{
  "best_metric": 0.47524943947792053,
  "best_model_checkpoint": "./qlora-NIPS-data-400k_01/checkpoint-2752",
  "epoch": 1.532080723729993,
  "eval_steps": 64,
  "global_step": 2752,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 3.125e-07,
      "loss": 0.8165,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.25e-07,
      "loss": 0.9879,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 9.375000000000001e-07,
      "loss": 0.9669,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.25e-06,
      "loss": 0.9109,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.5625e-06,
      "loss": 0.941,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.9453,
      "step": 6
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.1875000000000002e-06,
      "loss": 0.9347,
      "step": 7
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.5e-06,
      "loss": 0.8695,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.8125e-06,
      "loss": 0.8352,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.125e-06,
      "loss": 0.8564,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4375e-06,
      "loss": 0.8302,
      "step": 11
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.9371,
      "step": 12
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.8189,
      "step": 13
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.9387,
      "step": 14
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.7479,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-06,
      "loss": 0.798,
      "step": 16
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.3125e-06,
      "loss": 0.7361,
      "step": 17
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.625e-06,
      "loss": 1.0068,
      "step": 18
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.9375e-06,
      "loss": 0.6944,
      "step": 19
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-06,
      "loss": 0.7299,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.5625e-06,
      "loss": 0.8062,
      "step": 21
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.875e-06,
      "loss": 0.6909,
      "step": 22
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.1875e-06,
      "loss": 0.7106,
      "step": 23
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.7108,
      "step": 24
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.8125e-06,
      "loss": 0.6372,
      "step": 25
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.6847,
      "step": 26
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.4375e-06,
      "loss": 0.5598,
      "step": 27
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.8044,
      "step": 28
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.0625e-06,
      "loss": 0.6563,
      "step": 29
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.8439,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.6875e-06,
      "loss": 0.7044,
      "step": 31
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 0.6,
      "step": 32
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999998053117552e-06,
      "loss": 0.6321,
      "step": 33
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999992212471724e-06,
      "loss": 0.8344,
      "step": 34
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999982478067059e-06,
      "loss": 0.673,
      "step": 35
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999968849911145e-06,
      "loss": 0.6557,
      "step": 36
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999951328014591e-06,
      "loss": 0.6997,
      "step": 37
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999929912391043e-06,
      "loss": 0.5216,
      "step": 38
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999904603057179e-06,
      "loss": 0.7094,
      "step": 39
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999875400032708e-06,
      "loss": 0.521,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999842303340373e-06,
      "loss": 0.6753,
      "step": 41
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999805313005946e-06,
      "loss": 0.6763,
      "step": 42
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999764429058236e-06,
      "loss": 0.5527,
      "step": 43
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.99971965152908e-06,
      "loss": 0.6219,
      "step": 44
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999670980453348e-06,
      "loss": 0.5754,
      "step": 45
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999618415868944e-06,
      "loss": 0.4872,
      "step": 46
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999561957816803e-06,
      "loss": 0.5978,
      "step": 47
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.99950160634089e-06,
      "loss": 0.634,
      "step": 48
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999437361488208e-06,
      "loss": 0.6225,
      "step": 49
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999369223308784e-06,
      "loss": 0.5999,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999297191855681e-06,
      "loss": 0.5386,
      "step": 51
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999221267184993e-06,
      "loss": 0.5576,
      "step": 52
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999141449355852e-06,
      "loss": 0.5008,
      "step": 53
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999057738430409e-06,
      "loss": 0.5041,
      "step": 54
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.99897013447386e-06,
      "loss": 0.4938,
      "step": 55
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998878637554423e-06,
      "loss": 0.5274,
      "step": 56
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998783247743353e-06,
      "loss": 0.4732,
      "step": 57
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998683965114937e-06,
      "loss": 0.5957,
      "step": 58
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998580789746488e-06,
      "loss": 0.6783,
      "step": 59
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998473721718355e-06,
      "loss": 0.5587,
      "step": 60
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998362761113921e-06,
      "loss": 0.5989,
      "step": 61
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.998247908019594e-06,
      "loss": 0.5879,
      "step": 62
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998129162524814e-06,
      "loss": 0.6438,
      "step": 63
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998006524722059e-06,
      "loss": 0.574,
      "step": 64
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.6096213459968567,
      "eval_runtime": 79.0621,
      "eval_samples_per_second": 25.297,
      "eval_steps_per_second": 12.648,
      "step": 64
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.997879994706832e-06,
      "loss": 0.6231,
      "step": 65
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.997749572577667e-06,
      "loss": 0.4819,
      "step": 66
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.99761525843613e-06,
      "loss": 0.5907,
      "step": 67
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.997477052386824e-06,
      "loss": 0.6023,
      "step": 68
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.997334954537369e-06,
      "loss": 0.5224,
      "step": 69
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.997188964998433e-06,
      "loss": 0.4894,
      "step": 70
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.9970390838837e-06,
      "loss": 0.4819,
      "step": 71
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.996885311309892e-06,
      "loss": 0.5552,
      "step": 72
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.99672764739676e-06,
      "loss": 0.5185,
      "step": 73
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.996566092267084e-06,
      "loss": 0.4946,
      "step": 74
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.996400646046677e-06,
      "loss": 0.5452,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.996231308864382e-06,
      "loss": 0.5336,
      "step": 76
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.996058080852067e-06,
      "loss": 0.5169,
      "step": 77
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.995880962144636e-06,
      "loss": 0.5177,
      "step": 78
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.99569995288002e-06,
      "loss": 0.5767,
      "step": 79
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.995515053199184e-06,
      "loss": 0.5936,
      "step": 80
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.995326263246114e-06,
      "loss": 0.4862,
      "step": 81
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.995133583167833e-06,
      "loss": 0.6151,
      "step": 82
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.994937013114392e-06,
      "loss": 0.4828,
      "step": 83
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.994736553238871e-06,
      "loss": 0.7472,
      "step": 84
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.994532203697376e-06,
      "loss": 0.5064,
      "step": 85
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.994323964649047e-06,
      "loss": 0.6282,
      "step": 86
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.994111836256049e-06,
      "loss": 0.5719,
      "step": 87
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.993895818683578e-06,
      "loss": 0.6454,
      "step": 88
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.993675912099862e-06,
      "loss": 0.6099,
      "step": 89
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.99345211667615e-06,
      "loss": 0.7713,
      "step": 90
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.993224432586723e-06,
      "loss": 0.6204,
      "step": 91
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.992992860008893e-06,
      "loss": 0.6609,
      "step": 92
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.992757399122996e-06,
      "loss": 0.5075,
      "step": 93
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.9925180501124e-06,
      "loss": 0.5341,
      "step": 94
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.992274813163496e-06,
      "loss": 0.4679,
      "step": 95
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.992027688465707e-06,
      "loss": 0.478,
      "step": 96
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.991776676211483e-06,
      "loss": 0.5733,
      "step": 97
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.9915217765963e-06,
      "loss": 0.5646,
      "step": 98
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.99126298981866e-06,
      "loss": 0.6954,
      "step": 99
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.991000316080095e-06,
      "loss": 0.5218,
      "step": 100
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.990733755585164e-06,
      "loss": 0.6288,
      "step": 101
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.990463308541452e-06,
      "loss": 0.5681,
      "step": 102
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.99018897515957e-06,
      "loss": 0.5465,
      "step": 103
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.989910755653154e-06,
      "loss": 0.4622,
      "step": 104
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.989628650238869e-06,
      "loss": 0.5905,
      "step": 105
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.989342659136407e-06,
      "loss": 0.5715,
      "step": 106
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.989052782568484e-06,
      "loss": 0.6149,
      "step": 107
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.988759020760841e-06,
      "loss": 0.5057,
      "step": 108
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.988461373942248e-06,
      "loss": 0.5731,
      "step": 109
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.988159842344498e-06,
      "loss": 0.6775,
      "step": 110
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.987854426202405e-06,
      "loss": 0.4528,
      "step": 111
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.987545125753818e-06,
      "loss": 0.5313,
      "step": 112
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.987231941239607e-06,
      "loss": 0.5012,
      "step": 113
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.986914872903659e-06,
      "loss": 0.5323,
      "step": 114
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.986593920992897e-06,
      "loss": 0.5824,
      "step": 115
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.986269085757263e-06,
      "loss": 0.5856,
      "step": 116
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.98594036744972e-06,
      "loss": 0.5003,
      "step": 117
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.98560776632626e-06,
      "loss": 0.5618,
      "step": 118
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.985271282645899e-06,
      "loss": 0.5268,
      "step": 119
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.98493091667067e-06,
      "loss": 0.5912,
      "step": 120
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.984586668665641e-06,
      "loss": 0.5691,
      "step": 121
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.98423853889889e-06,
      "loss": 0.5789,
      "step": 122
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.983886527641529e-06,
      "loss": 0.5838,
      "step": 123
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.983530635167682e-06,
      "loss": 0.5829,
      "step": 124
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.983170861754508e-06,
      "loss": 0.504,
      "step": 125
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.982807207682175e-06,
      "loss": 0.5618,
      "step": 126
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.982439673233885e-06,
      "loss": 0.5363,
      "step": 127
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.982068258695854e-06,
      "loss": 0.5211,
      "step": 128
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.5722420811653137,
      "eval_runtime": 79.0808,
      "eval_samples_per_second": 25.291,
      "eval_steps_per_second": 12.645,
      "step": 128
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.981692964357322e-06,
      "loss": 0.4428,
      "step": 129
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.981313790510552e-06,
      "loss": 0.5178,
      "step": 130
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.980930737450824e-06,
      "loss": 0.493,
      "step": 131
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.980543805476447e-06,
      "loss": 0.4824,
      "step": 132
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.980152994888738e-06,
      "loss": 0.6084,
      "step": 133
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.97975830599205e-06,
      "loss": 0.6085,
      "step": 134
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.979359739093742e-06,
      "loss": 0.5294,
      "step": 135
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.978957294504202e-06,
      "loss": 0.6388,
      "step": 136
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.978550972536834e-06,
      "loss": 0.5521,
      "step": 137
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.978140773508063e-06,
      "loss": 0.6245,
      "step": 138
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.977726697737333e-06,
      "loss": 0.5264,
      "step": 139
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.977308745547105e-06,
      "loss": 0.5898,
      "step": 140
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.976886917262862e-06,
      "loss": 0.55,
      "step": 141
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.976461213213104e-06,
      "loss": 0.6395,
      "step": 142
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.976031633729348e-06,
      "loss": 0.4954,
      "step": 143
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.975598179146133e-06,
      "loss": 0.5925,
      "step": 144
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.975160849801012e-06,
      "loss": 0.5295,
      "step": 145
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.974719646034553e-06,
      "loss": 0.479,
      "step": 146
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.974274568190349e-06,
      "loss": 0.4406,
      "step": 147
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.973825616615004e-06,
      "loss": 0.4649,
      "step": 148
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.973372791658141e-06,
      "loss": 0.5074,
      "step": 149
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.972916093672399e-06,
      "loss": 0.611,
      "step": 150
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.972455523013431e-06,
      "loss": 0.7048,
      "step": 151
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.971991080039912e-06,
      "loss": 0.5418,
      "step": 152
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.971522765113522e-06,
      "loss": 0.5401,
      "step": 153
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.971050578598969e-06,
      "loss": 0.5127,
      "step": 154
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.970574520863964e-06,
      "loss": 0.4859,
      "step": 155
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.970094592279243e-06,
      "loss": 0.4579,
      "step": 156
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.96961079321855e-06,
      "loss": 0.4316,
      "step": 157
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.969123124058646e-06,
      "loss": 0.4774,
      "step": 158
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.968631585179303e-06,
      "loss": 0.4922,
      "step": 159
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.968136176963308e-06,
      "loss": 0.4985,
      "step": 160
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.967636899796464e-06,
      "loss": 0.5246,
      "step": 161
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.967133754067581e-06,
      "loss": 0.4993,
      "step": 162
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.96662674016849e-06,
      "loss": 0.4475,
      "step": 163
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.966115858494028e-06,
      "loss": 0.5884,
      "step": 164
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.965601109442043e-06,
      "loss": 0.4568,
      "step": 165
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.9650824934134e-06,
      "loss": 0.5504,
      "step": 166
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.964560010811972e-06,
      "loss": 0.5667,
      "step": 167
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.964033662044643e-06,
      "loss": 0.541,
      "step": 168
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.96350344752131e-06,
      "loss": 0.5365,
      "step": 169
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.962969367654879e-06,
      "loss": 0.6509,
      "step": 170
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.962431422861264e-06,
      "loss": 0.4116,
      "step": 171
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.961889613559396e-06,
      "loss": 0.5437,
      "step": 172
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.961343940171206e-06,
      "loss": 0.5639,
      "step": 173
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.960794403121641e-06,
      "loss": 0.533,
      "step": 174
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.960241002838653e-06,
      "loss": 0.5721,
      "step": 175
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.959683739753206e-06,
      "loss": 0.5189,
      "step": 176
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.95912261429927e-06,
      "loss": 0.7846,
      "step": 177
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.958557626913822e-06,
      "loss": 0.5217,
      "step": 178
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.957988778036848e-06,
      "loss": 0.4911,
      "step": 179
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.95741606811134e-06,
      "loss": 0.5656,
      "step": 180
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.956839497583299e-06,
      "loss": 0.5082,
      "step": 181
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.956259066901733e-06,
      "loss": 0.5715,
      "step": 182
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.95567477651865e-06,
      "loss": 0.5325,
      "step": 183
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.955086626889068e-06,
      "loss": 0.4987,
      "step": 184
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.954494618471013e-06,
      "loss": 0.4316,
      "step": 185
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.953898751725512e-06,
      "loss": 0.5406,
      "step": 186
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.953299027116598e-06,
      "loss": 0.5589,
      "step": 187
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.95269544511131e-06,
      "loss": 0.4994,
      "step": 188
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.952088006179686e-06,
      "loss": 0.4837,
      "step": 189
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.951476710794773e-06,
      "loss": 0.4763,
      "step": 190
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.950861559432617e-06,
      "loss": 0.5583,
      "step": 191
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.950242552572272e-06,
      "loss": 0.4809,
      "step": 192
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.5548442006111145,
      "eval_runtime": 79.0785,
      "eval_samples_per_second": 25.291,
      "eval_steps_per_second": 12.646,
      "step": 192
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.94961969069579e-06,
      "loss": 0.5525,
      "step": 193
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.948992974288227e-06,
      "loss": 0.5509,
      "step": 194
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.94836240383764e-06,
      "loss": 0.5044,
      "step": 195
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.94772797983509e-06,
      "loss": 0.5086,
      "step": 196
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.94708970277463e-06,
      "loss": 0.572,
      "step": 197
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.946447573153326e-06,
      "loss": 0.6107,
      "step": 198
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.945801591471235e-06,
      "loss": 0.5338,
      "step": 199
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.945151758231421e-06,
      "loss": 0.4725,
      "step": 200
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.944498073939941e-06,
      "loss": 0.4825,
      "step": 201
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.943840539105853e-06,
      "loss": 0.4891,
      "step": 202
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.943179154241216e-06,
      "loss": 0.4407,
      "step": 203
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.942513919861083e-06,
      "loss": 0.5643,
      "step": 204
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.941844836483512e-06,
      "loss": 0.536,
      "step": 205
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.941171904629547e-06,
      "loss": 0.3997,
      "step": 206
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.940495124823241e-06,
      "loss": 0.6168,
      "step": 207
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.939814497591637e-06,
      "loss": 0.5239,
      "step": 208
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.939130023464774e-06,
      "loss": 0.4141,
      "step": 209
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.4626,
      "step": 210
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.937749536660415e-06,
      "loss": 0.6568,
      "step": 211
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.937053525057977e-06,
      "loss": 0.4568,
      "step": 212
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.936353668710398e-06,
      "loss": 0.4536,
      "step": 213
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.935649968162688e-06,
      "loss": 0.4452,
      "step": 214
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.934942423962862e-06,
      "loss": 0.5018,
      "step": 215
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.93423103666192e-06,
      "loss": 0.7585,
      "step": 216
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.933515806813856e-06,
      "loss": 0.4969,
      "step": 217
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.932796734975659e-06,
      "loss": 0.4988,
      "step": 218
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.932073821707306e-06,
      "loss": 0.4645,
      "step": 219
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.931347067571773e-06,
      "loss": 0.5931,
      "step": 220
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.930616473135015e-06,
      "loss": 0.513,
      "step": 221
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.92988203896599e-06,
      "loss": 0.5187,
      "step": 222
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.929143765636636e-06,
      "loss": 0.5182,
      "step": 223
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.928401653721891e-06,
      "loss": 0.5049,
      "step": 224
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.927655703799673e-06,
      "loss": 0.5474,
      "step": 225
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.926905916450892e-06,
      "loss": 0.5561,
      "step": 226
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.926152292259452e-06,
      "loss": 0.4641,
      "step": 227
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.925394831812236e-06,
      "loss": 0.4612,
      "step": 228
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.92463353569912e-06,
      "loss": 0.4559,
      "step": 229
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.923868404512963e-06,
      "loss": 0.5295,
      "step": 230
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.923099438849619e-06,
      "loss": 0.4425,
      "step": 231
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.922326639307918e-06,
      "loss": 0.4513,
      "step": 232
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.92155000648968e-06,
      "loss": 0.4194,
      "step": 233
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.920769540999711e-06,
      "loss": 0.5773,
      "step": 234
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.9199852434458e-06,
      "loss": 0.3653,
      "step": 235
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.919197114438722e-06,
      "loss": 0.5329,
      "step": 236
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.918405154592234e-06,
      "loss": 0.5181,
      "step": 237
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.91760936452308e-06,
      "loss": 0.5247,
      "step": 238
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.91680974485098e-06,
      "loss": 0.4108,
      "step": 239
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.91600629619864e-06,
      "loss": 0.5234,
      "step": 240
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.915199019191752e-06,
      "loss": 0.4957,
      "step": 241
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.914387914458983e-06,
      "loss": 0.4754,
      "step": 242
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.913572982631983e-06,
      "loss": 0.5077,
      "step": 243
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.912754224345382e-06,
      "loss": 0.5102,
      "step": 244
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.911931640236791e-06,
      "loss": 0.4394,
      "step": 245
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.911105230946802e-06,
      "loss": 0.5577,
      "step": 246
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.910274997118982e-06,
      "loss": 0.4796,
      "step": 247
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.909440939399877e-06,
      "loss": 0.5317,
      "step": 248
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.908603058439012e-06,
      "loss": 0.6023,
      "step": 249
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.907761354888892e-06,
      "loss": 0.4779,
      "step": 250
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.906915829404993e-06,
      "loss": 0.4293,
      "step": 251
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.906066482645774e-06,
      "loss": 0.49,
      "step": 252
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.905213315272661e-06,
      "loss": 0.4866,
      "step": 253
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.904356327950067e-06,
      "loss": 0.5765,
      "step": 254
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.90349552134537e-06,
      "loss": 0.4825,
      "step": 255
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.902630896128924e-06,
      "loss": 0.4868,
      "step": 256
    },
    {
      "epoch": 0.14,
      "eval_loss": 0.5429680943489075,
      "eval_runtime": 79.0584,
      "eval_samples_per_second": 25.298,
      "eval_steps_per_second": 12.649,
      "step": 256
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.90176245297406e-06,
      "loss": 0.5433,
      "step": 257
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.900890192557082e-06,
      "loss": 0.5151,
      "step": 258
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.900014115557266e-06,
      "loss": 0.4419,
      "step": 259
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.899134222656857e-06,
      "loss": 0.5391,
      "step": 260
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.898250514541075e-06,
      "loss": 0.4374,
      "step": 261
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.89736299189811e-06,
      "loss": 0.5194,
      "step": 262
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.896471655419124e-06,
      "loss": 0.6542,
      "step": 263
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.895576505798248e-06,
      "loss": 0.4405,
      "step": 264
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.894677543732583e-06,
      "loss": 0.6172,
      "step": 265
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.893774769922196e-06,
      "loss": 0.5801,
      "step": 266
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.892868185070125e-06,
      "loss": 0.4615,
      "step": 267
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.891957789882378e-06,
      "loss": 0.5452,
      "step": 268
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.891043585067925e-06,
      "loss": 0.5684,
      "step": 269
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.890125571338709e-06,
      "loss": 0.5419,
      "step": 270
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.889203749409635e-06,
      "loss": 0.4965,
      "step": 271
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.888278119998573e-06,
      "loss": 0.4844,
      "step": 272
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.887348683826362e-06,
      "loss": 0.5648,
      "step": 273
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.886415441616799e-06,
      "loss": 0.6049,
      "step": 274
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.885478394096653e-06,
      "loss": 0.4377,
      "step": 275
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.88453754199565e-06,
      "loss": 0.5485,
      "step": 276
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.883592886046486e-06,
      "loss": 0.5001,
      "step": 277
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.88264442698481e-06,
      "loss": 0.4296,
      "step": 278
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.881692165549239e-06,
      "loss": 0.5512,
      "step": 279
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.88073610248135e-06,
      "loss": 0.408,
      "step": 280
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.879776238525678e-06,
      "loss": 0.6324,
      "step": 281
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.878812574429722e-06,
      "loss": 0.5345,
      "step": 282
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.877845110943937e-06,
      "loss": 0.4709,
      "step": 283
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.876873848821738e-06,
      "loss": 0.4904,
      "step": 284
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.8758987888195e-06,
      "loss": 0.5782,
      "step": 285
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.874919931696551e-06,
      "loss": 0.494,
      "step": 286
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.873937278215181e-06,
      "loss": 0.5124,
      "step": 287
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.872950829140633e-06,
      "loss": 0.479,
      "step": 288
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.87196058524111e-06,
      "loss": 0.427,
      "step": 289
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.870966547287765e-06,
      "loss": 0.5456,
      "step": 290
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.869968716054707e-06,
      "loss": 0.5136,
      "step": 291
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.868967092319003e-06,
      "loss": 0.4721,
      "step": 292
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.867961676860668e-06,
      "loss": 0.4731,
      "step": 293
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.866952470462674e-06,
      "loss": 0.4966,
      "step": 294
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.865939473910941e-06,
      "loss": 0.3889,
      "step": 295
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.864922687994347e-06,
      "loss": 0.4427,
      "step": 296
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.863902113504713e-06,
      "loss": 0.6879,
      "step": 297
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.862877751236816e-06,
      "loss": 0.462,
      "step": 298
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.861849601988384e-06,
      "loss": 0.5933,
      "step": 299
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.860817666560088e-06,
      "loss": 0.55,
      "step": 300
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.85978194575555e-06,
      "loss": 0.4847,
      "step": 301
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.858742440381343e-06,
      "loss": 0.4662,
      "step": 302
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.857699151246985e-06,
      "loss": 0.5489,
      "step": 303
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.856652079164936e-06,
      "loss": 0.4433,
      "step": 304
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.855601224950614e-06,
      "loss": 0.5152,
      "step": 305
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.854546589422368e-06,
      "loss": 0.4854,
      "step": 306
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.853488173401504e-06,
      "loss": 0.5628,
      "step": 307
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.852425977712263e-06,
      "loss": 0.4533,
      "step": 308
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.851360003181836e-06,
      "loss": 0.4805,
      "step": 309
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.850290250640351e-06,
      "loss": 0.4837,
      "step": 310
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.84921672092088e-06,
      "loss": 0.4298,
      "step": 311
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.848139414859441e-06,
      "loss": 0.4588,
      "step": 312
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.847058333294989e-06,
      "loss": 0.4971,
      "step": 313
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.845973477069417e-06,
      "loss": 0.5583,
      "step": 314
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.844884847027561e-06,
      "loss": 0.5648,
      "step": 315
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.843792444017195e-06,
      "loss": 0.4825,
      "step": 316
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.842696268889032e-06,
      "loss": 0.4509,
      "step": 317
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.841596322496722e-06,
      "loss": 0.4739,
      "step": 318
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.84049260569685e-06,
      "loss": 0.4976,
      "step": 319
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.839385119348938e-06,
      "loss": 0.4764,
      "step": 320
    },
    {
      "epoch": 0.18,
      "eval_loss": 0.5333892107009888,
      "eval_runtime": 79.0623,
      "eval_samples_per_second": 25.297,
      "eval_steps_per_second": 12.648,
      "step": 320
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.838273864315446e-06,
      "loss": 0.5165,
      "step": 321
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.837158841461767e-06,
      "loss": 0.5403,
      "step": 322
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.836040051656228e-06,
      "loss": 0.5087,
      "step": 323
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.83491749577009e-06,
      "loss": 0.4615,
      "step": 324
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.833791174677546e-06,
      "loss": 0.4195,
      "step": 325
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.832661089255722e-06,
      "loss": 0.4603,
      "step": 326
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.831527240384677e-06,
      "loss": 0.5733,
      "step": 327
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.8303896289474e-06,
      "loss": 0.5571,
      "step": 328
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.829248255829804e-06,
      "loss": 0.4694,
      "step": 329
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.828103121920742e-06,
      "loss": 0.4807,
      "step": 330
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.826954228111988e-06,
      "loss": 0.482,
      "step": 331
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.825801575298248e-06,
      "loss": 0.446,
      "step": 332
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.824645164377154e-06,
      "loss": 0.4578,
      "step": 333
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.823484996249263e-06,
      "loss": 0.547,
      "step": 334
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.822321071818057e-06,
      "loss": 0.5362,
      "step": 335
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.821153391989951e-06,
      "loss": 0.5058,
      "step": 336
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.819981957674273e-06,
      "loss": 0.5525,
      "step": 337
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.81880676978329e-06,
      "loss": 0.5829,
      "step": 338
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.817627829232173e-06,
      "loss": 0.4493,
      "step": 339
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.816445136939032e-06,
      "loss": 0.4462,
      "step": 340
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.81525869382489e-06,
      "loss": 0.4681,
      "step": 341
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.814068500813692e-06,
      "loss": 0.389,
      "step": 342
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.812874558832306e-06,
      "loss": 0.5339,
      "step": 343
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.811676868810517e-06,
      "loss": 0.5293,
      "step": 344
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.810475431681033e-06,
      "loss": 0.5142,
      "step": 345
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.809270248379472e-06,
      "loss": 0.5726,
      "step": 346
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.808061319844376e-06,
      "loss": 0.5037,
      "step": 347
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.806848647017201e-06,
      "loss": 0.4902,
      "step": 348
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.805632230842322e-06,
      "loss": 0.539,
      "step": 349
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.804412072267025e-06,
      "loss": 0.4967,
      "step": 350
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.80318817224151e-06,
      "loss": 0.5119,
      "step": 351
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.801960531718898e-06,
      "loss": 0.4291,
      "step": 352
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.800729151655213e-06,
      "loss": 0.5343,
      "step": 353
