150570015057,
      "grad_norm": 0.0,
      "learning_rate": 4.357345640935145e-06,
      "loss": 0.7635,
      "step": 3255
    },
    {
      "epoch": 0.700365670036567,
      "grad_norm": 0.0,
      "learning_rate": 4.351594785966662e-06,
      "loss": 0.8364,
      "step": 3256
    },
    {
      "epoch": 0.700580770058077,
      "grad_norm": 0.0,
      "learning_rate": 4.345846672983355e-06,
      "loss": 0.8391,
      "step": 3257
    },
    {
      "epoch": 0.700795870079587,
      "grad_norm": 0.0,
      "learning_rate": 4.340101304775607e-06,
      "loss": 0.8575,
      "step": 3258
    },
    {
      "epoch": 0.701010970101097,
      "grad_norm": 0.0,
      "learning_rate": 4.334358684132478e-06,
      "loss": 0.7963,
      "step": 3259
    },
    {
      "epoch": 0.701226070122607,
      "grad_norm": 0.0,
      "learning_rate": 4.3286188138416915e-06,
      "loss": 0.7969,
      "step": 3260
    },
    {
      "epoch": 0.701441170144117,
      "grad_norm": 0.0,
      "learning_rate": 4.322881696689629e-06,
      "loss": 0.8534,
      "step": 3261
    },
    {
      "epoch": 0.701656270165627,
      "grad_norm": 0.0,
      "learning_rate": 4.317147335461343e-06,
      "loss": 0.7832,
      "step": 3262
    },
    {
      "epoch": 0.701871370187137,
      "grad_norm": 0.0,
      "learning_rate": 4.3114157329405494e-06,
      "loss": 0.7804,
      "step": 3263
    },
    {
      "epoch": 0.702086470208647,
      "grad_norm": 0.0,
      "learning_rate": 4.305686891909612e-06,
      "loss": 0.8536,
      "step": 3264
    },
    {
      "epoch": 0.702301570230157,
      "grad_norm": 0.0,
      "learning_rate": 4.299960815149572e-06,
      "loss": 0.8466,
      "step": 3265
    },
    {
      "epoch": 0.702516670251667,
      "grad_norm": 0.0,
      "learning_rate": 4.294237505440112e-06,
      "loss": 0.8537,
      "step": 3266
    },
    {
      "epoch": 0.702731770273177,
      "grad_norm": 0.0,
      "learning_rate": 4.288516965559584e-06,
      "loss": 0.8203,
      "step": 3267
    },
    {
      "epoch": 0.702946870294687,
      "grad_norm": 0.0,
      "learning_rate": 4.282799198284992e-06,
      "loss": 0.8788,
      "step": 3268
    },
    {
      "epoch": 0.703161970316197,
      "grad_norm": 0.0,
      "learning_rate": 4.27708420639199e-06,
      "loss": 0.8223,
      "step": 3269
    },
    {
      "epoch": 0.703377070337707,
      "grad_norm": 0.0,
      "learning_rate": 4.271371992654884e-06,
      "loss": 0.7915,
      "step": 3270
    },
    {
      "epoch": 0.7035921703592171,
      "grad_norm": 0.0,
      "learning_rate": 4.265662559846641e-06,
      "loss": 0.7878,
      "step": 3271
    },
    {
      "epoch": 0.703807270380727,
      "grad_norm": 0.0,
      "learning_rate": 4.2599559107388645e-06,
      "loss": 0.7698,
      "step": 3272
    },
    {
      "epoch": 0.704022370402237,
      "grad_norm": 0.0,
      "learning_rate": 4.2542520481018225e-06,
      "loss": 0.8609,
      "step": 3273
    },
    {
      "epoch": 0.7042374704237471,
      "grad_norm": 0.0,
      "learning_rate": 4.248550974704413e-06,
      "loss": 0.8342,
      "step": 3274
    },
    {
      "epoch": 0.704452570445257,
      "grad_norm": 0.0,
      "learning_rate": 4.242852693314193e-06,
      "loss": 0.8291,
      "step": 3275
    },
    {
      "epoch": 0.704667670466767,
      "grad_norm": 0.0,
      "learning_rate": 4.237157206697363e-06,
      "loss": 0.8225,
      "step": 3276
    },
    {
      "epoch": 0.7048827704882771,
      "grad_norm": 0.0,
      "learning_rate": 4.231464517618757e-06,
      "loss": 0.8702,
      "step": 3277
    },
    {
      "epoch": 0.705097870509787,
      "grad_norm": 0.0,
      "learning_rate": 4.225774628841864e-06,
      "loss": 0.8674,
      "step": 3278
    },
    {
      "epoch": 0.705312970531297,
      "grad_norm": 0.0,
      "learning_rate": 4.220087543128799e-06,
      "loss": 0.8267,
      "step": 3279
    },
    {
      "epoch": 0.7055280705528071,
      "grad_norm": 0.0,
      "learning_rate": 4.21440326324033e-06,
      "loss": 0.8774,
      "step": 3280
    },
    {
      "epoch": 0.705743170574317,
      "grad_norm": 0.0,
      "learning_rate": 4.2087217919358604e-06,
      "loss": 0.8217,
      "step": 3281
    },
    {
      "epoch": 0.705958270595827,
      "grad_norm": 0.0,
      "learning_rate": 4.203043131973419e-06,
      "loss": 0.8131,
      "step": 3282
    },
    {
      "epoch": 0.7061733706173371,
      "grad_norm": 0.0,
      "learning_rate": 4.197367286109685e-06,
      "loss": 0.8325,
      "step": 3283
    },
    {
      "epoch": 0.706388470638847,
      "grad_norm": 0.0,
      "learning_rate": 4.191694257099962e-06,
      "loss": 0.7934,
      "step": 3284
    },
    {
      "epoch": 0.7066035706603571,
      "grad_norm": 0.0,
      "learning_rate": 4.1860240476981855e-06,
      "loss": 0.7856,
      "step": 3285
    },
    {
      "epoch": 0.7068186706818671,
      "grad_norm": 0.0,
      "learning_rate": 4.180356660656931e-06,
      "loss": 0.8334,
      "step": 3286
    },
    {
      "epoch": 0.707033770703377,
      "grad_norm": 0.0,
      "learning_rate": 4.174692098727392e-06,
      "loss": 0.8568,
      "step": 3287
    },
    {
      "epoch": 0.7072488707248871,
      "grad_norm": 0.0,
      "learning_rate": 4.1690303646594e-06,
      "loss": 0.8201,
      "step": 3288
    },
    {
      "epoch": 0.7074639707463971,
      "grad_norm": 0.0,
      "learning_rate": 4.163371461201415e-06,
      "loss": 0.8519,
      "step": 3289
    },
    {
      "epoch": 0.707679070767907,
      "grad_norm": 0.0,
      "learning_rate": 4.157715391100509e-06,
      "loss": 0.8896,
      "step": 3290
    },
    {
      "epoch": 0.7078941707894171,
      "grad_norm": 0.0,
      "learning_rate": 4.152062157102399e-06,
      "loss": 0.8375,
      "step": 3291
    },
    {
      "epoch": 0.7081092708109271,
      "grad_norm": 0.0,
      "learning_rate": 4.146411761951405e-06,
      "loss": 0.8848,
      "step": 3292
    },
    {
      "epoch": 0.708324370832437,
      "grad_norm": 0.0,
      "learning_rate": 4.1407642083904805e-06,
      "loss": 0.8201,
      "step": 3293
    },
    {
      "epoch": 0.7085394708539471,
      "grad_norm": 0.0,
      "learning_rate": 4.135119499161203e-06,
      "loss": 0.8292,
      "step": 3294
    },
    {
      "epoch": 0.7087545708754571,
      "grad_norm": 0.0,
      "learning_rate": 4.129477637003756e-06,
      "loss": 0.8072,
      "step": 3295
    },
    {
      "epoch": 0.708969670896967,
      "grad_norm": 0.0,
      "learning_rate": 4.123838624656954e-06,
      "loss": 0.8247,
      "step": 3296
    },
    {
      "epoch": 0.7091847709184771,
      "grad_norm": 0.0,
      "learning_rate": 4.1182024648582174e-06,
      "loss": 0.759,
      "step": 3297
    },
    {
      "epoch": 0.7093998709399871,
      "grad_norm": 0.0,
      "learning_rate": 4.112569160343592e-06,
      "loss": 0.8011,
      "step": 3298
    },
    {
      "epoch": 0.7096149709614971,
      "grad_norm": 0.0,
      "learning_rate": 4.10693871384773e-06,
      "loss": 0.8264,
      "step": 3299
    },
    {
      "epoch": 0.7098300709830071,
      "grad_norm": 0.0,
      "learning_rate": 4.101311128103895e-06,
      "loss": 0.8311,
      "step": 3300
    },
    {
      "epoch": 0.7100451710045171,
      "grad_norm": 0.0,
      "learning_rate": 4.095686405843974e-06,
      "loss": 0.8887,
      "step": 3301
    },
    {
      "epoch": 0.7102602710260271,
      "grad_norm": 0.0,
      "learning_rate": 4.090064549798445e-06,
      "loss": 0.8738,
      "step": 3302
    },
    {
      "epoch": 0.7104753710475371,
      "grad_norm": 0.0,
      "learning_rate": 4.084445562696412e-06,
      "loss": 0.8054,
      "step": 3303
    },
    {
      "epoch": 0.7106904710690471,
      "grad_norm": 0.0,
      "learning_rate": 4.07882944726558e-06,
      "loss": 0.8303,
      "step": 3304
    },
    {
      "epoch": 0.7109055710905571,
      "grad_norm": 0.0,
      "learning_rate": 4.073216206232253e-06,
      "loss": 0.806,
      "step": 3305
    },
    {
      "epoch": 0.7111206711120671,
      "grad_norm": 0.0,
      "learning_rate": 4.067605842321351e-06,
      "loss": 0.8685,
      "step": 3306
    },
    {
      "epoch": 0.7113357711335772,
      "grad_norm": 0.0,
      "learning_rate": 4.061998358256395e-06,
      "loss": 0.8806,
      "step": 3307
    },
    {
      "epoch": 0.7115508711550871,
      "grad_norm": 0.0,
      "learning_rate": 4.056393756759498e-06,
      "loss": 0.7702,
      "step": 3308
    },
    {
      "epoch": 0.7117659711765971,
      "grad_norm": 0.0,
      "learning_rate": 4.050792040551388e-06,
      "loss": 0.8558,
      "step": 3309
    },
    {
      "epoch": 0.7119810711981072,
      "grad_norm": 0.0,
      "learning_rate": 4.045193212351377e-06,
      "loss": 0.8164,
      "step": 3310
    },
    {
      "epoch": 0.7121961712196171,
      "grad_norm": 0.0,
      "learning_rate": 4.039597274877388e-06,
      "loss": 0.7863,
      "step": 3311
    },
    {
      "epoch": 0.7124112712411271,
      "grad_norm": 0.0,
      "learning_rate": 4.03400423084594e-06,
      "loss": 0.7666,
      "step": 3312
    },
    {
      "epoch": 0.7126263712626372,
      "grad_norm": 0.0,
      "learning_rate": 4.028414082972141e-06,
      "loss": 0.8864,
      "step": 3313
    },
    {
      "epoch": 0.7128414712841471,
      "grad_norm": 0.0,
      "learning_rate": 4.022826833969692e-06,
      "loss": 0.8624,
      "step": 3314
    },
    {
      "epoch": 0.7130565713056571,
      "grad_norm": 0.0,
      "learning_rate": 4.017242486550889e-06,
      "loss": 0.8247,
      "step": 3315
    },
    {
      "epoch": 0.7132716713271672,
      "grad_norm": 0.0,
      "learning_rate": 4.011661043426624e-06,
      "loss": 0.9008,
      "step": 3316
    },
    {
      "epoch": 0.7134867713486771,
      "grad_norm": 0.0,
      "learning_rate": 4.006082507306379e-06,
      "loss": 0.7618,
      "step": 3317
    },
    {
      "epoch": 0.7137018713701871,
      "grad_norm": 0.0,
      "learning_rate": 4.000506880898215e-06,
      "loss": 0.9087,
      "step": 3318
    },
    {
      "epoch": 0.7139169713916972,
      "grad_norm": 0.0,
      "learning_rate": 3.99493416690879e-06,
      "loss": 0.8943,
      "step": 3319
    },
    {
      "epoch": 0.7141320714132071,
      "grad_norm": 0.0,
      "learning_rate": 3.98936436804335e-06,
      "loss": 0.9004,
      "step": 3320
    },
    {
      "epoch": 0.7143471714347172,
      "grad_norm": 0.0,
      "learning_rate": 3.983797487005713e-06,
      "loss": 0.8558,
      "step": 3321
    },
    {
      "epoch": 0.7145622714562272,
      "grad_norm": 0.0,
      "learning_rate": 3.978233526498296e-06,
      "loss": 0.8216,
      "step": 3322
    },
    {
      "epoch": 0.7147773714777371,
      "grad_norm": 0.0,
      "learning_rate": 3.972672489222083e-06,
      "loss": 0.8233,
      "step": 3323
    },
    {
      "epoch": 0.7149924714992472,
      "grad_norm": 0.0,
      "learning_rate": 3.967114377876654e-06,
      "loss": 0.8813,
      "step": 3324
    },
    {
      "epoch": 0.7152075715207572,
      "grad_norm": 0.0,
      "learning_rate": 3.961559195160162e-06,
      "loss": 0.8895,
      "step": 3325
    },
    {
      "epoch": 0.7154226715422671,
      "grad_norm": 0.0,
      "learning_rate": 3.956006943769331e-06,
      "loss": 0.816,
      "step": 3326
    },
    {
      "epoch": 0.7156377715637772,
      "grad_norm": 0.0,
      "learning_rate": 3.950457626399475e-06,
      "loss": 0.8239,
      "step": 3327
    },
    {
      "epoch": 0.7158528715852872,
      "grad_norm": 0.0,
      "learning_rate": 3.944911245744479e-06,
      "loss": 0.778,
      "step": 3328
    },
    {
      "epoch": 0.7160679716067971,
      "grad_norm": 0.0,
      "learning_rate": 3.9393678044967895e-06,
      "loss": 0.8406,
      "step": 3329
    },
    {
      "epoch": 0.7162830716283072,
      "grad_norm": 0.0,
      "learning_rate": 3.933827305347452e-06,
      "loss": 0.816,
      "step": 3330
    },
    {
      "epoch": 0.7164981716498172,
      "grad_norm": 0.0,
      "learning_rate": 3.928289750986059e-06,
      "loss": 0.8223,
      "step": 3331
    },
    {
      "epoch": 0.7167132716713271,
      "grad_norm": 0.0,
      "learning_rate": 3.922755144100786e-06,
      "loss": 0.811,
      "step": 3332
    },
    {
      "epoch": 0.7169283716928372,
      "grad_norm": 0.0,
      "learning_rate": 3.917223487378381e-06,
      "loss": 0.8512,
      "step": 3333
    },
    {
      "epoch": 0.7171434717143472,
      "grad_norm": 0.0,
      "learning_rate": 3.911694783504148e-06,
      "loss": 0.7962,
      "step": 3334
    },
    {
      "epoch": 0.7173585717358572,
      "grad_norm": 0.0,
      "learning_rate": 3.9061690351619675e-06,
      "loss": 0.8626,
      "step": 3335
    },
    {
      "epoch": 0.7175736717573672,
      "grad_norm": 0.0,
      "learning_rate": 3.900646245034277e-06,
      "loss": 0.837,
      "step": 3336
    },
    {
      "epoch": 0.7177887717788772,
      "grad_norm": 0.0,
      "learning_rate": 3.895126415802088e-06,
      "loss": 0.7966,
      "step": 3337
    },
    {
      "epoch": 0.7180038718003872,
      "grad_norm": 0.0,
      "learning_rate": 3.889609550144968e-06,
      "loss": 0.799,
      "step": 3338
    },
    {
      "epoch": 0.7182189718218972,
      "grad_norm": 0.0,
      "learning_rate": 3.8840956507410445e-06,
      "loss": 0.7719,
      "step": 3339
    },
    {
      "epoch": 0.7184340718434071,
      "grad_norm": 0.0,
      "learning_rate": 3.878584720267011e-06,
      "loss": 0.8472,
      "step": 3340
    },
    {
      "epoch": 0.7186491718649172,
      "grad_norm": 0.0,
      "learning_rate": 3.873076761398112e-06,
      "loss": 0.8382,
      "step": 3341
    },
    {
      "epoch": 0.7188642718864272,
      "grad_norm": 0.0,
      "learning_rate": 3.86757177680816e-06,
      "loss": 0.88,
      "step": 3342
    },
    {
      "epoch": 0.7190793719079371,
      "grad_norm": 0.0,
      "learning_rate": 3.862069769169514e-06,
      "loss": 0.8857,
      "step": 3343
    },
    {
      "epoch": 0.7192944719294472,
      "grad_norm": 0.0,
      "learning_rate": 3.856570741153087e-06,
      "loss": 0.8266,
      "step": 3344
    },
    {
      "epoch": 0.7195095719509572,
      "grad_norm": 0.0,
      "learning_rate": 3.851074695428357e-06,
      "loss": 0.871,
      "step": 3345
    },
    {
      "epoch": 0.7197246719724671,
      "grad_norm": 0.0,
      "learning_rate": 3.845581634663342e-06,
      "loss": 0.7817,
      "step": 3346
    },
    {
      "epoch": 0.7199397719939772,
      "grad_norm": 0.0,
      "learning_rate": 3.840091561524618e-06,
      "loss": 0.8528,
      "step": 3347
    },
    {
      "epoch": 0.7201548720154872,
      "grad_norm": 0.0,
      "learning_rate": 3.834604478677313e-06,
      "loss": 0.7806,
      "step": 3348
    },
    {
      "epoch": 0.7203699720369972,
      "grad_norm": 0.0,
      "learning_rate": 3.829120388785091e-06,
      "loss": 0.8111,
      "step": 3349
    },
    {
      "epoch": 0.7205850720585072,
      "grad_norm": 0.0,
      "learning_rate": 3.823639294510176e-06,
      "loss": 0.8745,
      "step": 3350
    },
    {
      "epoch": 0.7208001720800172,
      "grad_norm": 0.0,
      "learning_rate": 3.818161198513337e-06,
      "loss": 0.8368,
      "step": 3351
    },
    {
      "epoch": 0.7210152721015272,
      "grad_norm": 0.0,
      "learning_rate": 3.8126861034538754e-06,
      "loss": 0.8364,
      "step": 3352
    },
    {
      "epoch": 0.7212303721230372,
      "grad_norm": 0.0,
      "learning_rate": 3.8072140119896507e-06,
      "loss": 0.8579,
      "step": 3353
    },
    {
      "epoch": 0.7214454721445472,
      "grad_norm": 0.0,
      "learning_rate": 3.801744926777051e-06,
      "loss": 0.8299,
      "step": 3354
    },
    {
      "epoch": 0.7216605721660572,
      "grad_norm": 0.0,
      "learning_rate": 3.7962788504710203e-06,
      "loss": 0.8566,
      "step": 3355
    },
    {
      "epoch": 0.7218756721875672,
      "grad_norm": 0.0,
      "learning_rate": 3.790815785725024e-06,
      "loss": 0.7475,
      "step": 3356
    },
    {
      "epoch": 0.7220907722090772,
      "grad_norm": 0.0,
      "learning_rate": 3.785355735191082e-06,
      "loss": 0.8159,
      "step": 3357
    },
    {
      "epoch": 0.7223058722305872,
      "grad_norm": 0.0,
      "learning_rate": 3.779898701519741e-06,
      "loss": 0.7605,
      "step": 3358
    },
    {
      "epoch": 0.7225209722520972,
      "grad_norm": 0.0,
      "learning_rate": 3.774444687360083e-06,
      "loss": 0.8236,
      "step": 3359
    },
    {
      "epoch": 0.7227360722736073,
      "grad_norm": 0.0,
      "learning_rate": 3.7689936953597306e-06,
      "loss": 0.8237,
      "step": 3360
    },
    {
      "epoch": 0.7229511722951172,
      "grad_norm": 0.0,
      "learning_rate": 3.763545728164838e-06,
      "loss": 0.8342,
      "step": 3361
    },
    {
      "epoch": 0.7231662723166272,
      "grad_norm": 0.0,
      "learning_rate": 3.7581007884200826e-06,
      "loss": 0.8037,
      "step": 3362
    },
    {
      "epoch": 0.7233813723381373,
      "grad_norm": 0.0,
      "learning_rate": 3.7526588787686835e-06,
      "loss": 0.8401,
      "step": 3363
    },
    {
      "epoch": 0.7235964723596472,
      "grad_norm": 0.0,
      "learning_rate": 3.747220001852384e-06,
      "loss": 0.8922,
      "step": 3364
    },
    {
      "epoch": 0.7238115723811572,
      "grad_norm": 0.0,
      "learning_rate": 3.7417841603114515e-06,
      "loss": 0.814,
      "step": 3365
    },
    {
      "epoch": 0.7240266724026673,
      "grad_norm": 0.0,
      "learning_rate": 3.736351356784689e-06,
      "loss": 0.8206,
      "step": 3366
    },
    {
      "epoch": 0.7242417724241772,
      "grad_norm": 0.0,
      "learning_rate": 3.7309215939094114e-06,
      "loss": 0.8887,
      "step": 3367
    },
    {
      "epoch": 0.7244568724456872,
      "grad_norm": 0.0,
      "learning_rate": 3.7254948743214693e-06,
      "loss": 0.8598,
      "step": 3368
    },
    {
      "epoch": 0.7246719724671973,
      "grad_norm": 0.0,
      "learning_rate": 3.7200712006552354e-06,
      "loss": 0.789,
      "step": 3369
    },
    {
      "epoch": 0.7248870724887072,
      "grad_norm": 0.0,
      "learning_rate": 3.7146505755435967e-06,
      "loss": 0.8248,
      "step": 3370
    },
    {
      "epoch": 0.7251021725102172,
      "grad_norm": 0.0,
      "learning_rate": 3.709233001617961e-06,
      "loss": 0.7695,
      "step": 3371
    },
    {
      "epoch": 0.7253172725317273,
      "grad_norm": 0.0,
      "learning_rate": 3.7038184815082634e-06,
      "loss": 0.8132,
      "step": 3372
    },
    {
      "epoch": 0.7255323725532372,
      "grad_norm": 0.0,
      "learning_rate": 3.6984070178429456e-06,
      "loss": 0.8293,
      "step": 3373
    },
    {
      "epoch": 0.7257474725747473,
      "grad_norm": 0.0,
      "learning_rate": 3.692998613248977e-06,
      "loss": 0.8438,
      "step": 3374
    },
    {
      "epoch": 0.7259625725962573,
      "grad_norm": 0.0,
      "learning_rate": 3.687593270351828e-06,
      "loss": 0.8776,
      "step": 3375
    },
    {
      "epoch": 0.7261776726177672,
      "grad_norm": 0.0,
      "learning_rate": 3.682190991775495e-06,
      "loss": 0.8213,
      "step": 3376
    },
    {
      "epoch": 0.7263927726392773,
      "grad_norm": 0.0,
      "learning_rate": 3.676791780142487e-06,
      "loss": 0.8083,
      "step": 3377
    },
    {
      "epoch": 0.7266078726607873,
      "grad_norm": 0.0,
      "learning_rate": 3.6713956380738115e-06,
      "loss": 0.8426,
      "step": 3378
    },
    {
      "epoch": 0.7268229726822972,
      "grad_norm": 0.0,
      "learning_rate": 3.666002568189e-06,
      "loss": 0.8066,
      "step": 3379
    },
    {
      "epoch": 0.7270380727038073,
      "grad_norm": 0.0,
      "learning_rate": 3.6606125731060815e-06,
      "loss": 0.8539,
      "step": 3380
    },
    {
      "epoch": 0.7272531727253173,
      "grad_norm": 0.0,
      "learning_rate": 3.6552256554416e-06,
      "loss": 0.8771,
      "step": 3381
    },
    {
      "epoch": 0.7274682727468272,
      "grad_norm": 0.0,
      "learning_rate": 3.6498418178106065e-06,
      "loss": 0.8585,
      "step": 3382
    },
    {
      "epoch": 0.7276833727683373,
      "grad_norm": 0.0,
      "learning_rate": 3.6444610628266466e-06,
      "loss": 0.8217,
      "step": 3383
    },
    {
      "epoch": 0.7278984727898473,
      "grad_norm": 0.0,
      "learning_rate": 3.639083393101783e-06,
      "loss": 0.8084,
      "step": 3384
    },
    {
      "epoch": 0.7281135728113572,
      "grad_norm": 0.0,
      "learning_rate": 3.6337088112465725e-06,
      "loss": 0.8811,
      "step": 3385
    },
    {
      "epoch": 0.7283286728328673,
      "grad_norm": 0.0,
      "learning_rate": 3.6283373198700676e-06,
      "loss": 0.8207,
      "step": 3386
    },
    {
      "epoch": 0.7285437728543773,
      "grad_norm": 0.0,
      "learning_rate": 3.6229689215798357e-06,
      "loss": 0.8738,
      "step": 3387
    },
    {
      "epoch": 0.7287588728758873,
      "grad_norm": 0.0,
      "learning_rate": 3.6176036189819273e-06,
      "loss": 0.8333,
      "step": 3388
    },
    {
      "epoch": 0.7289739728973973,
      "grad_norm": 0.0,
      "learning_rate": 3.6122414146809014e-06,
      "loss": 0.8149,
      "step": 3389
    },
    {
      "epoch": 0.7291890729189073,
      "grad_norm": 0.0,
      "learning_rate": 3.6068823112798114e-06,
      "loss": 0.8419,
      "step": 3390
    },
    {
      "epoch": 0.7294041729404173,
      "grad_norm": 0.0,
      "learning_rate": 3.6015263113801945e-06,
      "loss": 0.8335,
      "step": 3391
    },
    {
      "epoch": 0.7296192729619273,
      "grad_norm": 0.0,
      "learning_rate": 3.596173417582097e-06,
      "loss": 0.8499,
      "step": 3392
    },
    {
      "epoch": 0.7298343729834373,
      "grad_norm": 0.0,
      "learning_rate": 3.590823632484045e-06,
      "loss": 0.6876,
      "step": 3393
    },
    {
      "epoch": 0.7300494730049473,
      "grad_norm": 0.0,
      "learning_rate": 3.5854769586830608e-06,
      "loss": 0.8232,
      "step": 3394
    },
    {
      "epoch": 0.7302645730264573,
      "grad_norm": 0.0,
      "learning_rate": 3.5801333987746612e-06,
      "loss": 0.8097,
      "step": 3395
    },
    {
      "epoch": 0.7304796730479673,
      "grad_norm": 0.0,
      "learning_rate": 3.5747929553528384e-06,
      "loss": 0.847,
      "step": 3396
    },
    {
      "epoch": 0.7306947730694773,
      "grad_norm": 0.0,
      "learning_rate": 3.569455631010086e-06,
      "loss": 0.8704,
      "step": 3397
    },
    {
      "epoch": 0.7309098730909873,
      "grad_norm": 0.0,
      "learning_rate": 3.564121428337369e-06,
      "loss": 0.8367,
      "step": 3398
    },
    {
      "epoch": 0.7311249731124974,
      "grad_norm": 0.0,
      "learning_rate": 3.5587903499241538e-06,
      "loss": 0.7935,
      "step": 3399
    },
    {
      "epoch": 0.7313400731340073,
      "grad_norm": 0.0,
      "learning_rate": 3.5534623983583772e-06,
      "loss": 0.8496,
      "step": 3400
    },
    {
      "epoch": 0.7315551731555173,
      "grad_norm": 0.0,
      "learning_rate": 3.548137576226458e-06,
      "loss": 0.8425,
      "step": 3401
    },
    {
      "epoch": 0.7317702731770274,
      "grad_norm": 0.0,
      "learning_rate": 3.5428158861133055e-06,
      "loss": 0.7775,
      "step": 3402
    },
    {
      "epoch": 0.7319853731985373,
      "grad_norm": 0.0,
      "learning_rate": 3.5374973306022987e-06,
      "loss": 0.8436,
      "step": 3403
    },
    {
      "epoch": 0.7322004732200473,
      "grad_norm": 0.0,
      "learning_rate": 3.532181912275301e-06,
      "loss": 0.8115,
      "step": 3404
    },
    {
      "epoch": 0.7324155732415574,
      "grad_norm": 0.0,
      "learning_rate": 3.5268696337126553e-06,
      "loss": 0.8104,
      "step": 3405
    },
    {
      "epoch": 0.7326306732630673,
      "grad_norm": 0.0,
      "learning_rate": 3.5215604974931685e-06,
      "loss": 0.8203,
      "step": 3406
    },
    {
      "epoch": 0.7328457732845773,
      "grad_norm": 0.0,
      "learning_rate": 3.5162545061941332e-06,
      "loss": 0.8042,
      "step": 3407
    },
    {
      "epoch": 0.7330608733060874,
      "grad_norm": 0.0,
      "learning_rate": 3.510951662391314e-06,
      "loss": 0.7958,
      "step": 3408
    },
    {
      "epoch": 0.7332759733275973,
      "grad_norm": 0.0,
      "learning_rate": 3.5056519686589417e-06,
      "loss": 0.8623,
      "step": 3409
    },
    {
      "epoch": 0.7334910733491073,
      "grad_norm": 0.0,
      "learning_rate": 3.5003554275697247e-06,
      "loss": 0.8751,
      "step": 3410
    },
    {
      "epoch": 0.7337061733706174,
      "grad_norm": 0.0,
      "learning_rate": 3.4950620416948343e-06,
      "loss": 0.839,
      "step": 3411
    },
    {
      "epoch": 0.7339212733921273,
      "grad_norm": 0.0,
      "learning_rate": 3.4897718136039148e-06,
      "loss": 0.8098,
      "step": 3412
    },
    {
      "epoch": 0.7341363734136374,
      "grad_norm": 0.0,
      "learning_rate": 3.4844847458650797e-06,
      "loss": 0.8747,
      "step": 3413
    },
    {
      "epoch": 0.7343514734351474,
      "grad_norm": 0.0,
      "learning_rate": 3.479200841044903e-06,
      "loss": 0.8372,
      "step": 3414
    },
    {
      "epoch": 0.7345665734566573,
      "grad_norm": 0.0,
      "learning_rate": 3.4739201017084243e-06,
      "loss": 0.7872,
      "step": 3415
    },
    {
      "epoch": 0.7347816734781674,
      "grad_norm": 0.0,
      "learning_rate": 3.468642530419144e-06,
      "loss": 0.8288,
      "step": 3416
    },
    {
      "epoch": 0.7349967734996774,
      "grad_norm": 0.0,
      "learning_rate": 3.4633681297390333e-06,
      "loss": 0.8468,
      "step": 3417
    },
    {
      "epoch": 0.7352118735211873,
      "grad_norm": 0.0,
      "learning_rate": 3.4580969022285193e-06,
      "loss": 0.7515,
      "step": 3418
    },
    {
      "epoch": 0.7354269735426974,
      "grad_norm": 0.0,
      "learning_rate": 3.4528288504464847e-06,
      "loss": 0.8337,
      "step": 3419
    },
    {
      "epoch": 0.7356420735642074,
      "grad_norm": 0.0,
      "learning_rate": 3.4475639769502765e-06,
      "loss": 0.8437,
      "step": 3420
    },
    {
      "epoch": 0.7358571735857173,
      "grad_norm": 0.0,
      "learning_rate": 3.4423022842957e-06,
      "loss": 0.8372,
      "step": 3421
    },
    {
      "epoch": 0.7360722736072274,
      "grad_norm": 0.0,
      "learning_rate": 3.437043775037007e-06,
      "loss": 0.7745,
      "step": 3422
    },
    {
      "epoch": 0.7362873736287374,
      "grad_norm": 0.0,
      "learning_rate": 3.431788451726914e-06,
      "loss": 0.7843,
      "step": 3423
    },
    {
      "epoch": 0.7365024736502473,
      "grad_norm": 0.0,
      "learning_rate": 3.4265363169165843e-06,
      "loss": 0.8468,
      "step": 3424
    },
    {
      "epoch": 0.7367175736717574,
      "grad_norm": 0.0,
      "learning_rate": 3.4212873731556364e-06,
      "loss": 0.782,
      "step": 3425
    },
    {
      "epoch": 0.7369326736932673,
      "grad_norm": 0.0,
      "learning_rate": 3.4160416229921432e-06,
      "loss": 0.8175,
      "step": 3426
    },
    {
      "epoch": 0.7371477737147774,
      "grad_norm": 0.0,
      "learning_rate": 3.4107990689726166e-06,
      "loss": 0.859,
      "step": 3427
    },
    {
      "epoch": 0.7373628737362874,
      "grad_norm": 0.0,
      "learning_rate": 3.4055597136420314e-06,
      "loss": 0.8347,
      "step": 3428
    },
    {
      "epoch": 0.7375779737577973,
      "grad_norm": 0.0,
      "learning_rate": 3.4003235595437976e-06,
      "loss": 0.8341,
      "step": 3429
    },
