{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 956,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010460251046025104,
      "grad_norm": 0.0654296875,
      "learning_rate": 9.989539748953976e-06,
      "loss": 1.3539,
      "step": 1
    },
    {
      "epoch": 0.0020920502092050207,
      "grad_norm": 0.0703125,
      "learning_rate": 9.97907949790795e-06,
      "loss": 1.3735,
      "step": 2
    },
    {
      "epoch": 0.0031380753138075313,
      "grad_norm": 0.068359375,
      "learning_rate": 9.968619246861927e-06,
      "loss": 1.3617,
      "step": 3
    },
    {
      "epoch": 0.0041841004184100415,
      "grad_norm": 0.0693359375,
      "learning_rate": 9.958158995815901e-06,
      "loss": 1.3627,
      "step": 4
    },
    {
      "epoch": 0.005230125523012552,
      "grad_norm": 0.07275390625,
      "learning_rate": 9.947698744769874e-06,
      "loss": 1.3958,
      "step": 5
    },
    {
      "epoch": 0.006276150627615063,
      "grad_norm": 0.0791015625,
      "learning_rate": 9.937238493723849e-06,
      "loss": 1.4488,
      "step": 6
    },
    {
      "epoch": 0.007322175732217573,
      "grad_norm": 0.08203125,
      "learning_rate": 9.926778242677825e-06,
      "loss": 1.5159,
      "step": 7
    },
    {
      "epoch": 0.008368200836820083,
      "grad_norm": 0.08056640625,
      "learning_rate": 9.9163179916318e-06,
      "loss": 1.3692,
      "step": 8
    },
    {
      "epoch": 0.009414225941422594,
      "grad_norm": 0.08642578125,
      "learning_rate": 9.905857740585775e-06,
      "loss": 1.4844,
      "step": 9
    },
    {
      "epoch": 0.010460251046025104,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.89539748953975e-06,
      "loss": 1.3112,
      "step": 10
    },
    {
      "epoch": 0.011506276150627616,
      "grad_norm": 0.08642578125,
      "learning_rate": 9.884937238493724e-06,
      "loss": 1.3439,
      "step": 11
    },
    {
      "epoch": 0.012552301255230125,
      "grad_norm": 0.0849609375,
      "learning_rate": 9.874476987447699e-06,
      "loss": 1.2908,
      "step": 12
    },
    {
      "epoch": 0.013598326359832637,
      "grad_norm": 0.09326171875,
      "learning_rate": 9.864016736401675e-06,
      "loss": 1.4019,
      "step": 13
    },
    {
      "epoch": 0.014644351464435146,
      "grad_norm": 0.10205078125,
      "learning_rate": 9.85355648535565e-06,
      "loss": 1.4304,
      "step": 14
    },
    {
      "epoch": 0.015690376569037656,
      "grad_norm": 0.10400390625,
      "learning_rate": 9.843096234309623e-06,
      "loss": 1.3832,
      "step": 15
    },
    {
      "epoch": 0.016736401673640166,
      "grad_norm": 0.10546875,
      "learning_rate": 9.8326359832636e-06,
      "loss": 1.3762,
      "step": 16
    },
    {
      "epoch": 0.01778242677824268,
      "grad_norm": 0.11083984375,
      "learning_rate": 9.822175732217574e-06,
      "loss": 1.4184,
      "step": 17
    },
    {
      "epoch": 0.01882845188284519,
      "grad_norm": 0.10888671875,
      "learning_rate": 9.811715481171549e-06,
      "loss": 1.3479,
      "step": 18
    },
    {
      "epoch": 0.0198744769874477,
      "grad_norm": 0.11669921875,
      "learning_rate": 9.801255230125524e-06,
      "loss": 1.3843,
      "step": 19
    },
    {
      "epoch": 0.02092050209205021,
      "grad_norm": 0.11279296875,
      "learning_rate": 9.790794979079498e-06,
      "loss": 1.2884,
      "step": 20
    },
    {
      "epoch": 0.021966527196652718,
      "grad_norm": 0.11083984375,
      "learning_rate": 9.780334728033473e-06,
      "loss": 1.2661,
      "step": 21
    },
    {
      "epoch": 0.02301255230125523,
      "grad_norm": 0.12109375,
      "learning_rate": 9.76987447698745e-06,
      "loss": 1.3819,
      "step": 22
    },
    {
      "epoch": 0.02405857740585774,
      "grad_norm": 0.1240234375,
      "learning_rate": 9.759414225941424e-06,
      "loss": 1.3495,
      "step": 23
    },
    {
      "epoch": 0.02510460251046025,
      "grad_norm": 0.138671875,
      "learning_rate": 9.748953974895399e-06,
      "loss": 1.4215,
      "step": 24
    },
    {
      "epoch": 0.02615062761506276,
      "grad_norm": 0.12890625,
      "learning_rate": 9.738493723849372e-06,
      "loss": 1.3194,
      "step": 25
    },
    {
      "epoch": 0.027196652719665274,
      "grad_norm": 0.13671875,
      "learning_rate": 9.728033472803348e-06,
      "loss": 1.3578,
      "step": 26
    },
    {
      "epoch": 0.028242677824267783,
      "grad_norm": 0.140625,
      "learning_rate": 9.717573221757323e-06,
      "loss": 1.387,
      "step": 27
    },
    {
      "epoch": 0.029288702928870293,
      "grad_norm": 0.1435546875,
      "learning_rate": 9.707112970711298e-06,
      "loss": 1.3147,
      "step": 28
    },
    {
      "epoch": 0.030334728033472803,
      "grad_norm": 0.14453125,
      "learning_rate": 9.696652719665272e-06,
      "loss": 1.3406,
      "step": 29
    },
    {
      "epoch": 0.03138075313807531,
      "grad_norm": 0.162109375,
      "learning_rate": 9.686192468619247e-06,
      "loss": 1.4469,
      "step": 30
    },
    {
      "epoch": 0.032426778242677826,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.675732217573223e-06,
      "loss": 1.4211,
      "step": 31
    },
    {
      "epoch": 0.03347280334728033,
      "grad_norm": 0.173828125,
      "learning_rate": 9.665271966527198e-06,
      "loss": 1.3831,
      "step": 32
    },
    {
      "epoch": 0.034518828451882845,
      "grad_norm": 0.1552734375,
      "learning_rate": 9.654811715481173e-06,
      "loss": 1.3041,
      "step": 33
    },
    {
      "epoch": 0.03556485355648536,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.644351464435148e-06,
      "loss": 1.3841,
      "step": 34
    },
    {
      "epoch": 0.036610878661087864,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.633891213389122e-06,
      "loss": 1.4276,
      "step": 35
    },
    {
      "epoch": 0.03765690376569038,
      "grad_norm": 0.1689453125,
      "learning_rate": 9.623430962343097e-06,
      "loss": 1.3756,
      "step": 36
    },
    {
      "epoch": 0.038702928870292884,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.612970711297072e-06,
      "loss": 1.3659,
      "step": 37
    },
    {
      "epoch": 0.0397489539748954,
      "grad_norm": 0.171875,
      "learning_rate": 9.602510460251046e-06,
      "loss": 1.2676,
      "step": 38
    },
    {
      "epoch": 0.04079497907949791,
      "grad_norm": 0.185546875,
      "learning_rate": 9.592050209205021e-06,
      "loss": 1.4119,
      "step": 39
    },
    {
      "epoch": 0.04184100418410042,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.581589958158996e-06,
      "loss": 1.3362,
      "step": 40
    },
    {
      "epoch": 0.04288702928870293,
      "grad_norm": 0.1796875,
      "learning_rate": 9.571129707112972e-06,
      "loss": 1.3414,
      "step": 41
    },
    {
      "epoch": 0.043933054393305436,
      "grad_norm": 0.181640625,
      "learning_rate": 9.560669456066947e-06,
      "loss": 1.3427,
      "step": 42
    },
    {
      "epoch": 0.04497907949790795,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.550209205020922e-06,
      "loss": 1.24,
      "step": 43
    },
    {
      "epoch": 0.04602510460251046,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.539748953974896e-06,
      "loss": 1.2545,
      "step": 44
    },
    {
      "epoch": 0.04707112970711297,
      "grad_norm": 0.177734375,
      "learning_rate": 9.529288702928871e-06,
      "loss": 1.2704,
      "step": 45
    },
    {
      "epoch": 0.04811715481171548,
      "grad_norm": 0.1865234375,
      "learning_rate": 9.518828451882846e-06,
      "loss": 1.3125,
      "step": 46
    },
    {
      "epoch": 0.049163179916317995,
      "grad_norm": 0.19921875,
      "learning_rate": 9.50836820083682e-06,
      "loss": 1.3691,
      "step": 47
    },
    {
      "epoch": 0.0502092050209205,
      "grad_norm": 0.1806640625,
      "learning_rate": 9.497907949790795e-06,
      "loss": 1.2617,
      "step": 48
    },
    {
      "epoch": 0.051255230125523014,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.48744769874477e-06,
      "loss": 1.2553,
      "step": 49
    },
    {
      "epoch": 0.05230125523012552,
      "grad_norm": 0.18359375,
      "learning_rate": 9.476987447698746e-06,
      "loss": 1.2782,
      "step": 50
    },
    {
      "epoch": 0.053347280334728034,
      "grad_norm": 0.1845703125,
      "learning_rate": 9.466527196652721e-06,
      "loss": 1.2378,
      "step": 51
    },
    {
      "epoch": 0.05439330543933055,
      "grad_norm": 0.181640625,
      "learning_rate": 9.456066945606696e-06,
      "loss": 1.2952,
      "step": 52
    },
    {
      "epoch": 0.05543933054393305,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.44560669456067e-06,
      "loss": 1.3791,
      "step": 53
    },
    {
      "epoch": 0.056485355648535567,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.435146443514645e-06,
      "loss": 1.247,
      "step": 54
    },
    {
      "epoch": 0.05753138075313807,
      "grad_norm": 0.1806640625,
      "learning_rate": 9.42468619246862e-06,
      "loss": 1.2619,
      "step": 55
    },
    {
      "epoch": 0.058577405857740586,
      "grad_norm": 0.1845703125,
      "learning_rate": 9.414225941422594e-06,
      "loss": 1.2701,
      "step": 56
    },
    {
      "epoch": 0.0596234309623431,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.40376569037657e-06,
      "loss": 1.344,
      "step": 57
    },
    {
      "epoch": 0.060669456066945605,
      "grad_norm": 0.1845703125,
      "learning_rate": 9.393305439330544e-06,
      "loss": 1.3005,
      "step": 58
    },
    {
      "epoch": 0.06171548117154812,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.38284518828452e-06,
      "loss": 1.2562,
      "step": 59
    },
    {
      "epoch": 0.06276150627615062,
      "grad_norm": 0.1806640625,
      "learning_rate": 9.372384937238495e-06,
      "loss": 1.2649,
      "step": 60
    },
    {
      "epoch": 0.06380753138075314,
      "grad_norm": 0.171875,
      "learning_rate": 9.36192468619247e-06,
      "loss": 1.1994,
      "step": 61
    },
    {
      "epoch": 0.06485355648535565,
      "grad_norm": 0.177734375,
      "learning_rate": 9.351464435146444e-06,
      "loss": 1.2614,
      "step": 62
    },
    {
      "epoch": 0.06589958158995816,
      "grad_norm": 0.173828125,
      "learning_rate": 9.341004184100419e-06,
      "loss": 1.1784,
      "step": 63
    },
    {
      "epoch": 0.06694560669456066,
      "grad_norm": 0.17578125,
      "learning_rate": 9.330543933054394e-06,
      "loss": 1.2676,
      "step": 64
    },
    {
      "epoch": 0.06799163179916318,
      "grad_norm": 0.169921875,
      "learning_rate": 9.320083682008368e-06,
      "loss": 1.2065,
      "step": 65
    },
    {
      "epoch": 0.06903765690376569,
      "grad_norm": 0.1708984375,
      "learning_rate": 9.309623430962343e-06,
      "loss": 1.1899,
      "step": 66
    },
    {
      "epoch": 0.0700836820083682,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.299163179916318e-06,
      "loss": 1.1644,
      "step": 67
    },
    {
      "epoch": 0.07112970711297072,
      "grad_norm": 0.1884765625,
      "learning_rate": 9.288702928870293e-06,
      "loss": 1.2813,
      "step": 68
    },
    {
      "epoch": 0.07217573221757322,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.278242677824269e-06,
      "loss": 1.2005,
      "step": 69
    },
    {
      "epoch": 0.07322175732217573,
      "grad_norm": 0.181640625,
      "learning_rate": 9.267782426778244e-06,
      "loss": 1.1998,
      "step": 70
    },
    {
      "epoch": 0.07426778242677824,
      "grad_norm": 0.1669921875,
      "learning_rate": 9.257322175732218e-06,
      "loss": 1.1805,
      "step": 71
    },
    {
      "epoch": 0.07531380753138076,
      "grad_norm": 0.158203125,
      "learning_rate": 9.246861924686193e-06,
      "loss": 1.136,
      "step": 72
    },
    {
      "epoch": 0.07635983263598327,
      "grad_norm": 0.1865234375,
      "learning_rate": 9.236401673640168e-06,
      "loss": 1.2927,
      "step": 73
    },
    {
      "epoch": 0.07740585774058577,
      "grad_norm": 0.15625,
      "learning_rate": 9.225941422594144e-06,
      "loss": 1.192,
      "step": 74
    },
    {
      "epoch": 0.07845188284518828,
      "grad_norm": 0.17578125,
      "learning_rate": 9.215481171548117e-06,
      "loss": 1.2096,
      "step": 75
    },
    {
      "epoch": 0.0794979079497908,
      "grad_norm": 0.1689453125,
      "learning_rate": 9.205020920502092e-06,
      "loss": 1.2029,
      "step": 76
    },
    {
      "epoch": 0.08054393305439331,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.194560669456067e-06,
      "loss": 1.1622,
      "step": 77
    },
    {
      "epoch": 0.08158995815899582,
      "grad_norm": 0.1943359375,
      "learning_rate": 9.184100418410043e-06,
      "loss": 1.2874,
      "step": 78
    },
    {
      "epoch": 0.08263598326359832,
      "grad_norm": 0.189453125,
      "learning_rate": 9.173640167364018e-06,
      "loss": 1.2342,
      "step": 79
    },
    {
      "epoch": 0.08368200836820083,
      "grad_norm": 0.1650390625,
      "learning_rate": 9.163179916317992e-06,
      "loss": 1.1455,
      "step": 80
    },
    {
      "epoch": 0.08472803347280335,
      "grad_norm": 0.20703125,
      "learning_rate": 9.152719665271967e-06,
      "loss": 1.1988,
      "step": 81
    },
    {
      "epoch": 0.08577405857740586,
      "grad_norm": 0.154296875,
      "learning_rate": 9.142259414225942e-06,
      "loss": 1.1205,
      "step": 82
    },
    {
      "epoch": 0.08682008368200837,
      "grad_norm": 0.154296875,
      "learning_rate": 9.131799163179918e-06,
      "loss": 1.1656,
      "step": 83
    },
    {
      "epoch": 0.08786610878661087,
      "grad_norm": 0.18359375,
      "learning_rate": 9.121338912133893e-06,
      "loss": 1.2441,
      "step": 84
    },
    {
      "epoch": 0.08891213389121339,
      "grad_norm": 0.1630859375,
      "learning_rate": 9.110878661087866e-06,
      "loss": 1.1709,
      "step": 85
    },
    {
      "epoch": 0.0899581589958159,
      "grad_norm": 0.1455078125,
      "learning_rate": 9.10041841004184e-06,
      "loss": 1.1165,
      "step": 86
    },
    {
      "epoch": 0.09100418410041841,
      "grad_norm": 0.15234375,
      "learning_rate": 9.089958158995817e-06,
      "loss": 1.1295,
      "step": 87
    },
    {
      "epoch": 0.09205020920502092,
      "grad_norm": 0.1474609375,
      "learning_rate": 9.079497907949792e-06,
      "loss": 1.1215,
      "step": 88
    },
    {
      "epoch": 0.09309623430962342,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.069037656903766e-06,
      "loss": 1.0779,
      "step": 89
    },
    {
      "epoch": 0.09414225941422594,
      "grad_norm": 0.1513671875,
      "learning_rate": 9.058577405857741e-06,
      "loss": 1.1276,
      "step": 90
    },
    {
      "epoch": 0.09518828451882845,
      "grad_norm": 0.146484375,
      "learning_rate": 9.048117154811716e-06,
      "loss": 1.1666,
      "step": 91
    },
    {
      "epoch": 0.09623430962343096,
      "grad_norm": 0.134765625,
      "learning_rate": 9.03765690376569e-06,
      "loss": 1.0678,
      "step": 92
    },
    {
      "epoch": 0.09728033472803348,
      "grad_norm": 0.1474609375,
      "learning_rate": 9.027196652719667e-06,
      "loss": 1.1405,
      "step": 93
    },
    {
      "epoch": 0.09832635983263599,
      "grad_norm": 0.142578125,
      "learning_rate": 9.016736401673642e-06,
      "loss": 1.1121,
      "step": 94
    },
    {
      "epoch": 0.09937238493723849,
      "grad_norm": 0.150390625,
      "learning_rate": 9.006276150627615e-06,
      "loss": 1.1483,
      "step": 95
    },
    {
      "epoch": 0.100418410041841,
      "grad_norm": 0.1416015625,
      "learning_rate": 8.995815899581591e-06,
      "loss": 1.1121,
      "step": 96
    },
    {
      "epoch": 0.10146443514644352,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.985355648535566e-06,
      "loss": 1.1645,
      "step": 97
    },
    {
      "epoch": 0.10251046025104603,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.97489539748954e-06,
      "loss": 1.1407,
      "step": 98
    },
    {
      "epoch": 0.10355648535564854,
      "grad_norm": 0.173828125,
      "learning_rate": 8.964435146443515e-06,
      "loss": 1.262,
      "step": 99
    },
    {
      "epoch": 0.10460251046025104,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.95397489539749e-06,
      "loss": 1.1214,
      "step": 100
    },
    {
      "epoch": 0.10564853556485355,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.943514644351465e-06,
      "loss": 1.1976,
      "step": 101
    },
    {
      "epoch": 0.10669456066945607,
      "grad_norm": 0.14453125,
      "learning_rate": 8.933054393305441e-06,
      "loss": 1.1277,
      "step": 102
    },
    {
      "epoch": 0.10774058577405858,
      "grad_norm": 0.11767578125,
      "learning_rate": 8.922594142259416e-06,
      "loss": 1.0382,
      "step": 103
    },
    {
      "epoch": 0.1087866108786611,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.91213389121339e-06,
      "loss": 1.1274,
      "step": 104
    },
    {
      "epoch": 0.1098326359832636,
      "grad_norm": 0.1328125,
      "learning_rate": 8.901673640167363e-06,
      "loss": 1.1292,
      "step": 105
    },
    {
      "epoch": 0.1108786610878661,
      "grad_norm": 0.11669921875,
      "learning_rate": 8.89121338912134e-06,
      "loss": 1.0558,
      "step": 106
    },
    {
      "epoch": 0.11192468619246862,
      "grad_norm": 0.12890625,
      "learning_rate": 8.880753138075315e-06,
      "loss": 1.078,
      "step": 107
    },
    {
      "epoch": 0.11297071129707113,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.87029288702929e-06,
      "loss": 1.0754,
      "step": 108
    },
    {
      "epoch": 0.11401673640167365,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.859832635983264e-06,
      "loss": 1.0902,
      "step": 109
    },
    {
      "epoch": 0.11506276150627615,
      "grad_norm": 0.1201171875,
      "learning_rate": 8.849372384937239e-06,
      "loss": 1.0939,
      "step": 110
    },
    {
      "epoch": 0.11610878661087866,
      "grad_norm": 0.130859375,
      "learning_rate": 8.838912133891215e-06,
      "loss": 1.1431,
      "step": 111
    },
    {
      "epoch": 0.11715481171548117,
      "grad_norm": 0.138671875,
      "learning_rate": 8.82845188284519e-06,
      "loss": 1.1457,
      "step": 112
    },
    {
      "epoch": 0.11820083682008369,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.817991631799164e-06,
      "loss": 1.1114,
      "step": 113
    },
    {
      "epoch": 0.1192468619246862,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.80753138075314e-06,
      "loss": 1.1537,
      "step": 114
    },
    {
      "epoch": 0.1202928870292887,
      "grad_norm": 0.134765625,
      "learning_rate": 8.797071129707114e-06,
      "loss": 1.1379,
      "step": 115
    },
    {
      "epoch": 0.12133891213389121,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.786610878661089e-06,
      "loss": 1.0569,
      "step": 116
    },
    {
      "epoch": 0.12238493723849372,
      "grad_norm": 0.1142578125,
      "learning_rate": 8.776150627615063e-06,
      "loss": 1.1038,
      "step": 117
    },
    {
      "epoch": 0.12343096234309624,
      "grad_norm": 0.09375,
      "learning_rate": 8.765690376569038e-06,
      "loss": 0.9908,
      "step": 118
    },
    {
      "epoch": 0.12447698744769875,
      "grad_norm": 0.134765625,
      "learning_rate": 8.755230125523013e-06,
      "loss": 1.1526,
      "step": 119
    },
    {
      "epoch": 0.12552301255230125,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.744769874476987e-06,