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.799494033009396e-06,
      "loss": 0.5804,
      "step": 354
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.798255176743303e-06,
      "loss": 0.4254,
      "step": 355
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.797012583821695e-06,
      "loss": 0.4979,
      "step": 356
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.795766255212242e-06,
      "loss": 0.4761,
      "step": 357
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.79451619188553e-06,
      "loss": 0.441,
      "step": 358
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.793262394815051e-06,
      "loss": 0.586,
      "step": 359
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.792004864977198e-06,
      "loss": 0.4655,
      "step": 360
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.79074360335128e-06,
      "loss": 0.4901,
      "step": 361
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.789478610919508e-06,
      "loss": 0.4765,
      "step": 362
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.788209888666995e-06,
      "loss": 0.5146,
      "step": 363
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.786937437581766e-06,
      "loss": 0.4925,
      "step": 364
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.785661258654746e-06,
      "loss": 0.545,
      "step": 365
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.784381352879762e-06,
      "loss": 0.422,
      "step": 366
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.783097721253543e-06,
      "loss": 0.4474,
      "step": 367
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.781810364775724e-06,
      "loss": 0.4775,
      "step": 368
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.780519284448835e-06,
      "loss": 0.5141,
      "step": 369
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.779224481278312e-06,
      "loss": 0.4978,
      "step": 370
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.777925956272481e-06,
      "loss": 0.4719,
      "step": 371
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.77662371044258e-06,
      "loss": 0.5397,
      "step": 372
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.775317744802729e-06,
      "loss": 0.4311,
      "step": 373
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.774008060369958e-06,
      "loss": 0.5017,
      "step": 374
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.772694658164186e-06,
      "loss": 0.4376,
      "step": 375
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.77137753920823e-06,
      "loss": 0.4582,
      "step": 376
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.770056704527797e-06,
      "loss": 0.5169,
      "step": 377
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.768732155151494e-06,
      "loss": 0.5127,
      "step": 378
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.767403892110816e-06,
      "loss": 0.5328,
      "step": 379
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.766071916440154e-06,
      "loss": 0.4869,
      "step": 380
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.764736229176785e-06,
      "loss": 0.5272,
      "step": 381
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.763396831360884e-06,
      "loss": 0.4626,
      "step": 382
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.762053724035505e-06,
      "loss": 0.4893,
      "step": 383
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.760706908246603e-06,
      "loss": 0.4378,
      "step": 384
    },
    {
      "epoch": 0.21,
      "eval_loss": 0.52733314037323,
      "eval_runtime": 78.9925,
      "eval_samples_per_second": 25.319,
      "eval_steps_per_second": 12.659,
      "step": 384
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.75935638504301e-06,
      "loss": 0.3876,
      "step": 385
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.758002155476454e-06,
      "loss": 0.4443,
      "step": 386
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.756644220601541e-06,
      "loss": 0.5849,
      "step": 387
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.4614,
      "step": 388
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.753917239159519e-06,
      "loss": 0.3794,
      "step": 389
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.752548194716054e-06,
      "loss": 0.5652,
      "step": 390
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.751175449211524e-06,
      "loss": 0.4911,
      "step": 391
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.749799003714954e-06,
      "loss": 0.4656,
      "step": 392
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.748418859298261e-06,
      "loss": 0.5126,
      "step": 393
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.747035017036232e-06,
      "loss": 0.5257,
      "step": 394
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.74564747800654e-06,
      "loss": 0.4614,
      "step": 395
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.744256243289735e-06,
      "loss": 0.5432,
      "step": 396
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.742861313969246e-06,
      "loss": 0.4705,
      "step": 397
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.741462691131377e-06,
      "loss": 0.5083,
      "step": 398
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.74006037586531e-06,
      "loss": 0.4593,
      "step": 399
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.738654369263103e-06,
      "loss": 0.5706,
      "step": 400
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.737244672419688e-06,
      "loss": 0.4994,
      "step": 401
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.735831286432869e-06,
      "loss": 0.3964,
      "step": 402
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.734414212403325e-06,
      "loss": 0.3949,
      "step": 403
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.732993451434607e-06,
      "loss": 0.4325,
      "step": 404
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.731569004633137e-06,
      "loss": 0.4194,
      "step": 405
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.730140873108207e-06,
      "loss": 0.4538,
      "step": 406
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.728709057971979e-06,
      "loss": 0.4599,
      "step": 407
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.727273560339483e-06,
      "loss": 0.4611,
      "step": 408
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.725834381328618e-06,
      "loss": 0.5658,
      "step": 409
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.724391522060148e-06,
      "loss": 0.4589,
      "step": 410
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.722944983657703e-06,
      "loss": 0.5236,
      "step": 411
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.721494767247779e-06,
      "loss": 0.4316,
      "step": 412
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.72004087395974e-06,
      "loss": 0.4586,
      "step": 413
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.718583304925805e-06,
      "loss": 0.4327,
      "step": 414
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.717122061281064e-06,
      "loss": 0.5219,
      "step": 415
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.715657144163463e-06,
      "loss": 0.4618,
      "step": 416
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.71418855471381e-06,
      "loss": 0.4045,
      "step": 417
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.712716294075775e-06,
      "loss": 0.4632,
      "step": 418
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.711240363395885e-06,
      "loss": 0.453,
      "step": 419
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.709760763823524e-06,
      "loss": 0.4425,
      "step": 420
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.708277496510936e-06,
      "loss": 0.526,
      "step": 421
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.70679056261322e-06,
      "loss": 0.4649,
      "step": 422
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.705299963288328e-06,
      "loss": 0.4632,
      "step": 423
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.703805699697072e-06,
      "loss": 0.4537,
      "step": 424
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.702307773003112e-06,
      "loss": 0.4325,
      "step": 425
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.700806184372964e-06,
      "loss": 0.5727,
      "step": 426
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.699300934975993e-06,
      "loss": 0.5295,
      "step": 427
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.697792025984417e-06,
      "loss": 0.4778,
      "step": 428
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.696279458573305e-06,
      "loss": 0.4528,
      "step": 429
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.694763233920572e-06,
      "loss": 0.4456,
      "step": 430
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.693243353206982e-06,
      "loss": 0.5799,
      "step": 431
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.691719817616148e-06,
      "loss": 0.454,
      "step": 432
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.690192628334526e-06,
      "loss": 0.438,
      "step": 433
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.688661786551423e-06,
      "loss": 0.5527,
      "step": 434
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.687127293458982e-06,
      "loss": 0.4025,
      "step": 435
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.685589150252197e-06,
      "loss": 0.5557,
      "step": 436
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.6840473581289e-06,
      "loss": 0.4371,
      "step": 437
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.682501918289767e-06,
      "loss": 0.5403,
      "step": 438
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.680952831938313e-06,
      "loss": 0.5306,
      "step": 439
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.679400100280896e-06,
      "loss": 0.5356,
      "step": 440
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.677843724526708e-06,
      "loss": 0.5173,
      "step": 441
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.676283705887783e-06,
      "loss": 0.5046,
      "step": 442
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.674720045578987e-06,
      "loss": 0.4796,
      "step": 443
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.67315274481803e-06,
      "loss": 0.4871,
      "step": 444
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.671581804825447e-06,
      "loss": 0.407,
      "step": 445
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.670007226824617e-06,
      "loss": 0.4888,
      "step": 446
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.668429012041742e-06,
      "loss": 0.5067,
      "step": 447
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.666847161705866e-06,
      "loss": 0.5995,
      "step": 448
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.5218273997306824,
      "eval_runtime": 79.0535,
      "eval_samples_per_second": 25.299,
      "eval_steps_per_second": 12.65,
      "step": 448
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.665261677048858e-06,
      "loss": 0.4061,
      "step": 449
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.66367255930542e-06,
      "loss": 0.5822,
      "step": 450
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.662079809713082e-06,
      "loss": 0.5444,
      "step": 451
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.660483429512198e-06,
      "loss": 0.377,
      "step": 452
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.65888341994596e-06,
      "loss": 0.4275,
      "step": 453
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.657279782260377e-06,
      "loss": 0.4568,
      "step": 454
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.655672517704289e-06,
      "loss": 0.5082,
      "step": 455
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.654061627529354e-06,
      "loss": 0.6113,
      "step": 456
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.652447112990063e-06,
      "loss": 0.4176,
      "step": 457
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.650828975343718e-06,
      "loss": 0.4993,
      "step": 458
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.649207215850451e-06,
      "loss": 0.4822,
      "step": 459
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.647581835773215e-06,
      "loss": 0.5142,
      "step": 460
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.645952836377775e-06,
      "loss": 0.5042,
      "step": 461
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.644320218932723e-06,
      "loss": 0.4374,
      "step": 462
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.642683984709462e-06,
      "loss": 0.4981,
      "step": 463
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.641044134982215e-06,
      "loss": 0.3573,
      "step": 464
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.63940067102802e-06,
      "loss": 0.4793,
      "step": 465
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.63775359412673e-06,
      "loss": 0.4492,
      "step": 466
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.63610290556101e-06,
      "loss": 0.5817,
      "step": 467
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.63444860661634e-06,
      "loss": 0.4782,
      "step": 468
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.632790698581009e-06,
      "loss": 0.4295,
      "step": 469
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.631129182746118e-06,
      "loss": 0.5782,
      "step": 470
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.629464060405578e-06,
      "loss": 0.4553,
      "step": 471
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.627795332856107e-06,
      "loss": 0.4803,
      "step": 472
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.626123001397232e-06,
      "loss": 0.666,
      "step": 473
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.624447067331287e-06,
      "loss": 0.4582,
      "step": 474
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.622767531963408e-06,
      "loss": 0.4907,
      "step": 475
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.621084396601544e-06,
      "loss": 0.3915,
      "step": 476
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.5525,
      "step": 477
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.617707331141632e-06,
      "loss": 0.4334,
      "step": 478
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.61601340367349e-06,
      "loss": 0.498,
      "step": 479
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.614315881471154e-06,
      "loss": 0.5104,
      "step": 480
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.612614765856577e-06,
      "loss": 0.4234,
      "step": 481
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.61091005815451e-06,
      "loss": 0.679,
      "step": 482
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.609201759692494e-06,
      "loss": 0.3989,
      "step": 483
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.607489871800877e-06,
      "loss": 0.4432,
      "step": 484
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.605774395812795e-06,
      "loss": 0.5664,
      "step": 485
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.604055333064179e-06,
      "loss": 0.442,
      "step": 486
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.602332684893754e-06,
      "loss": 0.5539,
      "step": 487
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.600606452643037e-06,
      "loss": 0.416,
      "step": 488
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.59887663765634e-06,
      "loss": 0.4381,
      "step": 489
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.597143241280758e-06,
      "loss": 0.4374,
      "step": 490
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.59540626486618e-06,
      "loss": 0.5936,
      "step": 491
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.59366570976528e-06,
      "loss": 0.4979,
      "step": 492
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.591921577333525e-06,
      "loss": 0.4883,
      "step": 493
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.590173868929158e-06,
      "loss": 0.6121,
      "step": 494
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.588422585913215e-06,
      "loss": 0.4331,
      "step": 495
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.586667729649513e-06,
      "loss": 0.4896,
      "step": 496
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.584909301504649e-06,
      "loss": 0.4802,
      "step": 497
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.583147302848006e-06,
      "loss": 0.5995,
      "step": 498
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.581381735051745e-06,
      "loss": 0.4577,
      "step": 499
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.57961259949081e-06,
      "loss": 0.4358,
      "step": 500
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.577839897542916e-06,
      "loss": 0.4232,
      "step": 501
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.576063630588563e-06,
      "loss": 0.4274,
      "step": 502
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.574283800011025e-06,
      "loss": 0.43,
      "step": 503
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.572500407196348e-06,
      "loss": 0.5231,
      "step": 504
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.570713453533353e-06,
      "loss": 0.504,
      "step": 505
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.568922940413642e-06,
      "loss": 0.4734,
      "step": 506
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.567128869231575e-06,
      "loss": 0.5759,
      "step": 507
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.565331241384294e-06,
      "loss": 0.5026,
      "step": 508
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.563530058271706e-06,
      "loss": 0.6291,
      "step": 509
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.561725321296489e-06,
      "loss": 0.5096,
      "step": 510
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.559917031864088e-06,
      "loss": 0.5221,
      "step": 511
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.55810519138271e-06,
      "loss": 0.4673,
      "step": 512
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.5173125863075256,
      "eval_runtime": 79.0589,
      "eval_samples_per_second": 25.298,
      "eval_steps_per_second": 12.649,
      "step": 512
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.556289801263334e-06,
      "loss": 0.4624,
      "step": 513
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.5544708629197e-06,
      "loss": 0.5445,
      "step": 514
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.552648377768313e-06,
      "loss": 0.5187,
      "step": 515
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.550822347228436e-06,
      "loss": 0.4261,
      "step": 516
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.548992772722097e-06,
      "loss": 0.4895,
      "step": 517
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.547159655674083e-06,
      "loss": 0.3719,
      "step": 518
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.545322997511938e-06,
      "loss": 0.5412,
      "step": 519
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.543482799665969e-06,
      "loss": 0.4062,
      "step": 520
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.54163906356923e-06,
      "loss": 0.5126,
      "step": 521
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.53979179065754e-06,
      "loss": 0.6525,
      "step": 522
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.537940982369464e-06,
      "loss": 0.4508,
      "step": 523
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.53608664014633e-06,
      "loss": 0.4279,
      "step": 524
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.53422876543221e-06,
      "loss": 0.4578,
      "step": 525
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.532367359673928e-06,
      "loss": 0.5106,
      "step": 526
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.530502424321062e-06,
      "loss": 0.4304,
      "step": 527
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.528633960825933e-06,
      "loss": 0.4933,
      "step": 528
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.526761970643614e-06,
      "loss": 0.4807,
      "step": 529
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.524886455231926e-06,
      "loss": 0.3713,
      "step": 530
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.523007416051426e-06,
      "loss": 0.5254,
      "step": 531
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.521124854565425e-06,
      "loss": 0.4147,
      "step": 532
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.519238772239974e-06,
      "loss": 0.49,
      "step": 533
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.517349170543864e-06,
      "loss": 0.467,
      "step": 534
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.515456050948629e-06,
      "loss": 0.4349,
      "step": 535
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.51355941492854e-06,
      "loss": 0.5455,
      "step": 536
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.511659263960607e-06,
      "loss": 0.5025,
      "step": 537
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.509755599524582e-06,
      "loss": 0.6229,
      "step": 538
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.507848423102946e-06,
      "loss": 0.4356,
      "step": 539
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.50593773618092e-06,
      "loss": 0.4261,
      "step": 540
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.504023540246455e-06,
      "loss": 0.5435,
      "step": 541
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.50210583679024e-06,
      "loss": 0.63,
      "step": 542
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.500184627305691e-06,
      "loss": 0.4592,
      "step": 543
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.498259913288953e-06,
      "loss": 0.4834,
      "step": 544
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.49633169623891e-06,
      "loss": 0.4609,
      "step": 545
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.494399977657158e-06,
      "loss": 0.4195,
      "step": 546
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.492464759048033e-06,
      "loss": 0.494,
      "step": 547
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.490526041918593e-06,
      "loss": 0.4989,
      "step": 548
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.488583827778617e-06,
      "loss": 0.4121,
      "step": 549
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.486638118140612e-06,
      "loss": 0.5701,
      "step": 550
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.484688914519804e-06,
      "loss": 0.4414,
      "step": 551
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.482736218434144e-06,
      "loss": 0.4481,
      "step": 552
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.480780031404295e-06,
      "loss": 0.4334,
      "step": 553
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.478820354953649e-06,
      "loss": 0.361,
      "step": 554
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.476857190608305e-06,
      "loss": 0.3428,
      "step": 555
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.474890539897085e-06,
      "loss": 0.6267,
      "step": 556
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.472920404351527e-06,
      "loss": 0.5139,
      "step": 557
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.470946785505875e-06,
      "loss": 0.5036,
      "step": 558
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.468969684897093e-06,
      "loss": 0.5598,
      "step": 559
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.466989104064853e-06,
      "loss": 0.4888,
      "step": 560
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.46500504455154e-06,
      "loss": 0.4919,
      "step": 561
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.463017507902245e-06,
      "loss": 0.714,
      "step": 562
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.461026495664767e-06,
      "loss": 0.383,
      "step": 563
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.459032009389616e-06,
      "loss": 0.5042,
      "step": 564
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.457034050630001e-06,
      "loss": 0.3919,
      "step": 565
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.4336,
      "step": 566
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.453027721883751e-06,
      "loss": 0.4926,
      "step": 567
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.451019355017057e-06,
      "loss": 0.4066,
      "step": 568
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.449007521905779e-06,
      "loss": 0.5266,
      "step": 569
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.446992224116635e-06,
      "loss": 0.534,
      "step": 570
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.444973463219049e-06,
      "loss": 0.6115,
      "step": 571
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.442951240785135e-06,
      "loss": 0.6032,
      "step": 572
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.440925558389704e-06,
      "loss": 0.3988,
      "step": 573
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.438896417610263e-06,
      "loss": 0.5284,
      "step": 574
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.436863820027011e-06,
      "loss": 0.432,
      "step": 575
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.43482776722284e-06,
      "loss": 0.553,
      "step": 576
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.5129925608634949,
      "eval_runtime": 79.0333,
      "eval_samples_per_second": 25.306,
      "eval_steps_per_second": 12.653,
      "step": 576
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.432788260783333e-06,
      "loss": 0.4708,
      "step": 577
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.430745302296758e-06,
      "loss": 0.5288,
      "step": 578
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.428698893354079e-06,
      "loss": 0.4339,
      "step": 579
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.426649035548942e-06,
      "loss": 0.3875,
      "step": 580
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.424595730477677e-06,
      "loss": 0.4599,
      "step": 581
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.422538979739307e-06,
      "loss": 0.5069,
      "step": 582
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.42047878493553e-06,
      "loss": 0.4862,
      "step": 583
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.418415147670725e-06,
      "loss": 0.488,
      "step": 584
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.416348069551962e-06,
      "loss": 0.5114,
      "step": 585
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.41427755218898e-06,
      "loss": 0.4549,
      "step": 586
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.412203597194204e-06,
      "loss": 0.498,
      "step": 587
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.41012620618273e-06,
      "loss": 0.5201,
      "step": 588
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.408045380772333e-06,
      "loss": 0.5406,
      "step": 589
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.40596112258346e-06,
      "loss": 0.4097,
      "step": 590
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.403873433239237e-06,
      "loss": 0.3922,
      "step": 591
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.401782314365458e-06,
      "loss": 0.4817,
      "step": 592
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.399687767590585e-06,
      "loss": 0.5372,
      "step": 593
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.397589794545754e-06,
      "loss": 0.454,
      "step": 594
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.395488396864769e-06,
      "loss": 0.4681,
      "step": 595
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.393383576184099e-06,
      "loss": 0.472,
      "step": 596
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.391275334142879e-06,
      "loss": 0.5719,
      "step": 597
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.389163672382909e-06,
      "loss": 0.5548,
      "step": 598
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.38704859254865e-06,
      "loss": 0.5349,
      "step": 599
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.38493009628723e-06,
      "loss": 0.4724,
      "step": 600
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.382808185248433e-06,
      "loss": 0.4712,
      "step": 601
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.380682861084703e-06,
      "loss": 0.489,
      "step": 602
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.378554125451142e-06,
      "loss": 0.4214,
      "step": 603
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.37642198000551e-06,
      "loss": 0.5649,
      "step": 604
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.374286426408221e-06,
      "loss": 0.4534,
      "step": 605
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.372147466322347e-06,
      "loss": 0.5382,
      "step": 606
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.370005101413605e-06,
      "loss": 0.6463,
      "step": 607
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.36785933335037e-06,
      "loss": 0.4572,
      "step": 608
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.365710163803666e-06,
      "loss": 0.4539,
      "step": 609
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.363557594447165e-06,
      "loss": 0.486,
      "step": 610
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.361401626957185e-06,
      "loss": 0.5371,
      "step": 611
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.359242263012693e-06,
      "loss": 0.6162,
      "step": 612
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.357079504295304e-06,
      "loss": 0.4469,
      "step": 613
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.354913352489266e-06,
      "loss": 0.5328,
      "step": 614
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.352743809281479e-06,
      "loss": 0.4376,
      "step": 615
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.350570876361483e-06,
      "loss": 0.5159,
      "step": 616
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.348394555421454e-06,
      "loss": 0.4736,
      "step": 617
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.34621484815621e-06,
      "loss": 0.4567,
      "step": 618
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.344031756263201e-06,
      "loss": 0.4941,
      "step": 619
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.341845281442522e-06,
      "loss": 0.4231,
      "step": 620
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.339655425396892e-06,
      "loss": 0.3752,
      "step": 621
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.33746218983167e-06,
      "loss": 0.4356,
      "step": 622
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.335265576454845e-06,
      "loss": 0.4406,
      "step": 623
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.333065586977036e-06,
      "loss": 0.4107,
      "step": 624
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.330862223111489e-06,
      "loss": 0.5,
      "step": 625
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.328655486574083e-06,
      "loss": 0.5116,
      "step": 626
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.32644537908332e-06,
      "loss": 0.4533,
      "step": 627
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.324231902360327e-06,
      "loss": 0.5655,
      "step": 628
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.322015058128857e-06,
      "loss": 0.4198,
      "step": 629
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.319794848115283e-06,
      "loss": 0.4835,
      "step": 630
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.317571274048599e-06,
      "loss": 0.4838,
      "step": 631
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.315344337660422e-06,
      "loss": 0.4864,
      "step": 632
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.313114040684984e-06,
      "loss": 0.4495,
      "step": 633
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.310880384859138e-06,
      "loss": 0.4633,
      "step": 634
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.308643371922345e-06,
      "loss": 0.42,
      "step": 635
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.306403003616691e-06,
      "loss": 0.409,
      "step": 636
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.304159281686867e-06,
      "loss": 0.4202,
      "step": 637
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.301912207880179e-06,
      "loss": 0.457,
      "step": 638
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.299661783946542e-06,
      "loss": 0.4817,
      "step": 639
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.297408011638477e-06,
      "loss": 0.3932,
      "step": 640
    },
    {
      "epoch": 0.36,
      "eval_loss": 0.5109407305717468,
      "eval_runtime": 79.0623,
      "eval_samples_per_second": 25.297,
      "eval_steps_per_second": 12.648,
      "step": 640
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.29515089271112e-06,
      "loss": 0.4037,
      "step": 641
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.29289042892221e-06,
      "loss": 0.4449,
      "step": 642
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.290626622032085e-06,
      "loss": 0.4967,
      "step": 643
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.288359473803697e-06,
      "loss": 0.4316,
      "step": 644
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.28608898600259e-06,
      "loss": 0.4156,
      "step": 645
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.283815160396917e-06,
      "loss": 0.5678,
      "step": 646
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.281537998757421e-06,
      "loss": 0.5437,
      "step": 647
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.279257502857455e-06,
      "loss": 0.4657,
      "step": 648
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.276973674472958e-06,
      "loss": 0.4187,
      "step": 649
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.274686515382467e-06,
      "loss": 0.4608,
      "step": 650
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.272396027367118e-06,
      "loss": 0.5023,
      "step": 651
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.270102212210632e-06,
      "loss": 0.6111,
      "step": 652
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.267805071699328e-06,
      "loss": 0.5555,
      "step": 653
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.265504607622106e-06,
      "loss": 0.5164,
      "step": 654
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.263200821770462e-06,
      "loss": 0.5746,
      "step": 655
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.260893715938477e-06,
      "loss": 0.4507,
      "step": 656
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.258583291922814e-06,
      "loss": 0.4599,
      "step": 657
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.256269551522725e-06,
      "loss": 0.5245,
      "step": 658
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.253952496540041e-06,
      "loss": 0.4965,
      "step": 659
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.251632128779178e-06,
      "loss": 0.4129,
      "step": 660
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.249308450047126e-06,
      "loss": 0.5154,
      "step": 661
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.246981462153456e-06,
      "loss": 0.5286,
      "step": 662
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.24465116691032e-06,
      "loss": 0.4595,
      "step": 663
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.24231756613244e-06,
      "loss": 0.4507,
      "step": 664
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.239980661637117e-06,
      "loss": 0.588,
      "step": 665
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.23764045524422e-06,
      "loss": 0.4717,
      "step": 666
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.235296948776194e-06,
      "loss": 0.5443,
      "step": 667
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.232950144058049e-06,
      "loss": 0.4229,
      "step": 668
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.230600042917366e-06,
      "loss": 0.5067,
      "step": 669
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.228246647184295e-06,
      "loss": 0.5375,
      "step": 670
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.225889958691552e-06,
      "loss": 0.5049,
      "step": 671
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.223529979274411e-06,
      "loss": 0.4193,
      "step": 672
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.221166710770714e-06,
      "loss": 0.4661,
      "step": 673
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.218800155020865e-06,
      "loss": 0.4787,
      "step": 674
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.216430313867827e-06,
      "loss": 0.5159,
      "step": 675
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.214057189157119e-06,
      "loss": 0.4753,
      "step": 676
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.211680782736818e-06,
      "loss": 0.4939,
      "step": 677
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.209301096457559e-06,
      "loss": 0.4682,
      "step": 678
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.20691813217253e-06,
      "loss": 0.4909,
      "step": 679
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.20453189173747e-06,
      "loss": 0.4317,
      "step": 680
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.202142377010675e-06,
      "loss": 0.4362,
      "step": 681
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.19974958985298e-06,
      "loss": 0.444,
      "step": 682
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.197353532127782e-06,
      "loss": 0.4589,
      "step": 683
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.194954205701013e-06,
      "loss": 0.447,
      "step": 684
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.192551612441159e-06,
      "loss": 0.4635,
      "step": 685
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.190145754219244e-06,
      "loss": 0.4625,
      "step": 686
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.18773663290884e-06,
      "loss": 0.5798,
      "step": 687
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.185324250386054e-06,
      "loss": 0.5442,
      "step": 688
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.18290860852954e-06,
      "loss": 0.4713,
      "step": 689
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.180489709220481e-06,
      "loss": 0.3612,
      "step": 690
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.178067554342609e-06,
      "loss": 0.5119,
      "step": 691
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.175642145782179e-06,
      "loss": 0.4144,
      "step": 692
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.173213485427986e-06,
      "loss": 0.4464,
      "step": 693
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.17078157517136e-06,
      "loss": 0.4973,
      "step": 694
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.168346416906153e-06,
      "loss": 0.3995,
      "step": 695
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.165908012528754e-06,
      "loss": 0.3953,
      "step": 696
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.16346636393808e-06,
      "loss": 0.5573,
      "step": 697
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.16102147303557e-06,
      "loss": 0.4327,
      "step": 698
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.158573341725189e-06,
      "loss": 0.4509,
      "step": 699
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.156121971913428e-06,
