{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.0,
  "global_step": 616,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 12.2148,
      "step": 1
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.105263157894737e-06,
      "loss": 12.0312,
      "step": 2
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.157894736842105e-06,
      "loss": 12.3086,
      "step": 3
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.210526315789474e-06,
      "loss": 12.1172,
      "step": 4
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.263157894736842e-06,
      "loss": 12.0117,
      "step": 5
    },
    {
      "epoch": 0.08,
      "learning_rate": 6.31578947368421e-06,
      "loss": 12.2656,
      "step": 6
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.368421052631579e-06,
      "loss": 12.125,
      "step": 7
    },
    {
      "epoch": 0.1,
      "learning_rate": 8.421052631578948e-06,
      "loss": 11.2266,
      "step": 8
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.473684210526315e-06,
      "loss": 11.1523,
      "step": 9
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 9.5234,
      "step": 10
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1578947368421053e-05,
      "loss": 9.4688,
      "step": 11
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.263157894736842e-05,
      "loss": 9.25,
      "step": 12
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.3684210526315791e-05,
      "loss": 7.7285,
      "step": 13
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.4736842105263159e-05,
      "loss": 7.6367,
      "step": 14
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.578947368421053e-05,
      "loss": 7.4844,
      "step": 15
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.6842105263157896e-05,
      "loss": 7.2422,
      "step": 16
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7894736842105264e-05,
      "loss": 7.0938,
      "step": 17
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.894736842105263e-05,
      "loss": 6.7266,
      "step": 18
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 6.5234,
      "step": 19
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9999861541352416e-05,
      "loss": 6.3477,
      "step": 20
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9999446169243816e-05,
      "loss": 6.127,
      "step": 21
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9998753895176576e-05,
      "loss": 5.8555,
      "step": 22
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.999778473832096e-05,
      "loss": 5.7402,
      "step": 23
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9996538725514597e-05,
      "loss": 5.5605,
      "step": 24
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.999501589126174e-05,
      "loss": 5.4199,
      "step": 25
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9993216277732302e-05,
      "loss": 5.3242,
      "step": 26
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.999113993476069e-05,
      "loss": 5.2148,
      "step": 27
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9988786919844437e-05,
      "loss": 5.1016,
      "step": 28
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9986157298142595e-05,
      "loss": 5.0488,
      "step": 29
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9983251142473935e-05,
      "loss": 4.9258,
      "step": 30
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9980068533314937e-05,
      "loss": 4.9531,
      "step": 31
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9976609558797545e-05,
      "loss": 4.8535,
      "step": 32
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9972874314706755e-05,
      "loss": 4.8203,
      "step": 33
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9968862904477936e-05,
      "loss": 4.8535,
      "step": 34
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9964575439193966e-05,
      "loss": 4.7168,
      "step": 35
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.996001203758218e-05,
      "loss": 4.6875,
      "step": 36
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.995517282601106e-05,
      "loss": 4.6172,
      "step": 37
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9950057938486745e-05,
      "loss": 4.6523,
      "step": 38
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.994466751664932e-05,
      "loss": 4.5195,
      "step": 39
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.993900170976888e-05,
      "loss": 4.5117,
      "step": 40
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9933060674741422e-05,
      "loss": 4.4141,
      "step": 41
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9926844576084483e-05,
      "loss": 4.3398,
      "step": 42
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.992035358593258e-05,
      "loss": 4.3232,
      "step": 43
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.991358788403246e-05,
      "loss": 4.2305,
      "step": 44
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.990654765773811e-05,
      "loss": 4.1641,
      "step": 45
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.9899233102005573e-05,
      "loss": 4.0674,
      "step": 46
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9891644419387545e-05,
      "loss": 3.915,
      "step": 47
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9883781820027777e-05,
      "loss": 3.7822,
      "step": 48
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.987564552165524e-05,
      "loss": 3.709,
      "step": 49
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.9867235749578108e-05,
      "loss": 3.4131,
      "step": 50
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.9858552736677516e-05,
      "loss": 3.1318,
      "step": 51
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.984959672340111e-05,
      "loss": 2.834,
      "step": 52
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.984036795775638e-05,
      "loss": 2.5654,
      "step": 53
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.9830866695303817e-05,
      "loss": 2.417,
      "step": 54
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.9821093199149806e-05,
      "loss": 2.1909,
      "step": 55
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.981104773993936e-05,
      "loss": 2.2568,
      "step": 56
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.980073059584862e-05,
      "loss": 2.2744,
      "step": 57
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.9790142052577148e-05,
      "loss": 2.0771,
      "step": 58
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.977928240334002e-05,
      "loss": 2.1729,
      "step": 59
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.9768151948859705e-05,
      "loss": 2.123,
      "step": 60
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.9756750997357738e-05,
      "loss": 2.0356,
      "step": 61
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9745079864546184e-05,
      "loss": 2.0142,
      "step": 62
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.97331388736189e-05,
      "loss": 2.061,
      "step": 63
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.972092835524257e-05,
      "loss": 2.0508,
      "step": 64
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.9708448647547575e-05,
      "loss": 2.0171,
      "step": 65
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.9695700096118594e-05,
      "loss": 2.1284,
      "step": 66
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.9682683053985073e-05,
      "loss": 2.0166,
      "step": 67
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.966939788161142e-05,
      "loss": 2.062,
      "step": 68
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9655844946887035e-05,
      "loss": 2.0142,
      "step": 69
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.9642024625116117e-05,
      "loss": 2.0103,
      "step": 70
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.9627937299007286e-05,
      "loss": 1.9956,
      "step": 71
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.961358335866296e-05,
      "loss": 1.9868,
      "step": 72
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.959896320156857e-05,
      "loss": 2.0435,
      "step": 73
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.958407723258156e-05,
      "loss": 2.0112,
      "step": 74
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9568925863920155e-05,
      "loss": 2.0908,
      "step": 75
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.955350951515195e-05,
      "loss": 1.9795,
      "step": 76
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9537828613182314e-05,
      "loss": 2.0112,
      "step": 77
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.9521883592242537e-05,
      "loss": 2.0459,
      "step": 78
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.950567489387783e-05,
      "loss": 2.0117,
      "step": 79
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.9489202966935084e-05,
