{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 24.137931034482758,
  "eval_steps": 50,
  "global_step": 700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 4.955975532531738,
      "learning_rate": 0.0,
      "loss": 1.5752,
      "step": 1
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 3.4605445861816406,
      "learning_rate": 1.2500000000000002e-07,
      "loss": 1.6145,
      "step": 2
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 3.622607946395874,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 1.6033,
      "step": 3
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 6.3884453773498535,
      "learning_rate": 3.75e-07,
      "loss": 1.6219,
      "step": 4
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 3.212383270263672,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.5958,
      "step": 5
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 6.6238508224487305,
      "learning_rate": 6.25e-07,
      "loss": 1.6268,
      "step": 6
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 7.565281391143799,
      "learning_rate": 7.5e-07,
      "loss": 1.6133,
      "step": 7
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 11.01657772064209,
      "learning_rate": 8.750000000000001e-07,
      "loss": 1.6061,
      "step": 8
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 8.828049659729004,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.6352,
      "step": 9
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 3.2700514793395996,
      "learning_rate": 1.125e-06,
      "loss": 1.5985,
      "step": 10
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 7.4744086265563965,
      "learning_rate": 1.25e-06,
      "loss": 1.6075,
      "step": 11
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 5.462789535522461,
      "learning_rate": 1.3750000000000002e-06,
      "loss": 1.6174,
      "step": 12
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 3.593034267425537,
      "learning_rate": 1.5e-06,
      "loss": 1.6083,
      "step": 13
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 4.305792331695557,
      "learning_rate": 1.6250000000000001e-06,
      "loss": 1.5997,
      "step": 14
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 8.656575202941895,
      "learning_rate": 1.7500000000000002e-06,
      "loss": 1.6283,
      "step": 15
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 6.6580491065979,
      "learning_rate": 1.875e-06,
      "loss": 1.5906,
      "step": 16
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 1.8648769855499268,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.5667,
      "step": 17
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 8.883339881896973,
      "learning_rate": 2.1250000000000004e-06,
      "loss": 1.5999,
      "step": 18
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 3.959343194961548,
      "learning_rate": 2.25e-06,
      "loss": 1.5751,
      "step": 19
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 5.278771877288818,
      "learning_rate": 2.375e-06,
      "loss": 1.6216,
      "step": 20
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 3.716949224472046,
      "learning_rate": 2.5e-06,
      "loss": 1.613,
      "step": 21
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 1.6308438777923584,
      "learning_rate": 2.625e-06,
      "loss": 1.5822,
      "step": 22
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 1.8407272100448608,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 1.5369,
      "step": 23
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 7.528758525848389,
      "learning_rate": 2.8750000000000004e-06,
      "loss": 1.584,
      "step": 24
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 3.860356569290161,
      "learning_rate": 3e-06,
      "loss": 1.58,
      "step": 25
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 1.7416785955429077,
      "learning_rate": 3.125e-06,
      "loss": 1.5714,
      "step": 26
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 4.024614334106445,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 1.5905,
      "step": 27
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 5.027279376983643,
      "learning_rate": 3.3750000000000003e-06,
      "loss": 1.5906,
      "step": 28
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.354084014892578,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 1.5781,
      "step": 29
    },
    {
      "epoch": 1.0344827586206897,
      "grad_norm": 1.950278401374817,
      "learning_rate": 3.625e-06,
      "loss": 1.5747,
      "step": 30
    },
    {
      "epoch": 1.0689655172413792,
      "grad_norm": 3.3933162689208984,
      "learning_rate": 3.75e-06,
      "loss": 1.5642,
      "step": 31
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 4.551718235015869,
      "learning_rate": 3.875e-06,
      "loss": 1.5872,
      "step": 32
    },
    {
      "epoch": 1.1379310344827587,
      "grad_norm": 3.4449899196624756,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.5755,
      "step": 33
    },
    {
      "epoch": 1.1724137931034484,
      "grad_norm": 2.4862890243530273,
      "learning_rate": 4.125e-06,
      "loss": 1.5661,
      "step": 34
    },
    {
      "epoch": 1.206896551724138,
      "grad_norm": 3.3560233116149902,
      "learning_rate": 4.250000000000001e-06,
      "loss": 1.5813,
      "step": 35
    },
    {
      "epoch": 1.2413793103448276,
      "grad_norm": 1.7421871423721313,
      "learning_rate": 4.375e-06,
      "loss": 1.5653,
      "step": 36
    },
    {
      "epoch": 1.2758620689655173,
      "grad_norm": 1.8098750114440918,
      "learning_rate": 4.5e-06,
      "loss": 1.5686,
      "step": 37
    },
    {
      "epoch": 1.3103448275862069,
      "grad_norm": 5.115236759185791,
      "learning_rate": 4.625e-06,
      "loss": 1.5678,
      "step": 38
    },
    {
      "epoch": 1.3448275862068966,
      "grad_norm": 3.044386625289917,
      "learning_rate": 4.75e-06,
      "loss": 1.5634,
      "step": 39
    },
    {
      "epoch": 1.3793103448275863,
      "grad_norm": 5.8161139488220215,
      "learning_rate": 4.875000000000001e-06,
      "loss": 1.5846,
      "step": 40
    },
    {
      "epoch": 1.4137931034482758,
      "grad_norm": 1.6947190761566162,
      "learning_rate": 5e-06,
      "loss": 1.5512,
      "step": 41
    },
    {
      "epoch": 1.4482758620689655,
      "grad_norm": 3.2046549320220947,
      "learning_rate": 5.125e-06,
      "loss": 1.5579,
      "step": 42
    },
    {
      "epoch": 1.4827586206896552,
      "grad_norm": 2.9457404613494873,
      "learning_rate": 5.25e-06,
      "loss": 1.5661,
      "step": 43
    },
    {
      "epoch": 1.5172413793103448,
      "grad_norm": 1.604246735572815,
      "learning_rate": 5.375e-06,
      "loss": 1.5584,
      "step": 44
    },
    {
      "epoch": 1.5517241379310345,
      "grad_norm": 1.433769702911377,
      "learning_rate": 5.500000000000001e-06,
      "loss": 1.5398,
      "step": 45
    },
    {
      "epoch": 1.5862068965517242,
      "grad_norm": 1.4531933069229126,
      "learning_rate": 5.625e-06,
      "loss": 1.5459,
      "step": 46
    },
    {
      "epoch": 1.6206896551724137,
      "grad_norm": 3.0887269973754883,
      "learning_rate": 5.750000000000001e-06,
      "loss": 1.5475,
      "step": 47
    },
    {
      "epoch": 1.6551724137931034,
      "grad_norm": 1.7311522960662842,
      "learning_rate": 5.875e-06,
      "loss": 1.5395,
      "step": 48
    },
    {
      "epoch": 1.6896551724137931,
      "grad_norm": 1.57200026512146,
      "learning_rate": 6e-06,
      "loss": 1.5355,
      "step": 49
    },
    {
      "epoch": 1.7241379310344827,
      "grad_norm": 6.282898426055908,
      "learning_rate": 6.125e-06,
      "loss": 1.5479,
      "step": 50
    },
    {
      "epoch": 1.7586206896551724,
      "grad_norm": 5.877266883850098,
      "learning_rate": 6.25e-06,
      "loss": 1.535,
      "step": 51
    },
    {
      "epoch": 1.793103448275862,
      "grad_norm": 2.238494396209717,
      "learning_rate": 6.375000000000001e-06,
      "loss": 1.5439,
      "step": 52
    },
    {
      "epoch": 1.8275862068965516,
      "grad_norm": 3.176295518875122,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 1.5411,
      "step": 53
    },
    {
      "epoch": 1.8620689655172413,
      "grad_norm": 1.5491082668304443,
      "learning_rate": 6.625000000000001e-06,
      "loss": 1.5279,
      "step": 54
    },
    {
      "epoch": 1.896551724137931,
      "grad_norm": 2.956325054168701,
      "learning_rate": 6.750000000000001e-06,
      "loss": 1.5381,
      "step": 55
    },
    {
      "epoch": 1.9310344827586206,
      "grad_norm": 5.207773685455322,
      "learning_rate": 6.875000000000001e-06,
      "loss": 1.5499,
      "step": 56
    },
    {
      "epoch": 1.9655172413793105,
      "grad_norm": 1.360984444618225,
      "learning_rate": 7.000000000000001e-06,
      "loss": 1.5294,
      "step": 57
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9358643293380737,
      "learning_rate": 7.1249999999999995e-06,
      "loss": 1.5492,
      "step": 58
    },
    {
      "epoch": 2.0344827586206895,
      "grad_norm": 1.5760064125061035,
      "learning_rate": 7.25e-06,
      "loss": 1.5279,
      "step": 59
    },
    {
      "epoch": 2.0689655172413794,
      "grad_norm": 2.5182933807373047,
      "learning_rate": 7.375e-06,
      "loss": 1.5272,
      "step": 60
    },
    {
      "epoch": 2.103448275862069,
      "grad_norm": 1.9873909950256348,
      "learning_rate": 7.5e-06,
      "loss": 1.5298,
      "step": 61
    },
    {
      "epoch": 2.1379310344827585,
      "grad_norm": 1.4847761392593384,
      "learning_rate": 7.625e-06,
      "loss": 1.5129,
      "step": 62
    },
    {
      "epoch": 2.1724137931034484,
      "grad_norm": 1.4767472743988037,
      "learning_rate": 7.75e-06,
      "loss": 1.5255,
      "step": 63
    },
    {
      "epoch": 2.206896551724138,
      "grad_norm": 4.426845550537109,
      "learning_rate": 7.875e-06,
      "loss": 1.5277,
      "step": 64
    },
    {
      "epoch": 2.2413793103448274,
      "grad_norm": 2.6555216312408447,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.5278,
      "step": 65
    },
    {
      "epoch": 2.2758620689655173,
      "grad_norm": 2.148672580718994,
      "learning_rate": 8.125000000000001e-06,
      "loss": 1.5202,
      "step": 66
    },
    {
      "epoch": 2.310344827586207,
      "grad_norm": 2.529113531112671,
      "learning_rate": 8.25e-06,
      "loss": 1.533,
      "step": 67
    },
    {
      "epoch": 2.344827586206897,
      "grad_norm": 2.6022942066192627,
      "learning_rate": 8.375e-06,
      "loss": 1.5221,
      "step": 68
    },
    {
      "epoch": 2.3793103448275863,
      "grad_norm": 4.527515888214111,
      "learning_rate": 8.500000000000002e-06,
      "loss": 1.5143,
      "step": 69
    },
    {
      "epoch": 2.413793103448276,
      "grad_norm": 3.554746627807617,
      "learning_rate": 8.625e-06,
      "loss": 1.5105,
      "step": 70
    },
    {
      "epoch": 2.4482758620689653,
      "grad_norm": 2.6532673835754395,
      "learning_rate": 8.75e-06,
      "loss": 1.4933,
      "step": 71
    },
    {
      "epoch": 2.4827586206896552,
      "grad_norm": 3.799637794494629,
      "learning_rate": 8.875e-06,
      "loss": 1.5067,
      "step": 72
    },
    {
      "epoch": 2.5172413793103448,
      "grad_norm": 3.7445719242095947,
      "learning_rate": 9e-06,
      "loss": 1.5193,
      "step": 73
    },
    {
      "epoch": 2.5517241379310347,
      "grad_norm": 1.5109844207763672,
      "learning_rate": 9.125e-06,
      "loss": 1.5219,
      "step": 74
    },
    {
      "epoch": 2.586206896551724,
      "grad_norm": 1.2224950790405273,
      "learning_rate": 9.25e-06,
      "loss": 1.4971,
      "step": 75
    },
    {
      "epoch": 2.6206896551724137,
      "grad_norm": 5.30598783493042,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.4901,
      "step": 76
    },
    {
      "epoch": 2.655172413793103,
      "grad_norm": 1.4342738389968872,
      "learning_rate": 9.5e-06,
      "loss": 1.5118,
      "step": 77
    },
    {
      "epoch": 2.689655172413793,
      "grad_norm": 4.92336368560791,
      "learning_rate": 9.625e-06,
      "loss": 1.4859,
      "step": 78
    },
    {
      "epoch": 2.7241379310344827,
      "grad_norm": 10.244889259338379,
      "learning_rate": 9.750000000000002e-06,
      "loss": 1.5147,
      "step": 79
    },
    {
      "epoch": 2.7586206896551726,
      "grad_norm": 5.509783744812012,
      "learning_rate": 9.875000000000001e-06,
      "loss": 1.5108,
      "step": 80
    },
    {
      "epoch": 2.793103448275862,
      "grad_norm": 1.6187537908554077,
      "learning_rate": 1e-05,
      "loss": 1.4753,
      "step": 81
    },
    {
      "epoch": 2.8275862068965516,
      "grad_norm": 1.2228670120239258,
      "learning_rate": 1.0125e-05,
      "loss": 1.5034,
      "step": 82
    },
    {
      "epoch": 2.862068965517241,
      "grad_norm": 2.5575168132781982,
      "learning_rate": 1.025e-05,
      "loss": 1.515,
      "step": 83
    },
    {
      "epoch": 2.896551724137931,
      "grad_norm": 2.906822919845581,
      "learning_rate": 1.0375e-05,
      "loss": 1.5243,
      "step": 84
    },
    {
      "epoch": 2.9310344827586206,
      "grad_norm": 2.380941152572632,
      "learning_rate": 1.05e-05,
      "loss": 1.4967,
      "step": 85
    },
    {
      "epoch": 2.9655172413793105,
      "grad_norm": 2.5720791816711426,
      "learning_rate": 1.0625e-05,
      "loss": 1.5065,
      "step": 86
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.6703929901123047,
      "learning_rate": 1.075e-05,
      "loss": 1.4984,
      "step": 87
    },
    {
      "epoch": 3.0344827586206895,
      "grad_norm": 3.5005722045898438,
      "learning_rate": 1.0875e-05,
      "loss": 1.5122,
      "step": 88
    },
    {
      "epoch": 3.0689655172413794,
      "grad_norm": 4.101701736450195,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 1.5057,
      "step": 89
    },
    {
      "epoch": 3.103448275862069,
      "grad_norm": 1.8032540082931519,
      "learning_rate": 1.1125000000000001e-05,
      "loss": 1.4953,
      "step": 90
    },
    {
      "epoch": 3.1379310344827585,
      "grad_norm": 1.8906192779541016,
      "learning_rate": 1.125e-05,
      "loss": 1.4885,
      "step": 91
    },
    {
      "epoch": 3.1724137931034484,
      "grad_norm": 4.843202114105225,
      "learning_rate": 1.1375e-05,
      "loss": 1.4819,
      "step": 92
    },
    {
      "epoch": 3.206896551724138,
      "grad_norm": 2.6482224464416504,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 1.4766,
      "step": 93