      "loss": 0.4193,
      "step": 700
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.153667365509299e-06,
      "loss": 0.416,
      "step": 701
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.151209524424333e-06,
      "loss": 0.4706,
      "step": 702
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.148748450572582e-06,
      "loss": 0.4612,
      "step": 703
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.146284145870614e-06,
      "loss": 0.4819,
      "step": 704
    },
    {
      "epoch": 0.39,
      "eval_loss": 0.5074958801269531,
      "eval_runtime": 79.0525,
      "eval_samples_per_second": 25.3,
      "eval_steps_per_second": 12.65,
      "step": 704
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.143816612237512e-06,
      "loss": 0.4877,
      "step": 705
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.141345851594878e-06,
      "loss": 0.7266,
      "step": 706
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.138871865866824e-06,
      "loss": 0.4263,
      "step": 707
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.136394656979972e-06,
      "loss": 0.4577,
      "step": 708
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.133914226863457e-06,
      "loss": 0.4712,
      "step": 709
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.13143057744892e-06,
      "loss": 0.4442,
      "step": 710
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.128943710670513e-06,
      "loss": 0.5147,
      "step": 711
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.126453628464889e-06,
      "loss": 0.4974,
      "step": 712
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.123960332771205e-06,
      "loss": 0.5558,
      "step": 713
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.121463825531127e-06,
      "loss": 0.4443,
      "step": 714
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.118964108688813e-06,
      "loss": 0.4904,
      "step": 715
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.11646118419093e-06,
      "loss": 0.449,
      "step": 716
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.113955053986632e-06,
      "loss": 0.495,
      "step": 717
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.111445720027579e-06,
      "loss": 0.4644,
      "step": 718
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.108933184267924e-06,
      "loss": 0.4241,
      "step": 719
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.106417448664307e-06,
      "loss": 0.5364,
      "step": 720
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.103898515175866e-06,
      "loss": 0.5056,
      "step": 721
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.10137638576423e-06,
      "loss": 0.583,
      "step": 722
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.098851062393514e-06,
      "loss": 0.5559,
      "step": 723
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.096322547030321e-06,
      "loss": 0.4065,
      "step": 724
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.09379084164374e-06,
      "loss": 0.5348,
      "step": 725
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.091255948205342e-06,
      "loss": 0.4559,
      "step": 726
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.088717868689186e-06,
      "loss": 0.4272,
      "step": 727
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.086176605071806e-06,
      "loss": 0.3592,
      "step": 728
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.083632159332222e-06,
      "loss": 0.5802,
      "step": 729
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.081084533451926e-06,
      "loss": 0.4552,
      "step": 730
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.07853372941489e-06,
      "loss": 0.5569,
      "step": 731
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.07597974920756e-06,
      "loss": 0.374,
      "step": 732
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.073422594818857e-06,
      "loss": 0.4882,
      "step": 733
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.070862268240173e-06,
      "loss": 0.4218,
      "step": 734
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.068298771465366e-06,
      "loss": 0.5483,
      "step": 735
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.06573210649077e-06,
      "loss": 0.3646,
      "step": 736
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.063162275315182e-06,
      "loss": 0.5453,
      "step": 737
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.060589279939867e-06,
      "loss": 0.3796,
      "step": 738
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.058013122368552e-06,
      "loss": 0.439,
      "step": 739
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.055433804607427e-06,
      "loss": 0.4964,
      "step": 740
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.052851328665142e-06,
      "loss": 0.4182,
      "step": 741
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.05026569655281e-06,
      "loss": 0.5296,
      "step": 742
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.047676910284002e-06,
      "loss": 0.4461,
      "step": 743
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.5544,
      "step": 744
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.0424898833435e-06,
      "loss": 0.5806,
      "step": 745
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.03989164671122e-06,
      "loss": 0.4681,
      "step": 746
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.037290264001286e-06,
      "loss": 0.5145,
      "step": 747
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.034685737239528e-06,
      "loss": 0.4642,
      "step": 748
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.03207806845423e-06,
      "loss": 0.5336,
      "step": 749
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.029467259676126e-06,
      "loss": 0.558,
      "step": 750
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.026853312938384e-06,
      "loss": 0.4999,
      "step": 751
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.02423623027663e-06,
      "loss": 0.4367,
      "step": 752
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.02161601372892e-06,
      "loss": 0.5415,
      "step": 753
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.018992665335757e-06,
      "loss": 0.488,
      "step": 754
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.01636618714008e-06,
      "loss": 0.4526,
      "step": 755
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.01373658118727e-06,
      "loss": 0.3681,
      "step": 756
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.011103849525139e-06,
      "loss": 0.4311,
      "step": 757
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.008467994203932e-06,
      "loss": 0.6587,
      "step": 758
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.005829017276333e-06,
      "loss": 0.464,
      "step": 759
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.003186920797452e-06,
      "loss": 0.4982,
      "step": 760
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.00054170682483e-06,
      "loss": 0.4554,
      "step": 761
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.997893377418432e-06,
      "loss": 0.5452,
      "step": 762
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.995241934640656e-06,
      "loss": 0.4707,
      "step": 763
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.99258738055632e-06,
      "loss": 0.5127,
      "step": 764
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.989929717232664e-06,
      "loss": 0.5426,
      "step": 765
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.987268946739354e-06,
      "loss": 0.5481,
      "step": 766
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.984605071148471e-06,
      "loss": 0.5295,
      "step": 767
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.981938092534516e-06,
      "loss": 0.5594,
      "step": 768
    },
    {
      "epoch": 0.43,
      "eval_loss": 0.504974365234375,
      "eval_runtime": 79.068,
      "eval_samples_per_second": 25.295,
      "eval_steps_per_second": 12.647,
      "step": 768
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.97926801297441e-06,
      "loss": 0.4932,
      "step": 769
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.976594834547482e-06,
      "loss": 0.4049,
      "step": 770
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.973918559335477e-06,
      "loss": 0.549,
      "step": 771
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.971239189422555e-06,
      "loss": 0.4987,
      "step": 772
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.968556726895282e-06,
      "loss": 0.4435,
      "step": 773
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.965871173842632e-06,
      "loss": 0.4575,
      "step": 774
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.963182532355992e-06,
      "loss": 0.4201,
      "step": 775
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.960490804529145e-06,
      "loss": 0.5323,
      "step": 776
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.957795992458285e-06,
      "loss": 0.4752,
      "step": 777
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.955098098242003e-06,
      "loss": 0.5505,
      "step": 778
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.952397123981293e-06,
      "loss": 0.5072,
      "step": 779
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.949693071779546e-06,
      "loss": 0.4484,
      "step": 780
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.946985943742553e-06,
      "loss": 0.5666,
      "step": 781
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.944275741978495e-06,
      "loss": 0.4926,
      "step": 782
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.941562468597951e-06,
      "loss": 0.4328,
      "step": 783
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.938846125713891e-06,
      "loss": 0.4424,
      "step": 784
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.936126715441675e-06,
      "loss": 0.4422,
      "step": 785
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.933404239899052e-06,
      "loss": 0.4216,
      "step": 786
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.93067870120616e-06,
      "loss": 0.4774,
      "step": 787
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.927950101485514e-06,
      "loss": 0.496,
      "step": 788
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.925218442862025e-06,
      "loss": 0.5359,
      "step": 789
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.922483727462978e-06,
      "loss": 0.4117,
      "step": 790
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.919745957418042e-06,
      "loss": 0.4577,
      "step": 791
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.917005134859263e-06,
      "loss": 0.5282,
      "step": 792
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.914261261921065e-06,
      "loss": 0.4537,
      "step": 793
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.911514340740246e-06,
      "loss": 0.445,
      "step": 794
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.908764373455981e-06,
      "loss": 0.4598,
      "step": 795
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.906011362209812e-06,
      "loss": 0.4584,
      "step": 796
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.90325530914566e-06,
      "loss": 0.6345,
      "step": 797
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.900496216409804e-06,
      "loss": 0.5062,
      "step": 798
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.897734086150897e-06,
      "loss": 0.427,
      "step": 799
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.89496892051996e-06,
      "loss": 0.5119,
      "step": 800
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.89220072167037e-06,
      "loss": 0.4858,
      "step": 801
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.889429491757872e-06,
      "loss": 0.4411,
      "step": 802
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.886655232940567e-06,
      "loss": 0.4341,
      "step": 803
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.88387794737892e-06,
      "loss": 0.5323,
      "step": 804
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.881097637235751e-06,
      "loss": 0.485,
      "step": 805
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.878314304676233e-06,
      "loss": 0.4557,
      "step": 806
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.875527951867895e-06,
      "loss": 0.4362,
      "step": 807
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.872738580980615e-06,
      "loss": 0.4202,
      "step": 808
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.86994619418663e-06,
      "loss": 0.4877,
      "step": 809
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.867150793660512e-06,
      "loss": 0.4149,
      "step": 810
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.864352381579196e-06,
      "loss": 0.6014,
      "step": 811
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.861550960121946e-06,
      "loss": 0.5263,
      "step": 812
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.85874653147038e-06,
      "loss": 0.4834,
      "step": 813
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.855939097808457e-06,
      "loss": 0.5017,
      "step": 814
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.853128661322473e-06,
      "loss": 0.4023,
      "step": 815
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.850315224201063e-06,
      "loss": 0.4753,
      "step": 816
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.8474987886352e-06,
      "loss": 0.5752,
      "step": 817
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.844679356818193e-06,
      "loss": 0.5207,
      "step": 818
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.84185693094568e-06,
      "loss": 0.5477,
      "step": 819
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.839031513215637e-06,
      "loss": 0.4876,
      "step": 820
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.836203105828364e-06,
      "loss": 0.504,
      "step": 821
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.833371710986493e-06,
      "loss": 0.3766,
      "step": 822
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.830537330894982e-06,
      "loss": 0.4425,
      "step": 823
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.827699967761108e-06,
      "loss": 0.427,
      "step": 824
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.824859623794484e-06,
      "loss": 0.4392,
      "step": 825
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.822016301207029e-06,
      "loss": 0.4516,
      "step": 826
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.819170002212992e-06,
      "loss": 0.4879,
      "step": 827
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.816320729028937e-06,
      "loss": 0.4256,
      "step": 828
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.813468483873745e-06,
      "loss": 0.5723,
      "step": 829
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.810613268968606e-06,
      "loss": 0.5121,
      "step": 830
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.807755086537032e-06,
      "loss": 0.4122,
      "step": 831
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.804893938804839e-06,
      "loss": 0.3747,
      "step": 832
    },
    {
      "epoch": 0.46,
      "eval_loss": 0.502413809299469,
      "eval_runtime": 79.0567,
      "eval_samples_per_second": 25.298,
      "eval_steps_per_second": 12.649,
      "step": 832
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.802029828000157e-06,
      "loss": 0.3845,
      "step": 833
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.799162756353415e-06,
      "loss": 0.4314,
      "step": 834
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.796292726097358e-06,
      "loss": 0.468,
      "step": 835
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.79341973946703e-06,
      "loss": 0.5254,
      "step": 836
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.790543798699778e-06,
      "loss": 0.3681,
      "step": 837
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.787664906035249e-06,
      "loss": 0.4798,
      "step": 838
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.784783063715388e-06,
      "loss": 0.3947,
      "step": 839
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.78189827398444e-06,
      "loss": 0.4217,
      "step": 840
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.779010539088943e-06,
      "loss": 0.4979,
      "step": 841
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.77611986127773e-06,
      "loss": 0.406,
      "step": 842
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.773226242801922e-06,
      "loss": 0.4741,
      "step": 843
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.770329685914938e-06,
      "loss": 0.5108,
      "step": 844
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.767430192872477e-06,
      "loss": 0.4786,
      "step": 845
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.764527765932527e-06,
      "loss": 0.5394,
      "step": 846
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.761622407355364e-06,
      "loss": 0.4309,
      "step": 847
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.758714119403542e-06,
      "loss": 0.3953,
      "step": 848
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.755802904341903e-06,
      "loss": 0.5524,
      "step": 849
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.752888764437561e-06,
      "loss": 0.4188,
      "step": 850
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.749971701959912e-06,
      "loss": 0.504,
      "step": 851
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.747051719180626e-06,
      "loss": 0.4233,
      "step": 852
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.744128818373652e-06,
      "loss": 0.4552,
      "step": 853
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.741203001815204e-06,
      "loss": 0.4908,
      "step": 854
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.738274271783772e-06,
      "loss": 0.442,
      "step": 855
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.73534263056011e-06,
      "loss": 0.4016,
      "step": 856
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.732408080427247e-06,
      "loss": 0.4578,
      "step": 857
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.729470623670472e-06,
      "loss": 0.4717,
      "step": 858
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.726530262577334e-06,
      "loss": 0.5276,
      "step": 859
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.723586999437655e-06,
      "loss": 0.3864,
      "step": 860
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.720640836543503e-06,
      "loss": 0.4352,
      "step": 861
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.717691776189214e-06,
      "loss": 0.5054,
      "step": 862
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.71473982067138e-06,
      "loss": 0.394,
      "step": 863
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.711784972288839e-06,
      "loss": 0.4152,
      "step": 864
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.708827233342696e-06,
      "loss": 0.515,
      "step": 865
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.70586660613629e-06,
      "loss": 0.4025,
      "step": 866
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.702903092975226e-06,
      "loss": 0.4176,
      "step": 867
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.699936696167346e-06,
      "loss": 0.5085,
      "step": 868
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.696967418022738e-06,
      "loss": 0.4368,
      "step": 869
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.69399526085374e-06,
      "loss": 0.498,
      "step": 870
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.691020226974926e-06,
      "loss": 0.4616,
      "step": 871
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.688042318703111e-06,
      "loss": 0.4799,
      "step": 872
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.685061538357356e-06,
      "loss": 0.3882,
      "step": 873
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.682077888258945e-06,
      "loss": 0.4237,
      "step": 874
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.679091370731408e-06,
      "loss": 0.6244,
      "step": 875
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.676101988100505e-06,
      "loss": 0.4264,
      "step": 876
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.673109742694227e-06,
      "loss": 0.4557,
      "step": 877
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.670114636842792e-06,
      "loss": 0.4589,
      "step": 878
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.66711667287865e-06,
      "loss": 0.4562,
      "step": 879
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.66411585313647e-06,
      "loss": 0.4146,
      "step": 880
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.661112179953155e-06,
      "loss": 0.4252,
      "step": 881
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.65810565566782e-06,
      "loss": 0.4908,
      "step": 882
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.655096282621806e-06,
      "loss": 0.4877,
      "step": 883
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.652084063158674e-06,
      "loss": 0.381,
      "step": 884
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.649068999624195e-06,
      "loss": 0.3952,
      "step": 885
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.646051094366361e-06,
      "loss": 0.3716,
      "step": 886
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.643030349735373e-06,
      "loss": 0.4655,
      "step": 887
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.640006768083646e-06,
      "loss": 0.4874,
      "step": 888
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.636980351765804e-06,
      "loss": 0.5176,
      "step": 889
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.633951103138675e-06,
      "loss": 0.4857,
      "step": 890
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.6309190245613e-06,
      "loss": 0.4058,
      "step": 891
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.627884118394913e-06,
      "loss": 0.3992,
      "step": 892
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.624846387002964e-06,
      "loss": 0.4105,
      "step": 893
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.621805832751087e-06,
      "loss": 0.4018,
      "step": 894
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.618762458007129e-06,
      "loss": 0.5145,
      "step": 895
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.615716265141123e-06,
      "loss": 0.4922,
      "step": 896
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.5009724497795105,
      "eval_runtime": 79.0048,
      "eval_samples_per_second": 25.315,
      "eval_steps_per_second": 12.657,
      "step": 896
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.612667256525305e-06,
      "loss": 0.5074,
      "step": 897
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.609615434534094e-06,
      "loss": 0.4199,
      "step": 898
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.60656080154411e-06,
      "loss": 0.5297,
      "step": 899
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.603503359934157e-06,
      "loss": 0.413,
      "step": 900
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.600443112085226e-06,
      "loss": 0.4565,
      "step": 901
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.597380060380493e-06,
      "loss": 0.4626,
      "step": 902
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.594314207205321e-06,
      "loss": 0.4675,
      "step": 903
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.59124555494725e-06,
      "loss": 0.4311,
      "step": 904
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.588174105996003e-06,
      "loss": 0.4565,
      "step": 905
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.58509986274348e-06,
      "loss": 0.4687,
      "step": 906
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.582022827583758e-06,
      "loss": 0.3826,
      "step": 907
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.578943002913086e-06,
      "loss": 0.4284,
      "step": 908
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.575860391129887e-06,
      "loss": 0.4999,
      "step": 909
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.572774994634755e-06,
      "loss": 0.4384,
      "step": 910
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.569686815830451e-06,
      "loss": 0.5591,
      "step": 911
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.566595857121902e-06,
      "loss": 0.507,
      "step": 912
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.563502120916202e-06,
      "loss": 0.4694,
      "step": 913
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.560405609622611e-06,
      "loss": 0.5248,
      "step": 914
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.557306325652542e-06,
      "loss": 0.422,
      "step": 915
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.554204271419573e-06,
      "loss": 0.547,
      "step": 916
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.551099449339438e-06,
      "loss": 0.4738,
      "step": 917
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.547991861830026e-06,
      "loss": 0.3952,
      "step": 918
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.54488151131138e-06,
      "loss": 0.7104,
      "step": 919
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.541768400205696e-06,
      "loss": 0.4882,
      "step": 920
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.53865253093732e-06,
      "loss": 0.6037,
      "step": 921
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.5965,
      "step": 922
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.532412527620595e-06,
      "loss": 0.4412,
      "step": 923
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.529288398431671e-06,
      "loss": 0.5441,
      "step": 924
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.526161520798892e-06,
      "loss": 0.4231,
      "step": 925
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.523031897157322e-06,
      "loss": 0.5148,
      "step": 926
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.519899529944166e-06,
      "loss": 0.4912,
      "step": 927
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.516764421598762e-06,
      "loss": 0.4143,
      "step": 928
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.513626574562588e-06,
      "loss": 0.4422,
      "step": 929
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.51048599127925e-06,
      "loss": 0.4825,
      "step": 930
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.507342674194486e-06,
      "loss": 0.4386,
      "step": 931
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.504196625756166e-06,
      "loss": 0.5102,
      "step": 932
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.501047848414282e-06,
      "loss": 0.3804,
      "step": 933
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.497896344620955e-06,
      "loss": 0.5053,
      "step": 934
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.494742116830428e-06,
      "loss": 0.4246,
      "step": 935
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.491585167499065e-06,
      "loss": 0.5186,
      "step": 936
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.48842549908535e-06,
      "loss": 0.4454,
      "step": 937
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.485263114049883e-06,
      "loss": 0.386,
      "step": 938
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.482098014855382e-06,
      "loss": 0.4822,
      "step": 939
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.478930203966677e-06,
      "loss": 0.4938,
      "step": 940
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.475759683850712e-06,
      "loss": 0.4549,
      "step": 941
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.472586456976534e-06,
      "loss": 0.4913,
      "step": 942
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.469410525815309e-06,
      "loss": 0.4477,
      "step": 943
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.4662318928403e-06,
      "loss": 0.4313,
      "step": 944
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.463050560526875e-06,
      "loss": 0.4206,
      "step": 945
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.45986653135251e-06,
      "loss": 0.4459,
      "step": 946
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.456679807796774e-06,
      "loss": 0.4615,
      "step": 947
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.453490392341339e-06,
      "loss": 0.4388,
      "step": 948
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.45029828746997e-06,
      "loss": 0.3408,
      "step": 949
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.447103495668531e-06,
      "loss": 0.5004,
      "step": 950
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.443906019424975e-06,
      "loss": 0.5057,
      "step": 951
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.440705861229344e-06,
      "loss": 0.5581,
      "step": 952
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.437503023573774e-06,
      "loss": 0.4501,
      "step": 953
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.43429750895248e-06,
      "loss": 0.4208,
      "step": 954
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.43108931986177e-06,
      "loss": 0.4856,
      "step": 955
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.427878458800028e-06,
      "loss": 0.461,
      "step": 956
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.424664928267724e-06,
      "loss": 0.3956,
      "step": 957
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.421448730767402e-06,
      "loss": 0.4776,
      "step": 958
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.418229868803689e-06,
      "loss": 0.5098,
      "step": 959
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.41500834488328e-06,
      "loss": 0.3946,
      "step": 960
    },
    {
      "epoch": 0.53,
      "eval_loss": 0.49997323751449585,
      "eval_runtime": 79.0697,
      "eval_samples_per_second": 25.294,
      "eval_steps_per_second": 12.647,
      "step": 960
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.411784161514949e-06,
      "loss": 0.6309,
      "step": 961
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.408557321209534e-06,
      "loss": 0.4127,
      "step": 962
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.405327826479951e-06,
      "loss": 0.4121,
      "step": 963
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.402095679841178e-06,
      "loss": 0.4841,
      "step": 964
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.398860883810256e-06,
      "loss": 0.469,
      "step": 965
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.395623440906295e-06,
      "loss": 0.54,
      "step": 966
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.392383353650463e-06,
      "loss": 0.4004,
      "step": 967
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.389140624565986e-06,
      "loss": 0.4693,
      "step": 968
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.385895256178149e-06,
      "loss": 0.4523,
      "step": 969
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.382647251014294e-06,
      "loss": 0.654,
      "step": 970
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.379396611603814e-06,
      "loss": 0.4355,
      "step": 971
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.376143340478153e-06,
      "loss": 0.4371,
      "step": 972
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.372887440170807e-06,
      "loss": 0.4495,
      "step": 973
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.369628913217319e-06,
      "loss": 0.6127,
      "step": 974
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.366367762155272e-06,
      "loss": 0.4929,
      "step": 975
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.363103989524302e-06,
      "loss": 0.4094,
      "step": 976
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.35983759786608e-06,
      "loss": 0.446,
      "step": 977
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.356568589724319e-06,
      "loss": 0.4032,
      "step": 978
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.353296967644766e-06,
      "loss": 0.5866,
      "step": 979
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.350022734175209e-06,
      "loss": 0.469,
      "step": 980
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.346745891865467e-06,
      "loss": 0.5413,
      "step": 981
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.34346644326739e-06,
      "loss": 0.4428,
      "step": 982
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.340184390934858e-06,
      "loss": 0.5432,
      "step": 983
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.33689973742378e-06,
      "loss": 0.4444,
      "step": 984
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.333612485292091e-06,
      "loss": 0.4367,
      "step": 985
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.330322637099745e-06,
      "loss": 0.5782,
      "step": 986
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.327030195408723e-06,
      "loss": 0.4174,
      "step": 987
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.323735162783023e-06,
      "loss": 0.5431,
      "step": 988
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.320437541788662e-06,
      "loss": 0.5407,
      "step": 989
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.317137334993672e-06,
      "loss": 0.3769,
      "step": 990
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.313834544968101e-06,
      "loss": 0.3686,
      "step": 991
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.310529174284004e-06,
      "loss": 0.5097,
      "step": 992
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.307221225515448e-06,
      "loss": 0.4515,
      "step": 993
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.303910701238508e-06,
      "loss": 0.4732,
      "step": 994
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.300597604031267e-06,
      "loss": 0.4239,
      "step": 995
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.297281936473806e-06,
      "loss": 0.532,
      "step": 996
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.293963701148215e-06,
      "loss": 0.4585,
      "step": 997
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.290642900638577e-06,
      "loss": 0.4016,
      "step": 998
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.287319537530976e-06,
      "loss": 0.4253,
      "step": 999
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.28399361441349e-06,
      "loss": 0.4045,
      "step": 1000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.280665133876193e-06,
      "loss": 0.6461,
      "step": 1001
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.277334098511147e-06,
      "loss": 0.4068,
      "step": 1002
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.274000510912407e-06,
      "loss": 0.4512,
      "step": 1003
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.270664373676017e-06,
      "loss": 0.37,
      "step": 1004
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.267325689399999e-06,
      "loss": 0.5165,
      "step": 1005
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.263984460684366e-06,
      "loss": 0.4661,
      "step": 1006
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.260640690131108e-06,
      "loss": 0.462,
      "step": 1007
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.257294380344199e-06,
      "loss": 0.4892,
      "step": 1008
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.253945533929585e-06,
      "loss": 0.48,
      "step": 1009
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.250594153495195e-06,
      "loss": 0.4191,
      "step": 1010
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.247240241650918e-06,
      "loss": 0.4479,
      "step": 1011
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.243883801008632e-06,
      "loss": 0.3947,
      "step": 1012
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.240524834182169e-06,
      "loss": 0.4222,
      "step": 1013
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.237163343787336e-06,
      "loss": 0.5066,
      "step": 1014
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.233799332441903e-06,
      "loss": 0.5237,
      "step": 1015
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.230432802765607e-06,
      "loss": 0.4713,
      "step": 1016
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.227063757380141e-06,
      "loss": 0.5122,
      "step": 1017
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.223692198909157e-06,
      "loss": 0.467,
      "step": 1018
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.22031812997827e-06,
      "loss": 0.4335,
      "step": 1019
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.216941553215043e-06,
      "loss": 0.4371,
      "step": 1020
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.213562471249e-06,
      "loss": 0.5562,
      "step": 1021
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.210180886711603e-06,
      "loss": 0.4283,
      "step": 1022
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.206796802236278e-06,
      "loss": 0.4516,
      "step": 1023
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.203410220458389e-06,
      "loss": 0.4622,
      "step": 1024
    },
    {
      "epoch": 0.57,
      "eval_loss": 0.4968355894088745,
      "eval_runtime": 79.0631,
      "eval_samples_per_second": 25.296,
      "eval_steps_per_second": 12.648,
      "step": 1024
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.200021144015247e-06,
      "loss": 0.4479,
      "step": 1025
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.196629575546104e-06,
      "loss": 0.4073,
      "step": 1026
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.193235517692154e-06,
      "loss": 0.5608,
      "step": 1027
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.189838973096532e-06,
      "loss": 0.381,
      "step": 1028
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.186439944404303e-06,
      "loss": 0.4881,
      "step": 1029
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.183038434262476e-06,
      "loss": 0.4857,
      "step": 1030
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.179634445319984e-06,
      "loss": 0.5112,
      "step": 1031
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.176227980227693e-06,
      "loss": 0.4367,
      "step": 1032
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.172819041638401e-06,
      "loss": 0.4327,
      "step": 1033
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.169407632206826e-06,
      "loss": 0.4172,
      "step": 1034
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.165993754589613e-06,
      "loss": 0.5877,
      "step": 1035
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.162577411445332e-06,
      "loss": 0.4797,
      "step": 1036
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.159158605434468e-06,
      "loss": 0.4082,
      "step": 1037
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.15573733921943e-06,
      "loss": 0.3613,
      "step": 1038
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.152313615464533e-06,
      "loss": 0.3938,
      "step": 1039
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.148887436836018e-06,
      "loss": 0.5206,
      "step": 1040
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.14545880600203e-06,
      "loss": 0.4458,
      "step": 1041
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.142027725632622e-06,
      "loss": 0.4768,
      "step": 1042
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.138594198399765e-06,
      "loss": 0.4599,
      "step": 1043
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.135158226977321e-06,
      "loss": 0.474,
      "step": 1044