      "loss": 1.0158,
      "step": 120
    },
    {
      "epoch": 0.12656903765690378,
      "grad_norm": 0.12890625,
      "learning_rate": 8.734309623430964e-06,
      "loss": 1.1142,
      "step": 121
    },
    {
      "epoch": 0.12761506276150628,
      "grad_norm": 0.1181640625,
      "learning_rate": 8.723849372384939e-06,
      "loss": 1.0441,
      "step": 122
    },
    {
      "epoch": 0.12866108786610878,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.713389121338913e-06,
      "loss": 1.1059,
      "step": 123
    },
    {
      "epoch": 0.1297071129707113,
      "grad_norm": 0.11328125,
      "learning_rate": 8.702928870292888e-06,
      "loss": 1.0326,
      "step": 124
    },
    {
      "epoch": 0.1307531380753138,
      "grad_norm": 0.1201171875,
      "learning_rate": 8.692468619246863e-06,
      "loss": 1.0525,
      "step": 125
    },
    {
      "epoch": 0.13179916317991633,
      "grad_norm": 0.1103515625,
      "learning_rate": 8.682008368200837e-06,
      "loss": 1.0494,
      "step": 126
    },
    {
      "epoch": 0.13284518828451883,
      "grad_norm": 0.10302734375,
      "learning_rate": 8.671548117154812e-06,
      "loss": 1.0117,
      "step": 127
    },
    {
      "epoch": 0.13389121338912133,
      "grad_norm": 0.12890625,
      "learning_rate": 8.661087866108787e-06,
      "loss": 1.1009,
      "step": 128
    },
    {
      "epoch": 0.13493723849372385,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.650627615062761e-06,
      "loss": 1.0536,
      "step": 129
    },
    {
      "epoch": 0.13598326359832635,
      "grad_norm": 0.1162109375,
      "learning_rate": 8.640167364016738e-06,
      "loss": 1.0448,
      "step": 130
    },
    {
      "epoch": 0.13702928870292888,
      "grad_norm": 0.11181640625,
      "learning_rate": 8.629707112970713e-06,
      "loss": 1.0978,
      "step": 131
    },
    {
      "epoch": 0.13807531380753138,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.619246861924687e-06,
      "loss": 1.0372,
      "step": 132
    },
    {
      "epoch": 0.13912133891213388,
      "grad_norm": 0.10009765625,
      "learning_rate": 8.608786610878662e-06,
      "loss": 0.9844,
      "step": 133
    },
    {
      "epoch": 0.1401673640167364,
      "grad_norm": 0.09423828125,
      "learning_rate": 8.598326359832637e-06,
      "loss": 0.9349,
      "step": 134
    },
    {
      "epoch": 0.1412133891213389,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.587866108786611e-06,
      "loss": 1.1065,
      "step": 135
    },
    {
      "epoch": 0.14225941422594143,
      "grad_norm": 0.12158203125,
      "learning_rate": 8.577405857740586e-06,
      "loss": 1.1182,
      "step": 136
    },
    {
      "epoch": 0.14330543933054393,
      "grad_norm": 0.12109375,
      "learning_rate": 8.56694560669456e-06,
      "loss": 1.1216,
      "step": 137
    },
    {
      "epoch": 0.14435146443514643,
      "grad_norm": 0.10791015625,
      "learning_rate": 8.556485355648536e-06,
      "loss": 0.9742,
      "step": 138
    },
    {
      "epoch": 0.14539748953974896,
      "grad_norm": 0.09375,
      "learning_rate": 8.546025104602512e-06,
      "loss": 0.9678,
      "step": 139
    },
    {
      "epoch": 0.14644351464435146,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.535564853556487e-06,
      "loss": 1.0091,
      "step": 140
    },
    {
      "epoch": 0.14748953974895398,
      "grad_norm": 0.109375,
      "learning_rate": 8.525104602510461e-06,
      "loss": 1.059,
      "step": 141
    },
    {
      "epoch": 0.14853556485355648,
      "grad_norm": 0.125,
      "learning_rate": 8.514644351464436e-06,
      "loss": 1.0616,
      "step": 142
    },
    {
      "epoch": 0.14958158995815898,
      "grad_norm": 0.11767578125,
      "learning_rate": 8.50418410041841e-06,
      "loss": 1.0624,
      "step": 143
    },
    {
      "epoch": 0.1506276150627615,
      "grad_norm": 0.111328125,
      "learning_rate": 8.493723849372385e-06,
      "loss": 1.0442,
      "step": 144
    },
    {
      "epoch": 0.151673640167364,
      "grad_norm": 0.115234375,
      "learning_rate": 8.48326359832636e-06,
      "loss": 1.049,
      "step": 145
    },
    {
      "epoch": 0.15271966527196654,
      "grad_norm": 0.111328125,
      "learning_rate": 8.472803347280335e-06,
      "loss": 1.0247,
      "step": 146
    },
    {
      "epoch": 0.15376569037656904,
      "grad_norm": 0.10791015625,
      "learning_rate": 8.46234309623431e-06,
      "loss": 0.9802,
      "step": 147
    },
    {
      "epoch": 0.15481171548117154,
      "grad_norm": 0.11572265625,
      "learning_rate": 8.451882845188284e-06,
      "loss": 1.0421,
      "step": 148
    },
    {
      "epoch": 0.15585774058577406,
      "grad_norm": 0.10888671875,
      "learning_rate": 8.44142259414226e-06,
      "loss": 1.0267,
      "step": 149
    },
    {
      "epoch": 0.15690376569037656,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.430962343096235e-06,
      "loss": 0.9916,
      "step": 150
    },
    {
      "epoch": 0.1579497907949791,
      "grad_norm": 0.09521484375,
      "learning_rate": 8.42050209205021e-06,
      "loss": 0.9676,
      "step": 151
    },
    {
      "epoch": 0.1589958158995816,
      "grad_norm": 0.09912109375,
      "learning_rate": 8.410041841004185e-06,
      "loss": 0.9879,
      "step": 152
    },
    {
      "epoch": 0.1600418410041841,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.39958158995816e-06,
      "loss": 1.0094,
      "step": 153
    },
    {
      "epoch": 0.16108786610878661,
      "grad_norm": 0.1142578125,
      "learning_rate": 8.389121338912136e-06,
      "loss": 0.9737,
      "step": 154
    },
    {
      "epoch": 0.16213389121338911,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.378661087866109e-06,
      "loss": 1.002,
      "step": 155
    },
    {
      "epoch": 0.16317991631799164,
      "grad_norm": 0.1142578125,
      "learning_rate": 8.368200836820084e-06,
      "loss": 1.0412,
      "step": 156
    },
    {
      "epoch": 0.16422594142259414,
      "grad_norm": 0.11474609375,
      "learning_rate": 8.357740585774058e-06,
      "loss": 1.0494,
      "step": 157
    },
    {
      "epoch": 0.16527196652719664,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.347280334728035e-06,
      "loss": 0.9618,
      "step": 158
    },
    {
      "epoch": 0.16631799163179917,
      "grad_norm": 0.107421875,
      "learning_rate": 8.33682008368201e-06,
      "loss": 1.0424,
      "step": 159
    },
    {
      "epoch": 0.16736401673640167,
      "grad_norm": 0.107421875,
      "learning_rate": 8.326359832635984e-06,
      "loss": 0.9916,
      "step": 160
    },
    {
      "epoch": 0.1684100418410042,
      "grad_norm": 0.09228515625,
      "learning_rate": 8.315899581589959e-06,
      "loss": 0.9611,
      "step": 161
    },
    {
      "epoch": 0.1694560669456067,
      "grad_norm": 0.10595703125,
      "learning_rate": 8.305439330543934e-06,
      "loss": 1.0187,
      "step": 162
    },
    {
      "epoch": 0.1705020920502092,
      "grad_norm": 0.10693359375,
      "learning_rate": 8.294979079497908e-06,
      "loss": 1.0063,
      "step": 163
    },
    {
      "epoch": 0.17154811715481172,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.284518828451885e-06,
      "loss": 1.0615,
      "step": 164
    },
    {
      "epoch": 0.17259414225941422,
      "grad_norm": 0.126953125,
      "learning_rate": 8.274058577405858e-06,
      "loss": 1.0572,
      "step": 165
    },
    {
      "epoch": 0.17364016736401675,
      "grad_norm": 0.10009765625,
      "learning_rate": 8.263598326359832e-06,
      "loss": 0.9887,
      "step": 166
    },
    {
      "epoch": 0.17468619246861924,
      "grad_norm": 0.109375,
      "learning_rate": 8.253138075313809e-06,
      "loss": 0.983,
      "step": 167
    },
    {
      "epoch": 0.17573221757322174,
      "grad_norm": 0.1083984375,
      "learning_rate": 8.242677824267783e-06,
      "loss": 1.0023,
      "step": 168
    },
    {
      "epoch": 0.17677824267782427,
      "grad_norm": 0.109375,
      "learning_rate": 8.232217573221758e-06,
      "loss": 1.0193,
      "step": 169
    },
    {
      "epoch": 0.17782426778242677,
      "grad_norm": 0.119140625,
      "learning_rate": 8.221757322175733e-06,
      "loss": 1.0689,
      "step": 170
    },
    {
      "epoch": 0.1788702928870293,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.211297071129708e-06,
      "loss": 1.0083,
      "step": 171
    },
    {
      "epoch": 0.1799163179916318,
      "grad_norm": 0.1103515625,
      "learning_rate": 8.200836820083682e-06,
      "loss": 0.989,
      "step": 172
    },
    {
      "epoch": 0.1809623430962343,
      "grad_norm": 0.11328125,
      "learning_rate": 8.190376569037659e-06,
      "loss": 0.9852,
      "step": 173
    },
    {
      "epoch": 0.18200836820083682,
      "grad_norm": 0.09765625,
      "learning_rate": 8.179916317991633e-06,
      "loss": 0.9449,
      "step": 174
    },
    {
      "epoch": 0.18305439330543932,
      "grad_norm": 0.10595703125,
      "learning_rate": 8.169456066945606e-06,
      "loss": 0.9565,
      "step": 175
    },
    {
      "epoch": 0.18410041841004185,
      "grad_norm": 0.11474609375,
      "learning_rate": 8.158995815899581e-06,
      "loss": 1.0036,
      "step": 176
    },
    {
      "epoch": 0.18514644351464435,
      "grad_norm": 0.11474609375,
      "learning_rate": 8.148535564853557e-06,
      "loss": 1.0193,
      "step": 177
    },
    {
      "epoch": 0.18619246861924685,
      "grad_norm": 0.09814453125,
      "learning_rate": 8.138075313807532e-06,
      "loss": 0.9806,
      "step": 178
    },
    {
      "epoch": 0.18723849372384938,
      "grad_norm": 0.10693359375,
      "learning_rate": 8.127615062761507e-06,
      "loss": 0.9989,
      "step": 179
    },
    {
      "epoch": 0.18828451882845187,
      "grad_norm": 0.1376953125,
      "learning_rate": 8.117154811715482e-06,
      "loss": 0.9812,
      "step": 180
    },
    {
      "epoch": 0.1893305439330544,
      "grad_norm": 0.10546875,
      "learning_rate": 8.106694560669456e-06,
      "loss": 0.9767,
      "step": 181
    },
    {
      "epoch": 0.1903765690376569,
      "grad_norm": 0.1064453125,
      "learning_rate": 8.096234309623433e-06,
      "loss": 0.9743,
      "step": 182
    },
    {
      "epoch": 0.19142259414225943,
      "grad_norm": 0.1103515625,
      "learning_rate": 8.085774058577407e-06,
      "loss": 0.9414,
      "step": 183
    },
    {
      "epoch": 0.19246861924686193,
      "grad_norm": 0.10400390625,
      "learning_rate": 8.075313807531382e-06,
      "loss": 1.0257,
      "step": 184
    },
    {
      "epoch": 0.19351464435146443,
      "grad_norm": 0.11376953125,
      "learning_rate": 8.064853556485355e-06,
      "loss": 0.9335,
      "step": 185
    },
    {
      "epoch": 0.19456066945606695,
      "grad_norm": 0.1005859375,
      "learning_rate": 8.054393305439332e-06,
      "loss": 0.9864,
      "step": 186
    },
    {
      "epoch": 0.19560669456066945,
      "grad_norm": 0.103515625,
      "learning_rate": 8.043933054393306e-06,
      "loss": 0.961,
      "step": 187
    },
    {
      "epoch": 0.19665271966527198,
      "grad_norm": 0.1103515625,
      "learning_rate": 8.033472803347281e-06,
      "loss": 1.0023,
      "step": 188
    },
    {
      "epoch": 0.19769874476987448,
      "grad_norm": 0.1064453125,
      "learning_rate": 8.023012552301256e-06,
      "loss": 0.9762,
      "step": 189
    },
    {
      "epoch": 0.19874476987447698,
      "grad_norm": 0.1103515625,
      "learning_rate": 8.01255230125523e-06,
      "loss": 1.0018,
      "step": 190
    },
    {
      "epoch": 0.1997907949790795,
      "grad_norm": 0.09375,
      "learning_rate": 8.002092050209205e-06,
      "loss": 0.9536,
      "step": 191
    },
    {
      "epoch": 0.200836820083682,
      "grad_norm": 0.0947265625,
      "learning_rate": 7.991631799163181e-06,
      "loss": 0.9631,
      "step": 192
    },
    {
      "epoch": 0.20188284518828453,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.981171548117156e-06,
      "loss": 0.9963,
      "step": 193
    },
    {
      "epoch": 0.20292887029288703,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.97071129707113e-06,
      "loss": 0.9915,
      "step": 194
    },
    {
      "epoch": 0.20397489539748953,
      "grad_norm": 0.1220703125,
      "learning_rate": 7.960251046025106e-06,
      "loss": 0.9648,
      "step": 195
    },
    {
      "epoch": 0.20502092050209206,
      "grad_norm": 0.09765625,
      "learning_rate": 7.94979079497908e-06,
      "loss": 0.9624,
      "step": 196
    },
    {
      "epoch": 0.20606694560669456,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.939330543933055e-06,
      "loss": 1.0037,
      "step": 197
    },
    {
      "epoch": 0.20711297071129708,
      "grad_norm": 0.1005859375,
      "learning_rate": 7.92887029288703e-06,
      "loss": 0.9593,
      "step": 198
    },
    {
      "epoch": 0.20815899581589958,
      "grad_norm": 0.1162109375,
      "learning_rate": 7.918410041841004e-06,
      "loss": 0.9875,
      "step": 199
    },
    {
      "epoch": 0.20920502092050208,
      "grad_norm": 0.10546875,
      "learning_rate": 7.907949790794979e-06,
      "loss": 0.919,
      "step": 200
    },
    {
      "epoch": 0.2102510460251046,
      "grad_norm": 0.12890625,
      "learning_rate": 7.897489539748955e-06,
      "loss": 0.9507,
      "step": 201
    },
    {
      "epoch": 0.2112970711297071,
      "grad_norm": 0.12890625,
      "learning_rate": 7.88702928870293e-06,
      "loss": 1.0072,
      "step": 202
    },
    {
      "epoch": 0.21234309623430964,
      "grad_norm": 0.11865234375,
      "learning_rate": 7.876569037656905e-06,
      "loss": 0.9883,
      "step": 203
    },
    {
      "epoch": 0.21338912133891214,
      "grad_norm": 0.10009765625,
      "learning_rate": 7.86610878661088e-06,
      "loss": 0.9561,
      "step": 204
    },
    {
      "epoch": 0.21443514644351463,
      "grad_norm": 0.1142578125,
      "learning_rate": 7.855648535564854e-06,
      "loss": 0.9691,
      "step": 205
    },
    {
      "epoch": 0.21548117154811716,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.845188284518829e-06,
      "loss": 0.963,
      "step": 206
    },
    {
      "epoch": 0.21652719665271966,
      "grad_norm": 0.103515625,
      "learning_rate": 7.834728033472804e-06,
      "loss": 0.9599,
      "step": 207
    },
    {
      "epoch": 0.2175732217573222,
      "grad_norm": 0.0859375,
      "learning_rate": 7.824267782426778e-06,
      "loss": 0.8706,
      "step": 208
    },
    {
      "epoch": 0.2186192468619247,
      "grad_norm": 0.09912109375,
      "learning_rate": 7.813807531380753e-06,
      "loss": 0.954,
      "step": 209
    },
    {
      "epoch": 0.2196652719665272,
      "grad_norm": 0.11328125,
      "learning_rate": 7.80334728033473e-06,
      "loss": 1.0155,
      "step": 210
    },
    {
      "epoch": 0.22071129707112971,
      "grad_norm": 0.09912109375,
      "learning_rate": 7.792887029288704e-06,
      "loss": 0.9794,
      "step": 211
    },
    {
      "epoch": 0.2217573221757322,
      "grad_norm": 0.11376953125,
      "learning_rate": 7.782426778242679e-06,
      "loss": 0.9667,
      "step": 212
    },
    {
      "epoch": 0.22280334728033474,
      "grad_norm": 0.09375,
      "learning_rate": 7.771966527196654e-06,
      "loss": 0.8688,
      "step": 213
    },
    {
      "epoch": 0.22384937238493724,
      "grad_norm": 0.09765625,
      "learning_rate": 7.761506276150628e-06,
      "loss": 0.9605,
      "step": 214
    },
    {
      "epoch": 0.22489539748953974,
      "grad_norm": 0.0869140625,
      "learning_rate": 7.751046025104603e-06,
      "loss": 0.9244,
      "step": 215
    },
    {
      "epoch": 0.22594142259414227,
      "grad_norm": 0.1064453125,
      "learning_rate": 7.740585774058578e-06,
      "loss": 0.9182,
      "step": 216
    },
    {
      "epoch": 0.22698744769874477,
      "grad_norm": 0.12890625,
      "learning_rate": 7.730125523012552e-06,
      "loss": 0.9618,
      "step": 217
    },
    {
      "epoch": 0.2280334728033473,
      "grad_norm": 0.08203125,
      "learning_rate": 7.719665271966527e-06,
      "loss": 0.8844,
      "step": 218
    },
    {
      "epoch": 0.2290794979079498,
      "grad_norm": 0.111328125,
      "learning_rate": 7.709205020920504e-06,
      "loss": 0.9747,
      "step": 219
    },
    {
      "epoch": 0.2301255230125523,
      "grad_norm": 0.111328125,
      "learning_rate": 7.698744769874478e-06,
      "loss": 0.981,
      "step": 220
    },
    {
      "epoch": 0.23117154811715482,
      "grad_norm": 0.10595703125,
      "learning_rate": 7.688284518828453e-06,
      "loss": 0.9593,
      "step": 221
    },
    {
      "epoch": 0.23221757322175732,
      "grad_norm": 0.09228515625,
      "learning_rate": 7.677824267782428e-06,
      "loss": 0.9102,
      "step": 222
    },
    {
      "epoch": 0.23326359832635984,
      "grad_norm": 0.09765625,
      "learning_rate": 7.667364016736402e-06,
      "loss": 0.9207,
      "step": 223
    },
    {
      "epoch": 0.23430962343096234,
      "grad_norm": 0.08642578125,
      "learning_rate": 7.656903765690377e-06,
      "loss": 0.9065,
      "step": 224
    },
    {
      "epoch": 0.23535564853556484,
      "grad_norm": 0.1083984375,
      "learning_rate": 7.646443514644352e-06,
      "loss": 0.9559,
      "step": 225
    },
    {
      "epoch": 0.23640167364016737,
      "grad_norm": 0.107421875,
      "learning_rate": 7.635983263598326e-06,
      "loss": 0.963,
      "step": 226
    },
    {
      "epoch": 0.23744769874476987,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.625523012552302e-06,
      "loss": 0.9821,
      "step": 227
    },
    {
      "epoch": 0.2384937238493724,
      "grad_norm": 0.09814453125,
      "learning_rate": 7.615062761506277e-06,
      "loss": 0.9338,
      "step": 228
    },
    {
      "epoch": 0.2395397489539749,
      "grad_norm": 0.119140625,
      "learning_rate": 7.6046025104602514e-06,
      "loss": 1.0077,
      "step": 229
    },
    {
      "epoch": 0.2405857740585774,
      "grad_norm": 0.1015625,
      "learning_rate": 7.594142259414227e-06,
      "loss": 0.9009,
      "step": 230
    },
    {
      "epoch": 0.24163179916317992,
      "grad_norm": 0.10400390625,
      "learning_rate": 7.583682008368202e-06,
      "loss": 0.9378,
      "step": 231
    },
    {
      "epoch": 0.24267782426778242,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.573221757322176e-06,
      "loss": 0.928,
      "step": 232
    },
    {
      "epoch": 0.24372384937238495,
      "grad_norm": 0.11572265625,
      "learning_rate": 7.562761506276152e-06,
      "loss": 0.977,
      "step": 233
    },
    {
      "epoch": 0.24476987447698745,
      "grad_norm": 0.115234375,
      "learning_rate": 7.552301255230127e-06,
      "loss": 0.9679,
      "step": 234
    },
    {
      "epoch": 0.24581589958158995,
      "grad_norm": 0.10009765625,
      "learning_rate": 7.5418410041841005e-06,
      "loss": 0.9019,
      "step": 235
    },
    {
      "epoch": 0.24686192468619247,
      "grad_norm": 0.107421875,
      "learning_rate": 7.531380753138075e-06,
      "loss": 0.9495,
      "step": 236
    },
    {
      "epoch": 0.24790794979079497,
      "grad_norm": 0.08935546875,
      "learning_rate": 7.520920502092051e-06,
      "loss": 0.9171,
      "step": 237
    },
    {
      "epoch": 0.2489539748953975,
      "grad_norm": 0.10009765625,
      "learning_rate": 7.5104602510460255e-06,
      "loss": 0.9258,
      "step": 238
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.9288,
      "step": 239
    },
    {
      "epoch": 0.2510460251046025,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.489539748953976e-06,