    },
    {
      "epoch": 3.2413793103448274,
      "grad_norm": 2.58847975730896,
      "learning_rate": 1.1625000000000001e-05,
      "loss": 1.4883,
      "step": 94
    },
    {
      "epoch": 3.2758620689655173,
      "grad_norm": 2.6886796951293945,
      "learning_rate": 1.175e-05,
      "loss": 1.4912,
      "step": 95
    },
    {
      "epoch": 3.310344827586207,
      "grad_norm": 4.70560359954834,
      "learning_rate": 1.1875e-05,
      "loss": 1.4835,
      "step": 96
    },
    {
      "epoch": 3.344827586206897,
      "grad_norm": 1.3306467533111572,
      "learning_rate": 1.2e-05,
      "loss": 1.4993,
      "step": 97
    },
    {
      "epoch": 3.3793103448275863,
      "grad_norm": 6.546566009521484,
      "learning_rate": 1.2125e-05,
      "loss": 1.4564,
      "step": 98
    },
    {
      "epoch": 3.413793103448276,
      "grad_norm": 2.113272190093994,
      "learning_rate": 1.225e-05,
      "loss": 1.4483,
      "step": 99
    },
    {
      "epoch": 3.4482758620689653,
      "grad_norm": 3.0866124629974365,
      "learning_rate": 1.2375000000000001e-05,
      "loss": 1.4824,
      "step": 100
    },
    {
      "epoch": 3.4827586206896552,
      "grad_norm": 1.7086529731750488,
      "learning_rate": 1.25e-05,
      "loss": 1.4845,
      "step": 101
    },
    {
      "epoch": 3.5172413793103448,
      "grad_norm": 1.3498625755310059,
      "learning_rate": 1.2625e-05,
      "loss": 1.4632,
      "step": 102
    },
    {
      "epoch": 3.5517241379310347,
      "grad_norm": 2.0302608013153076,
      "learning_rate": 1.2750000000000002e-05,
      "loss": 1.4623,
      "step": 103
    },
    {
      "epoch": 3.586206896551724,
      "grad_norm": 1.4253138303756714,
      "learning_rate": 1.2875000000000001e-05,
      "loss": 1.4459,
      "step": 104
    },
    {
      "epoch": 3.6206896551724137,
      "grad_norm": 2.8165297508239746,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 1.448,
      "step": 105
    },
    {
      "epoch": 3.655172413793103,
      "grad_norm": 2.0474743843078613,
      "learning_rate": 1.3125e-05,
      "loss": 1.4324,
      "step": 106
    },
    {
      "epoch": 3.689655172413793,
      "grad_norm": 1.4272180795669556,
      "learning_rate": 1.3250000000000002e-05,
      "loss": 1.4328,
      "step": 107
    },
    {
      "epoch": 3.7241379310344827,
      "grad_norm": 5.992876052856445,
      "learning_rate": 1.3375000000000002e-05,
      "loss": 1.4314,
      "step": 108
    },
    {
      "epoch": 3.7586206896551726,
      "grad_norm": 3.8637657165527344,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 1.4809,
      "step": 109
    },
    {
      "epoch": 3.793103448275862,
      "grad_norm": 4.207685947418213,
      "learning_rate": 1.3625e-05,
      "loss": 1.4811,
      "step": 110
    },
    {
      "epoch": 3.8275862068965516,
      "grad_norm": 1.545461654663086,
      "learning_rate": 1.3750000000000002e-05,
      "loss": 1.4942,
      "step": 111
    },
    {
      "epoch": 3.862068965517241,
      "grad_norm": 1.6922414302825928,
      "learning_rate": 1.3875000000000002e-05,
      "loss": 1.4439,
      "step": 112
    },
    {
      "epoch": 3.896551724137931,
      "grad_norm": 1.6125301122665405,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 1.4745,
      "step": 113
    },
    {
      "epoch": 3.9310344827586206,
      "grad_norm": 3.392108678817749,
      "learning_rate": 1.4125e-05,
      "loss": 1.4744,
      "step": 114
    },
    {
      "epoch": 3.9655172413793105,
      "grad_norm": 2.7309863567352295,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 1.4439,
      "step": 115
    },
    {
      "epoch": 4.0,
      "grad_norm": 6.303447246551514,
      "learning_rate": 1.4374999999999999e-05,
      "loss": 1.461,
      "step": 116
    },
    {
      "epoch": 4.0344827586206895,
      "grad_norm": 1.5163921117782593,
      "learning_rate": 1.45e-05,
      "loss": 1.4461,
      "step": 117
    },
    {
      "epoch": 4.068965517241379,
      "grad_norm": 2.4618422985076904,
      "learning_rate": 1.4625e-05,
      "loss": 1.4361,
      "step": 118
    },
    {
      "epoch": 4.103448275862069,
      "grad_norm": 2.786285877227783,
      "learning_rate": 1.475e-05,
      "loss": 1.4183,
      "step": 119
    },
    {
      "epoch": 4.137931034482759,
      "grad_norm": 5.21203088760376,
      "learning_rate": 1.4875e-05,
      "loss": 1.4407,
      "step": 120
    },
    {
      "epoch": 4.172413793103448,
      "grad_norm": 1.920630931854248,
      "learning_rate": 1.5e-05,
      "loss": 1.3928,
      "step": 121
    },
    {
      "epoch": 4.206896551724138,
      "grad_norm": 2.916647434234619,
      "learning_rate": 1.5125e-05,
      "loss": 1.409,
      "step": 122
    },
    {
      "epoch": 4.241379310344827,
      "grad_norm": 4.050491809844971,
      "learning_rate": 1.525e-05,
      "loss": 1.4555,
      "step": 123
    },
    {
      "epoch": 4.275862068965517,
      "grad_norm": 3.479999303817749,
      "learning_rate": 1.5375e-05,
      "loss": 1.396,
      "step": 124
    },
    {
      "epoch": 4.310344827586207,
      "grad_norm": 5.414281368255615,
      "learning_rate": 1.55e-05,
      "loss": 1.3829,
      "step": 125
    },
    {
      "epoch": 4.344827586206897,
      "grad_norm": 4.517972946166992,
      "learning_rate": 1.5625e-05,
      "loss": 1.3182,
      "step": 126
    },
    {
      "epoch": 4.379310344827586,
      "grad_norm": 10.26574993133545,
      "learning_rate": 1.575e-05,
      "loss": 1.4205,
      "step": 127
    },
    {
      "epoch": 4.413793103448276,
      "grad_norm": 2.3254988193511963,
      "learning_rate": 1.5875e-05,
      "loss": 1.2768,
      "step": 128
    },
    {
      "epoch": 4.448275862068965,
      "grad_norm": 5.00213098526001,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.3419,
      "step": 129
    },
    {
      "epoch": 4.482758620689655,
      "grad_norm": 11.136003494262695,
      "learning_rate": 1.6125000000000002e-05,
      "loss": 1.2928,
      "step": 130
    },
    {
      "epoch": 4.517241379310345,
      "grad_norm": 6.113260269165039,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 1.2989,
      "step": 131
    },
    {
      "epoch": 4.551724137931035,
      "grad_norm": 5.029887676239014,
      "learning_rate": 1.6375e-05,
      "loss": 1.3367,
      "step": 132
    },
    {
      "epoch": 4.586206896551724,
      "grad_norm": 15.060640335083008,
      "learning_rate": 1.65e-05,
      "loss": 1.4661,
      "step": 133
    },
    {
      "epoch": 4.620689655172414,
      "grad_norm": 18.83147430419922,
      "learning_rate": 1.6625e-05,
      "loss": 1.5725,
      "step": 134
    },
    {
      "epoch": 4.655172413793103,
      "grad_norm": 8.334407806396484,
      "learning_rate": 1.675e-05,
      "loss": 1.3389,
      "step": 135
    },
    {
      "epoch": 4.689655172413794,
      "grad_norm": 5.727024555206299,
      "learning_rate": 1.6875000000000004e-05,
      "loss": 1.3305,
      "step": 136
    },
    {
      "epoch": 4.724137931034483,
      "grad_norm": 6.0667009353637695,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 1.2844,
      "step": 137
    },
    {
      "epoch": 4.758620689655173,
      "grad_norm": 8.634448051452637,
      "learning_rate": 1.7125000000000003e-05,
      "loss": 1.2871,
      "step": 138
    },
    {
      "epoch": 4.793103448275862,
      "grad_norm": 6.196165561676025,
      "learning_rate": 1.725e-05,
      "loss": 1.2925,
      "step": 139
    },
    {
      "epoch": 4.827586206896552,
      "grad_norm": 4.187297821044922,
      "learning_rate": 1.7375e-05,
      "loss": 1.2614,
      "step": 140
    },
    {
      "epoch": 4.862068965517241,
      "grad_norm": 4.917987823486328,
      "learning_rate": 1.75e-05,
      "loss": 1.2843,
      "step": 141
    },
    {
      "epoch": 4.896551724137931,
      "grad_norm": 3.2452642917633057,
      "learning_rate": 1.7625e-05,
      "loss": 1.2741,
      "step": 142
    },
    {
      "epoch": 4.931034482758621,
      "grad_norm": 6.927144527435303,
      "learning_rate": 1.775e-05,
      "loss": 1.2967,
      "step": 143
    },
    {
      "epoch": 4.9655172413793105,
      "grad_norm": 4.876591682434082,
      "learning_rate": 1.7875e-05,
      "loss": 1.1916,
      "step": 144
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.315114736557007,
      "learning_rate": 1.8e-05,
      "loss": 1.1966,
      "step": 145
    },
    {
      "epoch": 5.0344827586206895,
      "grad_norm": 4.320967674255371,
      "learning_rate": 1.8125e-05,
      "loss": 1.2583,
      "step": 146
    },
    {
      "epoch": 5.068965517241379,
      "grad_norm": 3.7428784370422363,
      "learning_rate": 1.825e-05,
      "loss": 1.2523,
      "step": 147
    },
    {
      "epoch": 5.103448275862069,
      "grad_norm": 4.975188255310059,
      "learning_rate": 1.8375e-05,
      "loss": 1.2224,
      "step": 148
    },
    {
      "epoch": 5.137931034482759,
      "grad_norm": 5.255687713623047,
      "learning_rate": 1.85e-05,
      "loss": 1.1885,
      "step": 149
    },
    {
      "epoch": 5.172413793103448,
      "grad_norm": 7.562132835388184,
      "learning_rate": 1.8625000000000002e-05,
      "loss": 1.1543,
      "step": 150
    },
    {
      "epoch": 5.206896551724138,
      "grad_norm": 4.356451988220215,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 1.2019,
      "step": 151
    },
    {
      "epoch": 5.241379310344827,
      "grad_norm": 8.989137649536133,
      "learning_rate": 1.8875e-05,
      "loss": 1.378,
      "step": 152
    },
    {
      "epoch": 5.275862068965517,
      "grad_norm": 2.729421615600586,
      "learning_rate": 1.9e-05,
      "loss": 1.1547,
      "step": 153
    },
    {
      "epoch": 5.310344827586207,
      "grad_norm": 2.701746702194214,
      "learning_rate": 1.9125e-05,
      "loss": 1.184,
      "step": 154
    },
    {
      "epoch": 5.344827586206897,
      "grad_norm": 4.610611915588379,
      "learning_rate": 1.925e-05,
      "loss": 1.2343,
      "step": 155
    },
    {
      "epoch": 5.379310344827586,
      "grad_norm": 5.182763576507568,
      "learning_rate": 1.9375e-05,
      "loss": 1.189,
      "step": 156
    },
    {
      "epoch": 5.413793103448276,
      "grad_norm": 5.774525165557861,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 1.2143,
      "step": 157
    },
    {
      "epoch": 5.448275862068965,
      "grad_norm": 3.1842565536499023,
      "learning_rate": 1.9625000000000003e-05,
      "loss": 1.1613,
      "step": 158
    },
    {
      "epoch": 5.482758620689655,
      "grad_norm": 5.885169982910156,
      "learning_rate": 1.9750000000000002e-05,
      "loss": 1.24,
      "step": 159
    },
    {
      "epoch": 5.517241379310345,
      "grad_norm": 8.06775951385498,
      "learning_rate": 1.9875000000000002e-05,
      "loss": 1.1973,
      "step": 160
    },
    {
      "epoch": 5.551724137931035,
      "grad_norm": 3.398749589920044,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 161
    },
    {
      "epoch": 5.586206896551724,
      "grad_norm": 3.8745975494384766,
      "learning_rate": 2.0125e-05,
      "loss": 1.1218,
      "step": 162
    },
    {
      "epoch": 5.620689655172414,
      "grad_norm": 6.615961074829102,
      "learning_rate": 2.025e-05,
      "loss": 1.1056,
      "step": 163
    },
    {
      "epoch": 5.655172413793103,
      "grad_norm": 4.613058090209961,
      "learning_rate": 2.0375e-05,
      "loss": 1.0493,
      "step": 164
    },
    {
      "epoch": 5.689655172413794,
      "grad_norm": 4.730791091918945,
      "learning_rate": 2.05e-05,
      "loss": 1.0686,
      "step": 165
    },
    {
      "epoch": 5.724137931034483,
      "grad_norm": 6.086816310882568,
      "learning_rate": 2.0625e-05,
      "loss": 1.084,
      "step": 166
    },
    {
      "epoch": 5.758620689655173,
      "grad_norm": 5.453216552734375,
      "learning_rate": 2.075e-05,
      "loss": 1.0637,
      "step": 167
    },
    {
      "epoch": 5.793103448275862,
      "grad_norm": 4.013542652130127,
      "learning_rate": 2.0875e-05,
      "loss": 1.0962,
      "step": 168
    },
    {
      "epoch": 5.827586206896552,
      "grad_norm": 3.824021816253662,
      "learning_rate": 2.1e-05,
      "loss": 1.0632,
      "step": 169
    },
    {
      "epoch": 5.862068965517241,
      "grad_norm": 3.00637149810791,
      "learning_rate": 2.1125000000000002e-05,
      "loss": 1.0575,
      "step": 170
    },
    {
      "epoch": 5.896551724137931,
      "grad_norm": 4.230837821960449,
      "learning_rate": 2.125e-05,
      "loss": 1.0095,
      "step": 171
    },
    {
      "epoch": 5.931034482758621,
      "grad_norm": 4.6836371421813965,
      "learning_rate": 2.1375e-05,
      "loss": 1.1063,
      "step": 172
    },
    {
      "epoch": 5.9655172413793105,
      "grad_norm": 3.750352382659912,
      "learning_rate": 2.15e-05,
      "loss": 1.1189,
      "step": 173
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.330487251281738,
      "learning_rate": 2.1625e-05,
      "loss": 1.0574,
      "step": 174
    },
    {
      "epoch": 6.0344827586206895,
      "grad_norm": 5.845319747924805,
      "learning_rate": 2.175e-05,
      "loss": 1.08,
      "step": 175
    },
    {
      "epoch": 6.068965517241379,
      "grad_norm": 3.027203321456909,
      "learning_rate": 2.1875e-05,
      "loss": 0.9774,
      "step": 176
    },
    {
      "epoch": 6.103448275862069,
      "grad_norm": 3.455016851425171,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 1.0531,
      "step": 177
    },
    {
      "epoch": 6.137931034482759,
      "grad_norm": 3.488523483276367,
      "learning_rate": 2.2125000000000002e-05,
      "loss": 0.9497,
      "step": 178
    },
    {
      "epoch": 6.172413793103448,
      "grad_norm": 3.371523141860962,
      "learning_rate": 2.2250000000000002e-05,
      "loss": 1.0177,
      "step": 179
    },
    {
      "epoch": 6.206896551724138,
      "grad_norm": 5.724728107452393,
      "learning_rate": 2.2375000000000002e-05,
      "loss": 1.1333,
      "step": 180
    },
    {
      "epoch": 6.241379310344827,
      "grad_norm": 3.2824714183807373,
      "learning_rate": 2.25e-05,
      "loss": 1.0047,
      "step": 181
    },
    {
      "epoch": 6.275862068965517,
      "grad_norm": 7.460841178894043,
      "learning_rate": 2.2625e-05,
      "loss": 1.1217,
      "step": 182
    },
    {
      "epoch": 6.310344827586207,
      "grad_norm": 4.130681037902832,
      "learning_rate": 2.275e-05,
      "loss": 1.066,
      "step": 183
    },
    {
      "epoch": 6.344827586206897,
      "grad_norm": 3.287168025970459,
      "learning_rate": 2.2875e-05,
      "loss": 0.9983,
      "step": 184
    },
    {
      "epoch": 6.379310344827586,
