{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0011273957158964,
  "eval_steps": 500,
  "global_step": 888,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0011273957158962795,
      "grad_norm": 8.9375,
      "learning_rate": 1.1363636363636364e-07,
      "loss": 1.7846,
      "step": 1
    },
    {
      "epoch": 0.002254791431792559,
      "grad_norm": 4.28125,
      "learning_rate": 2.2727272727272729e-07,
      "loss": 1.9124,
      "step": 2
    },
    {
      "epoch": 0.0033821871476888386,
      "grad_norm": 4.28125,
      "learning_rate": 3.4090909090909096e-07,
      "loss": 1.791,
      "step": 3
    },
    {
      "epoch": 0.004509582863585118,
      "grad_norm": 5.28125,
      "learning_rate": 4.5454545454545457e-07,
      "loss": 1.705,
      "step": 4
    },
    {
      "epoch": 0.005636978579481398,
      "grad_norm": 4.15625,
      "learning_rate": 5.681818181818182e-07,
      "loss": 1.749,
      "step": 5
    },
    {
      "epoch": 0.006764374295377677,
      "grad_norm": 5.3125,
      "learning_rate": 6.818181818181819e-07,
      "loss": 1.6386,
      "step": 6
    },
    {
      "epoch": 0.007891770011273957,
      "grad_norm": 3.046875,
      "learning_rate": 7.954545454545455e-07,
      "loss": 1.8087,
      "step": 7
    },
    {
      "epoch": 0.009019165727170236,
      "grad_norm": 3.265625,
      "learning_rate": 9.090909090909091e-07,
      "loss": 1.8114,
      "step": 8
    },
    {
      "epoch": 0.010146561443066516,
      "grad_norm": 2.828125,
      "learning_rate": 1.0227272727272729e-06,
      "loss": 1.841,
      "step": 9
    },
    {
      "epoch": 0.011273957158962795,
      "grad_norm": 3.15625,
      "learning_rate": 1.1363636363636364e-06,
      "loss": 1.7005,
      "step": 10
    },
    {
      "epoch": 0.012401352874859075,
      "grad_norm": 2.375,
      "learning_rate": 1.25e-06,
      "loss": 1.7904,
      "step": 11
    },
    {
      "epoch": 0.013528748590755355,
      "grad_norm": 2.546875,
      "learning_rate": 1.3636363636363638e-06,
      "loss": 1.7428,
      "step": 12
    },
    {
      "epoch": 0.014656144306651634,
      "grad_norm": 2.453125,
      "learning_rate": 1.4772727272727275e-06,
      "loss": 1.8844,
      "step": 13
    },
    {
      "epoch": 0.015783540022547914,
      "grad_norm": 2.140625,
      "learning_rate": 1.590909090909091e-06,
      "loss": 1.8013,
      "step": 14
    },
    {
      "epoch": 0.016910935738444193,
      "grad_norm": 2.4375,
      "learning_rate": 1.7045454545454548e-06,
      "loss": 1.6395,
      "step": 15
    },
    {
      "epoch": 0.018038331454340473,
      "grad_norm": 2.421875,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 1.82,
      "step": 16
    },
    {
      "epoch": 0.019165727170236752,
      "grad_norm": 2.5,
      "learning_rate": 1.931818181818182e-06,
      "loss": 1.5336,
      "step": 17
    },
    {
      "epoch": 0.020293122886133032,
      "grad_norm": 2.4375,
      "learning_rate": 2.0454545454545457e-06,
      "loss": 1.7984,
      "step": 18
    },
    {
      "epoch": 0.02142051860202931,
      "grad_norm": 3.8125,
      "learning_rate": 2.1590909090909092e-06,
      "loss": 1.4523,
      "step": 19
    },
    {
      "epoch": 0.02254791431792559,
      "grad_norm": 2.28125,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.5846,
      "step": 20
    },
    {
      "epoch": 0.02367531003382187,
      "grad_norm": 2.59375,
      "learning_rate": 2.3863636363636363e-06,
      "loss": 1.7654,
      "step": 21
    },
    {
      "epoch": 0.02480270574971815,
      "grad_norm": 2.5625,
      "learning_rate": 2.5e-06,
      "loss": 1.7708,
      "step": 22
    },
    {
      "epoch": 0.02593010146561443,
      "grad_norm": 2.65625,
      "learning_rate": 2.6136363636363637e-06,
      "loss": 1.6936,
      "step": 23
    },
    {
      "epoch": 0.02705749718151071,
      "grad_norm": 2.734375,
      "learning_rate": 2.7272727272727276e-06,
      "loss": 1.7375,
      "step": 24
    },
    {
      "epoch": 0.02818489289740699,
      "grad_norm": 2.9375,
      "learning_rate": 2.840909090909091e-06,
      "loss": 1.5302,
      "step": 25
    },
    {
      "epoch": 0.029312288613303268,
      "grad_norm": 2.9375,
      "learning_rate": 2.954545454545455e-06,
      "loss": 1.6362,
      "step": 26
    },
    {
      "epoch": 0.030439684329199548,
      "grad_norm": 3.1875,
      "learning_rate": 3.068181818181818e-06,
      "loss": 1.7483,
      "step": 27
    },
    {
      "epoch": 0.03156708004509583,
      "grad_norm": 3.296875,
      "learning_rate": 3.181818181818182e-06,
      "loss": 1.7535,
      "step": 28
    },
    {
      "epoch": 0.03269447576099211,
      "grad_norm": 3.828125,
      "learning_rate": 3.2954545454545456e-06,
      "loss": 1.6826,
      "step": 29
    },
    {
      "epoch": 0.033821871476888386,
      "grad_norm": 4.03125,
      "learning_rate": 3.4090909090909096e-06,
      "loss": 1.6137,
      "step": 30
    },
    {
      "epoch": 0.03494926719278467,
      "grad_norm": 3.3125,
      "learning_rate": 3.5227272727272726e-06,
      "loss": 1.5402,
      "step": 31
    },
    {
      "epoch": 0.036076662908680945,
      "grad_norm": 3.84375,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 1.7778,
      "step": 32
    },
    {
      "epoch": 0.03720405862457723,
      "grad_norm": 4.625,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 1.7418,
      "step": 33
    },
    {
      "epoch": 0.038331454340473504,
      "grad_norm": 4.75,
      "learning_rate": 3.863636363636364e-06,
      "loss": 1.8357,
      "step": 34
    },
    {
      "epoch": 0.03945885005636979,
      "grad_norm": 4.03125,
      "learning_rate": 3.9772727272727275e-06,
      "loss": 1.6101,
      "step": 35
    },
    {
      "epoch": 0.040586245772266064,
      "grad_norm": 4.375,
      "learning_rate": 4.0909090909090915e-06,
      "loss": 1.7671,
      "step": 36
    },
    {
      "epoch": 0.04171364148816235,
      "grad_norm": 3.734375,
      "learning_rate": 4.204545454545455e-06,
      "loss": 1.8664,
      "step": 37
    },
    {
      "epoch": 0.04284103720405862,
      "grad_norm": 5.4375,
      "learning_rate": 4.3181818181818185e-06,
      "loss": 1.7168,
      "step": 38
    },
    {
      "epoch": 0.043968432919954906,
      "grad_norm": 3.34375,
      "learning_rate": 4.4318181818181824e-06,
      "loss": 1.6126,
      "step": 39
    },
    {
      "epoch": 0.04509582863585118,
      "grad_norm": 5.21875,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.8064,
      "step": 40
    },
    {
      "epoch": 0.046223224351747465,
      "grad_norm": 3.140625,
      "learning_rate": 4.6590909090909095e-06,
      "loss": 1.765,
      "step": 41
    },
    {
      "epoch": 0.04735062006764374,
      "grad_norm": 4.6875,
      "learning_rate": 4.7727272727272725e-06,
      "loss": 1.5978,
      "step": 42
    },
    {
      "epoch": 0.048478015783540024,
      "grad_norm": 4.6875,
      "learning_rate": 4.8863636363636365e-06,
      "loss": 1.7776,
      "step": 43
    },
    {
      "epoch": 0.0496054114994363,
      "grad_norm": 6.46875,
      "learning_rate": 5e-06,
      "loss": 1.7274,
      "step": 44
    },
    {
      "epoch": 0.05073280721533258,
      "grad_norm": 4.21875,
      "learning_rate": 5.113636363636364e-06,
      "loss": 1.7375,
      "step": 45
    },
    {
      "epoch": 0.05186020293122886,
      "grad_norm": 5.9375,
      "learning_rate": 5.2272727272727274e-06,
      "loss": 1.6734,
      "step": 46
    },
    {
      "epoch": 0.05298759864712514,
      "grad_norm": 5.5625,
      "learning_rate": 5.340909090909091e-06,
      "loss": 1.8996,
      "step": 47
    },
    {
      "epoch": 0.05411499436302142,
      "grad_norm": 7.0625,
      "learning_rate": 5.454545454545455e-06,
      "loss": 1.7567,
      "step": 48
    },
    {
      "epoch": 0.0552423900789177,
      "grad_norm": 3.875,
      "learning_rate": 5.568181818181819e-06,
      "loss": 1.7517,
      "step": 49
    },
    {
      "epoch": 0.05636978579481398,
      "grad_norm": 6.84375,
      "learning_rate": 5.681818181818182e-06,
      "loss": 1.825,
      "step": 50
    },
    {
      "epoch": 0.05749718151071026,
      "grad_norm": 4.59375,
      "learning_rate": 5.795454545454546e-06,
      "loss": 1.7975,
      "step": 51
    },
    {
      "epoch": 0.058624577226606536,
      "grad_norm": 6.6875,
      "learning_rate": 5.90909090909091e-06,
      "loss": 1.7209,
      "step": 52
    },
    {
      "epoch": 0.05975197294250282,
      "grad_norm": 4.34375,
      "learning_rate": 6.022727272727274e-06,
      "loss": 1.5089,
      "step": 53
    },
    {
      "epoch": 0.060879368658399095,
      "grad_norm": 6.84375,
      "learning_rate": 6.136363636363636e-06,
      "loss": 1.685,
      "step": 54
    },
    {
      "epoch": 0.06200676437429538,
      "grad_norm": 4.6875,
      "learning_rate": 6.25e-06,
      "loss": 1.459,
      "step": 55
    },
    {
      "epoch": 0.06313416009019165,
      "grad_norm": 7.34375,
      "learning_rate": 6.363636363636364e-06,
      "loss": 1.7602,
      "step": 56
    },
    {
      "epoch": 0.06426155580608793,
      "grad_norm": 4.78125,
      "learning_rate": 6.477272727272728e-06,
      "loss": 1.7697,
      "step": 57
    },
    {
      "epoch": 0.06538895152198422,
      "grad_norm": 7.28125,
      "learning_rate": 6.590909090909091e-06,
      "loss": 1.7208,
      "step": 58
    },
    {
      "epoch": 0.0665163472378805,
      "grad_norm": 4.75,
      "learning_rate": 6.704545454545455e-06,
      "loss": 1.7793,
      "step": 59
    },
    {
      "epoch": 0.06764374295377677,
      "grad_norm": 7.84375,
      "learning_rate": 6.818181818181819e-06,
      "loss": 1.874,
      "step": 60
    },
    {
      "epoch": 0.06877113866967305,
      "grad_norm": 5.09375,
      "learning_rate": 6.931818181818183e-06,
      "loss": 1.6342,
      "step": 61
    },
    {
      "epoch": 0.06989853438556934,
      "grad_norm": 6.84375,
      "learning_rate": 7.045454545454545e-06,
      "loss": 1.7773,
      "step": 62
    },
    {
      "epoch": 0.07102593010146561,
      "grad_norm": 4.71875,
      "learning_rate": 7.159090909090909e-06,
      "loss": 1.8439,
      "step": 63
    },
    {
      "epoch": 0.07215332581736189,
      "grad_norm": 6.40625,
      "learning_rate": 7.272727272727273e-06,
      "loss": 1.7297,
      "step": 64
    },
    {
      "epoch": 0.07328072153325817,
      "grad_norm": 4.4375,
      "learning_rate": 7.386363636363637e-06,
      "loss": 1.4491,
      "step": 65
    },
    {
      "epoch": 0.07440811724915446,
      "grad_norm": 6.125,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.671,
      "step": 66
    },
    {
      "epoch": 0.07553551296505073,
      "grad_norm": 5.5,
      "learning_rate": 7.613636363636364e-06,
      "loss": 1.7045,
      "step": 67
    },
    {
      "epoch": 0.07666290868094701,
      "grad_norm": 6.4375,
      "learning_rate": 7.727272727272727e-06,
      "loss": 1.4962,
      "step": 68
    },
    {
      "epoch": 0.0777903043968433,
      "grad_norm": 4.09375,
      "learning_rate": 7.840909090909091e-06,
      "loss": 1.4349,
      "step": 69
    },
    {
      "epoch": 0.07891770011273957,
      "grad_norm": 3.890625,
      "learning_rate": 7.954545454545455e-06,
      "loss": 1.6128,
      "step": 70
    },
    {
      "epoch": 0.08004509582863585,
      "grad_norm": 5.9375,
      "learning_rate": 8.068181818181819e-06,
      "loss": 1.7444,
      "step": 71
    },
    {
      "epoch": 0.08117249154453213,
      "grad_norm": 6.03125,
      "learning_rate": 8.181818181818183e-06,
      "loss": 1.6638,
      "step": 72
    },
    {
      "epoch": 0.08229988726042842,
      "grad_norm": 4.34375,
      "learning_rate": 8.295454545454547e-06,
      "loss": 1.6338,
      "step": 73
    },
    {
      "epoch": 0.0834272829763247,
      "grad_norm": 5.1875,
      "learning_rate": 8.40909090909091e-06,
      "loss": 1.7443,
      "step": 74
    },
    {
      "epoch": 0.08455467869222097,
      "grad_norm": 3.890625,
      "learning_rate": 8.522727272727273e-06,
      "loss": 1.535,
      "step": 75
    },
    {
      "epoch": 0.08568207440811725,
      "grad_norm": 5.5625,
      "learning_rate": 8.636363636363637e-06,
      "loss": 1.4609,
      "step": 76
    },
    {
      "epoch": 0.08680947012401354,
      "grad_norm": 4.21875,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.7159,
      "step": 77
    },
    {
      "epoch": 0.08793686583990981,
      "grad_norm": 6.8125,
      "learning_rate": 8.863636363636365e-06,
      "loss": 1.7649,
      "step": 78
    },
    {
      "epoch": 0.08906426155580609,
      "grad_norm": 4.3125,
      "learning_rate": 8.977272727272727e-06,
      "loss": 1.8751,
      "step": 79
    },
    {
      "epoch": 0.09019165727170236,
      "grad_norm": 7.1875,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.8044,
      "step": 80
    },
    {
      "epoch": 0.09131905298759865,
      "grad_norm": 4.1875,
      "learning_rate": 9.204545454545455e-06,
      "loss": 1.6525,
      "step": 81
    },
    {
      "epoch": 0.09244644870349493,
      "grad_norm": 5.96875,
      "learning_rate": 9.318181818181819e-06,
      "loss": 1.5917,
      "step": 82
    },
    {
      "epoch": 0.0935738444193912,
      "grad_norm": 5.625,
      "learning_rate": 9.431818181818183e-06,
      "loss": 1.7145,
      "step": 83
    },
    {
      "epoch": 0.09470124013528748,
      "grad_norm": 6.96875,
      "learning_rate": 9.545454545454545e-06,
      "loss": 1.8438,
      "step": 84
    },
    {
      "epoch": 0.09582863585118377,
      "grad_norm": 5.03125,
      "learning_rate": 9.659090909090909e-06,
      "loss": 1.6567,
      "step": 85
    },
    {
      "epoch": 0.09695603156708005,
      "grad_norm": 6.40625,
      "learning_rate": 9.772727272727273e-06,
      "loss": 1.7732,
      "step": 86
    },
    {
      "epoch": 0.09808342728297632,
      "grad_norm": 4.625,
      "learning_rate": 9.886363636363637e-06,
      "loss": 1.6559,
      "step": 87
    },
    {
      "epoch": 0.0992108229988726,
      "grad_norm": 6.3125,
      "learning_rate": 1e-05,
      "loss": 1.7145,
      "step": 88
    },
    {
      "epoch": 0.10033821871476889,
      "grad_norm": 4.4375,
      "learning_rate": 9.9994065634087e-06,
      "loss": 1.6582,
      "step": 89
    },
    {
      "epoch": 0.10146561443066517,
      "grad_norm": 7.15625,
      "learning_rate": 9.99881249257808e-06,
      "loss": 1.8273,
      "step": 90
    },
    {
      "epoch": 0.10259301014656144,
      "grad_norm": 4.59375,
      "learning_rate": 9.998217786490822e-06,
      "loss": 1.6587,
      "step": 91
    },
    {
      "epoch": 0.10372040586245772,
      "grad_norm": 6.3125,
      "learning_rate": 9.997622444127439e-06,
      "loss": 1.5284,
      "step": 92
    },
    {
      "epoch": 0.10484780157835401,
      "grad_norm": 4.59375,
      "learning_rate": 9.997026464466251e-06,
      "loss": 1.5952,
      "step": 93
    },
    {
      "epoch": 0.10597519729425028,
      "grad_norm": 6.9375,
      "learning_rate": 9.9964298464834e-06,
      "loss": 1.6027,
      "step": 94
    },
    {
      "epoch": 0.10710259301014656,
      "grad_norm": 3.875,
      "learning_rate": 9.995832589152826e-06,
      "loss": 1.614,
      "step": 95
    },
    {
      "epoch": 0.10822998872604284,
      "grad_norm": 5.875,
      "learning_rate": 9.995234691446273e-06,
      "loss": 1.6792,
      "step": 96
    },
    {
      "epoch": 0.10935738444193913,
      "grad_norm": 4.6875,
      "learning_rate": 9.994636152333274e-06,
      "loss": 1.7316,
      "step": 97
    },
    {
      "epoch": 0.1104847801578354,
      "grad_norm": 6.84375,
      "learning_rate": 9.994036970781159e-06,
      "loss": 1.7538,
      "step": 98
    },
    {
      "epoch": 0.11161217587373168,
      "grad_norm": 5.375,
      "learning_rate": 9.993437145755027e-06,
      "loss": 1.8301,
      "step": 99
    },
    {
      "epoch": 0.11273957158962795,
      "grad_norm": 6.5625,
      "learning_rate": 9.992836676217766e-06,
      "loss": 1.7217,
      "step": 100
    },
    {
      "epoch": 0.11386696730552424,
      "grad_norm": 4.375,
      "learning_rate": 9.992235561130025e-06,
      "loss": 1.6602,
      "step": 101
    },
    {
      "epoch": 0.11499436302142052,
      "grad_norm": 6.53125,
      "learning_rate": 9.991633799450222e-06,
      "loss": 1.7215,
      "step": 102
    },
    {
      "epoch": 0.1161217587373168,
      "grad_norm": 5.09375,
      "learning_rate": 9.99103139013453e-06,
      "loss": 1.6249,
      "step": 103
    },
    {
      "epoch": 0.11724915445321307,
      "grad_norm": 6.34375,
      "learning_rate": 9.990428332136875e-06,
      "loss": 1.6093,
      "step": 104
    },
    {
      "epoch": 0.11837655016910936,
      "grad_norm": 3.75,
      "learning_rate": 9.98982462440893e-06,
      "loss": 1.4658,
      "step": 105
    },
    {
      "epoch": 0.11950394588500564,
      "grad_norm": 5.71875,
      "learning_rate": 9.989220265900107e-06,
      "loss": 1.6491,
      "step": 106
    },
    {
      "epoch": 0.12063134160090191,
      "grad_norm": 4.84375,
      "learning_rate": 9.988615255557554e-06,
      "loss": 1.6891,
      "step": 107
    },
    {
      "epoch": 0.12175873731679819,
      "grad_norm": 6.125,
      "learning_rate": 9.98800959232614e-06,
      "loss": 1.6079,
      "step": 108
    },
    {
      "epoch": 0.12288613303269448,
      "grad_norm": 3.734375,
      "learning_rate": 9.987403275148463e-06,
      "loss": 1.7634,
      "step": 109
    },
    {
      "epoch": 0.12401352874859076,
      "grad_norm": 5.3125,
      "learning_rate": 9.98679630296483e-06,
      "loss": 1.7401,
      "step": 110
    },
    {
      "epoch": 0.12514092446448705,
      "grad_norm": 5.875,
      "learning_rate": 9.986188674713266e-06,
      "loss": 1.7283,
      "step": 111
    },
    {
      "epoch": 0.1262683201803833,
