{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.942724458204334,
  "eval_steps": 50,
  "global_step": 5168,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 6e-06,
      "loss": 1.3528,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.2e-05,
      "loss": 1.3449,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.8e-05,
      "loss": 1.2907,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4e-05,
      "loss": 1.3997,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 3e-05,
      "loss": 1.3751,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.6e-05,
      "loss": 1.3271,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2e-05,
      "loss": 1.2757,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8e-05,
      "loss": 1.2364,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 1.1987,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 6e-05,
      "loss": 1.314,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.999999443547221e-05,
      "loss": 1.1383,
      "step": 11
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.999997774189092e-05,
      "loss": 1.1465,
      "step": 12
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.999994991926231e-05,
      "loss": 1.1017,
      "step": 13
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.99999109675967e-05,
      "loss": 1.1974,
      "step": 14
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.9999860886908546e-05,
      "loss": 1.1773,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.9999799677216426e-05,
      "loss": 1.119,
      "step": 16
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.999972733854306e-05,
      "loss": 1.1467,
      "step": 17
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.999964387091525e-05,
      "loss": 1.0846,
      "step": 18
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.999954927436399e-05,
      "loss": 0.9089,
      "step": 19
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.9999443548924355e-05,
      "loss": 1.0313,
      "step": 20
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999932669463558e-05,
      "loss": 1.053,
      "step": 21
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.9999198711541e-05,
      "loss": 1.0601,
      "step": 22
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.9999059599688106e-05,
      "loss": 1.1186,
      "step": 23
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.9998909359128495e-05,
      "loss": 0.9551,
      "step": 24
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.99987479899179e-05,
      "loss": 1.026,
      "step": 25
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999857549211619e-05,
      "loss": 0.956,
      "step": 26
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999839186578736e-05,
      "loss": 0.9429,
      "step": 27
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.9998197110999523e-05,
      "loss": 1.1377,
      "step": 28
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999799122782492e-05,
      "loss": 1.0053,
      "step": 29
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999777421633994e-05,
      "loss": 1.053,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999754607662509e-05,
      "loss": 0.9801,
      "step": 31
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.999730680876498e-05,
      "loss": 1.0854,
      "step": 32
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.99970564128484e-05,
      "loss": 1.0073,
      "step": 33
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999679488896822e-05,
      "loss": 1.0055,
      "step": 34
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999652223722146e-05,
      "loss": 1.077,
      "step": 35
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999623845770926e-05,
      "loss": 1.0049,
      "step": 36
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.9995943550536915e-05,
      "loss": 1.037,
      "step": 37
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.9995637515813805e-05,
      "loss": 1.0184,
      "step": 38
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.9995320353653467e-05,
      "loss": 1.0514,
      "step": 39
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.9994992064173556e-05,
      "loss": 0.9558,
      "step": 40
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999465264749586e-05,
      "loss": 1.0158,
      "step": 41
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999430210374629e-05,
      "loss": 1.0178,
      "step": 42
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999394043305488e-05,
      "loss": 0.9685,
      "step": 43
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.9993567635555814e-05,
      "loss": 0.9661,
      "step": 44
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.999318371138738e-05,
      "loss": 1.0643,
      "step": 45
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.9992788660691994e-05,
      "loss": 0.9666,
      "step": 46
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.999238248361622e-05,
      "loss": 0.9656,
      "step": 47
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.999196518031074e-05,
      "loss": 0.9473,
      "step": 48
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.999153675093034e-05,
      "loss": 1.0628,
      "step": 49
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.999109719563396e-05,
      "loss": 1.0203,
      "step": 50
    },
    {
      "epoch": 0.04,
      "eval_loss": 1.0092148780822754,
      "eval_runtime": 14.3039,
      "eval_samples_per_second": 5.733,
      "eval_steps_per_second": 5.733,
      "step": 50
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.999064651458468e-05,
      "loss": 0.9855,
      "step": 51
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.999018470794967e-05,
      "loss": 1.0158,
      "step": 52
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.9989711775900255e-05,
      "loss": 1.1334,
      "step": 53
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.998922771861187e-05,
      "loss": 0.9961,
      "step": 54
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.998873253626409e-05,
      "loss": 0.9928,
      "step": 55
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.998822622904061e-05,
      "loss": 1.0314,
      "step": 56
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.998770879712925e-05,
      "loss": 0.9761,
      "step": 57
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.998718024072198e-05,
      "loss": 1.0083,
      "step": 58
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.9986640560014846e-05,
      "loss": 0.952,
      "step": 59
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.9986089755208086e-05,
      "loss": 0.9763,
      "step": 60
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998552782650601e-05,
      "loss": 1.0366,
      "step": 61
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.9984954774117075e-05,
      "loss": 1.0147,
      "step": 62
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998437059825389e-05,
      "loss": 1.0294,
      "step": 63
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998377529913313e-05,
      "loss": 1.0446,
      "step": 64
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998316887697566e-05,
      "loss": 0.9677,
      "step": 65
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.9982551332006425e-05,
      "loss": 1.1324,
      "step": 66
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998192266445453e-05,
      "loss": 1.0193,
      "step": 67
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998128287455318e-05,
      "loss": 0.9358,
      "step": 68
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.998063196253973e-05,
      "loss": 0.899,
      "step": 69
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.997996992865563e-05,
      "loss": 0.924,
      "step": 70
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.9979296773146486e-05,
      "loss": 0.959,
      "step": 71
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997861249626202e-05,
      "loss": 0.9175,
      "step": 72
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997791709825606e-05,
      "loss": 0.9295,
      "step": 73
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.9977210579386595e-05,
      "loss": 0.9717,
      "step": 74
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997649293991571e-05,
      "loss": 0.9112,
      "step": 75
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.9975764180109634e-05,
      "loss": 0.9751,
      "step": 76
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997502430023871e-05,
      "loss": 1.0222,
      "step": 77
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997427330057741e-05,
      "loss": 0.9514,
      "step": 78
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997351118140434e-05,
      "loss": 0.9049,
      "step": 79
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.9972737943002204e-05,
      "loss": 1.0462,
      "step": 80
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997195358565787e-05,
      "loss": 1.0072,
      "step": 81
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.99711581096623e-05,
      "loss": 0.8974,
      "step": 82
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.997035151531057e-05,
      "loss": 0.9284,
      "step": 83
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996953380290194e-05,
      "loss": 1.0063,
      "step": 84
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.9968704972739736e-05,
      "loss": 0.9174,
      "step": 85
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.9967865025131417e-05,
      "loss": 0.9491,
      "step": 86
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.99670139603886e-05,
      "loss": 0.9228,
      "step": 87
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996615177882699e-05,
      "loss": 0.9828,
      "step": 88
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.9965278480766434e-05,
      "loss": 0.9737,
      "step": 89
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996439406653088e-05,
      "loss": 0.9446,
      "step": 90
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996349853644845e-05,
      "loss": 0.8599,
      "step": 91
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996259189085134e-05,
      "loss": 0.92,
      "step": 92
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996167413007588e-05,
      "loss": 0.9635,
      "step": 93
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.996074525446254e-05,
      "loss": 0.9384,
      "step": 94
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.99598052643559e-05,
      "loss": 0.9687,
      "step": 95
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.995885416010467e-05,
      "loss": 0.9615,
      "step": 96
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.995789194206168e-05,
      "loss": 0.9118,
      "step": 97
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.995691861058389e-05,
      "loss": 0.9355,
      "step": 98
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.995593416603235e-05,
      "loss": 0.9507,
      "step": 99
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.995493860877228e-05,
      "loss": 0.9366,
      "step": 100
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.9740197658538818,
      "eval_runtime": 14.3321,
      "eval_samples_per_second": 5.721,
      "eval_steps_per_second": 5.721,
      "step": 100
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.9953931939172995e-05,
      "loss": 1.0358,
      "step": 101
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.9952914157607934e-05,
      "loss": 0.9567,
      "step": 102
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.995188526445466e-05,
      "loss": 0.9613,
      "step": 103
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.9950845260094875e-05,
      "loss": 0.9635,
      "step": 104
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.994979414491438e-05,
      "loss": 0.9957,
      "step": 105
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.9948731919303095e-05,
      "loss": 0.9309,
      "step": 106
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.9947658583655087e-05,
      "loss": 0.9494,
      "step": 107
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.994657413836852e-05,
      "loss": 0.9189,
      "step": 108
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.9945478583845686e-05,
      "loss": 0.9197,
      "step": 109
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.994437192049303e-05,
      "loss": 0.9303,
      "step": 110
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.994325414872104e-05,
      "loss": 1.039,
      "step": 111
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.994212526894441e-05,
      "loss": 0.9252,
      "step": 112
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.9940985281581914e-05,
      "loss": 0.9062,
      "step": 113
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.993983418705644e-05,
      "loss": 0.9566,
      "step": 114
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.993867198579503e-05,
      "loss": 0.9317,
      "step": 115
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.99374986782288e-05,
      "loss": 0.9046,
      "step": 116
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.9936314264793025e-05,
      "loss": 0.9501,
      "step": 117
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.993511874592708e-05,
      "loss": 0.9414,
      "step": 118
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.993391212207446e-05,
      "loss": 0.9857,
      "step": 119
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.993269439368279e-05,
      "loss": 0.907,
      "step": 120
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.9931465561203805e-05,
      "loss": 0.9606,
      "step": 121
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.993022562509337e-05,
      "loss": 0.8788,
      "step": 122
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.9928974585811446e-05,
      "loss": 0.9131,
      "step": 123
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.9927712443822156e-05,
      "loss": 1.1033,
      "step": 124
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.9926439199593695e-05,
      "loss": 0.9284,
      "step": 125
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.992515485359841e-05,
      "loss": 0.9336,
      "step": 126
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.992385940631273e-05,
      "loss": 0.9167,
      "step": 127
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.9922552858217255e-05,
      "loss": 0.9539,
      "step": 128
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.992123520979665e-05,
      "loss": 0.9554,
      "step": 129
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.991990646153973e-05,
      "loss": 0.9995,
      "step": 130
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.991856661393942e-05,
      "loss": 0.9189,
      "step": 131
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.9917215667492765e-05,
      "loss": 1.0112,
      "step": 132
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.9915853622700904e-05,
      "loss": 0.9819,
      "step": 133
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.991448048006914e-05,
      "loss": 0.9668,
      "step": 134
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.991309624010684e-05,
      "loss": 0.9559,
      "step": 135
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9911700903327534e-05,
      "loss": 0.8581,
      "step": 136
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.991029447024884e-05,
      "loss": 1.008,
      "step": 137
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.99088769413925e-05,
      "loss": 0.9794,
      "step": 138
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9907448317284366e-05,
      "loss": 0.9236,
      "step": 139
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9906008598454426e-05,
      "loss": 0.9119,
      "step": 140
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9904557785436764e-05,
      "loss": 0.9184,
      "step": 141
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.990309587876958e-05,
      "loss": 0.9256,
      "step": 142
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9901622878995217e-05,
      "loss": 0.9921,
      "step": 143
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.990013878666008e-05,
      "loss": 0.9549,
      "step": 144
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.989864360231474e-05,
      "loss": 0.9589,
      "step": 145
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.989713732651386e-05,
      "loss": 0.9441,
      "step": 146
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9895619959816226e-05,
      "loss": 0.9112,
      "step": 147
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9894091502784726e-05,
      "loss": 0.9471,
      "step": 148
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.989255195598636e-05,
      "loss": 0.8758,
      "step": 149
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.989100131999227e-05,
      "loss": 0.9341,
      "step": 150
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.9579644799232483,
      "eval_runtime": 14.3455,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 150
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9889439595377686e-05,
      "loss": 1.0078,
      "step": 151
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.988786678272196e-05,
      "loss": 0.9289,
      "step": 152
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9886282882608555e-05,
      "loss": 0.9127,
      "step": 153
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9884687895625036e-05,
      "loss": 0.9824,
      "step": 154
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.988308182236312e-05,
      "loss": 0.9168,
      "step": 155
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.988146466341859e-05,
      "loss": 0.915,
      "step": 156
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9879836419391356e-05,
      "loss": 0.8693,
      "step": 157
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.987819709088547e-05,
      "loss": 1.0086,
      "step": 158
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9876546678509035e-05,
      "loss": 0.8625,
      "step": 159
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9874885182874336e-05,
      "loss": 0.9213,
      "step": 160
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.9873212604597726e-05,
      "loss": 0.8849,
      "step": 161
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.9871528944299666e-05,
      "loss": 0.8412,
      "step": 162
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.986983420260475e-05,
      "loss": 0.9405,
      "step": 163
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.986812838014168e-05,
      "loss": 0.9492,
      "step": 164
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.9866411477543254e-05,
      "loss": 0.9946,
      "step": 165
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.986468349544639e-05,
      "loss": 0.8268,
      "step": 166
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.986294443449212e-05,
      "loss": 0.874,
      "step": 167
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.986119429532557e-05,
      "loss": 0.8546,
      "step": 168
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.985943307859599e-05,
      "loss": 0.8573,
      "step": 169
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.9857660784956754e-05,
      "loss": 0.9122,
      "step": 170
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.985587741506531e-05,
      "loss": 0.9282,
      "step": 171
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.985408296958323e-05,
      "loss": 0.8424,
      "step": 172
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.98522774491762e-05,
      "loss": 0.8827,
      "step": 173
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.985046085451401e-05,
      "loss": 0.886,
      "step": 174
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.984863318627057e-05,
      "loss": 0.8363,
      "step": 175
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.984679444512388e-05,
      "loss": 0.9607,
      "step": 176
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.984494463175605e-05,
      "loss": 0.97,
      "step": 177
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.984308374685331e-05,
      "loss": 0.8727,
      "step": 178
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.984121179110598e-05,
      "loss": 0.9653,
      "step": 179
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.983932876520851e-05,
      "loss": 0.9436,
      "step": 180
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.9837434669859435e-05,
      "loss": 0.9764,
      "step": 181
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.983552950576141e-05,
      "loss": 0.867,
      "step": 182
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.983361327362118e-05,
      "loss": 0.8783,
      "step": 183
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.983168597414962e-05,
      "loss": 0.9174,
      "step": 184
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.9829747608061686e-05,
      "loss": 0.9021,
      "step": 185
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.9827798176076464e-05,
      "loss": 0.9501,
      "step": 186
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.9825837678917114e-05,
      "loss": 0.9088,
      "step": 187
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.982386611731094e-05,
      "loss": 0.9459,
      "step": 188
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.9821883491989313e-05,
      "loss": 0.9144,
      "step": 189
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.981988980368773e-05,
      "loss": 0.8691,
      "step": 190
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.981788505314579e-05,
      "loss": 0.8543,
      "step": 191
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.981586924110719e-05,
      "loss": 0.966,
      "step": 192
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.981384236831973e-05,
      "loss": 0.9013,
      "step": 193
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.981180443553531e-05,
      "loss": 0.9596,
      "step": 194
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.980975544350995e-05,
      "loss": 0.9167,
      "step": 195
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.9807695393003765e-05,
      "loss": 0.9444,
      "step": 196
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.9805624284780955e-05,
      "loss": 0.9015,
      "step": 197
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.980354211960985e-05,
      "loss": 0.8981,
      "step": 198
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.980144889826285e-05,
      "loss": 0.9672,
      "step": 199
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.979934462151649e-05,
      "loss": 0.9574,
      "step": 200
    },
    {
      "epoch": 0.15,
      "eval_loss": 0.9466971755027771,
      "eval_runtime": 14.339,
      "eval_samples_per_second": 5.719,
      "eval_steps_per_second": 5.719,
      "step": 200
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.979722929015139e-05,
      "loss": 1.0925,
      "step": 201
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.9795102904952266e-05,
      "loss": 0.8727,
      "step": 202
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.9792965466707936e-05,
      "loss": 0.9032,
      "step": 203
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.9790816976211334e-05,
      "loss": 0.9172,
      "step": 204
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.978865743425947e-05,
      "loss": 0.9406,
      "step": 205
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.978648684165347e-05,
      "loss": 0.8997,
      "step": 206
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.978430519919856e-05,
      "loss": 0.881,
      "step": 207
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.978211250770406e-05,
      "loss": 0.9005,
      "step": 208
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.9779908767983376e-05,
      "loss": 0.9645,
      "step": 209
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.977769398085405e-05,
      "loss": 0.9006,
      "step": 210
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.977546814713768e-05,
      "loss": 0.9797,
      "step": 211
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.977323126765999e-05,
      "loss": 0.9193,
      "step": 212
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.977098334325078e-05,
      "loss": 0.9866,
      "step": 213
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.976872437474397e-05,
      "loss": 1.107,
      "step": 214
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.9766454362977564e-05,
      "loss": 0.9714,
      "step": 215
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.9764173308793666e-05,
      "loss": 0.9226,
      "step": 216
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.9761881213038465e-05,
      "loss": 1.001,
      "step": 217
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.975957807656228e-05,
      "loss": 0.9818,
      "step": 218
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.975726390021947e-05,
      "loss": 0.8939,
      "step": 219
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.975493868486854e-05,
      "loss": 0.8813,
      "step": 220
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.975260243137208e-05,
      "loss": 0.9387,
      "step": 221
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.9750255140596756e-05,
      "loss": 0.8727,
      "step": 222
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.9747896813413335e-05,
      "loss": 0.8764,
      "step": 223
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.974552745069669e-05,
      "loss": 0.8334,
      "step": 224
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.974314705332577e-05,
      "loss": 0.964,
      "step": 225
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.974075562218364e-05,
      "loss": 0.8916,
      "step": 226
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.973835315815744e-05,
      "loss": 0.9837,
      "step": 227
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.9735939662138406e-05,
      "loss": 0.9812,
      "step": 228
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.9733515135021864e-05,
      "loss": 0.908,
      "step": 229
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.9731079577707245e-05,
      "loss": 0.8595,
      "step": 230
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.9728632991098076e-05,
      "loss": 0.9381,
      "step": 231
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.9726175376101943e-05,
      "loss": 0.9119,
      "step": 232
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.972370673363056e-05,
      "loss": 0.8801,
      "step": 233
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.972122706459969e-05,
      "loss": 0.8782,
      "step": 234
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.971873636992925e-05,
      "loss": 0.9719,
      "step": 235
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.971623465054317e-05,
      "loss": 0.9726,
      "step": 236
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.971372190736955e-05,
      "loss": 0.8699,
      "step": 237
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.9711198141340505e-05,
      "loss": 0.9727,
      "step": 238
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.970866335339228e-05,
      "loss": 0.9451,
      "step": 239
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.970611754446522e-05,
      "loss": 0.9522,
      "step": 240
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.970356071550371e-05,
      "loss": 0.9961,
      "step": 241
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.970099286745628e-05,
      "loss": 0.8883,
      "step": 242
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.969841400127551e-05,
      "loss": 0.9045,
      "step": 243
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.969582411791807e-05,
      "loss": 0.9338,
      "step": 244
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.9693223218344734e-05,
      "loss": 0.9816,
      "step": 245
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.9690611303520356e-05,
      "loss": 0.9122,
      "step": 246
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.9687988374413874e-05,
      "loss": 0.8637,
      "step": 247
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.968535443199831e-05,
      "loss": 0.9587,
      "step": 248
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.9682709477250775e-05,
      "loss": 0.9989,
      "step": 249
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.968005351115245e-05,
      "loss": 0.9073,
      "step": 250
    },
    {
      "epoch": 0.19,
      "eval_loss": 0.9367482662200928,
      "eval_runtime": 14.3489,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 250
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.967738653468864e-05,
      "loss": 0.9046,
      "step": 251
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.967470854884869e-05,
      "loss": 0.9685,
      "step": 252
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.967201955462606e-05,
      "loss": 0.9696,
      "step": 253
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.966931955301827e-05,
      "loss": 0.9046,
      "step": 254
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.9666608545026946e-05,
      "loss": 0.9596,
      "step": 255
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.9663886531657785e-05,
      "loss": 0.9658,
      "step": 256
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.9661153513920566e-05,
      "loss": 0.8072,
      "step": 257
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.965840949282915e-05,
      "loss": 0.9069,
      "step": 258
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.965565446940149e-05,
      "loss": 0.8832,
      "step": 259
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.96528884446596e-05,
      "loss": 0.9458,
      "step": 260
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.9650111419629606e-05,
      "loss": 0.956,
      "step": 261
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.9647323395341675e-05,
      "loss": 0.8134,
      "step": 262
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.96445243728301e-05,
      "loss": 0.9372,
      "step": 263
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.964171435313321e-05,
      "loss": 1.0882,
      "step": 264
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.963889333729345e-05,
      "loss": 0.9947,
      "step": 265
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.963606132635731e-05,
      "loss": 0.9149,
      "step": 266
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.96332183213754e-05,
      "loss": 0.9228,
      "step": 267
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.963036432340237e-05,
      "loss": 0.9393,
      "step": 268
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.9627499333496965e-05,
      "loss": 0.8971,
      "step": 269
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.962462335272201e-05,
      "loss": 0.8505,
      "step": 270
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.96217363821444e-05,
      "loss": 0.9163,
      "step": 271
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.961883842283511e-05,
      "loss": 0.8878,
      "step": 272
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.961592947586919e-05,
      "loss": 0.9367,
      "step": 273
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.9613009542325765e-05,
      "loss": 0.9018,
      "step": 274
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.9610078623288056e-05,
      "loss": 0.9076,
      "step": 275
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.9607136719843326e-05,
      "loss": 0.9455,
      "step": 276
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.9604183833082926e-05,
      "loss": 0.859,
      "step": 277
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.96012199641023e-05,
      "loss": 0.9633,
      "step": 278
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.959824511400093e-05,
      "loss": 0.8642,
      "step": 279
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.9595259283882404e-05,
      "loss": 0.9142,
      "step": 280
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.959226247485438e-05,
      "loss": 0.8873,
      "step": 281
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.9589254688028556e-05,
      "loss": 0.9313,
      "step": 282
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.958623592452074e-05,
      "loss": 0.8899,
      "step": 283
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.95832061854508e-05,
      "loss": 0.9243,
      "step": 284
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.9580165471942676e-05,
      "loss": 0.9604,
      "step": 285
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.9577113785124364e-05,
      "loss": 0.9449,
      "step": 286
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.957405112612795e-05,
      "loss": 0.8803,
      "step": 287
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.957097749608959e-05,
      "loss": 0.9243,
      "step": 288
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.95678928961495e-05,
      "loss": 0.8947,
      "step": 289
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.9564797327451975e-05,
      "loss": 0.8416,
      "step": 290
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.9561690791145356e-05,
      "loss": 0.8503,
      "step": 291
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.955857328838209e-05,
      "loss": 0.9263,
      "step": 292
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.955544482031866e-05,
      "loss": 0.9338,
      "step": 293
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.955230538811564e-05,
      "loss": 0.8688,
      "step": 294
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.954915499293765e-05,
      "loss": 0.8991,
      "step": 295
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.954599363595339e-05,
      "loss": 0.9422,
      "step": 296
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.954282131833562e-05,
      "loss": 0.9208,
      "step": 297
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.953963804126119e-05,
      "loss": 0.8805,
      "step": 298
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.953644380591097e-05,
      "loss": 0.8969,
      "step": 299
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.953323861346993e-05,
      "loss": 0.9326,
      "step": 300
    },
    {
      "epoch": 0.23,
      "eval_loss": 0.9325690865516663,
      "eval_runtime": 14.3495,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 300
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.9530022465127104e-05,
      "loss": 0.9887,
      "step": 301
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.952679536207558e-05,
      "loss": 0.9263,
      "step": 302
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.9523557305512495e-05,
      "loss": 0.9236,
      "step": 303
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.952030829663907e-05,
      "loss": 0.8056,
      "step": 304
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.951704833666061e-05,
      "loss": 0.9275,
      "step": 305
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.951377742678644e-05,
      "loss": 0.8818,
      "step": 306
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.951049556822996e-05,
      "loss": 0.9973,
      "step": 307
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.950720276220863e-05,
      "loss": 0.9542,
      "step": 308
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.9503899009944006e-05,
      "loss": 0.9337,
      "step": 309
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.950058431266165e-05,
      "loss": 0.9551,
      "step": 310
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.949725867159121e-05,
      "loss": 0.8586,
      "step": 311
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.949392208796642e-05,
      "loss": 0.8877,
      "step": 312
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.9490574563025014e-05,
      "loss": 0.8719,
      "step": 313
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.9487216098008846e-05,
      "loss": 0.9017,
      "step": 314
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.948384669416379e-05,
      "loss": 0.9108,
      "step": 315
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.948046635273978e-05,
      "loss": 0.8661,
      "step": 316
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.947707507499083e-05,
      "loss": 0.9431,
      "step": 317
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.947367286217499e-05,
      "loss": 0.8706,
      "step": 318
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.947025971555438e-05,
      "loss": 0.9148,
      "step": 319
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.946683563639516e-05,
      "loss": 0.9137,
      "step": 320
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.9463400625967564e-05,
      "loss": 0.9359,
      "step": 321
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.945995468554587e-05,
      "loss": 1.0004,
      "step": 322
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.945649781640841e-05,
      "loss": 0.915,
      "step": 323
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.945303001983758e-05,
      "loss": 0.841,
      "step": 324
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.9449551297119825e-05,
      "loss": 0.9262,
      "step": 325
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.944606164954563e-05,
      "loss": 0.8751,
      "step": 326
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.9442561078409554e-05,
      "loss": 0.9131,
      "step": 327
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.9439049585010195e-05,
      "loss": 0.9083,
      "step": 328
    },
    {
      "epoch": 0.25,
      "learning_rate": 5.94355271706502e-05,
      "loss": 0.9454,
      "step": 329
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.94319938366363e-05,
      "loss": 0.965,
      "step": 330
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9428449584279216e-05,
      "loss": 0.9193,
      "step": 331
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.942489441489377e-05,
      "loss": 0.8919,
      "step": 332
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9421328329798834e-05,
      "loss": 0.9617,
      "step": 333
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.941775133031729e-05,
      "loss": 0.9865,
      "step": 334
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9414163417776094e-05,
      "loss": 0.8605,
      "step": 335
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9410564593506254e-05,
      "loss": 0.8571,
      "step": 336
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.940695485884282e-05,
      "loss": 0.9212,
      "step": 337
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9403334215124895e-05,
      "loss": 0.9161,
      "step": 338
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9399702663695626e-05,
      "loss": 0.8645,
      "step": 339
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9396060205902186e-05,
      "loss": 0.8923,
      "step": 340
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.939240684309583e-05,
      "loss": 0.8962,
      "step": 341
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.9388742576631825e-05,
      "loss": 0.8472,
      "step": 342
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.938506740786952e-05,
      "loss": 0.8725,
      "step": 343
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.938138133817226e-05,
      "loss": 0.8891,
      "step": 344
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.937768436890749e-05,
      "loss": 0.9247,
      "step": 345
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.937397650144665e-05,
      "loss": 0.8729,
      "step": 346
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.937025773716525e-05,
      "loss": 0.9109,
      "step": 347
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.936652807744283e-05,
      "loss": 0.999,
      "step": 348
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.936278752366297e-05,
      "loss": 0.8707,
      "step": 349
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.935903607721332e-05,
      "loss": 0.823,
      "step": 350
    },
    {
      "epoch": 0.27,
      "eval_loss": 0.9278062582015991,
      "eval_runtime": 14.3458,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 350
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.9355273739485516e-05,
      "loss": 0.8985,
      "step": 351
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.935150051187531e-05,
      "loss": 1.0011,
      "step": 352
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.9347716395782404e-05,
      "loss": 0.9014,
      "step": 353
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.934392139261061e-05,
      "loss": 0.874,
      "step": 354
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.9340115503767755e-05,
      "loss": 0.9043,
      "step": 355
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.933629873066571e-05,
      "loss": 0.9621,
      "step": 356
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.933247107472036e-05,
      "loss": 0.8471,
      "step": 357
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.932863253735165e-05,
      "loss": 0.9379,
      "step": 358
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.9324783119983564e-05,
      "loss": 0.9738,
      "step": 359
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.932092282404411e-05,
      "loss": 0.8806,
      "step": 360
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.931705165096534e-05,
      "loss": 0.9184,
      "step": 361
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.931316960218334e-05,
      "loss": 0.9544,
      "step": 362
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.930927667913821e-05,
      "loss": 0.941,
      "step": 363
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.930537288327412e-05,
      "loss": 0.8859,
      "step": 364
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.930145821603925e-05,
      "loss": 0.9242,
      "step": 365
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.929753267888582e-05,
      "loss": 0.8827,
      "step": 366
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.9293596273270074e-05,
      "loss": 1.0236,
      "step": 367
    },
    {
      "epoch": 0.28,
      "learning_rate": 5.92896490006523e-05,
      "loss": 0.8813,
      "step": 368
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.9285690862496814e-05,
      "loss": 0.9873,
      "step": 369
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.928172186027195e-05,
      "loss": 0.9933,
      "step": 370
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.92777419954501e-05,
      "loss": 0.909,
      "step": 371
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.927375126950765e-05,
      "loss": 0.909,
      "step": 372
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.926974968392505e-05,
      "loss": 0.9498,
      "step": 373
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.926573724018676e-05,
      "loss": 0.9165,
      "step": 374
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.926171393978126e-05,
      "loss": 0.8861,
      "step": 375
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.9257679784201076e-05,
      "loss": 0.9242,
      "step": 376
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.9253634774942736e-05,
      "loss": 0.9005,
      "step": 377
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.9249578913506835e-05,
      "loss": 0.848,
      "step": 378
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.924551220139796e-05,
      "loss": 0.9174,
      "step": 379
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.9241434640124745e-05,
      "loss": 0.9594,
      "step": 380
    },
    {
      "epoch": 0.29,
      "learning_rate": 5.92373462311998e-05,
      "loss": 0.8323,
      "step": 381
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.9233246976139843e-05,
      "loss": 0.9557,
      "step": 382
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.9229136876465537e-05,
      "loss": 0.8413,
      "step": 383
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.922501593370161e-05,
      "loss": 0.9098,
      "step": 384
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.92208841493768e-05,
      "loss": 0.9021,
      "step": 385
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.921674152502387e-05,
      "loss": 0.8922,
      "step": 386
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.921258806217961e-05,
      "loss": 0.991,
      "step": 387
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.92084237623848e-05,
      "loss": 0.9665,
      "step": 388
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.92042486271843e-05,
      "loss": 0.9502,
      "step": 389
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.920006265812693e-05,
      "loss": 0.942,
      "step": 390
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.919586585676556e-05,
      "loss": 0.9667,
      "step": 391
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.919165822465706e-05,
      "loss": 0.8824,
      "step": 392
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.9187439763362343e-05,
      "loss": 0.9201,
      "step": 393
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.918321047444632e-05,
      "loss": 1.009,
      "step": 394
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.917897035947793e-05,
      "loss": 0.8835,
      "step": 395
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.9174719420030106e-05,
      "loss": 0.8517,
      "step": 396
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.917045765767983e-05,
      "loss": 0.9925,
      "step": 397
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.9166185074008076e-05,
      "loss": 0.9388,
      "step": 398
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.9161901670599836e-05,
      "loss": 0.8664,
      "step": 399
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.915760744904412e-05,
      "loss": 0.905,
      "step": 400
    },
    {
      "epoch": 0.31,
      "eval_loss": 0.9229549169540405,
      "eval_runtime": 14.344,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 400
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.915330241093394e-05,
      "loss": 0.9731,
      "step": 401
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.914898655786635e-05,
      "loss": 0.8997,
      "step": 402
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.914465989144238e-05,
      "loss": 1.0159,
      "step": 403
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.9140322413267084e-05,
      "loss": 0.8836,
      "step": 404
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.913597412494954e-05,
      "loss": 0.9752,
      "step": 405
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.913161502810282e-05,
      "loss": 0.899,
      "step": 406
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.912724512434402e-05,
      "loss": 0.9651,
      "step": 407
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.912286441529422e-05,
      "loss": 0.8707,
      "step": 408
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.911847290257854e-05,
      "loss": 0.8244,
      "step": 409
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.9114070587826095e-05,
      "loss": 0.9235,
      "step": 410
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.9109657472669995e-05,
      "loss": 0.8612,
      "step": 411
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.910523355874737e-05,
      "loss": 0.8909,
      "step": 412
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.910079884769935e-05,
      "loss": 0.9639,
      "step": 413
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.9096353341171085e-05,
      "loss": 0.9975,
      "step": 414
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.90918970408117e-05,
      "loss": 0.8918,
      "step": 415
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.9087429948274356e-05,
      "loss": 0.9594,
      "step": 416
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.90829520652162e-05,
      "loss": 0.868,
      "step": 417
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.907846339329838e-05,
      "loss": 0.9264,
      "step": 418
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.907396393418607e-05,
      "loss": 0.9042,
      "step": 419
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.90694536895484e-05,
      "loss": 0.9016,
      "step": 420
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.9064932661058554e-05,
      "loss": 0.9021,
      "step": 421
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.906040085039368e-05,
      "loss": 0.9378,
      "step": 422
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.905585825923493e-05,
      "loss": 0.9121,
      "step": 423
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.905130488926748e-05,
      "loss": 0.9005,
      "step": 424
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.9046740742180475e-05,
      "loss": 0.8844,
      "step": 425
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.904216581966707e-05,
      "loss": 0.8901,
      "step": 426
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.9037580123424414e-05,
      "loss": 0.9657,
      "step": 427
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.903298365515367e-05,
      "loss": 0.9243,
      "step": 428
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.902837641655996e-05,
      "loss": 0.9058,
      "step": 429
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.902375840935245e-05,
      "loss": 0.8608,
      "step": 430
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.901912963524426e-05,
      "loss": 1.02,
      "step": 431
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.901449009595253e-05,
      "loss": 0.9886,
      "step": 432
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.9009839793198364e-05,
      "loss": 0.9695,
      "step": 433
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.9005178728706905e-05,
      "loss": 0.8573,
      "step": 434
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.900050690420724e-05,
      "loss": 0.9105,
      "step": 435
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.899582432143247e-05,
      "loss": 0.9772,
      "step": 436
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.89911309821197e-05,
      "loss": 0.89,
      "step": 437
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.898642688801e-05,
      "loss": 0.8906,
      "step": 438
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.8981712040848446e-05,
      "loss": 0.9414,
      "step": 439
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.897698644238409e-05,
      "loss": 0.857,
      "step": 440
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.897225009436998e-05,
      "loss": 0.834,
      "step": 441
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.896750299856317e-05,
      "loss": 0.8266,
      "step": 442
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.896274515672467e-05,
      "loss": 0.9454,
      "step": 443
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.895797657061948e-05,
      "loss": 0.8975,
      "step": 444
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.89531972420166e-05,
      "loss": 0.886,
      "step": 445
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.894840717268903e-05,
      "loss": 0.9169,
      "step": 446
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.894360636441371e-05,
      "loss": 0.8902,
      "step": 447
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.893879481897161e-05,
      "loss": 0.9005,
      "step": 448
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.893397253814765e-05,
      "loss": 0.9576,
      "step": 449
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.8929139523730735e-05,
      "loss": 0.8854,
      "step": 450
    },
    {
      "epoch": 0.35,
      "eval_loss": 0.9176666736602783,
      "eval_runtime": 14.3438,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 450
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.892429577751378e-05,
      "loss": 0.8726,
      "step": 451
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.8919441301293654e-05,
      "loss": 0.8685,
      "step": 452
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.8914576096871215e-05,
      "loss": 0.9095,
      "step": 453
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.89097001660513e-05,
      "loss": 0.9005,
      "step": 454
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.8904813510642724e-05,
      "loss": 0.9191,
      "step": 455
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.889991613245829e-05,
      "loss": 0.9875,
      "step": 456
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.889500803331476e-05,
      "loss": 0.886,
      "step": 457
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.88900892150329e-05,
      "loss": 0.9026,
      "step": 458
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8885159679437414e-05,
      "loss": 0.9143,
      "step": 459
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.888021942835702e-05,
      "loss": 0.9503,
      "step": 460
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8875268463624406e-05,
      "loss": 0.989,
      "step": 461
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.88703067870762e-05,
      "loss": 0.8855,
      "step": 462
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8865334400553045e-05,
      "loss": 0.9237,
      "step": 463
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.886035130589953e-05,
      "loss": 0.9647,
      "step": 464
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.885535750496423e-05,
      "loss": 0.9134,
      "step": 465
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8850352999599685e-05,
      "loss": 0.9029,
      "step": 466
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8845337791662415e-05,
      "loss": 0.8842,
      "step": 467
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8840311883012904e-05,
      "loss": 0.8617,
      "step": 468
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.88352752755156e-05,
      "loss": 0.9173,
      "step": 469
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.883022797103893e-05,
      "loss": 0.8786,
      "step": 470
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8825169971455286e-05,
      "loss": 0.8795,
      "step": 471
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.882010127864102e-05,
      "loss": 0.8865,
      "step": 472
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8815021894476465e-05,
      "loss": 0.8709,
      "step": 473
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.880993182084591e-05,
      "loss": 0.923,
      "step": 474
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.880483105963761e-05,
      "loss": 0.826,
      "step": 475
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8799719612743794e-05,
      "loss": 1.0261,
      "step": 476
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.879459748206064e-05,
      "loss": 0.8337,
      "step": 477
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8789464669488294e-05,
      "loss": 0.8766,
      "step": 478
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.878432117693088e-05,
      "loss": 0.8939,
      "step": 479
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.877916700629647e-05,
      "loss": 0.8606,
      "step": 480
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.877400215949708e-05,
      "loss": 0.905,
      "step": 481
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.876882663844873e-05,
      "loss": 0.8498,
      "step": 482
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.876364044507136e-05,
      "loss": 0.9171,
      "step": 483
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.87584435812889e-05,
      "loss": 0.8669,
      "step": 484
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.8753236049029205e-05,
      "loss": 0.9325,
      "step": 485
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.874801785022412e-05,
      "loss": 0.8923,
      "step": 486
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.874278898680941e-05,
      "loss": 0.8181,
      "step": 487
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.8737549460724854e-05,
      "loss": 0.9525,
      "step": 488
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.873229927391412e-05,
      "loss": 0.8537,
      "step": 489
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.8727038428324876e-05,
      "loss": 0.9342,
      "step": 490
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.872176692590873e-05,
      "loss": 0.921,
      "step": 491
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.871648476862124e-05,
      "loss": 0.8135,
      "step": 492
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.8711191958421923e-05,
      "loss": 0.9138,
      "step": 493
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.8705888497274236e-05,
      "loss": 0.9068,
      "step": 494
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.870057438714562e-05,
      "loss": 0.8974,
      "step": 495
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.869524963000741e-05,
      "loss": 0.9027,
      "step": 496
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.868991422783495e-05,
      "loss": 0.9058,
      "step": 497
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.8684568182607486e-05,
      "loss": 0.8869,
      "step": 498
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.8679211496308245e-05,
      "loss": 0.9143,
      "step": 499
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.8673844170924385e-05,
      "loss": 0.8421,
      "step": 500
    },
    {
      "epoch": 0.39,
      "eval_loss": 0.9152392745018005,
      "eval_runtime": 14.354,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 500
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.8668466208447025e-05,
      "loss": 0.909,
      "step": 501
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.86630776108712e-05,
      "loss": 0.9393,
      "step": 502
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.865767838019592e-05,
      "loss": 0.8721,
      "step": 503
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.865226851842414e-05,
      "loss": 0.8537,
      "step": 504
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.864684802756274e-05,
      "loss": 0.8315,
      "step": 505
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.864141690962254e-05,
      "loss": 0.8782,
      "step": 506
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.8635975166618324e-05,
      "loss": 0.951,
      "step": 507
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.863052280056881e-05,
      "loss": 0.8717,
      "step": 508
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.862505981349665e-05,
      "loss": 0.9315,
      "step": 509
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.861958620742845e-05,
      "loss": 0.9339,
      "step": 510
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.861410198439472e-05,
      "loss": 0.9994,
      "step": 511
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.8608607146429966e-05,
      "loss": 0.9129,
      "step": 512
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.8603101695572585e-05,
      "loss": 0.8851,
      "step": 513
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.8597585633864924e-05,
      "loss": 0.8472,
      "step": 514
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.859205896335327e-05,
      "loss": 0.8707,
      "step": 515
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.858652168608785e-05,
      "loss": 0.8989,
      "step": 516
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.85809738041228e-05,
      "loss": 0.8433,
      "step": 517
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.857541531951623e-05,
      "loss": 0.9697,
      "step": 518
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.8569846234330156e-05,
      "loss": 0.9184,
      "step": 519
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.856426655063053e-05,
      "loss": 0.897,
      "step": 520
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.855867627048725e-05,
      "loss": 0.9788,
      "step": 521
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.8553075395974125e-05,
      "loss": 0.9255,
      "step": 522
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.85474639291689e-05,
      "loss": 0.8777,
      "step": 523
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.8541841872153256e-05,
      "loss": 0.8764,
      "step": 524
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.85362092270128e-05,
      "loss": 0.8822,
      "step": 525
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.853056599583707e-05,
      "loss": 1.0576,
      "step": 526
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.8524912180719526e-05,
      "loss": 0.8794,
      "step": 527
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.851924778375754e-05,
      "loss": 0.9405,
      "step": 528
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.851357280705245e-05,
      "loss": 0.8427,
      "step": 529
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.8507887252709475e-05,
      "loss": 0.982,
      "step": 530
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.850219112283779e-05,
      "loss": 0.8303,
      "step": 531
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.849648441955047e-05,
      "loss": 0.9272,
      "step": 532
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.849076714496452e-05,
      "loss": 0.8439,
      "step": 533
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.848503930120088e-05,
      "loss": 0.8927,
      "step": 534
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.8479300890384386e-05,
      "loss": 0.8113,
      "step": 535
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.847355191464382e-05,
      "loss": 0.9015,
      "step": 536
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.846779237611187e-05,
      "loss": 0.8994,
      "step": 537
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8462022276925134e-05,
      "loss": 0.8438,
      "step": 538
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8456241619224144e-05,
      "loss": 0.9202,
      "step": 539
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8450450405153344e-05,
      "loss": 0.8469,
      "step": 540
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.844464863686109e-05,
      "loss": 0.8866,
      "step": 541
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.843883631649965e-05,
      "loss": 0.9187,
      "step": 542
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.843301344622523e-05,
      "loss": 0.9825,
      "step": 543
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8427180028197904e-05,
      "loss": 0.9502,
      "step": 544
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.84213360645817e-05,
      "loss": 0.8491,
      "step": 545
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8415481557544555e-05,
      "loss": 0.9165,
      "step": 546
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.840961650925829e-05,
      "loss": 0.8767,
      "step": 547
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.840374092189866e-05,
      "loss": 0.8619,
      "step": 548
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.839785479764532e-05,
      "loss": 0.9788,
      "step": 549
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.839195813868185e-05,
      "loss": 0.9077,
      "step": 550
    },
    {
      "epoch": 0.43,
      "eval_loss": 0.9117449522018433,
      "eval_runtime": 14.3485,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 550
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.838605094719571e-05,
      "loss": 0.9642,
      "step": 551
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.838013322537828e-05,
      "loss": 0.8709,
      "step": 552
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.8374204975424865e-05,
      "loss": 0.9959,
      "step": 553
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.836826619953465e-05,
      "loss": 0.8574,
      "step": 554
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.8362316899910726e-05,
      "loss": 0.8509,
      "step": 555
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.835635707876011e-05,
      "loss": 0.8414,
      "step": 556
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.8350386738293695e-05,
      "loss": 0.9044,
      "step": 557
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.83444058807263e-05,
      "loss": 1.0612,
      "step": 558
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.833841450827663e-05,
      "loss": 0.8717,
      "step": 559
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.833241262316729e-05,
      "loss": 0.8887,
      "step": 560
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.83264002276248e-05,
      "loss": 0.9073,
      "step": 561
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.832037732387957e-05,
      "loss": 0.9127,
      "step": 562
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.83143439141659e-05,
      "loss": 0.852,
      "step": 563
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.8308300000721996e-05,
      "loss": 0.9675,
      "step": 564
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.830224558578997e-05,
      "loss": 0.8847,
      "step": 565
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.82961806716158e-05,
      "loss": 0.8906,
      "step": 566
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.82901052604494e-05,
      "loss": 0.7885,
      "step": 567
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.8284019354544544e-05,
      "loss": 0.8548,
      "step": 568
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.827792295615891e-05,
      "loss": 0.8745,
      "step": 569
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.827181606755408e-05,
      "loss": 0.9804,
      "step": 570
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.8265698690995514e-05,
      "loss": 0.9415,
      "step": 571
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.825957082875255e-05,
      "loss": 0.8596,
      "step": 572
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.825343248309845e-05,
      "loss": 0.9504,
      "step": 573
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.824728365631035e-05,
      "loss": 0.8962,
      "step": 574
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.8241124350669254e-05,
      "loss": 0.8937,
      "step": 575
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.823495456846009e-05,
      "loss": 0.9165,
      "step": 576
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.822877431197163e-05,
      "loss": 0.8571,
      "step": 577
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.822258358349656e-05,
      "loss": 0.9668,
      "step": 578
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.8216382385331474e-05,
      "loss": 0.9476,
      "step": 579
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.8210170719776796e-05,
      "loss": 0.9074,
      "step": 580
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.820394858913685e-05,
      "loss": 0.8916,
      "step": 581
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.819771599571987e-05,
      "loss": 0.8664,
      "step": 582
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.819147294183795e-05,
      "loss": 0.9233,
      "step": 583
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.8185219429807054e-05,
      "loss": 0.8575,
      "step": 584
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.8178955461947046e-05,
      "loss": 0.8917,
      "step": 585
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.817268104058167e-05,
      "loss": 0.8712,
      "step": 586
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.816639616803851e-05,
      "loss": 0.8683,
      "step": 587
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.816010084664909e-05,
      "loss": 0.9248,
      "step": 588
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.815379507874876e-05,
      "loss": 0.9097,
      "step": 589
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.814747886667676e-05,
      "loss": 0.8979,
      "step": 590
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.814115221277621e-05,
      "loss": 0.8521,
      "step": 591
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.81348151193941e-05,
      "loss": 0.8079,
      "step": 592
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.812846758888127e-05,
      "loss": 0.9141,
      "step": 593
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.8122109623592495e-05,
      "loss": 0.8798,
      "step": 594
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.811574122588635e-05,
      "loss": 0.8782,
      "step": 595
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.8109362398125325e-05,
      "loss": 0.9246,
      "step": 596
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.8102973142675756e-05,
      "loss": 0.9663,
      "step": 597
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.809657346190785e-05,
      "loss": 0.9701,
      "step": 598
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.8090163358195705e-05,
      "loss": 0.919,
      "step": 599
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.8083742833917254e-05,
      "loss": 0.9248,
      "step": 600
    },
    {
      "epoch": 0.46,
      "eval_loss": 0.9059345126152039,
      "eval_runtime": 14.3408,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 600
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.807731189145431e-05,
      "loss": 0.9101,
      "step": 601
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.807087053319256e-05,
      "loss": 0.8648,
      "step": 602
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.806441876152153e-05,
      "loss": 0.8518,
      "step": 603
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.805795657883465e-05,
      "loss": 0.8208,
      "step": 604
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.805148398752914e-05,
      "loss": 0.937,
      "step": 605
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.8045000990006184e-05,
      "loss": 0.868,
      "step": 606
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.803850758867072e-05,
      "loss": 0.8612,
      "step": 607
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.8032003785931634e-05,
      "loss": 0.8796,
      "step": 608
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.8025489584201604e-05,
      "loss": 0.8998,
      "step": 609
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.801896498589722e-05,
      "loss": 0.8788,
      "step": 610
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.801242999343888e-05,
      "loss": 0.8722,
      "step": 611
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.8005884609250866e-05,
      "loss": 0.908,
      "step": 612
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.799932883576132e-05,
      "loss": 0.7785,
      "step": 613
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.7992762675402206e-05,
      "loss": 0.8833,
      "step": 614
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.798618613060939e-05,
      "loss": 0.9345,
      "step": 615
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.7979599203822534e-05,
      "loss": 0.8909,
      "step": 616
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.7973001897485216e-05,
      "loss": 0.8355,
      "step": 617
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.79663942140448e-05,
      "loss": 0.9329,
      "step": 618
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.795977615595254e-05,
      "loss": 0.8372,
      "step": 619
