{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1272,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0023584905660377358,
      "grad_norm": 2.078125,
      "learning_rate": 0.0,
      "loss": 1.1393,
      "step": 1
    },
    {
      "epoch": 0.0047169811320754715,
      "grad_norm": 2.375,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.2078,
      "step": 2
    },
    {
      "epoch": 0.007075471698113208,
      "grad_norm": 2.390625,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.2325,
      "step": 3
    },
    {
      "epoch": 0.009433962264150943,
      "grad_norm": 2.0625,
      "learning_rate": 1.5e-06,
      "loss": 1.0869,
      "step": 4
    },
    {
      "epoch": 0.01179245283018868,
      "grad_norm": 2.25,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.1706,
      "step": 5
    },
    {
      "epoch": 0.014150943396226415,
      "grad_norm": 2.15625,
      "learning_rate": 2.5e-06,
      "loss": 1.1445,
      "step": 6
    },
    {
      "epoch": 0.01650943396226415,
      "grad_norm": 2.203125,
      "learning_rate": 3e-06,
      "loss": 1.188,
      "step": 7
    },
    {
      "epoch": 0.018867924528301886,
      "grad_norm": 1.84375,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 1.0233,
      "step": 8
    },
    {
      "epoch": 0.02122641509433962,
      "grad_norm": 2.328125,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.208,
      "step": 9
    },
    {
      "epoch": 0.02358490566037736,
      "grad_norm": 1.828125,
      "learning_rate": 4.5e-06,
      "loss": 1.0994,
      "step": 10
    },
    {
      "epoch": 0.025943396226415096,
      "grad_norm": 2.046875,
      "learning_rate": 5e-06,
      "loss": 1.2144,
      "step": 11
    },
    {
      "epoch": 0.02830188679245283,
      "grad_norm": 1.8515625,
      "learning_rate": 5.500000000000001e-06,
      "loss": 1.1439,
      "step": 12
    },
    {
      "epoch": 0.030660377358490566,
      "grad_norm": 1.6484375,
      "learning_rate": 6e-06,
      "loss": 1.1021,
      "step": 13
    },
    {
      "epoch": 0.0330188679245283,
      "grad_norm": 1.5703125,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 1.0857,
      "step": 14
    },
    {
      "epoch": 0.03537735849056604,
      "grad_norm": 1.5390625,
      "learning_rate": 7.000000000000001e-06,
      "loss": 1.11,
      "step": 15
    },
    {
      "epoch": 0.03773584905660377,
      "grad_norm": 1.1875,
      "learning_rate": 7.5e-06,
      "loss": 0.989,
      "step": 16
    },
    {
      "epoch": 0.04009433962264151,
      "grad_norm": 1.0703125,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.0047,
      "step": 17
    },
    {
      "epoch": 0.04245283018867924,
      "grad_norm": 1.03125,
      "learning_rate": 8.500000000000002e-06,
      "loss": 1.0146,
      "step": 18
    },
    {
      "epoch": 0.04481132075471698,
      "grad_norm": 0.96484375,
      "learning_rate": 9e-06,
      "loss": 1.0734,
      "step": 19
    },
    {
      "epoch": 0.04716981132075472,
      "grad_norm": 0.9453125,
      "learning_rate": 9.5e-06,
      "loss": 1.1424,
      "step": 20
    },
    {
      "epoch": 0.049528301886792456,
      "grad_norm": 0.64453125,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 21
    },
    {
      "epoch": 0.05188679245283019,
      "grad_norm": 0.52734375,
      "learning_rate": 1.05e-05,
      "loss": 0.9029,
      "step": 22
    },
    {
      "epoch": 0.054245283018867926,
      "grad_norm": 0.51953125,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.9315,
      "step": 23
    },
    {
      "epoch": 0.05660377358490566,
      "grad_norm": 0.5078125,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 0.941,
      "step": 24
    },
    {
      "epoch": 0.0589622641509434,
      "grad_norm": 0.484375,
      "learning_rate": 1.2e-05,
      "loss": 0.9998,
      "step": 25
    },
    {
      "epoch": 0.06132075471698113,
      "grad_norm": 0.423828125,
      "learning_rate": 1.25e-05,
      "loss": 0.8876,
      "step": 26
    },
    {
      "epoch": 0.06367924528301887,
      "grad_norm": 0.37890625,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.8619,
      "step": 27
    },
    {
      "epoch": 0.0660377358490566,
      "grad_norm": 0.322265625,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.8723,
      "step": 28
    },
    {
      "epoch": 0.06839622641509434,
      "grad_norm": 0.3046875,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.8539,
      "step": 29
    },
    {
      "epoch": 0.07075471698113207,
      "grad_norm": 0.31640625,
      "learning_rate": 1.45e-05,
      "loss": 0.9227,
      "step": 30
    },
    {
      "epoch": 0.07311320754716981,
      "grad_norm": 0.275390625,
      "learning_rate": 1.5e-05,
      "loss": 0.8836,
      "step": 31
    },
    {
      "epoch": 0.07547169811320754,
      "grad_norm": 0.263671875,
      "learning_rate": 1.55e-05,
      "loss": 0.8709,
      "step": 32
    },
    {
      "epoch": 0.07783018867924528,
      "grad_norm": 0.23828125,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.8388,
      "step": 33
    },
    {
      "epoch": 0.08018867924528301,
      "grad_norm": 0.251953125,
      "learning_rate": 1.65e-05,
      "loss": 0.8783,
      "step": 34
    },
    {
      "epoch": 0.08254716981132075,
      "grad_norm": 0.2333984375,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.8681,
      "step": 35
    },
    {
      "epoch": 0.08490566037735849,
      "grad_norm": 0.2265625,
      "learning_rate": 1.75e-05,
      "loss": 0.9061,
      "step": 36
    },
    {
      "epoch": 0.08726415094339622,
      "grad_norm": 0.228515625,
      "learning_rate": 1.8e-05,
      "loss": 0.906,
      "step": 37
    },
    {
      "epoch": 0.08962264150943396,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.85e-05,
      "loss": 0.7402,
      "step": 38
    },
    {
      "epoch": 0.09198113207547169,
      "grad_norm": 0.330078125,
      "learning_rate": 1.9e-05,
      "loss": 0.8487,
      "step": 39
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 0.8298,
      "step": 40
    },
    {
      "epoch": 0.09669811320754718,
      "grad_norm": 0.2021484375,
      "learning_rate": 2e-05,
      "loss": 0.8189,
      "step": 41
    },
    {
      "epoch": 0.09905660377358491,
      "grad_norm": 0.19140625,
      "learning_rate": 2.05e-05,
      "loss": 0.8084,
      "step": 42
    },
    {
      "epoch": 0.10141509433962265,
      "grad_norm": 0.2109375,
      "learning_rate": 2.1e-05,
      "loss": 0.8351,
      "step": 43
    },
    {
      "epoch": 0.10377358490566038,
      "grad_norm": 0.169921875,
      "learning_rate": 2.15e-05,
      "loss": 0.7304,
      "step": 44
    },
    {
      "epoch": 0.10613207547169812,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.7718,
      "step": 45
    },
    {
      "epoch": 0.10849056603773585,
      "grad_norm": 0.20703125,
      "learning_rate": 2.25e-05,
      "loss": 0.8098,
      "step": 46
    },
    {
      "epoch": 0.11084905660377359,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.8681,
      "step": 47
    },
    {
      "epoch": 0.11320754716981132,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.35e-05,
      "loss": 0.7755,
      "step": 48
    },
    {
      "epoch": 0.11556603773584906,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.4e-05,
      "loss": 0.7966,
      "step": 49
    },
    {
      "epoch": 0.1179245283018868,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.45e-05,
      "loss": 0.7335,
      "step": 50
    },
    {
      "epoch": 0.12028301886792453,
      "grad_norm": 0.21484375,
      "learning_rate": 2.5e-05,
      "loss": 0.769,
      "step": 51
    },
    {
      "epoch": 0.12264150943396226,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.7142,
      "step": 52
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.7055,
      "step": 53
    },
    {
      "epoch": 0.12735849056603774,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 0.6973,
      "step": 54
    },
    {
      "epoch": 0.12971698113207547,
      "grad_norm": 0.208984375,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.7549,
      "step": 55
    },
    {
      "epoch": 0.1320754716981132,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.6616,
      "step": 56
    },
    {
      "epoch": 0.13443396226415094,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.6457,
      "step": 57
    },
    {
      "epoch": 0.13679245283018868,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 0.643,
      "step": 58
    },
    {
      "epoch": 0.1391509433962264,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.9e-05,
      "loss": 0.5876,
      "step": 59
    },
    {
      "epoch": 0.14150943396226415,
      "grad_norm": 0.244140625,
      "learning_rate": 2.95e-05,
      "loss": 0.6435,
      "step": 60
    },
    {
      "epoch": 0.14386792452830188,
      "grad_norm": 0.2080078125,
      "learning_rate": 3e-05,
      "loss": 0.6658,
      "step": 61
    },
    {
      "epoch": 0.14622641509433962,
      "grad_norm": 0.2294921875,
      "learning_rate": 3.05e-05,
      "loss": 0.6052,
      "step": 62
    },
    {
      "epoch": 0.14858490566037735,
      "grad_norm": 0.212890625,
      "learning_rate": 3.1e-05,
      "loss": 0.5963,
      "step": 63
    },
    {
      "epoch": 0.1509433962264151,
      "grad_norm": 0.2451171875,
      "learning_rate": 3.15e-05,
      "loss": 0.5722,
      "step": 64
    },
    {
      "epoch": 0.15330188679245282,
      "grad_norm": 0.2294921875,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.5896,
      "step": 65
    },
    {
      "epoch": 0.15566037735849056,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.5465,
      "step": 66
    },
    {
      "epoch": 0.1580188679245283,
      "grad_norm": 0.248046875,
      "learning_rate": 3.3e-05,
      "loss": 0.4922,
      "step": 67
    },
    {
      "epoch": 0.16037735849056603,
      "grad_norm": 0.263671875,
      "learning_rate": 3.35e-05,
      "loss": 0.5559,
      "step": 68
    },
    {
      "epoch": 0.16273584905660377,
      "grad_norm": 0.29296875,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.4788,
      "step": 69
    },
    {
      "epoch": 0.1650943396226415,
      "grad_norm": 0.2265625,
      "learning_rate": 3.45e-05,
      "loss": 0.4556,
      "step": 70
    },
    {
      "epoch": 0.16745283018867924,
      "grad_norm": 0.349609375,
      "learning_rate": 3.5e-05,
      "loss": 0.4199,
      "step": 71
    },
    {
      "epoch": 0.16981132075471697,
      "grad_norm": 0.279296875,
      "learning_rate": 3.55e-05,
      "loss": 0.4406,
      "step": 72
    },
    {
      "epoch": 0.1721698113207547,
      "grad_norm": 0.3125,
      "learning_rate": 3.6e-05,
      "loss": 0.3566,
      "step": 73
    },
    {
      "epoch": 0.17452830188679244,
      "grad_norm": 0.2412109375,
      "learning_rate": 3.65e-05,
      "loss": 0.3495,
      "step": 74
    },
    {
      "epoch": 0.17688679245283018,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.7e-05,
      "loss": 0.3582,
      "step": 75
    },
    {
      "epoch": 0.1792452830188679,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.3605,
      "step": 76
    },
    {
      "epoch": 0.18160377358490565,
      "grad_norm": 0.2412109375,
      "learning_rate": 3.8e-05,
      "loss": 0.3715,
      "step": 77
    },
    {
      "epoch": 0.18396226415094338,
      "grad_norm": 0.15234375,
      "learning_rate": 3.85e-05,
      "loss": 0.3467,
      "step": 78
    },
    {
      "epoch": 0.18632075471698112,
      "grad_norm": 0.166015625,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.3252,
      "step": 79
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 0.17578125,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 0.3505,
      "step": 80
    },
    {
      "epoch": 0.19103773584905662,
      "grad_norm": 0.1552734375,
      "learning_rate": 4e-05,
      "loss": 0.3695,
      "step": 81
    },
    {
      "epoch": 0.19339622641509435,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.05e-05,
      "loss": 0.3293,
      "step": 82
    },
    {
      "epoch": 0.1957547169811321,
      "grad_norm": 0.1357421875,
      "learning_rate": 4.1e-05,
      "loss": 0.3556,
      "step": 83
    },
    {
      "epoch": 0.19811320754716982,
      "grad_norm": 0.15625,
      "learning_rate": 4.15e-05,
      "loss": 0.2957,
      "step": 84
    },
    {
      "epoch": 0.20047169811320756,
      "grad_norm": 0.13671875,
      "learning_rate": 4.2e-05,
      "loss": 0.321,
      "step": 85
    },
    {
      "epoch": 0.2028301886792453,
      "grad_norm": 0.248046875,
      "learning_rate": 4.25e-05,
      "loss": 0.2577,
      "step": 86
    },
    {
      "epoch": 0.20518867924528303,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.3e-05,
      "loss": 0.2795,
      "step": 87
    },
    {
      "epoch": 0.20754716981132076,
      "grad_norm": 0.1396484375,
      "learning_rate": 4.35e-05,
      "loss": 0.3022,
      "step": 88
    },
    {
      "epoch": 0.2099056603773585,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3011,
      "step": 89
    },
    {
      "epoch": 0.21226415094339623,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 0.3602,
      "step": 90
    },
    {
      "epoch": 0.21462264150943397,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.5e-05,
      "loss": 0.3046,
      "step": 91
    },
    {
      "epoch": 0.2169811320754717,
      "grad_norm": 0.15234375,
      "learning_rate": 4.55e-05,
      "loss": 0.2893,
      "step": 92
    },
    {
      "epoch": 0.21933962264150944,
      "grad_norm": 0.13671875,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2794,
      "step": 93
    },
    {
      "epoch": 0.22169811320754718,
      "grad_norm": 0.142578125,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 0.2873,
      "step": 94
    },
    {
      "epoch": 0.2240566037735849,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.7e-05,
      "loss": 0.2767,
      "step": 95
    },
    {
      "epoch": 0.22641509433962265,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.75e-05,
      "loss": 0.2628,
      "step": 96
    },
    {
      "epoch": 0.22877358490566038,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.8e-05,
      "loss": 0.261,
      "step": 97
    },
    {
      "epoch": 0.23113207547169812,
      "grad_norm": 0.1455078125,
      "learning_rate": 4.85e-05,
      "loss": 0.3502,
      "step": 98
    },
    {
      "epoch": 0.23349056603773585,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.9e-05,
      "loss": 0.25,
      "step": 99
    },
    {
      "epoch": 0.2358490566037736,
      "grad_norm": 0.1435546875,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 0.2733,
      "step": 100
    },
    {
      "epoch": 0.23820754716981132,
      "grad_norm": 0.1943359375,
      "learning_rate": 5e-05,
      "loss": 0.3216,
      "step": 101
    },
    {
      "epoch": 0.24056603773584906,
      "grad_norm": 0.15234375,
      "learning_rate": 4.998792270531401e-05,
      "loss": 0.2627,
      "step": 102
    },
    {
      "epoch": 0.2429245283018868,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.997584541062802e-05,
      "loss": 0.3059,
      "step": 103
    },
    {
      "epoch": 0.24528301886792453,
      "grad_norm": 0.14453125,
      "learning_rate": 4.9963768115942036e-05,
      "loss": 0.3102,
      "step": 104
    },
    {
      "epoch": 0.24764150943396226,
      "grad_norm": 0.140625,
      "learning_rate": 4.9951690821256045e-05,
      "loss": 0.3018,
      "step": 105
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.15625,
      "learning_rate": 4.9939613526570054e-05,
      "loss": 0.3077,
      "step": 106
    },
    {
      "epoch": 0.25235849056603776,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.9927536231884056e-05,
      "loss": 0.2928,
      "step": 107
    },
    {
      "epoch": 0.25471698113207547,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.9915458937198065e-05,
      "loss": 0.2573,
      "step": 108
    },
    {
      "epoch": 0.25707547169811323,
      "grad_norm": 0.154296875,
      "learning_rate": 4.990338164251208e-05,
      "loss": 0.244,
      "step": 109
    },
    {
      "epoch": 0.25943396226415094,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.989130434782609e-05,
      "loss": 0.2972,
      "step": 110
    },
    {
      "epoch": 0.2617924528301887,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.98792270531401e-05,
      "loss": 0.3234,
      "step": 111
    },
    {
      "epoch": 0.2641509433962264,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.986714975845411e-05,
      "loss": 0.2901,
      "step": 112
    },
    {
      "epoch": 0.2665094339622642,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.985507246376812e-05,
      "loss": 0.3195,
      "step": 113
    },
    {
      "epoch": 0.2688679245283019,
      "grad_norm": 0.1484375,
      "learning_rate": 4.984299516908213e-05,
      "loss": 0.2951,
      "step": 114
    },
    {
      "epoch": 0.27122641509433965,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.983091787439614e-05,
      "loss": 0.2182,
      "step": 115
    },
    {
      "epoch": 0.27358490566037735,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.981884057971015e-05,
      "loss": 0.2401,
      "step": 116
    },
    {
      "epoch": 0.2759433962264151,
      "grad_norm": 0.162109375,
      "learning_rate": 4.980676328502415e-05,
      "loss": 0.2992,
      "step": 117
    },
    {
      "epoch": 0.2783018867924528,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.979468599033817e-05,
      "loss": 0.3225,
      "step": 118
    },
    {
      "epoch": 0.2806603773584906,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.9782608695652176e-05,
      "loss": 0.2677,
      "step": 119
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 0.15234375,
      "learning_rate": 4.9770531400966185e-05,
      "loss": 0.2855,
      "step": 120
    },
    {
      "epoch": 0.28537735849056606,
      "grad_norm": 0.14453125,
      "learning_rate": 4.9758454106280194e-05,
      "loss": 0.2437,
      "step": 121
    },
    {
      "epoch": 0.28773584905660377,
      "grad_norm": 0.1494140625,
      "learning_rate": 4.97463768115942e-05,
      "loss": 0.3062,
      "step": 122
    },
    {
      "epoch": 0.29009433962264153,
      "grad_norm": 0.16015625,
      "learning_rate": 4.973429951690822e-05,
      "loss": 0.2605,
      "step": 123
    },
    {
      "epoch": 0.29245283018867924,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.972222222222223e-05,
      "loss": 0.3214,
      "step": 124
    },
    {
      "epoch": 0.294811320754717,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.9710144927536237e-05,
      "loss": 0.2647,
      "step": 125
    },
    {
      "epoch": 0.2971698113207547,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.9698067632850245e-05,
      "loss": 0.2728,
      "step": 126
    },
    {
      "epoch": 0.29952830188679247,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.9685990338164254e-05,
      "loss": 0.2834,
      "step": 127
    },
    {
      "epoch": 0.3018867924528302,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.967391304347826e-05,
      "loss": 0.2503,
      "step": 128
    },
    {
      "epoch": 0.30424528301886794,
      "grad_norm": 0.15234375,
      "learning_rate": 4.966183574879227e-05,
      "loss": 0.2993,
      "step": 129
    },
    {
      "epoch": 0.30660377358490565,
      "grad_norm": 0.154296875,
      "learning_rate": 4.964975845410628e-05,
      "loss": 0.2504,
      "step": 130
    },
    {
      "epoch": 0.3089622641509434,
      "grad_norm": 0.162109375,
      "learning_rate": 4.963768115942029e-05,
      "loss": 0.3401,
      "step": 131
    },
    {
      "epoch": 0.3113207547169811,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.96256038647343e-05,
      "loss": 0.3325,
      "step": 132
    },
    {
      "epoch": 0.3136792452830189,
      "grad_norm": 0.1494140625,
      "learning_rate": 4.9613526570048315e-05,
      "loss": 0.2741,
      "step": 133
    },
    {
      "epoch": 0.3160377358490566,
      "grad_norm": 0.173828125,
      "learning_rate": 4.9601449275362324e-05,
      "loss": 0.2868,
      "step": 134
    },
    {
      "epoch": 0.31839622641509435,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.958937198067633e-05,
      "loss": 0.3247,
      "step": 135
    },
    {
      "epoch": 0.32075471698113206,
      "grad_norm": 0.146484375,
      "learning_rate": 4.957729468599034e-05,
      "loss": 0.2917,
      "step": 136
    },
    {
      "epoch": 0.3231132075471698,
      "grad_norm": 0.16015625,
      "learning_rate": 4.956521739130435e-05,
      "loss": 0.2444,
      "step": 137
    },
    {
      "epoch": 0.32547169811320753,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.955314009661836e-05,
      "loss": 0.2287,
      "step": 138
    },
    {
      "epoch": 0.3278301886792453,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.954106280193237e-05,
      "loss": 0.3229,
      "step": 139
    },
    {
      "epoch": 0.330188679245283,
      "grad_norm": 0.1640625,
      "learning_rate": 4.952898550724638e-05,
      "loss": 0.2322,
      "step": 140
    },
    {
      "epoch": 0.33254716981132076,
      "grad_norm": 0.162109375,
      "learning_rate": 4.9516908212560386e-05,
      "loss": 0.2876,
      "step": 141
    },
    {
      "epoch": 0.33490566037735847,
      "grad_norm": 0.1494140625,
      "learning_rate": 4.95048309178744e-05,
      "loss": 0.2596,
      "step": 142
    },
    {
      "epoch": 0.33726415094339623,
      "grad_norm": 0.1875,
      "learning_rate": 4.949275362318841e-05,
      "loss": 0.2916,
      "step": 143
    },
    {
      "epoch": 0.33962264150943394,
      "grad_norm": 0.1484375,
      "learning_rate": 4.948067632850242e-05,
      "loss": 0.2786,
      "step": 144
    },
    {
      "epoch": 0.3419811320754717,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.946859903381643e-05,
      "loss": 0.2569,
      "step": 145
    },
    {
      "epoch": 0.3443396226415094,
      "grad_norm": 0.158203125,
      "learning_rate": 4.945652173913044e-05,
      "loss": 0.3018,
      "step": 146
    },
    {
      "epoch": 0.3466981132075472,
      "grad_norm": 0.1875,
      "learning_rate": 4.9444444444444446e-05,
      "loss": 0.2931,
      "step": 147
    },
    {
      "epoch": 0.3490566037735849,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.9432367149758455e-05,
      "loss": 0.2308,
      "step": 148
    },
    {
      "epoch": 0.35141509433962265,
      "grad_norm": 0.16796875,
      "learning_rate": 4.9420289855072464e-05,
      "loss": 0.2381,
      "step": 149
    },
    {
      "epoch": 0.35377358490566035,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.940821256038647e-05,
      "loss": 0.2218,
      "step": 150
    },
    {
      "epoch": 0.3561320754716981,
      "grad_norm": 0.169921875,
      "learning_rate": 4.939613526570048e-05,
      "loss": 0.2723,
      "step": 151
    },
    {
      "epoch": 0.3584905660377358,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.93840579710145e-05,
      "loss": 0.2527,
      "step": 152
    },
    {
      "epoch": 0.3608490566037736,
      "grad_norm": 0.1640625,
      "learning_rate": 4.9371980676328506e-05,
      "loss": 0.3174,
      "step": 153
    },
    {
      "epoch": 0.3632075471698113,
      "grad_norm": 0.2109375,
      "learning_rate": 4.9359903381642515e-05,
      "loss": 0.3323,
      "step": 154
    },
    {
      "epoch": 0.36556603773584906,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.9347826086956524e-05,
      "loss": 0.255,
      "step": 155
    },
    {
      "epoch": 0.36792452830188677,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.933574879227053e-05,
      "loss": 0.297,
      "step": 156
    },
    {
      "epoch": 0.37028301886792453,
      "grad_norm": 0.15234375,
      "learning_rate": 4.932367149758454e-05,
      "loss": 0.2665,
      "step": 157
    },
    {
      "epoch": 0.37264150943396224,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.931159420289855e-05,
      "loss": 0.2348,
      "step": 158
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.929951690821256e-05,
      "loss": 0.2566,
      "step": 159
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 0.169921875,
      "learning_rate": 4.928743961352657e-05,
      "loss": 0.2809,
      "step": 160
    },
    {
      "epoch": 0.37971698113207547,
      "grad_norm": 0.19140625,
      "learning_rate": 4.9275362318840584e-05,
      "loss": 0.2623,
      "step": 161
    },
    {
      "epoch": 0.38207547169811323,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.926328502415459e-05,
      "loss": 0.269,
      "step": 162
    },
    {
      "epoch": 0.38443396226415094,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.92512077294686e-05,
      "loss": 0.2316,
      "step": 163
    },
    {
      "epoch": 0.3867924528301887,
      "grad_norm": 0.173828125,
      "learning_rate": 4.923913043478261e-05,
      "loss": 0.204,
      "step": 164
    },
    {
      "epoch": 0.3891509433962264,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.922705314009662e-05,
      "loss": 0.2431,
      "step": 165
    },
    {
      "epoch": 0.3915094339622642,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.9214975845410636e-05,
      "loss": 0.3052,
      "step": 166
    },
    {
      "epoch": 0.3938679245283019,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.920289855072464e-05,
      "loss": 0.3119,
      "step": 167
    },
    {
      "epoch": 0.39622641509433965,
      "grad_norm": 0.181640625,
      "learning_rate": 4.919082125603865e-05,
      "loss": 0.3632,
      "step": 168
    },
    {
      "epoch": 0.39858490566037735,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.9178743961352656e-05,
      "loss": 0.2698,
