{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999159027836179,
  "eval_steps": 500,
  "global_step": 5945,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016819443276427552,
      "grad_norm": NaN,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.2492,
      "step": 1
    },
    {
      "epoch": 0.00033638886552855103,
      "grad_norm": NaN,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0,
      "step": 2
    },
    {
      "epoch": 0.0005045832982928265,
      "grad_norm": NaN,
      "learning_rate": 3e-06,
      "loss": 0.0,
      "step": 3
    },
    {
      "epoch": 0.0006727777310571021,
      "grad_norm": NaN,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0,
      "step": 4
    },
    {
      "epoch": 0.0008409721638213775,
      "grad_norm": NaN,
      "learning_rate": 5e-06,
      "loss": 0.0,
      "step": 5
    },
    {
      "epoch": 0.001009166596585653,
      "grad_norm": NaN,
      "learning_rate": 6e-06,
      "loss": 0.0,
      "step": 6
    },
    {
      "epoch": 0.0011773610293499286,
      "grad_norm": NaN,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.0,
      "step": 7
    },
    {
      "epoch": 0.0013455554621142041,
      "grad_norm": NaN,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0,
      "step": 8
    },
    {
      "epoch": 0.0015137498948784795,
      "grad_norm": NaN,
      "learning_rate": 9e-06,
      "loss": 0.0,
      "step": 9
    },
    {
      "epoch": 0.001681944327642755,
      "grad_norm": NaN,
      "learning_rate": 1e-05,
      "loss": 0.0,
      "step": 10
    },
    {
      "epoch": 0.0018501387604070306,
      "grad_norm": NaN,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.0,
      "step": 11
    },
    {
      "epoch": 0.002018333193171306,
      "grad_norm": NaN,
      "learning_rate": 1.2e-05,
      "loss": 0.0,
      "step": 12
    },
    {
      "epoch": 0.0021865276259355813,
      "grad_norm": NaN,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.0,
      "step": 13
    },
    {
      "epoch": 0.002354722058699857,
      "grad_norm": NaN,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.0,
      "step": 14
    },
    {
      "epoch": 0.0025229164914641325,
      "grad_norm": NaN,
      "learning_rate": 1.5e-05,
      "loss": 0.0,
      "step": 15
    },
    {
      "epoch": 0.0026911109242284082,
      "grad_norm": NaN,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0,
      "step": 16
    },
    {
      "epoch": 0.0028593053569926836,
      "grad_norm": NaN,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.0,
      "step": 17
    },
    {
      "epoch": 0.003027499789756959,
      "grad_norm": NaN,
      "learning_rate": 1.8e-05,
      "loss": 0.0,
      "step": 18
    },
    {
      "epoch": 0.0031956942225212347,
      "grad_norm": NaN,
      "learning_rate": 1.9e-05,
      "loss": 0.0,
      "step": 19
    },
    {
      "epoch": 0.00336388865528551,
      "grad_norm": NaN,
      "learning_rate": 2e-05,
      "loss": 0.0,
      "step": 20
    },
    {
      "epoch": 0.0035320830880497854,
      "grad_norm": NaN,
      "learning_rate": 2.1e-05,
      "loss": 0.0,
      "step": 21
    },
    {
      "epoch": 0.0037002775208140612,
      "grad_norm": NaN,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.0,
      "step": 22
    },
    {
      "epoch": 0.0038684719535783366,
      "grad_norm": NaN,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.0,
      "step": 23
    },
    {
      "epoch": 0.004036666386342612,
      "grad_norm": NaN,
      "learning_rate": 2.4e-05,
      "loss": 0.0,
      "step": 24
    },
    {
      "epoch": 0.004204860819106888,
      "grad_norm": NaN,
      "learning_rate": 2.5e-05,
      "loss": 0.0,
      "step": 25
    },
    {
      "epoch": 0.004373055251871163,
      "grad_norm": NaN,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.0,
      "step": 26
    },
    {
      "epoch": 0.004541249684635438,
      "grad_norm": NaN,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.0,
      "step": 27
    },
    {
      "epoch": 0.004709444117399714,
      "grad_norm": NaN,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.0,
      "step": 28
    },
    {
      "epoch": 0.00487763855016399,
      "grad_norm": NaN,
      "learning_rate": 2.9e-05,
      "loss": 0.0,
      "step": 29
    },
    {
      "epoch": 0.005045832982928265,
      "grad_norm": NaN,
      "learning_rate": 3e-05,
      "loss": 0.0,
      "step": 30
    },
    {
      "epoch": 0.005214027415692541,
      "grad_norm": NaN,
      "learning_rate": 3.1e-05,
      "loss": 0.0,
      "step": 31
    },
    {
      "epoch": 0.0053822218484568165,
      "grad_norm": NaN,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.0,
      "step": 32
    },
    {
      "epoch": 0.005550416281221091,
      "grad_norm": NaN,
      "learning_rate": 3.3e-05,
      "loss": 0.0,
      "step": 33
    },
    {
      "epoch": 0.005718610713985367,
      "grad_norm": NaN,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.0,
      "step": 34
    },
    {
      "epoch": 0.005886805146749643,
      "grad_norm": NaN,
      "learning_rate": 3.5e-05,
      "loss": 0.0,
      "step": 35
    },
    {
      "epoch": 0.006054999579513918,
      "grad_norm": NaN,
      "learning_rate": 3.6e-05,
      "loss": 0.0,
      "step": 36
    },
    {
      "epoch": 0.006223194012278194,
      "grad_norm": NaN,
      "learning_rate": 3.7e-05,
      "loss": 0.0,
      "step": 37
    },
    {
      "epoch": 0.0063913884450424695,
      "grad_norm": NaN,
      "learning_rate": 3.8e-05,
      "loss": 0.0,
      "step": 38
    },
    {
      "epoch": 0.006559582877806744,
      "grad_norm": NaN,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.0,
      "step": 39
    },
    {
      "epoch": 0.00672777731057102,
      "grad_norm": NaN,
      "learning_rate": 4e-05,
      "loss": 0.0,
      "step": 40
    },
    {
      "epoch": 0.006895971743335296,
      "grad_norm": NaN,
      "learning_rate": 4.1e-05,
      "loss": 0.0,
      "step": 41
    },
    {
      "epoch": 0.007064166176099571,
      "grad_norm": NaN,
      "learning_rate": 4.2e-05,
      "loss": 0.0,
      "step": 42
    },
    {
      "epoch": 0.007232360608863847,
      "grad_norm": NaN,
      "learning_rate": 4.3e-05,
      "loss": 0.0,
      "step": 43
    },
    {
      "epoch": 0.0074005550416281225,
      "grad_norm": NaN,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0,
      "step": 44
    },
    {
      "epoch": 0.007568749474392397,
      "grad_norm": NaN,
      "learning_rate": 4.5e-05,
      "loss": 0.0,
      "step": 45
    },
    {
      "epoch": 0.007736943907156673,
      "grad_norm": NaN,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.0,
      "step": 46
    },
    {
      "epoch": 0.007905138339920948,
      "grad_norm": NaN,
      "learning_rate": 4.7e-05,
      "loss": 0.0,
      "step": 47
    },
    {
      "epoch": 0.008073332772685224,
      "grad_norm": NaN,
      "learning_rate": 4.8e-05,
      "loss": 0.0,
      "step": 48
    },
    {
      "epoch": 0.0082415272054495,
      "grad_norm": NaN,
      "learning_rate": 4.9e-05,
      "loss": 0.0,
      "step": 49
    },
    {
      "epoch": 0.008409721638213775,
      "grad_norm": NaN,
      "learning_rate": 5e-05,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 0.008577916070978051,
      "grad_norm": NaN,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 0.0,
      "step": 51
    },
    {
      "epoch": 0.008746110503742325,
      "grad_norm": NaN,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.0,
      "step": 52
    },
    {
      "epoch": 0.008914304936506601,
      "grad_norm": NaN,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.0,
      "step": 53
    },
    {
      "epoch": 0.009082499369270877,
      "grad_norm": NaN,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.0,
      "step": 54
    },
    {
      "epoch": 0.009250693802035153,
      "grad_norm": NaN,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.0,
      "step": 55
    },
    {
      "epoch": 0.009418888234799428,
      "grad_norm": NaN,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.0,
      "step": 56
    },
    {
      "epoch": 0.009587082667563704,
      "grad_norm": NaN,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.0,
      "step": 57
    },
    {
      "epoch": 0.00975527710032798,
      "grad_norm": NaN,
      "learning_rate": 5.8e-05,
      "loss": 0.0,
      "step": 58
    },
    {
      "epoch": 0.009923471533092254,
      "grad_norm": NaN,
      "learning_rate": 5.9e-05,
      "loss": 0.0,
      "step": 59
    },
    {
      "epoch": 0.01009166596585653,
      "grad_norm": NaN,
      "learning_rate": 6e-05,
      "loss": 0.0,
      "step": 60
    },
    {
      "epoch": 0.010259860398620806,
      "grad_norm": NaN,
      "learning_rate": 6.1e-05,
      "loss": 0.0,
      "step": 61
    },
    {
      "epoch": 0.010428054831385081,
      "grad_norm": NaN,
      "learning_rate": 6.2e-05,
      "loss": 0.0,
      "step": 62
    },
    {
      "epoch": 0.010596249264149357,
      "grad_norm": NaN,
      "learning_rate": 6.3e-05,
      "loss": 0.0,
      "step": 63
    },
    {
      "epoch": 0.010764443696913633,
      "grad_norm": NaN,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.0,
      "step": 64
    },
    {
      "epoch": 0.010932638129677907,
      "grad_norm": NaN,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.0,
      "step": 65
    },
    {
      "epoch": 0.011100832562442183,
      "grad_norm": NaN,
      "learning_rate": 6.6e-05,
      "loss": 0.0,
      "step": 66
    },
    {
      "epoch": 0.011269026995206459,
      "grad_norm": NaN,
      "learning_rate": 6.7e-05,
      "loss": 0.0,
      "step": 67
    },
    {
      "epoch": 0.011437221427970734,
      "grad_norm": NaN,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.0,
      "step": 68
    },
    {
      "epoch": 0.01160541586073501,
      "grad_norm": NaN,
      "learning_rate": 6.9e-05,
      "loss": 0.0,
      "step": 69
    },
    {
      "epoch": 0.011773610293499286,
      "grad_norm": NaN,
      "learning_rate": 7e-05,
      "loss": 0.0,
      "step": 70
    },
    {
      "epoch": 0.01194180472626356,
      "grad_norm": NaN,
      "learning_rate": 7.1e-05,
      "loss": 0.0,
      "step": 71
    },
    {
      "epoch": 0.012109999159027836,
      "grad_norm": NaN,
      "learning_rate": 7.2e-05,
      "loss": 0.0,
      "step": 72
    },
    {
      "epoch": 0.012278193591792112,
      "grad_norm": NaN,
      "learning_rate": 7.3e-05,
      "loss": 0.0,
      "step": 73
    },
    {
      "epoch": 0.012446388024556387,
      "grad_norm": NaN,
      "learning_rate": 7.4e-05,
      "loss": 0.0,
      "step": 74
    },
    {
      "epoch": 0.012614582457320663,
      "grad_norm": NaN,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.0,
      "step": 75
    },
    {
      "epoch": 0.012782776890084939,
      "grad_norm": NaN,
      "learning_rate": 7.6e-05,
      "loss": 0.0,
      "step": 76
    },
    {
      "epoch": 0.012950971322849213,
      "grad_norm": NaN,
      "learning_rate": 7.7e-05,
      "loss": 0.0,
      "step": 77
    },
    {
      "epoch": 0.013119165755613489,
      "grad_norm": NaN,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.0,
      "step": 78
    },
    {
      "epoch": 0.013287360188377765,
      "grad_norm": NaN,
      "learning_rate": 7.900000000000001e-05,
      "loss": 0.0,
      "step": 79
    },
    {
      "epoch": 0.01345555462114204,
      "grad_norm": NaN,
      "learning_rate": 8e-05,
      "loss": 0.0,
      "step": 80
    },
    {
      "epoch": 0.013623749053906316,
      "grad_norm": NaN,
      "learning_rate": 8.1e-05,
      "loss": 0.0,
      "step": 81
    },
    {
      "epoch": 0.013791943486670592,
      "grad_norm": NaN,
      "learning_rate": 8.2e-05,
      "loss": 0.0,
      "step": 82
    },
    {
      "epoch": 0.013960137919434866,
      "grad_norm": NaN,
      "learning_rate": 8.3e-05,
      "loss": 0.0,
      "step": 83
    },
    {
      "epoch": 0.014128332352199142,
      "grad_norm": NaN,
      "learning_rate": 8.4e-05,
      "loss": 0.0,
      "step": 84
    },
    {
      "epoch": 0.014296526784963418,
      "grad_norm": NaN,
      "learning_rate": 8.5e-05,
      "loss": 0.0,
      "step": 85
    },
    {
      "epoch": 0.014464721217727693,
      "grad_norm": NaN,
      "learning_rate": 8.6e-05,
      "loss": 0.0,
      "step": 86
    },
    {
      "epoch": 0.014632915650491969,
      "grad_norm": NaN,
      "learning_rate": 8.7e-05,
      "loss": 0.0,
      "step": 87
    },
    {
      "epoch": 0.014801110083256245,
      "grad_norm": NaN,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.0,
      "step": 88
    },
    {
      "epoch": 0.014969304516020519,
      "grad_norm": NaN,
      "learning_rate": 8.900000000000001e-05,
      "loss": 0.0,
      "step": 89
    },
    {
      "epoch": 0.015137498948784795,
      "grad_norm": NaN,
      "learning_rate": 9e-05,
      "loss": 0.0,
      "step": 90
    },
    {
      "epoch": 0.01530569338154907,
      "grad_norm": NaN,
      "learning_rate": 9.1e-05,
      "loss": 0.0,
      "step": 91
    },
    {
      "epoch": 0.015473887814313346,
      "grad_norm": NaN,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.0,
      "step": 92
    },
    {
      "epoch": 0.01564208224707762,
      "grad_norm": NaN,
      "learning_rate": 9.300000000000001e-05,
      "loss": 0.0,
      "step": 93
    },
    {
      "epoch": 0.015810276679841896,
      "grad_norm": NaN,
      "learning_rate": 9.4e-05,
      "loss": 0.0,
      "step": 94
    },
    {
      "epoch": 0.015978471112606172,
      "grad_norm": NaN,
      "learning_rate": 9.5e-05,
      "loss": 0.0,
      "step": 95
    },
    {
      "epoch": 0.016146665545370448,
      "grad_norm": NaN,
      "learning_rate": 9.6e-05,
      "loss": 0.0,
      "step": 96
    },
    {
      "epoch": 0.016314859978134724,
      "grad_norm": NaN,
      "learning_rate": 9.7e-05,
      "loss": 0.0,
      "step": 97
    },
    {
      "epoch": 0.016483054410899,
      "grad_norm": NaN,
      "learning_rate": 9.8e-05,
      "loss": 0.0,
      "step": 98
    },
    {
      "epoch": 0.016651248843663275,
      "grad_norm": NaN,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.0,
      "step": 99
    },
    {
      "epoch": 0.01681944327642755,
      "grad_norm": NaN,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 100
    },
    {
      "epoch": 0.016987637709191827,
      "grad_norm": NaN,
      "learning_rate": 9.999999277778003e-05,
      "loss": 0.0,
      "step": 101
    },
    {
      "epoch": 0.017155832141956102,
      "grad_norm": NaN,
      "learning_rate": 9.999997111112216e-05,
      "loss": 0.0,
      "step": 102
    },
    {
      "epoch": 0.017324026574720378,
      "grad_norm": NaN,
      "learning_rate": 9.999993500003267e-05,
      "loss": 0.0,
      "step": 103
    },
    {
      "epoch": 0.01749222100748465,
      "grad_norm": NaN,
      "learning_rate": 9.999988444452199e-05,
      "loss": 0.0,
      "step": 104
    },
    {
      "epoch": 0.017660415440248926,
      "grad_norm": NaN,
      "learning_rate": 9.999981944460473e-05,
      "loss": 0.0,
      "step": 105
    },
    {
      "epoch": 0.017828609873013202,
      "grad_norm": NaN,
      "learning_rate": 9.999974000029966e-05,
      "loss": 0.0,
      "step": 106
    },
    {
      "epoch": 0.017996804305777478,
      "grad_norm": NaN,
      "learning_rate": 9.999964611162974e-05,
      "loss": 0.0,
      "step": 107
    },
    {
      "epoch": 0.018164998738541754,
      "grad_norm": NaN,
      "learning_rate": 9.999953777862207e-05,
      "loss": 0.0,
      "step": 108
    },
    {
      "epoch": 0.01833319317130603,
      "grad_norm": NaN,
      "learning_rate": 9.999941500130797e-05,
      "loss": 0.0,
      "step": 109
    },
    {
      "epoch": 0.018501387604070305,
      "grad_norm": NaN,
      "learning_rate": 9.99992777797229e-05,
      "loss": 0.0,
      "step": 110
    },
    {
      "epoch": 0.01866958203683458,
      "grad_norm": NaN,
      "learning_rate": 9.999912611390651e-05,
      "loss": 0.0,
      "step": 111
    },
    {
      "epoch": 0.018837776469598857,
      "grad_norm": NaN,
      "learning_rate": 9.999896000390261e-05,
      "loss": 0.0,
      "step": 112
    },
    {
      "epoch": 0.019005970902363133,
      "grad_norm": NaN,
      "learning_rate": 9.999877944975917e-05,
      "loss": 0.0,
      "step": 113
    },
    {
      "epoch": 0.01917416533512741,
      "grad_norm": NaN,
      "learning_rate": 9.999858445152839e-05,
      "loss": 0.0,
      "step": 114
    },
    {
      "epoch": 0.019342359767891684,
      "grad_norm": NaN,
      "learning_rate": 9.999837500926656e-05,
      "loss": 0.0,
      "step": 115
    },
    {
      "epoch": 0.01951055420065596,
      "grad_norm": NaN,
      "learning_rate": 9.99981511230342e-05,
      "loss": 0.0,
      "step": 116
    },
    {
      "epoch": 0.019678748633420232,
      "grad_norm": NaN,
      "learning_rate": 9.999791279289601e-05,
      "loss": 0.0,
      "step": 117
    },
    {
      "epoch": 0.019846943066184508,
      "grad_norm": NaN,
      "learning_rate": 9.999766001892081e-05,
      "loss": 0.0,
      "step": 118
    },
    {
      "epoch": 0.020015137498948784,
      "grad_norm": NaN,
      "learning_rate": 9.999739280118163e-05,
      "loss": 0.0,
      "step": 119
    },
    {
      "epoch": 0.02018333193171306,
      "grad_norm": NaN,
      "learning_rate": 9.999711113975568e-05,
      "loss": 0.0,
      "step": 120
    },
    {
      "epoch": 0.020351526364477335,
      "grad_norm": NaN,
      "learning_rate": 9.999681503472433e-05,
      "loss": 0.0,
      "step": 121
    },
    {
      "epoch": 0.02051972079724161,
      "grad_norm": NaN,
      "learning_rate": 9.99965044861731e-05,
      "loss": 0.0,
      "step": 122
    },
    {
      "epoch": 0.020687915230005887,
      "grad_norm": NaN,
      "learning_rate": 9.999617949419174e-05,
      "loss": 0.0,
      "step": 123
    },
    {
      "epoch": 0.020856109662770163,
      "grad_norm": NaN,
      "learning_rate": 9.999584005887407e-05,
      "loss": 0.0,
      "step": 124
    },
    {
      "epoch": 0.02102430409553444,
      "grad_norm": NaN,
      "learning_rate": 9.999548618031823e-05,
      "loss": 0.0,
      "step": 125
    },
    {
      "epoch": 0.021192498528298714,
      "grad_norm": NaN,
      "learning_rate": 9.99951178586264e-05,
      "loss": 0.0,
      "step": 126
    },
    {
      "epoch": 0.02136069296106299,
      "grad_norm": NaN,
      "learning_rate": 9.9994735093905e-05,
      "loss": 0.0,
      "step": 127
    },
    {
      "epoch": 0.021528887393827266,
      "grad_norm": NaN,
      "learning_rate": 9.999433788626461e-05,
      "loss": 0.0,
      "step": 128
    },
    {
      "epoch": 0.02169708182659154,
      "grad_norm": NaN,
      "learning_rate": 9.999392623581997e-05,
      "loss": 0.0,
      "step": 129
    },
    {
      "epoch": 0.021865276259355814,
      "grad_norm": NaN,
      "learning_rate": 9.999350014269e-05,
      "loss": 0.0,
      "step": 130
    },
    {
      "epoch": 0.02203347069212009,
      "grad_norm": NaN,
      "learning_rate": 9.999305960699781e-05,
      "loss": 0.0,
      "step": 131
    },
    {
      "epoch": 0.022201665124884366,
      "grad_norm": NaN,
      "learning_rate": 9.999260462887064e-05,
      "loss": 0.0,
      "step": 132
    },
    {
      "epoch": 0.02236985955764864,
      "grad_norm": NaN,
      "learning_rate": 9.999213520843994e-05,
      "loss": 0.0,
      "step": 133
    },
    {
      "epoch": 0.022538053990412917,
      "grad_norm": NaN,
      "learning_rate": 9.999165134584133e-05,
      "loss": 0.0,
      "step": 134
    },
    {
      "epoch": 0.022706248423177193,
      "grad_norm": NaN,
      "learning_rate": 9.999115304121457e-05,
      "loss": 0.0,
      "step": 135
    },
    {
      "epoch": 0.02287444285594147,
      "grad_norm": NaN,
      "learning_rate": 9.999064029470366e-05,
      "loss": 0.0,
      "step": 136
    },
    {
      "epoch": 0.023042637288705745,
      "grad_norm": NaN,
      "learning_rate": 9.999011310645668e-05,
      "loss": 0.0,
      "step": 137
    },
    {
      "epoch": 0.02321083172147002,
      "grad_norm": NaN,
      "learning_rate": 9.998957147662594e-05,
      "loss": 0.0,
      "step": 138
    },
    {
      "epoch": 0.023379026154234296,
      "grad_norm": NaN,
      "learning_rate": 9.998901540536792e-05,
      "loss": 0.0,
      "step": 139
    },
    {
      "epoch": 0.023547220586998572,
      "grad_norm": NaN,
      "learning_rate": 9.998844489284327e-05,
      "loss": 0.0,
      "step": 140
    },
    {
      "epoch": 0.023715415019762844,
      "grad_norm": NaN,
      "learning_rate": 9.998785993921678e-05,
      "loss": 0.0,
      "step": 141
    },
    {
      "epoch": 0.02388360945252712,
      "grad_norm": NaN,
      "learning_rate": 9.998726054465744e-05,
      "loss": 0.0,
      "step": 142
    },
    {
      "epoch": 0.024051803885291396,
      "grad_norm": NaN,
      "learning_rate": 9.998664670933844e-05,
      "loss": 0.0,
      "step": 143
    },
    {
      "epoch": 0.02421999831805567,
      "grad_norm": NaN,
      "learning_rate": 9.998601843343707e-05,
      "loss": 0.0,
      "step": 144
    },
    {
      "epoch": 0.024388192750819947,
      "grad_norm": NaN,
      "learning_rate": 9.998537571713487e-05,
      "loss": 0.0,
      "step": 145
    },
    {
      "epoch": 0.024556387183584223,
      "grad_norm": NaN,
      "learning_rate": 9.998471856061747e-05,
      "loss": 0.0,
      "step": 146
    },
    {
      "epoch": 0.0247245816163485,
      "grad_norm": NaN,
      "learning_rate": 9.998404696407476e-05,
      "loss": 0.0,
      "step": 147
    },
    {
      "epoch": 0.024892776049112775,
      "grad_norm": NaN,
      "learning_rate": 9.998336092770073e-05,
      "loss": 0.0,
      "step": 148
    },
    {
      "epoch": 0.02506097048187705,
      "grad_norm": NaN,
      "learning_rate": 9.998266045169356e-05,
      "loss": 0.0,
      "step": 149
    },
    {
      "epoch": 0.025229164914641326,
      "grad_norm": NaN,
      "learning_rate": 9.998194553625563e-05,
      "loss": 0.0,
      "step": 150
    },
    {
      "epoch": 0.025397359347405602,
      "grad_norm": NaN,
      "learning_rate": 9.998121618159346e-05,
      "loss": 0.0,
      "step": 151
    },
    {
      "epoch": 0.025565553780169878,
      "grad_norm": NaN,
      "learning_rate": 9.998047238791777e-05,
      "loss": 0.0,
      "step": 152
    },
    {
      "epoch": 0.02573374821293415,
      "grad_norm": NaN,
      "learning_rate": 9.997971415544341e-05,
      "loss": 0.0,
      "step": 153
    },
    {
      "epoch": 0.025901942645698426,
      "grad_norm": NaN,
      "learning_rate": 9.997894148438944e-05,
      "loss": 0.0,
      "step": 154
    },
    {
      "epoch": 0.026070137078462702,
      "grad_norm": NaN,
      "learning_rate": 9.997815437497908e-05,
      "loss": 0.0,
      "step": 155
    },
    {
      "epoch": 0.026238331511226978,
      "grad_norm": NaN,
      "learning_rate": 9.997735282743969e-05,
      "loss": 0.0,
      "step": 156
    },
    {
      "epoch": 0.026406525943991253,
      "grad_norm": NaN,
      "learning_rate": 9.997653684200286e-05,
      "loss": 0.0,
      "step": 157
    },
    {
      "epoch": 0.02657472037675553,
      "grad_norm": NaN,
      "learning_rate": 9.99757064189043e-05,
      "loss": 0.0,
      "step": 158
    },
    {
      "epoch": 0.026742914809519805,
      "grad_norm": NaN,
      "learning_rate": 9.997486155838392e-05,
      "loss": 0.0,
      "step": 159
    },
    {
      "epoch": 0.02691110924228408,
      "grad_norm": NaN,
      "learning_rate": 9.997400226068578e-05,
      "loss": 0.0,
      "step": 160
    },
    {
      "epoch": 0.027079303675048357,
      "grad_norm": NaN,
      "learning_rate": 9.997312852605814e-05,
      "loss": 0.0,
      "step": 161
    },
    {
      "epoch": 0.027247498107812632,
      "grad_norm": NaN,
      "learning_rate": 9.997224035475339e-05,
      "loss": 0.0,
      "step": 162
    },
    {
      "epoch": 0.027415692540576908,
      "grad_norm": NaN,
      "learning_rate": 9.997133774702812e-05,
      "loss": 0.0,
      "step": 163
    },
    {
      "epoch": 0.027583886973341184,
      "grad_norm": NaN,
      "learning_rate": 9.997042070314309e-05,
      "loss": 0.0,
      "step": 164
    },
    {
      "epoch": 0.027752081406105456,
      "grad_norm": NaN,
      "learning_rate": 9.996948922336323e-05,
      "loss": 0.0,
      "step": 165
    },
    {
      "epoch": 0.027920275838869732,
      "grad_norm": NaN,
      "learning_rate": 9.996854330795761e-05,
      "loss": 0.0,
      "step": 166
    },
    {
      "epoch": 0.028088470271634008,
      "grad_norm": NaN,
      "learning_rate": 9.996758295719951e-05,
      "loss": 0.0,
      "step": 167
    },
    {
      "epoch": 0.028256664704398284,
      "grad_norm": NaN,
      "learning_rate": 9.996660817136636e-05,
      "loss": 0.0,
      "step": 168
    },
    {
      "epoch": 0.02842485913716256,
      "grad_norm": NaN,
      "learning_rate": 9.996561895073976e-05,
      "loss": 0.0,
      "step": 169
    },
    {
      "epoch": 0.028593053569926835,
      "grad_norm": NaN,
      "learning_rate": 9.996461529560553e-05,
      "loss": 0.0,
      "step": 170
    },
    {
      "epoch": 0.02876124800269111,
      "grad_norm": NaN,
      "learning_rate": 9.996359720625354e-05,
      "loss": 0.0,
      "step": 171
    },
    {
      "epoch": 0.028929442435455387,
      "grad_norm": NaN,
      "learning_rate": 9.996256468297795e-05,
      "loss": 0.0,
      "step": 172
    },
    {
      "epoch": 0.029097636868219662,
      "grad_norm": NaN,
      "learning_rate": 9.996151772607704e-05,
      "loss": 0.0,
      "step": 173
    },
    {
      "epoch": 0.029265831300983938,
      "grad_norm": NaN,
      "learning_rate": 9.996045633585326e-05,
      "loss": 0.0,
      "step": 174
    },
    {
      "epoch": 0.029434025733748214,
      "grad_norm": NaN,
      "learning_rate": 9.995938051261324e-05,
      "loss": 0.0,
      "step": 175
    },
    {
      "epoch": 0.02960222016651249,
      "grad_norm": NaN,
      "learning_rate": 9.995829025666775e-05,
      "loss": 0.0,
      "step": 176
    },
    {
      "epoch": 0.029770414599276766,
      "grad_norm": NaN,
      "learning_rate": 9.995718556833178e-05,
      "loss": 0.0,
      "step": 177
    },
    {
      "epoch": 0.029938609032041038,
      "grad_norm": NaN,
      "learning_rate": 9.995606644792446e-05,
      "loss": 0.0,
      "step": 178
    },
    {
      "epoch": 0.030106803464805314,
      "grad_norm": NaN,
      "learning_rate": 9.995493289576907e-05,
      "loss": 0.0,
      "step": 179
    },
    {
      "epoch": 0.03027499789756959,
      "grad_norm": NaN,
      "learning_rate": 9.99537849121931e-05,
      "loss": 0.0,
      "step": 180
    },
    {
      "epoch": 0.030443192330333865,
      "grad_norm": NaN,
      "learning_rate": 9.995262249752817e-05,
      "loss": 0.0,
      "step": 181
    },
    {
      "epoch": 0.03061138676309814,
      "grad_norm": NaN,
      "learning_rate": 9.995144565211012e-05,
      "loss": 0.0,
      "step": 182
    },
    {
      "epoch": 0.030779581195862417,
      "grad_norm": NaN,
      "learning_rate": 9.99502543762789e-05,
      "loss": 0.0,
      "step": 183
    },
    {
      "epoch": 0.030947775628626693,
      "grad_norm": NaN,
      "learning_rate": 9.994904867037867e-05,
      "loss": 0.0,
      "step": 184
    },
    {
      "epoch": 0.03111597006139097,
      "grad_norm": NaN,
      "learning_rate": 9.994782853475774e-05,
      "loss": 0.0,
      "step": 185
    },
    {
      "epoch": 0.03128416449415524,
      "grad_norm": NaN,
      "learning_rate": 9.994659396976859e-05,
      "loss": 0.0,
      "step": 186
    },
    {
      "epoch": 0.03145235892691952,
      "grad_norm": NaN,
      "learning_rate": 9.994534497576787e-05,
      "loss": 0.0,
      "step": 187
    },
    {
      "epoch": 0.03162055335968379,
      "grad_norm": NaN,
      "learning_rate": 9.994408155311642e-05,
      "loss": 0.0,
      "step": 188
    },
    {
      "epoch": 0.03178874779244807,
      "grad_norm": NaN,
      "learning_rate": 9.994280370217922e-05,
      "loss": 0.0,
      "step": 189
    },
    {
      "epoch": 0.031956942225212344,
      "grad_norm": NaN,
      "learning_rate": 9.99415114233254e-05,
      "loss": 0.0,
      "step": 190
    },
    {
      "epoch": 0.03212513665797662,
      "grad_norm": NaN,
      "learning_rate": 9.994020471692833e-05,
      "loss": 0.0,
      "step": 191
    },
    {
      "epoch": 0.032293331090740895,
      "grad_norm": NaN,
      "learning_rate": 9.993888358336545e-05,
      "loss": 0.0,
      "step": 192
    },
    {
      "epoch": 0.032461525523505175,
      "grad_norm": NaN,
      "learning_rate": 9.993754802301847e-05,
      "loss": 0.0,
      "step": 193
    },
    {
      "epoch": 0.03262971995626945,
      "grad_norm": NaN,
      "learning_rate": 9.993619803627321e-05,
      "loss": 0.0,
      "step": 194
    },
    {
      "epoch": 0.032797914389033726,
      "grad_norm": NaN,
      "learning_rate": 9.993483362351963e-05,
      "loss": 0.0,
      "step": 195
    },
    {
      "epoch": 0.032966108821798,
      "grad_norm": NaN,
      "learning_rate": 9.993345478515194e-05,
      "loss": 0.0,
      "step": 196
    },
    {
      "epoch": 0.03313430325456227,
      "grad_norm": NaN,
      "learning_rate": 9.993206152156844e-05,
      "loss": 0.0,
      "step": 197
    },
    {
      "epoch": 0.03330249768732655,
      "grad_norm": NaN,
      "learning_rate": 9.993065383317163e-05,
      "loss": 0.0,
      "step": 198
    },
    {
      "epoch": 0.03347069212009082,
      "grad_norm": NaN,
      "learning_rate": 9.992923172036819e-05,
      "loss": 0.0,
      "step": 199
    },
    {
      "epoch": 0.0336388865528551,
      "grad_norm": NaN,
      "learning_rate": 9.992779518356896e-05,
      "loss": 0.0,
      "step": 200
    },
    {
      "epoch": 0.033807080985619374,
      "grad_norm": NaN,
      "learning_rate": 9.99263442231889e-05,
      "loss": 0.0,
      "step": 201
    },
    {
      "epoch": 0.03397527541838365,
      "grad_norm": NaN,
      "learning_rate": 9.99248788396472e-05,
      "loss": 0.0,
      "step": 202
    },
    {
      "epoch": 0.034143469851147926,
      "grad_norm": NaN,
      "learning_rate": 9.992339903336722e-05,
      "loss": 0.0,
      "step": 203
    },
    {
      "epoch": 0.034311664283912205,
      "grad_norm": NaN,
      "learning_rate": 9.992190480477641e-05,
      "loss": 0.0,
      "step": 204
    },
    {
      "epoch": 0.03447985871667648,
      "grad_norm": NaN,
      "learning_rate": 9.992039615430648e-05,
      "loss": 0.0,
      "step": 205
    },
    {
      "epoch": 0.034648053149440756,
      "grad_norm": NaN,
      "learning_rate": 9.991887308239322e-05,
      "loss": 0.0,
      "step": 206
    },
    {
      "epoch": 0.03481624758220503,
      "grad_norm": NaN,
      "learning_rate": 9.991733558947667e-05,
      "loss": 0.0,
      "step": 207
    },
    {
      "epoch": 0.0349844420149693,
      "grad_norm": NaN,
      "learning_rate": 9.991578367600096e-05,
      "loss": 0.0,
      "step": 208
    },
    {
      "epoch": 0.03515263644773358,
      "grad_norm": NaN,
      "learning_rate": 9.991421734241444e-05,
      "loss": 0.0,
      "step": 209
    },
    {
      "epoch": 0.03532083088049785,
      "grad_norm": NaN,
      "learning_rate": 9.99126365891696e-05,
      "loss": 0.0,
      "step": 210
    },
    {
      "epoch": 0.03548902531326213,
      "grad_norm": NaN,
      "learning_rate": 9.991104141672309e-05,
      "loss": 0.0,
      "step": 211
    },
    {
      "epoch": 0.035657219746026404,
      "grad_norm": NaN,
      "learning_rate": 9.990943182553579e-05,
      "loss": 0.0,
      "step": 212
    },
    {
      "epoch": 0.035825414178790684,
      "grad_norm": NaN,
      "learning_rate": 9.990780781607261e-05,
      "loss": 0.0,
      "step": 213
    },
    {
      "epoch": 0.035993608611554956,
      "grad_norm": NaN,
      "learning_rate": 9.990616938880278e-05,
      "loss": 0.0,
      "step": 214
    },
    {
      "epoch": 0.036161803044319235,
      "grad_norm": NaN,
      "learning_rate": 9.990451654419957e-05,
      "loss": 0.0,
      "step": 215
    },
    {
      "epoch": 0.03632999747708351,
      "grad_norm": NaN,
      "learning_rate": 9.990284928274051e-05,
      "loss": 0.0,
      "step": 216
    },
    {
      "epoch": 0.03649819190984779,
      "grad_norm": NaN,
      "learning_rate": 9.990116760490723e-05,
      "loss": 0.0,
      "step": 217
    },
    {
      "epoch": 0.03666638634261206,
      "grad_norm": NaN,
      "learning_rate": 9.989947151118555e-05,
      "loss": 0.0,
      "step": 218
    },
    {
      "epoch": 0.03683458077537634,
      "grad_norm": NaN,
      "learning_rate": 9.989776100206548e-05,
      "loss": 0.0,
      "step": 219
    },
    {
      "epoch": 0.03700277520814061,
      "grad_norm": NaN,
      "learning_rate": 9.989603607804112e-05,
      "loss": 0.0,
      "step": 220
    },
    {
      "epoch": 0.03717096964090488,
      "grad_norm": NaN,
      "learning_rate": 9.98942967396108e-05,
      "loss": 0.0,
      "step": 221
    },
    {
      "epoch": 0.03733916407366916,
      "grad_norm": NaN,
      "learning_rate": 9.989254298727702e-05,
      "loss": 0.0,
      "step": 222
    },
    {
      "epoch": 0.037507358506433434,
      "grad_norm": NaN,
      "learning_rate": 9.989077482154638e-05,
      "loss": 0.0,
      "step": 223
    },
    {
      "epoch": 0.037675552939197714,
      "grad_norm": NaN,
      "learning_rate": 9.988899224292971e-05,
      "loss": 0.0,
      "step": 224
    },
    {
      "epoch": 0.037843747371961986,
      "grad_norm": NaN,
      "learning_rate": 9.988719525194198e-05,
      "loss": 0.0,
      "step": 225
    },
    {
      "epoch": 0.038011941804726265,
      "grad_norm": NaN,
      "learning_rate": 9.988538384910231e-05,
      "loss": 0.0,
      "step": 226
    },
    {
      "epoch": 0.03818013623749054,
      "grad_norm": NaN,
      "learning_rate": 9.988355803493398e-05,
      "loss": 0.0,
      "step": 227
    },
    {
      "epoch": 0.03834833067025482,
      "grad_norm": NaN,
      "learning_rate": 9.988171780996446e-05,
      "loss": 0.0,
      "step": 228
    },
    {
      "epoch": 0.03851652510301909,
      "grad_norm": NaN,
      "learning_rate": 9.987986317472539e-05,
      "loss": 0.0,
      "step": 229
    },
    {
      "epoch": 0.03868471953578337,
      "grad_norm": NaN,
      "learning_rate": 9.987799412975252e-05,
      "loss": 0.0,
      "step": 230
    },
    {
      "epoch": 0.03885291396854764,
      "grad_norm": NaN,
      "learning_rate": 9.987611067558582e-05,
      "loss": 0.0,
      "step": 231
    },
    {
      "epoch": 0.03902110840131192,
      "grad_norm": NaN,
      "learning_rate": 9.987421281276939e-05,
      "loss": 0.0,
      "step": 232
    },
    {
      "epoch": 0.03918930283407619,
      "grad_norm": NaN,
      "learning_rate": 9.98723005418515e-05,
      "loss": 0.0,
      "step": 233
    },
    {
      "epoch": 0.039357497266840465,
      "grad_norm": NaN,
      "learning_rate": 9.987037386338458e-05,
      "loss": 0.0,
      "step": 234
    },
    {
      "epoch": 0.039525691699604744,
      "grad_norm": NaN,
      "learning_rate": 9.986843277792523e-05,
      "loss": 0.0,
      "step": 235
    },
    {
      "epoch": 0.039693886132369016,
      "grad_norm": NaN,
      "learning_rate": 9.986647728603422e-05,
      "loss": 0.0,
      "step": 236
    },
    {
      "epoch": 0.039862080565133295,
      "grad_norm": NaN,
      "learning_rate": 9.986450738827646e-05,
      "loss": 0.0,
      "step": 237
    },
    {
      "epoch": 0.04003027499789757,
      "grad_norm": NaN,
      "learning_rate": 9.986252308522101e-05,
      "loss": 0.0,
      "step": 238
    },
    {
      "epoch": 0.04019846943066185,
      "grad_norm": NaN,
      "learning_rate": 9.986052437744115e-05,
      "loss": 0.0,
      "step": 239
    },
    {
      "epoch": 0.04036666386342612,
      "grad_norm": NaN,
      "learning_rate": 9.985851126551428e-05,
      "loss": 0.0,
      "step": 240
    },
    {
      "epoch": 0.0405348582961904,
      "grad_norm": NaN,
      "learning_rate": 9.985648375002192e-05,
      "loss": 0.0,
      "step": 241
    },
    {
      "epoch": 0.04070305272895467,
      "grad_norm": NaN,
      "learning_rate": 9.985444183154986e-05,
      "loss": 0.0,
      "step": 242
    },
    {
      "epoch": 0.04087124716171895,
      "grad_norm": NaN,
      "learning_rate": 9.985238551068794e-05,
      "loss": 0.0,
      "step": 243
    },
    {
      "epoch": 0.04103944159448322,
      "grad_norm": NaN,
      "learning_rate": 9.985031478803022e-05,
      "loss": 0.0,
      "step": 244
    },
    {
      "epoch": 0.041207636027247495,
      "grad_norm": NaN,
      "learning_rate": 9.984822966417494e-05,
      "loss": 0.0,
      "step": 245
    },
    {
      "epoch": 0.041375830460011774,
      "grad_norm": NaN,
      "learning_rate": 9.984613013972442e-05,
      "loss": 0.0,
      "step": 246
    },
    {
      "epoch": 0.041544024892776046,
      "grad_norm": NaN,
      "learning_rate": 9.98440162152852e-05,
      "loss": 0.0,
      "step": 247
    },
    {
      "epoch": 0.041712219325540326,
      "grad_norm": NaN,
      "learning_rate": 9.9841887891468e-05,
      "loss": 0.0,
      "step": 248
    },
    {
      "epoch": 0.0418804137583046,
      "grad_norm": NaN,
      "learning_rate": 9.983974516888763e-05,
      "loss": 0.0,
      "step": 249
    },
    {
      "epoch": 0.04204860819106888,
      "grad_norm": NaN,
      "learning_rate": 9.983758804816314e-05,
      "loss": 0.0,
      "step": 250
    },
    {
      "epoch": 0.04221680262383315,
      "grad_norm": NaN,
      "learning_rate": 9.983541652991766e-05,
      "loss": 0.0,
      "step": 251
    },
    {
      "epoch": 0.04238499705659743,
      "grad_norm": NaN,
      "learning_rate": 9.983323061477854e-05,
      "loss": 0.0,
      "step": 252
    },
    {
      "epoch": 0.0425531914893617,
      "grad_norm": NaN,
      "learning_rate": 9.983103030337727e-05,
      "loss": 0.0,
      "step": 253
    },
    {
      "epoch": 0.04272138592212598,
      "grad_norm": NaN,
      "learning_rate": 9.982881559634947e-05,
      "loss": 0.0,
      "step": 254
    },
    {
      "epoch": 0.04288958035489025,
      "grad_norm": NaN,
      "learning_rate": 9.982658649433497e-05,
      "loss": 0.0,
      "step": 255
    },
    {
      "epoch": 0.04305777478765453,
      "grad_norm": NaN,
      "learning_rate": 9.982434299797772e-05,
      "loss": 0.0,
      "step": 256
    },
    {
      "epoch": 0.043225969220418804,
      "grad_norm": NaN,
      "learning_rate": 9.982208510792583e-05,
      "loss": 0.0,
      "step": 257
    },
    {
      "epoch": 0.04339416365318308,
      "grad_norm": NaN,
      "learning_rate": 9.98198128248316e-05,
      "loss": 0.0,
      "step": 258
    },
    {
      "epoch": 0.043562358085947356,
      "grad_norm": NaN,
      "learning_rate": 9.981752614935145e-05,
      "loss": 0.0,
      "step": 259
    },
    {
      "epoch": 0.04373055251871163,
      "grad_norm": NaN,
      "learning_rate": 9.9815225082146e-05,
      "loss": 0.0,
      "step": 260
    },
    {
      "epoch": 0.04389874695147591,
      "grad_norm": NaN,
      "learning_rate": 9.981290962387998e-05,
      "loss": 0.0,
      "step": 261
    },
    {
      "epoch": 0.04406694138424018,
      "grad_norm": NaN,
      "learning_rate": 9.98105797752223e-05,
      "loss": 0.0,
      "step": 262
    },
    {
      "epoch": 0.04423513581700446,
      "grad_norm": NaN,
      "learning_rate": 9.980823553684604e-05,
      "loss": 0.0,
      "step": 263
    },
    {
      "epoch": 0.04440333024976873,
      "grad_norm": NaN,
      "learning_rate": 9.980587690942841e-05,
      "loss": 0.0,
      "step": 264
    },
    {
      "epoch": 0.04457152468253301,
      "grad_norm": NaN,
      "learning_rate": 9.980350389365081e-05,
      "loss": 0.0,
      "step": 265
    },
    {
      "epoch": 0.04473971911529728,
      "grad_norm": NaN,
      "learning_rate": 9.980111649019876e-05,
      "loss": 0.0,
      "step": 266
    },
    {
      "epoch": 0.04490791354806156,
      "grad_norm": NaN,
      "learning_rate": 9.979871469976196e-05,
      "loss": 0.0,
      "step": 267
    },
    {
      "epoch": 0.045076107980825834,
      "grad_norm": NaN,
      "learning_rate": 9.979629852303426e-05,
      "loss": 0.0,
      "step": 268
    },
    {
      "epoch": 0.04524430241359011,
      "grad_norm": NaN,
      "learning_rate": 9.979386796071366e-05,
      "loss": 0.0,
      "step": 269
    },
    {
      "epoch": 0.045412496846354386,
      "grad_norm": NaN,
      "learning_rate": 9.979142301350235e-05,
      "loss": 0.0,
      "step": 270
    },
    {
      "epoch": 0.04558069127911866,
      "grad_norm": NaN,
      "learning_rate": 9.978896368210662e-05,
      "loss": 0.0,
      "step": 271
    },
    {
      "epoch": 0.04574888571188294,
      "grad_norm": NaN,
      "learning_rate": 9.978648996723695e-05,
      "loss": 0.0,
      "step": 272
    },
    {
      "epoch": 0.04591708014464721,
      "grad_norm": NaN,
      "learning_rate": 9.978400186960795e-05,
      "loss": 0.0,
      "step": 273
    },
    {
      "epoch": 0.04608527457741149,
      "grad_norm": NaN,
      "learning_rate": 9.978149938993843e-05,
      "loss": 0.0,
      "step": 274
    },
    {
      "epoch": 0.04625346901017576,
      "grad_norm": NaN,
      "learning_rate": 9.977898252895134e-05,
      "loss": 0.0,
      "step": 275
    },
    {
      "epoch": 0.04642166344294004,
      "grad_norm": NaN,
      "learning_rate": 9.977645128737374e-05,
      "loss": 0.0,
      "step": 276
    },
    {
      "epoch": 0.04658985787570431,
      "grad_norm": NaN,
      "learning_rate": 9.97739056659369e-05,
      "loss": 0.0,
      "step": 277
    },
    {
      "epoch": 0.04675805230846859,
      "grad_norm": NaN,
      "learning_rate": 9.977134566537621e-05,
      "loss": 0.0,
      "step": 278
    },
    {
      "epoch": 0.046926246741232865,
      "grad_norm": NaN,
      "learning_rate": 9.976877128643122e-05,
      "loss": 0.0,
      "step": 279
    },
    {
      "epoch": 0.047094441173997144,
      "grad_norm": NaN,
      "learning_rate": 9.976618252984564e-05,
      "loss": 0.0,
      "step": 280
    },
    {
      "epoch": 0.047262635606761416,
      "grad_norm": NaN,
      "learning_rate": 9.976357939636735e-05,
      "loss": 0.0,
      "step": 281
    },
    {
      "epoch": 0.04743083003952569,
      "grad_norm": NaN,
      "learning_rate": 9.976096188674837e-05,
      "loss": 0.0,
      "step": 282
    },
    {
      "epoch": 0.04759902447228997,
      "grad_norm": NaN,
      "learning_rate": 9.975833000174485e-05,
      "loss": 0.0,
      "step": 283
    },
    {
      "epoch": 0.04776721890505424,
      "grad_norm": NaN,
      "learning_rate": 9.97556837421171e-05,
      "loss": 0.0,
      "step": 284
    },
    {
      "epoch": 0.04793541333781852,
      "grad_norm": NaN,
      "learning_rate": 9.975302310862964e-05,
      "loss": 0.0,
      "step": 285
    },
    {
      "epoch": 0.04810360777058279,
      "grad_norm": NaN,
      "learning_rate": 9.975034810205105e-05,
      "loss": 0.0,
      "step": 286
    },
    {
      "epoch": 0.04827180220334707,
      "grad_norm": NaN,
      "learning_rate": 9.974765872315414e-05,
      "loss": 0.0,
      "step": 287
    },
    {
      "epoch": 0.04843999663611134,
      "grad_norm": NaN,
      "learning_rate": 9.974495497271583e-05,
      "loss": 0.0,
      "step": 288
    },
    {
      "epoch": 0.04860819106887562,
      "grad_norm": NaN,
      "learning_rate": 9.97422368515172e-05,
      "loss": 0.0,
      "step": 289
    },
    {
      "epoch": 0.048776385501639895,
      "grad_norm": NaN,
      "learning_rate": 9.973950436034348e-05,
      "loss": 0.0,
      "step": 290
    },
    {
      "epoch": 0.048944579934404174,
      "grad_norm": NaN,
      "learning_rate": 9.973675749998407e-05,
      "loss": 0.0,
      "step": 291
    },
    {
      "epoch": 0.049112774367168446,
      "grad_norm": NaN,
      "learning_rate": 9.97339962712325e-05,
      "loss": 0.0,
      "step": 292
    },
    {
      "epoch": 0.049280968799932726,
      "grad_norm": NaN,
      "learning_rate": 9.973122067488647e-05,
      "loss": 0.0,
      "step": 293
    },
    {
      "epoch": 0.049449163232697,
      "grad_norm": NaN,
      "learning_rate": 9.97284307117478e-05,
      "loss": 0.0,
      "step": 294
    },
    {
      "epoch": 0.04961735766546127,
      "grad_norm": NaN,
      "learning_rate": 9.972562638262249e-05,
      "loss": 0.0,
      "step": 295
    },
    {
      "epoch": 0.04978555209822555,
      "grad_norm": NaN,
      "learning_rate": 9.972280768832068e-05,
      "loss": 0.0,
      "step": 296
    },
    {
      "epoch": 0.04995374653098982,
      "grad_norm": NaN,
      "learning_rate": 9.971997462965666e-05,
      "loss": 0.0,
      "step": 297
    },
    {
      "epoch": 0.0501219409637541,
      "grad_norm": NaN,
      "learning_rate": 9.971712720744885e-05,
      "loss": 0.0,
      "step": 298
    },
    {
      "epoch": 0.05029013539651837,
      "grad_norm": NaN,
      "learning_rate": 9.971426542251986e-05,
      "loss": 0.0,
      "step": 299
    },
    {
      "epoch": 0.05045832982928265,
      "grad_norm": NaN,
      "learning_rate": 9.971138927569642e-05,
      "loss": 0.0,
      "step": 300
    },
    {
      "epoch": 0.050626524262046925,
      "grad_norm": NaN,
      "learning_rate": 9.970849876780942e-05,
      "loss": 0.0,
      "step": 301
    },
    {
      "epoch": 0.050794718694811204,
      "grad_norm": NaN,
      "learning_rate": 9.970559389969389e-05,
      "loss": 0.0,
      "step": 302
    },
    {
      "epoch": 0.05096291312757548,
      "grad_norm": NaN,
      "learning_rate": 9.970267467218904e-05,
      "loss": 0.0,
      "step": 303
    },
    {
      "epoch": 0.051131107560339756,
      "grad_norm": NaN,
      "learning_rate": 9.969974108613816e-05,
      "loss": 0.0,
      "step": 304
    },
    {
      "epoch": 0.05129930199310403,
      "grad_norm": NaN,
      "learning_rate": 9.969679314238874e-05,
      "loss": 0.0,
      "step": 305
    },
    {
      "epoch": 0.0514674964258683,
      "grad_norm": NaN,
      "learning_rate": 9.969383084179243e-05,
      "loss": 0.0,
      "step": 306
    },
    {
      "epoch": 0.05163569085863258,
      "grad_norm": NaN,
      "learning_rate": 9.969085418520498e-05,
      "loss": 0.0,
      "step": 307
    },
    {
      "epoch": 0.05180388529139685,
      "grad_norm": NaN,
      "learning_rate": 9.968786317348632e-05,
      "loss": 0.0,
      "step": 308
    },
    {
      "epoch": 0.05197207972416113,
      "grad_norm": NaN,
      "learning_rate": 9.968485780750056e-05,
      "loss": 0.0,
      "step": 309
    },
    {
      "epoch": 0.052140274156925404,
      "grad_norm": NaN,
      "learning_rate": 9.968183808811586e-05,
      "loss": 0.0,
      "step": 310
    },
    {
      "epoch": 0.05230846858968968,
      "grad_norm": NaN,
      "learning_rate": 9.96788040162046e-05,
      "loss": 0.0,
      "step": 311
    },
    {
      "epoch": 0.052476663022453955,
      "grad_norm": NaN,
      "learning_rate": 9.96757555926433e-05,
      "loss": 0.0,
      "step": 312
    },
    {
      "epoch": 0.052644857455218234,
      "grad_norm": NaN,
      "learning_rate": 9.967269281831262e-05,
      "loss": 0.0,
      "step": 313
    },
    {
      "epoch": 0.05281305188798251,
      "grad_norm": NaN,
      "learning_rate": 9.966961569409735e-05,
      "loss": 0.0,
      "step": 314
    },
    {
      "epoch": 0.052981246320746786,
      "grad_norm": NaN,
      "learning_rate": 9.96665242208864e-05,
      "loss": 0.0,
      "step": 315
    },
    {
      "epoch": 0.05314944075351106,
      "grad_norm": NaN,
      "learning_rate": 9.966341839957293e-05,
      "loss": 0.0,
      "step": 316
    },
    {
      "epoch": 0.05331763518627534,
      "grad_norm": NaN,
      "learning_rate": 9.966029823105416e-05,
      "loss": 0.0,
      "step": 317
    },
    {
      "epoch": 0.05348582961903961,
      "grad_norm": NaN,
      "learning_rate": 9.965716371623143e-05,
      "loss": 0.0,
      "step": 318
    },
    {
      "epoch": 0.05365402405180388,
      "grad_norm": NaN,
      "learning_rate": 9.965401485601031e-05,
      "loss": 0.0,
      "step": 319
    },
    {
      "epoch": 0.05382221848456816,
      "grad_norm": NaN,
      "learning_rate": 9.965085165130046e-05,
      "loss": 0.0,
      "step": 320
    },
    {
      "epoch": 0.053990412917332434,
      "grad_norm": NaN,
      "learning_rate": 9.964767410301568e-05,
      "loss": 0.0,
      "step": 321
    },
    {
      "epoch": 0.05415860735009671,
      "grad_norm": NaN,
      "learning_rate": 9.964448221207392e-05,
      "loss": 0.0,
      "step": 322
    },
    {
      "epoch": 0.054326801782860985,
      "grad_norm": NaN,
      "learning_rate": 9.964127597939732e-05,
      "loss": 0.0,
      "step": 323
    },
    {
      "epoch": 0.054494996215625265,
      "grad_norm": NaN,
      "learning_rate": 9.963805540591211e-05,
      "loss": 0.0,
      "step": 324
    },
    {
      "epoch": 0.05466319064838954,
      "grad_norm": NaN,
      "learning_rate": 9.963482049254866e-05,
      "loss": 0.0,
      "step": 325
    },
    {
      "epoch": 0.054831385081153816,
      "grad_norm": NaN,
      "learning_rate": 9.963157124024152e-05,
      "loss": 0.0,
      "step": 326
    },
    {
      "epoch": 0.05499957951391809,
      "grad_norm": NaN,
      "learning_rate": 9.962830764992934e-05,
      "loss": 0.0,
      "step": 327
    },
    {
      "epoch": 0.05516777394668237,
      "grad_norm": NaN,
      "learning_rate": 9.962502972255497e-05,
      "loss": 0.0,
      "step": 328
    },
    {
      "epoch": 0.05533596837944664,
      "grad_norm": NaN,
      "learning_rate": 9.962173745906533e-05,
      "loss": 0.0,
      "step": 329
    },
    {
      "epoch": 0.05550416281221091,
      "grad_norm": NaN,
      "learning_rate": 9.961843086041154e-05,
      "loss": 0.0,
      "step": 330
    },
    {
      "epoch": 0.05567235724497519,
      "grad_norm": NaN,
      "learning_rate": 9.961510992754882e-05,
      "loss": 0.0,
      "step": 331
    },
    {
      "epoch": 0.055840551677739464,
      "grad_norm": NaN,
      "learning_rate": 9.961177466143658e-05,
      "loss": 0.0,
      "step": 332
    },
    {
      "epoch": 0.05600874611050374,
      "grad_norm": NaN,
      "learning_rate": 9.960842506303831e-05,
      "loss": 0.0,
      "step": 333
    },
    {
      "epoch": 0.056176940543268016,
      "grad_norm": NaN,
      "learning_rate": 9.960506113332169e-05,
      "loss": 0.0,
      "step": 334
    },
    {
      "epoch": 0.056345134976032295,
      "grad_norm": NaN,
      "learning_rate": 9.960168287325852e-05,
      "loss": 0.0,
      "step": 335
    },
    {
      "epoch": 0.05651332940879657,
      "grad_norm": NaN,
      "learning_rate": 9.959829028382473e-05,
      "loss": 0.0,
      "step": 336
    },
    {
      "epoch": 0.056681523841560846,
      "grad_norm": NaN,
      "learning_rate": 9.959488336600042e-05,
      "loss": 0.0,
      "step": 337
    },
    {
      "epoch": 0.05684971827432512,
      "grad_norm": NaN,
      "learning_rate": 9.95914621207698e-05,
      "loss": 0.0,
      "step": 338
    },
    {
      "epoch": 0.0570179127070894,
      "grad_norm": NaN,
      "learning_rate": 9.958802654912123e-05,
      "loss": 0.0,
      "step": 339
    },
    {
      "epoch": 0.05718610713985367,
      "grad_norm": NaN,
      "learning_rate": 9.95845766520472e-05,
      "loss": 0.0,
      "step": 340
    },
    {
      "epoch": 0.05735430157261795,
      "grad_norm": NaN,
      "learning_rate": 9.958111243054436e-05,
      "loss": 0.0,
      "step": 341
    },
    {
      "epoch": 0.05752249600538222,
      "grad_norm": NaN,
      "learning_rate": 9.957763388561347e-05,
      "loss": 0.0,
      "step": 342
    },
    {
      "epoch": 0.057690690438146494,
      "grad_norm": NaN,
      "learning_rate": 9.957414101825945e-05,
      "loss": 0.0,
      "step": 343
    },
    {
      "epoch": 0.05785888487091077,
      "grad_norm": NaN,
      "learning_rate": 9.957063382949137e-05,
      "loss": 0.0,
      "step": 344
    },
    {
      "epoch": 0.058027079303675046,
      "grad_norm": NaN,
      "learning_rate": 9.95671123203224e-05,
      "loss": 0.0,
      "step": 345
    },
    {
      "epoch": 0.058195273736439325,
      "grad_norm": NaN,
      "learning_rate": 9.956357649176984e-05,
      "loss": 0.0,
      "step": 346
    },
    {
      "epoch": 0.0583634681692036,
      "grad_norm": NaN,
      "learning_rate": 9.95600263448552e-05,
      "loss": 0.0,
      "step": 347
    },
    {
      "epoch": 0.058531662601967877,
      "grad_norm": NaN,
      "learning_rate": 9.955646188060404e-05,
      "loss": 0.0,
      "step": 348
    },
    {
      "epoch": 0.05869985703473215,
      "grad_norm": NaN,
      "learning_rate": 9.955288310004612e-05,
      "loss": 0.0,
      "step": 349
    },
    {
      "epoch": 0.05886805146749643,
      "grad_norm": NaN,
      "learning_rate": 9.95492900042153e-05,
      "loss": 0.0,
      "step": 350
    },
    {
      "epoch": 0.0590362459002607,
      "grad_norm": NaN,
      "learning_rate": 9.954568259414956e-05,
      "loss": 0.0,
      "step": 351
    },
    {
      "epoch": 0.05920444033302498,
      "grad_norm": NaN,
      "learning_rate": 9.954206087089107e-05,
      "loss": 0.0,
      "step": 352
    },
    {
      "epoch": 0.05937263476578925,
      "grad_norm": NaN,
      "learning_rate": 9.95384248354861e-05,
      "loss": 0.0,
      "step": 353
    },
    {
      "epoch": 0.05954082919855353,
      "grad_norm": NaN,
      "learning_rate": 9.953477448898506e-05,
      "loss": 0.0,
      "step": 354
    },
    {
      "epoch": 0.059709023631317804,
      "grad_norm": NaN,
      "learning_rate": 9.953110983244248e-05,
      "loss": 0.0,
      "step": 355
    },
    {
      "epoch": 0.059877218064082076,
      "grad_norm": NaN,
      "learning_rate": 9.952743086691705e-05,
      "loss": 0.0,
      "step": 356
    },
    {
      "epoch": 0.060045412496846355,
      "grad_norm": NaN,
      "learning_rate": 9.952373759347157e-05,
      "loss": 0.0,
      "step": 357
    },
    {
      "epoch": 0.06021360692961063,
      "grad_norm": NaN,
      "learning_rate": 9.952003001317303e-05,
      "loss": 0.0,
      "step": 358
    },
    {
      "epoch": 0.06038180136237491,
      "grad_norm": NaN,
      "learning_rate": 9.951630812709245e-05,
      "loss": 0.0,
      "step": 359
    },
    {
      "epoch": 0.06054999579513918,
      "grad_norm": NaN,
      "learning_rate": 9.951257193630507e-05,
      "loss": 0.0,
      "step": 360
    },
    {
      "epoch": 0.06071819022790346,
      "grad_norm": NaN,
      "learning_rate": 9.950882144189024e-05,
      "loss": 0.0,
      "step": 361
    },
    {
      "epoch": 0.06088638466066773,
      "grad_norm": NaN,
      "learning_rate": 9.950505664493144e-05,
      "loss": 0.0,
      "step": 362
    },
    {
      "epoch": 0.06105457909343201,
      "grad_norm": NaN,
      "learning_rate": 9.950127754651624e-05,
      "loss": 0.0,
      "step": 363
    },
    {
      "epoch": 0.06122277352619628,
      "grad_norm": NaN,
      "learning_rate": 9.949748414773641e-05,
      "loss": 0.0,
      "step": 364
    },
    {
      "epoch": 0.06139096795896056,
      "grad_norm": NaN,
      "learning_rate": 9.949367644968781e-05,
      "loss": 0.0,
      "step": 365
    },
    {
      "epoch": 0.061559162391724834,
      "grad_norm": NaN,
      "learning_rate": 9.948985445347046e-05,
      "loss": 0.0,
      "step": 366
    },
    {
      "epoch": 0.061727356824489106,
      "grad_norm": NaN,
      "learning_rate": 9.948601816018846e-05,
      "loss": 0.0,
      "step": 367
    },
    {
      "epoch": 0.061895551257253385,
      "grad_norm": NaN,
      "learning_rate": 9.948216757095008e-05,
      "loss": 0.0,
      "step": 368
    },
    {
      "epoch": 0.06206374569001766,
      "grad_norm": NaN,
      "learning_rate": 9.947830268686773e-05,
      "loss": 0.0,
      "step": 369
    },
    {
      "epoch": 0.06223194012278194,
      "grad_norm": NaN,
      "learning_rate": 9.947442350905793e-05,
      "loss": 0.0,
      "step": 370
    },
    {
      "epoch": 0.06240013455554621,
      "grad_norm": NaN,
      "learning_rate": 9.947053003864134e-05,
      "loss": 0.0,
      "step": 371
    },
    {
      "epoch": 0.06256832898831048,
      "grad_norm": NaN,
      "learning_rate": 9.946662227674268e-05,
      "loss": 0.0,
      "step": 372
    },
    {
      "epoch": 0.06273652342107476,
      "grad_norm": NaN,
      "learning_rate": 9.946270022449093e-05,
      "loss": 0.0,
      "step": 373
    },
    {
      "epoch": 0.06290471785383904,
      "grad_norm": NaN,
      "learning_rate": 9.945876388301909e-05,
      "loss": 0.0,
      "step": 374
    },
    {
      "epoch": 0.06307291228660332,
      "grad_norm": NaN,
      "learning_rate": 9.945481325346434e-05,
      "loss": 0.0,
      "step": 375
    },
    {
      "epoch": 0.06324110671936758,
      "grad_norm": NaN,
      "learning_rate": 9.945084833696797e-05,
      "loss": 0.0,
      "step": 376
    },
    {
      "epoch": 0.06340930115213186,
      "grad_norm": NaN,
      "learning_rate": 9.944686913467539e-05,
      "loss": 0.0,
      "step": 377
    },
    {
      "epoch": 0.06357749558489614,
      "grad_norm": NaN,
      "learning_rate": 9.944287564773615e-05,
      "loss": 0.0,
      "step": 378
    },
    {
      "epoch": 0.06374569001766041,
      "grad_norm": NaN,
      "learning_rate": 9.943886787730394e-05,
      "loss": 0.0,
      "step": 379
    },
    {
      "epoch": 0.06391388445042469,
      "grad_norm": NaN,
      "learning_rate": 9.943484582453653e-05,
      "loss": 0.0,
      "step": 380
    },
    {
      "epoch": 0.06408207888318897,
      "grad_norm": NaN,
      "learning_rate": 9.943080949059588e-05,
      "loss": 0.0,
      "step": 381
    },
    {
      "epoch": 0.06425027331595325,
      "grad_norm": NaN,
      "learning_rate": 9.942675887664802e-05,
      "loss": 0.0,
      "step": 382
    },
    {
      "epoch": 0.06441846774871751,
      "grad_norm": NaN,
      "learning_rate": 9.942269398386313e-05,
      "loss": 0.0,
      "step": 383
    },
    {
      "epoch": 0.06458666218148179,
      "grad_norm": NaN,
      "learning_rate": 9.941861481341551e-05,
      "loss": 0.0,
      "step": 384
    },
    {
      "epoch": 0.06475485661424607,
      "grad_norm": NaN,
      "learning_rate": 9.941452136648359e-05,
      "loss": 0.0,
      "step": 385
    },
    {
      "epoch": 0.06492305104701035,
      "grad_norm": NaN,
      "learning_rate": 9.941041364424993e-05,
      "loss": 0.0,
      "step": 386
    },
    {
      "epoch": 0.06509124547977461,
      "grad_norm": NaN,
      "learning_rate": 9.94062916479012e-05,
      "loss": 0.0,
      "step": 387
    },
    {
      "epoch": 0.0652594399125389,
      "grad_norm": NaN,
      "learning_rate": 9.940215537862818e-05,
      "loss": 0.0,
      "step": 388
    },
    {
      "epoch": 0.06542763434530317,
      "grad_norm": NaN,
      "learning_rate": 9.939800483762581e-05,
      "loss": 0.0,
      "step": 389
    },
    {
      "epoch": 0.06559582877806745,
      "grad_norm": NaN,
      "learning_rate": 9.939384002609315e-05,
      "loss": 0.0,
      "step": 390
    },
    {
      "epoch": 0.06576402321083172,
      "grad_norm": NaN,
      "learning_rate": 9.938966094523332e-05,
      "loss": 0.0,
      "step": 391
    },
    {
      "epoch": 0.065932217643596,
      "grad_norm": NaN,
      "learning_rate": 9.938546759625367e-05,
      "loss": 0.0,
      "step": 392
    },
    {
      "epoch": 0.06610041207636028,
      "grad_norm": NaN,
      "learning_rate": 9.938125998036555e-05,
      "loss": 0.0,
      "step": 393
    },
    {
      "epoch": 0.06626860650912454,
      "grad_norm": NaN,
      "learning_rate": 9.937703809878455e-05,
      "loss": 0.0,
      "step": 394
    },
    {
      "epoch": 0.06643680094188882,
      "grad_norm": NaN,
      "learning_rate": 9.93728019527303e-05,
      "loss": 0.0,
      "step": 395
    },
    {
      "epoch": 0.0666049953746531,
      "grad_norm": NaN,
      "learning_rate": 9.936855154342655e-05,
      "loss": 0.0,
      "step": 396
    },
    {
      "epoch": 0.06677318980741738,
      "grad_norm": NaN,
      "learning_rate": 9.936428687210122e-05,
      "loss": 0.0,
      "step": 397
    },
    {
      "epoch": 0.06694138424018165,
      "grad_norm": NaN,
      "learning_rate": 9.936000793998634e-05,
      "loss": 0.0,
      "step": 398
    },
    {
      "epoch": 0.06710957867294592,
      "grad_norm": NaN,
      "learning_rate": 9.935571474831801e-05,
      "loss": 0.0,
      "step": 399
    },
    {
      "epoch": 0.0672777731057102,
      "grad_norm": NaN,
      "learning_rate": 9.935140729833651e-05,
      "loss": 0.0,
      "step": 400
    },
    {
      "epoch": 0.06744596753847448,
      "grad_norm": NaN,
      "learning_rate": 9.934708559128622e-05,
      "loss": 0.0,
      "step": 401
    },
    {
      "epoch": 0.06761416197123875,
      "grad_norm": NaN,
      "learning_rate": 9.934274962841561e-05,
      "loss": 0.0,
      "step": 402
    },
    {
      "epoch": 0.06778235640400303,
      "grad_norm": NaN,
      "learning_rate": 9.933839941097732e-05,
      "loss": 0.0,
      "step": 403
    },
    {
      "epoch": 0.0679505508367673,
      "grad_norm": NaN,
      "learning_rate": 9.933403494022806e-05,
      "loss": 0.0,
      "step": 404
    },
    {
      "epoch": 0.06811874526953157,
      "grad_norm": NaN,
      "learning_rate": 9.932965621742867e-05,
      "loss": 0.0,
      "step": 405
    },
    {
      "epoch": 0.06828693970229585,
      "grad_norm": NaN,
      "learning_rate": 9.932526324384412e-05,
      "loss": 0.0,
      "step": 406
    },
    {
      "epoch": 0.06845513413506013,
      "grad_norm": NaN,
      "learning_rate": 9.93208560207435e-05,
      "loss": 0.0,
      "step": 407
    },
    {
      "epoch": 0.06862332856782441,
      "grad_norm": NaN,
      "learning_rate": 9.93164345494e-05,
      "loss": 0.0,
      "step": 408
    },
    {
      "epoch": 0.06879152300058868,
      "grad_norm": NaN,
      "learning_rate": 9.931199883109093e-05,
      "loss": 0.0,
      "step": 409
    },
    {
      "epoch": 0.06895971743335295,
      "grad_norm": NaN,
      "learning_rate": 9.930754886709772e-05,
      "loss": 0.0,
      "step": 410
    },
    {
      "epoch": 0.06912791186611723,
      "grad_norm": NaN,
      "learning_rate": 9.930308465870593e-05,
      "loss": 0.0,
      "step": 411
    },
    {
      "epoch": 0.06929610629888151,
      "grad_norm": NaN,
      "learning_rate": 9.92986062072052e-05,
      "loss": 0.0,
      "step": 412
    },
    {
      "epoch": 0.06946430073164578,
      "grad_norm": NaN,
      "learning_rate": 9.92941135138893e-05,
      "loss": 0.0,
      "step": 413
    },
    {
      "epoch": 0.06963249516441006,
      "grad_norm": NaN,
      "learning_rate": 9.928960658005617e-05,
      "loss": 0.0,
      "step": 414
    },
    {
      "epoch": 0.06980068959717434,
      "grad_norm": NaN,
      "learning_rate": 9.928508540700774e-05,
      "loss": 0.0,
      "step": 415
    },
    {
      "epoch": 0.0699688840299386,
      "grad_norm": NaN,
      "learning_rate": 9.928054999605018e-05,
      "loss": 0.0,
      "step": 416
    },
    {
      "epoch": 0.07013707846270288,
      "grad_norm": NaN,
      "learning_rate": 9.92760003484937e-05,
      "loss": 0.0,
      "step": 417
    },
    {
      "epoch": 0.07030527289546716,
      "grad_norm": NaN,
      "learning_rate": 9.927143646565262e-05,
      "loss": 0.0,
      "step": 418
    },
    {
      "epoch": 0.07047346732823144,
      "grad_norm": NaN,
      "learning_rate": 9.926685834884543e-05,
      "loss": 0.0,
      "step": 419
    },
    {
      "epoch": 0.0706416617609957,
      "grad_norm": NaN,
      "learning_rate": 9.926226599939468e-05,
      "loss": 0.0,
      "step": 420
    },
    {
      "epoch": 0.07080985619375998,
      "grad_norm": NaN,
      "learning_rate": 9.925765941862706e-05,
      "loss": 0.0,
      "step": 421
    },
    {
      "epoch": 0.07097805062652426,
      "grad_norm": NaN,
      "learning_rate": 9.925303860787335e-05,
      "loss": 0.0,
      "step": 422
    },
    {
      "epoch": 0.07114624505928854,
      "grad_norm": NaN,
      "learning_rate": 9.924840356846845e-05,
      "loss": 0.0,
      "step": 423
    },
    {
      "epoch": 0.07131443949205281,
      "grad_norm": NaN,
      "learning_rate": 9.924375430175136e-05,
      "loss": 0.0,
      "step": 424
    },
    {
      "epoch": 0.07148263392481709,
      "grad_norm": NaN,
      "learning_rate": 9.923909080906524e-05,
      "loss": 0.0,
      "step": 425
    },
    {
      "epoch": 0.07165082835758137,
      "grad_norm": NaN,
      "learning_rate": 9.923441309175727e-05,
      "loss": 0.0,
      "step": 426
    },
    {
      "epoch": 0.07181902279034565,
      "grad_norm": NaN,
      "learning_rate": 9.922972115117883e-05,
      "loss": 0.0,
      "step": 427
    },
    {
      "epoch": 0.07198721722310991,
      "grad_norm": NaN,
      "learning_rate": 9.922501498868535e-05,
      "loss": 0.0,
      "step": 428
    },
    {
      "epoch": 0.07215541165587419,
      "grad_norm": NaN,
      "learning_rate": 9.92202946056364e-05,
      "loss": 0.0,
      "step": 429
    },
    {
      "epoch": 0.07232360608863847,
      "grad_norm": NaN,
      "learning_rate": 9.921556000339563e-05,
      "loss": 0.0,
      "step": 430
    },
    {
      "epoch": 0.07249180052140274,
      "grad_norm": NaN,
      "learning_rate": 9.921081118333081e-05,
      "loss": 0.0,
      "step": 431
    },
    {
      "epoch": 0.07265999495416701,
      "grad_norm": NaN,
      "learning_rate": 9.920604814681385e-05,
      "loss": 0.0,
      "step": 432
    },
    {
      "epoch": 0.0728281893869313,
      "grad_norm": NaN,
      "learning_rate": 9.920127089522071e-05,
      "loss": 0.0,
      "step": 433
    },
    {
      "epoch": 0.07299638381969557,
      "grad_norm": NaN,
      "learning_rate": 9.919647942993148e-05,
      "loss": 0.0,
      "step": 434
    },
    {
      "epoch": 0.07316457825245984,
      "grad_norm": NaN,
      "learning_rate": 9.919167375233041e-05,
      "loss": 0.0,
      "step": 435
    },
    {
      "epoch": 0.07333277268522412,
      "grad_norm": NaN,
      "learning_rate": 9.918685386380573e-05,
      "loss": 0.0,
      "step": 436
    },
    {
      "epoch": 0.0735009671179884,
      "grad_norm": NaN,
      "learning_rate": 9.918201976574992e-05,
      "loss": 0.0,
      "step": 437
    },
    {
      "epoch": 0.07366916155075268,
      "grad_norm": NaN,
      "learning_rate": 9.917717145955947e-05,
      "loss": 0.0,
      "step": 438
    },
    {
      "epoch": 0.07383735598351694,
      "grad_norm": NaN,
      "learning_rate": 9.917230894663499e-05,
      "loss": 0.0,
      "step": 439
    },
    {
      "epoch": 0.07400555041628122,
      "grad_norm": NaN,
      "learning_rate": 9.916743222838123e-05,
      "loss": 0.0,
      "step": 440
    },
    {
      "epoch": 0.0741737448490455,
      "grad_norm": NaN,
      "learning_rate": 9.9162541306207e-05,
      "loss": 0.0,
      "step": 441
    },
    {
      "epoch": 0.07434193928180977,
      "grad_norm": NaN,
      "learning_rate": 9.915763618152522e-05,
      "loss": 0.0,
      "step": 442
    },
    {
      "epoch": 0.07451013371457404,
      "grad_norm": NaN,
      "learning_rate": 9.915271685575297e-05,
      "loss": 0.0,
      "step": 443
    },
    {
      "epoch": 0.07467832814733832,
      "grad_norm": NaN,
      "learning_rate": 9.914778333031135e-05,
      "loss": 0.0,
      "step": 444
    },
    {
      "epoch": 0.0748465225801026,
      "grad_norm": NaN,
      "learning_rate": 9.914283560662562e-05,
      "loss": 0.0,
      "step": 445
    },
    {
      "epoch": 0.07501471701286687,
      "grad_norm": NaN,
      "learning_rate": 9.91378736861251e-05,
      "loss": 0.0,
      "step": 446
    },
    {
      "epoch": 0.07518291144563115,
      "grad_norm": NaN,
      "learning_rate": 9.913289757024327e-05,
      "loss": 0.0,
      "step": 447
    },
    {
      "epoch": 0.07535110587839543,
      "grad_norm": NaN,
      "learning_rate": 9.912790726041763e-05,
      "loss": 0.0,
      "step": 448
    },
    {
      "epoch": 0.0755193003111597,
      "grad_norm": NaN,
      "learning_rate": 9.912290275808986e-05,
      "loss": 0.0,
      "step": 449
    },
    {
      "epoch": 0.07568749474392397,
      "grad_norm": NaN,
      "learning_rate": 9.911788406470569e-05,
      "loss": 0.0,
      "step": 450
    },
    {
      "epoch": 0.07585568917668825,
      "grad_norm": NaN,
      "learning_rate": 9.911285118171496e-05,
      "loss": 0.0,
      "step": 451
    },
    {
      "epoch": 0.07602388360945253,
      "grad_norm": NaN,
      "learning_rate": 9.910780411057164e-05,
      "loss": 0.0,
      "step": 452
    },
    {
      "epoch": 0.0761920780422168,
      "grad_norm": NaN,
      "learning_rate": 9.910274285273373e-05,
      "loss": 0.0,
      "step": 453
    },
    {
      "epoch": 0.07636027247498108,
      "grad_norm": NaN,
      "learning_rate": 9.909766740966339e-05,
      "loss": 0.0,
      "step": 454
    },
    {
      "epoch": 0.07652846690774535,
      "grad_norm": NaN,
      "learning_rate": 9.909257778282685e-05,
      "loss": 0.0,
      "step": 455
    },
    {
      "epoch": 0.07669666134050963,
      "grad_norm": NaN,
      "learning_rate": 9.908747397369449e-05,
      "loss": 0.0,
      "step": 456
    },
    {
      "epoch": 0.0768648557732739,
      "grad_norm": NaN,
      "learning_rate": 9.908235598374069e-05,
      "loss": 0.0,
      "step": 457
    },
    {
      "epoch": 0.07703305020603818,
      "grad_norm": NaN,
      "learning_rate": 9.9077223814444e-05,
      "loss": 0.0,
      "step": 458
    },
    {
      "epoch": 0.07720124463880246,
      "grad_norm": NaN,
      "learning_rate": 9.907207746728706e-05,
      "loss": 0.0,
      "step": 459
    },
    {
      "epoch": 0.07736943907156674,
      "grad_norm": NaN,
      "learning_rate": 9.906691694375656e-05,
      "loss": 0.0,
      "step": 460
    },
    {
      "epoch": 0.077537633504331,
      "grad_norm": NaN,
      "learning_rate": 9.906174224534335e-05,
      "loss": 0.0,
      "step": 461
    },
    {
      "epoch": 0.07770582793709528,
      "grad_norm": NaN,
      "learning_rate": 9.905655337354231e-05,
      "loss": 0.0,
      "step": 462
    },
    {
      "epoch": 0.07787402236985956,
      "grad_norm": NaN,
      "learning_rate": 9.905135032985249e-05,
      "loss": 0.0,
      "step": 463
    },
    {
      "epoch": 0.07804221680262384,
      "grad_norm": NaN,
      "learning_rate": 9.904613311577695e-05,
      "loss": 0.0,
      "step": 464
    },
    {
      "epoch": 0.0782104112353881,
      "grad_norm": NaN,
      "learning_rate": 9.90409017328229e-05,
      "loss": 0.0,
      "step": 465
    },
    {
      "epoch": 0.07837860566815238,
      "grad_norm": NaN,
      "learning_rate": 9.903565618250165e-05,
      "loss": 0.0,
      "step": 466
    },
    {
      "epoch": 0.07854680010091666,
      "grad_norm": NaN,
      "learning_rate": 9.903039646632855e-05,
      "loss": 0.0,
      "step": 467
    },
    {
      "epoch": 0.07871499453368093,
      "grad_norm": NaN,
      "learning_rate": 9.902512258582307e-05,
      "loss": 0.0,
      "step": 468
    },
    {
      "epoch": 0.07888318896644521,
      "grad_norm": NaN,
      "learning_rate": 9.901983454250878e-05,
      "loss": 0.0,
      "step": 469
    },
    {
      "epoch": 0.07905138339920949,
      "grad_norm": NaN,
      "learning_rate": 9.901453233791337e-05,
      "loss": 0.0,
      "step": 470
    },
    {
      "epoch": 0.07921957783197377,
      "grad_norm": NaN,
      "learning_rate": 9.900921597356856e-05,
      "loss": 0.0,
      "step": 471
    },
    {
      "epoch": 0.07938777226473803,
      "grad_norm": NaN,
      "learning_rate": 9.900388545101018e-05,
      "loss": 0.0,
      "step": 472
    },
    {
      "epoch": 0.07955596669750231,
      "grad_norm": NaN,
      "learning_rate": 9.899854077177816e-05,
      "loss": 0.0,
      "step": 473
    },
    {
      "epoch": 0.07972416113026659,
      "grad_norm": NaN,
      "learning_rate": 9.899318193741654e-05,
      "loss": 0.0,
      "step": 474
    },
    {
      "epoch": 0.07989235556303087,
      "grad_norm": NaN,
      "learning_rate": 9.89878089494734e-05,
      "loss": 0.0,
      "step": 475
    },
    {
      "epoch": 0.08006054999579514,
      "grad_norm": NaN,
      "learning_rate": 9.898242180950097e-05,
      "loss": 0.0,
      "step": 476
    },
    {
      "epoch": 0.08022874442855941,
      "grad_norm": NaN,
      "learning_rate": 9.89770205190555e-05,
      "loss": 0.0,
      "step": 477
    },
    {
      "epoch": 0.0803969388613237,
      "grad_norm": NaN,
      "learning_rate": 9.897160507969738e-05,
      "loss": 0.0,
      "step": 478
    },
    {
      "epoch": 0.08056513329408796,
      "grad_norm": NaN,
      "learning_rate": 9.896617549299106e-05,
      "loss": 0.0,
      "step": 479
    },
    {
      "epoch": 0.08073332772685224,
      "grad_norm": NaN,
      "learning_rate": 9.896073176050508e-05,
      "loss": 0.0,
      "step": 480
    },
    {
      "epoch": 0.08090152215961652,
      "grad_norm": NaN,
      "learning_rate": 9.895527388381212e-05,
      "loss": 0.0,
      "step": 481
    },
    {
      "epoch": 0.0810697165923808,
      "grad_norm": NaN,
      "learning_rate": 9.894980186448885e-05,
      "loss": 0.0,
      "step": 482
    },
    {
      "epoch": 0.08123791102514506,
      "grad_norm": NaN,
      "learning_rate": 9.894431570411608e-05,
      "loss": 0.0,
      "step": 483
    },
    {
      "epoch": 0.08140610545790934,
      "grad_norm": NaN,
      "learning_rate": 9.893881540427872e-05,
      "loss": 0.0,
      "step": 484
    },
    {
      "epoch": 0.08157429989067362,
      "grad_norm": NaN,
      "learning_rate": 9.893330096656574e-05,
      "loss": 0.0,
      "step": 485
    },
    {
      "epoch": 0.0817424943234379,
      "grad_norm": NaN,
      "learning_rate": 9.892777239257019e-05,
      "loss": 0.0,
      "step": 486
    },
    {
      "epoch": 0.08191068875620217,
      "grad_norm": NaN,
      "learning_rate": 9.892222968388922e-05,
      "loss": 0.0,
      "step": 487
    },
    {
      "epoch": 0.08207888318896644,
      "grad_norm": NaN,
      "learning_rate": 9.891667284212404e-05,
      "loss": 0.0,
      "step": 488
    },
    {
      "epoch": 0.08224707762173072,
      "grad_norm": NaN,
      "learning_rate": 9.891110186887999e-05,
      "loss": 0.0,
      "step": 489
    },
    {
      "epoch": 0.08241527205449499,
      "grad_norm": NaN,
      "learning_rate": 9.890551676576645e-05,
      "loss": 0.0,
      "step": 490
    },
    {
      "epoch": 0.08258346648725927,
      "grad_norm": NaN,
      "learning_rate": 9.889991753439689e-05,
      "loss": 0.0,
      "step": 491
    },
    {
      "epoch": 0.08275166092002355,
      "grad_norm": NaN,
      "learning_rate": 9.889430417638885e-05,
      "loss": 0.0,
      "step": 492
    },
    {
      "epoch": 0.08291985535278783,
      "grad_norm": NaN,
      "learning_rate": 9.888867669336399e-05,
      "loss": 0.0,
      "step": 493
    },
    {
      "epoch": 0.08308804978555209,
      "grad_norm": NaN,
      "learning_rate": 9.888303508694802e-05,
      "loss": 0.0,
      "step": 494
    },
    {
      "epoch": 0.08325624421831637,
      "grad_norm": NaN,
      "learning_rate": 9.887737935877072e-05,
      "loss": 0.0,
      "step": 495
    },
    {
      "epoch": 0.08342443865108065,
      "grad_norm": NaN,
      "learning_rate": 9.8871709510466e-05,
      "loss": 0.0,
      "step": 496
    },
    {
      "epoch": 0.08359263308384493,
      "grad_norm": NaN,
      "learning_rate": 9.886602554367179e-05,
      "loss": 0.0,
      "step": 497
    },
    {
      "epoch": 0.0837608275166092,
      "grad_norm": NaN,
      "learning_rate": 9.886032746003013e-05,
      "loss": 0.0,
      "step": 498
    },
    {
      "epoch": 0.08392902194937348,
      "grad_norm": NaN,
      "learning_rate": 9.885461526118713e-05,
      "loss": 0.0,
      "step": 499
    },
    {
      "epoch": 0.08409721638213775,
      "grad_norm": NaN,
      "learning_rate": 9.8848888948793e-05,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 0.08426541081490203,
      "grad_norm": NaN,
      "learning_rate": 9.884314852450198e-05,
      "loss": 0.0,
      "step": 501
    },
    {
      "epoch": 0.0844336052476663,
      "grad_norm": NaN,
      "learning_rate": 9.883739398997243e-05,
      "loss": 0.0,
      "step": 502
    },
    {
      "epoch": 0.08460179968043058,
      "grad_norm": NaN,
      "learning_rate": 9.883162534686675e-05,
      "loss": 0.0,
      "step": 503
    },
    {
      "epoch": 0.08476999411319486,
      "grad_norm": NaN,
      "learning_rate": 9.882584259685147e-05,
      "loss": 0.0,
      "step": 504
    },
    {
      "epoch": 0.08493818854595912,
      "grad_norm": NaN,
      "learning_rate": 9.882004574159713e-05,
      "loss": 0.0,
      "step": 505
    },
    {
      "epoch": 0.0851063829787234,
      "grad_norm": NaN,
      "learning_rate": 9.88142347827784e-05,
      "loss": 0.0,
      "step": 506
    },
    {
      "epoch": 0.08527457741148768,
      "grad_norm": NaN,
      "learning_rate": 9.8808409722074e-05,
      "loss": 0.0,
      "step": 507
    },
    {
      "epoch": 0.08544277184425196,
      "grad_norm": NaN,
      "learning_rate": 9.88025705611667e-05,
      "loss": 0.0,
      "step": 508
    },
    {
      "epoch": 0.08561096627701623,
      "grad_norm": NaN,
      "learning_rate": 9.87967173017434e-05,
      "loss": 0.0,
      "step": 509
    },
    {
      "epoch": 0.0857791607097805,
      "grad_norm": NaN,
      "learning_rate": 9.879084994549503e-05,
      "loss": 0.0,
      "step": 510
    },
    {
      "epoch": 0.08594735514254478,
      "grad_norm": NaN,
      "learning_rate": 9.878496849411659e-05,
      "loss": 0.0,
      "step": 511
    },
    {
      "epoch": 0.08611554957530906,
      "grad_norm": NaN,
      "learning_rate": 9.877907294930717e-05,
      "loss": 0.0,
      "step": 512
    },
    {
      "epoch": 0.08628374400807333,
      "grad_norm": NaN,
      "learning_rate": 9.877316331276995e-05,
      "loss": 0.0,
      "step": 513
    },
    {
      "epoch": 0.08645193844083761,
      "grad_norm": NaN,
      "learning_rate": 9.876723958621213e-05,
      "loss": 0.0,
      "step": 514
    },
    {
      "epoch": 0.08662013287360189,
      "grad_norm": NaN,
      "learning_rate": 9.8761301771345e-05,
      "loss": 0.0,
      "step": 515
    },
    {
      "epoch": 0.08678832730636615,
      "grad_norm": NaN,
      "learning_rate": 9.875534986988397e-05,
      "loss": 0.0,
      "step": 516
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": NaN,
      "learning_rate": 9.874938388354844e-05,
      "loss": 0.0,
      "step": 517
    },
    {
      "epoch": 0.08712471617189471,
      "grad_norm": NaN,
      "learning_rate": 9.874340381406194e-05,
      "loss": 0.0,
      "step": 518
    },
    {
      "epoch": 0.08729291060465899,
      "grad_norm": NaN,
      "learning_rate": 9.873740966315203e-05,
      "loss": 0.0,
      "step": 519
    },
    {
      "epoch": 0.08746110503742326,
      "grad_norm": NaN,
      "learning_rate": 9.873140143255036e-05,
      "loss": 0.0,
      "step": 520
    },
    {
      "epoch": 0.08762929947018754,
      "grad_norm": NaN,
      "learning_rate": 9.872537912399264e-05,
      "loss": 0.0,
      "step": 521
    },
    {
      "epoch": 0.08779749390295181,
      "grad_norm": NaN,
      "learning_rate": 9.871934273921864e-05,
      "loss": 0.0,
      "step": 522
    },
    {
      "epoch": 0.0879656883357161,
      "grad_norm": NaN,
      "learning_rate": 9.871329227997221e-05,
      "loss": 0.0,
      "step": 523
    },
    {
      "epoch": 0.08813388276848036,
      "grad_norm": NaN,
      "learning_rate": 9.870722774800127e-05,
      "loss": 0.0,
      "step": 524
    },
    {
      "epoch": 0.08830207720124464,
      "grad_norm": NaN,
      "learning_rate": 9.87011491450578e-05,
      "loss": 0.0,
      "step": 525
    },
    {
      "epoch": 0.08847027163400892,
      "grad_norm": NaN,
      "learning_rate": 9.86950564728978e-05,
      "loss": 0.0,
      "step": 526
    },
    {
      "epoch": 0.08863846606677318,
      "grad_norm": NaN,
      "learning_rate": 9.86889497332814e-05,
      "loss": 0.0,
      "step": 527
    },
    {
      "epoch": 0.08880666049953746,
      "grad_norm": NaN,
      "learning_rate": 9.868282892797279e-05,
      "loss": 0.0,
      "step": 528
    },
    {
      "epoch": 0.08897485493230174,
      "grad_norm": NaN,
      "learning_rate": 9.867669405874016e-05,
      "loss": 0.0,
      "step": 529
    },
    {
      "epoch": 0.08914304936506602,
      "grad_norm": NaN,
      "learning_rate": 9.867054512735584e-05,
      "loss": 0.0,
      "step": 530
    },
    {
      "epoch": 0.08931124379783029,
      "grad_norm": NaN,
      "learning_rate": 9.866438213559617e-05,
      "loss": 0.0,
      "step": 531
    },
    {
      "epoch": 0.08947943823059457,
      "grad_norm": NaN,
      "learning_rate": 9.865820508524157e-05,
      "loss": 0.0,
      "step": 532
    },
    {
      "epoch": 0.08964763266335884,
      "grad_norm": NaN,
      "learning_rate": 9.865201397807653e-05,
      "loss": 0.0,
      "step": 533
    },
    {
      "epoch": 0.08981582709612312,
      "grad_norm": NaN,
      "learning_rate": 9.864580881588959e-05,
      "loss": 0.0,
      "step": 534
    },
    {
      "epoch": 0.08998402152888739,
      "grad_norm": NaN,
      "learning_rate": 9.863958960047335e-05,
      "loss": 0.0,
      "step": 535
    },
    {
      "epoch": 0.09015221596165167,
      "grad_norm": NaN,
      "learning_rate": 9.863335633362445e-05,
      "loss": 0.0,
      "step": 536
    },
    {
      "epoch": 0.09032041039441595,
      "grad_norm": NaN,
      "learning_rate": 9.862710901714365e-05,
      "loss": 0.0,
      "step": 537
    },
    {
      "epoch": 0.09048860482718021,
      "grad_norm": NaN,
      "learning_rate": 9.86208476528357e-05,
      "loss": 0.0,
      "step": 538
    },
    {
      "epoch": 0.09065679925994449,
      "grad_norm": NaN,
      "learning_rate": 9.861457224250944e-05,
      "loss": 0.0,
      "step": 539
    },
    {
      "epoch": 0.09082499369270877,
      "grad_norm": NaN,
      "learning_rate": 9.860828278797778e-05,
      "loss": 0.0,
      "step": 540
    },
    {
      "epoch": 0.09099318812547305,
      "grad_norm": NaN,
      "learning_rate": 9.860197929105768e-05,
      "loss": 0.0,
      "step": 541
    },
    {
      "epoch": 0.09116138255823732,
      "grad_norm": NaN,
      "learning_rate": 9.859566175357011e-05,
      "loss": 0.0,
      "step": 542
    },
    {
      "epoch": 0.0913295769910016,
      "grad_norm": NaN,
      "learning_rate": 9.858933017734018e-05,
      "loss": 0.0,
      "step": 543
    },
    {
      "epoch": 0.09149777142376588,
      "grad_norm": NaN,
      "learning_rate": 9.858298456419699e-05,
      "loss": 0.0,
      "step": 544
    },
    {
      "epoch": 0.09166596585653015,
      "grad_norm": NaN,
      "learning_rate": 9.857662491597373e-05,
      "loss": 0.0,
      "step": 545
    },
    {
      "epoch": 0.09183416028929442,
      "grad_norm": NaN,
      "learning_rate": 9.85702512345076e-05,
      "loss": 0.0,
      "step": 546
    },
    {
      "epoch": 0.0920023547220587,
      "grad_norm": NaN,
      "learning_rate": 9.856386352163992e-05,
      "loss": 0.0,
      "step": 547
    },
    {
      "epoch": 0.09217054915482298,
      "grad_norm": NaN,
      "learning_rate": 9.8557461779216e-05,
      "loss": 0.0,
      "step": 548
    },
    {
      "epoch": 0.09233874358758726,
      "grad_norm": NaN,
      "learning_rate": 9.855104600908527e-05,
      "loss": 0.0,
      "step": 549
    },
    {
      "epoch": 0.09250693802035152,
      "grad_norm": NaN,
      "learning_rate": 9.854461621310113e-05,
      "loss": 0.0,
      "step": 550
    },
    {
      "epoch": 0.0926751324531158,
      "grad_norm": NaN,
      "learning_rate": 9.853817239312112e-05,
      "loss": 0.0,
      "step": 551
    },
    {
      "epoch": 0.09284332688588008,
      "grad_norm": NaN,
      "learning_rate": 9.853171455100675e-05,
      "loss": 0.0,
      "step": 552
    },
    {
      "epoch": 0.09301152131864435,
      "grad_norm": NaN,
      "learning_rate": 9.852524268862363e-05,
      "loss": 0.0,
      "step": 553
    },
    {
      "epoch": 0.09317971575140863,
      "grad_norm": NaN,
      "learning_rate": 9.851875680784142e-05,
      "loss": 0.0,
      "step": 554
    },
    {
      "epoch": 0.0933479101841729,
      "grad_norm": NaN,
      "learning_rate": 9.85122569105338e-05,
      "loss": 0.0,
      "step": 555
    },
    {
      "epoch": 0.09351610461693718,
      "grad_norm": NaN,
      "learning_rate": 9.850574299857854e-05,
      "loss": 0.0,
      "step": 556
    },
    {
      "epoch": 0.09368429904970145,
      "grad_norm": NaN,
      "learning_rate": 9.849921507385741e-05,
      "loss": 0.0,
      "step": 557
    },
    {
      "epoch": 0.09385249348246573,
      "grad_norm": NaN,
      "learning_rate": 9.849267313825628e-05,
      "loss": 0.0,
      "step": 558
    },
    {
      "epoch": 0.09402068791523001,
      "grad_norm": NaN,
      "learning_rate": 9.8486117193665e-05,
      "loss": 0.0,
      "step": 559
    },
    {
      "epoch": 0.09418888234799429,
      "grad_norm": NaN,
      "learning_rate": 9.847954724197757e-05,
      "loss": 0.0,
      "step": 560
    },
    {
      "epoch": 0.09435707678075855,
      "grad_norm": NaN,
      "learning_rate": 9.847296328509194e-05,
      "loss": 0.0,
      "step": 561
    },
    {
      "epoch": 0.09452527121352283,
      "grad_norm": NaN,
      "learning_rate": 9.846636532491014e-05,
      "loss": 0.0,
      "step": 562
    },
    {
      "epoch": 0.09469346564628711,
      "grad_norm": NaN,
      "learning_rate": 9.845975336333825e-05,
      "loss": 0.0,
      "step": 563
    },
    {
      "epoch": 0.09486166007905138,
      "grad_norm": NaN,
      "learning_rate": 9.845312740228638e-05,
      "loss": 0.0,
      "step": 564
    },
    {
      "epoch": 0.09502985451181566,
      "grad_norm": NaN,
      "learning_rate": 9.844648744366873e-05,
      "loss": 0.0,
      "step": 565
    },
    {
      "epoch": 0.09519804894457994,
      "grad_norm": NaN,
      "learning_rate": 9.843983348940348e-05,
      "loss": 0.0,
      "step": 566
    },
    {
      "epoch": 0.09536624337734421,
      "grad_norm": NaN,
      "learning_rate": 9.84331655414129e-05,
      "loss": 0.0,
      "step": 567
    },
    {
      "epoch": 0.09553443781010848,
      "grad_norm": NaN,
      "learning_rate": 9.842648360162327e-05,
      "loss": 0.0,
      "step": 568
    },
    {
      "epoch": 0.09570263224287276,
      "grad_norm": NaN,
      "learning_rate": 9.841978767196495e-05,
      "loss": 0.0,
      "step": 569
    },
    {
      "epoch": 0.09587082667563704,
      "grad_norm": NaN,
      "learning_rate": 9.84130777543723e-05,
      "loss": 0.0,
      "step": 570
    },
    {
      "epoch": 0.09603902110840132,
      "grad_norm": NaN,
      "learning_rate": 9.840635385078372e-05,
      "loss": 0.0,
      "step": 571
    },
    {
      "epoch": 0.09620721554116558,
      "grad_norm": NaN,
      "learning_rate": 9.839961596314172e-05,
      "loss": 0.0,
      "step": 572
    },
    {
      "epoch": 0.09637540997392986,
      "grad_norm": NaN,
      "learning_rate": 9.839286409339276e-05,
      "loss": 0.0,
      "step": 573
    },
    {
      "epoch": 0.09654360440669414,
      "grad_norm": NaN,
      "learning_rate": 9.838609824348742e-05,
      "loss": 0.0,
      "step": 574
    },
    {
      "epoch": 0.09671179883945841,
      "grad_norm": NaN,
      "learning_rate": 9.837931841538022e-05,
      "loss": 0.0,
      "step": 575
    },
    {
      "epoch": 0.09687999327222269,
      "grad_norm": NaN,
      "learning_rate": 9.837252461102981e-05,
      "loss": 0.0,
      "step": 576
    },
    {
      "epoch": 0.09704818770498697,
      "grad_norm": NaN,
      "learning_rate": 9.836571683239887e-05,
      "loss": 0.0,
      "step": 577
    },
    {
      "epoch": 0.09721638213775124,
      "grad_norm": NaN,
      "learning_rate": 9.835889508145405e-05,
      "loss": 0.0,
      "step": 578
    },
    {
      "epoch": 0.09738457657051551,
      "grad_norm": NaN,
      "learning_rate": 9.83520593601661e-05,
      "loss": 0.0,
      "step": 579
    },
    {
      "epoch": 0.09755277100327979,
      "grad_norm": NaN,
      "learning_rate": 9.834520967050975e-05,
      "loss": 0.0,
      "step": 580
    },
    {
      "epoch": 0.09772096543604407,
      "grad_norm": NaN,
      "learning_rate": 9.833834601446386e-05,
      "loss": 0.0,
      "step": 581
    },
    {
      "epoch": 0.09788915986880835,
      "grad_norm": NaN,
      "learning_rate": 9.833146839401119e-05,
      "loss": 0.0,
      "step": 582
    },
    {
      "epoch": 0.09805735430157261,
      "grad_norm": NaN,
      "learning_rate": 9.832457681113866e-05,
      "loss": 0.0,
      "step": 583
    },
    {
      "epoch": 0.09822554873433689,
      "grad_norm": NaN,
      "learning_rate": 9.831767126783717e-05,
      "loss": 0.0,
      "step": 584
    },
    {
      "epoch": 0.09839374316710117,
      "grad_norm": NaN,
      "learning_rate": 9.831075176610163e-05,
      "loss": 0.0,
      "step": 585
    },
    {
      "epoch": 0.09856193759986545,
      "grad_norm": NaN,
      "learning_rate": 9.8303818307931e-05,
      "loss": 0.0,
      "step": 586
    },
    {
      "epoch": 0.09873013203262972,
      "grad_norm": NaN,
      "learning_rate": 9.829687089532831e-05,
      "loss": 0.0,
      "step": 587
    },
    {
      "epoch": 0.098898326465394,
      "grad_norm": NaN,
      "learning_rate": 9.828990953030058e-05,
      "loss": 0.0,
      "step": 588
    },
    {
      "epoch": 0.09906652089815828,
      "grad_norm": NaN,
      "learning_rate": 9.828293421485885e-05,
      "loss": 0.0,
      "step": 589
    },
    {
      "epoch": 0.09923471533092254,
      "grad_norm": NaN,
      "learning_rate": 9.827594495101823e-05,
      "loss": 0.0,
      "step": 590
    },
    {
      "epoch": 0.09940290976368682,
      "grad_norm": NaN,
      "learning_rate": 9.826894174079785e-05,
      "loss": 0.0,
      "step": 591
    },
    {
      "epoch": 0.0995711041964511,
      "grad_norm": NaN,
      "learning_rate": 9.826192458622083e-05,
      "loss": 0.0,
      "step": 592
    },
    {
      "epoch": 0.09973929862921538,
      "grad_norm": NaN,
      "learning_rate": 9.825489348931435e-05,
      "loss": 0.0,
      "step": 593
    },
    {
      "epoch": 0.09990749306197964,
      "grad_norm": NaN,
      "learning_rate": 9.824784845210966e-05,
      "loss": 0.0,
      "step": 594
    },
    {
      "epoch": 0.10007568749474392,
      "grad_norm": NaN,
      "learning_rate": 9.824078947664193e-05,
      "loss": 0.0,
      "step": 595
    },
    {
      "epoch": 0.1002438819275082,
      "grad_norm": NaN,
      "learning_rate": 9.823371656495046e-05,
      "loss": 0.0,
      "step": 596
    },
    {
      "epoch": 0.10041207636027248,
      "grad_norm": NaN,
      "learning_rate": 9.822662971907852e-05,
      "loss": 0.0,
      "step": 597
    },
    {
      "epoch": 0.10058027079303675,
      "grad_norm": NaN,
      "learning_rate": 9.821952894107343e-05,
      "loss": 0.0,
      "step": 598
    },
    {
      "epoch": 0.10074846522580103,
      "grad_norm": NaN,
      "learning_rate": 9.821241423298651e-05,
      "loss": 0.0,
      "step": 599
    },
    {
      "epoch": 0.1009166596585653,
      "grad_norm": NaN,
      "learning_rate": 9.820528559687312e-05,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 0.10108485409132957,
      "grad_norm": NaN,
      "learning_rate": 9.819814303479267e-05,
      "loss": 0.0,
      "step": 601
    },
    {
      "epoch": 0.10125304852409385,
      "grad_norm": NaN,
      "learning_rate": 9.819098654880854e-05,
      "loss": 0.0,
      "step": 602
    },
    {
      "epoch": 0.10142124295685813,
      "grad_norm": NaN,
      "learning_rate": 9.818381614098817e-05,
      "loss": 0.0,
      "step": 603
    },
    {
      "epoch": 0.10158943738962241,
      "grad_norm": NaN,
      "learning_rate": 9.817663181340299e-05,
      "loss": 0.0,
      "step": 604
    },
    {
      "epoch": 0.10175763182238667,
      "grad_norm": NaN,
      "learning_rate": 9.81694335681285e-05,
      "loss": 0.0,
      "step": 605
    },
    {
      "epoch": 0.10192582625515095,
      "grad_norm": NaN,
      "learning_rate": 9.816222140724418e-05,
      "loss": 0.0,
      "step": 606
    },
    {
      "epoch": 0.10209402068791523,
      "grad_norm": NaN,
      "learning_rate": 9.815499533283354e-05,
      "loss": 0.0,
      "step": 607
    },
    {
      "epoch": 0.10226221512067951,
      "grad_norm": NaN,
      "learning_rate": 9.814775534698413e-05,
      "loss": 0.0,
      "step": 608
    },
    {
      "epoch": 0.10243040955344378,
      "grad_norm": NaN,
      "learning_rate": 9.814050145178747e-05,
      "loss": 0.0,
      "step": 609
    },
    {
      "epoch": 0.10259860398620806,
      "grad_norm": NaN,
      "learning_rate": 9.813323364933915e-05,
      "loss": 0.0,
      "step": 610
    },
    {
      "epoch": 0.10276679841897234,
      "grad_norm": NaN,
      "learning_rate": 9.812595194173875e-05,
      "loss": 0.0,
      "step": 611
    },
    {
      "epoch": 0.1029349928517366,
      "grad_norm": NaN,
      "learning_rate": 9.811865633108987e-05,
      "loss": 0.0,
      "step": 612
    },
    {
      "epoch": 0.10310318728450088,
      "grad_norm": NaN,
      "learning_rate": 9.811134681950016e-05,
      "loss": 0.0,
      "step": 613
    },
    {
      "epoch": 0.10327138171726516,
      "grad_norm": NaN,
      "learning_rate": 9.810402340908121e-05,
      "loss": 0.0,
      "step": 614
    },
    {
      "epoch": 0.10343957615002944,
      "grad_norm": NaN,
      "learning_rate": 9.80966861019487e-05,
      "loss": 0.0,
      "step": 615
    },
    {
      "epoch": 0.1036077705827937,
      "grad_norm": NaN,
      "learning_rate": 9.808933490022229e-05,
      "loss": 0.0,
      "step": 616
    },
    {
      "epoch": 0.10377596501555798,
      "grad_norm": NaN,
      "learning_rate": 9.808196980602566e-05,
      "loss": 0.0,
      "step": 617
    },
    {
      "epoch": 0.10394415944832226,
      "grad_norm": NaN,
      "learning_rate": 9.807459082148648e-05,
      "loss": 0.0,
      "step": 618
    },
    {
      "epoch": 0.10411235388108654,
      "grad_norm": NaN,
      "learning_rate": 9.806719794873651e-05,
      "loss": 0.0,
      "step": 619
    },
    {
      "epoch": 0.10428054831385081,
      "grad_norm": NaN,
      "learning_rate": 9.805979118991142e-05,
      "loss": 0.0,
      "step": 620
    },
    {
      "epoch": 0.10444874274661509,
      "grad_norm": NaN,
      "learning_rate": 9.805237054715095e-05,
      "loss": 0.0,
      "step": 621
    },
    {
      "epoch": 0.10461693717937937,
      "grad_norm": NaN,
      "learning_rate": 9.804493602259885e-05,
      "loss": 0.0,
      "step": 622
    },
    {
      "epoch": 0.10478513161214364,
      "grad_norm": NaN,
      "learning_rate": 9.803748761840287e-05,
      "loss": 0.0,
      "step": 623
    },
    {
      "epoch": 0.10495332604490791,
      "grad_norm": NaN,
      "learning_rate": 9.803002533671478e-05,
      "loss": 0.0,
      "step": 624
    },
    {
      "epoch": 0.10512152047767219,
      "grad_norm": NaN,
      "learning_rate": 9.802254917969032e-05,
      "loss": 0.0,
      "step": 625
    },
    {
      "epoch": 0.10528971491043647,
      "grad_norm": NaN,
      "learning_rate": 9.801505914948929e-05,
      "loss": 0.0,
      "step": 626
    },
    {
      "epoch": 0.10545790934320073,
      "grad_norm": NaN,
      "learning_rate": 9.800755524827548e-05,
      "loss": 0.0,
      "step": 627
    },
    {
      "epoch": 0.10562610377596501,
      "grad_norm": NaN,
      "learning_rate": 9.800003747821667e-05,
      "loss": 0.0,
      "step": 628
    },
    {
      "epoch": 0.10579429820872929,
      "grad_norm": NaN,
      "learning_rate": 9.799250584148466e-05,
      "loss": 0.0,
      "step": 629
    },
    {
      "epoch": 0.10596249264149357,
      "grad_norm": NaN,
      "learning_rate": 9.798496034025526e-05,
      "loss": 0.0,
      "step": 630
    },
    {
      "epoch": 0.10613068707425784,
      "grad_norm": NaN,
      "learning_rate": 9.797740097670827e-05,
      "loss": 0.0,
      "step": 631
    },
    {
      "epoch": 0.10629888150702212,
      "grad_norm": NaN,
      "learning_rate": 9.796982775302755e-05,
      "loss": 0.0,
      "step": 632
    },
    {
      "epoch": 0.1064670759397864,
      "grad_norm": NaN,
      "learning_rate": 9.796224067140085e-05,
      "loss": 0.0,
      "step": 633
    },
    {
      "epoch": 0.10663527037255068,
      "grad_norm": NaN,
      "learning_rate": 9.795463973402003e-05,
      "loss": 0.0,
      "step": 634
    },
    {
      "epoch": 0.10680346480531494,
      "grad_norm": NaN,
      "learning_rate": 9.794702494308093e-05,
      "loss": 0.0,
      "step": 635
    },
    {
      "epoch": 0.10697165923807922,
      "grad_norm": NaN,
      "learning_rate": 9.793939630078335e-05,
      "loss": 0.0,
      "step": 636
    },
    {
      "epoch": 0.1071398536708435,
      "grad_norm": NaN,
      "learning_rate": 9.793175380933113e-05,
      "loss": 0.0,
      "step": 637
    },
    {
      "epoch": 0.10730804810360776,
      "grad_norm": NaN,
      "learning_rate": 9.792409747093212e-05,
      "loss": 0.0,
      "step": 638
    },
    {
      "epoch": 0.10747624253637204,
      "grad_norm": NaN,
      "learning_rate": 9.791642728779811e-05,
      "loss": 0.0,
      "step": 639
    },
    {
      "epoch": 0.10764443696913632,
      "grad_norm": NaN,
      "learning_rate": 9.790874326214496e-05,
      "loss": 0.0,
      "step": 640
    },
    {
      "epoch": 0.1078126314019006,
      "grad_norm": NaN,
      "learning_rate": 9.790104539619248e-05,
      "loss": 0.0,
      "step": 641
    },
    {
      "epoch": 0.10798082583466487,
      "grad_norm": NaN,
      "learning_rate": 9.789333369216451e-05,
      "loss": 0.0,
      "step": 642
    },
    {
      "epoch": 0.10814902026742915,
      "grad_norm": NaN,
      "learning_rate": 9.788560815228889e-05,
      "loss": 0.0,
      "step": 643
    },
    {
      "epoch": 0.10831721470019343,
      "grad_norm": NaN,
      "learning_rate": 9.787786877879741e-05,
      "loss": 0.0,
      "step": 644
    },
    {
      "epoch": 0.1084854091329577,
      "grad_norm": NaN,
      "learning_rate": 9.78701155739259e-05,
      "loss": 0.0,
      "step": 645
    },
    {
      "epoch": 0.10865360356572197,
      "grad_norm": NaN,
      "learning_rate": 9.786234853991418e-05,
      "loss": 0.0,
      "step": 646
    },
    {
      "epoch": 0.10882179799848625,
      "grad_norm": NaN,
      "learning_rate": 9.785456767900607e-05,
      "loss": 0.0,
      "step": 647
    },
    {
      "epoch": 0.10898999243125053,
      "grad_norm": NaN,
      "learning_rate": 9.784677299344934e-05,
      "loss": 0.0,
      "step": 648
    },
    {
      "epoch": 0.1091581868640148,
      "grad_norm": NaN,
      "learning_rate": 9.78389644854958e-05,
      "loss": 0.0,
      "step": 649
    },
    {
      "epoch": 0.10932638129677907,
      "grad_norm": NaN,
      "learning_rate": 9.783114215740128e-05,
      "loss": 0.0,
      "step": 650
    },
    {
      "epoch": 0.10949457572954335,
      "grad_norm": NaN,
      "learning_rate": 9.782330601142551e-05,
      "loss": 0.0,
      "step": 651
    },
    {
      "epoch": 0.10966277016230763,
      "grad_norm": NaN,
      "learning_rate": 9.781545604983228e-05,
      "loss": 0.0,
      "step": 652
    },
    {
      "epoch": 0.1098309645950719,
      "grad_norm": NaN,
      "learning_rate": 9.780759227488936e-05,
      "loss": 0.0,
      "step": 653
    },
    {
      "epoch": 0.10999915902783618,
      "grad_norm": NaN,
      "learning_rate": 9.779971468886853e-05,
      "loss": 0.0,
      "step": 654
    },
    {
      "epoch": 0.11016735346060046,
      "grad_norm": NaN,
      "learning_rate": 9.779182329404549e-05,
      "loss": 0.0,
      "step": 655
    },
    {
      "epoch": 0.11033554789336474,
      "grad_norm": NaN,
      "learning_rate": 9.778391809270002e-05,
      "loss": 0.0,
      "step": 656
    },
    {
      "epoch": 0.110503742326129,
      "grad_norm": NaN,
      "learning_rate": 9.77759990871158e-05,
      "loss": 0.0,
      "step": 657
    },
    {
      "epoch": 0.11067193675889328,
      "grad_norm": NaN,
      "learning_rate": 9.776806627958057e-05,
      "loss": 0.0,
      "step": 658
    },
    {
      "epoch": 0.11084013119165756,
      "grad_norm": NaN,
      "learning_rate": 9.776011967238604e-05,
      "loss": 0.0,
      "step": 659
    },
    {
      "epoch": 0.11100832562442182,
      "grad_norm": NaN,
      "learning_rate": 9.775215926782788e-05,
      "loss": 0.0,
      "step": 660
    },
    {
      "epoch": 0.1111765200571861,
      "grad_norm": NaN,
      "learning_rate": 9.774418506820575e-05,
      "loss": 0.0,
      "step": 661
    },
    {
      "epoch": 0.11134471448995038,
      "grad_norm": NaN,
      "learning_rate": 9.773619707582333e-05,
      "loss": 0.0,
      "step": 662
    },
    {
      "epoch": 0.11151290892271466,
      "grad_norm": NaN,
      "learning_rate": 9.772819529298824e-05,
      "loss": 0.0,
      "step": 663
    },
    {
      "epoch": 0.11168110335547893,
      "grad_norm": NaN,
      "learning_rate": 9.772017972201213e-05,
      "loss": 0.0,
      "step": 664
    },
    {
      "epoch": 0.11184929778824321,
      "grad_norm": NaN,
      "learning_rate": 9.771215036521057e-05,
      "loss": 0.0,
      "step": 665
    },
    {
      "epoch": 0.11201749222100749,
      "grad_norm": NaN,
      "learning_rate": 9.770410722490319e-05,
      "loss": 0.0,
      "step": 666
    },
    {
      "epoch": 0.11218568665377177,
      "grad_norm": NaN,
      "learning_rate": 9.769605030341357e-05,
      "loss": 0.0,
      "step": 667
    },
    {
      "epoch": 0.11235388108653603,
      "grad_norm": NaN,
      "learning_rate": 9.768797960306921e-05,
      "loss": 0.0,
      "step": 668
    },
    {
      "epoch": 0.11252207551930031,
      "grad_norm": NaN,
      "learning_rate": 9.76798951262017e-05,
      "loss": 0.0,
      "step": 669
    },
    {
      "epoch": 0.11269026995206459,
      "grad_norm": NaN,
      "learning_rate": 9.767179687514654e-05,
      "loss": 0.0,
      "step": 670
    },
    {
      "epoch": 0.11285846438482887,
      "grad_norm": NaN,
      "learning_rate": 9.76636848522432e-05,
      "loss": 0.0,
      "step": 671
    },
    {
      "epoch": 0.11302665881759313,
      "grad_norm": NaN,
      "learning_rate": 9.765555905983517e-05,
      "loss": 0.0,
      "step": 672
    },
    {
      "epoch": 0.11319485325035741,
      "grad_norm": NaN,
      "learning_rate": 9.764741950026991e-05,
      "loss": 0.0,
      "step": 673
    },
    {
      "epoch": 0.11336304768312169,
      "grad_norm": NaN,
      "learning_rate": 9.763926617589883e-05,
      "loss": 0.0,
      "step": 674
    },
    {
      "epoch": 0.11353124211588596,
      "grad_norm": NaN,
      "learning_rate": 9.763109908907734e-05,
      "loss": 0.0,
      "step": 675
    },
    {
      "epoch": 0.11369943654865024,
      "grad_norm": NaN,
      "learning_rate": 9.762291824216484e-05,
      "loss": 0.0,
      "step": 676
    },
    {
      "epoch": 0.11386763098141452,
      "grad_norm": NaN,
      "learning_rate": 9.761472363752465e-05,
      "loss": 0.0,
      "step": 677
    },
    {
      "epoch": 0.1140358254141788,
      "grad_norm": NaN,
      "learning_rate": 9.760651527752413e-05,
      "loss": 0.0,
      "step": 678
    },
    {
      "epoch": 0.11420401984694306,
      "grad_norm": NaN,
      "learning_rate": 9.759829316453456e-05,
      "loss": 0.0,
      "step": 679
    },
    {
      "epoch": 0.11437221427970734,
      "grad_norm": NaN,
      "learning_rate": 9.759005730093123e-05,
      "loss": 0.0,
      "step": 680
    },
    {
      "epoch": 0.11454040871247162,
      "grad_norm": NaN,
      "learning_rate": 9.758180768909337e-05,
      "loss": 0.0,
      "step": 681
    },
    {
      "epoch": 0.1147086031452359,
      "grad_norm": NaN,
      "learning_rate": 9.757354433140425e-05,
      "loss": 0.0,
      "step": 682
    },
    {
      "epoch": 0.11487679757800016,
      "grad_norm": NaN,
      "learning_rate": 9.7565267230251e-05,
      "loss": 0.0,
      "step": 683
    },
    {
      "epoch": 0.11504499201076444,
      "grad_norm": NaN,
      "learning_rate": 9.75569763880248e-05,
      "loss": 0.0,
      "step": 684
    },
    {
      "epoch": 0.11521318644352872,
      "grad_norm": NaN,
      "learning_rate": 9.754867180712082e-05,
      "loss": 0.0,
      "step": 685
    },
    {
      "epoch": 0.11538138087629299,
      "grad_norm": NaN,
      "learning_rate": 9.75403534899381e-05,
      "loss": 0.0,
      "step": 686
    },
    {
      "epoch": 0.11554957530905727,
      "grad_norm": NaN,
      "learning_rate": 9.753202143887977e-05,
      "loss": 0.0,
      "step": 687
    },
    {
      "epoch": 0.11571776974182155,
      "grad_norm": NaN,
      "learning_rate": 9.752367565635281e-05,
      "loss": 0.0,
      "step": 688
    },
    {
      "epoch": 0.11588596417458583,
      "grad_norm": NaN,
      "learning_rate": 9.751531614476826e-05,
      "loss": 0.0,
      "step": 689
    },
    {
      "epoch": 0.11605415860735009,
      "grad_norm": NaN,
      "learning_rate": 9.750694290654108e-05,
      "loss": 0.0,
      "step": 690
    },
    {
      "epoch": 0.11622235304011437,
      "grad_norm": NaN,
      "learning_rate": 9.749855594409018e-05,
      "loss": 0.0,
      "step": 691
    },
    {
      "epoch": 0.11639054747287865,
      "grad_norm": NaN,
      "learning_rate": 9.749015525983852e-05,
      "loss": 0.0,
      "step": 692
    },
    {
      "epoch": 0.11655874190564293,
      "grad_norm": NaN,
      "learning_rate": 9.74817408562129e-05,
      "loss": 0.0,
      "step": 693
    },
    {
      "epoch": 0.1167269363384072,
      "grad_norm": NaN,
      "learning_rate": 9.747331273564418e-05,
      "loss": 0.0,
      "step": 694
    },
    {
      "epoch": 0.11689513077117147,
      "grad_norm": NaN,
      "learning_rate": 9.746487090056713e-05,
      "loss": 0.0,
      "step": 695
    },
    {
      "epoch": 0.11706332520393575,
      "grad_norm": NaN,
      "learning_rate": 9.745641535342053e-05,
      "loss": 0.0,
      "step": 696
    },
    {
      "epoch": 0.11723151963670002,
      "grad_norm": NaN,
      "learning_rate": 9.744794609664707e-05,
      "loss": 0.0,
      "step": 697
    },
    {
      "epoch": 0.1173997140694643,
      "grad_norm": NaN,
      "learning_rate": 9.743946313269344e-05,
      "loss": 0.0,
      "step": 698
    },
    {
      "epoch": 0.11756790850222858,
      "grad_norm": NaN,
      "learning_rate": 9.743096646401024e-05,
      "loss": 0.0,
      "step": 699
    },
    {
      "epoch": 0.11773610293499286,
      "grad_norm": NaN,
      "learning_rate": 9.742245609305212e-05,
      "loss": 0.0,
      "step": 700
    },
    {
      "epoch": 0.11790429736775712,
      "grad_norm": NaN,
      "learning_rate": 9.741393202227758e-05,
      "loss": 0.0,
      "step": 701
    },
    {
      "epoch": 0.1180724918005214,
      "grad_norm": NaN,
      "learning_rate": 9.740539425414912e-05,
      "loss": 0.0,
      "step": 702
    },
    {
      "epoch": 0.11824068623328568,
      "grad_norm": NaN,
      "learning_rate": 9.739684279113326e-05,
      "loss": 0.0,
      "step": 703
    },
    {
      "epoch": 0.11840888066604996,
      "grad_norm": NaN,
      "learning_rate": 9.73882776357004e-05,
      "loss": 0.0,
      "step": 704
    },
    {
      "epoch": 0.11857707509881422,
      "grad_norm": NaN,
      "learning_rate": 9.73796987903249e-05,
      "loss": 0.0,
      "step": 705
    },
    {
      "epoch": 0.1187452695315785,
      "grad_norm": NaN,
      "learning_rate": 9.737110625748508e-05,
      "loss": 0.0,
      "step": 706
    },
    {
      "epoch": 0.11891346396434278,
      "grad_norm": NaN,
      "learning_rate": 9.736250003966327e-05,
      "loss": 0.0,
      "step": 707
    },
    {
      "epoch": 0.11908165839710706,
      "grad_norm": NaN,
      "learning_rate": 9.73538801393457e-05,
      "loss": 0.0,
      "step": 708
    },
    {
      "epoch": 0.11924985282987133,
      "grad_norm": NaN,
      "learning_rate": 9.734524655902253e-05,
      "loss": 0.0,
      "step": 709
    },
    {
      "epoch": 0.11941804726263561,
      "grad_norm": NaN,
      "learning_rate": 9.733659930118793e-05,
      "loss": 0.0,
      "step": 710
    },
    {
      "epoch": 0.11958624169539989,
      "grad_norm": NaN,
      "learning_rate": 9.732793836833999e-05,
      "loss": 0.0,
      "step": 711
    },
    {
      "epoch": 0.11975443612816415,
      "grad_norm": NaN,
      "learning_rate": 9.731926376298077e-05,
      "loss": 0.0,
      "step": 712
    },
    {
      "epoch": 0.11992263056092843,
      "grad_norm": NaN,
      "learning_rate": 9.731057548761626e-05,
      "loss": 0.0,
      "step": 713
    },
    {
      "epoch": 0.12009082499369271,
      "grad_norm": NaN,
      "learning_rate": 9.73018735447564e-05,
      "loss": 0.0,
      "step": 714
    },
    {
      "epoch": 0.12025901942645699,
      "grad_norm": NaN,
      "learning_rate": 9.729315793691507e-05,
      "loss": 0.0,
      "step": 715
    },
    {
      "epoch": 0.12042721385922125,
      "grad_norm": NaN,
      "learning_rate": 9.728442866661013e-05,
      "loss": 0.0,
      "step": 716
    },
    {
      "epoch": 0.12059540829198553,
      "grad_norm": NaN,
      "learning_rate": 9.727568573636337e-05,
      "loss": 0.0,
      "step": 717
    },
    {
      "epoch": 0.12076360272474981,
      "grad_norm": NaN,
      "learning_rate": 9.72669291487005e-05,
      "loss": 0.0,
      "step": 718
    },
    {
      "epoch": 0.12093179715751409,
      "grad_norm": NaN,
      "learning_rate": 9.725815890615124e-05,
      "loss": 0.0,
      "step": 719
    },
    {
      "epoch": 0.12109999159027836,
      "grad_norm": NaN,
      "learning_rate": 9.724937501124918e-05,
      "loss": 0.0,
      "step": 720
    },
    {
      "epoch": 0.12126818602304264,
      "grad_norm": NaN,
      "learning_rate": 9.72405774665319e-05,
      "loss": 0.0,
      "step": 721
    },
    {
      "epoch": 0.12143638045580692,
      "grad_norm": NaN,
      "learning_rate": 9.723176627454094e-05,
      "loss": 0.0,
      "step": 722
    },
    {
      "epoch": 0.12160457488857118,
      "grad_norm": NaN,
      "learning_rate": 9.722294143782171e-05,
      "loss": 0.0,
      "step": 723
    },
    {
      "epoch": 0.12177276932133546,
      "grad_norm": NaN,
      "learning_rate": 9.721410295892363e-05,
      "loss": 0.0,
      "step": 724
    },
    {
      "epoch": 0.12194096375409974,
      "grad_norm": NaN,
      "learning_rate": 9.720525084040005e-05,
      "loss": 0.0,
      "step": 725
    },
    {
      "epoch": 0.12210915818686402,
      "grad_norm": NaN,
      "learning_rate": 9.71963850848082e-05,
      "loss": 0.0,
      "step": 726
    },
    {
      "epoch": 0.12227735261962829,
      "grad_norm": NaN,
      "learning_rate": 9.718750569470934e-05,
      "loss": 0.0,
      "step": 727
    },
    {
      "epoch": 0.12244554705239256,
      "grad_norm": NaN,
      "learning_rate": 9.717861267266862e-05,
      "loss": 0.0,
      "step": 728
    },
    {
      "epoch": 0.12261374148515684,
      "grad_norm": NaN,
      "learning_rate": 9.716970602125512e-05,
      "loss": 0.0,
      "step": 729
    },
    {
      "epoch": 0.12278193591792112,
      "grad_norm": NaN,
      "learning_rate": 9.716078574304189e-05,
      "loss": 0.0,
      "step": 730
    },
    {
      "epoch": 0.12295013035068539,
      "grad_norm": NaN,
      "learning_rate": 9.715185184060588e-05,
      "loss": 0.0,
      "step": 731
    },
    {
      "epoch": 0.12311832478344967,
      "grad_norm": NaN,
      "learning_rate": 9.714290431652803e-05,
      "loss": 0.0,
      "step": 732
    },
    {
      "epoch": 0.12328651921621395,
      "grad_norm": NaN,
      "learning_rate": 9.713394317339313e-05,
      "loss": 0.0,
      "step": 733
    },
    {
      "epoch": 0.12345471364897821,
      "grad_norm": NaN,
      "learning_rate": 9.712496841379e-05,
      "loss": 0.0,
      "step": 734
    },
    {
      "epoch": 0.12362290808174249,
      "grad_norm": NaN,
      "learning_rate": 9.711598004031131e-05,
      "loss": 0.0,
      "step": 735
    },
    {
      "epoch": 0.12379110251450677,
      "grad_norm": NaN,
      "learning_rate": 9.710697805555372e-05,
      "loss": 0.0,
      "step": 736
    },
    {
      "epoch": 0.12395929694727105,
      "grad_norm": NaN,
      "learning_rate": 9.709796246211777e-05,
      "loss": 0.0,
      "step": 737
    },
    {
      "epoch": 0.12412749138003532,
      "grad_norm": NaN,
      "learning_rate": 9.708893326260801e-05,
      "loss": 0.0,
      "step": 738
    },
    {
      "epoch": 0.1242956858127996,
      "grad_norm": NaN,
      "learning_rate": 9.707989045963288e-05,
      "loss": 0.0,
      "step": 739
    },
    {
      "epoch": 0.12446388024556387,
      "grad_norm": NaN,
      "learning_rate": 9.70708340558047e-05,
      "loss": 0.0,
      "step": 740
    },
    {
      "epoch": 0.12463207467832815,
      "grad_norm": NaN,
      "learning_rate": 9.706176405373979e-05,
      "loss": 0.0,
      "step": 741
    },
    {
      "epoch": 0.12480026911109242,
      "grad_norm": NaN,
      "learning_rate": 9.705268045605837e-05,
      "loss": 0.0,
      "step": 742
    },
    {
      "epoch": 0.1249684635438567,
      "grad_norm": NaN,
      "learning_rate": 9.704358326538457e-05,
      "loss": 0.0,
      "step": 743
    },
    {
      "epoch": 0.12513665797662096,
      "grad_norm": NaN,
      "learning_rate": 9.70344724843465e-05,
      "loss": 0.0,
      "step": 744
    },
    {
      "epoch": 0.12530485240938524,
      "grad_norm": NaN,
      "learning_rate": 9.702534811557615e-05,
      "loss": 0.0,
      "step": 745
    },
    {
      "epoch": 0.12547304684214952,
      "grad_norm": NaN,
      "learning_rate": 9.701621016170943e-05,
      "loss": 0.0,
      "step": 746
    },
    {
      "epoch": 0.1256412412749138,
      "grad_norm": NaN,
      "learning_rate": 9.70070586253862e-05,
      "loss": 0.0,
      "step": 747
    },
    {
      "epoch": 0.12580943570767808,
      "grad_norm": NaN,
      "learning_rate": 9.699789350925026e-05,
      "loss": 0.0,
      "step": 748
    },
    {
      "epoch": 0.12597763014044236,
      "grad_norm": NaN,
      "learning_rate": 9.698871481594927e-05,
      "loss": 0.0,
      "step": 749
    },
    {
      "epoch": 0.12614582457320664,
      "grad_norm": NaN,
      "learning_rate": 9.69795225481349e-05,
      "loss": 0.0,
      "step": 750
    },
    {
      "epoch": 0.1263140190059709,
      "grad_norm": NaN,
      "learning_rate": 9.697031670846265e-05,