      "loss": 0.9427,
      "step": 240
    },
    {
      "epoch": 0.252092050209205,
      "grad_norm": 0.109375,
      "learning_rate": 7.4790794979079504e-06,
      "loss": 0.939,
      "step": 241
    },
    {
      "epoch": 0.25313807531380755,
      "grad_norm": 0.1171875,
      "learning_rate": 7.468619246861926e-06,
      "loss": 0.9618,
      "step": 242
    },
    {
      "epoch": 0.25418410041841005,
      "grad_norm": 0.10546875,
      "learning_rate": 7.458158995815901e-06,
      "loss": 0.9305,
      "step": 243
    },
    {
      "epoch": 0.25523012552301255,
      "grad_norm": 0.09423828125,
      "learning_rate": 7.4476987447698746e-06,
      "loss": 0.9138,
      "step": 244
    },
    {
      "epoch": 0.25627615062761505,
      "grad_norm": 0.126953125,
      "learning_rate": 7.437238493723849e-06,
      "loss": 0.9676,
      "step": 245
    },
    {
      "epoch": 0.25732217573221755,
      "grad_norm": 0.095703125,
      "learning_rate": 7.426778242677825e-06,
      "loss": 0.9479,
      "step": 246
    },
    {
      "epoch": 0.2583682008368201,
      "grad_norm": 0.11279296875,
      "learning_rate": 7.4163179916317995e-06,
      "loss": 0.9357,
      "step": 247
    },
    {
      "epoch": 0.2594142259414226,
      "grad_norm": 0.09033203125,
      "learning_rate": 7.405857740585774e-06,
      "loss": 0.8945,
      "step": 248
    },
    {
      "epoch": 0.2604602510460251,
      "grad_norm": 0.1025390625,
      "learning_rate": 7.39539748953975e-06,
      "loss": 0.9294,
      "step": 249
    },
    {
      "epoch": 0.2615062761506276,
      "grad_norm": 0.0986328125,
      "learning_rate": 7.3849372384937245e-06,
      "loss": 0.9292,
      "step": 250
    },
    {
      "epoch": 0.2625523012552301,
      "grad_norm": 0.095703125,
      "learning_rate": 7.374476987447699e-06,
      "loss": 0.9701,
      "step": 251
    },
    {
      "epoch": 0.26359832635983266,
      "grad_norm": 0.1123046875,
      "learning_rate": 7.364016736401675e-06,
      "loss": 0.9327,
      "step": 252
    },
    {
      "epoch": 0.26464435146443516,
      "grad_norm": 0.09423828125,
      "learning_rate": 7.3535564853556494e-06,
      "loss": 0.9195,
      "step": 253
    },
    {
      "epoch": 0.26569037656903766,
      "grad_norm": 0.10986328125,
      "learning_rate": 7.343096234309623e-06,
      "loss": 0.9268,
      "step": 254
    },
    {
      "epoch": 0.26673640167364016,
      "grad_norm": 0.1220703125,
      "learning_rate": 7.332635983263599e-06,
      "loss": 1.0125,
      "step": 255
    },
    {
      "epoch": 0.26778242677824265,
      "grad_norm": 0.09912109375,
      "learning_rate": 7.3221757322175736e-06,
      "loss": 0.9131,
      "step": 256
    },
    {
      "epoch": 0.2688284518828452,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.311715481171548e-06,
      "loss": 0.9215,
      "step": 257
    },
    {
      "epoch": 0.2698744769874477,
      "grad_norm": 0.099609375,
      "learning_rate": 7.301255230125524e-06,
      "loss": 0.9177,
      "step": 258
    },
    {
      "epoch": 0.2709205020920502,
      "grad_norm": 0.0810546875,
      "learning_rate": 7.2907949790794985e-06,
      "loss": 0.8573,
      "step": 259
    },
    {
      "epoch": 0.2719665271966527,
      "grad_norm": 0.09521484375,
      "learning_rate": 7.280334728033473e-06,
      "loss": 0.9278,
      "step": 260
    },
    {
      "epoch": 0.2730125523012552,
      "grad_norm": 0.1103515625,
      "learning_rate": 7.269874476987449e-06,
      "loss": 0.9361,
      "step": 261
    },
    {
      "epoch": 0.27405857740585776,
      "grad_norm": 0.0927734375,
      "learning_rate": 7.2594142259414235e-06,
      "loss": 0.8595,
      "step": 262
    },
    {
      "epoch": 0.27510460251046026,
      "grad_norm": 0.1123046875,
      "learning_rate": 7.248953974895398e-06,
      "loss": 0.9602,
      "step": 263
    },
    {
      "epoch": 0.27615062761506276,
      "grad_norm": 0.107421875,
      "learning_rate": 7.238493723849372e-06,
      "loss": 0.9332,
      "step": 264
    },
    {
      "epoch": 0.27719665271966526,
      "grad_norm": 0.12158203125,
      "learning_rate": 7.228033472803348e-06,
      "loss": 0.943,
      "step": 265
    },
    {
      "epoch": 0.27824267782426776,
      "grad_norm": 0.09814453125,
      "learning_rate": 7.217573221757322e-06,
      "loss": 0.8968,
      "step": 266
    },
    {
      "epoch": 0.2792887029288703,
      "grad_norm": 0.11083984375,
      "learning_rate": 7.207112970711298e-06,
      "loss": 0.9311,
      "step": 267
    },
    {
      "epoch": 0.2803347280334728,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.1966527196652726e-06,
      "loss": 0.9304,
      "step": 268
    },
    {
      "epoch": 0.2813807531380753,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.186192468619247e-06,
      "loss": 0.8889,
      "step": 269
    },
    {
      "epoch": 0.2824267782426778,
      "grad_norm": 0.09765625,
      "learning_rate": 7.175732217573223e-06,
      "loss": 0.9103,
      "step": 270
    },
    {
      "epoch": 0.2834728033472803,
      "grad_norm": 0.09375,
      "learning_rate": 7.1652719665271975e-06,
      "loss": 0.9061,
      "step": 271
    },
    {
      "epoch": 0.28451882845188287,
      "grad_norm": 0.1123046875,
      "learning_rate": 7.154811715481172e-06,
      "loss": 0.9643,
      "step": 272
    },
    {
      "epoch": 0.28556485355648537,
      "grad_norm": 0.087890625,
      "learning_rate": 7.144351464435148e-06,
      "loss": 0.8873,
      "step": 273
    },
    {
      "epoch": 0.28661087866108786,
      "grad_norm": 0.09130859375,
      "learning_rate": 7.133891213389122e-06,
      "loss": 0.9066,
      "step": 274
    },
    {
      "epoch": 0.28765690376569036,
      "grad_norm": 0.10986328125,
      "learning_rate": 7.123430962343096e-06,
      "loss": 0.8907,
      "step": 275
    },
    {
      "epoch": 0.28870292887029286,
      "grad_norm": 0.13671875,
      "learning_rate": 7.112970711297071e-06,
      "loss": 1.017,
      "step": 276
    },
    {
      "epoch": 0.2897489539748954,
      "grad_norm": 0.11279296875,
      "learning_rate": 7.102510460251047e-06,
      "loss": 0.9662,
      "step": 277
    },
    {
      "epoch": 0.2907949790794979,
      "grad_norm": 0.10595703125,
      "learning_rate": 7.092050209205021e-06,
      "loss": 0.9488,
      "step": 278
    },
    {
      "epoch": 0.2918410041841004,
      "grad_norm": 0.12353515625,
      "learning_rate": 7.081589958158996e-06,
      "loss": 0.9343,
      "step": 279
    },
    {
      "epoch": 0.2928870292887029,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.0711297071129716e-06,
      "loss": 0.8936,
      "step": 280
    },
    {
      "epoch": 0.2939330543933054,
      "grad_norm": 0.08251953125,
      "learning_rate": 7.060669456066946e-06,
      "loss": 0.8293,
      "step": 281
    },
    {
      "epoch": 0.29497907949790797,
      "grad_norm": 0.1015625,
      "learning_rate": 7.050209205020922e-06,
      "loss": 0.8911,
      "step": 282
    },
    {
      "epoch": 0.29602510460251047,
      "grad_norm": 0.1171875,
      "learning_rate": 7.0397489539748965e-06,
      "loss": 0.9673,
      "step": 283
    },
    {
      "epoch": 0.29707112970711297,
      "grad_norm": 0.095703125,
      "learning_rate": 7.02928870292887e-06,
      "loss": 0.8736,
      "step": 284
    },
    {
      "epoch": 0.29811715481171547,
      "grad_norm": 0.10546875,
      "learning_rate": 7.018828451882845e-06,
      "loss": 0.9277,
      "step": 285
    },
    {
      "epoch": 0.29916317991631797,
      "grad_norm": 0.1201171875,
      "learning_rate": 7.008368200836821e-06,
      "loss": 0.9491,
      "step": 286
    },
    {
      "epoch": 0.3002092050209205,
      "grad_norm": 0.103515625,
      "learning_rate": 6.997907949790795e-06,
      "loss": 0.9451,
      "step": 287
    },
    {
      "epoch": 0.301255230125523,
      "grad_norm": 0.10791015625,
      "learning_rate": 6.98744769874477e-06,
      "loss": 0.9345,
      "step": 288
    },
    {
      "epoch": 0.3023012552301255,
      "grad_norm": 0.09326171875,
      "learning_rate": 6.976987447698746e-06,
      "loss": 0.8919,
      "step": 289
    },
    {
      "epoch": 0.303347280334728,
      "grad_norm": 0.095703125,
      "learning_rate": 6.96652719665272e-06,
      "loss": 0.8962,
      "step": 290
    },
    {
      "epoch": 0.3043933054393305,
      "grad_norm": 0.111328125,
      "learning_rate": 6.956066945606695e-06,
      "loss": 0.8902,
      "step": 291
    },
    {
      "epoch": 0.3054393305439331,
      "grad_norm": 0.09765625,
      "learning_rate": 6.9456066945606706e-06,
      "loss": 0.8771,
      "step": 292
    },
    {
      "epoch": 0.3064853556485356,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.935146443514645e-06,
      "loss": 0.9338,
      "step": 293
    },
    {
      "epoch": 0.3075313807531381,
      "grad_norm": 0.080078125,
      "learning_rate": 6.924686192468619e-06,
      "loss": 0.8572,
      "step": 294
    },
    {
      "epoch": 0.30857740585774057,
      "grad_norm": 0.109375,
      "learning_rate": 6.914225941422595e-06,
      "loss": 0.9397,
      "step": 295
    },
    {
      "epoch": 0.30962343096234307,
      "grad_norm": 0.08935546875,
      "learning_rate": 6.903765690376569e-06,
      "loss": 0.8715,
      "step": 296
    },
    {
      "epoch": 0.3106694560669456,
      "grad_norm": 0.10205078125,
      "learning_rate": 6.893305439330544e-06,
      "loss": 0.9247,
      "step": 297
    },
    {
      "epoch": 0.3117154811715481,
      "grad_norm": 0.1357421875,
      "learning_rate": 6.88284518828452e-06,
      "loss": 0.9162,
      "step": 298
    },
    {
      "epoch": 0.3127615062761506,
      "grad_norm": 0.1015625,
      "learning_rate": 6.872384937238494e-06,
      "loss": 0.9123,
      "step": 299
    },
    {
      "epoch": 0.3138075313807531,
      "grad_norm": 0.11474609375,
      "learning_rate": 6.861924686192469e-06,
      "loss": 0.8794,
      "step": 300
    },
    {
      "epoch": 0.3148535564853556,
      "grad_norm": 0.10986328125,
      "learning_rate": 6.851464435146445e-06,
      "loss": 0.911,
      "step": 301
    },
    {
      "epoch": 0.3158995815899582,
      "grad_norm": 0.09814453125,
      "learning_rate": 6.841004184100419e-06,
      "loss": 0.8873,
      "step": 302
    },
    {
      "epoch": 0.3169456066945607,
      "grad_norm": 0.259765625,
      "learning_rate": 6.830543933054394e-06,
      "loss": 0.9175,
      "step": 303
    },
    {
      "epoch": 0.3179916317991632,
      "grad_norm": 0.11083984375,
      "learning_rate": 6.820083682008368e-06,
      "loss": 0.9741,
      "step": 304
    },
    {
      "epoch": 0.3190376569037657,
      "grad_norm": 0.10595703125,
      "learning_rate": 6.809623430962343e-06,
      "loss": 0.9271,
      "step": 305
    },
    {
      "epoch": 0.3200836820083682,
      "grad_norm": 0.11572265625,
      "learning_rate": 6.799163179916318e-06,
      "loss": 0.914,
      "step": 306
    },
    {
      "epoch": 0.32112970711297073,
      "grad_norm": 0.09765625,
      "learning_rate": 6.788702928870293e-06,
      "loss": 0.9197,
      "step": 307
    },
    {
      "epoch": 0.32217573221757323,
      "grad_norm": 0.1103515625,
      "learning_rate": 6.778242677824268e-06,
      "loss": 0.9172,
      "step": 308
    },
    {
      "epoch": 0.32322175732217573,
      "grad_norm": 0.1279296875,
      "learning_rate": 6.767782426778243e-06,
      "loss": 1.0448,
      "step": 309
    },
    {
      "epoch": 0.32426778242677823,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.757322175732219e-06,
      "loss": 0.8812,
      "step": 310
    },
    {
      "epoch": 0.3253138075313807,
      "grad_norm": 0.111328125,
      "learning_rate": 6.746861924686193e-06,
      "loss": 0.8771,
      "step": 311
    },
    {
      "epoch": 0.3263598326359833,
      "grad_norm": 0.10302734375,
      "learning_rate": 6.736401673640168e-06,
      "loss": 0.8931,
      "step": 312
    },
    {
      "epoch": 0.3274058577405858,
      "grad_norm": 0.099609375,
      "learning_rate": 6.725941422594144e-06,
      "loss": 0.8973,
      "step": 313
    },
    {
      "epoch": 0.3284518828451883,
      "grad_norm": 0.1279296875,
      "learning_rate": 6.7154811715481175e-06,
      "loss": 0.9766,
      "step": 314
    },
    {
      "epoch": 0.3294979079497908,
      "grad_norm": 0.0966796875,
      "learning_rate": 6.705020920502092e-06,
      "loss": 0.8771,
      "step": 315
    },
    {
      "epoch": 0.3305439330543933,
      "grad_norm": 0.107421875,
      "learning_rate": 6.694560669456067e-06,
      "loss": 0.9193,
      "step": 316
    },
    {
      "epoch": 0.33158995815899583,
      "grad_norm": 0.1005859375,
      "learning_rate": 6.684100418410042e-06,
      "loss": 0.9222,
      "step": 317
    },
    {
      "epoch": 0.33263598326359833,
      "grad_norm": 0.09130859375,
      "learning_rate": 6.673640167364017e-06,
      "loss": 0.8725,
      "step": 318
    },
    {
      "epoch": 0.33368200836820083,
      "grad_norm": 0.09375,
      "learning_rate": 6.663179916317992e-06,
      "loss": 0.8838,
      "step": 319
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 0.09130859375,
      "learning_rate": 6.652719665271967e-06,
      "loss": 0.907,
      "step": 320
    },
    {
      "epoch": 0.33577405857740583,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.642259414225942e-06,
      "loss": 0.8938,
      "step": 321
    },
    {
      "epoch": 0.3368200836820084,
      "grad_norm": 0.125,
      "learning_rate": 6.631799163179918e-06,
      "loss": 0.9461,
      "step": 322
    },
    {
      "epoch": 0.3378661087866109,
      "grad_norm": 0.095703125,
      "learning_rate": 6.621338912133892e-06,
      "loss": 0.8944,
      "step": 323
    },
    {
      "epoch": 0.3389121338912134,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.610878661087866e-06,
      "loss": 0.878,
      "step": 324
    },
    {
      "epoch": 0.3399581589958159,
      "grad_norm": 0.11865234375,
      "learning_rate": 6.600418410041841e-06,
      "loss": 0.8597,
      "step": 325
    },
    {
      "epoch": 0.3410041841004184,
      "grad_norm": 0.08740234375,
      "learning_rate": 6.5899581589958165e-06,
      "loss": 0.8925,
      "step": 326
    },
    {
      "epoch": 0.34205020920502094,
      "grad_norm": 0.09326171875,
      "learning_rate": 6.579497907949791e-06,
      "loss": 0.891,
      "step": 327
    },
    {
      "epoch": 0.34309623430962344,
      "grad_norm": 0.126953125,
      "learning_rate": 6.569037656903766e-06,
      "loss": 0.9515,
      "step": 328
    },
    {
      "epoch": 0.34414225941422594,
      "grad_norm": 0.10546875,
      "learning_rate": 6.558577405857741e-06,
      "loss": 0.9099,
      "step": 329
    },
    {
      "epoch": 0.34518828451882844,
      "grad_norm": 0.09228515625,
      "learning_rate": 6.548117154811716e-06,
      "loss": 0.8978,
      "step": 330
    },
    {
      "epoch": 0.34623430962343094,
      "grad_norm": 0.09765625,
      "learning_rate": 6.537656903765691e-06,
      "loss": 0.8619,
      "step": 331
    },
    {
      "epoch": 0.3472803347280335,
      "grad_norm": 0.1162109375,
      "learning_rate": 6.527196652719666e-06,
      "loss": 0.946,
      "step": 332
    },
    {
      "epoch": 0.348326359832636,
      "grad_norm": 0.091796875,
      "learning_rate": 6.516736401673641e-06,
      "loss": 0.817,
      "step": 333
    },
    {
      "epoch": 0.3493723849372385,
      "grad_norm": 0.08154296875,
      "learning_rate": 6.506276150627615e-06,
      "loss": 0.8626,
      "step": 334
    },
    {
      "epoch": 0.350418410041841,
      "grad_norm": 0.0869140625,
      "learning_rate": 6.4958158995815905e-06,
      "loss": 0.8684,
      "step": 335
    },
    {
      "epoch": 0.3514644351464435,
      "grad_norm": 0.09033203125,
      "learning_rate": 6.485355648535565e-06,
      "loss": 0.8923,
      "step": 336
    },
    {
      "epoch": 0.35251046025104604,
      "grad_norm": 0.1220703125,
      "learning_rate": 6.47489539748954e-06,
      "loss": 0.9581,
      "step": 337
    },
    {
      "epoch": 0.35355648535564854,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.4644351464435155e-06,
      "loss": 0.8977,
      "step": 338
    },
    {
      "epoch": 0.35460251046025104,
      "grad_norm": 0.09814453125,
      "learning_rate": 6.45397489539749e-06,
      "loss": 0.8681,
      "step": 339
    },
    {
      "epoch": 0.35564853556485354,
      "grad_norm": 0.09765625,
      "learning_rate": 6.443514644351465e-06,
      "loss": 0.8527,
      "step": 340
    },
    {
      "epoch": 0.35669456066945604,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.43305439330544e-06,
      "loss": 0.8743,
      "step": 341
    },
    {
      "epoch": 0.3577405857740586,
      "grad_norm": 0.107421875,
      "learning_rate": 6.422594142259415e-06,
      "loss": 0.9322,
      "step": 342
    },
    {
      "epoch": 0.3587866108786611,
      "grad_norm": 0.09228515625,
      "learning_rate": 6.41213389121339e-06,
      "loss": 0.8788,
      "step": 343
    },
    {
      "epoch": 0.3598326359832636,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.401673640167364e-06,
      "loss": 0.8711,
      "step": 344
    },
    {
      "epoch": 0.3608786610878661,
      "grad_norm": 0.1103515625,
      "learning_rate": 6.391213389121339e-06,
      "loss": 0.8806,
      "step": 345
    },
    {
      "epoch": 0.3619246861924686,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.380753138075314e-06,
      "loss": 0.8456,
      "step": 346
    },
    {
      "epoch": 0.36297071129707115,
      "grad_norm": 0.091796875,
      "learning_rate": 6.370292887029289e-06,
      "loss": 0.8777,
      "step": 347
    },
    {
      "epoch": 0.36401673640167365,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.359832635983264e-06,
      "loss": 0.9072,
      "step": 348
    },
    {
      "epoch": 0.36506276150627615,
      "grad_norm": 0.109375,
      "learning_rate": 6.349372384937239e-06,
      "loss": 0.9098,
      "step": 349
    },
    {
      "epoch": 0.36610878661087864,
      "grad_norm": 0.0859375,
      "learning_rate": 6.3389121338912145e-06,
      "loss": 0.845,
      "step": 350
    },
    {
      "epoch": 0.36715481171548114,
      "grad_norm": 0.09326171875,
      "learning_rate": 6.328451882845189e-06,
      "loss": 0.8812,
      "step": 351
    },
    {
      "epoch": 0.3682008368200837,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.317991631799164e-06,
      "loss": 0.8963,
      "step": 352
    },
    {
      "epoch": 0.3692468619246862,
      "grad_norm": 0.0966796875,
      "learning_rate": 6.307531380753139e-06,
      "loss": 0.9016,
      "step": 353
    },
    {
      "epoch": 0.3702928870292887,
      "grad_norm": 0.09326171875,
      "learning_rate": 6.297071129707113e-06,
      "loss": 0.8604,
      "step": 354
    },
    {
      "epoch": 0.3713389121338912,
      "grad_norm": 0.1005859375,
      "learning_rate": 6.286610878661088e-06,
      "loss": 0.9167,
      "step": 355
    },
    {
      "epoch": 0.3723849372384937,
      "grad_norm": 0.10498046875,
      "learning_rate": 6.276150627615063e-06,
      "loss": 0.8783,
      "step": 356
    },
    {
      "epoch": 0.37343096234309625,
      "grad_norm": 0.08056640625,
      "learning_rate": 6.265690376569038e-06,
      "loss": 0.8728,
      "step": 357
    },
    {
      "epoch": 0.37447698744769875,
      "grad_norm": 0.0966796875,
      "learning_rate": 6.255230125523013e-06,
      "loss": 0.8673,
      "step": 358
    },
    {
      "epoch": 0.37552301255230125,
      "grad_norm": 0.1044921875,
      "learning_rate": 6.244769874476988e-06,