      "grad_norm": 4.71875,
      "learning_rate": 9.985580389329489e-06,
      "loss": 1.4814,
      "step": 112
    },
    {
      "epoch": 0.1273957158962796,
      "grad_norm": 4.5,
      "learning_rate": 9.98497144574692e-06,
      "loss": 1.6475,
      "step": 113
    },
    {
      "epoch": 0.12852311161217586,
      "grad_norm": 6.03125,
      "learning_rate": 9.984361842896669e-06,
      "loss": 1.7061,
      "step": 114
    },
    {
      "epoch": 0.12965050732807215,
      "grad_norm": 3.890625,
      "learning_rate": 9.98375157970753e-06,
      "loss": 1.5777,
      "step": 115
    },
    {
      "epoch": 0.13077790304396844,
      "grad_norm": 5.71875,
      "learning_rate": 9.983140655105975e-06,
      "loss": 1.5693,
      "step": 116
    },
    {
      "epoch": 0.1319052987598647,
      "grad_norm": 4.25,
      "learning_rate": 9.982529068016147e-06,
      "loss": 1.84,
      "step": 117
    },
    {
      "epoch": 0.133032694475761,
      "grad_norm": 5.34375,
      "learning_rate": 9.981916817359856e-06,
      "loss": 1.6239,
      "step": 118
    },
    {
      "epoch": 0.13416009019165728,
      "grad_norm": 5.0625,
      "learning_rate": 9.981303902056572e-06,
      "loss": 1.6975,
      "step": 119
    },
    {
      "epoch": 0.13528748590755355,
      "grad_norm": 5.28125,
      "learning_rate": 9.980690321023414e-06,
      "loss": 1.6421,
      "step": 120
    },
    {
      "epoch": 0.13641488162344984,
      "grad_norm": 4.625,
      "learning_rate": 9.98007607317515e-06,
      "loss": 1.6319,
      "step": 121
    },
    {
      "epoch": 0.1375422773393461,
      "grad_norm": 6.0625,
      "learning_rate": 9.979461157424189e-06,
      "loss": 1.7249,
      "step": 122
    },
    {
      "epoch": 0.1386696730552424,
      "grad_norm": 4.96875,
      "learning_rate": 9.97884557268057e-06,
      "loss": 1.7171,
      "step": 123
    },
    {
      "epoch": 0.13979706877113868,
      "grad_norm": 5.375,
      "learning_rate": 9.97822931785196e-06,
      "loss": 1.6007,
      "step": 124
    },
    {
      "epoch": 0.14092446448703494,
      "grad_norm": 4.375,
      "learning_rate": 9.97761239184365e-06,
      "loss": 1.4696,
      "step": 125
    },
    {
      "epoch": 0.14205186020293123,
      "grad_norm": 5.71875,
      "learning_rate": 9.976994793558542e-06,
      "loss": 1.6035,
      "step": 126
    },
    {
      "epoch": 0.14317925591882752,
      "grad_norm": 3.890625,
      "learning_rate": 9.976376521897147e-06,
      "loss": 1.6808,
      "step": 127
    },
    {
      "epoch": 0.14430665163472378,
      "grad_norm": 5.03125,
      "learning_rate": 9.975757575757577e-06,
      "loss": 1.575,
      "step": 128
    },
    {
      "epoch": 0.14543404735062007,
      "grad_norm": 5.1875,
      "learning_rate": 9.975137954035535e-06,
      "loss": 1.5836,
      "step": 129
    },
    {
      "epoch": 0.14656144306651633,
      "grad_norm": 5.5,
      "learning_rate": 9.974517655624318e-06,
      "loss": 1.6954,
      "step": 130
    },
    {
      "epoch": 0.14768883878241262,
      "grad_norm": 3.453125,
      "learning_rate": 9.9738966794148e-06,
      "loss": 1.6806,
      "step": 131
    },
    {
      "epoch": 0.1488162344983089,
      "grad_norm": 4.96875,
      "learning_rate": 9.973275024295434e-06,
      "loss": 1.496,
      "step": 132
    },
    {
      "epoch": 0.14994363021420518,
      "grad_norm": 5.0625,
      "learning_rate": 9.972652689152234e-06,
      "loss": 1.6856,
      "step": 133
    },
    {
      "epoch": 0.15107102593010147,
      "grad_norm": 5.46875,
      "learning_rate": 9.972029672868784e-06,
      "loss": 1.7261,
      "step": 134
    },
    {
      "epoch": 0.15219842164599776,
      "grad_norm": 3.921875,
      "learning_rate": 9.971405974326217e-06,
      "loss": 1.6694,
      "step": 135
    },
    {
      "epoch": 0.15332581736189402,
      "grad_norm": 6.25,
      "learning_rate": 9.970781592403216e-06,
      "loss": 1.7033,
      "step": 136
    },
    {
      "epoch": 0.1544532130777903,
      "grad_norm": 4.78125,
      "learning_rate": 9.970156525976004e-06,
      "loss": 1.9388,
      "step": 137
    },
    {
      "epoch": 0.1555806087936866,
      "grad_norm": 4.6875,
      "learning_rate": 9.969530773918343e-06,
      "loss": 1.6945,
      "step": 138
    },
    {
      "epoch": 0.15670800450958286,
      "grad_norm": 4.53125,
      "learning_rate": 9.968904335101519e-06,
      "loss": 1.559,
      "step": 139
    },
    {
      "epoch": 0.15783540022547915,
      "grad_norm": 5.21875,
      "learning_rate": 9.96827720839434e-06,
      "loss": 1.6408,
      "step": 140
    },
    {
      "epoch": 0.1589627959413754,
      "grad_norm": 4.59375,
      "learning_rate": 9.967649392663127e-06,
      "loss": 1.5835,
      "step": 141
    },
    {
      "epoch": 0.1600901916572717,
      "grad_norm": 5.34375,
      "learning_rate": 9.967020886771712e-06,
      "loss": 1.5748,
      "step": 142
    },
    {
      "epoch": 0.161217587373168,
      "grad_norm": 3.90625,
      "learning_rate": 9.966391689581424e-06,
      "loss": 1.6636,
      "step": 143
    },
    {
      "epoch": 0.16234498308906425,
      "grad_norm": 5.59375,
      "learning_rate": 9.965761799951089e-06,
      "loss": 1.6184,
      "step": 144
    },
    {
      "epoch": 0.16347237880496054,
      "grad_norm": 5.59375,
      "learning_rate": 9.965131216737015e-06,
      "loss": 1.7388,
      "step": 145
    },
    {
      "epoch": 0.16459977452085683,
      "grad_norm": 4.65625,
      "learning_rate": 9.964499938792999e-06,
      "loss": 1.5444,
      "step": 146
    },
    {
      "epoch": 0.1657271702367531,
      "grad_norm": 3.75,
      "learning_rate": 9.963867964970299e-06,
      "loss": 1.5246,
      "step": 147
    },
    {
      "epoch": 0.1668545659526494,
      "grad_norm": 5.53125,
      "learning_rate": 9.963235294117648e-06,
      "loss": 1.6923,
      "step": 148
    },
    {
      "epoch": 0.16798196166854565,
      "grad_norm": 4.0625,
      "learning_rate": 9.962601925081234e-06,
      "loss": 1.5901,
      "step": 149
    },
    {
      "epoch": 0.16910935738444194,
      "grad_norm": 4.875,
      "learning_rate": 9.9619678567047e-06,
      "loss": 1.6415,
      "step": 150
    },
    {
      "epoch": 0.17023675310033823,
      "grad_norm": 3.921875,
      "learning_rate": 9.96133308782913e-06,
      "loss": 1.5451,
      "step": 151
    },
    {
      "epoch": 0.1713641488162345,
      "grad_norm": 4.59375,
      "learning_rate": 9.96069761729305e-06,
      "loss": 1.5919,
      "step": 152
    },
    {
      "epoch": 0.17249154453213078,
      "grad_norm": 5.40625,
      "learning_rate": 9.960061443932414e-06,
      "loss": 1.8012,
      "step": 153
    },
    {
      "epoch": 0.17361894024802707,
      "grad_norm": 5.84375,
      "learning_rate": 9.959424566580598e-06,
      "loss": 1.7756,
      "step": 154
    },
    {
      "epoch": 0.17474633596392333,
      "grad_norm": 3.875,
      "learning_rate": 9.958786984068403e-06,
      "loss": 1.6429,
      "step": 155
    },
    {
      "epoch": 0.17587373167981962,
      "grad_norm": 5.5,
      "learning_rate": 9.95814869522403e-06,
      "loss": 1.7847,
      "step": 156
    },
    {
      "epoch": 0.17700112739571588,
      "grad_norm": 4.625,
      "learning_rate": 9.957509698873084e-06,
      "loss": 1.5883,
      "step": 157
    },
    {
      "epoch": 0.17812852311161217,
      "grad_norm": 5.15625,
      "learning_rate": 9.95686999383857e-06,
      "loss": 1.7321,
      "step": 158
    },
    {
      "epoch": 0.17925591882750846,
      "grad_norm": 4.3125,
      "learning_rate": 9.956229578940879e-06,
      "loss": 1.5098,
      "step": 159
    },
    {
      "epoch": 0.18038331454340473,
      "grad_norm": 5.5625,
      "learning_rate": 9.95558845299778e-06,
      "loss": 1.5735,
      "step": 160
    },
    {
      "epoch": 0.18151071025930102,
      "grad_norm": 4.84375,
      "learning_rate": 9.954946614824416e-06,
      "loss": 1.6569,
      "step": 161
    },
    {
      "epoch": 0.1826381059751973,
      "grad_norm": 4.9375,
      "learning_rate": 9.954304063233296e-06,
      "loss": 1.5892,
      "step": 162
    },
    {
      "epoch": 0.18376550169109357,
      "grad_norm": 4.59375,
      "learning_rate": 9.953660797034291e-06,
      "loss": 1.5893,
      "step": 163
    },
    {
      "epoch": 0.18489289740698986,
      "grad_norm": 5.34375,
      "learning_rate": 9.953016815034619e-06,
      "loss": 1.6609,
      "step": 164
    },
    {
      "epoch": 0.18602029312288612,
      "grad_norm": 4.75,
      "learning_rate": 9.952372116038845e-06,
      "loss": 1.6546,
      "step": 165
    },
    {
      "epoch": 0.1871476888387824,
      "grad_norm": 5.15625,
      "learning_rate": 9.951726698848868e-06,
      "loss": 1.4648,
      "step": 166
    },
    {
      "epoch": 0.1882750845546787,
      "grad_norm": 4.28125,
      "learning_rate": 9.951080562263918e-06,
      "loss": 1.7456,
      "step": 167
    },
    {
      "epoch": 0.18940248027057496,
      "grad_norm": 5.65625,
      "learning_rate": 9.950433705080546e-06,
      "loss": 1.8155,
      "step": 168
    },
    {
      "epoch": 0.19052987598647125,
      "grad_norm": 3.609375,
      "learning_rate": 9.949786126092617e-06,
      "loss": 1.6109,
      "step": 169
    },
    {
      "epoch": 0.19165727170236754,
      "grad_norm": 4.9375,
      "learning_rate": 9.949137824091304e-06,
      "loss": 1.8276,
      "step": 170
    },
    {
      "epoch": 0.1927846674182638,
      "grad_norm": 4.875,
      "learning_rate": 9.948488797865079e-06,
      "loss": 1.7163,
      "step": 171
    },
    {
      "epoch": 0.1939120631341601,
      "grad_norm": 6.1875,
      "learning_rate": 9.947839046199703e-06,
      "loss": 1.8181,
      "step": 172
    },
    {
      "epoch": 0.19503945885005636,
      "grad_norm": 4.25,
      "learning_rate": 9.947188567878224e-06,
      "loss": 1.646,
      "step": 173
    },
    {
      "epoch": 0.19616685456595265,
      "grad_norm": 4.625,
      "learning_rate": 9.946537361680966e-06,
      "loss": 1.7897,
      "step": 174
    },
    {
      "epoch": 0.19729425028184894,
      "grad_norm": 4.34375,
      "learning_rate": 9.945885426385521e-06,
      "loss": 1.6442,
      "step": 175
    },
    {
      "epoch": 0.1984216459977452,
      "grad_norm": 4.625,
      "learning_rate": 9.945232760766741e-06,
      "loss": 1.5793,
      "step": 176
    },
    {
      "epoch": 0.1995490417136415,
      "grad_norm": 4.4375,
      "learning_rate": 9.944579363596737e-06,
      "loss": 1.7043,
      "step": 177
    },
    {
      "epoch": 0.20067643742953778,
      "grad_norm": 6.125,
      "learning_rate": 9.94392523364486e-06,
      "loss": 1.6059,
      "step": 178
    },
    {
      "epoch": 0.20180383314543404,
      "grad_norm": 4.9375,
      "learning_rate": 9.9432703696777e-06,
      "loss": 1.5164,
      "step": 179
    },
    {
      "epoch": 0.20293122886133033,
      "grad_norm": 4.90625,
      "learning_rate": 9.942614770459082e-06,
      "loss": 1.7232,
      "step": 180
    },
    {
      "epoch": 0.2040586245772266,
      "grad_norm": 4.46875,
      "learning_rate": 9.941958434750047e-06,
      "loss": 1.6029,
      "step": 181
    },
    {
      "epoch": 0.20518602029312288,
      "grad_norm": 5.0,
      "learning_rate": 9.941301361308855e-06,
      "loss": 1.7425,
      "step": 182
    },
    {
      "epoch": 0.20631341600901917,
      "grad_norm": 3.28125,
      "learning_rate": 9.940643548890972e-06,
      "loss": 1.4423,
      "step": 183
    },
    {
      "epoch": 0.20744081172491544,
      "grad_norm": 4.75,
      "learning_rate": 9.939984996249064e-06,
      "loss": 1.6006,
      "step": 184
    },
    {
      "epoch": 0.20856820744081173,
      "grad_norm": 3.71875,
      "learning_rate": 9.939325702132984e-06,
      "loss": 1.5647,
      "step": 185
    },
    {
      "epoch": 0.20969560315670802,
      "grad_norm": 4.875,
      "learning_rate": 9.938665665289775e-06,
      "loss": 1.6318,
      "step": 186
    },
    {
      "epoch": 0.21082299887260428,
      "grad_norm": 4.96875,
      "learning_rate": 9.93800488446365e-06,
      "loss": 1.7127,
      "step": 187
    },
    {
      "epoch": 0.21195039458850057,
      "grad_norm": 4.5625,
      "learning_rate": 9.937343358395991e-06,
      "loss": 1.5561,
      "step": 188
    },
    {
      "epoch": 0.21307779030439683,
      "grad_norm": 4.6875,
      "learning_rate": 9.936681085825341e-06,
      "loss": 1.6196,
      "step": 189
    },
    {
      "epoch": 0.21420518602029312,
      "grad_norm": 4.9375,
      "learning_rate": 9.936018065487392e-06,
      "loss": 1.7327,
      "step": 190
    },
    {
      "epoch": 0.2153325817361894,
      "grad_norm": 4.90625,
      "learning_rate": 9.935354296114982e-06,
      "loss": 1.6876,
      "step": 191
    },
    {
      "epoch": 0.21645997745208567,
      "grad_norm": 4.3125,
      "learning_rate": 9.934689776438082e-06,
      "loss": 1.4816,
      "step": 192
    },
    {
      "epoch": 0.21758737316798196,
      "grad_norm": 4.875,
      "learning_rate": 9.93402450518379e-06,
      "loss": 1.6501,
      "step": 193
    },
    {
      "epoch": 0.21871476888387825,
      "grad_norm": 4.71875,
      "learning_rate": 9.933358481076325e-06,
      "loss": 1.6442,
      "step": 194
    },
    {
      "epoch": 0.21984216459977451,
      "grad_norm": 4.4375,
      "learning_rate": 9.932691702837015e-06,
      "loss": 1.6382,
      "step": 195
    },
    {
      "epoch": 0.2209695603156708,
      "grad_norm": 4.65625,
      "learning_rate": 9.93202416918429e-06,
      "loss": 1.4606,
      "step": 196
    },
    {
      "epoch": 0.22209695603156707,
      "grad_norm": 4.375,
      "learning_rate": 9.931355878833681e-06,
      "loss": 1.7238,
      "step": 197
    },
    {
      "epoch": 0.22322435174746336,
      "grad_norm": 5.0625,
      "learning_rate": 9.930686830497795e-06,
      "loss": 1.6136,
      "step": 198
    },
    {
      "epoch": 0.22435174746335965,
      "grad_norm": 4.90625,
      "learning_rate": 9.930017022886325e-06,
      "loss": 1.7028,
      "step": 199
    },
    {
      "epoch": 0.2254791431792559,
      "grad_norm": 4.1875,
      "learning_rate": 9.929346454706032e-06,
      "loss": 1.4346,
      "step": 200
    },
    {
      "epoch": 0.2266065388951522,
      "grad_norm": 4.3125,
      "learning_rate": 9.928675124660734e-06,
      "loss": 1.6348,
      "step": 201
    },
    {
      "epoch": 0.2277339346110485,
      "grad_norm": 2.4375,
      "learning_rate": 9.928003031451308e-06,
      "loss": 1.823,
      "step": 202
    },
    {
      "epoch": 0.22886133032694475,
      "grad_norm": 2.8125,
      "learning_rate": 9.927330173775673e-06,
      "loss": 1.6165,
      "step": 203
    },
    {
      "epoch": 0.22998872604284104,
      "grad_norm": 3.0625,
      "learning_rate": 9.926656550328782e-06,
      "loss": 1.5784,
      "step": 204
    },
    {
      "epoch": 0.23111612175873733,
      "grad_norm": 3.5,
      "learning_rate": 9.92598215980262e-06,
      "loss": 1.662,
      "step": 205
    },
    {
      "epoch": 0.2322435174746336,
      "grad_norm": 2.875,
      "learning_rate": 9.92530700088619e-06,
      "loss": 1.7793,
      "step": 206
    },
    {
      "epoch": 0.23337091319052988,
      "grad_norm": 3.03125,
      "learning_rate": 9.924631072265503e-06,
      "loss": 1.7705,
      "step": 207
    },
    {
      "epoch": 0.23449830890642615,
      "grad_norm": 3.546875,
      "learning_rate": 9.923954372623576e-06,
      "loss": 1.7613,
      "step": 208
    },
    {
      "epoch": 0.23562570462232243,
      "grad_norm": 3.359375,
      "learning_rate": 9.923276900640417e-06,
      "loss": 1.7128,
      "step": 209
    },
    {
      "epoch": 0.23675310033821872,
      "grad_norm": 3.171875,
      "learning_rate": 9.922598654993023e-06,
      "loss": 1.6944,
      "step": 210
    },
    {
      "epoch": 0.237880496054115,
      "grad_norm": 3.40625,
      "learning_rate": 9.921919634355362e-06,
      "loss": 1.4668,
      "step": 211
    },
    {
      "epoch": 0.23900789177001128,
      "grad_norm": 3.59375,
      "learning_rate": 9.921239837398375e-06,
      "loss": 1.7813,
      "step": 212
    },
    {
      "epoch": 0.24013528748590757,
      "grad_norm": 3.515625,
      "learning_rate": 9.92055926278996e-06,
      "loss": 1.6753,
      "step": 213
    },
    {
      "epoch": 0.24126268320180383,
      "grad_norm": 3.703125,
      "learning_rate": 9.919877909194964e-06,
      "loss": 1.6827,
      "step": 214
    },
    {
      "epoch": 0.24239007891770012,
      "grad_norm": 4.75,
      "learning_rate": 9.91919577527518e-06,
      "loss": 1.9023,
      "step": 215
    },
    {
      "epoch": 0.24351747463359638,
      "grad_norm": 3.53125,
      "learning_rate": 9.918512859689332e-06,
      "loss": 1.5267,
      "step": 216
    },
    {
      "epoch": 0.24464487034949267,
      "grad_norm": 3.625,
      "learning_rate": 9.917829161093064e-06,
      "loss": 1.7207,
      "step": 217
    },
    {
      "epoch": 0.24577226606538896,
      "grad_norm": 3.703125,
      "learning_rate": 9.917144678138942e-06,
      "loss": 1.4898,
      "step": 218
    },
    {
      "epoch": 0.24689966178128522,
      "grad_norm": 4.0625,
      "learning_rate": 9.916459409476437e-06,
      "loss": 1.7543,
      "step": 219
    },
    {
      "epoch": 0.2480270574971815,
      "grad_norm": 4.3125,
      "learning_rate": 9.915773353751915e-06,
      "loss": 1.7978,
      "step": 220
    },
    {
      "epoch": 0.2491544532130778,
      "grad_norm": 3.9375,
      "learning_rate": 9.915086509608632e-06,
      "loss": 1.9315,
      "step": 221
    },
    {
      "epoch": 0.2502818489289741,
      "grad_norm": 4.0,
      "learning_rate": 9.914398875686726e-06,
      "loss": 1.5695,
      "step": 222
    },
    {
      "epoch": 0.25140924464487036,
      "grad_norm": 3.953125,