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.13171981404107e-06,
      "loss": 0.5071,
      "step": 1045
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.12827896226868e-06,
      "loss": 0.4471,
      "step": 1046
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.12483567433973e-06,
      "loss": 0.4503,
      "step": 1047
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.121389952935687e-06,
      "loss": 0.3892,
      "step": 1048
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.11794180073992e-06,
      "loss": 0.4338,
      "step": 1049
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.114491220437684e-06,
      "loss": 0.4124,
      "step": 1050
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.11103821471613e-06,
      "loss": 0.4689,
      "step": 1051
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.107582786264299e-06,
      "loss": 0.4968,
      "step": 1052
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.104124937773112e-06,
      "loss": 0.4157,
      "step": 1053
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.10066467193538e-06,
      "loss": 0.4093,
      "step": 1054
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.0972019914458e-06,
      "loss": 0.4376,
      "step": 1055
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.09373689900094e-06,
      "loss": 0.5009,
      "step": 1056
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.09026939729925e-06,
      "loss": 0.433,
      "step": 1057
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.086799489041065e-06,
      "loss": 0.4158,
      "step": 1058
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.083327176928576e-06,
      "loss": 0.4419,
      "step": 1059
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.079852463665861e-06,
      "loss": 0.3399,
      "step": 1060
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.076375351958865e-06,
      "loss": 0.4379,
      "step": 1061
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.072895844515398e-06,
      "loss": 1.0759,
      "step": 1062
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.069413944045137e-06,
      "loss": 0.4376,
      "step": 1063
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.065929653259622e-06,
      "loss": 0.4842,
      "step": 1064
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.062442974872253e-06,
      "loss": 0.4847,
      "step": 1065
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.058953911598295e-06,
      "loss": 0.4131,
      "step": 1066
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.055462466154862e-06,
      "loss": 0.5139,
      "step": 1067
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.05196864126093e-06,
      "loss": 0.4698,
      "step": 1068
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.048472439637328e-06,
      "loss": 0.6227,
      "step": 1069
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.044973864006728e-06,
      "loss": 0.4018,
      "step": 1070
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.04147291709366e-06,
      "loss": 0.4721,
      "step": 1071
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.037969601624495e-06,
      "loss": 0.3703,
      "step": 1072
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.034463920327453e-06,
      "loss": 0.525,
      "step": 1073
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.030955875932592e-06,
      "loss": 0.3308,
      "step": 1074
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.02744547117181e-06,
      "loss": 0.5135,
      "step": 1075
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.023932708778848e-06,
      "loss": 0.5037,
      "step": 1076
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.020417591489279e-06,
      "loss": 0.5133,
      "step": 1077
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.016900122040512e-06,
      "loss": 0.4979,
      "step": 1078
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.013380303171786e-06,
      "loss": 0.4266,
      "step": 1079
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.009858137624171e-06,
      "loss": 0.4597,
      "step": 1080
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.006333628140564e-06,
      "loss": 0.4658,
      "step": 1081
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.002806777465685e-06,
      "loss": 0.4762,
      "step": 1082
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.999277588346083e-06,
      "loss": 0.51,
      "step": 1083
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.995746063530122e-06,
      "loss": 0.5668,
      "step": 1084
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.992212205767988e-06,
      "loss": 0.4348,
      "step": 1085
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.988676017811685e-06,
      "loss": 0.4274,
      "step": 1086
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.985137502415027e-06,
      "loss": 0.3972,
      "step": 1087
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.981596662333644e-06,
      "loss": 0.4296,
      "step": 1088
    },
    {
      "epoch": 0.61,
      "eval_loss": 0.49543190002441406,
      "eval_runtime": 79.0563,
      "eval_samples_per_second": 25.298,
      "eval_steps_per_second": 12.649,
      "step": 1088
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.978053500324978e-06,
      "loss": 0.5159,
      "step": 1089
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.974508019148277e-06,
      "loss": 0.5845,
      "step": 1090
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.970960221564592e-06,
      "loss": 0.3935,
      "step": 1091
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.967410110336782e-06,
      "loss": 0.4928,
      "step": 1092
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.963857688229509e-06,
      "loss": 0.4336,
      "step": 1093
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.96030295800923e-06,
      "loss": 0.5113,
      "step": 1094
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.956745922444202e-06,
      "loss": 0.4901,
      "step": 1095
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.953186584304476e-06,
      "loss": 0.4196,
      "step": 1096
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.9496249463619e-06,
      "loss": 0.4333,
      "step": 1097
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.946061011390108e-06,
      "loss": 0.4436,
      "step": 1098
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.942494782164527e-06,
      "loss": 0.3922,
      "step": 1099
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.5854,
      "step": 1100
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.935355452062625e-06,
      "loss": 0.4624,
      "step": 1101
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.931782356746076e-06,
      "loss": 0.5078,
      "step": 1102
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.928206978295284e-06,
      "loss": 0.4675,
      "step": 1103
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.924629319494583e-06,
      "loss": 0.3855,
      "step": 1104
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.921049383130083e-06,
      "loss": 0.4333,
      "step": 1105
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.917467171989677e-06,
      "loss": 0.3946,
      "step": 1106
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.913882688863015e-06,
      "loss": 0.3799,
      "step": 1107
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.910295936541528e-06,
      "loss": 0.466,
      "step": 1108
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.906706917818411e-06,
      "loss": 0.5435,
      "step": 1109
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.90311563548862e-06,
      "loss": 0.4728,
      "step": 1110
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.89952209234888e-06,
      "loss": 0.4511,
      "step": 1111
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.895926291197667e-06,
      "loss": 0.4434,
      "step": 1112
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.89232823483523e-06,
      "loss": 0.4419,
      "step": 1113
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.888727926063562e-06,
      "loss": 0.6481,
      "step": 1114
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.885125367686413e-06,
      "loss": 0.4306,
      "step": 1115
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.88152056250929e-06,
      "loss": 0.4244,
      "step": 1116
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.877913513339444e-06,
      "loss": 0.4517,
      "step": 1117
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.874304222985873e-06,
      "loss": 0.5515,
      "step": 1118
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.870692694259324e-06,
      "loss": 0.4579,
      "step": 1119
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.867078929972287e-06,
      "loss": 0.4716,
      "step": 1120
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.863462932938993e-06,
      "loss": 0.4773,
      "step": 1121
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.859844705975405e-06,
      "loss": 0.3817,
      "step": 1122
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.856224251899233e-06,
      "loss": 0.5636,
      "step": 1123
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.852601573529914e-06,
      "loss": 0.4406,
      "step": 1124
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.848976673688622e-06,
      "loss": 0.4908,
      "step": 1125
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.845349555198257e-06,
      "loss": 0.4338,
      "step": 1126
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.841720220883446e-06,
      "loss": 0.3604,
      "step": 1127
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.838088673570548e-06,
      "loss": 0.4637,
      "step": 1128
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.834454916087637e-06,
      "loss": 0.3874,
      "step": 1129
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.830818951264518e-06,
      "loss": 0.5322,
      "step": 1130
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.827180781932704e-06,
      "loss": 0.4614,
      "step": 1131
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.823540410925434e-06,
      "loss": 0.4935,
      "step": 1132
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.819897841077656e-06,
      "loss": 0.4765,
      "step": 1133
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.816253075226034e-06,
      "loss": 0.5037,
      "step": 1134
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.812606116208936e-06,
      "loss": 0.5034,
      "step": 1135
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.808956966866446e-06,
      "loss": 0.4319,
      "step": 1136
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.80530563004035e-06,
      "loss": 0.4039,
      "step": 1137
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.801652108574136e-06,
      "loss": 0.5039,
      "step": 1138
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.797996405312995e-06,
      "loss": 0.4602,
      "step": 1139
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.794338523103814e-06,
      "loss": 0.5567,
      "step": 1140
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.790678464795186e-06,
      "loss": 0.3797,
      "step": 1141
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.787016233237387e-06,
      "loss": 0.4737,
      "step": 1142
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.783351831282393e-06,
      "loss": 0.4798,
      "step": 1143
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.779685261783867e-06,
      "loss": 0.3969,
      "step": 1144
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.776016527597162e-06,
      "loss": 0.4099,
      "step": 1145
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.772345631579316e-06,
      "loss": 0.3877,
      "step": 1146
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.768672576589046e-06,
      "loss": 0.5284,
      "step": 1147
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.764997365486761e-06,
      "loss": 0.5848,
      "step": 1148
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.76132000113454e-06,
      "loss": 0.5105,
      "step": 1149
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.757640486396138e-06,
      "loss": 0.5121,
      "step": 1150
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.753958824136994e-06,
      "loss": 0.5719,
      "step": 1151
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.750275017224208e-06,
      "loss": 0.4529,
      "step": 1152
    },
    {
      "epoch": 0.64,
      "eval_loss": 0.49458709359169006,
      "eval_runtime": 79.0646,
      "eval_samples_per_second": 25.296,
      "eval_steps_per_second": 12.648,
      "step": 1152
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.746589068526557e-06,
      "loss": 0.4456,
      "step": 1153
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.742900980914487e-06,
      "loss": 0.3804,
      "step": 1154
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.739210757260107e-06,
      "loss": 0.437,
      "step": 1155
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.735518400437187e-06,
      "loss": 0.3498,
      "step": 1156
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.731823913321162e-06,
      "loss": 0.4302,
      "step": 1157
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.728127298789126e-06,
      "loss": 0.3999,
      "step": 1158
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.724428559719827e-06,
      "loss": 0.4317,
      "step": 1159
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.720727698993671e-06,
      "loss": 0.4866,
      "step": 1160
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.717024719492711e-06,
      "loss": 0.4725,
      "step": 1161
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.713319624100657e-06,
      "loss": 0.4561,
      "step": 1162
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.70961241570286e-06,
      "loss": 0.4244,
      "step": 1163
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.705903097186323e-06,
      "loss": 0.4528,
      "step": 1164
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.702191671439686e-06,
      "loss": 0.497,
      "step": 1165
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.698478141353233e-06,
      "loss": 0.3872,
      "step": 1166
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.69476250981889e-06,
      "loss": 0.4151,
      "step": 1167
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.691044779730211e-06,
      "loss": 0.4621,
      "step": 1168
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.687324953982394e-06,
      "loss": 0.4741,
      "step": 1169
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.68360303547226e-06,
      "loss": 0.3984,
      "step": 1170
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.679879027098268e-06,
      "loss": 0.4979,
      "step": 1171
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.676152931760496e-06,
      "loss": 0.3878,
      "step": 1172
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.67242475236066e-06,
      "loss": 0.4647,
      "step": 1173
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.668694491802083e-06,
      "loss": 0.4248,
      "step": 1174
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.664962152989717e-06,
      "loss": 0.463,
      "step": 1175
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.661227738830137e-06,
      "loss": 0.4919,
      "step": 1176
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.657491252231525e-06,
      "loss": 0.586,
      "step": 1177
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.653752696103683e-06,
      "loss": 0.4906,
      "step": 1178
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.650012073358023e-06,
      "loss": 0.4685,
      "step": 1179
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.646269386907564e-06,
      "loss": 0.3917,
      "step": 1180
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.642524639666938e-06,
      "loss": 0.4543,
      "step": 1181
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.638777834552372e-06,
      "loss": 0.5682,
      "step": 1182
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.635028974481706e-06,
      "loss": 0.5002,
      "step": 1183
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.631278062374377e-06,
      "loss": 0.3762,
      "step": 1184
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.627525101151414e-06,
      "loss": 0.4412,
      "step": 1185
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.623770093735452e-06,
      "loss": 0.4408,
      "step": 1186
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.620013043050713e-06,
      "loss": 0.5146,
      "step": 1187
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.61625395202301e-06,
      "loss": 0.4715,
      "step": 1188
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.612492823579744e-06,
      "loss": 0.4515,
      "step": 1189
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.6087296606499116e-06,
      "loss": 0.4075,
      "step": 1190
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.604964466164082e-06,
      "loss": 0.4631,
      "step": 1191
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.601197243054411e-06,
      "loss": 0.3973,
      "step": 1192
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.59742799425464e-06,
      "loss": 0.4643,
      "step": 1193
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.593656722700078e-06,
      "loss": 0.37,
      "step": 1194
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.589883431327613e-06,
      "loss": 0.4275,
      "step": 1195
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.586108123075713e-06,
      "loss": 0.4104,
      "step": 1196
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.582330800884405e-06,
      "loss": 0.4119,
      "step": 1197
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.578551467695293e-06,
      "loss": 0.4737,
      "step": 1198
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.574770126451542e-06,
      "loss": 0.4718,
      "step": 1199
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.570986780097884e-06,
      "loss": 0.4336,
      "step": 1200
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.567201431580612e-06,
      "loss": 0.4839,
      "step": 1201
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.563414083847573e-06,
      "loss": 0.4366,
      "step": 1202
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.559624739848183e-06,
      "loss": 0.625,
      "step": 1203
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.555833402533398e-06,
      "loss": 0.3674,
      "step": 1204
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.5520400748557355e-06,
      "loss": 0.4802,
      "step": 1205
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.548244759769261e-06,
      "loss": 0.3986,
      "step": 1206
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.544447460229587e-06,
      "loss": 0.4536,
      "step": 1207
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.540648179193875e-06,
      "loss": 0.4875,
      "step": 1208
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.53684691962082e-06,
      "loss": 0.4989,
      "step": 1209
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.533043684470671e-06,
      "loss": 0.5284,
      "step": 1210
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.529238476705203e-06,
      "loss": 0.4,
      "step": 1211
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.525431299287737e-06,
      "loss": 0.5043,
      "step": 1212
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.521622155183123e-06,
      "loss": 0.4732,
      "step": 1213
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.517811047357741e-06,
      "loss": 0.3518,
      "step": 1214
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.5139979787795046e-06,
      "loss": 0.5385,
      "step": 1215
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.51018295241785e-06,
      "loss": 0.5455,
      "step": 1216
    },
    {
      "epoch": 0.68,
      "eval_loss": 0.49334439635276794,
      "eval_runtime": 78.9943,
      "eval_samples_per_second": 25.318,
      "eval_steps_per_second": 12.659,
      "step": 1216
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.506365971243746e-06,
      "loss": 0.4587,
      "step": 1217
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.5025470382296725e-06,
      "loss": 0.4452,
      "step": 1218
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.498726156349636e-06,
      "loss": 0.4379,
      "step": 1219
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.494903328579161e-06,
      "loss": 0.3855,
      "step": 1220
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.491078557895285e-06,
      "loss": 0.4488,
      "step": 1221
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.4872518472765594e-06,
      "loss": 0.444,
      "step": 1222
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.483423199703049e-06,
      "loss": 0.4687,
      "step": 1223
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.479592618156321e-06,
      "loss": 0.5197,
      "step": 1224
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.475760105619455e-06,
      "loss": 0.4596,
      "step": 1225
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.471925665077029e-06,
      "loss": 0.5691,
      "step": 1226
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.4680892995151264e-06,
      "loss": 0.6313,
      "step": 1227
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.46425101192133e-06,
      "loss": 0.4174,
      "step": 1228
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.460410805284715e-06,
      "loss": 0.4197,
      "step": 1229
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.4565686825958525e-06,
      "loss": 0.5392,
      "step": 1230
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.452724646846811e-06,
      "loss": 0.4352,
      "step": 1231
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.4488787010311425e-06,
      "loss": 0.4175,
      "step": 1232
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.44503084814389e-06,
      "loss": 0.428,
      "step": 1233
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.441181091181579e-06,
      "loss": 0.4832,
      "step": 1234
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.43732943314222e-06,
      "loss": 0.4748,
      "step": 1235
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.433475877025302e-06,
      "loss": 0.5485,
      "step": 1236
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.429620425831795e-06,
      "loss": 0.4898,
      "step": 1237
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.425763082564143e-06,
      "loss": 0.5095,
      "step": 1238
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.421903850226263e-06,
      "loss": 0.4471,
      "step": 1239
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.418042731823544e-06,
      "loss": 0.3856,
      "step": 1240
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.414179730362842e-06,
      "loss": 0.4075,
      "step": 1241
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.4103148488524824e-06,
      "loss": 0.4797,
      "step": 1242
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.4064480903022515e-06,
      "loss": 0.6903,
      "step": 1243
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.402579457723402e-06,
      "loss": 0.4427,
      "step": 1244
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.3987089541286415e-06,
      "loss": 0.471,
      "step": 1245
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.394836582532136e-06,
      "loss": 0.4306,
      "step": 1246
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.390962345949506e-06,
      "loss": 0.4024,
      "step": 1247
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.387086247397826e-06,
      "loss": 0.4317,
      "step": 1248
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.383208289895617e-06,
      "loss": 0.5006,
      "step": 1249
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.379328476462853e-06,
      "loss": 0.4547,
      "step": 1250
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.375446810120948e-06,
      "loss": 0.4946,
      "step": 1251
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.371563293892761e-06,
      "loss": 0.5074,
      "step": 1252
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.367677930802595e-06,
      "loss": 0.5549,
      "step": 1253
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.363790723876186e-06,
      "loss": 0.5479,
      "step": 1254
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.359901676140707e-06,
      "loss": 0.4794,
      "step": 1255
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.356010790624767e-06,
      "loss": 0.5399,
      "step": 1256
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.3521180703584025e-06,
      "loss": 0.4504,
      "step": 1257
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.348223518373085e-06,
      "loss": 0.5727,
      "step": 1258
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.344327137701706e-06,
      "loss": 0.429,
      "step": 1259
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.3404289313785815e-06,
      "loss": 0.4925,
      "step": 1260
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.336528902439453e-06,
      "loss": 0.4118,
      "step": 1261
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.3326270539214826e-06,
      "loss": 0.4786,
      "step": 1262
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.328723388863244e-06,
      "loss": 0.4895,
      "step": 1263
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.324817910304728e-06,
      "loss": 0.5722,
      "step": 1264
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.320910621287336e-06,
      "loss": 0.3969,
      "step": 1265
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.3170015248538825e-06,
      "loss": 0.437,
      "step": 1266
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.3130906240485886e-06,
      "loss": 0.4701,
      "step": 1267
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.309177921917081e-06,
      "loss": 0.4369,
      "step": 1268
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.3052634215063855e-06,
      "loss": 0.4554,
      "step": 1269
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.301347125864931e-06,
      "loss": 0.5166,
      "step": 1270
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.297429038042548e-06,
      "loss": 0.3408,
      "step": 1271
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.293509161090453e-06,
      "loss": 0.4286,
      "step": 1272
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.2895874980612665e-06,
      "loss": 0.4327,
      "step": 1273
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.285664052008996e-06,
      "loss": 0.4011,
      "step": 1274
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.281738825989033e-06,
      "loss": 0.53,
      "step": 1275
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.277811823058163e-06,
      "loss": 0.5385,
      "step": 1276
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.273883046274547e-06,
      "loss": 0.4879,
      "step": 1277
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.4308,
      "step": 1278
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.266020183388651e-06,
      "loss": 0.4427,
      "step": 1279
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.262086103409598e-06,
      "loss": 0.5217,
      "step": 1280
    },
    {
      "epoch": 0.71,
      "eval_loss": 0.49165645241737366,
      "eval_runtime": 78.9963,
      "eval_samples_per_second": 25.318,
      "eval_steps_per_second": 12.659,
      "step": 1280
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.25815026182425e-06,
      "loss": 0.5713,
      "step": 1281
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.2542126616976596e-06,
      "loss": 0.4726,
      "step": 1282
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.250273306096243e-06,
      "loss": 0.5628,
      "step": 1283
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.2463321980877845e-06,
      "loss": 0.3487,
      "step": 1284
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.2423893407414335e-06,
      "loss": 0.405,
      "step": 1285
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.238444737127701e-06,
      "loss": 0.4786,
      "step": 1286
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.234498390318461e-06,
      "loss": 0.5226,
      "step": 1287
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.230550303386942e-06,
      "loss": 0.544,
      "step": 1288
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.226600479407728e-06,
      "loss": 0.4612,
      "step": 1289
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.2226489214567574e-06,
      "loss": 0.4325,
      "step": 1290
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.218695632611317e-06,
      "loss": 0.4665,
      "step": 1291
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.214740615950041e-06,
      "loss": 0.4193,
      "step": 1292
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.210783874552911e-06,
      "loss": 0.5188,
      "step": 1293
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.206825411501255e-06,
      "loss": 0.4224,
      "step": 1294
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.202865229877732e-06,
      "loss": 0.4951,
      "step": 1295
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.198903332766347e-06,
      "loss": 0.4333,
      "step": 1296
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.194939723252442e-06,
      "loss": 0.5102,
      "step": 1297
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.190974404422688e-06,
      "loss": 0.4639,
      "step": 1298
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.187007379365087e-06,
      "loss": 0.4319,
      "step": 1299
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.183038651168974e-06,
      "loss": 0.4255,
      "step": 1300
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.179068222925007e-06,
      "loss": 0.4235,
      "step": 1301
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.175096097725169e-06,
      "loss": 0.4696,
      "step": 1302
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.171122278662765e-06,
      "loss": 0.473,
      "step": 1303
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.167146768832418e-06,
      "loss": 0.4114,
      "step": 1304
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.163169571330066e-06,
      "loss": 0.4984,
      "step": 1305
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.159190689252966e-06,
      "loss": 0.5283,
      "step": 1306
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.155210125699683e-06,
      "loss": 0.4758,
      "step": 1307
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.151227883770094e-06,
      "loss": 0.4078,
      "step": 1308
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.147243966565381e-06,
      "loss": 0.3839,
      "step": 1309
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.143258377188031e-06,
      "loss": 0.4721,
      "step": 1310
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.139271118741835e-06,
      "loss": 0.5766,
      "step": 1311
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.135282194331881e-06,
      "loss": 0.4266,
      "step": 1312
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.131291607064555e-06,
      "loss": 0.5066,
      "step": 1313
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.127299360047541e-06,
      "loss": 0.536,
      "step": 1314
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.123305456389812e-06,
      "loss": 0.4678,
      "step": 1315
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.119309899201633e-06,
      "loss": 0.4715,
      "step": 1316
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.1153126915945535e-06,
      "loss": 0.4902,
      "step": 1317
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.111313836681415e-06,
      "loss": 0.4812,
      "step": 1318
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.107313337576335e-06,
      "loss": 0.405,
      "step": 1319
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.103311197394716e-06,
      "loss": 0.3706,
      "step": 1320
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.099307419253232e-06,
      "loss": 0.6804,
      "step": 1321
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.095302006269842e-06,
      "loss": 0.4172,
      "step": 1322
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.0912949615637724e-06,
      "loss": 0.5116,
      "step": 1323
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.08728628825552e-06,
      "loss": 0.4629,
      "step": 1324
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.083275989466852e-06,
      "loss": 0.6083,
      "step": 1325
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.0792640683208e-06,
      "loss": 0.5238,
      "step": 1326
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.07525052794166e-06,
      "loss": 0.5577,
      "step": 1327
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.071235371454988e-06,
      "loss": 0.5366,
      "step": 1328
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.067218601987598e-06,
      "loss": 0.3161,
      "step": 1329
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.063200222667566e-06,
      "loss": 0.4827,
      "step": 1330
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.059180236624209e-06,
      "loss": 0.3874,
      "step": 1331
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.05515864698811e-06,
      "loss": 0.4561,
      "step": 1332
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.0511354568910905e-06,
      "loss": 0.3867,
      "step": 1333
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.047110669466224e-06,
      "loss": 0.5818,
      "step": 1334
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.043084287847822e-06,
      "loss": 0.3442,
      "step": 1335
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.0390563151714455e-06,
      "loss": 0.5089,
      "step": 1336
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.035026754573888e-06,
      "loss": 0.4837,
      "step": 1337
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.030995609193181e-06,
      "loss": 0.4531,
      "step": 1338
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.026962882168593e-06,
      "loss": 0.4756,
      "step": 1339
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.022928576640622e-06,
      "loss": 0.5753,
      "step": 1340
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.018892695750991e-06,
      "loss": 0.5848,
      "step": 1341
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.014855242642662e-06,
      "loss": 0.4696,
      "step": 1342
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.010816220459808e-06,
      "loss": 0.4408,
      "step": 1343
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.006775632347832e-06,
      "loss": 0.434,
      "step": 1344
    },
    {
      "epoch": 0.75,
      "eval_loss": 0.48941290378570557,
      "eval_runtime": 79.0625,
      "eval_samples_per_second": 25.296,
      "eval_steps_per_second": 12.648,
      "step": 1344
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.002733481453352e-06,
      "loss": 0.4684,
      "step": 1345
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.998689770924207e-06,
      "loss": 0.6309,
      "step": 1346
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.994644503909449e-06,
      "loss": 0.4669,
      "step": 1347
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.990597683559338e-06,
      "loss": 0.4434,
      "step": 1348
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.986549313025351e-06,
      "loss": 0.611,
      "step": 1349
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.982499395460169e-06,
      "loss": 0.4371,
      "step": 1350
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.978447934017675e-06,
      "loss": 0.4609,
      "step": 1351
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.974394931852957e-06,
      "loss": 0.5696,
      "step": 1352
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.970340392122303e-06,
      "loss": 0.4799,
      "step": 1353
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.9662843179832e-06,
      "loss": 0.4745,
      "step": 1354
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.962226712594324e-06,
      "loss": 0.4129,
      "step": 1355
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.9581675791155505e-06,
      "loss": 0.502,
      "step": 1356
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.954106920707939e-06,
      "loss": 0.4251,
      "step": 1357
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.950044740533742e-06,
      "loss": 0.5216,
      "step": 1358
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.945981041756392e-06,
      "loss": 0.4818,
      "step": 1359
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.94191582754051e-06,
      "loss": 0.403,
      "step": 1360
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.93784910105189e-06,
      "loss": 0.467,
      "step": 1361
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.933780865457508e-06,
      "loss": 0.3722,
      "step": 1362
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.929711123925516e-06,
      "loss": 0.516,
      "step": 1363
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.925639879625236e-06,
      "loss": 0.5042,
      "step": 1364
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.921567135727165e-06,
      "loss": 0.4154,
      "step": 1365
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.917492895402959e-06,
      "loss": 0.3835,
      "step": 1366
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.4476,
      "step": 1367
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.909339938168625e-06,
      "loss": 0.4544,
      "step": 1368
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.905261227607632e-06,
      "loss": 0.4718,
      "step": 1369
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.901181033318782e-06,
      "loss": 0.3975,
      "step": 1370
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.89709935847954e-06,
      "loss": 0.433,
      "step": 1371
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.893016206268518e-06,
      "loss": 0.4605,
      "step": 1372
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.888931579865485e-06,
      "loss": 0.45,
      "step": 1373
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.884845482451357e-06,
      "loss": 0.4418,
      "step": 1374
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.880757917208195e-06,
      "loss": 0.3925,
      "step": 1375
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.8766688873191985e-06,
      "loss": 0.4268,
      "step": 1376
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.872578395968717e-06,
      "loss": 0.4647,
      "step": 1377
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.868486446342228e-06,
      "loss": 0.4837,
      "step": 1378
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.864393041626354e-06,
      "loss": 0.4731,
      "step": 1379
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.860298185008844e-06,
      "loss": 0.3435,
      "step": 1380
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.856201879678579e-06,
      "loss": 0.6136,
      "step": 1381
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.85210412882557e-06,
      "loss": 0.4892,
      "step": 1382
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.848004935640951e-06,
      "loss": 0.4354,
      "step": 1383
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.8439043033169825e-06,
      "loss": 0.4073,
      "step": 1384
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.839802235047046e-06,
      "loss": 0.5129,
      "step": 1385
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.835698734025635e-06,
      "loss": 0.4031,
      "step": 1386
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.831593803448366e-06,
      "loss": 0.5611,
      "step": 1387