      "grad_norm": 4.445807456970215,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 1.1108,
      "step": 185
    },
    {
      "epoch": 6.413793103448276,
      "grad_norm": 4.80116081237793,
      "learning_rate": 2.3125000000000003e-05,
      "loss": 1.0195,
      "step": 186
    },
    {
      "epoch": 6.448275862068965,
      "grad_norm": 3.579498529434204,
      "learning_rate": 2.3250000000000003e-05,
      "loss": 1.0323,
      "step": 187
    },
    {
      "epoch": 6.482758620689655,
      "grad_norm": 3.3527798652648926,
      "learning_rate": 2.3375000000000002e-05,
      "loss": 0.976,
      "step": 188
    },
    {
      "epoch": 6.517241379310345,
      "grad_norm": 2.929718017578125,
      "learning_rate": 2.35e-05,
      "loss": 0.9717,
      "step": 189
    },
    {
      "epoch": 6.551724137931035,
      "grad_norm": 4.6633758544921875,
      "learning_rate": 2.3624999999999998e-05,
      "loss": 1.1071,
      "step": 190
    },
    {
      "epoch": 6.586206896551724,
      "grad_norm": 4.373714923858643,
      "learning_rate": 2.375e-05,
      "loss": 0.9349,
      "step": 191
    },
    {
      "epoch": 6.620689655172414,
      "grad_norm": 4.665045738220215,
      "learning_rate": 2.3875e-05,
      "loss": 0.9615,
      "step": 192
    },
    {
      "epoch": 6.655172413793103,
      "grad_norm": 7.3372907638549805,
      "learning_rate": 2.4e-05,
      "loss": 0.9949,
      "step": 193
    },
    {
      "epoch": 6.689655172413794,
      "grad_norm": 4.059077739715576,
      "learning_rate": 2.4125e-05,
      "loss": 0.9306,
      "step": 194
    },
    {
      "epoch": 6.724137931034483,
      "grad_norm": 5.908609390258789,
      "learning_rate": 2.425e-05,
      "loss": 0.7885,
      "step": 195
    },
    {
      "epoch": 6.758620689655173,
      "grad_norm": 6.472893238067627,
      "learning_rate": 2.4375e-05,
      "loss": 1.1013,
      "step": 196
    },
    {
      "epoch": 6.793103448275862,
      "grad_norm": 4.9746575355529785,
      "learning_rate": 2.45e-05,
      "loss": 1.0222,
      "step": 197
    },
    {
      "epoch": 6.827586206896552,
      "grad_norm": 5.179764270782471,
      "learning_rate": 2.4625000000000002e-05,
      "loss": 0.9958,
      "step": 198
    },
    {
      "epoch": 6.862068965517241,
      "grad_norm": 7.71660041809082,
      "learning_rate": 2.4750000000000002e-05,
      "loss": 1.05,
      "step": 199
    },
    {
      "epoch": 6.896551724137931,
      "grad_norm": 6.319756507873535,
      "learning_rate": 2.4875e-05,
      "loss": 1.0519,
      "step": 200
    },
    {
      "epoch": 6.931034482758621,
      "grad_norm": 4.988058567047119,
      "learning_rate": 2.5e-05,
      "loss": 0.998,
      "step": 201
    },
    {
      "epoch": 6.9655172413793105,
      "grad_norm": 5.239701747894287,
      "learning_rate": 2.4999980961416097e-05,
      "loss": 1.0276,
      "step": 202
    },
    {
      "epoch": 7.0,
      "grad_norm": 9.979659080505371,
      "learning_rate": 2.499992384572238e-05,
      "loss": 1.005,
      "step": 203
    },
    {
      "epoch": 7.0344827586206895,
      "grad_norm": 5.444819927215576,
      "learning_rate": 2.4999828653092835e-05,
      "loss": 1.0034,
      "step": 204
    },
    {
      "epoch": 7.068965517241379,
      "grad_norm": 5.544732570648193,
      "learning_rate": 2.4999695383817435e-05,
      "loss": 1.0074,
      "step": 205
    },
    {
      "epoch": 7.103448275862069,
      "grad_norm": 7.6183762550354,
      "learning_rate": 2.499952403830214e-05,
      "loss": 0.9798,
      "step": 206
    },
    {
      "epoch": 7.137931034482759,
      "grad_norm": 4.80704927444458,
      "learning_rate": 2.4999314617068904e-05,
      "loss": 0.9429,
      "step": 207
    },
    {
      "epoch": 7.172413793103448,
      "grad_norm": 5.107629776000977,
      "learning_rate": 2.4999067120755652e-05,
      "loss": 1.0054,
      "step": 208
    },
    {
      "epoch": 7.206896551724138,
      "grad_norm": 5.443073272705078,
      "learning_rate": 2.4998781550116305e-05,
      "loss": 1.0603,
      "step": 209
    },
    {
      "epoch": 7.241379310344827,
      "grad_norm": 2.814347267150879,
      "learning_rate": 2.499845790602076e-05,
      "loss": 1.0818,
      "step": 210
    },
    {
      "epoch": 7.275862068965517,
      "grad_norm": 4.260725021362305,
      "learning_rate": 2.4998096189454893e-05,
      "loss": 0.9751,
      "step": 211
    },
    {
      "epoch": 7.310344827586207,
      "grad_norm": 3.0248708724975586,
      "learning_rate": 2.4997696401520555e-05,
      "loss": 0.9156,
      "step": 212
    },
    {
      "epoch": 7.344827586206897,
      "grad_norm": 6.022355079650879,
      "learning_rate": 2.499725854343557e-05,
      "loss": 0.9621,
      "step": 213
    },
    {
      "epoch": 7.379310344827586,
      "grad_norm": 2.480070114135742,
      "learning_rate": 2.4996782616533732e-05,
      "loss": 0.9835,
      "step": 214
    },
    {
      "epoch": 7.413793103448276,
      "grad_norm": 2.478194236755371,
      "learning_rate": 2.499626862226479e-05,
      "loss": 0.8196,
      "step": 215
    },
    {
      "epoch": 7.448275862068965,
      "grad_norm": 2.9012348651885986,
      "learning_rate": 2.4995716562194465e-05,
      "loss": 0.9136,
      "step": 216
    },
    {
      "epoch": 7.482758620689655,
      "grad_norm": 2.776332378387451,
      "learning_rate": 2.499512643800443e-05,
      "loss": 0.874,
      "step": 217
    },
    {
      "epoch": 7.517241379310345,
      "grad_norm": 3.0665347576141357,
      "learning_rate": 2.4994498251492302e-05,
      "loss": 0.8381,
      "step": 218
    },
    {
      "epoch": 7.551724137931035,
      "grad_norm": 3.6337053775787354,
      "learning_rate": 2.4993832004571646e-05,
      "loss": 0.9145,
      "step": 219
    },
    {
      "epoch": 7.586206896551724,
      "grad_norm": 3.960299015045166,
      "learning_rate": 2.4993127699271966e-05,
      "loss": 0.8541,
      "step": 220
    },
    {
      "epoch": 7.620689655172414,
      "grad_norm": 4.1685967445373535,
      "learning_rate": 2.49923853377387e-05,
      "loss": 0.9458,
      "step": 221
    },
    {
      "epoch": 7.655172413793103,
      "grad_norm": 3.217474937438965,
      "learning_rate": 2.4991604922233204e-05,
      "loss": 0.842,
      "step": 222
    },
    {
      "epoch": 7.689655172413794,
      "grad_norm": 3.9884676933288574,
      "learning_rate": 2.4990786455132764e-05,
      "loss": 0.8746,
      "step": 223
    },
    {
      "epoch": 7.724137931034483,
      "grad_norm": 6.753876209259033,
      "learning_rate": 2.4989929938930576e-05,
      "loss": 0.8353,
      "step": 224
    },
    {
      "epoch": 7.758620689655173,
      "grad_norm": 4.734915256500244,
      "learning_rate": 2.498903537623573e-05,
      "loss": 0.8488,
      "step": 225
    },
    {
      "epoch": 7.793103448275862,
      "grad_norm": 2.8307461738586426,
      "learning_rate": 2.4988102769773227e-05,
      "loss": 0.8886,
      "step": 226
    },
    {
      "epoch": 7.827586206896552,
      "grad_norm": 4.811133861541748,
      "learning_rate": 2.4987132122383936e-05,
      "loss": 0.8869,
      "step": 227
    },
    {
      "epoch": 7.862068965517241,
      "grad_norm": 6.6021728515625,
      "learning_rate": 2.4986123437024627e-05,
      "loss": 1.0391,
      "step": 228
    },
    {
      "epoch": 7.896551724137931,
      "grad_norm": 8.46661376953125,
      "learning_rate": 2.4985076716767927e-05,
      "loss": 0.9857,
      "step": 229
    },
    {
      "epoch": 7.931034482758621,
      "grad_norm": 7.210445880889893,
      "learning_rate": 2.4983991964802327e-05,
      "loss": 0.8793,
      "step": 230
    },
    {
      "epoch": 7.9655172413793105,
      "grad_norm": 4.752058506011963,
      "learning_rate": 2.4982869184432174e-05,
      "loss": 0.8634,
      "step": 231
    },
    {
      "epoch": 8.0,
      "grad_norm": 8.460209846496582,
      "learning_rate": 2.498170837907765e-05,
      "loss": 0.9404,
      "step": 232
    },
    {
      "epoch": 8.03448275862069,
      "grad_norm": 4.385580539703369,
      "learning_rate": 2.4980509552274765e-05,
      "loss": 0.8939,
      "step": 233
    },
    {
      "epoch": 8.068965517241379,
      "grad_norm": 4.554064750671387,
      "learning_rate": 2.4979272707675356e-05,
      "loss": 0.8854,
      "step": 234
    },
    {
      "epoch": 8.10344827586207,
      "grad_norm": 5.013895034790039,
      "learning_rate": 2.497799784904707e-05,
      "loss": 0.8524,
      "step": 235
    },
    {
      "epoch": 8.137931034482758,
      "grad_norm": 2.559938669204712,
      "learning_rate": 2.4976684980273338e-05,
      "loss": 0.8763,
      "step": 236
    },
    {
      "epoch": 8.172413793103448,
      "grad_norm": 2.896611452102661,
      "learning_rate": 2.4975334105353396e-05,
      "loss": 0.8625,
      "step": 237
    },
    {
      "epoch": 8.206896551724139,
      "grad_norm": 3.1764471530914307,
      "learning_rate": 2.497394522840224e-05,
      "loss": 0.8599,
      "step": 238
    },
    {
      "epoch": 8.241379310344827,
      "grad_norm": 2.926391363143921,
      "learning_rate": 2.4972518353650626e-05,
      "loss": 0.8583,
      "step": 239
    },
    {
      "epoch": 8.275862068965518,
      "grad_norm": 3.4869906902313232,
      "learning_rate": 2.497105348544507e-05,
      "loss": 0.8417,
      "step": 240
    },
    {
      "epoch": 8.310344827586206,
      "grad_norm": 5.803765296936035,
      "learning_rate": 2.4969550628247805e-05,
      "loss": 0.8914,
      "step": 241
    },
    {
      "epoch": 8.344827586206897,
      "grad_norm": 3.910322904586792,
      "learning_rate": 2.49680097866368e-05,
      "loss": 0.8737,
      "step": 242
    },
    {
      "epoch": 8.379310344827585,
      "grad_norm": 3.2522225379943848,
      "learning_rate": 2.4966430965305727e-05,
      "loss": 0.8915,
      "step": 243
    },
    {
      "epoch": 8.413793103448276,
      "grad_norm": 3.7491557598114014,
      "learning_rate": 2.4964814169063948e-05,
      "loss": 0.9399,
      "step": 244
    },
    {
      "epoch": 8.448275862068966,
      "grad_norm": 2.59580397605896,
      "learning_rate": 2.4963159402836506e-05,
      "loss": 0.9132,
      "step": 245
    },
    {
      "epoch": 8.482758620689655,
      "grad_norm": 2.840510368347168,
      "learning_rate": 2.49614666716641e-05,
      "loss": 0.8135,
      "step": 246
    },
    {
      "epoch": 8.517241379310345,
      "grad_norm": 6.813579082489014,
      "learning_rate": 2.495973598070309e-05,
      "loss": 0.8306,
      "step": 247
    },
    {
      "epoch": 8.551724137931034,
      "grad_norm": 5.595628261566162,
      "learning_rate": 2.4957967335225456e-05,
      "loss": 0.9031,
      "step": 248
    },
    {
      "epoch": 8.586206896551724,
      "grad_norm": 3.344548463821411,
      "learning_rate": 2.4956160740618806e-05,
      "loss": 0.8205,
      "step": 249
    },
    {
      "epoch": 8.620689655172415,
      "grad_norm": 2.66579270362854,
      "learning_rate": 2.495431620238633e-05,
      "loss": 0.8087,
      "step": 250
    },
    {
      "epoch": 8.655172413793103,
      "grad_norm": 3.5464556217193604,
      "learning_rate": 2.495243372614682e-05,
      "loss": 0.7636,
      "step": 251
    },
    {
      "epoch": 8.689655172413794,
      "grad_norm": 5.313910484313965,
      "learning_rate": 2.495051331763462e-05,
      "loss": 0.8267,
      "step": 252
    },
    {
      "epoch": 8.724137931034482,
      "grad_norm": 5.469503879547119,
      "learning_rate": 2.494855498269963e-05,
      "loss": 0.7766,
      "step": 253
    },
    {
      "epoch": 8.758620689655173,
      "grad_norm": 4.186700820922852,
      "learning_rate": 2.4946558727307277e-05,
      "loss": 0.8521,
      "step": 254
    },
    {
      "epoch": 8.793103448275861,
      "grad_norm": 3.4536843299865723,
      "learning_rate": 2.4944524557538503e-05,
      "loss": 0.7532,
      "step": 255
    },
    {
      "epoch": 8.827586206896552,
      "grad_norm": 3.1947126388549805,
      "learning_rate": 2.4942452479589735e-05,
      "loss": 0.716,
      "step": 256
    },
    {
      "epoch": 8.862068965517242,
      "grad_norm": 4.5636138916015625,
      "learning_rate": 2.494034249977289e-05,
      "loss": 0.9678,
      "step": 257
    },
    {
      "epoch": 8.89655172413793,
      "grad_norm": 4.235171794891357,
      "learning_rate": 2.4938194624515333e-05,
      "loss": 0.9419,
      "step": 258
    },
    {
      "epoch": 8.931034482758621,
      "grad_norm": 4.488134860992432,
      "learning_rate": 2.4936008860359854e-05,
      "loss": 0.9512,
      "step": 259
    },
    {
      "epoch": 8.96551724137931,
      "grad_norm": 5.502348899841309,
      "learning_rate": 2.4933785213964677e-05,
      "loss": 0.8107,
      "step": 260
    },
    {
      "epoch": 9.0,
      "grad_norm": 3.3806426525115967,
      "learning_rate": 2.4931523692103418e-05,
      "loss": 0.8593,
      "step": 261
    },
    {
      "epoch": 9.03448275862069,
      "grad_norm": 5.362583160400391,
      "learning_rate": 2.492922430166506e-05,
      "loss": 0.9019,
      "step": 262
    },
    {
      "epoch": 9.068965517241379,
      "grad_norm": 6.098759651184082,
      "learning_rate": 2.4926887049653943e-05,
      "loss": 0.9845,
      "step": 263
    },
    {
      "epoch": 9.10344827586207,
      "grad_norm": 3.2926342487335205,
      "learning_rate": 2.492451194318975e-05,
      "loss": 0.8941,
      "step": 264
    },
    {
      "epoch": 9.137931034482758,
      "grad_norm": 3.6239383220672607,
      "learning_rate": 2.4922098989507454e-05,
      "loss": 0.9016,
      "step": 265
    },
    {
      "epoch": 9.172413793103448,
      "grad_norm": 2.945098400115967,
      "learning_rate": 2.4919648195957344e-05,
      "loss": 0.8249,
      "step": 266
    },
    {
      "epoch": 9.206896551724139,
      "grad_norm": 5.63911247253418,
      "learning_rate": 2.4917159570004954e-05,
      "loss": 0.8786,
      "step": 267
    },
    {
      "epoch": 9.241379310344827,
      "grad_norm": 4.707492351531982,
      "learning_rate": 2.491463311923108e-05,
      "loss": 0.9624,
      "step": 268
    },
    {
      "epoch": 9.275862068965518,
      "grad_norm": 3.5989744663238525,
      "learning_rate": 2.491206885133171e-05,
      "loss": 0.7872,
      "step": 269
    },
    {
      "epoch": 9.310344827586206,
      "grad_norm": 3.1965887546539307,
      "learning_rate": 2.490946677411807e-05,
      "loss": 0.8428,
      "step": 270
    },
    {
      "epoch": 9.344827586206897,
      "grad_norm": 3.745168924331665,
      "learning_rate": 2.4906826895516528e-05,
      "loss": 0.8836,
      "step": 271
    },
    {
      "epoch": 9.379310344827585,