      "learning_rate": 9.913710450623204e-06,
      "loss": 1.8259,
      "step": 223
    },
    {
      "epoch": 0.2525366403607666,
      "grad_norm": 3.25,
      "learning_rate": 9.913021233051932e-06,
      "loss": 1.3427,
      "step": 224
    },
    {
      "epoch": 0.25366403607666294,
      "grad_norm": 3.578125,
      "learning_rate": 9.912331221603636e-06,
      "loss": 1.6302,
      "step": 225
    },
    {
      "epoch": 0.2547914317925592,
      "grad_norm": 3.71875,
      "learning_rate": 9.911640414905878e-06,
      "loss": 1.7424,
      "step": 226
    },
    {
      "epoch": 0.25591882750845546,
      "grad_norm": 3.59375,
      "learning_rate": 9.910948811583061e-06,
      "loss": 1.5013,
      "step": 227
    },
    {
      "epoch": 0.2570462232243517,
      "grad_norm": 4.15625,
      "learning_rate": 9.910256410256412e-06,
      "loss": 1.9034,
      "step": 228
    },
    {
      "epoch": 0.25817361894024804,
      "grad_norm": 3.171875,
      "learning_rate": 9.909563209543969e-06,
      "loss": 1.3779,
      "step": 229
    },
    {
      "epoch": 0.2593010146561443,
      "grad_norm": 4.0625,
      "learning_rate": 9.908869208060583e-06,
      "loss": 1.5715,
      "step": 230
    },
    {
      "epoch": 0.26042841037204056,
      "grad_norm": 4.0,
      "learning_rate": 9.908174404417903e-06,
      "loss": 1.7055,
      "step": 231
    },
    {
      "epoch": 0.2615558060879369,
      "grad_norm": 4.03125,
      "learning_rate": 9.907478797224365e-06,
      "loss": 1.6714,
      "step": 232
    },
    {
      "epoch": 0.26268320180383314,
      "grad_norm": 3.90625,
      "learning_rate": 9.906782385085183e-06,
      "loss": 1.5676,
      "step": 233
    },
    {
      "epoch": 0.2638105975197294,
      "grad_norm": 3.4375,
      "learning_rate": 9.906085166602342e-06,
      "loss": 1.595,
      "step": 234
    },
    {
      "epoch": 0.2649379932356257,
      "grad_norm": 3.671875,
      "learning_rate": 9.90538714037459e-06,
      "loss": 1.5537,
      "step": 235
    },
    {
      "epoch": 0.266065388951522,
      "grad_norm": 3.71875,
      "learning_rate": 9.904688304997424e-06,
      "loss": 1.7156,
      "step": 236
    },
    {
      "epoch": 0.26719278466741825,
      "grad_norm": 3.9375,
      "learning_rate": 9.903988659063084e-06,
      "loss": 1.573,
      "step": 237
    },
    {
      "epoch": 0.26832018038331457,
      "grad_norm": 3.4375,
      "learning_rate": 9.903288201160543e-06,
      "loss": 1.5044,
      "step": 238
    },
    {
      "epoch": 0.26944757609921083,
      "grad_norm": 4.0625,
      "learning_rate": 9.902586929875493e-06,
      "loss": 1.8448,
      "step": 239
    },
    {
      "epoch": 0.2705749718151071,
      "grad_norm": 4.21875,
      "learning_rate": 9.901884843790343e-06,
      "loss": 1.702,
      "step": 240
    },
    {
      "epoch": 0.2717023675310034,
      "grad_norm": 3.5,
      "learning_rate": 9.90118194148421e-06,
      "loss": 1.6115,
      "step": 241
    },
    {
      "epoch": 0.27282976324689967,
      "grad_norm": 4.65625,
      "learning_rate": 9.900478221532895e-06,
      "loss": 1.5315,
      "step": 242
    },
    {
      "epoch": 0.27395715896279593,
      "grad_norm": 3.03125,
      "learning_rate": 9.899773682508892e-06,
      "loss": 1.4657,
      "step": 243
    },
    {
      "epoch": 0.2750845546786922,
      "grad_norm": 3.296875,
      "learning_rate": 9.899068322981368e-06,
      "loss": 1.6866,
      "step": 244
    },
    {
      "epoch": 0.2762119503945885,
      "grad_norm": 4.0625,
      "learning_rate": 9.898362141516154e-06,
      "loss": 1.5703,
      "step": 245
    },
    {
      "epoch": 0.2773393461104848,
      "grad_norm": 4.0,
      "learning_rate": 9.897655136675737e-06,
      "loss": 1.8035,
      "step": 246
    },
    {
      "epoch": 0.27846674182638104,
      "grad_norm": 4.71875,
      "learning_rate": 9.89694730701925e-06,
      "loss": 1.7236,
      "step": 247
    },
    {
      "epoch": 0.27959413754227735,
      "grad_norm": 3.953125,
      "learning_rate": 9.896238651102465e-06,
      "loss": 1.734,
      "step": 248
    },
    {
      "epoch": 0.2807215332581736,
      "grad_norm": 3.671875,
      "learning_rate": 9.895529167477777e-06,
      "loss": 1.614,
      "step": 249
    },
    {
      "epoch": 0.2818489289740699,
      "grad_norm": 4.1875,
      "learning_rate": 9.894818854694197e-06,
      "loss": 1.6502,
      "step": 250
    },
    {
      "epoch": 0.2829763246899662,
      "grad_norm": 4.1875,
      "learning_rate": 9.894107711297344e-06,
      "loss": 1.7718,
      "step": 251
    },
    {
      "epoch": 0.28410372040586246,
      "grad_norm": 3.5625,
      "learning_rate": 9.893395735829434e-06,
      "loss": 1.5139,
      "step": 252
    },
    {
      "epoch": 0.2852311161217587,
      "grad_norm": 3.1875,
      "learning_rate": 9.89268292682927e-06,
      "loss": 1.549,
      "step": 253
    },
    {
      "epoch": 0.28635851183765504,
      "grad_norm": 3.71875,
      "learning_rate": 9.891969282832227e-06,
      "loss": 1.5342,
      "step": 254
    },
    {
      "epoch": 0.2874859075535513,
      "grad_norm": 3.546875,
      "learning_rate": 9.891254802370254e-06,
      "loss": 1.6014,
      "step": 255
    },
    {
      "epoch": 0.28861330326944756,
      "grad_norm": 3.46875,
      "learning_rate": 9.890539483971854e-06,
      "loss": 1.5965,
      "step": 256
    },
    {
      "epoch": 0.2897406989853439,
      "grad_norm": 4.3125,
      "learning_rate": 9.889823326162072e-06,
      "loss": 1.5295,
      "step": 257
    },
    {
      "epoch": 0.29086809470124014,
      "grad_norm": 3.5625,
      "learning_rate": 9.889106327462492e-06,
      "loss": 1.6969,
      "step": 258
    },
    {
      "epoch": 0.2919954904171364,
      "grad_norm": 3.90625,
      "learning_rate": 9.888388486391228e-06,
      "loss": 1.7436,
      "step": 259
    },
    {
      "epoch": 0.29312288613303267,
      "grad_norm": 3.5625,
      "learning_rate": 9.887669801462905e-06,
      "loss": 1.5908,
      "step": 260
    },
    {
      "epoch": 0.294250281848929,
      "grad_norm": 3.453125,
      "learning_rate": 9.886950271188657e-06,
      "loss": 1.7394,
      "step": 261
    },
    {
      "epoch": 0.29537767756482525,
      "grad_norm": 3.515625,
      "learning_rate": 9.88622989407611e-06,
      "loss": 1.2984,
      "step": 262
    },
    {
      "epoch": 0.2965050732807215,
      "grad_norm": 3.140625,
      "learning_rate": 9.885508668629376e-06,
      "loss": 1.7264,
      "step": 263
    },
    {
      "epoch": 0.2976324689966178,
      "grad_norm": 3.953125,
      "learning_rate": 9.884786593349045e-06,
      "loss": 1.5456,
      "step": 264
    },
    {
      "epoch": 0.2987598647125141,
      "grad_norm": 4.03125,
      "learning_rate": 9.884063666732169e-06,
      "loss": 1.5851,
      "step": 265
    },
    {
      "epoch": 0.29988726042841035,
      "grad_norm": 3.703125,
      "learning_rate": 9.883339887272253e-06,
      "loss": 1.7294,
      "step": 266
    },
    {
      "epoch": 0.30101465614430667,
      "grad_norm": 4.0,
      "learning_rate": 9.882615253459245e-06,
      "loss": 1.6875,
      "step": 267
    },
    {
      "epoch": 0.30214205186020293,
      "grad_norm": 3.625,
      "learning_rate": 9.881889763779528e-06,
      "loss": 1.7715,
      "step": 268
    },
    {
      "epoch": 0.3032694475760992,
      "grad_norm": 4.28125,
      "learning_rate": 9.88116341671591e-06,
      "loss": 1.6943,
      "step": 269
    },
    {
      "epoch": 0.3043968432919955,
      "grad_norm": 4.34375,
      "learning_rate": 9.880436210747604e-06,
      "loss": 1.8057,
      "step": 270
    },
    {
      "epoch": 0.3055242390078918,
      "grad_norm": 3.875,
      "learning_rate": 9.879708144350228e-06,
      "loss": 1.6853,
      "step": 271
    },
    {
      "epoch": 0.30665163472378804,
      "grad_norm": 3.4375,
      "learning_rate": 9.878979215995793e-06,
      "loss": 1.4301,
      "step": 272
    },
    {
      "epoch": 0.30777903043968435,
      "grad_norm": 2.984375,
      "learning_rate": 9.878249424152682e-06,
      "loss": 1.5873,
      "step": 273
    },
    {
      "epoch": 0.3089064261555806,
      "grad_norm": 4.03125,
      "learning_rate": 9.877518767285658e-06,
      "loss": 1.7891,
      "step": 274
    },
    {
      "epoch": 0.3100338218714769,
      "grad_norm": 3.484375,
      "learning_rate": 9.876787243855835e-06,
      "loss": 1.4832,
      "step": 275
    },
    {
      "epoch": 0.3111612175873732,
      "grad_norm": 4.0,
      "learning_rate": 9.876054852320675e-06,
      "loss": 1.6482,
      "step": 276
    },
    {
      "epoch": 0.31228861330326946,
      "grad_norm": 4.125,
      "learning_rate": 9.875321591133981e-06,
      "loss": 1.5463,
      "step": 277
    },
    {
      "epoch": 0.3134160090191657,
      "grad_norm": 3.671875,
      "learning_rate": 9.874587458745876e-06,
      "loss": 1.621,
      "step": 278
    },
    {
      "epoch": 0.314543404735062,
      "grad_norm": 4.0625,
      "learning_rate": 9.873852453602802e-06,
      "loss": 1.6288,
      "step": 279
    },
    {
      "epoch": 0.3156708004509583,
      "grad_norm": 3.921875,
      "learning_rate": 9.873116574147503e-06,
      "loss": 1.6868,
      "step": 280
    },
    {
      "epoch": 0.31679819616685456,
      "grad_norm": 3.953125,
      "learning_rate": 9.872379818819018e-06,
      "loss": 1.8418,
      "step": 281
    },
    {
      "epoch": 0.3179255918827508,
      "grad_norm": 3.796875,
      "learning_rate": 9.871642186052666e-06,
      "loss": 1.8009,
      "step": 282
    },
    {
      "epoch": 0.31905298759864714,
      "grad_norm": 3.8125,
      "learning_rate": 9.87090367428004e-06,
      "loss": 1.7026,
      "step": 283
    },
    {
      "epoch": 0.3201803833145434,
      "grad_norm": 3.140625,
      "learning_rate": 9.87016428192899e-06,
      "loss": 1.7357,
      "step": 284
    },
    {
      "epoch": 0.32130777903043967,
      "grad_norm": 3.71875,
      "learning_rate": 9.86942400742361e-06,
      "loss": 1.7802,
      "step": 285
    },
    {
      "epoch": 0.322435174746336,
      "grad_norm": 3.28125,
      "learning_rate": 9.868682849184243e-06,
      "loss": 1.5126,
      "step": 286
    },
    {
      "epoch": 0.32356257046223225,
      "grad_norm": 3.28125,
      "learning_rate": 9.867940805627448e-06,
      "loss": 1.6296,
      "step": 287
    },
    {
      "epoch": 0.3246899661781285,
      "grad_norm": 3.5625,
      "learning_rate": 9.867197875166004e-06,
      "loss": 1.6254,
      "step": 288
    },
    {
      "epoch": 0.3258173618940248,
      "grad_norm": 3.734375,
      "learning_rate": 9.86645405620889e-06,
      "loss": 1.6511,
      "step": 289
    },
    {
      "epoch": 0.3269447576099211,
      "grad_norm": 3.59375,
      "learning_rate": 9.865709347161282e-06,
      "loss": 1.5809,
      "step": 290
    },
    {
      "epoch": 0.32807215332581735,
      "grad_norm": 3.296875,
      "learning_rate": 9.864963746424534e-06,
      "loss": 1.6914,
      "step": 291
    },
    {
      "epoch": 0.32919954904171367,
      "grad_norm": 3.109375,
      "learning_rate": 9.864217252396166e-06,
      "loss": 1.5188,
      "step": 292
    },
    {
      "epoch": 0.33032694475760993,
      "grad_norm": 3.234375,
      "learning_rate": 9.863469863469864e-06,
      "loss": 1.4525,
      "step": 293
    },
    {
      "epoch": 0.3314543404735062,
      "grad_norm": 3.328125,
      "learning_rate": 9.862721578035453e-06,
      "loss": 1.8035,
      "step": 294
    },
    {
      "epoch": 0.33258173618940245,
      "grad_norm": 3.015625,
      "learning_rate": 9.861972394478897e-06,
      "loss": 1.6225,
      "step": 295
    },
    {
      "epoch": 0.3337091319052988,
      "grad_norm": 3.890625,
      "learning_rate": 9.86122231118228e-06,
      "loss": 1.6615,
      "step": 296
    },
    {
      "epoch": 0.33483652762119503,
      "grad_norm": 4.125,
      "learning_rate": 9.8604713265238e-06,
      "loss": 1.9232,
      "step": 297
    },
    {
      "epoch": 0.3359639233370913,
      "grad_norm": 3.640625,
      "learning_rate": 9.859719438877757e-06,
      "loss": 1.5466,
      "step": 298
    },
    {
      "epoch": 0.3370913190529876,
      "grad_norm": 3.6875,
      "learning_rate": 9.858966646614532e-06,
      "loss": 1.5936,
      "step": 299
    },
    {
      "epoch": 0.3382187147688839,
      "grad_norm": 3.875,
      "learning_rate": 9.85821294810059e-06,
      "loss": 1.7918,
      "step": 300
    },
    {
      "epoch": 0.33934611048478014,
      "grad_norm": 3.625,
      "learning_rate": 9.857458341698455e-06,
      "loss": 1.6119,
      "step": 301
    },
    {
      "epoch": 0.34047350620067646,
      "grad_norm": 3.59375,
      "learning_rate": 9.856702825766707e-06,
      "loss": 1.5585,
      "step": 302
    },
    {
      "epoch": 0.3416009019165727,
      "grad_norm": 3.640625,
      "learning_rate": 9.855946398659967e-06,
      "loss": 1.6617,
      "step": 303
    },
    {
      "epoch": 0.342728297632469,
      "grad_norm": 3.640625,
      "learning_rate": 9.855189058728882e-06,
      "loss": 1.6031,
      "step": 304
    },
    {
      "epoch": 0.3438556933483653,
      "grad_norm": 3.578125,
      "learning_rate": 9.854430804320119e-06,
      "loss": 1.7147,
      "step": 305
    },
    {
      "epoch": 0.34498308906426156,
      "grad_norm": 3.671875,
      "learning_rate": 9.853671633776347e-06,
      "loss": 1.5599,
      "step": 306
    },
    {
      "epoch": 0.3461104847801578,
      "grad_norm": 3.734375,
      "learning_rate": 9.85291154543623e-06,
      "loss": 1.7042,
      "step": 307
    },
    {
      "epoch": 0.34723788049605414,
      "grad_norm": 3.578125,
      "learning_rate": 9.852150537634409e-06,
      "loss": 1.5787,
      "step": 308
    },
    {
      "epoch": 0.3483652762119504,
      "grad_norm": 3.765625,
      "learning_rate": 9.8513886087015e-06,
      "loss": 1.6713,
      "step": 309
    },
    {
      "epoch": 0.34949267192784667,
      "grad_norm": 3.453125,
      "learning_rate": 9.85062575696407e-06,
      "loss": 1.6092,
      "step": 310
    },
    {
      "epoch": 0.3506200676437429,
      "grad_norm": 3.25,
      "learning_rate": 9.849861980744634e-06,
      "loss": 1.3859,
      "step": 311
    },
    {
      "epoch": 0.35174746335963925,
      "grad_norm": 3.078125,
      "learning_rate": 9.849097278361627e-06,
      "loss": 1.478,
      "step": 312
    },
    {
      "epoch": 0.3528748590755355,
      "grad_norm": 3.359375,
      "learning_rate": 9.848331648129425e-06,
      "loss": 1.6706,
      "step": 313
    },
    {
      "epoch": 0.35400225479143177,
      "grad_norm": 3.609375,
      "learning_rate": 9.847565088358289e-06,
      "loss": 1.6807,
      "step": 314
    },
    {
      "epoch": 0.3551296505073281,
      "grad_norm": 3.109375,
      "learning_rate": 9.846797597354391e-06,
      "loss": 1.5461,
      "step": 315
    },
    {
      "epoch": 0.35625704622322435,
      "grad_norm": 2.96875,
      "learning_rate": 9.846029173419774e-06,
      "loss": 1.4548,
      "step": 316
    },
    {
      "epoch": 0.3573844419391206,
      "grad_norm": 3.578125,
      "learning_rate": 9.845259814852356e-06,
      "loss": 1.6827,
      "step": 317
    },
    {
      "epoch": 0.35851183765501693,
      "grad_norm": 3.0,
      "learning_rate": 9.84448951994591e-06,
      "loss": 1.6626,
      "step": 318
    },
    {
      "epoch": 0.3596392333709132,
      "grad_norm": 3.34375,
      "learning_rate": 9.843718286990056e-06,
      "loss": 1.8132,
      "step": 319
    },
    {
      "epoch": 0.36076662908680945,
      "grad_norm": 3.109375,
      "learning_rate": 9.842946114270241e-06,
      "loss": 1.5748,
      "step": 320
    },
    {
      "epoch": 0.36189402480270577,
      "grad_norm": 3.296875,
      "learning_rate": 9.842173000067737e-06,
      "loss": 1.5703,
      "step": 321
    },
    {
      "epoch": 0.36302142051860203,
      "grad_norm": 3.75,
      "learning_rate": 9.84139894265962e-06,
      "loss": 1.529,
      "step": 322
    },
    {
      "epoch": 0.3641488162344983,
      "grad_norm": 3.671875,
      "learning_rate": 9.840623940318752e-06,
      "loss": 1.6112,
      "step": 323
    },
    {
      "epoch": 0.3652762119503946,
      "grad_norm": 4.125,
      "learning_rate": 9.83984799131379e-06,
      "loss": 1.9493,
      "step": 324
    },
    {
      "epoch": 0.3664036076662909,
      "grad_norm": 3.359375,
      "learning_rate": 9.839071093909148e-06,
      "loss": 1.5269,
      "step": 325
    },
    {
      "epoch": 0.36753100338218714,
      "grad_norm": 3.71875,
      "learning_rate": 9.838293246364996e-06,
      "loss": 1.7027,
      "step": 326
    },
    {
      "epoch": 0.3686583990980834,
      "grad_norm": 3.84375,
      "learning_rate": 9.83751444693725e-06,
      "loss": 1.7282,
      "step": 327
    },
    {
      "epoch": 0.3697857948139797,
      "grad_norm": 3.21875,
      "learning_rate": 9.836734693877552e-06,
      "loss": 1.3475,
      "step": 328
    },
    {
      "epoch": 0.370913190529876,
      "grad_norm": 3.328125,
      "learning_rate": 9.83595398543326e-06,
      "loss": 1.6058,
      "step": 329
    },
    {
      "epoch": 0.37204058624577224,
      "grad_norm": 3.890625,
      "learning_rate": 9.835172319847434e-06,
      "loss": 1.519,
      "step": 330
    },
    {
      "epoch": 0.37316798196166856,
      "grad_norm": 3.265625,
      "learning_rate": 9.834389695358823e-06,
      "loss": 1.6139,
      "step": 331
    },
    {
      "epoch": 0.3742953776775648,
      "grad_norm": 3.203125,
      "learning_rate": 9.833606110201856e-06,
      "loss": 1.6625,
      "step": 332
    },
    {
      "epoch": 0.3754227733934611,
      "grad_norm": 4.125,
      "learning_rate": 9.83282156260662e-06,
      "loss": 1.6857,
      "step": 333
    },
    {
      "epoch": 0.3765501691093574,
      "grad_norm": 3.390625,