      "loss": 2.0156,
      "step": 80
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.947246826755044e-05,
      "loss": 2.0547,
      "step": 81
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.945547125913667e-05,
      "loss": 1.9639,
      "step": 82
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.943821241237034e-05,
      "loss": 2.019,
      "step": 83
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.9420692205178753e-05,
      "loss": 1.9771,
      "step": 84
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.9402911122726756e-05,
      "loss": 1.9492,
      "step": 85
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.9384869657403277e-05,
      "loss": 1.9702,
      "step": 86
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.9366568308807685e-05,
      "loss": 1.9946,
      "step": 87
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.9348007583735985e-05,
      "loss": 1.9854,
      "step": 88
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.9329187996166747e-05,
      "loss": 1.959,
      "step": 89
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.9310110067246905e-05,
      "loss": 1.9722,
      "step": 90
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.9290774325277305e-05,
      "loss": 2.0376,
      "step": 91
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.9271181305698084e-05,
      "loss": 1.9834,
      "step": 92
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.9251331551073843e-05,
      "loss": 2.0049,
      "step": 93
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.923122561107861e-05,
      "loss": 1.9824,
      "step": 94
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.9210864042480645e-05,
      "loss": 1.9624,
      "step": 95
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.9190247409126993e-05,
      "loss": 1.9395,
      "step": 96
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.916937628192789e-05,
      "loss": 1.9746,
      "step": 97
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.9148251238840947e-05,
      "loss": 1.9507,
      "step": 98
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.9126872864855142e-05,
      "loss": 2.0054,
      "step": 99
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.9105241751974624e-05,
      "loss": 1.9409,
      "step": 100
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.9083358499202323e-05,
      "loss": 1.9912,
      "step": 101
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.9061223712523352e-05,
      "loss": 1.9404,
      "step": 102
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.903883800488824e-05,
      "loss": 1.9102,
      "step": 103
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.9016201996195943e-05,
      "loss": 1.9248,
      "step": 104
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.8993316313276694e-05,
      "loss": 1.8984,
      "step": 105
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.8970181589874637e-05,
      "loss": 1.9331,
      "step": 106
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.894679846663027e-05,
      "loss": 1.9561,
      "step": 107
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.8923167591062723e-05,
      "loss": 1.8901,
      "step": 108
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.8899289617551803e-05,
      "loss": 1.9922,
      "step": 109
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8875165207319902e-05,
      "loss": 1.9277,
      "step": 110
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8850795028413658e-05,
      "loss": 1.9185,
      "step": 111
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.882617975568547e-05,
      "loss": 1.9575,
      "step": 112
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.880132007077482e-05,
      "loss": 1.957,
      "step": 113
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.8776216662089373e-05,
      "loss": 1.8984,
      "step": 114
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.875087022478594e-05,
      "loss": 1.9429,
      "step": 115
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.8725281460751198e-05,
      "loss": 1.8701,
      "step": 116
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.869945107858228e-05,
      "loss": 1.9497,
      "step": 117
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.867337979356715e-05,
      "loss": 1.8921,
      "step": 118
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.8647068327664774e-05,
      "loss": 1.8569,
      "step": 119
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.8620517409485148e-05,
      "loss": 1.8882,
      "step": 120
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.8593727774269122e-05,
      "loss": 1.8765,
      "step": 121
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.8566700163868027e-05,
      "loss": 1.9282,
      "step": 122
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.8539435326723135e-05,
      "loss": 1.8384,
      "step": 123
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.851193401784495e-05,
      "loss": 1.9185,
      "step": 124
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.848419699879227e-05,
      "loss": 1.834,
      "step": 125
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.845622503765113e-05,
      "loss": 1.8657,
      "step": 126
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.842801890901351e-05,
      "loss": 1.8457,
      "step": 127
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.8399579393955893e-05,
      "loss": 1.7671,
      "step": 128
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.837090728001764e-05,
      "loss": 1.8462,
      "step": 129
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.834200336117918e-05,
      "loss": 1.8296,
      "step": 130
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.8312868437840002e-05,
      "loss": 1.8262,
      "step": 131
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.8283503316796536e-05,
      "loss": 1.835,
      "step": 132
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.8253908811219764e-05,
      "loss": 1.8979,
      "step": 133
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.822408574063273e-05,
      "loss": 1.8496,
      "step": 134
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.8194034930887842e-05,
      "loss": 1.8252,
      "step": 135
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.8163757214143993e-05,
      "loss": 1.7812,
      "step": 136
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.8133253428843524e-05,
      "loss": 1.8364,
      "step": 137
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.810252441968901e-05,
      "loss": 1.8013,
      "step": 138
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.8071571037619856e-05,
      "loss": 1.8203,
      "step": 139
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.804039413978875e-05,
      "loss": 1.7729,
      "step": 140
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.8008994589537913e-05,
      "loss": 1.8491,
      "step": 141
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.7977373256375194e-05,
      "loss": 1.7998,
      "step": 142
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.7945531015950008e-05,
      "loss": 1.8364,
      "step": 143
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.791346875002905e-05,
      "loss": 1.8125,
      "step": 144
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.7881187346471924e-05,
      "loss": 1.832,
      "step": 145
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.784868769920653e-05,
      "loss": 1.8271,
      "step": 146
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7815970708204296e-05,
      "loss": 1.7959,
      "step": 147
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.77830372794553e-05,
      "loss": 1.7798,
      "step": 148
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.774988832494314e-05,
      "loss": 1.7651,
      "step": 149
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7716524762619695e-05,
      "loss": 1.8076,
      "step": 150
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7682947516379706e-05,
      "loss": 1.8379,
      "step": 151
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7649157516035205e-05,
      "loss": 1.8228,
      "step": 152
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.7615155697289734e-05,
      "loss": 1.7783,
      "step": 153
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.7580943001712457e-05,
      "loss": 1.8188,
      "step": 154
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.7546520376712093e-05,
      "loss": 1.7974,
      "step": 155
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.7511888775510662e-05,
      "loss": 1.7964,
      "step": 156
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.7477049157117093e-05,
      "loss": 1.7515,
      "step": 157
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.744200248630068e-05,
      "loss": 1.7725,
      "step": 158
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.7406749733564344e-05,