    {
      "epoch": 0.7377930737793074,
      "grad_norm": 0.0,
      "learning_rate": 3.395090609219772e-06,
      "loss": 0.8031,
      "step": 3430
    },
    {
      "epoch": 0.7380081738008174,
      "grad_norm": 0.0,
      "learning_rate": 3.389860865210267e-06,
      "loss": 0.8352,
      "step": 3431
    },
    {
      "epoch": 0.7382232738223273,
      "grad_norm": 0.0,
      "learning_rate": 3.384634330054024e-06,
      "loss": 0.8204,
      "step": 3432
    },
    {
      "epoch": 0.7384383738438374,
      "grad_norm": 0.0,
      "learning_rate": 3.379411006288237e-06,
      "loss": 0.8,
      "step": 3433
    },
    {
      "epoch": 0.7386534738653474,
      "grad_norm": 0.0,
      "learning_rate": 3.3741908964485415e-06,
      "loss": 0.9057,
      "step": 3434
    },
    {
      "epoch": 0.7388685738868573,
      "grad_norm": 0.0,
      "learning_rate": 3.368974003069003e-06,
      "loss": 0.8363,
      "step": 3435
    },
    {
      "epoch": 0.7390836739083674,
      "grad_norm": 0.0,
      "learning_rate": 3.3637603286821376e-06,
      "loss": 0.849,
      "step": 3436
    },
    {
      "epoch": 0.7392987739298774,
      "grad_norm": 0.0,
      "learning_rate": 3.358549875818887e-06,
      "loss": 0.8682,
      "step": 3437
    },
    {
      "epoch": 0.7395138739513873,
      "grad_norm": 0.0,
      "learning_rate": 3.3533426470086384e-06,
      "loss": 0.8456,
      "step": 3438
    },
    {
      "epoch": 0.7397289739728974,
      "grad_norm": 0.0,
      "learning_rate": 3.348138644779212e-06,
      "loss": 0.79,
      "step": 3439
    },
    {
      "epoch": 0.7399440739944074,
      "grad_norm": 0.0,
      "learning_rate": 3.3429378716568563e-06,
      "loss": 0.8464,
      "step": 3440
    },
    {
      "epoch": 0.7401591740159174,
      "grad_norm": 0.0,
      "learning_rate": 3.3377403301662605e-06,
      "loss": 0.7899,
      "step": 3441
    },
    {
      "epoch": 0.7403742740374274,
      "grad_norm": 0.0,
      "learning_rate": 3.332546022830535e-06,
      "loss": 0.7862,
      "step": 3442
    },
    {
      "epoch": 0.7405893740589374,
      "grad_norm": 0.0,
      "learning_rate": 3.327354952171232e-06,
      "loss": 0.7663,
      "step": 3443
    },
    {
      "epoch": 0.7408044740804474,
      "grad_norm": 0.0,
      "learning_rate": 3.322167120708324e-06,
      "loss": 0.8582,
      "step": 3444
    },
    {
      "epoch": 0.7410195741019574,
      "grad_norm": 0.0,
      "learning_rate": 3.316982530960209e-06,
      "loss": 0.772,
      "step": 3445
    },
    {
      "epoch": 0.7412346741234674,
      "grad_norm": 0.0,
      "learning_rate": 3.3118011854437226e-06,
      "loss": 0.7947,
      "step": 3446
    },
    {
      "epoch": 0.7414497741449774,
      "grad_norm": 0.0,
      "learning_rate": 3.3066230866741124e-06,
      "loss": 0.8254,
      "step": 3447
    },
    {
      "epoch": 0.7416648741664874,
      "grad_norm": 0.0,
      "learning_rate": 3.301448237165058e-06,
      "loss": 0.7802,
      "step": 3448
    },
    {
      "epoch": 0.7418799741879974,
      "grad_norm": 0.0,
      "learning_rate": 3.296276639428665e-06,
      "loss": 0.8157,
      "step": 3449
    },
    {
      "epoch": 0.7420950742095074,
      "grad_norm": 0.0,
      "learning_rate": 3.2911082959754472e-06,
      "loss": 0.8011,
      "step": 3450
    },
    {
      "epoch": 0.7423101742310174,
      "grad_norm": 0.0,
      "learning_rate": 3.2859432093143497e-06,
      "loss": 0.8719,
      "step": 3451
    },
    {
      "epoch": 0.7425252742525275,
      "grad_norm": 0.0,
      "learning_rate": 3.2807813819527377e-06,
      "loss": 0.8872,
      "step": 3452
    },
    {
      "epoch": 0.7427403742740374,
      "grad_norm": 0.0,
      "learning_rate": 3.2756228163963845e-06,
      "loss": 0.84,
      "step": 3453
    },
    {
      "epoch": 0.7429554742955474,
      "grad_norm": 0.0,
      "learning_rate": 3.2704675151494893e-06,
      "loss": 0.8048,
      "step": 3454
    },
    {
      "epoch": 0.7431705743170575,
      "grad_norm": 0.0,
      "learning_rate": 3.265315480714657e-06,
      "loss": 0.7626,
      "step": 3455
    },
    {
      "epoch": 0.7433856743385674,
      "grad_norm": 0.0,
      "learning_rate": 3.2601667155929207e-06,
      "loss": 0.9071,
      "step": 3456
    },
    {
      "epoch": 0.7436007743600774,
      "grad_norm": 0.0,
      "learning_rate": 3.2550212222837106e-06,
      "loss": 0.8086,
      "step": 3457
    },
    {
      "epoch": 0.7438158743815875,
      "grad_norm": 0.0,
      "learning_rate": 3.2498790032848836e-06,
      "loss": 0.9005,
      "step": 3458
    },
    {
      "epoch": 0.7440309744030974,
      "grad_norm": 0.0,
      "learning_rate": 3.244740061092695e-06,
      "loss": 0.7694,
      "step": 3459
    },
    {
      "epoch": 0.7442460744246074,
      "grad_norm": 0.0,
      "learning_rate": 3.239604398201811e-06,
      "loss": 0.8148,
      "step": 3460
    },
    {
      "epoch": 0.7444611744461175,
      "grad_norm": 0.0,
      "learning_rate": 3.2344720171053134e-06,
      "loss": 0.7759,
      "step": 3461
    },
    {
      "epoch": 0.7446762744676274,
      "grad_norm": 0.0,
      "learning_rate": 3.2293429202946904e-06,
      "loss": 0.8366,
      "step": 3462
    },
    {
      "epoch": 0.7448913744891374,
      "grad_norm": 0.0,
      "learning_rate": 3.224217110259823e-06,
      "loss": 0.8395,
      "step": 3463
    },
    {
      "epoch": 0.7451064745106475,
      "grad_norm": 0.0,
      "learning_rate": 3.2190945894890103e-06,
      "loss": 0.7455,
      "step": 3464
    },
    {
      "epoch": 0.7453215745321574,
      "grad_norm": 0.0,
      "learning_rate": 3.213975360468954e-06,
      "loss": 0.7668,
      "step": 3465
    },
    {
      "epoch": 0.7455366745536675,
      "grad_norm": 0.0,
      "learning_rate": 3.208859425684745e-06,
      "loss": 0.8569,
      "step": 3466
    },
    {
      "epoch": 0.7457517745751775,
      "grad_norm": 0.0,
      "learning_rate": 3.2037467876198925e-06,
      "loss": 0.8355,
      "step": 3467
    },
    {
      "epoch": 0.7459668745966874,
      "grad_norm": 0.0,
      "learning_rate": 3.1986374487562887e-06,
      "loss": 0.7813,
      "step": 3468
    },
    {
      "epoch": 0.7461819746181975,
      "grad_norm": 0.0,
      "learning_rate": 3.1935314115742376e-06,
      "loss": 0.8324,
      "step": 3469
    },
    {
      "epoch": 0.7463970746397075,
      "grad_norm": 0.0,
      "learning_rate": 3.1884286785524356e-06,
      "loss": 0.825,
      "step": 3470
    },
    {
      "epoch": 0.7466121746612174,
      "grad_norm": 0.0,
      "learning_rate": 3.1833292521679725e-06,
      "loss": 0.8461,
      "step": 3471
    },
    {
      "epoch": 0.7468272746827275,
      "grad_norm": 0.0,
      "learning_rate": 3.178233134896331e-06,
      "loss": 0.7733,
      "step": 3472
    },
    {
      "epoch": 0.7470423747042375,
      "grad_norm": 0.0,
      "learning_rate": 3.1731403292114006e-06,
      "loss": 0.8148,
      "step": 3473
    },
    {
      "epoch": 0.7472574747257474,
      "grad_norm": 0.0,
      "learning_rate": 3.1680508375854434e-06,
      "loss": 0.8552,
      "step": 3474
    },
    {
      "epoch": 0.7474725747472575,
      "grad_norm": 0.0,
      "learning_rate": 3.162964662489132e-06,
      "loss": 0.7629,
      "step": 3475
    },
    {
      "epoch": 0.7476876747687675,
      "grad_norm": 0.0,
      "learning_rate": 3.1578818063915128e-06,
      "loss": 0.8359,
      "step": 3476
    },
    {
      "epoch": 0.7479027747902774,
      "grad_norm": 0.0,
      "learning_rate": 3.1528022717600338e-06,
      "loss": 0.7786,
      "step": 3477
    },
    {
      "epoch": 0.7481178748117875,
      "grad_norm": 0.0,
      "learning_rate": 3.1477260610605255e-06,
      "loss": 0.8366,
      "step": 3478
    },
    {
      "epoch": 0.7483329748332975,
      "grad_norm": 0.0,
      "learning_rate": 3.1426531767572e-06,
      "loss": 0.873,
      "step": 3479
    },
    {
      "epoch": 0.7485480748548075,
      "grad_norm": 0.0,
      "learning_rate": 3.1375836213126653e-06,
      "loss": 0.8455,
      "step": 3480
    },
    {
      "epoch": 0.7487631748763175,
      "grad_norm": 0.0,
      "learning_rate": 3.132517397187902e-06,
      "loss": 0.866,
      "step": 3481
    },
    {
      "epoch": 0.7489782748978275,
      "grad_norm": 0.0,
      "learning_rate": 3.1274545068422803e-06,
      "loss": 0.8536,
      "step": 3482
    },
    {
      "epoch": 0.7491933749193375,
      "grad_norm": 0.0,
      "learning_rate": 3.122394952733554e-06,
      "loss": 0.8552,
      "step": 3483
    },
    {
      "epoch": 0.7494084749408475,
      "grad_norm": 0.0,
      "learning_rate": 3.1173387373178487e-06,
      "loss": 0.9048,
      "step": 3484
    },
    {
      "epoch": 0.7496235749623575,
      "grad_norm": 0.0,
      "learning_rate": 3.112285863049681e-06,
      "loss": 0.8447,
      "step": 3485
    },
    {
      "epoch": 0.7498386749838675,
      "grad_norm": 0.0,
      "learning_rate": 3.1072363323819345e-06,
      "loss": 0.8221,
      "step": 3486
    },
    {
      "epoch": 0.7500537750053775,
      "grad_norm": 0.0,
      "learning_rate": 3.102190147765871e-06,
      "loss": 0.8438,
      "step": 3487
    },
    {
      "epoch": 0.7502688750268875,
      "grad_norm": 0.0,
      "learning_rate": 3.09714731165114e-06,
      "loss": 0.8687,
      "step": 3488
    },
    {
      "epoch": 0.7504839750483975,
      "grad_norm": 0.0,
      "learning_rate": 3.0921078264857474e-06,
      "loss": 0.7132,
      "step": 3489
    },
    {
      "epoch": 0.7506990750699075,
      "grad_norm": 0.0,
      "learning_rate": 3.0870716947160872e-06,
      "loss": 0.8303,
      "step": 3490
    },
    {
      "epoch": 0.7509141750914176,
      "grad_norm": 0.0,
      "learning_rate": 3.0820389187869214e-06,
      "loss": 0.8486,
      "step": 3491
    },
    {
      "epoch": 0.7511292751129275,
      "grad_norm": 0.0,
      "learning_rate": 3.0770095011413757e-06,
      "loss": 0.8279,
      "step": 3492
    },
    {
      "epoch": 0.7513443751344375,
      "grad_norm": 0.0,
      "learning_rate": 3.0719834442209596e-06,
      "loss": 0.9094,
      "step": 3493
    },
    {
      "epoch": 0.7515594751559476,
      "grad_norm": 0.0,
      "learning_rate": 3.066960750465533e-06,
      "loss": 0.8591,
      "step": 3494
    },
    {
      "epoch": 0.7517745751774575,
      "grad_norm": 0.0,
      "learning_rate": 3.061941422313339e-06,
      "loss": 0.8609,
      "step": 3495
    },
    {
      "epoch": 0.7519896751989675,
      "grad_norm": 0.0,
      "learning_rate": 3.0569254622009846e-06,
      "loss": 0.8767,
      "step": 3496
    },
    {
      "epoch": 0.7522047752204776,
      "grad_norm": 0.0,
      "learning_rate": 3.0519128725634294e-06,
      "loss": 0.8506,
      "step": 3497
    },
    {
      "epoch": 0.7524198752419875,
      "grad_norm": 0.0,
      "learning_rate": 3.0469036558340147e-06,
      "loss": 0.7673,
      "step": 3498
    },
    {
      "epoch": 0.7526349752634975,
      "grad_norm": 0.0,
      "learning_rate": 3.0418978144444267e-06,
      "loss": 0.8089,
      "step": 3499
    },
    {
      "epoch": 0.7528500752850076,
      "grad_norm": 0.0,
      "learning_rate": 3.036895350824729e-06,
      "loss": 0.7183,
      "step": 3500
    },
    {
      "epoch": 0.7530651753065175,
      "grad_norm": 0.0,
      "learning_rate": 3.0318962674033358e-06,
      "loss": 0.801,
      "step": 3501
    },
    {
      "epoch": 0.7532802753280275,
      "grad_norm": 0.0,
      "learning_rate": 3.0269005666070185e-06,
      "loss": 0.8857,
      "step": 3502
    },
    {
      "epoch": 0.7534953753495376,
      "grad_norm": 0.0,
      "learning_rate": 3.021908250860918e-06,
      "loss": 0.8514,
      "step": 3503
    },
    {
      "epoch": 0.7537104753710475,
      "grad_norm": 0.0,
      "learning_rate": 3.0169193225885162e-06,
      "loss": 0.874,
      "step": 3504
    },
    {
      "epoch": 0.7539255753925576,
      "grad_norm": 0.0,
      "learning_rate": 3.011933784211666e-06,
      "loss": 0.8266,
      "step": 3505
    },
    {
      "epoch": 0.7541406754140676,
      "grad_norm": 0.0,
      "learning_rate": 3.006951638150567e-06,
      "loss": 0.853,
      "step": 3506
    },
    {
      "epoch": 0.7543557754355775,
      "grad_norm": 0.0,
      "learning_rate": 3.0019728868237695e-06,
      "loss": 0.8087,
      "step": 3507
    },
    {
      "epoch": 0.7545708754570876,
      "grad_norm": 0.0,
      "learning_rate": 2.9969975326481816e-06,
      "loss": 0.8405,
      "step": 3508
    },
    {
      "epoch": 0.7547859754785975,
      "grad_norm": 0.0,
      "learning_rate": 2.9920255780390615e-06,
      "loss": 0.8382,
      "step": 3509
    },
    {
      "epoch": 0.7550010755001075,
      "grad_norm": 0.0,
      "learning_rate": 2.9870570254100097e-06,
      "loss": 0.8585,
      "step": 3510
    },
    {
      "epoch": 0.7552161755216176,
      "grad_norm": 0.0,
      "learning_rate": 2.9820918771729877e-06,
      "loss": 0.8345,
      "step": 3511
    },
    {
      "epoch": 0.7554312755431275,
      "grad_norm": 0.0,
      "learning_rate": 2.977130135738292e-06,
      "loss": 0.836,
      "step": 3512
    },
    {
      "epoch": 0.7556463755646375,
      "grad_norm": 0.0,
      "learning_rate": 2.972171803514573e-06,
      "loss": 0.8135,
      "step": 3513
    },
    {
      "epoch": 0.7558614755861476,
      "grad_norm": 0.0,
      "learning_rate": 2.9672168829088267e-06,
      "loss": 0.7953,
      "step": 3514
    },
    {
      "epoch": 0.7560765756076575,
      "grad_norm": 0.0,
      "learning_rate": 2.9622653763263874e-06,
      "loss": 0.7773,
      "step": 3515
    },
    {
      "epoch": 0.7562916756291675,
      "grad_norm": 0.0,
      "learning_rate": 2.9573172861709366e-06,
      "loss": 0.8097,
      "step": 3516
    },
    {
      "epoch": 0.7565067756506776,
      "grad_norm": 0.0,
      "learning_rate": 2.9523726148444896e-06,
      "loss": 0.8094,
      "step": 3517
    },
    {
      "epoch": 0.7567218756721875,
      "grad_norm": 0.0,
      "learning_rate": 2.947431364747414e-06,
      "loss": 0.8642,
      "step": 3518
    },
    {
      "epoch": 0.7569369756936976,
      "grad_norm": 0.0,
      "learning_rate": 2.9424935382784115e-06,
      "loss": 0.8092,
      "step": 3519
    },
    {
      "epoch": 0.7571520757152076,
      "grad_norm": 0.0,
      "learning_rate": 2.937559137834517e-06,
      "loss": 0.878,
      "step": 3520
    },
    {
      "epoch": 0.7573671757367175,
      "grad_norm": 0.0,
      "learning_rate": 2.9326281658111066e-06,
      "loss": 0.812,
      "step": 3521
    },
    {
      "epoch": 0.7575822757582276,
      "grad_norm": 0.0,
      "learning_rate": 2.927700624601898e-06,
      "loss": 0.8289,
      "step": 3522
    },
    {
      "epoch": 0.7577973757797376,
      "grad_norm": 0.0,
      "learning_rate": 2.9227765165989287e-06,
      "loss": 0.862,
      "step": 3523
    },
    {
      "epoch": 0.7580124758012475,
      "grad_norm": 0.0,
      "learning_rate": 2.9178558441925843e-06,
      "loss": 0.8011,
      "step": 3524
    },
    {
      "epoch": 0.7582275758227576,
      "grad_norm": 0.0,
      "learning_rate": 2.912938609771572e-06,
      "loss": 0.8331,
      "step": 3525
    },
    {
      "epoch": 0.7584426758442676,
      "grad_norm": 0.0,
      "learning_rate": 2.9080248157229353e-06,
      "loss": 0.8612,
      "step": 3526
    },
    {
      "epoch": 0.7586577758657775,
      "grad_norm": 0.0,
      "learning_rate": 2.903114464432051e-06,
      "loss": 0.808,
      "step": 3527
    },
    {
      "epoch": 0.7588728758872876,
      "grad_norm": 0.0,
      "learning_rate": 2.898207558282613e-06,
      "loss": 0.8355,
      "step": 3528
    },
    {
      "epoch": 0.7590879759087976,
      "grad_norm": 0.0,
      "learning_rate": 2.8933040996566585e-06,
      "loss": 0.8005,
      "step": 3529
    },
    {
      "epoch": 0.7593030759303075,
      "grad_norm": 0.0,
      "learning_rate": 2.8884040909345378e-06,
      "loss": 0.8613,
      "step": 3530
    },
    {
      "epoch": 0.7595181759518176,
      "grad_norm": 0.0,
      "learning_rate": 2.8835075344949294e-06,
      "loss": 0.8992,
      "step": 3531
    },
    {
      "epoch": 0.7597332759733276,
      "grad_norm": 0.0,
      "learning_rate": 2.8786144327148426e-06,
      "loss": 0.8036,
      "step": 3532
    },
    {
      "epoch": 0.7599483759948376,
      "grad_norm": 0.0,
      "learning_rate": 2.8737247879696007e-06,
      "loss": 0.7626,
      "step": 3533
    },
    {
      "epoch": 0.7601634760163476,
      "grad_norm": 0.0,
      "learning_rate": 2.868838602632854e-06,
      "loss": 0.8128,
      "step": 3534
    },
    {
      "epoch": 0.7603785760378576,
      "grad_norm": 0.0,
      "learning_rate": 2.8639558790765775e-06,
      "loss": 0.7767,
      "step": 3535
    },
    {
      "epoch": 0.7605936760593676,
      "grad_norm": 0.0,
      "learning_rate": 2.859076619671053e-06,
      "loss": 0.8617,
      "step": 3536
    },
    {
      "epoch": 0.7608087760808776,
      "grad_norm": 0.0,
      "learning_rate": 2.8542008267848963e-06,
      "loss": 0.781,
      "step": 3537
    },
    {
      "epoch": 0.7610238761023876,
      "grad_norm": 0.0,
      "learning_rate": 2.849328502785025e-06,
      "loss": 0.8324,
      "step": 3538
    },
    {
      "epoch": 0.7612389761238976,
      "grad_norm": 0.0,
      "learning_rate": 2.844459650036683e-06,
      "loss": 0.8358,
      "step": 3539
    },
    {
      "epoch": 0.7614540761454076,
      "grad_norm": 0.0,
      "learning_rate": 2.8395942709034297e-06,
      "loss": 0.8146,
      "step": 3540
    },
    {
      "epoch": 0.7616691761669176,
      "grad_norm": 0.0,
      "learning_rate": 2.8347323677471296e-06,
      "loss": 0.8573,
      "step": 3541
    },
    {
      "epoch": 0.7618842761884276,
      "grad_norm": 0.0,
      "learning_rate": 2.829873942927971e-06,
      "loss": 0.8516,
      "step": 3542
    },
    {
      "epoch": 0.7620993762099376,
      "grad_norm": 0.0,
      "learning_rate": 2.8250189988044396e-06,
      "loss": 0.8017,
      "step": 3543
    },
    {
      "epoch": 0.7623144762314477,
      "grad_norm": 0.0,
      "learning_rate": 2.820167537733348e-06,
      "loss": 0.7607,
      "step": 3544
    },
    {
      "epoch": 0.7625295762529576,
      "grad_norm": 0.0,
      "learning_rate": 2.815319562069806e-06,
      "loss": 0.7945,
      "step": 3545
    },
    {
      "epoch": 0.7627446762744676,
      "grad_norm": 0.0,
      "learning_rate": 2.8104750741672314e-06,
      "loss": 0.7912,
      "step": 3546
    },
    {
      "epoch": 0.7629597762959777,
      "grad_norm": 0.0,
      "learning_rate": 2.805634076377356e-06,
      "loss": 0.8817,
      "step": 3547
    },
    {
      "epoch": 0.7631748763174876,
      "grad_norm": 0.0,
      "learning_rate": 2.8007965710502162e-06,
      "loss": 0.8237,
      "step": 3548
    },
    {
      "epoch": 0.7633899763389976,
      "grad_norm": 0.0,
      "learning_rate": 2.7959625605341446e-06,
      "loss": 0.8073,
      "step": 3549
    },
    {
      "epoch": 0.7636050763605077,
      "grad_norm": 0.0,
      "learning_rate": 2.7911320471757887e-06,
      "loss": 0.8401,
      "step": 3550
    },
    {
      "epoch": 0.7638201763820176,
      "grad_norm": 0.0,
      "learning_rate": 2.7863050333200894e-06,
      "loss": 0.8643,
      "step": 3551
    },
    {
      "epoch": 0.7640352764035276,
      "grad_norm": 0.0,
      "learning_rate": 2.781481521310292e-06,
      "loss": 0.8028,
      "step": 3552
    },
    {
      "epoch": 0.7642503764250377,
      "grad_norm": 0.0,
      "learning_rate": 2.776661513487947e-06,
      "loss": 0.8299,
      "step": 3553
    },
    {
      "epoch": 0.7644654764465476,
      "grad_norm": 0.0,
      "learning_rate": 2.771845012192892e-06,
      "loss": 0.797,
      "step": 3554
    },
    {
      "epoch": 0.7646805764680576,
      "grad_norm": 0.0,
      "learning_rate": 2.767032019763277e-06,
      "loss": 0.7717,
      "step": 3555
    },
    {
      "epoch": 0.7648956764895677,
      "grad_norm": 0.0,
      "learning_rate": 2.762222538535533e-06,
      "loss": 0.7942,
      "step": 3556
    },
    {
      "epoch": 0.7651107765110776,
      "grad_norm": 0.0,
      "learning_rate": 2.757416570844401e-06,
      "loss": 0.8323,
      "step": 3557
    },
    {
      "epoch": 0.7653258765325877,
      "grad_norm": 0.0,
      "learning_rate": 2.752614119022904e-06,
      "loss": 0.852,
      "step": 3558
    },
    {
      "epoch": 0.7655409765540977,
      "grad_norm": 0.0,
      "learning_rate": 2.7478151854023704e-06,
      "loss": 0.8417,
      "step": 3559
    },
    {
      "epoch": 0.7657560765756076,
      "grad_norm": 0.0,
      "learning_rate": 2.743019772312412e-06,
      "loss": 0.8497,
      "step": 3560
    },
    {
      "epoch": 0.7659711765971177,
      "grad_norm": 0.0,
      "learning_rate": 2.738227882080928e-06,
      "loss": 0.8251,
      "step": 3561
    },
    {
      "epoch": 0.7661862766186277,
      "grad_norm": 0.0,
      "learning_rate": 2.73343951703412e-06,
      "loss": 0.7889,
      "step": 3562
    },
    {
      "epoch": 0.7664013766401376,
      "grad_norm": 0.0,
      "learning_rate": 2.728654679496473e-06,
      "loss": 0.8425,
      "step": 3563
    },
    {
      "epoch": 0.7666164766616477,
      "grad_norm": 0.0,
      "learning_rate": 2.7238733717907518e-06,
      "loss": 0.7793,
      "step": 3564
    },
    {
      "epoch": 0.7668315766831577,
      "grad_norm": 0.0,
      "learning_rate": 2.7190955962380184e-06,
      "loss": 0.7888,
      "step": 3565
    },
    {
      "epoch": 0.7670466767046676,
      "grad_norm": 0.0,
      "learning_rate": 2.7143213551576177e-06,
      "loss": 0.7984,
      "step": 3566
    },
    {
      "epoch": 0.7672617767261777,
      "grad_norm": 0.0,
      "learning_rate": 2.709550650867172e-06,
      "loss": 0.8385,
      "step": 3567
    },
    {
      "epoch": 0.7674768767476877,
      "grad_norm": 0.0,
      "learning_rate": 2.7047834856825972e-06,
      "loss": 0.7361,
      "step": 3568
    },
    {
      "epoch": 0.7676919767691976,
      "grad_norm": 0.0,
      "learning_rate": 2.7000198619180795e-06,
      "loss": 0.7814,
      "step": 3569
    },
    {
      "epoch": 0.7679070767907077,
      "grad_norm": 0.0,
      "learning_rate": 2.695259781886096e-06,
      "loss": 0.8246,
      "step": 3570
    },
    {
      "epoch": 0.7681221768122177,
      "grad_norm": 0.0,
      "learning_rate": 2.6905032478974025e-06,
      "loss": 0.7805,
      "step": 3571
    },
    {
      "epoch": 0.7683372768337277,
      "grad_norm": 0.0,
      "learning_rate": 2.6857502622610267e-06,
      "loss": 0.7737,
      "step": 3572
    },
    {
      "epoch": 0.7685523768552377,
      "grad_norm": 0.0,
      "learning_rate": 2.681000827284275e-06,
      "loss": 0.7525,
      "step": 3573
    },
    {
      "epoch": 0.7687674768767477,
      "grad_norm": 0.0,
      "learning_rate": 2.6762549452727383e-06,
      "loss": 0.8155,
      "step": 3574
    },
    {
      "epoch": 0.7689825768982577,
      "grad_norm": 0.0,
      "learning_rate": 2.671512618530272e-06,
      "loss": 0.8516,
      "step": 3575
    },
    {
      "epoch": 0.7691976769197677,
      "grad_norm": 0.0,
      "learning_rate": 2.6667738493590165e-06,
      "loss": 0.8182,
      "step": 3576
    },
    {
      "epoch": 0.7694127769412777,
      "grad_norm": 0.0,
      "learning_rate": 2.6620386400593733e-06,
      "loss": 0.8184,
      "step": 3577
    },
    {
      "epoch": 0.7696278769627877,
      "grad_norm": 0.0,
      "learning_rate": 2.6573069929300243e-06,
      "loss": 0.8455,
      "step": 3578
    },
    {
      "epoch": 0.7698429769842977,
      "grad_norm": 0.0,
      "learning_rate": 2.6525789102679256e-06,
      "loss": 0.76,
      "step": 3579
    },
    {
      "epoch": 0.7700580770058078,
      "grad_norm": 0.0,
      "learning_rate": 2.647854394368288e-06,
      "loss": 0.7416,
      "step": 3580
    },
    {
      "epoch": 0.7702731770273177,
      "grad_norm": 0.0,
      "learning_rate": 2.6431334475246074e-06,
      "loss": 0.7753,
      "step": 3581
    },
    {
      "epoch": 0.7704882770488277,
      "grad_norm": 0.0,
      "learning_rate": 2.6384160720286334e-06,
      "loss": 0.7652,
      "step": 3582
    },
    {
      "epoch": 0.7707033770703378,
      "grad_norm": 0.0,
      "learning_rate": 2.6337022701703907e-06,
      "loss": 0.8328,
      "step": 3583
    },
    {
      "epoch": 0.7709184770918477,
      "grad_norm": 0.0,
      "learning_rate": 2.628992044238172e-06,
      "loss": 0.831,
      "step": 3584
    },
    {
      "epoch": 0.7711335771133577,
      "grad_norm": 0.0,
      "learning_rate": 2.624285396518521e-06,
      "loss": 0.8608,
      "step": 3585
    },
    {
      "epoch": 0.7713486771348678,
      "grad_norm": 0.0,
      "learning_rate": 2.6195823292962587e-06,
      "loss": 0.8395,
      "step": 3586
    },
    {
      "epoch": 0.7715637771563777,
      "grad_norm": 0.0,
      "learning_rate": 2.6148828448544574e-06,
      "loss": 0.7714,
      "step": 3587
    },
    {
      "epoch": 0.7717788771778877,
      "grad_norm": 0.0,
      "learning_rate": 2.610186945474453e-06,
      "loss": 0.812,
      "step": 3588
    },
    {
      "epoch": 0.7719939771993978,
      "grad_norm": 0.0,
      "learning_rate": 2.6054946334358478e-06,
      "loss": 0.8331,
      "step": 3589
    },
    {
      "epoch": 0.7722090772209077,
      "grad_norm": 0.0,
      "learning_rate": 2.6008059110164906e-06,
      "loss": 0.8208,
      "step": 3590
    },
    {
      "epoch": 0.7724241772424177,
      "grad_norm": 0.0,
      "learning_rate": 2.596120780492497e-06,
      "loss": 0.7524,
      "step": 3591
    },
    {
      "epoch": 0.7726392772639278,
      "grad_norm": 0.0,
      "learning_rate": 2.5914392441382408e-06,
      "loss": 0.7573,
      "step": 3592
    },
    {
      "epoch": 0.7728543772854377,
      "grad_norm": 0.0,
      "learning_rate": 2.5867613042263384e-06,
      "loss": 0.88,
      "step": 3593
    },
    {
      "epoch": 0.7730694773069477,
      "grad_norm": 0.0,
      "learning_rate": 2.5820869630276747e-06,
      "loss": 0.8658,
      "step": 3594
    },
    {
      "epoch": 0.7732845773284577,
      "grad_norm": 0.0,
      "learning_rate": 2.5774162228113765e-06,
      "loss": 0.8739,
      "step": 3595
    },
    {
      "epoch": 0.7734996773499677,
      "grad_norm": 0.0,
      "learning_rate": 2.572749085844829e-06,
      "loss": 0.8657,
      "step": 3596
    },
    {
      "epoch": 0.7737147773714778,
      "grad_norm": 0.0,
      "learning_rate": 2.568085554393669e-06,
      "loss": 0.8226,
      "step": 3597
    },
    {
      "epoch": 0.7739298773929877,
      "grad_norm": 0.0,
      "learning_rate": 2.5634256307217766e-06,
      "loss": 0.814,
      "step": 3598
    },
    {
      "epoch": 0.7741449774144977,
      "grad_norm": 0.0,
      "learning_rate": 2.5587693170912876e-06,
      "loss": 0.7922,
      "step": 3599
    },
    {
      "epoch": 0.7743600774360078,
      "grad_norm": 0.0,
      "learning_rate": 2.5541166157625783e-06,
      "loss": 0.7878,
      "step": 3600
    },
    {
      "epoch": 0.7745751774575177,
      "grad_norm": 0.0,
      "learning_rate": 2.5494675289942806e-06,
      "loss": 0.8881,
      "step": 3601
    },
    {
      "epoch": 0.7747902774790277,
      "grad_norm": 0.0,
      "learning_rate": 2.5448220590432604e-06,
      "loss": 0.8459,
      "step": 3602
    },
    {
      "epoch": 0.7750053775005378,
      "grad_norm": 0.0,
      "learning_rate": 2.54018020816464e-06,
      "loss": 0.8535,
      "step": 3603
    },
    {
      "epoch": 0.7752204775220477,
      "grad_norm": 0.0,