      "learning_rate": 9.832036050798853e-06,
      "loss": 1.634,
      "step": 334
    },
    {
      "epoch": 0.37767756482525366,
      "grad_norm": 3.359375,
      "learning_rate": 9.831249572999932e-06,
      "loss": 1.6545,
      "step": 335
    },
    {
      "epoch": 0.3788049605411499,
      "grad_norm": 3.75,
      "learning_rate": 9.830462127426853e-06,
      "loss": 1.5383,
      "step": 336
    },
    {
      "epoch": 0.37993235625704624,
      "grad_norm": 3.25,
      "learning_rate": 9.829673712292223e-06,
      "loss": 1.6299,
      "step": 337
    },
    {
      "epoch": 0.3810597519729425,
      "grad_norm": 3.15625,
      "learning_rate": 9.828884325804245e-06,
      "loss": 1.5428,
      "step": 338
    },
    {
      "epoch": 0.38218714768883877,
      "grad_norm": 3.90625,
      "learning_rate": 9.828093966166702e-06,
      "loss": 1.5365,
      "step": 339
    },
    {
      "epoch": 0.3833145434047351,
      "grad_norm": 3.59375,
      "learning_rate": 9.827302631578949e-06,
      "loss": 1.7071,
      "step": 340
    },
    {
      "epoch": 0.38444193912063135,
      "grad_norm": 2.875,
      "learning_rate": 9.826510320235892e-06,
      "loss": 1.3473,
      "step": 341
    },
    {
      "epoch": 0.3855693348365276,
      "grad_norm": 2.890625,
      "learning_rate": 9.825717030327982e-06,
      "loss": 1.6839,
      "step": 342
    },
    {
      "epoch": 0.38669673055242393,
      "grad_norm": 3.46875,
      "learning_rate": 9.824922760041196e-06,
      "loss": 1.6082,
      "step": 343
    },
    {
      "epoch": 0.3878241262683202,
      "grad_norm": 3.46875,
      "learning_rate": 9.824127507557021e-06,
      "loss": 1.5861,
      "step": 344
    },
    {
      "epoch": 0.38895152198421645,
      "grad_norm": 3.71875,
      "learning_rate": 9.823331271052452e-06,
      "loss": 1.692,
      "step": 345
    },
    {
      "epoch": 0.3900789177001127,
      "grad_norm": 3.859375,
      "learning_rate": 9.82253404869996e-06,
      "loss": 1.7168,
      "step": 346
    },
    {
      "epoch": 0.39120631341600903,
      "grad_norm": 3.8125,
      "learning_rate": 9.821735838667494e-06,
      "loss": 1.6199,
      "step": 347
    },
    {
      "epoch": 0.3923337091319053,
      "grad_norm": 3.453125,
      "learning_rate": 9.820936639118459e-06,
      "loss": 1.4607,
      "step": 348
    },
    {
      "epoch": 0.39346110484780156,
      "grad_norm": 3.375,
      "learning_rate": 9.820136448211703e-06,
      "loss": 1.454,
      "step": 349
    },
    {
      "epoch": 0.3945885005636979,
      "grad_norm": 4.125,
      "learning_rate": 9.819335264101504e-06,
      "loss": 1.691,
      "step": 350
    },
    {
      "epoch": 0.39571589627959414,
      "grad_norm": 3.3125,
      "learning_rate": 9.818533084937556e-06,
      "loss": 1.6374,
      "step": 351
    },
    {
      "epoch": 0.3968432919954904,
      "grad_norm": 3.875,
      "learning_rate": 9.817729908864955e-06,
      "loss": 1.7745,
      "step": 352
    },
    {
      "epoch": 0.3979706877113867,
      "grad_norm": 3.421875,
      "learning_rate": 9.816925734024179e-06,
      "loss": 1.6846,
      "step": 353
    },
    {
      "epoch": 0.399098083427283,
      "grad_norm": 2.875,
      "learning_rate": 9.816120558551085e-06,
      "loss": 1.5797,
      "step": 354
    },
    {
      "epoch": 0.40022547914317924,
      "grad_norm": 3.0625,
      "learning_rate": 9.815314380576883e-06,
      "loss": 1.7882,
      "step": 355
    },
    {
      "epoch": 0.40135287485907556,
      "grad_norm": 3.171875,
      "learning_rate": 9.814507198228129e-06,
      "loss": 1.7044,
      "step": 356
    },
    {
      "epoch": 0.4024802705749718,
      "grad_norm": 3.109375,
      "learning_rate": 9.813699009626705e-06,
      "loss": 1.5796,
      "step": 357
    },
    {
      "epoch": 0.4036076662908681,
      "grad_norm": 2.890625,
      "learning_rate": 9.812889812889814e-06,
      "loss": 1.7575,
      "step": 358
    },
    {
      "epoch": 0.4047350620067644,
      "grad_norm": 2.59375,
      "learning_rate": 9.812079606129951e-06,
      "loss": 1.5751,
      "step": 359
    },
    {
      "epoch": 0.40586245772266066,
      "grad_norm": 3.671875,
      "learning_rate": 9.8112683874549e-06,
      "loss": 1.6335,
      "step": 360
    },
    {
      "epoch": 0.4069898534385569,
      "grad_norm": 3.453125,
      "learning_rate": 9.810456154967715e-06,
      "loss": 1.774,
      "step": 361
    },
    {
      "epoch": 0.4081172491544532,
      "grad_norm": 3.359375,
      "learning_rate": 9.80964290676671e-06,
      "loss": 1.7743,
      "step": 362
    },
    {
      "epoch": 0.4092446448703495,
      "grad_norm": 3.234375,
      "learning_rate": 9.80882864094543e-06,
      "loss": 1.5289,
      "step": 363
    },
    {
      "epoch": 0.41037204058624577,
      "grad_norm": 3.5,
      "learning_rate": 9.808013355592656e-06,
      "loss": 1.5161,
      "step": 364
    },
    {
      "epoch": 0.41149943630214203,
      "grad_norm": 3.21875,
      "learning_rate": 9.807197048792372e-06,
      "loss": 1.6062,
      "step": 365
    },
    {
      "epoch": 0.41262683201803835,
      "grad_norm": 3.40625,
      "learning_rate": 9.806379718623765e-06,
      "loss": 1.6808,
      "step": 366
    },
    {
      "epoch": 0.4137542277339346,
      "grad_norm": 3.71875,
      "learning_rate": 9.805561363161197e-06,
      "loss": 1.651,
      "step": 367
    },
    {
      "epoch": 0.41488162344983087,
      "grad_norm": 3.09375,
      "learning_rate": 9.8047419804742e-06,
      "loss": 1.4252,
      "step": 368
    },
    {
      "epoch": 0.4160090191657272,
      "grad_norm": 3.4375,
      "learning_rate": 9.803921568627453e-06,
      "loss": 1.5857,
      "step": 369
    },
    {
      "epoch": 0.41713641488162345,
      "grad_norm": 3.25,
      "learning_rate": 9.803100125680771e-06,
      "loss": 1.6415,
      "step": 370
    },
    {
      "epoch": 0.4182638105975197,
      "grad_norm": 4.46875,
      "learning_rate": 9.802277649689093e-06,
      "loss": 1.6668,
      "step": 371
    },
    {
      "epoch": 0.41939120631341603,
      "grad_norm": 3.0625,
      "learning_rate": 9.80145413870246e-06,
      "loss": 1.6097,
      "step": 372
    },
    {
      "epoch": 0.4205186020293123,
      "grad_norm": 3.015625,
      "learning_rate": 9.800629590766002e-06,
      "loss": 1.4352,
      "step": 373
    },
    {
      "epoch": 0.42164599774520856,
      "grad_norm": 3.78125,
      "learning_rate": 9.799804003919921e-06,
      "loss": 1.8216,
      "step": 374
    },
    {
      "epoch": 0.4227733934611049,
      "grad_norm": 4.09375,
      "learning_rate": 9.798977376199483e-06,
      "loss": 1.756,
      "step": 375
    },
    {
      "epoch": 0.42390078917700114,
      "grad_norm": 4.0625,
      "learning_rate": 9.798149705634989e-06,
      "loss": 1.7512,
      "step": 376
    },
    {
      "epoch": 0.4250281848928974,
      "grad_norm": 3.640625,
      "learning_rate": 9.797320990251772e-06,
      "loss": 1.6677,
      "step": 377
    },
    {
      "epoch": 0.42615558060879366,
      "grad_norm": 3.71875,
      "learning_rate": 9.796491228070176e-06,
      "loss": 1.6386,
      "step": 378
    },
    {
      "epoch": 0.42728297632469,
      "grad_norm": 3.046875,
      "learning_rate": 9.795660417105541e-06,
      "loss": 1.6434,
      "step": 379
    },
    {
      "epoch": 0.42841037204058624,
      "grad_norm": 2.90625,
      "learning_rate": 9.794828555368185e-06,
      "loss": 1.7351,
      "step": 380
    },
    {
      "epoch": 0.4295377677564825,
      "grad_norm": 3.703125,
      "learning_rate": 9.793995640863391e-06,
      "loss": 1.6731,
      "step": 381
    },
    {
      "epoch": 0.4306651634723788,
      "grad_norm": 3.546875,
      "learning_rate": 9.79316167159139e-06,
      "loss": 1.6669,
      "step": 382
    },
    {
      "epoch": 0.4317925591882751,
      "grad_norm": 3.21875,
      "learning_rate": 9.792326645547344e-06,
      "loss": 1.5247,
      "step": 383
    },
    {
      "epoch": 0.43291995490417134,
      "grad_norm": 3.4375,
      "learning_rate": 9.79149056072133e-06,
      "loss": 1.6837,
      "step": 384
    },
    {
      "epoch": 0.43404735062006766,
      "grad_norm": 3.453125,
      "learning_rate": 9.79065341509833e-06,
      "loss": 1.647,
      "step": 385
    },
    {
      "epoch": 0.4351747463359639,
      "grad_norm": 3.84375,
      "learning_rate": 9.789815206658204e-06,
      "loss": 1.5363,
      "step": 386
    },
    {
      "epoch": 0.4363021420518602,
      "grad_norm": 3.765625,
      "learning_rate": 9.78897593337568e-06,
      "loss": 1.5102,
      "step": 387
    },
    {
      "epoch": 0.4374295377677565,
      "grad_norm": 3.375,
      "learning_rate": 9.78813559322034e-06,
      "loss": 1.6933,
      "step": 388
    },
    {
      "epoch": 0.43855693348365277,
      "grad_norm": 3.046875,
      "learning_rate": 9.787294184156596e-06,
      "loss": 1.5482,
      "step": 389
    },
    {
      "epoch": 0.43968432919954903,
      "grad_norm": 3.609375,
      "learning_rate": 9.786451704143686e-06,
      "loss": 1.6291,
      "step": 390
    },
    {
      "epoch": 0.44081172491544535,
      "grad_norm": 3.453125,
      "learning_rate": 9.78560815113564e-06,
      "loss": 1.7273,
      "step": 391
    },
    {
      "epoch": 0.4419391206313416,
      "grad_norm": 3.453125,
      "learning_rate": 9.78476352308128e-06,
      "loss": 1.5634,
      "step": 392
    },
    {
      "epoch": 0.44306651634723787,
      "grad_norm": 3.203125,
      "learning_rate": 9.783917817924195e-06,
      "loss": 1.5487,
      "step": 393
    },
    {
      "epoch": 0.44419391206313413,
      "grad_norm": 3.515625,
      "learning_rate": 9.783071033602722e-06,
      "loss": 1.6889,
      "step": 394
    },
    {
      "epoch": 0.44532130777903045,
      "grad_norm": 3.40625,
      "learning_rate": 9.78222316804994e-06,
      "loss": 1.5318,
      "step": 395
    },
    {
      "epoch": 0.4464487034949267,
      "grad_norm": 3.78125,
      "learning_rate": 9.781374219193641e-06,
      "loss": 1.7132,
      "step": 396
    },
    {
      "epoch": 0.447576099210823,
      "grad_norm": 3.3125,
      "learning_rate": 9.780524184956318e-06,
      "loss": 1.547,
      "step": 397
    },
    {
      "epoch": 0.4487034949267193,
      "grad_norm": 2.9375,
      "learning_rate": 9.779673063255153e-06,
      "loss": 1.6435,
      "step": 398
    },
    {
      "epoch": 0.44983089064261556,
      "grad_norm": 3.3125,
      "learning_rate": 9.778820852001991e-06,
      "loss": 1.4575,
      "step": 399
    },
    {
      "epoch": 0.4509582863585118,
      "grad_norm": 3.53125,
      "learning_rate": 9.777967549103331e-06,
      "loss": 1.601,
      "step": 400
    },
    {
      "epoch": 0.45208568207440814,
      "grad_norm": 3.390625,
      "learning_rate": 9.777113152460302e-06,
      "loss": 1.7805,
      "step": 401
    },
    {
      "epoch": 0.4532130777903044,
      "grad_norm": 3.046875,
      "learning_rate": 9.776257659968649e-06,
      "loss": 1.6517,
      "step": 402
    },
    {
      "epoch": 0.45434047350620066,
      "grad_norm": 2.875,
      "learning_rate": 9.775401069518717e-06,
      "loss": 1.5515,
      "step": 403
    },
    {
      "epoch": 0.455467869222097,
      "grad_norm": 3.015625,
      "learning_rate": 9.774543378995435e-06,
      "loss": 1.5371,
      "step": 404
    },
    {
      "epoch": 0.45659526493799324,
      "grad_norm": 3.3125,
      "learning_rate": 9.77368458627829e-06,
      "loss": 1.8825,
      "step": 405
    },
    {
      "epoch": 0.4577226606538895,
      "grad_norm": 2.90625,
      "learning_rate": 9.772824689241322e-06,
      "loss": 1.6773,
      "step": 406
    },
    {
      "epoch": 0.4588500563697858,
      "grad_norm": 2.890625,
      "learning_rate": 9.771963685753093e-06,
      "loss": 1.6354,
      "step": 407
    },
    {
      "epoch": 0.4599774520856821,
      "grad_norm": 3.1875,
      "learning_rate": 9.771101573676682e-06,
      "loss": 1.6786,
      "step": 408
    },
    {
      "epoch": 0.46110484780157834,
      "grad_norm": 3.484375,
      "learning_rate": 9.770238350869659e-06,
      "loss": 1.7191,
      "step": 409
    },
    {
      "epoch": 0.46223224351747466,
      "grad_norm": 3.328125,
      "learning_rate": 9.76937401518407e-06,
      "loss": 1.4773,
      "step": 410
    },
    {
      "epoch": 0.4633596392333709,
      "grad_norm": 2.84375,
      "learning_rate": 9.768508564466424e-06,
      "loss": 1.4653,
      "step": 411
    },
    {
      "epoch": 0.4644870349492672,
      "grad_norm": 3.375,
      "learning_rate": 9.767641996557659e-06,
      "loss": 1.5449,
      "step": 412
    },
    {
      "epoch": 0.46561443066516345,
      "grad_norm": 3.484375,
      "learning_rate": 9.766774309293147e-06,
      "loss": 1.7697,
      "step": 413
    },
    {
      "epoch": 0.46674182638105977,
      "grad_norm": 3.859375,
      "learning_rate": 9.765905500502657e-06,
      "loss": 1.58,
      "step": 414
    },
    {
      "epoch": 0.46786922209695603,
      "grad_norm": 3.140625,
      "learning_rate": 9.765035568010348e-06,
      "loss": 1.5477,
      "step": 415
    },
    {
      "epoch": 0.4689966178128523,
      "grad_norm": 2.875,
      "learning_rate": 9.764164509634744e-06,
      "loss": 1.5618,
      "step": 416
    },
    {
      "epoch": 0.4701240135287486,
      "grad_norm": 2.59375,
      "learning_rate": 9.76329232318872e-06,
      "loss": 1.3428,
      "step": 417
    },
    {
      "epoch": 0.47125140924464487,
      "grad_norm": 3.3125,
      "learning_rate": 9.762419006479482e-06,
      "loss": 1.7728,
      "step": 418
    },
    {
      "epoch": 0.47237880496054113,
      "grad_norm": 4.09375,
      "learning_rate": 9.761544557308551e-06,
      "loss": 1.6734,
      "step": 419
    },
    {
      "epoch": 0.47350620067643745,
      "grad_norm": 6.65625,
      "learning_rate": 9.760668973471743e-06,
      "loss": 1.4851,
      "step": 420
    },
    {
      "epoch": 0.4746335963923337,
      "grad_norm": 3.484375,
      "learning_rate": 9.759792252759144e-06,
      "loss": 1.665,
      "step": 421
    },
    {
      "epoch": 0.47576099210823,
      "grad_norm": 3.375,
      "learning_rate": 9.758914392955104e-06,
      "loss": 1.529,
      "step": 422
    },
    {
      "epoch": 0.4768883878241263,
      "grad_norm": 3.296875,
      "learning_rate": 9.75803539183821e-06,
      "loss": 1.6007,
      "step": 423
    },
    {
      "epoch": 0.47801578354002255,
      "grad_norm": 3.890625,
      "learning_rate": 9.757155247181269e-06,
      "loss": 1.5566,
      "step": 424
    },
    {
      "epoch": 0.4791431792559188,
      "grad_norm": 3.1875,
      "learning_rate": 9.756273956751286e-06,
      "loss": 1.5672,
      "step": 425
    },
    {
      "epoch": 0.48027057497181513,
      "grad_norm": 3.890625,
      "learning_rate": 9.755391518309452e-06,
      "loss": 1.7351,
      "step": 426
    },
    {
      "epoch": 0.4813979706877114,
      "grad_norm": 7.71875,
      "learning_rate": 9.754507929611125e-06,
      "loss": 1.558,
      "step": 427
    },
    {
      "epoch": 0.48252536640360766,
      "grad_norm": 3.40625,
      "learning_rate": 9.753623188405798e-06,
      "loss": 1.6279,
      "step": 428
    },
    {
      "epoch": 0.4836527621195039,
      "grad_norm": 3.609375,
      "learning_rate": 9.752737292437097e-06,
      "loss": 1.6238,
      "step": 429
    },
    {
      "epoch": 0.48478015783540024,
      "grad_norm": 3.875,
      "learning_rate": 9.751850239442752e-06,
      "loss": 1.4702,
      "step": 430
    },
    {
      "epoch": 0.4859075535512965,
      "grad_norm": 3.3125,
      "learning_rate": 9.750962027154578e-06,
      "loss": 1.7224,
      "step": 431
    },
    {
      "epoch": 0.48703494926719276,
      "grad_norm": 3.125,
      "learning_rate": 9.75007265329846e-06,
      "loss": 1.5142,
      "step": 432
    },
    {
      "epoch": 0.4881623449830891,
      "grad_norm": 3.03125,
      "learning_rate": 9.74918211559433e-06,
      "loss": 1.687,
      "step": 433
    },
    {
      "epoch": 0.48928974069898534,
      "grad_norm": 3.828125,
      "learning_rate": 9.748290411756148e-06,
      "loss": 1.7972,
      "step": 434
    },
    {
      "epoch": 0.4904171364148816,
      "grad_norm": 3.4375,
      "learning_rate": 9.747397539491884e-06,
      "loss": 1.523,
      "step": 435
    },
    {
      "epoch": 0.4915445321307779,
      "grad_norm": 4.0625,
      "learning_rate": 9.746503496503497e-06,
      "loss": 1.5665,
      "step": 436
    },
    {
      "epoch": 0.4926719278466742,
      "grad_norm": 3.96875,
      "learning_rate": 9.745608280486916e-06,
      "loss": 1.6972,
      "step": 437
    },
    {
      "epoch": 0.49379932356257045,
      "grad_norm": 3.90625,
      "learning_rate": 9.744711889132021e-06,
      "loss": 1.5423,
      "step": 438
    },
    {
      "epoch": 0.49492671927846676,
      "grad_norm": 3.1875,
      "learning_rate": 9.743814320122619e-06,
      "loss": 1.5759,
      "step": 439
    },
    {
      "epoch": 0.496054114994363,
      "grad_norm": 4.09375,
      "learning_rate": 9.742915571136431e-06,
      "loss": 1.6851,
      "step": 440
    },
    {
      "epoch": 0.4971815107102593,
      "grad_norm": 3.515625,
      "learning_rate": 9.742015639845063e-06,
      "loss": 1.5385,
      "step": 441
    },
    {
      "epoch": 0.4983089064261556,
      "grad_norm": 4.21875,
      "learning_rate": 9.741114523913999e-06,
      "loss": 1.6369,
      "step": 442
    },
    {
      "epoch": 0.49943630214205187,
      "grad_norm": 3.703125,
      "learning_rate": 9.740212221002563e-06,
      "loss": 1.7675,
      "step": 443
    },
    {
      "epoch": 0.5005636978579482,
      "grad_norm": 3.796875,
      "learning_rate": 9.739308728763914e-06,
      "loss": 1.7441,
      "step": 444
    },
    {
      "epoch": 0.5016910935738444,
      "grad_norm": 3.453125,