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.827487446511967e-06,
      "loss": 0.4085,
      "step": 1388
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.823379666414272e-06,
      "loss": 0.4726,
      "step": 1389
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.819270466354231e-06,
      "loss": 0.6781,
      "step": 1390
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.8151598495318926e-06,
      "loss": 0.5026,
      "step": 1391
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.811047819148413e-06,
      "loss": 0.394,
      "step": 1392
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.806934378406049e-06,
      "loss": 0.394,
      "step": 1393
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.8028195305081535e-06,
      "loss": 0.4413,
      "step": 1394
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.798703278659179e-06,
      "loss": 0.5482,
      "step": 1395
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.7945856260646645e-06,
      "loss": 0.4122,
      "step": 1396
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.7904665759312475e-06,
      "loss": 0.642,
      "step": 1397
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.78634613146665e-06,
      "loss": 0.561,
      "step": 1398
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.782224295879682e-06,
      "loss": 0.4374,
      "step": 1399
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.778101072380231e-06,
      "loss": 0.3864,
      "step": 1400
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.773976464179274e-06,
      "loss": 0.4501,
      "step": 1401
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.769850474488859e-06,
      "loss": 0.3571,
      "step": 1402
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.765723106522114e-06,
      "loss": 0.4486,
      "step": 1403
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.761594363493238e-06,
      "loss": 0.4879,
      "step": 1404
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.757464248617505e-06,
      "loss": 0.4468,
      "step": 1405
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.7533327651112505e-06,
      "loss": 0.3945,
      "step": 1406
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.74919991619188e-06,
      "loss": 0.4992,
      "step": 1407
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.745065705077865e-06,
      "loss": 0.5255,
      "step": 1408
    },
    {
      "epoch": 0.78,
      "eval_loss": 0.488463819026947,
      "eval_runtime": 79.0636,
      "eval_samples_per_second": 25.296,
      "eval_steps_per_second": 12.648,
      "step": 1408
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.740930134988733e-06,
      "loss": 0.4916,
      "step": 1409
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.7367932091450695e-06,
      "loss": 0.3928,
      "step": 1410
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.73265493076852e-06,
      "loss": 0.4388,
      "step": 1411
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.728515303081782e-06,
      "loss": 0.5768,
      "step": 1412
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.724374329308602e-06,
      "loss": 0.5574,
      "step": 1413
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.720232012673773e-06,
      "loss": 0.4227,
      "step": 1414
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.716088356403141e-06,
      "loss": 0.4695,
      "step": 1415
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.711943363723588e-06,
      "loss": 0.4835,
      "step": 1416
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.70779703786304e-06,
      "loss": 0.4218,
      "step": 1417
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.703649382050459e-06,
      "loss": 0.4258,
      "step": 1418
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.699500399515847e-06,
      "loss": 0.5633,
      "step": 1419
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.695350093490235e-06,
      "loss": 0.5216,
      "step": 1420
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.691198467205686e-06,
      "loss": 0.5059,
      "step": 1421
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.687045523895292e-06,
      "loss": 0.4912,
      "step": 1422
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.68289126679317e-06,
      "loss": 0.4111,
      "step": 1423
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.67873569913446e-06,
      "loss": 0.4827,
      "step": 1424
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.674578824155322e-06,
      "loss": 0.4265,
      "step": 1425
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.670420645092936e-06,
      "loss": 0.4249,
      "step": 1426
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.666261165185496e-06,
      "loss": 0.4651,
      "step": 1427
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.662100387672208e-06,
      "loss": 0.6199,
      "step": 1428
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.657938315793291e-06,
      "loss": 0.4907,
      "step": 1429
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.653774952789971e-06,
      "loss": 0.3854,
      "step": 1430
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.649610301904481e-06,
      "loss": 0.3661,
      "step": 1431
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.64544436638005e-06,
      "loss": 0.4742,
      "step": 1432
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.641277149460918e-06,
      "loss": 0.4358,
      "step": 1433
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.637108654392316e-06,
      "loss": 0.4414,
      "step": 1434
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.632938884420471e-06,
      "loss": 0.3538,
      "step": 1435
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.628767842792605e-06,
      "loss": 0.4284,
      "step": 1436
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.6245955327569285e-06,
      "loss": 0.4779,
      "step": 1437
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.620421957562641e-06,
      "loss": 0.5848,
      "step": 1438
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.616247120459925e-06,
      "loss": 0.4742,
      "step": 1439
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.612071024699948e-06,
      "loss": 0.3899,
      "step": 1440
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.607893673534859e-06,
      "loss": 0.4604,
      "step": 1441
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.603715070217779e-06,
      "loss": 0.386,
      "step": 1442
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.59953521800281e-06,
      "loss": 0.4713,
      "step": 1443
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.595354120145026e-06,
      "loss": 0.4086,
      "step": 1444
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.591171779900467e-06,
      "loss": 0.4814,
      "step": 1445
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.586988200526143e-06,
      "loss": 0.4076,
      "step": 1446
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.58280338528003e-06,
      "loss": 0.4269,
      "step": 1447
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.578617337421065e-06,
      "loss": 0.3715,
      "step": 1448
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.574430060209144e-06,
      "loss": 0.4577,
      "step": 1449
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.570241556905124e-06,
      "loss": 0.4777,
      "step": 1450
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.566051830770813e-06,
      "loss": 0.4118,
      "step": 1451
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.561860885068972e-06,
      "loss": 0.4323,
      "step": 1452
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.557668723063313e-06,
      "loss": 0.5271,
      "step": 1453
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.553475348018496e-06,
      "loss": 0.4592,
      "step": 1454
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.5492807632001225e-06,
      "loss": 0.4402,
      "step": 1455
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.577,
      "step": 1456
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.540887977309829e-06,
      "loss": 0.4373,
      "step": 1457
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.536689782773816e-06,
      "loss": 0.5503,
      "step": 1458
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.532490391536057e-06,
      "loss": 0.4831,
      "step": 1459
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.528289806866838e-06,
      "loss": 0.5276,
      "step": 1460
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.524088032037381e-06,
      "loss": 0.5026,
      "step": 1461
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.519885070319827e-06,
      "loss": 0.4116,
      "step": 1462
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.515680924987244e-06,
      "loss": 0.4229,
      "step": 1463
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.511475599313627e-06,
      "loss": 0.456,
      "step": 1464
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.507269096573883e-06,
      "loss": 0.4871,
      "step": 1465
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.503061420043839e-06,
      "loss": 0.4992,
      "step": 1466
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.498852573000236e-06,
      "loss": 0.3877,
      "step": 1467
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.494642558720726e-06,
      "loss": 0.4326,
      "step": 1468
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.490431380483871e-06,
      "loss": 0.4138,
      "step": 1469
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.486219041569135e-06,
      "loss": 0.4574,
      "step": 1470
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.482005545256895e-06,
      "loss": 0.4502,
      "step": 1471
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.477790894828422e-06,
      "loss": 0.5583,
      "step": 1472
    },
    {
      "epoch": 0.82,
      "eval_loss": 0.48751842975616455,
      "eval_runtime": 79.0763,
      "eval_samples_per_second": 25.292,
      "eval_steps_per_second": 12.646,
      "step": 1472
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.473575093565884e-06,
      "loss": 0.4911,
      "step": 1473
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.469358144752352e-06,
      "loss": 0.4992,
      "step": 1474
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.465140051671787e-06,
      "loss": 0.3911,
      "step": 1475
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.460920817609041e-06,
      "loss": 0.6187,
      "step": 1476
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.456700445849857e-06,
      "loss": 0.5671,
      "step": 1477
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.45247893968086e-06,
      "loss": 0.4398,
      "step": 1478
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.4482563023895604e-06,
      "loss": 0.4364,
      "step": 1479
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.44403253726435e-06,
      "loss": 0.4036,
      "step": 1480
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.4398076475945e-06,
      "loss": 0.3898,
      "step": 1481
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.435581636670154e-06,
      "loss": 0.5082,
      "step": 1482
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.431354507782332e-06,
      "loss": 0.437,
      "step": 1483
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.427126264222921e-06,
      "loss": 0.5702,
      "step": 1484
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.422896909284679e-06,
      "loss": 0.464,
      "step": 1485
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.418666446261231e-06,
      "loss": 0.3991,
      "step": 1486
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.414434878447061e-06,
      "loss": 0.4089,
      "step": 1487
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.410202209137515e-06,
      "loss": 0.4743,
      "step": 1488
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.405968441628797e-06,
      "loss": 0.4559,
      "step": 1489
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.401733579217966e-06,
      "loss": 0.3916,
      "step": 1490
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.397497625202933e-06,
      "loss": 0.5026,
      "step": 1491
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.393260582882462e-06,
      "loss": 0.4787,
      "step": 1492
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.38902245555616e-06,
      "loss": 0.5419,
      "step": 1493
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.384783246524482e-06,
      "loss": 0.5211,
      "step": 1494
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.380542959088725e-06,
      "loss": 0.4329,
      "step": 1495
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.376301596551025e-06,
      "loss": 0.5373,
      "step": 1496
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.372059162214358e-06,
      "loss": 0.4461,
      "step": 1497
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.367815659382528e-06,
      "loss": 0.4189,
      "step": 1498
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.3635710913601785e-06,
      "loss": 0.6807,
      "step": 1499
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.35932546145278e-06,
      "loss": 0.4635,
      "step": 1500
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.355078772966627e-06,
      "loss": 0.4117,
      "step": 1501
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.350831029208844e-06,
      "loss": 0.4895,
      "step": 1502
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.346582233487371e-06,
      "loss": 0.4224,
      "step": 1503
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.342332389110971e-06,
      "loss": 0.4848,
      "step": 1504
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.338081499389224e-06,
      "loss": 0.3652,
      "step": 1505
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.3338295676325215e-06,
      "loss": 0.4416,
      "step": 1506
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.329576597152072e-06,
      "loss": 0.4245,
      "step": 1507
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.325322591259884e-06,
      "loss": 0.4525,
      "step": 1508
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.321067553268778e-06,
      "loss": 0.4369,
      "step": 1509
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.316811486492382e-06,
      "loss": 0.5112,
      "step": 1510
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.312554394245115e-06,
      "loss": 0.4335,
      "step": 1511
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.308296279842204e-06,
      "loss": 0.4807,
      "step": 1512
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.304037146599666e-06,
      "loss": 0.5819,
      "step": 1513
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.299776997834316e-06,
      "loss": 0.396,
      "step": 1514
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.295515836863755e-06,
      "loss": 0.3796,
      "step": 1515
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.291253667006377e-06,
      "loss": 0.4243,
      "step": 1516
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.28699049158136e-06,
      "loss": 0.4189,
      "step": 1517
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.2827263139086614e-06,
      "loss": 0.4517,
      "step": 1518
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.278461137309023e-06,
      "loss": 0.4383,
      "step": 1519
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.274194965103966e-06,
      "loss": 0.5542,
      "step": 1520
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.2699278006157836e-06,
      "loss": 0.4615,
      "step": 1521
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.265659647167542e-06,
      "loss": 0.5744,
      "step": 1522
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.261390508083081e-06,
      "loss": 0.5591,
      "step": 1523
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.257120386687002e-06,
      "loss": 0.4373,
      "step": 1524
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.252849286304676e-06,
      "loss": 0.4307,
      "step": 1525
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.248577210262235e-06,
      "loss": 0.3613,
      "step": 1526
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.244304161886574e-06,
      "loss": 0.3678,
      "step": 1527
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.240030144505339e-06,
      "loss": 0.3932,
      "step": 1528
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.235755161446933e-06,
      "loss": 0.5042,
      "step": 1529
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.231479216040512e-06,
      "loss": 0.427,
      "step": 1530
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.227202311615985e-06,
      "loss": 0.471,
      "step": 1531
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.222924451504001e-06,
      "loss": 0.4478,
      "step": 1532
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.218645639035955e-06,
      "loss": 0.429,
      "step": 1533
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.214365877543988e-06,
      "loss": 0.4227,
      "step": 1534
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.210085170360975e-06,
      "loss": 0.3987,
      "step": 1535
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.205803520820531e-06,
      "loss": 0.4137,
      "step": 1536
    },
    {
      "epoch": 0.86,
      "eval_loss": 0.4866933524608612,
      "eval_runtime": 79.0675,
      "eval_samples_per_second": 25.295,
      "eval_steps_per_second": 12.647,
      "step": 1536
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.2015209322570025e-06,
      "loss": 0.4201,
      "step": 1537
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.1972374080054675e-06,
      "loss": 0.3712,
      "step": 1538
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.192952951401735e-06,
      "loss": 0.4686,
      "step": 1539
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.1886675657823346e-06,
      "loss": 0.5032,
      "step": 1540
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.184381254484529e-06,
      "loss": 0.4106,
      "step": 1541
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.180094020846291e-06,
      "loss": 0.5976,
      "step": 1542
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.175805868206317e-06,
      "loss": 0.4167,
      "step": 1543
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.171516799904022e-06,
      "loss": 0.4771,
      "step": 1544
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.1672268192795285e-06,
      "loss": 0.459,
      "step": 1545
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.16293592967367e-06,
      "loss": 0.4181,
      "step": 1546
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.158644134427994e-06,
      "loss": 0.4465,
      "step": 1547
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.154351436884743e-06,
      "loss": 0.5784,
      "step": 1548
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.150057840386874e-06,
      "loss": 0.4164,
      "step": 1549
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.145763348278032e-06,
      "loss": 0.591,
      "step": 1550
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.141467963902571e-06,
      "loss": 0.4313,
      "step": 1551
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.1371716906055336e-06,
      "loss": 0.4164,
      "step": 1552
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.132874531732651e-06,
      "loss": 0.5759,
      "step": 1553
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.128576490630353e-06,
      "loss": 0.3884,
      "step": 1554
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.1242775706457505e-06,
      "loss": 0.4643,
      "step": 1555
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.1199777751266384e-06,
      "loss": 0.4743,
      "step": 1556
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.1156771074214995e-06,
      "loss": 0.4236,
      "step": 1557
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.111375570879488e-06,
      "loss": 0.3739,
      "step": 1558
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.107073168850439e-06,
      "loss": 0.4911,
      "step": 1559
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.1027699046848595e-06,
      "loss": 0.4368,
      "step": 1560
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.098465781733934e-06,
      "loss": 0.4548,
      "step": 1561
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.094160803349508e-06,
      "loss": 0.4463,
      "step": 1562
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.089854972884096e-06,
      "loss": 0.3895,
      "step": 1563
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.085548293690875e-06,
      "loss": 0.5945,
      "step": 1564
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.0812407691236874e-06,
      "loss": 0.5059,
      "step": 1565
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.076932402537029e-06,
      "loss": 0.355,
      "step": 1566
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.0726231972860535e-06,
      "loss": 0.4693,
      "step": 1567
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.068313156726568e-06,
      "loss": 0.4988,
      "step": 1568
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.064002284215027e-06,
      "loss": 0.4216,
      "step": 1569
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.059690583108539e-06,
      "loss": 0.4221,
      "step": 1570
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.05537805676485e-06,
      "loss": 0.5473,
      "step": 1571
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.051064708542357e-06,
      "loss": 0.3785,
      "step": 1572
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.046750541800089e-06,
      "loss": 0.4055,
      "step": 1573
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.0424355598977184e-06,
      "loss": 0.4534,
      "step": 1574
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.038119766195549e-06,
      "loss": 0.4937,
      "step": 1575
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.033803164054519e-06,
      "loss": 0.4004,
      "step": 1576
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.029485756836195e-06,
      "loss": 0.3485,
      "step": 1577
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.025167547902771e-06,
      "loss": 0.4675,
      "step": 1578
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.020848540617064e-06,
      "loss": 0.4024,
      "step": 1579
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.0165287383425145e-06,
      "loss": 0.4664,
      "step": 1580
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.01220814444318e-06,
      "loss": 0.4579,
      "step": 1581
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.00788676228374e-06,
      "loss": 0.4457,
      "step": 1582
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.00356459522948e-06,
      "loss": 0.4118,
      "step": 1583
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.999241646646301e-06,
      "loss": 0.4726,
      "step": 1584
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.994917919900711e-06,
      "loss": 0.3583,
      "step": 1585
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.990593418359828e-06,
      "loss": 0.4327,
      "step": 1586
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.986268145391369e-06,
      "loss": 0.4514,
      "step": 1587
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.981942104363653e-06,
      "loss": 0.4582,
      "step": 1588
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.977615298645597e-06,
      "loss": 0.4608,
      "step": 1589
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.973287731606716e-06,
      "loss": 0.4485,
      "step": 1590
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.968959406617113e-06,
      "loss": 0.4111,
      "step": 1591
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.964630327047485e-06,
      "loss": 0.3841,
      "step": 1592
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.9603004962691155e-06,
      "loss": 0.417,
      "step": 1593
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.955969917653874e-06,
      "loss": 0.4437,
      "step": 1594
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.95163859457421e-06,
      "loss": 0.753,
      "step": 1595
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.947306530403156e-06,
      "loss": 0.3783,
      "step": 1596
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.9429737285143185e-06,
      "loss": 0.3842,
      "step": 1597
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.938640192281883e-06,
      "loss": 0.3771,
      "step": 1598
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.934305925080597e-06,
      "loss": 0.4039,
      "step": 1599
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.92997093028579e-06,
      "loss": 0.3999,
      "step": 1600
    },
    {
      "epoch": 0.89,
      "eval_loss": 0.4855106472969055,
      "eval_runtime": 79.065,
      "eval_samples_per_second": 25.296,
      "eval_steps_per_second": 12.648,
      "step": 1600
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.925635211273351e-06,
      "loss": 0.3541,
      "step": 1601
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.921298771419731e-06,
      "loss": 0.532,
      "step": 1602
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.916961614101951e-06,
      "loss": 0.5708,
      "step": 1603
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.912623742697579e-06,
      "loss": 0.4266,
      "step": 1604
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.908285160584749e-06,
      "loss": 0.3768,
      "step": 1605
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.9039458711421435e-06,
      "loss": 0.4223,
      "step": 1606
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.8996058777489985e-06,
      "loss": 0.3968,
      "step": 1607
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.895265183785097e-06,
      "loss": 0.3859,
      "step": 1608
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.890923792630764e-06,
      "loss": 0.409,
      "step": 1609
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.8865817076668745e-06,
      "loss": 0.4364,
      "step": 1610
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.882238932274837e-06,
      "loss": 0.5137,
      "step": 1611
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.877895469836604e-06,
      "loss": 0.5755,
      "step": 1612
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.8735513237346585e-06,
      "loss": 0.496,
      "step": 1613
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.869206497352016e-06,
      "loss": 0.5097,
      "step": 1614
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.864860994072223e-06,
      "loss": 0.4512,
      "step": 1615
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.860514817279355e-06,
      "loss": 0.3992,
      "step": 1616
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.85616797035801e-06,
      "loss": 0.434,
      "step": 1617
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.851820456693307e-06,
      "loss": 0.5344,
      "step": 1618
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.847472279670885e-06,
      "loss": 0.4182,
      "step": 1619
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.843123442676899e-06,
      "loss": 0.4328,
      "step": 1620
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.838773949098021e-06,
      "loss": 0.4202,
      "step": 1621
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.8344238023214305e-06,
      "loss": 0.4788,
      "step": 1622
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.8300730057348186e-06,
      "loss": 0.4243,
      "step": 1623
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.82572156272638e-06,
      "loss": 0.3784,
      "step": 1624
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.821369476684814e-06,
      "loss": 0.4242,
      "step": 1625
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.817016750999321e-06,
      "loss": 0.4043,
      "step": 1626
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.8126633890595984e-06,
      "loss": 0.5093,
      "step": 1627
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.808309394255842e-06,
      "loss": 0.4947,
      "step": 1628
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.803954769978734e-06,
      "loss": 0.4161,
      "step": 1629
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.799599519619455e-06,
      "loss": 0.4541,
      "step": 1630
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.795243646569667e-06,
      "loss": 0.5662,
      "step": 1631
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.790887154221521e-06,
      "loss": 0.4003,
      "step": 1632
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.786530045967646e-06,
      "loss": 0.3762,
      "step": 1633
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.4051,
      "step": 1634
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.777813995315635e-06,
      "loss": 0.3985,
      "step": 1635
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.773455059705149e-06,
      "loss": 0.4577,
      "step": 1636
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.76909552176423e-06,
      "loss": 0.4084,
      "step": 1637
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.7647353848878825e-06,
      "loss": 0.3867,
      "step": 1638
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.760374652471576e-06,
      "loss": 0.4887,
      "step": 1639
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.756013327911242e-06,
      "loss": 0.4193,
      "step": 1640
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.751651414603278e-06,
      "loss": 0.3908,
      "step": 1641
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.747288915944533e-06,
      "loss": 0.4584,
      "step": 1642
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.74292583533232e-06,
      "loss": 0.4853,
      "step": 1643
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.738562176164399e-06,
      "loss": 0.4908,
      "step": 1644
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.734197941838984e-06,
      "loss": 0.3903,
      "step": 1645
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.729833135754732e-06,
      "loss": 0.5031,
      "step": 1646
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.725467761310751e-06,
      "loss": 0.4886,
      "step": 1647
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.721101821906592e-06,
      "loss": 0.3953,
      "step": 1648
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.7167353209422395e-06,
      "loss": 0.4238,
      "step": 1649
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.71236826181812e-06,
      "loss": 0.4743,
      "step": 1650
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.708000647935093e-06,
      "loss": 0.4382,
      "step": 1651
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.703632482694453e-06,
      "loss": 0.453,
      "step": 1652
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.6992637694979215e-06,
      "loss": 0.4232,
      "step": 1653
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.694894511747644e-06,
      "loss": 0.4411,
      "step": 1654
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.690524712846197e-06,
      "loss": 0.3937,
      "step": 1655
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.686154376196572e-06,
      "loss": 0.4248,
      "step": 1656
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.681783505202182e-06,
      "loss": 0.5285,
      "step": 1657
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.6774121032668565e-06,
      "loss": 0.4817,
      "step": 1658
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.673040173794836e-06,
      "loss": 0.4444,
      "step": 1659
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.668667720190776e-06,
      "loss": 0.5655,
      "step": 1660
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.664294745859734e-06,
      "loss": 0.4417,
      "step": 1661
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.659921254207183e-06,
      "loss": 0.427,
      "step": 1662
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.655547248638989e-06,
      "loss": 0.4148,
      "step": 1663
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.651172732561419e-06,
      "loss": 0.4806,
      "step": 1664
    },
    {
      "epoch": 0.93,
      "eval_loss": 0.4841112494468689,
      "eval_runtime": 79.003,
      "eval_samples_per_second": 25.316,
      "eval_steps_per_second": 12.658,
      "step": 1664
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.6467977093811444e-06,
      "loss": 0.4794,
      "step": 1665
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.642422182505227e-06,
      "loss": 0.4033,
      "step": 1666
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.638046155341121e-06,
      "loss": 0.501,
      "step": 1667
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.633669631296671e-06,
      "loss": 0.5882,
      "step": 1668
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.629292613780106e-06,
      "loss": 0.4682,
      "step": 1669
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.624915106200045e-06,
      "loss": 0.4699,
      "step": 1670
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.620537111965483e-06,
      "loss": 0.4465,
      "step": 1671
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.616158634485793e-06,
      "loss": 0.4153,
      "step": 1672
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.611779677170735e-06,
      "loss": 0.464,
      "step": 1673
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.607400243430428e-06,
      "loss": 0.5238,
      "step": 1674
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.603020336675372e-06,
      "loss": 0.5194,
      "step": 1675
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.598639960316432e-06,
      "loss": 0.5894,
      "step": 1676
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.59425911776484e-06,
      "loss": 0.4334,
      "step": 1677
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.589877812432189e-06,
      "loss": 0.3735,
      "step": 1678
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.5854960477304345e-06,
      "loss": 0.5154,
      "step": 1679
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.581113827071889e-06,
      "loss": 0.3754,
      "step": 1680
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.576731153869218e-06,
      "loss": 0.4164,
      "step": 1681
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.572348031535442e-06,
      "loss": 0.6036,
      "step": 1682
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.567964463483932e-06,
      "loss": 0.4684,
      "step": 1683
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.563580453128406e-06,
      "loss": 0.4153,
      "step": 1684
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.559196003882919e-06,
      "loss": 0.4932,
      "step": 1685
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.554811119161879e-06,
      "loss": 0.4632,
      "step": 1686
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.5504258023800286e-06,
      "loss": 0.6013,
      "step": 1687
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.546040056952443e-06,
      "loss": 0.498,
      "step": 1688
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.5416538862945375e-06,
      "loss": 0.4372,
      "step": 1689
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.537267293822054e-06,
      "loss": 0.4122,
      "step": 1690
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.532880282951064e-06,
      "loss": 0.6001,
      "step": 1691
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.528492857097966e-06,
      "loss": 0.4273,
      "step": 1692
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.524105019679481e-06,
      "loss": 0.4002,
      "step": 1693
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.51971677411265e-06,
      "loss": 0.4065,
      "step": 1694
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.515328123814834e-06,
      "loss": 0.4755,
      "step": 1695
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.510939072203704e-06,
      "loss": 0.4396,
      "step": 1696
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.506549622697251e-06,
      "loss": 0.4022,
      "step": 1697
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.502159778713769e-06,
      "loss": 0.4969,
      "step": 1698
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.497769543671865e-06,
      "loss": 0.4394,
      "step": 1699
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.493378920990443e-06,
      "loss": 0.5264,
      "step": 1700
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.4889879140887185e-06,
      "loss": 0.4291,
      "step": 1701
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.484596526386198e-06,
      "loss": 0.465,
      "step": 1702
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.480204761302688e-06,
      "loss": 0.4368,
      "step": 1703
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.475812622258292e-06,
      "loss": 0.6286,
      "step": 1704
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.471420112673397e-06,
      "loss": 0.4125,
      "step": 1705
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.467027235968684e-06,
      "loss": 0.432,
      "step": 1706
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.46263399556512e-06,
      "loss": 0.6613,
      "step": 1707
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.458240394883954e-06,
      "loss": 0.5218,
      "step": 1708
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.453846437346716e-06,
      "loss": 0.4571,
      "step": 1709
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.44945212637521e-06,
      "loss": 0.5158,
      "step": 1710
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.445057465391523e-06,
      "loss": 0.4752,
      "step": 1711
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.44066245781801e-06,
      "loss": 0.4911,
      "step": 1712
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.436267107077294e-06,
      "loss": 0.5847,
      "step": 1713
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.431871416592268e-06,
      "loss": 0.4705,
      "step": 1714
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.42747538978609e-06,
      "loss": 0.4396,
      "step": 1715
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.423079030082179e-06,
      "loss": 0.3881,
      "step": 1716
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.418682340904211e-06,
      "loss": 0.4572,
      "step": 1717
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.414285325676124e-06,
      "loss": 0.4426,
      "step": 1718
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.409887987822107e-06,
      "loss": 0.4033,
      "step": 1719
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.405490330766595e-06,
      "loss": 0.4772,
      "step": 1720
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.401092357934281e-06,
      "loss": 0.3757,
      "step": 1721
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.396694072750099e-06,
      "loss": 0.4384,
      "step": 1722
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.392295478639226e-06,
      "loss": 0.4637,
      "step": 1723
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.387896579027079e-06,
      "loss": 0.4069,
      "step": 1724
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.383497377339316e-06,
      "loss": 0.4781,
      "step": 1725
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.379097877001828e-06,
      "loss": 0.3822,
      "step": 1726
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.374698081440737e-06,
      "loss": 0.4408,
      "step": 1727
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.3702979940824e-06,
      "loss": 0.371,
      "step": 1728
    },
    {
      "epoch": 0.96,
      "eval_loss": 0.4832473695278168,
      "eval_runtime": 79.0679,
      "eval_samples_per_second": 25.295,
      "eval_steps_per_second": 12.647,
      "step": 1728