      "loss": 0.0,
      "step": 751
    },
    {
      "epoch": 0.12648221343873517,
      "grad_norm": NaN,
      "learning_rate": 9.6961097299592e-05,
      "loss": 0.0,
      "step": 752
    },
    {
      "epoch": 0.12665040787149945,
      "grad_norm": NaN,
      "learning_rate": 9.695186432418635e-05,
      "loss": 0.0,
      "step": 753
    },
    {
      "epoch": 0.12681860230426373,
      "grad_norm": NaN,
      "learning_rate": 9.694261778491297e-05,
      "loss": 0.0,
      "step": 754
    },
    {
      "epoch": 0.126986796737028,
      "grad_norm": NaN,
      "learning_rate": 9.693335768444311e-05,
      "loss": 0.0,
      "step": 755
    },
    {
      "epoch": 0.1271549911697923,
      "grad_norm": NaN,
      "learning_rate": 9.69240840254519e-05,
      "loss": 0.0,
      "step": 756
    },
    {
      "epoch": 0.12732318560255657,
      "grad_norm": NaN,
      "learning_rate": 9.69147968106184e-05,
      "loss": 0.0,
      "step": 757
    },
    {
      "epoch": 0.12749138003532082,
      "grad_norm": NaN,
      "learning_rate": 9.690549604262555e-05,
      "loss": 0.0,
      "step": 758
    },
    {
      "epoch": 0.1276595744680851,
      "grad_norm": NaN,
      "learning_rate": 9.689618172416028e-05,
      "loss": 0.0,
      "step": 759
    },
    {
      "epoch": 0.12782776890084938,
      "grad_norm": NaN,
      "learning_rate": 9.688685385791339e-05,
      "loss": 0.0,
      "step": 760
    },
    {
      "epoch": 0.12799596333361365,
      "grad_norm": NaN,
      "learning_rate": 9.687751244657957e-05,
      "loss": 0.0,
      "step": 761
    },
    {
      "epoch": 0.12816415776637793,
      "grad_norm": NaN,
      "learning_rate": 9.686815749285745e-05,
      "loss": 0.0,
      "step": 762
    },
    {
      "epoch": 0.1283323521991422,
      "grad_norm": NaN,
      "learning_rate": 9.68587889994496e-05,
      "loss": 0.0,
      "step": 763
    },
    {
      "epoch": 0.1285005466319065,
      "grad_norm": NaN,
      "learning_rate": 9.684940696906246e-05,
      "loss": 0.0,
      "step": 764
    },
    {
      "epoch": 0.12866874106467077,
      "grad_norm": NaN,
      "learning_rate": 9.684001140440639e-05,
      "loss": 0.0,
      "step": 765
    },
    {
      "epoch": 0.12883693549743502,
      "grad_norm": NaN,
      "learning_rate": 9.683060230819565e-05,
      "loss": 0.0,
      "step": 766
    },
    {
      "epoch": 0.1290051299301993,
      "grad_norm": NaN,
      "learning_rate": 9.682117968314846e-05,
      "loss": 0.0,
      "step": 767
    },
    {
      "epoch": 0.12917332436296358,
      "grad_norm": NaN,
      "learning_rate": 9.681174353198687e-05,
      "loss": 0.0,
      "step": 768
    },
    {
      "epoch": 0.12934151879572786,
      "grad_norm": NaN,
      "learning_rate": 9.680229385743689e-05,
      "loss": 0.0,
      "step": 769
    },
    {
      "epoch": 0.12950971322849214,
      "grad_norm": NaN,
      "learning_rate": 9.679283066222845e-05,
      "loss": 0.0,
      "step": 770
    },
    {
      "epoch": 0.12967790766125642,
      "grad_norm": NaN,
      "learning_rate": 9.678335394909532e-05,
      "loss": 0.0,
      "step": 771
    },
    {
      "epoch": 0.1298461020940207,
      "grad_norm": NaN,
      "learning_rate": 9.677386372077524e-05,
      "loss": 0.0,
      "step": 772
    },
    {
      "epoch": 0.13001429652678495,
      "grad_norm": NaN,
      "learning_rate": 9.676435998000983e-05,
      "loss": 0.0,
      "step": 773
    },
    {
      "epoch": 0.13018249095954923,
      "grad_norm": NaN,
      "learning_rate": 9.675484272954462e-05,
      "loss": 0.0,
      "step": 774
    },
    {
      "epoch": 0.1303506853923135,
      "grad_norm": NaN,
      "learning_rate": 9.674531197212903e-05,
      "loss": 0.0,
      "step": 775
    },
    {
      "epoch": 0.1305188798250778,
      "grad_norm": NaN,
      "learning_rate": 9.673576771051639e-05,
      "loss": 0.0,
      "step": 776
    },
    {
      "epoch": 0.13068707425784207,
      "grad_norm": NaN,
      "learning_rate": 9.672620994746394e-05,
      "loss": 0.0,
      "step": 777
    },
    {
      "epoch": 0.13085526869060635,
      "grad_norm": NaN,
      "learning_rate": 9.671663868573277e-05,
      "loss": 0.0,
      "step": 778
    },
    {
      "epoch": 0.13102346312337063,
      "grad_norm": NaN,
      "learning_rate": 9.670705392808796e-05,
      "loss": 0.0,
      "step": 779
    },
    {
      "epoch": 0.1311916575561349,
      "grad_norm": NaN,
      "learning_rate": 9.669745567729841e-05,
      "loss": 0.0,
      "step": 780
    },
    {
      "epoch": 0.13135985198889916,
      "grad_norm": NaN,
      "learning_rate": 9.668784393613696e-05,
      "loss": 0.0,
      "step": 781
    },
    {
      "epoch": 0.13152804642166344,
      "grad_norm": NaN,
      "learning_rate": 9.667821870738033e-05,
      "loss": 0.0,
      "step": 782
    },
    {
      "epoch": 0.13169624085442772,
      "grad_norm": NaN,
      "learning_rate": 9.666857999380914e-05,
      "loss": 0.0,
      "step": 783
    },
    {
      "epoch": 0.131864435287192,
      "grad_norm": NaN,
      "learning_rate": 9.665892779820791e-05,
      "loss": 0.0,
      "step": 784
    },
    {
      "epoch": 0.13203262971995627,
      "grad_norm": NaN,
      "learning_rate": 9.664926212336505e-05,
      "loss": 0.0,
      "step": 785
    },
    {
      "epoch": 0.13220082415272055,
      "grad_norm": NaN,
      "learning_rate": 9.663958297207286e-05,
      "loss": 0.0,
      "step": 786
    },
    {
      "epoch": 0.13236901858548483,
      "grad_norm": NaN,
      "learning_rate": 9.662989034712755e-05,
      "loss": 0.0,
      "step": 787
    },
    {
      "epoch": 0.13253721301824908,
      "grad_norm": NaN,
      "learning_rate": 9.66201842513292e-05,
      "loss": 0.0,
      "step": 788
    },
    {
      "epoch": 0.13270540745101336,
      "grad_norm": NaN,
      "learning_rate": 9.66104646874818e-05,
      "loss": 0.0,
      "step": 789
    },
    {
      "epoch": 0.13287360188377764,
      "grad_norm": NaN,
      "learning_rate": 9.66007316583932e-05,
      "loss": 0.0,
      "step": 790
    },
    {
      "epoch": 0.13304179631654192,
      "grad_norm": NaN,
      "learning_rate": 9.65909851668752e-05,
      "loss": 0.0,
      "step": 791
    },
    {
      "epoch": 0.1332099907493062,
      "grad_norm": NaN,
      "learning_rate": 9.658122521574344e-05,
      "loss": 0.0,
      "step": 792
    },
    {
      "epoch": 0.13337818518207048,
      "grad_norm": NaN,
      "learning_rate": 9.657145180781743e-05,
      "loss": 0.0,
      "step": 793
    },
    {
      "epoch": 0.13354637961483476,
      "grad_norm": NaN,
      "learning_rate": 9.656166494592064e-05,
      "loss": 0.0,
      "step": 794
    },
    {
      "epoch": 0.133714574047599,
      "grad_norm": NaN,
      "learning_rate": 9.655186463288038e-05,
      "loss": 0.0,
      "step": 795
    },
    {
      "epoch": 0.1338827684803633,
      "grad_norm": NaN,
      "learning_rate": 9.654205087152781e-05,
      "loss": 0.0,
      "step": 796
    },
    {
      "epoch": 0.13405096291312757,
      "grad_norm": NaN,
      "learning_rate": 9.653222366469808e-05,
      "loss": 0.0,
      "step": 797
    },
    {
      "epoch": 0.13421915734589185,
      "grad_norm": NaN,
      "learning_rate": 9.65223830152301e-05,
      "loss": 0.0,
      "step": 798
    },
    {
      "epoch": 0.13438735177865613,
      "grad_norm": NaN,
      "learning_rate": 9.651252892596673e-05,
      "loss": 0.0,
      "step": 799
    },
    {
      "epoch": 0.1345555462114204,
      "grad_norm": NaN,
      "learning_rate": 9.650266139975474e-05,
      "loss": 0.0,
      "step": 800
    },
    {
      "epoch": 0.1347237406441847,
      "grad_norm": NaN,
      "learning_rate": 9.649278043944474e-05,
      "loss": 0.0,
      "step": 801
    },
    {
      "epoch": 0.13489193507694897,
      "grad_norm": NaN,
      "learning_rate": 9.648288604789121e-05,
      "loss": 0.0,
      "step": 802
    },
    {
      "epoch": 0.13506012950971322,
      "grad_norm": NaN,
      "learning_rate": 9.647297822795256e-05,
      "loss": 0.0,
      "step": 803
    },
    {
      "epoch": 0.1352283239424775,
      "grad_norm": NaN,
      "learning_rate": 9.6463056982491e-05,
      "loss": 0.0,
      "step": 804
    },
    {
      "epoch": 0.13539651837524178,
      "grad_norm": NaN,
      "learning_rate": 9.645312231437271e-05,
      "loss": 0.0,
      "step": 805
    },
    {
      "epoch": 0.13556471280800605,
      "grad_norm": NaN,
      "learning_rate": 9.644317422646768e-05,
      "loss": 0.0,
      "step": 806
    },
    {
      "epoch": 0.13573290724077033,
      "grad_norm": NaN,
      "learning_rate": 9.643321272164981e-05,
      "loss": 0.0,
      "step": 807
    },
    {
      "epoch": 0.1359011016735346,
      "grad_norm": NaN,
      "learning_rate": 9.642323780279687e-05,
      "loss": 0.0,
      "step": 808
    },
    {
      "epoch": 0.1360692961062989,
      "grad_norm": NaN,
      "learning_rate": 9.641324947279049e-05,
      "loss": 0.0,
      "step": 809
    },
    {
      "epoch": 0.13623749053906314,
      "grad_norm": NaN,
      "learning_rate": 9.640324773451622e-05,
      "loss": 0.0,
      "step": 810
    },
    {
      "epoch": 0.13640568497182742,
      "grad_norm": NaN,
      "learning_rate": 9.63932325908634e-05,
      "loss": 0.0,
      "step": 811
    },
    {
      "epoch": 0.1365738794045917,
      "grad_norm": NaN,
      "learning_rate": 9.638320404472533e-05,
      "loss": 0.0,
      "step": 812
    },
    {
      "epoch": 0.13674207383735598,
      "grad_norm": NaN,
      "learning_rate": 9.637316209899912e-05,
      "loss": 0.0,
      "step": 813
    },
    {
      "epoch": 0.13691026827012026,
      "grad_norm": NaN,
      "learning_rate": 9.63631067565858e-05,
      "loss": 0.0,
      "step": 814
    },
    {
      "epoch": 0.13707846270288454,
      "grad_norm": NaN,
      "learning_rate": 9.635303802039024e-05,
      "loss": 0.0,
      "step": 815
    },
    {
      "epoch": 0.13724665713564882,
      "grad_norm": NaN,
      "learning_rate": 9.634295589332117e-05,
      "loss": 0.0,
      "step": 816
    },
    {
      "epoch": 0.1374148515684131,
      "grad_norm": NaN,
      "learning_rate": 9.633286037829121e-05,
      "loss": 0.0,
      "step": 817
    },
    {
      "epoch": 0.13758304600117735,
      "grad_norm": NaN,
      "learning_rate": 9.632275147821686e-05,
      "loss": 0.0,
      "step": 818
    },
    {
      "epoch": 0.13775124043394163,
      "grad_norm": NaN,
      "learning_rate": 9.631262919601844e-05,
      "loss": 0.0,
      "step": 819
    },
    {
      "epoch": 0.1379194348667059,
      "grad_norm": NaN,
      "learning_rate": 9.630249353462018e-05,
      "loss": 0.0,
      "step": 820
    },
    {
      "epoch": 0.1380876292994702,
      "grad_norm": NaN,
      "learning_rate": 9.629234449695015e-05,
      "loss": 0.0,
      "step": 821
    },
    {
      "epoch": 0.13825582373223447,
      "grad_norm": NaN,
      "learning_rate": 9.62821820859403e-05,
      "loss": 0.0,
      "step": 822
    },
    {
      "epoch": 0.13842401816499875,
      "grad_norm": NaN,
      "learning_rate": 9.627200630452644e-05,
      "loss": 0.0,
      "step": 823
    },
    {
      "epoch": 0.13859221259776303,
      "grad_norm": NaN,
      "learning_rate": 9.626181715564823e-05,
      "loss": 0.0,
      "step": 824
    },
    {
      "epoch": 0.13876040703052728,
      "grad_norm": NaN,
      "learning_rate": 9.625161464224922e-05,
      "loss": 0.0,
      "step": 825
    },
    {
      "epoch": 0.13892860146329156,
      "grad_norm": NaN,
      "learning_rate": 9.624139876727678e-05,
      "loss": 0.0,
      "step": 826
    },
    {
      "epoch": 0.13909679589605584,
      "grad_norm": NaN,
      "learning_rate": 9.623116953368216e-05,
      "loss": 0.0,
      "step": 827
    },
    {
      "epoch": 0.13926499032882012,
      "grad_norm": NaN,
      "learning_rate": 9.62209269444205e-05,
      "loss": 0.0,
      "step": 828
    },
    {
      "epoch": 0.1394331847615844,
      "grad_norm": NaN,
      "learning_rate": 9.621067100245073e-05,
      "loss": 0.0,
      "step": 829
    },
    {
      "epoch": 0.13960137919434867,
      "grad_norm": NaN,
      "learning_rate": 9.62004017107357e-05,
      "loss": 0.0,
      "step": 830
    },
    {
      "epoch": 0.13976957362711295,
      "grad_norm": NaN,
      "learning_rate": 9.61901190722421e-05,
      "loss": 0.0,
      "step": 831
    },
    {
      "epoch": 0.1399377680598772,
      "grad_norm": NaN,
      "learning_rate": 9.617982308994045e-05,
      "loss": 0.0,
      "step": 832
    },
    {
      "epoch": 0.14010596249264148,
      "grad_norm": NaN,
      "learning_rate": 9.616951376680516e-05,
      "loss": 0.0,
      "step": 833
    },
    {
      "epoch": 0.14027415692540576,
      "grad_norm": NaN,
      "learning_rate": 9.615919110581447e-05,
      "loss": 0.0,
      "step": 834
    },
    {
      "epoch": 0.14044235135817004,
      "grad_norm": NaN,
      "learning_rate": 9.614885510995047e-05,
      "loss": 0.0,
      "step": 835
    },
    {
      "epoch": 0.14061054579093432,
      "grad_norm": NaN,
      "learning_rate": 9.613850578219914e-05,
      "loss": 0.0,
      "step": 836
    },
    {
      "epoch": 0.1407787402236986,
      "grad_norm": NaN,
      "learning_rate": 9.612814312555027e-05,
      "loss": 0.0,
      "step": 837
    },
    {
      "epoch": 0.14094693465646288,
      "grad_norm": NaN,
      "learning_rate": 9.611776714299751e-05,
      "loss": 0.0,
      "step": 838
    },
    {
      "epoch": 0.14111512908922716,
      "grad_norm": NaN,
      "learning_rate": 9.610737783753838e-05,
      "loss": 0.0,
      "step": 839
    },
    {
      "epoch": 0.1412833235219914,
      "grad_norm": NaN,
      "learning_rate": 9.60969752121742e-05,
      "loss": 0.0,
      "step": 840
    },
    {
      "epoch": 0.1414515179547557,
      "grad_norm": NaN,
      "learning_rate": 9.608655926991021e-05,
      "loss": 0.0,
      "step": 841
    },
    {
      "epoch": 0.14161971238751997,
      "grad_norm": NaN,
      "learning_rate": 9.607613001375546e-05,
      "loss": 0.0,
      "step": 842
    },
    {
      "epoch": 0.14178790682028425,
      "grad_norm": NaN,
      "learning_rate": 9.606568744672282e-05,
      "loss": 0.0,
      "step": 843
    },
    {
      "epoch": 0.14195610125304853,
      "grad_norm": NaN,
      "learning_rate": 9.605523157182904e-05,
      "loss": 0.0,
      "step": 844
    },
    {
      "epoch": 0.1421242956858128,
      "grad_norm": NaN,
      "learning_rate": 9.604476239209472e-05,
      "loss": 0.0,
      "step": 845
    },
    {
      "epoch": 0.1422924901185771,
      "grad_norm": NaN,
      "learning_rate": 9.603427991054426e-05,
      "loss": 0.0,
      "step": 846
    },
    {
      "epoch": 0.14246068455134134,
      "grad_norm": NaN,
      "learning_rate": 9.602378413020596e-05,
      "loss": 0.0,
      "step": 847
    },
    {
      "epoch": 0.14262887898410562,
      "grad_norm": NaN,
      "learning_rate": 9.601327505411191e-05,
      "loss": 0.0,
      "step": 848
    },
    {
      "epoch": 0.1427970734168699,
      "grad_norm": NaN,
      "learning_rate": 9.600275268529807e-05,
      "loss": 0.0,
      "step": 849
    },
    {
      "epoch": 0.14296526784963418,
      "grad_norm": NaN,
      "learning_rate": 9.599221702680425e-05,
      "loss": 0.0,
      "step": 850
    },
    {
      "epoch": 0.14313346228239845,
      "grad_norm": NaN,
      "learning_rate": 9.598166808167408e-05,
      "loss": 0.0,
      "step": 851
    },
    {
      "epoch": 0.14330165671516273,
      "grad_norm": NaN,
      "learning_rate": 9.597110585295502e-05,
      "loss": 0.0,
      "step": 852
    },
    {
      "epoch": 0.143469851147927,
      "grad_norm": NaN,
      "learning_rate": 9.596053034369837e-05,
      "loss": 0.0,
      "step": 853
    },
    {
      "epoch": 0.1436380455806913,
      "grad_norm": NaN,
      "learning_rate": 9.594994155695931e-05,
      "loss": 0.0,
      "step": 854
    },
    {
      "epoch": 0.14380624001345554,
      "grad_norm": NaN,
      "learning_rate": 9.593933949579678e-05,
      "loss": 0.0,
      "step": 855
    },
    {
      "epoch": 0.14397443444621982,
      "grad_norm": NaN,
      "learning_rate": 9.592872416327365e-05,
      "loss": 0.0,
      "step": 856
    },
    {
      "epoch": 0.1441426288789841,
      "grad_norm": NaN,
      "learning_rate": 9.591809556245652e-05,
      "loss": 0.0,
      "step": 857
    },
    {
      "epoch": 0.14431082331174838,
      "grad_norm": NaN,
      "learning_rate": 9.59074536964159e-05,
      "loss": 0.0,
      "step": 858
    },
    {
      "epoch": 0.14447901774451266,
      "grad_norm": NaN,
      "learning_rate": 9.58967985682261e-05,
      "loss": 0.0,
      "step": 859
    },
    {
      "epoch": 0.14464721217727694,
      "grad_norm": NaN,
      "learning_rate": 9.588613018096529e-05,
      "loss": 0.0,
      "step": 860
    },
    {
      "epoch": 0.14481540661004122,
      "grad_norm": NaN,
      "learning_rate": 9.587544853771539e-05,
      "loss": 0.0,
      "step": 861
    },
    {
      "epoch": 0.14498360104280547,
      "grad_norm": NaN,
      "learning_rate": 9.586475364156226e-05,
      "loss": 0.0,
      "step": 862
    },
    {
      "epoch": 0.14515179547556975,
      "grad_norm": NaN,
      "learning_rate": 9.585404549559551e-05,
      "loss": 0.0,
      "step": 863
    },
    {
      "epoch": 0.14531998990833403,
      "grad_norm": NaN,
      "learning_rate": 9.584332410290861e-05,
      "loss": 0.0,
      "step": 864
    },
    {
      "epoch": 0.1454881843410983,
      "grad_norm": NaN,
      "learning_rate": 9.583258946659885e-05,
      "loss": 0.0,
      "step": 865
    },
    {
      "epoch": 0.1456563787738626,
      "grad_norm": NaN,
      "learning_rate": 9.582184158976736e-05,
      "loss": 0.0,
      "step": 866
    },
    {
      "epoch": 0.14582457320662687,
      "grad_norm": NaN,
      "learning_rate": 9.581108047551905e-05,
      "loss": 0.0,
      "step": 867
    },
    {
      "epoch": 0.14599276763939115,
      "grad_norm": NaN,
      "learning_rate": 9.580030612696273e-05,
      "loss": 0.0,
      "step": 868
    },
    {
      "epoch": 0.1461609620721554,
      "grad_norm": NaN,
      "learning_rate": 9.578951854721095e-05,
      "loss": 0.0,
      "step": 869
    },
    {
      "epoch": 0.14632915650491968,
      "grad_norm": NaN,
      "learning_rate": 9.577871773938011e-05,
      "loss": 0.0,
      "step": 870
    },
    {
      "epoch": 0.14649735093768396,
      "grad_norm": NaN,
      "learning_rate": 9.57679037065905e-05,
      "loss": 0.0,
      "step": 871
    },
    {
      "epoch": 0.14666554537044824,
      "grad_norm": NaN,
      "learning_rate": 9.575707645196612e-05,
      "loss": 0.0,
      "step": 872
    },
    {
      "epoch": 0.14683373980321252,
      "grad_norm": NaN,
      "learning_rate": 9.574623597863488e-05,
      "loss": 0.0,
      "step": 873
    },
    {
      "epoch": 0.1470019342359768,
      "grad_norm": NaN,
      "learning_rate": 9.573538228972844e-05,
      "loss": 0.0,
      "step": 874
    },
    {
      "epoch": 0.14717012866874107,
      "grad_norm": NaN,
      "learning_rate": 9.572451538838232e-05,
      "loss": 0.0,
      "step": 875
    },
    {
      "epoch": 0.14733832310150535,
      "grad_norm": NaN,
      "learning_rate": 9.571363527773585e-05,
      "loss": 0.0,
      "step": 876
    },
    {
      "epoch": 0.1475065175342696,
      "grad_norm": NaN,
      "learning_rate": 9.570274196093217e-05,
      "loss": 0.0,
      "step": 877
    },
    {
      "epoch": 0.14767471196703388,
      "grad_norm": NaN,
      "learning_rate": 9.569183544111824e-05,
      "loss": 0.0,
      "step": 878
    },
    {
      "epoch": 0.14784290639979816,
      "grad_norm": NaN,
      "learning_rate": 9.568091572144484e-05,
      "loss": 0.0,
      "step": 879
    },
    {
      "epoch": 0.14801110083256244,
      "grad_norm": NaN,
      "learning_rate": 9.566998280506653e-05,
      "loss": 0.0,
      "step": 880
    },
    {
      "epoch": 0.14817929526532672,
      "grad_norm": NaN,
      "learning_rate": 9.565903669514172e-05,
      "loss": 0.0,
      "step": 881
    },
    {
      "epoch": 0.148347489698091,
      "grad_norm": NaN,
      "learning_rate": 9.564807739483262e-05,
      "loss": 0.0,
      "step": 882
    },
    {
      "epoch": 0.14851568413085528,
      "grad_norm": NaN,
      "learning_rate": 9.563710490730525e-05,
      "loss": 0.0,
      "step": 883
    },
    {
      "epoch": 0.14868387856361953,
      "grad_norm": NaN,
      "learning_rate": 9.562611923572944e-05,
      "loss": 0.0,
      "step": 884
    },
    {
      "epoch": 0.1488520729963838,
      "grad_norm": NaN,
      "learning_rate": 9.561512038327882e-05,
      "loss": 0.0,
      "step": 885
    },
    {
      "epoch": 0.1490202674291481,
      "grad_norm": NaN,
      "learning_rate": 9.560410835313084e-05,
      "loss": 0.0,
      "step": 886
    },
    {
      "epoch": 0.14918846186191237,
      "grad_norm": NaN,
      "learning_rate": 9.559308314846675e-05,
      "loss": 0.0,
      "step": 887
    },
    {
      "epoch": 0.14935665629467665,
      "grad_norm": NaN,
      "learning_rate": 9.558204477247163e-05,
      "loss": 0.0,
      "step": 888
    },
    {
      "epoch": 0.14952485072744093,
      "grad_norm": NaN,
      "learning_rate": 9.55709932283343e-05,
      "loss": 0.0,
      "step": 889
    },
    {
      "epoch": 0.1496930451602052,
      "grad_norm": NaN,
      "learning_rate": 9.555992851924746e-05,
      "loss": 0.0,
      "step": 890
    },
    {
      "epoch": 0.1498612395929695,
      "grad_norm": NaN,
      "learning_rate": 9.554885064840758e-05,
      "loss": 0.0,
      "step": 891
    },
    {
      "epoch": 0.15002943402573374,
      "grad_norm": NaN,
      "learning_rate": 9.55377596190149e-05,
      "loss": 0.0,
      "step": 892
    },
    {
      "epoch": 0.15019762845849802,
      "grad_norm": NaN,
      "learning_rate": 9.552665543427354e-05,
      "loss": 0.0,
      "step": 893
    },
    {
      "epoch": 0.1503658228912623,
      "grad_norm": NaN,
      "learning_rate": 9.551553809739135e-05,
      "loss": 0.0,
      "step": 894
    },
    {
      "epoch": 0.15053401732402658,
      "grad_norm": NaN,
      "learning_rate": 9.550440761158e-05,
      "loss": 0.0,
      "step": 895
    },
    {
      "epoch": 0.15070221175679085,
      "grad_norm": NaN,
      "learning_rate": 9.549326398005495e-05,
      "loss": 0.0,
      "step": 896
    },
    {
      "epoch": 0.15087040618955513,
      "grad_norm": NaN,
      "learning_rate": 9.548210720603552e-05,
      "loss": 0.0,
      "step": 897
    },
    {
      "epoch": 0.1510386006223194,
      "grad_norm": NaN,
      "learning_rate": 9.547093729274474e-05,
      "loss": 0.0,
      "step": 898
    },
    {
      "epoch": 0.15120679505508366,
      "grad_norm": NaN,
      "learning_rate": 9.545975424340948e-05,
      "loss": 0.0,
      "step": 899
    },
    {
      "epoch": 0.15137498948784794,
      "grad_norm": NaN,
      "learning_rate": 9.544855806126041e-05,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 0.15154318392061222,
      "grad_norm": NaN,
      "learning_rate": 9.543734874953194e-05,
      "loss": 0.0,
      "step": 901
    },
    {
      "epoch": 0.1517113783533765,
      "grad_norm": NaN,
      "learning_rate": 9.542612631146237e-05,
      "loss": 0.0,
      "step": 902
    },
    {
      "epoch": 0.15187957278614078,
      "grad_norm": NaN,
      "learning_rate": 9.54148907502937e-05,
      "loss": 0.0,
      "step": 903
    },
    {
      "epoch": 0.15204776721890506,
      "grad_norm": NaN,
      "learning_rate": 9.540364206927175e-05,
      "loss": 0.0,
      "step": 904
    },
    {
      "epoch": 0.15221596165166934,
      "grad_norm": NaN,
      "learning_rate": 9.539238027164619e-05,
      "loss": 0.0,
      "step": 905
    },
    {
      "epoch": 0.1523841560844336,
      "grad_norm": NaN,
      "learning_rate": 9.538110536067037e-05,
      "loss": 0.0,
      "step": 906
    },
    {
      "epoch": 0.15255235051719787,
      "grad_norm": NaN,
      "learning_rate": 9.536981733960152e-05,
      "loss": 0.0,
      "step": 907
    },
    {
      "epoch": 0.15272054494996215,
      "grad_norm": NaN,
      "learning_rate": 9.53585162117006e-05,
      "loss": 0.0,
      "step": 908
    },
    {
      "epoch": 0.15288873938272643,
      "grad_norm": NaN,
      "learning_rate": 9.53472019802324e-05,
      "loss": 0.0,
      "step": 909
    },
    {
      "epoch": 0.1530569338154907,
      "grad_norm": NaN,
      "learning_rate": 9.533587464846548e-05,
      "loss": 0.0,
      "step": 910
    },
    {
      "epoch": 0.153225128248255,
      "grad_norm": NaN,
      "learning_rate": 9.532453421967213e-05,
      "loss": 0.0,
      "step": 911
    },
    {
      "epoch": 0.15339332268101927,
      "grad_norm": NaN,
      "learning_rate": 9.531318069712854e-05,
      "loss": 0.0,
      "step": 912
    },
    {
      "epoch": 0.15356151711378355,
      "grad_norm": NaN,
      "learning_rate": 9.530181408411456e-05,
      "loss": 0.0,
      "step": 913
    },
    {
      "epoch": 0.1537297115465478,
      "grad_norm": NaN,
      "learning_rate": 9.529043438391392e-05,
      "loss": 0.0,
      "step": 914
    },
    {
      "epoch": 0.15389790597931208,
      "grad_norm": NaN,
      "learning_rate": 9.527904159981404e-05,
      "loss": 0.0,
      "step": 915
    },
    {
      "epoch": 0.15406610041207636,
      "grad_norm": NaN,
      "learning_rate": 9.526763573510622e-05,
      "loss": 0.0,
      "step": 916
    },
    {
      "epoch": 0.15423429484484064,
      "grad_norm": NaN,
      "learning_rate": 9.525621679308546e-05,
      "loss": 0.0,
      "step": 917
    },
    {
      "epoch": 0.15440248927760492,
      "grad_norm": NaN,
      "learning_rate": 9.524478477705057e-05,
      "loss": 0.0,
      "step": 918
    },
    {
      "epoch": 0.1545706837103692,
      "grad_norm": NaN,
      "learning_rate": 9.523333969030413e-05,
      "loss": 0.0,
      "step": 919
    },
    {
      "epoch": 0.15473887814313347,
      "grad_norm": NaN,
      "learning_rate": 9.522188153615248e-05,
      "loss": 0.0,
      "step": 920
    },
    {
      "epoch": 0.15490707257589773,
      "grad_norm": NaN,
      "learning_rate": 9.521041031790578e-05,
      "loss": 0.0,
      "step": 921
    },
    {
      "epoch": 0.155075267008662,
      "grad_norm": NaN,
      "learning_rate": 9.519892603887792e-05,
      "loss": 0.0,
      "step": 922
    },
    {
      "epoch": 0.15524346144142628,
      "grad_norm": NaN,
      "learning_rate": 9.51874287023866e-05,
      "loss": 0.0,
      "step": 923
    },
    {
      "epoch": 0.15541165587419056,
      "grad_norm": NaN,
      "learning_rate": 9.517591831175323e-05,
      "loss": 0.0,
      "step": 924
    },
    {
      "epoch": 0.15557985030695484,
      "grad_norm": NaN,
      "learning_rate": 9.516439487030308e-05,
      "loss": 0.0,
      "step": 925
    },
    {
      "epoch": 0.15574804473971912,
      "grad_norm": NaN,
      "learning_rate": 9.51528583813651e-05,
      "loss": 0.0,
      "step": 926
    },
    {
      "epoch": 0.1559162391724834,
      "grad_norm": NaN,
      "learning_rate": 9.51413088482721e-05,
      "loss": 0.0,
      "step": 927
    },
    {
      "epoch": 0.15608443360524768,
      "grad_norm": NaN,
      "learning_rate": 9.512974627436058e-05,
      "loss": 0.0,
      "step": 928
    },
    {
      "epoch": 0.15625262803801193,
      "grad_norm": NaN,
      "learning_rate": 9.511817066297082e-05,
      "loss": 0.0,
      "step": 929
    },
    {
      "epoch": 0.1564208224707762,
      "grad_norm": NaN,
      "learning_rate": 9.510658201744693e-05,
      "loss": 0.0,
      "step": 930
    },
    {
      "epoch": 0.1565890169035405,
      "grad_norm": NaN,
      "learning_rate": 9.509498034113671e-05,
      "loss": 0.0,
      "step": 931
    },
    {
      "epoch": 0.15675721133630477,
      "grad_norm": NaN,
      "learning_rate": 9.508336563739178e-05,
      "loss": 0.0,
      "step": 932
    },
    {
      "epoch": 0.15692540576906905,
      "grad_norm": NaN,
      "learning_rate": 9.507173790956746e-05,
      "loss": 0.0,
      "step": 933
    },
    {
      "epoch": 0.15709360020183333,
      "grad_norm": NaN,
      "learning_rate": 9.506009716102288e-05,
      "loss": 0.0,
      "step": 934
    },
    {
      "epoch": 0.1572617946345976,
      "grad_norm": NaN,
      "learning_rate": 9.504844339512095e-05,
      "loss": 0.0,
      "step": 935
    },
    {
      "epoch": 0.15742998906736186,
      "grad_norm": NaN,
      "learning_rate": 9.50367766152283e-05,
      "loss": 0.0,
      "step": 936
    },
    {
      "epoch": 0.15759818350012614,
      "grad_norm": NaN,
      "learning_rate": 9.502509682471532e-05,
      "loss": 0.0,
      "step": 937
    },
    {
      "epoch": 0.15776637793289042,
      "grad_norm": NaN,
      "learning_rate": 9.501340402695617e-05,
      "loss": 0.0,
      "step": 938
    },
    {
      "epoch": 0.1579345723656547,
      "grad_norm": NaN,
      "learning_rate": 9.50016982253288e-05,
      "loss": 0.0,
      "step": 939
    },
    {
      "epoch": 0.15810276679841898,
      "grad_norm": NaN,
      "learning_rate": 9.498997942321483e-05,
      "loss": 0.0,
      "step": 940
    },
    {
      "epoch": 0.15827096123118325,
      "grad_norm": NaN,
      "learning_rate": 9.497824762399974e-05,
      "loss": 0.0,
      "step": 941
    },
    {
      "epoch": 0.15843915566394753,
      "grad_norm": NaN,
      "learning_rate": 9.496650283107269e-05,
      "loss": 0.0,
      "step": 942
    },
    {
      "epoch": 0.15860735009671179,
      "grad_norm": NaN,
      "learning_rate": 9.495474504782661e-05,
      "loss": 0.0,
      "step": 943
    },
    {
      "epoch": 0.15877554452947606,
      "grad_norm": NaN,
      "learning_rate": 9.494297427765824e-05,
      "loss": 0.0,
      "step": 944
    },
    {
      "epoch": 0.15894373896224034,
      "grad_norm": NaN,
      "learning_rate": 9.493119052396797e-05,
      "loss": 0.0,
      "step": 945
    },
    {
      "epoch": 0.15911193339500462,
      "grad_norm": NaN,
      "learning_rate": 9.491939379016003e-05,
      "loss": 0.0,
      "step": 946
    },
    {
      "epoch": 0.1592801278277689,
      "grad_norm": NaN,
      "learning_rate": 9.490758407964234e-05,
      "loss": 0.0,
      "step": 947
    },
    {
      "epoch": 0.15944832226053318,
      "grad_norm": NaN,
      "learning_rate": 9.489576139582661e-05,
      "loss": 0.0,
      "step": 948
    },
    {
      "epoch": 0.15961651669329746,
      "grad_norm": NaN,
      "learning_rate": 9.488392574212827e-05,
      "loss": 0.0,
      "step": 949
    },
    {
      "epoch": 0.15978471112606174,
      "grad_norm": NaN,
      "learning_rate": 9.487207712196651e-05,
      "loss": 0.0,
      "step": 950
    },
    {
      "epoch": 0.159952905558826,
      "grad_norm": NaN,
      "learning_rate": 9.486021553876426e-05,
      "loss": 0.0,
      "step": 951
    },
    {
      "epoch": 0.16012109999159027,
      "grad_norm": NaN,
      "learning_rate": 9.484834099594822e-05,
      "loss": 0.0,
      "step": 952
    },
    {
      "epoch": 0.16028929442435455,
      "grad_norm": NaN,
      "learning_rate": 9.483645349694879e-05,
      "loss": 0.0,
      "step": 953
    },
    {
      "epoch": 0.16045748885711883,
      "grad_norm": NaN,
      "learning_rate": 9.482455304520013e-05,
      "loss": 0.0,
      "step": 954
    },
    {
      "epoch": 0.1606256832898831,
      "grad_norm": NaN,
      "learning_rate": 9.481263964414016e-05,
      "loss": 0.0,
      "step": 955
    },
    {
      "epoch": 0.1607938777226474,
      "grad_norm": NaN,
      "learning_rate": 9.480071329721054e-05,
      "loss": 0.0,
      "step": 956
    },
    {
      "epoch": 0.16096207215541167,
      "grad_norm": NaN,
      "learning_rate": 9.478877400785664e-05,
      "loss": 0.0,
      "step": 957
    },
    {
      "epoch": 0.16113026658817592,
      "grad_norm": NaN,
      "learning_rate": 9.47768217795276e-05,
      "loss": 0.0,
      "step": 958
    },
    {
      "epoch": 0.1612984610209402,
      "grad_norm": NaN,
      "learning_rate": 9.476485661567626e-05,
      "loss": 0.0,
      "step": 959
    },
    {
      "epoch": 0.16146665545370448,
      "grad_norm": NaN,
      "learning_rate": 9.475287851975925e-05,
      "loss": 0.0,
      "step": 960
    },
    {
      "epoch": 0.16163484988646876,
      "grad_norm": NaN,
      "learning_rate": 9.474088749523689e-05,
      "loss": 0.0,
      "step": 961
    },
    {
      "epoch": 0.16180304431923304,
      "grad_norm": NaN,
      "learning_rate": 9.472888354557327e-05,
      "loss": 0.0,
      "step": 962
    },
    {
      "epoch": 0.16197123875199732,
      "grad_norm": NaN,
      "learning_rate": 9.471686667423618e-05,
      "loss": 0.0,
      "step": 963
    },
    {
      "epoch": 0.1621394331847616,
      "grad_norm": NaN,
      "learning_rate": 9.470483688469716e-05,
      "loss": 0.0,
      "step": 964
    },
    {
      "epoch": 0.16230762761752587,
      "grad_norm": NaN,
      "learning_rate": 9.469279418043148e-05,
      "loss": 0.0,
      "step": 965
    },
    {
      "epoch": 0.16247582205029013,
      "grad_norm": NaN,
      "learning_rate": 9.468073856491818e-05,
      "loss": 0.0,
      "step": 966
    },
    {
      "epoch": 0.1626440164830544,
      "grad_norm": NaN,
      "learning_rate": 9.466867004163992e-05,
      "loss": 0.0,
      "step": 967
    },
    {
      "epoch": 0.16281221091581868,
      "grad_norm": NaN,
      "learning_rate": 9.465658861408324e-05,
      "loss": 0.0,
      "step": 968
    },
    {
      "epoch": 0.16298040534858296,
      "grad_norm": NaN,
      "learning_rate": 9.464449428573827e-05,
      "loss": 0.0,
      "step": 969
    },
    {
      "epoch": 0.16314859978134724,
      "grad_norm": NaN,
      "learning_rate": 9.463238706009896e-05,
      "loss": 0.0,
      "step": 970
    },
    {
      "epoch": 0.16331679421411152,
      "grad_norm": NaN,
      "learning_rate": 9.462026694066293e-05,
      "loss": 0.0,
      "step": 971
    },
    {
      "epoch": 0.1634849886468758,
      "grad_norm": NaN,
      "learning_rate": 9.460813393093157e-05,
      "loss": 0.0,
      "step": 972
    },
    {
      "epoch": 0.16365318307964005,
      "grad_norm": NaN,
      "learning_rate": 9.459598803440994e-05,
      "loss": 0.0,
      "step": 973
    },
    {
      "epoch": 0.16382137751240433,
      "grad_norm": NaN,
      "learning_rate": 9.458382925460689e-05,
      "loss": 0.0,
      "step": 974
    },
    {
      "epoch": 0.1639895719451686,
      "grad_norm": NaN,
      "learning_rate": 9.457165759503493e-05,
      "loss": 0.0,
      "step": 975
    },
    {
      "epoch": 0.1641577663779329,
      "grad_norm": NaN,
      "learning_rate": 9.455947305921031e-05,
      "loss": 0.0,
      "step": 976
    },
    {
      "epoch": 0.16432596081069717,
      "grad_norm": NaN,
      "learning_rate": 9.454727565065302e-05,
      "loss": 0.0,
      "step": 977
    },
    {
      "epoch": 0.16449415524346145,
      "grad_norm": NaN,
      "learning_rate": 9.453506537288677e-05,
      "loss": 0.0,
      "step": 978
    },
    {
      "epoch": 0.16466234967622573,
      "grad_norm": NaN,
      "learning_rate": 9.452284222943894e-05,
      "loss": 0.0,
      "step": 979
    },
    {
      "epoch": 0.16483054410898998,
      "grad_norm": NaN,
      "learning_rate": 9.451060622384067e-05,
      "loss": 0.0,
      "step": 980
    },
    {
      "epoch": 0.16499873854175426,
      "grad_norm": NaN,
      "learning_rate": 9.449835735962682e-05,
      "loss": 0.0,
      "step": 981
    },
    {
      "epoch": 0.16516693297451854,
      "grad_norm": NaN,
      "learning_rate": 9.448609564033593e-05,
      "loss": 0.0,
      "step": 982
    },
    {
      "epoch": 0.16533512740728282,
      "grad_norm": NaN,
      "learning_rate": 9.447382106951029e-05,
      "loss": 0.0,
      "step": 983
    },
    {
      "epoch": 0.1655033218400471,
      "grad_norm": NaN,
      "learning_rate": 9.446153365069587e-05,
      "loss": 0.0,
      "step": 984
    },
    {
      "epoch": 0.16567151627281138,
      "grad_norm": NaN,
      "learning_rate": 9.444923338744239e-05,
      "loss": 0.0,
      "step": 985
    },
    {
      "epoch": 0.16583971070557565,
      "grad_norm": NaN,
      "learning_rate": 9.443692028330325e-05,
      "loss": 0.0,
      "step": 986
    },
    {
      "epoch": 0.16600790513833993,
      "grad_norm": NaN,
      "learning_rate": 9.442459434183554e-05,
      "loss": 0.0,
      "step": 987
    },
    {
      "epoch": 0.16617609957110419,
      "grad_norm": NaN,
      "learning_rate": 9.441225556660011e-05,
      "loss": 0.0,
      "step": 988
    },
    {
      "epoch": 0.16634429400386846,
      "grad_norm": NaN,
      "learning_rate": 9.43999039611615e-05,
      "loss": 0.0,
      "step": 989
    },
    {
      "epoch": 0.16651248843663274,
      "grad_norm": NaN,
      "learning_rate": 9.438753952908794e-05,
      "loss": 0.0,
      "step": 990
    },
    {
      "epoch": 0.16668068286939702,
      "grad_norm": NaN,
      "learning_rate": 9.437516227395139e-05,
      "loss": 0.0,
      "step": 991
    },
    {
      "epoch": 0.1668488773021613,
      "grad_norm": NaN,
      "learning_rate": 9.436277219932747e-05,
      "loss": 0.0,
      "step": 992
    },
    {
      "epoch": 0.16701707173492558,
      "grad_norm": NaN,
      "learning_rate": 9.435036930879557e-05,
      "loss": 0.0,
      "step": 993
    },
    {
      "epoch": 0.16718526616768986,
      "grad_norm": NaN,
      "learning_rate": 9.433795360593874e-05,
      "loss": 0.0,
      "step": 994
    },
    {
      "epoch": 0.1673534606004541,
      "grad_norm": NaN,
      "learning_rate": 9.432552509434369e-05,
      "loss": 0.0,
      "step": 995
    },
    {
      "epoch": 0.1675216550332184,
      "grad_norm": NaN,
      "learning_rate": 9.431308377760094e-05,
      "loss": 0.0,
      "step": 996
    },
    {
      "epoch": 0.16768984946598267,
      "grad_norm": NaN,
      "learning_rate": 9.430062965930462e-05,
      "loss": 0.0,
      "step": 997
    },
    {
      "epoch": 0.16785804389874695,
      "grad_norm": NaN,
      "learning_rate": 9.428816274305258e-05,
      "loss": 0.0,
      "step": 998
    },
    {
      "epoch": 0.16802623833151123,
      "grad_norm": NaN,
      "learning_rate": 9.427568303244639e-05,
      "loss": 0.0,
      "step": 999
    },
    {
      "epoch": 0.1681944327642755,
      "grad_norm": NaN,
      "learning_rate": 9.426319053109127e-05,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 0.1683626271970398,
      "grad_norm": NaN,
      "learning_rate": 9.425068524259619e-05,
      "loss": 0.0,
      "step": 1001
    },
    {
      "epoch": 0.16853082162980407,
      "grad_norm": NaN,
      "learning_rate": 9.423816717057379e-05,
      "loss": 0.0,
      "step": 1002
    },
    {
      "epoch": 0.16869901606256832,
      "grad_norm": NaN,
      "learning_rate": 9.422563631864038e-05,
      "loss": 0.0,
      "step": 1003
    },
    {
      "epoch": 0.1688672104953326,
      "grad_norm": NaN,
      "learning_rate": 9.4213092690416e-05,
      "loss": 0.0,
      "step": 1004
    },
    {
      "epoch": 0.16903540492809688,
      "grad_norm": NaN,
      "learning_rate": 9.420053628952434e-05,
      "loss": 0.0,
      "step": 1005
    },
    {
      "epoch": 0.16920359936086116,
      "grad_norm": NaN,
      "learning_rate": 9.418796711959284e-05,
      "loss": 0.0,
      "step": 1006
    },
    {
      "epoch": 0.16937179379362544,
      "grad_norm": NaN,
      "learning_rate": 9.417538518425258e-05,
      "loss": 0.0,
      "step": 1007
    },
    {
      "epoch": 0.16953998822638972,
      "grad_norm": NaN,
      "learning_rate": 9.41627904871383e-05,
      "loss": 0.0,
      "step": 1008
    },
    {
      "epoch": 0.169708182659154,
      "grad_norm": NaN,
      "learning_rate": 9.415018303188851e-05,
      "loss": 0.0,
      "step": 1009
    },
    {
      "epoch": 0.16987637709191825,
      "grad_norm": NaN,
      "learning_rate": 9.413756282214537e-05,
      "loss": 0.0,
      "step": 1010
    },
    {
      "epoch": 0.17004457152468253,
      "grad_norm": NaN,
      "learning_rate": 9.412492986155471e-05,
      "loss": 0.0,
      "step": 1011
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": NaN,
      "learning_rate": 9.411228415376601e-05,
      "loss": 0.0,
      "step": 1012
    },
    {
      "epoch": 0.17038096039021108,
      "grad_norm": NaN,
      "learning_rate": 9.409962570243252e-05,
      "loss": 0.0,
      "step": 1013
    },
    {
      "epoch": 0.17054915482297536,
      "grad_norm": NaN,
      "learning_rate": 9.408695451121109e-05,
      "loss": 0.0,
      "step": 1014
    },
    {
      "epoch": 0.17071734925573964,
      "grad_norm": NaN,
      "learning_rate": 9.407427058376233e-05,
      "loss": 0.0,
      "step": 1015
    },
    {
      "epoch": 0.17088554368850392,
      "grad_norm": NaN,
      "learning_rate": 9.406157392375044e-05,
      "loss": 0.0,
      "step": 1016
    },
    {
      "epoch": 0.17105373812126817,
      "grad_norm": NaN,
      "learning_rate": 9.404886453484336e-05,
      "loss": 0.0,
      "step": 1017
    },
    {
      "epoch": 0.17122193255403245,
      "grad_norm": NaN,
      "learning_rate": 9.403614242071271e-05,
      "loss": 0.0,
      "step": 1018
    },
    {
      "epoch": 0.17139012698679673,
      "grad_norm": NaN,
      "learning_rate": 9.402340758503373e-05,
      "loss": 0.0,
      "step": 1019
    },
    {
      "epoch": 0.171558321419561,
      "grad_norm": NaN,
      "learning_rate": 9.40106600314854e-05,
      "loss": 0.0,
      "step": 1020
    },
    {
      "epoch": 0.1717265158523253,
      "grad_norm": NaN,
      "learning_rate": 9.399789976375032e-05,
      "loss": 0.0,
      "step": 1021
    },
    {
      "epoch": 0.17189471028508957,
      "grad_norm": NaN,
      "learning_rate": 9.398512678551481e-05,
      "loss": 0.0,
      "step": 1022
    },
    {
      "epoch": 0.17206290471785385,
      "grad_norm": NaN,
      "learning_rate": 9.397234110046883e-05,
      "loss": 0.0,
      "step": 1023
    },
    {
      "epoch": 0.17223109915061813,
      "grad_norm": NaN,
      "learning_rate": 9.395954271230604e-05,
      "loss": 0.0,
      "step": 1024
    },
    {
      "epoch": 0.17239929358338238,
      "grad_norm": NaN,
      "learning_rate": 9.394673162472373e-05,
      "loss": 0.0,
      "step": 1025
    },
    {
      "epoch": 0.17256748801614666,
      "grad_norm": NaN,
      "learning_rate": 9.393390784142288e-05,
      "loss": 0.0,
      "step": 1026
    },
    {
      "epoch": 0.17273568244891094,
      "grad_norm": NaN,
      "learning_rate": 9.392107136610814e-05,
      "loss": 0.0,
      "step": 1027
    },
    {
      "epoch": 0.17290387688167522,
      "grad_norm": NaN,
      "learning_rate": 9.390822220248782e-05,
      "loss": 0.0,
      "step": 1028
    },
    {
      "epoch": 0.1730720713144395,
      "grad_norm": NaN,
      "learning_rate": 9.389536035427394e-05,
      "loss": 0.0,
      "step": 1029
    },
    {
      "epoch": 0.17324026574720378,
      "grad_norm": NaN,
      "learning_rate": 9.388248582518208e-05,
      "loss": 0.0,
      "step": 1030
    },
    {
      "epoch": 0.17340846017996805,
      "grad_norm": NaN,
      "learning_rate": 9.386959861893158e-05,
      "loss": 0.0,
      "step": 1031
    },
    {
      "epoch": 0.1735766546127323,
      "grad_norm": NaN,
      "learning_rate": 9.385669873924541e-05,
      "loss": 0.0,
      "step": 1032
    },
    {
      "epoch": 0.17374484904549659,
      "grad_norm": NaN,
      "learning_rate": 9.384378618985019e-05,
      "loss": 0.0,
      "step": 1033
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": NaN,
      "learning_rate": 9.383086097447624e-05,
      "loss": 0.0,
      "step": 1034
    },
    {
      "epoch": 0.17408123791102514,
      "grad_norm": NaN,
      "learning_rate": 9.381792309685746e-05,
      "loss": 0.0,
      "step": 1035
    },
    {
      "epoch": 0.17424943234378942,
      "grad_norm": NaN,
      "learning_rate": 9.38049725607315e-05,
      "loss": 0.0,
      "step": 1036
    },
    {
      "epoch": 0.1744176267765537,
      "grad_norm": NaN,
      "learning_rate": 9.379200936983963e-05,
      "loss": 0.0,
      "step": 1037
    },
    {
      "epoch": 0.17458582120931798,
      "grad_norm": NaN,
      "learning_rate": 9.377903352792672e-05,
      "loss": 0.0,
      "step": 1038
    },
    {
      "epoch": 0.17475401564208223,
      "grad_norm": NaN,
      "learning_rate": 9.37660450387414e-05,
      "loss": 0.0,
      "step": 1039
    },
    {
      "epoch": 0.1749222100748465,
      "grad_norm": NaN,
      "learning_rate": 9.375304390603585e-05,
      "loss": 0.0,
      "step": 1040
    },
    {
      "epoch": 0.1750904045076108,
      "grad_norm": NaN,
      "learning_rate": 9.3740030133566e-05,
      "loss": 0.0,
      "step": 1041
    },
    {
      "epoch": 0.17525859894037507,
      "grad_norm": NaN,
      "learning_rate": 9.372700372509135e-05,
      "loss": 0.0,
      "step": 1042
    },
    {
      "epoch": 0.17542679337313935,
      "grad_norm": NaN,
      "learning_rate": 9.37139646843751e-05,
      "loss": 0.0,
      "step": 1043
    },
    {
      "epoch": 0.17559498780590363,
      "grad_norm": NaN,
      "learning_rate": 9.370091301518409e-05,
      "loss": 0.0,
      "step": 1044
    },
    {
      "epoch": 0.1757631822386679,
      "grad_norm": NaN,
      "learning_rate": 9.368784872128878e-05,
      "loss": 0.0,
      "step": 1045
    },
    {
      "epoch": 0.1759313766714322,
      "grad_norm": NaN,
      "learning_rate": 9.367477180646329e-05,
      "loss": 0.0,
      "step": 1046
    },
    {
      "epoch": 0.17609957110419644,
      "grad_norm": NaN,
      "learning_rate": 9.366168227448541e-05,
      "loss": 0.0,
      "step": 1047
    },
    {
      "epoch": 0.17626776553696072,
      "grad_norm": NaN,
      "learning_rate": 9.364858012913658e-05,
      "loss": 0.0,
      "step": 1048
    },
    {
      "epoch": 0.176435959969725,
      "grad_norm": NaN,
      "learning_rate": 9.363546537420184e-05,
      "loss": 0.0,
      "step": 1049
    },
    {
      "epoch": 0.17660415440248928,
      "grad_norm": NaN,
      "learning_rate": 9.36223380134699e-05,
      "loss": 0.0,
      "step": 1050
    },
    {
      "epoch": 0.17677234883525356,
      "grad_norm": NaN,
      "learning_rate": 9.36091980507331e-05,
      "loss": 0.0,
      "step": 1051
    },
    {
      "epoch": 0.17694054326801784,
      "grad_norm": NaN,
      "learning_rate": 9.359604548978742e-05,
      "loss": 0.0,
      "step": 1052
    },
    {
      "epoch": 0.17710873770078212,
      "grad_norm": NaN,
      "learning_rate": 9.358288033443251e-05,
      "loss": 0.0,
      "step": 1053
    },
    {
      "epoch": 0.17727693213354637,
      "grad_norm": NaN,
      "learning_rate": 9.356970258847164e-05,
      "loss": 0.0,
      "step": 1054
    },
    {
      "epoch": 0.17744512656631065,
      "grad_norm": NaN,
      "learning_rate": 9.35565122557117e-05,
      "loss": 0.0,
      "step": 1055
    },
    {
      "epoch": 0.17761332099907493,
      "grad_norm": NaN,
      "learning_rate": 9.354330933996321e-05,
      "loss": 0.0,
      "step": 1056
    },
    {
      "epoch": 0.1777815154318392,
      "grad_norm": NaN,
      "learning_rate": 9.353009384504038e-05,
      "loss": 0.0,
      "step": 1057
    },
    {
      "epoch": 0.17794970986460348,
      "grad_norm": NaN,
      "learning_rate": 9.3516865774761e-05,
      "loss": 0.0,
      "step": 1058
    },
    {
      "epoch": 0.17811790429736776,
      "grad_norm": NaN,
      "learning_rate": 9.350362513294651e-05,
      "loss": 0.0,
      "step": 1059
    },
    {
      "epoch": 0.17828609873013204,
      "grad_norm": NaN,
      "learning_rate": 9.3490371923422e-05,
      "loss": 0.0,
      "step": 1060
    },
    {
      "epoch": 0.17845429316289632,
      "grad_norm": NaN,
      "learning_rate": 9.347710615001614e-05,
      "loss": 0.0,
      "step": 1061
    },
    {
      "epoch": 0.17862248759566057,
      "grad_norm": NaN,
      "learning_rate": 9.346382781656131e-05,
      "loss": 0.0,
      "step": 1062
    },
    {
      "epoch": 0.17879068202842485,
      "grad_norm": NaN,
      "learning_rate": 9.345053692689344e-05,
      "loss": 0.0,
      "step": 1063
    },
    {
      "epoch": 0.17895887646118913,
      "grad_norm": NaN,
      "learning_rate": 9.343723348485212e-05,
      "loss": 0.0,
      "step": 1064
    },
    {
      "epoch": 0.1791270708939534,
      "grad_norm": NaN,
      "learning_rate": 9.342391749428057e-05,
      "loss": 0.0,
      "step": 1065
    },
    {
      "epoch": 0.1792952653267177,
      "grad_norm": NaN,
      "learning_rate": 9.341058895902563e-05,
      "loss": 0.0,
      "step": 1066
    },
    {
      "epoch": 0.17946345975948197,
      "grad_norm": NaN,
      "learning_rate": 9.339724788293778e-05,
      "loss": 0.0,
      "step": 1067
    },
    {
      "epoch": 0.17963165419224625,
      "grad_norm": NaN,
      "learning_rate": 9.338389426987108e-05,
      "loss": 0.0,
      "step": 1068
    },
    {
      "epoch": 0.1797998486250105,
      "grad_norm": NaN,
      "learning_rate": 9.337052812368325e-05,
      "loss": 0.0,
      "step": 1069
    },
    {
      "epoch": 0.17996804305777478,
      "grad_norm": NaN,
      "learning_rate": 9.335714944823564e-05,
      "loss": 0.0,
      "step": 1070
    },
    {
      "epoch": 0.18013623749053906,
      "grad_norm": NaN,
      "learning_rate": 9.334375824739317e-05,
      "loss": 0.0,
      "step": 1071
    },
    {
      "epoch": 0.18030443192330334,
      "grad_norm": NaN,
      "learning_rate": 9.333035452502444e-05,
      "loss": 0.0,
      "step": 1072
    },
    {
      "epoch": 0.18047262635606762,
      "grad_norm": NaN,
      "learning_rate": 9.331693828500159e-05,
      "loss": 0.0,
      "step": 1073
    },
    {
      "epoch": 0.1806408207888319,
      "grad_norm": NaN,
      "learning_rate": 9.330350953120045e-05,
      "loss": 0.0,
      "step": 1074
    },
    {
      "epoch": 0.18080901522159618,
      "grad_norm": NaN,
      "learning_rate": 9.329006826750044e-05,
      "loss": 0.0,
      "step": 1075
    },
    {
      "epoch": 0.18097720965436043,
      "grad_norm": NaN,
      "learning_rate": 9.327661449778457e-05,
      "loss": 0.0,
      "step": 1076
    },
    {
      "epoch": 0.1811454040871247,
      "grad_norm": NaN,
      "learning_rate": 9.326314822593953e-05,
      "loss": 0.0,
      "step": 1077
    },
    {
      "epoch": 0.18131359851988899,
      "grad_norm": NaN,
      "learning_rate": 9.324966945585551e-05,
      "loss": 0.0,
      "step": 1078
    },
    {
      "epoch": 0.18148179295265326,
      "grad_norm": NaN,
      "learning_rate": 9.323617819142643e-05,
      "loss": 0.0,
      "step": 1079
    },
    {
      "epoch": 0.18164998738541754,
      "grad_norm": NaN,
      "learning_rate": 9.322267443654972e-05,
      "loss": 0.0,
      "step": 1080
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": NaN,
      "learning_rate": 9.320915819512651e-05,
      "loss": 0.0,
      "step": 1081
    },
    {
      "epoch": 0.1819863762509461,
      "grad_norm": NaN,
      "learning_rate": 9.319562947106146e-05,
      "loss": 0.0,
      "step": 1082
    },
    {
      "epoch": 0.18215457068371038,
      "grad_norm": NaN,
      "learning_rate": 9.318208826826287e-05,
      "loss": 0.0,
      "step": 1083
    },
    {
      "epoch": 0.18232276511647463,
      "grad_norm": NaN,
      "learning_rate": 9.316853459064265e-05,
      "loss": 0.0,
      "step": 1084
    },
    {
      "epoch": 0.1824909595492389,
      "grad_norm": NaN,
      "learning_rate": 9.31549684421163e-05,
      "loss": 0.0,
      "step": 1085
    },
    {
      "epoch": 0.1826591539820032,
      "grad_norm": NaN,
      "learning_rate": 9.314138982660295e-05,
      "loss": 0.0,
      "step": 1086
    },
    {
      "epoch": 0.18282734841476747,
      "grad_norm": NaN,
      "learning_rate": 9.312779874802526e-05,
      "loss": 0.0,
      "step": 1087
    },
    {
      "epoch": 0.18299554284753175,
      "grad_norm": NaN,
      "learning_rate": 9.311419521030959e-05,
      "loss": 0.0,
      "step": 1088
    },
    {
      "epoch": 0.18316373728029603,
      "grad_norm": NaN,
      "learning_rate": 9.310057921738582e-05,
      "loss": 0.0,
      "step": 1089
    },
    {
      "epoch": 0.1833319317130603,
      "grad_norm": NaN,
      "learning_rate": 9.308695077318747e-05,
      "loss": 0.0,
      "step": 1090
    },
    {
      "epoch": 0.18350012614582456,
      "grad_norm": NaN,
      "learning_rate": 9.307330988165165e-05,
      "loss": 0.0,
      "step": 1091
    },
    {
      "epoch": 0.18366832057858884,
      "grad_norm": NaN,
      "learning_rate": 9.305965654671904e-05,
      "loss": 0.0,
      "step": 1092
    },
    {
      "epoch": 0.18383651501135312,
      "grad_norm": NaN,
      "learning_rate": 9.304599077233395e-05,
      "loss": 0.0,
      "step": 1093
    },
    {
      "epoch": 0.1840047094441174,
      "grad_norm": NaN,
      "learning_rate": 9.30323125624443e-05,
      "loss": 0.0,
      "step": 1094
    },
    {
      "epoch": 0.18417290387688168,
      "grad_norm": NaN,
      "learning_rate": 9.30186219210015e-05,
      "loss": 0.0,
      "step": 1095
    },
    {
      "epoch": 0.18434109830964596,
      "grad_norm": NaN,
      "learning_rate": 9.30049188519607e-05,
      "loss": 0.0,
      "step": 1096
    },
    {
      "epoch": 0.18450929274241024,
      "grad_norm": NaN,
      "learning_rate": 9.29912033592805e-05,
      "loss": 0.0,
      "step": 1097
    },
    {
      "epoch": 0.18467748717517452,
      "grad_norm": NaN,
      "learning_rate": 9.297747544692319e-05,
      "loss": 0.0,
      "step": 1098
    },
    {
      "epoch": 0.18484568160793877,
      "grad_norm": NaN,
      "learning_rate": 9.29637351188546e-05,
      "loss": 0.0,
      "step": 1099
    },
    {
      "epoch": 0.18501387604070305,
      "grad_norm": NaN,
      "learning_rate": 9.294998237904414e-05,
      "loss": 0.0,
      "step": 1100
    },
    {
      "epoch": 0.18518207047346733,
      "grad_norm": NaN,
      "learning_rate": 9.293621723146485e-05,
      "loss": 0.0,
      "step": 1101
    },
    {
      "epoch": 0.1853502649062316,
      "grad_norm": NaN,
      "learning_rate": 9.292243968009331e-05,
      "loss": 0.0,
      "step": 1102
    },
    {
      "epoch": 0.18551845933899588,
      "grad_norm": NaN,
      "learning_rate": 9.290864972890971e-05,
      "loss": 0.0,
      "step": 1103
    },
    {
      "epoch": 0.18568665377176016,
      "grad_norm": NaN,
      "learning_rate": 9.28948473818978e-05,
      "loss": 0.0,
      "step": 1104
    },
    {
      "epoch": 0.18585484820452444,
      "grad_norm": NaN,
      "learning_rate": 9.288103264304496e-05,
      "loss": 0.0,
      "step": 1105
    },
    {
      "epoch": 0.1860230426372887,
      "grad_norm": NaN,
      "learning_rate": 9.286720551634205e-05,
      "loss": 0.0,
      "step": 1106
    },
    {
      "epoch": 0.18619123707005297,
      "grad_norm": NaN,
      "learning_rate": 9.285336600578361e-05,
      "loss": 0.0,
      "step": 1107
    },
    {
      "epoch": 0.18635943150281725,
      "grad_norm": NaN,
      "learning_rate": 9.283951411536773e-05,
      "loss": 0.0,
      "step": 1108
    },
    {
      "epoch": 0.18652762593558153,
      "grad_norm": NaN,
      "learning_rate": 9.282564984909604e-05,
      "loss": 0.0,
      "step": 1109
    },
    {
      "epoch": 0.1866958203683458,
      "grad_norm": NaN,
      "learning_rate": 9.28117732109738e-05,
      "loss": 0.0,
      "step": 1110
    },
    {
      "epoch": 0.1868640148011101,
      "grad_norm": NaN,
      "learning_rate": 9.279788420500979e-05,
      "loss": 0.0,
      "step": 1111
    },
    {
      "epoch": 0.18703220923387437,
      "grad_norm": NaN,
      "learning_rate": 9.278398283521639e-05,
      "loss": 0.0,
      "step": 1112
    },
    {
      "epoch": 0.18720040366663862,
      "grad_norm": NaN,
      "learning_rate": 9.277006910560958e-05,
      "loss": 0.0,
      "step": 1113
    },
    {
      "epoch": 0.1873685980994029,
      "grad_norm": NaN,
      "learning_rate": 9.275614302020883e-05,
      "loss": 0.0,
      "step": 1114
    },
    {
      "epoch": 0.18753679253216718,
      "grad_norm": NaN,
      "learning_rate": 9.274220458303727e-05,
      "loss": 0.0,
      "step": 1115
    },
    {
      "epoch": 0.18770498696493146,
      "grad_norm": NaN,
      "learning_rate": 9.272825379812153e-05,
      "loss": 0.0,
      "step": 1116
    },
    {
      "epoch": 0.18787318139769574,
      "grad_norm": NaN,
      "learning_rate": 9.271429066949186e-05,
      "loss": 0.0,
      "step": 1117
    },
    {
      "epoch": 0.18804137583046002,
      "grad_norm": NaN,
      "learning_rate": 9.270031520118205e-05,
      "loss": 0.0,
      "step": 1118
    },
    {
      "epoch": 0.1882095702632243,
      "grad_norm": NaN,
      "learning_rate": 9.268632739722947e-05,
      "loss": 0.0,
      "step": 1119
    },
    {
      "epoch": 0.18837776469598858,
      "grad_norm": NaN,
      "learning_rate": 9.267232726167499e-05,
      "loss": 0.0,
      "step": 1120
    },
    {
      "epoch": 0.18854595912875283,
      "grad_norm": NaN,
      "learning_rate": 9.265831479856313e-05,
      "loss": 0.0,
      "step": 1121
    },
    {
      "epoch": 0.1887141535615171,
      "grad_norm": NaN,
      "learning_rate": 9.264429001194193e-05,
      "loss": 0.0,
      "step": 1122
    },
    {
      "epoch": 0.18888234799428139,
      "grad_norm": NaN,
      "learning_rate": 9.263025290586299e-05,
      "loss": 0.0,
      "step": 1123
    },
    {
      "epoch": 0.18905054242704566,
      "grad_norm": NaN,
      "learning_rate": 9.261620348438145e-05,
      "loss": 0.0,
      "step": 1124
    },
    {
      "epoch": 0.18921873685980994,
      "grad_norm": NaN,
      "learning_rate": 9.260214175155606e-05,
      "loss": 0.0,
      "step": 1125
    },
    {
      "epoch": 0.18938693129257422,
      "grad_norm": NaN,
      "learning_rate": 9.258806771144911e-05,
      "loss": 0.0,
      "step": 1126
    },
    {
      "epoch": 0.1895551257253385,
      "grad_norm": NaN,
      "learning_rate": 9.257398136812638e-05,
      "loss": 0.0,
      "step": 1127
    },
    {
      "epoch": 0.18972332015810275,
      "grad_norm": NaN,
      "learning_rate": 9.255988272565732e-05,
      "loss": 0.0,
      "step": 1128
    },
    {
      "epoch": 0.18989151459086703,
      "grad_norm": NaN,
      "learning_rate": 9.254577178811482e-05,
      "loss": 0.0,
      "step": 1129
    },
    {
      "epoch": 0.1900597090236313,
      "grad_norm": NaN,
      "learning_rate": 9.25316485595754e-05,
      "loss": 0.0,
      "step": 1130
    },
    {
      "epoch": 0.1902279034563956,
      "grad_norm": NaN,
      "learning_rate": 9.251751304411909e-05,
      "loss": 0.0,
      "step": 1131
    },
    {
      "epoch": 0.19039609788915987,
      "grad_norm": NaN,
      "learning_rate": 9.250336524582948e-05,
      "loss": 0.0,
      "step": 1132
    },
    {
      "epoch": 0.19056429232192415,
      "grad_norm": NaN,
      "learning_rate": 9.248920516879371e-05,
      "loss": 0.0,
      "step": 1133
    },
    {
      "epoch": 0.19073248675468843,
      "grad_norm": NaN,
      "learning_rate": 9.24750328171025e-05,
      "loss": 0.0,
      "step": 1134
    },
    {
      "epoch": 0.1909006811874527,
      "grad_norm": NaN,
      "learning_rate": 9.246084819485004e-05,
      "loss": 0.0,
      "step": 1135
    },
    {
      "epoch": 0.19106887562021696,
      "grad_norm": NaN,
      "learning_rate": 9.244665130613411e-05,
      "loss": 0.0,
      "step": 1136
    },
    {
      "epoch": 0.19123707005298124,
      "grad_norm": NaN,
      "learning_rate": 9.243244215505607e-05,
      "loss": 0.0,
      "step": 1137
    },
    {
      "epoch": 0.19140526448574552,
      "grad_norm": NaN,
      "learning_rate": 9.241822074572076e-05,
      "loss": 0.0,
      "step": 1138
    },
    {
      "epoch": 0.1915734589185098,
      "grad_norm": NaN,
      "learning_rate": 9.240398708223656e-05,
      "loss": 0.0,
      "step": 1139
    },
    {
      "epoch": 0.19174165335127408,
      "grad_norm": NaN,
      "learning_rate": 9.238974116871548e-05,
      "loss": 0.0,
      "step": 1140
    },
    {
      "epoch": 0.19190984778403836,
      "grad_norm": NaN,
      "learning_rate": 9.237548300927294e-05,
      "loss": 0.0,
      "step": 1141
    },
    {
      "epoch": 0.19207804221680264,
      "grad_norm": NaN,
      "learning_rate": 9.236121260802802e-05,
      "loss": 0.0,
      "step": 1142
    },
    {
      "epoch": 0.1922462366495669,
      "grad_norm": NaN,
      "learning_rate": 9.234692996910323e-05,
      "loss": 0.0,
      "step": 1143
    },
    {
      "epoch": 0.19241443108233117,
      "grad_norm": NaN,
      "learning_rate": 9.233263509662469e-05,
      "loss": 0.0,
      "step": 1144
    },
    {
      "epoch": 0.19258262551509545,
      "grad_norm": NaN,
      "learning_rate": 9.231832799472202e-05,
      "loss": 0.0,
      "step": 1145
    },
    {
      "epoch": 0.19275081994785973,
      "grad_norm": NaN,
      "learning_rate": 9.23040086675284e-05,
      "loss": 0.0,
      "step": 1146
    },
    {
      "epoch": 0.192919014380624,
      "grad_norm": NaN,
      "learning_rate": 9.228967711918048e-05,
      "loss": 0.0,
      "step": 1147
    },
    {
      "epoch": 0.19308720881338828,
      "grad_norm": NaN,
      "learning_rate": 9.227533335381854e-05,
      "loss": 0.0,
      "step": 1148
    },
    {
      "epoch": 0.19325540324615256,
      "grad_norm": NaN,
      "learning_rate": 9.22609773755863e-05,
      "loss": 0.0,
      "step": 1149
    },
    {
      "epoch": 0.19342359767891681,
      "grad_norm": NaN,
      "learning_rate": 9.224660918863104e-05,
      "loss": 0.0,
      "step": 1150
    },
    {
      "epoch": 0.1935917921116811,
      "grad_norm": NaN,
      "learning_rate": 9.223222879710356e-05,
      "loss": 0.0,
      "step": 1151
    },
    {
      "epoch": 0.19375998654444537,
      "grad_norm": NaN,
      "learning_rate": 9.221783620515823e-05,
      "loss": 0.0,
      "step": 1152
    },
    {
      "epoch": 0.19392818097720965,
      "grad_norm": NaN,
      "learning_rate": 9.220343141695287e-05,
      "loss": 0.0,
      "step": 1153
    },
    {
      "epoch": 0.19409637540997393,
      "grad_norm": NaN,
      "learning_rate": 9.218901443664889e-05,
      "loss": 0.0,
      "step": 1154
    },
    {
      "epoch": 0.1942645698427382,
      "grad_norm": NaN,
      "learning_rate": 9.217458526841118e-05,
      "loss": 0.0,
      "step": 1155
    },
    {
      "epoch": 0.1944327642755025,
      "grad_norm": NaN,
      "learning_rate": 9.216014391640818e-05,
      "loss": 0.0,
      "step": 1156
    },
    {
      "epoch": 0.19460095870826677,
      "grad_norm": NaN,
      "learning_rate": 9.214569038481183e-05,
      "loss": 0.0,
      "step": 1157
    },
    {
      "epoch": 0.19476915314103102,
      "grad_norm": NaN,
      "learning_rate": 9.213122467779756e-05,
      "loss": 0.0,
      "step": 1158
    },
    {
      "epoch": 0.1949373475737953,
      "grad_norm": NaN,
      "learning_rate": 9.211674679954439e-05,
      "loss": 0.0,
      "step": 1159
    },
    {
      "epoch": 0.19510554200655958,
      "grad_norm": NaN,
      "learning_rate": 9.21022567542348e-05,
      "loss": 0.0,
      "step": 1160
    },
    {
      "epoch": 0.19527373643932386,
      "grad_norm": NaN,
      "learning_rate": 9.20877545460548e-05,
      "loss": 0.0,
      "step": 1161
    },
    {
      "epoch": 0.19544193087208814,
      "grad_norm": NaN,
      "learning_rate": 9.207324017919393e-05,
      "loss": 0.0,
      "step": 1162
    },
    {
      "epoch": 0.19561012530485242,
      "grad_norm": NaN,
      "learning_rate": 9.205871365784521e-05,
      "loss": 0.0,
      "step": 1163
    },
    {
      "epoch": 0.1957783197376167,
      "grad_norm": NaN,
      "learning_rate": 9.204417498620522e-05,
      "loss": 0.0,
      "step": 1164
    },
    {
      "epoch": 0.19594651417038095,
      "grad_norm": NaN,
      "learning_rate": 9.202962416847398e-05,
      "loss": 0.0,
      "step": 1165
    },
    {
      "epoch": 0.19611470860314523,
      "grad_norm": NaN,
      "learning_rate": 9.201506120885507e-05,
      "loss": 0.0,
      "step": 1166
    },
    {
      "epoch": 0.1962829030359095,
      "grad_norm": NaN,
      "learning_rate": 9.20004861115556e-05,
      "loss": 0.0,
      "step": 1167
    },
    {
      "epoch": 0.19645109746867379,
      "grad_norm": NaN,
      "learning_rate": 9.198589888078611e-05,
      "loss": 0.0,
      "step": 1168
    },
    {
      "epoch": 0.19661929190143806,
      "grad_norm": NaN,
      "learning_rate": 9.19712995207607e-05,
      "loss": 0.0,
      "step": 1169
    },
    {
      "epoch": 0.19678748633420234,
      "grad_norm": NaN,
      "learning_rate": 9.195668803569696e-05,
      "loss": 0.0,
      "step": 1170
    },
    {
      "epoch": 0.19695568076696662,
      "grad_norm": NaN,
      "learning_rate": 9.194206442981601e-05,
      "loss": 0.0,
      "step": 1171
    },
    {
      "epoch": 0.1971238751997309,
      "grad_norm": NaN,
      "learning_rate": 9.192742870734241e-05,
      "loss": 0.0,
      "step": 1172
    },
    {
      "epoch": 0.19729206963249515,
      "grad_norm": NaN,
      "learning_rate": 9.191278087250426e-05,
      "loss": 0.0,
      "step": 1173
    },
    {
      "epoch": 0.19746026406525943,
      "grad_norm": NaN,
      "learning_rate": 9.189812092953319e-05,
      "loss": 0.0,
      "step": 1174
    },
    {
      "epoch": 0.1976284584980237,
      "grad_norm": NaN,
      "learning_rate": 9.188344888266425e-05,
      "loss": 0.0,
      "step": 1175
    },
    {
      "epoch": 0.197796652930788,
      "grad_norm": NaN,
      "learning_rate": 9.186876473613605e-05,
      "loss": 0.0,
      "step": 1176
    },
    {
      "epoch": 0.19796484736355227,
      "grad_norm": NaN,
      "learning_rate": 9.185406849419067e-05,
      "loss": 0.0,
      "step": 1177
    },
    {
      "epoch": 0.19813304179631655,
      "grad_norm": NaN,
      "learning_rate": 9.183936016107371e-05,
      "loss": 0.0,
      "step": 1178
    },
    {
      "epoch": 0.19830123622908083,
      "grad_norm": NaN,
      "learning_rate": 9.182463974103422e-05,
      "loss": 0.0,
      "step": 1179
    },
    {
      "epoch": 0.19846943066184508,
      "grad_norm": NaN,
      "learning_rate": 9.180990723832478e-05,
      "loss": 0.0,
      "step": 1180
    },
    {
      "epoch": 0.19863762509460936,
      "grad_norm": NaN,
      "learning_rate": 9.179516265720143e-05,
      "loss": 0.0,
      "step": 1181
    },
    {
      "epoch": 0.19880581952737364,
      "grad_norm": NaN,
      "learning_rate": 9.178040600192371e-05,
      "loss": 0.0,
      "step": 1182
    },
    {
      "epoch": 0.19897401396013792,
      "grad_norm": NaN,
      "learning_rate": 9.176563727675467e-05,
      "loss": 0.0,
      "step": 1183
    },
    {
      "epoch": 0.1991422083929022,
      "grad_norm": NaN,
      "learning_rate": 9.175085648596083e-05,
      "loss": 0.0,
      "step": 1184
    },
    {
      "epoch": 0.19931040282566648,
      "grad_norm": NaN,
      "learning_rate": 9.173606363381219e-05,
      "loss": 0.0,
      "step": 1185
    },
    {
      "epoch": 0.19947859725843076,
      "grad_norm": NaN,
      "learning_rate": 9.172125872458223e-05,
      "loss": 0.0,
      "step": 1186
    },
    {
      "epoch": 0.199646791691195,
      "grad_norm": NaN,
      "learning_rate": 9.170644176254792e-05,
      "loss": 0.0,
      "step": 1187
    },
    {
      "epoch": 0.1998149861239593,
      "grad_norm": NaN,
      "learning_rate": 9.169161275198975e-05,
      "loss": 0.0,
      "step": 1188
    },
    {
      "epoch": 0.19998318055672357,
      "grad_norm": NaN,
      "learning_rate": 9.16767716971916e-05,
      "loss": 0.0,
      "step": 1189
    },
    {
      "epoch": 0.20015137498948785,
      "grad_norm": NaN,
      "learning_rate": 9.166191860244093e-05,
      "loss": 0.0,
      "step": 1190
    },
    {
      "epoch": 0.20031956942225213,
      "grad_norm": NaN,
      "learning_rate": 9.16470534720286e-05,
      "loss": 0.0,
      "step": 1191
    },
    {
      "epoch": 0.2004877638550164,
      "grad_norm": NaN,
      "learning_rate": 9.163217631024901e-05,
      "loss": 0.0,
      "step": 1192
    },
    {
      "epoch": 0.20065595828778068,
      "grad_norm": NaN,
      "learning_rate": 9.161728712139998e-05,
      "loss": 0.0,
      "step": 1193
    },
    {
      "epoch": 0.20082415272054496,
      "grad_norm": NaN,
      "learning_rate": 9.160238590978286e-05,
      "loss": 0.0,
      "step": 1194
    },
    {
      "epoch": 0.20099234715330921,
      "grad_norm": NaN,
      "learning_rate": 9.158747267970242e-05,
      "loss": 0.0,
      "step": 1195
    },
    {
      "epoch": 0.2011605415860735,
      "grad_norm": NaN,
      "learning_rate": 9.157254743546691e-05,
      "loss": 0.0,
      "step": 1196
    },
    {
      "epoch": 0.20132873601883777,
      "grad_norm": NaN,
      "learning_rate": 9.15576101813881e-05,
      "loss": 0.0,
      "step": 1197
    },
    {
      "epoch": 0.20149693045160205,
      "grad_norm": NaN,
      "learning_rate": 9.154266092178117e-05,
      "loss": 0.0,
      "step": 1198
    },
    {
      "epoch": 0.20166512488436633,
      "grad_norm": NaN,
      "learning_rate": 9.152769966096482e-05,
      "loss": 0.0,
      "step": 1199
    },
    {
      "epoch": 0.2018333193171306,
      "grad_norm": NaN,
      "learning_rate": 9.151272640326116e-05,
      "loss": 0.0,
      "step": 1200
    },
    {
      "epoch": 0.2020015137498949,
      "grad_norm": NaN,
      "learning_rate": 9.149774115299584e-05,
      "loss": 0.0,
      "step": 1201
    },
    {
      "epoch": 0.20216970818265914,
      "grad_norm": NaN,
      "learning_rate": 9.148274391449788e-05,
      "loss": 0.0,
      "step": 1202
    },
    {
      "epoch": 0.20233790261542342,
      "grad_norm": NaN,
      "learning_rate": 9.146773469209982e-05,
      "loss": 0.0,
      "step": 1203
    },
    {
      "epoch": 0.2025060970481877,
      "grad_norm": NaN,
      "learning_rate": 9.145271349013769e-05,
      "loss": 0.0,
      "step": 1204
    },
    {
      "epoch": 0.20267429148095198,
      "grad_norm": NaN,
      "learning_rate": 9.143768031295094e-05,
      "loss": 0.0,
      "step": 1205
    },
    {
      "epoch": 0.20284248591371626,
      "grad_norm": NaN,
      "learning_rate": 9.142263516488245e-05,
      "loss": 0.0,
      "step": 1206
    },
    {
      "epoch": 0.20301068034648054,
      "grad_norm": NaN,
      "learning_rate": 9.140757805027864e-05,
      "loss": 0.0,
      "step": 1207
    },
    {
      "epoch": 0.20317887477924482,
      "grad_norm": NaN,
      "learning_rate": 9.139250897348933e-05,
      "loss": 0.0,
      "step": 1208
    },
    {
      "epoch": 0.2033470692120091,
      "grad_norm": NaN,
      "learning_rate": 9.137742793886779e-05,
      "loss": 0.0,
      "step": 1209
    },
    {
      "epoch": 0.20351526364477335,
      "grad_norm": NaN,
      "learning_rate": 9.136233495077077e-05,
      "loss": 0.0,
      "step": 1210
    },
    {
      "epoch": 0.20368345807753763,
      "grad_norm": NaN,
      "learning_rate": 9.134723001355847e-05,
      "loss": 0.0,
      "step": 1211
    },
    {
      "epoch": 0.2038516525103019,
      "grad_norm": NaN,
      "learning_rate": 9.133211313159453e-05,
      "loss": 0.0,
      "step": 1212
    },
    {
      "epoch": 0.20401984694306619,
      "grad_norm": NaN,
      "learning_rate": 9.131698430924605e-05,
      "loss": 0.0,
      "step": 1213
    },
    {
      "epoch": 0.20418804137583046,
      "grad_norm": NaN,
      "learning_rate": 9.130184355088359e-05,
      "loss": 0.0,
      "step": 1214
    },
    {
      "epoch": 0.20435623580859474,
      "grad_norm": NaN,
      "learning_rate": 9.128669086088112e-05,
      "loss": 0.0,
      "step": 1215
    },
    {
      "epoch": 0.20452443024135902,
      "grad_norm": NaN,
      "learning_rate": 9.127152624361612e-05,
      "loss": 0.0,
      "step": 1216
    },
    {
      "epoch": 0.20469262467412327,
      "grad_norm": NaN,
      "learning_rate": 9.125634970346945e-05,
      "loss": 0.0,
      "step": 1217
    },
    {
      "epoch": 0.20486081910688755,
      "grad_norm": NaN,
      "learning_rate": 9.124116124482545e-05,
      "loss": 0.0,
      "step": 1218
    },
    {
      "epoch": 0.20502901353965183,
      "grad_norm": NaN,
      "learning_rate": 9.122596087207187e-05,
      "loss": 0.0,
      "step": 1219
    },
    {
      "epoch": 0.2051972079724161,
      "grad_norm": NaN,
      "learning_rate": 9.121074858959997e-05,
      "loss": 0.0,
      "step": 1220
    },
    {
      "epoch": 0.2053654024051804,
      "grad_norm": NaN,
      "learning_rate": 9.11955244018044e-05,
      "loss": 0.0,
      "step": 1221
    },
    {
      "epoch": 0.20553359683794467,
      "grad_norm": NaN,
      "learning_rate": 9.118028831308322e-05,
      "loss": 0.0,
      "step": 1222
    },
    {
      "epoch": 0.20570179127070895,
      "grad_norm": NaN,
      "learning_rate": 9.116504032783802e-05,
      "loss": 0.0,
      "step": 1223
    },
    {
      "epoch": 0.2058699857034732,
      "grad_norm": NaN,
      "learning_rate": 9.114978045047372e-05,
      "loss": 0.0,
      "step": 1224
    },
    {
      "epoch": 0.20603818013623748,
      "grad_norm": NaN,
      "learning_rate": 9.113450868539876e-05,
      "loss": 0.0,
      "step": 1225
    },
    {
      "epoch": 0.20620637456900176,
      "grad_norm": NaN,
      "learning_rate": 9.111922503702498e-05,
      "loss": 0.0,
      "step": 1226
    },
    {
      "epoch": 0.20637456900176604,
      "grad_norm": NaN,
      "learning_rate": 9.110392950976764e-05,
      "loss": 0.0,
      "step": 1227
    },
    {
      "epoch": 0.20654276343453032,
      "grad_norm": NaN,
      "learning_rate": 9.108862210804548e-05,
      "loss": 0.0,
      "step": 1228
    },
    {
      "epoch": 0.2067109578672946,
      "grad_norm": NaN,
      "learning_rate": 9.107330283628058e-05,
      "loss": 0.0,
      "step": 1229
    },
    {
      "epoch": 0.20687915230005888,
      "grad_norm": NaN,
      "learning_rate": 9.105797169889854e-05,
      "loss": 0.0,
      "step": 1230
    },
    {
      "epoch": 0.20704734673282316,
      "grad_norm": NaN,
      "learning_rate": 9.104262870032837e-05,
      "loss": 0.0,
      "step": 1231
    },
    {
      "epoch": 0.2072155411655874,
      "grad_norm": NaN,
      "learning_rate": 9.102727384500247e-05,
      "loss": 0.0,
      "step": 1232
    },
    {
      "epoch": 0.2073837355983517,
      "grad_norm": NaN,
      "learning_rate": 9.101190713735667e-05,
      "loss": 0.0,
      "step": 1233
    },
    {
      "epoch": 0.20755193003111597,
      "grad_norm": NaN,
      "learning_rate": 9.099652858183028e-05,
      "loss": 0.0,
      "step": 1234
    },
    {
      "epoch": 0.20772012446388025,
      "grad_norm": NaN,
      "learning_rate": 9.098113818286595e-05,
      "loss": 0.0,
      "step": 1235
    },
    {
      "epoch": 0.20788831889664453,
      "grad_norm": NaN,
      "learning_rate": 9.096573594490982e-05,
      "loss": 0.0,
      "step": 1236
    },
    {
      "epoch": 0.2080565133294088,
      "grad_norm": NaN,
      "learning_rate": 9.095032187241143e-05,
      "loss": 0.0,
      "step": 1237
    },
    {
      "epoch": 0.20822470776217308,
      "grad_norm": NaN,
      "learning_rate": 9.09348959698237e-05,
      "loss": 0.0,
      "step": 1238
    },
    {
      "epoch": 0.20839290219493734,
      "grad_norm": NaN,
      "learning_rate": 9.091945824160304e-05,
      "loss": 0.0,
      "step": 1239
    },
    {
      "epoch": 0.20856109662770161,
      "grad_norm": NaN,
      "learning_rate": 9.09040086922092e-05,
      "loss": 0.0,
      "step": 1240
    },
    {
      "epoch": 0.2087292910604659,
      "grad_norm": NaN,
      "learning_rate": 9.088854732610543e-05,
      "loss": 0.0,
      "step": 1241
    },
    {
      "epoch": 0.20889748549323017,
      "grad_norm": NaN,
      "learning_rate": 9.08730741477583e-05,
      "loss": 0.0,
      "step": 1242
    },
    {
      "epoch": 0.20906567992599445,
      "grad_norm": NaN,
      "learning_rate": 9.085758916163785e-05,
      "loss": 0.0,
      "step": 1243
    },
    {
      "epoch": 0.20923387435875873,
      "grad_norm": NaN,
      "learning_rate": 9.084209237221752e-05,
      "loss": 0.0,
      "step": 1244
    },
    {
      "epoch": 0.209402068791523,
      "grad_norm": NaN,
      "learning_rate": 9.082658378397417e-05,
      "loss": 0.0,
      "step": 1245
    },
    {
      "epoch": 0.2095702632242873,
      "grad_norm": NaN,
      "learning_rate": 9.081106340138803e-05,
      "loss": 0.0,
      "step": 1246
    },
    {
      "epoch": 0.20973845765705154,
      "grad_norm": NaN,
      "learning_rate": 9.07955312289428e-05,
      "loss": 0.0,
      "step": 1247
    },
    {
      "epoch": 0.20990665208981582,
      "grad_norm": NaN,
      "learning_rate": 9.077998727112554e-05,
      "loss": 0.0,
      "step": 1248
    },
    {
      "epoch": 0.2100748465225801,
      "grad_norm": NaN,
      "learning_rate": 9.076443153242671e-05,
      "loss": 0.0,
      "step": 1249
    },
    {
      "epoch": 0.21024304095534438,
      "grad_norm": NaN,
      "learning_rate": 9.07488640173402e-05,
      "loss": 0.0,
      "step": 1250
    },
    {
      "epoch": 0.21041123538810866,
      "grad_norm": NaN,
      "learning_rate": 9.073328473036328e-05,
      "loss": 0.0,
      "step": 1251
    },
    {
      "epoch": 0.21057942982087294,
      "grad_norm": NaN,
      "learning_rate": 9.071769367599665e-05,
      "loss": 0.0,
      "step": 1252
    },
    {
      "epoch": 0.21074762425363722,
      "grad_norm": NaN,
      "learning_rate": 9.070209085874437e-05,
      "loss": 0.0,
      "step": 1253
    },
    {
      "epoch": 0.21091581868640147,
      "grad_norm": NaN,
      "learning_rate": 9.068647628311395e-05,
      "loss": 0.0,
      "step": 1254
    },
    {
      "epoch": 0.21108401311916575,
      "grad_norm": NaN,
      "learning_rate": 9.067084995361623e-05,
      "loss": 0.0,
      "step": 1255
    },
    {
      "epoch": 0.21125220755193003,
      "grad_norm": NaN,
      "learning_rate": 9.065521187476551e-05,
      "loss": 0.0,
      "step": 1256
    },
    {
      "epoch": 0.2114204019846943,
      "grad_norm": NaN,
      "learning_rate": 9.063956205107944e-05,
      "loss": 0.0,
      "step": 1257
    },
    {
      "epoch": 0.21158859641745859,
      "grad_norm": NaN,
      "learning_rate": 9.062390048707907e-05,
      "loss": 0.0,
      "step": 1258
    },
    {
      "epoch": 0.21175679085022286,
      "grad_norm": NaN,
      "learning_rate": 9.060822718728888e-05,
      "loss": 0.0,
      "step": 1259
    },
    {
      "epoch": 0.21192498528298714,
      "grad_norm": NaN,
      "learning_rate": 9.059254215623668e-05,
      "loss": 0.0,
      "step": 1260
    },
    {
      "epoch": 0.2120931797157514,
      "grad_norm": NaN,
      "learning_rate": 9.057684539845372e-05,
      "loss": 0.0,
      "step": 1261
    },
    {
      "epoch": 0.21226137414851567,
      "grad_norm": NaN,
      "learning_rate": 9.056113691847461e-05,
      "loss": 0.0,
      "step": 1262
    },
    {
      "epoch": 0.21242956858127995,
      "grad_norm": NaN,
      "learning_rate": 9.054541672083736e-05,
      "loss": 0.0,
      "step": 1263
    },
    {
      "epoch": 0.21259776301404423,
      "grad_norm": NaN,
      "learning_rate": 9.052968481008334e-05,
      "loss": 0.0,
      "step": 1264
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": NaN,
      "learning_rate": 9.051394119075736e-05,
      "loss": 0.0,
      "step": 1265
    },
    {
      "epoch": 0.2129341518795728,
      "grad_norm": NaN,
      "learning_rate": 9.049818586740755e-05,
      "loss": 0.0,
      "step": 1266
    },
    {
      "epoch": 0.21310234631233707,
      "grad_norm": NaN,
      "learning_rate": 9.048241884458543e-05,
      "loss": 0.0,
      "step": 1267
    },
    {
      "epoch": 0.21327054074510135,
      "grad_norm": NaN,
      "learning_rate": 9.046664012684594e-05,
      "loss": 0.0,
      "step": 1268
    },
    {
      "epoch": 0.2134387351778656,
      "grad_norm": NaN,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.0,
      "step": 1269
    },
    {
      "epoch": 0.21360692961062988,
      "grad_norm": NaN,
      "learning_rate": 9.04350476248514e-05,
      "loss": 0.0,
      "step": 1270
    },
    {
      "epoch": 0.21377512404339416,
      "grad_norm": NaN,
      "learning_rate": 9.041923384972306e-05,
      "loss": 0.0,
      "step": 1271
    },
    {
      "epoch": 0.21394331847615844,
      "grad_norm": NaN,
      "learning_rate": 9.040340839793078e-05,
      "loss": 0.0,
      "step": 1272
    },
    {
      "epoch": 0.21411151290892272,
      "grad_norm": NaN,
      "learning_rate": 9.038757127404638e-05,
      "loss": 0.0,
      "step": 1273
    },
    {
      "epoch": 0.214279707341687,
      "grad_norm": NaN,
      "learning_rate": 9.037172248264498e-05,
      "loss": 0.0,
      "step": 1274
    },
    {
      "epoch": 0.21444790177445128,
      "grad_norm": NaN,
      "learning_rate": 9.035586202830517e-05,
      "loss": 0.0,
      "step": 1275
    },
    {
      "epoch": 0.21461609620721553,
      "grad_norm": NaN,
      "learning_rate": 9.033998991560881e-05,
      "loss": 0.0,
      "step": 1276
    },
    {
      "epoch": 0.2147842906399798,
      "grad_norm": NaN,
      "learning_rate": 9.032410614914119e-05,
      "loss": 0.0,
      "step": 1277
    },
    {
      "epoch": 0.2149524850727441,
      "grad_norm": NaN,
      "learning_rate": 9.030821073349098e-05,
      "loss": 0.0,
      "step": 1278
    },
    {
      "epoch": 0.21512067950550837,
      "grad_norm": NaN,
      "learning_rate": 9.029230367325017e-05,
      "loss": 0.0,
      "step": 1279
    },
    {
      "epoch": 0.21528887393827265,
      "grad_norm": NaN,
      "learning_rate": 9.027638497301412e-05,
      "loss": 0.0,
      "step": 1280
    },
    {
      "epoch": 0.21545706837103692,
      "grad_norm": NaN,
      "learning_rate": 9.026045463738155e-05,
      "loss": 0.0,
      "step": 1281
    },
    {
      "epoch": 0.2156252628038012,
      "grad_norm": NaN,
      "learning_rate": 9.024451267095461e-05,
      "loss": 0.0,
      "step": 1282
    },
    {
      "epoch": 0.21579345723656548,
      "grad_norm": NaN,
      "learning_rate": 9.022855907833871e-05,
      "loss": 0.0,
      "step": 1283
    },
    {
      "epoch": 0.21596165166932974,
      "grad_norm": NaN,
      "learning_rate": 9.021259386414269e-05,
      "loss": 0.0,
      "step": 1284
    },
    {
      "epoch": 0.21612984610209401,
      "grad_norm": NaN,
      "learning_rate": 9.01966170329787e-05,
      "loss": 0.0,
      "step": 1285
    },
    {
      "epoch": 0.2162980405348583,
      "grad_norm": NaN,
      "learning_rate": 9.018062858946227e-05,
      "loss": 0.0,
      "step": 1286
    },
    {
      "epoch": 0.21646623496762257,
      "grad_norm": NaN,
      "learning_rate": 9.01646285382123e-05,
      "loss": 0.0,
      "step": 1287
    },
    {
      "epoch": 0.21663442940038685,
      "grad_norm": NaN,
      "learning_rate": 9.014861688385101e-05,
      "loss": 0.0,
      "step": 1288
    },
    {
      "epoch": 0.21680262383315113,
      "grad_norm": NaN,
      "learning_rate": 9.0132593631004e-05,
      "loss": 0.0,
      "step": 1289
    },
    {
      "epoch": 0.2169708182659154,
      "grad_norm": NaN,
      "learning_rate": 9.011655878430019e-05,
      "loss": 0.0,
      "step": 1290
    },
    {
      "epoch": 0.21713901269867966,
      "grad_norm": NaN,
      "learning_rate": 9.010051234837187e-05,
      "loss": 0.0,
      "step": 1291
    },
    {
      "epoch": 0.21730720713144394,
      "grad_norm": NaN,
      "learning_rate": 9.00844543278547e-05,
      "loss": 0.0,
      "step": 1292
    },
    {
      "epoch": 0.21747540156420822,
      "grad_norm": NaN,
      "learning_rate": 9.006838472738766e-05,
      "loss": 0.0,
      "step": 1293
    },
    {
      "epoch": 0.2176435959969725,
      "grad_norm": NaN,
      "learning_rate": 9.005230355161304e-05,
      "loss": 0.0,
      "step": 1294
    },
    {
      "epoch": 0.21781179042973678,
      "grad_norm": NaN,
      "learning_rate": 9.003621080517654e-05,
      "loss": 0.0,
      "step": 1295
    },
    {
      "epoch": 0.21797998486250106,
      "grad_norm": NaN,
      "learning_rate": 9.002010649272716e-05,
      "loss": 0.0,
      "step": 1296
    },
    {
      "epoch": 0.21814817929526534,
      "grad_norm": NaN,
      "learning_rate": 9.000399061891728e-05,
      "loss": 0.0,
      "step": 1297
    },
    {
      "epoch": 0.2183163737280296,
      "grad_norm": NaN,
      "learning_rate": 8.998786318840258e-05,
      "loss": 0.0,
      "step": 1298
    },
    {
      "epoch": 0.21848456816079387,
      "grad_norm": NaN,
      "learning_rate": 8.99717242058421e-05,
      "loss": 0.0,
      "step": 1299
    },
    {
      "epoch": 0.21865276259355815,
      "grad_norm": NaN,
      "learning_rate": 8.99555736758982e-05,
      "loss": 0.0,
      "step": 1300
    },
    {
      "epoch": 0.21882095702632243,
      "grad_norm": NaN,
      "learning_rate": 8.993941160323658e-05,
      "loss": 0.0,
      "step": 1301
    },
    {
      "epoch": 0.2189891514590867,
      "grad_norm": NaN,
      "learning_rate": 8.992323799252631e-05,
      "loss": 0.0,
      "step": 1302
    },
    {
      "epoch": 0.21915734589185099,
      "grad_norm": NaN,
      "learning_rate": 8.990705284843974e-05,
      "loss": 0.0,
      "step": 1303
    },
    {
      "epoch": 0.21932554032461526,
      "grad_norm": NaN,
      "learning_rate": 8.98908561756526e-05,
      "loss": 0.0,
      "step": 1304
    },
    {
      "epoch": 0.21949373475737954,
      "grad_norm": NaN,
      "learning_rate": 8.987464797884391e-05,
      "loss": 0.0,
      "step": 1305
    },
    {
      "epoch": 0.2196619291901438,
      "grad_norm": NaN,
      "learning_rate": 8.985842826269604e-05,
      "loss": 0.0,
      "step": 1306
    },
    {
      "epoch": 0.21983012362290807,
      "grad_norm": NaN,
      "learning_rate": 8.984219703189468e-05,
      "loss": 0.0,
      "step": 1307
    },
    {
      "epoch": 0.21999831805567235,
      "grad_norm": NaN,
      "learning_rate": 8.982595429112885e-05,
      "loss": 0.0,
      "step": 1308
    },
    {
      "epoch": 0.22016651248843663,
      "grad_norm": NaN,
      "learning_rate": 8.980970004509091e-05,
      "loss": 0.0,
      "step": 1309
    },
    {
      "epoch": 0.2203347069212009,
      "grad_norm": NaN,
      "learning_rate": 8.979343429847653e-05,
      "loss": 0.0,
      "step": 1310
    },
    {
      "epoch": 0.2205029013539652,
      "grad_norm": NaN,
      "learning_rate": 8.977715705598469e-05,
      "loss": 0.0,
      "step": 1311
    },
    {
      "epoch": 0.22067109578672947,
      "grad_norm": NaN,
      "learning_rate": 8.976086832231769e-05,
      "loss": 0.0,
      "step": 1312
    },
    {
      "epoch": 0.22083929021949372,
      "grad_norm": NaN,
      "learning_rate": 8.97445681021812e-05,
      "loss": 0.0,
      "step": 1313
    },
    {
      "epoch": 0.221007484652258,
      "grad_norm": NaN,
      "learning_rate": 8.972825640028412e-05,
      "loss": 0.0,
      "step": 1314
    },
    {
      "epoch": 0.22117567908502228,
      "grad_norm": NaN,
      "learning_rate": 8.971193322133877e-05,
      "loss": 0.0,
      "step": 1315
    },
    {
      "epoch": 0.22134387351778656,
      "grad_norm": NaN,
      "learning_rate": 8.969559857006072e-05,
      "loss": 0.0,
      "step": 1316
    },
    {
      "epoch": 0.22151206795055084,
      "grad_norm": NaN,
      "learning_rate": 8.967925245116884e-05,
      "loss": 0.0,
      "step": 1317
    },
    {
      "epoch": 0.22168026238331512,
      "grad_norm": NaN,
      "learning_rate": 8.966289486938538e-05,
      "loss": 0.0,
      "step": 1318
    },
    {
      "epoch": 0.2218484568160794,
      "grad_norm": NaN,
      "learning_rate": 8.964652582943582e-05,
      "loss": 0.0,
      "step": 1319
    },
    {
      "epoch": 0.22201665124884365,
      "grad_norm": NaN,
      "learning_rate": 8.963014533604903e-05,
      "loss": 0.0,
      "step": 1320
    },
    {
      "epoch": 0.22218484568160793,
      "grad_norm": NaN,
      "learning_rate": 8.961375339395712e-05,
      "loss": 0.0,
      "step": 1321
    },
    {
      "epoch": 0.2223530401143722,
      "grad_norm": NaN,
      "learning_rate": 8.959735000789558e-05,
      "loss": 0.0,
      "step": 1322
    },
    {
      "epoch": 0.2225212345471365,
      "grad_norm": NaN,
      "learning_rate": 8.958093518260312e-05,
      "loss": 0.0,
      "step": 1323
    },
    {
      "epoch": 0.22268942897990077,
      "grad_norm": NaN,
      "learning_rate": 8.956450892282182e-05,
      "loss": 0.0,
      "step": 1324
    },
    {
      "epoch": 0.22285762341266505,
      "grad_norm": NaN,
      "learning_rate": 8.954807123329704e-05,
      "loss": 0.0,
      "step": 1325
    },
    {
      "epoch": 0.22302581784542932,
      "grad_norm": NaN,
      "learning_rate": 8.953162211877744e-05,
      "loss": 0.0,
      "step": 1326
    },
    {
      "epoch": 0.2231940122781936,
      "grad_norm": NaN,
      "learning_rate": 8.9515161584015e-05,
      "loss": 0.0,
      "step": 1327
    },
    {
      "epoch": 0.22336220671095786,
      "grad_norm": NaN,
      "learning_rate": 8.949868963376496e-05,
      "loss": 0.0,
      "step": 1328
    },
    {
      "epoch": 0.22353040114372213,
      "grad_norm": NaN,
      "learning_rate": 8.94822062727859e-05,
      "loss": 0.0,
      "step": 1329
    },
    {
      "epoch": 0.22369859557648641,
      "grad_norm": NaN,
      "learning_rate": 8.946571150583969e-05,
      "loss": 0.0,
      "step": 1330
    },
    {
      "epoch": 0.2238667900092507,
      "grad_norm": NaN,
      "learning_rate": 8.944920533769146e-05,
      "loss": 0.0,
      "step": 1331
    },
    {
      "epoch": 0.22403498444201497,
      "grad_norm": NaN,
      "learning_rate": 8.943268777310964e-05,
      "loss": 0.0,
      "step": 1332
    },
    {
      "epoch": 0.22420317887477925,
      "grad_norm": NaN,
      "learning_rate": 8.941615881686603e-05,
      "loss": 0.0,
      "step": 1333
    },
    {
      "epoch": 0.22437137330754353,
      "grad_norm": NaN,
      "learning_rate": 8.939961847373559e-05,
      "loss": 0.0,
      "step": 1334
    },
    {
      "epoch": 0.22453956774030778,
      "grad_norm": NaN,
      "learning_rate": 8.93830667484967e-05,
      "loss": 0.0,
      "step": 1335
    },
    {
      "epoch": 0.22470776217307206,
      "grad_norm": NaN,
      "learning_rate": 8.936650364593093e-05,
      "loss": 0.0,
      "step": 1336
    },
    {
      "epoch": 0.22487595660583634,
      "grad_norm": NaN,
      "learning_rate": 8.934992917082318e-05,
      "loss": 0.0,
      "step": 1337
    },
    {
      "epoch": 0.22504415103860062,
      "grad_norm": NaN,
      "learning_rate": 8.933334332796165e-05,
      "loss": 0.0,
      "step": 1338
    },
    {
      "epoch": 0.2252123454713649,
      "grad_norm": NaN,
      "learning_rate": 8.931674612213778e-05,
      "loss": 0.0,
      "step": 1339
    },
    {
      "epoch": 0.22538053990412918,
      "grad_norm": NaN,
      "learning_rate": 8.930013755814634e-05,
      "loss": 0.0,
      "step": 1340
    },
    {
      "epoch": 0.22554873433689346,
      "grad_norm": NaN,
      "learning_rate": 8.928351764078536e-05,
      "loss": 0.0,
      "step": 1341
    },
    {
      "epoch": 0.22571692876965774,
      "grad_norm": NaN,
      "learning_rate": 8.926688637485609e-05,
      "loss": 0.0,
      "step": 1342
    },
    {
      "epoch": 0.225885123202422,
      "grad_norm": NaN,
      "learning_rate": 8.92502437651632e-05,
      "loss": 0.0,
      "step": 1343
    },
    {
      "epoch": 0.22605331763518627,
      "grad_norm": NaN,
      "learning_rate": 8.923358981651451e-05,
      "loss": 0.0,
      "step": 1344
    },
    {
      "epoch": 0.22622151206795055,
      "grad_norm": NaN,
      "learning_rate": 8.921692453372114e-05,
      "loss": 0.0,
      "step": 1345
    },
    {
      "epoch": 0.22638970650071483,
      "grad_norm": NaN,
      "learning_rate": 8.920024792159754e-05,
      "loss": 0.0,
      "step": 1346
    },
    {
      "epoch": 0.2265579009334791,
      "grad_norm": NaN,
      "learning_rate": 8.91835599849614e-05,
      "loss": 0.0,
      "step": 1347
    },
    {
      "epoch": 0.22672609536624339,
      "grad_norm": NaN,
      "learning_rate": 8.916686072863362e-05,
      "loss": 0.0,
      "step": 1348
    },
    {
      "epoch": 0.22689428979900766,
      "grad_norm": NaN,
      "learning_rate": 8.915015015743851e-05,
      "loss": 0.0,
      "step": 1349
    },
    {
      "epoch": 0.22706248423177192,
      "grad_norm": NaN,
      "learning_rate": 8.913342827620351e-05,
      "loss": 0.0,
      "step": 1350
    },
    {
      "epoch": 0.2272306786645362,
      "grad_norm": NaN,
      "learning_rate": 8.911669508975941e-05,
      "loss": 0.0,
      "step": 1351
    },
    {
      "epoch": 0.22739887309730047,
      "grad_norm": NaN,
      "learning_rate": 8.90999506029402e-05,
      "loss": 0.0,
      "step": 1352
    },
    {
      "epoch": 0.22756706753006475,
      "grad_norm": NaN,
      "learning_rate": 8.908319482058325e-05,
      "loss": 0.0,
      "step": 1353
    },
    {
      "epoch": 0.22773526196282903,
      "grad_norm": NaN,
      "learning_rate": 8.906642774752905e-05,
      "loss": 0.0,
      "step": 1354
    },
    {
      "epoch": 0.2279034563955933,
      "grad_norm": NaN,
      "learning_rate": 8.904964938862144e-05,
      "loss": 0.0,
      "step": 1355
    },
    {
      "epoch": 0.2280716508283576,
      "grad_norm": NaN,
      "learning_rate": 8.903285974870751e-05,
      "loss": 0.0,
      "step": 1356
    },
    {
      "epoch": 0.22823984526112184,
      "grad_norm": NaN,
      "learning_rate": 8.901605883263759e-05,
      "loss": 0.0,
      "step": 1357
    },
    {
      "epoch": 0.22840803969388612,
      "grad_norm": NaN,
      "learning_rate": 8.899924664526527e-05,
      "loss": 0.0,
      "step": 1358
    },
    {
      "epoch": 0.2285762341266504,
      "grad_norm": NaN,
      "learning_rate": 8.898242319144743e-05,
      "loss": 0.0,
      "step": 1359
    },
    {
      "epoch": 0.22874442855941468,
      "grad_norm": NaN,
      "learning_rate": 8.896558847604414e-05,
      "loss": 0.0,
      "step": 1360
    },
    {
      "epoch": 0.22891262299217896,
      "grad_norm": NaN,
      "learning_rate": 8.894874250391878e-05,
      "loss": 0.0,
      "step": 1361
    },
    {
      "epoch": 0.22908081742494324,
      "grad_norm": NaN,
      "learning_rate": 8.893188527993796e-05,
      "loss": 0.0,
      "step": 1362
    },
    {
      "epoch": 0.22924901185770752,
      "grad_norm": NaN,
      "learning_rate": 8.891501680897156e-05,
      "loss": 0.0,
      "step": 1363
    },
    {
      "epoch": 0.2294172062904718,
      "grad_norm": NaN,
      "learning_rate": 8.889813709589266e-05,
      "loss": 0.0,
      "step": 1364
    },
    {
      "epoch": 0.22958540072323605,
      "grad_norm": NaN,
      "learning_rate": 8.888124614557764e-05,
      "loss": 0.0,
      "step": 1365
    },
    {
      "epoch": 0.22975359515600033,
      "grad_norm": NaN,
      "learning_rate": 8.886434396290611e-05,
      "loss": 0.0,
      "step": 1366
    },
    {
      "epoch": 0.2299217895887646,
      "grad_norm": NaN,
      "learning_rate": 8.884743055276092e-05,
      "loss": 0.0,
      "step": 1367
    },
    {
      "epoch": 0.2300899840215289,
      "grad_norm": NaN,
      "learning_rate": 8.883050592002815e-05,
      "loss": 0.0,
      "step": 1368
    },
    {
      "epoch": 0.23025817845429317,
      "grad_norm": NaN,
      "learning_rate": 8.881357006959715e-05,
      "loss": 0.0,
      "step": 1369
    },
    {
      "epoch": 0.23042637288705745,
      "grad_norm": NaN,
      "learning_rate": 8.879662300636048e-05,
      "loss": 0.0,
      "step": 1370
    },
    {
      "epoch": 0.23059456731982172,
      "grad_norm": NaN,
      "learning_rate": 8.8779664735214e-05,
      "loss": 0.0,
      "step": 1371
    },
    {
      "epoch": 0.23076276175258598,
      "grad_norm": NaN,
      "learning_rate": 8.876269526105672e-05,
      "loss": 0.0,
      "step": 1372
    },
    {
      "epoch": 0.23093095618535026,
      "grad_norm": NaN,
      "learning_rate": 8.874571458879095e-05,
      "loss": 0.0,
      "step": 1373
    },
    {
      "epoch": 0.23109915061811453,
      "grad_norm": NaN,
      "learning_rate": 8.87287227233222e-05,
      "loss": 0.0,
      "step": 1374
    },
    {
      "epoch": 0.23126734505087881,
      "grad_norm": NaN,
      "learning_rate": 8.871171966955924e-05,
      "loss": 0.0,
      "step": 1375
    },
    {
      "epoch": 0.2314355394836431,
      "grad_norm": NaN,
      "learning_rate": 8.869470543241408e-05,
      "loss": 0.0,
      "step": 1376
    },
    {
      "epoch": 0.23160373391640737,
      "grad_norm": NaN,
      "learning_rate": 8.867768001680192e-05,
      "loss": 0.0,
      "step": 1377
    },
    {
      "epoch": 0.23177192834917165,
      "grad_norm": NaN,
      "learning_rate": 8.866064342764123e-05,
      "loss": 0.0,
      "step": 1378
    },
    {
      "epoch": 0.23194012278193593,
      "grad_norm": NaN,
      "learning_rate": 8.864359566985365e-05,
      "loss": 0.0,
      "step": 1379
    },
    {
      "epoch": 0.23210831721470018,
      "grad_norm": NaN,
      "learning_rate": 8.862653674836413e-05,
      "loss": 0.0,
      "step": 1380
    },
    {
      "epoch": 0.23227651164746446,
      "grad_norm": NaN,
      "learning_rate": 8.860946666810079e-05,
      "loss": 0.0,
      "step": 1381
    },
    {
      "epoch": 0.23244470608022874,
      "grad_norm": NaN,
      "learning_rate": 8.859238543399498e-05,
      "loss": 0.0,
      "step": 1382
    },
    {
      "epoch": 0.23261290051299302,
      "grad_norm": NaN,
      "learning_rate": 8.857529305098127e-05,
      "loss": 0.0,
      "step": 1383
    },
    {
      "epoch": 0.2327810949457573,
      "grad_norm": NaN,
      "learning_rate": 8.855818952399746e-05,
      "loss": 0.0,
      "step": 1384
    },
    {
      "epoch": 0.23294928937852158,
      "grad_norm": NaN,
      "learning_rate": 8.854107485798458e-05,
      "loss": 0.0,
      "step": 1385
    },
    {
      "epoch": 0.23311748381128586,
      "grad_norm": NaN,
      "learning_rate": 8.852394905788685e-05,
      "loss": 0.0,
      "step": 1386
    },
    {
      "epoch": 0.2332856782440501,
      "grad_norm": NaN,
      "learning_rate": 8.850681212865175e-05,
      "loss": 0.0,
      "step": 1387
    },
    {
      "epoch": 0.2334538726768144,
      "grad_norm": NaN,
      "learning_rate": 8.848966407522992e-05,
      "loss": 0.0,
      "step": 1388
    },
    {
      "epoch": 0.23362206710957867,
      "grad_norm": NaN,
      "learning_rate": 8.847250490257524e-05,
      "loss": 0.0,
      "step": 1389
    },
    {
      "epoch": 0.23379026154234295,
      "grad_norm": NaN,
      "learning_rate": 8.84553346156448e-05,
      "loss": 0.0,
      "step": 1390
    },
    {
      "epoch": 0.23395845597510723,
      "grad_norm": NaN,
      "learning_rate": 8.843815321939893e-05,
      "loss": 0.0,
      "step": 1391
    },
    {
      "epoch": 0.2341266504078715,
      "grad_norm": NaN,
      "learning_rate": 8.842096071880113e-05,
      "loss": 0.0,
      "step": 1392
    },
    {
      "epoch": 0.23429484484063579,
      "grad_norm": NaN,
      "learning_rate": 8.840375711881811e-05,
      "loss": 0.0,
      "step": 1393
    },
    {
      "epoch": 0.23446303927340004,
      "grad_norm": NaN,
      "learning_rate": 8.838654242441981e-05,
      "loss": 0.0,
      "step": 1394
    },
    {
      "epoch": 0.23463123370616432,
      "grad_norm": NaN,
      "learning_rate": 8.836931664057935e-05,
      "loss": 0.0,
      "step": 1395
    },
    {
      "epoch": 0.2347994281389286,
      "grad_norm": NaN,
      "learning_rate": 8.83520797722731e-05,
      "loss": 0.0,
      "step": 1396
    },
    {
      "epoch": 0.23496762257169287,
      "grad_norm": NaN,
      "learning_rate": 8.833483182448052e-05,
      "loss": 0.0,
      "step": 1397
    },
    {
      "epoch": 0.23513581700445715,
      "grad_norm": NaN,
      "learning_rate": 8.831757280218444e-05,
      "loss": 0.0,
      "step": 1398
    },
    {
      "epoch": 0.23530401143722143,
      "grad_norm": NaN,
      "learning_rate": 8.830030271037076e-05,
      "loss": 0.0,
      "step": 1399
    },
    {
      "epoch": 0.2354722058699857,
      "grad_norm": NaN,
      "learning_rate": 8.82830215540286e-05,
      "loss": 0.0,
      "step": 1400
    },
    {
      "epoch": 0.23564040030275,
      "grad_norm": NaN,
      "learning_rate": 8.826572933815032e-05,
      "loss": 0.0,
      "step": 1401
    },
    {
      "epoch": 0.23580859473551424,
      "grad_norm": NaN,
      "learning_rate": 8.824842606773142e-05,
      "loss": 0.0,
      "step": 1402
    },
    {
      "epoch": 0.23597678916827852,
      "grad_norm": NaN,
      "learning_rate": 8.823111174777064e-05,
      "loss": 0.0,
      "step": 1403
    },
    {
      "epoch": 0.2361449836010428,
      "grad_norm": NaN,
      "learning_rate": 8.82137863832699e-05,
      "loss": 0.0,
      "step": 1404
    },
    {
      "epoch": 0.23631317803380708,
      "grad_norm": NaN,
      "learning_rate": 8.819644997923427e-05,
      "loss": 0.0,
      "step": 1405
    },
    {
      "epoch": 0.23648137246657136,
      "grad_norm": NaN,
      "learning_rate": 8.817910254067209e-05,
      "loss": 0.0,
      "step": 1406
    },
    {
      "epoch": 0.23664956689933564,
      "grad_norm": NaN,
      "learning_rate": 8.816174407259481e-05,
      "loss": 0.0,
      "step": 1407
    },
    {
      "epoch": 0.23681776133209992,
      "grad_norm": NaN,
      "learning_rate": 8.81443745800171e-05,
      "loss": 0.0,
      "step": 1408
    },
    {
      "epoch": 0.23698595576486417,
      "grad_norm": NaN,
      "learning_rate": 8.812699406795682e-05,
      "loss": 0.0,
      "step": 1409
    },
    {
      "epoch": 0.23715415019762845,
      "grad_norm": NaN,
      "learning_rate": 8.810960254143499e-05,
      "loss": 0.0,
      "step": 1410
    },
    {
      "epoch": 0.23732234463039273,
      "grad_norm": NaN,
      "learning_rate": 8.809220000547584e-05,
      "loss": 0.0,
      "step": 1411
    },
    {
      "epoch": 0.237490539063157,
      "grad_norm": NaN,
      "learning_rate": 8.807478646510677e-05,
      "loss": 0.0,
      "step": 1412
    },
    {
      "epoch": 0.2376587334959213,
      "grad_norm": NaN,
      "learning_rate": 8.805736192535835e-05,
      "loss": 0.0,
      "step": 1413
    },
    {
      "epoch": 0.23782692792868557,
      "grad_norm": NaN,
      "learning_rate": 8.803992639126433e-05,
      "loss": 0.0,
      "step": 1414
    },
    {
      "epoch": 0.23799512236144985,
      "grad_norm": NaN,
      "learning_rate": 8.802247986786167e-05,
      "loss": 0.0,
      "step": 1415
    },
    {
      "epoch": 0.23816331679421412,
      "grad_norm": NaN,
      "learning_rate": 8.800502236019044e-05,
      "loss": 0.0,
      "step": 1416
    },
    {
      "epoch": 0.23833151122697838,
      "grad_norm": NaN,
      "learning_rate": 8.798755387329392e-05,
      "loss": 0.0,
      "step": 1417
    },
    {
      "epoch": 0.23849970565974266,
      "grad_norm": NaN,
      "learning_rate": 8.797007441221859e-05,
      "loss": 0.0,
      "step": 1418
    },
    {
      "epoch": 0.23866790009250693,
      "grad_norm": NaN,
      "learning_rate": 8.795258398201403e-05,
      "loss": 0.0,
      "step": 1419
    },
    {
      "epoch": 0.23883609452527121,
      "grad_norm": NaN,
      "learning_rate": 8.793508258773307e-05,
      "loss": 0.0,
      "step": 1420
    },
    {
      "epoch": 0.2390042889580355,
      "grad_norm": NaN,
      "learning_rate": 8.791757023443163e-05,
      "loss": 0.0,
      "step": 1421
    },
    {
      "epoch": 0.23917248339079977,
      "grad_norm": NaN,
      "learning_rate": 8.790004692716887e-05,
      "loss": 0.0,
      "step": 1422
    },
    {
      "epoch": 0.23934067782356405,
      "grad_norm": NaN,
      "learning_rate": 8.788251267100704e-05,
      "loss": 0.0,
      "step": 1423
    },
    {
      "epoch": 0.2395088722563283,
      "grad_norm": NaN,
      "learning_rate": 8.786496747101163e-05,
      "loss": 0.0,
      "step": 1424
    },
    {
      "epoch": 0.23967706668909258,
      "grad_norm": NaN,
      "learning_rate": 8.78474113322512e-05,
      "loss": 0.0,
      "step": 1425
    },
    {
      "epoch": 0.23984526112185686,
      "grad_norm": NaN,
      "learning_rate": 8.782984425979757e-05,
      "loss": 0.0,
      "step": 1426
    },
    {
      "epoch": 0.24001345555462114,
      "grad_norm": NaN,
      "learning_rate": 8.781226625872564e-05,
      "loss": 0.0,
      "step": 1427
    },
    {
      "epoch": 0.24018164998738542,
      "grad_norm": NaN,
      "learning_rate": 8.779467733411353e-05,
      "loss": 0.0,
      "step": 1428
    },
    {
      "epoch": 0.2403498444201497,
      "grad_norm": NaN,
      "learning_rate": 8.777707749104244e-05,
      "loss": 0.0,
      "step": 1429
    },
    {
      "epoch": 0.24051803885291398,
      "grad_norm": NaN,
      "learning_rate": 8.775946673459681e-05,
      "loss": 0.0,
      "step": 1430
    },
    {
      "epoch": 0.24068623328567823,
      "grad_norm": NaN,
      "learning_rate": 8.774184506986416e-05,
      "loss": 0.0,
      "step": 1431
    },
    {
      "epoch": 0.2408544277184425,
      "grad_norm": NaN,
      "learning_rate": 8.772421250193521e-05,
      "loss": 0.0,
      "step": 1432
    },
    {
      "epoch": 0.2410226221512068,
      "grad_norm": NaN,
      "learning_rate": 8.77065690359038e-05,
      "loss": 0.0,
      "step": 1433
    },
    {
      "epoch": 0.24119081658397107,
      "grad_norm": NaN,
      "learning_rate": 8.768891467686693e-05,
      "loss": 0.0,
      "step": 1434
    },
    {
      "epoch": 0.24135901101673535,
      "grad_norm": NaN,
      "learning_rate": 8.767124942992475e-05,
      "loss": 0.0,
      "step": 1435
    },
    {
      "epoch": 0.24152720544949963,
      "grad_norm": NaN,
      "learning_rate": 8.765357330018056e-05,
      "loss": 0.0,
      "step": 1436
    },
    {
      "epoch": 0.2416953998822639,
      "grad_norm": NaN,
      "learning_rate": 8.763588629274077e-05,
      "loss": 0.0,
      "step": 1437
    },
    {
      "epoch": 0.24186359431502819,
      "grad_norm": NaN,
      "learning_rate": 8.761818841271498e-05,
      "loss": 0.0,
      "step": 1438
    },
    {
      "epoch": 0.24203178874779244,
      "grad_norm": NaN,
      "learning_rate": 8.760047966521591e-05,
      "loss": 0.0,
      "step": 1439
    },
    {
      "epoch": 0.24219998318055672,
      "grad_norm": NaN,
      "learning_rate": 8.758276005535943e-05,
      "loss": 0.0,
      "step": 1440
    },
    {
      "epoch": 0.242368177613321,
      "grad_norm": NaN,
      "learning_rate": 8.75650295882645e-05,
      "loss": 0.0,
      "step": 1441
    },
    {
      "epoch": 0.24253637204608527,
      "grad_norm": NaN,
      "learning_rate": 8.75472882690533e-05,
      "loss": 0.0,
      "step": 1442
    },
    {
      "epoch": 0.24270456647884955,
      "grad_norm": NaN,
      "learning_rate": 8.752953610285104e-05,
      "loss": 0.0,
      "step": 1443
    },
    {
      "epoch": 0.24287276091161383,
      "grad_norm": NaN,
      "learning_rate": 8.751177309478618e-05,
      "loss": 0.0,
      "step": 1444
    },
    {
      "epoch": 0.2430409553443781,
      "grad_norm": NaN,
      "learning_rate": 8.74939992499902e-05,
      "loss": 0.0,
      "step": 1445
    },
    {
      "epoch": 0.24320914977714236,
      "grad_norm": NaN,
      "learning_rate": 8.747621457359782e-05,
      "loss": 0.0,
      "step": 1446
    },
    {
      "epoch": 0.24337734420990664,
      "grad_norm": NaN,
      "learning_rate": 8.745841907074682e-05,
      "loss": 0.0,
      "step": 1447
    },
    {
      "epoch": 0.24354553864267092,
      "grad_norm": NaN,
      "learning_rate": 8.744061274657806e-05,
      "loss": 0.0,
      "step": 1448
    },
    {
      "epoch": 0.2437137330754352,
      "grad_norm": NaN,
      "learning_rate": 8.742279560623567e-05,
      "loss": 0.0,
      "step": 1449
    },
    {
      "epoch": 0.24388192750819948,
      "grad_norm": NaN,
      "learning_rate": 8.740496765486679e-05,
      "loss": 0.0,
      "step": 1450
    },
    {
      "epoch": 0.24405012194096376,
      "grad_norm": NaN,
      "learning_rate": 8.738712889762171e-05,
      "loss": 0.0,
      "step": 1451
    },
    {
      "epoch": 0.24421831637372804,
      "grad_norm": NaN,
      "learning_rate": 8.736927933965385e-05,
      "loss": 0.0,
      "step": 1452
    },
    {
      "epoch": 0.24438651080649232,
      "grad_norm": NaN,
      "learning_rate": 8.735141898611974e-05,
      "loss": 0.0,
      "step": 1453
    },
    {
      "epoch": 0.24455470523925657,
      "grad_norm": NaN,
      "learning_rate": 8.733354784217905e-05,
      "loss": 0.0,
      "step": 1454
    },
    {
      "epoch": 0.24472289967202085,
      "grad_norm": NaN,
      "learning_rate": 8.731566591299455e-05,
      "loss": 0.0,
      "step": 1455
    },
    {
      "epoch": 0.24489109410478513,
      "grad_norm": NaN,
      "learning_rate": 8.729777320373213e-05,
      "loss": 0.0,
      "step": 1456
    },
    {
      "epoch": 0.2450592885375494,
      "grad_norm": NaN,
      "learning_rate": 8.727986971956079e-05,
      "loss": 0.0,
      "step": 1457
    },
    {
      "epoch": 0.2452274829703137,
      "grad_norm": NaN,
      "learning_rate": 8.726195546565263e-05,
      "loss": 0.0,
      "step": 1458
    },
    {
      "epoch": 0.24539567740307797,
      "grad_norm": NaN,
      "learning_rate": 8.724403044718292e-05,
      "loss": 0.0,
      "step": 1459
    },
    {
      "epoch": 0.24556387183584225,
      "grad_norm": NaN,
      "learning_rate": 8.722609466932996e-05,
      "loss": 0.0,
      "step": 1460
    },
    {
      "epoch": 0.2457320662686065,
      "grad_norm": NaN,
      "learning_rate": 8.72081481372752e-05,
      "loss": 0.0,
      "step": 1461
    },
    {
      "epoch": 0.24590026070137078,
      "grad_norm": NaN,
      "learning_rate": 8.719019085620321e-05,
      "loss": 0.0,
      "step": 1462
    },
    {
      "epoch": 0.24606845513413506,
      "grad_norm": NaN,
      "learning_rate": 8.717222283130163e-05,
      "loss": 0.0,
      "step": 1463
    },
    {
      "epoch": 0.24623664956689933,
      "grad_norm": NaN,
      "learning_rate": 8.715424406776123e-05,
      "loss": 0.0,
      "step": 1464
    },
    {
      "epoch": 0.24640484399966361,
      "grad_norm": NaN,
      "learning_rate": 8.713625457077585e-05,
      "loss": 0.0,
      "step": 1465
    },
    {
      "epoch": 0.2465730384324279,
      "grad_norm": NaN,
      "learning_rate": 8.71182543455425e-05,
      "loss": 0.0,
      "step": 1466
    },
    {
      "epoch": 0.24674123286519217,
      "grad_norm": NaN,
      "learning_rate": 8.710024339726121e-05,
      "loss": 0.0,
      "step": 1467
    },
    {
      "epoch": 0.24690942729795642,
      "grad_norm": NaN,
      "learning_rate": 8.708222173113514e-05,
      "loss": 0.0,
      "step": 1468
    },
    {
      "epoch": 0.2470776217307207,
      "grad_norm": NaN,
      "learning_rate": 8.706418935237057e-05,
      "loss": 0.0,
      "step": 1469
    },
    {
      "epoch": 0.24724581616348498,
      "grad_norm": NaN,
      "learning_rate": 8.704614626617683e-05,
      "loss": 0.0,
      "step": 1470
    },
    {
      "epoch": 0.24741401059624926,
      "grad_norm": NaN,
      "learning_rate": 8.702809247776638e-05,
      "loss": 0.0,
      "step": 1471
    },
    {
      "epoch": 0.24758220502901354,
      "grad_norm": NaN,
      "learning_rate": 8.701002799235475e-05,
      "loss": 0.0,
      "step": 1472
    },
    {
      "epoch": 0.24775039946177782,
      "grad_norm": NaN,
      "learning_rate": 8.699195281516056e-05,
      "loss": 0.0,
      "step": 1473
    },
    {
      "epoch": 0.2479185938945421,
      "grad_norm": NaN,
      "learning_rate": 8.697386695140554e-05,
      "loss": 0.0,
      "step": 1474
    },
    {
      "epoch": 0.24808678832730638,
      "grad_norm": NaN,
      "learning_rate": 8.69557704063145e-05,
      "loss": 0.0,
      "step": 1475
    },
    {
      "epoch": 0.24825498276007063,
      "grad_norm": NaN,
      "learning_rate": 8.693766318511532e-05,
      "loss": 0.0,
      "step": 1476
    },
    {
      "epoch": 0.2484231771928349,
      "grad_norm": NaN,
      "learning_rate": 8.691954529303895e-05,
      "loss": 0.0,
      "step": 1477
    },
    {
      "epoch": 0.2485913716255992,
      "grad_norm": NaN,
      "learning_rate": 8.690141673531948e-05,
      "loss": 0.0,
      "step": 1478
    },
    {
      "epoch": 0.24875956605836347,
      "grad_norm": NaN,
      "learning_rate": 8.688327751719403e-05,
      "loss": 0.0,
      "step": 1479
    },
    {
      "epoch": 0.24892776049112775,
      "grad_norm": NaN,
      "learning_rate": 8.686512764390283e-05,
      "loss": 0.0,
      "step": 1480
    },
    {
      "epoch": 0.24909595492389203,
      "grad_norm": NaN,
      "learning_rate": 8.684696712068916e-05,
      "loss": 0.0,
      "step": 1481
    },
    {
      "epoch": 0.2492641493566563,
      "grad_norm": NaN,
      "learning_rate": 8.68287959527994e-05,
      "loss": 0.0,
      "step": 1482
    },
    {
      "epoch": 0.24943234378942056,
      "grad_norm": NaN,
      "learning_rate": 8.6810614145483e-05,
      "loss": 0.0,
      "step": 1483
    },
    {
      "epoch": 0.24960053822218484,
      "grad_norm": NaN,
      "learning_rate": 8.679242170399246e-05,
      "loss": 0.0,
      "step": 1484
    },
    {
      "epoch": 0.24976873265494912,
      "grad_norm": NaN,
      "learning_rate": 8.677421863358341e-05,
      "loss": 0.0,
      "step": 1485
    },
    {
      "epoch": 0.2499369270877134,
      "grad_norm": NaN,
      "learning_rate": 8.675600493951448e-05,
      "loss": 0.0,
      "step": 1486
    },
    {
      "epoch": 0.25010512152047765,
      "grad_norm": NaN,
      "learning_rate": 8.673778062704741e-05,
      "loss": 0.0,
      "step": 1487
    },
    {
      "epoch": 0.2502733159532419,
      "grad_norm": NaN,
      "learning_rate": 8.671954570144701e-05,
      "loss": 0.0,
      "step": 1488
    },
    {
      "epoch": 0.2504415103860062,
      "grad_norm": NaN,
      "learning_rate": 8.670130016798115e-05,
      "loss": 0.0,
      "step": 1489
    },
    {
      "epoch": 0.2506097048187705,
      "grad_norm": NaN,
      "learning_rate": 8.668304403192074e-05,
      "loss": 0.0,
      "step": 1490
    },
    {
      "epoch": 0.25077789925153476,
      "grad_norm": NaN,
      "learning_rate": 8.666477729853977e-05,
      "loss": 0.0,
      "step": 1491
    },
    {
      "epoch": 0.25094609368429904,
      "grad_norm": NaN,
      "learning_rate": 8.664649997311531e-05,
      "loss": 0.0,
      "step": 1492
    },
    {
      "epoch": 0.2511142881170633,
      "grad_norm": NaN,
      "learning_rate": 8.662821206092748e-05,
      "loss": 0.0,
      "step": 1493
    },
    {
      "epoch": 0.2512824825498276,
      "grad_norm": NaN,