      "learning_rate": 9.738404044845022e-06,
      "loss": 1.6207,
      "step": 445
    },
    {
      "epoch": 0.5028184892897407,
      "grad_norm": 3.859375,
      "learning_rate": 9.737498166886642e-06,
      "loss": 1.7834,
      "step": 446
    },
    {
      "epoch": 0.503945885005637,
      "grad_norm": 3.625,
      "learning_rate": 9.736591092523296e-06,
      "loss": 1.6147,
      "step": 447
    },
    {
      "epoch": 0.5050732807215332,
      "grad_norm": 3.578125,
      "learning_rate": 9.735682819383259e-06,
      "loss": 1.6568,
      "step": 448
    },
    {
      "epoch": 0.5062006764374295,
      "grad_norm": 4.0625,
      "learning_rate": 9.734773345088532e-06,
      "loss": 1.8356,
      "step": 449
    },
    {
      "epoch": 0.5073280721533259,
      "grad_norm": 3.40625,
      "learning_rate": 9.733862667254817e-06,
      "loss": 1.6871,
      "step": 450
    },
    {
      "epoch": 0.5084554678692221,
      "grad_norm": 4.0625,
      "learning_rate": 9.732950783491503e-06,
      "loss": 1.7135,
      "step": 451
    },
    {
      "epoch": 0.5095828635851184,
      "grad_norm": 3.703125,
      "learning_rate": 9.73203769140165e-06,
      "loss": 1.4805,
      "step": 452
    },
    {
      "epoch": 0.5107102593010147,
      "grad_norm": 3.015625,
      "learning_rate": 9.731123388581952e-06,
      "loss": 1.4773,
      "step": 453
    },
    {
      "epoch": 0.5118376550169109,
      "grad_norm": 3.734375,
      "learning_rate": 9.730207872622734e-06,
      "loss": 1.6121,
      "step": 454
    },
    {
      "epoch": 0.5129650507328072,
      "grad_norm": 3.953125,
      "learning_rate": 9.729291141107916e-06,
      "loss": 1.4887,
      "step": 455
    },
    {
      "epoch": 0.5140924464487034,
      "grad_norm": 3.4375,
      "learning_rate": 9.728373191615e-06,
      "loss": 1.3623,
      "step": 456
    },
    {
      "epoch": 0.5152198421645998,
      "grad_norm": 3.84375,
      "learning_rate": 9.727454021715045e-06,
      "loss": 1.5935,
      "step": 457
    },
    {
      "epoch": 0.5163472378804961,
      "grad_norm": 4.53125,
      "learning_rate": 9.726533628972654e-06,
      "loss": 1.5401,
      "step": 458
    },
    {
      "epoch": 0.5174746335963923,
      "grad_norm": 3.84375,
      "learning_rate": 9.725612010945937e-06,
      "loss": 1.774,
      "step": 459
    },
    {
      "epoch": 0.5186020293122886,
      "grad_norm": 3.640625,
      "learning_rate": 9.724689165186502e-06,
      "loss": 1.3738,
      "step": 460
    },
    {
      "epoch": 0.5197294250281849,
      "grad_norm": 3.984375,
      "learning_rate": 9.72376508923943e-06,
      "loss": 1.5934,
      "step": 461
    },
    {
      "epoch": 0.5208568207440811,
      "grad_norm": 3.28125,
      "learning_rate": 9.722839780643251e-06,
      "loss": 1.4424,
      "step": 462
    },
    {
      "epoch": 0.5219842164599775,
      "grad_norm": 4.40625,
      "learning_rate": 9.721913236929923e-06,
      "loss": 1.4921,
      "step": 463
    },
    {
      "epoch": 0.5231116121758738,
      "grad_norm": 3.421875,
      "learning_rate": 9.720985455624816e-06,
      "loss": 1.7314,
      "step": 464
    },
    {
      "epoch": 0.52423900789177,
      "grad_norm": 3.84375,
      "learning_rate": 9.720056434246678e-06,
      "loss": 1.6544,
      "step": 465
    },
    {
      "epoch": 0.5253664036076663,
      "grad_norm": 3.640625,
      "learning_rate": 9.719126170307623e-06,
      "loss": 1.4869,
      "step": 466
    },
    {
      "epoch": 0.5264937993235626,
      "grad_norm": 3.359375,
      "learning_rate": 9.71819466131311e-06,
      "loss": 1.6399,
      "step": 467
    },
    {
      "epoch": 0.5276211950394588,
      "grad_norm": 3.828125,
      "learning_rate": 9.717261904761906e-06,
      "loss": 1.6606,
      "step": 468
    },
    {
      "epoch": 0.5287485907553551,
      "grad_norm": 3.40625,
      "learning_rate": 9.71632789814608e-06,
      "loss": 1.5784,
      "step": 469
    },
    {
      "epoch": 0.5298759864712514,
      "grad_norm": 3.703125,
      "learning_rate": 9.715392638950977e-06,
      "loss": 1.5559,
      "step": 470
    },
    {
      "epoch": 0.5310033821871477,
      "grad_norm": 3.640625,
      "learning_rate": 9.714456124655185e-06,
      "loss": 1.8239,
      "step": 471
    },
    {
      "epoch": 0.532130777903044,
      "grad_norm": 3.71875,
      "learning_rate": 9.713518352730529e-06,
      "loss": 1.477,
      "step": 472
    },
    {
      "epoch": 0.5332581736189402,
      "grad_norm": 4.25,
      "learning_rate": 9.712579320642032e-06,
      "loss": 1.6339,
      "step": 473
    },
    {
      "epoch": 0.5343855693348365,
      "grad_norm": 3.9375,
      "learning_rate": 9.7116390258479e-06,
      "loss": 1.6942,
      "step": 474
    },
    {
      "epoch": 0.5355129650507328,
      "grad_norm": 3.796875,
      "learning_rate": 9.710697465799508e-06,
      "loss": 1.4921,
      "step": 475
    },
    {
      "epoch": 0.5366403607666291,
      "grad_norm": 3.859375,
      "learning_rate": 9.709754637941354e-06,
      "loss": 1.5555,
      "step": 476
    },
    {
      "epoch": 0.5377677564825254,
      "grad_norm": 3.609375,
      "learning_rate": 9.708810539711057e-06,
      "loss": 1.6788,
      "step": 477
    },
    {
      "epoch": 0.5388951521984217,
      "grad_norm": 3.921875,
      "learning_rate": 9.707865168539326e-06,
      "loss": 1.6697,
      "step": 478
    },
    {
      "epoch": 0.5400225479143179,
      "grad_norm": 4.03125,
      "learning_rate": 9.706918521849937e-06,
      "loss": 1.7856,
      "step": 479
    },
    {
      "epoch": 0.5411499436302142,
      "grad_norm": 4.71875,
      "learning_rate": 9.705970597059707e-06,
      "loss": 1.8509,
      "step": 480
    },
    {
      "epoch": 0.5422773393461104,
      "grad_norm": 3.40625,
      "learning_rate": 9.705021391578475e-06,
      "loss": 1.4216,
      "step": 481
    },
    {
      "epoch": 0.5434047350620068,
      "grad_norm": 3.71875,
      "learning_rate": 9.704070902809075e-06,
      "loss": 1.6996,
      "step": 482
    },
    {
      "epoch": 0.5445321307779031,
      "grad_norm": 3.953125,
      "learning_rate": 9.703119128147315e-06,
      "loss": 1.76,
      "step": 483
    },
    {
      "epoch": 0.5456595264937993,
      "grad_norm": 4.34375,
      "learning_rate": 9.70216606498195e-06,
      "loss": 1.666,
      "step": 484
    },
    {
      "epoch": 0.5467869222096956,
      "grad_norm": 3.609375,
      "learning_rate": 9.701211710694666e-06,
      "loss": 1.6628,
      "step": 485
    },
    {
      "epoch": 0.5479143179255919,
      "grad_norm": 4.25,
      "learning_rate": 9.700256062660039e-06,
      "loss": 1.818,
      "step": 486
    },
    {
      "epoch": 0.5490417136414881,
      "grad_norm": 3.3125,
      "learning_rate": 9.699299118245535e-06,
      "loss": 1.6174,
      "step": 487
    },
    {
      "epoch": 0.5501691093573844,
      "grad_norm": 3.671875,
      "learning_rate": 9.698340874811463e-06,
      "loss": 1.7608,
      "step": 488
    },
    {
      "epoch": 0.5512965050732808,
      "grad_norm": 3.828125,
      "learning_rate": 9.697381329710966e-06,
      "loss": 1.5091,
      "step": 489
    },
    {
      "epoch": 0.552423900789177,
      "grad_norm": 3.671875,
      "learning_rate": 9.696420480289986e-06,
      "loss": 1.4943,
      "step": 490
    },
    {
      "epoch": 0.5535512965050733,
      "grad_norm": 3.734375,
      "learning_rate": 9.695458323887253e-06,
      "loss": 1.52,
      "step": 491
    },
    {
      "epoch": 0.5546786922209695,
      "grad_norm": 3.734375,
      "learning_rate": 9.69449485783424e-06,
      "loss": 1.5239,
      "step": 492
    },
    {
      "epoch": 0.5558060879368658,
      "grad_norm": 3.953125,
      "learning_rate": 9.693530079455165e-06,
      "loss": 1.7351,
      "step": 493
    },
    {
      "epoch": 0.5569334836527621,
      "grad_norm": 4.03125,
      "learning_rate": 9.692563986066939e-06,
      "loss": 1.7989,
      "step": 494
    },
    {
      "epoch": 0.5580608793686584,
      "grad_norm": 4.15625,
      "learning_rate": 9.691596574979163e-06,
      "loss": 1.421,
      "step": 495
    },
    {
      "epoch": 0.5591882750845547,
      "grad_norm": 3.53125,
      "learning_rate": 9.690627843494087e-06,
      "loss": 1.5376,
      "step": 496
    },
    {
      "epoch": 0.560315670800451,
      "grad_norm": 3.390625,
      "learning_rate": 9.689657788906594e-06,
      "loss": 1.5293,
      "step": 497
    },
    {
      "epoch": 0.5614430665163472,
      "grad_norm": 3.390625,
      "learning_rate": 9.688686408504178e-06,
      "loss": 1.5411,
      "step": 498
    },
    {
      "epoch": 0.5625704622322435,
      "grad_norm": 3.6875,
      "learning_rate": 9.687713699566902e-06,
      "loss": 1.758,
      "step": 499
    },
    {
      "epoch": 0.5636978579481398,
      "grad_norm": 4.3125,
      "learning_rate": 9.686739659367397e-06,
      "loss": 1.8494,
      "step": 500
    },
    {
      "epoch": 0.5648252536640361,
      "grad_norm": 4.1875,
      "learning_rate": 9.685764285170814e-06,
      "loss": 1.5636,
      "step": 501
    },
    {
      "epoch": 0.5659526493799324,
      "grad_norm": 3.671875,
      "learning_rate": 9.684787574234811e-06,
      "loss": 1.7566,
      "step": 502
    },
    {
      "epoch": 0.5670800450958287,
      "grad_norm": 3.71875,
      "learning_rate": 9.683809523809525e-06,
      "loss": 1.6848,
      "step": 503
    },
    {
      "epoch": 0.5682074408117249,
      "grad_norm": 3.296875,
      "learning_rate": 9.682830131137545e-06,
      "loss": 1.3602,
      "step": 504
    },
    {
      "epoch": 0.5693348365276212,
      "grad_norm": 3.390625,
      "learning_rate": 9.681849393453878e-06,
      "loss": 1.4902,
      "step": 505
    },
    {
      "epoch": 0.5704622322435174,
      "grad_norm": 3.984375,
      "learning_rate": 9.680867307985952e-06,
      "loss": 1.622,
      "step": 506
    },
    {
      "epoch": 0.5715896279594137,
      "grad_norm": 3.96875,
      "learning_rate": 9.67988387195355e-06,
      "loss": 1.4971,
      "step": 507
    },
    {
      "epoch": 0.5727170236753101,
      "grad_norm": 3.6875,
      "learning_rate": 9.678899082568808e-06,
      "loss": 1.5791,
      "step": 508
    },
    {
      "epoch": 0.5738444193912063,
      "grad_norm": 4.4375,
      "learning_rate": 9.677912937036189e-06,
      "loss": 1.5487,
      "step": 509
    },
    {
      "epoch": 0.5749718151071026,
      "grad_norm": 3.90625,
      "learning_rate": 9.676925432552444e-06,
      "loss": 1.5266,
      "step": 510
    },
    {
      "epoch": 0.5760992108229989,
      "grad_norm": 4.21875,
      "learning_rate": 9.675936566306596e-06,
      "loss": 1.65,
      "step": 511
    },
    {
      "epoch": 0.5772266065388951,
      "grad_norm": 3.9375,
      "learning_rate": 9.674946335479915e-06,
      "loss": 1.5408,
      "step": 512
    },
    {
      "epoch": 0.5783540022547914,
      "grad_norm": 3.34375,
      "learning_rate": 9.673954737245878e-06,
      "loss": 1.5965,
      "step": 513
    },
    {
      "epoch": 0.5794813979706878,
      "grad_norm": 3.265625,
      "learning_rate": 9.672961768770153e-06,
      "loss": 1.2959,
      "step": 514
    },
    {
      "epoch": 0.580608793686584,
      "grad_norm": 3.890625,
      "learning_rate": 9.671967427210572e-06,
      "loss": 1.6373,
      "step": 515
    },
    {
      "epoch": 0.5817361894024803,
      "grad_norm": 3.59375,
      "learning_rate": 9.670971709717099e-06,
      "loss": 1.5402,
      "step": 516
    },
    {
      "epoch": 0.5828635851183765,
      "grad_norm": 3.515625,
      "learning_rate": 9.669974613431804e-06,
      "loss": 1.4471,
      "step": 517
    },
    {
      "epoch": 0.5839909808342728,
      "grad_norm": 3.53125,
      "learning_rate": 9.668976135488838e-06,
      "loss": 1.7388,
      "step": 518
    },
    {
      "epoch": 0.5851183765501691,
      "grad_norm": 4.03125,
      "learning_rate": 9.667976273014407e-06,
      "loss": 1.5746,
      "step": 519
    },
    {
      "epoch": 0.5862457722660653,
      "grad_norm": 3.28125,
      "learning_rate": 9.666975023126736e-06,
      "loss": 1.5283,
      "step": 520
    },
    {
      "epoch": 0.5873731679819617,
      "grad_norm": 3.78125,
      "learning_rate": 9.66597238293605e-06,
      "loss": 1.44,
      "step": 521
    },
    {
      "epoch": 0.588500563697858,
      "grad_norm": 4.25,
      "learning_rate": 9.664968349544543e-06,
      "loss": 1.6131,
      "step": 522
    },
    {
      "epoch": 0.5896279594137542,
      "grad_norm": 3.8125,
      "learning_rate": 9.66396292004635e-06,
      "loss": 1.5181,
      "step": 523
    },
    {
      "epoch": 0.5907553551296505,
      "grad_norm": 3.796875,
      "learning_rate": 9.662956091527523e-06,
      "loss": 1.643,
      "step": 524
    },
    {
      "epoch": 0.5918827508455468,
      "grad_norm": 4.03125,
      "learning_rate": 9.661947861065986e-06,
      "loss": 1.6147,
      "step": 525
    },
    {
      "epoch": 0.593010146561443,
      "grad_norm": 3.46875,
      "learning_rate": 9.660938225731538e-06,
      "loss": 1.5732,
      "step": 526
    },
    {
      "epoch": 0.5941375422773394,
      "grad_norm": 3.875,
      "learning_rate": 9.659927182585793e-06,
      "loss": 1.6491,
      "step": 527
    },
    {
      "epoch": 0.5952649379932357,
      "grad_norm": 3.65625,
      "learning_rate": 9.65891472868217e-06,
      "loss": 1.6974,
      "step": 528
    },
    {
      "epoch": 0.5963923337091319,
      "grad_norm": 3.59375,
      "learning_rate": 9.65790086106586e-06,
      "loss": 1.6503,
      "step": 529
    },
    {
      "epoch": 0.5975197294250282,
      "grad_norm": 4.09375,
      "learning_rate": 9.656885576773793e-06,
      "loss": 1.4981,
      "step": 530
    },
    {
      "epoch": 0.5986471251409244,
      "grad_norm": 4.0,
      "learning_rate": 9.655868872834615e-06,
      "loss": 1.9282,
      "step": 531
    },
    {
      "epoch": 0.5997745208568207,
      "grad_norm": 3.515625,
      "learning_rate": 9.654850746268657e-06,
      "loss": 1.8477,
      "step": 532
    },
    {
      "epoch": 0.6009019165727171,
      "grad_norm": 3.765625,
      "learning_rate": 9.653831194087905e-06,
      "loss": 1.5327,
      "step": 533
    },
    {
      "epoch": 0.6020293122886133,
      "grad_norm": 3.734375,
      "learning_rate": 9.652810213295968e-06,
      "loss": 1.7506,
      "step": 534
    },
    {
      "epoch": 0.6031567080045096,
      "grad_norm": 3.59375,
      "learning_rate": 9.65178780088806e-06,
      "loss": 1.4972,
      "step": 535
    },
    {
      "epoch": 0.6042841037204059,
      "grad_norm": 3.6875,
      "learning_rate": 9.650763953850952e-06,
      "loss": 1.7595,
      "step": 536
    },
    {
      "epoch": 0.6054114994363021,
      "grad_norm": 3.890625,
      "learning_rate": 9.649738669162963e-06,
      "loss": 1.7174,
      "step": 537
    },
    {
      "epoch": 0.6065388951521984,
      "grad_norm": 3.375,
      "learning_rate": 9.648711943793913e-06,
      "loss": 1.497,
      "step": 538
    },
    {
      "epoch": 0.6076662908680946,
      "grad_norm": 3.765625,
      "learning_rate": 9.647683774705103e-06,
      "loss": 1.7673,
      "step": 539
    },
    {
      "epoch": 0.608793686583991,
      "grad_norm": 3.640625,
      "learning_rate": 9.646654158849282e-06,
      "loss": 1.7369,
      "step": 540
    },
    {
      "epoch": 0.6099210822998873,
      "grad_norm": 3.65625,
      "learning_rate": 9.645623093170618e-06,
      "loss": 1.7056,
      "step": 541
    },
    {
      "epoch": 0.6110484780157835,
      "grad_norm": 3.4375,
      "learning_rate": 9.644590574604668e-06,
      "loss": 1.6402,
      "step": 542
    },
    {
      "epoch": 0.6121758737316798,
      "grad_norm": 2.90625,
      "learning_rate": 9.643556600078342e-06,
      "loss": 1.4158,
      "step": 543
    },
    {
      "epoch": 0.6133032694475761,
      "grad_norm": 4.15625,
      "learning_rate": 9.642521166509878e-06,
      "loss": 1.7481,
      "step": 544
    },
    {
      "epoch": 0.6144306651634723,
      "grad_norm": 3.796875,
      "learning_rate": 9.641484270808819e-06,
      "loss": 1.7715,
      "step": 545
    },
    {
      "epoch": 0.6155580608793687,
      "grad_norm": 3.8125,
      "learning_rate": 9.640445909875961e-06,
      "loss": 1.474,
      "step": 546
    },
    {
      "epoch": 0.616685456595265,
      "grad_norm": 3.8125,
      "learning_rate": 9.639406080603348e-06,
      "loss": 1.5572,
      "step": 547
    },
    {
      "epoch": 0.6178128523111612,
      "grad_norm": 3.765625,
      "learning_rate": 9.638364779874214e-06,
      "loss": 1.7144,
      "step": 548
    },
    {
      "epoch": 0.6189402480270575,
      "grad_norm": 3.671875,
      "learning_rate": 9.637322004562979e-06,
      "loss": 1.7477,
      "step": 549
    },
    {
      "epoch": 0.6200676437429538,
      "grad_norm": 3.875,
      "learning_rate": 9.636277751535192e-06,
      "loss": 1.7134,
      "step": 550
    },
    {
      "epoch": 0.62119503945885,
      "grad_norm": 4.1875,
      "learning_rate": 9.635232017647523e-06,
      "loss": 1.7736,
      "step": 551
    },
    {
      "epoch": 0.6223224351747464,
      "grad_norm": 3.734375,
      "learning_rate": 9.634184799747714e-06,
      "loss": 1.6046,
      "step": 552
    },
    {
      "epoch": 0.6234498308906427,
      "grad_norm": 3.3125,
      "learning_rate": 9.633136094674556e-06,
      "loss": 1.4566,
      "step": 553
    },
    {
      "epoch": 0.6245772266065389,
      "grad_norm": 4.75,
      "learning_rate": 9.632085899257857e-06,
      "loss": 1.4509,
      "step": 554
    },
    {
      "epoch": 0.6257046223224352,
      "grad_norm": 3.453125,
      "learning_rate": 9.631034210318402e-06,
      "loss": 1.5229,
      "step": 555
    },
    {
      "epoch": 0.6268320180383314,
      "grad_norm": 3.53125,
      "learning_rate": 9.629981024667933e-06,