      "grad_norm": 4.135939598083496,
      "learning_rate": 2.490414922356861e-05,
      "loss": 0.8569,
      "step": 272
    },
    {
      "epoch": 9.413793103448276,
      "grad_norm": 3.0350985527038574,
      "learning_rate": 2.4901433766430975e-05,
      "loss": 0.8092,
      "step": 273
    },
    {
      "epoch": 9.448275862068966,
      "grad_norm": 2.5709338188171387,
      "learning_rate": 2.4898680532375374e-05,
      "loss": 0.8073,
      "step": 274
    },
    {
      "epoch": 9.482758620689655,
      "grad_norm": 3.360161542892456,
      "learning_rate": 2.489588952978863e-05,
      "loss": 0.7794,
      "step": 275
    },
    {
      "epoch": 9.517241379310345,
      "grad_norm": 3.463740587234497,
      "learning_rate": 2.4893060767172632e-05,
      "loss": 0.8216,
      "step": 276
    },
    {
      "epoch": 9.551724137931034,
      "grad_norm": 4.866174221038818,
      "learning_rate": 2.489019425314427e-05,
      "loss": 0.821,
      "step": 277
    },
    {
      "epoch": 9.586206896551724,
      "grad_norm": 4.300073146820068,
      "learning_rate": 2.4887289996435452e-05,
      "loss": 0.8998,
      "step": 278
    },
    {
      "epoch": 9.620689655172415,
      "grad_norm": 7.732598304748535,
      "learning_rate": 2.4884348005893045e-05,
      "loss": 0.8335,
      "step": 279
    },
    {
      "epoch": 9.655172413793103,
      "grad_norm": 6.1583027839660645,
      "learning_rate": 2.488136829047886e-05,
      "loss": 0.8181,
      "step": 280
    },
    {
      "epoch": 9.689655172413794,
      "grad_norm": 3.6511595249176025,
      "learning_rate": 2.487835085926963e-05,
      "loss": 0.8059,
      "step": 281
    },
    {
      "epoch": 9.724137931034482,
      "grad_norm": 3.68314790725708,
      "learning_rate": 2.487529572145697e-05,
      "loss": 0.8394,
      "step": 282
    },
    {
      "epoch": 9.758620689655173,
      "grad_norm": 3.6726181507110596,
      "learning_rate": 2.4872202886347362e-05,
      "loss": 0.7453,
      "step": 283
    },
    {
      "epoch": 9.793103448275861,
      "grad_norm": 2.719984531402588,
      "learning_rate": 2.486907236336212e-05,
      "loss": 0.8272,
      "step": 284
    },
    {
      "epoch": 9.827586206896552,
      "grad_norm": 3.7963995933532715,
      "learning_rate": 2.4865904162037358e-05,
      "loss": 0.8991,
      "step": 285
    },
    {
      "epoch": 9.862068965517242,
      "grad_norm": 2.754535436630249,
      "learning_rate": 2.4862698292023963e-05,
      "loss": 0.7064,
      "step": 286
    },
    {
      "epoch": 9.89655172413793,
      "grad_norm": 3.0903878211975098,
      "learning_rate": 2.4859454763087577e-05,
      "loss": 0.735,
      "step": 287
    },
    {
      "epoch": 9.931034482758621,
      "grad_norm": 4.279144287109375,
      "learning_rate": 2.4856173585108544e-05,
      "loss": 0.8074,
      "step": 288
    },
    {
      "epoch": 9.96551724137931,
      "grad_norm": 4.23798942565918,
      "learning_rate": 2.4852854768081912e-05,
      "loss": 0.7802,
      "step": 289
    },
    {
      "epoch": 10.0,
      "grad_norm": 3.5624747276306152,
      "learning_rate": 2.4849498322117364e-05,
      "loss": 0.8029,
      "step": 290
    },
    {
      "epoch": 10.03448275862069,
      "grad_norm": 2.764233350753784,
      "learning_rate": 2.4846104257439222e-05,
      "loss": 0.7497,
      "step": 291
    },
    {
      "epoch": 10.068965517241379,
      "grad_norm": 4.359961032867432,
      "learning_rate": 2.4842672584386396e-05,
      "loss": 0.7794,
      "step": 292
    },
    {
      "epoch": 10.10344827586207,
      "grad_norm": 5.876353740692139,
      "learning_rate": 2.483920331341235e-05,
      "loss": 0.8483,
      "step": 293
    },
    {
      "epoch": 10.137931034482758,
      "grad_norm": 4.351016998291016,
      "learning_rate": 2.4835696455085093e-05,
      "loss": 0.7924,
      "step": 294
    },
    {
      "epoch": 10.172413793103448,
      "grad_norm": 3.5949246883392334,
      "learning_rate": 2.483215202008712e-05,
      "loss": 0.8293,
      "step": 295
    },
    {
      "epoch": 10.206896551724139,
      "grad_norm": 3.226196765899658,
      "learning_rate": 2.4828570019215396e-05,
      "loss": 0.7934,
      "step": 296
    },
    {
      "epoch": 10.241379310344827,
      "grad_norm": 5.124762535095215,
      "learning_rate": 2.4824950463381314e-05,
      "loss": 0.8055,
      "step": 297
    },
    {
      "epoch": 10.275862068965518,
      "grad_norm": 3.658576011657715,
      "learning_rate": 2.482129336361067e-05,
      "loss": 0.6952,
      "step": 298
    },
    {
      "epoch": 10.310344827586206,
      "grad_norm": 3.146904945373535,
      "learning_rate": 2.481759873104363e-05,
      "loss": 0.714,
      "step": 299
    },
    {
      "epoch": 10.344827586206897,
      "grad_norm": 2.9892313480377197,
      "learning_rate": 2.4813866576934676e-05,
      "loss": 0.7746,
      "step": 300
    },
    {
      "epoch": 10.379310344827585,
      "grad_norm": 2.505178928375244,
      "learning_rate": 2.4810096912652604e-05,
      "loss": 0.6921,
      "step": 301
    },
    {
      "epoch": 10.413793103448276,
      "grad_norm": 2.44994854927063,
      "learning_rate": 2.480628974968046e-05,
      "loss": 0.7294,
      "step": 302
    },
    {
      "epoch": 10.448275862068966,
      "grad_norm": 3.606544256210327,
      "learning_rate": 2.4802445099615525e-05,
      "loss": 0.8161,
      "step": 303
    },
    {
      "epoch": 10.482758620689655,
      "grad_norm": 3.289849042892456,
      "learning_rate": 2.479856297416927e-05,
      "loss": 0.7754,
      "step": 304
    },
    {
      "epoch": 10.517241379310345,
      "grad_norm": 3.646949529647827,
      "learning_rate": 2.4794643385167327e-05,
      "loss": 0.8014,
      "step": 305
    },
    {
      "epoch": 10.551724137931034,
      "grad_norm": 6.4262847900390625,
      "learning_rate": 2.4790686344549436e-05,
      "loss": 0.8158,
      "step": 306
    },
    {
      "epoch": 10.586206896551724,
      "grad_norm": 3.1850662231445312,
      "learning_rate": 2.478669186436943e-05,
      "loss": 0.7479,
      "step": 307
    },
    {
      "epoch": 10.620689655172415,
      "grad_norm": 3.0070602893829346,
      "learning_rate": 2.478265995679519e-05,
      "loss": 0.769,
      "step": 308
    },
    {
      "epoch": 10.655172413793103,
      "grad_norm": 6.722799777984619,
      "learning_rate": 2.4778590634108613e-05,
      "loss": 0.7574,
      "step": 309
    },
    {
      "epoch": 10.689655172413794,
      "grad_norm": 3.4408349990844727,
      "learning_rate": 2.4774483908705546e-05,
      "loss": 0.6743,
      "step": 310
    },
    {
      "epoch": 10.724137931034482,
      "grad_norm": 2.932551383972168,
      "learning_rate": 2.4770339793095802e-05,
      "loss": 0.7759,
      "step": 311
    },
    {
      "epoch": 10.758620689655173,
      "grad_norm": 3.669299602508545,
      "learning_rate": 2.4766158299903062e-05,
      "loss": 0.8693,
      "step": 312
    },
    {
      "epoch": 10.793103448275861,
      "grad_norm": 3.167391061782837,
      "learning_rate": 2.4761939441864895e-05,
      "loss": 0.7359,
      "step": 313
    },
    {
      "epoch": 10.827586206896552,
      "grad_norm": 2.894587993621826,
      "learning_rate": 2.4757683231832662e-05,
      "loss": 0.6821,
      "step": 314
    },
    {
      "epoch": 10.862068965517242,
      "grad_norm": 3.0310933589935303,
      "learning_rate": 2.4753389682771523e-05,
      "loss": 0.7143,
      "step": 315
    },
    {
      "epoch": 10.89655172413793,
      "grad_norm": 3.1652767658233643,
      "learning_rate": 2.474905880776037e-05,
      "loss": 0.7949,
      "step": 316
    },
    {
      "epoch": 10.931034482758621,
      "grad_norm": 5.710330009460449,
      "learning_rate": 2.47446906199918e-05,
      "loss": 0.7518,
      "step": 317
    },
    {
      "epoch": 10.96551724137931,
      "grad_norm": 5.877900123596191,
      "learning_rate": 2.4740285132772072e-05,
      "loss": 0.8513,
      "step": 318
    },
    {
      "epoch": 11.0,
      "grad_norm": 3.645622730255127,
      "learning_rate": 2.4735842359521064e-05,
      "loss": 0.8153,
      "step": 319
    },
    {
      "epoch": 11.03448275862069,
      "grad_norm": 3.9911906719207764,
      "learning_rate": 2.4731362313772233e-05,
      "loss": 0.7651,
      "step": 320
    },
    {
      "epoch": 11.068965517241379,
      "grad_norm": 4.0215935707092285,
      "learning_rate": 2.4726845009172572e-05,
      "loss": 0.7985,
      "step": 321
    },
    {
      "epoch": 11.10344827586207,
      "grad_norm": 2.402211904525757,
      "learning_rate": 2.4722290459482578e-05,
      "loss": 0.7054,
      "step": 322
    },
    {
      "epoch": 11.137931034482758,
      "grad_norm": 2.782041549682617,
      "learning_rate": 2.47176986785762e-05,
      "loss": 0.7909,
      "step": 323
    },
    {
      "epoch": 11.172413793103448,
      "grad_norm": 4.322781085968018,
      "learning_rate": 2.47130696804408e-05,
      "loss": 0.747,
      "step": 324
    },
    {
      "epoch": 11.206896551724139,
      "grad_norm": 5.149921417236328,
      "learning_rate": 2.47084034791771e-05,
      "loss": 0.8314,
      "step": 325
    },
    {
      "epoch": 11.241379310344827,
      "grad_norm": 3.920372486114502,
      "learning_rate": 2.4703700088999167e-05,
      "loss": 0.8623,
      "step": 326
    },
    {
      "epoch": 11.275862068965518,
      "grad_norm": 2.6054768562316895,
      "learning_rate": 2.4698959524234346e-05,
      "loss": 0.7899,
      "step": 327
    },
    {
      "epoch": 11.310344827586206,
      "grad_norm": 4.032412528991699,
      "learning_rate": 2.4694181799323206e-05,
      "loss": 0.8142,
      "step": 328
    },
    {
      "epoch": 11.344827586206897,
      "grad_norm": 3.573333978652954,
      "learning_rate": 2.468936692881954e-05,
      "loss": 0.7309,
      "step": 329
    },
    {
      "epoch": 11.379310344827585,
      "grad_norm": 7.383831977844238,
      "learning_rate": 2.4684514927390274e-05,
      "loss": 0.8518,
      "step": 330
    },
    {
      "epoch": 11.413793103448276,
      "grad_norm": 3.8350508213043213,
      "learning_rate": 2.4679625809815443e-05,
      "loss": 0.7778,
      "step": 331
    },
    {
      "epoch": 11.448275862068966,
      "grad_norm": 5.659679889678955,
      "learning_rate": 2.467469959098815e-05,
      "loss": 0.7818,
      "step": 332
    },
    {
      "epoch": 11.482758620689655,
      "grad_norm": 6.893414497375488,
      "learning_rate": 2.4669736285914505e-05,
      "loss": 0.7871,
      "step": 333
    },
    {
      "epoch": 11.517241379310345,
      "grad_norm": 3.4787838459014893,
      "learning_rate": 2.4664735909713606e-05,
      "loss": 0.6967,
      "step": 334
    },
    {
      "epoch": 11.551724137931034,
      "grad_norm": 3.2118844985961914,
      "learning_rate": 2.465969847761746e-05,
      "loss": 0.6598,
      "step": 335
    },
    {
      "epoch": 11.586206896551724,
      "grad_norm": 3.6841318607330322,
      "learning_rate": 2.4654624004970957e-05,
      "loss": 0.7323,
      "step": 336
    },
    {
      "epoch": 11.620689655172415,
      "grad_norm": 4.46773624420166,
      "learning_rate": 2.464951250723183e-05,
      "loss": 0.6713,
      "step": 337
    },
    {
      "epoch": 11.655172413793103,
      "grad_norm": 2.266669988632202,
      "learning_rate": 2.4644363999970576e-05,
      "loss": 0.6425,
      "step": 338
    },
    {
      "epoch": 11.689655172413794,
      "grad_norm": 5.7526140213012695,
      "learning_rate": 2.4639178498870452e-05,
      "loss": 0.6836,
      "step": 339
    },
    {
      "epoch": 11.724137931034482,
      "grad_norm": 5.038420677185059,
      "learning_rate": 2.4633956019727385e-05,
      "loss": 0.7308,
      "step": 340
    },
    {
      "epoch": 11.758620689655173,
      "grad_norm": 5.116992473602295,
      "learning_rate": 2.4628696578449956e-05,
      "loss": 0.7882,
      "step": 341
    },
    {
      "epoch": 11.793103448275861,
      "grad_norm": 4.496801376342773,
      "learning_rate": 2.4623400191059335e-05,
      "loss": 0.7327,
      "step": 342
    },
    {
      "epoch": 11.827586206896552,
      "grad_norm": 5.261680603027344,
      "learning_rate": 2.4618066873689238e-05,
      "loss": 0.8085,
      "step": 343
    },
    {
      "epoch": 11.862068965517242,
      "grad_norm": 4.4358367919921875,
      "learning_rate": 2.461269664258587e-05,
      "loss": 0.6914,
      "step": 344
    },
    {
      "epoch": 11.89655172413793,
      "grad_norm": 3.8224196434020996,
      "learning_rate": 2.4607289514107888e-05,
      "loss": 0.6859,
      "step": 345
    },
    {
      "epoch": 11.931034482758621,
      "grad_norm": 3.285635471343994,
      "learning_rate": 2.460184550472635e-05,
      "loss": 0.8106,
      "step": 346
    },
    {
      "epoch": 11.96551724137931,
      "grad_norm": 4.5762939453125,
      "learning_rate": 2.4596364631024643e-05,
      "loss": 0.7377,
      "step": 347
    },
    {
      "epoch": 12.0,
      "grad_norm": 3.780836820602417,
      "learning_rate": 2.459084690969846e-05,
      "loss": 0.707,
      "step": 348
    },
    {
      "epoch": 12.03448275862069,
      "grad_norm": 4.315356731414795,
      "learning_rate": 2.4585292357555746e-05,
      "loss": 0.7106,
      "step": 349
    },
    {
      "epoch": 12.068965517241379,
      "grad_norm": 3.3244175910949707,
      "learning_rate": 2.457970099151662e-05,
      "loss": 0.699,
      "step": 350
    },
    {
      "epoch": 12.10344827586207,
      "grad_norm": 2.7675392627716064,
      "learning_rate": 2.4574072828613354e-05,
      "loss": 0.6546,
      "step": 351
    },
    {
      "epoch": 12.137931034482758,
      "grad_norm": 3.6822872161865234,
      "learning_rate": 2.4568407885990313e-05,
      "loss": 0.6953,
      "step": 352
    },
    {
      "epoch": 12.172413793103448,
      "grad_norm": 4.7930731773376465,
      "learning_rate": 2.4562706180903894e-05,
      "loss": 0.7863,
      "step": 353
    },
    {
      "epoch": 12.206896551724139,
      "grad_norm": 3.7039031982421875,
      "learning_rate": 2.4556967730722478e-05,
      "loss": 0.7581,
      "step": 354
    },
    {
      "epoch": 12.241379310344827,
      "grad_norm": 5.0022053718566895,
      "learning_rate": 2.455119255292638e-05,
      "loss": 0.8091,
      "step": 355
    },
    {
      "epoch": 12.275862068965518,
      "grad_norm": 3.0869357585906982,
      "learning_rate": 2.4545380665107786e-05,
      "loss": 0.7677,
      "step": 356
    },
    {
      "epoch": 12.310344827586206,
      "grad_norm": 3.7048325538635254,
      "learning_rate": 2.453953208497073e-05,
      "loss": 0.7886,
      "step": 357
    },
    {
      "epoch": 12.344827586206897,