      "loss": 0.9517,
      "step": 359
    },
    {
      "epoch": 0.37656903765690375,
      "grad_norm": 0.099609375,
      "learning_rate": 6.234309623430963e-06,
      "loss": 0.9099,
      "step": 360
    },
    {
      "epoch": 0.3776150627615063,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.223849372384938e-06,
      "loss": 0.893,
      "step": 361
    },
    {
      "epoch": 0.3786610878661088,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.213389121338913e-06,
      "loss": 0.8721,
      "step": 362
    },
    {
      "epoch": 0.3797071129707113,
      "grad_norm": 0.09765625,
      "learning_rate": 6.202928870292888e-06,
      "loss": 0.9031,
      "step": 363
    },
    {
      "epoch": 0.3807531380753138,
      "grad_norm": 0.091796875,
      "learning_rate": 6.192468619246862e-06,
      "loss": 0.9104,
      "step": 364
    },
    {
      "epoch": 0.3817991631799163,
      "grad_norm": 0.10986328125,
      "learning_rate": 6.182008368200837e-06,
      "loss": 0.8794,
      "step": 365
    },
    {
      "epoch": 0.38284518828451886,
      "grad_norm": 0.109375,
      "learning_rate": 6.171548117154812e-06,
      "loss": 0.9106,
      "step": 366
    },
    {
      "epoch": 0.38389121338912136,
      "grad_norm": 0.09228515625,
      "learning_rate": 6.161087866108787e-06,
      "loss": 0.8426,
      "step": 367
    },
    {
      "epoch": 0.38493723849372385,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.150627615062762e-06,
      "loss": 0.9084,
      "step": 368
    },
    {
      "epoch": 0.38598326359832635,
      "grad_norm": 0.10791015625,
      "learning_rate": 6.140167364016737e-06,
      "loss": 0.9031,
      "step": 369
    },
    {
      "epoch": 0.38702928870292885,
      "grad_norm": 0.0908203125,
      "learning_rate": 6.129707112970712e-06,
      "loss": 0.8625,
      "step": 370
    },
    {
      "epoch": 0.3880753138075314,
      "grad_norm": 0.15234375,
      "learning_rate": 6.119246861924687e-06,
      "loss": 1.0223,
      "step": 371
    },
    {
      "epoch": 0.3891213389121339,
      "grad_norm": 0.09228515625,
      "learning_rate": 6.108786610878662e-06,
      "loss": 0.8818,
      "step": 372
    },
    {
      "epoch": 0.3901673640167364,
      "grad_norm": 0.107421875,
      "learning_rate": 6.098326359832637e-06,
      "loss": 0.9229,
      "step": 373
    },
    {
      "epoch": 0.3912133891213389,
      "grad_norm": 0.0859375,
      "learning_rate": 6.087866108786611e-06,
      "loss": 0.8092,
      "step": 374
    },
    {
      "epoch": 0.3922594142259414,
      "grad_norm": 0.1181640625,
      "learning_rate": 6.0774058577405855e-06,
      "loss": 0.9239,
      "step": 375
    },
    {
      "epoch": 0.39330543933054396,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.066945606694561e-06,
      "loss": 0.8912,
      "step": 376
    },
    {
      "epoch": 0.39435146443514646,
      "grad_norm": 0.1005859375,
      "learning_rate": 6.056485355648536e-06,
      "loss": 0.8451,
      "step": 377
    },
    {
      "epoch": 0.39539748953974896,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.046025104602511e-06,
      "loss": 0.8595,
      "step": 378
    },
    {
      "epoch": 0.39644351464435146,
      "grad_norm": 0.0947265625,
      "learning_rate": 6.035564853556486e-06,
      "loss": 0.8707,
      "step": 379
    },
    {
      "epoch": 0.39748953974895396,
      "grad_norm": 0.0908203125,
      "learning_rate": 6.025104602510461e-06,
      "loss": 0.8899,
      "step": 380
    },
    {
      "epoch": 0.3985355648535565,
      "grad_norm": 0.10595703125,
      "learning_rate": 6.014644351464436e-06,
      "loss": 0.9368,
      "step": 381
    },
    {
      "epoch": 0.399581589958159,
      "grad_norm": 0.09912109375,
      "learning_rate": 6.004184100418411e-06,
      "loss": 0.8832,
      "step": 382
    },
    {
      "epoch": 0.4006276150627615,
      "grad_norm": 0.103515625,
      "learning_rate": 5.993723849372386e-06,
      "loss": 0.8722,
      "step": 383
    },
    {
      "epoch": 0.401673640167364,
      "grad_norm": 0.10107421875,
      "learning_rate": 5.9832635983263595e-06,
      "loss": 0.894,
      "step": 384
    },
    {
      "epoch": 0.4027196652719665,
      "grad_norm": 0.1044921875,
      "learning_rate": 5.972803347280335e-06,
      "loss": 0.9334,
      "step": 385
    },
    {
      "epoch": 0.40376569037656906,
      "grad_norm": 0.10009765625,
      "learning_rate": 5.96234309623431e-06,
      "loss": 0.9141,
      "step": 386
    },
    {
      "epoch": 0.40481171548117156,
      "grad_norm": 0.0830078125,
      "learning_rate": 5.9518828451882845e-06,
      "loss": 0.8375,
      "step": 387
    },
    {
      "epoch": 0.40585774058577406,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.94142259414226e-06,
      "loss": 0.8767,
      "step": 388
    },
    {
      "epoch": 0.40690376569037656,
      "grad_norm": 0.0888671875,
      "learning_rate": 5.930962343096235e-06,
      "loss": 0.8585,
      "step": 389
    },
    {
      "epoch": 0.40794979079497906,
      "grad_norm": 0.09228515625,
      "learning_rate": 5.92050209205021e-06,
      "loss": 0.8905,
      "step": 390
    },
    {
      "epoch": 0.4089958158995816,
      "grad_norm": 0.10986328125,
      "learning_rate": 5.910041841004185e-06,
      "loss": 0.9263,
      "step": 391
    },
    {
      "epoch": 0.4100418410041841,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.89958158995816e-06,
      "loss": 0.8629,
      "step": 392
    },
    {
      "epoch": 0.4110878661087866,
      "grad_norm": 0.1025390625,
      "learning_rate": 5.889121338912135e-06,
      "loss": 0.8862,
      "step": 393
    },
    {
      "epoch": 0.4121338912133891,
      "grad_norm": 0.11669921875,
      "learning_rate": 5.878661087866109e-06,
      "loss": 0.8961,
      "step": 394
    },
    {
      "epoch": 0.4131799163179916,
      "grad_norm": 0.087890625,
      "learning_rate": 5.868200836820084e-06,
      "loss": 0.9038,
      "step": 395
    },
    {
      "epoch": 0.41422594142259417,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.8577405857740585e-06,
      "loss": 0.8358,
      "step": 396
    },
    {
      "epoch": 0.41527196652719667,
      "grad_norm": 0.11279296875,
      "learning_rate": 5.847280334728034e-06,
      "loss": 0.9427,
      "step": 397
    },
    {
      "epoch": 0.41631799163179917,
      "grad_norm": 0.09130859375,
      "learning_rate": 5.836820083682009e-06,
      "loss": 0.8776,
      "step": 398
    },
    {
      "epoch": 0.41736401673640167,
      "grad_norm": 0.11083984375,
      "learning_rate": 5.8263598326359835e-06,
      "loss": 0.923,
      "step": 399
    },
    {
      "epoch": 0.41841004184100417,
      "grad_norm": 0.126953125,
      "learning_rate": 5.815899581589959e-06,
      "loss": 0.96,
      "step": 400
    },
    {
      "epoch": 0.4194560669456067,
      "grad_norm": 0.1025390625,
      "learning_rate": 5.805439330543934e-06,
      "loss": 0.8968,
      "step": 401
    },
    {
      "epoch": 0.4205020920502092,
      "grad_norm": 0.0966796875,
      "learning_rate": 5.7949790794979084e-06,
      "loss": 0.9069,
      "step": 402
    },
    {
      "epoch": 0.4215481171548117,
      "grad_norm": 0.11181640625,
      "learning_rate": 5.784518828451884e-06,
      "loss": 0.8935,
      "step": 403
    },
    {
      "epoch": 0.4225941422594142,
      "grad_norm": 0.10400390625,
      "learning_rate": 5.774058577405858e-06,
      "loss": 0.8852,
      "step": 404
    },
    {
      "epoch": 0.4236401673640167,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.7635983263598325e-06,
      "loss": 0.8771,
      "step": 405
    },
    {
      "epoch": 0.4246861924686193,
      "grad_norm": 0.09716796875,
      "learning_rate": 5.753138075313808e-06,
      "loss": 0.904,
      "step": 406
    },
    {
      "epoch": 0.42573221757322177,
      "grad_norm": 0.1025390625,
      "learning_rate": 5.742677824267783e-06,
      "loss": 0.9387,
      "step": 407
    },
    {
      "epoch": 0.42677824267782427,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.7322175732217575e-06,
      "loss": 0.9,
      "step": 408
    },
    {
      "epoch": 0.42782426778242677,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.721757322175733e-06,
      "loss": 0.913,
      "step": 409
    },
    {
      "epoch": 0.42887029288702927,
      "grad_norm": 0.08837890625,
      "learning_rate": 5.711297071129708e-06,
      "loss": 0.8625,
      "step": 410
    },
    {
      "epoch": 0.4299163179916318,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.7008368200836825e-06,
      "loss": 0.9078,
      "step": 411
    },
    {
      "epoch": 0.4309623430962343,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.690376569037658e-06,
      "loss": 0.8397,
      "step": 412
    },
    {
      "epoch": 0.4320083682008368,
      "grad_norm": 0.0888671875,
      "learning_rate": 5.679916317991633e-06,
      "loss": 0.8702,
      "step": 413
    },
    {
      "epoch": 0.4330543933054393,
      "grad_norm": 0.10498046875,
      "learning_rate": 5.669456066945607e-06,
      "loss": 0.8992,
      "step": 414
    },
    {
      "epoch": 0.4341004184100418,
      "grad_norm": 0.109375,
      "learning_rate": 5.658995815899581e-06,
      "loss": 0.9321,
      "step": 415
    },
    {
      "epoch": 0.4351464435146444,
      "grad_norm": 0.09130859375,
      "learning_rate": 5.648535564853557e-06,
      "loss": 0.8505,
      "step": 416
    },
    {
      "epoch": 0.4361924686192469,
      "grad_norm": 0.09814453125,
      "learning_rate": 5.6380753138075315e-06,
      "loss": 0.9053,
      "step": 417
    },
    {
      "epoch": 0.4372384937238494,
      "grad_norm": 0.115234375,
      "learning_rate": 5.627615062761507e-06,
      "loss": 0.9518,
      "step": 418
    },
    {
      "epoch": 0.4382845188284519,
      "grad_norm": 0.08642578125,
      "learning_rate": 5.617154811715482e-06,
      "loss": 0.8712,
      "step": 419
    },
    {
      "epoch": 0.4393305439330544,
      "grad_norm": 0.08154296875,
      "learning_rate": 5.6066945606694565e-06,
      "loss": 0.8438,
      "step": 420
    },
    {
      "epoch": 0.44037656903765693,
      "grad_norm": 0.10546875,
      "learning_rate": 5.596234309623432e-06,
      "loss": 0.8503,
      "step": 421
    },
    {
      "epoch": 0.44142259414225943,
      "grad_norm": 0.095703125,
      "learning_rate": 5.585774058577407e-06,
      "loss": 0.8564,
      "step": 422
    },
    {
      "epoch": 0.4424686192468619,
      "grad_norm": 0.0908203125,
      "learning_rate": 5.5753138075313815e-06,
      "loss": 0.8633,
      "step": 423
    },
    {
      "epoch": 0.4435146443514644,
      "grad_norm": 0.11279296875,
      "learning_rate": 5.564853556485355e-06,
      "loss": 0.8951,
      "step": 424
    },
    {
      "epoch": 0.4445606694560669,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.554393305439331e-06,
      "loss": 0.8833,
      "step": 425
    },
    {
      "epoch": 0.4456066945606695,
      "grad_norm": 0.10107421875,
      "learning_rate": 5.543933054393306e-06,
      "loss": 0.866,
      "step": 426
    },
    {
      "epoch": 0.446652719665272,
      "grad_norm": 0.09375,
      "learning_rate": 5.53347280334728e-06,
      "loss": 0.8752,
      "step": 427
    },
    {
      "epoch": 0.4476987447698745,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.523012552301256e-06,
      "loss": 0.8856,
      "step": 428
    },
    {
      "epoch": 0.448744769874477,
      "grad_norm": 0.09619140625,
      "learning_rate": 5.5125523012552305e-06,
      "loss": 0.8931,
      "step": 429
    },
    {
      "epoch": 0.4497907949790795,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.502092050209205e-06,
      "loss": 0.8161,
      "step": 430
    },
    {
      "epoch": 0.45083682008368203,
      "grad_norm": 0.08740234375,
      "learning_rate": 5.491631799163181e-06,
      "loss": 0.8387,
      "step": 431
    },
    {
      "epoch": 0.45188284518828453,
      "grad_norm": 0.087890625,
      "learning_rate": 5.4811715481171555e-06,
      "loss": 0.831,
      "step": 432
    },
    {
      "epoch": 0.45292887029288703,
      "grad_norm": 0.138671875,
      "learning_rate": 5.470711297071131e-06,
      "loss": 0.8379,
      "step": 433
    },
    {
      "epoch": 0.45397489539748953,
      "grad_norm": 0.1083984375,
      "learning_rate": 5.460251046025105e-06,
      "loss": 0.8598,
      "step": 434
    },
    {
      "epoch": 0.45502092050209203,
      "grad_norm": 0.0888671875,
      "learning_rate": 5.44979079497908e-06,
      "loss": 0.8678,
      "step": 435
    },
    {
      "epoch": 0.4560669456066946,
      "grad_norm": 0.091796875,
      "learning_rate": 5.439330543933054e-06,
      "loss": 0.8765,
      "step": 436
    },
    {
      "epoch": 0.4571129707112971,
      "grad_norm": 0.09228515625,
      "learning_rate": 5.42887029288703e-06,
      "loss": 0.8965,
      "step": 437
    },
    {
      "epoch": 0.4581589958158996,
      "grad_norm": 0.107421875,
      "learning_rate": 5.418410041841005e-06,
      "loss": 0.9177,
      "step": 438
    },
    {
      "epoch": 0.4592050209205021,
      "grad_norm": 0.10595703125,
      "learning_rate": 5.407949790794979e-06,
      "loss": 0.8782,
      "step": 439
    },
    {
      "epoch": 0.4602510460251046,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.397489539748955e-06,
      "loss": 0.846,
      "step": 440
    },
    {
      "epoch": 0.46129707112970714,
      "grad_norm": 2.046875,
      "learning_rate": 5.3870292887029296e-06,
      "loss": 0.8594,
      "step": 441
    },
    {
      "epoch": 0.46234309623430964,
      "grad_norm": 0.11865234375,
      "learning_rate": 5.376569037656904e-06,
      "loss": 0.8708,
      "step": 442
    },
    {
      "epoch": 0.46338912133891214,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.36610878661088e-06,
      "loss": 0.8774,
      "step": 443
    },
    {
      "epoch": 0.46443514644351463,
      "grad_norm": 0.103515625,
      "learning_rate": 5.355648535564854e-06,
      "loss": 0.8599,
      "step": 444
    },
    {
      "epoch": 0.46548117154811713,
      "grad_norm": 0.1025390625,
      "learning_rate": 5.345188284518828e-06,
      "loss": 0.8744,
      "step": 445
    },
    {
      "epoch": 0.4665271966527197,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.334728033472804e-06,
      "loss": 0.9502,
      "step": 446
    },
    {
      "epoch": 0.4675732217573222,
      "grad_norm": 0.08935546875,
      "learning_rate": 5.324267782426779e-06,
      "loss": 0.8594,
      "step": 447
    },
    {
      "epoch": 0.4686192468619247,
      "grad_norm": 0.083984375,
      "learning_rate": 5.313807531380753e-06,
      "loss": 0.825,
      "step": 448
    },
    {
      "epoch": 0.4696652719665272,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.303347280334729e-06,
      "loss": 0.9047,
      "step": 449
    },
    {
      "epoch": 0.4707112970711297,
      "grad_norm": 0.099609375,
      "learning_rate": 5.292887029288704e-06,
      "loss": 0.8469,
      "step": 450
    },
    {
      "epoch": 0.47175732217573224,
      "grad_norm": 0.08984375,
      "learning_rate": 5.282426778242678e-06,
      "loss": 0.8386,
      "step": 451
    },
    {
      "epoch": 0.47280334728033474,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.271966527196654e-06,
      "loss": 0.8811,
      "step": 452
    },
    {
      "epoch": 0.47384937238493724,
      "grad_norm": 0.09423828125,
      "learning_rate": 5.2615062761506286e-06,
      "loss": 0.8811,
      "step": 453
    },
    {
      "epoch": 0.47489539748953974,
      "grad_norm": 0.08642578125,
      "learning_rate": 5.251046025104602e-06,
      "loss": 0.8556,
      "step": 454
    },
    {
      "epoch": 0.47594142259414224,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.240585774058577e-06,
      "loss": 0.8514,
      "step": 455
    },
    {
      "epoch": 0.4769874476987448,
      "grad_norm": 0.0888671875,
      "learning_rate": 5.230125523012553e-06,
      "loss": 0.8779,
      "step": 456
    },
    {
      "epoch": 0.4780334728033473,
      "grad_norm": 0.08935546875,
      "learning_rate": 5.219665271966527e-06,
      "loss": 0.8411,
      "step": 457
    },
    {
      "epoch": 0.4790794979079498,
      "grad_norm": 0.10498046875,
      "learning_rate": 5.209205020920503e-06,
      "loss": 0.8882,
      "step": 458
    },
    {
      "epoch": 0.4801255230125523,
      "grad_norm": 0.08984375,
      "learning_rate": 5.198744769874478e-06,
      "loss": 0.8204,
      "step": 459
    },
    {
      "epoch": 0.4811715481171548,
      "grad_norm": 0.10400390625,
      "learning_rate": 5.188284518828452e-06,
      "loss": 0.8705,
      "step": 460
    },
    {
      "epoch": 0.48221757322175735,
      "grad_norm": 0.10107421875,
      "learning_rate": 5.177824267782428e-06,
      "loss": 0.8861,
      "step": 461
    },
    {
      "epoch": 0.48326359832635984,
      "grad_norm": 0.08642578125,
      "learning_rate": 5.167364016736403e-06,
      "loss": 0.8654,
      "step": 462
    },
    {
      "epoch": 0.48430962343096234,
      "grad_norm": 0.11474609375,
      "learning_rate": 5.156903765690377e-06,
      "loss": 0.9339,
      "step": 463
    },
    {
      "epoch": 0.48535564853556484,
      "grad_norm": 0.10693359375,
      "learning_rate": 5.146443514644351e-06,
      "loss": 0.8879,
      "step": 464
    },
    {
      "epoch": 0.48640167364016734,
      "grad_norm": 0.09228515625,
      "learning_rate": 5.135983263598327e-06,
      "loss": 0.8486,
      "step": 465
    },
    {
      "epoch": 0.4874476987447699,
      "grad_norm": 0.099609375,
      "learning_rate": 5.125523012552301e-06,
      "loss": 0.8765,
      "step": 466
    },
    {
      "epoch": 0.4884937238493724,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.115062761506276e-06,
      "loss": 0.8455,
      "step": 467
    },
    {
      "epoch": 0.4895397489539749,
      "grad_norm": 0.09619140625,
      "learning_rate": 5.104602510460252e-06,
      "loss": 0.8361,
      "step": 468
    },
    {
      "epoch": 0.4905857740585774,
      "grad_norm": 0.1142578125,
      "learning_rate": 5.094142259414226e-06,
      "loss": 0.9174,
      "step": 469
    },
    {
      "epoch": 0.4916317991631799,
      "grad_norm": 0.1015625,
      "learning_rate": 5.083682008368201e-06,
      "loss": 0.8871,
      "step": 470
    },
    {
      "epoch": 0.49267782426778245,
      "grad_norm": 0.10107421875,
      "learning_rate": 5.073221757322177e-06,
      "loss": 0.9009,
      "step": 471
    },
    {
      "epoch": 0.49372384937238495,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.062761506276151e-06,
      "loss": 0.8671,
      "step": 472
    },
    {
      "epoch": 0.49476987447698745,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.052301255230127e-06,
      "loss": 0.8869,
      "step": 473
    },
    {
      "epoch": 0.49581589958158995,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.041841004184101e-06,
      "loss": 0.8629,
      "step": 474
    },
    {
      "epoch": 0.49686192468619245,
      "grad_norm": 0.09765625,
      "learning_rate": 5.0313807531380754e-06,
      "loss": 0.8809,
      "step": 475
    },
    {
      "epoch": 0.497907949790795,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.02092050209205e-06,
      "loss": 0.8865,
      "step": 476
    },
    {
      "epoch": 0.4989539748953975,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.010460251046026e-06,
      "loss": 0.8911,
      "step": 477
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1025390625,
      "learning_rate": 5e-06,
      "loss": 0.8873,
      "step": 478
    },
    {
      "epoch": 0.5010460251046025,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.989539748953975e-06,