      "loss": 1.4461,
      "step": 556
    },
    {
      "epoch": 0.6279594137542277,
      "grad_norm": 3.875,
      "learning_rate": 9.628926339109108e-06,
      "loss": 1.6644,
      "step": 557
    },
    {
      "epoch": 0.629086809470124,
      "grad_norm": 3.828125,
      "learning_rate": 9.627870150435472e-06,
      "loss": 1.7888,
      "step": 558
    },
    {
      "epoch": 0.6302142051860203,
      "grad_norm": 3.5625,
      "learning_rate": 9.626812455431425e-06,
      "loss": 1.4916,
      "step": 559
    },
    {
      "epoch": 0.6313416009019166,
      "grad_norm": 3.5,
      "learning_rate": 9.625753250872185e-06,
      "loss": 1.5074,
      "step": 560
    },
    {
      "epoch": 0.6324689966178129,
      "grad_norm": 3.484375,
      "learning_rate": 9.624692533523766e-06,
      "loss": 1.8062,
      "step": 561
    },
    {
      "epoch": 0.6335963923337091,
      "grad_norm": 3.5625,
      "learning_rate": 9.623630300142927e-06,
      "loss": 1.5502,
      "step": 562
    },
    {
      "epoch": 0.6347237880496054,
      "grad_norm": 2.859375,
      "learning_rate": 9.622566547477156e-06,
      "loss": 1.6944,
      "step": 563
    },
    {
      "epoch": 0.6358511837655016,
      "grad_norm": 3.65625,
      "learning_rate": 9.621501272264631e-06,
      "loss": 1.59,
      "step": 564
    },
    {
      "epoch": 0.636978579481398,
      "grad_norm": 3.75,
      "learning_rate": 9.620434471234185e-06,
      "loss": 1.623,
      "step": 565
    },
    {
      "epoch": 0.6381059751972943,
      "grad_norm": 3.46875,
      "learning_rate": 9.619366141105272e-06,
      "loss": 1.6178,
      "step": 566
    },
    {
      "epoch": 0.6392333709131905,
      "grad_norm": 3.53125,
      "learning_rate": 9.618296278587935e-06,
      "loss": 1.7501,
      "step": 567
    },
    {
      "epoch": 0.6403607666290868,
      "grad_norm": 3.953125,
      "learning_rate": 9.617224880382776e-06,
      "loss": 1.5141,
      "step": 568
    },
    {
      "epoch": 0.6414881623449831,
      "grad_norm": 3.4375,
      "learning_rate": 9.616151943180911e-06,
      "loss": 1.7371,
      "step": 569
    },
    {
      "epoch": 0.6426155580608793,
      "grad_norm": 3.421875,
      "learning_rate": 9.615077463663952e-06,
      "loss": 1.6816,
      "step": 570
    },
    {
      "epoch": 0.6437429537767756,
      "grad_norm": 3.28125,
      "learning_rate": 9.614001438503957e-06,
      "loss": 1.7284,
      "step": 571
    },
    {
      "epoch": 0.644870349492672,
      "grad_norm": 2.90625,
      "learning_rate": 9.612923864363404e-06,
      "loss": 1.3685,
      "step": 572
    },
    {
      "epoch": 0.6459977452085682,
      "grad_norm": 3.3125,
      "learning_rate": 9.611844737895158e-06,
      "loss": 1.5401,
      "step": 573
    },
    {
      "epoch": 0.6471251409244645,
      "grad_norm": 3.46875,
      "learning_rate": 9.610764055742431e-06,
      "loss": 1.4004,
      "step": 574
    },
    {
      "epoch": 0.6482525366403608,
      "grad_norm": 3.765625,
      "learning_rate": 9.609681814538753e-06,
      "loss": 1.5518,
      "step": 575
    },
    {
      "epoch": 0.649379932356257,
      "grad_norm": 3.296875,
      "learning_rate": 9.608598010907925e-06,
      "loss": 1.5126,
      "step": 576
    },
    {
      "epoch": 0.6505073280721533,
      "grad_norm": 3.84375,
      "learning_rate": 9.607512641464002e-06,
      "loss": 1.682,
      "step": 577
    },
    {
      "epoch": 0.6516347237880497,
      "grad_norm": 4.03125,
      "learning_rate": 9.606425702811246e-06,
      "loss": 1.6966,
      "step": 578
    },
    {
      "epoch": 0.6527621195039459,
      "grad_norm": 3.53125,
      "learning_rate": 9.60533719154409e-06,
      "loss": 1.7269,
      "step": 579
    },
    {
      "epoch": 0.6538895152198422,
      "grad_norm": 3.40625,
      "learning_rate": 9.604247104247105e-06,
      "loss": 1.5854,
      "step": 580
    },
    {
      "epoch": 0.6550169109357384,
      "grad_norm": 3.40625,
      "learning_rate": 9.60315543749497e-06,
      "loss": 1.4829,
      "step": 581
    },
    {
      "epoch": 0.6561443066516347,
      "grad_norm": 3.1875,
      "learning_rate": 9.602062187852426e-06,
      "loss": 1.4663,
      "step": 582
    },
    {
      "epoch": 0.657271702367531,
      "grad_norm": 3.359375,
      "learning_rate": 9.600967351874244e-06,
      "loss": 1.7147,
      "step": 583
    },
    {
      "epoch": 0.6583990980834273,
      "grad_norm": 3.1875,
      "learning_rate": 9.599870926105196e-06,
      "loss": 1.5217,
      "step": 584
    },
    {
      "epoch": 0.6595264937993236,
      "grad_norm": 3.796875,
      "learning_rate": 9.598772907080004e-06,
      "loss": 1.5732,
      "step": 585
    },
    {
      "epoch": 0.6606538895152199,
      "grad_norm": 3.390625,
      "learning_rate": 9.597673291323317e-06,
      "loss": 1.5924,
      "step": 586
    },
    {
      "epoch": 0.6617812852311161,
      "grad_norm": 4.25,
      "learning_rate": 9.596572075349665e-06,
      "loss": 1.6177,
      "step": 587
    },
    {
      "epoch": 0.6629086809470124,
      "grad_norm": 3.609375,
      "learning_rate": 9.59546925566343e-06,
      "loss": 1.7268,
      "step": 588
    },
    {
      "epoch": 0.6640360766629086,
      "grad_norm": 3.84375,
      "learning_rate": 9.594364828758805e-06,
      "loss": 1.6693,
      "step": 589
    },
    {
      "epoch": 0.6651634723788049,
      "grad_norm": 3.359375,
      "learning_rate": 9.593258791119754e-06,
      "loss": 1.5739,
      "step": 590
    },
    {
      "epoch": 0.6662908680947013,
      "grad_norm": 3.59375,
      "learning_rate": 9.592151139219979e-06,
      "loss": 1.61,
      "step": 591
    },
    {
      "epoch": 0.6674182638105975,
      "grad_norm": 3.90625,
      "learning_rate": 9.591041869522884e-06,
      "loss": 1.5201,
      "step": 592
    },
    {
      "epoch": 0.6685456595264938,
      "grad_norm": 4.25,
      "learning_rate": 9.589930978481527e-06,
      "loss": 1.5204,
      "step": 593
    },
    {
      "epoch": 0.6696730552423901,
      "grad_norm": 3.203125,
      "learning_rate": 9.5888184625386e-06,
      "loss": 1.6602,
      "step": 594
    },
    {
      "epoch": 0.6708004509582863,
      "grad_norm": 3.53125,
      "learning_rate": 9.587704318126373e-06,
      "loss": 1.5429,
      "step": 595
    },
    {
      "epoch": 0.6719278466741826,
      "grad_norm": 3.65625,
      "learning_rate": 9.586588541666668e-06,
      "loss": 1.6456,
      "step": 596
    },
    {
      "epoch": 0.673055242390079,
      "grad_norm": 3.640625,
      "learning_rate": 9.585471129570814e-06,
      "loss": 1.5501,
      "step": 597
    },
    {
      "epoch": 0.6741826381059752,
      "grad_norm": 3.703125,
      "learning_rate": 9.58435207823961e-06,
      "loss": 1.5508,
      "step": 598
    },
    {
      "epoch": 0.6753100338218715,
      "grad_norm": 4.1875,
      "learning_rate": 9.583231384063291e-06,
      "loss": 1.4796,
      "step": 599
    },
    {
      "epoch": 0.6764374295377678,
      "grad_norm": 3.859375,
      "learning_rate": 9.582109043421484e-06,
      "loss": 1.7035,
      "step": 600
    },
    {
      "epoch": 0.677564825253664,
      "grad_norm": 3.921875,
      "learning_rate": 9.580985052683167e-06,
      "loss": 1.8507,
      "step": 601
    },
    {
      "epoch": 0.6786922209695603,
      "grad_norm": 3.078125,
      "learning_rate": 9.579859408206638e-06,
      "loss": 1.4633,
      "step": 602
    },
    {
      "epoch": 0.6798196166854565,
      "grad_norm": 3.25,
      "learning_rate": 9.578732106339468e-06,
      "loss": 1.7404,
      "step": 603
    },
    {
      "epoch": 0.6809470124013529,
      "grad_norm": 3.390625,
      "learning_rate": 9.577603143418468e-06,
      "loss": 1.5974,
      "step": 604
    },
    {
      "epoch": 0.6820744081172492,
      "grad_norm": 3.5,
      "learning_rate": 9.57647251576964e-06,
      "loss": 1.8548,
      "step": 605
    },
    {
      "epoch": 0.6832018038331454,
      "grad_norm": 4.65625,
      "learning_rate": 9.575340219708148e-06,
      "loss": 1.628,
      "step": 606
    },
    {
      "epoch": 0.6843291995490417,
      "grad_norm": 3.34375,
      "learning_rate": 9.574206251538272e-06,
      "loss": 1.8579,
      "step": 607
    },
    {
      "epoch": 0.685456595264938,
      "grad_norm": 5.40625,
      "learning_rate": 9.573070607553366e-06,
      "loss": 1.5727,
      "step": 608
    },
    {
      "epoch": 0.6865839909808342,
      "grad_norm": 3.390625,
      "learning_rate": 9.571933284035823e-06,
      "loss": 1.6927,
      "step": 609
    },
    {
      "epoch": 0.6877113866967306,
      "grad_norm": 2.90625,
      "learning_rate": 9.57079427725703e-06,
      "loss": 1.5644,
      "step": 610
    },
    {
      "epoch": 0.6888387824126269,
      "grad_norm": 3.8125,
      "learning_rate": 9.569653583477332e-06,
      "loss": 1.7526,
      "step": 611
    },
    {
      "epoch": 0.6899661781285231,
      "grad_norm": 3.671875,
      "learning_rate": 9.568511198945983e-06,
      "loss": 1.6372,
      "step": 612
    },
    {
      "epoch": 0.6910935738444194,
      "grad_norm": 3.921875,
      "learning_rate": 9.567367119901113e-06,
      "loss": 1.5723,
      "step": 613
    },
    {
      "epoch": 0.6922209695603156,
      "grad_norm": 3.375,
      "learning_rate": 9.566221342569686e-06,
      "loss": 1.7095,
      "step": 614
    },
    {
      "epoch": 0.6933483652762119,
      "grad_norm": 3.578125,
      "learning_rate": 9.565073863167451e-06,
      "loss": 1.9733,
      "step": 615
    },
    {
      "epoch": 0.6944757609921083,
      "grad_norm": 2.953125,
      "learning_rate": 9.563924677898912e-06,
      "loss": 1.5899,
      "step": 616
    },
    {
      "epoch": 0.6956031567080045,
      "grad_norm": 3.671875,
      "learning_rate": 9.56277378295727e-06,
      "loss": 1.6525,
      "step": 617
    },
    {
      "epoch": 0.6967305524239008,
      "grad_norm": 3.703125,
      "learning_rate": 9.5616211745244e-06,
      "loss": 1.6769,
      "step": 618
    },
    {
      "epoch": 0.6978579481397971,
      "grad_norm": 3.234375,
      "learning_rate": 9.5604668487708e-06,
      "loss": 1.7715,
      "step": 619
    },
    {
      "epoch": 0.6989853438556933,
      "grad_norm": 3.546875,
      "learning_rate": 9.559310801855534e-06,
      "loss": 1.6866,
      "step": 620
    },
    {
      "epoch": 0.7001127395715896,
      "grad_norm": 3.59375,
      "learning_rate": 9.55815302992622e-06,
      "loss": 1.7819,
      "step": 621
    },
    {
      "epoch": 0.7012401352874859,
      "grad_norm": 3.125,
      "learning_rate": 9.556993529118966e-06,
      "loss": 1.4556,
      "step": 622
    },
    {
      "epoch": 0.7023675310033822,
      "grad_norm": 3.078125,
      "learning_rate": 9.555832295558323e-06,
      "loss": 1.7588,
      "step": 623
    },
    {
      "epoch": 0.7034949267192785,
      "grad_norm": 3.671875,
      "learning_rate": 9.554669325357261e-06,
      "loss": 1.5501,
      "step": 624
    },
    {
      "epoch": 0.7046223224351748,
      "grad_norm": 3.546875,
      "learning_rate": 9.553504614617112e-06,
      "loss": 1.7801,
      "step": 625
    },
    {
      "epoch": 0.705749718151071,
      "grad_norm": 3.328125,
      "learning_rate": 9.552338159427526e-06,
      "loss": 1.4868,
      "step": 626
    },
    {
      "epoch": 0.7068771138669673,
      "grad_norm": 2.984375,
      "learning_rate": 9.551169955866435e-06,
      "loss": 1.4296,
      "step": 627
    },
    {
      "epoch": 0.7080045095828635,
      "grad_norm": 2.8125,
      "learning_rate": 9.55e-06,
      "loss": 1.2028,
      "step": 628
    },
    {
      "epoch": 0.7091319052987599,
      "grad_norm": 2.671875,
      "learning_rate": 9.54882828788258e-06,
      "loss": 1.7153,
      "step": 629
    },
    {
      "epoch": 0.7102593010146562,
      "grad_norm": 3.546875,
      "learning_rate": 9.54765481555667e-06,
      "loss": 1.6981,
      "step": 630
    },
    {
      "epoch": 0.7113866967305524,
      "grad_norm": 4.0,
      "learning_rate": 9.54647957905287e-06,
      "loss": 1.6865,
      "step": 631
    },
    {
      "epoch": 0.7125140924464487,
      "grad_norm": 3.78125,
      "learning_rate": 9.545302574389837e-06,
      "loss": 1.7228,
      "step": 632
    },
    {
      "epoch": 0.713641488162345,
      "grad_norm": 3.984375,
      "learning_rate": 9.544123797574238e-06,
      "loss": 1.7625,
      "step": 633
    },
    {
      "epoch": 0.7147688838782412,
      "grad_norm": 2.90625,
      "learning_rate": 9.542943244600703e-06,
      "loss": 1.3438,
      "step": 634
    },
    {
      "epoch": 0.7158962795941376,
      "grad_norm": 2.890625,
      "learning_rate": 9.54176091145179e-06,
      "loss": 1.4803,
      "step": 635
    },
    {
      "epoch": 0.7170236753100339,
      "grad_norm": 3.265625,
      "learning_rate": 9.540576794097922e-06,
      "loss": 1.7673,
      "step": 636
    },
    {
      "epoch": 0.7181510710259301,
      "grad_norm": 3.421875,
      "learning_rate": 9.539390888497358e-06,
      "loss": 1.521,
      "step": 637
    },
    {
      "epoch": 0.7192784667418264,
      "grad_norm": 3.875,
      "learning_rate": 9.53820319059614e-06,
      "loss": 1.7496,
      "step": 638
    },
    {
      "epoch": 0.7204058624577226,
      "grad_norm": 3.71875,
      "learning_rate": 9.537013696328042e-06,
      "loss": 1.5867,
      "step": 639
    },
    {
      "epoch": 0.7215332581736189,
      "grad_norm": 3.171875,
      "learning_rate": 9.535822401614532e-06,
      "loss": 1.6582,
      "step": 640
    },
    {
      "epoch": 0.7226606538895152,
      "grad_norm": 4.0,
      "learning_rate": 9.534629302364722e-06,
      "loss": 1.6313,
      "step": 641
    },
    {
      "epoch": 0.7237880496054115,
      "grad_norm": 3.484375,
      "learning_rate": 9.533434394475325e-06,
      "loss": 1.6702,
      "step": 642
    },
    {
      "epoch": 0.7249154453213078,
      "grad_norm": 3.21875,
      "learning_rate": 9.532237673830594e-06,
      "loss": 1.6134,
      "step": 643
    },
    {
      "epoch": 0.7260428410372041,
      "grad_norm": 3.140625,
      "learning_rate": 9.531039136302295e-06,
      "loss": 1.6143,
      "step": 644
    },
    {
      "epoch": 0.7271702367531003,
      "grad_norm": 3.859375,
      "learning_rate": 9.529838777749642e-06,
      "loss": 1.7458,
      "step": 645
    },
    {
      "epoch": 0.7282976324689966,
      "grad_norm": 4.09375,
      "learning_rate": 9.52863659401926e-06,
      "loss": 1.7044,
      "step": 646
    },
    {
      "epoch": 0.7294250281848929,
      "grad_norm": 3.40625,
      "learning_rate": 9.527432580945135e-06,
      "loss": 1.5526,
      "step": 647
    },
    {
      "epoch": 0.7305524239007892,
      "grad_norm": 3.34375,
      "learning_rate": 9.526226734348563e-06,
      "loss": 1.5698,
      "step": 648
    },
    {
      "epoch": 0.7316798196166855,
      "grad_norm": 3.359375,
      "learning_rate": 9.525019050038102e-06,
      "loss": 1.4824,
      "step": 649
    },
    {
      "epoch": 0.7328072153325818,
      "grad_norm": 3.796875,
      "learning_rate": 9.523809523809525e-06,
      "loss": 1.7792,
      "step": 650
    },
    {
      "epoch": 0.733934611048478,
      "grad_norm": 3.140625,
      "learning_rate": 9.522598151445775e-06,
      "loss": 1.6709,
      "step": 651
    },
    {
      "epoch": 0.7350620067643743,
      "grad_norm": 3.796875,
      "learning_rate": 9.521384928716906e-06,
      "loss": 1.6095,
      "step": 652
    },
    {
      "epoch": 0.7361894024802705,
      "grad_norm": 3.59375,
      "learning_rate": 9.520169851380043e-06,
      "loss": 1.731,
      "step": 653
    },
    {
      "epoch": 0.7373167981961668,
      "grad_norm": 3.75,
      "learning_rate": 9.518952915179332e-06,
      "loss": 1.6662,
      "step": 654
    },
    {
      "epoch": 0.7384441939120632,
      "grad_norm": 3.46875,
      "learning_rate": 9.51773411584588e-06,
      "loss": 1.5384,
      "step": 655
    },
    {
      "epoch": 0.7395715896279594,
      "grad_norm": 4.0,
      "learning_rate": 9.516513449097719e-06,
      "loss": 1.493,
      "step": 656
    },
    {
      "epoch": 0.7406989853438557,
      "grad_norm": 3.46875,
      "learning_rate": 9.51529091063975e-06,
      "loss": 1.6095,
      "step": 657
    },
    {
      "epoch": 0.741826381059752,
      "grad_norm": 3.1875,
      "learning_rate": 9.514066496163684e-06,
      "loss": 1.3806,
      "step": 658
    },
    {
      "epoch": 0.7429537767756482,
      "grad_norm": 3.53125,
      "learning_rate": 9.51284020134801e-06,
      "loss": 1.8906,
      "step": 659
    },
    {
      "epoch": 0.7440811724915445,
      "grad_norm": 3.640625,
      "learning_rate": 9.511612021857924e-06,
      "loss": 1.7183,
      "step": 660
    },
    {
      "epoch": 0.7452085682074409,
      "grad_norm": 3.328125,
      "learning_rate": 9.510381953345297e-06,
      "loss": 1.5711,
      "step": 661
    },
    {
      "epoch": 0.7463359639233371,
      "grad_norm": 3.40625,
      "learning_rate": 9.509149991448606e-06,
      "loss": 1.5501,
      "step": 662
    },
    {
      "epoch": 0.7474633596392334,
      "grad_norm": 3.3125,
      "learning_rate": 9.507916131792897e-06,
      "loss": 1.5851,
      "step": 663
    },
    {
      "epoch": 0.7485907553551296,
      "grad_norm": 3.3125,
      "learning_rate": 9.506680369989723e-06,
      "loss": 1.6575,
      "step": 664
    },
    {
      "epoch": 0.7497181510710259,
      "grad_norm": 3.53125,
      "learning_rate": 9.505442701637096e-06,
      "loss": 1.3892,
      "step": 665
    },
    {
      "epoch": 0.7508455467869222,
      "grad_norm": 10.1875,
      "learning_rate": 9.504203122319438e-06,
      "loss": 1.7219,
      "step": 666
    },
    {
      "epoch": 0.7519729425028185,
      "grad_norm": 4.0,
      "learning_rate": 9.502961627607521e-06,
      "loss": 1.8462,
      "step": 667
    },