      "grad_norm": 4.7608561515808105,
      "learning_rate": 2.4533646830330986e-05,
      "loss": 0.9105,
      "step": 358
    },
    {
      "epoch": 12.379310344827585,
      "grad_norm": 5.352680683135986,
      "learning_rate": 2.452772491911607e-05,
      "loss": 0.8114,
      "step": 359
    },
    {
      "epoch": 12.413793103448276,
      "grad_norm": 8.703157424926758,
      "learning_rate": 2.4521766369365142e-05,
      "loss": 0.8285,
      "step": 360
    },
    {
      "epoch": 12.448275862068966,
      "grad_norm": 3.2992141246795654,
      "learning_rate": 2.4515771199228987e-05,
      "loss": 0.7644,
      "step": 361
    },
    {
      "epoch": 12.482758620689655,
      "grad_norm": 6.761632919311523,
      "learning_rate": 2.450973942696993e-05,
      "loss": 0.8596,
      "step": 362
    },
    {
      "epoch": 12.517241379310345,
      "grad_norm": 4.267969608306885,
      "learning_rate": 2.450367107096179e-05,
      "loss": 0.7883,
      "step": 363
    },
    {
      "epoch": 12.551724137931034,
      "grad_norm": 3.685096025466919,
      "learning_rate": 2.449756614968984e-05,
      "loss": 0.6954,
      "step": 364
    },
    {
      "epoch": 12.586206896551724,
      "grad_norm": 3.5344960689544678,
      "learning_rate": 2.449142468175072e-05,
      "loss": 0.6647,
      "step": 365
    },
    {
      "epoch": 12.620689655172415,
      "grad_norm": 4.839688777923584,
      "learning_rate": 2.4485246685852413e-05,
      "loss": 0.7058,
      "step": 366
    },
    {
      "epoch": 12.655172413793103,
      "grad_norm": 3.455937623977661,
      "learning_rate": 2.4479032180814166e-05,
      "loss": 0.6476,
      "step": 367
    },
    {
      "epoch": 12.689655172413794,
      "grad_norm": 4.819273948669434,
      "learning_rate": 2.447278118556644e-05,
      "loss": 0.6687,
      "step": 368
    },
    {
      "epoch": 12.724137931034482,
      "grad_norm": 3.5306191444396973,
      "learning_rate": 2.446649371915084e-05,
      "loss": 0.709,
      "step": 369
    },
    {
      "epoch": 12.758620689655173,
      "grad_norm": 5.589993953704834,
      "learning_rate": 2.4460169800720095e-05,
      "loss": 0.8271,
      "step": 370
    },
    {
      "epoch": 12.793103448275861,
      "grad_norm": 2.941460132598877,
      "learning_rate": 2.4453809449537947e-05,
      "loss": 0.7747,
      "step": 371
    },
    {
      "epoch": 12.827586206896552,
      "grad_norm": 2.6910247802734375,
      "learning_rate": 2.4447412684979127e-05,
      "loss": 0.6994,
      "step": 372
    },
    {
      "epoch": 12.862068965517242,
      "grad_norm": 4.3834662437438965,
      "learning_rate": 2.4440979526529295e-05,
      "loss": 0.7782,
      "step": 373
    },
    {
      "epoch": 12.89655172413793,
      "grad_norm": 5.008590221405029,
      "learning_rate": 2.4434509993784972e-05,
      "loss": 0.8679,
      "step": 374
    },
    {
      "epoch": 12.931034482758621,
      "grad_norm": 3.8566648960113525,
      "learning_rate": 2.4428004106453462e-05,
      "loss": 0.8427,
      "step": 375
    },
    {
      "epoch": 12.96551724137931,
      "grad_norm": 4.028674125671387,
      "learning_rate": 2.4421461884352836e-05,
      "loss": 0.7371,
      "step": 376
    },
    {
      "epoch": 13.0,
      "grad_norm": 5.727315902709961,
      "learning_rate": 2.4414883347411836e-05,
      "loss": 0.7163,
      "step": 377
    },
    {
      "epoch": 13.03448275862069,
      "grad_norm": 2.517099618911743,
      "learning_rate": 2.440826851566983e-05,
      "loss": 0.6465,
      "step": 378
    },
    {
      "epoch": 13.068965517241379,
      "grad_norm": 2.401848316192627,
      "learning_rate": 2.4401617409276735e-05,
      "loss": 0.6307,
      "step": 379
    },
    {
      "epoch": 13.10344827586207,
      "grad_norm": 3.5928995609283447,
      "learning_rate": 2.439493004849298e-05,
      "loss": 0.665,
      "step": 380
    },
    {
      "epoch": 13.137931034482758,
      "grad_norm": 3.128108024597168,
      "learning_rate": 2.438820645368942e-05,
      "loss": 0.6785,
      "step": 381
    },
    {
      "epoch": 13.172413793103448,
      "grad_norm": 2.5367393493652344,
      "learning_rate": 2.4381446645347297e-05,
      "loss": 0.6353,
      "step": 382
    },
    {
      "epoch": 13.206896551724139,
      "grad_norm": 3.4342269897460938,
      "learning_rate": 2.4374650644058156e-05,
      "loss": 0.6067,
      "step": 383
    },
    {
      "epoch": 13.241379310344827,
      "grad_norm": 2.9581339359283447,
      "learning_rate": 2.43678184705238e-05,
      "loss": 0.5938,
      "step": 384
    },
    {
      "epoch": 13.275862068965518,
      "grad_norm": 4.147696018218994,
      "learning_rate": 2.4360950145556208e-05,
      "loss": 0.6053,
      "step": 385
    },
    {
      "epoch": 13.310344827586206,
      "grad_norm": 3.097829580307007,
      "learning_rate": 2.4354045690077492e-05,
      "loss": 0.7284,
      "step": 386
    },
    {
      "epoch": 13.344827586206897,
      "grad_norm": 3.5135653018951416,
      "learning_rate": 2.4347105125119824e-05,
      "loss": 0.6605,
      "step": 387
    },
    {
      "epoch": 13.379310344827585,
      "grad_norm": 3.659688949584961,
      "learning_rate": 2.4340128471825362e-05,
      "loss": 0.6748,
      "step": 388
    },
    {
      "epoch": 13.413793103448276,
      "grad_norm": 2.355313539505005,
      "learning_rate": 2.4333115751446208e-05,
      "loss": 0.6609,
      "step": 389
    },
    {
      "epoch": 13.448275862068966,
      "grad_norm": 4.517484664916992,
      "learning_rate": 2.4326066985344318e-05,
      "loss": 0.6709,
      "step": 390
    },
    {
      "epoch": 13.482758620689655,
      "grad_norm": 5.471557140350342,
      "learning_rate": 2.4318982194991463e-05,
      "loss": 0.7848,
      "step": 391
    },
    {
      "epoch": 13.517241379310345,
      "grad_norm": 3.947722911834717,
      "learning_rate": 2.4311861401969138e-05,
      "loss": 0.7373,
      "step": 392
    },
    {
      "epoch": 13.551724137931034,
      "grad_norm": 3.1544623374938965,
      "learning_rate": 2.4304704627968515e-05,
      "loss": 0.688,
      "step": 393
    },
    {
      "epoch": 13.586206896551724,
      "grad_norm": 3.4366607666015625,
      "learning_rate": 2.429751189479037e-05,
      "loss": 0.6537,
      "step": 394
    },
    {
      "epoch": 13.620689655172415,
      "grad_norm": 2.357426404953003,
      "learning_rate": 2.429028322434501e-05,
      "loss": 0.5864,
      "step": 395
    },
    {
      "epoch": 13.655172413793103,
      "grad_norm": 2.7362890243530273,
      "learning_rate": 2.4283018638652234e-05,
      "loss": 0.6294,
      "step": 396
    },
    {
      "epoch": 13.689655172413794,
      "grad_norm": 2.7302050590515137,
      "learning_rate": 2.427571815984121e-05,
      "loss": 0.6314,
      "step": 397
    },
    {
      "epoch": 13.724137931034482,
      "grad_norm": 3.3525278568267822,
      "learning_rate": 2.4268381810150474e-05,
      "loss": 0.5765,
      "step": 398
    },
    {
      "epoch": 13.758620689655173,
      "grad_norm": 5.29491662979126,
      "learning_rate": 2.426100961192782e-05,
      "loss": 0.6679,
      "step": 399
    },
    {
      "epoch": 13.793103448275861,
      "grad_norm": 3.3575775623321533,
      "learning_rate": 2.4253601587630236e-05,
      "loss": 0.6177,
      "step": 400
    },
    {
      "epoch": 13.827586206896552,
      "grad_norm": 3.106152057647705,
      "learning_rate": 2.4246157759823855e-05,
      "loss": 0.6268,
      "step": 401
    },
    {
      "epoch": 13.862068965517242,
      "grad_norm": 3.3294615745544434,
      "learning_rate": 2.4238678151183863e-05,
      "loss": 0.6302,
      "step": 402
    },
    {
      "epoch": 13.89655172413793,
      "grad_norm": 3.1822283267974854,
      "learning_rate": 2.423116278449445e-05,
      "loss": 0.626,
      "step": 403
    },
    {
      "epoch": 13.931034482758621,
      "grad_norm": 2.9840409755706787,
      "learning_rate": 2.4223611682648724e-05,
      "loss": 0.6787,
      "step": 404
    },
    {
      "epoch": 13.96551724137931,
      "grad_norm": 2.9849910736083984,
      "learning_rate": 2.4216024868648644e-05,
      "loss": 0.6664,
      "step": 405
    },
    {
      "epoch": 14.0,
      "grad_norm": 4.7554731369018555,
      "learning_rate": 2.4208402365604972e-05,
      "loss": 0.6893,
      "step": 406
    },
    {
      "epoch": 14.03448275862069,
      "grad_norm": 4.694082260131836,
      "learning_rate": 2.420074419673717e-05,
      "loss": 0.6614,
      "step": 407
    },
    {
      "epoch": 14.068965517241379,
      "grad_norm": 4.404138088226318,
      "learning_rate": 2.4193050385373344e-05,
      "loss": 0.6874,
      "step": 408
    },
    {
      "epoch": 14.10344827586207,
      "grad_norm": 3.568782091140747,
      "learning_rate": 2.418532095495018e-05,
      "loss": 0.6397,
      "step": 409
    },
    {
      "epoch": 14.137931034482758,
      "grad_norm": 3.321381092071533,
      "learning_rate": 2.417755592901287e-05,
      "loss": 0.7479,
      "step": 410
    },
    {
      "epoch": 14.172413793103448,
      "grad_norm": 2.899473190307617,
      "learning_rate": 2.4169755331215023e-05,
      "loss": 0.5792,
      "step": 411
    },
    {
      "epoch": 14.206896551724139,
      "grad_norm": 5.696684837341309,
      "learning_rate": 2.4161919185318617e-05,
      "loss": 0.6417,
      "step": 412
    },
    {
      "epoch": 14.241379310344827,
      "grad_norm": 3.3947043418884277,
      "learning_rate": 2.4154047515193904e-05,
      "loss": 0.596,
      "step": 413
    },
    {
      "epoch": 14.275862068965518,
      "grad_norm": 4.2483229637146,
      "learning_rate": 2.4146140344819363e-05,
      "loss": 0.6651,
      "step": 414
    },
    {
      "epoch": 14.310344827586206,
      "grad_norm": 4.062285423278809,
      "learning_rate": 2.4138197698281606e-05,
      "loss": 0.6853,
      "step": 415
    },
    {
      "epoch": 14.344827586206897,
      "grad_norm": 3.0232627391815186,
      "learning_rate": 2.413021959977531e-05,
      "loss": 0.6696,
      "step": 416
    },
    {
      "epoch": 14.379310344827585,
      "grad_norm": 2.5746848583221436,
      "learning_rate": 2.4122206073603142e-05,
      "loss": 0.7158,
      "step": 417
    },
    {
      "epoch": 14.413793103448276,
      "grad_norm": 5.224025249481201,
      "learning_rate": 2.4114157144175703e-05,
      "loss": 0.7015,
      "step": 418
    },
    {
      "epoch": 14.448275862068966,
      "grad_norm": 3.374368906021118,
      "learning_rate": 2.4106072836011422e-05,
      "loss": 0.6896,
      "step": 419
    },
    {
      "epoch": 14.482758620689655,
      "grad_norm": 2.7048842906951904,
      "learning_rate": 2.40979531737365e-05,
      "loss": 0.6628,
      "step": 420
    },
    {
      "epoch": 14.517241379310345,
      "grad_norm": 3.953042984008789,
      "learning_rate": 2.4089798182084845e-05,
      "loss": 0.6238,
      "step": 421
    },
    {
      "epoch": 14.551724137931034,
      "grad_norm": 2.6758952140808105,
      "learning_rate": 2.4081607885897966e-05,
      "loss": 0.6396,
      "step": 422
    },
    {
      "epoch": 14.586206896551724,
      "grad_norm": 3.827179193496704,
      "learning_rate": 2.407338231012494e-05,
      "loss": 0.6959,
      "step": 423
    },
    {
      "epoch": 14.620689655172415,
      "grad_norm": 6.4605607986450195,
      "learning_rate": 2.406512147982228e-05,
      "loss": 0.7085,
      "step": 424
    },
    {
      "epoch": 14.655172413793103,
      "grad_norm": 6.189979076385498,
      "learning_rate": 2.4056825420153917e-05,
      "loss": 0.7617,
      "step": 425
    },
    {
      "epoch": 14.689655172413794,
      "grad_norm": 5.246057987213135,
      "learning_rate": 2.4048494156391087e-05,
      "loss": 0.7481,
      "step": 426
    },
    {
      "epoch": 14.724137931034482,
      "grad_norm": 2.7838568687438965,
      "learning_rate": 2.4040127713912264e-05,
      "loss": 0.634,
      "step": 427
    },
    {
      "epoch": 14.758620689655173,
      "grad_norm": 2.8083300590515137,
      "learning_rate": 2.403172611820308e-05,
      "loss": 0.6767,
      "step": 428
    },
    {
      "epoch": 14.793103448275861,
      "grad_norm": 3.925990343093872,
      "learning_rate": 2.4023289394856257e-05,
      "loss": 0.7427,
      "step": 429
    },
    {
      "epoch": 14.827586206896552,
      "grad_norm": 2.5433623790740967,
      "learning_rate": 2.401481756957152e-05,
      "loss": 0.6591,
      "step": 430
    },
    {
      "epoch": 14.862068965517242,
      "grad_norm": 4.543763160705566,
      "learning_rate": 2.4006310668155508e-05,
      "loss": 0.6728,
      "step": 431
    },
    {
      "epoch": 14.89655172413793,
      "grad_norm": 4.649848461151123,
      "learning_rate": 2.3997768716521723e-05,
      "loss": 0.741,
      "step": 432
    },
    {
      "epoch": 14.931034482758621,
      "grad_norm": 3.5682265758514404,
      "learning_rate": 2.398919174069043e-05,
      "loss": 0.6855,
      "step": 433
    },
    {
      "epoch": 14.96551724137931,
      "grad_norm": 2.9493958950042725,
      "learning_rate": 2.398057976678859e-05,
      "loss": 0.6561,
      "step": 434
    },
    {
      "epoch": 15.0,
      "grad_norm": 2.569502115249634,
      "learning_rate": 2.3971932821049765e-05,
      "loss": 0.6723,
      "step": 435
    },
    {
      "epoch": 15.03448275862069,
      "grad_norm": 2.2379672527313232,
      "learning_rate": 2.396325092981405e-05,
      "loss": 0.6323,
      "step": 436
    },
    {
      "epoch": 15.068965517241379,
      "grad_norm": 3.066236972808838,
      "learning_rate": 2.3954534119527996e-05,
      "loss": 0.7536,
      "step": 437
    },
    {
      "epoch": 15.10344827586207,
      "grad_norm": 3.0116641521453857,
      "learning_rate": 2.3945782416744517e-05,
      "loss": 0.7323,
      "step": 438
    },
    {
      "epoch": 15.137931034482758,
      "grad_norm": 4.455348491668701,
      "learning_rate": 2.3936995848122812e-05,
      "loss": 0.6905,
      "step": 439
    },
    {
      "epoch": 15.172413793103448,
      "grad_norm": 3.218564033508301,
      "learning_rate": 2.3928174440428297e-05,
      "loss": 0.663,
      "step": 440
    },
    {
      "epoch": 15.206896551724139,
      "grad_norm": 3.6186459064483643,
      "learning_rate": 2.391931822053251e-05,
      "loss": 0.6548,
      "step": 441
    },
    {
      "epoch": 15.241379310344827,
      "grad_norm": 2.9447264671325684,
      "learning_rate": 2.3910427215413036e-05,
      "loss": 0.6127,
      "step": 442
    },
    {
      "epoch": 15.275862068965518,
      "grad_norm": 2.7375073432922363,
      "learning_rate": 2.390150145215341e-05,
      "loss": 0.6404,
      "step": 443
    },