      "loss": 0.8808,
      "step": 479
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.979079497907951e-06,
      "loss": 0.8761,
      "step": 480
    },
    {
      "epoch": 0.5031380753138075,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.9686192468619245e-06,
      "loss": 0.8068,
      "step": 481
    },
    {
      "epoch": 0.50418410041841,
      "grad_norm": 0.095703125,
      "learning_rate": 4.9581589958159e-06,
      "loss": 0.8576,
      "step": 482
    },
    {
      "epoch": 0.5052301255230126,
      "grad_norm": 0.09033203125,
      "learning_rate": 4.947698744769875e-06,
      "loss": 0.8834,
      "step": 483
    },
    {
      "epoch": 0.5062761506276151,
      "grad_norm": 0.08837890625,
      "learning_rate": 4.9372384937238495e-06,
      "loss": 0.8536,
      "step": 484
    },
    {
      "epoch": 0.5073221757322176,
      "grad_norm": 0.1123046875,
      "learning_rate": 4.926778242677825e-06,
      "loss": 0.9277,
      "step": 485
    },
    {
      "epoch": 0.5083682008368201,
      "grad_norm": 0.08203125,
      "learning_rate": 4.9163179916318e-06,
      "loss": 0.8232,
      "step": 486
    },
    {
      "epoch": 0.5094142259414226,
      "grad_norm": 0.08984375,
      "learning_rate": 4.9058577405857744e-06,
      "loss": 0.8615,
      "step": 487
    },
    {
      "epoch": 0.5104602510460251,
      "grad_norm": 0.11474609375,
      "learning_rate": 4.895397489539749e-06,
      "loss": 0.8843,
      "step": 488
    },
    {
      "epoch": 0.5115062761506276,
      "grad_norm": 0.087890625,
      "learning_rate": 4.884937238493725e-06,
      "loss": 0.8163,
      "step": 489
    },
    {
      "epoch": 0.5125523012552301,
      "grad_norm": 0.08984375,
      "learning_rate": 4.874476987447699e-06,
      "loss": 0.8757,
      "step": 490
    },
    {
      "epoch": 0.5135983263598326,
      "grad_norm": 0.095703125,
      "learning_rate": 4.864016736401674e-06,
      "loss": 0.8783,
      "step": 491
    },
    {
      "epoch": 0.5146443514644351,
      "grad_norm": 0.08642578125,
      "learning_rate": 4.853556485355649e-06,
      "loss": 0.8225,
      "step": 492
    },
    {
      "epoch": 0.5156903765690377,
      "grad_norm": 0.099609375,
      "learning_rate": 4.8430962343096235e-06,
      "loss": 0.8611,
      "step": 493
    },
    {
      "epoch": 0.5167364016736402,
      "grad_norm": 0.1064453125,
      "learning_rate": 4.832635983263599e-06,
      "loss": 0.9104,
      "step": 494
    },
    {
      "epoch": 0.5177824267782427,
      "grad_norm": 0.08984375,
      "learning_rate": 4.822175732217574e-06,
      "loss": 0.8532,
      "step": 495
    },
    {
      "epoch": 0.5188284518828452,
      "grad_norm": 0.10009765625,
      "learning_rate": 4.8117154811715485e-06,
      "loss": 0.8623,
      "step": 496
    },
    {
      "epoch": 0.5198744769874477,
      "grad_norm": 0.0859375,
      "learning_rate": 4.801255230125523e-06,
      "loss": 0.8516,
      "step": 497
    },
    {
      "epoch": 0.5209205020920502,
      "grad_norm": 0.0986328125,
      "learning_rate": 4.790794979079498e-06,
      "loss": 0.865,
      "step": 498
    },
    {
      "epoch": 0.5219665271966527,
      "grad_norm": 0.0849609375,
      "learning_rate": 4.7803347280334734e-06,
      "loss": 0.8243,
      "step": 499
    },
    {
      "epoch": 0.5230125523012552,
      "grad_norm": 0.09619140625,
      "learning_rate": 4.769874476987448e-06,
      "loss": 0.8641,
      "step": 500
    },
    {
      "epoch": 0.5240585774058577,
      "grad_norm": 0.09423828125,
      "learning_rate": 4.759414225941423e-06,
      "loss": 0.8721,
      "step": 501
    },
    {
      "epoch": 0.5251046025104602,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.7489539748953976e-06,
      "loss": 0.8669,
      "step": 502
    },
    {
      "epoch": 0.5261506276150628,
      "grad_norm": 0.10009765625,
      "learning_rate": 4.738493723849373e-06,
      "loss": 0.8812,
      "step": 503
    },
    {
      "epoch": 0.5271966527196653,
      "grad_norm": 0.09375,
      "learning_rate": 4.728033472803348e-06,
      "loss": 0.8685,
      "step": 504
    },
    {
      "epoch": 0.5282426778242678,
      "grad_norm": 0.0859375,
      "learning_rate": 4.7175732217573225e-06,
      "loss": 0.841,
      "step": 505
    },
    {
      "epoch": 0.5292887029288703,
      "grad_norm": 0.10791015625,
      "learning_rate": 4.707112970711297e-06,
      "loss": 0.9178,
      "step": 506
    },
    {
      "epoch": 0.5303347280334728,
      "grad_norm": 0.09033203125,
      "learning_rate": 4.696652719665272e-06,
      "loss": 0.8741,
      "step": 507
    },
    {
      "epoch": 0.5313807531380753,
      "grad_norm": 0.09765625,
      "learning_rate": 4.6861924686192475e-06,
      "loss": 0.802,
      "step": 508
    },
    {
      "epoch": 0.5324267782426778,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.675732217573222e-06,
      "loss": 0.8591,
      "step": 509
    },
    {
      "epoch": 0.5334728033472803,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.665271966527197e-06,
      "loss": 0.8691,
      "step": 510
    },
    {
      "epoch": 0.5345188284518828,
      "grad_norm": 0.10693359375,
      "learning_rate": 4.654811715481172e-06,
      "loss": 0.8785,
      "step": 511
    },
    {
      "epoch": 0.5355648535564853,
      "grad_norm": 0.08984375,
      "learning_rate": 4.644351464435146e-06,
      "loss": 0.841,
      "step": 512
    },
    {
      "epoch": 0.5366108786610879,
      "grad_norm": 0.09716796875,
      "learning_rate": 4.633891213389122e-06,
      "loss": 0.885,
      "step": 513
    },
    {
      "epoch": 0.5376569037656904,
      "grad_norm": 0.10888671875,
      "learning_rate": 4.6234309623430966e-06,
      "loss": 0.8998,
      "step": 514
    },
    {
      "epoch": 0.5387029288702929,
      "grad_norm": 0.09619140625,
      "learning_rate": 4.612970711297072e-06,
      "loss": 0.8567,
      "step": 515
    },
    {
      "epoch": 0.5397489539748954,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.602510460251046e-06,
      "loss": 0.8909,
      "step": 516
    },
    {
      "epoch": 0.5407949790794979,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.5920502092050215e-06,
      "loss": 0.8617,
      "step": 517
    },
    {
      "epoch": 0.5418410041841004,
      "grad_norm": 0.10302734375,
      "learning_rate": 4.581589958158996e-06,
      "loss": 0.8556,
      "step": 518
    },
    {
      "epoch": 0.5428870292887029,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.571129707112971e-06,
      "loss": 0.8341,
      "step": 519
    },
    {
      "epoch": 0.5439330543933054,
      "grad_norm": 0.09521484375,
      "learning_rate": 4.5606694560669465e-06,
      "loss": 0.8571,
      "step": 520
    },
    {
      "epoch": 0.5449790794979079,
      "grad_norm": 0.09423828125,
      "learning_rate": 4.55020920502092e-06,
      "loss": 0.8406,
      "step": 521
    },
    {
      "epoch": 0.5460251046025104,
      "grad_norm": 0.10205078125,
      "learning_rate": 4.539748953974896e-06,
      "loss": 0.8728,
      "step": 522
    },
    {
      "epoch": 0.547071129707113,
      "grad_norm": 0.1005859375,
      "learning_rate": 4.529288702928871e-06,
      "loss": 0.8657,
      "step": 523
    },
    {
      "epoch": 0.5481171548117155,
      "grad_norm": 0.10400390625,
      "learning_rate": 4.518828451882845e-06,
      "loss": 0.8691,
      "step": 524
    },
    {
      "epoch": 0.549163179916318,
      "grad_norm": 0.0986328125,
      "learning_rate": 4.508368200836821e-06,
      "loss": 0.8224,
      "step": 525
    },
    {
      "epoch": 0.5502092050209205,
      "grad_norm": 0.109375,
      "learning_rate": 4.4979079497907956e-06,
      "loss": 0.871,
      "step": 526
    },
    {
      "epoch": 0.551255230125523,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.48744769874477e-06,
      "loss": 0.8908,
      "step": 527
    },
    {
      "epoch": 0.5523012552301255,
      "grad_norm": 0.095703125,
      "learning_rate": 4.476987447698745e-06,
      "loss": 0.8415,
      "step": 528
    },
    {
      "epoch": 0.553347280334728,
      "grad_norm": 0.08984375,
      "learning_rate": 4.4665271966527205e-06,
      "loss": 0.8476,
      "step": 529
    },
    {
      "epoch": 0.5543933054393305,
      "grad_norm": 0.08740234375,
      "learning_rate": 4.456066945606695e-06,
      "loss": 0.8253,
      "step": 530
    },
    {
      "epoch": 0.555439330543933,
      "grad_norm": 0.103515625,
      "learning_rate": 4.44560669456067e-06,
      "loss": 0.8927,
      "step": 531
    },
    {
      "epoch": 0.5564853556485355,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.435146443514645e-06,
      "loss": 0.9062,
      "step": 532
    },
    {
      "epoch": 0.5575313807531381,
      "grad_norm": 0.09716796875,
      "learning_rate": 4.424686192468619e-06,
      "loss": 0.8632,
      "step": 533
    },
    {
      "epoch": 0.5585774058577406,
      "grad_norm": 0.10546875,
      "learning_rate": 4.414225941422595e-06,
      "loss": 0.8825,
      "step": 534
    },
    {
      "epoch": 0.5596234309623431,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.40376569037657e-06,
      "loss": 0.821,
      "step": 535
    },
    {
      "epoch": 0.5606694560669456,
      "grad_norm": 0.11376953125,
      "learning_rate": 4.393305439330544e-06,
      "loss": 0.9028,
      "step": 536
    },
    {
      "epoch": 0.5617154811715481,
      "grad_norm": 0.0986328125,
      "learning_rate": 4.382845188284519e-06,
      "loss": 0.8516,
      "step": 537
    },
    {
      "epoch": 0.5627615062761506,
      "grad_norm": 0.09423828125,
      "learning_rate": 4.372384937238494e-06,
      "loss": 0.8684,
      "step": 538
    },
    {
      "epoch": 0.5638075313807531,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.361924686192469e-06,
      "loss": 0.8528,
      "step": 539
    },
    {
      "epoch": 0.5648535564853556,
      "grad_norm": 0.11279296875,
      "learning_rate": 4.351464435146444e-06,
      "loss": 0.8934,
      "step": 540
    },
    {
      "epoch": 0.5658995815899581,
      "grad_norm": 0.091796875,
      "learning_rate": 4.341004184100419e-06,
      "loss": 0.8286,
      "step": 541
    },
    {
      "epoch": 0.5669456066945606,
      "grad_norm": 0.09375,
      "learning_rate": 4.330543933054393e-06,
      "loss": 0.7909,
      "step": 542
    },
    {
      "epoch": 0.5679916317991632,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.320083682008369e-06,
      "loss": 0.8797,
      "step": 543
    },
    {
      "epoch": 0.5690376569037657,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.309623430962344e-06,
      "loss": 0.8559,
      "step": 544
    },
    {
      "epoch": 0.5700836820083682,
      "grad_norm": 0.11962890625,
      "learning_rate": 4.299163179916318e-06,
      "loss": 0.8863,
      "step": 545
    },
    {
      "epoch": 0.5711297071129707,
      "grad_norm": 0.10888671875,
      "learning_rate": 4.288702928870293e-06,
      "loss": 0.8825,
      "step": 546
    },
    {
      "epoch": 0.5721757322175732,
      "grad_norm": 0.09130859375,
      "learning_rate": 4.278242677824268e-06,
      "loss": 0.8801,
      "step": 547
    },
    {
      "epoch": 0.5732217573221757,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.267782426778243e-06,
      "loss": 0.9142,
      "step": 548
    },
    {
      "epoch": 0.5742677824267782,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.257322175732218e-06,
      "loss": 0.906,
      "step": 549
    },
    {
      "epoch": 0.5753138075313807,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.246861924686193e-06,
      "loss": 0.9119,
      "step": 550
    },
    {
      "epoch": 0.5763598326359832,
      "grad_norm": 0.099609375,
      "learning_rate": 4.236401673640167e-06,
      "loss": 0.8944,
      "step": 551
    },
    {
      "epoch": 0.5774058577405857,
      "grad_norm": 0.1064453125,
      "learning_rate": 4.225941422594142e-06,
      "loss": 0.8915,
      "step": 552
    },
    {
      "epoch": 0.5784518828451883,
      "grad_norm": 0.095703125,
      "learning_rate": 4.215481171548118e-06,
      "loss": 0.8774,
      "step": 553
    },
    {
      "epoch": 0.5794979079497908,
      "grad_norm": 0.09912109375,
      "learning_rate": 4.205020920502092e-06,
      "loss": 0.8404,
      "step": 554
    },
    {
      "epoch": 0.5805439330543933,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.194560669456068e-06,
      "loss": 0.8677,
      "step": 555
    },
    {
      "epoch": 0.5815899581589958,
      "grad_norm": 0.099609375,
      "learning_rate": 4.184100418410042e-06,
      "loss": 0.8739,
      "step": 556
    },
    {
      "epoch": 0.5826359832635983,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.173640167364017e-06,
      "loss": 0.8615,
      "step": 557
    },
    {
      "epoch": 0.5836820083682008,
      "grad_norm": 0.10546875,
      "learning_rate": 4.163179916317992e-06,
      "loss": 0.8825,
      "step": 558
    },
    {
      "epoch": 0.5847280334728033,
      "grad_norm": 0.09228515625,
      "learning_rate": 4.152719665271967e-06,
      "loss": 0.8407,
      "step": 559
    },
    {
      "epoch": 0.5857740585774058,
      "grad_norm": 0.11083984375,
      "learning_rate": 4.142259414225942e-06,
      "loss": 0.8957,
      "step": 560
    },
    {
      "epoch": 0.5868200836820083,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.131799163179916e-06,
      "loss": 0.8527,
      "step": 561
    },
    {
      "epoch": 0.5878661087866108,
      "grad_norm": 0.08056640625,
      "learning_rate": 4.121338912133892e-06,
      "loss": 0.8654,
      "step": 562
    },
    {
      "epoch": 0.5889121338912134,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.110878661087866e-06,
      "loss": 0.8628,
      "step": 563
    },
    {
      "epoch": 0.5899581589958159,
      "grad_norm": 0.09326171875,
      "learning_rate": 4.100418410041841e-06,
      "loss": 0.8385,
      "step": 564
    },
    {
      "epoch": 0.5910041841004184,
      "grad_norm": 0.1103515625,
      "learning_rate": 4.089958158995817e-06,
      "loss": 0.8834,
      "step": 565
    },
    {
      "epoch": 0.5920502092050209,
      "grad_norm": 0.0869140625,
      "learning_rate": 4.0794979079497905e-06,
      "loss": 0.8654,
      "step": 566
    },
    {
      "epoch": 0.5930962343096234,
      "grad_norm": 0.1181640625,
      "learning_rate": 4.069037656903766e-06,
      "loss": 0.9128,
      "step": 567
    },
    {
      "epoch": 0.5941422594142259,
      "grad_norm": 0.11865234375,
      "learning_rate": 4.058577405857741e-06,
      "loss": 0.9384,
      "step": 568
    },
    {
      "epoch": 0.5951882845188284,
      "grad_norm": 0.09375,
      "learning_rate": 4.048117154811716e-06,
      "loss": 0.8636,
      "step": 569
    },
    {
      "epoch": 0.5962343096234309,
      "grad_norm": 0.09375,
      "learning_rate": 4.037656903765691e-06,
      "loss": 0.8303,
      "step": 570
    },
    {
      "epoch": 0.5972803347280334,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.027196652719666e-06,
      "loss": 0.8769,
      "step": 571
    },
    {
      "epoch": 0.5983263598326359,
      "grad_norm": 0.09375,
      "learning_rate": 4.0167364016736405e-06,
      "loss": 0.8639,
      "step": 572
    },
    {
      "epoch": 0.5993723849372385,
      "grad_norm": 0.08544921875,
      "learning_rate": 4.006276150627615e-06,
      "loss": 0.854,
      "step": 573
    },
    {
      "epoch": 0.600418410041841,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.995815899581591e-06,
      "loss": 0.8573,
      "step": 574
    },
    {
      "epoch": 0.6014644351464435,
      "grad_norm": 0.08203125,
      "learning_rate": 3.985355648535565e-06,
      "loss": 0.8373,
      "step": 575
    },
    {
      "epoch": 0.602510460251046,
      "grad_norm": 0.109375,
      "learning_rate": 3.97489539748954e-06,
      "loss": 0.8973,
      "step": 576
    },
    {
      "epoch": 0.6035564853556485,
      "grad_norm": 0.09423828125,
      "learning_rate": 3.964435146443515e-06,
      "loss": 0.8573,
      "step": 577
    },
    {
      "epoch": 0.604602510460251,
      "grad_norm": 0.0908203125,
      "learning_rate": 3.9539748953974895e-06,
      "loss": 0.8742,
      "step": 578
    },
    {
      "epoch": 0.6056485355648535,
      "grad_norm": 0.142578125,
      "learning_rate": 3.943514644351465e-06,
      "loss": 0.9369,
      "step": 579
    },
    {
      "epoch": 0.606694560669456,
      "grad_norm": 0.08447265625,
      "learning_rate": 3.93305439330544e-06,
      "loss": 0.8354,
      "step": 580
    },
    {
      "epoch": 0.6077405857740585,
      "grad_norm": 0.130859375,
      "learning_rate": 3.9225941422594145e-06,
      "loss": 0.9565,
      "step": 581
    },
    {
      "epoch": 0.608786610878661,
      "grad_norm": 0.11865234375,
      "learning_rate": 3.912133891213389e-06,
      "loss": 0.8563,
      "step": 582
    },
    {
      "epoch": 0.6098326359832636,
      "grad_norm": 0.1005859375,
      "learning_rate": 3.901673640167365e-06,
      "loss": 0.8788,
      "step": 583
    },
    {
      "epoch": 0.6108786610878661,
      "grad_norm": 0.08740234375,
      "learning_rate": 3.8912133891213395e-06,
      "loss": 0.8446,
      "step": 584
    },
    {
      "epoch": 0.6119246861924686,
      "grad_norm": 0.103515625,
      "learning_rate": 3.880753138075314e-06,
      "loss": 0.8968,
      "step": 585
    },
    {
      "epoch": 0.6129707112970711,
      "grad_norm": 0.1142578125,
      "learning_rate": 3.870292887029289e-06,
      "loss": 0.9485,
      "step": 586
    },
    {
      "epoch": 0.6140167364016736,
      "grad_norm": 0.10302734375,
      "learning_rate": 3.8598326359832636e-06,
      "loss": 0.8737,
      "step": 587
    },
    {
      "epoch": 0.6150627615062761,
      "grad_norm": 0.11083984375,
      "learning_rate": 3.849372384937239e-06,
      "loss": 0.9064,
      "step": 588
    },
    {
      "epoch": 0.6161087866108786,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.838912133891214e-06,
      "loss": 0.866,
      "step": 589
    },
    {
      "epoch": 0.6171548117154811,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.8284518828451885e-06,
      "loss": 0.8705,
      "step": 590
    },
    {
      "epoch": 0.6182008368200836,
      "grad_norm": 0.11376953125,
      "learning_rate": 3.817991631799163e-06,
      "loss": 0.9165,
      "step": 591
    },
    {
      "epoch": 0.6192468619246861,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.8075313807531384e-06,
      "loss": 0.8676,
      "step": 592
    },
    {
      "epoch": 0.6202928870292888,
      "grad_norm": 0.08251953125,
      "learning_rate": 3.7970711297071135e-06,
      "loss": 0.8063,
      "step": 593
    },
    {
      "epoch": 0.6213389121338913,
      "grad_norm": 0.1162109375,
      "learning_rate": 3.786610878661088e-06,
      "loss": 0.8948,
      "step": 594
    },
    {
      "epoch": 0.6223849372384938,
      "grad_norm": 0.12353515625,
      "learning_rate": 3.7761506276150633e-06,
      "loss": 0.9441,
      "step": 595
    },
    {
      "epoch": 0.6234309623430963,
      "grad_norm": 0.09912109375,
      "learning_rate": 3.7656903765690376e-06,
      "loss": 0.8766,
      "step": 596
    },
    {
      "epoch": 0.6244769874476988,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.7552301255230127e-06,
      "loss": 0.8703,
      "step": 597
    },
    {
      "epoch": 0.6255230125523012,
      "grad_norm": 0.08251953125,
      "learning_rate": 3.744769874476988e-06,
      "loss": 0.8276,
      "step": 598
    },
    {
      "epoch": 0.6265690376569037,
      "grad_norm": 0.10205078125,