    {
      "epoch": 0.7531003382187148,
      "grad_norm": 4.78125,
      "learning_rate": 9.501718213058419e-06,
      "loss": 1.6425,
      "step": 668
    },
    {
      "epoch": 0.7542277339346111,
      "grad_norm": 3.453125,
      "learning_rate": 9.50047287421546e-06,
      "loss": 1.6238,
      "step": 669
    },
    {
      "epoch": 0.7553551296505073,
      "grad_norm": 3.25,
      "learning_rate": 9.499225606608158e-06,
      "loss": 1.6365,
      "step": 670
    },
    {
      "epoch": 0.7564825253664036,
      "grad_norm": 3.46875,
      "learning_rate": 9.497976405752176e-06,
      "loss": 1.6166,
      "step": 671
    },
    {
      "epoch": 0.7576099210822999,
      "grad_norm": 3.4375,
      "learning_rate": 9.49672526714926e-06,
      "loss": 1.7892,
      "step": 672
    },
    {
      "epoch": 0.7587373167981961,
      "grad_norm": 3.984375,
      "learning_rate": 9.495472186287193e-06,
      "loss": 1.4993,
      "step": 673
    },
    {
      "epoch": 0.7598647125140925,
      "grad_norm": 3.859375,
      "learning_rate": 9.49421715863974e-06,
      "loss": 1.7426,
      "step": 674
    },
    {
      "epoch": 0.7609921082299888,
      "grad_norm": 3.578125,
      "learning_rate": 9.492960179666582e-06,
      "loss": 1.6136,
      "step": 675
    },
    {
      "epoch": 0.762119503945885,
      "grad_norm": 3.953125,
      "learning_rate": 9.49170124481328e-06,
      "loss": 1.6111,
      "step": 676
    },
    {
      "epoch": 0.7632468996617813,
      "grad_norm": 3.1875,
      "learning_rate": 9.490440349511206e-06,
      "loss": 1.6796,
      "step": 677
    },
    {
      "epoch": 0.7643742953776775,
      "grad_norm": 3.15625,
      "learning_rate": 9.489177489177491e-06,
      "loss": 1.5719,
      "step": 678
    },
    {
      "epoch": 0.7655016910935738,
      "grad_norm": 2.78125,
      "learning_rate": 9.487912659214972e-06,
      "loss": 1.2953,
      "step": 679
    },
    {
      "epoch": 0.7666290868094702,
      "grad_norm": 3.5,
      "learning_rate": 9.48664585501214e-06,
      "loss": 1.5232,
      "step": 680
    },
    {
      "epoch": 0.7677564825253664,
      "grad_norm": 3.515625,
      "learning_rate": 9.48537707194307e-06,
      "loss": 1.6088,
      "step": 681
    },
    {
      "epoch": 0.7688838782412627,
      "grad_norm": 3.5625,
      "learning_rate": 9.48410630536738e-06,
      "loss": 1.6768,
      "step": 682
    },
    {
      "epoch": 0.770011273957159,
      "grad_norm": 3.1875,
      "learning_rate": 9.482833550630161e-06,
      "loss": 1.5304,
      "step": 683
    },
    {
      "epoch": 0.7711386696730552,
      "grad_norm": 3.15625,
      "learning_rate": 9.481558803061935e-06,
      "loss": 1.5584,
      "step": 684
    },
    {
      "epoch": 0.7722660653889515,
      "grad_norm": 2.875,
      "learning_rate": 9.480282057978584e-06,
      "loss": 1.6804,
      "step": 685
    },
    {
      "epoch": 0.7733934611048479,
      "grad_norm": 2.84375,
      "learning_rate": 9.479003310681304e-06,
      "loss": 1.4617,
      "step": 686
    },
    {
      "epoch": 0.7745208568207441,
      "grad_norm": 3.140625,
      "learning_rate": 9.477722556456536e-06,
      "loss": 1.5673,
      "step": 687
    },
    {
      "epoch": 0.7756482525366404,
      "grad_norm": 3.484375,
      "learning_rate": 9.476439790575917e-06,
      "loss": 1.6792,
      "step": 688
    },
    {
      "epoch": 0.7767756482525366,
      "grad_norm": 3.390625,
      "learning_rate": 9.475155008296219e-06,
      "loss": 1.6166,
      "step": 689
    },
    {
      "epoch": 0.7779030439684329,
      "grad_norm": 2.890625,
      "learning_rate": 9.47386820485929e-06,
      "loss": 1.2754,
      "step": 690
    },
    {
      "epoch": 0.7790304396843292,
      "grad_norm": 3.46875,
      "learning_rate": 9.472579375491998e-06,
      "loss": 1.7385,
      "step": 691
    },
    {
      "epoch": 0.7801578354002254,
      "grad_norm": 3.5625,
      "learning_rate": 9.471288515406164e-06,
      "loss": 1.5246,
      "step": 692
    },
    {
      "epoch": 0.7812852311161218,
      "grad_norm": 2.875,
      "learning_rate": 9.469995619798513e-06,
      "loss": 1.4468,
      "step": 693
    },
    {
      "epoch": 0.7824126268320181,
      "grad_norm": 3.5,
      "learning_rate": 9.468700683850606e-06,
      "loss": 1.4542,
      "step": 694
    },
    {
      "epoch": 0.7835400225479143,
      "grad_norm": 4.0625,
      "learning_rate": 9.46740370272879e-06,
      "loss": 1.7879,
      "step": 695
    },
    {
      "epoch": 0.7846674182638106,
      "grad_norm": 3.59375,
      "learning_rate": 9.466104671584126e-06,
      "loss": 1.6231,
      "step": 696
    },
    {
      "epoch": 0.7857948139797069,
      "grad_norm": 4.4375,
      "learning_rate": 9.464803585552334e-06,
      "loss": 1.5563,
      "step": 697
    },
    {
      "epoch": 0.7869222096956031,
      "grad_norm": 3.203125,
      "learning_rate": 9.46350043975374e-06,
      "loss": 1.6716,
      "step": 698
    },
    {
      "epoch": 0.7880496054114995,
      "grad_norm": 3.4375,
      "learning_rate": 9.462195229293196e-06,
      "loss": 1.6035,
      "step": 699
    },
    {
      "epoch": 0.7891770011273957,
      "grad_norm": 3.078125,
      "learning_rate": 9.460887949260043e-06,
      "loss": 1.6049,
      "step": 700
    },
    {
      "epoch": 0.790304396843292,
      "grad_norm": 3.484375,
      "learning_rate": 9.459578594728026e-06,
      "loss": 1.6692,
      "step": 701
    },
    {
      "epoch": 0.7914317925591883,
      "grad_norm": 3.515625,
      "learning_rate": 9.45826716075525e-06,
      "loss": 1.7533,
      "step": 702
    },
    {
      "epoch": 0.7925591882750845,
      "grad_norm": 4.40625,
      "learning_rate": 9.456953642384106e-06,
      "loss": 1.5232,
      "step": 703
    },
    {
      "epoch": 0.7936865839909808,
      "grad_norm": 3.28125,
      "learning_rate": 9.455638034641217e-06,
      "loss": 1.6223,
      "step": 704
    },
    {
      "epoch": 0.7948139797068771,
      "grad_norm": 3.71875,
      "learning_rate": 9.454320332537366e-06,
      "loss": 1.5986,
      "step": 705
    },
    {
      "epoch": 0.7959413754227734,
      "grad_norm": 3.46875,
      "learning_rate": 9.453000531067446e-06,
      "loss": 1.7529,
      "step": 706
    },
    {
      "epoch": 0.7970687711386697,
      "grad_norm": 3.515625,
      "learning_rate": 9.451678625210383e-06,
      "loss": 1.5458,
      "step": 707
    },
    {
      "epoch": 0.798196166854566,
      "grad_norm": 3.671875,
      "learning_rate": 9.450354609929078e-06,
      "loss": 1.6905,
      "step": 708
    },
    {
      "epoch": 0.7993235625704622,
      "grad_norm": 3.109375,
      "learning_rate": 9.449028480170349e-06,
      "loss": 1.3862,
      "step": 709
    },
    {
      "epoch": 0.8004509582863585,
      "grad_norm": 3.390625,
      "learning_rate": 9.447700230864857e-06,
      "loss": 1.7146,
      "step": 710
    },
    {
      "epoch": 0.8015783540022547,
      "grad_norm": 4.375,
      "learning_rate": 9.446369856927043e-06,
      "loss": 1.805,
      "step": 711
    },
    {
      "epoch": 0.8027057497181511,
      "grad_norm": 4.84375,
      "learning_rate": 9.445037353255072e-06,
      "loss": 1.5799,
      "step": 712
    },
    {
      "epoch": 0.8038331454340474,
      "grad_norm": 4.15625,
      "learning_rate": 9.443702714730754e-06,
      "loss": 1.7286,
      "step": 713
    },
    {
      "epoch": 0.8049605411499436,
      "grad_norm": 2.796875,
      "learning_rate": 9.442365936219491e-06,
      "loss": 1.6373,
      "step": 714
    },
    {
      "epoch": 0.8060879368658399,
      "grad_norm": 3.734375,
      "learning_rate": 9.441027012570208e-06,
      "loss": 1.6383,
      "step": 715
    },
    {
      "epoch": 0.8072153325817362,
      "grad_norm": 3.765625,
      "learning_rate": 9.439685938615275e-06,
      "loss": 1.7575,
      "step": 716
    },
    {
      "epoch": 0.8083427282976324,
      "grad_norm": 2.96875,
      "learning_rate": 9.438342709170463e-06,
      "loss": 1.5325,
      "step": 717
    },
    {
      "epoch": 0.8094701240135288,
      "grad_norm": 3.46875,
      "learning_rate": 9.436997319034853e-06,
      "loss": 1.6242,
      "step": 718
    },
    {
      "epoch": 0.8105975197294251,
      "grad_norm": 3.625,
      "learning_rate": 9.435649762990788e-06,
      "loss": 1.5506,
      "step": 719
    },
    {
      "epoch": 0.8117249154453213,
      "grad_norm": 3.15625,
      "learning_rate": 9.434300035803795e-06,
      "loss": 1.6569,
      "step": 720
    },
    {
      "epoch": 0.8128523111612176,
      "grad_norm": 3.640625,
      "learning_rate": 9.432948132222521e-06,
      "loss": 1.5688,
      "step": 721
    },
    {
      "epoch": 0.8139797068771139,
      "grad_norm": 3.96875,
      "learning_rate": 9.431594046978663e-06,
      "loss": 1.9711,
      "step": 722
    },
    {
      "epoch": 0.8151071025930101,
      "grad_norm": 4.3125,
      "learning_rate": 9.4302377747869e-06,
      "loss": 1.8151,
      "step": 723
    },
    {
      "epoch": 0.8162344983089064,
      "grad_norm": 3.53125,
      "learning_rate": 9.428879310344828e-06,
      "loss": 1.7613,
      "step": 724
    },
    {
      "epoch": 0.8173618940248027,
      "grad_norm": 4.25,
      "learning_rate": 9.427518648332885e-06,
      "loss": 1.7276,
      "step": 725
    },
    {
      "epoch": 0.818489289740699,
      "grad_norm": 3.015625,
      "learning_rate": 9.426155783414283e-06,
      "loss": 1.7445,
      "step": 726
    },
    {
      "epoch": 0.8196166854565953,
      "grad_norm": 3.234375,
      "learning_rate": 9.424790710234945e-06,
      "loss": 1.7383,
      "step": 727
    },
    {
      "epoch": 0.8207440811724915,
      "grad_norm": 3.328125,
      "learning_rate": 9.423423423423425e-06,
      "loss": 1.65,
      "step": 728
    },
    {
      "epoch": 0.8218714768883878,
      "grad_norm": 3.15625,
      "learning_rate": 9.42205391759084e-06,
      "loss": 1.7405,
      "step": 729
    },
    {
      "epoch": 0.8229988726042841,
      "grad_norm": 3.15625,
      "learning_rate": 9.420682187330808e-06,
      "loss": 1.5457,
      "step": 730
    },
    {
      "epoch": 0.8241262683201804,
      "grad_norm": 2.921875,
      "learning_rate": 9.419308227219363e-06,
      "loss": 1.572,
      "step": 731
    },
    {
      "epoch": 0.8252536640360767,
      "grad_norm": 3.25,
      "learning_rate": 9.417932031814896e-06,
      "loss": 1.8325,
      "step": 732
    },
    {
      "epoch": 0.826381059751973,
      "grad_norm": 3.65625,
      "learning_rate": 9.416553595658075e-06,
      "loss": 1.6288,
      "step": 733
    },
    {
      "epoch": 0.8275084554678692,
      "grad_norm": 3.453125,
      "learning_rate": 9.415172913271773e-06,
      "loss": 1.8794,
      "step": 734
    },
    {
      "epoch": 0.8286358511837655,
      "grad_norm": 3.375,
      "learning_rate": 9.413789979161006e-06,
      "loss": 1.5804,
      "step": 735
    },
    {
      "epoch": 0.8297632468996617,
      "grad_norm": 3.359375,
      "learning_rate": 9.412404787812843e-06,
      "loss": 1.6024,
      "step": 736
    },
    {
      "epoch": 0.830890642615558,
      "grad_norm": 3.578125,
      "learning_rate": 9.411017333696343e-06,
      "loss": 1.8214,
      "step": 737
    },
    {
      "epoch": 0.8320180383314544,
      "grad_norm": 3.390625,
      "learning_rate": 9.409627611262488e-06,
      "loss": 1.7035,
      "step": 738
    },
    {
      "epoch": 0.8331454340473506,
      "grad_norm": 3.640625,
      "learning_rate": 9.408235614944095e-06,
      "loss": 1.7314,
      "step": 739
    },
    {
      "epoch": 0.8342728297632469,
      "grad_norm": 3.21875,
      "learning_rate": 9.40684133915575e-06,
      "loss": 1.5981,
      "step": 740
    },
    {
      "epoch": 0.8354002254791432,
      "grad_norm": 3.609375,
      "learning_rate": 9.405444778293727e-06,
      "loss": 1.7295,
      "step": 741
    },
    {
      "epoch": 0.8365276211950394,
      "grad_norm": 3.375,
      "learning_rate": 9.404045926735922e-06,
      "loss": 1.6989,
      "step": 742
    },
    {
      "epoch": 0.8376550169109357,
      "grad_norm": 3.765625,
      "learning_rate": 9.40264477884177e-06,
      "loss": 1.7439,
      "step": 743
    },
    {
      "epoch": 0.8387824126268321,
      "grad_norm": 2.828125,
      "learning_rate": 9.401241328952173e-06,
      "loss": 1.4783,
      "step": 744
    },
    {
      "epoch": 0.8399098083427283,
      "grad_norm": 3.375,
      "learning_rate": 9.399835571389422e-06,
      "loss": 1.7843,
      "step": 745
    },
    {
      "epoch": 0.8410372040586246,
      "grad_norm": 3.265625,
      "learning_rate": 9.398427500457122e-06,
      "loss": 1.5326,
      "step": 746
    },
    {
      "epoch": 0.8421645997745209,
      "grad_norm": 3.046875,
      "learning_rate": 9.397017110440115e-06,
      "loss": 1.5508,
      "step": 747
    },
    {
      "epoch": 0.8432919954904171,
      "grad_norm": 3.359375,
      "learning_rate": 9.395604395604396e-06,
      "loss": 1.6956,
      "step": 748
    },
    {
      "epoch": 0.8444193912063134,
      "grad_norm": 2.96875,
      "learning_rate": 9.39418935019705e-06,
      "loss": 1.4156,
      "step": 749
    },
    {
      "epoch": 0.8455467869222097,
      "grad_norm": 3.875,
      "learning_rate": 9.392771968446159e-06,
      "loss": 1.8742,
      "step": 750
    },
    {
      "epoch": 0.846674182638106,
      "grad_norm": 3.703125,
      "learning_rate": 9.391352244560728e-06,
      "loss": 1.6635,
      "step": 751
    },
    {
      "epoch": 0.8478015783540023,
      "grad_norm": 3.09375,
      "learning_rate": 9.389930172730614e-06,
      "loss": 1.4239,
      "step": 752
    },
    {
      "epoch": 0.8489289740698985,
      "grad_norm": 3.484375,
      "learning_rate": 9.388505747126438e-06,
      "loss": 1.6623,
      "step": 753
    },
    {
      "epoch": 0.8500563697857948,
      "grad_norm": 3.796875,
      "learning_rate": 9.387078961899505e-06,
      "loss": 1.5389,
      "step": 754
    },
    {
      "epoch": 0.8511837655016911,
      "grad_norm": 3.78125,
      "learning_rate": 9.385649811181727e-06,
      "loss": 1.804,
      "step": 755
    },
    {
      "epoch": 0.8523111612175873,
      "grad_norm": 3.59375,
      "learning_rate": 9.384218289085547e-06,
      "loss": 1.6367,
      "step": 756
    },
    {
      "epoch": 0.8534385569334837,
      "grad_norm": 3.84375,
      "learning_rate": 9.382784389703847e-06,
      "loss": 1.3663,
      "step": 757
    },
    {
      "epoch": 0.85456595264938,
      "grad_norm": 3.453125,
      "learning_rate": 9.38134810710988e-06,
      "loss": 1.5136,
      "step": 758
    },
    {
      "epoch": 0.8556933483652762,
      "grad_norm": 3.59375,
      "learning_rate": 9.379909435357177e-06,
      "loss": 1.5709,
      "step": 759
    },
    {
      "epoch": 0.8568207440811725,
      "grad_norm": 3.21875,
      "learning_rate": 9.378468368479467e-06,
      "loss": 1.6866,
      "step": 760
    },
    {
      "epoch": 0.8579481397970687,
      "grad_norm": 3.09375,
      "learning_rate": 9.377024900490605e-06,
      "loss": 1.6377,
      "step": 761
    },
    {
      "epoch": 0.859075535512965,
      "grad_norm": 3.078125,
      "learning_rate": 9.375579025384474e-06,
      "loss": 1.5065,
      "step": 762
    },
    {
      "epoch": 0.8602029312288614,
      "grad_norm": 3.1875,
      "learning_rate": 9.37413073713491e-06,
      "loss": 1.7366,
      "step": 763
    },
    {
      "epoch": 0.8613303269447576,
      "grad_norm": 3.109375,
      "learning_rate": 9.372680029695621e-06,
      "loss": 1.4707,
      "step": 764
    },
    {
      "epoch": 0.8624577226606539,
      "grad_norm": 3.515625,
      "learning_rate": 9.371226897000094e-06,
      "loss": 1.6906,
      "step": 765
    },
    {
      "epoch": 0.8635851183765502,
      "grad_norm": 3.703125,
      "learning_rate": 9.369771332961518e-06,
      "loss": 1.5372,
      "step": 766
    },
    {
      "epoch": 0.8647125140924464,
      "grad_norm": 3.5625,
      "learning_rate": 9.368313331472695e-06,
      "loss": 1.8378,
      "step": 767
    },
    {
      "epoch": 0.8658399098083427,
      "grad_norm": 3.296875,
      "learning_rate": 9.36685288640596e-06,
      "loss": 1.4621,
      "step": 768
    },
    {
      "epoch": 0.8669673055242391,
      "grad_norm": 2.84375,
      "learning_rate": 9.365389991613086e-06,
      "loss": 1.7535,
      "step": 769
    },
    {
      "epoch": 0.8680947012401353,
      "grad_norm": 3.109375,
      "learning_rate": 9.363924640925202e-06,
      "loss": 1.7386,
      "step": 770
    },
    {
      "epoch": 0.8692220969560316,
      "grad_norm": 3.171875,
      "learning_rate": 9.362456828152714e-06,
      "loss": 1.4926,
      "step": 771
    },
    {
      "epoch": 0.8703494926719278,
      "grad_norm": 3.390625,
      "learning_rate": 9.360986547085203e-06,
      "loss": 1.5779,
      "step": 772
    },
    {
      "epoch": 0.8714768883878241,
      "grad_norm": 3.34375,
      "learning_rate": 9.359513791491353e-06,
      "loss": 1.502,
      "step": 773
    },
    {
      "epoch": 0.8726042841037204,
      "grad_norm": 3.453125,
      "learning_rate": 9.358038555118849e-06,
      "loss": 1.7193,
      "step": 774
    },
    {
      "epoch": 0.8737316798196166,
      "grad_norm": 3.015625,
      "learning_rate": 9.356560831694299e-06,
      "loss": 1.4015,
      "step": 775
    },
    {
      "epoch": 0.874859075535513,
      "grad_norm": 3.25,
      "learning_rate": 9.355080614923134e-06,
      "loss": 1.5895,
      "step": 776
    },
    {
      "epoch": 0.8759864712514093,
      "grad_norm": 7.9375,
      "learning_rate": 9.35359789848954e-06,
      "loss": 1.5285,
      "step": 777
    },
    {
      "epoch": 0.8771138669673055,
      "grad_norm": 3.265625,
      "learning_rate": 9.35211267605634e-06,
      "loss": 1.4308,
      "step": 778
    },
    {
      "epoch": 0.8782412626832018,