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.795314772566352e-05,
      "loss": 0.9102,
      "step": 620
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.794650892563669e-05,
      "loss": 0.8206,
      "step": 621
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.793985975833483e-05,
      "loss": 0.9362,
      "step": 622
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.793320022622456e-05,
      "loss": 0.9391,
      "step": 623
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.792653033177638e-05,
      "loss": 0.8853,
      "step": 624
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.791985007746459e-05,
      "loss": 0.9109,
      "step": 625
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.791315946576737e-05,
      "loss": 0.9542,
      "step": 626
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.790645849916672e-05,
      "loss": 0.8725,
      "step": 627
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.789974718014848e-05,
      "loss": 0.8199,
      "step": 628
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.789302551120235e-05,
      "loss": 0.8111,
      "step": 629
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.788629349482185e-05,
      "loss": 0.9484,
      "step": 630
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.787955113350436e-05,
      "loss": 0.9182,
      "step": 631
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.787279842975106e-05,
      "loss": 0.9239,
      "step": 632
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.7866035386067006e-05,
      "loss": 0.9145,
      "step": 633
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.7859262004961065e-05,
      "loss": 0.8298,
      "step": 634
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.7852478288945966e-05,
      "loss": 0.899,
      "step": 635
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.784568424053823e-05,
      "loss": 0.8257,
      "step": 636
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.7838879862258265e-05,
      "loss": 1.0265,
      "step": 637
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.783206515663025e-05,
      "loss": 0.898,
      "step": 638
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.7825240126182254e-05,
      "loss": 0.8298,
      "step": 639
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.7818404773446126e-05,
      "loss": 0.9161,
      "step": 640
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.781155910095757e-05,
      "loss": 0.9423,
      "step": 641
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.7804703111256125e-05,
      "loss": 0.8903,
      "step": 642
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.779783680688514e-05,
      "loss": 0.9239,
      "step": 643
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.7790960190391804e-05,
      "loss": 0.9202,
      "step": 644
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.7784073264327115e-05,
      "loss": 0.8425,
      "step": 645
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.777717603124593e-05,
      "loss": 0.8335,
      "step": 646
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.777026849370687e-05,
      "loss": 0.9355,
      "step": 647
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.776335065427244e-05,
      "loss": 0.8531,
      "step": 648
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.775642251550893e-05,
      "loss": 0.9,
      "step": 649
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.7749484079986456e-05,
      "loss": 0.8229,
      "step": 650
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.9039155840873718,
      "eval_runtime": 14.346,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 650
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.7742535350278983e-05,
      "loss": 0.8567,
      "step": 651
    },
    {
      "epoch": 0.5,
      "learning_rate": 5.773557632896424e-05,
      "loss": 0.8965,
      "step": 652
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.772860701862383e-05,
      "loss": 0.8232,
      "step": 653
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.772162742184313e-05,
      "loss": 0.872,
      "step": 654
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.771463754121136e-05,
      "loss": 0.8713,
      "step": 655
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.770763737932155e-05,
      "loss": 0.8131,
      "step": 656
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.7700626938770524e-05,
      "loss": 0.8987,
      "step": 657
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.769360622215896e-05,
      "loss": 0.9732,
      "step": 658
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.768657523209129e-05,
      "loss": 0.9619,
      "step": 659
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.767953397117581e-05,
      "loss": 0.8869,
      "step": 660
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.767248244202461e-05,
      "loss": 0.8991,
      "step": 661
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.766542064725357e-05,
      "loss": 0.9926,
      "step": 662
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.76583485894824e-05,
      "loss": 0.8709,
      "step": 663
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.7651266271334626e-05,
      "loss": 0.8714,
      "step": 664
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.764417369543755e-05,
      "loss": 0.9332,
      "step": 665
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.763707086442229e-05,
      "loss": 0.9,
      "step": 666
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.762995778092378e-05,
      "loss": 0.8855,
      "step": 667
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.762283444758075e-05,
      "loss": 0.9365,
      "step": 668
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.761570086703573e-05,
      "loss": 0.8918,
      "step": 669
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.7608557041935055e-05,
      "loss": 0.9494,
      "step": 670
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.760140297492886e-05,
      "loss": 0.8959,
      "step": 671
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.759423866867108e-05,
      "loss": 0.8714,
      "step": 672
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.7587064125819444e-05,
      "loss": 0.9425,
      "step": 673
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.757987934903548e-05,
      "loss": 0.9575,
      "step": 674
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.7572684340984526e-05,
      "loss": 0.931,
      "step": 675
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.756547910433568e-05,
      "loss": 0.8957,
      "step": 676
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.755826364176188e-05,
      "loss": 0.8972,
      "step": 677
    },
    {
      "epoch": 0.52,
      "learning_rate": 5.755103795593983e-05,
      "loss": 0.9123,
      "step": 678
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7543802049550026e-05,
      "loss": 0.8302,
      "step": 679
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7536555925276766e-05,
      "loss": 0.8715,
      "step": 680
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.752929958580814e-05,
      "loss": 0.8265,
      "step": 681
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7522033033836004e-05,
      "loss": 0.8706,
      "step": 682
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7514756272056025e-05,
      "loss": 0.8956,
      "step": 683
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7507469303167674e-05,
      "loss": 0.9307,
      "step": 684
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7500172129874165e-05,
      "loss": 0.8777,
      "step": 685
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7492864754882527e-05,
      "loss": 0.8308,
      "step": 686
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.748554718090356e-05,
      "loss": 0.8611,
      "step": 687
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.747821941065186e-05,
      "loss": 0.8826,
      "step": 688
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.747088144684581e-05,
      "loss": 0.9537,
      "step": 689
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.7463533292207535e-05,
      "loss": 0.8979,
      "step": 690
    },
    {
      "epoch": 0.53,
      "learning_rate": 5.745617494946299e-05,
      "loss": 0.8807,
      "step": 691
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7448806421341884e-05,
      "loss": 0.8096,
      "step": 692
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.744142771057771e-05,
      "loss": 0.8452,
      "step": 693
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.743403881990773e-05,
      "loss": 0.9678,
      "step": 694
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7426639752073e-05,
      "loss": 0.8519,
      "step": 695
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.741923050981834e-05,
      "loss": 0.8548,
      "step": 696
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7411811095892344e-05,
      "loss": 0.8179,
      "step": 697
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.740438151304737e-05,
      "loss": 0.864,
      "step": 698
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.739694176403958e-05,
      "loss": 0.9448,
      "step": 699
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7389491851628865e-05,
      "loss": 0.9499,
      "step": 700
    },
    {
      "epoch": 0.54,
      "eval_loss": 0.9012778401374817,
      "eval_runtime": 14.3419,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 700
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7382031778578935e-05,
      "loss": 0.8603,
      "step": 701
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7374561547657214e-05,
      "loss": 0.8941,
      "step": 702
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7367081161634936e-05,
      "loss": 0.909,
      "step": 703
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.7359590623287085e-05,
      "loss": 0.9642,
      "step": 704
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.735208993539242e-05,
      "loss": 0.9763,
      "step": 705
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7344579100733463e-05,
      "loss": 0.8517,
      "step": 706
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7337058122096474e-05,
      "loss": 0.8502,
      "step": 707
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.732952700227154e-05,
      "loss": 0.9014,
      "step": 708
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7321985744052426e-05,
      "loss": 0.9876,
      "step": 709
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.731443435023673e-05,
      "loss": 0.8273,
      "step": 710
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7306872823625776e-05,
      "loss": 0.8698,
      "step": 711
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7299301167024636e-05,
      "loss": 0.9286,
      "step": 712
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.729171938324218e-05,
      "loss": 0.917,
      "step": 713
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7284127475091005e-05,
      "loss": 0.855,
      "step": 714
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.727652544538746e-05,
      "loss": 0.8091,
      "step": 715
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7268913296951664e-05,
      "loss": 0.873,
      "step": 716
    },
    {
      "epoch": 0.55,
      "learning_rate": 5.7261291032607484e-05,
      "loss": 0.8522,
      "step": 717
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.725365865518254e-05,
      "loss": 1.0029,
      "step": 718
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.72460161675082e-05,
      "loss": 0.8087,
      "step": 719
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.72383635724196e-05,
      "loss": 0.8944,
      "step": 720
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.72307008727556e-05,
      "loss": 0.8752,
      "step": 721
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.722302807135882e-05,
      "loss": 0.8779,
      "step": 722
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.7215345171075635e-05,
      "loss": 0.9601,
      "step": 723
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.720765217475616e-05,
      "loss": 0.9212,
      "step": 724
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.719994908525424e-05,
      "loss": 0.9622,
      "step": 725
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.719223590542749e-05,
      "loss": 0.8445,
      "step": 726
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.718451263813725e-05,
      "loss": 0.862,
      "step": 727
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.717677928624862e-05,
      "loss": 0.9523,
      "step": 728
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.7169035852630415e-05,
      "loss": 0.8818,
      "step": 729
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.7161282340155227e-05,
      "loss": 0.8504,
      "step": 730
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.7153518751699345e-05,
      "loss": 0.8917,
      "step": 731
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.7145745090142826e-05,
      "loss": 0.8896,
      "step": 732
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.713796135836945e-05,
      "loss": 0.8805,
      "step": 733
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.713016755926674e-05,
      "loss": 0.8733,
      "step": 734
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.712236369572595e-05,
      "loss": 0.7904,
      "step": 735
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.711454977064205e-05,
      "loss": 0.8426,
      "step": 736
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.710672578691379e-05,
      "loss": 0.8387,
      "step": 737
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.70988917474436e-05,
      "loss": 0.8782,
      "step": 738
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.709104765513767e-05,
      "loss": 0.8335,
      "step": 739
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.708319351290591e-05,
      "loss": 0.8842,
      "step": 740
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.707532932366197e-05,
      "loss": 0.9647,
      "step": 741
    },
    {
      "epoch": 0.57,
      "learning_rate": 5.7067455090323195e-05,
      "loss": 0.8426,
      "step": 742
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.705957081581069e-05,
      "loss": 0.7764,
      "step": 743
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.705167650304927e-05,
      "loss": 0.8897,
      "step": 744
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.7043772154967484e-05,
      "loss": 0.8388,
      "step": 745
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.7035857774497595e-05,
      "loss": 0.8251,
      "step": 746
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.702793336457558e-05,
      "loss": 0.866,
      "step": 747
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.7019998928141144e-05,
      "loss": 0.9179,
      "step": 748
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.7012054468137725e-05,
      "loss": 0.8813,
      "step": 749
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.700409998751246e-05,
      "loss": 0.8964,
      "step": 750
    },
    {
      "epoch": 0.58,
      "eval_loss": 0.8968020081520081,
      "eval_runtime": 14.3411,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 750
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.699613548921622e-05,
      "loss": 0.8096,
      "step": 751
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.698816097620357e-05,
      "loss": 0.8947,
      "step": 752
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.69801764514328e-05,
      "loss": 0.8738,
      "step": 753
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.697218191786593e-05,
      "loss": 0.8185,
      "step": 754
    },
    {
      "epoch": 0.58,
      "learning_rate": 5.696417737846869e-05,
      "loss": 0.8905,
      "step": 755
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.695616283621049e-05,
      "loss": 0.8541,
      "step": 756
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.6948138294064484e-05,
      "loss": 1.0064,
      "step": 757
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.694010375500752e-05,
      "loss": 0.8539,
      "step": 758
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.693205922202016e-05,
      "loss": 0.9046,
      "step": 759
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.692400469808668e-05,
      "loss": 0.8888,
      "step": 760
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.6915940186195044e-05,
      "loss": 0.8379,
      "step": 761
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.6907865689336936e-05,
      "loss": 0.8347,
      "step": 762
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.6899781210507754e-05,
      "loss": 0.9094,
      "step": 763
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.6891686752706556e-05,
      "loss": 0.9074,
      "step": 764
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.688358231893615e-05,
      "loss": 0.8133,
      "step": 765
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.687546791220303e-05,
      "loss": 0.9195,
      "step": 766
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.686734353551737e-05,
      "loss": 0.8915,
      "step": 767
    },
    {
      "epoch": 0.59,
      "learning_rate": 5.685920919189307e-05,
      "loss": 0.8723,
      "step": 768
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.685106488434771e-05,
      "loss": 0.8626,
      "step": 769
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.684291061590258e-05,
      "loss": 0.8588,
      "step": 770
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.683474638958264e-05,
      "loss": 0.8745,
      "step": 771
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.6826572208416576e-05,
      "loss": 0.9324,
      "step": 772
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.681838807543675e-05,
      "loss": 0.8791,
      "step": 773
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.681019399367921e-05,
      "loss": 0.8418,
      "step": 774
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.680198996618371e-05,
      "loss": 0.8458,
      "step": 775
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.6793775995993676e-05,
      "loss": 0.8241,
      "step": 776
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.678555208615626e-05,
      "loss": 0.9038,
      "step": 777
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.677731823972223e-05,
      "loss": 0.7262,
      "step": 778
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.676907445974612e-05,
      "loss": 0.8631,
      "step": 779
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.676082074928609e-05,
      "loss": 0.943,
      "step": 780
    },
    {
      "epoch": 0.6,
      "learning_rate": 5.675255711140402e-05,
      "loss": 0.857,
      "step": 781
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.674428354916547e-05,
      "loss": 0.8835,
      "step": 782
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.673600006563964e-05,
      "loss": 0.8119,
      "step": 783
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.672770666389946e-05,
      "loss": 0.8788,
      "step": 784
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.6719403347021524e-05,
      "loss": 0.8798,
      "step": 785
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.6711090118086096e-05,
      "loss": 0.9142,
      "step": 786
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.670276698017714e-05,
      "loss": 0.863,
      "step": 787
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.669443393638223e-05,
      "loss": 0.9426,
      "step": 788
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.668609098979271e-05,
      "loss": 0.9019,
      "step": 789
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.6677738143503534e-05,
      "loss": 0.8161,
      "step": 790
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.6669375400613355e-05,
      "loss": 0.9096,
      "step": 791
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.666100276422446e-05,
      "loss": 0.9592,
      "step": 792
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.6652620237442866e-05,
      "loss": 0.8469,
      "step": 793
    },
    {
      "epoch": 0.61,
      "learning_rate": 5.6644227823378207e-05,
      "loss": 0.9142,
      "step": 794
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.663582552514381e-05,
      "loss": 0.8349,
      "step": 795
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.662741334585666e-05,
      "loss": 0.9619,
      "step": 796
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.661899128863742e-05,
      "loss": 0.8727,
      "step": 797
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.6610559356610396e-05,
      "loss": 0.9841,
      "step": 798
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.660211755290357e-05,
      "loss": 0.8318,
      "step": 799
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.6593665880648606e-05,
      "loss": 0.8897,
      "step": 800
    },
    {
      "epoch": 0.62,
      "eval_loss": 0.8941949009895325,
      "eval_runtime": 14.3548,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 800
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.6585204342980783e-05,
      "loss": 0.8329,
      "step": 801
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.6576732943039074e-05,
      "loss": 0.8769,
      "step": 802
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.65682516839661e-05,
      "loss": 0.8992,
      "step": 803
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.6559760568908145e-05,
      "loss": 0.8372,
      "step": 804
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.655125960101515e-05,
      "loss": 0.8693,
      "step": 805
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.65427487834407e-05,
      "loss": 0.8151,
      "step": 806
    },
    {
      "epoch": 0.62,
      "learning_rate": 5.653422811934204e-05,
      "loss": 0.8688,
      "step": 807
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.652569761188006e-05,
      "loss": 0.952,
      "step": 808
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.651715726421933e-05,
      "loss": 0.8595,
      "step": 809
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.6508607079528034e-05,
      "loss": 0.8442,
      "step": 810
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.6500047060978024e-05,
      "loss": 0.9144,
      "step": 811
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.6491477211744796e-05,
      "loss": 0.9222,
      "step": 812
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.64828975350075e-05,
      "loss": 0.8602,
      "step": 813
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.647430803394893e-05,
      "loss": 0.9584,
      "step": 814
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.646570871175551e-05,
      "loss": 0.9033,
      "step": 815
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.645709957161732e-05,
      "loss": 0.9517,
      "step": 816
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.644848061672809e-05,
      "loss": 0.8506,
      "step": 817
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.643985185028516e-05,
      "loss": 0.874,
      "step": 818
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.643121327548956e-05,
      "loss": 0.9528,
      "step": 819
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.6422564895545904e-05,
      "loss": 0.843,
      "step": 820
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6413906713662475e-05,
      "loss": 0.8059,
      "step": 821
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.64052387330512e-05,
      "loss": 0.8422,
      "step": 822
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6396560956927604e-05,
      "loss": 0.9148,
      "step": 823
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.638787338851088e-05,
      "loss": 0.8774,
      "step": 824
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6379176031023854e-05,
      "loss": 0.939,
      "step": 825
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6370468887692945e-05,
      "loss": 0.8618,
      "step": 826
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.636175196174825e-05,
      "loss": 0.891,
      "step": 827
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.635302525642348e-05,
      "loss": 0.8987,
      "step": 828
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.634428877495595e-05,
      "loss": 0.9016,
      "step": 829
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6335542520586617e-05,
      "loss": 0.8609,
      "step": 830
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.632678649656009e-05,
      "loss": 0.8138,
      "step": 831
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.631802070612455e-05,
      "loss": 0.8953,
      "step": 832
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.630924515253185e-05,
      "loss": 0.8122,
      "step": 833
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.630045983903743e-05,
      "loss": 0.9477,
      "step": 834
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.629166476890038e-05,
      "loss": 0.8222,
      "step": 835
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.628285994538338e-05,
      "loss": 0.9617,
      "step": 836
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.627404537175275e-05,
      "loss": 0.865,
      "step": 837
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.626522105127842e-05,
      "loss": 0.7969,
      "step": 838
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.6256386987233936e-05,
      "loss": 0.9645,
      "step": 839
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.624754318289645e-05,
      "loss": 0.8427,
      "step": 840
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.6238689641546736e-05,
      "loss": 0.831,
      "step": 841
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.6229826366469183e-05,
      "loss": 0.8311,
      "step": 842
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.622095336095179e-05,
      "loss": 0.9052,
      "step": 843
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.621207062828616e-05,
      "loss": 0.9529,
      "step": 844
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.62031781717675e-05,
      "loss": 0.8941,
      "step": 845
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.619427599469464e-05,
      "loss": 0.8498,
      "step": 846
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.618536410037002e-05,
      "loss": 0.8236,
      "step": 847
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.6176442492099635e-05,
      "loss": 0.9967,
      "step": 848
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.616751117319315e-05,
      "loss": 0.9366,
      "step": 849
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.61585701469638e-05,
      "loss": 0.9187,
      "step": 850
    },
    {
      "epoch": 0.66,
      "eval_loss": 0.8949863910675049,
      "eval_runtime": 14.3422,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 850
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.614961941672842e-05,
      "loss": 0.8379,
      "step": 851
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.614065898580745e-05,
      "loss": 0.9213,
      "step": 852
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.6131688857524925e-05,
      "loss": 0.8992,
      "step": 853
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.6122709035208485e-05,
      "loss": 0.904,
      "step": 854
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.611371952218936e-05,
      "loss": 0.8267,
      "step": 855
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.610472032180237e-05,
      "loss": 0.904,
      "step": 856
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.609571143738595e-05,
      "loss": 0.8653,
      "step": 857
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.60866928722821e-05,
      "loss": 0.9218,
      "step": 858
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.607766462983642e-05,
      "loss": 0.9337,
      "step": 859
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.6068626713398116e-05,
      "loss": 0.8817,
      "step": 860
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.6059579126319965e-05,
      "loss": 0.8493,
      "step": 861
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.605052187195834e-05,
      "loss": 0.8802,
      "step": 862
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.604145495367318e-05,
      "loss": 0.8963,
      "step": 863
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.603237837482805e-05,
      "loss": 0.9126,
      "step": 864
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.602329213879007e-05,
      "loss": 0.9063,
      "step": 865
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.601419624892994e-05,
      "loss": 0.8126,
      "step": 866
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.6005090708621946e-05,
      "loss": 0.8346,
      "step": 867
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.5995975521243966e-05,
      "loss": 0.8174,
      "step": 868
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.598685069017745e-05,
      "loss": 0.8269,
      "step": 869
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.5977716218807416e-05,
      "loss": 0.8838,
      "step": 870
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.596857211052245e-05,
      "loss": 0.9552,
      "step": 871
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.5959418368714764e-05,
      "loss": 0.8258,
      "step": 872
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.5950254996780084e-05,
      "loss": 0.8131,
      "step": 873
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.594108199811774e-05,
      "loss": 0.9021,
      "step": 874
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.593189937613062e-05,
      "loss": 0.8106,
      "step": 875
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.592270713422519e-05,
      "loss": 0.7695,
      "step": 876
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.5913505275811485e-05,
      "loss": 0.9251,
      "step": 877
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.5904293804303097e-05,
      "loss": 0.8606,
      "step": 878
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.58950727231172e-05,
      "loss": 0.8502,
      "step": 879
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.588584203567453e-05,
      "loss": 0.8431,
      "step": 880
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.5876601745399364e-05,
      "loss": 0.8561,
      "step": 881
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.586735185571957e-05,
      "loss": 0.8766,
      "step": 882
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.585809237006657e-05,
      "loss": 0.8661,
      "step": 883
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.584882329187534e-05,
      "loss": 0.8532,
      "step": 884
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.58395446245844e-05,
      "loss": 0.861,
      "step": 885
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.5830256371635874e-05,
      "loss": 0.9342,
      "step": 886
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.582095853647538e-05,
      "loss": 0.9382,
      "step": 887
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.581165112255215e-05,
      "loss": 0.8305,
      "step": 888
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.580233413331891e-05,
      "loss": 0.7964,
      "step": 889
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.579300757223199e-05,
      "loss": 0.8303,
      "step": 890
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.578367144275126e-05,
      "loss": 0.8899,
      "step": 891
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.5774325748340104e-05,
      "loss": 0.8736,
      "step": 892
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.57649704924655e-05,
      "loss": 0.9197,
      "step": 893
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.575560567859795e-05,
      "loss": 0.8775,
      "step": 894
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.5746231310211495e-05,
      "loss": 0.8338,
      "step": 895
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.5736847390783746e-05,
      "loss": 0.8329,
      "step": 896
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.572745392379583e-05,
      "loss": 0.7831,
      "step": 897
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.5718050912732435e-05,
      "loss": 0.795,
      "step": 898
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.570863836108178e-05,
      "loss": 0.9468,
      "step": 899
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.5699216272335615e-05,
      "loss": 0.7855,
      "step": 900
    },
    {
      "epoch": 0.7,
      "eval_loss": 0.8913570046424866,
      "eval_runtime": 14.3474,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 900
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.568978464998925e-05,
      "loss": 0.7732,
      "step": 901
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.568034349754152e-05,
      "loss": 0.9285,
      "step": 902
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.5670892818494795e-05,
      "loss": 0.8993,
      "step": 903
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.566143261635497e-05,
      "loss": 0.9271,
      "step": 904
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.5651962894631494e-05,
      "loss": 0.884,
      "step": 905
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.564248365683733e-05,
      "loss": 0.8917,
      "step": 906
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.5632994906488986e-05,
      "loss": 0.8178,
      "step": 907
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.562349664710646e-05,
      "loss": 0.8853,
      "step": 908
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.5613988882213357e-05,
      "loss": 0.905,
      "step": 909
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.560447161533671e-05,
      "loss": 0.882,
      "step": 910
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.559494485000715e-05,
      "loss": 0.9204,
      "step": 911
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.55854085897588e-05,
      "loss": 0.9245,
      "step": 912
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.557586283812932e-05,
      "loss": 0.9107,
      "step": 913
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.556630759865987e-05,
      "loss": 0.899,
      "step": 914
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.555674287489516e-05,
      "loss": 0.8228,
      "step": 915
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.554716867038339e-05,
      "loss": 0.8602,
      "step": 916
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.553758498867628e-05,
      "loss": 0.9472,
      "step": 917
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.55279918333291e-05,
      "loss": 0.8463,
      "step": 918
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.551838920790059e-05,
      "loss": 0.8159,
      "step": 919
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.550877711595302e-05,
      "loss": 0.8532,
      "step": 920
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.5499155561052184e-05,
      "loss": 0.852,
      "step": 921
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.548952454676737e-05,
      "loss": 0.9073,
      "step": 922
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.5479884076671376e-05,
      "loss": 0.8982,
      "step": 923
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.547023415434053e-05,
      "loss": 0.9325,
      "step": 924
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5460574783354634e-05,
      "loss": 0.9771,
      "step": 925
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.545090596729702e-05,
      "loss": 0.8209,
      "step": 926
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5441227709754495e-05,
      "loss": 0.8407,
      "step": 927
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5431540014317414e-05,
      "loss": 0.8746,
      "step": 928
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5421842884579594e-05,
      "loss": 0.8438,
      "step": 929
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.541213632413836e-05,
      "loss": 0.9129,
      "step": 930
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.540242033659456e-05,
      "loss": 0.8874,
      "step": 931
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5392694925552494e-05,
      "loss": 0.898,
      "step": 932
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5382960094619996e-05,
      "loss": 0.8126,
      "step": 933
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.537321584740839e-05,
      "loss": 0.9165,
      "step": 934
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5363462187532467e-05,
      "loss": 0.863,
      "step": 935
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5353699118610534e-05,
      "loss": 0.8454,
      "step": 936
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.534392664426439e-05,
      "loss": 0.9201,
      "step": 937
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.5334144768119316e-05,
      "loss": 0.8392,
      "step": 938
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.532435349380407e-05,
      "loss": 0.9189,
      "step": 939
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.531455282495091e-05,
      "loss": 0.8695,
      "step": 940
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.530474276519557e-05,
      "loss": 1.0612,
      "step": 941
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.5294923318177296e-05,
      "loss": 0.8439,
      "step": 942
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.528509448753877e-05,
      "loss": 0.8951,
      "step": 943
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.527525627692618e-05,
      "loss": 0.8056,
      "step": 944
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.5265408689989207e-05,
      "loss": 0.8821,
      "step": 945
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.525555173038098e-05,
      "loss": 0.8263,
      "step": 946
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.524568540175813e-05,
      "loss": 0.9063,
      "step": 947
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.5235809707780767e-05,
      "loss": 0.822,
      "step": 948
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.522592465211244e-05,
      "loss": 0.8603,
      "step": 949
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.521603023842019e-05,
      "loss": 0.8748,
      "step": 950
    },
    {
      "epoch": 0.74,
      "eval_loss": 0.8888280391693115,
      "eval_runtime": 14.3504,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 950
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.520612647037456e-05,
      "loss": 0.925,
      "step": 951
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.519621335164951e-05,
      "loss": 0.8473,
      "step": 952
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.518629088592251e-05,
      "loss": 0.8636,
      "step": 953
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.517635907687448e-05,
      "loss": 0.829,
      "step": 954
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.5166417928189804e-05,
      "loss": 0.8522,
      "step": 955
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.515646744355635e-05,
      "loss": 0.9416,
      "step": 956
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.51465076266654e-05,
      "loss": 0.9227,
      "step": 957
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.513653848121177e-05,
      "loss": 0.885,
      "step": 958
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.5126560010893676e-05,
      "loss": 0.8619,
      "step": 959
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.511657221941283e-05,
      "loss": 0.9174,
      "step": 960
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.5106575110474384e-05,
      "loss": 0.8525,
      "step": 961
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.509656868778695e-05,
      "loss": 0.8293,
      "step": 962
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.5086552955062585e-05,
      "loss": 0.905,
      "step": 963
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.507652791601683e-05,
      "loss": 0.8448,
      "step": 964
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.506649357436866e-05,
      "loss": 0.9456,
      "step": 965
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.505644993384047e-05,
      "loss": 0.8988,
      "step": 966
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.5046396998158165e-05,
      "loss": 0.8796,
      "step": 967
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.5036334771051056e-05,
      "loss": 0.8541,
      "step": 968
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.5026263256251906e-05,
      "loss": 0.9154,
      "step": 969
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.501618245749694e-05,
      "loss": 0.9356,
      "step": 970
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.5006092378525824e-05,
      "loss": 0.9185,
      "step": 971
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.499599302308163e-05,
      "loss": 0.8636,
      "step": 972
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.498588439491093e-05,
      "loss": 0.8768,
      "step": 973
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.49757664977637e-05,
      "loss": 0.8873,
      "step": 974
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.496563933539336e-05,
      "loss": 0.8334,
      "step": 975
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.4955502911556764e-05,
      "loss": 0.7884,
      "step": 976
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.4945357230014216e-05,
      "loss": 0.8446,
      "step": 977
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.493520229452943e-05,
      "loss": 0.8913,
      "step": 978
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.4925038108869576e-05,
      "loss": 0.8677,
      "step": 979
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.491486467680525e-05,
      "loss": 0.9592,
      "step": 980
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.490468200211047e-05,
      "loss": 0.8444,
      "step": 981
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.489449008856268e-05,
      "loss": 0.962,
      "step": 982
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.488428893994278e-05,
      "loss": 0.8356,
      "step": 983
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.4874078560035054e-05,
      "loss": 0.8641,
      "step": 984
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.486385895262725e-05,
      "loss": 0.7904,
      "step": 985
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.48536301215105e-05,
      "loss": 0.8758,
      "step": 986
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.4843392070479405e-05,
      "loss": 0.8741,
      "step": 987
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.483314480333193e-05,
      "loss": 0.8517,
      "step": 988
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.4822888323869506e-05,
      "loss": 0.9177,
      "step": 989
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.481262263589696e-05,
      "loss": 0.8776,
      "step": 990
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.480234774322254e-05,
      "loss": 0.9352,
      "step": 991
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.4792063649657905e-05,
      "loss": 0.8994,
      "step": 992
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.478177035901814e-05,
      "loss": 0.8878,
      "step": 993
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.4771467875121724e-05,
      "loss": 0.8775,
      "step": 994
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.476115620179054e-05,
      "loss": 0.8977,
      "step": 995
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.475083534284993e-05,
      "loss": 1.0301,
      "step": 996
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.4740505302128564e-05,
      "loss": 0.9255,
      "step": 997
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.47301660834586e-05,
      "loss": 0.8594,
      "step": 998
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.471981769067554e-05,
      "loss": 0.8798,
      "step": 999
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.470946012761832e-05,
      "loss": 0.8359,
      "step": 1000
    },
    {
      "epoch": 0.77,
      "eval_loss": 0.8880231380462646,
      "eval_runtime": 14.3483,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 1000
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.469909339812927e-05,
      "loss": 0.8456,
      "step": 1001
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.468871750605412e-05,
      "loss": 0.8645,
      "step": 1002
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.467833245524199e-05,
      "loss": 0.9429,
      "step": 1003
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.4667938249545416e-05,
      "loss": 0.976,
      "step": 1004
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.4657534892820326e-05,
      "loss": 0.8414,
      "step": 1005
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.4647122388926034e-05,
      "loss": 0.8938,
      "step": 1006
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.463670074172524e-05,
      "loss": 0.8265,
      "step": 1007
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.4626269955084066e-05,
      "loss": 0.9194,
      "step": 1008
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.4615830032871995e-05,
      "loss": 0.889,
      "step": 1009
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.46053809789619e-05,
      "loss": 0.918,
      "step": 1010
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.459492279723007e-05,
      "loss": 0.8926,
      "step": 1011
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.458445549155615e-05,
      "loss": 0.8545,
      "step": 1012
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.457397906582318e-05,
      "loss": 0.8266,
      "step": 1013
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.4563493523917594e-05,
      "loss": 0.8977,
      "step": 1014
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.455299886972919e-05,
      "loss": 0.8406,
      "step": 1015
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.454249510715115e-05,
      "loss": 0.8098,
      "step": 1016
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4531982240080054e-05,
      "loss": 0.893,
      "step": 1017
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.452146027241583e-05,
      "loss": 0.8664,
      "step": 1018
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4510929208061805e-05,
      "loss": 0.8563,
      "step": 1019
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4500389050924665e-05,
      "loss": 0.9148,
      "step": 1020
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.448983980491448e-05,
      "loss": 0.8875,
      "step": 1021
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4479281473944695e-05,
      "loss": 0.9022,
      "step": 1022
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.446871406193211e-05,
      "loss": 0.9037,
      "step": 1023
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4458137572796904e-05,
      "loss": 0.898,
      "step": 1024
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4447552010462624e-05,
      "loss": 0.8346,
      "step": 1025
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.4436957378856184e-05,
      "loss": 0.8174,
      "step": 1026
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.442635368190784e-05,
      "loss": 0.7377,
      "step": 1027
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.441574092355126e-05,
      "loss": 0.8535,
      "step": 1028
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.440511910772342e-05,
      "loss": 0.8878,
      "step": 1029
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.4394488238364676e-05,
      "loss": 0.8778,
      "step": 1030
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.4383848319418764e-05,
      "loss": 0.9867,
      "step": 1031
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.4373199354832754e-05,
      "loss": 0.7847,
      "step": 1032
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.436254134855707e-05,
      "loss": 0.8753,
      "step": 1033
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.4351874304545505e-05,
      "loss": 0.8297,
      "step": 1034
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.434119822675519e-05,
      "loss": 0.8327,
      "step": 1035
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.433051311914661e-05,
      "loss": 0.8878,
      "step": 1036
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.431981898568361e-05,
      "loss": 0.9493,
      "step": 1037
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.430911583033339e-05,
      "loss": 0.8616,
      "step": 1038
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.429840365706645e-05,
      "loss": 0.8094,
      "step": 1039
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.42876824698567e-05,
      "loss": 0.7852,
      "step": 1040
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.427695227268135e-05,
      "loss": 0.8982,
      "step": 1041
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.426621306952097e-05,
      "loss": 0.8993,
      "step": 1042
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.425546486435945e-05,
      "loss": 0.7602,
      "step": 1043
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.424470766118406e-05,
      "loss": 0.8864,
      "step": 1044
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.423394146398537e-05,
      "loss": 0.9181,
      "step": 1045
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.42231662767573e-05,
      "loss": 0.8539,
      "step": 1046
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.4212382103497106e-05,
      "loss": 0.8353,
      "step": 1047
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.4201588948205395e-05,
      "loss": 0.9303,
      "step": 1048
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.419078681488606e-05,
      "loss": 0.8489,
      "step": 1049
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.4179975707546365e-05,
      "loss": 0.9785,
      "step": 1050
    },
    {
      "epoch": 0.81,
      "eval_loss": 0.8838340640068054,
      "eval_runtime": 14.349,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 1050
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.416915563019689e-05,
      "loss": 1.0054,
      "step": 1051
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.415832658685154e-05,
      "loss": 0.916,
      "step": 1052
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.414748858152755e-05,
      "loss": 0.8414,
      "step": 1053
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.413664161824549e-05,
      "loss": 0.7944,
      "step": 1054
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.412578570102923e-05,
      "loss": 0.8821,
      "step": 1055
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.411492083390598e-05,
      "loss": 0.8735,
      "step": 1056
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.410404702090625e-05,
      "loss": 0.9337,
      "step": 1057
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.4093164266063905e-05,
      "loss": 0.8168,
      "step": 1058
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.4082272573416076e-05,
      "loss": 0.9905,
      "step": 1059
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.407137194700326e-05,
      "loss": 0.895,
      "step": 1060
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.406046239086924e-05,
      "loss": 0.8701,
      "step": 1061
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.404954390906112e-05,
      "loss": 0.8583,
      "step": 1062
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.403861650562931e-05,
      "loss": 0.8807,
      "step": 1063
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.402768018462754e-05,
      "loss": 0.8533,
      "step": 1064
    },
    {
      "epoch": 0.82,
      "learning_rate": 5.4016734950112825e-05,
      "loss": 0.9211,
      "step": 1065
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.400578080614552e-05,
      "loss": 0.8886,
      "step": 1066
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.3994817756789244e-05,
      "loss": 0.8251,
      "step": 1067
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.398384580611097e-05,
      "loss": 0.9983,
      "step": 1068
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.397286495818093e-05,
      "loss": 0.9099,
      "step": 1069
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.3961875217072685e-05,
      "loss": 1.0173,
      "step": 1070
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.395087658686307e-05,
      "loss": 0.8843,
      "step": 1071
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.393986907163224e-05,
      "loss": 1.0166,
      "step": 1072
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.3928852675463625e-05,
      "loss": 0.8313,
      "step": 1073
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.391782740244397e-05,
      "loss": 0.8441,
      "step": 1074
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.3906793256663314e-05,
      "loss": 0.8513,
      "step": 1075
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.389575024221496e-05,
      "loss": 0.9339,
      "step": 1076
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.388469836319553e-05,
      "loss": 0.8151,
      "step": 1077
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.387363762370491e-05,
      "loss": 0.8495,
      "step": 1078
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.3862568027846304e-05,
      "loss": 0.9172,
      "step": 1079
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.385148957972617e-05,
      "loss": 0.8647,
      "step": 1080
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.384040228345427e-05,
      "loss": 0.8994,
      "step": 1081
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.3829306143143634e-05,
      "loss": 0.8578,
      "step": 1082
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.38182011629106e-05,
      "loss": 0.8323,
      "step": 1083
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.3807087346874743e-05,
      "loss": 1.0322,
      "step": 1084
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.379596469915895e-05,
      "loss": 0.8803,
      "step": 1085
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.378483322388937e-05,
      "loss": 0.9787,
      "step": 1086
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.3773692925195435e-05,
      "loss": 0.9307,
      "step": 1087
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.376254380720984e-05,
      "loss": 0.8441,
      "step": 1088
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.375138587406855e-05,
      "loss": 0.8948,
      "step": 1089
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.3740219129910835e-05,
      "loss": 0.8514,
      "step": 1090
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.372904357887918e-05,
      "loss": 0.8839,
      "step": 1091
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.371785922511936e-05,
      "loss": 0.8411,
      "step": 1092
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.370666607278043e-05,
      "loss": 0.8117,
      "step": 1093
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.36954641260147e-05,
      "loss": 0.8656,
      "step": 1094
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.3684253388977724e-05,
      "loss": 0.8813,
      "step": 1095
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.3673033865828354e-05,
      "loss": 0.8953,
      "step": 1096
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.366180556072866e-05,
      "loss": 0.8153,
      "step": 1097
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.365056847784399e-05,
      "loss": 0.9304,
      "step": 1098
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.363932262134297e-05,
      "loss": 0.9051,
      "step": 1099
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.362806799539744e-05,
      "loss": 0.7917,
      "step": 1100
    },
    {
      "epoch": 0.85,
      "eval_loss": 0.8823765516281128,
      "eval_runtime": 14.3449,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 1100
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.361680460418252e-05,
      "loss": 0.8093,
      "step": 1101
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.360553245187656e-05,
      "loss": 0.8427,
      "step": 1102
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.3594251542661194e-05,
      "loss": 0.8185,
      "step": 1103
    },
    {
      "epoch": 0.85,
      "learning_rate": 5.358296188072126e-05,
      "loss": 0.8468,
      "step": 1104
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.3571663470244896e-05,
      "loss": 0.8316,
      "step": 1105
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.356035631542344e-05,
      "loss": 0.9634,
      "step": 1106
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.3549040420451476e-05,
      "loss": 0.8637,
      "step": 1107
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.353771578952688e-05,
      "loss": 0.8595,
      "step": 1108
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.35263824268507e-05,
      "loss": 0.9663,
      "step": 1109
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.351504033662728e-05,
      "loss": 0.9065,
      "step": 1110
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.350368952306417e-05,
      "loss": 0.9395,
      "step": 1111
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.349232999037216e-05,
      "loss": 0.849,
      "step": 1112
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.3480961742765284e-05,
      "loss": 0.8452,
      "step": 1113
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.346958478446081e-05,
      "loss": 0.8252,
      "step": 1114
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.3458199119679216e-05,
      "loss": 0.861,
      "step": 1115
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.344680475264423e-05,
      "loss": 0.8347,
      "step": 1116
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.343540168758282e-05,
      "loss": 0.76,
      "step": 1117
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.342398992872514e-05,
      "loss": 0.8713,
      "step": 1118
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.341256948030461e-05,
      "loss": 0.845,
      "step": 1119
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.3401140346557854e-05,
      "loss": 0.8518,
      "step": 1120
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.338970253172471e-05,
      "loss": 1.0125,
      "step": 1121
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.337825604004826e-05,
      "loss": 0.9063,
      "step": 1122
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.336680087577479e-05,
      "loss": 0.7845,
      "step": 1123
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.3355337043153796e-05,
      "loss": 0.8583,
      "step": 1124
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.334386454643801e-05,
      "loss": 0.8584,
      "step": 1125
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.333238338988336e-05,
      "loss": 0.8493,
      "step": 1126
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.3320893577749e-05,
      "loss": 0.9009,
      "step": 1127
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.330939511429729e-05,
      "loss": 0.8671,
      "step": 1128
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.329788800379378e-05,
      "loss": 0.9104,
      "step": 1129
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.328637225050727e-05,
      "loss": 0.8349,
      "step": 1130
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.327484785870973e-05,
      "loss": 0.8712,
      "step": 1131
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.326331483267635e-05,
      "loss": 0.8756,
      "step": 1132
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.3251773176685506e-05,
      "loss": 0.9156,
      "step": 1133
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.3240222895018806e-05,
      "loss": 0.8102,
      "step": 1134
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.322866399196104e-05,
      "loss": 0.9215,
      "step": 1135
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.321709647180018e-05,
      "loss": 0.8451,
      "step": 1136
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.3205520338827435e-05,
      "loss": 0.809,
      "step": 1137
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.319393559733718e-05,
      "loss": 0.808,
      "step": 1138
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.318234225162697e-05,
      "loss": 0.7634,
      "step": 1139
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.317074030599758e-05,
      "loss": 0.8315,
      "step": 1140
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.315912976475298e-05,
      "loss": 0.8239,
      "step": 1141
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.314751063220032e-05,
      "loss": 0.9905,
      "step": 1142
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.313588291264991e-05,
      "loss": 0.8341,
      "step": 1143
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.312424661041528e-05,
      "loss": 0.8208,
      "step": 1144
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.311260172981312e-05,
      "loss": 0.9058,
      "step": 1145
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.3100948275163325e-05,
      "loss": 0.8294,
      "step": 1146
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.3089286250788975e-05,
      "loss": 0.8013,
      "step": 1147
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.307761566101629e-05,
      "loss": 0.8528,
      "step": 1148
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.306593651017469e-05,
      "loss": 0.9322,
      "step": 1149
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.305424880259679e-05,
      "loss": 0.8404,
      "step": 1150
    },
    {
      "epoch": 0.89,
      "eval_loss": 0.8795799016952515,
      "eval_runtime": 14.3398,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 1150
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.304255254261834e-05,
      "loss": 0.8116,
      "step": 1151
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.303084773457831e-05,
      "loss": 0.8656,
      "step": 1152
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.30191343828188e-05,
      "loss": 0.8394,
      "step": 1153
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.3007412491685085e-05,
      "loss": 0.8543,
      "step": 1154
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.299568206552564e-05,
      "loss": 0.8538,
      "step": 1155
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.298394310869207e-05,
      "loss": 0.8461,
      "step": 1156
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.297219562553917e-05,
      "loss": 0.9063,
      "step": 1157
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.296043962042487e-05,
      "loss": 0.8471,
      "step": 1158
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.294867509771029e-05,
      "loss": 0.9213,
      "step": 1159
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.293690206175968e-05,
      "loss": 0.8441,
      "step": 1160
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.2925120516940495e-05,
      "loss": 0.8693,
      "step": 1161
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.29133304676233e-05,
      "loss": 0.8417,
      "step": 1162
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.290153191818183e-05,
      "loss": 0.8573,
      "step": 1163
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.2889724872992984e-05,
      "loss": 0.8977,
      "step": 1164
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.287790933643679e-05,
      "loss": 0.8754,
      "step": 1165
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.2866085312896455e-05,
      "loss": 0.8506,
      "step": 1166
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.2854252806758315e-05,
      "loss": 0.8955,
      "step": 1167
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.284241182241186e-05,
      "loss": 0.8315,
      "step": 1168
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.2830562364249705e-05,
      "loss": 0.9084,
      "step": 1169
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.281870443666764e-05,
      "loss": 0.8364,
      "step": 1170
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.280683804406459e-05,
      "loss": 0.8322,
      "step": 1171
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.2794963190842605e-05,
      "loss": 0.9101,
      "step": 1172
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.278307988140688e-05,
      "loss": 0.839,
      "step": 1173
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.277118812016575e-05,
      "loss": 0.7663,
      "step": 1174
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.275928791153069e-05,
      "loss": 0.9053,
      "step": 1175
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.2747379259916293e-05,
      "loss": 0.8727,
      "step": 1176
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.27354621697403e-05,
      "loss": 0.8697,
      "step": 1177
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.272353664542358e-05,
      "loss": 0.9178,
      "step": 1178
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.271160269139011e-05,
      "loss": 0.8797,
      "step": 1179
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.2699660312067026e-05,
      "loss": 0.8795,
      "step": 1180
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.268770951188458e-05,
      "loss": 0.9538,
      "step": 1181
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.2675750295276124e-05,
      "loss": 0.8571,
      "step": 1182
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.2663782666678165e-05,
      "loss": 0.7752,
      "step": 1183
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.265180663053032e-05,
      "loss": 0.8515,
      "step": 1184
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.263982219127531e-05,
      "loss": 0.9467,
      "step": 1185
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.262782935335898e-05,
      "loss": 0.8506,
      "step": 1186
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.261582812123031e-05,
      "loss": 0.9126,
      "step": 1187
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.260381849934138e-05,
      "loss": 0.8524,
      "step": 1188
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.2591800492147366e-05,
      "loss": 0.9179,
      "step": 1189
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.2579774104106587e-05,
      "loss": 0.8502,
      "step": 1190
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.256773933968044e-05,
      "loss": 0.8448,
      "step": 1191
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.2555696203333455e-05,
      "loss": 0.776,
      "step": 1192
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.254364469953325e-05,
      "loss": 0.8378,
      "step": 1193
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.2531584832750555e-05,
      "loss": 0.8589,
      "step": 1194
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.251951660745921e-05,
      "loss": 0.8958,
      "step": 1195
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.2507440028136125e-05,
      "loss": 0.8813,
      "step": 1196
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.249535509926135e-05,
      "loss": 0.8846,
      "step": 1197
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.248326182531801e-05,
      "loss": 0.86,
      "step": 1198
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.247116021079231e-05,
      "loss": 0.8941,
      "step": 1199
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.245905026017361e-05,
      "loss": 0.881,
      "step": 1200
    },
    {
      "epoch": 0.93,
      "eval_loss": 0.87925785779953,
      "eval_runtime": 14.3586,
      "eval_samples_per_second": 5.711,
      "eval_steps_per_second": 5.711,
      "step": 1200
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.244693197795427e-05,
      "loss": 0.8463,
      "step": 1201
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.243480536862983e-05,
      "loss": 0.8611,
      "step": 1202
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.242267043669886e-05,
      "loss": 0.9424,
      "step": 1203
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.2410527186663046e-05,
      "loss": 0.8505,
      "step": 1204
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.239837562302715e-05,
      "loss": 0.8556,
      "step": 1205
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.238621575029902e-05,
      "loss": 0.9157,
      "step": 1206
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.2374047572989585e-05,
      "loss": 0.8603,
      "step": 1207
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.2361871095612854e-05,
      "loss": 0.8713,
      "step": 1208
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.234968632268592e-05,
      "loss": 0.866,
      "step": 1209
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.2337493258728944e-05,
      "loss": 0.8145,
      "step": 1210
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.2325291908265177e-05,
      "loss": 0.8519,
      "step": 1211
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.2313082275820934e-05,
      "loss": 0.8288,
      "step": 1212
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.230086436592559e-05,
      "loss": 0.8545,
      "step": 1213
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.228863818311163e-05,
      "loss": 0.9122,
      "step": 1214
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.2276403731914564e-05,
      "loss": 0.8539,
      "step": 1215
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.2264161016873e-05,
      "loss": 0.7631,
      "step": 1216
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.225191004252858e-05,
      "loss": 0.8867,
      "step": 1217
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.223965081342606e-05,
      "loss": 0.9395,
      "step": 1218
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.222738333411321e-05,
      "loss": 0.8129,
      "step": 1219
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.221510760914088e-05,
      "loss": 0.823,
      "step": 1220
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.220282364306297e-05,
      "loss": 0.8681,
      "step": 1221
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.2190531440436464e-05,
      "loss": 0.829,
      "step": 1222
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.2178231005821365e-05,
      "loss": 0.837,
      "step": 1223
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.216592234378076e-05,
      "loss": 0.8937,
      "step": 1224
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.2153605458880765e-05,
      "loss": 0.8753,
      "step": 1225
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.214128035569056e-05,
      "loss": 0.8563,
      "step": 1226
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.212894703878237e-05,
      "loss": 0.8824,
      "step": 1227
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.211660551273147e-05,
      "loss": 0.8809,
      "step": 1228
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.210425578211617e-05,
      "loss": 0.9764,
      "step": 1229
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.209189785151784e-05,
      "loss": 0.8432,
      "step": 1230
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.207953172552088e-05,
      "loss": 0.8064,
      "step": 1231
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.2067157408712734e-05,
      "loss": 0.7645,
      "step": 1232
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.205477490568388e-05,
      "loss": 0.891,
      "step": 1233