      "learning_rate": 3.734309623430963e-06,
      "loss": 0.8466,
      "step": 599
    },
    {
      "epoch": 0.6276150627615062,
      "grad_norm": 0.10009765625,
      "learning_rate": 3.7238493723849373e-06,
      "loss": 0.8471,
      "step": 600
    },
    {
      "epoch": 0.6286610878661087,
      "grad_norm": 0.09033203125,
      "learning_rate": 3.7133891213389124e-06,
      "loss": 0.8262,
      "step": 601
    },
    {
      "epoch": 0.6297071129707112,
      "grad_norm": 0.11865234375,
      "learning_rate": 3.702928870292887e-06,
      "loss": 0.9015,
      "step": 602
    },
    {
      "epoch": 0.6307531380753139,
      "grad_norm": 0.0927734375,
      "learning_rate": 3.6924686192468622e-06,
      "loss": 0.8548,
      "step": 603
    },
    {
      "epoch": 0.6317991631799164,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.6820083682008374e-06,
      "loss": 0.7881,
      "step": 604
    },
    {
      "epoch": 0.6328451882845189,
      "grad_norm": 0.11572265625,
      "learning_rate": 3.6715481171548117e-06,
      "loss": 0.9528,
      "step": 605
    },
    {
      "epoch": 0.6338912133891214,
      "grad_norm": 0.09765625,
      "learning_rate": 3.6610878661087868e-06,
      "loss": 0.8503,
      "step": 606
    },
    {
      "epoch": 0.6349372384937239,
      "grad_norm": 0.10107421875,
      "learning_rate": 3.650627615062762e-06,
      "loss": 0.8578,
      "step": 607
    },
    {
      "epoch": 0.6359832635983264,
      "grad_norm": 0.08544921875,
      "learning_rate": 3.6401673640167366e-06,
      "loss": 0.759,
      "step": 608
    },
    {
      "epoch": 0.6370292887029289,
      "grad_norm": 0.10888671875,
      "learning_rate": 3.6297071129707117e-06,
      "loss": 0.8866,
      "step": 609
    },
    {
      "epoch": 0.6380753138075314,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.619246861924686e-06,
      "loss": 0.8999,
      "step": 610
    },
    {
      "epoch": 0.6391213389121339,
      "grad_norm": 0.11865234375,
      "learning_rate": 3.608786610878661e-06,
      "loss": 0.8906,
      "step": 611
    },
    {
      "epoch": 0.6401673640167364,
      "grad_norm": 0.1025390625,
      "learning_rate": 3.5983263598326363e-06,
      "loss": 0.8221,
      "step": 612
    },
    {
      "epoch": 0.641213389121339,
      "grad_norm": 0.0908203125,
      "learning_rate": 3.5878661087866114e-06,
      "loss": 0.8639,
      "step": 613
    },
    {
      "epoch": 0.6422594142259415,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.577405857740586e-06,
      "loss": 0.8807,
      "step": 614
    },
    {
      "epoch": 0.643305439330544,
      "grad_norm": 0.091796875,
      "learning_rate": 3.566945606694561e-06,
      "loss": 0.8815,
      "step": 615
    },
    {
      "epoch": 0.6443514644351465,
      "grad_norm": 0.09619140625,
      "learning_rate": 3.5564853556485355e-06,
      "loss": 0.8524,
      "step": 616
    },
    {
      "epoch": 0.645397489539749,
      "grad_norm": 0.107421875,
      "learning_rate": 3.5460251046025107e-06,
      "loss": 0.8805,
      "step": 617
    },
    {
      "epoch": 0.6464435146443515,
      "grad_norm": 0.0927734375,
      "learning_rate": 3.5355648535564858e-06,
      "loss": 0.8298,
      "step": 618
    },
    {
      "epoch": 0.647489539748954,
      "grad_norm": 0.08984375,
      "learning_rate": 3.525104602510461e-06,
      "loss": 0.8323,
      "step": 619
    },
    {
      "epoch": 0.6485355648535565,
      "grad_norm": 0.10546875,
      "learning_rate": 3.514644351464435e-06,
      "loss": 0.8698,
      "step": 620
    },
    {
      "epoch": 0.649581589958159,
      "grad_norm": 0.091796875,
      "learning_rate": 3.5041841004184103e-06,
      "loss": 0.8337,
      "step": 621
    },
    {
      "epoch": 0.6506276150627615,
      "grad_norm": 0.109375,
      "learning_rate": 3.493723849372385e-06,
      "loss": 0.881,
      "step": 622
    },
    {
      "epoch": 0.6516736401673641,
      "grad_norm": 0.1015625,
      "learning_rate": 3.48326359832636e-06,
      "loss": 0.8694,
      "step": 623
    },
    {
      "epoch": 0.6527196652719666,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.4728033472803353e-06,
      "loss": 0.9134,
      "step": 624
    },
    {
      "epoch": 0.6537656903765691,
      "grad_norm": 0.08837890625,
      "learning_rate": 3.4623430962343096e-06,
      "loss": 0.8696,
      "step": 625
    },
    {
      "epoch": 0.6548117154811716,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.4518828451882847e-06,
      "loss": 0.8623,
      "step": 626
    },
    {
      "epoch": 0.6558577405857741,
      "grad_norm": 0.1103515625,
      "learning_rate": 3.44142259414226e-06,
      "loss": 0.8869,
      "step": 627
    },
    {
      "epoch": 0.6569037656903766,
      "grad_norm": 0.095703125,
      "learning_rate": 3.4309623430962345e-06,
      "loss": 0.8641,
      "step": 628
    },
    {
      "epoch": 0.6579497907949791,
      "grad_norm": 0.099609375,
      "learning_rate": 3.4205020920502097e-06,
      "loss": 0.8798,
      "step": 629
    },
    {
      "epoch": 0.6589958158995816,
      "grad_norm": 0.08837890625,
      "learning_rate": 3.410041841004184e-06,
      "loss": 0.8221,
      "step": 630
    },
    {
      "epoch": 0.6600418410041841,
      "grad_norm": 0.1103515625,
      "learning_rate": 3.399581589958159e-06,
      "loss": 0.8785,
      "step": 631
    },
    {
      "epoch": 0.6610878661087866,
      "grad_norm": 0.099609375,
      "learning_rate": 3.389121338912134e-06,
      "loss": 0.8835,
      "step": 632
    },
    {
      "epoch": 0.6621338912133892,
      "grad_norm": 0.08935546875,
      "learning_rate": 3.3786610878661093e-06,
      "loss": 0.818,
      "step": 633
    },
    {
      "epoch": 0.6631799163179917,
      "grad_norm": 0.1015625,
      "learning_rate": 3.368200836820084e-06,
      "loss": 0.854,
      "step": 634
    },
    {
      "epoch": 0.6642259414225942,
      "grad_norm": 0.10498046875,
      "learning_rate": 3.3577405857740587e-06,
      "loss": 0.853,
      "step": 635
    },
    {
      "epoch": 0.6652719665271967,
      "grad_norm": 0.099609375,
      "learning_rate": 3.3472803347280334e-06,
      "loss": 0.8782,
      "step": 636
    },
    {
      "epoch": 0.6663179916317992,
      "grad_norm": 0.10400390625,
      "learning_rate": 3.3368200836820086e-06,
      "loss": 0.8837,
      "step": 637
    },
    {
      "epoch": 0.6673640167364017,
      "grad_norm": 0.10986328125,
      "learning_rate": 3.3263598326359837e-06,
      "loss": 0.8869,
      "step": 638
    },
    {
      "epoch": 0.6684100418410042,
      "grad_norm": 0.099609375,
      "learning_rate": 3.315899581589959e-06,
      "loss": 0.875,
      "step": 639
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 0.09228515625,
      "learning_rate": 3.305439330543933e-06,
      "loss": 0.8555,
      "step": 640
    },
    {
      "epoch": 0.6705020920502092,
      "grad_norm": 0.10595703125,
      "learning_rate": 3.2949790794979082e-06,
      "loss": 0.9065,
      "step": 641
    },
    {
      "epoch": 0.6715481171548117,
      "grad_norm": 0.1162109375,
      "learning_rate": 3.284518828451883e-06,
      "loss": 0.895,
      "step": 642
    },
    {
      "epoch": 0.6725941422594143,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.274058577405858e-06,
      "loss": 0.8262,
      "step": 643
    },
    {
      "epoch": 0.6736401673640168,
      "grad_norm": 0.1328125,
      "learning_rate": 3.263598326359833e-06,
      "loss": 0.9373,
      "step": 644
    },
    {
      "epoch": 0.6746861924686193,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.2531380753138075e-06,
      "loss": 0.8639,
      "step": 645
    },
    {
      "epoch": 0.6757322175732218,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.2426778242677826e-06,
      "loss": 0.8723,
      "step": 646
    },
    {
      "epoch": 0.6767782426778243,
      "grad_norm": 0.10595703125,
      "learning_rate": 3.2322175732217577e-06,
      "loss": 0.8464,
      "step": 647
    },
    {
      "epoch": 0.6778242677824268,
      "grad_norm": 0.08251953125,
      "learning_rate": 3.2217573221757324e-06,
      "loss": 0.8299,
      "step": 648
    },
    {
      "epoch": 0.6788702928870293,
      "grad_norm": 0.1162109375,
      "learning_rate": 3.2112970711297076e-06,
      "loss": 0.9213,
      "step": 649
    },
    {
      "epoch": 0.6799163179916318,
      "grad_norm": 0.10107421875,
      "learning_rate": 3.200836820083682e-06,
      "loss": 0.8463,
      "step": 650
    },
    {
      "epoch": 0.6809623430962343,
      "grad_norm": 0.09619140625,
      "learning_rate": 3.190376569037657e-06,
      "loss": 0.8579,
      "step": 651
    },
    {
      "epoch": 0.6820083682008368,
      "grad_norm": 0.09619140625,
      "learning_rate": 3.179916317991632e-06,
      "loss": 0.8196,
      "step": 652
    },
    {
      "epoch": 0.6830543933054394,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.1694560669456072e-06,
      "loss": 0.8815,
      "step": 653
    },
    {
      "epoch": 0.6841004184100419,
      "grad_norm": 0.09423828125,
      "learning_rate": 3.158995815899582e-06,
      "loss": 0.815,
      "step": 654
    },
    {
      "epoch": 0.6851464435146444,
      "grad_norm": 0.08837890625,
      "learning_rate": 3.1485355648535566e-06,
      "loss": 0.7707,
      "step": 655
    },
    {
      "epoch": 0.6861924686192469,
      "grad_norm": 0.130859375,
      "learning_rate": 3.1380753138075313e-06,
      "loss": 0.8292,
      "step": 656
    },
    {
      "epoch": 0.6872384937238494,
      "grad_norm": 0.08984375,
      "learning_rate": 3.1276150627615065e-06,
      "loss": 0.7917,
      "step": 657
    },
    {
      "epoch": 0.6882845188284519,
      "grad_norm": 0.099609375,
      "learning_rate": 3.1171548117154816e-06,
      "loss": 0.8956,
      "step": 658
    },
    {
      "epoch": 0.6893305439330544,
      "grad_norm": 0.11328125,
      "learning_rate": 3.1066945606694563e-06,
      "loss": 0.9042,
      "step": 659
    },
    {
      "epoch": 0.6903765690376569,
      "grad_norm": 0.08837890625,
      "learning_rate": 3.096234309623431e-06,
      "loss": 0.8367,
      "step": 660
    },
    {
      "epoch": 0.6914225941422594,
      "grad_norm": 0.09228515625,
      "learning_rate": 3.085774058577406e-06,
      "loss": 0.8347,
      "step": 661
    },
    {
      "epoch": 0.6924686192468619,
      "grad_norm": 0.0908203125,
      "learning_rate": 3.075313807531381e-06,
      "loss": 0.8567,
      "step": 662
    },
    {
      "epoch": 0.6935146443514645,
      "grad_norm": 0.091796875,
      "learning_rate": 3.064853556485356e-06,
      "loss": 0.8927,
      "step": 663
    },
    {
      "epoch": 0.694560669456067,
      "grad_norm": 0.1064453125,
      "learning_rate": 3.054393305439331e-06,
      "loss": 0.8954,
      "step": 664
    },
    {
      "epoch": 0.6956066945606695,
      "grad_norm": 0.1083984375,
      "learning_rate": 3.0439330543933054e-06,
      "loss": 0.8552,
      "step": 665
    },
    {
      "epoch": 0.696652719665272,
      "grad_norm": 0.111328125,
      "learning_rate": 3.0334728033472805e-06,
      "loss": 0.8976,
      "step": 666
    },
    {
      "epoch": 0.6976987447698745,
      "grad_norm": 0.12353515625,
      "learning_rate": 3.0230125523012556e-06,
      "loss": 0.8636,
      "step": 667
    },
    {
      "epoch": 0.698744769874477,
      "grad_norm": 0.0927734375,
      "learning_rate": 3.0125523012552303e-06,
      "loss": 0.8367,
      "step": 668
    },
    {
      "epoch": 0.6997907949790795,
      "grad_norm": 0.091796875,
      "learning_rate": 3.0020920502092055e-06,
      "loss": 0.8557,
      "step": 669
    },
    {
      "epoch": 0.700836820083682,
      "grad_norm": 0.1044921875,
      "learning_rate": 2.9916317991631798e-06,
      "loss": 0.8191,
      "step": 670
    },
    {
      "epoch": 0.7018828451882845,
      "grad_norm": 0.09423828125,
      "learning_rate": 2.981171548117155e-06,
      "loss": 0.8189,
      "step": 671
    },
    {
      "epoch": 0.702928870292887,
      "grad_norm": 0.09912109375,
      "learning_rate": 2.97071129707113e-06,
      "loss": 0.8531,
      "step": 672
    },
    {
      "epoch": 0.7039748953974896,
      "grad_norm": 0.1025390625,
      "learning_rate": 2.960251046025105e-06,
      "loss": 0.8707,
      "step": 673
    },
    {
      "epoch": 0.7050209205020921,
      "grad_norm": 0.0869140625,
      "learning_rate": 2.94979079497908e-06,
      "loss": 0.864,
      "step": 674
    },
    {
      "epoch": 0.7060669456066946,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.9393305439330546e-06,
      "loss": 0.8408,
      "step": 675
    },
    {
      "epoch": 0.7071129707112971,
      "grad_norm": 0.09765625,
      "learning_rate": 2.9288702928870293e-06,
      "loss": 0.8563,
      "step": 676
    },
    {
      "epoch": 0.7081589958158996,
      "grad_norm": 0.1162109375,
      "learning_rate": 2.9184100418410044e-06,
      "loss": 0.8916,
      "step": 677
    },
    {
      "epoch": 0.7092050209205021,
      "grad_norm": 0.08984375,
      "learning_rate": 2.9079497907949795e-06,
      "loss": 0.825,
      "step": 678
    },
    {
      "epoch": 0.7102510460251046,
      "grad_norm": 0.09375,
      "learning_rate": 2.8974895397489542e-06,
      "loss": 0.8608,
      "step": 679
    },
    {
      "epoch": 0.7112970711297071,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.887029288702929e-06,
      "loss": 0.8671,
      "step": 680
    },
    {
      "epoch": 0.7123430962343096,
      "grad_norm": 0.1103515625,
      "learning_rate": 2.876569037656904e-06,
      "loss": 0.8897,
      "step": 681
    },
    {
      "epoch": 0.7133891213389121,
      "grad_norm": 0.11669921875,
      "learning_rate": 2.8661087866108788e-06,
      "loss": 0.9009,
      "step": 682
    },
    {
      "epoch": 0.7144351464435147,
      "grad_norm": 0.125,
      "learning_rate": 2.855648535564854e-06,
      "loss": 0.9556,
      "step": 683
    },
    {
      "epoch": 0.7154811715481172,
      "grad_norm": 0.08154296875,
      "learning_rate": 2.845188284518829e-06,
      "loss": 0.8242,
      "step": 684
    },
    {
      "epoch": 0.7165271966527197,
      "grad_norm": 0.1044921875,
      "learning_rate": 2.8347280334728033e-06,
      "loss": 0.8645,
      "step": 685
    },
    {
      "epoch": 0.7175732217573222,
      "grad_norm": 0.08251953125,
      "learning_rate": 2.8242677824267784e-06,
      "loss": 0.7855,
      "step": 686
    },
    {
      "epoch": 0.7186192468619247,
      "grad_norm": 0.10546875,
      "learning_rate": 2.8138075313807536e-06,
      "loss": 0.8884,
      "step": 687
    },
    {
      "epoch": 0.7196652719665272,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.8033472803347283e-06,
      "loss": 0.8042,
      "step": 688
    },
    {
      "epoch": 0.7207112970711297,
      "grad_norm": 0.0888671875,
      "learning_rate": 2.7928870292887034e-06,
      "loss": 0.8291,
      "step": 689
    },
    {
      "epoch": 0.7217573221757322,
      "grad_norm": 0.10888671875,
      "learning_rate": 2.7824267782426777e-06,
      "loss": 0.9078,
      "step": 690
    },
    {
      "epoch": 0.7228033472803347,
      "grad_norm": 0.111328125,
      "learning_rate": 2.771966527196653e-06,
      "loss": 0.8689,
      "step": 691
    },
    {
      "epoch": 0.7238493723849372,
      "grad_norm": 0.10107421875,
      "learning_rate": 2.761506276150628e-06,
      "loss": 0.9107,
      "step": 692
    },
    {
      "epoch": 0.7248953974895398,
      "grad_norm": 0.1015625,
      "learning_rate": 2.7510460251046026e-06,
      "loss": 0.8694,
      "step": 693
    },
    {
      "epoch": 0.7259414225941423,
      "grad_norm": 0.0888671875,
      "learning_rate": 2.7405857740585778e-06,
      "loss": 0.8606,
      "step": 694
    },
    {
      "epoch": 0.7269874476987448,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.7301255230125525e-06,
      "loss": 0.838,
      "step": 695
    },
    {
      "epoch": 0.7280334728033473,
      "grad_norm": 0.10498046875,
      "learning_rate": 2.719665271966527e-06,
      "loss": 0.8472,
      "step": 696
    },
    {
      "epoch": 0.7290794979079498,
      "grad_norm": 0.10009765625,
      "learning_rate": 2.7092050209205023e-06,
      "loss": 0.8877,
      "step": 697
    },
    {
      "epoch": 0.7301255230125523,
      "grad_norm": 0.09228515625,
      "learning_rate": 2.6987447698744774e-06,
      "loss": 0.8396,
      "step": 698
    },
    {
      "epoch": 0.7311715481171548,
      "grad_norm": 0.10009765625,
      "learning_rate": 2.688284518828452e-06,
      "loss": 0.8664,
      "step": 699
    },
    {
      "epoch": 0.7322175732217573,
      "grad_norm": 0.09912109375,
      "learning_rate": 2.677824267782427e-06,
      "loss": 0.8484,
      "step": 700
    },
    {
      "epoch": 0.7332635983263598,
      "grad_norm": 0.0927734375,
      "learning_rate": 2.667364016736402e-06,
      "loss": 0.8336,
      "step": 701
    },
    {
      "epoch": 0.7343096234309623,
      "grad_norm": 0.09228515625,
      "learning_rate": 2.6569037656903767e-06,
      "loss": 0.8289,
      "step": 702
    },
    {
      "epoch": 0.7353556485355649,
      "grad_norm": 0.09814453125,
      "learning_rate": 2.646443514644352e-06,
      "loss": 0.8339,
      "step": 703
    },
    {
      "epoch": 0.7364016736401674,
      "grad_norm": 0.0869140625,
      "learning_rate": 2.635983263598327e-06,
      "loss": 0.849,
      "step": 704
    },
    {
      "epoch": 0.7374476987447699,
      "grad_norm": 0.10205078125,
      "learning_rate": 2.625523012552301e-06,
      "loss": 0.8722,
      "step": 705
    },
    {
      "epoch": 0.7384937238493724,
      "grad_norm": 0.1015625,
      "learning_rate": 2.6150627615062763e-06,
      "loss": 0.8463,
      "step": 706
    },
    {
      "epoch": 0.7395397489539749,
      "grad_norm": 0.08984375,
      "learning_rate": 2.6046025104602515e-06,
      "loss": 0.8233,
      "step": 707
    },
    {
      "epoch": 0.7405857740585774,
      "grad_norm": 0.10400390625,
      "learning_rate": 2.594142259414226e-06,
      "loss": 0.8976,
      "step": 708
    },
    {
      "epoch": 0.7416317991631799,
      "grad_norm": 0.09228515625,
      "learning_rate": 2.5836820083682013e-06,
      "loss": 0.8638,
      "step": 709
    },
    {
      "epoch": 0.7426778242677824,
      "grad_norm": 0.1123046875,
      "learning_rate": 2.5732217573221756e-06,
      "loss": 0.8638,
      "step": 710
    },
    {
      "epoch": 0.7437238493723849,
      "grad_norm": 0.0869140625,
      "learning_rate": 2.5627615062761507e-06,
      "loss": 0.8585,
      "step": 711
    },
    {
      "epoch": 0.7447698744769874,
      "grad_norm": 0.09814453125,
      "learning_rate": 2.552301255230126e-06,
      "loss": 0.8522,
      "step": 712
    },
    {
      "epoch": 0.74581589958159,
      "grad_norm": 0.115234375,
      "learning_rate": 2.5418410041841005e-06,
      "loss": 0.8869,
      "step": 713
    },
    {
      "epoch": 0.7468619246861925,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.5313807531380757e-06,
      "loss": 0.8304,
      "step": 714
    },
    {
      "epoch": 0.747907949790795,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.5209205020920504e-06,
      "loss": 0.8229,
      "step": 715
    },
    {
      "epoch": 0.7489539748953975,
      "grad_norm": 0.103515625,
      "learning_rate": 2.510460251046025e-06,
      "loss": 0.8414,
      "step": 716
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.109375,
      "learning_rate": 2.5e-06,
      "loss": 0.855,
      "step": 717
    },
    {
      "epoch": 0.7510460251046025,
      "grad_norm": 0.1064453125,
      "learning_rate": 2.4895397489539753e-06,
      "loss": 0.8528,
      "step": 718