      "learning_rate": 2.535541978611775e-06,
      "loss": 0.8069,
      "step": 3604
    },
    {
      "epoch": 0.7754355775435577,
      "grad_norm": 0.0,
      "learning_rate": 2.5309073726362656e-06,
      "loss": 0.8135,
      "step": 3605
    },
    {
      "epoch": 0.7756506775650678,
      "grad_norm": 0.0,
      "learning_rate": 2.5262763924879543e-06,
      "loss": 0.8008,
      "step": 3606
    },
    {
      "epoch": 0.7758657775865777,
      "grad_norm": 0.0,
      "learning_rate": 2.5216490404149286e-06,
      "loss": 0.8362,
      "step": 3607
    },
    {
      "epoch": 0.7760808776080877,
      "grad_norm": 0.0,
      "learning_rate": 2.517025318663502e-06,
      "loss": 0.8018,
      "step": 3608
    },
    {
      "epoch": 0.7762959776295978,
      "grad_norm": 0.0,
      "learning_rate": 2.5124052294782353e-06,
      "loss": 0.8111,
      "step": 3609
    },
    {
      "epoch": 0.7765110776511077,
      "grad_norm": 0.0,
      "learning_rate": 2.5077887751019294e-06,
      "loss": 0.8349,
      "step": 3610
    },
    {
      "epoch": 0.7767261776726178,
      "grad_norm": 0.0,
      "learning_rate": 2.5031759577756077e-06,
      "loss": 0.8328,
      "step": 3611
    },
    {
      "epoch": 0.7769412776941278,
      "grad_norm": 0.0,
      "learning_rate": 2.4985667797385403e-06,
      "loss": 0.84,
      "step": 3612
    },
    {
      "epoch": 0.7771563777156377,
      "grad_norm": 0.0,
      "learning_rate": 2.493961243228221e-06,
      "loss": 0.7584,
      "step": 3613
    },
    {
      "epoch": 0.7773714777371478,
      "grad_norm": 0.0,
      "learning_rate": 2.4893593504803824e-06,
      "loss": 0.7885,
      "step": 3614
    },
    {
      "epoch": 0.7775865777586578,
      "grad_norm": 0.0,
      "learning_rate": 2.484761103728991e-06,
      "loss": 0.8381,
      "step": 3615
    },
    {
      "epoch": 0.7778016777801677,
      "grad_norm": 0.0,
      "learning_rate": 2.4801665052062342e-06,
      "loss": 0.7956,
      "step": 3616
    },
    {
      "epoch": 0.7780167778016778,
      "grad_norm": 0.0,
      "learning_rate": 2.47557555714253e-06,
      "loss": 0.8226,
      "step": 3617
    },
    {
      "epoch": 0.7782318778231878,
      "grad_norm": 0.0,
      "learning_rate": 2.4709882617665337e-06,
      "loss": 0.8299,
      "step": 3618
    },
    {
      "epoch": 0.7784469778446977,
      "grad_norm": 0.0,
      "learning_rate": 2.466404621305115e-06,
      "loss": 0.868,
      "step": 3619
    },
    {
      "epoch": 0.7786620778662078,
      "grad_norm": 0.0,
      "learning_rate": 2.461824637983381e-06,
      "loss": 0.8727,
      "step": 3620
    },
    {
      "epoch": 0.7788771778877178,
      "grad_norm": 0.0,
      "learning_rate": 2.4572483140246517e-06,
      "loss": 0.8722,
      "step": 3621
    },
    {
      "epoch": 0.7790922779092277,
      "grad_norm": 0.0,
      "learning_rate": 2.452675651650479e-06,
      "loss": 0.8033,
      "step": 3622
    },
    {
      "epoch": 0.7793073779307378,
      "grad_norm": 0.0,
      "learning_rate": 2.44810665308064e-06,
      "loss": 0.7941,
      "step": 3623
    },
    {
      "epoch": 0.7795224779522478,
      "grad_norm": 0.0,
      "learning_rate": 2.4435413205331204e-06,
      "loss": 0.8158,
      "step": 3624
    },
    {
      "epoch": 0.7797375779737578,
      "grad_norm": 0.0,
      "learning_rate": 2.4389796562241408e-06,
      "loss": 0.8143,
      "step": 3625
    },
    {
      "epoch": 0.7799526779952678,
      "grad_norm": 0.0,
      "learning_rate": 2.434421662368128e-06,
      "loss": 0.8271,
      "step": 3626
    },
    {
      "epoch": 0.7801677780167778,
      "grad_norm": 0.0,
      "learning_rate": 2.4298673411777387e-06,
      "loss": 0.8798,
      "step": 3627
    },
    {
      "epoch": 0.7803828780382878,
      "grad_norm": 0.0,
      "learning_rate": 2.4253166948638394e-06,
      "loss": 0.8493,
      "step": 3628
    },
    {
      "epoch": 0.7805979780597978,
      "grad_norm": 0.0,
      "learning_rate": 2.4207697256355144e-06,
      "loss": 0.831,
      "step": 3629
    },
    {
      "epoch": 0.7808130780813078,
      "grad_norm": 0.0,
      "learning_rate": 2.4162264357000652e-06,
      "loss": 0.8398,
      "step": 3630
    },
    {
      "epoch": 0.7810281781028178,
      "grad_norm": 0.0,
      "learning_rate": 2.411686827263003e-06,
      "loss": 0.8504,
      "step": 3631
    },
    {
      "epoch": 0.7812432781243278,
      "grad_norm": 0.0,
      "learning_rate": 2.4071509025280526e-06,
      "loss": 0.8095,
      "step": 3632
    },
    {
      "epoch": 0.7814583781458379,
      "grad_norm": 0.0,
      "learning_rate": 2.402618663697157e-06,
      "loss": 0.8074,
      "step": 3633
    },
    {
      "epoch": 0.7816734781673478,
      "grad_norm": 0.0,
      "learning_rate": 2.3980901129704593e-06,
      "loss": 0.8497,
      "step": 3634
    },
    {
      "epoch": 0.7818885781888578,
      "grad_norm": 0.0,
      "learning_rate": 2.3935652525463206e-06,
      "loss": 0.9335,
      "step": 3635
    },
    {
      "epoch": 0.7821036782103679,
      "grad_norm": 0.0,
      "learning_rate": 2.3890440846213103e-06,
      "loss": 0.8081,
      "step": 3636
    },
    {
      "epoch": 0.7823187782318778,
      "grad_norm": 0.0,
      "learning_rate": 2.384526611390198e-06,
      "loss": 0.794,
      "step": 3637
    },
    {
      "epoch": 0.7825338782533878,
      "grad_norm": 0.0,
      "learning_rate": 2.3800128350459695e-06,
      "loss": 0.8225,
      "step": 3638
    },
    {
      "epoch": 0.7827489782748979,
      "grad_norm": 0.0,
      "learning_rate": 2.375502757779806e-06,
      "loss": 0.7981,
      "step": 3639
    },
    {
      "epoch": 0.7829640782964078,
      "grad_norm": 0.0,
      "learning_rate": 2.370996381781101e-06,
      "loss": 0.8056,
      "step": 3640
    },
    {
      "epoch": 0.7831791783179178,
      "grad_norm": 0.0,
      "learning_rate": 2.3664937092374497e-06,
      "loss": 0.8251,
      "step": 3641
    },
    {
      "epoch": 0.7833942783394279,
      "grad_norm": 0.0,
      "learning_rate": 2.3619947423346447e-06,
      "loss": 0.8447,
      "step": 3642
    },
    {
      "epoch": 0.7836093783609378,
      "grad_norm": 0.0,
      "learning_rate": 2.3574994832566868e-06,
      "loss": 0.8418,
      "step": 3643
    },
    {
      "epoch": 0.7838244783824478,
      "grad_norm": 0.0,
      "learning_rate": 2.3530079341857682e-06,
      "loss": 0.7876,
      "step": 3644
    },
    {
      "epoch": 0.7840395784039579,
      "grad_norm": 0.0,
      "learning_rate": 2.3485200973022904e-06,
      "loss": 0.9171,
      "step": 3645
    },
    {
      "epoch": 0.7842546784254678,
      "grad_norm": 0.0,
      "learning_rate": 2.3440359747848452e-06,
      "loss": 0.875,
      "step": 3646
    },
    {
      "epoch": 0.7844697784469779,
      "grad_norm": 0.0,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.8066,
      "step": 3647
    },
    {
      "epoch": 0.7846848784684879,
      "grad_norm": 0.0,
      "learning_rate": 2.3350788815534063e-06,
      "loss": 0.8182,
      "step": 3648
    },
    {
      "epoch": 0.7848999784899978,
      "grad_norm": 0.0,
      "learning_rate": 2.330605915187587e-06,
      "loss": 0.8235,
      "step": 3649
    },
    {
      "epoch": 0.7851150785115079,
      "grad_norm": 0.0,
      "learning_rate": 2.3261366718841307e-06,
      "loss": 0.8242,
      "step": 3650
    },
    {
      "epoch": 0.7853301785330179,
      "grad_norm": 0.0,
      "learning_rate": 2.3216711538126134e-06,
      "loss": 0.7175,
      "step": 3651
    },
    {
      "epoch": 0.7855452785545278,
      "grad_norm": 0.0,
      "learning_rate": 2.3172093631407864e-06,
      "loss": 0.8065,
      "step": 3652
    },
    {
      "epoch": 0.7857603785760379,
      "grad_norm": 0.0,
      "learning_rate": 2.312751302034605e-06,
      "loss": 0.7602,
      "step": 3653
    },
    {
      "epoch": 0.7859754785975479,
      "grad_norm": 0.0,
      "learning_rate": 2.3082969726582107e-06,
      "loss": 0.8188,
      "step": 3654
    },
    {
      "epoch": 0.7861905786190578,
      "grad_norm": 0.0,
      "learning_rate": 2.303846377173925e-06,
      "loss": 0.8467,
      "step": 3655
    },
    {
      "epoch": 0.7864056786405679,
      "grad_norm": 0.0,
      "learning_rate": 2.29939951774227e-06,
      "loss": 0.8326,
      "step": 3656
    },
    {
      "epoch": 0.7866207786620779,
      "grad_norm": 0.0,
      "learning_rate": 2.2949563965219424e-06,
      "loss": 0.8065,
      "step": 3657
    },
    {
      "epoch": 0.7868358786835878,
      "grad_norm": 0.0,
      "learning_rate": 2.290517015669833e-06,
      "loss": 0.797,
      "step": 3658
    },
    {
      "epoch": 0.7870509787050979,
      "grad_norm": 0.0,
      "learning_rate": 2.2860813773410105e-06,
      "loss": 0.8049,
      "step": 3659
    },
    {
      "epoch": 0.7872660787266079,
      "grad_norm": 0.0,
      "learning_rate": 2.2816494836887326e-06,
      "loss": 0.8465,
      "step": 3660
    },
    {
      "epoch": 0.7874811787481178,
      "grad_norm": 0.0,
      "learning_rate": 2.2772213368644334e-06,
      "loss": 0.8664,
      "step": 3661
    },
    {
      "epoch": 0.7876962787696279,
      "grad_norm": 0.0,
      "learning_rate": 2.2727969390177306e-06,
      "loss": 0.8373,
      "step": 3662
    },
    {
      "epoch": 0.7879113787911379,
      "grad_norm": 0.0,
      "learning_rate": 2.2683762922964214e-06,
      "loss": 0.8594,
      "step": 3663
    },
    {
      "epoch": 0.7881264788126479,
      "grad_norm": 0.0,
      "learning_rate": 2.263959398846488e-06,
      "loss": 0.8283,
      "step": 3664
    },
    {
      "epoch": 0.7883415788341579,
      "grad_norm": 0.0,
      "learning_rate": 2.25954626081208e-06,
      "loss": 0.8552,
      "step": 3665
    },
    {
      "epoch": 0.7885566788556679,
      "grad_norm": 0.0,
      "learning_rate": 2.25513688033553e-06,
      "loss": 0.8011,
      "step": 3666
    },
    {
      "epoch": 0.7887717788771779,
      "grad_norm": 0.0,
      "learning_rate": 2.2507312595573505e-06,
      "loss": 0.8083,
      "step": 3667
    },
    {
      "epoch": 0.7889868788986879,
      "grad_norm": 0.0,
      "learning_rate": 2.2463294006162185e-06,
      "loss": 0.861,
      "step": 3668
    },
    {
      "epoch": 0.7892019789201979,
      "grad_norm": 0.0,
      "learning_rate": 2.2419313056489954e-06,
      "loss": 0.7971,
      "step": 3669
    },
    {
      "epoch": 0.7894170789417079,
      "grad_norm": 0.0,
      "learning_rate": 2.237536976790706e-06,
      "loss": 0.8752,
      "step": 3670
    },
    {
      "epoch": 0.7896321789632179,
      "grad_norm": 0.0,
      "learning_rate": 2.2331464161745532e-06,
      "loss": 0.784,
      "step": 3671
    },
    {
      "epoch": 0.789847278984728,
      "grad_norm": 0.0,
      "learning_rate": 2.228759625931912e-06,
      "loss": 0.7979,
      "step": 3672
    },
    {
      "epoch": 0.7900623790062379,
      "grad_norm": 0.0,
      "learning_rate": 2.22437660819232e-06,
      "loss": 0.839,
      "step": 3673
    },
    {
      "epoch": 0.7902774790277479,
      "grad_norm": 0.0,
      "learning_rate": 2.2199973650834903e-06,
      "loss": 0.8222,
      "step": 3674
    },
    {
      "epoch": 0.790492579049258,
      "grad_norm": 0.0,
      "learning_rate": 2.2156218987313003e-06,
      "loss": 0.7779,
      "step": 3675
    },
    {
      "epoch": 0.7907076790707679,
      "grad_norm": 0.0,
      "learning_rate": 2.21125021125979e-06,
      "loss": 0.8214,
      "step": 3676
    },
    {
      "epoch": 0.7909227790922779,
      "grad_norm": 0.0,
      "learning_rate": 2.206882304791176e-06,
      "loss": 0.8106,
      "step": 3677
    },
    {
      "epoch": 0.7911378791137879,
      "grad_norm": 0.0,
      "learning_rate": 2.2025181814458276e-06,
      "loss": 0.8231,
      "step": 3678
    },
    {
      "epoch": 0.7913529791352979,
      "grad_norm": 0.0,
      "learning_rate": 2.1981578433422844e-06,
      "loss": 0.8598,
      "step": 3679
    },
    {
      "epoch": 0.7915680791568079,
      "grad_norm": 0.0,
      "learning_rate": 2.1938012925972507e-06,
      "loss": 0.8025,
      "step": 3680
    },
    {
      "epoch": 0.7917831791783179,
      "grad_norm": 0.0,
      "learning_rate": 2.189448531325582e-06,
      "loss": 0.7999,
      "step": 3681
    },
    {
      "epoch": 0.7919982791998279,
      "grad_norm": 0.0,
      "learning_rate": 2.1850995616403083e-06,
      "loss": 0.7811,
      "step": 3682
    },
    {
      "epoch": 0.7922133792213379,
      "grad_norm": 0.0,
      "learning_rate": 2.1807543856526037e-06,
      "loss": 0.7173,
      "step": 3683
    },
    {
      "epoch": 0.7924284792428479,
      "grad_norm": 0.0,
      "learning_rate": 2.1764130054718123e-06,
      "loss": 0.8026,
      "step": 3684
    },
    {
      "epoch": 0.7926435792643579,
      "grad_norm": 0.0,
      "learning_rate": 2.1720754232054333e-06,
      "loss": 0.792,
      "step": 3685
    },
    {
      "epoch": 0.792858679285868,
      "grad_norm": 0.0,
      "learning_rate": 2.167741640959117e-06,
      "loss": 0.7994,
      "step": 3686
    },
    {
      "epoch": 0.7930737793073779,
      "grad_norm": 0.0,
      "learning_rate": 2.163411660836676e-06,
      "loss": 0.8022,
      "step": 3687
    },
    {
      "epoch": 0.7932888793288879,
      "grad_norm": 0.0,
      "learning_rate": 2.15908548494007e-06,
      "loss": 0.8241,
      "step": 3688
    },
    {
      "epoch": 0.793503979350398,
      "grad_norm": 0.0,
      "learning_rate": 2.1547631153694193e-06,
      "loss": 0.7797,
      "step": 3689
    },
    {
      "epoch": 0.7937190793719079,
      "grad_norm": 0.0,
      "learning_rate": 2.1504445542229923e-06,
      "loss": 0.8033,
      "step": 3690
    },
    {
      "epoch": 0.7939341793934179,
      "grad_norm": 0.0,
      "learning_rate": 2.1461298035972055e-06,
      "loss": 0.802,
      "step": 3691
    },
    {
      "epoch": 0.794149279414928,
      "grad_norm": 0.0,
      "learning_rate": 2.1418188655866314e-06,
      "loss": 0.7829,
      "step": 3692
    },
    {
      "epoch": 0.7943643794364379,
      "grad_norm": 0.0,
      "learning_rate": 2.137511742283992e-06,
      "loss": 0.8226,
      "step": 3693
    },
    {
      "epoch": 0.7945794794579479,
      "grad_norm": 0.0,
      "learning_rate": 2.1332084357801508e-06,
      "loss": 0.7864,
      "step": 3694
    },
    {
      "epoch": 0.794794579479458,
      "grad_norm": 0.0,
      "learning_rate": 2.128908948164128e-06,
      "loss": 0.829,
      "step": 3695
    },
    {
      "epoch": 0.7950096795009679,
      "grad_norm": 0.0,
      "learning_rate": 2.124613281523079e-06,
      "loss": 0.8479,
      "step": 3696
    },
    {
      "epoch": 0.7952247795224779,
      "grad_norm": 0.0,
      "learning_rate": 2.120321437942312e-06,
      "loss": 0.8191,
      "step": 3697
    },
    {
      "epoch": 0.795439879543988,
      "grad_norm": 0.0,
      "learning_rate": 2.11603341950528e-06,
      "loss": 0.8431,
      "step": 3698
    },
    {
      "epoch": 0.7956549795654979,
      "grad_norm": 0.0,
      "learning_rate": 2.1117492282935726e-06,
      "loss": 0.822,
      "step": 3699
    },
    {
      "epoch": 0.795870079587008,
      "grad_norm": 0.0,
      "learning_rate": 2.1074688663869302e-06,
      "loss": 0.7946,
      "step": 3700
    },
    {
      "epoch": 0.796085179608518,
      "grad_norm": 0.0,
      "learning_rate": 2.103192335863222e-06,
      "loss": 0.8457,
      "step": 3701
    },
    {
      "epoch": 0.7963002796300279,
      "grad_norm": 0.0,
      "learning_rate": 2.0989196387984724e-06,
      "loss": 0.7922,
      "step": 3702
    },
    {
      "epoch": 0.796515379651538,
      "grad_norm": 0.0,
      "learning_rate": 2.09465077726683e-06,
      "loss": 0.8217,
      "step": 3703
    },
    {
      "epoch": 0.796730479673048,
      "grad_norm": 0.0,
      "learning_rate": 2.090385753340596e-06,
      "loss": 0.7897,
      "step": 3704
    },
    {
      "epoch": 0.7969455796945579,
      "grad_norm": 0.0,
      "learning_rate": 2.0861245690901953e-06,
      "loss": 0.8298,
      "step": 3705
    },
    {
      "epoch": 0.797160679716068,
      "grad_norm": 0.0,
      "learning_rate": 2.0818672265841955e-06,
      "loss": 0.8627,
      "step": 3706
    },
    {
      "epoch": 0.797375779737578,
      "grad_norm": 0.0,
      "learning_rate": 2.0776137278893005e-06,
      "loss": 0.7784,
      "step": 3707
    },
    {
      "epoch": 0.7975908797590879,
      "grad_norm": 0.0,
      "learning_rate": 2.0733640750703475e-06,
      "loss": 0.8231,
      "step": 3708
    },
    {
      "epoch": 0.797805979780598,
      "grad_norm": 0.0,
      "learning_rate": 2.069118270190301e-06,
      "loss": 0.819,
      "step": 3709
    },
    {
      "epoch": 0.798021079802108,
      "grad_norm": 0.0,
      "learning_rate": 2.064876315310265e-06,
      "loss": 0.8809,
      "step": 3710
    },
    {
      "epoch": 0.7982361798236179,
      "grad_norm": 0.0,
      "learning_rate": 2.060638212489474e-06,
      "loss": 0.8523,
      "step": 3711
    },
    {
      "epoch": 0.798451279845128,
      "grad_norm": 0.0,
      "learning_rate": 2.056403963785284e-06,
      "loss": 0.8283,
      "step": 3712
    },
    {
      "epoch": 0.798666379866638,
      "grad_norm": 0.0,
      "learning_rate": 2.0521735712531934e-06,
      "loss": 0.8472,
      "step": 3713
    },
    {
      "epoch": 0.798881479888148,
      "grad_norm": 0.0,
      "learning_rate": 2.0479470369468146e-06,
      "loss": 0.8135,
      "step": 3714
    },
    {
      "epoch": 0.799096579909658,
      "grad_norm": 0.0,
      "learning_rate": 2.043724362917897e-06,
      "loss": 0.8779,
      "step": 3715
    },
    {
      "epoch": 0.799311679931168,
      "grad_norm": 0.0,
      "learning_rate": 2.0395055512163166e-06,
      "loss": 0.8614,
      "step": 3716
    },
    {
      "epoch": 0.799526779952678,
      "grad_norm": 0.0,
      "learning_rate": 2.0352906038900666e-06,
      "loss": 0.8691,
      "step": 3717
    },
    {
      "epoch": 0.799741879974188,
      "grad_norm": 0.0,
      "learning_rate": 2.0310795229852653e-06,
      "loss": 0.8217,
      "step": 3718
    },
    {
      "epoch": 0.799956979995698,
      "grad_norm": 0.0,
      "learning_rate": 2.0268723105461653e-06,
      "loss": 0.7685,
      "step": 3719
    },
    {
      "epoch": 0.800172080017208,
      "grad_norm": 0.0,
      "learning_rate": 2.0226689686151244e-06,
      "loss": 0.8518,
      "step": 3720
    },
    {
      "epoch": 0.800387180038718,
      "grad_norm": 0.0,
      "learning_rate": 2.018469499232639e-06,
      "loss": 0.7874,
      "step": 3721
    },
    {
      "epoch": 0.800602280060228,
      "grad_norm": 0.0,
      "learning_rate": 2.0142739044373093e-06,
      "loss": 0.8481,
      "step": 3722
    },
    {
      "epoch": 0.800817380081738,
      "grad_norm": 0.0,
      "learning_rate": 2.010082186265866e-06,
      "loss": 0.8537,
      "step": 3723
    },
    {
      "epoch": 0.801032480103248,
      "grad_norm": 0.0,
      "learning_rate": 2.0058943467531556e-06,
      "loss": 0.843,
      "step": 3724
    },
    {
      "epoch": 0.801247580124758,
      "grad_norm": 0.0,
      "learning_rate": 2.0017103879321355e-06,
      "loss": 0.8251,
      "step": 3725
    },
    {
      "epoch": 0.801462680146268,
      "grad_norm": 0.0,
      "learning_rate": 1.9975303118338894e-06,
      "loss": 0.8339,
      "step": 3726
    },
    {
      "epoch": 0.801677780167778,
      "grad_norm": 0.0,
      "learning_rate": 1.993354120487605e-06,
      "loss": 0.8119,
      "step": 3727
    },
    {
      "epoch": 0.8018928801892881,
      "grad_norm": 0.0,
      "learning_rate": 1.9891818159205933e-06,
      "loss": 0.8556,
      "step": 3728
    },
    {
      "epoch": 0.802107980210798,
      "grad_norm": 0.0,
      "learning_rate": 1.985013400158278e-06,
      "loss": 0.8342,
      "step": 3729
    },
    {
      "epoch": 0.802323080232308,
      "grad_norm": 0.0,
      "learning_rate": 1.980848875224185e-06,
      "loss": 0.8769,
      "step": 3730
    },
    {
      "epoch": 0.8025381802538181,
      "grad_norm": 0.0,
      "learning_rate": 1.976688243139965e-06,
      "loss": 0.8412,
      "step": 3731
    },
    {
      "epoch": 0.802753280275328,
      "grad_norm": 0.0,
      "learning_rate": 1.9725315059253704e-06,
      "loss": 0.8078,
      "step": 3732
    },
    {
      "epoch": 0.802968380296838,
      "grad_norm": 0.0,
      "learning_rate": 1.9683786655982607e-06,
      "loss": 0.8519,
      "step": 3733
    },
    {
      "epoch": 0.8031834803183481,
      "grad_norm": 0.0,
      "learning_rate": 1.9642297241746146e-06,
      "loss": 0.8288,
      "step": 3734
    },
    {
      "epoch": 0.803398580339858,
      "grad_norm": 0.0,
      "learning_rate": 1.960084683668504e-06,
      "loss": 0.8144,
      "step": 3735
    },
    {
      "epoch": 0.803613680361368,
      "grad_norm": 0.0,
      "learning_rate": 1.9559435460921207e-06,
      "loss": 0.8163,
      "step": 3736
    },
    {
      "epoch": 0.8038287803828781,
      "grad_norm": 0.0,
      "learning_rate": 1.951806313455754e-06,
      "loss": 0.8689,
      "step": 3737
    },
    {
      "epoch": 0.804043880404388,
      "grad_norm": 0.0,
      "learning_rate": 1.947672987767798e-06,
      "loss": 0.7966,
      "step": 3738
    },
    {
      "epoch": 0.804258980425898,
      "grad_norm": 0.0,
      "learning_rate": 1.943543571034754e-06,
      "loss": 0.8034,
      "step": 3739
    },
    {
      "epoch": 0.8044740804474081,
      "grad_norm": 0.0,
      "learning_rate": 1.9394180652612193e-06,
      "loss": 0.8835,
      "step": 3740
    },
    {
      "epoch": 0.804689180468918,
      "grad_norm": 0.0,
      "learning_rate": 1.935296472449898e-06,
      "loss": 0.8549,
      "step": 3741
    },
    {
      "epoch": 0.8049042804904281,
      "grad_norm": 0.0,
      "learning_rate": 1.931178794601596e-06,
      "loss": 0.7959,
      "step": 3742
    },
    {
      "epoch": 0.8051193805119381,
      "grad_norm": 0.0,
      "learning_rate": 1.9270650337152107e-06,
      "loss": 0.8529,
      "step": 3743
    },
    {
      "epoch": 0.805334480533448,
      "grad_norm": 0.0,
      "learning_rate": 1.9229551917877485e-06,
      "loss": 0.789,
      "step": 3744
    },
    {
      "epoch": 0.8055495805549581,
      "grad_norm": 0.0,
      "learning_rate": 1.9188492708143035e-06,
      "loss": 0.7346,
      "step": 3745
    },
    {
      "epoch": 0.8057646805764681,
      "grad_norm": 0.0,
      "learning_rate": 1.9147472727880757e-06,
      "loss": 0.8305,
      "step": 3746
    },
    {
      "epoch": 0.805979780597978,
      "grad_norm": 0.0,
      "learning_rate": 1.910649199700353e-06,
      "loss": 0.7864,
      "step": 3747
    },
    {
      "epoch": 0.8061948806194881,
      "grad_norm": 0.0,
      "learning_rate": 1.9065550535405186e-06,
      "loss": 0.8718,
      "step": 3748
    },
    {
      "epoch": 0.8064099806409981,
      "grad_norm": 0.0,
      "learning_rate": 1.902464836296054e-06,
      "loss": 0.8523,
      "step": 3749
    },
    {
      "epoch": 0.806625080662508,
      "grad_norm": 0.0,
      "learning_rate": 1.8983785499525342e-06,
      "loss": 0.7193,
      "step": 3750
    },
    {
      "epoch": 0.8068401806840181,
      "grad_norm": 0.0,
      "learning_rate": 1.894296196493618e-06,
      "loss": 0.8429,
      "step": 3751
    },
    {
      "epoch": 0.8070552807055281,
      "grad_norm": 0.0,
      "learning_rate": 1.8902177779010645e-06,
      "loss": 0.7944,
      "step": 3752
    },
    {
      "epoch": 0.807270380727038,
      "grad_norm": 0.0,
      "learning_rate": 1.8861432961547133e-06,
      "loss": 0.8179,
      "step": 3753
    },
    {
      "epoch": 0.8074854807485481,
      "grad_norm": 0.0,
      "learning_rate": 1.8820727532325e-06,
      "loss": 0.8205,
      "step": 3754
    },
    {
      "epoch": 0.8077005807700581,
      "grad_norm": 0.0,
      "learning_rate": 1.8780061511104487e-06,
      "loss": 0.8612,
      "step": 3755
    },
    {
      "epoch": 0.8079156807915681,
      "grad_norm": 0.0,
      "learning_rate": 1.873943491762662e-06,
      "loss": 0.7069,
      "step": 3756
    },
    {
      "epoch": 0.8081307808130781,
      "grad_norm": 0.0,
      "learning_rate": 1.8698847771613393e-06,
      "loss": 0.8023,
      "step": 3757
    },
    {
      "epoch": 0.8083458808345881,
      "grad_norm": 0.0,
      "learning_rate": 1.8658300092767546e-06,
      "loss": 0.8138,
      "step": 3758
    },
    {
      "epoch": 0.8085609808560981,
      "grad_norm": 0.0,
      "learning_rate": 1.8617791900772742e-06,
      "loss": 0.8523,
      "step": 3759
    },
    {
      "epoch": 0.8087760808776081,
      "grad_norm": 0.0,
      "learning_rate": 1.8577323215293463e-06,
      "loss": 0.8854,
      "step": 3760
    },
    {
      "epoch": 0.8089911808991181,
      "grad_norm": 0.0,
      "learning_rate": 1.8536894055974974e-06,
      "loss": 0.8802,
      "step": 3761
    },
    {
      "epoch": 0.8092062809206281,
      "grad_norm": 0.0,
      "learning_rate": 1.8496504442443375e-06,
      "loss": 0.7475,
      "step": 3762
    },
    {
      "epoch": 0.8094213809421381,
      "grad_norm": 0.0,
      "learning_rate": 1.8456154394305536e-06,
      "loss": 0.8383,
      "step": 3763
    },
    {
      "epoch": 0.809636480963648,
      "grad_norm": 0.0,
      "learning_rate": 1.8415843931149192e-06,
      "loss": 0.8523,
      "step": 3764
    },
    {
      "epoch": 0.8098515809851581,
      "grad_norm": 0.0,
      "learning_rate": 1.8375573072542819e-06,
      "loss": 0.7621,
      "step": 3765
    },
    {
      "epoch": 0.8100666810066681,
      "grad_norm": 0.0,
      "learning_rate": 1.8335341838035635e-06,
      "loss": 0.8384,
      "step": 3766
    },
    {
      "epoch": 0.810281781028178,
      "grad_norm": 0.0,
      "learning_rate": 1.829515024715769e-06,
      "loss": 0.7647,
      "step": 3767
    },
    {
      "epoch": 0.8104968810496881,
      "grad_norm": 0.0,
      "learning_rate": 1.825499831941977e-06,
      "loss": 0.8093,
      "step": 3768
    },
    {
      "epoch": 0.8107119810711981,
      "grad_norm": 0.0,
      "learning_rate": 1.8214886074313343e-06,
      "loss": 0.8168,
      "step": 3769
    },
    {
      "epoch": 0.8109270810927081,
      "grad_norm": 0.0,
      "learning_rate": 1.817481353131071e-06,
      "loss": 0.7762,
      "step": 3770
    },
    {
      "epoch": 0.8111421811142181,
      "grad_norm": 0.0,
      "learning_rate": 1.8134780709864808e-06,
      "loss": 0.7641,
      "step": 3771
    },
    {
      "epoch": 0.8113572811357281,
      "grad_norm": 0.0,
      "learning_rate": 1.8094787629409361e-06,
      "loss": 0.8354,
      "step": 3772
    },
    {
      "epoch": 0.8115723811572381,
      "grad_norm": 0.0,
      "learning_rate": 1.8054834309358792e-06,
      "loss": 0.8101,
      "step": 3773
    },
    {
      "epoch": 0.8117874811787481,
      "grad_norm": 0.0,
      "learning_rate": 1.801492076910817e-06,
      "loss": 0.8,
      "step": 3774
    },
    {
      "epoch": 0.8120025812002581,
      "grad_norm": 0.0,
      "learning_rate": 1.7975047028033333e-06,
      "loss": 0.839,
      "step": 3775
    },
    {
      "epoch": 0.8122176812217681,
      "grad_norm": 0.0,
      "learning_rate": 1.7935213105490722e-06,
      "loss": 0.7587,
      "step": 3776
    },
    {
      "epoch": 0.8124327812432781,
      "grad_norm": 0.0,
      "learning_rate": 1.7895419020817462e-06,
      "loss": 0.8239,
      "step": 3777
    },
    {
      "epoch": 0.8126478812647882,
      "grad_norm": 0.0,
      "learning_rate": 1.7855664793331406e-06,
      "loss": 0.856,
      "step": 3778
    },
    {