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.3658976183533984e-06,
      "loss": 0.4176,
      "step": 1729
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.361496957680535e-06,
      "loss": 0.5219,
      "step": 1730
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.357096015490838e-06,
      "loss": 0.5209,
      "step": 1731
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.352694795211555e-06,
      "loss": 0.3885,
      "step": 1732
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.348293300270151e-06,
      "loss": 0.5024,
      "step": 1733
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.3438915340943e-06,
      "loss": 0.4721,
      "step": 1734
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.339489500111893e-06,
      "loss": 0.4118,
      "step": 1735
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.335087201751026e-06,
      "loss": 0.6414,
      "step": 1736
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.330684642440003e-06,
      "loss": 0.4665,
      "step": 1737
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.3262818256073275e-06,
      "loss": 0.394,
      "step": 1738
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.32187875468171e-06,
      "loss": 0.4341,
      "step": 1739
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.317475433092054e-06,
      "loss": 0.5417,
      "step": 1740
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.313071864267456e-06,
      "loss": 0.5497,
      "step": 1741
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.308668051637213e-06,
      "loss": 0.5217,
      "step": 1742
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.304263998630804e-06,
      "loss": 0.4734,
      "step": 1743
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.299859708677901e-06,
      "loss": 0.4926,
      "step": 1744
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.295455185208355e-06,
      "loss": 0.5995,
      "step": 1745
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.291050431652203e-06,
      "loss": 0.4434,
      "step": 1746
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.28664545143966e-06,
      "loss": 0.329,
      "step": 1747
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.282240248001118e-06,
      "loss": 0.4623,
      "step": 1748
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.277834824767144e-06,
      "loss": 0.4014,
      "step": 1749
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.27342918516847e-06,
      "loss": 0.3895,
      "step": 1750
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.269023332636004e-06,
      "loss": 0.5486,
      "step": 1751
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.2646172706008154e-06,
      "loss": 0.4791,
      "step": 1752
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.26021100249414e-06,
      "loss": 0.5405,
      "step": 1753
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.255804531747371e-06,
      "loss": 0.3997,
      "step": 1754
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.251397861792061e-06,
      "loss": 0.3934,
      "step": 1755
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.2469909960599164e-06,
      "loss": 0.3693,
      "step": 1756
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.242583937982798e-06,
      "loss": 0.4616,
      "step": 1757
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.238176690992716e-06,
      "loss": 0.454,
      "step": 1758
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.233769258521825e-06,
      "loss": 0.4503,
      "step": 1759
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.229361644002429e-06,
      "loss": 0.4075,
      "step": 1760
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.224953850866967e-06,
      "loss": 0.4082,
      "step": 1761
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.220545882548024e-06,
      "loss": 0.4809,
      "step": 1762
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.2161377424783185e-06,
      "loss": 0.5632,
      "step": 1763
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.211729434090702e-06,
      "loss": 0.6447,
      "step": 1764
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.20732096081816e-06,
      "loss": 0.4682,
      "step": 1765
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.2029123260937985e-06,
      "loss": 0.4268,
      "step": 1766
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.198503533350859e-06,
      "loss": 0.4684,
      "step": 1767
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.1940945860227035e-06,
      "loss": 0.4492,
      "step": 1768
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.189685487542809e-06,
      "loss": 0.4802,
      "step": 1769
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.185276241344777e-06,
      "loss": 0.5059,
      "step": 1770
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.18086685086232e-06,
      "loss": 0.4059,
      "step": 1771
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.176457319529264e-06,
      "loss": 0.4225,
      "step": 1772
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.172047650779543e-06,
      "loss": 0.4073,
      "step": 1773
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.167637848047202e-06,
      "loss": 0.3704,
      "step": 1774
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.16322791476639e-06,
      "loss": 0.461,
      "step": 1775
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.15881785437135e-06,
      "loss": 0.5892,
      "step": 1776
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.154407670296434e-06,
      "loss": 0.4746,
      "step": 1777
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.149997365976085e-06,
      "loss": 0.4459,
      "step": 1778
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.1455869448448395e-06,
      "loss": 0.4559,
      "step": 1779
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.141176410337327e-06,
      "loss": 0.467,
      "step": 1780
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.1367657658882656e-06,
      "loss": 0.3881,
      "step": 1781
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.132355014932455e-06,
      "loss": 0.3435,
      "step": 1782
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.127944160904782e-06,
      "loss": 0.4764,
      "step": 1783
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.123533207240212e-06,
      "loss": 0.4607,
      "step": 1784
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.119122157373792e-06,
      "loss": 0.4577,
      "step": 1785
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.1147110147406344e-06,
      "loss": 0.4485,
      "step": 1786
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.1102997827759324e-06,
      "loss": 0.4513,
      "step": 1787
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1058884649149465e-06,
      "loss": 0.4062,
      "step": 1788
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1014770645930035e-06,
      "loss": 0.3971,
      "step": 1789
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.097065585245493e-06,
      "loss": 0.4959,
      "step": 1790
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.09265403030787e-06,
      "loss": 0.3373,
      "step": 1791
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.088242403215644e-06,
      "loss": 0.445,
      "step": 1792
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.48162105679512024,
      "eval_runtime": 79.0721,
      "eval_samples_per_second": 25.293,
      "eval_steps_per_second": 12.647,
      "step": 1792
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.083830707404382e-06,
      "loss": 0.4665,
      "step": 1793
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.07941894630971e-06,
      "loss": 0.5717,
      "step": 1794
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.075007123367296e-06,
      "loss": 0.5432,
      "step": 1795
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.070595242012861e-06,
      "loss": 0.3956,
      "step": 1796
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.06618330568217e-06,
      "loss": 0.4424,
      "step": 1797
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.061771317811033e-06,
      "loss": 0.379,
      "step": 1798
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.057359281835298e-06,
      "loss": 0.4075,
      "step": 1799
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.052947201190852e-06,
      "loss": 0.3503,
      "step": 1800
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.0485350793136155e-06,
      "loss": 0.4418,
      "step": 1801
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.0441229196395416e-06,
      "loss": 0.5076,
      "step": 1802
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.039710725604612e-06,
      "loss": 0.406,
      "step": 1803
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.035298500644835e-06,
      "loss": 0.5392,
      "step": 1804
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.030886248196249e-06,
      "loss": 0.4233,
      "step": 1805
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.026473971694902e-06,
      "loss": 0.4608,
      "step": 1806
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.022061674576871e-06,
      "loss": 0.4107,
      "step": 1807
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0176493602782465e-06,
      "loss": 0.4151,
      "step": 1808
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.013237032235129e-06,
      "loss": 0.4327,
      "step": 1809
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.008824693883635e-06,
      "loss": 0.383,
      "step": 1810
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.004412348659882e-06,
      "loss": 0.442,
      "step": 1811
    },
    {
      "epoch": 1.01,
      "learning_rate": 5e-06,
      "loss": 0.4637,
      "step": 1812
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.995587651340119e-06,
      "loss": 0.437,
      "step": 1813
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.991175306116368e-06,
      "loss": 0.5213,
      "step": 1814
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.986762967764872e-06,
      "loss": 0.4203,
      "step": 1815
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.982350639721754e-06,
      "loss": 0.3875,
      "step": 1816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.97793832542313e-06,
      "loss": 0.4118,
      "step": 1817
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.973526028305099e-06,
      "loss": 0.3781,
      "step": 1818
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.969113751803754e-06,
      "loss": 0.3544,
      "step": 1819
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.9647014993551665e-06,
      "loss": 0.4225,
      "step": 1820
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.960289274395391e-06,
      "loss": 0.3735,
      "step": 1821
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.955877080360462e-06,
      "loss": 0.4382,
      "step": 1822
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.951464920686384e-06,
      "loss": 0.4448,
      "step": 1823
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.947052798809148e-06,
      "loss": 0.5788,
      "step": 1824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.942640718164703e-06,
      "loss": 0.4622,
      "step": 1825
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.938228682188969e-06,
      "loss": 0.3957,
      "step": 1826
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.933816694317832e-06,
      "loss": 0.5289,
      "step": 1827
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.9294047579871415e-06,
      "loss": 0.4823,
      "step": 1828
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.924992876632707e-06,
      "loss": 0.5287,
      "step": 1829
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.920581053690291e-06,
      "loss": 0.4187,
      "step": 1830
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.916169292595618e-06,
      "loss": 0.5265,
      "step": 1831
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.911757596784358e-06,
      "loss": 0.295,
      "step": 1832
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.907345969692132e-06,
      "loss": 0.4476,
      "step": 1833
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.902934414754508e-06,
      "loss": 0.4152,
      "step": 1834
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.898522935406998e-06,
      "loss": 0.4682,
      "step": 1835
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.894111535085056e-06,
      "loss": 0.4039,
      "step": 1836
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.889700217224068e-06,
      "loss": 0.4159,
      "step": 1837
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.885288985259366e-06,
      "loss": 0.4638,
      "step": 1838
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.880877842626209e-06,
      "loss": 0.4095,
      "step": 1839
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.876466792759789e-06,
      "loss": 0.3816,
      "step": 1840
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.87205583909522e-06,
      "loss": 0.3787,
      "step": 1841
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.867644985067548e-06,
      "loss": 0.3846,
      "step": 1842
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.863234234111738e-06,
      "loss": 0.4268,
      "step": 1843
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.8588235896626736e-06,
      "loss": 0.3361,
      "step": 1844
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.854413055155162e-06,
      "loss": 0.5196,
      "step": 1845
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.850002634023917e-06,
      "loss": 0.514,
      "step": 1846
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.845592329703568e-06,
      "loss": 0.3857,
      "step": 1847
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.841182145628652e-06,
      "loss": 0.4017,
      "step": 1848
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.836772085233613e-06,
      "loss": 0.4456,
      "step": 1849
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.832362151952799e-06,
      "loss": 0.4021,
      "step": 1850
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.827952349220458e-06,
      "loss": 0.4035,
      "step": 1851
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.823542680470738e-06,
      "loss": 0.4473,
      "step": 1852
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.819133149137681e-06,
      "loss": 0.3988,
      "step": 1853
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.814723758655224e-06,
      "loss": 0.4668,
      "step": 1854
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.8103145124571924e-06,
      "loss": 0.3953,
      "step": 1855
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.8059054139773e-06,
      "loss": 0.377,
      "step": 1856
    },
    {
      "epoch": 1.03,
      "eval_loss": 0.4816921651363373,
      "eval_runtime": 78.992,
      "eval_samples_per_second": 25.319,
      "eval_steps_per_second": 12.66,
      "step": 1856
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.801496466649143e-06,
      "loss": 0.4473,
      "step": 1857
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.797087673906202e-06,
      "loss": 0.4279,
      "step": 1858
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.7926790391818425e-06,
      "loss": 0.3945,
      "step": 1859
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.788270565909299e-06,
      "loss": 0.4274,
      "step": 1860
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.783862257521683e-06,
      "loss": 0.372,
      "step": 1861
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.779454117451978e-06,
      "loss": 0.45,
      "step": 1862
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7750461491330355e-06,
      "loss": 0.4835,
      "step": 1863
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7706383559975744e-06,
      "loss": 0.5581,
      "step": 1864
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.766230741478176e-06,
      "loss": 0.3291,
      "step": 1865
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.761823309007286e-06,
      "loss": 0.4185,
      "step": 1866
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.757416062017203e-06,
      "loss": 0.4287,
      "step": 1867
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.753009003940084e-06,
      "loss": 0.452,
      "step": 1868
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.74860213820794e-06,
      "loss": 0.522,
      "step": 1869
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7441954682526304e-06,
      "loss": 0.4329,
      "step": 1870
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.739788997505861e-06,
      "loss": 0.4385,
      "step": 1871
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7353827293991845e-06,
      "loss": 0.4344,
      "step": 1872
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.730976667363997e-06,
      "loss": 0.4832,
      "step": 1873
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7265708148315305e-06,
      "loss": 0.4113,
      "step": 1874
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.722165175232858e-06,
      "loss": 0.3888,
      "step": 1875
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.717759751998883e-06,
      "loss": 0.5397,
      "step": 1876
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.713354548560342e-06,
      "loss": 0.4875,
      "step": 1877
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.7089495683477995e-06,
      "loss": 0.4555,
      "step": 1878
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.704544814791646e-06,
      "loss": 0.5812,
      "step": 1879
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.700140291322102e-06,
      "loss": 0.3515,
      "step": 1880
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.695736001369197e-06,
      "loss": 0.4913,
      "step": 1881
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.691331948362789e-06,
      "loss": 0.4367,
      "step": 1882
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.6869281357325455e-06,
      "loss": 0.3654,
      "step": 1883
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.682524566907949e-06,
      "loss": 0.4848,
      "step": 1884
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.678121245318291e-06,
      "loss": 0.4493,
      "step": 1885
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.673718174392673e-06,
      "loss": 0.3786,
      "step": 1886
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.669315357559999e-06,
      "loss": 0.5583,
      "step": 1887
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.664912798248975e-06,
      "loss": 0.3587,
      "step": 1888
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.660510499888109e-06,
      "loss": 0.4157,
      "step": 1889
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.656108465905702e-06,
      "loss": 0.4966,
      "step": 1890
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.651706699729851e-06,
      "loss": 0.3724,
      "step": 1891
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.647305204788445e-06,
      "loss": 0.3293,
      "step": 1892
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.642903984509163e-06,
      "loss": 0.3938,
      "step": 1893
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.6385030423194665e-06,
      "loss": 0.3774,
      "step": 1894
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.634102381646603e-06,
      "loss": 0.3741,
      "step": 1895
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.629702005917601e-06,
      "loss": 0.5314,
      "step": 1896
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.625301918559264e-06,
      "loss": 0.5001,
      "step": 1897
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.620902122998175e-06,
      "loss": 0.427,
      "step": 1898
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.616502622660684e-06,
      "loss": 0.3953,
      "step": 1899
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.612103420972922e-06,
      "loss": 0.4087,
      "step": 1900
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.6077045213607765e-06,
      "loss": 0.4102,
      "step": 1901
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.603305927249902e-06,
      "loss": 0.3437,
      "step": 1902
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.59890764206572e-06,
      "loss": 0.4181,
      "step": 1903
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.594509669233406e-06,
      "loss": 0.4848,
      "step": 1904
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.590112012177896e-06,
      "loss": 0.3539,
      "step": 1905
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.585714674323876e-06,
      "loss": 0.3929,
      "step": 1906
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.5813176590957896e-06,
      "loss": 0.4022,
      "step": 1907
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.576920969917822e-06,
      "loss": 0.4843,
      "step": 1908
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.5725246102139105e-06,
      "loss": 0.4022,
      "step": 1909
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.568128583407733e-06,
      "loss": 0.3719,
      "step": 1910
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.563732892922707e-06,
      "loss": 0.4042,
      "step": 1911
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.559337542181993e-06,
      "loss": 0.4194,
      "step": 1912
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.554942534608477e-06,
      "loss": 0.3634,
      "step": 1913
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.5505478736247895e-06,
      "loss": 0.3739,
      "step": 1914
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.546153562653286e-06,
      "loss": 0.4017,
      "step": 1915
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.541759605116048e-06,
      "loss": 0.4117,
      "step": 1916
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.537366004434882e-06,
      "loss": 0.4781,
      "step": 1917
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.5329727640313186e-06,
      "loss": 0.5535,
      "step": 1918
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.5285798873266055e-06,
      "loss": 0.4021,
      "step": 1919
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.52418737774171e-06,
      "loss": 0.453,
      "step": 1920
    },
    {
      "epoch": 1.07,
      "eval_loss": 0.48085153102874756,
      "eval_runtime": 79.0676,
      "eval_samples_per_second": 25.295,
      "eval_steps_per_second": 12.647,
      "step": 1920
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.5197952386973125e-06,
      "loss": 0.4051,
      "step": 1921
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.5154034736138035e-06,
      "loss": 0.4278,
      "step": 1922
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.511012085911283e-06,
      "loss": 0.4636,
      "step": 1923
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.506621079009558e-06,
      "loss": 0.3963,
      "step": 1924
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.502230456328137e-06,
      "loss": 0.4217,
      "step": 1925
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.497840221286232e-06,
      "loss": 0.5512,
      "step": 1926
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.49345037730275e-06,
      "loss": 0.4074,
      "step": 1927
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.489060927796297e-06,
      "loss": 0.4281,
      "step": 1928
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.484671876185168e-06,
      "loss": 0.3836,
      "step": 1929
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.480283225887351e-06,
      "loss": 0.4008,
      "step": 1930
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.4758949803205204e-06,
      "loss": 0.434,
      "step": 1931
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.471507142902036e-06,
      "loss": 0.4881,
      "step": 1932
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.467119717048939e-06,
      "loss": 0.389,
      "step": 1933
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.462732706177947e-06,
      "loss": 0.4859,
      "step": 1934
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.458346113705463e-06,
      "loss": 0.4694,
      "step": 1935
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.453959943047558e-06,
      "loss": 0.3541,
      "step": 1936
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.449574197619973e-06,
      "loss": 0.3903,
      "step": 1937
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.445188880838122e-06,
      "loss": 0.4797,
      "step": 1938
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.440803996117082e-06,
      "loss": 0.4371,
      "step": 1939
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.436419546871598e-06,
      "loss": 0.4162,
      "step": 1940
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.432035536516069e-06,
      "loss": 0.5207,
      "step": 1941
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.427651968464559e-06,
      "loss": 0.4158,
      "step": 1942
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.423268846130784e-06,
      "loss": 0.5328,
      "step": 1943
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.418886172928113e-06,
      "loss": 0.4018,
      "step": 1944
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.414503952269566e-06,
      "loss": 0.4081,
      "step": 1945
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.4101221875678115e-06,
      "loss": 0.4556,
      "step": 1946
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.4057408822351625e-06,
      "loss": 0.5473,
      "step": 1947
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.401360039683568e-06,
      "loss": 0.5222,
      "step": 1948
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.396979663324629e-06,
      "loss": 0.4642,
      "step": 1949
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.3925997565695725e-06,
      "loss": 0.5368,
      "step": 1950
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.388220322829266e-06,
      "loss": 0.3985,
      "step": 1951
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.383841365514208e-06,
      "loss": 0.446,
      "step": 1952
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.37946288803452e-06,
      "loss": 0.406,
      "step": 1953
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.375084893799957e-06,
      "loss": 0.4504,
      "step": 1954
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.370707386219894e-06,
      "loss": 0.5131,
      "step": 1955
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.366330368703331e-06,
      "loss": 0.4085,
      "step": 1956
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.3619538446588804e-06,
      "loss": 0.4454,
      "step": 1957
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.357577817494774e-06,
      "loss": 0.4056,
      "step": 1958
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.353202290618857e-06,
      "loss": 0.4309,
      "step": 1959
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.348827267438583e-06,
      "loss": 0.3432,
      "step": 1960
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.344452751361015e-06,
      "loss": 0.4149,
      "step": 1961
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.340078745792818e-06,
      "loss": 0.406,
      "step": 1962
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.335705254140266e-06,
      "loss": 0.3429,
      "step": 1963
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.331332279809226e-06,
      "loss": 0.4313,
      "step": 1964
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.326959826205166e-06,
      "loss": 0.5938,
      "step": 1965
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.322587896733146e-06,
      "loss": 0.5207,
      "step": 1966
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.3182164947978215e-06,
      "loss": 0.3405,
      "step": 1967
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.313845623803431e-06,
      "loss": 0.3439,
      "step": 1968
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.309475287153804e-06,
      "loss": 0.4294,
      "step": 1969
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.305105488252356e-06,
      "loss": 0.5882,
      "step": 1970
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.300736230502079e-06,
      "loss": 0.3668,
      "step": 1971
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.296367517305548e-06,
      "loss": 0.5344,
      "step": 1972
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.291999352064908e-06,
      "loss": 0.5059,
      "step": 1973
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.2876317381818825e-06,
      "loss": 0.4205,
      "step": 1974
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.283264679057763e-06,
      "loss": 0.4242,
      "step": 1975
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.278898178093409e-06,
      "loss": 0.4613,
      "step": 1976
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.274532238689248e-06,
      "loss": 0.4237,
      "step": 1977
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.270166864245269e-06,
      "loss": 0.4136,
      "step": 1978
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.265802058161018e-06,
      "loss": 0.4024,
      "step": 1979
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.261437823835602e-06,
      "loss": 0.4328,
      "step": 1980
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.257074164667681e-06,
      "loss": 0.3954,
      "step": 1981
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.252711084055468e-06,
      "loss": 0.3743,
      "step": 1982
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.248348585396723e-06,
      "loss": 0.4173,
      "step": 1983
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.243986672088758e-06,
      "loss": 0.3562,
      "step": 1984
    },
    {
      "epoch": 1.1,
      "eval_loss": 0.47997310757637024,
      "eval_runtime": 79.0711,
      "eval_samples_per_second": 25.294,
      "eval_steps_per_second": 12.647,
      "step": 1984
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.2396253475284255e-06,
      "loss": 0.4787,
      "step": 1985
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.235264615112118e-06,
      "loss": 0.4035,
      "step": 1986
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.230904478235772e-06,
      "loss": 0.457,
      "step": 1987
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.226544940294853e-06,
      "loss": 0.5198,
      "step": 1988
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.2221860046843675e-06,
      "loss": 0.3725,
      "step": 1989
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.3857,
      "step": 1990
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.2134699540323554e-06,
      "loss": 0.3889,
      "step": 1991
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.209112845778481e-06,
      "loss": 0.3812,
      "step": 1992
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.204756353430335e-06,
      "loss": 0.4699,
      "step": 1993
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.200400480380546e-06,
      "loss": 0.5201,
      "step": 1994
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.196045230021267e-06,
      "loss": 0.5845,
      "step": 1995
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.1916906057441605e-06,
      "loss": 0.3614,
      "step": 1996
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.187336610940402e-06,
      "loss": 0.6447,
      "step": 1997
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.18298324900068e-06,
      "loss": 0.4447,
      "step": 1998
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.178630523315187e-06,
      "loss": 0.4828,
      "step": 1999
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.174278437273621e-06,
      "loss": 0.3913,
      "step": 2000
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.169926994265182e-06,
      "loss": 0.3707,
      "step": 2001
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.165576197678571e-06,
      "loss": 0.4049,
      "step": 2002
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.161226050901982e-06,
      "loss": 0.5539,
      "step": 2003
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.156876557323102e-06,
      "loss": 0.4837,
      "step": 2004
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.152527720329117e-06,
      "loss": 0.3919,
      "step": 2005
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.148179543306695e-06,
      "loss": 0.5429,
      "step": 2006
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.143832029641992e-06,
      "loss": 0.4633,
      "step": 2007
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.139485182720646e-06,
      "loss": 0.3848,
      "step": 2008
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.1351390059277784e-06,
      "loss": 0.3277,
      "step": 2009
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.130793502647987e-06,
      "loss": 0.4524,
      "step": 2010
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.126448676265343e-06,
      "loss": 0.3681,
      "step": 2011
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.122104530163397e-06,
      "loss": 0.4642,
      "step": 2012
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.117761067725164e-06,
      "loss": 0.4107,
      "step": 2013
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.113418292333127e-06,
      "loss": 0.4543,
      "step": 2014
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.109076207369238e-06,
      "loss": 0.4383,
      "step": 2015
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.104734816214906e-06,
      "loss": 0.4136,
      "step": 2016
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.100394122251002e-06,
      "loss": 0.3795,
      "step": 2017
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.0960541288578564e-06,
      "loss": 0.4359,
      "step": 2018
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.091714839415251e-06,
      "loss": 0.3642,
      "step": 2019
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.087376257302422e-06,
      "loss": 0.4413,
      "step": 2020
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.083038385898052e-06,
      "loss": 0.4774,
      "step": 2021
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.0787012285802695e-06,
      "loss": 0.4419,
      "step": 2022
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.074364788726653e-06,
      "loss": 0.5124,
      "step": 2023
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.0700290697142126e-06,
      "loss": 0.4675,
      "step": 2024
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.065694074919404e-06,
      "loss": 0.4112,
      "step": 2025
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.061359807718119e-06,
      "loss": 0.4941,
      "step": 2026
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.057026271485682e-06,
      "loss": 0.418,
      "step": 2027
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.052693469596845e-06,
      "loss": 0.4794,
      "step": 2028
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.048361405425791e-06,
      "loss": 0.4338,
      "step": 2029
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.0440300823461275e-06,
      "loss": 0.466,
      "step": 2030
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.039699503730886e-06,
      "loss": 0.4134,
      "step": 2031
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.035369672952516e-06,
      "loss": 0.4531,
      "step": 2032
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.031040593382889e-06,
      "loss": 0.3404,
      "step": 2033
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.026712268393286e-06,
      "loss": 0.4218,
      "step": 2034
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.022384701354404e-06,
      "loss": 0.4377,
      "step": 2035
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.018057895636349e-06,
      "loss": 0.5106,
      "step": 2036
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.013731854608633e-06,
      "loss": 0.388,
      "step": 2037
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.009406581640172e-06,
      "loss": 0.4399,
      "step": 2038
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.005082080099289e-06,
      "loss": 0.4516,
      "step": 2039
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.0007583533537005e-06,
      "loss": 0.4758,
      "step": 2040
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.996435404770522e-06,
      "loss": 0.4409,
      "step": 2041
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.992113237716261e-06,
      "loss": 0.4207,
      "step": 2042
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.987791855556821e-06,
      "loss": 0.5178,
      "step": 2043
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.983471261657488e-06,
      "loss": 0.3711,
      "step": 2044
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.979151459382936e-06,
      "loss": 0.5076,
      "step": 2045
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.974832452097229e-06,
      "loss": 0.6966,
      "step": 2046
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.9705142431638065e-06,
      "loss": 0.493,
      "step": 2047
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.966196835945483e-06,
      "loss": 0.3696,
      "step": 2048
    },
    {
      "epoch": 1.14,
      "eval_loss": 0.47977104783058167,
      "eval_runtime": 79.043,
      "eval_samples_per_second": 25.303,
      "eval_steps_per_second": 12.651,
      "step": 2048
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.9618802338044525e-06,
      "loss": 0.3903,
      "step": 2049
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.957564440102284e-06,
      "loss": 0.3543,
      "step": 2050
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.9532494581999135e-06,
      "loss": 0.4051,
      "step": 2051
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.948935291457645e-06,
      "loss": 0.4427,
      "step": 2052
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.944621943235151e-06,
      "loss": 0.4806,
      "step": 2053
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.940309416891463e-06,
      "loss": 0.4045,
      "step": 2054
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.935997715784974e-06,
      "loss": 0.4803,
      "step": 2055
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.931686843273434e-06,
      "loss": 0.5534,
      "step": 2056
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.927376802713948e-06,
      "loss": 0.3297,
      "step": 2057
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.923067597462972e-06,
      "loss": 0.4267,
      "step": 2058
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.9187592308763125e-06,
      "loss": 0.3894,
      "step": 2059
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.914451706309125e-06,
      "loss": 0.4414,
      "step": 2060
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.9101450271159055e-06,
      "loss": 0.4517,
      "step": 2061
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.905839196650494e-06,
      "loss": 0.413,
      "step": 2062
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.9015342182660675e-06,
      "loss": 0.4178,
      "step": 2063
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.897230095315141e-06,
      "loss": 0.3267,
      "step": 2064
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.892926831149564e-06,
      "loss": 0.3495,
      "step": 2065
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.888624429120512e-06,
      "loss": 0.4084,
      "step": 2066
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.884322892578503e-06,
      "loss": 0.4686,
      "step": 2067
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.880022224873362e-06,
      "loss": 0.4278,
      "step": 2068
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.875722429354252e-06,
      "loss": 0.4053,
      "step": 2069
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.871423509369648e-06,
      "loss": 0.46,
      "step": 2070