    },
    {
      "epoch": 0.752092050209205,
      "grad_norm": 0.0966796875,
      "learning_rate": 2.47907949790795e-06,
      "loss": 0.8417,
      "step": 719
    },
    {
      "epoch": 0.7531380753138075,
      "grad_norm": 0.0888671875,
      "learning_rate": 2.4686192468619247e-06,
      "loss": 0.861,
      "step": 720
    },
    {
      "epoch": 0.75418410041841,
      "grad_norm": 0.091796875,
      "learning_rate": 2.4581589958159e-06,
      "loss": 0.8343,
      "step": 721
    },
    {
      "epoch": 0.7552301255230126,
      "grad_norm": 0.10205078125,
      "learning_rate": 2.4476987447698746e-06,
      "loss": 0.8872,
      "step": 722
    },
    {
      "epoch": 0.7562761506276151,
      "grad_norm": 0.1123046875,
      "learning_rate": 2.4372384937238497e-06,
      "loss": 0.8499,
      "step": 723
    },
    {
      "epoch": 0.7573221757322176,
      "grad_norm": 0.10107421875,
      "learning_rate": 2.4267782426778244e-06,
      "loss": 0.8413,
      "step": 724
    },
    {
      "epoch": 0.7583682008368201,
      "grad_norm": 0.08251953125,
      "learning_rate": 2.4163179916317995e-06,
      "loss": 0.7787,
      "step": 725
    },
    {
      "epoch": 0.7594142259414226,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.4058577405857742e-06,
      "loss": 0.8697,
      "step": 726
    },
    {
      "epoch": 0.7604602510460251,
      "grad_norm": 0.0908203125,
      "learning_rate": 2.395397489539749e-06,
      "loss": 0.8112,
      "step": 727
    },
    {
      "epoch": 0.7615062761506276,
      "grad_norm": 0.08642578125,
      "learning_rate": 2.384937238493724e-06,
      "loss": 0.7977,
      "step": 728
    },
    {
      "epoch": 0.7625523012552301,
      "grad_norm": 0.10546875,
      "learning_rate": 2.3744769874476988e-06,
      "loss": 0.862,
      "step": 729
    },
    {
      "epoch": 0.7635983263598326,
      "grad_norm": 0.11279296875,
      "learning_rate": 2.364016736401674e-06,
      "loss": 0.8344,
      "step": 730
    },
    {
      "epoch": 0.7646443514644351,
      "grad_norm": 0.1025390625,
      "learning_rate": 2.3535564853556486e-06,
      "loss": 0.9037,
      "step": 731
    },
    {
      "epoch": 0.7656903765690377,
      "grad_norm": 0.1083984375,
      "learning_rate": 2.3430962343096237e-06,
      "loss": 0.9096,
      "step": 732
    },
    {
      "epoch": 0.7667364016736402,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.3326359832635984e-06,
      "loss": 0.7703,
      "step": 733
    },
    {
      "epoch": 0.7677824267782427,
      "grad_norm": 0.087890625,
      "learning_rate": 2.322175732217573e-06,
      "loss": 0.8288,
      "step": 734
    },
    {
      "epoch": 0.7688284518828452,
      "grad_norm": 0.09326171875,
      "learning_rate": 2.3117154811715483e-06,
      "loss": 0.8307,
      "step": 735
    },
    {
      "epoch": 0.7698744769874477,
      "grad_norm": 0.1328125,
      "learning_rate": 2.301255230125523e-06,
      "loss": 0.8689,
      "step": 736
    },
    {
      "epoch": 0.7709205020920502,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.290794979079498e-06,
      "loss": 0.8924,
      "step": 737
    },
    {
      "epoch": 0.7719665271966527,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.2803347280334732e-06,
      "loss": 0.7704,
      "step": 738
    },
    {
      "epoch": 0.7730125523012552,
      "grad_norm": 0.10693359375,
      "learning_rate": 2.269874476987448e-06,
      "loss": 0.8947,
      "step": 739
    },
    {
      "epoch": 0.7740585774058577,
      "grad_norm": 0.1044921875,
      "learning_rate": 2.2594142259414227e-06,
      "loss": 0.8601,
      "step": 740
    },
    {
      "epoch": 0.7751046025104602,
      "grad_norm": 0.10107421875,
      "learning_rate": 2.2489539748953978e-06,
      "loss": 0.8755,
      "step": 741
    },
    {
      "epoch": 0.7761506276150628,
      "grad_norm": 0.0947265625,
      "learning_rate": 2.2384937238493725e-06,
      "loss": 0.8213,
      "step": 742
    },
    {
      "epoch": 0.7771966527196653,
      "grad_norm": 0.08642578125,
      "learning_rate": 2.2280334728033476e-06,
      "loss": 0.8635,
      "step": 743
    },
    {
      "epoch": 0.7782426778242678,
      "grad_norm": 0.0888671875,
      "learning_rate": 2.2175732217573223e-06,
      "loss": 0.8436,
      "step": 744
    },
    {
      "epoch": 0.7792887029288703,
      "grad_norm": 0.07958984375,
      "learning_rate": 2.2071129707112974e-06,
      "loss": 0.8184,
      "step": 745
    },
    {
      "epoch": 0.7803347280334728,
      "grad_norm": 0.091796875,
      "learning_rate": 2.196652719665272e-06,
      "loss": 0.8549,
      "step": 746
    },
    {
      "epoch": 0.7813807531380753,
      "grad_norm": 0.0966796875,
      "learning_rate": 2.186192468619247e-06,
      "loss": 0.858,
      "step": 747
    },
    {
      "epoch": 0.7824267782426778,
      "grad_norm": 0.09765625,
      "learning_rate": 2.175732217573222e-06,
      "loss": 0.8773,
      "step": 748
    },
    {
      "epoch": 0.7834728033472803,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.1652719665271967e-06,
      "loss": 0.8562,
      "step": 749
    },
    {
      "epoch": 0.7845188284518828,
      "grad_norm": 0.0947265625,
      "learning_rate": 2.154811715481172e-06,
      "loss": 0.8862,
      "step": 750
    },
    {
      "epoch": 0.7855648535564853,
      "grad_norm": 0.08984375,
      "learning_rate": 2.1443514644351465e-06,
      "loss": 0.8171,
      "step": 751
    },
    {
      "epoch": 0.7866108786610879,
      "grad_norm": 0.09375,
      "learning_rate": 2.1338912133891217e-06,
      "loss": 0.8458,
      "step": 752
    },
    {
      "epoch": 0.7876569037656904,
      "grad_norm": 0.0859375,
      "learning_rate": 2.1234309623430964e-06,
      "loss": 0.8185,
      "step": 753
    },
    {
      "epoch": 0.7887029288702929,
      "grad_norm": 0.1220703125,
      "learning_rate": 2.112970711297071e-06,
      "loss": 0.9272,
      "step": 754
    },
    {
      "epoch": 0.7897489539748954,
      "grad_norm": 0.09912109375,
      "learning_rate": 2.102510460251046e-06,
      "loss": 0.8551,
      "step": 755
    },
    {
      "epoch": 0.7907949790794979,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.092050209205021e-06,
      "loss": 0.8283,
      "step": 756
    },
    {
      "epoch": 0.7918410041841004,
      "grad_norm": 0.0908203125,
      "learning_rate": 2.081589958158996e-06,
      "loss": 0.8593,
      "step": 757
    },
    {
      "epoch": 0.7928870292887029,
      "grad_norm": 0.091796875,
      "learning_rate": 2.071129707112971e-06,
      "loss": 0.8549,
      "step": 758
    },
    {
      "epoch": 0.7939330543933054,
      "grad_norm": 0.1064453125,
      "learning_rate": 2.060669456066946e-06,
      "loss": 0.8906,
      "step": 759
    },
    {
      "epoch": 0.7949790794979079,
      "grad_norm": 0.11376953125,
      "learning_rate": 2.0502092050209206e-06,
      "loss": 0.8726,
      "step": 760
    },
    {
      "epoch": 0.7960251046025104,
      "grad_norm": 0.1181640625,
      "learning_rate": 2.0397489539748953e-06,
      "loss": 0.8883,
      "step": 761
    },
    {
      "epoch": 0.797071129707113,
      "grad_norm": 0.0830078125,
      "learning_rate": 2.0292887029288704e-06,
      "loss": 0.801,
      "step": 762
    },
    {
      "epoch": 0.7981171548117155,
      "grad_norm": 0.08447265625,
      "learning_rate": 2.0188284518828455e-06,
      "loss": 0.7928,
      "step": 763
    },
    {
      "epoch": 0.799163179916318,
      "grad_norm": 0.0947265625,
      "learning_rate": 2.0083682008368202e-06,
      "loss": 0.8533,
      "step": 764
    },
    {
      "epoch": 0.8002092050209205,
      "grad_norm": 0.1103515625,
      "learning_rate": 1.9979079497907954e-06,
      "loss": 0.8839,
      "step": 765
    },
    {
      "epoch": 0.801255230125523,
      "grad_norm": 0.1025390625,
      "learning_rate": 1.98744769874477e-06,
      "loss": 0.8619,
      "step": 766
    },
    {
      "epoch": 0.8023012552301255,
      "grad_norm": 0.091796875,
      "learning_rate": 1.9769874476987448e-06,
      "loss": 0.84,
      "step": 767
    },
    {
      "epoch": 0.803347280334728,
      "grad_norm": 0.08837890625,
      "learning_rate": 1.96652719665272e-06,
      "loss": 0.8604,
      "step": 768
    },
    {
      "epoch": 0.8043933054393305,
      "grad_norm": 0.10888671875,
      "learning_rate": 1.9560669456066946e-06,
      "loss": 0.9023,
      "step": 769
    },
    {
      "epoch": 0.805439330543933,
      "grad_norm": 0.10302734375,
      "learning_rate": 1.9456066945606697e-06,
      "loss": 0.8494,
      "step": 770
    },
    {
      "epoch": 0.8064853556485355,
      "grad_norm": 0.09130859375,
      "learning_rate": 1.9351464435146444e-06,
      "loss": 0.8274,
      "step": 771
    },
    {
      "epoch": 0.8075313807531381,
      "grad_norm": 0.09130859375,
      "learning_rate": 1.9246861924686196e-06,
      "loss": 0.8647,
      "step": 772
    },
    {
      "epoch": 0.8085774058577406,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.9142259414225943e-06,
      "loss": 0.8367,
      "step": 773
    },
    {
      "epoch": 0.8096234309623431,
      "grad_norm": 0.1142578125,
      "learning_rate": 1.9037656903765692e-06,
      "loss": 0.9087,
      "step": 774
    },
    {
      "epoch": 0.8106694560669456,
      "grad_norm": 0.10888671875,
      "learning_rate": 1.893305439330544e-06,
      "loss": 0.8926,
      "step": 775
    },
    {
      "epoch": 0.8117154811715481,
      "grad_norm": 0.095703125,
      "learning_rate": 1.8828451882845188e-06,
      "loss": 0.865,
      "step": 776
    },
    {
      "epoch": 0.8127615062761506,
      "grad_norm": 0.1083984375,
      "learning_rate": 1.872384937238494e-06,
      "loss": 0.8925,
      "step": 777
    },
    {
      "epoch": 0.8138075313807531,
      "grad_norm": 0.1064453125,
      "learning_rate": 1.8619246861924686e-06,
      "loss": 0.863,
      "step": 778
    },
    {
      "epoch": 0.8148535564853556,
      "grad_norm": 0.10400390625,
      "learning_rate": 1.8514644351464436e-06,
      "loss": 0.8554,
      "step": 779
    },
    {
      "epoch": 0.8158995815899581,
      "grad_norm": 0.10888671875,
      "learning_rate": 1.8410041841004187e-06,
      "loss": 0.8646,
      "step": 780
    },
    {
      "epoch": 0.8169456066945606,
      "grad_norm": 0.09619140625,
      "learning_rate": 1.8305439330543934e-06,
      "loss": 0.8549,
      "step": 781
    },
    {
      "epoch": 0.8179916317991632,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.8200836820083683e-06,
      "loss": 0.8667,
      "step": 782
    },
    {
      "epoch": 0.8190376569037657,
      "grad_norm": 0.1025390625,
      "learning_rate": 1.809623430962343e-06,
      "loss": 0.8541,
      "step": 783
    },
    {
      "epoch": 0.8200836820083682,
      "grad_norm": 0.10302734375,
      "learning_rate": 1.7991631799163181e-06,
      "loss": 0.852,
      "step": 784
    },
    {
      "epoch": 0.8211297071129707,
      "grad_norm": 0.1181640625,
      "learning_rate": 1.788702928870293e-06,
      "loss": 0.8856,
      "step": 785
    },
    {
      "epoch": 0.8221757322175732,
      "grad_norm": 0.11328125,
      "learning_rate": 1.7782426778242678e-06,
      "loss": 0.8414,
      "step": 786
    },
    {
      "epoch": 0.8232217573221757,
      "grad_norm": 0.0888671875,
      "learning_rate": 1.7677824267782429e-06,
      "loss": 0.8754,
      "step": 787
    },
    {
      "epoch": 0.8242677824267782,
      "grad_norm": 0.12060546875,
      "learning_rate": 1.7573221757322176e-06,
      "loss": 0.9196,
      "step": 788
    },
    {
      "epoch": 0.8253138075313807,
      "grad_norm": 0.109375,
      "learning_rate": 1.7468619246861925e-06,
      "loss": 0.8947,
      "step": 789
    },
    {
      "epoch": 0.8263598326359832,
      "grad_norm": 0.0947265625,
      "learning_rate": 1.7364016736401676e-06,
      "loss": 0.8553,
      "step": 790
    },
    {
      "epoch": 0.8274058577405857,
      "grad_norm": 0.091796875,
      "learning_rate": 1.7259414225941423e-06,
      "loss": 0.8004,
      "step": 791
    },
    {
      "epoch": 0.8284518828451883,
      "grad_norm": 0.111328125,
      "learning_rate": 1.7154811715481173e-06,
      "loss": 0.9118,
      "step": 792
    },
    {
      "epoch": 0.8294979079497908,
      "grad_norm": 0.0927734375,
      "learning_rate": 1.705020920502092e-06,
      "loss": 0.8325,
      "step": 793
    },
    {
      "epoch": 0.8305439330543933,
      "grad_norm": 0.111328125,
      "learning_rate": 1.694560669456067e-06,
      "loss": 0.927,
      "step": 794
    },
    {
      "epoch": 0.8315899581589958,
      "grad_norm": 0.09521484375,
      "learning_rate": 1.684100418410042e-06,
      "loss": 0.8621,
      "step": 795
    },
    {
      "epoch": 0.8326359832635983,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.6736401673640167e-06,
      "loss": 0.8545,
      "step": 796
    },
    {
      "epoch": 0.8336820083682008,
      "grad_norm": 0.1240234375,
      "learning_rate": 1.6631799163179918e-06,
      "loss": 0.8687,
      "step": 797
    },
    {
      "epoch": 0.8347280334728033,
      "grad_norm": 0.080078125,
      "learning_rate": 1.6527196652719666e-06,
      "loss": 0.8196,
      "step": 798
    },
    {
      "epoch": 0.8357740585774058,
      "grad_norm": 0.19140625,
      "learning_rate": 1.6422594142259415e-06,
      "loss": 0.9357,
      "step": 799
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 0.10107421875,
      "learning_rate": 1.6317991631799166e-06,
      "loss": 0.8787,
      "step": 800
    },
    {
      "epoch": 0.8378661087866108,
      "grad_norm": 0.109375,
      "learning_rate": 1.6213389121338913e-06,
      "loss": 0.8675,
      "step": 801
    },
    {
      "epoch": 0.8389121338912134,
      "grad_norm": 0.08740234375,
      "learning_rate": 1.6108786610878662e-06,
      "loss": 0.8483,
      "step": 802
    },
    {
      "epoch": 0.8399581589958159,
      "grad_norm": 0.09326171875,
      "learning_rate": 1.600418410041841e-06,
      "loss": 0.8057,
      "step": 803
    },
    {
      "epoch": 0.8410041841004184,
      "grad_norm": 0.0791015625,
      "learning_rate": 1.589958158995816e-06,
      "loss": 0.8434,
      "step": 804
    },
    {
      "epoch": 0.8420502092050209,
      "grad_norm": 0.08251953125,
      "learning_rate": 1.579497907949791e-06,
      "loss": 0.8234,
      "step": 805
    },
    {
      "epoch": 0.8430962343096234,
      "grad_norm": 0.1025390625,
      "learning_rate": 1.5690376569037657e-06,
      "loss": 0.8473,
      "step": 806
    },
    {
      "epoch": 0.8441422594142259,
      "grad_norm": 0.09228515625,
      "learning_rate": 1.5585774058577408e-06,
      "loss": 0.84,
      "step": 807
    },
    {
      "epoch": 0.8451882845188284,
      "grad_norm": 0.0888671875,
      "learning_rate": 1.5481171548117155e-06,
      "loss": 0.8447,
      "step": 808
    },
    {
      "epoch": 0.8462343096234309,
      "grad_norm": 0.08984375,
      "learning_rate": 1.5376569037656904e-06,
      "loss": 0.8484,
      "step": 809
    },
    {
      "epoch": 0.8472803347280334,
      "grad_norm": 0.08984375,
      "learning_rate": 1.5271966527196656e-06,
      "loss": 0.8372,
      "step": 810
    },
    {
      "epoch": 0.8483263598326359,
      "grad_norm": 0.09375,
      "learning_rate": 1.5167364016736403e-06,
      "loss": 0.8022,
      "step": 811
    },
    {
      "epoch": 0.8493723849372385,
      "grad_norm": 0.091796875,
      "learning_rate": 1.5062761506276152e-06,
      "loss": 0.8403,
      "step": 812
    },
    {
      "epoch": 0.850418410041841,
      "grad_norm": 0.10546875,
      "learning_rate": 1.4958158995815899e-06,
      "loss": 0.8424,
      "step": 813
    },
    {
      "epoch": 0.8514644351464435,
      "grad_norm": 0.10302734375,
      "learning_rate": 1.485355648535565e-06,
      "loss": 0.8855,
      "step": 814
    },
    {
      "epoch": 0.852510460251046,
      "grad_norm": 0.1064453125,
      "learning_rate": 1.47489539748954e-06,
      "loss": 0.8923,
      "step": 815
    },
    {
      "epoch": 0.8535564853556485,
      "grad_norm": 0.0830078125,
      "learning_rate": 1.4644351464435146e-06,
      "loss": 0.8027,
      "step": 816
    },
    {
      "epoch": 0.854602510460251,
      "grad_norm": 0.08544921875,
      "learning_rate": 1.4539748953974898e-06,
      "loss": 0.8362,
      "step": 817
    },
    {
      "epoch": 0.8556485355648535,
      "grad_norm": 0.09375,
      "learning_rate": 1.4435146443514645e-06,
      "loss": 0.8504,
      "step": 818
    },
    {
      "epoch": 0.856694560669456,
      "grad_norm": 0.0888671875,
      "learning_rate": 1.4330543933054394e-06,
      "loss": 0.8441,
      "step": 819
    },
    {
      "epoch": 0.8577405857740585,
      "grad_norm": 0.09765625,
      "learning_rate": 1.4225941422594145e-06,
      "loss": 0.8415,
      "step": 820
    },
    {
      "epoch": 0.858786610878661,
      "grad_norm": 0.0986328125,
      "learning_rate": 1.4121338912133892e-06,
      "loss": 0.8819,
      "step": 821
    },
    {
      "epoch": 0.8598326359832636,
      "grad_norm": 0.09765625,
      "learning_rate": 1.4016736401673641e-06,
      "loss": 0.8536,
      "step": 822
    },
    {
      "epoch": 0.8608786610878661,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.3912133891213388e-06,
      "loss": 0.8666,
      "step": 823
    },
    {
      "epoch": 0.8619246861924686,
      "grad_norm": 0.1318359375,
      "learning_rate": 1.380753138075314e-06,
      "loss": 0.9692,
      "step": 824
    },
    {
      "epoch": 0.8629707112970711,
      "grad_norm": 0.095703125,
      "learning_rate": 1.3702928870292889e-06,
      "loss": 0.8637,
      "step": 825
    },
    {
      "epoch": 0.8640167364016736,
      "grad_norm": 0.09326171875,
      "learning_rate": 1.3598326359832636e-06,
      "loss": 0.8329,
      "step": 826
    },
    {
      "epoch": 0.8650627615062761,
      "grad_norm": 0.11376953125,
      "learning_rate": 1.3493723849372387e-06,
      "loss": 0.9496,
      "step": 827
    },
    {
      "epoch": 0.8661087866108786,
      "grad_norm": 0.09228515625,
      "learning_rate": 1.3389121338912134e-06,
      "loss": 0.8308,
      "step": 828
    },
    {
      "epoch": 0.8671548117154811,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.3284518828451883e-06,
      "loss": 0.8752,
      "step": 829
    },
    {
      "epoch": 0.8682008368200836,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.3179916317991635e-06,
      "loss": 0.8354,
      "step": 830
    },
    {
      "epoch": 0.8692468619246861,
      "grad_norm": 0.1005859375,
      "learning_rate": 1.3075313807531382e-06,
      "loss": 0.8603,
      "step": 831
    },
    {
      "epoch": 0.8702928870292888,
      "grad_norm": 0.0927734375,
      "learning_rate": 1.297071129707113e-06,
      "loss": 0.8431,
      "step": 832
    },
    {
      "epoch": 0.8713389121338913,
      "grad_norm": 0.091796875,
      "learning_rate": 1.2866108786610878e-06,
      "loss": 0.849,
      "step": 833
    },
    {
      "epoch": 0.8723849372384938,
      "grad_norm": 0.11376953125,
      "learning_rate": 1.276150627615063e-06,
      "loss": 0.8878,
      "step": 834
    },
    {
      "epoch": 0.8734309623430963,
      "grad_norm": 0.0908203125,
      "learning_rate": 1.2656903765690378e-06,
      "loss": 0.8557,
      "step": 835
    },
    {
      "epoch": 0.8744769874476988,
      "grad_norm": 0.1044921875,
      "learning_rate": 1.2552301255230125e-06,
      "loss": 0.858,
      "step": 836
    },
    {
      "epoch": 0.8755230125523012,
      "grad_norm": 0.10791015625,
      "learning_rate": 1.2447698744769877e-06,
      "loss": 0.888,
      "step": 837
    },
    {