      "loss": 1.7534,
      "step": 159
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.737129187511779e-05,
      "loss": 1.8408,
      "step": 160
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.7335629892850436e-05,
      "loss": 1.7686,
      "step": 161
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.729976477430425e-05,
      "loss": 1.7642,
      "step": 162
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.7263697512646397e-05,
      "loss": 1.8047,
      "step": 163
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.7227429106641726e-05,
      "loss": 1.8301,
      "step": 164
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.7190960560625127e-05,
      "loss": 1.7588,
      "step": 165
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.7154292884473712e-05,
      "loss": 1.7749,
      "step": 166
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.711742709357886e-05,
      "loss": 1.7251,
      "step": 167
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.708036420881807e-05,
      "loss": 1.7603,
      "step": 168
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7043105256526723e-05,
      "loss": 1.7339,
      "step": 169
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.7005651268469652e-05,
      "loss": 1.731,
      "step": 170
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6968003281812563e-05,
      "loss": 1.7598,
      "step": 171
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.693016233909332e-05,
      "loss": 1.7007,
      "step": 172
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.689212948819307e-05,
      "loss": 1.7183,
      "step": 173
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.6853905782307235e-05,
      "loss": 1.7173,
      "step": 174
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.681549227991634e-05,
      "loss": 1.7856,
      "step": 175
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.67768900447567e-05,
      "loss": 1.7329,
      "step": 176
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.6738100145790977e-05,
      "loss": 1.7578,
      "step": 177
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.6699123657178553e-05,
      "loss": 1.6846,
      "step": 178
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.6659961658245813e-05,
      "loss": 1.791,
      "step": 179
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.6620615233456235e-05,
      "loss": 1.7798,
      "step": 180
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.658108547238038e-05,
      "loss": 1.6987,
      "step": 181
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.6541373469665688e-05,
      "loss": 1.7202,
      "step": 182
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.6501480325006206e-05,
      "loss": 1.7285,
      "step": 183
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.64614071431121e-05,
      "loss": 1.7417,
      "step": 184
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.6421155033679085e-05,
      "loss": 1.79,
      "step": 185
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.6380725111357693e-05,
      "loss": 1.7876,
      "step": 186
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.634011849572239e-05,
      "loss": 1.7734,
      "step": 187
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.6299336311240593e-05,
      "loss": 1.7686,
      "step": 188
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.6258379687241533e-05,
      "loss": 1.7993,
      "step": 189
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6217249757884954e-05,
      "loss": 1.708,
      "step": 190
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6175947662129735e-05,
      "loss": 1.7065,
      "step": 191
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.6134474543702353e-05,
      "loss": 1.7324,
      "step": 192
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.609283155106517e-05,
      "loss": 1.7686,
      "step": 193
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.605101983738468e-05,
      "loss": 1.7563,
      "step": 194
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.6009040560499548e-05,
      "loss": 1.7373,
      "step": 195
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.596689488288856e-05,
      "loss": 1.7104,
      "step": 196
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.5924583971638416e-05,
      "loss": 1.7368,
      "step": 197
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.5882108998411427e-05,
      "loss": 1.7886,
      "step": 198
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.5839471139413065e-05,
      "loss": 1.6855,
      "step": 199
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5796671575359382e-05,
      "loss": 1.7158,
      "step": 200
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5753711491444336e-05,
      "loss": 1.7144,
      "step": 201
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.571059207730695e-05,
      "loss": 1.6909,
      "step": 202
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.5667314526998373e-05,
      "loss": 1.8003,
      "step": 203
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.5623880038948828e-05,
      "loss": 1.7231,
      "step": 204
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.55802898159344e-05,
      "loss": 1.6816,
      "step": 205
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.553654506504377e-05,
      "loss": 1.6826,
      "step": 206
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.5492646997644737e-05,
      "loss": 1.7085,
      "step": 207
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.5448596829350706e-05,
      "loss": 1.6797,
      "step": 208
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.540439577998703e-05,
      "loss": 1.708,
      "step": 209
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.5360045073557214e-05,
      "loss": 1.7036,
      "step": 210
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.5315545938209016e-05,
      "loss": 1.7129,
      "step": 211
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.527089960620046e-05,
      "loss": 1.6855,
      "step": 212
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.5226107313865701e-05,
      "loss": 1.645,
      "step": 213
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.5181170301580776e-05,
      "loss": 1.6982,
      "step": 214
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.5136089813729276e-05,
      "loss": 1.731,
      "step": 215
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.509086709866788e-05,
      "loss": 1.7192,
      "step": 216
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.5045503408691776e-05,
      "loss": 1.6982,
      "step": 217
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.7266,
      "step": 218
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.495435813266064e-05,
      "loss": 1.6958,
      "step": 219
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.4908579070575936e-05,
      "loss": 1.7056,
      "step": 220
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.4862664081447297e-05,
      "loss": 1.6943,
      "step": 221
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.4816614436740184e-05,
      "loss": 1.6724,
      "step": 222
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.4770431411648898e-05,
      "loss": 1.6641,
      "step": 223
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4724116285061278e-05,
      "loss": 1.7461,
      "step": 224
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.4677670339523285e-05,
      "loss": 1.7207,
      "step": 225
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.4631094861203478e-05,
      "loss": 1.7061,
      "step": 226
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.4584391139857407e-05,
      "loss": 1.6758,
      "step": 227
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.4537560468791889e-05,
      "loss": 1.73,
      "step": 228
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.4490604144829204e-05,
      "loss": 1.7314,
      "step": 229
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.4443523468271168e-05,
      "loss": 1.7114,
      "step": 230
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.4396319742863145e-05,
      "loss": 1.7212,
      "step": 231
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.4348994275757933e-05,
      "loss": 1.7036,
      "step": 232
    },
    {
      "epoch": 3.03,
      "learning_rate": 1.4301548377479562e-05,
      "loss": 1.71,
      "step": 233
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.4253983361887017e-05,
      "loss": 1.7432,
      "step": 234
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.4206300546137844e-05,
      "loss": 1.6992,
      "step": 235
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.415850125065168e-05,
      "loss": 1.7271,
      "step": 236
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.4110586799073684e-05,
      "loss": 1.6792,
      "step": 237