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.86712546826735e-06,
      "loss": 0.4252,
      "step": 2071
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.862828309394469e-06,
      "loss": 0.498,
      "step": 2072
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8585320360974284e-06,
      "loss": 0.4111,
      "step": 2073
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.854236651721967e-06,
      "loss": 0.4937,
      "step": 2074
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.8499421596131285e-06,
      "loss": 0.5975,
      "step": 2075
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.845648563115258e-06,
      "loss": 0.4655,
      "step": 2076
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.841355865572009e-06,
      "loss": 0.4114,
      "step": 2077
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.837064070326331e-06,
      "loss": 0.4998,
      "step": 2078
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.832773180720475e-06,
      "loss": 0.4823,
      "step": 2079
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.828483200095978e-06,
      "loss": 0.3938,
      "step": 2080
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.824194131793682e-06,
      "loss": 0.4028,
      "step": 2081
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.8199059791537105e-06,
      "loss": 0.419,
      "step": 2082
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.815618745515474e-06,
      "loss": 0.4091,
      "step": 2083
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.8113324342176667e-06,
      "loss": 0.4174,
      "step": 2084
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.807047048598268e-06,
      "loss": 0.4743,
      "step": 2085
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.802762591994534e-06,
      "loss": 0.5111,
      "step": 2086
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.7984790677429988e-06,
      "loss": 0.4911,
      "step": 2087
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.79419647917947e-06,
      "loss": 0.4131,
      "step": 2088
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.789914829639026e-06,
      "loss": 0.4214,
      "step": 2089
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.7856341224560134e-06,
      "loss": 0.4049,
      "step": 2090
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.781354360964046e-06,
      "loss": 0.4048,
      "step": 2091
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.777075548496001e-06,
      "loss": 0.487,
      "step": 2092
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7727976883840166e-06,
      "loss": 0.4457,
      "step": 2093
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7685207839594873e-06,
      "loss": 0.4197,
      "step": 2094
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.764244838553068e-06,
      "loss": 0.5666,
      "step": 2095
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.759969855494663e-06,
      "loss": 0.4012,
      "step": 2096
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.755695838113427e-06,
      "loss": 0.4081,
      "step": 2097
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7514227897377654e-06,
      "loss": 0.3682,
      "step": 2098
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7471507136953257e-06,
      "loss": 0.3913,
      "step": 2099
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.742879613313001e-06,
      "loss": 0.3529,
      "step": 2100
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.73860949191692e-06,
      "loss": 0.4607,
      "step": 2101
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7343403528324574e-06,
      "loss": 0.4774,
      "step": 2102
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.730072199384218e-06,
      "loss": 0.3231,
      "step": 2103
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.725805034896035e-06,
      "loss": 0.4231,
      "step": 2104
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7215388626909786e-06,
      "loss": 0.4627,
      "step": 2105
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7172736860913415e-06,
      "loss": 0.4107,
      "step": 2106
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.713009508418643e-06,
      "loss": 0.5991,
      "step": 2107
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7087463329936235e-06,
      "loss": 0.3532,
      "step": 2108
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.704484163136245e-06,
      "loss": 0.3626,
      "step": 2109
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.7002230021656847e-06,
      "loss": 0.4203,
      "step": 2110
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6959628534003345e-06,
      "loss": 0.4163,
      "step": 2111
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6917037201577977e-06,
      "loss": 0.3842,
      "step": 2112
    },
    {
      "epoch": 1.18,
      "eval_loss": 0.47938331961631775,
      "eval_runtime": 79.0621,
      "eval_samples_per_second": 25.297,
      "eval_steps_per_second": 12.648,
      "step": 2112
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.687445605754886e-06,
      "loss": 0.4291,
      "step": 2113
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6831885135076212e-06,
      "loss": 0.4098,
      "step": 2114
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6789324467312213e-06,
      "loss": 0.5303,
      "step": 2115
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.674677408740117e-06,
      "loss": 0.4741,
      "step": 2116
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6704234028479296e-06,
      "loss": 0.4727,
      "step": 2117
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6661704323674798e-06,
      "loss": 0.3629,
      "step": 2118
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6619185006107783e-06,
      "loss": 0.4881,
      "step": 2119
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.657667610889032e-06,
      "loss": 0.3419,
      "step": 2120
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6534177665126323e-06,
      "loss": 0.4587,
      "step": 2121
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.649168970791157e-06,
      "loss": 0.4485,
      "step": 2122
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.644921227033374e-06,
      "loss": 0.5088,
      "step": 2123
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.640674538547222e-06,
      "loss": 0.4387,
      "step": 2124
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6364289086398223e-06,
      "loss": 0.4418,
      "step": 2125
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6321843406174735e-06,
      "loss": 0.3886,
      "step": 2126
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6279408377856445e-06,
      "loss": 0.3471,
      "step": 2127
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.6236984034489765e-06,
      "loss": 0.474,
      "step": 2128
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.6194570409112756e-06,
      "loss": 0.4932,
      "step": 2129
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.615216753475519e-06,
      "loss": 0.3981,
      "step": 2130
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.610977544443841e-06,
      "loss": 0.3767,
      "step": 2131
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.6067394171175397e-06,
      "loss": 0.3832,
      "step": 2132
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.602502374797068e-06,
      "loss": 0.5161,
      "step": 2133
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.598266420782037e-06,
      "loss": 0.399,
      "step": 2134
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.594031558371206e-06,
      "loss": 0.4788,
      "step": 2135
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5897977908624855e-06,
      "loss": 0.4923,
      "step": 2136
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5855651215529397e-06,
      "loss": 0.4557,
      "step": 2137
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5813335537387707e-06,
      "loss": 0.3677,
      "step": 2138
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.577103090715322e-06,
      "loss": 0.3835,
      "step": 2139
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5728737357770814e-06,
      "loss": 0.509,
      "step": 2140
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5686454922176712e-06,
      "loss": 0.3728,
      "step": 2141
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.564418363329848e-06,
      "loss": 0.3659,
      "step": 2142
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5601923524055014e-06,
      "loss": 0.4407,
      "step": 2143
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.555967462735651e-06,
      "loss": 0.417,
      "step": 2144
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.5517436976104412e-06,
      "loss": 0.4087,
      "step": 2145
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.547521060319142e-06,
      "loss": 0.3435,
      "step": 2146
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5432995541501445e-06,
      "loss": 0.448,
      "step": 2147
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5390791823909597e-06,
      "loss": 0.4089,
      "step": 2148
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.534859948328215e-06,
      "loss": 0.4478,
      "step": 2149
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5306418552476483e-06,
      "loss": 0.4113,
      "step": 2150
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5264249064341173e-06,
      "loss": 0.4561,
      "step": 2151
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5222091051715803e-06,
      "loss": 0.4303,
      "step": 2152
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5179944547431054e-06,
      "loss": 0.4262,
      "step": 2153
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5137809584308656e-06,
      "loss": 0.4086,
      "step": 2154
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5095686195161315e-06,
      "loss": 0.403,
      "step": 2155
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5053574412792757e-06,
      "loss": 0.5776,
      "step": 2156
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.501147426999764e-06,
      "loss": 0.4188,
      "step": 2157
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.4969385799561623e-06,
      "loss": 0.4546,
      "step": 2158
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.4927309034261186e-06,
      "loss": 0.4383,
      "step": 2159
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.488524400686375e-06,
      "loss": 0.5172,
      "step": 2160
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.4843190750127577e-06,
      "loss": 0.5023,
      "step": 2161
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.480114929680176e-06,
      "loss": 0.3669,
      "step": 2162
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.475911967962622e-06,
      "loss": 0.4227,
      "step": 2163
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.4717101931331622e-06,
      "loss": 0.4871,
      "step": 2164
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.467509608463945e-06,
      "loss": 0.4419,
      "step": 2165
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.463310217226185e-06,
      "loss": 0.3818,
      "step": 2166
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4591120226901724e-06,
      "loss": 0.4711,
      "step": 2167
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.4019,
      "step": 2168
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.45071923679988e-06,
      "loss": 0.3963,
      "step": 2169
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4465246519815076e-06,
      "loss": 0.3923,
      "step": 2170
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.442331276936688e-06,
      "loss": 0.4705,
      "step": 2171
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4381391149310294e-06,
      "loss": 0.3239,
      "step": 2172
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.433948169229189e-06,
      "loss": 0.4513,
      "step": 2173
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.429758443094878e-06,
      "loss": 0.3776,
      "step": 2174
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.425569939790857e-06,
      "loss": 0.483,
      "step": 2175
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.421382662578937e-06,
      "loss": 0.3702,
      "step": 2176
    },
    {
      "epoch": 1.21,
      "eval_loss": 0.47821933031082153,
      "eval_runtime": 79.0597,
      "eval_samples_per_second": 25.297,
      "eval_steps_per_second": 12.649,
      "step": 2176
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.417196614719972e-06,
      "loss": 0.3529,
      "step": 2177
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4130117994738577e-06,
      "loss": 0.4366,
      "step": 2178
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.408828220099534e-06,
      "loss": 0.4435,
      "step": 2179
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4046458798549752e-06,
      "loss": 0.3455,
      "step": 2180
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.4004647819971902e-06,
      "loss": 0.4157,
      "step": 2181
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.3962849297822225e-06,
      "loss": 0.3265,
      "step": 2182
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.392106326465144e-06,
      "loss": 0.4165,
      "step": 2183
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3879289753000533e-06,
      "loss": 0.491,
      "step": 2184
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3837528795400766e-06,
      "loss": 0.4326,
      "step": 2185
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3795780424373608e-06,
      "loss": 0.3625,
      "step": 2186
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.375404467243073e-06,
      "loss": 0.4439,
      "step": 2187
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3712321572073957e-06,
      "loss": 0.4531,
      "step": 2188
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3670611155795298e-06,
      "loss": 0.3881,
      "step": 2189
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3628913456076863e-06,
      "loss": 0.3905,
      "step": 2190
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3587228505390825e-06,
      "loss": 0.4072,
      "step": 2191
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.35455563361995e-06,
      "loss": 0.4447,
      "step": 2192
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.350389698095521e-06,
      "loss": 0.3538,
      "step": 2193
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.34622504721003e-06,
      "loss": 0.4542,
      "step": 2194
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3420616842067104e-06,
      "loss": 0.4822,
      "step": 2195
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.337899612327794e-06,
      "loss": 0.456,
      "step": 2196
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.333738834814506e-06,
      "loss": 0.4579,
      "step": 2197
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.329579354907065e-06,
      "loss": 0.5906,
      "step": 2198
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.3254211758446784e-06,
      "loss": 0.5003,
      "step": 2199
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.321264300865541e-06,
      "loss": 0.4051,
      "step": 2200
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.317108733206831e-06,
      "loss": 0.4285,
      "step": 2201
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.3129544761047093e-06,
      "loss": 0.56,
      "step": 2202
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.3088015327943158e-06,
      "loss": 0.4704,
      "step": 2203
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.304649906509767e-06,
      "loss": 0.4086,
      "step": 2204
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.3004996004841537e-06,
      "loss": 0.4284,
      "step": 2205
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.296350617949542e-06,
      "loss": 0.4114,
      "step": 2206
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.292202962136962e-06,
      "loss": 0.4657,
      "step": 2207
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.2880566362764133e-06,
      "loss": 0.425,
      "step": 2208
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.28391164359686e-06,
      "loss": 0.3524,
      "step": 2209
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.2797679873262283e-06,
      "loss": 0.3081,
      "step": 2210
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.275625670691401e-06,
      "loss": 0.5051,
      "step": 2211
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.271484696918218e-06,
      "loss": 0.3896,
      "step": 2212
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.2673450692314797e-06,
      "loss": 0.4996,
      "step": 2213
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.263206790854932e-06,
      "loss": 0.3889,
      "step": 2214
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.2590698650112695e-06,
      "loss": 0.3673,
      "step": 2215
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.254934294922137e-06,
      "loss": 0.4506,
      "step": 2216
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.250800083808121e-06,
      "loss": 0.4594,
      "step": 2217
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.2466672348887524e-06,
      "loss": 0.4785,
      "step": 2218
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2425357513824973e-06,
      "loss": 0.4297,
      "step": 2219
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2384056365067628e-06,
      "loss": 0.3872,
      "step": 2220
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.234276893477888e-06,
      "loss": 0.2984,
      "step": 2221
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2301495255111426e-06,
      "loss": 0.3666,
      "step": 2222
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.226023535820728e-06,
      "loss": 0.4306,
      "step": 2223
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.22189892761977e-06,
      "loss": 0.4672,
      "step": 2224
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2177757041203213e-06,
      "loss": 0.4133,
      "step": 2225
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2136538685333497e-06,
      "loss": 0.4244,
      "step": 2226
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2095334240687524e-06,
      "loss": 0.358,
      "step": 2227
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.205414373935336e-06,
      "loss": 0.3794,
      "step": 2228
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.201296721340823e-06,
      "loss": 0.4733,
      "step": 2229
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1971804694918477e-06,
      "loss": 0.3562,
      "step": 2230
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1930656215939527e-06,
      "loss": 0.3395,
      "step": 2231
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1889521808515888e-06,
      "loss": 0.4555,
      "step": 2232
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1848401504681074e-06,
      "loss": 0.4442,
      "step": 2233
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.180729533645771e-06,
      "loss": 0.6133,
      "step": 2234
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1766203335857285e-06,
      "loss": 0.3892,
      "step": 2235
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.172512553488035e-06,
      "loss": 0.4528,
      "step": 2236
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.168406196551635e-06,
      "loss": 0.4993,
      "step": 2237
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1643012659743667e-06,
      "loss": 0.4625,
      "step": 2238
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.160197764952957e-06,
      "loss": 0.3343,
      "step": 2239
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.156095696683018e-06,
      "loss": 0.3385,
      "step": 2240
    },
    {
      "epoch": 1.25,
      "eval_loss": 0.47787922620773315,
      "eval_runtime": 79.0724,
      "eval_samples_per_second": 25.293,
      "eval_steps_per_second": 12.647,
      "step": 2240
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1519950643590504e-06,
      "loss": 0.4377,
      "step": 2241
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1478958711744324e-06,
      "loss": 0.3927,
      "step": 2242
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.143798120321423e-06,
      "loss": 0.4056,
      "step": 2243
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.139701814991158e-06,
      "loss": 0.3573,
      "step": 2244
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.135606958373648e-06,
      "loss": 0.4155,
      "step": 2245
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1315135536577735e-06,
      "loss": 0.4297,
      "step": 2246
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.127421604031284e-06,
      "loss": 0.4911,
      "step": 2247
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1233311126808015e-06,
      "loss": 0.3956,
      "step": 2248
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.119242082791807e-06,
      "loss": 0.3825,
      "step": 2249
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1151545175486443e-06,
      "loss": 0.4265,
      "step": 2250
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1110684201345164e-06,
      "loss": 0.4214,
      "step": 2251
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1069837937314846e-06,
      "loss": 0.3945,
      "step": 2252
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.1029006415204633e-06,
      "loss": 0.3982,
      "step": 2253
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.0988189666812185e-06,
      "loss": 0.3839,
      "step": 2254
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0947387723923696e-06,
      "loss": 0.4731,
      "step": 2255
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0906600618313774e-06,
      "loss": 0.4306,
      "step": 2256
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.427,
      "step": 2257
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.082507104597041e-06,
      "loss": 0.4002,
      "step": 2258
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0784328642728368e-06,
      "loss": 0.4924,
      "step": 2259
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.074360120374764e-06,
      "loss": 0.4624,
      "step": 2260
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.070288876074484e-06,
      "loss": 0.585,
      "step": 2261
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0662191345424925e-06,
      "loss": 0.4784,
      "step": 2262
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.062150898948111e-06,
      "loss": 0.4554,
      "step": 2263
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.058084172459491e-06,
      "loss": 0.3408,
      "step": 2264
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.054018958243609e-06,
      "loss": 0.639,
      "step": 2265
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0499552594662603e-06,
      "loss": 0.3945,
      "step": 2266
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.045893079292063e-06,
      "loss": 0.4473,
      "step": 2267
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0418324208844503e-06,
      "loss": 0.4531,
      "step": 2268
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0377732874056764e-06,
      "loss": 0.5022,
      "step": 2269
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.033715682016802e-06,
      "loss": 0.4481,
      "step": 2270
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0296596078776973e-06,
      "loss": 0.431,
      "step": 2271
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.0256050681470446e-06,
      "loss": 0.3536,
      "step": 2272
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.021552065982327e-06,
      "loss": 0.3972,
      "step": 2273
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.0175006045398336e-06,
      "loss": 0.3971,
      "step": 2274
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.0134506869746496e-06,
      "loss": 0.4013,
      "step": 2275
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.009402316440663e-06,
      "loss": 0.4417,
      "step": 2276
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.005355496090553e-06,
      "loss": 0.4321,
      "step": 2277
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.0013102290757935e-06,
      "loss": 0.3617,
      "step": 2278
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9972665185466488e-06,
      "loss": 0.3559,
      "step": 2279
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9932243676521696e-06,
      "loss": 0.384,
      "step": 2280
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.989183779540194e-06,
      "loss": 0.3812,
      "step": 2281
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9851447573573383e-06,
      "loss": 0.383,
      "step": 2282
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.981107304249008e-06,
      "loss": 0.4122,
      "step": 2283
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.97707142335938e-06,
      "loss": 0.4139,
      "step": 2284
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9730371178314086e-06,
      "loss": 0.5211,
      "step": 2285
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9690043908068212e-06,
      "loss": 0.5033,
      "step": 2286
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.964973245426115e-06,
      "loss": 0.4605,
      "step": 2287
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9609436848285566e-06,
      "loss": 0.3502,
      "step": 2288
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.9569157121521774e-06,
      "loss": 0.3847,
      "step": 2289
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.952889330533778e-06,
      "loss": 0.3923,
      "step": 2290
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9488645431089107e-06,
      "loss": 0.5326,
      "step": 2291
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9448413530118912e-06,
      "loss": 0.4058,
      "step": 2292
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.940819763375792e-06,
      "loss": 0.4958,
      "step": 2293
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9367997773324375e-06,
      "loss": 0.4781,
      "step": 2294
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9327813980124033e-06,
      "loss": 0.6368,
      "step": 2295
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9287646285450134e-06,
      "loss": 0.405,
      "step": 2296
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9247494720583415e-06,
      "loss": 0.403,
      "step": 2297
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.9207359316792013e-06,
      "loss": 0.4507,
      "step": 2298
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.916724010533149e-06,
      "loss": 0.3996,
      "step": 2299
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.912713711744481e-06,
      "loss": 0.481,
      "step": 2300
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.908705038436229e-06,
      "loss": 0.4139,
      "step": 2301
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.904697993730159e-06,
      "loss": 0.3972,
      "step": 2302
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.900692580746769e-06,
      "loss": 0.5579,
      "step": 2303
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8966888026052876e-06,
      "loss": 0.3156,
      "step": 2304
    },
    {
      "epoch": 1.28,
      "eval_loss": 0.47729530930519104,
      "eval_runtime": 79.0524,
      "eval_samples_per_second": 25.3,
      "eval_steps_per_second": 12.65,
      "step": 2304
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.892686662423667e-06,
      "loss": 0.4482,
      "step": 2305
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.888686163318587e-06,
      "loss": 0.3714,
      "step": 2306
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8846873084054478e-06,
      "loss": 0.3904,
      "step": 2307
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.88069010079837e-06,
      "loss": 0.399,
      "step": 2308
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.87669454361019e-06,
      "loss": 0.4646,
      "step": 2309
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8727006399524593e-06,
      "loss": 0.5023,
      "step": 2310
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8687083929354447e-06,
      "loss": 0.4367,
      "step": 2311
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8647178056681197e-06,
      "loss": 0.4828,
      "step": 2312
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.860728881258167e-06,
      "loss": 0.4218,
      "step": 2313
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8567416228119706e-06,
      "loss": 0.3807,
      "step": 2314
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.852756033434621e-06,
      "loss": 0.4543,
      "step": 2315
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8487721162299083e-06,
      "loss": 0.363,
      "step": 2316
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8447898743003166e-06,
      "loss": 0.4373,
      "step": 2317
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.840809310747035e-06,
      "loss": 0.3705,
      "step": 2318
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.836830428669935e-06,
      "loss": 0.4608,
      "step": 2319
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8328532311675836e-06,
      "loss": 0.5731,
      "step": 2320
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.828877721337236e-06,
      "loss": 0.472,
      "step": 2321
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8249039022748315e-06,
      "loss": 0.4497,
      "step": 2322
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8209317770749933e-06,
      "loss": 0.4557,
      "step": 2323
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8169613488310264e-06,
      "loss": 0.4044,
      "step": 2324
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8129926206349138e-06,
      "loss": 0.4339,
      "step": 2325
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.809025595577314e-06,
      "loss": 0.4175,
      "step": 2326
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8050602767475595e-06,
      "loss": 0.4106,
      "step": 2327
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8010966672336537e-06,
      "loss": 0.4692,
      "step": 2328
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.7971347701222705e-06,
      "loss": 0.4471,
      "step": 2329
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.793174588498748e-06,
      "loss": 0.4303,
      "step": 2330
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.789216125447088e-06,
      "loss": 0.4207,
      "step": 2331
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.785259384049959e-06,
      "loss": 0.4166,
      "step": 2332
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.7813043673886863e-06,
      "loss": 0.4056,
      "step": 2333
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.777351078543245e-06,
      "loss": 0.4625,
      "step": 2334
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.773399520592274e-06,
      "loss": 0.3848,
      "step": 2335
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.769449696613061e-06,
      "loss": 0.4279,
      "step": 2336
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.7655016096815395e-06,
      "loss": 0.3622,
      "step": 2337
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.7615552628722996e-06,
      "loss": 0.4183,
      "step": 2338
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.757610659258568e-06,
      "loss": 0.4726,
      "step": 2339
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.7536678019122176e-06,
      "loss": 0.4815,
      "step": 2340
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.749726693903758e-06,
      "loss": 0.4424,
      "step": 2341
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.745787338302341e-06,
      "loss": 0.4506,
      "step": 2342
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.7418497381757504e-06,
      "loss": 0.4391,
      "step": 2343
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.737913896590404e-06,
      "loss": 0.4017,
      "step": 2344
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.7339798166113507e-06,
      "loss": 0.4616,
      "step": 2345
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.4223,
      "step": 2346
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.726116953725454e-06,
      "loss": 0.5155,
      "step": 2347
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.722188176941839e-06,
      "loss": 0.5007,
      "step": 2348
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.7182611740109677e-06,
      "loss": 0.5037,
      "step": 2349
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.7143359479910057e-06,
      "loss": 0.3925,
      "step": 2350
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.7104125019387317e-06,
      "loss": 0.4059,
      "step": 2351
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.706490838909547e-06,
      "loss": 0.508,
      "step": 2352
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.7025709619574553e-06,
      "loss": 0.6276,
      "step": 2353
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.6986528741350705e-06,
      "loss": 0.3531,
      "step": 2354
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.6947365784936175e-06,
      "loss": 0.4624,
      "step": 2355
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.690822078082922e-06,
      "loss": 0.456,
      "step": 2356
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.686909375951413e-06,
      "loss": 0.3772,
      "step": 2357
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.6829984751461175e-06,
      "loss": 0.3668,
      "step": 2358
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.679089378712665e-06,
      "loss": 0.3909,
      "step": 2359
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.6751820896952742e-06,
      "loss": 0.3836,
      "step": 2360
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.671276611136757e-06,
      "loss": 0.4072,
      "step": 2361
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.6673729460785174e-06,
      "loss": 0.4144,
      "step": 2362
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.663471097560546e-06,
      "loss": 0.5006,
      "step": 2363
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6595710686214194e-06,
      "loss": 0.3627,
      "step": 2364
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6556728622982965e-06,
      "loss": 0.5265,
      "step": 2365
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6517764816269165e-06,
      "loss": 0.3724,
      "step": 2366
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.647881929641598e-06,
      "loss": 0.4783,
      "step": 2367
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.643989209375235e-06,
      "loss": 0.3899,
      "step": 2368
    },
    {
      "epoch": 1.32,
      "eval_loss": 0.476846843957901,
      "eval_runtime": 79.0365,
      "eval_samples_per_second": 25.305,
      "eval_steps_per_second": 12.652,
      "step": 2368
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.640098323859295e-06,
      "loss": 0.4562,
      "step": 2369
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6362092761238155e-06,
      "loss": 0.3885,
      "step": 2370
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.632322069197406e-06,
      "loss": 0.4253,
      "step": 2371
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.628436706107238e-06,
      "loss": 0.4015,
      "step": 2372
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6245531898790543e-06,
      "loss": 0.341,
      "step": 2373
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6206715235371492e-06,
      "loss": 0.4278,
      "step": 2374
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.616791710104385e-06,
      "loss": 0.4218,
      "step": 2375
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6129137526021777e-06,
      "loss": 0.5121,
      "step": 2376
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.609037654050497e-06,
      "loss": 0.4396,
      "step": 2377
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.605163417467867e-06,
      "loss": 0.47,
      "step": 2378
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6012910458713593e-06,
      "loss": 0.4495,
      "step": 2379
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.597420542276598e-06,
      "loss": 0.456,
      "step": 2380
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.593551909697748e-06,
      "loss": 0.4195,
      "step": 2381
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5896851511475184e-06,
      "loss": 0.4111,
      "step": 2382
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5858202696371594e-06,
      "loss": 0.4415,
      "step": 2383
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5819572681764594e-06,
      "loss": 0.51,
      "step": 2384
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.57809614977374e-06,
      "loss": 0.3458,
      "step": 2385
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.574236917435857e-06,
      "loss": 0.5063,
      "step": 2386
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5703795741682053e-06,
      "loss": 0.3979,
      "step": 2387
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5665241229746986e-06,
      "loss": 0.3208,
      "step": 2388
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5626705668577816e-06,
      "loss": 0.471,
      "step": 2389
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5588189088184222e-06,
      "loss": 0.3755,
      "step": 2390
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5549691518561116e-06,
      "loss": 0.3939,
      "step": 2391
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5511212989688587e-06,
      "loss": 0.3884,
      "step": 2392
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5472753531531906e-06,
      "loss": 0.3479,
      "step": 2393
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.543431317404149e-06,
      "loss": 0.3155,
      "step": 2394
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5395891947152886e-06,
      "loss": 0.4588,
      "step": 2395
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.535748988078673e-06,
      "loss": 0.3609,
      "step": 2396
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.5319107004848752e-06,
      "loss": 0.3852,
      "step": 2397
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.528074334922973e-06,
      "loss": 0.425,
      "step": 2398
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.5242398943805478e-06,
      "loss": 0.4779,
      "step": 2399
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.520407381843679e-06,
      "loss": 0.3981,
      "step": 2400
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.5165768002969514e-06,
      "loss": 0.4475,
      "step": 2401
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.5127481527234397e-06,
      "loss": 0.448,
      "step": 2402
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.508921442104715e-06,
      "loss": 0.4256,
      "step": 2403
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.5050966714208414e-06,
      "loss": 0.4126,
      "step": 2404
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.5012738436503665e-06,
      "loss": 0.396,
      "step": 2405
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.497452961770331e-06,
      "loss": 0.5452,
      "step": 2406
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.493634028756255e-06,
      "loss": 0.4957,
      "step": 2407
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.489817047582149e-06,
      "loss": 0.5408,
      "step": 2408
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.4860020212204967e-06,
      "loss": 0.3675,
      "step": 2409
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.4821889526422605e-06,
      "loss": 0.4573,
      "step": 2410
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.478377844816879e-06,
      "loss": 0.3737,
      "step": 2411
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.4745687007122636e-06,
      "loss": 0.4464,
      "step": 2412
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.470761523294798e-06,
      "loss": 0.3526,