      "step": 169
    },
    {
      "epoch": 0.4009433962264151,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.9166666666666665e-05,
      "loss": 0.2451,
      "step": 170
    },
    {
      "epoch": 0.4033018867924528,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.915458937198068e-05,
      "loss": 0.2578,
      "step": 171
    },
    {
      "epoch": 0.4056603773584906,
      "grad_norm": 0.181640625,
      "learning_rate": 4.914251207729469e-05,
      "loss": 0.2744,
      "step": 172
    },
    {
      "epoch": 0.4080188679245283,
      "grad_norm": 0.17578125,
      "learning_rate": 4.91304347826087e-05,
      "loss": 0.2876,
      "step": 173
    },
    {
      "epoch": 0.41037735849056606,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.911835748792271e-05,
      "loss": 0.2372,
      "step": 174
    },
    {
      "epoch": 0.41273584905660377,
      "grad_norm": 0.173828125,
      "learning_rate": 4.910628019323672e-05,
      "loss": 0.2721,
      "step": 175
    },
    {
      "epoch": 0.41509433962264153,
      "grad_norm": 0.16015625,
      "learning_rate": 4.909420289855073e-05,
      "loss": 0.2804,
      "step": 176
    },
    {
      "epoch": 0.41745283018867924,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.9082125603864734e-05,
      "loss": 0.2782,
      "step": 177
    },
    {
      "epoch": 0.419811320754717,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.907004830917874e-05,
      "loss": 0.2766,
      "step": 178
    },
    {
      "epoch": 0.4221698113207547,
      "grad_norm": 0.1640625,
      "learning_rate": 4.905797101449275e-05,
      "loss": 0.2826,
      "step": 179
    },
    {
      "epoch": 0.42452830188679247,
      "grad_norm": 0.162109375,
      "learning_rate": 4.904589371980677e-05,
      "loss": 0.2456,
      "step": 180
    },
    {
      "epoch": 0.4268867924528302,
      "grad_norm": 0.212890625,
      "learning_rate": 4.9033816425120776e-05,
      "loss": 0.2044,
      "step": 181
    },
    {
      "epoch": 0.42924528301886794,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.9021739130434785e-05,
      "loss": 0.2233,
      "step": 182
    },
    {
      "epoch": 0.43160377358490565,
      "grad_norm": 0.173828125,
      "learning_rate": 4.9009661835748794e-05,
      "loss": 0.2478,
      "step": 183
    },
    {
      "epoch": 0.4339622641509434,
      "grad_norm": 0.171875,
      "learning_rate": 4.89975845410628e-05,
      "loss": 0.2646,
      "step": 184
    },
    {
      "epoch": 0.4363207547169811,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.898550724637682e-05,
      "loss": 0.2858,
      "step": 185
    },
    {
      "epoch": 0.4386792452830189,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.897342995169083e-05,
      "loss": 0.2512,
      "step": 186
    },
    {
      "epoch": 0.4410377358490566,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.8961352657004836e-05,
      "loss": 0.2303,
      "step": 187
    },
    {
      "epoch": 0.44339622641509435,
      "grad_norm": 0.166015625,
      "learning_rate": 4.894927536231884e-05,
      "loss": 0.2185,
      "step": 188
    },
    {
      "epoch": 0.44575471698113206,
      "grad_norm": 0.158203125,
      "learning_rate": 4.893719806763285e-05,
      "loss": 0.299,
      "step": 189
    },
    {
      "epoch": 0.4481132075471698,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.892512077294686e-05,
      "loss": 0.2568,
      "step": 190
    },
    {
      "epoch": 0.45047169811320753,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.891304347826087e-05,
      "loss": 0.2343,
      "step": 191
    },
    {
      "epoch": 0.4528301886792453,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.890096618357488e-05,
      "loss": 0.2598,
      "step": 192
    },
    {
      "epoch": 0.455188679245283,
      "grad_norm": 0.16015625,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.2554,
      "step": 193
    },
    {
      "epoch": 0.45754716981132076,
      "grad_norm": 0.15625,
      "learning_rate": 4.88768115942029e-05,
      "loss": 0.2337,
      "step": 194
    },
    {
      "epoch": 0.45990566037735847,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.8864734299516914e-05,
      "loss": 0.2328,
      "step": 195
    },
    {
      "epoch": 0.46226415094339623,
      "grad_norm": 0.162109375,
      "learning_rate": 4.885265700483092e-05,
      "loss": 0.212,
      "step": 196
    },
    {
      "epoch": 0.46462264150943394,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.884057971014493e-05,
      "loss": 0.1992,
      "step": 197
    },
    {
      "epoch": 0.4669811320754717,
      "grad_norm": 0.185546875,
      "learning_rate": 4.8828502415458934e-05,
      "loss": 0.2715,
      "step": 198
    },
    {
      "epoch": 0.4693396226415094,
      "grad_norm": 0.169921875,
      "learning_rate": 4.881642512077295e-05,
      "loss": 0.2522,
      "step": 199
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.880434782608696e-05,
      "loss": 0.1978,
      "step": 200
    },
    {
      "epoch": 0.4740566037735849,
      "grad_norm": 0.1640625,
      "learning_rate": 4.879227053140097e-05,
      "loss": 0.1826,
      "step": 201
    },
    {
      "epoch": 0.47641509433962265,
      "grad_norm": 0.1640625,
      "learning_rate": 4.878019323671498e-05,
      "loss": 0.234,
      "step": 202
    },
    {
      "epoch": 0.47877358490566035,
      "grad_norm": 0.166015625,
      "learning_rate": 4.8768115942028986e-05,
      "loss": 0.2578,
      "step": 203
    },
    {
      "epoch": 0.4811320754716981,
      "grad_norm": 0.1875,
      "learning_rate": 4.8756038647343e-05,
      "loss": 0.2545,
      "step": 204
    },
    {
      "epoch": 0.4834905660377358,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.874396135265701e-05,
      "loss": 0.3121,
      "step": 205
    },
    {
      "epoch": 0.4858490566037736,
      "grad_norm": 0.16015625,
      "learning_rate": 4.873188405797102e-05,
      "loss": 0.2773,
      "step": 206
    },
    {
      "epoch": 0.4882075471698113,
      "grad_norm": 0.169921875,
      "learning_rate": 4.871980676328503e-05,
      "loss": 0.284,
      "step": 207
    },
    {
      "epoch": 0.49056603773584906,
      "grad_norm": 0.166015625,
      "learning_rate": 4.870772946859904e-05,
      "loss": 0.2421,
      "step": 208
    },
    {
      "epoch": 0.49292452830188677,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.8695652173913046e-05,
      "loss": 0.2508,
      "step": 209
    },
    {
      "epoch": 0.49528301886792453,
      "grad_norm": 0.15234375,
      "learning_rate": 4.8683574879227055e-05,
      "loss": 0.2496,
      "step": 210
    },
    {
      "epoch": 0.49764150943396224,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.8671497584541064e-05,
      "loss": 0.2707,
      "step": 211
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.865942028985507e-05,
      "loss": 0.2424,
      "step": 212
    },
    {
      "epoch": 0.5023584905660378,
      "grad_norm": 0.1640625,
      "learning_rate": 4.864734299516908e-05,
      "loss": 0.2821,
      "step": 213
    },
    {
      "epoch": 0.5047169811320755,
      "grad_norm": 0.15625,
      "learning_rate": 4.86352657004831e-05,
      "loss": 0.2835,
      "step": 214
    },
    {
      "epoch": 0.5070754716981132,
      "grad_norm": 0.15625,
      "learning_rate": 4.8623188405797106e-05,
      "loss": 0.2487,
      "step": 215
    },
    {
      "epoch": 0.5094339622641509,
      "grad_norm": 0.1640625,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 0.2029,
      "step": 216
    },
    {
      "epoch": 0.5117924528301887,
      "grad_norm": 0.15625,
      "learning_rate": 4.8599033816425124e-05,
      "loss": 0.1669,
      "step": 217
    },
    {
      "epoch": 0.5141509433962265,
      "grad_norm": 0.173828125,
      "learning_rate": 4.858695652173913e-05,
      "loss": 0.2651,
      "step": 218
    },
    {
      "epoch": 0.5165094339622641,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.857487922705314e-05,
      "loss": 0.2581,
      "step": 219
    },
    {
      "epoch": 0.5188679245283019,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.856280193236715e-05,
      "loss": 0.2689,
      "step": 220
    },
    {
      "epoch": 0.5212264150943396,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.855072463768116e-05,
      "loss": 0.2661,
      "step": 221
    },
    {
      "epoch": 0.5235849056603774,
      "grad_norm": 0.166015625,
      "learning_rate": 4.853864734299517e-05,
      "loss": 0.1716,
      "step": 222
    },
    {
      "epoch": 0.5259433962264151,
      "grad_norm": 0.1796875,
      "learning_rate": 4.8526570048309184e-05,
      "loss": 0.2253,
      "step": 223
    },
    {
      "epoch": 0.5283018867924528,
      "grad_norm": 0.177734375,
      "learning_rate": 4.851449275362319e-05,
      "loss": 0.2869,
      "step": 224
    },
    {
      "epoch": 0.5306603773584906,
      "grad_norm": 0.173828125,
      "learning_rate": 4.85024154589372e-05,
      "loss": 0.227,
      "step": 225
    },
    {
      "epoch": 0.5330188679245284,
      "grad_norm": 0.173828125,
      "learning_rate": 4.849033816425121e-05,
      "loss": 0.2509,
      "step": 226
    },
    {
      "epoch": 0.535377358490566,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.847826086956522e-05,
      "loss": 0.2557,
      "step": 227
    },
    {
      "epoch": 0.5377358490566038,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.846618357487923e-05,
      "loss": 0.2634,
      "step": 228
    },
    {
      "epoch": 0.5400943396226415,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.845410628019324e-05,
      "loss": 0.2225,
      "step": 229
    },
    {
      "epoch": 0.5424528301886793,
      "grad_norm": 0.18359375,
      "learning_rate": 4.844202898550725e-05,
      "loss": 0.2496,
      "step": 230
    },
    {
      "epoch": 0.5448113207547169,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.8429951690821256e-05,
      "loss": 0.2011,
      "step": 231
    },
    {
      "epoch": 0.5471698113207547,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.8417874396135264e-05,
      "loss": 0.2566,
      "step": 232
    },
    {
      "epoch": 0.5495283018867925,
      "grad_norm": 0.14453125,
      "learning_rate": 4.840579710144928e-05,
      "loss": 0.3044,
      "step": 233
    },
    {
      "epoch": 0.5518867924528302,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.839371980676329e-05,
      "loss": 0.2312,
      "step": 234
    },
    {
      "epoch": 0.5542452830188679,
      "grad_norm": 0.17578125,
      "learning_rate": 4.83816425120773e-05,
      "loss": 0.2843,
      "step": 235
    },
    {
      "epoch": 0.5566037735849056,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.836956521739131e-05,
      "loss": 0.2653,
      "step": 236
    },
    {
      "epoch": 0.5589622641509434,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.8357487922705316e-05,
      "loss": 0.2075,
      "step": 237
    },
    {
      "epoch": 0.5613207547169812,
      "grad_norm": 0.162109375,
      "learning_rate": 4.8345410628019325e-05,
      "loss": 0.2522,
      "step": 238
    },
    {
      "epoch": 0.5636792452830188,
      "grad_norm": 0.166015625,
      "learning_rate": 4.8333333333333334e-05,
      "loss": 0.2598,
      "step": 239
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 0.17578125,
      "learning_rate": 4.832125603864734e-05,
      "loss": 0.2634,
      "step": 240
    },
    {
      "epoch": 0.5683962264150944,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.830917874396135e-05,
      "loss": 0.2398,
      "step": 241
    },
    {
      "epoch": 0.5707547169811321,
      "grad_norm": 0.177734375,
      "learning_rate": 4.829710144927537e-05,
      "loss": 0.2774,
      "step": 242
    },
    {
      "epoch": 0.5731132075471698,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.8285024154589376e-05,
      "loss": 0.2017,
      "step": 243
    },
    {
      "epoch": 0.5754716981132075,
      "grad_norm": 0.1298828125,
      "learning_rate": 4.8272946859903385e-05,
      "loss": 0.2774,
      "step": 244
    },
    {
      "epoch": 0.5778301886792453,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.8260869565217394e-05,
      "loss": 0.2557,
      "step": 245
    },
    {
      "epoch": 0.5801886792452831,
      "grad_norm": 0.169921875,
      "learning_rate": 4.82487922705314e-05,
      "loss": 0.2128,
      "step": 246
    },
    {
      "epoch": 0.5825471698113207,
      "grad_norm": 0.1640625,
      "learning_rate": 4.823671497584542e-05,
      "loss": 0.2628,
      "step": 247
    },
    {
      "epoch": 0.5849056603773585,
      "grad_norm": 0.171875,
      "learning_rate": 4.822463768115942e-05,
      "loss": 0.2388,
      "step": 248
    },
    {
      "epoch": 0.5872641509433962,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.821256038647343e-05,
      "loss": 0.2672,
      "step": 249
    },
    {
      "epoch": 0.589622641509434,
      "grad_norm": 0.166015625,
      "learning_rate": 4.820048309178744e-05,
      "loss": 0.1935,
      "step": 250
    },
    {
      "epoch": 0.5919811320754716,
      "grad_norm": 0.16796875,
      "learning_rate": 4.818840579710145e-05,
      "loss": 0.2481,
      "step": 251
    },
    {
      "epoch": 0.5943396226415094,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.817632850241546e-05,
      "loss": 0.2388,
      "step": 252
    },
    {
      "epoch": 0.5966981132075472,
      "grad_norm": 0.1640625,
      "learning_rate": 4.816425120772947e-05,
      "loss": 0.239,
      "step": 253
    },
    {
      "epoch": 0.5990566037735849,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.815217391304348e-05,
      "loss": 0.2519,
      "step": 254
    },
    {
      "epoch": 0.6014150943396226,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.814009661835749e-05,
      "loss": 0.2811,
      "step": 255
    },
    {
      "epoch": 0.6037735849056604,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.81280193236715e-05,
      "loss": 0.2092,
      "step": 256
    },
    {
      "epoch": 0.6061320754716981,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.8115942028985514e-05,
      "loss": 0.2322,
      "step": 257
    },
    {
      "epoch": 0.6084905660377359,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.8103864734299516e-05,
      "loss": 0.2925,
      "step": 258
    },
    {
      "epoch": 0.6108490566037735,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.8091787439613525e-05,
      "loss": 0.2686,
      "step": 259
    },
    {
      "epoch": 0.6132075471698113,
      "grad_norm": 0.185546875,
      "learning_rate": 4.8079710144927534e-05,
      "loss": 0.2795,
      "step": 260
    },
    {
      "epoch": 0.6155660377358491,
      "grad_norm": 0.166015625,
      "learning_rate": 4.806763285024155e-05,
      "loss": 0.2429,
      "step": 261
    },
    {
      "epoch": 0.6179245283018868,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.805555555555556e-05,
      "loss": 0.2307,
      "step": 262
    },
    {
      "epoch": 0.6202830188679245,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.804347826086957e-05,
      "loss": 0.2156,
      "step": 263
    },
    {
      "epoch": 0.6226415094339622,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.803140096618358e-05,
      "loss": 0.2741,
      "step": 264
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.173828125,
      "learning_rate": 4.8019323671497586e-05,
      "loss": 0.2278,
      "step": 265
    },
    {
      "epoch": 0.6273584905660378,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.80072463768116e-05,
      "loss": 0.2745,
      "step": 266
    },
    {
      "epoch": 0.6297169811320755,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.799516908212561e-05,
      "loss": 0.2378,
      "step": 267
    },
    {
      "epoch": 0.6320754716981132,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.798309178743962e-05,
      "loss": 0.3189,
      "step": 268
    },
    {
      "epoch": 0.6344339622641509,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.797101449275362e-05,
      "loss": 0.2461,
      "step": 269
    },
    {
      "epoch": 0.6367924528301887,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.795893719806763e-05,
      "loss": 0.2148,
      "step": 270
    },
    {
      "epoch": 0.6391509433962265,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.7946859903381646e-05,
      "loss": 0.2671,
      "step": 271
    },
    {
      "epoch": 0.6415094339622641,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.7934782608695655e-05,
      "loss": 0.2864,
      "step": 272
    },
    {
      "epoch": 0.6438679245283019,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.7922705314009664e-05,
      "loss": 0.258,
      "step": 273
    },
    {
      "epoch": 0.6462264150943396,
      "grad_norm": 0.1640625,
      "learning_rate": 4.791062801932367e-05,
      "loss": 0.2192,
      "step": 274
    },
    {
      "epoch": 0.6485849056603774,
      "grad_norm": 0.177734375,
      "learning_rate": 4.789855072463768e-05,
      "loss": 0.1993,
      "step": 275
    },
    {
      "epoch": 0.6509433962264151,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.78864734299517e-05,
      "loss": 0.2081,
      "step": 276
    },
    {
      "epoch": 0.6533018867924528,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.7874396135265706e-05,
      "loss": 0.2475,
      "step": 277
    },
    {
      "epoch": 0.6556603773584906,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.7862318840579715e-05,
      "loss": 0.2806,
      "step": 278
    },
    {
      "epoch": 0.6580188679245284,
      "grad_norm": 0.15625,
      "learning_rate": 4.785024154589372e-05,
      "loss": 0.189,
      "step": 279
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 0.171875,
      "learning_rate": 4.783816425120773e-05,
      "loss": 0.2643,
      "step": 280
    },
    {
      "epoch": 0.6627358490566038,
      "grad_norm": 0.171875,
      "learning_rate": 4.782608695652174e-05,
      "loss": 0.2901,
      "step": 281
    },
    {
      "epoch": 0.6650943396226415,
      "grad_norm": 0.1640625,
      "learning_rate": 4.781400966183575e-05,
      "loss": 0.2147,
      "step": 282
    },
    {
      "epoch": 0.6674528301886793,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.780193236714976e-05,
      "loss": 0.283,
      "step": 283
    },
    {
      "epoch": 0.6698113207547169,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.778985507246377e-05,
      "loss": 0.2162,
      "step": 284
    },
    {
      "epoch": 0.6721698113207547,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.7777777777777784e-05,
      "loss": 0.2791,
      "step": 285
    },
    {
      "epoch": 0.6745283018867925,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.776570048309179e-05,
      "loss": 0.2493,
      "step": 286
    },
    {
      "epoch": 0.6768867924528302,
      "grad_norm": 0.1953125,
      "learning_rate": 4.77536231884058e-05,
      "loss": 0.213,
      "step": 287
    },
    {
      "epoch": 0.6792452830188679,
      "grad_norm": 0.17578125,
      "learning_rate": 4.774154589371981e-05,
      "loss": 0.2256,
      "step": 288
    },
    {
      "epoch": 0.6816037735849056,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.772946859903382e-05,
      "loss": 0.2777,
      "step": 289
    },
    {
      "epoch": 0.6839622641509434,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.771739130434783e-05,
      "loss": 0.2146,
      "step": 290
    },
    {
      "epoch": 0.6863207547169812,
      "grad_norm": 0.185546875,
      "learning_rate": 4.770531400966184e-05,
      "loss": 0.2479,
      "step": 291
    },
    {
      "epoch": 0.6886792452830188,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.7693236714975847e-05,
      "loss": 0.1894,
      "step": 292
    },
    {
      "epoch": 0.6910377358490566,
      "grad_norm": 0.181640625,
      "learning_rate": 4.7681159420289855e-05,
      "loss": 0.2686,
      "step": 293
    },
    {
      "epoch": 0.6933962264150944,
      "grad_norm": 0.158203125,
      "learning_rate": 4.7669082125603864e-05,
      "loss": 0.2776,
      "step": 294
    },
    {
      "epoch": 0.6957547169811321,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.765700483091788e-05,
      "loss": 0.2537,
      "step": 295
    },
    {
      "epoch": 0.6981132075471698,
      "grad_norm": 0.220703125,
      "learning_rate": 4.764492753623189e-05,
      "loss": 0.2174,
      "step": 296
    },
    {
      "epoch": 0.7004716981132075,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.76328502415459e-05,
      "loss": 0.2367,
      "step": 297
    },
    {
      "epoch": 0.7028301886792453,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.762077294685991e-05,
      "loss": 0.2322,
      "step": 298
    },
    {
      "epoch": 0.7051886792452831,
      "grad_norm": 0.201171875,
      "learning_rate": 4.7608695652173916e-05,
      "loss": 0.2931,
      "step": 299
    },
    {
      "epoch": 0.7075471698113207,
      "grad_norm": 0.1796875,
      "learning_rate": 4.7596618357487925e-05,
      "loss": 0.2955,
      "step": 300
    },
    {
      "epoch": 0.7099056603773585,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.7584541062801933e-05,
      "loss": 0.2001,
      "step": 301
    },
    {
      "epoch": 0.7122641509433962,
      "grad_norm": 0.162109375,
      "learning_rate": 4.757246376811594e-05,
      "loss": 0.2024,
      "step": 302
    },
    {
      "epoch": 0.714622641509434,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.756038647342995e-05,
      "loss": 0.23,
      "step": 303
    },
    {
      "epoch": 0.7169811320754716,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.754830917874397e-05,
      "loss": 0.2711,
      "step": 304
    },
    {
      "epoch": 0.7193396226415094,
      "grad_norm": 0.21484375,
      "learning_rate": 4.7536231884057976e-05,
      "loss": 0.2395,
      "step": 305
    },
    {
      "epoch": 0.7216981132075472,
      "grad_norm": 0.177734375,
      "learning_rate": 4.7524154589371985e-05,
      "loss": 0.2339,
      "step": 306
    },
    {
      "epoch": 0.7240566037735849,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.7512077294685994e-05,
      "loss": 0.1925,
      "step": 307
    },
    {
      "epoch": 0.7264150943396226,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.75e-05,
      "loss": 0.2169,
      "step": 308
    },
    {
      "epoch": 0.7287735849056604,
      "grad_norm": 0.162109375,
      "learning_rate": 4.748792270531401e-05,
      "loss": 0.2367,
      "step": 309
    },
    {
      "epoch": 0.7311320754716981,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.747584541062802e-05,
      "loss": 0.2168,
      "step": 310
    },
    {
      "epoch": 0.7334905660377359,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.746376811594203e-05,
      "loss": 0.2884,
      "step": 311
    },
    {
      "epoch": 0.7358490566037735,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.745169082125604e-05,
      "loss": 0.2613,
      "step": 312
    },
    {
      "epoch": 0.7382075471698113,
      "grad_norm": 0.193359375,
      "learning_rate": 4.743961352657005e-05,
      "loss": 0.2507,
      "step": 313
    },
    {
      "epoch": 0.7405660377358491,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.742753623188406e-05,
      "loss": 0.2155,
      "step": 314
    },
    {
      "epoch": 0.7429245283018868,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.741545893719807e-05,
      "loss": 0.1892,
      "step": 315
    },
    {
      "epoch": 0.7452830188679245,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.740338164251208e-05,
      "loss": 0.2686,
      "step": 316
    },
    {
      "epoch": 0.7476415094339622,
      "grad_norm": 0.16796875,
      "learning_rate": 4.739130434782609e-05,
      "loss": 0.2424,
      "step": 317
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.73792270531401e-05,
      "loss": 0.2529,
      "step": 318
    },
    {
      "epoch": 0.7523584905660378,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.736714975845411e-05,
      "loss": 0.2634,
      "step": 319
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.7355072463768116e-05,
      "loss": 0.2029,
      "step": 320
    },
    {
      "epoch": 0.7570754716981132,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.7342995169082125e-05,
      "loss": 0.269,
      "step": 321
    },
    {
      "epoch": 0.7594339622641509,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.7330917874396134e-05,
      "loss": 0.2411,
      "step": 322
    },
    {
      "epoch": 0.7617924528301887,
      "grad_norm": 0.1796875,
      "learning_rate": 4.731884057971015e-05,
      "loss": 0.2295,
      "step": 323
    },
    {
      "epoch": 0.7641509433962265,
      "grad_norm": 0.169921875,
      "learning_rate": 4.730676328502416e-05,
      "loss": 0.1881,
      "step": 324
    },
    {
      "epoch": 0.7665094339622641,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.729468599033817e-05,
      "loss": 0.1975,
      "step": 325
    },
    {
      "epoch": 0.7688679245283019,
      "grad_norm": 0.177734375,
      "learning_rate": 4.7282608695652177e-05,
      "loss": 0.267,
      "step": 326
    },
    {
      "epoch": 0.7712264150943396,
      "grad_norm": 0.181640625,
      "learning_rate": 4.7270531400966185e-05,
      "loss": 0.2515,
      "step": 327
    },
    {
      "epoch": 0.7735849056603774,
      "grad_norm": 0.166015625,
      "learning_rate": 4.72584541062802e-05,
      "loss": 0.3147,
      "step": 328
    },
    {
      "epoch": 0.7759433962264151,
      "grad_norm": 0.16015625,
      "learning_rate": 4.72463768115942e-05,
      "loss": 0.1665,
      "step": 329
    },
    {
      "epoch": 0.7783018867924528,