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.204238422102784e-05,
      "loss": 0.7995,
      "step": 1234
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.2029985359341165e-05,
      "loss": 0.8309,
      "step": 1235
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.201757832522345e-05,
      "loss": 1.0696,
      "step": 1236
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.200516312327731e-05,
      "loss": 0.8111,
      "step": 1237
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.199273975810839e-05,
      "loss": 0.8308,
      "step": 1238
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.198030823432537e-05,
      "loss": 0.8864,
      "step": 1239
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.196786855653996e-05,
      "loss": 0.8606,
      "step": 1240
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.195542072936688e-05,
      "loss": 0.797,
      "step": 1241
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.194296475742389e-05,
      "loss": 0.8634,
      "step": 1242
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.193050064533176e-05,
      "loss": 0.8548,
      "step": 1243
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.191802839771428e-05,
      "loss": 0.8668,
      "step": 1244
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.1905548019198264e-05,
      "loss": 0.8793,
      "step": 1245
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.189305951441353e-05,
      "loss": 0.8059,
      "step": 1246
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.188056288799294e-05,
      "loss": 0.8551,
      "step": 1247
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.186805814457233e-05,
      "loss": 0.8569,
      "step": 1248
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.1855545288790586e-05,
      "loss": 0.8936,
      "step": 1249
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.184302432528956e-05,
      "loss": 0.925,
      "step": 1250
    },
    {
      "epoch": 0.97,
      "eval_loss": 0.8768937587738037,
      "eval_runtime": 14.3545,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 1250
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.1830495258714156e-05,
      "loss": 0.79,
      "step": 1251
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.181795809371225e-05,
      "loss": 0.8115,
      "step": 1252
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.180541283493474e-05,
      "loss": 0.8788,
      "step": 1253
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.179285948703552e-05,
      "loss": 0.8397,
      "step": 1254
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.178029805467149e-05,
      "loss": 0.8189,
      "step": 1255
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.1767728542502545e-05,
      "loss": 0.9749,
      "step": 1256
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.1755150955191586e-05,
      "loss": 0.8964,
      "step": 1257
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.174256529740449e-05,
      "loss": 0.8835,
      "step": 1258
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.172997157381014e-05,
      "loss": 0.7787,
      "step": 1259
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.171736978908042e-05,
      "loss": 0.9683,
      "step": 1260
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.1704759947890185e-05,
      "loss": 0.906,
      "step": 1261
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.16921420549173e-05,
      "loss": 0.8747,
      "step": 1262
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.167951611484259e-05,
      "loss": 0.7947,
      "step": 1263
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.16668821323499e-05,
      "loss": 0.8383,
      "step": 1264
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.1654240112126034e-05,
      "loss": 0.8896,
      "step": 1265
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.164159005886078e-05,
      "loss": 0.8375,
      "step": 1266
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.162893197724692e-05,
      "loss": 0.8575,
      "step": 1267
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.1616265871980175e-05,
      "loss": 0.8488,
      "step": 1268
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.16035917477593e-05,
      "loss": 0.8398,
      "step": 1269
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.159090960928599e-05,
      "loss": 0.8146,
      "step": 1270
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.157821946126491e-05,
      "loss": 0.9755,
      "step": 1271
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.156552130840372e-05,
      "loss": 0.9845,
      "step": 1272
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.1552815155413015e-05,
      "loss": 0.8869,
      "step": 1273
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.154010100700639e-05,
      "loss": 0.8351,
      "step": 1274
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.152737886790041e-05,
      "loss": 0.8805,
      "step": 1275
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.151464874281455e-05,
      "loss": 0.9241,
      "step": 1276
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.150191063647132e-05,
      "loss": 0.8213,
      "step": 1277
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.148916455359613e-05,
      "loss": 0.828,
      "step": 1278
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.14764104989174e-05,
      "loss": 0.8004,
      "step": 1279
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.1463648477166474e-05,
      "loss": 0.7979,
      "step": 1280
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.145087849307765e-05,
      "loss": 0.771,
      "step": 1281
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.14381005513882e-05,
      "loss": 0.9042,
      "step": 1282
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.142531465683834e-05,
      "loss": 0.7922,
      "step": 1283
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.141252081417123e-05,
      "loss": 0.827,
      "step": 1284
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.139971902813297e-05,
      "loss": 0.8541,
      "step": 1285
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.138690930347265e-05,
      "loss": 0.9865,
      "step": 1286
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.137409164494225e-05,
      "loss": 0.885,
      "step": 1287
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.136126605729673e-05,
      "loss": 0.8943,
      "step": 1288
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.134843254529397e-05,
      "loss": 0.8419,
      "step": 1289
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1335591113694806e-05,
      "loss": 0.8346,
      "step": 1290
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1322741767263e-05,
      "loss": 0.8761,
      "step": 1291
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1309884510765256e-05,
      "loss": 0.8206,
      "step": 1292
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.129701934897121e-05,
      "loss": 0.8221,
      "step": 1293
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1284146286653444e-05,
      "loss": 0.8702,
      "step": 1294
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.127126532858745e-05,
      "loss": 0.8543,
      "step": 1295
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1258376479551645e-05,
      "loss": 0.8241,
      "step": 1296
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.12454797443274e-05,
      "loss": 0.8338,
      "step": 1297
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.1232575127699005e-05,
      "loss": 1.1177,
      "step": 1298
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.1219662634453654e-05,
      "loss": 0.8439,
      "step": 1299
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.120674226938148e-05,
      "loss": 0.8789,
      "step": 1300
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.8753229975700378,
      "eval_runtime": 14.3493,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 1300
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.119381403727554e-05,
      "loss": 0.8942,
      "step": 1301
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.118087794293178e-05,
      "loss": 0.9067,
      "step": 1302
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.1167933991149104e-05,
      "loss": 0.8373,
      "step": 1303
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.11549821867293e-05,
      "loss": 0.8963,
      "step": 1304
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.114202253447708e-05,
      "loss": 0.8233,
      "step": 1305
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.112905503920007e-05,
      "loss": 0.8927,
      "step": 1306
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.11160797057088e-05,
      "loss": 0.7887,
      "step": 1307
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.110309653881672e-05,
      "loss": 0.8514,
      "step": 1308
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.109010554334016e-05,
      "loss": 0.8772,
      "step": 1309
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.107710672409838e-05,
      "loss": 0.829,
      "step": 1310
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.106410008591354e-05,
      "loss": 0.8759,
      "step": 1311
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.105108563361068e-05,
      "loss": 1.0493,
      "step": 1312
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.103806337201776e-05,
      "loss": 0.8611,
      "step": 1313
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.102503330596562e-05,
      "loss": 0.8037,
      "step": 1314
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.101199544028801e-05,
      "loss": 0.6513,
      "step": 1315
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.099894977982157e-05,
      "loss": 0.7341,
      "step": 1316
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.0985896329405826e-05,
      "loss": 0.7717,
      "step": 1317
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.0972835093883204e-05,
      "loss": 0.7947,
      "step": 1318
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.0959766078099e-05,
      "loss": 0.7887,
      "step": 1319
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.094668928690141e-05,
      "loss": 0.7444,
      "step": 1320
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0933604725141525e-05,
      "loss": 0.7055,
      "step": 1321
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0920512397673296e-05,
      "loss": 0.6733,
      "step": 1322
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.090741230935355e-05,
      "loss": 0.6929,
      "step": 1323
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0894304465042036e-05,
      "loss": 0.7772,
      "step": 1324
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.088118886960132e-05,
      "loss": 0.7842,
      "step": 1325
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0868065527896905e-05,
      "loss": 0.7613,
      "step": 1326
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0854934444797126e-05,
      "loss": 0.8024,
      "step": 1327
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.084179562517319e-05,
      "loss": 0.7318,
      "step": 1328
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0828649073899204e-05,
      "loss": 0.7364,
      "step": 1329
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.081549479585211e-05,
      "loss": 0.7271,
      "step": 1330
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0802332795911745e-05,
      "loss": 0.8116,
      "step": 1331
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.078916307896079e-05,
      "loss": 0.6577,
      "step": 1332
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.0775985649884784e-05,
      "loss": 0.6727,
      "step": 1333
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0762800513572144e-05,
      "loss": 0.704,
      "step": 1334
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0749607674914164e-05,
      "loss": 0.7942,
      "step": 1335
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.073640713880493e-05,
      "loss": 0.7264,
      "step": 1336
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0723198910141466e-05,
      "loss": 0.769,
      "step": 1337
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0709982993823586e-05,
      "loss": 0.8758,
      "step": 1338
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0696759394753986e-05,
      "loss": 0.706,
      "step": 1339
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.06835281178382e-05,
      "loss": 0.7596,
      "step": 1340
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0670289167984616e-05,
      "loss": 0.6511,
      "step": 1341
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.065704255010448e-05,
      "loss": 0.7497,
      "step": 1342
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.064378826911185e-05,
      "loss": 0.6703,
      "step": 1343
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0630526329923657e-05,
      "loss": 0.7381,
      "step": 1344
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.061725673745967e-05,
      "loss": 0.7463,
      "step": 1345
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.0603979496642476e-05,
      "loss": 0.7884,
      "step": 1346
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.0590694612397516e-05,
      "loss": 0.6742,
      "step": 1347
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.057740208965308e-05,
      "loss": 0.7837,
      "step": 1348
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.056410193334026e-05,
      "loss": 0.7239,
      "step": 1349
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.0550794148393e-05,
      "loss": 0.7165,
      "step": 1350
    },
    {
      "epoch": 1.03,
      "eval_loss": 0.8818119764328003,
      "eval_runtime": 14.3524,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 1350
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.053747873974807e-05,
      "loss": 0.7394,
      "step": 1351
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.052415571234505e-05,
      "loss": 0.7359,
      "step": 1352
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.05108250711264e-05,
      "loss": 0.7452,
      "step": 1353
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.049748682103734e-05,
      "loss": 0.7395,
      "step": 1354
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.048414096702594e-05,
      "loss": 0.7465,
      "step": 1355
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.0470787514043116e-05,
      "loss": 0.6899,
      "step": 1356
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.0457426467042556e-05,
      "loss": 0.7488,
      "step": 1357
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.0444057830980795e-05,
      "loss": 0.7735,
      "step": 1358
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.043068161081717e-05,
      "loss": 0.7023,
      "step": 1359
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.041729781151385e-05,
      "loss": 0.6328,
      "step": 1360
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.040390643803579e-05,
      "loss": 0.6947,
      "step": 1361
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.039050749535079e-05,
      "loss": 0.7527,
      "step": 1362
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.0377100988429407e-05,
      "loss": 0.7863,
      "step": 1363
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.036368692224505e-05,
      "loss": 0.7439,
      "step": 1364
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.035026530177391e-05,
      "loss": 0.7687,
      "step": 1365
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.033683613199499e-05,
      "loss": 0.79,
      "step": 1366
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.032339941789009e-05,
      "loss": 0.7709,
      "step": 1367
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.03099551644438e-05,
      "loss": 0.7457,
      "step": 1368
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.029650337664352e-05,
      "loss": 0.8041,
      "step": 1369
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.028304405947944e-05,
      "loss": 0.7605,
      "step": 1370
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.026957721794453e-05,
      "loss": 0.7704,
      "step": 1371
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.02561028570346e-05,
      "loss": 0.6882,
      "step": 1372
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.024262098174818e-05,
      "loss": 0.7087,
      "step": 1373
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.022913159708663e-05,
      "loss": 0.7001,
      "step": 1374
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.021563470805408e-05,
      "loss": 0.741,
      "step": 1375
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.020213031965748e-05,
      "loss": 0.7177,
      "step": 1376
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.01886184369065e-05,
      "loss": 0.7357,
      "step": 1377
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.017509906481363e-05,
      "loss": 0.693,
      "step": 1378
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.016157220839415e-05,
      "loss": 0.7577,
      "step": 1379
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.014803787266608e-05,
      "loss": 0.743,
      "step": 1380
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.013449606265023e-05,
      "loss": 0.7679,
      "step": 1381
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.01209467833702e-05,
      "loss": 0.7466,
      "step": 1382
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.010739003985234e-05,
      "loss": 0.666,
      "step": 1383
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.009382583712577e-05,
      "loss": 0.8433,
      "step": 1384
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.0080254180222383e-05,
      "loss": 0.7525,
      "step": 1385
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.006667507417684e-05,
      "loss": 0.7486,
      "step": 1386
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.005308852402657e-05,
      "loss": 0.7169,
      "step": 1387
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.003949453481174e-05,
      "loss": 0.6822,
      "step": 1388
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.0025893111575306e-05,
      "loss": 0.7147,
      "step": 1389
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.001228425936295e-05,
      "loss": 0.8178,
      "step": 1390
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.999866798322314e-05,
      "loss": 0.7474,
      "step": 1391
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.998504428820709e-05,
      "loss": 0.7511,
      "step": 1392
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.997141317936876e-05,
      "loss": 0.7417,
      "step": 1393
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.995777466176485e-05,
      "loss": 0.7704,
      "step": 1394
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.9944128740454835e-05,
      "loss": 0.6597,
      "step": 1395
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.993047542050091e-05,
      "loss": 0.788,
      "step": 1396
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.991681470696804e-05,
      "loss": 0.7183,
      "step": 1397
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9903146604923906e-05,
      "loss": 0.7814,
      "step": 1398
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9889471119438955e-05,
      "loss": 0.7289,
      "step": 1399
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.987578825558635e-05,
      "loss": 0.7512,
      "step": 1400
    },
    {
      "epoch": 1.07,
      "eval_loss": 0.8847687244415283,
      "eval_runtime": 14.346,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 1400
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9862098018442014e-05,
      "loss": 0.6941,
      "step": 1401
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9848400413084594e-05,
      "loss": 0.696,
      "step": 1402
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9834695444595456e-05,
      "loss": 0.7361,
      "step": 1403
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.982098311805874e-05,
      "loss": 0.6625,
      "step": 1404
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.980726343856125e-05,
      "loss": 0.7242,
      "step": 1405
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.979353641119259e-05,
      "loss": 0.7827,
      "step": 1406
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.977980204104503e-05,
      "loss": 0.681,
      "step": 1407
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9766060333213595e-05,
      "loss": 0.694,
      "step": 1408
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.975231129279603e-05,
      "loss": 0.6905,
      "step": 1409
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.973855492489281e-05,
      "loss": 0.754,
      "step": 1410
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.972479123460709e-05,
      "loss": 0.7068,
      "step": 1411
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.971102022704478e-05,
      "loss": 0.7554,
      "step": 1412
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.969724190731448e-05,
      "loss": 0.7502,
      "step": 1413
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.968345628052752e-05,
      "loss": 0.7346,
      "step": 1414
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.9669663351797945e-05,
      "loss": 0.7053,
      "step": 1415
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.965586312624247e-05,
      "loss": 0.6501,
      "step": 1416
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.9642055608980565e-05,
      "loss": 0.7445,
      "step": 1417
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.962824080513438e-05,
      "loss": 0.7211,
      "step": 1418
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.961441871982877e-05,
      "loss": 0.7008,
      "step": 1419
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.960058935819129e-05,
      "loss": 0.7268,
      "step": 1420
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.958675272535221e-05,
      "loss": 0.6951,
      "step": 1421
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.957290882644447e-05,
      "loss": 0.7443,
      "step": 1422
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.955905766660373e-05,
      "loss": 0.6985,
      "step": 1423
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.954519925096834e-05,
      "loss": 0.7468,
      "step": 1424
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.953133358467931e-05,
      "loss": 0.6917,
      "step": 1425
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9517460672880404e-05,
      "loss": 0.7756,
      "step": 1426
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9503580520718e-05,
      "loss": 0.7002,
      "step": 1427
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.948969313334122e-05,
      "loss": 0.7434,
      "step": 1428
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.947579851590185e-05,
      "loss": 0.7617,
      "step": 1429
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9461896673554336e-05,
      "loss": 0.7514,
      "step": 1430
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9447987611455835e-05,
      "loss": 0.7324,
      "step": 1431
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9434071334766166e-05,
      "loss": 0.7889,
      "step": 1432
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.942014784864784e-05,
      "loss": 0.7827,
      "step": 1433
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.940621715826602e-05,
      "loss": 0.7607,
      "step": 1434
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9392279268788566e-05,
      "loss": 0.7496,
      "step": 1435
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.9378334185385975e-05,
      "loss": 0.695,
      "step": 1436
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9364381913231465e-05,
      "loss": 0.7978,
      "step": 1437
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.935042245750087e-05,
      "loss": 0.6955,
      "step": 1438
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.933645582337271e-05,
      "loss": 0.6942,
      "step": 1439
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9322482016028166e-05,
      "loss": 0.7097,
      "step": 1440
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.930850104065108e-05,
      "loss": 0.7042,
      "step": 1441
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.929451290242797e-05,
      "loss": 0.7893,
      "step": 1442
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9280517606547964e-05,
      "loss": 0.7096,
      "step": 1443
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9266515158202906e-05,
      "loss": 0.6997,
      "step": 1444
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9252505562587246e-05,
      "loss": 0.791,
      "step": 1445
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.923848882489812e-05,
      "loss": 0.7428,
      "step": 1446
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9224464950335274e-05,
      "loss": 0.7297,
      "step": 1447
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.921043394410114e-05,
      "loss": 0.7527,
      "step": 1448
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.919639581140077e-05,
      "loss": 0.7556,
      "step": 1449
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.918235055744187e-05,
      "loss": 0.7828,
      "step": 1450
    },
    {
      "epoch": 1.11,
      "eval_loss": 0.885636568069458,
      "eval_runtime": 14.3466,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 1450
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.91682981874348e-05,
      "loss": 0.7871,
      "step": 1451
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.915423870659253e-05,
      "loss": 0.6896,
      "step": 1452
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9140172120130696e-05,
      "loss": 0.7367,
      "step": 1453
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9126098433267556e-05,
      "loss": 0.7803,
      "step": 1454
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9112017651224e-05,
      "loss": 0.8131,
      "step": 1455
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9097929779223564e-05,
      "loss": 0.6623,
      "step": 1456
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.90838348224924e-05,
      "loss": 0.6911,
      "step": 1457
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9069732786259285e-05,
      "loss": 0.6775,
      "step": 1458
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.905562367575564e-05,
      "loss": 0.6836,
      "step": 1459
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.904150749621551e-05,
      "loss": 0.7788,
      "step": 1460
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9027384252875534e-05,
      "loss": 0.7467,
      "step": 1461
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9013253950975e-05,
      "loss": 0.7021,
      "step": 1462
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.8999116595755806e-05,
      "loss": 0.7976,
      "step": 1463
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.8984972192462456e-05,
      "loss": 0.8429,
      "step": 1464
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.89708207463421e-05,
      "loss": 0.7204,
      "step": 1465
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.8956662262644454e-05,
      "loss": 0.7229,
      "step": 1466
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.894249674662189e-05,
      "loss": 0.7653,
      "step": 1467
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.892832420352935e-05,
      "loss": 0.7093,
      "step": 1468
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.891414463862442e-05,
      "loss": 0.7237,
      "step": 1469
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.889995805716726e-05,
      "loss": 0.6645,
      "step": 1470
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.888576446442065e-05,
      "loss": 0.7388,
      "step": 1471
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.8871563865649966e-05,
      "loss": 0.8006,
      "step": 1472
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.885735626612318e-05,
      "loss": 0.7611,
      "step": 1473
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.8843141671110864e-05,
      "loss": 0.8196,
      "step": 1474
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.8828920085886184e-05,
      "loss": 0.7549,
      "step": 1475
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.88146915157249e-05,
      "loss": 0.7273,
      "step": 1476
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.880045596590538e-05,
      "loss": 0.6948,
      "step": 1477
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8786213441708545e-05,
      "loss": 0.7454,
      "step": 1478
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.877196394841793e-05,
      "loss": 0.7132,
      "step": 1479
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8757707491319647e-05,
      "loss": 0.7683,
      "step": 1480
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8743444075702396e-05,
      "loss": 0.6431,
      "step": 1481
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.872917370685745e-05,
      "loss": 0.7342,
      "step": 1482
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8714896390078675e-05,
      "loss": 0.7534,
      "step": 1483
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.87006121306625e-05,
      "loss": 0.7165,
      "step": 1484
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8686320933907936e-05,
      "loss": 0.7064,
      "step": 1485
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8672022805116564e-05,
      "loss": 0.7236,
      "step": 1486
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8657717749592554e-05,
      "loss": 0.7311,
      "step": 1487
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.8643405772642614e-05,
      "loss": 0.7372,
      "step": 1488
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.8629086879576034e-05,
      "loss": 0.6571,
      "step": 1489
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.8614761075704695e-05,
      "loss": 0.7603,
      "step": 1490
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.8600428366343e-05,
      "loss": 0.7019,
      "step": 1491
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.858608875680794e-05,
      "loss": 0.7244,
      "step": 1492
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.857174225241905e-05,
      "loss": 0.6965,
      "step": 1493
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.855738885849846e-05,
      "loss": 0.7637,
      "step": 1494
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.854302858037079e-05,
      "loss": 0.7041,
      "step": 1495
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.8528661423363264e-05,
      "loss": 0.7087,
      "step": 1496
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.851428739280566e-05,
      "loss": 0.7082,
      "step": 1497
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.8499906494030275e-05,
      "loss": 0.7166,
      "step": 1498
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.848551873237198e-05,
      "loss": 0.7699,
      "step": 1499
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.8471124113168165e-05,
      "loss": 0.7222,
      "step": 1500
    },
    {
      "epoch": 1.14,
      "eval_loss": 0.88495934009552,
      "eval_runtime": 14.3402,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 1500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.845672264175881e-05,
      "loss": 0.737,
      "step": 1501
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.844231432348637e-05,
      "loss": 0.7304,
      "step": 1502
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.84278991636959e-05,
      "loss": 0.772,
      "step": 1503
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.841347716773498e-05,
      "loss": 0.7159,
      "step": 1504
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.8399048340953685e-05,
      "loss": 0.7205,
      "step": 1505
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.838461268870469e-05,
      "loss": 0.7608,
      "step": 1506
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.837017021634314e-05,
      "loss": 0.7214,
      "step": 1507
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.835572092922676e-05,
      "loss": 0.7385,
      "step": 1508
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.834126483271576e-05,
      "loss": 0.7403,
      "step": 1509
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.8326801932172913e-05,
      "loss": 0.7305,
      "step": 1510
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.8312332232963486e-05,
      "loss": 0.7631,
      "step": 1511
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.829785574045529e-05,
      "loss": 0.7355,
      "step": 1512
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.8283372460018656e-05,
      "loss": 0.7653,
      "step": 1513
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.826888239702641e-05,
      "loss": 0.7166,
      "step": 1514
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.8254385556853916e-05,
      "loss": 0.7706,
      "step": 1515
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.823988194487904e-05,
      "loss": 0.7545,
      "step": 1516
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.822537156648218e-05,
      "loss": 0.7319,
      "step": 1517
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.8210854427046206e-05,
      "loss": 0.7191,
      "step": 1518
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.8196330531956545e-05,
      "loss": 0.8004,
      "step": 1519
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.818179988660109e-05,
      "loss": 0.7634,
      "step": 1520
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.816726249637026e-05,
      "loss": 0.6883,
      "step": 1521
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.815271836665696e-05,
      "loss": 0.7044,
      "step": 1522
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.8138167502856616e-05,
      "loss": 0.7565,
      "step": 1523
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.8123609910367124e-05,
      "loss": 0.7528,
      "step": 1524
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.810904559458892e-05,
      "loss": 0.7931,
      "step": 1525
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.809447456092487e-05,
      "loss": 0.7617,
      "step": 1526
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.80798968147804e-05,
      "loss": 0.7203,
      "step": 1527
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.806531236156338e-05,
      "loss": 0.7867,
      "step": 1528
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.8050721206684186e-05,
      "loss": 0.7332,
      "step": 1529
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.8036123355555675e-05,
      "loss": 0.7947,
      "step": 1530
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.8021518813593195e-05,
      "loss": 0.7698,
      "step": 1531
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.8006907586214564e-05,
      "loss": 0.8222,
      "step": 1532
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.79922896788401e-05,
      "loss": 0.7849,
      "step": 1533
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.797766509689257e-05,
      "loss": 0.6832,
      "step": 1534
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.7963033845797246e-05,
      "loss": 0.696,
      "step": 1535
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.7948395930981864e-05,
      "loss": 0.7331,
      "step": 1536
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.7933751357876615e-05,
      "loss": 0.7088,
      "step": 1537
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.791910013191419e-05,
      "loss": 0.7354,
      "step": 1538
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.790444225852971e-05,
      "loss": 0.7325,
      "step": 1539
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.7889777743160816e-05,
      "loss": 0.7238,
      "step": 1540
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.787510659124756e-05,
      "loss": 0.7859,
      "step": 1541
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.786042880823247e-05,
      "loss": 0.7204,
      "step": 1542
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.7845744399560575e-05,
      "loss": 0.7611,
      "step": 1543
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.7831053370679305e-05,
      "loss": 0.6948,
      "step": 1544
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.7816355727038555e-05,
      "loss": 0.718,
      "step": 1545
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.7801651474090716e-05,
      "loss": 0.7753,
      "step": 1546
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.7786940617290606e-05,
      "loss": 0.7239,
      "step": 1547
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.777222316209546e-05,
      "loss": 0.802,
      "step": 1548
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.7757499113965005e-05,
      "loss": 0.761,
      "step": 1549
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.77427684783614e-05,
      "loss": 0.6756,
      "step": 1550
    },
    {
      "epoch": 1.18,
      "eval_loss": 0.8857229948043823,
      "eval_runtime": 14.3477,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 1550
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.772803126074925e-05,
      "loss": 0.7224,
      "step": 1551
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.77132874665956e-05,
      "loss": 0.7972,
      "step": 1552
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.769853710136993e-05,
      "loss": 0.7825,
      "step": 1553
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.768378017054416e-05,
      "loss": 0.7337,
      "step": 1554
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.7669016679592644e-05,
      "loss": 0.7354,
      "step": 1555
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.765424663399218e-05,
      "loss": 0.7571,
      "step": 1556
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.763947003922199e-05,
      "loss": 0.7014,
      "step": 1557
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.7624686900763716e-05,
      "loss": 0.718,
      "step": 1558
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.760989722410144e-05,
      "loss": 0.716,
      "step": 1559
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.759510101472168e-05,
      "loss": 0.7214,
      "step": 1560
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.758029827811334e-05,
      "loss": 0.7252,
      "step": 1561
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.756548901976779e-05,
      "loss": 0.8031,
      "step": 1562
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.7550673245178794e-05,
      "loss": 0.7485,
      "step": 1563
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.753585095984253e-05,
      "loss": 0.6565,
      "step": 1564
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.7521022169257615e-05,
      "loss": 0.7743,
      "step": 1565
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7506186878925036e-05,
      "loss": 0.7199,
      "step": 1566
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.749134509434824e-05,
      "loss": 0.7236,
      "step": 1567
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7476496821033055e-05,
      "loss": 0.7317,
      "step": 1568
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.746164206448774e-05,
      "loss": 0.6961,
      "step": 1569
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.744678083022291e-05,
      "loss": 0.7468,
      "step": 1570
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7431913123751633e-05,
      "loss": 0.7202,
      "step": 1571
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7417038950589367e-05,
      "loss": 0.7342,
      "step": 1572
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.740215831625395e-05,
      "loss": 0.7087,
      "step": 1573
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.738727122626564e-05,
      "loss": 0.7099,
      "step": 1574
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7372377686147066e-05,
      "loss": 0.7821,
      "step": 1575
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.735747770142327e-05,
      "loss": 0.7052,
      "step": 1576
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.734257127762167e-05,
      "loss": 0.7284,
      "step": 1577
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7327658420272096e-05,
      "loss": 0.6882,
      "step": 1578
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.731273913490673e-05,
      "loss": 0.7918,
      "step": 1579
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.729781342706017e-05,
      "loss": 0.737,
      "step": 1580
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.728288130226938e-05,
      "loss": 0.8181,
      "step": 1581
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.72679427660737e-05,
      "loss": 0.6916,
      "step": 1582
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.7252997824014876e-05,
      "loss": 0.7605,
      "step": 1583
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.7238046481636984e-05,
      "loss": 0.7349,
      "step": 1584
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.722308874448653e-05,
      "loss": 0.784,
      "step": 1585
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.720812461811234e-05,
      "loss": 0.7272,
      "step": 1586
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.719315410806565e-05,
      "loss": 0.7939,
      "step": 1587
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.7178177219900045e-05,
      "loss": 0.7328,
      "step": 1588
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.7163193959171467e-05,
      "loss": 0.7205,
      "step": 1589
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.714820433143825e-05,
      "loss": 0.9286,
      "step": 1590
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.713320834226106e-05,
      "loss": 0.6858,
      "step": 1591
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.7118205997202943e-05,
      "loss": 0.7277,
      "step": 1592
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.7103197301829306e-05,
      "loss": 0.753,
      "step": 1593
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.708818226170788e-05,
      "loss": 0.6704,
      "step": 1594
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.70731608824088e-05,
      "loss": 0.7733,
      "step": 1595
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.705813316950451e-05,
      "loss": 0.6913,
      "step": 1596
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.704309912856982e-05,
      "loss": 0.8382,
      "step": 1597
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.702805876518188e-05,
      "loss": 0.7542,
      "step": 1598
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.7013012084920194e-05,
      "loss": 0.7539,
      "step": 1599
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.699795909336661e-05,
      "loss": 0.7387,
      "step": 1600
    },
    {
      "epoch": 1.22,
      "eval_loss": 0.8828137516975403,
      "eval_runtime": 14.3544,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 1600
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.6982899796105326e-05,
      "loss": 0.7901,
      "step": 1601
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.6967834198722834e-05,
      "loss": 0.7365,
      "step": 1602
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.695276230680804e-05,
      "loss": 0.8082,
      "step": 1603
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.69376841259521e-05,
      "loss": 0.7551,
      "step": 1604
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.6922599661748564e-05,
      "loss": 0.8186,
      "step": 1605
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.690750891979331e-05,
      "loss": 0.7537,
      "step": 1606
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.6892411905684485e-05,
      "loss": 0.7469,
      "step": 1607
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.6877308625022645e-05,
      "loss": 0.7054,
      "step": 1608
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.686219908341062e-05,
      "loss": 0.8492,
      "step": 1609
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.684708328645357e-05,
      "loss": 0.7773,
      "step": 1610
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.683196123975898e-05,
      "loss": 0.793,
      "step": 1611
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.681683294893665e-05,
      "loss": 0.6801,
      "step": 1612
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.680169841959871e-05,
      "loss": 0.7246,
      "step": 1613
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.678655765735959e-05,
      "loss": 0.7162,
      "step": 1614
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.677141066783604e-05,
      "loss": 0.8533,
      "step": 1615
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.6756257456647104e-05,
      "loss": 0.7476,
      "step": 1616
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.674109802941415e-05,
      "loss": 0.7158,
      "step": 1617
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.6725932391760856e-05,
      "loss": 0.7867,
      "step": 1618
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.6710760549313196e-05,
      "loss": 0.708,
      "step": 1619
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.669558250769943e-05,
      "loss": 0.73,
      "step": 1620
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.668039827255016e-05,
      "loss": 0.6983,
      "step": 1621
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.666520784949824e-05,
      "loss": 0.6917,
      "step": 1622
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.665001124417884e-05,
      "loss": 0.8406,
      "step": 1623
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.663480846222943e-05,
      "loss": 0.7394,
      "step": 1624
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.6619599509289755e-05,
      "loss": 0.7888,
      "step": 1625
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.660438439100187e-05,
      "loss": 0.7496,
      "step": 1626
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.658916311301008e-05,
      "loss": 0.7271,
      "step": 1627
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.6573935680961025e-05,
      "loss": 0.7143,
      "step": 1628
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.655870210050361e-05,
      "loss": 0.804,
      "step": 1629
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.6543462377288975e-05,
      "loss": 0.7344,
      "step": 1630
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.652821651697061e-05,
      "loss": 0.6774,
      "step": 1631
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6512964525204244e-05,
      "loss": 0.7849,
      "step": 1632
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6497706407647884e-05,
      "loss": 0.6897,
      "step": 1633
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6482442169961805e-05,
      "loss": 0.7676,
      "step": 1634
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.646717181780856e-05,
      "loss": 0.7603,
      "step": 1635
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6451895356852976e-05,
      "loss": 0.7619,
      "step": 1636
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.643661279276214e-05,
      "loss": 0.7411,
      "step": 1637
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6421324131205384e-05,
      "loss": 0.7921,
      "step": 1638
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.640602937785434e-05,
      "loss": 0.7097,
      "step": 1639
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6390728538382875e-05,
      "loss": 0.7223,
      "step": 1640
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.637542161846712e-05,
      "loss": 0.8295,
      "step": 1641
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.6360108623785446e-05,
      "loss": 0.7343,
      "step": 1642
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.634478956001851e-05,
      "loss": 0.8296,
      "step": 1643
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.63294644328492e-05,
      "loss": 0.7694,
      "step": 1644
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.631413324796264e-05,
      "loss": 0.7122,
      "step": 1645
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6298796011046245e-05,
      "loss": 0.7635,
      "step": 1646
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6283452727789614e-05,
      "loss": 0.7729,
      "step": 1647
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.626810340388465e-05,
      "loss": 0.7372,
      "step": 1648
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6252748045025456e-05,
      "loss": 0.6914,
      "step": 1649
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.623738665690837e-05,
      "loss": 0.6987,
      "step": 1650
    },
    {
      "epoch": 1.26,
      "eval_loss": 0.8809465169906616,
      "eval_runtime": 14.349,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 1650
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6222019245232035e-05,
      "loss": 0.7009,
      "step": 1651
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6206645815697216e-05,
      "loss": 0.7336,
      "step": 1652
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.619126637400702e-05,
      "loss": 0.72,
      "step": 1653
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6175880925866695e-05,
      "loss": 0.83,
      "step": 1654
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.616048947698379e-05,
      "loss": 0.7477,
      "step": 1655
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.6145092033068035e-05,
      "loss": 0.7777,
      "step": 1656
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.612968859983139e-05,
      "loss": 0.7489,
      "step": 1657
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.611427918298806e-05,
      "loss": 0.7405,
      "step": 1658
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.609886378825444e-05,
      "loss": 0.7424,
      "step": 1659
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.608344242134916e-05,
      "loss": 0.7945,
      "step": 1660
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.606801508799307e-05,
      "loss": 0.7446,
      "step": 1661
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.605258179390922e-05,
      "loss": 0.7478,
      "step": 1662
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.603714254482286e-05,
      "loss": 0.7911,
      "step": 1663
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.6021697346461485e-05,
      "loss": 0.8057,
      "step": 1664
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.600624620455478e-05,
      "loss": 0.7531,
      "step": 1665
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.5990789124834615e-05,
      "loss": 0.7571,
      "step": 1666
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.597532611303509e-05,
      "loss": 0.8031,
      "step": 1667
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.59598571748925e-05,
      "loss": 0.6981,
      "step": 1668
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.594438231614533e-05,
      "loss": 0.7431,
      "step": 1669
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.5928901542534255e-05,
      "loss": 0.8732,
      "step": 1670
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.5913414859802164e-05,
      "loss": 0.7353,
      "step": 1671
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.589792227369413e-05,
      "loss": 0.7699,
      "step": 1672
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.588242378995742e-05,
      "loss": 0.7053,
      "step": 1673
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.586691941434147e-05,
      "loss": 0.6829,
      "step": 1674
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.585140915259793e-05,
      "loss": 0.7506,
      "step": 1675
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.5835893010480596e-05,
      "loss": 0.7665,
      "step": 1676
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.582037099374549e-05,
      "loss": 0.756,
      "step": 1677
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.5804843108150796e-05,
      "loss": 0.7316,
      "step": 1678
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.5789309359456846e-05,
      "loss": 0.7229,
      "step": 1679
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.577376975342619e-05,
      "loss": 0.7673,
      "step": 1680
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.575822429582353e-05,
      "loss": 0.7197,
      "step": 1681
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.574267299241573e-05,
      "loss": 0.6894,
      "step": 1682
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.5727115848971866e-05,
      "loss": 0.7528,
      "step": 1683
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.57115528712631e-05,
      "loss": 0.6868,
      "step": 1684
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.569598406506285e-05,
      "loss": 0.7193,
      "step": 1685
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.568040943614662e-05,
      "loss": 0.6883,
      "step": 1686
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.566482899029213e-05,
      "loss": 0.6781,
      "step": 1687
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.564924273327923e-05,
      "loss": 0.7749,
      "step": 1688
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.5633650670889925e-05,
      "loss": 0.696,
      "step": 1689
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.5618052808908386e-05,
      "loss": 0.8012,
      "step": 1690
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.560244915312093e-05,
      "loss": 0.6993,
      "step": 1691
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.5586839709316004e-05,
      "loss": 0.724,
      "step": 1692
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.557122448328423e-05,
      "loss": 0.7526,
      "step": 1693
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.5555603480818384e-05,
      "loss": 0.8284,
      "step": 1694
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.553997670771334e-05,
      "loss": 0.7312,
      "step": 1695
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.5524344169766166e-05,
      "loss": 0.712,
      "step": 1696
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.550870587277601e-05,
      "loss": 0.7641,
      "step": 1697
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.54930618225442e-05,
      "loss": 0.735,
      "step": 1698
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.54774120248742e-05,
      "loss": 0.79,
      "step": 1699
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.5461756485571576e-05,
      "loss": 0.7261,
      "step": 1700
    },
    {
      "epoch": 1.3,
      "eval_loss": 0.8800485730171204,
      "eval_runtime": 14.3484,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 1700
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.544609521044405e-05,
      "loss": 0.7465,
      "step": 1701
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.543042820530145e-05,
      "loss": 0.7304,
      "step": 1702
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.541475547595576e-05,
      "loss": 0.7478,
      "step": 1703
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.5399077028221045e-05,
      "loss": 0.7386,
      "step": 1704
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.538339286791354e-05,
      "loss": 0.6747,
      "step": 1705
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.5367703000851564e-05,
      "loss": 0.6497,
      "step": 1706
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.535200743285556e-05,
      "loss": 0.7422,
      "step": 1707
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.5336306169748104e-05,
      "loss": 0.808,
      "step": 1708
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.532059921735384e-05,
      "loss": 0.7441,
      "step": 1709
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.5304886581499586e-05,
      "loss": 0.7086,
      "step": 1710
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.528916826801422e-05,
      "loss": 0.8303,
      "step": 1711
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.5273444282728735e-05,
      "loss": 0.7517,
      "step": 1712
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.525771463147625e-05,
      "loss": 0.8047,
      "step": 1713
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.524197932009196e-05,
      "loss": 0.7621,
      "step": 1714
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.522623835441318e-05,
      "loss": 0.7005,
      "step": 1715
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.5210491740279304e-05,
      "loss": 0.678,
      "step": 1716
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.519473948353183e-05,
      "loss": 0.7293,
      "step": 1717
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.517898159001435e-05,
      "loss": 0.918,
      "step": 1718
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.516321806557255e-05,
      "loss": 0.7949,
      "step": 1719
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.51474489160542e-05,
      "loss": 0.7624,
      "step": 1720
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.5131674147309156e-05,
      "loss": 0.7626,
      "step": 1721
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.5115893765189365e-05,
      "loss": 0.7318,
      "step": 1722
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.510010777554884e-05,
      "loss": 0.7063,
      "step": 1723
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.5084316184243715e-05,
      "loss": 0.8252,
      "step": 1724
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.506851899713213e-05,
      "loss": 0.7273,
      "step": 1725
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.505271622007438e-05,
      "loss": 0.7217,
      "step": 1726
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.503690785893278e-05,
      "loss": 0.7197,
      "step": 1727
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.502109391957175e-05,
      "loss": 0.8053,
      "step": 1728
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.500527440785774e-05,
      "loss": 0.7812,
      "step": 1729
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.498944932965931e-05,
      "loss": 0.7997,
      "step": 1730
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.497361869084706e-05,
      "loss": 0.7701,
      "step": 1731
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.4957782497293656e-05,
      "loss": 0.7056,
      "step": 1732
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.494194075487384e-05,
      "loss": 0.7982,
      "step": 1733
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.4926093469464376e-05,
      "loss": 0.6936,
      "step": 1734
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.491024064694413e-05,
      "loss": 0.7162,
      "step": 1735
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.489438229319398e-05,
      "loss": 0.7375,
      "step": 1736
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.487851841409691e-05,
      "loss": 0.7426,
      "step": 1737
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.4862649015537885e-05,
      "loss": 0.8237,
      "step": 1738
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.4846774103403954e-05,
      "loss": 0.7955,
      "step": 1739
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.483089368358424e-05,
      "loss": 0.6942,
      "step": 1740
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.481500776196984e-05,
      "loss": 0.8526,
      "step": 1741
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.4799116344453964e-05,
      "loss": 0.7007,
      "step": 1742
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.47832194369318e-05,
      "loss": 0.7394,
      "step": 1743
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.476731704530063e-05,
      "loss": 0.6913,
      "step": 1744
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.4751409175459715e-05,
      "loss": 0.7522,
      "step": 1745
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.473549583331039e-05,
      "loss": 0.7878,
      "step": 1746
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.471957702475599e-05,
      "loss": 0.711,
      "step": 1747
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.47036527557019e-05,
      "loss": 0.7426,
      "step": 1748
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.468772303205552e-05,
      "loss": 0.758,
      "step": 1749
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.4671787859726284e-05,
      "loss": 0.768,
      "step": 1750
    },
    {
      "epoch": 1.34,
      "eval_loss": 0.8819646239280701,
      "eval_runtime": 14.358,
      "eval_samples_per_second": 5.711,
      "eval_steps_per_second": 5.711,
      "step": 1750
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.465584724462563e-05,
      "loss": 0.757,
      "step": 1751
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.4639901192667024e-05,
      "loss": 0.7058,
      "step": 1752
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.462394970976596e-05,
      "loss": 0.7538,
      "step": 1753
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.460799280183992e-05,
      "loss": 0.7675,
      "step": 1754
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.4592030474808437e-05,
      "loss": 0.6985,
      "step": 1755
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.457606273459301e-05,
      "loss": 0.7848,
      "step": 1756
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.456008958711717e-05,
      "loss": 0.7108,
      "step": 1757
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.4544111038306465e-05,
      "loss": 0.7294,
      "step": 1758
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.4528127094088425e-05,
      "loss": 0.7471,
      "step": 1759
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.4512137760392595e-05,
      "loss": 0.7586,
      "step": 1760
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.449614304315051e-05,
      "loss": 0.6787,
      "step": 1761
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.44801429482957e-05,
      "loss": 0.7876,
      "step": 1762
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.446413748176371e-05,
      "loss": 0.7457,
      "step": 1763
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.444812664949206e-05,
      "loss": 0.6915,
      "step": 1764
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.443211045742026e-05,
      "loss": 0.7195,
      "step": 1765
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.4416088911489814e-05,
      "loss": 0.7385,
      "step": 1766
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.440006201764422e-05,
      "loss": 0.742,
      "step": 1767
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.4384029781828934e-05,
      "loss": 0.6314,
      "step": 1768
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.436799220999143e-05,
      "loss": 0.7231,
      "step": 1769
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.435194930808113e-05,
      "loss": 0.7251,
      "step": 1770
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.433590108204945e-05,
      "loss": 0.7702,
      "step": 1771
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.431984753784977e-05,
      "loss": 0.686,
      "step": 1772
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.4303788681437465e-05,
      "loss": 0.8027,
      "step": 1773
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.4287724518769854e-05,
      "loss": 0.7391,
      "step": 1774
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.427165505580622e-05,
      "loss": 0.748,
      "step": 1775
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.4255580298507864e-05,
      "loss": 0.8088,
      "step": 1776
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.423950025283799e-05,
      "loss": 0.7445,
      "step": 1777
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.422341492476179e-05,
      "loss": 0.7149,
      "step": 1778
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.420732432024642e-05,
      "loss": 0.6865,
      "step": 1779
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.419122844526099e-05,
      "loss": 0.7929,
      "step": 1780
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.417512730577656e-05,
      "loss": 0.6532,
      "step": 1781
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.415902090776614e-05,
      "loss": 0.8499,
      "step": 1782
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.4142909257204704e-05,
      "loss": 0.7482,
      "step": 1783
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.412679236006917e-05,
      "loss": 0.6911,
      "step": 1784
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.411067022233838e-05,
      "loss": 0.6651,
      "step": 1785
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.409454284999317e-05,
      "loss": 0.8793,
      "step": 1786
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.4078410249016264e-05,
      "loss": 0.7101,
      "step": 1787
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.4062272425392364e-05,
      "loss": 0.6418,
      "step": 1788
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.4046129385108074e-05,
      "loss": 0.7553,
      "step": 1789
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.402998113415198e-05,
      "loss": 0.7615,
      "step": 1790
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.4013827678514546e-05,
      "loss": 0.7921,
      "step": 1791
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.399766902418823e-05,
      "loss": 0.7226,
      "step": 1792
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.398150517716735e-05,
      "loss": 0.7441,
      "step": 1793
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.396533614344821e-05,
      "loss": 0.7324,
      "step": 1794
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.394916192902899e-05,
      "loss": 0.7043,
      "step": 1795
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.393298253990984e-05,
      "loss": 0.798,
      "step": 1796
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.391679798209278e-05,
      "loss": 0.7324,
      "step": 1797
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.3900608261581785e-05,
      "loss": 0.7754,
      "step": 1798
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.388441338438274e-05,
      "loss": 0.7147,
      "step": 1799
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.3868213356503405e-05,
      "loss": 0.6835,
      "step": 1800
    },
    {
      "epoch": 1.38,
      "eval_loss": 0.8801754713058472,
      "eval_runtime": 14.3453,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 1800
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.385200818395351e-05,
      "loss": 0.659,
      "step": 1801
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.383579787274464e-05,
      "loss": 0.7386,
      "step": 1802
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.3819582428890336e-05,
      "loss": 0.8155,
      "step": 1803
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.3803361858406e-05,
      "loss": 0.7306,
      "step": 1804
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.3787136167308954e-05,
      "loss": 0.7055,
      "step": 1805
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.377090536161843e-05,
      "loss": 0.7933,
      "step": 1806
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.375466944735552e-05,
      "loss": 0.7755,
      "step": 1807
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.373842843054327e-05,
      "loss": 0.715,
      "step": 1808
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.372218231720656e-05,
      "loss": 0.7815,
      "step": 1809
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.370593111337221e-05,
      "loss": 0.7794,
      "step": 1810
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.368967482506888e-05,
      "loss": 0.7055,
      "step": 1811
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.367341345832716e-05,
      "loss": 0.7412,
      "step": 1812
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.365714701917949e-05,
      "loss": 0.7351,
      "step": 1813
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.364087551366022e-05,
      "loss": 0.7823,
      "step": 1814
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.362459894780556e-05,
      "loss": 0.7907,
      "step": 1815
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.36083173276536e-05,
      "loss": 0.7397,
      "step": 1816
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.359203065924433e-05,
      "loss": 0.7556,
      "step": 1817
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.357573894861956e-05,
      "loss": 0.72,
      "step": 1818
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.355944220182302e-05,
      "loss": 0.788,
      "step": 1819
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.3543140424900284e-05,
      "loss": 0.7287,
      "step": 1820
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.352683362389881e-05,
      "loss": 0.7075,
      "step": 1821
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.3510521804867885e-05,
      "loss": 0.8168,
      "step": 1822
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.349420497385871e-05,
      "loss": 0.7443,
      "step": 1823
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.347788313692428e-05,
      "loss": 0.7385,
      "step": 1824
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.3461556300119505e-05,
      "loss": 0.6783,
      "step": 1825
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.344522446950114e-05,
      "loss": 0.737,
      "step": 1826
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.342888765112775e-05,
      "loss": 0.8081,
      "step": 1827
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.3412545851059804e-05,
      "loss": 0.7325,
      "step": 1828
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.339619907535957e-05,
      "loss": 0.778,
      "step": 1829
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.337984733009122e-05,
      "loss": 0.7726,
      "step": 1830
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.336349062132072e-05,
      "loss": 0.8172,
      "step": 1831
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.33471289551159e-05,
      "loss": 0.8095,
      "step": 1832
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.33307623375464e-05,
      "loss": 0.6837,
      "step": 1833
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.331439077468374e-05,
      "loss": 0.7845,
      "step": 1834
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.329801427260126e-05,
      "loss": 0.7773,
      "step": 1835
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.328163283737411e-05,
      "loss": 0.7771,
      "step": 1836
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.3265246475079306e-05,
      "loss": 0.7585,
      "step": 1837
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.324885519179566e-05,
      "loss": 0.7593,
      "step": 1838
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.3232458993603815e-05,
      "loss": 0.7254,
      "step": 1839
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.321605788658626e-05,
      "loss": 0.7671,
      "step": 1840
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.319965187682727e-05,
      "loss": 0.643,
      "step": 1841
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.3183240970412986e-05,
      "loss": 0.6654,
      "step": 1842
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.316682517343131e-05,
      "loss": 0.8231,
      "step": 1843
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.3150404491972006e-05,
      "loss": 0.7317,
      "step": 1844
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.313397893212661e-05,