      "epoch": 0.8128629812862981,
      "grad_norm": 0.0,
      "learning_rate": 1.7815950442330965e-06,
      "loss": 0.8396,
      "step": 3779
    },
    {
      "epoch": 0.8130780813078081,
      "grad_norm": 0.0,
      "learning_rate": 1.777627598709526e-06,
      "loss": 0.8679,
      "step": 3780
    },
    {
      "epoch": 0.8132931813293182,
      "grad_norm": 0.0,
      "learning_rate": 1.773664144688405e-06,
      "loss": 0.8356,
      "step": 3781
    },
    {
      "epoch": 0.8135082813508281,
      "grad_norm": 0.0,
      "learning_rate": 1.7697046840937638e-06,
      "loss": 0.7967,
      "step": 3782
    },
    {
      "epoch": 0.8137233813723381,
      "grad_norm": 0.0,
      "learning_rate": 1.7657492188477033e-06,
      "loss": 0.8001,
      "step": 3783
    },
    {
      "epoch": 0.8139384813938482,
      "grad_norm": 0.0,
      "learning_rate": 1.7617977508703777e-06,
      "loss": 0.8136,
      "step": 3784
    },
    {
      "epoch": 0.8141535814153581,
      "grad_norm": 0.0,
      "learning_rate": 1.7578502820800048e-06,
      "loss": 0.774,
      "step": 3785
    },
    {
      "epoch": 0.8143686814368681,
      "grad_norm": 0.0,
      "learning_rate": 1.7539068143928638e-06,
      "loss": 0.9083,
      "step": 3786
    },
    {
      "epoch": 0.8145837814583782,
      "grad_norm": 0.0,
      "learning_rate": 1.7499673497232827e-06,
      "loss": 0.8309,
      "step": 3787
    },
    {
      "epoch": 0.8147988814798881,
      "grad_norm": 0.0,
      "learning_rate": 1.746031889983657e-06,
      "loss": 0.8301,
      "step": 3788
    },
    {
      "epoch": 0.8150139815013981,
      "grad_norm": 0.0,
      "learning_rate": 1.7421004370844263e-06,
      "loss": 0.7986,
      "step": 3789
    },
    {
      "epoch": 0.8152290815229082,
      "grad_norm": 0.0,
      "learning_rate": 1.7381729929340974e-06,
      "loss": 0.833,
      "step": 3790
    },
    {
      "epoch": 0.8154441815444181,
      "grad_norm": 0.0,
      "learning_rate": 1.7342495594392227e-06,
      "loss": 0.83,
      "step": 3791
    },
    {
      "epoch": 0.8156592815659282,
      "grad_norm": 0.0,
      "learning_rate": 1.7303301385044058e-06,
      "loss": 0.7958,
      "step": 3792
    },
    {
      "epoch": 0.8158743815874382,
      "grad_norm": 0.0,
      "learning_rate": 1.726414732032311e-06,
      "loss": 0.7968,
      "step": 3793
    },
    {
      "epoch": 0.8160894816089481,
      "grad_norm": 0.0,
      "learning_rate": 1.7225033419236502e-06,
      "loss": 0.7709,
      "step": 3794
    },
    {
      "epoch": 0.8163045816304582,
      "grad_norm": 0.0,
      "learning_rate": 1.7185959700771816e-06,
      "loss": 0.8252,
      "step": 3795
    },
    {
      "epoch": 0.8165196816519682,
      "grad_norm": 0.0,
      "learning_rate": 1.7146926183897194e-06,
      "loss": 0.8365,
      "step": 3796
    },
    {
      "epoch": 0.8167347816734781,
      "grad_norm": 0.0,
      "learning_rate": 1.7107932887561196e-06,
      "loss": 0.7973,
      "step": 3797
    },
    {
      "epoch": 0.8169498816949882,
      "grad_norm": 0.0,
      "learning_rate": 1.7068979830692889e-06,
      "loss": 0.7536,
      "step": 3798
    },
    {
      "epoch": 0.8171649817164982,
      "grad_norm": 0.0,
      "learning_rate": 1.703006703220186e-06,
      "loss": 0.8416,
      "step": 3799
    },
    {
      "epoch": 0.8173800817380081,
      "grad_norm": 0.0,
      "learning_rate": 1.6991194510978015e-06,
      "loss": 0.8067,
      "step": 3800
    },
    {
      "epoch": 0.8175951817595182,
      "grad_norm": 0.0,
      "learning_rate": 1.6952362285891866e-06,
      "loss": 0.7911,
      "step": 3801
    },
    {
      "epoch": 0.8178102817810282,
      "grad_norm": 0.0,
      "learning_rate": 1.691357037579423e-06,
      "loss": 0.8292,
      "step": 3802
    },
    {
      "epoch": 0.8180253818025381,
      "grad_norm": 0.0,
      "learning_rate": 1.6874818799516468e-06,
      "loss": 0.7846,
      "step": 3803
    },
    {
      "epoch": 0.8182404818240482,
      "grad_norm": 0.0,
      "learning_rate": 1.6836107575870254e-06,
      "loss": 0.8502,
      "step": 3804
    },
    {
      "epoch": 0.8184555818455582,
      "grad_norm": 0.0,
      "learning_rate": 1.6797436723647775e-06,
      "loss": 0.8255,
      "step": 3805
    },
    {
      "epoch": 0.8186706818670682,
      "grad_norm": 0.0,
      "learning_rate": 1.675880626162153e-06,
      "loss": 0.8836,
      "step": 3806
    },
    {
      "epoch": 0.8188857818885782,
      "grad_norm": 0.0,
      "learning_rate": 1.6720216208544448e-06,
      "loss": 0.755,
      "step": 3807
    },
    {
      "epoch": 0.8191008819100882,
      "grad_norm": 0.0,
      "learning_rate": 1.6681666583149846e-06,
      "loss": 0.8141,
      "step": 3808
    },
    {
      "epoch": 0.8193159819315982,
      "grad_norm": 0.0,
      "learning_rate": 1.6643157404151468e-06,
      "loss": 0.8652,
      "step": 3809
    },
    {
      "epoch": 0.8195310819531082,
      "grad_norm": 0.0,
      "learning_rate": 1.6604688690243288e-06,
      "loss": 0.8787,
      "step": 3810
    },
    {
      "epoch": 0.8197461819746182,
      "grad_norm": 0.0,
      "learning_rate": 1.6566260460099749e-06,
      "loss": 0.8508,
      "step": 3811
    },
    {
      "epoch": 0.8199612819961282,
      "grad_norm": 0.0,
      "learning_rate": 1.652787273237565e-06,
      "loss": 0.7976,
      "step": 3812
    },
    {
      "epoch": 0.8201763820176382,
      "grad_norm": 0.0,
      "learning_rate": 1.648952552570603e-06,
      "loss": 0.7554,
      "step": 3813
    },
    {
      "epoch": 0.8203914820391482,
      "grad_norm": 0.0,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 0.7877,
      "step": 3814
    },
    {
      "epoch": 0.8206065820606582,
      "grad_norm": 0.0,
      "learning_rate": 1.6412952749972355e-06,
      "loss": 0.8204,
      "step": 3815
    },
    {
      "epoch": 0.8208216820821682,
      "grad_norm": 0.0,
      "learning_rate": 1.6374727218080078e-06,
      "loss": 0.8325,
      "step": 3816
    },
    {
      "epoch": 0.8210367821036783,
      "grad_norm": 0.0,
      "learning_rate": 1.633654228158592e-06,
      "loss": 0.7842,
      "step": 3817
    },
    {
      "epoch": 0.8212518821251882,
      "grad_norm": 0.0,
      "learning_rate": 1.6298397959026513e-06,
      "loss": 0.7846,
      "step": 3818
    },
    {
      "epoch": 0.8214669821466982,
      "grad_norm": 0.0,
      "learning_rate": 1.6260294268918763e-06,
      "loss": 0.7951,
      "step": 3819
    },
    {
      "epoch": 0.8216820821682083,
      "grad_norm": 0.0,
      "learning_rate": 1.6222231229759945e-06,
      "loss": 0.8135,
      "step": 3820
    },
    {
      "epoch": 0.8218971821897182,
      "grad_norm": 0.0,
      "learning_rate": 1.6184208860027473e-06,
      "loss": 0.8149,
      "step": 3821
    },
    {
      "epoch": 0.8221122822112282,
      "grad_norm": 0.0,
      "learning_rate": 1.6146227178179142e-06,
      "loss": 0.8447,
      "step": 3822
    },
    {
      "epoch": 0.8223273822327383,
      "grad_norm": 0.0,
      "learning_rate": 1.610828620265289e-06,
      "loss": 0.8233,
      "step": 3823
    },
    {
      "epoch": 0.8225424822542482,
      "grad_norm": 0.0,
      "learning_rate": 1.6070385951866951e-06,
      "loss": 0.796,
      "step": 3824
    },
    {
      "epoch": 0.8227575822757582,
      "grad_norm": 0.0,
      "learning_rate": 1.603252644421982e-06,
      "loss": 0.836,
      "step": 3825
    },
    {
      "epoch": 0.8229726822972683,
      "grad_norm": 0.0,
      "learning_rate": 1.5994707698090118e-06,
      "loss": 0.7795,
      "step": 3826
    },
    {
      "epoch": 0.8231877823187782,
      "grad_norm": 0.0,
      "learning_rate": 1.595692973183679e-06,
      "loss": 0.8969,
      "step": 3827
    },
    {
      "epoch": 0.8234028823402882,
      "grad_norm": 0.0,
      "learning_rate": 1.5919192563798857e-06,
      "loss": 0.8371,
      "step": 3828
    },
    {
      "epoch": 0.8236179823617983,
      "grad_norm": 0.0,
      "learning_rate": 1.5881496212295654e-06,
      "loss": 0.858,
      "step": 3829
    },
    {
      "epoch": 0.8238330823833082,
      "grad_norm": 0.0,
      "learning_rate": 1.5843840695626666e-06,
      "loss": 0.8495,
      "step": 3830
    },
    {
      "epoch": 0.8240481824048183,
      "grad_norm": 0.0,
      "learning_rate": 1.5806226032071503e-06,
      "loss": 0.8382,
      "step": 3831
    },
    {
      "epoch": 0.8242632824263283,
      "grad_norm": 0.0,
      "learning_rate": 1.5768652239890014e-06,
      "loss": 0.8301,
      "step": 3832
    },
    {
      "epoch": 0.8244783824478382,
      "grad_norm": 0.0,
      "learning_rate": 1.5731119337322176e-06,
      "loss": 0.7716,
      "step": 3833
    },
    {
      "epoch": 0.8246934824693483,
      "grad_norm": 0.0,
      "learning_rate": 1.569362734258807e-06,
      "loss": 0.7937,
      "step": 3834
    },
    {
      "epoch": 0.8249085824908583,
      "grad_norm": 0.0,
      "learning_rate": 1.5656176273888025e-06,
      "loss": 0.8507,
      "step": 3835
    },
    {
      "epoch": 0.8251236825123682,
      "grad_norm": 0.0,
      "learning_rate": 1.5618766149402387e-06,
      "loss": 0.868,
      "step": 3836
    },
    {
      "epoch": 0.8253387825338783,
      "grad_norm": 0.0,
      "learning_rate": 1.5581396987291719e-06,
      "loss": 0.7863,
      "step": 3837
    },
    {
      "epoch": 0.8255538825553883,
      "grad_norm": 0.0,
      "learning_rate": 1.5544068805696666e-06,
      "loss": 0.8192,
      "step": 3838
    },
    {
      "epoch": 0.8257689825768982,
      "grad_norm": 0.0,
      "learning_rate": 1.5506781622737943e-06,
      "loss": 0.771,
      "step": 3839
    },
    {
      "epoch": 0.8259840825984083,
      "grad_norm": 0.0,
      "learning_rate": 1.5469535456516437e-06,
      "loss": 0.7906,
      "step": 3840
    },
    {
      "epoch": 0.8261991826199183,
      "grad_norm": 0.0,
      "learning_rate": 1.5432330325113043e-06,
      "loss": 0.8307,
      "step": 3841
    },
    {
      "epoch": 0.8264142826414282,
      "grad_norm": 0.0,
      "learning_rate": 1.5395166246588777e-06,
      "loss": 0.8545,
      "step": 3842
    },
    {
      "epoch": 0.8266293826629383,
      "grad_norm": 0.0,
      "learning_rate": 1.5358043238984755e-06,
      "loss": 0.7877,
      "step": 3843
    },
    {
      "epoch": 0.8268444826844483,
      "grad_norm": 0.0,
      "learning_rate": 1.532096132032208e-06,
      "loss": 0.8378,
      "step": 3844
    },
    {
      "epoch": 0.8270595827059583,
      "grad_norm": 0.0,
      "learning_rate": 1.5283920508602e-06,
      "loss": 0.8509,
      "step": 3845
    },
    {
      "epoch": 0.8272746827274683,
      "grad_norm": 0.0,
      "learning_rate": 1.524692082180571e-06,
      "loss": 0.8141,
      "step": 3846
    },
    {
      "epoch": 0.8274897827489782,
      "grad_norm": 0.0,
      "learning_rate": 1.520996227789454e-06,
      "loss": 0.8449,
      "step": 3847
    },
    {
      "epoch": 0.8277048827704883,
      "grad_norm": 0.0,
      "learning_rate": 1.5173044894809764e-06,
      "loss": 0.8414,
      "step": 3848
    },
    {
      "epoch": 0.8279199827919983,
      "grad_norm": 0.0,
      "learning_rate": 1.5136168690472696e-06,
      "loss": 0.7887,
      "step": 3849
    },
    {
      "epoch": 0.8281350828135082,
      "grad_norm": 0.0,
      "learning_rate": 1.50993336827847e-06,
      "loss": 0.8195,
      "step": 3850
    },
    {
      "epoch": 0.8283501828350183,
      "grad_norm": 0.0,
      "learning_rate": 1.5062539889627138e-06,
      "loss": 0.7593,
      "step": 3851
    },
    {
      "epoch": 0.8285652828565283,
      "grad_norm": 0.0,
      "learning_rate": 1.5025787328861275e-06,
      "loss": 0.8195,
      "step": 3852
    },
    {
      "epoch": 0.8287803828780382,
      "grad_norm": 0.0,
      "learning_rate": 1.4989076018328497e-06,
      "loss": 0.7996,
      "step": 3853
    },
    {
      "epoch": 0.8289954828995483,
      "grad_norm": 0.0,
      "learning_rate": 1.4952405975850025e-06,
      "loss": 0.8357,
      "step": 3854
    },
    {
      "epoch": 0.8292105829210583,
      "grad_norm": 0.0,
      "learning_rate": 1.4915777219227156e-06,
      "loss": 0.8535,
      "step": 3855
    },
    {
      "epoch": 0.8294256829425682,
      "grad_norm": 0.0,
      "learning_rate": 1.4879189766241131e-06,
      "loss": 0.7645,
      "step": 3856
    },
    {
      "epoch": 0.8296407829640783,
      "grad_norm": 0.0,
      "learning_rate": 1.4842643634653054e-06,
      "loss": 0.7997,
      "step": 3857
    },
    {
      "epoch": 0.8298558829855883,
      "grad_norm": 0.0,
      "learning_rate": 1.4806138842204088e-06,
      "loss": 0.7804,
      "step": 3858
    },
    {
      "epoch": 0.8300709830070983,
      "grad_norm": 0.0,
      "learning_rate": 1.4769675406615213e-06,
      "loss": 0.7987,
      "step": 3859
    },
    {
      "epoch": 0.8302860830286083,
      "grad_norm": 0.0,
      "learning_rate": 1.4733253345587429e-06,
      "loss": 0.8513,
      "step": 3860
    },
    {
      "epoch": 0.8305011830501183,
      "grad_norm": 0.0,
      "learning_rate": 1.4696872676801621e-06,
      "loss": 0.8596,
      "step": 3861
    },
    {
      "epoch": 0.8307162830716283,
      "grad_norm": 0.0,
      "learning_rate": 1.4660533417918554e-06,
      "loss": 0.8581,
      "step": 3862
    },
    {
      "epoch": 0.8309313830931383,
      "grad_norm": 0.0,
      "learning_rate": 1.4624235586578895e-06,
      "loss": 0.7849,
      "step": 3863
    },
    {
      "epoch": 0.8311464831146483,
      "grad_norm": 0.0,
      "learning_rate": 1.4587979200403213e-06,
      "loss": 0.8473,
      "step": 3864
    },
    {
      "epoch": 0.8313615831361583,
      "grad_norm": 0.0,
      "learning_rate": 1.4551764276991965e-06,
      "loss": 0.8544,
      "step": 3865
    },
    {
      "epoch": 0.8315766831576683,
      "grad_norm": 0.0,
      "learning_rate": 1.4515590833925508e-06,
      "loss": 0.7443,
      "step": 3866
    },
    {
      "epoch": 0.8317917831791783,
      "grad_norm": 0.0,
      "learning_rate": 1.4479458888763964e-06,
      "loss": 0.8218,
      "step": 3867
    },
    {
      "epoch": 0.8320068832006883,
      "grad_norm": 0.0,
      "learning_rate": 1.4443368459047413e-06,
      "loss": 0.7359,
      "step": 3868
    },
    {
      "epoch": 0.8322219832221983,
      "grad_norm": 0.0,
      "learning_rate": 1.4407319562295763e-06,
      "loss": 0.7972,
      "step": 3869
    },
    {
      "epoch": 0.8324370832437084,
      "grad_norm": 0.0,
      "learning_rate": 1.437131221600868e-06,
      "loss": 0.7999,
      "step": 3870
    },
    {
      "epoch": 0.8326521832652183,
      "grad_norm": 0.0,
      "learning_rate": 1.4335346437665787e-06,
      "loss": 0.8433,
      "step": 3871
    },
    {
      "epoch": 0.8328672832867283,
      "grad_norm": 0.0,
      "learning_rate": 1.4299422244726403e-06,
      "loss": 0.8353,
      "step": 3872
    },
    {
      "epoch": 0.8330823833082384,
      "grad_norm": 0.0,
      "learning_rate": 1.4263539654629732e-06,
      "loss": 0.8036,
      "step": 3873
    },
    {
      "epoch": 0.8332974833297483,
      "grad_norm": 0.0,
      "learning_rate": 1.4227698684794799e-06,
      "loss": 0.8485,
      "step": 3874
    },
    {
      "epoch": 0.8335125833512583,
      "grad_norm": 0.0,
      "learning_rate": 1.4191899352620342e-06,
      "loss": 0.7747,
      "step": 3875
    },
    {
      "epoch": 0.8337276833727684,
      "grad_norm": 0.0,
      "learning_rate": 1.415614167548498e-06,
      "loss": 0.8214,
      "step": 3876
    },
    {
      "epoch": 0.8339427833942783,
      "grad_norm": 0.0,
      "learning_rate": 1.4120425670747061e-06,
      "loss": 0.7688,
      "step": 3877
    },
    {
      "epoch": 0.8341578834157883,
      "grad_norm": 0.0,
      "learning_rate": 1.4084751355744675e-06,
      "loss": 0.7696,
      "step": 3878
    },
    {
      "epoch": 0.8343729834372984,
      "grad_norm": 0.0,
      "learning_rate": 1.4049118747795754e-06,
      "loss": 0.7973,
      "step": 3879
    },
    {
      "epoch": 0.8345880834588083,
      "grad_norm": 0.0,
      "learning_rate": 1.4013527864197883e-06,
      "loss": 0.8324,
      "step": 3880
    },
    {
      "epoch": 0.8348031834803183,
      "grad_norm": 0.0,
      "learning_rate": 1.3977978722228502e-06,
      "loss": 0.8056,
      "step": 3881
    },
    {
      "epoch": 0.8350182835018284,
      "grad_norm": 0.0,
      "learning_rate": 1.3942471339144726e-06,
      "loss": 0.8496,
      "step": 3882
    },
    {
      "epoch": 0.8352333835233383,
      "grad_norm": 0.0,
      "learning_rate": 1.3907005732183364e-06,
      "loss": 0.8699,
      "step": 3883
    },
    {
      "epoch": 0.8354484835448484,
      "grad_norm": 0.0,
      "learning_rate": 1.387158191856105e-06,
      "loss": 0.838,
      "step": 3884
    },
    {
      "epoch": 0.8356635835663584,
      "grad_norm": 0.0,
      "learning_rate": 1.3836199915474003e-06,
      "loss": 0.8418,
      "step": 3885
    },
    {
      "epoch": 0.8358786835878683,
      "grad_norm": 0.0,
      "learning_rate": 1.3800859740098238e-06,
      "loss": 0.8623,
      "step": 3886
    },
    {
      "epoch": 0.8360937836093784,
      "grad_norm": 0.0,
      "learning_rate": 1.3765561409589468e-06,
      "loss": 0.7972,
      "step": 3887
    },
    {
      "epoch": 0.8363088836308884,
      "grad_norm": 0.0,
      "learning_rate": 1.3730304941082994e-06,
      "loss": 0.8416,
      "step": 3888
    },
    {
      "epoch": 0.8365239836523983,
      "grad_norm": 0.0,
      "learning_rate": 1.3695090351693929e-06,
      "loss": 0.874,
      "step": 3889
    },
    {
      "epoch": 0.8367390836739084,
      "grad_norm": 0.0,
      "learning_rate": 1.365991765851693e-06,
      "loss": 0.7933,
      "step": 3890
    },
    {
      "epoch": 0.8369541836954184,
      "grad_norm": 0.0,
      "learning_rate": 1.3624786878626416e-06,
      "loss": 0.8634,
      "step": 3891
    },
    {
      "epoch": 0.8371692837169283,
      "grad_norm": 0.0,
      "learning_rate": 1.3589698029076415e-06,
      "loss": 0.7776,
      "step": 3892
    },
    {
      "epoch": 0.8373843837384384,
      "grad_norm": 0.0,
      "learning_rate": 1.3554651126900564e-06,
      "loss": 0.8601,
      "step": 3893
    },
    {
      "epoch": 0.8375994837599484,
      "grad_norm": 0.0,
      "learning_rate": 1.3519646189112202e-06,
      "loss": 0.7494,
      "step": 3894
    },
    {
      "epoch": 0.8378145837814583,
      "grad_norm": 0.0,
      "learning_rate": 1.34846832327043e-06,
      "loss": 0.7402,
      "step": 3895
    },
    {
      "epoch": 0.8380296838029684,
      "grad_norm": 0.0,
      "learning_rate": 1.3449762274649358e-06,
      "loss": 0.7525,
      "step": 3896
    },
    {
      "epoch": 0.8382447838244784,
      "grad_norm": 0.0,
      "learning_rate": 1.3414883331899608e-06,
      "loss": 0.8283,
      "step": 3897
    },
    {
      "epoch": 0.8384598838459884,
      "grad_norm": 0.0,
      "learning_rate": 1.3380046421386795e-06,
      "loss": 0.7787,
      "step": 3898
    },
    {
      "epoch": 0.8386749838674984,
      "grad_norm": 0.0,
      "learning_rate": 1.3345251560022288e-06,
      "loss": 0.8363,
      "step": 3899
    },
    {
      "epoch": 0.8388900838890084,
      "grad_norm": 0.0,
      "learning_rate": 1.331049876469709e-06,
      "loss": 0.7725,
      "step": 3900
    },
    {
      "epoch": 0.8391051839105184,
      "grad_norm": 0.0,
      "learning_rate": 1.3275788052281692e-06,
      "loss": 0.7987,
      "step": 3901
    },
    {
      "epoch": 0.8393202839320284,
      "grad_norm": 0.0,
      "learning_rate": 1.324111943962626e-06,
      "loss": 0.7787,
      "step": 3902
    },
    {
      "epoch": 0.8395353839535384,
      "grad_norm": 0.0,
      "learning_rate": 1.3206492943560401e-06,
      "loss": 0.7812,
      "step": 3903
    },
    {
      "epoch": 0.8397504839750484,
      "grad_norm": 0.0,
      "learning_rate": 1.3171908580893399e-06,
      "loss": 0.817,
      "step": 3904
    },
    {
      "epoch": 0.8399655839965584,
      "grad_norm": 0.0,
      "learning_rate": 1.3137366368413995e-06,
      "loss": 0.8107,
      "step": 3905
    },
    {
      "epoch": 0.8401806840180684,
      "grad_norm": 0.0,
      "learning_rate": 1.310286632289055e-06,
      "loss": 0.7542,
      "step": 3906
    },
    {
      "epoch": 0.8403957840395784,
      "grad_norm": 0.0,
      "learning_rate": 1.3068408461070868e-06,
      "loss": 0.8749,
      "step": 3907
    },
    {
      "epoch": 0.8406108840610884,
      "grad_norm": 0.0,
      "learning_rate": 1.3033992799682304e-06,
      "loss": 0.8151,
      "step": 3908
    },
    {
      "epoch": 0.8408259840825985,
      "grad_norm": 0.0,
      "learning_rate": 1.299961935543176e-06,
      "loss": 0.8267,
      "step": 3909
    },
    {
      "epoch": 0.8410410841041084,
      "grad_norm": 0.0,
      "learning_rate": 1.296528814500565e-06,
      "loss": 0.8204,
      "step": 3910
    },
    {
      "epoch": 0.8412561841256184,
      "grad_norm": 0.0,
      "learning_rate": 1.2930999185069804e-06,
      "loss": 0.7437,
      "step": 3911
    },
    {
      "epoch": 0.8414712841471285,
      "grad_norm": 0.0,
      "learning_rate": 1.2896752492269627e-06,
      "loss": 0.7713,
      "step": 3912
    },
    {
      "epoch": 0.8416863841686384,
      "grad_norm": 0.0,
      "learning_rate": 1.2862548083229998e-06,
      "loss": 0.7972,
      "step": 3913
    },
    {
      "epoch": 0.8419014841901484,
      "grad_norm": 0.0,
      "learning_rate": 1.2828385974555203e-06,
      "loss": 0.798,
      "step": 3914
    },
    {
      "epoch": 0.8421165842116585,
      "grad_norm": 0.0,
      "learning_rate": 1.2794266182829073e-06,
      "loss": 0.8511,
      "step": 3915
    },
    {
      "epoch": 0.8423316842331684,
      "grad_norm": 0.0,
      "learning_rate": 1.276018872461482e-06,
      "loss": 0.8369,
      "step": 3916
    },
    {
      "epoch": 0.8425467842546784,
      "grad_norm": 0.0,
      "learning_rate": 1.2726153616455173e-06,
      "loss": 0.7553,
      "step": 3917
    },
    {
      "epoch": 0.8427618842761885,
      "grad_norm": 0.0,
      "learning_rate": 1.2692160874872295e-06,
      "loss": 0.8025,
      "step": 3918
    },
    {
      "epoch": 0.8429769842976984,
      "grad_norm": 0.0,
      "learning_rate": 1.265821051636773e-06,
      "loss": 0.767,
      "step": 3919
    },
    {
      "epoch": 0.8431920843192084,
      "grad_norm": 0.0,
      "learning_rate": 1.2624302557422475e-06,
      "loss": 0.801,
      "step": 3920
    },
    {
      "epoch": 0.8434071843407185,
      "grad_norm": 0.0,
      "learning_rate": 1.2590437014496993e-06,
      "loss": 0.8205,
      "step": 3921
    },
    {
      "epoch": 0.8436222843622284,
      "grad_norm": 0.0,
      "learning_rate": 1.2556613904031055e-06,
      "loss": 0.8021,
      "step": 3922
    },
    {
      "epoch": 0.8438373843837385,
      "grad_norm": 0.0,
      "learning_rate": 1.2522833242443954e-06,
      "loss": 0.8669,
      "step": 3923
    },
    {
      "epoch": 0.8440524844052485,
      "grad_norm": 0.0,
      "learning_rate": 1.2489095046134248e-06,
      "loss": 0.821,
      "step": 3924
    },
    {
      "epoch": 0.8442675844267584,
      "grad_norm": 0.0,
      "learning_rate": 1.245539933147999e-06,
      "loss": 0.7905,
      "step": 3925
    },
    {
      "epoch": 0.8444826844482685,
      "grad_norm": 0.0,
      "learning_rate": 1.2421746114838573e-06,
      "loss": 0.7471,
      "step": 3926
    },
    {
      "epoch": 0.8446977844697785,
      "grad_norm": 0.0,
      "learning_rate": 1.238813541254672e-06,
      "loss": 0.7795,
      "step": 3927
    },
    {
      "epoch": 0.8449128844912884,
      "grad_norm": 0.0,
      "learning_rate": 1.235456724092059e-06,
      "loss": 0.8178,
      "step": 3928
    },
    {
      "epoch": 0.8451279845127985,
      "grad_norm": 0.0,
      "learning_rate": 1.2321041616255613e-06,
      "loss": 0.7856,
      "step": 3929
    },
    {
      "epoch": 0.8453430845343085,
      "grad_norm": 0.0,
      "learning_rate": 1.228755855482663e-06,
      "loss": 0.8766,
      "step": 3930
    },
    {
      "epoch": 0.8455581845558184,
      "grad_norm": 0.0,
      "learning_rate": 1.22541180728878e-06,
      "loss": 0.7484,
      "step": 3931
    },
    {
      "epoch": 0.8457732845773285,
      "grad_norm": 0.0,
      "learning_rate": 1.2220720186672585e-06,
      "loss": 0.7785,
      "step": 3932
    },
    {
      "epoch": 0.8459883845988384,
      "grad_norm": 0.0,
      "learning_rate": 1.218736491239384e-06,
      "loss": 0.8349,
      "step": 3933
    },
    {
      "epoch": 0.8462034846203484,
      "grad_norm": 0.0,
      "learning_rate": 1.2154052266243655e-06,
      "loss": 0.8415,
      "step": 3934
    },
    {
      "epoch": 0.8464185846418585,
      "grad_norm": 0.0,
      "learning_rate": 1.2120782264393427e-06,
      "loss": 0.7593,
      "step": 3935
    },
    {
      "epoch": 0.8466336846633684,
      "grad_norm": 0.0,
      "learning_rate": 1.208755492299395e-06,
      "loss": 0.7303,
      "step": 3936
    },
    {
      "epoch": 0.8468487846848785,
      "grad_norm": 0.0,
      "learning_rate": 1.2054370258175186e-06,
      "loss": 0.7787,
      "step": 3937
    },
    {
      "epoch": 0.8470638847063885,
      "grad_norm": 0.0,
      "learning_rate": 1.2021228286046461e-06,
      "loss": 0.798,
      "step": 3938
    },
    {
      "epoch": 0.8472789847278984,
      "grad_norm": 0.0,
      "learning_rate": 1.198812902269636e-06,
      "loss": 0.8734,
      "step": 3939
    },
    {
      "epoch": 0.8474940847494085,
      "grad_norm": 0.0,
      "learning_rate": 1.1955072484192697e-06,
      "loss": 0.7668,
      "step": 3940
    },
    {
      "epoch": 0.8477091847709185,
      "grad_norm": 0.0,
      "learning_rate": 1.1922058686582616e-06,
      "loss": 0.8076,
      "step": 3941
    },
    {
      "epoch": 0.8479242847924284,
      "grad_norm": 0.0,
      "learning_rate": 1.1889087645892427e-06,
      "loss": 0.7927,
      "step": 3942
    },
    {
      "epoch": 0.8481393848139385,
      "grad_norm": 0.0,
      "learning_rate": 1.1856159378127752e-06,
      "loss": 0.8189,
      "step": 3943
    },
    {
      "epoch": 0.8483544848354485,
      "grad_norm": 0.0,
      "learning_rate": 1.1823273899273436e-06,
      "loss": 0.8118,
      "step": 3944
    },
    {
      "epoch": 0.8485695848569584,
      "grad_norm": 0.0,
      "learning_rate": 1.179043122529352e-06,
      "loss": 0.7799,
      "step": 3945
    },
    {
      "epoch": 0.8487846848784685,
      "grad_norm": 0.0,
      "learning_rate": 1.1757631372131307e-06,
      "loss": 0.8211,
      "step": 3946
    },
    {
      "epoch": 0.8489997848999785,
      "grad_norm": 0.0,
      "learning_rate": 1.1724874355709259e-06,
      "loss": 0.7756,
      "step": 3947
    },
    {
      "epoch": 0.8492148849214884,
      "grad_norm": 0.0,
      "learning_rate": 1.1692160191929125e-06,
      "loss": 0.8393,
      "step": 3948
    },
    {
      "epoch": 0.8494299849429985,
      "grad_norm": 0.0,
      "learning_rate": 1.1659488896671777e-06,
      "loss": 0.7857,
      "step": 3949
    },
    {
      "epoch": 0.8496450849645085,
      "grad_norm": 0.0,
      "learning_rate": 1.1626860485797298e-06,
      "loss": 0.8501,
      "step": 3950
    },
    {
      "epoch": 0.8498601849860185,
      "grad_norm": 0.0,
      "learning_rate": 1.1594274975144948e-06,
      "loss": 0.7444,
      "step": 3951
    },
    {
      "epoch": 0.8500752850075285,
      "grad_norm": 0.0,
      "learning_rate": 1.156173238053322e-06,
      "loss": 0.7623,
      "step": 3952
    },
    {
      "epoch": 0.8502903850290385,