      "grad_norm": 0.17578125,
      "learning_rate": 4.723429951690821e-05,
      "loss": 0.2565,
      "step": 330
    },
    {
      "epoch": 0.7806603773584906,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.2862,
      "step": 331
    },
    {
      "epoch": 0.7830188679245284,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.721014492753623e-05,
      "loss": 0.2767,
      "step": 332
    },
    {
      "epoch": 0.785377358490566,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.7198067632850246e-05,
      "loss": 0.2646,
      "step": 333
    },
    {
      "epoch": 0.7877358490566038,
      "grad_norm": 0.181640625,
      "learning_rate": 4.7185990338164255e-05,
      "loss": 0.2759,
      "step": 334
    },
    {
      "epoch": 0.7900943396226415,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.7173913043478264e-05,
      "loss": 0.2441,
      "step": 335
    },
    {
      "epoch": 0.7924528301886793,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.716183574879227e-05,
      "loss": 0.1741,
      "step": 336
    },
    {
      "epoch": 0.7948113207547169,
      "grad_norm": 0.181640625,
      "learning_rate": 4.714975845410628e-05,
      "loss": 0.209,
      "step": 337
    },
    {
      "epoch": 0.7971698113207547,
      "grad_norm": 0.189453125,
      "learning_rate": 4.71376811594203e-05,
      "loss": 0.2408,
      "step": 338
    },
    {
      "epoch": 0.7995283018867925,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.71256038647343e-05,
      "loss": 0.2426,
      "step": 339
    },
    {
      "epoch": 0.8018867924528302,
      "grad_norm": 0.181640625,
      "learning_rate": 4.711352657004831e-05,
      "loss": 0.2649,
      "step": 340
    },
    {
      "epoch": 0.8042452830188679,
      "grad_norm": 0.158203125,
      "learning_rate": 4.710144927536232e-05,
      "loss": 0.2854,
      "step": 341
    },
    {
      "epoch": 0.8066037735849056,
      "grad_norm": 0.18359375,
      "learning_rate": 4.708937198067633e-05,
      "loss": 0.2435,
      "step": 342
    },
    {
      "epoch": 0.8089622641509434,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.707729468599034e-05,
      "loss": 0.2589,
      "step": 343
    },
    {
      "epoch": 0.8113207547169812,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.706521739130435e-05,
      "loss": 0.1894,
      "step": 344
    },
    {
      "epoch": 0.8136792452830188,
      "grad_norm": 0.169921875,
      "learning_rate": 4.705314009661836e-05,
      "loss": 0.214,
      "step": 345
    },
    {
      "epoch": 0.8160377358490566,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.704106280193237e-05,
      "loss": 0.2477,
      "step": 346
    },
    {
      "epoch": 0.8183962264150944,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.7028985507246384e-05,
      "loss": 0.2263,
      "step": 347
    },
    {
      "epoch": 0.8207547169811321,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.701690821256039e-05,
      "loss": 0.2686,
      "step": 348
    },
    {
      "epoch": 0.8231132075471698,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.7004830917874395e-05,
      "loss": 0.2559,
      "step": 349
    },
    {
      "epoch": 0.8254716981132075,
      "grad_norm": 0.203125,
      "learning_rate": 4.6992753623188404e-05,
      "loss": 0.2542,
      "step": 350
    },
    {
      "epoch": 0.8278301886792453,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.698067632850241e-05,
      "loss": 0.2565,
      "step": 351
    },
    {
      "epoch": 0.8301886792452831,
      "grad_norm": 0.1640625,
      "learning_rate": 4.696859903381643e-05,
      "loss": 0.263,
      "step": 352
    },
    {
      "epoch": 0.8325471698113207,
      "grad_norm": 0.17578125,
      "learning_rate": 4.695652173913044e-05,
      "loss": 0.2339,
      "step": 353
    },
    {
      "epoch": 0.8349056603773585,
      "grad_norm": 0.181640625,
      "learning_rate": 4.6944444444444446e-05,
      "loss": 0.2714,
      "step": 354
    },
    {
      "epoch": 0.8372641509433962,
      "grad_norm": 0.166015625,
      "learning_rate": 4.6932367149758455e-05,
      "loss": 0.246,
      "step": 355
    },
    {
      "epoch": 0.839622641509434,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.6920289855072464e-05,
      "loss": 0.2264,
      "step": 356
    },
    {
      "epoch": 0.8419811320754716,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.690821256038648e-05,
      "loss": 0.2231,
      "step": 357
    },
    {
      "epoch": 0.8443396226415094,
      "grad_norm": 0.189453125,
      "learning_rate": 4.689613526570049e-05,
      "loss": 0.2739,
      "step": 358
    },
    {
      "epoch": 0.8466981132075472,
      "grad_norm": 0.177734375,
      "learning_rate": 4.68840579710145e-05,
      "loss": 0.2105,
      "step": 359
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.68719806763285e-05,
      "loss": 0.2152,
      "step": 360
    },
    {
      "epoch": 0.8514150943396226,
      "grad_norm": 0.205078125,
      "learning_rate": 4.6859903381642516e-05,
      "loss": 0.2439,
      "step": 361
    },
    {
      "epoch": 0.8537735849056604,
      "grad_norm": 0.16796875,
      "learning_rate": 4.6847826086956524e-05,
      "loss": 0.2251,
      "step": 362
    },
    {
      "epoch": 0.8561320754716981,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.683574879227053e-05,
      "loss": 0.1995,
      "step": 363
    },
    {
      "epoch": 0.8584905660377359,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.682367149758454e-05,
      "loss": 0.2365,
      "step": 364
    },
    {
      "epoch": 0.8608490566037735,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.681159420289855e-05,
      "loss": 0.2455,
      "step": 365
    },
    {
      "epoch": 0.8632075471698113,
      "grad_norm": 0.1640625,
      "learning_rate": 4.679951690821257e-05,
      "loss": 0.2196,
      "step": 366
    },
    {
      "epoch": 0.8655660377358491,
      "grad_norm": 0.1640625,
      "learning_rate": 4.6787439613526576e-05,
      "loss": 0.2039,
      "step": 367
    },
    {
      "epoch": 0.8679245283018868,
      "grad_norm": 0.1484375,
      "learning_rate": 4.6775362318840585e-05,
      "loss": 0.2099,
      "step": 368
    },
    {
      "epoch": 0.8702830188679245,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.6763285024154594e-05,
      "loss": 0.2659,
      "step": 369
    },
    {
      "epoch": 0.8726415094339622,
      "grad_norm": 0.19140625,
      "learning_rate": 4.6751207729468596e-05,
      "loss": 0.2325,
      "step": 370
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.185546875,
      "learning_rate": 4.673913043478261e-05,
      "loss": 0.2674,
      "step": 371
    },
    {
      "epoch": 0.8773584905660378,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.672705314009662e-05,
      "loss": 0.2474,
      "step": 372
    },
    {
      "epoch": 0.8797169811320755,
      "grad_norm": 0.17578125,
      "learning_rate": 4.671497584541063e-05,
      "loss": 0.2295,
      "step": 373
    },
    {
      "epoch": 0.8820754716981132,
      "grad_norm": 0.162109375,
      "learning_rate": 4.670289855072464e-05,
      "loss": 0.1892,
      "step": 374
    },
    {
      "epoch": 0.8844339622641509,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.669082125603865e-05,
      "loss": 0.1869,
      "step": 375
    },
    {
      "epoch": 0.8867924528301887,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.667874396135266e-05,
      "loss": 0.2925,
      "step": 376
    },
    {
      "epoch": 0.8891509433962265,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.2628,
      "step": 377
    },
    {
      "epoch": 0.8915094339622641,
      "grad_norm": 0.158203125,
      "learning_rate": 4.665458937198068e-05,
      "loss": 0.2342,
      "step": 378
    },
    {
      "epoch": 0.8938679245283019,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.664251207729469e-05,
      "loss": 0.2401,
      "step": 379
    },
    {
      "epoch": 0.8962264150943396,
      "grad_norm": 0.1640625,
      "learning_rate": 4.66304347826087e-05,
      "loss": 0.2705,
      "step": 380
    },
    {
      "epoch": 0.8985849056603774,
      "grad_norm": 0.18359375,
      "learning_rate": 4.661835748792271e-05,
      "loss": 0.1537,
      "step": 381
    },
    {
      "epoch": 0.9009433962264151,
      "grad_norm": 0.189453125,
      "learning_rate": 4.6606280193236716e-05,
      "loss": 0.233,
      "step": 382
    },
    {
      "epoch": 0.9033018867924528,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.6594202898550725e-05,
      "loss": 0.223,
      "step": 383
    },
    {
      "epoch": 0.9056603773584906,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.6582125603864734e-05,
      "loss": 0.2686,
      "step": 384
    },
    {
      "epoch": 0.9080188679245284,
      "grad_norm": 0.18359375,
      "learning_rate": 4.657004830917875e-05,
      "loss": 0.218,
      "step": 385
    },
    {
      "epoch": 0.910377358490566,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.655797101449276e-05,
      "loss": 0.2451,
      "step": 386
    },
    {
      "epoch": 0.9127358490566038,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.654589371980677e-05,
      "loss": 0.2005,
      "step": 387
    },
    {
      "epoch": 0.9150943396226415,
      "grad_norm": 0.1640625,
      "learning_rate": 4.6533816425120776e-05,
      "loss": 0.2196,
      "step": 388
    },
    {
      "epoch": 0.9174528301886793,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.6521739130434785e-05,
      "loss": 0.2112,
      "step": 389
    },
    {
      "epoch": 0.9198113207547169,
      "grad_norm": 0.171875,
      "learning_rate": 4.6509661835748794e-05,
      "loss": 0.2125,
      "step": 390
    },
    {
      "epoch": 0.9221698113207547,
      "grad_norm": 0.171875,
      "learning_rate": 4.64975845410628e-05,
      "loss": 0.2552,
      "step": 391
    },
    {
      "epoch": 0.9245283018867925,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.648550724637681e-05,
      "loss": 0.2555,
      "step": 392
    },
    {
      "epoch": 0.9268867924528302,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.647342995169082e-05,
      "loss": 0.2233,
      "step": 393
    },
    {
      "epoch": 0.9292452830188679,
      "grad_norm": 0.171875,
      "learning_rate": 4.646135265700483e-05,
      "loss": 0.2453,
      "step": 394
    },
    {
      "epoch": 0.9316037735849056,
      "grad_norm": 0.1640625,
      "learning_rate": 4.6449275362318846e-05,
      "loss": 0.2835,
      "step": 395
    },
    {
      "epoch": 0.9339622641509434,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.6437198067632854e-05,
      "loss": 0.1948,
      "step": 396
    },
    {
      "epoch": 0.9363207547169812,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.642512077294686e-05,
      "loss": 0.2161,
      "step": 397
    },
    {
      "epoch": 0.9386792452830188,
      "grad_norm": 0.16015625,
      "learning_rate": 4.641304347826087e-05,
      "loss": 0.2034,
      "step": 398
    },
    {
      "epoch": 0.9410377358490566,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.640096618357488e-05,
      "loss": 0.2251,
      "step": 399
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 0.16796875,
      "learning_rate": 4.638888888888889e-05,
      "loss": 0.2275,
      "step": 400
    },
    {
      "epoch": 0.9457547169811321,
      "grad_norm": 0.2099609375,
      "learning_rate": 4.63768115942029e-05,
      "loss": 0.2455,
      "step": 401
    },
    {
      "epoch": 0.9481132075471698,
      "grad_norm": 0.173828125,
      "learning_rate": 4.636473429951691e-05,
      "loss": 0.233,
      "step": 402
    },
    {
      "epoch": 0.9504716981132075,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.635265700483092e-05,
      "loss": 0.2564,
      "step": 403
    },
    {
      "epoch": 0.9528301886792453,
      "grad_norm": 0.166015625,
      "learning_rate": 4.634057971014493e-05,
      "loss": 0.1985,
      "step": 404
    },
    {
      "epoch": 0.9551886792452831,
      "grad_norm": 0.1796875,
      "learning_rate": 4.632850241545894e-05,
      "loss": 0.1993,
      "step": 405
    },
    {
      "epoch": 0.9575471698113207,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.631642512077295e-05,
      "loss": 0.2276,
      "step": 406
    },
    {
      "epoch": 0.9599056603773585,
      "grad_norm": 0.1953125,
      "learning_rate": 4.630434782608696e-05,
      "loss": 0.2415,
      "step": 407
    },
    {
      "epoch": 0.9622641509433962,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.629227053140097e-05,
      "loss": 0.1818,
      "step": 408
    },
    {
      "epoch": 0.964622641509434,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.628019323671498e-05,
      "loss": 0.266,
      "step": 409
    },
    {
      "epoch": 0.9669811320754716,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.6268115942028986e-05,
      "loss": 0.2183,
      "step": 410
    },
    {
      "epoch": 0.9693396226415094,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.6256038647342995e-05,
      "loss": 0.2608,
      "step": 411
    },
    {
      "epoch": 0.9716981132075472,
      "grad_norm": 0.16796875,
      "learning_rate": 4.6243961352657004e-05,
      "loss": 0.2197,
      "step": 412
    },
    {
      "epoch": 0.9740566037735849,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.623188405797101e-05,
      "loss": 0.2073,
      "step": 413
    },
    {
      "epoch": 0.9764150943396226,
      "grad_norm": 0.171875,
      "learning_rate": 4.621980676328503e-05,
      "loss": 0.2685,
      "step": 414
    },
    {
      "epoch": 0.9787735849056604,
      "grad_norm": 0.162109375,
      "learning_rate": 4.620772946859904e-05,
      "loss": 0.274,
      "step": 415
    },
    {
      "epoch": 0.9811320754716981,
      "grad_norm": 0.169921875,
      "learning_rate": 4.6195652173913046e-05,
      "loss": 0.1816,
      "step": 416
    },
    {
      "epoch": 0.9834905660377359,
      "grad_norm": 0.169921875,
      "learning_rate": 4.6183574879227055e-05,
      "loss": 0.2717,
      "step": 417
    },
    {
      "epoch": 0.9858490566037735,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.6171497584541064e-05,
      "loss": 0.2059,
      "step": 418
    },
    {
      "epoch": 0.9882075471698113,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.615942028985508e-05,
      "loss": 0.2238,
      "step": 419
    },
    {
      "epoch": 0.9905660377358491,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.614734299516908e-05,
      "loss": 0.1817,
      "step": 420
    },
    {
      "epoch": 0.9929245283018868,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.613526570048309e-05,
      "loss": 0.2255,
      "step": 421
    },
    {
      "epoch": 0.9952830188679245,
      "grad_norm": 0.171875,
      "learning_rate": 4.61231884057971e-05,
      "loss": 0.2199,
      "step": 422
    },
    {
      "epoch": 0.9976415094339622,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.6111111111111115e-05,
      "loss": 0.2055,
      "step": 423
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.369140625,
      "learning_rate": 4.6099033816425124e-05,
      "loss": 0.1884,
      "step": 424
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.20209836959838867,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 1.9322,
      "eval_samples_per_second": 1.553,
      "eval_steps_per_second": 1.553,
      "step": 424
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.14613842964172363,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 1.3863,
      "eval_samples_per_second": 2.885,
      "eval_steps_per_second": 2.885,
      "step": 424
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.2947477698326111,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 3.0807,
      "eval_samples_per_second": 1.298,
      "eval_steps_per_second": 1.298,
      "step": 424
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.3528926372528076,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 2.902,
      "eval_samples_per_second": 1.378,
      "eval_steps_per_second": 1.378,
      "step": 424
    },
    {
      "epoch": 1.0023584905660377,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.608695652173913e-05,
      "loss": 0.2101,
      "step": 425
    },
    {
      "epoch": 1.0047169811320755,
      "grad_norm": 0.1953125,
      "learning_rate": 4.607487922705314e-05,
      "loss": 0.2346,
      "step": 426
    },
    {
      "epoch": 1.0070754716981132,
      "grad_norm": 0.16796875,
      "learning_rate": 4.606280193236715e-05,
      "loss": 0.2147,
      "step": 427
    },
    {
      "epoch": 1.009433962264151,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.605072463768117e-05,
      "loss": 0.1795,
      "step": 428
    },
    {
      "epoch": 1.0117924528301887,
      "grad_norm": 0.1640625,
      "learning_rate": 4.6038647342995176e-05,
      "loss": 0.2032,
      "step": 429
    },
    {
      "epoch": 1.0141509433962264,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.602657004830918e-05,
      "loss": 0.2545,
      "step": 430
    },
    {
      "epoch": 1.0165094339622642,
      "grad_norm": 0.169921875,
      "learning_rate": 4.601449275362319e-05,
      "loss": 0.2582,
      "step": 431
    },
    {
      "epoch": 1.0188679245283019,
      "grad_norm": 0.1796875,
      "learning_rate": 4.6002415458937196e-05,
      "loss": 0.2086,
      "step": 432
    },
    {
      "epoch": 1.0212264150943395,
      "grad_norm": 0.177734375,
      "learning_rate": 4.599033816425121e-05,
      "loss": 0.1966,
      "step": 433
    },
    {
      "epoch": 1.0235849056603774,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.597826086956522e-05,
      "loss": 0.2325,
      "step": 434
    },
    {
      "epoch": 1.025943396226415,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.596618357487923e-05,
      "loss": 0.216,
      "step": 435
    },
    {
      "epoch": 1.028301886792453,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.595410628019324e-05,
      "loss": 0.2214,
      "step": 436
    },
    {
      "epoch": 1.0306603773584906,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.594202898550725e-05,
      "loss": 0.1825,
      "step": 437
    },
    {
      "epoch": 1.0330188679245282,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.592995169082126e-05,
      "loss": 0.258,
      "step": 438
    },
    {
      "epoch": 1.0353773584905661,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.591787439613527e-05,
      "loss": 0.2459,
      "step": 439
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 0.1796875,
      "learning_rate": 4.590579710144928e-05,
      "loss": 0.1915,
      "step": 440
    },
    {
      "epoch": 1.0400943396226414,
      "grad_norm": 0.189453125,
      "learning_rate": 4.589371980676328e-05,
      "loss": 0.2758,
      "step": 441
    },
    {
      "epoch": 1.0424528301886793,
      "grad_norm": 0.173828125,
      "learning_rate": 4.588164251207729e-05,
      "loss": 0.1211,
      "step": 442
    },
    {
      "epoch": 1.044811320754717,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.586956521739131e-05,
      "loss": 0.2225,
      "step": 443
    },
    {
      "epoch": 1.0471698113207548,
      "grad_norm": 0.181640625,
      "learning_rate": 4.5857487922705316e-05,
      "loss": 0.2099,
      "step": 444
    },
    {
      "epoch": 1.0495283018867925,
      "grad_norm": 0.177734375,
      "learning_rate": 4.5845410628019325e-05,
      "loss": 0.2346,
      "step": 445
    },
    {
      "epoch": 1.0518867924528301,
      "grad_norm": 0.166015625,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.216,
      "step": 446
    },
    {
      "epoch": 1.054245283018868,
      "grad_norm": 0.177734375,
      "learning_rate": 4.582125603864735e-05,
      "loss": 0.1919,
      "step": 447
    },
    {
      "epoch": 1.0566037735849056,
      "grad_norm": 0.205078125,
      "learning_rate": 4.580917874396136e-05,
      "loss": 0.2289,
      "step": 448
    },
    {
      "epoch": 1.0589622641509433,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.579710144927537e-05,
      "loss": 0.244,
      "step": 449
    },
    {
      "epoch": 1.0613207547169812,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.5785024154589376e-05,
      "loss": 0.2763,
      "step": 450
    },
    {
      "epoch": 1.0636792452830188,
      "grad_norm": 0.19140625,
      "learning_rate": 4.577294685990338e-05,
      "loss": 0.2623,
      "step": 451
    },
    {
      "epoch": 1.0660377358490567,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.5760869565217394e-05,
      "loss": 0.2189,
      "step": 452
    },
    {
      "epoch": 1.0683962264150944,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.57487922705314e-05,
      "loss": 0.2407,
      "step": 453
    },
    {
      "epoch": 1.070754716981132,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.573671497584541e-05,
      "loss": 0.2225,
      "step": 454
    },
    {
      "epoch": 1.0731132075471699,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.572463768115942e-05,
      "loss": 0.1959,
      "step": 455
    },
    {
      "epoch": 1.0754716981132075,
      "grad_norm": 0.169921875,
      "learning_rate": 4.571256038647343e-05,
      "loss": 0.2003,
      "step": 456
    },
    {
      "epoch": 1.0778301886792452,
      "grad_norm": 0.169921875,
      "learning_rate": 4.5700483091787445e-05,
      "loss": 0.2682,
      "step": 457
    },
    {
      "epoch": 1.080188679245283,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.5688405797101454e-05,
      "loss": 0.2264,
      "step": 458
    },
    {
      "epoch": 1.0825471698113207,
      "grad_norm": 0.177734375,
      "learning_rate": 4.567632850241546e-05,
      "loss": 0.2599,
      "step": 459
    },
    {
      "epoch": 1.0849056603773586,
      "grad_norm": 0.17578125,
      "learning_rate": 4.566425120772947e-05,
      "loss": 0.2701,
      "step": 460
    },
    {
      "epoch": 1.0872641509433962,
      "grad_norm": 0.169921875,
      "learning_rate": 4.565217391304348e-05,
      "loss": 0.1767,
      "step": 461
    },
    {
      "epoch": 1.0896226415094339,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.564009661835749e-05,
      "loss": 0.1988,
      "step": 462
    },
    {
      "epoch": 1.0919811320754718,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.56280193236715e-05,
      "loss": 0.1861,
      "step": 463
    },
    {
      "epoch": 1.0943396226415094,
      "grad_norm": 0.162109375,
      "learning_rate": 4.561594202898551e-05,
      "loss": 0.2402,
      "step": 464
    },
    {
      "epoch": 1.0966981132075473,
      "grad_norm": 0.1875,
      "learning_rate": 4.560386473429952e-05,
      "loss": 0.2588,
      "step": 465
    },
    {
      "epoch": 1.099056603773585,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.559178743961353e-05,
      "loss": 0.2163,
      "step": 466
    },
    {
      "epoch": 1.1014150943396226,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.557971014492754e-05,
      "loss": 0.2164,
      "step": 467
    },
    {
      "epoch": 1.1037735849056605,
      "grad_norm": 0.2373046875,
      "learning_rate": 4.556763285024155e-05,
      "loss": 0.1767,
      "step": 468
    },
    {
      "epoch": 1.1061320754716981,
      "grad_norm": 0.18359375,
      "learning_rate": 4.555555555555556e-05,
      "loss": 0.2001,
      "step": 469
    },
    {
      "epoch": 1.1084905660377358,
      "grad_norm": 0.185546875,
      "learning_rate": 4.554347826086957e-05,
      "loss": 0.2175,
      "step": 470
    },
    {
      "epoch": 1.1108490566037736,
      "grad_norm": 0.16796875,
      "learning_rate": 4.553140096618358e-05,
      "loss": 0.1817,
      "step": 471
    },
    {
      "epoch": 1.1132075471698113,
      "grad_norm": 0.173828125,
      "learning_rate": 4.5519323671497586e-05,
      "loss": 0.2323,
      "step": 472
    },
    {
      "epoch": 1.115566037735849,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.5507246376811595e-05,
      "loss": 0.1766,
      "step": 473
    },
    {
      "epoch": 1.1179245283018868,
      "grad_norm": 0.171875,
      "learning_rate": 4.5495169082125604e-05,
      "loss": 0.1928,
      "step": 474
    },
    {
      "epoch": 1.1202830188679245,
      "grad_norm": 0.169921875,
      "learning_rate": 4.548309178743961e-05,
      "loss": 0.211,
      "step": 475
    },
    {
      "epoch": 1.1226415094339623,
      "grad_norm": 0.15625,
      "learning_rate": 4.547101449275363e-05,
      "loss": 0.2076,
      "step": 476
    },
    {
      "epoch": 1.125,
      "grad_norm": 0.17578125,
      "learning_rate": 4.545893719806764e-05,
      "loss": 0.2427,
      "step": 477
    },
    {
      "epoch": 1.1273584905660377,
      "grad_norm": 0.173828125,
      "learning_rate": 4.5446859903381646e-05,
      "loss": 0.2044,
      "step": 478
    },
    {
      "epoch": 1.1297169811320755,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.5434782608695655e-05,
      "loss": 0.2126,
      "step": 479
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.5422705314009664e-05,
      "loss": 0.2192,
      "step": 480
    },
    {
      "epoch": 1.134433962264151,
      "grad_norm": 0.173828125,
      "learning_rate": 4.541062801932367e-05,
      "loss": 0.2358,
      "step": 481
    },
    {
      "epoch": 1.1367924528301887,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.539855072463768e-05,
      "loss": 0.2581,
      "step": 482
    },
    {
      "epoch": 1.1391509433962264,
      "grad_norm": 0.1796875,
      "learning_rate": 4.538647342995169e-05,
      "loss": 0.1859,
      "step": 483
    },
    {
      "epoch": 1.1415094339622642,
      "grad_norm": 0.171875,
      "learning_rate": 4.53743961352657e-05,
      "loss": 0.2458,
      "step": 484
    },
    {