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.4062558518237893e-05,
      "loss": 1.73,
      "step": 238
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.4014417738130464e-05,
      "loss": 1.7192,
      "step": 239
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.3966165791852862e-05,
      "loss": 1.7476,
      "step": 240
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.3917804015584932e-05,
      "loss": 1.6958,
      "step": 241
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.3869333748547901e-05,
      "loss": 1.6865,
      "step": 242
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.3820756332967294e-05,
      "loss": 1.668,
      "step": 243
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.3772073114035762e-05,
      "loss": 1.6826,
      "step": 244
    },
    {
      "epoch": 3.18,
      "learning_rate": 1.3723285439875836e-05,
      "loss": 1.7227,
      "step": 245
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.3674394661502595e-05,
      "loss": 1.7163,
      "step": 246
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.3625402132786247e-05,
      "loss": 1.6606,
      "step": 247
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.3576309210414646e-05,
      "loss": 1.7085,
      "step": 248
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.352711725385572e-05,
      "loss": 1.668,
      "step": 249
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.3477827625319826e-05,
      "loss": 1.7173,
      "step": 250
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.3428441689722023e-05,
      "loss": 1.7656,
      "step": 251
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.3378960814644283e-05,
      "loss": 1.6812,
      "step": 252
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.3329386370297615e-05,
      "loss": 1.6953,
      "step": 253
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.3279719729484117e-05,
      "loss": 1.665,
      "step": 254
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.3229962267558982e-05,
      "loss": 1.6587,
      "step": 255
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.3180115362392383e-05,
      "loss": 1.6797,
      "step": 256
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.3130180394331335e-05,
      "loss": 1.6992,
      "step": 257
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.3080158746161468e-05,
      "loss": 1.6567,
      "step": 258
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.3030051803068729e-05,
      "loss": 1.6641,
      "step": 259
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.2979860952601038e-05,
      "loss": 1.6841,
      "step": 260
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.2929587584629845e-05,
      "loss": 1.6777,
      "step": 261
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.2879233091311667e-05,
      "loss": 1.7065,
      "step": 262
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.2828798867049504e-05,
      "loss": 1.6997,
      "step": 263
    },
    {
      "epoch": 3.43,
      "learning_rate": 1.2778286308454255e-05,
      "loss": 1.6704,
      "step": 264
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.2727696814306034e-05,
      "loss": 1.6489,
      "step": 265
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.2677031785515423e-05,
      "loss": 1.6777,
      "step": 266
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.26262926250847e-05,
      "loss": 1.6284,
      "step": 267
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.2575480738068971e-05,
      "loss": 1.6445,
      "step": 268
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.2524597531537261e-05,
      "loss": 1.626,
      "step": 269
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.2473644414533573e-05,
      "loss": 1.626,
      "step": 270
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.2422622798037833e-05,
      "loss": 1.6919,
      "step": 271
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.2371534094926852e-05,
      "loss": 1.6602,
      "step": 272
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.232037971993517e-05,
      "loss": 1.6401,
      "step": 273
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.2269161089615902e-05,
      "loss": 1.7026,
      "step": 274
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2217879622301514e-05,
      "loss": 1.6875,
      "step": 275
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2166536738064523e-05,
      "loss": 1.6646,
      "step": 276
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.2115133858678192e-05,
      "loss": 1.6631,
      "step": 277
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.2063672407577154e-05,
      "loss": 1.6196,
      "step": 278
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.2012153809817992e-05,
      "loss": 1.6606,
      "step": 279
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.1960579492039783e-05,
      "loss": 1.6719,
      "step": 280
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.1908950882424581e-05,
      "loss": 1.6958,
      "step": 281
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.1857269410657883e-05,
      "loss": 1.645,
      "step": 282
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.1805536507889021e-05,
      "loss": 1.6782,
      "step": 283
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.1753753606691554e-05,
      "loss": 1.6724,
      "step": 284
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.1701922141023566e-05,
      "loss": 1.6108,
      "step": 285
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.1650043546187994e-05,
      "loss": 1.6313,
      "step": 286
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.1598119258792848e-05,
      "loss": 1.647,
      "step": 287
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.1546150716711448e-05,
      "loss": 1.6816,
      "step": 288
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.1494139359042612e-05,
      "loss": 1.6846,
      "step": 289
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.1442086626070781e-05,
      "loss": 1.6602,
      "step": 290
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.1389993959226163e-05,
      "loss": 1.6133,
      "step": 291
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.1337862801044792e-05,
      "loss": 1.6997,
      "step": 292
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.1285694595128606e-05,
      "loss": 1.6172,
      "step": 293
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.123349078610545e-05,
      "loss": 1.6479,
      "step": 294
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1181252819589081e-05,
      "loss": 1.6851,
      "step": 295
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.1128982142139142e-05,
      "loss": 1.6533,
      "step": 296
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.1076680201221093e-05,
      "loss": 1.6367,
      "step": 297
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.1024348445166133e-05,
      "loss": 1.6426,
      "step": 298
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.0971988323131099e-05,
      "loss": 1.6509,
      "step": 299
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.091960128505833e-05,
      "loss": 1.6997,
      "step": 300
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.086718878163551e-05,
      "loss": 1.6187,
      "step": 301
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.0814752264255508e-05,
      "loss": 1.6914,
      "step": 302
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.0762293184976178e-05,
      "loss": 1.6421,
      "step": 303
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.070981299648016e-05,
      "loss": 1.6631,
      "step": 304
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.0657313152034634e-05,
      "loss": 1.7046,
      "step": 305
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.0604795105451096e-05,
      "loss": 1.5845,
      "step": 306
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.0552260311045082e-05,
      "loss": 1.6621,
      "step": 307
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.0499710223595913e-05,
      "loss": 1.6782,
      "step": 308
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.0447146298306394e-05,
      "loss": 1.6611,
      "step": 309
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.0394569990762528e-05,
      "loss": 1.6626,
      "step": 310
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.0341982756893203e-05,
      "loss": 1.6406,
      "step": 311
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.0289386052929874e-05,
      "loss": 1.6455,
      "step": 312
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.0236781335366239e-05,
      "loss": 1.7051,
      "step": 313
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.0184170060917914e-05,
      "loss": 1.5967,
      "step": 314
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.0131553686482077e-05,
      "loss": 1.6772,
      "step": 315
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.0078933669097135e-05,