      "grad_norm": 0.0,
      "learning_rate": 1.1529232717759676e-06,
      "loss": 0.8747,
      "step": 3953
    },
    {
      "epoch": 0.8505054850505485,
      "grad_norm": 0.0,
      "learning_rate": 1.1496776002601128e-06,
      "loss": 0.8333,
      "step": 3954
    },
    {
      "epoch": 0.8507205850720585,
      "grad_norm": 0.0,
      "learning_rate": 1.1464362250813454e-06,
      "loss": 0.8083,
      "step": 3955
    },
    {
      "epoch": 0.8509356850935685,
      "grad_norm": 0.0,
      "learning_rate": 1.1431991478131754e-06,
      "loss": 0.8023,
      "step": 3956
    },
    {
      "epoch": 0.8511507851150785,
      "grad_norm": 0.0,
      "learning_rate": 1.1399663700270225e-06,
      "loss": 0.8206,
      "step": 3957
    },
    {
      "epoch": 0.8513658851365885,
      "grad_norm": 0.0,
      "learning_rate": 1.1367378932922179e-06,
      "loss": 0.8124,
      "step": 3958
    },
    {
      "epoch": 0.8515809851580985,
      "grad_norm": 0.0,
      "learning_rate": 1.1335137191760094e-06,
      "loss": 0.9286,
      "step": 3959
    },
    {
      "epoch": 0.8517960851796085,
      "grad_norm": 0.0,
      "learning_rate": 1.13029384924355e-06,
      "loss": 0.8024,
      "step": 3960
    },
    {
      "epoch": 0.8520111852011185,
      "grad_norm": 0.0,
      "learning_rate": 1.1270782850579077e-06,
      "loss": 0.8163,
      "step": 3961
    },
    {
      "epoch": 0.8522262852226286,
      "grad_norm": 0.0,
      "learning_rate": 1.1238670281800635e-06,
      "loss": 0.7791,
      "step": 3962
    },
    {
      "epoch": 0.8524413852441385,
      "grad_norm": 0.0,
      "learning_rate": 1.1206600801688983e-06,
      "loss": 0.8231,
      "step": 3963
    },
    {
      "epoch": 0.8526564852656485,
      "grad_norm": 0.0,
      "learning_rate": 1.1174574425812068e-06,
      "loss": 0.8355,
      "step": 3964
    },
    {
      "epoch": 0.8528715852871586,
      "grad_norm": 0.0,
      "learning_rate": 1.1142591169716931e-06,
      "loss": 0.8274,
      "step": 3965
    },
    {
      "epoch": 0.8530866853086685,
      "grad_norm": 0.0,
      "learning_rate": 1.1110651048929621e-06,
      "loss": 0.8579,
      "step": 3966
    },
    {
      "epoch": 0.8533017853301785,
      "grad_norm": 0.0,
      "learning_rate": 1.1078754078955322e-06,
      "loss": 0.8434,
      "step": 3967
    },
    {
      "epoch": 0.8535168853516886,
      "grad_norm": 0.0,
      "learning_rate": 1.1046900275278183e-06,
      "loss": 0.8402,
      "step": 3968
    },
    {
      "epoch": 0.8537319853731985,
      "grad_norm": 0.0,
      "learning_rate": 1.1015089653361488e-06,
      "loss": 0.8539,
      "step": 3969
    },
    {
      "epoch": 0.8539470853947085,
      "grad_norm": 0.0,
      "learning_rate": 1.0983322228647509e-06,
      "loss": 0.7604,
      "step": 3970
    },
    {
      "epoch": 0.8541621854162186,
      "grad_norm": 0.0,
      "learning_rate": 1.0951598016557542e-06,
      "loss": 0.7819,
      "step": 3971
    },
    {
      "epoch": 0.8543772854377285,
      "grad_norm": 0.0,
      "learning_rate": 1.0919917032491935e-06,
      "loss": 0.7882,
      "step": 3972
    },
    {
      "epoch": 0.8545923854592385,
      "grad_norm": 0.0,
      "learning_rate": 1.0888279291830017e-06,
      "loss": 0.7842,
      "step": 3973
    },
    {
      "epoch": 0.8548074854807486,
      "grad_norm": 0.0,
      "learning_rate": 1.085668480993015e-06,
      "loss": 0.7693,
      "step": 3974
    },
    {
      "epoch": 0.8550225855022585,
      "grad_norm": 0.0,
      "learning_rate": 1.0825133602129722e-06,
      "loss": 0.8036,
      "step": 3975
    },
    {
      "epoch": 0.8552376855237686,
      "grad_norm": 0.0,
      "learning_rate": 1.0793625683745024e-06,
      "loss": 0.8822,
      "step": 3976
    },
    {
      "epoch": 0.8554527855452786,
      "grad_norm": 0.0,
      "learning_rate": 1.0762161070071441e-06,
      "loss": 0.8464,
      "step": 3977
    },
    {
      "epoch": 0.8556678855667885,
      "grad_norm": 0.0,
      "learning_rate": 1.073073977638327e-06,
      "loss": 0.8268,
      "step": 3978
    },
    {
      "epoch": 0.8558829855882986,
      "grad_norm": 0.0,
      "learning_rate": 1.0699361817933761e-06,
      "loss": 0.7827,
      "step": 3979
    },
    {
      "epoch": 0.8560980856098086,
      "grad_norm": 0.0,
      "learning_rate": 1.0668027209955212e-06,
      "loss": 0.7975,
      "step": 3980
    },
    {
      "epoch": 0.8563131856313185,
      "grad_norm": 0.0,
      "learning_rate": 1.0636735967658785e-06,
      "loss": 0.8543,
      "step": 3981
    },
    {
      "epoch": 0.8565282856528286,
      "grad_norm": 0.0,
      "learning_rate": 1.060548810623464e-06,
      "loss": 0.793,
      "step": 3982
    },
    {
      "epoch": 0.8567433856743386,
      "grad_norm": 0.0,
      "learning_rate": 1.057428364085189e-06,
      "loss": 0.8201,
      "step": 3983
    },
    {
      "epoch": 0.8569584856958485,
      "grad_norm": 0.0,
      "learning_rate": 1.0543122586658528e-06,
      "loss": 0.844,
      "step": 3984
    },
    {
      "epoch": 0.8571735857173586,
      "grad_norm": 0.0,
      "learning_rate": 1.051200495878153e-06,
      "loss": 0.764,
      "step": 3985
    },
    {
      "epoch": 0.8573886857388686,
      "grad_norm": 0.0,
      "learning_rate": 1.0480930772326735e-06,
      "loss": 0.8134,
      "step": 3986
    },
    {
      "epoch": 0.8576037857603785,
      "grad_norm": 0.0,
      "learning_rate": 1.0449900042378957e-06,
      "loss": 0.8264,
      "step": 3987
    },
    {
      "epoch": 0.8578188857818886,
      "grad_norm": 0.0,
      "learning_rate": 1.0418912784001878e-06,
      "loss": 0.8634,
      "step": 3988
    },
    {
      "epoch": 0.8580339858033986,
      "grad_norm": 0.0,
      "learning_rate": 1.0387969012238064e-06,
      "loss": 0.8592,
      "step": 3989
    },
    {
      "epoch": 0.8582490858249086,
      "grad_norm": 0.0,
      "learning_rate": 1.0357068742109e-06,
      "loss": 0.8129,
      "step": 3990
    },
    {
      "epoch": 0.8584641858464186,
      "grad_norm": 0.0,
      "learning_rate": 1.0326211988615031e-06,
      "loss": 0.7132,
      "step": 3991
    },
    {
      "epoch": 0.8586792858679286,
      "grad_norm": 0.0,
      "learning_rate": 1.029539876673541e-06,
      "loss": 0.8411,
      "step": 3992
    },
    {
      "epoch": 0.8588943858894386,
      "grad_norm": 0.0,
      "learning_rate": 1.0264629091428213e-06,
      "loss": 0.7886,
      "step": 3993
    },
    {
      "epoch": 0.8591094859109486,
      "grad_norm": 0.0,
      "learning_rate": 1.0233902977630383e-06,
      "loss": 0.863,
      "step": 3994
    },
    {
      "epoch": 0.8593245859324586,
      "grad_norm": 0.0,
      "learning_rate": 1.0203220440257756e-06,
      "loss": 0.8503,
      "step": 3995
    },
    {
      "epoch": 0.8595396859539686,
      "grad_norm": 0.0,
      "learning_rate": 1.0172581494204992e-06,
      "loss": 0.7925,
      "step": 3996
    },
    {
      "epoch": 0.8597547859754786,
      "grad_norm": 0.0,
      "learning_rate": 1.0141986154345573e-06,
      "loss": 0.8009,
      "step": 3997
    },
    {
      "epoch": 0.8599698859969886,
      "grad_norm": 0.0,
      "learning_rate": 1.0111434435531852e-06,
      "loss": 0.8016,
      "step": 3998
    },
    {
      "epoch": 0.8601849860184986,
      "grad_norm": 0.0,
      "learning_rate": 1.0080926352594945e-06,
      "loss": 0.8077,
      "step": 3999
    },
    {
      "epoch": 0.8604000860400086,
      "grad_norm": 0.0,
      "learning_rate": 1.0050461920344846e-06,
      "loss": 0.7732,
      "step": 4000
    },
    {
      "epoch": 0.8606151860615187,
      "grad_norm": 0.0,
      "learning_rate": 1.0020041153570349e-06,
      "loss": 0.769,
      "step": 4001
    },
    {
      "epoch": 0.8608302860830286,
      "grad_norm": 0.0,
      "learning_rate": 9.989664067038996e-07,
      "loss": 0.8817,
      "step": 4002
    },
    {
      "epoch": 0.8610453861045386,
      "grad_norm": 0.0,
      "learning_rate": 9.959330675497202e-07,
      "loss": 0.7599,
      "step": 4003
    },
    {
      "epoch": 0.8612604861260487,
      "grad_norm": 0.0,
      "learning_rate": 9.929040993670113e-07,
      "loss": 0.7567,
      "step": 4004
    },
    {
      "epoch": 0.8614755861475586,
      "grad_norm": 0.0,
      "learning_rate": 9.898795036261688e-07,
      "loss": 0.7729,
      "step": 4005
    },
    {
      "epoch": 0.8616906861690686,
      "grad_norm": 0.0,
      "learning_rate": 9.868592817954637e-07,
      "loss": 0.7936,
      "step": 4006
    },
    {
      "epoch": 0.8619057861905787,
      "grad_norm": 0.0,
      "learning_rate": 9.838434353410476e-07,
      "loss": 0.7803,
      "step": 4007
    },
    {
      "epoch": 0.8621208862120886,
      "grad_norm": 0.0,
      "learning_rate": 9.808319657269417e-07,
      "loss": 0.8321,
      "step": 4008
    },
    {
      "epoch": 0.8623359862335986,
      "grad_norm": 0.0,
      "learning_rate": 9.77824874415051e-07,
      "loss": 0.8148,
      "step": 4009
    },
    {
      "epoch": 0.8625510862551087,
      "grad_norm": 0.0,
      "learning_rate": 9.748221628651445e-07,
      "loss": 0.856,
      "step": 4010
    },
    {
      "epoch": 0.8627661862766186,
      "grad_norm": 0.0,
      "learning_rate": 9.718238325348761e-07,
      "loss": 0.8032,
      "step": 4011
    },
    {
      "epoch": 0.8629812862981286,
      "grad_norm": 0.0,
      "learning_rate": 9.68829884879764e-07,
      "loss": 0.7729,
      "step": 4012
    },
    {
      "epoch": 0.8631963863196387,
      "grad_norm": 0.0,
      "learning_rate": 9.658403213532042e-07,
      "loss": 0.7833,
      "step": 4013
    },
    {
      "epoch": 0.8634114863411486,
      "grad_norm": 0.0,
      "learning_rate": 9.628551434064648e-07,
      "loss": 0.7611,
      "step": 4014
    },
    {
      "epoch": 0.8636265863626587,
      "grad_norm": 0.0,
      "learning_rate": 9.598743524886777e-07,
      "loss": 0.8537,
      "step": 4015
    },
    {
      "epoch": 0.8638416863841686,
      "grad_norm": 0.0,
      "learning_rate": 9.568979500468567e-07,
      "loss": 0.8364,
      "step": 4016
    },
    {
      "epoch": 0.8640567864056786,
      "grad_norm": 0.0,
      "learning_rate": 9.539259375258737e-07,
      "loss": 0.8443,
      "step": 4017
    },
    {
      "epoch": 0.8642718864271887,
      "grad_norm": 0.0,
      "learning_rate": 9.509583163684755e-07,
      "loss": 0.8187,
      "step": 4018
    },
    {
      "epoch": 0.8644869864486986,
      "grad_norm": 0.0,
      "learning_rate": 9.47995088015281e-07,
      "loss": 0.8419,
      "step": 4019
    },
    {
      "epoch": 0.8647020864702086,
      "grad_norm": 0.0,
      "learning_rate": 9.450362539047675e-07,
      "loss": 0.8777,
      "step": 4020
    },
    {
      "epoch": 0.8649171864917187,
      "grad_norm": 0.0,
      "learning_rate": 9.420818154732836e-07,
      "loss": 0.7702,
      "step": 4021
    },
    {
      "epoch": 0.8651322865132286,
      "grad_norm": 0.0,
      "learning_rate": 9.391317741550465e-07,
      "loss": 0.7896,
      "step": 4022
    },
    {
      "epoch": 0.8653473865347386,
      "grad_norm": 0.0,
      "learning_rate": 9.361861313821341e-07,
      "loss": 0.7701,
      "step": 4023
    },
    {
      "epoch": 0.8655624865562487,
      "grad_norm": 0.0,
      "learning_rate": 9.332448885844936e-07,
      "loss": 0.8665,
      "step": 4024
    },
    {
      "epoch": 0.8657775865777586,
      "grad_norm": 0.0,
      "learning_rate": 9.303080471899318e-07,
      "loss": 0.8586,
      "step": 4025
    },
    {
      "epoch": 0.8659926865992686,
      "grad_norm": 0.0,
      "learning_rate": 9.273756086241226e-07,
      "loss": 0.808,
      "step": 4026
    },
    {
      "epoch": 0.8662077866207787,
      "grad_norm": 0.0,
      "learning_rate": 9.244475743106019e-07,
      "loss": 0.81,
      "step": 4027
    },
    {
      "epoch": 0.8664228866422886,
      "grad_norm": 0.0,
      "learning_rate": 9.215239456707636e-07,
      "loss": 0.7797,
      "step": 4028
    },
    {
      "epoch": 0.8666379866637987,
      "grad_norm": 0.0,
      "learning_rate": 9.186047241238715e-07,
      "loss": 0.7424,
      "step": 4029
    },
    {
      "epoch": 0.8668530866853087,
      "grad_norm": 0.0,
      "learning_rate": 9.156899110870376e-07,
      "loss": 0.7831,
      "step": 4030
    },
    {
      "epoch": 0.8670681867068186,
      "grad_norm": 0.0,
      "learning_rate": 9.127795079752455e-07,
      "loss": 0.8355,
      "step": 4031
    },
    {
      "epoch": 0.8672832867283287,
      "grad_norm": 0.0,
      "learning_rate": 9.098735162013339e-07,
      "loss": 0.7995,
      "step": 4032
    },
    {
      "epoch": 0.8674983867498387,
      "grad_norm": 0.0,
      "learning_rate": 9.069719371759966e-07,
      "loss": 0.8055,
      "step": 4033
    },
    {
      "epoch": 0.8677134867713486,
      "grad_norm": 0.0,
      "learning_rate": 9.040747723077903e-07,
      "loss": 0.7507,
      "step": 4034
    },
    {
      "epoch": 0.8679285867928587,
      "grad_norm": 0.0,
      "learning_rate": 9.011820230031243e-07,
      "loss": 0.772,
      "step": 4035
    },
    {
      "epoch": 0.8681436868143687,
      "grad_norm": 0.0,
      "learning_rate": 8.982936906662698e-07,
      "loss": 0.8713,
      "step": 4036
    },
    {
      "epoch": 0.8683587868358786,
      "grad_norm": 0.0,
      "learning_rate": 8.954097766993497e-07,
      "loss": 0.8435,
      "step": 4037
    },
    {
      "epoch": 0.8685738868573887,
      "grad_norm": 0.0,
      "learning_rate": 8.925302825023407e-07,
      "loss": 0.8354,
      "step": 4038
    },
    {
      "epoch": 0.8687889868788987,
      "grad_norm": 0.0,
      "learning_rate": 8.896552094730781e-07,
      "loss": 0.8245,
      "step": 4039
    },
    {
      "epoch": 0.8690040869004086,
      "grad_norm": 0.0,
      "learning_rate": 8.86784559007251e-07,
      "loss": 0.7636,
      "step": 4040
    },
    {
      "epoch": 0.8692191869219187,
      "grad_norm": 0.0,
      "learning_rate": 8.839183324983969e-07,
      "loss": 0.8425,
      "step": 4041
    },
    {
      "epoch": 0.8694342869434287,
      "grad_norm": 0.0,
      "learning_rate": 8.810565313379116e-07,
      "loss": 0.7812,
      "step": 4042
    },
    {
      "epoch": 0.8696493869649387,
      "grad_norm": 0.0,
      "learning_rate": 8.781991569150361e-07,
      "loss": 0.805,
      "step": 4043
    },
    {
      "epoch": 0.8698644869864487,
      "grad_norm": 0.0,
      "learning_rate": 8.753462106168675e-07,
      "loss": 0.8906,
      "step": 4044
    },
    {
      "epoch": 0.8700795870079587,
      "grad_norm": 0.0,
      "learning_rate": 8.724976938283558e-07,
      "loss": 0.7649,
      "step": 4045
    },
    {
      "epoch": 0.8702946870294687,
      "grad_norm": 0.0,
      "learning_rate": 8.696536079322904e-07,
      "loss": 0.8462,
      "step": 4046
    },
    {
      "epoch": 0.8705097870509787,
      "grad_norm": 0.0,
      "learning_rate": 8.668139543093201e-07,
      "loss": 0.8447,
      "step": 4047
    },
    {
      "epoch": 0.8707248870724887,
      "grad_norm": 0.0,
      "learning_rate": 8.639787343379369e-07,
      "loss": 0.8451,
      "step": 4048
    },
    {
      "epoch": 0.8709399870939987,
      "grad_norm": 0.0,
      "learning_rate": 8.61147949394483e-07,
      "loss": 0.7721,
      "step": 4049
    },
    {
      "epoch": 0.8711550871155087,
      "grad_norm": 0.0,
      "learning_rate": 8.583216008531425e-07,
      "loss": 0.8067,
      "step": 4050
    },
    {
      "epoch": 0.8713701871370187,
      "grad_norm": 0.0,
      "learning_rate": 8.554996900859558e-07,
      "loss": 0.7619,
      "step": 4051
    },
    {
      "epoch": 0.8715852871585287,
      "grad_norm": 0.0,
      "learning_rate": 8.526822184627959e-07,
      "loss": 0.8198,
      "step": 4052
    },
    {
      "epoch": 0.8718003871800387,
      "grad_norm": 0.0,
      "learning_rate": 8.498691873513942e-07,
      "loss": 0.8013,
      "step": 4053
    },
    {
      "epoch": 0.8720154872015488,
      "grad_norm": 0.0,
      "learning_rate": 8.470605981173163e-07,
      "loss": 0.7863,
      "step": 4054
    },
    {
      "epoch": 0.8722305872230587,
      "grad_norm": 0.0,
      "learning_rate": 8.442564521239782e-07,
      "loss": 0.7414,
      "step": 4055
    },
    {
      "epoch": 0.8724456872445687,
      "grad_norm": 0.0,
      "learning_rate": 8.414567507326321e-07,
      "loss": 0.7875,
      "step": 4056
    },
    {
      "epoch": 0.8726607872660788,
      "grad_norm": 0.0,
      "learning_rate": 8.386614953023786e-07,
      "loss": 0.8031,
      "step": 4057
    },
    {
      "epoch": 0.8728758872875887,
      "grad_norm": 0.0,
      "learning_rate": 8.358706871901612e-07,
      "loss": 0.7616,
      "step": 4058
    },
    {
      "epoch": 0.8730909873090987,
      "grad_norm": 0.0,
      "learning_rate": 8.330843277507561e-07,
      "loss": 0.8623,
      "step": 4059
    },
    {
      "epoch": 0.8733060873306088,
      "grad_norm": 0.0,
      "learning_rate": 8.303024183367892e-07,
      "loss": 0.77,
      "step": 4060
    },
    {
      "epoch": 0.8735211873521187,
      "grad_norm": 0.0,
      "learning_rate": 8.27524960298719e-07,
      "loss": 0.7915,
      "step": 4061
    },
    {
      "epoch": 0.8737362873736287,
      "grad_norm": 0.0,
      "learning_rate": 8.247519549848482e-07,
      "loss": 0.8058,
      "step": 4062
    },
    {
      "epoch": 0.8739513873951388,
      "grad_norm": 0.0,
      "learning_rate": 8.219834037413165e-07,
      "loss": 0.768,
      "step": 4063
    },
    {
      "epoch": 0.8741664874166487,
      "grad_norm": 0.0,
      "learning_rate": 8.192193079121003e-07,
      "loss": 0.82,
      "step": 4064
    },
    {
      "epoch": 0.8743815874381587,
      "grad_norm": 0.0,
      "learning_rate": 8.164596688390114e-07,
      "loss": 0.8474,
      "step": 4065
    },
    {
      "epoch": 0.8745966874596688,
      "grad_norm": 0.0,
      "learning_rate": 8.13704487861704e-07,
      "loss": 0.7552,
      "step": 4066
    },
    {
      "epoch": 0.8748117874811787,
      "grad_norm": 0.0,
      "learning_rate": 8.109537663176603e-07,
      "loss": 0.8302,
      "step": 4067
    },
    {
      "epoch": 0.8750268875026888,
      "grad_norm": 0.0,
      "learning_rate": 8.082075055422067e-07,
      "loss": 0.824,
      "step": 4068
    },
    {
      "epoch": 0.8752419875241988,
      "grad_norm": 0.0,
      "learning_rate": 8.054657068684957e-07,
      "loss": 0.7457,
      "step": 4069
    },
    {
      "epoch": 0.8754570875457087,
      "grad_norm": 0.0,
      "learning_rate": 8.027283716275181e-07,
      "loss": 0.8522,
      "step": 4070
    },
    {
      "epoch": 0.8756721875672188,
      "grad_norm": 0.0,
      "learning_rate": 7.999955011480997e-07,
      "loss": 0.8092,
      "step": 4071
    },
    {
      "epoch": 0.8758872875887288,
      "grad_norm": 0.0,
      "learning_rate": 7.972670967568929e-07,
      "loss": 0.8704,
      "step": 4072
    },
    {
      "epoch": 0.8761023876102387,
      "grad_norm": 0.0,
      "learning_rate": 7.945431597783904e-07,
      "loss": 0.8121,
      "step": 4073
    },
    {
      "epoch": 0.8763174876317488,
      "grad_norm": 0.0,
      "learning_rate": 7.918236915349054e-07,
      "loss": 0.8276,
      "step": 4074
    },
    {
      "epoch": 0.8765325876532588,
      "grad_norm": 0.0,
      "learning_rate": 7.891086933465908e-07,
      "loss": 0.8533,
      "step": 4075
    },
    {
      "epoch": 0.8767476876747687,
      "grad_norm": 0.0,
      "learning_rate": 7.86398166531428e-07,
      "loss": 0.7981,
      "step": 4076
    },
    {
      "epoch": 0.8769627876962788,
      "grad_norm": 0.0,
      "learning_rate": 7.836921124052244e-07,
      "loss": 0.8322,
      "step": 4077
    },
    {
      "epoch": 0.8771778877177888,
      "grad_norm": 0.0,
      "learning_rate": 7.809905322816192e-07,
      "loss": 0.8279,
      "step": 4078
    },
    {
      "epoch": 0.8773929877392987,
      "grad_norm": 0.0,
      "learning_rate": 7.782934274720777e-07,
      "loss": 0.7878,
      "step": 4079
    },
    {
      "epoch": 0.8776080877608088,
      "grad_norm": 0.0,
      "learning_rate": 7.756007992858928e-07,
      "loss": 0.871,
      "step": 4080
    },
    {
      "epoch": 0.8778231877823188,
      "grad_norm": 0.0,
      "learning_rate": 7.729126490301885e-07,
      "loss": 0.7948,
      "step": 4081
    },
    {
      "epoch": 0.8780382878038288,
      "grad_norm": 0.0,
      "learning_rate": 7.70228978009907e-07,
      "loss": 0.8043,
      "step": 4082
    },
    {
      "epoch": 0.8782533878253388,
      "grad_norm": 0.0,
      "learning_rate": 7.675497875278226e-07,
      "loss": 0.823,
      "step": 4083
    },
    {
      "epoch": 0.8784684878468488,
      "grad_norm": 0.0,
      "learning_rate": 7.648750788845338e-07,
      "loss": 0.8113,
      "step": 4084
    },
    {
      "epoch": 0.8786835878683588,
      "grad_norm": 0.0,
      "learning_rate": 7.62204853378461e-07,
      "loss": 0.8034,
      "step": 4085
    },
    {
      "epoch": 0.8788986878898688,
      "grad_norm": 0.0,
      "learning_rate": 7.595391123058515e-07,
      "loss": 0.7887,
      "step": 4086
    },
    {
      "epoch": 0.8791137879113788,
      "grad_norm": 0.0,
      "learning_rate": 7.568778569607693e-07,
      "loss": 0.8328,
      "step": 4087
    },
    {
      "epoch": 0.8793288879328888,
      "grad_norm": 0.0,
      "learning_rate": 7.542210886351087e-07,
      "loss": 0.7811,
      "step": 4088
    },
    {
      "epoch": 0.8795439879543988,
      "grad_norm": 0.0,
      "learning_rate": 7.515688086185824e-07,
      "loss": 0.7833,
      "step": 4089
    },
    {
      "epoch": 0.8797590879759088,
      "grad_norm": 0.0,
      "learning_rate": 7.489210181987228e-07,
      "loss": 0.8512,
      "step": 4090
    },
    {
      "epoch": 0.8799741879974188,
      "grad_norm": 0.0,
      "learning_rate": 7.46277718660885e-07,
      "loss": 0.7635,
      "step": 4091
    },
    {
      "epoch": 0.8801892880189288,
      "grad_norm": 0.0,
      "learning_rate": 7.436389112882414e-07,
      "loss": 0.8277,
      "step": 4092
    },
    {
      "epoch": 0.8804043880404389,
      "grad_norm": 0.0,
      "learning_rate": 7.410045973617896e-07,
      "loss": 0.8393,
      "step": 4093
    },
    {
      "epoch": 0.8806194880619488,
      "grad_norm": 0.0,
      "learning_rate": 7.38374778160339e-07,
      "loss": 0.8259,
      "step": 4094
    },
    {
      "epoch": 0.8808345880834588,
      "grad_norm": 0.0,
      "learning_rate": 7.357494549605181e-07,
      "loss": 0.8248,
      "step": 4095
    },
    {
      "epoch": 0.8810496881049689,
      "grad_norm": 0.0,
      "learning_rate": 7.331286290367779e-07,
      "loss": 0.8801,
      "step": 4096
    },
    {
      "epoch": 0.8812647881264788,
      "grad_norm": 0.0,
      "learning_rate": 7.305123016613825e-07,
      "loss": 0.761,
      "step": 4097
    },
    {
      "epoch": 0.8814798881479888,
      "grad_norm": 0.0,
      "learning_rate": 7.279004741044104e-07,
      "loss": 0.7518,
      "step": 4098
    },
    {
      "epoch": 0.8816949881694989,
      "grad_norm": 0.0,
      "learning_rate": 7.25293147633761e-07,
      "loss": 0.886,
      "step": 4099
    },
    {
      "epoch": 0.8819100881910088,
      "grad_norm": 0.0,
      "learning_rate": 7.226903235151439e-07,
      "loss": 0.768,
      "step": 4100
    },
    {
      "epoch": 0.8821251882125188,
      "grad_norm": 0.0,
      "learning_rate": 7.20092003012084e-07,
      "loss": 0.8768,
      "step": 4101
    },
    {
      "epoch": 0.8823402882340288,
      "grad_norm": 0.0,
      "learning_rate": 7.174981873859232e-07,
      "loss": 0.7819,
      "step": 4102
    },
    {
      "epoch": 0.8825553882555388,
      "grad_norm": 0.0,
      "learning_rate": 7.149088778958125e-07,
      "loss": 0.8711,
      "step": 4103
    },
    {
      "epoch": 0.8827704882770488,
      "grad_norm": 0.0,
      "learning_rate": 7.123240757987182e-07,
      "loss": 0.8119,
      "step": 4104
    },
    {
      "epoch": 0.8829855882985588,
      "grad_norm": 0.0,
      "learning_rate": 7.097437823494147e-07,
      "loss": 0.7867,
      "step": 4105
    },
    {
      "epoch": 0.8832006883200688,
      "grad_norm": 0.0,
      "learning_rate": 7.07167998800491e-07,
      "loss": 0.8603,
      "step": 4106
    },
    {
      "epoch": 0.8834157883415789,
      "grad_norm": 0.0,
      "learning_rate": 7.045967264023502e-07,
      "loss": 0.8052,
      "step": 4107
    },
    {
      "epoch": 0.8836308883630888,
      "grad_norm": 0.0,
      "learning_rate": 7.020299664031971e-07,
      "loss": 0.7536,
      "step": 4108
    },
    {
      "epoch": 0.8838459883845988,
      "grad_norm": 0.0,
      "learning_rate": 6.994677200490508e-07,
      "loss": 0.7889,
      "step": 4109
    },
    {
      "epoch": 0.8840610884061089,
      "grad_norm": 0.0,
      "learning_rate": 6.969099885837405e-07,
      "loss": 0.7652,
      "step": 4110
    },
    {
      "epoch": 0.8842761884276188,
      "grad_norm": 0.0,
      "learning_rate": 6.943567732488999e-07,
      "loss": 0.8275,
      "step": 4111
    },
    {
      "epoch": 0.8844912884491288,
      "grad_norm": 0.0,
      "learning_rate": 6.91808075283975e-07,
      "loss": 0.8425,
      "step": 4112
    },
    {
      "epoch": 0.8847063884706389,
      "grad_norm": 0.0,
      "learning_rate": 6.892638959262121e-07,
      "loss": 0.8343,
      "step": 4113
    },
    {
      "epoch": 0.8849214884921488,
      "grad_norm": 0.0,
      "learning_rate": 6.867242364106719e-07,
      "loss": 0.7309,
      "step": 4114
    },
    {
      "epoch": 0.8851365885136588,
      "grad_norm": 0.0,
      "learning_rate": 6.84189097970217e-07,
      "loss": 0.8466,
      "step": 4115
    },
    {
      "epoch": 0.8853516885351689,
      "grad_norm": 0.0,
      "learning_rate": 6.816584818355132e-07,
      "loss": 0.8423,
      "step": 4116
    },
    {
      "epoch": 0.8855667885566788,
      "grad_norm": 0.0,
      "learning_rate": 6.791323892350354e-07,
      "loss": 0.8956,
      "step": 4117
    },
    {
      "epoch": 0.8857818885781888,
      "grad_norm": 0.0,
      "learning_rate": 6.766108213950584e-07,
      "loss": 0.7747,
      "step": 4118
    },
    {
      "epoch": 0.8859969885996989,
      "grad_norm": 0.0,
      "learning_rate": 6.740937795396641e-07,
      "loss": 0.7725,
      "step": 4119
    },
    {
      "epoch": 0.8862120886212088,
      "grad_norm": 0.0,
      "learning_rate": 6.715812648907361e-07,
      "loss": 0.8863,
      "step": 4120
    },
    {
      "epoch": 0.8864271886427189,
      "grad_norm": 0.0,
      "learning_rate": 6.690732786679588e-07,
      "loss": 0.7911,
      "step": 4121
    },
    {
      "epoch": 0.8866422886642289,
      "grad_norm": 0.0,
      "learning_rate": 6.665698220888206e-07,
      "loss": 0.8432,
      "step": 4122
    },
    {
      "epoch": 0.8868573886857388,
      "grad_norm": 0.0,
      "learning_rate": 6.640708963686082e-07,
      "loss": 0.8159,
      "step": 4123
    },
    {
      "epoch": 0.8870724887072489,
      "grad_norm": 0.0,
      "learning_rate": 6.615765027204101e-07,
      "loss": 0.8106,
      "step": 4124
    },
    {
      "epoch": 0.8872875887287589,
      "grad_norm": 0.0,
      "learning_rate": 6.590866423551179e-07,
      "loss": 0.7319,
      "step": 4125
    },
    {
      "epoch": 0.8875026887502688,
      "grad_norm": 0.0,
      "learning_rate": 6.566013164814167e-07,
      "loss": 0.7563,
      "step": 4126
    },
    {
      "epoch": 0.8877177887717789,
      "grad_norm": 0.0,
      "learning_rate": 6.541205263057936e-07,
      "loss": 0.8665,
      "step": 4127
    },
    {