      "epoch": 1.1438679245283019,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.5362318840579715e-05,
      "loss": 0.2227,
      "step": 485
    },
    {
      "epoch": 1.1462264150943395,
      "grad_norm": 0.169921875,
      "learning_rate": 4.5350241545893724e-05,
      "loss": 0.195,
      "step": 486
    },
    {
      "epoch": 1.1485849056603774,
      "grad_norm": 0.201171875,
      "learning_rate": 4.533816425120773e-05,
      "loss": 0.2042,
      "step": 487
    },
    {
      "epoch": 1.150943396226415,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.532608695652174e-05,
      "loss": 0.1949,
      "step": 488
    },
    {
      "epoch": 1.1533018867924527,
      "grad_norm": 0.17578125,
      "learning_rate": 4.531400966183575e-05,
      "loss": 0.2532,
      "step": 489
    },
    {
      "epoch": 1.1556603773584906,
      "grad_norm": 0.171875,
      "learning_rate": 4.530193236714976e-05,
      "loss": 0.1716,
      "step": 490
    },
    {
      "epoch": 1.1580188679245282,
      "grad_norm": 0.171875,
      "learning_rate": 4.528985507246377e-05,
      "loss": 0.2335,
      "step": 491
    },
    {
      "epoch": 1.1603773584905661,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.527777777777778e-05,
      "loss": 0.2686,
      "step": 492
    },
    {
      "epoch": 1.1627358490566038,
      "grad_norm": 0.18359375,
      "learning_rate": 4.5265700483091786e-05,
      "loss": 0.2406,
      "step": 493
    },
    {
      "epoch": 1.1650943396226414,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.5253623188405795e-05,
      "loss": 0.2407,
      "step": 494
    },
    {
      "epoch": 1.1674528301886793,
      "grad_norm": 0.169921875,
      "learning_rate": 4.524154589371981e-05,
      "loss": 0.2021,
      "step": 495
    },
    {
      "epoch": 1.169811320754717,
      "grad_norm": 0.171875,
      "learning_rate": 4.522946859903382e-05,
      "loss": 0.1804,
      "step": 496
    },
    {
      "epoch": 1.1721698113207548,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.521739130434783e-05,
      "loss": 0.1606,
      "step": 497
    },
    {
      "epoch": 1.1745283018867925,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.520531400966184e-05,
      "loss": 0.1924,
      "step": 498
    },
    {
      "epoch": 1.1768867924528301,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.519323671497585e-05,
      "loss": 0.2865,
      "step": 499
    },
    {
      "epoch": 1.179245283018868,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.518115942028986e-05,
      "loss": 0.2153,
      "step": 500
    },
    {
      "epoch": 1.1816037735849056,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.5169082125603865e-05,
      "loss": 0.2222,
      "step": 501
    },
    {
      "epoch": 1.1839622641509433,
      "grad_norm": 0.1640625,
      "learning_rate": 4.5157004830917873e-05,
      "loss": 0.194,
      "step": 502
    },
    {
      "epoch": 1.1863207547169812,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.514492753623188e-05,
      "loss": 0.2364,
      "step": 503
    },
    {
      "epoch": 1.1886792452830188,
      "grad_norm": 0.18359375,
      "learning_rate": 4.513285024154589e-05,
      "loss": 0.1862,
      "step": 504
    },
    {
      "epoch": 1.1910377358490567,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.512077294685991e-05,
      "loss": 0.1753,
      "step": 505
    },
    {
      "epoch": 1.1933962264150944,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.5108695652173916e-05,
      "loss": 0.1964,
      "step": 506
    },
    {
      "epoch": 1.195754716981132,
      "grad_norm": 0.185546875,
      "learning_rate": 4.5096618357487925e-05,
      "loss": 0.1917,
      "step": 507
    },
    {
      "epoch": 1.1981132075471699,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.5084541062801934e-05,
      "loss": 0.206,
      "step": 508
    },
    {
      "epoch": 1.2004716981132075,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.507246376811595e-05,
      "loss": 0.2206,
      "step": 509
    },
    {
      "epoch": 1.2028301886792452,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.506038647342996e-05,
      "loss": 0.1989,
      "step": 510
    },
    {
      "epoch": 1.205188679245283,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.504830917874396e-05,
      "loss": 0.1985,
      "step": 511
    },
    {
      "epoch": 1.2075471698113207,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.503623188405797e-05,
      "loss": 0.2389,
      "step": 512
    },
    {
      "epoch": 1.2099056603773586,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.502415458937198e-05,
      "loss": 0.1828,
      "step": 513
    },
    {
      "epoch": 1.2122641509433962,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.5012077294685994e-05,
      "loss": 0.2247,
      "step": 514
    },
    {
      "epoch": 1.2146226415094339,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.5e-05,
      "loss": 0.2458,
      "step": 515
    },
    {
      "epoch": 1.2169811320754718,
      "grad_norm": 0.1875,
      "learning_rate": 4.498792270531401e-05,
      "loss": 0.2359,
      "step": 516
    },
    {
      "epoch": 1.2193396226415094,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.497584541062802e-05,
      "loss": 0.2013,
      "step": 517
    },
    {
      "epoch": 1.2216981132075473,
      "grad_norm": 0.173828125,
      "learning_rate": 4.496376811594203e-05,
      "loss": 0.2619,
      "step": 518
    },
    {
      "epoch": 1.224056603773585,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.4951690821256045e-05,
      "loss": 0.2226,
      "step": 519
    },
    {
      "epoch": 1.2264150943396226,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.4939613526570054e-05,
      "loss": 0.2212,
      "step": 520
    },
    {
      "epoch": 1.2287735849056605,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.492753623188406e-05,
      "loss": 0.1572,
      "step": 521
    },
    {
      "epoch": 1.2311320754716981,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.4915458937198065e-05,
      "loss": 0.1803,
      "step": 522
    },
    {
      "epoch": 1.2334905660377358,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.4903381642512074e-05,
      "loss": 0.2811,
      "step": 523
    },
    {
      "epoch": 1.2358490566037736,
      "grad_norm": 0.271484375,
      "learning_rate": 4.489130434782609e-05,
      "loss": 0.2537,
      "step": 524
    },
    {
      "epoch": 1.2382075471698113,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.48792270531401e-05,
      "loss": 0.2208,
      "step": 525
    },
    {
      "epoch": 1.240566037735849,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.486714975845411e-05,
      "loss": 0.2783,
      "step": 526
    },
    {
      "epoch": 1.2429245283018868,
      "grad_norm": 0.173828125,
      "learning_rate": 4.4855072463768117e-05,
      "loss": 0.1809,
      "step": 527
    },
    {
      "epoch": 1.2452830188679245,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.484299516908213e-05,
      "loss": 0.2384,
      "step": 528
    },
    {
      "epoch": 1.2476415094339623,
      "grad_norm": 0.1640625,
      "learning_rate": 4.483091787439614e-05,
      "loss": 0.2009,
      "step": 529
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.181640625,
      "learning_rate": 4.481884057971015e-05,
      "loss": 0.2203,
      "step": 530
    },
    {
      "epoch": 1.2523584905660377,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.480676328502416e-05,
      "loss": 0.2287,
      "step": 531
    },
    {
      "epoch": 1.2547169811320755,
      "grad_norm": 0.18359375,
      "learning_rate": 4.479468599033816e-05,
      "loss": 0.2094,
      "step": 532
    },
    {
      "epoch": 1.2570754716981132,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.478260869565218e-05,
      "loss": 0.275,
      "step": 533
    },
    {
      "epoch": 1.259433962264151,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.4770531400966186e-05,
      "loss": 0.2273,
      "step": 534
    },
    {
      "epoch": 1.2617924528301887,
      "grad_norm": 0.169921875,
      "learning_rate": 4.4758454106280195e-05,
      "loss": 0.1819,
      "step": 535
    },
    {
      "epoch": 1.2641509433962264,
      "grad_norm": 0.17578125,
      "learning_rate": 4.4746376811594203e-05,
      "loss": 0.2954,
      "step": 536
    },
    {
      "epoch": 1.2665094339622642,
      "grad_norm": 0.173828125,
      "learning_rate": 4.473429951690821e-05,
      "loss": 0.2257,
      "step": 537
    },
    {
      "epoch": 1.2688679245283019,
      "grad_norm": 0.177734375,
      "learning_rate": 4.472222222222223e-05,
      "loss": 0.2202,
      "step": 538
    },
    {
      "epoch": 1.2712264150943398,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.471014492753624e-05,
      "loss": 0.1869,
      "step": 539
    },
    {
      "epoch": 1.2735849056603774,
      "grad_norm": 0.177734375,
      "learning_rate": 4.4698067632850246e-05,
      "loss": 0.2246,
      "step": 540
    },
    {
      "epoch": 1.275943396226415,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.4685990338164255e-05,
      "loss": 0.2148,
      "step": 541
    },
    {
      "epoch": 1.2783018867924527,
      "grad_norm": 0.185546875,
      "learning_rate": 4.4673913043478264e-05,
      "loss": 0.2224,
      "step": 542
    },
    {
      "epoch": 1.2806603773584906,
      "grad_norm": 0.1796875,
      "learning_rate": 4.466183574879227e-05,
      "loss": 0.1854,
      "step": 543
    },
    {
      "epoch": 1.2830188679245282,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.464975845410628e-05,
      "loss": 0.1901,
      "step": 544
    },
    {
      "epoch": 1.2853773584905661,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.463768115942029e-05,
      "loss": 0.1689,
      "step": 545
    },
    {
      "epoch": 1.2877358490566038,
      "grad_norm": 0.1953125,
      "learning_rate": 4.46256038647343e-05,
      "loss": 0.2246,
      "step": 546
    },
    {
      "epoch": 1.2900943396226414,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.4613526570048315e-05,
      "loss": 0.1902,
      "step": 547
    },
    {
      "epoch": 1.2924528301886793,
      "grad_norm": 0.16796875,
      "learning_rate": 4.4601449275362324e-05,
      "loss": 0.235,
      "step": 548
    },
    {
      "epoch": 1.294811320754717,
      "grad_norm": 0.16796875,
      "learning_rate": 4.458937198067633e-05,
      "loss": 0.1851,
      "step": 549
    },
    {
      "epoch": 1.2971698113207548,
      "grad_norm": 0.17578125,
      "learning_rate": 4.457729468599034e-05,
      "loss": 0.2243,
      "step": 550
    },
    {
      "epoch": 1.2995283018867925,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.456521739130435e-05,
      "loss": 0.2029,
      "step": 551
    },
    {
      "epoch": 1.3018867924528301,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.455314009661836e-05,
      "loss": 0.2436,
      "step": 552
    },
    {
      "epoch": 1.304245283018868,
      "grad_norm": 0.18359375,
      "learning_rate": 4.454106280193237e-05,
      "loss": 0.2127,
      "step": 553
    },
    {
      "epoch": 1.3066037735849056,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.452898550724638e-05,
      "loss": 0.2274,
      "step": 554
    },
    {
      "epoch": 1.3089622641509435,
      "grad_norm": 0.173828125,
      "learning_rate": 4.4516908212560386e-05,
      "loss": 0.1664,
      "step": 555
    },
    {
      "epoch": 1.3113207547169812,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.4504830917874395e-05,
      "loss": 0.1755,
      "step": 556
    },
    {
      "epoch": 1.3136792452830188,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.449275362318841e-05,
      "loss": 0.2096,
      "step": 557
    },
    {
      "epoch": 1.3160377358490565,
      "grad_norm": 0.171875,
      "learning_rate": 4.448067632850242e-05,
      "loss": 0.2465,
      "step": 558
    },
    {
      "epoch": 1.3183962264150944,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.446859903381643e-05,
      "loss": 0.1882,
      "step": 559
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 0.171875,
      "learning_rate": 4.445652173913044e-05,
      "loss": 0.1661,
      "step": 560
    },
    {
      "epoch": 1.3231132075471699,
      "grad_norm": 0.181640625,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.2361,
      "step": 561
    },
    {
      "epoch": 1.3254716981132075,
      "grad_norm": 0.2099609375,
      "learning_rate": 4.4432367149758455e-05,
      "loss": 0.2356,
      "step": 562
    },
    {
      "epoch": 1.3278301886792452,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.4420289855072464e-05,
      "loss": 0.1663,
      "step": 563
    },
    {
      "epoch": 1.330188679245283,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.440821256038647e-05,
      "loss": 0.1963,
      "step": 564
    },
    {
      "epoch": 1.3325471698113207,
      "grad_norm": 0.181640625,
      "learning_rate": 4.439613526570048e-05,
      "loss": 0.2084,
      "step": 565
    },
    {
      "epoch": 1.3349056603773586,
      "grad_norm": 0.1953125,
      "learning_rate": 4.438405797101449e-05,
      "loss": 0.1783,
      "step": 566
    },
    {
      "epoch": 1.3372641509433962,
      "grad_norm": 0.1875,
      "learning_rate": 4.437198067632851e-05,
      "loss": 0.216,
      "step": 567
    },
    {
      "epoch": 1.3396226415094339,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.4359903381642516e-05,
      "loss": 0.1546,
      "step": 568
    },
    {
      "epoch": 1.3419811320754718,
      "grad_norm": 0.1875,
      "learning_rate": 4.4347826086956525e-05,
      "loss": 0.252,
      "step": 569
    },
    {
      "epoch": 1.3443396226415094,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.4335748792270534e-05,
      "loss": 0.2514,
      "step": 570
    },
    {
      "epoch": 1.3466981132075473,
      "grad_norm": 0.189453125,
      "learning_rate": 4.432367149758454e-05,
      "loss": 0.1867,
      "step": 571
    },
    {
      "epoch": 1.349056603773585,
      "grad_norm": 0.19140625,
      "learning_rate": 4.431159420289855e-05,
      "loss": 0.2068,
      "step": 572
    },
    {
      "epoch": 1.3514150943396226,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.429951690821256e-05,
      "loss": 0.1882,
      "step": 573
    },
    {
      "epoch": 1.3537735849056602,
      "grad_norm": 0.169921875,
      "learning_rate": 4.428743961352657e-05,
      "loss": 0.2432,
      "step": 574
    },
    {
      "epoch": 1.3561320754716981,
      "grad_norm": 0.1796875,
      "learning_rate": 4.427536231884058e-05,
      "loss": 0.1826,
      "step": 575
    },
    {
      "epoch": 1.3584905660377358,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.4263285024154594e-05,
      "loss": 0.2079,
      "step": 576
    },
    {
      "epoch": 1.3608490566037736,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.42512077294686e-05,
      "loss": 0.1799,
      "step": 577
    },
    {
      "epoch": 1.3632075471698113,
      "grad_norm": 0.1796875,
      "learning_rate": 4.423913043478261e-05,
      "loss": 0.2276,
      "step": 578
    },
    {
      "epoch": 1.365566037735849,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.422705314009662e-05,
      "loss": 0.198,
      "step": 579
    },
    {
      "epoch": 1.3679245283018868,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.421497584541063e-05,
      "loss": 0.2243,
      "step": 580
    },
    {
      "epoch": 1.3702830188679245,
      "grad_norm": 0.2001953125,
      "learning_rate": 4.4202898550724645e-05,
      "loss": 0.1526,
      "step": 581
    },
    {
      "epoch": 1.3726415094339623,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.419082125603865e-05,
      "loss": 0.217,
      "step": 582
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.4178743961352656e-05,
      "loss": 0.2284,
      "step": 583
    },
    {
      "epoch": 1.3773584905660377,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.4166666666666665e-05,
      "loss": 0.2343,
      "step": 584
    },
    {
      "epoch": 1.3797169811320755,
      "grad_norm": 0.173828125,
      "learning_rate": 4.4154589371980674e-05,
      "loss": 0.2199,
      "step": 585
    },
    {
      "epoch": 1.3820754716981132,
      "grad_norm": 0.208984375,
      "learning_rate": 4.414251207729469e-05,
      "loss": 0.1973,
      "step": 586
    },
    {
      "epoch": 1.384433962264151,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.41304347826087e-05,
      "loss": 0.2089,
      "step": 587
    },
    {
      "epoch": 1.3867924528301887,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.411835748792271e-05,
      "loss": 0.2843,
      "step": 588
    },
    {
      "epoch": 1.3891509433962264,
      "grad_norm": 0.1796875,
      "learning_rate": 4.4106280193236716e-05,
      "loss": 0.2258,
      "step": 589
    },
    {
      "epoch": 1.3915094339622642,
      "grad_norm": 0.193359375,
      "learning_rate": 4.409420289855073e-05,
      "loss": 0.1735,
      "step": 590
    },
    {
      "epoch": 1.3938679245283019,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.408212560386474e-05,
      "loss": 0.2161,
      "step": 591
    },
    {
      "epoch": 1.3962264150943398,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.407004830917874e-05,
      "loss": 0.1804,
      "step": 592
    },
    {
      "epoch": 1.3985849056603774,
      "grad_norm": 0.177734375,
      "learning_rate": 4.405797101449275e-05,
      "loss": 0.2205,
      "step": 593
    },
    {
      "epoch": 1.400943396226415,
      "grad_norm": 0.171875,
      "learning_rate": 4.404589371980676e-05,
      "loss": 0.1507,
      "step": 594
    },
    {
      "epoch": 1.4033018867924527,
      "grad_norm": 0.169921875,
      "learning_rate": 4.403381642512078e-05,
      "loss": 0.1623,
      "step": 595
    },
    {
      "epoch": 1.4056603773584906,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.4021739130434786e-05,
      "loss": 0.244,
      "step": 596
    },
    {
      "epoch": 1.4080188679245282,
      "grad_norm": 0.185546875,
      "learning_rate": 4.4009661835748794e-05,
      "loss": 0.2622,
      "step": 597
    },
    {
      "epoch": 1.4103773584905661,
      "grad_norm": 0.18359375,
      "learning_rate": 4.39975845410628e-05,
      "loss": 0.198,
      "step": 598
    },
    {
      "epoch": 1.4127358490566038,
      "grad_norm": 0.189453125,
      "learning_rate": 4.398550724637681e-05,
      "loss": 0.2169,
      "step": 599
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.397342995169083e-05,
      "loss": 0.2353,
      "step": 600
    },
    {
      "epoch": 1.4174528301886793,
      "grad_norm": 0.185546875,
      "learning_rate": 4.396135265700484e-05,
      "loss": 0.2098,
      "step": 601
    },
    {
      "epoch": 1.419811320754717,
      "grad_norm": 0.177734375,
      "learning_rate": 4.3949275362318846e-05,
      "loss": 0.2388,
      "step": 602
    },
    {
      "epoch": 1.4221698113207548,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.393719806763285e-05,
      "loss": 0.2057,
      "step": 603
    },
    {
      "epoch": 1.4245283018867925,
      "grad_norm": 0.18359375,
      "learning_rate": 4.392512077294686e-05,
      "loss": 0.2122,
      "step": 604
    },
    {
      "epoch": 1.4268867924528301,
      "grad_norm": 0.17578125,
      "learning_rate": 4.391304347826087e-05,
      "loss": 0.2538,
      "step": 605
    },
    {
      "epoch": 1.429245283018868,
      "grad_norm": 0.189453125,
      "learning_rate": 4.390096618357488e-05,
      "loss": 0.185,
      "step": 606
    },
    {
      "epoch": 1.4316037735849056,
      "grad_norm": 0.166015625,
      "learning_rate": 4.388888888888889e-05,
      "loss": 0.1923,
      "step": 607
    },
    {
      "epoch": 1.4339622641509435,
      "grad_norm": 0.19140625,
      "learning_rate": 4.38768115942029e-05,
      "loss": 0.2381,
      "step": 608
    },
    {
      "epoch": 1.4363207547169812,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.3864734299516915e-05,
      "loss": 0.2313,
      "step": 609
    },
    {
      "epoch": 1.4386792452830188,
      "grad_norm": 0.18359375,
      "learning_rate": 4.3852657004830924e-05,
      "loss": 0.1926,
      "step": 610
    },
    {
      "epoch": 1.4410377358490565,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.384057971014493e-05,
      "loss": 0.1756,
      "step": 611
    },
    {
      "epoch": 1.4433962264150944,
      "grad_norm": 0.16796875,
      "learning_rate": 4.382850241545894e-05,
      "loss": 0.213,
      "step": 612
    },
    {
      "epoch": 1.445754716981132,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.3816425120772944e-05,
      "loss": 0.2409,
      "step": 613
    },
    {
      "epoch": 1.4481132075471699,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.380434782608696e-05,
      "loss": 0.2701,
      "step": 614
    },
    {
      "epoch": 1.4504716981132075,
      "grad_norm": 0.1953125,
      "learning_rate": 4.379227053140097e-05,
      "loss": 0.2253,
      "step": 615
    },
    {
      "epoch": 1.4528301886792452,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.378019323671498e-05,
      "loss": 0.1752,
      "step": 616
    },
    {
      "epoch": 1.455188679245283,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.3768115942028986e-05,
      "loss": 0.2302,
      "step": 617
    },
    {
      "epoch": 1.4575471698113207,
      "grad_norm": 0.169921875,
      "learning_rate": 4.3756038647342995e-05,
      "loss": 0.2329,
      "step": 618
    },
    {
      "epoch": 1.4599056603773586,
      "grad_norm": 0.19140625,
      "learning_rate": 4.374396135265701e-05,
      "loss": 0.2356,
      "step": 619
    },
    {
      "epoch": 1.4622641509433962,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.373188405797102e-05,
      "loss": 0.2693,
      "step": 620
    },
    {
      "epoch": 1.4646226415094339,
      "grad_norm": 0.185546875,
      "learning_rate": 4.371980676328503e-05,
      "loss": 0.1653,
      "step": 621
    },
    {
      "epoch": 1.4669811320754718,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.370772946859904e-05,
      "loss": 0.2275,
      "step": 622
    },
    {
      "epoch": 1.4693396226415094,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.3695652173913046e-05,
      "loss": 0.244,
      "step": 623
    },
    {
      "epoch": 1.4716981132075473,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.3683574879227055e-05,
      "loss": 0.2125,
      "step": 624
    },
    {
      "epoch": 1.474056603773585,
      "grad_norm": 0.17578125,
      "learning_rate": 4.3671497584541064e-05,
      "loss": 0.2369,
      "step": 625
    },
    {
      "epoch": 1.4764150943396226,
      "grad_norm": 0.193359375,
      "learning_rate": 4.365942028985507e-05,
      "loss": 0.1618,
      "step": 626
    },
    {
      "epoch": 1.4787735849056602,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.364734299516908e-05,
      "loss": 0.246,
      "step": 627
    },
    {
      "epoch": 1.4811320754716981,
      "grad_norm": 0.19140625,
      "learning_rate": 4.363526570048309e-05,
      "loss": 0.2355,
      "step": 628
    },
    {
      "epoch": 1.4834905660377358,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.362318840579711e-05,
      "loss": 0.1971,
      "step": 629
    },
    {
      "epoch": 1.4858490566037736,
      "grad_norm": 0.193359375,
      "learning_rate": 4.3611111111111116e-05,
      "loss": 0.2319,
      "step": 630
    },
    {
      "epoch": 1.4882075471698113,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.3599033816425124e-05,
      "loss": 0.2107,
      "step": 631
    },
    {
      "epoch": 1.490566037735849,
      "grad_norm": 0.17578125,
      "learning_rate": 4.3586956521739133e-05,
      "loss": 0.2178,
      "step": 632
    },
    {
      "epoch": 1.4929245283018868,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.357487922705314e-05,
      "loss": 0.1887,
      "step": 633
    },
    {
      "epoch": 1.4952830188679245,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.356280193236715e-05,
      "loss": 0.1996,
      "step": 634
    },
    {
      "epoch": 1.4976415094339623,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.355072463768116e-05,
      "loss": 0.325,
      "step": 635
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.353864734299517e-05,
      "loss": 0.1693,
      "step": 636
    },
    {
      "epoch": 1.5023584905660377,
      "grad_norm": 0.17578125,
      "learning_rate": 4.352657004830918e-05,
      "loss": 0.2586,
      "step": 637
    },
    {
      "epoch": 1.5047169811320755,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.3514492753623194e-05,
      "loss": 0.2369,
      "step": 638
    },
    {
      "epoch": 1.5070754716981132,
      "grad_norm": 0.185546875,
      "learning_rate": 4.35024154589372e-05,
      "loss": 0.1628,
      "step": 639
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 0.17578125,
      "learning_rate": 4.349033816425121e-05,
      "loss": 0.2164,
      "step": 640
    },
    {
      "epoch": 1.5117924528301887,
      "grad_norm": 0.197265625,
      "learning_rate": 4.347826086956522e-05,
      "loss": 0.2098,
      "step": 641
    },
    {
      "epoch": 1.5141509433962264,
      "grad_norm": 0.19140625,
      "learning_rate": 4.346618357487923e-05,
      "loss": 0.2351,
      "step": 642
    },
    {
      "epoch": 1.516509433962264,
      "grad_norm": 0.171875,
      "learning_rate": 4.345410628019324e-05,
      "loss": 0.1725,
      "step": 643
    },
    {
      "epoch": 1.5188679245283019,
      "grad_norm": 0.1796875,
      "learning_rate": 4.344202898550725e-05,
      "loss": 0.292,
      "step": 644
    },
    {
      "epoch": 1.5212264150943398,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.3429951690821256e-05,