      "loss": 1.625,
      "step": 316
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.002631146590238e-05,
      "loss": 1.6572,
      "step": 317
    },
    {
      "epoch": 4.13,
      "learning_rate": 9.973688534097624e-06,
      "loss": 1.6694,
      "step": 318
    },
    {
      "epoch": 4.14,
      "learning_rate": 9.92106633090287e-06,
      "loss": 1.6377,
      "step": 319
    },
    {
      "epoch": 4.16,
      "learning_rate": 9.868446313517927e-06,
      "loss": 1.6782,
      "step": 320
    },
    {
      "epoch": 4.17,
      "learning_rate": 9.815829939082087e-06,
      "loss": 1.6147,
      "step": 321
    },
    {
      "epoch": 4.18,
      "learning_rate": 9.763218664633763e-06,
      "loss": 1.6826,
      "step": 322
    },
    {
      "epoch": 4.19,
      "learning_rate": 9.710613947070127e-06,
      "loss": 1.7041,
      "step": 323
    },
    {
      "epoch": 4.21,
      "learning_rate": 9.658017243106802e-06,
      "loss": 1.6343,
      "step": 324
    },
    {
      "epoch": 4.22,
      "learning_rate": 9.605430009237474e-06,
      "loss": 1.6724,
      "step": 325
    },
    {
      "epoch": 4.23,
      "learning_rate": 9.552853701693606e-06,
      "loss": 1.6812,
      "step": 326
    },
    {
      "epoch": 4.25,
      "learning_rate": 9.50028977640409e-06,
      "loss": 1.6289,
      "step": 327
    },
    {
      "epoch": 4.26,
      "learning_rate": 9.44773968895492e-06,
      "loss": 1.6313,
      "step": 328
    },
    {
      "epoch": 4.27,
      "learning_rate": 9.395204894548907e-06,
      "loss": 1.6274,
      "step": 329
    },
    {
      "epoch": 4.29,
      "learning_rate": 9.342686847965367e-06,
      "loss": 1.6572,
      "step": 330
    },
    {
      "epoch": 4.3,
      "learning_rate": 9.290187003519841e-06,
      "loss": 1.6333,
      "step": 331
    },
    {
      "epoch": 4.31,
      "learning_rate": 9.237706815023824e-06,
      "loss": 1.687,
      "step": 332
    },
    {
      "epoch": 4.32,
      "learning_rate": 9.185247735744495e-06,
      "loss": 1.6626,
      "step": 333
    },
    {
      "epoch": 4.34,
      "learning_rate": 9.132811218364494e-06,
      "loss": 1.6431,
      "step": 334
    },
    {
      "epoch": 4.35,
      "learning_rate": 9.080398714941672e-06,
      "loss": 1.6562,
      "step": 335
    },
    {
      "epoch": 4.36,
      "learning_rate": 9.028011676868901e-06,
      "loss": 1.6714,
      "step": 336
    },
    {
      "epoch": 4.38,
      "learning_rate": 8.975651554833869e-06,
      "loss": 1.604,
      "step": 337
    },
    {
      "epoch": 4.39,
      "learning_rate": 8.92331979877891e-06,
      "loss": 1.6719,
      "step": 338
    },
    {
      "epoch": 4.4,
      "learning_rate": 8.871017857860863e-06,
      "loss": 1.707,
      "step": 339
    },
    {
      "epoch": 4.42,
      "learning_rate": 8.81874718041092e-06,
      "loss": 1.647,
      "step": 340
    },
    {
      "epoch": 4.43,
      "learning_rate": 8.766509213894552e-06,
      "loss": 1.6675,
      "step": 341
    },
    {
      "epoch": 4.44,
      "learning_rate": 8.714305404871397e-06,
      "loss": 1.6636,
      "step": 342
    },
    {
      "epoch": 4.45,
      "learning_rate": 8.662137198955211e-06,
      "loss": 1.6768,
      "step": 343
    },
    {
      "epoch": 4.47,
      "learning_rate": 8.610006040773844e-06,
      "loss": 1.5864,
      "step": 344
    },
    {
      "epoch": 4.48,
      "learning_rate": 8.557913373929222e-06,
      "loss": 1.6304,
      "step": 345
    },
    {
      "epoch": 4.49,
      "learning_rate": 8.50586064095739e-06,
      "loss": 1.6289,
      "step": 346
    },
    {
      "epoch": 4.51,
      "learning_rate": 8.453849283288554e-06,
      "loss": 1.6436,
      "step": 347
    },
    {
      "epoch": 4.52,
      "learning_rate": 8.401880741207155e-06,
      "loss": 1.6221,
      "step": 348
    },
    {
      "epoch": 4.53,
      "learning_rate": 8.349956453812009e-06,
      "loss": 1.6904,
      "step": 349
    },
    {
      "epoch": 4.55,
      "learning_rate": 8.298077858976435e-06,
      "loss": 1.5898,
      "step": 350
    },
    {
      "epoch": 4.56,
      "learning_rate": 8.246246393308448e-06,
      "loss": 1.667,
      "step": 351
    },
    {
      "epoch": 4.57,
      "learning_rate": 8.194463492110982e-06,
      "loss": 1.6543,
      "step": 352
    },
    {
      "epoch": 4.58,
      "learning_rate": 8.142730589342119e-06,
      "loss": 1.6572,
      "step": 353
    },
    {
      "epoch": 4.6,
      "learning_rate": 8.091049117575424e-06,
      "loss": 1.6685,
      "step": 354
    },
    {
      "epoch": 4.61,
      "learning_rate": 8.03942050796022e-06,
      "loss": 1.6484,
      "step": 355
    },
    {
      "epoch": 4.62,
      "learning_rate": 7.98784619018201e-06,
      "loss": 1.5405,
      "step": 356
    },
    {
      "epoch": 4.64,
      "learning_rate": 7.93632759242285e-06,
      "loss": 1.644,
      "step": 357
    },
    {
      "epoch": 4.65,
      "learning_rate": 7.884866141321811e-06,
      "loss": 1.6206,
      "step": 358
    },
    {
      "epoch": 4.66,
      "learning_rate": 7.833463261935482e-06,
      "loss": 1.6079,
      "step": 359
    },
    {
      "epoch": 4.68,
      "learning_rate": 7.782120377698489e-06,
      "loss": 1.6108,
      "step": 360
    },
    {
      "epoch": 4.69,
      "learning_rate": 7.730838910384098e-06,
      "loss": 1.5625,
      "step": 361
    },
    {
      "epoch": 4.7,
      "learning_rate": 7.679620280064837e-06,
      "loss": 1.647,
      "step": 362
    },
    {
      "epoch": 4.71,
      "learning_rate": 7.6284659050731525e-06,
      "loss": 1.5493,
      "step": 363
    },
    {
      "epoch": 4.73,
      "learning_rate": 7.57737720196217e-06,
      "loss": 1.6362,
      "step": 364
    },
    {
      "epoch": 4.74,
      "learning_rate": 7.526355585466432e-06,
      "loss": 1.6294,
      "step": 365
    },
    {
      "epoch": 4.75,
      "learning_rate": 7.4754024684627405e-06,
      "loss": 1.6675,
      "step": 366
    },
    {
      "epoch": 4.77,
      "learning_rate": 7.424519261931036e-06,
      "loss": 1.6519,
      "step": 367
    },
    {
      "epoch": 4.78,
      "learning_rate": 7.373707374915303e-06,
      "loss": 1.6807,
      "step": 368
    },
    {
      "epoch": 4.79,
      "learning_rate": 7.322968214484583e-06,
      "loss": 1.6221,
      "step": 369
    },
    {
      "epoch": 4.81,
      "learning_rate": 7.27230318569397e-06,
      "loss": 1.6523,
      "step": 370
    },
    {
      "epoch": 4.82,
      "learning_rate": 7.221713691545746e-06,
      "loss": 1.6118,
      "step": 371
    },
    {
      "epoch": 4.83,
      "learning_rate": 7.171201132950502e-06,
      "loss": 1.6279,
      "step": 372
    },
    {
      "epoch": 4.84,
      "learning_rate": 7.1207669086883366e-06,
      "loss": 1.6416,
      "step": 373
    },
    {
      "epoch": 4.86,
      "learning_rate": 7.070412415370158e-06,
      "loss": 1.605,
      "step": 374
    },
    {
      "epoch": 4.87,
      "learning_rate": 7.020139047398966e-06,
      "loss": 1.627,
      "step": 375
    },
    {
      "epoch": 4.88,
      "learning_rate": 6.969948196931272e-06,
      "loss": 1.6123,
      "step": 376
    },
    {
      "epoch": 4.9,
      "learning_rate": 6.919841253838537e-06,
      "loss": 1.6333,
      "step": 377
    },
    {
      "epoch": 4.91,
      "learning_rate": 6.869819605668669e-06,
      "loss": 1.5981,
      "step": 378
    },
    {
      "epoch": 4.92,
      "learning_rate": 6.819884637607619e-06,
      "loss": 1.646,
      "step": 379
    },
    {
      "epoch": 4.94,
      "learning_rate": 6.770037732441019e-06,
      "loss": 1.6641,
      "step": 380
    },
    {
      "epoch": 4.95,
      "learning_rate": 6.720280270515882e-06,
      "loss": 1.6362,
      "step": 381
    },
    {
      "epoch": 4.96,
      "learning_rate": 6.670613629702391e-06,
      "loss": 1.6562,
      "step": 382
    },
    {
      "epoch": 4.97,
      "learning_rate": 6.62103918535572e-06,
      "loss": 1.6772,
      "step": 383
    },
    {
      "epoch": 4.99,
      "learning_rate": 6.5715583102779815e-06,
      "loss": 1.6729,
      "step": 384
    },
    {
      "epoch": 5.0,
      "learning_rate": 6.522172374680177e-06,
      "loss": 1.6597,
      "step": 385
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.472882746144282e-06,
      "loss": 1.6348,
      "step": 386
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.423690789585359e-06,
      "loss": 1.6108,
      "step": 387
    },
    {
      "epoch": 5.04,
      "learning_rate": 6.374597867213756e-06,
      "loss": 1.6421,
      "step": 388
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.3256053384974105e-06,
      "loss": 1.6455,
      "step": 389
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.276714560124166e-06,
      "loss": 1.6616,
      "step": 390
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.2279268859642396e-06,
      "loss": 1.6162,
      "step": 391
    },
    {
      "epoch": 5.09,
      "learning_rate": 6.179243667032709e-06,