      "learning_rate": 8.660991356725946e-05,
      "loss": 0.0,
      "step": 1494
    },
    {
      "epoch": 0.2514506769825919,
      "grad_norm": NaN,
      "learning_rate": 8.659160449739745e-05,
      "loss": 0.0,
      "step": 1495
    },
    {
      "epoch": 0.25161887141535616,
      "grad_norm": NaN,
      "learning_rate": 8.657328485663074e-05,
      "loss": 0.0,
      "step": 1496
    },
    {
      "epoch": 0.25178706584812044,
      "grad_norm": NaN,
      "learning_rate": 8.65549546502517e-05,
      "loss": 0.0,
      "step": 1497
    },
    {
      "epoch": 0.2519552602808847,
      "grad_norm": NaN,
      "learning_rate": 8.653661388355568e-05,
      "loss": 0.0,
      "step": 1498
    },
    {
      "epoch": 0.252123454713649,
      "grad_norm": NaN,
      "learning_rate": 8.651826256184117e-05,
      "loss": 0.0,
      "step": 1499
    },
    {
      "epoch": 0.2522916491464133,
      "grad_norm": NaN,
      "learning_rate": 8.649990069040961e-05,
      "loss": 0.0,
      "step": 1500
    },
    {
      "epoch": 0.25245984357917756,
      "grad_norm": NaN,
      "learning_rate": 8.648152827456557e-05,
      "loss": 0.0,
      "step": 1501
    },
    {
      "epoch": 0.2526280380119418,
      "grad_norm": NaN,
      "learning_rate": 8.646314531961663e-05,
      "loss": 0.0,
      "step": 1502
    },
    {
      "epoch": 0.25279623244470606,
      "grad_norm": NaN,
      "learning_rate": 8.644475183087341e-05,
      "loss": 0.0,
      "step": 1503
    },
    {
      "epoch": 0.25296442687747034,
      "grad_norm": NaN,
      "learning_rate": 8.64263478136496e-05,
      "loss": 0.0,
      "step": 1504
    },
    {
      "epoch": 0.2531326213102346,
      "grad_norm": NaN,
      "learning_rate": 8.64079332732619e-05,
      "loss": 0.0,
      "step": 1505
    },
    {
      "epoch": 0.2533008157429989,
      "grad_norm": NaN,
      "learning_rate": 8.638950821503007e-05,
      "loss": 0.0,
      "step": 1506
    },
    {
      "epoch": 0.2534690101757632,
      "grad_norm": NaN,
      "learning_rate": 8.63710726442769e-05,
      "loss": 0.0,
      "step": 1507
    },
    {
      "epoch": 0.25363720460852746,
      "grad_norm": NaN,
      "learning_rate": 8.635262656632823e-05,
      "loss": 0.0,
      "step": 1508
    },
    {
      "epoch": 0.25380539904129173,
      "grad_norm": NaN,
      "learning_rate": 8.633416998651292e-05,
      "loss": 0.0,
      "step": 1509
    },
    {
      "epoch": 0.253973593474056,
      "grad_norm": NaN,
      "learning_rate": 8.631570291016284e-05,
      "loss": 0.0,
      "step": 1510
    },
    {
      "epoch": 0.2541417879068203,
      "grad_norm": NaN,
      "learning_rate": 8.629722534261298e-05,
      "loss": 0.0,
      "step": 1511
    },
    {
      "epoch": 0.2543099823395846,
      "grad_norm": NaN,
      "learning_rate": 8.627873728920127e-05,
      "loss": 0.0,
      "step": 1512
    },
    {
      "epoch": 0.25447817677234885,
      "grad_norm": NaN,
      "learning_rate": 8.626023875526868e-05,
      "loss": 0.0,
      "step": 1513
    },
    {
      "epoch": 0.25464637120511313,
      "grad_norm": NaN,
      "learning_rate": 8.624172974615926e-05,
      "loss": 0.0,
      "step": 1514
    },
    {
      "epoch": 0.2548145656378774,
      "grad_norm": NaN,
      "learning_rate": 8.622321026722005e-05,
      "loss": 0.0,
      "step": 1515
    },
    {
      "epoch": 0.25498276007064163,
      "grad_norm": NaN,
      "learning_rate": 8.620468032380111e-05,
      "loss": 0.0,
      "step": 1516
    },
    {
      "epoch": 0.2551509545034059,
      "grad_norm": NaN,
      "learning_rate": 8.618613992125556e-05,
      "loss": 0.0,
      "step": 1517
    },
    {
      "epoch": 0.2553191489361702,
      "grad_norm": NaN,
      "learning_rate": 8.616758906493948e-05,
      "loss": 0.0,
      "step": 1518
    },
    {
      "epoch": 0.25548734336893447,
      "grad_norm": NaN,
      "learning_rate": 8.614902776021202e-05,
      "loss": 0.0,
      "step": 1519
    },
    {
      "epoch": 0.25565553780169875,
      "grad_norm": NaN,
      "learning_rate": 8.613045601243534e-05,
      "loss": 0.0,
      "step": 1520
    },
    {
      "epoch": 0.25582373223446303,
      "grad_norm": NaN,
      "learning_rate": 8.611187382697458e-05,
      "loss": 0.0,
      "step": 1521
    },
    {
      "epoch": 0.2559919266672273,
      "grad_norm": NaN,
      "learning_rate": 8.609328120919797e-05,
      "loss": 0.0,
      "step": 1522
    },
    {
      "epoch": 0.2561601210999916,
      "grad_norm": NaN,
      "learning_rate": 8.607467816447669e-05,
      "loss": 0.0,
      "step": 1523
    },
    {
      "epoch": 0.25632831553275587,
      "grad_norm": NaN,
      "learning_rate": 8.605606469818495e-05,
      "loss": 0.0,
      "step": 1524
    },
    {
      "epoch": 0.25649650996552015,
      "grad_norm": NaN,
      "learning_rate": 8.603744081569994e-05,
      "loss": 0.0,
      "step": 1525
    },
    {
      "epoch": 0.2566647043982844,
      "grad_norm": NaN,
      "learning_rate": 8.601880652240195e-05,
      "loss": 0.0,
      "step": 1526
    },
    {
      "epoch": 0.2568328988310487,
      "grad_norm": NaN,
      "learning_rate": 8.600016182367417e-05,
      "loss": 0.0,
      "step": 1527
    },
    {
      "epoch": 0.257001093263813,
      "grad_norm": NaN,
      "learning_rate": 8.598150672490288e-05,
      "loss": 0.0,
      "step": 1528
    },
    {
      "epoch": 0.25716928769657726,
      "grad_norm": NaN,
      "learning_rate": 8.596284123147731e-05,
      "loss": 0.0,
      "step": 1529
    },
    {
      "epoch": 0.25733748212934154,
      "grad_norm": NaN,
      "learning_rate": 8.59441653487897e-05,
      "loss": 0.0,
      "step": 1530
    },
    {
      "epoch": 0.25750567656210577,
      "grad_norm": NaN,
      "learning_rate": 8.592547908223532e-05,
      "loss": 0.0,
      "step": 1531
    },
    {
      "epoch": 0.25767387099487005,
      "grad_norm": NaN,
      "learning_rate": 8.590678243721245e-05,
      "loss": 0.0,
      "step": 1532
    },
    {
      "epoch": 0.2578420654276343,
      "grad_norm": NaN,
      "learning_rate": 8.58880754191223e-05,
      "loss": 0.0,
      "step": 1533
    },
    {
      "epoch": 0.2580102598603986,
      "grad_norm": NaN,
      "learning_rate": 8.586935803336911e-05,
      "loss": 0.0,
      "step": 1534
    },
    {
      "epoch": 0.2581784542931629,
      "grad_norm": NaN,
      "learning_rate": 8.585063028536016e-05,
      "loss": 0.0,
      "step": 1535
    },
    {
      "epoch": 0.25834664872592716,
      "grad_norm": NaN,
      "learning_rate": 8.583189218050567e-05,
      "loss": 0.0,
      "step": 1536
    },
    {
      "epoch": 0.25851484315869144,
      "grad_norm": NaN,
      "learning_rate": 8.581314372421887e-05,
      "loss": 0.0,
      "step": 1537
    },
    {
      "epoch": 0.2586830375914557,
      "grad_norm": NaN,
      "learning_rate": 8.579438492191597e-05,
      "loss": 0.0,
      "step": 1538
    },
    {
      "epoch": 0.25885123202422,
      "grad_norm": NaN,
      "learning_rate": 8.57756157790162e-05,
      "loss": 0.0,
      "step": 1539
    },
    {
      "epoch": 0.2590194264569843,
      "grad_norm": NaN,
      "learning_rate": 8.575683630094173e-05,
      "loss": 0.0,
      "step": 1540
    },
    {
      "epoch": 0.25918762088974856,
      "grad_norm": NaN,
      "learning_rate": 8.573804649311776e-05,
      "loss": 0.0,
      "step": 1541
    },
    {
      "epoch": 0.25935581532251284,
      "grad_norm": NaN,
      "learning_rate": 8.571924636097243e-05,
      "loss": 0.0,
      "step": 1542
    },
    {
      "epoch": 0.2595240097552771,
      "grad_norm": NaN,
      "learning_rate": 8.570043590993693e-05,
      "loss": 0.0,
      "step": 1543
    },
    {
      "epoch": 0.2596922041880414,
      "grad_norm": NaN,
      "learning_rate": 8.568161514544535e-05,
      "loss": 0.0,
      "step": 1544
    },
    {
      "epoch": 0.2598603986208057,
      "grad_norm": NaN,
      "learning_rate": 8.566278407293481e-05,
      "loss": 0.0,
      "step": 1545
    },
    {
      "epoch": 0.2600285930535699,
      "grad_norm": NaN,
      "learning_rate": 8.56439426978454e-05,
      "loss": 0.0,
      "step": 1546
    },
    {
      "epoch": 0.2601967874863342,
      "grad_norm": NaN,
      "learning_rate": 8.562509102562018e-05,
      "loss": 0.0,
      "step": 1547
    },
    {
      "epoch": 0.26036498191909846,
      "grad_norm": NaN,
      "learning_rate": 8.560622906170518e-05,
      "loss": 0.0,
      "step": 1548
    },
    {
      "epoch": 0.26053317635186274,
      "grad_norm": NaN,
      "learning_rate": 8.558735681154943e-05,
      "loss": 0.0,
      "step": 1549
    },
    {
      "epoch": 0.260701370784627,
      "grad_norm": NaN,
      "learning_rate": 8.55684742806049e-05,
      "loss": 0.0,
      "step": 1550
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": NaN,
      "learning_rate": 8.554958147432652e-05,
      "loss": 0.0,
      "step": 1551
    },
    {
      "epoch": 0.2610377596501556,
      "grad_norm": NaN,
      "learning_rate": 8.553067839817225e-05,
      "loss": 0.0,
      "step": 1552
    },
    {
      "epoch": 0.26120595408291986,
      "grad_norm": NaN,
      "learning_rate": 8.551176505760294e-05,
      "loss": 0.0,
      "step": 1553
    },
    {
      "epoch": 0.26137414851568413,
      "grad_norm": NaN,
      "learning_rate": 8.549284145808248e-05,
      "loss": 0.0,
      "step": 1554
    },
    {
      "epoch": 0.2615423429484484,
      "grad_norm": NaN,
      "learning_rate": 8.547390760507765e-05,
      "loss": 0.0,
      "step": 1555
    },
    {
      "epoch": 0.2617105373812127,
      "grad_norm": NaN,
      "learning_rate": 8.545496350405825e-05,
      "loss": 0.0,
      "step": 1556
    },
    {
      "epoch": 0.261878731813977,
      "grad_norm": NaN,
      "learning_rate": 8.543600916049701e-05,
      "loss": 0.0,
      "step": 1557
    },
    {
      "epoch": 0.26204692624674125,
      "grad_norm": NaN,
      "learning_rate": 8.541704457986963e-05,
      "loss": 0.0,
      "step": 1558
    },
    {
      "epoch": 0.26221512067950553,
      "grad_norm": NaN,
      "learning_rate": 8.539806976765476e-05,
      "loss": 0.0,
      "step": 1559
    },
    {
      "epoch": 0.2623833151122698,
      "grad_norm": NaN,
      "learning_rate": 8.537908472933402e-05,
      "loss": 0.0,
      "step": 1560
    },
    {
      "epoch": 0.26255150954503403,
      "grad_norm": NaN,
      "learning_rate": 8.536008947039197e-05,
      "loss": 0.0,
      "step": 1561
    },
    {
      "epoch": 0.2627197039777983,
      "grad_norm": NaN,
      "learning_rate": 8.534108399631612e-05,
      "loss": 0.0,
      "step": 1562
    },
    {
      "epoch": 0.2628878984105626,
      "grad_norm": NaN,
      "learning_rate": 8.532206831259696e-05,
      "loss": 0.0,
      "step": 1563
    },
    {
      "epoch": 0.26305609284332687,
      "grad_norm": NaN,
      "learning_rate": 8.530304242472788e-05,
      "loss": 0.0,
      "step": 1564
    },
    {
      "epoch": 0.26322428727609115,
      "grad_norm": NaN,
      "learning_rate": 8.528400633820528e-05,
      "loss": 0.0,
      "step": 1565
    },
    {
      "epoch": 0.26339248170885543,
      "grad_norm": NaN,
      "learning_rate": 8.526496005852845e-05,
      "loss": 0.0,
      "step": 1566
    },
    {
      "epoch": 0.2635606761416197,
      "grad_norm": NaN,
      "learning_rate": 8.524590359119961e-05,
      "loss": 0.0,
      "step": 1567
    },
    {
      "epoch": 0.263728870574384,
      "grad_norm": NaN,
      "learning_rate": 8.522683694172404e-05,
      "loss": 0.0,
      "step": 1568
    },
    {
      "epoch": 0.26389706500714827,
      "grad_norm": NaN,
      "learning_rate": 8.520776011560983e-05,
      "loss": 0.0,
      "step": 1569
    },
    {
      "epoch": 0.26406525943991255,
      "grad_norm": NaN,
      "learning_rate": 8.518867311836808e-05,
      "loss": 0.0,
      "step": 1570
    },
    {
      "epoch": 0.2642334538726768,
      "grad_norm": NaN,
      "learning_rate": 8.516957595551279e-05,
      "loss": 0.0,
      "step": 1571
    },
    {
      "epoch": 0.2644016483054411,
      "grad_norm": NaN,
      "learning_rate": 8.515046863256094e-05,
      "loss": 0.0,
      "step": 1572
    },
    {
      "epoch": 0.2645698427382054,
      "grad_norm": NaN,
      "learning_rate": 8.513135115503239e-05,
      "loss": 0.0,
      "step": 1573
    },
    {
      "epoch": 0.26473803717096966,
      "grad_norm": NaN,
      "learning_rate": 8.511222352845e-05,
      "loss": 0.0,
      "step": 1574
    },
    {
      "epoch": 0.26490623160373394,
      "grad_norm": NaN,
      "learning_rate": 8.50930857583395e-05,
      "loss": 0.0,
      "step": 1575
    },
    {
      "epoch": 0.26507442603649817,
      "grad_norm": NaN,
      "learning_rate": 8.507393785022961e-05,
      "loss": 0.0,
      "step": 1576
    },
    {
      "epoch": 0.26524262046926245,
      "grad_norm": NaN,
      "learning_rate": 8.505477980965191e-05,
      "loss": 0.0,
      "step": 1577
    },
    {
      "epoch": 0.2654108149020267,
      "grad_norm": NaN,
      "learning_rate": 8.503561164214095e-05,
      "loss": 0.0,
      "step": 1578
    },
    {
      "epoch": 0.265579009334791,
      "grad_norm": NaN,
      "learning_rate": 8.501643335323424e-05,
      "loss": 0.0,
      "step": 1579
    },
    {
      "epoch": 0.2657472037675553,
      "grad_norm": NaN,
      "learning_rate": 8.499724494847212e-05,
      "loss": 0.0,
      "step": 1580
    },
    {
      "epoch": 0.26591539820031956,
      "grad_norm": NaN,
      "learning_rate": 8.497804643339792e-05,
      "loss": 0.0,
      "step": 1581
    },
    {
      "epoch": 0.26608359263308384,
      "grad_norm": NaN,
      "learning_rate": 8.495883781355789e-05,
      "loss": 0.0,
      "step": 1582
    },
    {
      "epoch": 0.2662517870658481,
      "grad_norm": NaN,
      "learning_rate": 8.493961909450118e-05,
      "loss": 0.0,
      "step": 1583
    },
    {
      "epoch": 0.2664199814986124,
      "grad_norm": NaN,
      "learning_rate": 8.492039028177986e-05,
      "loss": 0.0,
      "step": 1584
    },
    {
      "epoch": 0.2665881759313767,
      "grad_norm": NaN,
      "learning_rate": 8.490115138094889e-05,
      "loss": 0.0,
      "step": 1585
    },
    {
      "epoch": 0.26675637036414096,
      "grad_norm": NaN,
      "learning_rate": 8.488190239756623e-05,
      "loss": 0.0,
      "step": 1586
    },
    {
      "epoch": 0.26692456479690524,
      "grad_norm": NaN,
      "learning_rate": 8.486264333719267e-05,
      "loss": 0.0,
      "step": 1587
    },
    {
      "epoch": 0.2670927592296695,
      "grad_norm": NaN,
      "learning_rate": 8.484337420539193e-05,
      "loss": 0.0,
      "step": 1588
    },
    {
      "epoch": 0.2672609536624338,
      "grad_norm": NaN,
      "learning_rate": 8.482409500773066e-05,
      "loss": 0.0,
      "step": 1589
    },
    {
      "epoch": 0.267429148095198,
      "grad_norm": NaN,
      "learning_rate": 8.480480574977836e-05,
      "loss": 0.0,
      "step": 1590
    },
    {
      "epoch": 0.2675973425279623,
      "grad_norm": NaN,
      "learning_rate": 8.478550643710754e-05,
      "loss": 0.0,
      "step": 1591
    },
    {
      "epoch": 0.2677655369607266,
      "grad_norm": NaN,
      "learning_rate": 8.476619707529354e-05,
      "loss": 0.0,
      "step": 1592
    },
    {
      "epoch": 0.26793373139349086,
      "grad_norm": NaN,
      "learning_rate": 8.474687766991459e-05,
      "loss": 0.0,
      "step": 1593
    },
    {
      "epoch": 0.26810192582625514,
      "grad_norm": NaN,
      "learning_rate": 8.472754822655187e-05,
      "loss": 0.0,
      "step": 1594
    },
    {
      "epoch": 0.2682701202590194,
      "grad_norm": NaN,
      "learning_rate": 8.470820875078944e-05,
      "loss": 0.0,
      "step": 1595
    },
    {
      "epoch": 0.2684383146917837,
      "grad_norm": NaN,
      "learning_rate": 8.468885924821426e-05,
      "loss": 0.0,
      "step": 1596
    },
    {
      "epoch": 0.268606509124548,
      "grad_norm": NaN,
      "learning_rate": 8.466949972441618e-05,
      "loss": 0.0,
      "step": 1597
    },
    {
      "epoch": 0.26877470355731226,
      "grad_norm": NaN,
      "learning_rate": 8.465013018498795e-05,
      "loss": 0.0,
      "step": 1598
    },
    {
      "epoch": 0.26894289799007653,
      "grad_norm": NaN,
      "learning_rate": 8.463075063552521e-05,
      "loss": 0.0,
      "step": 1599
    },
    {
      "epoch": 0.2691110924228408,
      "grad_norm": NaN,
      "learning_rate": 8.46113610816265e-05,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 0.2692792868556051,
      "grad_norm": NaN,
      "learning_rate": 8.459196152889325e-05,
      "loss": 0.0,
      "step": 1601
    },
    {
      "epoch": 0.2694474812883694,
      "grad_norm": NaN,
      "learning_rate": 8.457255198292977e-05,
      "loss": 0.0,
      "step": 1602
    },
    {
      "epoch": 0.26961567572113365,
      "grad_norm": NaN,
      "learning_rate": 8.455313244934324e-05,
      "loss": 0.0,
      "step": 1603
    },
    {
      "epoch": 0.26978387015389793,
      "grad_norm": NaN,
      "learning_rate": 8.453370293374376e-05,
      "loss": 0.0,
      "step": 1604
    },
    {
      "epoch": 0.26995206458666215,
      "grad_norm": NaN,
      "learning_rate": 8.451426344174433e-05,
      "loss": 0.0,
      "step": 1605
    },
    {
      "epoch": 0.27012025901942643,
      "grad_norm": NaN,
      "learning_rate": 8.449481397896075e-05,
      "loss": 0.0,
      "step": 1606
    },
    {
      "epoch": 0.2702884534521907,
      "grad_norm": NaN,
      "learning_rate": 8.44753545510118e-05,
      "loss": 0.0,
      "step": 1607
    },
    {
      "epoch": 0.270456647884955,
      "grad_norm": NaN,
      "learning_rate": 8.445588516351902e-05,
      "loss": 0.0,
      "step": 1608
    },
    {
      "epoch": 0.27062484231771927,
      "grad_norm": NaN,
      "learning_rate": 8.443640582210697e-05,
      "loss": 0.0,
      "step": 1609
    },
    {
      "epoch": 0.27079303675048355,
      "grad_norm": NaN,
      "learning_rate": 8.4416916532403e-05,
      "loss": 0.0,
      "step": 1610
    },
    {
      "epoch": 0.27096123118324783,
      "grad_norm": NaN,
      "learning_rate": 8.439741730003732e-05,
      "loss": 0.0,
      "step": 1611
    },
    {
      "epoch": 0.2711294256160121,
      "grad_norm": NaN,
      "learning_rate": 8.437790813064305e-05,
      "loss": 0.0,
      "step": 1612
    },
    {
      "epoch": 0.2712976200487764,
      "grad_norm": NaN,
      "learning_rate": 8.43583890298562e-05,
      "loss": 0.0,
      "step": 1613
    },
    {
      "epoch": 0.27146581448154067,
      "grad_norm": NaN,
      "learning_rate": 8.433886000331557e-05,
      "loss": 0.0,
      "step": 1614
    },
    {
      "epoch": 0.27163400891430495,
      "grad_norm": NaN,
      "learning_rate": 8.431932105666291e-05,
      "loss": 0.0,
      "step": 1615
    },
    {
      "epoch": 0.2718022033470692,
      "grad_norm": NaN,
      "learning_rate": 8.429977219554282e-05,
      "loss": 0.0,
      "step": 1616
    },
    {
      "epoch": 0.2719703977798335,
      "grad_norm": NaN,
      "learning_rate": 8.42802134256027e-05,
      "loss": 0.0,
      "step": 1617
    },
    {
      "epoch": 0.2721385922125978,
      "grad_norm": NaN,
      "learning_rate": 8.426064475249287e-05,
      "loss": 0.0,
      "step": 1618
    },
    {
      "epoch": 0.27230678664536206,
      "grad_norm": NaN,
      "learning_rate": 8.424106618186653e-05,
      "loss": 0.0,
      "step": 1619
    },
    {
      "epoch": 0.2724749810781263,
      "grad_norm": NaN,
      "learning_rate": 8.422147771937969e-05,
      "loss": 0.0,
      "step": 1620
    },
    {
      "epoch": 0.27264317551089057,
      "grad_norm": NaN,
      "learning_rate": 8.420187937069124e-05,
      "loss": 0.0,
      "step": 1621
    },
    {
      "epoch": 0.27281136994365485,
      "grad_norm": NaN,
      "learning_rate": 8.418227114146291e-05,
      "loss": 0.0,
      "step": 1622
    },
    {
      "epoch": 0.2729795643764191,
      "grad_norm": NaN,
      "learning_rate": 8.416265303735932e-05,
      "loss": 0.0,
      "step": 1623
    },
    {
      "epoch": 0.2731477588091834,
      "grad_norm": NaN,
      "learning_rate": 8.414302506404791e-05,
      "loss": 0.0,
      "step": 1624
    },
    {
      "epoch": 0.2733159532419477,
      "grad_norm": NaN,
      "learning_rate": 8.412338722719899e-05,
      "loss": 0.0,
      "step": 1625
    },
    {
      "epoch": 0.27348414767471196,
      "grad_norm": NaN,
      "learning_rate": 8.41037395324857e-05,
      "loss": 0.0,
      "step": 1626
    },
    {
      "epoch": 0.27365234210747624,
      "grad_norm": NaN,
      "learning_rate": 8.408408198558403e-05,
      "loss": 0.0,
      "step": 1627
    },
    {
      "epoch": 0.2738205365402405,
      "grad_norm": NaN,
      "learning_rate": 8.406441459217285e-05,
      "loss": 0.0,
      "step": 1628
    },
    {
      "epoch": 0.2739887309730048,
      "grad_norm": NaN,
      "learning_rate": 8.404473735793384e-05,
      "loss": 0.0,
      "step": 1629
    },
    {
      "epoch": 0.2741569254057691,
      "grad_norm": NaN,
      "learning_rate": 8.402505028855152e-05,
      "loss": 0.0,
      "step": 1630
    },
    {
      "epoch": 0.27432511983853336,
      "grad_norm": NaN,
      "learning_rate": 8.400535338971326e-05,
      "loss": 0.0,
      "step": 1631
    },
    {
      "epoch": 0.27449331427129764,
      "grad_norm": NaN,
      "learning_rate": 8.398564666710932e-05,
      "loss": 0.0,
      "step": 1632
    },
    {
      "epoch": 0.2746615087040619,
      "grad_norm": NaN,
      "learning_rate": 8.396593012643272e-05,
      "loss": 0.0,
      "step": 1633
    },
    {
      "epoch": 0.2748297031368262,
      "grad_norm": NaN,
      "learning_rate": 8.394620377337931e-05,
      "loss": 0.0,
      "step": 1634
    },
    {
      "epoch": 0.2749978975695904,
      "grad_norm": NaN,
      "learning_rate": 8.392646761364789e-05,
      "loss": 0.0,
      "step": 1635
    },
    {
      "epoch": 0.2751660920023547,
      "grad_norm": NaN,
      "learning_rate": 8.390672165293995e-05,
      "loss": 0.0,
      "step": 1636
    },
    {
      "epoch": 0.275334286435119,
      "grad_norm": NaN,
      "learning_rate": 8.388696589695991e-05,
      "loss": 0.0,
      "step": 1637
    },
    {
      "epoch": 0.27550248086788326,
      "grad_norm": NaN,
      "learning_rate": 8.386720035141497e-05,
      "loss": 0.0,
      "step": 1638
    },
    {
      "epoch": 0.27567067530064754,
      "grad_norm": NaN,
      "learning_rate": 8.38474250220152e-05,
      "loss": 0.0,
      "step": 1639
    },
    {
      "epoch": 0.2758388697334118,
      "grad_norm": NaN,
      "learning_rate": 8.382763991447344e-05,
      "loss": 0.0,
      "step": 1640
    },
    {
      "epoch": 0.2760070641661761,
      "grad_norm": NaN,
      "learning_rate": 8.38078450345054e-05,
      "loss": 0.0,
      "step": 1641
    },
    {
      "epoch": 0.2761752585989404,
      "grad_norm": NaN,
      "learning_rate": 8.37880403878296e-05,
      "loss": 0.0,
      "step": 1642
    },
    {
      "epoch": 0.27634345303170466,
      "grad_norm": NaN,
      "learning_rate": 8.376822598016739e-05,
      "loss": 0.0,
      "step": 1643
    },
    {
      "epoch": 0.27651164746446893,
      "grad_norm": NaN,
      "learning_rate": 8.37484018172429e-05,
      "loss": 0.0,
      "step": 1644
    },
    {
      "epoch": 0.2766798418972332,
      "grad_norm": NaN,
      "learning_rate": 8.372856790478315e-05,
      "loss": 0.0,
      "step": 1645
    },
    {
      "epoch": 0.2768480363299975,
      "grad_norm": NaN,
      "learning_rate": 8.37087242485179e-05,
      "loss": 0.0,
      "step": 1646
    },
    {
      "epoch": 0.2770162307627618,
      "grad_norm": NaN,
      "learning_rate": 8.368887085417977e-05,
      "loss": 0.0,
      "step": 1647
    },
    {
      "epoch": 0.27718442519552605,
      "grad_norm": NaN,
      "learning_rate": 8.366900772750421e-05,
      "loss": 0.0,
      "step": 1648
    },
    {
      "epoch": 0.27735261962829033,
      "grad_norm": NaN,
      "learning_rate": 8.364913487422942e-05,
      "loss": 0.0,
      "step": 1649
    },
    {
      "epoch": 0.27752081406105455,
      "grad_norm": NaN,
      "learning_rate": 8.362925230009646e-05,
      "loss": 0.0,
      "step": 1650
    },
    {
      "epoch": 0.27768900849381883,
      "grad_norm": NaN,
      "learning_rate": 8.360936001084917e-05,
      "loss": 0.0,
      "step": 1651
    },
    {
      "epoch": 0.2778572029265831,
      "grad_norm": NaN,
      "learning_rate": 8.358945801223422e-05,
      "loss": 0.0,
      "step": 1652
    },
    {
      "epoch": 0.2780253973593474,
      "grad_norm": NaN,
      "learning_rate": 8.356954631000107e-05,
      "loss": 0.0,
      "step": 1653
    },
    {
      "epoch": 0.27819359179211167,
      "grad_norm": NaN,
      "learning_rate": 8.354962490990202e-05,
      "loss": 0.0,
      "step": 1654
    },
    {
      "epoch": 0.27836178622487595,
      "grad_norm": NaN,
      "learning_rate": 8.352969381769208e-05,
      "loss": 0.0,
      "step": 1655
    },
    {
      "epoch": 0.27852998065764023,
      "grad_norm": NaN,
      "learning_rate": 8.350975303912916e-05,
      "loss": 0.0,
      "step": 1656
    },
    {
      "epoch": 0.2786981750904045,
      "grad_norm": NaN,
      "learning_rate": 8.348980257997392e-05,
      "loss": 0.0,
      "step": 1657
    },
    {
      "epoch": 0.2788663695231688,
      "grad_norm": NaN,
      "learning_rate": 8.346984244598981e-05,
      "loss": 0.0,
      "step": 1658
    },
    {
      "epoch": 0.27903456395593307,
      "grad_norm": NaN,
      "learning_rate": 8.344987264294312e-05,
      "loss": 0.0,
      "step": 1659
    },
    {
      "epoch": 0.27920275838869735,
      "grad_norm": NaN,
      "learning_rate": 8.342989317660289e-05,
      "loss": 0.0,
      "step": 1660
    },
    {
      "epoch": 0.2793709528214616,
      "grad_norm": NaN,
      "learning_rate": 8.340990405274091e-05,
      "loss": 0.0,
      "step": 1661
    },
    {
      "epoch": 0.2795391472542259,
      "grad_norm": NaN,
      "learning_rate": 8.338990527713191e-05,
      "loss": 0.0,
      "step": 1662
    },
    {
      "epoch": 0.2797073416869902,
      "grad_norm": NaN,
      "learning_rate": 8.336989685555324e-05,
      "loss": 0.0,
      "step": 1663
    },
    {
      "epoch": 0.2798755361197544,
      "grad_norm": NaN,
      "learning_rate": 8.334987879378516e-05,
      "loss": 0.0,
      "step": 1664
    },
    {
      "epoch": 0.2800437305525187,
      "grad_norm": NaN,
      "learning_rate": 8.33298510976106e-05,
      "loss": 0.0,
      "step": 1665
    },
    {
      "epoch": 0.28021192498528297,
      "grad_norm": NaN,
      "learning_rate": 8.33098137728154e-05,
      "loss": 0.0,
      "step": 1666
    },
    {
      "epoch": 0.28038011941804725,
      "grad_norm": NaN,
      "learning_rate": 8.328976682518807e-05,
      "loss": 0.0,
      "step": 1667
    },
    {
      "epoch": 0.2805483138508115,
      "grad_norm": NaN,
      "learning_rate": 8.326971026052e-05,
      "loss": 0.0,
      "step": 1668
    },
    {
      "epoch": 0.2807165082835758,
      "grad_norm": NaN,
      "learning_rate": 8.324964408460525e-05,
      "loss": 0.0,
      "step": 1669
    },
    {
      "epoch": 0.2808847027163401,
      "grad_norm": NaN,
      "learning_rate": 8.322956830324076e-05,
      "loss": 0.0,
      "step": 1670
    },
    {
      "epoch": 0.28105289714910436,
      "grad_norm": NaN,
      "learning_rate": 8.320948292222617e-05,
      "loss": 0.0,
      "step": 1671
    },
    {
      "epoch": 0.28122109158186864,
      "grad_norm": NaN,
      "learning_rate": 8.318938794736392e-05,
      "loss": 0.0,
      "step": 1672
    },
    {
      "epoch": 0.2813892860146329,
      "grad_norm": NaN,
      "learning_rate": 8.316928338445927e-05,
      "loss": 0.0,
      "step": 1673
    },
    {
      "epoch": 0.2815574804473972,
      "grad_norm": NaN,
      "learning_rate": 8.314916923932014e-05,
      "loss": 0.0,
      "step": 1674
    },
    {
      "epoch": 0.2817256748801615,
      "grad_norm": NaN,
      "learning_rate": 8.312904551775731e-05,
      "loss": 0.0,
      "step": 1675
    },
    {
      "epoch": 0.28189386931292576,
      "grad_norm": NaN,
      "learning_rate": 8.31089122255843e-05,
      "loss": 0.0,
      "step": 1676
    },
    {
      "epoch": 0.28206206374569004,
      "grad_norm": NaN,
      "learning_rate": 8.308876936861739e-05,
      "loss": 0.0,
      "step": 1677
    },
    {
      "epoch": 0.2822302581784543,
      "grad_norm": NaN,
      "learning_rate": 8.306861695267563e-05,
      "loss": 0.0,
      "step": 1678
    },
    {
      "epoch": 0.28239845261121854,
      "grad_norm": NaN,
      "learning_rate": 8.304845498358081e-05,
      "loss": 0.0,
      "step": 1679
    },
    {
      "epoch": 0.2825666470439828,
      "grad_norm": NaN,
      "learning_rate": 8.302828346715752e-05,
      "loss": 0.0,
      "step": 1680
    },
    {
      "epoch": 0.2827348414767471,
      "grad_norm": NaN,
      "learning_rate": 8.300810240923307e-05,
      "loss": 0.0,
      "step": 1681
    },
    {
      "epoch": 0.2829030359095114,
      "grad_norm": NaN,
      "learning_rate": 8.298791181563754e-05,
      "loss": 0.0,
      "step": 1682
    },
    {
      "epoch": 0.28307123034227566,
      "grad_norm": NaN,
      "learning_rate": 8.296771169220378e-05,
      "loss": 0.0,
      "step": 1683
    },
    {
      "epoch": 0.28323942477503994,
      "grad_norm": NaN,
      "learning_rate": 8.294750204476736e-05,
      "loss": 0.0,
      "step": 1684
    },
    {
      "epoch": 0.2834076192078042,
      "grad_norm": NaN,
      "learning_rate": 8.292728287916665e-05,
      "loss": 0.0,
      "step": 1685
    },
    {
      "epoch": 0.2835758136405685,
      "grad_norm": NaN,
      "learning_rate": 8.290705420124272e-05,
      "loss": 0.0,
      "step": 1686
    },
    {
      "epoch": 0.2837440080733328,
      "grad_norm": NaN,
      "learning_rate": 8.28868160168394e-05,
      "loss": 0.0,
      "step": 1687
    },
    {
      "epoch": 0.28391220250609706,
      "grad_norm": NaN,
      "learning_rate": 8.286656833180329e-05,
      "loss": 0.0,
      "step": 1688
    },
    {
      "epoch": 0.28408039693886133,
      "grad_norm": NaN,
      "learning_rate": 8.284631115198371e-05,
      "loss": 0.0,
      "step": 1689
    },
    {
      "epoch": 0.2842485913716256,
      "grad_norm": NaN,
      "learning_rate": 8.282604448323277e-05,
      "loss": 0.0,
      "step": 1690
    },
    {
      "epoch": 0.2844167858043899,
      "grad_norm": NaN,
      "learning_rate": 8.280576833140522e-05,
      "loss": 0.0,
      "step": 1691
    },
    {
      "epoch": 0.2845849802371542,
      "grad_norm": NaN,
      "learning_rate": 8.278548270235865e-05,
      "loss": 0.0,
      "step": 1692
    },
    {
      "epoch": 0.28475317466991845,
      "grad_norm": NaN,
      "learning_rate": 8.276518760195335e-05,
      "loss": 0.0,
      "step": 1693
    },
    {
      "epoch": 0.2849213691026827,
      "grad_norm": NaN,
      "learning_rate": 8.274488303605234e-05,
      "loss": 0.0,
      "step": 1694
    },
    {
      "epoch": 0.28508956353544695,
      "grad_norm": NaN,
      "learning_rate": 8.272456901052139e-05,
      "loss": 0.0,
      "step": 1695
    },
    {
      "epoch": 0.28525775796821123,
      "grad_norm": NaN,
      "learning_rate": 8.270424553122899e-05,
      "loss": 0.0,
      "step": 1696
    },
    {
      "epoch": 0.2854259524009755,
      "grad_norm": NaN,
      "learning_rate": 8.268391260404635e-05,
      "loss": 0.0,
      "step": 1697
    },
    {
      "epoch": 0.2855941468337398,
      "grad_norm": NaN,
      "learning_rate": 8.266357023484745e-05,
      "loss": 0.0,
      "step": 1698
    },
    {
      "epoch": 0.28576234126650407,
      "grad_norm": NaN,
      "learning_rate": 8.264321842950898e-05,
      "loss": 0.0,
      "step": 1699
    },
    {
      "epoch": 0.28593053569926835,
      "grad_norm": NaN,
      "learning_rate": 8.262285719391029e-05,
      "loss": 0.0,
      "step": 1700
    },
    {
      "epoch": 0.28609873013203263,
      "grad_norm": NaN,
      "learning_rate": 8.260248653393359e-05,
      "loss": 0.0,
      "step": 1701
    },
    {
      "epoch": 0.2862669245647969,
      "grad_norm": NaN,
      "learning_rate": 8.258210645546368e-05,
      "loss": 0.0,
      "step": 1702
    },
    {
      "epoch": 0.2864351189975612,
      "grad_norm": NaN,
      "learning_rate": 8.256171696438817e-05,
      "loss": 0.0,
      "step": 1703
    },
    {
      "epoch": 0.28660331343032547,
      "grad_norm": NaN,
      "learning_rate": 8.254131806659733e-05,
      "loss": 0.0,
      "step": 1704
    },
    {
      "epoch": 0.28677150786308975,
      "grad_norm": NaN,
      "learning_rate": 8.252090976798419e-05,
      "loss": 0.0,
      "step": 1705
    },
    {
      "epoch": 0.286939702295854,
      "grad_norm": NaN,
      "learning_rate": 8.250049207444446e-05,
      "loss": 0.0,
      "step": 1706
    },
    {
      "epoch": 0.2871078967286183,
      "grad_norm": NaN,
      "learning_rate": 8.248006499187661e-05,
      "loss": 0.0,
      "step": 1707
    },
    {
      "epoch": 0.2872760911613826,
      "grad_norm": NaN,
      "learning_rate": 8.245962852618176e-05,
      "loss": 0.0,
      "step": 1708
    },
    {
      "epoch": 0.2874442855941468,
      "grad_norm": NaN,
      "learning_rate": 8.243918268326382e-05,
      "loss": 0.0,
      "step": 1709
    },
    {
      "epoch": 0.2876124800269111,
      "grad_norm": NaN,
      "learning_rate": 8.241872746902935e-05,
      "loss": 0.0,
      "step": 1710
    },
    {
      "epoch": 0.28778067445967537,
      "grad_norm": NaN,
      "learning_rate": 8.23982628893876e-05,
      "loss": 0.0,
      "step": 1711
    },
    {
      "epoch": 0.28794886889243965,
      "grad_norm": NaN,
      "learning_rate": 8.237778895025059e-05,
      "loss": 0.0,
      "step": 1712
    },
    {
      "epoch": 0.2881170633252039,
      "grad_norm": NaN,
      "learning_rate": 8.235730565753301e-05,
      "loss": 0.0,
      "step": 1713
    },
    {
      "epoch": 0.2882852577579682,
      "grad_norm": NaN,
      "learning_rate": 8.233681301715225e-05,
      "loss": 0.0,
      "step": 1714
    },
    {
      "epoch": 0.2884534521907325,
      "grad_norm": NaN,
      "learning_rate": 8.23163110350284e-05,
      "loss": 0.0,
      "step": 1715
    },
    {
      "epoch": 0.28862164662349676,
      "grad_norm": NaN,
      "learning_rate": 8.229579971708426e-05,
      "loss": 0.0,
      "step": 1716
    },
    {
      "epoch": 0.28878984105626104,
      "grad_norm": NaN,
      "learning_rate": 8.22752790692453e-05,
      "loss": 0.0,
      "step": 1717
    },
    {
      "epoch": 0.2889580354890253,
      "grad_norm": NaN,
      "learning_rate": 8.225474909743973e-05,
      "loss": 0.0,
      "step": 1718
    },
    {
      "epoch": 0.2891262299217896,
      "grad_norm": NaN,
      "learning_rate": 8.223420980759842e-05,
      "loss": 0.0,
      "step": 1719
    },
    {
      "epoch": 0.2892944243545539,
      "grad_norm": NaN,
      "learning_rate": 8.221366120565493e-05,
      "loss": 0.0,
      "step": 1720
    },
    {
      "epoch": 0.28946261878731816,
      "grad_norm": NaN,
      "learning_rate": 8.219310329754554e-05,
      "loss": 0.0,
      "step": 1721
    },
    {
      "epoch": 0.28963081322008244,
      "grad_norm": NaN,
      "learning_rate": 8.21725360892092e-05,
      "loss": 0.0,
      "step": 1722
    },
    {
      "epoch": 0.2897990076528467,
      "grad_norm": NaN,
      "learning_rate": 8.215195958658752e-05,
      "loss": 0.0,
      "step": 1723
    },
    {
      "epoch": 0.28996720208561094,
      "grad_norm": NaN,
      "learning_rate": 8.213137379562485e-05,
      "loss": 0.0,
      "step": 1724
    },
    {
      "epoch": 0.2901353965183752,
      "grad_norm": NaN,
      "learning_rate": 8.211077872226819e-05,
      "loss": 0.0,
      "step": 1725
    },
    {
      "epoch": 0.2903035909511395,
      "grad_norm": NaN,
      "learning_rate": 8.20901743724672e-05,
      "loss": 0.0,
      "step": 1726
    },
    {
      "epoch": 0.2904717853839038,
      "grad_norm": NaN,
      "learning_rate": 8.206956075217428e-05,
      "loss": 0.0,
      "step": 1727
    },
    {
      "epoch": 0.29063997981666806,
      "grad_norm": NaN,
      "learning_rate": 8.204893786734444e-05,
      "loss": 0.0,
      "step": 1728
    },
    {
      "epoch": 0.29080817424943234,
      "grad_norm": NaN,
      "learning_rate": 8.202830572393543e-05,
      "loss": 0.0,
      "step": 1729
    },
    {
      "epoch": 0.2909763686821966,
      "grad_norm": NaN,
      "learning_rate": 8.200766432790764e-05,
      "loss": 0.0,
      "step": 1730
    },
    {
      "epoch": 0.2911445631149609,
      "grad_norm": NaN,
      "learning_rate": 8.198701368522412e-05,
      "loss": 0.0,
      "step": 1731
    },
    {
      "epoch": 0.2913127575477252,
      "grad_norm": NaN,
      "learning_rate": 8.196635380185063e-05,
      "loss": 0.0,
      "step": 1732
    },
    {
      "epoch": 0.29148095198048946,
      "grad_norm": NaN,
      "learning_rate": 8.194568468375558e-05,
      "loss": 0.0,
      "step": 1733
    },
    {
      "epoch": 0.29164914641325373,
      "grad_norm": NaN,
      "learning_rate": 8.192500633691001e-05,
      "loss": 0.0,
      "step": 1734
    },
    {
      "epoch": 0.291817340846018,
      "grad_norm": NaN,
      "learning_rate": 8.190431876728772e-05,
      "loss": 0.0,
      "step": 1735
    },
    {
      "epoch": 0.2919855352787823,
      "grad_norm": NaN,
      "learning_rate": 8.188362198086506e-05,
      "loss": 0.0,
      "step": 1736
    },
    {
      "epoch": 0.2921537297115466,
      "grad_norm": NaN,
      "learning_rate": 8.186291598362114e-05,
      "loss": 0.0,
      "step": 1737
    },
    {
      "epoch": 0.2923219241443108,
      "grad_norm": NaN,
      "learning_rate": 8.184220078153768e-05,
      "loss": 0.0,
      "step": 1738
    },
    {
      "epoch": 0.2924901185770751,
      "grad_norm": NaN,
      "learning_rate": 8.182147638059906e-05,
      "loss": 0.0,
      "step": 1739
    },
    {
      "epoch": 0.29265831300983935,
      "grad_norm": NaN,
      "learning_rate": 8.180074278679233e-05,
      "loss": 0.0,
      "step": 1740
    },
    {
      "epoch": 0.29282650744260363,
      "grad_norm": NaN,
      "learning_rate": 8.178000000610722e-05,
      "loss": 0.0,
      "step": 1741
    },
    {
      "epoch": 0.2929947018753679,
      "grad_norm": NaN,
      "learning_rate": 8.175924804453604e-05,
      "loss": 0.0,
      "step": 1742
    },
    {
      "epoch": 0.2931628963081322,
      "grad_norm": NaN,
      "learning_rate": 8.173848690807385e-05,
      "loss": 0.0,
      "step": 1743
    },
    {
      "epoch": 0.29333109074089647,
      "grad_norm": NaN,
      "learning_rate": 8.171771660271825e-05,
      "loss": 0.0,
      "step": 1744
    },
    {
      "epoch": 0.29349928517366075,
      "grad_norm": NaN,
      "learning_rate": 8.169693713446959e-05,
      "loss": 0.0,
      "step": 1745
    },
    {
      "epoch": 0.29366747960642503,
      "grad_norm": NaN,
      "learning_rate": 8.167614850933083e-05,
      "loss": 0.0,
      "step": 1746
    },
    {
      "epoch": 0.2938356740391893,
      "grad_norm": NaN,
      "learning_rate": 8.165535073330757e-05,
      "loss": 0.0,
      "step": 1747
    },
    {
      "epoch": 0.2940038684719536,
      "grad_norm": NaN,
      "learning_rate": 8.163454381240803e-05,
      "loss": 0.0,
      "step": 1748
    },
    {
      "epoch": 0.29417206290471787,
      "grad_norm": NaN,
      "learning_rate": 8.161372775264311e-05,
      "loss": 0.0,
      "step": 1749
    },
    {
      "epoch": 0.29434025733748215,
      "grad_norm": NaN,
      "learning_rate": 8.159290256002632e-05,
      "loss": 0.0,
      "step": 1750
    },
    {
      "epoch": 0.2945084517702464,
      "grad_norm": NaN,
      "learning_rate": 8.157206824057386e-05,
      "loss": 0.0,
      "step": 1751
    },
    {
      "epoch": 0.2946766462030107,
      "grad_norm": NaN,
      "learning_rate": 8.155122480030453e-05,
      "loss": 0.0,
      "step": 1752
    },
    {
      "epoch": 0.29484484063577493,
      "grad_norm": NaN,
      "learning_rate": 8.153037224523972e-05,
      "loss": 0.0,
      "step": 1753
    },
    {
      "epoch": 0.2950130350685392,
      "grad_norm": NaN,
      "learning_rate": 8.150951058140354e-05,
      "loss": 0.0,
      "step": 1754
    },
    {
      "epoch": 0.2951812295013035,
      "grad_norm": NaN,
      "learning_rate": 8.148863981482267e-05,
      "loss": 0.0,
      "step": 1755
    },
    {
      "epoch": 0.29534942393406777,
      "grad_norm": NaN,
      "learning_rate": 8.146775995152646e-05,
      "loss": 0.0,
      "step": 1756
    },
    {
      "epoch": 0.29551761836683205,
      "grad_norm": NaN,
      "learning_rate": 8.144687099754685e-05,
      "loss": 0.0,
      "step": 1757
    },
    {
      "epoch": 0.2956858127995963,
      "grad_norm": NaN,
      "learning_rate": 8.142597295891846e-05,
      "loss": 0.0,
      "step": 1758
    },
    {
      "epoch": 0.2958540072323606,
      "grad_norm": NaN,
      "learning_rate": 8.140506584167845e-05,
      "loss": 0.0,
      "step": 1759
    },
    {
      "epoch": 0.2960222016651249,
      "grad_norm": NaN,
      "learning_rate": 8.138414965186667e-05,
      "loss": 0.0,
      "step": 1760
    },
    {
      "epoch": 0.29619039609788916,
      "grad_norm": NaN,
      "learning_rate": 8.13632243955256e-05,
      "loss": 0.0,
      "step": 1761
    },
    {
      "epoch": 0.29635859053065344,
      "grad_norm": NaN,
      "learning_rate": 8.134229007870027e-05,
      "loss": 0.0,
      "step": 1762
    },
    {
      "epoch": 0.2965267849634177,
      "grad_norm": NaN,
      "learning_rate": 8.132134670743838e-05,
      "loss": 0.0,
      "step": 1763
    },
    {
      "epoch": 0.296694979396182,
      "grad_norm": NaN,
      "learning_rate": 8.130039428779028e-05,
      "loss": 0.0,
      "step": 1764
    },
    {
      "epoch": 0.2968631738289463,
      "grad_norm": NaN,
      "learning_rate": 8.127943282580884e-05,
      "loss": 0.0,
      "step": 1765
    },
    {
      "epoch": 0.29703136826171056,
      "grad_norm": NaN,
      "learning_rate": 8.12584623275496e-05,
      "loss": 0.0,
      "step": 1766
    },
    {
      "epoch": 0.29719956269447484,
      "grad_norm": NaN,
      "learning_rate": 8.123748279907071e-05,
      "loss": 0.0,
      "step": 1767
    },
    {
      "epoch": 0.29736775712723906,
      "grad_norm": NaN,
      "learning_rate": 8.121649424643293e-05,
      "loss": 0.0,
      "step": 1768
    },
    {
      "epoch": 0.29753595156000334,
      "grad_norm": NaN,
      "learning_rate": 8.119549667569959e-05,
      "loss": 0.0,
      "step": 1769
    },
    {
      "epoch": 0.2977041459927676,
      "grad_norm": NaN,
      "learning_rate": 8.117449009293668e-05,
      "loss": 0.0,
      "step": 1770
    },
    {
      "epoch": 0.2978723404255319,
      "grad_norm": NaN,
      "learning_rate": 8.115347450421275e-05,
      "loss": 0.0,
      "step": 1771
    },
    {
      "epoch": 0.2980405348582962,
      "grad_norm": NaN,
      "learning_rate": 8.113244991559898e-05,
      "loss": 0.0,
      "step": 1772
    },
    {
      "epoch": 0.29820872929106046,
      "grad_norm": NaN,
      "learning_rate": 8.111141633316914e-05,
      "loss": 0.0,
      "step": 1773
    },
    {
      "epoch": 0.29837692372382474,
      "grad_norm": NaN,
      "learning_rate": 8.109037376299958e-05,
      "loss": 0.0,
      "step": 1774
    },
    {
      "epoch": 0.298545118156589,
      "grad_norm": NaN,
      "learning_rate": 8.106932221116928e-05,
      "loss": 0.0,
      "step": 1775
    },
    {
      "epoch": 0.2987133125893533,
      "grad_norm": NaN,
      "learning_rate": 8.104826168375979e-05,
      "loss": 0.0,
      "step": 1776
    },
    {
      "epoch": 0.2988815070221176,
      "grad_norm": NaN,
      "learning_rate": 8.102719218685524e-05,
      "loss": 0.0,
      "step": 1777
    },
    {
      "epoch": 0.29904970145488186,
      "grad_norm": NaN,
      "learning_rate": 8.100611372654241e-05,
      "loss": 0.0,
      "step": 1778
    },
    {
      "epoch": 0.29921789588764613,
      "grad_norm": NaN,
      "learning_rate": 8.09850263089106e-05,
      "loss": 0.0,
      "step": 1779
    },
    {
      "epoch": 0.2993860903204104,
      "grad_norm": NaN,
      "learning_rate": 8.096392994005177e-05,
      "loss": 0.0,
      "step": 1780
    },
    {
      "epoch": 0.2995542847531747,
      "grad_norm": NaN,
      "learning_rate": 8.094282462606036e-05,
      "loss": 0.0,
      "step": 1781
    },
    {
      "epoch": 0.299722479185939,
      "grad_norm": NaN,
      "learning_rate": 8.092171037303352e-05,
      "loss": 0.0,
      "step": 1782
    },
    {
      "epoch": 0.2998906736187032,
      "grad_norm": NaN,
      "learning_rate": 8.090058718707087e-05,
      "loss": 0.0,
      "step": 1783
    },
    {
      "epoch": 0.3000588680514675,
      "grad_norm": NaN,
      "learning_rate": 8.087945507427473e-05,
      "loss": 0.0,
      "step": 1784
    },
    {
      "epoch": 0.30022706248423175,
      "grad_norm": NaN,
      "learning_rate": 8.085831404074984e-05,
      "loss": 0.0,
      "step": 1785
    },
    {
      "epoch": 0.30039525691699603,
      "grad_norm": NaN,
      "learning_rate": 8.08371640926037e-05,
      "loss": 0.0,
      "step": 1786
    },
    {
      "epoch": 0.3005634513497603,
      "grad_norm": NaN,
      "learning_rate": 8.081600523594622e-05,
      "loss": 0.0,
      "step": 1787
    },
    {
      "epoch": 0.3007316457825246,
      "grad_norm": NaN,
      "learning_rate": 8.079483747689e-05,
      "loss": 0.0,
      "step": 1788
    },
    {
      "epoch": 0.30089984021528887,
      "grad_norm": NaN,
      "learning_rate": 8.077366082155016e-05,
      "loss": 0.0,
      "step": 1789
    },
    {
      "epoch": 0.30106803464805315,
      "grad_norm": NaN,
      "learning_rate": 8.075247527604437e-05,
      "loss": 0.0,
      "step": 1790
    },
    {
      "epoch": 0.30123622908081743,
      "grad_norm": NaN,
      "learning_rate": 8.073128084649294e-05,
      "loss": 0.0,
      "step": 1791
    },
    {
      "epoch": 0.3014044235135817,
      "grad_norm": NaN,
      "learning_rate": 8.071007753901868e-05,
      "loss": 0.0,
      "step": 1792
    },
    {
      "epoch": 0.301572617946346,
      "grad_norm": NaN,
      "learning_rate": 8.068886535974699e-05,
      "loss": 0.0,
      "step": 1793
    },
    {
      "epoch": 0.30174081237911027,
      "grad_norm": NaN,
      "learning_rate": 8.066764431480583e-05,
      "loss": 0.0,
      "step": 1794
    },
    {
      "epoch": 0.30190900681187455,
      "grad_norm": NaN,
      "learning_rate": 8.064641441032572e-05,
      "loss": 0.0,
      "step": 1795
    },
    {
      "epoch": 0.3020772012446388,
      "grad_norm": NaN,
      "learning_rate": 8.062517565243977e-05,
      "loss": 0.0,
      "step": 1796
    },
    {
      "epoch": 0.30224539567740305,
      "grad_norm": NaN,
      "learning_rate": 8.060392804728357e-05,
      "loss": 0.0,
      "step": 1797
    },
    {
      "epoch": 0.30241359011016733,
      "grad_norm": NaN,
      "learning_rate": 8.058267160099535e-05,
      "loss": 0.0,
      "step": 1798
    },
    {
      "epoch": 0.3025817845429316,
      "grad_norm": NaN,
      "learning_rate": 8.056140631971586e-05,
      "loss": 0.0,
      "step": 1799
    },
    {
      "epoch": 0.3027499789756959,
      "grad_norm": NaN,
      "learning_rate": 8.054013220958838e-05,
      "loss": 0.0,
      "step": 1800
    },
    {
      "epoch": 0.30291817340846017,
      "grad_norm": NaN,
      "learning_rate": 8.051884927675877e-05,
      "loss": 0.0,
      "step": 1801
    },
    {
      "epoch": 0.30308636784122445,
      "grad_norm": NaN,
      "learning_rate": 8.049755752737545e-05,
      "loss": 0.0,
      "step": 1802
    },
    {
      "epoch": 0.3032545622739887,
      "grad_norm": NaN,
      "learning_rate": 8.047625696758933e-05,
      "loss": 0.0,
      "step": 1803
    },
    {
      "epoch": 0.303422756706753,
      "grad_norm": NaN,
      "learning_rate": 8.045494760355395e-05,
      "loss": 0.0,
      "step": 1804
    },
    {
      "epoch": 0.3035909511395173,
      "grad_norm": NaN,
      "learning_rate": 8.043362944142531e-05,
      "loss": 0.0,
      "step": 1805
    },
    {
      "epoch": 0.30375914557228156,
      "grad_norm": NaN,
      "learning_rate": 8.0412302487362e-05,
      "loss": 0.0,
      "step": 1806
    },
    {
      "epoch": 0.30392734000504584,
      "grad_norm": NaN,
      "learning_rate": 8.039096674752514e-05,
      "loss": 0.0,
      "step": 1807
    },
    {
      "epoch": 0.3040955344378101,
      "grad_norm": NaN,
      "learning_rate": 8.036962222807838e-05,
      "loss": 0.0,
      "step": 1808
    },
    {
      "epoch": 0.3042637288705744,
      "grad_norm": NaN,
      "learning_rate": 8.034826893518793e-05,
      "loss": 0.0,
      "step": 1809
    },
    {
      "epoch": 0.3044319233033387,
      "grad_norm": NaN,
      "learning_rate": 8.03269068750225e-05,
      "loss": 0.0,
      "step": 1810
    },
    {
      "epoch": 0.30460011773610296,
      "grad_norm": NaN,
      "learning_rate": 8.030553605375335e-05,
      "loss": 0.0,
      "step": 1811
    },
    {
      "epoch": 0.3047683121688672,
      "grad_norm": NaN,
      "learning_rate": 8.028415647755427e-05,
      "loss": 0.0,
      "step": 1812
    },
    {
      "epoch": 0.30493650660163146,
      "grad_norm": NaN,
      "learning_rate": 8.026276815260161e-05,
      "loss": 0.0,
      "step": 1813
    },
    {
      "epoch": 0.30510470103439574,
      "grad_norm": NaN,
      "learning_rate": 8.024137108507417e-05,
      "loss": 0.0,
      "step": 1814
    },
    {
      "epoch": 0.30527289546716,
      "grad_norm": NaN,
      "learning_rate": 8.021996528115335e-05,
      "loss": 0.0,
      "step": 1815
    },
    {
      "epoch": 0.3054410898999243,
      "grad_norm": NaN,
      "learning_rate": 8.019855074702304e-05,
      "loss": 0.0,
      "step": 1816
    },
    {
      "epoch": 0.3056092843326886,
      "grad_norm": NaN,
      "learning_rate": 8.017712748886967e-05,
      "loss": 0.0,
      "step": 1817
    },
    {
      "epoch": 0.30577747876545286,
      "grad_norm": NaN,
      "learning_rate": 8.015569551288217e-05,
      "loss": 0.0,
      "step": 1818
    },
    {
      "epoch": 0.30594567319821714,
      "grad_norm": NaN,
      "learning_rate": 8.013425482525201e-05,
      "loss": 0.0,
      "step": 1819
    },
    {
      "epoch": 0.3061138676309814,
      "grad_norm": NaN,
      "learning_rate": 8.011280543217313e-05,
      "loss": 0.0,
      "step": 1820
    },
    {
      "epoch": 0.3062820620637457,
      "grad_norm": NaN,
      "learning_rate": 8.009134733984206e-05,
      "loss": 0.0,
      "step": 1821
    },
    {
      "epoch": 0.30645025649651,
      "grad_norm": NaN,
      "learning_rate": 8.006988055445778e-05,
      "loss": 0.0,
      "step": 1822
    },
    {
      "epoch": 0.30661845092927426,
      "grad_norm": NaN,
      "learning_rate": 8.004840508222182e-05,
      "loss": 0.0,
      "step": 1823
    },
    {
      "epoch": 0.30678664536203853,
      "grad_norm": NaN,
      "learning_rate": 8.002692092933818e-05,
      "loss": 0.0,
      "step": 1824
    },
    {
      "epoch": 0.3069548397948028,
      "grad_norm": NaN,
      "learning_rate": 8.000542810201341e-05,
      "loss": 0.0,
      "step": 1825
    },
    {
      "epoch": 0.3071230342275671,
      "grad_norm": NaN,
      "learning_rate": 7.998392660645655e-05,
      "loss": 0.0,
      "step": 1826
    },
    {
      "epoch": 0.3072912286603313,
      "grad_norm": NaN,
      "learning_rate": 7.996241644887912e-05,
      "loss": 0.0,
      "step": 1827
    },
    {
      "epoch": 0.3074594230930956,
      "grad_norm": NaN,
      "learning_rate": 7.994089763549517e-05,
      "loss": 0.0,
      "step": 1828
    },
    {
      "epoch": 0.3076276175258599,
      "grad_norm": NaN,
      "learning_rate": 7.991937017252126e-05,
      "loss": 0.0,
      "step": 1829
    },
    {
      "epoch": 0.30779581195862415,
      "grad_norm": NaN,
      "learning_rate": 7.989783406617643e-05,
      "loss": 0.0,
      "step": 1830
    },
    {
      "epoch": 0.30796400639138843,
      "grad_norm": NaN,
      "learning_rate": 7.98762893226822e-05,
      "loss": 0.0,
      "step": 1831
    },
    {
      "epoch": 0.3081322008241527,
      "grad_norm": NaN,
      "learning_rate": 7.985473594826263e-05,
      "loss": 0.0,
      "step": 1832
    },
    {
      "epoch": 0.308300395256917,
      "grad_norm": NaN,
      "learning_rate": 7.983317394914425e-05,
      "loss": 0.0,
      "step": 1833
    },
    {
      "epoch": 0.30846858968968127,
      "grad_norm": NaN,
      "learning_rate": 7.981160333155605e-05,
      "loss": 0.0,
      "step": 1834
    },
    {
      "epoch": 0.30863678412244555,
      "grad_norm": NaN,
      "learning_rate": 7.979002410172954e-05,
      "loss": 0.0,
      "step": 1835
    },
    {
      "epoch": 0.30880497855520983,
      "grad_norm": NaN,
      "learning_rate": 7.976843626589875e-05,
      "loss": 0.0,
      "step": 1836
    },
    {
      "epoch": 0.3089731729879741,
      "grad_norm": NaN,
      "learning_rate": 7.974683983030016e-05,
      "loss": 0.0,
      "step": 1837
    },
    {
      "epoch": 0.3091413674207384,
      "grad_norm": NaN,
      "learning_rate": 7.972523480117271e-05,
      "loss": 0.0,
      "step": 1838
    },
    {
      "epoch": 0.30930956185350267,
      "grad_norm": NaN,
      "learning_rate": 7.970362118475787e-05,
      "loss": 0.0,
      "step": 1839
    },
    {
      "epoch": 0.30947775628626695,
      "grad_norm": NaN,
      "learning_rate": 7.968199898729958e-05,
      "loss": 0.0,
      "step": 1840
    },
    {
      "epoch": 0.3096459507190312,
      "grad_norm": NaN,
      "learning_rate": 7.966036821504424e-05,
      "loss": 0.0,
      "step": 1841
    },
    {
      "epoch": 0.30981414515179545,
      "grad_norm": NaN,
      "learning_rate": 7.96387288742407e-05,
      "loss": 0.0,
      "step": 1842
    },
    {
      "epoch": 0.30998233958455973,
      "grad_norm": NaN,
      "learning_rate": 7.96170809711404e-05,
      "loss": 0.0,
      "step": 1843
    },
    {
      "epoch": 0.310150534017324,
      "grad_norm": NaN,
      "learning_rate": 7.959542451199713e-05,
      "loss": 0.0,
      "step": 1844
    },
    {
      "epoch": 0.3103187284500883,
      "grad_norm": NaN,
      "learning_rate": 7.957375950306721e-05,
      "loss": 0.0,
      "step": 1845
    },
    {
      "epoch": 0.31048692288285257,
      "grad_norm": NaN,
      "learning_rate": 7.955208595060941e-05,
      "loss": 0.0,
      "step": 1846
    },
    {
      "epoch": 0.31065511731561685,
      "grad_norm": NaN,
      "learning_rate": 7.953040386088497e-05,
      "loss": 0.0,
      "step": 1847
    },
    {
      "epoch": 0.3108233117483811,
      "grad_norm": NaN,
      "learning_rate": 7.950871324015763e-05,
      "loss": 0.0,
      "step": 1848
    },
    {
      "epoch": 0.3109915061811454,
      "grad_norm": NaN,
      "learning_rate": 7.948701409469354e-05,
      "loss": 0.0,
      "step": 1849
    },
    {
      "epoch": 0.3111597006139097,
      "grad_norm": NaN,
      "learning_rate": 7.946530643076138e-05,
      "loss": 0.0,
      "step": 1850
    },
    {
      "epoch": 0.31132789504667396,
      "grad_norm": NaN,
      "learning_rate": 7.944359025463219e-05,
      "loss": 0.0,
      "step": 1851
    },
    {
      "epoch": 0.31149608947943824,
      "grad_norm": NaN,
      "learning_rate": 7.942186557257958e-05,
      "loss": 0.0,
      "step": 1852
    },
    {
      "epoch": 0.3116642839122025,
      "grad_norm": NaN,
      "learning_rate": 7.940013239087953e-05,
      "loss": 0.0,
      "step": 1853
    },
    {
      "epoch": 0.3118324783449668,
      "grad_norm": NaN,
      "learning_rate": 7.937839071581055e-05,
      "loss": 0.0,
      "step": 1854
    },
    {
      "epoch": 0.3120006727777311,
      "grad_norm": NaN,
      "learning_rate": 7.935664055365355e-05,
      "loss": 0.0,
      "step": 1855
    },
    {
      "epoch": 0.31216886721049536,
      "grad_norm": NaN,
      "learning_rate": 7.933488191069189e-05,
      "loss": 0.0,
      "step": 1856
    },
    {
      "epoch": 0.3123370616432596,
      "grad_norm": NaN,
      "learning_rate": 7.931311479321144e-05,
      "loss": 0.0,
      "step": 1857
    },
    {
      "epoch": 0.31250525607602386,
      "grad_norm": NaN,
      "learning_rate": 7.929133920750044e-05,
      "loss": 0.0,
      "step": 1858
    },
    {
      "epoch": 0.31267345050878814,
      "grad_norm": NaN,
      "learning_rate": 7.926955515984961e-05,
      "loss": 0.0,
      "step": 1859
    },
    {
      "epoch": 0.3128416449415524,
      "grad_norm": NaN,
      "learning_rate": 7.924776265655216e-05,
      "loss": 0.0,
      "step": 1860
    },
    {
      "epoch": 0.3130098393743167,
      "grad_norm": NaN,
      "learning_rate": 7.922596170390365e-05,
      "loss": 0.0,
      "step": 1861
    },
    {
      "epoch": 0.313178033807081,
      "grad_norm": NaN,
      "learning_rate": 7.920415230820218e-05,
      "loss": 0.0,
      "step": 1862
    },
    {
      "epoch": 0.31334622823984526,
      "grad_norm": NaN,
      "learning_rate": 7.918233447574818e-05,
      "loss": 0.0,
      "step": 1863
    },
    {
      "epoch": 0.31351442267260954,
      "grad_norm": NaN,
      "learning_rate": 7.916050821284462e-05,
      "loss": 0.0,
      "step": 1864
    },
    {
      "epoch": 0.3136826171053738,
      "grad_norm": NaN,
      "learning_rate": 7.913867352579688e-05,
      "loss": 0.0,
      "step": 1865
    },
    {
      "epoch": 0.3138508115381381,
      "grad_norm": NaN,
      "learning_rate": 7.911683042091269e-05,
      "loss": 0.0,
      "step": 1866
    },
    {
      "epoch": 0.3140190059709024,
      "grad_norm": NaN,
      "learning_rate": 7.909497890450235e-05,
      "loss": 0.0,
      "step": 1867
    },
    {
      "epoch": 0.31418720040366666,
      "grad_norm": NaN,
      "learning_rate": 7.907311898287847e-05,
      "loss": 0.0,
      "step": 1868
    },
    {
      "epoch": 0.31435539483643093,
      "grad_norm": NaN,
      "learning_rate": 7.905125066235616e-05,
      "loss": 0.0,
      "step": 1869
    },
    {
      "epoch": 0.3145235892691952,
      "grad_norm": NaN,
      "learning_rate": 7.902937394925293e-05,
      "loss": 0.0,
      "step": 1870
    },
    {
      "epoch": 0.31469178370195944,
      "grad_norm": NaN,
      "learning_rate": 7.90074888498887e-05,
      "loss": 0.0,
      "step": 1871
    },
    {
      "epoch": 0.3148599781347237,
      "grad_norm": NaN,
      "learning_rate": 7.898559537058586e-05,
      "loss": 0.0,
      "step": 1872
    },
    {
      "epoch": 0.315028172567488,
      "grad_norm": NaN,
      "learning_rate": 7.896369351766916e-05,
      "loss": 0.0,
      "step": 1873
    },
    {
      "epoch": 0.3151963670002523,
      "grad_norm": NaN,
      "learning_rate": 7.894178329746582e-05,
      "loss": 0.0,
      "step": 1874
    },
    {
      "epoch": 0.31536456143301655,
      "grad_norm": NaN,
      "learning_rate": 7.891986471630546e-05,
      "loss": 0.0,
      "step": 1875
    },
    {
      "epoch": 0.31553275586578083,
      "grad_norm": NaN,
      "learning_rate": 7.889793778052011e-05,
      "loss": 0.0,
      "step": 1876
    },
    {
      "epoch": 0.3157009502985451,
      "grad_norm": NaN,
      "learning_rate": 7.887600249644418e-05,
      "loss": 0.0,
      "step": 1877
    },
    {
      "epoch": 0.3158691447313094,
      "grad_norm": NaN,
      "learning_rate": 7.88540588704146e-05,
      "loss": 0.0,
      "step": 1878
    },
    {
      "epoch": 0.31603733916407367,
      "grad_norm": NaN,
      "learning_rate": 7.883210690877055e-05,
      "loss": 0.0,
      "step": 1879
    },
    {
      "epoch": 0.31620553359683795,
      "grad_norm": NaN,
      "learning_rate": 7.881014661785379e-05,
      "loss": 0.0,
      "step": 1880
    },
    {
      "epoch": 0.31637372802960223,
      "grad_norm": NaN,
      "learning_rate": 7.878817800400833e-05,
      "loss": 0.0,
      "step": 1881
    },
    {
      "epoch": 0.3165419224623665,
      "grad_norm": NaN,
      "learning_rate": 7.876620107358073e-05,
      "loss": 0.0,
      "step": 1882
    },
    {
      "epoch": 0.3167101168951308,
      "grad_norm": NaN,
      "learning_rate": 7.874421583291982e-05,
      "loss": 0.0,
      "step": 1883
    },
    {
      "epoch": 0.31687831132789507,
      "grad_norm": NaN,
      "learning_rate": 7.872222228837691e-05,
      "loss": 0.0,
      "step": 1884
    },
    {
      "epoch": 0.31704650576065935,
      "grad_norm": NaN,
      "learning_rate": 7.870022044630569e-05,
      "loss": 0.0,
      "step": 1885
    },
    {
      "epoch": 0.31721470019342357,
      "grad_norm": NaN,
      "learning_rate": 7.867821031306223e-05,
      "loss": 0.0,
      "step": 1886
    },
    {
      "epoch": 0.31738289462618785,
      "grad_norm": NaN,
      "learning_rate": 7.865619189500504e-05,
      "loss": 0.0,
      "step": 1887
    },
    {
      "epoch": 0.31755108905895213,
      "grad_norm": NaN,
      "learning_rate": 7.863416519849499e-05,
      "loss": 0.0,
      "step": 1888
    },
    {
      "epoch": 0.3177192834917164,
      "grad_norm": NaN,
      "learning_rate": 7.861213022989531e-05,
      "loss": 0.0,
      "step": 1889
    },
    {
      "epoch": 0.3178874779244807,
      "grad_norm": NaN,
      "learning_rate": 7.85900869955717e-05,
      "loss": 0.0,
      "step": 1890
    },
    {
      "epoch": 0.31805567235724497,
      "grad_norm": NaN,
      "learning_rate": 7.856803550189218e-05,
      "loss": 0.0,
      "step": 1891
    },
    {
      "epoch": 0.31822386679000925,
      "grad_norm": NaN,
      "learning_rate": 7.854597575522717e-05,
      "loss": 0.0,
      "step": 1892
    },
    {
      "epoch": 0.3183920612227735,
      "grad_norm": NaN,
      "learning_rate": 7.85239077619495e-05,
      "loss": 0.0,
      "step": 1893
    },
    {
      "epoch": 0.3185602556555378,
      "grad_norm": NaN,
      "learning_rate": 7.850183152843439e-05,
      "loss": 0.0,
      "step": 1894
    },
    {
      "epoch": 0.3187284500883021,
      "grad_norm": NaN,
      "learning_rate": 7.847974706105935e-05,
      "loss": 0.0,
      "step": 1895
    },
    {
      "epoch": 0.31889664452106636,
      "grad_norm": NaN,
      "learning_rate": 7.845765436620441e-05,
      "loss": 0.0,
      "step": 1896
    },
    {
      "epoch": 0.31906483895383064,
      "grad_norm": NaN,
      "learning_rate": 7.843555345025186e-05,
      "loss": 0.0,
      "step": 1897
    },
    {
      "epoch": 0.3192330333865949,
      "grad_norm": NaN,
      "learning_rate": 7.841344431958641e-05,
      "loss": 0.0,
      "step": 1898
    },
    {
      "epoch": 0.3194012278193592,
      "grad_norm": NaN,
      "learning_rate": 7.839132698059515e-05,
      "loss": 0.0,
      "step": 1899
    },
    {
      "epoch": 0.3195694222521235,
      "grad_norm": NaN,
      "learning_rate": 7.836920143966751e-05,
      "loss": 0.0,
      "step": 1900
    },
    {
      "epoch": 0.3197376166848877,
      "grad_norm": NaN,
      "learning_rate": 7.834706770319536e-05,
      "loss": 0.0,
      "step": 1901
    },
    {
      "epoch": 0.319905811117652,
      "grad_norm": NaN,
      "learning_rate": 7.832492577757282e-05,
      "loss": 0.0,
      "step": 1902
    },
    {
      "epoch": 0.32007400555041626,
      "grad_norm": NaN,
      "learning_rate": 7.83027756691965e-05,
      "loss": 0.0,
      "step": 1903
    },
    {
      "epoch": 0.32024219998318054,
      "grad_norm": NaN,
      "learning_rate": 7.828061738446528e-05,
      "loss": 0.0,
      "step": 1904
    },
    {
      "epoch": 0.3204103944159448,
      "grad_norm": NaN,
      "learning_rate": 7.825845092978048e-05,
      "loss": 0.0,
      "step": 1905
    },
    {
      "epoch": 0.3205785888487091,
      "grad_norm": NaN,
      "learning_rate": 7.823627631154571e-05,
      "loss": 0.0,
      "step": 1906
    },
    {
      "epoch": 0.3207467832814734,
      "grad_norm": NaN,
      "learning_rate": 7.821409353616699e-05,
      "loss": 0.0,
      "step": 1907
    },
    {
      "epoch": 0.32091497771423766,
      "grad_norm": NaN,
      "learning_rate": 7.819190261005264e-05,
      "loss": 0.0,
      "step": 1908
    },
    {
      "epoch": 0.32108317214700194,
      "grad_norm": NaN,
      "learning_rate": 7.816970353961341e-05,
      "loss": 0.0,
      "step": 1909
    },
    {
      "epoch": 0.3212513665797662,
      "grad_norm": NaN,
      "learning_rate": 7.814749633126235e-05,
      "loss": 0.0,
      "step": 1910
    },
    {
      "epoch": 0.3214195610125305,
      "grad_norm": NaN,
      "learning_rate": 7.812528099141486e-05,
      "loss": 0.0,
      "step": 1911
    },
    {
      "epoch": 0.3215877554452948,
      "grad_norm": NaN,
      "learning_rate": 7.810305752648871e-05,
      "loss": 0.0,
      "step": 1912
    },
    {
      "epoch": 0.32175594987805906,
      "grad_norm": NaN,
      "learning_rate": 7.808082594290402e-05,
      "loss": 0.0,
      "step": 1913
    },
    {
      "epoch": 0.32192414431082333,
      "grad_norm": NaN,
      "learning_rate": 7.805858624708323e-05,
      "loss": 0.0,
      "step": 1914
    },
    {
      "epoch": 0.3220923387435876,
      "grad_norm": NaN,
      "learning_rate": 7.803633844545114e-05,
      "loss": 0.0,
      "step": 1915
    },
    {
      "epoch": 0.32226053317635184,
      "grad_norm": NaN,
      "learning_rate": 7.801408254443493e-05,
      "loss": 0.0,
      "step": 1916
    },
    {
      "epoch": 0.3224287276091161,
      "grad_norm": NaN,
      "learning_rate": 7.799181855046403e-05,
      "loss": 0.0,
      "step": 1917
    },
    {
      "epoch": 0.3225969220418804,
      "grad_norm": NaN,
      "learning_rate": 7.796954646997028e-05,
      "loss": 0.0,
      "step": 1918
    },
    {
      "epoch": 0.3227651164746447,
      "grad_norm": NaN,
      "learning_rate": 7.794726630938782e-05,
      "loss": 0.0,
      "step": 1919
    },
    {
      "epoch": 0.32293331090740895,
      "grad_norm": NaN,
      "learning_rate": 7.792497807515317e-05,
      "loss": 0.0,
      "step": 1920
    },
    {
      "epoch": 0.32310150534017323,
      "grad_norm": NaN,
      "learning_rate": 7.790268177370512e-05,
      "loss": 0.0,
      "step": 1921
    },
    {
      "epoch": 0.3232696997729375,
      "grad_norm": NaN,
      "learning_rate": 7.788037741148483e-05,
      "loss": 0.0,
      "step": 1922
    },
    {
      "epoch": 0.3234378942057018,
      "grad_norm": NaN,
      "learning_rate": 7.78580649949358e-05,
      "loss": 0.0,
      "step": 1923
    },
    {
      "epoch": 0.32360608863846607,
      "grad_norm": NaN,
      "learning_rate": 7.78357445305038e-05,
      "loss": 0.0,
      "step": 1924
    },
    {
      "epoch": 0.32377428307123035,
      "grad_norm": NaN,
      "learning_rate": 7.7813416024637e-05,
      "loss": 0.0,
      "step": 1925
    },
    {
      "epoch": 0.32394247750399463,
      "grad_norm": NaN,
      "learning_rate": 7.779107948378582e-05,
      "loss": 0.0,
      "step": 1926
    },
    {
      "epoch": 0.3241106719367589,
      "grad_norm": NaN,
      "learning_rate": 7.776873491440307e-05,
      "loss": 0.0,
      "step": 1927
    },
    {
      "epoch": 0.3242788663695232,
      "grad_norm": NaN,
      "learning_rate": 7.774638232294382e-05,
      "loss": 0.0,
      "step": 1928
    },
    {
      "epoch": 0.32444706080228747,
      "grad_norm": NaN,
      "learning_rate": 7.772402171586552e-05,
      "loss": 0.0,
      "step": 1929
    },
    {
      "epoch": 0.32461525523505175,
      "grad_norm": NaN,
      "learning_rate": 7.770165309962785e-05,
      "loss": 0.0,
      "step": 1930
    },
    {
      "epoch": 0.32478344966781597,
      "grad_norm": NaN,
      "learning_rate": 7.767927648069289e-05,
      "loss": 0.0,
      "step": 1931
    },
    {
      "epoch": 0.32495164410058025,
      "grad_norm": NaN,
      "learning_rate": 7.765689186552496e-05,
      "loss": 0.0,
      "step": 1932
    },
    {
      "epoch": 0.32511983853334453,
      "grad_norm": NaN,
      "learning_rate": 7.763449926059076e-05,
      "loss": 0.0,
      "step": 1933
    },
    {
      "epoch": 0.3252880329661088,
      "grad_norm": NaN,
      "learning_rate": 7.761209867235924e-05,
      "loss": 0.0,
      "step": 1934
    },
    {
      "epoch": 0.3254562273988731,
      "grad_norm": NaN,
      "learning_rate": 7.75896901073017e-05,
      "loss": 0.0,
      "step": 1935
    },
    {
      "epoch": 0.32562442183163737,
      "grad_norm": NaN,
      "learning_rate": 7.756727357189169e-05,
      "loss": 0.0,
      "step": 1936
    },
    {
      "epoch": 0.32579261626440165,
      "grad_norm": NaN,
      "learning_rate": 7.754484907260513e-05,
      "loss": 0.0,
      "step": 1937
    },
    {
      "epoch": 0.3259608106971659,
      "grad_norm": NaN,
      "learning_rate": 7.75224166159202e-05,
      "loss": 0.0,
      "step": 1938
    },
    {
      "epoch": 0.3261290051299302,
      "grad_norm": NaN,
      "learning_rate": 7.749997620831737e-05,
      "loss": 0.0,
      "step": 1939
    },
    {
      "epoch": 0.3262971995626945,
      "grad_norm": NaN,
      "learning_rate": 7.747752785627942e-05,
      "loss": 0.0,
      "step": 1940
    },
    {
      "epoch": 0.32646539399545876,
      "grad_norm": NaN,
      "learning_rate": 7.745507156629145e-05,
      "loss": 0.0,
      "step": 1941
    },
    {
      "epoch": 0.32663358842822304,
      "grad_norm": NaN,
      "learning_rate": 7.743260734484081e-05,
      "loss": 0.0,
      "step": 1942
    },
    {
      "epoch": 0.3268017828609873,
      "grad_norm": NaN,
      "learning_rate": 7.741013519841718e-05,
      "loss": 0.0,
      "step": 1943
    },
    {
      "epoch": 0.3269699772937516,
      "grad_norm": NaN,
      "learning_rate": 7.73876551335125e-05,
      "loss": 0.0,
      "step": 1944
    },
    {
      "epoch": 0.3271381717265158,
      "grad_norm": NaN,
      "learning_rate": 7.736516715662101e-05,
      "loss": 0.0,
      "step": 1945
    },
    {
      "epoch": 0.3273063661592801,
      "grad_norm": NaN,
      "learning_rate": 7.734267127423924e-05,
      "loss": 0.0,
      "step": 1946
    },
    {
      "epoch": 0.3274745605920444,
      "grad_norm": NaN,
      "learning_rate": 7.7320167492866e-05,
      "loss": 0.0,
      "step": 1947
    },
    {
      "epoch": 0.32764275502480866,
      "grad_norm": NaN,
      "learning_rate": 7.729765581900235e-05,
      "loss": 0.0,
      "step": 1948
    },
    {
      "epoch": 0.32781094945757294,
      "grad_norm": NaN,
      "learning_rate": 7.72751362591517e-05,
      "loss": 0.0,
      "step": 1949
    },
    {
      "epoch": 0.3279791438903372,
      "grad_norm": NaN,
      "learning_rate": 7.725260881981971e-05,
      "loss": 0.0,
      "step": 1950
    },
    {
      "epoch": 0.3281473383231015,
      "grad_norm": NaN,
      "learning_rate": 7.723007350751423e-05,
      "loss": 0.0,
      "step": 1951
    },
    {
      "epoch": 0.3283155327558658,
      "grad_norm": NaN,
      "learning_rate": 7.720753032874555e-05,
      "loss": 0.0,
      "step": 1952
    },
    {
      "epoch": 0.32848372718863006,
      "grad_norm": NaN,
      "learning_rate": 7.718497929002608e-05,
      "loss": 0.0,
      "step": 1953
    },
    {
      "epoch": 0.32865192162139434,
      "grad_norm": NaN,
      "learning_rate": 7.716242039787059e-05,
      "loss": 0.0,
      "step": 1954
    },
    {
      "epoch": 0.3288201160541586,
      "grad_norm": NaN,
      "learning_rate": 7.713985365879606e-05,
      "loss": 0.0,
      "step": 1955
    },
    {
      "epoch": 0.3289883104869229,
      "grad_norm": NaN,
      "learning_rate": 7.711727907932181e-05,
      "loss": 0.0,
      "step": 1956
    },
    {
      "epoch": 0.3291565049196872,
      "grad_norm": NaN,
      "learning_rate": 7.709469666596937e-05,
      "loss": 0.0,
      "step": 1957
    },
    {
      "epoch": 0.32932469935245146,
      "grad_norm": NaN,
      "learning_rate": 7.707210642526253e-05,
      "loss": 0.0,
      "step": 1958
    },
    {
      "epoch": 0.32949289378521573,
      "grad_norm": NaN,
      "learning_rate": 7.704950836372738e-05,
      "loss": 0.0,
      "step": 1959
    },
    {
      "epoch": 0.32966108821797996,
      "grad_norm": NaN,
      "learning_rate": 7.702690248789223e-05,
      "loss": 0.0,
      "step": 1960
    },
    {
      "epoch": 0.32982928265074424,
      "grad_norm": NaN,
      "learning_rate": 7.700428880428768e-05,
      "loss": 0.0,
      "step": 1961
    },
    {
      "epoch": 0.3299974770835085,
      "grad_norm": NaN,
      "learning_rate": 7.698166731944654e-05,
      "loss": 0.0,
      "step": 1962
    },
    {
      "epoch": 0.3301656715162728,
      "grad_norm": NaN,
      "learning_rate": 7.695903803990392e-05,
      "loss": 0.0,
      "step": 1963
    },
    {
      "epoch": 0.3303338659490371,
      "grad_norm": NaN,
      "learning_rate": 7.693640097219719e-05,
      "loss": 0.0,
      "step": 1964
    },
    {
      "epoch": 0.33050206038180135,
      "grad_norm": NaN,
      "learning_rate": 7.691375612286591e-05,
      "loss": 0.0,
      "step": 1965
    },
    {
      "epoch": 0.33067025481456563,
      "grad_norm": NaN,
      "learning_rate": 7.689110349845195e-05,
      "loss": 0.0,
      "step": 1966
    },
    {
      "epoch": 0.3308384492473299,
      "grad_norm": NaN,
      "learning_rate": 7.686844310549936e-05,
      "loss": 0.0,
      "step": 1967
    },
    {
      "epoch": 0.3310066436800942,
      "grad_norm": NaN,
      "learning_rate": 7.684577495055453e-05,
      "loss": 0.0,
      "step": 1968
    },
    {
      "epoch": 0.33117483811285847,
      "grad_norm": NaN,
      "learning_rate": 7.682309904016601e-05,
      "loss": 0.0,
      "step": 1969
    },
    {
      "epoch": 0.33134303254562275,
      "grad_norm": NaN,
      "learning_rate": 7.68004153808846e-05,
      "loss": 0.0,
      "step": 1970
    },
    {
      "epoch": 0.33151122697838703,
      "grad_norm": NaN,
      "learning_rate": 7.677772397926337e-05,
      "loss": 0.0,
      "step": 1971
    },
    {
      "epoch": 0.3316794214111513,
      "grad_norm": NaN,
      "learning_rate": 7.675502484185761e-05,
      "loss": 0.0,
      "step": 1972
    },
    {
      "epoch": 0.3318476158439156,
      "grad_norm": NaN,
      "learning_rate": 7.673231797522484e-05,
      "loss": 0.0,
      "step": 1973
    },
    {
      "epoch": 0.33201581027667987,
      "grad_norm": NaN,
      "learning_rate": 7.670960338592483e-05,
      "loss": 0.0,
      "step": 1974
    },
    {
      "epoch": 0.3321840047094441,
      "grad_norm": NaN,
      "learning_rate": 7.668688108051957e-05,
      "loss": 0.0,
      "step": 1975
    },
    {
      "epoch": 0.33235219914220837,
      "grad_norm": NaN,
      "learning_rate": 7.666415106557327e-05,
      "loss": 0.0,
      "step": 1976
    },
    {
      "epoch": 0.33252039357497265,
      "grad_norm": NaN,
      "learning_rate": 7.664141334765238e-05,
      "loss": 0.0,
      "step": 1977
    },
    {
      "epoch": 0.33268858800773693,
      "grad_norm": NaN,
      "learning_rate": 7.66186679333256e-05,
      "loss": 0.0,
      "step": 1978
    },
    {
      "epoch": 0.3328567824405012,
      "grad_norm": NaN,
      "learning_rate": 7.659591482916378e-05,
      "loss": 0.0,
      "step": 1979
    },
    {
      "epoch": 0.3330249768732655,
      "grad_norm": NaN,
      "learning_rate": 7.657315404174006e-05,
      "loss": 0.0,
      "step": 1980
    },
    {
      "epoch": 0.33319317130602977,
      "grad_norm": NaN,
      "learning_rate": 7.655038557762977e-05,
      "loss": 0.0,
      "step": 1981
    },
    {
      "epoch": 0.33336136573879405,
      "grad_norm": NaN,
      "learning_rate": 7.652760944341046e-05,
      "loss": 0.0,
      "step": 1982
    },
    {
      "epoch": 0.3335295601715583,
      "grad_norm": NaN,
      "learning_rate": 7.650482564566193e-05,
      "loss": 0.0,
      "step": 1983
    },
    {
      "epoch": 0.3336977546043226,
      "grad_norm": NaN,
      "learning_rate": 7.648203419096613e-05,
      "loss": 0.0,
      "step": 1984
    },
    {
      "epoch": 0.3338659490370869,
      "grad_norm": NaN,
      "learning_rate": 7.645923508590726e-05,
      "loss": 0.0,
      "step": 1985
    },
    {
      "epoch": 0.33403414346985116,
      "grad_norm": NaN,
      "learning_rate": 7.643642833707174e-05,
      "loss": 0.0,
      "step": 1986
    },
    {
      "epoch": 0.33420233790261544,
      "grad_norm": NaN,
      "learning_rate": 7.64136139510482e-05,
      "loss": 0.0,
      "step": 1987
    },
    {
      "epoch": 0.3343705323353797,
      "grad_norm": NaN,
      "learning_rate": 7.639079193442742e-05,
      "loss": 0.0,
      "step": 1988
    },
    {
      "epoch": 0.334538726768144,
      "grad_norm": NaN,
      "learning_rate": 7.636796229380243e-05,
      "loss": 0.0,
      "step": 1989
    },
    {
      "epoch": 0.3347069212009082,
      "grad_norm": NaN,
      "learning_rate": 7.63451250357685e-05,
      "loss": 0.0,
      "step": 1990
    },
    {
      "epoch": 0.3348751156336725,
      "grad_norm": NaN,
      "learning_rate": 7.6322280166923e-05,
      "loss": 0.0,
      "step": 1991
    },
    {
      "epoch": 0.3350433100664368,
      "grad_norm": NaN,
      "learning_rate": 7.629942769386562e-05,
      "loss": 0.0,
      "step": 1992
    },
    {
      "epoch": 0.33521150449920106,
      "grad_norm": NaN,
      "learning_rate": 7.627656762319812e-05,
      "loss": 0.0,
      "step": 1993
    },
    {
      "epoch": 0.33537969893196534,
      "grad_norm": NaN,
      "learning_rate": 7.625369996152456e-05,
      "loss": 0.0,
      "step": 1994
    },
    {
      "epoch": 0.3355478933647296,
      "grad_norm": NaN,
      "learning_rate": 7.623082471545113e-05,
      "loss": 0.0,
      "step": 1995
    },
    {
      "epoch": 0.3357160877974939,
      "grad_norm": NaN,
      "learning_rate": 7.620794189158626e-05,
      "loss": 0.0,
      "step": 1996
    },
    {
      "epoch": 0.3358842822302582,
      "grad_norm": NaN,
      "learning_rate": 7.618505149654051e-05,
      "loss": 0.0,
      "step": 1997
    },
    {
      "epoch": 0.33605247666302246,
      "grad_norm": NaN,
      "learning_rate": 7.616215353692668e-05,
      "loss": 0.0,
      "step": 1998
    },
    {
      "epoch": 0.33622067109578674,
      "grad_norm": NaN,
      "learning_rate": 7.613924801935973e-05,
      "loss": 0.0,
      "step": 1999
    },
    {
      "epoch": 0.336388865528551,
      "grad_norm": NaN,
      "learning_rate": 7.611633495045679e-05,
      "loss": 0.0,
      "step": 2000
    },
    {
      "epoch": 0.3365570599613153,
      "grad_norm": NaN,
      "learning_rate": 7.609341433683721e-05,
      "loss": 0.0,
      "step": 2001
    },
    {
      "epoch": 0.3367252543940796,
      "grad_norm": NaN,
      "learning_rate": 7.60704861851225e-05,
      "loss": 0.0,
      "step": 2002
    },
    {
      "epoch": 0.33689344882684386,
      "grad_norm": NaN,
      "learning_rate": 7.604755050193631e-05,
      "loss": 0.0,
      "step": 2003
    },
    {
      "epoch": 0.33706164325960813,
      "grad_norm": NaN,
      "learning_rate": 7.602460729390455e-05,
      "loss": 0.0,
      "step": 2004
    },
    {
      "epoch": 0.33722983769237236,
      "grad_norm": NaN,
      "learning_rate": 7.600165656765523e-05,
      "loss": 0.0,
      "step": 2005
    },
    {
      "epoch": 0.33739803212513664,
      "grad_norm": NaN,
      "learning_rate": 7.597869832981857e-05,
      "loss": 0.0,
      "step": 2006
    },
    {
      "epoch": 0.3375662265579009,
      "grad_norm": NaN,
      "learning_rate": 7.595573258702693e-05,
      "loss": 0.0,
      "step": 2007
    },
    {
      "epoch": 0.3377344209906652,
      "grad_norm": NaN,
      "learning_rate": 7.593275934591487e-05,
      "loss": 0.0,
      "step": 2008
    },
    {
      "epoch": 0.3379026154234295,
      "grad_norm": NaN,
      "learning_rate": 7.59097786131191e-05,
      "loss": 0.0,
      "step": 2009
    },
    {
      "epoch": 0.33807080985619375,
      "grad_norm": NaN,
      "learning_rate": 7.588679039527848e-05,
      "loss": 0.0,
      "step": 2010
    },
    {
      "epoch": 0.33823900428895803,
      "grad_norm": NaN,
      "learning_rate": 7.586379469903408e-05,
      "loss": 0.0,
      "step": 2011
    },
    {
      "epoch": 0.3384071987217223,
      "grad_norm": NaN,
      "learning_rate": 7.584079153102908e-05,
      "loss": 0.0,
      "step": 2012
    },
    {
      "epoch": 0.3385753931544866,
      "grad_norm": NaN,
      "learning_rate": 7.581778089790885e-05,
      "loss": 0.0,
      "step": 2013
    },
    {
      "epoch": 0.33874358758725087,
      "grad_norm": NaN,
      "learning_rate": 7.579476280632088e-05,
      "loss": 0.0,
      "step": 2014
    },
    {
      "epoch": 0.33891178202001515,
      "grad_norm": NaN,
      "learning_rate": 7.577173726291487e-05,
      "loss": 0.0,
      "step": 2015
    },
    {
      "epoch": 0.33907997645277943,
      "grad_norm": NaN,
      "learning_rate": 7.574870427434261e-05,
      "loss": 0.0,
      "step": 2016
    },
    {
      "epoch": 0.3392481708855437,
      "grad_norm": NaN,
      "learning_rate": 7.572566384725809e-05,
      "loss": 0.0,
      "step": 2017
    },
    {
      "epoch": 0.339416365318308,
      "grad_norm": NaN,
      "learning_rate": 7.570261598831743e-05,
      "loss": 0.0,
      "step": 2018
    },
    {
      "epoch": 0.3395845597510722,
      "grad_norm": NaN,
      "learning_rate": 7.567956070417889e-05,
      "loss": 0.0,
      "step": 2019
    },
    {
      "epoch": 0.3397527541838365,
      "grad_norm": NaN,
      "learning_rate": 7.565649800150289e-05,
      "loss": 0.0,
      "step": 2020
    },
    {
      "epoch": 0.33992094861660077,
      "grad_norm": NaN,
      "learning_rate": 7.5633427886952e-05,
      "loss": 0.0,
      "step": 2021
    },
    {
      "epoch": 0.34008914304936505,
      "grad_norm": NaN,
      "learning_rate": 7.561035036719088e-05,
      "loss": 0.0,
      "step": 2022
    },
    {
      "epoch": 0.34025733748212933,
      "grad_norm": NaN,
      "learning_rate": 7.558726544888642e-05,
      "loss": 0.0,
      "step": 2023
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": NaN,
      "learning_rate": 7.556417313870754e-05,
      "loss": 0.0,
      "step": 2024
    },
    {
      "epoch": 0.3405937263476579,
      "grad_norm": NaN,
      "learning_rate": 7.55410734433254e-05,
      "loss": 0.0,
      "step": 2025
    },
    {
      "epoch": 0.34076192078042217,
      "grad_norm": NaN,
      "learning_rate": 7.55179663694132e-05,
      "loss": 0.0,
      "step": 2026
    },
    {
      "epoch": 0.34093011521318645,
      "grad_norm": NaN,
      "learning_rate": 7.549485192364635e-05,
      "loss": 0.0,
      "step": 2027
    },
    {
      "epoch": 0.3410983096459507,
      "grad_norm": NaN,
      "learning_rate": 7.547173011270232e-05,
      "loss": 0.0,
      "step": 2028
    },
    {
      "epoch": 0.341266504078715,
      "grad_norm": NaN,
      "learning_rate": 7.544860094326078e-05,
      "loss": 0.0,
      "step": 2029
    },
    {
      "epoch": 0.3414346985114793,
      "grad_norm": NaN,
      "learning_rate": 7.542546442200345e-05,
      "loss": 0.0,
      "step": 2030
    },
    {
      "epoch": 0.34160289294424356,
      "grad_norm": NaN,
      "learning_rate": 7.540232055561423e-05,
      "loss": 0.0,
      "step": 2031
    },
    {
      "epoch": 0.34177108737700784,
      "grad_norm": NaN,
      "learning_rate": 7.537916935077914e-05,
      "loss": 0.0,
      "step": 2032
    },
    {
      "epoch": 0.3419392818097721,
      "grad_norm": NaN,
      "learning_rate": 7.535601081418627e-05,
      "loss": 0.0,
      "step": 2033
    },
    {
      "epoch": 0.34210747624253635,
      "grad_norm": NaN,
      "learning_rate": 7.533284495252589e-05,
      "loss": 0.0,
      "step": 2034
    },
    {
      "epoch": 0.3422756706753006,
      "grad_norm": NaN,
      "learning_rate": 7.530967177249036e-05,
      "loss": 0.0,
      "step": 2035
    },
    {
      "epoch": 0.3424438651080649,
      "grad_norm": NaN,
      "learning_rate": 7.528649128077412e-05,
      "loss": 0.0,
      "step": 2036
    },
    {
      "epoch": 0.3426120595408292,
      "grad_norm": NaN,
      "learning_rate": 7.526330348407378e-05,
      "loss": 0.0,
      "step": 2037
    },
    {
      "epoch": 0.34278025397359346,
      "grad_norm": NaN,
      "learning_rate": 7.524010838908802e-05,
      "loss": 0.0,
      "step": 2038
    },
    {
      "epoch": 0.34294844840635774,
      "grad_norm": NaN,
      "learning_rate": 7.521690600251766e-05,
      "loss": 0.0,
      "step": 2039
    },
    {
      "epoch": 0.343116642839122,
      "grad_norm": NaN,
      "learning_rate": 7.519369633106559e-05,
      "loss": 0.0,
      "step": 2040
    },
    {
      "epoch": 0.3432848372718863,
      "grad_norm": NaN,
      "learning_rate": 7.517047938143684e-05,
      "loss": 0.0,
      "step": 2041
    },
    {
      "epoch": 0.3434530317046506,
      "grad_norm": NaN,
      "learning_rate": 7.514725516033854e-05,
      "loss": 0.0,
      "step": 2042
    },
    {
      "epoch": 0.34362122613741486,
      "grad_norm": NaN,
      "learning_rate": 7.512402367447986e-05,
      "loss": 0.0,
      "step": 2043
    },
    {
      "epoch": 0.34378942057017914,
      "grad_norm": NaN,
      "learning_rate": 7.510078493057218e-05,
      "loss": 0.0,
      "step": 2044
    },
    {
      "epoch": 0.3439576150029434,
      "grad_norm": NaN,
      "learning_rate": 7.507753893532886e-05,
      "loss": 0.0,
      "step": 2045
    },
    {
      "epoch": 0.3441258094357077,
      "grad_norm": NaN,
      "learning_rate": 7.505428569546542e-05,
      "loss": 0.0,
      "step": 2046
    },
    {
      "epoch": 0.344294003868472,
      "grad_norm": NaN,
      "learning_rate": 7.503102521769949e-05,
      "loss": 0.0,
      "step": 2047
    },
    {
      "epoch": 0.34446219830123626,
      "grad_norm": NaN,
      "learning_rate": 7.500775750875071e-05,
      "loss": 0.0,
      "step": 2048
    },
    {
      "epoch": 0.3446303927340005,
      "grad_norm": NaN,
      "learning_rate": 7.49844825753409e-05,
      "loss": 0.0,
      "step": 2049
    },
    {
      "epoch": 0.34479858716676476,
      "grad_norm": NaN,
      "learning_rate": 7.496120042419393e-05,
      "loss": 0.0,
      "step": 2050
    },
    {
      "epoch": 0.34496678159952904,
      "grad_norm": NaN,
      "learning_rate": 7.493791106203571e-05,
      "loss": 0.0,
      "step": 2051
    },
    {
      "epoch": 0.3451349760322933,
      "grad_norm": NaN,
      "learning_rate": 7.491461449559434e-05,
      "loss": 0.0,
      "step": 2052
    },
    {
      "epoch": 0.3453031704650576,
      "grad_norm": NaN,
      "learning_rate": 7.489131073159987e-05,
      "loss": 0.0,
      "step": 2053
    },
    {
      "epoch": 0.3454713648978219,
      "grad_norm": NaN,
      "learning_rate": 7.486799977678455e-05,
      "loss": 0.0,
      "step": 2054
    },
    {
      "epoch": 0.34563955933058615,
      "grad_norm": NaN,
      "learning_rate": 7.484468163788262e-05,
      "loss": 0.0,
      "step": 2055
    },
    {
      "epoch": 0.34580775376335043,
      "grad_norm": NaN,
      "learning_rate": 7.482135632163045e-05,
      "loss": 0.0,
      "step": 2056
    },
    {
      "epoch": 0.3459759481961147,
      "grad_norm": NaN,
      "learning_rate": 7.479802383476643e-05,
      "loss": 0.0,
      "step": 2057
    },
    {
      "epoch": 0.346144142628879,
      "grad_norm": NaN,
      "learning_rate": 7.477468418403111e-05,
      "loss": 0.0,
      "step": 2058
    },
    {
      "epoch": 0.34631233706164327,
      "grad_norm": NaN,
      "learning_rate": 7.475133737616699e-05,
      "loss": 0.0,
      "step": 2059
    },
    {
      "epoch": 0.34648053149440755,
      "grad_norm": NaN,
      "learning_rate": 7.472798341791877e-05,
      "loss": 0.0,
      "step": 2060
    },
    {
      "epoch": 0.34664872592717183,
      "grad_norm": NaN,
      "learning_rate": 7.470462231603308e-05,
      "loss": 0.0,
      "step": 2061
    },
    {
      "epoch": 0.3468169203599361,
      "grad_norm": NaN,
      "learning_rate": 7.468125407725872e-05,
      "loss": 0.0,
      "step": 2062
    },
    {
      "epoch": 0.3469851147927004,
      "grad_norm": NaN,
      "learning_rate": 7.465787870834652e-05,
      "loss": 0.0,
      "step": 2063
    },
    {
      "epoch": 0.3471533092254646,
      "grad_norm": NaN,
      "learning_rate": 7.463449621604932e-05,
      "loss": 0.0,
      "step": 2064
    },
    {
      "epoch": 0.3473215036582289,
      "grad_norm": NaN,
      "learning_rate": 7.461110660712209e-05,
      "loss": 0.0,
      "step": 2065
    },
    {
      "epoch": 0.34748969809099317,
      "grad_norm": NaN,
      "learning_rate": 7.458770988832183e-05,
      "loss": 0.0,
      "step": 2066
    },
    {
      "epoch": 0.34765789252375745,
      "grad_norm": NaN,
      "learning_rate": 7.456430606640757e-05,
      "loss": 0.0,
      "step": 2067
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": NaN,
      "learning_rate": 7.454089514814044e-05,
      "loss": 0.0,
      "step": 2068
    },
    {
      "epoch": 0.347994281389286,
      "grad_norm": NaN,
      "learning_rate": 7.451747714028356e-05,
      "loss": 0.0,
      "step": 2069
    },
    {
      "epoch": 0.3481624758220503,
      "grad_norm": NaN,
      "learning_rate": 7.449405204960216e-05,
      "loss": 0.0,
      "step": 2070
    },
    {
      "epoch": 0.34833067025481457,
      "grad_norm": NaN,
      "learning_rate": 7.447061988286346e-05,
      "loss": 0.0,
      "step": 2071
    },
    {
      "epoch": 0.34849886468757885,
      "grad_norm": NaN,
      "learning_rate": 7.444718064683676e-05,
      "loss": 0.0,
      "step": 2072
    },
    {
      "epoch": 0.3486670591203431,
      "grad_norm": NaN,
      "learning_rate": 7.442373434829341e-05,
      "loss": 0.0,
      "step": 2073
    },
    {
      "epoch": 0.3488352535531074,
      "grad_norm": NaN,
      "learning_rate": 7.440028099400677e-05,
      "loss": 0.0,
      "step": 2074
    },
    {
      "epoch": 0.3490034479858717,
      "grad_norm": NaN,
      "learning_rate": 7.437682059075223e-05,
      "loss": 0.0,
      "step": 2075
    },
    {
      "epoch": 0.34917164241863596,
      "grad_norm": NaN,
      "learning_rate": 7.435335314530727e-05,
      "loss": 0.0,
      "step": 2076
    },
    {
      "epoch": 0.34933983685140024,
      "grad_norm": NaN,
      "learning_rate": 7.432987866445135e-05,
      "loss": 0.0,
      "step": 2077
    },
    {
      "epoch": 0.34950803128416447,
      "grad_norm": NaN,
      "learning_rate": 7.4306397154966e-05,
      "loss": 0.0,
      "step": 2078
    },
    {
      "epoch": 0.34967622571692875,
      "grad_norm": NaN,
      "learning_rate": 7.428290862363477e-05,
      "loss": 0.0,
      "step": 2079
    },
    {
      "epoch": 0.349844420149693,
      "grad_norm": NaN,
      "learning_rate": 7.42594130772432e-05,
      "loss": 0.0,
      "step": 2080
    },
    {
      "epoch": 0.3500126145824573,
      "grad_norm": NaN,
      "learning_rate": 7.423591052257893e-05,
      "loss": 0.0,
      "step": 2081
    },
    {
      "epoch": 0.3501808090152216,
      "grad_norm": NaN,
      "learning_rate": 7.421240096643157e-05,
      "loss": 0.0,
      "step": 2082
    },
    {
      "epoch": 0.35034900344798586,
      "grad_norm": NaN,
      "learning_rate": 7.418888441559275e-05,
      "loss": 0.0,
      "step": 2083
    },
    {
      "epoch": 0.35051719788075014,
      "grad_norm": NaN,
      "learning_rate": 7.416536087685617e-05,
      "loss": 0.0,
      "step": 2084
    },
    {
      "epoch": 0.3506853923135144,
      "grad_norm": NaN,
      "learning_rate": 7.414183035701749e-05,
      "loss": 0.0,
      "step": 2085
    },
    {
      "epoch": 0.3508535867462787,
      "grad_norm": NaN,
      "learning_rate": 7.411829286287442e-05,
      "loss": 0.0,
      "step": 2086
    },
    {
      "epoch": 0.351021781179043,
      "grad_norm": NaN,
      "learning_rate": 7.409474840122669e-05,
      "loss": 0.0,
      "step": 2087
    },
    {
      "epoch": 0.35118997561180726,
      "grad_norm": NaN,
      "learning_rate": 7.407119697887602e-05,
      "loss": 0.0,
      "step": 2088
    },
    {
      "epoch": 0.35135817004457154,
      "grad_norm": NaN,
      "learning_rate": 7.404763860262616e-05,
      "loss": 0.0,
      "step": 2089
    },
    {
      "epoch": 0.3515263644773358,
      "grad_norm": NaN,
      "learning_rate": 7.402407327928285e-05,
      "loss": 0.0,
      "step": 2090
    },
    {
      "epoch": 0.3516945589101001,
      "grad_norm": NaN,
      "learning_rate": 7.400050101565387e-05,
      "loss": 0.0,
      "step": 2091
    },
    {
      "epoch": 0.3518627533428644,
      "grad_norm": NaN,
      "learning_rate": 7.397692181854895e-05,
      "loss": 0.0,
      "step": 2092
    },
    {
      "epoch": 0.3520309477756286,
      "grad_norm": NaN,
      "learning_rate": 7.395333569477988e-05,
      "loss": 0.0,
      "step": 2093
    },
    {
      "epoch": 0.3521991422083929,
      "grad_norm": NaN,
      "learning_rate": 7.392974265116043e-05,
      "loss": 0.0,
      "step": 2094
    },
    {
      "epoch": 0.35236733664115716,
      "grad_norm": NaN,
      "learning_rate": 7.390614269450634e-05,
      "loss": 0.0,
      "step": 2095
    },
    {
      "epoch": 0.35253553107392144,
      "grad_norm": NaN,
      "learning_rate": 7.38825358316354e-05,
      "loss": 0.0,
      "step": 2096
    },
    {
      "epoch": 0.3527037255066857,
      "grad_norm": NaN,
      "learning_rate": 7.385892206936735e-05,
      "loss": 0.0,
      "step": 2097
    },
    {
      "epoch": 0.35287191993945,
      "grad_norm": NaN,
      "learning_rate": 7.383530141452396e-05,
      "loss": 0.0,
      "step": 2098
    },
    {
      "epoch": 0.3530401143722143,
      "grad_norm": NaN,
      "learning_rate": 7.381167387392895e-05,
      "loss": 0.0,
      "step": 2099
    },
    {
      "epoch": 0.35320830880497855,
      "grad_norm": NaN,
      "learning_rate": 7.378803945440808e-05,
      "loss": 0.0,
      "step": 2100
    },
    {
      "epoch": 0.35337650323774283,
      "grad_norm": NaN,
      "learning_rate": 7.376439816278904e-05,
      "loss": 0.0,
      "step": 2101
    },
    {
      "epoch": 0.3535446976705071,
      "grad_norm": NaN,
      "learning_rate": 7.374075000590155e-05,
      "loss": 0.0,
      "step": 2102
    },
    {
      "epoch": 0.3537128921032714,
      "grad_norm": NaN,
      "learning_rate": 7.371709499057729e-05,
      "loss": 0.0,
      "step": 2103
    },
    {
      "epoch": 0.35388108653603567,
      "grad_norm": NaN,
      "learning_rate": 7.369343312364993e-05,
      "loss": 0.0,
      "step": 2104
    },
    {
      "epoch": 0.35404928096879995,
      "grad_norm": NaN,
      "learning_rate": 7.366976441195515e-05,
      "loss": 0.0,
      "step": 2105
    },
    {
      "epoch": 0.35421747540156423,
      "grad_norm": NaN,
      "learning_rate": 7.364608886233052e-05,
      "loss": 0.0,
      "step": 2106
    },
    {
      "epoch": 0.3543856698343285,
      "grad_norm": NaN,
      "learning_rate": 7.362240648161568e-05,
      "loss": 0.0,
      "step": 2107
    },
    {
      "epoch": 0.35455386426709273,
      "grad_norm": NaN,
      "learning_rate": 7.359871727665219e-05,
      "loss": 0.0,
      "step": 2108
    },
    {
      "epoch": 0.354722058699857,
      "grad_norm": NaN,
      "learning_rate": 7.35750212542836e-05,
      "loss": 0.0,
      "step": 2109
    },
    {
      "epoch": 0.3548902531326213,
      "grad_norm": NaN,
      "learning_rate": 7.355131842135543e-05,
      "loss": 0.0,
      "step": 2110
    },
    {
      "epoch": 0.35505844756538557,
      "grad_norm": NaN,
      "learning_rate": 7.352760878471516e-05,
      "loss": 0.0,
      "step": 2111
    },
    {
      "epoch": 0.35522664199814985,
      "grad_norm": NaN,
      "learning_rate": 7.350389235121223e-05,
      "loss": 0.0,
      "step": 2112
    },
    {
      "epoch": 0.35539483643091413,
      "grad_norm": NaN,
      "learning_rate": 7.348016912769806e-05,
      "loss": 0.0,
      "step": 2113
    },
    {
      "epoch": 0.3555630308636784,
      "grad_norm": NaN,
      "learning_rate": 7.345643912102601e-05,
      "loss": 0.0,
      "step": 2114
    },
    {
      "epoch": 0.3557312252964427,
      "grad_norm": NaN,
      "learning_rate": 7.343270233805144e-05,
      "loss": 0.0,
      "step": 2115
    },
    {
      "epoch": 0.35589941972920697,
      "grad_norm": NaN,
      "learning_rate": 7.340895878563162e-05,
      "loss": 0.0,
      "step": 2116
    },
    {
      "epoch": 0.35606761416197125,
      "grad_norm": NaN,
      "learning_rate": 7.338520847062582e-05,
      "loss": 0.0,
      "step": 2117
    },
    {
      "epoch": 0.3562358085947355,
      "grad_norm": NaN,
      "learning_rate": 7.336145139989518e-05,
      "loss": 0.0,
      "step": 2118
    },
    {
      "epoch": 0.3564040030274998,
      "grad_norm": NaN,
      "learning_rate": 7.333768758030292e-05,
      "loss": 0.0,
      "step": 2119
    },
    {
      "epoch": 0.3565721974602641,
      "grad_norm": NaN,
      "learning_rate": 7.331391701871413e-05,
      "loss": 0.0,
      "step": 2120
    },
    {
      "epoch": 0.35674039189302836,
      "grad_norm": NaN,
      "learning_rate": 7.329013972199583e-05,
      "loss": 0.0,
      "step": 2121
    },
    {
      "epoch": 0.35690858632579264,
      "grad_norm": NaN,
      "learning_rate": 7.326635569701703e-05,
      "loss": 0.0,
      "step": 2122
    },
    {
      "epoch": 0.35707678075855687,
      "grad_norm": NaN,
      "learning_rate": 7.324256495064867e-05,
      "loss": 0.0,
      "step": 2123
    },
    {
      "epoch": 0.35724497519132115,
      "grad_norm": NaN,
      "learning_rate": 7.321876748976361e-05,
      "loss": 0.0,
      "step": 2124
    },
    {
      "epoch": 0.3574131696240854,
      "grad_norm": NaN,
      "learning_rate": 7.319496332123671e-05,
      "loss": 0.0,
      "step": 2125
    },
    {
      "epoch": 0.3575813640568497,
      "grad_norm": NaN,
      "learning_rate": 7.317115245194469e-05,
      "loss": 0.0,
      "step": 2126
    },
    {
      "epoch": 0.357749558489614,
      "grad_norm": NaN,
      "learning_rate": 7.314733488876626e-05,
      "loss": 0.0,
      "step": 2127
    },
    {
      "epoch": 0.35791775292237826,
      "grad_norm": NaN,
      "learning_rate": 7.312351063858206e-05,
      "loss": 0.0,
      "step": 2128
    },
    {
      "epoch": 0.35808594735514254,
      "grad_norm": NaN,
      "learning_rate": 7.309967970827462e-05,
      "loss": 0.0,
      "step": 2129
    },
    {
      "epoch": 0.3582541417879068,
      "grad_norm": NaN,
      "learning_rate": 7.307584210472844e-05,
      "loss": 0.0,
      "step": 2130
    },
    {
      "epoch": 0.3584223362206711,
      "grad_norm": NaN,
      "learning_rate": 7.305199783482997e-05,
      "loss": 0.0,
      "step": 2131
    },
    {
      "epoch": 0.3585905306534354,
      "grad_norm": NaN,
      "learning_rate": 7.302814690546749e-05,
      "loss": 0.0,
      "step": 2132
    },
    {
      "epoch": 0.35875872508619966,
      "grad_norm": NaN,
      "learning_rate": 7.30042893235313e-05,
      "loss": 0.0,
      "step": 2133
    },
    {
      "epoch": 0.35892691951896394,
      "grad_norm": NaN,
      "learning_rate": 7.29804250959136e-05,
      "loss": 0.0,
      "step": 2134
    },
    {
      "epoch": 0.3590951139517282,
      "grad_norm": NaN,
      "learning_rate": 7.295655422950849e-05,
      "loss": 0.0,
      "step": 2135
    },
    {
      "epoch": 0.3592633083844925,
      "grad_norm": NaN,
      "learning_rate": 7.293267673121197e-05,
      "loss": 0.0,
      "step": 2136
    },
    {
      "epoch": 0.3594315028172568,
      "grad_norm": NaN,
      "learning_rate": 7.290879260792203e-05,
      "loss": 0.0,
      "step": 2137
    },
    {
      "epoch": 0.359599697250021,
      "grad_norm": NaN,
      "learning_rate": 7.288490186653848e-05,
      "loss": 0.0,
      "step": 2138
    },
    {
      "epoch": 0.3597678916827853,
      "grad_norm": NaN,
      "learning_rate": 7.286100451396312e-05,
      "loss": 0.0,
      "step": 2139
    },
    {
      "epoch": 0.35993608611554956,
      "grad_norm": NaN,
      "learning_rate": 7.283710055709959e-05,
      "loss": 0.0,
      "step": 2140
    },
    {
      "epoch": 0.36010428054831384,
      "grad_norm": NaN,
      "learning_rate": 7.281319000285352e-05,
      "loss": 0.0,
      "step": 2141
    },
    {
      "epoch": 0.3602724749810781,
      "grad_norm": NaN,
      "learning_rate": 7.278927285813238e-05,
      "loss": 0.0,
      "step": 2142
    },
    {
      "epoch": 0.3604406694138424,
      "grad_norm": NaN,
      "learning_rate": 7.276534912984556e-05,
      "loss": 0.0,
      "step": 2143
    },
    {
      "epoch": 0.3606088638466067,
      "grad_norm": NaN,
      "learning_rate": 7.274141882490435e-05,
      "loss": 0.0,
      "step": 2144
    },
    {
      "epoch": 0.36077705827937095,
      "grad_norm": NaN,
      "learning_rate": 7.271748195022199e-05,
      "loss": 0.0,
      "step": 2145
    },
    {
      "epoch": 0.36094525271213523,
      "grad_norm": NaN,
      "learning_rate": 7.269353851271352e-05,
      "loss": 0.0,
      "step": 2146
    },
    {
      "epoch": 0.3611134471448995,
      "grad_norm": NaN,
      "learning_rate": 7.266958851929598e-05,
      "loss": 0.0,
      "step": 2147
    },
    {
      "epoch": 0.3612816415776638,
      "grad_norm": NaN,
      "learning_rate": 7.26456319768882e-05,
      "loss": 0.0,
      "step": 2148
    },
    {
      "epoch": 0.36144983601042807,
      "grad_norm": NaN,
      "learning_rate": 7.2621668892411e-05,
      "loss": 0.0,
      "step": 2149
    },
    {
      "epoch": 0.36161803044319235,
      "grad_norm": NaN,
      "learning_rate": 7.259769927278705e-05,
      "loss": 0.0,
      "step": 2150
    },
    {
      "epoch": 0.36178622487595663,
      "grad_norm": NaN,
      "learning_rate": 7.257372312494088e-05,
      "loss": 0.0,
      "step": 2151
    },
    {
      "epoch": 0.36195441930872085,
      "grad_norm": NaN,
      "learning_rate": 7.254974045579893e-05,
      "loss": 0.0,
      "step": 2152
    },
    {
      "epoch": 0.36212261374148513,
      "grad_norm": NaN,
      "learning_rate": 7.252575127228954e-05,
      "loss": 0.0,
      "step": 2153
    },
    {
      "epoch": 0.3622908081742494,
      "grad_norm": NaN,
      "learning_rate": 7.250175558134291e-05,
      "loss": 0.0,
      "step": 2154
    },
    {
      "epoch": 0.3624590026070137,
      "grad_norm": NaN,
      "learning_rate": 7.247775338989111e-05,
      "loss": 0.0,
      "step": 2155
    },
    {
      "epoch": 0.36262719703977797,
      "grad_norm": NaN,
      "learning_rate": 7.245374470486814e-05,
      "loss": 0.0,
      "step": 2156
    },
    {
      "epoch": 0.36279539147254225,
      "grad_norm": NaN,
      "learning_rate": 7.242972953320984e-05,
      "loss": 0.0,
      "step": 2157
    },
    {
      "epoch": 0.36296358590530653,
      "grad_norm": NaN,
      "learning_rate": 7.240570788185388e-05,
      "loss": 0.0,
      "step": 2158
    },
    {
      "epoch": 0.3631317803380708,
      "grad_norm": NaN,
      "learning_rate": 7.238167975773987e-05,
      "loss": 0.0,
      "step": 2159
    },
    {
      "epoch": 0.3632999747708351,
      "grad_norm": NaN,
      "learning_rate": 7.235764516780927e-05,
      "loss": 0.0,
      "step": 2160
    },
    {
      "epoch": 0.36346816920359937,
      "grad_norm": NaN,
      "learning_rate": 7.23336041190054e-05,
      "loss": 0.0,
      "step": 2161
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": NaN,
      "learning_rate": 7.230955661827346e-05,
      "loss": 0.0,
      "step": 2162
    },
    {
      "epoch": 0.3638045580691279,
      "grad_norm": NaN,
      "learning_rate": 7.228550267256048e-05,
      "loss": 0.0,
      "step": 2163
    },
    {
      "epoch": 0.3639727525018922,
      "grad_norm": NaN,
      "learning_rate": 7.22614422888154e-05,
      "loss": 0.0,
      "step": 2164
    },
    {
      "epoch": 0.3641409469346565,
      "grad_norm": NaN,
      "learning_rate": 7.223737547398898e-05,
      "loss": 0.0,
      "step": 2165
    },
    {
      "epoch": 0.36430914136742076,
      "grad_norm": NaN,
      "learning_rate": 7.221330223503387e-05,
      "loss": 0.0,
      "step": 2166
    },
    {
      "epoch": 0.364477335800185,
      "grad_norm": NaN,
      "learning_rate": 7.218922257890454e-05,
      "loss": 0.0,
      "step": 2167
    },
    {
      "epoch": 0.36464553023294927,
      "grad_norm": NaN,
      "learning_rate": 7.216513651255735e-05,
      "loss": 0.0,
      "step": 2168
    },
    {
      "epoch": 0.36481372466571355,
      "grad_norm": NaN,
      "learning_rate": 7.214104404295045e-05,
      "loss": 0.0,
      "step": 2169
    },
    {
      "epoch": 0.3649819190984778,
      "grad_norm": NaN,
      "learning_rate": 7.211694517704394e-05,
      "loss": 0.0,
      "step": 2170
    },
    {
      "epoch": 0.3651501135312421,
      "grad_norm": NaN,
      "learning_rate": 7.20928399217997e-05,
      "loss": 0.0,
      "step": 2171
    },
    {
      "epoch": 0.3653183079640064,
      "grad_norm": NaN,
      "learning_rate": 7.206872828418146e-05,
      "loss": 0.0,
      "step": 2172
    },
    {
      "epoch": 0.36548650239677066,
      "grad_norm": NaN,
      "learning_rate": 7.20446102711548e-05,
      "loss": 0.0,
      "step": 2173
    },
    {
      "epoch": 0.36565469682953494,
      "grad_norm": NaN,
      "learning_rate": 7.202048588968715e-05,
      "loss": 0.0,
      "step": 2174
    },
    {
      "epoch": 0.3658228912622992,
      "grad_norm": NaN,
      "learning_rate": 7.199635514674778e-05,
      "loss": 0.0,
      "step": 2175
    },
    {
      "epoch": 0.3659910856950635,
      "grad_norm": NaN,
      "learning_rate": 7.197221804930776e-05,
      "loss": 0.0,
      "step": 2176
    },
    {
      "epoch": 0.3661592801278278,
      "grad_norm": NaN,
      "learning_rate": 7.194807460434005e-05,
      "loss": 0.0,
      "step": 2177
    },
    {
      "epoch": 0.36632747456059206,
      "grad_norm": NaN,
      "learning_rate": 7.192392481881944e-05,
      "loss": 0.0,
      "step": 2178
    },
    {
      "epoch": 0.36649566899335634,
      "grad_norm": NaN,
      "learning_rate": 7.189976869972248e-05,
      "loss": 0.0,
      "step": 2179
    },
    {
      "epoch": 0.3666638634261206,
      "grad_norm": NaN,
      "learning_rate": 7.187560625402766e-05,
      "loss": 0.0,
      "step": 2180
    },
    {
      "epoch": 0.3668320578588849,
      "grad_norm": NaN,
      "learning_rate": 7.18514374887152e-05,
      "loss": 0.0,
      "step": 2181
    },
    {
      "epoch": 0.3670002522916491,
      "grad_norm": NaN,
      "learning_rate": 7.18272624107672e-05,
      "loss": 0.0,
      "step": 2182
    },
    {
      "epoch": 0.3671684467244134,
      "grad_norm": NaN,
      "learning_rate": 7.180308102716758e-05,
      "loss": 0.0,
      "step": 2183
    },
    {
      "epoch": 0.3673366411571777,
      "grad_norm": NaN,
      "learning_rate": 7.177889334490206e-05,
      "loss": 0.0,
      "step": 2184
    },
    {
      "epoch": 0.36750483558994196,
      "grad_norm": NaN,
      "learning_rate": 7.175469937095818e-05,
      "loss": 0.0,
      "step": 2185
    },
    {
      "epoch": 0.36767303002270624,
      "grad_norm": NaN,
      "learning_rate": 7.173049911232533e-05,
      "loss": 0.0,
      "step": 2186
    },
    {
      "epoch": 0.3678412244554705,
      "grad_norm": NaN,
      "learning_rate": 7.170629257599468e-05,
      "loss": 0.0,
      "step": 2187
    },
    {
      "epoch": 0.3680094188882348,
      "grad_norm": NaN,
      "learning_rate": 7.168207976895922e-05,
      "loss": 0.0,
      "step": 2188
    },
    {
      "epoch": 0.3681776133209991,
      "grad_norm": NaN,
      "learning_rate": 7.165786069821376e-05,
      "loss": 0.0,
      "step": 2189
    },
    {
      "epoch": 0.36834580775376335,
      "grad_norm": NaN,
      "learning_rate": 7.163363537075495e-05,
      "loss": 0.0,
      "step": 2190
    },
    {
      "epoch": 0.36851400218652763,
      "grad_norm": NaN,
      "learning_rate": 7.160940379358117e-05,
      "loss": 0.0,
      "step": 2191
    },
    {
      "epoch": 0.3686821966192919,
      "grad_norm": NaN,
      "learning_rate": 7.158516597369269e-05,
      "loss": 0.0,
      "step": 2192
    },
    {
      "epoch": 0.3688503910520562,
      "grad_norm": NaN,
      "learning_rate": 7.156092191809152e-05,
      "loss": 0.0,
      "step": 2193
    },
    {
      "epoch": 0.36901858548482047,
      "grad_norm": NaN,
      "learning_rate": 7.153667163378151e-05,
      "loss": 0.0,
      "step": 2194
    },
    {
      "epoch": 0.36918677991758475,
      "grad_norm": NaN,
      "learning_rate": 7.151241512776829e-05,
      "loss": 0.0,
      "step": 2195
    },
    {
      "epoch": 0.36935497435034903,
      "grad_norm": NaN,
      "learning_rate": 7.148815240705928e-05,
      "loss": 0.0,
      "step": 2196
    },
    {
      "epoch": 0.36952316878311325,
      "grad_norm": NaN,
      "learning_rate": 7.146388347866372e-05,
      "loss": 0.0,
      "step": 2197
    },
    {
      "epoch": 0.36969136321587753,
      "grad_norm": NaN,
      "learning_rate": 7.143960834959264e-05,
      "loss": 0.0,
      "step": 2198
    },
    {
      "epoch": 0.3698595576486418,
      "grad_norm": NaN,
      "learning_rate": 7.141532702685886e-05,
      "loss": 0.0,
      "step": 2199
    },
    {
      "epoch": 0.3700277520814061,
      "grad_norm": NaN,
      "learning_rate": 7.139103951747695e-05,
      "loss": 0.0,
      "step": 2200
    },
    {
      "epoch": 0.37019594651417037,
      "grad_norm": NaN,
      "learning_rate": 7.136674582846333e-05,
      "loss": 0.0,
      "step": 2201
    },
    {
      "epoch": 0.37036414094693465,
      "grad_norm": NaN,
      "learning_rate": 7.134244596683614e-05,
      "loss": 0.0,
      "step": 2202
    },
    {
      "epoch": 0.37053233537969893,
      "grad_norm": NaN,
      "learning_rate": 7.131813993961537e-05,
      "loss": 0.0,
      "step": 2203
    },
    {
      "epoch": 0.3707005298124632,
      "grad_norm": NaN,
      "learning_rate": 7.129382775382276e-05,
      "loss": 0.0,
      "step": 2204
    },
    {
      "epoch": 0.3708687242452275,
      "grad_norm": NaN,
      "learning_rate": 7.12695094164818e-05,
      "loss": 0.0,
      "step": 2205
    },
    {
      "epoch": 0.37103691867799177,
      "grad_norm": NaN,
      "learning_rate": 7.12451849346178e-05,
      "loss": 0.0,
      "step": 2206
    },
    {
      "epoch": 0.37120511311075605,
      "grad_norm": NaN,
      "learning_rate": 7.122085431525785e-05,
      "loss": 0.0,
      "step": 2207
    },
    {
      "epoch": 0.3713733075435203,
      "grad_norm": NaN,
      "learning_rate": 7.119651756543075e-05,
      "loss": 0.0,
      "step": 2208
    },
    {
      "epoch": 0.3715415019762846,
      "grad_norm": NaN,
      "learning_rate": 7.117217469216716e-05,
      "loss": 0.0,
      "step": 2209
    },
    {
      "epoch": 0.3717096964090489,
      "grad_norm": NaN,
      "learning_rate": 7.114782570249943e-05,
      "loss": 0.0,
      "step": 2210
    },
    {
      "epoch": 0.37187789084181316,
      "grad_norm": NaN,
      "learning_rate": 7.112347060346173e-05,
      "loss": 0.0,
      "step": 2211
    },
    {
      "epoch": 0.3720460852745774,
      "grad_norm": NaN,
      "learning_rate": 7.109910940208999e-05,
      "loss": 0.0,
      "step": 2212
    },
    {
      "epoch": 0.37221427970734167,
      "grad_norm": NaN,
      "learning_rate": 7.107474210542185e-05,
      "loss": 0.0,
      "step": 2213
    },
    {
      "epoch": 0.37238247414010595,
      "grad_norm": NaN,
      "learning_rate": 7.105036872049675e-05,
      "loss": 0.0,
      "step": 2214
    },
    {
      "epoch": 0.3725506685728702,
      "grad_norm": NaN,
      "learning_rate": 7.102598925435593e-05,
      "loss": 0.0,
      "step": 2215
    },
    {
      "epoch": 0.3727188630056345,
      "grad_norm": NaN,
      "learning_rate": 7.100160371404229e-05,
      "loss": 0.0,
      "step": 2216
    },
    {
      "epoch": 0.3728870574383988,
      "grad_norm": NaN,
      "learning_rate": 7.09772121066006e-05,
      "loss": 0.0,
      "step": 2217
    },
    {
      "epoch": 0.37305525187116306,
      "grad_norm": NaN,
      "learning_rate": 7.095281443907726e-05,
      "loss": 0.0,
      "step": 2218
    },
    {
      "epoch": 0.37322344630392734,
      "grad_norm": NaN,
      "learning_rate": 7.092841071852055e-05,
      "loss": 0.0,
      "step": 2219
    },
    {
      "epoch": 0.3733916407366916,
      "grad_norm": NaN,
      "learning_rate": 7.090400095198035e-05,
      "loss": 0.0,
      "step": 2220
    },
    {
      "epoch": 0.3735598351694559,
      "grad_norm": NaN,
      "learning_rate": 7.087958514650846e-05,
      "loss": 0.0,
      "step": 2221
    },
    {
      "epoch": 0.3737280296022202,
      "grad_norm": NaN,
      "learning_rate": 7.085516330915825e-05,
      "loss": 0.0,
      "step": 2222
    },
    {
      "epoch": 0.37389622403498446,
      "grad_norm": NaN,
      "learning_rate": 7.083073544698498e-05,
      "loss": 0.0,
      "step": 2223
    },
    {
      "epoch": 0.37406441846774874,
      "grad_norm": NaN,
      "learning_rate": 7.080630156704553e-05,
      "loss": 0.0,
      "step": 2224
    },
    {
      "epoch": 0.374232612900513,
      "grad_norm": NaN,
      "learning_rate": 7.078186167639862e-05,
      "loss": 0.0,
      "step": 2225
    },
    {
      "epoch": 0.37440080733327724,
      "grad_norm": NaN,
      "learning_rate": 7.075741578210463e-05,
      "loss": 0.0,
      "step": 2226
    },
    {
      "epoch": 0.3745690017660415,
      "grad_norm": NaN,
      "learning_rate": 7.073296389122572e-05,
      "loss": 0.0,
      "step": 2227
    },
    {
      "epoch": 0.3747371961988058,
      "grad_norm": NaN,
      "learning_rate": 7.070850601082577e-05,
      "loss": 0.0,
      "step": 2228
    },
    {
      "epoch": 0.3749053906315701,
      "grad_norm": NaN,
      "learning_rate": 7.068404214797038e-05,
      "loss": 0.0,
      "step": 2229
    },
    {
      "epoch": 0.37507358506433436,
      "grad_norm": NaN,
      "learning_rate": 7.065957230972689e-05,
      "loss": 0.0,
      "step": 2230
    },
    {
      "epoch": 0.37524177949709864,
      "grad_norm": NaN,
      "learning_rate": 7.063509650316435e-05,
      "loss": 0.0,
      "step": 2231
    },
    {
      "epoch": 0.3754099739298629,
      "grad_norm": NaN,
      "learning_rate": 7.061061473535358e-05,
      "loss": 0.0,
      "step": 2232
    },
    {
      "epoch": 0.3755781683626272,
      "grad_norm": NaN,
      "learning_rate": 7.058612701336705e-05,
      "loss": 0.0,
      "step": 2233
    },
    {
      "epoch": 0.3757463627953915,
      "grad_norm": NaN,
      "learning_rate": 7.056163334427899e-05,
      "loss": 0.0,
      "step": 2234
    },
    {
      "epoch": 0.37591455722815575,
      "grad_norm": NaN,
      "learning_rate": 7.053713373516538e-05,
      "loss": 0.0,
      "step": 2235
    },
    {
      "epoch": 0.37608275166092003,
      "grad_norm": NaN,
      "learning_rate": 7.051262819310385e-05,