    {
      "epoch": 15.310344827586206,
      "grad_norm": 2.91684889793396,
      "learning_rate": 2.3892540957943067e-05,
      "loss": 0.629,
      "step": 444
    },
    {
      "epoch": 15.344827586206897,
      "grad_norm": 4.3465399742126465,
      "learning_rate": 2.3883545760077215e-05,
      "loss": 0.7172,
      "step": 445
    },
    {
      "epoch": 15.379310344827585,
      "grad_norm": 6.414719104766846,
      "learning_rate": 2.3874515885956792e-05,
      "loss": 0.6642,
      "step": 446
    },
    {
      "epoch": 15.413793103448276,
      "grad_norm": 2.770498752593994,
      "learning_rate": 2.386545136308836e-05,
      "loss": 0.7017,
      "step": 447
    },
    {
      "epoch": 15.448275862068966,
      "grad_norm": 3.6498284339904785,
      "learning_rate": 2.3856352219084024e-05,
      "loss": 0.6538,
      "step": 448
    },
    {
      "epoch": 15.482758620689655,
      "grad_norm": 2.958104372024536,
      "learning_rate": 2.384721848166136e-05,
      "loss": 0.5846,
      "step": 449
    },
    {
      "epoch": 15.517241379310345,
      "grad_norm": 2.3549644947052,
      "learning_rate": 2.3838050178643312e-05,
      "loss": 0.5435,
      "step": 450
    },
    {
      "epoch": 15.551724137931034,
      "grad_norm": 3.659111261367798,
      "learning_rate": 2.3828847337958126e-05,
      "loss": 0.6394,
      "step": 451
    },
    {
      "epoch": 15.586206896551724,
      "grad_norm": 2.7539238929748535,
      "learning_rate": 2.3819609987639247e-05,
      "loss": 0.6072,
      "step": 452
    },
    {
      "epoch": 15.620689655172415,
      "grad_norm": 3.189476728439331,
      "learning_rate": 2.3810338155825245e-05,
      "loss": 0.6523,
      "step": 453
    },
    {
      "epoch": 15.655172413793103,
      "grad_norm": 2.3295481204986572,
      "learning_rate": 2.3801031870759732e-05,
      "loss": 0.6303,
      "step": 454
    },
    {
      "epoch": 15.689655172413794,
      "grad_norm": 2.7341103553771973,
      "learning_rate": 2.379169116079126e-05,
      "loss": 0.6822,
      "step": 455
    },
    {
      "epoch": 15.724137931034482,
      "grad_norm": 2.9363160133361816,
      "learning_rate": 2.378231605437326e-05,
      "loss": 0.6329,
      "step": 456
    },
    {
      "epoch": 15.758620689655173,
      "grad_norm": 5.169081211090088,
      "learning_rate": 2.3772906580063924e-05,
      "loss": 0.644,
      "step": 457
    },
    {
      "epoch": 15.793103448275861,
      "grad_norm": 4.046895503997803,
      "learning_rate": 2.3763462766526145e-05,
      "loss": 0.6442,
      "step": 458
    },
    {
      "epoch": 15.827586206896552,
      "grad_norm": 5.600619316101074,
      "learning_rate": 2.3753984642527423e-05,
      "loss": 0.6732,
      "step": 459
    },
    {
      "epoch": 15.862068965517242,
      "grad_norm": 2.420551061630249,
      "learning_rate": 2.3744472236939753e-05,
      "loss": 0.5782,
      "step": 460
    },
    {
      "epoch": 15.89655172413793,
      "grad_norm": 4.496841907501221,
      "learning_rate": 2.3734925578739588e-05,
      "loss": 0.6463,
      "step": 461
    },
    {
      "epoch": 15.931034482758621,
      "grad_norm": 5.044814586639404,
      "learning_rate": 2.3725344697007696e-05,
      "loss": 0.5673,
      "step": 462
    },
    {
      "epoch": 15.96551724137931,
      "grad_norm": 4.280508518218994,
      "learning_rate": 2.3715729620929106e-05,
      "loss": 0.5989,
      "step": 463
    },
    {
      "epoch": 16.0,
      "grad_norm": 3.7099342346191406,
      "learning_rate": 2.3706080379793007e-05,
      "loss": 0.641,
      "step": 464
    },
    {
      "epoch": 16.03448275862069,
      "grad_norm": 4.294964790344238,
      "learning_rate": 2.3696397002992663e-05,
      "loss": 0.6666,
      "step": 465
    },
    {
      "epoch": 16.06896551724138,
      "grad_norm": 3.043022632598877,
      "learning_rate": 2.3686679520025314e-05,
      "loss": 0.6193,
      "step": 466
    },
    {
      "epoch": 16.103448275862068,
      "grad_norm": 2.8048079013824463,
      "learning_rate": 2.36769279604921e-05,
      "loss": 0.6389,
      "step": 467
    },
    {
      "epoch": 16.137931034482758,
      "grad_norm": 4.923339366912842,
      "learning_rate": 2.366714235409797e-05,
      "loss": 0.6676,
      "step": 468
    },
    {
      "epoch": 16.17241379310345,
      "grad_norm": 2.6823081970214844,
      "learning_rate": 2.365732273065157e-05,
      "loss": 0.6144,
      "step": 469
    },
    {
      "epoch": 16.20689655172414,
      "grad_norm": 2.74808406829834,
      "learning_rate": 2.3647469120065177e-05,
      "loss": 0.5502,
      "step": 470
    },
    {
      "epoch": 16.24137931034483,
      "grad_norm": 4.341047286987305,
      "learning_rate": 2.36375815523546e-05,
      "loss": 0.6018,
      "step": 471
    },
    {
      "epoch": 16.275862068965516,
      "grad_norm": 3.325765371322632,
      "learning_rate": 2.3627660057639082e-05,
      "loss": 0.6211,
      "step": 472
    },
    {
      "epoch": 16.310344827586206,
      "grad_norm": 4.053094387054443,
      "learning_rate": 2.361770466614122e-05,
      "loss": 0.7121,
      "step": 473
    },
    {
      "epoch": 16.344827586206897,
      "grad_norm": 3.4117484092712402,
      "learning_rate": 2.3607715408186863e-05,
      "loss": 0.6648,
      "step": 474
    },
    {
      "epoch": 16.379310344827587,
      "grad_norm": 5.172271251678467,
      "learning_rate": 2.3597692314205016e-05,
      "loss": 0.659,
      "step": 475
    },
    {
      "epoch": 16.413793103448278,
      "grad_norm": 3.7227883338928223,
      "learning_rate": 2.358763541472777e-05,
      "loss": 0.6244,
      "step": 476
    },
    {
      "epoch": 16.448275862068964,
      "grad_norm": 2.213212013244629,
      "learning_rate": 2.3577544740390184e-05,
      "loss": 0.6075,
      "step": 477
    },
    {
      "epoch": 16.482758620689655,
      "grad_norm": 2.4992291927337646,
      "learning_rate": 2.35674203219302e-05,
      "loss": 0.6339,
      "step": 478
    },
    {
      "epoch": 16.517241379310345,
      "grad_norm": 5.619305610656738,
      "learning_rate": 2.355726219018855e-05,
      "loss": 0.6946,
      "step": 479
    },
    {
      "epoch": 16.551724137931036,
      "grad_norm": 3.0107035636901855,
      "learning_rate": 2.354707037610867e-05,
      "loss": 0.6593,
      "step": 480
    },
    {
      "epoch": 16.586206896551722,
      "grad_norm": 2.2236149311065674,
      "learning_rate": 2.353684491073659e-05,
      "loss": 0.615,
      "step": 481
    },
    {
      "epoch": 16.620689655172413,
      "grad_norm": 2.139848470687866,
      "learning_rate": 2.3526585825220848e-05,
      "loss": 0.6075,
      "step": 482
    },
    {
      "epoch": 16.655172413793103,
      "grad_norm": 2.2389163970947266,
      "learning_rate": 2.35162931508124e-05,
      "loss": 0.6533,
      "step": 483
    },
    {
      "epoch": 16.689655172413794,
      "grad_norm": 3.55932879447937,
      "learning_rate": 2.3505966918864525e-05,
      "loss": 0.7435,
      "step": 484
    },
    {
      "epoch": 16.724137931034484,
      "grad_norm": 2.86104416847229,
      "learning_rate": 2.3495607160832707e-05,
      "loss": 0.6867,
      "step": 485
    },
    {
      "epoch": 16.75862068965517,
      "grad_norm": 3.4844839572906494,
      "learning_rate": 2.3485213908274567e-05,
      "loss": 0.571,
      "step": 486
    },
    {
      "epoch": 16.79310344827586,
      "grad_norm": 2.288844108581543,
      "learning_rate": 2.3474787192849756e-05,
      "loss": 0.6348,
      "step": 487
    },
    {
      "epoch": 16.82758620689655,
      "grad_norm": 2.6627936363220215,
      "learning_rate": 2.346432704631986e-05,
      "loss": 0.5483,
      "step": 488
    },
    {
      "epoch": 16.862068965517242,
      "grad_norm": 3.744588851928711,
      "learning_rate": 2.3453833500548295e-05,
      "loss": 0.6306,
      "step": 489
    },
    {
      "epoch": 16.896551724137932,
      "grad_norm": 2.5902559757232666,
      "learning_rate": 2.3443306587500225e-05,
      "loss": 0.6857,
      "step": 490
    },
    {
      "epoch": 16.93103448275862,
      "grad_norm": 3.003737449645996,
      "learning_rate": 2.3432746339242448e-05,
      "loss": 0.6444,
      "step": 491
    },
    {
      "epoch": 16.96551724137931,
      "grad_norm": 2.010629653930664,
      "learning_rate": 2.342215278794332e-05,
      "loss": 0.5778,
      "step": 492
    },
    {
      "epoch": 17.0,
      "grad_norm": 3.528261423110962,
      "learning_rate": 2.341152596587262e-05,
      "loss": 0.6003,
      "step": 493
    },
    {
      "epoch": 17.03448275862069,
      "grad_norm": 4.4283223152160645,
      "learning_rate": 2.340086590540151e-05,
      "loss": 0.6058,
      "step": 494
    },
    {
      "epoch": 17.06896551724138,
      "grad_norm": 2.131556272506714,
      "learning_rate": 2.339017263900237e-05,
      "loss": 0.6174,
      "step": 495
    },
    {
      "epoch": 17.103448275862068,
      "grad_norm": 3.118192672729492,
      "learning_rate": 2.3379446199248747e-05,
      "loss": 0.5973,
      "step": 496
    },
    {
      "epoch": 17.137931034482758,
      "grad_norm": 4.915523052215576,
      "learning_rate": 2.3368686618815238e-05,
      "loss": 0.6529,
      "step": 497
    },
    {
      "epoch": 17.17241379310345,
      "grad_norm": 3.0185019969940186,
      "learning_rate": 2.335789393047739e-05,
      "loss": 0.5984,
      "step": 498
    },
    {
      "epoch": 17.20689655172414,
      "grad_norm": 2.78558087348938,
      "learning_rate": 2.334706816711161e-05,
      "loss": 0.6387,
      "step": 499
    },
    {
      "epoch": 17.24137931034483,
      "grad_norm": 6.381213665008545,
      "learning_rate": 2.3336209361695035e-05,
      "loss": 0.6722,
      "step": 500
    },
    {
      "epoch": 17.275862068965516,
      "grad_norm": 3.779954195022583,
      "learning_rate": 2.3325317547305485e-05,
      "loss": 0.6373,
      "step": 501
    },
    {
      "epoch": 17.310344827586206,
      "grad_norm": 4.086780548095703,
      "learning_rate": 2.3314392757121308e-05,
      "loss": 0.6152,
      "step": 502
    },
    {
      "epoch": 17.344827586206897,
      "grad_norm": 4.311843395233154,
      "learning_rate": 2.3303435024421312e-05,
      "loss": 0.6094,
      "step": 503
    },
    {
      "epoch": 17.379310344827587,
      "grad_norm": 4.192368507385254,
      "learning_rate": 2.3292444382584648e-05,
      "loss": 0.6398,
      "step": 504
    },
    {
      "epoch": 17.413793103448278,
      "grad_norm": 1.923520565032959,
      "learning_rate": 2.328142086509072e-05,
      "loss": 0.6108,
      "step": 505
    },
    {
      "epoch": 17.448275862068964,
      "grad_norm": 3.6873779296875,
      "learning_rate": 2.3270364505519073e-05,
      "loss": 0.6146,
      "step": 506
    },
    {
      "epoch": 17.482758620689655,
      "grad_norm": 2.5435547828674316,
      "learning_rate": 2.32592753375493e-05,
      "loss": 0.5429,
      "step": 507
    },
    {
      "epoch": 17.517241379310345,
      "grad_norm": 4.662963390350342,
      "learning_rate": 2.324815339496092e-05,
      "loss": 0.5818,
      "step": 508
    },
    {
      "epoch": 17.551724137931036,
      "grad_norm": 3.448092222213745,
      "learning_rate": 2.3236998711633307e-05,
      "loss": 0.5538,
      "step": 509
    },
    {
      "epoch": 17.586206896551722,
      "grad_norm": 2.9241552352905273,
      "learning_rate": 2.322581132154556e-05,
      "loss": 0.5671,
      "step": 510
    },
    {
      "epoch": 17.620689655172413,
      "grad_norm": 3.708261489868164,
      "learning_rate": 2.3214591258776404e-05,
      "loss": 0.5787,
      "step": 511
    },
    {
      "epoch": 17.655172413793103,
      "grad_norm": 3.2863380908966064,
      "learning_rate": 2.3203338557504105e-05,
      "loss": 0.5985,
      "step": 512
    },
    {
      "epoch": 17.689655172413794,
      "grad_norm": 2.276576042175293,
      "learning_rate": 2.3192053252006335e-05,
      "loss": 0.5104,
      "step": 513
    },
    {
      "epoch": 17.724137931034484,
      "grad_norm": 4.1456780433654785,
      "learning_rate": 2.3180735376660094e-05,
      "loss": 0.5687,
      "step": 514
    },
    {
      "epoch": 17.75862068965517,
      "grad_norm": 4.008676528930664,
      "learning_rate": 2.3169384965941592e-05,
      "loss": 0.6583,
      "step": 515
    },
    {
      "epoch": 17.79310344827586,
      "grad_norm": 3.410106658935547,
      "learning_rate": 2.3158002054426153e-05,
      "loss": 0.6284,
      "step": 516
    },
    {
      "epoch": 17.82758620689655,
      "grad_norm": 3.3623507022857666,
      "learning_rate": 2.3146586676788095e-05,
      "loss": 0.5826,
      "step": 517
    },
    {
      "epoch": 17.862068965517242,
      "grad_norm": 2.8540868759155273,
      "learning_rate": 2.3135138867800642e-05,
      "loss": 0.6098,
      "step": 518
    },
    {
      "epoch": 17.896551724137932,
      "grad_norm": 5.287265777587891,
      "learning_rate": 2.3123658662335802e-05,
      "loss": 0.6344,
      "step": 519
    },
    {
      "epoch": 17.93103448275862,
      "grad_norm": 2.628873586654663,
      "learning_rate": 2.311214609536427e-05,
      "loss": 0.5716,
      "step": 520
    },
    {
      "epoch": 17.96551724137931,
      "grad_norm": 3.3482089042663574,
      "learning_rate": 2.3100601201955324e-05,
      "loss": 0.6347,
      "step": 521
    },
    {
      "epoch": 18.0,
      "grad_norm": 2.661550521850586,
      "learning_rate": 2.308902401727672e-05,
      "loss": 0.6066,
      "step": 522
    },
    {
      "epoch": 18.03448275862069,
      "grad_norm": 3.0608065128326416,
      "learning_rate": 2.3077414576594553e-05,
      "loss": 0.639,
      "step": 523
    },
    {
      "epoch": 18.06896551724138,
      "grad_norm": 5.210854530334473,
      "learning_rate": 2.3065772915273203e-05,
      "loss": 0.6772,
      "step": 524
    },
    {
      "epoch": 18.103448275862068,
      "grad_norm": 3.0067954063415527,
      "learning_rate": 2.305409906877519e-05,
      "loss": 0.6215,
      "step": 525
    },
    {
      "epoch": 18.137931034482758,
      "grad_norm": 3.596611261367798,
      "learning_rate": 2.3042393072661074e-05,
      "loss": 0.6174,
      "step": 526
    },
    {
      "epoch": 18.17241379310345,
      "grad_norm": 2.643958806991577,
      "learning_rate": 2.3030654962589346e-05,
      "loss": 0.6218,
      "step": 527
    },
    {
      "epoch": 18.20689655172414,
      "grad_norm": 2.740034818649292,
      "learning_rate": 2.3018884774316328e-05,
      "loss": 0.5306,
      "step": 528
    },
    {
      "epoch": 18.24137931034483,
      "grad_norm": 2.9877166748046875,
      "learning_rate": 2.3007082543696055e-05,
      "loss": 0.573,
      "step": 529