      "loss": 0.6867,
      "step": 1845
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.311754849998852e-05,
      "loss": 0.798,
      "step": 1846
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.3101113201652885e-05,
      "loss": 0.7644,
      "step": 1847
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.308467304321667e-05,
      "loss": 0.7841,
      "step": 1848
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.306822803077869e-05,
      "loss": 0.8223,
      "step": 1849
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.3051778170439495e-05,
      "loss": 0.7596,
      "step": 1850
    },
    {
      "epoch": 1.41,
      "eval_loss": 0.8806047439575195,
      "eval_runtime": 14.3462,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 1850
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.303532346830149e-05,
      "loss": 0.7071,
      "step": 1851
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.301886393046884e-05,
      "loss": 0.7555,
      "step": 1852
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.3002399563047524e-05,
      "loss": 0.7187,
      "step": 1853
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.298593037214528e-05,
      "loss": 0.7665,
      "step": 1854
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.296945636387169e-05,
      "loss": 0.6921,
      "step": 1855
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.2952977544338084e-05,
      "loss": 0.7753,
      "step": 1856
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.293649391965757e-05,
      "loss": 0.7765,
      "step": 1857
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.2920005495945064e-05,
      "loss": 0.7259,
      "step": 1858
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.290351227931726e-05,
      "loss": 0.6778,
      "step": 1859
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.28870142758926e-05,
      "loss": 0.8319,
      "step": 1860
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.287051149179135e-05,
      "loss": 0.7737,
      "step": 1861
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.285400393313551e-05,
      "loss": 0.8024,
      "step": 1862
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.283749160604887e-05,
      "loss": 0.7461,
      "step": 1863
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.282097451665697e-05,
      "loss": 0.7285,
      "step": 1864
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.280445267108715e-05,
      "loss": 0.7428,
      "step": 1865
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.2787926075468476e-05,
      "loss": 0.7819,
      "step": 1866
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.277139473593181e-05,
      "loss": 0.7342,
      "step": 1867
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.275485865860974e-05,
      "loss": 0.7081,
      "step": 1868
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.273831784963666e-05,
      "loss": 0.7559,
      "step": 1869
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.272177231514866e-05,
      "loss": 0.7107,
      "step": 1870
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.2705222061283626e-05,
      "loss": 0.8007,
      "step": 1871
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.2688667094181193e-05,
      "loss": 0.7506,
      "step": 1872
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.26721074199827e-05,
      "loss": 0.7798,
      "step": 1873
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.265554304483129e-05,
      "loss": 0.6875,
      "step": 1874
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.263897397487182e-05,
      "loss": 0.6977,
      "step": 1875
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.262240021625088e-05,
      "loss": 0.8111,
      "step": 1876
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.260582177511684e-05,
      "loss": 0.7052,
      "step": 1877
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.258923865761975e-05,
      "loss": 0.6943,
      "step": 1878
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2572650869911446e-05,
      "loss": 0.8229,
      "step": 1879
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2556058418145474e-05,
      "loss": 0.6964,
      "step": 1880
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2539461308477095e-05,
      "loss": 0.7208,
      "step": 1881
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2522859547063324e-05,
      "loss": 0.7412,
      "step": 1882
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2506253140062896e-05,
      "loss": 0.7003,
      "step": 1883
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2489642093636255e-05,
      "loss": 0.7393,
      "step": 1884
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.24730264139456e-05,
      "loss": 0.6988,
      "step": 1885
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.245640610715479e-05,
      "loss": 0.6991,
      "step": 1886
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.243978117942945e-05,
      "loss": 0.6828,
      "step": 1887
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2423151636936924e-05,
      "loss": 0.7587,
      "step": 1888
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.240651748584622e-05,
      "loss": 0.7437,
      "step": 1889
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.23898787323281e-05,
      "loss": 0.7268,
      "step": 1890
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.237323538255501e-05,
      "loss": 0.7614,
      "step": 1891
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.235658744270113e-05,
      "loss": 0.8012,
      "step": 1892
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.233993491894229e-05,
      "loss": 0.7332,
      "step": 1893
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.2323277817456074e-05,
      "loss": 0.6872,
      "step": 1894
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.230661614442173e-05,
      "loss": 0.753,
      "step": 1895
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.2289949906020224e-05,
      "loss": 0.7763,
      "step": 1896
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.227327910843419e-05,
      "loss": 0.9111,
      "step": 1897
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.2256603757848e-05,
      "loss": 0.6724,
      "step": 1898
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.223992386044766e-05,
      "loss": 0.7276,
      "step": 1899
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.222323942242087e-05,
      "loss": 0.7167,
      "step": 1900
    },
    {
      "epoch": 1.45,
      "eval_loss": 0.880465567111969,
      "eval_runtime": 14.3504,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 1900
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.2206550449957075e-05,
      "loss": 0.6746,
      "step": 1901
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.2189856949247334e-05,
      "loss": 0.7185,
      "step": 1902
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.217315892648441e-05,
      "loss": 0.7052,
      "step": 1903
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.2156456387862744e-05,
      "loss": 0.7014,
      "step": 1904
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.2139749339578456e-05,
      "loss": 0.7096,
      "step": 1905
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.212303778782934e-05,
      "loss": 0.6532,
      "step": 1906
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.210632173881483e-05,
      "loss": 0.6804,
      "step": 1907
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.208960119873609e-05,
      "loss": 0.7285,
      "step": 1908
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.207287617379589e-05,
      "loss": 0.7519,
      "step": 1909
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.20561466701987e-05,
      "loss": 0.6983,
      "step": 1910
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.203941269415062e-05,
      "loss": 0.6491,
      "step": 1911
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.202267425185946e-05,
      "loss": 0.7058,
      "step": 1912
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.200593134953463e-05,
      "loss": 0.6879,
      "step": 1913
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.198918399338722e-05,
      "loss": 0.6981,
      "step": 1914
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.197243218962999e-05,
      "loss": 0.6831,
      "step": 1915
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.1955675944477305e-05,
      "loss": 0.8011,
      "step": 1916
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.193891526414523e-05,
      "loss": 0.7055,
      "step": 1917
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.192215015485144e-05,
      "loss": 0.6869,
      "step": 1918
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.190538062281526e-05,
      "loss": 0.7511,
      "step": 1919
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.188860667425766e-05,
      "loss": 0.7659,
      "step": 1920
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.1871828315401254e-05,
      "loss": 0.7837,
      "step": 1921
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.1855045552470275e-05,
      "loss": 0.6628,
      "step": 1922
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.183825839169061e-05,
      "loss": 0.7228,
      "step": 1923
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.182146683928975e-05,
      "loss": 0.7484,
      "step": 1924
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.180467090149684e-05,
      "loss": 0.727,
      "step": 1925
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.178787058454266e-05,
      "loss": 0.7187,
      "step": 1926
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.177106589465957e-05,
      "loss": 0.7845,
      "step": 1927
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.175425683808161e-05,
      "loss": 0.7089,
      "step": 1928
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.17374434210444e-05,
      "loss": 0.7895,
      "step": 1929
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.172062564978518e-05,
      "loss": 0.7893,
      "step": 1930
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.170380353054282e-05,
      "loss": 0.7338,
      "step": 1931
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.1686977069557796e-05,
      "loss": 0.7355,
      "step": 1932
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.167014627307219e-05,
      "loss": 0.7342,
      "step": 1933
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.165331114732971e-05,
      "loss": 0.783,
      "step": 1934
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.163647169857565e-05,
      "loss": 0.7681,
      "step": 1935
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.161962793305692e-05,
      "loss": 0.7921,
      "step": 1936
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.160277985702202e-05,
      "loss": 0.759,
      "step": 1937
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.158592747672106e-05,
      "loss": 0.7394,
      "step": 1938
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.1569070798405734e-05,
      "loss": 0.7978,
      "step": 1939
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.155220982832935e-05,
      "loss": 0.7239,
      "step": 1940
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.1535344572746795e-05,
      "loss": 0.7764,
      "step": 1941
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.151847503791455e-05,
      "loss": 0.7925,
      "step": 1942
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.150160123009067e-05,
      "loss": 0.7438,
      "step": 1943
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.148472315553482e-05,
      "loss": 0.7996,
      "step": 1944
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.146784082050823e-05,
      "loss": 0.759,
      "step": 1945
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.145095423127371e-05,
      "loss": 0.8259,
      "step": 1946
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.143406339409567e-05,
      "loss": 0.703,
      "step": 1947
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.141716831524004e-05,
      "loss": 0.7543,
      "step": 1948
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.1400269000974406e-05,
      "loss": 0.6806,
      "step": 1949
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.1383365457567854e-05,
      "loss": 0.7199,
      "step": 1950
    },
    {
      "epoch": 1.49,
      "eval_loss": 0.8791970610618591,
      "eval_runtime": 14.3503,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 1950
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.1366457691291086e-05,
      "loss": 0.7911,
      "step": 1951
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.1349545708416335e-05,
      "loss": 0.7181,
      "step": 1952
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.1332629515217416e-05,
      "loss": 0.7496,
      "step": 1953
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.1315709117969714e-05,
      "loss": 0.6741,
      "step": 1954
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.129878452295016e-05,
      "loss": 0.7684,
      "step": 1955
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.128185573643724e-05,
      "loss": 0.746,
      "step": 1956
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.1264922764711e-05,
      "loss": 0.7304,
      "step": 1957
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.124798561405305e-05,
      "loss": 0.7331,
      "step": 1958
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.123104429074653e-05,
      "loss": 0.7801,
      "step": 1959
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.121409880107615e-05,
      "loss": 0.6875,
      "step": 1960
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.119714915132813e-05,
      "loss": 0.8197,
      "step": 1961
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.118019534779029e-05,
      "loss": 0.6903,
      "step": 1962
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.116323739675192e-05,
      "loss": 0.7223,
      "step": 1963
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.1146275304503914e-05,
      "loss": 0.7784,
      "step": 1964
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.1129309077338666e-05,
      "loss": 0.6828,
      "step": 1965
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.1112338721550105e-05,
      "loss": 0.7611,
      "step": 1966
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.109536424343371e-05,
      "loss": 0.7042,
      "step": 1967
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.1078385649286474e-05,
      "loss": 0.7418,
      "step": 1968
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.106140294540691e-05,
      "loss": 0.7404,
      "step": 1969
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.104441613809508e-05,
      "loss": 0.8114,
      "step": 1970
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.1027425233652556e-05,
      "loss": 0.7308,
      "step": 1971
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.101043023838241e-05,
      "loss": 0.7404,
      "step": 1972
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.0993431158589285e-05,
      "loss": 0.6959,
      "step": 1973
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.097642800057927e-05,
      "loss": 0.7533,
      "step": 1974
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.095942077066001e-05,
      "loss": 0.7524,
      "step": 1975
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.094240947514067e-05,
      "loss": 0.6258,
      "step": 1976
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.092539412033188e-05,
      "loss": 0.7516,
      "step": 1977
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.090837471254582e-05,
      "loss": 0.604,
      "step": 1978
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.089135125809614e-05,
      "loss": 0.7791,
      "step": 1979
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.087432376329802e-05,
      "loss": 0.7904,
      "step": 1980
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.0857292234468116e-05,
      "loss": 0.8409,
      "step": 1981
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.084025667792459e-05,
      "loss": 0.7395,
      "step": 1982
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.0823217099987095e-05,
      "loss": 0.7387,
      "step": 1983
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.080617350697678e-05,
      "loss": 0.7727,
      "step": 1984
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.0789125905216285e-05,
      "loss": 0.7293,
      "step": 1985
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.077207430102973e-05,
      "loss": 0.7676,
      "step": 1986
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.075501870074272e-05,
      "loss": 0.7224,
      "step": 1987
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.0737959110682354e-05,
      "loss": 0.7492,
      "step": 1988
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.0720895537177194e-05,
      "loss": 0.6893,
      "step": 1989
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.070382798655731e-05,
      "loss": 0.6783,
      "step": 1990
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.068675646515418e-05,
      "loss": 0.8031,
      "step": 1991
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.066968097930085e-05,
      "loss": 0.7665,
      "step": 1992
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.065260153533176e-05,
      "loss": 0.7154,
      "step": 1993
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.063551813958286e-05,
      "loss": 0.7497,
      "step": 1994
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.061843079839154e-05,
      "loss": 0.6996,
      "step": 1995
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.0601339518096674e-05,
      "loss": 0.7618,
      "step": 1996
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.058424430503859e-05,
      "loss": 0.7502,
      "step": 1997
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.0567145165559064e-05,
      "loss": 0.6962,
      "step": 1998
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.055004210600135e-05,
      "loss": 0.6929,
      "step": 1999
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.0532935132710136e-05,
      "loss": 0.7051,
      "step": 2000
    },
    {
      "epoch": 1.53,
      "eval_loss": 0.8783687353134155,
      "eval_runtime": 14.35,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 2000
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.051582425203158e-05,
      "loss": 0.753,
      "step": 2001
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.049870947031326e-05,
      "loss": 0.6482,
      "step": 2002
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.048159079390425e-05,
      "loss": 0.7264,
      "step": 2003
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.046446822915502e-05,
      "loss": 0.6637,
      "step": 2004
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.044734178241751e-05,
      "loss": 0.7223,
      "step": 2005
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.043021146004508e-05,
      "loss": 0.7129,
      "step": 2006
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.041307726839256e-05,
      "loss": 0.7459,
      "step": 2007
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0395939213816185e-05,
      "loss": 0.7958,
      "step": 2008
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0378797302673636e-05,
      "loss": 0.7825,
      "step": 2009
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0361651541324014e-05,
      "loss": 0.6921,
      "step": 2010
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0344501936127866e-05,
      "loss": 0.7348,
      "step": 2011
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0327348493447155e-05,
      "loss": 0.7457,
      "step": 2012
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0310191219645273e-05,
      "loss": 0.8222,
      "step": 2013
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0293030121087016e-05,
      "loss": 0.774,
      "step": 2014
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.027586520413863e-05,
      "loss": 0.7647,
      "step": 2015
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.025869647516773e-05,
      "loss": 0.7541,
      "step": 2016
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.0241523940543406e-05,
      "loss": 0.76,
      "step": 2017
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.02243476066361e-05,
      "loss": 0.6979,
      "step": 2018
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.020716747981772e-05,
      "loss": 0.8027,
      "step": 2019
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.0189983566461517e-05,
      "loss": 0.7101,
      "step": 2020
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.017279587294221e-05,
      "loss": 0.7482,
      "step": 2021
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.015560440563588e-05,
      "loss": 0.7238,
      "step": 2022
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.013840917092002e-05,
      "loss": 0.757,
      "step": 2023
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.0121210175173534e-05,
      "loss": 0.7086,
      "step": 2024
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.010400742477669e-05,
      "loss": 0.7082,
      "step": 2025
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.0086800926111175e-05,
      "loss": 0.7279,
      "step": 2026
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.0069590685560054e-05,
      "loss": 0.7088,
      "step": 2027
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.00523767095078e-05,
      "loss": 0.6843,
      "step": 2028
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.003515900434024e-05,
      "loss": 0.805,
      "step": 2029
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.0017937576444596e-05,
      "loss": 0.7841,
      "step": 2030
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.0000712432209496e-05,
      "loss": 0.6871,
      "step": 2031
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.9983483578024913e-05,
      "loss": 0.7337,
      "step": 2032
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.996625102028221e-05,
      "loss": 0.6758,
      "step": 2033
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.9949014765374124e-05,
      "loss": 0.7373,
      "step": 2034
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.993177481969477e-05,
      "loss": 0.7768,
      "step": 2035
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.991453118963961e-05,
      "loss": 0.7875,
      "step": 2036
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.98972838816055e-05,
      "loss": 0.6129,
      "step": 2037
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.9880032901990646e-05,
      "loss": 0.683,
      "step": 2038
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.9862778257194616e-05,
      "loss": 0.673,
      "step": 2039
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.9845519953618344e-05,
      "loss": 0.7505,
      "step": 2040
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.982825799766411e-05,
      "loss": 0.6152,
      "step": 2041
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.981099239573557e-05,
      "loss": 0.6636,
      "step": 2042
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.97937231542377e-05,
      "loss": 0.6902,
      "step": 2043
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.977645027957687e-05,
      "loss": 0.7391,
      "step": 2044
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.975917377816074e-05,
      "loss": 0.7081,
      "step": 2045
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.974189365639838e-05,
      "loss": 0.7384,
      "step": 2046
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.972460992070014e-05,
      "loss": 0.7121,
      "step": 2047
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.970732257747777e-05,
      "loss": 0.7628,
      "step": 2048
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.969003163314432e-05,
      "loss": 0.6477,
      "step": 2049
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.967273709411418e-05,
      "loss": 0.7225,
      "step": 2050
    },
    {
      "epoch": 1.57,
      "eval_loss": 0.8779468536376953,
      "eval_runtime": 14.3436,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 2050
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.965543896680309e-05,
      "loss": 0.8159,
      "step": 2051
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.96381372576281e-05,
      "loss": 0.6992,
      "step": 2052
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.962083197300761e-05,
      "loss": 0.7359,
      "step": 2053
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.9603523119361326e-05,
      "loss": 0.7284,
      "step": 2054
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.958621070311029e-05,
      "loss": 0.8001,
      "step": 2055
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.9568894730676875e-05,
      "loss": 0.676,
      "step": 2056
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.955157520848475e-05,
      "loss": 0.7533,
      "step": 2057
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.953425214295891e-05,
      "loss": 0.707,
      "step": 2058
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.951692554052567e-05,
      "loss": 0.7499,
      "step": 2059
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.949959540761266e-05,
      "loss": 0.784,
      "step": 2060
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.9482261750648804e-05,
      "loss": 0.7383,
      "step": 2061
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.946492457606436e-05,
      "loss": 0.7745,
      "step": 2062
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.944758389029085e-05,
      "loss": 0.7593,
      "step": 2063
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.943023969976114e-05,
      "loss": 0.8077,
      "step": 2064
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.941289201090938e-05,
      "loss": 0.7238,
      "step": 2065
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.9395540830170997e-05,
      "loss": 0.76,
      "step": 2066
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.937818616398276e-05,
      "loss": 0.7098,
      "step": 2067
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.936082801878268e-05,
      "loss": 0.7177,
      "step": 2068
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.93434664010101e-05,
      "loss": 0.7975,
      "step": 2069
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.932610131710562e-05,
      "loss": 0.7274,
      "step": 2070
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.930873277351115e-05,
      "loss": 0.7774,
      "step": 2071
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.929136077666987e-05,
      "loss": 0.6873,
      "step": 2072
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.927398533302623e-05,
      "loss": 0.6809,
      "step": 2073
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.9256606449026e-05,
      "loss": 0.7809,
      "step": 2074
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.923922413111617e-05,
      "loss": 0.7155,
      "step": 2075
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.9221838385745054e-05,
      "loss": 0.6659,
      "step": 2076
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.92044492193622e-05,
      "loss": 0.7595,
      "step": 2077
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.9187056638418466e-05,
      "loss": 0.7589,
      "step": 2078
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.916966064936592e-05,
      "loss": 0.7004,
      "step": 2079
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.915226125865795e-05,
      "loss": 0.6819,
      "step": 2080
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.9134858472749175e-05,
      "loss": 0.745,
      "step": 2081
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.911745229809548e-05,
      "loss": 0.7628,
      "step": 2082
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.9100042741154e-05,
      "loss": 0.7381,
      "step": 2083
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.908262980838315e-05,
      "loss": 0.7889,
      "step": 2084
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.9065213506242565e-05,
      "loss": 0.7337,
      "step": 2085
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.904779384119316e-05,
      "loss": 0.792,
      "step": 2086
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.903037081969706e-05,
      "loss": 0.7448,
      "step": 2087
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.9012944448217676e-05,
      "loss": 0.727,
      "step": 2088
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.899551473321963e-05,
      "loss": 0.7817,
      "step": 2089
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.897808168116881e-05,
      "loss": 0.6766,
      "step": 2090
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.896064529853233e-05,
      "loss": 0.6844,
      "step": 2091
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.894320559177853e-05,
      "loss": 0.7576,
      "step": 2092
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.8925762567376985e-05,
      "loss": 0.7731,
      "step": 2093
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.890831623179854e-05,
      "loss": 0.6669,
      "step": 2094
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.88908665915152e-05,
      "loss": 0.7537,
      "step": 2095
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.887341365300025e-05,
      "loss": 0.705,
      "step": 2096
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8855957422728175e-05,
      "loss": 0.7689,
      "step": 2097
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.883849790717469e-05,
      "loss": 0.6996,
      "step": 2098
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.882103511281673e-05,
      "loss": 0.7233,
      "step": 2099
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8803569046132434e-05,
      "loss": 0.8217,
      "step": 2100
    },
    {
      "epoch": 1.61,
      "eval_loss": 0.8748719096183777,
      "eval_runtime": 14.342,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 2100
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.878609971360117e-05,
      "loss": 0.7065,
      "step": 2101
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.87686271217035e-05,
      "loss": 0.8111,
      "step": 2102
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8751151276921215e-05,
      "loss": 0.6709,
      "step": 2103
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.87336721857373e-05,
      "loss": 0.6551,
      "step": 2104
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8716189854635946e-05,
      "loss": 0.6731,
      "step": 2105
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.869870429010255e-05,
      "loss": 0.8014,
      "step": 2106
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.868121549862371e-05,
      "loss": 0.8961,
      "step": 2107
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.866372348668721e-05,
      "loss": 0.7963,
      "step": 2108
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8646228260782037e-05,
      "loss": 0.7041,
      "step": 2109
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8628729827398376e-05,
      "loss": 0.6732,
      "step": 2110
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.861122819302757e-05,
      "loss": 0.8084,
      "step": 2111
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.859372336416222e-05,
      "loss": 0.733,
      "step": 2112
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8576215347296024e-05,
      "loss": 0.6857,
      "step": 2113
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.855870414892392e-05,
      "loss": 0.802,
      "step": 2114
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8541189775542024e-05,
      "loss": 0.808,
      "step": 2115
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8523672233647596e-05,
      "loss": 0.7284,
      "step": 2116
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.850615152973911e-05,
      "loss": 0.7368,
      "step": 2117
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8488627670316175e-05,
      "loss": 0.7844,
      "step": 2118
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8471100661879614e-05,
      "loss": 0.7998,
      "step": 2119
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.845357051093138e-05,
      "loss": 0.7325,
      "step": 2120
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.843603722397461e-05,
      "loss": 0.772,
      "step": 2121
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8418500807513614e-05,
      "loss": 0.6408,
      "step": 2122
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.840096126805383e-05,
      "loss": 0.7253,
      "step": 2123
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.838341861210189e-05,
      "loss": 0.811,
      "step": 2124
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.836587284616557e-05,
      "loss": 0.6745,
      "step": 2125
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8348323976753776e-05,
      "loss": 0.78,
      "step": 2126
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.83307720103766e-05,
      "loss": 0.6676,
      "step": 2127
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.831321695354528e-05,
      "loss": 0.7881,
      "step": 2128
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.829565881277217e-05,
      "loss": 0.7599,
      "step": 2129
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.827809759457079e-05,
      "loss": 0.8214,
      "step": 2130
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.826053330545582e-05,
      "loss": 0.8498,
      "step": 2131
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.824296595194303e-05,
      "loss": 0.7856,
      "step": 2132
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.822539554054938e-05,
      "loss": 0.6694,
      "step": 2133
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8207822077792906e-05,
      "loss": 0.7356,
      "step": 2134
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8190245570192855e-05,
      "loss": 0.7686,
      "step": 2135
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8172666024269515e-05,
      "loss": 0.725,
      "step": 2136
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.815508344654437e-05,
      "loss": 0.8098,
      "step": 2137
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.813749784353999e-05,
      "loss": 0.6692,
      "step": 2138
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.81199092217801e-05,
      "loss": 0.7352,
      "step": 2139
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8102317587789504e-05,
      "loss": 0.8148,
      "step": 2140
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.808472294809414e-05,
      "loss": 0.7402,
      "step": 2141
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.806712530922109e-05,
      "loss": 0.7222,
      "step": 2142
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.80495246776985e-05,
      "loss": 0.7175,
      "step": 2143
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8031921060055674e-05,
      "loss": 0.6854,
      "step": 2144
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.801431446282298e-05,
      "loss": 0.7106,
      "step": 2145
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.799670489253192e-05,
      "loss": 0.7621,
      "step": 2146
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.7979092355715086e-05,
      "loss": 0.7628,
      "step": 2147
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7961476858906174e-05,
      "loss": 0.7768,
      "step": 2148
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7943858408639986e-05,
      "loss": 0.6192,
      "step": 2149
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7926237011452405e-05,
      "loss": 0.8074,
      "step": 2150
    },
    {
      "epoch": 1.65,
      "eval_loss": 0.874337911605835,
      "eval_runtime": 14.3556,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 2150
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7908612673880423e-05,
      "loss": 0.7607,
      "step": 2151
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.78909854024621e-05,
      "loss": 0.7807,
      "step": 2152
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.787335520373662e-05,
      "loss": 0.6967,
      "step": 2153
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.785572208424421e-05,
      "loss": 0.7388,
      "step": 2154
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7838086050526214e-05,
      "loss": 0.6968,
      "step": 2155
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.782044710912504e-05,
      "loss": 0.7528,
      "step": 2156
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7802805266584196e-05,
      "loss": 0.7052,
      "step": 2157
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7785160529448226e-05,
      "loss": 0.8198,
      "step": 2158
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.776751290426279e-05,
      "loss": 0.743,
      "step": 2159
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.7749862397574584e-05,
      "loss": 0.7135,
      "step": 2160
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.773220901593141e-05,
      "loss": 0.737,
      "step": 2161
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7714552765882104e-05,
      "loss": 0.7455,
      "step": 2162
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7696893653976584e-05,
      "loss": 0.7869,
      "step": 2163
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7679231686765824e-05,
      "loss": 0.7369,
      "step": 2164
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.766156687080186e-05,
      "loss": 0.6684,
      "step": 2165
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.764389921263777e-05,
      "loss": 0.7192,
      "step": 2166
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7626228718827727e-05,
      "loss": 0.6806,
      "step": 2167
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.760855539592689e-05,
      "loss": 0.7308,
      "step": 2168
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.759087925049154e-05,
      "loss": 0.7946,
      "step": 2169
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.757320028907895e-05,
      "loss": 0.7007,
      "step": 2170
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7555518518247465e-05,
      "loss": 0.7634,
      "step": 2171
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.753783394455647e-05,
      "loss": 0.7298,
      "step": 2172
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.752014657456636e-05,
      "loss": 0.7925,
      "step": 2173
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.750245641483863e-05,
      "loss": 0.7565,
      "step": 2174
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7484763471935746e-05,
      "loss": 0.7023,
      "step": 2175
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.746706775242125e-05,
      "loss": 0.7649,
      "step": 2176
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7449369262859676e-05,
      "loss": 0.7626,
      "step": 2177
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7431668009816626e-05,
      "loss": 0.7032,
      "step": 2178
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7413963999858694e-05,
      "loss": 0.7374,
      "step": 2179
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7396257239553523e-05,
      "loss": 0.6999,
      "step": 2180
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.737854773546975e-05,
      "loss": 0.7976,
      "step": 2181
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.736083549417706e-05,
      "loss": 0.6926,
      "step": 2182
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.734312052224612e-05,
      "loss": 0.8153,
      "step": 2183
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7325402826248635e-05,
      "loss": 0.6844,
      "step": 2184
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.730768241275732e-05,
      "loss": 0.7263,
      "step": 2185
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.728995928834587e-05,
      "loss": 0.696,
      "step": 2186
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.727223345958902e-05,
      "loss": 0.7529,
      "step": 2187
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.72545049330625e-05,
      "loss": 0.7448,
      "step": 2188
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.723677371534302e-05,
      "loss": 0.7359,
      "step": 2189
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.721903981300832e-05,
      "loss": 0.7707,
      "step": 2190
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.7201303232637105e-05,
      "loss": 0.7122,
      "step": 2191
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.7183563980809095e-05,
      "loss": 0.7209,
      "step": 2192
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.7165822064104994e-05,
      "loss": 0.7218,
      "step": 2193
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.71480774891065e-05,
      "loss": 0.7939,
      "step": 2194
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.713033026239627e-05,
      "loss": 0.7417,
      "step": 2195
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.7112580390557994e-05,
      "loss": 0.7829,
      "step": 2196
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.70948278801763e-05,
      "loss": 0.6892,
      "step": 2197
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.707707273783681e-05,
      "loss": 0.7992,
      "step": 2198
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.705931497012614e-05,
      "loss": 0.7951,
      "step": 2199
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.704155458363184e-05,
      "loss": 0.7397,
      "step": 2200
    },
    {
      "epoch": 1.69,
      "eval_loss": 0.874525249004364,
      "eval_runtime": 14.3508,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 2200
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.702379158494246e-05,
      "loss": 0.6982,
      "step": 2201
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.700602598064752e-05,
      "loss": 0.6808,
      "step": 2202
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.69882577773375e-05,
      "loss": 0.6832,
      "step": 2203
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.697048698160384e-05,
      "loss": 0.7731,
      "step": 2204
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.6952713600038945e-05,
      "loss": 0.7637,
      "step": 2205
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.693493763923619e-05,
      "loss": 0.7465,
      "step": 2206
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.691715910578988e-05,
      "loss": 0.8485,
      "step": 2207
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.6899378006295306e-05,
      "loss": 0.7475,
      "step": 2208
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.688159434734869e-05,
      "loss": 0.7456,
      "step": 2209
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.6863808135547206e-05,
      "loss": 0.7612,
      "step": 2210
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.684601937748898e-05,
      "loss": 0.7709,
      "step": 2211
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.682822807977309e-05,
      "loss": 0.7485,
      "step": 2212
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.681043424899953e-05,
      "loss": 0.715,
      "step": 2213
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.679263789176928e-05,
      "loss": 0.7119,
      "step": 2214
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.677483901468421e-05,
      "loss": 0.8468,
      "step": 2215
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.675703762434713e-05,
      "loss": 0.6934,
      "step": 2216
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.6739233727361814e-05,
      "loss": 0.7069,
      "step": 2217
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.672142733033294e-05,
      "loss": 0.7676,
      "step": 2218
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.670361843986613e-05,
      "loss": 0.7247,
      "step": 2219
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.6685807062567904e-05,
      "loss": 0.8125,
      "step": 2220
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.666799320504574e-05,
      "loss": 0.7855,
      "step": 2221
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.665017687390801e-05,
      "loss": 0.693,
      "step": 2222
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.663235807576402e-05,
      "loss": 0.7568,
      "step": 2223
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.661453681722397e-05,
      "loss": 0.7067,
      "step": 2224
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.659671310489899e-05,
      "loss": 0.6745,
      "step": 2225
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.657888694540112e-05,
      "loss": 0.6845,
      "step": 2226
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.65610583453433e-05,
      "loss": 0.8272,
      "step": 2227
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.6543227311339394e-05,
      "loss": 0.7136,
      "step": 2228
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.6525393850004124e-05,
      "loss": 0.7754,
      "step": 2229
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.6507557967953174e-05,
      "loss": 0.7365,
      "step": 2230
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.648971967180308e-05,
      "loss": 0.7406,
      "step": 2231
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.647187896817129e-05,
      "loss": 0.6714,
      "step": 2232
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.645403586367613e-05,
      "loss": 0.7315,
      "step": 2233
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.643619036493686e-05,
      "loss": 0.7555,
      "step": 2234
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.641834247857357e-05,
      "loss": 0.7113,
      "step": 2235
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.640049221120729e-05,
      "loss": 0.76,
      "step": 2236
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.638263956945988e-05,
      "loss": 0.7909,
      "step": 2237
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.6364784559954126e-05,
      "loss": 0.7669,
      "step": 2238
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.634692718931367e-05,
      "loss": 0.6865,
      "step": 2239
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.632906746416303e-05,
      "loss": 0.7323,
      "step": 2240
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.631120539112761e-05,
      "loss": 0.7683,
      "step": 2241
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.629334097683367e-05,
      "loss": 0.7143,
      "step": 2242
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.627547422790835e-05,
      "loss": 0.6917,
      "step": 2243
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.625760515097964e-05,
      "loss": 0.6976,
      "step": 2244
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.623973375267642e-05,
      "loss": 0.8428,
      "step": 2245
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.622186003962841e-05,
      "loss": 0.6989,
      "step": 2246
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.620398401846621e-05,
      "loss": 0.7715,
      "step": 2247
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.618610569582122e-05,
      "loss": 0.704,
      "step": 2248
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.616822507832578e-05,
      "loss": 0.7357,
      "step": 2249
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.615034217261301e-05,
      "loss": 0.6683,
      "step": 2250
    },
    {
      "epoch": 1.72,
      "eval_loss": 0.8741868138313293,
      "eval_runtime": 14.3489,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 2250
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.613245698531691e-05,
      "loss": 0.8096,
      "step": 2251
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.6114569523072325e-05,
      "loss": 0.6569,
      "step": 2252
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.6096679792514944e-05,
      "loss": 0.7241,
      "step": 2253
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.6078787800281284e-05,
      "loss": 0.6897,
      "step": 2254
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.6060893553008716e-05,
      "loss": 0.7376,
      "step": 2255
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.604299705733544e-05,
      "loss": 0.6707,
      "step": 2256
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.6025098319900495e-05,
      "loss": 0.6857,
      "step": 2257
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.600719734734376e-05,
      "loss": 0.7581,
      "step": 2258
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.598929414630591e-05,
      "loss": 0.7359,
      "step": 2259
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.5971388723428485e-05,
      "loss": 0.727,
      "step": 2260
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.595348108535382e-05,
      "loss": 0.7271,
      "step": 2261
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.5935571238725105e-05,
      "loss": 0.7773,
      "step": 2262
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.5917659190186304e-05,
      "loss": 0.693,
      "step": 2263
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.589974494638225e-05,
      "loss": 0.9146,
      "step": 2264
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.588182851395855e-05,
      "loss": 0.7574,
      "step": 2265
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5863909899561626e-05,
      "loss": 0.6707,
      "step": 2266
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.584598910983874e-05,
      "loss": 0.6941,
      "step": 2267
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.582806615143792e-05,
      "loss": 0.7662,
      "step": 2268
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5810141031008037e-05,
      "loss": 0.6642,
      "step": 2269
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5792213755198734e-05,
      "loss": 0.7531,
      "step": 2270
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5774284330660484e-05,
      "loss": 0.7816,
      "step": 2271
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.575635276404451e-05,
      "loss": 0.6823,
      "step": 2272
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5738419062002895e-05,
      "loss": 0.8505,
      "step": 2273
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5720483231188446e-05,
      "loss": 0.7108,
      "step": 2274
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.5702545278254805e-05,
      "loss": 0.738,
      "step": 2275
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.56846052098564e-05,
      "loss": 0.7765,
      "step": 2276
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.56666630326484e-05,
      "loss": 0.8621,
      "step": 2277
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.564871875328682e-05,
      "loss": 0.6656,
      "step": 2278
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.56307723784284e-05,
      "loss": 0.6752,
      "step": 2279
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.5612823914730695e-05,
      "loss": 0.6731,
      "step": 2280
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.5594873368852016e-05,
      "loss": 0.7253,
      "step": 2281
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.557692074745144e-05,
      "loss": 0.7647,
      "step": 2282
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.555896605718885e-05,
      "loss": 0.7819,
      "step": 2283
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.5541009304724835e-05,
      "loss": 0.6938,
      "step": 2284
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.552305049672082e-05,
      "loss": 0.6909,
      "step": 2285
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.550508963983893e-05,
      "loss": 0.8039,
      "step": 2286
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.548712674074209e-05,
      "loss": 0.7594,
      "step": 2287
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.546916180609397e-05,
      "loss": 0.7575,
      "step": 2288
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.5451194842558994e-05,
      "loss": 0.7307,
      "step": 2289
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.543322585680235e-05,
      "loss": 0.8042,
      "step": 2290
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.5415254855489935e-05,
      "loss": 0.7168,
      "step": 2291
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.539728184528847e-05,
      "loss": 0.7916,
      "step": 2292
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.537930683286533e-05,
      "loss": 0.7244,
      "step": 2293
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.536132982488871e-05,
      "loss": 0.7403,
      "step": 2294
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.53433508280275e-05,
      "loss": 0.6849,
      "step": 2295
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.532536984895135e-05,
      "loss": 0.6976,
      "step": 2296
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.5307386894330645e-05,
      "loss": 0.8175,
      "step": 2297
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.5289401970836466e-05,
      "loss": 0.749,
      "step": 2298
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.527141508514068e-05,
      "loss": 0.6959,
      "step": 2299
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.525342624391584e-05,
      "loss": 0.7296,
      "step": 2300
    },
    {
      "epoch": 1.76,
      "eval_loss": 0.8722487688064575,
      "eval_runtime": 14.3453,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 2300
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.523543545383525e-05,
      "loss": 0.718,
      "step": 2301
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.521744272157292e-05,
      "loss": 0.7951,
      "step": 2302
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.519944805380359e-05,
      "loss": 0.7069,
      "step": 2303
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.518145145720272e-05,
      "loss": 0.696,
      "step": 2304
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.5163452938446476e-05,
      "loss": 0.7295,
      "step": 2305
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.514545250421174e-05,
      "loss": 0.6443,
      "step": 2306
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.512745016117611e-05,
      "loss": 0.759,
      "step": 2307
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.5109445916017875e-05,
      "loss": 0.7487,
      "step": 2308
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.509143977541607e-05,
      "loss": 0.7059,
      "step": 2309
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.507343174605038e-05,
      "loss": 0.6763,
      "step": 2310
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.505542183460124e-05,
      "loss": 0.7448,
      "step": 2311
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.503741004774974e-05,
      "loss": 0.7249,
      "step": 2312
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.50193963921777e-05,
      "loss": 0.7392,
      "step": 2313
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.500138087456761e-05,
      "loss": 0.6984,
      "step": 2314
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.4983363501602666e-05,
      "loss": 0.7075,
      "step": 2315
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.496534427996673e-05,
      "loss": 0.7537,
      "step": 2316
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.4947323216344405e-05,
      "loss": 0.7529,
      "step": 2317
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.49293003174209e-05,
      "loss": 0.7683,
      "step": 2318
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.491127558988216e-05,
      "loss": 0.7035,
      "step": 2319
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.4893249040414775e-05,
      "loss": 0.7591,
      "step": 2320
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.487522067570605e-05,
      "loss": 0.6909,
      "step": 2321
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.4857190502443934e-05,
      "loss": 0.7193,
      "step": 2322
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.483915852731704e-05,
      "loss": 0.7378,
      "step": 2323
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.482112475701467e-05,
      "loss": 0.8269,
      "step": 2324
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.48030891982268e-05,
      "loss": 0.7279,
      "step": 2325
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.478505185764405e-05,
      "loss": 0.7341,
      "step": 2326
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.476701274195768e-05,
      "loss": 0.6962,
      "step": 2327
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.4748971857859664e-05,
      "loss": 0.7611,
      "step": 2328
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.4730929212042585e-05,
      "loss": 0.8383,
      "step": 2329
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.471288481119971e-05,
      "loss": 0.6253,
      "step": 2330
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.469483866202493e-05,
      "loss": 0.8265,
      "step": 2331
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.4676790771212805e-05,
      "loss": 0.7709,
      "step": 2332
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.4658741145458535e-05,
      "loss": 0.7836,
      "step": 2333
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.4640689791457955e-05,
      "loss": 0.6994,
      "step": 2334
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.4622636715907566e-05,
      "loss": 0.7138,
      "step": 2335
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.4604581925504463e-05,
      "loss": 0.721,
      "step": 2336
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.458652542694644e-05,
      "loss": 0.7797,
      "step": 2337
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.456846722693186e-05,
      "loss": 0.7303,
      "step": 2338
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.455040733215977e-05,
      "loss": 0.688,
      "step": 2339
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.45323457493298e-05,
      "loss": 0.7265,
      "step": 2340
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.451428248514224e-05,
      "loss": 0.7352,
      "step": 2341
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4496217546298e-05,
      "loss": 0.701,
      "step": 2342
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.447815093949859e-05,
      "loss": 0.7293,
      "step": 2343
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.446008267144616e-05,
      "loss": 0.7546,
      "step": 2344
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.444201274884345e-05,
      "loss": 0.7629,
      "step": 2345
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.442394117839386e-05,
      "loss": 0.6579,
      "step": 2346
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4405867966801365e-05,
      "loss": 0.8261,
      "step": 2347
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.438779312077056e-05,
      "loss": 0.8232,
      "step": 2348
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.436971664700664e-05,
      "loss": 0.727,
      "step": 2349
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4351638552215394e-05,
      "loss": 0.7235,
      "step": 2350
    },
    {
      "epoch": 1.8,
      "eval_loss": 0.8719328045845032,
      "eval_runtime": 14.35,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 2350
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.433355884310324e-05,
      "loss": 0.6884,
      "step": 2351
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4315477526377184e-05,
      "loss": 0.7136,
      "step": 2352
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.429739460874483e-05,
      "loss": 0.7712,
      "step": 2353
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.427931009691434e-05,
      "loss": 0.8043,
      "step": 2354
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.426122399759454e-05,
      "loss": 0.6661,
      "step": 2355
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.424313631749478e-05,
      "loss": 0.6867,
      "step": 2356
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.422504706332502e-05,
      "loss": 0.7852,
      "step": 2357
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.420695624179581e-05,
      "loss": 0.7094,
      "step": 2358
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.418886385961827e-05,
      "loss": 0.7098,
      "step": 2359
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4170769923504115e-05,
      "loss": 0.7893,
      "step": 2360
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.41526744401656e-05,
      "loss": 0.7238,
      "step": 2361
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.413457741631562e-05,
      "loss": 0.7307,
      "step": 2362
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4116478858667566e-05,
      "loss": 0.6876,
      "step": 2363
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.409837877393545e-05,
      "loss": 0.7864,
      "step": 2364
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.408027716883383e-05,
      "loss": 0.7425,
      "step": 2365
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4062174050077824e-05,
      "loss": 0.7402,
      "step": 2366
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4044069424383134e-05,
      "loss": 0.797,
      "step": 2367
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4025963298466e-05,
      "loss": 0.7309,
      "step": 2368
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.400785567904322e-05,
      "loss": 0.8338,
      "step": 2369
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.398974657283215e-05,
      "loss": 0.7133,
      "step": 2370
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3971635986550706e-05,
      "loss": 0.6727,
      "step": 2371
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3953523926917345e-05,
      "loss": 0.7403,
      "step": 2372
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.393541040065106e-05,
      "loss": 0.7016,
      "step": 2373
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3917295414471416e-05,
      "loss": 0.683,
      "step": 2374
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3899178975098476e-05,
      "loss": 0.7222,
      "step": 2375
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.38810610892529e-05,
      "loss": 0.6936,
      "step": 2376
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.386294176365584e-05,
      "loss": 0.7931,
      "step": 2377
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.384482100502899e-05,
      "loss": 0.7135,
      "step": 2378
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.382669882009458e-05,
      "loss": 0.6877,
      "step": 2379
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.380857521557539e-05,
      "loss": 0.7479,
      "step": 2380
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.379045019819468e-05,
      "loss": 0.6577,
      "step": 2381
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.377232377467628e-05,
      "loss": 0.6909,
      "step": 2382
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.375419595174451e-05,
      "loss": 0.75,
      "step": 2383
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.373606673612423e-05,
      "loss": 0.6937,
      "step": 2384
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.371793613454081e-05,
      "loss": 0.7231,
      "step": 2385
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3699804153720125e-05,
      "loss": 0.7523,
      "step": 2386
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.368167080038857e-05,
      "loss": 0.6748,
      "step": 2387
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.366353608127304e-05,
      "loss": 0.7125,
      "step": 2388
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3645400003100976e-05,
      "loss": 0.7545,
      "step": 2389
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.362726257260026e-05,
      "loss": 0.7688,
      "step": 2390
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.360912379649932e-05,
      "loss": 0.775,
      "step": 2391
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.359098368152708e-05,
      "loss": 0.7643,
      "step": 2392
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.3572842234412926e-05,
      "loss": 0.7118,
      "step": 2393
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.355469946188679e-05,
      "loss": 0.7742,
      "step": 2394
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.353655537067905e-05,
      "loss": 0.7185,
      "step": 2395
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.351840996752061e-05,
      "loss": 0.7421,
      "step": 2396
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.3500263259142826e-05,
      "loss": 0.7133,
      "step": 2397
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.348211525227758e-05,
      "loss": 0.7382,
      "step": 2398
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.346396595365718e-05,
      "loss": 0.713,
      "step": 2399
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.344581537001447e-05,
      "loss": 0.775,
      "step": 2400
    },
    {
      "epoch": 1.84,
      "eval_loss": 0.8707707524299622,
      "eval_runtime": 14.3493,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 2400
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.342766350808273e-05,
      "loss": 0.8548,
      "step": 2401
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.340951037459573e-05,
      "loss": 0.8524,
      "step": 2402
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.339135597628772e-05,
      "loss": 0.7088,
      "step": 2403
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.3373200319893405e-05,
      "loss": 0.7444,
      "step": 2404
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.335504341214796e-05,
      "loss": 0.7358,
      "step": 2405
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.333688525978704e-05,
      "loss": 0.7417,
      "step": 2406
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.331872586954672e-05,
      "loss": 0.6915,
      "step": 2407
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.330056524816359e-05,
      "loss": 0.7291,
      "step": 2408
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.328240340237466e-05,
      "loss": 0.7125,
      "step": 2409
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.326424033891739e-05,
      "loss": 0.7418,
      "step": 2410
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.324607606452972e-05,
      "loss": 0.82,
      "step": 2411
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.3227910585950025e-05,
      "loss": 0.7193,
      "step": 2412
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.320974390991711e-05,
      "loss": 0.6835,
      "step": 2413
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.3191576043170264e-05,
      "loss": 0.6781,
      "step": 2414
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.317340699244917e-05,
      "loss": 0.811,
      "step": 2415
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.315523676449399e-05,
      "loss": 0.7081,
      "step": 2416
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.313706536604529e-05,
      "loss": 0.7807,
      "step": 2417
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.31188928038441e-05,
      "loss": 0.6861,
      "step": 2418
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.310071908463187e-05,
      "loss": 0.7163,
      "step": 2419
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.308254421515047e-05,
      "loss": 0.682,
      "step": 2420
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.3064368202142204e-05,
      "loss": 0.7581,
      "step": 2421
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.304619105234981e-05,
      "loss": 0.6423,
      "step": 2422
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.3028012772516426e-05,
      "loss": 0.7495,
      "step": 2423
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.3009833369385634e-05,
      "loss": 0.7508,
      "step": 2424
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.299165284970141e-05,
      "loss": 0.7066,
      "step": 2425
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.297347122020817e-05,
      "loss": 0.7708,
      "step": 2426
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.29552884876507e-05,
      "loss": 0.7372,
      "step": 2427
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.293710465877424e-05,
      "loss": 0.9199,
      "step": 2428
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2918919740324414e-05,
      "loss": 0.7787,
      "step": 2429
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.290073373904726e-05,
      "loss": 0.705,
      "step": 2430
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2882546661689204e-05,
      "loss": 0.71,
      "step": 2431
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2864358514997084e-05,
      "loss": 0.7439,
      "step": 2432
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2846169305718124e-05,
      "loss": 0.7932,
      "step": 2433
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.282797904059995e-05,
      "loss": 0.7699,
      "step": 2434
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.280978772639058e-05,
      "loss": 0.8039,
      "step": 2435
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2791595369838416e-05,
      "loss": 0.7152,
      "step": 2436
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.277340197769226e-05,
      "loss": 0.7636,
      "step": 2437
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.275520755670127e-05,
      "loss": 0.7868,
      "step": 2438
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2737012113615024e-05,
      "loss": 0.7617,
      "step": 2439
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2718815655183444e-05,
      "loss": 0.7344,
      "step": 2440
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2700618188156835e-05,
      "loss": 0.7598,
      "step": 2441
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2682419719285914e-05,
      "loss": 0.7146,
      "step": 2442
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.26642202553217e-05,
      "loss": 0.6427,
      "step": 2443
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2646019803015665e-05,
      "loss": 0.7954,
      "step": 2444
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2627818369119566e-05,
      "loss": 0.7503,
      "step": 2445
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.260961596038559e-05,
      "loss": 0.7025,
      "step": 2446
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.259141258356624e-05,
      "loss": 0.746,
      "step": 2447
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2573208245414404e-05,
      "loss": 0.6502,
      "step": 2448
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2555002952683314e-05,
      "loss": 0.6968,
      "step": 2449
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.253679671212656e-05,
      "loss": 0.7331,
      "step": 2450
    },
    {
      "epoch": 1.88,
      "eval_loss": 0.8711597323417664,
      "eval_runtime": 14.3483,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 2450
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.25185895304981e-05,
      "loss": 0.7648,
      "step": 2451
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2500381414552195e-05,
      "loss": 0.7235,
      "step": 2452
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.24821723710435e-05,
      "loss": 0.6853,
      "step": 2453
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2463962406727e-05,
      "loss": 0.6442,
      "step": 2454
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2445751528358024e-05,
      "loss": 0.7097,
      "step": 2455