      "loss": 0.0,
      "step": 2236
    },
    {
      "epoch": 0.3762509460936843,
      "grad_norm": NaN,
      "learning_rate": 7.048811672517379e-05,
      "loss": 0.0,
      "step": 2237
    },
    {
      "epoch": 0.3764191405264486,
      "grad_norm": NaN,
      "learning_rate": 7.04635993384563e-05,
      "loss": 0.0,
      "step": 2238
    },
    {
      "epoch": 0.37658733495921287,
      "grad_norm": NaN,
      "learning_rate": 7.043907604003415e-05,
      "loss": 0.0,
      "step": 2239
    },
    {
      "epoch": 0.37675552939197715,
      "grad_norm": NaN,
      "learning_rate": 7.041454683699189e-05,
      "loss": 0.0,
      "step": 2240
    },
    {
      "epoch": 0.3769237238247414,
      "grad_norm": NaN,
      "learning_rate": 7.039001173641568e-05,
      "loss": 0.0,
      "step": 2241
    },
    {
      "epoch": 0.37709191825750565,
      "grad_norm": NaN,
      "learning_rate": 7.036547074539347e-05,
      "loss": 0.0,
      "step": 2242
    },
    {
      "epoch": 0.37726011269026993,
      "grad_norm": NaN,
      "learning_rate": 7.034092387101487e-05,
      "loss": 0.0,
      "step": 2243
    },
    {
      "epoch": 0.3774283071230342,
      "grad_norm": NaN,
      "learning_rate": 7.031637112037118e-05,
      "loss": 0.0,
      "step": 2244
    },
    {
      "epoch": 0.3775965015557985,
      "grad_norm": NaN,
      "learning_rate": 7.029181250055545e-05,
      "loss": 0.0,
      "step": 2245
    },
    {
      "epoch": 0.37776469598856277,
      "grad_norm": NaN,
      "learning_rate": 7.026724801866235e-05,
      "loss": 0.0,
      "step": 2246
    },
    {
      "epoch": 0.37793289042132705,
      "grad_norm": NaN,
      "learning_rate": 7.024267768178831e-05,
      "loss": 0.0,
      "step": 2247
    },
    {
      "epoch": 0.37810108485409133,
      "grad_norm": NaN,
      "learning_rate": 7.021810149703143e-05,
      "loss": 0.0,
      "step": 2248
    },
    {
      "epoch": 0.3782692792868556,
      "grad_norm": NaN,
      "learning_rate": 7.019351947149148e-05,
      "loss": 0.0,
      "step": 2249
    },
    {
      "epoch": 0.3784374737196199,
      "grad_norm": NaN,
      "learning_rate": 7.016893161226994e-05,
      "loss": 0.0,
      "step": 2250
    },
    {
      "epoch": 0.37860566815238417,
      "grad_norm": NaN,
      "learning_rate": 7.014433792646995e-05,
      "loss": 0.0,
      "step": 2251
    },
    {
      "epoch": 0.37877386258514845,
      "grad_norm": NaN,
      "learning_rate": 7.011973842119637e-05,
      "loss": 0.0,
      "step": 2252
    },
    {
      "epoch": 0.3789420570179127,
      "grad_norm": NaN,
      "learning_rate": 7.009513310355571e-05,
      "loss": 0.0,
      "step": 2253
    },
    {
      "epoch": 0.379110251450677,
      "grad_norm": NaN,
      "learning_rate": 7.007052198065618e-05,
      "loss": 0.0,
      "step": 2254
    },
    {
      "epoch": 0.3792784458834413,
      "grad_norm": NaN,
      "learning_rate": 7.004590505960765e-05,
      "loss": 0.0,
      "step": 2255
    },
    {
      "epoch": 0.3794466403162055,
      "grad_norm": NaN,
      "learning_rate": 7.002128234752166e-05,
      "loss": 0.0,
      "step": 2256
    },
    {
      "epoch": 0.3796148347489698,
      "grad_norm": NaN,
      "learning_rate": 6.999665385151147e-05,
      "loss": 0.0,
      "step": 2257
    },
    {
      "epoch": 0.37978302918173407,
      "grad_norm": NaN,
      "learning_rate": 6.997201957869195e-05,
      "loss": 0.0,
      "step": 2258
    },
    {
      "epoch": 0.37995122361449835,
      "grad_norm": NaN,
      "learning_rate": 6.994737953617968e-05,
      "loss": 0.0,
      "step": 2259
    },
    {
      "epoch": 0.3801194180472626,
      "grad_norm": NaN,
      "learning_rate": 6.992273373109288e-05,
      "loss": 0.0,
      "step": 2260
    },
    {
      "epoch": 0.3802876124800269,
      "grad_norm": NaN,
      "learning_rate": 6.989808217055145e-05,
      "loss": 0.0,
      "step": 2261
    },
    {
      "epoch": 0.3804558069127912,
      "grad_norm": NaN,
      "learning_rate": 6.987342486167697e-05,
      "loss": 0.0,
      "step": 2262
    },
    {
      "epoch": 0.38062400134555546,
      "grad_norm": NaN,
      "learning_rate": 6.984876181159261e-05,
      "loss": 0.0,
      "step": 2263
    },
    {
      "epoch": 0.38079219577831974,
      "grad_norm": NaN,
      "learning_rate": 6.98240930274233e-05,
      "loss": 0.0,
      "step": 2264
    },
    {
      "epoch": 0.380960390211084,
      "grad_norm": NaN,
      "learning_rate": 6.979941851629555e-05,
      "loss": 0.0,
      "step": 2265
    },
    {
      "epoch": 0.3811285846438483,
      "grad_norm": NaN,
      "learning_rate": 6.977473828533757e-05,
      "loss": 0.0,
      "step": 2266
    },
    {
      "epoch": 0.3812967790766126,
      "grad_norm": NaN,
      "learning_rate": 6.975005234167917e-05,
      "loss": 0.0,
      "step": 2267
    },
    {
      "epoch": 0.38146497350937686,
      "grad_norm": NaN,
      "learning_rate": 6.972536069245189e-05,
      "loss": 0.0,
      "step": 2268
    },
    {
      "epoch": 0.38163316794214114,
      "grad_norm": NaN,
      "learning_rate": 6.970066334478882e-05,
      "loss": 0.0,
      "step": 2269
    },
    {
      "epoch": 0.3818013623749054,
      "grad_norm": NaN,
      "learning_rate": 6.967596030582478e-05,
      "loss": 0.0,
      "step": 2270
    },
    {
      "epoch": 0.38196955680766964,
      "grad_norm": NaN,
      "learning_rate": 6.965125158269619e-05,
      "loss": 0.0,
      "step": 2271
    },
    {
      "epoch": 0.3821377512404339,
      "grad_norm": NaN,
      "learning_rate": 6.962653718254111e-05,
      "loss": 0.0,
      "step": 2272
    },
    {
      "epoch": 0.3823059456731982,
      "grad_norm": NaN,
      "learning_rate": 6.960181711249929e-05,
      "loss": 0.0,
      "step": 2273
    },
    {
      "epoch": 0.3824741401059625,
      "grad_norm": NaN,
      "learning_rate": 6.957709137971204e-05,
      "loss": 0.0,
      "step": 2274
    },
    {
      "epoch": 0.38264233453872676,
      "grad_norm": NaN,
      "learning_rate": 6.955235999132238e-05,
      "loss": 0.0,
      "step": 2275
    },
    {
      "epoch": 0.38281052897149104,
      "grad_norm": NaN,
      "learning_rate": 6.952762295447491e-05,
      "loss": 0.0,
      "step": 2276
    },
    {
      "epoch": 0.3829787234042553,
      "grad_norm": NaN,
      "learning_rate": 6.950288027631588e-05,
      "loss": 0.0,
      "step": 2277
    },
    {
      "epoch": 0.3831469178370196,
      "grad_norm": NaN,
      "learning_rate": 6.947813196399321e-05,
      "loss": 0.0,
      "step": 2278
    },
    {
      "epoch": 0.3833151122697839,
      "grad_norm": NaN,
      "learning_rate": 6.945337802465636e-05,
      "loss": 0.0,
      "step": 2279
    },
    {
      "epoch": 0.38348330670254815,
      "grad_norm": NaN,
      "learning_rate": 6.942861846545651e-05,
      "loss": 0.0,
      "step": 2280
    },
    {
      "epoch": 0.38365150113531243,
      "grad_norm": NaN,
      "learning_rate": 6.940385329354639e-05,
      "loss": 0.0,
      "step": 2281
    },
    {
      "epoch": 0.3838196955680767,
      "grad_norm": NaN,
      "learning_rate": 6.937908251608038e-05,
      "loss": 0.0,
      "step": 2282
    },
    {
      "epoch": 0.383987890000841,
      "grad_norm": NaN,
      "learning_rate": 6.93543061402145e-05,
      "loss": 0.0,
      "step": 2283
    },
    {
      "epoch": 0.38415608443360527,
      "grad_norm": NaN,
      "learning_rate": 6.932952417310634e-05,
      "loss": 0.0,
      "step": 2284
    },
    {
      "epoch": 0.38432427886636955,
      "grad_norm": NaN,
      "learning_rate": 6.930473662191517e-05,
      "loss": 0.0,
      "step": 2285
    },
    {
      "epoch": 0.3844924732991338,
      "grad_norm": NaN,
      "learning_rate": 6.92799434938018e-05,
      "loss": 0.0,
      "step": 2286
    },
    {
      "epoch": 0.38466066773189805,
      "grad_norm": NaN,
      "learning_rate": 6.925514479592872e-05,
      "loss": 0.0,
      "step": 2287
    },
    {
      "epoch": 0.38482886216466233,
      "grad_norm": NaN,
      "learning_rate": 6.923034053545996e-05,
      "loss": 0.0,
      "step": 2288
    },
    {
      "epoch": 0.3849970565974266,
      "grad_norm": NaN,
      "learning_rate": 6.920553071956122e-05,
      "loss": 0.0,
      "step": 2289
    },
    {
      "epoch": 0.3851652510301909,
      "grad_norm": NaN,
      "learning_rate": 6.918071535539978e-05,
      "loss": 0.0,
      "step": 2290
    },
    {
      "epoch": 0.38533344546295517,
      "grad_norm": NaN,
      "learning_rate": 6.915589445014448e-05,
      "loss": 0.0,
      "step": 2291
    },
    {
      "epoch": 0.38550163989571945,
      "grad_norm": NaN,
      "learning_rate": 6.913106801096586e-05,
      "loss": 0.0,
      "step": 2292
    },
    {
      "epoch": 0.38566983432848373,
      "grad_norm": NaN,
      "learning_rate": 6.910623604503596e-05,
      "loss": 0.0,
      "step": 2293
    },
    {
      "epoch": 0.385838028761248,
      "grad_norm": NaN,
      "learning_rate": 6.908139855952846e-05,
      "loss": 0.0,
      "step": 2294
    },
    {
      "epoch": 0.3860062231940123,
      "grad_norm": NaN,
      "learning_rate": 6.905655556161865e-05,
      "loss": 0.0,
      "step": 2295
    },
    {
      "epoch": 0.38617441762677657,
      "grad_norm": NaN,
      "learning_rate": 6.903170705848339e-05,
      "loss": 0.0,
      "step": 2296
    },
    {
      "epoch": 0.38634261205954085,
      "grad_norm": NaN,
      "learning_rate": 6.900685305730112e-05,
      "loss": 0.0,
      "step": 2297
    },
    {
      "epoch": 0.3865108064923051,
      "grad_norm": NaN,
      "learning_rate": 6.89819935652519e-05,
      "loss": 0.0,
      "step": 2298
    },
    {
      "epoch": 0.3866790009250694,
      "grad_norm": NaN,
      "learning_rate": 6.895712858951735e-05,
      "loss": 0.0,
      "step": 2299
    },
    {
      "epoch": 0.38684719535783363,
      "grad_norm": NaN,
      "learning_rate": 6.893225813728067e-05,
      "loss": 0.0,
      "step": 2300
    },
    {
      "epoch": 0.3870153897905979,
      "grad_norm": NaN,
      "learning_rate": 6.890738221572668e-05,
      "loss": 0.0,
      "step": 2301
    },
    {
      "epoch": 0.3871835842233622,
      "grad_norm": NaN,
      "learning_rate": 6.888250083204173e-05,
      "loss": 0.0,
      "step": 2302
    },
    {
      "epoch": 0.38735177865612647,
      "grad_norm": NaN,
      "learning_rate": 6.885761399341379e-05,
      "loss": 0.0,
      "step": 2303
    },
    {
      "epoch": 0.38751997308889075,
      "grad_norm": NaN,
      "learning_rate": 6.883272170703237e-05,
      "loss": 0.0,
      "step": 2304
    },
    {
      "epoch": 0.387688167521655,
      "grad_norm": NaN,
      "learning_rate": 6.880782398008862e-05,
      "loss": 0.0,
      "step": 2305
    },
    {
      "epoch": 0.3878563619544193,
      "grad_norm": NaN,
      "learning_rate": 6.878292081977516e-05,
      "loss": 0.0,
      "step": 2306
    },
    {
      "epoch": 0.3880245563871836,
      "grad_norm": NaN,
      "learning_rate": 6.875801223328628e-05,
      "loss": 0.0,
      "step": 2307
    },
    {
      "epoch": 0.38819275081994786,
      "grad_norm": NaN,
      "learning_rate": 6.873309822781773e-05,
      "loss": 0.0,
      "step": 2308
    },
    {
      "epoch": 0.38836094525271214,
      "grad_norm": NaN,
      "learning_rate": 6.870817881056695e-05,
      "loss": 0.0,
      "step": 2309
    },
    {
      "epoch": 0.3885291396854764,
      "grad_norm": NaN,
      "learning_rate": 6.868325398873284e-05,
      "loss": 0.0,
      "step": 2310
    },
    {
      "epoch": 0.3886973341182407,
      "grad_norm": NaN,
      "learning_rate": 6.865832376951594e-05,
      "loss": 0.0,
      "step": 2311
    },
    {
      "epoch": 0.388865528551005,
      "grad_norm": NaN,
      "learning_rate": 6.863338816011826e-05,
      "loss": 0.0,
      "step": 2312
    },
    {
      "epoch": 0.38903372298376926,
      "grad_norm": NaN,
      "learning_rate": 6.860844716774346e-05,
      "loss": 0.0,
      "step": 2313
    },
    {
      "epoch": 0.38920191741653354,
      "grad_norm": NaN,
      "learning_rate": 6.85835007995967e-05,
      "loss": 0.0,
      "step": 2314
    },
    {
      "epoch": 0.38937011184929776,
      "grad_norm": NaN,
      "learning_rate": 6.855854906288471e-05,
      "loss": 0.0,
      "step": 2315
    },
    {
      "epoch": 0.38953830628206204,
      "grad_norm": NaN,
      "learning_rate": 6.853359196481576e-05,
      "loss": 0.0,
      "step": 2316
    },
    {
      "epoch": 0.3897065007148263,
      "grad_norm": NaN,
      "learning_rate": 6.850862951259968e-05,
      "loss": 0.0,
      "step": 2317
    },
    {
      "epoch": 0.3898746951475906,
      "grad_norm": NaN,
      "learning_rate": 6.848366171344784e-05,
      "loss": 0.0,
      "step": 2318
    },
    {
      "epoch": 0.3900428895803549,
      "grad_norm": NaN,
      "learning_rate": 6.845868857457316e-05,
      "loss": 0.0,
      "step": 2319
    },
    {
      "epoch": 0.39021108401311916,
      "grad_norm": NaN,
      "learning_rate": 6.84337101031901e-05,
      "loss": 0.0,
      "step": 2320
    },
    {
      "epoch": 0.39037927844588344,
      "grad_norm": NaN,
      "learning_rate": 6.840872630651467e-05,
      "loss": 0.0,
      "step": 2321
    },
    {
      "epoch": 0.3905474728786477,
      "grad_norm": NaN,
      "learning_rate": 6.838373719176439e-05,
      "loss": 0.0,
      "step": 2322
    },
    {
      "epoch": 0.390715667311412,
      "grad_norm": NaN,
      "learning_rate": 6.835874276615835e-05,
      "loss": 0.0,
      "step": 2323
    },
    {
      "epoch": 0.3908838617441763,
      "grad_norm": NaN,
      "learning_rate": 6.833374303691714e-05,
      "loss": 0.0,
      "step": 2324
    },
    {
      "epoch": 0.39105205617694055,
      "grad_norm": NaN,
      "learning_rate": 6.830873801126293e-05,
      "loss": 0.0,
      "step": 2325
    },
    {
      "epoch": 0.39122025060970483,
      "grad_norm": NaN,
      "learning_rate": 6.828372769641938e-05,
      "loss": 0.0,
      "step": 2326
    },
    {
      "epoch": 0.3913884450424691,
      "grad_norm": NaN,
      "learning_rate": 6.825871209961168e-05,
      "loss": 0.0,
      "step": 2327
    },
    {
      "epoch": 0.3915566394752334,
      "grad_norm": NaN,
      "learning_rate": 6.823369122806656e-05,
      "loss": 0.0,
      "step": 2328
    },
    {
      "epoch": 0.39172483390799767,
      "grad_norm": NaN,
      "learning_rate": 6.820866508901229e-05,
      "loss": 0.0,
      "step": 2329
    },
    {
      "epoch": 0.3918930283407619,
      "grad_norm": NaN,
      "learning_rate": 6.818363368967862e-05,
      "loss": 0.0,
      "step": 2330
    },
    {
      "epoch": 0.3920612227735262,
      "grad_norm": NaN,
      "learning_rate": 6.815859703729683e-05,
      "loss": 0.0,
      "step": 2331
    },
    {
      "epoch": 0.39222941720629045,
      "grad_norm": NaN,
      "learning_rate": 6.813355513909976e-05,
      "loss": 0.0,
      "step": 2332
    },
    {
      "epoch": 0.39239761163905473,
      "grad_norm": NaN,
      "learning_rate": 6.810850800232172e-05,
      "loss": 0.0,
      "step": 2333
    },
    {
      "epoch": 0.392565806071819,
      "grad_norm": NaN,
      "learning_rate": 6.808345563419853e-05,
      "loss": 0.0,
      "step": 2334
    },
    {
      "epoch": 0.3927340005045833,
      "grad_norm": NaN,
      "learning_rate": 6.805839804196757e-05,
      "loss": 0.0,
      "step": 2335
    },
    {
      "epoch": 0.39290219493734757,
      "grad_norm": NaN,
      "learning_rate": 6.803333523286766e-05,
      "loss": 0.0,
      "step": 2336
    },
    {
      "epoch": 0.39307038937011185,
      "grad_norm": NaN,
      "learning_rate": 6.800826721413919e-05,
      "loss": 0.0,
      "step": 2337
    },
    {
      "epoch": 0.39323858380287613,
      "grad_norm": NaN,
      "learning_rate": 6.798319399302404e-05,
      "loss": 0.0,
      "step": 2338
    },
    {
      "epoch": 0.3934067782356404,
      "grad_norm": NaN,
      "learning_rate": 6.795811557676557e-05,
      "loss": 0.0,
      "step": 2339
    },
    {
      "epoch": 0.3935749726684047,
      "grad_norm": NaN,
      "learning_rate": 6.793303197260864e-05,
      "loss": 0.0,
      "step": 2340
    },
    {
      "epoch": 0.39374316710116897,
      "grad_norm": NaN,
      "learning_rate": 6.790794318779964e-05,
      "loss": 0.0,
      "step": 2341
    },
    {
      "epoch": 0.39391136153393325,
      "grad_norm": NaN,
      "learning_rate": 6.788284922958643e-05,
      "loss": 0.0,
      "step": 2342
    },
    {
      "epoch": 0.3940795559666975,
      "grad_norm": NaN,
      "learning_rate": 6.785775010521837e-05,
      "loss": 0.0,
      "step": 2343
    },
    {
      "epoch": 0.3942477503994618,
      "grad_norm": NaN,
      "learning_rate": 6.783264582194635e-05,
      "loss": 0.0,
      "step": 2344
    },
    {
      "epoch": 0.39441594483222603,
      "grad_norm": NaN,
      "learning_rate": 6.780753638702267e-05,
      "loss": 0.0,
      "step": 2345
    },
    {
      "epoch": 0.3945841392649903,
      "grad_norm": NaN,
      "learning_rate": 6.778242180770118e-05,
      "loss": 0.0,
      "step": 2346
    },
    {
      "epoch": 0.3947523336977546,
      "grad_norm": NaN,
      "learning_rate": 6.775730209123722e-05,
      "loss": 0.0,
      "step": 2347
    },
    {
      "epoch": 0.39492052813051887,
      "grad_norm": NaN,
      "learning_rate": 6.773217724488756e-05,
      "loss": 0.0,
      "step": 2348
    },
    {
      "epoch": 0.39508872256328315,
      "grad_norm": NaN,
      "learning_rate": 6.770704727591053e-05,
      "loss": 0.0,
      "step": 2349
    },
    {
      "epoch": 0.3952569169960474,
      "grad_norm": NaN,
      "learning_rate": 6.768191219156586e-05,
      "loss": 0.0,
      "step": 2350
    },
    {
      "epoch": 0.3954251114288117,
      "grad_norm": NaN,
      "learning_rate": 6.765677199911479e-05,
      "loss": 0.0,
      "step": 2351
    },
    {
      "epoch": 0.395593305861576,
      "grad_norm": NaN,
      "learning_rate": 6.763162670582007e-05,
      "loss": 0.0,
      "step": 2352
    },
    {
      "epoch": 0.39576150029434026,
      "grad_norm": NaN,
      "learning_rate": 6.760647631894589e-05,
      "loss": 0.0,
      "step": 2353
    },
    {
      "epoch": 0.39592969472710454,
      "grad_norm": NaN,
      "learning_rate": 6.75813208457579e-05,
      "loss": 0.0,
      "step": 2354
    },
    {
      "epoch": 0.3960978891598688,
      "grad_norm": NaN,
      "learning_rate": 6.755616029352323e-05,
      "loss": 0.0,
      "step": 2355
    },
    {
      "epoch": 0.3962660835926331,
      "grad_norm": NaN,
      "learning_rate": 6.753099466951049e-05,
      "loss": 0.0,
      "step": 2356
    },
    {
      "epoch": 0.3964342780253974,
      "grad_norm": NaN,
      "learning_rate": 6.750582398098976e-05,
      "loss": 0.0,
      "step": 2357
    },
    {
      "epoch": 0.39660247245816166,
      "grad_norm": NaN,
      "learning_rate": 6.748064823523255e-05,
      "loss": 0.0,
      "step": 2358
    },
    {
      "epoch": 0.3967706668909259,
      "grad_norm": NaN,
      "learning_rate": 6.745546743951187e-05,
      "loss": 0.0,
      "step": 2359
    },
    {
      "epoch": 0.39693886132369016,
      "grad_norm": NaN,
      "learning_rate": 6.743028160110215e-05,
      "loss": 0.0,
      "step": 2360
    },
    {
      "epoch": 0.39710705575645444,
      "grad_norm": NaN,
      "learning_rate": 6.740509072727931e-05,
      "loss": 0.0,
      "step": 2361
    },
    {
      "epoch": 0.3972752501892187,
      "grad_norm": NaN,
      "learning_rate": 6.73798948253207e-05,
      "loss": 0.0,
      "step": 2362
    },
    {
      "epoch": 0.397443444621983,
      "grad_norm": NaN,
      "learning_rate": 6.735469390250515e-05,
      "loss": 0.0,
      "step": 2363
    },
    {
      "epoch": 0.3976116390547473,
      "grad_norm": NaN,
      "learning_rate": 6.732948796611292e-05,
      "loss": 0.0,
      "step": 2364
    },
    {
      "epoch": 0.39777983348751156,
      "grad_norm": NaN,
      "learning_rate": 6.73042770234257e-05,
      "loss": 0.0,
      "step": 2365
    },
    {
      "epoch": 0.39794802792027584,
      "grad_norm": NaN,
      "learning_rate": 6.727906108172668e-05,
      "loss": 0.0,
      "step": 2366
    },
    {
      "epoch": 0.3981162223530401,
      "grad_norm": NaN,
      "learning_rate": 6.725384014830045e-05,
      "loss": 0.0,
      "step": 2367
    },
    {
      "epoch": 0.3982844167858044,
      "grad_norm": NaN,
      "learning_rate": 6.722861423043305e-05,
      "loss": 0.0,
      "step": 2368
    },
    {
      "epoch": 0.3984526112185687,
      "grad_norm": NaN,
      "learning_rate": 6.720338333541196e-05,
      "loss": 0.0,
      "step": 2369
    },
    {
      "epoch": 0.39862080565133295,
      "grad_norm": NaN,
      "learning_rate": 6.717814747052613e-05,
      "loss": 0.0,
      "step": 2370
    },
    {
      "epoch": 0.39878900008409723,
      "grad_norm": NaN,
      "learning_rate": 6.715290664306589e-05,
      "loss": 0.0,
      "step": 2371
    },
    {
      "epoch": 0.3989571945168615,
      "grad_norm": NaN,
      "learning_rate": 6.712766086032306e-05,
      "loss": 0.0,
      "step": 2372
    },
    {
      "epoch": 0.3991253889496258,
      "grad_norm": NaN,
      "learning_rate": 6.710241012959085e-05,
      "loss": 0.0,
      "step": 2373
    },
    {
      "epoch": 0.39929358338239,
      "grad_norm": NaN,
      "learning_rate": 6.70771544581639e-05,
      "loss": 0.0,
      "step": 2374
    },
    {
      "epoch": 0.3994617778151543,
      "grad_norm": NaN,
      "learning_rate": 6.70518938533383e-05,
      "loss": 0.0,
      "step": 2375
    },
    {
      "epoch": 0.3996299722479186,
      "grad_norm": NaN,
      "learning_rate": 6.702662832241156e-05,
      "loss": 0.0,
      "step": 2376
    },
    {
      "epoch": 0.39979816668068285,
      "grad_norm": NaN,
      "learning_rate": 6.70013578726826e-05,
      "loss": 0.0,
      "step": 2377
    },
    {
      "epoch": 0.39996636111344713,
      "grad_norm": NaN,
      "learning_rate": 6.69760825114518e-05,
      "loss": 0.0,
      "step": 2378
    },
    {
      "epoch": 0.4001345555462114,
      "grad_norm": NaN,
      "learning_rate": 6.695080224602088e-05,
      "loss": 0.0,
      "step": 2379
    },
    {
      "epoch": 0.4003027499789757,
      "grad_norm": NaN,
      "learning_rate": 6.692551708369307e-05,
      "loss": 0.0,
      "step": 2380
    },
    {
      "epoch": 0.40047094441173997,
      "grad_norm": NaN,
      "learning_rate": 6.690022703177294e-05,
      "loss": 0.0,
      "step": 2381
    },
    {
      "epoch": 0.40063913884450425,
      "grad_norm": NaN,
      "learning_rate": 6.687493209756653e-05,
      "loss": 0.0,
      "step": 2382
    },
    {
      "epoch": 0.40080733327726853,
      "grad_norm": NaN,
      "learning_rate": 6.684963228838122e-05,
      "loss": 0.0,
      "step": 2383
    },
    {
      "epoch": 0.4009755277100328,
      "grad_norm": NaN,
      "learning_rate": 6.682432761152589e-05,
      "loss": 0.0,
      "step": 2384
    },
    {
      "epoch": 0.4011437221427971,
      "grad_norm": NaN,
      "learning_rate": 6.679901807431073e-05,
      "loss": 0.0,
      "step": 2385
    },
    {
      "epoch": 0.40131191657556137,
      "grad_norm": NaN,
      "learning_rate": 6.677370368404744e-05,
      "loss": 0.0,
      "step": 2386
    },
    {
      "epoch": 0.40148011100832565,
      "grad_norm": NaN,
      "learning_rate": 6.674838444804902e-05,
      "loss": 0.0,
      "step": 2387
    },
    {
      "epoch": 0.4016483054410899,
      "grad_norm": NaN,
      "learning_rate": 6.672306037362991e-05,
      "loss": 0.0,
      "step": 2388
    },
    {
      "epoch": 0.40181649987385415,
      "grad_norm": NaN,
      "learning_rate": 6.669773146810599e-05,
      "loss": 0.0,
      "step": 2389
    },
    {
      "epoch": 0.40198469430661843,
      "grad_norm": NaN,
      "learning_rate": 6.667239773879447e-05,
      "loss": 0.0,
      "step": 2390
    },
    {
      "epoch": 0.4021528887393827,
      "grad_norm": NaN,
      "learning_rate": 6.664705919301399e-05,
      "loss": 0.0,
      "step": 2391
    },
    {
      "epoch": 0.402321083172147,
      "grad_norm": NaN,
      "learning_rate": 6.662171583808455e-05,
      "loss": 0.0,
      "step": 2392
    },
    {
      "epoch": 0.40248927760491127,
      "grad_norm": NaN,
      "learning_rate": 6.659636768132759e-05,
      "loss": 0.0,
      "step": 2393
    },
    {
      "epoch": 0.40265747203767555,
      "grad_norm": NaN,
      "learning_rate": 6.657101473006588e-05,
      "loss": 0.0,
      "step": 2394
    },
    {
      "epoch": 0.4028256664704398,
      "grad_norm": NaN,
      "learning_rate": 6.654565699162364e-05,
      "loss": 0.0,
      "step": 2395
    },
    {
      "epoch": 0.4029938609032041,
      "grad_norm": NaN,
      "learning_rate": 6.652029447332641e-05,
      "loss": 0.0,
      "step": 2396
    },
    {
      "epoch": 0.4031620553359684,
      "grad_norm": NaN,
      "learning_rate": 6.649492718250115e-05,
      "loss": 0.0,
      "step": 2397
    },
    {
      "epoch": 0.40333024976873266,
      "grad_norm": NaN,
      "learning_rate": 6.646955512647616e-05,
      "loss": 0.0,
      "step": 2398
    },
    {
      "epoch": 0.40349844420149694,
      "grad_norm": NaN,
      "learning_rate": 6.64441783125812e-05,
      "loss": 0.0,
      "step": 2399
    },
    {
      "epoch": 0.4036666386342612,
      "grad_norm": NaN,
      "learning_rate": 6.641879674814729e-05,
      "loss": 0.0,
      "step": 2400
    },
    {
      "epoch": 0.4038348330670255,
      "grad_norm": NaN,
      "learning_rate": 6.63934104405069e-05,
      "loss": 0.0,
      "step": 2401
    },
    {
      "epoch": 0.4040030274997898,
      "grad_norm": NaN,
      "learning_rate": 6.636801939699384e-05,
      "loss": 0.0,
      "step": 2402
    },
    {
      "epoch": 0.40417122193255406,
      "grad_norm": NaN,
      "learning_rate": 6.634262362494332e-05,
      "loss": 0.0,
      "step": 2403
    },
    {
      "epoch": 0.4043394163653183,
      "grad_norm": NaN,
      "learning_rate": 6.631722313169188e-05,
      "loss": 0.0,
      "step": 2404
    },
    {
      "epoch": 0.40450761079808256,
      "grad_norm": NaN,
      "learning_rate": 6.629181792457745e-05,
      "loss": 0.0,
      "step": 2405
    },
    {
      "epoch": 0.40467580523084684,
      "grad_norm": NaN,
      "learning_rate": 6.626640801093929e-05,
      "loss": 0.0,
      "step": 2406
    },
    {
      "epoch": 0.4048439996636111,
      "grad_norm": NaN,
      "learning_rate": 6.624099339811805e-05,
      "loss": 0.0,
      "step": 2407
    },
    {
      "epoch": 0.4050121940963754,
      "grad_norm": NaN,
      "learning_rate": 6.621557409345572e-05,
      "loss": 0.0,
      "step": 2408
    },
    {
      "epoch": 0.4051803885291397,
      "grad_norm": NaN,
      "learning_rate": 6.619015010429568e-05,
      "loss": 0.0,
      "step": 2409
    },
    {
      "epoch": 0.40534858296190396,
      "grad_norm": NaN,
      "learning_rate": 6.616472143798261e-05,
      "loss": 0.0,
      "step": 2410
    },
    {
      "epoch": 0.40551677739466824,
      "grad_norm": NaN,
      "learning_rate": 6.613928810186257e-05,
      "loss": 0.0,
      "step": 2411
    },
    {
      "epoch": 0.4056849718274325,
      "grad_norm": NaN,
      "learning_rate": 6.611385010328294e-05,
      "loss": 0.0,
      "step": 2412
    },
    {
      "epoch": 0.4058531662601968,
      "grad_norm": NaN,
      "learning_rate": 6.608840744959255e-05,
      "loss": 0.0,
      "step": 2413
    },
    {
      "epoch": 0.4060213606929611,
      "grad_norm": NaN,
      "learning_rate": 6.60629601481414e-05,
      "loss": 0.0,
      "step": 2414
    },
    {
      "epoch": 0.40618955512572535,
      "grad_norm": NaN,
      "learning_rate": 6.603750820628102e-05,
      "loss": 0.0,
      "step": 2415
    },
    {
      "epoch": 0.40635774955848963,
      "grad_norm": NaN,
      "learning_rate": 6.601205163136412e-05,
      "loss": 0.0,
      "step": 2416
    },
    {
      "epoch": 0.4065259439912539,
      "grad_norm": NaN,
      "learning_rate": 6.598659043074487e-05,
      "loss": 0.0,
      "step": 2417
    },
    {
      "epoch": 0.4066941384240182,
      "grad_norm": NaN,
      "learning_rate": 6.596112461177869e-05,
      "loss": 0.0,
      "step": 2418
    },
    {
      "epoch": 0.4068623328567824,
      "grad_norm": NaN,
      "learning_rate": 6.59356541818224e-05,
      "loss": 0.0,
      "step": 2419
    },
    {
      "epoch": 0.4070305272895467,
      "grad_norm": NaN,
      "learning_rate": 6.591017914823409e-05,
      "loss": 0.0,
      "step": 2420
    },
    {
      "epoch": 0.407198721722311,
      "grad_norm": NaN,
      "learning_rate": 6.588469951837323e-05,
      "loss": 0.0,
      "step": 2421
    },
    {
      "epoch": 0.40736691615507525,
      "grad_norm": NaN,
      "learning_rate": 6.58592152996006e-05,
      "loss": 0.0,
      "step": 2422
    },
    {
      "epoch": 0.40753511058783953,
      "grad_norm": NaN,
      "learning_rate": 6.58337264992783e-05,
      "loss": 0.0,
      "step": 2423
    },
    {
      "epoch": 0.4077033050206038,
      "grad_norm": NaN,
      "learning_rate": 6.580823312476976e-05,
      "loss": 0.0,
      "step": 2424
    },
    {
      "epoch": 0.4078714994533681,
      "grad_norm": NaN,
      "learning_rate": 6.578273518343975e-05,
      "loss": 0.0,
      "step": 2425
    },
    {
      "epoch": 0.40803969388613237,
      "grad_norm": NaN,
      "learning_rate": 6.57572326826543e-05,
      "loss": 0.0,
      "step": 2426
    },
    {
      "epoch": 0.40820788831889665,
      "grad_norm": NaN,
      "learning_rate": 6.573172562978084e-05,
      "loss": 0.0,
      "step": 2427
    },
    {
      "epoch": 0.40837608275166093,
      "grad_norm": NaN,
      "learning_rate": 6.570621403218803e-05,
      "loss": 0.0,
      "step": 2428
    },
    {
      "epoch": 0.4085442771844252,
      "grad_norm": NaN,
      "learning_rate": 6.568069789724593e-05,
      "loss": 0.0,
      "step": 2429
    },
    {
      "epoch": 0.4087124716171895,
      "grad_norm": NaN,
      "learning_rate": 6.565517723232583e-05,
      "loss": 0.0,
      "step": 2430
    },
    {
      "epoch": 0.40888066604995377,
      "grad_norm": NaN,
      "learning_rate": 6.562965204480039e-05,
      "loss": 0.0,
      "step": 2431
    },
    {
      "epoch": 0.40904886048271805,
      "grad_norm": NaN,
      "learning_rate": 6.560412234204351e-05,
      "loss": 0.0,
      "step": 2432
    },
    {
      "epoch": 0.40921705491548227,
      "grad_norm": NaN,
      "learning_rate": 6.557858813143048e-05,
      "loss": 0.0,
      "step": 2433
    },
    {
      "epoch": 0.40938524934824655,
      "grad_norm": NaN,
      "learning_rate": 6.55530494203378e-05,
      "loss": 0.0,
      "step": 2434
    },
    {
      "epoch": 0.40955344378101083,
      "grad_norm": NaN,
      "learning_rate": 6.552750621614337e-05,
      "loss": 0.0,
      "step": 2435
    },
    {
      "epoch": 0.4097216382137751,
      "grad_norm": NaN,
      "learning_rate": 6.550195852622631e-05,
      "loss": 0.0,
      "step": 2436
    },
    {
      "epoch": 0.4098898326465394,
      "grad_norm": NaN,
      "learning_rate": 6.547640635796707e-05,
      "loss": 0.0,
      "step": 2437
    },
    {
      "epoch": 0.41005802707930367,
      "grad_norm": NaN,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0,
      "step": 2438
    },
    {
      "epoch": 0.41022622151206795,
      "grad_norm": NaN,
      "learning_rate": 6.542528861595025e-05,
      "loss": 0.0,
      "step": 2439
    },
    {
      "epoch": 0.4103944159448322,
      "grad_norm": NaN,
      "learning_rate": 6.539972305696002e-05,
      "loss": 0.0,
      "step": 2440
    },
    {
      "epoch": 0.4105626103775965,
      "grad_norm": NaN,
      "learning_rate": 6.537415304916231e-05,
      "loss": 0.0,
      "step": 2441
    },
    {
      "epoch": 0.4107308048103608,
      "grad_norm": NaN,
      "learning_rate": 6.534857859994394e-05,
      "loss": 0.0,
      "step": 2442
    },
    {
      "epoch": 0.41089899924312506,
      "grad_norm": NaN,
      "learning_rate": 6.532299971669316e-05,
      "loss": 0.0,
      "step": 2443
    },
    {
      "epoch": 0.41106719367588934,
      "grad_norm": NaN,
      "learning_rate": 6.529741640679936e-05,
      "loss": 0.0,
      "step": 2444
    },
    {
      "epoch": 0.4112353881086536,
      "grad_norm": NaN,
      "learning_rate": 6.527182867765332e-05,
      "loss": 0.0,
      "step": 2445
    },
    {
      "epoch": 0.4114035825414179,
      "grad_norm": NaN,
      "learning_rate": 6.524623653664703e-05,
      "loss": 0.0,
      "step": 2446
    },
    {
      "epoch": 0.4115717769741822,
      "grad_norm": NaN,
      "learning_rate": 6.522063999117379e-05,
      "loss": 0.0,
      "step": 2447
    },
    {
      "epoch": 0.4117399714069464,
      "grad_norm": NaN,
      "learning_rate": 6.51950390486281e-05,
      "loss": 0.0,
      "step": 2448
    },
    {
      "epoch": 0.4119081658397107,
      "grad_norm": NaN,
      "learning_rate": 6.516943371640583e-05,
      "loss": 0.0,
      "step": 2449
    },
    {
      "epoch": 0.41207636027247496,
      "grad_norm": NaN,
      "learning_rate": 6.514382400190409e-05,
      "loss": 0.0,
      "step": 2450
    },
    {
      "epoch": 0.41224455470523924,
      "grad_norm": NaN,
      "learning_rate": 6.51182099125212e-05,
      "loss": 0.0,
      "step": 2451
    },
    {
      "epoch": 0.4124127491380035,
      "grad_norm": NaN,
      "learning_rate": 6.509259145565681e-05,
      "loss": 0.0,
      "step": 2452
    },
    {
      "epoch": 0.4125809435707678,
      "grad_norm": NaN,
      "learning_rate": 6.506696863871178e-05,
      "loss": 0.0,
      "step": 2453
    },
    {
      "epoch": 0.4127491380035321,
      "grad_norm": NaN,
      "learning_rate": 6.504134146908828e-05,
      "loss": 0.0,
      "step": 2454
    },
    {
      "epoch": 0.41291733243629636,
      "grad_norm": NaN,
      "learning_rate": 6.501570995418969e-05,
      "loss": 0.0,
      "step": 2455
    },
    {
      "epoch": 0.41308552686906064,
      "grad_norm": NaN,
      "learning_rate": 6.499007410142069e-05,
      "loss": 0.0,
      "step": 2456
    },
    {
      "epoch": 0.4132537213018249,
      "grad_norm": NaN,
      "learning_rate": 6.496443391818719e-05,
      "loss": 0.0,
      "step": 2457
    },
    {
      "epoch": 0.4134219157345892,
      "grad_norm": NaN,
      "learning_rate": 6.493878941189633e-05,
      "loss": 0.0,
      "step": 2458
    },
    {
      "epoch": 0.4135901101673535,
      "grad_norm": NaN,
      "learning_rate": 6.491314058995654e-05,
      "loss": 0.0,
      "step": 2459
    },
    {
      "epoch": 0.41375830460011775,
      "grad_norm": NaN,
      "learning_rate": 6.488748745977746e-05,
      "loss": 0.0,
      "step": 2460
    },
    {
      "epoch": 0.41392649903288203,
      "grad_norm": NaN,
      "learning_rate": 6.486183002877001e-05,
      "loss": 0.0,
      "step": 2461
    },
    {
      "epoch": 0.4140946934656463,
      "grad_norm": NaN,
      "learning_rate": 6.483616830434632e-05,
      "loss": 0.0,
      "step": 2462
    },
    {
      "epoch": 0.41426288789841054,
      "grad_norm": NaN,
      "learning_rate": 6.481050229391978e-05,
      "loss": 0.0,
      "step": 2463
    },
    {
      "epoch": 0.4144310823311748,
      "grad_norm": NaN,
      "learning_rate": 6.478483200490503e-05,
      "loss": 0.0,
      "step": 2464
    },
    {
      "epoch": 0.4145992767639391,
      "grad_norm": NaN,
      "learning_rate": 6.475915744471792e-05,
      "loss": 0.0,
      "step": 2465
    },
    {
      "epoch": 0.4147674711967034,
      "grad_norm": NaN,
      "learning_rate": 6.473347862077552e-05,
      "loss": 0.0,
      "step": 2466
    },
    {
      "epoch": 0.41493566562946765,
      "grad_norm": NaN,
      "learning_rate": 6.470779554049615e-05,
      "loss": 0.0,
      "step": 2467
    },
    {
      "epoch": 0.41510386006223193,
      "grad_norm": NaN,
      "learning_rate": 6.468210821129942e-05,
      "loss": 0.0,
      "step": 2468
    },
    {
      "epoch": 0.4152720544949962,
      "grad_norm": NaN,
      "learning_rate": 6.465641664060605e-05,
      "loss": 0.0,
      "step": 2469
    },
    {
      "epoch": 0.4154402489277605,
      "grad_norm": NaN,
      "learning_rate": 6.46307208358381e-05,
      "loss": 0.0,
      "step": 2470
    },
    {
      "epoch": 0.41560844336052477,
      "grad_norm": NaN,
      "learning_rate": 6.460502080441874e-05,
      "loss": 0.0,
      "step": 2471
    },
    {
      "epoch": 0.41577663779328905,
      "grad_norm": NaN,
      "learning_rate": 6.45793165537725e-05,
      "loss": 0.0,
      "step": 2472
    },
    {
      "epoch": 0.41594483222605333,
      "grad_norm": NaN,
      "learning_rate": 6.455360809132496e-05,
      "loss": 0.0,
      "step": 2473
    },
    {
      "epoch": 0.4161130266588176,
      "grad_norm": NaN,
      "learning_rate": 6.452789542450309e-05,
      "loss": 0.0,
      "step": 2474
    },
    {
      "epoch": 0.4162812210915819,
      "grad_norm": NaN,
      "learning_rate": 6.450217856073494e-05,
      "loss": 0.0,
      "step": 2475
    },
    {
      "epoch": 0.41644941552434617,
      "grad_norm": NaN,
      "learning_rate": 6.447645750744984e-05,
      "loss": 0.0,
      "step": 2476
    },
    {
      "epoch": 0.41661760995711045,
      "grad_norm": NaN,
      "learning_rate": 6.44507322720783e-05,
      "loss": 0.0,
      "step": 2477
    },
    {
      "epoch": 0.41678580438987467,
      "grad_norm": NaN,
      "learning_rate": 6.442500286205207e-05,
      "loss": 0.0,
      "step": 2478
    },
    {
      "epoch": 0.41695399882263895,
      "grad_norm": NaN,
      "learning_rate": 6.439926928480408e-05,
      "loss": 0.0,
      "step": 2479
    },
    {
      "epoch": 0.41712219325540323,
      "grad_norm": NaN,
      "learning_rate": 6.437353154776849e-05,
      "loss": 0.0,
      "step": 2480
    },
    {
      "epoch": 0.4172903876881675,
      "grad_norm": NaN,
      "learning_rate": 6.43477896583806e-05,
      "loss": 0.0,
      "step": 2481
    },
    {
      "epoch": 0.4174585821209318,
      "grad_norm": NaN,
      "learning_rate": 6.4322043624077e-05,
      "loss": 0.0,
      "step": 2482
    },
    {
      "epoch": 0.41762677655369607,
      "grad_norm": NaN,
      "learning_rate": 6.42962934522954e-05,
      "loss": 0.0,
      "step": 2483
    },
    {
      "epoch": 0.41779497098646035,
      "grad_norm": NaN,
      "learning_rate": 6.427053915047477e-05,
      "loss": 0.0,
      "step": 2484
    },
    {
      "epoch": 0.4179631654192246,
      "grad_norm": NaN,
      "learning_rate": 6.424478072605522e-05,
      "loss": 0.0,
      "step": 2485
    },
    {
      "epoch": 0.4181313598519889,
      "grad_norm": NaN,
      "learning_rate": 6.421901818647807e-05,
      "loss": 0.0,
      "step": 2486
    },
    {
      "epoch": 0.4182995542847532,
      "grad_norm": NaN,
      "learning_rate": 6.41932515391858e-05,
      "loss": 0.0,
      "step": 2487
    },
    {
      "epoch": 0.41846774871751746,
      "grad_norm": NaN,
      "learning_rate": 6.416748079162216e-05,
      "loss": 0.0,
      "step": 2488
    },
    {
      "epoch": 0.41863594315028174,
      "grad_norm": NaN,
      "learning_rate": 6.414170595123198e-05,
      "loss": 0.0,
      "step": 2489
    },
    {
      "epoch": 0.418804137583046,
      "grad_norm": NaN,
      "learning_rate": 6.411592702546136e-05,
      "loss": 0.0,
      "step": 2490
    },
    {
      "epoch": 0.4189723320158103,
      "grad_norm": NaN,
      "learning_rate": 6.409014402175754e-05,
      "loss": 0.0,
      "step": 2491
    },
    {
      "epoch": 0.4191405264485746,
      "grad_norm": NaN,
      "learning_rate": 6.406435694756892e-05,
      "loss": 0.0,
      "step": 2492
    },
    {
      "epoch": 0.4193087208813388,
      "grad_norm": NaN,
      "learning_rate": 6.403856581034511e-05,
      "loss": 0.0,
      "step": 2493
    },
    {
      "epoch": 0.4194769153141031,
      "grad_norm": NaN,
      "learning_rate": 6.401277061753689e-05,
      "loss": 0.0,
      "step": 2494
    },
    {
      "epoch": 0.41964510974686736,
      "grad_norm": NaN,
      "learning_rate": 6.398697137659618e-05,
      "loss": 0.0,
      "step": 2495
    },
    {
      "epoch": 0.41981330417963164,
      "grad_norm": NaN,
      "learning_rate": 6.39611680949761e-05,
      "loss": 0.0,
      "step": 2496
    },
    {
      "epoch": 0.4199814986123959,
      "grad_norm": NaN,
      "learning_rate": 6.393536078013091e-05,
      "loss": 0.0,
      "step": 2497
    },
    {
      "epoch": 0.4201496930451602,
      "grad_norm": NaN,
      "learning_rate": 6.390954943951612e-05,
      "loss": 0.0,
      "step": 2498
    },
    {
      "epoch": 0.4203178874779245,
      "grad_norm": NaN,
      "learning_rate": 6.388373408058827e-05,
      "loss": 0.0,
      "step": 2499
    },
    {
      "epoch": 0.42048608191068876,
      "grad_norm": NaN,
      "learning_rate": 6.385791471080514e-05,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 0.42065427634345304,
      "grad_norm": NaN,
      "learning_rate": 6.383209133762569e-05,
      "loss": 0.0,
      "step": 2501
    },
    {
      "epoch": 0.4208224707762173,
      "grad_norm": NaN,
      "learning_rate": 6.380626396850997e-05,
      "loss": 0.0,
      "step": 2502
    },
    {
      "epoch": 0.4209906652089816,
      "grad_norm": NaN,
      "learning_rate": 6.378043261091922e-05,
      "loss": 0.0,
      "step": 2503
    },
    {
      "epoch": 0.4211588596417459,
      "grad_norm": NaN,
      "learning_rate": 6.375459727231585e-05,
      "loss": 0.0,
      "step": 2504
    },
    {
      "epoch": 0.42132705407451015,
      "grad_norm": NaN,
      "learning_rate": 6.37287579601634e-05,
      "loss": 0.0,
      "step": 2505
    },
    {
      "epoch": 0.42149524850727443,
      "grad_norm": NaN,
      "learning_rate": 6.370291468192652e-05,
      "loss": 0.0,
      "step": 2506
    },
    {
      "epoch": 0.42166344294003866,
      "grad_norm": NaN,
      "learning_rate": 6.367706744507109e-05,
      "loss": 0.0,
      "step": 2507
    },
    {
      "epoch": 0.42183163737280294,
      "grad_norm": NaN,
      "learning_rate": 6.365121625706405e-05,
      "loss": 0.0,
      "step": 2508
    },
    {
      "epoch": 0.4219998318055672,
      "grad_norm": NaN,
      "learning_rate": 6.362536112537354e-05,
      "loss": 0.0,
      "step": 2509
    },
    {
      "epoch": 0.4221680262383315,
      "grad_norm": NaN,
      "learning_rate": 6.359950205746881e-05,
      "loss": 0.0,
      "step": 2510
    },
    {
      "epoch": 0.4223362206710958,
      "grad_norm": NaN,
      "learning_rate": 6.357363906082028e-05,
      "loss": 0.0,
      "step": 2511
    },
    {
      "epoch": 0.42250441510386005,
      "grad_norm": NaN,
      "learning_rate": 6.354777214289944e-05,
      "loss": 0.0,
      "step": 2512
    },
    {
      "epoch": 0.42267260953662433,
      "grad_norm": NaN,
      "learning_rate": 6.352190131117899e-05,
      "loss": 0.0,
      "step": 2513
    },
    {
      "epoch": 0.4228408039693886,
      "grad_norm": NaN,
      "learning_rate": 6.349602657313268e-05,
      "loss": 0.0,
      "step": 2514
    },
    {
      "epoch": 0.4230089984021529,
      "grad_norm": NaN,
      "learning_rate": 6.347014793623547e-05,
      "loss": 0.0,
      "step": 2515
    },
    {
      "epoch": 0.42317719283491717,
      "grad_norm": NaN,
      "learning_rate": 6.34442654079634e-05,
      "loss": 0.0,
      "step": 2516
    },
    {
      "epoch": 0.42334538726768145,
      "grad_norm": NaN,
      "learning_rate": 6.341837899579363e-05,
      "loss": 0.0,
      "step": 2517
    },
    {
      "epoch": 0.42351358170044573,
      "grad_norm": NaN,
      "learning_rate": 6.339248870720447e-05,
      "loss": 0.0,
      "step": 2518
    },
    {
      "epoch": 0.42368177613321,
      "grad_norm": NaN,
      "learning_rate": 6.336659454967532e-05,
      "loss": 0.0,
      "step": 2519
    },
    {
      "epoch": 0.4238499705659743,
      "grad_norm": NaN,
      "learning_rate": 6.334069653068671e-05,
      "loss": 0.0,
      "step": 2520
    },
    {
      "epoch": 0.42401816499873857,
      "grad_norm": NaN,
      "learning_rate": 6.331479465772032e-05,
      "loss": 0.0,
      "step": 2521
    },
    {
      "epoch": 0.4241863594315028,
      "grad_norm": NaN,
      "learning_rate": 6.328888893825888e-05,
      "loss": 0.0,
      "step": 2522
    },
    {
      "epoch": 0.42435455386426707,
      "grad_norm": NaN,
      "learning_rate": 6.326297937978627e-05,
      "loss": 0.0,
      "step": 2523
    },
    {
      "epoch": 0.42452274829703135,
      "grad_norm": NaN,
      "learning_rate": 6.323706598978746e-05,
      "loss": 0.0,
      "step": 2524
    },
    {
      "epoch": 0.42469094272979563,
      "grad_norm": NaN,
      "learning_rate": 6.321114877574856e-05,
      "loss": 0.0,
      "step": 2525
    },
    {
      "epoch": 0.4248591371625599,
      "grad_norm": NaN,
      "learning_rate": 6.318522774515674e-05,
      "loss": 0.0,
      "step": 2526
    },
    {
      "epoch": 0.4250273315953242,
      "grad_norm": NaN,
      "learning_rate": 6.315930290550032e-05,
      "loss": 0.0,
      "step": 2527
    },
    {
      "epoch": 0.42519552602808847,
      "grad_norm": NaN,
      "learning_rate": 6.313337426426867e-05,
      "loss": 0.0,
      "step": 2528
    },
    {
      "epoch": 0.42536372046085275,
      "grad_norm": NaN,
      "learning_rate": 6.310744182895231e-05,
      "loss": 0.0,
      "step": 2529
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": NaN,
      "learning_rate": 6.308150560704281e-05,
      "loss": 0.0,
      "step": 2530
    },
    {
      "epoch": 0.4257001093263813,
      "grad_norm": NaN,
      "learning_rate": 6.305556560603286e-05,
      "loss": 0.0,
      "step": 2531
    },
    {
      "epoch": 0.4258683037591456,
      "grad_norm": NaN,
      "learning_rate": 6.302962183341626e-05,
      "loss": 0.0,
      "step": 2532
    },
    {
      "epoch": 0.42603649819190986,
      "grad_norm": NaN,
      "learning_rate": 6.300367429668783e-05,
      "loss": 0.0,
      "step": 2533
    },
    {
      "epoch": 0.42620469262467414,
      "grad_norm": NaN,
      "learning_rate": 6.297772300334355e-05,
      "loss": 0.0,
      "step": 2534
    },
    {
      "epoch": 0.4263728870574384,
      "grad_norm": NaN,
      "learning_rate": 6.295176796088045e-05,
      "loss": 0.0,
      "step": 2535
    },
    {
      "epoch": 0.4265410814902027,
      "grad_norm": NaN,
      "learning_rate": 6.292580917679665e-05,
      "loss": 0.0,
      "step": 2536
    },
    {
      "epoch": 0.4267092759229669,
      "grad_norm": NaN,
      "learning_rate": 6.289984665859136e-05,
      "loss": 0.0,
      "step": 2537
    },
    {
      "epoch": 0.4268774703557312,
      "grad_norm": NaN,
      "learning_rate": 6.287388041376486e-05,
      "loss": 0.0,
      "step": 2538
    },
    {
      "epoch": 0.4270456647884955,
      "grad_norm": NaN,
      "learning_rate": 6.284791044981851e-05,
      "loss": 0.0,
      "step": 2539
    },
    {
      "epoch": 0.42721385922125976,
      "grad_norm": NaN,
      "learning_rate": 6.282193677425474e-05,
      "loss": 0.0,
      "step": 2540
    },
    {
      "epoch": 0.42738205365402404,
      "grad_norm": NaN,
      "learning_rate": 6.279595939457705e-05,
      "loss": 0.0,
      "step": 2541
    },
    {
      "epoch": 0.4275502480867883,
      "grad_norm": NaN,
      "learning_rate": 6.276997831829e-05,
      "loss": 0.0,
      "step": 2542
    },
    {
      "epoch": 0.4277184425195526,
      "grad_norm": NaN,
      "learning_rate": 6.274399355289923e-05,
      "loss": 0.0,
      "step": 2543
    },
    {
      "epoch": 0.4278866369523169,
      "grad_norm": NaN,
      "learning_rate": 6.271800510591148e-05,
      "loss": 0.0,
      "step": 2544
    },
    {
      "epoch": 0.42805483138508116,
      "grad_norm": NaN,
      "learning_rate": 6.269201298483451e-05,
      "loss": 0.0,
      "step": 2545
    },
    {
      "epoch": 0.42822302581784544,
      "grad_norm": NaN,
      "learning_rate": 6.266601719717715e-05,
      "loss": 0.0,
      "step": 2546
    },
    {
      "epoch": 0.4283912202506097,
      "grad_norm": NaN,
      "learning_rate": 6.264001775044929e-05,
      "loss": 0.0,
      "step": 2547
    },
    {
      "epoch": 0.428559414683374,
      "grad_norm": NaN,
      "learning_rate": 6.261401465216188e-05,
      "loss": 0.0,
      "step": 2548
    },
    {
      "epoch": 0.4287276091161383,
      "grad_norm": NaN,
      "learning_rate": 6.25880079098269e-05,
      "loss": 0.0,
      "step": 2549
    },
    {
      "epoch": 0.42889580354890255,
      "grad_norm": NaN,
      "learning_rate": 6.256199753095745e-05,
      "loss": 0.0,
      "step": 2550
    },
    {
      "epoch": 0.42906399798166683,
      "grad_norm": NaN,
      "learning_rate": 6.25359835230676e-05,
      "loss": 0.0,
      "step": 2551
    },
    {
      "epoch": 0.42923219241443106,
      "grad_norm": NaN,
      "learning_rate": 6.250996589367255e-05,
      "loss": 0.0,
      "step": 2552
    },
    {
      "epoch": 0.42940038684719534,
      "grad_norm": NaN,
      "learning_rate": 6.248394465028844e-05,
      "loss": 0.0,
      "step": 2553
    },
    {
      "epoch": 0.4295685812799596,
      "grad_norm": NaN,
      "learning_rate": 6.245791980043257e-05,
      "loss": 0.0,
      "step": 2554
    },
    {
      "epoch": 0.4297367757127239,
      "grad_norm": NaN,
      "learning_rate": 6.24318913516232e-05,
      "loss": 0.0,
      "step": 2555
    },
    {
      "epoch": 0.4299049701454882,
      "grad_norm": NaN,
      "learning_rate": 6.240585931137966e-05,
      "loss": 0.0,
      "step": 2556
    },
    {
      "epoch": 0.43007316457825245,
      "grad_norm": NaN,
      "learning_rate": 6.237982368722232e-05,
      "loss": 0.0,
      "step": 2557
    },
    {
      "epoch": 0.43024135901101673,
      "grad_norm": NaN,
      "learning_rate": 6.235378448667257e-05,
      "loss": 0.0,
      "step": 2558
    },
    {
      "epoch": 0.430409553443781,
      "grad_norm": NaN,
      "learning_rate": 6.232774171725287e-05,
      "loss": 0.0,
      "step": 2559
    },
    {
      "epoch": 0.4305777478765453,
      "grad_norm": NaN,
      "learning_rate": 6.230169538648667e-05,
      "loss": 0.0,
      "step": 2560
    },
    {
      "epoch": 0.43074594230930957,
      "grad_norm": NaN,
      "learning_rate": 6.227564550189844e-05,
      "loss": 0.0,
      "step": 2561
    },
    {
      "epoch": 0.43091413674207385,
      "grad_norm": NaN,
      "learning_rate": 6.224959207101372e-05,
      "loss": 0.0,
      "step": 2562
    },
    {
      "epoch": 0.43108233117483813,
      "grad_norm": NaN,
      "learning_rate": 6.222353510135906e-05,
      "loss": 0.0,
      "step": 2563
    },
    {
      "epoch": 0.4312505256076024,
      "grad_norm": NaN,
      "learning_rate": 6.219747460046203e-05,
      "loss": 0.0,
      "step": 2564
    },
    {
      "epoch": 0.4314187200403667,
      "grad_norm": NaN,
      "learning_rate": 6.217141057585119e-05,
      "loss": 0.0,
      "step": 2565
    },
    {
      "epoch": 0.43158691447313097,
      "grad_norm": NaN,
      "learning_rate": 6.214534303505617e-05,
      "loss": 0.0,
      "step": 2566
    },
    {
      "epoch": 0.4317551089058952,
      "grad_norm": NaN,
      "learning_rate": 6.211927198560759e-05,
      "loss": 0.0,
      "step": 2567
    },
    {
      "epoch": 0.43192330333865947,
      "grad_norm": NaN,
      "learning_rate": 6.209319743503706e-05,
      "loss": 0.0,
      "step": 2568
    },
    {
      "epoch": 0.43209149777142375,
      "grad_norm": NaN,
      "learning_rate": 6.206711939087727e-05,
      "loss": 0.0,
      "step": 2569
    },
    {
      "epoch": 0.43225969220418803,
      "grad_norm": NaN,
      "learning_rate": 6.204103786066183e-05,
      "loss": 0.0,
      "step": 2570
    },
    {
      "epoch": 0.4324278866369523,
      "grad_norm": NaN,
      "learning_rate": 6.201495285192542e-05,
      "loss": 0.0,
      "step": 2571
    },
    {
      "epoch": 0.4325960810697166,
      "grad_norm": NaN,
      "learning_rate": 6.19888643722037e-05,
      "loss": 0.0,
      "step": 2572
    },
    {
      "epoch": 0.43276427550248087,
      "grad_norm": NaN,
      "learning_rate": 6.196277242903336e-05,
      "loss": 0.0,
      "step": 2573
    },
    {
      "epoch": 0.43293246993524515,
      "grad_norm": NaN,
      "learning_rate": 6.193667702995205e-05,
      "loss": 0.0,
      "step": 2574
    },
    {
      "epoch": 0.4331006643680094,
      "grad_norm": NaN,
      "learning_rate": 6.191057818249844e-05,
      "loss": 0.0,
      "step": 2575
    },
    {
      "epoch": 0.4332688588007737,
      "grad_norm": NaN,
      "learning_rate": 6.18844758942122e-05,
      "loss": 0.0,
      "step": 2576
    },
    {
      "epoch": 0.433437053233538,
      "grad_norm": NaN,
      "learning_rate": 6.185837017263399e-05,
      "loss": 0.0,
      "step": 2577
    },
    {
      "epoch": 0.43360524766630226,
      "grad_norm": NaN,
      "learning_rate": 6.183226102530547e-05,
      "loss": 0.0,
      "step": 2578
    },
    {
      "epoch": 0.43377344209906654,
      "grad_norm": NaN,
      "learning_rate": 6.180614845976926e-05,
      "loss": 0.0,
      "step": 2579
    },
    {
      "epoch": 0.4339416365318308,
      "grad_norm": NaN,
      "learning_rate": 6.178003248356898e-05,
      "loss": 0.0,
      "step": 2580
    },
    {
      "epoch": 0.43410983096459504,
      "grad_norm": NaN,
      "learning_rate": 6.175391310424928e-05,
      "loss": 0.0,
      "step": 2581
    },
    {
      "epoch": 0.4342780253973593,
      "grad_norm": NaN,
      "learning_rate": 6.172779032935573e-05,
      "loss": 0.0,
      "step": 2582
    },
    {
      "epoch": 0.4344462198301236,
      "grad_norm": NaN,
      "learning_rate": 6.17016641664349e-05,
      "loss": 0.0,
      "step": 2583
    },
    {
      "epoch": 0.4346144142628879,
      "grad_norm": NaN,
      "learning_rate": 6.167553462303438e-05,
      "loss": 0.0,
      "step": 2584
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": NaN,
      "learning_rate": 6.164940170670266e-05,
      "loss": 0.0,
      "step": 2585
    },
    {
      "epoch": 0.43495080312841644,
      "grad_norm": NaN,
      "learning_rate": 6.162326542498928e-05,
      "loss": 0.0,
      "step": 2586
    },
    {
      "epoch": 0.4351189975611807,
      "grad_norm": NaN,
      "learning_rate": 6.159712578544472e-05,
      "loss": 0.0,
      "step": 2587
    },
    {
      "epoch": 0.435287191993945,
      "grad_norm": NaN,
      "learning_rate": 6.157098279562041e-05,
      "loss": 0.0,
      "step": 2588
    },
    {
      "epoch": 0.4354553864267093,
      "grad_norm": NaN,
      "learning_rate": 6.154483646306875e-05,
      "loss": 0.0,
      "step": 2589
    },
    {
      "epoch": 0.43562358085947356,
      "grad_norm": NaN,
      "learning_rate": 6.151868679534316e-05,
      "loss": 0.0,
      "step": 2590
    },
    {
      "epoch": 0.43579177529223784,
      "grad_norm": NaN,
      "learning_rate": 6.149253379999796e-05,
      "loss": 0.0,
      "step": 2591
    },
    {
      "epoch": 0.4359599697250021,
      "grad_norm": NaN,
      "learning_rate": 6.146637748458849e-05,
      "loss": 0.0,
      "step": 2592
    },
    {
      "epoch": 0.4361281641577664,
      "grad_norm": NaN,
      "learning_rate": 6.144021785667098e-05,
      "loss": 0.0,
      "step": 2593
    },
    {
      "epoch": 0.4362963585905307,
      "grad_norm": NaN,
      "learning_rate": 6.141405492380268e-05,
      "loss": 0.0,
      "step": 2594
    },
    {
      "epoch": 0.43646455302329495,
      "grad_norm": NaN,
      "learning_rate": 6.138788869354176e-05,
      "loss": 0.0,
      "step": 2595
    },
    {
      "epoch": 0.4366327474560592,
      "grad_norm": NaN,
      "learning_rate": 6.136171917344733e-05,
      "loss": 0.0,
      "step": 2596
    },
    {
      "epoch": 0.43680094188882346,
      "grad_norm": NaN,
      "learning_rate": 6.13355463710795e-05,
      "loss": 0.0,
      "step": 2597
    },
    {
      "epoch": 0.43696913632158774,
      "grad_norm": NaN,
      "learning_rate": 6.13093702939993e-05,
      "loss": 0.0,
      "step": 2598
    },
    {
      "epoch": 0.437137330754352,
      "grad_norm": NaN,
      "learning_rate": 6.128319094976868e-05,
      "loss": 0.0,
      "step": 2599
    },
    {
      "epoch": 0.4373055251871163,
      "grad_norm": NaN,
      "learning_rate": 6.125700834595057e-05,
      "loss": 0.0,
      "step": 2600
    },
    {
      "epoch": 0.4374737196198806,
      "grad_norm": NaN,
      "learning_rate": 6.123082249010885e-05,
      "loss": 0.0,
      "step": 2601
    },
    {
      "epoch": 0.43764191405264485,
      "grad_norm": NaN,
      "learning_rate": 6.120463338980829e-05,
      "loss": 0.0,
      "step": 2602
    },
    {
      "epoch": 0.43781010848540913,
      "grad_norm": NaN,
      "learning_rate": 6.117844105261465e-05,
      "loss": 0.0,
      "step": 2603
    },
    {
      "epoch": 0.4379783029181734,
      "grad_norm": NaN,
      "learning_rate": 6.115224548609459e-05,
      "loss": 0.0,
      "step": 2604
    },
    {
      "epoch": 0.4381464973509377,
      "grad_norm": NaN,
      "learning_rate": 6.112604669781572e-05,
      "loss": 0.0,
      "step": 2605
    },
    {
      "epoch": 0.43831469178370197,
      "grad_norm": NaN,
      "learning_rate": 6.109984469534659e-05,
      "loss": 0.0,
      "step": 2606
    },
    {
      "epoch": 0.43848288621646625,
      "grad_norm": NaN,
      "learning_rate": 6.107363948625665e-05,
      "loss": 0.0,
      "step": 2607
    },
    {
      "epoch": 0.43865108064923053,
      "grad_norm": NaN,
      "learning_rate": 6.104743107811629e-05,
      "loss": 0.0,
      "step": 2608
    },
    {
      "epoch": 0.4388192750819948,
      "grad_norm": NaN,
      "learning_rate": 6.102121947849683e-05,
      "loss": 0.0,
      "step": 2609
    },
    {
      "epoch": 0.4389874695147591,
      "grad_norm": NaN,
      "learning_rate": 6.09950046949705e-05,
      "loss": 0.0,
      "step": 2610
    },
    {
      "epoch": 0.4391556639475233,
      "grad_norm": NaN,
      "learning_rate": 6.0968786735110486e-05,
      "loss": 0.0,
      "step": 2611
    },
    {
      "epoch": 0.4393238583802876,
      "grad_norm": NaN,
      "learning_rate": 6.094256560649081e-05,
      "loss": 0.0,
      "step": 2612
    },
    {
      "epoch": 0.43949205281305187,
      "grad_norm": NaN,
      "learning_rate": 6.091634131668652e-05,
      "loss": 0.0,
      "step": 2613
    },
    {
      "epoch": 0.43966024724581615,
      "grad_norm": NaN,
      "learning_rate": 6.089011387327348e-05,
      "loss": 0.0,
      "step": 2614
    },
    {
      "epoch": 0.43982844167858043,
      "grad_norm": NaN,
      "learning_rate": 6.086388328382853e-05,
      "loss": 0.0,
      "step": 2615
    },
    {
      "epoch": 0.4399966361113447,
      "grad_norm": NaN,
      "learning_rate": 6.0837649555929376e-05,
      "loss": 0.0,
      "step": 2616
    },
    {
      "epoch": 0.440164830544109,
      "grad_norm": NaN,
      "learning_rate": 6.081141269715466e-05,
      "loss": 0.0,
      "step": 2617
    },
    {
      "epoch": 0.44033302497687327,
      "grad_norm": NaN,
      "learning_rate": 6.0785172715083895e-05,
      "loss": 0.0,
      "step": 2618
    },
    {
      "epoch": 0.44050121940963755,
      "grad_norm": NaN,
      "learning_rate": 6.0758929617297545e-05,
      "loss": 0.0,
      "step": 2619
    },
    {
      "epoch": 0.4406694138424018,
      "grad_norm": NaN,
      "learning_rate": 6.0732683411376935e-05,
      "loss": 0.0,
      "step": 2620
    },
    {
      "epoch": 0.4408376082751661,
      "grad_norm": NaN,
      "learning_rate": 6.07064341049043e-05,
      "loss": 0.0,
      "step": 2621
    },
    {
      "epoch": 0.4410058027079304,
      "grad_norm": NaN,
      "learning_rate": 6.068018170546276e-05,
      "loss": 0.0,
      "step": 2622
    },
    {
      "epoch": 0.44117399714069466,
      "grad_norm": NaN,
      "learning_rate": 6.065392622063637e-05,
      "loss": 0.0,
      "step": 2623
    },
    {
      "epoch": 0.44134219157345894,
      "grad_norm": NaN,
      "learning_rate": 6.0627667658010015e-05,
      "loss": 0.0,
      "step": 2624
    },
    {
      "epoch": 0.4415103860062232,
      "grad_norm": NaN,
      "learning_rate": 6.060140602516952e-05,
      "loss": 0.0,
      "step": 2625
    },
    {
      "epoch": 0.44167858043898744,
      "grad_norm": NaN,
      "learning_rate": 6.0575141329701545e-05,
      "loss": 0.0,
      "step": 2626
    },
    {
      "epoch": 0.4418467748717517,
      "grad_norm": NaN,
      "learning_rate": 6.054887357919371e-05,
      "loss": 0.0,
      "step": 2627
    },
    {
      "epoch": 0.442014969304516,
      "grad_norm": NaN,
      "learning_rate": 6.0522602781234426e-05,
      "loss": 0.0,
      "step": 2628
    },
    {
      "epoch": 0.4421831637372803,
      "grad_norm": NaN,
      "learning_rate": 6.049632894341308e-05,
      "loss": 0.0,
      "step": 2629
    },
    {
      "epoch": 0.44235135817004456,
      "grad_norm": NaN,
      "learning_rate": 6.047005207331986e-05,
      "loss": 0.0,
      "step": 2630
    },
    {
      "epoch": 0.44251955260280884,
      "grad_norm": NaN,
      "learning_rate": 6.0443772178545874e-05,
      "loss": 0.0,
      "step": 2631
    },
    {
      "epoch": 0.4426877470355731,
      "grad_norm": NaN,
      "learning_rate": 6.041748926668308e-05,
      "loss": 0.0,
      "step": 2632
    },
    {
      "epoch": 0.4428559414683374,
      "grad_norm": NaN,
      "learning_rate": 6.039120334532432e-05,
      "loss": 0.0,
      "step": 2633
    },
    {
      "epoch": 0.4430241359011017,
      "grad_norm": NaN,
      "learning_rate": 6.0364914422063304e-05,
      "loss": 0.0,
      "step": 2634
    },
    {
      "epoch": 0.44319233033386596,
      "grad_norm": NaN,
      "learning_rate": 6.0338622504494604e-05,
      "loss": 0.0,
      "step": 2635
    },
    {
      "epoch": 0.44336052476663024,
      "grad_norm": NaN,
      "learning_rate": 6.031232760021366e-05,
      "loss": 0.0,
      "step": 2636
    },
    {
      "epoch": 0.4435287191993945,
      "grad_norm": NaN,
      "learning_rate": 6.0286029716816774e-05,
      "loss": 0.0,
      "step": 2637
    },
    {
      "epoch": 0.4436969136321588,
      "grad_norm": NaN,
      "learning_rate": 6.025972886190111e-05,
      "loss": 0.0,
      "step": 2638
    },
    {
      "epoch": 0.4438651080649231,
      "grad_norm": NaN,
      "learning_rate": 6.023342504306471e-05,
      "loss": 0.0,
      "step": 2639
    },
    {
      "epoch": 0.4440333024976873,
      "grad_norm": NaN,
      "learning_rate": 6.0207118267906415e-05,
      "loss": 0.0,
      "step": 2640
    },
    {
      "epoch": 0.4442014969304516,
      "grad_norm": NaN,
      "learning_rate": 6.018080854402599e-05,
      "loss": 0.0,
      "step": 2641
    },
    {
      "epoch": 0.44436969136321586,
      "grad_norm": NaN,
      "learning_rate": 6.0154495879024e-05,
      "loss": 0.0,
      "step": 2642
    },
    {
      "epoch": 0.44453788579598014,
      "grad_norm": NaN,
      "learning_rate": 6.012818028050189e-05,
      "loss": 0.0,
      "step": 2643
    },
    {
      "epoch": 0.4447060802287444,
      "grad_norm": NaN,
      "learning_rate": 6.010186175606195e-05,
      "loss": 0.0,
      "step": 2644
    },
    {
      "epoch": 0.4448742746615087,
      "grad_norm": NaN,
      "learning_rate": 6.0075540313307296e-05,
      "loss": 0.0,
      "step": 2645
    },
    {
      "epoch": 0.445042469094273,
      "grad_norm": NaN,
      "learning_rate": 6.004921595984189e-05,
      "loss": 0.0,
      "step": 2646
    },
    {
      "epoch": 0.44521066352703725,
      "grad_norm": NaN,
      "learning_rate": 6.002288870327055e-05,
      "loss": 0.0,
      "step": 2647
    },
    {
      "epoch": 0.44537885795980153,
      "grad_norm": NaN,
      "learning_rate": 5.999655855119893e-05,
      "loss": 0.0,
      "step": 2648
    },
    {
      "epoch": 0.4455470523925658,
      "grad_norm": NaN,
      "learning_rate": 5.9970225511233504e-05,
      "loss": 0.0,
      "step": 2649
    },
    {
      "epoch": 0.4457152468253301,
      "grad_norm": NaN,
      "learning_rate": 5.994388959098162e-05,
      "loss": 0.0,
      "step": 2650
    },
    {
      "epoch": 0.44588344125809437,
      "grad_norm": NaN,
      "learning_rate": 5.9917550798051384e-05,
      "loss": 0.0,
      "step": 2651
    },
    {
      "epoch": 0.44605163569085865,
      "grad_norm": NaN,
      "learning_rate": 5.989120914005183e-05,
      "loss": 0.0,
      "step": 2652
    },
    {
      "epoch": 0.44621983012362293,
      "grad_norm": NaN,
      "learning_rate": 5.986486462459273e-05,
      "loss": 0.0,
      "step": 2653
    },
    {
      "epoch": 0.4463880245563872,
      "grad_norm": NaN,
      "learning_rate": 5.983851725928474e-05,
      "loss": 0.0,
      "step": 2654
    },
    {
      "epoch": 0.44655621898915143,
      "grad_norm": NaN,
      "learning_rate": 5.98121670517393e-05,
      "loss": 0.0,
      "step": 2655
    },
    {
      "epoch": 0.4467244134219157,
      "grad_norm": NaN,
      "learning_rate": 5.978581400956872e-05,
      "loss": 0.0,
      "step": 2656
    },
    {
      "epoch": 0.44689260785468,
      "grad_norm": NaN,
      "learning_rate": 5.975945814038606e-05,
      "loss": 0.0,
      "step": 2657
    },
    {
      "epoch": 0.44706080228744427,
      "grad_norm": NaN,
      "learning_rate": 5.973309945180526e-05,
      "loss": 0.0,
      "step": 2658
    },
    {
      "epoch": 0.44722899672020855,
      "grad_norm": NaN,
      "learning_rate": 5.970673795144105e-05,
      "loss": 0.0,
      "step": 2659
    },
    {
      "epoch": 0.44739719115297283,
      "grad_norm": NaN,
      "learning_rate": 5.968037364690897e-05,
      "loss": 0.0,
      "step": 2660
    },
    {
      "epoch": 0.4475653855857371,
      "grad_norm": NaN,
      "learning_rate": 5.965400654582536e-05,
      "loss": 0.0,
      "step": 2661
    },
    {
      "epoch": 0.4477335800185014,
      "grad_norm": NaN,
      "learning_rate": 5.962763665580741e-05,
      "loss": 0.0,
      "step": 2662
    },
    {
      "epoch": 0.44790177445126567,
      "grad_norm": NaN,
      "learning_rate": 5.960126398447304e-05,
      "loss": 0.0,
      "step": 2663
    },
    {
      "epoch": 0.44806996888402995,
      "grad_norm": NaN,
      "learning_rate": 5.957488853944106e-05,
      "loss": 0.0,
      "step": 2664
    },
    {
      "epoch": 0.4482381633167942,
      "grad_norm": NaN,
      "learning_rate": 5.954851032833103e-05,
      "loss": 0.0,
      "step": 2665
    },
    {
      "epoch": 0.4484063577495585,
      "grad_norm": NaN,
      "learning_rate": 5.9522129358763315e-05,
      "loss": 0.0,
      "step": 2666
    },
    {
      "epoch": 0.4485745521823228,
      "grad_norm": NaN,
      "learning_rate": 5.9495745638359066e-05,
      "loss": 0.0,
      "step": 2667
    },
    {
      "epoch": 0.44874274661508706,
      "grad_norm": NaN,
      "learning_rate": 5.946935917474028e-05,
      "loss": 0.0,
      "step": 2668
    },
    {
      "epoch": 0.44891094104785134,
      "grad_norm": NaN,
      "learning_rate": 5.944296997552967e-05,
      "loss": 0.0,
      "step": 2669
    },
    {
      "epoch": 0.44907913548061557,
      "grad_norm": NaN,
      "learning_rate": 5.941657804835081e-05,
      "loss": 0.0,
      "step": 2670
    },
    {
      "epoch": 0.44924732991337984,
      "grad_norm": NaN,
      "learning_rate": 5.939018340082804e-05,
      "loss": 0.0,
      "step": 2671
    },
    {
      "epoch": 0.4494155243461441,
      "grad_norm": NaN,
      "learning_rate": 5.9363786040586455e-05,
      "loss": 0.0,
      "step": 2672
    },
    {
      "epoch": 0.4495837187789084,
      "grad_norm": NaN,
      "learning_rate": 5.933738597525195e-05,
      "loss": 0.0,
      "step": 2673
    },
    {
      "epoch": 0.4497519132116727,
      "grad_norm": NaN,
      "learning_rate": 5.9310983212451234e-05,
      "loss": 0.0,
      "step": 2674
    },
    {
      "epoch": 0.44992010764443696,
      "grad_norm": NaN,
      "learning_rate": 5.9284577759811744e-05,
      "loss": 0.0,
      "step": 2675
    },
    {
      "epoch": 0.45008830207720124,
      "grad_norm": NaN,
      "learning_rate": 5.925816962496175e-05,
      "loss": 0.0,
      "step": 2676
    },
    {
      "epoch": 0.4502564965099655,
      "grad_norm": NaN,
      "learning_rate": 5.923175881553022e-05,
      "loss": 0.0,
      "step": 2677
    },
    {
      "epoch": 0.4504246909427298,
      "grad_norm": NaN,
      "learning_rate": 5.9205345339147e-05,
      "loss": 0.0,
      "step": 2678
    },
    {
      "epoch": 0.4505928853754941,
      "grad_norm": NaN,
      "learning_rate": 5.91789292034426e-05,
      "loss": 0.0,
      "step": 2679
    },
    {
      "epoch": 0.45076107980825836,
      "grad_norm": NaN,
      "learning_rate": 5.915251041604837e-05,
      "loss": 0.0,
      "step": 2680
    },
    {
      "epoch": 0.45092927424102264,
      "grad_norm": NaN,
      "learning_rate": 5.9126088984596394e-05,
      "loss": 0.0,
      "step": 2681
    },
    {
      "epoch": 0.4510974686737869,
      "grad_norm": NaN,
      "learning_rate": 5.9099664916719535e-05,
      "loss": 0.0,
      "step": 2682
    },
    {
      "epoch": 0.4512656631065512,
      "grad_norm": NaN,
      "learning_rate": 5.9073238220051394e-05,
      "loss": 0.0,
      "step": 2683
    },
    {
      "epoch": 0.4514338575393155,
      "grad_norm": NaN,
      "learning_rate": 5.904680890222636e-05,
      "loss": 0.0,
      "step": 2684
    },
    {
      "epoch": 0.4516020519720797,
      "grad_norm": NaN,
      "learning_rate": 5.902037697087957e-05,
      "loss": 0.0,
      "step": 2685
    },
    {
      "epoch": 0.451770246404844,
      "grad_norm": NaN,
      "learning_rate": 5.89939424336469e-05,
      "loss": 0.0,
      "step": 2686
    },
    {
      "epoch": 0.45193844083760826,
      "grad_norm": NaN,
      "learning_rate": 5.896750529816499e-05,
      "loss": 0.0,
      "step": 2687
    },
    {
      "epoch": 0.45210663527037254,
      "grad_norm": NaN,
      "learning_rate": 5.8941065572071255e-05,
      "loss": 0.0,
      "step": 2688
    },
    {
      "epoch": 0.4522748297031368,
      "grad_norm": NaN,
      "learning_rate": 5.891462326300381e-05,
      "loss": 0.0,
      "step": 2689
    },
    {
      "epoch": 0.4524430241359011,
      "grad_norm": NaN,
      "learning_rate": 5.8888178378601565e-05,
      "loss": 0.0,
      "step": 2690
    },
    {
      "epoch": 0.4526112185686654,
      "grad_norm": NaN,
      "learning_rate": 5.886173092650414e-05,
      "loss": 0.0,
      "step": 2691
    },
    {
      "epoch": 0.45277941300142965,
      "grad_norm": NaN,
      "learning_rate": 5.883528091435191e-05,
      "loss": 0.0,
      "step": 2692
    },
    {
      "epoch": 0.45294760743419393,
      "grad_norm": NaN,
      "learning_rate": 5.880882834978597e-05,
      "loss": 0.0,
      "step": 2693
    },
    {
      "epoch": 0.4531158018669582,
      "grad_norm": NaN,
      "learning_rate": 5.87823732404482e-05,
      "loss": 0.0,
      "step": 2694
    },
    {
      "epoch": 0.4532839962997225,
      "grad_norm": NaN,
      "learning_rate": 5.875591559398116e-05,
      "loss": 0.0,
      "step": 2695
    },
    {
      "epoch": 0.45345219073248677,
      "grad_norm": NaN,
      "learning_rate": 5.872945541802818e-05,
      "loss": 0.0,
      "step": 2696
    },
    {
      "epoch": 0.45362038516525105,
      "grad_norm": NaN,
      "learning_rate": 5.8702992720233296e-05,
      "loss": 0.0,
      "step": 2697
    },
    {
      "epoch": 0.45378857959801533,
      "grad_norm": NaN,
      "learning_rate": 5.867652750824131e-05,
      "loss": 0.0,
      "step": 2698
    },
    {
      "epoch": 0.4539567740307796,
      "grad_norm": NaN,
      "learning_rate": 5.865005978969771e-05,
      "loss": 0.0,
      "step": 2699
    },
    {
      "epoch": 0.45412496846354383,
      "grad_norm": NaN,
      "learning_rate": 5.862358957224872e-05,
      "loss": 0.0,
      "step": 2700
    },
    {
      "epoch": 0.4542931628963081,
      "grad_norm": NaN,
      "learning_rate": 5.859711686354129e-05,
      "loss": 0.0,
      "step": 2701
    },
    {
      "epoch": 0.4544613573290724,
      "grad_norm": NaN,
      "learning_rate": 5.8570641671223084e-05,
      "loss": 0.0,
      "step": 2702
    },
    {
      "epoch": 0.45462955176183667,
      "grad_norm": NaN,
      "learning_rate": 5.85441640029425e-05,
      "loss": 0.0,
      "step": 2703
    },
    {
      "epoch": 0.45479774619460095,
      "grad_norm": NaN,
      "learning_rate": 5.851768386634863e-05,
      "loss": 0.0,
      "step": 2704
    },
    {
      "epoch": 0.45496594062736523,
      "grad_norm": NaN,
      "learning_rate": 5.84912012690913e-05,
      "loss": 0.0,
      "step": 2705
    },
    {
      "epoch": 0.4551341350601295,
      "grad_norm": NaN,
      "learning_rate": 5.846471621882103e-05,
      "loss": 0.0,
      "step": 2706
    },
    {
      "epoch": 0.4553023294928938,
      "grad_norm": NaN,
      "learning_rate": 5.843822872318906e-05,
      "loss": 0.0,
      "step": 2707
    },
    {
      "epoch": 0.45547052392565807,
      "grad_norm": NaN,
      "learning_rate": 5.841173878984731e-05,
      "loss": 0.0,
      "step": 2708
    },
    {
      "epoch": 0.45563871835842235,
      "grad_norm": NaN,
      "learning_rate": 5.838524642644846e-05,
      "loss": 0.0,
      "step": 2709
    },
    {
      "epoch": 0.4558069127911866,
      "grad_norm": NaN,
      "learning_rate": 5.8358751640645835e-05,
      "loss": 0.0,
      "step": 2710
    },
    {
      "epoch": 0.4559751072239509,
      "grad_norm": NaN,
      "learning_rate": 5.833225444009348e-05,
      "loss": 0.0,
      "step": 2711
    },
    {
      "epoch": 0.4561433016567152,
      "grad_norm": NaN,
      "learning_rate": 5.830575483244613e-05,
      "loss": 0.0,
      "step": 2712
    },
    {
      "epoch": 0.45631149608947946,
      "grad_norm": NaN,
      "learning_rate": 5.827925282535926e-05,
      "loss": 0.0,
      "step": 2713
    },
    {
      "epoch": 0.4564796905222437,
      "grad_norm": NaN,
      "learning_rate": 5.8252748426488966e-05,
      "loss": 0.0,
      "step": 2714
    },
    {
      "epoch": 0.45664788495500797,
      "grad_norm": NaN,
      "learning_rate": 5.8226241643492096e-05,
      "loss": 0.0,
      "step": 2715
    },
    {
      "epoch": 0.45681607938777224,
      "grad_norm": NaN,
      "learning_rate": 5.819973248402615e-05,
      "loss": 0.0,
      "step": 2716
    },
    {
      "epoch": 0.4569842738205365,
      "grad_norm": NaN,
      "learning_rate": 5.8173220955749343e-05,
      "loss": 0.0,
      "step": 2717
    },
    {
      "epoch": 0.4571524682533008,
      "grad_norm": NaN,
      "learning_rate": 5.814670706632054e-05,
      "loss": 0.0,
      "step": 2718
    },
    {
      "epoch": 0.4573206626860651,
      "grad_norm": NaN,
      "learning_rate": 5.8120190823399324e-05,
      "loss": 0.0,
      "step": 2719
    },
    {
      "epoch": 0.45748885711882936,
      "grad_norm": NaN,
      "learning_rate": 5.80936722346459e-05,
      "loss": 0.0,
      "step": 2720
    },
    {
      "epoch": 0.45765705155159364,
      "grad_norm": NaN,
      "learning_rate": 5.806715130772125e-05,
      "loss": 0.0,
      "step": 2721
    },
    {
      "epoch": 0.4578252459843579,
      "grad_norm": NaN,
      "learning_rate": 5.804062805028693e-05,
      "loss": 0.0,
      "step": 2722
    },
    {
      "epoch": 0.4579934404171222,
      "grad_norm": NaN,
      "learning_rate": 5.8014102470005236e-05,
      "loss": 0.0,
      "step": 2723
    },
    {
      "epoch": 0.4581616348498865,
      "grad_norm": NaN,
      "learning_rate": 5.7987574574539096e-05,
      "loss": 0.0,
      "step": 2724
    },
    {
      "epoch": 0.45832982928265076,
      "grad_norm": NaN,
      "learning_rate": 5.796104437155213e-05,
      "loss": 0.0,
      "step": 2725
    },
    {
      "epoch": 0.45849802371541504,
      "grad_norm": NaN,
      "learning_rate": 5.79345118687086e-05,
      "loss": 0.0,
      "step": 2726
    },
    {
      "epoch": 0.4586662181481793,
      "grad_norm": NaN,
      "learning_rate": 5.790797707367348e-05,
      "loss": 0.0,
      "step": 2727
    },
    {
      "epoch": 0.4588344125809436,
      "grad_norm": NaN,
      "learning_rate": 5.788143999411236e-05,
      "loss": 0.0,
      "step": 2728
    },
    {
      "epoch": 0.4590026070137078,
      "grad_norm": NaN,
      "learning_rate": 5.78549006376915e-05,
      "loss": 0.0,
      "step": 2729
    },
    {
      "epoch": 0.4591708014464721,
      "grad_norm": NaN,
      "learning_rate": 5.7828359012077814e-05,
      "loss": 0.0,
      "step": 2730
    },
    {
      "epoch": 0.4593389958792364,
      "grad_norm": NaN,
      "learning_rate": 5.7801815124938905e-05,
      "loss": 0.0,
      "step": 2731
    },
    {
      "epoch": 0.45950719031200066,
      "grad_norm": NaN,
      "learning_rate": 5.777526898394298e-05,
      "loss": 0.0,
      "step": 2732
    },
    {
      "epoch": 0.45967538474476494,
      "grad_norm": NaN,
      "learning_rate": 5.774872059675894e-05,
      "loss": 0.0,
      "step": 2733
    },
    {
      "epoch": 0.4598435791775292,
      "grad_norm": NaN,
      "learning_rate": 5.772216997105631e-05,
      "loss": 0.0,
      "step": 2734
    },
    {
      "epoch": 0.4600117736102935,
      "grad_norm": NaN,
      "learning_rate": 5.769561711450527e-05,
      "loss": 0.0,
      "step": 2735
    },
    {
      "epoch": 0.4601799680430578,
      "grad_norm": NaN,
      "learning_rate": 5.7669062034776634e-05,
      "loss": 0.0,
      "step": 2736
    },
    {
      "epoch": 0.46034816247582205,
      "grad_norm": NaN,
      "learning_rate": 5.764250473954189e-05,
      "loss": 0.0,
      "step": 2737
    },
    {
      "epoch": 0.46051635690858633,
      "grad_norm": NaN,
      "learning_rate": 5.761594523647312e-05,
      "loss": 0.0,
      "step": 2738
    },
    {
      "epoch": 0.4606845513413506,
      "grad_norm": NaN,
      "learning_rate": 5.758938353324308e-05,
      "loss": 0.0,
      "step": 2739
    },
    {
      "epoch": 0.4608527457741149,
      "grad_norm": NaN,
      "learning_rate": 5.7562819637525124e-05,
      "loss": 0.0,
      "step": 2740
    },
    {
      "epoch": 0.46102094020687917,
      "grad_norm": NaN,
      "learning_rate": 5.7536253556993306e-05,
      "loss": 0.0,
      "step": 2741
    },
    {
      "epoch": 0.46118913463964345,
      "grad_norm": NaN,
      "learning_rate": 5.750968529932223e-05,
      "loss": 0.0,
      "step": 2742
    },
    {
      "epoch": 0.46135732907240773,
      "grad_norm": NaN,
      "learning_rate": 5.7483114872187193e-05,
      "loss": 0.0,
      "step": 2743
    },
    {
      "epoch": 0.46152552350517195,
      "grad_norm": NaN,
      "learning_rate": 5.745654228326407e-05,
      "loss": 0.0,
      "step": 2744
    },
    {
      "epoch": 0.46169371793793623,
      "grad_norm": NaN,
      "learning_rate": 5.742996754022942e-05,
      "loss": 0.0,
      "step": 2745
    },
    {
      "epoch": 0.4618619123707005,
      "grad_norm": NaN,
      "learning_rate": 5.7403390650760356e-05,
      "loss": 0.0,
      "step": 2746
    },
    {
      "epoch": 0.4620301068034648,
      "grad_norm": NaN,
      "learning_rate": 5.737681162253468e-05,
      "loss": 0.0,
      "step": 2747
    },
    {
      "epoch": 0.46219830123622907,
      "grad_norm": NaN,
      "learning_rate": 5.7350230463230725e-05,
      "loss": 0.0,
      "step": 2748
    },
    {
      "epoch": 0.46236649566899335,
      "grad_norm": NaN,
      "learning_rate": 5.732364718052752e-05,
      "loss": 0.0,
      "step": 2749
    },
    {
      "epoch": 0.46253469010175763,
      "grad_norm": NaN,
      "learning_rate": 5.729706178210468e-05,
      "loss": 0.0,
      "step": 2750
    },
    {
      "epoch": 0.4627028845345219,
      "grad_norm": NaN,
      "learning_rate": 5.7270474275642406e-05,
      "loss": 0.0,
      "step": 2751
    },
    {
      "epoch": 0.4628710789672862,
      "grad_norm": NaN,
      "learning_rate": 5.724388466882157e-05,
      "loss": 0.0,
      "step": 2752
    },
    {
      "epoch": 0.46303927340005047,
      "grad_norm": NaN,
      "learning_rate": 5.7217292969323576e-05,
      "loss": 0.0,
      "step": 2753
    },
    {
      "epoch": 0.46320746783281475,
      "grad_norm": NaN,
      "learning_rate": 5.719069918483049e-05,
      "loss": 0.0,
      "step": 2754
    },
    {
      "epoch": 0.463375662265579,
      "grad_norm": NaN,
      "learning_rate": 5.716410332302493e-05,
      "loss": 0.0,
      "step": 2755
    },
    {
      "epoch": 0.4635438566983433,
      "grad_norm": NaN,
      "learning_rate": 5.713750539159018e-05,
      "loss": 0.0,
      "step": 2756
    },
    {
      "epoch": 0.4637120511311076,
      "grad_norm": NaN,
      "learning_rate": 5.711090539821008e-05,
      "loss": 0.0,
      "step": 2757
    },
    {
      "epoch": 0.46388024556387186,
      "grad_norm": NaN,
      "learning_rate": 5.708430335056905e-05,
      "loss": 0.0,
      "step": 2758
    },
    {
      "epoch": 0.4640484399966361,
      "grad_norm": NaN,
      "learning_rate": 5.7057699256352114e-05,
      "loss": 0.0,
      "step": 2759
    },
    {
      "epoch": 0.46421663442940037,
      "grad_norm": NaN,
      "learning_rate": 5.7031093123244925e-05,
      "loss": 0.0,
      "step": 2760
    },
    {
      "epoch": 0.46438482886216464,
      "grad_norm": NaN,
      "learning_rate": 5.700448495893368e-05,
      "loss": 0.0,
      "step": 2761
    },
    {
      "epoch": 0.4645530232949289,
      "grad_norm": NaN,
      "learning_rate": 5.697787477110519e-05,
      "loss": 0.0,
      "step": 2762
    },
    {
      "epoch": 0.4647212177276932,
      "grad_norm": NaN,
      "learning_rate": 5.695126256744683e-05,
      "loss": 0.0,
      "step": 2763
    },
    {
      "epoch": 0.4648894121604575,
      "grad_norm": NaN,
      "learning_rate": 5.692464835564658e-05,
      "loss": 0.0,
      "step": 2764
    },
    {
      "epoch": 0.46505760659322176,
      "grad_norm": NaN,
      "learning_rate": 5.689803214339298e-05,
      "loss": 0.0,
      "step": 2765
    },
    {
      "epoch": 0.46522580102598604,
      "grad_norm": NaN,
      "learning_rate": 5.687141393837516e-05,
      "loss": 0.0,
      "step": 2766
    },
    {
      "epoch": 0.4653939954587503,
      "grad_norm": NaN,
      "learning_rate": 5.68447937482828e-05,
      "loss": 0.0,
      "step": 2767
    },
    {
      "epoch": 0.4655621898915146,
      "grad_norm": NaN,
      "learning_rate": 5.6818171580806214e-05,
      "loss": 0.0,
      "step": 2768
    },
    {
      "epoch": 0.4657303843242789,
      "grad_norm": NaN,
      "learning_rate": 5.6791547443636205e-05,
      "loss": 0.0,
      "step": 2769
    },
    {
      "epoch": 0.46589857875704316,
      "grad_norm": NaN,
      "learning_rate": 5.676492134446422e-05,
      "loss": 0.0,
      "step": 2770
    },
    {
      "epoch": 0.46606677318980744,
      "grad_norm": NaN,
      "learning_rate": 5.6738293290982224e-05,
      "loss": 0.0,
      "step": 2771
    },
    {
      "epoch": 0.4662349676225717,
      "grad_norm": NaN,
      "learning_rate": 5.6711663290882776e-05,
      "loss": 0.0,
      "step": 2772
    },
    {
      "epoch": 0.466403162055336,
      "grad_norm": NaN,
      "learning_rate": 5.668503135185897e-05,
      "loss": 0.0,
      "step": 2773
    },
    {
      "epoch": 0.4665713564881002,
      "grad_norm": NaN,
      "learning_rate": 5.66583974816045e-05,
      "loss": 0.0,
      "step": 2774
    },
    {
      "epoch": 0.4667395509208645,
      "grad_norm": NaN,
      "learning_rate": 5.6631761687813564e-05,
      "loss": 0.0,
      "step": 2775
    },
    {
      "epoch": 0.4669077453536288,
      "grad_norm": NaN,
      "learning_rate": 5.6605123978180963e-05,
      "loss": 0.0,
      "step": 2776
    },
    {
      "epoch": 0.46707593978639306,
      "grad_norm": NaN,
      "learning_rate": 5.657848436040202e-05,
      "loss": 0.0,
      "step": 2777
    },
    {
      "epoch": 0.46724413421915734,
      "grad_norm": NaN,
      "learning_rate": 5.655184284217263e-05,
      "loss": 0.0,
      "step": 2778
    },
    {
      "epoch": 0.4674123286519216,
      "grad_norm": NaN,
      "learning_rate": 5.6525199431189226e-05,
      "loss": 0.0,
      "step": 2779
    },
    {
      "epoch": 0.4675805230846859,
      "grad_norm": NaN,
      "learning_rate": 5.64985541351488e-05,
      "loss": 0.0,
      "step": 2780
    },
    {
      "epoch": 0.4677487175174502,
      "grad_norm": NaN,
      "learning_rate": 5.6471906961748856e-05,
      "loss": 0.0,
      "step": 2781
    },
    {
      "epoch": 0.46791691195021445,
      "grad_norm": NaN,
      "learning_rate": 5.64452579186875e-05,
      "loss": 0.0,
      "step": 2782
    },
    {
      "epoch": 0.46808510638297873,
      "grad_norm": NaN,
      "learning_rate": 5.6418607013663314e-05,
      "loss": 0.0,
      "step": 2783
    },
    {
      "epoch": 0.468253300815743,
      "grad_norm": NaN,
      "learning_rate": 5.6391954254375465e-05,
      "loss": 0.0,
      "step": 2784
    },
    {
      "epoch": 0.4684214952485073,
      "grad_norm": NaN,
      "learning_rate": 5.6365299648523605e-05,
      "loss": 0.0,
      "step": 2785
    },
    {
      "epoch": 0.46858968968127157,
      "grad_norm": NaN,
      "learning_rate": 5.6338643203807986e-05,
      "loss": 0.0,
      "step": 2786
    },
    {
      "epoch": 0.46875788411403585,
      "grad_norm": NaN,
      "learning_rate": 5.631198492792933e-05,
      "loss": 0.0,
      "step": 2787
    },
    {
      "epoch": 0.4689260785468001,
      "grad_norm": NaN,
      "learning_rate": 5.6285324828588934e-05,
      "loss": 0.0,
      "step": 2788
    },
    {
      "epoch": 0.46909427297956435,
      "grad_norm": NaN,
      "learning_rate": 5.625866291348859e-05,
      "loss": 0.0,
      "step": 2789
    },
    {
      "epoch": 0.46926246741232863,
      "grad_norm": NaN,
      "learning_rate": 5.6231999190330645e-05,
      "loss": 0.0,
      "step": 2790
    },
    {
      "epoch": 0.4694306618450929,
      "grad_norm": NaN,
      "learning_rate": 5.6205333666817925e-05,
      "loss": 0.0,
      "step": 2791
    },
    {
      "epoch": 0.4695988562778572,
      "grad_norm": NaN,
      "learning_rate": 5.617866635065382e-05,
      "loss": 0.0,
      "step": 2792
    },
    {
      "epoch": 0.46976705071062147,
      "grad_norm": NaN,
      "learning_rate": 5.615199724954221e-05,
      "loss": 0.0,
      "step": 2793
    },
    {
      "epoch": 0.46993524514338575,
      "grad_norm": NaN,
      "learning_rate": 5.6125326371187504e-05,
      "loss": 0.0,
      "step": 2794
    },
    {
      "epoch": 0.47010343957615003,
      "grad_norm": NaN,
      "learning_rate": 5.6098653723294604e-05,
      "loss": 0.0,
      "step": 2795
    },
    {
      "epoch": 0.4702716340089143,
      "grad_norm": NaN,
      "learning_rate": 5.607197931356896e-05,
      "loss": 0.0,
      "step": 2796
    },
    {
      "epoch": 0.4704398284416786,
      "grad_norm": NaN,
      "learning_rate": 5.60453031497165e-05,
      "loss": 0.0,
      "step": 2797
    },
    {
      "epoch": 0.47060802287444287,
      "grad_norm": NaN,
      "learning_rate": 5.6018625239443666e-05,
      "loss": 0.0,
      "step": 2798
    },
    {
      "epoch": 0.47077621730720715,
      "grad_norm": NaN,
      "learning_rate": 5.599194559045743e-05,
      "loss": 0.0,
      "step": 2799
    },
    {
      "epoch": 0.4709444117399714,
      "grad_norm": NaN,
      "learning_rate": 5.596526421046521e-05,
      "loss": 0.0,
      "step": 2800
    },
    {
      "epoch": 0.4711126061727357,
      "grad_norm": NaN,
      "learning_rate": 5.593858110717498e-05,
      "loss": 0.0,
      "step": 2801
    },
    {
      "epoch": 0.4712808006055,
      "grad_norm": NaN,
      "learning_rate": 5.5911896288295183e-05,
      "loss": 0.0,
      "step": 2802
    },
    {
      "epoch": 0.4714489950382642,
      "grad_norm": NaN,
      "learning_rate": 5.588520976153477e-05,
      "loss": 0.0,
      "step": 2803
    },
    {
      "epoch": 0.4716171894710285,
      "grad_norm": NaN,
      "learning_rate": 5.585852153460318e-05,
      "loss": 0.0,
      "step": 2804
    },
    {
      "epoch": 0.47178538390379277,
      "grad_norm": NaN,
      "learning_rate": 5.583183161521033e-05,
      "loss": 0.0,
      "step": 2805
    },
    {
      "epoch": 0.47195357833655704,
      "grad_norm": NaN,
      "learning_rate": 5.5805140011066636e-05,
      "loss": 0.0,
      "step": 2806
    },
    {
      "epoch": 0.4721217727693213,
      "grad_norm": NaN,
      "learning_rate": 5.5778446729883026e-05,
      "loss": 0.0,
      "step": 2807
    },
    {
      "epoch": 0.4722899672020856,
      "grad_norm": NaN,
      "learning_rate": 5.5751751779370874e-05,
      "loss": 0.0,
      "step": 2808
    },
    {
      "epoch": 0.4724581616348499,
      "grad_norm": NaN,
      "learning_rate": 5.572505516724207e-05,
      "loss": 0.0,
      "step": 2809
    },
    {
      "epoch": 0.47262635606761416,
      "grad_norm": NaN,
      "learning_rate": 5.5698356901208925e-05,
      "loss": 0.0,
      "step": 2810
    },
    {
      "epoch": 0.47279455050037844,
      "grad_norm": NaN,
      "learning_rate": 5.567165698898432e-05,
      "loss": 0.0,
      "step": 2811
    },
    {
      "epoch": 0.4729627449331427,
      "grad_norm": NaN,
      "learning_rate": 5.564495543828153e-05,
      "loss": 0.0,
      "step": 2812
    },
    {
      "epoch": 0.473130939365907,
      "grad_norm": NaN,
      "learning_rate": 5.561825225681435e-05,
      "loss": 0.0,
      "step": 2813
    },
    {
      "epoch": 0.4732991337986713,
      "grad_norm": NaN,
      "learning_rate": 5.5591547452297e-05,
      "loss": 0.0,
      "step": 2814
    },
    {
      "epoch": 0.47346732823143556,
      "grad_norm": NaN,
      "learning_rate": 5.5564841032444234e-05,
      "loss": 0.0,
      "step": 2815
    },
    {
      "epoch": 0.47363552266419984,
      "grad_norm": NaN,
      "learning_rate": 5.5538133004971216e-05,
      "loss": 0.0,
      "step": 2816
    },
    {
      "epoch": 0.4738037170969641,
      "grad_norm": NaN,
      "learning_rate": 5.551142337759362e-05,
      "loss": 0.0,
      "step": 2817
    },
    {
      "epoch": 0.47397191152972834,
      "grad_norm": NaN,
      "learning_rate": 5.5484712158027515e-05,
      "loss": 0.0,
      "step": 2818
    },
    {
      "epoch": 0.4741401059624926,
      "grad_norm": NaN,
      "learning_rate": 5.545799935398952e-05,
      "loss": 0.0,
      "step": 2819
    },
    {
      "epoch": 0.4743083003952569,
      "grad_norm": NaN,
      "learning_rate": 5.543128497319664e-05,
      "loss": 0.0,
      "step": 2820
    },
    {
      "epoch": 0.4744764948280212,
      "grad_norm": NaN,
      "learning_rate": 5.540456902336637e-05,
      "loss": 0.0,
      "step": 2821
    },
    {
      "epoch": 0.47464468926078546,
      "grad_norm": NaN,
      "learning_rate": 5.5377851512216635e-05,
      "loss": 0.0,
      "step": 2822
    },
    {
      "epoch": 0.47481288369354974,
      "grad_norm": NaN,
      "learning_rate": 5.535113244746585e-05,
      "loss": 0.0,
      "step": 2823
    },
    {
      "epoch": 0.474981078126314,
      "grad_norm": NaN,
      "learning_rate": 5.532441183683281e-05,
      "loss": 0.0,
      "step": 2824
    },
    {
      "epoch": 0.4751492725590783,
      "grad_norm": NaN,
      "learning_rate": 5.529768968803686e-05,
      "loss": 0.0,
      "step": 2825
    },
    {
      "epoch": 0.4753174669918426,
      "grad_norm": NaN,
      "learning_rate": 5.5270966008797664e-05,
      "loss": 0.0,
      "step": 2826
    },
    {
      "epoch": 0.47548566142460685,
      "grad_norm": NaN,
      "learning_rate": 5.5244240806835445e-05,
      "loss": 0.0,
      "step": 2827
    },
    {
      "epoch": 0.47565385585737113,
      "grad_norm": NaN,
      "learning_rate": 5.521751408987078e-05,
      "loss": 0.0,
      "step": 2828
    },
    {
      "epoch": 0.4758220502901354,
      "grad_norm": NaN,
      "learning_rate": 5.519078586562475e-05,
      "loss": 0.0,
      "step": 2829
    },
    {
      "epoch": 0.4759902447228997,
      "grad_norm": NaN,
      "learning_rate": 5.516405614181883e-05,
      "loss": 0.0,
      "step": 2830
    },
    {
      "epoch": 0.47615843915566397,
      "grad_norm": NaN,
      "learning_rate": 5.513732492617492e-05,
      "loss": 0.0,
      "step": 2831
    },
    {
      "epoch": 0.47632663358842825,
      "grad_norm": NaN,
      "learning_rate": 5.511059222641537e-05,
      "loss": 0.0,
      "step": 2832
    },
    {
      "epoch": 0.4764948280211925,
      "grad_norm": NaN,
      "learning_rate": 5.508385805026299e-05,
      "loss": 0.0,
      "step": 2833
    },
    {
      "epoch": 0.47666302245395675,
      "grad_norm": NaN,
      "learning_rate": 5.5057122405440945e-05,
      "loss": 0.0,
      "step": 2834
    },
    {
      "epoch": 0.47683121688672103,
      "grad_norm": NaN,
      "learning_rate": 5.5030385299672884e-05,
      "loss": 0.0,
      "step": 2835
    },
    {
      "epoch": 0.4769994113194853,
      "grad_norm": NaN,
      "learning_rate": 5.500364674068285e-05,
      "loss": 0.0,
      "step": 2836
    },
    {
      "epoch": 0.4771676057522496,
      "grad_norm": NaN,
      "learning_rate": 5.497690673619532e-05,
      "loss": 0.0,
      "step": 2837
    },
    {
      "epoch": 0.47733580018501387,
      "grad_norm": NaN,
      "learning_rate": 5.4950165293935177e-05,
      "loss": 0.0,
      "step": 2838
    },
    {
      "epoch": 0.47750399461777815,
      "grad_norm": NaN,
      "learning_rate": 5.4923422421627736e-05,
      "loss": 0.0,
      "step": 2839
    },
    {
      "epoch": 0.47767218905054243,
      "grad_norm": NaN,
      "learning_rate": 5.489667812699869e-05,
      "loss": 0.0,
      "step": 2840
    },
    {
      "epoch": 0.4778403834833067,
      "grad_norm": NaN,
      "learning_rate": 5.4869932417774196e-05,
      "loss": 0.0,
      "step": 2841
    },
    {
      "epoch": 0.478008577916071,
      "grad_norm": NaN,
      "learning_rate": 5.4843185301680755e-05,
      "loss": 0.0,
      "step": 2842
    },
    {
      "epoch": 0.47817677234883527,
      "grad_norm": NaN,
      "learning_rate": 5.4816436786445335e-05,
      "loss": 0.0,
      "step": 2843
    },
    {
      "epoch": 0.47834496678159955,
      "grad_norm": NaN,
      "learning_rate": 5.4789686879795263e-05,
      "loss": 0.0,
      "step": 2844
    },
    {
      "epoch": 0.4785131612143638,
      "grad_norm": NaN,
      "learning_rate": 5.476293558945832e-05,
      "loss": 0.0,
      "step": 2845
    },
    {
      "epoch": 0.4786813556471281,
      "grad_norm": NaN,
      "learning_rate": 5.4736182923162605e-05,
      "loss": 0.0,
      "step": 2846
    },
    {
      "epoch": 0.4788495500798924,
      "grad_norm": NaN,
      "learning_rate": 5.470942888863672e-05,
      "loss": 0.0,
      "step": 2847
    },
    {
      "epoch": 0.4790177445126566,
      "grad_norm": NaN,
      "learning_rate": 5.468267349360957e-05,
      "loss": 0.0,
      "step": 2848
    },
    {
      "epoch": 0.4791859389454209,
      "grad_norm": NaN,
      "learning_rate": 5.465591674581049e-05,
      "loss": 0.0,
      "step": 2849
    },
    {
      "epoch": 0.47935413337818517,
      "grad_norm": NaN,
      "learning_rate": 5.4629158652969216e-05,
      "loss": 0.0,
      "step": 2850
    },
    {
      "epoch": 0.47952232781094944,
      "grad_norm": NaN,
      "learning_rate": 5.460239922281586e-05,
      "loss": 0.0,
      "step": 2851
    },
    {
      "epoch": 0.4796905222437137,
      "grad_norm": NaN,
      "learning_rate": 5.457563846308091e-05,
      "loss": 0.0,
      "step": 2852
    },
    {
      "epoch": 0.479858716676478,
      "grad_norm": NaN,
      "learning_rate": 5.454887638149526e-05,
      "loss": 0.0,
      "step": 2853
    },
    {
      "epoch": 0.4800269111092423,
      "grad_norm": NaN,
      "learning_rate": 5.4522112985790175e-05,
      "loss": 0.0,
      "step": 2854
    },
    {
      "epoch": 0.48019510554200656,
      "grad_norm": NaN,
      "learning_rate": 5.44953482836973e-05,
      "loss": 0.0,
      "step": 2855
    },
    {
      "epoch": 0.48036329997477084,
      "grad_norm": NaN,
      "learning_rate": 5.446858228294865e-05,
      "loss": 0.0,
      "step": 2856
    },
    {
      "epoch": 0.4805314944075351,
      "grad_norm": NaN,
      "learning_rate": 5.444181499127663e-05,
      "loss": 0.0,
      "step": 2857
    },
    {
      "epoch": 0.4806996888402994,
      "grad_norm": NaN,
      "learning_rate": 5.441504641641403e-05,
      "loss": 0.0,
      "step": 2858
    },
    {
      "epoch": 0.4808678832730637,
      "grad_norm": NaN,
      "learning_rate": 5.4388276566093955e-05,
      "loss": 0.0,
      "step": 2859
    },
    {
      "epoch": 0.48103607770582796,
      "grad_norm": NaN,
      "learning_rate": 5.436150544804994e-05,
      "loss": 0.0,
      "step": 2860
    },
    {
      "epoch": 0.48120427213859224,
      "grad_norm": NaN,
      "learning_rate": 5.433473307001584e-05,
      "loss": 0.0,
      "step": 2861
    },
    {
      "epoch": 0.48137246657135646,
      "grad_norm": NaN,
      "learning_rate": 5.4307959439725916e-05,
      "loss": 0.0,
      "step": 2862
    },
    {
      "epoch": 0.48154066100412074,
      "grad_norm": NaN,
      "learning_rate": 5.4281184564914754e-05,
      "loss": 0.0,
      "step": 2863
    },
    {
      "epoch": 0.481708855436885,
      "grad_norm": NaN,
      "learning_rate": 5.425440845331733e-05,
      "loss": 0.0,
      "step": 2864
    },
    {
      "epoch": 0.4818770498696493,
      "grad_norm": NaN,
      "learning_rate": 5.4227631112668955e-05,
      "loss": 0.0,
      "step": 2865
    },
    {
      "epoch": 0.4820452443024136,
      "grad_norm": NaN,
      "learning_rate": 5.420085255070531e-05,
      "loss": 0.0,
      "step": 2866
    },
    {
      "epoch": 0.48221343873517786,
      "grad_norm": NaN,
      "learning_rate": 5.41740727751624e-05,
      "loss": 0.0,
      "step": 2867
    },
    {
      "epoch": 0.48238163316794214,
      "grad_norm": NaN,
      "learning_rate": 5.4147291793776636e-05,
      "loss": 0.0,
      "step": 2868
    },
    {
      "epoch": 0.4825498276007064,
      "grad_norm": NaN,
      "learning_rate": 5.412050961428471e-05,
      "loss": 0.0,
      "step": 2869
    },
    {
      "epoch": 0.4827180220334707,
      "grad_norm": NaN,
      "learning_rate": 5.409372624442373e-05,
      "loss": 0.0,
      "step": 2870
    },
    {
      "epoch": 0.482886216466235,
      "grad_norm": NaN,
      "learning_rate": 5.406694169193106e-05,
      "loss": 0.0,
      "step": 2871
    },
    {
      "epoch": 0.48305441089899925,
      "grad_norm": NaN,
      "learning_rate": 5.4040155964544506e-05,
      "loss": 0.0,
      "step": 2872
    },
    {
      "epoch": 0.48322260533176353,
      "grad_norm": NaN,
      "learning_rate": 5.401336907000214e-05,
      "loss": 0.0,
      "step": 2873
    },
    {
      "epoch": 0.4833907997645278,
      "grad_norm": NaN,
      "learning_rate": 5.39865810160424e-05,
      "loss": 0.0,
      "step": 2874
    },
    {
      "epoch": 0.4835589941972921,
      "grad_norm": NaN,
      "learning_rate": 5.395979181040406e-05,
      "loss": 0.0,
      "step": 2875
    },
    {
      "epoch": 0.48372718863005637,
      "grad_norm": NaN,
      "learning_rate": 5.3933001460826225e-05,
      "loss": 0.0,
      "step": 2876
    },
    {
      "epoch": 0.4838953830628206,
      "grad_norm": NaN,
      "learning_rate": 5.3906209975048317e-05,
      "loss": 0.0,
      "step": 2877
    },
    {
      "epoch": 0.4840635774955849,
      "grad_norm": NaN,
      "learning_rate": 5.38794173608101e-05,
      "loss": 0.0,
      "step": 2878
    },
    {
      "epoch": 0.48423177192834915,
      "grad_norm": NaN,
      "learning_rate": 5.385262362585165e-05,
      "loss": 0.0,
      "step": 2879
    },
    {
      "epoch": 0.48439996636111343,
      "grad_norm": NaN,
      "learning_rate": 5.38258287779134e-05,
      "loss": 0.0,
      "step": 2880
    },
    {
      "epoch": 0.4845681607938777,
      "grad_norm": NaN,
      "learning_rate": 5.379903282473605e-05,
      "loss": 0.0,
      "step": 2881
    },
    {
      "epoch": 0.484736355226642,
      "grad_norm": NaN,
      "learning_rate": 5.377223577406069e-05,
      "loss": 0.0,
      "step": 2882
    },
    {
      "epoch": 0.48490454965940627,
      "grad_norm": NaN,
      "learning_rate": 5.374543763362865e-05,
      "loss": 0.0,
      "step": 2883
    },
    {
      "epoch": 0.48507274409217055,
      "grad_norm": NaN,
      "learning_rate": 5.371863841118163e-05,
      "loss": 0.0,
      "step": 2884
    },
    {
      "epoch": 0.48524093852493483,
      "grad_norm": NaN,
      "learning_rate": 5.3691838114461626e-05,
      "loss": 0.0,
      "step": 2885
    },
    {
      "epoch": 0.4854091329576991,
      "grad_norm": NaN,
      "learning_rate": 5.366503675121095e-05,
      "loss": 0.0,
      "step": 2886
    },
    {
      "epoch": 0.4855773273904634,
      "grad_norm": NaN,
      "learning_rate": 5.363823432917221e-05,
      "loss": 0.0,
      "step": 2887
    },
    {
      "epoch": 0.48574552182322767,
      "grad_norm": NaN,
      "learning_rate": 5.361143085608832e-05,
      "loss": 0.0,
      "step": 2888
    },
    {
      "epoch": 0.48591371625599195,
      "grad_norm": NaN,
      "learning_rate": 5.3584626339702495e-05,
      "loss": 0.0,
      "step": 2889
    },
    {
      "epoch": 0.4860819106887562,
      "grad_norm": NaN,
      "learning_rate": 5.355782078775827e-05,
      "loss": 0.0,
      "step": 2890
    },
    {
      "epoch": 0.4862501051215205,
      "grad_norm": NaN,
      "learning_rate": 5.3531014207999475e-05,
      "loss": 0.0,
      "step": 2891
    },
    {
      "epoch": 0.4864182995542847,
      "grad_norm": NaN,
      "learning_rate": 5.3504206608170224e-05,
      "loss": 0.0,
      "step": 2892
    },
    {
      "epoch": 0.486586493987049,
      "grad_norm": NaN,
      "learning_rate": 5.347739799601493e-05,
      "loss": 0.0,
      "step": 2893
    },
    {
      "epoch": 0.4867546884198133,
      "grad_norm": NaN,
      "learning_rate": 5.345058837927831e-05,
      "loss": 0.0,
      "step": 2894
    },
    {
      "epoch": 0.48692288285257757,
      "grad_norm": NaN,
      "learning_rate": 5.342377776570535e-05,
      "loss": 0.0,
      "step": 2895
    },
    {
      "epoch": 0.48709107728534184,
      "grad_norm": NaN,
      "learning_rate": 5.339696616304135e-05,
      "loss": 0.0,
      "step": 2896
    },
    {
      "epoch": 0.4872592717181061,
      "grad_norm": NaN,
      "learning_rate": 5.3370153579031876e-05,
      "loss": 0.0,
      "step": 2897
    },
    {
      "epoch": 0.4874274661508704,
      "grad_norm": NaN,
      "learning_rate": 5.334334002142278e-05,
      "loss": 0.0,
      "step": 2898
    },
    {
      "epoch": 0.4875956605836347,
      "grad_norm": NaN,
      "learning_rate": 5.33165254979602e-05,
      "loss": 0.0,
      "step": 2899
    },
    {
      "epoch": 0.48776385501639896,
      "grad_norm": NaN,
      "learning_rate": 5.3289710016390535e-05,
      "loss": 0.0,
      "step": 2900
    },
    {
      "epoch": 0.48793204944916324,
      "grad_norm": NaN,
      "learning_rate": 5.326289358446051e-05,
      "loss": 0.0,
      "step": 2901
    },
    {
      "epoch": 0.4881002438819275,
      "grad_norm": NaN,
      "learning_rate": 5.323607620991706e-05,
      "loss": 0.0,
      "step": 2902
    },
    {
      "epoch": 0.4882684383146918,
      "grad_norm": NaN,
      "learning_rate": 5.320925790050746e-05,
      "loss": 0.0,
      "step": 2903
    },
    {
      "epoch": 0.4884366327474561,
      "grad_norm": NaN,
      "learning_rate": 5.318243866397917e-05,
      "loss": 0.0,
      "step": 2904
    },
    {
      "epoch": 0.48860482718022036,
      "grad_norm": NaN,
      "learning_rate": 5.315561850808002e-05,
      "loss": 0.0,
      "step": 2905
    },
    {
      "epoch": 0.48877302161298464,
      "grad_norm": NaN,
      "learning_rate": 5.312879744055802e-05,
      "loss": 0.0,
      "step": 2906
    },
    {
      "epoch": 0.48894121604574886,
      "grad_norm": NaN,
      "learning_rate": 5.310197546916149e-05,
      "loss": 0.0,
      "step": 2907
    },
    {
      "epoch": 0.48910941047851314,
      "grad_norm": NaN,
      "learning_rate": 5.3075152601638976e-05,
      "loss": 0.0,
      "step": 2908
    },
    {
      "epoch": 0.4892776049112774,
      "grad_norm": NaN,
      "learning_rate": 5.304832884573934e-05,
      "loss": 0.0,
      "step": 2909
    },
    {
      "epoch": 0.4894457993440417,
      "grad_norm": NaN,
      "learning_rate": 5.302150420921161e-05,
      "loss": 0.0,
      "step": 2910
    },
    {
      "epoch": 0.489613993776806,
      "grad_norm": NaN,
      "learning_rate": 5.2994678699805186e-05,
      "loss": 0.0,
      "step": 2911
    },
    {
      "epoch": 0.48978218820957026,
      "grad_norm": NaN,
      "learning_rate": 5.296785232526962e-05,
      "loss": 0.0,
      "step": 2912
    },
    {
      "epoch": 0.48995038264233454,
      "grad_norm": NaN,
      "learning_rate": 5.2941025093354756e-05,
      "loss": 0.0,
      "step": 2913
    },
    {
      "epoch": 0.4901185770750988,
      "grad_norm": NaN,
      "learning_rate": 5.291419701181068e-05,
      "loss": 0.0,
      "step": 2914
    },
    {
      "epoch": 0.4902867715078631,
      "grad_norm": NaN,
      "learning_rate": 5.288736808838774e-05,
      "loss": 0.0,
      "step": 2915
    },
    {
      "epoch": 0.4904549659406274,
      "grad_norm": NaN,
      "learning_rate": 5.2860538330836504e-05,
      "loss": 0.0,
      "step": 2916
    },
    {
      "epoch": 0.49062316037339165,
      "grad_norm": NaN,
      "learning_rate": 5.283370774690778e-05,
      "loss": 0.0,
      "step": 2917
    },
    {
      "epoch": 0.49079135480615593,
      "grad_norm": NaN,
      "learning_rate": 5.280687634435261e-05,
      "loss": 0.0,
      "step": 2918
    },
    {
      "epoch": 0.4909595492389202,
      "grad_norm": NaN,
      "learning_rate": 5.2780044130922325e-05,
      "loss": 0.0,
      "step": 2919
    },
    {
      "epoch": 0.4911277436716845,
      "grad_norm": NaN,
      "learning_rate": 5.2753211114368406e-05,
      "loss": 0.0,
      "step": 2920
    },
    {
      "epoch": 0.4912959381044487,
      "grad_norm": NaN,
      "learning_rate": 5.2726377302442656e-05,
      "loss": 0.0,
      "step": 2921
    },
    {
      "epoch": 0.491464132537213,
      "grad_norm": NaN,
      "learning_rate": 5.269954270289702e-05,
      "loss": 0.0,
      "step": 2922
    },
    {
      "epoch": 0.4916323269699773,
      "grad_norm": NaN,
      "learning_rate": 5.2672707323483747e-05,
      "loss": 0.0,
      "step": 2923
    },
    {
      "epoch": 0.49180052140274155,
      "grad_norm": NaN,
      "learning_rate": 5.2645871171955275e-05,
      "loss": 0.0,
      "step": 2924
    },
    {
      "epoch": 0.49196871583550583,
      "grad_norm": NaN,
      "learning_rate": 5.2619034256064245e-05,
      "loss": 0.0,
      "step": 2925
    },
    {
      "epoch": 0.4921369102682701,
      "grad_norm": NaN,
      "learning_rate": 5.259219658356355e-05,
      "loss": 0.0,
      "step": 2926
    },
    {
      "epoch": 0.4923051047010344,
      "grad_norm": NaN,
      "learning_rate": 5.256535816220629e-05,
      "loss": 0.0,
      "step": 2927
    },
    {
      "epoch": 0.49247329913379867,
      "grad_norm": NaN,
      "learning_rate": 5.25385189997458e-05,
      "loss": 0.0,
      "step": 2928
    },
    {
      "epoch": 0.49264149356656295,
      "grad_norm": NaN,
      "learning_rate": 5.251167910393561e-05,
      "loss": 0.0,
      "step": 2929
    },
    {
      "epoch": 0.49280968799932723,
      "grad_norm": NaN,
      "learning_rate": 5.248483848252944e-05,
      "loss": 0.0,
      "step": 2930
    },
    {
      "epoch": 0.4929778824320915,
      "grad_norm": NaN,
      "learning_rate": 5.245799714328128e-05,
      "loss": 0.0,
      "step": 2931
    },
    {
      "epoch": 0.4931460768648558,
      "grad_norm": NaN,
      "learning_rate": 5.2431155093945275e-05,
      "loss": 0.0,
      "step": 2932
    },
    {
      "epoch": 0.49331427129762007,
      "grad_norm": NaN,
      "learning_rate": 5.2404312342275784e-05,
      "loss": 0.0,
      "step": 2933
    },
    {
      "epoch": 0.49348246573038435,
      "grad_norm": NaN,
      "learning_rate": 5.2377468896027403e-05,
      "loss": 0.0,
      "step": 2934
    },
    {
      "epoch": 0.4936506601631486,
      "grad_norm": NaN,
      "learning_rate": 5.2350624762954884e-05,
      "loss": 0.0,
      "step": 2935
    },
    {
      "epoch": 0.49381885459591285,
      "grad_norm": NaN,
      "learning_rate": 5.232377995081319e-05,
      "loss": 0.0,
      "step": 2936
    },
    {
      "epoch": 0.4939870490286771,
      "grad_norm": NaN,
      "learning_rate": 5.229693446735751e-05,
      "loss": 0.0,
      "step": 2937
    },
    {
      "epoch": 0.4941552434614414,
      "grad_norm": NaN,
      "learning_rate": 5.227008832034317e-05,
      "loss": 0.0,
      "step": 2938
    },
    {
      "epoch": 0.4943234378942057,
      "grad_norm": NaN,
      "learning_rate": 5.2243241517525754e-05,
      "loss": 0.0,
      "step": 2939
    },
    {
      "epoch": 0.49449163232696997,
      "grad_norm": NaN,
      "learning_rate": 5.221639406666098e-05,
      "loss": 0.0,
      "step": 2940
    },
    {
      "epoch": 0.49465982675973424,
      "grad_norm": NaN,
      "learning_rate": 5.2189545975504795e-05,
      "loss": 0.0,
      "step": 2941
    },
    {
      "epoch": 0.4948280211924985,
      "grad_norm": NaN,
      "learning_rate": 5.21626972518133e-05,
      "loss": 0.0,
      "step": 2942
    },
    {
      "epoch": 0.4949962156252628,
      "grad_norm": NaN,
      "learning_rate": 5.2135847903342795e-05,
      "loss": 0.0,
      "step": 2943
    },
    {
      "epoch": 0.4951644100580271,
      "grad_norm": NaN,
      "learning_rate": 5.2108997937849743e-05,
      "loss": 0.0,
      "step": 2944
    },
    {
      "epoch": 0.49533260449079136,
      "grad_norm": NaN,
      "learning_rate": 5.208214736309082e-05,
      "loss": 0.0,
      "step": 2945
    },
    {
      "epoch": 0.49550079892355564,
      "grad_norm": NaN,
      "learning_rate": 5.205529618682283e-05,
      "loss": 0.0,
      "step": 2946
    },
    {
      "epoch": 0.4956689933563199,
      "grad_norm": NaN,
      "learning_rate": 5.2028444416802814e-05,
      "loss": 0.0,
      "step": 2947
    },
    {
      "epoch": 0.4958371877890842,
      "grad_norm": NaN,
      "learning_rate": 5.200159206078792e-05,
      "loss": 0.0,
      "step": 2948
    },
    {
      "epoch": 0.4960053822218485,
      "grad_norm": NaN,
      "learning_rate": 5.197473912653549e-05,
      "loss": 0.0,
      "step": 2949
    },
    {
      "epoch": 0.49617357665461276,
      "grad_norm": NaN,
      "learning_rate": 5.194788562180305e-05,
      "loss": 0.0,
      "step": 2950
    },
    {
      "epoch": 0.496341771087377,
      "grad_norm": NaN,
      "learning_rate": 5.192103155434825e-05,
      "loss": 0.0,
      "step": 2951
    },
    {
      "epoch": 0.49650996552014126,
      "grad_norm": NaN,
      "learning_rate": 5.189417693192898e-05,
      "loss": 0.0,
      "step": 2952
    },
    {
      "epoch": 0.49667815995290554,
      "grad_norm": NaN,
      "learning_rate": 5.186732176230321e-05,
      "loss": 0.0,
      "step": 2953
    },
    {
      "epoch": 0.4968463543856698,
      "grad_norm": NaN,
      "learning_rate": 5.184046605322909e-05,
      "loss": 0.0,
      "step": 2954
    },
    {
      "epoch": 0.4970145488184341,
      "grad_norm": NaN,
      "learning_rate": 5.1813609812464936e-05,
      "loss": 0.0,
      "step": 2955
    },
    {
      "epoch": 0.4971827432511984,
      "grad_norm": NaN,
      "learning_rate": 5.178675304776923e-05,
      "loss": 0.0,
      "step": 2956
    },
    {
      "epoch": 0.49735093768396266,
      "grad_norm": NaN,
      "learning_rate": 5.1759895766900566e-05,
      "loss": 0.0,
      "step": 2957
    },
    {
      "epoch": 0.49751913211672694,
      "grad_norm": NaN,
      "learning_rate": 5.173303797761775e-05,
      "loss": 0.0,
      "step": 2958
    },
    {
      "epoch": 0.4976873265494912,
      "grad_norm": NaN,
      "learning_rate": 5.170617968767965e-05,
      "loss": 0.0,
      "step": 2959
    },
    {
      "epoch": 0.4978555209822555,
      "grad_norm": NaN,
      "learning_rate": 5.1679320904845365e-05,
      "loss": 0.0,
      "step": 2960
    },
    {
      "epoch": 0.4980237154150198,
      "grad_norm": NaN,
      "learning_rate": 5.165246163687406e-05,
      "loss": 0.0,
      "step": 2961
    },
    {
      "epoch": 0.49819190984778405,
      "grad_norm": NaN,
      "learning_rate": 5.162560189152512e-05,
      "loss": 0.0,
      "step": 2962
    },
    {
      "epoch": 0.49836010428054833,
      "grad_norm": NaN,
      "learning_rate": 5.1598741676557995e-05,
      "loss": 0.0,
      "step": 2963
    },
    {
      "epoch": 0.4985282987133126,
      "grad_norm": NaN,
      "learning_rate": 5.157188099973231e-05,
      "loss": 0.0,
      "step": 2964
    },
    {
      "epoch": 0.4986964931460769,
      "grad_norm": NaN,
      "learning_rate": 5.15450198688078e-05,
      "loss": 0.0,
      "step": 2965
    },
    {
      "epoch": 0.4988646875788411,
      "grad_norm": NaN,
      "learning_rate": 5.151815829154436e-05,
      "loss": 0.0,
      "step": 2966
    },
    {
      "epoch": 0.4990328820116054,
      "grad_norm": NaN,
      "learning_rate": 5.149129627570198e-05,
      "loss": 0.0,
      "step": 2967
    },
    {
      "epoch": 0.4992010764443697,
      "grad_norm": NaN,
      "learning_rate": 5.1464433829040834e-05,
      "loss": 0.0,
      "step": 2968
    },
    {
      "epoch": 0.49936927087713395,
      "grad_norm": NaN,
      "learning_rate": 5.143757095932113e-05,
      "loss": 0.0,
      "step": 2969
    },
    {
      "epoch": 0.49953746530989823,
      "grad_norm": NaN,
      "learning_rate": 5.14107076743033e-05,
      "loss": 0.0,
      "step": 2970
    },
    {
      "epoch": 0.4997056597426625,
      "grad_norm": NaN,
      "learning_rate": 5.1383843981747824e-05,
      "loss": 0.0,
      "step": 2971
    },
    {
      "epoch": 0.4998738541754268,
      "grad_norm": NaN,
      "learning_rate": 5.1356979889415324e-05,
      "loss": 0.0,
      "step": 2972
    },
    {
      "epoch": 0.500042048608191,
      "grad_norm": NaN,
      "learning_rate": 5.1330115405066517e-05,
      "loss": 0.0,
      "step": 2973
    },
    {
      "epoch": 0.5002102430409553,
      "grad_norm": NaN,
      "learning_rate": 5.1303250536462277e-05,
      "loss": 0.0,
      "step": 2974
    },
    {
      "epoch": 0.5003784374737196,
      "grad_norm": NaN,
      "learning_rate": 5.127638529136355e-05,
      "loss": 0.0,
      "step": 2975
    },
    {
      "epoch": 0.5005466319064839,
      "grad_norm": NaN,
      "learning_rate": 5.124951967753141e-05,
      "loss": 0.0,
      "step": 2976
    },
    {
      "epoch": 0.5007148263392481,
      "grad_norm": NaN,
      "learning_rate": 5.122265370272703e-05,
      "loss": 0.0,