      "loss": 0.2255,
      "step": 645
    },
    {
      "epoch": 1.5235849056603774,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.3417874396135265e-05,
      "loss": 0.2405,
      "step": 646
    },
    {
      "epoch": 1.525943396226415,
      "grad_norm": 0.171875,
      "learning_rate": 4.3405797101449274e-05,
      "loss": 0.1365,
      "step": 647
    },
    {
      "epoch": 1.5283018867924527,
      "grad_norm": 0.1796875,
      "learning_rate": 4.339371980676329e-05,
      "loss": 0.226,
      "step": 648
    },
    {
      "epoch": 1.5306603773584906,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.33816425120773e-05,
      "loss": 0.2234,
      "step": 649
    },
    {
      "epoch": 1.5330188679245285,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.336956521739131e-05,
      "loss": 0.1871,
      "step": 650
    },
    {
      "epoch": 1.5353773584905661,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.3357487922705316e-05,
      "loss": 0.1553,
      "step": 651
    },
    {
      "epoch": 1.5377358490566038,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.3345410628019325e-05,
      "loss": 0.1986,
      "step": 652
    },
    {
      "epoch": 1.5400943396226414,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 0.1834,
      "step": 653
    },
    {
      "epoch": 1.5424528301886793,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.332125603864734e-05,
      "loss": 0.2358,
      "step": 654
    },
    {
      "epoch": 1.544811320754717,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.330917874396135e-05,
      "loss": 0.2054,
      "step": 655
    },
    {
      "epoch": 1.5471698113207548,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.329710144927536e-05,
      "loss": 0.2357,
      "step": 656
    },
    {
      "epoch": 1.5495283018867925,
      "grad_norm": 0.203125,
      "learning_rate": 4.3285024154589376e-05,
      "loss": 0.1766,
      "step": 657
    },
    {
      "epoch": 1.5518867924528301,
      "grad_norm": 0.181640625,
      "learning_rate": 4.3272946859903385e-05,
      "loss": 0.2229,
      "step": 658
    },
    {
      "epoch": 1.5542452830188678,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.3260869565217394e-05,
      "loss": 0.2601,
      "step": 659
    },
    {
      "epoch": 1.5566037735849056,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.32487922705314e-05,
      "loss": 0.186,
      "step": 660
    },
    {
      "epoch": 1.5589622641509435,
      "grad_norm": 0.173828125,
      "learning_rate": 4.323671497584541e-05,
      "loss": 0.2052,
      "step": 661
    },
    {
      "epoch": 1.5613207547169812,
      "grad_norm": 0.177734375,
      "learning_rate": 4.322463768115943e-05,
      "loss": 0.208,
      "step": 662
    },
    {
      "epoch": 1.5636792452830188,
      "grad_norm": 0.1875,
      "learning_rate": 4.321256038647343e-05,
      "loss": 0.229,
      "step": 663
    },
    {
      "epoch": 1.5660377358490565,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.320048309178744e-05,
      "loss": 0.1981,
      "step": 664
    },
    {
      "epoch": 1.5683962264150944,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.318840579710145e-05,
      "loss": 0.237,
      "step": 665
    },
    {
      "epoch": 1.5707547169811322,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.317632850241546e-05,
      "loss": 0.1616,
      "step": 666
    },
    {
      "epoch": 1.5731132075471699,
      "grad_norm": 0.173828125,
      "learning_rate": 4.316425120772947e-05,
      "loss": 0.2627,
      "step": 667
    },
    {
      "epoch": 1.5754716981132075,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.315217391304348e-05,
      "loss": 0.1916,
      "step": 668
    },
    {
      "epoch": 1.5778301886792452,
      "grad_norm": 0.197265625,
      "learning_rate": 4.314009661835749e-05,
      "loss": 0.2466,
      "step": 669
    },
    {
      "epoch": 1.580188679245283,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.31280193236715e-05,
      "loss": 0.214,
      "step": 670
    },
    {
      "epoch": 1.5825471698113207,
      "grad_norm": 0.1796875,
      "learning_rate": 4.3115942028985515e-05,
      "loss": 0.2344,
      "step": 671
    },
    {
      "epoch": 1.5849056603773586,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.3103864734299524e-05,
      "loss": 0.1622,
      "step": 672
    },
    {
      "epoch": 1.5872641509433962,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.3091787439613526e-05,
      "loss": 0.2049,
      "step": 673
    },
    {
      "epoch": 1.5896226415094339,
      "grad_norm": 0.166015625,
      "learning_rate": 4.3079710144927535e-05,
      "loss": 0.2044,
      "step": 674
    },
    {
      "epoch": 1.5919811320754715,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.3067632850241544e-05,
      "loss": 0.2177,
      "step": 675
    },
    {
      "epoch": 1.5943396226415094,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.1978,
      "step": 676
    },
    {
      "epoch": 1.5966981132075473,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.304347826086957e-05,
      "loss": 0.2,
      "step": 677
    },
    {
      "epoch": 1.599056603773585,
      "grad_norm": 0.189453125,
      "learning_rate": 4.303140096618358e-05,
      "loss": 0.2146,
      "step": 678
    },
    {
      "epoch": 1.6014150943396226,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.3019323671497586e-05,
      "loss": 0.2307,
      "step": 679
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.3007246376811595e-05,
      "loss": 0.2075,
      "step": 680
    },
    {
      "epoch": 1.6061320754716981,
      "grad_norm": 0.185546875,
      "learning_rate": 4.299516908212561e-05,
      "loss": 0.2119,
      "step": 681
    },
    {
      "epoch": 1.608490566037736,
      "grad_norm": 0.208984375,
      "learning_rate": 4.298309178743962e-05,
      "loss": 0.1721,
      "step": 682
    },
    {
      "epoch": 1.6108490566037736,
      "grad_norm": 0.1875,
      "learning_rate": 4.297101449275363e-05,
      "loss": 0.2322,
      "step": 683
    },
    {
      "epoch": 1.6132075471698113,
      "grad_norm": 0.177734375,
      "learning_rate": 4.295893719806763e-05,
      "loss": 0.1744,
      "step": 684
    },
    {
      "epoch": 1.615566037735849,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.294685990338164e-05,
      "loss": 0.2662,
      "step": 685
    },
    {
      "epoch": 1.6179245283018868,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.2934782608695655e-05,
      "loss": 0.2608,
      "step": 686
    },
    {
      "epoch": 1.6202830188679245,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.2922705314009664e-05,
      "loss": 0.213,
      "step": 687
    },
    {
      "epoch": 1.6226415094339623,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.291062801932367e-05,
      "loss": 0.2192,
      "step": 688
    },
    {
      "epoch": 1.625,
      "grad_norm": 0.173828125,
      "learning_rate": 4.289855072463768e-05,
      "loss": 0.2234,
      "step": 689
    },
    {
      "epoch": 1.6273584905660377,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.288647342995169e-05,
      "loss": 0.191,
      "step": 690
    },
    {
      "epoch": 1.6297169811320755,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.2874396135265707e-05,
      "loss": 0.158,
      "step": 691
    },
    {
      "epoch": 1.6320754716981132,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.2862318840579715e-05,
      "loss": 0.2012,
      "step": 692
    },
    {
      "epoch": 1.634433962264151,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.2850241545893724e-05,
      "loss": 0.1813,
      "step": 693
    },
    {
      "epoch": 1.6367924528301887,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.2838164251207726e-05,
      "loss": 0.2173,
      "step": 694
    },
    {
      "epoch": 1.6391509433962264,
      "grad_norm": 0.19140625,
      "learning_rate": 4.282608695652174e-05,
      "loss": 0.1685,
      "step": 695
    },
    {
      "epoch": 1.641509433962264,
      "grad_norm": 0.19140625,
      "learning_rate": 4.281400966183575e-05,
      "loss": 0.2017,
      "step": 696
    },
    {
      "epoch": 1.6438679245283019,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.280193236714976e-05,
      "loss": 0.2093,
      "step": 697
    },
    {
      "epoch": 1.6462264150943398,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.278985507246377e-05,
      "loss": 0.1914,
      "step": 698
    },
    {
      "epoch": 1.6485849056603774,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.277777777777778e-05,
      "loss": 0.161,
      "step": 699
    },
    {
      "epoch": 1.650943396226415,
      "grad_norm": 0.173828125,
      "learning_rate": 4.2765700483091793e-05,
      "loss": 0.2604,
      "step": 700
    },
    {
      "epoch": 1.6533018867924527,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.27536231884058e-05,
      "loss": 0.2281,
      "step": 701
    },
    {
      "epoch": 1.6556603773584906,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.274154589371981e-05,
      "loss": 0.2372,
      "step": 702
    },
    {
      "epoch": 1.6580188679245285,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.272946859903382e-05,
      "loss": 0.2395,
      "step": 703
    },
    {
      "epoch": 1.6603773584905661,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.271739130434783e-05,
      "loss": 0.246,
      "step": 704
    },
    {
      "epoch": 1.6627358490566038,
      "grad_norm": 0.177734375,
      "learning_rate": 4.270531400966184e-05,
      "loss": 0.2408,
      "step": 705
    },
    {
      "epoch": 1.6650943396226414,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.269323671497585e-05,
      "loss": 0.1842,
      "step": 706
    },
    {
      "epoch": 1.6674528301886793,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.2681159420289856e-05,
      "loss": 0.171,
      "step": 707
    },
    {
      "epoch": 1.669811320754717,
      "grad_norm": 0.1875,
      "learning_rate": 4.2669082125603865e-05,
      "loss": 0.2483,
      "step": 708
    },
    {
      "epoch": 1.6721698113207548,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.2657004830917874e-05,
      "loss": 0.2174,
      "step": 709
    },
    {
      "epoch": 1.6745283018867925,
      "grad_norm": 0.1875,
      "learning_rate": 4.264492753623189e-05,
      "loss": 0.1855,
      "step": 710
    },
    {
      "epoch": 1.6768867924528301,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.26328502415459e-05,
      "loss": 0.2511,
      "step": 711
    },
    {
      "epoch": 1.6792452830188678,
      "grad_norm": 0.189453125,
      "learning_rate": 4.262077294685991e-05,
      "loss": 0.251,
      "step": 712
    },
    {
      "epoch": 1.6816037735849056,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.2608695652173916e-05,
      "loss": 0.1684,
      "step": 713
    },
    {
      "epoch": 1.6839622641509435,
      "grad_norm": 0.166015625,
      "learning_rate": 4.2596618357487925e-05,
      "loss": 0.2031,
      "step": 714
    },
    {
      "epoch": 1.6863207547169812,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.2584541062801934e-05,
      "loss": 0.1838,
      "step": 715
    },
    {
      "epoch": 1.6886792452830188,
      "grad_norm": 0.1953125,
      "learning_rate": 4.257246376811594e-05,
      "loss": 0.1721,
      "step": 716
    },
    {
      "epoch": 1.6910377358490565,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.256038647342995e-05,
      "loss": 0.1933,
      "step": 717
    },
    {
      "epoch": 1.6933962264150944,
      "grad_norm": 0.177734375,
      "learning_rate": 4.254830917874396e-05,
      "loss": 0.2081,
      "step": 718
    },
    {
      "epoch": 1.6957547169811322,
      "grad_norm": 0.17578125,
      "learning_rate": 4.2536231884057976e-05,
      "loss": 0.1809,
      "step": 719
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 0.16796875,
      "learning_rate": 4.2524154589371985e-05,
      "loss": 0.2066,
      "step": 720
    },
    {
      "epoch": 1.7004716981132075,
      "grad_norm": 0.177734375,
      "learning_rate": 4.2512077294685994e-05,
      "loss": 0.2154,
      "step": 721
    },
    {
      "epoch": 1.7028301886792452,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.25e-05,
      "loss": 0.1604,
      "step": 722
    },
    {
      "epoch": 1.705188679245283,
      "grad_norm": 0.193359375,
      "learning_rate": 4.248792270531401e-05,
      "loss": 0.2113,
      "step": 723
    },
    {
      "epoch": 1.7075471698113207,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.247584541062802e-05,
      "loss": 0.2057,
      "step": 724
    },
    {
      "epoch": 1.7099056603773586,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.246376811594203e-05,
      "loss": 0.2194,
      "step": 725
    },
    {
      "epoch": 1.7122641509433962,
      "grad_norm": 0.173828125,
      "learning_rate": 4.245169082125604e-05,
      "loss": 0.1841,
      "step": 726
    },
    {
      "epoch": 1.7146226415094339,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.243961352657005e-05,
      "loss": 0.218,
      "step": 727
    },
    {
      "epoch": 1.7169811320754715,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.2427536231884057e-05,
      "loss": 0.2445,
      "step": 728
    },
    {
      "epoch": 1.7193396226415094,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.241545893719807e-05,
      "loss": 0.2042,
      "step": 729
    },
    {
      "epoch": 1.7216981132075473,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.240338164251208e-05,
      "loss": 0.226,
      "step": 730
    },
    {
      "epoch": 1.724056603773585,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.239130434782609e-05,
      "loss": 0.1803,
      "step": 731
    },
    {
      "epoch": 1.7264150943396226,
      "grad_norm": 0.18359375,
      "learning_rate": 4.23792270531401e-05,
      "loss": 0.2406,
      "step": 732
    },
    {
      "epoch": 1.7287735849056602,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.236714975845411e-05,
      "loss": 0.2102,
      "step": 733
    },
    {
      "epoch": 1.7311320754716981,
      "grad_norm": 0.166015625,
      "learning_rate": 4.235507246376812e-05,
      "loss": 0.2038,
      "step": 734
    },
    {
      "epoch": 1.733490566037736,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.2342995169082126e-05,
      "loss": 0.239,
      "step": 735
    },
    {
      "epoch": 1.7358490566037736,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.2330917874396135e-05,
      "loss": 0.2345,
      "step": 736
    },
    {
      "epoch": 1.7382075471698113,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.2318840579710143e-05,
      "loss": 0.2081,
      "step": 737
    },
    {
      "epoch": 1.740566037735849,
      "grad_norm": 0.2001953125,
      "learning_rate": 4.230676328502416e-05,
      "loss": 0.3349,
      "step": 738
    },
    {
      "epoch": 1.7429245283018868,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.229468599033817e-05,
      "loss": 0.2072,
      "step": 739
    },
    {
      "epoch": 1.7452830188679245,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.228260869565218e-05,
      "loss": 0.2703,
      "step": 740
    },
    {
      "epoch": 1.7476415094339623,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.2270531400966186e-05,
      "loss": 0.2552,
      "step": 741
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.181640625,
      "learning_rate": 4.2258454106280195e-05,
      "loss": 0.1623,
      "step": 742
    },
    {
      "epoch": 1.7523584905660377,
      "grad_norm": 0.181640625,
      "learning_rate": 4.224637681159421e-05,
      "loss": 0.2243,
      "step": 743
    },
    {
      "epoch": 1.7547169811320755,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.223429951690821e-05,
      "loss": 0.184,
      "step": 744
    },
    {
      "epoch": 1.7570754716981132,
      "grad_norm": 0.181640625,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.2278,
      "step": 745
    },
    {
      "epoch": 1.759433962264151,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.221014492753623e-05,
      "loss": 0.151,
      "step": 746
    },
    {
      "epoch": 1.7617924528301887,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.219806763285024e-05,
      "loss": 0.1757,
      "step": 747
    },
    {
      "epoch": 1.7641509433962264,
      "grad_norm": 0.177734375,
      "learning_rate": 4.2185990338164255e-05,
      "loss": 0.2068,
      "step": 748
    },
    {
      "epoch": 1.766509433962264,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.2173913043478264e-05,
      "loss": 0.186,
      "step": 749
    },
    {
      "epoch": 1.7688679245283019,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.216183574879227e-05,
      "loss": 0.1743,
      "step": 750
    },
    {
      "epoch": 1.7712264150943398,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.214975845410628e-05,
      "loss": 0.2309,
      "step": 751
    },
    {
      "epoch": 1.7735849056603774,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.213768115942029e-05,
      "loss": 0.2385,
      "step": 752
    },
    {
      "epoch": 1.775943396226415,
      "grad_norm": 0.18359375,
      "learning_rate": 4.2125603864734306e-05,
      "loss": 0.2294,
      "step": 753
    },
    {
      "epoch": 1.7783018867924527,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.211352657004831e-05,
      "loss": 0.2488,
      "step": 754
    },
    {
      "epoch": 1.7806603773584906,
      "grad_norm": 0.251953125,
      "learning_rate": 4.210144927536232e-05,
      "loss": 0.185,
      "step": 755
    },
    {
      "epoch": 1.7830188679245285,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.2089371980676326e-05,
      "loss": 0.1946,
      "step": 756
    },
    {
      "epoch": 1.7853773584905661,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.207729468599034e-05,
      "loss": 0.2336,
      "step": 757
    },
    {
      "epoch": 1.7877358490566038,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.206521739130435e-05,
      "loss": 0.2102,
      "step": 758
    },
    {
      "epoch": 1.7900943396226414,
      "grad_norm": 0.173828125,
      "learning_rate": 4.205314009661836e-05,
      "loss": 0.1886,
      "step": 759
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 0.1435546875,
      "learning_rate": 4.204106280193237e-05,
      "loss": 0.2648,
      "step": 760
    },
    {
      "epoch": 1.794811320754717,
      "grad_norm": 0.1953125,
      "learning_rate": 4.202898550724638e-05,
      "loss": 0.2237,
      "step": 761
    },
    {
      "epoch": 1.7971698113207548,
      "grad_norm": 0.171875,
      "learning_rate": 4.201690821256039e-05,
      "loss": 0.1726,
      "step": 762
    },
    {
      "epoch": 1.7995283018867925,
      "grad_norm": 0.171875,
      "learning_rate": 4.20048309178744e-05,
      "loss": 0.1945,
      "step": 763
    },
    {
      "epoch": 1.8018867924528301,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.199275362318841e-05,
      "loss": 0.193,
      "step": 764
    },
    {
      "epoch": 1.8042452830188678,
      "grad_norm": 0.169921875,
      "learning_rate": 4.198067632850241e-05,
      "loss": 0.2376,
      "step": 765
    },
    {
      "epoch": 1.8066037735849056,
      "grad_norm": 0.171875,
      "learning_rate": 4.196859903381642e-05,
      "loss": 0.186,
      "step": 766
    },
    {
      "epoch": 1.8089622641509435,
      "grad_norm": 0.19921875,
      "learning_rate": 4.195652173913044e-05,
      "loss": 0.2072,
      "step": 767
    },
    {
      "epoch": 1.8113207547169812,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.194444444444445e-05,
      "loss": 0.1168,
      "step": 768
    },
    {
      "epoch": 1.8136792452830188,
      "grad_norm": 0.185546875,
      "learning_rate": 4.1932367149758456e-05,
      "loss": 0.22,
      "step": 769
    },
    {
      "epoch": 1.8160377358490565,
      "grad_norm": 0.171875,
      "learning_rate": 4.1920289855072465e-05,
      "loss": 0.1496,
      "step": 770
    },
    {
      "epoch": 1.8183962264150944,
      "grad_norm": 0.189453125,
      "learning_rate": 4.1908212560386474e-05,
      "loss": 0.2285,
      "step": 771
    },
    {
      "epoch": 1.8207547169811322,
      "grad_norm": 0.2001953125,
      "learning_rate": 4.189613526570049e-05,
      "loss": 0.2242,
      "step": 772
    },
    {
      "epoch": 1.8231132075471699,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.18840579710145e-05,
      "loss": 0.1414,
      "step": 773
    },
    {
      "epoch": 1.8254716981132075,
      "grad_norm": 0.1640625,
      "learning_rate": 4.187198067632851e-05,
      "loss": 0.2634,
      "step": 774
    },
    {
      "epoch": 1.8278301886792452,
      "grad_norm": 0.1796875,
      "learning_rate": 4.185990338164251e-05,
      "loss": 0.208,
      "step": 775
    },
    {
      "epoch": 1.830188679245283,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.1847826086956525e-05,
      "loss": 0.2143,
      "step": 776
    },
    {
      "epoch": 1.8325471698113207,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.1835748792270534e-05,
      "loss": 0.1597,
      "step": 777
    },
    {
      "epoch": 1.8349056603773586,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.182367149758454e-05,
      "loss": 0.2352,
      "step": 778
    },
    {
      "epoch": 1.8372641509433962,
      "grad_norm": 0.185546875,
      "learning_rate": 4.181159420289855e-05,
      "loss": 0.2212,
      "step": 779
    },
    {
      "epoch": 1.8396226415094339,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.179951690821256e-05,
      "loss": 0.1893,
      "step": 780
    },
    {
      "epoch": 1.8419811320754715,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.1787439613526576e-05,
      "loss": 0.1695,
      "step": 781
    },
    {
      "epoch": 1.8443396226415094,
      "grad_norm": 0.189453125,
      "learning_rate": 4.1775362318840585e-05,
      "loss": 0.2045,
      "step": 782
    },
    {
      "epoch": 1.8466981132075473,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.1763285024154594e-05,
      "loss": 0.1643,
      "step": 783
    },
    {
      "epoch": 1.849056603773585,
      "grad_norm": 0.16015625,
      "learning_rate": 4.17512077294686e-05,
      "loss": 0.2276,
      "step": 784
    },
    {
      "epoch": 1.8514150943396226,
      "grad_norm": 0.169921875,
      "learning_rate": 4.1739130434782605e-05,
      "loss": 0.1776,
      "step": 785
    },
    {
      "epoch": 1.8537735849056602,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.172705314009662e-05,
      "loss": 0.2207,
      "step": 786
    },
    {
      "epoch": 1.8561320754716981,
      "grad_norm": 0.189453125,
      "learning_rate": 4.171497584541063e-05,
      "loss": 0.2373,
      "step": 787
    },
    {
      "epoch": 1.858490566037736,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.170289855072464e-05,
      "loss": 0.2157,
      "step": 788
    },
    {
      "epoch": 1.8608490566037736,
      "grad_norm": 0.177734375,
      "learning_rate": 4.169082125603865e-05,
      "loss": 0.2092,
      "step": 789
    },
    {
      "epoch": 1.8632075471698113,
      "grad_norm": 0.1796875,
      "learning_rate": 4.1678743961352656e-05,
      "loss": 0.1924,
      "step": 790
    },
    {
      "epoch": 1.865566037735849,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.1777,
      "step": 791
    },
    {
      "epoch": 1.8679245283018868,
      "grad_norm": 0.177734375,
      "learning_rate": 4.165458937198068e-05,
      "loss": 0.2097,
      "step": 792
    },
    {
      "epoch": 1.8702830188679245,
      "grad_norm": 0.177734375,
      "learning_rate": 4.164251207729469e-05,
      "loss": 0.2004,
      "step": 793
    },
    {
      "epoch": 1.8726415094339623,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.16304347826087e-05,
      "loss": 0.1682,
      "step": 794
    },
    {
      "epoch": 1.875,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.161835748792271e-05,
      "loss": 0.2233,
      "step": 795
    },
    {
      "epoch": 1.8773584905660377,
      "grad_norm": 0.201171875,
      "learning_rate": 4.1606280193236717e-05,
      "loss": 0.2327,
      "step": 796
    },
    {
      "epoch": 1.8797169811320755,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.1594202898550726e-05,
      "loss": 0.2283,
      "step": 797
    },
    {
      "epoch": 1.8820754716981132,
      "grad_norm": 0.162109375,
      "learning_rate": 4.1582125603864734e-05,
      "loss": 0.1803,
      "step": 798
    },
    {
      "epoch": 1.884433962264151,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.157004830917874e-05,
      "loss": 0.2033,
      "step": 799
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 0.177734375,
      "learning_rate": 4.155797101449276e-05,
      "loss": 0.196,
      "step": 800
    },
    {
      "epoch": 1.8891509433962264,
      "grad_norm": 0.19921875,
      "learning_rate": 4.154589371980677e-05,
      "loss": 0.2487,
      "step": 801
    },
    {
      "epoch": 1.891509433962264,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.153381642512078e-05,
      "loss": 0.1973,
      "step": 802
    },
    {
      "epoch": 1.8938679245283019,
      "grad_norm": 0.189453125,
      "learning_rate": 4.1521739130434786e-05,
      "loss": 0.1681,
      "step": 803
    },
    {
      "epoch": 1.8962264150943398,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.1509661835748795e-05,
      "loss": 0.235,
      "step": 804
    },
    {
      "epoch": 1.8985849056603774,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.1497584541062804e-05,
      "loss": 0.1762,
      "step": 805
    },
    {
      "epoch": 1.900943396226415,
      "grad_norm": 0.173828125,