      "epoch": 0.8879328887932889,
      "grad_norm": 0.0,
      "learning_rate": 6.516442730325378e-07,
      "loss": 0.7753,
      "step": 4128
    },
    {
      "epoch": 0.8881479888147988,
      "grad_norm": 0.0,
      "learning_rate": 6.491725578637254e-07,
      "loss": 0.7733,
      "step": 4129
    },
    {
      "epoch": 0.8883630888363089,
      "grad_norm": 0.0,
      "learning_rate": 6.467053819992419e-07,
      "loss": 0.8197,
      "step": 4130
    },
    {
      "epoch": 0.8885781888578189,
      "grad_norm": 0.0,
      "learning_rate": 6.442427466367595e-07,
      "loss": 0.7809,
      "step": 4131
    },
    {
      "epoch": 0.8887932888793288,
      "grad_norm": 0.0,
      "learning_rate": 6.417846529717509e-07,
      "loss": 0.8343,
      "step": 4132
    },
    {
      "epoch": 0.8890083889008389,
      "grad_norm": 0.0,
      "learning_rate": 6.393311021974858e-07,
      "loss": 0.834,
      "step": 4133
    },
    {
      "epoch": 0.8892234889223489,
      "grad_norm": 0.0,
      "learning_rate": 6.368820955050214e-07,
      "loss": 0.8609,
      "step": 4134
    },
    {
      "epoch": 0.8894385889438589,
      "grad_norm": 0.0,
      "learning_rate": 6.344376340832181e-07,
      "loss": 0.8688,
      "step": 4135
    },
    {
      "epoch": 0.8896536889653689,
      "grad_norm": 0.0,
      "learning_rate": 6.319977191187232e-07,
      "loss": 0.7604,
      "step": 4136
    },
    {
      "epoch": 0.8898687889868789,
      "grad_norm": 0.0,
      "learning_rate": 6.295623517959793e-07,
      "loss": 0.8473,
      "step": 4137
    },
    {
      "epoch": 0.8900838890083889,
      "grad_norm": 0.0,
      "learning_rate": 6.271315332972227e-07,
      "loss": 0.8738,
      "step": 4138
    },
    {
      "epoch": 0.8902989890298989,
      "grad_norm": 0.0,
      "learning_rate": 6.247052648024765e-07,
      "loss": 0.8354,
      "step": 4139
    },
    {
      "epoch": 0.8905140890514089,
      "grad_norm": 0.0,
      "learning_rate": 6.222835474895594e-07,
      "loss": 0.7946,
      "step": 4140
    },
    {
      "epoch": 0.8907291890729189,
      "grad_norm": 0.0,
      "learning_rate": 6.198663825340845e-07,
      "loss": 0.8237,
      "step": 4141
    },
    {
      "epoch": 0.8909442890944289,
      "grad_norm": 0.0,
      "learning_rate": 6.174537711094442e-07,
      "loss": 0.7563,
      "step": 4142
    },
    {
      "epoch": 0.891159389115939,
      "grad_norm": 0.0,
      "learning_rate": 6.150457143868305e-07,
      "loss": 0.7942,
      "step": 4143
    },
    {
      "epoch": 0.8913744891374489,
      "grad_norm": 0.0,
      "learning_rate": 6.126422135352184e-07,
      "loss": 0.7924,
      "step": 4144
    },
    {
      "epoch": 0.8915895891589589,
      "grad_norm": 0.0,
      "learning_rate": 6.102432697213734e-07,
      "loss": 0.7968,
      "step": 4145
    },
    {
      "epoch": 0.891804689180469,
      "grad_norm": 0.0,
      "learning_rate": 6.078488841098518e-07,
      "loss": 0.8032,
      "step": 4146
    },
    {
      "epoch": 0.8920197892019789,
      "grad_norm": 0.0,
      "learning_rate": 6.054590578629893e-07,
      "loss": 0.746,
      "step": 4147
    },
    {
      "epoch": 0.8922348892234889,
      "grad_norm": 0.0,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.8373,
      "step": 4148
    },
    {
      "epoch": 0.892449989244999,
      "grad_norm": 0.0,
      "learning_rate": 6.006930881015449e-07,
      "loss": 0.8345,
      "step": 4149
    },
    {
      "epoch": 0.8926650892665089,
      "grad_norm": 0.0,
      "learning_rate": 5.983169469005745e-07,
      "loss": 0.8288,
      "step": 4150
    },
    {
      "epoch": 0.8928801892880189,
      "grad_norm": 0.0,
      "learning_rate": 5.959453696914874e-07,
      "loss": 0.8162,
      "step": 4151
    },
    {
      "epoch": 0.893095289309529,
      "grad_norm": 0.0,
      "learning_rate": 5.93578357625555e-07,
      "loss": 0.7956,
      "step": 4152
    },
    {
      "epoch": 0.8933103893310389,
      "grad_norm": 0.0,
      "learning_rate": 5.912159118518246e-07,
      "loss": 0.8879,
      "step": 4153
    },
    {
      "epoch": 0.8935254893525489,
      "grad_norm": 0.0,
      "learning_rate": 5.888580335171368e-07,
      "loss": 0.7797,
      "step": 4154
    },
    {
      "epoch": 0.893740589374059,
      "grad_norm": 0.0,
      "learning_rate": 5.865047237661059e-07,
      "loss": 0.8232,
      "step": 4155
    },
    {
      "epoch": 0.8939556893955689,
      "grad_norm": 0.0,
      "learning_rate": 5.841559837411348e-07,
      "loss": 0.7893,
      "step": 4156
    },
    {
      "epoch": 0.894170789417079,
      "grad_norm": 0.0,
      "learning_rate": 5.818118145824036e-07,
      "loss": 0.7299,
      "step": 4157
    },
    {
      "epoch": 0.894385889438589,
      "grad_norm": 0.0,
      "learning_rate": 5.794722174278766e-07,
      "loss": 0.8014,
      "step": 4158
    },
    {
      "epoch": 0.8946009894600989,
      "grad_norm": 0.0,
      "learning_rate": 5.771371934132986e-07,
      "loss": 0.8195,
      "step": 4159
    },
    {
      "epoch": 0.894816089481609,
      "grad_norm": 0.0,
      "learning_rate": 5.748067436721893e-07,
      "loss": 0.8043,
      "step": 4160
    },
    {
      "epoch": 0.895031189503119,
      "grad_norm": 0.0,
      "learning_rate": 5.724808693358574e-07,
      "loss": 0.899,
      "step": 4161
    },
    {
      "epoch": 0.8952462895246289,
      "grad_norm": 0.0,
      "learning_rate": 5.701595715333807e-07,
      "loss": 0.7971,
      "step": 4162
    },
    {
      "epoch": 0.895461389546139,
      "grad_norm": 0.0,
      "learning_rate": 5.678428513916212e-07,
      "loss": 0.8225,
      "step": 4163
    },
    {
      "epoch": 0.895676489567649,
      "grad_norm": 0.0,
      "learning_rate": 5.655307100352181e-07,
      "loss": 0.8432,
      "step": 4164
    },
    {
      "epoch": 0.8958915895891589,
      "grad_norm": 0.0,
      "learning_rate": 5.632231485865857e-07,
      "loss": 0.8206,
      "step": 4165
    },
    {
      "epoch": 0.896106689610669,
      "grad_norm": 0.0,
      "learning_rate": 5.609201681659149e-07,
      "loss": 0.7885,
      "step": 4166
    },
    {
      "epoch": 0.896321789632179,
      "grad_norm": 0.0,
      "learning_rate": 5.586217698911766e-07,
      "loss": 0.7987,
      "step": 4167
    },
    {
      "epoch": 0.8965368896536889,
      "grad_norm": 0.0,
      "learning_rate": 5.56327954878112e-07,
      "loss": 0.8203,
      "step": 4168
    },
    {
      "epoch": 0.896751989675199,
      "grad_norm": 0.0,
      "learning_rate": 5.540387242402434e-07,
      "loss": 0.822,
      "step": 4169
    },
    {
      "epoch": 0.896967089696709,
      "grad_norm": 0.0,
      "learning_rate": 5.517540790888609e-07,
      "loss": 0.7397,
      "step": 4170
    },
    {
      "epoch": 0.897182189718219,
      "grad_norm": 0.0,
      "learning_rate": 5.49474020533034e-07,
      "loss": 0.7822,
      "step": 4171
    },
    {
      "epoch": 0.897397289739729,
      "grad_norm": 0.0,
      "learning_rate": 5.471985496796051e-07,
      "loss": 0.8293,
      "step": 4172
    },
    {
      "epoch": 0.897612389761239,
      "grad_norm": 0.0,
      "learning_rate": 5.449276676331849e-07,
      "loss": 0.7993,
      "step": 4173
    },
    {
      "epoch": 0.897827489782749,
      "grad_norm": 0.0,
      "learning_rate": 5.42661375496164e-07,
      "loss": 0.7558,
      "step": 4174
    },
    {
      "epoch": 0.898042589804259,
      "grad_norm": 0.0,
      "learning_rate": 5.403996743686957e-07,
      "loss": 0.8414,
      "step": 4175
    },
    {
      "epoch": 0.898257689825769,
      "grad_norm": 0.0,
      "learning_rate": 5.381425653487115e-07,
      "loss": 0.7719,
      "step": 4176
    },
    {
      "epoch": 0.898472789847279,
      "grad_norm": 0.0,
      "learning_rate": 5.358900495319142e-07,
      "loss": 0.863,
      "step": 4177
    },
    {
      "epoch": 0.898687889868789,
      "grad_norm": 0.0,
      "learning_rate": 5.33642128011771e-07,
      "loss": 0.7549,
      "step": 4178
    },
    {
      "epoch": 0.898902989890299,
      "grad_norm": 0.0,
      "learning_rate": 5.313988018795246e-07,
      "loss": 0.766,
      "step": 4179
    },
    {
      "epoch": 0.899118089911809,
      "grad_norm": 0.0,
      "learning_rate": 5.291600722241841e-07,
      "loss": 0.6876,
      "step": 4180
    },
    {
      "epoch": 0.899333189933319,
      "grad_norm": 0.0,
      "learning_rate": 5.269259401325256e-07,
      "loss": 0.8227,
      "step": 4181
    },
    {
      "epoch": 0.899548289954829,
      "grad_norm": 0.0,
      "learning_rate": 5.246964066890981e-07,
      "loss": 0.8183,
      "step": 4182
    },
    {
      "epoch": 0.899763389976339,
      "grad_norm": 0.0,
      "learning_rate": 5.224714729762137e-07,
      "loss": 0.8426,
      "step": 4183
    },
    {
      "epoch": 0.899978489997849,
      "grad_norm": 0.0,
      "learning_rate": 5.20251140073953e-07,
      "loss": 0.8602,
      "step": 4184
    },
    {
      "epoch": 0.900193590019359,
      "grad_norm": 0.0,
      "learning_rate": 5.180354090601669e-07,
      "loss": 0.7665,
      "step": 4185
    },
    {
      "epoch": 0.900408690040869,
      "grad_norm": 0.0,
      "learning_rate": 5.158242810104652e-07,
      "loss": 0.83,
      "step": 4186
    },
    {
      "epoch": 0.900623790062379,
      "grad_norm": 0.0,
      "learning_rate": 5.136177569982293e-07,
      "loss": 0.7985,
      "step": 4187
    },
    {
      "epoch": 0.900838890083889,
      "grad_norm": 0.0,
      "learning_rate": 5.114158380946022e-07,
      "loss": 0.8612,
      "step": 4188
    },
    {
      "epoch": 0.901053990105399,
      "grad_norm": 0.0,
      "learning_rate": 5.092185253684923e-07,
      "loss": 0.8068,
      "step": 4189
    },
    {
      "epoch": 0.901269090126909,
      "grad_norm": 0.0,
      "learning_rate": 5.070258198865741e-07,
      "loss": 0.7839,
      "step": 4190
    },
    {
      "epoch": 0.901484190148419,
      "grad_norm": 0.0,
      "learning_rate": 5.048377227132816e-07,
      "loss": 0.8212,
      "step": 4191
    },
    {
      "epoch": 0.901699290169929,
      "grad_norm": 0.0,
      "learning_rate": 5.026542349108155e-07,
      "loss": 0.7825,
      "step": 4192
    },
    {
      "epoch": 0.901914390191439,
      "grad_norm": 0.0,
      "learning_rate": 5.004753575391341e-07,
      "loss": 0.8149,
      "step": 4193
    },
    {
      "epoch": 0.902129490212949,
      "grad_norm": 0.0,
      "learning_rate": 4.98301091655965e-07,
      "loss": 0.8179,
      "step": 4194
    },
    {
      "epoch": 0.902344590234459,
      "grad_norm": 0.0,
      "learning_rate": 4.961314383167903e-07,
      "loss": 0.849,
      "step": 4195
    },
    {
      "epoch": 0.902559690255969,
      "grad_norm": 0.0,
      "learning_rate": 4.939663985748544e-07,
      "loss": 0.8117,
      "step": 4196
    },
    {
      "epoch": 0.902774790277479,
      "grad_norm": 0.0,
      "learning_rate": 4.918059734811642e-07,
      "loss": 0.8658,
      "step": 4197
    },
    {
      "epoch": 0.902989890298989,
      "grad_norm": 0.0,
      "learning_rate": 4.896501640844864e-07,
      "loss": 0.8302,
      "step": 4198
    },
    {
      "epoch": 0.9032049903204991,
      "grad_norm": 0.0,
      "learning_rate": 4.874989714313449e-07,
      "loss": 0.8427,
      "step": 4199
    },
    {
      "epoch": 0.903420090342009,
      "grad_norm": 0.0,
      "learning_rate": 4.853523965660245e-07,
      "loss": 0.8094,
      "step": 4200
    },
    {
      "epoch": 0.903635190363519,
      "grad_norm": 0.0,
      "learning_rate": 4.832104405305659e-07,
      "loss": 0.8133,
      "step": 4201
    },
    {
      "epoch": 0.9038502903850291,
      "grad_norm": 0.0,
      "learning_rate": 4.810731043647699e-07,
      "loss": 0.8374,
      "step": 4202
    },
    {
      "epoch": 0.904065390406539,
      "grad_norm": 0.0,
      "learning_rate": 4.78940389106195e-07,
      "loss": 0.8354,
      "step": 4203
    },
    {
      "epoch": 0.904280490428049,
      "grad_norm": 0.0,
      "learning_rate": 4.768122957901522e-07,
      "loss": 0.7873,
      "step": 4204
    },
    {
      "epoch": 0.9044955904495591,
      "grad_norm": 0.0,
      "learning_rate": 4.7468882544971615e-07,
      "loss": 0.772,
      "step": 4205
    },
    {
      "epoch": 0.904710690471069,
      "grad_norm": 0.0,
      "learning_rate": 4.72569979115709e-07,
      "loss": 0.8248,
      "step": 4206
    },
    {
      "epoch": 0.904925790492579,
      "grad_norm": 0.0,
      "learning_rate": 4.704557578167146e-07,
      "loss": 0.8531,
      "step": 4207
    },
    {
      "epoch": 0.9051408905140891,
      "grad_norm": 0.0,
      "learning_rate": 4.683461625790697e-07,
      "loss": 0.8226,
      "step": 4208
    },
    {
      "epoch": 0.905355990535599,
      "grad_norm": 0.0,
      "learning_rate": 4.662411944268641e-07,
      "loss": 0.8305,
      "step": 4209
    },
    {
      "epoch": 0.905571090557109,
      "grad_norm": 0.0,
      "learning_rate": 4.6414085438194034e-07,
      "loss": 0.7503,
      "step": 4210
    },
    {
      "epoch": 0.9057861905786191,
      "grad_norm": 0.0,
      "learning_rate": 4.620451434638995e-07,
      "loss": 0.7532,
      "step": 4211
    },
    {
      "epoch": 0.906001290600129,
      "grad_norm": 0.0,
      "learning_rate": 4.599540626900889e-07,
      "loss": 0.8496,
      "step": 4212
    },
    {
      "epoch": 0.9062163906216391,
      "grad_norm": 0.0,
      "learning_rate": 4.5786761307561546e-07,
      "loss": 0.846,
      "step": 4213
    },
    {
      "epoch": 0.9064314906431491,
      "grad_norm": 0.0,
      "learning_rate": 4.5578579563333006e-07,
      "loss": 0.8604,
      "step": 4214
    },
    {
      "epoch": 0.906646590664659,
      "grad_norm": 0.0,
      "learning_rate": 4.5370861137383984e-07,
      "loss": 0.7918,
      "step": 4215
    },
    {
      "epoch": 0.9068616906861691,
      "grad_norm": 0.0,
      "learning_rate": 4.5163606130550376e-07,
      "loss": 0.8504,
      "step": 4216
    },
    {
      "epoch": 0.9070767907076791,
      "grad_norm": 0.0,
      "learning_rate": 4.495681464344259e-07,
      "loss": 0.8171,
      "step": 4217
    },
    {
      "epoch": 0.907291890729189,
      "grad_norm": 0.0,
      "learning_rate": 4.4750486776446664e-07,
      "loss": 0.7802,
      "step": 4218
    },
    {
      "epoch": 0.9075069907506991,
      "grad_norm": 0.0,
      "learning_rate": 4.4544622629722924e-07,
      "loss": 0.7425,
      "step": 4219
    },
    {
      "epoch": 0.9077220907722091,
      "grad_norm": 0.0,
      "learning_rate": 4.4339222303207086e-07,
      "loss": 0.8488,
      "step": 4220
    },
    {
      "epoch": 0.907937190793719,
      "grad_norm": 0.0,
      "learning_rate": 4.413428589660962e-07,
      "loss": 0.8511,
      "step": 4221
    },
    {
      "epoch": 0.9081522908152291,
      "grad_norm": 0.0,
      "learning_rate": 4.39298135094155e-07,
      "loss": 0.8597,
      "step": 4222
    },
    {
      "epoch": 0.9083673908367391,
      "grad_norm": 0.0,
      "learning_rate": 4.372580524088488e-07,
      "loss": 0.8315,
      "step": 4223
    },
    {
      "epoch": 0.908582490858249,
      "grad_norm": 0.0,
      "learning_rate": 4.352226119005232e-07,
      "loss": 0.8143,
      "step": 4224
    },
    {
      "epoch": 0.9087975908797591,
      "grad_norm": 0.0,
      "learning_rate": 4.331918145572678e-07,
      "loss": 0.8062,
      "step": 4225
    },
    {
      "epoch": 0.9090126909012691,
      "grad_norm": 0.0,
      "learning_rate": 4.311656613649251e-07,
      "loss": 0.8176,
      "step": 4226
    },
    {
      "epoch": 0.9092277909227791,
      "grad_norm": 0.0,
      "learning_rate": 4.2914415330707726e-07,
      "loss": 0.7844,
      "step": 4227
    },
    {
      "epoch": 0.9094428909442891,
      "grad_norm": 0.0,
      "learning_rate": 4.2712729136505257e-07,
      "loss": 0.771,
      "step": 4228
    },
    {
      "epoch": 0.9096579909657991,
      "grad_norm": 0.0,
      "learning_rate": 4.2511507651792906e-07,
      "loss": 0.8222,
      "step": 4229
    },
    {
      "epoch": 0.9098730909873091,
      "grad_norm": 0.0,
      "learning_rate": 4.2310750974251967e-07,
      "loss": 0.7852,
      "step": 4230
    },
    {
      "epoch": 0.9100881910088191,
      "grad_norm": 0.0,
      "learning_rate": 4.211045920133894e-07,
      "loss": 0.8249,
      "step": 4231
    },
    {
      "epoch": 0.9103032910303291,
      "grad_norm": 0.0,
      "learning_rate": 4.191063243028415e-07,
      "loss": 0.8198,
      "step": 4232
    },
    {
      "epoch": 0.9105183910518391,
      "grad_norm": 0.0,
      "learning_rate": 4.171127075809223e-07,
      "loss": 0.8217,
      "step": 4233
    },
    {
      "epoch": 0.9107334910733491,
      "grad_norm": 0.0,
      "learning_rate": 4.151237428154231e-07,
      "loss": 0.7882,
      "step": 4234
    },
    {
      "epoch": 0.9109485910948592,
      "grad_norm": 0.0,
      "learning_rate": 4.13139430971875e-07,
      "loss": 0.8204,
      "step": 4235
    },
    {
      "epoch": 0.9111636911163691,
      "grad_norm": 0.0,
      "learning_rate": 4.111597730135497e-07,
      "loss": 0.836,
      "step": 4236
    },
    {
      "epoch": 0.9113787911378791,
      "grad_norm": 0.0,
      "learning_rate": 4.091847699014606e-07,
      "loss": 0.8045,
      "step": 4237
    },
    {
      "epoch": 0.9115938911593892,
      "grad_norm": 0.0,
      "learning_rate": 4.072144225943631e-07,
      "loss": 0.7969,
      "step": 4238
    },
    {
      "epoch": 0.9118089911808991,
      "grad_norm": 0.0,
      "learning_rate": 4.052487320487486e-07,
      "loss": 0.8162,
      "step": 4239
    },
    {
      "epoch": 0.9120240912024091,
      "grad_norm": 0.0,
      "learning_rate": 4.0328769921884946e-07,
      "loss": 0.8347,
      "step": 4240
    },
    {
      "epoch": 0.9122391912239192,
      "grad_norm": 0.0,
      "learning_rate": 4.0133132505663977e-07,
      "loss": 0.7668,
      "step": 4241
    },
    {
      "epoch": 0.9124542912454291,
      "grad_norm": 0.0,
      "learning_rate": 3.993796105118297e-07,
      "loss": 0.7102,
      "step": 4242
    },
    {
      "epoch": 0.9126693912669391,
      "grad_norm": 0.0,
      "learning_rate": 3.9743255653186595e-07,
      "loss": 0.8539,
      "step": 4243
    },
    {
      "epoch": 0.9128844912884492,
      "grad_norm": 0.0,
      "learning_rate": 3.9549016406193686e-07,
      "loss": 0.7692,
      "step": 4244
    },
    {
      "epoch": 0.9130995913099591,
      "grad_norm": 0.0,
      "learning_rate": 3.935524340449615e-07,
      "loss": 0.7315,
      "step": 4245
    },
    {
      "epoch": 0.9133146913314691,
      "grad_norm": 0.0,
      "learning_rate": 3.916193674216029e-07,
      "loss": 0.7986,
      "step": 4246
    },
    {
      "epoch": 0.9135297913529792,
      "grad_norm": 0.0,
      "learning_rate": 3.8969096513025826e-07,
      "loss": 0.7818,
      "step": 4247
    },
    {
      "epoch": 0.9137448913744891,
      "grad_norm": 0.0,
      "learning_rate": 3.8776722810705436e-07,
      "loss": 0.8394,
      "step": 4248
    },
    {
      "epoch": 0.9139599913959991,
      "grad_norm": 0.0,
      "learning_rate": 3.85848157285863e-07,
      "loss": 0.7887,
      "step": 4249
    },
    {
      "epoch": 0.9141750914175092,
      "grad_norm": 0.0,
      "learning_rate": 3.839337535982812e-07,
      "loss": 0.847,
      "step": 4250
    },
    {
      "epoch": 0.9143901914390191,
      "grad_norm": 0.0,
      "learning_rate": 3.8202401797364896e-07,
      "loss": 0.8032,
      "step": 4251
    },
    {
      "epoch": 0.9146052914605292,
      "grad_norm": 0.0,
      "learning_rate": 3.801189513390335e-07,
      "loss": 0.8157,
      "step": 4252
    },
    {
      "epoch": 0.9148203914820392,
      "grad_norm": 0.0,
      "learning_rate": 3.7821855461924075e-07,
      "loss": 0.8074,
      "step": 4253
    },
    {
      "epoch": 0.9150354915035491,
      "grad_norm": 0.0,
      "learning_rate": 3.763228287368037e-07,
      "loss": 0.8222,
      "step": 4254
    },
    {
      "epoch": 0.9152505915250592,
      "grad_norm": 0.0,
      "learning_rate": 3.744317746119952e-07,
      "loss": 0.7964,
      "step": 4255
    },
    {
      "epoch": 0.9154656915465692,
      "grad_norm": 0.0,
      "learning_rate": 3.7254539316281424e-07,
      "loss": 0.7682,
      "step": 4256
    },
    {
      "epoch": 0.9156807915680791,
      "grad_norm": 0.0,
      "learning_rate": 3.7066368530499384e-07,
      "loss": 0.8052,
      "step": 4257
    },
    {
      "epoch": 0.9158958915895892,
      "grad_norm": 0.0,
      "learning_rate": 3.687866519519978e-07,
      "loss": 0.7966,
      "step": 4258
    },
    {
      "epoch": 0.9161109916110992,
      "grad_norm": 0.0,
      "learning_rate": 3.6691429401502055e-07,
      "loss": 0.724,
      "step": 4259
    },
    {
      "epoch": 0.9163260916326091,
      "grad_norm": 0.0,
      "learning_rate": 3.650466124029894e-07,
      "loss": 0.8302,
      "step": 4260
    },
    {
      "epoch": 0.9165411916541192,
      "grad_norm": 0.0,
      "learning_rate": 3.6318360802255703e-07,
      "loss": 0.7997,
      "step": 4261
    },
    {
      "epoch": 0.9167562916756292,
      "grad_norm": 0.0,
      "learning_rate": 3.61325281778111e-07,
      "loss": 0.8212,
      "step": 4262
    },
    {
      "epoch": 0.9169713916971391,
      "grad_norm": 0.0,
      "learning_rate": 3.594716345717608e-07,
      "loss": 0.8085,
      "step": 4263
    },
    {
      "epoch": 0.9171864917186492,
      "grad_norm": 0.0,
      "learning_rate": 3.576226673033512e-07,
      "loss": 0.7907,
      "step": 4264
    },
    {
      "epoch": 0.9174015917401592,
      "grad_norm": 0.0,
      "learning_rate": 3.557783808704529e-07,
      "loss": 0.7865,
      "step": 4265
    },
    {
      "epoch": 0.9176166917616692,
      "grad_norm": 0.0,
      "learning_rate": 3.5393877616836327e-07,
      "loss": 0.7946,
      "step": 4266
    },
    {
      "epoch": 0.9178317917831792,
      "grad_norm": 0.0,
      "learning_rate": 3.521038540901056e-07,
      "loss": 0.8091,
      "step": 4267
    },
    {
      "epoch": 0.9180468918046892,
      "grad_norm": 0.0,
      "learning_rate": 3.502736155264352e-07,
      "loss": 0.7733,
      "step": 4268
    },
    {
      "epoch": 0.9182619918261992,
      "grad_norm": 0.0,
      "learning_rate": 3.484480613658281e-07,
      "loss": 0.7767,
      "step": 4269
    },
    {
      "epoch": 0.9184770918477092,
      "grad_norm": 0.0,
      "learning_rate": 3.4662719249449196e-07,
      "loss": 0.802,
      "step": 4270
    },
    {
      "epoch": 0.9186921918692191,
      "grad_norm": 0.0,
      "learning_rate": 3.4481100979635304e-07,
      "loss": 0.7932,
      "step": 4271
    },
    {
      "epoch": 0.9189072918907292,
      "grad_norm": 0.0,
      "learning_rate": 3.429995141530695e-07,
      "loss": 0.865,
      "step": 4272
    },
    {
      "epoch": 0.9191223919122392,
      "grad_norm": 0.0,
      "learning_rate": 3.4119270644402015e-07,
      "loss": 0.8355,
      "step": 4273
    },
    {
      "epoch": 0.9193374919337491,
      "grad_norm": 0.0,
      "learning_rate": 3.393905875463088e-07,
      "loss": 0.8175,
      "step": 4274
    },
    {
      "epoch": 0.9195525919552592,
      "grad_norm": 0.0,
      "learning_rate": 3.375931583347647e-07,
      "loss": 0.7882,
      "step": 4275
    },
    {
      "epoch": 0.9197676919767692,
      "grad_norm": 0.0,
      "learning_rate": 3.3580041968193867e-07,
      "loss": 0.7872,
      "step": 4276
    },
    {
      "epoch": 0.9199827919982791,
      "grad_norm": 0.0,
      "learning_rate": 3.340123724581046e-07,
      "loss": 0.8211,
      "step": 4277
    },
    {
      "epoch": 0.9201978920197892,
      "grad_norm": 0.0,
      "learning_rate": 3.3222901753126037e-07,
      "loss": 0.8118,
      "step": 4278
    },
    {
      "epoch": 0.9204129920412992,
      "grad_norm": 0.0,
      "learning_rate": 3.3045035576712456e-07,
      "loss": 0.8198,
      "step": 4279
    },
    {
      "epoch": 0.9206280920628092,
      "grad_norm": 0.0,
      "learning_rate": 3.286763880291399e-07,
      "loss": 0.7968,
      "step": 4280
    },
    {
      "epoch": 0.9208431920843192,
      "grad_norm": 0.0,
      "learning_rate": 3.269071151784664e-07,
      "loss": 0.7719,
      "step": 4281
    },
    {
      "epoch": 0.9210582921058292,
      "grad_norm": 0.0,
      "learning_rate": 3.25142538073987e-07,
      "loss": 0.8361,
      "step": 4282
    },
    {
      "epoch": 0.9212733921273392,
      "grad_norm": 0.0,
      "learning_rate": 3.2338265757230647e-07,
      "loss": 0.8345,
      "step": 4283
    },
    {
      "epoch": 0.9214884921488492,
      "grad_norm": 0.0,
      "learning_rate": 3.216274745277481e-07,
      "loss": 0.811,
      "step": 4284
    },
    {
      "epoch": 0.9217035921703592,
      "grad_norm": 0.0,
      "learning_rate": 3.198769897923548e-07,
      "loss": 0.7996,
      "step": 4285
    },
    {
      "epoch": 0.9219186921918692,
      "grad_norm": 0.0,
      "learning_rate": 3.1813120421589126e-07,
      "loss": 0.805,
      "step": 4286
    },
    {
      "epoch": 0.9221337922133792,
      "grad_norm": 0.0,
      "learning_rate": 3.1639011864583625e-07,
      "loss": 0.7861,
      "step": 4287
    },
    {
      "epoch": 0.9223488922348893,
      "grad_norm": 0.0,
      "learning_rate": 3.146537339273914e-07,
      "loss": 0.7956,
      "step": 4288
    },
    {
      "epoch": 0.9225639922563992,
      "grad_norm": 0.0,
      "learning_rate": 3.129220509034725e-07,
      "loss": 0.8571,
      "step": 4289
    },
    {
      "epoch": 0.9227790922779092,
      "grad_norm": 0.0,
      "learning_rate": 3.1119507041471595e-07,
      "loss": 0.7926,
      "step": 4290
    },
    {
      "epoch": 0.9229941922994193,
      "grad_norm": 0.0,
      "learning_rate": 3.0947279329947564e-07,
      "loss": 0.7925,
      "step": 4291
    },
    {
      "epoch": 0.9232092923209292,
      "grad_norm": 0.0,
      "learning_rate": 3.0775522039381724e-07,
      "loss": 0.8116,
      "step": 4292
    },
    {
      "epoch": 0.9234243923424392,
      "grad_norm": 0.0,
      "learning_rate": 3.0604235253153145e-07,
      "loss": 0.8038,
      "step": 4293
    },
    {
      "epoch": 0.9236394923639493,
      "grad_norm": 0.0,
      "learning_rate": 3.043341905441155e-07,
      "loss": 0.8493,
      "step": 4294
    },
    {
      "epoch": 0.9238545923854592,
      "grad_norm": 0.0,
      "learning_rate": 3.0263073526079046e-07,
      "loss": 0.8447,
      "step": 4295
    },
    {
      "epoch": 0.9240696924069692,
      "grad_norm": 0.0,
      "learning_rate": 3.0093198750848483e-07,
      "loss": 0.8189,
      "step": 4296
    },
    {
      "epoch": 0.9242847924284793,
      "grad_norm": 0.0,
      "learning_rate": 2.9923794811184794e-07,
      "loss": 0.7559,
      "step": 4297
    },
    {
      "epoch": 0.9244998924499892,
      "grad_norm": 0.0,
      "learning_rate": 2.9754861789324074e-07,
      "loss": 0.8318,
      "step": 4298
    },
    {
      "epoch": 0.9247149924714992,
      "grad_norm": 0.0,
      "learning_rate": 2.9586399767273956e-07,
      "loss": 0.8647,
      "step": 4299
    },
    {
      "epoch": 0.9249300924930093,
      "grad_norm": 0.0,
      "learning_rate": 2.9418408826813125e-07,
      "loss": 0.8733,
      "step": 4300
    },
    {
      "epoch": 0.9251451925145192,
      "grad_norm": 0.0,
      "learning_rate": 2.9250889049492136e-07,
      "loss": 0.8184,
      "step": 4301
    },
    {
      "epoch": 0.9253602925360292,
      "grad_norm": 0.0,
      "learning_rate": 2.9083840516632155e-07,
      "loss": 0.79,
      "step": 4302
    },
    {
      "epoch": 0.9255753925575393,