      "loss": 1.6646,
      "step": 392
    },
    {
      "epoch": 5.1,
      "learning_rate": 6.130666251452102e-06,
      "loss": 1.6445,
      "step": 393
    },
    {
      "epoch": 5.12,
      "learning_rate": 6.082195984415069e-06,
      "loss": 1.6299,
      "step": 394
    },
    {
      "epoch": 5.13,
      "learning_rate": 6.03383420814714e-06,
      "loss": 1.6221,
      "step": 395
    },
    {
      "epoch": 5.14,
      "learning_rate": 5.9855822618695385e-06,
      "loss": 1.647,
      "step": 396
    },
    {
      "epoch": 5.16,
      "learning_rate": 5.937441481762112e-06,
      "loss": 1.6147,
      "step": 397
    },
    {
      "epoch": 5.17,
      "learning_rate": 5.889413200926317e-06,
      "loss": 1.6025,
      "step": 398
    },
    {
      "epoch": 5.18,
      "learning_rate": 5.841498749348322e-06,
      "loss": 1.6064,
      "step": 399
    },
    {
      "epoch": 5.19,
      "learning_rate": 5.793699453862161e-06,
      "loss": 1.6587,
      "step": 400
    },
    {
      "epoch": 5.21,
      "learning_rate": 5.746016638112986e-06,
      "loss": 1.6255,
      "step": 401
    },
    {
      "epoch": 5.22,
      "learning_rate": 5.698451622520442e-06,
      "loss": 1.6523,
      "step": 402
    },
    {
      "epoch": 5.23,
      "learning_rate": 5.651005724242072e-06,
      "loss": 1.6367,
      "step": 403
    },
    {
      "epoch": 5.25,
      "learning_rate": 5.603680257136857e-06,
      "loss": 1.6006,
      "step": 404
    },
    {
      "epoch": 5.26,
      "learning_rate": 5.556476531728836e-06,
      "loss": 1.6294,
      "step": 405
    },
    {
      "epoch": 5.27,
      "learning_rate": 5.509395855170798e-06,
      "loss": 1.6284,
      "step": 406
    },
    {
      "epoch": 5.29,
      "learning_rate": 5.4624395312081125e-06,
      "loss": 1.6392,
      "step": 407
    },
    {
      "epoch": 5.3,
      "learning_rate": 5.415608860142593e-06,
      "loss": 1.625,
      "step": 408
    },
    {
      "epoch": 5.31,
      "learning_rate": 5.368905138796523e-06,
      "loss": 1.6162,
      "step": 409
    },
    {
      "epoch": 5.32,
      "learning_rate": 5.322329660476715e-06,
      "loss": 1.5752,
      "step": 410
    },
    {
      "epoch": 5.34,
      "learning_rate": 5.275883714938726e-06,
      "loss": 1.6655,
      "step": 411
    },
    {
      "epoch": 5.35,
      "learning_rate": 5.2295685883511086e-06,
      "loss": 1.5972,
      "step": 412
    },
    {
      "epoch": 5.36,
      "learning_rate": 5.183385563259819e-06,
      "loss": 1.6421,
      "step": 413
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.137335918552702e-06,
      "loss": 1.5869,
      "step": 414
    },
    {
      "epoch": 5.39,
      "learning_rate": 5.091420929424065e-06,
      "loss": 1.6333,
      "step": 415
    },
    {
      "epoch": 5.4,
      "learning_rate": 5.045641867339361e-06,
      "loss": 1.6445,
      "step": 416
    },
    {
      "epoch": 5.42,
      "learning_rate": 5.000000000000003e-06,
      "loss": 1.6597,
      "step": 417
    },
    {
      "epoch": 5.43,
      "learning_rate": 4.954496591308227e-06,
      "loss": 1.6387,
      "step": 418
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.909132901332122e-06,
      "loss": 1.6489,
      "step": 419
    },
    {
      "epoch": 5.45,
      "learning_rate": 4.863910186270726e-06,
      "loss": 1.6318,
      "step": 420
    },
    {
      "epoch": 5.47,
      "learning_rate": 4.818829698419225e-06,
      "loss": 1.6841,
      "step": 421
    },
    {
      "epoch": 5.48,
      "learning_rate": 4.773892686134301e-06,
      "loss": 1.666,
      "step": 422
    },
    {
      "epoch": 5.49,
      "learning_rate": 4.729100393799538e-06,
      "loss": 1.6162,
      "step": 423
    },
    {
      "epoch": 5.51,
      "learning_rate": 4.684454061790987e-06,
      "loss": 1.5957,
      "step": 424
    },
    {
      "epoch": 5.52,
      "learning_rate": 4.639954926442792e-06,
      "loss": 1.6201,
      "step": 425
    },
    {
      "epoch": 5.53,
      "learning_rate": 4.5956042200129725e-06,
      "loss": 1.6533,
      "step": 426
    },
    {
      "epoch": 5.55,
      "learning_rate": 4.551403170649299e-06,
      "loss": 1.624,
      "step": 427
    },
    {
      "epoch": 5.56,
      "learning_rate": 4.507353002355269e-06,
      "loss": 1.604,
      "step": 428
    },
    {
      "epoch": 5.57,
      "learning_rate": 4.4634549349562315e-06,
      "loss": 1.6089,
      "step": 429
    },
    {
      "epoch": 5.58,
      "learning_rate": 4.4197101840656e-06,
      "loss": 1.5962,
      "step": 430
    },
    {
      "epoch": 5.6,
      "learning_rate": 4.376119961051175e-06,
      "loss": 1.5962,
      "step": 431
    },
    {
      "epoch": 5.61,
      "learning_rate": 4.33268547300163e-06,
      "loss": 1.6313,
      "step": 432
    },
    {
      "epoch": 5.62,
      "learning_rate": 4.289407922693053e-06,
      "loss": 1.6626,
      "step": 433
    },
    {
      "epoch": 5.64,
      "learning_rate": 4.2462885085556635e-06,
      "loss": 1.5796,
      "step": 434
    },
    {
      "epoch": 5.65,
      "learning_rate": 4.203328424640619e-06,
      "loss": 1.6836,
      "step": 435
    },
    {
      "epoch": 5.66,
      "learning_rate": 4.1605288605869365e-06,
      "loss": 1.6675,
      "step": 436
    },
    {
      "epoch": 5.68,
      "learning_rate": 4.117891001588574e-06,
      "loss": 1.6807,
      "step": 437
    },
    {
      "epoch": 5.69,
      "learning_rate": 4.075416028361584e-06,
      "loss": 1.6167,
      "step": 438
    },
    {
      "epoch": 5.7,
      "learning_rate": 4.033105117111441e-06,
      "loss": 1.6851,
      "step": 439
    },
    {
      "epoch": 5.71,
      "learning_rate": 3.9909594395004545e-06,
      "loss": 1.6191,
      "step": 440
    },
    {
      "epoch": 5.73,
      "learning_rate": 3.948980162615323e-06,
      "loss": 1.6362,
      "step": 441
    },
    {
      "epoch": 5.74,
      "learning_rate": 3.907168448934836e-06,
      "loss": 1.5825,
      "step": 442
    },
    {
      "epoch": 5.75,
      "learning_rate": 3.865525456297652e-06,
      "loss": 1.6182,
      "step": 443
    },
    {
      "epoch": 5.77,
      "learning_rate": 3.824052337870263e-06,
      "loss": 1.5908,
      "step": 444
    },
    {
      "epoch": 5.78,
      "learning_rate": 3.7827502421150497e-06,
      "loss": 1.6162,
      "step": 445
    },
    {
      "epoch": 5.79,
      "learning_rate": 3.741620312758469e-06,
      "loss": 1.6021,
      "step": 446
    },
    {
      "epoch": 5.81,
      "learning_rate": 3.7006636887594095e-06,
      "loss": 1.6479,
      "step": 447
    },
    {
      "epoch": 5.82,
      "learning_rate": 3.6598815042776135e-06,
      "loss": 1.6294,
      "step": 448
    },
    {
      "epoch": 5.83,
      "learning_rate": 3.619274888642309e-06,
      "loss": 1.6914,
      "step": 449
    },
    {
      "epoch": 5.84,
      "learning_rate": 3.578844966320917e-06,
      "loss": 1.6226,
      "step": 450
    },
    {
      "epoch": 5.86,
      "learning_rate": 3.5385928568879012e-06,
      "loss": 1.6196,
      "step": 451
    },
    {
      "epoch": 5.87,
      "learning_rate": 3.4985196749937976e-06,
      "loss": 1.5977,
      "step": 452
    },
    {
      "epoch": 5.88,
      "learning_rate": 3.458626530334316e-06,
      "loss": 1.5786,
      "step": 453
    },
    {
      "epoch": 5.9,
      "learning_rate": 3.4189145276196244e-06,
      "loss": 1.6113,
      "step": 454
    },
    {
      "epoch": 5.91,
      "learning_rate": 3.3793847665437674e-06,
      "loss": 1.6025,
      "step": 455
    },
    {
      "epoch": 5.92,
      "learning_rate": 3.340038341754189e-06,
      "loss": 1.6191,
      "step": 456
    },
    {
      "epoch": 5.94,
      "learning_rate": 3.300876342821451e-06,
      "loss": 1.604,
      "step": 457
    },
    {
      "epoch": 5.95,
      "learning_rate": 3.2618998542090263e-06,
      "loss": 1.6274,
      "step": 458
    },
    {
      "epoch": 5.96,
      "learning_rate": 3.2231099552433e-06,
      "loss": 1.6543,
      "step": 459
    },
    {
      "epoch": 5.97,
      "learning_rate": 3.1845077200836638e-06,
      "loss": 1.6265,
      "step": 460
    },
    {
      "epoch": 5.99,
      "learning_rate": 3.1460942176927666e-06,
      "loss": 1.6123,
      "step": 461
    },
    {
      "epoch": 6.0,
      "learning_rate": 3.107870511806934e-06,
      "loss": 1.6401,
      "step": 462
    },
    {
      "epoch": 6.01,
      "learning_rate": 3.0698376609066828e-06,
      "loss": 1.6094,
      "step": 463
    },
    {
      "epoch": 6.03,
      "learning_rate": 3.0319967181874366e-06,
      "loss": 1.5859,
      "step": 464
    },
    {
      "epoch": 6.04,
      "learning_rate": 2.9943487315303486e-06,
      "loss": 1.6182,
      "step": 465
    },
    {
      "epoch": 6.05,
      "learning_rate": 2.9568947434732777e-06,
      "loss": 1.6196,
      "step": 466
    },
    {
      "epoch": 6.06,
      "learning_rate": 2.919635791181934e-06,
      "loss": 1.6367,
      "step": 467
    },
    {