      "learning_rate": 4.148550724637681e-05,
      "loss": 0.183,
      "step": 806
    },
    {
      "epoch": 1.9033018867924527,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.147342995169082e-05,
      "loss": 0.2162,
      "step": 807
    },
    {
      "epoch": 1.9056603773584906,
      "grad_norm": 0.185546875,
      "learning_rate": 4.146135265700483e-05,
      "loss": 0.2272,
      "step": 808
    },
    {
      "epoch": 1.9080188679245285,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.144927536231884e-05,
      "loss": 0.1966,
      "step": 809
    },
    {
      "epoch": 1.9103773584905661,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.1437198067632855e-05,
      "loss": 0.2346,
      "step": 810
    },
    {
      "epoch": 1.9127358490566038,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.1425120772946864e-05,
      "loss": 0.2566,
      "step": 811
    },
    {
      "epoch": 1.9150943396226414,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.141304347826087e-05,
      "loss": 0.2432,
      "step": 812
    },
    {
      "epoch": 1.9174528301886793,
      "grad_norm": 0.193359375,
      "learning_rate": 4.140096618357488e-05,
      "loss": 0.2137,
      "step": 813
    },
    {
      "epoch": 1.919811320754717,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.138888888888889e-05,
      "loss": 0.2406,
      "step": 814
    },
    {
      "epoch": 1.9221698113207548,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.13768115942029e-05,
      "loss": 0.1665,
      "step": 815
    },
    {
      "epoch": 1.9245283018867925,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.136473429951691e-05,
      "loss": 0.1815,
      "step": 816
    },
    {
      "epoch": 1.9268867924528301,
      "grad_norm": 0.189453125,
      "learning_rate": 4.135265700483092e-05,
      "loss": 0.2658,
      "step": 817
    },
    {
      "epoch": 1.9292452830188678,
      "grad_norm": 0.1953125,
      "learning_rate": 4.1340579710144926e-05,
      "loss": 0.2009,
      "step": 818
    },
    {
      "epoch": 1.9316037735849056,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.132850241545894e-05,
      "loss": 0.1572,
      "step": 819
    },
    {
      "epoch": 1.9339622641509435,
      "grad_norm": 0.19140625,
      "learning_rate": 4.131642512077295e-05,
      "loss": 0.2028,
      "step": 820
    },
    {
      "epoch": 1.9363207547169812,
      "grad_norm": 0.17578125,
      "learning_rate": 4.130434782608696e-05,
      "loss": 0.2114,
      "step": 821
    },
    {
      "epoch": 1.9386792452830188,
      "grad_norm": 0.171875,
      "learning_rate": 4.129227053140097e-05,
      "loss": 0.1673,
      "step": 822
    },
    {
      "epoch": 1.9410377358490565,
      "grad_norm": 0.1796875,
      "learning_rate": 4.128019323671498e-05,
      "loss": 0.1817,
      "step": 823
    },
    {
      "epoch": 1.9433962264150944,
      "grad_norm": 0.1796875,
      "learning_rate": 4.1268115942028986e-05,
      "loss": 0.2004,
      "step": 824
    },
    {
      "epoch": 1.9457547169811322,
      "grad_norm": 0.189453125,
      "learning_rate": 4.1256038647342995e-05,
      "loss": 0.1819,
      "step": 825
    },
    {
      "epoch": 1.9481132075471699,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.1243961352657004e-05,
      "loss": 0.2353,
      "step": 826
    },
    {
      "epoch": 1.9504716981132075,
      "grad_norm": 0.185546875,
      "learning_rate": 4.123188405797101e-05,
      "loss": 0.223,
      "step": 827
    },
    {
      "epoch": 1.9528301886792452,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.121980676328502e-05,
      "loss": 0.1905,
      "step": 828
    },
    {
      "epoch": 1.955188679245283,
      "grad_norm": 0.2109375,
      "learning_rate": 4.120772946859904e-05,
      "loss": 0.2472,
      "step": 829
    },
    {
      "epoch": 1.9575471698113207,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.119565217391305e-05,
      "loss": 0.195,
      "step": 830
    },
    {
      "epoch": 1.9599056603773586,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.1183574879227056e-05,
      "loss": 0.2392,
      "step": 831
    },
    {
      "epoch": 1.9622641509433962,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.1171497584541064e-05,
      "loss": 0.2165,
      "step": 832
    },
    {
      "epoch": 1.9646226415094339,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.115942028985507e-05,
      "loss": 0.1971,
      "step": 833
    },
    {
      "epoch": 1.9669811320754715,
      "grad_norm": 0.1640625,
      "learning_rate": 4.114734299516909e-05,
      "loss": 0.1785,
      "step": 834
    },
    {
      "epoch": 1.9693396226415094,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.113526570048309e-05,
      "loss": 0.2001,
      "step": 835
    },
    {
      "epoch": 1.9716981132075473,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.11231884057971e-05,
      "loss": 0.2403,
      "step": 836
    },
    {
      "epoch": 1.974056603773585,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.111111111111111e-05,
      "loss": 0.2136,
      "step": 837
    },
    {
      "epoch": 1.9764150943396226,
      "grad_norm": 0.1796875,
      "learning_rate": 4.1099033816425125e-05,
      "loss": 0.1629,
      "step": 838
    },
    {
      "epoch": 1.9787735849056602,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.1086956521739134e-05,
      "loss": 0.1848,
      "step": 839
    },
    {
      "epoch": 1.9811320754716981,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.107487922705314e-05,
      "loss": 0.2127,
      "step": 840
    },
    {
      "epoch": 1.983490566037736,
      "grad_norm": 0.185546875,
      "learning_rate": 4.106280193236715e-05,
      "loss": 0.1934,
      "step": 841
    },
    {
      "epoch": 1.9858490566037736,
      "grad_norm": 0.177734375,
      "learning_rate": 4.105072463768116e-05,
      "loss": 0.1965,
      "step": 842
    },
    {
      "epoch": 1.9882075471698113,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.1038647342995176e-05,
      "loss": 0.2302,
      "step": 843
    },
    {
      "epoch": 1.990566037735849,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.1026570048309185e-05,
      "loss": 0.192,
      "step": 844
    },
    {
      "epoch": 1.9929245283018868,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.101449275362319e-05,
      "loss": 0.2492,
      "step": 845
    },
    {
      "epoch": 1.9952830188679245,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.1002415458937196e-05,
      "loss": 0.1668,
      "step": 846
    },
    {
      "epoch": 1.9976415094339623,
      "grad_norm": 0.15625,
      "learning_rate": 4.0990338164251205e-05,
      "loss": 0.2165,
      "step": 847
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.359375,
      "learning_rate": 4.097826086956522e-05,
      "loss": 0.1189,
      "step": 848
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.1894041895866394,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 1.9363,
      "eval_samples_per_second": 1.549,
      "eval_steps_per_second": 1.549,
      "step": 848
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.1335233747959137,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 1.3886,
      "eval_samples_per_second": 2.881,
      "eval_steps_per_second": 2.881,
      "step": 848
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.28316783905029297,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 3.0794,
      "eval_samples_per_second": 1.299,
      "eval_steps_per_second": 1.299,
      "step": 848
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.3653801679611206,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 2.9031,
      "eval_samples_per_second": 1.378,
      "eval_steps_per_second": 1.378,
      "step": 848
    },
    {
      "epoch": 2.0023584905660377,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.096618357487923e-05,
      "loss": 0.1316,
      "step": 849
    },
    {
      "epoch": 2.0047169811320753,
      "grad_norm": 0.173828125,
      "learning_rate": 4.095410628019324e-05,
      "loss": 0.2495,
      "step": 850
    },
    {
      "epoch": 2.0070754716981134,
      "grad_norm": 0.20703125,
      "learning_rate": 4.094202898550725e-05,
      "loss": 0.212,
      "step": 851
    },
    {
      "epoch": 2.009433962264151,
      "grad_norm": 0.17578125,
      "learning_rate": 4.0929951690821256e-05,
      "loss": 0.1642,
      "step": 852
    },
    {
      "epoch": 2.0117924528301887,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.091787439613527e-05,
      "loss": 0.2194,
      "step": 853
    },
    {
      "epoch": 2.0141509433962264,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.090579710144928e-05,
      "loss": 0.1889,
      "step": 854
    },
    {
      "epoch": 2.016509433962264,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.089371980676329e-05,
      "loss": 0.1623,
      "step": 855
    },
    {
      "epoch": 2.018867924528302,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.088164251207729e-05,
      "loss": 0.2157,
      "step": 856
    },
    {
      "epoch": 2.0212264150943398,
      "grad_norm": 0.1689453125,
      "learning_rate": 4.086956521739131e-05,
      "loss": 0.2043,
      "step": 857
    },
    {
      "epoch": 2.0235849056603774,
      "grad_norm": 0.1953125,
      "learning_rate": 4.0857487922705316e-05,
      "loss": 0.2281,
      "step": 858
    },
    {
      "epoch": 2.025943396226415,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.0845410628019325e-05,
      "loss": 0.1865,
      "step": 859
    },
    {
      "epoch": 2.0283018867924527,
      "grad_norm": 0.197265625,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 0.2043,
      "step": 860
    },
    {
      "epoch": 2.0306603773584904,
      "grad_norm": 0.1875,
      "learning_rate": 4.082125603864734e-05,
      "loss": 0.1371,
      "step": 861
    },
    {
      "epoch": 2.0330188679245285,
      "grad_norm": 0.169921875,
      "learning_rate": 4.080917874396136e-05,
      "loss": 0.1811,
      "step": 862
    },
    {
      "epoch": 2.035377358490566,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.079710144927537e-05,
      "loss": 0.2127,
      "step": 863
    },
    {
      "epoch": 2.0377358490566038,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.078502415458938e-05,
      "loss": 0.2162,
      "step": 864
    },
    {
      "epoch": 2.0400943396226414,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.0772946859903386e-05,
      "loss": 0.2194,
      "step": 865
    },
    {
      "epoch": 2.042452830188679,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.076086956521739e-05,
      "loss": 0.181,
      "step": 866
    },
    {
      "epoch": 2.044811320754717,
      "grad_norm": 0.193359375,
      "learning_rate": 4.0748792270531403e-05,
      "loss": 0.1707,
      "step": 867
    },
    {
      "epoch": 2.047169811320755,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.073671497584541e-05,
      "loss": 0.1745,
      "step": 868
    },
    {
      "epoch": 2.0495283018867925,
      "grad_norm": 0.1796875,
      "learning_rate": 4.072463768115942e-05,
      "loss": 0.1778,
      "step": 869
    },
    {
      "epoch": 2.05188679245283,
      "grad_norm": 0.185546875,
      "learning_rate": 4.071256038647343e-05,
      "loss": 0.2681,
      "step": 870
    },
    {
      "epoch": 2.0542452830188678,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.070048309178744e-05,
      "loss": 0.1439,
      "step": 871
    },
    {
      "epoch": 2.056603773584906,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.0688405797101455e-05,
      "loss": 0.1901,
      "step": 872
    },
    {
      "epoch": 2.0589622641509435,
      "grad_norm": 0.181640625,
      "learning_rate": 4.0676328502415464e-05,
      "loss": 0.1252,
      "step": 873
    },
    {
      "epoch": 2.061320754716981,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.066425120772947e-05,
      "loss": 0.1836,
      "step": 874
    },
    {
      "epoch": 2.063679245283019,
      "grad_norm": 0.173828125,
      "learning_rate": 4.065217391304348e-05,
      "loss": 0.1753,
      "step": 875
    },
    {
      "epoch": 2.0660377358490565,
      "grad_norm": 0.185546875,
      "learning_rate": 4.064009661835749e-05,
      "loss": 0.1759,
      "step": 876
    },
    {
      "epoch": 2.068396226415094,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.06280193236715e-05,
      "loss": 0.1619,
      "step": 877
    },
    {
      "epoch": 2.0707547169811322,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.061594202898551e-05,
      "loss": 0.1735,
      "step": 878
    },
    {
      "epoch": 2.07311320754717,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.060386473429952e-05,
      "loss": 0.1604,
      "step": 879
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.0591787439613526e-05,
      "loss": 0.1954,
      "step": 880
    },
    {
      "epoch": 2.077830188679245,
      "grad_norm": 0.185546875,
      "learning_rate": 4.057971014492754e-05,
      "loss": 0.1961,
      "step": 881
    },
    {
      "epoch": 2.080188679245283,
      "grad_norm": 0.18359375,
      "learning_rate": 4.056763285024155e-05,
      "loss": 0.1376,
      "step": 882
    },
    {
      "epoch": 2.082547169811321,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.055555555555556e-05,
      "loss": 0.1498,
      "step": 883
    },
    {
      "epoch": 2.0849056603773586,
      "grad_norm": 0.201171875,
      "learning_rate": 4.054347826086957e-05,
      "loss": 0.2109,
      "step": 884
    },
    {
      "epoch": 2.0872641509433962,
      "grad_norm": 0.189453125,
      "learning_rate": 4.053140096618358e-05,
      "loss": 0.1731,
      "step": 885
    },
    {
      "epoch": 2.089622641509434,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.0519323671497586e-05,
      "loss": 0.1901,
      "step": 886
    },
    {
      "epoch": 2.0919811320754715,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.0507246376811595e-05,
      "loss": 0.2425,
      "step": 887
    },
    {
      "epoch": 2.0943396226415096,
      "grad_norm": 0.189453125,
      "learning_rate": 4.0495169082125604e-05,
      "loss": 0.152,
      "step": 888
    },
    {
      "epoch": 2.0966981132075473,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.048309178743961e-05,
      "loss": 0.1956,
      "step": 889
    },
    {
      "epoch": 2.099056603773585,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.047101449275362e-05,
      "loss": 0.1292,
      "step": 890
    },
    {
      "epoch": 2.1014150943396226,
      "grad_norm": 0.1875,
      "learning_rate": 4.045893719806764e-05,
      "loss": 0.1988,
      "step": 891
    },
    {
      "epoch": 2.1037735849056602,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.0446859903381646e-05,
      "loss": 0.1922,
      "step": 892
    },
    {
      "epoch": 2.106132075471698,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.0434782608695655e-05,
      "loss": 0.1292,
      "step": 893
    },
    {
      "epoch": 2.108490566037736,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.0422705314009664e-05,
      "loss": 0.1904,
      "step": 894
    },
    {
      "epoch": 2.1108490566037736,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.041062801932367e-05,
      "loss": 0.3066,
      "step": 895
    },
    {
      "epoch": 2.1132075471698113,
      "grad_norm": 0.1875,
      "learning_rate": 4.039855072463768e-05,
      "loss": 0.1962,
      "step": 896
    },
    {
      "epoch": 2.115566037735849,
      "grad_norm": 0.1875,
      "learning_rate": 4.038647342995169e-05,
      "loss": 0.1817,
      "step": 897
    },
    {
      "epoch": 2.1179245283018866,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.03743961352657e-05,
      "loss": 0.2305,
      "step": 898
    },
    {
      "epoch": 2.1202830188679247,
      "grad_norm": 0.19140625,
      "learning_rate": 4.036231884057971e-05,
      "loss": 0.1749,
      "step": 899
    },
    {
      "epoch": 2.1226415094339623,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.0350241545893725e-05,
      "loss": 0.1475,
      "step": 900
    },
    {
      "epoch": 2.125,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.0338164251207733e-05,
      "loss": 0.2368,
      "step": 901
    },
    {
      "epoch": 2.1273584905660377,
      "grad_norm": 0.1875,
      "learning_rate": 4.032608695652174e-05,
      "loss": 0.21,
      "step": 902
    },
    {
      "epoch": 2.1297169811320753,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.031400966183575e-05,
      "loss": 0.1789,
      "step": 903
    },
    {
      "epoch": 2.1320754716981134,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.030193236714976e-05,
      "loss": 0.1687,
      "step": 904
    },
    {
      "epoch": 2.134433962264151,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.028985507246377e-05,
      "loss": 0.1722,
      "step": 905
    },
    {
      "epoch": 2.1367924528301887,
      "grad_norm": 0.169921875,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.211,
      "step": 906
    },
    {
      "epoch": 2.1391509433962264,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.026570048309179e-05,
      "loss": 0.1953,
      "step": 907
    },
    {
      "epoch": 2.141509433962264,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.0253623188405796e-05,
      "loss": 0.1817,
      "step": 908
    },
    {
      "epoch": 2.143867924528302,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.0241545893719805e-05,
      "loss": 0.1319,
      "step": 909
    },
    {
      "epoch": 2.1462264150943398,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.022946859903382e-05,
      "loss": 0.2152,
      "step": 910
    },
    {
      "epoch": 2.1485849056603774,
      "grad_norm": 0.1875,
      "learning_rate": 4.021739130434783e-05,
      "loss": 0.1915,
      "step": 911
    },
    {
      "epoch": 2.150943396226415,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.020531400966184e-05,
      "loss": 0.1755,
      "step": 912
    },
    {
      "epoch": 2.1533018867924527,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.019323671497585e-05,
      "loss": 0.2014,
      "step": 913
    },
    {
      "epoch": 2.1556603773584904,
      "grad_norm": 0.224609375,
      "learning_rate": 4.0181159420289856e-05,
      "loss": 0.1947,
      "step": 914
    },
    {
      "epoch": 2.1580188679245285,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.016908212560387e-05,
      "loss": 0.2218,
      "step": 915
    },
    {
      "epoch": 2.160377358490566,
      "grad_norm": 0.1875,
      "learning_rate": 4.0157004830917874e-05,
      "loss": 0.1455,
      "step": 916
    },
    {
      "epoch": 2.1627358490566038,
      "grad_norm": 0.2177734375,
      "learning_rate": 4.014492753623188e-05,
      "loss": 0.1925,
      "step": 917
    },
    {
      "epoch": 2.1650943396226414,
      "grad_norm": 0.22265625,
      "learning_rate": 4.013285024154589e-05,
      "loss": 0.253,
      "step": 918
    },
    {
      "epoch": 2.167452830188679,
      "grad_norm": 0.203125,
      "learning_rate": 4.012077294685991e-05,
      "loss": 0.2179,
      "step": 919
    },
    {
      "epoch": 2.169811320754717,
      "grad_norm": 0.21484375,
      "learning_rate": 4.0108695652173916e-05,
      "loss": 0.1851,
      "step": 920
    },
    {
      "epoch": 2.172169811320755,
      "grad_norm": 0.17578125,
      "learning_rate": 4.0096618357487925e-05,
      "loss": 0.208,
      "step": 921
    },
    {
      "epoch": 2.1745283018867925,
      "grad_norm": 0.181640625,
      "learning_rate": 4.0084541062801934e-05,
      "loss": 0.141,
      "step": 922
    },
    {
      "epoch": 2.17688679245283,
      "grad_norm": 0.193359375,
      "learning_rate": 4.007246376811594e-05,
      "loss": 0.1958,
      "step": 923
    },
    {
      "epoch": 2.1792452830188678,
      "grad_norm": 0.1875,
      "learning_rate": 4.006038647342996e-05,
      "loss": 0.1791,
      "step": 924
    },
    {
      "epoch": 2.1816037735849054,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.004830917874397e-05,
      "loss": 0.1796,
      "step": 925
    },
    {
      "epoch": 2.1839622641509435,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.003623188405797e-05,
      "loss": 0.1603,
      "step": 926
    },
    {
      "epoch": 2.186320754716981,
      "grad_norm": 0.2099609375,
      "learning_rate": 4.002415458937198e-05,
      "loss": 0.1712,
      "step": 927
    },
    {
      "epoch": 2.188679245283019,
      "grad_norm": 0.1875,
      "learning_rate": 4.001207729468599e-05,
      "loss": 0.1966,
      "step": 928
    },
    {
      "epoch": 2.1910377358490565,
      "grad_norm": 0.1943359375,
      "learning_rate": 4e-05,
      "loss": 0.1698,
      "step": 929
    },
    {
      "epoch": 2.1933962264150946,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.998792270531401e-05,
      "loss": 0.1401,
      "step": 930
    },
    {
      "epoch": 2.1957547169811322,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.997584541062802e-05,
      "loss": 0.1628,
      "step": 931
    },
    {
      "epoch": 2.19811320754717,
      "grad_norm": 0.205078125,
      "learning_rate": 3.996376811594203e-05,
      "loss": 0.1943,
      "step": 932
    },
    {
      "epoch": 2.2004716981132075,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.995169082125604e-05,
      "loss": 0.211,
      "step": 933
    },
    {
      "epoch": 2.202830188679245,
      "grad_norm": 0.17578125,
      "learning_rate": 3.9939613526570055e-05,
      "loss": 0.2247,
      "step": 934
    },
    {
      "epoch": 2.205188679245283,
      "grad_norm": 0.19140625,
      "learning_rate": 3.9927536231884064e-05,
      "loss": 0.1668,
      "step": 935
    },
    {
      "epoch": 2.207547169811321,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.991545893719807e-05,
      "loss": 0.2115,
      "step": 936
    },
    {
      "epoch": 2.2099056603773586,
      "grad_norm": 0.1796875,
      "learning_rate": 3.9903381642512075e-05,
      "loss": 0.1576,
      "step": 937
    },
    {
      "epoch": 2.2122641509433962,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.9891304347826083e-05,
      "loss": 0.1442,
      "step": 938
    },
    {
      "epoch": 2.214622641509434,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.98792270531401e-05,
      "loss": 0.1752,
      "step": 939
    },
    {
      "epoch": 2.2169811320754715,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.986714975845411e-05,
      "loss": 0.1627,
      "step": 940
    },
    {
      "epoch": 2.2193396226415096,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.985507246376812e-05,
      "loss": 0.1738,
      "step": 941
    },
    {
      "epoch": 2.2216981132075473,
      "grad_norm": 0.185546875,
      "learning_rate": 3.9842995169082126e-05,
      "loss": 0.2142,
      "step": 942
    },
    {
      "epoch": 2.224056603773585,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.983091787439614e-05,
      "loss": 0.1611,
      "step": 943
    },
    {
      "epoch": 2.2264150943396226,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.981884057971015e-05,
      "loss": 0.2196,
      "step": 944
    },
    {
      "epoch": 2.2287735849056602,
      "grad_norm": 0.17578125,
      "learning_rate": 3.980676328502416e-05,
      "loss": 0.1625,
      "step": 945
    },
    {
      "epoch": 2.231132075471698,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.979468599033817e-05,
      "loss": 0.2,
      "step": 946
    },
    {
      "epoch": 2.233490566037736,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.978260869565217e-05,
      "loss": 0.2023,
      "step": 947
    },
    {
      "epoch": 2.2358490566037736,
      "grad_norm": 0.201171875,
      "learning_rate": 3.9770531400966186e-05,
      "loss": 0.2394,
      "step": 948
    },
    {
      "epoch": 2.2382075471698113,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.9758454106280195e-05,
      "loss": 0.1778,
      "step": 949
    },
    {
      "epoch": 2.240566037735849,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.9746376811594204e-05,
      "loss": 0.1623,
      "step": 950
    },
    {
      "epoch": 2.2429245283018866,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.973429951690821e-05,
      "loss": 0.1717,
      "step": 951
    },
    {
      "epoch": 2.2452830188679247,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.972222222222222e-05,
      "loss": 0.1812,
      "step": 952
    },
    {
      "epoch": 2.2476415094339623,
      "grad_norm": 0.2255859375,
      "learning_rate": 3.971014492753624e-05,
      "loss": 0.2304,
      "step": 953
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.20703125,
      "learning_rate": 3.9698067632850246e-05,
      "loss": 0.2065,
      "step": 954
    },
    {
      "epoch": 2.2523584905660377,
      "grad_norm": 0.20703125,
      "learning_rate": 3.9685990338164255e-05,
      "loss": 0.1719,
      "step": 955
    },
    {
      "epoch": 2.2547169811320753,
      "grad_norm": 0.18359375,
      "learning_rate": 3.9673913043478264e-05,
      "loss": 0.1461,
      "step": 956
    },
    {
      "epoch": 2.2570754716981134,
      "grad_norm": 0.189453125,
      "learning_rate": 3.966183574879227e-05,
      "loss": 0.1944,
      "step": 957
    },
    {
      "epoch": 2.259433962264151,
      "grad_norm": 0.2080078125,
      "learning_rate": 3.964975845410628e-05,
      "loss": 0.2106,
      "step": 958
    },
    {
      "epoch": 2.2617924528301887,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.963768115942029e-05,
      "loss": 0.191,
      "step": 959
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.96256038647343e-05,
      "loss": 0.1871,
      "step": 960
    },
    {
      "epoch": 2.266509433962264,
      "grad_norm": 0.193359375,