      "step": 2977
    },
    {
      "epoch": 0.5008830207720124,
      "grad_norm": NaN,
      "learning_rate": 5.119578737471169e-05,
      "loss": 0.0,
      "step": 2978
    },
    {
      "epoch": 0.5010512152047767,
      "grad_norm": NaN,
      "learning_rate": 5.116892070124677e-05,
      "loss": 0.0,
      "step": 2979
    },
    {
      "epoch": 0.501219409637541,
      "grad_norm": NaN,
      "learning_rate": 5.114205369009375e-05,
      "loss": 0.0,
      "step": 2980
    },
    {
      "epoch": 0.5013876040703052,
      "grad_norm": NaN,
      "learning_rate": 5.1115186349014224e-05,
      "loss": 0.0,
      "step": 2981
    },
    {
      "epoch": 0.5015557985030695,
      "grad_norm": NaN,
      "learning_rate": 5.108831868576984e-05,
      "loss": 0.0,
      "step": 2982
    },
    {
      "epoch": 0.5017239929358338,
      "grad_norm": NaN,
      "learning_rate": 5.1061450708122385e-05,
      "loss": 0.0,
      "step": 2983
    },
    {
      "epoch": 0.5018921873685981,
      "grad_norm": NaN,
      "learning_rate": 5.103458242383371e-05,
      "loss": 0.0,
      "step": 2984
    },
    {
      "epoch": 0.5020603818013624,
      "grad_norm": NaN,
      "learning_rate": 5.100771384066575e-05,
      "loss": 0.0,
      "step": 2985
    },
    {
      "epoch": 0.5022285762341266,
      "grad_norm": NaN,
      "learning_rate": 5.0980844966380545e-05,
      "loss": 0.0,
      "step": 2986
    },
    {
      "epoch": 0.5023967706668909,
      "grad_norm": NaN,
      "learning_rate": 5.0953975808740226e-05,
      "loss": 0.0,
      "step": 2987
    },
    {
      "epoch": 0.5025649650996552,
      "grad_norm": NaN,
      "learning_rate": 5.0927106375506984e-05,
      "loss": 0.0,
      "step": 2988
    },
    {
      "epoch": 0.5027331595324195,
      "grad_norm": NaN,
      "learning_rate": 5.090023667444309e-05,
      "loss": 0.0,
      "step": 2989
    },
    {
      "epoch": 0.5029013539651838,
      "grad_norm": NaN,
      "learning_rate": 5.087336671331091e-05,
      "loss": 0.0,
      "step": 2990
    },
    {
      "epoch": 0.503069548397948,
      "grad_norm": NaN,
      "learning_rate": 5.084649649987285e-05,
      "loss": 0.0,
      "step": 2991
    },
    {
      "epoch": 0.5032377428307123,
      "grad_norm": NaN,
      "learning_rate": 5.081962604189145e-05,
      "loss": 0.0,
      "step": 2992
    },
    {
      "epoch": 0.5034059372634766,
      "grad_norm": NaN,
      "learning_rate": 5.079275534712926e-05,
      "loss": 0.0,
      "step": 2993
    },
    {
      "epoch": 0.5035741316962409,
      "grad_norm": NaN,
      "learning_rate": 5.076588442334894e-05,
      "loss": 0.0,
      "step": 2994
    },
    {
      "epoch": 0.5037423261290052,
      "grad_norm": NaN,
      "learning_rate": 5.073901327831317e-05,
      "loss": 0.0,
      "step": 2995
    },
    {
      "epoch": 0.5039105205617694,
      "grad_norm": NaN,
      "learning_rate": 5.071214191978476e-05,
      "loss": 0.0,
      "step": 2996
    },
    {
      "epoch": 0.5040787149945337,
      "grad_norm": NaN,
      "learning_rate": 5.0685270355526506e-05,
      "loss": 0.0,
      "step": 2997
    },
    {
      "epoch": 0.504246909427298,
      "grad_norm": NaN,
      "learning_rate": 5.065839859330134e-05,
      "loss": 0.0,
      "step": 2998
    },
    {
      "epoch": 0.5044151038600623,
      "grad_norm": NaN,
      "learning_rate": 5.06315266408722e-05,
      "loss": 0.0,
      "step": 2999
    },
    {
      "epoch": 0.5045832982928266,
      "grad_norm": NaN,
      "learning_rate": 5.060465450600207e-05,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 0.5047514927255908,
      "grad_norm": NaN,
      "learning_rate": 5.057778219645404e-05,
      "loss": 0.0,
      "step": 3001
    },
    {
      "epoch": 0.5049196871583551,
      "grad_norm": NaN,
      "learning_rate": 5.0550909719991194e-05,
      "loss": 0.0,
      "step": 3002
    },
    {
      "epoch": 0.5050878815911193,
      "grad_norm": NaN,
      "learning_rate": 5.052403708437669e-05,
      "loss": 0.0,
      "step": 3003
    },
    {
      "epoch": 0.5052560760238836,
      "grad_norm": NaN,
      "learning_rate": 5.0497164297373755e-05,
      "loss": 0.0,
      "step": 3004
    },
    {
      "epoch": 0.5054242704566478,
      "grad_norm": NaN,
      "learning_rate": 5.047029136674563e-05,
      "loss": 0.0,
      "step": 3005
    },
    {
      "epoch": 0.5055924648894121,
      "grad_norm": NaN,
      "learning_rate": 5.0443418300255577e-05,
      "loss": 0.0,
      "step": 3006
    },
    {
      "epoch": 0.5057606593221764,
      "grad_norm": NaN,
      "learning_rate": 5.041654510566696e-05,
      "loss": 0.0,
      "step": 3007
    },
    {
      "epoch": 0.5059288537549407,
      "grad_norm": NaN,
      "learning_rate": 5.038967179074312e-05,
      "loss": 0.0,
      "step": 3008
    },
    {
      "epoch": 0.506097048187705,
      "grad_norm": NaN,
      "learning_rate": 5.036279836324746e-05,
      "loss": 0.0,
      "step": 3009
    },
    {
      "epoch": 0.5062652426204692,
      "grad_norm": NaN,
      "learning_rate": 5.033592483094344e-05,
      "loss": 0.0,
      "step": 3010
    },
    {
      "epoch": 0.5064334370532335,
      "grad_norm": NaN,
      "learning_rate": 5.030905120159448e-05,
      "loss": 0.0,
      "step": 3011
    },
    {
      "epoch": 0.5066016314859978,
      "grad_norm": NaN,
      "learning_rate": 5.028217748296409e-05,
      "loss": 0.0,
      "step": 3012
    },
    {
      "epoch": 0.5067698259187621,
      "grad_norm": NaN,
      "learning_rate": 5.025530368281579e-05,
      "loss": 0.0,
      "step": 3013
    },
    {
      "epoch": 0.5069380203515264,
      "grad_norm": NaN,
      "learning_rate": 5.0228429808913105e-05,
      "loss": 0.0,
      "step": 3014
    },
    {
      "epoch": 0.5071062147842906,
      "grad_norm": NaN,
      "learning_rate": 5.0201555869019626e-05,
      "loss": 0.0,
      "step": 3015
    },
    {
      "epoch": 0.5072744092170549,
      "grad_norm": NaN,
      "learning_rate": 5.017468187089889e-05,
      "loss": 0.0,
      "step": 3016
    },
    {
      "epoch": 0.5074426036498192,
      "grad_norm": NaN,
      "learning_rate": 5.014780782231453e-05,
      "loss": 0.0,
      "step": 3017
    },
    {
      "epoch": 0.5076107980825835,
      "grad_norm": NaN,
      "learning_rate": 5.012093373103015e-05,
      "loss": 0.0,
      "step": 3018
    },
    {
      "epoch": 0.5077789925153477,
      "grad_norm": NaN,
      "learning_rate": 5.0094059604809375e-05,
      "loss": 0.0,
      "step": 3019
    },
    {
      "epoch": 0.507947186948112,
      "grad_norm": NaN,
      "learning_rate": 5.006718545141582e-05,
      "loss": 0.0,
      "step": 3020
    },
    {
      "epoch": 0.5081153813808763,
      "grad_norm": NaN,
      "learning_rate": 5.004031127861315e-05,
      "loss": 0.0,
      "step": 3021
    },
    {
      "epoch": 0.5082835758136406,
      "grad_norm": NaN,
      "learning_rate": 5.0013437094165e-05,
      "loss": 0.0,
      "step": 3022
    },
    {
      "epoch": 0.5084517702464049,
      "grad_norm": NaN,
      "learning_rate": 4.998656290583502e-05,
      "loss": 0.0,
      "step": 3023
    },
    {
      "epoch": 0.5086199646791691,
      "grad_norm": NaN,
      "learning_rate": 4.9959688721386874e-05,
      "loss": 0.0,
      "step": 3024
    },
    {
      "epoch": 0.5087881591119334,
      "grad_norm": NaN,
      "learning_rate": 4.993281454858418e-05,
      "loss": 0.0,
      "step": 3025
    },
    {
      "epoch": 0.5089563535446977,
      "grad_norm": NaN,
      "learning_rate": 4.990594039519063e-05,
      "loss": 0.0,
      "step": 3026
    },
    {
      "epoch": 0.509124547977462,
      "grad_norm": NaN,
      "learning_rate": 4.987906626896985e-05,
      "loss": 0.0,
      "step": 3027
    },
    {
      "epoch": 0.5092927424102263,
      "grad_norm": NaN,
      "learning_rate": 4.985219217768548e-05,
      "loss": 0.0,
      "step": 3028
    },
    {
      "epoch": 0.5094609368429905,
      "grad_norm": NaN,
      "learning_rate": 4.982531812910111e-05,
      "loss": 0.0,
      "step": 3029
    },
    {
      "epoch": 0.5096291312757548,
      "grad_norm": NaN,
      "learning_rate": 4.979844413098039e-05,
      "loss": 0.0,
      "step": 3030
    },
    {
      "epoch": 0.5097973257085191,
      "grad_norm": NaN,
      "learning_rate": 4.97715701910869e-05,
      "loss": 0.0,
      "step": 3031
    },
    {
      "epoch": 0.5099655201412833,
      "grad_norm": NaN,
      "learning_rate": 4.9744696317184234e-05,
      "loss": 0.0,
      "step": 3032
    },
    {
      "epoch": 0.5101337145740475,
      "grad_norm": NaN,
      "learning_rate": 4.971782251703591e-05,
      "loss": 0.0,
      "step": 3033
    },
    {
      "epoch": 0.5103019090068118,
      "grad_norm": NaN,
      "learning_rate": 4.969094879840553e-05,
      "loss": 0.0,
      "step": 3034
    },
    {
      "epoch": 0.5104701034395761,
      "grad_norm": NaN,
      "learning_rate": 4.966407516905657e-05,
      "loss": 0.0,
      "step": 3035
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": NaN,
      "learning_rate": 4.963720163675255e-05,
      "loss": 0.0,
      "step": 3036
    },
    {
      "epoch": 0.5108064923051047,
      "grad_norm": NaN,
      "learning_rate": 4.9610328209256886e-05,
      "loss": 0.0,
      "step": 3037
    },
    {
      "epoch": 0.5109746867378689,
      "grad_norm": NaN,
      "learning_rate": 4.958345489433305e-05,
      "loss": 0.0,
      "step": 3038
    },
    {
      "epoch": 0.5111428811706332,
      "grad_norm": NaN,
      "learning_rate": 4.955658169974443e-05,
      "loss": 0.0,
      "step": 3039
    },
    {
      "epoch": 0.5113110756033975,
      "grad_norm": NaN,
      "learning_rate": 4.95297086332544e-05,
      "loss": 0.0,
      "step": 3040
    },
    {
      "epoch": 0.5114792700361618,
      "grad_norm": NaN,
      "learning_rate": 4.950283570262624e-05,
      "loss": 0.0,
      "step": 3041
    },
    {
      "epoch": 0.5116474644689261,
      "grad_norm": NaN,
      "learning_rate": 4.947596291562331e-05,
      "loss": 0.0,
      "step": 3042
    },
    {
      "epoch": 0.5118156589016903,
      "grad_norm": NaN,
      "learning_rate": 4.9449090280008824e-05,
      "loss": 0.0,
      "step": 3043
    },
    {
      "epoch": 0.5119838533344546,
      "grad_norm": NaN,
      "learning_rate": 4.942221780354599e-05,
      "loss": 0.0,
      "step": 3044
    },
    {
      "epoch": 0.5121520477672189,
      "grad_norm": NaN,
      "learning_rate": 4.9395345493997944e-05,
      "loss": 0.0,
      "step": 3045
    },
    {
      "epoch": 0.5123202421999832,
      "grad_norm": NaN,
      "learning_rate": 4.936847335912782e-05,
      "loss": 0.0,
      "step": 3046
    },
    {
      "epoch": 0.5124884366327475,
      "grad_norm": NaN,
      "learning_rate": 4.934160140669867e-05,
      "loss": 0.0,
      "step": 3047
    },
    {
      "epoch": 0.5126566310655117,
      "grad_norm": NaN,
      "learning_rate": 4.93147296444735e-05,
      "loss": 0.0,
      "step": 3048
    },
    {
      "epoch": 0.512824825498276,
      "grad_norm": NaN,
      "learning_rate": 4.9287858080215265e-05,
      "loss": 0.0,
      "step": 3049
    },
    {
      "epoch": 0.5129930199310403,
      "grad_norm": NaN,
      "learning_rate": 4.9260986721686826e-05,
      "loss": 0.0,
      "step": 3050
    },
    {
      "epoch": 0.5131612143638046,
      "grad_norm": NaN,
      "learning_rate": 4.923411557665108e-05,
      "loss": 0.0,
      "step": 3051
    },
    {
      "epoch": 0.5133294087965689,
      "grad_norm": NaN,
      "learning_rate": 4.920724465287075e-05,
      "loss": 0.0,
      "step": 3052
    },
    {
      "epoch": 0.5134976032293331,
      "grad_norm": NaN,
      "learning_rate": 4.9180373958108566e-05,
      "loss": 0.0,
      "step": 3053
    },
    {
      "epoch": 0.5136657976620974,
      "grad_norm": NaN,
      "learning_rate": 4.915350350012714e-05,
      "loss": 0.0,
      "step": 3054
    },
    {
      "epoch": 0.5138339920948617,
      "grad_norm": NaN,
      "learning_rate": 4.91266332866891e-05,
      "loss": 0.0,
      "step": 3055
    },
    {
      "epoch": 0.514002186527626,
      "grad_norm": NaN,
      "learning_rate": 4.9099763325556916e-05,
      "loss": 0.0,
      "step": 3056
    },
    {
      "epoch": 0.5141703809603902,
      "grad_norm": NaN,
      "learning_rate": 4.907289362449303e-05,
      "loss": 0.0,
      "step": 3057
    },
    {
      "epoch": 0.5143385753931545,
      "grad_norm": NaN,
      "learning_rate": 4.904602419125978e-05,
      "loss": 0.0,
      "step": 3058
    },
    {
      "epoch": 0.5145067698259188,
      "grad_norm": NaN,
      "learning_rate": 4.901915503361946e-05,
      "loss": 0.0,
      "step": 3059
    },
    {
      "epoch": 0.5146749642586831,
      "grad_norm": NaN,
      "learning_rate": 4.899228615933427e-05,
      "loss": 0.0,
      "step": 3060
    },
    {
      "epoch": 0.5148431586914474,
      "grad_norm": NaN,
      "learning_rate": 4.8965417576166316e-05,
      "loss": 0.0,
      "step": 3061
    },
    {
      "epoch": 0.5150113531242115,
      "grad_norm": NaN,
      "learning_rate": 4.893854929187762e-05,
      "loss": 0.0,
      "step": 3062
    },
    {
      "epoch": 0.5151795475569758,
      "grad_norm": NaN,
      "learning_rate": 4.8911681314230164e-05,
      "loss": 0.0,
      "step": 3063
    },
    {
      "epoch": 0.5153477419897401,
      "grad_norm": NaN,
      "learning_rate": 4.888481365098579e-05,
      "loss": 0.0,
      "step": 3064
    },
    {
      "epoch": 0.5155159364225044,
      "grad_norm": NaN,
      "learning_rate": 4.885794630990625e-05,
      "loss": 0.0,
      "step": 3065
    },
    {
      "epoch": 0.5156841308552687,
      "grad_norm": NaN,
      "learning_rate": 4.883107929875323e-05,
      "loss": 0.0,
      "step": 3066
    },
    {
      "epoch": 0.5158523252880329,
      "grad_norm": NaN,
      "learning_rate": 4.880421262528832e-05,
      "loss": 0.0,
      "step": 3067
    },
    {
      "epoch": 0.5160205197207972,
      "grad_norm": NaN,
      "learning_rate": 4.8777346297272986e-05,
      "loss": 0.0,
      "step": 3068
    },
    {
      "epoch": 0.5161887141535615,
      "grad_norm": NaN,
      "learning_rate": 4.875048032246861e-05,
      "loss": 0.0,
      "step": 3069
    },
    {
      "epoch": 0.5163569085863258,
      "grad_norm": NaN,
      "learning_rate": 4.872361470863646e-05,
      "loss": 0.0,
      "step": 3070
    },
    {
      "epoch": 0.51652510301909,
      "grad_norm": NaN,
      "learning_rate": 4.8696749463537735e-05,
      "loss": 0.0,
      "step": 3071
    },
    {
      "epoch": 0.5166932974518543,
      "grad_norm": NaN,
      "learning_rate": 4.8669884594933495e-05,
      "loss": 0.0,
      "step": 3072
    },
    {
      "epoch": 0.5168614918846186,
      "grad_norm": NaN,
      "learning_rate": 4.8643020110584694e-05,
      "loss": 0.0,
      "step": 3073
    },
    {
      "epoch": 0.5170296863173829,
      "grad_norm": NaN,
      "learning_rate": 4.861615601825219e-05,
      "loss": 0.0,
      "step": 3074
    },
    {
      "epoch": 0.5171978807501472,
      "grad_norm": NaN,
      "learning_rate": 4.858929232569671e-05,
      "loss": 0.0,
      "step": 3075
    },
    {
      "epoch": 0.5173660751829114,
      "grad_norm": NaN,
      "learning_rate": 4.856242904067887e-05,
      "loss": 0.0,
      "step": 3076
    },
    {
      "epoch": 0.5175342696156757,
      "grad_norm": NaN,
      "learning_rate": 4.853556617095919e-05,
      "loss": 0.0,
      "step": 3077
    },
    {
      "epoch": 0.51770246404844,
      "grad_norm": NaN,
      "learning_rate": 4.8508703724298016e-05,
      "loss": 0.0,
      "step": 3078
    },
    {
      "epoch": 0.5178706584812043,
      "grad_norm": NaN,
      "learning_rate": 4.848184170845565e-05,
      "loss": 0.0,
      "step": 3079
    },
    {
      "epoch": 0.5180388529139686,
      "grad_norm": NaN,
      "learning_rate": 4.845498013119221e-05,
      "loss": 0.0,
      "step": 3080
    },
    {
      "epoch": 0.5182070473467328,
      "grad_norm": NaN,
      "learning_rate": 4.842811900026771e-05,
      "loss": 0.0,
      "step": 3081
    },
    {
      "epoch": 0.5183752417794971,
      "grad_norm": NaN,
      "learning_rate": 4.8401258323442016e-05,
      "loss": 0.0,
      "step": 3082
    },
    {
      "epoch": 0.5185434362122614,
      "grad_norm": NaN,
      "learning_rate": 4.837439810847489e-05,
      "loss": 0.0,
      "step": 3083
    },
    {
      "epoch": 0.5187116306450257,
      "grad_norm": NaN,
      "learning_rate": 4.8347538363125947e-05,
      "loss": 0.0,
      "step": 3084
    },
    {
      "epoch": 0.51887982507779,
      "grad_norm": NaN,
      "learning_rate": 4.832067909515467e-05,
      "loss": 0.0,
      "step": 3085
    },
    {
      "epoch": 0.5190480195105542,
      "grad_norm": NaN,
      "learning_rate": 4.8293820312320356e-05,
      "loss": 0.0,
      "step": 3086
    },
    {
      "epoch": 0.5192162139433185,
      "grad_norm": NaN,
      "learning_rate": 4.826696202238227e-05,
      "loss": 0.0,
      "step": 3087
    },
    {
      "epoch": 0.5193844083760828,
      "grad_norm": NaN,
      "learning_rate": 4.8240104233099445e-05,
      "loss": 0.0,
      "step": 3088
    },
    {
      "epoch": 0.5195526028088471,
      "grad_norm": NaN,
      "learning_rate": 4.8213246952230794e-05,
      "loss": 0.0,
      "step": 3089
    },
    {
      "epoch": 0.5197207972416114,
      "grad_norm": NaN,
      "learning_rate": 4.818639018753506e-05,
      "loss": 0.0,
      "step": 3090
    },
    {
      "epoch": 0.5198889916743756,
      "grad_norm": NaN,
      "learning_rate": 4.8159533946770916e-05,
      "loss": 0.0,
      "step": 3091
    },
    {
      "epoch": 0.5200571861071398,
      "grad_norm": NaN,
      "learning_rate": 4.8132678237696805e-05,
      "loss": 0.0,
      "step": 3092
    },
    {
      "epoch": 0.5202253805399041,
      "grad_norm": NaN,
      "learning_rate": 4.810582306807103e-05,
      "loss": 0.0,
      "step": 3093
    },
    {
      "epoch": 0.5203935749726684,
      "grad_norm": NaN,
      "learning_rate": 4.807896844565174e-05,
      "loss": 0.0,
      "step": 3094
    },
    {
      "epoch": 0.5205617694054326,
      "grad_norm": NaN,
      "learning_rate": 4.8052114378196964e-05,
      "loss": 0.0,
      "step": 3095
    },
    {
      "epoch": 0.5207299638381969,
      "grad_norm": NaN,
      "learning_rate": 4.802526087346453e-05,
      "loss": 0.0,
      "step": 3096
    },
    {
      "epoch": 0.5208981582709612,
      "grad_norm": NaN,
      "learning_rate": 4.7998407939212104e-05,
      "loss": 0.0,
      "step": 3097
    },
    {
      "epoch": 0.5210663527037255,
      "grad_norm": NaN,
      "learning_rate": 4.797155558319721e-05,
      "loss": 0.0,
      "step": 3098
    },
    {
      "epoch": 0.5212345471364898,
      "grad_norm": NaN,
      "learning_rate": 4.7944703813177166e-05,
      "loss": 0.0,
      "step": 3099
    },
    {
      "epoch": 0.521402741569254,
      "grad_norm": NaN,
      "learning_rate": 4.7917852636909185e-05,
      "loss": 0.0,
      "step": 3100
    },
    {
      "epoch": 0.5215709360020183,
      "grad_norm": NaN,
      "learning_rate": 4.789100206215027e-05,
      "loss": 0.0,
      "step": 3101
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": NaN,
      "learning_rate": 4.786415209665723e-05,
      "loss": 0.0,
      "step": 3102
    },
    {
      "epoch": 0.5219073248675469,
      "grad_norm": NaN,
      "learning_rate": 4.7837302748186705e-05,
      "loss": 0.0,
      "step": 3103
    },
    {
      "epoch": 0.5220755193003112,
      "grad_norm": NaN,
      "learning_rate": 4.781045402449522e-05,
      "loss": 0.0,
      "step": 3104
    },
    {
      "epoch": 0.5222437137330754,
      "grad_norm": NaN,
      "learning_rate": 4.778360593333903e-05,
      "loss": 0.0,
      "step": 3105
    },
    {
      "epoch": 0.5224119081658397,
      "grad_norm": NaN,
      "learning_rate": 4.775675848247427e-05,
      "loss": 0.0,
      "step": 3106
    },
    {
      "epoch": 0.522580102598604,
      "grad_norm": NaN,
      "learning_rate": 4.772991167965683e-05,
      "loss": 0.0,
      "step": 3107
    },
    {
      "epoch": 0.5227482970313683,
      "grad_norm": NaN,
      "learning_rate": 4.770306553264251e-05,
      "loss": 0.0,
      "step": 3108
    },
    {
      "epoch": 0.5229164914641325,
      "grad_norm": NaN,
      "learning_rate": 4.767622004918682e-05,
      "loss": 0.0,
      "step": 3109
    },
    {
      "epoch": 0.5230846858968968,
      "grad_norm": NaN,
      "learning_rate": 4.7649375237045135e-05,
      "loss": 0.0,
      "step": 3110
    },
    {
      "epoch": 0.5232528803296611,
      "grad_norm": NaN,
      "learning_rate": 4.762253110397261e-05,
      "loss": 0.0,
      "step": 3111
    },
    {
      "epoch": 0.5234210747624254,
      "grad_norm": NaN,
      "learning_rate": 4.759568765772422e-05,
      "loss": 0.0,
      "step": 3112
    },
    {
      "epoch": 0.5235892691951897,
      "grad_norm": NaN,
      "learning_rate": 4.756884490605474e-05,
      "loss": 0.0,
      "step": 3113
    },
    {
      "epoch": 0.523757463627954,
      "grad_norm": NaN,
      "learning_rate": 4.754200285671874e-05,
      "loss": 0.0,
      "step": 3114
    },
    {
      "epoch": 0.5239256580607182,
      "grad_norm": NaN,
      "learning_rate": 4.7515161517470556e-05,
      "loss": 0.0,
      "step": 3115
    },
    {
      "epoch": 0.5240938524934825,
      "grad_norm": NaN,
      "learning_rate": 4.748832089606441e-05,
      "loss": 0.0,
      "step": 3116
    },
    {
      "epoch": 0.5242620469262468,
      "grad_norm": NaN,
      "learning_rate": 4.746148100025421e-05,
      "loss": 0.0,
      "step": 3117
    },
    {
      "epoch": 0.5244302413590111,
      "grad_norm": NaN,
      "learning_rate": 4.743464183779372e-05,
      "loss": 0.0,
      "step": 3118
    },
    {
      "epoch": 0.5245984357917753,
      "grad_norm": NaN,
      "learning_rate": 4.7407803416436456e-05,
      "loss": 0.0,
      "step": 3119
    },
    {
      "epoch": 0.5247666302245396,
      "grad_norm": NaN,
      "learning_rate": 4.7380965743935766e-05,
      "loss": 0.0,
      "step": 3120
    },
    {
      "epoch": 0.5249348246573038,
      "grad_norm": NaN,
      "learning_rate": 4.735412882804474e-05,
      "loss": 0.0,
      "step": 3121
    },
    {
      "epoch": 0.5251030190900681,
      "grad_norm": NaN,
      "learning_rate": 4.732729267651626e-05,
      "loss": 0.0,
      "step": 3122
    },
    {
      "epoch": 0.5252712135228323,
      "grad_norm": NaN,
      "learning_rate": 4.730045729710298e-05,
      "loss": 0.0,
      "step": 3123
    },
    {
      "epoch": 0.5254394079555966,
      "grad_norm": NaN,
      "learning_rate": 4.7273622697557356e-05,
      "loss": 0.0,
      "step": 3124
    },
    {
      "epoch": 0.5256076023883609,
      "grad_norm": NaN,
      "learning_rate": 4.7246788885631605e-05,
      "loss": 0.0,
      "step": 3125
    },
    {
      "epoch": 0.5257757968211252,
      "grad_norm": NaN,
      "learning_rate": 4.72199558690777e-05,
      "loss": 0.0,
      "step": 3126
    },
    {
      "epoch": 0.5259439912538895,
      "grad_norm": NaN,
      "learning_rate": 4.719312365564739e-05,
      "loss": 0.0,
      "step": 3127
    },
    {
      "epoch": 0.5261121856866537,
      "grad_norm": NaN,
      "learning_rate": 4.716629225309224e-05,
      "loss": 0.0,
      "step": 3128
    },
    {
      "epoch": 0.526280380119418,
      "grad_norm": NaN,
      "learning_rate": 4.713946166916351e-05,
      "loss": 0.0,
      "step": 3129
    },
    {
      "epoch": 0.5264485745521823,
      "grad_norm": NaN,
      "learning_rate": 4.711263191161227e-05,
      "loss": 0.0,
      "step": 3130
    },
    {
      "epoch": 0.5266167689849466,
      "grad_norm": NaN,
      "learning_rate": 4.7085802988189315e-05,
      "loss": 0.0,
      "step": 3131
    },
    {
      "epoch": 0.5267849634177109,
      "grad_norm": NaN,
      "learning_rate": 4.7058974906645256e-05,
      "loss": 0.0,
      "step": 3132
    },
    {
      "epoch": 0.5269531578504751,
      "grad_norm": NaN,
      "learning_rate": 4.7032147674730395e-05,
      "loss": 0.0,
      "step": 3133
    },
    {
      "epoch": 0.5271213522832394,
      "grad_norm": NaN,
      "learning_rate": 4.700532130019483e-05,
      "loss": 0.0,
      "step": 3134
    },
    {
      "epoch": 0.5272895467160037,
      "grad_norm": NaN,
      "learning_rate": 4.697849579078838e-05,
      "loss": 0.0,
      "step": 3135
    },
    {
      "epoch": 0.527457741148768,
      "grad_norm": NaN,
      "learning_rate": 4.695167115426068e-05,
      "loss": 0.0,
      "step": 3136
    },
    {
      "epoch": 0.5276259355815323,
      "grad_norm": NaN,
      "learning_rate": 4.6924847398361035e-05,
      "loss": 0.0,
      "step": 3137
    },
    {
      "epoch": 0.5277941300142965,
      "grad_norm": NaN,
      "learning_rate": 4.689802453083853e-05,
      "loss": 0.0,
      "step": 3138
    },
    {
      "epoch": 0.5279623244470608,
      "grad_norm": NaN,
      "learning_rate": 4.687120255944199e-05,
      "loss": 0.0,
      "step": 3139
    },
    {
      "epoch": 0.5281305188798251,
      "grad_norm": NaN,
      "learning_rate": 4.684438149191999e-05,
      "loss": 0.0,
      "step": 3140
    },
    {
      "epoch": 0.5282987133125894,
      "grad_norm": NaN,
      "learning_rate": 4.681756133602084e-05,
      "loss": 0.0,
      "step": 3141
    },
    {
      "epoch": 0.5284669077453537,
      "grad_norm": NaN,
      "learning_rate": 4.6790742099492565e-05,
      "loss": 0.0,
      "step": 3142
    },
    {
      "epoch": 0.5286351021781179,
      "grad_norm": NaN,
      "learning_rate": 4.6763923790082935e-05,
      "loss": 0.0,
      "step": 3143
    },
    {
      "epoch": 0.5288032966108822,
      "grad_norm": NaN,
      "learning_rate": 4.67371064155395e-05,
      "loss": 0.0,
      "step": 3144
    },
    {
      "epoch": 0.5289714910436465,
      "grad_norm": NaN,
      "learning_rate": 4.671028998360947e-05,
      "loss": 0.0,
      "step": 3145
    },
    {
      "epoch": 0.5291396854764108,
      "grad_norm": NaN,
      "learning_rate": 4.6683474502039826e-05,
      "loss": 0.0,
      "step": 3146
    },
    {
      "epoch": 0.529307879909175,
      "grad_norm": NaN,
      "learning_rate": 4.665665997857723e-05,
      "loss": 0.0,
      "step": 3147
    },
    {
      "epoch": 0.5294760743419393,
      "grad_norm": NaN,
      "learning_rate": 4.6629846420968136e-05,
      "loss": 0.0,
      "step": 3148
    },
    {
      "epoch": 0.5296442687747036,
      "grad_norm": NaN,
      "learning_rate": 4.660303383695866e-05,
      "loss": 0.0,
      "step": 3149
    },
    {
      "epoch": 0.5298124632074679,
      "grad_norm": NaN,
      "learning_rate": 4.6576222234294656e-05,
      "loss": 0.0,
      "step": 3150
    },
    {
      "epoch": 0.5299806576402321,
      "grad_norm": NaN,
      "learning_rate": 4.654941162072171e-05,
      "loss": 0.0,
      "step": 3151
    },
    {
      "epoch": 0.5301488520729963,
      "grad_norm": NaN,
      "learning_rate": 4.652260200398507e-05,
      "loss": 0.0,
      "step": 3152
    },
    {
      "epoch": 0.5303170465057606,
      "grad_norm": NaN,
      "learning_rate": 4.649579339182978e-05,
      "loss": 0.0,
      "step": 3153
    },
    {
      "epoch": 0.5304852409385249,
      "grad_norm": NaN,
      "learning_rate": 4.646898579200054e-05,
      "loss": 0.0,
      "step": 3154
    },
    {
      "epoch": 0.5306534353712892,
      "grad_norm": NaN,
      "learning_rate": 4.644217921224175e-05,
      "loss": 0.0,
      "step": 3155
    },
    {
      "epoch": 0.5308216298040535,
      "grad_norm": NaN,
      "learning_rate": 4.641537366029751e-05,
      "loss": 0.0,
      "step": 3156
    },
    {
      "epoch": 0.5309898242368177,
      "grad_norm": NaN,
      "learning_rate": 4.63885691439117e-05,
      "loss": 0.0,
      "step": 3157
    },
    {
      "epoch": 0.531158018669582,
      "grad_norm": NaN,
      "learning_rate": 4.6361765670827805e-05,
      "loss": 0.0,
      "step": 3158
    },
    {
      "epoch": 0.5313262131023463,
      "grad_norm": NaN,
      "learning_rate": 4.633496324878906e-05,
      "loss": 0.0,
      "step": 3159
    },
    {
      "epoch": 0.5314944075351106,
      "grad_norm": NaN,
      "learning_rate": 4.630816188553837e-05,
      "loss": 0.0,
      "step": 3160
    },
    {
      "epoch": 0.5316626019678748,
      "grad_norm": NaN,
      "learning_rate": 4.6281361588818376e-05,
      "loss": 0.0,
      "step": 3161
    },
    {
      "epoch": 0.5318307964006391,
      "grad_norm": NaN,
      "learning_rate": 4.625456236637137e-05,
      "loss": 0.0,
      "step": 3162
    },
    {
      "epoch": 0.5319989908334034,
      "grad_norm": NaN,
      "learning_rate": 4.6227764225939337e-05,
      "loss": 0.0,
      "step": 3163
    },
    {
      "epoch": 0.5321671852661677,
      "grad_norm": NaN,
      "learning_rate": 4.620096717526395e-05,
      "loss": 0.0,
      "step": 3164
    },
    {
      "epoch": 0.532335379698932,
      "grad_norm": NaN,
      "learning_rate": 4.617417122208661e-05,
      "loss": 0.0,
      "step": 3165
    },
    {
      "epoch": 0.5325035741316962,
      "grad_norm": NaN,
      "learning_rate": 4.614737637414836e-05,
      "loss": 0.0,
      "step": 3166
    },
    {
      "epoch": 0.5326717685644605,
      "grad_norm": NaN,
      "learning_rate": 4.612058263918992e-05,
      "loss": 0.0,
      "step": 3167
    },
    {
      "epoch": 0.5328399629972248,
      "grad_norm": NaN,
      "learning_rate": 4.6093790024951695e-05,
      "loss": 0.0,
      "step": 3168
    },
    {
      "epoch": 0.5330081574299891,
      "grad_norm": NaN,
      "learning_rate": 4.606699853917379e-05,
      "loss": 0.0,
      "step": 3169
    },
    {
      "epoch": 0.5331763518627534,
      "grad_norm": NaN,
      "learning_rate": 4.6040208189595954e-05,
      "loss": 0.0,
      "step": 3170
    },
    {
      "epoch": 0.5333445462955176,
      "grad_norm": NaN,
      "learning_rate": 4.601341898395762e-05,
      "loss": 0.0,
      "step": 3171
    },
    {
      "epoch": 0.5335127407282819,
      "grad_norm": NaN,
      "learning_rate": 4.5986630929997866e-05,
      "loss": 0.0,
      "step": 3172
    },
    {
      "epoch": 0.5336809351610462,
      "grad_norm": NaN,
      "learning_rate": 4.59598440354555e-05,
      "loss": 0.0,
      "step": 3173
    },
    {
      "epoch": 0.5338491295938105,
      "grad_norm": NaN,
      "learning_rate": 4.593305830806895e-05,
      "loss": 0.0,
      "step": 3174
    },
    {
      "epoch": 0.5340173240265748,
      "grad_norm": NaN,
      "learning_rate": 4.5906273755576304e-05,
      "loss": 0.0,
      "step": 3175
    },
    {
      "epoch": 0.534185518459339,
      "grad_norm": NaN,
      "learning_rate": 4.58794903857153e-05,
      "loss": 0.0,
      "step": 3176
    },
    {
      "epoch": 0.5343537128921033,
      "grad_norm": NaN,
      "learning_rate": 4.585270820622338e-05,
      "loss": 0.0,
      "step": 3177
    },
    {
      "epoch": 0.5345219073248676,
      "grad_norm": NaN,
      "learning_rate": 4.582592722483761e-05,
      "loss": 0.0,
      "step": 3178
    },
    {
      "epoch": 0.5346901017576319,
      "grad_norm": NaN,
      "learning_rate": 4.579914744929472e-05,
      "loss": 0.0,
      "step": 3179
    },
    {
      "epoch": 0.534858296190396,
      "grad_norm": NaN,
      "learning_rate": 4.577236888733105e-05,
      "loss": 0.0,
      "step": 3180
    },
    {
      "epoch": 0.5350264906231603,
      "grad_norm": NaN,
      "learning_rate": 4.5745591546682674e-05,
      "loss": 0.0,
      "step": 3181
    },
    {
      "epoch": 0.5351946850559246,
      "grad_norm": NaN,
      "learning_rate": 4.571881543508526e-05,
      "loss": 0.0,
      "step": 3182
    },
    {
      "epoch": 0.5353628794886889,
      "grad_norm": NaN,
      "learning_rate": 4.569204056027411e-05,
      "loss": 0.0,
      "step": 3183
    },
    {
      "epoch": 0.5355310739214532,
      "grad_norm": NaN,
      "learning_rate": 4.566526692998416e-05,
      "loss": 0.0,
      "step": 3184
    },
    {
      "epoch": 0.5356992683542174,
      "grad_norm": NaN,
      "learning_rate": 4.563849455195007e-05,
      "loss": 0.0,
      "step": 3185
    },
    {
      "epoch": 0.5358674627869817,
      "grad_norm": NaN,
      "learning_rate": 4.5611723433906056e-05,
      "loss": 0.0,
      "step": 3186
    },
    {
      "epoch": 0.536035657219746,
      "grad_norm": NaN,
      "learning_rate": 4.5584953583585985e-05,
      "loss": 0.0,
      "step": 3187
    },
    {
      "epoch": 0.5362038516525103,
      "grad_norm": NaN,
      "learning_rate": 4.555818500872336e-05,
      "loss": 0.0,
      "step": 3188
    },
    {
      "epoch": 0.5363720460852746,
      "grad_norm": NaN,
      "learning_rate": 4.553141771705135e-05,
      "loss": 0.0,
      "step": 3189
    },
    {
      "epoch": 0.5365402405180388,
      "grad_norm": NaN,
      "learning_rate": 4.550465171630271e-05,
      "loss": 0.0,
      "step": 3190
    },
    {
      "epoch": 0.5367084349508031,
      "grad_norm": NaN,
      "learning_rate": 4.547788701420984e-05,
      "loss": 0.0,
      "step": 3191
    },
    {
      "epoch": 0.5368766293835674,
      "grad_norm": NaN,
      "learning_rate": 4.5451123618504744e-05,
      "loss": 0.0,
      "step": 3192
    },
    {
      "epoch": 0.5370448238163317,
      "grad_norm": NaN,
      "learning_rate": 4.54243615369191e-05,
      "loss": 0.0,
      "step": 3193
    },
    {
      "epoch": 0.537213018249096,
      "grad_norm": NaN,
      "learning_rate": 4.539760077718416e-05,
      "loss": 0.0,
      "step": 3194
    },
    {
      "epoch": 0.5373812126818602,
      "grad_norm": NaN,
      "learning_rate": 4.5370841347030796e-05,
      "loss": 0.0,
      "step": 3195
    },
    {
      "epoch": 0.5375494071146245,
      "grad_norm": NaN,
      "learning_rate": 4.534408325418952e-05,
      "loss": 0.0,
      "step": 3196
    },
    {
      "epoch": 0.5377176015473888,
      "grad_norm": NaN,
      "learning_rate": 4.531732650639045e-05,
      "loss": 0.0,
      "step": 3197
    },
    {
      "epoch": 0.5378857959801531,
      "grad_norm": NaN,
      "learning_rate": 4.5290571111363295e-05,
      "loss": 0.0,
      "step": 3198
    },
    {
      "epoch": 0.5380539904129173,
      "grad_norm": NaN,
      "learning_rate": 4.52638170768374e-05,
      "loss": 0.0,
      "step": 3199
    },
    {
      "epoch": 0.5382221848456816,
      "grad_norm": NaN,
      "learning_rate": 4.523706441054171e-05,
      "loss": 0.0,
      "step": 3200
    },
    {
      "epoch": 0.5383903792784459,
      "grad_norm": NaN,
      "learning_rate": 4.5210313120204735e-05,
      "loss": 0.0,
      "step": 3201
    },
    {
      "epoch": 0.5385585737112102,
      "grad_norm": NaN,
      "learning_rate": 4.5183563213554676e-05,
      "loss": 0.0,
      "step": 3202
    },
    {
      "epoch": 0.5387267681439745,
      "grad_norm": NaN,
      "learning_rate": 4.5156814698319264e-05,
      "loss": 0.0,
      "step": 3203
    },
    {
      "epoch": 0.5388949625767387,
      "grad_norm": NaN,
      "learning_rate": 4.513006758222583e-05,
      "loss": 0.0,
      "step": 3204
    },
    {
      "epoch": 0.539063157009503,
      "grad_norm": NaN,
      "learning_rate": 4.5103321873001314e-05,
      "loss": 0.0,
      "step": 3205
    },
    {
      "epoch": 0.5392313514422673,
      "grad_norm": NaN,
      "learning_rate": 4.507657757837228e-05,
      "loss": 0.0,
      "step": 3206
    },
    {
      "epoch": 0.5393995458750316,
      "grad_norm": NaN,
      "learning_rate": 4.5049834706064835e-05,
      "loss": 0.0,
      "step": 3207
    },
    {
      "epoch": 0.5395677403077959,
      "grad_norm": NaN,
      "learning_rate": 4.50230932638047e-05,
      "loss": 0.0,
      "step": 3208
    },
    {
      "epoch": 0.5397359347405601,
      "grad_norm": NaN,
      "learning_rate": 4.499635325931715e-05,
      "loss": 0.0,
      "step": 3209
    },
    {
      "epoch": 0.5399041291733243,
      "grad_norm": NaN,
      "learning_rate": 4.496961470032713e-05,
      "loss": 0.0,
      "step": 3210
    },
    {
      "epoch": 0.5400723236060886,
      "grad_norm": NaN,
      "learning_rate": 4.4942877594559067e-05,
      "loss": 0.0,
      "step": 3211
    },
    {
      "epoch": 0.5402405180388529,
      "grad_norm": NaN,
      "learning_rate": 4.4916141949737034e-05,
      "loss": 0.0,
      "step": 3212
    },
    {
      "epoch": 0.5404087124716171,
      "grad_norm": NaN,
      "learning_rate": 4.488940777358462e-05,
      "loss": 0.0,
      "step": 3213
    },
    {
      "epoch": 0.5405769069043814,
      "grad_norm": NaN,
      "learning_rate": 4.4862675073825086e-05,
      "loss": 0.0,
      "step": 3214
    },
    {
      "epoch": 0.5407451013371457,
      "grad_norm": NaN,
      "learning_rate": 4.483594385818118e-05,
      "loss": 0.0,
      "step": 3215
    },
    {
      "epoch": 0.54091329576991,
      "grad_norm": NaN,
      "learning_rate": 4.4809214134375254e-05,
      "loss": 0.0,
      "step": 3216
    },
    {
      "epoch": 0.5410814902026743,
      "grad_norm": NaN,
      "learning_rate": 4.478248591012921e-05,
      "loss": 0.0,
      "step": 3217
    },
    {
      "epoch": 0.5412496846354385,
      "grad_norm": NaN,
      "learning_rate": 4.4755759193164566e-05,
      "loss": 0.0,
      "step": 3218
    },
    {
      "epoch": 0.5414178790682028,
      "grad_norm": NaN,
      "learning_rate": 4.472903399120235e-05,
      "loss": 0.0,
      "step": 3219
    },
    {
      "epoch": 0.5415860735009671,
      "grad_norm": NaN,
      "learning_rate": 4.470231031196317e-05,
      "loss": 0.0,
      "step": 3220
    },
    {
      "epoch": 0.5417542679337314,
      "grad_norm": NaN,
      "learning_rate": 4.467558816316718e-05,
      "loss": 0.0,
      "step": 3221
    },
    {
      "epoch": 0.5419224623664957,
      "grad_norm": NaN,
      "learning_rate": 4.464886755253416e-05,
      "loss": 0.0,
      "step": 3222
    },
    {
      "epoch": 0.5420906567992599,
      "grad_norm": NaN,
      "learning_rate": 4.462214848778337e-05,
      "loss": 0.0,
      "step": 3223
    },
    {
      "epoch": 0.5422588512320242,
      "grad_norm": NaN,
      "learning_rate": 4.459543097663364e-05,
      "loss": 0.0,
      "step": 3224
    },
    {
      "epoch": 0.5424270456647885,
      "grad_norm": NaN,
      "learning_rate": 4.456871502680336e-05,
      "loss": 0.0,
      "step": 3225
    },
    {
      "epoch": 0.5425952400975528,
      "grad_norm": NaN,
      "learning_rate": 4.454200064601048e-05,
      "loss": 0.0,
      "step": 3226
    },
    {
      "epoch": 0.5427634345303171,
      "grad_norm": NaN,
      "learning_rate": 4.451528784197249e-05,
      "loss": 0.0,
      "step": 3227
    },
    {
      "epoch": 0.5429316289630813,
      "grad_norm": NaN,
      "learning_rate": 4.4488576622406404e-05,
      "loss": 0.0,
      "step": 3228
    },
    {
      "epoch": 0.5430998233958456,
      "grad_norm": NaN,
      "learning_rate": 4.4461866995028776e-05,
      "loss": 0.0,
      "step": 3229
    },
    {
      "epoch": 0.5432680178286099,
      "grad_norm": NaN,
      "learning_rate": 4.443515896755577e-05,
      "loss": 0.0,
      "step": 3230
    },
    {
      "epoch": 0.5434362122613742,
      "grad_norm": NaN,
      "learning_rate": 4.4408452547703005e-05,
      "loss": 0.0,
      "step": 3231
    },
    {
      "epoch": 0.5436044066941385,
      "grad_norm": NaN,
      "learning_rate": 4.4381747743185675e-05,
      "loss": 0.0,
      "step": 3232
    },
    {
      "epoch": 0.5437726011269027,
      "grad_norm": NaN,
      "learning_rate": 4.4355044561718476e-05,
      "loss": 0.0,
      "step": 3233
    },
    {
      "epoch": 0.543940795559667,
      "grad_norm": NaN,
      "learning_rate": 4.4328343011015695e-05,
      "loss": 0.0,
      "step": 3234
    },
    {
      "epoch": 0.5441089899924313,
      "grad_norm": NaN,
      "learning_rate": 4.4301643098791086e-05,
      "loss": 0.0,
      "step": 3235
    },
    {
      "epoch": 0.5442771844251956,
      "grad_norm": NaN,
      "learning_rate": 4.427494483275796e-05,
      "loss": 0.0,
      "step": 3236
    },
    {
      "epoch": 0.5444453788579598,
      "grad_norm": NaN,
      "learning_rate": 4.4248248220629124e-05,
      "loss": 0.0,
      "step": 3237
    },
    {
      "epoch": 0.5446135732907241,
      "grad_norm": NaN,
      "learning_rate": 4.422155327011698e-05,
      "loss": 0.0,
      "step": 3238
    },
    {
      "epoch": 0.5447817677234883,
      "grad_norm": NaN,
      "learning_rate": 4.4194859988933376e-05,
      "loss": 0.0,
      "step": 3239
    },
    {
      "epoch": 0.5449499621562526,
      "grad_norm": NaN,
      "learning_rate": 4.4168168384789694e-05,
      "loss": 0.0,
      "step": 3240
    },
    {
      "epoch": 0.5451181565890169,
      "grad_norm": NaN,
      "learning_rate": 4.4141478465396836e-05,
      "loss": 0.0,
      "step": 3241
    },
    {
      "epoch": 0.5452863510217811,
      "grad_norm": NaN,
      "learning_rate": 4.411479023846524e-05,
      "loss": 0.0,
      "step": 3242
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": NaN,
      "learning_rate": 4.408810371170483e-05,
      "loss": 0.0,
      "step": 3243
    },
    {
      "epoch": 0.5456227398873097,
      "grad_norm": NaN,
      "learning_rate": 4.406141889282504e-05,
      "loss": 0.0,
      "step": 3244
    },
    {
      "epoch": 0.545790934320074,
      "grad_norm": NaN,
      "learning_rate": 4.4034735789534795e-05,
      "loss": 0.0,
      "step": 3245
    },
    {
      "epoch": 0.5459591287528383,
      "grad_norm": NaN,
      "learning_rate": 4.400805440954258e-05,
      "loss": 0.0,
      "step": 3246
    },
    {
      "epoch": 0.5461273231856025,
      "grad_norm": NaN,
      "learning_rate": 4.398137476055634e-05,
      "loss": 0.0,
      "step": 3247
    },
    {
      "epoch": 0.5462955176183668,
      "grad_norm": NaN,
      "learning_rate": 4.395469685028352e-05,
      "loss": 0.0,
      "step": 3248
    },
    {
      "epoch": 0.5464637120511311,
      "grad_norm": NaN,
      "learning_rate": 4.392802068643106e-05,
      "loss": 0.0,
      "step": 3249
    },
    {
      "epoch": 0.5466319064838954,
      "grad_norm": NaN,
      "learning_rate": 4.39013462767054e-05,
      "loss": 0.0,
      "step": 3250
    },
    {
      "epoch": 0.5468001009166596,
      "grad_norm": NaN,
      "learning_rate": 4.3874673628812515e-05,
      "loss": 0.0,
      "step": 3251
    },
    {
      "epoch": 0.5469682953494239,
      "grad_norm": NaN,
      "learning_rate": 4.38480027504578e-05,
      "loss": 0.0,
      "step": 3252
    },
    {
      "epoch": 0.5471364897821882,
      "grad_norm": NaN,
      "learning_rate": 4.382133364934619e-05,
      "loss": 0.0,
      "step": 3253
    },
    {
      "epoch": 0.5473046842149525,
      "grad_norm": NaN,
      "learning_rate": 4.379466633318208e-05,
      "loss": 0.0,
      "step": 3254
    },
    {
      "epoch": 0.5474728786477168,
      "grad_norm": NaN,
      "learning_rate": 4.376800080966936e-05,
      "loss": 0.0,
      "step": 3255
    },
    {
      "epoch": 0.547641073080481,
      "grad_norm": NaN,
      "learning_rate": 4.3741337086511414e-05,
      "loss": 0.0,
      "step": 3256
    },
    {
      "epoch": 0.5478092675132453,
      "grad_norm": NaN,
      "learning_rate": 4.371467517141108e-05,
      "loss": 0.0,
      "step": 3257
    },
    {
      "epoch": 0.5479774619460096,
      "grad_norm": NaN,
      "learning_rate": 4.368801507207067e-05,
      "loss": 0.0,
      "step": 3258
    },
    {
      "epoch": 0.5481456563787739,
      "grad_norm": NaN,
      "learning_rate": 4.3661356796192025e-05,
      "loss": 0.0,
      "step": 3259
    },
    {
      "epoch": 0.5483138508115382,
      "grad_norm": NaN,
      "learning_rate": 4.363470035147641e-05,
      "loss": 0.0,
      "step": 3260
    },
    {
      "epoch": 0.5484820452443024,
      "grad_norm": NaN,
      "learning_rate": 4.360804574562456e-05,
      "loss": 0.0,
      "step": 3261
    },
    {
      "epoch": 0.5486502396770667,
      "grad_norm": NaN,
      "learning_rate": 4.358139298633669e-05,
      "loss": 0.0,
      "step": 3262
    },
    {
      "epoch": 0.548818434109831,
      "grad_norm": NaN,
      "learning_rate": 4.355474208131251e-05,
      "loss": 0.0,
      "step": 3263
    },
    {
      "epoch": 0.5489866285425953,
      "grad_norm": NaN,
      "learning_rate": 4.352809303825115e-05,
      "loss": 0.0,
      "step": 3264
    },
    {
      "epoch": 0.5491548229753596,
      "grad_norm": NaN,
      "learning_rate": 4.350144586485122e-05,
      "loss": 0.0,
      "step": 3265
    },
    {
      "epoch": 0.5493230174081238,
      "grad_norm": NaN,
      "learning_rate": 4.347480056881077e-05,
      "loss": 0.0,
      "step": 3266
    },
    {
      "epoch": 0.5494912118408881,
      "grad_norm": NaN,
      "learning_rate": 4.3448157157827376e-05,
      "loss": 0.0,
      "step": 3267
    },
    {
      "epoch": 0.5496594062736524,
      "grad_norm": NaN,
      "learning_rate": 4.342151563959799e-05,
      "loss": 0.0,
      "step": 3268
    },
    {
      "epoch": 0.5498276007064166,
      "grad_norm": NaN,
      "learning_rate": 4.3394876021819055e-05,
      "loss": 0.0,
      "step": 3269
    },
    {
      "epoch": 0.5499957951391808,
      "grad_norm": NaN,
      "learning_rate": 4.336823831218644e-05,
      "loss": 0.0,
      "step": 3270
    },
    {
      "epoch": 0.5501639895719451,
      "grad_norm": NaN,
      "learning_rate": 4.334160251839551e-05,
      "loss": 0.0,
      "step": 3271
    },
    {
      "epoch": 0.5503321840047094,
      "grad_norm": NaN,
      "learning_rate": 4.331496864814103e-05,
      "loss": 0.0,
      "step": 3272
    },
    {
      "epoch": 0.5505003784374737,
      "grad_norm": NaN,
      "learning_rate": 4.328833670911724e-05,
      "loss": 0.0,
      "step": 3273
    },
    {
      "epoch": 0.550668572870238,
      "grad_norm": NaN,
      "learning_rate": 4.3261706709017774e-05,
      "loss": 0.0,
      "step": 3274
    },
    {
      "epoch": 0.5508367673030022,
      "grad_norm": NaN,
      "learning_rate": 4.323507865553579e-05,
      "loss": 0.0,
      "step": 3275
    },
    {
      "epoch": 0.5510049617357665,
      "grad_norm": NaN,
      "learning_rate": 4.3208452556363807e-05,
      "loss": 0.0,
      "step": 3276
    },
    {
      "epoch": 0.5511731561685308,
      "grad_norm": NaN,
      "learning_rate": 4.318182841919382e-05,
      "loss": 0.0,
      "step": 3277
    },
    {
      "epoch": 0.5513413506012951,
      "grad_norm": NaN,
      "learning_rate": 4.31552062517172e-05,
      "loss": 0.0,
      "step": 3278
    },
    {
      "epoch": 0.5515095450340594,
      "grad_norm": NaN,
      "learning_rate": 4.312858606162485e-05,
      "loss": 0.0,
      "step": 3279
    },
    {
      "epoch": 0.5516777394668236,
      "grad_norm": NaN,
      "learning_rate": 4.310196785660703e-05,
      "loss": 0.0,
      "step": 3280
    },
    {
      "epoch": 0.5518459338995879,
      "grad_norm": NaN,
      "learning_rate": 4.307535164435344e-05,
      "loss": 0.0,
      "step": 3281
    },
    {
      "epoch": 0.5520141283323522,
      "grad_norm": NaN,
      "learning_rate": 4.304873743255318e-05,
      "loss": 0.0,
      "step": 3282
    },
    {
      "epoch": 0.5521823227651165,
      "grad_norm": NaN,
      "learning_rate": 4.302212522889482e-05,
      "loss": 0.0,
      "step": 3283
    },
    {
      "epoch": 0.5523505171978808,
      "grad_norm": NaN,
      "learning_rate": 4.299551504106634e-05,
      "loss": 0.0,
      "step": 3284
    },
    {
      "epoch": 0.552518711630645,
      "grad_norm": NaN,
      "learning_rate": 4.29689068767551e-05,
      "loss": 0.0,
      "step": 3285
    },
    {
      "epoch": 0.5526869060634093,
      "grad_norm": NaN,
      "learning_rate": 4.294230074364789e-05,
      "loss": 0.0,
      "step": 3286
    },
    {
      "epoch": 0.5528551004961736,
      "grad_norm": NaN,
      "learning_rate": 4.2915696649430964e-05,
      "loss": 0.0,
      "step": 3287
    },
    {
      "epoch": 0.5530232949289379,
      "grad_norm": NaN,
      "learning_rate": 4.288909460178993e-05,
      "loss": 0.0,
      "step": 3288
    },
    {
      "epoch": 0.5531914893617021,
      "grad_norm": NaN,
      "learning_rate": 4.286249460840982e-05,
      "loss": 0.0,
      "step": 3289
    },
    {
      "epoch": 0.5533596837944664,
      "grad_norm": NaN,
      "learning_rate": 4.283589667697506e-05,
      "loss": 0.0,
      "step": 3290
    },
    {
      "epoch": 0.5535278782272307,
      "grad_norm": NaN,
      "learning_rate": 4.280930081516952e-05,
      "loss": 0.0,
      "step": 3291
    },
    {
      "epoch": 0.553696072659995,
      "grad_norm": NaN,
      "learning_rate": 4.278270703067644e-05,
      "loss": 0.0,
      "step": 3292
    },
    {
      "epoch": 0.5538642670927593,
      "grad_norm": NaN,
      "learning_rate": 4.2756115331178454e-05,
      "loss": 0.0,
      "step": 3293
    },
    {
      "epoch": 0.5540324615255235,
      "grad_norm": NaN,
      "learning_rate": 4.2729525724357585e-05,
      "loss": 0.0,
      "step": 3294
    },
    {
      "epoch": 0.5542006559582878,
      "grad_norm": NaN,
      "learning_rate": 4.2702938217895334e-05,
      "loss": 0.0,
      "step": 3295
    },
    {
      "epoch": 0.5543688503910521,
      "grad_norm": NaN,
      "learning_rate": 4.267635281947249e-05,
      "loss": 0.0,
      "step": 3296
    },
    {
      "epoch": 0.5545370448238164,
      "grad_norm": NaN,
      "learning_rate": 4.264976953676929e-05,
      "loss": 0.0,
      "step": 3297
    },
    {
      "epoch": 0.5547052392565807,
      "grad_norm": NaN,
      "learning_rate": 4.262318837746535e-05,
      "loss": 0.0,
      "step": 3298
    },
    {
      "epoch": 0.5548734336893448,
      "grad_norm": NaN,
      "learning_rate": 4.259660934923965e-05,
      "loss": 0.0,
      "step": 3299
    },
    {
      "epoch": 0.5550416281221091,
      "grad_norm": NaN,
      "learning_rate": 4.257003245977059e-05,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 0.5552098225548734,
      "grad_norm": NaN,
      "learning_rate": 4.2543457716735935e-05,
      "loss": 0.0,
      "step": 3301
    },
    {
      "epoch": 0.5553780169876377,
      "grad_norm": NaN,
      "learning_rate": 4.251688512781283e-05,
      "loss": 0.0,
      "step": 3302
    },
    {
      "epoch": 0.555546211420402,
      "grad_norm": NaN,
      "learning_rate": 4.249031470067778e-05,
      "loss": 0.0,
      "step": 3303
    },
    {
      "epoch": 0.5557144058531662,
      "grad_norm": NaN,
      "learning_rate": 4.2463746443006706e-05,
      "loss": 0.0,
      "step": 3304
    },
    {
      "epoch": 0.5558826002859305,
      "grad_norm": NaN,
      "learning_rate": 4.243718036247488e-05,
      "loss": 0.0,
      "step": 3305
    },
    {
      "epoch": 0.5560507947186948,
      "grad_norm": NaN,
      "learning_rate": 4.241061646675695e-05,
      "loss": 0.0,
      "step": 3306
    },
    {
      "epoch": 0.5562189891514591,
      "grad_norm": NaN,
      "learning_rate": 4.238405476352689e-05,
      "loss": 0.0,
      "step": 3307
    },
    {
      "epoch": 0.5563871835842233,
      "grad_norm": NaN,
      "learning_rate": 4.235749526045813e-05,
      "loss": 0.0,
      "step": 3308
    },
    {
      "epoch": 0.5565553780169876,
      "grad_norm": NaN,
      "learning_rate": 4.233093796522338e-05,
      "loss": 0.0,
      "step": 3309
    },
    {
      "epoch": 0.5567235724497519,
      "grad_norm": NaN,
      "learning_rate": 4.2304382885494753e-05,
      "loss": 0.0,
      "step": 3310
    },
    {
      "epoch": 0.5568917668825162,
      "grad_norm": NaN,
      "learning_rate": 4.227783002894369e-05,
      "loss": 0.0,
      "step": 3311
    },
    {
      "epoch": 0.5570599613152805,
      "grad_norm": NaN,
      "learning_rate": 4.225127940324107e-05,
      "loss": 0.0,
      "step": 3312
    },
    {
      "epoch": 0.5572281557480447,
      "grad_norm": NaN,
      "learning_rate": 4.222473101605703e-05,
      "loss": 0.0,
      "step": 3313
    },
    {
      "epoch": 0.557396350180809,
      "grad_norm": NaN,
      "learning_rate": 4.219818487506112e-05,
      "loss": 0.0,
      "step": 3314
    },
    {
      "epoch": 0.5575645446135733,
      "grad_norm": NaN,
      "learning_rate": 4.2171640987922184e-05,
      "loss": 0.0,
      "step": 3315
    },
    {
      "epoch": 0.5577327390463376,
      "grad_norm": NaN,
      "learning_rate": 4.214509936230851e-05,
      "loss": 0.0,
      "step": 3316
    },
    {
      "epoch": 0.5579009334791019,
      "grad_norm": NaN,
      "learning_rate": 4.2118560005887656e-05,
      "loss": 0.0,
      "step": 3317
    },
    {
      "epoch": 0.5580691279118661,
      "grad_norm": NaN,
      "learning_rate": 4.209202292632653e-05,
      "loss": 0.0,
      "step": 3318
    },
    {
      "epoch": 0.5582373223446304,
      "grad_norm": NaN,
      "learning_rate": 4.20654881312914e-05,
      "loss": 0.0,
      "step": 3319
    },
    {
      "epoch": 0.5584055167773947,
      "grad_norm": NaN,
      "learning_rate": 4.203895562844789e-05,
      "loss": 0.0,
      "step": 3320
    },
    {
      "epoch": 0.558573711210159,
      "grad_norm": NaN,
      "learning_rate": 4.201242542546092e-05,
      "loss": 0.0,
      "step": 3321
    },
    {
      "epoch": 0.5587419056429233,
      "grad_norm": NaN,
      "learning_rate": 4.198589752999479e-05,
      "loss": 0.0,
      "step": 3322
    },
    {
      "epoch": 0.5589101000756875,
      "grad_norm": NaN,
      "learning_rate": 4.195937194971307e-05,
      "loss": 0.0,
      "step": 3323
    },
    {
      "epoch": 0.5590782945084518,
      "grad_norm": NaN,
      "learning_rate": 4.193284869227876e-05,
      "loss": 0.0,
      "step": 3324
    },
    {
      "epoch": 0.5592464889412161,
      "grad_norm": NaN,
      "learning_rate": 4.1906327765354105e-05,
      "loss": 0.0,
      "step": 3325
    },
    {
      "epoch": 0.5594146833739804,
      "grad_norm": NaN,
      "learning_rate": 4.18798091766007e-05,
      "loss": 0.0,
      "step": 3326
    },
    {
      "epoch": 0.5595828778067446,
      "grad_norm": NaN,
      "learning_rate": 4.185329293367947e-05,
      "loss": 0.0,
      "step": 3327
    },
    {
      "epoch": 0.5597510722395088,
      "grad_norm": NaN,
      "learning_rate": 4.182677904425067e-05,
      "loss": 0.0,
      "step": 3328
    },
    {
      "epoch": 0.5599192666722731,
      "grad_norm": NaN,
      "learning_rate": 4.180026751597386e-05,
      "loss": 0.0,
      "step": 3329
    },
    {
      "epoch": 0.5600874611050374,
      "grad_norm": NaN,
      "learning_rate": 4.177375835650792e-05,
      "loss": 0.0,
      "step": 3330
    },
    {
      "epoch": 0.5602556555378017,
      "grad_norm": NaN,
      "learning_rate": 4.174725157351103e-05,
      "loss": 0.0,
      "step": 3331
    },
    {
      "epoch": 0.5604238499705659,
      "grad_norm": NaN,
      "learning_rate": 4.172074717464075e-05,
      "loss": 0.0,
      "step": 3332
    },
    {
      "epoch": 0.5605920444033302,
      "grad_norm": NaN,
      "learning_rate": 4.169424516755387e-05,
      "loss": 0.0,
      "step": 3333
    },
    {
      "epoch": 0.5607602388360945,
      "grad_norm": NaN,
      "learning_rate": 4.166774555990654e-05,
      "loss": 0.0,
      "step": 3334
    },
    {
      "epoch": 0.5609284332688588,
      "grad_norm": NaN,
      "learning_rate": 4.164124835935418e-05,
      "loss": 0.0,
      "step": 3335
    },
    {
      "epoch": 0.561096627701623,
      "grad_norm": NaN,
      "learning_rate": 4.161475357355155e-05,
      "loss": 0.0,
      "step": 3336
    },
    {
      "epoch": 0.5612648221343873,
      "grad_norm": NaN,
      "learning_rate": 4.15882612101527e-05,
      "loss": 0.0,
      "step": 3337
    },
    {
      "epoch": 0.5614330165671516,
      "grad_norm": NaN,
      "learning_rate": 4.156177127681096e-05,
      "loss": 0.0,
      "step": 3338
    },
    {
      "epoch": 0.5616012109999159,
      "grad_norm": NaN,
      "learning_rate": 4.153528378117897e-05,
      "loss": 0.0,
      "step": 3339
    },
    {
      "epoch": 0.5617694054326802,
      "grad_norm": NaN,
      "learning_rate": 4.150879873090871e-05,
      "loss": 0.0,
      "step": 3340
    },
    {
      "epoch": 0.5619375998654444,
      "grad_norm": NaN,
      "learning_rate": 4.1482316133651375e-05,
      "loss": 0.0,
      "step": 3341
    },
    {
      "epoch": 0.5621057942982087,
      "grad_norm": NaN,
      "learning_rate": 4.1455835997057525e-05,
      "loss": 0.0,
      "step": 3342
    },
    {
      "epoch": 0.562273988730973,
      "grad_norm": NaN,
      "learning_rate": 4.142935832877692e-05,
      "loss": 0.0,
      "step": 3343
    },
    {
      "epoch": 0.5624421831637373,
      "grad_norm": NaN,
      "learning_rate": 4.1402883136458725e-05,
      "loss": 0.0,
      "step": 3344
    },
    {
      "epoch": 0.5626103775965016,
      "grad_norm": NaN,
      "learning_rate": 4.137641042775129e-05,
      "loss": 0.0,
      "step": 3345
    },
    {
      "epoch": 0.5627785720292658,
      "grad_norm": NaN,
      "learning_rate": 4.134994021030231e-05,
      "loss": 0.0,
      "step": 3346
    },
    {
      "epoch": 0.5629467664620301,
      "grad_norm": NaN,
      "learning_rate": 4.1323472491758694e-05,
      "loss": 0.0,
      "step": 3347
    },
    {
      "epoch": 0.5631149608947944,
      "grad_norm": NaN,
      "learning_rate": 4.12970072797667e-05,
      "loss": 0.0,
      "step": 3348
    },
    {
      "epoch": 0.5632831553275587,
      "grad_norm": NaN,
      "learning_rate": 4.1270544581971826e-05,
      "loss": 0.0,
      "step": 3349
    },
    {
      "epoch": 0.563451349760323,
      "grad_norm": NaN,
      "learning_rate": 4.124408440601885e-05,
      "loss": 0.0,
      "step": 3350
    },
    {
      "epoch": 0.5636195441930872,
      "grad_norm": NaN,
      "learning_rate": 4.121762675955182e-05,
      "loss": 0.0,
      "step": 3351
    },
    {
      "epoch": 0.5637877386258515,
      "grad_norm": NaN,
      "learning_rate": 4.119117165021402e-05,
      "loss": 0.0,
      "step": 3352
    },
    {
      "epoch": 0.5639559330586158,
      "grad_norm": NaN,
      "learning_rate": 4.11647190856481e-05,
      "loss": 0.0,
      "step": 3353
    },
    {
      "epoch": 0.5641241274913801,
      "grad_norm": NaN,
      "learning_rate": 4.1138269073495864e-05,
      "loss": 0.0,
      "step": 3354
    },
    {
      "epoch": 0.5642923219241444,
      "grad_norm": NaN,
      "learning_rate": 4.1111821621398446e-05,
      "loss": 0.0,
      "step": 3355
    },
    {
      "epoch": 0.5644605163569086,
      "grad_norm": NaN,
      "learning_rate": 4.108537673699619e-05,
      "loss": 0.0,
      "step": 3356
    },
    {
      "epoch": 0.5646287107896729,
      "grad_norm": NaN,
      "learning_rate": 4.105893442792876e-05,
      "loss": 0.0,
      "step": 3357
    },
    {
      "epoch": 0.5647969052224371,
      "grad_norm": NaN,
      "learning_rate": 4.103249470183502e-05,
      "loss": 0.0,
      "step": 3358
    },
    {
      "epoch": 0.5649650996552014,
      "grad_norm": NaN,
      "learning_rate": 4.1006057566353126e-05,
      "loss": 0.0,
      "step": 3359
    },
    {
      "epoch": 0.5651332940879656,
      "grad_norm": NaN,
      "learning_rate": 4.0979623029120436e-05,
      "loss": 0.0,
      "step": 3360
    },
    {
      "epoch": 0.5653014885207299,
      "grad_norm": NaN,
      "learning_rate": 4.0953191097773645e-05,
      "loss": 0.0,
      "step": 3361
    },
    {
      "epoch": 0.5654696829534942,
      "grad_norm": NaN,
      "learning_rate": 4.092676177994862e-05,
      "loss": 0.0,
      "step": 3362
    },
    {
      "epoch": 0.5656378773862585,
      "grad_norm": NaN,
      "learning_rate": 4.0900335083280484e-05,
      "loss": 0.0,
      "step": 3363
    },
    {
      "epoch": 0.5658060718190228,
      "grad_norm": NaN,
      "learning_rate": 4.087391101540361e-05,
      "loss": 0.0,
      "step": 3364
    },
    {
      "epoch": 0.565974266251787,
      "grad_norm": NaN,
      "learning_rate": 4.084748958395164e-05,
      "loss": 0.0,
      "step": 3365
    },
    {
      "epoch": 0.5661424606845513,
      "grad_norm": NaN,
      "learning_rate": 4.082107079655741e-05,
      "loss": 0.0,
      "step": 3366
    },
    {
      "epoch": 0.5663106551173156,
      "grad_norm": NaN,
      "learning_rate": 4.0794654660853024e-05,
      "loss": 0.0,
      "step": 3367
    },
    {
      "epoch": 0.5664788495500799,
      "grad_norm": NaN,
      "learning_rate": 4.0768241184469775e-05,
      "loss": 0.0,
      "step": 3368
    },
    {
      "epoch": 0.5666470439828442,
      "grad_norm": NaN,
      "learning_rate": 4.074183037503827e-05,
      "loss": 0.0,
      "step": 3369
    },
    {
      "epoch": 0.5668152384156084,
      "grad_norm": NaN,
      "learning_rate": 4.071542224018827e-05,
      "loss": 0.0,
      "step": 3370
    },
    {
      "epoch": 0.5669834328483727,
      "grad_norm": NaN,
      "learning_rate": 4.068901678754879e-05,
      "loss": 0.0,
      "step": 3371
    },
    {
      "epoch": 0.567151627281137,
      "grad_norm": NaN,
      "learning_rate": 4.066261402474805e-05,
      "loss": 0.0,
      "step": 3372
    },
    {
      "epoch": 0.5673198217139013,
      "grad_norm": NaN,
      "learning_rate": 4.063621395941356e-05,
      "loss": 0.0,
      "step": 3373
    },
    {
      "epoch": 0.5674880161466656,
      "grad_norm": NaN,
      "learning_rate": 4.0609816599171966e-05,
      "loss": 0.0,
      "step": 3374
    },
    {
      "epoch": 0.5676562105794298,
      "grad_norm": NaN,
      "learning_rate": 4.058342195164919e-05,
      "loss": 0.0,
      "step": 3375
    },
    {
      "epoch": 0.5678244050121941,
      "grad_norm": NaN,
      "learning_rate": 4.055703002447033e-05,
      "loss": 0.0,
      "step": 3376
    },
    {
      "epoch": 0.5679925994449584,
      "grad_norm": NaN,
      "learning_rate": 4.053064082525974e-05,
      "loss": 0.0,
      "step": 3377
    },
    {
      "epoch": 0.5681607938777227,
      "grad_norm": NaN,
      "learning_rate": 4.0504254361640946e-05,
      "loss": 0.0,
      "step": 3378
    },
    {
      "epoch": 0.568328988310487,
      "grad_norm": NaN,
      "learning_rate": 4.047787064123671e-05,
      "loss": 0.0,
      "step": 3379
    },
    {
      "epoch": 0.5684971827432512,
      "grad_norm": NaN,
      "learning_rate": 4.045148967166897e-05,
      "loss": 0.0,
      "step": 3380
    },
    {
      "epoch": 0.5686653771760155,
      "grad_norm": NaN,
      "learning_rate": 4.042511146055894e-05,
      "loss": 0.0,
      "step": 3381
    },
    {
      "epoch": 0.5688335716087798,
      "grad_norm": NaN,
      "learning_rate": 4.039873601552696e-05,
      "loss": 0.0,
      "step": 3382
    },
    {
      "epoch": 0.5690017660415441,
      "grad_norm": NaN,
      "learning_rate": 4.037236334419261e-05,
      "loss": 0.0,
      "step": 3383
    },
    {
      "epoch": 0.5691699604743083,
      "grad_norm": NaN,
      "learning_rate": 4.034599345417464e-05,
      "loss": 0.0,
      "step": 3384
    },
    {
      "epoch": 0.5693381549070726,
      "grad_norm": NaN,
      "learning_rate": 4.031962635309104e-05,
      "loss": 0.0,
      "step": 3385
    },
    {
      "epoch": 0.5695063493398369,
      "grad_norm": NaN,
      "learning_rate": 4.029326204855896e-05,
      "loss": 0.0,
      "step": 3386
    },
    {
      "epoch": 0.5696745437726011,
      "grad_norm": NaN,
      "learning_rate": 4.0266900548194755e-05,
      "loss": 0.0,
      "step": 3387
    },
    {
      "epoch": 0.5698427382053654,
      "grad_norm": NaN,
      "learning_rate": 4.024054185961394e-05,
      "loss": 0.0,
      "step": 3388
    },
    {
      "epoch": 0.5700109326381296,
      "grad_norm": NaN,
      "learning_rate": 4.02141859904313e-05,
      "loss": 0.0,
      "step": 3389
    },
    {
      "epoch": 0.5701791270708939,
      "grad_norm": NaN,
      "learning_rate": 4.0187832948260705e-05,
      "loss": 0.0,
      "step": 3390
    },
    {
      "epoch": 0.5703473215036582,
      "grad_norm": NaN,
      "learning_rate": 4.016148274071528e-05,
      "loss": 0.0,
      "step": 3391
    },
    {
      "epoch": 0.5705155159364225,
      "grad_norm": NaN,
      "learning_rate": 4.0135135375407276e-05,
      "loss": 0.0,
      "step": 3392
    },
    {
      "epoch": 0.5706837103691867,
      "grad_norm": NaN,
      "learning_rate": 4.0108790859948184e-05,
      "loss": 0.0,
      "step": 3393
    },
    {
      "epoch": 0.570851904801951,
      "grad_norm": NaN,
      "learning_rate": 4.008244920194862e-05,
      "loss": 0.0,
      "step": 3394
    },
    {
      "epoch": 0.5710200992347153,
      "grad_norm": NaN,
      "learning_rate": 4.005611040901841e-05,
      "loss": 0.0,
      "step": 3395
    },
    {
      "epoch": 0.5711882936674796,
      "grad_norm": NaN,
      "learning_rate": 4.0029774488766494e-05,
      "loss": 0.0,
      "step": 3396
    },
    {
      "epoch": 0.5713564881002439,
      "grad_norm": NaN,
      "learning_rate": 4.000344144880108e-05,
      "loss": 0.0,
      "step": 3397
    },
    {
      "epoch": 0.5715246825330081,
      "grad_norm": NaN,
      "learning_rate": 3.997711129672946e-05,
      "loss": 0.0,
      "step": 3398
    },
    {
      "epoch": 0.5716928769657724,
      "grad_norm": NaN,
      "learning_rate": 3.995078404015813e-05,
      "loss": 0.0,
      "step": 3399
    },
    {
      "epoch": 0.5718610713985367,
      "grad_norm": NaN,
      "learning_rate": 3.992445968669273e-05,
      "loss": 0.0,
      "step": 3400
    },
    {
      "epoch": 0.572029265831301,
      "grad_norm": NaN,
      "learning_rate": 3.989813824393806e-05,
      "loss": 0.0,
      "step": 3401
    },
    {
      "epoch": 0.5721974602640653,
      "grad_norm": NaN,
      "learning_rate": 3.9871819719498116e-05,
      "loss": 0.0,
      "step": 3402
    },
    {
      "epoch": 0.5723656546968295,
      "grad_norm": NaN,
      "learning_rate": 3.984550412097601e-05,
      "loss": 0.0,
      "step": 3403
    },
    {
      "epoch": 0.5725338491295938,
      "grad_norm": NaN,
      "learning_rate": 3.981919145597404e-05,
      "loss": 0.0,
      "step": 3404
    },
    {
      "epoch": 0.5727020435623581,
      "grad_norm": NaN,
      "learning_rate": 3.979288173209359e-05,
      "loss": 0.0,
      "step": 3405
    },
    {
      "epoch": 0.5728702379951224,
      "grad_norm": NaN,
      "learning_rate": 3.976657495693531e-05,
      "loss": 0.0,
      "step": 3406
    },
    {
      "epoch": 0.5730384324278867,
      "grad_norm": NaN,
      "learning_rate": 3.97402711380989e-05,
      "loss": 0.0,
      "step": 3407
    },
    {
      "epoch": 0.5732066268606509,
      "grad_norm": NaN,
      "learning_rate": 3.9713970283183244e-05,
      "loss": 0.0,
      "step": 3408
    },
    {
      "epoch": 0.5733748212934152,
      "grad_norm": NaN,
      "learning_rate": 3.9687672399786344e-05,
      "loss": 0.0,
      "step": 3409
    },
    {
      "epoch": 0.5735430157261795,
      "grad_norm": NaN,
      "learning_rate": 3.96613774955054e-05,
      "loss": 0.0,
      "step": 3410
    },
    {
      "epoch": 0.5737112101589438,
      "grad_norm": NaN,
      "learning_rate": 3.96350855779367e-05,
      "loss": 0.0,
      "step": 3411
    },
    {
      "epoch": 0.573879404591708,
      "grad_norm": NaN,
      "learning_rate": 3.960879665467569e-05,
      "loss": 0.0,
      "step": 3412
    },
    {
      "epoch": 0.5740475990244723,
      "grad_norm": NaN,
      "learning_rate": 3.958251073331693e-05,
      "loss": 0.0,
      "step": 3413
    },
    {
      "epoch": 0.5742157934572366,
      "grad_norm": NaN,
      "learning_rate": 3.955622782145413e-05,
      "loss": 0.0,
      "step": 3414
    },
    {
      "epoch": 0.5743839878900009,
      "grad_norm": NaN,
      "learning_rate": 3.9529947926680144e-05,
      "loss": 0.0,
      "step": 3415
    },
    {
      "epoch": 0.5745521823227652,
      "grad_norm": NaN,
      "learning_rate": 3.950367105658693e-05,
      "loss": 0.0,
      "step": 3416
    },
    {
      "epoch": 0.5747203767555293,
      "grad_norm": NaN,
      "learning_rate": 3.9477397218765565e-05,
      "loss": 0.0,
      "step": 3417
    },
    {
      "epoch": 0.5748885711882936,
      "grad_norm": NaN,
      "learning_rate": 3.9451126420806304e-05,
      "loss": 0.0,
      "step": 3418
    },
    {
      "epoch": 0.5750567656210579,
      "grad_norm": NaN,
      "learning_rate": 3.942485867029847e-05,
      "loss": 0.0,
      "step": 3419
    },
    {
      "epoch": 0.5752249600538222,
      "grad_norm": NaN,
      "learning_rate": 3.939859397483051e-05,
      "loss": 0.0,
      "step": 3420
    },
    {
      "epoch": 0.5753931544865865,
      "grad_norm": NaN,
      "learning_rate": 3.937233234198999e-05,
      "loss": 0.0,
      "step": 3421
    },
    {
      "epoch": 0.5755613489193507,
      "grad_norm": NaN,
      "learning_rate": 3.9346073779363646e-05,
      "loss": 0.0,
      "step": 3422
    },
    {
      "epoch": 0.575729543352115,
      "grad_norm": NaN,
      "learning_rate": 3.931981829453725e-05,
      "loss": 0.0,
      "step": 3423
    },
    {
      "epoch": 0.5758977377848793,
      "grad_norm": NaN,
      "learning_rate": 3.929356589509572e-05,
      "loss": 0.0,
      "step": 3424
    },
    {
      "epoch": 0.5760659322176436,
      "grad_norm": NaN,
      "learning_rate": 3.926731658862307e-05,
      "loss": 0.0,
      "step": 3425
    },
    {
      "epoch": 0.5762341266504079,
      "grad_norm": NaN,
      "learning_rate": 3.924107038270246e-05,
      "loss": 0.0,
      "step": 3426
    },
    {
      "epoch": 0.5764023210831721,
      "grad_norm": NaN,
      "learning_rate": 3.921482728491611e-05,
      "loss": 0.0,
      "step": 3427
    },
    {
      "epoch": 0.5765705155159364,
      "grad_norm": NaN,
      "learning_rate": 3.918858730284536e-05,
      "loss": 0.0,
      "step": 3428
    },
    {
      "epoch": 0.5767387099487007,
      "grad_norm": NaN,
      "learning_rate": 3.9162350444070636e-05,
      "loss": 0.0,
      "step": 3429
    },
    {
      "epoch": 0.576906904381465,
      "grad_norm": NaN,
      "learning_rate": 3.913611671617148e-05,
      "loss": 0.0,
      "step": 3430
    },
    {
      "epoch": 0.5770750988142292,
      "grad_norm": NaN,
      "learning_rate": 3.910988612672653e-05,
      "loss": 0.0,
      "step": 3431
    },
    {
      "epoch": 0.5772432932469935,
      "grad_norm": NaN,
      "learning_rate": 3.90836586833135e-05,
      "loss": 0.0,
      "step": 3432
    },
    {
      "epoch": 0.5774114876797578,
      "grad_norm": NaN,
      "learning_rate": 3.905743439350918e-05,
      "loss": 0.0,
      "step": 3433
    },
    {
      "epoch": 0.5775796821125221,
      "grad_norm": NaN,
      "learning_rate": 3.903121326488953e-05,
      "loss": 0.0,
      "step": 3434
    },
    {
      "epoch": 0.5777478765452864,
      "grad_norm": NaN,
      "learning_rate": 3.900499530502951e-05,
      "loss": 0.0,
      "step": 3435
    },
    {
      "epoch": 0.5779160709780506,
      "grad_norm": NaN,
      "learning_rate": 3.897878052150319e-05,
      "loss": 0.0,
      "step": 3436
    },
    {
      "epoch": 0.5780842654108149,
      "grad_norm": NaN,
      "learning_rate": 3.8952568921883713e-05,
      "loss": 0.0,
      "step": 3437
    },
    {
      "epoch": 0.5782524598435792,
      "grad_norm": NaN,
      "learning_rate": 3.8926360513743356e-05,
      "loss": 0.0,
      "step": 3438
    },
    {
      "epoch": 0.5784206542763435,
      "grad_norm": NaN,
      "learning_rate": 3.890015530465342e-05,
      "loss": 0.0,
      "step": 3439
    },
    {
      "epoch": 0.5785888487091078,
      "grad_norm": NaN,
      "learning_rate": 3.887395330218429e-05,
      "loss": 0.0,
      "step": 3440
    },
    {
      "epoch": 0.578757043141872,
      "grad_norm": NaN,
      "learning_rate": 3.884775451390542e-05,
      "loss": 0.0,
      "step": 3441
    },
    {
      "epoch": 0.5789252375746363,
      "grad_norm": NaN,
      "learning_rate": 3.882155894738536e-05,
      "loss": 0.0,
      "step": 3442
    },
    {
      "epoch": 0.5790934320074006,
      "grad_norm": NaN,
      "learning_rate": 3.8795366610191725e-05,
      "loss": 0.0,
      "step": 3443
    },
    {
      "epoch": 0.5792616264401649,
      "grad_norm": NaN,
      "learning_rate": 3.876917750989117e-05,
      "loss": 0.0,
      "step": 3444
    },
    {
      "epoch": 0.5794298208729292,
      "grad_norm": NaN,
      "learning_rate": 3.8742991654049426e-05,
      "loss": 0.0,
      "step": 3445
    },
    {
      "epoch": 0.5795980153056934,
      "grad_norm": NaN,
      "learning_rate": 3.871680905023133e-05,
      "loss": 0.0,
      "step": 3446
    },
    {
      "epoch": 0.5797662097384576,
      "grad_norm": NaN,
      "learning_rate": 3.869062970600071e-05,
      "loss": 0.0,
      "step": 3447
    },
    {
      "epoch": 0.5799344041712219,
      "grad_norm": NaN,
      "learning_rate": 3.866445362892051e-05,
      "loss": 0.0,
      "step": 3448
    },
    {
      "epoch": 0.5801025986039862,
      "grad_norm": NaN,
      "learning_rate": 3.863828082655268e-05,
      "loss": 0.0,
      "step": 3449
    },
    {
      "epoch": 0.5802707930367504,
      "grad_norm": NaN,
      "learning_rate": 3.8612111306458254e-05,
      "loss": 0.0,
      "step": 3450
    },
    {
      "epoch": 0.5804389874695147,
      "grad_norm": NaN,
      "learning_rate": 3.8585945076197325e-05,
      "loss": 0.0,
      "step": 3451
    },
    {
      "epoch": 0.580607181902279,
      "grad_norm": NaN,
      "learning_rate": 3.855978214332903e-05,
      "loss": 0.0,
      "step": 3452
    },
    {
      "epoch": 0.5807753763350433,
      "grad_norm": NaN,
      "learning_rate": 3.8533622515411525e-05,
      "loss": 0.0,
      "step": 3453
    },
    {
      "epoch": 0.5809435707678076,
      "grad_norm": NaN,
      "learning_rate": 3.8507466200002027e-05,
      "loss": 0.0,
      "step": 3454
    },
    {
      "epoch": 0.5811117652005718,
      "grad_norm": NaN,
      "learning_rate": 3.8481313204656844e-05,
      "loss": 0.0,
      "step": 3455
    },
    {
      "epoch": 0.5812799596333361,
      "grad_norm": NaN,
      "learning_rate": 3.845516353693126e-05,
      "loss": 0.0,
      "step": 3456
    },
    {
      "epoch": 0.5814481540661004,
      "grad_norm": NaN,
      "learning_rate": 3.842901720437962e-05,
      "loss": 0.0,
      "step": 3457
    },
    {
      "epoch": 0.5816163484988647,
      "grad_norm": NaN,
      "learning_rate": 3.8402874214555296e-05,
      "loss": 0.0,
      "step": 3458
    },
    {
      "epoch": 0.581784542931629,
      "grad_norm": NaN,
      "learning_rate": 3.837673457501072e-05,
      "loss": 0.0,
      "step": 3459
    },
    {
      "epoch": 0.5819527373643932,
      "grad_norm": NaN,
      "learning_rate": 3.835059829329735e-05,
      "loss": 0.0,
      "step": 3460
    },
    {
      "epoch": 0.5821209317971575,
      "grad_norm": NaN,
      "learning_rate": 3.832446537696564e-05,
      "loss": 0.0,
      "step": 3461
    },
    {
      "epoch": 0.5822891262299218,
      "grad_norm": NaN,
      "learning_rate": 3.82983358335651e-05,
      "loss": 0.0,
      "step": 3462
    },
    {
      "epoch": 0.5824573206626861,
      "grad_norm": NaN,
      "learning_rate": 3.827220967064428e-05,
      "loss": 0.0,
      "step": 3463
    },
    {
      "epoch": 0.5826255150954504,
      "grad_norm": NaN,
      "learning_rate": 3.824608689575073e-05,
      "loss": 0.0,
      "step": 3464
    },
    {
      "epoch": 0.5827937095282146,
      "grad_norm": NaN,
      "learning_rate": 3.8219967516431034e-05,
      "loss": 0.0,
      "step": 3465
    },
    {
      "epoch": 0.5829619039609789,
      "grad_norm": NaN,
      "learning_rate": 3.819385154023075e-05,
      "loss": 0.0,
      "step": 3466
    },
    {
      "epoch": 0.5831300983937432,
      "grad_norm": NaN,
      "learning_rate": 3.816773897469454e-05,
      "loss": 0.0,
      "step": 3467
    },
    {
      "epoch": 0.5832982928265075,
      "grad_norm": NaN,
      "learning_rate": 3.8141629827366015e-05,
      "loss": 0.0,
      "step": 3468
    },
    {
      "epoch": 0.5834664872592717,
      "grad_norm": NaN,
      "learning_rate": 3.811552410578781e-05,
      "loss": 0.0,
      "step": 3469
    },
    {
      "epoch": 0.583634681692036,
      "grad_norm": NaN,
      "learning_rate": 3.808942181750156e-05,
      "loss": 0.0,
      "step": 3470
    },
    {
      "epoch": 0.5838028761248003,
      "grad_norm": NaN,
      "learning_rate": 3.806332297004797e-05,
      "loss": 0.0,
      "step": 3471
    },
    {
      "epoch": 0.5839710705575646,
      "grad_norm": NaN,
      "learning_rate": 3.803722757096666e-05,
      "loss": 0.0,
      "step": 3472
    },
    {
      "epoch": 0.5841392649903289,
      "grad_norm": NaN,
      "learning_rate": 3.801113562779631e-05,
      "loss": 0.0,
      "step": 3473
    },
    {
      "epoch": 0.5843074594230931,
      "grad_norm": NaN,
      "learning_rate": 3.7985047148074585e-05,
      "loss": 0.0,
      "step": 3474
    },
    {
      "epoch": 0.5844756538558574,
      "grad_norm": NaN,
      "learning_rate": 3.7958962139338176e-05,
      "loss": 0.0,
      "step": 3475
    },
    {
      "epoch": 0.5846438482886216,
      "grad_norm": NaN,
      "learning_rate": 3.793288060912275e-05,
      "loss": 0.0,
      "step": 3476
    },
    {
      "epoch": 0.5848120427213859,
      "grad_norm": NaN,
      "learning_rate": 3.790680256496294e-05,
      "loss": 0.0,
      "step": 3477
    },
    {
      "epoch": 0.5849802371541502,
      "grad_norm": NaN,
      "learning_rate": 3.788072801439242e-05,
      "loss": 0.0,
      "step": 3478
    },
    {
      "epoch": 0.5851484315869144,
      "grad_norm": NaN,
      "learning_rate": 3.785465696494384e-05,
      "loss": 0.0,
      "step": 3479
    },
    {
      "epoch": 0.5853166260196787,
      "grad_norm": NaN,
      "learning_rate": 3.7828589424148816e-05,
      "loss": 0.0,
      "step": 3480
    },
    {
      "epoch": 0.585484820452443,
      "grad_norm": NaN,
      "learning_rate": 3.7802525399538e-05,
      "loss": 0.0,
      "step": 3481
    },
    {
      "epoch": 0.5856530148852073,
      "grad_norm": NaN,
      "learning_rate": 3.777646489864094e-05,
      "loss": 0.0,
      "step": 3482
    },
    {
      "epoch": 0.5858212093179715,
      "grad_norm": NaN,
      "learning_rate": 3.775040792898629e-05,
      "loss": 0.0,
      "step": 3483
    },
    {
      "epoch": 0.5859894037507358,
      "grad_norm": NaN,
      "learning_rate": 3.7724354498101575e-05,
      "loss": 0.0,
      "step": 3484
    },
    {
      "epoch": 0.5861575981835001,
      "grad_norm": NaN,
      "learning_rate": 3.769830461351336e-05,
      "loss": 0.0,
      "step": 3485
    },
    {
      "epoch": 0.5863257926162644,
      "grad_norm": NaN,
      "learning_rate": 3.767225828274714e-05,
      "loss": 0.0,
      "step": 3486
    },
    {
      "epoch": 0.5864939870490287,
      "grad_norm": NaN,
      "learning_rate": 3.764621551332744e-05,
      "loss": 0.0,
      "step": 3487
    },
    {
      "epoch": 0.5866621814817929,
      "grad_norm": NaN,
      "learning_rate": 3.76201763127777e-05,
      "loss": 0.0,
      "step": 3488
    },
    {
      "epoch": 0.5868303759145572,
      "grad_norm": NaN,
      "learning_rate": 3.759414068862036e-05,
      "loss": 0.0,
      "step": 3489
    },
    {
      "epoch": 0.5869985703473215,
      "grad_norm": NaN,
      "learning_rate": 3.756810864837681e-05,
      "loss": 0.0,
      "step": 3490
    },
    {
      "epoch": 0.5871667647800858,
      "grad_norm": NaN,
      "learning_rate": 3.754208019956744e-05,
      "loss": 0.0,
      "step": 3491
    },
    {
      "epoch": 0.5873349592128501,
      "grad_norm": NaN,
      "learning_rate": 3.751605534971157e-05,
      "loss": 0.0,
      "step": 3492
    },
    {
      "epoch": 0.5875031536456143,
      "grad_norm": NaN,
      "learning_rate": 3.749003410632748e-05,
      "loss": 0.0,
      "step": 3493
    },
    {
      "epoch": 0.5876713480783786,
      "grad_norm": NaN,
      "learning_rate": 3.7464016476932394e-05,
      "loss": 0.0,
      "step": 3494
    },
    {
      "epoch": 0.5878395425111429,
      "grad_norm": NaN,
      "learning_rate": 3.7438002469042565e-05,
      "loss": 0.0,
      "step": 3495
    },
    {
      "epoch": 0.5880077369439072,
      "grad_norm": NaN,
      "learning_rate": 3.7411992090173107e-05,
      "loss": 0.0,
      "step": 3496
    },
    {
      "epoch": 0.5881759313766715,
      "grad_norm": NaN,
      "learning_rate": 3.738598534783815e-05,
      "loss": 0.0,
      "step": 3497
    },
    {
      "epoch": 0.5883441258094357,
      "grad_norm": NaN,
      "learning_rate": 3.735998224955073e-05,
      "loss": 0.0,
      "step": 3498
    },
    {
      "epoch": 0.5885123202422,
      "grad_norm": NaN,
      "learning_rate": 3.733398280282285e-05,
      "loss": 0.0,
      "step": 3499
    },
    {
      "epoch": 0.5886805146749643,
      "grad_norm": NaN,
      "learning_rate": 3.730798701516549e-05,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 0.5888487091077286,
      "grad_norm": NaN,
      "learning_rate": 3.7281994894088524e-05,
      "loss": 0.0,
      "step": 3501
    },
    {
      "epoch": 0.5890169035404929,
      "grad_norm": NaN,
      "learning_rate": 3.725600644710078e-05,
      "loss": 0.0,
      "step": 3502
    },
    {
      "epoch": 0.5891850979732571,
      "grad_norm": NaN,
      "learning_rate": 3.723002168171002e-05,
      "loss": 0.0,
      "step": 3503
    },
    {
      "epoch": 0.5893532924060214,
      "grad_norm": NaN,
      "learning_rate": 3.720404060542297e-05,
      "loss": 0.0,
      "step": 3504
    },
    {
      "epoch": 0.5895214868387857,
      "grad_norm": NaN,
      "learning_rate": 3.717806322574527e-05,
      "loss": 0.0,
      "step": 3505
    },
    {
      "epoch": 0.5896896812715499,
      "grad_norm": NaN,
      "learning_rate": 3.71520895501815e-05,
      "loss": 0.0,
      "step": 3506
    },
    {
      "epoch": 0.5898578757043141,
      "grad_norm": NaN,
      "learning_rate": 3.712611958623514e-05,
      "loss": 0.0,
      "step": 3507
    },
    {
      "epoch": 0.5900260701370784,
      "grad_norm": NaN,
      "learning_rate": 3.7100153341408636e-05,
      "loss": 0.0,
      "step": 3508
    },
    {
      "epoch": 0.5901942645698427,
      "grad_norm": NaN,
      "learning_rate": 3.707419082320336e-05,
      "loss": 0.0,
      "step": 3509
    },
    {
      "epoch": 0.590362459002607,
      "grad_norm": NaN,
      "learning_rate": 3.704823203911957e-05,
      "loss": 0.0,
      "step": 3510
    },
    {
      "epoch": 0.5905306534353713,
      "grad_norm": NaN,
      "learning_rate": 3.702227699665646e-05,
      "loss": 0.0,
      "step": 3511
    },
    {
      "epoch": 0.5906988478681355,
      "grad_norm": NaN,
      "learning_rate": 3.6996325703312176e-05,
      "loss": 0.0,
      "step": 3512
    },
    {
      "epoch": 0.5908670423008998,
      "grad_norm": NaN,
      "learning_rate": 3.6970378166583755e-05,
      "loss": 0.0,
      "step": 3513
    },
    {
      "epoch": 0.5910352367336641,
      "grad_norm": NaN,
      "learning_rate": 3.6944434393967144e-05,
      "loss": 0.0,
      "step": 3514
    },
    {
      "epoch": 0.5912034311664284,
      "grad_norm": NaN,
      "learning_rate": 3.691849439295719e-05,
      "loss": 0.0,
      "step": 3515
    },
    {
      "epoch": 0.5913716255991927,
      "grad_norm": NaN,
      "learning_rate": 3.68925581710477e-05,
      "loss": 0.0,
      "step": 3516
    },
    {
      "epoch": 0.5915398200319569,
      "grad_norm": NaN,
      "learning_rate": 3.686662573573134e-05,
      "loss": 0.0,
      "step": 3517
    },
    {
      "epoch": 0.5917080144647212,
      "grad_norm": NaN,
      "learning_rate": 3.6840697094499705e-05,
      "loss": 0.0,
      "step": 3518
    },
    {
      "epoch": 0.5918762088974855,
      "grad_norm": NaN,
      "learning_rate": 3.681477225484326e-05,
      "loss": 0.0,
      "step": 3519
    },
    {
      "epoch": 0.5920444033302498,
      "grad_norm": NaN,
      "learning_rate": 3.678885122425145e-05,
      "loss": 0.0,
      "step": 3520
    },
    {
      "epoch": 0.592212597763014,
      "grad_norm": NaN,
      "learning_rate": 3.676293401021256e-05,
      "loss": 0.0,
      "step": 3521
    },
    {
      "epoch": 0.5923807921957783,
      "grad_norm": NaN,
      "learning_rate": 3.673702062021376e-05,
      "loss": 0.0,
      "step": 3522
    },
    {
      "epoch": 0.5925489866285426,
      "grad_norm": NaN,
      "learning_rate": 3.671111106174113e-05,
      "loss": 0.0,
      "step": 3523
    },
    {
      "epoch": 0.5927171810613069,
      "grad_norm": NaN,
      "learning_rate": 3.668520534227969e-05,
      "loss": 0.0,
      "step": 3524
    },
    {
      "epoch": 0.5928853754940712,
      "grad_norm": NaN,
      "learning_rate": 3.66593034693133e-05,
      "loss": 0.0,
      "step": 3525
    },
    {
      "epoch": 0.5930535699268354,
      "grad_norm": NaN,
      "learning_rate": 3.66334054503247e-05,
      "loss": 0.0,
      "step": 3526
    },
    {
      "epoch": 0.5932217643595997,
      "grad_norm": NaN,
      "learning_rate": 3.660751129279554e-05,
      "loss": 0.0,
      "step": 3527
    },
    {
      "epoch": 0.593389958792364,
      "grad_norm": NaN,
      "learning_rate": 3.658162100420638e-05,
      "loss": 0.0,
      "step": 3528
    },
    {
      "epoch": 0.5935581532251283,
      "grad_norm": NaN,
      "learning_rate": 3.655573459203662e-05,
      "loss": 0.0,
      "step": 3529
    },
    {
      "epoch": 0.5937263476578926,
      "grad_norm": NaN,
      "learning_rate": 3.6529852063764545e-05,
      "loss": 0.0,
      "step": 3530
    },
    {
      "epoch": 0.5938945420906568,
      "grad_norm": NaN,
      "learning_rate": 3.6503973426867317e-05,
      "loss": 0.0,
      "step": 3531
    },
    {
      "epoch": 0.5940627365234211,
      "grad_norm": NaN,
      "learning_rate": 3.647809868882102e-05,
      "loss": 0.0,
      "step": 3532
    },
    {
      "epoch": 0.5942309309561854,
      "grad_norm": NaN,
      "learning_rate": 3.645222785710056e-05,
      "loss": 0.0,
      "step": 3533
    },
    {
      "epoch": 0.5943991253889497,
      "grad_norm": NaN,
      "learning_rate": 3.6426360939179735e-05,
      "loss": 0.0,
      "step": 3534
    },
    {
      "epoch": 0.5945673198217138,
      "grad_norm": NaN,
      "learning_rate": 3.6400497942531186e-05,
      "loss": 0.0,
      "step": 3535
    },
    {
      "epoch": 0.5947355142544781,
      "grad_norm": NaN,
      "learning_rate": 3.637463887462647e-05,
      "loss": 0.0,
      "step": 3536
    },
    {
      "epoch": 0.5949037086872424,
      "grad_norm": NaN,
      "learning_rate": 3.6348783742935966e-05,
      "loss": 0.0,
      "step": 3537
    },
    {
      "epoch": 0.5950719031200067,
      "grad_norm": NaN,
      "learning_rate": 3.632293255492894e-05,
      "loss": 0.0,
      "step": 3538
    },
    {
      "epoch": 0.595240097552771,
      "grad_norm": NaN,
      "learning_rate": 3.629708531807348e-05,
      "loss": 0.0,
      "step": 3539
    },
    {
      "epoch": 0.5954082919855352,
      "grad_norm": NaN,
      "learning_rate": 3.627124203983662e-05,
      "loss": 0.0,
      "step": 3540
    },
    {
      "epoch": 0.5955764864182995,
      "grad_norm": NaN,
      "learning_rate": 3.624540272768416e-05,
      "loss": 0.0,
      "step": 3541
    },
    {
      "epoch": 0.5957446808510638,
      "grad_norm": NaN,
      "learning_rate": 3.621956738908079e-05,
      "loss": 0.0,
      "step": 3542
    },
    {
      "epoch": 0.5959128752838281,
      "grad_norm": NaN,
      "learning_rate": 3.6193736031490046e-05,
      "loss": 0.0,
      "step": 3543
    },
    {
      "epoch": 0.5960810697165924,
      "grad_norm": NaN,
      "learning_rate": 3.616790866237433e-05,
      "loss": 0.0,
      "step": 3544
    },
    {
      "epoch": 0.5962492641493566,
      "grad_norm": NaN,
      "learning_rate": 3.614208528919487e-05,
      "loss": 0.0,
      "step": 3545
    },
    {
      "epoch": 0.5964174585821209,
      "grad_norm": NaN,
      "learning_rate": 3.611626591941175e-05,
      "loss": 0.0,
      "step": 3546
    },
    {
      "epoch": 0.5965856530148852,
      "grad_norm": NaN,
      "learning_rate": 3.609045056048389e-05,
      "loss": 0.0,
      "step": 3547
    },
    {
      "epoch": 0.5967538474476495,
      "grad_norm": NaN,
      "learning_rate": 3.606463921986908e-05,
      "loss": 0.0,
      "step": 3548
    },
    {
      "epoch": 0.5969220418804138,
      "grad_norm": NaN,
      "learning_rate": 3.6038831905023916e-05,
      "loss": 0.0,
      "step": 3549
    },
    {
      "epoch": 0.597090236313178,
      "grad_norm": NaN,
      "learning_rate": 3.601302862340384e-05,
      "loss": 0.0,
      "step": 3550
    },
    {
      "epoch": 0.5972584307459423,
      "grad_norm": NaN,
      "learning_rate": 3.598722938246314e-05,
      "loss": 0.0,
      "step": 3551
    },
    {
      "epoch": 0.5974266251787066,
      "grad_norm": NaN,
      "learning_rate": 3.5961434189654896e-05,
      "loss": 0.0,
      "step": 3552
    },
    {
      "epoch": 0.5975948196114709,
      "grad_norm": NaN,
      "learning_rate": 3.593564305243109e-05,
      "loss": 0.0,
      "step": 3553
    },
    {
      "epoch": 0.5977630140442352,
      "grad_norm": NaN,
      "learning_rate": 3.590985597824248e-05,
      "loss": 0.0,
      "step": 3554
    },
    {
      "epoch": 0.5979312084769994,
      "grad_norm": NaN,
      "learning_rate": 3.5884072974538654e-05,
      "loss": 0.0,
      "step": 3555
    },
    {
      "epoch": 0.5980994029097637,
      "grad_norm": NaN,
      "learning_rate": 3.5858294048768016e-05,
      "loss": 0.0,
      "step": 3556
    },
    {
      "epoch": 0.598267597342528,
      "grad_norm": NaN,
      "learning_rate": 3.5832519208377855e-05,
      "loss": 0.0,
      "step": 3557
    },
    {
      "epoch": 0.5984357917752923,
      "grad_norm": NaN,
      "learning_rate": 3.580674846081421e-05,
      "loss": 0.0,
      "step": 3558
    },
    {
      "epoch": 0.5986039862080565,
      "grad_norm": NaN,
      "learning_rate": 3.578098181352196e-05,
      "loss": 0.0,
      "step": 3559
    },
    {
      "epoch": 0.5987721806408208,
      "grad_norm": NaN,
      "learning_rate": 3.57552192739448e-05,
      "loss": 0.0,
      "step": 3560
    },
    {
      "epoch": 0.5989403750735851,
      "grad_norm": NaN,
      "learning_rate": 3.572946084952524e-05,
      "loss": 0.0,
      "step": 3561
    },
    {
      "epoch": 0.5991085695063494,
      "grad_norm": NaN,
      "learning_rate": 3.57037065477046e-05,
      "loss": 0.0,
      "step": 3562
    },
    {
      "epoch": 0.5992767639391137,
      "grad_norm": NaN,
      "learning_rate": 3.567795637592301e-05,
      "loss": 0.0,
      "step": 3563
    },
    {
      "epoch": 0.599444958371878,
      "grad_norm": NaN,
      "learning_rate": 3.56522103416194e-05,
      "loss": 0.0,
      "step": 3564
    },
    {
      "epoch": 0.5996131528046421,
      "grad_norm": NaN,
      "learning_rate": 3.562646845223153e-05,
      "loss": 0.0,
      "step": 3565
    },
    {
      "epoch": 0.5997813472374064,
      "grad_norm": NaN,
      "learning_rate": 3.560073071519593e-05,
      "loss": 0.0,
      "step": 3566
    },
    {
      "epoch": 0.5999495416701707,
      "grad_norm": NaN,
      "learning_rate": 3.557499713794795e-05,
      "loss": 0.0,
      "step": 3567
    },
    {
      "epoch": 0.600117736102935,
      "grad_norm": NaN,
      "learning_rate": 3.55492677279217e-05,
      "loss": 0.0,
      "step": 3568
    },
    {
      "epoch": 0.6002859305356992,
      "grad_norm": NaN,
      "learning_rate": 3.5523542492550175e-05,
      "loss": 0.0,
      "step": 3569
    },
    {
      "epoch": 0.6004541249684635,
      "grad_norm": NaN,
      "learning_rate": 3.5497821439265074e-05,
      "loss": 0.0,
      "step": 3570
    },
    {
      "epoch": 0.6006223194012278,
      "grad_norm": NaN,
      "learning_rate": 3.5472104575496925e-05,
      "loss": 0.0,
      "step": 3571
    },
    {
      "epoch": 0.6007905138339921,
      "grad_norm": NaN,
      "learning_rate": 3.5446391908675033e-05,
      "loss": 0.0,
      "step": 3572
    },
    {
      "epoch": 0.6009587082667563,
      "grad_norm": NaN,
      "learning_rate": 3.542068344622752e-05,
      "loss": 0.0,
      "step": 3573
    },
    {
      "epoch": 0.6011269026995206,
      "grad_norm": NaN,
      "learning_rate": 3.5394979195581254e-05,
      "loss": 0.0,
      "step": 3574
    },
    {
      "epoch": 0.6012950971322849,
      "grad_norm": NaN,
      "learning_rate": 3.536927916416193e-05,
      "loss": 0.0,
      "step": 3575
    },
    {
      "epoch": 0.6014632915650492,
      "grad_norm": NaN,
      "learning_rate": 3.534358335939394e-05,
      "loss": 0.0,
      "step": 3576
    },
    {
      "epoch": 0.6016314859978135,
      "grad_norm": NaN,
      "learning_rate": 3.531789178870059e-05,
      "loss": 0.0,
      "step": 3577
    },
    {
      "epoch": 0.6017996804305777,
      "grad_norm": NaN,
      "learning_rate": 3.529220445950385e-05,
      "loss": 0.0,
      "step": 3578
    },
    {
      "epoch": 0.601967874863342,
      "grad_norm": NaN,
      "learning_rate": 3.526652137922451e-05,
      "loss": 0.0,
      "step": 3579
    },
    {
      "epoch": 0.6021360692961063,
      "grad_norm": NaN,
      "learning_rate": 3.52408425552821e-05,
      "loss": 0.0,
      "step": 3580
    },
    {
      "epoch": 0.6023042637288706,
      "grad_norm": NaN,
      "learning_rate": 3.5215167995094975e-05,
      "loss": 0.0,
      "step": 3581
    },
    {
      "epoch": 0.6024724581616349,
      "grad_norm": NaN,
      "learning_rate": 3.518949770608022e-05,
      "loss": 0.0,
      "step": 3582
    },
    {
      "epoch": 0.6026406525943991,
      "grad_norm": NaN,
      "learning_rate": 3.5163831695653694e-05,
      "loss": 0.0,
      "step": 3583
    },
    {
      "epoch": 0.6028088470271634,
      "grad_norm": NaN,
      "learning_rate": 3.513816997123e-05,
      "loss": 0.0,
      "step": 3584
    },
    {
      "epoch": 0.6029770414599277,
      "grad_norm": NaN,
      "learning_rate": 3.511251254022255e-05,
      "loss": 0.0,
      "step": 3585
    },
    {
      "epoch": 0.603145235892692,
      "grad_norm": NaN,
      "learning_rate": 3.508685941004348e-05,
      "loss": 0.0,
      "step": 3586
    },
    {
      "epoch": 0.6033134303254563,
      "grad_norm": NaN,
      "learning_rate": 3.5061210588103686e-05,
      "loss": 0.0,
      "step": 3587
    },
    {
      "epoch": 0.6034816247582205,
      "grad_norm": NaN,
      "learning_rate": 3.503556608181282e-05,
      "loss": 0.0,
      "step": 3588
    },
    {
      "epoch": 0.6036498191909848,
      "grad_norm": NaN,
      "learning_rate": 3.5009925898579314e-05,
      "loss": 0.0,
      "step": 3589
    },
    {
      "epoch": 0.6038180136237491,
      "grad_norm": NaN,
      "learning_rate": 3.498429004581032e-05,
      "loss": 0.0,
      "step": 3590
    },
    {
      "epoch": 0.6039862080565134,
      "grad_norm": NaN,
      "learning_rate": 3.495865853091175e-05,
      "loss": 0.0,
      "step": 3591
    },
    {
      "epoch": 0.6041544024892777,
      "grad_norm": NaN,
      "learning_rate": 3.4933031361288226e-05,
      "loss": 0.0,
      "step": 3592
    },
    {
      "epoch": 0.6043225969220419,
      "grad_norm": NaN,
      "learning_rate": 3.490740854434321e-05,
      "loss": 0.0,
      "step": 3593
    },
    {
      "epoch": 0.6044907913548061,
      "grad_norm": NaN,
      "learning_rate": 3.488179008747881e-05,
      "loss": 0.0,
      "step": 3594
    },
    {
      "epoch": 0.6046589857875704,
      "grad_norm": NaN,
      "learning_rate": 3.485617599809593e-05,
      "loss": 0.0,
      "step": 3595
    },
    {
      "epoch": 0.6048271802203347,
      "grad_norm": NaN,
      "learning_rate": 3.483056628359416e-05,
      "loss": 0.0,
      "step": 3596
    },
    {
      "epoch": 0.6049953746530989,
      "grad_norm": NaN,
      "learning_rate": 3.48049609513719e-05,
      "loss": 0.0,
      "step": 3597
    },
    {
      "epoch": 0.6051635690858632,
      "grad_norm": NaN,
      "learning_rate": 3.477936000882623e-05,
      "loss": 0.0,
      "step": 3598
    },
    {
      "epoch": 0.6053317635186275,
      "grad_norm": NaN,
      "learning_rate": 3.475376346335298e-05,
      "loss": 0.0,
      "step": 3599
    },
    {
      "epoch": 0.6054999579513918,
      "grad_norm": NaN,
      "learning_rate": 3.4728171322346694e-05,
      "loss": 0.0,
      "step": 3600
    },
    {
      "epoch": 0.605668152384156,
      "grad_norm": NaN,
      "learning_rate": 3.470258359320064e-05,
      "loss": 0.0,
      "step": 3601
    },
    {
      "epoch": 0.6058363468169203,
      "grad_norm": NaN,
      "learning_rate": 3.467700028330686e-05,
      "loss": 0.0,
      "step": 3602
    },
    {
      "epoch": 0.6060045412496846,
      "grad_norm": NaN,
      "learning_rate": 3.465142140005607e-05,
      "loss": 0.0,
      "step": 3603
    },
    {
      "epoch": 0.6061727356824489,
      "grad_norm": NaN,
      "learning_rate": 3.4625846950837725e-05,
      "loss": 0.0,
      "step": 3604
    },
    {
      "epoch": 0.6063409301152132,
      "grad_norm": NaN,
      "learning_rate": 3.460027694303998e-05,
      "loss": 0.0,
      "step": 3605
    },
    {
      "epoch": 0.6065091245479775,
      "grad_norm": NaN,
      "learning_rate": 3.457471138404975e-05,
      "loss": 0.0,
      "step": 3606
    },
    {
      "epoch": 0.6066773189807417,
      "grad_norm": NaN,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.0,
      "step": 3607
    },
    {
      "epoch": 0.606845513413506,
      "grad_norm": NaN,
      "learning_rate": 3.452359364203294e-05,
      "loss": 0.0,
      "step": 3608
    },
    {
      "epoch": 0.6070137078462703,
      "grad_norm": NaN,
      "learning_rate": 3.449804147377369e-05,
      "loss": 0.0,
      "step": 3609
    },
    {
      "epoch": 0.6071819022790346,
      "grad_norm": NaN,
      "learning_rate": 3.447249378385663e-05,
      "loss": 0.0,
      "step": 3610
    },
    {
      "epoch": 0.6073500967117988,
      "grad_norm": NaN,
      "learning_rate": 3.444695057966221e-05,
      "loss": 0.0,
      "step": 3611
    },
    {
      "epoch": 0.6075182911445631,
      "grad_norm": NaN,
      "learning_rate": 3.442141186856955e-05,
      "loss": 0.0,
      "step": 3612
    },
    {
      "epoch": 0.6076864855773274,
      "grad_norm": NaN,
      "learning_rate": 3.43958776579565e-05,
      "loss": 0.0,
      "step": 3613
    },
    {
      "epoch": 0.6078546800100917,
      "grad_norm": NaN,
      "learning_rate": 3.437034795519963e-05,
      "loss": 0.0,
      "step": 3614
    },
    {
      "epoch": 0.608022874442856,
      "grad_norm": NaN,
      "learning_rate": 3.434482276767418e-05,
      "loss": 0.0,
      "step": 3615
    },
    {
      "epoch": 0.6081910688756202,
      "grad_norm": NaN,
      "learning_rate": 3.431930210275409e-05,
      "loss": 0.0,
      "step": 3616
    },
    {
      "epoch": 0.6083592633083845,
      "grad_norm": NaN,
      "learning_rate": 3.4293785967811964e-05,
      "loss": 0.0,
      "step": 3617
    },
    {
      "epoch": 0.6085274577411488,
      "grad_norm": NaN,
      "learning_rate": 3.426827437021917e-05,
      "loss": 0.0,
      "step": 3618
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": NaN,
      "learning_rate": 3.424276731734571e-05,
      "loss": 0.0,
      "step": 3619
    },
    {
      "epoch": 0.6088638466066774,
      "grad_norm": NaN,
      "learning_rate": 3.4217264816560276e-05,
      "loss": 0.0,
      "step": 3620
    },
    {
      "epoch": 0.6090320410394416,
      "grad_norm": NaN,
      "learning_rate": 3.4191766875230234e-05,
      "loss": 0.0,
      "step": 3621
    },
    {
      "epoch": 0.6092002354722059,
      "grad_norm": NaN,
      "learning_rate": 3.416627350072171e-05,
      "loss": 0.0,
      "step": 3622
    },
    {
      "epoch": 0.6093684299049702,
      "grad_norm": NaN,
      "learning_rate": 3.414078470039942e-05,
      "loss": 0.0,
      "step": 3623
    },
    {
      "epoch": 0.6095366243377344,
      "grad_norm": NaN,
      "learning_rate": 3.411530048162679e-05,
      "loss": 0.0,
      "step": 3624
    },
    {
      "epoch": 0.6097048187704986,
      "grad_norm": NaN,
      "learning_rate": 3.408982085176592e-05,
      "loss": 0.0,
      "step": 3625
    },
    {
      "epoch": 0.6098730132032629,
      "grad_norm": NaN,
      "learning_rate": 3.406434581817762e-05,
      "loss": 0.0,
      "step": 3626
    },
    {
      "epoch": 0.6100412076360272,
      "grad_norm": NaN,
      "learning_rate": 3.403887538822132e-05,
      "loss": 0.0,
      "step": 3627
    },
    {
      "epoch": 0.6102094020687915,
      "grad_norm": NaN,
      "learning_rate": 3.401340956925515e-05,
      "loss": 0.0,
      "step": 3628
    },
    {
      "epoch": 0.6103775965015558,
      "grad_norm": NaN,
      "learning_rate": 3.398794836863588e-05,
      "loss": 0.0,
      "step": 3629
    },
    {
      "epoch": 0.61054579093432,
      "grad_norm": NaN,
      "learning_rate": 3.3962491793719e-05,
      "loss": 0.0,
      "step": 3630
    },
    {
      "epoch": 0.6107139853670843,
      "grad_norm": NaN,
      "learning_rate": 3.39370398518586e-05,
      "loss": 0.0,
      "step": 3631
    },
    {
      "epoch": 0.6108821797998486,
      "grad_norm": NaN,
      "learning_rate": 3.391159255040748e-05,
      "loss": 0.0,
      "step": 3632
    },
    {
      "epoch": 0.6110503742326129,
      "grad_norm": NaN,
      "learning_rate": 3.388614989671705e-05,
      "loss": 0.0,
      "step": 3633
    },
    {
      "epoch": 0.6112185686653772,
      "grad_norm": NaN,
      "learning_rate": 3.386071189813744e-05,
      "loss": 0.0,
      "step": 3634
    },
    {
      "epoch": 0.6113867630981414,
      "grad_norm": NaN,
      "learning_rate": 3.38352785620174e-05,
      "loss": 0.0,
      "step": 3635
    },
    {
      "epoch": 0.6115549575309057,
      "grad_norm": NaN,
      "learning_rate": 3.380984989570433e-05,
      "loss": 0.0,
      "step": 3636
    },
    {
      "epoch": 0.61172315196367,
      "grad_norm": NaN,
      "learning_rate": 3.378442590654427e-05,
      "loss": 0.0,
      "step": 3637
    },
    {
      "epoch": 0.6118913463964343,
      "grad_norm": NaN,
      "learning_rate": 3.375900660188196e-05,
      "loss": 0.0,
      "step": 3638
    },
    {
      "epoch": 0.6120595408291986,
      "grad_norm": NaN,
      "learning_rate": 3.373359198906072e-05,
      "loss": 0.0,
      "step": 3639
    },
    {
      "epoch": 0.6122277352619628,
      "grad_norm": NaN,
      "learning_rate": 3.3708182075422566e-05,
      "loss": 0.0,
      "step": 3640
    },
    {
      "epoch": 0.6123959296947271,
      "grad_norm": NaN,
      "learning_rate": 3.368277686830812e-05,
      "loss": 0.0,
      "step": 3641
    },
    {
      "epoch": 0.6125641241274914,
      "grad_norm": NaN,
      "learning_rate": 3.3657376375056684e-05,
      "loss": 0.0,
      "step": 3642
    },
    {
      "epoch": 0.6127323185602557,
      "grad_norm": NaN,
      "learning_rate": 3.363198060300616e-05,
      "loss": 0.0,
      "step": 3643
    },
    {
      "epoch": 0.61290051299302,
      "grad_norm": NaN,
      "learning_rate": 3.360658955949312e-05,
      "loss": 0.0,
      "step": 3644
    },
    {
      "epoch": 0.6130687074257842,
      "grad_norm": NaN,
      "learning_rate": 3.358120325185272e-05,
      "loss": 0.0,
      "step": 3645
    },
    {
      "epoch": 0.6132369018585485,
      "grad_norm": NaN,
      "learning_rate": 3.355582168741882e-05,
      "loss": 0.0,
      "step": 3646
    },
    {
      "epoch": 0.6134050962913128,
      "grad_norm": NaN,
      "learning_rate": 3.353044487352384e-05,
      "loss": 0.0,
      "step": 3647
    },
    {
      "epoch": 0.6135732907240771,
      "grad_norm": NaN,
      "learning_rate": 3.3505072817498876e-05,
      "loss": 0.0,
      "step": 3648
    },
    {
      "epoch": 0.6137414851568413,
      "grad_norm": NaN,
      "learning_rate": 3.347970552667361e-05,
      "loss": 0.0,
      "step": 3649
    },
    {
      "epoch": 0.6139096795896056,
      "grad_norm": NaN,
      "learning_rate": 3.345434300837636e-05,
      "loss": 0.0,
      "step": 3650
    },
    {
      "epoch": 0.6140778740223699,
      "grad_norm": NaN,
      "learning_rate": 3.342898526993412e-05,
      "loss": 0.0,
      "step": 3651
    },
    {
      "epoch": 0.6142460684551342,
      "grad_norm": NaN,
      "learning_rate": 3.340363231867243e-05,
      "loss": 0.0,
      "step": 3652
    },
    {
      "epoch": 0.6144142628878985,
      "grad_norm": NaN,
      "learning_rate": 3.3378284161915465e-05,
      "loss": 0.0,
      "step": 3653
    },
    {
      "epoch": 0.6145824573206626,
      "grad_norm": NaN,
      "learning_rate": 3.335294080698603e-05,
      "loss": 0.0,
      "step": 3654
    },
    {
      "epoch": 0.6147506517534269,
      "grad_norm": NaN,
      "learning_rate": 3.332760226120554e-05,
      "loss": 0.0,
      "step": 3655
    },
    {
      "epoch": 0.6149188461861912,
      "grad_norm": NaN,
      "learning_rate": 3.330226853189402e-05,
      "loss": 0.0,
      "step": 3656
    },
    {
      "epoch": 0.6150870406189555,
      "grad_norm": NaN,
      "learning_rate": 3.32769396263701e-05,
      "loss": 0.0,
      "step": 3657
    },
    {
      "epoch": 0.6152552350517198,
      "grad_norm": NaN,
      "learning_rate": 3.325161555195099e-05,
      "loss": 0.0,
      "step": 3658
    },
    {
      "epoch": 0.615423429484484,
      "grad_norm": NaN,
      "learning_rate": 3.3226296315952566e-05,
      "loss": 0.0,
      "step": 3659
    },
    {
      "epoch": 0.6155916239172483,
      "grad_norm": NaN,
      "learning_rate": 3.3200981925689264e-05,
      "loss": 0.0,
      "step": 3660
    },
    {
      "epoch": 0.6157598183500126,
      "grad_norm": NaN,
      "learning_rate": 3.3175672388474136e-05,
      "loss": 0.0,
      "step": 3661
    },
    {
      "epoch": 0.6159280127827769,
      "grad_norm": NaN,
      "learning_rate": 3.315036771161878e-05,
      "loss": 0.0,
      "step": 3662
    },
    {
      "epoch": 0.6160962072155411,
      "grad_norm": NaN,
      "learning_rate": 3.3125067902433485e-05,
      "loss": 0.0,
      "step": 3663
    },
    {
      "epoch": 0.6162644016483054,
      "grad_norm": NaN,
      "learning_rate": 3.3099772968227066e-05,
      "loss": 0.0,
      "step": 3664
    },
    {
      "epoch": 0.6164325960810697,
      "grad_norm": NaN,
      "learning_rate": 3.307448291630695e-05,
      "loss": 0.0,
      "step": 3665
    },
    {
      "epoch": 0.616600790513834,
      "grad_norm": NaN,
      "learning_rate": 3.304919775397912e-05,
      "loss": 0.0,
      "step": 3666
    },
    {
      "epoch": 0.6167689849465983,
      "grad_norm": NaN,
      "learning_rate": 3.3023917488548215e-05,
      "loss": 0.0,
      "step": 3667
    },
    {
      "epoch": 0.6169371793793625,
      "grad_norm": NaN,
      "learning_rate": 3.29986421273174e-05,
      "loss": 0.0,
      "step": 3668
    },
    {
      "epoch": 0.6171053738121268,
      "grad_norm": NaN,
      "learning_rate": 3.297337167758846e-05,
      "loss": 0.0,
      "step": 3669
    },
    {
      "epoch": 0.6172735682448911,
      "grad_norm": NaN,
      "learning_rate": 3.29481061466617e-05,
      "loss": 0.0,
      "step": 3670
    },
    {
      "epoch": 0.6174417626776554,
      "grad_norm": NaN,
      "learning_rate": 3.292284554183611e-05,
      "loss": 0.0,
      "step": 3671
    },
    {
      "epoch": 0.6176099571104197,
      "grad_norm": NaN,
      "learning_rate": 3.2897589870409164e-05,
      "loss": 0.0,
      "step": 3672
    },
    {
      "epoch": 0.6177781515431839,
      "grad_norm": NaN,
      "learning_rate": 3.287233913967695e-05,
      "loss": 0.0,
      "step": 3673
    },
    {
      "epoch": 0.6179463459759482,
      "grad_norm": NaN,
      "learning_rate": 3.284709335693411e-05,
      "loss": 0.0,
      "step": 3674
    },
    {
      "epoch": 0.6181145404087125,
      "grad_norm": NaN,
      "learning_rate": 3.282185252947387e-05,
      "loss": 0.0,
      "step": 3675
    },
    {
      "epoch": 0.6182827348414768,
      "grad_norm": NaN,
      "learning_rate": 3.279661666458804e-05,
      "loss": 0.0,
      "step": 3676
    },
    {
      "epoch": 0.6184509292742411,
      "grad_norm": NaN,
      "learning_rate": 3.2771385769566975e-05,
      "loss": 0.0,
      "step": 3677
    },
    {
      "epoch": 0.6186191237070053,
      "grad_norm": NaN,
      "learning_rate": 3.274615985169955e-05,
      "loss": 0.0,
      "step": 3678
    },
    {
      "epoch": 0.6187873181397696,
      "grad_norm": NaN,
      "learning_rate": 3.272093891827332e-05,
      "loss": 0.0,
      "step": 3679
    },
    {
      "epoch": 0.6189555125725339,
      "grad_norm": NaN,
      "learning_rate": 3.26957229765743e-05,
      "loss": 0.0,
      "step": 3680
    },
    {
      "epoch": 0.6191237070052982,
      "grad_norm": NaN,
      "learning_rate": 3.26705120338871e-05,
      "loss": 0.0,
      "step": 3681
    },
    {
      "epoch": 0.6192919014380625,
      "grad_norm": NaN,
      "learning_rate": 3.2645306097494855e-05,
      "loss": 0.0,
      "step": 3682
    },
    {
      "epoch": 0.6194600958708266,
      "grad_norm": NaN,
      "learning_rate": 3.2620105174679305e-05,
      "loss": 0.0,
      "step": 3683
    },
    {
      "epoch": 0.6196282903035909,
      "grad_norm": NaN,
      "learning_rate": 3.259490927272071e-05,
      "loss": 0.0,
      "step": 3684
    },
    {
      "epoch": 0.6197964847363552,
      "grad_norm": NaN,
      "learning_rate": 3.256971839889787e-05,
      "loss": 0.0,
      "step": 3685
    },
    {
      "epoch": 0.6199646791691195,
      "grad_norm": NaN,
      "learning_rate": 3.254453256048813e-05,
      "loss": 0.0,
      "step": 3686
    },
    {
      "epoch": 0.6201328736018837,
      "grad_norm": NaN,
      "learning_rate": 3.251935176476745e-05,
      "loss": 0.0,
      "step": 3687
    },
    {
      "epoch": 0.620301068034648,
      "grad_norm": NaN,
      "learning_rate": 3.2494176019010244e-05,
      "loss": 0.0,
      "step": 3688
    },
    {
      "epoch": 0.6204692624674123,
      "grad_norm": NaN,
      "learning_rate": 3.246900533048952e-05,
      "loss": 0.0,
      "step": 3689
    },
    {
      "epoch": 0.6206374569001766,
      "grad_norm": NaN,
      "learning_rate": 3.244383970647677e-05,
      "loss": 0.0,
      "step": 3690
    },
    {
      "epoch": 0.6208056513329409,
      "grad_norm": NaN,
      "learning_rate": 3.241867915424211e-05,
      "loss": 0.0,
      "step": 3691
    },
    {
      "epoch": 0.6209738457657051,
      "grad_norm": NaN,
      "learning_rate": 3.2393523681054114e-05,
      "loss": 0.0,
      "step": 3692
    },
    {
      "epoch": 0.6211420401984694,
      "grad_norm": NaN,
      "learning_rate": 3.236837329417993e-05,
      "loss": 0.0,
      "step": 3693
    },
    {
      "epoch": 0.6213102346312337,
      "grad_norm": NaN,
      "learning_rate": 3.2343228000885206e-05,
      "loss": 0.0,
      "step": 3694
    },
    {
      "epoch": 0.621478429063998,
      "grad_norm": NaN,
      "learning_rate": 3.231808780843416e-05,
      "loss": 0.0,
      "step": 3695
    },
    {
      "epoch": 0.6216466234967623,
      "grad_norm": NaN,
      "learning_rate": 3.229295272408949e-05,
      "loss": 0.0,
      "step": 3696
    },
    {
      "epoch": 0.6218148179295265,
      "grad_norm": NaN,
      "learning_rate": 3.226782275511245e-05,
      "loss": 0.0,
      "step": 3697
    },
    {
      "epoch": 0.6219830123622908,
      "grad_norm": NaN,
      "learning_rate": 3.22426979087628e-05,
      "loss": 0.0,
      "step": 3698
    },
    {
      "epoch": 0.6221512067950551,
      "grad_norm": NaN,
      "learning_rate": 3.2217578192298815e-05,
      "loss": 0.0,
      "step": 3699
    },
    {
      "epoch": 0.6223194012278194,
      "grad_norm": NaN,
      "learning_rate": 3.219246361297733e-05,
      "loss": 0.0,
      "step": 3700
    },
    {
      "epoch": 0.6224875956605836,
      "grad_norm": NaN,
      "learning_rate": 3.216735417805366e-05,
      "loss": 0.0,
      "step": 3701
    },
    {
      "epoch": 0.6226557900933479,
      "grad_norm": NaN,
      "learning_rate": 3.214224989478163e-05,
      "loss": 0.0,
      "step": 3702
    },
    {
      "epoch": 0.6228239845261122,
      "grad_norm": NaN,
      "learning_rate": 3.2117150770413574e-05,
      "loss": 0.0,
      "step": 3703
    },
    {
      "epoch": 0.6229921789588765,
      "grad_norm": NaN,
      "learning_rate": 3.209205681220037e-05,
      "loss": 0.0,
      "step": 3704
    },
    {
      "epoch": 0.6231603733916408,
      "grad_norm": NaN,
      "learning_rate": 3.2066968027391374e-05,
      "loss": 0.0,
      "step": 3705
    },
    {
      "epoch": 0.623328567824405,
      "grad_norm": NaN,
      "learning_rate": 3.204188442323445e-05,
      "loss": 0.0,
      "step": 3706
    },
    {
      "epoch": 0.6234967622571693,
      "grad_norm": NaN,
      "learning_rate": 3.201680600697596e-05,
      "loss": 0.0,
      "step": 3707
    },
    {
      "epoch": 0.6236649566899336,
      "grad_norm": NaN,
      "learning_rate": 3.1991732785860805e-05,
      "loss": 0.0,
      "step": 3708
    },
    {
      "epoch": 0.6238331511226979,
      "grad_norm": NaN,
      "learning_rate": 3.196666476713235e-05,
      "loss": 0.0,
      "step": 3709
    },
    {
      "epoch": 0.6240013455554622,
      "grad_norm": NaN,
      "learning_rate": 3.194160195803245e-05,
      "loss": 0.0,
      "step": 3710
    },
    {
      "epoch": 0.6241695399882264,
      "grad_norm": NaN,
      "learning_rate": 3.1916544365801484e-05,
      "loss": 0.0,
      "step": 3711
    },
    {
      "epoch": 0.6243377344209907,
      "grad_norm": NaN,
      "learning_rate": 3.18914919976783e-05,
      "loss": 0.0,
      "step": 3712
    },
    {
      "epoch": 0.6245059288537549,
      "grad_norm": NaN,
      "learning_rate": 3.1866444860900256e-05,
      "loss": 0.0,
      "step": 3713
    },
    {
      "epoch": 0.6246741232865192,
      "grad_norm": NaN,
      "learning_rate": 3.184140296270318e-05,
      "loss": 0.0,
      "step": 3714
    },
    {
      "epoch": 0.6248423177192834,
      "grad_norm": NaN,
      "learning_rate": 3.1816366310321385e-05,
      "loss": 0.0,
      "step": 3715
    },
    {
      "epoch": 0.6250105121520477,
      "grad_norm": NaN,
      "learning_rate": 3.179133491098772e-05,
      "loss": 0.0,
      "step": 3716
    },
    {
      "epoch": 0.625178706584812,
      "grad_norm": NaN,
      "learning_rate": 3.176630877193344e-05,
      "loss": 0.0,
      "step": 3717
    },
    {
      "epoch": 0.6253469010175763,
      "grad_norm": NaN,
      "learning_rate": 3.174128790038833e-05,
      "loss": 0.0,
      "step": 3718
    },
    {
      "epoch": 0.6255150954503406,
      "grad_norm": NaN,
      "learning_rate": 3.171627230358063e-05,
      "loss": 0.0,
      "step": 3719
    },
    {
      "epoch": 0.6256832898831048,
      "grad_norm": NaN,
      "learning_rate": 3.169126198873708e-05,
      "loss": 0.0,
      "step": 3720
    },
    {