      "grad_norm": 0.0,
      "learning_rate": 2.891726330932598e-07,
      "loss": 0.8054,
      "step": 4303
    },
    {
      "epoch": 0.9257904925790492,
      "grad_norm": 0.0,
      "learning_rate": 2.875115750843771e-07,
      "loss": 0.7541,
      "step": 4304
    },
    {
      "epoch": 0.9260055926005593,
      "grad_norm": 0.0,
      "learning_rate": 2.8585523194602283e-07,
      "loss": 0.7893,
      "step": 4305
    },
    {
      "epoch": 0.9262206926220693,
      "grad_norm": 0.0,
      "learning_rate": 2.842036044822605e-07,
      "loss": 0.7887,
      "step": 4306
    },
    {
      "epoch": 0.9264357926435792,
      "grad_norm": 0.0,
      "learning_rate": 2.825566934948598e-07,
      "loss": 0.7705,
      "step": 4307
    },
    {
      "epoch": 0.9266508926650893,
      "grad_norm": 0.0,
      "learning_rate": 2.8091449978330685e-07,
      "loss": 0.8365,
      "step": 4308
    },
    {
      "epoch": 0.9268659926865993,
      "grad_norm": 0.0,
      "learning_rate": 2.792770241447962e-07,
      "loss": 0.7824,
      "step": 4309
    },
    {
      "epoch": 0.9270810927081092,
      "grad_norm": 0.0,
      "learning_rate": 2.776442673742274e-07,
      "loss": 0.8426,
      "step": 4310
    },
    {
      "epoch": 0.9272961927296193,
      "grad_norm": 0.0,
      "learning_rate": 2.7601623026421445e-07,
      "loss": 0.8674,
      "step": 4311
    },
    {
      "epoch": 0.9275112927511293,
      "grad_norm": 0.0,
      "learning_rate": 2.7439291360507824e-07,
      "loss": 0.7953,
      "step": 4312
    },
    {
      "epoch": 0.9277263927726392,
      "grad_norm": 0.0,
      "learning_rate": 2.727743181848463e-07,
      "loss": 0.7899,
      "step": 4313
    },
    {
      "epoch": 0.9279414927941493,
      "grad_norm": 0.0,
      "learning_rate": 2.711604447892602e-07,
      "loss": 0.8423,
      "step": 4314
    },
    {
      "epoch": 0.9281565928156593,
      "grad_norm": 0.0,
      "learning_rate": 2.6955129420176193e-07,
      "loss": 0.8094,
      "step": 4315
    },
    {
      "epoch": 0.9283716928371692,
      "grad_norm": 0.0,
      "learning_rate": 2.6794686720350573e-07,
      "loss": 0.8859,
      "step": 4316
    },
    {
      "epoch": 0.9285867928586793,
      "grad_norm": 0.0,
      "learning_rate": 2.6634716457335177e-07,
      "loss": 0.8587,
      "step": 4317
    },
    {
      "epoch": 0.9288018928801893,
      "grad_norm": 0.0,
      "learning_rate": 2.6475218708786445e-07,
      "loss": 0.8471,
      "step": 4318
    },
    {
      "epoch": 0.9290169929016993,
      "grad_norm": 0.0,
      "learning_rate": 2.631619355213189e-07,
      "loss": 0.8288,
      "step": 4319
    },
    {
      "epoch": 0.9292320929232093,
      "grad_norm": 0.0,
      "learning_rate": 2.615764106456897e-07,
      "loss": 0.766,
      "step": 4320
    },
    {
      "epoch": 0.9294471929447193,
      "grad_norm": 0.0,
      "learning_rate": 2.5999561323066356e-07,
      "loss": 0.7892,
      "step": 4321
    },
    {
      "epoch": 0.9296622929662293,
      "grad_norm": 0.0,
      "learning_rate": 2.584195440436299e-07,
      "loss": 0.8235,
      "step": 4322
    },
    {
      "epoch": 0.9298773929877393,
      "grad_norm": 0.0,
      "learning_rate": 2.568482038496789e-07,
      "loss": 0.7996,
      "step": 4323
    },
    {
      "epoch": 0.9300924930092493,
      "grad_norm": 0.0,
      "learning_rate": 2.5528159341161264e-07,
      "loss": 0.789,
      "step": 4324
    },
    {
      "epoch": 0.9303075930307593,
      "grad_norm": 0.0,
      "learning_rate": 2.537197134899294e-07,
      "loss": 0.7742,
      "step": 4325
    },
    {
      "epoch": 0.9305226930522693,
      "grad_norm": 0.0,
      "learning_rate": 2.521625648428361e-07,
      "loss": 0.7958,
      "step": 4326
    },
    {
      "epoch": 0.9307377930737794,
      "grad_norm": 0.0,
      "learning_rate": 2.506101482262424e-07,
      "loss": 0.797,
      "step": 4327
    },
    {
      "epoch": 0.9309528930952893,
      "grad_norm": 0.0,
      "learning_rate": 2.490624643937567e-07,
      "loss": 0.7892,
      "step": 4328
    },
    {
      "epoch": 0.9311679931167993,
      "grad_norm": 0.0,
      "learning_rate": 2.475195140966957e-07,
      "loss": 0.7658,
      "step": 4329
    },
    {
      "epoch": 0.9313830931383094,
      "grad_norm": 0.0,
      "learning_rate": 2.459812980840759e-07,
      "loss": 0.7597,
      "step": 4330
    },
    {
      "epoch": 0.9315981931598193,
      "grad_norm": 0.0,
      "learning_rate": 2.4444781710261324e-07,
      "loss": 0.8205,
      "step": 4331
    },
    {
      "epoch": 0.9318132931813293,
      "grad_norm": 0.0,
      "learning_rate": 2.429190718967289e-07,
      "loss": 0.83,
      "step": 4332
    },
    {
      "epoch": 0.9320283932028394,
      "grad_norm": 0.0,
      "learning_rate": 2.4139506320854154e-07,
      "loss": 0.8226,
      "step": 4333
    },
    {
      "epoch": 0.9322434932243493,
      "grad_norm": 0.0,
      "learning_rate": 2.398757917778727e-07,
      "loss": 0.8127,
      "step": 4334
    },
    {
      "epoch": 0.9324585932458593,
      "grad_norm": 0.0,
      "learning_rate": 2.3836125834224566e-07,
      "loss": 0.7189,
      "step": 4335
    },
    {
      "epoch": 0.9326736932673694,
      "grad_norm": 0.0,
      "learning_rate": 2.3685146363688018e-07,
      "loss": 0.8453,
      "step": 4336
    },
    {
      "epoch": 0.9328887932888793,
      "grad_norm": 0.0,
      "learning_rate": 2.3534640839469768e-07,
      "loss": 0.7494,
      "step": 4337
    },
    {
      "epoch": 0.9331038933103893,
      "grad_norm": 0.0,
      "learning_rate": 2.3384609334631713e-07,
      "loss": 0.7766,
      "step": 4338
    },
    {
      "epoch": 0.9333189933318994,
      "grad_norm": 0.0,
      "learning_rate": 2.3235051922005926e-07,
      "loss": 0.7919,
      "step": 4339
    },
    {
      "epoch": 0.9335340933534093,
      "grad_norm": 0.0,
      "learning_rate": 2.3085968674194214e-07,
      "loss": 0.8438,
      "step": 4340
    },
    {
      "epoch": 0.9337491933749194,
      "grad_norm": 0.0,
      "learning_rate": 2.2937359663567916e-07,
      "loss": 0.6956,
      "step": 4341
    },
    {
      "epoch": 0.9339642933964294,
      "grad_norm": 0.0,
      "learning_rate": 2.278922496226854e-07,
      "loss": 0.8082,
      "step": 4342
    },
    {
      "epoch": 0.9341793934179393,
      "grad_norm": 0.0,
      "learning_rate": 2.2641564642207347e-07,
      "loss": 0.7913,
      "step": 4343
    },
    {
      "epoch": 0.9343944934394494,
      "grad_norm": 0.0,
      "learning_rate": 2.2494378775064997e-07,
      "loss": 0.7968,
      "step": 4344
    },
    {
      "epoch": 0.9346095934609594,
      "grad_norm": 0.0,
      "learning_rate": 2.2347667432292109e-07,
      "loss": 0.8354,
      "step": 4345
    },
    {
      "epoch": 0.9348246934824693,
      "grad_norm": 0.0,
      "learning_rate": 2.2201430685108715e-07,
      "loss": 0.7958,
      "step": 4346
    },
    {
      "epoch": 0.9350397935039794,
      "grad_norm": 0.0,
      "learning_rate": 2.20556686045047e-07,
      "loss": 0.7768,
      "step": 4347
    },
    {
      "epoch": 0.9352548935254894,
      "grad_norm": 0.0,
      "learning_rate": 2.1910381261239456e-07,
      "loss": 0.777,
      "step": 4348
    },
    {
      "epoch": 0.9354699935469993,
      "grad_norm": 0.0,
      "learning_rate": 2.1765568725841679e-07,
      "loss": 0.825,
      "step": 4349
    },
    {
      "epoch": 0.9356850935685094,
      "grad_norm": 0.0,
      "learning_rate": 2.1621231068610028e-07,
      "loss": 0.7826,
      "step": 4350
    },
    {
      "epoch": 0.9359001935900194,
      "grad_norm": 0.0,
      "learning_rate": 2.1477368359612117e-07,
      "loss": 0.8391,
      "step": 4351
    },
    {
      "epoch": 0.9361152936115293,
      "grad_norm": 0.0,
      "learning_rate": 2.1333980668685416e-07,
      "loss": 0.8354,
      "step": 4352
    },
    {
      "epoch": 0.9363303936330394,
      "grad_norm": 0.0,
      "learning_rate": 2.1191068065436472e-07,
      "loss": 0.8087,
      "step": 4353
    },
    {
      "epoch": 0.9365454936545493,
      "grad_norm": 0.0,
      "learning_rate": 2.1048630619241563e-07,
      "loss": 0.8986,
      "step": 4354
    },
    {
      "epoch": 0.9367605936760593,
      "grad_norm": 0.0,
      "learning_rate": 2.0906668399246045e-07,
      "loss": 0.844,
      "step": 4355
    },
    {
      "epoch": 0.9369756936975694,
      "grad_norm": 0.0,
      "learning_rate": 2.0765181474364572e-07,
      "loss": 0.7914,
      "step": 4356
    },
    {
      "epoch": 0.9371907937190793,
      "grad_norm": 0.0,
      "learning_rate": 2.06241699132812e-07,
      "loss": 0.8216,
      "step": 4357
    },
    {
      "epoch": 0.9374058937405894,
      "grad_norm": 0.0,
      "learning_rate": 2.0483633784449285e-07,
      "loss": 0.7929,
      "step": 4358
    },
    {
      "epoch": 0.9376209937620994,
      "grad_norm": 0.0,
      "learning_rate": 2.0343573156091035e-07,
      "loss": 0.7977,
      "step": 4359
    },
    {
      "epoch": 0.9378360937836093,
      "grad_norm": 0.0,
      "learning_rate": 2.020398809619817e-07,
      "loss": 0.8777,
      "step": 4360
    },
    {
      "epoch": 0.9380511938051194,
      "grad_norm": 0.0,
      "learning_rate": 2.0064878672531485e-07,
      "loss": 0.7604,
      "step": 4361
    },
    {
      "epoch": 0.9382662938266294,
      "grad_norm": 0.0,
      "learning_rate": 1.9926244952620743e-07,
      "loss": 0.8351,
      "step": 4362
    },
    {
      "epoch": 0.9384813938481393,
      "grad_norm": 0.0,
      "learning_rate": 1.978808700376489e-07,
      "loss": 0.7879,
      "step": 4363
    },
    {
      "epoch": 0.9386964938696494,
      "grad_norm": 0.0,
      "learning_rate": 1.9650404893031938e-07,
      "loss": 0.7998,
      "step": 4364
    },
    {
      "epoch": 0.9389115938911594,
      "grad_norm": 0.0,
      "learning_rate": 1.9513198687258873e-07,
      "loss": 0.8151,
      "step": 4365
    },
    {
      "epoch": 0.9391266939126693,
      "grad_norm": 0.0,
      "learning_rate": 1.9376468453051633e-07,
      "loss": 0.8546,
      "step": 4366
    },
    {
      "epoch": 0.9393417939341794,
      "grad_norm": 0.0,
      "learning_rate": 1.9240214256785128e-07,
      "loss": 0.8536,
      "step": 4367
    },
    {
      "epoch": 0.9395568939556894,
      "grad_norm": 0.0,
      "learning_rate": 1.9104436164602892e-07,
      "loss": 0.7866,
      "step": 4368
    },
    {
      "epoch": 0.9397719939771993,
      "grad_norm": 0.0,
      "learning_rate": 1.8969134242418086e-07,
      "loss": 0.8532,
      "step": 4369
    },
    {
      "epoch": 0.9399870939987094,
      "grad_norm": 0.0,
      "learning_rate": 1.883430855591173e-07,
      "loss": 0.8375,
      "step": 4370
    },
    {
      "epoch": 0.9402021940202194,
      "grad_norm": 0.0,
      "learning_rate": 1.8699959170534577e-07,
      "loss": 0.8233,
      "step": 4371
    },
    {
      "epoch": 0.9404172940417294,
      "grad_norm": 0.0,
      "learning_rate": 1.8566086151505568e-07,
      "loss": 0.8074,
      "step": 4372
    },
    {
      "epoch": 0.9406323940632394,
      "grad_norm": 0.0,
      "learning_rate": 1.84326895638125e-07,
      "loss": 0.7656,
      "step": 4373
    },
    {
      "epoch": 0.9408474940847494,
      "grad_norm": 0.0,
      "learning_rate": 1.8299769472212125e-07,
      "loss": 0.7247,
      "step": 4374
    },
    {
      "epoch": 0.9410625941062594,
      "grad_norm": 0.0,
      "learning_rate": 1.8167325941229496e-07,
      "loss": 0.7518,
      "step": 4375
    },
    {
      "epoch": 0.9412776941277694,
      "grad_norm": 0.0,
      "learning_rate": 1.803535903515885e-07,
      "loss": 0.8295,
      "step": 4376
    },
    {
      "epoch": 0.9414927941492794,
      "grad_norm": 0.0,
      "learning_rate": 1.790386881806261e-07,
      "loss": 0.8591,
      "step": 4377
    },
    {
      "epoch": 0.9417078941707894,
      "grad_norm": 0.0,
      "learning_rate": 1.7772855353771823e-07,
      "loss": 0.8226,
      "step": 4378
    },
    {
      "epoch": 0.9419229941922994,
      "grad_norm": 0.0,
      "learning_rate": 1.7642318705886284e-07,
      "loss": 0.7837,
      "step": 4379
    },
    {
      "epoch": 0.9421380942138095,
      "grad_norm": 0.0,
      "learning_rate": 1.7512258937774307e-07,
      "loss": 0.8499,
      "step": 4380
    },
    {
      "epoch": 0.9423531942353194,
      "grad_norm": 0.0,
      "learning_rate": 1.7382676112572605e-07,
      "loss": 0.8081,
      "step": 4381
    },
    {
      "epoch": 0.9425682942568294,
      "grad_norm": 0.0,
      "learning_rate": 1.7253570293186418e-07,
      "loss": 0.8097,
      "step": 4382
    },
    {
      "epoch": 0.9427833942783395,
      "grad_norm": 0.0,
      "learning_rate": 1.7124941542289276e-07,
      "loss": 0.8137,
      "step": 4383
    },
    {
      "epoch": 0.9429984942998494,
      "grad_norm": 0.0,
      "learning_rate": 1.6996789922323454e-07,
      "loss": 0.7784,
      "step": 4384
    },
    {
      "epoch": 0.9432135943213594,
      "grad_norm": 0.0,
      "learning_rate": 1.68691154954993e-07,
      "loss": 0.7903,
      "step": 4385
    },
    {
      "epoch": 0.9434286943428695,
      "grad_norm": 0.0,
      "learning_rate": 1.6741918323795458e-07,
      "loss": 0.744,
      "step": 4386
    },
    {
      "epoch": 0.9436437943643794,
      "grad_norm": 0.0,
      "learning_rate": 1.6615198468959426e-07,
      "loss": 0.8748,
      "step": 4387
    },
    {
      "epoch": 0.9438588943858894,
      "grad_norm": 0.0,
      "learning_rate": 1.6488955992506218e-07,
      "loss": 0.8597,
      "step": 4388
    },
    {
      "epoch": 0.9440739944073995,
      "grad_norm": 0.0,
      "learning_rate": 1.6363190955719922e-07,
      "loss": 0.8796,
      "step": 4389
    },
    {
      "epoch": 0.9442890944289094,
      "grad_norm": 0.0,
      "learning_rate": 1.6237903419651923e-07,
      "loss": 0.8066,
      "step": 4390
    },
    {
      "epoch": 0.9445041944504194,
      "grad_norm": 0.0,
      "learning_rate": 1.6113093445122685e-07,
      "loss": 0.7905,
      "step": 4391
    },
    {
      "epoch": 0.9447192944719295,
      "grad_norm": 0.0,
      "learning_rate": 1.5988761092720517e-07,
      "loss": 0.837,
      "step": 4392
    },
    {
      "epoch": 0.9449343944934394,
      "grad_norm": 0.0,
      "learning_rate": 1.5864906422801584e-07,
      "loss": 0.7734,
      "step": 4393
    },
    {
      "epoch": 0.9451494945149495,
      "grad_norm": 0.0,
      "learning_rate": 1.574152949549057e-07,
      "loss": 0.7933,
      "step": 4394
    },
    {
      "epoch": 0.9453645945364595,
      "grad_norm": 0.0,
      "learning_rate": 1.561863037068001e-07,
      "loss": 0.7818,
      "step": 4395
    },
    {
      "epoch": 0.9455796945579694,
      "grad_norm": 0.0,
      "learning_rate": 1.5496209108030735e-07,
      "loss": 0.8053,
      "step": 4396
    },
    {
      "epoch": 0.9457947945794795,
      "grad_norm": 0.0,
      "learning_rate": 1.5374265766971097e-07,
      "loss": 0.8403,
      "step": 4397
    },
    {
      "epoch": 0.9460098946009895,
      "grad_norm": 0.0,
      "learning_rate": 1.5252800406698076e-07,
      "loss": 0.7811,
      "step": 4398
    },
    {
      "epoch": 0.9462249946224994,
      "grad_norm": 0.0,
      "learning_rate": 1.5131813086176063e-07,
      "loss": 0.8032,
      "step": 4399
    },
    {
      "epoch": 0.9464400946440095,
      "grad_norm": 0.0,
      "learning_rate": 1.5011303864137962e-07,
      "loss": 0.8645,
      "step": 4400
    },
    {
      "epoch": 0.9466551946655195,
      "grad_norm": 0.0,
      "learning_rate": 1.4891272799083979e-07,
      "loss": 0.8273,
      "step": 4401
    },
    {
      "epoch": 0.9468702946870294,
      "grad_norm": 0.0,
      "learning_rate": 1.4771719949282725e-07,
      "loss": 0.7975,
      "step": 4402
    },
    {
      "epoch": 0.9470853947085395,
      "grad_norm": 0.0,
      "learning_rate": 1.4652645372770226e-07,
      "loss": 0.7821,
      "step": 4403
    },
    {
      "epoch": 0.9473004947300495,
      "grad_norm": 0.0,
      "learning_rate": 1.4534049127350568e-07,
      "loss": 0.8101,
      "step": 4404
    },
    {
      "epoch": 0.9475155947515594,
      "grad_norm": 0.0,
      "learning_rate": 1.4415931270595708e-07,
      "loss": 0.7644,
      "step": 4405
    },
    {
      "epoch": 0.9477306947730695,
      "grad_norm": 0.0,
      "learning_rate": 1.4298291859845215e-07,
      "loss": 0.8539,
      "step": 4406
    },
    {
      "epoch": 0.9479457947945795,
      "grad_norm": 0.0,
      "learning_rate": 1.4181130952206524e-07,
      "loss": 0.8091,
      "step": 4407
    },
    {
      "epoch": 0.9481608948160895,
      "grad_norm": 0.0,
      "learning_rate": 1.406444860455447e-07,
      "loss": 0.7821,
      "step": 4408
    },
    {
      "epoch": 0.9483759948375995,
      "grad_norm": 0.0,
      "learning_rate": 1.394824487353208e-07,
      "loss": 0.7887,
      "step": 4409
    },
    {
      "epoch": 0.9485910948591095,
      "grad_norm": 0.0,
      "learning_rate": 1.3832519815549671e-07,
      "loss": 0.8039,
      "step": 4410
    },
    {
      "epoch": 0.9488061948806195,
      "grad_norm": 0.0,
      "learning_rate": 1.371727348678531e-07,
      "loss": 0.822,
      "step": 4411
    },
    {
      "epoch": 0.9490212949021295,
      "grad_norm": 0.0,
      "learning_rate": 1.3602505943184575e-07,
      "loss": 0.8456,
      "step": 4412
    },
    {
      "epoch": 0.9492363949236395,
      "grad_norm": 0.0,
      "learning_rate": 1.3488217240460788e-07,
      "loss": 0.7968,
      "step": 4413
    },
    {
      "epoch": 0.9494514949451495,
      "grad_norm": 0.0,
      "learning_rate": 1.3374407434094682e-07,
      "loss": 0.8276,
      "step": 4414
    },
    {
      "epoch": 0.9496665949666595,
      "grad_norm": 0.0,
      "learning_rate": 1.326107657933473e-07,
      "loss": 0.8872,
      "step": 4415
    },
    {
      "epoch": 0.9498816949881695,
      "grad_norm": 0.0,
      "learning_rate": 1.3148224731196367e-07,
      "loss": 0.8576,
      "step": 4416
    },
    {
      "epoch": 0.9500967950096795,
      "grad_norm": 0.0,
      "learning_rate": 1.3035851944463106e-07,
      "loss": 0.7301,
      "step": 4417
    },
    {
      "epoch": 0.9503118950311895,
      "grad_norm": 0.0,
      "learning_rate": 1.2923958273685755e-07,
      "loss": 0.8582,
      "step": 4418
    },
    {
      "epoch": 0.9505269950526996,
      "grad_norm": 0.0,
      "learning_rate": 1.2812543773182195e-07,
      "loss": 0.797,
      "step": 4419
    },
    {
      "epoch": 0.9507420950742095,
      "grad_norm": 0.0,
      "learning_rate": 1.2701608497038055e-07,
      "loss": 0.8463,
      "step": 4420
    },
    {
      "epoch": 0.9509571950957195,
      "grad_norm": 0.0,
      "learning_rate": 1.2591152499106253e-07,
      "loss": 0.7557,
      "step": 4421
    },
    {
      "epoch": 0.9511722951172296,
      "grad_norm": 0.0,
      "learning_rate": 1.2481175833006898e-07,
      "loss": 0.7777,
      "step": 4422
    },
    {
      "epoch": 0.9513873951387395,
      "grad_norm": 0.0,
      "learning_rate": 1.237167855212773e-07,
      "loss": 0.7364,
      "step": 4423
    },
    {
      "epoch": 0.9516024951602495,
      "grad_norm": 0.0,
      "learning_rate": 1.2262660709623232e-07,
      "loss": 0.7885,
      "step": 4424
    },
    {
      "epoch": 0.9518175951817596,
      "grad_norm": 0.0,
      "learning_rate": 1.2154122358415842e-07,
      "loss": 0.7781,
      "step": 4425
    },
    {
      "epoch": 0.9520326952032695,
      "grad_norm": 0.0,
      "learning_rate": 1.2046063551194643e-07,
      "loss": 0.7535,
      "step": 4426
    },
    {
      "epoch": 0.9522477952247795,
      "grad_norm": 0.0,
      "learning_rate": 1.1938484340416113e-07,
      "loss": 0.8175,
      "step": 4427
    },
    {
      "epoch": 0.9524628952462896,
      "grad_norm": 0.0,
      "learning_rate": 1.1831384778304033e-07,
      "loss": 0.7991,
      "step": 4428
    },
    {
      "epoch": 0.9526779952677995,
      "grad_norm": 0.0,
      "learning_rate": 1.1724764916849263e-07,
      "loss": 0.8475,
      "step": 4429
    },
    {
      "epoch": 0.9528930952893095,
      "grad_norm": 0.0,
      "learning_rate": 1.1618624807809841e-07,
      "loss": 0.7924,
      "step": 4430
    },
    {
      "epoch": 0.9531081953108196,
      "grad_norm": 0.0,
      "learning_rate": 1.1512964502710778e-07,
      "loss": 0.7432,
      "step": 4431
    },
    {
      "epoch": 0.9533232953323295,
      "grad_norm": 0.0,
      "learning_rate": 1.140778405284415e-07,
      "loss": 0.7831,
      "step": 4432
    },
    {
      "epoch": 0.9535383953538396,
      "grad_norm": 0.0,
      "learning_rate": 1.1303083509269452e-07,
      "loss": 0.8146,
      "step": 4433
    },
    {
      "epoch": 0.9537534953753496,
      "grad_norm": 0.0,
      "learning_rate": 1.1198862922812693e-07,
      "loss": 0.7749,
      "step": 4434
    },
    {
      "epoch": 0.9539685953968595,
      "grad_norm": 0.0,
      "learning_rate": 1.1095122344067289e-07,
      "loss": 0.8212,
      "step": 4435
    },
    {
      "epoch": 0.9541836954183696,
      "grad_norm": 0.0,
      "learning_rate": 1.0991861823393513e-07,
      "loss": 0.825,
      "step": 4436
    },
    {
      "epoch": 0.9543987954398796,
      "grad_norm": 0.0,
      "learning_rate": 1.088908141091849e-07,
      "loss": 0.833,
      "step": 4437
    },
    {
      "epoch": 0.9546138954613895,
      "grad_norm": 0.0,
      "learning_rate": 1.0786781156536418e-07,
      "loss": 0.7561,
      "step": 4438
    },
    {
      "epoch": 0.9548289954828996,
      "grad_norm": 0.0,
      "learning_rate": 1.0684961109908354e-07,
      "loss": 0.8169,
      "step": 4439
    },
    {
      "epoch": 0.9550440955044095,
      "grad_norm": 0.0,
      "learning_rate": 1.0583621320462201e-07,
      "loss": 0.7723,
      "step": 4440
    },
    {
      "epoch": 0.9552591955259195,
      "grad_norm": 0.0,
      "learning_rate": 1.048276183739283e-07,
      "loss": 0.8317,
      "step": 4441
    },
    {
      "epoch": 0.9554742955474296,
      "grad_norm": 0.0,
      "learning_rate": 1.0382382709661853e-07,
      "loss": 0.8565,
      "step": 4442
    },
    {
      "epoch": 0.9556893955689395,
      "grad_norm": 0.0,
      "learning_rate": 1.0282483985997627e-07,
      "loss": 0.7884,
      "step": 4443
    },
    {
      "epoch": 0.9559044955904495,
      "grad_norm": 0.0,
      "learning_rate": 1.018306571489569e-07,
      "loss": 0.824,
      "step": 4444
    },
    {
      "epoch": 0.9561195956119596,
      "grad_norm": 0.0,
      "learning_rate": 1.008412794461766e-07,
      "loss": 0.7593,
      "step": 4445
    },
    {
      "epoch": 0.9563346956334695,
      "grad_norm": 0.0,
      "learning_rate": 9.985670723192675e-08,
      "loss": 0.7813,
      "step": 4446
    },
    {
      "epoch": 0.9565497956549796,
      "grad_norm": 0.0,
      "learning_rate": 9.887694098415946e-08,
      "loss": 0.7934,
      "step": 4447
    },
    {
      "epoch": 0.9567648956764896,
      "grad_norm": 0.0,
      "learning_rate": 9.790198117849758e-08,
      "loss": 0.8128,
      "step": 4448
    },
    {
      "epoch": 0.9569799956979995,
      "grad_norm": 0.0,
      "learning_rate": 9.693182828823034e-08,
      "loss": 0.7936,
      "step": 4449
    },
    {
      "epoch": 0.9571950957195096,
      "grad_norm": 0.0,
      "learning_rate": 9.596648278431209e-08,
      "loss": 0.8732,
      "step": 4450
    },
    {
      "epoch": 0.9574101957410196,
      "grad_norm": 0.0,
      "learning_rate": 9.500594513536354e-08,
      "loss": 0.7803,
      "step": 4451
    },
    {
      "epoch": 0.9576252957625295,
      "grad_norm": 0.0,
      "learning_rate": 9.405021580767171e-08,
      "loss": 0.774,
      "step": 4452
    },
    {
      "epoch": 0.9578403957840396,
      "grad_norm": 0.0,
      "learning_rate": 9.309929526519212e-08,
      "loss": 0.8055,
      "step": 4453
    },
    {
      "epoch": 0.9580554958055496,
      "grad_norm": 0.0,
      "learning_rate": 9.215318396954109e-08,
      "loss": 0.8656,
      "step": 4454
    },
    {
      "epoch": 0.9582705958270595,
      "grad_norm": 0.0,
      "learning_rate": 9.121188238000456e-08,
      "loss": 0.8178,
      "step": 4455
    },
    {
      "epoch": 0.9584856958485696,
      "grad_norm": 0.0,
      "learning_rate": 9.027539095353144e-08,
      "loss": 0.7503,
      "step": 4456
    },
    {
      "epoch": 0.9587007958700796,
      "grad_norm": 0.0,
      "learning_rate": 8.934371014473475e-08,
      "loss": 0.7969,
      "step": 4457
    },
    {
      "epoch": 0.9589158958915895,
      "grad_norm": 0.0,
      "learning_rate": 8.841684040589494e-08,
      "loss": 0.8282,
      "step": 4458
    },
    {
      "epoch": 0.9591309959130996,
      "grad_norm": 0.0,
      "learning_rate": 8.749478218695429e-08,
      "loss": 0.7865,
      "step": 4459
    },
    {
      "epoch": 0.9593460959346096,
      "grad_norm": 0.0,
      "learning_rate": 8.657753593552142e-08,
      "loss": 0.7417,
      "step": 4460
    },
    {
      "epoch": 0.9595611959561196,
      "grad_norm": 0.0,
      "learning_rate": 8.566510209686796e-08,
      "loss": 0.7899,
      "step": 4461
    },
    {
      "epoch": 0.9597762959776296,
      "grad_norm": 0.0,
      "learning_rate": 8.475748111393068e-08,
      "loss": 0.794,
      "step": 4462
    },
    {
      "epoch": 0.9599913959991396,
      "grad_norm": 0.0,
      "learning_rate": 8.385467342730602e-08,
      "loss": 0.7789,
      "step": 4463
    },
    {
      "epoch": 0.9602064960206496,
      "grad_norm": 0.0,
      "learning_rate": 8.29566794752601e-08,
      "loss": 0.8641,
      "step": 4464
    },
    {
      "epoch": 0.9604215960421596,
      "grad_norm": 0.0,
      "learning_rate": 8.20634996937153e-08,
      "loss": 0.8061,
      "step": 4465
    },
    {
      "epoch": 0.9606366960636696,
      "grad_norm": 0.0,
      "learning_rate": 8.117513451626258e-08,
      "loss": 0.7789,
      "step": 4466
    },
    {
      "epoch": 0.9608517960851796,
      "grad_norm": 0.0,
      "learning_rate": 8.029158437415474e-08,
      "loss": 0.8158,
      "step": 4467
    },
    {
      "epoch": 0.9610668961066896,
      "grad_norm": 0.0,
      "learning_rate": 7.941284969630314e-08,
      "loss": 0.7708,
      "step": 4468
    },
    {
      "epoch": 0.9612819961281996,
      "grad_norm": 0.0,
      "learning_rate": 7.853893090928654e-08,
      "loss": 0.7141,
      "step": 4469
    },
    {
      "epoch": 0.9614970961497096,
      "grad_norm": 0.0,
      "learning_rate": 7.766982843734339e-08,
      "loss": 0.8325,
      "step": 4470
    },
    {
      "epoch": 0.9617121961712196,
      "grad_norm": 0.0,
      "learning_rate": 7.680554270237283e-08,
      "loss": 0.8071,
      "step": 4471
    },
    {
      "epoch": 0.9619272961927297,
      "grad_norm": 0.0,
      "learning_rate": 7.594607412393928e-08,
      "loss": 0.8739,
      "step": 4472
    },
    {
      "epoch": 0.9621423962142396,
      "grad_norm": 0.0,
      "learning_rate": 7.509142311926565e-08,
      "loss": 0.8425,
      "step": 4473
    },
    {
      "epoch": 0.9623574962357496,
      "grad_norm": 0.0,
      "learning_rate": 7.424159010323784e-08,
      "loss": 0.8335,
      "step": 4474
    },
    {
      "epoch": 0.9625725962572597,
      "grad_norm": 0.0,
      "learning_rate": 7.33965754884014e-08,
      "loss": 0.8169,
      "step": 4475
    },
    {
      "epoch": 0.9627876962787696,
      "grad_norm": 0.0,
      "learning_rate": 7.255637968496487e-08,
      "loss": 0.745,
      "step": 4476
    },
    {
      "epoch": 0.9630027963002796,
      "grad_norm": 0.0,
      "learning_rate": 7.172100310079533e-08,