      "grad_norm": 4.125,
      "learning_rate": 9.35062494126492e-06,
      "loss": 1.9012,
      "step": 779
    },
    {
      "epoch": 0.8793686583990981,
      "grad_norm": 3.515625,
      "learning_rate": 9.34913468773514e-06,
      "loss": 1.6944,
      "step": 780
    },
    {
      "epoch": 0.8804960541149943,
      "grad_norm": 4.125,
      "learning_rate": 9.347641909065236e-06,
      "loss": 2.0202,
      "step": 781
    },
    {
      "epoch": 0.8816234498308907,
      "grad_norm": 3.46875,
      "learning_rate": 9.346146598831732e-06,
      "loss": 1.8744,
      "step": 782
    },
    {
      "epoch": 0.882750845546787,
      "grad_norm": 3.828125,
      "learning_rate": 9.344648750589346e-06,
      "loss": 1.7919,
      "step": 783
    },
    {
      "epoch": 0.8838782412626832,
      "grad_norm": 3.40625,
      "learning_rate": 9.343148357870895e-06,
      "loss": 1.7713,
      "step": 784
    },
    {
      "epoch": 0.8850056369785795,
      "grad_norm": 3.484375,
      "learning_rate": 9.341645414187212e-06,
      "loss": 1.698,
      "step": 785
    },
    {
      "epoch": 0.8861330326944757,
      "grad_norm": 3.5,
      "learning_rate": 9.340139913027038e-06,
      "loss": 1.6789,
      "step": 786
    },
    {
      "epoch": 0.887260428410372,
      "grad_norm": 3.40625,
      "learning_rate": 9.338631847856941e-06,
      "loss": 1.7609,
      "step": 787
    },
    {
      "epoch": 0.8883878241262683,
      "grad_norm": 3.25,
      "learning_rate": 9.337121212121212e-06,
      "loss": 1.4006,
      "step": 788
    },
    {
      "epoch": 0.8895152198421646,
      "grad_norm": 3.078125,
      "learning_rate": 9.33560799924178e-06,
      "loss": 1.7291,
      "step": 789
    },
    {
      "epoch": 0.8906426155580609,
      "grad_norm": 3.1875,
      "learning_rate": 9.3340922026181e-06,
      "loss": 1.4335,
      "step": 790
    },
    {
      "epoch": 0.8917700112739572,
      "grad_norm": 3.125,
      "learning_rate": 9.332573815627079e-06,
      "loss": 1.7791,
      "step": 791
    },
    {
      "epoch": 0.8928974069898534,
      "grad_norm": 3.359375,
      "learning_rate": 9.331052831622958e-06,
      "loss": 1.8336,
      "step": 792
    },
    {
      "epoch": 0.8940248027057497,
      "grad_norm": 3.828125,
      "learning_rate": 9.329529243937234e-06,
      "loss": 1.6809,
      "step": 793
    },
    {
      "epoch": 0.895152198421646,
      "grad_norm": 3.109375,
      "learning_rate": 9.328003045878548e-06,
      "loss": 1.7117,
      "step": 794
    },
    {
      "epoch": 0.8962795941375423,
      "grad_norm": 3.5,
      "learning_rate": 9.32647423073259e-06,
      "loss": 1.7183,
      "step": 795
    },
    {
      "epoch": 0.8974069898534386,
      "grad_norm": 3.09375,
      "learning_rate": 9.324942791762014e-06,
      "loss": 1.6755,
      "step": 796
    },
    {
      "epoch": 0.8985343855693348,
      "grad_norm": 3.078125,
      "learning_rate": 9.323408722206318e-06,
      "loss": 1.6148,
      "step": 797
    },
    {
      "epoch": 0.8996617812852311,
      "grad_norm": 3.53125,
      "learning_rate": 9.321872015281758e-06,
      "loss": 1.7439,
      "step": 798
    },
    {
      "epoch": 0.9007891770011274,
      "grad_norm": 3.09375,
      "learning_rate": 9.320332664181245e-06,
      "loss": 1.5594,
      "step": 799
    },
    {
      "epoch": 0.9019165727170236,
      "grad_norm": 3.546875,
      "learning_rate": 9.318790662074246e-06,
      "loss": 1.6319,
      "step": 800
    },
    {
      "epoch": 0.90304396843292,
      "grad_norm": 3.96875,
      "learning_rate": 9.317246002106675e-06,
      "loss": 1.5949,
      "step": 801
    },
    {
      "epoch": 0.9041713641488163,
      "grad_norm": 3.203125,
      "learning_rate": 9.315698677400806e-06,
      "loss": 1.7452,
      "step": 802
    },
    {
      "epoch": 0.9052987598647125,
      "grad_norm": 3.53125,
      "learning_rate": 9.314148681055156e-06,
      "loss": 1.4137,
      "step": 803
    },
    {
      "epoch": 0.9064261555806088,
      "grad_norm": 2.734375,
      "learning_rate": 9.312596006144394e-06,
      "loss": 1.471,
      "step": 804
    },
    {
      "epoch": 0.9075535512965051,
      "grad_norm": 3.4375,
      "learning_rate": 9.311040645719229e-06,
      "loss": 1.5523,
      "step": 805
    },
    {
      "epoch": 0.9086809470124013,
      "grad_norm": 3.078125,
      "learning_rate": 9.30948259280631e-06,
      "loss": 1.6964,
      "step": 806
    },
    {
      "epoch": 0.9098083427282976,
      "grad_norm": 4.4375,
      "learning_rate": 9.307921840408125e-06,
      "loss": 1.7243,
      "step": 807
    },
    {
      "epoch": 0.910935738444194,
      "grad_norm": 2.78125,
      "learning_rate": 9.306358381502892e-06,
      "loss": 1.5565,
      "step": 808
    },
    {
      "epoch": 0.9120631341600902,
      "grad_norm": 2.859375,
      "learning_rate": 9.304792209044453e-06,
      "loss": 1.7734,
      "step": 809
    },
    {
      "epoch": 0.9131905298759865,
      "grad_norm": 2.78125,
      "learning_rate": 9.30322331596217e-06,
      "loss": 1.5408,
      "step": 810
    },
    {
      "epoch": 0.9143179255918827,
      "grad_norm": 3.265625,
      "learning_rate": 9.301651695160824e-06,
      "loss": 1.6421,
      "step": 811
    },
    {
      "epoch": 0.915445321307779,
      "grad_norm": 3.0625,
      "learning_rate": 9.300077339520497e-06,
      "loss": 1.5401,
      "step": 812
    },
    {
      "epoch": 0.9165727170236753,
      "grad_norm": 2.765625,
      "learning_rate": 9.29850024189647e-06,
      "loss": 1.629,
      "step": 813
    },
    {
      "epoch": 0.9177001127395716,
      "grad_norm": 4.15625,
      "learning_rate": 9.296920395119119e-06,
      "loss": 1.7156,
      "step": 814
    },
    {
      "epoch": 0.9188275084554679,
      "grad_norm": 2.78125,
      "learning_rate": 9.295337791993797e-06,
      "loss": 1.5553,
      "step": 815
    },
    {
      "epoch": 0.9199549041713642,
      "grad_norm": 2.984375,
      "learning_rate": 9.293752425300738e-06,
      "loss": 1.7641,
      "step": 816
    },
    {
      "epoch": 0.9210822998872604,
      "grad_norm": 3.25,
      "learning_rate": 9.292164287794931e-06,
      "loss": 1.5463,
      "step": 817
    },
    {
      "epoch": 0.9222096956031567,
      "grad_norm": 3.234375,
      "learning_rate": 9.290573372206025e-06,
      "loss": 1.6798,
      "step": 818
    },
    {
      "epoch": 0.923337091319053,
      "grad_norm": 3.234375,
      "learning_rate": 9.288979671238206e-06,
      "loss": 1.6036,
      "step": 819
    },
    {
      "epoch": 0.9244644870349493,
      "grad_norm": 3.09375,
      "learning_rate": 9.287383177570093e-06,
      "loss": 1.4552,
      "step": 820
    },
    {
      "epoch": 0.9255918827508456,
      "grad_norm": 3.1875,
      "learning_rate": 9.285783883854625e-06,
      "loss": 1.5538,
      "step": 821
    },
    {
      "epoch": 0.9267192784667418,
      "grad_norm": 3.484375,
      "learning_rate": 9.28418178271894e-06,
      "loss": 1.5624,
      "step": 822
    },
    {
      "epoch": 0.9278466741826381,
      "grad_norm": 3.265625,
      "learning_rate": 9.282576866764276e-06,
      "loss": 1.513,
      "step": 823
    },
    {
      "epoch": 0.9289740698985344,
      "grad_norm": 3.15625,
      "learning_rate": 9.280969128565847e-06,
      "loss": 1.6548,
      "step": 824
    },
    {
      "epoch": 0.9301014656144306,
      "grad_norm": 3.453125,
      "learning_rate": 9.27935856067273e-06,
      "loss": 1.866,
      "step": 825
    },
    {
      "epoch": 0.9312288613303269,
      "grad_norm": 3.328125,
      "learning_rate": 9.277745155607753e-06,
      "loss": 1.5337,
      "step": 826
    },
    {
      "epoch": 0.9323562570462233,
      "grad_norm": 2.984375,
      "learning_rate": 9.276128905867374e-06,
      "loss": 1.6386,
      "step": 827
    },
    {
      "epoch": 0.9334836527621195,
      "grad_norm": 2.953125,
      "learning_rate": 9.27450980392157e-06,
      "loss": 1.7085,
      "step": 828
    },
    {
      "epoch": 0.9346110484780158,
      "grad_norm": 2.625,
      "learning_rate": 9.27288784221372e-06,
      "loss": 1.5551,
      "step": 829
    },
    {
      "epoch": 0.9357384441939121,
      "grad_norm": 2.703125,
      "learning_rate": 9.27126301316048e-06,
      "loss": 1.5283,
      "step": 830
    },
    {
      "epoch": 0.9368658399098083,
      "grad_norm": 4.1875,
      "learning_rate": 9.269635309151678e-06,
      "loss": 1.7804,
      "step": 831
    },
    {
      "epoch": 0.9379932356257046,
      "grad_norm": 3.546875,
      "learning_rate": 9.26800472255018e-06,
      "loss": 1.6568,
      "step": 832
    },
    {
      "epoch": 0.939120631341601,
      "grad_norm": 3.125,
      "learning_rate": 9.266371245691778e-06,
      "loss": 1.5002,
      "step": 833
    },
    {
      "epoch": 0.9402480270574972,
      "grad_norm": 2.9375,
      "learning_rate": 9.264734870885078e-06,
      "loss": 1.6161,
      "step": 834
    },
    {
      "epoch": 0.9413754227733935,
      "grad_norm": 3.40625,
      "learning_rate": 9.263095590411364e-06,
      "loss": 1.6907,
      "step": 835
    },
    {
      "epoch": 0.9425028184892897,
      "grad_norm": 3.046875,
      "learning_rate": 9.261453396524487e-06,
      "loss": 1.5391,
      "step": 836
    },
    {
      "epoch": 0.943630214205186,
      "grad_norm": 2.390625,
      "learning_rate": 9.259808281450736e-06,
      "loss": 1.4727,
      "step": 837
    },
    {
      "epoch": 0.9447576099210823,
      "grad_norm": 3.21875,
      "learning_rate": 9.258160237388725e-06,
      "loss": 1.5999,
      "step": 838
    },
    {
      "epoch": 0.9458850056369785,
      "grad_norm": 3.375,
      "learning_rate": 9.256509256509257e-06,
      "loss": 1.6536,
      "step": 839
    },
    {
      "epoch": 0.9470124013528749,
      "grad_norm": 3.421875,
      "learning_rate": 9.254855330955212e-06,
      "loss": 1.6861,
      "step": 840
    },
    {
      "epoch": 0.9481397970687712,
      "grad_norm": 3.375,
      "learning_rate": 9.253198452841416e-06,
      "loss": 1.5666,
      "step": 841
    },
    {
      "epoch": 0.9492671927846674,
      "grad_norm": 3.28125,
      "learning_rate": 9.251538614254517e-06,
      "loss": 1.5615,
      "step": 842
    },
    {
      "epoch": 0.9503945885005637,
      "grad_norm": 3.375,
      "learning_rate": 9.249875807252857e-06,
      "loss": 1.6008,
      "step": 843
    },
    {
      "epoch": 0.95152198421646,
      "grad_norm": 3.40625,
      "learning_rate": 9.248210023866349e-06,
      "loss": 1.6909,
      "step": 844
    },
    {
      "epoch": 0.9526493799323562,
      "grad_norm": 3.359375,
      "learning_rate": 9.246541256096348e-06,
      "loss": 1.6791,
      "step": 845
    },
    {
      "epoch": 0.9537767756482526,
      "grad_norm": 3.46875,
      "learning_rate": 9.24486949591552e-06,
      "loss": 1.4178,
      "step": 846
    },
    {
      "epoch": 0.9549041713641488,
      "grad_norm": 2.890625,
      "learning_rate": 9.243194735267724e-06,
      "loss": 1.506,
      "step": 847
    },
    {
      "epoch": 0.9560315670800451,
      "grad_norm": 3.6875,
      "learning_rate": 9.241516966067866e-06,
      "loss": 1.4603,
      "step": 848
    },
    {
      "epoch": 0.9571589627959414,
      "grad_norm": 3.671875,
      "learning_rate": 9.239836180201778e-06,
      "loss": 1.5911,
      "step": 849
    },
    {
      "epoch": 0.9582863585118376,
      "grad_norm": 3.40625,
      "learning_rate": 9.238152369526095e-06,
      "loss": 1.7094,
      "step": 850
    },
    {
      "epoch": 0.9594137542277339,
      "grad_norm": 3.828125,
      "learning_rate": 9.236465525868108e-06,
      "loss": 1.6584,
      "step": 851
    },
    {
      "epoch": 0.9605411499436303,
      "grad_norm": 3.453125,
      "learning_rate": 9.23477564102564e-06,
      "loss": 1.738,
      "step": 852
    },
    {
      "epoch": 0.9616685456595265,
      "grad_norm": 3.28125,
      "learning_rate": 9.233082706766918e-06,
      "loss": 1.6107,
      "step": 853
    },
    {
      "epoch": 0.9627959413754228,
      "grad_norm": 3.078125,
      "learning_rate": 9.231386714830424e-06,
      "loss": 1.5704,
      "step": 854
    },
    {
      "epoch": 0.963923337091319,
      "grad_norm": 3.296875,
      "learning_rate": 9.229687656924776e-06,
      "loss": 1.6499,
      "step": 855
    },
    {
      "epoch": 0.9650507328072153,
      "grad_norm": 3.265625,
      "learning_rate": 9.227985524728589e-06,
      "loss": 1.7179,
      "step": 856
    },
    {
      "epoch": 0.9661781285231116,
      "grad_norm": 3.140625,
      "learning_rate": 9.226280309890331e-06,
      "loss": 1.7951,
      "step": 857
    },
    {
      "epoch": 0.9673055242390078,
      "grad_norm": 3.4375,
      "learning_rate": 9.224572004028199e-06,
      "loss": 1.6809,
      "step": 858
    },
    {
      "epoch": 0.9684329199549042,
      "grad_norm": 5.15625,
      "learning_rate": 9.222860598729967e-06,
      "loss": 1.6217,
      "step": 859
    },
    {
      "epoch": 0.9695603156708005,
      "grad_norm": 3.0,
      "learning_rate": 9.221146085552865e-06,
      "loss": 1.6503,
      "step": 860
    },
    {
      "epoch": 0.9706877113866967,
      "grad_norm": 2.703125,
      "learning_rate": 9.219428456023428e-06,
      "loss": 1.4897,
      "step": 861
    },
    {
      "epoch": 0.971815107102593,
      "grad_norm": 2.671875,
      "learning_rate": 9.217707701637357e-06,
      "loss": 1.6116,
      "step": 862
    },
    {
      "epoch": 0.9729425028184893,
      "grad_norm": 3.203125,
      "learning_rate": 9.215983813859384e-06,
      "loss": 1.8738,
      "step": 863
    },
    {
      "epoch": 0.9740698985343855,
      "grad_norm": 3.234375,
      "learning_rate": 9.214256784123128e-06,
      "loss": 1.5292,
      "step": 864
    },
    {
      "epoch": 0.9751972942502819,
      "grad_norm": 3.421875,
      "learning_rate": 9.212526603830953e-06,
      "loss": 1.7098,
      "step": 865
    },
    {
      "epoch": 0.9763246899661782,
      "grad_norm": 3.046875,
      "learning_rate": 9.210793264353825e-06,
      "loss": 1.3685,
      "step": 866
    },
    {
      "epoch": 0.9774520856820744,
      "grad_norm": 3.453125,
      "learning_rate": 9.209056757031171e-06,
      "loss": 1.7727,
      "step": 867
    },
    {
      "epoch": 0.9785794813979707,
      "grad_norm": 3.40625,
      "learning_rate": 9.207317073170733e-06,
      "loss": 1.5483,
      "step": 868
    },
    {
      "epoch": 0.979706877113867,
      "grad_norm": 3.203125,
      "learning_rate": 9.205574204048417e-06,
      "loss": 1.4888,
      "step": 869
    },
    {
      "epoch": 0.9808342728297632,
      "grad_norm": 3.078125,
      "learning_rate": 9.203828140908166e-06,
      "loss": 1.6672,
      "step": 870
    },
    {
      "epoch": 0.9819616685456595,
      "grad_norm": 3.453125,
      "learning_rate": 9.202078874961784e-06,
      "loss": 1.6311,
      "step": 871
    },
    {
      "epoch": 0.9830890642615558,
      "grad_norm": 3.140625,
      "learning_rate": 9.200326397388821e-06,
      "loss": 1.5886,
      "step": 872
    },
    {
      "epoch": 0.9842164599774521,
      "grad_norm": 3.265625,
      "learning_rate": 9.198570699336397e-06,
      "loss": 1.6038,
      "step": 873
    },
    {
      "epoch": 0.9853438556933484,
      "grad_norm": 3.0,
      "learning_rate": 9.196811771919068e-06,
      "loss": 1.5722,
      "step": 874
    },
    {
      "epoch": 0.9864712514092446,
      "grad_norm": 3.765625,
      "learning_rate": 9.195049606218677e-06,
      "loss": 1.9187,
      "step": 875
    },
    {
      "epoch": 0.9875986471251409,
      "grad_norm": 3.40625,
      "learning_rate": 9.193284193284193e-06,
      "loss": 1.9184,
      "step": 876
    },
    {
      "epoch": 0.9887260428410372,
      "grad_norm": 3.734375,
      "learning_rate": 9.191515524131571e-06,
      "loss": 1.6248,
      "step": 877
    },
    {
      "epoch": 0.9898534385569335,
      "grad_norm": 2.890625,
      "learning_rate": 9.18974358974359e-06,
      "loss": 1.3455,
      "step": 878
    },
    {
      "epoch": 0.9909808342728298,
      "grad_norm": 3.703125,
      "learning_rate": 9.187968381069706e-06,
      "loss": 1.7755,
      "step": 879
    },
    {
      "epoch": 0.992108229988726,
      "grad_norm": 3.703125,
      "learning_rate": 9.186189889025895e-06,
      "loss": 1.6777,
      "step": 880
    },
    {
      "epoch": 0.9932356257046223,
      "grad_norm": 3.046875,
      "learning_rate": 9.184408104494499e-06,
      "loss": 1.5603,
      "step": 881
    },
    {
      "epoch": 0.9943630214205186,
      "grad_norm": 3.0625,
      "learning_rate": 9.182623018324069e-06,
      "loss": 1.3597,
      "step": 882
    },
    {
      "epoch": 0.9954904171364148,
      "grad_norm": 3.109375,
      "learning_rate": 9.180834621329212e-06,
      "loss": 1.726,
      "step": 883
    },
    {
      "epoch": 0.9966178128523112,
      "grad_norm": 2.96875,
      "learning_rate": 9.179042904290431e-06,
      "loss": 1.7267,
      "step": 884
    },
    {
      "epoch": 0.9977452085682075,
      "grad_norm": 2.421875,
      "learning_rate": 9.177247857953961e-06,
      "loss": 1.5367,
      "step": 885
    },
    {
      "epoch": 0.9988726042841037,
      "grad_norm": 2.59375,
      "learning_rate": 9.175449473031619e-06,
      "loss": 1.5634,
      "step": 886
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.984375,
      "learning_rate": 9.173647740200642e-06,
      "loss": 1.5659,
      "step": 887
    },
    {
      "epoch": 1.0011273957158964,
      "grad_norm": 3.34375,
      "learning_rate": 9.171842650103519e-06,
      "loss": 1.696,
      "step": 888
    }
  ],
  "logging_steps": 1,
  "max_steps": 1774,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 444,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 8.084609946353664e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}