      "learning_rate": 3.961352657004831e-05,
      "loss": 0.2197,
      "step": 961
    },
    {
      "epoch": 2.268867924528302,
      "grad_norm": 0.201171875,
      "learning_rate": 3.9601449275362324e-05,
      "loss": 0.2007,
      "step": 962
    },
    {
      "epoch": 2.2712264150943398,
      "grad_norm": 0.19140625,
      "learning_rate": 3.958937198067633e-05,
      "loss": 0.1725,
      "step": 963
    },
    {
      "epoch": 2.2735849056603774,
      "grad_norm": 0.173828125,
      "learning_rate": 3.957729468599034e-05,
      "loss": 0.2017,
      "step": 964
    },
    {
      "epoch": 2.275943396226415,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.956521739130435e-05,
      "loss": 0.1637,
      "step": 965
    },
    {
      "epoch": 2.2783018867924527,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.955314009661836e-05,
      "loss": 0.2261,
      "step": 966
    },
    {
      "epoch": 2.2806603773584904,
      "grad_norm": 0.20703125,
      "learning_rate": 3.954106280193237e-05,
      "loss": 0.2283,
      "step": 967
    },
    {
      "epoch": 2.2830188679245285,
      "grad_norm": 0.197265625,
      "learning_rate": 3.952898550724638e-05,
      "loss": 0.2031,
      "step": 968
    },
    {
      "epoch": 2.285377358490566,
      "grad_norm": 0.1875,
      "learning_rate": 3.951690821256039e-05,
      "loss": 0.1723,
      "step": 969
    },
    {
      "epoch": 2.2877358490566038,
      "grad_norm": 0.181640625,
      "learning_rate": 3.9504830917874396e-05,
      "loss": 0.1876,
      "step": 970
    },
    {
      "epoch": 2.2900943396226414,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.9492753623188405e-05,
      "loss": 0.245,
      "step": 971
    },
    {
      "epoch": 2.292452830188679,
      "grad_norm": 0.19921875,
      "learning_rate": 3.948067632850242e-05,
      "loss": 0.2104,
      "step": 972
    },
    {
      "epoch": 2.294811320754717,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.946859903381643e-05,
      "loss": 0.1803,
      "step": 973
    },
    {
      "epoch": 2.297169811320755,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.945652173913044e-05,
      "loss": 0.175,
      "step": 974
    },
    {
      "epoch": 2.2995283018867925,
      "grad_norm": 0.212890625,
      "learning_rate": 3.944444444444445e-05,
      "loss": 0.22,
      "step": 975
    },
    {
      "epoch": 2.30188679245283,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.9432367149758456e-05,
      "loss": 0.152,
      "step": 976
    },
    {
      "epoch": 2.3042452830188678,
      "grad_norm": 0.19921875,
      "learning_rate": 3.9420289855072465e-05,
      "loss": 0.1653,
      "step": 977
    },
    {
      "epoch": 2.3066037735849054,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.9408212560386474e-05,
      "loss": 0.1997,
      "step": 978
    },
    {
      "epoch": 2.3089622641509435,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.939613526570048e-05,
      "loss": 0.1874,
      "step": 979
    },
    {
      "epoch": 2.311320754716981,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.938405797101449e-05,
      "loss": 0.1651,
      "step": 980
    },
    {
      "epoch": 2.313679245283019,
      "grad_norm": 0.18359375,
      "learning_rate": 3.937198067632851e-05,
      "loss": 0.1969,
      "step": 981
    },
    {
      "epoch": 2.3160377358490565,
      "grad_norm": 0.181640625,
      "learning_rate": 3.9359903381642516e-05,
      "loss": 0.1625,
      "step": 982
    },
    {
      "epoch": 2.3183962264150946,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.9347826086956525e-05,
      "loss": 0.1762,
      "step": 983
    },
    {
      "epoch": 2.3207547169811322,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.9335748792270534e-05,
      "loss": 0.2141,
      "step": 984
    },
    {
      "epoch": 2.32311320754717,
      "grad_norm": 0.18359375,
      "learning_rate": 3.932367149758454e-05,
      "loss": 0.1638,
      "step": 985
    },
    {
      "epoch": 2.3254716981132075,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.931159420289855e-05,
      "loss": 0.1823,
      "step": 986
    },
    {
      "epoch": 2.327830188679245,
      "grad_norm": 0.18359375,
      "learning_rate": 3.929951690821256e-05,
      "loss": 0.2013,
      "step": 987
    },
    {
      "epoch": 2.330188679245283,
      "grad_norm": 0.203125,
      "learning_rate": 3.928743961352657e-05,
      "loss": 0.1541,
      "step": 988
    },
    {
      "epoch": 2.332547169811321,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.927536231884058e-05,
      "loss": 0.1919,
      "step": 989
    },
    {
      "epoch": 2.3349056603773586,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.926328502415459e-05,
      "loss": 0.2139,
      "step": 990
    },
    {
      "epoch": 2.3372641509433962,
      "grad_norm": 0.193359375,
      "learning_rate": 3.92512077294686e-05,
      "loss": 0.1868,
      "step": 991
    },
    {
      "epoch": 2.339622641509434,
      "grad_norm": 0.23828125,
      "learning_rate": 3.923913043478261e-05,
      "loss": 0.1597,
      "step": 992
    },
    {
      "epoch": 2.3419811320754715,
      "grad_norm": 0.189453125,
      "learning_rate": 3.922705314009662e-05,
      "loss": 0.1834,
      "step": 993
    },
    {
      "epoch": 2.3443396226415096,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.921497584541063e-05,
      "loss": 0.1888,
      "step": 994
    },
    {
      "epoch": 2.3466981132075473,
      "grad_norm": 0.20703125,
      "learning_rate": 3.920289855072464e-05,
      "loss": 0.1685,
      "step": 995
    },
    {
      "epoch": 2.349056603773585,
      "grad_norm": 0.2060546875,
      "learning_rate": 3.9190821256038654e-05,
      "loss": 0.178,
      "step": 996
    },
    {
      "epoch": 2.3514150943396226,
      "grad_norm": 0.193359375,
      "learning_rate": 3.9178743961352657e-05,
      "loss": 0.1286,
      "step": 997
    },
    {
      "epoch": 2.3537735849056602,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 0.1988,
      "step": 998
    },
    {
      "epoch": 2.356132075471698,
      "grad_norm": 0.208984375,
      "learning_rate": 3.9154589371980674e-05,
      "loss": 0.1895,
      "step": 999
    },
    {
      "epoch": 2.358490566037736,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.914251207729468e-05,
      "loss": 0.2011,
      "step": 1000
    },
    {
      "epoch": 2.3608490566037736,
      "grad_norm": 0.189453125,
      "learning_rate": 3.91304347826087e-05,
      "loss": 0.1247,
      "step": 1001
    },
    {
      "epoch": 2.3632075471698113,
      "grad_norm": 0.2109375,
      "learning_rate": 3.911835748792271e-05,
      "loss": 0.2061,
      "step": 1002
    },
    {
      "epoch": 2.365566037735849,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.910628019323672e-05,
      "loss": 0.1819,
      "step": 1003
    },
    {
      "epoch": 2.3679245283018866,
      "grad_norm": 0.224609375,
      "learning_rate": 3.9094202898550726e-05,
      "loss": 0.1506,
      "step": 1004
    },
    {
      "epoch": 2.3702830188679247,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.908212560386474e-05,
      "loss": 0.1879,
      "step": 1005
    },
    {
      "epoch": 2.3726415094339623,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.907004830917875e-05,
      "loss": 0.2187,
      "step": 1006
    },
    {
      "epoch": 2.375,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.905797101449275e-05,
      "loss": 0.1355,
      "step": 1007
    },
    {
      "epoch": 2.3773584905660377,
      "grad_norm": 0.1953125,
      "learning_rate": 3.904589371980676e-05,
      "loss": 0.2423,
      "step": 1008
    },
    {
      "epoch": 2.3797169811320753,
      "grad_norm": 0.2060546875,
      "learning_rate": 3.903381642512077e-05,
      "loss": 0.1776,
      "step": 1009
    },
    {
      "epoch": 2.3820754716981134,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.9021739130434786e-05,
      "loss": 0.2058,
      "step": 1010
    },
    {
      "epoch": 2.384433962264151,
      "grad_norm": 0.15625,
      "learning_rate": 3.9009661835748795e-05,
      "loss": 0.2455,
      "step": 1011
    },
    {
      "epoch": 2.3867924528301887,
      "grad_norm": 0.212890625,
      "learning_rate": 3.8997584541062804e-05,
      "loss": 0.2017,
      "step": 1012
    },
    {
      "epoch": 2.3891509433962264,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.898550724637681e-05,
      "loss": 0.1582,
      "step": 1013
    },
    {
      "epoch": 2.391509433962264,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.897342995169082e-05,
      "loss": 0.2063,
      "step": 1014
    },
    {
      "epoch": 2.393867924528302,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.896135265700484e-05,
      "loss": 0.1723,
      "step": 1015
    },
    {
      "epoch": 2.3962264150943398,
      "grad_norm": 0.19921875,
      "learning_rate": 3.8949275362318846e-05,
      "loss": 0.2177,
      "step": 1016
    },
    {
      "epoch": 2.3985849056603774,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.8937198067632855e-05,
      "loss": 0.1976,
      "step": 1017
    },
    {
      "epoch": 2.400943396226415,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.892512077294686e-05,
      "loss": 0.2074,
      "step": 1018
    },
    {
      "epoch": 2.4033018867924527,
      "grad_norm": 0.189453125,
      "learning_rate": 3.8913043478260866e-05,
      "loss": 0.1525,
      "step": 1019
    },
    {
      "epoch": 2.4056603773584904,
      "grad_norm": 0.19140625,
      "learning_rate": 3.890096618357488e-05,
      "loss": 0.1773,
      "step": 1020
    },
    {
      "epoch": 2.4080188679245285,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.1383,
      "step": 1021
    },
    {
      "epoch": 2.410377358490566,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.88768115942029e-05,
      "loss": 0.2056,
      "step": 1022
    },
    {
      "epoch": 2.4127358490566038,
      "grad_norm": 0.18359375,
      "learning_rate": 3.886473429951691e-05,
      "loss": 0.212,
      "step": 1023
    },
    {
      "epoch": 2.4150943396226414,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.8852657004830924e-05,
      "loss": 0.2115,
      "step": 1024
    },
    {
      "epoch": 2.417452830188679,
      "grad_norm": 0.189453125,
      "learning_rate": 3.884057971014493e-05,
      "loss": 0.1863,
      "step": 1025
    },
    {
      "epoch": 2.419811320754717,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.882850241545894e-05,
      "loss": 0.1729,
      "step": 1026
    },
    {
      "epoch": 2.422169811320755,
      "grad_norm": 0.181640625,
      "learning_rate": 3.881642512077295e-05,
      "loss": 0.2195,
      "step": 1027
    },
    {
      "epoch": 2.4245283018867925,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.880434782608695e-05,
      "loss": 0.2295,
      "step": 1028
    },
    {
      "epoch": 2.42688679245283,
      "grad_norm": 0.189453125,
      "learning_rate": 3.879227053140097e-05,
      "loss": 0.154,
      "step": 1029
    },
    {
      "epoch": 2.4292452830188678,
      "grad_norm": 0.201171875,
      "learning_rate": 3.878019323671498e-05,
      "loss": 0.1936,
      "step": 1030
    },
    {
      "epoch": 2.4316037735849054,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.876811594202899e-05,
      "loss": 0.1821,
      "step": 1031
    },
    {
      "epoch": 2.4339622641509435,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.8756038647342996e-05,
      "loss": 0.1862,
      "step": 1032
    },
    {
      "epoch": 2.436320754716981,
      "grad_norm": 0.1875,
      "learning_rate": 3.8743961352657004e-05,
      "loss": 0.1795,
      "step": 1033
    },
    {
      "epoch": 2.438679245283019,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.873188405797102e-05,
      "loss": 0.1741,
      "step": 1034
    },
    {
      "epoch": 2.4410377358490565,
      "grad_norm": 0.18359375,
      "learning_rate": 3.871980676328503e-05,
      "loss": 0.2044,
      "step": 1035
    },
    {
      "epoch": 2.4433962264150946,
      "grad_norm": 0.166015625,
      "learning_rate": 3.870772946859904e-05,
      "loss": 0.2205,
      "step": 1036
    },
    {
      "epoch": 2.4457547169811322,
      "grad_norm": 0.1494140625,
      "learning_rate": 3.869565217391305e-05,
      "loss": 0.2513,
      "step": 1037
    },
    {
      "epoch": 2.44811320754717,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.8683574879227056e-05,
      "loss": 0.2266,
      "step": 1038
    },
    {
      "epoch": 2.4504716981132075,
      "grad_norm": 0.18359375,
      "learning_rate": 3.8671497584541065e-05,
      "loss": 0.1408,
      "step": 1039
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 0.1953125,
      "learning_rate": 3.8659420289855074e-05,
      "loss": 0.2024,
      "step": 1040
    },
    {
      "epoch": 2.455188679245283,
      "grad_norm": 0.1875,
      "learning_rate": 3.864734299516908e-05,
      "loss": 0.2021,
      "step": 1041
    },
    {
      "epoch": 2.457547169811321,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.863526570048309e-05,
      "loss": 0.1989,
      "step": 1042
    },
    {
      "epoch": 2.4599056603773586,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.862318840579711e-05,
      "loss": 0.1857,
      "step": 1043
    },
    {
      "epoch": 2.4622641509433962,
      "grad_norm": 0.19921875,
      "learning_rate": 3.8611111111111116e-05,
      "loss": 0.2184,
      "step": 1044
    },
    {
      "epoch": 2.464622641509434,
      "grad_norm": 0.1875,
      "learning_rate": 3.8599033816425125e-05,
      "loss": 0.2195,
      "step": 1045
    },
    {
      "epoch": 2.4669811320754715,
      "grad_norm": 0.1875,
      "learning_rate": 3.8586956521739134e-05,
      "loss": 0.2195,
      "step": 1046
    },
    {
      "epoch": 2.4693396226415096,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.857487922705314e-05,
      "loss": 0.1646,
      "step": 1047
    },
    {
      "epoch": 2.4716981132075473,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.856280193236715e-05,
      "loss": 0.1978,
      "step": 1048
    },
    {
      "epoch": 2.474056603773585,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.855072463768116e-05,
      "loss": 0.1261,
      "step": 1049
    },
    {
      "epoch": 2.4764150943396226,
      "grad_norm": 0.19140625,
      "learning_rate": 3.853864734299517e-05,
      "loss": 0.2194,
      "step": 1050
    },
    {
      "epoch": 2.4787735849056602,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.852657004830918e-05,
      "loss": 0.2035,
      "step": 1051
    },
    {
      "epoch": 2.481132075471698,
      "grad_norm": 0.1875,
      "learning_rate": 3.851449275362319e-05,
      "loss": 0.2113,
      "step": 1052
    },
    {
      "epoch": 2.483490566037736,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.85024154589372e-05,
      "loss": 0.1497,
      "step": 1053
    },
    {
      "epoch": 2.4858490566037736,
      "grad_norm": 0.1875,
      "learning_rate": 3.849033816425121e-05,
      "loss": 0.2233,
      "step": 1054
    },
    {
      "epoch": 2.4882075471698113,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.847826086956522e-05,
      "loss": 0.2332,
      "step": 1055
    },
    {
      "epoch": 2.490566037735849,
      "grad_norm": 0.185546875,
      "learning_rate": 3.846618357487923e-05,
      "loss": 0.1666,
      "step": 1056
    },
    {
      "epoch": 2.4929245283018866,
      "grad_norm": 0.20703125,
      "learning_rate": 3.845410628019324e-05,
      "loss": 0.2022,
      "step": 1057
    },
    {
      "epoch": 2.4952830188679247,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.844202898550725e-05,
      "loss": 0.1991,
      "step": 1058
    },
    {
      "epoch": 2.4976415094339623,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.8429951690821256e-05,
      "loss": 0.1872,
      "step": 1059
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.1708984375,
      "learning_rate": 3.8417874396135265e-05,
      "loss": 0.1632,
      "step": 1060
    },
    {
      "epoch": 2.5023584905660377,
      "grad_norm": 0.310546875,
      "learning_rate": 3.8405797101449274e-05,
      "loss": 0.2266,
      "step": 1061
    },
    {
      "epoch": 2.5047169811320753,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.839371980676328e-05,
      "loss": 0.1728,
      "step": 1062
    },
    {
      "epoch": 2.507075471698113,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.83816425120773e-05,
      "loss": 0.1447,
      "step": 1063
    },
    {
      "epoch": 2.509433962264151,
      "grad_norm": 0.18359375,
      "learning_rate": 3.836956521739131e-05,
      "loss": 0.1357,
      "step": 1064
    },
    {
      "epoch": 2.5117924528301887,
      "grad_norm": 0.177734375,
      "learning_rate": 3.835748792270532e-05,
      "loss": 0.2192,
      "step": 1065
    },
    {
      "epoch": 2.5141509433962264,
      "grad_norm": 0.19140625,
      "learning_rate": 3.8345410628019326e-05,
      "loss": 0.1527,
      "step": 1066
    },
    {
      "epoch": 2.516509433962264,
      "grad_norm": 0.177734375,
      "learning_rate": 3.8333333333333334e-05,
      "loss": 0.1365,
      "step": 1067
    },
    {
      "epoch": 2.518867924528302,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.8321256038647343e-05,
      "loss": 0.1819,
      "step": 1068
    },
    {
      "epoch": 2.5212264150943398,
      "grad_norm": 0.197265625,
      "learning_rate": 3.830917874396135e-05,
      "loss": 0.1597,
      "step": 1069
    },
    {
      "epoch": 2.5235849056603774,
      "grad_norm": 0.17578125,
      "learning_rate": 3.829710144927536e-05,
      "loss": 0.173,
      "step": 1070
    },
    {
      "epoch": 2.525943396226415,
      "grad_norm": 0.24609375,
      "learning_rate": 3.828502415458937e-05,
      "loss": 0.1972,
      "step": 1071
    },
    {
      "epoch": 2.5283018867924527,
      "grad_norm": 0.166015625,
      "learning_rate": 3.8272946859903386e-05,
      "loss": 0.1766,
      "step": 1072
    },
    {
      "epoch": 2.5306603773584904,
      "grad_norm": 0.197265625,
      "learning_rate": 3.8260869565217395e-05,
      "loss": 0.1998,
      "step": 1073
    },
    {
      "epoch": 2.5330188679245285,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.8248792270531404e-05,
      "loss": 0.1345,
      "step": 1074
    },
    {
      "epoch": 2.535377358490566,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.823671497584541e-05,
      "loss": 0.1899,
      "step": 1075
    },
    {
      "epoch": 2.5377358490566038,
      "grad_norm": 0.1875,
      "learning_rate": 3.822463768115942e-05,
      "loss": 0.1778,
      "step": 1076
    },
    {
      "epoch": 2.5400943396226414,
      "grad_norm": 0.19140625,
      "learning_rate": 3.821256038647344e-05,
      "loss": 0.1904,
      "step": 1077
    },
    {
      "epoch": 2.5424528301886795,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.820048309178744e-05,
      "loss": 0.1916,
      "step": 1078
    },
    {
      "epoch": 2.544811320754717,
      "grad_norm": 0.19140625,
      "learning_rate": 3.818840579710145e-05,
      "loss": 0.1687,
      "step": 1079
    },
    {
      "epoch": 2.547169811320755,
      "grad_norm": 0.197265625,
      "learning_rate": 3.817632850241546e-05,
      "loss": 0.1759,
      "step": 1080
    },
    {
      "epoch": 2.5495283018867925,
      "grad_norm": 0.177734375,
      "learning_rate": 3.8164251207729466e-05,
      "loss": 0.2125,
      "step": 1081
    },
    {
      "epoch": 2.55188679245283,
      "grad_norm": 0.19140625,
      "learning_rate": 3.815217391304348e-05,
      "loss": 0.1455,
      "step": 1082
    },
    {
      "epoch": 2.5542452830188678,
      "grad_norm": 0.18359375,
      "learning_rate": 3.814009661835749e-05,
      "loss": 0.1759,
      "step": 1083
    },
    {
      "epoch": 2.5566037735849054,
      "grad_norm": 0.21484375,
      "learning_rate": 3.81280193236715e-05,
      "loss": 0.1886,
      "step": 1084
    },
    {
      "epoch": 2.5589622641509435,
      "grad_norm": 0.181640625,
      "learning_rate": 3.811594202898551e-05,
      "loss": 0.1697,
      "step": 1085
    },
    {
      "epoch": 2.561320754716981,
      "grad_norm": 0.177734375,
      "learning_rate": 3.8103864734299524e-05,
      "loss": 0.2173,
      "step": 1086
    },
    {
      "epoch": 2.563679245283019,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.809178743961353e-05,
      "loss": 0.1971,
      "step": 1087
    },
    {
      "epoch": 2.5660377358490565,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.8079710144927535e-05,
      "loss": 0.1928,
      "step": 1088
    },
    {
      "epoch": 2.5683962264150946,
      "grad_norm": 0.189453125,
      "learning_rate": 3.8067632850241544e-05,
      "loss": 0.2107,
      "step": 1089
    },
    {
      "epoch": 2.5707547169811322,
      "grad_norm": 0.193359375,
      "learning_rate": 3.805555555555555e-05,
      "loss": 0.1423,
      "step": 1090
    },
    {
      "epoch": 2.57311320754717,
      "grad_norm": 0.1796875,
      "learning_rate": 3.804347826086957e-05,
      "loss": 0.1914,
      "step": 1091
    },
    {
      "epoch": 2.5754716981132075,
      "grad_norm": 0.185546875,
      "learning_rate": 3.803140096618358e-05,
      "loss": 0.2311,
      "step": 1092
    },
    {
      "epoch": 2.577830188679245,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.8019323671497586e-05,
      "loss": 0.2126,
      "step": 1093
    },
    {
      "epoch": 2.580188679245283,
      "grad_norm": 0.181640625,
      "learning_rate": 3.8007246376811595e-05,
      "loss": 0.1811,
      "step": 1094
    },
    {
      "epoch": 2.5825471698113205,
      "grad_norm": 0.185546875,
      "learning_rate": 3.7995169082125604e-05,
      "loss": 0.1393,
      "step": 1095
    },
    {
      "epoch": 2.5849056603773586,
      "grad_norm": 0.197265625,
      "learning_rate": 3.798309178743962e-05,
      "loss": 0.1603,
      "step": 1096
    },
    {
      "epoch": 2.5872641509433962,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.797101449275363e-05,
      "loss": 0.2051,
      "step": 1097
    },
    {
      "epoch": 2.589622641509434,
      "grad_norm": 0.1796875,
      "learning_rate": 3.795893719806764e-05,
      "loss": 0.2001,
      "step": 1098
    },
    {
      "epoch": 2.5919811320754715,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.794685990338164e-05,
      "loss": 0.2188,
      "step": 1099
    },
    {
      "epoch": 2.5943396226415096,
      "grad_norm": 0.19921875,
      "learning_rate": 3.793478260869565e-05,
      "loss": 0.2174,
      "step": 1100
    },
    {
      "epoch": 2.5966981132075473,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.7922705314009665e-05,
      "loss": 0.166,
      "step": 1101
    },
    {
      "epoch": 2.599056603773585,
      "grad_norm": 0.185546875,
      "learning_rate": 3.7910628019323673e-05,
      "loss": 0.1941,
      "step": 1102
    },
    {
      "epoch": 2.6014150943396226,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.789855072463768e-05,
      "loss": 0.1978,
      "step": 1103
    },
    {
      "epoch": 2.6037735849056602,
      "grad_norm": 0.20703125,
      "learning_rate": 3.788647342995169e-05,
      "loss": 0.1516,
      "step": 1104
    },
    {
      "epoch": 2.606132075471698,
      "grad_norm": 0.177734375,
      "learning_rate": 3.787439613526571e-05,
      "loss": 0.2202,
      "step": 1105
    },
    {
      "epoch": 2.608490566037736,
      "grad_norm": 0.18359375,
      "learning_rate": 3.7862318840579716e-05,
      "loss": 0.1613,
      "step": 1106
    },
    {
      "epoch": 2.6108490566037736,
      "grad_norm": 0.189453125,
      "learning_rate": 3.7850241545893725e-05,
      "loss": 0.1819,
      "step": 1107
    },
    {
      "epoch": 2.6132075471698113,
      "grad_norm": 0.185546875,
      "learning_rate": 3.7838164251207734e-05,
      "loss": 0.1657,
      "step": 1108
    },
    {
      "epoch": 2.615566037735849,
      "grad_norm": 0.177734375,
      "learning_rate": 3.7826086956521736e-05,
      "loss": 0.1895,
      "step": 1109
    },
    {
      "epoch": 2.617924528301887,
      "grad_norm": 0.173828125,
      "learning_rate": 3.781400966183575e-05,
      "loss": 0.1536,
      "step": 1110
    },
    {
      "epoch": 2.6202830188679247,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.780193236714976e-05,
      "loss": 0.1289,
      "step": 1111
    },
    {
      "epoch": 2.6226415094339623,
      "grad_norm": 0.185546875,
      "learning_rate": 3.778985507246377e-05,
      "loss": 0.158,
      "step": 1112
    },
    {
      "epoch": 2.625,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.777777777777778e-05,
      "loss": 0.1955,
      "step": 1113
    },
    {
      "epoch": 2.6273584905660377,
      "grad_norm": 0.205078125,
      "learning_rate": 3.776570048309179e-05,
      "loss": 0.1632,
      "step": 1114
    },
    {
      "epoch": 2.6297169811320753,
      "grad_norm": 0.185546875,
      "learning_rate": 3.77536231884058e-05,
      "loss": 0.1701,
      "step": 1115
    },
    {
      "epoch": 2.632075471698113,
      "grad_norm": 0.1953125,
      "learning_rate": 3.774154589371981e-05,
      "loss": 0.1332,
      "step": 1116
    },
    {
      "epoch": 2.634433962264151,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.772946859903382e-05,
      "loss": 0.1832,
      "step": 1117
    },
    {
      "epoch": 2.6367924528301887,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.771739130434783e-05,
      "loss": 0.194,
      "step": 1118
    },
    {
      "epoch": 2.6391509433962264,
      "grad_norm": 0.2109375,
      "learning_rate": 3.770531400966184e-05,
      "loss": 0.1961,
      "step": 1119
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 0.2060546875,