    },
    {
      "epoch": 18.275862068965516,
      "grad_norm": 2.206543445587158,
      "learning_rate": 2.299524830668017e-05,
      "loss": 0.5803,
      "step": 530
    },
    {
      "epoch": 18.310344827586206,
      "grad_norm": 3.766697883605957,
      "learning_rate": 2.2983382099317803e-05,
      "loss": 0.7642,
      "step": 531
    },
    {
      "epoch": 18.344827586206897,
      "grad_norm": 2.516557455062866,
      "learning_rate": 2.2971483957755487e-05,
      "loss": 0.588,
      "step": 532
    },
    {
      "epoch": 18.379310344827587,
      "grad_norm": 5.181385517120361,
      "learning_rate": 2.295955391823702e-05,
      "loss": 0.7196,
      "step": 533
    },
    {
      "epoch": 18.413793103448278,
      "grad_norm": 6.308480739593506,
      "learning_rate": 2.2947592017103383e-05,
      "loss": 0.7492,
      "step": 534
    },
    {
      "epoch": 18.448275862068964,
      "grad_norm": 5.336180210113525,
      "learning_rate": 2.2935598290792583e-05,
      "loss": 0.6505,
      "step": 535
    },
    {
      "epoch": 18.482758620689655,
      "grad_norm": 3.4817123413085938,
      "learning_rate": 2.2923572775839603e-05,
      "loss": 0.624,
      "step": 536
    },
    {
      "epoch": 18.517241379310345,
      "grad_norm": 3.327723503112793,
      "learning_rate": 2.2911515508876243e-05,
      "loss": 0.5971,
      "step": 537
    },
    {
      "epoch": 18.551724137931036,
      "grad_norm": 4.9538350105285645,
      "learning_rate": 2.2899426526631033e-05,
      "loss": 0.6098,
      "step": 538
    },
    {
      "epoch": 18.586206896551722,
      "grad_norm": 4.590574264526367,
      "learning_rate": 2.2887305865929104e-05,
      "loss": 0.6799,
      "step": 539
    },
    {
      "epoch": 18.620689655172413,
      "grad_norm": 3.821331262588501,
      "learning_rate": 2.2875153563692094e-05,
      "loss": 0.6897,
      "step": 540
    },
    {
      "epoch": 18.655172413793103,
      "grad_norm": 3.8579297065734863,
      "learning_rate": 2.286296965693802e-05,
      "loss": 0.6482,
      "step": 541
    },
    {
      "epoch": 18.689655172413794,
      "grad_norm": 6.096618175506592,
      "learning_rate": 2.285075418278118e-05,
      "loss": 0.6636,
      "step": 542
    },
    {
      "epoch": 18.724137931034484,
      "grad_norm": 5.061996936798096,
      "learning_rate": 2.283850717843202e-05,
      "loss": 0.6584,
      "step": 543
    },
    {
      "epoch": 18.75862068965517,
      "grad_norm": 3.448591709136963,
      "learning_rate": 2.2826228681197047e-05,
      "loss": 0.6224,
      "step": 544
    },
    {
      "epoch": 18.79310344827586,
      "grad_norm": 5.197028160095215,
      "learning_rate": 2.281391872847869e-05,
      "loss": 0.5659,
      "step": 545
    },
    {
      "epoch": 18.82758620689655,
      "grad_norm": 4.6534318923950195,
      "learning_rate": 2.2801577357775193e-05,
      "loss": 0.6029,
      "step": 546
    },
    {
      "epoch": 18.862068965517242,
      "grad_norm": 4.554355144500732,
      "learning_rate": 2.2789204606680524e-05,
      "loss": 0.681,
      "step": 547
    },
    {
      "epoch": 18.896551724137932,
      "grad_norm": 6.15927791595459,
      "learning_rate": 2.2776800512884218e-05,
      "loss": 0.7005,
      "step": 548
    },
    {
      "epoch": 18.93103448275862,
      "grad_norm": 4.163325786590576,
      "learning_rate": 2.2764365114171303e-05,
      "loss": 0.6504,
      "step": 549
    },
    {
      "epoch": 18.96551724137931,
      "grad_norm": 5.173964500427246,
      "learning_rate": 2.2751898448422155e-05,
      "loss": 0.6507,
      "step": 550
    },
    {
      "epoch": 19.0,
      "grad_norm": 3.4791982173919678,
      "learning_rate": 2.27394005536124e-05,
      "loss": 0.6203,
      "step": 551
    },
    {
      "epoch": 19.03448275862069,
      "grad_norm": 2.631436347961426,
      "learning_rate": 2.2726871467812795e-05,
      "loss": 0.6327,
      "step": 552
    },
    {
      "epoch": 19.06896551724138,
      "grad_norm": 4.092041969299316,
      "learning_rate": 2.2714311229189105e-05,
      "loss": 0.649,
      "step": 553
    },
    {
      "epoch": 19.103448275862068,
      "grad_norm": 3.6334736347198486,
      "learning_rate": 2.2701719876002004e-05,
      "loss": 0.6024,
      "step": 554
    },
    {
      "epoch": 19.137931034482758,
      "grad_norm": 3.8364288806915283,
      "learning_rate": 2.268909744660693e-05,
      "loss": 0.5991,
      "step": 555
    },
    {
      "epoch": 19.17241379310345,
      "grad_norm": 4.198896408081055,
      "learning_rate": 2.267644397945399e-05,
      "loss": 0.6068,
      "step": 556
    },
    {
      "epoch": 19.20689655172414,
      "grad_norm": 3.1704394817352295,
      "learning_rate": 2.2663759513087846e-05,
      "loss": 0.6155,
      "step": 557
    },
    {
      "epoch": 19.24137931034483,
      "grad_norm": 5.422239303588867,
      "learning_rate": 2.2651044086147578e-05,
      "loss": 0.6836,
      "step": 558
    },
    {
      "epoch": 19.275862068965516,
      "grad_norm": 3.0339040756225586,
      "learning_rate": 2.2638297737366583e-05,
      "loss": 0.5523,
      "step": 559
    },
    {
      "epoch": 19.310344827586206,
      "grad_norm": 2.4835903644561768,
      "learning_rate": 2.2625520505572452e-05,
      "loss": 0.6802,
      "step": 560
    },
    {
      "epoch": 19.344827586206897,
      "grad_norm": 4.562428951263428,
      "learning_rate": 2.2612712429686845e-05,
      "loss": 0.7499,
      "step": 561
    },
    {
      "epoch": 19.379310344827587,
      "grad_norm": 4.1088409423828125,
      "learning_rate": 2.259987354872538e-05,
      "loss": 0.6736,
      "step": 562
    },
    {
      "epoch": 19.413793103448278,
      "grad_norm": 3.155371904373169,
      "learning_rate": 2.2587003901797528e-05,
      "loss": 0.6435,
      "step": 563
    },
    {
      "epoch": 19.448275862068964,
      "grad_norm": 4.05208158493042,
      "learning_rate": 2.2574103528106448e-05,
      "loss": 0.6947,
      "step": 564
    },
    {
      "epoch": 19.482758620689655,
      "grad_norm": 2.9079976081848145,
      "learning_rate": 2.2561172466948925e-05,
      "loss": 0.6085,
      "step": 565
    },
    {
      "epoch": 19.517241379310345,
      "grad_norm": 2.7722418308258057,
      "learning_rate": 2.2548210757715216e-05,
      "loss": 0.6012,
      "step": 566
    },
    {
      "epoch": 19.551724137931036,
      "grad_norm": 3.1863250732421875,
      "learning_rate": 2.2535218439888933e-05,
      "loss": 0.6131,
      "step": 567
    },
    {
      "epoch": 19.586206896551722,
      "grad_norm": 4.646592140197754,
      "learning_rate": 2.2522195553046926e-05,
      "loss": 0.6314,
      "step": 568
    },
    {
      "epoch": 19.620689655172413,
      "grad_norm": 5.8424577713012695,
      "learning_rate": 2.2509142136859168e-05,
      "loss": 0.6064,
      "step": 569
    },
    {
      "epoch": 19.655172413793103,
      "grad_norm": 3.509737968444824,
      "learning_rate": 2.2496058231088635e-05,
      "loss": 0.5589,
      "step": 570
    },
    {
      "epoch": 19.689655172413794,
      "grad_norm": 6.115649700164795,
      "learning_rate": 2.248294387559116e-05,
      "loss": 0.646,
      "step": 571
    },
    {
      "epoch": 19.724137931034484,
      "grad_norm": 2.8004534244537354,
      "learning_rate": 2.246979911031536e-05,
      "loss": 0.5837,
      "step": 572
    },
    {
      "epoch": 19.75862068965517,
      "grad_norm": 2.3154537677764893,
      "learning_rate": 2.2456623975302454e-05,
      "loss": 0.5972,
      "step": 573
    },
    {
      "epoch": 19.79310344827586,
      "grad_norm": 2.849769115447998,
      "learning_rate": 2.2443418510686198e-05,
      "loss": 0.6159,
      "step": 574
    },
    {
      "epoch": 19.82758620689655,
      "grad_norm": 2.4394686222076416,
      "learning_rate": 2.2430182756692728e-05,
      "loss": 0.5879,
      "step": 575
    },
    {
      "epoch": 19.862068965517242,
      "grad_norm": 1.768813133239746,
      "learning_rate": 2.2416916753640442e-05,
      "loss": 0.5802,
      "step": 576
    },
    {
      "epoch": 19.896551724137932,
      "grad_norm": 2.7482104301452637,
      "learning_rate": 2.2403620541939884e-05,
      "loss": 0.5606,
      "step": 577
    },
    {
      "epoch": 19.93103448275862,
      "grad_norm": 1.9933055639266968,
      "learning_rate": 2.2390294162093627e-05,
      "loss": 0.5557,
      "step": 578
    },
    {
      "epoch": 19.96551724137931,
      "grad_norm": 3.114577293395996,
      "learning_rate": 2.2376937654696133e-05,
      "loss": 0.5388,
      "step": 579
    },
    {
      "epoch": 20.0,
      "grad_norm": 2.786837339401245,
      "learning_rate": 2.2363551060433636e-05,
      "loss": 0.5458,
      "step": 580
    },
    {
      "epoch": 20.03448275862069,
      "grad_norm": 3.6182146072387695,
      "learning_rate": 2.2350134420084023e-05,
      "loss": 0.6075,
      "step": 581
    },
    {
      "epoch": 20.06896551724138,
      "grad_norm": 4.18286657333374,
      "learning_rate": 2.233668777451672e-05,
      "loss": 0.6064,
      "step": 582
    },
    {
      "epoch": 20.103448275862068,
      "grad_norm": 2.0372018814086914,
      "learning_rate": 2.2323211164692526e-05,
      "loss": 0.575,
      "step": 583
    },
    {
      "epoch": 20.137931034482758,
      "grad_norm": 2.5157814025878906,
      "learning_rate": 2.2309704631663536e-05,
      "loss": 0.5397,
      "step": 584
    },
    {
      "epoch": 20.17241379310345,
      "grad_norm": 2.8599259853363037,
      "learning_rate": 2.2296168216573e-05,
      "loss": 0.6183,
      "step": 585
    },
    {
      "epoch": 20.20689655172414,
      "grad_norm": 2.6183788776397705,
      "learning_rate": 2.2282601960655175e-05,
      "loss": 0.5912,
      "step": 586
    },
    {
      "epoch": 20.24137931034483,
      "grad_norm": 2.3617968559265137,
      "learning_rate": 2.2269005905235234e-05,
      "loss": 0.6095,
      "step": 587
    },
    {
      "epoch": 20.275862068965516,
      "grad_norm": 2.041109323501587,
      "learning_rate": 2.2255380091729124e-05,
      "loss": 0.5709,
      "step": 588
    },
    {
      "epoch": 20.310344827586206,
      "grad_norm": 2.473818778991699,
      "learning_rate": 2.224172456164343e-05,
      "loss": 0.622,
      "step": 589
    },
    {
      "epoch": 20.344827586206897,
      "grad_norm": 4.0362324714660645,
      "learning_rate": 2.2228039356575265e-05,
      "loss": 0.6281,
      "step": 590
    },
    {
      "epoch": 20.379310344827587,
      "grad_norm": 4.358520984649658,
      "learning_rate": 2.221432451821214e-05,
      "loss": 0.6563,
      "step": 591
    },
    {
      "epoch": 20.413793103448278,
      "grad_norm": 2.3213906288146973,
      "learning_rate": 2.2200580088331825e-05,
      "loss": 0.577,
      "step": 592
    },
    {
      "epoch": 20.448275862068964,
      "grad_norm": 4.955238342285156,
      "learning_rate": 2.2186806108802248e-05,
      "loss": 0.6418,
      "step": 593
    },
    {
      "epoch": 20.482758620689655,
      "grad_norm": 5.53913688659668,
      "learning_rate": 2.217300262158133e-05,
      "loss": 0.5817,
      "step": 594
    },
    {
      "epoch": 20.517241379310345,
      "grad_norm": 3.345353603363037,
      "learning_rate": 2.215916966871689e-05,
      "loss": 0.5873,
      "step": 595
    },
    {
      "epoch": 20.551724137931036,
      "grad_norm": 3.257923126220703,
      "learning_rate": 2.2145307292346502e-05,
      "loss": 0.5968,
      "step": 596
    },
    {
      "epoch": 20.586206896551722,
      "grad_norm": 3.00583553314209,
      "learning_rate": 2.213141553469737e-05,
      "loss": 0.5637,
      "step": 597
    },
    {
      "epoch": 20.620689655172413,
      "grad_norm": 3.975980043411255,
      "learning_rate": 2.211749443808619e-05,
      "loss": 0.6064,
      "step": 598
    },
    {
      "epoch": 20.655172413793103,
      "grad_norm": 4.438958644866943,
      "learning_rate": 2.2103544044919045e-05,
      "loss": 0.6115,
      "step": 599
    },
    {
      "epoch": 20.689655172413794,
      "grad_norm": 5.422476768493652,
      "learning_rate": 2.208956439769125e-05,
      "loss": 0.6534,
      "step": 600
    },
    {
      "epoch": 20.724137931034484,
      "grad_norm": 2.1388237476348877,
      "learning_rate": 2.2075555538987227e-05,
      "loss": 0.5648,
      "step": 601
    },
    {
      "epoch": 20.75862068965517,
      "grad_norm": 3.457695245742798,
      "learning_rate": 2.20615175114804e-05,
      "loss": 0.53,
      "step": 602
    },
    {
      "epoch": 20.79310344827586,
      "grad_norm": 4.380722999572754,
      "learning_rate": 2.2047450357933032e-05,
      "loss": 0.6238,
      "step": 603
    },
    {
      "epoch": 20.82758620689655,
      "grad_norm": 3.4675967693328857,
      "learning_rate": 2.2033354121196102e-05,
      "loss": 0.594,
      "step": 604
    },
    {
      "epoch": 20.862068965517242,
      "grad_norm": 6.335079669952393,
      "learning_rate": 2.201922884420921e-05,
      "loss": 0.5905,
      "step": 605
    },
    {
      "epoch": 20.896551724137932,
      "grad_norm": 4.242478370666504,
      "learning_rate": 2.200507457000039e-05,
      "loss": 0.5376,
      "step": 606
    },
    {
      "epoch": 20.93103448275862,
      "grad_norm": 2.4837803840637207,
      "learning_rate": 2.1990891341686008e-05,
      "loss": 0.5069,
      "step": 607
    },
    {
      "epoch": 20.96551724137931,
      "grad_norm": 2.1897082328796387,
      "learning_rate": 2.1976679202470654e-05,
      "loss": 0.5472,
      "step": 608
    },
    {
      "epoch": 21.0,
      "grad_norm": 2.4400973320007324,
      "learning_rate": 2.1962438195646958e-05,
      "loss": 0.547,
      "step": 609
    },
    {
      "epoch": 21.03448275862069,
      "grad_norm": 4.650160789489746,
      "learning_rate": 2.1948168364595497e-05,
      "loss": 0.6903,
      "step": 610
    },
    {
      "epoch": 21.06896551724138,
      "grad_norm": 3.0969042778015137,
      "learning_rate": 2.1933869752784654e-05,
      "loss": 0.5143,
      "step": 611
    },
    {
      "epoch": 21.103448275862068,
      "grad_norm": 4.73034143447876,
      "learning_rate": 2.1919542403770476e-05,
      "loss": 0.6413,
      "step": 612
    },
    {
      "epoch": 21.137931034482758,
      "grad_norm": 2.4023869037628174,
      "learning_rate": 2.1905186361196556e-05,
      "loss": 0.5607,
      "step": 613
    },
    {
      "epoch": 21.17241379310345,
      "grad_norm": 3.85475754737854,
      "learning_rate": 2.189080166879389e-05,
      "loss": 0.6169,
      "step": 614
    },
    {
      "epoch": 21.20689655172414,
      "grad_norm": 2.8691437244415283,
      "learning_rate": 2.1876388370380745e-05,
      "loss": 0.4884,