      "epoch": 6.08,
      "learning_rate": 2.882572906421145e-06,
      "loss": 1.7124,
      "step": 468
    },
    {
      "epoch": 6.09,
      "learning_rate": 2.8457071155262885e-06,
      "loss": 1.623,
      "step": 469
    },
    {
      "epoch": 6.1,
      "learning_rate": 2.809039439374878e-06,
      "loss": 1.5874,
      "step": 470
    },
    {
      "epoch": 6.12,
      "learning_rate": 2.7725708933582785e-06,
      "loss": 1.6362,
      "step": 471
    },
    {
      "epoch": 6.13,
      "learning_rate": 2.7363024873536093e-06,
      "loss": 1.6221,
      "step": 472
    },
    {
      "epoch": 6.14,
      "learning_rate": 2.700235225695752e-06,
      "loss": 1.6416,
      "step": 473
    },
    {
      "epoch": 6.16,
      "learning_rate": 2.6643701071495644e-06,
      "loss": 1.668,
      "step": 474
    },
    {
      "epoch": 6.17,
      "learning_rate": 2.628708124882212e-06,
      "loss": 1.5928,
      "step": 475
    },
    {
      "epoch": 6.18,
      "learning_rate": 2.5932502664356553e-06,
      "loss": 1.6172,
      "step": 476
    },
    {
      "epoch": 6.19,
      "learning_rate": 2.5579975136993253e-06,
      "loss": 1.6162,
      "step": 477
    },
    {
      "epoch": 6.21,
      "learning_rate": 2.52295084288291e-06,
      "loss": 1.6636,
      "step": 478
    },
    {
      "epoch": 6.22,
      "learning_rate": 2.4881112244893403e-06,
      "loss": 1.6748,
      "step": 479
    },
    {
      "epoch": 6.23,
      "learning_rate": 2.453479623287909e-06,
      "loss": 1.6167,
      "step": 480
    },
    {
      "epoch": 6.25,
      "learning_rate": 2.419056998287547e-06,
      "loss": 1.6763,
      "step": 481
    },
    {
      "epoch": 6.26,
      "learning_rate": 2.3848443027102706e-06,
      "loss": 1.6587,
      "step": 482
    },
    {
      "epoch": 6.27,
      "learning_rate": 2.3508424839647994e-06,
      "loss": 1.6538,
      "step": 483
    },
    {
      "epoch": 6.29,
      "learning_rate": 2.3170524836202936e-06,
      "loss": 1.5952,
      "step": 484
    },
    {
      "epoch": 6.3,
      "learning_rate": 2.2834752373803094e-06,
      "loss": 1.6348,
      "step": 485
    },
    {
      "epoch": 6.31,
      "learning_rate": 2.250111675056863e-06,
      "loss": 1.6074,
      "step": 486
    },
    {
      "epoch": 6.32,
      "learning_rate": 2.216962720544703e-06,
      "loss": 1.6284,
      "step": 487
    },
    {
      "epoch": 6.34,
      "learning_rate": 2.184029291795705e-06,
      "loss": 1.6143,
      "step": 488
    },
    {
      "epoch": 6.35,
      "learning_rate": 2.151312300793473e-06,
      "loss": 1.6323,
      "step": 489
    },
    {
      "epoch": 6.36,
      "learning_rate": 2.118812653528077e-06,
      "loss": 1.6387,
      "step": 490
    },
    {
      "epoch": 6.38,
      "learning_rate": 2.086531249970952e-06,
      "loss": 1.6016,
      "step": 491
    },
    {
      "epoch": 6.39,
      "learning_rate": 2.0544689840499988e-06,
      "loss": 1.6616,
      "step": 492
    },
    {
      "epoch": 6.4,
      "learning_rate": 2.022626743624807e-06,
      "loss": 1.6211,
      "step": 493
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.991005410462089e-06,
      "loss": 1.6504,
      "step": 494
    },
    {
      "epoch": 6.43,
      "learning_rate": 1.9596058602112533e-06,
      "loss": 1.6748,
      "step": 495
    },
    {
      "epoch": 6.44,
      "learning_rate": 1.928428962380148e-06,
      "loss": 1.6597,
      "step": 496
    },
    {
      "epoch": 6.45,
      "learning_rate": 1.8974755803109968e-06,
      "loss": 1.6133,
      "step": 497
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.866746571156479e-06,
      "loss": 1.6294,
      "step": 498
    },
    {
      "epoch": 6.48,
      "learning_rate": 1.8362427858560094e-06,
      "loss": 1.6074,
      "step": 499
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.8059650691121611e-06,
      "loss": 1.645,
      "step": 500
    },
    {
      "epoch": 6.51,
      "learning_rate": 1.7759142593672707e-06,
      "loss": 1.5884,
      "step": 501
    },
    {
      "epoch": 6.52,
      "learning_rate": 1.74609118878024e-06,
      "loss": 1.6245,
      "step": 502
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.7164966832034668e-06,
      "loss": 1.6309,
      "step": 503
    },
    {
      "epoch": 6.55,
      "learning_rate": 1.6871315621599982e-06,
      "loss": 1.6035,
      "step": 504
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.6579966388208257e-06,
      "loss": 1.5688,
      "step": 505
    },
    {
      "epoch": 6.57,
      "learning_rate": 1.6290927199823604e-06,
      "loss": 1.5762,
      "step": 506
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.6004206060441096e-06,
      "loss": 1.6323,
      "step": 507
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.5719810909864941e-06,
      "loss": 1.5884,
      "step": 508
    },
    {
      "epoch": 6.61,
      "learning_rate": 1.543774962348874e-06,
      "loss": 1.6382,
      "step": 509
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.5158030012077329e-06,
      "loss": 1.6279,
      "step": 510
    },
    {
      "epoch": 6.64,
      "learning_rate": 1.4880659821550547e-06,
      "loss": 1.6304,
      "step": 511
    },
    {
      "epoch": 6.65,
      "learning_rate": 1.4605646732768685e-06,
      "loss": 1.6289,
      "step": 512
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.4332998361319783e-06,
      "loss": 1.5889,
      "step": 513
    },
    {
      "epoch": 6.68,
      "learning_rate": 1.4062722257308803e-06,
      "loss": 1.6221,
      "step": 514
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.3794825905148557e-06,
      "loss": 1.604,
      "step": 515
    },
    {
      "epoch": 6.7,
      "learning_rate": 1.3529316723352303e-06,
      "loss": 1.6099,
      "step": 516
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.3266202064328548e-06,
      "loss": 1.6045,
      "step": 517
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.3005489214177213e-06,
      "loss": 1.6289,
      "step": 518
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.2747185392488048e-06,
      "loss": 1.6519,
      "step": 519
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.249129775214064e-06,
      "loss": 1.6338,
      "step": 520
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.2237833379106257e-06,
      "loss": 1.6196,
      "step": 521
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.1986799292251816e-06,
      "loss": 1.6104,
      "step": 522
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.1738202443145307e-06,
      "loss": 1.6309,
      "step": 523
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.1492049715863464e-06,
      "loss": 1.5845,
      "step": 524
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.1248347926801029e-06,
      "loss": 1.582,
      "step": 525
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.100710382448198e-06,
      "loss": 1.6553,
      "step": 526
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.0768324089372816e-06,
      "loss": 1.5771,
      "step": 527
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.053201533369731e-06,
      "loss": 1.6611,
      "step": 528
    },
    {
      "epoch": 6.87,
      "learning_rate": 1.029818410125365e-06,
      "loss": 1.6128,
      "step": 529
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.0066836867233087e-06,
      "loss": 1.5957,
      "step": 530
    },
    {
      "epoch": 6.9,
      "learning_rate": 9.837980038040607e-07,
      "loss": 1.6299,
      "step": 531
    },
    {
      "epoch": 6.91,
      "learning_rate": 9.611619951117657e-07,
      "loss": 1.6147,
      "step": 532
    },
    {
      "epoch": 6.92,
      "learning_rate": 9.387762874766515e-07,
      "loss": 1.5864,
      "step": 533
    },
    {
      "epoch": 6.94,
      "learning_rate": 9.166415007976803e-07,
      "loss": 1.6245,
      "step": 534
    },
    {
      "epoch": 6.95,
      "learning_rate": 8.94758248025378e-07,
      "loss": 1.5781,
      "step": 535
    },
    {
      "epoch": 6.96,
      "learning_rate": 8.7312713514486e-07,
      "loss": 1.5845,
      "step": 536
    },
    {
      "epoch": 6.97,
      "learning_rate": 8.517487611590558e-07,
      "loss": 1.624,
      "step": 537
    },
    {
      "epoch": 6.99,
      "learning_rate": 8.306237180721121e-07,
      "loss": 1.5811,
      "step": 538
    },
    {
      "epoch": 7.0,
      "learning_rate": 8.097525908730108e-07,
      "loss": 1.5898,
      "step": 539
    },
    {
      "epoch": 7.01,
      "learning_rate": 7.891359575193613e-07,
      "loss": 1.5542,
      "step": 540
    },
    {
      "epoch": 7.03,
      "learning_rate": 7.687743889213939e-07,
      "loss": 1.6382,
      "step": 541
    },
    {
      "epoch": 7.04,
      "learning_rate": 7.486684489261609e-07,
      "loss": 1.6597,
      "step": 542
    },
    {
      "epoch": 7.05,
      "learning_rate": 7.288186943019171e-07,
      "loss": 1.5918,
      "step": 543
    },
    {
      "epoch": 7.06,
      "learning_rate": 7.092256747226944e-07,