      "learning_rate": 3.769323671497585e-05,
      "loss": 0.2415,
      "step": 1120
    },
    {
      "epoch": 2.643867924528302,
      "grad_norm": 0.177734375,
      "learning_rate": 3.7681159420289856e-05,
      "loss": 0.1738,
      "step": 1121
    },
    {
      "epoch": 2.6462264150943398,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.7669082125603865e-05,
      "loss": 0.1866,
      "step": 1122
    },
    {
      "epoch": 2.6485849056603774,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.7657004830917874e-05,
      "loss": 0.1936,
      "step": 1123
    },
    {
      "epoch": 2.650943396226415,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.764492753623188e-05,
      "loss": 0.1426,
      "step": 1124
    },
    {
      "epoch": 2.6533018867924527,
      "grad_norm": 0.201171875,
      "learning_rate": 3.76328502415459e-05,
      "loss": 0.1443,
      "step": 1125
    },
    {
      "epoch": 2.6556603773584904,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.762077294685991e-05,
      "loss": 0.1687,
      "step": 1126
    },
    {
      "epoch": 2.6580188679245285,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.7608695652173917e-05,
      "loss": 0.1995,
      "step": 1127
    },
    {
      "epoch": 2.660377358490566,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.7596618357487925e-05,
      "loss": 0.2061,
      "step": 1128
    },
    {
      "epoch": 2.6627358490566038,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.7584541062801934e-05,
      "loss": 0.2042,
      "step": 1129
    },
    {
      "epoch": 2.6650943396226414,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.757246376811594e-05,
      "loss": 0.2026,
      "step": 1130
    },
    {
      "epoch": 2.6674528301886795,
      "grad_norm": 0.212890625,
      "learning_rate": 3.756038647342995e-05,
      "loss": 0.2111,
      "step": 1131
    },
    {
      "epoch": 2.669811320754717,
      "grad_norm": 0.25,
      "learning_rate": 3.754830917874396e-05,
      "loss": 0.2498,
      "step": 1132
    },
    {
      "epoch": 2.672169811320755,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.753623188405797e-05,
      "loss": 0.1967,
      "step": 1133
    },
    {
      "epoch": 2.6745283018867925,
      "grad_norm": 0.189453125,
      "learning_rate": 3.7524154589371986e-05,
      "loss": 0.1896,
      "step": 1134
    },
    {
      "epoch": 2.67688679245283,
      "grad_norm": 0.18359375,
      "learning_rate": 3.7512077294685995e-05,
      "loss": 0.2118,
      "step": 1135
    },
    {
      "epoch": 2.6792452830188678,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.1803,
      "step": 1136
    },
    {
      "epoch": 2.6816037735849054,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.748792270531401e-05,
      "loss": 0.1656,
      "step": 1137
    },
    {
      "epoch": 2.6839622641509435,
      "grad_norm": 0.185546875,
      "learning_rate": 3.747584541062802e-05,
      "loss": 0.1922,
      "step": 1138
    },
    {
      "epoch": 2.686320754716981,
      "grad_norm": 0.1796875,
      "learning_rate": 3.746376811594203e-05,
      "loss": 0.1798,
      "step": 1139
    },
    {
      "epoch": 2.688679245283019,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.745169082125604e-05,
      "loss": 0.2159,
      "step": 1140
    },
    {
      "epoch": 2.6910377358490565,
      "grad_norm": 0.181640625,
      "learning_rate": 3.743961352657005e-05,
      "loss": 0.2046,
      "step": 1141
    },
    {
      "epoch": 2.6933962264150946,
      "grad_norm": 0.181640625,
      "learning_rate": 3.742753623188406e-05,
      "loss": 0.1221,
      "step": 1142
    },
    {
      "epoch": 2.6957547169811322,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.7415458937198066e-05,
      "loss": 0.2146,
      "step": 1143
    },
    {
      "epoch": 2.69811320754717,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.740338164251208e-05,
      "loss": 0.2348,
      "step": 1144
    },
    {
      "epoch": 2.7004716981132075,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.739130434782609e-05,
      "loss": 0.2273,
      "step": 1145
    },
    {
      "epoch": 2.702830188679245,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.73792270531401e-05,
      "loss": 0.1446,
      "step": 1146
    },
    {
      "epoch": 2.705188679245283,
      "grad_norm": 0.17578125,
      "learning_rate": 3.736714975845411e-05,
      "loss": 0.1826,
      "step": 1147
    },
    {
      "epoch": 2.7075471698113205,
      "grad_norm": 0.177734375,
      "learning_rate": 3.735507246376812e-05,
      "loss": 0.2079,
      "step": 1148
    },
    {
      "epoch": 2.7099056603773586,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.7342995169082126e-05,
      "loss": 0.1618,
      "step": 1149
    },
    {
      "epoch": 2.7122641509433962,
      "grad_norm": 0.201171875,
      "learning_rate": 3.7330917874396135e-05,
      "loss": 0.1877,
      "step": 1150
    },
    {
      "epoch": 2.714622641509434,
      "grad_norm": 0.19921875,
      "learning_rate": 3.7318840579710144e-05,
      "loss": 0.1798,
      "step": 1151
    },
    {
      "epoch": 2.7169811320754715,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.730676328502415e-05,
      "loss": 0.1939,
      "step": 1152
    },
    {
      "epoch": 2.7193396226415096,
      "grad_norm": 0.193359375,
      "learning_rate": 3.729468599033817e-05,
      "loss": 0.2008,
      "step": 1153
    },
    {
      "epoch": 2.7216981132075473,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.728260869565218e-05,
      "loss": 0.1885,
      "step": 1154
    },
    {
      "epoch": 2.724056603773585,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.7270531400966186e-05,
      "loss": 0.1843,
      "step": 1155
    },
    {
      "epoch": 2.7264150943396226,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.7258454106280195e-05,
      "loss": 0.1905,
      "step": 1156
    },
    {
      "epoch": 2.7287735849056602,
      "grad_norm": 0.189453125,
      "learning_rate": 3.7246376811594204e-05,
      "loss": 0.191,
      "step": 1157
    },
    {
      "epoch": 2.731132075471698,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.723429951690822e-05,
      "loss": 0.2004,
      "step": 1158
    },
    {
      "epoch": 2.733490566037736,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.722222222222222e-05,
      "loss": 0.1479,
      "step": 1159
    },
    {
      "epoch": 2.7358490566037736,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.721014492753623e-05,
      "loss": 0.1937,
      "step": 1160
    },
    {
      "epoch": 2.7382075471698113,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.719806763285024e-05,
      "loss": 0.116,
      "step": 1161
    },
    {
      "epoch": 2.740566037735849,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.718599033816425e-05,
      "loss": 0.1278,
      "step": 1162
    },
    {
      "epoch": 2.742924528301887,
      "grad_norm": 0.185546875,
      "learning_rate": 3.7173913043478264e-05,
      "loss": 0.1988,
      "step": 1163
    },
    {
      "epoch": 2.7452830188679247,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.716183574879227e-05,
      "loss": 0.2031,
      "step": 1164
    },
    {
      "epoch": 2.7476415094339623,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.714975845410628e-05,
      "loss": 0.1813,
      "step": 1165
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.185546875,
      "learning_rate": 3.713768115942029e-05,
      "loss": 0.17,
      "step": 1166
    },
    {
      "epoch": 2.7523584905660377,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.71256038647343e-05,
      "loss": 0.1505,
      "step": 1167
    },
    {
      "epoch": 2.7547169811320753,
      "grad_norm": 0.189453125,
      "learning_rate": 3.7113526570048316e-05,
      "loss": 0.2367,
      "step": 1168
    },
    {
      "epoch": 2.757075471698113,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.710144927536232e-05,
      "loss": 0.1698,
      "step": 1169
    },
    {
      "epoch": 2.759433962264151,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.708937198067633e-05,
      "loss": 0.2041,
      "step": 1170
    },
    {
      "epoch": 2.7617924528301887,
      "grad_norm": 0.193359375,
      "learning_rate": 3.7077294685990336e-05,
      "loss": 0.1773,
      "step": 1171
    },
    {
      "epoch": 2.7641509433962264,
      "grad_norm": 0.19140625,
      "learning_rate": 3.706521739130435e-05,
      "loss": 0.192,
      "step": 1172
    },
    {
      "epoch": 2.766509433962264,
      "grad_norm": 0.19921875,
      "learning_rate": 3.705314009661836e-05,
      "loss": 0.1752,
      "step": 1173
    },
    {
      "epoch": 2.768867924528302,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.704106280193237e-05,
      "loss": 0.18,
      "step": 1174
    },
    {
      "epoch": 2.7712264150943398,
      "grad_norm": 0.171875,
      "learning_rate": 3.702898550724638e-05,
      "loss": 0.1704,
      "step": 1175
    },
    {
      "epoch": 2.7735849056603774,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.701690821256039e-05,
      "loss": 0.163,
      "step": 1176
    },
    {
      "epoch": 2.775943396226415,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.70048309178744e-05,
      "loss": 0.1524,
      "step": 1177
    },
    {
      "epoch": 2.7783018867924527,
      "grad_norm": 0.193359375,
      "learning_rate": 3.699275362318841e-05,
      "loss": 0.222,
      "step": 1178
    },
    {
      "epoch": 2.7806603773584904,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.698067632850242e-05,
      "loss": 0.2208,
      "step": 1179
    },
    {
      "epoch": 2.7830188679245285,
      "grad_norm": 0.1875,
      "learning_rate": 3.696859903381642e-05,
      "loss": 0.1686,
      "step": 1180
    },
    {
      "epoch": 2.785377358490566,
      "grad_norm": 0.1953125,
      "learning_rate": 3.695652173913043e-05,
      "loss": 0.163,
      "step": 1181
    },
    {
      "epoch": 2.7877358490566038,
      "grad_norm": 0.181640625,
      "learning_rate": 3.694444444444445e-05,
      "loss": 0.1728,
      "step": 1182
    },
    {
      "epoch": 2.7900943396226414,
      "grad_norm": 0.208984375,
      "learning_rate": 3.6932367149758456e-05,
      "loss": 0.2207,
      "step": 1183
    },
    {
      "epoch": 2.7924528301886795,
      "grad_norm": 0.2255859375,
      "learning_rate": 3.6920289855072465e-05,
      "loss": 0.1958,
      "step": 1184
    },
    {
      "epoch": 2.794811320754717,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.6908212560386474e-05,
      "loss": 0.217,
      "step": 1185
    },
    {
      "epoch": 2.797169811320755,
      "grad_norm": 0.19921875,
      "learning_rate": 3.689613526570048e-05,
      "loss": 0.2034,
      "step": 1186
    },
    {
      "epoch": 2.7995283018867925,
      "grad_norm": 0.1748046875,
      "learning_rate": 3.68840579710145e-05,
      "loss": 0.2082,
      "step": 1187
    },
    {
      "epoch": 2.80188679245283,
      "grad_norm": 0.189453125,
      "learning_rate": 3.687198067632851e-05,
      "loss": 0.1625,
      "step": 1188
    },
    {
      "epoch": 2.8042452830188678,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.6859903381642516e-05,
      "loss": 0.1966,
      "step": 1189
    },
    {
      "epoch": 2.8066037735849054,
      "grad_norm": 0.18359375,
      "learning_rate": 3.684782608695652e-05,
      "loss": 0.1426,
      "step": 1190
    },
    {
      "epoch": 2.8089622641509435,
      "grad_norm": 0.181640625,
      "learning_rate": 3.6835748792270534e-05,
      "loss": 0.2127,
      "step": 1191
    },
    {
      "epoch": 2.811320754716981,
      "grad_norm": 0.19921875,
      "learning_rate": 3.682367149758454e-05,
      "loss": 0.197,
      "step": 1192
    },
    {
      "epoch": 2.813679245283019,
      "grad_norm": 0.18359375,
      "learning_rate": 3.681159420289855e-05,
      "loss": 0.1788,
      "step": 1193
    },
    {
      "epoch": 2.8160377358490565,
      "grad_norm": 0.1875,
      "learning_rate": 3.679951690821256e-05,
      "loss": 0.1827,
      "step": 1194
    },
    {
      "epoch": 2.8183962264150946,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.678743961352657e-05,
      "loss": 0.2266,
      "step": 1195
    },
    {
      "epoch": 2.8207547169811322,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.6775362318840586e-05,
      "loss": 0.2276,
      "step": 1196
    },
    {
      "epoch": 2.82311320754717,
      "grad_norm": 0.1767578125,
      "learning_rate": 3.6763285024154594e-05,
      "loss": 0.199,
      "step": 1197
    },
    {
      "epoch": 2.8254716981132075,
      "grad_norm": 0.1875,
      "learning_rate": 3.67512077294686e-05,
      "loss": 0.1716,
      "step": 1198
    },
    {
      "epoch": 2.827830188679245,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.673913043478261e-05,
      "loss": 0.1785,
      "step": 1199
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.672705314009662e-05,
      "loss": 0.1574,
      "step": 1200
    },
    {
      "epoch": 2.8325471698113205,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.671497584541063e-05,
      "loss": 0.2298,
      "step": 1201
    },
    {
      "epoch": 2.8349056603773586,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.670289855072464e-05,
      "loss": 0.1538,
      "step": 1202
    },
    {
      "epoch": 2.8372641509433962,
      "grad_norm": 0.203125,
      "learning_rate": 3.669082125603865e-05,
      "loss": 0.1283,
      "step": 1203
    },
    {
      "epoch": 2.839622641509434,
      "grad_norm": 0.205078125,
      "learning_rate": 3.667874396135266e-05,
      "loss": 0.2051,
      "step": 1204
    },
    {
      "epoch": 2.8419811320754715,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.18,
      "step": 1205
    },
    {
      "epoch": 2.8443396226415096,
      "grad_norm": 0.259765625,
      "learning_rate": 3.665458937198068e-05,
      "loss": 0.1889,
      "step": 1206
    },
    {
      "epoch": 2.8466981132075473,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.664251207729469e-05,
      "loss": 0.2161,
      "step": 1207
    },
    {
      "epoch": 2.849056603773585,
      "grad_norm": 0.1953125,
      "learning_rate": 3.66304347826087e-05,
      "loss": 0.1915,
      "step": 1208
    },
    {
      "epoch": 2.8514150943396226,
      "grad_norm": 0.19921875,
      "learning_rate": 3.661835748792271e-05,
      "loss": 0.1385,
      "step": 1209
    },
    {
      "epoch": 2.8537735849056602,
      "grad_norm": 0.203125,
      "learning_rate": 3.660628019323672e-05,
      "loss": 0.1742,
      "step": 1210
    },
    {
      "epoch": 2.856132075471698,
      "grad_norm": 0.201171875,
      "learning_rate": 3.6594202898550726e-05,
      "loss": 0.1811,
      "step": 1211
    },
    {
      "epoch": 2.858490566037736,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.6582125603864735e-05,
      "loss": 0.2164,
      "step": 1212
    },
    {
      "epoch": 2.8608490566037736,
      "grad_norm": 0.19921875,
      "learning_rate": 3.6570048309178744e-05,
      "loss": 0.1637,
      "step": 1213
    },
    {
      "epoch": 2.8632075471698113,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.655797101449275e-05,
      "loss": 0.1676,
      "step": 1214
    },
    {
      "epoch": 2.865566037735849,
      "grad_norm": 0.185546875,
      "learning_rate": 3.654589371980677e-05,
      "loss": 0.1623,
      "step": 1215
    },
    {
      "epoch": 2.867924528301887,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.653381642512078e-05,
      "loss": 0.162,
      "step": 1216
    },
    {
      "epoch": 2.8702830188679247,
      "grad_norm": 0.2080078125,
      "learning_rate": 3.6521739130434786e-05,
      "loss": 0.2494,
      "step": 1217
    },
    {
      "epoch": 2.8726415094339623,
      "grad_norm": 0.3125,
      "learning_rate": 3.6509661835748795e-05,
      "loss": 0.1946,
      "step": 1218
    },
    {
      "epoch": 2.875,
      "grad_norm": 0.18359375,
      "learning_rate": 3.6497584541062804e-05,
      "loss": 0.1741,
      "step": 1219
    },
    {
      "epoch": 2.8773584905660377,
      "grad_norm": 0.185546875,
      "learning_rate": 3.648550724637681e-05,
      "loss": 0.1485,
      "step": 1220
    },
    {
      "epoch": 2.8797169811320753,
      "grad_norm": 0.224609375,
      "learning_rate": 3.647342995169082e-05,
      "loss": 0.142,
      "step": 1221
    },
    {
      "epoch": 2.882075471698113,
      "grad_norm": 0.18359375,
      "learning_rate": 3.646135265700483e-05,
      "loss": 0.1576,
      "step": 1222
    },
    {
      "epoch": 2.884433962264151,
      "grad_norm": 0.435546875,
      "learning_rate": 3.644927536231884e-05,
      "loss": 0.1523,
      "step": 1223
    },
    {
      "epoch": 2.8867924528301887,
      "grad_norm": 0.220703125,
      "learning_rate": 3.643719806763285e-05,
      "loss": 0.234,
      "step": 1224
    },
    {
      "epoch": 2.8891509433962264,
      "grad_norm": 0.318359375,
      "learning_rate": 3.6425120772946864e-05,
      "loss": 0.2233,
      "step": 1225
    },
    {
      "epoch": 2.891509433962264,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.641304347826087e-05,
      "loss": 0.1666,
      "step": 1226
    },
    {
      "epoch": 2.893867924528302,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.640096618357488e-05,
      "loss": 0.167,
      "step": 1227
    },
    {
      "epoch": 2.8962264150943398,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.638888888888889e-05,
      "loss": 0.1722,
      "step": 1228
    },
    {
      "epoch": 2.8985849056603774,
      "grad_norm": 0.25390625,
      "learning_rate": 3.63768115942029e-05,
      "loss": 0.1925,
      "step": 1229
    },
    {
      "epoch": 2.900943396226415,
      "grad_norm": 0.19140625,
      "learning_rate": 3.636473429951691e-05,
      "loss": 0.1325,
      "step": 1230
    },
    {
      "epoch": 2.9033018867924527,
      "grad_norm": 0.1875,
      "learning_rate": 3.635265700483092e-05,
      "loss": 0.233,
      "step": 1231
    },
    {
      "epoch": 2.9056603773584904,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.6340579710144927e-05,
      "loss": 0.1781,
      "step": 1232
    },
    {
      "epoch": 2.9080188679245285,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.6328502415458936e-05,
      "loss": 0.2029,
      "step": 1233
    },
    {
      "epoch": 2.910377358490566,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.631642512077295e-05,
      "loss": 0.1959,
      "step": 1234
    },
    {
      "epoch": 2.9127358490566038,
      "grad_norm": 0.193359375,
      "learning_rate": 3.630434782608696e-05,
      "loss": 0.2121,
      "step": 1235
    },
    {
      "epoch": 2.9150943396226414,
      "grad_norm": 0.1953125,
      "learning_rate": 3.629227053140097e-05,
      "loss": 0.1859,
      "step": 1236
    },
    {
      "epoch": 2.9174528301886795,
      "grad_norm": 0.185546875,
      "learning_rate": 3.628019323671498e-05,
      "loss": 0.1822,
      "step": 1237
    },
    {
      "epoch": 2.919811320754717,
      "grad_norm": 0.185546875,
      "learning_rate": 3.626811594202899e-05,
      "loss": 0.1592,
      "step": 1238
    },
    {
      "epoch": 2.922169811320755,
      "grad_norm": 0.17578125,
      "learning_rate": 3.6256038647343e-05,
      "loss": 0.1637,
      "step": 1239
    },
    {
      "epoch": 2.9245283018867925,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.6243961352657005e-05,
      "loss": 0.1613,
      "step": 1240
    },
    {
      "epoch": 2.92688679245283,
      "grad_norm": 0.2060546875,
      "learning_rate": 3.6231884057971014e-05,
      "loss": 0.1936,
      "step": 1241
    },
    {
      "epoch": 2.9292452830188678,
      "grad_norm": 0.181640625,
      "learning_rate": 3.621980676328502e-05,
      "loss": 0.1408,
      "step": 1242
    },
    {
      "epoch": 2.9316037735849054,
      "grad_norm": 0.19921875,
      "learning_rate": 3.620772946859903e-05,
      "loss": 0.198,
      "step": 1243
    },
    {
      "epoch": 2.9339622641509435,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.619565217391305e-05,
      "loss": 0.1584,
      "step": 1244
    },
    {
      "epoch": 2.936320754716981,
      "grad_norm": 0.203125,
      "learning_rate": 3.6183574879227056e-05,
      "loss": 0.1821,
      "step": 1245
    },
    {
      "epoch": 2.938679245283019,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.6171497584541065e-05,
      "loss": 0.1955,
      "step": 1246
    },
    {
      "epoch": 2.9410377358490565,
      "grad_norm": 0.23828125,
      "learning_rate": 3.6159420289855074e-05,
      "loss": 0.1995,
      "step": 1247
    },
    {
      "epoch": 2.9433962264150946,
      "grad_norm": 0.2109375,
      "learning_rate": 3.614734299516908e-05,
      "loss": 0.178,
      "step": 1248
    },
    {
      "epoch": 2.9457547169811322,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.61352657004831e-05,
      "loss": 0.2233,
      "step": 1249
    },
    {
      "epoch": 2.94811320754717,
      "grad_norm": 0.396484375,
      "learning_rate": 3.61231884057971e-05,
      "loss": 0.1746,
      "step": 1250
    },
    {
      "epoch": 2.9504716981132075,
      "grad_norm": 0.19921875,
      "learning_rate": 3.611111111111111e-05,
      "loss": 0.1922,
      "step": 1251
    },
    {
      "epoch": 2.952830188679245,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.609903381642512e-05,
      "loss": 0.2223,
      "step": 1252
    },
    {
      "epoch": 2.955188679245283,
      "grad_norm": 0.18359375,
      "learning_rate": 3.6086956521739134e-05,
      "loss": 0.1854,
      "step": 1253
    },
    {
      "epoch": 2.9575471698113205,
      "grad_norm": 0.197265625,
      "learning_rate": 3.607487922705314e-05,
      "loss": 0.1685,
      "step": 1254
    },
    {
      "epoch": 2.9599056603773586,
      "grad_norm": 0.18359375,
      "learning_rate": 3.606280193236715e-05,
      "loss": 0.1933,
      "step": 1255
    },
    {
      "epoch": 2.9622641509433962,
      "grad_norm": 0.20703125,
      "learning_rate": 3.605072463768116e-05,
      "loss": 0.311,
      "step": 1256
    },
    {
      "epoch": 2.964622641509434,
      "grad_norm": 0.19140625,
      "learning_rate": 3.603864734299517e-05,
      "loss": 0.2193,
      "step": 1257
    },
    {
      "epoch": 2.9669811320754715,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.6026570048309185e-05,
      "loss": 0.213,
      "step": 1258
    },
    {
      "epoch": 2.9693396226415096,
      "grad_norm": 0.189453125,
      "learning_rate": 3.6014492753623194e-05,
      "loss": 0.1949,
      "step": 1259
    },
    {
      "epoch": 2.9716981132075473,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.6002415458937196e-05,
      "loss": 0.2059,
      "step": 1260
    },
    {
      "epoch": 2.974056603773585,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.5990338164251205e-05,
      "loss": 0.1922,
      "step": 1261
    },
    {
      "epoch": 2.9764150943396226,
      "grad_norm": 0.208984375,
      "learning_rate": 3.5978260869565214e-05,
      "loss": 0.1772,
      "step": 1262
    },
    {
      "epoch": 2.9787735849056602,
      "grad_norm": 0.70703125,
      "learning_rate": 3.596618357487923e-05,
      "loss": 0.1584,
      "step": 1263
    },
    {
      "epoch": 2.981132075471698,
      "grad_norm": 0.2392578125,
      "learning_rate": 3.595410628019324e-05,
      "loss": 0.2285,
      "step": 1264
    },
    {
      "epoch": 2.983490566037736,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.594202898550725e-05,
      "loss": 0.1494,
      "step": 1265
    },
    {
      "epoch": 2.9858490566037736,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.592995169082126e-05,
      "loss": 0.224,
      "step": 1266
    },
    {
      "epoch": 2.9882075471698113,
      "grad_norm": 0.205078125,
      "learning_rate": 3.5917874396135266e-05,
      "loss": 0.1966,
      "step": 1267
    },
    {
      "epoch": 2.990566037735849,
      "grad_norm": 0.189453125,
      "learning_rate": 3.590579710144928e-05,
      "loss": 0.1913,
      "step": 1268
    },
    {
      "epoch": 2.992924528301887,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.589371980676329e-05,
      "loss": 0.2177,
      "step": 1269
    },
    {
      "epoch": 2.9952830188679247,
      "grad_norm": 0.2109375,
      "learning_rate": 3.58816425120773e-05,
      "loss": 0.153,
      "step": 1270
    },
    {
      "epoch": 2.9976415094339623,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.58695652173913e-05,
      "loss": 0.1582,
      "step": 1271
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.41015625,
      "learning_rate": 3.585748792270532e-05,
      "loss": 0.2775,
      "step": 1272
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.18447954952716827,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 1.926,
      "eval_samples_per_second": 1.558,
      "eval_steps_per_second": 1.558,
      "step": 1272
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.1286066472530365,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 1.3892,
      "eval_samples_per_second": 2.879,
      "eval_steps_per_second": 2.879,
      "step": 1272
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.2791191041469574,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 3.0854,
      "eval_samples_per_second": 1.296,
      "eval_steps_per_second": 1.296,
      "step": 1272
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.3917909860610962,
      "eval_model_preparation_time": 0.0079,
      "eval_runtime": 2.8983,
      "eval_samples_per_second": 1.38,
      "eval_steps_per_second": 1.38,
      "step": 1272
    }
  ],
  "logging_steps": 1,
  "max_steps": 4240,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6752781420553994e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}