      "loss": 0.8217,
      "step": 4477
    },
    {
      "epoch": 0.9632178963217897,
      "grad_norm": 0.0,
      "learning_rate": 7.08904461414217e-08,
      "loss": 0.7923,
      "step": 4478
    },
    {
      "epoch": 0.9634329963432996,
      "grad_norm": 0.0,
      "learning_rate": 7.006470921003261e-08,
      "loss": 0.854,
      "step": 4479
    },
    {
      "epoch": 0.9636480963648096,
      "grad_norm": 0.0,
      "learning_rate": 6.92437927074796e-08,
      "loss": 0.7779,
      "step": 4480
    },
    {
      "epoch": 0.9638631963863197,
      "grad_norm": 0.0,
      "learning_rate": 6.842769703226836e-08,
      "loss": 0.8496,
      "step": 4481
    },
    {
      "epoch": 0.9640782964078296,
      "grad_norm": 0.0,
      "learning_rate": 6.761642258056977e-08,
      "loss": 0.8233,
      "step": 4482
    },
    {
      "epoch": 0.9642933964293396,
      "grad_norm": 0.0,
      "learning_rate": 6.680996974621212e-08,
      "loss": 0.8487,
      "step": 4483
    },
    {
      "epoch": 0.9645084964508497,
      "grad_norm": 0.0,
      "learning_rate": 6.600833892068336e-08,
      "loss": 0.8127,
      "step": 4484
    },
    {
      "epoch": 0.9647235964723596,
      "grad_norm": 0.0,
      "learning_rate": 6.521153049312889e-08,
      "loss": 0.6648,
      "step": 4485
    },
    {
      "epoch": 0.9649386964938697,
      "grad_norm": 0.0,
      "learning_rate": 6.441954485035707e-08,
      "loss": 0.7624,
      "step": 4486
    },
    {
      "epoch": 0.9651537965153797,
      "grad_norm": 0.0,
      "learning_rate": 6.363238237683034e-08,
      "loss": 0.7468,
      "step": 4487
    },
    {
      "epoch": 0.9653688965368896,
      "grad_norm": 0.0,
      "learning_rate": 6.285004345467305e-08,
      "loss": 0.8262,
      "step": 4488
    },
    {
      "epoch": 0.9655839965583997,
      "grad_norm": 0.0,
      "learning_rate": 6.207252846366807e-08,
      "loss": 0.7949,
      "step": 4489
    },
    {
      "epoch": 0.9657990965799097,
      "grad_norm": 0.0,
      "learning_rate": 6.129983778125348e-08,
      "loss": 0.8098,
      "step": 4490
    },
    {
      "epoch": 0.9660141966014196,
      "grad_norm": 0.0,
      "learning_rate": 6.053197178252923e-08,
      "loss": 0.779,
      "step": 4491
    },
    {
      "epoch": 0.9662292966229297,
      "grad_norm": 0.0,
      "learning_rate": 5.976893084025048e-08,
      "loss": 0.7781,
      "step": 4492
    },
    {
      "epoch": 0.9664443966444397,
      "grad_norm": 0.0,
      "learning_rate": 5.901071532483204e-08,
      "loss": 0.8476,
      "step": 4493
    },
    {
      "epoch": 0.9666594966659496,
      "grad_norm": 0.0,
      "learning_rate": 5.825732560434283e-08,
      "loss": 0.7867,
      "step": 4494
    },
    {
      "epoch": 0.9668745966874597,
      "grad_norm": 0.0,
      "learning_rate": 5.7508762044514724e-08,
      "loss": 0.8179,
      "step": 4495
    },
    {
      "epoch": 0.9670896967089697,
      "grad_norm": 0.0,
      "learning_rate": 5.676502500873038e-08,
      "loss": 0.8339,
      "step": 4496
    },
    {
      "epoch": 0.9673047967304796,
      "grad_norm": 0.0,
      "learning_rate": 5.602611485803433e-08,
      "loss": 0.8987,
      "step": 4497
    },
    {
      "epoch": 0.9675198967519897,
      "grad_norm": 0.0,
      "learning_rate": 5.529203195112631e-08,
      "loss": 0.7883,
      "step": 4498
    },
    {
      "epoch": 0.9677349967734997,
      "grad_norm": 0.0,
      "learning_rate": 5.456277664436127e-08,
      "loss": 0.894,
      "step": 4499
    },
    {
      "epoch": 0.9679500967950097,
      "grad_norm": 0.0,
      "learning_rate": 5.38383492917538e-08,
      "loss": 0.7084,
      "step": 4500
    },
    {
      "epoch": 0.9681651968165197,
      "grad_norm": 0.0,
      "learning_rate": 5.311875024497037e-08,
      "loss": 0.7535,
      "step": 4501
    },
    {
      "epoch": 0.9683802968380297,
      "grad_norm": 0.0,
      "learning_rate": 5.240397985333823e-08,
      "loss": 0.839,
      "step": 4502
    },
    {
      "epoch": 0.9685953968595397,
      "grad_norm": 0.0,
      "learning_rate": 5.1694038463838695e-08,
      "loss": 0.7117,
      "step": 4503
    },
    {
      "epoch": 0.9688104968810497,
      "grad_norm": 0.0,
      "learning_rate": 5.0988926421106086e-08,
      "loss": 0.8232,
      "step": 4504
    },
    {
      "epoch": 0.9690255969025597,
      "grad_norm": 0.0,
      "learning_rate": 5.0288644067435496e-08,
      "loss": 0.8127,
      "step": 4505
    },
    {
      "epoch": 0.9692406969240697,
      "grad_norm": 0.0,
      "learning_rate": 4.959319174277388e-08,
      "loss": 0.8261,
      "step": 4506
    },
    {
      "epoch": 0.9694557969455797,
      "grad_norm": 0.0,
      "learning_rate": 4.890256978472452e-08,
      "loss": 0.8089,
      "step": 4507
    },
    {
      "epoch": 0.9696708969670897,
      "grad_norm": 0.0,
      "learning_rate": 4.8216778528545895e-08,
      "loss": 0.8538,
      "step": 4508
    },
    {
      "epoch": 0.9698859969885997,
      "grad_norm": 0.0,
      "learning_rate": 4.753581830715171e-08,
      "loss": 0.7969,
      "step": 4509
    },
    {
      "epoch": 0.9701010970101097,
      "grad_norm": 0.0,
      "learning_rate": 4.685968945110864e-08,
      "loss": 0.79,
      "step": 4510
    },
    {
      "epoch": 0.9703161970316198,
      "grad_norm": 0.0,
      "learning_rate": 4.618839228864192e-08,
      "loss": 0.7731,
      "step": 4511
    },
    {
      "epoch": 0.9705312970531297,
      "grad_norm": 0.0,
      "learning_rate": 4.552192714562642e-08,
      "loss": 0.8501,
      "step": 4512
    },
    {
      "epoch": 0.9707463970746397,
      "grad_norm": 0.0,
      "learning_rate": 4.486029434559558e-08,
      "loss": 0.8021,
      "step": 4513
    },
    {
      "epoch": 0.9709614970961498,
      "grad_norm": 0.0,
      "learning_rate": 4.420349420973358e-08,
      "loss": 0.8187,
      "step": 4514
    },
    {
      "epoch": 0.9711765971176597,
      "grad_norm": 0.0,
      "learning_rate": 4.3551527056880926e-08,
      "loss": 0.7364,
      "step": 4515
    },
    {
      "epoch": 0.9713916971391697,
      "grad_norm": 0.0,
      "learning_rate": 4.290439320353224e-08,
      "loss": 0.8003,
      "step": 4516
    },
    {
      "epoch": 0.9716067971606798,
      "grad_norm": 0.0,
      "learning_rate": 4.226209296383288e-08,
      "loss": 0.7776,
      "step": 4517
    },
    {
      "epoch": 0.9718218971821897,
      "grad_norm": 0.0,
      "learning_rate": 4.1624626649584555e-08,
      "loss": 0.7657,
      "step": 4518
    },
    {
      "epoch": 0.9720369972036997,
      "grad_norm": 0.0,
      "learning_rate": 4.099199457024194e-08,
      "loss": 0.8124,
      "step": 4519
    },
    {
      "epoch": 0.9722520972252098,
      "grad_norm": 0.0,
      "learning_rate": 4.036419703291272e-08,
      "loss": 0.794,
      "step": 4520
    },
    {
      "epoch": 0.9724671972467197,
      "grad_norm": 0.0,
      "learning_rate": 3.974123434235644e-08,
      "loss": 0.7758,
      "step": 4521
    },
    {
      "epoch": 0.9726822972682297,
      "grad_norm": 0.0,
      "learning_rate": 3.912310680098785e-08,
      "loss": 0.8238,
      "step": 4522
    },
    {
      "epoch": 0.9728973972897397,
      "grad_norm": 0.0,
      "learning_rate": 3.850981470887361e-08,
      "loss": 0.8327,
      "step": 4523
    },
    {
      "epoch": 0.9731124973112497,
      "grad_norm": 0.0,
      "learning_rate": 3.79013583637311e-08,
      "loss": 0.8015,
      "step": 4524
    },
    {
      "epoch": 0.9733275973327598,
      "grad_norm": 0.0,
      "learning_rate": 3.729773806093295e-08,
      "loss": 0.7736,
      "step": 4525
    },
    {
      "epoch": 0.9735426973542697,
      "grad_norm": 0.0,
      "learning_rate": 3.669895409350366e-08,
      "loss": 0.7459,
      "step": 4526
    },
    {
      "epoch": 0.9737577973757797,
      "grad_norm": 0.0,
      "learning_rate": 3.610500675211848e-08,
      "loss": 0.8383,
      "step": 4527
    },
    {
      "epoch": 0.9739728973972898,
      "grad_norm": 0.0,
      "learning_rate": 3.551589632510455e-08,
      "loss": 0.8388,
      "step": 4528
    },
    {
      "epoch": 0.9741879974187997,
      "grad_norm": 0.0,
      "learning_rate": 3.493162309844533e-08,
      "loss": 0.7762,
      "step": 4529
    },
    {
      "epoch": 0.9744030974403097,
      "grad_norm": 0.0,
      "learning_rate": 3.435218735576951e-08,
      "loss": 0.7883,
      "step": 4530
    },
    {
      "epoch": 0.9746181974618198,
      "grad_norm": 0.0,
      "learning_rate": 3.3777589378363177e-08,
      "loss": 0.7619,
      "step": 4531
    },
    {
      "epoch": 0.9748332974833297,
      "grad_norm": 0.0,
      "learning_rate": 3.3207829445159876e-08,
      "loss": 0.7375,
      "step": 4532
    },
    {
      "epoch": 0.9750483975048397,
      "grad_norm": 0.0,
      "learning_rate": 3.2642907832747263e-08,
      "loss": 0.8142,
      "step": 4533
    },
    {
      "epoch": 0.9752634975263498,
      "grad_norm": 0.0,
      "learning_rate": 3.2082824815362624e-08,
      "loss": 0.8737,
      "step": 4534
    },
    {
      "epoch": 0.9754785975478597,
      "grad_norm": 0.0,
      "learning_rate": 3.152758066489514e-08,
      "loss": 0.8554,
      "step": 4535
    },
    {
      "epoch": 0.9756936975693697,
      "grad_norm": 0.0,
      "learning_rate": 3.097717565088476e-08,
      "loss": 0.8545,
      "step": 4536
    },
    {
      "epoch": 0.9759087975908798,
      "grad_norm": 0.0,
      "learning_rate": 3.0431610040522194e-08,
      "loss": 0.7435,
      "step": 4537
    },
    {
      "epoch": 0.9761238976123897,
      "grad_norm": 0.0,
      "learning_rate": 2.989088409865004e-08,
      "loss": 0.8471,
      "step": 4538
    },
    {
      "epoch": 0.9763389976338998,
      "grad_norm": 0.0,
      "learning_rate": 2.9354998087759436e-08,
      "loss": 0.8589,
      "step": 4539
    },
    {
      "epoch": 0.9765540976554098,
      "grad_norm": 0.0,
      "learning_rate": 2.8823952267994503e-08,
      "loss": 0.7563,
      "step": 4540
    },
    {
      "epoch": 0.9767691976769197,
      "grad_norm": 0.0,
      "learning_rate": 2.8297746897146816e-08,
      "loss": 0.8363,
      "step": 4541
    },
    {
      "epoch": 0.9769842976984298,
      "grad_norm": 0.0,
      "learning_rate": 2.7776382230659816e-08,
      "loss": 0.8268,
      "step": 4542
    },
    {
      "epoch": 0.9771993977199398,
      "grad_norm": 0.0,
      "learning_rate": 2.7259858521627713e-08,
      "loss": 0.8139,
      "step": 4543
    },
    {
      "epoch": 0.9774144977414497,
      "grad_norm": 0.0,
      "learning_rate": 2.674817602079327e-08,
      "loss": 0.8172,
      "step": 4544
    },
    {
      "epoch": 0.9776295977629598,
      "grad_norm": 0.0,
      "learning_rate": 2.6241334976550016e-08,
      "loss": 0.8174,
      "step": 4545
    },
    {
      "epoch": 0.9778446977844698,
      "grad_norm": 0.0,
      "learning_rate": 2.5739335634941132e-08,
      "loss": 0.8742,
      "step": 4546
    },
    {
      "epoch": 0.9780597978059797,
      "grad_norm": 0.0,
      "learning_rate": 2.5242178239658356e-08,
      "loss": 0.8882,
      "step": 4547
    },
    {
      "epoch": 0.9782748978274898,
      "grad_norm": 0.0,
      "learning_rate": 2.4749863032045297e-08,
      "loss": 0.8035,
      "step": 4548
    },
    {
      "epoch": 0.9784899978489998,
      "grad_norm": 0.0,
      "learning_rate": 2.4262390251091894e-08,
      "loss": 0.8578,
      "step": 4549
    },
    {
      "epoch": 0.9787050978705097,
      "grad_norm": 0.0,
      "learning_rate": 2.3779760133441076e-08,
      "loss": 0.7764,
      "step": 4550
    },
    {
      "epoch": 0.9789201978920198,
      "grad_norm": 0.0,
      "learning_rate": 2.330197291338099e-08,
      "loss": 0.8403,
      "step": 4551
    },
    {
      "epoch": 0.9791352979135298,
      "grad_norm": 0.0,
      "learning_rate": 2.282902882285054e-08,
      "loss": 0.7085,
      "step": 4552
    },
    {
      "epoch": 0.9793503979350398,
      "grad_norm": 0.0,
      "learning_rate": 2.2360928091439415e-08,
      "loss": 0.8266,
      "step": 4553
    },
    {
      "epoch": 0.9795654979565498,
      "grad_norm": 0.0,
      "learning_rate": 2.189767094638251e-08,
      "loss": 0.8903,
      "step": 4554
    },
    {
      "epoch": 0.9797805979780598,
      "grad_norm": 0.0,
      "learning_rate": 2.143925761256549e-08,
      "loss": 0.7984,
      "step": 4555
    },
    {
      "epoch": 0.9799956979995698,
      "grad_norm": 0.0,
      "learning_rate": 2.0985688312524787e-08,
      "loss": 0.7936,
      "step": 4556
    },
    {
      "epoch": 0.9802107980210798,
      "grad_norm": 0.0,
      "learning_rate": 2.0536963266438725e-08,
      "loss": 0.7951,
      "step": 4557
    },
    {
      "epoch": 0.9804258980425898,
      "grad_norm": 0.0,
      "learning_rate": 2.009308269214194e-08,
      "loss": 0.8012,
      "step": 4558
    },
    {
      "epoch": 0.9806409980640998,
      "grad_norm": 0.0,
      "learning_rate": 1.965404680511207e-08,
      "loss": 0.8782,
      "step": 4559
    },
    {
      "epoch": 0.9808560980856098,
      "grad_norm": 0.0,
      "learning_rate": 1.9219855818477517e-08,
      "loss": 0.7616,
      "step": 4560
    },
    {
      "epoch": 0.9810711981071198,
      "grad_norm": 0.0,
      "learning_rate": 1.8790509943011903e-08,
      "loss": 0.865,
      "step": 4561
    },
    {
      "epoch": 0.9812862981286298,
      "grad_norm": 0.0,
      "learning_rate": 1.8366009387140726e-08,
      "loss": 0.743,
      "step": 4562
    },
    {
      "epoch": 0.9815013981501398,
      "grad_norm": 0.0,
      "learning_rate": 1.7946354356933594e-08,
      "loss": 0.8056,
      "step": 4563
    },
    {
      "epoch": 0.9817164981716499,
      "grad_norm": 0.0,
      "learning_rate": 1.7531545056110875e-08,
      "loss": 0.7515,
      "step": 4564
    },
    {
      "epoch": 0.9819315981931598,
      "grad_norm": 0.0,
      "learning_rate": 1.712158168603928e-08,
      "loss": 0.7716,
      "step": 4565
    },
    {
      "epoch": 0.9821466982146698,
      "grad_norm": 0.0,
      "learning_rate": 1.671646444573183e-08,
      "loss": 0.8555,
      "step": 4566
    },
    {
      "epoch": 0.9823617982361799,
      "grad_norm": 0.0,
      "learning_rate": 1.63161935318501e-08,
      "loss": 0.8278,
      "step": 4567
    },
    {
      "epoch": 0.9825768982576898,
      "grad_norm": 0.0,
      "learning_rate": 1.592076913870644e-08,
      "loss": 0.8405,
      "step": 4568
    },
    {
      "epoch": 0.9827919982791998,
      "grad_norm": 0.0,
      "learning_rate": 1.5530191458252854e-08,
      "loss": 0.8082,
      "step": 4569
    },
    {
      "epoch": 0.9830070983007099,
      "grad_norm": 0.0,
      "learning_rate": 1.514446068009656e-08,
      "loss": 0.7409,
      "step": 4570
    },
    {
      "epoch": 0.9832221983222198,
      "grad_norm": 0.0,
      "learning_rate": 1.4763576991487782e-08,
      "loss": 0.8378,
      "step": 4571
    },
    {
      "epoch": 0.9834372983437298,
      "grad_norm": 0.0,
      "learning_rate": 1.4387540577321946e-08,
      "loss": 0.7721,
      "step": 4572
    },
    {
      "epoch": 0.9836523983652399,
      "grad_norm": 0.0,
      "learning_rate": 1.4016351620145252e-08,
      "loss": 0.7708,
      "step": 4573
    },
    {
      "epoch": 0.9838674983867498,
      "grad_norm": 0.0,
      "learning_rate": 1.3650010300150229e-08,
      "loss": 0.8197,
      "step": 4574
    },
    {
      "epoch": 0.9840825984082598,
      "grad_norm": 0.0,
      "learning_rate": 1.3288516795173511e-08,
      "loss": 0.722,
      "step": 4575
    },
    {
      "epoch": 0.9842976984297699,
      "grad_norm": 0.0,
      "learning_rate": 1.2931871280700281e-08,
      "loss": 0.8083,
      "step": 4576
    },
    {
      "epoch": 0.9845127984512798,
      "grad_norm": 0.0,
      "learning_rate": 1.2580073929863157e-08,
      "loss": 0.736,
      "step": 4577
    },
    {
      "epoch": 0.9847278984727899,
      "grad_norm": 0.0,
      "learning_rate": 1.2233124913438865e-08,
      "loss": 0.812,
      "step": 4578
    },
    {
      "epoch": 0.9849429984942999,
      "grad_norm": 0.0,
      "learning_rate": 1.1891024399851569e-08,
      "loss": 0.8172,
      "step": 4579
    },
    {
      "epoch": 0.9851580985158098,
      "grad_norm": 0.0,
      "learning_rate": 1.1553772555172871e-08,
      "loss": 0.8418,
      "step": 4580
    },
    {
      "epoch": 0.9853731985373199,
      "grad_norm": 0.0,
      "learning_rate": 1.1221369543119587e-08,
      "loss": 0.8536,
      "step": 4581
    },
    {
      "epoch": 0.9855882985588299,
      "grad_norm": 0.0,
      "learning_rate": 1.0893815525054863e-08,
      "loss": 0.8517,
      "step": 4582
    },
    {
      "epoch": 0.9858033985803398,
      "grad_norm": 0.0,
      "learning_rate": 1.057111065998706e-08,
      "loss": 0.7984,
      "step": 4583
    },
    {
      "epoch": 0.9860184986018499,
      "grad_norm": 0.0,
      "learning_rate": 1.0253255104571979e-08,
      "loss": 0.7738,
      "step": 4584
    },
    {
      "epoch": 0.9862335986233599,
      "grad_norm": 0.0,
      "learning_rate": 9.940249013111747e-09,
      "loss": 0.8217,
      "step": 4585
    },
    {
      "epoch": 0.9864486986448698,
      "grad_norm": 0.0,
      "learning_rate": 9.632092537551485e-09,
      "loss": 0.7469,
      "step": 4586
    },
    {
      "epoch": 0.9866637986663799,
      "grad_norm": 0.0,
      "learning_rate": 9.328785827483756e-09,
      "loss": 0.8047,
      "step": 4587
    },
    {
      "epoch": 0.9868788986878899,
      "grad_norm": 0.0,
      "learning_rate": 9.030329030148554e-09,
      "loss": 0.8359,
      "step": 4588
    },
    {
      "epoch": 0.9870939987093998,
      "grad_norm": 0.0,
      "learning_rate": 8.736722290429988e-09,
      "loss": 0.7484,
      "step": 4589
    },
    {
      "epoch": 0.9873090987309099,
      "grad_norm": 0.0,
      "learning_rate": 8.447965750857378e-09,
      "loss": 0.836,
      "step": 4590
    },
    {
      "epoch": 0.9875241987524199,
      "grad_norm": 0.0,
      "learning_rate": 8.164059551605264e-09,
      "loss": 0.7724,
      "step": 4591
    },
    {
      "epoch": 0.9877392987739299,
      "grad_norm": 0.0,
      "learning_rate": 7.885003830494508e-09,
      "loss": 0.8716,
      "step": 4592
    },
    {
      "epoch": 0.9879543987954399,
      "grad_norm": 0.0,
      "learning_rate": 7.610798722991197e-09,
      "loss": 0.7755,
      "step": 4593
    },
    {
      "epoch": 0.9881694988169499,
      "grad_norm": 0.0,
      "learning_rate": 7.3414443622066286e-09,
      "loss": 0.8367,
      "step": 4594
    },
    {
      "epoch": 0.9883845988384599,
      "grad_norm": 0.0,
      "learning_rate": 7.07694087889621e-09,
      "loss": 0.8298,
      "step": 4595
    },
    {
      "epoch": 0.9885996988599699,
      "grad_norm": 0.0,
      "learning_rate": 6.8172884014638955e-09,
      "loss": 0.8246,
      "step": 4596
    },
    {
      "epoch": 0.9888147988814799,
      "grad_norm": 0.0,
      "learning_rate": 6.562487055955524e-09,
      "loss": 0.7173,
      "step": 4597
    },
    {
      "epoch": 0.9890298989029899,
      "grad_norm": 0.0,
      "learning_rate": 6.312536966062155e-09,
      "loss": 0.8021,
      "step": 4598
    },
    {
      "epoch": 0.9892449989244999,
      "grad_norm": 0.0,
      "learning_rate": 6.067438253121171e-09,
      "loss": 0.8216,
      "step": 4599
    },
    {
      "epoch": 0.98946009894601,
      "grad_norm": 0.0,
      "learning_rate": 5.827191036114066e-09,
      "loss": 0.8065,
      "step": 4600
    },
    {
      "epoch": 0.9896751989675199,
      "grad_norm": 0.0,
      "learning_rate": 5.591795431667546e-09,
      "loss": 0.8293,
      "step": 4601
    },
    {
      "epoch": 0.9898902989890299,
      "grad_norm": 0.0,
      "learning_rate": 5.3612515540524264e-09,
      "loss": 0.8094,
      "step": 4602
    },
    {
      "epoch": 0.99010539901054,
      "grad_norm": 0.0,
      "learning_rate": 5.135559515186961e-09,
      "loss": 0.8434,
      "step": 4603
    },
    {
      "epoch": 0.9903204990320499,
      "grad_norm": 0.0,
      "learning_rate": 4.9147194246290666e-09,
      "loss": 0.8218,
      "step": 4604
    },
    {
      "epoch": 0.9905355990535599,
      "grad_norm": 0.0,
      "learning_rate": 4.69873138958632e-09,
      "loss": 0.8026,
      "step": 4605
    },
    {
      "epoch": 0.99075069907507,
      "grad_norm": 0.0,
      "learning_rate": 4.487595514909293e-09,
      "loss": 0.8501,
      "step": 4606
    },
    {
      "epoch": 0.9909657990965799,
      "grad_norm": 0.0,
      "learning_rate": 4.281311903090446e-09,
      "loss": 0.8883,
      "step": 4607
    },
    {
      "epoch": 0.9911808991180899,
      "grad_norm": 0.0,
      "learning_rate": 4.079880654270785e-09,
      "loss": 0.8517,
      "step": 4608
    },
    {
      "epoch": 0.9913959991395999,
      "grad_norm": 0.0,
      "learning_rate": 3.883301866232092e-09,
      "loss": 0.8193,
      "step": 4609
    },
    {
      "epoch": 0.9916110991611099,
      "grad_norm": 0.0,
      "learning_rate": 3.6915756344046983e-09,
      "loss": 0.8327,
      "step": 4610
    },
    {
      "epoch": 0.9918261991826199,
      "grad_norm": 0.0,
      "learning_rate": 3.5047020518597096e-09,
      "loss": 0.8376,
      "step": 4611
    },
    {
      "epoch": 0.9920412992041299,
      "grad_norm": 0.0,
      "learning_rate": 3.32268120931456e-09,
      "loss": 0.8305,
      "step": 4612
    },
    {
      "epoch": 0.9922563992256399,
      "grad_norm": 0.0,
      "learning_rate": 3.1455131951285688e-09,
      "loss": 0.8133,
      "step": 4613
    },
    {
      "epoch": 0.99247149924715,
      "grad_norm": 0.0,
      "learning_rate": 2.9731980953084936e-09,
      "loss": 0.8692,
      "step": 4614
    },
    {
      "epoch": 0.9926865992686599,
      "grad_norm": 0.0,
      "learning_rate": 2.8057359935029784e-09,
      "loss": 0.8527,
      "step": 4615
    },
    {
      "epoch": 0.9929016992901699,
      "grad_norm": 0.0,
      "learning_rate": 2.6431269710058827e-09,
      "loss": 0.8769,
      "step": 4616
    },
    {
      "epoch": 0.99311679931168,
      "grad_norm": 0.0,
      "learning_rate": 2.4853711067551746e-09,
      "loss": 0.8188,
      "step": 4617
    },
    {
      "epoch": 0.9933318993331899,
      "grad_norm": 0.0,
      "learning_rate": 2.332468477330707e-09,
      "loss": 0.8034,
      "step": 4618
    },
    {
      "epoch": 0.9935469993546999,
      "grad_norm": 0.0,
      "learning_rate": 2.184419156959772e-09,
      "loss": 0.7694,
      "step": 4619
    },
    {
      "epoch": 0.99376209937621,
      "grad_norm": 0.0,
      "learning_rate": 2.0412232175126555e-09,
      "loss": 0.8007,
      "step": 4620
    },
    {
      "epoch": 0.9939771993977199,
      "grad_norm": 0.0,
      "learning_rate": 1.9028807285015326e-09,
      "loss": 0.7558,
      "step": 4621
    },
    {
      "epoch": 0.9941922994192299,
      "grad_norm": 0.0,
      "learning_rate": 1.7693917570837938e-09,
      "loss": 0.8414,
      "step": 4622
    },
    {
      "epoch": 0.99440739944074,
      "grad_norm": 0.0,
      "learning_rate": 1.6407563680620465e-09,
      "loss": 0.8665,
      "step": 4623
    },
    {
      "epoch": 0.9946224994622499,
      "grad_norm": 0.0,
      "learning_rate": 1.5169746238807847e-09,
      "loss": 0.8333,
      "step": 4624
    },
    {
      "epoch": 0.9948375994837599,
      "grad_norm": 0.0,
      "learning_rate": 1.3980465846297198e-09,
      "loss": 0.788,
      "step": 4625
    },
    {
      "epoch": 0.99505269950527,
      "grad_norm": 0.0,
      "learning_rate": 1.2839723080404486e-09,
      "loss": 0.846,
      "step": 4626
    },
    {
      "epoch": 0.9952677995267799,
      "grad_norm": 0.0,
      "learning_rate": 1.1747518494920063e-09,
      "loss": 0.7818,
      "step": 4627
    },
    {
      "epoch": 0.99548289954829,
      "grad_norm": 0.0,
      "learning_rate": 1.070385262003093e-09,
      "loss": 0.8021,
      "step": 4628
    },
    {
      "epoch": 0.9956979995698,
      "grad_norm": 0.0,
      "learning_rate": 9.708725962376265e-10,
      "loss": 0.8078,
      "step": 4629
    },
    {
      "epoch": 0.9959130995913099,
      "grad_norm": 0.0,
      "learning_rate": 8.76213900504741e-10,
      "loss": 0.8436,
      "step": 4630
    },
    {
      "epoch": 0.99612819961282,
      "grad_norm": 0.0,
      "learning_rate": 7.864092207554575e-10,
      "loss": 0.7927,
      "step": 4631
    },
    {
      "epoch": 0.99634329963433,
      "grad_norm": 0.0,
      "learning_rate": 7.014586005837931e-10,
      "loss": 0.8431,
      "step": 4632
    },
    {
      "epoch": 0.9965583996558399,
      "grad_norm": 0.0,
      "learning_rate": 6.213620812300924e-10,
      "loss": 0.8312,
      "step": 4633
    },
    {
      "epoch": 0.99677349967735,
      "grad_norm": 0.0,
      "learning_rate": 5.461197015765863e-10,
      "loss": 0.8907,
      "step": 4634
    },
    {
      "epoch": 0.99698859969886,
      "grad_norm": 0.0,
      "learning_rate": 4.757314981485017e-10,
      "loss": 0.7517,
      "step": 4635
    },
    {
      "epoch": 0.9972036997203699,
      "grad_norm": 0.0,
      "learning_rate": 4.10197505115173e-10,
      "loss": 0.7974,
      "step": 4636
    },
    {
      "epoch": 0.99741879974188,
      "grad_norm": 0.0,
      "learning_rate": 3.4951775429004073e-10,
      "loss": 0.8225,
      "step": 4637
    },
    {
      "epoch": 0.99763389976339,
      "grad_norm": 0.0,
      "learning_rate": 2.9369227513065255e-10,
      "loss": 0.8476,
      "step": 4638
    },
    {
      "epoch": 0.9978489997848999,
      "grad_norm": 0.0,
      "learning_rate": 2.4272109473644225e-10,
      "loss": 0.8227,
      "step": 4639
    },
    {
      "epoch": 0.99806409980641,
      "grad_norm": 0.0,
      "learning_rate": 1.9660423784984007e-10,
      "loss": 0.7886,
      "step": 4640
    },
    {
      "epoch": 0.99827919982792,
      "grad_norm": 0.0,
      "learning_rate": 1.553417268607138e-10,
      "loss": 0.8299,
      "step": 4641
    },
    {
      "epoch": 0.99849429984943,
      "grad_norm": 0.0,
      "learning_rate": 1.189335817963766e-10,
      "loss": 0.8425,
      "step": 4642
    },
    {
      "epoch": 0.99870939987094,
      "grad_norm": 0.0,
      "learning_rate": 8.737982033379944e-11,
      "loss": 0.8007,
      "step": 4643
    },
    {
      "epoch": 0.99892449989245,
      "grad_norm": 0.0,
      "learning_rate": 6.068045778961917e-11,
      "loss": 0.8463,
      "step": 4644
    },
    {
      "epoch": 0.99913959991396,
      "grad_norm": 0.0,
      "learning_rate": 3.88355071245794e-11,
      "loss": 0.7867,
      "step": 4645
    },
    {
      "epoch": 0.99935469993547,
      "grad_norm": 0.0,
      "learning_rate": 2.184497894353044e-11,
      "loss": 0.8235,
      "step": 4646
    },
    {
      "epoch": 0.99956979995698,
      "grad_norm": 0.0,
      "learning_rate": 9.708881494319145e-12,
      "loss": 0.8366,
      "step": 4647
    },
    {
      "epoch": 0.99978489997849,
      "grad_norm": 0.0,
      "learning_rate": 2.4272206677888876e-12,
      "loss": 0.814,
      "step": 4648
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 4649
    },
    {
      "epoch": 1.0,
      "step": 4649,
      "total_flos": 2.7743596673706754e+19,
      "train_loss": 0.8857983078308376,
      "train_runtime": 80782.8926,
      "train_samples_per_second": 18.413,
      "train_steps_per_second": 0.058
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4649,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 3000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.7743596673706754e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}