      "loss": 1.6226,
      "step": 544
    },
    {
      "epoch": 7.08,
      "learning_rate": 6.89889932753095e-07,
      "loss": 1.563,
      "step": 545
    },
    {
      "epoch": 7.09,
      "learning_rate": 6.708120038332533e-07,
      "loss": 1.6348,
      "step": 546
    },
    {
      "epoch": 7.1,
      "learning_rate": 6.519924162640168e-07,
      "loss": 1.6089,
      "step": 547
    },
    {
      "epoch": 7.12,
      "learning_rate": 6.334316911923155e-07,
      "loss": 1.6143,
      "step": 548
    },
    {
      "epoch": 7.13,
      "learning_rate": 6.151303425967259e-07,
      "loss": 1.6396,
      "step": 549
    },
    {
      "epoch": 7.14,
      "learning_rate": 5.970888772732453e-07,
      "loss": 1.6387,
      "step": 550
    },
    {
      "epoch": 7.16,
      "learning_rate": 5.793077948212478e-07,
      "loss": 1.5835,
      "step": 551
    },
    {
      "epoch": 7.17,
      "learning_rate": 5.617875876296641e-07,
      "loss": 1.6489,
      "step": 552
    },
    {
      "epoch": 7.18,
      "learning_rate": 5.445287408633304e-07,
      "loss": 1.6318,
      "step": 553
    },
    {
      "epoch": 7.19,
      "learning_rate": 5.27531732449561e-07,
      "loss": 1.645,
      "step": 554
    },
    {
      "epoch": 7.21,
      "learning_rate": 5.107970330649204e-07,
      "loss": 1.5996,
      "step": 555
    },
    {
      "epoch": 7.22,
      "learning_rate": 4.943251061221721e-07,
      "loss": 1.5962,
      "step": 556
    },
    {
      "epoch": 7.23,
      "learning_rate": 4.78116407757464e-07,
      "loss": 1.6211,
      "step": 557
    },
    {
      "epoch": 7.25,
      "learning_rate": 4.6217138681769026e-07,
      "loss": 1.6011,
      "step": 558
    },
    {
      "epoch": 7.26,
      "learning_rate": 4.464904848480522e-07,
      "loss": 1.6392,
      "step": 559
    },
    {
      "epoch": 7.27,
      "learning_rate": 4.310741360798498e-07,
      "loss": 1.6265,
      "step": 560
    },
    {
      "epoch": 7.29,
      "learning_rate": 4.1592276741844075e-07,
      "loss": 1.6255,
      "step": 561
    },
    {
      "epoch": 7.3,
      "learning_rate": 4.0103679843142895e-07,
      "loss": 1.6196,
      "step": 562
    },
    {
      "epoch": 7.31,
      "learning_rate": 3.864166413370429e-07,
      "loss": 1.6201,
      "step": 563
    },
    {
      "epoch": 7.32,
      "learning_rate": 3.720627009927158e-07,
      "loss": 1.6396,
      "step": 564
    },
    {
      "epoch": 7.34,
      "learning_rate": 3.5797537488388326e-07,
      "loss": 1.6553,
      "step": 565
    },
    {
      "epoch": 7.35,
      "learning_rate": 3.441550531129667e-07,
      "loss": 1.6431,
      "step": 566
    },
    {
      "epoch": 7.36,
      "learning_rate": 3.3060211838858104e-07,
      "loss": 1.6362,
      "step": 567
    },
    {
      "epoch": 7.38,
      "learning_rate": 3.1731694601492834e-07,
      "loss": 1.5654,
      "step": 568
    },
    {
      "epoch": 7.39,
      "learning_rate": 3.042999038814076e-07,
      "loss": 1.6074,
      "step": 569
    },
    {
      "epoch": 7.4,
      "learning_rate": 2.915513524524294e-07,
      "loss": 1.6094,
      "step": 570
    },
    {
      "epoch": 7.42,
      "learning_rate": 2.790716447574304e-07,
      "loss": 1.6758,
      "step": 571
    },
    {
      "epoch": 7.43,
      "learning_rate": 2.668611263811016e-07,
      "loss": 1.6313,
      "step": 572
    },
    {
      "epoch": 7.44,
      "learning_rate": 2.5492013545381666e-07,
      "loss": 1.5835,
      "step": 573
    },
    {
      "epoch": 7.45,
      "learning_rate": 2.4324900264226405e-07,
      "loss": 1.5972,
      "step": 574
    },
    {
      "epoch": 7.47,
      "learning_rate": 2.3184805114029872e-07,
      "loss": 1.6689,
      "step": 575
    },
    {
      "epoch": 7.48,
      "learning_rate": 2.2071759665998282e-07,
      "loss": 1.6304,
      "step": 576
    },
    {
      "epoch": 7.49,
      "learning_rate": 2.098579474228546e-07,
      "loss": 1.6035,
      "step": 577
    },
    {
      "epoch": 7.51,
      "learning_rate": 1.9926940415138206e-07,
      "loss": 1.5952,
      "step": 578
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.8895226006064084e-07,
      "loss": 1.584,
      "step": 579
    },
    {
      "epoch": 7.53,
      "learning_rate": 1.7890680085019597e-07,
      "loss": 1.6064,
      "step": 580
    },
    {
      "epoch": 7.55,
      "learning_rate": 1.6913330469618628e-07,
      "loss": 1.6235,
      "step": 581
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.5963204224362261e-07,
      "loss": 1.6294,
      "step": 582
    },
    {
      "epoch": 7.57,
      "learning_rate": 1.504032765988961e-07,
      "loss": 1.6055,
      "step": 583
    },
    {
      "epoch": 7.58,
      "learning_rate": 1.4144726332248726e-07,
      "loss": 1.6353,
      "step": 584
    },
    {
      "epoch": 7.6,
      "learning_rate": 1.327642504218951e-07,
      "loss": 1.6108,
      "step": 585
    },
    {
      "epoch": 7.61,
      "learning_rate": 1.2435447834476254e-07,
      "loss": 1.6201,
      "step": 586
    },
    {
      "epoch": 7.62,
      "learning_rate": 1.1621817997222507e-07,
      "loss": 1.6128,
      "step": 587
    },
    {
      "epoch": 7.64,
      "learning_rate": 1.0835558061245587e-07,
      "loss": 1.6196,
      "step": 588
    },
    {
      "epoch": 7.65,
      "learning_rate": 1.0076689799442874e-07,
      "loss": 1.6621,
      "step": 589
    },
    {
      "epoch": 7.66,
      "learning_rate": 9.34523422618916e-08,
      "loss": 1.6216,
      "step": 590
    },
    {
      "epoch": 7.68,
      "learning_rate": 8.641211596754129e-08,
      "loss": 1.6289,
      "step": 591
    },
    {
      "epoch": 7.69,
      "learning_rate": 7.964641406742135e-08,
      "loss": 1.6279,
      "step": 592
    },
    {
      "epoch": 7.7,
      "learning_rate": 7.315542391551966e-08,
      "loss": 1.6187,
      "step": 593
    },
    {
      "epoch": 7.71,
      "learning_rate": 6.693932525857927e-08,
      "loss": 1.6445,
      "step": 594
    },
    {
      "epoch": 7.73,
      "learning_rate": 6.099829023112236e-08,
      "loss": 1.6226,
      "step": 595
    },
    {
      "epoch": 7.74,
      "learning_rate": 5.533248335068409e-08,
      "loss": 1.6025,
      "step": 596
    },
    {
      "epoch": 7.75,
      "learning_rate": 4.994206151325509e-08,
      "loss": 1.5981,
      "step": 597
    },
    {
      "epoch": 7.77,
      "learning_rate": 4.482717398894165e-08,
      "loss": 1.6479,
      "step": 598
    },
    {
      "epoch": 7.78,
      "learning_rate": 3.998796241782232e-08,
      "loss": 1.6494,
      "step": 599
    },
    {
      "epoch": 7.79,
      "learning_rate": 3.5424560806036625e-08,
      "loss": 1.6328,
      "step": 600
    },
    {
      "epoch": 7.81,
      "learning_rate": 3.1137095522068006e-08,
      "loss": 1.5732,
      "step": 601
    },
    {
      "epoch": 7.82,
      "learning_rate": 2.7125685293245552e-08,
      "loss": 1.6196,
      "step": 602
    },
    {
      "epoch": 7.83,
      "learning_rate": 2.3390441202455484e-08,
      "loss": 1.5894,
      "step": 603
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.993146668506585e-08,
      "loss": 1.6172,
      "step": 604
    },
    {
      "epoch": 7.86,
      "learning_rate": 1.6748857526066588e-08,
      "loss": 1.604,
      "step": 605
    },
    {
      "epoch": 7.87,
      "learning_rate": 1.3842701857406104e-08,
      "loss": 1.6172,
      "step": 606
    },
    {
      "epoch": 7.88,
      "learning_rate": 1.1213080155564327e-08,
      "loss": 1.6377,
      "step": 607
    },
    {
      "epoch": 7.9,
      "learning_rate": 8.860065239311155e-09,
      "loss": 1.6064,
      "step": 608
    },
    {
      "epoch": 7.91,
      "learning_rate": 6.783722267701409e-09,
      "loss": 1.6211,
      "step": 609
    },
    {
      "epoch": 7.92,
      "learning_rate": 4.984108738261828e-09,
      "loss": 1.6274,
      "step": 610
    },
    {
      "epoch": 7.94,
      "learning_rate": 3.4612744854045645e-09,
      "loss": 1.6328,
      "step": 611
    },
    {
      "epoch": 7.95,
      "learning_rate": 2.215261679042735e-09,
      "loss": 1.583,
      "step": 612
    },
    {
      "epoch": 7.96,
      "learning_rate": 1.246104823426908e-09,
      "loss": 1.6318,
      "step": 613
    },
    {
      "epoch": 7.97,
      "learning_rate": 5.538307561858691e-10,
      "loss": 1.6265,
      "step": 614
    },
    {
      "epoch": 7.99,
      "learning_rate": 1.3845864758610384e-10,
      "loss": 1.605,
      "step": 615
    },
    {
      "epoch": 8.0,
      "learning_rate": 0.0,
      "loss": 1.6025,
      "step": 616
    },
    {
      "epoch": 8.0,
      "step": 616,
      "total_flos": 1.5114021399418634e+18,
      "train_loss": 2.115578391335227,
      "train_runtime": 34978.7912,
      "train_samples_per_second": 2.252,
      "train_steps_per_second": 0.018
    }
  ],
  "max_steps": 616,
  "num_train_epochs": 8,
  "total_flos": 1.5114021399418634e+18,
  "trial_name": null,
  "trial_params": null
}