      "epoch": 0.6258514843158691,
      "grad_norm": NaN,
      "learning_rate": 3.1666256963082876e-05,
      "loss": 0.0,
      "step": 3721
    },
    {
      "epoch": 0.6260196787486334,
      "grad_norm": NaN,
      "learning_rate": 3.164125723384168e-05,
      "loss": 0.0,
      "step": 3722
    },
    {
      "epoch": 0.6261878731813977,
      "grad_norm": NaN,
      "learning_rate": 3.161626280823562e-05,
      "loss": 0.0,
      "step": 3723
    },
    {
      "epoch": 0.626356067614162,
      "grad_norm": NaN,
      "learning_rate": 3.1591273693485345e-05,
      "loss": 0.0,
      "step": 3724
    },
    {
      "epoch": 0.6265242620469262,
      "grad_norm": NaN,
      "learning_rate": 3.156628989680991e-05,
      "loss": 0.0,
      "step": 3725
    },
    {
      "epoch": 0.6266924564796905,
      "grad_norm": NaN,
      "learning_rate": 3.154131142542686e-05,
      "loss": 0.0,
      "step": 3726
    },
    {
      "epoch": 0.6268606509124548,
      "grad_norm": NaN,
      "learning_rate": 3.151633828655216e-05,
      "loss": 0.0,
      "step": 3727
    },
    {
      "epoch": 0.6270288453452191,
      "grad_norm": NaN,
      "learning_rate": 3.149137048740032e-05,
      "loss": 0.0,
      "step": 3728
    },
    {
      "epoch": 0.6271970397779834,
      "grad_norm": NaN,
      "learning_rate": 3.146640803518425e-05,
      "loss": 0.0,
      "step": 3729
    },
    {
      "epoch": 0.6273652342107476,
      "grad_norm": NaN,
      "learning_rate": 3.1441450937115305e-05,
      "loss": 0.0,
      "step": 3730
    },
    {
      "epoch": 0.6275334286435119,
      "grad_norm": NaN,
      "learning_rate": 3.14164992004033e-05,
      "loss": 0.0,
      "step": 3731
    },
    {
      "epoch": 0.6277016230762762,
      "grad_norm": NaN,
      "learning_rate": 3.139155283225654e-05,
      "loss": 0.0,
      "step": 3732
    },
    {
      "epoch": 0.6278698175090405,
      "grad_norm": NaN,
      "learning_rate": 3.136661183988175e-05,
      "loss": 0.0,
      "step": 3733
    },
    {
      "epoch": 0.6280380119418048,
      "grad_norm": NaN,
      "learning_rate": 3.134167623048409e-05,
      "loss": 0.0,
      "step": 3734
    },
    {
      "epoch": 0.628206206374569,
      "grad_norm": NaN,
      "learning_rate": 3.1316746011267155e-05,
      "loss": 0.0,
      "step": 3735
    },
    {
      "epoch": 0.6283744008073333,
      "grad_norm": NaN,
      "learning_rate": 3.129182118943306e-05,
      "loss": 0.0,
      "step": 3736
    },
    {
      "epoch": 0.6285425952400976,
      "grad_norm": NaN,
      "learning_rate": 3.126690177218228e-05,
      "loss": 0.0,
      "step": 3737
    },
    {
      "epoch": 0.6287107896728619,
      "grad_norm": NaN,
      "learning_rate": 3.1241987766713754e-05,
      "loss": 0.0,
      "step": 3738
    },
    {
      "epoch": 0.6288789841056261,
      "grad_norm": NaN,
      "learning_rate": 3.1217079180224846e-05,
      "loss": 0.0,
      "step": 3739
    },
    {
      "epoch": 0.6290471785383904,
      "grad_norm": NaN,
      "learning_rate": 3.119217601991139e-05,
      "loss": 0.0,
      "step": 3740
    },
    {
      "epoch": 0.6292153729711547,
      "grad_norm": NaN,
      "learning_rate": 3.1167278292967624e-05,
      "loss": 0.0,
      "step": 3741
    },
    {
      "epoch": 0.6293835674039189,
      "grad_norm": NaN,
      "learning_rate": 3.1142386006586234e-05,
      "loss": 0.0,
      "step": 3742
    },
    {
      "epoch": 0.6295517618366832,
      "grad_norm": NaN,
      "learning_rate": 3.111749916795828e-05,
      "loss": 0.0,
      "step": 3743
    },
    {
      "epoch": 0.6297199562694474,
      "grad_norm": NaN,
      "learning_rate": 3.1092617784273336e-05,
      "loss": 0.0,
      "step": 3744
    },
    {
      "epoch": 0.6298881507022117,
      "grad_norm": NaN,
      "learning_rate": 3.1067741862719346e-05,
      "loss": 0.0,
      "step": 3745
    },
    {
      "epoch": 0.630056345134976,
      "grad_norm": NaN,
      "learning_rate": 3.104287141048267e-05,
      "loss": 0.0,
      "step": 3746
    },
    {
      "epoch": 0.6302245395677403,
      "grad_norm": NaN,
      "learning_rate": 3.1018006434748113e-05,
      "loss": 0.0,
      "step": 3747
    },
    {
      "epoch": 0.6303927340005046,
      "grad_norm": NaN,
      "learning_rate": 3.099314694269889e-05,
      "loss": 0.0,
      "step": 3748
    },
    {
      "epoch": 0.6305609284332688,
      "grad_norm": NaN,
      "learning_rate": 3.096829294151662e-05,
      "loss": 0.0,
      "step": 3749
    },
    {
      "epoch": 0.6307291228660331,
      "grad_norm": NaN,
      "learning_rate": 3.0943444438381366e-05,
      "loss": 0.0,
      "step": 3750
    },
    {
      "epoch": 0.6308973172987974,
      "grad_norm": NaN,
      "learning_rate": 3.091860144047155e-05,
      "loss": 0.0,
      "step": 3751
    },
    {
      "epoch": 0.6310655117315617,
      "grad_norm": NaN,
      "learning_rate": 3.089376395496405e-05,
      "loss": 0.0,
      "step": 3752
    },
    {
      "epoch": 0.631233706164326,
      "grad_norm": NaN,
      "learning_rate": 3.0868931989034155e-05,
      "loss": 0.0,
      "step": 3753
    },
    {
      "epoch": 0.6314019005970902,
      "grad_norm": NaN,
      "learning_rate": 3.084410554985553e-05,
      "loss": 0.0,
      "step": 3754
    },
    {
      "epoch": 0.6315700950298545,
      "grad_norm": NaN,
      "learning_rate": 3.0819284644600246e-05,
      "loss": 0.0,
      "step": 3755
    },
    {
      "epoch": 0.6317382894626188,
      "grad_norm": NaN,
      "learning_rate": 3.079446928043878e-05,
      "loss": 0.0,
      "step": 3756
    },
    {
      "epoch": 0.6319064838953831,
      "grad_norm": NaN,
      "learning_rate": 3.076965946454005e-05,
      "loss": 0.0,
      "step": 3757
    },
    {
      "epoch": 0.6320746783281473,
      "grad_norm": NaN,
      "learning_rate": 3.07448552040713e-05,
      "loss": 0.0,
      "step": 3758
    },
    {
      "epoch": 0.6322428727609116,
      "grad_norm": NaN,
      "learning_rate": 3.072005650619821e-05,
      "loss": 0.0,
      "step": 3759
    },
    {
      "epoch": 0.6324110671936759,
      "grad_norm": NaN,
      "learning_rate": 3.0695263378084834e-05,
      "loss": 0.0,
      "step": 3760
    },
    {
      "epoch": 0.6325792616264402,
      "grad_norm": NaN,
      "learning_rate": 3.0670475826893664e-05,
      "loss": 0.0,
      "step": 3761
    },
    {
      "epoch": 0.6327474560592045,
      "grad_norm": NaN,
      "learning_rate": 3.064569385978552e-05,
      "loss": 0.0,
      "step": 3762
    },
    {
      "epoch": 0.6329156504919687,
      "grad_norm": NaN,
      "learning_rate": 3.0620917483919637e-05,
      "loss": 0.0,
      "step": 3763
    },
    {
      "epoch": 0.633083844924733,
      "grad_norm": NaN,
      "learning_rate": 3.0596146706453616e-05,
      "loss": 0.0,
      "step": 3764
    },
    {
      "epoch": 0.6332520393574973,
      "grad_norm": NaN,
      "learning_rate": 3.05713815345435e-05,
      "loss": 0.0,
      "step": 3765
    },
    {
      "epoch": 0.6334202337902616,
      "grad_norm": NaN,
      "learning_rate": 3.054662197534363e-05,
      "loss": 0.0,
      "step": 3766
    },
    {
      "epoch": 0.6335884282230259,
      "grad_norm": NaN,
      "learning_rate": 3.0521868036006804e-05,
      "loss": 0.0,
      "step": 3767
    },
    {
      "epoch": 0.6337566226557901,
      "grad_norm": NaN,
      "learning_rate": 3.0497119723684108e-05,
      "loss": 0.0,
      "step": 3768
    },
    {
      "epoch": 0.6339248170885544,
      "grad_norm": NaN,
      "learning_rate": 3.0472377045525097e-05,
      "loss": 0.0,
      "step": 3769
    },
    {
      "epoch": 0.6340930115213187,
      "grad_norm": NaN,
      "learning_rate": 3.0447640008677635e-05,
      "loss": 0.0,
      "step": 3770
    },
    {
      "epoch": 0.634261205954083,
      "grad_norm": NaN,
      "learning_rate": 3.0422908620287972e-05,
      "loss": 0.0,
      "step": 3771
    },
    {
      "epoch": 0.6344294003868471,
      "grad_norm": NaN,
      "learning_rate": 3.0398182887500716e-05,
      "loss": 0.0,
      "step": 3772
    },
    {
      "epoch": 0.6345975948196114,
      "grad_norm": NaN,
      "learning_rate": 3.0373462817458886e-05,
      "loss": 0.0,
      "step": 3773
    },
    {
      "epoch": 0.6347657892523757,
      "grad_norm": NaN,
      "learning_rate": 3.0348748417303823e-05,
      "loss": 0.0,
      "step": 3774
    },
    {
      "epoch": 0.63493398368514,
      "grad_norm": NaN,
      "learning_rate": 3.0324039694175233e-05,
      "loss": 0.0,
      "step": 3775
    },
    {
      "epoch": 0.6351021781179043,
      "grad_norm": NaN,
      "learning_rate": 3.0299336655211184e-05,
      "loss": 0.0,
      "step": 3776
    },
    {
      "epoch": 0.6352703725506685,
      "grad_norm": NaN,
      "learning_rate": 3.0274639307548125e-05,
      "loss": 0.0,
      "step": 3777
    },
    {
      "epoch": 0.6354385669834328,
      "grad_norm": NaN,
      "learning_rate": 3.0249947658320832e-05,
      "loss": 0.0,
      "step": 3778
    },
    {
      "epoch": 0.6356067614161971,
      "grad_norm": NaN,
      "learning_rate": 3.022526171466245e-05,
      "loss": 0.0,
      "step": 3779
    },
    {
      "epoch": 0.6357749558489614,
      "grad_norm": NaN,
      "learning_rate": 3.0200581483704448e-05,
      "loss": 0.0,
      "step": 3780
    },
    {
      "epoch": 0.6359431502817257,
      "grad_norm": NaN,
      "learning_rate": 3.0175906972576708e-05,
      "loss": 0.0,
      "step": 3781
    },
    {
      "epoch": 0.6361113447144899,
      "grad_norm": NaN,
      "learning_rate": 3.01512381884074e-05,
      "loss": 0.0,
      "step": 3782
    },
    {
      "epoch": 0.6362795391472542,
      "grad_norm": NaN,
      "learning_rate": 3.0126575138323064e-05,
      "loss": 0.0,
      "step": 3783
    },
    {
      "epoch": 0.6364477335800185,
      "grad_norm": NaN,
      "learning_rate": 3.010191782944855e-05,
      "loss": 0.0,
      "step": 3784
    },
    {
      "epoch": 0.6366159280127828,
      "grad_norm": NaN,
      "learning_rate": 3.0077266268907124e-05,
      "loss": 0.0,
      "step": 3785
    },
    {
      "epoch": 0.636784122445547,
      "grad_norm": NaN,
      "learning_rate": 3.005262046382033e-05,
      "loss": 0.0,
      "step": 3786
    },
    {
      "epoch": 0.6369523168783113,
      "grad_norm": NaN,
      "learning_rate": 3.0027980421308062e-05,
      "loss": 0.0,
      "step": 3787
    },
    {
      "epoch": 0.6371205113110756,
      "grad_norm": NaN,
      "learning_rate": 3.000334614848854e-05,
      "loss": 0.0,
      "step": 3788
    },
    {
      "epoch": 0.6372887057438399,
      "grad_norm": NaN,
      "learning_rate": 2.9978717652478344e-05,
      "loss": 0.0,
      "step": 3789
    },
    {
      "epoch": 0.6374569001766042,
      "grad_norm": NaN,
      "learning_rate": 2.9954094940392375e-05,
      "loss": 0.0,
      "step": 3790
    },
    {
      "epoch": 0.6376250946093684,
      "grad_norm": NaN,
      "learning_rate": 2.9929478019343842e-05,
      "loss": 0.0,
      "step": 3791
    },
    {
      "epoch": 0.6377932890421327,
      "grad_norm": NaN,
      "learning_rate": 2.9904866896444295e-05,
      "loss": 0.0,
      "step": 3792
    },
    {
      "epoch": 0.637961483474897,
      "grad_norm": NaN,
      "learning_rate": 2.9880261578803636e-05,
      "loss": 0.0,
      "step": 3793
    },
    {
      "epoch": 0.6381296779076613,
      "grad_norm": NaN,
      "learning_rate": 2.9855662073530055e-05,
      "loss": 0.0,
      "step": 3794
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": NaN,
      "learning_rate": 2.9831068387730076e-05,
      "loss": 0.0,
      "step": 3795
    },
    {
      "epoch": 0.6384660667731898,
      "grad_norm": NaN,
      "learning_rate": 2.9806480528508517e-05,
      "loss": 0.0,
      "step": 3796
    },
    {
      "epoch": 0.6386342612059541,
      "grad_norm": NaN,
      "learning_rate": 2.978189850296857e-05,
      "loss": 0.0,
      "step": 3797
    },
    {
      "epoch": 0.6388024556387184,
      "grad_norm": NaN,
      "learning_rate": 2.975732231821169e-05,
      "loss": 0.0,
      "step": 3798
    },
    {
      "epoch": 0.6389706500714827,
      "grad_norm": NaN,
      "learning_rate": 2.9732751981337658e-05,
      "loss": 0.0,
      "step": 3799
    },
    {
      "epoch": 0.639138844504247,
      "grad_norm": NaN,
      "learning_rate": 2.9708187499444574e-05,
      "loss": 0.0,
      "step": 3800
    },
    {
      "epoch": 0.6393070389370111,
      "grad_norm": NaN,
      "learning_rate": 2.9683628879628816e-05,
      "loss": 0.0,
      "step": 3801
    },
    {
      "epoch": 0.6394752333697754,
      "grad_norm": NaN,
      "learning_rate": 2.965907612898514e-05,
      "loss": 0.0,
      "step": 3802
    },
    {
      "epoch": 0.6396434278025397,
      "grad_norm": NaN,
      "learning_rate": 2.9634529254606542e-05,
      "loss": 0.0,
      "step": 3803
    },
    {
      "epoch": 0.639811622235304,
      "grad_norm": NaN,
      "learning_rate": 2.960998826358433e-05,
      "loss": 0.0,
      "step": 3804
    },
    {
      "epoch": 0.6399798166680682,
      "grad_norm": NaN,
      "learning_rate": 2.9585453163008124e-05,
      "loss": 0.0,
      "step": 3805
    },
    {
      "epoch": 0.6401480111008325,
      "grad_norm": NaN,
      "learning_rate": 2.9560923959965854e-05,
      "loss": 0.0,
      "step": 3806
    },
    {
      "epoch": 0.6403162055335968,
      "grad_norm": NaN,
      "learning_rate": 2.9536400661543717e-05,
      "loss": 0.0,
      "step": 3807
    },
    {
      "epoch": 0.6404843999663611,
      "grad_norm": NaN,
      "learning_rate": 2.9511883274826224e-05,
      "loss": 0.0,
      "step": 3808
    },
    {
      "epoch": 0.6406525943991254,
      "grad_norm": NaN,
      "learning_rate": 2.948737180689615e-05,
      "loss": 0.0,
      "step": 3809
    },
    {
      "epoch": 0.6408207888318896,
      "grad_norm": NaN,
      "learning_rate": 2.946286626483463e-05,
      "loss": 0.0,
      "step": 3810
    },
    {
      "epoch": 0.6409889832646539,
      "grad_norm": NaN,
      "learning_rate": 2.9438366655721018e-05,
      "loss": 0.0,
      "step": 3811
    },
    {
      "epoch": 0.6411571776974182,
      "grad_norm": NaN,
      "learning_rate": 2.9413872986632972e-05,
      "loss": 0.0,
      "step": 3812
    },
    {
      "epoch": 0.6413253721301825,
      "grad_norm": NaN,
      "learning_rate": 2.938938526464644e-05,
      "loss": 0.0,
      "step": 3813
    },
    {
      "epoch": 0.6414935665629468,
      "grad_norm": NaN,
      "learning_rate": 2.936490349683566e-05,
      "loss": 0.0,
      "step": 3814
    },
    {
      "epoch": 0.641661760995711,
      "grad_norm": NaN,
      "learning_rate": 2.934042769027313e-05,
      "loss": 0.0,
      "step": 3815
    },
    {
      "epoch": 0.6418299554284753,
      "grad_norm": NaN,
      "learning_rate": 2.931595785202964e-05,
      "loss": 0.0,
      "step": 3816
    },
    {
      "epoch": 0.6419981498612396,
      "grad_norm": NaN,
      "learning_rate": 2.9291493989174234e-05,
      "loss": 0.0,
      "step": 3817
    },
    {
      "epoch": 0.6421663442940039,
      "grad_norm": NaN,
      "learning_rate": 2.9267036108774282e-05,
      "loss": 0.0,
      "step": 3818
    },
    {
      "epoch": 0.6423345387267682,
      "grad_norm": NaN,
      "learning_rate": 2.9242584217895373e-05,
      "loss": 0.0,
      "step": 3819
    },
    {
      "epoch": 0.6425027331595324,
      "grad_norm": NaN,
      "learning_rate": 2.9218138323601397e-05,
      "loss": 0.0,
      "step": 3820
    },
    {
      "epoch": 0.6426709275922967,
      "grad_norm": NaN,
      "learning_rate": 2.9193698432954463e-05,
      "loss": 0.0,
      "step": 3821
    },
    {
      "epoch": 0.642839122025061,
      "grad_norm": NaN,
      "learning_rate": 2.916926455301504e-05,
      "loss": 0.0,
      "step": 3822
    },
    {
      "epoch": 0.6430073164578253,
      "grad_norm": NaN,
      "learning_rate": 2.914483669084176e-05,
      "loss": 0.0,
      "step": 3823
    },
    {
      "epoch": 0.6431755108905896,
      "grad_norm": NaN,
      "learning_rate": 2.912041485349157e-05,
      "loss": 0.0,
      "step": 3824
    },
    {
      "epoch": 0.6433437053233538,
      "grad_norm": NaN,
      "learning_rate": 2.909599904801964e-05,
      "loss": 0.0,
      "step": 3825
    },
    {
      "epoch": 0.6435118997561181,
      "grad_norm": NaN,
      "learning_rate": 2.907158928147947e-05,
      "loss": 0.0,
      "step": 3826
    },
    {
      "epoch": 0.6436800941888824,
      "grad_norm": NaN,
      "learning_rate": 2.9047185560922745e-05,
      "loss": 0.0,
      "step": 3827
    },
    {
      "epoch": 0.6438482886216467,
      "grad_norm": NaN,
      "learning_rate": 2.902278789339943e-05,
      "loss": 0.0,
      "step": 3828
    },
    {
      "epoch": 0.644016483054411,
      "grad_norm": NaN,
      "learning_rate": 2.899839628595771e-05,
      "loss": 0.0,
      "step": 3829
    },
    {
      "epoch": 0.6441846774871752,
      "grad_norm": NaN,
      "learning_rate": 2.8974010745644087e-05,
      "loss": 0.0,
      "step": 3830
    },
    {
      "epoch": 0.6443528719199394,
      "grad_norm": NaN,
      "learning_rate": 2.8949631279503264e-05,
      "loss": 0.0,
      "step": 3831
    },
    {
      "epoch": 0.6445210663527037,
      "grad_norm": NaN,
      "learning_rate": 2.892525789457818e-05,
      "loss": 0.0,
      "step": 3832
    },
    {
      "epoch": 0.644689260785468,
      "grad_norm": NaN,
      "learning_rate": 2.8900890597910023e-05,
      "loss": 0.0,
      "step": 3833
    },
    {
      "epoch": 0.6448574552182322,
      "grad_norm": NaN,
      "learning_rate": 2.8876529396538264e-05,
      "loss": 0.0,
      "step": 3834
    },
    {
      "epoch": 0.6450256496509965,
      "grad_norm": NaN,
      "learning_rate": 2.8852174297500567e-05,
      "loss": 0.0,
      "step": 3835
    },
    {
      "epoch": 0.6451938440837608,
      "grad_norm": NaN,
      "learning_rate": 2.882782530783285e-05,
      "loss": 0.0,
      "step": 3836
    },
    {
      "epoch": 0.6453620385165251,
      "grad_norm": NaN,
      "learning_rate": 2.880348243456926e-05,
      "loss": 0.0,
      "step": 3837
    },
    {
      "epoch": 0.6455302329492894,
      "grad_norm": NaN,
      "learning_rate": 2.877914568474218e-05,
      "loss": 0.0,
      "step": 3838
    },
    {
      "epoch": 0.6456984273820536,
      "grad_norm": NaN,
      "learning_rate": 2.8754815065382225e-05,
      "loss": 0.0,
      "step": 3839
    },
    {
      "epoch": 0.6458666218148179,
      "grad_norm": NaN,
      "learning_rate": 2.8730490583518234e-05,
      "loss": 0.0,
      "step": 3840
    },
    {
      "epoch": 0.6460348162475822,
      "grad_norm": NaN,
      "learning_rate": 2.870617224617726e-05,
      "loss": 0.0,
      "step": 3841
    },
    {
      "epoch": 0.6462030106803465,
      "grad_norm": NaN,
      "learning_rate": 2.8681860060384648e-05,
      "loss": 0.0,
      "step": 3842
    },
    {
      "epoch": 0.6463712051131107,
      "grad_norm": NaN,
      "learning_rate": 2.865755403316388e-05,
      "loss": 0.0,
      "step": 3843
    },
    {
      "epoch": 0.646539399545875,
      "grad_norm": NaN,
      "learning_rate": 2.8633254171536704e-05,
      "loss": 0.0,
      "step": 3844
    },
    {
      "epoch": 0.6467075939786393,
      "grad_norm": NaN,
      "learning_rate": 2.8608960482523056e-05,
      "loss": 0.0,
      "step": 3845
    },
    {
      "epoch": 0.6468757884114036,
      "grad_norm": NaN,
      "learning_rate": 2.858467297314115e-05,
      "loss": 0.0,
      "step": 3846
    },
    {
      "epoch": 0.6470439828441679,
      "grad_norm": NaN,
      "learning_rate": 2.8560391650407358e-05,
      "loss": 0.0,
      "step": 3847
    },
    {
      "epoch": 0.6472121772769321,
      "grad_norm": NaN,
      "learning_rate": 2.8536116521336287e-05,
      "loss": 0.0,
      "step": 3848
    },
    {
      "epoch": 0.6473803717096964,
      "grad_norm": NaN,
      "learning_rate": 2.8511847592940738e-05,
      "loss": 0.0,
      "step": 3849
    },
    {
      "epoch": 0.6475485661424607,
      "grad_norm": NaN,
      "learning_rate": 2.848758487223172e-05,
      "loss": 0.0,
      "step": 3850
    },
    {
      "epoch": 0.647716760575225,
      "grad_norm": NaN,
      "learning_rate": 2.84633283662185e-05,
      "loss": 0.0,
      "step": 3851
    },
    {
      "epoch": 0.6478849550079893,
      "grad_norm": NaN,
      "learning_rate": 2.843907808190849e-05,
      "loss": 0.0,
      "step": 3852
    },
    {
      "epoch": 0.6480531494407535,
      "grad_norm": NaN,
      "learning_rate": 2.8414834026307323e-05,
      "loss": 0.0,
      "step": 3853
    },
    {
      "epoch": 0.6482213438735178,
      "grad_norm": NaN,
      "learning_rate": 2.8390596206418817e-05,
      "loss": 0.0,
      "step": 3854
    },
    {
      "epoch": 0.6483895383062821,
      "grad_norm": NaN,
      "learning_rate": 2.8366364629245057e-05,
      "loss": 0.0,
      "step": 3855
    },
    {
      "epoch": 0.6485577327390464,
      "grad_norm": NaN,
      "learning_rate": 2.8342139301786236e-05,
      "loss": 0.0,
      "step": 3856
    },
    {
      "epoch": 0.6487259271718107,
      "grad_norm": NaN,
      "learning_rate": 2.8317920231040796e-05,
      "loss": 0.0,
      "step": 3857
    },
    {
      "epoch": 0.6488941216045749,
      "grad_norm": NaN,
      "learning_rate": 2.829370742400534e-05,
      "loss": 0.0,
      "step": 3858
    },
    {
      "epoch": 0.6490623160373392,
      "grad_norm": NaN,
      "learning_rate": 2.8269500887674687e-05,
      "loss": 0.0,
      "step": 3859
    },
    {
      "epoch": 0.6492305104701035,
      "grad_norm": NaN,
      "learning_rate": 2.8245300629041838e-05,
      "loss": 0.0,
      "step": 3860
    },
    {
      "epoch": 0.6493987049028677,
      "grad_norm": NaN,
      "learning_rate": 2.822110665509796e-05,
      "loss": 0.0,
      "step": 3861
    },
    {
      "epoch": 0.6495668993356319,
      "grad_norm": NaN,
      "learning_rate": 2.819691897283242e-05,
      "loss": 0.0,
      "step": 3862
    },
    {
      "epoch": 0.6497350937683962,
      "grad_norm": NaN,
      "learning_rate": 2.81727375892328e-05,
      "loss": 0.0,
      "step": 3863
    },
    {
      "epoch": 0.6499032882011605,
      "grad_norm": NaN,
      "learning_rate": 2.814856251128481e-05,
      "loss": 0.0,
      "step": 3864
    },
    {
      "epoch": 0.6500714826339248,
      "grad_norm": NaN,
      "learning_rate": 2.8124393745972354e-05,
      "loss": 0.0,
      "step": 3865
    },
    {
      "epoch": 0.6502396770666891,
      "grad_norm": NaN,
      "learning_rate": 2.8100231300277514e-05,
      "loss": 0.0,
      "step": 3866
    },
    {
      "epoch": 0.6504078714994533,
      "grad_norm": NaN,
      "learning_rate": 2.8076075181180576e-05,
      "loss": 0.0,
      "step": 3867
    },
    {
      "epoch": 0.6505760659322176,
      "grad_norm": NaN,
      "learning_rate": 2.8051925395659955e-05,
      "loss": 0.0,
      "step": 3868
    },
    {
      "epoch": 0.6507442603649819,
      "grad_norm": NaN,
      "learning_rate": 2.8027781950692254e-05,
      "loss": 0.0,
      "step": 3869
    },
    {
      "epoch": 0.6509124547977462,
      "grad_norm": NaN,
      "learning_rate": 2.800364485325223e-05,
      "loss": 0.0,
      "step": 3870
    },
    {
      "epoch": 0.6510806492305105,
      "grad_norm": NaN,
      "learning_rate": 2.797951411031285e-05,
      "loss": 0.0,
      "step": 3871
    },
    {
      "epoch": 0.6512488436632747,
      "grad_norm": NaN,
      "learning_rate": 2.79553897288452e-05,
      "loss": 0.0,
      "step": 3872
    },
    {
      "epoch": 0.651417038096039,
      "grad_norm": NaN,
      "learning_rate": 2.793127171581854e-05,
      "loss": 0.0,
      "step": 3873
    },
    {
      "epoch": 0.6515852325288033,
      "grad_norm": NaN,
      "learning_rate": 2.7907160078200308e-05,
      "loss": 0.0,
      "step": 3874
    },
    {
      "epoch": 0.6517534269615676,
      "grad_norm": NaN,
      "learning_rate": 2.7883054822956068e-05,
      "loss": 0.0,
      "step": 3875
    },
    {
      "epoch": 0.6519216213943319,
      "grad_norm": NaN,
      "learning_rate": 2.7858955957049566e-05,
      "loss": 0.0,
      "step": 3876
    },
    {
      "epoch": 0.6520898158270961,
      "grad_norm": NaN,
      "learning_rate": 2.7834863487442693e-05,
      "loss": 0.0,
      "step": 3877
    },
    {
      "epoch": 0.6522580102598604,
      "grad_norm": NaN,
      "learning_rate": 2.781077742109548e-05,
      "loss": 0.0,
      "step": 3878
    },
    {
      "epoch": 0.6524262046926247,
      "grad_norm": NaN,
      "learning_rate": 2.7786697764966152e-05,
      "loss": 0.0,
      "step": 3879
    },
    {
      "epoch": 0.652594399125389,
      "grad_norm": NaN,
      "learning_rate": 2.7762624526011038e-05,
      "loss": 0.0,
      "step": 3880
    },
    {
      "epoch": 0.6527625935581532,
      "grad_norm": NaN,
      "learning_rate": 2.7738557711184622e-05,
      "loss": 0.0,
      "step": 3881
    },
    {
      "epoch": 0.6529307879909175,
      "grad_norm": NaN,
      "learning_rate": 2.7714497327439524e-05,
      "loss": 0.0,
      "step": 3882
    },
    {
      "epoch": 0.6530989824236818,
      "grad_norm": NaN,
      "learning_rate": 2.7690443381726554e-05,
      "loss": 0.0,
      "step": 3883
    },
    {
      "epoch": 0.6532671768564461,
      "grad_norm": NaN,
      "learning_rate": 2.766639588099461e-05,
      "loss": 0.0,
      "step": 3884
    },
    {
      "epoch": 0.6534353712892104,
      "grad_norm": NaN,
      "learning_rate": 2.7642354832190747e-05,
      "loss": 0.0,
      "step": 3885
    },
    {
      "epoch": 0.6536035657219746,
      "grad_norm": NaN,
      "learning_rate": 2.761832024226013e-05,
      "loss": 0.0,
      "step": 3886
    },
    {
      "epoch": 0.6537717601547389,
      "grad_norm": NaN,
      "learning_rate": 2.7594292118146137e-05,
      "loss": 0.0,
      "step": 3887
    },
    {
      "epoch": 0.6539399545875032,
      "grad_norm": NaN,
      "learning_rate": 2.757027046679018e-05,
      "loss": 0.0,
      "step": 3888
    },
    {
      "epoch": 0.6541081490202675,
      "grad_norm": NaN,
      "learning_rate": 2.754625529513186e-05,
      "loss": 0.0,
      "step": 3889
    },
    {
      "epoch": 0.6542763434530317,
      "grad_norm": NaN,
      "learning_rate": 2.752224661010887e-05,
      "loss": 0.0,
      "step": 3890
    },
    {
      "epoch": 0.6544445378857959,
      "grad_norm": NaN,
      "learning_rate": 2.7498244418657094e-05,
      "loss": 0.0,
      "step": 3891
    },
    {
      "epoch": 0.6546127323185602,
      "grad_norm": NaN,
      "learning_rate": 2.7474248727710467e-05,
      "loss": 0.0,
      "step": 3892
    },
    {
      "epoch": 0.6547809267513245,
      "grad_norm": NaN,
      "learning_rate": 2.745025954420108e-05,
      "loss": 0.0,
      "step": 3893
    },
    {
      "epoch": 0.6549491211840888,
      "grad_norm": NaN,
      "learning_rate": 2.7426276875059143e-05,
      "loss": 0.0,
      "step": 3894
    },
    {
      "epoch": 0.655117315616853,
      "grad_norm": NaN,
      "learning_rate": 2.740230072721297e-05,
      "loss": 0.0,
      "step": 3895
    },
    {
      "epoch": 0.6552855100496173,
      "grad_norm": NaN,
      "learning_rate": 2.7378331107589007e-05,
      "loss": 0.0,
      "step": 3896
    },
    {
      "epoch": 0.6554537044823816,
      "grad_norm": NaN,
      "learning_rate": 2.7354368023111822e-05,
      "loss": 0.0,
      "step": 3897
    },
    {
      "epoch": 0.6556218989151459,
      "grad_norm": NaN,
      "learning_rate": 2.7330411480704054e-05,
      "loss": 0.0,
      "step": 3898
    },
    {
      "epoch": 0.6557900933479102,
      "grad_norm": NaN,
      "learning_rate": 2.7306461487286484e-05,
      "loss": 0.0,
      "step": 3899
    },
    {
      "epoch": 0.6559582877806744,
      "grad_norm": NaN,
      "learning_rate": 2.7282518049778027e-05,
      "loss": 0.0,
      "step": 3900
    },
    {
      "epoch": 0.6561264822134387,
      "grad_norm": NaN,
      "learning_rate": 2.7258581175095654e-05,
      "loss": 0.0,
      "step": 3901
    },
    {
      "epoch": 0.656294676646203,
      "grad_norm": NaN,
      "learning_rate": 2.7234650870154465e-05,
      "loss": 0.0,
      "step": 3902
    },
    {
      "epoch": 0.6564628710789673,
      "grad_norm": NaN,
      "learning_rate": 2.721072714186763e-05,
      "loss": 0.0,
      "step": 3903
    },
    {
      "epoch": 0.6566310655117316,
      "grad_norm": NaN,
      "learning_rate": 2.718680999714649e-05,
      "loss": 0.0,
      "step": 3904
    },
    {
      "epoch": 0.6567992599444958,
      "grad_norm": NaN,
      "learning_rate": 2.7162899442900414e-05,
      "loss": 0.0,
      "step": 3905
    },
    {
      "epoch": 0.6569674543772601,
      "grad_norm": NaN,
      "learning_rate": 2.7138995486036906e-05,
      "loss": 0.0,
      "step": 3906
    },
    {
      "epoch": 0.6571356488100244,
      "grad_norm": NaN,
      "learning_rate": 2.711509813346152e-05,
      "loss": 0.0,
      "step": 3907
    },
    {
      "epoch": 0.6573038432427887,
      "grad_norm": NaN,
      "learning_rate": 2.7091207392077977e-05,
      "loss": 0.0,
      "step": 3908
    },
    {
      "epoch": 0.657472037675553,
      "grad_norm": NaN,
      "learning_rate": 2.7067323268788024e-05,
      "loss": 0.0,
      "step": 3909
    },
    {
      "epoch": 0.6576402321083172,
      "grad_norm": NaN,
      "learning_rate": 2.7043445770491526e-05,
      "loss": 0.0,
      "step": 3910
    },
    {
      "epoch": 0.6578084265410815,
      "grad_norm": NaN,
      "learning_rate": 2.7019574904086393e-05,
      "loss": 0.0,
      "step": 3911
    },
    {
      "epoch": 0.6579766209738458,
      "grad_norm": NaN,
      "learning_rate": 2.6995710676468695e-05,
      "loss": 0.0,
      "step": 3912
    },
    {
      "epoch": 0.6581448154066101,
      "grad_norm": NaN,
      "learning_rate": 2.697185309453252e-05,
      "loss": 0.0,
      "step": 3913
    },
    {
      "epoch": 0.6583130098393744,
      "grad_norm": NaN,
      "learning_rate": 2.694800216517005e-05,
      "loss": 0.0,
      "step": 3914
    },
    {
      "epoch": 0.6584812042721386,
      "grad_norm": NaN,
      "learning_rate": 2.6924157895271563e-05,
      "loss": 0.0,
      "step": 3915
    },
    {
      "epoch": 0.6586493987049029,
      "grad_norm": NaN,
      "learning_rate": 2.6900320291725396e-05,
      "loss": 0.0,
      "step": 3916
    },
    {
      "epoch": 0.6588175931376672,
      "grad_norm": NaN,
      "learning_rate": 2.687648936141796e-05,
      "loss": 0.0,
      "step": 3917
    },
    {
      "epoch": 0.6589857875704315,
      "grad_norm": NaN,
      "learning_rate": 2.6852665111233755e-05,
      "loss": 0.0,
      "step": 3918
    },
    {
      "epoch": 0.6591539820031957,
      "grad_norm": NaN,
      "learning_rate": 2.6828847548055313e-05,
      "loss": 0.0,
      "step": 3919
    },
    {
      "epoch": 0.6593221764359599,
      "grad_norm": NaN,
      "learning_rate": 2.6805036678763307e-05,
      "loss": 0.0,
      "step": 3920
    },
    {
      "epoch": 0.6594903708687242,
      "grad_norm": NaN,
      "learning_rate": 2.67812325102364e-05,
      "loss": 0.0,
      "step": 3921
    },
    {
      "epoch": 0.6596585653014885,
      "grad_norm": NaN,
      "learning_rate": 2.6757435049351353e-05,
      "loss": 0.0,
      "step": 3922
    },
    {
      "epoch": 0.6598267597342528,
      "grad_norm": NaN,
      "learning_rate": 2.6733644302982975e-05,
      "loss": 0.0,
      "step": 3923
    },
    {
      "epoch": 0.659994954167017,
      "grad_norm": NaN,
      "learning_rate": 2.6709860278004172e-05,
      "loss": 0.0,
      "step": 3924
    },
    {
      "epoch": 0.6601631485997813,
      "grad_norm": NaN,
      "learning_rate": 2.6686082981285876e-05,
      "loss": 0.0,
      "step": 3925
    },
    {
      "epoch": 0.6603313430325456,
      "grad_norm": NaN,
      "learning_rate": 2.6662312419697077e-05,
      "loss": 0.0,
      "step": 3926
    },
    {
      "epoch": 0.6604995374653099,
      "grad_norm": NaN,
      "learning_rate": 2.6638548600104805e-05,
      "loss": 0.0,
      "step": 3927
    },
    {
      "epoch": 0.6606677318980742,
      "grad_norm": NaN,
      "learning_rate": 2.6614791529374196e-05,
      "loss": 0.0,
      "step": 3928
    },
    {
      "epoch": 0.6608359263308384,
      "grad_norm": NaN,
      "learning_rate": 2.6591041214368385e-05,
      "loss": 0.0,
      "step": 3929
    },
    {
      "epoch": 0.6610041207636027,
      "grad_norm": NaN,
      "learning_rate": 2.6567297661948565e-05,
      "loss": 0.0,
      "step": 3930
    },
    {
      "epoch": 0.661172315196367,
      "grad_norm": NaN,
      "learning_rate": 2.6543560878973996e-05,
      "loss": 0.0,
      "step": 3931
    },
    {
      "epoch": 0.6613405096291313,
      "grad_norm": NaN,
      "learning_rate": 2.6519830872301965e-05,
      "loss": 0.0,
      "step": 3932
    },
    {
      "epoch": 0.6615087040618955,
      "grad_norm": NaN,
      "learning_rate": 2.6496107648787794e-05,
      "loss": 0.0,
      "step": 3933
    },
    {
      "epoch": 0.6616768984946598,
      "grad_norm": NaN,
      "learning_rate": 2.6472391215284868e-05,
      "loss": 0.0,
      "step": 3934
    },
    {
      "epoch": 0.6618450929274241,
      "grad_norm": NaN,
      "learning_rate": 2.644868157864458e-05,
      "loss": 0.0,
      "step": 3935
    },
    {
      "epoch": 0.6620132873601884,
      "grad_norm": NaN,
      "learning_rate": 2.642497874571641e-05,
      "loss": 0.0,
      "step": 3936
    },
    {
      "epoch": 0.6621814817929527,
      "grad_norm": NaN,
      "learning_rate": 2.6401282723347826e-05,
      "loss": 0.0,
      "step": 3937
    },
    {
      "epoch": 0.6623496762257169,
      "grad_norm": NaN,
      "learning_rate": 2.637759351838434e-05,
      "loss": 0.0,
      "step": 3938
    },
    {
      "epoch": 0.6625178706584812,
      "grad_norm": NaN,
      "learning_rate": 2.6353911137669483e-05,
      "loss": 0.0,
      "step": 3939
    },
    {
      "epoch": 0.6626860650912455,
      "grad_norm": NaN,
      "learning_rate": 2.6330235588044865e-05,
      "loss": 0.0,
      "step": 3940
    },
    {
      "epoch": 0.6628542595240098,
      "grad_norm": NaN,
      "learning_rate": 2.630656687635007e-05,
      "loss": 0.0,
      "step": 3941
    },
    {
      "epoch": 0.6630224539567741,
      "grad_norm": NaN,
      "learning_rate": 2.628290500942272e-05,
      "loss": 0.0,
      "step": 3942
    },
    {
      "epoch": 0.6631906483895383,
      "grad_norm": NaN,
      "learning_rate": 2.6259249994098455e-05,
      "loss": 0.0,
      "step": 3943
    },
    {
      "epoch": 0.6633588428223026,
      "grad_norm": NaN,
      "learning_rate": 2.6235601837210966e-05,
      "loss": 0.0,
      "step": 3944
    },
    {
      "epoch": 0.6635270372550669,
      "grad_norm": NaN,
      "learning_rate": 2.6211960545591936e-05,
      "loss": 0.0,
      "step": 3945
    },
    {
      "epoch": 0.6636952316878312,
      "grad_norm": NaN,
      "learning_rate": 2.618832612607106e-05,
      "loss": 0.0,
      "step": 3946
    },
    {
      "epoch": 0.6638634261205955,
      "grad_norm": NaN,
      "learning_rate": 2.6164698585476056e-05,
      "loss": 0.0,
      "step": 3947
    },
    {
      "epoch": 0.6640316205533597,
      "grad_norm": NaN,
      "learning_rate": 2.6141077930632646e-05,
      "loss": 0.0,
      "step": 3948
    },
    {
      "epoch": 0.6641998149861239,
      "grad_norm": NaN,
      "learning_rate": 2.6117464168364603e-05,
      "loss": 0.0,
      "step": 3949
    },
    {
      "epoch": 0.6643680094188882,
      "grad_norm": NaN,
      "learning_rate": 2.6093857305493664e-05,
      "loss": 0.0,
      "step": 3950
    },
    {
      "epoch": 0.6645362038516525,
      "grad_norm": NaN,
      "learning_rate": 2.607025734883959e-05,
      "loss": 0.0,
      "step": 3951
    },
    {
      "epoch": 0.6647043982844167,
      "grad_norm": NaN,
      "learning_rate": 2.604666430522013e-05,
      "loss": 0.0,
      "step": 3952
    },
    {
      "epoch": 0.664872592717181,
      "grad_norm": NaN,
      "learning_rate": 2.6023078181451067e-05,
      "loss": 0.0,
      "step": 3953
    },
    {
      "epoch": 0.6650407871499453,
      "grad_norm": NaN,
      "learning_rate": 2.5999498984346155e-05,
      "loss": 0.0,
      "step": 3954
    },
    {
      "epoch": 0.6652089815827096,
      "grad_norm": NaN,
      "learning_rate": 2.597592672071717e-05,
      "loss": 0.0,
      "step": 3955
    },
    {
      "epoch": 0.6653771760154739,
      "grad_norm": NaN,
      "learning_rate": 2.5952361397373848e-05,
      "loss": 0.0,
      "step": 3956
    },
    {
      "epoch": 0.6655453704482381,
      "grad_norm": NaN,
      "learning_rate": 2.592880302112399e-05,
      "loss": 0.0,
      "step": 3957
    },
    {
      "epoch": 0.6657135648810024,
      "grad_norm": NaN,
      "learning_rate": 2.5905251598773323e-05,
      "loss": 0.0,
      "step": 3958
    },
    {
      "epoch": 0.6658817593137667,
      "grad_norm": NaN,
      "learning_rate": 2.5881707137125594e-05,
      "loss": 0.0,
      "step": 3959
    },
    {
      "epoch": 0.666049953746531,
      "grad_norm": NaN,
      "learning_rate": 2.5858169642982515e-05,
      "loss": 0.0,
      "step": 3960
    },
    {
      "epoch": 0.6662181481792953,
      "grad_norm": NaN,
      "learning_rate": 2.5834639123143844e-05,
      "loss": 0.0,
      "step": 3961
    },
    {
      "epoch": 0.6663863426120595,
      "grad_norm": NaN,
      "learning_rate": 2.5811115584407253e-05,
      "loss": 0.0,
      "step": 3962
    },
    {
      "epoch": 0.6665545370448238,
      "grad_norm": NaN,
      "learning_rate": 2.578759903356845e-05,
      "loss": 0.0,
      "step": 3963
    },
    {
      "epoch": 0.6667227314775881,
      "grad_norm": NaN,
      "learning_rate": 2.5764089477421067e-05,
      "loss": 0.0,
      "step": 3964
    },
    {
      "epoch": 0.6668909259103524,
      "grad_norm": NaN,
      "learning_rate": 2.5740586922756792e-05,
      "loss": 0.0,
      "step": 3965
    },
    {
      "epoch": 0.6670591203431167,
      "grad_norm": NaN,
      "learning_rate": 2.5717091376365243e-05,
      "loss": 0.0,
      "step": 3966
    },
    {
      "epoch": 0.6672273147758809,
      "grad_norm": NaN,
      "learning_rate": 2.5693602845034003e-05,
      "loss": 0.0,
      "step": 3967
    },
    {
      "epoch": 0.6673955092086452,
      "grad_norm": NaN,
      "learning_rate": 2.5670121335548657e-05,
      "loss": 0.0,
      "step": 3968
    },
    {
      "epoch": 0.6675637036414095,
      "grad_norm": NaN,
      "learning_rate": 2.5646646854692746e-05,
      "loss": 0.0,
      "step": 3969
    },
    {
      "epoch": 0.6677318980741738,
      "grad_norm": NaN,
      "learning_rate": 2.562317940924779e-05,
      "loss": 0.0,
      "step": 3970
    },
    {
      "epoch": 0.667900092506938,
      "grad_norm": NaN,
      "learning_rate": 2.559971900599326e-05,
      "loss": 0.0,
      "step": 3971
    },
    {
      "epoch": 0.6680682869397023,
      "grad_norm": NaN,
      "learning_rate": 2.55762656517066e-05,
      "loss": 0.0,
      "step": 3972
    },
    {
      "epoch": 0.6682364813724666,
      "grad_norm": NaN,
      "learning_rate": 2.5552819353163244e-05,
      "loss": 0.0,
      "step": 3973
    },
    {
      "epoch": 0.6684046758052309,
      "grad_norm": NaN,
      "learning_rate": 2.5529380117136554e-05,
      "loss": 0.0,
      "step": 3974
    },
    {
      "epoch": 0.6685728702379952,
      "grad_norm": NaN,
      "learning_rate": 2.5505947950397863e-05,
      "loss": 0.0,
      "step": 3975
    },
    {
      "epoch": 0.6687410646707594,
      "grad_norm": NaN,
      "learning_rate": 2.548252285971644e-05,
      "loss": 0.0,
      "step": 3976
    },
    {
      "epoch": 0.6689092591035237,
      "grad_norm": NaN,
      "learning_rate": 2.545910485185957e-05,
      "loss": 0.0,
      "step": 3977
    },
    {
      "epoch": 0.669077453536288,
      "grad_norm": NaN,
      "learning_rate": 2.5435693933592432e-05,
      "loss": 0.0,
      "step": 3978
    },
    {
      "epoch": 0.6692456479690522,
      "grad_norm": NaN,
      "learning_rate": 2.5412290111678184e-05,
      "loss": 0.0,
      "step": 3979
    },
    {
      "epoch": 0.6694138424018165,
      "grad_norm": NaN,
      "learning_rate": 2.5388893392877904e-05,
      "loss": 0.0,
      "step": 3980
    },
    {
      "epoch": 0.6695820368345807,
      "grad_norm": NaN,
      "learning_rate": 2.5365503783950685e-05,
      "loss": 0.0,
      "step": 3981
    },
    {
      "epoch": 0.669750231267345,
      "grad_norm": NaN,
      "learning_rate": 2.5342121291653498e-05,
      "loss": 0.0,
      "step": 3982
    },
    {
      "epoch": 0.6699184257001093,
      "grad_norm": NaN,
      "learning_rate": 2.5318745922741282e-05,
      "loss": 0.0,
      "step": 3983
    },
    {
      "epoch": 0.6700866201328736,
      "grad_norm": NaN,
      "learning_rate": 2.529537768396691e-05,
      "loss": 0.0,
      "step": 3984
    },
    {
      "epoch": 0.6702548145656378,
      "grad_norm": NaN,
      "learning_rate": 2.5272016582081236e-05,
      "loss": 0.0,
      "step": 3985
    },
    {
      "epoch": 0.6704230089984021,
      "grad_norm": NaN,
      "learning_rate": 2.5248662623832997e-05,
      "loss": 0.0,
      "step": 3986
    },
    {
      "epoch": 0.6705912034311664,
      "grad_norm": NaN,
      "learning_rate": 2.5225315815968896e-05,
      "loss": 0.0,
      "step": 3987
    },
    {
      "epoch": 0.6707593978639307,
      "grad_norm": NaN,
      "learning_rate": 2.520197616523357e-05,
      "loss": 0.0,
      "step": 3988
    },
    {
      "epoch": 0.670927592296695,
      "grad_norm": NaN,
      "learning_rate": 2.5178643678369572e-05,
      "loss": 0.0,
      "step": 3989
    },
    {
      "epoch": 0.6710957867294592,
      "grad_norm": NaN,
      "learning_rate": 2.51553183621174e-05,
      "loss": 0.0,
      "step": 3990
    },
    {
      "epoch": 0.6712639811622235,
      "grad_norm": NaN,
      "learning_rate": 2.5132000223215478e-05,
      "loss": 0.0,
      "step": 3991
    },
    {
      "epoch": 0.6714321755949878,
      "grad_norm": NaN,
      "learning_rate": 2.5108689268400132e-05,
      "loss": 0.0,
      "step": 3992
    },
    {
      "epoch": 0.6716003700277521,
      "grad_norm": NaN,
      "learning_rate": 2.5085385504405678e-05,
      "loss": 0.0,
      "step": 3993
    },
    {
      "epoch": 0.6717685644605164,
      "grad_norm": NaN,
      "learning_rate": 2.506208893796429e-05,
      "loss": 0.0,
      "step": 3994
    },
    {
      "epoch": 0.6719367588932806,
      "grad_norm": NaN,
      "learning_rate": 2.5038799575806095e-05,
      "loss": 0.0,
      "step": 3995
    },
    {
      "epoch": 0.6721049533260449,
      "grad_norm": NaN,
      "learning_rate": 2.5015517424659095e-05,
      "loss": 0.0,
      "step": 3996
    },
    {
      "epoch": 0.6722731477588092,
      "grad_norm": NaN,
      "learning_rate": 2.499224249124929e-05,
      "loss": 0.0,
      "step": 3997
    },
    {
      "epoch": 0.6724413421915735,
      "grad_norm": NaN,
      "learning_rate": 2.4968974782300524e-05,
      "loss": 0.0,
      "step": 3998
    },
    {
      "epoch": 0.6726095366243378,
      "grad_norm": NaN,
      "learning_rate": 2.4945714304534585e-05,
      "loss": 0.0,
      "step": 3999
    },
    {
      "epoch": 0.672777731057102,
      "grad_norm": NaN,
      "learning_rate": 2.4922461064671156e-05,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 0.6729459254898663,
      "grad_norm": NaN,
      "learning_rate": 2.4899215069427823e-05,
      "loss": 0.0,
      "step": 4001
    },
    {
      "epoch": 0.6731141199226306,
      "grad_norm": NaN,
      "learning_rate": 2.487597632552013e-05,
      "loss": 0.0,
      "step": 4002
    },
    {
      "epoch": 0.6732823143553949,
      "grad_norm": NaN,
      "learning_rate": 2.4852744839661467e-05,
      "loss": 0.0,
      "step": 4003
    },
    {
      "epoch": 0.6734505087881592,
      "grad_norm": NaN,
      "learning_rate": 2.4829520618563162e-05,
      "loss": 0.0,
      "step": 4004
    },
    {
      "epoch": 0.6736187032209234,
      "grad_norm": NaN,
      "learning_rate": 2.4806303668934422e-05,
      "loss": 0.0,
      "step": 4005
    },
    {
      "epoch": 0.6737868976536877,
      "grad_norm": NaN,
      "learning_rate": 2.4783093997482364e-05,
      "loss": 0.0,
      "step": 4006
    },
    {
      "epoch": 0.673955092086452,
      "grad_norm": NaN,
      "learning_rate": 2.4759891610912005e-05,
      "loss": 0.0,
      "step": 4007
    },
    {
      "epoch": 0.6741232865192163,
      "grad_norm": NaN,
      "learning_rate": 2.4736696515926256e-05,
      "loss": 0.0,
      "step": 4008
    },
    {
      "epoch": 0.6742914809519804,
      "grad_norm": NaN,
      "learning_rate": 2.4713508719225896e-05,
      "loss": 0.0,
      "step": 4009
    },
    {
      "epoch": 0.6744596753847447,
      "grad_norm": NaN,
      "learning_rate": 2.4690328227509668e-05,
      "loss": 0.0,
      "step": 4010
    },
    {
      "epoch": 0.674627869817509,
      "grad_norm": NaN,
      "learning_rate": 2.4667155047474127e-05,
      "loss": 0.0,
      "step": 4011
    },
    {
      "epoch": 0.6747960642502733,
      "grad_norm": NaN,
      "learning_rate": 2.464398918581375e-05,
      "loss": 0.0,
      "step": 4012
    },
    {
      "epoch": 0.6749642586830376,
      "grad_norm": NaN,
      "learning_rate": 2.4620830649220873e-05,
      "loss": 0.0,
      "step": 4013
    },
    {
      "epoch": 0.6751324531158018,
      "grad_norm": NaN,
      "learning_rate": 2.4597679444385772e-05,
      "loss": 0.0,
      "step": 4014
    },
    {
      "epoch": 0.6753006475485661,
      "grad_norm": NaN,
      "learning_rate": 2.4574535577996566e-05,
      "loss": 0.0,
      "step": 4015
    },
    {
      "epoch": 0.6754688419813304,
      "grad_norm": NaN,
      "learning_rate": 2.4551399056739245e-05,
      "loss": 0.0,
      "step": 4016
    },
    {
      "epoch": 0.6756370364140947,
      "grad_norm": NaN,
      "learning_rate": 2.4528269887297677e-05,
      "loss": 0.0,
      "step": 4017
    },
    {
      "epoch": 0.675805230846859,
      "grad_norm": NaN,
      "learning_rate": 2.4505148076353662e-05,
      "loss": 0.0,
      "step": 4018
    },
    {
      "epoch": 0.6759734252796232,
      "grad_norm": NaN,
      "learning_rate": 2.4482033630586804e-05,
      "loss": 0.0,
      "step": 4019
    },
    {
      "epoch": 0.6761416197123875,
      "grad_norm": NaN,
      "learning_rate": 2.4458926556674615e-05,
      "loss": 0.0,
      "step": 4020
    },
    {
      "epoch": 0.6763098141451518,
      "grad_norm": NaN,
      "learning_rate": 2.443582686129245e-05,
      "loss": 0.0,
      "step": 4021
    },
    {
      "epoch": 0.6764780085779161,
      "grad_norm": NaN,
      "learning_rate": 2.4412734551113585e-05,
      "loss": 0.0,
      "step": 4022
    },
    {
      "epoch": 0.6766462030106803,
      "grad_norm": NaN,
      "learning_rate": 2.4389649632809115e-05,
      "loss": 0.0,
      "step": 4023
    },
    {
      "epoch": 0.6768143974434446,
      "grad_norm": NaN,
      "learning_rate": 2.4366572113048014e-05,
      "loss": 0.0,
      "step": 4024
    },
    {
      "epoch": 0.6769825918762089,
      "grad_norm": NaN,
      "learning_rate": 2.434350199849712e-05,
      "loss": 0.0,
      "step": 4025
    },
    {
      "epoch": 0.6771507863089732,
      "grad_norm": NaN,
      "learning_rate": 2.432043929582113e-05,
      "loss": 0.0,
      "step": 4026
    },
    {
      "epoch": 0.6773189807417375,
      "grad_norm": NaN,
      "learning_rate": 2.4297384011682595e-05,
      "loss": 0.0,
      "step": 4027
    },
    {
      "epoch": 0.6774871751745017,
      "grad_norm": NaN,
      "learning_rate": 2.427433615274194e-05,
      "loss": 0.0,
      "step": 4028
    },
    {
      "epoch": 0.677655369607266,
      "grad_norm": NaN,
      "learning_rate": 2.4251295725657404e-05,
      "loss": 0.0,
      "step": 4029
    },
    {
      "epoch": 0.6778235640400303,
      "grad_norm": NaN,
      "learning_rate": 2.422826273708515e-05,
      "loss": 0.0,
      "step": 4030
    },
    {
      "epoch": 0.6779917584727946,
      "grad_norm": NaN,
      "learning_rate": 2.420523719367913e-05,
      "loss": 0.0,
      "step": 4031
    },
    {
      "epoch": 0.6781599529055589,
      "grad_norm": NaN,
      "learning_rate": 2.418221910209117e-05,
      "loss": 0.0,
      "step": 4032
    },
    {
      "epoch": 0.6783281473383231,
      "grad_norm": NaN,
      "learning_rate": 2.415920846897091e-05,
      "loss": 0.0,
      "step": 4033
    },
    {
      "epoch": 0.6784963417710874,
      "grad_norm": NaN,
      "learning_rate": 2.413620530096592e-05,
      "loss": 0.0,
      "step": 4034
    },
    {
      "epoch": 0.6786645362038517,
      "grad_norm": NaN,
      "learning_rate": 2.411320960472152e-05,
      "loss": 0.0,
      "step": 4035
    },
    {
      "epoch": 0.678832730636616,
      "grad_norm": NaN,
      "learning_rate": 2.409022138688092e-05,
      "loss": 0.0,
      "step": 4036
    },
    {
      "epoch": 0.6790009250693803,
      "grad_norm": NaN,
      "learning_rate": 2.4067240654085132e-05,
      "loss": 0.0,
      "step": 4037
    },
    {
      "epoch": 0.6791691195021444,
      "grad_norm": NaN,
      "learning_rate": 2.4044267412973078e-05,
      "loss": 0.0,
      "step": 4038
    },
    {
      "epoch": 0.6793373139349087,
      "grad_norm": NaN,
      "learning_rate": 2.4021301670181446e-05,
      "loss": 0.0,
      "step": 4039
    },
    {
      "epoch": 0.679505508367673,
      "grad_norm": NaN,
      "learning_rate": 2.3998343432344784e-05,
      "loss": 0.0,
      "step": 4040
    },
    {
      "epoch": 0.6796737028004373,
      "grad_norm": NaN,
      "learning_rate": 2.3975392706095446e-05,
      "loss": 0.0,
      "step": 4041
    },
    {
      "epoch": 0.6798418972332015,
      "grad_norm": NaN,
      "learning_rate": 2.3952449498063683e-05,
      "loss": 0.0,
      "step": 4042
    },
    {
      "epoch": 0.6800100916659658,
      "grad_norm": NaN,
      "learning_rate": 2.3929513814877514e-05,
      "loss": 0.0,
      "step": 4043
    },
    {
      "epoch": 0.6801782860987301,
      "grad_norm": NaN,
      "learning_rate": 2.39065856631628e-05,
      "loss": 0.0,
      "step": 4044
    },
    {
      "epoch": 0.6803464805314944,
      "grad_norm": NaN,
      "learning_rate": 2.388366504954322e-05,
      "loss": 0.0,
      "step": 4045
    },
    {
      "epoch": 0.6805146749642587,
      "grad_norm": NaN,
      "learning_rate": 2.386075198064029e-05,
      "loss": 0.0,
      "step": 4046
    },
    {
      "epoch": 0.6806828693970229,
      "grad_norm": NaN,
      "learning_rate": 2.383784646307334e-05,
      "loss": 0.0,
      "step": 4047
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": NaN,
      "learning_rate": 2.3814948503459507e-05,
      "loss": 0.0,
      "step": 4048
    },
    {
      "epoch": 0.6810192582625515,
      "grad_norm": NaN,
      "learning_rate": 2.379205810841376e-05,
      "loss": 0.0,
      "step": 4049
    },
    {
      "epoch": 0.6811874526953158,
      "grad_norm": NaN,
      "learning_rate": 2.3769175284548868e-05,
      "loss": 0.0,
      "step": 4050
    },
    {
      "epoch": 0.68135564712808,
      "grad_norm": NaN,
      "learning_rate": 2.374630003847545e-05,
      "loss": 0.0,
      "step": 4051
    },
    {
      "epoch": 0.6815238415608443,
      "grad_norm": NaN,
      "learning_rate": 2.372343237680189e-05,
      "loss": 0.0,
      "step": 4052
    },
    {
      "epoch": 0.6816920359936086,
      "grad_norm": NaN,
      "learning_rate": 2.3700572306134412e-05,
      "loss": 0.0,
      "step": 4053
    },
    {
      "epoch": 0.6818602304263729,
      "grad_norm": NaN,
      "learning_rate": 2.3677719833076996e-05,
      "loss": 0.0,
      "step": 4054
    },
    {
      "epoch": 0.6820284248591372,
      "grad_norm": NaN,
      "learning_rate": 2.3654874964231518e-05,
      "loss": 0.0,
      "step": 4055
    },
    {
      "epoch": 0.6821966192919015,
      "grad_norm": NaN,
      "learning_rate": 2.3632037706197573e-05,
      "loss": 0.0,
      "step": 4056
    },
    {
      "epoch": 0.6823648137246657,
      "grad_norm": NaN,
      "learning_rate": 2.36092080655726e-05,
      "loss": 0.0,
      "step": 4057
    },
    {
      "epoch": 0.68253300815743,
      "grad_norm": NaN,
      "learning_rate": 2.358638604895181e-05,
      "loss": 0.0,
      "step": 4058
    },
    {
      "epoch": 0.6827012025901943,
      "grad_norm": NaN,
      "learning_rate": 2.3563571662928252e-05,
      "loss": 0.0,
      "step": 4059
    },
    {
      "epoch": 0.6828693970229586,
      "grad_norm": NaN,
      "learning_rate": 2.354076491409274e-05,
      "loss": 0.0,
      "step": 4060
    },
    {
      "epoch": 0.6830375914557228,
      "grad_norm": NaN,
      "learning_rate": 2.3517965809033882e-05,
      "loss": 0.0,
      "step": 4061
    },
    {
      "epoch": 0.6832057858884871,
      "grad_norm": NaN,
      "learning_rate": 2.3495174354338084e-05,
      "loss": 0.0,
      "step": 4062
    },
    {
      "epoch": 0.6833739803212514,
      "grad_norm": NaN,
      "learning_rate": 2.3472390556589546e-05,
      "loss": 0.0,
      "step": 4063
    },
    {
      "epoch": 0.6835421747540157,
      "grad_norm": NaN,
      "learning_rate": 2.3449614422370253e-05,
      "loss": 0.0,
      "step": 4064
    },
    {
      "epoch": 0.68371036918678,
      "grad_norm": NaN,
      "learning_rate": 2.3426845958259968e-05,
      "loss": 0.0,
      "step": 4065
    },
    {
      "epoch": 0.6838785636195442,
      "grad_norm": NaN,
      "learning_rate": 2.3404085170836233e-05,
      "loss": 0.0,
      "step": 4066
    },
    {
      "epoch": 0.6840467580523085,
      "grad_norm": NaN,
      "learning_rate": 2.3381332066674422e-05,
      "loss": 0.0,
      "step": 4067
    },
    {
      "epoch": 0.6842149524850727,
      "grad_norm": NaN,
      "learning_rate": 2.3358586652347624e-05,
      "loss": 0.0,
      "step": 4068
    },
    {
      "epoch": 0.684383146917837,
      "grad_norm": NaN,
      "learning_rate": 2.3335848934426746e-05,
      "loss": 0.0,
      "step": 4069
    },
    {
      "epoch": 0.6845513413506013,
      "grad_norm": NaN,
      "learning_rate": 2.3313118919480436e-05,
      "loss": 0.0,
      "step": 4070
    },
    {
      "epoch": 0.6847195357833655,
      "grad_norm": NaN,
      "learning_rate": 2.329039661407518e-05,
      "loss": 0.0,
      "step": 4071
    },
    {
      "epoch": 0.6848877302161298,
      "grad_norm": NaN,
      "learning_rate": 2.3267682024775173e-05,
      "loss": 0.0,
      "step": 4072
    },
    {
      "epoch": 0.6850559246488941,
      "grad_norm": NaN,
      "learning_rate": 2.3244975158142413e-05,
      "loss": 0.0,
      "step": 4073
    },
    {
      "epoch": 0.6852241190816584,
      "grad_norm": NaN,
      "learning_rate": 2.3222276020736634e-05,
      "loss": 0.0,
      "step": 4074
    },
    {
      "epoch": 0.6853923135144226,
      "grad_norm": NaN,
      "learning_rate": 2.3199584619115406e-05,
      "loss": 0.0,
      "step": 4075
    },
    {
      "epoch": 0.6855605079471869,
      "grad_norm": NaN,
      "learning_rate": 2.3176900959834004e-05,
      "loss": 0.0,
      "step": 4076
    },
    {
      "epoch": 0.6857287023799512,
      "grad_norm": NaN,
      "learning_rate": 2.315422504944547e-05,
      "loss": 0.0,
      "step": 4077
    },
    {
      "epoch": 0.6858968968127155,
      "grad_norm": NaN,
      "learning_rate": 2.3131556894500616e-05,
      "loss": 0.0,
      "step": 4078
    },
    {
      "epoch": 0.6860650912454798,
      "grad_norm": NaN,
      "learning_rate": 2.310889650154805e-05,
      "loss": 0.0,
      "step": 4079
    },
    {
      "epoch": 0.686233285678244,
      "grad_norm": NaN,
      "learning_rate": 2.308624387713409e-05,
      "loss": 0.0,
      "step": 4080
    },
    {
      "epoch": 0.6864014801110083,
      "grad_norm": NaN,
      "learning_rate": 2.3063599027802818e-05,
      "loss": 0.0,
      "step": 4081
    },
    {
      "epoch": 0.6865696745437726,
      "grad_norm": NaN,
      "learning_rate": 2.3040961960096086e-05,
      "loss": 0.0,
      "step": 4082
    },
    {
      "epoch": 0.6867378689765369,
      "grad_norm": NaN,
      "learning_rate": 2.3018332680553477e-05,
      "loss": 0.0,
      "step": 4083
    },
    {
      "epoch": 0.6869060634093012,
      "grad_norm": NaN,
      "learning_rate": 2.2995711195712354e-05,
      "loss": 0.0,
      "step": 4084
    },
    {
      "epoch": 0.6870742578420654,
      "grad_norm": NaN,
      "learning_rate": 2.2973097512107796e-05,
      "loss": 0.0,
      "step": 4085
    },
    {
      "epoch": 0.6872424522748297,
      "grad_norm": NaN,
      "learning_rate": 2.2950491636272632e-05,
      "loss": 0.0,
      "step": 4086
    },
    {
      "epoch": 0.687410646707594,
      "grad_norm": NaN,
      "learning_rate": 2.292789357473748e-05,
      "loss": 0.0,
      "step": 4087
    },
    {
      "epoch": 0.6875788411403583,
      "grad_norm": NaN,
      "learning_rate": 2.290530333403065e-05,
      "loss": 0.0,
      "step": 4088
    },
    {
      "epoch": 0.6877470355731226,
      "grad_norm": NaN,
      "learning_rate": 2.28827209206782e-05,
      "loss": 0.0,
      "step": 4089
    },
    {
      "epoch": 0.6879152300058868,
      "grad_norm": NaN,
      "learning_rate": 2.2860146341203937e-05,
      "loss": 0.0,
      "step": 4090
    },
    {
      "epoch": 0.6880834244386511,
      "grad_norm": NaN,
      "learning_rate": 2.2837579602129426e-05,
      "loss": 0.0,
      "step": 4091
    },
    {
      "epoch": 0.6882516188714154,
      "grad_norm": NaN,
      "learning_rate": 2.281502070997394e-05,
      "loss": 0.0,
      "step": 4092
    },
    {
      "epoch": 0.6884198133041797,
      "grad_norm": NaN,
      "learning_rate": 2.2792469671254475e-05,
      "loss": 0.0,
      "step": 4093
    },
    {
      "epoch": 0.688588007736944,
      "grad_norm": NaN,
      "learning_rate": 2.276992649248576e-05,
      "loss": 0.0,
      "step": 4094
    },
    {
      "epoch": 0.6887562021697082,
      "grad_norm": NaN,
      "learning_rate": 2.2747391180180307e-05,
      "loss": 0.0,
      "step": 4095
    },
    {
      "epoch": 0.6889243966024725,
      "grad_norm": NaN,
      "learning_rate": 2.2724863740848296e-05,
      "loss": 0.0,
      "step": 4096
    },
    {
      "epoch": 0.6890925910352367,
      "grad_norm": NaN,
      "learning_rate": 2.2702344180997647e-05,
      "loss": 0.0,
      "step": 4097
    },
    {
      "epoch": 0.689260785468001,
      "grad_norm": NaN,
      "learning_rate": 2.2679832507134018e-05,
      "loss": 0.0,
      "step": 4098
    },
    {
      "epoch": 0.6894289799007652,
      "grad_norm": NaN,
      "learning_rate": 2.2657328725760774e-05,
      "loss": 0.0,
      "step": 4099
    },
    {
      "epoch": 0.6895971743335295,
      "grad_norm": NaN,
      "learning_rate": 2.2634832843379006e-05,
      "loss": 0.0,
      "step": 4100
    },
    {
      "epoch": 0.6897653687662938,
      "grad_norm": NaN,
      "learning_rate": 2.2612344866487523e-05,
      "loss": 0.0,
      "step": 4101
    },
    {
      "epoch": 0.6899335631990581,
      "grad_norm": NaN,
      "learning_rate": 2.2589864801582843e-05,
      "loss": 0.0,
      "step": 4102
    },
    {
      "epoch": 0.6901017576318224,
      "grad_norm": NaN,
      "learning_rate": 2.2567392655159197e-05,
      "loss": 0.0,
      "step": 4103
    },
    {
      "epoch": 0.6902699520645866,
      "grad_norm": NaN,
      "learning_rate": 2.254492843370857e-05,
      "loss": 0.0,
      "step": 4104
    },
    {
      "epoch": 0.6904381464973509,
      "grad_norm": NaN,
      "learning_rate": 2.2522472143720597e-05,
      "loss": 0.0,
      "step": 4105
    },
    {
      "epoch": 0.6906063409301152,
      "grad_norm": NaN,
      "learning_rate": 2.2500023791682662e-05,
      "loss": 0.0,
      "step": 4106
    },
    {
      "epoch": 0.6907745353628795,
      "grad_norm": NaN,
      "learning_rate": 2.2477583384079816e-05,
      "loss": 0.0,
      "step": 4107
    },
    {
      "epoch": 0.6909427297956438,
      "grad_norm": NaN,
      "learning_rate": 2.245515092739488e-05,
      "loss": 0.0,
      "step": 4108
    },
    {
      "epoch": 0.691110924228408,
      "grad_norm": NaN,
      "learning_rate": 2.2432726428108318e-05,
      "loss": 0.0,
      "step": 4109
    },
    {
      "epoch": 0.6912791186611723,
      "grad_norm": NaN,
      "learning_rate": 2.2410309892698322e-05,
      "loss": 0.0,
      "step": 4110
    },
    {
      "epoch": 0.6914473130939366,
      "grad_norm": NaN,
      "learning_rate": 2.238790132764076e-05,
      "loss": 0.0,
      "step": 4111
    },
    {
      "epoch": 0.6916155075267009,
      "grad_norm": NaN,
      "learning_rate": 2.2365500739409244e-05,
      "loss": 0.0,
      "step": 4112
    },
    {
      "epoch": 0.6917837019594651,
      "grad_norm": NaN,
      "learning_rate": 2.234310813447505e-05,
      "loss": 0.0,
      "step": 4113
    },
    {
      "epoch": 0.6919518963922294,
      "grad_norm": NaN,
      "learning_rate": 2.2320723519307136e-05,
      "loss": 0.0,
      "step": 4114
    },
    {
      "epoch": 0.6921200908249937,
      "grad_norm": NaN,
      "learning_rate": 2.2298346900372146e-05,
      "loss": 0.0,
      "step": 4115
    },
    {
      "epoch": 0.692288285257758,
      "grad_norm": NaN,
      "learning_rate": 2.2275978284134486e-05,
      "loss": 0.0,
      "step": 4116
    },
    {
      "epoch": 0.6924564796905223,
      "grad_norm": NaN,
      "learning_rate": 2.2253617677056172e-05,
      "loss": 0.0,
      "step": 4117
    },
    {
      "epoch": 0.6926246741232865,
      "grad_norm": NaN,
      "learning_rate": 2.2231265085596938e-05,
      "loss": 0.0,
      "step": 4118
    },
    {
      "epoch": 0.6927928685560508,
      "grad_norm": NaN,
      "learning_rate": 2.220892051621419e-05,
      "loss": 0.0,
      "step": 4119
    },
    {
      "epoch": 0.6929610629888151,
      "grad_norm": NaN,
      "learning_rate": 2.2186583975363022e-05,
      "loss": 0.0,
      "step": 4120
    },
    {
      "epoch": 0.6931292574215794,
      "grad_norm": NaN,
      "learning_rate": 2.216425546949622e-05,
      "loss": 0.0,
      "step": 4121
    },
    {
      "epoch": 0.6932974518543437,
      "grad_norm": NaN,
      "learning_rate": 2.214193500506423e-05,
      "loss": 0.0,
      "step": 4122
    },
    {
      "epoch": 0.6934656462871079,
      "grad_norm": NaN,
      "learning_rate": 2.2119622588515176e-05,
      "loss": 0.0,
      "step": 4123
    },
    {
      "epoch": 0.6936338407198722,
      "grad_norm": NaN,
      "learning_rate": 2.2097318226294893e-05,
      "loss": 0.0,
      "step": 4124
    },
    {
      "epoch": 0.6938020351526365,
      "grad_norm": NaN,
      "learning_rate": 2.207502192484685e-05,
      "loss": 0.0,
      "step": 4125
    },
    {
      "epoch": 0.6939702295854008,
      "grad_norm": NaN,
      "learning_rate": 2.2052733690612193e-05,
      "loss": 0.0,
      "step": 4126
    },
    {
      "epoch": 0.694138424018165,
      "grad_norm": NaN,
      "learning_rate": 2.2030453530029725e-05,
      "loss": 0.0,
      "step": 4127
    },
    {
      "epoch": 0.6943066184509292,
      "grad_norm": NaN,
      "learning_rate": 2.2008181449535976e-05,
      "loss": 0.0,
      "step": 4128
    },
    {
      "epoch": 0.6944748128836935,
      "grad_norm": NaN,
      "learning_rate": 2.1985917455565082e-05,
      "loss": 0.0,
      "step": 4129
    },
    {
      "epoch": 0.6946430073164578,
      "grad_norm": NaN,
      "learning_rate": 2.1963661554548853e-05,
      "loss": 0.0,
      "step": 4130
    },
    {
      "epoch": 0.6948112017492221,
      "grad_norm": NaN,
      "learning_rate": 2.194141375291676e-05,
      "loss": 0.0,
      "step": 4131
    },
    {
      "epoch": 0.6949793961819863,
      "grad_norm": NaN,
      "learning_rate": 2.191917405709598e-05,
      "loss": 0.0,
      "step": 4132
    },
    {
      "epoch": 0.6951475906147506,
      "grad_norm": NaN,
      "learning_rate": 2.1896942473511294e-05,
      "loss": 0.0,
      "step": 4133
    },
    {
      "epoch": 0.6953157850475149,
      "grad_norm": NaN,
      "learning_rate": 2.1874719008585154e-05,
      "loss": 0.0,
      "step": 4134
    },
    {
      "epoch": 0.6954839794802792,
      "grad_norm": NaN,
      "learning_rate": 2.185250366873765e-05,
      "loss": 0.0,
      "step": 4135
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": NaN,
      "learning_rate": 2.1830296460386584e-05,
      "loss": 0.0,
      "step": 4136
    },
    {
      "epoch": 0.6958203683458077,
      "grad_norm": NaN,
      "learning_rate": 2.1808097389947356e-05,
      "loss": 0.0,
      "step": 4137
    },
    {
      "epoch": 0.695988562778572,
      "grad_norm": NaN,
      "learning_rate": 2.1785906463833023e-05,
      "loss": 0.0,
      "step": 4138
    },
    {
      "epoch": 0.6961567572113363,
      "grad_norm": NaN,
      "learning_rate": 2.1763723688454298e-05,
      "loss": 0.0,
      "step": 4139
    },
    {
      "epoch": 0.6963249516441006,
      "grad_norm": NaN,
      "learning_rate": 2.174154907021953e-05,
      "loss": 0.0,
      "step": 4140
    },
    {
      "epoch": 0.6964931460768649,
      "grad_norm": NaN,
      "learning_rate": 2.171938261553472e-05,
      "loss": 0.0,
      "step": 4141
    },
    {
      "epoch": 0.6966613405096291,
      "grad_norm": NaN,
      "learning_rate": 2.1697224330803522e-05,
      "loss": 0.0,
      "step": 4142
    },
    {
      "epoch": 0.6968295349423934,
      "grad_norm": NaN,
      "learning_rate": 2.1675074222427183e-05,
      "loss": 0.0,
      "step": 4143
    },
    {
      "epoch": 0.6969977293751577,
      "grad_norm": NaN,
      "learning_rate": 2.1652932296804663e-05,
      "loss": 0.0,
      "step": 4144
    },
    {
      "epoch": 0.697165923807922,
      "grad_norm": NaN,
      "learning_rate": 2.1630798560332492e-05,
      "loss": 0.0,
      "step": 4145
    },
    {
      "epoch": 0.6973341182406863,
      "grad_norm": NaN,
      "learning_rate": 2.160867301940487e-05,
      "loss": 0.0,
      "step": 4146
    },
    {
      "epoch": 0.6975023126734505,
      "grad_norm": NaN,
      "learning_rate": 2.1586555680413606e-05,
      "loss": 0.0,
      "step": 4147
    },
    {
      "epoch": 0.6976705071062148,
      "grad_norm": NaN,
      "learning_rate": 2.156444654974814e-05,
      "loss": 0.0,
      "step": 4148
    },
    {
      "epoch": 0.6978387015389791,
      "grad_norm": NaN,
      "learning_rate": 2.154234563379559e-05,
      "loss": 0.0,
      "step": 4149
    },
    {
      "epoch": 0.6980068959717434,
      "grad_norm": NaN,
      "learning_rate": 2.1520252938940643e-05,
      "loss": 0.0,
      "step": 4150
    },
    {
      "epoch": 0.6981750904045076,
      "grad_norm": NaN,
      "learning_rate": 2.1498168471565627e-05,
      "loss": 0.0,
      "step": 4151
    },
    {
      "epoch": 0.6983432848372719,
      "grad_norm": NaN,
      "learning_rate": 2.147609223805049e-05,
      "loss": 0.0,
      "step": 4152
    },
    {
      "epoch": 0.6985114792700362,
      "grad_norm": NaN,
      "learning_rate": 2.145402424477283e-05,
      "loss": 0.0,
      "step": 4153
    },
    {
      "epoch": 0.6986796737028005,
      "grad_norm": NaN,
      "learning_rate": 2.143196449810783e-05,
      "loss": 0.0,
      "step": 4154
    },
    {
      "epoch": 0.6988478681355648,
      "grad_norm": NaN,
      "learning_rate": 2.140991300442831e-05,
      "loss": 0.0,
      "step": 4155
    },
    {
      "epoch": 0.6990160625683289,
      "grad_norm": NaN,
      "learning_rate": 2.13878697701047e-05,
      "loss": 0.0,
      "step": 4156
    },
    {
      "epoch": 0.6991842570010932,
      "grad_norm": NaN,
      "learning_rate": 2.1365834801505036e-05,
      "loss": 0.0,
      "step": 4157
    },
    {
      "epoch": 0.6993524514338575,
      "grad_norm": NaN,
      "learning_rate": 2.134380810499497e-05,
      "loss": 0.0,
      "step": 4158
    },
    {
      "epoch": 0.6995206458666218,
      "grad_norm": NaN,
      "learning_rate": 2.1321789686937787e-05,
      "loss": 0.0,
      "step": 4159
    },
    {
      "epoch": 0.699688840299386,
      "grad_norm": NaN,
      "learning_rate": 2.1299779553694323e-05,
      "loss": 0.0,
      "step": 4160
    },
    {
      "epoch": 0.6998570347321503,
      "grad_norm": NaN,
      "learning_rate": 2.1277777711623104e-05,
      "loss": 0.0,
      "step": 4161
    },
    {
      "epoch": 0.7000252291649146,
      "grad_norm": NaN,
      "learning_rate": 2.12557841670802e-05,
      "loss": 0.0,
      "step": 4162
    },
    {
      "epoch": 0.7001934235976789,
      "grad_norm": NaN,
      "learning_rate": 2.1233798926419295e-05,
      "loss": 0.0,
      "step": 4163
    },
    {
      "epoch": 0.7003616180304432,
      "grad_norm": NaN,
      "learning_rate": 2.121182199599166e-05,
      "loss": 0.0,
      "step": 4164
    },
    {
      "epoch": 0.7005298124632074,
      "grad_norm": NaN,
      "learning_rate": 2.1189853382146226e-05,
      "loss": 0.0,
      "step": 4165
    },
    {
      "epoch": 0.7006980068959717,
      "grad_norm": NaN,
      "learning_rate": 2.1167893091229452e-05,
      "loss": 0.0,
      "step": 4166
    },
    {
      "epoch": 0.700866201328736,
      "grad_norm": NaN,
      "learning_rate": 2.1145941129585435e-05,
      "loss": 0.0,
      "step": 4167
    },
    {
      "epoch": 0.7010343957615003,
      "grad_norm": NaN,
      "learning_rate": 2.1123997503555813e-05,
      "loss": 0.0,
      "step": 4168
    },
    {
      "epoch": 0.7012025901942646,
      "grad_norm": NaN,
      "learning_rate": 2.1102062219479906e-05,
      "loss": 0.0,
      "step": 4169
    },
    {
      "epoch": 0.7013707846270288,
      "grad_norm": NaN,
      "learning_rate": 2.1080135283694547e-05,
      "loss": 0.0,
      "step": 4170
    },
    {
      "epoch": 0.7015389790597931,
      "grad_norm": NaN,
      "learning_rate": 2.105821670253419e-05,
      "loss": 0.0,
      "step": 4171
    },
    {
      "epoch": 0.7017071734925574,
      "grad_norm": NaN,
      "learning_rate": 2.1036306482330837e-05,
      "loss": 0.0,
      "step": 4172
    },
    {
      "epoch": 0.7018753679253217,
      "grad_norm": NaN,
      "learning_rate": 2.1014404629414148e-05,
      "loss": 0.0,
      "step": 4173
    },
    {
      "epoch": 0.702043562358086,
      "grad_norm": NaN,
      "learning_rate": 2.09925111501113e-05,
      "loss": 0.0,
      "step": 4174
    },
    {
      "epoch": 0.7022117567908502,
      "grad_norm": NaN,
      "learning_rate": 2.0970626050747082e-05,
      "loss": 0.0,
      "step": 4175
    },
    {
      "epoch": 0.7023799512236145,
      "grad_norm": NaN,
      "learning_rate": 2.0948749337643852e-05,
      "loss": 0.0,
      "step": 4176
    },
    {
      "epoch": 0.7025481456563788,
      "grad_norm": NaN,
      "learning_rate": 2.0926881017121547e-05,
      "loss": 0.0,
      "step": 4177
    },
    {
      "epoch": 0.7027163400891431,
      "grad_norm": NaN,
      "learning_rate": 2.0905021095497674e-05,
      "loss": 0.0,
      "step": 4178
    },
    {
      "epoch": 0.7028845345219074,
      "grad_norm": NaN,
      "learning_rate": 2.0883169579087324e-05,
      "loss": 0.0,
      "step": 4179
    },
    {
      "epoch": 0.7030527289546716,
      "grad_norm": NaN,
      "learning_rate": 2.0861326474203143e-05,
      "loss": 0.0,
      "step": 4180
    },
    {
      "epoch": 0.7032209233874359,
      "grad_norm": NaN,
      "learning_rate": 2.0839491787155387e-05,
      "loss": 0.0,
      "step": 4181
    },
    {
      "epoch": 0.7033891178202002,
      "grad_norm": NaN,
      "learning_rate": 2.0817665524251835e-05,
      "loss": 0.0,
      "step": 4182
    },
    {
      "epoch": 0.7035573122529645,
      "grad_norm": NaN,
      "learning_rate": 2.0795847691797853e-05,
      "loss": 0.0,
      "step": 4183
    },
    {
      "epoch": 0.7037255066857288,
      "grad_norm": NaN,
      "learning_rate": 2.0774038296096347e-05,
      "loss": 0.0,
      "step": 4184
    },
    {
      "epoch": 0.703893701118493,
      "grad_norm": NaN,
      "learning_rate": 2.075223734344785e-05,
      "loss": 0.0,
      "step": 4185
    },
    {
      "epoch": 0.7040618955512572,
      "grad_norm": NaN,
      "learning_rate": 2.0730444840150392e-05,
      "loss": 0.0,
      "step": 4186
    },
    {
      "epoch": 0.7042300899840215,
      "grad_norm": NaN,
      "learning_rate": 2.0708660792499578e-05,
      "loss": 0.0,
      "step": 4187
    },
    {
      "epoch": 0.7043982844167858,
      "grad_norm": NaN,
      "learning_rate": 2.0686885206788565e-05,
      "loss": 0.0,
      "step": 4188
    },
    {
      "epoch": 0.70456647884955,
      "grad_norm": NaN,
      "learning_rate": 2.0665118089308107e-05,
      "loss": 0.0,
      "step": 4189
    },
    {
      "epoch": 0.7047346732823143,
      "grad_norm": NaN,
      "learning_rate": 2.0643359446346462e-05,
      "loss": 0.0,
      "step": 4190
    },
    {
      "epoch": 0.7049028677150786,
      "grad_norm": NaN,
      "learning_rate": 2.0621609284189458e-05,
      "loss": 0.0,
      "step": 4191
    },
    {
      "epoch": 0.7050710621478429,
      "grad_norm": NaN,
      "learning_rate": 2.0599867609120477e-05,
      "loss": 0.0,
      "step": 4192
    },
    {
      "epoch": 0.7052392565806072,
      "grad_norm": NaN,
      "learning_rate": 2.0578134427420443e-05,
      "loss": 0.0,
      "step": 4193
    },
    {
      "epoch": 0.7054074510133714,
      "grad_norm": NaN,
      "learning_rate": 2.0556409745367828e-05,
      "loss": 0.0,
      "step": 4194
    },
    {
      "epoch": 0.7055756454461357,
      "grad_norm": NaN,
      "learning_rate": 2.053469356923865e-05,
      "loss": 0.0,
      "step": 4195
    },
    {
      "epoch": 0.7057438398789,
      "grad_norm": NaN,
      "learning_rate": 2.0512985905306453e-05,
      "loss": 0.0,
      "step": 4196
    },
    {
      "epoch": 0.7059120343116643,
      "grad_norm": NaN,
      "learning_rate": 2.0491286759842377e-05,
      "loss": 0.0,
      "step": 4197
    },
    {
      "epoch": 0.7060802287444286,
      "grad_norm": NaN,
      "learning_rate": 2.046959613911503e-05,
      "loss": 0.0,
      "step": 4198
    },
    {
      "epoch": 0.7062484231771928,
      "grad_norm": NaN,
      "learning_rate": 2.044791404939061e-05,
      "loss": 0.0,
      "step": 4199
    },
    {
      "epoch": 0.7064166176099571,
      "grad_norm": NaN,
      "learning_rate": 2.0426240496932814e-05,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 0.7065848120427214,
      "grad_norm": NaN,
      "learning_rate": 2.0404575488002873e-05,
      "loss": 0.0,
      "step": 4201
    },
    {
      "epoch": 0.7067530064754857,
      "grad_norm": NaN,
      "learning_rate": 2.0382919028859605e-05,
      "loss": 0.0,
      "step": 4202
    },
    {
      "epoch": 0.70692120090825,
      "grad_norm": NaN,
      "learning_rate": 2.0361271125759302e-05,
      "loss": 0.0,
      "step": 4203
    },
    {
      "epoch": 0.7070893953410142,
      "grad_norm": NaN,
      "learning_rate": 2.0339631784955792e-05,
      "loss": 0.0,
      "step": 4204
    },
    {
      "epoch": 0.7072575897737785,
      "grad_norm": NaN,
      "learning_rate": 2.0318001012700433e-05,
      "loss": 0.0,
      "step": 4205
    },
    {
      "epoch": 0.7074257842065428,
      "grad_norm": NaN,
      "learning_rate": 2.0296378815242135e-05,
      "loss": 0.0,
      "step": 4206
    },
    {
      "epoch": 0.7075939786393071,
      "grad_norm": NaN,
      "learning_rate": 2.0274765198827302e-05,
      "loss": 0.0,
      "step": 4207
    },
    {
      "epoch": 0.7077621730720713,
      "grad_norm": NaN,
      "learning_rate": 2.0253160169699858e-05,
      "loss": 0.0,
      "step": 4208
    },
    {
      "epoch": 0.7079303675048356,
      "grad_norm": NaN,
      "learning_rate": 2.0231563734101243e-05,
      "loss": 0.0,
      "step": 4209
    },
    {
      "epoch": 0.7080985619375999,
      "grad_norm": NaN,
      "learning_rate": 2.0209975898270454e-05,
      "loss": 0.0,
      "step": 4210
    },
    {
      "epoch": 0.7082667563703642,
      "grad_norm": NaN,
      "learning_rate": 2.0188396668443965e-05,
      "loss": 0.0,
      "step": 4211
    },
    {
      "epoch": 0.7084349508031285,
      "grad_norm": NaN,
      "learning_rate": 2.016682605085577e-05,
      "loss": 0.0,
      "step": 4212
    },
    {
      "epoch": 0.7086031452358927,
      "grad_norm": NaN,
      "learning_rate": 2.014526405173738e-05,
      "loss": 0.0,
      "step": 4213
    },
    {
      "epoch": 0.708771339668657,
      "grad_norm": NaN,
      "learning_rate": 2.0123710677317803e-05,
      "loss": 0.0,
      "step": 4214
    },
    {
      "epoch": 0.7089395341014213,
      "grad_norm": NaN,
      "learning_rate": 2.0102165933823585e-05,
      "loss": 0.0,
      "step": 4215
    },
    {
      "epoch": 0.7091077285341855,
      "grad_norm": NaN,
      "learning_rate": 2.0080629827478753e-05,
      "loss": 0.0,
      "step": 4216
    },
    {
      "epoch": 0.7092759229669497,
      "grad_norm": NaN,
      "learning_rate": 2.0059102364504828e-05,
      "loss": 0.0,
      "step": 4217
    },
    {
      "epoch": 0.709444117399714,
      "grad_norm": NaN,
      "learning_rate": 2.0037583551120892e-05,
      "loss": 0.0,
      "step": 4218
    },
    {
      "epoch": 0.7096123118324783,
      "grad_norm": NaN,
      "learning_rate": 2.0016073393543473e-05,
      "loss": 0.0,
      "step": 4219
    },
    {
      "epoch": 0.7097805062652426,
      "grad_norm": NaN,
      "learning_rate": 1.99945718979866e-05,
      "loss": 0.0,
      "step": 4220
    },
    {
      "epoch": 0.7099487006980069,
      "grad_norm": NaN,
      "learning_rate": 1.9973079070661822e-05,
      "loss": 0.0,
      "step": 4221
    },
    {
      "epoch": 0.7101168951307711,
      "grad_norm": NaN,
      "learning_rate": 1.9951594917778187e-05,
      "loss": 0.0,
      "step": 4222
    },
    {
      "epoch": 0.7102850895635354,
      "grad_norm": NaN,
      "learning_rate": 1.9930119445542227e-05,
      "loss": 0.0,
      "step": 4223
    },
    {
      "epoch": 0.7104532839962997,
      "grad_norm": NaN,
      "learning_rate": 1.990865266015795e-05,
      "loss": 0.0,
      "step": 4224
    },
    {
      "epoch": 0.710621478429064,
      "grad_norm": NaN,
      "learning_rate": 1.9887194567826868e-05,
      "loss": 0.0,
      "step": 4225
    },
    {
      "epoch": 0.7107896728618283,
      "grad_norm": NaN,
      "learning_rate": 1.9865745174748e-05,
      "loss": 0.0,
      "step": 4226
    },
    {
      "epoch": 0.7109578672945925,
      "grad_norm": NaN,
      "learning_rate": 1.9844304487117836e-05,
      "loss": 0.0,
      "step": 4227
    },
    {
      "epoch": 0.7111260617273568,
      "grad_norm": NaN,
      "learning_rate": 1.9822872511130336e-05,
      "loss": 0.0,
      "step": 4228
    },
    {
      "epoch": 0.7112942561601211,
      "grad_norm": NaN,
      "learning_rate": 1.980144925297695e-05,
      "loss": 0.0,
      "step": 4229
    },
    {
      "epoch": 0.7114624505928854,
      "grad_norm": NaN,
      "learning_rate": 1.978003471884665e-05,
      "loss": 0.0,
      "step": 4230
    },
    {
      "epoch": 0.7116306450256497,
      "grad_norm": NaN,
      "learning_rate": 1.9758628914925837e-05,
      "loss": 0.0,
      "step": 4231
    },
    {
      "epoch": 0.7117988394584139,
      "grad_norm": NaN,
      "learning_rate": 1.9737231847398408e-05,
      "loss": 0.0,
      "step": 4232
    },
    {
      "epoch": 0.7119670338911782,
      "grad_norm": NaN,
      "learning_rate": 1.9715843522445737e-05,
      "loss": 0.0,
      "step": 4233
    },
    {
      "epoch": 0.7121352283239425,
      "grad_norm": NaN,
      "learning_rate": 1.9694463946246667e-05,
      "loss": 0.0,
      "step": 4234
    },
    {
      "epoch": 0.7123034227567068,
      "grad_norm": NaN,
      "learning_rate": 1.967309312497752e-05,
      "loss": 0.0,
      "step": 4235
    },
    {
      "epoch": 0.712471617189471,
      "grad_norm": NaN,
      "learning_rate": 1.965173106481209e-05,
      "loss": 0.0,
      "step": 4236
    },
    {
      "epoch": 0.7126398116222353,
      "grad_norm": NaN,
      "learning_rate": 1.963037777192162e-05,
      "loss": 0.0,
      "step": 4237
    },
    {
      "epoch": 0.7128080060549996,
      "grad_norm": NaN,
      "learning_rate": 1.960903325247487e-05,
      "loss": 0.0,
      "step": 4238
    },
    {
      "epoch": 0.7129762004877639,
      "grad_norm": NaN,
      "learning_rate": 1.9587697512638015e-05,
      "loss": 0.0,
      "step": 4239
    },
    {
      "epoch": 0.7131443949205282,
      "grad_norm": NaN,
      "learning_rate": 1.956637055857471e-05,
      "loss": 0.0,
      "step": 4240
    },
    {
      "epoch": 0.7133125893532924,
      "grad_norm": NaN,
      "learning_rate": 1.9545052396446055e-05,
      "loss": 0.0,
      "step": 4241
    },
    {
      "epoch": 0.7134807837860567,
      "grad_norm": NaN,
      "learning_rate": 1.9523743032410667e-05,
      "loss": 0.0,
      "step": 4242
    },
    {
      "epoch": 0.713648978218821,
      "grad_norm": NaN,
      "learning_rate": 1.9502442472624567e-05,
      "loss": 0.0,
      "step": 4243
    },
    {
      "epoch": 0.7138171726515853,
      "grad_norm": NaN,
      "learning_rate": 1.9481150723241236e-05,
      "loss": 0.0,
      "step": 4244
    },
    {
      "epoch": 0.7139853670843495,
      "grad_norm": NaN,
      "learning_rate": 1.9459867790411617e-05,
      "loss": 0.0,
      "step": 4245
    },
    {
      "epoch": 0.7141535615171137,
      "grad_norm": NaN,
      "learning_rate": 1.9438593680284145e-05,
      "loss": 0.0,
      "step": 4246
    },
    {
      "epoch": 0.714321755949878,
      "grad_norm": NaN,
      "learning_rate": 1.9417328399004646e-05,
      "loss": 0.0,
      "step": 4247
    },
    {
      "epoch": 0.7144899503826423,
      "grad_norm": NaN,
      "learning_rate": 1.939607195271643e-05,
      "loss": 0.0,
      "step": 4248
    },
    {
      "epoch": 0.7146581448154066,
      "grad_norm": NaN,
      "learning_rate": 1.9374824347560243e-05,
      "loss": 0.0,
      "step": 4249
    },
    {
      "epoch": 0.7148263392481709,
      "grad_norm": NaN,
      "learning_rate": 1.9353585589674285e-05,
      "loss": 0.0,
      "step": 4250
    },
    {
      "epoch": 0.7149945336809351,
      "grad_norm": NaN,
      "learning_rate": 1.9332355685194182e-05,
      "loss": 0.0,
      "step": 4251
    },
    {
      "epoch": 0.7151627281136994,
      "grad_norm": NaN,
      "learning_rate": 1.931113464025303e-05,
      "loss": 0.0,
      "step": 4252
    },
    {
      "epoch": 0.7153309225464637,
      "grad_norm": NaN,
      "learning_rate": 1.928992246098134e-05,
      "loss": 0.0,
      "step": 4253
    },
    {
      "epoch": 0.715499116979228,
      "grad_norm": NaN,
      "learning_rate": 1.926871915350707e-05,
      "loss": 0.0,
      "step": 4254
    },
    {
      "epoch": 0.7156673114119922,
      "grad_norm": NaN,
      "learning_rate": 1.9247524723955635e-05,
      "loss": 0.0,
      "step": 4255
    },
    {
      "epoch": 0.7158355058447565,
      "grad_norm": NaN,
      "learning_rate": 1.9226339178449865e-05,
      "loss": 0.0,
      "step": 4256
    },
    {
      "epoch": 0.7160037002775208,
      "grad_norm": NaN,
      "learning_rate": 1.920516252311002e-05,
      "loss": 0.0,
      "step": 4257
    },
    {
      "epoch": 0.7161718947102851,
      "grad_norm": NaN,
      "learning_rate": 1.918399476405378e-05,
      "loss": 0.0,
      "step": 4258
    },
    {
      "epoch": 0.7163400891430494,
      "grad_norm": NaN,
      "learning_rate": 1.9162835907396314e-05,
      "loss": 0.0,
      "step": 4259
    },
    {
      "epoch": 0.7165082835758136,
      "grad_norm": NaN,
      "learning_rate": 1.914168595925016e-05,
      "loss": 0.0,
      "step": 4260
    },
    {
      "epoch": 0.7166764780085779,
      "grad_norm": NaN,
      "learning_rate": 1.9120544925725298e-05,
      "loss": 0.0,
      "step": 4261
    },
    {
      "epoch": 0.7168446724413422,
      "grad_norm": NaN,
      "learning_rate": 1.909941281292912e-05,
      "loss": 0.0,
      "step": 4262
    },
    {
      "epoch": 0.7170128668741065,
      "grad_norm": NaN,
      "learning_rate": 1.907828962696649e-05,
      "loss": 0.0,
      "step": 4263
    },
    {
      "epoch": 0.7171810613068708,
      "grad_norm": NaN,
      "learning_rate": 1.9057175373939645e-05,
      "loss": 0.0,
      "step": 4264
    },
    {
      "epoch": 0.717349255739635,
      "grad_norm": NaN,
      "learning_rate": 1.9036070059948252e-05,
      "loss": 0.0,
      "step": 4265
    },
    {
      "epoch": 0.7175174501723993,
      "grad_norm": NaN,
      "learning_rate": 1.9014973691089387e-05,
      "loss": 0.0,
      "step": 4266
    },
    {
      "epoch": 0.7176856446051636,
      "grad_norm": NaN,
      "learning_rate": 1.899388627345759e-05,
      "loss": 0.0,
      "step": 4267
    },
    {
      "epoch": 0.7178538390379279,
      "grad_norm": NaN,
      "learning_rate": 1.897280781314476e-05,
      "loss": 0.0,
      "step": 4268
    },
    {
      "epoch": 0.7180220334706922,
      "grad_norm": NaN,
      "learning_rate": 1.8951738316240226e-05,
      "loss": 0.0,
      "step": 4269
    },
    {
      "epoch": 0.7181902279034564,
      "grad_norm": NaN,
      "learning_rate": 1.893067778883073e-05,
      "loss": 0.0,
      "step": 4270
    },
    {
      "epoch": 0.7183584223362207,
      "grad_norm": NaN,
      "learning_rate": 1.8909626237000432e-05,
      "loss": 0.0,
      "step": 4271
    },
    {
      "epoch": 0.718526616768985,
      "grad_norm": NaN,
      "learning_rate": 1.8888583666830876e-05,
      "loss": 0.0,
      "step": 4272
    },
    {
      "epoch": 0.7186948112017493,
      "grad_norm": NaN,
      "learning_rate": 1.886755008440103e-05,
      "loss": 0.0,
      "step": 4273
    },
    {
      "epoch": 0.7188630056345136,
      "grad_norm": NaN,
      "learning_rate": 1.8846525495787247e-05,
      "loss": 0.0,
      "step": 4274
    },
    {
      "epoch": 0.7190312000672777,
      "grad_norm": NaN,
      "learning_rate": 1.8825509907063327e-05,
      "loss": 0.0,
      "step": 4275
    },
    {
      "epoch": 0.719199394500042,
      "grad_norm": NaN,
      "learning_rate": 1.8804503324300422e-05,
      "loss": 0.0,
      "step": 4276
    },
    {
      "epoch": 0.7193675889328063,
      "grad_norm": NaN,
      "learning_rate": 1.8783505753567094e-05,
      "loss": 0.0,
      "step": 4277
    },
    {
      "epoch": 0.7195357833655706,
      "grad_norm": NaN,
      "learning_rate": 1.8762517200929292e-05,
      "loss": 0.0,
      "step": 4278
    },
    {
      "epoch": 0.7197039777983348,
      "grad_norm": NaN,
      "learning_rate": 1.8741537672450405e-05,
      "loss": 0.0,
      "step": 4279
    },
    {
      "epoch": 0.7198721722310991,
      "grad_norm": NaN,
      "learning_rate": 1.872056717419117e-05,
      "loss": 0.0,
      "step": 4280
    },
    {
      "epoch": 0.7200403666638634,
      "grad_norm": NaN,
      "learning_rate": 1.869960571220974e-05,
      "loss": 0.0,
      "step": 4281
    },
    {
      "epoch": 0.7202085610966277,
      "grad_norm": NaN,
      "learning_rate": 1.8678653292561605e-05,
      "loss": 0.0,
      "step": 4282
    },
    {
      "epoch": 0.720376755529392,
      "grad_norm": NaN,
      "learning_rate": 1.8657709921299736e-05,
      "loss": 0.0,
      "step": 4283
    },
    {
      "epoch": 0.7205449499621562,
      "grad_norm": NaN,
      "learning_rate": 1.8636775604474416e-05,
      "loss": 0.0,
      "step": 4284
    },
    {
      "epoch": 0.7207131443949205,
      "grad_norm": NaN,
      "learning_rate": 1.861585034813334e-05,
      "loss": 0.0,
      "step": 4285
    },
    {
      "epoch": 0.7208813388276848,
      "grad_norm": NaN,
      "learning_rate": 1.859493415832157e-05,
      "loss": 0.0,
      "step": 4286
    },
    {
      "epoch": 0.7210495332604491,
      "grad_norm": NaN,
      "learning_rate": 1.857402704108157e-05,
      "loss": 0.0,
      "step": 4287
    },
    {
      "epoch": 0.7212177276932134,
      "grad_norm": NaN,
      "learning_rate": 1.8553129002453163e-05,
      "loss": 0.0,
      "step": 4288
    },
    {
      "epoch": 0.7213859221259776,
      "grad_norm": NaN,
      "learning_rate": 1.8532240048473566e-05,
      "loss": 0.0,
      "step": 4289
    },
    {
      "epoch": 0.7215541165587419,
      "grad_norm": NaN,
      "learning_rate": 1.851136018517734e-05,
      "loss": 0.0,
      "step": 4290
    },
    {
      "epoch": 0.7217223109915062,
      "grad_norm": NaN,
      "learning_rate": 1.8490489418596478e-05,
      "loss": 0.0,
      "step": 4291
    },
    {
      "epoch": 0.7218905054242705,
      "grad_norm": NaN,
      "learning_rate": 1.84696277547603e-05,
      "loss": 0.0,
      "step": 4292
    },
    {
      "epoch": 0.7220586998570347,
      "grad_norm": NaN,
      "learning_rate": 1.8448775199695502e-05,
      "loss": 0.0,
      "step": 4293
    },
    {
      "epoch": 0.722226894289799,
      "grad_norm": NaN,
      "learning_rate": 1.8427931759426137e-05,
      "loss": 0.0,
      "step": 4294
    },
    {
      "epoch": 0.7223950887225633,
      "grad_norm": NaN,
      "learning_rate": 1.8407097439973674e-05,
      "loss": 0.0,
      "step": 4295
    },
    {
      "epoch": 0.7225632831553276,
      "grad_norm": NaN,
      "learning_rate": 1.8386272247356904e-05,
      "loss": 0.0,
      "step": 4296
    },
    {
      "epoch": 0.7227314775880919,
      "grad_norm": NaN,
      "learning_rate": 1.836545618759199e-05,
      "loss": 0.0,
      "step": 4297
    },
    {
      "epoch": 0.7228996720208561,
      "grad_norm": NaN,
      "learning_rate": 1.8344649266692455e-05,
      "loss": 0.0,
      "step": 4298
    },
    {
      "epoch": 0.7230678664536204,
      "grad_norm": NaN,
      "learning_rate": 1.8323851490669163e-05,
      "loss": 0.0,
      "step": 4299
    },
    {
      "epoch": 0.7232360608863847,
      "grad_norm": NaN,
      "learning_rate": 1.8303062865530406e-05,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 0.723404255319149,
      "grad_norm": NaN,
      "learning_rate": 1.828228339728176e-05,
      "loss": 0.0,
      "step": 4301
    },
    {
      "epoch": 0.7235724497519133,
      "grad_norm": NaN,
      "learning_rate": 1.826151309192618e-05,
      "loss": 0.0,
      "step": 4302
    },
    {
      "epoch": 0.7237406441846775,
      "grad_norm": NaN,
      "learning_rate": 1.8240751955463958e-05,
      "loss": 0.0,
      "step": 4303
    },
    {
      "epoch": 0.7239088386174417,
      "grad_norm": NaN,
      "learning_rate": 1.821999999389279e-05,
      "loss": 0.0,
      "step": 4304
    },
    {
      "epoch": 0.724077033050206,
      "grad_norm": NaN,
      "learning_rate": 1.8199257213207672e-05,
      "loss": 0.0,
      "step": 4305
    },
    {
      "epoch": 0.7242452274829703,
      "grad_norm": NaN,
      "learning_rate": 1.817852361940095e-05,
      "loss": 0.0,
      "step": 4306
    },
    {
      "epoch": 0.7244134219157345,
      "grad_norm": NaN,
      "learning_rate": 1.8157799218462335e-05,
      "loss": 0.0,
      "step": 4307
    },
    {
      "epoch": 0.7245816163484988,
      "grad_norm": NaN,
      "learning_rate": 1.8137084016378875e-05,
      "loss": 0.0,
      "step": 4308
    },
    {
      "epoch": 0.7247498107812631,
      "grad_norm": NaN,
      "learning_rate": 1.8116378019134957e-05,
      "loss": 0.0,
      "step": 4309
    },
    {
      "epoch": 0.7249180052140274,
      "grad_norm": NaN,
      "learning_rate": 1.809568123271231e-05,
      "loss": 0.0,
      "step": 4310
    },
    {
      "epoch": 0.7250861996467917,
      "grad_norm": NaN,
      "learning_rate": 1.8074993663089988e-05,
      "loss": 0.0,
      "step": 4311
    },
    {
      "epoch": 0.7252543940795559,
      "grad_norm": NaN,
      "learning_rate": 1.8054315316244442e-05,
      "loss": 0.0,
      "step": 4312
    },
    {
      "epoch": 0.7254225885123202,
      "grad_norm": NaN,
      "learning_rate": 1.8033646198149377e-05,
      "loss": 0.0,
      "step": 4313
    },
    {
      "epoch": 0.7255907829450845,
      "grad_norm": NaN,
      "learning_rate": 1.801298631477589e-05,
      "loss": 0.0,
      "step": 4314
    },
    {
      "epoch": 0.7257589773778488,
      "grad_norm": NaN,
      "learning_rate": 1.7992335672092362e-05,
      "loss": 0.0,
      "step": 4315
    },
    {
      "epoch": 0.7259271718106131,
      "grad_norm": NaN,
      "learning_rate": 1.7971694276064564e-05,
      "loss": 0.0,
      "step": 4316
    },
    {
      "epoch": 0.7260953662433773,
      "grad_norm": NaN,
      "learning_rate": 1.7951062132655562e-05,
      "loss": 0.0,
      "step": 4317
    },
    {
      "epoch": 0.7262635606761416,
      "grad_norm": NaN,
      "learning_rate": 1.7930439247825737e-05,
      "loss": 0.0,
      "step": 4318
    },
    {
      "epoch": 0.7264317551089059,
      "grad_norm": NaN,
      "learning_rate": 1.7909825627532796e-05,
      "loss": 0.0,
      "step": 4319
    },
    {
      "epoch": 0.7265999495416702,
      "grad_norm": NaN,
      "learning_rate": 1.7889221277731816e-05,
      "loss": 0.0,
      "step": 4320
    },
    {
      "epoch": 0.7267681439744345,
      "grad_norm": NaN,
      "learning_rate": 1.786862620437515e-05,
      "loss": 0.0,
      "step": 4321
    },
    {
      "epoch": 0.7269363384071987,
      "grad_norm": NaN,
      "learning_rate": 1.784804041341248e-05,
      "loss": 0.0,
      "step": 4322
    },
    {
      "epoch": 0.727104532839963,
      "grad_norm": NaN,
      "learning_rate": 1.782746391079081e-05,
      "loss": 0.0,
      "step": 4323
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": NaN,
      "learning_rate": 1.780689670245447e-05,
      "loss": 0.0,
      "step": 4324
    },
    {
      "epoch": 0.7274409217054916,
      "grad_norm": NaN,
      "learning_rate": 1.7786338794345082e-05,
      "loss": 0.0,
      "step": 4325
    },
    {
      "epoch": 0.7276091161382559,
      "grad_norm": NaN,
      "learning_rate": 1.7765790192401604e-05,
      "loss": 0.0,
      "step": 4326
    },
    {
      "epoch": 0.7277773105710201,
      "grad_norm": NaN,
      "learning_rate": 1.7745250902560278e-05,
      "loss": 0.0,
      "step": 4327
    },
    {
      "epoch": 0.7279455050037844,
      "grad_norm": NaN,
      "learning_rate": 1.772472093075471e-05,
      "loss": 0.0,
      "step": 4328
    },
    {
      "epoch": 0.7281136994365487,
      "grad_norm": NaN,
      "learning_rate": 1.7704200282915766e-05,
      "loss": 0.0,
      "step": 4329
    },
    {
      "epoch": 0.728281893869313,
      "grad_norm": NaN,
      "learning_rate": 1.7683688964971625e-05,
      "loss": 0.0,
      "step": 4330
    },
    {
      "epoch": 0.7284500883020772,
      "grad_norm": NaN,
      "learning_rate": 1.7663186982847757e-05,
      "loss": 0.0,
      "step": 4331
    },
    {
      "epoch": 0.7286182827348415,
      "grad_norm": NaN,
      "learning_rate": 1.7642694342466998e-05,
      "loss": 0.0,
      "step": 4332
    },
    {
      "epoch": 0.7287864771676058,
      "grad_norm": NaN,
      "learning_rate": 1.7622211049749422e-05,
      "loss": 0.0,
      "step": 4333
    },
    {
      "epoch": 0.72895467160037,
      "grad_norm": NaN,
      "learning_rate": 1.7601737110612416e-05,
      "loss": 0.0,
      "step": 4334
    },
    {
      "epoch": 0.7291228660331343,
      "grad_norm": NaN,
      "learning_rate": 1.7581272530970667e-05,
      "loss": 0.0,
      "step": 4335
    },
    {
      "epoch": 0.7292910604658985,
      "grad_norm": NaN,
      "learning_rate": 1.7560817316736183e-05,
      "loss": 0.0,
      "step": 4336
    },
    {
      "epoch": 0.7294592548986628,
      "grad_norm": NaN,
      "learning_rate": 1.754037147381824e-05,
      "loss": 0.0,
      "step": 4337
    },
    {
      "epoch": 0.7296274493314271,
      "grad_norm": NaN,
      "learning_rate": 1.7519935008123412e-05,
      "loss": 0.0,
      "step": 4338
    },
    {
      "epoch": 0.7297956437641914,
      "grad_norm": NaN,
      "learning_rate": 1.749950792555554e-05,
      "loss": 0.0,
      "step": 4339
    },
    {
      "epoch": 0.7299638381969556,
      "grad_norm": NaN,
      "learning_rate": 1.7479090232015822e-05,
      "loss": 0.0,
      "step": 4340
    },
    {
      "epoch": 0.7301320326297199,
      "grad_norm": NaN,
      "learning_rate": 1.7458681933402682e-05,
      "loss": 0.0,
      "step": 4341
    },
    {
      "epoch": 0.7303002270624842,
      "grad_norm": NaN,
      "learning_rate": 1.7438283035611846e-05,
      "loss": 0.0,
      "step": 4342
    },
    {
      "epoch": 0.7304684214952485,
      "grad_norm": NaN,
      "learning_rate": 1.741789354453633e-05,
      "loss": 0.0,
      "step": 4343
    },
    {
      "epoch": 0.7306366159280128,
      "grad_norm": NaN,
      "learning_rate": 1.7397513466066428e-05,
      "loss": 0.0,
      "step": 4344
    },
    {
      "epoch": 0.730804810360777,
      "grad_norm": NaN,
      "learning_rate": 1.7377142806089717e-05,
      "loss": 0.0,
      "step": 4345
    },
    {
      "epoch": 0.7309730047935413,
      "grad_norm": NaN,
      "learning_rate": 1.7356781570491053e-05,
      "loss": 0.0,
      "step": 4346
    },
    {
      "epoch": 0.7311411992263056,
      "grad_norm": NaN,
      "learning_rate": 1.7336429765152568e-05,
      "loss": 0.0,
      "step": 4347
    },
    {
      "epoch": 0.7313093936590699,
      "grad_norm": NaN,
      "learning_rate": 1.7316087395953655e-05,
      "loss": 0.0,
      "step": 4348
    },
    {
      "epoch": 0.7314775880918342,
      "grad_norm": NaN,
      "learning_rate": 1.7295754468771024e-05,
      "loss": 0.0,
      "step": 4349
    },
    {
      "epoch": 0.7316457825245984,
      "grad_norm": NaN,
      "learning_rate": 1.7275430989478626e-05,
      "loss": 0.0,
      "step": 4350
    },
    {
      "epoch": 0.7318139769573627,
      "grad_norm": NaN,
      "learning_rate": 1.7255116963947676e-05,
      "loss": 0.0,
      "step": 4351
    },
    {
      "epoch": 0.731982171390127,
      "grad_norm": NaN,
      "learning_rate": 1.7234812398046653e-05,
      "loss": 0.0,
      "step": 4352
    },
    {
      "epoch": 0.7321503658228913,
      "grad_norm": NaN,
      "learning_rate": 1.7214517297641353e-05,
      "loss": 0.0,
      "step": 4353
    },
    {
      "epoch": 0.7323185602556556,
      "grad_norm": NaN,
      "learning_rate": 1.719423166859479e-05,
      "loss": 0.0,
      "step": 4354
    },
    {
      "epoch": 0.7324867546884198,
      "grad_norm": NaN,
      "learning_rate": 1.7173955516767255e-05,
      "loss": 0.0,
      "step": 4355
    },
    {
      "epoch": 0.7326549491211841,
      "grad_norm": NaN,
      "learning_rate": 1.7153688848016277e-05,
      "loss": 0.0,
      "step": 4356
    },
    {
      "epoch": 0.7328231435539484,
      "grad_norm": NaN,
      "learning_rate": 1.713343166819671e-05,
      "loss": 0.0,
      "step": 4357
    },
    {
      "epoch": 0.7329913379867127,
      "grad_norm": NaN,
      "learning_rate": 1.7113183983160602e-05,
      "loss": 0.0,
      "step": 4358
    },
    {
      "epoch": 0.733159532419477,
      "grad_norm": NaN,
      "learning_rate": 1.7092945798757294e-05,
      "loss": 0.0,
      "step": 4359
    },
    {
      "epoch": 0.7333277268522412,
      "grad_norm": NaN,
      "learning_rate": 1.707271712083335e-05,
      "loss": 0.0,
      "step": 4360
    },
    {
      "epoch": 0.7334959212850055,
      "grad_norm": NaN,
      "learning_rate": 1.7052497955232633e-05,
      "loss": 0.0,
      "step": 4361
    },
    {
      "epoch": 0.7336641157177698,
      "grad_norm": NaN,
      "learning_rate": 1.7032288307796225e-05,
      "loss": 0.0,
      "step": 4362
    },
    {
      "epoch": 0.733832310150534,
      "grad_norm": NaN,
      "learning_rate": 1.7012088184362467e-05,
      "loss": 0.0,
      "step": 4363
    },
    {
      "epoch": 0.7340005045832982,
      "grad_norm": NaN,
      "learning_rate": 1.699189759076695e-05,
      "loss": 0.0,
      "step": 4364
    },
    {
      "epoch": 0.7341686990160625,
      "grad_norm": NaN,
      "learning_rate": 1.6971716532842504e-05,
      "loss": 0.0,
      "step": 4365
    },
    {
      "epoch": 0.7343368934488268,
      "grad_norm": NaN,
      "learning_rate": 1.6951545016419207e-05,
      "loss": 0.0,
      "step": 4366
    },
    {
      "epoch": 0.7345050878815911,
      "grad_norm": NaN,
      "learning_rate": 1.69313830473244e-05,
      "loss": 0.0,
      "step": 4367
    },
    {
      "epoch": 0.7346732823143554,
      "grad_norm": NaN,
      "learning_rate": 1.6911230631382618e-05,
      "loss": 0.0,
      "step": 4368
    },
    {
      "epoch": 0.7348414767471196,
      "grad_norm": NaN,
      "learning_rate": 1.6891087774415714e-05,
      "loss": 0.0,
      "step": 4369
    },
    {
      "epoch": 0.7350096711798839,
      "grad_norm": NaN,
      "learning_rate": 1.6870954482242707e-05,
      "loss": 0.0,
      "step": 4370
    },
    {
      "epoch": 0.7351778656126482,
      "grad_norm": NaN,
      "learning_rate": 1.6850830760679887e-05,
      "loss": 0.0,
      "step": 4371
    },
    {
      "epoch": 0.7353460600454125,
      "grad_norm": NaN,
      "learning_rate": 1.683071661554075e-05,
      "loss": 0.0,
      "step": 4372
    },
    {
      "epoch": 0.7355142544781768,
      "grad_norm": NaN,
      "learning_rate": 1.6810612052636078e-05,
      "loss": 0.0,
      "step": 4373
    },
    {
      "epoch": 0.735682448910941,
      "grad_norm": NaN,
      "learning_rate": 1.6790517077773845e-05,
      "loss": 0.0,
      "step": 4374
    },
    {
      "epoch": 0.7358506433437053,
      "grad_norm": NaN,
      "learning_rate": 1.677043169675926e-05,
      "loss": 0.0,
      "step": 4375
    },
    {
      "epoch": 0.7360188377764696,
      "grad_norm": NaN,
      "learning_rate": 1.675035591539475e-05,
      "loss": 0.0,
      "step": 4376
    },
    {
      "epoch": 0.7361870322092339,
      "grad_norm": NaN,
      "learning_rate": 1.6730289739480015e-05,
      "loss": 0.0,
      "step": 4377
    },
    {
      "epoch": 0.7363552266419982,
      "grad_norm": NaN,
      "learning_rate": 1.671023317481193e-05,
      "loss": 0.0,
      "step": 4378
    },
    {
      "epoch": 0.7365234210747624,
      "grad_norm": NaN,
      "learning_rate": 1.6690186227184618e-05,
      "loss": 0.0,
      "step": 4379
    },
    {
      "epoch": 0.7366916155075267,
      "grad_norm": NaN,
      "learning_rate": 1.667014890238941e-05,
      "loss": 0.0,
      "step": 4380
    },
    {
      "epoch": 0.736859809940291,
      "grad_norm": NaN,
      "learning_rate": 1.665012120621487e-05,
      "loss": 0.0,
      "step": 4381
    },
    {
      "epoch": 0.7370280043730553,
      "grad_norm": NaN,
      "learning_rate": 1.6630103144446775e-05,
      "loss": 0.0,
      "step": 4382
    },
    {
      "epoch": 0.7371961988058195,
      "grad_norm": NaN,
      "learning_rate": 1.6610094722868114e-05,
      "loss": 0.0,
      "step": 4383
    },
    {
      "epoch": 0.7373643932385838,
      "grad_norm": NaN,
      "learning_rate": 1.659009594725908e-05,
      "loss": 0.0,
      "step": 4384
    },
    {
      "epoch": 0.7375325876713481,
      "grad_norm": NaN,
      "learning_rate": 1.6570106823397142e-05,
      "loss": 0.0,
      "step": 4385
    },
    {
      "epoch": 0.7377007821041124,
      "grad_norm": NaN,
      "learning_rate": 1.6550127357056893e-05,
      "loss": 0.0,
      "step": 4386
    },
    {
      "epoch": 0.7378689765368767,
      "grad_norm": NaN,
      "learning_rate": 1.65301575540102e-05,
      "loss": 0.0,
      "step": 4387
    },
    {
      "epoch": 0.7380371709696409,
      "grad_norm": NaN,
      "learning_rate": 1.6510197420026086e-05,
      "loss": 0.0,
      "step": 4388
    },
    {
      "epoch": 0.7382053654024052,
      "grad_norm": NaN,
      "learning_rate": 1.6490246960870847e-05,
      "loss": 0.0,
      "step": 4389
    },
    {
      "epoch": 0.7383735598351695,
      "grad_norm": NaN,
      "learning_rate": 1.647030618230793e-05,
      "loss": 0.0,
      "step": 4390
    },
    {
      "epoch": 0.7385417542679338,
      "grad_norm": NaN,
      "learning_rate": 1.6450375090098003e-05,
      "loss": 0.0,
      "step": 4391
    },
    {
      "epoch": 0.7387099487006981,
      "grad_norm": NaN,
      "learning_rate": 1.643045368999892e-05,
      "loss": 0.0,
      "step": 4392
    },
    {
      "epoch": 0.7388781431334622,
      "grad_norm": NaN,
      "learning_rate": 1.641054198776578e-05,
      "loss": 0.0,
      "step": 4393
    },
    {
      "epoch": 0.7390463375662265,
      "grad_norm": NaN,
      "learning_rate": 1.6390639989150834e-05,
      "loss": 0.0,
      "step": 4394
    },
    {
      "epoch": 0.7392145319989908,
      "grad_norm": NaN,
      "learning_rate": 1.6370747699903556e-05,
      "loss": 0.0,
      "step": 4395
    },
    {
      "epoch": 0.7393827264317551,
      "grad_norm": NaN,
      "learning_rate": 1.6350865125770576e-05,
      "loss": 0.0,
      "step": 4396
    },
    {
      "epoch": 0.7395509208645193,
      "grad_norm": NaN,
      "learning_rate": 1.6330992272495786e-05,
      "loss": 0.0,
      "step": 4397
    },
    {
      "epoch": 0.7397191152972836,
      "grad_norm": NaN,
      "learning_rate": 1.6311129145820218e-05,
      "loss": 0.0,
      "step": 4398
    },
    {
      "epoch": 0.7398873097300479,
      "grad_norm": NaN,
      "learning_rate": 1.62912757514821e-05,
      "loss": 0.0,
      "step": 4399
    },
    {
      "epoch": 0.7400555041628122,
      "grad_norm": NaN,
      "learning_rate": 1.627143209521686e-05,
      "loss": 0.0,
      "step": 4400
    },
    {
      "epoch": 0.7402236985955765,
      "grad_norm": NaN,
      "learning_rate": 1.62515981827571e-05,
      "loss": 0.0,
      "step": 4401
    },
    {
      "epoch": 0.7403918930283407,
      "grad_norm": NaN,
      "learning_rate": 1.623177401983263e-05,
      "loss": 0.0,
      "step": 4402
    },
    {
      "epoch": 0.740560087461105,
      "grad_norm": NaN,
      "learning_rate": 1.6211959612170414e-05,
      "loss": 0.0,
      "step": 4403
    },
    {
      "epoch": 0.7407282818938693,
      "grad_norm": NaN,
      "learning_rate": 1.6192154965494616e-05,
      "loss": 0.0,
      "step": 4404
    },
    {
      "epoch": 0.7408964763266336,
      "grad_norm": NaN,
      "learning_rate": 1.6172360085526565e-05,
      "loss": 0.0,
      "step": 4405
    },
    {
      "epoch": 0.7410646707593979,
      "grad_norm": NaN,
      "learning_rate": 1.6152574977984814e-05,
      "loss": 0.0,
      "step": 4406
    },
    {
      "epoch": 0.7412328651921621,
      "grad_norm": NaN,
      "learning_rate": 1.613279964858504e-05,
      "loss": 0.0,
      "step": 4407
    },
    {
      "epoch": 0.7414010596249264,
      "grad_norm": NaN,
      "learning_rate": 1.6113034103040107e-05,
      "loss": 0.0,
      "step": 4408
    },
    {
      "epoch": 0.7415692540576907,
      "grad_norm": NaN,
      "learning_rate": 1.6093278347060055e-05,
      "loss": 0.0,
      "step": 4409
    },
    {
      "epoch": 0.741737448490455,
      "grad_norm": NaN,
      "learning_rate": 1.607353238635212e-05,
      "loss": 0.0,
      "step": 4410
    },
    {
      "epoch": 0.7419056429232193,
      "grad_norm": NaN,
      "learning_rate": 1.6053796226620688e-05,
      "loss": 0.0,
      "step": 4411
    },
    {
      "epoch": 0.7420738373559835,
      "grad_norm": NaN,
      "learning_rate": 1.6034069873567303e-05,
      "loss": 0.0,
      "step": 4412
    },
    {
      "epoch": 0.7422420317887478,
      "grad_norm": NaN,
      "learning_rate": 1.6014353332890676e-05,
      "loss": 0.0,
      "step": 4413
    },
    {
      "epoch": 0.7424102262215121,
      "grad_norm": NaN,
      "learning_rate": 1.599464661028673e-05,
      "loss": 0.0,
      "step": 4414
    },
    {
      "epoch": 0.7425784206542764,
      "grad_norm": NaN,
      "learning_rate": 1.5974949711448488e-05,
      "loss": 0.0,
      "step": 4415
    },
    {
      "epoch": 0.7427466150870407,
      "grad_norm": NaN,
      "learning_rate": 1.5955262642066175e-05,
      "loss": 0.0,
      "step": 4416
    },
    {
      "epoch": 0.7429148095198049,
      "grad_norm": NaN,
      "learning_rate": 1.593558540782717e-05,
      "loss": 0.0,
      "step": 4417
    },
    {
      "epoch": 0.7430830039525692,
      "grad_norm": NaN,
      "learning_rate": 1.5915918014415986e-05,
      "loss": 0.0,
      "step": 4418
    },
    {
      "epoch": 0.7432511983853335,
      "grad_norm": NaN,
      "learning_rate": 1.5896260467514336e-05,
      "loss": 0.0,
      "step": 4419
    },
    {
      "epoch": 0.7434193928180978,
      "grad_norm": NaN,
      "learning_rate": 1.5876612772801042e-05,
      "loss": 0.0,
      "step": 4420
    },
    {
      "epoch": 0.743587587250862,
      "grad_norm": NaN,
      "learning_rate": 1.58569749359521e-05,
      "loss": 0.0,
      "step": 4421
    },
    {
      "epoch": 0.7437557816836263,
      "grad_norm": NaN,
      "learning_rate": 1.5837346962640697e-05,
      "loss": 0.0,
      "step": 4422
    },
    {
      "epoch": 0.7439239761163905,
      "grad_norm": NaN,
      "learning_rate": 1.5817728858537107e-05,
      "loss": 0.0,
      "step": 4423
    },
    {
      "epoch": 0.7440921705491548,
      "grad_norm": NaN,
      "learning_rate": 1.5798120629308788e-05,
      "loss": 0.0,
      "step": 4424
    },
    {
      "epoch": 0.744260364981919,
      "grad_norm": NaN,
      "learning_rate": 1.5778522280620323e-05,
      "loss": 0.0,
      "step": 4425
    },
    {
      "epoch": 0.7444285594146833,
      "grad_norm": NaN,
      "learning_rate": 1.575893381813348e-05,
      "loss": 0.0,
      "step": 4426
    },
    {
      "epoch": 0.7445967538474476,
      "grad_norm": NaN,
      "learning_rate": 1.573935524750714e-05,
      "loss": 0.0,
      "step": 4427
    },
    {
      "epoch": 0.7447649482802119,
      "grad_norm": NaN,
      "learning_rate": 1.5719786574397328e-05,
      "loss": 0.0,
      "step": 4428
    },
    {
      "epoch": 0.7449331427129762,
      "grad_norm": NaN,
      "learning_rate": 1.570022780445719e-05,
      "loss": 0.0,
      "step": 4429
    },
    {
      "epoch": 0.7451013371457404,
      "grad_norm": NaN,
      "learning_rate": 1.5680678943337084e-05,
      "loss": 0.0,
      "step": 4430
    },
    {
      "epoch": 0.7452695315785047,
      "grad_norm": NaN,
      "learning_rate": 1.5661139996684433e-05,
      "loss": 0.0,
      "step": 4431
    },
    {
      "epoch": 0.745437726011269,
      "grad_norm": NaN,
      "learning_rate": 1.5641610970143816e-05,
      "loss": 0.0,
      "step": 4432
    },
    {
      "epoch": 0.7456059204440333,
      "grad_norm": NaN,
      "learning_rate": 1.5622091869356937e-05,
      "loss": 0.0,
      "step": 4433
    },
    {
      "epoch": 0.7457741148767976,
      "grad_norm": NaN,
      "learning_rate": 1.560258269996268e-05,
      "loss": 0.0,
      "step": 4434
    },
    {
      "epoch": 0.7459423093095618,
      "grad_norm": NaN,
      "learning_rate": 1.558308346759701e-05,
      "loss": 0.0,
      "step": 4435
    },
    {
      "epoch": 0.7461105037423261,
      "grad_norm": NaN,
      "learning_rate": 1.556359417789303e-05,
      "loss": 0.0,
      "step": 4436
    },
    {
      "epoch": 0.7462786981750904,
      "grad_norm": NaN,
      "learning_rate": 1.5544114836480984e-05,
      "loss": 0.0,
      "step": 4437
    },
    {
      "epoch": 0.7464468926078547,
      "grad_norm": NaN,
      "learning_rate": 1.5524645448988235e-05,
      "loss": 0.0,
      "step": 4438
    },
    {
      "epoch": 0.746615087040619,
      "grad_norm": NaN,
      "learning_rate": 1.550518602103927e-05,
      "loss": 0.0,
      "step": 4439
    },
    {
      "epoch": 0.7467832814733832,
      "grad_norm": NaN,
      "learning_rate": 1.5485736558255697e-05,
      "loss": 0.0,
      "step": 4440
    },
    {
      "epoch": 0.7469514759061475,
      "grad_norm": NaN,
      "learning_rate": 1.5466297066256235e-05,
      "loss": 0.0,
      "step": 4441
    },
    {
      "epoch": 0.7471196703389118,
      "grad_norm": NaN,
      "learning_rate": 1.544686755065677e-05,
      "loss": 0.0,
      "step": 4442
    },
    {
      "epoch": 0.7472878647716761,
      "grad_norm": NaN,
      "learning_rate": 1.542744801707025e-05,
      "loss": 0.0,
      "step": 4443
    },
    {
      "epoch": 0.7474560592044404,
      "grad_norm": NaN,
      "learning_rate": 1.5408038471106762e-05,
      "loss": 0.0,
      "step": 4444
    },
    {
      "epoch": 0.7476242536372046,
      "grad_norm": NaN,
      "learning_rate": 1.5388638918373495e-05,
      "loss": 0.0,
      "step": 4445
    },
    {
      "epoch": 0.7477924480699689,
      "grad_norm": NaN,
      "learning_rate": 1.5369249364474787e-05,
      "loss": 0.0,
      "step": 4446
    },
    {
      "epoch": 0.7479606425027332,
      "grad_norm": NaN,
      "learning_rate": 1.534986981501205e-05,
      "loss": 0.0,
      "step": 4447
    },
    {
      "epoch": 0.7481288369354975,
      "grad_norm": NaN,
      "learning_rate": 1.5330500275583825e-05,
      "loss": 0.0,
      "step": 4448
    },
    {
      "epoch": 0.7482970313682618,
      "grad_norm": NaN,
      "learning_rate": 1.5311140751785745e-05,
      "loss": 0.0,
      "step": 4449
    },
    {
      "epoch": 0.748465225801026,
      "grad_norm": NaN,
      "learning_rate": 1.529179124921055e-05,
      "loss": 0.0,
      "step": 4450
    },
    {
      "epoch": 0.7486334202337903,
      "grad_norm": NaN,
      "learning_rate": 1.527245177344813e-05,
      "loss": 0.0,
      "step": 4451
    },
    {
      "epoch": 0.7488016146665545,
      "grad_norm": NaN,
      "learning_rate": 1.5253122330085417e-05,
      "loss": 0.0,
      "step": 4452
    },
    {
      "epoch": 0.7489698090993188,
      "grad_norm": NaN,
      "learning_rate": 1.5233802924706475e-05,
      "loss": 0.0,
      "step": 4453
    },
    {
      "epoch": 0.749138003532083,
      "grad_norm": NaN,
      "learning_rate": 1.521449356289245e-05,
      "loss": 0.0,
      "step": 4454
    },
    {
      "epoch": 0.7493061979648473,
      "grad_norm": NaN,
      "learning_rate": 1.5195194250221629e-05,
      "loss": 0.0,
      "step": 4455
    },
    {
      "epoch": 0.7494743923976116,
      "grad_norm": NaN,
      "learning_rate": 1.5175904992269352e-05,
      "loss": 0.0,
      "step": 4456
    },
    {
      "epoch": 0.7496425868303759,
      "grad_norm": NaN,
      "learning_rate": 1.5156625794608076e-05,
      "loss": 0.0,
      "step": 4457
    },
    {
      "epoch": 0.7498107812631402,
      "grad_norm": NaN,
      "learning_rate": 1.5137356662807333e-05,
      "loss": 0.0,
      "step": 4458
    },
    {
      "epoch": 0.7499789756959044,
      "grad_norm": NaN,
      "learning_rate": 1.5118097602433773e-05,
      "loss": 0.0,
      "step": 4459
    },
    {
      "epoch": 0.7501471701286687,
      "grad_norm": NaN,
      "learning_rate": 1.5098848619051109e-05,
      "loss": 0.0,
      "step": 4460
    },
    {
      "epoch": 0.750315364561433,
      "grad_norm": NaN,
      "learning_rate": 1.5079609718220166e-05,
      "loss": 0.0,
      "step": 4461
    },
    {
      "epoch": 0.7504835589941973,