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.242753974269221e-05,
      "loss": 0.754,
      "step": 2456
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.240932705648558e-05,
      "loss": 0.6783,
      "step": 2457
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.2391113476494466e-05,
      "loss": 0.7127,
      "step": 2458
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.237289900947552e-05,
      "loss": 0.7112,
      "step": 2459
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.235468366218574e-05,
      "loss": 0.8238,
      "step": 2460
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.233646744138245e-05,
      "loss": 0.7006,
      "step": 2461
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.2318250353823296e-05,
      "loss": 0.6848,
      "step": 2462
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.230003240626624e-05,
      "loss": 0.7262,
      "step": 2463
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.2281813605469575e-05,
      "loss": 0.6914,
      "step": 2464
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.226359395819189e-05,
      "loss": 0.7282,
      "step": 2465
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.2245373471192097e-05,
      "loss": 0.7219,
      "step": 2466
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.222715215122945e-05,
      "loss": 0.6659,
      "step": 2467
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.220893000506346e-05,
      "loss": 0.7114,
      "step": 2468
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.219070703945398e-05,
      "loss": 0.7773,
      "step": 2469
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.2172483261161146e-05,
      "loss": 0.726,
      "step": 2470
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.2154258676945414e-05,
      "loss": 0.6556,
      "step": 2471
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.2136033293567534e-05,
      "loss": 0.6932,
      "step": 2472
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.211780711778855e-05,
      "loss": 0.7253,
      "step": 2473
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.20995801563698e-05,
      "loss": 0.7703,
      "step": 2474
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.208135241607289e-05,
      "loss": 0.6738,
      "step": 2475
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.206312390365978e-05,
      "loss": 0.6872,
      "step": 2476
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.2044894625892634e-05,
      "loss": 0.7278,
      "step": 2477
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.202666458953396e-05,
      "loss": 0.7858,
      "step": 2478
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.200843380134653e-05,
      "loss": 0.7241,
      "step": 2479
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1990202268093385e-05,
      "loss": 0.7537,
      "step": 2480
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1971969996537854e-05,
      "loss": 0.8413,
      "step": 2481
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1953736993443524e-05,
      "loss": 0.7354,
      "step": 2482
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1935503265574286e-05,
      "loss": 0.751,
      "step": 2483
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.191726881969426e-05,
      "loss": 0.7645,
      "step": 2484
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.1899033662567865e-05,
      "loss": 0.7077,
      "step": 2485
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.1880797800959755e-05,
      "loss": 0.8488,
      "step": 2486
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.186256124163488e-05,
      "loss": 0.8058,
      "step": 2487
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.1844323991358416e-05,
      "loss": 0.7449,
      "step": 2488
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.182608605689582e-05,
      "loss": 0.7286,
      "step": 2489
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.180784744501277e-05,
      "loss": 0.7205,
      "step": 2490
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.178960816247524e-05,
      "loss": 0.7877,
      "step": 2491
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.177136821604942e-05,
      "loss": 0.7518,
      "step": 2492
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.175312761250174e-05,
      "loss": 0.7734,
      "step": 2493
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.173488635859893e-05,
      "loss": 0.6869,
      "step": 2494
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.171664446110788e-05,
      "loss": 0.7254,
      "step": 2495
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.1698401926795774e-05,
      "loss": 0.7733,
      "step": 2496
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.168015876243003e-05,
      "loss": 0.6565,
      "step": 2497
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.1661914974778276e-05,
      "loss": 0.6776,
      "step": 2498
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.164367057060839e-05,
      "loss": 0.785,
      "step": 2499
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.1625425556688456e-05,
      "loss": 0.7885,
      "step": 2500
    },
    {
      "epoch": 1.92,
      "eval_loss": 0.8706058263778687,
      "eval_runtime": 14.3411,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 2500
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.1607179939786816e-05,
      "loss": 0.6953,
      "step": 2501
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.158893372667201e-05,
      "loss": 0.6852,
      "step": 2502
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.157068692411282e-05,
      "loss": 0.7211,
      "step": 2503
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.155243953887822e-05,
      "loss": 0.7909,
      "step": 2504
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.153419157773743e-05,
      "loss": 0.6928,
      "step": 2505
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.1515943047459864e-05,
      "loss": 0.6628,
      "step": 2506
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.149769395481514e-05,
      "loss": 0.7463,
      "step": 2507
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.1479444306573115e-05,
      "loss": 0.7208,
      "step": 2508
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.146119410950382e-05,
      "loss": 0.7075,
      "step": 2509
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.1442943370377524e-05,
      "loss": 0.7011,
      "step": 2510
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.142469209596465e-05,
      "loss": 0.7123,
      "step": 2511
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.140644029303586e-05,
      "loss": 0.7436,
      "step": 2512
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.1388187968362004e-05,
      "loss": 0.7856,
      "step": 2513
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.1369935128714105e-05,
      "loss": 0.7904,
      "step": 2514
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.135168178086341e-05,
      "loss": 0.7109,
      "step": 2515
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.133342793158132e-05,
      "loss": 0.7581,
      "step": 2516
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.131517358763945e-05,
      "loss": 0.7204,
      "step": 2517
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.129691875580957e-05,
      "loss": 0.6838,
      "step": 2518
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.127866344286367e-05,
      "loss": 0.7611,
      "step": 2519
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.126040765557388e-05,
      "loss": 0.725,
      "step": 2520
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.124215140071254e-05,
      "loss": 0.749,
      "step": 2521
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.1223894685052114e-05,
      "loss": 0.7351,
      "step": 2522
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.1205637515365304e-05,
      "loss": 0.7703,
      "step": 2523
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.118737989842494e-05,
      "loss": 0.7747,
      "step": 2524
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.1169121841003994e-05,
      "loss": 0.7681,
      "step": 2525
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.115086334987566e-05,
      "loss": 0.6811,
      "step": 2526
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.113260443181325e-05,
      "loss": 0.7567,
      "step": 2527
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.1114345093590254e-05,
      "loss": 0.723,
      "step": 2528
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.1096085341980304e-05,
      "loss": 0.7981,
      "step": 2529
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.10778251837572e-05,
      "loss": 0.7345,
      "step": 2530
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.105956462569488e-05,
      "loss": 0.7504,
      "step": 2531
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.104130367456745e-05,
      "loss": 0.7419,
      "step": 2532
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.102304233714913e-05,
      "loss": 0.7536,
      "step": 2533
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.100478062021431e-05,
      "loss": 0.7329,
      "step": 2534
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.0986518530537516e-05,
      "loss": 0.6955,
      "step": 2535
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.096825607489339e-05,
      "loss": 0.7978,
      "step": 2536
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.094999326005676e-05,
      "loss": 0.7556,
      "step": 2537
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.093173009280252e-05,
      "loss": 0.7235,
      "step": 2538
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.091346657990577e-05,
      "loss": 0.7558,
      "step": 2539
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.0895202728141655e-05,
      "loss": 0.6487,
      "step": 2540
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.0876938544285516e-05,
      "loss": 0.7524,
      "step": 2541
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.085867403511278e-05,
      "loss": 0.7866,
      "step": 2542
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.0840409207399014e-05,
      "loss": 0.7008,
      "step": 2543
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.082214406791989e-05,
      "loss": 0.7024,
      "step": 2544
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.080387862345119e-05,
      "loss": 0.7114,
      "step": 2545
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.0785612880768845e-05,
      "loss": 0.8164,
      "step": 2546
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.076734684664884e-05,
      "loss": 0.7579,
      "step": 2547
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.074908052786731e-05,
      "loss": 0.6991,
      "step": 2548
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.07308139312005e-05,
      "loss": 0.7083,
      "step": 2549
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.071254706342471e-05,
      "loss": 0.7514,
      "step": 2550
    },
    {
      "epoch": 1.96,
      "eval_loss": 0.8684533834457397,
      "eval_runtime": 14.3463,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 2550
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.06942799313164e-05,
      "loss": 0.6215,
      "step": 2551
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.067601254165209e-05,
      "loss": 0.7677,
      "step": 2552
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.065774490120842e-05,
      "loss": 0.7675,
      "step": 2553
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.063947701676208e-05,
      "loss": 0.6963,
      "step": 2554
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.062120889508991e-05,
      "loss": 0.7135,
      "step": 2555
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.0602940542968794e-05,
      "loss": 0.6898,
      "step": 2556
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.0584671967175714e-05,
      "loss": 0.7861,
      "step": 2557
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.056640317448774e-05,
      "loss": 0.7293,
      "step": 2558
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.054813417168202e-05,
      "loss": 0.7435,
      "step": 2559
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.0529864965535786e-05,
      "loss": 0.7507,
      "step": 2560
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.051159556282632e-05,
      "loss": 0.8022,
      "step": 2561
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0493325970331007e-05,
      "loss": 0.7034,
      "step": 2562
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0475056194827294e-05,
      "loss": 0.7615,
      "step": 2563
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0456786243092684e-05,
      "loss": 0.7714,
      "step": 2564
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.043851612190476e-05,
      "loss": 0.8021,
      "step": 2565
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0420245838041158e-05,
      "loss": 0.7452,
      "step": 2566
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0401975398279584e-05,
      "loss": 0.7123,
      "step": 2567
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0383704809397782e-05,
      "loss": 0.7361,
      "step": 2568
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.036543407817358e-05,
      "loss": 0.7082,
      "step": 2569
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0347163211384843e-05,
      "loss": 0.8203,
      "step": 2570
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.032889221580948e-05,
      "loss": 0.7166,
      "step": 2571
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0310621098225462e-05,
      "loss": 0.7441,
      "step": 2572
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0292349865410793e-05,
      "loss": 0.7075,
      "step": 2573
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0274078524143532e-05,
      "loss": 0.771,
      "step": 2574
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0255807081201757e-05,
      "loss": 0.7102,
      "step": 2575
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0237535543363618e-05,
      "loss": 0.7595,
      "step": 2576
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.021926391740726e-05,
      "loss": 0.7126,
      "step": 2577
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.02009922101109e-05,
      "loss": 0.6812,
      "step": 2578
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.018272042825275e-05,
      "loss": 0.6923,
      "step": 2579
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.016444857861108e-05,
      "loss": 0.7798,
      "step": 2580
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.014617666796417e-05,
      "loss": 0.6313,
      "step": 2581
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0127904703090304e-05,
      "loss": 0.7093,
      "step": 2582
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0109632690767825e-05,
      "loss": 0.7608,
      "step": 2583
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.009136063777507e-05,
      "loss": 0.6978,
      "step": 2584
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0073088550890394e-05,
      "loss": 0.7979,
      "step": 2585
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.0054816436892167e-05,
      "loss": 0.7086,
      "step": 2586
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.003654430255877e-05,
      "loss": 0.6954,
      "step": 2587
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.001827215466858e-05,
      "loss": 0.7304,
      "step": 2588
    },
    {
      "epoch": 1.99,
      "learning_rate": 3e-05,
      "loss": 0.7189,
      "step": 2589
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.998172784533142e-05,
      "loss": 0.7015,
      "step": 2590
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.9963455697441242e-05,
      "loss": 0.7471,
      "step": 2591
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.994518356310784e-05,
      "loss": 0.6747,
      "step": 2592
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.9926911449109608e-05,
      "loss": 0.7449,
      "step": 2593
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.9908639362224933e-05,
      "loss": 0.7641,
      "step": 2594
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.9890367309232187e-05,
      "loss": 0.714,
      "step": 2595
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.98720952969097e-05,
      "loss": 0.6854,
      "step": 2596
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.985382333203584e-05,
      "loss": 0.682,
      "step": 2597
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.9835551421388932e-05,
      "loss": 0.8492,
      "step": 2598
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.9817279571747252e-05,
      "loss": 0.6973,
      "step": 2599
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9799007789889106e-05,
      "loss": 0.7594,
      "step": 2600
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.8686913251876831,
      "eval_runtime": 14.3406,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 2600
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.978073608259274e-05,
      "loss": 0.6863,
      "step": 2601
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.976246445663639e-05,
      "loss": 0.7559,
      "step": 2602
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9744192918798248e-05,
      "loss": 0.723,
      "step": 2603
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9725921475856476e-05,
      "loss": 0.7271,
      "step": 2604
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.970765013458921e-05,
      "loss": 0.6807,
      "step": 2605
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9689378901774542e-05,
      "loss": 0.7644,
      "step": 2606
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9671107784190526e-05,
      "loss": 0.7413,
      "step": 2607
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.965283678861516e-05,
      "loss": 0.6663,
      "step": 2608
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9634565921826424e-05,
      "loss": 0.7648,
      "step": 2609
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.961629519060222e-05,
      "loss": 0.7261,
      "step": 2610
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9598024601720418e-05,
      "loss": 0.7171,
      "step": 2611
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9579754161958847e-05,
      "loss": 0.691,
      "step": 2612
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9561483878095246e-05,
      "loss": 0.7714,
      "step": 2613
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9543213756907318e-05,
      "loss": 0.6975,
      "step": 2614
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9524943805172708e-05,
      "loss": 0.8116,
      "step": 2615
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9506674029669005e-05,
      "loss": 0.7574,
      "step": 2616
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.948840443717369e-05,
      "loss": 0.6962,
      "step": 2617
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9470135034464222e-05,
      "loss": 0.6662,
      "step": 2618
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9451865828317977e-05,
      "loss": 0.6775,
      "step": 2619
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9433596825512267e-05,
      "loss": 0.7147,
      "step": 2620
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.941532803282429e-05,
      "loss": 0.7323,
      "step": 2621
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9397059457031214e-05,
      "loss": 0.7034,
      "step": 2622
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9378791104910102e-05,
      "loss": 0.7488,
      "step": 2623
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9360522983237923e-05,
      "loss": 0.7982,
      "step": 2624
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.934225509879159e-05,
      "loss": 0.7421,
      "step": 2625
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9323987458347905e-05,
      "loss": 0.733,
      "step": 2626
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9305720068683606e-05,
      "loss": 0.6955,
      "step": 2627
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9287452936575293e-05,
      "loss": 0.7172,
      "step": 2628
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9269186068799505e-05,
      "loss": 0.7588,
      "step": 2629
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9250919472132682e-05,
      "loss": 0.6847,
      "step": 2630
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9232653153351165e-05,
      "loss": 0.5645,
      "step": 2631
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.921438711923116e-05,
      "loss": 0.6111,
      "step": 2632
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9196121376548802e-05,
      "loss": 0.6269,
      "step": 2633
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9177855932080118e-05,
      "loss": 0.6904,
      "step": 2634
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9159590792600988e-05,
      "loss": 0.6205,
      "step": 2635
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9141325964887218e-05,
      "loss": 0.5846,
      "step": 2636
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9123061455714485e-05,
      "loss": 0.5674,
      "step": 2637
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9104797271858356e-05,
      "loss": 0.5638,
      "step": 2638
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9086533420094244e-05,
      "loss": 0.6014,
      "step": 2639
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9068269907197477e-05,
      "loss": 0.6399,
      "step": 2640
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.905000673994325e-05,
      "loss": 0.5834,
      "step": 2641
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9031743925106613e-05,
      "loss": 0.5745,
      "step": 2642
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9013481469462496e-05,
      "loss": 0.6563,
      "step": 2643
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.8995219379785692e-05,
      "loss": 0.651,
      "step": 2644
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.8976957662850877e-05,
      "loss": 0.5702,
      "step": 2645
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.8958696325432556e-05,
      "loss": 0.5643,
      "step": 2646
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.894043537430512e-05,
      "loss": 0.7604,
      "step": 2647
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.892217481624281e-05,
      "loss": 0.5186,
      "step": 2648
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.8903914658019704e-05,
      "loss": 0.6252,
      "step": 2649
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8885654906409755e-05,
      "loss": 0.5374,
      "step": 2650
    },
    {
      "epoch": 2.02,
      "eval_loss": 0.8936876654624939,
      "eval_runtime": 14.3816,
      "eval_samples_per_second": 5.702,
      "eval_steps_per_second": 5.702,
      "step": 2650
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.886739556818675e-05,
      "loss": 0.5928,
      "step": 2651
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.884913665012435e-05,
      "loss": 0.5716,
      "step": 2652
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.883087815899601e-05,
      "loss": 0.563,
      "step": 2653
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.881262010157507e-05,
      "loss": 0.5517,
      "step": 2654
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8794362484634688e-05,
      "loss": 0.5891,
      "step": 2655
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8776105314947884e-05,
      "loss": 0.5421,
      "step": 2656
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8757848599287468e-05,
      "loss": 0.5782,
      "step": 2657
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8739592344426117e-05,
      "loss": 0.6855,
      "step": 2658
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8721336557136342e-05,
      "loss": 0.5735,
      "step": 2659
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.870308124419043e-05,
      "loss": 0.6183,
      "step": 2660
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.8684826412360556e-05,
      "loss": 0.6124,
      "step": 2661
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.866657206841868e-05,
      "loss": 0.6108,
      "step": 2662
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8648318219136603e-05,
      "loss": 0.5706,
      "step": 2663
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8630064871285896e-05,
      "loss": 0.6317,
      "step": 2664
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8611812031637997e-05,
      "loss": 0.5691,
      "step": 2665
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8593559706964148e-05,
      "loss": 0.5968,
      "step": 2666
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8575307904035357e-05,
      "loss": 0.5814,
      "step": 2667
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8557056629622484e-05,
      "loss": 0.6678,
      "step": 2668
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8538805890496175e-05,
      "loss": 0.5998,
      "step": 2669
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8520555693426893e-05,
      "loss": 0.613,
      "step": 2670
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8502306045184864e-05,
      "loss": 0.6022,
      "step": 2671
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.848405695254014e-05,
      "loss": 0.5138,
      "step": 2672
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8465808422262583e-05,
      "loss": 0.5793,
      "step": 2673
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8447560461121784e-05,
      "loss": 0.5879,
      "step": 2674
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.8429313075887184e-05,
      "loss": 0.5664,
      "step": 2675
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8411066273327988e-05,
      "loss": 0.6732,
      "step": 2676
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8392820060213192e-05,
      "loss": 0.5823,
      "step": 2677
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8374574443311552e-05,
      "loss": 0.6178,
      "step": 2678
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.835632942939162e-05,
      "loss": 0.5375,
      "step": 2679
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8338085025221725e-05,
      "loss": 0.5923,
      "step": 2680
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8319841237569972e-05,
      "loss": 0.5758,
      "step": 2681
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.830159807320423e-05,
      "loss": 0.5979,
      "step": 2682
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.828335553889213e-05,
      "loss": 0.6743,
      "step": 2683
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8265113641401087e-05,
      "loss": 0.6088,
      "step": 2684
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.824687238749826e-05,
      "loss": 0.6633,
      "step": 2685
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.8228631783950593e-05,
      "loss": 0.5681,
      "step": 2686
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.821039183752477e-05,
      "loss": 0.6499,
      "step": 2687
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.819215255498724e-05,
      "loss": 0.5744,
      "step": 2688
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.8173913943104194e-05,
      "loss": 0.6159,
      "step": 2689
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.8155676008641586e-05,
      "loss": 0.6013,
      "step": 2690
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.813743875836513e-05,
      "loss": 0.6353,
      "step": 2691
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.811920219904025e-05,
      "loss": 0.5899,
      "step": 2692
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.8100966337432144e-05,
      "loss": 0.6301,
      "step": 2693
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.808273118030574e-05,
      "loss": 0.589,
      "step": 2694
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.8064496734425725e-05,
      "loss": 0.6085,
      "step": 2695
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.804626300655648e-05,
      "loss": 0.5627,
      "step": 2696
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.8028030003462154e-05,
      "loss": 0.5612,
      "step": 2697
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.800979773190663e-05,
      "loss": 0.5324,
      "step": 2698
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.7991566198653477e-05,
      "loss": 0.5823,
      "step": 2699
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.7973335410466038e-05,
      "loss": 0.6379,
      "step": 2700
    },
    {
      "epoch": 2.05,
      "eval_loss": 0.8998726010322571,
      "eval_runtime": 14.3538,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 2700
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.7955105374107364e-05,
      "loss": 0.6455,
      "step": 2701
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7936876096340234e-05,
      "loss": 0.6517,
      "step": 2702
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7918647583927108e-05,
      "loss": 0.5715,
      "step": 2703
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7900419843630206e-05,
      "loss": 0.6363,
      "step": 2704
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7882192882211446e-05,
      "loss": 0.5727,
      "step": 2705
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.786396670643247e-05,
      "loss": 0.6682,
      "step": 2706
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7845741323054587e-05,
      "loss": 0.6217,
      "step": 2707
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.782751673883886e-05,
      "loss": 0.5931,
      "step": 2708
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7809292960546038e-05,
      "loss": 0.5728,
      "step": 2709
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7791069994936552e-05,
      "loss": 0.551,
      "step": 2710
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.777284784877056e-05,
      "loss": 0.5774,
      "step": 2711
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7754626528807898e-05,
      "loss": 0.6234,
      "step": 2712
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.7736406041808124e-05,
      "loss": 0.5676,
      "step": 2713
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7718186394530437e-05,
      "loss": 0.8027,
      "step": 2714
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.769996759373376e-05,
      "loss": 0.5916,
      "step": 2715
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.768174964617671e-05,
      "loss": 0.6363,
      "step": 2716
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7663532558617556e-05,
      "loss": 0.5935,
      "step": 2717
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7645316337814268e-05,
      "loss": 0.6129,
      "step": 2718
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7627100990524484e-05,
      "loss": 0.5754,
      "step": 2719
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7608886523505543e-05,
      "loss": 0.5744,
      "step": 2720
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7590672943514418e-05,
      "loss": 0.5536,
      "step": 2721
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7572460257307794e-05,
      "loss": 0.6329,
      "step": 2722
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7554248471641988e-05,
      "loss": 0.6687,
      "step": 2723
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7536037593273003e-05,
      "loss": 0.6175,
      "step": 2724
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.75178276289565e-05,
      "loss": 0.5727,
      "step": 2725
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.7499618585447807e-05,
      "loss": 0.6494,
      "step": 2726
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7481410469501918e-05,
      "loss": 0.7224,
      "step": 2727
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.746320328787344e-05,
      "loss": 0.5289,
      "step": 2728
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7444997047316688e-05,
      "loss": 0.5391,
      "step": 2729
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7426791754585595e-05,
      "loss": 0.6347,
      "step": 2730
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.740858741643377e-05,
      "loss": 0.6468,
      "step": 2731
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7390384039614414e-05,
      "loss": 0.6058,
      "step": 2732
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.737218163088043e-05,
      "loss": 0.6161,
      "step": 2733
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7353980196984347e-05,
      "loss": 0.6023,
      "step": 2734
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.73357797446783e-05,
      "loss": 0.5994,
      "step": 2735
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7317580280714098e-05,
      "loss": 0.6671,
      "step": 2736
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.729938181184316e-05,
      "loss": 0.5402,
      "step": 2737
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7281184344816567e-05,
      "loss": 0.5937,
      "step": 2738
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.7262987886384984e-05,
      "loss": 0.6044,
      "step": 2739
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7244792443298727e-05,
      "loss": 0.5476,
      "step": 2740
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.722659802230775e-05,
      "loss": 0.6518,
      "step": 2741
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7208404630161585e-05,
      "loss": 0.595,
      "step": 2742
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7190212273609425e-05,
      "loss": 0.5223,
      "step": 2743
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.717202095940005e-05,
      "loss": 0.5877,
      "step": 2744
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7153830694281888e-05,
      "loss": 0.6255,
      "step": 2745
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7135641485002928e-05,
      "loss": 0.6133,
      "step": 2746
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.71174533383108e-05,
      "loss": 0.6193,
      "step": 2747
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7099266260952752e-05,
      "loss": 0.5772,
      "step": 2748
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.708108025967559e-05,
      "loss": 0.5591,
      "step": 2749
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7062895341225764e-05,
      "loss": 0.5491,
      "step": 2750
    },
    {
      "epoch": 2.09,
      "eval_loss": 0.9033382534980774,
      "eval_runtime": 14.3416,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 2750
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.70447115123493e-05,
      "loss": 0.5747,
      "step": 2751
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.7026528779791843e-05,
      "loss": 0.5851,
      "step": 2752
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.7008347150298596e-05,
      "loss": 0.5736,
      "step": 2753
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.699016663061437e-05,
      "loss": 0.6265,
      "step": 2754
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.697198722748357e-05,
      "loss": 0.5627,
      "step": 2755
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6953808947650197e-05,
      "loss": 0.6119,
      "step": 2756
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6935631797857804e-05,
      "loss": 0.6276,
      "step": 2757
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6917455784849536e-05,
      "loss": 0.5264,
      "step": 2758
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.689928091536814e-05,
      "loss": 0.6396,
      "step": 2759
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6881107196155902e-05,
      "loss": 0.5978,
      "step": 2760
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.686293463395471e-05,
      "loss": 0.6071,
      "step": 2761
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6844763235506018e-05,
      "loss": 0.6495,
      "step": 2762
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6826593007550837e-05,
      "loss": 0.6878,
      "step": 2763
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.680842395682974e-05,
      "loss": 0.634,
      "step": 2764
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.6790256090082885e-05,
      "loss": 0.6406,
      "step": 2765
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6772089414049983e-05,
      "loss": 0.6206,
      "step": 2766
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6753923935470283e-05,
      "loss": 0.6578,
      "step": 2767
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.673575966108261e-05,
      "loss": 0.5676,
      "step": 2768
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.671759659762534e-05,
      "loss": 0.6249,
      "step": 2769
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6699434751836415e-05,
      "loss": 0.5571,
      "step": 2770
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.668127413045328e-05,
      "loss": 0.5793,
      "step": 2771
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6663114740212966e-05,
      "loss": 0.6899,
      "step": 2772
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6644956587852046e-05,
      "loss": 0.5512,
      "step": 2773
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6626799680106604e-05,
      "loss": 0.6398,
      "step": 2774
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6608644023712282e-05,
      "loss": 0.5954,
      "step": 2775
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.659048962540427e-05,
      "loss": 0.6063,
      "step": 2776
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6572336491917286e-05,
      "loss": 0.5792,
      "step": 2777
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.6554184629985542e-05,
      "loss": 0.6353,
      "step": 2778
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.653603404634282e-05,
      "loss": 0.5436,
      "step": 2779
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6517884747722422e-05,
      "loss": 0.592,
      "step": 2780
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6499736740857176e-05,
      "loss": 0.6646,
      "step": 2781
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6481590032479394e-05,
      "loss": 0.6369,
      "step": 2782
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6463444629320945e-05,
      "loss": 0.6086,
      "step": 2783
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.644530053811322e-05,
      "loss": 0.6461,
      "step": 2784
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.642715776558708e-05,
      "loss": 0.5938,
      "step": 2785
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6409016318472927e-05,
      "loss": 0.5772,
      "step": 2786
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6390876203500675e-05,
      "loss": 0.58,
      "step": 2787
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6372737427399744e-05,
      "loss": 0.6184,
      "step": 2788
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6354599996899026e-05,
      "loss": 0.6524,
      "step": 2789
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.633646391872695e-05,
      "loss": 0.627,
      "step": 2790
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.6318329199611436e-05,
      "loss": 0.5859,
      "step": 2791
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6300195846279883e-05,
      "loss": 0.6431,
      "step": 2792
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.62820638654592e-05,
      "loss": 0.5897,
      "step": 2793
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6263933263875772e-05,
      "loss": 0.6215,
      "step": 2794
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6245804048255494e-05,
      "loss": 0.5959,
      "step": 2795
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6227676225323724e-05,
      "loss": 0.6082,
      "step": 2796
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6209549801805326e-05,
      "loss": 0.6161,
      "step": 2797
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6191424784424622e-05,
      "loss": 0.6139,
      "step": 2798
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6173301179905422e-05,
      "loss": 0.5439,
      "step": 2799
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6155178994971012e-05,
      "loss": 0.6083,
      "step": 2800
    },
    {
      "epoch": 2.13,
      "eval_loss": 0.9041104912757874,
      "eval_runtime": 14.3545,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 2800
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.613705823634416e-05,
      "loss": 0.6854,
      "step": 2801
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6118938910747107e-05,
      "loss": 0.5262,
      "step": 2802
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6100821024901525e-05,
      "loss": 0.6416,
      "step": 2803
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.6082704585528595e-05,
      "loss": 0.5976,
      "step": 2804
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.6064589599348937e-05,
      "loss": 0.6369,
      "step": 2805
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.604647607308267e-05,
      "loss": 0.5811,
      "step": 2806
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.60283640134493e-05,
      "loss": 0.589,
      "step": 2807
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.6010253427167852e-05,
      "loss": 0.5554,
      "step": 2808
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.5992144320956794e-05,
      "loss": 0.5743,
      "step": 2809
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.597403670153401e-05,
      "loss": 0.5694,
      "step": 2810
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.5955930575616867e-05,
      "loss": 0.5731,
      "step": 2811
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.593782594992217e-05,
      "loss": 0.5578,
      "step": 2812
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.591972283116618e-05,
      "loss": 0.6372,
      "step": 2813
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.5901621226064556e-05,
      "loss": 0.646,
      "step": 2814
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.5883521141332436e-05,
      "loss": 0.5677,
      "step": 2815
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.586542258368439e-05,
      "loss": 0.5946,
      "step": 2816
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.58473255598344e-05,
      "loss": 0.5843,
      "step": 2817
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5829230076495893e-05,
      "loss": 0.5809,
      "step": 2818
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.581113614038173e-05,
      "loss": 0.6055,
      "step": 2819
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5793043758204202e-05,
      "loss": 0.5956,
      "step": 2820
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5774952936674985e-05,
      "loss": 0.6063,
      "step": 2821
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5756863682505223e-05,
      "loss": 0.6525,
      "step": 2822
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.573877600240547e-05,
      "loss": 0.6246,
      "step": 2823
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5720689903085663e-05,
      "loss": 0.684,
      "step": 2824
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.570260539125518e-05,
      "loss": 0.5721,
      "step": 2825
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5684522473622814e-05,
      "loss": 0.5971,
      "step": 2826
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.566644115689676e-05,
      "loss": 0.6161,
      "step": 2827
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5648361447784615e-05,
      "loss": 0.5995,
      "step": 2828
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.5630283352993372e-05,
      "loss": 0.5209,
      "step": 2829
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.561220687922944e-05,
      "loss": 0.5873,
      "step": 2830
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5594132033198633e-05,
      "loss": 0.6365,
      "step": 2831
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5576058821606142e-05,
      "loss": 0.6769,
      "step": 2832
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.555798725115655e-05,
      "loss": 0.561,
      "step": 2833
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5539917328553857e-05,
      "loss": 0.6373,
      "step": 2834
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.552184906050142e-05,
      "loss": 0.5981,
      "step": 2835
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5503782453702006e-05,
      "loss": 0.5912,
      "step": 2836
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5485717514857764e-05,
      "loss": 0.5749,
      "step": 2837
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.546765425067021e-05,
      "loss": 0.5876,
      "step": 2838
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.544959266784024e-05,
      "loss": 0.5807,
      "step": 2839
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.543153277306814e-05,
      "loss": 0.5597,
      "step": 2840
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5413474573053573e-05,
      "loss": 0.6189,
      "step": 2841
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5395418074495538e-05,
      "loss": 0.5448,
      "step": 2842
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.5377363284092446e-05,
      "loss": 0.6513,
      "step": 2843
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5359310208542046e-05,
      "loss": 0.6204,
      "step": 2844
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.534125885454148e-05,
      "loss": 0.5146,
      "step": 2845
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5323209228787204e-05,
      "loss": 0.6063,
      "step": 2846
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5305161337975073e-05,
      "loss": 0.5784,
      "step": 2847
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5287115188800305e-05,
      "loss": 0.6321,
      "step": 2848
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.526907078795742e-05,
      "loss": 0.6497,
      "step": 2849
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.525102814214034e-05,
      "loss": 0.5829,
      "step": 2850
    },
    {
      "epoch": 2.17,
      "eval_loss": 0.905153214931488,
      "eval_runtime": 14.3448,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 2850
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5232987258042318e-05,
      "loss": 0.6258,
      "step": 2851
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5214948142355965e-05,
      "loss": 0.5905,
      "step": 2852
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5196910801773202e-05,
      "loss": 0.5787,
      "step": 2853
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5178875242985325e-05,
      "loss": 0.6242,
      "step": 2854
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.5160841472682962e-05,
      "loss": 0.6317,
      "step": 2855
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.514280949755608e-05,
      "loss": 0.5466,
      "step": 2856
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.5124779324293956e-05,
      "loss": 0.6364,
      "step": 2857
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.5106750959585226e-05,
      "loss": 0.6252,
      "step": 2858
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.508872441011786e-05,
      "loss": 0.6701,
      "step": 2859
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.5070699682579113e-05,
      "loss": 0.6097,
      "step": 2860
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.5052676783655604e-05,
      "loss": 0.6498,
      "step": 2861
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.503465572003326e-05,
      "loss": 0.6243,
      "step": 2862
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.5016636498397346e-05,
      "loss": 0.6384,
      "step": 2863
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.4998619125432396e-05,
      "loss": 0.5734,
      "step": 2864
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.4980603607822306e-05,
      "loss": 0.6025,
      "step": 2865
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.496258995225027e-05,
      "loss": 0.617,
      "step": 2866
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.4944578165398766e-05,
      "loss": 0.6016,
      "step": 2867
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.492656825394962e-05,
      "loss": 0.6734,
      "step": 2868
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.4908560224583936e-05,
      "loss": 0.6572,
      "step": 2869
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4890554083982127e-05,
      "loss": 0.6031,
      "step": 2870
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.48725498388239e-05,
      "loss": 0.6178,
      "step": 2871
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.485454749578827e-05,
      "loss": 0.5554,
      "step": 2872
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4836547061553536e-05,
      "loss": 0.5415,
      "step": 2873
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4818548542797287e-05,
      "loss": 0.5968,
      "step": 2874
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4800551946196413e-05,
      "loss": 0.6102,
      "step": 2875
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.478255727842708e-05,
      "loss": 0.5373,
      "step": 2876
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4764564546164762e-05,
      "loss": 0.5539,
      "step": 2877
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4746573756084168e-05,
      "loss": 0.6017,
      "step": 2878
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4728584914859326e-05,
      "loss": 0.5984,
      "step": 2879
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.4710598029163532e-05,
      "loss": 0.6559,
      "step": 2880
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.469261310566937e-05,
      "loss": 0.5774,
      "step": 2881
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.467463015104865e-05,
      "loss": 0.5702,
      "step": 2882
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4656649171972494e-05,
      "loss": 0.5903,
      "step": 2883
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.46386701751113e-05,
      "loss": 0.6162,
      "step": 2884
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4620693167134675e-05,
      "loss": 0.5902,
      "step": 2885
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.460271815471154e-05,
      "loss": 0.5377,
      "step": 2886
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.458474514451006e-05,
      "loss": 0.6121,
      "step": 2887
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4566774143197664e-05,
      "loss": 0.6143,
      "step": 2888
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4548805157441007e-05,
      "loss": 0.5832,
      "step": 2889
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.453083819390603e-05,
      "loss": 0.6006,
      "step": 2890
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.451287325925792e-05,
      "loss": 0.5906,
      "step": 2891
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4494910360161078e-05,
      "loss": 0.5795,
      "step": 2892
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4476949503279188e-05,
      "loss": 0.6184,
      "step": 2893
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.445899069527516e-05,
      "loss": 0.5631,
      "step": 2894
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4441033942811166e-05,
      "loss": 0.6644,
      "step": 2895
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4423079252548558e-05,
      "loss": 0.5795,
      "step": 2896
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4405126631147985e-05,
      "loss": 0.559,
      "step": 2897
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4387176085269316e-05,
      "loss": 0.5527,
      "step": 2898
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4369227621571605e-05,
      "loss": 0.6151,
      "step": 2899
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4351281246713186e-05,
      "loss": 0.5595,
      "step": 2900
    },
    {
      "epoch": 2.21,
      "eval_loss": 0.9085999131202698,
      "eval_runtime": 14.3469,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 2900
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4333336967351597e-05,
      "loss": 0.5704,
      "step": 2901
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.431539479014361e-05,
      "loss": 0.5123,
      "step": 2902
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4297454721745197e-05,
      "loss": 0.6675,
      "step": 2903
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4279516768811552e-05,
      "loss": 0.5088,
      "step": 2904
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4261580937997107e-05,
      "loss": 0.5999,
      "step": 2905
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4243647235955486e-05,
      "loss": 0.5043,
      "step": 2906
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.4225715669339528e-05,
      "loss": 0.6215,
      "step": 2907
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4207786244801264e-05,
      "loss": 0.628,
      "step": 2908
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4189858968991972e-05,
      "loss": 0.6282,
      "step": 2909
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4171933848562084e-05,
      "loss": 0.6051,
      "step": 2910
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4154010890161266e-05,
      "loss": 0.5628,
      "step": 2911
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4136090100438382e-05,
      "loss": 0.599,
      "step": 2912
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4118171486041467e-05,
      "loss": 0.5825,
      "step": 2913
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4100255053617755e-05,
      "loss": 0.6682,
      "step": 2914
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4082340809813694e-05,
      "loss": 0.6411,
      "step": 2915
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.406442876127491e-05,
      "loss": 0.6583,
      "step": 2916
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4046518914646184e-05,
      "loss": 0.6163,
      "step": 2917
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4028611276571523e-05,
      "loss": 0.5946,
      "step": 2918
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.4010705853694093e-05,
      "loss": 0.671,
      "step": 2919
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.3992802652656254e-05,
      "loss": 0.6034,
      "step": 2920
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3974901680099507e-05,
      "loss": 0.5853,
      "step": 2921
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.395700294266456e-05,
      "loss": 0.6516,
      "step": 2922
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3939106446991296e-05,
      "loss": 0.6608,
      "step": 2923
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3921212199718727e-05,
      "loss": 0.5666,
      "step": 2924
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3903320207485064e-05,
      "loss": 0.6092,
      "step": 2925
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3885430476927676e-05,
      "loss": 0.4994,
      "step": 2926
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3867543014683102e-05,
      "loss": 0.6762,
      "step": 2927
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3849657827387003e-05,
      "loss": 0.6333,
      "step": 2928
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3831774921674226e-05,
      "loss": 0.645,
      "step": 2929
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3813894304178774e-05,
      "loss": 0.6126,
      "step": 2930
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.379601598153381e-05,
      "loss": 0.5588,
      "step": 2931
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.377813996037159e-05,
      "loss": 0.5403,
      "step": 2932
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.3760266247323575e-05,
      "loss": 0.6224,
      "step": 2933
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3742394849020366e-05,
      "loss": 0.5754,
      "step": 2934
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3724525772091665e-05,
      "loss": 0.6211,
      "step": 2935
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3706659023166337e-05,
      "loss": 0.6441,
      "step": 2936
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.368879460887239e-05,
      "loss": 0.5636,
      "step": 2937
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3670932535836978e-05,
      "loss": 0.6451,
      "step": 2938
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.365307281068634e-05,
      "loss": 0.5636,
      "step": 2939
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.363521544004588e-05,
      "loss": 0.58,
      "step": 2940
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3617360430540124e-05,
      "loss": 0.6007,
      "step": 2941
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3599507788792716e-05,
      "loss": 0.6116,
      "step": 2942
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3581657521426436e-05,
      "loss": 0.5281,
      "step": 2943
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3563809635063143e-05,
      "loss": 0.5465,
      "step": 2944
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.3545964136323874e-05,
      "loss": 0.6325,
      "step": 2945
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.352812103182872e-05,
      "loss": 0.5512,
      "step": 2946
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.351028032819693e-05,
      "loss": 0.6117,
      "step": 2947
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3492442032046838e-05,
      "loss": 0.6463,
      "step": 2948
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.347460614999588e-05,
      "loss": 0.5281,
      "step": 2949
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3456772688660618e-05,
      "loss": 0.5765,
      "step": 2950
    },
    {
      "epoch": 2.25,
      "eval_loss": 0.9085644483566284,
      "eval_runtime": 14.3485,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 2950
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3438941654656697e-05,
      "loss": 0.5705,
      "step": 2951
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3421113054598887e-05,
      "loss": 0.607,
      "step": 2952
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.340328689510102e-05,
      "loss": 0.6229,
      "step": 2953
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3385463182776036e-05,
      "loss": 0.5911,
      "step": 2954
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.336764192423598e-05,
      "loss": 0.5967,
      "step": 2955
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3349823126092e-05,
      "loss": 0.5535,
      "step": 2956
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.333200679495426e-05,
      "loss": 0.5435,
      "step": 2957
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.3314192937432094e-05,
      "loss": 0.6352,
      "step": 2958
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.329638156013388e-05,
      "loss": 0.5732,
      "step": 2959
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3278572669667065e-05,
      "loss": 0.5584,
      "step": 2960
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.326076627263819e-05,
      "loss": 0.6902,
      "step": 2961
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.324296237565287e-05,
      "loss": 0.5794,
      "step": 2962
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3225160985315805e-05,
      "loss": 0.5742,
      "step": 2963
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3207362108230723e-05,
      "loss": 0.5454,
      "step": 2964
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.318956575100046e-05,
      "loss": 0.5335,
      "step": 2965
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.317177192022692e-05,
      "loss": 0.5979,
      "step": 2966
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3153980622511023e-05,
      "loss": 0.6382,
      "step": 2967
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.31361918644528e-05,
      "loss": 0.5771,
      "step": 2968
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3118405652651313e-05,
      "loss": 0.6288,
      "step": 2969
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3100621993704703e-05,
      "loss": 0.579,
      "step": 2970
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3082840894210122e-05,
      "loss": 0.5602,
      "step": 2971
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.3065062360763814e-05,
      "loss": 0.5536,
      "step": 2972
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.304728639996106e-05,
      "loss": 0.6347,
      "step": 2973
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.3029513018396166e-05,
      "loss": 0.5767,
      "step": 2974
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.3011742222662502e-05,
      "loss": 0.6306,
      "step": 2975
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.299397401935248e-05,
      "loss": 0.6053,
      "step": 2976
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.2976208415057542e-05,
      "loss": 0.7341,
      "step": 2977
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.2958445416368173e-05,
      "loss": 0.688,
      "step": 2978
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.294068502987387e-05,
      "loss": 0.6279,
      "step": 2979
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.2922927262163186e-05,
      "loss": 0.5841,
      "step": 2980
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.2905172119823706e-05,
      "loss": 0.5905,
      "step": 2981
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.2887419609442007e-05,
      "loss": 0.6358,
      "step": 2982
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.286966973760373e-05,
      "loss": 0.6523,
      "step": 2983
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.2851922510893513e-05,
      "loss": 0.6374,
      "step": 2984
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.283417793589501e-05,
      "loss": 0.5353,
      "step": 2985
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2816436019190906e-05,
      "loss": 0.5695,
      "step": 2986
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.27986967673629e-05,
      "loss": 0.6418,
      "step": 2987
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2780960186991693e-05,
      "loss": 0.5923,
      "step": 2988
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2763226284656983e-05,
      "loss": 0.5739,
      "step": 2989
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2745495066937504e-05,
      "loss": 0.6539,
      "step": 2990
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2727766540410988e-05,
      "loss": 0.5703,
      "step": 2991
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.271004071165414e-05,
      "loss": 0.6099,
      "step": 2992
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.269231758724269e-05,
      "loss": 0.5153,
      "step": 2993
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2674597173751363e-05,
      "loss": 0.542,
      "step": 2994
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2656879477753886e-05,
      "loss": 0.6197,
      "step": 2995
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2639164505822946e-05,
      "loss": 0.6217,
      "step": 2996
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.262145226453025e-05,
      "loss": 0.6485,
      "step": 2997
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.2603742760446488e-05,
      "loss": 0.6085,
      "step": 2998
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2586036000141308e-05,
      "loss": 0.6064,
      "step": 2999
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.256833199018338e-05,
      "loss": 0.694,
      "step": 3000
    },
    {
      "epoch": 2.29,
      "eval_loss": 0.9075261950492859,
      "eval_runtime": 14.3473,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 3000
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2550630737140322e-05,
      "loss": 0.6436,
      "step": 3001
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2532932247578762e-05,
      "loss": 0.6168,
      "step": 3002
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2515236528064255e-05,
      "loss": 0.6122,
      "step": 3003
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2497543585161372e-05,
      "loss": 0.6597,
      "step": 3004
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.247985342543363e-05,
      "loss": 0.582,
      "step": 3005
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2462166055443548e-05,
      "loss": 0.5542,
      "step": 3006
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.244448148175254e-05,
      "loss": 0.5326,
      "step": 3007
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.242679971092105e-05,
      "loss": 0.6424,
      "step": 3008
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.240912074950847e-05,
      "loss": 0.5515,
      "step": 3009
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.239144460407311e-05,
      "loss": 0.6692,
      "step": 3010
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.2373771281172282e-05,
      "loss": 0.6823,
      "step": 3011
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.235610078736222e-05,
      "loss": 0.606,
      "step": 3012
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2338433129198152e-05,
      "loss": 0.589,
      "step": 3013
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.232076831323418e-05,
      "loss": 0.6252,
      "step": 3014
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2303106346023418e-05,
      "loss": 0.6025,
      "step": 3015
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2285447234117905e-05,
      "loss": 0.566,
      "step": 3016
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2267790984068597e-05,
      "loss": 0.6043,
      "step": 3017
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.225013760242542e-05,
      "loss": 0.5377,
      "step": 3018
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2232487095737222e-05,
      "loss": 0.584,
      "step": 3019
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2214839470551786e-05,
      "loss": 0.634,
      "step": 3020
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2197194733415812e-05,
      "loss": 0.5745,
      "step": 3021
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.217955289087496e-05,
      "loss": 0.6153,
      "step": 3022
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.216191394947379e-05,
      "loss": 0.5693,
      "step": 3023
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.2144277915755798e-05,
      "loss": 0.6879,
      "step": 3024
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.2126644796263388e-05,
      "loss": 0.5973,
      "step": 3025
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.2109014597537898e-05,
      "loss": 0.5959,
      "step": 3026
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.2091387326119588e-05,
      "loss": 0.6062,
      "step": 3027
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.2073762988547604e-05,
      "loss": 0.6106,
      "step": 3028
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.205614159136002e-05,
      "loss": 0.5692,
      "step": 3029
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.2038523141093828e-05,
      "loss": 0.5497,
      "step": 3030
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.2020907644284926e-05,
      "loss": 0.6053,
      "step": 3031
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.200329510746809e-05,
      "loss": 0.5601,
      "step": 3032
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.1985685537177024e-05,
      "loss": 0.6151,
      "step": 3033
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.196807893994434e-05,
      "loss": 0.5764,
      "step": 3034
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.19504753223015e-05,
      "loss": 0.621,
      "step": 3035
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.1932874690778913e-05,
      "loss": 0.6613,
      "step": 3036
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1915277051905854e-05,
      "loss": 0.7091,
      "step": 3037
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.189768241221051e-05,
      "loss": 0.594,
      "step": 3038
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.188009077821991e-05,
      "loss": 0.612,
      "step": 3039
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1862502156460004e-05,
      "loss": 0.5979,
      "step": 3040
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1844916553455643e-05,
      "loss": 0.6355,
      "step": 3041
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1827333975730493e-05,
      "loss": 0.5733,
      "step": 3042
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1809754429807157e-05,
      "loss": 0.5755,
      "step": 3043
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.179217792220709e-05,
      "loss": 0.5168,
      "step": 3044
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1774604459450638e-05,
      "loss": 0.6007,
      "step": 3045
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1757034048056975e-05,
      "loss": 0.5934,
      "step": 3046
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1739466694544185e-05,
      "loss": 0.5363,
      "step": 3047
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.17219024054292e-05,
      "loss": 0.5928,
      "step": 3048
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1704341187227838e-05,
      "loss": 0.5662,
      "step": 3049
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1686783046454726e-05,
      "loss": 0.5591,
      "step": 3050
    },
    {
      "epoch": 2.33,
      "eval_loss": 0.906229555606842,
      "eval_runtime": 14.3529,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 3050
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1669227989623395e-05,
      "loss": 0.5646,
      "step": 3051
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.165167602324623e-05,
      "loss": 0.583,
      "step": 3052
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1634127153834445e-05,
      "loss": 0.6597,
      "step": 3053
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1616581387898112e-05,
      "loss": 0.6094,
      "step": 3054
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.159903873194617e-05,
      "loss": 0.6876,
      "step": 3055
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.158149919248639e-05,
      "loss": 0.5709,
      "step": 3056
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1563962776025387e-05,
      "loss": 0.5947,
      "step": 3057
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1546429489068625e-05,
      "loss": 0.5505,
      "step": 3058
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1528899338120394e-05,
      "loss": 0.6635,
      "step": 3059
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.151137232968383e-05,
      "loss": 0.5761,
      "step": 3060
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1493848470260897e-05,
      "loss": 0.5479,
      "step": 3061
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.1476327766352412e-05,
      "loss": 0.6027,
      "step": 3062
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1458810224457987e-05,
      "loss": 0.5955,
      "step": 3063
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.144129585107608e-05,
      "loss": 0.5644,
      "step": 3064
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1423784652703978e-05,
      "loss": 0.6675,
      "step": 3065
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1406276635837794e-05,
      "loss": 0.6008,
      "step": 3066
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.138877180697243e-05,