      "step": 2413
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.4669563155293307e-06,
      "loss": 0.4378,
      "step": 2414
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.4631530803791807e-06,
      "loss": 0.4835,
      "step": 2415
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4593518208061274e-06,
      "loss": 0.4259,
      "step": 2416
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.455552539770414e-06,
      "loss": 0.3427,
      "step": 2417
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4517552402307405e-06,
      "loss": 0.3497,
      "step": 2418
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4479599251442665e-06,
      "loss": 0.4096,
      "step": 2419
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4441665974666047e-06,
      "loss": 0.4215,
      "step": 2420
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4403752601518175e-06,
      "loss": 0.4491,
      "step": 2421
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.436585916152426e-06,
      "loss": 0.4117,
      "step": 2422
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.432798568419389e-06,
      "loss": 0.483,
      "step": 2423
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4290132199021178e-06,
      "loss": 0.3828,
      "step": 2424
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4252298735484597e-06,
      "loss": 0.4361,
      "step": 2425
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4214485323047088e-06,
      "loss": 0.3909,
      "step": 2426
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4176691991155966e-06,
      "loss": 0.4437,
      "step": 2427
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4138918769242875e-06,
      "loss": 0.501,
      "step": 2428
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4101165686723866e-06,
      "loss": 0.459,
      "step": 2429
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.406343277299924e-06,
      "loss": 0.4824,
      "step": 2430
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.4025720057453615e-06,
      "loss": 0.4581,
      "step": 2431
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.3988027569455895e-06,
      "loss": 0.4315,
      "step": 2432
    },
    {
      "epoch": 1.35,
      "eval_loss": 0.47689923644065857,
      "eval_runtime": 79.0707,
      "eval_samples_per_second": 25.294,
      "eval_steps_per_second": 12.647,
      "step": 2432
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.39503553383592e-06,
      "loss": 0.4849,
      "step": 2433
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.39127033935009e-06,
      "loss": 0.7741,
      "step": 2434
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.387507176420256e-06,
      "loss": 0.4214,
      "step": 2435
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3837460479769924e-06,
      "loss": 0.402,
      "step": 2436
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.379986956949289e-06,
      "loss": 0.462,
      "step": 2437
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.376229906264549e-06,
      "loss": 0.4555,
      "step": 2438
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3724748988485868e-06,
      "loss": 0.513,
      "step": 2439
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3687219376256255e-06,
      "loss": 0.3828,
      "step": 2440
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3649710255182957e-06,
      "loss": 0.4535,
      "step": 2441
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.361222165447628e-06,
      "loss": 0.4558,
      "step": 2442
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3574753603330642e-06,
      "loss": 0.3768,
      "step": 2443
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3537306130924375e-06,
      "loss": 0.388,
      "step": 2444
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3499879266419794e-06,
      "loss": 0.5035,
      "step": 2445
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3462473038963187e-06,
      "loss": 0.369,
      "step": 2446
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3425087477684767e-06,
      "loss": 0.4675,
      "step": 2447
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3387722611698653e-06,
      "loss": 0.5124,
      "step": 2448
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.335037847010283e-06,
      "loss": 0.3859,
      "step": 2449
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.331305508197919e-06,
      "loss": 0.4558,
      "step": 2450
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.3275752476393422e-06,
      "loss": 0.4682,
      "step": 2451
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.323847068239504e-06,
      "loss": 0.453,
      "step": 2452
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.320120972901734e-06,
      "loss": 0.4584,
      "step": 2453
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.3163969645277413e-06,
      "loss": 0.4736,
      "step": 2454
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.3126750460176083e-06,
      "loss": 0.4117,
      "step": 2455
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.30895522026979e-06,
      "loss": 0.5084,
      "step": 2456
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.305237490181112e-06,
      "loss": 0.5928,
      "step": 2457
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.3015218586467675e-06,
      "loss": 0.3987,
      "step": 2458
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2978083285603152e-06,
      "loss": 0.4678,
      "step": 2459
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2940969028136782e-06,
      "loss": 0.5132,
      "step": 2460
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.29038758429714e-06,
      "loss": 0.3496,
      "step": 2461
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2866803758993446e-06,
      "loss": 0.4094,
      "step": 2462
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2829752805072886e-06,
      "loss": 0.432,
      "step": 2463
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2792723010063317e-06,
      "loss": 0.4894,
      "step": 2464
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.275571440280175e-06,
      "loss": 0.4355,
      "step": 2465
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.271872701210876e-06,
      "loss": 0.3728,
      "step": 2466
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2681760866788397e-06,
      "loss": 0.4038,
      "step": 2467
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.264481599562815e-06,
      "loss": 0.3476,
      "step": 2468
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.2607892427398953e-06,
      "loss": 0.4453,
      "step": 2469
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2570990190855125e-06,
      "loss": 0.3613,
      "step": 2470
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2534109314734425e-06,
      "loss": 0.3733,
      "step": 2471
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2497249827757933e-06,
      "loss": 0.4425,
      "step": 2472
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.246041175863008e-06,
      "loss": 0.4024,
      "step": 2473
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2423595136038624e-06,
      "loss": 0.4193,
      "step": 2474
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2386799988654618e-06,
      "loss": 0.4502,
      "step": 2475
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.235002634513241e-06,
      "loss": 0.5101,
      "step": 2476
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.231327423410954e-06,
      "loss": 0.3996,
      "step": 2477
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.227654368420686e-06,
      "loss": 0.3873,
      "step": 2478
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.223983472402839e-06,
      "loss": 0.4096,
      "step": 2479
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.220314738216134e-06,
      "loss": 0.4165,
      "step": 2480
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2166481687176083e-06,
      "loss": 0.4091,
      "step": 2481
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2129837667626147e-06,
      "loss": 0.4249,
      "step": 2482
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2093215352048143e-06,
      "loss": 0.3608,
      "step": 2483
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.205661476896187e-06,
      "loss": 0.4661,
      "step": 2484
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.2020035946870085e-06,
      "loss": 0.359,
      "step": 2485
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.1983478914258665e-06,
      "loss": 0.4717,
      "step": 2486
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.1946943699596516e-06,
      "loss": 0.3714,
      "step": 2487
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.191043033133555e-06,
      "loss": 0.4198,
      "step": 2488
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.187393883791065e-06,
      "loss": 0.4664,
      "step": 2489
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.183746924773967e-06,
      "loss": 0.442,
      "step": 2490
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1801021589223436e-06,
      "loss": 0.3956,
      "step": 2491
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.176459589074566e-06,
      "loss": 0.4861,
      "step": 2492
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1728192180672957e-06,
      "loss": 0.4872,
      "step": 2493
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.169181048735483e-06,
      "loss": 0.4795,
      "step": 2494
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1655450839123625e-06,
      "loss": 0.4176,
      "step": 2495
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1619113264294557e-06,
      "loss": 0.4753,
      "step": 2496
    },
    {
      "epoch": 1.39,
      "eval_loss": 0.4765991270542145,
      "eval_runtime": 79.0531,
      "eval_samples_per_second": 25.299,
      "eval_steps_per_second": 12.65,
      "step": 2496
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.158279779116555e-06,
      "loss": 0.418,
      "step": 2497
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.154650444801745e-06,
      "loss": 0.4203,
      "step": 2498
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1510233263113794e-06,
      "loss": 0.4953,
      "step": 2499
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1473984264700863e-06,
      "loss": 0.406,
      "step": 2500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1437757481007684e-06,
      "loss": 0.4151,
      "step": 2501
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1401552940245962e-06,
      "loss": 0.4235,
      "step": 2502
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.1365370670610098e-06,
      "loss": 0.4312,
      "step": 2503
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.132921070027714e-06,
      "loss": 0.4622,
      "step": 2504
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.129307305740677e-06,
      "loss": 0.4273,
      "step": 2505
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.1256957770141294e-06,
      "loss": 0.4483,
      "step": 2506
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.122086486660559e-06,
      "loss": 0.3463,
      "step": 2507
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.1184794374907116e-06,
      "loss": 0.3926,
      "step": 2508
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.114874632313588e-06,
      "loss": 0.3904,
      "step": 2509
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.11127207393644e-06,
      "loss": 0.4709,
      "step": 2510
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.1076717651647705e-06,
      "loss": 0.3942,
      "step": 2511
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.1040737088023323e-06,
      "loss": 0.3691,
      "step": 2512
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.1004779076511223e-06,
      "loss": 0.3874,
      "step": 2513
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.0968843645113807e-06,
      "loss": 0.4921,
      "step": 2514
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.0932930821815895e-06,
      "loss": 0.4009,
      "step": 2515
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.089704063458473e-06,
      "loss": 0.4769,
      "step": 2516
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.086117311136987e-06,
      "loss": 0.4293,
      "step": 2517
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.082532828010324e-06,
      "loss": 0.508,
      "step": 2518
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.0789506168699163e-06,
      "loss": 0.3258,
      "step": 2519
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.0753706805054186e-06,
      "loss": 0.6457,
      "step": 2520
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.0717930217047172e-06,
      "loss": 0.4265,
      "step": 2521
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.068217643253925e-06,
      "loss": 0.4333,
      "step": 2522
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.064644547937378e-06,
      "loss": 0.4008,
      "step": 2523
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.6114,
      "step": 2524
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0575052178354743e-06,
      "loss": 0.3296,
      "step": 2525
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0539389886098924e-06,
      "loss": 0.4072,
      "step": 2526
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0503750536381016e-06,
      "loss": 0.3746,
      "step": 2527
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0468134156955255e-06,
      "loss": 0.4114,
      "step": 2528
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0432540775558007e-06,
      "loss": 0.3637,
      "step": 2529
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.039697041990773e-06,
      "loss": 0.3933,
      "step": 2530
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0361423117704935e-06,
      "loss": 0.417,
      "step": 2531
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0325898896632178e-06,
      "loss": 0.4618,
      "step": 2532
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.029039778435409e-06,
      "loss": 0.4529,
      "step": 2533
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0254919808517243e-06,
      "loss": 0.5244,
      "step": 2534
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0219464996750216e-06,
      "loss": 0.4429,
      "step": 2535
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0184033376663576e-06,
      "loss": 0.4097,
      "step": 2536
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0148624975849755e-06,
      "loss": 0.3762,
      "step": 2537
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0113239821883184e-06,
      "loss": 0.4311,
      "step": 2538
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.007787794232013e-06,
      "loss": 0.355,
      "step": 2539
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0042539364698795e-06,
      "loss": 0.4331,
      "step": 2540
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0007224116539186e-06,
      "loss": 0.5037,
      "step": 2541
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.997193222534316e-06,
      "loss": 0.3886,
      "step": 2542
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.993666371859438e-06,
      "loss": 0.5631,
      "step": 2543
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.99014186237583e-06,
      "loss": 0.4159,
      "step": 2544
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.986619696828215e-06,
      "loss": 0.4283,
      "step": 2545
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.983099877959489e-06,
      "loss": 0.5237,
      "step": 2546
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9795824085107217e-06,
      "loss": 0.3623,
      "step": 2547
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9760672912211536e-06,
      "loss": 0.4862,
      "step": 2548
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9725545288281917e-06,
      "loss": 0.5032,
      "step": 2549
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9690441240674108e-06,
      "loss": 0.5753,
      "step": 2550
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9655360796725494e-06,
      "loss": 0.3757,
      "step": 2551
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.962030398375506e-06,
      "loss": 0.4116,
      "step": 2552
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9585270829063402e-06,
      "loss": 0.3628,
      "step": 2553
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.955026135993272e-06,
      "loss": 0.4151,
      "step": 2554
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.951527560362673e-06,
      "loss": 0.5582,
      "step": 2555
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9480313587390712e-06,
      "loss": 0.338,
      "step": 2556
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9445375338451405e-06,
      "loss": 0.3716,
      "step": 2557
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9410460884017084e-06,
      "loss": 0.4195,
      "step": 2558
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9375570251277492e-06,
      "loss": 0.3733,
      "step": 2559
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9340703467403793e-06,
      "loss": 0.3612,
      "step": 2560
    },
    {
      "epoch": 1.43,
      "eval_loss": 0.4759904146194458,
      "eval_runtime": 79.0373,
      "eval_samples_per_second": 25.305,
      "eval_steps_per_second": 12.652,
      "step": 2560
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9305860559548633e-06,
      "loss": 0.5373,
      "step": 2561
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.927104155484602e-06,
      "loss": 0.4354,
      "step": 2562
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9236246480411346e-06,
      "loss": 0.3893,
      "step": 2563
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9201475363341394e-06,
      "loss": 0.4291,
      "step": 2564
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9166728230714256e-06,
      "loss": 0.4705,
      "step": 2565
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.913200510958938e-06,
      "loss": 0.4103,
      "step": 2566
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9097306027007495e-06,
      "loss": 0.486,
      "step": 2567
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9062631009990612e-06,
      "loss": 0.3693,
      "step": 2568
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.9027980085542007e-06,
      "loss": 0.4377,
      "step": 2569
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.89933532806462e-06,
      "loss": 0.4442,
      "step": 2570
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8958750622268902e-06,
      "loss": 0.3024,
      "step": 2571
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8924172137357038e-06,
      "loss": 0.3489,
      "step": 2572
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8889617852838715e-06,
      "loss": 0.5231,
      "step": 2573
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8855087795623168e-06,
      "loss": 0.4414,
      "step": 2574
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.882058199260081e-06,
      "loss": 0.3936,
      "step": 2575
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8786100470643142e-06,
      "loss": 0.3693,
      "step": 2576
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8751643256602714e-06,
      "loss": 0.5956,
      "step": 2577
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8717210377313206e-06,
      "loss": 0.3859,
      "step": 2578
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8682801859589323e-06,
      "loss": 0.4571,
      "step": 2579
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8648417730226797e-06,
      "loss": 0.5014,
      "step": 2580
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8614058016002356e-06,
      "loss": 0.6371,
      "step": 2581
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8579722743673773e-06,
      "loss": 0.4462,
      "step": 2582
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8545411939979713e-06,
      "loss": 0.4454,
      "step": 2583
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8511125631639825e-06,
      "loss": 0.4269,
      "step": 2584
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.847686384535467e-06,
      "loss": 0.509,
      "step": 2585
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.844262660780572e-06,
      "loss": 0.4799,
      "step": 2586
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.840841394565534e-06,
      "loss": 0.4246,
      "step": 2587
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8374225885546693e-06,
      "loss": 0.4669,
      "step": 2588
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.834006245410388e-06,
      "loss": 0.3843,
      "step": 2589
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8305923677931759e-06,
      "loss": 0.3908,
      "step": 2590
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8271809583616007e-06,
      "loss": 0.3061,
      "step": 2591
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8237720197723075e-06,
      "loss": 0.4274,
      "step": 2592
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8203655546800175e-06,
      "loss": 0.5125,
      "step": 2593
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8169615657375255e-06,
      "loss": 0.4961,
      "step": 2594
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8135600555956979e-06,
      "loss": 0.3968,
      "step": 2595
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.8101610269034709e-06,
      "loss": 0.3974,
      "step": 2596
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.806764482307848e-06,
      "loss": 0.3824,
      "step": 2597
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.8033704244538985e-06,
      "loss": 0.3547,
      "step": 2598
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7999788559847554e-06,
      "loss": 0.4052,
      "step": 2599
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7965897795416125e-06,
      "loss": 0.3393,
      "step": 2600
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7932031977637238e-06,
      "loss": 0.4047,
      "step": 2601
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.789819113288397e-06,
      "loss": 0.3696,
      "step": 2602
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7864375287510022e-06,
      "loss": 0.475,
      "step": 2603
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.783058446784957e-06,
      "loss": 0.5213,
      "step": 2604
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7796818700217306e-06,
      "loss": 0.4471,
      "step": 2605
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.776307801090843e-06,
      "loss": 0.3942,
      "step": 2606
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.772936242619862e-06,
      "loss": 0.4769,
      "step": 2607
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7695671972343953e-06,
      "loss": 0.4158,
      "step": 2608
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7662006675580968e-06,
      "loss": 0.5425,
      "step": 2609
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7628366562126653e-06,
      "loss": 0.4325,
      "step": 2610
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7594751658178328e-06,
      "loss": 0.4576,
      "step": 2611
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.75611619899137e-06,
      "loss": 0.4143,
      "step": 2612
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.7527597583490825e-06,
      "loss": 0.4791,
      "step": 2613
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7494058465048075e-06,
      "loss": 0.4527,
      "step": 2614
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7460544660704154e-06,
      "loss": 0.3059,
      "step": 2615
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7427056196558024e-06,
      "loss": 0.3655,
      "step": 2616
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7393593098688933e-06,
      "loss": 0.4191,
      "step": 2617
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7360155393156364e-06,
      "loss": 0.5172,
      "step": 2618
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7326743106000033e-06,
      "loss": 0.4365,
      "step": 2619
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7293356263239858e-06,
      "loss": 0.4525,
      "step": 2620
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7259994890875936e-06,
      "loss": 0.4577,
      "step": 2621
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7226659014888548e-06,
      "loss": 0.4734,
      "step": 2622
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.719334866123808e-06,
      "loss": 0.5502,
      "step": 2623
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7160063855865105e-06,
      "loss": 0.5175,
      "step": 2624
    },
    {
      "epoch": 1.46,
      "eval_loss": 0.47565123438835144,
      "eval_runtime": 78.9965,
      "eval_samples_per_second": 25.318,
      "eval_steps_per_second": 12.659,
      "step": 2624
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7126804624690246e-06,
      "loss": 0.4067,
      "step": 2625
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7093570993614233e-06,
      "loss": 0.5484,
      "step": 2626
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.706036298851787e-06,
      "loss": 0.4386,
      "step": 2627
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7027180635261953e-06,
      "loss": 0.3809,
      "step": 2628
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.6994023959687339e-06,
      "loss": 0.3623,
      "step": 2629
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.6960892987614929e-06,
      "loss": 0.5497,
      "step": 2630
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.692778774484554e-06,
      "loss": 0.4193,
      "step": 2631
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.689470825715998e-06,
      "loss": 0.4006,
      "step": 2632
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6861654550319007e-06,
      "loss": 0.4205,
      "step": 2633
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6828626650063284e-06,
      "loss": 0.414,
      "step": 2634
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.679562458211339e-06,
      "loss": 0.3661,
      "step": 2635
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6762648372169787e-06,
      "loss": 0.3705,
      "step": 2636
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.672969804591279e-06,
      "loss": 0.3947,
      "step": 2637
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6696773629002572e-06,
      "loss": 0.4348,
      "step": 2638
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6663875147079112e-06,
      "loss": 0.4339,
      "step": 2639
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6631002625762205e-06,
      "loss": 0.5902,
      "step": 2640
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.659815609065143e-06,
      "loss": 0.4159,
      "step": 2641
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6565335567326112e-06,
      "loss": 0.4191,
      "step": 2642
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6532541081345326e-06,
      "loss": 0.494,
      "step": 2643
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6499772658247908e-06,
      "loss": 0.492,
      "step": 2644
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.646703032355234e-06,
      "loss": 0.4351,
      "step": 2645
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6434314102756821e-06,
      "loss": 0.4629,
      "step": 2646
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6401624021339218e-06,
      "loss": 0.42,
      "step": 2647
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6368960104756997e-06,
      "loss": 0.4729,
      "step": 2648
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.63363223784473e-06,
      "loss": 0.3772,
      "step": 2649
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6303710867826828e-06,
      "loss": 0.4271,
      "step": 2650
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6271125598291938e-06,
      "loss": 0.4481,
      "step": 2651
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6238566595218475e-06,
      "loss": 0.4612,
      "step": 2652
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6206033883961874e-06,
      "loss": 0.3422,
      "step": 2653
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6173527489857071e-06,
      "loss": 0.4721,
      "step": 2654
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6141047438218516e-06,
      "loss": 0.5051,
      "step": 2655
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6108593754340157e-06,
      "loss": 0.4345,
      "step": 2656
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.6076166463495384e-06,
      "loss": 0.4031,
      "step": 2657
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.604376559093706e-06,
      "loss": 0.369,
      "step": 2658
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.601139116189745e-06,
      "loss": 0.393,
      "step": 2659
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.597904320158824e-06,
      "loss": 0.4328,
      "step": 2660
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.59467217352005e-06,
      "loss": 0.3647,
      "step": 2661
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.591442678790467e-06,
      "loss": 0.365,
      "step": 2662
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.5882158384850538e-06,
      "loss": 0.4331,
      "step": 2663
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.5849916551167198e-06,
      "loss": 0.3882,
      "step": 2664
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.5817701311963107e-06,
      "loss": 0.4735,
      "step": 2665
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.5785512692325971e-06,
      "loss": 0.4916,
      "step": 2666
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.575335071732278e-06,
      "loss": 0.3409,
      "step": 2667
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5721215411999735e-06,
      "loss": 0.3519,
      "step": 2668
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5689106801382325e-06,
      "loss": 0.4869,
      "step": 2669
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.565702491047522e-06,
      "loss": 0.405,
      "step": 2670
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5624969764262277e-06,
      "loss": 0.3853,
      "step": 2671
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5592941387706562e-06,
      "loss": 0.4369,
      "step": 2672
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5560939805750252e-06,
      "loss": 0.41,
      "step": 2673
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5528965043314686e-06,
      "loss": 0.4911,
      "step": 2674
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5497017125300296e-06,
      "loss": 0.3724,
      "step": 2675
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5465096076586617e-06,
      "loss": 0.4152,
      "step": 2676
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5433201922032266e-06,
      "loss": 0.4406,
      "step": 2677
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5401334686474906e-06,
      "loss": 0.4661,
      "step": 2678
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.536949439473125e-06,
      "loss": 0.4045,
      "step": 2679
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.533768107159701e-06,
      "loss": 0.3836,
      "step": 2680
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5305894741846917e-06,
      "loss": 0.4505,
      "step": 2681
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5274135430234654e-06,
      "loss": 0.438,
      "step": 2682
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5242403161492896e-06,
      "loss": 0.4252,
      "step": 2683
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5210697960333238e-06,
      "loss": 0.3738,
      "step": 2684
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5179019851446175e-06,
      "loss": 0.4161,
      "step": 2685
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.5147368859501166e-06,
      "loss": 0.3852,
      "step": 2686
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.5115745009146521e-06,
      "loss": 0.4777,
      "step": 2687
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.5084148325009368e-06,
      "loss": 0.3961,
      "step": 2688
    },
    {
      "epoch": 1.5,
      "eval_loss": 0.4754595160484314,
      "eval_runtime": 79.0565,
      "eval_samples_per_second": 25.298,
      "eval_steps_per_second": 12.649,
      "step": 2688
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.5052578831695735e-06,
      "loss": 0.5124,
      "step": 2689
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.5021036553790468e-06,
      "loss": 0.4557,
      "step": 2690
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4989521515857198e-06,
      "loss": 0.4848,
      "step": 2691
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4958033742438348e-06,
      "loss": 0.446,
      "step": 2692
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4926573258055137e-06,
      "loss": 0.3954,
      "step": 2693
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.489514008720751e-06,
      "loss": 0.4421,
      "step": 2694
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4863734254374128e-06,
      "loss": 0.4869,
      "step": 2695
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4832355784012386e-06,
      "loss": 0.4623,
      "step": 2696
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4801004700558358e-06,
      "loss": 0.3748,
      "step": 2697
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4769681028426808e-06,
      "loss": 0.3576,
      "step": 2698
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.473838479201109e-06,
      "loss": 0.3677,
      "step": 2699
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4707116015683299e-06,
      "loss": 0.4205,
      "step": 2700
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.467587472379406e-06,
      "loss": 0.5296,
      "step": 2701
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.3719,
      "step": 2702
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4613474690626828e-06,
      "loss": 0.4484,
      "step": 2703
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4582315997943048e-06,
      "loss": 0.3277,
      "step": 2704
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.455118488688621e-06,
      "loss": 0.3874,
      "step": 2705
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.452008138169974e-06,
      "loss": 0.4009,
      "step": 2706
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.448900550660564e-06,
      "loss": 0.3671,
      "step": 2707
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.445795728580429e-06,
      "loss": 0.4595,
      "step": 2708
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.44269367434746e-06,
      "loss": 0.489,
      "step": 2709
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4395943903773912e-06,
      "loss": 0.4443,
      "step": 2710
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4364978790837986e-06,
      "loss": 0.3647,
      "step": 2711
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4334041428781003e-06,
      "loss": 0.379,
      "step": 2712
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4303131841695506e-06,
      "loss": 0.4421,
      "step": 2713
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.427225005365246e-06,
      "loss": 0.4752,
      "step": 2714
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4241396088701137e-06,
      "loss": 0.3919,
      "step": 2715
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.421056997086915e-06,
      "loss": 0.4545,
      "step": 2716
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4179771724162428e-06,
      "loss": 0.3758,
      "step": 2717
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4149001372565218e-06,
      "loss": 0.3767,
      "step": 2718
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4118258940039998e-06,
      "loss": 0.3833,
      "step": 2719
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4087544450527513e-06,
      "loss": 0.507,
      "step": 2720
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4056857927946805e-06,
      "loss": 0.3929,
      "step": 2721
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.4026199396195078e-06,
      "loss": 0.4098,
      "step": 2722
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3995568879147754e-06,
      "loss": 0.3677,
      "step": 2723
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3964966400658437e-06,
      "loss": 0.4965,
      "step": 2724
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3934391984558905e-06,
      "loss": 0.4745,
      "step": 2725
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.390384565465907e-06,
      "loss": 0.3315,
      "step": 2726
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3873327434746975e-06,
      "loss": 0.4222,
      "step": 2727
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3842837348588784e-06,
      "loss": 0.4512,
      "step": 2728
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.381237541992873e-06,
      "loss": 0.4284,
      "step": 2729
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3781941672489147e-06,
      "loss": 0.4781,
      "step": 2730
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.375153612997039e-06,
      "loss": 0.4951,
      "step": 2731
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3721158816050872e-06,
      "loss": 0.4159,
      "step": 2732
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3690809754387024e-06,
      "loss": 0.3596,
      "step": 2733
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3660488968613246e-06,
      "loss": 0.4519,
      "step": 2734
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3630196482341968e-06,
      "loss": 0.4024,
      "step": 2735
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.359993231916354e-06,
      "loss": 0.4117,
      "step": 2736
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3569696502646274e-06,
      "loss": 0.4206,
      "step": 2737
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3539489056336413e-06,
      "loss": 0.354,
      "step": 2738
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.350931000375807e-06,
      "loss": 0.4002,
      "step": 2739
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3479159368413286e-06,
      "loss": 0.4159,
      "step": 2740
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3449037173781937e-06,
      "loss": 0.4466,
      "step": 2741
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3418943443321807e-06,
      "loss": 0.4315,
      "step": 2742
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3388878200468463e-06,
      "loss": 0.4186,
      "step": 2743
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3358841468635303e-06,
      "loss": 0.3751,
      "step": 2744
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.332883327121352e-06,
      "loss": 0.5403,
      "step": 2745
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.329885363157209e-06,
      "loss": 0.4858,
      "step": 2746
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.326890257305774e-06,
      "loss": 0.4054,
      "step": 2747
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3238980118994954e-06,
      "loss": 0.4199,
      "step": 2748
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3209086292685924e-06,
      "loss": 0.3462,
      "step": 2749
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3179221117410568e-06,
      "loss": 0.4446,
      "step": 2750
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3149384616426468e-06,
      "loss": 0.4224,
      "step": 2751
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3119576812968893e-06,
      "loss": 0.3776,
      "step": 2752
    },
    {
      "epoch": 1.53,
      "eval_loss": 0.47524943947792053,
      "eval_runtime": 79.0196,
      "eval_samples_per_second": 25.31,
      "eval_steps_per_second": 12.655,
      "step": 2752
    }
  ],
  "logging_steps": 1,
  "max_steps": 3592,
  "num_train_epochs": 2,
  "save_steps": 64,
  "total_flos": 7.785437361478828e+18,
  "trial_name": null,
  "trial_params": null
}