      "grad_norm": NaN,
      "learning_rate": 1.5060380905498833e-05,
      "loss": 0.0,
      "step": 4462
    },
    {
      "epoch": 0.7506517534269616,
      "grad_norm": NaN,
      "learning_rate": 1.504116218644212e-05,
      "loss": 0.0,
      "step": 4463
    },
    {
      "epoch": 0.7508199478597258,
      "grad_norm": NaN,
      "learning_rate": 1.5021953566602093e-05,
      "loss": 0.0,
      "step": 4464
    },
    {
      "epoch": 0.7509881422924901,
      "grad_norm": NaN,
      "learning_rate": 1.5002755051527906e-05,
      "loss": 0.0,
      "step": 4465
    },
    {
      "epoch": 0.7511563367252544,
      "grad_norm": NaN,
      "learning_rate": 1.4983566646765773e-05,
      "loss": 0.0,
      "step": 4466
    },
    {
      "epoch": 0.7513245311580187,
      "grad_norm": NaN,
      "learning_rate": 1.4964388357859044e-05,
      "loss": 0.0,
      "step": 4467
    },
    {
      "epoch": 0.751492725590783,
      "grad_norm": NaN,
      "learning_rate": 1.4945220190348102e-05,
      "loss": 0.0,
      "step": 4468
    },
    {
      "epoch": 0.7516609200235472,
      "grad_norm": NaN,
      "learning_rate": 1.492606214977041e-05,
      "loss": 0.0,
      "step": 4469
    },
    {
      "epoch": 0.7518291144563115,
      "grad_norm": NaN,
      "learning_rate": 1.4906914241660492e-05,
      "loss": 0.0,
      "step": 4470
    },
    {
      "epoch": 0.7519973088890758,
      "grad_norm": NaN,
      "learning_rate": 1.4887776471550002e-05,
      "loss": 0.0,
      "step": 4471
    },
    {
      "epoch": 0.7521655033218401,
      "grad_norm": NaN,
      "learning_rate": 1.4868648844967615e-05,
      "loss": 0.0,
      "step": 4472
    },
    {
      "epoch": 0.7523336977546043,
      "grad_norm": NaN,
      "learning_rate": 1.484953136743908e-05,
      "loss": 0.0,
      "step": 4473
    },
    {
      "epoch": 0.7525018921873686,
      "grad_norm": NaN,
      "learning_rate": 1.4830424044487223e-05,
      "loss": 0.0,
      "step": 4474
    },
    {
      "epoch": 0.7526700866201329,
      "grad_norm": NaN,
      "learning_rate": 1.4811326881631937e-05,
      "loss": 0.0,
      "step": 4475
    },
    {
      "epoch": 0.7528382810528972,
      "grad_norm": NaN,
      "learning_rate": 1.4792239884390186e-05,
      "loss": 0.0,
      "step": 4476
    },
    {
      "epoch": 0.7530064754856615,
      "grad_norm": NaN,
      "learning_rate": 1.477316305827598e-05,
      "loss": 0.0,
      "step": 4477
    },
    {
      "epoch": 0.7531746699184257,
      "grad_norm": NaN,
      "learning_rate": 1.4754096408800383e-05,
      "loss": 0.0,
      "step": 4478
    },
    {
      "epoch": 0.75334286435119,
      "grad_norm": NaN,
      "learning_rate": 1.4735039941471573e-05,
      "loss": 0.0,
      "step": 4479
    },
    {
      "epoch": 0.7535110587839543,
      "grad_norm": NaN,
      "learning_rate": 1.4715993661794736e-05,
      "loss": 0.0,
      "step": 4480
    },
    {
      "epoch": 0.7536792532167186,
      "grad_norm": NaN,
      "learning_rate": 1.469695757527213e-05,
      "loss": 0.0,
      "step": 4481
    },
    {
      "epoch": 0.7538474476494827,
      "grad_norm": NaN,
      "learning_rate": 1.4677931687403046e-05,
      "loss": 0.0,
      "step": 4482
    },
    {
      "epoch": 0.754015642082247,
      "grad_norm": NaN,
      "learning_rate": 1.4658916003683882e-05,
      "loss": 0.0,
      "step": 4483
    },
    {
      "epoch": 0.7541838365150113,
      "grad_norm": NaN,
      "learning_rate": 1.4639910529608042e-05,
      "loss": 0.0,
      "step": 4484
    },
    {
      "epoch": 0.7543520309477756,
      "grad_norm": NaN,
      "learning_rate": 1.4620915270665992e-05,
      "loss": 0.0,
      "step": 4485
    },
    {
      "epoch": 0.7545202253805399,
      "grad_norm": NaN,
      "learning_rate": 1.460193023234524e-05,
      "loss": 0.0,
      "step": 4486
    },
    {
      "epoch": 0.7546884198133041,
      "grad_norm": NaN,
      "learning_rate": 1.4582955420130372e-05,
      "loss": 0.0,
      "step": 4487
    },
    {
      "epoch": 0.7548566142460684,
      "grad_norm": NaN,
      "learning_rate": 1.4563990839502995e-05,
      "loss": 0.0,
      "step": 4488
    },
    {
      "epoch": 0.7550248086788327,
      "grad_norm": NaN,
      "learning_rate": 1.454503649594176e-05,
      "loss": 0.0,
      "step": 4489
    },
    {
      "epoch": 0.755193003111597,
      "grad_norm": NaN,
      "learning_rate": 1.4526092394922347e-05,
      "loss": 0.0,
      "step": 4490
    },
    {
      "epoch": 0.7553611975443613,
      "grad_norm": NaN,
      "learning_rate": 1.450715854191752e-05,
      "loss": 0.0,
      "step": 4491
    },
    {
      "epoch": 0.7555293919771255,
      "grad_norm": NaN,
      "learning_rate": 1.4488234942397056e-05,
      "loss": 0.0,
      "step": 4492
    },
    {
      "epoch": 0.7556975864098898,
      "grad_norm": NaN,
      "learning_rate": 1.4469321601827757e-05,
      "loss": 0.0,
      "step": 4493
    },
    {
      "epoch": 0.7558657808426541,
      "grad_norm": NaN,
      "learning_rate": 1.4450418525673481e-05,
      "loss": 0.0,
      "step": 4494
    },
    {
      "epoch": 0.7560339752754184,
      "grad_norm": NaN,
      "learning_rate": 1.4431525719395117e-05,
      "loss": 0.0,
      "step": 4495
    },
    {
      "epoch": 0.7562021697081827,
      "grad_norm": NaN,
      "learning_rate": 1.4412643188450581e-05,
      "loss": 0.0,
      "step": 4496
    },
    {
      "epoch": 0.7563703641409469,
      "grad_norm": NaN,
      "learning_rate": 1.4393770938294825e-05,
      "loss": 0.0,
      "step": 4497
    },
    {
      "epoch": 0.7565385585737112,
      "grad_norm": NaN,
      "learning_rate": 1.4374908974379837e-05,
      "loss": 0.0,
      "step": 4498
    },
    {
      "epoch": 0.7567067530064755,
      "grad_norm": NaN,
      "learning_rate": 1.4356057302154608e-05,
      "loss": 0.0,
      "step": 4499
    },
    {
      "epoch": 0.7568749474392398,
      "grad_norm": NaN,
      "learning_rate": 1.43372159270652e-05,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 0.757043141872004,
      "grad_norm": NaN,
      "learning_rate": 1.4318384854554667e-05,
      "loss": 0.0,
      "step": 4501
    },
    {
      "epoch": 0.7572113363047683,
      "grad_norm": NaN,
      "learning_rate": 1.4299564090063088e-05,
      "loss": 0.0,
      "step": 4502
    },
    {
      "epoch": 0.7573795307375326,
      "grad_norm": NaN,
      "learning_rate": 1.4280753639027566e-05,
      "loss": 0.0,
      "step": 4503
    },
    {
      "epoch": 0.7575477251702969,
      "grad_norm": NaN,
      "learning_rate": 1.4261953506882252e-05,
      "loss": 0.0,
      "step": 4504
    },
    {
      "epoch": 0.7577159196030612,
      "grad_norm": NaN,
      "learning_rate": 1.4243163699058282e-05,
      "loss": 0.0,
      "step": 4505
    },
    {
      "epoch": 0.7578841140358255,
      "grad_norm": NaN,
      "learning_rate": 1.4224384220983817e-05,
      "loss": 0.0,
      "step": 4506
    },
    {
      "epoch": 0.7580523084685897,
      "grad_norm": NaN,
      "learning_rate": 1.4205615078084028e-05,
      "loss": 0.0,
      "step": 4507
    },
    {
      "epoch": 0.758220502901354,
      "grad_norm": NaN,
      "learning_rate": 1.418685627578114e-05,
      "loss": 0.0,
      "step": 4508
    },
    {
      "epoch": 0.7583886973341183,
      "grad_norm": NaN,
      "learning_rate": 1.4168107819494342e-05,
      "loss": 0.0,
      "step": 4509
    },
    {
      "epoch": 0.7585568917668826,
      "grad_norm": NaN,
      "learning_rate": 1.4149369714639853e-05,
      "loss": 0.0,
      "step": 4510
    },
    {
      "epoch": 0.7587250861996467,
      "grad_norm": NaN,
      "learning_rate": 1.4130641966630903e-05,
      "loss": 0.0,
      "step": 4511
    },
    {
      "epoch": 0.758893280632411,
      "grad_norm": NaN,
      "learning_rate": 1.4111924580877733e-05,
      "loss": 0.0,
      "step": 4512
    },
    {
      "epoch": 0.7590614750651753,
      "grad_norm": NaN,
      "learning_rate": 1.4093217562787575e-05,
      "loss": 0.0,
      "step": 4513
    },
    {
      "epoch": 0.7592296694979396,
      "grad_norm": NaN,
      "learning_rate": 1.4074520917764688e-05,
      "loss": 0.0,
      "step": 4514
    },
    {
      "epoch": 0.7593978639307039,
      "grad_norm": NaN,
      "learning_rate": 1.4055834651210309e-05,
      "loss": 0.0,
      "step": 4515
    },
    {
      "epoch": 0.7595660583634681,
      "grad_norm": NaN,
      "learning_rate": 1.4037158768522712e-05,
      "loss": 0.0,
      "step": 4516
    },
    {
      "epoch": 0.7597342527962324,
      "grad_norm": NaN,
      "learning_rate": 1.4018493275097139e-05,
      "loss": 0.0,
      "step": 4517
    },
    {
      "epoch": 0.7599024472289967,
      "grad_norm": NaN,
      "learning_rate": 1.3999838176325847e-05,
      "loss": 0.0,
      "step": 4518
    },
    {
      "epoch": 0.760070641661761,
      "grad_norm": NaN,
      "learning_rate": 1.3981193477598059e-05,
      "loss": 0.0,
      "step": 4519
    },
    {
      "epoch": 0.7602388360945252,
      "grad_norm": NaN,
      "learning_rate": 1.3962559184300062e-05,
      "loss": 0.0,
      "step": 4520
    },
    {
      "epoch": 0.7604070305272895,
      "grad_norm": NaN,
      "learning_rate": 1.3943935301815076e-05,
      "loss": 0.0,
      "step": 4521
    },
    {
      "epoch": 0.7605752249600538,
      "grad_norm": NaN,
      "learning_rate": 1.3925321835523326e-05,
      "loss": 0.0,
      "step": 4522
    },
    {
      "epoch": 0.7607434193928181,
      "grad_norm": NaN,
      "learning_rate": 1.3906718790802026e-05,
      "loss": 0.0,
      "step": 4523
    },
    {
      "epoch": 0.7609116138255824,
      "grad_norm": NaN,
      "learning_rate": 1.3888126173025412e-05,
      "loss": 0.0,
      "step": 4524
    },
    {
      "epoch": 0.7610798082583466,
      "grad_norm": NaN,
      "learning_rate": 1.386954398756467e-05,
      "loss": 0.0,
      "step": 4525
    },
    {
      "epoch": 0.7612480026911109,
      "grad_norm": NaN,
      "learning_rate": 1.3850972239787985e-05,
      "loss": 0.0,
      "step": 4526
    },
    {
      "epoch": 0.7614161971238752,
      "grad_norm": NaN,
      "learning_rate": 1.3832410935060513e-05,
      "loss": 0.0,
      "step": 4527
    },
    {
      "epoch": 0.7615843915566395,
      "grad_norm": NaN,
      "learning_rate": 1.3813860078744439e-05,
      "loss": 0.0,
      "step": 4528
    },
    {
      "epoch": 0.7617525859894038,
      "grad_norm": NaN,
      "learning_rate": 1.379531967619888e-05,
      "loss": 0.0,
      "step": 4529
    },
    {
      "epoch": 0.761920780422168,
      "grad_norm": NaN,
      "learning_rate": 1.3776789732779954e-05,
      "loss": 0.0,
      "step": 4530
    },
    {
      "epoch": 0.7620889748549323,
      "grad_norm": NaN,
      "learning_rate": 1.3758270253840744e-05,
      "loss": 0.0,
      "step": 4531
    },
    {
      "epoch": 0.7622571692876966,
      "grad_norm": NaN,
      "learning_rate": 1.373976124473133e-05,
      "loss": 0.0,
      "step": 4532
    },
    {
      "epoch": 0.7624253637204609,
      "grad_norm": NaN,
      "learning_rate": 1.3721262710798754e-05,
      "loss": 0.0,
      "step": 4533
    },
    {
      "epoch": 0.7625935581532252,
      "grad_norm": NaN,
      "learning_rate": 1.370277465738704e-05,
      "loss": 0.0,
      "step": 4534
    },
    {
      "epoch": 0.7627617525859894,
      "grad_norm": NaN,
      "learning_rate": 1.3684297089837155e-05,
      "loss": 0.0,
      "step": 4535
    },
    {
      "epoch": 0.7629299470187537,
      "grad_norm": NaN,
      "learning_rate": 1.3665830013487102e-05,
      "loss": 0.0,
      "step": 4536
    },
    {
      "epoch": 0.763098141451518,
      "grad_norm": NaN,
      "learning_rate": 1.3647373433671784e-05,
      "loss": 0.0,
      "step": 4537
    },
    {
      "epoch": 0.7632663358842823,
      "grad_norm": NaN,
      "learning_rate": 1.3628927355723114e-05,
      "loss": 0.0,
      "step": 4538
    },
    {
      "epoch": 0.7634345303170466,
      "grad_norm": NaN,
      "learning_rate": 1.3610491784969937e-05,
      "loss": 0.0,
      "step": 4539
    },
    {
      "epoch": 0.7636027247498108,
      "grad_norm": NaN,
      "learning_rate": 1.3592066726738113e-05,
      "loss": 0.0,
      "step": 4540
    },
    {
      "epoch": 0.763770919182575,
      "grad_norm": NaN,
      "learning_rate": 1.357365218635041e-05,
      "loss": 0.0,
      "step": 4541
    },
    {
      "epoch": 0.7639391136153393,
      "grad_norm": NaN,
      "learning_rate": 1.35552481691266e-05,
      "loss": 0.0,
      "step": 4542
    },
    {
      "epoch": 0.7641073080481036,
      "grad_norm": NaN,
      "learning_rate": 1.3536854680383375e-05,
      "loss": 0.0,
      "step": 4543
    },
    {
      "epoch": 0.7642755024808678,
      "grad_norm": NaN,
      "learning_rate": 1.3518471725434434e-05,
      "loss": 0.0,
      "step": 4544
    },
    {
      "epoch": 0.7644436969136321,
      "grad_norm": NaN,
      "learning_rate": 1.3500099309590397e-05,
      "loss": 0.0,
      "step": 4545
    },
    {
      "epoch": 0.7646118913463964,
      "grad_norm": NaN,
      "learning_rate": 1.3481737438158847e-05,
      "loss": 0.0,
      "step": 4546
    },
    {
      "epoch": 0.7647800857791607,
      "grad_norm": NaN,
      "learning_rate": 1.3463386116444327e-05,
      "loss": 0.0,
      "step": 4547
    },
    {
      "epoch": 0.764948280211925,
      "grad_norm": NaN,
      "learning_rate": 1.3445045349748304e-05,
      "loss": 0.0,
      "step": 4548
    },
    {
      "epoch": 0.7651164746446892,
      "grad_norm": NaN,
      "learning_rate": 1.3426715143369256e-05,
      "loss": 0.0,
      "step": 4549
    },
    {
      "epoch": 0.7652846690774535,
      "grad_norm": NaN,
      "learning_rate": 1.340839550260256e-05,
      "loss": 0.0,
      "step": 4550
    },
    {
      "epoch": 0.7654528635102178,
      "grad_norm": NaN,
      "learning_rate": 1.3390086432740551e-05,
      "loss": 0.0,
      "step": 4551
    },
    {
      "epoch": 0.7656210579429821,
      "grad_norm": NaN,
      "learning_rate": 1.3371787939072522e-05,
      "loss": 0.0,
      "step": 4552
    },
    {
      "epoch": 0.7657892523757464,
      "grad_norm": NaN,
      "learning_rate": 1.335350002688469e-05,
      "loss": 0.0,
      "step": 4553
    },
    {
      "epoch": 0.7659574468085106,
      "grad_norm": NaN,
      "learning_rate": 1.3335222701460243e-05,
      "loss": 0.0,
      "step": 4554
    },
    {
      "epoch": 0.7661256412412749,
      "grad_norm": NaN,
      "learning_rate": 1.3316955968079287e-05,
      "loss": 0.0,
      "step": 4555
    },
    {
      "epoch": 0.7662938356740392,
      "grad_norm": NaN,
      "learning_rate": 1.3298699832018862e-05,
      "loss": 0.0,
      "step": 4556
    },
    {
      "epoch": 0.7664620301068035,
      "grad_norm": NaN,
      "learning_rate": 1.3280454298552997e-05,
      "loss": 0.0,
      "step": 4557
    },
    {
      "epoch": 0.7666302245395678,
      "grad_norm": NaN,
      "learning_rate": 1.32622193729526e-05,
      "loss": 0.0,
      "step": 4558
    },
    {
      "epoch": 0.766798418972332,
      "grad_norm": NaN,
      "learning_rate": 1.3243995060485537e-05,
      "loss": 0.0,
      "step": 4559
    },
    {
      "epoch": 0.7669666134050963,
      "grad_norm": NaN,
      "learning_rate": 1.3225781366416595e-05,
      "loss": 0.0,
      "step": 4560
    },
    {
      "epoch": 0.7671348078378606,
      "grad_norm": NaN,
      "learning_rate": 1.3207578296007539e-05,
      "loss": 0.0,
      "step": 4561
    },
    {
      "epoch": 0.7673030022706249,
      "grad_norm": NaN,
      "learning_rate": 1.3189385854517012e-05,
      "loss": 0.0,
      "step": 4562
    },
    {
      "epoch": 0.7674711967033891,
      "grad_norm": NaN,
      "learning_rate": 1.3171204047200614e-05,
      "loss": 0.0,
      "step": 4563
    },
    {
      "epoch": 0.7676393911361534,
      "grad_norm": NaN,
      "learning_rate": 1.3153032879310839e-05,
      "loss": 0.0,
      "step": 4564
    },
    {
      "epoch": 0.7678075855689177,
      "grad_norm": NaN,
      "learning_rate": 1.3134872356097173e-05,
      "loss": 0.0,
      "step": 4565
    },
    {
      "epoch": 0.767975780001682,
      "grad_norm": NaN,
      "learning_rate": 1.311672248280597e-05,
      "loss": 0.0,
      "step": 4566
    },
    {
      "epoch": 0.7681439744344463,
      "grad_norm": NaN,
      "learning_rate": 1.309858326468053e-05,
      "loss": 0.0,
      "step": 4567
    },
    {
      "epoch": 0.7683121688672105,
      "grad_norm": NaN,
      "learning_rate": 1.3080454706961059e-05,
      "loss": 0.0,
      "step": 4568
    },
    {
      "epoch": 0.7684803632999748,
      "grad_norm": NaN,
      "learning_rate": 1.3062336814884702e-05,
      "loss": 0.0,
      "step": 4569
    },
    {
      "epoch": 0.7686485577327391,
      "grad_norm": NaN,
      "learning_rate": 1.3044229593685514e-05,
      "loss": 0.0,
      "step": 4570
    },
    {
      "epoch": 0.7688167521655033,
      "grad_norm": NaN,
      "learning_rate": 1.3026133048594469e-05,
      "loss": 0.0,
      "step": 4571
    },
    {
      "epoch": 0.7689849465982675,
      "grad_norm": NaN,
      "learning_rate": 1.3008047184839445e-05,
      "loss": 0.0,
      "step": 4572
    },
    {
      "epoch": 0.7691531410310318,
      "grad_norm": NaN,
      "learning_rate": 1.2989972007645263e-05,
      "loss": 0.0,
      "step": 4573
    },
    {
      "epoch": 0.7693213354637961,
      "grad_norm": NaN,
      "learning_rate": 1.2971907522233633e-05,
      "loss": 0.0,
      "step": 4574
    },
    {
      "epoch": 0.7694895298965604,
      "grad_norm": NaN,
      "learning_rate": 1.2953853733823185e-05,
      "loss": 0.0,
      "step": 4575
    },
    {
      "epoch": 0.7696577243293247,
      "grad_norm": NaN,
      "learning_rate": 1.2935810647629438e-05,
      "loss": 0.0,
      "step": 4576
    },
    {
      "epoch": 0.769825918762089,
      "grad_norm": NaN,
      "learning_rate": 1.2917778268864861e-05,
      "loss": 0.0,
      "step": 4577
    },
    {
      "epoch": 0.7699941131948532,
      "grad_norm": NaN,
      "learning_rate": 1.2899756602738805e-05,
      "loss": 0.0,
      "step": 4578
    },
    {
      "epoch": 0.7701623076276175,
      "grad_norm": NaN,
      "learning_rate": 1.2881745654457512e-05,
      "loss": 0.0,
      "step": 4579
    },
    {
      "epoch": 0.7703305020603818,
      "grad_norm": NaN,
      "learning_rate": 1.2863745429224144e-05,
      "loss": 0.0,
      "step": 4580
    },
    {
      "epoch": 0.7704986964931461,
      "grad_norm": NaN,
      "learning_rate": 1.2845755932238778e-05,
      "loss": 0.0,
      "step": 4581
    },
    {
      "epoch": 0.7706668909259103,
      "grad_norm": NaN,
      "learning_rate": 1.282777716869838e-05,
      "loss": 0.0,
      "step": 4582
    },
    {
      "epoch": 0.7708350853586746,
      "grad_norm": NaN,
      "learning_rate": 1.2809809143796803e-05,
      "loss": 0.0,
      "step": 4583
    },
    {
      "epoch": 0.7710032797914389,
      "grad_norm": NaN,
      "learning_rate": 1.2791851862724791e-05,
      "loss": 0.0,
      "step": 4584
    },
    {
      "epoch": 0.7711714742242032,
      "grad_norm": NaN,
      "learning_rate": 1.2773905330670038e-05,
      "loss": 0.0,
      "step": 4585
    },
    {
      "epoch": 0.7713396686569675,
      "grad_norm": NaN,
      "learning_rate": 1.2755969552817083e-05,
      "loss": 0.0,
      "step": 4586
    },
    {
      "epoch": 0.7715078630897317,
      "grad_norm": NaN,
      "learning_rate": 1.2738044534347365e-05,
      "loss": 0.0,
      "step": 4587
    },
    {
      "epoch": 0.771676057522496,
      "grad_norm": NaN,
      "learning_rate": 1.2720130280439218e-05,
      "loss": 0.0,
      "step": 4588
    },
    {
      "epoch": 0.7718442519552603,
      "grad_norm": NaN,
      "learning_rate": 1.270222679626788e-05,
      "loss": 0.0,
      "step": 4589
    },
    {
      "epoch": 0.7720124463880246,
      "grad_norm": NaN,
      "learning_rate": 1.2684334087005462e-05,
      "loss": 0.0,
      "step": 4590
    },
    {
      "epoch": 0.7721806408207889,
      "grad_norm": NaN,
      "learning_rate": 1.2666452157820963e-05,
      "loss": 0.0,
      "step": 4591
    },
    {
      "epoch": 0.7723488352535531,
      "grad_norm": NaN,
      "learning_rate": 1.2648581013880267e-05,
      "loss": 0.0,
      "step": 4592
    },
    {
      "epoch": 0.7725170296863174,
      "grad_norm": NaN,
      "learning_rate": 1.2630720660346163e-05,
      "loss": 0.0,
      "step": 4593
    },
    {
      "epoch": 0.7726852241190817,
      "grad_norm": NaN,
      "learning_rate": 1.2612871102378304e-05,
      "loss": 0.0,
      "step": 4594
    },
    {
      "epoch": 0.772853418551846,
      "grad_norm": NaN,
      "learning_rate": 1.2595032345133229e-05,
      "loss": 0.0,
      "step": 4595
    },
    {
      "epoch": 0.7730216129846103,
      "grad_norm": NaN,
      "learning_rate": 1.2577204393764331e-05,
      "loss": 0.0,
      "step": 4596
    },
    {
      "epoch": 0.7731898074173745,
      "grad_norm": NaN,
      "learning_rate": 1.2559387253421934e-05,
      "loss": 0.0,
      "step": 4597
    },
    {
      "epoch": 0.7733580018501388,
      "grad_norm": NaN,
      "learning_rate": 1.2541580929253204e-05,
      "loss": 0.0,
      "step": 4598
    },
    {
      "epoch": 0.7735261962829031,
      "grad_norm": NaN,
      "learning_rate": 1.252378542640219e-05,
      "loss": 0.0,
      "step": 4599
    },
    {
      "epoch": 0.7736943907156673,
      "grad_norm": NaN,
      "learning_rate": 1.2506000750009806e-05,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 0.7738625851484315,
      "grad_norm": NaN,
      "learning_rate": 1.2488226905213829e-05,
      "loss": 0.0,
      "step": 4601
    },
    {
      "epoch": 0.7740307795811958,
      "grad_norm": NaN,
      "learning_rate": 1.247046389714896e-05,
      "loss": 0.0,
      "step": 4602
    },
    {
      "epoch": 0.7741989740139601,
      "grad_norm": NaN,
      "learning_rate": 1.245271173094672e-05,
      "loss": 0.0,
      "step": 4603
    },
    {
      "epoch": 0.7743671684467244,
      "grad_norm": NaN,
      "learning_rate": 1.2434970411735502e-05,
      "loss": 0.0,
      "step": 4604
    },
    {
      "epoch": 0.7745353628794887,
      "grad_norm": NaN,
      "learning_rate": 1.2417239944640586e-05,
      "loss": 0.0,
      "step": 4605
    },
    {
      "epoch": 0.7747035573122529,
      "grad_norm": NaN,
      "learning_rate": 1.2399520334784092e-05,
      "loss": 0.0,
      "step": 4606
    },
    {
      "epoch": 0.7748717517450172,
      "grad_norm": NaN,
      "learning_rate": 1.2381811587285031e-05,
      "loss": 0.0,
      "step": 4607
    },
    {
      "epoch": 0.7750399461777815,
      "grad_norm": NaN,
      "learning_rate": 1.2364113707259251e-05,
      "loss": 0.0,
      "step": 4608
    },
    {
      "epoch": 0.7752081406105458,
      "grad_norm": NaN,
      "learning_rate": 1.2346426699819458e-05,
      "loss": 0.0,
      "step": 4609
    },
    {
      "epoch": 0.77537633504331,
      "grad_norm": NaN,
      "learning_rate": 1.2328750570075265e-05,
      "loss": 0.0,
      "step": 4610
    },
    {
      "epoch": 0.7755445294760743,
      "grad_norm": NaN,
      "learning_rate": 1.2311085323133087e-05,
      "loss": 0.0,
      "step": 4611
    },
    {
      "epoch": 0.7757127239088386,
      "grad_norm": NaN,
      "learning_rate": 1.229343096409622e-05,
      "loss": 0.0,
      "step": 4612
    },
    {
      "epoch": 0.7758809183416029,
      "grad_norm": NaN,
      "learning_rate": 1.2275787498064795e-05,
      "loss": 0.0,
      "step": 4613
    },
    {
      "epoch": 0.7760491127743672,
      "grad_norm": NaN,
      "learning_rate": 1.2258154930135846e-05,
      "loss": 0.0,
      "step": 4614
    },
    {
      "epoch": 0.7762173072071314,
      "grad_norm": NaN,
      "learning_rate": 1.2240533265403198e-05,
      "loss": 0.0,
      "step": 4615
    },
    {
      "epoch": 0.7763855016398957,
      "grad_norm": NaN,
      "learning_rate": 1.2222922508957563e-05,
      "loss": 0.0,
      "step": 4616
    },
    {
      "epoch": 0.77655369607266,
      "grad_norm": NaN,
      "learning_rate": 1.2205322665886471e-05,
      "loss": 0.0,
      "step": 4617
    },
    {
      "epoch": 0.7767218905054243,
      "grad_norm": NaN,
      "learning_rate": 1.2187733741274354e-05,
      "loss": 0.0,
      "step": 4618
    },
    {
      "epoch": 0.7768900849381886,
      "grad_norm": NaN,
      "learning_rate": 1.2170155740202432e-05,
      "loss": 0.0,
      "step": 4619
    },
    {
      "epoch": 0.7770582793709528,
      "grad_norm": NaN,
      "learning_rate": 1.2152588667748805e-05,
      "loss": 0.0,
      "step": 4620
    },
    {
      "epoch": 0.7772264738037171,
      "grad_norm": NaN,
      "learning_rate": 1.2135032528988377e-05,
      "loss": 0.0,
      "step": 4621
    },
    {
      "epoch": 0.7773946682364814,
      "grad_norm": NaN,
      "learning_rate": 1.2117487328992955e-05,
      "loss": 0.0,
      "step": 4622
    },
    {
      "epoch": 0.7775628626692457,
      "grad_norm": NaN,
      "learning_rate": 1.2099953072831138e-05,
      "loss": 0.0,
      "step": 4623
    },
    {
      "epoch": 0.77773105710201,
      "grad_norm": NaN,
      "learning_rate": 1.208242976556837e-05,
      "loss": 0.0,
      "step": 4624
    },
    {
      "epoch": 0.7778992515347742,
      "grad_norm": NaN,
      "learning_rate": 1.2064917412266946e-05,
      "loss": 0.0,
      "step": 4625
    },
    {
      "epoch": 0.7780674459675385,
      "grad_norm": NaN,
      "learning_rate": 1.204741601798598e-05,
      "loss": 0.0,
      "step": 4626
    },
    {
      "epoch": 0.7782356404003028,
      "grad_norm": NaN,
      "learning_rate": 1.2029925587781432e-05,
      "loss": 0.0,
      "step": 4627
    },
    {
      "epoch": 0.7784038348330671,
      "grad_norm": NaN,
      "learning_rate": 1.2012446126706094e-05,
      "loss": 0.0,
      "step": 4628
    },
    {
      "epoch": 0.7785720292658314,
      "grad_norm": NaN,
      "learning_rate": 1.1994977639809574e-05,
      "loss": 0.0,
      "step": 4629
    },
    {
      "epoch": 0.7787402236985955,
      "grad_norm": NaN,
      "learning_rate": 1.1977520132138342e-05,
      "loss": 0.0,
      "step": 4630
    },
    {
      "epoch": 0.7789084181313598,
      "grad_norm": NaN,
      "learning_rate": 1.1960073608735673e-05,
      "loss": 0.0,
      "step": 4631
    },
    {
      "epoch": 0.7790766125641241,
      "grad_norm": NaN,
      "learning_rate": 1.1942638074641665e-05,
      "loss": 0.0,
      "step": 4632
    },
    {
      "epoch": 0.7792448069968884,
      "grad_norm": NaN,
      "learning_rate": 1.1925213534893232e-05,
      "loss": 0.0,
      "step": 4633
    },
    {
      "epoch": 0.7794130014296526,
      "grad_norm": NaN,
      "learning_rate": 1.1907799994524166e-05,
      "loss": 0.0,
      "step": 4634
    },
    {
      "epoch": 0.7795811958624169,
      "grad_norm": NaN,
      "learning_rate": 1.1890397458565022e-05,
      "loss": 0.0,
      "step": 4635
    },
    {
      "epoch": 0.7797493902951812,
      "grad_norm": NaN,
      "learning_rate": 1.1873005932043202e-05,
      "loss": 0.0,
      "step": 4636
    },
    {
      "epoch": 0.7799175847279455,
      "grad_norm": NaN,
      "learning_rate": 1.1855625419982907e-05,
      "loss": 0.0,
      "step": 4637
    },
    {
      "epoch": 0.7800857791607098,
      "grad_norm": NaN,
      "learning_rate": 1.18382559274052e-05,
      "loss": 0.0,
      "step": 4638
    },
    {
      "epoch": 0.780253973593474,
      "grad_norm": NaN,
      "learning_rate": 1.1820897459327918e-05,
      "loss": 0.0,
      "step": 4639
    },
    {
      "epoch": 0.7804221680262383,
      "grad_norm": NaN,
      "learning_rate": 1.1803550020765736e-05,
      "loss": 0.0,
      "step": 4640
    },
    {
      "epoch": 0.7805903624590026,
      "grad_norm": NaN,
      "learning_rate": 1.1786213616730108e-05,
      "loss": 0.0,
      "step": 4641
    },
    {
      "epoch": 0.7807585568917669,
      "grad_norm": NaN,
      "learning_rate": 1.176888825222936e-05,
      "loss": 0.0,
      "step": 4642
    },
    {
      "epoch": 0.7809267513245312,
      "grad_norm": NaN,
      "learning_rate": 1.175157393226859e-05,
      "loss": 0.0,
      "step": 4643
    },
    {
      "epoch": 0.7810949457572954,
      "grad_norm": NaN,
      "learning_rate": 1.1734270661849695e-05,
      "loss": 0.0,
      "step": 4644
    },
    {
      "epoch": 0.7812631401900597,
      "grad_norm": NaN,
      "learning_rate": 1.171697844597141e-05,
      "loss": 0.0,
      "step": 4645
    },
    {
      "epoch": 0.781431334622824,
      "grad_norm": NaN,
      "learning_rate": 1.1699697289629258e-05,
      "loss": 0.0,
      "step": 4646
    },
    {
      "epoch": 0.7815995290555883,
      "grad_norm": NaN,
      "learning_rate": 1.168242719781557e-05,
      "loss": 0.0,
      "step": 4647
    },
    {
      "epoch": 0.7817677234883526,
      "grad_norm": NaN,
      "learning_rate": 1.1665168175519486e-05,
      "loss": 0.0,
      "step": 4648
    },
    {
      "epoch": 0.7819359179211168,
      "grad_norm": NaN,
      "learning_rate": 1.1647920227726938e-05,
      "loss": 0.0,
      "step": 4649
    },
    {
      "epoch": 0.7821041123538811,
      "grad_norm": NaN,
      "learning_rate": 1.1630683359420652e-05,
      "loss": 0.0,
      "step": 4650
    },
    {
      "epoch": 0.7822723067866454,
      "grad_norm": NaN,
      "learning_rate": 1.16134575755802e-05,
      "loss": 0.0,
      "step": 4651
    },
    {
      "epoch": 0.7824405012194097,
      "grad_norm": NaN,
      "learning_rate": 1.1596242881181901e-05,
      "loss": 0.0,
      "step": 4652
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": NaN,
      "learning_rate": 1.157903928119889e-05,
      "loss": 0.0,
      "step": 4653
    },
    {
      "epoch": 0.7827768900849382,
      "grad_norm": NaN,
      "learning_rate": 1.156184678060107e-05,
      "loss": 0.0,
      "step": 4654
    },
    {
      "epoch": 0.7829450845177025,
      "grad_norm": NaN,
      "learning_rate": 1.1544665384355202e-05,
      "loss": 0.0,
      "step": 4655
    },
    {
      "epoch": 0.7831132789504668,
      "grad_norm": NaN,
      "learning_rate": 1.1527495097424778e-05,
      "loss": 0.0,
      "step": 4656
    },
    {
      "epoch": 0.7832814733832311,
      "grad_norm": NaN,
      "learning_rate": 1.1510335924770105e-05,
      "loss": 0.0,
      "step": 4657
    },
    {
      "epoch": 0.7834496678159953,
      "grad_norm": NaN,
      "learning_rate": 1.149318787134826e-05,
      "loss": 0.0,
      "step": 4658
    },
    {
      "epoch": 0.7836178622487595,
      "grad_norm": NaN,
      "learning_rate": 1.1476050942113148e-05,
      "loss": 0.0,
      "step": 4659
    },
    {
      "epoch": 0.7837860566815238,
      "grad_norm": NaN,
      "learning_rate": 1.1458925142015431e-05,
      "loss": 0.0,
      "step": 4660
    },
    {
      "epoch": 0.7839542511142881,
      "grad_norm": NaN,
      "learning_rate": 1.1441810476002552e-05,
      "loss": 0.0,
      "step": 4661
    },
    {
      "epoch": 0.7841224455470523,
      "grad_norm": NaN,
      "learning_rate": 1.1424706949018754e-05,
      "loss": 0.0,
      "step": 4662
    },
    {
      "epoch": 0.7842906399798166,
      "grad_norm": NaN,
      "learning_rate": 1.1407614566005048e-05,
      "loss": 0.0,
      "step": 4663
    },
    {
      "epoch": 0.7844588344125809,
      "grad_norm": NaN,
      "learning_rate": 1.1390533331899234e-05,
      "loss": 0.0,
      "step": 4664
    },
    {
      "epoch": 0.7846270288453452,
      "grad_norm": NaN,
      "learning_rate": 1.1373463251635892e-05,
      "loss": 0.0,
      "step": 4665
    },
    {
      "epoch": 0.7847952232781095,
      "grad_norm": NaN,
      "learning_rate": 1.135640433014636e-05,
      "loss": 0.0,
      "step": 4666
    },
    {
      "epoch": 0.7849634177108737,
      "grad_norm": NaN,
      "learning_rate": 1.1339356572358795e-05,
      "loss": 0.0,
      "step": 4667
    },
    {
      "epoch": 0.785131612143638,
      "grad_norm": NaN,
      "learning_rate": 1.1322319983198093e-05,
      "loss": 0.0,
      "step": 4668
    },
    {
      "epoch": 0.7852998065764023,
      "grad_norm": NaN,
      "learning_rate": 1.1305294567585933e-05,
      "loss": 0.0,
      "step": 4669
    },
    {
      "epoch": 0.7854680010091666,
      "grad_norm": NaN,
      "learning_rate": 1.1288280330440753e-05,
      "loss": 0.0,
      "step": 4670
    },
    {
      "epoch": 0.7856361954419309,
      "grad_norm": NaN,
      "learning_rate": 1.1271277276677805e-05,
      "loss": 0.0,
      "step": 4671
    },
    {
      "epoch": 0.7858043898746951,
      "grad_norm": NaN,
      "learning_rate": 1.1254285411209065e-05,
      "loss": 0.0,
      "step": 4672
    },
    {
      "epoch": 0.7859725843074594,
      "grad_norm": NaN,
      "learning_rate": 1.1237304738943294e-05,
      "loss": 0.0,
      "step": 4673
    },
    {
      "epoch": 0.7861407787402237,
      "grad_norm": NaN,
      "learning_rate": 1.1220335264785997e-05,
      "loss": 0.0,
      "step": 4674
    },
    {
      "epoch": 0.786308973172988,
      "grad_norm": NaN,
      "learning_rate": 1.1203376993639508e-05,
      "loss": 0.0,
      "step": 4675
    },
    {
      "epoch": 0.7864771676057523,
      "grad_norm": NaN,
      "learning_rate": 1.1186429930402852e-05,
      "loss": 0.0,
      "step": 4676
    },
    {
      "epoch": 0.7866453620385165,
      "grad_norm": NaN,
      "learning_rate": 1.1169494079971854e-05,
      "loss": 0.0,
      "step": 4677
    },
    {
      "epoch": 0.7868135564712808,
      "grad_norm": NaN,
      "learning_rate": 1.1152569447239075e-05,
      "loss": 0.0,
      "step": 4678
    },
    {
      "epoch": 0.7869817509040451,
      "grad_norm": NaN,
      "learning_rate": 1.1135656037093877e-05,
      "loss": 0.0,
      "step": 4679
    },
    {
      "epoch": 0.7871499453368094,
      "grad_norm": NaN,
      "learning_rate": 1.1118753854422353e-05,
      "loss": 0.0,
      "step": 4680
    },
    {
      "epoch": 0.7873181397695737,
      "grad_norm": NaN,
      "learning_rate": 1.110186290410734e-05,
      "loss": 0.0,
      "step": 4681
    },
    {
      "epoch": 0.7874863342023379,
      "grad_norm": NaN,
      "learning_rate": 1.1084983191028452e-05,
      "loss": 0.0,
      "step": 4682
    },
    {
      "epoch": 0.7876545286351022,
      "grad_norm": NaN,
      "learning_rate": 1.1068114720062044e-05,
      "loss": 0.0,
      "step": 4683
    },
    {
      "epoch": 0.7878227230678665,
      "grad_norm": NaN,
      "learning_rate": 1.1051257496081235e-05,
      "loss": 0.0,
      "step": 4684
    },
    {
      "epoch": 0.7879909175006308,
      "grad_norm": NaN,
      "learning_rate": 1.103441152395588e-05,
      "loss": 0.0,
      "step": 4685
    },
    {
      "epoch": 0.788159111933395,
      "grad_norm": NaN,
      "learning_rate": 1.1017576808552581e-05,
      "loss": 0.0,
      "step": 4686
    },
    {
      "epoch": 0.7883273063661593,
      "grad_norm": NaN,
      "learning_rate": 1.1000753354734733e-05,
      "loss": 0.0,
      "step": 4687
    },
    {
      "epoch": 0.7884955007989236,
      "grad_norm": NaN,
      "learning_rate": 1.0983941167362422e-05,
      "loss": 0.0,
      "step": 4688
    },
    {
      "epoch": 0.7886636952316878,
      "grad_norm": NaN,
      "learning_rate": 1.0967140251292507e-05,
      "loss": 0.0,
      "step": 4689
    },
    {
      "epoch": 0.7888318896644521,
      "grad_norm": NaN,
      "learning_rate": 1.0950350611378562e-05,
      "loss": 0.0,
      "step": 4690
    },
    {
      "epoch": 0.7890000840972163,
      "grad_norm": NaN,
      "learning_rate": 1.0933572252470958e-05,
      "loss": 0.0,
      "step": 4691
    },
    {
      "epoch": 0.7891682785299806,
      "grad_norm": NaN,
      "learning_rate": 1.0916805179416761e-05,
      "loss": 0.0,
      "step": 4692
    },
    {
      "epoch": 0.7893364729627449,
      "grad_norm": NaN,
      "learning_rate": 1.0900049397059792e-05,
      "loss": 0.0,
      "step": 4693
    },
    {
      "epoch": 0.7895046673955092,
      "grad_norm": NaN,
      "learning_rate": 1.0883304910240594e-05,
      "loss": 0.0,
      "step": 4694
    },
    {
      "epoch": 0.7896728618282735,
      "grad_norm": NaN,
      "learning_rate": 1.0866571723796487e-05,
      "loss": 0.0,
      "step": 4695
    },
    {
      "epoch": 0.7898410562610377,
      "grad_norm": NaN,
      "learning_rate": 1.0849849842561494e-05,
      "loss": 0.0,
      "step": 4696
    },
    {
      "epoch": 0.790009250693802,
      "grad_norm": NaN,
      "learning_rate": 1.0833139271366371e-05,
      "loss": 0.0,
      "step": 4697
    },
    {
      "epoch": 0.7901774451265663,
      "grad_norm": NaN,
      "learning_rate": 1.081644001503862e-05,
      "loss": 0.0,
      "step": 4698
    },
    {
      "epoch": 0.7903456395593306,
      "grad_norm": NaN,
      "learning_rate": 1.0799752078402469e-05,
      "loss": 0.0,
      "step": 4699
    },
    {
      "epoch": 0.7905138339920948,
      "grad_norm": NaN,
      "learning_rate": 1.078307546627887e-05,
      "loss": 0.0,
      "step": 4700
    },
    {
      "epoch": 0.7906820284248591,
      "grad_norm": NaN,
      "learning_rate": 1.076641018348552e-05,
      "loss": 0.0,
      "step": 4701
    },
    {
      "epoch": 0.7908502228576234,
      "grad_norm": NaN,
      "learning_rate": 1.0749756234836821e-05,
      "loss": 0.0,
      "step": 4702
    },
    {
      "epoch": 0.7910184172903877,
      "grad_norm": NaN,
      "learning_rate": 1.073311362514391e-05,
      "loss": 0.0,
      "step": 4703
    },
    {
      "epoch": 0.791186611723152,
      "grad_norm": NaN,
      "learning_rate": 1.0716482359214664e-05,
      "loss": 0.0,
      "step": 4704
    },
    {
      "epoch": 0.7913548061559162,
      "grad_norm": NaN,
      "learning_rate": 1.069986244185367e-05,
      "loss": 0.0,
      "step": 4705
    },
    {
      "epoch": 0.7915230005886805,
      "grad_norm": NaN,
      "learning_rate": 1.0683253877862225e-05,
      "loss": 0.0,
      "step": 4706
    },
    {
      "epoch": 0.7916911950214448,
      "grad_norm": NaN,
      "learning_rate": 1.0666656672038355e-05,
      "loss": 0.0,
      "step": 4707
    },
    {
      "epoch": 0.7918593894542091,
      "grad_norm": NaN,
      "learning_rate": 1.0650070829176823e-05,
      "loss": 0.0,
      "step": 4708
    },
    {
      "epoch": 0.7920275838869734,
      "grad_norm": NaN,
      "learning_rate": 1.0633496354069084e-05,
      "loss": 0.0,
      "step": 4709
    },
    {
      "epoch": 0.7921957783197376,
      "grad_norm": NaN,
      "learning_rate": 1.061693325150332e-05,
      "loss": 0.0,
      "step": 4710
    },
    {
      "epoch": 0.7923639727525019,
      "grad_norm": NaN,
      "learning_rate": 1.0600381526264408e-05,
      "loss": 0.0,
      "step": 4711
    },
    {
      "epoch": 0.7925321671852662,
      "grad_norm": NaN,
      "learning_rate": 1.0583841183133986e-05,
      "loss": 0.0,
      "step": 4712
    },
    {
      "epoch": 0.7927003616180305,
      "grad_norm": NaN,
      "learning_rate": 1.0567312226890364e-05,
      "loss": 0.0,
      "step": 4713
    },
    {
      "epoch": 0.7928685560507948,
      "grad_norm": NaN,
      "learning_rate": 1.0550794662308566e-05,
      "loss": 0.0,
      "step": 4714
    },
    {
      "epoch": 0.793036750483559,
      "grad_norm": NaN,
      "learning_rate": 1.0534288494160316e-05,
      "loss": 0.0,
      "step": 4715
    },
    {
      "epoch": 0.7932049449163233,
      "grad_norm": NaN,
      "learning_rate": 1.0517793727214098e-05,
      "loss": 0.0,
      "step": 4716
    },
    {
      "epoch": 0.7933731393490876,
      "grad_norm": NaN,
      "learning_rate": 1.0501310366235045e-05,
      "loss": 0.0,
      "step": 4717
    },
    {
      "epoch": 0.7935413337818518,
      "grad_norm": NaN,
      "learning_rate": 1.0484838415985015e-05,
      "loss": 0.0,
      "step": 4718
    },
    {
      "epoch": 0.793709528214616,
      "grad_norm": NaN,
      "learning_rate": 1.0468377881222568e-05,
      "loss": 0.0,
      "step": 4719
    },
    {
      "epoch": 0.7938777226473803,
      "grad_norm": NaN,
      "learning_rate": 1.0451928766702979e-05,
      "loss": 0.0,
      "step": 4720
    },
    {
      "epoch": 0.7940459170801446,
      "grad_norm": NaN,
      "learning_rate": 1.04354910771782e-05,
      "loss": 0.0,
      "step": 4721
    },
    {
      "epoch": 0.7942141115129089,
      "grad_norm": NaN,
      "learning_rate": 1.04190648173969e-05,
      "loss": 0.0,
      "step": 4722
    },
    {
      "epoch": 0.7943823059456732,
      "grad_norm": NaN,
      "learning_rate": 1.0402649992104434e-05,
      "loss": 0.0,
      "step": 4723
    },
    {
      "epoch": 0.7945505003784374,
      "grad_norm": NaN,
      "learning_rate": 1.0386246606042882e-05,
      "loss": 0.0,
      "step": 4724
    },
    {
      "epoch": 0.7947186948112017,
      "grad_norm": NaN,
      "learning_rate": 1.0369854663950983e-05,
      "loss": 0.0,
      "step": 4725
    },
    {
      "epoch": 0.794886889243966,
      "grad_norm": NaN,
      "learning_rate": 1.0353474170564188e-05,
      "loss": 0.0,
      "step": 4726
    },
    {
      "epoch": 0.7950550836767303,
      "grad_norm": NaN,
      "learning_rate": 1.0337105130614627e-05,
      "loss": 0.0,
      "step": 4727
    },
    {
      "epoch": 0.7952232781094946,
      "grad_norm": NaN,
      "learning_rate": 1.032074754883116e-05,
      "loss": 0.0,
      "step": 4728
    },
    {
      "epoch": 0.7953914725422588,
      "grad_norm": NaN,
      "learning_rate": 1.0304401429939293e-05,
      "loss": 0.0,
      "step": 4729
    },
    {
      "epoch": 0.7955596669750231,
      "grad_norm": NaN,
      "learning_rate": 1.0288066778661231e-05,
      "loss": 0.0,
      "step": 4730
    },
    {
      "epoch": 0.7957278614077874,
      "grad_norm": NaN,
      "learning_rate": 1.0271743599715865e-05,
      "loss": 0.0,
      "step": 4731
    },
    {
      "epoch": 0.7958960558405517,
      "grad_norm": NaN,
      "learning_rate": 1.0255431897818807e-05,
      "loss": 0.0,
      "step": 4732
    },
    {
      "epoch": 0.796064250273316,
      "grad_norm": NaN,
      "learning_rate": 1.0239131677682313e-05,
      "loss": 0.0,
      "step": 4733
    },
    {
      "epoch": 0.7962324447060802,
      "grad_norm": NaN,
      "learning_rate": 1.0222842944015326e-05,
      "loss": 0.0,
      "step": 4734
    },
    {
      "epoch": 0.7964006391388445,
      "grad_norm": NaN,
      "learning_rate": 1.0206565701523479e-05,
      "loss": 0.0,
      "step": 4735
    },
    {
      "epoch": 0.7965688335716088,
      "grad_norm": NaN,
      "learning_rate": 1.0190299954909093e-05,
      "loss": 0.0,
      "step": 4736
    },
    {
      "epoch": 0.7967370280043731,
      "grad_norm": NaN,
      "learning_rate": 1.0174045708871154e-05,
      "loss": 0.0,
      "step": 4737
    },
    {
      "epoch": 0.7969052224371374,
      "grad_norm": NaN,
      "learning_rate": 1.015780296810534e-05,
      "loss": 0.0,
      "step": 4738
    },
    {
      "epoch": 0.7970734168699016,
      "grad_norm": NaN,
      "learning_rate": 1.0141571737303968e-05,
      "loss": 0.0,
      "step": 4739
    },
    {
      "epoch": 0.7972416113026659,
      "grad_norm": NaN,
      "learning_rate": 1.0125352021156098e-05,
      "loss": 0.0,
      "step": 4740
    },
    {
      "epoch": 0.7974098057354302,
      "grad_norm": NaN,
      "learning_rate": 1.0109143824347411e-05,
      "loss": 0.0,
      "step": 4741
    },
    {
      "epoch": 0.7975780001681945,
      "grad_norm": NaN,
      "learning_rate": 1.0092947151560267e-05,
      "loss": 0.0,
      "step": 4742
    },
    {
      "epoch": 0.7977461946009587,
      "grad_norm": NaN,
      "learning_rate": 1.007676200747369e-05,
      "loss": 0.0,
      "step": 4743
    },
    {
      "epoch": 0.797914389033723,
      "grad_norm": NaN,
      "learning_rate": 1.0060588396763421e-05,
      "loss": 0.0,
      "step": 4744
    },
    {
      "epoch": 0.7980825834664873,
      "grad_norm": NaN,
      "learning_rate": 1.0044426324101813e-05,
      "loss": 0.0,
      "step": 4745
    },
    {
      "epoch": 0.7982507778992516,
      "grad_norm": NaN,
      "learning_rate": 1.0028275794157915e-05,
      "loss": 0.0,
      "step": 4746
    },
    {
      "epoch": 0.7984189723320159,
      "grad_norm": NaN,
      "learning_rate": 1.0012136811597428e-05,
      "loss": 0.0,
      "step": 4747
    },
    {
      "epoch": 0.79858716676478,
      "grad_norm": NaN,
      "learning_rate": 9.996009381082717e-06,
      "loss": 0.0,
      "step": 4748
    },
    {
      "epoch": 0.7987553611975443,
      "grad_norm": NaN,
      "learning_rate": 9.979893507272836e-06,
      "loss": 0.0,
      "step": 4749
    },
    {
      "epoch": 0.7989235556303086,
      "grad_norm": NaN,
      "learning_rate": 9.963789194823469e-06,
      "loss": 0.0,
      "step": 4750
    },
    {
      "epoch": 0.7990917500630729,
      "grad_norm": NaN,
      "learning_rate": 9.94769644838698e-06,
      "loss": 0.0,
      "step": 4751
    },
    {
      "epoch": 0.7992599444958371,
      "grad_norm": NaN,
      "learning_rate": 9.931615272612354e-06,
      "loss": 0.0,
      "step": 4752
    },
    {
      "epoch": 0.7994281389286014,
      "grad_norm": NaN,
      "learning_rate": 9.915545672145299e-06,
      "loss": 0.0,
      "step": 4753
    },
    {
      "epoch": 0.7995963333613657,
      "grad_norm": NaN,
      "learning_rate": 9.899487651628132e-06,
      "loss": 0.0,
      "step": 4754
    },
    {
      "epoch": 0.79976452779413,
      "grad_norm": NaN,
      "learning_rate": 9.883441215699823e-06,
      "loss": 0.0,
      "step": 4755
    },
    {
      "epoch": 0.7999327222268943,
      "grad_norm": NaN,
      "learning_rate": 9.867406368996023e-06,
      "loss": 0.0,
      "step": 4756
    },
    {
      "epoch": 0.8001009166596585,
      "grad_norm": NaN,
      "learning_rate": 9.85138311614901e-06,
      "loss": 0.0,
      "step": 4757
    },
    {
      "epoch": 0.8002691110924228,
      "grad_norm": NaN,
      "learning_rate": 9.835371461787724e-06,
      "loss": 0.0,
      "step": 4758
    },
    {
      "epoch": 0.8004373055251871,
      "grad_norm": NaN,
      "learning_rate": 9.819371410537747e-06,
      "loss": 0.0,
      "step": 4759
    },
    {
      "epoch": 0.8006054999579514,
      "grad_norm": NaN,
      "learning_rate": 9.803382967021318e-06,
      "loss": 0.0,
      "step": 4760
    },
    {
      "epoch": 0.8007736943907157,
      "grad_norm": NaN,
      "learning_rate": 9.787406135857329e-06,
      "loss": 0.0,
      "step": 4761
    },
    {
      "epoch": 0.8009418888234799,
      "grad_norm": NaN,
      "learning_rate": 9.7714409216613e-06,
      "loss": 0.0,
      "step": 4762
    },
    {
      "epoch": 0.8011100832562442,
      "grad_norm": NaN,
      "learning_rate": 9.755487329045404e-06,
      "loss": 0.0,
      "step": 4763
    },
    {
      "epoch": 0.8012782776890085,
      "grad_norm": NaN,
      "learning_rate": 9.739545362618446e-06,
      "loss": 0.0,
      "step": 4764
    },
    {
      "epoch": 0.8014464721217728,
      "grad_norm": NaN,
      "learning_rate": 9.7236150269859e-06,
      "loss": 0.0,
      "step": 4765
    },
    {
      "epoch": 0.8016146665545371,
      "grad_norm": NaN,
      "learning_rate": 9.707696326749848e-06,
      "loss": 0.0,
      "step": 4766
    },
    {
      "epoch": 0.8017828609873013,
      "grad_norm": NaN,
      "learning_rate": 9.691789266509028e-06,
      "loss": 0.0,
      "step": 4767
    },
    {
      "epoch": 0.8019510554200656,
      "grad_norm": NaN,
      "learning_rate": 9.675893850858803e-06,
      "loss": 0.0,
      "step": 4768
    },
    {
      "epoch": 0.8021192498528299,
      "grad_norm": NaN,
      "learning_rate": 9.660010084391197e-06,
      "loss": 0.0,
      "step": 4769
    },
    {
      "epoch": 0.8022874442855942,
      "grad_norm": NaN,
      "learning_rate": 9.644137971694844e-06,
      "loss": 0.0,
      "step": 4770
    },
    {
      "epoch": 0.8024556387183585,
      "grad_norm": NaN,
      "learning_rate": 9.628277517355021e-06,
      "loss": 0.0,
      "step": 4771
    },
    {
      "epoch": 0.8026238331511227,
      "grad_norm": NaN,
      "learning_rate": 9.61242872595362e-06,
      "loss": 0.0,
      "step": 4772
    },
    {
      "epoch": 0.802792027583887,
      "grad_norm": NaN,
      "learning_rate": 9.59659160206921e-06,
      "loss": 0.0,
      "step": 4773
    },
    {
      "epoch": 0.8029602220166513,
      "grad_norm": NaN,
      "learning_rate": 9.580766150276937e-06,
      "loss": 0.0,
      "step": 4774
    },
    {
      "epoch": 0.8031284164494156,
      "grad_norm": NaN,
      "learning_rate": 9.564952375148606e-06,
      "loss": 0.0,
      "step": 4775
    },
    {
      "epoch": 0.8032966108821799,
      "grad_norm": NaN,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0,
      "step": 4776
    },
    {
      "epoch": 0.8034648053149441,
      "grad_norm": NaN,
      "learning_rate": 9.533359873154068e-06,
      "loss": 0.0,
      "step": 4777
    },
    {
      "epoch": 0.8036329997477083,
      "grad_norm": NaN,
      "learning_rate": 9.517581155414584e-06,
      "loss": 0.0,
      "step": 4778
    },
    {
      "epoch": 0.8038011941804726,
      "grad_norm": NaN,
      "learning_rate": 9.501814132592474e-06,
      "loss": 0.0,
      "step": 4779
    },
    {
      "epoch": 0.8039693886132369,
      "grad_norm": NaN,
      "learning_rate": 9.486058809242642e-06,
      "loss": 0.0,
      "step": 4780
    },
    {
      "epoch": 0.8041375830460011,
      "grad_norm": NaN,
      "learning_rate": 9.470315189916656e-06,
      "loss": 0.0,
      "step": 4781
    },
    {
      "epoch": 0.8043057774787654,
      "grad_norm": NaN,
      "learning_rate": 9.454583279162648e-06,
      "loss": 0.0,
      "step": 4782
    },
    {
      "epoch": 0.8044739719115297,
      "grad_norm": NaN,
      "learning_rate": 9.438863081525396e-06,
      "loss": 0.0,
      "step": 4783
    },
    {
      "epoch": 0.804642166344294,
      "grad_norm": NaN,
      "learning_rate": 9.423154601546281e-06,
      "loss": 0.0,
      "step": 4784
    },
    {
      "epoch": 0.8048103607770583,
      "grad_norm": NaN,
      "learning_rate": 9.407457843763324e-06,
      "loss": 0.0,
      "step": 4785
    },
    {
      "epoch": 0.8049785552098225,
      "grad_norm": NaN,
      "learning_rate": 9.39177281271113e-06,
      "loss": 0.0,
      "step": 4786
    },
    {
      "epoch": 0.8051467496425868,
      "grad_norm": NaN,
      "learning_rate": 9.376099512920939e-06,
      "loss": 0.0,
      "step": 4787
    },
    {
      "epoch": 0.8053149440753511,
      "grad_norm": NaN,
      "learning_rate": 9.360437948920564e-06,
      "loss": 0.0,
      "step": 4788
    },
    {
      "epoch": 0.8054831385081154,
      "grad_norm": NaN,
      "learning_rate": 9.3447881252345e-06,
      "loss": 0.0,
      "step": 4789
    },
    {
      "epoch": 0.8056513329408796,
      "grad_norm": NaN,
      "learning_rate": 9.329150046383772e-06,
      "loss": 0.0,
      "step": 4790
    },
    {
      "epoch": 0.8058195273736439,
      "grad_norm": NaN,
      "learning_rate": 9.313523716886063e-06,
      "loss": 0.0,
      "step": 4791
    },
    {
      "epoch": 0.8059877218064082,
      "grad_norm": NaN,
      "learning_rate": 9.297909141255633e-06,
      "loss": 0.0,
      "step": 4792
    },
    {
      "epoch": 0.8061559162391725,
      "grad_norm": NaN,
      "learning_rate": 9.282306324003364e-06,
      "loss": 0.0,
      "step": 4793
    },
    {
      "epoch": 0.8063241106719368,
      "grad_norm": NaN,
      "learning_rate": 9.266715269636733e-06,
      "loss": 0.0,
      "step": 4794
    },
    {
      "epoch": 0.806492305104701,
      "grad_norm": NaN,
      "learning_rate": 9.251135982659825e-06,
      "loss": 0.0,
      "step": 4795
    },
    {
      "epoch": 0.8066604995374653,
      "grad_norm": NaN,
      "learning_rate": 9.235568467573313e-06,
      "loss": 0.0,
      "step": 4796
    },
    {
      "epoch": 0.8068286939702296,
      "grad_norm": NaN,
      "learning_rate": 9.220012728874473e-06,
      "loss": 0.0,
      "step": 4797
    },
    {
      "epoch": 0.8069968884029939,
      "grad_norm": NaN,
      "learning_rate": 9.204468771057206e-06,
      "loss": 0.0,
      "step": 4798
    },
    {
      "epoch": 0.8071650828357582,
      "grad_norm": NaN,
      "learning_rate": 9.188936598611975e-06,
      "loss": 0.0,
      "step": 4799
    },
    {
      "epoch": 0.8073332772685224,
      "grad_norm": NaN,
      "learning_rate": 9.173416216025854e-06,
      "loss": 0.0,
      "step": 4800
    },
    {
      "epoch": 0.8075014717012867,
      "grad_norm": NaN,
      "learning_rate": 9.157907627782485e-06,
      "loss": 0.0,
      "step": 4801
    },
    {
      "epoch": 0.807669666134051,
      "grad_norm": NaN,
      "learning_rate": 9.142410838362164e-06,
      "loss": 0.0,
      "step": 4802
    },
    {
      "epoch": 0.8078378605668153,
      "grad_norm": NaN,
      "learning_rate": 9.12692585224172e-06,
      "loss": 0.0,
      "step": 4803
    },
    {
      "epoch": 0.8080060549995796,
      "grad_norm": NaN,
      "learning_rate": 9.111452673894588e-06,
      "loss": 0.0,
      "step": 4804
    },
    {
      "epoch": 0.8081742494323438,
      "grad_norm": NaN,
      "learning_rate": 9.095991307790786e-06,
      "loss": 0.0,
      "step": 4805
    },
    {
      "epoch": 0.8083424438651081,
      "grad_norm": NaN,
      "learning_rate": 9.08054175839696e-06,
      "loss": 0.0,
      "step": 4806
    },
    {
      "epoch": 0.8085106382978723,
      "grad_norm": NaN,
      "learning_rate": 9.065104030176297e-06,
      "loss": 0.0,
      "step": 4807
    },
    {
      "epoch": 0.8086788327306366,
      "grad_norm": NaN,
      "learning_rate": 9.049678127588585e-06,
      "loss": 0.0,
      "step": 4808
    },
    {
      "epoch": 0.8088470271634008,
      "grad_norm": NaN,
      "learning_rate": 9.034264055090175e-06,
      "loss": 0.0,
      "step": 4809
    },
    {
      "epoch": 0.8090152215961651,
      "grad_norm": NaN,
      "learning_rate": 9.01886181713405e-06,
      "loss": 0.0,
      "step": 4810
    },
    {
      "epoch": 0.8091834160289294,
      "grad_norm": NaN,
      "learning_rate": 9.003471418169734e-06,
      "loss": 0.0,
      "step": 4811
    },
    {
      "epoch": 0.8093516104616937,
      "grad_norm": NaN,
      "learning_rate": 8.988092862643332e-06,
      "loss": 0.0,
      "step": 4812
    },
    {
      "epoch": 0.809519804894458,
      "grad_norm": NaN,
      "learning_rate": 8.972726154997547e-06,
      "loss": 0.0,
      "step": 4813
    },
    {
      "epoch": 0.8096879993272222,
      "grad_norm": NaN,
      "learning_rate": 8.957371299671641e-06,
      "loss": 0.0,
      "step": 4814
    },
    {
      "epoch": 0.8098561937599865,
      "grad_norm": NaN,
      "learning_rate": 8.942028301101468e-06,
      "loss": 0.0,
      "step": 4815
    },
    {
      "epoch": 0.8100243881927508,
      "grad_norm": NaN,
      "learning_rate": 8.926697163719438e-06,
      "loss": 0.0,
      "step": 4816
    },
    {
      "epoch": 0.8101925826255151,
      "grad_norm": NaN,
      "learning_rate": 8.911377891954543e-06,
      "loss": 0.0,
      "step": 4817
    },
    {
      "epoch": 0.8103607770582794,
      "grad_norm": NaN,
      "learning_rate": 8.89607049023236e-06,
      "loss": 0.0,
      "step": 4818
    },
    {
      "epoch": 0.8105289714910436,
      "grad_norm": NaN,
      "learning_rate": 8.88077496297503e-06,
      "loss": 0.0,
      "step": 4819
    },
    {
      "epoch": 0.8106971659238079,
      "grad_norm": NaN,
      "learning_rate": 8.865491314601249e-06,
      "loss": 0.0,
      "step": 4820
    },
    {
      "epoch": 0.8108653603565722,
      "grad_norm": NaN,
      "learning_rate": 8.850219549526278e-06,
      "loss": 0.0,
      "step": 4821
    },
    {
      "epoch": 0.8110335547893365,
      "grad_norm": NaN,
      "learning_rate": 8.834959672161991e-06,
      "loss": 0.0,
      "step": 4822
    },
    {
      "epoch": 0.8112017492221008,
      "grad_norm": NaN,
      "learning_rate": 8.819711686916783e-06,
      "loss": 0.0,
      "step": 4823
    },
    {
      "epoch": 0.811369943654865,
      "grad_norm": NaN,
      "learning_rate": 8.804475598195616e-06,
      "loss": 0.0,
      "step": 4824
    },
    {
      "epoch": 0.8115381380876293,
      "grad_norm": NaN,
      "learning_rate": 8.789251410400023e-06,
      "loss": 0.0,
      "step": 4825
    },
    {
      "epoch": 0.8117063325203936,
      "grad_norm": NaN,
      "learning_rate": 8.774039127928124e-06,
      "loss": 0.0,
      "step": 4826
    },
    {
      "epoch": 0.8118745269531579,
      "grad_norm": NaN,
      "learning_rate": 8.758838755174564e-06,
      "loss": 0.0,
      "step": 4827
    },
    {
      "epoch": 0.8120427213859222,
      "grad_norm": NaN,
      "learning_rate": 8.743650296530559e-06,
      "loss": 0.0,
      "step": 4828
    },
    {
      "epoch": 0.8122109158186864,
      "grad_norm": NaN,
      "learning_rate": 8.728473756383887e-06,
      "loss": 0.0,
      "step": 4829
    },
    {
      "epoch": 0.8123791102514507,
      "grad_norm": NaN,
      "learning_rate": 8.71330913911888e-06,
      "loss": 0.0,
      "step": 4830
    },
    {
      "epoch": 0.812547304684215,
      "grad_norm": NaN,
      "learning_rate": 8.698156449116424e-06,
      "loss": 0.0,
      "step": 4831
    },
    {
      "epoch": 0.8127154991169793,
      "grad_norm": NaN,
      "learning_rate": 8.68301569075396e-06,
      "loss": 0.0,
      "step": 4832
    },
    {
      "epoch": 0.8128836935497435,
      "grad_norm": NaN,
      "learning_rate": 8.667886868405478e-06,
      "loss": 0.0,
      "step": 4833
    },
    {
      "epoch": 0.8130518879825078,
      "grad_norm": NaN,
      "learning_rate": 8.652769986441544e-06,
      "loss": 0.0,
      "step": 4834
    },
    {
      "epoch": 0.8132200824152721,
      "grad_norm": NaN,
      "learning_rate": 8.637665049229243e-06,
      "loss": 0.0,
      "step": 4835
    },
    {
      "epoch": 0.8133882768480364,
      "grad_norm": NaN,
      "learning_rate": 8.622572061132227e-06,
      "loss": 0.0,
      "step": 4836
    },
    {
      "epoch": 0.8135564712808006,
      "grad_norm": NaN,
      "learning_rate": 8.607491026510678e-06,
      "loss": 0.0,
      "step": 4837
    },
    {
      "epoch": 0.8137246657135648,
      "grad_norm": NaN,
      "learning_rate": 8.592421949721364e-06,
      "loss": 0.0,
      "step": 4838
    },
    {
      "epoch": 0.8138928601463291,
      "grad_norm": NaN,
      "learning_rate": 8.577364835117552e-06,
      "loss": 0.0,
      "step": 4839
    },
    {
      "epoch": 0.8140610545790934,
      "grad_norm": NaN,
      "learning_rate": 8.56231968704908e-06,
      "loss": 0.0,
      "step": 4840
    },
    {
      "epoch": 0.8142292490118577,
      "grad_norm": NaN,
      "learning_rate": 8.547286509862306e-06,
      "loss": 0.0,
      "step": 4841
    },
    {
      "epoch": 0.814397443444622,
      "grad_norm": NaN,
      "learning_rate": 8.532265307900178e-06,
      "loss": 0.0,
      "step": 4842
    },
    {
      "epoch": 0.8145656378773862,
      "grad_norm": NaN,
      "learning_rate": 8.51725608550214e-06,
      "loss": 0.0,
      "step": 4843
    },
    {
      "epoch": 0.8147338323101505,
      "grad_norm": NaN,
      "learning_rate": 8.50225884700418e-06,
      "loss": 0.0,
      "step": 4844
    },
    {
      "epoch": 0.8149020267429148,
      "grad_norm": NaN,
      "learning_rate": 8.487273596738832e-06,
      "loss": 0.0,
      "step": 4845
    },
    {
      "epoch": 0.8150702211756791,
      "grad_norm": NaN,
      "learning_rate": 8.472300339035178e-06,
      "loss": 0.0,
      "step": 4846
    },
    {
      "epoch": 0.8152384156084433,
      "grad_norm": NaN,
      "learning_rate": 8.457339078218823e-06,
      "loss": 0.0,
      "step": 4847
    },
    {
      "epoch": 0.8154066100412076,
      "grad_norm": NaN,
      "learning_rate": 8.442389818611906e-06,
      "loss": 0.0,
      "step": 4848
    },
    {
      "epoch": 0.8155748044739719,
      "grad_norm": NaN,
      "learning_rate": 8.427452564533094e-06,
      "loss": 0.0,
      "step": 4849
    },
    {
      "epoch": 0.8157429989067362,
      "grad_norm": NaN,
      "learning_rate": 8.412527320297598e-06,
      "loss": 0.0,
      "step": 4850
    },
    {
      "epoch": 0.8159111933395005,
      "grad_norm": NaN,
      "learning_rate": 8.397614090217155e-06,
      "loss": 0.0,
      "step": 4851
    },
    {
      "epoch": 0.8160793877722647,
      "grad_norm": NaN,
      "learning_rate": 8.382712878600023e-06,
      "loss": 0.0,
      "step": 4852
    },
    {
      "epoch": 0.816247582205029,
      "grad_norm": NaN,
      "learning_rate": 8.367823689751009e-06,
      "loss": 0.0,
      "step": 4853
    },
    {
      "epoch": 0.8164157766377933,
      "grad_norm": NaN,
      "learning_rate": 8.352946527971406e-06,
      "loss": 0.0,
      "step": 4854
    },
    {
      "epoch": 0.8165839710705576,
      "grad_norm": NaN,
      "learning_rate": 8.338081397559088e-06,
      "loss": 0.0,
      "step": 4855
    },
    {
      "epoch": 0.8167521655033219,
      "grad_norm": NaN,
      "learning_rate": 8.323228302808417e-06,
      "loss": 0.0,
      "step": 4856
    },
    {
      "epoch": 0.8169203599360861,
      "grad_norm": NaN,
      "learning_rate": 8.308387248010279e-06,
      "loss": 0.0,
      "step": 4857
    },
    {
      "epoch": 0.8170885543688504,
      "grad_norm": NaN,
      "learning_rate": 8.293558237452082e-06,
      "loss": 0.0,
      "step": 4858
    },
    {
      "epoch": 0.8172567488016147,
      "grad_norm": NaN,
      "learning_rate": 8.278741275417784e-06,
      "loss": 0.0,
      "step": 4859
    },
    {
      "epoch": 0.817424943234379,
      "grad_norm": NaN,
      "learning_rate": 8.263936366187824e-06,
      "loss": 0.0,
      "step": 4860
    },
    {
      "epoch": 0.8175931376671433,
      "grad_norm": NaN,
      "learning_rate": 8.24914351403918e-06,
      "loss": 0.0,
      "step": 4861
    },
    {
      "epoch": 0.8177613320999075,
      "grad_norm": NaN,
      "learning_rate": 8.234362723245326e-06,
      "loss": 0.0,
      "step": 4862
    },
    {
      "epoch": 0.8179295265326718,
      "grad_norm": NaN,
      "learning_rate": 8.219593998076292e-06,
      "loss": 0.0,
      "step": 4863
    },
    {
      "epoch": 0.8180977209654361,
      "grad_norm": NaN,
      "learning_rate": 8.204837342798589e-06,
      "loss": 0.0,
      "step": 4864
    },
    {
      "epoch": 0.8182659153982004,
      "grad_norm": NaN,
      "learning_rate": 8.19009276167524e-06,
      "loss": 0.0,
      "step": 4865
    },
    {
      "epoch": 0.8184341098309645,
      "grad_norm": NaN,
      "learning_rate": 8.17536025896578e-06,
      "loss": 0.0,
      "step": 4866
    },
    {
      "epoch": 0.8186023042637288,
      "grad_norm": NaN,
      "learning_rate": 8.160639838926293e-06,
      "loss": 0.0,
      "step": 4867
    },
    {
      "epoch": 0.8187704986964931,
      "grad_norm": NaN,
      "learning_rate": 8.145931505809329e-06,
      "loss": 0.0,
      "step": 4868
    },
    {
      "epoch": 0.8189386931292574,
      "grad_norm": NaN,
      "learning_rate": 8.131235263863957e-06,
      "loss": 0.0,
      "step": 4869
    },
    {
      "epoch": 0.8191068875620217,
      "grad_norm": NaN,
      "learning_rate": 8.116551117335763e-06,
      "loss": 0.0,
      "step": 4870
    },
    {
      "epoch": 0.8192750819947859,
      "grad_norm": NaN,
      "learning_rate": 8.101879070466833e-06,
      "loss": 0.0,
      "step": 4871
    },
    {
      "epoch": 0.8194432764275502,
      "grad_norm": NaN,
      "learning_rate": 8.087219127495748e-06,
      "loss": 0.0,
      "step": 4872
    },
    {
      "epoch": 0.8196114708603145,
      "grad_norm": NaN,
      "learning_rate": 8.07257129265761e-06,
      "loss": 0.0,
      "step": 4873
    },
    {
      "epoch": 0.8197796652930788,
      "grad_norm": NaN,
      "learning_rate": 8.057935570184e-06,
      "loss": 0.0,
      "step": 4874
    },
    {
      "epoch": 0.819947859725843,
      "grad_norm": NaN,
      "learning_rate": 8.043311964303036e-06,
      "loss": 0.0,
      "step": 4875
    },
    {
      "epoch": 0.8201160541586073,
      "grad_norm": NaN,
      "learning_rate": 8.028700479239303e-06,
      "loss": 0.0,
      "step": 4876
    },
    {
      "epoch": 0.8202842485913716,
      "grad_norm": NaN,
      "learning_rate": 8.014101119213902e-06,
      "loss": 0.0,
      "step": 4877
    },
    {
      "epoch": 0.8204524430241359,
      "grad_norm": NaN,
      "learning_rate": 7.999513888444399e-06,
      "loss": 0.0,
      "step": 4878
    },
    {
      "epoch": 0.8206206374569002,
      "grad_norm": NaN,
      "learning_rate": 7.98493879114492e-06,
      "loss": 0.0,
      "step": 4879
    },
    {
      "epoch": 0.8207888318896644,
      "grad_norm": NaN,
      "learning_rate": 7.970375831526023e-06,
      "loss": 0.0,
      "step": 4880
    },
    {
      "epoch": 0.8209570263224287,
      "grad_norm": NaN,
      "learning_rate": 7.955825013794793e-06,
      "loss": 0.0,
      "step": 4881
    },
    {
      "epoch": 0.821125220755193,
      "grad_norm": NaN,
      "learning_rate": 7.94128634215478e-06,
      "loss": 0.0,
      "step": 4882
    },
    {
      "epoch": 0.8212934151879573,
      "grad_norm": NaN,
      "learning_rate": 7.926759820806067e-06,
      "loss": 0.0,
      "step": 4883
    },
    {
      "epoch": 0.8214616096207216,
      "grad_norm": NaN,
      "learning_rate": 7.912245453945199e-06,
      "loss": 0.0,
      "step": 4884
    },
    {
      "epoch": 0.8216298040534858,
      "grad_norm": NaN,
      "learning_rate": 7.89774324576521e-06,
      "loss": 0.0,
      "step": 4885
    },
    {
      "epoch": 0.8217979984862501,
      "grad_norm": NaN,
      "learning_rate": 7.883253200455627e-06,
      "loss": 0.0,
      "step": 4886
    },
    {
      "epoch": 0.8219661929190144,
      "grad_norm": NaN,
      "learning_rate": 7.868775322202454e-06,
      "loss": 0.0,
      "step": 4887
    },
    {
      "epoch": 0.8221343873517787,
      "grad_norm": NaN,
      "learning_rate": 7.854309615188198e-06,
      "loss": 0.0,
      "step": 4888
    },
    {
      "epoch": 0.822302581784543,
      "grad_norm": NaN,
      "learning_rate": 7.839856083591834e-06,
      "loss": 0.0,
      "step": 4889
    },
    {
      "epoch": 0.8224707762173072,
      "grad_norm": NaN,
      "learning_rate": 7.825414731588816e-06,
      "loss": 0.0,
      "step": 4890
    },
    {
      "epoch": 0.8226389706500715,
      "grad_norm": NaN,
      "learning_rate": 7.81098556335111e-06,
      "loss": 0.0,
      "step": 4891
    },
    {
      "epoch": 0.8228071650828358,
      "grad_norm": NaN,
      "learning_rate": 7.796568583047132e-06,
      "loss": 0.0,
      "step": 4892
    },
    {
      "epoch": 0.8229753595156001,
      "grad_norm": NaN,
      "learning_rate": 7.782163794841784e-06,
      "loss": 0.0,
      "step": 4893
    },
    {
      "epoch": 0.8231435539483644,
      "grad_norm": NaN,
      "learning_rate": 7.76777120289644e-06,
      "loss": 0.0,
      "step": 4894
    },
    {
      "epoch": 0.8233117483811286,
      "grad_norm": NaN,
      "learning_rate": 7.753390811368971e-06,
      "loss": 0.0,
      "step": 4895
    },
    {
      "epoch": 0.8234799428138928,
      "grad_norm": NaN,
      "learning_rate": 7.739022624413717e-06,
      "loss": 0.0,
      "step": 4896
    },
    {
      "epoch": 0.8236481372466571,
      "grad_norm": NaN,
      "learning_rate": 7.724666646181472e-06,
      "loss": 0.0,
      "step": 4897
    },
    {
      "epoch": 0.8238163316794214,
      "grad_norm": NaN,
      "learning_rate": 7.710322880819521e-06,
      "loss": 0.0,
      "step": 4898
    },
    {
      "epoch": 0.8239845261121856,
      "grad_norm": NaN,
      "learning_rate": 7.695991332471608e-06,
      "loss": 0.0,
      "step": 4899
    },
    {
      "epoch": 0.8241527205449499,
      "grad_norm": NaN,
      "learning_rate": 7.681672005277979e-06,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 0.8243209149777142,
      "grad_norm": NaN,
      "learning_rate": 7.667364903375313e-06,
      "loss": 0.0,
      "step": 4901
    },
    {
      "epoch": 0.8244891094104785,
      "grad_norm": NaN,
      "learning_rate": 7.653070030896774e-06,
      "loss": 0.0,
      "step": 4902
    },
    {
      "epoch": 0.8246573038432428,
      "grad_norm": NaN,
      "learning_rate": 7.638787391971986e-06,
      "loss": 0.0,
      "step": 4903
    },
    {
      "epoch": 0.824825498276007,
      "grad_norm": NaN,
      "learning_rate": 7.624516990727049e-06,
      "loss": 0.0,
      "step": 4904
    },
    {
      "epoch": 0.8249936927087713,
      "grad_norm": NaN,
      "learning_rate": 7.610258831284528e-06,
      "loss": 0.0,
      "step": 4905
    },
    {
      "epoch": 0.8251618871415356,
      "grad_norm": NaN,
      "learning_rate": 7.596012917763434e-06,
      "loss": 0.0,
      "step": 4906
    },
    {
      "epoch": 0.8253300815742999,
      "grad_norm": NaN,
      "learning_rate": 7.581779254279259e-06,
      "loss": 0.0,
      "step": 4907
    },
    {
      "epoch": 0.8254982760070642,
      "grad_norm": NaN,
      "learning_rate": 7.5675578449439455e-06,
      "loss": 0.0,
      "step": 4908
    },
    {
      "epoch": 0.8256664704398284,
      "grad_norm": NaN,
      "learning_rate": 7.553348693865897e-06,
      "loss": 0.0,
      "step": 4909
    },
    {
      "epoch": 0.8258346648725927,
      "grad_norm": NaN,
      "learning_rate": 7.539151805149985e-06,
      "loss": 0.0,
      "step": 4910
    },
    {
      "epoch": 0.826002859305357,
      "grad_norm": NaN,
      "learning_rate": 7.524967182897513e-06,
      "loss": 0.0,
      "step": 4911
    },
    {
      "epoch": 0.8261710537381213,
      "grad_norm": NaN,
      "learning_rate": 7.510794831206291e-06,
      "loss": 0.0,
      "step": 4912
    },
    {
      "epoch": 0.8263392481708856,
      "grad_norm": NaN,
      "learning_rate": 7.496634754170534e-06,
      "loss": 0.0,
      "step": 4913
    },
    {
      "epoch": 0.8265074426036498,
      "grad_norm": NaN,
      "learning_rate": 7.48248695588093e-06,
      "loss": 0.0,
      "step": 4914
    },
    {
      "epoch": 0.8266756370364141,
      "grad_norm": NaN,
      "learning_rate": 7.468351440424609e-06,
      "loss": 0.0,
      "step": 4915
    },
    {
      "epoch": 0.8268438314691784,
      "grad_norm": NaN,
      "learning_rate": 7.454228211885184e-06,
      "loss": 0.0,
      "step": 4916
    },
    {
      "epoch": 0.8270120259019427,
      "grad_norm": NaN,
      "learning_rate": 7.440117274342695e-06,
      "loss": 0.0,
      "step": 4917
    },
    {
      "epoch": 0.827180220334707,
      "grad_norm": NaN,
      "learning_rate": 7.426018631873621e-06,
      "loss": 0.0,
      "step": 4918
    },
    {
      "epoch": 0.8273484147674712,
      "grad_norm": NaN,
      "learning_rate": 7.411932288550899e-06,
      "loss": 0.0,
      "step": 4919
    },
    {
      "epoch": 0.8275166092002355,
      "grad_norm": NaN,
      "learning_rate": 7.397858248443939e-06,
      "loss": 0.0,
      "step": 4920
    },
    {
      "epoch": 0.8276848036329998,
      "grad_norm": NaN,
      "learning_rate": 7.383796515618557e-06,
      "loss": 0.0,
      "step": 4921
    },
    {
      "epoch": 0.8278529980657641,
      "grad_norm": NaN,
      "learning_rate": 7.369747094137036e-06,
      "loss": 0.0,
      "step": 4922
    },
    {
      "epoch": 0.8280211924985283,
      "grad_norm": NaN,
      "learning_rate": 7.35570998805809e-06,
      "loss": 0.0,
      "step": 4923
    },
    {
      "epoch": 0.8281893869312926,
      "grad_norm": NaN,
      "learning_rate": 7.341685201436887e-06,
      "loss": 0.0,
      "step": 4924
    },
    {
      "epoch": 0.8283575813640569,
      "grad_norm": NaN,
      "learning_rate": 7.327672738325025e-06,
      "loss": 0.0,
      "step": 4925
    },
    {
      "epoch": 0.8285257757968211,
      "grad_norm": NaN,
      "learning_rate": 7.313672602770561e-06,
      "loss": 0.0,
      "step": 4926
    },
    {
      "epoch": 0.8286939702295854,
      "grad_norm": NaN,
      "learning_rate": 7.2996847988179466e-06,
      "loss": 0.0,
      "step": 4927
    },
    {
      "epoch": 0.8288621646623496,
      "grad_norm": NaN,
      "learning_rate": 7.2857093305081394e-06,
      "loss": 0.0,
      "step": 4928
    },
    {
      "epoch": 0.8290303590951139,
      "grad_norm": NaN,
      "learning_rate": 7.271746201878477e-06,
      "loss": 0.0,
      "step": 4929
    },
    {
      "epoch": 0.8291985535278782,
      "grad_norm": NaN,
      "learning_rate": 7.257795416962753e-06,
      "loss": 0.0,
      "step": 4930
    },
    {
      "epoch": 0.8293667479606425,
      "grad_norm": NaN,
      "learning_rate": 7.2438569797911806e-06,
      "loss": 0.0,
      "step": 4931
    },
    {
      "epoch": 0.8295349423934067,
      "grad_norm": NaN,
      "learning_rate": 7.229930894390446e-06,
      "loss": 0.0,
      "step": 4932
    },
    {
      "epoch": 0.829703136826171,
      "grad_norm": NaN,
      "learning_rate": 7.216017164783617e-06,
      "loss": 0.0,
      "step": 4933
    },
    {
      "epoch": 0.8298713312589353,
      "grad_norm": NaN,
      "learning_rate": 7.202115794990228e-06,
      "loss": 0.0,
      "step": 4934
    },
    {
      "epoch": 0.8300395256916996,
      "grad_norm": NaN,
      "learning_rate": 7.188226789026209e-06,
      "loss": 0.0,
      "step": 4935
    },
    {
      "epoch": 0.8302077201244639,
      "grad_norm": NaN,
      "learning_rate": 7.174350150903958e-06,
      "loss": 0.0,
      "step": 4936
    },
    {
      "epoch": 0.8303759145572281,
      "grad_norm": NaN,
      "learning_rate": 7.160485884632278e-06,
      "loss": 0.0,
      "step": 4937
    },
    {
      "epoch": 0.8305441089899924,
      "grad_norm": NaN,
      "learning_rate": 7.1466339942163955e-06,
      "loss": 0.0,
      "step": 4938
    },
    {
      "epoch": 0.8307123034227567,
      "grad_norm": NaN,
      "learning_rate": 7.132794483657957e-06,
      "loss": 0.0,
      "step": 4939
    },
    {
      "epoch": 0.830880497855521,
      "grad_norm": NaN,
      "learning_rate": 7.118967356955059e-06,
      "loss": 0.0,
      "step": 4940
    },
    {
      "epoch": 0.8310486922882853,
      "grad_norm": NaN,
      "learning_rate": 7.105152618102195e-06,
      "loss": 0.0,
      "step": 4941
    },
    {
      "epoch": 0.8312168867210495,
      "grad_norm": NaN,
      "learning_rate": 7.091350271090297e-06,
      "loss": 0.0,
      "step": 4942
    },
    {
      "epoch": 0.8313850811538138,
      "grad_norm": NaN,
      "learning_rate": 7.077560319906695e-06,
      "loss": 0.0,
      "step": 4943
    },
    {
      "epoch": 0.8315532755865781,
      "grad_norm": NaN,
      "learning_rate": 7.06378276853516e-06,
      "loss": 0.0,
      "step": 4944
    },
    {
      "epoch": 0.8317214700193424,
      "grad_norm": NaN,
      "learning_rate": 7.0500176209558735e-06,
      "loss": 0.0,
      "step": 4945
    },
    {
      "epoch": 0.8318896644521067,
      "grad_norm": NaN,
      "learning_rate": 7.036264881145427e-06,
      "loss": 0.0,
      "step": 4946
    },
    {
      "epoch": 0.8320578588848709,
      "grad_norm": NaN,
      "learning_rate": 7.0225245530768315e-06,
      "loss": 0.0,
      "step": 4947
    },
    {
      "epoch": 0.8322260533176352,
      "grad_norm": NaN,
      "learning_rate": 7.0087966407195064e-06,
      "loss": 0.0,
      "step": 4948
    },
    {
      "epoch": 0.8323942477503995,
      "grad_norm": NaN,
      "learning_rate": 6.9950811480393175e-06,
      "loss": 0.0,
      "step": 4949
    },
    {
      "epoch": 0.8325624421831638,
      "grad_norm": NaN,
      "learning_rate": 6.981378078998502e-06,
      "loss": 0.0,
      "step": 4950
    },
    {
      "epoch": 0.832730636615928,
      "grad_norm": NaN,
      "learning_rate": 6.96768743755572e-06,
      "loss": 0.0,
      "step": 4951
    },
    {
      "epoch": 0.8328988310486923,
      "grad_norm": NaN,
      "learning_rate": 6.954009227666042e-06,
      "loss": 0.0,
      "step": 4952
    },
    {
      "epoch": 0.8330670254814566,
      "grad_norm": NaN,
      "learning_rate": 6.940343453280962e-06,
      "loss": 0.0,
      "step": 4953
    },
    {
      "epoch": 0.8332352199142209,
      "grad_norm": NaN,
      "learning_rate": 6.926690118348361e-06,
      "loss": 0.0,
      "step": 4954
    },
    {
      "epoch": 0.8334034143469851,
      "grad_norm": NaN,
      "learning_rate": 6.91304922681254e-06,
      "loss": 0.0,
      "step": 4955
    },
    {
      "epoch": 0.8335716087797493,
      "grad_norm": NaN,
      "learning_rate": 6.899420782614185e-06,
      "loss": 0.0,
      "step": 4956
    },
    {
      "epoch": 0.8337398032125136,
      "grad_norm": NaN,
      "learning_rate": 6.88580478969042e-06,
      "loss": 0.0,
      "step": 4957
    },
    {
      "epoch": 0.8339079976452779,
      "grad_norm": NaN,
      "learning_rate": 6.872201251974747e-06,
      "loss": 0.0,
      "step": 4958
    },
    {
      "epoch": 0.8340761920780422,
      "grad_norm": NaN,
      "learning_rate": 6.858610173397067e-06,
      "loss": 0.0,
      "step": 4959
    },
    {
      "epoch": 0.8342443865108065,
      "grad_norm": NaN,
      "learning_rate": 6.8450315578836965e-06,
      "loss": 0.0,
      "step": 4960
    },
    {
      "epoch": 0.8344125809435707,
      "grad_norm": NaN,
      "learning_rate": 6.831465409357346e-06,
      "loss": 0.0,
      "step": 4961
    },
    {
      "epoch": 0.834580775376335,
      "grad_norm": NaN,
      "learning_rate": 6.817911731737131e-06,
      "loss": 0.0,
      "step": 4962
    },
    {
      "epoch": 0.8347489698090993,
      "grad_norm": NaN,
      "learning_rate": 6.8043705289385475e-06,
      "loss": 0.0,
      "step": 4963
    },
    {
      "epoch": 0.8349171642418636,
      "grad_norm": NaN,
      "learning_rate": 6.790841804873499e-06,
      "loss": 0.0,
      "step": 4964
    },
    {
      "epoch": 0.8350853586746279,
      "grad_norm": NaN,
      "learning_rate": 6.777325563450282e-06,
      "loss": 0.0,
      "step": 4965
    },
    {
      "epoch": 0.8352535531073921,
      "grad_norm": NaN,
      "learning_rate": 6.763821808573589e-06,
      "loss": 0.0,
      "step": 4966
    },
    {
      "epoch": 0.8354217475401564,
      "grad_norm": NaN,
      "learning_rate": 6.750330544144501e-06,
      "loss": 0.0,
      "step": 4967
    },
    {
      "epoch": 0.8355899419729207,
      "grad_norm": NaN,
      "learning_rate": 6.736851774060482e-06,
      "loss": 0.0,
      "step": 4968
    },
    {
      "epoch": 0.835758136405685,
      "grad_norm": NaN,
      "learning_rate": 6.723385502215429e-06,
      "loss": 0.0,
      "step": 4969
    },
    {
      "epoch": 0.8359263308384492,
      "grad_norm": NaN,
      "learning_rate": 6.709931732499569e-06,
      "loss": 0.0,
      "step": 4970
    },
    {
      "epoch": 0.8360945252712135,
      "grad_norm": NaN,
      "learning_rate": 6.69649046879956e-06,
      "loss": 0.0,
      "step": 4971
    },
    {
      "epoch": 0.8362627197039778,
      "grad_norm": NaN,
      "learning_rate": 6.6830617149984175e-06,
      "loss": 0.0,
      "step": 4972
    },
    {
      "epoch": 0.8364309141367421,
      "grad_norm": NaN,
      "learning_rate": 6.669645474975578e-06,
      "loss": 0.0,
      "step": 4973
    },
    {
      "epoch": 0.8365991085695064,
      "grad_norm": NaN,
      "learning_rate": 6.6562417526068345e-06,
      "loss": 0.0,
      "step": 4974
    },
    {
      "epoch": 0.8367673030022706,
      "grad_norm": NaN,
      "learning_rate": 6.642850551764368e-06,
      "loss": 0.0,
      "step": 4975
    },
    {
      "epoch": 0.8369354974350349,
      "grad_norm": NaN,
      "learning_rate": 6.629471876316739e-06,
      "loss": 0.0,
      "step": 4976
    },
    {
      "epoch": 0.8371036918677992,
      "grad_norm": NaN,
      "learning_rate": 6.616105730128919e-06,
      "loss": 0.0,
      "step": 4977
    },
    {
      "epoch": 0.8372718863005635,
      "grad_norm": NaN,
      "learning_rate": 6.602752117062228e-06,
      "loss": 0.0,
      "step": 4978
    },
    {
      "epoch": 0.8374400807333278,
      "grad_norm": NaN,
      "learning_rate": 6.589411040974369e-06,
      "loss": 0.0,
      "step": 4979
    },
    {
      "epoch": 0.837608275166092,
      "grad_norm": NaN,
      "learning_rate": 6.576082505719433e-06,
      "loss": 0.0,
      "step": 4980
    },
    {
      "epoch": 0.8377764695988563,
      "grad_norm": NaN,
      "learning_rate": 6.562766515147889e-06,
      "loss": 0.0,
      "step": 4981
    },
    {
      "epoch": 0.8379446640316206,
      "grad_norm": NaN,
      "learning_rate": 6.549463073106571e-06,
      "loss": 0.0,
      "step": 4982
    },
    {
      "epoch": 0.8381128584643849,
      "grad_norm": NaN,
      "learning_rate": 6.536172183438699e-06,
      "loss": 0.0,
      "step": 4983
    },
    {
      "epoch": 0.8382810528971492,
      "grad_norm": NaN,
      "learning_rate": 6.522893849983852e-06,
      "loss": 0.0,
      "step": 4984
    },
    {
      "epoch": 0.8384492473299133,
      "grad_norm": NaN,
      "learning_rate": 6.509628076578006e-06,
      "loss": 0.0,
      "step": 4985
    },
    {
      "epoch": 0.8386174417626776,
      "grad_norm": NaN,
      "learning_rate": 6.496374867053495e-06,
      "loss": 0.0,
      "step": 4986
    },
    {
      "epoch": 0.8387856361954419,
      "grad_norm": NaN,
      "learning_rate": 6.483134225239013e-06,
      "loss": 0.0,
      "step": 4987
    },
    {
      "epoch": 0.8389538306282062,
      "grad_norm": NaN,
      "learning_rate": 6.46990615495962e-06,
      "loss": 0.0,
      "step": 4988
    },
    {
      "epoch": 0.8391220250609704,
      "grad_norm": NaN,
      "learning_rate": 6.456690660036796e-06,
      "loss": 0.0,
      "step": 4989
    },
    {
      "epoch": 0.8392902194937347,
      "grad_norm": NaN,
      "learning_rate": 6.443487744288318e-06,
      "loss": 0.0,
      "step": 4990
    },
    {
      "epoch": 0.839458413926499,
      "grad_norm": NaN,
      "learning_rate": 6.430297411528374e-06,
      "loss": 0.0,
      "step": 4991
    },
    {
      "epoch": 0.8396266083592633,
      "grad_norm": NaN,
      "learning_rate": 6.41711966556749e-06,
      "loss": 0.0,
      "step": 4992
    },
    {
      "epoch": 0.8397948027920276,
      "grad_norm": NaN,
      "learning_rate": 6.4039545102125845e-06,
      "loss": 0.0,
      "step": 4993
    },
    {
      "epoch": 0.8399629972247918,
      "grad_norm": NaN,
      "learning_rate": 6.390801949266917e-06,
      "loss": 0.0,
      "step": 4994
    },
    {
      "epoch": 0.8401311916575561,
      "grad_norm": NaN,
      "learning_rate": 6.3776619865301206e-06,
      "loss": 0.0,
      "step": 4995
    },
    {
      "epoch": 0.8402993860903204,
      "grad_norm": NaN,
      "learning_rate": 6.364534625798174e-06,
      "loss": 0.0,
      "step": 4996
    },
    {
      "epoch": 0.8404675805230847,
      "grad_norm": NaN,
      "learning_rate": 6.351419870863417e-06,
      "loss": 0.0,
      "step": 4997
    },
    {
      "epoch": 0.840635774955849,
      "grad_norm": NaN,
      "learning_rate": 6.338317725514581e-06,
      "loss": 0.0,
      "step": 4998
    },
    {
      "epoch": 0.8408039693886132,
      "grad_norm": NaN,
      "learning_rate": 6.325228193536714e-06,
      "loss": 0.0,
      "step": 4999
    },
    {
      "epoch": 0.8409721638213775,
      "grad_norm": NaN,
      "learning_rate": 6.312151278711237e-06,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 0.8411403582541418,
      "grad_norm": NaN,
      "learning_rate": 6.299086984815927e-06,
      "loss": 0.0,
      "step": 5001
    },
    {
      "epoch": 0.8413085526869061,
      "grad_norm": NaN,
      "learning_rate": 6.286035315624905e-06,
      "loss": 0.0,
      "step": 5002
    },
    {
      "epoch": 0.8414767471196704,
      "grad_norm": NaN,
      "learning_rate": 6.272996274908654e-06,
      "loss": 0.0,
      "step": 5003
    },
    {
      "epoch": 0.8416449415524346,
      "grad_norm": NaN,
      "learning_rate": 6.259969866434012e-06,
      "loss": 0.0,
      "step": 5004
    },
    {
      "epoch": 0.8418131359851989,
      "grad_norm": NaN,
      "learning_rate": 6.246956093964146e-06,
      "loss": 0.0,
      "step": 5005
    },
    {
      "epoch": 0.8419813304179632,
      "grad_norm": NaN,
      "learning_rate": 6.233954961258615e-06,
      "loss": 0.0,
      "step": 5006
    },
    {
      "epoch": 0.8421495248507275,
      "grad_norm": NaN,
      "learning_rate": 6.220966472073287e-06,
      "loss": 0.0,
      "step": 5007
    },
    {
      "epoch": 0.8423177192834918,
      "grad_norm": NaN,
      "learning_rate": 6.207990630160393e-06,
      "loss": 0.0,
      "step": 5008
    },
    {
      "epoch": 0.842485913716256,
      "grad_norm": NaN,
      "learning_rate": 6.195027439268497e-06,
      "loss": 0.0,
      "step": 5009
    },
    {
      "epoch": 0.8426541081490203,
      "grad_norm": NaN,
      "learning_rate": 6.182076903142536e-06,
      "loss": 0.0,
      "step": 5010
    },
    {
      "epoch": 0.8428223025817846,
      "grad_norm": NaN,
      "learning_rate": 6.169139025523773e-06,
      "loss": 0.0,
      "step": 5011
    },
    {
      "epoch": 0.8429904970145489,
      "grad_norm": NaN,
      "learning_rate": 6.156213810149813e-06,
      "loss": 0.0,
      "step": 5012
    },
    {
      "epoch": 0.8431586914473131,
      "grad_norm": NaN,
      "learning_rate": 6.143301260754591e-06,
      "loss": 0.0,
      "step": 5013
    },
    {
      "epoch": 0.8433268858800773,
      "grad_norm": NaN,
      "learning_rate": 6.130401381068424e-06,
      "loss": 0.0,
      "step": 5014
    },
    {
      "epoch": 0.8434950803128416,
      "grad_norm": NaN,
      "learning_rate": 6.117514174817929e-06,
      "loss": 0.0,
      "step": 5015
    },
    {
      "epoch": 0.8436632747456059,
      "grad_norm": NaN,
      "learning_rate": 6.104639645726079e-06,
      "loss": 0.0,
      "step": 5016
    },
    {
      "epoch": 0.8438314691783702,
      "grad_norm": NaN,
      "learning_rate": 6.091777797512177e-06,
      "loss": 0.0,
      "step": 5017
    },
    {
      "epoch": 0.8439996636111344,
      "grad_norm": NaN,
      "learning_rate": 6.078928633891867e-06,
      "loss": 0.0,
      "step": 5018
    },
    {
      "epoch": 0.8441678580438987,
      "grad_norm": NaN,
      "learning_rate": 6.066092158577136e-06,
      "loss": 0.0,
      "step": 5019
    },
    {
      "epoch": 0.844336052476663,
      "grad_norm": NaN,
      "learning_rate": 6.053268375276289e-06,
      "loss": 0.0,
      "step": 5020
    },
    {
      "epoch": 0.8445042469094273,
      "grad_norm": NaN,
      "learning_rate": 6.040457287693963e-06,
      "loss": 0.0,
      "step": 5021
    },
    {
      "epoch": 0.8446724413421915,
      "grad_norm": NaN,
      "learning_rate": 6.027658899531169e-06,
      "loss": 0.0,
      "step": 5022
    },
    {
      "epoch": 0.8448406357749558,
      "grad_norm": NaN,
      "learning_rate": 6.014873214485195e-06,
      "loss": 0.0,
      "step": 5023
    },
    {
      "epoch": 0.8450088302077201,
      "grad_norm": NaN,
      "learning_rate": 6.002100236249691e-06,
      "loss": 0.0,
      "step": 5024
    },
    {
      "epoch": 0.8451770246404844,
      "grad_norm": NaN,
      "learning_rate": 5.989339968514612e-06,
      "loss": 0.0,
      "step": 5025
    },
    {
      "epoch": 0.8453452190732487,
      "grad_norm": NaN,
      "learning_rate": 5.976592414966281e-06,
      "loss": 0.0,
      "step": 5026
    },
    {
      "epoch": 0.845513413506013,
      "grad_norm": NaN,
      "learning_rate": 5.963857579287308e-06,
      "loss": 0.0,
      "step": 5027
    },
    {
      "epoch": 0.8456816079387772,
      "grad_norm": NaN,
      "learning_rate": 5.9511354651566485e-06,
      "loss": 0.0,
      "step": 5028
    },
    {
      "epoch": 0.8458498023715415,
      "grad_norm": NaN,
      "learning_rate": 5.938426076249565e-06,
      "loss": 0.0,
      "step": 5029
    },
    {
      "epoch": 0.8460179968043058,
      "grad_norm": NaN,
      "learning_rate": 5.925729416237685e-06,
      "loss": 0.0,
      "step": 5030
    },
    {
      "epoch": 0.8461861912370701,
      "grad_norm": NaN,
      "learning_rate": 5.913045488788915e-06,
      "loss": 0.0,
      "step": 5031
    },
    {
      "epoch": 0.8463543856698343,
      "grad_norm": NaN,
      "learning_rate": 5.900374297567502e-06,
      "loss": 0.0,
      "step": 5032
    },
    {
      "epoch": 0.8465225801025986,
      "grad_norm": NaN,
      "learning_rate": 5.887715846233993e-06,
      "loss": 0.0,
      "step": 5033
    },
    {
      "epoch": 0.8466907745353629,
      "grad_norm": NaN,
      "learning_rate": 5.8750701384453095e-06,
      "loss": 0.0,
      "step": 5034
    },
    {
      "epoch": 0.8468589689681272,
      "grad_norm": NaN,
      "learning_rate": 5.86243717785463e-06,
      "loss": 0.0,
      "step": 5035
    },
    {
      "epoch": 0.8470271634008915,
      "grad_norm": NaN,
      "learning_rate": 5.849816968111482e-06,
      "loss": 0.0,
      "step": 5036
    },
    {
      "epoch": 0.8471953578336557,
      "grad_norm": NaN,
      "learning_rate": 5.837209512861702e-06,
      "loss": 0.0,
      "step": 5037
    },
    {
      "epoch": 0.84736355226642,
      "grad_norm": NaN,
      "learning_rate": 5.824614815747442e-06,
      "loss": 0.0,
      "step": 5038
    },
    {
      "epoch": 0.8475317466991843,
      "grad_norm": NaN,
      "learning_rate": 5.81203288040717e-06,
      "loss": 0.0,
      "step": 5039
    },
    {
      "epoch": 0.8476999411319486,
      "grad_norm": NaN,
      "learning_rate": 5.799463710475667e-06,
      "loss": 0.0,
      "step": 5040
    },
    {
      "epoch": 0.8478681355647129,
      "grad_norm": NaN,
      "learning_rate": 5.786907309584011e-06,
      "loss": 0.0,
      "step": 5041
    },
    {
      "epoch": 0.8480363299974771,
      "grad_norm": NaN,
      "learning_rate": 5.774363681359624e-06,
      "loss": 0.0,
      "step": 5042
    },
    {
      "epoch": 0.8482045244302414,
      "grad_norm": NaN,
      "learning_rate": 5.761832829426223e-06,
      "loss": 0.0,
      "step": 5043
    },
    {
      "epoch": 0.8483727188630056,
      "grad_norm": NaN,
      "learning_rate": 5.749314757403812e-06,
      "loss": 0.0,
      "step": 5044
    },
    {
      "epoch": 0.8485409132957699,
      "grad_norm": NaN,
      "learning_rate": 5.736809468908727e-06,
      "loss": 0.0,
      "step": 5045
    },
    {
      "epoch": 0.8487091077285341,
      "grad_norm": NaN,
      "learning_rate": 5.724316967553622e-06,
      "loss": 0.0,
      "step": 5046
    },
    {
      "epoch": 0.8488773021612984,
      "grad_norm": NaN,
      "learning_rate": 5.711837256947427e-06,
      "loss": 0.0,
      "step": 5047
    },
    {
      "epoch": 0.8490454965940627,
      "grad_norm": NaN,
      "learning_rate": 5.6993703406953924e-06,
      "loss": 0.0,
      "step": 5048
    },
    {
      "epoch": 0.849213691026827,
      "grad_norm": NaN,
      "learning_rate": 5.686916222399069e-06,
      "loss": 0.0,
      "step": 5049
    },
    {
      "epoch": 0.8493818854595913,
      "grad_norm": NaN,
      "learning_rate": 5.674474905656307e-06,
      "loss": 0.0,
      "step": 5050
    },
    {
      "epoch": 0.8495500798923555,
      "grad_norm": NaN,
      "learning_rate": 5.662046394061277e-06,
      "loss": 0.0,
      "step": 5051
    },
    {
      "epoch": 0.8497182743251198,
      "grad_norm": NaN,
      "learning_rate": 5.649630691204433e-06,
      "loss": 0.0,
      "step": 5052
    },
    {
      "epoch": 0.8498864687578841,
      "grad_norm": NaN,
      "learning_rate": 5.637227800672529e-06,
      "loss": 0.0,
      "step": 5053
    },
    {
      "epoch": 0.8500546631906484,
      "grad_norm": NaN,
      "learning_rate": 5.62483772604862e-06,
      "loss": 0.0,
      "step": 5054
    },
    {
      "epoch": 0.8502228576234127,
      "grad_norm": NaN,
      "learning_rate": 5.612460470912067e-06,
      "loss": 0.0,
      "step": 5055
    },
    {
      "epoch": 0.8503910520561769,
      "grad_norm": NaN,
      "learning_rate": 5.6000960388385095e-06,
      "loss": 0.0,
      "step": 5056
    },
    {
      "epoch": 0.8505592464889412,
      "grad_norm": NaN,
      "learning_rate": 5.587744433399905e-06,
      "loss": 0.0,
      "step": 5057
    },
    {
      "epoch": 0.8507274409217055,
      "grad_norm": NaN,
      "learning_rate": 5.5754056581644735e-06,
      "loss": 0.0,
      "step": 5058
    },
    {
      "epoch": 0.8508956353544698,
      "grad_norm": NaN,
      "learning_rate": 5.563079716696773e-06,
      "loss": 0.0,
      "step": 5059
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": NaN,
      "learning_rate": 5.550766612557623e-06,
      "loss": 0.0,
      "step": 5060
    },
    {
      "epoch": 0.8512320242199983,
      "grad_norm": NaN,
      "learning_rate": 5.538466349304133e-06,
      "loss": 0.0,
      "step": 5061
    },
    {
      "epoch": 0.8514002186527626,
      "grad_norm": NaN,
      "learning_rate": 5.526178930489711e-06,
      "loss": 0.0,
      "step": 5062
    },
    {
      "epoch": 0.8515684130855269,
      "grad_norm": NaN,
      "learning_rate": 5.513904359664074e-06,
      "loss": 0.0,
      "step": 5063
    },
    {
      "epoch": 0.8517366075182912,
      "grad_norm": NaN,
      "learning_rate": 5.501642640373189e-06,
      "loss": 0.0,
      "step": 5064
    },
    {
      "epoch": 0.8519048019510554,
      "grad_norm": NaN,
      "learning_rate": 5.489393776159335e-06,
      "loss": 0.0,
      "step": 5065
    },
    {
      "epoch": 0.8520729963838197,
      "grad_norm": NaN,
      "learning_rate": 5.477157770561064e-06,
      "loss": 0.0,
      "step": 5066
    },
    {
      "epoch": 0.852241190816584,
      "grad_norm": NaN,
      "learning_rate": 5.4649346271132385e-06,
      "loss": 0.0,
      "step": 5067
    },
    {
      "epoch": 0.8524093852493483,
      "grad_norm": NaN,
      "learning_rate": 5.452724349346977e-06,
      "loss": 0.0,
      "step": 5068
    },
    {
      "epoch": 0.8525775796821126,
      "grad_norm": NaN,
      "learning_rate": 5.440526940789698e-06,
      "loss": 0.0,
      "step": 5069
    },
    {
      "epoch": 0.8527457741148768,
      "grad_norm": NaN,
      "learning_rate": 5.428342404965076e-06,
      "loss": 0.0,
      "step": 5070
    },
    {
      "epoch": 0.8529139685476411,
      "grad_norm": NaN,
      "learning_rate": 5.416170745393112e-06,
      "loss": 0.0,
      "step": 5071
    },
    {
      "epoch": 0.8530821629804054,
      "grad_norm": NaN,
      "learning_rate": 5.404011965590055e-06,
      "loss": 0.0,
      "step": 5072
    },
    {
      "epoch": 0.8532503574131696,
      "grad_norm": NaN,
      "learning_rate": 5.391866069068436e-06,
      "loss": 0.0,
      "step": 5073
    },
    {
      "epoch": 0.8534185518459338,
      "grad_norm": NaN,
      "learning_rate": 5.379733059337067e-06,
      "loss": 0.0,
      "step": 5074
    },
    {
      "epoch": 0.8535867462786981,
      "grad_norm": NaN,
      "learning_rate": 5.367612939901045e-06,
      "loss": 0.0,
      "step": 5075
    },
    {
      "epoch": 0.8537549407114624,
      "grad_norm": NaN,
      "learning_rate": 5.355505714261732e-06,
      "loss": 0.0,
      "step": 5076
    },
    {
      "epoch": 0.8539231351442267,
      "grad_norm": NaN,
      "learning_rate": 5.343411385916769e-06,
      "loss": 0.0,
      "step": 5077
    },
    {
      "epoch": 0.854091329576991,
      "grad_norm": NaN,
      "learning_rate": 5.331329958360071e-06,
      "loss": 0.0,
      "step": 5078
    },
    {
      "epoch": 0.8542595240097552,
      "grad_norm": NaN,
      "learning_rate": 5.319261435081834e-06,
      "loss": 0.0,
      "step": 5079
    },
    {
      "epoch": 0.8544277184425195,
      "grad_norm": NaN,
      "learning_rate": 5.307205819568517e-06,
      "loss": 0.0,
      "step": 5080
    },
    {
      "epoch": 0.8545959128752838,
      "grad_norm": NaN,
      "learning_rate": 5.295163115302854e-06,
      "loss": 0.0,
      "step": 5081
    },
    {
      "epoch": 0.8547641073080481,
      "grad_norm": NaN,
      "learning_rate": 5.283133325763828e-06,
      "loss": 0.0,
      "step": 5082
    },
    {
      "epoch": 0.8549323017408124,
      "grad_norm": NaN,
      "learning_rate": 5.271116454426739e-06,
      "loss": 0.0,
      "step": 5083
    },
    {
      "epoch": 0.8551004961735766,
      "grad_norm": NaN,
      "learning_rate": 5.259112504763114e-06,
      "loss": 0.0,
      "step": 5084
    },
    {
      "epoch": 0.8552686906063409,
      "grad_norm": NaN,
      "learning_rate": 5.247121480240763e-06,
      "loss": 0.0,
      "step": 5085
    },
    {
      "epoch": 0.8554368850391052,
      "grad_norm": NaN,
      "learning_rate": 5.235143384323743e-06,
      "loss": 0.0,
      "step": 5086
    },
    {
      "epoch": 0.8556050794718695,
      "grad_norm": NaN,
      "learning_rate": 5.2231782204724136e-06,
      "loss": 0.0,
      "step": 5087
    },
    {
      "epoch": 0.8557732739046338,
      "grad_norm": NaN,
      "learning_rate": 5.211225992143365e-06,
      "loss": 0.0,
      "step": 5088
    },
    {
      "epoch": 0.855941468337398,
      "grad_norm": NaN,
      "learning_rate": 5.199286702789468e-06,
      "loss": 0.0,
      "step": 5089
    },
    {
      "epoch": 0.8561096627701623,
      "grad_norm": NaN,
      "learning_rate": 5.187360355859838e-06,
      "loss": 0.0,
      "step": 5090
    },
    {
      "epoch": 0.8562778572029266,
      "grad_norm": NaN,
      "learning_rate": 5.175446954799873e-06,
      "loss": 0.0,
      "step": 5091
    },
    {
      "epoch": 0.8564460516356909,
      "grad_norm": NaN,
      "learning_rate": 5.163546503051226e-06,
      "loss": 0.0,
      "step": 5092
    },
    {
      "epoch": 0.8566142460684552,
      "grad_norm": NaN,
      "learning_rate": 5.151659004051795e-06,
      "loss": 0.0,
      "step": 5093
    },
    {
      "epoch": 0.8567824405012194,
      "grad_norm": NaN,
      "learning_rate": 5.139784461235747e-06,
      "loss": 0.0,
      "step": 5094
    },
    {
      "epoch": 0.8569506349339837,
      "grad_norm": NaN,
      "learning_rate": 5.127922878033503e-06,
      "loss": 0.0,
      "step": 5095
    },
    {
      "epoch": 0.857118829366748,
      "grad_norm": NaN,
      "learning_rate": 5.116074257871744e-06,
      "loss": 0.0,
      "step": 5096
    },
    {
      "epoch": 0.8572870237995123,
      "grad_norm": NaN,
      "learning_rate": 5.1042386041734055e-06,
      "loss": 0.0,
      "step": 5097
    },
    {
      "epoch": 0.8574552182322766,
      "grad_norm": NaN,
      "learning_rate": 5.092415920357674e-06,
      "loss": 0.0,
      "step": 5098
    },
    {
      "epoch": 0.8576234126650408,
      "grad_norm": NaN,
      "learning_rate": 5.080606209839977e-06,
      "loss": 0.0,
      "step": 5099
    },
    {
      "epoch": 0.8577916070978051,
      "grad_norm": NaN,
      "learning_rate": 5.068809476032033e-06,
      "loss": 0.0,
      "step": 5100
    },
    {
      "epoch": 0.8579598015305694,
      "grad_norm": NaN,
      "learning_rate": 5.0570257223417684e-06,
      "loss": 0.0,
      "step": 5101
    },
    {
      "epoch": 0.8581279959633337,
      "grad_norm": NaN,
      "learning_rate": 5.045254952173384e-06,
      "loss": 0.0,
      "step": 5102
    },
    {
      "epoch": 0.8582961903960978,
      "grad_norm": NaN,
      "learning_rate": 5.033497168927315e-06,
      "loss": 0.0,
      "step": 5103
    },
    {
      "epoch": 0.8584643848288621,
      "grad_norm": NaN,
      "learning_rate": 5.021752376000266e-06,
      "loss": 0.0,
      "step": 5104
    },
    {
      "epoch": 0.8586325792616264,
      "grad_norm": NaN,
      "learning_rate": 5.010020576785174e-06,
      "loss": 0.0,
      "step": 5105
    },
    {
      "epoch": 0.8588007736943907,
      "grad_norm": NaN,
      "learning_rate": 4.998301774671216e-06,
      "loss": 0.0,
      "step": 5106
    },
    {
      "epoch": 0.858968968127155,
      "grad_norm": NaN,
      "learning_rate": 4.9865959730438225e-06,
      "loss": 0.0,
      "step": 5107
    },
    {
      "epoch": 0.8591371625599192,
      "grad_norm": NaN,
      "learning_rate": 4.974903175284678e-06,
      "loss": 0.0,
      "step": 5108
    },
    {
      "epoch": 0.8593053569926835,
      "grad_norm": NaN,
      "learning_rate": 4.963223384771703e-06,
      "loss": 0.0,
      "step": 5109
    },
    {
      "epoch": 0.8594735514254478,
      "grad_norm": NaN,
      "learning_rate": 4.951556604879048e-06,
      "loss": 0.0,
      "step": 5110
    },
    {
      "epoch": 0.8596417458582121,
      "grad_norm": NaN,
      "learning_rate": 4.939902838977117e-06,
      "loss": 0.0,
      "step": 5111
    },
    {
      "epoch": 0.8598099402909763,
      "grad_norm": NaN,
      "learning_rate": 4.928262090432556e-06,
      "loss": 0.0,
      "step": 5112
    },
    {
      "epoch": 0.8599781347237406,
      "grad_norm": NaN,
      "learning_rate": 4.916634362608241e-06,
      "loss": 0.0,
      "step": 5113
    },
    {
      "epoch": 0.8601463291565049,
      "grad_norm": NaN,
      "learning_rate": 4.905019658863297e-06,
      "loss": 0.0,
      "step": 5114
    },
    {
      "epoch": 0.8603145235892692,
      "grad_norm": NaN,
      "learning_rate": 4.8934179825530725e-06,
      "loss": 0.0,
      "step": 5115
    },
    {
      "epoch": 0.8604827180220335,
      "grad_norm": NaN,
      "learning_rate": 4.8818293370291816e-06,
      "loss": 0.0,
      "step": 5116
    },
    {
      "epoch": 0.8606509124547977,
      "grad_norm": NaN,
      "learning_rate": 4.87025372563944e-06,
      "loss": 0.0,
      "step": 5117
    },
    {
      "epoch": 0.860819106887562,
      "grad_norm": NaN,
      "learning_rate": 4.858691151727918e-06,
      "loss": 0.0,
      "step": 5118
    },
    {
      "epoch": 0.8609873013203263,
      "grad_norm": NaN,
      "learning_rate": 4.847141618634898e-06,
      "loss": 0.0,
      "step": 5119
    },
    {
      "epoch": 0.8611554957530906,
      "grad_norm": NaN,
      "learning_rate": 4.8356051296969355e-06,
      "loss": 0.0,
      "step": 5120
    },
    {
      "epoch": 0.8613236901858549,
      "grad_norm": NaN,
      "learning_rate": 4.8240816882467786e-06,
      "loss": 0.0,
      "step": 5121
    },
    {
      "epoch": 0.8614918846186191,
      "grad_norm": NaN,
      "learning_rate": 4.812571297613422e-06,
      "loss": 0.0,
      "step": 5122
    },
    {
      "epoch": 0.8616600790513834,
      "grad_norm": NaN,
      "learning_rate": 4.801073961122082e-06,
      "loss": 0.0,
      "step": 5123
    },
    {
      "epoch": 0.8618282734841477,
      "grad_norm": NaN,
      "learning_rate": 4.789589682094225e-06,
      "loss": 0.0,
      "step": 5124
    },
    {
      "epoch": 0.861996467916912,
      "grad_norm": NaN,
      "learning_rate": 4.778118463847525e-06,
      "loss": 0.0,
      "step": 5125
    },
    {
      "epoch": 0.8621646623496763,
      "grad_norm": NaN,
      "learning_rate": 4.766660309695886e-06,
      "loss": 0.0,
      "step": 5126
    },
    {
      "epoch": 0.8623328567824405,
      "grad_norm": NaN,
      "learning_rate": 4.755215222949433e-06,
      "loss": 0.0,
      "step": 5127
    },
    {
      "epoch": 0.8625010512152048,
      "grad_norm": NaN,
      "learning_rate": 4.743783206914537e-06,
      "loss": 0.0,
      "step": 5128
    },
    {
      "epoch": 0.8626692456479691,
      "grad_norm": NaN,
      "learning_rate": 4.73236426489378e-06,
      "loss": 0.0,
      "step": 5129
    },
    {
      "epoch": 0.8628374400807334,
      "grad_norm": NaN,
      "learning_rate": 4.720958400185954e-06,
      "loss": 0.0,
      "step": 5130
    },
    {
      "epoch": 0.8630056345134977,
      "grad_norm": NaN,
      "learning_rate": 4.709565616086098e-06,
      "loss": 0.0,
      "step": 5131
    },
    {
      "epoch": 0.8631738289462619,
      "grad_norm": NaN,
      "learning_rate": 4.69818591588545e-06,
      "loss": 0.0,
      "step": 5132
    },
    {
      "epoch": 0.8633420233790261,
      "grad_norm": NaN,
      "learning_rate": 4.686819302871481e-06,
      "loss": 0.0,
      "step": 5133
    },
    {
      "epoch": 0.8635102178117904,
      "grad_norm": NaN,
      "learning_rate": 4.675465780327876e-06,
      "loss": 0.0,
      "step": 5134
    },
    {
      "epoch": 0.8636784122445547,
      "grad_norm": NaN,
      "learning_rate": 4.664125351534537e-06,
      "loss": 0.0,
      "step": 5135
    },
    {
      "epoch": 0.8638466066773189,
      "grad_norm": NaN,
      "learning_rate": 4.652798019767607e-06,
      "loss": 0.0,
      "step": 5136
    },
    {
      "epoch": 0.8640148011100832,
      "grad_norm": NaN,
      "learning_rate": 4.641483788299405e-06,
      "loss": 0.0,
      "step": 5137
    },
    {
      "epoch": 0.8641829955428475,
      "grad_norm": NaN,
      "learning_rate": 4.630182660398496e-06,
      "loss": 0.0,
      "step": 5138
    },
    {
      "epoch": 0.8643511899756118,
      "grad_norm": NaN,
      "learning_rate": 4.618894639329629e-06,
      "loss": 0.0,
      "step": 5139
    },
    {
      "epoch": 0.8645193844083761,
      "grad_norm": NaN,
      "learning_rate": 4.607619728353818e-06,
      "loss": 0.0,
      "step": 5140
    },
    {
      "epoch": 0.8646875788411403,
      "grad_norm": NaN,
      "learning_rate": 4.596357930728245e-06,
      "loss": 0.0,
      "step": 5141
    },
    {
      "epoch": 0.8648557732739046,
      "grad_norm": NaN,
      "learning_rate": 4.585109249706315e-06,
      "loss": 0.0,
      "step": 5142
    },
    {
      "epoch": 0.8650239677066689,
      "grad_norm": NaN,
      "learning_rate": 4.573873688537639e-06,
      "loss": 0.0,
      "step": 5143
    },
    {
      "epoch": 0.8651921621394332,
      "grad_norm": NaN,
      "learning_rate": 4.562651250468059e-06,
      "loss": 0.0,
      "step": 5144
    },
    {
      "epoch": 0.8653603565721975,
      "grad_norm": NaN,
      "learning_rate": 4.551441938739603e-06,
      "loss": 0.0,
      "step": 5145
    },
    {
      "epoch": 0.8655285510049617,
      "grad_norm": NaN,
      "learning_rate": 4.540245756590522e-06,
      "loss": 0.0,
      "step": 5146
    },
    {
      "epoch": 0.865696745437726,
      "grad_norm": NaN,
      "learning_rate": 4.529062707255261e-06,
      "loss": 0.0,
      "step": 5147
    },
    {
      "epoch": 0.8658649398704903,
      "grad_norm": NaN,
      "learning_rate": 4.517892793964484e-06,
      "loss": 0.0,
      "step": 5148
    },
    {
      "epoch": 0.8660331343032546,
      "grad_norm": NaN,
      "learning_rate": 4.5067360199450456e-06,
      "loss": 0.0,
      "step": 5149
    },
    {
      "epoch": 0.8662013287360188,
      "grad_norm": NaN,
      "learning_rate": 4.495592388420017e-06,
      "loss": 0.0,
      "step": 5150
    },
    {
      "epoch": 0.8663695231687831,
      "grad_norm": NaN,
      "learning_rate": 4.484461902608672e-06,
      "loss": 0.0,
      "step": 5151
    },
    {
      "epoch": 0.8665377176015474,
      "grad_norm": NaN,
      "learning_rate": 4.473344565726467e-06,
      "loss": 0.0,
      "step": 5152
    },
    {
      "epoch": 0.8667059120343117,
      "grad_norm": NaN,
      "learning_rate": 4.462240380985106e-06,
      "loss": 0.0,
      "step": 5153
    },
    {
      "epoch": 0.866874106467076,
      "grad_norm": NaN,
      "learning_rate": 4.451149351592437e-06,
      "loss": 0.0,
      "step": 5154
    },
    {
      "epoch": 0.8670423008998402,
      "grad_norm": NaN,
      "learning_rate": 4.44007148075255e-06,
      "loss": 0.0,
      "step": 5155
    },
    {
      "epoch": 0.8672104953326045,
      "grad_norm": NaN,
      "learning_rate": 4.429006771665706e-06,
      "loss": 0.0,
      "step": 5156
    },
    {
      "epoch": 0.8673786897653688,
      "grad_norm": NaN,
      "learning_rate": 4.417955227528381e-06,
      "loss": 0.0,
      "step": 5157
    },
    {
      "epoch": 0.8675468841981331,
      "grad_norm": NaN,
      "learning_rate": 4.406916851533249e-06,
      "loss": 0.0,
      "step": 5158
    },
    {
      "epoch": 0.8677150786308974,
      "grad_norm": NaN,
      "learning_rate": 4.395891646869166e-06,
      "loss": 0.0,
      "step": 5159
    },
    {
      "epoch": 0.8678832730636616,
      "grad_norm": NaN,
      "learning_rate": 4.384879616721182e-06,
      "loss": 0.0,
      "step": 5160
    },
    {
      "epoch": 0.8680514674964259,
      "grad_norm": NaN,
      "learning_rate": 4.373880764270566e-06,
      "loss": 0.0,
      "step": 5161
    },
    {
      "epoch": 0.8682196619291901,
      "grad_norm": NaN,
      "learning_rate": 4.362895092694757e-06,
      "loss": 0.0,
      "step": 5162
    },
    {
      "epoch": 0.8683878563619544,
      "grad_norm": NaN,
      "learning_rate": 4.351922605167386e-06,
      "loss": 0.0,
      "step": 5163
    },
    {
      "epoch": 0.8685560507947186,
      "grad_norm": NaN,
      "learning_rate": 4.340963304858281e-06,
      "loss": 0.0,
      "step": 5164
    },
    {
      "epoch": 0.8687242452274829,
      "grad_norm": NaN,
      "learning_rate": 4.330017194933478e-06,
      "loss": 0.0,
      "step": 5165
    },
    {
      "epoch": 0.8688924396602472,
      "grad_norm": NaN,
      "learning_rate": 4.319084278555169e-06,
      "loss": 0.0,
      "step": 5166
    },
    {
      "epoch": 0.8690606340930115,
      "grad_norm": NaN,
      "learning_rate": 4.308164558881761e-06,
      "loss": 0.0,
      "step": 5167
    },
    {
      "epoch": 0.8692288285257758,
      "grad_norm": NaN,
      "learning_rate": 4.297258039067831e-06,
      "loss": 0.0,
      "step": 5168
    },
    {
      "epoch": 0.86939702295854,
      "grad_norm": NaN,
      "learning_rate": 4.286364722264158e-06,
      "loss": 0.0,
      "step": 5169
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": NaN,
      "learning_rate": 4.275484611617691e-06,
      "loss": 0.0,
      "step": 5170
    },
    {
      "epoch": 0.8697334118240686,
      "grad_norm": NaN,
      "learning_rate": 4.264617710271579e-06,
      "loss": 0.0,
      "step": 5171
    },
    {
      "epoch": 0.8699016062568329,
      "grad_norm": NaN,
      "learning_rate": 4.253764021365131e-06,
      "loss": 0.0,
      "step": 5172
    },
    {
      "epoch": 0.8700698006895972,
      "grad_norm": NaN,
      "learning_rate": 4.242923548033878e-06,
      "loss": 0.0,
      "step": 5173
    },
    {
      "epoch": 0.8702379951223614,
      "grad_norm": NaN,
      "learning_rate": 4.23209629340951e-06,
      "loss": 0.0,
      "step": 5174
    },
    {
      "epoch": 0.8704061895551257,
      "grad_norm": NaN,
      "learning_rate": 4.221282260619891e-06,
      "loss": 0.0,
      "step": 5175
    },
    {
      "epoch": 0.87057438398789,
      "grad_norm": NaN,
      "learning_rate": 4.210481452789061e-06,
      "loss": 0.0,
      "step": 5176
    },
    {
      "epoch": 0.8707425784206543,
      "grad_norm": NaN,
      "learning_rate": 4.19969387303728e-06,
      "loss": 0.0,
      "step": 5177
    },
    {
      "epoch": 0.8709107728534186,
      "grad_norm": NaN,
      "learning_rate": 4.1889195244809495e-06,
      "loss": 0.0,
      "step": 5178
    },
    {
      "epoch": 0.8710789672861828,
      "grad_norm": NaN,
      "learning_rate": 4.178158410232652e-06,
      "loss": 0.0,
      "step": 5179
    },
    {
      "epoch": 0.8712471617189471,
      "grad_norm": NaN,
      "learning_rate": 4.167410533401151e-06,
      "loss": 0.0,
      "step": 5180
    },
    {
      "epoch": 0.8714153561517114,
      "grad_norm": NaN,
      "learning_rate": 4.1566758970913945e-06,
      "loss": 0.0,
      "step": 5181
    },
    {
      "epoch": 0.8715835505844757,
      "grad_norm": NaN,
      "learning_rate": 4.145954504404498e-06,
      "loss": 0.0,
      "step": 5182
    },
    {
      "epoch": 0.87175174501724,
      "grad_norm": NaN,
      "learning_rate": 4.1352463584377556e-06,
      "loss": 0.0,
      "step": 5183
    },
    {
      "epoch": 0.8719199394500042,
      "grad_norm": NaN,
      "learning_rate": 4.12455146228461e-06,
      "loss": 0.0,
      "step": 5184
    },
    {
      "epoch": 0.8720881338827685,
      "grad_norm": NaN,
      "learning_rate": 4.113869819034727e-06,
      "loss": 0.0,
      "step": 5185
    },
    {
      "epoch": 0.8722563283155328,
      "grad_norm": NaN,
      "learning_rate": 4.103201431773896e-06,
      "loss": 0.0,
      "step": 5186
    },
    {
      "epoch": 0.8724245227482971,
      "grad_norm": NaN,
      "learning_rate": 4.0925463035841e-06,
      "loss": 0.0,
      "step": 5187
    },
    {
      "epoch": 0.8725927171810614,
      "grad_norm": NaN,
      "learning_rate": 4.081904437543482e-06,
      "loss": 0.0,
      "step": 5188
    },
    {
      "epoch": 0.8727609116138256,
      "grad_norm": NaN,
      "learning_rate": 4.071275836726357e-06,
      "loss": 0.0,
      "step": 5189
    },
    {
      "epoch": 0.8729291060465899,
      "grad_norm": NaN,
      "learning_rate": 4.060660504203217e-06,
      "loss": 0.0,
      "step": 5190
    },
    {
      "epoch": 0.8730973004793542,
      "grad_norm": NaN,
      "learning_rate": 4.0500584430407056e-06,
      "loss": 0.0,
      "step": 5191
    },
    {
      "epoch": 0.8732654949121184,
      "grad_norm": NaN,
      "learning_rate": 4.039469656301631e-06,
      "loss": 0.0,
      "step": 5192
    },
    {
      "epoch": 0.8734336893448826,
      "grad_norm": NaN,
      "learning_rate": 4.028894147044992e-06,
      "loss": 0.0,
      "step": 5193
    },
    {
      "epoch": 0.8736018837776469,
      "grad_norm": NaN,
      "learning_rate": 4.018331918325935e-06,
      "loss": 0.0,
      "step": 5194
    },
    {
      "epoch": 0.8737700782104112,
      "grad_norm": NaN,
      "learning_rate": 4.007782973195756e-06,
      "loss": 0.0,
      "step": 5195
    },
    {
      "epoch": 0.8739382726431755,
      "grad_norm": NaN,
      "learning_rate": 3.997247314701935e-06,
      "loss": 0.0,
      "step": 5196
    },
    {
      "epoch": 0.8741064670759398,
      "grad_norm": NaN,
      "learning_rate": 3.9867249458880984e-06,
      "loss": 0.0,
      "step": 5197
    },
    {
      "epoch": 0.874274661508704,
      "grad_norm": NaN,
      "learning_rate": 3.976215869794054e-06,
      "loss": 0.0,
      "step": 5198
    },
    {
      "epoch": 0.8744428559414683,
      "grad_norm": NaN,
      "learning_rate": 3.965720089455749e-06,
      "loss": 0.0,
      "step": 5199
    },
    {
      "epoch": 0.8746110503742326,
      "grad_norm": NaN,
      "learning_rate": 3.955237607905299e-06,
      "loss": 0.0,
      "step": 5200
    },
    {
      "epoch": 0.8747792448069969,
      "grad_norm": NaN,
      "learning_rate": 3.9447684281709585e-06,
      "loss": 0.0,
      "step": 5201
    },
    {
      "epoch": 0.8749474392397611,
      "grad_norm": NaN,
      "learning_rate": 3.934312553277186e-06,
      "loss": 0.0,
      "step": 5202
    },
    {
      "epoch": 0.8751156336725254,
      "grad_norm": NaN,
      "learning_rate": 3.923869986244549e-06,
      "loss": 0.0,
      "step": 5203
    },
    {
      "epoch": 0.8752838281052897,
      "grad_norm": NaN,
      "learning_rate": 3.91344073008979e-06,
      "loss": 0.0,
      "step": 5204
    },
    {
      "epoch": 0.875452022538054,