      "loss": 0.6442,
      "step": 3067
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1371270172601633e-05,
      "loss": 0.7093,
      "step": 3068
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1353771739217965e-05,
      "loss": 0.5655,
      "step": 3069
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1336276513312803e-05,
      "loss": 0.6183,
      "step": 3070
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1318784501376297e-05,
      "loss": 0.5789,
      "step": 3071
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1301295709897448e-05,
      "loss": 0.6655,
      "step": 3072
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1283810145364052e-05,
      "loss": 0.616,
      "step": 3073
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.1266327814262707e-05,
      "loss": 0.5883,
      "step": 3074
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.124884872307879e-05,
      "loss": 0.5327,
      "step": 3075
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.12313728782965e-05,
      "loss": 0.6298,
      "step": 3076
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1213900286398845e-05,
      "loss": 0.5597,
      "step": 3077
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1196430953867574e-05,
      "loss": 0.6047,
      "step": 3078
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.117896488718327e-05,
      "loss": 0.6192,
      "step": 3079
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1161502092825307e-05,
      "loss": 0.6419,
      "step": 3080
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.114404257727183e-05,
      "loss": 0.6228,
      "step": 3081
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.112658634699976e-05,
      "loss": 0.6059,
      "step": 3082
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1109133408484804e-05,
      "loss": 0.5824,
      "step": 3083
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1091683768201475e-05,
      "loss": 0.6081,
      "step": 3084
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1074237432623013e-05,
      "loss": 0.5523,
      "step": 3085
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.105679440822148e-05,
      "loss": 0.5775,
      "step": 3086
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1039354701467674e-05,
      "loss": 0.5576,
      "step": 3087
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.1021918318831196e-05,
      "loss": 0.6467,
      "step": 3088
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.100448526678037e-05,
      "loss": 0.5698,
      "step": 3089
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.098705555178233e-05,
      "loss": 0.6207,
      "step": 3090
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0969629180302946e-05,
      "loss": 0.5943,
      "step": 3091
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0952206158806852e-05,
      "loss": 0.5748,
      "step": 3092
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.093478649375744e-05,
      "loss": 0.5901,
      "step": 3093
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0917370191616856e-05,
      "loss": 0.5843,
      "step": 3094
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0899957258846003e-05,
      "loss": 0.5722,
      "step": 3095
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.088254770190453e-05,
      "loss": 0.5657,
      "step": 3096
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0865141527250833e-05,
      "loss": 0.5318,
      "step": 3097
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0847738741342054e-05,
      "loss": 0.6612,
      "step": 3098
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0830339350634086e-05,
      "loss": 0.5564,
      "step": 3099
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0812943361581543e-05,
      "loss": 0.6341,
      "step": 3100
    },
    {
      "epoch": 2.36,
      "eval_loss": 0.9065569043159485,
      "eval_runtime": 14.3498,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 3100
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.0795550780637793e-05,
      "loss": 0.5488,
      "step": 3101
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0778161614254955e-05,
      "loss": 0.5638,
      "step": 3102
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0760775868883837e-05,
      "loss": 0.6025,
      "step": 3103
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.074339355097401e-05,
      "loss": 0.674,
      "step": 3104
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.072601466697377e-05,
      "loss": 0.5713,
      "step": 3105
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0708639223330147e-05,
      "loss": 0.5525,
      "step": 3106
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0691267226488857e-05,
      "loss": 0.6261,
      "step": 3107
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.067389868289438e-05,
      "loss": 0.6238,
      "step": 3108
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.065653359898991e-05,
      "loss": 0.5731,
      "step": 3109
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0639171981217323e-05,
      "loss": 0.5786,
      "step": 3110
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0621813836017245e-05,
      "loss": 0.6901,
      "step": 3111
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0604459169828995e-05,
      "loss": 0.602,
      "step": 3112
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.0587107989090632e-05,
      "loss": 0.5832,
      "step": 3113
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.056976030023886e-05,
      "loss": 0.6939,
      "step": 3114
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.055241610970915e-05,
      "loss": 0.6049,
      "step": 3115
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0535075423935654e-05,
      "loss": 0.5772,
      "step": 3116
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.05177382493512e-05,
      "loss": 0.6658,
      "step": 3117
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0500404592387347e-05,
      "loss": 0.6142,
      "step": 3118
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.048307445947433e-05,
      "loss": 0.5908,
      "step": 3119
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0465747857041104e-05,
      "loss": 0.6085,
      "step": 3120
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0448424791515265e-05,
      "loss": 0.5741,
      "step": 3121
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.043110526932313e-05,
      "loss": 0.567,
      "step": 3122
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0413789296889704e-05,
      "loss": 0.5422,
      "step": 3123
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0396476880638683e-05,
      "loss": 0.6032,
      "step": 3124
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0379168026992398e-05,
      "loss": 0.6015,
      "step": 3125
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0361862742371898e-05,
      "loss": 0.5577,
      "step": 3126
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0344561033196917e-05,
      "loss": 0.5523,
      "step": 3127
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0327262905885827e-05,
      "loss": 0.583,
      "step": 3128
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0309968366855685e-05,
      "loss": 0.6013,
      "step": 3129
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.029267742252223e-05,
      "loss": 0.5704,
      "step": 3130
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0275390079299863e-05,
      "loss": 0.5918,
      "step": 3131
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0258106343601634e-05,
      "loss": 0.5738,
      "step": 3132
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.024082622183927e-05,
      "loss": 0.6191,
      "step": 3133
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0223549720423146e-05,
      "loss": 0.6428,
      "step": 3134
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.02062768457623e-05,
      "loss": 0.6291,
      "step": 3135
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0189007604264433e-05,
      "loss": 0.6252,
      "step": 3136
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.017174200233589e-05,
      "loss": 0.6708,
      "step": 3137
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0154480046381664e-05,
      "loss": 0.5716,
      "step": 3138
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.013722174280539e-05,
      "loss": 0.5821,
      "step": 3139
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0119967098009356e-05,
      "loss": 0.5909,
      "step": 3140
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.010271611839451e-05,
      "loss": 0.5737,
      "step": 3141
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.00854688103604e-05,
      "loss": 0.5459,
      "step": 3142
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0068225180305237e-05,
      "loss": 0.5638,
      "step": 3143
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0050985234625874e-05,
      "loss": 0.6002,
      "step": 3144
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0033748979717798e-05,
      "loss": 0.6695,
      "step": 3145
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.001651642197509e-05,
      "loss": 0.581,
      "step": 3146
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.9999287567790502e-05,
      "loss": 0.5315,
      "step": 3147
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.9982062423555396e-05,
      "loss": 0.5499,
      "step": 3148
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.996484099565977e-05,
      "loss": 0.5997,
      "step": 3149
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.99476232904922e-05,
      "loss": 0.5486,
      "step": 3150
    },
    {
      "epoch": 2.4,
      "eval_loss": 0.9071575403213501,
      "eval_runtime": 14.3477,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 3150
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.9930409314439937e-05,
      "loss": 0.6121,
      "step": 3151
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.9913199073888834e-05,
      "loss": 0.6106,
      "step": 3152
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.9895992575223316e-05,
      "loss": 0.6597,
      "step": 3153
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.987878982482647e-05,
      "loss": 0.5484,
      "step": 3154
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.986159082907997e-05,
      "loss": 0.6308,
      "step": 3155
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9844395594364126e-05,
      "loss": 0.5924,
      "step": 3156
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9827204127057795e-05,
      "loss": 0.5815,
      "step": 3157
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.981001643353848e-05,
      "loss": 0.5499,
      "step": 3158
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9792832520182298e-05,
      "loss": 0.6628,
      "step": 3159
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9775652393363903e-05,
      "loss": 0.5569,
      "step": 3160
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.97584760594566e-05,
      "loss": 0.5519,
      "step": 3161
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9741303524832265e-05,
      "loss": 0.5764,
      "step": 3162
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.972413479586138e-05,
      "loss": 0.5958,
      "step": 3163
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9706969878912985e-05,
      "loss": 0.5879,
      "step": 3164
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9689808780354725e-05,
      "loss": 0.6505,
      "step": 3165
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9672651506552846e-05,
      "loss": 0.561,
      "step": 3166
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9655498063872132e-05,
      "loss": 0.6174,
      "step": 3167
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9638348458675994e-05,
      "loss": 0.6962,
      "step": 3168
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9621202697326376e-05,
      "loss": 0.5529,
      "step": 3169
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9604060786183826e-05,
      "loss": 0.5938,
      "step": 3170
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9586922731607446e-05,
      "loss": 0.6585,
      "step": 3171
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9569788539954922e-05,
      "loss": 0.558,
      "step": 3172
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9552658217582505e-05,
      "loss": 0.6476,
      "step": 3173
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.953553177084499e-05,
      "loss": 0.5614,
      "step": 3174
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9518409206095758e-05,
      "loss": 0.6483,
      "step": 3175
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.950129052968674e-05,
      "loss": 0.5673,
      "step": 3176
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9484175747968436e-05,
      "loss": 0.5428,
      "step": 3177
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9467064867289876e-05,
      "loss": 0.6594,
      "step": 3178
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.944995789399866e-05,
      "loss": 0.6755,
      "step": 3179
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9432854834440934e-05,
      "loss": 0.6057,
      "step": 3180
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.941575569496142e-05,
      "loss": 0.7042,
      "step": 3181
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.939866048190333e-05,
      "loss": 0.5819,
      "step": 3182
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9381569201608457e-05,
      "loss": 0.601,
      "step": 3183
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9364481860417154e-05,
      "loss": 0.6223,
      "step": 3184
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9347398464668246e-05,
      "loss": 0.6301,
      "step": 3185
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9330319020699153e-05,
      "loss": 0.6266,
      "step": 3186
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9313243534845816e-05,
      "loss": 0.6186,
      "step": 3187
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.929617201344271e-05,
      "loss": 0.6003,
      "step": 3188
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9279104462822808e-05,
      "loss": 0.6221,
      "step": 3189
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9262040889317647e-05,
      "loss": 0.5672,
      "step": 3190
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.924498129925729e-05,
      "loss": 0.5846,
      "step": 3191
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9227925698970278e-05,
      "loss": 0.6981,
      "step": 3192
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9210874094783717e-05,
      "loss": 0.5918,
      "step": 3193
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9193826493023218e-05,
      "loss": 0.5876,
      "step": 3194
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9176782900012913e-05,
      "loss": 0.5801,
      "step": 3195
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9159743322075418e-05,
      "loss": 0.6721,
      "step": 3196
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9142707765531886e-05,
      "loss": 0.5567,
      "step": 3197
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9125676236701978e-05,
      "loss": 0.5771,
      "step": 3198
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9108648741903866e-05,
      "loss": 0.5769,
      "step": 3199
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9091625287454184e-05,
      "loss": 0.641,
      "step": 3200
    },
    {
      "epoch": 2.44,
      "eval_loss": 0.9058852791786194,
      "eval_runtime": 14.3555,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 3200
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.907460587966812e-05,
      "loss": 0.5727,
      "step": 3201
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.905759052485934e-05,
      "loss": 0.6113,
      "step": 3202
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9040579229339992e-05,
      "loss": 0.5969,
      "step": 3203
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.9023571999420737e-05,
      "loss": 0.6155,
      "step": 3204
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.900656884141072e-05,
      "loss": 0.6422,
      "step": 3205
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8989569761617586e-05,
      "loss": 0.5747,
      "step": 3206
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8972574766347446e-05,
      "loss": 0.5759,
      "step": 3207
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.895558386190492e-05,
      "loss": 0.621,
      "step": 3208
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.89385970545931e-05,
      "loss": 0.5791,
      "step": 3209
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8921614350713538e-05,
      "loss": 0.6098,
      "step": 3210
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8904635756566296e-05,
      "loss": 0.5322,
      "step": 3211
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.88876612784499e-05,
      "loss": 0.6069,
      "step": 3212
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8870690922661346e-05,
      "loss": 0.611,
      "step": 3213
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8853724695496088e-05,
      "loss": 0.5967,
      "step": 3214
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.883676260324808e-05,
      "loss": 0.5501,
      "step": 3215
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8819804652209728e-05,
      "loss": 0.5892,
      "step": 3216
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8802850848671874e-05,
      "loss": 0.5804,
      "step": 3217
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8785901198923857e-05,
      "loss": 0.5864,
      "step": 3218
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8768955709253467e-05,
      "loss": 0.6703,
      "step": 3219
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8752014385946962e-05,
      "loss": 0.6634,
      "step": 3220
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8735077235289005e-05,
      "loss": 0.6034,
      "step": 3221
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8718144263562765e-05,
      "loss": 0.638,
      "step": 3222
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.870121547704984e-05,
      "loss": 0.5426,
      "step": 3223
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.868429088203029e-05,
      "loss": 0.5767,
      "step": 3224
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8667370484782585e-05,
      "loss": 0.5417,
      "step": 3225
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8650454291583667e-05,
      "loss": 0.5813,
      "step": 3226
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8633542308708925e-05,
      "loss": 0.5934,
      "step": 3227
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8616634542432148e-05,
      "loss": 0.6255,
      "step": 3228
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8599730999025596e-05,
      "loss": 0.5974,
      "step": 3229
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8582831684759953e-05,
      "loss": 0.5871,
      "step": 3230
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8565936605904348e-05,
      "loss": 0.5884,
      "step": 3231
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.854904576872629e-05,
      "loss": 0.5542,
      "step": 3232
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.853215917949177e-05,
      "loss": 0.5966,
      "step": 3233
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8515276844465187e-05,
      "loss": 0.5297,
      "step": 3234
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.849839876990933e-05,
      "loss": 0.5905,
      "step": 3235
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.848152496208545e-05,
      "loss": 0.6258,
      "step": 3236
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8464655427253203e-05,
      "loss": 0.5889,
      "step": 3237
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.844779017167066e-05,
      "loss": 0.5518,
      "step": 3238
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.843092920159427e-05,
      "loss": 0.6263,
      "step": 3239
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.841407252327895e-05,
      "loss": 0.6179,
      "step": 3240
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.839722014297799e-05,
      "loss": 0.5499,
      "step": 3241
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8380372066943085e-05,
      "loss": 0.6303,
      "step": 3242
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8363528301424356e-05,
      "loss": 0.6232,
      "step": 3243
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.834668885267029e-05,
      "loss": 0.6119,
      "step": 3244
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8329853726927812e-05,
      "loss": 0.5255,
      "step": 3245
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8313022930442212e-05,
      "loss": 0.6864,
      "step": 3246
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8296196469457184e-05,
      "loss": 0.6222,
      "step": 3247
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8279374350214827e-05,
      "loss": 0.6073,
      "step": 3248
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8262556578955608e-05,
      "loss": 0.6065,
      "step": 3249
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.824574316191839e-05,
      "loss": 0.5855,
      "step": 3250
    },
    {
      "epoch": 2.48,
      "eval_loss": 0.9090210199356079,
      "eval_runtime": 14.3451,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 3250
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8228934105340423e-05,
      "loss": 0.585,
      "step": 3251
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.821212941545735e-05,
      "loss": 0.5573,
      "step": 3252
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8195329098503163e-05,
      "loss": 0.6837,
      "step": 3253
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8178533160710254e-05,
      "loss": 0.6412,
      "step": 3254
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8161741608309394e-05,
      "loss": 0.633,
      "step": 3255
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.814495444752973e-05,
      "loss": 0.5447,
      "step": 3256
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.812817168459875e-05,
      "loss": 0.6011,
      "step": 3257
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8111393325742337e-05,
      "loss": 0.6036,
      "step": 3258
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8094619377184747e-05,
      "loss": 0.5918,
      "step": 3259
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8077849845148566e-05,
      "loss": 0.4989,
      "step": 3260
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.806108473585477e-05,
      "loss": 0.6214,
      "step": 3261
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8044324055522693e-05,
      "loss": 0.5756,
      "step": 3262
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8027567810370027e-05,
      "loss": 0.5693,
      "step": 3263
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8010816006612787e-05,
      "loss": 0.5922,
      "step": 3264
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7994068650465376e-05,
      "loss": 0.5738,
      "step": 3265
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7977325748140553e-05,
      "loss": 0.5966,
      "step": 3266
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.796058730584938e-05,
      "loss": 0.5411,
      "step": 3267
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7943853329801308e-05,
      "loss": 0.5935,
      "step": 3268
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.792712382620411e-05,
      "loss": 0.5495,
      "step": 3269
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.791039880126392e-05,
      "loss": 0.569,
      "step": 3270
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.789367826118517e-05,
      "loss": 0.6894,
      "step": 3271
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7876962212170672e-05,
      "loss": 0.6205,
      "step": 3272
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7860250660421546e-05,
      "loss": 0.5763,
      "step": 3273
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7843543612137264e-05,
      "loss": 0.6898,
      "step": 3274
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7826841073515598e-05,
      "loss": 0.5341,
      "step": 3275
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7810143050752668e-05,
      "loss": 0.6114,
      "step": 3276
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7793449550042924e-05,
      "loss": 0.5766,
      "step": 3277
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.777676057757913e-05,
      "loss": 0.6656,
      "step": 3278
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.776007613955235e-05,
      "loss": 0.5544,
      "step": 3279
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7743396242152004e-05,
      "loss": 0.628,
      "step": 3280
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.772672089156581e-05,
      "loss": 0.5557,
      "step": 3281
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.771005009397978e-05,
      "loss": 0.5997,
      "step": 3282
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7693383855578276e-05,
      "loss": 0.6359,
      "step": 3283
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7676722182543938e-05,
      "loss": 0.5732,
      "step": 3284
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7660065081057717e-05,
      "loss": 0.6305,
      "step": 3285
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.764341255729888e-05,
      "loss": 0.5848,
      "step": 3286
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.762676461744499e-05,
      "loss": 0.6282,
      "step": 3287
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7610121267671912e-05,
      "loss": 0.522,
      "step": 3288
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7593482514153787e-05,
      "loss": 0.5833,
      "step": 3289
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7576848363063084e-05,
      "loss": 0.5671,
      "step": 3290
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7560218820570555e-05,
      "loss": 0.6436,
      "step": 3291
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7543593892845222e-05,
      "loss": 0.6065,
      "step": 3292
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.752697358605441e-05,
      "loss": 0.5963,
      "step": 3293
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7510357906363743e-05,
      "loss": 0.5893,
      "step": 3294
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.749374685993711e-05,
      "loss": 0.5871,
      "step": 3295
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.747714045293668e-05,
      "loss": 0.5747,
      "step": 3296
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7460538691522916e-05,
      "loss": 0.5912,
      "step": 3297
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.744394158185453e-05,
      "loss": 0.5828,
      "step": 3298
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7427349130088552e-05,
      "loss": 0.5952,
      "step": 3299
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.741076134238025e-05,
      "loss": 0.5486,
      "step": 3300
    },
    {
      "epoch": 2.52,
      "eval_loss": 0.907595694065094,
      "eval_runtime": 14.353,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 3300
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.739417822488317e-05,
      "loss": 0.6628,
      "step": 3301
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7377599783749125e-05,
      "loss": 0.617,
      "step": 3302
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7361026025128188e-05,
      "loss": 0.6024,
      "step": 3303
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7344456955168717e-05,
      "loss": 0.6275,
      "step": 3304
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.732789258001731e-05,
      "loss": 0.5949,
      "step": 3305
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7311332905818825e-05,
      "loss": 0.5461,
      "step": 3306
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7294777938716366e-05,
      "loss": 0.5974,
      "step": 3307
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7278227684851336e-05,
      "loss": 0.5508,
      "step": 3308
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7261682150363342e-05,
      "loss": 0.5807,
      "step": 3309
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7245141341390257e-05,
      "loss": 0.6137,
      "step": 3310
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7228605264068186e-05,
      "loss": 0.5247,
      "step": 3311
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7212073924531522e-05,
      "loss": 0.5881,
      "step": 3312
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7195547328912858e-05,
      "loss": 0.5315,
      "step": 3313
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7179025483343035e-05,
      "loss": 0.5863,
      "step": 3314
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7162508393951133e-05,
      "loss": 0.5146,
      "step": 3315
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7145996066864504e-05,
      "loss": 0.6345,
      "step": 3316
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7129488508208655e-05,
      "loss": 0.6453,
      "step": 3317
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7112985724107405e-05,
      "loss": 0.5522,
      "step": 3318
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7096487720682742e-05,
      "loss": 0.5875,
      "step": 3319
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.707999450405495e-05,
      "loss": 0.6068,
      "step": 3320
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.706350608034244e-05,
      "loss": 0.5976,
      "step": 3321
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.704702245566193e-05,
      "loss": 0.6185,
      "step": 3322
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.7030543636128303e-05,
      "loss": 0.5451,
      "step": 3323
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.7014069627854728e-05,
      "loss": 0.6703,
      "step": 3324
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6997600436952484e-05,
      "loss": 0.6138,
      "step": 3325
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6981136069531164e-05,
      "loss": 0.5277,
      "step": 3326
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.696467653169852e-05,
      "loss": 0.5794,
      "step": 3327
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6948221829560513e-05,
      "loss": 0.6227,
      "step": 3328
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.693177196922132e-05,
      "loss": 0.579,
      "step": 3329
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6915326956783333e-05,
      "loss": 0.5596,
      "step": 3330
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6898886798347133e-05,
      "loss": 0.5813,
      "step": 3331
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.688245150001149e-05,
      "loss": 0.573,
      "step": 3332
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6866021067873382e-05,
      "loss": 0.6701,
      "step": 3333
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6849595508028e-05,
      "loss": 0.528,
      "step": 3334
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6833174826568695e-05,
      "loss": 0.5731,
      "step": 3335
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6816759029587026e-05,
      "loss": 0.5534,
      "step": 3336
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6800348123172723e-05,
      "loss": 0.5863,
      "step": 3337
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6783942113413746e-05,
      "loss": 0.5668,
      "step": 3338
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.676754100639619e-05,
      "loss": 0.634,
      "step": 3339
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6751144808204354e-05,
      "loss": 0.6415,
      "step": 3340
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6734753524920703e-05,
      "loss": 0.5339,
      "step": 3341
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6718367162625883e-05,
      "loss": 0.6817,
      "step": 3342
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6701985727398738e-05,
      "loss": 0.5812,
      "step": 3343
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.668560922531625e-05,
      "loss": 0.614,
      "step": 3344
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6669237662453607e-05,
      "loss": 0.5759,
      "step": 3345
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6652871044884105e-05,
      "loss": 0.6313,
      "step": 3346
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.6636509378679276e-05,
      "loss": 0.5804,
      "step": 3347
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6620152669908762e-05,
      "loss": 0.6101,
      "step": 3348
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6603800924640425e-05,
      "loss": 0.6043,
      "step": 3349
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6587454148940194e-05,
      "loss": 0.6025,
      "step": 3350
    },
    {
      "epoch": 2.56,
      "eval_loss": 0.908098042011261,
      "eval_runtime": 14.3445,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 3350
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6571112348872248e-05,
      "loss": 0.602,
      "step": 3351
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.655477553049887e-05,
      "loss": 0.5954,
      "step": 3352
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6538443699880493e-05,
      "loss": 0.6449,
      "step": 3353
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.652211686307572e-05,
      "loss": 0.5597,
      "step": 3354
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.65057950261413e-05,
      "loss": 0.5606,
      "step": 3355
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.648947819513212e-05,
      "loss": 0.5849,
      "step": 3356
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6473166376101203e-05,
      "loss": 0.5761,
      "step": 3357
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.645685957509971e-05,
      "loss": 0.5668,
      "step": 3358
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.6440557798176997e-05,
      "loss": 0.4722,
      "step": 3359
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.642426105138045e-05,
      "loss": 0.533,
      "step": 3360
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.640796934075568e-05,
      "loss": 0.5844,
      "step": 3361
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6391682672346393e-05,
      "loss": 0.5912,
      "step": 3362
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6375401052194457e-05,
      "loss": 0.6103,
      "step": 3363
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6359124486339785e-05,
      "loss": 0.5787,
      "step": 3364
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6342852980820518e-05,
      "loss": 0.6329,
      "step": 3365
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6326586541672856e-05,
      "loss": 0.5982,
      "step": 3366
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6310325174931135e-05,
      "loss": 0.5884,
      "step": 3367
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.62940688866278e-05,
      "loss": 0.6544,
      "step": 3368
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.627781768279344e-05,
      "loss": 0.6326,
      "step": 3369
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6261571569456738e-05,
      "loss": 0.5663,
      "step": 3370
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6245330552644485e-05,
      "loss": 0.6215,
      "step": 3371
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.6229094638381577e-05,
      "loss": 0.5641,
      "step": 3372
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.621286383269105e-05,
      "loss": 0.6133,
      "step": 3373
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.619663814159401e-05,
      "loss": 0.5856,
      "step": 3374
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.618041757110966e-05,
      "loss": 0.5646,
      "step": 3375
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6164202127255356e-05,
      "loss": 0.5948,
      "step": 3376
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6147991816046497e-05,
      "loss": 0.5951,
      "step": 3377
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6131786643496597e-05,
      "loss": 0.5664,
      "step": 3378
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6115586615617265e-05,
      "loss": 0.6127,
      "step": 3379
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.609939173841821e-05,
      "loss": 0.5749,
      "step": 3380
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.608320201790722e-05,
      "loss": 0.5864,
      "step": 3381
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6067017460090167e-05,
      "loss": 0.5969,
      "step": 3382
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6050838070971e-05,
      "loss": 0.5588,
      "step": 3383
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.6034663856551805e-05,
      "loss": 0.5762,
      "step": 3384
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.601849482283265e-05,
      "loss": 0.6047,
      "step": 3385
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.6002330975811777e-05,
      "loss": 0.5702,
      "step": 3386
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5986172321485446e-05,
      "loss": 0.5586,
      "step": 3387
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5970018865848033e-05,
      "loss": 0.5577,
      "step": 3388
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5953870614891924e-05,
      "loss": 0.616,
      "step": 3389
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5937727574607645e-05,
      "loss": 0.6479,
      "step": 3390
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.592158975098374e-05,
      "loss": 0.6141,
      "step": 3391
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.590545715000684e-05,
      "loss": 0.6664,
      "step": 3392
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5889329777661613e-05,
      "loss": 0.5308,
      "step": 3393
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.587320763993084e-05,
      "loss": 0.6092,
      "step": 3394
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5857090742795304e-05,
      "loss": 0.6331,
      "step": 3395
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.584097909223387e-05,
      "loss": 0.5804,
      "step": 3396
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.582487269422344e-05,
      "loss": 0.6378,
      "step": 3397
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.5808771554739015e-05,
      "loss": 0.5128,
      "step": 3398
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5792675679753587e-05,
      "loss": 0.5748,
      "step": 3399
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.577658507523822e-05,
      "loss": 0.558,
      "step": 3400
    },
    {
      "epoch": 2.6,
      "eval_loss": 0.9083759188652039,
      "eval_runtime": 14.3453,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 3400
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5760499747162016e-05,
      "loss": 0.5766,
      "step": 3401
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.574441970149215e-05,
      "loss": 0.5501,
      "step": 3402
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5728344944193783e-05,
      "loss": 0.6261,
      "step": 3403
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.571227548123016e-05,
      "loss": 0.654,
      "step": 3404
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.569621131856254e-05,
      "loss": 0.5286,
      "step": 3405
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5680152462150244e-05,
      "loss": 0.6014,
      "step": 3406
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5664098917950556e-05,
      "loss": 0.6788,
      "step": 3407
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5648050691918877e-05,
      "loss": 0.6296,
      "step": 3408
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5632007790008583e-05,
      "loss": 0.5897,
      "step": 3409
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5615970218171064e-05,
      "loss": 0.5874,
      "step": 3410
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.559993798235579e-05,
      "loss": 0.5793,
      "step": 3411
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.558391108851019e-05,
      "loss": 0.5708,
      "step": 3412
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.556788954257975e-05,
      "loss": 0.5608,
      "step": 3413
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5551873350507938e-05,
      "loss": 0.6154,
      "step": 3414
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.553586251823629e-05,
      "loss": 0.6339,
      "step": 3415
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5519857051704303e-05,
      "loss": 0.5688,
      "step": 3416
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.55038569568495e-05,
      "loss": 0.6575,
      "step": 3417
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5487862239607403e-05,
      "loss": 0.6371,
      "step": 3418
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5471872905911573e-05,
      "loss": 0.5904,
      "step": 3419
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5455888961693533e-05,
      "loss": 0.608,
      "step": 3420
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.543991041288283e-05,
      "loss": 0.5953,
      "step": 3421
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.5423937265406987e-05,
      "loss": 0.6094,
      "step": 3422
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.540796952519156e-05,
      "loss": 0.6297,
      "step": 3423
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.539200719816007e-05,
      "loss": 0.562,
      "step": 3424
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5376050290234044e-05,
      "loss": 0.6066,
      "step": 3425
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5360098807332967e-05,
      "loss": 0.6244,
      "step": 3426
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.534415275537438e-05,
      "loss": 0.6127,
      "step": 3427
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5328212140273717e-05,
      "loss": 0.6094,
      "step": 3428
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.531227696794448e-05,
      "loss": 0.577,
      "step": 3429
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5296347244298098e-05,
      "loss": 0.5745,
      "step": 3430
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5280422975244022e-05,
      "loss": 0.6225,
      "step": 3431
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5264504166689613e-05,
      "loss": 0.579,
      "step": 3432
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5248590824540292e-05,
      "loss": 0.5709,
      "step": 3433
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5232682954699381e-05,
      "loss": 0.564,
      "step": 3434
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5216780563068204e-05,
      "loss": 0.5882,
      "step": 3435
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.520088365554604e-05,
      "loss": 0.625,
      "step": 3436
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.5184992238030162e-05,
      "loss": 0.5922,
      "step": 3437
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5169106316415776e-05,
      "loss": 0.593,
      "step": 3438
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5153225896596053e-05,
      "loss": 0.6239,
      "step": 3439
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5137350984462125e-05,
      "loss": 0.747,
      "step": 3440
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5121481585903104e-05,
      "loss": 0.6566,
      "step": 3441
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.510561770680602e-05,
      "loss": 0.5386,
      "step": 3442
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5089759353055882e-05,
      "loss": 0.6635,
      "step": 3443
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5073906530535623e-05,
      "loss": 0.5453,
      "step": 3444
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5058059245126168e-05,
      "loss": 0.5597,
      "step": 3445
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5042217502706347e-05,
      "loss": 0.5607,
      "step": 3446
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.502638130915295e-05,
      "loss": 0.5538,
      "step": 3447
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.5010550670340688e-05,
      "loss": 0.6558,
      "step": 3448
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.499472559214226e-05,
      "loss": 0.5725,
      "step": 3449
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.497890608042826e-05,
      "loss": 0.5873,
      "step": 3450
    },
    {
      "epoch": 2.63,
      "eval_loss": 0.9048791527748108,
      "eval_runtime": 14.3528,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 3450
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4963092141067224e-05,
      "loss": 0.5807,
      "step": 3451
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.494728377992563e-05,
      "loss": 0.5394,
      "step": 3452
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4931481002867871e-05,
      "loss": 0.5891,
      "step": 3453
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4915683815756295e-05,
      "loss": 0.5895,
      "step": 3454
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4899892224451159e-05,
      "loss": 0.5391,
      "step": 3455
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4884106234810645e-05,
      "loss": 0.6293,
      "step": 3456
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4868325852690842e-05,
      "loss": 0.608,
      "step": 3457
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4852551083945803e-05,
      "loss": 0.6653,
      "step": 3458
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4836781934427452e-05,
      "loss": 0.6244,
      "step": 3459
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4821018409985656e-05,
      "loss": 0.5834,
      "step": 3460
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.480526051646817e-05,
      "loss": 0.6888,
      "step": 3461
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4789508259720698e-05,
      "loss": 0.5337,
      "step": 3462
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.4773761645586822e-05,
      "loss": 0.6155,
      "step": 3463
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4758020679908042e-05,
      "loss": 0.591,
      "step": 3464
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4742285368523744e-05,
      "loss": 0.6043,
      "step": 3465
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4726555717271273e-05,
      "loss": 0.6346,
      "step": 3466
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4710831731985786e-05,
      "loss": 0.5885,
      "step": 3467
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4695113418500422e-05,
      "loss": 0.6175,
      "step": 3468
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4679400782646157e-05,
      "loss": 0.5534,
      "step": 3469
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4663693830251916e-05,
      "loss": 0.5954,
      "step": 3470
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4647992567144441e-05,
      "loss": 0.6197,
      "step": 3471
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4632296999148443e-05,
      "loss": 0.5884,
      "step": 3472
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4616607132086454e-05,
      "loss": 0.539,
      "step": 3473
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.460092297177896e-05,
      "loss": 0.632,
      "step": 3474
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.4585244524044245e-05,
      "loss": 0.605,
      "step": 3475
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.456957179469855e-05,
      "loss": 0.6022,
      "step": 3476
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.455390478955596e-05,
      "loss": 0.5616,
      "step": 3477
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4538243514428434e-05,
      "loss": 0.5772,
      "step": 3478
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4522587975125804e-05,
      "loss": 0.6183,
      "step": 3479
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.45069381774558e-05,
      "loss": 0.528,
      "step": 3480
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4491294127224e-05,
      "loss": 0.6007,
      "step": 3481
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.447565583023385e-05,
      "loss": 0.6282,
      "step": 3482
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4460023292286655e-05,
      "loss": 0.6965,
      "step": 3483
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.444439651918162e-05,
      "loss": 0.5822,
      "step": 3484
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4428775516715771e-05,
      "loss": 0.5682,
      "step": 3485
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4413160290684008e-05,
      "loss": 0.6245,
      "step": 3486
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4397550846879081e-05,
      "loss": 0.5639,
      "step": 3487
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.438194719109162e-05,
      "loss": 0.6741,
      "step": 3488
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.4366349329110078e-05,
      "loss": 0.5751,
      "step": 3489
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4350757266720765e-05,
      "loss": 0.654,
      "step": 3490
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4335171009707879e-05,
      "loss": 0.5383,
      "step": 3491
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4319590563853379e-05,
      "loss": 0.6399,
      "step": 3492
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.430401593493716e-05,
      "loss": 0.5816,
      "step": 3493
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4288447128736894e-05,
      "loss": 0.5703,
      "step": 3494
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4272884151028151e-05,
      "loss": 0.5643,
      "step": 3495
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4257327007584264e-05,
      "loss": 0.649,
      "step": 3496
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4241775704176473e-05,
      "loss": 0.5697,
      "step": 3497
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4226230246573803e-05,
      "loss": 0.595,
      "step": 3498
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4210690640543159e-05,
      "loss": 0.6709,
      "step": 3499
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4195156891849205e-05,
      "loss": 0.5555,
      "step": 3500
    },
    {
      "epoch": 2.67,
      "eval_loss": 0.9056723713874817,
      "eval_runtime": 14.3411,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 3500
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4179629006254506e-05,
      "loss": 0.6475,
      "step": 3501
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4164106989519407e-05,
      "loss": 0.6426,
      "step": 3502
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4148590847402082e-05,
      "loss": 0.5947,
      "step": 3503
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.413308058565853e-05,
      "loss": 0.555,
      "step": 3504
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4117576210042585e-05,
      "loss": 0.5855,
      "step": 3505
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4102077726305874e-05,
      "loss": 0.5417,
      "step": 3506
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4086585140197843e-05,
      "loss": 0.7197,
      "step": 3507
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.407109845746575e-05,
      "loss": 0.5591,
      "step": 3508
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4055617683854692e-05,
      "loss": 0.6209,
      "step": 3509
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4040142825107507e-05,
      "loss": 0.5858,
      "step": 3510
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4024673886964919e-05,
      "loss": 0.5444,
      "step": 3511
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4009210875165385e-05,
      "loss": 0.5672,
      "step": 3512
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.3993753795445237e-05,
      "loss": 0.5621,
      "step": 3513
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.3978302653538518e-05,
      "loss": 0.6263,
      "step": 3514
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.3962857455177146e-05,
      "loss": 0.5688,
      "step": 3515
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3947418206090798e-05,
      "loss": 0.5692,
      "step": 3516
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3931984912006943e-05,
      "loss": 0.5796,
      "step": 3517
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.391655757865084e-05,
      "loss": 0.5717,
      "step": 3518
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3901136211745563e-05,
      "loss": 0.6872,
      "step": 3519
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.388572081701195e-05,
      "loss": 0.6278,
      "step": 3520
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3870311400168619e-05,
      "loss": 0.6047,
      "step": 3521
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.385490796693197e-05,
      "loss": 0.6102,
      "step": 3522
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3839510523016217e-05,
      "loss": 0.6044,
      "step": 3523
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.382411907413331e-05,
      "loss": 0.6657,
      "step": 3524
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3808733625992987e-05,
      "loss": 0.5726,
      "step": 3525
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.379335418430278e-05,
      "loss": 0.618,
      "step": 3526
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3777980754767974e-05,
      "loss": 0.5411,
      "step": 3527
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3762613343091625e-05,
      "loss": 0.5698,
      "step": 3528
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3747251954974546e-05,
      "loss": 0.5969,
      "step": 3529
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3731896596115351e-05,
      "loss": 0.6127,
      "step": 3530
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3716547272210387e-05,
      "loss": 0.7299,
      "step": 3531
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3701203988953765e-05,
      "loss": 0.5356,
      "step": 3532
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3685866752037354e-05,
      "loss": 0.6114,
      "step": 3533
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3670535567150815e-05,
      "loss": 0.64,
      "step": 3534
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3655210439981493e-05,
      "loss": 0.5486,
      "step": 3535
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.363989137621456e-05,
      "loss": 0.5491,
      "step": 3536
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3624578381532882e-05,
      "loss": 0.635,
      "step": 3537
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3609271461617138e-05,
      "loss": 0.5431,
      "step": 3538
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.3593970622145663e-05,
      "loss": 0.6112,
      "step": 3539
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.357867586879462e-05,
      "loss": 0.639,
      "step": 3540
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3563387207237875e-05,
      "loss": 0.6052,
      "step": 3541
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3548104643147034e-05,
      "loss": 0.5503,
      "step": 3542
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.353282818219144e-05,
      "loss": 0.6592,
      "step": 3543
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3517557830038202e-05,
      "loss": 0.5926,
      "step": 3544
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3502293592352126e-05,
      "loss": 0.5705,
      "step": 3545
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.348703547479576e-05,
      "loss": 0.587,
      "step": 3546
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3471783483029387e-05,
      "loss": 0.6793,
      "step": 3547
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3456537622711023e-05,
      "loss": 0.6059,
      "step": 3548
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3441297899496402e-05,
      "loss": 0.6137,
      "step": 3549
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3426064319038976e-05,
      "loss": 0.6205,
      "step": 3550
    },
    {
      "epoch": 2.71,
      "eval_loss": 0.9059410691261292,
      "eval_runtime": 14.3406,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 3550
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3410836886989913e-05,
      "loss": 0.5936,
      "step": 3551
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3395615608998148e-05,
      "loss": 0.5553,
      "step": 3552
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.338040049071025e-05,
      "loss": 0.6752,
      "step": 3553
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3365191537770579e-05,
      "loss": 0.6192,
      "step": 3554
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3349988755821158e-05,
      "loss": 0.5962,
      "step": 3555
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3334792150501774e-05,
      "loss": 0.6288,
      "step": 3556
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3319601727449843e-05,
      "loss": 0.6113,
      "step": 3557
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3304417492300571e-05,
      "loss": 0.6498,
      "step": 3558
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3289239450686814e-05,
      "loss": 0.5813,
      "step": 3559
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3274067608239144e-05,
      "loss": 0.6116,
      "step": 3560
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3258901970585852e-05,
      "loss": 0.563,
      "step": 3561
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3243742543352906e-05,
      "loss": 0.5701,
      "step": 3562
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3228589332163971e-05,
      "loss": 0.5902,
      "step": 3563
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3213442342640407e-05,
      "loss": 0.5551,
      "step": 3564
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.319830158040129e-05,
      "loss": 0.6071,
      "step": 3565
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3183167051063353e-05,
      "loss": 0.5931,
      "step": 3566
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3168038760241032e-05,
      "loss": 0.5593,
      "step": 3567
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3152916713546433e-05,
      "loss": 0.5882,
      "step": 3568
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3137800916589385e-05,
      "loss": 0.5759,
      "step": 3569
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3122691374977357e-05,
      "loss": 0.6559,
      "step": 3570
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3107588094315518e-05,
      "loss": 0.5545,
      "step": 3571
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3092491080206696e-05,
      "loss": 0.6007,
      "step": 3572
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3077400338251431e-05,
      "loss": 0.6261,
      "step": 3573
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3062315874047902e-05,
      "loss": 0.6515,
      "step": 3574
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3047237693191971e-05,
      "loss": 0.6975,
      "step": 3575
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3032165801277159e-05,
      "loss": 0.581,
      "step": 3576
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.301710020389469e-05,
      "loss": 0.6248,
      "step": 3577
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.3002040906633389e-05,
      "loss": 0.6062,
      "step": 3578
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.298698791507981e-05,
      "loss": 0.6022,
      "step": 3579
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.297194123481812e-05,
      "loss": 0.6216,
      "step": 3580
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2956900871430193e-05,
      "loss": 0.5706,
      "step": 3581
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2941866830495494e-05,
      "loss": 0.613,
      "step": 3582
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.29268391175912e-05,
      "loss": 0.5545,
      "step": 3583
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2911817738292118e-05,
      "loss": 0.6109,
      "step": 3584
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2896802698170704e-05,
      "loss": 0.6183,
      "step": 3585
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2881794002797055e-05,
      "loss": 0.6032,
      "step": 3586
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2866791657738945e-05,
      "loss": 0.5208,
      "step": 3587
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2851795668561761e-05,
      "loss": 0.5462,
      "step": 3588
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2836806040828545e-05,
      "loss": 0.6395,
      "step": 3589
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2821822780099962e-05,
      "loss": 0.5436,
      "step": 3590
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2806845891934354e-05,
      "loss": 0.5503,
      "step": 3591
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2791875381887664e-05,
      "loss": 0.5321,
      "step": 3592
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2776911255513478e-05,
      "loss": 0.6581,
      "step": 3593
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2761953518363014e-05,
      "loss": 0.5771,
      "step": 3594
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.274700217598513e-05,
      "loss": 0.7166,
      "step": 3595
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2732057233926302e-05,
      "loss": 0.5576,
      "step": 3596
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.271711869773063e-05,
      "loss": 0.6107,
      "step": 3597
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2702186572939831e-05,
      "loss": 0.5712,
      "step": 3598
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2687260865093272e-05,
      "loss": 0.6404,
      "step": 3599
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.267234157972791e-05,
      "loss": 0.6291,
      "step": 3600
    },
    {
      "epoch": 2.75,
      "eval_loss": 0.9052521586418152,
      "eval_runtime": 14.3415,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 3600
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2657428722378334e-05,
      "loss": 0.6017,
      "step": 3601
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2642522298576742e-05,
      "loss": 0.5599,
      "step": 3602
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2627622313852935e-05,
      "loss": 0.6406,
      "step": 3603
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2612728773734367e-05,
      "loss": 0.6409,
      "step": 3604
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.2597841683746053e-05,
      "loss": 0.5935,
      "step": 3605
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.258296104941064e-05,
      "loss": 0.5246,
      "step": 3606
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2568086876248363e-05,
      "loss": 0.5919,
      "step": 3607
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2553219169777094e-05,
      "loss": 0.5394,
      "step": 3608
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2538357935512275e-05,
      "loss": 0.6268,
      "step": 3609
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.252350317896695e-05,
      "loss": 0.6874,
      "step": 3610
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2508654905651761e-05,
      "loss": 0.6049,
      "step": 3611
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2493813121074969e-05,
      "loss": 0.5941,
      "step": 3612
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2478977830742399e-05,
      "loss": 0.5641,
      "step": 3613
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2464149040157475e-05,
      "loss": 0.521,
      "step": 3614
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2449326754821206e-05,
      "loss": 0.5742,
      "step": 3615
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2434510980232217e-05,
      "loss": 0.5801,
      "step": 3616
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2419701721886658e-05,
      "loss": 0.5714,
      "step": 3617
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.240489898527833e-05,
      "loss": 0.5798,
      "step": 3618
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2390102775898553e-05,
      "loss": 0.5716,
      "step": 3619
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2375313099236293e-05,
      "loss": 0.6022,
      "step": 3620
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2360529960778012e-05,
      "loss": 0.5843,
      "step": 3621
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2345753366007823e-05,
      "loss": 0.5831,
      "step": 3622
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.233098332040735e-05,
      "loss": 0.5647,
      "step": 3623
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2316219829455851e-05,
      "loss": 0.6177,
      "step": 3624
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2301462898630071e-05,
      "loss": 0.7447,
      "step": 3625
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2286712533404402e-05,
      "loss": 0.5572,
      "step": 3626
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2271968739250752e-05,
      "loss": 0.5132,
      "step": 3627
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2257231521638607e-05,
      "loss": 0.6098,
      "step": 3628
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2242500886035e-05,
      "loss": 0.6158,
      "step": 3629
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2227776837904549e-05,
      "loss": 0.5807,
      "step": 3630
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.2213059382709407e-05,
      "loss": 0.6877,
      "step": 3631
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2198348525909289e-05,
      "loss": 0.6984,
      "step": 3632
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2183644272961441e-05,
      "loss": 0.5508,
      "step": 3633
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2168946629320703e-05,
      "loss": 0.6051,
      "step": 3634
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2154255600439432e-05,
      "loss": 0.5263,
      "step": 3635
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.213957119176753e-05,
      "loss": 0.6008,
      "step": 3636
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2124893408752446e-05,
      "loss": 0.5954,
      "step": 3637
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2110222256839192e-05,
      "loss": 0.628,
      "step": 3638
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2095557741470293e-05,
      "loss": 0.5883,
      "step": 3639
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2080899868085814e-05,
      "loss": 0.5936,
      "step": 3640
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2066248642123398e-05,
      "loss": 0.5739,
      "step": 3641
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2051604069018141e-05,
      "loss": 0.5981,
      "step": 3642
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2036966154202755e-05,
      "loss": 0.6666,
      "step": 3643
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.2022334903107424e-05,
      "loss": 0.5844,
      "step": 3644
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.2007710321159912e-05,
      "loss": 0.5977,
      "step": 3645
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1993092413785436e-05,
      "loss": 0.5888,
      "step": 3646
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.197848118640681e-05,
      "loss": 0.5912,
      "step": 3647
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1963876644444322e-05,
      "loss": 0.6142,
      "step": 3648
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1949278793315823e-05,
      "loss": 0.6093,
      "step": 3649
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1934687638436624e-05,
      "loss": 0.5679,
      "step": 3650
    },
    {
      "epoch": 2.79,
      "eval_loss": 0.9059380292892456,
      "eval_runtime": 14.3423,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 3650
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1920103185219605e-05,
      "loss": 0.5687,
      "step": 3651
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1905525439075134e-05,
      "loss": 0.5371,
      "step": 3652
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1890954405411095e-05,
      "loss": 0.5354,
      "step": 3653
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.187639008963287e-05,
      "loss": 0.6213,
      "step": 3654
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1861832497143388e-05,
      "loss": 0.5827,
      "step": 3655
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.184728163334304e-05,
      "loss": 0.577,
      "step": 3656
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1832737503629747e-05,
      "loss": 0.6303,
      "step": 3657
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1818200113398906e-05,
      "loss": 0.6221,
      "step": 3658
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1803669468043465e-05,
      "loss": 0.6339,
      "step": 3659
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1789145572953792e-05,
      "loss": 0.6036,
      "step": 3660
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.177462843351783e-05,
      "loss": 0.6029,
      "step": 3661
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1760118055120956e-05,
      "loss": 0.5578,
      "step": 3662
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.17456144431461e-05,
      "loss": 0.6308,
      "step": 3663
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1731117602973597e-05,
      "loss": 0.5752,
      "step": 3664
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1716627539981352e-05,
      "loss": 0.6848,
      "step": 3665
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1702144259544717e-05,
      "loss": 0.5306,
      "step": 3666
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1687667767036524e-05,
      "loss": 0.5727,
      "step": 3667
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1673198067827093e-05,
      "loss": 0.5628,
      "step": 3668
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1658735167284248e-05,
      "loss": 0.5469,
      "step": 3669
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1644279070773254e-05,
      "loss": 0.5808,
      "step": 3670
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1629829783656867e-05,
      "loss": 0.5938,
      "step": 3671
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1615387311295316e-05,
      "loss": 0.5555,
      "step": 3672
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1600951659046315e-05,
      "loss": 0.6515,
      "step": 3673
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1586522832265031e-05,
      "loss": 0.5515,
      "step": 3674
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1572100836304095e-05,
      "loss": 0.6039,
      "step": 3675
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1557685676513631e-05,
      "loss": 0.6676,
      "step": 3676
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1543277358241198e-05,
      "loss": 0.5765,
      "step": 3677
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1528875886831836e-05,
      "loss": 0.6707,
      "step": 3678
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1514481267628018e-05,
      "loss": 0.5824,