      "epoch": 0.8765690376569037,
      "grad_norm": 0.09814453125,
      "learning_rate": 1.2343096234309624e-06,
      "loss": 0.8368,
      "step": 838
    },
    {
      "epoch": 0.8776150627615062,
      "grad_norm": 0.1015625,
      "learning_rate": 1.2238493723849373e-06,
      "loss": 0.8634,
      "step": 839
    },
    {
      "epoch": 0.8786610878661087,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.2133891213389122e-06,
      "loss": 0.85,
      "step": 840
    },
    {
      "epoch": 0.8797071129707112,
      "grad_norm": 0.10302734375,
      "learning_rate": 1.2029288702928871e-06,
      "loss": 0.9003,
      "step": 841
    },
    {
      "epoch": 0.8807531380753139,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.192468619246862e-06,
      "loss": 0.8719,
      "step": 842
    },
    {
      "epoch": 0.8817991631799164,
      "grad_norm": 0.10498046875,
      "learning_rate": 1.182008368200837e-06,
      "loss": 0.8888,
      "step": 843
    },
    {
      "epoch": 0.8828451882845189,
      "grad_norm": 0.09521484375,
      "learning_rate": 1.1715481171548119e-06,
      "loss": 0.8276,
      "step": 844
    },
    {
      "epoch": 0.8838912133891214,
      "grad_norm": 0.12353515625,
      "learning_rate": 1.1610878661087866e-06,
      "loss": 0.8328,
      "step": 845
    },
    {
      "epoch": 0.8849372384937239,
      "grad_norm": 0.11376953125,
      "learning_rate": 1.1506276150627615e-06,
      "loss": 0.8723,
      "step": 846
    },
    {
      "epoch": 0.8859832635983264,
      "grad_norm": 0.1181640625,
      "learning_rate": 1.1401673640167366e-06,
      "loss": 0.9427,
      "step": 847
    },
    {
      "epoch": 0.8870292887029289,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.1297071129707113e-06,
      "loss": 0.879,
      "step": 848
    },
    {
      "epoch": 0.8880753138075314,
      "grad_norm": 0.1025390625,
      "learning_rate": 1.1192468619246862e-06,
      "loss": 0.8537,
      "step": 849
    },
    {
      "epoch": 0.8891213389121339,
      "grad_norm": 0.11962890625,
      "learning_rate": 1.1087866108786612e-06,
      "loss": 0.878,
      "step": 850
    },
    {
      "epoch": 0.8901673640167364,
      "grad_norm": 0.1142578125,
      "learning_rate": 1.098326359832636e-06,
      "loss": 0.8979,
      "step": 851
    },
    {
      "epoch": 0.891213389121339,
      "grad_norm": 0.1015625,
      "learning_rate": 1.087866108786611e-06,
      "loss": 0.8928,
      "step": 852
    },
    {
      "epoch": 0.8922594142259415,
      "grad_norm": 0.08740234375,
      "learning_rate": 1.077405857740586e-06,
      "loss": 0.828,
      "step": 853
    },
    {
      "epoch": 0.893305439330544,
      "grad_norm": 0.09326171875,
      "learning_rate": 1.0669456066945608e-06,
      "loss": 0.8874,
      "step": 854
    },
    {
      "epoch": 0.8943514644351465,
      "grad_norm": 0.0947265625,
      "learning_rate": 1.0564853556485355e-06,
      "loss": 0.8917,
      "step": 855
    },
    {
      "epoch": 0.895397489539749,
      "grad_norm": 0.10009765625,
      "learning_rate": 1.0460251046025104e-06,
      "loss": 0.892,
      "step": 856
    },
    {
      "epoch": 0.8964435146443515,
      "grad_norm": 0.11279296875,
      "learning_rate": 1.0355648535564856e-06,
      "loss": 0.8623,
      "step": 857
    },
    {
      "epoch": 0.897489539748954,
      "grad_norm": 0.11474609375,
      "learning_rate": 1.0251046025104603e-06,
      "loss": 0.911,
      "step": 858
    },
    {
      "epoch": 0.8985355648535565,
      "grad_norm": 0.103515625,
      "learning_rate": 1.0146443514644352e-06,
      "loss": 0.8716,
      "step": 859
    },
    {
      "epoch": 0.899581589958159,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.0041841004184101e-06,
      "loss": 0.8211,
      "step": 860
    },
    {
      "epoch": 0.9006276150627615,
      "grad_norm": 0.1103515625,
      "learning_rate": 9.93723849372385e-07,
      "loss": 0.9194,
      "step": 861
    },
    {
      "epoch": 0.9016736401673641,
      "grad_norm": 0.0966796875,
      "learning_rate": 9.8326359832636e-07,
      "loss": 0.8673,
      "step": 862
    },
    {
      "epoch": 0.9027196652719666,
      "grad_norm": 0.0869140625,
      "learning_rate": 9.728033472803349e-07,
      "loss": 0.8245,
      "step": 863
    },
    {
      "epoch": 0.9037656903765691,
      "grad_norm": 0.10400390625,
      "learning_rate": 9.623430962343098e-07,
      "loss": 0.821,
      "step": 864
    },
    {
      "epoch": 0.9048117154811716,
      "grad_norm": 0.10546875,
      "learning_rate": 9.518828451882846e-07,
      "loss": 0.879,
      "step": 865
    },
    {
      "epoch": 0.9058577405857741,
      "grad_norm": 0.10302734375,
      "learning_rate": 9.414225941422594e-07,
      "loss": 0.852,
      "step": 866
    },
    {
      "epoch": 0.9069037656903766,
      "grad_norm": 0.0947265625,
      "learning_rate": 9.309623430962343e-07,
      "loss": 0.8116,
      "step": 867
    },
    {
      "epoch": 0.9079497907949791,
      "grad_norm": 0.09765625,
      "learning_rate": 9.205020920502093e-07,
      "loss": 0.8576,
      "step": 868
    },
    {
      "epoch": 0.9089958158995816,
      "grad_norm": 0.107421875,
      "learning_rate": 9.100418410041842e-07,
      "loss": 0.8879,
      "step": 869
    },
    {
      "epoch": 0.9100418410041841,
      "grad_norm": 0.111328125,
      "learning_rate": 8.995815899581591e-07,
      "loss": 0.8837,
      "step": 870
    },
    {
      "epoch": 0.9110878661087866,
      "grad_norm": 0.09326171875,
      "learning_rate": 8.891213389121339e-07,
      "loss": 0.8464,
      "step": 871
    },
    {
      "epoch": 0.9121338912133892,
      "grad_norm": 0.09521484375,
      "learning_rate": 8.786610878661088e-07,
      "loss": 0.8495,
      "step": 872
    },
    {
      "epoch": 0.9131799163179917,
      "grad_norm": 0.11767578125,
      "learning_rate": 8.682008368200838e-07,
      "loss": 0.9009,
      "step": 873
    },
    {
      "epoch": 0.9142259414225942,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.577405857740586e-07,
      "loss": 0.9155,
      "step": 874
    },
    {
      "epoch": 0.9152719665271967,
      "grad_norm": 0.0947265625,
      "learning_rate": 8.472803347280335e-07,
      "loss": 0.8656,
      "step": 875
    },
    {
      "epoch": 0.9163179916317992,
      "grad_norm": 0.1044921875,
      "learning_rate": 8.368200836820084e-07,
      "loss": 0.854,
      "step": 876
    },
    {
      "epoch": 0.9173640167364017,
      "grad_norm": 0.10009765625,
      "learning_rate": 8.263598326359833e-07,
      "loss": 0.9088,
      "step": 877
    },
    {
      "epoch": 0.9184100418410042,
      "grad_norm": 0.08642578125,
      "learning_rate": 8.158995815899583e-07,
      "loss": 0.827,
      "step": 878
    },
    {
      "epoch": 0.9194560669456067,
      "grad_norm": 0.11572265625,
      "learning_rate": 8.054393305439331e-07,
      "loss": 0.8497,
      "step": 879
    },
    {
      "epoch": 0.9205020920502092,
      "grad_norm": 0.12109375,
      "learning_rate": 7.94979079497908e-07,
      "loss": 0.949,
      "step": 880
    },
    {
      "epoch": 0.9215481171548117,
      "grad_norm": 0.11328125,
      "learning_rate": 7.845188284518828e-07,
      "loss": 0.8879,
      "step": 881
    },
    {
      "epoch": 0.9225941422594143,
      "grad_norm": 0.09912109375,
      "learning_rate": 7.740585774058578e-07,
      "loss": 0.8189,
      "step": 882
    },
    {
      "epoch": 0.9236401673640168,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.635983263598328e-07,
      "loss": 0.8702,
      "step": 883
    },
    {
      "epoch": 0.9246861924686193,
      "grad_norm": 0.1083984375,
      "learning_rate": 7.531380753138076e-07,
      "loss": 0.8866,
      "step": 884
    },
    {
      "epoch": 0.9257322175732218,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.426778242677825e-07,
      "loss": 0.8771,
      "step": 885
    },
    {
      "epoch": 0.9267782426778243,
      "grad_norm": 0.11181640625,
      "learning_rate": 7.322175732217573e-07,
      "loss": 0.8748,
      "step": 886
    },
    {
      "epoch": 0.9278242677824268,
      "grad_norm": 0.10107421875,
      "learning_rate": 7.217573221757322e-07,
      "loss": 0.8465,
      "step": 887
    },
    {
      "epoch": 0.9288702928870293,
      "grad_norm": 0.0927734375,
      "learning_rate": 7.112970711297073e-07,
      "loss": 0.8548,
      "step": 888
    },
    {
      "epoch": 0.9299163179916318,
      "grad_norm": 0.099609375,
      "learning_rate": 7.008368200836821e-07,
      "loss": 0.8595,
      "step": 889
    },
    {
      "epoch": 0.9309623430962343,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.90376569037657e-07,
      "loss": 0.8246,
      "step": 890
    },
    {
      "epoch": 0.9320083682008368,
      "grad_norm": 0.1220703125,
      "learning_rate": 6.799163179916318e-07,
      "loss": 0.8459,
      "step": 891
    },
    {
      "epoch": 0.9330543933054394,
      "grad_norm": 0.1064453125,
      "learning_rate": 6.694560669456067e-07,
      "loss": 0.8714,
      "step": 892
    },
    {
      "epoch": 0.9341004184100419,
      "grad_norm": 0.09716796875,
      "learning_rate": 6.589958158995817e-07,
      "loss": 0.8102,
      "step": 893
    },
    {
      "epoch": 0.9351464435146444,
      "grad_norm": 0.0927734375,
      "learning_rate": 6.485355648535565e-07,
      "loss": 0.8503,
      "step": 894
    },
    {
      "epoch": 0.9361924686192469,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.380753138075315e-07,
      "loss": 0.8772,
      "step": 895
    },
    {
      "epoch": 0.9372384937238494,
      "grad_norm": 0.10791015625,
      "learning_rate": 6.276150627615063e-07,
      "loss": 0.8826,
      "step": 896
    },
    {
      "epoch": 0.9382845188284519,
      "grad_norm": 0.1025390625,
      "learning_rate": 6.171548117154812e-07,
      "loss": 0.8729,
      "step": 897
    },
    {
      "epoch": 0.9393305439330544,
      "grad_norm": 0.09326171875,
      "learning_rate": 6.066945606694561e-07,
      "loss": 0.8303,
      "step": 898
    },
    {
      "epoch": 0.9403765690376569,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.96234309623431e-07,
      "loss": 0.9165,
      "step": 899
    },
    {
      "epoch": 0.9414225941422594,
      "grad_norm": 0.1123046875,
      "learning_rate": 5.857740585774059e-07,
      "loss": 0.8438,
      "step": 900
    },
    {
      "epoch": 0.9424686192468619,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.753138075313807e-07,
      "loss": 0.8579,
      "step": 901
    },
    {
      "epoch": 0.9435146443514645,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.648535564853557e-07,
      "loss": 0.8644,
      "step": 902
    },
    {
      "epoch": 0.944560669456067,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.543933054393306e-07,
      "loss": 0.8297,
      "step": 903
    },
    {
      "epoch": 0.9456066945606695,
      "grad_norm": 0.1201171875,
      "learning_rate": 5.439330543933055e-07,
      "loss": 0.8491,
      "step": 904
    },
    {
      "epoch": 0.946652719665272,
      "grad_norm": 0.0869140625,
      "learning_rate": 5.334728033472804e-07,
      "loss": 0.8045,
      "step": 905
    },
    {
      "epoch": 0.9476987447698745,
      "grad_norm": 0.1064453125,
      "learning_rate": 5.230125523012552e-07,
      "loss": 0.854,
      "step": 906
    },
    {
      "epoch": 0.948744769874477,
      "grad_norm": 0.11376953125,
      "learning_rate": 5.125523012552301e-07,
      "loss": 0.8927,
      "step": 907
    },
    {
      "epoch": 0.9497907949790795,
      "grad_norm": 0.091796875,
      "learning_rate": 5.020920502092051e-07,
      "loss": 0.8389,
      "step": 908
    },
    {
      "epoch": 0.950836820083682,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.9163179916318e-07,
      "loss": 0.8362,
      "step": 909
    },
    {
      "epoch": 0.9518828451882845,
      "grad_norm": 0.1328125,
      "learning_rate": 4.811715481171549e-07,
      "loss": 0.9468,
      "step": 910
    },
    {
      "epoch": 0.952928870292887,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.707112970711297e-07,
      "loss": 0.8647,
      "step": 911
    },
    {
      "epoch": 0.9539748953974896,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.6025104602510467e-07,
      "loss": 0.8814,
      "step": 912
    },
    {
      "epoch": 0.9550209205020921,
      "grad_norm": 0.0908203125,
      "learning_rate": 4.4979079497907954e-07,
      "loss": 0.8166,
      "step": 913
    },
    {
      "epoch": 0.9560669456066946,
      "grad_norm": 0.1025390625,
      "learning_rate": 4.393305439330544e-07,
      "loss": 0.8781,
      "step": 914
    },
    {
      "epoch": 0.9571129707112971,
      "grad_norm": 0.09619140625,
      "learning_rate": 4.288702928870293e-07,
      "loss": 0.8205,
      "step": 915
    },
    {
      "epoch": 0.9581589958158996,
      "grad_norm": 0.1005859375,
      "learning_rate": 4.184100418410042e-07,
      "loss": 0.8635,
      "step": 916
    },
    {
      "epoch": 0.9592050209205021,
      "grad_norm": 0.0986328125,
      "learning_rate": 4.0794979079497915e-07,
      "loss": 0.8748,
      "step": 917
    },
    {
      "epoch": 0.9602510460251046,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.97489539748954e-07,
      "loss": 0.8512,
      "step": 918
    },
    {
      "epoch": 0.9612970711297071,
      "grad_norm": 0.09033203125,
      "learning_rate": 3.870292887029289e-07,
      "loss": 0.8284,
      "step": 919
    },
    {
      "epoch": 0.9623430962343096,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.765690376569038e-07,
      "loss": 0.839,
      "step": 920
    },
    {
      "epoch": 0.9633891213389121,
      "grad_norm": 0.1064453125,
      "learning_rate": 3.6610878661087866e-07,
      "loss": 0.8857,
      "step": 921
    },
    {
      "epoch": 0.9644351464435147,
      "grad_norm": 0.12158203125,
      "learning_rate": 3.5564853556485363e-07,
      "loss": 0.9004,
      "step": 922
    },
    {
      "epoch": 0.9654811715481172,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.451882845188285e-07,
      "loss": 0.8857,
      "step": 923
    },
    {
      "epoch": 0.9665271966527197,
      "grad_norm": 0.1142578125,
      "learning_rate": 3.3472803347280335e-07,
      "loss": 0.89,
      "step": 924
    },
    {
      "epoch": 0.9675732217573222,
      "grad_norm": 0.08203125,
      "learning_rate": 3.2426778242677827e-07,
      "loss": 0.8128,
      "step": 925
    },
    {
      "epoch": 0.9686192468619247,
      "grad_norm": 0.1064453125,
      "learning_rate": 3.1380753138075313e-07,
      "loss": 0.8726,
      "step": 926
    },
    {
      "epoch": 0.9696652719665272,
      "grad_norm": 0.1015625,
      "learning_rate": 3.0334728033472805e-07,
      "loss": 0.8422,
      "step": 927
    },
    {
      "epoch": 0.9707112970711297,
      "grad_norm": 0.103515625,
      "learning_rate": 2.9288702928870297e-07,
      "loss": 0.8852,
      "step": 928
    },
    {
      "epoch": 0.9717573221757322,
      "grad_norm": 0.1064453125,
      "learning_rate": 2.8242677824267783e-07,
      "loss": 0.8818,
      "step": 929
    },
    {
      "epoch": 0.9728033472803347,
      "grad_norm": 0.111328125,
      "learning_rate": 2.7196652719665275e-07,
      "loss": 0.8707,
      "step": 930
    },
    {
      "epoch": 0.9738493723849372,
      "grad_norm": 0.10888671875,
      "learning_rate": 2.615062761506276e-07,
      "loss": 0.832,
      "step": 931
    },
    {
      "epoch": 0.9748953974895398,
      "grad_norm": 0.09814453125,
      "learning_rate": 2.5104602510460253e-07,
      "loss": 0.8398,
      "step": 932
    },
    {
      "epoch": 0.9759414225941423,
      "grad_norm": 0.1025390625,
      "learning_rate": 2.4058577405857745e-07,
      "loss": 0.8542,
      "step": 933
    },
    {
      "epoch": 0.9769874476987448,
      "grad_norm": 0.09912109375,
      "learning_rate": 2.3012552301255234e-07,
      "loss": 0.8708,
      "step": 934
    },
    {
      "epoch": 0.9780334728033473,
      "grad_norm": 0.10205078125,
      "learning_rate": 2.196652719665272e-07,
      "loss": 0.8673,
      "step": 935
    },
    {
      "epoch": 0.9790794979079498,
      "grad_norm": 0.10888671875,
      "learning_rate": 2.092050209205021e-07,
      "loss": 0.9128,
      "step": 936
    },
    {
      "epoch": 0.9801255230125523,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.98744769874477e-07,
      "loss": 0.8649,
      "step": 937
    },
    {
      "epoch": 0.9811715481171548,
      "grad_norm": 0.11474609375,
      "learning_rate": 1.882845188284519e-07,
      "loss": 0.894,
      "step": 938
    },
    {
      "epoch": 0.9822175732217573,
      "grad_norm": 0.10302734375,
      "learning_rate": 1.7782426778242681e-07,
      "loss": 0.8683,
      "step": 939
    },
    {
      "epoch": 0.9832635983263598,
      "grad_norm": 0.1328125,
      "learning_rate": 1.6736401673640168e-07,
      "loss": 0.9564,
      "step": 940
    },
    {
      "epoch": 0.9843096234309623,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.5690376569037657e-07,
      "loss": 0.855,
      "step": 941
    },
    {
      "epoch": 0.9853556485355649,
      "grad_norm": 0.107421875,
      "learning_rate": 1.4644351464435148e-07,
      "loss": 0.8664,
      "step": 942
    },
    {
      "epoch": 0.9864016736401674,
      "grad_norm": 0.09375,
      "learning_rate": 1.3598326359832637e-07,
      "loss": 0.8839,
      "step": 943
    },
    {
      "epoch": 0.9874476987447699,
      "grad_norm": 0.1005859375,
      "learning_rate": 1.2552301255230126e-07,
      "loss": 0.8598,
      "step": 944
    },
    {
      "epoch": 0.9884937238493724,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.1506276150627617e-07,
      "loss": 0.8796,
      "step": 945
    },
    {
      "epoch": 0.9895397489539749,
      "grad_norm": 0.0869140625,
      "learning_rate": 1.0460251046025104e-07,
      "loss": 0.8252,
      "step": 946
    },
    {
      "epoch": 0.9905857740585774,
      "grad_norm": 0.10400390625,
      "learning_rate": 9.414225941422595e-08,
      "loss": 0.8631,
      "step": 947
    },
    {
      "epoch": 0.9916317991631799,
      "grad_norm": 0.115234375,
      "learning_rate": 8.368200836820084e-08,
      "loss": 0.9105,
      "step": 948
    },
    {
      "epoch": 0.9926778242677824,
      "grad_norm": 0.1064453125,
      "learning_rate": 7.322175732217574e-08,
      "loss": 0.8963,
      "step": 949
    },
    {
      "epoch": 0.9937238493723849,
      "grad_norm": 0.10205078125,
      "learning_rate": 6.276150627615063e-08,
      "loss": 0.8922,
      "step": 950
    },
    {
      "epoch": 0.9947698744769874,
      "grad_norm": 0.1259765625,
      "learning_rate": 5.230125523012552e-08,
      "loss": 0.9242,
      "step": 951
    },
    {
      "epoch": 0.99581589958159,
      "grad_norm": 0.10986328125,
      "learning_rate": 4.184100418410042e-08,
      "loss": 0.871,
      "step": 952
    },
    {
      "epoch": 0.9968619246861925,
      "grad_norm": 0.1240234375,
      "learning_rate": 3.1380753138075316e-08,
      "loss": 0.912,
      "step": 953
    },
    {
      "epoch": 0.997907949790795,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.092050209205021e-08,
      "loss": 0.8917,
      "step": 954
    },
    {
      "epoch": 0.9989539748953975,
      "grad_norm": 0.083984375,
      "learning_rate": 1.0460251046025105e-08,
      "loss": 0.8154,
      "step": 955
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0,
      "loss": 0.7838,
      "step": 956
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 956,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.084254995575472e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}