      "grad_norm": NaN,
      "learning_rate": 3.9030247878258046e-06,
      "loss": 0.0,
      "step": 5205
    },
    {
      "epoch": 0.8756202169708183,
      "grad_norm": NaN,
      "learning_rate": 3.89262216246164e-06,
      "loss": 0.0,
      "step": 5206
    },
    {
      "epoch": 0.8757884114035825,
      "grad_norm": NaN,
      "learning_rate": 3.882232857002505e-06,
      "loss": 0.0,
      "step": 5207
    },
    {
      "epoch": 0.8759566058363468,
      "grad_norm": NaN,
      "learning_rate": 3.871856874449748e-06,
      "loss": 0.0,
      "step": 5208
    },
    {
      "epoch": 0.8761248002691111,
      "grad_norm": NaN,
      "learning_rate": 3.861494217800865e-06,
      "loss": 0.0,
      "step": 5209
    },
    {
      "epoch": 0.8762929947018754,
      "grad_norm": NaN,
      "learning_rate": 3.851144890049535e-06,
      "loss": 0.0,
      "step": 5210
    },
    {
      "epoch": 0.8764611891346397,
      "grad_norm": NaN,
      "learning_rate": 3.840808894185543e-06,
      "loss": 0.0,
      "step": 5211
    },
    {
      "epoch": 0.8766293835674039,
      "grad_norm": NaN,
      "learning_rate": 3.830486233194852e-06,
      "loss": 0.0,
      "step": 5212
    },
    {
      "epoch": 0.8767975780001682,
      "grad_norm": NaN,
      "learning_rate": 3.820176910059553e-06,
      "loss": 0.0,
      "step": 5213
    },
    {
      "epoch": 0.8769657724329325,
      "grad_norm": NaN,
      "learning_rate": 3.8098809277579027e-06,
      "loss": 0.0,
      "step": 5214
    },
    {
      "epoch": 0.8771339668656968,
      "grad_norm": NaN,
      "learning_rate": 3.7995982892643e-06,
      "loss": 0.0,
      "step": 5215
    },
    {
      "epoch": 0.8773021612984611,
      "grad_norm": NaN,
      "learning_rate": 3.789328997549274e-06,
      "loss": 0.0,
      "step": 5216
    },
    {
      "epoch": 0.8774703557312253,
      "grad_norm": NaN,
      "learning_rate": 3.7790730555795075e-06,
      "loss": 0.0,
      "step": 5217
    },
    {
      "epoch": 0.8776385501639896,
      "grad_norm": NaN,
      "learning_rate": 3.768830466317835e-06,
      "loss": 0.0,
      "step": 5218
    },
    {
      "epoch": 0.8778067445967539,
      "grad_norm": NaN,
      "learning_rate": 3.7586012327232244e-06,
      "loss": 0.0,
      "step": 5219
    },
    {
      "epoch": 0.8779749390295182,
      "grad_norm": NaN,
      "learning_rate": 3.7483853577507834e-06,
      "loss": 0.0,
      "step": 5220
    },
    {
      "epoch": 0.8781431334622823,
      "grad_norm": NaN,
      "learning_rate": 3.7381828443517563e-06,
      "loss": 0.0,
      "step": 5221
    },
    {
      "epoch": 0.8783113278950466,
      "grad_norm": NaN,
      "learning_rate": 3.7279936954735526e-06,
      "loss": 0.0,
      "step": 5222
    },
    {
      "epoch": 0.8784795223278109,
      "grad_norm": NaN,
      "learning_rate": 3.7178179140596957e-06,
      "loss": 0.0,
      "step": 5223
    },
    {
      "epoch": 0.8786477167605752,
      "grad_norm": NaN,
      "learning_rate": 3.7076555030498506e-06,
      "loss": 0.0,
      "step": 5224
    },
    {
      "epoch": 0.8788159111933395,
      "grad_norm": NaN,
      "learning_rate": 3.6975064653798297e-06,
      "loss": 0.0,
      "step": 5225
    },
    {
      "epoch": 0.8789841056261037,
      "grad_norm": NaN,
      "learning_rate": 3.687370803981571e-06,
      "loss": 0.0,
      "step": 5226
    },
    {
      "epoch": 0.879152300058868,
      "grad_norm": NaN,
      "learning_rate": 3.6772485217831555e-06,
      "loss": 0.0,
      "step": 5227
    },
    {
      "epoch": 0.8793204944916323,
      "grad_norm": NaN,
      "learning_rate": 3.6671396217087985e-06,
      "loss": 0.0,
      "step": 5228
    },
    {
      "epoch": 0.8794886889243966,
      "grad_norm": NaN,
      "learning_rate": 3.6570441066788363e-06,
      "loss": 0.0,
      "step": 5229
    },
    {
      "epoch": 0.8796568833571609,
      "grad_norm": NaN,
      "learning_rate": 3.64696197960977e-06,
      "loss": 0.0,
      "step": 5230
    },
    {
      "epoch": 0.8798250777899251,
      "grad_norm": NaN,
      "learning_rate": 3.6368932434142076e-06,
      "loss": 0.0,
      "step": 5231
    },
    {
      "epoch": 0.8799932722226894,
      "grad_norm": NaN,
      "learning_rate": 3.6268379010008846e-06,
      "loss": 0.0,
      "step": 5232
    },
    {
      "epoch": 0.8801614666554537,
      "grad_norm": NaN,
      "learning_rate": 3.6167959552746768e-06,
      "loss": 0.0,
      "step": 5233
    },
    {
      "epoch": 0.880329661088218,
      "grad_norm": NaN,
      "learning_rate": 3.6067674091366075e-06,
      "loss": 0.0,
      "step": 5234
    },
    {
      "epoch": 0.8804978555209823,
      "grad_norm": NaN,
      "learning_rate": 3.5967522654837994e-06,
      "loss": 0.0,
      "step": 5235
    },
    {
      "epoch": 0.8806660499537465,
      "grad_norm": NaN,
      "learning_rate": 3.5867505272095093e-06,
      "loss": 0.0,
      "step": 5236
    },
    {
      "epoch": 0.8808342443865108,
      "grad_norm": NaN,
      "learning_rate": 3.576762197203132e-06,
      "loss": 0.0,
      "step": 5237
    },
    {
      "epoch": 0.8810024388192751,
      "grad_norm": NaN,
      "learning_rate": 3.566787278350192e-06,
      "loss": 0.0,
      "step": 5238
    },
    {
      "epoch": 0.8811706332520394,
      "grad_norm": NaN,
      "learning_rate": 3.556825773532324e-06,
      "loss": 0.0,
      "step": 5239
    },
    {
      "epoch": 0.8813388276848036,
      "grad_norm": NaN,
      "learning_rate": 3.546877685627298e-06,
      "loss": 0.0,
      "step": 5240
    },
    {
      "epoch": 0.8815070221175679,
      "grad_norm": NaN,
      "learning_rate": 3.5369430175090033e-06,
      "loss": 0.0,
      "step": 5241
    },
    {
      "epoch": 0.8816752165503322,
      "grad_norm": NaN,
      "learning_rate": 3.5270217720474564e-06,
      "loss": 0.0,
      "step": 5242
    },
    {
      "epoch": 0.8818434109830965,
      "grad_norm": NaN,
      "learning_rate": 3.5171139521087916e-06,
      "loss": 0.0,
      "step": 5243
    },
    {
      "epoch": 0.8820116054158608,
      "grad_norm": NaN,
      "learning_rate": 3.5072195605552694e-06,
      "loss": 0.0,
      "step": 5244
    },
    {
      "epoch": 0.882179799848625,
      "grad_norm": NaN,
      "learning_rate": 3.4973386002452535e-06,
      "loss": 0.0,
      "step": 5245
    },
    {
      "epoch": 0.8823479942813893,
      "grad_norm": NaN,
      "learning_rate": 3.487471074033272e-06,
      "loss": 0.0,
      "step": 5246
    },
    {
      "epoch": 0.8825161887141536,
      "grad_norm": NaN,
      "learning_rate": 3.4776169847699213e-06,
      "loss": 0.0,
      "step": 5247
    },
    {
      "epoch": 0.8826843831469179,
      "grad_norm": NaN,
      "learning_rate": 3.467776335301942e-06,
      "loss": 0.0,
      "step": 5248
    },
    {
      "epoch": 0.8828525775796822,
      "grad_norm": NaN,
      "learning_rate": 3.457949128472193e-06,
      "loss": 0.0,
      "step": 5249
    },
    {
      "epoch": 0.8830207720124464,
      "grad_norm": NaN,
      "learning_rate": 3.4481353671196313e-06,
      "loss": 0.0,
      "step": 5250
    },
    {
      "epoch": 0.8831889664452106,
      "grad_norm": NaN,
      "learning_rate": 3.438335054079361e-06,
      "loss": 0.0,
      "step": 5251
    },
    {
      "epoch": 0.8833571608779749,
      "grad_norm": NaN,
      "learning_rate": 3.428548192182568e-06,
      "loss": 0.0,
      "step": 5252
    },
    {
      "epoch": 0.8835253553107392,
      "grad_norm": NaN,
      "learning_rate": 3.4187747842565733e-06,
      "loss": 0.0,
      "step": 5253
    },
    {
      "epoch": 0.8836935497435034,
      "grad_norm": NaN,
      "learning_rate": 3.4090148331248016e-06,
      "loss": 0.0,
      "step": 5254
    },
    {
      "epoch": 0.8838617441762677,
      "grad_norm": NaN,
      "learning_rate": 3.3992683416067984e-06,
      "loss": 0.0,
      "step": 5255
    },
    {
      "epoch": 0.884029938609032,
      "grad_norm": NaN,
      "learning_rate": 3.3895353125182106e-06,
      "loss": 0.0,
      "step": 5256
    },
    {
      "epoch": 0.8841981330417963,
      "grad_norm": NaN,
      "learning_rate": 3.3798157486708113e-06,
      "loss": 0.0,
      "step": 5257
    },
    {
      "epoch": 0.8843663274745606,
      "grad_norm": NaN,
      "learning_rate": 3.3701096528724494e-06,
      "loss": 0.0,
      "step": 5258
    },
    {
      "epoch": 0.8845345219073248,
      "grad_norm": NaN,
      "learning_rate": 3.3604170279271374e-06,
      "loss": 0.0,
      "step": 5259
    },
    {
      "epoch": 0.8847027163400891,
      "grad_norm": NaN,
      "learning_rate": 3.350737876634957e-06,
      "loss": 0.0,
      "step": 5260
    },
    {
      "epoch": 0.8848709107728534,
      "grad_norm": NaN,
      "learning_rate": 3.3410722017920947e-06,
      "loss": 0.0,
      "step": 5261
    },
    {
      "epoch": 0.8850391052056177,
      "grad_norm": NaN,
      "learning_rate": 3.331420006190866e-06,
      "loss": 0.0,
      "step": 5262
    },
    {
      "epoch": 0.885207299638382,
      "grad_norm": NaN,
      "learning_rate": 3.32178129261968e-06,
      "loss": 0.0,
      "step": 5263
    },
    {
      "epoch": 0.8853754940711462,
      "grad_norm": NaN,
      "learning_rate": 3.3121560638630535e-06,
      "loss": 0.0,
      "step": 5264
    },
    {
      "epoch": 0.8855436885039105,
      "grad_norm": NaN,
      "learning_rate": 3.302544322701606e-06,
      "loss": 0.0,
      "step": 5265
    },
    {
      "epoch": 0.8857118829366748,
      "grad_norm": NaN,
      "learning_rate": 3.292946071912051e-06,
      "loss": 0.0,
      "step": 5266
    },
    {
      "epoch": 0.8858800773694391,
      "grad_norm": NaN,
      "learning_rate": 3.2833613142672358e-06,
      "loss": 0.0,
      "step": 5267
    },
    {
      "epoch": 0.8860482718022034,
      "grad_norm": NaN,
      "learning_rate": 3.2737900525360853e-06,
      "loss": 0.0,
      "step": 5268
    },
    {
      "epoch": 0.8862164662349676,
      "grad_norm": NaN,
      "learning_rate": 3.264232289483621e-06,
      "loss": 0.0,
      "step": 5269
    },
    {
      "epoch": 0.8863846606677319,
      "grad_norm": NaN,
      "learning_rate": 3.2546880278709725e-06,
      "loss": 0.0,
      "step": 5270
    },
    {
      "epoch": 0.8865528551004962,
      "grad_norm": NaN,
      "learning_rate": 3.245157270455379e-06,
      "loss": 0.0,
      "step": 5271
    },
    {
      "epoch": 0.8867210495332605,
      "grad_norm": NaN,
      "learning_rate": 3.235640019990166e-06,
      "loss": 0.0,
      "step": 5272
    },
    {
      "epoch": 0.8868892439660248,
      "grad_norm": NaN,
      "learning_rate": 3.226136279224762e-06,
      "loss": 0.0,
      "step": 5273
    },
    {
      "epoch": 0.887057438398789,
      "grad_norm": NaN,
      "learning_rate": 3.2166460509046814e-06,
      "loss": 0.0,
      "step": 5274
    },
    {
      "epoch": 0.8872256328315533,
      "grad_norm": NaN,
      "learning_rate": 3.207169337771565e-06,
      "loss": 0.0,
      "step": 5275
    },
    {
      "epoch": 0.8873938272643176,
      "grad_norm": NaN,
      "learning_rate": 3.1977061425631117e-06,
      "loss": 0.0,
      "step": 5276
    },
    {
      "epoch": 0.8875620216970819,
      "grad_norm": NaN,
      "learning_rate": 3.18825646801314e-06,
      "loss": 0.0,
      "step": 5277
    },
    {
      "epoch": 0.8877302161298462,
      "grad_norm": NaN,
      "learning_rate": 3.1788203168515497e-06,
      "loss": 0.0,
      "step": 5278
    },
    {
      "epoch": 0.8878984105626104,
      "grad_norm": NaN,
      "learning_rate": 3.169397691804343e-06,
      "loss": 0.0,
      "step": 5279
    },
    {
      "epoch": 0.8880666049953746,
      "grad_norm": NaN,
      "learning_rate": 3.159988595593616e-06,
      "loss": 0.0,
      "step": 5280
    },
    {
      "epoch": 0.8882347994281389,
      "grad_norm": NaN,
      "learning_rate": 3.150593030937543e-06,
      "loss": 0.0,
      "step": 5281
    },
    {
      "epoch": 0.8884029938609032,
      "grad_norm": NaN,
      "learning_rate": 3.141211000550398e-06,
      "loss": 0.0,
      "step": 5282
    },
    {
      "epoch": 0.8885711882936674,
      "grad_norm": NaN,
      "learning_rate": 3.1318425071425463e-06,
      "loss": 0.0,
      "step": 5283
    },
    {
      "epoch": 0.8887393827264317,
      "grad_norm": NaN,
      "learning_rate": 3.1224875534204445e-06,
      "loss": 0.0,
      "step": 5284
    },
    {
      "epoch": 0.888907577159196,
      "grad_norm": NaN,
      "learning_rate": 3.1131461420866313e-06,
      "loss": 0.0,
      "step": 5285
    },
    {
      "epoch": 0.8890757715919603,
      "grad_norm": NaN,
      "learning_rate": 3.1038182758397207e-06,
      "loss": 0.0,
      "step": 5286
    },
    {
      "epoch": 0.8892439660247246,
      "grad_norm": NaN,
      "learning_rate": 3.094503957374456e-06,
      "loss": 0.0,
      "step": 5287
    },
    {
      "epoch": 0.8894121604574888,
      "grad_norm": NaN,
      "learning_rate": 3.0852031893816245e-06,
      "loss": 0.0,
      "step": 5288
    },
    {
      "epoch": 0.8895803548902531,
      "grad_norm": NaN,
      "learning_rate": 3.075915974548116e-06,
      "loss": 0.0,
      "step": 5289
    },
    {
      "epoch": 0.8897485493230174,
      "grad_norm": NaN,
      "learning_rate": 3.066642315556895e-06,
      "loss": 0.0,
      "step": 5290
    },
    {
      "epoch": 0.8899167437557817,
      "grad_norm": NaN,
      "learning_rate": 3.0573822150870344e-06,
      "loss": 0.0,
      "step": 5291
    },
    {
      "epoch": 0.890084938188546,
      "grad_norm": NaN,
      "learning_rate": 3.0481356758136624e-06,
      "loss": 0.0,
      "step": 5292
    },
    {
      "epoch": 0.8902531326213102,
      "grad_norm": NaN,
      "learning_rate": 3.038902700408003e-06,
      "loss": 0.0,
      "step": 5293
    },
    {
      "epoch": 0.8904213270540745,
      "grad_norm": NaN,
      "learning_rate": 3.0296832915373497e-06,
      "loss": 0.0,
      "step": 5294
    },
    {
      "epoch": 0.8905895214868388,
      "grad_norm": NaN,
      "learning_rate": 3.020477451865106e-06,
      "loss": 0.0,
      "step": 5295
    },
    {
      "epoch": 0.8907577159196031,
      "grad_norm": NaN,
      "learning_rate": 3.0112851840507215e-06,
      "loss": 0.0,
      "step": 5296
    },
    {
      "epoch": 0.8909259103523673,
      "grad_norm": NaN,
      "learning_rate": 3.002106490749751e-06,
      "loss": 0.0,
      "step": 5297
    },
    {
      "epoch": 0.8910941047851316,
      "grad_norm": NaN,
      "learning_rate": 2.992941374613806e-06,
      "loss": 0.0,
      "step": 5298
    },
    {
      "epoch": 0.8912622992178959,
      "grad_norm": NaN,
      "learning_rate": 2.9837898382905847e-06,
      "loss": 0.0,
      "step": 5299
    },
    {
      "epoch": 0.8914304936506602,
      "grad_norm": NaN,
      "learning_rate": 2.974651884423868e-06,
      "loss": 0.0,
      "step": 5300
    },
    {
      "epoch": 0.8915986880834245,
      "grad_norm": NaN,
      "learning_rate": 2.9655275156535103e-06,
      "loss": 0.0,
      "step": 5301
    },
    {
      "epoch": 0.8917668825161887,
      "grad_norm": NaN,
      "learning_rate": 2.956416734615436e-06,
      "loss": 0.0,
      "step": 5302
    },
    {
      "epoch": 0.891935076948953,
      "grad_norm": NaN,
      "learning_rate": 2.9473195439416457e-06,
      "loss": 0.0,
      "step": 5303
    },
    {
      "epoch": 0.8921032713817173,
      "grad_norm": NaN,
      "learning_rate": 2.9382359462602206e-06,
      "loss": 0.0,
      "step": 5304
    },
    {
      "epoch": 0.8922714658144816,
      "grad_norm": NaN,
      "learning_rate": 2.9291659441953114e-06,
      "loss": 0.0,
      "step": 5305
    },
    {
      "epoch": 0.8924396602472459,
      "grad_norm": NaN,
      "learning_rate": 2.920109540367133e-06,
      "loss": 0.0,
      "step": 5306
    },
    {
      "epoch": 0.8926078546800101,
      "grad_norm": NaN,
      "learning_rate": 2.911066737391982e-06,
      "loss": 0.0,
      "step": 5307
    },
    {
      "epoch": 0.8927760491127744,
      "grad_norm": NaN,
      "learning_rate": 2.9020375378822297e-06,
      "loss": 0.0,
      "step": 5308
    },
    {
      "epoch": 0.8929442435455387,
      "grad_norm": NaN,
      "learning_rate": 2.8930219444463005e-06,
      "loss": 0.0,
      "step": 5309
    },
    {
      "epoch": 0.8931124379783029,
      "grad_norm": NaN,
      "learning_rate": 2.8840199596887105e-06,
      "loss": 0.0,
      "step": 5310
    },
    {
      "epoch": 0.8932806324110671,
      "grad_norm": NaN,
      "learning_rate": 2.8750315862100142e-06,
      "loss": 0.0,
      "step": 5311
    },
    {
      "epoch": 0.8934488268438314,
      "grad_norm": NaN,
      "learning_rate": 2.8660568266068723e-06,
      "loss": 0.0,
      "step": 5312
    },
    {
      "epoch": 0.8936170212765957,
      "grad_norm": NaN,
      "learning_rate": 2.8570956834719776e-06,
      "loss": 0.0,
      "step": 5313
    },
    {
      "epoch": 0.89378521570936,
      "grad_norm": NaN,
      "learning_rate": 2.848148159394115e-06,
      "loss": 0.0,
      "step": 5314
    },
    {
      "epoch": 0.8939534101421243,
      "grad_norm": NaN,
      "learning_rate": 2.839214256958106e-06,
      "loss": 0.0,
      "step": 5315
    },
    {
      "epoch": 0.8941216045748885,
      "grad_norm": NaN,
      "learning_rate": 2.8302939787448746e-06,
      "loss": 0.0,
      "step": 5316
    },
    {
      "epoch": 0.8942897990076528,
      "grad_norm": NaN,
      "learning_rate": 2.8213873273313873e-06,
      "loss": 0.0,
      "step": 5317
    },
    {
      "epoch": 0.8944579934404171,
      "grad_norm": NaN,
      "learning_rate": 2.8124943052906638e-06,
      "loss": 0.0,
      "step": 5318
    },
    {
      "epoch": 0.8946261878731814,
      "grad_norm": NaN,
      "learning_rate": 2.80361491519181e-06,
      "loss": 0.0,
      "step": 5319
    },
    {
      "epoch": 0.8947943823059457,
      "grad_norm": NaN,
      "learning_rate": 2.794749159599974e-06,
      "loss": 0.0,
      "step": 5320
    },
    {
      "epoch": 0.8949625767387099,
      "grad_norm": NaN,
      "learning_rate": 2.7858970410763795e-06,
      "loss": 0.0,
      "step": 5321
    },
    {
      "epoch": 0.8951307711714742,
      "grad_norm": NaN,
      "learning_rate": 2.7770585621782973e-06,
      "loss": 0.0,
      "step": 5322
    },
    {
      "epoch": 0.8952989656042385,
      "grad_norm": NaN,
      "learning_rate": 2.7682337254590684e-06,
      "loss": 0.0,
      "step": 5323
    },
    {
      "epoch": 0.8954671600370028,
      "grad_norm": NaN,
      "learning_rate": 2.759422533468092e-06,
      "loss": 0.0,
      "step": 5324
    },
    {
      "epoch": 0.895635354469767,
      "grad_norm": NaN,
      "learning_rate": 2.750624988750822e-06,
      "loss": 0.0,
      "step": 5325
    },
    {
      "epoch": 0.8958035489025313,
      "grad_norm": NaN,
      "learning_rate": 2.7418410938487736e-06,
      "loss": 0.0,
      "step": 5326
    },
    {
      "epoch": 0.8959717433352956,
      "grad_norm": NaN,
      "learning_rate": 2.7330708512994964e-06,
      "loss": 0.0,
      "step": 5327
    },
    {
      "epoch": 0.8961399377680599,
      "grad_norm": NaN,
      "learning_rate": 2.7243142636366457e-06,
      "loss": 0.0,
      "step": 5328
    },
    {
      "epoch": 0.8963081322008242,
      "grad_norm": NaN,
      "learning_rate": 2.7155713333898825e-06,
      "loss": 0.0,
      "step": 5329
    },
    {
      "epoch": 0.8964763266335884,
      "grad_norm": NaN,
      "learning_rate": 2.706842063084941e-06,
      "loss": 0.0,
      "step": 5330
    },
    {
      "epoch": 0.8966445210663527,
      "grad_norm": NaN,
      "learning_rate": 2.6981264552436105e-06,
      "loss": 0.0,
      "step": 5331
    },
    {
      "epoch": 0.896812715499117,
      "grad_norm": NaN,
      "learning_rate": 2.689424512383748e-06,
      "loss": 0.0,
      "step": 5332
    },
    {
      "epoch": 0.8969809099318813,
      "grad_norm": NaN,
      "learning_rate": 2.680736237019227e-06,
      "loss": 0.0,
      "step": 5333
    },
    {
      "epoch": 0.8971491043646456,
      "grad_norm": NaN,
      "learning_rate": 2.6720616316600056e-06,
      "loss": 0.0,
      "step": 5334
    },
    {
      "epoch": 0.8973172987974098,
      "grad_norm": NaN,
      "learning_rate": 2.663400698812074e-06,
      "loss": 0.0,
      "step": 5335
    },
    {
      "epoch": 0.8974854932301741,
      "grad_norm": NaN,
      "learning_rate": 2.654753440977481e-06,
      "loss": 0.0,
      "step": 5336
    },
    {
      "epoch": 0.8976536876629384,
      "grad_norm": NaN,
      "learning_rate": 2.646119860654317e-06,
      "loss": 0.0,
      "step": 5337
    },
    {
      "epoch": 0.8978218820957027,
      "grad_norm": NaN,
      "learning_rate": 2.6374999603367367e-06,
      "loss": 0.0,
      "step": 5338
    },
    {
      "epoch": 0.897990076528467,
      "grad_norm": NaN,
      "learning_rate": 2.6288937425149205e-06,
      "loss": 0.0,
      "step": 5339
    },
    {
      "epoch": 0.8981582709612311,
      "grad_norm": NaN,
      "learning_rate": 2.620301209675119e-06,
      "loss": 0.0,
      "step": 5340
    },
    {
      "epoch": 0.8983264653939954,
      "grad_norm": NaN,
      "learning_rate": 2.611722364299618e-06,
      "loss": 0.0,
      "step": 5341
    },
    {
      "epoch": 0.8984946598267597,
      "grad_norm": NaN,
      "learning_rate": 2.6031572088667465e-06,
      "loss": 0.0,
      "step": 5342
    },
    {
      "epoch": 0.898662854259524,
      "grad_norm": NaN,
      "learning_rate": 2.5946057458508756e-06,
      "loss": 0.0,
      "step": 5343
    },
    {
      "epoch": 0.8988310486922882,
      "grad_norm": NaN,
      "learning_rate": 2.5860679777224394e-06,
      "loss": 0.0,
      "step": 5344
    },
    {
      "epoch": 0.8989992431250525,
      "grad_norm": NaN,
      "learning_rate": 2.577543906947899e-06,
      "loss": 0.0,
      "step": 5345
    },
    {
      "epoch": 0.8991674375578168,
      "grad_norm": NaN,
      "learning_rate": 2.5690335359897564e-06,
      "loss": 0.0,
      "step": 5346
    },
    {
      "epoch": 0.8993356319905811,
      "grad_norm": NaN,
      "learning_rate": 2.5605368673065733e-06,
      "loss": 0.0,
      "step": 5347
    },
    {
      "epoch": 0.8995038264233454,
      "grad_norm": NaN,
      "learning_rate": 2.55205390335293e-06,
      "loss": 0.0,
      "step": 5348
    },
    {
      "epoch": 0.8996720208561096,
      "grad_norm": NaN,
      "learning_rate": 2.5435846465794723e-06,
      "loss": 0.0,
      "step": 5349
    },
    {
      "epoch": 0.8998402152888739,
      "grad_norm": NaN,
      "learning_rate": 2.53512909943287e-06,
      "loss": 0.0,
      "step": 5350
    },
    {
      "epoch": 0.9000084097216382,
      "grad_norm": NaN,
      "learning_rate": 2.5266872643558316e-06,
      "loss": 0.0,
      "step": 5351
    },
    {
      "epoch": 0.9001766041544025,
      "grad_norm": NaN,
      "learning_rate": 2.518259143787105e-06,
      "loss": 0.0,
      "step": 5352
    },
    {
      "epoch": 0.9003447985871668,
      "grad_norm": NaN,
      "learning_rate": 2.5098447401614934e-06,
      "loss": 0.0,
      "step": 5353
    },
    {
      "epoch": 0.900512993019931,
      "grad_norm": NaN,
      "learning_rate": 2.501444055909813e-06,
      "loss": 0.0,
      "step": 5354
    },
    {
      "epoch": 0.9006811874526953,
      "grad_norm": NaN,
      "learning_rate": 2.493057093458934e-06,
      "loss": 0.0,
      "step": 5355
    },
    {
      "epoch": 0.9008493818854596,
      "grad_norm": NaN,
      "learning_rate": 2.484683855231751e-06,
      "loss": 0.0,
      "step": 5356
    },
    {
      "epoch": 0.9010175763182239,
      "grad_norm": NaN,
      "learning_rate": 2.4763243436472016e-06,
      "loss": 0.0,
      "step": 5357
    },
    {
      "epoch": 0.9011857707509882,
      "grad_norm": NaN,
      "learning_rate": 2.467978561120249e-06,
      "loss": 0.0,
      "step": 5358
    },
    {
      "epoch": 0.9013539651837524,
      "grad_norm": NaN,
      "learning_rate": 2.459646510061908e-06,
      "loss": 0.0,
      "step": 5359
    },
    {
      "epoch": 0.9015221596165167,
      "grad_norm": NaN,
      "learning_rate": 2.4513281928791985e-06,
      "loss": 0.0,
      "step": 5360
    },
    {
      "epoch": 0.901690354049281,
      "grad_norm": NaN,
      "learning_rate": 2.443023611975204e-06,
      "loss": 0.0,
      "step": 5361
    },
    {
      "epoch": 0.9018585484820453,
      "grad_norm": NaN,
      "learning_rate": 2.434732769749015e-06,
      "loss": 0.0,
      "step": 5362
    },
    {
      "epoch": 0.9020267429148096,
      "grad_norm": NaN,
      "learning_rate": 2.426455668595773e-06,
      "loss": 0.0,
      "step": 5363
    },
    {
      "epoch": 0.9021949373475738,
      "grad_norm": NaN,
      "learning_rate": 2.418192310906625e-06,
      "loss": 0.0,
      "step": 5364
    },
    {
      "epoch": 0.9023631317803381,
      "grad_norm": NaN,
      "learning_rate": 2.409942699068779e-06,
      "loss": 0.0,
      "step": 5365
    },
    {
      "epoch": 0.9025313262131024,
      "grad_norm": NaN,
      "learning_rate": 2.4017068354654503e-06,
      "loss": 0.0,
      "step": 5366
    },
    {
      "epoch": 0.9026995206458667,
      "grad_norm": NaN,
      "learning_rate": 2.3934847224758804e-06,
      "loss": 0.0,
      "step": 5367
    },
    {
      "epoch": 0.902867715078631,
      "grad_norm": NaN,
      "learning_rate": 2.385276362475347e-06,
      "loss": 0.0,
      "step": 5368
    },
    {
      "epoch": 0.9030359095113951,
      "grad_norm": NaN,
      "learning_rate": 2.3770817578351646e-06,
      "loss": 0.0,
      "step": 5369
    },
    {
      "epoch": 0.9032041039441594,
      "grad_norm": NaN,
      "learning_rate": 2.3689009109226556e-06,
      "loss": 0.0,
      "step": 5370
    },
    {
      "epoch": 0.9033722983769237,
      "grad_norm": NaN,
      "learning_rate": 2.3607338241011747e-06,
      "loss": 0.0,
      "step": 5371
    },
    {
      "epoch": 0.903540492809688,
      "grad_norm": NaN,
      "learning_rate": 2.3525804997300893e-06,
      "loss": 0.0,
      "step": 5372
    },
    {
      "epoch": 0.9037086872424522,
      "grad_norm": NaN,
      "learning_rate": 2.344440940164827e-06,
      "loss": 0.0,
      "step": 5373
    },
    {
      "epoch": 0.9038768816752165,
      "grad_norm": NaN,
      "learning_rate": 2.336315147756807e-06,
      "loss": 0.0,
      "step": 5374
    },
    {
      "epoch": 0.9040450761079808,
      "grad_norm": NaN,
      "learning_rate": 2.328203124853473e-06,
      "loss": 0.0,
      "step": 5375
    },
    {
      "epoch": 0.9042132705407451,
      "grad_norm": NaN,
      "learning_rate": 2.3201048737983013e-06,
      "loss": 0.0,
      "step": 5376
    },
    {
      "epoch": 0.9043814649735094,
      "grad_norm": NaN,
      "learning_rate": 2.3120203969307862e-06,
      "loss": 0.0,
      "step": 5377
    },
    {
      "epoch": 0.9045496594062736,
      "grad_norm": NaN,
      "learning_rate": 2.3039496965864436e-06,
      "loss": 0.0,
      "step": 5378
    },
    {
      "epoch": 0.9047178538390379,
      "grad_norm": NaN,
      "learning_rate": 2.2958927750968083e-06,
      "loss": 0.0,
      "step": 5379
    },
    {
      "epoch": 0.9048860482718022,
      "grad_norm": NaN,
      "learning_rate": 2.287849634789424e-06,
      "loss": 0.0,
      "step": 5380
    },
    {
      "epoch": 0.9050542427045665,
      "grad_norm": NaN,
      "learning_rate": 2.2798202779878818e-06,
      "loss": 0.0,
      "step": 5381
    },
    {
      "epoch": 0.9052224371373307,
      "grad_norm": NaN,
      "learning_rate": 2.2718047070117655e-06,
      "loss": 0.0,
      "step": 5382
    },
    {
      "epoch": 0.905390631570095,
      "grad_norm": NaN,
      "learning_rate": 2.2638029241766833e-06,
      "loss": 0.0,
      "step": 5383
    },
    {
      "epoch": 0.9055588260028593,
      "grad_norm": NaN,
      "learning_rate": 2.2558149317942536e-06,
      "loss": 0.0,
      "step": 5384
    },
    {
      "epoch": 0.9057270204356236,
      "grad_norm": NaN,
      "learning_rate": 2.2478407321721296e-06,
      "loss": 0.0,
      "step": 5385
    },
    {
      "epoch": 0.9058952148683879,
      "grad_norm": NaN,
      "learning_rate": 2.2398803276139636e-06,
      "loss": 0.0,
      "step": 5386
    },
    {
      "epoch": 0.9060634093011521,
      "grad_norm": NaN,
      "learning_rate": 2.2319337204194267e-06,
      "loss": 0.0,
      "step": 5387
    },
    {
      "epoch": 0.9062316037339164,
      "grad_norm": NaN,
      "learning_rate": 2.2240009128842e-06,
      "loss": 0.0,
      "step": 5388
    },
    {
      "epoch": 0.9063997981666807,
      "grad_norm": NaN,
      "learning_rate": 2.2160819072999885e-06,
      "loss": 0.0,
      "step": 5389
    },
    {
      "epoch": 0.906567992599445,
      "grad_norm": NaN,
      "learning_rate": 2.208176705954512e-06,
      "loss": 0.0,
      "step": 5390
    },
    {
      "epoch": 0.9067361870322093,
      "grad_norm": NaN,
      "learning_rate": 2.2002853111314783e-06,
      "loss": 0.0,
      "step": 5391
    },
    {
      "epoch": 0.9069043814649735,
      "grad_norm": NaN,
      "learning_rate": 2.1924077251106347e-06,
      "loss": 0.0,
      "step": 5392
    },
    {
      "epoch": 0.9070725758977378,
      "grad_norm": NaN,
      "learning_rate": 2.1845439501677222e-06,
      "loss": 0.0,
      "step": 5393
    },
    {
      "epoch": 0.9072407703305021,
      "grad_norm": NaN,
      "learning_rate": 2.176693988574502e-06,
      "loss": 0.0,
      "step": 5394
    },
    {
      "epoch": 0.9074089647632664,
      "grad_norm": NaN,
      "learning_rate": 2.168857842598737e-06,
      "loss": 0.0,
      "step": 5395
    },
    {
      "epoch": 0.9075771591960307,
      "grad_norm": NaN,
      "learning_rate": 2.161035514504195e-06,
      "loss": 0.0,
      "step": 5396
    },
    {
      "epoch": 0.9077453536287949,
      "grad_norm": NaN,
      "learning_rate": 2.1532270065506675e-06,
      "loss": 0.0,
      "step": 5397
    },
    {
      "epoch": 0.9079135480615592,
      "grad_norm": NaN,
      "learning_rate": 2.1454323209939455e-06,
      "loss": 0.0,
      "step": 5398
    },
    {
      "epoch": 0.9080817424943234,
      "grad_norm": NaN,
      "learning_rate": 2.137651460085821e-06,
      "loss": 0.0,
      "step": 5399
    },
    {
      "epoch": 0.9082499369270877,
      "grad_norm": NaN,
      "learning_rate": 2.129884426074108e-06,
      "loss": 0.0,
      "step": 5400
    },
    {
      "epoch": 0.9084181313598519,
      "grad_norm": NaN,
      "learning_rate": 2.1221312212025947e-06,
      "loss": 0.0,
      "step": 5401
    },
    {
      "epoch": 0.9085863257926162,
      "grad_norm": NaN,
      "learning_rate": 2.114391847711117e-06,
      "loss": 0.0,
      "step": 5402
    },
    {
      "epoch": 0.9087545202253805,
      "grad_norm": NaN,
      "learning_rate": 2.1066663078354866e-06,
      "loss": 0.0,
      "step": 5403
    },
    {
      "epoch": 0.9089227146581448,
      "grad_norm": NaN,
      "learning_rate": 2.0989546038075234e-06,
      "loss": 0.0,
      "step": 5404
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": NaN,
      "learning_rate": 2.091256737855046e-06,
      "loss": 0.0,
      "step": 5405
    },
    {
      "epoch": 0.9092591035236733,
      "grad_norm": NaN,
      "learning_rate": 2.083572712201898e-06,
      "loss": 0.0,
      "step": 5406
    },
    {
      "epoch": 0.9094272979564376,
      "grad_norm": NaN,
      "learning_rate": 2.075902529067897e-06,
      "loss": 0.0,
      "step": 5407
    },
    {
      "epoch": 0.9095954923892019,
      "grad_norm": NaN,
      "learning_rate": 2.068246190668871e-06,
      "loss": 0.0,
      "step": 5408
    },
    {
      "epoch": 0.9097636868219662,
      "grad_norm": NaN,
      "learning_rate": 2.060603699216651e-06,
      "loss": 0.0,
      "step": 5409
    },
    {
      "epoch": 0.9099318812547305,
      "grad_norm": NaN,
      "learning_rate": 2.0529750569190763e-06,
      "loss": 0.0,
      "step": 5410
    },
    {
      "epoch": 0.9101000756874947,
      "grad_norm": NaN,
      "learning_rate": 2.0453602659799677e-06,
      "loss": 0.0,
      "step": 5411
    },
    {
      "epoch": 0.910268270120259,
      "grad_norm": NaN,
      "learning_rate": 2.0377593285991594e-06,
      "loss": 0.0,
      "step": 5412
    },
    {
      "epoch": 0.9104364645530233,
      "grad_norm": NaN,
      "learning_rate": 2.0301722469724726e-06,
      "loss": 0.0,
      "step": 5413
    },
    {
      "epoch": 0.9106046589857876,
      "grad_norm": NaN,
      "learning_rate": 2.022599023291727e-06,
      "loss": 0.0,
      "step": 5414
    },
    {
      "epoch": 0.9107728534185519,
      "grad_norm": NaN,
      "learning_rate": 2.0150396597447496e-06,
      "loss": 0.0,
      "step": 5415
    },
    {
      "epoch": 0.9109410478513161,
      "grad_norm": NaN,
      "learning_rate": 2.0074941585153497e-06,
      "loss": 0.0,
      "step": 5416
    },
    {
      "epoch": 0.9111092422840804,
      "grad_norm": NaN,
      "learning_rate": 1.9999625217833384e-06,
      "loss": 0.0,
      "step": 5417
    },
    {
      "epoch": 0.9112774367168447,
      "grad_norm": NaN,
      "learning_rate": 1.992444751724526e-06,
      "loss": 0.0,
      "step": 5418
    },
    {
      "epoch": 0.911445631149609,
      "grad_norm": NaN,
      "learning_rate": 1.984940850510708e-06,
      "loss": 0.0,
      "step": 5419
    },
    {
      "epoch": 0.9116138255823732,
      "grad_norm": NaN,
      "learning_rate": 1.977450820309684e-06,
      "loss": 0.0,
      "step": 5420
    },
    {
      "epoch": 0.9117820200151375,
      "grad_norm": NaN,
      "learning_rate": 1.9699746632852234e-06,
      "loss": 0.0,
      "step": 5421
    },
    {
      "epoch": 0.9119502144479018,
      "grad_norm": NaN,
      "learning_rate": 1.9625123815971203e-06,
      "loss": 0.0,
      "step": 5422
    },
    {
      "epoch": 0.9121184088806661,
      "grad_norm": NaN,
      "learning_rate": 1.955063977401145e-06,
      "loss": 0.0,
      "step": 5423
    },
    {
      "epoch": 0.9122866033134304,
      "grad_norm": NaN,
      "learning_rate": 1.947629452849048e-06,
      "loss": 0.0,
      "step": 5424
    },
    {
      "epoch": 0.9124547977461946,
      "grad_norm": NaN,
      "learning_rate": 1.940208810088584e-06,
      "loss": 0.0,
      "step": 5425
    },
    {
      "epoch": 0.9126229921789589,
      "grad_norm": NaN,
      "learning_rate": 1.9328020512634936e-06,
      "loss": 0.0,
      "step": 5426
    },
    {
      "epoch": 0.9127911866117232,
      "grad_norm": NaN,
      "learning_rate": 1.9254091785135153e-06,
      "loss": 0.0,
      "step": 5427
    },
    {
      "epoch": 0.9129593810444874,
      "grad_norm": NaN,
      "learning_rate": 1.9180301939743516e-06,
      "loss": 0.0,
      "step": 5428
    },
    {
      "epoch": 0.9131275754772517,
      "grad_norm": NaN,
      "learning_rate": 1.9106650997777197e-06,
      "loss": 0.0,
      "step": 5429
    },
    {
      "epoch": 0.9132957699100159,
      "grad_norm": NaN,
      "learning_rate": 1.9033138980513066e-06,
      "loss": 0.0,
      "step": 5430
    },
    {
      "epoch": 0.9134639643427802,
      "grad_norm": NaN,
      "learning_rate": 1.8959765909187965e-06,
      "loss": 0.0,
      "step": 5431
    },
    {
      "epoch": 0.9136321587755445,
      "grad_norm": NaN,
      "learning_rate": 1.8886531804998553e-06,
      "loss": 0.0,
      "step": 5432
    },
    {
      "epoch": 0.9138003532083088,
      "grad_norm": NaN,
      "learning_rate": 1.8813436689101239e-06,
      "loss": 0.0,
      "step": 5433
    },
    {
      "epoch": 0.913968547641073,
      "grad_norm": NaN,
      "learning_rate": 1.8740480582612519e-06,
      "loss": 0.0,
      "step": 5434
    },
    {
      "epoch": 0.9141367420738373,
      "grad_norm": NaN,
      "learning_rate": 1.8667663506608534e-06,
      "loss": 0.0,
      "step": 5435
    },
    {
      "epoch": 0.9143049365066016,
      "grad_norm": NaN,
      "learning_rate": 1.859498548212535e-06,
      "loss": 0.0,
      "step": 5436
    },
    {
      "epoch": 0.9144731309393659,
      "grad_norm": NaN,
      "learning_rate": 1.8522446530158778e-06,
      "loss": 0.0,
      "step": 5437
    },
    {
      "epoch": 0.9146413253721302,
      "grad_norm": NaN,
      "learning_rate": 1.8450046671664555e-06,
      "loss": 0.0,
      "step": 5438
    },
    {
      "epoch": 0.9148095198048944,
      "grad_norm": NaN,
      "learning_rate": 1.8377785927558232e-06,
      "loss": 0.0,
      "step": 5439
    },
    {
      "epoch": 0.9149777142376587,
      "grad_norm": NaN,
      "learning_rate": 1.8305664318715054e-06,
      "loss": 0.0,
      "step": 5440
    },
    {
      "epoch": 0.915145908670423,
      "grad_norm": NaN,
      "learning_rate": 1.8233681865970077e-06,
      "loss": 0.0,
      "step": 5441
    },
    {
      "epoch": 0.9153141031031873,
      "grad_norm": NaN,
      "learning_rate": 1.8161838590118384e-06,
      "loss": 0.0,
      "step": 5442
    },
    {
      "epoch": 0.9154822975359516,
      "grad_norm": NaN,
      "learning_rate": 1.8090134511914658e-06,
      "loss": 0.0,
      "step": 5443
    },
    {
      "epoch": 0.9156504919687158,
      "grad_norm": NaN,
      "learning_rate": 1.8018569652073381e-06,
      "loss": 0.0,
      "step": 5444
    },
    {
      "epoch": 0.9158186864014801,
      "grad_norm": NaN,
      "learning_rate": 1.7947144031268737e-06,
      "loss": 0.0,
      "step": 5445
    },
    {
      "epoch": 0.9159868808342444,
      "grad_norm": NaN,
      "learning_rate": 1.7875857670134943e-06,
      "loss": 0.0,
      "step": 5446
    },
    {
      "epoch": 0.9161550752670087,
      "grad_norm": NaN,
      "learning_rate": 1.7804710589265805e-06,
      "loss": 0.0,
      "step": 5447
    },
    {
      "epoch": 0.916323269699773,
      "grad_norm": NaN,
      "learning_rate": 1.7733702809214825e-06,
      "loss": 0.0,
      "step": 5448
    },
    {
      "epoch": 0.9164914641325372,
      "grad_norm": NaN,
      "learning_rate": 1.7662834350495428e-06,
      "loss": 0.0,
      "step": 5449
    },
    {
      "epoch": 0.9166596585653015,
      "grad_norm": NaN,
      "learning_rate": 1.7592105233580736e-06,
      "loss": 0.0,
      "step": 5450
    },
    {
      "epoch": 0.9168278529980658,
      "grad_norm": NaN,
      "learning_rate": 1.7521515478903517e-06,
      "loss": 0.0,
      "step": 5451
    },
    {
      "epoch": 0.9169960474308301,
      "grad_norm": NaN,
      "learning_rate": 1.7451065106856458e-06,
      "loss": 0.0,
      "step": 5452
    },
    {
      "epoch": 0.9171642418635944,
      "grad_norm": NaN,
      "learning_rate": 1.7380754137791778e-06,
      "loss": 0.0,
      "step": 5453
    },
    {
      "epoch": 0.9173324362963586,
      "grad_norm": NaN,
      "learning_rate": 1.7310582592021562e-06,
      "loss": 0.0,
      "step": 5454
    },
    {
      "epoch": 0.9175006307291229,
      "grad_norm": NaN,
      "learning_rate": 1.7240550489817653e-06,
      "loss": 0.0,
      "step": 5455
    },
    {
      "epoch": 0.9176688251618872,
      "grad_norm": NaN,
      "learning_rate": 1.7170657851411476e-06,
      "loss": 0.0,
      "step": 5456
    },
    {
      "epoch": 0.9178370195946515,
      "grad_norm": NaN,
      "learning_rate": 1.7100904696994269e-06,
      "loss": 0.0,
      "step": 5457
    },
    {
      "epoch": 0.9180052140274156,
      "grad_norm": NaN,
      "learning_rate": 1.7031291046716856e-06,
      "loss": 0.0,
      "step": 5458
    },
    {
      "epoch": 0.9181734084601799,
      "grad_norm": NaN,
      "learning_rate": 1.6961816920689932e-06,
      "loss": 0.0,
      "step": 5459
    },
    {
      "epoch": 0.9183416028929442,
      "grad_norm": NaN,
      "learning_rate": 1.6892482338983828e-06,
      "loss": 0.0,
      "step": 5460
    },
    {
      "epoch": 0.9185097973257085,
      "grad_norm": NaN,
      "learning_rate": 1.6823287321628412e-06,
      "loss": 0.0,
      "step": 5461
    },
    {
      "epoch": 0.9186779917584728,
      "grad_norm": NaN,
      "learning_rate": 1.6754231888613304e-06,
      "loss": 0.0,
      "step": 5462
    },
    {
      "epoch": 0.918846186191237,
      "grad_norm": NaN,
      "learning_rate": 1.6685316059888046e-06,
      "loss": 0.0,
      "step": 5463
    },
    {
      "epoch": 0.9190143806240013,
      "grad_norm": NaN,
      "learning_rate": 1.6616539855361547e-06,
      "loss": 0.0,
      "step": 5464
    },
    {
      "epoch": 0.9191825750567656,
      "grad_norm": NaN,
      "learning_rate": 1.6547903294902468e-06,
      "loss": 0.0,
      "step": 5465
    },
    {
      "epoch": 0.9193507694895299,
      "grad_norm": NaN,
      "learning_rate": 1.647940639833917e-06,
      "loss": 0.0,
      "step": 5466
    },
    {
      "epoch": 0.9195189639222942,
      "grad_norm": NaN,
      "learning_rate": 1.6411049185459605e-06,
      "loss": 0.0,
      "step": 5467
    },
    {
      "epoch": 0.9196871583550584,
      "grad_norm": NaN,
      "learning_rate": 1.6342831676011416e-06,
      "loss": 0.0,
      "step": 5468
    },
    {
      "epoch": 0.9198553527878227,
      "grad_norm": NaN,
      "learning_rate": 1.62747538897019e-06,
      "loss": 0.0,
      "step": 5469
    },
    {
      "epoch": 0.920023547220587,
      "grad_norm": NaN,
      "learning_rate": 1.6206815846197877e-06,
      "loss": 0.0,
      "step": 5470
    },
    {
      "epoch": 0.9201917416533513,
      "grad_norm": NaN,
      "learning_rate": 1.6139017565126035e-06,
      "loss": 0.0,
      "step": 5471
    },
    {
      "epoch": 0.9203599360861155,
      "grad_norm": NaN,
      "learning_rate": 1.6071359066072433e-06,
      "loss": 0.0,
      "step": 5472
    },
    {
      "epoch": 0.9205281305188798,
      "grad_norm": NaN,
      "learning_rate": 1.6003840368582935e-06,
      "loss": 0.0,
      "step": 5473
    },
    {
      "epoch": 0.9206963249516441,
      "grad_norm": NaN,
      "learning_rate": 1.5936461492162823e-06,
      "loss": 0.0,
      "step": 5474
    },
    {
      "epoch": 0.9208645193844084,
      "grad_norm": NaN,
      "learning_rate": 1.58692224562772e-06,
      "loss": 0.0,
      "step": 5475
    },
    {
      "epoch": 0.9210327138171727,
      "grad_norm": NaN,
      "learning_rate": 1.5802123280350633e-06,
      "loss": 0.0,
      "step": 5476
    },
    {
      "epoch": 0.921200908249937,
      "grad_norm": NaN,
      "learning_rate": 1.5735163983767342e-06,
      "loss": 0.0,
      "step": 5477
    },
    {
      "epoch": 0.9213691026827012,
      "grad_norm": NaN,
      "learning_rate": 1.566834458587102e-06,
      "loss": 0.0,
      "step": 5478
    },
    {
      "epoch": 0.9215372971154655,
      "grad_norm": NaN,
      "learning_rate": 1.5601665105965168e-06,
      "loss": 0.0,
      "step": 5479
    },
    {
      "epoch": 0.9217054915482298,
      "grad_norm": NaN,
      "learning_rate": 1.5535125563312713e-06,
      "loss": 0.0,
      "step": 5480
    },
    {
      "epoch": 0.9218736859809941,
      "grad_norm": NaN,
      "learning_rate": 1.5468725977136168e-06,
      "loss": 0.0,
      "step": 5481
    },
    {
      "epoch": 0.9220418804137583,
      "grad_norm": NaN,
      "learning_rate": 1.5402466366617575e-06,
      "loss": 0.0,
      "step": 5482
    },
    {
      "epoch": 0.9222100748465226,
      "grad_norm": NaN,
      "learning_rate": 1.5336346750898678e-06,
      "loss": 0.0,
      "step": 5483
    },
    {
      "epoch": 0.9223782692792869,
      "grad_norm": NaN,
      "learning_rate": 1.52703671490807e-06,
      "loss": 0.0,
      "step": 5484
    },
    {
      "epoch": 0.9225464637120512,
      "grad_norm": NaN,
      "learning_rate": 1.5204527580224337e-06,
      "loss": 0.0,
      "step": 5485
    },
    {
      "epoch": 0.9227146581448155,
      "grad_norm": NaN,
      "learning_rate": 1.513882806334993e-06,
      "loss": 0.0,
      "step": 5486
    },
    {
      "epoch": 0.9228828525775797,
      "grad_norm": NaN,
      "learning_rate": 1.5073268617437352e-06,
      "loss": 0.0,
      "step": 5487
    },
    {
      "epoch": 0.9230510470103439,
      "grad_norm": NaN,
      "learning_rate": 1.500784926142601e-06,
      "loss": 0.0,
      "step": 5488
    },
    {
      "epoch": 0.9232192414431082,
      "grad_norm": NaN,
      "learning_rate": 1.4942570014214785e-06,
      "loss": 0.0,
      "step": 5489
    },
    {
      "epoch": 0.9233874358758725,
      "grad_norm": NaN,
      "learning_rate": 1.4877430894662036e-06,
      "loss": 0.0,
      "step": 5490
    },
    {
      "epoch": 0.9235556303086367,
      "grad_norm": NaN,
      "learning_rate": 1.481243192158588e-06,
      "loss": 0.0,
      "step": 5491
    },
    {
      "epoch": 0.923723824741401,
      "grad_norm": NaN,
      "learning_rate": 1.4747573113763735e-06,
      "loss": 0.0,
      "step": 5492
    },
    {
      "epoch": 0.9238920191741653,
      "grad_norm": NaN,
      "learning_rate": 1.4682854489932562e-06,
      "loss": 0.0,
      "step": 5493
    },
    {
      "epoch": 0.9240602136069296,
      "grad_norm": NaN,
      "learning_rate": 1.4618276068788849e-06,
      "loss": 0.0,
      "step": 5494
    },
    {
      "epoch": 0.9242284080396939,
      "grad_norm": NaN,
      "learning_rate": 1.4553837868988618e-06,
      "loss": 0.0,
      "step": 5495
    },
    {
      "epoch": 0.9243966024724581,
      "grad_norm": NaN,
      "learning_rate": 1.4489539909147365e-06,
      "loss": 0.0,
      "step": 5496
    },
    {
      "epoch": 0.9245647969052224,
      "grad_norm": NaN,
      "learning_rate": 1.4425382207839954e-06,
      "loss": 0.0,
      "step": 5497
    },
    {
      "epoch": 0.9247329913379867,
      "grad_norm": NaN,
      "learning_rate": 1.4361364783600895e-06,
      "loss": 0.0,
      "step": 5498
    },
    {
      "epoch": 0.924901185770751,
      "grad_norm": NaN,
      "learning_rate": 1.4297487654924002e-06,
      "loss": 0.0,
      "step": 5499
    },
    {
      "epoch": 0.9250693802035153,
      "grad_norm": NaN,
      "learning_rate": 1.423375084026285e-06,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 0.9252375746362795,
      "grad_norm": NaN,
      "learning_rate": 1.4170154358030151e-06,
      "loss": 0.0,
      "step": 5501
    },
    {
      "epoch": 0.9254057690690438,
      "grad_norm": NaN,
      "learning_rate": 1.410669822659827e-06,
      "loss": 0.0,
      "step": 5502
    },
    {
      "epoch": 0.9255739635018081,
      "grad_norm": NaN,
      "learning_rate": 1.404338246429887e-06,
      "loss": 0.0,
      "step": 5503
    },
    {
      "epoch": 0.9257421579345724,
      "grad_norm": NaN,
      "learning_rate": 1.3980207089423326e-06,
      "loss": 0.0,
      "step": 5504
    },
    {
      "epoch": 0.9259103523673367,
      "grad_norm": NaN,
      "learning_rate": 1.39171721202222e-06,
      "loss": 0.0,
      "step": 5505
    },
    {
      "epoch": 0.9260785468001009,
      "grad_norm": NaN,
      "learning_rate": 1.385427757490565e-06,
      "loss": 0.0,
      "step": 5506
    },
    {
      "epoch": 0.9262467412328652,
      "grad_norm": NaN,
      "learning_rate": 1.3791523471643141e-06,
      "loss": 0.0,
      "step": 5507
    },
    {
      "epoch": 0.9264149356656295,
      "grad_norm": NaN,
      "learning_rate": 1.3728909828563619e-06,
      "loss": 0.0,
      "step": 5508
    },
    {
      "epoch": 0.9265831300983938,
      "grad_norm": NaN,
      "learning_rate": 1.3666436663755555e-06,
      "loss": 0.0,
      "step": 5509
    },
    {
      "epoch": 0.926751324531158,
      "grad_norm": NaN,
      "learning_rate": 1.3604103995266682e-06,
      "loss": 0.0,
      "step": 5510
    },
    {
      "epoch": 0.9269195189639223,
      "grad_norm": NaN,
      "learning_rate": 1.3541911841104149e-06,
      "loss": 0.0,
      "step": 5511
    },
    {
      "epoch": 0.9270877133966866,
      "grad_norm": NaN,
      "learning_rate": 1.3479860219234697e-06,
      "loss": 0.0,
      "step": 5512
    },
    {
      "epoch": 0.9272559078294509,
      "grad_norm": NaN,
      "learning_rate": 1.3417949147584318e-06,
      "loss": 0.0,
      "step": 5513
    },
    {
      "epoch": 0.9274241022622152,
      "grad_norm": NaN,
      "learning_rate": 1.335617864403832e-06,
      "loss": 0.0,
      "step": 5514
    },
    {
      "epoch": 0.9275922966949794,
      "grad_norm": NaN,
      "learning_rate": 1.3294548726441592e-06,
      "loss": 0.0,
      "step": 5515
    },
    {
      "epoch": 0.9277604911277437,
      "grad_norm": NaN,
      "learning_rate": 1.3233059412598392e-06,
      "loss": 0.0,
      "step": 5516
    },
    {
      "epoch": 0.9279286855605079,
      "grad_norm": NaN,
      "learning_rate": 1.3171710720272234e-06,
      "loss": 0.0,
      "step": 5517
    },
    {
      "epoch": 0.9280968799932722,
      "grad_norm": NaN,
      "learning_rate": 1.3110502667185997e-06,
      "loss": 0.0,
      "step": 5518
    },
    {
      "epoch": 0.9282650744260365,
      "grad_norm": NaN,
      "learning_rate": 1.3049435271022037e-06,
      "loss": 0.0,
      "step": 5519
    },
    {
      "epoch": 0.9284332688588007,
      "grad_norm": NaN,
      "learning_rate": 1.298850854942213e-06,
      "loss": 0.0,
      "step": 5520
    },
    {
      "epoch": 0.928601463291565,
      "grad_norm": NaN,
      "learning_rate": 1.2927722519987306e-06,
      "loss": 0.0,
      "step": 5521
    },
    {
      "epoch": 0.9287696577243293,
      "grad_norm": NaN,
      "learning_rate": 1.2867077200277856e-06,
      "loss": 0.0,
      "step": 5522
    },
    {
      "epoch": 0.9289378521570936,
      "grad_norm": NaN,
      "learning_rate": 1.2806572607813649e-06,
      "loss": 0.0,
      "step": 5523
    },
    {
      "epoch": 0.9291060465898578,
      "grad_norm": NaN,
      "learning_rate": 1.2746208760073708e-06,
      "loss": 0.0,
      "step": 5524
    },
    {
      "epoch": 0.9292742410226221,
      "grad_norm": NaN,
      "learning_rate": 1.268598567449647e-06,
      "loss": 0.0,
      "step": 5525
    },
    {
      "epoch": 0.9294424354553864,
      "grad_norm": NaN,
      "learning_rate": 1.2625903368479796e-06,
      "loss": 0.0,
      "step": 5526
    },
    {
      "epoch": 0.9296106298881507,
      "grad_norm": NaN,
      "learning_rate": 1.2565961859380693e-06,
      "loss": 0.0,
      "step": 5527
    },
    {
      "epoch": 0.929778824320915,
      "grad_norm": NaN,
      "learning_rate": 1.250616116451564e-06,
      "loss": 0.0,
      "step": 5528
    },
    {
      "epoch": 0.9299470187536792,
      "grad_norm": NaN,
      "learning_rate": 1.2446501301160374e-06,
      "loss": 0.0,
      "step": 5529
    },
    {
      "epoch": 0.9301152131864435,
      "grad_norm": NaN,
      "learning_rate": 1.2386982286549998e-06,
      "loss": 0.0,
      "step": 5530
    },
    {
      "epoch": 0.9302834076192078,
      "grad_norm": NaN,
      "learning_rate": 1.2327604137878812e-06,
      "loss": 0.0,
      "step": 5531
    },
    {
      "epoch": 0.9304516020519721,
      "grad_norm": NaN,
      "learning_rate": 1.2268366872300597e-06,
      "loss": 0.0,
      "step": 5532
    },
    {
      "epoch": 0.9306197964847364,
      "grad_norm": NaN,
      "learning_rate": 1.2209270506928271e-06,
      "loss": 0.0,
      "step": 5533
    },
    {
      "epoch": 0.9307879909175006,
      "grad_norm": NaN,
      "learning_rate": 1.2150315058834184e-06,
      "loss": 0.0,
      "step": 5534
    },
    {
      "epoch": 0.9309561853502649,
      "grad_norm": NaN,
      "learning_rate": 1.2091500545049706e-06,
      "loss": 0.0,
      "step": 5535
    },
    {
      "epoch": 0.9311243797830292,
      "grad_norm": NaN,
      "learning_rate": 1.203282698256597e-06,
      "loss": 0.0,
      "step": 5536
    },
    {
      "epoch": 0.9312925742157935,
      "grad_norm": NaN,
      "learning_rate": 1.1974294388332918e-06,
      "loss": 0.0,
      "step": 5537
    },
    {
      "epoch": 0.9314607686485578,
      "grad_norm": NaN,
      "learning_rate": 1.1915902779260024e-06,
      "loss": 0.0,
      "step": 5538
    },
    {
      "epoch": 0.931628963081322,
      "grad_norm": NaN,
      "learning_rate": 1.1857652172215905e-06,
      "loss": 0.0,
      "step": 5539
    },
    {
      "epoch": 0.9317971575140863,
      "grad_norm": NaN,
      "learning_rate": 1.1799542584028656e-06,
      "loss": 0.0,
      "step": 5540
    },
    {
      "epoch": 0.9319653519468506,
      "grad_norm": NaN,
      "learning_rate": 1.1741574031485347e-06,
      "loss": 0.0,
      "step": 5541
    },
    {
      "epoch": 0.9321335463796149,
      "grad_norm": NaN,
      "learning_rate": 1.1683746531332529e-06,
      "loss": 0.0,
      "step": 5542
    },
    {
      "epoch": 0.9323017408123792,
      "grad_norm": NaN,
      "learning_rate": 1.1626060100275837e-06,
      "loss": 0.0,
      "step": 5543
    },
    {
      "epoch": 0.9324699352451434,
      "grad_norm": NaN,
      "learning_rate": 1.156851475498033e-06,
      "loss": 0.0,
      "step": 5544
    },
    {
      "epoch": 0.9326381296779077,
      "grad_norm": NaN,
      "learning_rate": 1.1511110512070155e-06,
      "loss": 0.0,
      "step": 5545
    },
    {
      "epoch": 0.932806324110672,
      "grad_norm": NaN,
      "learning_rate": 1.1453847388128712e-06,
      "loss": 0.0,
      "step": 5546
    },
    {
      "epoch": 0.9329745185434362,
      "grad_norm": NaN,
      "learning_rate": 1.1396725399698772e-06,
      "loss": 0.0,
      "step": 5547
    },
    {
      "epoch": 0.9331427129762004,
      "grad_norm": NaN,
      "learning_rate": 1.133974456328213e-06,
      "loss": 0.0,
      "step": 5548
    },
    {
      "epoch": 0.9333109074089647,
      "grad_norm": NaN,
      "learning_rate": 1.1282904895340064e-06,
      "loss": 0.0,
      "step": 5549
    },
    {
      "epoch": 0.933479101841729,
      "grad_norm": NaN,
      "learning_rate": 1.1226206412292773e-06,
      "loss": 0.0,
      "step": 5550
    },
    {
      "epoch": 0.9336472962744933,
      "grad_norm": NaN,
      "learning_rate": 1.1169649130519932e-06,
      "loss": 0.0,
      "step": 5551
    },
    {
      "epoch": 0.9338154907072576,
      "grad_norm": NaN,
      "learning_rate": 1.111323306636014e-06,
      "loss": 0.0,
      "step": 5552
    },
    {
      "epoch": 0.9339836851400218,
      "grad_norm": NaN,
      "learning_rate": 1.1056958236111525e-06,
      "loss": 0.0,
      "step": 5553
    },
    {
      "epoch": 0.9341518795727861,
      "grad_norm": NaN,
      "learning_rate": 1.1000824656031195e-06,
      "loss": 0.0,
      "step": 5554
    },
    {
      "epoch": 0.9343200740055504,
      "grad_norm": NaN,
      "learning_rate": 1.0944832342335564e-06,
      "loss": 0.0,
      "step": 5555
    },
    {
      "epoch": 0.9344882684383147,
      "grad_norm": NaN,
      "learning_rate": 1.0888981311200031e-06,
      "loss": 0.0,
      "step": 5556
    },
    {
      "epoch": 0.934656462871079,
      "grad_norm": NaN,
      "learning_rate": 1.083327157875952e-06,
      "loss": 0.0,
      "step": 5557
    },
    {
      "epoch": 0.9348246573038432,
      "grad_norm": NaN,
      "learning_rate": 1.0777703161107877e-06,
      "loss": 0.0,
      "step": 5558
    },
    {
      "epoch": 0.9349928517366075,
      "grad_norm": NaN,
      "learning_rate": 1.0722276074298154e-06,
      "loss": 0.0,
      "step": 5559
    },
    {
      "epoch": 0.9351610461693718,
      "grad_norm": NaN,
      "learning_rate": 1.0666990334342707e-06,
      "loss": 0.0,
      "step": 5560
    },
    {
      "epoch": 0.9353292406021361,
      "grad_norm": NaN,
      "learning_rate": 1.0611845957212873e-06,
      "loss": 0.0,
      "step": 5561
    },
    {
      "epoch": 0.9354974350349003,
      "grad_norm": NaN,
      "learning_rate": 1.0556842958839242e-06,
      "loss": 0.0,
      "step": 5562
    },
    {
      "epoch": 0.9356656294676646,
      "grad_norm": NaN,
      "learning_rate": 1.0501981355111656e-06,
      "loss": 0.0,
      "step": 5563
    },
    {
      "epoch": 0.9358338239004289,
      "grad_norm": NaN,
      "learning_rate": 1.0447261161878884e-06,
      "loss": 0.0,
      "step": 5564
    },
    {
      "epoch": 0.9360020183331932,
      "grad_norm": NaN,
      "learning_rate": 1.0392682394949116e-06,
      "loss": 0.0,
      "step": 5565
    },
    {
      "epoch": 0.9361702127659575,
      "grad_norm": NaN,
      "learning_rate": 1.0338245070089515e-06,
      "loss": 0.0,
      "step": 5566
    },
    {
      "epoch": 0.9363384071987217,
      "grad_norm": NaN,
      "learning_rate": 1.0283949203026332e-06,
      "loss": 0.0,
      "step": 5567
    },
    {
      "epoch": 0.936506601631486,
      "grad_norm": NaN,
      "learning_rate": 1.0229794809445081e-06,
      "loss": 0.0,
      "step": 5568
    },
    {
      "epoch": 0.9366747960642503,
      "grad_norm": NaN,
      "learning_rate": 1.0175781904990412e-06,
      "loss": 0.0,
      "step": 5569
    },
    {
      "epoch": 0.9368429904970146,
      "grad_norm": NaN,
      "learning_rate": 1.0121910505266008e-06,
      "loss": 0.0,
      "step": 5570
    },
    {
      "epoch": 0.9370111849297789,
      "grad_norm": NaN,
      "learning_rate": 1.00681806258347e-06,
      "loss": 0.0,
      "step": 5571
    },
    {
      "epoch": 0.9371793793625431,
      "grad_norm": NaN,
      "learning_rate": 1.0014592282218404e-06,
      "loss": 0.0,
      "step": 5572
    },
    {
      "epoch": 0.9373475737953074,
      "grad_norm": NaN,
      "learning_rate": 9.96114548989835e-07,
      "loss": 0.0,
      "step": 5573
    },
    {
      "epoch": 0.9375157682280717,
      "grad_norm": NaN,
      "learning_rate": 9.907840264314572e-07,
      "loss": 0.0,
      "step": 5574
    },
    {
      "epoch": 0.937683962660836,
      "grad_norm": NaN,
      "learning_rate": 9.854676620866366e-07,
      "loss": 0.0,
      "step": 5575
    },
    {
      "epoch": 0.9378521570936001,
      "grad_norm": NaN,
      "learning_rate": 9.80165457491211e-07,
      "loss": 0.0,
      "step": 5576
    },
    {
      "epoch": 0.9380203515263644,
      "grad_norm": NaN,
      "learning_rate": 9.74877414176939e-07,
      "loss": 0.0,
      "step": 5577
    },
    {
      "epoch": 0.9381885459591287,
      "grad_norm": NaN,
      "learning_rate": 9.696035336714648e-07,
      "loss": 0.0,
      "step": 5578
    },
    {
      "epoch": 0.938356740391893,
      "grad_norm": NaN,
      "learning_rate": 9.643438174983589e-07,
      "loss": 0.0,
      "step": 5579
    },
    {
      "epoch": 0.9385249348246573,
      "grad_norm": NaN,
      "learning_rate": 9.590982671770943e-07,
      "loss": 0.0,
      "step": 5580
    },
    {
      "epoch": 0.9386931292574215,
      "grad_norm": NaN,
      "learning_rate": 9.538668842230537e-07,
      "loss": 0.0,
      "step": 5581
    },
    {
      "epoch": 0.9388613236901858,
      "grad_norm": NaN,
      "learning_rate": 9.486496701475167e-07,
      "loss": 0.0,
      "step": 5582
    },
    {
      "epoch": 0.9390295181229501,
      "grad_norm": NaN,
      "learning_rate": 9.434466264576892e-07,
      "loss": 0.0,
      "step": 5583
    },
    {
      "epoch": 0.9391977125557144,
      "grad_norm": NaN,
      "learning_rate": 9.382577546566574e-07,
      "loss": 0.0,
      "step": 5584
    },
    {
      "epoch": 0.9393659069884787,
      "grad_norm": NaN,
      "learning_rate": 9.330830562434445e-07,
      "loss": 0.0,
      "step": 5585
    },
    {
      "epoch": 0.9395341014212429,
      "grad_norm": NaN,
      "learning_rate": 9.279225327129548e-07,
      "loss": 0.0,
      "step": 5586
    },
    {
      "epoch": 0.9397022958540072,
      "grad_norm": NaN,
      "learning_rate": 9.227761855560069e-07,
      "loss": 0.0,
      "step": 5587
    },
    {
      "epoch": 0.9398704902867715,
      "grad_norm": NaN,
      "learning_rate": 9.176440162593169e-07,
      "loss": 0.0,
      "step": 5588
    },
    {
      "epoch": 0.9400386847195358,
      "grad_norm": NaN,
      "learning_rate": 9.125260263055213e-07,
      "loss": 0.0,
      "step": 5589
    },
    {
      "epoch": 0.9402068791523001,
      "grad_norm": NaN,
      "learning_rate": 9.074222171731427e-07,
      "loss": 0.0,
      "step": 5590
    },
    {
      "epoch": 0.9403750735850643,
      "grad_norm": NaN,
      "learning_rate": 9.023325903366242e-07,
      "loss": 0.0,
      "step": 5591
    },
    {
      "epoch": 0.9405432680178286,
      "grad_norm": NaN,
      "learning_rate": 8.972571472662838e-07,
      "loss": 0.0,
      "step": 5592
    },
    {
      "epoch": 0.9407114624505929,
      "grad_norm": NaN,
      "learning_rate": 8.921958894283767e-07,
      "loss": 0.0,
      "step": 5593
    },
    {
      "epoch": 0.9408796568833572,
      "grad_norm": NaN,
      "learning_rate": 8.871488182850441e-07,
      "loss": 0.0,
      "step": 5594
    },
    {
      "epoch": 0.9410478513161215,
      "grad_norm": NaN,
      "learning_rate": 8.821159352943143e-07,
      "loss": 0.0,
      "step": 5595
    },
    {
      "epoch": 0.9412160457488857,
      "grad_norm": NaN,
      "learning_rate": 8.770972419101464e-07,
      "loss": 0.0,
      "step": 5596
    },
    {
      "epoch": 0.94138424018165,
      "grad_norm": NaN,
      "learning_rate": 8.720927395823697e-07,
      "loss": 0.0,
      "step": 5597
    },
    {
      "epoch": 0.9415524346144143,
      "grad_norm": NaN,
      "learning_rate": 8.671024297567388e-07,
      "loss": 0.0,
      "step": 5598
    },
    {
      "epoch": 0.9417206290471786,
      "grad_norm": NaN,
      "learning_rate": 8.621263138749002e-07,
      "loss": 0.0,
      "step": 5599
    },
    {
      "epoch": 0.9418888234799428,
      "grad_norm": NaN,
      "learning_rate": 8.571643933743879e-07,
      "loss": 0.0,
      "step": 5600
    },
    {
      "epoch": 0.9420570179127071,
      "grad_norm": NaN,
      "learning_rate": 8.522166696886547e-07,
      "loss": 0.0,
      "step": 5601
    },
    {
      "epoch": 0.9422252123454714,
      "grad_norm": NaN,
      "learning_rate": 8.472831442470408e-07,
      "loss": 0.0,
      "step": 5602
    },
    {
      "epoch": 0.9423934067782357,
      "grad_norm": NaN,
      "learning_rate": 8.423638184747784e-07,
      "loss": 0.0,
      "step": 5603
    },
    {
      "epoch": 0.942561601211,
      "grad_norm": NaN,
      "learning_rate": 8.374586937930196e-07,
      "loss": 0.0,
      "step": 5604
    },
    {
      "epoch": 0.9427297956437642,
      "grad_norm": NaN,
      "learning_rate": 8.325677716187807e-07,
      "loss": 0.0,
      "step": 5605
    },
    {
      "epoch": 0.9428979900765284,
      "grad_norm": NaN,
      "learning_rate": 8.276910533650151e-07,
      "loss": 0.0,
      "step": 5606
    },
    {
      "epoch": 0.9430661845092927,
      "grad_norm": NaN,
      "learning_rate": 8.2282854044054e-07,
      "loss": 0.0,
      "step": 5607
    },
    {
      "epoch": 0.943234378942057,
      "grad_norm": NaN,
      "learning_rate": 8.179802342500876e-07,
      "loss": 0.0,
      "step": 5608
    },
    {
      "epoch": 0.9434025733748213,
      "grad_norm": NaN,
      "learning_rate": 8.13146136194265e-07,
      "loss": 0.0,
      "step": 5609
    },
    {
      "epoch": 0.9435707678075855,
      "grad_norm": NaN,
      "learning_rate": 8.083262476696051e-07,
      "loss": 0.0,
      "step": 5610
    },
    {
      "epoch": 0.9437389622403498,
      "grad_norm": NaN,
      "learning_rate": 8.035205700685167e-07,
      "loss": 0.0,
      "step": 5611
    },
    {
      "epoch": 0.9439071566731141,
      "grad_norm": NaN,
      "learning_rate": 7.987291047793056e-07,
      "loss": 0.0,
      "step": 5612
    },
    {
      "epoch": 0.9440753511058784,
      "grad_norm": NaN,
      "learning_rate": 7.93951853186159e-07,
      "loss": 0.0,
      "step": 5613
    },
    {
      "epoch": 0.9442435455386426,
      "grad_norm": NaN,
      "learning_rate": 7.891888166691952e-07,
      "loss": 0.0,
      "step": 5614
    },
    {
      "epoch": 0.9444117399714069,
      "grad_norm": NaN,
      "learning_rate": 7.844399966043802e-07,
      "loss": 0.0,
      "step": 5615
    },
    {
      "epoch": 0.9445799344041712,
      "grad_norm": NaN,
      "learning_rate": 7.797053943636112e-07,
      "loss": 0.0,
      "step": 5616
    },
    {
      "epoch": 0.9447481288369355,
      "grad_norm": NaN,
      "learning_rate": 7.749850113146551e-07,
      "loss": 0.0,
      "step": 5617
    },
    {
      "epoch": 0.9449163232696998,
      "grad_norm": NaN,
      "learning_rate": 7.70278848821171e-07,
      "loss": 0.0,
      "step": 5618
    },
    {
      "epoch": 0.945084517702464,
      "grad_norm": NaN,
      "learning_rate": 7.655869082427269e-07,
      "loss": 0.0,
      "step": 5619
    },
    {
      "epoch": 0.9452527121352283,
      "grad_norm": NaN,
      "learning_rate": 7.609091909347721e-07,
      "loss": 0.0,
      "step": 5620
    },
    {
      "epoch": 0.9454209065679926,
      "grad_norm": NaN,
      "learning_rate": 7.56245698248631e-07,
      "loss": 0.0,
      "step": 5621
    },
    {
      "epoch": 0.9455891010007569,
      "grad_norm": NaN,
      "learning_rate": 7.515964315315538e-07,
      "loss": 0.0,
      "step": 5622
    },
    {
      "epoch": 0.9457572954335212,
      "grad_norm": NaN,
      "learning_rate": 7.46961392126655e-07,
      "loss": 0.0,
      "step": 5623
    },
    {
      "epoch": 0.9459254898662854,
      "grad_norm": NaN,
      "learning_rate": 7.423405813729467e-07,
      "loss": 0.0,
      "step": 5624
    },
    {
      "epoch": 0.9460936842990497,
      "grad_norm": NaN,
      "learning_rate": 7.377340006053169e-07,
      "loss": 0.0,
      "step": 5625
    },
    {
      "epoch": 0.946261878731814,
      "grad_norm": NaN,
      "learning_rate": 7.33141651154573e-07,
      "loss": 0.0,
      "step": 5626
    },
    {
      "epoch": 0.9464300731645783,
      "grad_norm": NaN,
      "learning_rate": 7.285635343473818e-07,
      "loss": 0.0,
      "step": 5627
    },
    {
      "epoch": 0.9465982675973426,
      "grad_norm": NaN,
      "learning_rate": 7.239996515063186e-07,
      "loss": 0.0,
      "step": 5628
    },
    {
      "epoch": 0.9467664620301068,
      "grad_norm": NaN,
      "learning_rate": 7.194500039498286e-07,
      "loss": 0.0,
      "step": 5629
    },
    {
      "epoch": 0.9469346564628711,
      "grad_norm": NaN,
      "learning_rate": 7.149145929922607e-07,
      "loss": 0.0,
      "step": 5630
    },
    {
      "epoch": 0.9471028508956354,
      "grad_norm": NaN,
      "learning_rate": 7.103934199438444e-07,
      "loss": 0.0,
      "step": 5631
    },
    {
      "epoch": 0.9472710453283997,
      "grad_norm": NaN,
      "learning_rate": 7.058864861106907e-07,
      "loss": 0.0,
      "step": 5632
    },
    {
      "epoch": 0.947439239761164,
      "grad_norm": NaN,
      "learning_rate": 7.013937927948022e-07,
      "loss": 0.0,
      "step": 5633
    },
    {
      "epoch": 0.9476074341939282,
      "grad_norm": NaN,
      "learning_rate": 6.969153412940743e-07,
      "loss": 0.0,
      "step": 5634
    },
    {
      "epoch": 0.9477756286266924,
      "grad_norm": NaN,
      "learning_rate": 6.924511329022831e-07,
      "loss": 0.0,
      "step": 5635
    },
    {
      "epoch": 0.9479438230594567,
      "grad_norm": NaN,
      "learning_rate": 6.880011689090804e-07,
      "loss": 0.0,
      "step": 5636
    },
    {
      "epoch": 0.948112017492221,
      "grad_norm": NaN,
      "learning_rate": 6.8356545060001e-07,
      "loss": 0.0,
      "step": 5637
    },
    {
      "epoch": 0.9482802119249852,
      "grad_norm": NaN,
      "learning_rate": 6.79143979256508e-07,
      "loss": 0.0,
      "step": 5638
    },
    {
      "epoch": 0.9484484063577495,
      "grad_norm": NaN,
      "learning_rate": 6.747367561558859e-07,
      "loss": 0.0,
      "step": 5639
    },
    {
      "epoch": 0.9486166007905138,
      "grad_norm": NaN,
      "learning_rate": 6.703437825713421e-07,
      "loss": 0.0,
      "step": 5640
    },
    {
      "epoch": 0.9487847952232781,
      "grad_norm": NaN,
      "learning_rate": 6.659650597719502e-07,
      "loss": 0.0,
      "step": 5641
    },
    {
      "epoch": 0.9489529896560424,
      "grad_norm": NaN,
      "learning_rate": 6.616005890226817e-07,
      "loss": 0.0,
      "step": 5642
    },
    {
      "epoch": 0.9491211840888066,
      "grad_norm": NaN,
      "learning_rate": 6.572503715843836e-07,
      "loss": 0.0,
      "step": 5643
    },
    {
      "epoch": 0.9492893785215709,
      "grad_norm": NaN,
      "learning_rate": 6.52914408713784e-07,
      "loss": 0.0,
      "step": 5644
    },
    {
      "epoch": 0.9494575729543352,
      "grad_norm": NaN,
      "learning_rate": 6.485927016634863e-07,
      "loss": 0.0,
      "step": 5645
    },
    {
      "epoch": 0.9496257673870995,
      "grad_norm": NaN,
      "learning_rate": 6.44285251681992e-07,
      "loss": 0.0,
      "step": 5646
    },
    {
      "epoch": 0.9497939618198638,
      "grad_norm": NaN,
      "learning_rate": 6.399920600136722e-07,
      "loss": 0.0,
      "step": 5647
    },
    {
      "epoch": 0.949962156252628,
      "grad_norm": NaN,
      "learning_rate": 6.357131278987849e-07,
      "loss": 0.0,
      "step": 5648
    },
    {
      "epoch": 0.9501303506853923,
      "grad_norm": NaN,
      "learning_rate": 6.314484565734636e-07,
      "loss": 0.0,
      "step": 5649
    },
    {
      "epoch": 0.9502985451181566,
      "grad_norm": NaN,
      "learning_rate": 6.271980472697225e-07,
      "loss": 0.0,
      "step": 5650
    },
    {
      "epoch": 0.9504667395509209,
      "grad_norm": NaN,
      "learning_rate": 6.229619012154575e-07,
      "loss": 0.0,
      "step": 5651
    },
    {
      "epoch": 0.9506349339836851,
      "grad_norm": NaN,
      "learning_rate": 6.18740019634445e-07,
      "loss": 0.0,
      "step": 5652
    },
    {
      "epoch": 0.9508031284164494,
      "grad_norm": NaN,
      "learning_rate": 6.145324037463429e-07,
      "loss": 0.0,
      "step": 5653
    },
    {
      "epoch": 0.9509713228492137,
      "grad_norm": NaN,
      "learning_rate": 6.103390547666788e-07,
      "loss": 0.0,
      "step": 5654
    },
    {
      "epoch": 0.951139517281978,
      "grad_norm": NaN,
      "learning_rate": 6.061599739068668e-07,
      "loss": 0.0,
      "step": 5655
    },
    {
      "epoch": 0.9513077117147423,
      "grad_norm": NaN,
      "learning_rate": 6.019951623741916e-07,
      "loss": 0.0,
      "step": 5656
    },
    {
      "epoch": 0.9514759061475065,
      "grad_norm": NaN,
      "learning_rate": 5.978446213718291e-07,
      "loss": 0.0,
      "step": 5657
    },
    {
      "epoch": 0.9516441005802708,
      "grad_norm": NaN,
      "learning_rate": 5.937083520988151e-07,
      "loss": 0.0,
      "step": 5658
    },
    {
      "epoch": 0.9518122950130351,
      "grad_norm": NaN,
      "learning_rate": 5.895863557500769e-07,
      "loss": 0.0,
      "step": 5659
    },
    {
      "epoch": 0.9519804894457994,
      "grad_norm": NaN,
      "learning_rate": 5.85478633516412e-07,
      "loss": 0.0,
      "step": 5660
    },
    {
      "epoch": 0.9521486838785637,
      "grad_norm": NaN,
      "learning_rate": 5.813851865844988e-07,
      "loss": 0.0,
      "step": 5661
    },
    {
      "epoch": 0.9523168783113279,
      "grad_norm": NaN,
      "learning_rate": 5.773060161368804e-07,
      "loss": 0.0,
      "step": 5662
    },
    {
      "epoch": 0.9524850727440922,
      "grad_norm": NaN,
      "learning_rate": 5.732411233519919e-07,
      "loss": 0.0,
      "step": 5663
    },
    {
      "epoch": 0.9526532671768565,
      "grad_norm": NaN,
      "learning_rate": 5.691905094041272e-07,
      "loss": 0.0,
      "step": 5664
    },
    {
      "epoch": 0.9528214616096207,
      "grad_norm": NaN,
      "learning_rate": 5.651541754634726e-07,
      "loss": 0.0,
      "step": 5665
    },
    {
      "epoch": 0.952989656042385,
      "grad_norm": NaN,
      "learning_rate": 5.611321226960675e-07,
      "loss": 0.0,
      "step": 5666
    },
    {
      "epoch": 0.9531578504751492,
      "grad_norm": NaN,
      "learning_rate": 5.571243522638547e-07,
      "loss": 0.0,
      "step": 5667
    },
    {
      "epoch": 0.9533260449079135,
      "grad_norm": NaN,
      "learning_rate": 5.531308653246192e-07,
      "loss": 0.0,
      "step": 5668
    },
    {
      "epoch": 0.9534942393406778,
      "grad_norm": NaN,
      "learning_rate": 5.491516630320381e-07,
      "loss": 0.0,
      "step": 5669
    },
    {
      "epoch": 0.9536624337734421,
      "grad_norm": NaN,
      "learning_rate": 5.451867465356641e-07,
      "loss": 0.0,
      "step": 5670
    },
    {
      "epoch": 0.9538306282062063,
      "grad_norm": NaN,
      "learning_rate": 5.412361169809088e-07,
      "loss": 0.0,
      "step": 5671
    },
    {
      "epoch": 0.9539988226389706,
      "grad_norm": NaN,
      "learning_rate": 5.372997755090759e-07,
      "loss": 0.0,
      "step": 5672
    },
    {
      "epoch": 0.9541670170717349,
      "grad_norm": NaN,
      "learning_rate": 5.333777232573223e-07,
      "loss": 0.0,
      "step": 5673
    },
    {
      "epoch": 0.9543352115044992,
      "grad_norm": NaN,
      "learning_rate": 5.294699613586862e-07,
      "loss": 0.0,
      "step": 5674
    },
    {
      "epoch": 0.9545034059372635,
      "grad_norm": NaN,
      "learning_rate": 5.255764909420757e-07,
      "loss": 0.0,
      "step": 5675
    },
    {
      "epoch": 0.9546716003700277,
      "grad_norm": NaN,
      "learning_rate": 5.216973131322689e-07,
      "loss": 0.0,
      "step": 5676
    },
    {
      "epoch": 0.954839794802792,
      "grad_norm": NaN,
      "learning_rate": 5.178324290499248e-07,
      "loss": 0.0,
      "step": 5677
    },
    {
      "epoch": 0.9550079892355563,
      "grad_norm": NaN,
      "learning_rate": 5.139818398115559e-07,
      "loss": 0.0,
      "step": 5678
    },
    {
      "epoch": 0.9551761836683206,
      "grad_norm": NaN,
      "learning_rate": 5.101455465295557e-07,
      "loss": 0.0,
      "step": 5679
    },
    {
      "epoch": 0.9553443781010849,
      "grad_norm": NaN,
      "learning_rate": 5.063235503121933e-07,
      "loss": 0.0,
      "step": 5680
    },
    {
      "epoch": 0.9555125725338491,
      "grad_norm": NaN,
      "learning_rate": 5.025158522635964e-07,
      "loss": 0.0,
      "step": 5681
    },
    {
      "epoch": 0.9556807669666134,
      "grad_norm": NaN,
      "learning_rate": 4.987224534837631e-07,
      "loss": 0.0,
      "step": 5682
    },
    {
      "epoch": 0.9558489613993777,
      "grad_norm": NaN,
      "learning_rate": 4.949433550685722e-07,
      "loss": 0.0,
      "step": 5683
    },
    {
      "epoch": 0.956017155832142,
      "grad_norm": NaN,
      "learning_rate": 4.911785581097561e-07,
      "loss": 0.0,
      "step": 5684
    },
    {
      "epoch": 0.9561853502649063,
      "grad_norm": NaN,
      "learning_rate": 4.874280636949225e-07,
      "loss": 0.0,
      "step": 5685
    },
    {
      "epoch": 0.9563535446976705,
      "grad_norm": NaN,
      "learning_rate": 4.836918729075435e-07,
      "loss": 0.0,
      "step": 5686
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": NaN,
      "learning_rate": 4.79969986826978e-07,
      "loss": 0.0,
      "step": 5687
    },
    {
      "epoch": 0.9566899335631991,
      "grad_norm": NaN,
      "learning_rate": 4.7626240652842155e-07,
      "loss": 0.0,
      "step": 5688
    },
    {
      "epoch": 0.9568581279959634,
      "grad_norm": NaN,
      "learning_rate": 4.7256913308295627e-07,
      "loss": 0.0,
      "step": 5689
    },
    {
      "epoch": 0.9570263224287276,
      "grad_norm": NaN,
      "learning_rate": 4.688901675575341e-07,
      "loss": 0.0,
      "step": 5690
    },
    {
      "epoch": 0.9571945168614919,
      "grad_norm": NaN,
      "learning_rate": 4.6522551101496057e-07,
      "loss": 0.0,
      "step": 5691
    },
    {
      "epoch": 0.9573627112942562,
      "grad_norm": NaN,
      "learning_rate": 4.6157516451391656e-07,
      "loss": 0.0,
      "step": 5692
    },
    {
      "epoch": 0.9575309057270205,
      "grad_norm": NaN,
      "learning_rate": 4.579391291089419e-07,
      "loss": 0.0,
      "step": 5693
    },
    {
      "epoch": 0.9576991001597848,
      "grad_norm": NaN,
      "learning_rate": 4.543174058504518e-07,
      "loss": 0.0,
      "step": 5694
    },
    {
      "epoch": 0.9578672945925489,
      "grad_norm": NaN,
      "learning_rate": 4.507099957847205e-07,
      "loss": 0.0,
      "step": 5695
    },
    {
      "epoch": 0.9580354890253132,
      "grad_norm": NaN,
      "learning_rate": 4.4711689995389216e-07,
      "loss": 0.0,
      "step": 5696
    },
    {
      "epoch": 0.9582036834580775,
      "grad_norm": NaN,
      "learning_rate": 4.435381193959587e-07,
      "loss": 0.0,
      "step": 5697
    },
    {
      "epoch": 0.9583718778908418,
      "grad_norm": NaN,
      "learning_rate": 4.3997365514480416e-07,
      "loss": 0.0,
      "step": 5698
    },
    {
      "epoch": 0.958540072323606,
      "grad_norm": NaN,
      "learning_rate": 4.364235082301549e-07,
      "loss": 0.0,
      "step": 5699
    },
    {
      "epoch": 0.9587082667563703,
      "grad_norm": NaN,
      "learning_rate": 4.3288767967760715e-07,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 0.9588764611891346,
      "grad_norm": NaN,
      "learning_rate": 4.293661705086327e-07,
      "loss": 0.0,
      "step": 5701
    },
    {
      "epoch": 0.9590446556218989,
      "grad_norm": NaN,
      "learning_rate": 4.258589817405401e-07,
      "loss": 0.0,
      "step": 5702
    },
    {
      "epoch": 0.9592128500546632,
      "grad_norm": NaN,
      "learning_rate": 4.2236611438652986e-07,
      "loss": 0.0,
      "step": 5703
    },
    {
      "epoch": 0.9593810444874274,
      "grad_norm": NaN,
      "learning_rate": 4.188875694556449e-07,
      "loss": 0.0,
      "step": 5704
    },
    {
      "epoch": 0.9595492389201917,
      "grad_norm": NaN,
      "learning_rate": 4.1542334795280355e-07,
      "loss": 0.0,
      "step": 5705
    },
    {
      "epoch": 0.959717433352956,
      "grad_norm": NaN,
      "learning_rate": 4.119734508787776e-07,
      "loss": 0.0,
      "step": 5706
    },
    {
      "epoch": 0.9598856277857203,
      "grad_norm": NaN,
      "learning_rate": 4.0853787923020303e-07,
      "loss": 0.0,
      "step": 5707
    },
    {
      "epoch": 0.9600538222184846,
      "grad_norm": NaN,
      "learning_rate": 4.0511663399958044e-07,
      "loss": 0.0,
      "step": 5708
    },
    {
      "epoch": 0.9602220166512488,
      "grad_norm": NaN,
      "learning_rate": 4.017097161752692e-07,
      "loss": 0.0,
      "step": 5709
    },
    {
      "epoch": 0.9603902110840131,
      "grad_norm": NaN,
      "learning_rate": 3.983171267414876e-07,
      "loss": 0.0,
      "step": 5710
    },
    {
      "epoch": 0.9605584055167774,
      "grad_norm": NaN,
      "learning_rate": 3.949388666783127e-07,
      "loss": 0.0,
      "step": 5711
    },
    {
      "epoch": 0.9607265999495417,
      "grad_norm": NaN,
      "learning_rate": 3.9157493696169724e-07,
      "loss": 0.0,
      "step": 5712
    },
    {
      "epoch": 0.960894794382306,
      "grad_norm": NaN,
      "learning_rate": 3.8822533856343044e-07,
      "loss": 0.0,
      "step": 5713
    },
    {
      "epoch": 0.9610629888150702,
      "grad_norm": NaN,
      "learning_rate": 3.848900724511828e-07,
      "loss": 0.0,
      "step": 5714
    },
    {
      "epoch": 0.9612311832478345,
      "grad_norm": NaN,
      "learning_rate": 3.815691395884724e-07,
      "loss": 0.0,
      "step": 5715
    },
    {
      "epoch": 0.9613993776805988,
      "grad_norm": NaN,
      "learning_rate": 3.782625409346763e-07,
      "loss": 0.0,
      "step": 5716
    },
    {
      "epoch": 0.9615675721133631,
      "grad_norm": NaN,
      "learning_rate": 3.749702774450414e-07,
      "loss": 0.0,
      "step": 5717
    },
    {
      "epoch": 0.9617357665461274,
      "grad_norm": NaN,
      "learning_rate": 3.7169235007065707e-07,
      "loss": 0.0,
      "step": 5718
    },
    {
      "epoch": 0.9619039609788916,
      "grad_norm": NaN,
      "learning_rate": 3.684287597584879e-07,
      "loss": 0.0,
      "step": 5719
    },
    {
      "epoch": 0.9620721554116559,
      "grad_norm": NaN,
      "learning_rate": 3.651795074513409e-07,
      "loss": 0.0,
      "step": 5720
    },
    {
      "epoch": 0.9622403498444202,
      "grad_norm": NaN,
      "learning_rate": 3.619445940878929e-07,
      "loss": 0.0,
      "step": 5721
    },
    {
      "epoch": 0.9624085442771845,
      "grad_norm": NaN,
      "learning_rate": 3.587240206026743e-07,
      "loss": 0.0,
      "step": 5722
    },
    {
      "epoch": 0.9625767387099488,
      "grad_norm": NaN,
      "learning_rate": 3.555177879260685e-07,
      "loss": 0.0,
      "step": 5723
    },
    {
      "epoch": 0.9627449331427129,
      "grad_norm": NaN,
      "learning_rate": 3.5232589698432907e-07,
      "loss": 0.0,
      "step": 5724
    },
    {
      "epoch": 0.9629131275754772,
      "grad_norm": NaN,
      "learning_rate": 3.491483486995517e-07,
      "loss": 0.0,
      "step": 5725
    },
    {
      "epoch": 0.9630813220082415,
      "grad_norm": NaN,
      "learning_rate": 3.459851439896966e-07,
      "loss": 0.0,
      "step": 5726
    },
    {
      "epoch": 0.9632495164410058,
      "grad_norm": NaN,
      "learning_rate": 3.428362837685717e-07,
      "loss": 0.0,
      "step": 5727
    },
    {
      "epoch": 0.96341771087377,
      "grad_norm": NaN,
      "learning_rate": 3.397017689458548e-07,
      "loss": 0.0,
      "step": 5728
    },
    {
      "epoch": 0.9635859053065343,
      "grad_norm": NaN,
      "learning_rate": 3.365816004270661e-07,
      "loss": 0.0,
      "step": 5729
    },
    {
      "epoch": 0.9637540997392986,
      "grad_norm": NaN,
      "learning_rate": 3.334757791135956e-07,
      "loss": 0.0,
      "step": 5730
    },
    {
      "epoch": 0.9639222941720629,
      "grad_norm": NaN,
      "learning_rate": 3.303843059026757e-07,
      "loss": 0.0,
      "step": 5731
    },
    {
      "epoch": 0.9640904886048272,
      "grad_norm": NaN,
      "learning_rate": 3.273071816873974e-07,
      "loss": 0.0,
      "step": 5732
    },
    {
      "epoch": 0.9642586830375914,
      "grad_norm": NaN,
      "learning_rate": 3.2424440735670526e-07,
      "loss": 0.0,
      "step": 5733
    },
    {
      "epoch": 0.9644268774703557,
      "grad_norm": NaN,
      "learning_rate": 3.211959837954026e-07,
      "loss": 0.0,
      "step": 5734
    },
    {
      "epoch": 0.96459507190312,
      "grad_norm": NaN,
      "learning_rate": 3.1816191188415166e-07,
      "loss": 0.0,
      "step": 5735
    },
    {
      "epoch": 0.9647632663358843,
      "grad_norm": NaN,
      "learning_rate": 3.151421924994513e-07,
      "loss": 0.0,
      "step": 5736
    },
    {
      "epoch": 0.9649314607686486,
      "grad_norm": NaN,
      "learning_rate": 3.121368265136704e-07,
      "loss": 0.0,
      "step": 5737
    },
    {
      "epoch": 0.9650996552014128,
      "grad_norm": NaN,
      "learning_rate": 3.091458147950255e-07,
      "loss": 0.0,
      "step": 5738
    },
    {
      "epoch": 0.9652678496341771,
      "grad_norm": NaN,
      "learning_rate": 3.0616915820758095e-07,
      "loss": 0.0,
      "step": 5739
    },
    {
      "epoch": 0.9654360440669414,
      "grad_norm": NaN,
      "learning_rate": 3.0320685761127123e-07,
      "loss": 0.0,
      "step": 5740
    },
    {
      "epoch": 0.9656042384997057,
      "grad_norm": NaN,
      "learning_rate": 3.002589138618561e-07,
      "loss": 0.0,
      "step": 5741
    },
    {
      "epoch": 0.96577243293247,
      "grad_norm": NaN,
      "learning_rate": 2.973253278109767e-07,
      "loss": 0.0,
      "step": 5742
    },
    {
      "epoch": 0.9659406273652342,
      "grad_norm": NaN,
      "learning_rate": 2.9440610030610494e-07,
      "loss": 0.0,
      "step": 5743
    },
    {
      "epoch": 0.9661088217979985,
      "grad_norm": NaN,
      "learning_rate": 2.91501232190583e-07,
      "loss": 0.0,
      "step": 5744
    },
    {
      "epoch": 0.9662770162307628,
      "grad_norm": NaN,
      "learning_rate": 2.8861072430358404e-07,
      "loss": 0.0,
      "step": 5745
    },
    {
      "epoch": 0.9664452106635271,
      "grad_norm": NaN,
      "learning_rate": 2.8573457748014564e-07,
      "loss": 0.0,
      "step": 5746
    },
    {
      "epoch": 0.9666134050962913,
      "grad_norm": NaN,
      "learning_rate": 2.8287279255115873e-07,
      "loss": 0.0,
      "step": 5747
    },
    {
      "epoch": 0.9667815995290556,
      "grad_norm": NaN,
      "learning_rate": 2.800253703433564e-07,
      "loss": 0.0,
      "step": 5748
    },
    {
      "epoch": 0.9669497939618199,
      "grad_norm": NaN,
      "learning_rate": 2.7719231167933067e-07,
      "loss": 0.0,
      "step": 5749
    },
    {
      "epoch": 0.9671179883945842,
      "grad_norm": NaN,
      "learning_rate": 2.743736173775213e-07,
      "loss": 0.0,
      "step": 5750
    },
    {
      "epoch": 0.9672861828273485,
      "grad_norm": NaN,
      "learning_rate": 2.715692882522103e-07,
      "loss": 0.0,
      "step": 5751
    },
    {
      "epoch": 0.9674543772601127,
      "grad_norm": NaN,
      "learning_rate": 2.687793251135384e-07,
      "loss": 0.0,
      "step": 5752
    },
    {
      "epoch": 0.967622571692877,
      "grad_norm": NaN,
      "learning_rate": 2.6600372876750544e-07,
      "loss": 0.0,
      "step": 5753
    },
    {
      "epoch": 0.9677907661256412,
      "grad_norm": NaN,
      "learning_rate": 2.6324250001593664e-07,
      "loss": 0.0,
      "step": 5754
    },
    {
      "epoch": 0.9679589605584055,
      "grad_norm": NaN,
      "learning_rate": 2.604956396565328e-07,
      "loss": 0.0,
      "step": 5755
    },
    {
      "epoch": 0.9681271549911697,
      "grad_norm": NaN,
      "learning_rate": 2.577631484828147e-07,
      "loss": 0.0,
      "step": 5756
    },
    {
      "epoch": 0.968295349423934,
      "grad_norm": NaN,
      "learning_rate": 2.550450272841842e-07,
      "loss": 0.0,
      "step": 5757
    },
    {
      "epoch": 0.9684635438566983,
      "grad_norm": NaN,
      "learning_rate": 2.523412768458688e-07,
      "loss": 0.0,
      "step": 5758
    },
    {
      "epoch": 0.9686317382894626,
      "grad_norm": NaN,
      "learning_rate": 2.4965189794895485e-07,
      "loss": 0.0,
      "step": 5759
    },
    {
      "epoch": 0.9687999327222269,
      "grad_norm": NaN,
      "learning_rate": 2.469768913703707e-07,
      "loss": 0.0,
      "step": 5760
    },
    {
      "epoch": 0.9689681271549911,
      "grad_norm": NaN,
      "learning_rate": 2.443162578828928e-07,
      "loss": 0.0,
      "step": 5761
    },
    {
      "epoch": 0.9691363215877554,
      "grad_norm": NaN,
      "learning_rate": 2.4166999825515625e-07,
      "loss": 0.0,
      "step": 5762
    },
    {
      "epoch": 0.9693045160205197,
      "grad_norm": NaN,
      "learning_rate": 2.3903811325163283e-07,
      "loss": 0.0,
      "step": 5763
    },
    {
      "epoch": 0.969472710453284,
      "grad_norm": NaN,
      "learning_rate": 2.3642060363264217e-07,
      "loss": 0.0,
      "step": 5764
    },
    {
      "epoch": 0.9696409048860483,
      "grad_norm": NaN,
      "learning_rate": 2.3381747015435163e-07,
      "loss": 0.0,
      "step": 5765
    },
    {
      "epoch": 0.9698090993188125,
      "grad_norm": NaN,
      "learning_rate": 2.312287135687874e-07,
      "loss": 0.0,
      "step": 5766
    },
    {
      "epoch": 0.9699772937515768,
      "grad_norm": NaN,
      "learning_rate": 2.2865433462380125e-07,
      "loss": 0.0,
      "step": 5767
    },
    {
      "epoch": 0.9701454881843411,
      "grad_norm": NaN,
      "learning_rate": 2.2609433406310941e-07,
      "loss": 0.0,
      "step": 5768
    },
    {
      "epoch": 0.9703136826171054,
      "grad_norm": NaN,
      "learning_rate": 2.2354871262626477e-07,
      "loss": 0.0,
      "step": 5769
    },
    {
      "epoch": 0.9704818770498697,
      "grad_norm": NaN,
      "learning_rate": 2.2101747104866788e-07,
      "loss": 0.0,
      "step": 5770
    },
    {
      "epoch": 0.9706500714826339,
      "grad_norm": NaN,
      "learning_rate": 2.185006100615672e-07,
      "loss": 0.0,
      "step": 5771
    },
    {
      "epoch": 0.9708182659153982,
      "grad_norm": NaN,
      "learning_rate": 2.159981303920533e-07,
      "loss": 0.0,
      "step": 5772
    },
    {
      "epoch": 0.9709864603481625,
      "grad_norm": NaN,
      "learning_rate": 2.1351003276307014e-07,
      "loss": 0.0,
      "step": 5773
    },
    {
      "epoch": 0.9711546547809268,
      "grad_norm": NaN,
      "learning_rate": 2.1103631789339272e-07,
      "loss": 0.0,
      "step": 5774
    },
    {
      "epoch": 0.971322849213691,
      "grad_norm": NaN,
      "learning_rate": 2.0857698649766055e-07,
      "loss": 0.0,
      "step": 5775
    },
    {
      "epoch": 0.9714910436464553,
      "grad_norm": NaN,
      "learning_rate": 2.061320392863386e-07,
      "loss": 0.0,
      "step": 5776
    },
    {
      "epoch": 0.9716592380792196,
      "grad_norm": NaN,
      "learning_rate": 2.0370147696574526e-07,
      "loss": 0.0,
      "step": 5777
    },
    {
      "epoch": 0.9718274325119839,
      "grad_norm": NaN,
      "learning_rate": 2.012853002380466e-07,
      "loss": 0.0,
      "step": 5778
    },
    {
      "epoch": 0.9719956269447482,
      "grad_norm": NaN,
      "learning_rate": 1.988835098012509e-07,
      "loss": 0.0,
      "step": 5779
    },
    {
      "epoch": 0.9721638213775124,
      "grad_norm": NaN,
      "learning_rate": 1.9649610634919767e-07,
      "loss": 0.0,
      "step": 5780
    },
    {
      "epoch": 0.9723320158102767,
      "grad_norm": NaN,
      "learning_rate": 1.9412309057159073e-07,
      "loss": 0.0,
      "step": 5781
    },
    {
      "epoch": 0.972500210243041,
      "grad_norm": NaN,
      "learning_rate": 1.9176446315397056e-07,
      "loss": 0.0,
      "step": 5782
    },
    {
      "epoch": 0.9726684046758052,
      "grad_norm": NaN,
      "learning_rate": 1.894202247777088e-07,
      "loss": 0.0,
      "step": 5783
    },
    {
      "epoch": 0.9728365991085695,
      "grad_norm": NaN,
      "learning_rate": 1.8709037612003045e-07,
      "loss": 0.0,
      "step": 5784
    },
    {
      "epoch": 0.9730047935413337,
      "grad_norm": NaN,
      "learning_rate": 1.8477491785400813e-07,
      "loss": 0.0,
      "step": 5785
    },
    {
      "epoch": 0.973172987974098,
      "grad_norm": NaN,
      "learning_rate": 1.8247385064855127e-07,
      "loss": 0.0,
      "step": 5786
    },
    {
      "epoch": 0.9733411824068623,
      "grad_norm": NaN,
      "learning_rate": 1.8018717516841143e-07,
      "loss": 0.0,
      "step": 5787
    },
    {
      "epoch": 0.9735093768396266,
      "grad_norm": NaN,
      "learning_rate": 1.779148920741769e-07,
      "loss": 0.0,
      "step": 5788
    },
    {
      "epoch": 0.9736775712723909,
      "grad_norm": NaN,
      "learning_rate": 1.7565700202229473e-07,
      "loss": 0.0,
      "step": 5789
    },
    {
      "epoch": 0.9738457657051551,
      "grad_norm": NaN,
      "learning_rate": 1.7341350566504323e-07,
      "loss": 0.0,
      "step": 5790
    },
    {
      "epoch": 0.9740139601379194,
      "grad_norm": NaN,
      "learning_rate": 1.7118440365053722e-07,
      "loss": 0.0,
      "step": 5791
    },
    {
      "epoch": 0.9741821545706837,
      "grad_norm": NaN,
      "learning_rate": 1.6896969662273944e-07,
      "loss": 0.0,
      "step": 5792
    },
    {
      "epoch": 0.974350349003448,
      "grad_norm": NaN,
      "learning_rate": 1.6676938522146023e-07,
      "loss": 0.0,
      "step": 5793
    },
    {
      "epoch": 0.9745185434362122,
      "grad_norm": NaN,
      "learning_rate": 1.645834700823412e-07,
      "loss": 0.0,
      "step": 5794
    },
    {
      "epoch": 0.9746867378689765,
      "grad_norm": NaN,
      "learning_rate": 1.6241195183686608e-07,
      "loss": 0.0,
      "step": 5795
    },
    {
      "epoch": 0.9748549323017408,
      "grad_norm": NaN,
      "learning_rate": 1.6025483111236638e-07,
      "loss": 0.0,
      "step": 5796
    },
    {
      "epoch": 0.9750231267345051,
      "grad_norm": NaN,
      "learning_rate": 1.581121085320103e-07,
      "loss": 0.0,
      "step": 5797
    },
    {
      "epoch": 0.9751913211672694,
      "grad_norm": NaN,
      "learning_rate": 1.5598378471480267e-07,
      "loss": 0.0,
      "step": 5798
    },
    {
      "epoch": 0.9753595156000336,
      "grad_norm": NaN,
      "learning_rate": 1.5386986027559613e-07,
      "loss": 0.0,
      "step": 5799
    },
    {
      "epoch": 0.9755277100327979,
      "grad_norm": NaN,
      "learning_rate": 1.5177033582507993e-07,
      "loss": 0.0,
      "step": 5800
    },
    {
      "epoch": 0.9756959044655622,
      "grad_norm": NaN,
      "learning_rate": 1.4968521196978002e-07,
      "loss": 0.0,
      "step": 5801
    },
    {
      "epoch": 0.9758640988983265,
      "grad_norm": NaN,
      "learning_rate": 1.4761448931206455e-07,
      "loss": 0.0,
      "step": 5802
    },
    {
      "epoch": 0.9760322933310908,
      "grad_norm": NaN,
      "learning_rate": 1.4555816845014948e-07,
      "loss": 0.0,
      "step": 5803
    },
    {
      "epoch": 0.976200487763855,
      "grad_norm": NaN,
      "learning_rate": 1.4351624997807623e-07,
      "loss": 0.0,
      "step": 5804
    },
    {
      "epoch": 0.9763686821966193,
      "grad_norm": NaN,
      "learning_rate": 1.4148873448573408e-07,
      "loss": 0.0,
      "step": 5805
    },
    {
      "epoch": 0.9765368766293836,
      "grad_norm": NaN,
      "learning_rate": 1.3947562255884338e-07,
      "loss": 0.0,
      "step": 5806
    },
    {
      "epoch": 0.9767050710621479,
      "grad_norm": NaN,
      "learning_rate": 1.374769147789834e-07,
      "loss": 0.0,
      "step": 5807
    },
    {
      "epoch": 0.9768732654949122,
      "grad_norm": NaN,
      "learning_rate": 1.3549261172354777e-07,
      "loss": 0.0,
      "step": 5808
    },
    {
      "epoch": 0.9770414599276764,
      "grad_norm": NaN,
      "learning_rate": 1.3352271396577798e-07,
      "loss": 0.0,
      "step": 5809
    },
    {
      "epoch": 0.9772096543604407,
      "grad_norm": NaN,
      "learning_rate": 1.3156722207476324e-07,
      "loss": 0.0,
      "step": 5810
    },
    {
      "epoch": 0.977377848793205,
      "grad_norm": NaN,
      "learning_rate": 1.2962613661541834e-07,
      "loss": 0.0,
      "step": 5811
    },
    {
      "epoch": 0.9775460432259693,
      "grad_norm": NaN,
      "learning_rate": 1.2769945814850582e-07,
      "loss": 0.0,
      "step": 5812
    },
    {
      "epoch": 0.9777142376587334,
      "grad_norm": NaN,
      "learning_rate": 1.2578718723061378e-07,
      "loss": 0.0,
      "step": 5813
    },
    {
      "epoch": 0.9778824320914977,
      "grad_norm": NaN,
      "learning_rate": 1.2388932441418367e-07,
      "loss": 0.0,
      "step": 5814
    },
    {
      "epoch": 0.978050626524262,
      "grad_norm": NaN,
      "learning_rate": 1.220058702474769e-07,
      "loss": 0.0,
      "step": 5815
    },
    {
      "epoch": 0.9782188209570263,
      "grad_norm": NaN,
      "learning_rate": 1.2013682527461379e-07,
      "loss": 0.0,
      "step": 5816
    },
    {
      "epoch": 0.9783870153897906,
      "grad_norm": NaN,
      "learning_rate": 1.1828219003553465e-07,
      "loss": 0.0,
      "step": 5817
    },
    {
      "epoch": 0.9785552098225548,
      "grad_norm": NaN,
      "learning_rate": 1.16441965066022e-07,
      "loss": 0.0,
      "step": 5818
    },
    {
      "epoch": 0.9787234042553191,
      "grad_norm": NaN,
      "learning_rate": 1.1461615089770062e-07,
      "loss": 0.0,
      "step": 5819
    },
    {
      "epoch": 0.9788915986880834,
      "grad_norm": NaN,
      "learning_rate": 1.1280474805802632e-07,
      "loss": 0.0,
      "step": 5820
    },
    {
      "epoch": 0.9790597931208477,
      "grad_norm": NaN,
      "learning_rate": 1.110077570702861e-07,
      "loss": 0.0,
      "step": 5821
    },
    {
      "epoch": 0.979227987553612,
      "grad_norm": NaN,
      "learning_rate": 1.0922517845362023e-07,
      "loss": 0.0,
      "step": 5822
    },
    {
      "epoch": 0.9793961819863762,
      "grad_norm": NaN,
      "learning_rate": 1.0745701272298902e-07,
      "loss": 0.0,
      "step": 5823
    },
    {
      "epoch": 0.9795643764191405,
      "grad_norm": NaN,
      "learning_rate": 1.0570326038920053e-07,
      "loss": 0.0,
      "step": 5824
    },
    {
      "epoch": 0.9797325708519048,
      "grad_norm": NaN,
      "learning_rate": 1.0396392195889393e-07,
      "loss": 0.0,
      "step": 5825
    },
    {
      "epoch": 0.9799007652846691,
      "grad_norm": NaN,
      "learning_rate": 1.022389979345395e-07,
      "loss": 0.0,
      "step": 5826
    },
    {
      "epoch": 0.9800689597174334,
      "grad_norm": NaN,
      "learning_rate": 1.0052848881444976e-07,
      "loss": 0.0,
      "step": 5827
    },
    {
      "epoch": 0.9802371541501976,
      "grad_norm": NaN,
      "learning_rate": 9.883239509277942e-08,
      "loss": 0.0,
      "step": 5828
    },
    {
      "epoch": 0.9804053485829619,
      "grad_norm": NaN,
      "learning_rate": 9.715071725949765e-08,
      "loss": 0.0,
      "step": 5829
    },
    {
      "epoch": 0.9805735430157262,
      "grad_norm": NaN,
      "learning_rate": 9.54834558004325e-08,
      "loss": 0.0,
      "step": 5830
    },
    {
      "epoch": 0.9807417374484905,
      "grad_norm": NaN,
      "learning_rate": 9.383061119723757e-08,
      "loss": 0.0,
      "step": 5831
    },
    {
      "epoch": 0.9809099318812547,
      "grad_norm": NaN,
      "learning_rate": 9.219218392739759e-08,
      "loss": 0.0,
      "step": 5832
    },
    {
      "epoch": 0.981078126314019,
      "grad_norm": NaN,
      "learning_rate": 9.056817446422839e-08,
      "loss": 0.0,
      "step": 5833
    },
    {
      "epoch": 0.9812463207467833,
      "grad_norm": NaN,
      "learning_rate": 8.895858327690464e-08,
      "loss": 0.0,
      "step": 5834
    },
    {
      "epoch": 0.9814145151795476,
      "grad_norm": NaN,
      "learning_rate": 8.736341083041e-08,
      "loss": 0.0,
      "step": 5835
    },
    {
      "epoch": 0.9815827096123119,
      "grad_norm": NaN,
      "learning_rate": 8.578265758557024e-08,
      "loss": 0.0,
      "step": 5836
    },
    {
      "epoch": 0.9817509040450761,
      "grad_norm": NaN,
      "learning_rate": 8.421632399904788e-08,
      "loss": 0.0,
      "step": 5837
    },
    {
      "epoch": 0.9819190984778404,
      "grad_norm": NaN,
      "learning_rate": 8.266441052334206e-08,
      "loss": 0.0,
      "step": 5838
    },
    {
      "epoch": 0.9820872929106047,
      "grad_norm": NaN,
      "learning_rate": 8.112691760677749e-08,
      "loss": 0.0,
      "step": 5839
    },
    {
      "epoch": 0.982255487343369,
      "grad_norm": NaN,
      "learning_rate": 7.960384569353219e-08,
      "loss": 0.0,
      "step": 5840
    },
    {
      "epoch": 0.9824236817761333,
      "grad_norm": NaN,
      "learning_rate": 7.809519522358755e-08,
      "loss": 0.0,
      "step": 5841
    },
    {
      "epoch": 0.9825918762088974,
      "grad_norm": NaN,
      "learning_rate": 7.660096663278938e-08,
      "loss": 0.0,
      "step": 5842
    },
    {
      "epoch": 0.9827600706416617,
      "grad_norm": NaN,
      "learning_rate": 7.512116035279237e-08,
      "loss": 0.0,
      "step": 5843
    },
    {
      "epoch": 0.982928265074426,
      "grad_norm": NaN,
      "learning_rate": 7.365577681110458e-08,
      "loss": 0.0,
      "step": 5844
    },
    {
      "epoch": 0.9830964595071903,
      "grad_norm": NaN,
      "learning_rate": 7.220481643105403e-08,
      "loss": 0.0,
      "step": 5845
    },
    {
      "epoch": 0.9832646539399545,
      "grad_norm": NaN,
      "learning_rate": 7.076827963181099e-08,
      "loss": 0.0,
      "step": 5846
    },
    {
      "epoch": 0.9834328483727188,
      "grad_norm": NaN,
      "learning_rate": 6.934616682837125e-08,
      "loss": 0.0,
      "step": 5847
    },
    {
      "epoch": 0.9836010428054831,
      "grad_norm": NaN,
      "learning_rate": 6.79384784315673e-08,
      "loss": 0.0,
      "step": 5848
    },
    {
      "epoch": 0.9837692372382474,
      "grad_norm": NaN,
      "learning_rate": 6.65452148480683e-08,
      "loss": 0.0,
      "step": 5849
    },
    {
      "epoch": 0.9839374316710117,
      "grad_norm": NaN,
      "learning_rate": 6.516637648036894e-08,
      "loss": 0.0,
      "step": 5850
    },
    {
      "epoch": 0.9841056261037759,
      "grad_norm": NaN,
      "learning_rate": 6.380196372680058e-08,
      "loss": 0.0,
      "step": 5851
    },
    {
      "epoch": 0.9842738205365402,
      "grad_norm": NaN,
      "learning_rate": 6.245197698152571e-08,
      "loss": 0.0,
      "step": 5852
    },
    {
      "epoch": 0.9844420149693045,
      "grad_norm": NaN,
      "learning_rate": 6.111641663454903e-08,
      "loss": 0.0,
      "step": 5853
    },
    {
      "epoch": 0.9846102094020688,
      "grad_norm": NaN,
      "learning_rate": 5.979528307168414e-08,
      "loss": 0.0,
      "step": 5854
    },
    {
      "epoch": 0.9847784038348331,
      "grad_norm": NaN,
      "learning_rate": 5.84885766746035e-08,
      "loss": 0.0,
      "step": 5855
    },
    {
      "epoch": 0.9849465982675973,
      "grad_norm": NaN,
      "learning_rate": 5.7196297820794054e-08,
      "loss": 0.0,
      "step": 5856
    },
    {
      "epoch": 0.9851147927003616,
      "grad_norm": NaN,
      "learning_rate": 5.591844688358494e-08,
      "loss": 0.0,
      "step": 5857
    },
    {
      "epoch": 0.9852829871331259,
      "grad_norm": NaN,
      "learning_rate": 5.465502423213087e-08,
      "loss": 0.0,
      "step": 5858
    },
    {
      "epoch": 0.9854511815658902,
      "grad_norm": NaN,
      "learning_rate": 5.340603023141766e-08,
      "loss": 0.0,
      "step": 5859
    },
    {
      "epoch": 0.9856193759986545,
      "grad_norm": NaN,
      "learning_rate": 5.217146524226779e-08,
      "loss": 0.0,
      "step": 5860
    },
    {
      "epoch": 0.9857875704314187,
      "grad_norm": NaN,
      "learning_rate": 5.0951329621340416e-08,
      "loss": 0.0,
      "step": 5861
    },
    {
      "epoch": 0.985955764864183,
      "grad_norm": NaN,
      "learning_rate": 4.9745623721109135e-08,
      "loss": 0.0,
      "step": 5862
    },
    {
      "epoch": 0.9861239592969473,
      "grad_norm": NaN,
      "learning_rate": 4.855434788988977e-08,
      "loss": 0.0,
      "step": 5863
    },
    {
      "epoch": 0.9862921537297116,
      "grad_norm": NaN,
      "learning_rate": 4.737750247183481e-08,
      "loss": 0.0,
      "step": 5864
    },
    {
      "epoch": 0.9864603481624759,
      "grad_norm": NaN,
      "learning_rate": 4.621508780691119e-08,
      "loss": 0.0,
      "step": 5865
    },
    {
      "epoch": 0.9866285425952401,
      "grad_norm": NaN,
      "learning_rate": 4.506710423093918e-08,
      "loss": 0.0,
      "step": 5866
    },
    {
      "epoch": 0.9867967370280044,
      "grad_norm": NaN,
      "learning_rate": 4.39335520755535e-08,
      "loss": 0.0,
      "step": 5867
    },
    {
      "epoch": 0.9869649314607687,
      "grad_norm": NaN,
      "learning_rate": 4.281443166822552e-08,
      "loss": 0.0,
      "step": 5868
    },
    {
      "epoch": 0.987133125893533,
      "grad_norm": NaN,
      "learning_rate": 4.1709743332252196e-08,
      "loss": 0.0,
      "step": 5869
    },
    {
      "epoch": 0.9873013203262972,
      "grad_norm": NaN,
      "learning_rate": 4.061948738677268e-08,
      "loss": 0.0,
      "step": 5870
    },
    {
      "epoch": 0.9874695147590615,
      "grad_norm": NaN,
      "learning_rate": 3.9543664146746154e-08,
      "loss": 0.0,
      "step": 5871
    },
    {
      "epoch": 0.9876377091918257,
      "grad_norm": NaN,
      "learning_rate": 3.8482273922962884e-08,
      "loss": 0.0,
      "step": 5872
    },
    {
      "epoch": 0.98780590362459,
      "grad_norm": NaN,
      "learning_rate": 3.743531702204983e-08,
      "loss": 0.0,
      "step": 5873
    },
    {
      "epoch": 0.9879740980573543,
      "grad_norm": NaN,
      "learning_rate": 3.6402793746465045e-08,
      "loss": 0.0,
      "step": 5874
    },
    {
      "epoch": 0.9881422924901185,
      "grad_norm": NaN,
      "learning_rate": 3.538470439448105e-08,
      "loss": 0.0,
      "step": 5875
    },
    {
      "epoch": 0.9883104869228828,
      "grad_norm": NaN,
      "learning_rate": 3.438104926022923e-08,
      "loss": 0.0,
      "step": 5876
    },
    {
      "epoch": 0.9884786813556471,
      "grad_norm": NaN,
      "learning_rate": 3.339182863363877e-08,
      "loss": 0.0,
      "step": 5877
    },
    {
      "epoch": 0.9886468757884114,
      "grad_norm": NaN,
      "learning_rate": 3.241704280049218e-08,
      "loss": 0.0,
      "step": 5878
    },
    {
      "epoch": 0.9888150702211757,
      "grad_norm": NaN,
      "learning_rate": 3.145669204239754e-08,
      "loss": 0.0,
      "step": 5879
    },
    {
      "epoch": 0.9889832646539399,
      "grad_norm": NaN,
      "learning_rate": 3.051077663677737e-08,
      "loss": 0.0,
      "step": 5880
    },
    {
      "epoch": 0.9891514590867042,
      "grad_norm": NaN,
      "learning_rate": 2.9579296856907523e-08,
      "loss": 0.0,
      "step": 5881
    },
    {
      "epoch": 0.9893196535194685,
      "grad_norm": NaN,
      "learning_rate": 2.86622529718783e-08,
      "loss": 0.0,
      "step": 5882
    },
    {
      "epoch": 0.9894878479522328,
      "grad_norm": NaN,
      "learning_rate": 2.775964524661667e-08,
      "loss": 0.0,
      "step": 5883
    },
    {
      "epoch": 0.989656042384997,
      "grad_norm": NaN,
      "learning_rate": 2.6871473941864067e-08,
      "loss": 0.0,
      "step": 5884
    },
    {
      "epoch": 0.9898242368177613,
      "grad_norm": NaN,
      "learning_rate": 2.599773931422078e-08,
      "loss": 0.0,
      "step": 5885
    },
    {
      "epoch": 0.9899924312505256,
      "grad_norm": NaN,
      "learning_rate": 2.5138441616079367e-08,
      "loss": 0.0,
      "step": 5886
    },
    {
      "epoch": 0.9901606256832899,
      "grad_norm": NaN,
      "learning_rate": 2.4293581095696794e-08,
      "loss": 0.0,
      "step": 5887
    },
    {
      "epoch": 0.9903288201160542,
      "grad_norm": NaN,
      "learning_rate": 2.346315799713894e-08,
      "loss": 0.0,
      "step": 5888
    },
    {
      "epoch": 0.9904970145488184,
      "grad_norm": NaN,
      "learning_rate": 2.264717256030835e-08,
      "loss": 0.0,
      "step": 5889
    },
    {
      "epoch": 0.9906652089815827,
      "grad_norm": NaN,
      "learning_rate": 2.1845625020927572e-08,
      "loss": 0.0,
      "step": 5890
    },
    {
      "epoch": 0.990833403414347,
      "grad_norm": NaN,
      "learning_rate": 2.105851561056138e-08,
      "loss": 0.0,
      "step": 5891
    },
    {
      "epoch": 0.9910015978471113,
      "grad_norm": NaN,
      "learning_rate": 2.0285844556588996e-08,
      "loss": 0.0,
      "step": 5892
    },
    {
      "epoch": 0.9911697922798756,
      "grad_norm": NaN,
      "learning_rate": 1.952761208223186e-08,
      "loss": 0.0,
      "step": 5893
    },
    {
      "epoch": 0.9913379867126398,
      "grad_norm": NaN,
      "learning_rate": 1.878381840653698e-08,
      "loss": 0.0,
      "step": 5894
    },
    {
      "epoch": 0.9915061811454041,
      "grad_norm": NaN,
      "learning_rate": 1.8054463744376914e-08,
      "loss": 0.0,
      "step": 5895
    },
    {
      "epoch": 0.9916743755781684,
      "grad_norm": NaN,
      "learning_rate": 1.7339548306449794e-08,
      "loss": 0.0,
      "step": 5896
    },
    {
      "epoch": 0.9918425700109327,
      "grad_norm": NaN,
      "learning_rate": 1.6639072299284852e-08,
      "loss": 0.0,
      "step": 5897
    },
    {
      "epoch": 0.992010764443697,
      "grad_norm": NaN,
      "learning_rate": 1.5953035925253547e-08,
      "loss": 0.0,
      "step": 5898
    },
    {
      "epoch": 0.9921789588764612,
      "grad_norm": NaN,
      "learning_rate": 1.528143938253068e-08,
      "loss": 0.0,
      "step": 5899
    },
    {
      "epoch": 0.9923471533092255,
      "grad_norm": NaN,
      "learning_rate": 1.4624282865144389e-08,
      "loss": 0.0,
      "step": 5900
    },
    {
      "epoch": 0.9925153477419898,
      "grad_norm": NaN,
      "learning_rate": 1.3981566562931702e-08,
      "loss": 0.0,
      "step": 5901
    },
    {
      "epoch": 0.992683542174754,
      "grad_norm": NaN,
      "learning_rate": 1.3353290661571871e-08,
      "loss": 0.0,
      "step": 5902
    },
    {
      "epoch": 0.9928517366075182,
      "grad_norm": NaN,
      "learning_rate": 1.2739455342558603e-08,
      "loss": 0.0,
      "step": 5903
    },
    {
      "epoch": 0.9930199310402825,
      "grad_norm": NaN,
      "learning_rate": 1.2140060783227824e-08,
      "loss": 0.0,
      "step": 5904
    },
    {
      "epoch": 0.9931881254730468,
      "grad_norm": NaN,
      "learning_rate": 1.155510715674102e-08,
      "loss": 0.0,
      "step": 5905
    },
    {
      "epoch": 0.9933563199058111,
      "grad_norm": NaN,
      "learning_rate": 1.098459463207968e-08,
      "loss": 0.0,
      "step": 5906
    },
    {
      "epoch": 0.9935245143385754,
      "grad_norm": NaN,
      "learning_rate": 1.042852337406197e-08,
      "loss": 0.0,
      "step": 5907
    },
    {
      "epoch": 0.9936927087713396,
      "grad_norm": NaN,
      "learning_rate": 9.88689354332606e-09,
      "loss": 0.0,
      "step": 5908
    },
    {
      "epoch": 0.9938609032041039,
      "grad_norm": NaN,
      "learning_rate": 9.359705296346776e-09,
      "loss": 0.0,
      "step": 5909
    },
    {
      "epoch": 0.9940290976368682,
      "grad_norm": NaN,
      "learning_rate": 8.846958785418968e-09,
      "loss": 0.0,
      "step": 5910
    },
    {
      "epoch": 0.9941972920696325,
      "grad_norm": NaN,
      "learning_rate": 8.34865415867414e-09,
      "loss": 0.0,
      "step": 5911
    },
    {
      "epoch": 0.9943654865023968,
      "grad_norm": NaN,
      "learning_rate": 7.86479156006381e-09,
      "loss": 0.0,
      "step": 5912
    },
    {
      "epoch": 0.994533680935161,
      "grad_norm": NaN,
      "learning_rate": 7.3953711293706096e-09,
      "loss": 0.0,
      "step": 5913
    },
    {
      "epoch": 0.9947018753679253,
      "grad_norm": NaN,
      "learning_rate": 6.940393002202727e-09,
      "loss": 0.0,
      "step": 5914
    },
    {
      "epoch": 0.9948700698006896,
      "grad_norm": NaN,
      "learning_rate": 6.4998573100050195e-09,
      "loss": 0.0,
      "step": 5915
    },
    {
      "epoch": 0.9950382642334539,
      "grad_norm": NaN,
      "learning_rate": 6.0737641800368e-09,
      "loss": 0.0,
      "step": 5916
    },
    {
      "epoch": 0.9952064586662182,
      "grad_norm": NaN,
      "learning_rate": 5.662113735394048e-09,
      "loss": 0.0,
      "step": 5917
    },
    {
      "epoch": 0.9953746530989824,
      "grad_norm": NaN,
      "learning_rate": 5.264906095003852e-09,
      "loss": 0.0,
      "step": 5918
    },
    {
      "epoch": 0.9955428475317467,
      "grad_norm": NaN,
      "learning_rate": 4.8821413736022115e-09,
      "loss": 0.0,
      "step": 5919
    },
    {
      "epoch": 0.995711041964511,
      "grad_norm": NaN,
      "learning_rate": 4.51381968177289e-09,
      "loss": 0.0,
      "step": 5920
    },
    {
      "epoch": 0.9958792363972753,
      "grad_norm": NaN,
      "learning_rate": 4.159941125925215e-09,
      "loss": 0.0,
      "step": 5921
    },
    {
      "epoch": 0.9960474308300395,
      "grad_norm": NaN,
      "learning_rate": 3.820505808277419e-09,
      "loss": 0.0,
      "step": 5922
    },
    {
      "epoch": 0.9962156252628038,
      "grad_norm": NaN,
      "learning_rate": 3.4955138269010534e-09,
      "loss": 0.0,
      "step": 5923
    },
    {
      "epoch": 0.9963838196955681,
      "grad_norm": NaN,
      "learning_rate": 3.184965275676577e-09,
      "loss": 0.0,
      "step": 5924
    },
    {
      "epoch": 0.9965520141283324,
      "grad_norm": NaN,
      "learning_rate": 2.8888602443211122e-09,
      "loss": 0.0,
      "step": 5925
    },
    {
      "epoch": 0.9967202085610967,
      "grad_norm": NaN,
      "learning_rate": 2.607198818371792e-09,
      "loss": 0.0,
      "step": 5926
    },
    {
      "epoch": 0.996888402993861,
      "grad_norm": NaN,
      "learning_rate": 2.3399810792024133e-09,
      "loss": 0.0,
      "step": 5927
    },
    {
      "epoch": 0.9970565974266252,
      "grad_norm": NaN,
      "learning_rate": 2.087207104001232e-09,
      "loss": 0.0,
      "step": 5928
    },
    {
      "epoch": 0.9972247918593895,
      "grad_norm": NaN,
      "learning_rate": 1.8488769658042693e-09,
      "loss": 0.0,
      "step": 5929
    },
    {
      "epoch": 0.9973929862921538,
      "grad_norm": NaN,
      "learning_rate": 1.624990733450904e-09,
      "loss": 0.0,
      "step": 5930
    },
    {
      "epoch": 0.997561180724918,
      "grad_norm": NaN,
      "learning_rate": 1.4155484716227296e-09,
      "loss": 0.0,
      "step": 5931
    },
    {
      "epoch": 0.9977293751576822,
      "grad_norm": NaN,
      "learning_rate": 1.220550240826901e-09,
      "loss": 0.0,
      "step": 5932
    },
    {
      "epoch": 0.9978975695904465,
      "grad_norm": NaN,
      "learning_rate": 1.0399960974016854e-09,
      "loss": 0.0,
      "step": 5933
    },
    {
      "epoch": 0.9980657640232108,
      "grad_norm": NaN,
      "learning_rate": 8.738860934942584e-10,
      "loss": 0.0,
      "step": 5934
    },
    {
      "epoch": 0.9982339584559751,
      "grad_norm": NaN,
      "learning_rate": 7.222202770995612e-10,
      "loss": 0.0,
      "step": 5935
    },
    {
      "epoch": 0.9984021528887393,
      "grad_norm": NaN,
      "learning_rate": 5.849986920325456e-10,
      "loss": 0.0,
      "step": 5936
    },
    {
      "epoch": 0.9985703473215036,
      "grad_norm": NaN,
      "learning_rate": 4.622213779392759e-10,
      "loss": 0.0,
      "step": 5937
    },
    {
      "epoch": 0.9987385417542679,
      "grad_norm": NaN,
      "learning_rate": 3.538883702747242e-10,
      "loss": 0.0,
      "step": 5938
    },
    {
      "epoch": 0.9989067361870322,
      "grad_norm": NaN,
      "learning_rate": 2.599997003471799e-10,
      "loss": 0.0,
      "step": 5939
    },
    {
      "epoch": 0.9990749306197965,
      "grad_norm": NaN,
      "learning_rate": 1.8055539527939148e-10,
      "loss": 0.0,
      "step": 5940
    },
    {
      "epoch": 0.9992431250525607,
      "grad_norm": NaN,
      "learning_rate": 1.155554780141177e-10,
      "loss": 0.0,
      "step": 5941
    },
    {
      "epoch": 0.999411319485325,
      "grad_norm": NaN,
      "learning_rate": 6.499996733633218e-11,
      "loss": 0.0,
      "step": 5942
    },
    {
      "epoch": 0.9995795139180893,
      "grad_norm": NaN,
      "learning_rate": 2.8888877851018792e-11,
      "loss": 0.0,
      "step": 5943
    },
    {
      "epoch": 0.9997477083508536,
      "grad_norm": NaN,
      "learning_rate": 7.2222199831717406e-12,
      "loss": 0.0,
      "step": 5944
    },
    {
      "epoch": 0.9999159027836179,
      "grad_norm": NaN,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 5945
    }
  ],
  "logging_steps": 1,
  "max_steps": 5945,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 239,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1191100899590144e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}