      "step": 3679
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.150009350596972e-05,
      "loss": 0.5383,
      "step": 3680
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.148571260719434e-05,
      "loss": 0.5431,
      "step": 3681
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1471338576636736e-05,
      "loss": 0.6214,
      "step": 3682
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.145697141962921e-05,
      "loss": 0.6609,
      "step": 3683
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1442611141501558e-05,
      "loss": 0.6077,
      "step": 3684
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1428257747580947e-05,
      "loss": 0.6439,
      "step": 3685
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1413911243192066e-05,
      "loss": 0.6439,
      "step": 3686
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1399571633657e-05,
      "loss": 0.5539,
      "step": 3687
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1385238924295316e-05,
      "loss": 0.6485,
      "step": 3688
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1370913120423967e-05,
      "loss": 0.5946,
      "step": 3689
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1356594227357395e-05,
      "loss": 0.5684,
      "step": 3690
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1342282250407456e-05,
      "loss": 0.6323,
      "step": 3691
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1327977194883439e-05,
      "loss": 0.6027,
      "step": 3692
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1313679066092064e-05,
      "loss": 0.5827,
      "step": 3693
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1299387869337503e-05,
      "loss": 0.5386,
      "step": 3694
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1285103609921328e-05,
      "loss": 0.6047,
      "step": 3695
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1270826293142553e-05,
      "loss": 0.6608,
      "step": 3696
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1256555924297602e-05,
      "loss": 0.639,
      "step": 3697
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1242292508680352e-05,
      "loss": 0.5559,
      "step": 3698
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1228036051582072e-05,
      "loss": 0.5678,
      "step": 3699
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.121378655829146e-05,
      "loss": 0.6411,
      "step": 3700
    },
    {
      "epoch": 2.83,
      "eval_loss": 0.906114935874939,
      "eval_runtime": 14.3464,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 3700
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.119954403409462e-05,
      "loss": 0.6612,
      "step": 3701
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1185308484275105e-05,
      "loss": 0.6263,
      "step": 3702
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.117107991411382e-05,
      "loss": 0.5612,
      "step": 3703
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1156858328889144e-05,
      "loss": 0.6053,
      "step": 3704
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1142643733876823e-05,
      "loss": 0.5707,
      "step": 3705
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1128436134350036e-05,
      "loss": 0.6268,
      "step": 3706
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1114235535579353e-05,
      "loss": 0.5895,
      "step": 3707
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.1100041942832744e-05,
      "loss": 0.6077,
      "step": 3708
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1085855361375584e-05,
      "loss": 0.6677,
      "step": 3709
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1071675796470645e-05,
      "loss": 0.5983,
      "step": 3710
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1057503253378112e-05,
      "loss": 0.6628,
      "step": 3711
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.104333773735555e-05,
      "loss": 0.6313,
      "step": 3712
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1029179253657908e-05,
      "loss": 0.6173,
      "step": 3713
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1015027807537538e-05,
      "loss": 0.5261,
      "step": 3714
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1000883404244197e-05,
      "loss": 0.5821,
      "step": 3715
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0986746049025007e-05,
      "loss": 0.5756,
      "step": 3716
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0972615747124476e-05,
      "loss": 0.6283,
      "step": 3717
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0958492503784494e-05,
      "loss": 0.587,
      "step": 3718
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0944376324244357e-05,
      "loss": 0.5634,
      "step": 3719
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.093026721374072e-05,
      "loss": 0.5603,
      "step": 3720
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0916165177507612e-05,
      "loss": 0.6072,
      "step": 3721
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0902070220776435e-05,
      "loss": 0.5906,
      "step": 3722
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0887982348776001e-05,
      "loss": 0.6014,
      "step": 3723
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0873901566732447e-05,
      "loss": 0.6379,
      "step": 3724
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0859827879869309e-05,
      "loss": 0.6358,
      "step": 3725
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0845761293407465e-05,
      "loss": 0.6473,
      "step": 3726
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0831701812565213e-05,
      "loss": 0.6799,
      "step": 3727
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.081764944255813e-05,
      "loss": 0.5858,
      "step": 3728
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0803604188599238e-05,
      "loss": 0.5516,
      "step": 3729
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0789566055898861e-05,
      "loss": 0.4996,
      "step": 3730
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0775535049664736e-05,
      "loss": 0.5279,
      "step": 3731
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0761511175101885e-05,
      "loss": 0.5545,
      "step": 3732
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.074749443741275e-05,
      "loss": 0.6114,
      "step": 3733
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.0733484841797097e-05,
      "loss": 0.6225,
      "step": 3734
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0719482393452039e-05,
      "loss": 0.6415,
      "step": 3735
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0705487097572035e-05,
      "loss": 0.6043,
      "step": 3736
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0691498959348917e-05,
      "loss": 0.5693,
      "step": 3737
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0677517983971838e-05,
      "loss": 0.5967,
      "step": 3738
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.06635441766273e-05,
      "loss": 0.5636,
      "step": 3739
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0649577542499132e-05,
      "loss": 0.6136,
      "step": 3740
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0635618086768535e-05,
      "loss": 0.5421,
      "step": 3741
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0621665814614023e-05,
      "loss": 0.5376,
      "step": 3742
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0607720731211447e-05,
      "loss": 0.6148,
      "step": 3743
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.059378284173398e-05,
      "loss": 0.5588,
      "step": 3744
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0579852151352166e-05,
      "loss": 0.5451,
      "step": 3745
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0565928665233839e-05,
      "loss": 0.5661,
      "step": 3746
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0552012388544178e-05,
      "loss": 0.5686,
      "step": 3747
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0538103326445668e-05,
      "loss": 0.5931,
      "step": 3748
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0524201484098157e-05,
      "loss": 0.63,
      "step": 3749
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0510306866658778e-05,
      "loss": 0.5585,
      "step": 3750
    },
    {
      "epoch": 2.87,
      "eval_loss": 0.9043093919754028,
      "eval_runtime": 14.3421,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 3750
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0496419479282e-05,
      "loss": 0.5921,
      "step": 3751
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0482539327119606e-05,
      "loss": 0.583,
      "step": 3752
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0468666415320683e-05,
      "loss": 0.5935,
      "step": 3753
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.045480074903167e-05,
      "loss": 0.5602,
      "step": 3754
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0440942333396264e-05,
      "loss": 0.5619,
      "step": 3755
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0427091173555536e-05,
      "loss": 0.5579,
      "step": 3756
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0413247274647793e-05,
      "loss": 0.6381,
      "step": 3757
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.039941064180871e-05,
      "loss": 0.5492,
      "step": 3758
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0385581280171235e-05,
      "loss": 0.6444,
      "step": 3759
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0371759194865627e-05,
      "loss": 0.5802,
      "step": 3760
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0357944391019434e-05,
      "loss": 0.5878,
      "step": 3761
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0344136873757531e-05,
      "loss": 0.5925,
      "step": 3762
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0330336648202063e-05,
      "loss": 0.6004,
      "step": 3763
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0316543719472482e-05,
      "loss": 0.5456,
      "step": 3764
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0302758092685519e-05,
      "loss": 0.6939,
      "step": 3765
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0288979772955234e-05,
      "loss": 0.579,
      "step": 3766
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0275208765392915e-05,
      "loss": 0.6215,
      "step": 3767
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0261445075107199e-05,
      "loss": 0.6072,
      "step": 3768
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0247688707203961e-05,
      "loss": 0.5236,
      "step": 3769
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0233939666786413e-05,
      "loss": 0.6105,
      "step": 3770
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0220197958954976e-05,
      "loss": 0.5301,
      "step": 3771
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0206463588807422e-05,
      "loss": 0.5763,
      "step": 3772
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0192736561438745e-05,
      "loss": 0.636,
      "step": 3773
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0179016881941276e-05,
      "loss": 0.5907,
      "step": 3774
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0165304555404542e-05,
      "loss": 0.568,
      "step": 3775
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0151599586915413e-05,
      "loss": 0.5842,
      "step": 3776
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0137901981557989e-05,
      "loss": 0.5999,
      "step": 3777
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0124211744413658e-05,
      "loss": 0.5907,
      "step": 3778
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0110528880561051e-05,
      "loss": 0.6715,
      "step": 3779
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.00968533950761e-05,
      "loss": 0.6059,
      "step": 3780
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.008318529303197e-05,
      "loss": 0.6302,
      "step": 3781
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.00695245794991e-05,
      "loss": 0.5651,
      "step": 3782
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0055871259545168e-05,
      "loss": 0.5258,
      "step": 3783
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0042225338235152e-05,
      "loss": 0.568,
      "step": 3784
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0028586820631246e-05,
      "loss": 0.6037,
      "step": 3785
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0014955711792914e-05,
      "loss": 0.5881,
      "step": 3786
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.0001332016776852e-05,
      "loss": 0.5571,
      "step": 3787
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.987715740637048e-06,
      "loss": 0.594,
      "step": 3788
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.974106888424703e-06,
      "loss": 0.5829,
      "step": 3789
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.960505465188254e-06,
      "loss": 0.5518,
      "step": 3790
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.946911475973439e-06,
      "loss": 0.6049,
      "step": 3791
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.933324925823157e-06,
      "loss": 0.609,
      "step": 3792
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.91974581977762e-06,
      "loss": 0.5701,
      "step": 3793
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.906174162874228e-06,
      "loss": 0.562,
      "step": 3794
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.892609960147674e-06,
      "loss": 0.6462,
      "step": 3795
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.879053216629801e-06,
      "loss": 0.5509,
      "step": 3796
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.865503937349772e-06,
      "loss": 0.5591,
      "step": 3797
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.851962127333922e-06,
      "loss": 0.6325,
      "step": 3798
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.838427791605862e-06,
      "loss": 0.5544,
      "step": 3799
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.824900935186368e-06,
      "loss": 0.6516,
      "step": 3800
    },
    {
      "epoch": 2.91,
      "eval_loss": 0.9059207439422607,
      "eval_runtime": 14.3535,
      "eval_samples_per_second": 5.713,
      "eval_steps_per_second": 5.713,
      "step": 3800
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.811381563093507e-06,
      "loss": 0.5745,
      "step": 3801
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.797869680342532e-06,
      "loss": 0.5914,
      "step": 3802
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.784365291945924e-06,
      "loss": 0.5786,
      "step": 3803
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.770868402913372e-06,
      "loss": 0.5796,
      "step": 3804
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.757379018251825e-06,
      "loss": 0.5948,
      "step": 3805
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.743897142965408e-06,
      "loss": 0.6105,
      "step": 3806
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.730422782055469e-06,
      "loss": 0.6793,
      "step": 3807
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.716955940520564e-06,
      "loss": 0.665,
      "step": 3808
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.703496623356496e-06,
      "loss": 0.5828,
      "step": 3809
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.690044835556207e-06,
      "loss": 0.5723,
      "step": 3810
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.676600582109922e-06,
      "loss": 0.6449,
      "step": 3811
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.66316386800501e-06,
      "loss": 0.6854,
      "step": 3812
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.6497346982261e-06,
      "loss": 0.614,
      "step": 3813
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.636313077754957e-06,
      "loss": 0.5608,
      "step": 3814
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.6228990115706e-06,
      "loss": 0.6233,
      "step": 3815
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.609492504649224e-06,
      "loss": 0.5902,
      "step": 3816
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.596093561964216e-06,
      "loss": 0.6262,
      "step": 3817
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.582702188486155e-06,
      "loss": 0.5959,
      "step": 3818
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.569318389182834e-06,
      "loss": 0.5693,
      "step": 3819
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.555942169019217e-06,
      "loss": 0.5494,
      "step": 3820
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.542573532957445e-06,
      "loss": 0.5787,
      "step": 3821
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.529212485956886e-06,
      "loss": 0.5736,
      "step": 3822
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.515859032974059e-06,
      "loss": 0.5702,
      "step": 3823
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.502513178962668e-06,
      "loss": 0.5456,
      "step": 3824
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.4891749288736e-06,
      "loss": 0.672,
      "step": 3825
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.475844287654944e-06,
      "loss": 0.5642,
      "step": 3826
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.462521260251939e-06,
      "loss": 0.5828,
      "step": 3827
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.44920585160701e-06,
      "loss": 0.5591,
      "step": 3828
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.435898066659738e-06,
      "loss": 0.5886,
      "step": 3829
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.422597910346922e-06,
      "loss": 0.5989,
      "step": 3830
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.409305387602482e-06,
      "loss": 0.6083,
      "step": 3831
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.396020503357529e-06,
      "loss": 0.5366,
      "step": 3832
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.382743262540334e-06,
      "loss": 0.6115,
      "step": 3833
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.369473670076353e-06,
      "loss": 0.576,
      "step": 3834
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.356211730888154e-06,
      "loss": 0.6165,
      "step": 3835
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.34295744989553e-06,
      "loss": 0.5958,
      "step": 3836
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.32971083201538e-06,
      "loss": 0.605,
      "step": 3837
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.316471882161813e-06,
      "loss": 0.627,
      "step": 3838
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.303240605246019e-06,
      "loss": 0.5733,
      "step": 3839
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.290017006176416e-06,
      "loss": 0.6037,
      "step": 3840
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.27680108985854e-06,
      "loss": 0.6875,
      "step": 3841
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.263592861195072e-06,
      "loss": 0.6029,
      "step": 3842
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.250392325085843e-06,
      "loss": 0.559,
      "step": 3843
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.237199486427854e-06,
      "loss": 0.6345,
      "step": 3844
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.224014350115226e-06,
      "loss": 0.6182,
      "step": 3845
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.210836921039226e-06,
      "loss": 0.5541,
      "step": 3846
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.197667204088258e-06,
      "loss": 0.6198,
      "step": 3847
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.18450520414789e-06,
      "loss": 0.5219,
      "step": 3848
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.171350926100801e-06,
      "loss": 0.6469,
      "step": 3849
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.158204374826817e-06,
      "loss": 0.5871,
      "step": 3850
    },
    {
      "epoch": 2.94,
      "eval_loss": 0.9054226279258728,
      "eval_runtime": 14.3461,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 3850
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.145065555202878e-06,
      "loss": 0.6075,
      "step": 3851
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.131934472103105e-06,
      "loss": 0.6292,
      "step": 3852
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.118811130398679e-06,
      "loss": 0.5667,
      "step": 3853
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.105695534957979e-06,
      "loss": 0.5684,
      "step": 3854
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.09258769064645e-06,
      "loss": 0.6368,
      "step": 3855
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.079487602326716e-06,
      "loss": 0.5738,
      "step": 3856
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.066395274858483e-06,
      "loss": 0.6018,
      "step": 3857
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.053310713098596e-06,
      "loss": 0.5888,
      "step": 3858
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.04023392190101e-06,
      "loss": 0.589,
      "step": 3859
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.027164906116801e-06,
      "loss": 0.5725,
      "step": 3860
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.014103670594179e-06,
      "loss": 0.5687,
      "step": 3861
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.001050220178438e-06,
      "loss": 0.6357,
      "step": 3862
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.988004559712001e-06,
      "loss": 0.567,
      "step": 3863
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.974966694034385e-06,
      "loss": 0.5838,
      "step": 3864
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.96193662798225e-06,
      "loss": 0.5474,
      "step": 3865
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.948914366389327e-06,
      "loss": 0.5768,
      "step": 3866
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.935899914086468e-06,
      "loss": 0.6574,
      "step": 3867
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.922893275901614e-06,
      "loss": 0.5629,
      "step": 3868
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.90989445665984e-06,
      "loss": 0.6632,
      "step": 3869
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.896903461183286e-06,
      "loss": 0.5861,
      "step": 3870
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.883920294291204e-06,
      "loss": 0.5676,
      "step": 3871
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.870944960799928e-06,
      "loss": 0.6304,
      "step": 3872
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.857977465522922e-06,
      "loss": 0.5878,
      "step": 3873
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.845017813270708e-06,
      "loss": 0.6011,
      "step": 3874
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.832066008850905e-06,
      "loss": 0.5809,
      "step": 3875
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.819122057068218e-06,
      "loss": 0.6539,
      "step": 3876
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.806185962724473e-06,
      "loss": 0.6274,
      "step": 3877
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.793257730618518e-06,
      "loss": 0.6392,
      "step": 3878
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.780337365546348e-06,
      "loss": 0.5652,
      "step": 3879
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.76742487230099e-06,
      "loss": 0.6125,
      "step": 3880
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.754520255672604e-06,
      "loss": 0.6336,
      "step": 3881
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.741623520448357e-06,
      "loss": 0.5429,
      "step": 3882
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.728734671412561e-06,
      "loss": 0.5838,
      "step": 3883
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.715853713346563e-06,
      "loss": 0.5835,
      "step": 3884
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.702980651028796e-06,
      "loss": 0.5584,
      "step": 3885
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.690115489234744e-06,
      "loss": 0.5617,
      "step": 3886
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.677258232737004e-06,
      "loss": 0.5369,
      "step": 3887
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.664408886305203e-06,
      "loss": 0.5444,
      "step": 3888
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.651567454706038e-06,
      "loss": 0.6399,
      "step": 3889
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.63873394270327e-06,
      "loss": 0.7163,
      "step": 3890
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.625908355057748e-06,
      "loss": 0.5222,
      "step": 3891
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.613090696527352e-06,
      "loss": 0.6317,
      "step": 3892
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.600280971867028e-06,
      "loss": 0.5712,
      "step": 3893
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.58747918582877e-06,
      "loss": 0.6105,
      "step": 3894
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.57468534316166e-06,
      "loss": 0.5155,
      "step": 3895
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.5618994486118e-06,
      "loss": 0.607,
      "step": 3896
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.549121506922351e-06,
      "loss": 0.5613,
      "step": 3897
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.536351522833524e-06,
      "loss": 0.6171,
      "step": 3898
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.523589501082597e-06,
      "loss": 0.5501,
      "step": 3899
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.510835446403867e-06,
      "loss": 0.6217,
      "step": 3900
    },
    {
      "epoch": 2.98,
      "eval_loss": 0.9043266773223877,
      "eval_runtime": 14.3172,
      "eval_samples_per_second": 5.727,
      "eval_steps_per_second": 5.727,
      "step": 3900
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.498089363528688e-06,
      "loss": 0.5806,
      "step": 3901
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.485351257185457e-06,
      "loss": 0.6112,
      "step": 3902
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.472621132099598e-06,
      "loss": 0.5947,
      "step": 3903
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.459898992993608e-06,
      "loss": 0.7329,
      "step": 3904
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.44718484458698e-06,
      "loss": 0.519,
      "step": 3905
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.434478691596293e-06,
      "loss": 0.6889,
      "step": 3906
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.42178053873509e-06,
      "loss": 0.5459,
      "step": 3907
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.409090390714013e-06,
      "loss": 0.6118,
      "step": 3908
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.396408252240704e-06,
      "loss": 0.5441,
      "step": 3909
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.383734128019832e-06,
      "loss": 0.5246,
      "step": 3910
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.371068022753088e-06,
      "loss": 0.6156,
      "step": 3911
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.35840994113922e-06,
      "loss": 0.6968,
      "step": 3912
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.345759887873966e-06,
      "loss": 0.5744,
      "step": 3913
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.333117867650102e-06,
      "loss": 0.5673,
      "step": 3914
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.320483885157403e-06,
      "loss": 0.5699,
      "step": 3915
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.307857945082703e-06,
      "loss": 0.6364,
      "step": 3916
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.295240052109817e-06,
      "loss": 0.5667,
      "step": 3917
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.282630210919587e-06,
      "loss": 0.6182,
      "step": 3918
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.270028426189857e-06,
      "loss": 0.59,
      "step": 3919
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.25743470259552e-06,
      "loss": 0.5168,
      "step": 3920
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.244849044808417e-06,
      "loss": 0.58,
      "step": 3921
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.232271457497453e-06,
      "loss": 0.6235,
      "step": 3922
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.219701945328503e-06,
      "loss": 0.7033,
      "step": 3923
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.207140512964487e-06,
      "loss": 0.5829,
      "step": 3924
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.194587165065263e-06,
      "loss": 0.6597,
      "step": 3925
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.182041906287755e-06,
      "loss": 0.6569,
      "step": 3926
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.169504741285854e-06,
      "loss": 0.5495,
      "step": 3927
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.156975674710449e-06,
      "loss": 0.6433,
      "step": 3928
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.144454711209417e-06,
      "loss": 0.5397,
      "step": 3929
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.13194185542767e-06,
      "loss": 0.5541,
      "step": 3930
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.119437112007065e-06,
      "loss": 0.5801,
      "step": 3931
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.106940485586477e-06,
      "loss": 0.5809,
      "step": 3932
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.09445198080174e-06,
      "loss": 0.6092,
      "step": 3933
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.081971602285725e-06,
      "loss": 0.6295,
      "step": 3934
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.069499354668247e-06,
      "loss": 0.5677,
      "step": 3935
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.057035242576116e-06,
      "loss": 0.7043,
      "step": 3936
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.044579270633115e-06,
      "loss": 0.5965,
      "step": 3937
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.032131443460038e-06,
      "loss": 0.5673,
      "step": 3938
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.019691765674629e-06,
      "loss": 0.6769,
      "step": 3939
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.007260241891604e-06,
      "loss": 0.5728,
      "step": 3940
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.99483687672269e-06,
      "loss": 0.6031,
      "step": 3941
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.98242167477655e-06,
      "loss": 0.5419,
      "step": 3942
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.970014640658836e-06,
      "loss": 0.6061,
      "step": 3943
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.957615778972158e-06,
      "loss": 0.5887,
      "step": 3944
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.945225094316131e-06,
      "loss": 0.6607,
      "step": 3945
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.932842591287271e-06,
      "loss": 0.5748,
      "step": 3946
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.920468274479126e-06,
      "loss": 0.5645,
      "step": 3947
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.908102148482159e-06,
      "loss": 0.596,
      "step": 3948
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.895744217883838e-06,
      "loss": 0.565,
      "step": 3949
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.883394487268535e-06,
      "loss": 0.5602,
      "step": 3950
    },
    {
      "epoch": 3.02,
      "eval_loss": 0.9035821557044983,
      "eval_runtime": 14.3463,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 3950
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.871052961217634e-06,
      "loss": 0.5245,
      "step": 3951
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.858719644309445e-06,
      "loss": 0.5808,
      "step": 3952
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.846394541119243e-06,
      "loss": 0.5081,
      "step": 3953
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.834077656219239e-06,
      "loss": 0.5423,
      "step": 3954
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.821768994178633e-06,
      "loss": 0.5067,
      "step": 3955
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.809468559563539e-06,
      "loss": 0.5306,
      "step": 3956
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.797176356937031e-06,
      "loss": 0.5312,
      "step": 3957
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.78489239085912e-06,
      "loss": 0.5373,
      "step": 3958
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.7726166658868e-06,
      "loss": 0.5267,
      "step": 3959
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.760349186573941e-06,
      "loss": 0.4885,
      "step": 3960
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.748089957471418e-06,
      "loss": 0.5107,
      "step": 3961
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.735838983127003e-06,
      "loss": 0.5501,
      "step": 3962
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.72359626808545e-06,
      "loss": 0.5287,
      "step": 3963
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.711361816888377e-06,
      "loss": 0.4511,
      "step": 3964
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.699135634074414e-06,
      "loss": 0.5497,
      "step": 3965
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.686917724179074e-06,
      "loss": 0.508,
      "step": 3966
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.674708091734839e-06,
      "loss": 0.4853,
      "step": 3967
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.66250674127106e-06,
      "loss": 0.5193,
      "step": 3968
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.65031367731409e-06,
      "loss": 0.5203,
      "step": 3969
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.638128904387153e-06,
      "loss": 0.5481,
      "step": 3970
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.625952427010415e-06,
      "loss": 0.4867,
      "step": 3971
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.613784249700981e-06,
      "loss": 0.4971,
      "step": 3972
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.60162437697285e-06,
      "loss": 0.5488,
      "step": 3973
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.589472813336956e-06,
      "loss": 0.4799,
      "step": 3974
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.577329563301134e-06,
      "loss": 0.5236,
      "step": 3975
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.56519463137017e-06,
      "loss": 0.5887,
      "step": 3976
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.5530680220457304e-06,
      "loss": 0.5515,
      "step": 3977
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.540949739826403e-06,
      "loss": 0.5423,
      "step": 3978
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.528839789207681e-06,
      "loss": 0.4713,
      "step": 3979
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.5167381746819955e-06,
      "loss": 0.5537,
      "step": 3980
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.504644900738654e-06,
      "loss": 0.5641,
      "step": 3981
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.492559971863883e-06,
      "loss": 0.4971,
      "step": 3982
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.480483392540795e-06,
      "loss": 0.4687,
      "step": 3983
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.468415167249453e-06,
      "loss": 0.4512,
      "step": 3984
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.456355300466753e-06,
      "loss": 0.5362,
      "step": 3985
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.444303796666553e-06,
      "loss": 0.472,
      "step": 3986
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.43226066031956e-06,
      "loss": 0.5349,
      "step": 3987
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.420225895893427e-06,
      "loss": 0.632,
      "step": 3988
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.408199507852636e-06,
      "loss": 0.5054,
      "step": 3989
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.396181500658627e-06,
      "loss": 0.5216,
      "step": 3990
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.384171878769686e-06,
      "loss": 0.5003,
      "step": 3991
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.372170646641025e-06,
      "loss": 0.4898,
      "step": 3992
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.360177808724696e-06,
      "loss": 0.4702,
      "step": 3993
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.348193369469687e-06,
      "loss": 0.51,
      "step": 3994
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.336217333321837e-06,
      "loss": 0.504,
      "step": 3995
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.324249704723881e-06,
      "loss": 0.5065,
      "step": 3996
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.3122904881154265e-06,
      "loss": 0.4929,
      "step": 3997
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.3003396879329745e-06,
      "loss": 0.5536,
      "step": 3998
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.2883973086098954e-06,
      "loss": 0.4505,
      "step": 3999
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.2764633545764326e-06,
      "loss": 0.5578,
      "step": 4000
    },
    {
      "epoch": 3.04,
      "eval_loss": 0.9381759762763977,
      "eval_runtime": 14.3513,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 4000
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.264537830259701e-06,
      "loss": 0.4959,
      "step": 4001
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.252620740083718e-06,
      "loss": 0.4544,
      "step": 4002
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.240712088469315e-06,
      "loss": 0.4971,
      "step": 4003
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.228811879834252e-06,
      "loss": 0.5625,
      "step": 4004
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.2169201185931185e-06,
      "loss": 0.5095,
      "step": 4005
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.2050368091573945e-06,
      "loss": 0.5144,
      "step": 4006
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.19316195593541e-06,
      "loss": 0.5595,
      "step": 4007
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.181295563332361e-06,
      "loss": 0.5293,
      "step": 4008
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.169437635750306e-06,
      "loss": 0.5166,
      "step": 4009
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.157588177588149e-06,
      "loss": 0.574,
      "step": 4010
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.145747193241692e-06,
      "loss": 0.4386,
      "step": 4011
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.13391468710355e-06,
      "loss": 0.4771,
      "step": 4012
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.122090663563216e-06,
      "loss": 0.4881,
      "step": 4013
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.11027512700702e-06,
      "loss": 0.4499,
      "step": 4014
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.098468081818173e-06,
      "loss": 0.5628,
      "step": 4015
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.086669532376706e-06,
      "loss": 0.4228,
      "step": 4016
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.07487948305951e-06,
      "loss": 0.4802,
      "step": 4017
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.063097938240313e-06,
      "loss": 0.5422,
      "step": 4018
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.051324902289715e-06,
      "loss": 0.5459,
      "step": 4019
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.039560379575134e-06,
      "loss": 0.549,
      "step": 4020
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.0278043744608374e-06,
      "loss": 0.5366,
      "step": 4021
    },
    {
      "epoch": 3.06,
      "learning_rate": 7.0160568913079245e-06,
      "loss": 0.5046,
      "step": 4022
    },
    {
      "epoch": 3.06,
      "learning_rate": 7.004317934474359e-06,
      "loss": 0.5232,
      "step": 4023
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.992587508314917e-06,
      "loss": 0.5323,
      "step": 4024
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.980865617181211e-06,
      "loss": 0.4624,
      "step": 4025
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.969152265421693e-06,
      "loss": 0.5528,
      "step": 4026
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.957447457381667e-06,
      "loss": 0.5099,
      "step": 4027
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.945751197403219e-06,
      "loss": 0.4854,
      "step": 4028
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.934063489825316e-06,
      "loss": 0.4918,
      "step": 4029
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.922384338983716e-06,
      "loss": 0.5288,
      "step": 4030
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.910713749211036e-06,
      "loss": 0.5227,
      "step": 4031
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.89905172483667e-06,
      "loss": 0.5143,
      "step": 4032
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.887398270186881e-06,
      "loss": 0.5992,
      "step": 4033
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.87575338958473e-06,
      "loss": 0.5098,
      "step": 4034
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.864117087350099e-06,
      "loss": 0.5214,
      "step": 4035
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.852489367799685e-06,
      "loss": 0.5205,
      "step": 4036
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.8408702352470165e-06,
      "loss": 0.5646,
      "step": 4037
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.82925969400242e-06,
      "loss": 0.475,
      "step": 4038
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.817657748373043e-06,
      "loss": 0.4389,
      "step": 4039
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.80606440266283e-06,
      "loss": 0.5151,
      "step": 4040
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.794479661172569e-06,
      "loss": 0.4867,
      "step": 4041
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.782903528199821e-06,
      "loss": 0.5579,
      "step": 4042
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.7713360080389695e-06,
      "loss": 0.4835,
      "step": 4043
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.759777104981193e-06,
      "loss": 0.5575,
      "step": 4044
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.748226823314494e-06,
      "loss": 0.471,
      "step": 4045
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.736685167323658e-06,
      "loss": 0.5656,
      "step": 4046
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.725152141290275e-06,
      "loss": 0.5214,
      "step": 4047
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.713627749492728e-06,
      "loss": 0.5121,
      "step": 4048
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.7021119962062174e-06,
      "loss": 0.5731,
      "step": 4049
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.690604885702716e-06,
      "loss": 0.497,
      "step": 4050
    },
    {
      "epoch": 3.08,
      "eval_loss": 0.9387401342391968,
      "eval_runtime": 14.344,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 4050
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.679106422251002e-06,
      "loss": 0.4846,
      "step": 4051
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.667616610116644e-06,
      "loss": 0.4672,
      "step": 4052
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.65613545356199e-06,
      "loss": 0.5067,
      "step": 4053
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.644662956846208e-06,
      "loss": 0.4681,
      "step": 4054
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.633199124225211e-06,
      "loss": 0.5084,
      "step": 4055
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.621743959951747e-06,
      "loss": 0.5475,
      "step": 4056
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.6102974682752906e-06,
      "loss": 0.5199,
      "step": 4057
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.598859653442154e-06,
      "loss": 0.5422,
      "step": 4058
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.587430519695395e-06,
      "loss": 0.5137,
      "step": 4059
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.5760100712748664e-06,
      "loss": 0.5315,
      "step": 4060
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.564598312417184e-06,
      "loss": 0.51,
      "step": 4061
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.553195247355768e-06,
      "loss": 0.4758,
      "step": 4062
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.5418008803207904e-06,
      "loss": 0.497,
      "step": 4063
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.530415215539202e-06,
      "loss": 0.5016,
      "step": 4064
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.519038257234715e-06,
      "loss": 0.4938,
      "step": 4065
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.5076700096278415e-06,
      "loss": 0.5128,
      "step": 4066
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.496310476935835e-06,
      "loss": 0.5107,
      "step": 4067
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.484959663372724e-06,
      "loss": 0.5294,
      "step": 4068
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.473617573149293e-06,
      "loss": 0.4878,
      "step": 4069
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.462284210473131e-06,
      "loss": 0.5281,
      "step": 4070
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.45095957954852e-06,
      "loss": 0.6168,
      "step": 4071
    },
    {
      "epoch": 3.09,
      "learning_rate": 6.439643684576571e-06,
      "loss": 0.5675,
      "step": 4072
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.428336529755104e-06,
      "loss": 0.5177,
      "step": 4073
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.417038119278743e-06,
      "loss": 0.4613,
      "step": 4074
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.4057484573388134e-06,
      "loss": 0.5663,
      "step": 4075
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.394467548123442e-06,
      "loss": 0.5379,
      "step": 4076
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.383195395817492e-06,
      "loss": 0.5132,
      "step": 4077
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.371932004602569e-06,
      "loss": 0.4835,
      "step": 4078
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.360677378657031e-06,
      "loss": 0.5682,
      "step": 4079
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.3494315221560086e-06,
      "loss": 0.5286,
      "step": 4080
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.338194439271348e-06,
      "loss": 0.5131,
      "step": 4081
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.326966134171658e-06,
      "loss": 0.473,
      "step": 4082
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.315746611022274e-06,
      "loss": 0.4894,
      "step": 4083
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.304535873985307e-06,
      "loss": 0.5091,
      "step": 4084
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.293333927219576e-06,
      "loss": 0.4924,
      "step": 4085
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.28214077488064e-06,
      "loss": 0.5076,
      "step": 4086
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.270956421120827e-06,
      "loss": 0.5344,
      "step": 4087
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.25978087008917e-06,
      "loss": 0.5017,
      "step": 4088
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.248614125931447e-06,
      "loss": 0.4968,
      "step": 4089
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.237456192790159e-06,
      "loss": 0.5414,
      "step": 4090
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.2263070748045644e-06,
      "loss": 0.5069,
      "step": 4091
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.21516677611063e-06,
      "loss": 0.5109,
      "step": 4092
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.204035300841056e-06,
      "loss": 0.5229,
      "step": 4093
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.192912653125258e-06,
      "loss": 0.4866,
      "step": 4094
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.1817988370894125e-06,
      "loss": 0.4266,
      "step": 4095
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.1706938568563645e-06,
      "loss": 0.5768,
      "step": 4096
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.159597716545733e-06,
      "loss": 0.5423,
      "step": 4097
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.148510420273826e-06,
      "loss": 0.494,
      "step": 4098
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.1374319721537045e-06,
      "loss": 0.4749,
      "step": 4099
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.1263623762950896e-06,
      "loss": 0.4496,
      "step": 4100
    },
    {
      "epoch": 3.12,
      "eval_loss": 0.9422940015792847,
      "eval_runtime": 14.3586,
      "eval_samples_per_second": 5.711,
      "eval_steps_per_second": 5.711,
      "step": 4100
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.1153016368044765e-06,
      "loss": 0.4763,
      "step": 4101
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.1042497577850444e-06,
      "loss": 0.4661,
      "step": 4102
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.093206743336692e-06,
      "loss": 0.4523,
      "step": 4103
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.082172597556024e-06,
      "loss": 0.502,
      "step": 4104
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.071147324536375e-06,
      "loss": 0.5289,
      "step": 4105
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.060130928367767e-06,
      "loss": 0.5435,
      "step": 4106
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.049123413136937e-06,
      "loss": 0.571,
      "step": 4107
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.038124782927319e-06,
      "loss": 0.5045,
      "step": 4108
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.02713504181908e-06,
      "loss": 0.5002,
      "step": 4109
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.016154193889034e-06,
      "loss": 0.5203,
      "step": 4110
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.00518224321076e-06,
      "loss": 0.5349,
      "step": 4111
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.99421919385449e-06,
      "loss": 0.4898,
      "step": 4112
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.983265049887189e-06,
      "loss": 0.5449,
      "step": 4113
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.97231981537247e-06,
      "loss": 0.5191,
      "step": 4114
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.961383494370695e-06,
      "loss": 0.5233,
      "step": 4115
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.950456090938877e-06,
      "loss": 0.5785,
      "step": 4116
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.939537609130765e-06,
      "loss": 0.5168,
      "step": 4117
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.928628052996743e-06,
      "loss": 0.5232,
      "step": 4118
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.9177274265839296e-06,
      "loss": 0.5422,
      "step": 4119
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.906835733936108e-06,
      "loss": 0.5077,
      "step": 4120
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.8959529790937485e-06,
      "loss": 0.5103,
      "step": 4121
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.885079166094025e-06,
      "loss": 0.467,
      "step": 4122
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.874214298970771e-06,
      "loss": 0.6323,
      "step": 4123
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.863358381754513e-06,
      "loss": 0.5215,
      "step": 4124
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.852511418472443e-06,
      "loss": 0.4783,
      "step": 4125
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.841673413148461e-06,
      "loss": 0.5032,
      "step": 4126
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.830844369803115e-06,
      "loss": 0.5011,
      "step": 4127
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.820024292453641e-06,
      "loss": 0.4567,
      "step": 4128
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.809213185113942e-06,
      "loss": 0.4952,
      "step": 4129
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.79841105179461e-06,
      "loss": 0.4838,
      "step": 4130
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.787617896502889e-06,
      "loss": 0.472,
      "step": 4131
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.776833723242704e-06,
      "loss": 0.5796,
      "step": 4132
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.76605853601463e-06,
      "loss": 0.4838,
      "step": 4133
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.755292338815951e-06,
      "loss": 0.5112,
      "step": 4134
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.744535135640551e-06,
      "loss": 0.499,
      "step": 4135
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.733786930479041e-06,
      "loss": 0.5057,
      "step": 4136
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.723047727318653e-06,
      "loss": 0.5464,
      "step": 4137
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.71231753014331e-06,
      "loss": 0.5288,
      "step": 4138
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.701596342933549e-06,
      "loss": 0.5125,
      "step": 4139
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.690884169666621e-06,
      "loss": 0.509,
      "step": 4140
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.680181014316383e-06,
      "loss": 0.5123,
      "step": 4141
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.669486880853396e-06,
      "loss": 0.5126,
      "step": 4142
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.658801773244812e-06,
      "loss": 0.4925,
      "step": 4143
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.648125695454499e-06,
      "loss": 0.4987,
      "step": 4144
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.63745865144293e-06,
      "loss": 0.4131,
      "step": 4145
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.626800645167248e-06,
      "loss": 0.5543,
      "step": 4146
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.6161516805812295e-06,
      "loss": 0.5329,
      "step": 4147
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.605511761635321e-06,
      "loss": 0.5664,
      "step": 4148
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.5948808922765894e-06,
      "loss": 0.5939,
      "step": 4149
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.584259076448749e-06,
      "loss": 0.5439,
      "step": 4150
    },
    {
      "epoch": 3.15,
      "eval_loss": 0.94086092710495,
      "eval_runtime": 14.3496,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 4150
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.573646318092155e-06,
      "loss": 0.5169,
      "step": 4151
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.563042621143821e-06,
      "loss": 0.4655,
      "step": 4152
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.552447989537378e-06,
      "loss": 0.4452,
      "step": 4153
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.541862427203101e-06,
      "loss": 0.5389,
      "step": 4154
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.53128593806789e-06,
      "loss": 0.5271,
      "step": 4155
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.5207185260553096e-06,
      "loss": 0.4631,
      "step": 4156
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.510160195085523e-06,
      "loss": 0.4633,
      "step": 4157
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.499610949075344e-06,
      "loss": 0.4656,
      "step": 4158
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.489070791938207e-06,
      "loss": 0.4656,
      "step": 4159
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.478539727584176e-06,
      "loss": 0.4684,
      "step": 4160
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.468017759919952e-06,
      "loss": 0.4781,
      "step": 4161
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.457504892848852e-06,
      "loss": 0.5559,
      "step": 4162
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.447001130270817e-06,
      "loss": 0.4065,
      "step": 4163
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.436506476082404e-06,
      "loss": 0.4903,
      "step": 4164
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.4260209341768174e-06,
      "loss": 0.5397,
      "step": 4165
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.415544508443854e-06,
      "loss": 0.5345,
      "step": 4166
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.4050772027699346e-06,
      "loss": 0.5459,
      "step": 4167
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.394619021038097e-06,
      "loss": 0.5731,
      "step": 4168
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.384169967128009e-06,
      "loss": 0.5271,
      "step": 4169
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.373730044915937e-06,
      "loss": 0.4676,
      "step": 4170
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.36329925827476e-06,
      "loss": 0.5555,
      "step": 4171
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.3528776110739665e-06,
      "loss": 0.5498,
      "step": 4172
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.342465107179672e-06,
      "loss": 0.6023,
      "step": 4173
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.332061750454582e-06,
      "loss": 0.5213,
      "step": 4174
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.321667544758014e-06,
      "loss": 0.498,
      "step": 4175
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.311282493945884e-06,
      "loss": 0.4768,
      "step": 4176
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.300906601870742e-06,
      "loss": 0.5262,
      "step": 4177
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.2905398723816834e-06,
      "loss": 0.5067,
      "step": 4178
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.280182309324465e-06,
      "loss": 0.5386,
      "step": 4179
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.2698339165414e-06,
      "loss": 0.5236,
      "step": 4180
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.259494697871441e-06,
      "loss": 0.5112,
      "step": 4181
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.2491646571500795e-06,
      "loss": 0.4823,
      "step": 4182
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.238843798209457e-06,
      "loss": 0.4662,
      "step": 4183
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.228532124878284e-06,
      "loss": 0.5354,
      "step": 4184
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.218229640981869e-06,
      "loss": 0.581,
      "step": 4185
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.207936350342092e-06,
      "loss": 0.5405,
      "step": 4186
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.197652256777463e-06,
      "loss": 0.5505,
      "step": 4187
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.187377364103044e-06,
      "loss": 0.5201,
      "step": 4188
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.177111676130499e-06,
      "loss": 0.5615,
      "step": 4189
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.166855196668072e-06,
      "loss": 0.5193,
      "step": 4190
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.156607929520603e-06,
      "loss": 0.4681,
      "step": 4191
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.1463698784895e-06,
      "loss": 0.5183,
      "step": 4192
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.136141047372756e-06,
      "loss": 0.5286,
      "step": 4193
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.1259214399649415e-06,
      "loss": 0.4635,
      "step": 4194
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.115711060057221e-06,
      "loss": 0.5008,
      "step": 4195
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.105509911437319e-06,
      "loss": 0.5026,
      "step": 4196
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.095317997889539e-06,
      "loss": 0.4896,
      "step": 4197
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.085135323194749e-06,
      "loss": 0.4951,
      "step": 4198
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.074961891130424e-06,
      "loss": 0.5077,
      "step": 4199
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.064797705470574e-06,
      "loss": 0.5637,
      "step": 4200
    },
    {
      "epoch": 3.19,
      "eval_loss": 0.9447075128555298,
      "eval_runtime": 14.3458,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 4200
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.054642769985783e-06,
      "loss": 0.4423,
      "step": 4201
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.044497088443238e-06,
      "loss": 0.559,
      "step": 4202
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.034360664606638e-06,
      "loss": 0.528,
      "step": 4203
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.024233502236299e-06,
      "loss": 0.5292,
      "step": 4204
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.014115605089063e-06,
      "loss": 0.6017,
      "step": 4205
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.0040069769183715e-06,
      "loss": 0.4566,
      "step": 4206
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.9939076214741845e-06,
      "loss": 0.5107,
      "step": 4207
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.983817542503063e-06,
      "loss": 0.5303,
      "step": 4208
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.973736743748097e-06,
      "loss": 0.5009,
      "step": 4209
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.963665228948956e-06,
      "loss": 0.5377,
      "step": 4210
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.953603001841837e-06,
      "loss": 0.4623,
      "step": 4211
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.9435500661595305e-06,
      "loss": 0.505,
      "step": 4212
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.93350642563135e-06,
      "loss": 0.5918,
      "step": 4213
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.923472083983169e-06,
      "loss": 0.5171,
      "step": 4214
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.9134470449374065e-06,
      "loss": 0.4564,
      "step": 4215
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.903431312213051e-06,
      "loss": 0.4986,
      "step": 4216
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.893424889525617e-06,
      "loss": 0.4995,
      "step": 4217
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.883427780587172e-06,
      "loss": 0.493,
      "step": 4218
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.873439989106319e-06,
      "loss": 0.5431,
      "step": 4219
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.86346151878824e-06,
      "loss": 0.5119,
      "step": 4220
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.8534923733345996e-06,
      "loss": 0.5189,
      "step": 4221
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.843532556443664e-06,
      "loss": 0.4911,
      "step": 4222
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.833582071810195e-06,
      "loss": 0.5441,
      "step": 4223
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.823640923125527e-06,
      "loss": 0.4547,
      "step": 4224
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.81370911407749e-06,
      "loss": 0.54,
      "step": 4225
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.803786648350493e-06,
      "loss": 0.5962,
      "step": 4226
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.793873529625449e-06,
      "loss": 0.5357,
      "step": 4227
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.783969761579815e-06,
      "loss": 0.5274,
      "step": 4228
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.77407534788757e-06,
      "loss": 0.4358,
      "step": 4229
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.76419029221924e-06,
      "loss": 0.5332,
      "step": 4230
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.754314598241867e-06,
      "loss": 0.4705,
      "step": 4231
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.744448269619021e-06,
      "loss": 0.5004,
      "step": 4232
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.734591310010794e-06,
      "loss": 0.4539,
      "step": 4233
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.7247437230738225e-06,
      "loss": 0.4801,
      "step": 4234
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.7149055124612374e-06,
      "loss": 0.4973,
      "step": 4235
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.705076681822707e-06,
      "loss": 0.5204,
      "step": 4236
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.695257234804427e-06,
      "loss": 0.4836,
      "step": 4237
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.685447175049094e-06,
      "loss": 0.4824,
      "step": 4238
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.675646506195936e-06,
      "loss": 0.507,
      "step": 4239
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.6658552318806834e-06,
      "loss": 0.539,
      "step": 4240
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.656073355735606e-06,
      "loss": 0.4499,
      "step": 4241
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.646300881389465e-06,
      "loss": 0.4859,
      "step": 4242
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.636537812467539e-06,
      "loss": 0.474,
      "step": 4243
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.626784152591613e-06,
      "loss": 0.5356,
      "step": 4244
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.617039905380011e-06,
      "loss": 0.5011,
      "step": 4245
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.60730507444751e-06,
      "loss": 0.521,
      "step": 4246
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.597579663405449e-06,
      "loss": 0.5467,
      "step": 4247
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.5878636758616335e-06,
      "loss": 0.5311,
      "step": 4248
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.578157115420413e-06,
      "loss": 0.5023,
      "step": 4249
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.568459985682587e-06,
      "loss": 0.4707,
      "step": 4250
    },
    {
      "epoch": 3.23,
      "eval_loss": 0.9437572956085205,
      "eval_runtime": 14.3515,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 4250
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.558772290245505e-06,
      "loss": 0.4917,
      "step": 4251
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.54909403270299e-06,
      "loss": 0.5035,
      "step": 4252
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.539425216645371e-06,
      "loss": 0.4971,
      "step": 4253
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.52976584565947e-06,
      "loss": 0.5078,
      "step": 4254
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.520115923328621e-06,
      "loss": 0.4958,
      "step": 4255
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.510475453232634e-06,
      "loss": 0.4591,
      "step": 4256
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.500844438947821e-06,
      "loss": 0.4818,
      "step": 4257
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.4912228840469806e-06,
      "loss": 0.5587,
      "step": 4258
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.481610792099421e-06,
      "loss": 0.5046,
      "step": 4259
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.472008166670907e-06,
      "loss": 0.5308,
      "step": 4260
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.462415011323722e-06,
      "loss": 0.4744,
      "step": 4261
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.452831329616616e-06,
      "loss": 0.5022,
      "step": 4262
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.443257125104851e-06,
      "loss": 0.5164,
      "step": 4263
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.433692401340135e-06,
      "loss": 0.4707,
      "step": 4264
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.424137161870691e-06,
      "loss": 0.5021,
      "step": 4265
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.4145914102412e-06,
      "loss": 0.4907,
      "step": 4266
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.40505514999286e-06,
      "loss": 0.424,
      "step": 4267
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.3955283846632956e-06,
      "loss": 0.4973,
      "step": 4268
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.386011117786654e-06,
      "loss": 0.4725,
      "step": 4269
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.376503352893537e-06,
      "loss": 0.552,
      "step": 4270
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.367005093511019e-06,
      "loss": 0.5539,
      "step": 4271
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.35751634316267e-06,
      "loss": 0.5317,
      "step": 4272
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.348037105368505e-06,
      "loss": 0.4836,
      "step": 4273
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.338567383645033e-06,
      "loss": 0.5294,
      "step": 4274
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.329107181505204e-06,
      "loss": 0.4604,
      "step": 4275
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.31965650245848e-06,
      "loss": 0.5289,
      "step": 4276
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.310215350010749e-06,
      "loss": 0.46,
      "step": 4277
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.300783727664389e-06,
      "loss": 0.533,
      "step": 4278
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.291361638918225e-06,
      "loss": 0.4954,
      "step": 4279
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.281949087267568e-06,
      "loss": 0.5406,
      "step": 4280
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.272546076204171e-06,
      "loss": 0.506,
      "step": 4281
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.263152609216257e-06,
      "loss": 0.5202,
      "step": 4282
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.2537686897885e-06,
      "loss": 0.5433,
      "step": 4283
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.24439432140206e-06,
      "loss": 0.4657,
      "step": 4284
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.235029507534503e-06,
      "loss": 0.481,
      "step": 4285
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.225674251659899e-06,
      "loss": 0.4553,
      "step": 4286
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.216328557248743e-06,
      "loss": 0.5034,
      "step": 4287
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.206992427768015e-06,
      "loss": 0.5046,
      "step": 4288
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.197665866681091e-06,
      "loss": 0.5104,
      "step": 4289
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.18834887744786e-06,
      "loss": 0.507,
      "step": 4290
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.179041463524617e-06,