      "step": 615
    },
    {
      "epoch": 21.24137931034483,
      "grad_norm": 2.3422751426696777,
      "learning_rate": 2.186194650986253e-05,
      "loss": 0.5269,
      "step": 616
    },
    {
      "epoch": 21.275862068965516,
      "grad_norm": 3.4676578044891357,
      "learning_rate": 2.184747613123165e-05,
      "loss": 0.5517,
      "step": 617
    },
    {
      "epoch": 21.310344827586206,
      "grad_norm": 3.613835573196411,
      "learning_rate": 2.1832977278567394e-05,
      "loss": 0.5919,
      "step": 618
    },
    {
      "epoch": 21.344827586206897,
      "grad_norm": 3.2445290088653564,
      "learning_rate": 2.181844999603578e-05,
      "loss": 0.5624,
      "step": 619
    },
    {
      "epoch": 21.379310344827587,
      "grad_norm": 3.5118792057037354,
      "learning_rate": 2.1803894327889425e-05,
      "loss": 0.587,
      "step": 620
    },
    {
      "epoch": 21.413793103448278,
      "grad_norm": 5.919454574584961,
      "learning_rate": 2.178931031846743e-05,
      "loss": 0.6432,
      "step": 621
    },
    {
      "epoch": 21.448275862068964,
      "grad_norm": 3.322352409362793,
      "learning_rate": 2.1774698012195206e-05,
      "loss": 0.6003,
      "step": 622
    },
    {
      "epoch": 21.482758620689655,
      "grad_norm": 6.281351566314697,
      "learning_rate": 2.1760057453584376e-05,
      "loss": 0.5805,
      "step": 623
    },
    {
      "epoch": 21.517241379310345,
      "grad_norm": 5.132048606872559,
      "learning_rate": 2.1745388687232624e-05,
      "loss": 0.5572,
      "step": 624
    },
    {
      "epoch": 21.551724137931036,
      "grad_norm": 2.332777976989746,
      "learning_rate": 2.1730691757823553e-05,
      "loss": 0.5633,
      "step": 625
    },
    {
      "epoch": 21.586206896551722,
      "grad_norm": 3.328850030899048,
      "learning_rate": 2.171596671012655e-05,
      "loss": 0.631,
      "step": 626
    },
    {
      "epoch": 21.620689655172413,
      "grad_norm": 2.524472713470459,
      "learning_rate": 2.1701213588996683e-05,
      "loss": 0.5458,
      "step": 627
    },
    {
      "epoch": 21.655172413793103,
      "grad_norm": 2.7530570030212402,
      "learning_rate": 2.16864324393745e-05,
      "loss": 0.5397,
      "step": 628
    },
    {
      "epoch": 21.689655172413794,
      "grad_norm": 4.604249954223633,
      "learning_rate": 2.1671623306285956e-05,
      "loss": 0.5842,
      "step": 629
    },
    {
      "epoch": 21.724137931034484,
      "grad_norm": 2.306530714035034,
      "learning_rate": 2.1656786234842237e-05,
      "loss": 0.5489,
      "step": 630
    },
    {
      "epoch": 21.75862068965517,
      "grad_norm": 3.085984230041504,
      "learning_rate": 2.1641921270239632e-05,
      "loss": 0.715,
      "step": 631
    },
    {
      "epoch": 21.79310344827586,
      "grad_norm": 3.6664793491363525,
      "learning_rate": 2.1627028457759408e-05,
      "loss": 0.5976,
      "step": 632
    },
    {
      "epoch": 21.82758620689655,
      "grad_norm": 3.8308627605438232,
      "learning_rate": 2.1612107842767647e-05,
      "loss": 0.5413,
      "step": 633
    },
    {
      "epoch": 21.862068965517242,
      "grad_norm": 3.766814947128296,
      "learning_rate": 2.1597159470715133e-05,
      "loss": 0.5849,
      "step": 634
    },
    {
      "epoch": 21.896551724137932,
      "grad_norm": 4.085217475891113,
      "learning_rate": 2.15821833871372e-05,
      "loss": 0.6535,
      "step": 635
    },
    {
      "epoch": 21.93103448275862,
      "grad_norm": 3.2848877906799316,
      "learning_rate": 2.1567179637653594e-05,
      "loss": 0.6572,
      "step": 636
    },
    {
      "epoch": 21.96551724137931,
      "grad_norm": 4.021419048309326,
      "learning_rate": 2.1552148267968347e-05,
      "loss": 0.5827,
      "step": 637
    },
    {
      "epoch": 22.0,
      "grad_norm": 3.385075569152832,
      "learning_rate": 2.1537089323869604e-05,
      "loss": 0.575,
      "step": 638
    },
    {
      "epoch": 22.03448275862069,
      "grad_norm": 3.6210548877716064,
      "learning_rate": 2.152200285122953e-05,
      "loss": 0.5036,
      "step": 639
    },
    {
      "epoch": 22.06896551724138,
      "grad_norm": 4.066530704498291,
      "learning_rate": 2.1506888896004133e-05,
      "loss": 0.5167,
      "step": 640
    },
    {
      "epoch": 22.103448275862068,
      "grad_norm": 2.326347589492798,
      "learning_rate": 2.1491747504233138e-05,
      "loss": 0.5364,
      "step": 641
    },
    {
      "epoch": 22.137931034482758,
      "grad_norm": 3.226869583129883,
      "learning_rate": 2.147657872203986e-05,
      "loss": 0.5291,
      "step": 642
    },
    {
      "epoch": 22.17241379310345,
      "grad_norm": 2.8205795288085938,
      "learning_rate": 2.1461382595631036e-05,
      "loss": 0.5034,
      "step": 643
    },
    {
      "epoch": 22.20689655172414,
      "grad_norm": 5.307386875152588,
      "learning_rate": 2.14461591712967e-05,
      "loss": 0.6833,
      "step": 644
    },
    {
      "epoch": 22.24137931034483,
      "grad_norm": 3.1979541778564453,
      "learning_rate": 2.1430908495410042e-05,
      "loss": 0.5307,
      "step": 645
    },
    {
      "epoch": 22.275862068965516,
      "grad_norm": 3.491663694381714,
      "learning_rate": 2.1415630614427272e-05,
      "loss": 0.5779,
      "step": 646
    },
    {
      "epoch": 22.310344827586206,
      "grad_norm": 2.6055986881256104,
      "learning_rate": 2.140032557488746e-05,
      "loss": 0.5429,
      "step": 647
    },
    {
      "epoch": 22.344827586206897,
      "grad_norm": 1.5949605703353882,
      "learning_rate": 2.1384993423412407e-05,
      "loss": 0.5109,
      "step": 648
    },
    {
      "epoch": 22.379310344827587,
      "grad_norm": 3.324326515197754,
      "learning_rate": 2.136963420670651e-05,
      "loss": 0.5277,
      "step": 649
    },
    {
      "epoch": 22.413793103448278,
      "grad_norm": 3.71690034866333,
      "learning_rate": 2.135424797155661e-05,
      "loss": 0.5901,
      "step": 650
    },
    {
      "epoch": 22.448275862068964,
      "grad_norm": 3.789421319961548,
      "learning_rate": 2.1338834764831845e-05,
      "loss": 0.5424,
      "step": 651
    },
    {
      "epoch": 22.482758620689655,
      "grad_norm": 3.0680930614471436,
      "learning_rate": 2.1323394633483514e-05,
      "loss": 0.6194,
      "step": 652
    },
    {
      "epoch": 22.517241379310345,
      "grad_norm": 2.454967975616455,
      "learning_rate": 2.1307927624544934e-05,
      "loss": 0.5856,
      "step": 653
    },
    {
      "epoch": 22.551724137931036,
      "grad_norm": 4.197128772735596,
      "learning_rate": 2.1292433785131298e-05,
      "loss": 0.6164,
      "step": 654
    },
    {
      "epoch": 22.586206896551722,
      "grad_norm": 4.163417339324951,
      "learning_rate": 2.1276913162439532e-05,
      "loss": 0.5499,
      "step": 655
    },
    {
      "epoch": 22.620689655172413,
      "grad_norm": 3.207831621170044,
      "learning_rate": 2.1261365803748138e-05,
      "loss": 0.5379,
      "step": 656
    },
    {
      "epoch": 22.655172413793103,
      "grad_norm": 2.1474342346191406,
      "learning_rate": 2.124579175641707e-05,
      "loss": 0.5198,
      "step": 657
    },
    {
      "epoch": 22.689655172413794,
      "grad_norm": 2.0802290439605713,
      "learning_rate": 2.1230191067887574e-05,
      "loss": 0.5988,
      "step": 658
    },
    {
      "epoch": 22.724137931034484,
      "grad_norm": 2.383214235305786,
      "learning_rate": 2.121456378568206e-05,
      "loss": 0.5644,
      "step": 659
    },
    {
      "epoch": 22.75862068965517,
      "grad_norm": 3.3427810668945312,
      "learning_rate": 2.1198909957403928e-05,
      "loss": 0.5382,
      "step": 660
    },
    {
      "epoch": 22.79310344827586,
      "grad_norm": 2.118605136871338,
      "learning_rate": 2.1183229630737467e-05,
      "loss": 0.5364,
      "step": 661
    },
    {
      "epoch": 22.82758620689655,
      "grad_norm": 2.1082427501678467,
      "learning_rate": 2.1167522853447664e-05,
      "loss": 0.5631,
      "step": 662
    },
    {
      "epoch": 22.862068965517242,
      "grad_norm": 2.0310721397399902,
      "learning_rate": 2.1151789673380086e-05,
      "loss": 0.583,
      "step": 663
    },
    {
      "epoch": 22.896551724137932,
      "grad_norm": 3.352940082550049,
      "learning_rate": 2.113603013846073e-05,
      "loss": 0.5694,
      "step": 664
    },
    {
      "epoch": 22.93103448275862,
      "grad_norm": 2.5591657161712646,
      "learning_rate": 2.1120244296695874e-05,
      "loss": 0.5605,
      "step": 665
    },
    {
      "epoch": 22.96551724137931,
      "grad_norm": 3.0828518867492676,
      "learning_rate": 2.1104432196171924e-05,
      "loss": 0.5105,
      "step": 666
    },
    {
      "epoch": 23.0,
      "grad_norm": 4.050001621246338,
      "learning_rate": 2.1088593885055288e-05,
      "loss": 0.5483,
      "step": 667
    },
    {
      "epoch": 23.03448275862069,
      "grad_norm": 3.4078168869018555,
      "learning_rate": 2.1072729411592206e-05,
      "loss": 0.5672,
      "step": 668
    },
    {
      "epoch": 23.06896551724138,
      "grad_norm": 4.16298246383667,
      "learning_rate": 2.105683882410861e-05,
      "loss": 0.6305,
      "step": 669
    },
    {
      "epoch": 23.103448275862068,
      "grad_norm": 4.402109146118164,
      "learning_rate": 2.1040922171009993e-05,
      "loss": 0.6278,
      "step": 670
    },
    {
      "epoch": 23.137931034482758,
      "grad_norm": 2.614933729171753,
      "learning_rate": 2.1024979500781232e-05,
      "loss": 0.5573,
      "step": 671
    },
    {
      "epoch": 23.17241379310345,
      "grad_norm": 5.566406726837158,
      "learning_rate": 2.1009010861986476e-05,
      "loss": 0.5987,
      "step": 672
    },
    {
      "epoch": 23.20689655172414,
      "grad_norm": 3.687830686569214,
      "learning_rate": 2.099301630326896e-05,
      "loss": 0.5332,
      "step": 673
    },
    {
      "epoch": 23.24137931034483,
      "grad_norm": 1.985662579536438,
      "learning_rate": 2.0976995873350887e-05,
      "loss": 0.5032,
      "step": 674
    },
    {
      "epoch": 23.275862068965516,
      "grad_norm": 3.1905019283294678,
      "learning_rate": 2.096094962103326e-05,
      "loss": 0.5556,
      "step": 675
    },
    {
      "epoch": 23.310344827586206,
      "grad_norm": 2.235889434814453,
      "learning_rate": 2.0944877595195755e-05,
      "loss": 0.5636,
      "step": 676
    },
    {
      "epoch": 23.344827586206897,
      "grad_norm": 4.098913669586182,
      "learning_rate": 2.092877984479654e-05,
      "loss": 0.5974,
      "step": 677
    },
    {
      "epoch": 23.379310344827587,
      "grad_norm": 5.365559101104736,
      "learning_rate": 2.091265641887217e-05,
      "loss": 0.5036,
      "step": 678
    },
    {
      "epoch": 23.413793103448278,
      "grad_norm": 3.486875534057617,
      "learning_rate": 2.089650736653738e-05,
      "loss": 0.5835,
      "step": 679
    },
    {
      "epoch": 23.448275862068964,
      "grad_norm": 6.258121967315674,
      "learning_rate": 2.088033273698499e-05,
      "loss": 0.6289,
      "step": 680
    },
    {
      "epoch": 23.482758620689655,
      "grad_norm": 3.1548774242401123,
      "learning_rate": 2.086413257948573e-05,
      "loss": 0.4888,
      "step": 681
    },
    {
      "epoch": 23.517241379310345,
      "grad_norm": 2.209894895553589,
      "learning_rate": 2.0847906943388085e-05,
      "loss": 0.4746,
      "step": 682
    },
    {
      "epoch": 23.551724137931036,
      "grad_norm": 2.9352316856384277,
      "learning_rate": 2.0831655878118155e-05,
      "loss": 0.525,
      "step": 683
    },
    {
      "epoch": 23.586206896551722,
      "grad_norm": 2.3592867851257324,
      "learning_rate": 2.081537943317951e-05,
      "loss": 0.4875,
      "step": 684
    },
    {
      "epoch": 23.620689655172413,
      "grad_norm": 3.657504081726074,
      "learning_rate": 2.0799077658153022e-05,
      "loss": 0.554,
      "step": 685
    },
    {
      "epoch": 23.655172413793103,
      "grad_norm": 6.461456298828125,
      "learning_rate": 2.0782750602696722e-05,
      "loss": 0.5205,
      "step": 686
    },
    {
      "epoch": 23.689655172413794,
      "grad_norm": 4.881630897521973,
      "learning_rate": 2.0766398316545648e-05,
      "loss": 0.5587,
      "step": 687
    },
    {
      "epoch": 23.724137931034484,
      "grad_norm": 2.7277519702911377,
      "learning_rate": 2.0750020849511712e-05,
      "loss": 0.546,
      "step": 688
    },
    {
      "epoch": 23.75862068965517,
      "grad_norm": 4.6229071617126465,
      "learning_rate": 2.0733618251483506e-05,
      "loss": 0.5404,
      "step": 689
    },
    {
      "epoch": 23.79310344827586,
      "grad_norm": 5.046594619750977,
      "learning_rate": 2.07171905724262e-05,
      "loss": 0.59,
      "step": 690
    },
    {
      "epoch": 23.82758620689655,
      "grad_norm": 4.2928853034973145,
      "learning_rate": 2.070073786238134e-05,
      "loss": 0.5288,
      "step": 691
    },
    {
      "epoch": 23.862068965517242,
      "grad_norm": 6.604857444763184,
      "learning_rate": 2.0684260171466745e-05,
      "loss": 0.6729,
      "step": 692
    },
    {
      "epoch": 23.896551724137932,
      "grad_norm": 7.314076900482178,
      "learning_rate": 2.066775754987632e-05,
      "loss": 0.6542,
      "step": 693
    },
    {
      "epoch": 23.93103448275862,
      "grad_norm": 3.7828738689422607,
      "learning_rate": 2.0651230047879905e-05,
      "loss": 0.6574,
      "step": 694
    },
    {
      "epoch": 23.96551724137931,
      "grad_norm": 3.6089084148406982,
      "learning_rate": 2.0634677715823137e-05,
      "loss": 0.6718,
      "step": 695
    },
    {
      "epoch": 24.0,
      "grad_norm": 3.206212043762207,
      "learning_rate": 2.0618100604127295e-05,
      "loss": 0.6149,
      "step": 696
    },
    {
      "epoch": 24.03448275862069,
      "grad_norm": 2.9614017009735107,
      "learning_rate": 2.0601498763289138e-05,
      "loss": 0.6372,
      "step": 697
    },
    {
      "epoch": 24.06896551724138,
      "grad_norm": 3.116159439086914,
      "learning_rate": 2.058487224388075e-05,
      "loss": 0.6268,
      "step": 698
    },
    {
      "epoch": 24.103448275862068,
      "grad_norm": 3.259319543838501,
      "learning_rate": 2.0568221096549384e-05,
      "loss": 0.5631,
      "step": 699
    },
    {
      "epoch": 24.137931034482758,
      "grad_norm": 2.3190183639526367,
      "learning_rate": 2.0551545372017332e-05,
      "loss": 0.5782,
      "step": 700
    }
  ],
  "logging_steps": 1,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 69,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}