      "loss": 0.5553,
      "step": 4291
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.169743628364133e-06,
      "loss": 0.5245,
      "step": 4292
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.160455375415594e-06,
      "loss": 0.4487,
      "step": 4293
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.151176708124662e-06,
      "loss": 0.4774,
      "step": 4294
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.141907629933432e-06,
      "loss": 0.5449,
      "step": 4295
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.132648144280431e-06,
      "loss": 0.447,
      "step": 4296
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.123398254600637e-06,
      "loss": 0.4581,
      "step": 4297
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.1141579643254766e-06,
      "loss": 0.493,
      "step": 4298
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.104927276882803e-06,
      "loss": 0.4358,
      "step": 4299
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.095706195696909e-06,
      "loss": 0.5757,
      "step": 4300
    },
    {
      "epoch": 3.27,
      "eval_loss": 0.9449428915977478,
      "eval_runtime": 14.3402,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 4300
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.086494724188519e-06,
      "loss": 0.4785,
      "step": 4301
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.0772928657748155e-06,
      "loss": 0.4567,
      "step": 4302
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.068100623869386e-06,
      "loss": 0.4898,
      "step": 4303
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.058918001882267e-06,
      "loss": 0.5259,
      "step": 4304
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.049745003219912e-06,
      "loss": 0.501,
      "step": 4305
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.040581631285233e-06,
      "loss": 0.4587,
      "step": 4306
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.031427889477543e-06,
      "loss": 0.4815,
      "step": 4307
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.0222837811925915e-06,
      "loss": 0.5081,
      "step": 4308
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.013149309822557e-06,
      "loss": 0.4893,
      "step": 4309
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.004024478756031e-06,
      "loss": 0.5365,
      "step": 4310
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.994909291378052e-06,
      "loss": 0.4735,
      "step": 4311
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.985803751070064e-06,
      "loss": 0.5642,
      "step": 4312
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.976707861209936e-06,
      "loss": 0.5087,
      "step": 4313
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.967621625171947e-06,
      "loss": 0.5827,
      "step": 4314
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.958545046326819e-06,
      "loss": 0.4956,
      "step": 4315
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.94947812804167e-06,
      "loss": 0.5531,
      "step": 4316
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.940420873680043e-06,
      "loss": 0.5632,
      "step": 4317
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.9313732866018856e-06,
      "loss": 0.5181,
      "step": 4318
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.9223353701635814e-06,
      "loss": 0.4832,
      "step": 4319
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.913307127717907e-06,
      "loss": 0.5163,
      "step": 4320
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.904288562614056e-06,
      "loss": 0.449,
      "step": 4321
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.8952796781976255e-06,
      "loss": 0.5648,
      "step": 4322
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.886280477810642e-06,
      "loss": 0.5083,
      "step": 4323
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.877290964791517e-06,
      "loss": 0.5009,
      "step": 4324
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.868311142475079e-06,
      "loss": 0.482,
      "step": 4325
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.859341014192549e-06,
      "loss": 0.4916,
      "step": 4326
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.850380583271587e-06,
      "loss": 0.5362,
      "step": 4327
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.841429853036202e-06,
      "loss": 0.543,
      "step": 4328
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.832488826806851e-06,
      "loss": 0.4839,
      "step": 4329
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.823557507900364e-06,
      "loss": 0.4839,
      "step": 4330
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.8146358996299958e-06,
      "loss": 0.4748,
      "step": 4331
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.8057240053053587e-06,
      "loss": 0.5226,
      "step": 4332
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.796821828232502e-06,
      "loss": 0.5804,
      "step": 4333
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.7879293717138464e-06,
      "loss": 0.5399,
      "step": 4334
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.7790466390482136e-06,
      "loss": 0.4839,
      "step": 4335
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.7701736335308147e-06,
      "loss": 0.5524,
      "step": 4336
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.761310358453266e-06,
      "loss": 0.4988,
      "step": 4337
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.752456817103558e-06,
      "loss": 0.5202,
      "step": 4338
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.743613012766073e-06,
      "loss": 0.5183,
      "step": 4339
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.7347789487215764e-06,
      "loss": 0.5323,
      "step": 4340
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.725954628247248e-06,
      "loss": 0.5466,
      "step": 4341
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.717140054616621e-06,
      "loss": 0.4966,
      "step": 4342
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.7083352310996243e-06,
      "loss": 0.489,
      "step": 4343
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.6995401609625677e-06,
      "loss": 0.5377,
      "step": 4344
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.690754847468155e-06,
      "loss": 0.5363,
      "step": 4345
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.681979293875456e-06,
      "loss": 0.5498,
      "step": 4346
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.673213503439923e-06,
      "loss": 0.4618,
      "step": 4347
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.664457479413381e-06,
      "loss": 0.5204,
      "step": 4348
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.6557112250440592e-06,
      "loss": 0.5212,
      "step": 4349
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.6469747435765253e-06,
      "loss": 0.5353,
      "step": 4350
    },
    {
      "epoch": 3.31,
      "eval_loss": 0.9444774985313416,
      "eval_runtime": 14.3476,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 4350
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.6382480382517434e-06,
      "loss": 0.4631,
      "step": 4351
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.6295311123070574e-06,
      "loss": 0.4902,
      "step": 4352
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.6208239689761536e-06,
      "loss": 0.4949,
      "step": 4353
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.6121266114891215e-06,
      "loss": 0.5668,
      "step": 4354
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.603439043072394e-06,
      "loss": 0.5824,
      "step": 4355
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.5947612669488096e-06,
      "loss": 0.5273,
      "step": 4356
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.586093286337523e-06,
      "loss": 0.5127,
      "step": 4357
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.5774351044540987e-06,
      "loss": 0.5054,
      "step": 4358
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.5687867245104442e-06,
      "loss": 0.4937,
      "step": 4359
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.560148149714837e-06,
      "loss": 0.4663,
      "step": 4360
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.5515193832719093e-06,
      "loss": 0.4732,
      "step": 4361
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.5429004283826783e-06,
      "loss": 0.5251,
      "step": 4362
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.534291288244492e-06,
      "loss": 0.4887,
      "step": 4363
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.5256919660510734e-06,
      "loss": 0.4687,
      "step": 4364
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.5171024649924967e-06,
      "loss": 0.4962,
      "step": 4365
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.508522788255203e-06,
      "loss": 0.4946,
      "step": 4366
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.499952939021983e-06,
      "loss": 0.555,
      "step": 4367
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.4913929204719743e-06,
      "loss": 0.5799,
      "step": 4368
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.482842735780676e-06,
      "loss": 0.5614,
      "step": 4369
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4743023881199476e-06,
      "loss": 0.4458,
      "step": 4370
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4657718806579684e-06,
      "loss": 0.649,
      "step": 4371
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4572512165593073e-06,
      "loss": 0.4734,
      "step": 4372
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4487403989848485e-06,
      "loss": 0.5176,
      "step": 4373
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4402394310918574e-06,
      "loss": 0.4958,
      "step": 4374
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4317483160339004e-06,
      "loss": 0.4513,
      "step": 4375
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4232670569609282e-06,
      "loss": 0.4725,
      "step": 4376
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.414795657019224e-06,
      "loss": 0.479,
      "step": 4377
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.4063341193514006e-06,
      "loss": 0.4531,
      "step": 4378
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.397882447096424e-06,
      "loss": 0.4733,
      "step": 4379
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.389440643389606e-06,
      "loss": 0.5684,
      "step": 4380
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.3810087113625854e-06,
      "loss": 0.5303,
      "step": 4381
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.3725866541433427e-06,
      "loss": 0.4703,
      "step": 4382
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.364174474856189e-06,
      "loss": 0.5112,
      "step": 4383
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.355772176621794e-06,
      "loss": 0.5545,
      "step": 4384
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.3473797625571355e-06,
      "loss": 0.5299,
      "step": 4385
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.3389972357755327e-06,
      "loss": 0.535,
      "step": 4386
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.330624599386648e-06,
      "loss": 0.5575,
      "step": 4387
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.3222618564964634e-06,
      "loss": 0.5457,
      "step": 4388
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.3139090102072876e-06,
      "loss": 0.53,
      "step": 4389
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.3055660636177653e-06,
      "loss": 0.4796,
      "step": 4390
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.2972330198228705e-06,
      "loss": 0.5438,
      "step": 4391
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.288909881913901e-06,
      "loss": 0.5086,
      "step": 4392
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.280596652978476e-06,
      "loss": 0.5005,
      "step": 4393
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.272293336100536e-06,
      "loss": 0.5418,
      "step": 4394
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.2639999343603668e-06,
      "loss": 0.5377,
      "step": 4395
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.2557164508345373e-06,
      "loss": 0.5745,
      "step": 4396
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.2474428885959793e-06,
      "loss": 0.487,
      "step": 4397
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.239179250713905e-06,
      "loss": 0.4599,
      "step": 4398
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.2309255402538886e-06,
      "loss": 0.583,
      "step": 4399
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.2226817602777705e-06,
      "loss": 0.514,
      "step": 4400
    },
    {
      "epoch": 3.35,
      "eval_loss": 0.9448750615119934,
      "eval_runtime": 14.3329,
      "eval_samples_per_second": 5.721,
      "eval_steps_per_second": 5.721,
      "step": 4400
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.214447913843751e-06,
      "loss": 0.5233,
      "step": 4401
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.2062240040063207e-06,
      "loss": 0.5139,
      "step": 4402
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.1980100338162954e-06,
      "loss": 0.5764,
      "step": 4403
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.189806006320791e-06,
      "loss": 0.5119,
      "step": 4404
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.1816119245632534e-06,
      "loss": 0.4828,
      "step": 4405
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.1734277915834275e-06,
      "loss": 0.5328,
      "step": 4406
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.1652536104173645e-06,
      "loss": 0.4846,
      "step": 4407
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.1570893840974256e-06,
      "loss": 0.4813,
      "step": 4408
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.1489351156522984e-06,
      "loss": 0.4631,
      "step": 4409
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.1407908081069326e-06,
      "loss": 0.5111,
      "step": 4410
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.132656464482635e-06,
      "loss": 0.5382,
      "step": 4411
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.1245320877969754e-06,
      "loss": 0.5365,
      "step": 4412
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.1164176810638557e-06,
      "loss": 0.4914,
      "step": 4413
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.1083132472934485e-06,
      "loss": 0.5305,
      "step": 4414
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.100218789492255e-06,
      "loss": 0.5583,
      "step": 4415
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.0921343106630605e-06,
      "loss": 0.5113,
      "step": 4416
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.084059813804956e-06,
      "loss": 0.5103,
      "step": 4417
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.0759953019133214e-06,
      "loss": 0.5624,
      "step": 4418
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.067940777979841e-06,
      "loss": 0.4565,
      "step": 4419
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.059896244992484e-06,
      "loss": 0.4928,
      "step": 4420
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.051861705935517e-06,
      "loss": 0.6079,
      "step": 4421
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.0438371637895124e-06,
      "loss": 0.5261,
      "step": 4422
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.0358226215313124e-06,
      "loss": 0.5021,
      "step": 4423
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.0278180821340686e-06,
      "loss": 0.594,
      "step": 4424
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.019823548567199e-06,
      "loss": 0.4998,
      "step": 4425
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.011839023796438e-06,
      "loss": 0.4864,
      "step": 4426
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.003864510783789e-06,
      "loss": 0.5094,
      "step": 4427
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.9959000124875437e-06,
      "loss": 0.5439,
      "step": 4428
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.9879455318622773e-06,
      "loss": 0.547,
      "step": 4429
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.9800010718588567e-06,
      "loss": 0.5289,
      "step": 4430
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.9720666354244275e-06,
      "loss": 0.4703,
      "step": 4431
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.964142225502412e-06,
      "loss": 0.5743,
      "step": 4432
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.956227845032512e-06,
      "loss": 0.4836,
      "step": 4433
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.948323496950732e-06,
      "loss": 0.4727,
      "step": 4434
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.9404291841893104e-06,
      "loss": 0.5062,
      "step": 4435
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.9325449096768107e-06,
      "loss": 0.5281,
      "step": 4436
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.9246706763380336e-06,
      "loss": 0.4666,
      "step": 4437
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.9168064870940923e-06,
      "loss": 0.4574,
      "step": 4438
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.90895234486233e-06,
      "loss": 0.5374,
      "step": 4439
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.9011082525564027e-06,
      "loss": 0.5363,
      "step": 4440
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.8932742130862087e-06,
      "loss": 0.512,
      "step": 4441
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.885450229357951e-06,
      "loss": 0.469,
      "step": 4442
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.8776363042740573e-06,
      "loss": 0.4633,
      "step": 4443
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.8698324407332654e-06,
      "loss": 0.4913,
      "step": 4444
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.862038641630553e-06,
      "loss": 0.4962,
      "step": 4445
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.854254909857178e-06,
      "loss": 0.5491,
      "step": 4446
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.8464812483006564e-06,
      "loss": 0.5032,
      "step": 4447
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.838717659844775e-06,
      "loss": 0.5215,
      "step": 4448
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.8309641473695835e-06,
      "loss": 0.5113,
      "step": 4449
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.8232207137513877e-06,
      "loss": 0.5102,
      "step": 4450
    },
    {
      "epoch": 3.39,
      "eval_loss": 0.9456427693367004,
      "eval_runtime": 14.3428,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 4450
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.8154873618627532e-06,
      "loss": 0.4937,
      "step": 4451
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.8077640945725214e-06,
      "loss": 0.4856,
      "step": 4452
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.8000509147457707e-06,
      "loss": 0.5426,
      "step": 4453
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.792347825243853e-06,
      "loss": 0.5286,
      "step": 4454
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.7846548289243677e-06,
      "loss": 0.6046,
      "step": 4455
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.77697192864118e-06,
      "loss": 0.4883,
      "step": 4456
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.7692991272444057e-06,
      "loss": 0.5447,
      "step": 4457
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.7616364275804052e-06,
      "loss": 0.465,
      "step": 4458
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.753983832491801e-06,
      "loss": 0.4957,
      "step": 4459
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.7463413448174625e-06,
      "loss": 0.5205,
      "step": 4460
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.7387089673925216e-06,
      "loss": 0.5338,
      "step": 4461
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.7310867030483433e-06,
      "loss": 0.4775,
      "step": 4462
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.7234745546125493e-06,
      "loss": 0.5117,
      "step": 4463
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.7158725249089976e-06,
      "loss": 0.4753,
      "step": 4464
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.7082806167578187e-06,
      "loss": 0.5277,
      "step": 4465
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.700698832975357e-06,
      "loss": 0.565,
      "step": 4466
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.693127176374235e-06,
      "loss": 0.5288,
      "step": 4467
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.68556564976327e-06,
      "loss": 0.5364,
      "step": 4468
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.678014255947577e-06,
      "loss": 0.5046,
      "step": 4469
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.6704729977284702e-06,
      "loss": 0.454,
      "step": 4470
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.6629418779035242e-06,
      "loss": 0.5701,
      "step": 4471
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.655420899266544e-06,
      "loss": 0.5433,
      "step": 4472
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.6479100646075816e-06,
      "loss": 0.4965,
      "step": 4473
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.640409376712918e-06,
      "loss": 0.5431,
      "step": 4474
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.6329188383650705e-06,
      "loss": 0.4767,
      "step": 4475
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.6254384523427866e-06,
      "loss": 0.4766,
      "step": 4476
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.6179682214210766e-06,
      "loss": 0.4702,
      "step": 4477
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.61050814837113e-06,
      "loss": 0.4728,
      "step": 4478
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.603058235960424e-06,
      "loss": 0.5029,
      "step": 4479
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5956184869526233e-06,
      "loss": 0.4902,
      "step": 4480
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5881889041076647e-06,
      "loss": 0.5067,
      "step": 4481
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5807694901816593e-06,
      "loss": 0.5276,
      "step": 4482
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.573360247926999e-06,
      "loss": 0.4755,
      "step": 4483
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5659611800922723e-06,
      "loss": 0.5369,
      "step": 4484
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5585722894222964e-06,
      "loss": 0.4655,
      "step": 4485
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.551193578658119e-06,
      "loss": 0.5118,
      "step": 4486
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.543825050537013e-06,
      "loss": 0.5045,
      "step": 4487
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.53646670779247e-06,
      "loss": 0.4929,
      "step": 4488
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5291185531542014e-06,
      "loss": 0.5191,
      "step": 4489
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.521780589348138e-06,
      "loss": 0.496,
      "step": 4490
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.514452819096439e-06,
      "loss": 0.4983,
      "step": 4491
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5071352451174788e-06,
      "loss": 0.5381,
      "step": 4492
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.49982787012584e-06,
      "loss": 0.4974,
      "step": 4493
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.492530696832325e-06,
      "loss": 0.5561,
      "step": 4494
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.4852437279439698e-06,
      "loss": 0.5025,
      "step": 4495
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.4779669661640005e-06,
      "loss": 0.4986,
      "step": 4496
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.47070041419187e-06,
      "loss": 0.4768,
      "step": 4497
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.4634440747232323e-06,
      "loss": 0.5543,
      "step": 4498
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.4561979504499742e-06,
      "loss": 0.4673,
      "step": 4499
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.4489620440601725e-06,
      "loss": 0.5367,
      "step": 4500
    },
    {
      "epoch": 3.43,
      "eval_loss": 0.9434195756912231,
      "eval_runtime": 14.3597,
      "eval_samples_per_second": 5.71,
      "eval_steps_per_second": 5.71,
      "step": 4500
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.4417363582381156e-06,
      "loss": 0.58,
      "step": 4501
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.434520895664324e-06,
      "loss": 0.517,
      "step": 4502
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.427315659015479e-06,
      "loss": 0.5209,
      "step": 4503
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.420120650964519e-06,
      "loss": 0.5118,
      "step": 4504
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.412935874180556e-06,
      "loss": 0.5075,
      "step": 4505
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.4057613313289253e-06,
      "loss": 0.5682,
      "step": 4506
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.3985970250711397e-06,
      "loss": 0.5144,
      "step": 4507
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.391442958064948e-06,
      "loss": 0.4869,
      "step": 4508
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.384299132964273e-06,
      "loss": 0.5276,
      "step": 4509
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.377165552419255e-06,
      "loss": 0.4492,
      "step": 4510
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.3700422190762204e-06,
      "loss": 0.5197,
      "step": 4511
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.3629291355777127e-06,
      "loss": 0.5223,
      "step": 4512
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.3558263045624575e-06,
      "loss": 0.4679,
      "step": 4513
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.3487337286653766e-06,
      "loss": 0.5166,
      "step": 4514
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.3416514105175923e-06,
      "loss": 0.5083,
      "step": 4515
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.3345793527464297e-06,
      "loss": 0.4924,
      "step": 4516
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.327517557975394e-06,
      "loss": 0.5181,
      "step": 4517
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.320466028824192e-06,
      "loss": 0.476,
      "step": 4518
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.3134247679087117e-06,
      "loss": 0.4558,
      "step": 4519
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.3063937778410536e-06,
      "loss": 0.5423,
      "step": 4520
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.299373061229475e-06,
      "loss": 0.4635,
      "step": 4521
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.292362620678452e-06,
      "loss": 0.4752,
      "step": 4522
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.2853624587886344e-06,
      "loss": 0.4719,
      "step": 4523
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.278372578156872e-06,
      "loss": 0.5032,
      "step": 4524
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2713929813761735e-06,
      "loss": 0.442,
      "step": 4525
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2644236710357578e-06,
      "loss": 0.5443,
      "step": 4526
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2574646497210236e-06,
      "loss": 0.5704,
      "step": 4527
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.250515920013543e-06,
      "loss": 0.4721,
      "step": 4528
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.243577484491073e-06,
      "loss": 0.5193,
      "step": 4529
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.236649345727565e-06,
      "loss": 0.5299,
      "step": 4530
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2297315062931333e-06,
      "loss": 0.4917,
      "step": 4531
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.222823968754075e-06,
      "loss": 0.4994,
      "step": 4532
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2159267356728808e-06,
      "loss": 0.5097,
      "step": 4533
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2090398096081966e-06,
      "loss": 0.5019,
      "step": 4534
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.2021631931148624e-06,
      "loss": 0.5477,
      "step": 4535
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.195296888743877e-06,
      "loss": 0.4684,
      "step": 4536
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.1884408990424307e-06,
      "loss": 0.5351,
      "step": 4537
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.181595226553882e-06,
      "loss": 0.5323,
      "step": 4538
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1747598738177556e-06,
      "loss": 0.4937,
      "step": 4539
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1679348433697467e-06,
      "loss": 0.5659,
      "step": 4540
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1611201377417366e-06,
      "loss": 0.5695,
      "step": 4541
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1543157594617634e-06,
      "loss": 0.5342,
      "step": 4542
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.147521711054038e-06,
      "loss": 0.5005,
      "step": 4543
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1407379950389305e-06,
      "loss": 0.4507,
      "step": 4544
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1339646139330048e-06,
      "loss": 0.4553,
      "step": 4545
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1272015702489478e-06,
      "loss": 0.5063,
      "step": 4546
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1204488664956524e-06,
      "loss": 0.5482,
      "step": 4547
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.113706505178148e-06,
      "loss": 0.4795,
      "step": 4548
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1069744887976574e-06,
      "loss": 0.4742,
      "step": 4549
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1002528198515223e-06,
      "loss": 0.47,
      "step": 4550
    },
    {
      "epoch": 3.46,
      "eval_loss": 0.944807767868042,
      "eval_runtime": 14.3478,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 4550
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0935415008332857e-06,
      "loss": 0.4588,
      "step": 4551
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.086840534232636e-06,
      "loss": 0.4914,
      "step": 4552
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0801499225354136e-06,
      "loss": 0.5375,
      "step": 4553
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.073469668223624e-06,
      "loss": 0.525,
      "step": 4554
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.066799773775442e-06,
      "loss": 0.5043,
      "step": 4555
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0601402416651806e-06,
      "loss": 0.5633,
      "step": 4556
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0534910743633173e-06,
      "loss": 0.5951,
      "step": 4557
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0468522743364793e-06,
      "loss": 0.5134,
      "step": 4558
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0402238440474695e-06,
      "loss": 0.6305,
      "step": 4559
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.033605785955205e-06,
      "loss": 0.5103,
      "step": 4560
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0269981025147876e-06,
      "loss": 0.4639,
      "step": 4561
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0204007961774573e-06,
      "loss": 0.4679,
      "step": 4562
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.0138138693906193e-06,
      "loss": 0.5578,
      "step": 4563
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.007237324597796e-06,
      "loss": 0.5198,
      "step": 4564
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.000671164238689e-06,
      "loss": 0.5171,
      "step": 4565
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.994115390749136e-06,
      "loss": 0.5398,
      "step": 4566
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.987570006561126e-06,
      "loss": 0.4728,
      "step": 4567
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.9810350141027857e-06,
      "loss": 0.5085,
      "step": 4568
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.974510415798396e-06,
      "loss": 0.4796,
      "step": 4569
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.9679962140683717e-06,
      "loss": 0.53,
      "step": 4570
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.9614924113292755e-06,
      "loss": 0.503,
      "step": 4571
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.954999009993824e-06,
      "loss": 0.4671,
      "step": 4572
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.9485160124708577e-06,
      "loss": 0.5553,
      "step": 4573
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.9420434211653625e-06,
      "loss": 0.6115,
      "step": 4574
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.935581238478468e-06,
      "loss": 0.4744,
      "step": 4575
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.929129466807442e-06,
      "loss": 0.5226,
      "step": 4576
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.922688108545688e-06,
      "loss": 0.5704,
      "step": 4577
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.9162571660827475e-06,
      "loss": 0.4935,
      "step": 4578
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.9098366418042922e-06,
      "loss": 0.4353,
      "step": 4579
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.9034265380921468e-06,
      "loss": 0.5353,
      "step": 4580
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8970268573242488e-06,
      "loss": 0.5143,
      "step": 4581
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8906376018746795e-06,
      "loss": 0.4759,
      "step": 4582
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8842587741136463e-06,
      "loss": 0.428,
      "step": 4583
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8778903764075095e-06,
      "loss": 0.5391,
      "step": 4584
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.871532411118726e-06,
      "loss": 0.4692,
      "step": 4585
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8651848806059125e-06,
      "loss": 0.4997,
      "step": 4586
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8588477872237951e-06,
      "loss": 0.4642,
      "step": 4587
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8525211333232472e-06,
      "loss": 0.5988,
      "step": 4588
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.8462049212512443e-06,
      "loss": 0.5027,
      "step": 4589
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.8398991533509124e-06,
      "loss": 0.5745,
      "step": 4590
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.8336038319614833e-06,
      "loss": 0.5008,
      "step": 4591
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.8273189594183425e-06,
      "loss": 0.5054,
      "step": 4592
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.821044538052955e-06,
      "loss": 0.5149,
      "step": 4593
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.8147805701929488e-06,
      "loss": 0.5751,
      "step": 4594
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.808527058162055e-06,
      "loss": 0.4616,
      "step": 4595
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.8022840042801315e-06,
      "loss": 0.5308,
      "step": 4596
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.7960514108631488e-06,
      "loss": 0.5475,
      "step": 4597
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.7898292802232109e-06,
      "loss": 0.5412,
      "step": 4598
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.7836176146685279e-06,
      "loss": 0.489,
      "step": 4599
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.7774164165034367e-06,
      "loss": 0.524,
      "step": 4600
    },
    {
      "epoch": 3.5,
      "eval_loss": 0.944672167301178,
      "eval_runtime": 14.3503,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 4600
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.771225688028374e-06,
      "loss": 0.4676,
      "step": 4601
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.765045431539919e-06,
      "loss": 0.5499,
      "step": 4602
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7588756493307512e-06,
      "loss": 0.5299,
      "step": 4603
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7527163436896564e-06,
      "loss": 0.4414,
      "step": 4604
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7465675169015472e-06,
      "loss": 0.4995,
      "step": 4605
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.740429171247452e-06,
      "loss": 0.5523,
      "step": 4606
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7343013090044923e-06,
      "loss": 0.5605,
      "step": 4607
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7281839324459236e-06,
      "loss": 0.5792,
      "step": 4608
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7220770438410904e-06,
      "loss": 0.5097,
      "step": 4609
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7159806454554538e-06,
      "loss": 0.4634,
      "step": 4610
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.709894739550598e-06,
      "loss": 0.5527,
      "step": 4611
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.7038193283841973e-06,
      "loss": 0.5177,
      "step": 4612
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.6977544142100355e-06,
      "loss": 0.4774,
      "step": 4613
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.6916999992780025e-06,
      "loss": 0.5394,
      "step": 4614
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.685656085834102e-06,
      "loss": 0.5265,
      "step": 4615
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6796226761204304e-06,
      "loss": 0.4282,
      "step": 4616
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6735997723752006e-06,
      "loss": 0.4686,
      "step": 4617
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6675873768327089e-06,
      "loss": 0.4784,
      "step": 4618
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.661585491723374e-06,
      "loss": 0.5249,
      "step": 4619
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6555941192737056e-06,
      "loss": 0.501,
      "step": 4620
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6496132617063086e-06,
      "loss": 0.5402,
      "step": 4621
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6436429212398918e-06,
      "loss": 0.4707,
      "step": 4622
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6376831000892734e-06,
      "loss": 0.6382,
      "step": 4623
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6317338004653548e-06,
      "loss": 0.532,
      "step": 4624
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6257950245751374e-06,
      "loss": 0.5265,
      "step": 4625
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6198667746217155e-06,
      "loss": 0.469,
      "step": 4626
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6139490528042965e-06,
      "loss": 0.5678,
      "step": 4627
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6080418613181513e-06,
      "loss": 0.5467,
      "step": 4628
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.6021452023546768e-06,
      "loss": 0.5393,
      "step": 4629
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5962590781013397e-06,
      "loss": 0.4937,
      "step": 4630
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.590383490741717e-06,
      "loss": 0.5032,
      "step": 4631
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5845184424554481e-06,
      "loss": 0.514,
      "step": 4632
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5786639354182998e-06,
      "loss": 0.4888,
      "step": 4633
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5728199718021007e-06,
      "loss": 0.53,
      "step": 4634
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5669865537747806e-06,
      "loss": 0.5523,
      "step": 4635
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5611636835003452e-06,
      "loss": 0.4948,
      "step": 4636
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5553513631389094e-06,
      "loss": 0.4526,
      "step": 4637
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5495495948466554e-06,
      "loss": 0.515,
      "step": 4638
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5437583807758548e-06,
      "loss": 0.5748,
      "step": 4639
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.537977723074866e-06,
      "loss": 0.499,
      "step": 4640
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.5322076238881332e-06,
      "loss": 0.4957,
      "step": 4641
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.5264480853561813e-06,
      "loss": 0.5637,
      "step": 4642
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.5206991096156176e-06,
      "loss": 0.5853,
      "step": 4643
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.514960698799126e-06,
      "loss": 0.5405,
      "step": 4644
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.5092328550354839e-06,
      "loss": 0.5099,
      "step": 4645
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.5035155804495382e-06,
      "loss": 0.4928,
      "step": 4646
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4978088771622157e-06,
      "loss": 0.5734,
      "step": 4647
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4921127472905228e-06,
      "loss": 0.5062,
      "step": 4648
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4864271929475525e-06,
      "loss": 0.5149,
      "step": 4649
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.480752216242458e-06,
      "loss": 0.504,
      "step": 4650
    },
    {
      "epoch": 3.54,
      "eval_loss": 0.9447042346000671,
      "eval_runtime": 14.3555,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 4650
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4750878192804785e-06,
      "loss": 0.5538,
      "step": 4651
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.469434004162936e-06,
      "loss": 0.5596,
      "step": 4652
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4637907729872002e-06,
      "loss": 0.4814,
      "step": 4653
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4581581278467493e-06,
      "loss": 0.4817,
      "step": 4654
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.452536070831102e-06,
      "loss": 0.4937,
      "step": 4655
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4469246040258832e-06,
      "loss": 0.4947,
      "step": 4656
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4413237295127546e-06,
      "loss": 0.5,
      "step": 4657
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4357334493694707e-06,
      "loss": 0.5048,
      "step": 4658
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4301537656698493e-06,
      "loss": 0.5052,
      "step": 4659
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4245846804837748e-06,
      "loss": 0.5198,
      "step": 4660
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.419026195877201e-06,
      "loss": 0.6259,
      "step": 4661
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4134783139121587e-06,
      "loss": 0.4372,
      "step": 4662
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.407941036646735e-06,
      "loss": 0.4683,
      "step": 4663
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.4024143661350797e-06,
      "loss": 0.534,
      "step": 4664
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.3968983044274132e-06,
      "loss": 0.5594,
      "step": 4665
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.3913928535700315e-06,
      "loss": 0.5486,
      "step": 4666
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.3858980156052748e-06,
      "loss": 0.4687,
      "step": 4667
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3804137925715554e-06,
      "loss": 0.429,
      "step": 4668
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3749401865033462e-06,
      "loss": 0.5219,
      "step": 4669
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3694771994311927e-06,
      "loss": 0.4545,
      "step": 4670
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3640248333816775e-06,
      "loss": 0.5293,
      "step": 4671
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3585830903774655e-06,
      "loss": 0.5407,
      "step": 4672
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3531519724372688e-06,
      "loss": 0.5841,
      "step": 4673
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3477314815758657e-06,
      "loss": 0.485,
      "step": 4674
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3423216198040778e-06,
      "loss": 0.4631,
      "step": 4675
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3369223891288029e-06,
      "loss": 0.516,
      "step": 4676
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3315337915529824e-06,
      "loss": 0.5005,
      "step": 4677
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.326155829075617e-06,
      "loss": 0.5028,
      "step": 4678
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.3207885036917577e-06,
      "loss": 0.4985,
      "step": 4679
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.3154318173925183e-06,
      "loss": 0.4913,
      "step": 4680
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.3100857721650594e-06,
      "loss": 0.5367,
      "step": 4681
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.3047503699925912e-06,
      "loss": 0.4765,
      "step": 4682
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2994256128543868e-06,
      "loss": 0.4372,
      "step": 4683
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2941115027257633e-06,
      "loss": 0.5074,
      "step": 4684
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2888080415780834e-06,
      "loss": 0.5406,
      "step": 4685
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.28351523137876e-06,
      "loss": 0.4948,
      "step": 4686
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.278233074091273e-06,
      "loss": 0.5555,
      "step": 4687
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2729615716751276e-06,
      "loss": 0.5495,
      "step": 4688
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.267700726085883e-06,
      "loss": 0.5922,
      "step": 4689
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2624505392751485e-06,
      "loss": 0.4911,
      "step": 4690
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2572110131905856e-06,
      "loss": 0.5128,
      "step": 4691
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.2519821497758866e-06,
      "loss": 0.539,
      "step": 4692
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2467639509707995e-06,
      "loss": 0.5165,
      "step": 4693
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2415564187111028e-06,
      "loss": 0.5115,
      "step": 4694
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2363595549286411e-06,
      "loss": 0.5435,
      "step": 4695
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2311733615512721e-06,
      "loss": 0.5227,
      "step": 4696
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2259978405029205e-06,
      "loss": 0.4805,
      "step": 4697
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2208329937035368e-06,
      "loss": 0.4365,
      "step": 4698
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.215678823069125e-06,
      "loss": 0.5108,
      "step": 4699
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.210535330511706e-06,
      "loss": 0.5358,
      "step": 4700
    },
    {
      "epoch": 3.58,
      "eval_loss": 0.9457272887229919,
      "eval_runtime": 14.3441,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 5.717,
      "step": 4700
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2054025179393668e-06,
      "loss": 0.5221,
      "step": 4701
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.2002803872562106e-06,
      "loss": 0.4986,
      "step": 4702
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.1951689403623912e-06,
      "loss": 0.6,
      "step": 4703
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.1900681791540913e-06,
      "loss": 0.4663,
      "step": 4704
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.1849781055235375e-06,
      "loss": 0.4943,
      "step": 4705
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.179898721358983e-06,
      "loss": 0.4627,
      "step": 4706
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.17483002854472e-06,
      "loss": 0.4807,
      "step": 4707
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1697720289610713e-06,
      "loss": 0.502,
      "step": 4708
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1647247244844062e-06,
      "loss": 0.5665,
      "step": 4709
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1596881169870998e-06,
      "loss": 0.4924,
      "step": 4710
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.154662208337588e-06,
      "loss": 0.4947,
      "step": 4711
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1496470004003156e-06,
      "loss": 0.5525,
      "step": 4712
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1446424950357782e-06,
      "loss": 0.5195,
      "step": 4713
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1396486941004768e-06,
      "loss": 0.5335,
      "step": 4714
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1346655994469602e-06,
      "loss": 0.4813,
      "step": 4715
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.129693212923799e-06,
      "loss": 0.5857,
      "step": 4716
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1247315363755984e-06,
      "loss": 0.5574,
      "step": 4717
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.1197805716429755e-06,
      "loss": 0.5268,
      "step": 4718
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.1148403205625879e-06,
      "loss": 0.5078,
      "step": 4719
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.1099107849671063e-06,
      "loss": 0.4982,
      "step": 4720
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.104991966685238e-06,
      "loss": 0.451,
      "step": 4721
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.1000838675417134e-06,
      "loss": 0.5352,
      "step": 4722
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.095186489357276e-06,
      "loss": 0.538,
      "step": 4723
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.0902998339487024e-06,
      "loss": 0.5549,
      "step": 4724
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.085423903128786e-06,
      "loss": 0.4083,
      "step": 4725
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.0805586987063465e-06,
      "loss": 0.4748,
      "step": 4726
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.0757042224862202e-06,
      "loss": 0.4864,
      "step": 4727
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.070860476269263e-06,
      "loss": 0.4862,
      "step": 4728
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.0660274618523547e-06,
      "loss": 0.4923,
      "step": 4729
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.0612051810283908e-06,
      "loss": 0.5071,
      "step": 4730
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.0563936355862868e-06,
      "loss": 0.555,
      "step": 4731
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.051592827310972e-06,
      "loss": 0.5332,
      "step": 4732
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.046802757983395e-06,
      "loss": 0.5416,
      "step": 4733
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0420234293805275e-06,
      "loss": 0.5126,
      "step": 4734
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0372548432753414e-06,
      "loss": 0.4967,
      "step": 4735
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.032497001436835e-06,
      "loss": 0.5313,
      "step": 4736
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0277499056300165e-06,
      "loss": 0.5192,
      "step": 4737
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0230135576159172e-06,
      "loss": 0.5191,
      "step": 4738
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0182879591515582e-06,
      "loss": 0.5066,
      "step": 4739
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0135731119900006e-06,
      "loss": 0.5198,
      "step": 4740
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0088690178802984e-06,
      "loss": 0.4426,
      "step": 4741
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.0041756785675294e-06,
      "loss": 0.4806,
      "step": 4742
    },
    {
      "epoch": 3.61,
      "learning_rate": 9.994930957927605e-07,
      "loss": 0.4909,
      "step": 4743
    },
    {
      "epoch": 3.61,
      "learning_rate": 9.94821271293096e-07,
      "loss": 0.5237,
      "step": 4744
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.901602068016325e-07,
      "loss": 0.5162,
      "step": 4745
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.85509904047477e-07,
      "loss": 0.5489,
      "step": 4746
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.808703647557394e-07,
      "loss": 0.5211,
      "step": 4747
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.76241590647553e-07,
      "loss": 0.5103,
      "step": 4748
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.71623583440041e-07,
      "loss": 0.5493,
      "step": 4749
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.670163448463399e-07,
      "loss": 0.5449,
      "step": 4750
    },
    {
      "epoch": 3.62,
      "eval_loss": 0.9459742307662964,
      "eval_runtime": 14.3507,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 4750
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.624198765755887e-07,
      "loss": 0.4921,
      "step": 4751
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.578341803329406e-07,
      "loss": 0.5269,
      "step": 4752
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.532592578195343e-07,
      "loss": 0.5256,
      "step": 4753
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.486951107325259e-07,
      "loss": 0.4509,
      "step": 4754
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.441417407650677e-07,
      "loss": 0.5172,
      "step": 4755
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.39599149606325e-07,
      "loss": 0.5019,
      "step": 4756
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.350673389414466e-07,
      "loss": 0.5122,
      "step": 4757
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.305463104515976e-07,
      "loss": 0.4649,
      "step": 4758
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.260360658139333e-07,
      "loss": 0.5641,
      "step": 4759
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.215366067016151e-07,
      "loss": 0.5098,
      "step": 4760
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.170479347838012e-07,
      "loss": 0.5367,
      "step": 4761
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.125700517256463e-07,
      "loss": 0.5283,
      "step": 4762
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.081029591883016e-07,
      "loss": 0.5683,
      "step": 4763
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.036466588289183e-07,
      "loss": 0.4846,
      "step": 4764
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.992011523006505e-07,
      "loss": 0.5061,
      "step": 4765
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.947664412526324e-07,
      "loss": 0.4873,
      "step": 4766
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.903425273300081e-07,
      "loss": 0.5471,
      "step": 4767
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.859294121739048e-07,
      "loss": 0.5063,
      "step": 4768
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.815270974214562e-07,
      "loss": 0.4516,
      "step": 4769
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.771355847057794e-07,
      "loss": 0.4781,
      "step": 4770
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.727548756559878e-07,
      "loss": 0.4753,
      "step": 4771
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.683849718971782e-07,
      "loss": 0.57,
      "step": 4772
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.640258750504604e-07,
      "loss": 0.5319,
      "step": 4773
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.596775867329176e-07,
      "loss": 0.529,
      "step": 4774
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.553401085576263e-07,
      "loss": 0.4777,
      "step": 4775
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.51013442133649e-07,
      "loss": 0.5303,
      "step": 4776
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.466975890660588e-07,
      "loss": 0.5168,
      "step": 4777
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.423925509558816e-07,
      "loss": 0.544,
      "step": 4778
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.380983294001665e-07,
      "loss": 0.4439,
      "step": 4779
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.338149259919226e-07,
      "loss": 0.4825,
      "step": 4780
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.29542342320172e-07,
      "loss": 0.4856,
      "step": 4781
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.252805799698937e-07,
      "loss": 0.5651,
      "step": 4782
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.210296405220763e-07,
      "loss": 0.5078,
      "step": 4783
    },
    {
      "epoch": 3.65,
      "learning_rate": 8.167895255536784e-07,
      "loss": 0.4934,
      "step": 4784
    },
    {
      "epoch": 3.65,
      "learning_rate": 8.125602366376583e-07,
      "loss": 0.561,
      "step": 4785
    },
    {
      "epoch": 3.65,
      "learning_rate": 8.083417753429412e-07,
      "loss": 0.5495,
      "step": 4786
    },
    {
      "epoch": 3.65,
      "learning_rate": 8.041341432344485e-07,
      "loss": 0.481,
      "step": 4787
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.999373418730749e-07,
      "loss": 0.62,
      "step": 4788
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.957513728157018e-07,
      "loss": 0.4478,
      "step": 4789
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.915762376151936e-07,
      "loss": 0.465,
      "step": 4790
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.874119378203981e-07,
      "loss": 0.5096,
      "step": 4791
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.832584749761328e-07,
      "loss": 0.5769,
      "step": 4792
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.791158506232054e-07,
      "loss": 0.4594,
      "step": 4793
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.749840662983931e-07,
      "loss": 0.525,
      "step": 4794
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.708631235344665e-07,
      "loss": 0.4271,
      "step": 4795
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.667530238601628e-07,
      "loss": 0.4954,
      "step": 4796
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.626537688001955e-07,
      "loss": 0.5469,
      "step": 4797
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.585653598752617e-07,
      "loss": 0.5052,
      "step": 4798
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.544877986020382e-07,
      "loss": 0.6034,
      "step": 4799
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.504210864931615e-07,
      "loss": 0.5942,
      "step": 4800
    },
    {
      "epoch": 3.66,
      "eval_loss": 0.9459019899368286,
      "eval_runtime": 14.3562,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 4800
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.463652250572617e-07,
      "loss": 0.5012,
      "step": 4801
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.423202157989351e-07,
      "loss": 0.4739,
      "step": 4802
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.382860602187447e-07,
      "loss": 0.5351,
      "step": 4803
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.342627598132434e-07,
      "loss": 0.4835,
      "step": 4804
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.302503160749441e-07,
      "loss": 0.5331,
      "step": 4805
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.262487304923459e-07,
      "loss": 0.5115,
      "step": 4806
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.222580045499017e-07,
      "loss": 0.6011,
      "step": 4807
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.182781397280469e-07,
      "loss": 0.5072,
      "step": 4808
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.143091375031874e-07,
      "loss": 0.5594,
      "step": 4809
    },
    {
      "epoch": 3.67,
      "learning_rate": 7.10350999347702e-07,
      "loss": 0.4873,
      "step": 4810
    },
    {
      "epoch": 3.67,
      "learning_rate": 7.064037267299295e-07,
      "loss": 0.5394,
      "step": 4811
    },
    {
      "epoch": 3.67,
      "learning_rate": 7.024673211141852e-07,
      "loss": 0.4625,
      "step": 4812
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.985417839607511e-07,
      "loss": 0.5255,
      "step": 4813
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.946271167258822e-07,
      "loss": 0.4801,
      "step": 4814
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.907233208617903e-07,
      "loss": 0.4665,
      "step": 4815
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.868303978166668e-07,
      "loss": 0.4915,
      "step": 4816
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.8294834903466e-07,
      "loss": 0.5059,
      "step": 4817
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.790771759558912e-07,
      "loss": 0.4765,
      "step": 4818
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.752168800164349e-07,
      "loss": 0.5319,
      "step": 4819
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.713674626483523e-07,
      "loss": 0.4708,
      "step": 4820
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.675289252796479e-07,
      "loss": 0.509,
      "step": 4821
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.637012693342992e-07,
      "loss": 0.4836,
      "step": 4822
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.598844962322436e-07,
      "loss": 0.4687,
      "step": 4823
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.560786073893921e-07,
      "loss": 0.5959,
      "step": 4824
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.522836042176017e-07,
      "loss": 0.475,
      "step": 4825
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.484994881246997e-07,
      "loss": 0.5421,
      "step": 4826
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.447262605144799e-07,
      "loss": 0.4752,
      "step": 4827
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.40963922786686e-07,
      "loss": 0.5123,
      "step": 4828
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.372124763370279e-07,
      "loss": 0.5249,
      "step": 4829
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.334719225571761e-07,
      "loss": 0.528,
      "step": 4830
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.297422628347538e-07,
      "loss": 0.4704,
      "step": 4831
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.260234985533475e-07,
      "loss": 0.6122,
      "step": 4832
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.223156310925104e-07,
      "loss": 0.5383,
      "step": 4833
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.186186618277357e-07,
      "loss": 0.6113,
      "step": 4834
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.149325921304861e-07,
      "loss": 0.5391,
      "step": 4835
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.112574233681745e-07,
      "loss": 0.539,
      "step": 4836
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.07593156904177e-07,
      "loss": 0.5451,
      "step": 4837
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.039397940978197e-07,
      "loss": 0.4878,
      "step": 4838
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.002973363043818e-07,
      "loss": 0.5373,
      "step": 4839
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.966657848751055e-07,
      "loss": 0.4803,
      "step": 4840
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.930451411571802e-07,
      "loss": 0.4926,
      "step": 4841
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.894354064937479e-07,
      "loss": 0.553,
      "step": 4842
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.858365822239109e-07,
      "loss": 0.4952,
      "step": 4843
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.822486696827145e-07,
      "loss": 0.5361,
      "step": 4844
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.786716702011708e-07,
      "loss": 0.4728,
      "step": 4845
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.751055851062248e-07,
      "loss": 0.5114,
      "step": 4846
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.715504157207851e-07,
      "loss": 0.4834,
      "step": 4847
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.680061633637035e-07,
      "loss": 0.5236,
      "step": 4848
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.644728293497947e-07,
      "loss": 0.4747,
      "step": 4849
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.60950414989807e-07,
      "loss": 0.4666,
      "step": 4850
    },
    {
      "epoch": 3.7,
      "eval_loss": 0.9457186460494995,
      "eval_runtime": 14.3519,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 4850
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.574389215904486e-07,
      "loss": 0.4997,
      "step": 4851
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.539383504543738e-07,
      "loss": 0.5024,
      "step": 4852
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.504487028801807e-07,
      "loss": 0.4384,
      "step": 4853
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.469699801624173e-07,
      "loss": 0.4654,
      "step": 4854
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.435021835915877e-07,
      "loss": 0.5046,
      "step": 4855
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.400453144541329e-07,
      "loss": 0.5208,
      "step": 4856
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.365993740324371e-07,
      "loss": 0.4566,
      "step": 4857
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.331643636048411e-07,
      "loss": 0.5915,
      "step": 4858
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.297402844456222e-07,
      "loss": 0.4562,
      "step": 4859
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.263271378250112e-07,
      "loss": 0.497,
      "step": 4860
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.229249250091717e-07,
      "loss": 0.5541,
      "step": 4861
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.195336472602207e-07,
      "loss": 0.5007,
      "step": 4862
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.161533058362155e-07,
      "loss": 0.5267,
      "step": 4863
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.127839019911562e-07,
      "loss": 0.4358,
      "step": 4864
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.094254369749863e-07,
      "loss": 0.4964,
      "step": 4865
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.060779120335856e-07,
      "loss": 0.5561,
      "step": 4866
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.027413284087878e-07,
      "loss": 0.5296,
      "step": 4867
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.994156873383593e-07,
      "loss": 0.5045,
      "step": 4868
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.961009900560032e-07,
      "loss": 0.4847,
      "step": 4869
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.927972377913692e-07,
      "loss": 0.4723,
      "step": 4870
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.89504431770047e-07,
      "loss": 0.4904,
      "step": 4871
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.862225732135662e-07,
      "loss": 0.5293,
      "step": 4872
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.829516633393926e-07,
      "loss": 0.469,
      "step": 4873
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.79691703360926e-07,
      "loss": 0.5481,
      "step": 4874
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.764426944875122e-07,
      "loss": 0.5785,
      "step": 4875
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.732046379244304e-07,
      "loss": 0.5256,
      "step": 4876
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.699775348728996e-07,
      "loss": 0.4552,
      "step": 4877
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.6676138653006907e-07,
      "loss": 0.5314,
      "step": 4878
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.6355619408903096e-07,
      "loss": 0.483,
      "step": 4879
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.6036195873881415e-07,
      "loss": 0.4782,
      "step": 4880
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.571786816643775e-07,
      "loss": 0.4721,
      "step": 4881
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.54006364046613e-07,
      "loss": 0.4962,
      "step": 4882
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.5084500706235267e-07,
      "loss": 0.4349,
      "step": 4883
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.476946118843617e-07,
      "loss": 0.4971,
      "step": 4884
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.445551796813385e-07,
      "loss": 0.4469,
      "step": 4885
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.414267116179116e-07,
      "loss": 0.507,
      "step": 4886
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.3830920885464253e-07,
      "loss": 0.5002,
      "step": 4887
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.352026725480296e-07,
      "loss": 0.4623,
      "step": 4888
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.3210710385049757e-07,
      "loss": 0.491,
      "step": 4889
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.290225039104079e-07,
      "loss": 0.504,
      "step": 4890
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.259488738720485e-07,
      "loss": 0.4984,
      "step": 4891
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.228862148756407e-07,
      "loss": 0.4771,
      "step": 4892
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.19834528057329e-07,
      "loss": 0.4822,
      "step": 4893
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.1679381454920116e-07,
      "loss": 0.4751,
      "step": 4894
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.1376407547926154e-07,
      "loss": 0.4559,
      "step": 4895
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.107453119714477e-07,
      "loss": 0.4867,
      "step": 4896
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.0773752514562724e-07,
      "loss": 0.4669,
      "step": 4897
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.0474071611759423e-07,
      "loss": 0.5254,
      "step": 4898
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.017548859990727e-07,
      "loss": 0.5205,
      "step": 4899
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.9878003589770674e-07,
      "loss": 0.5705,
      "step": 4900
    },
    {
      "epoch": 3.74,
      "eval_loss": 0.9464563727378845,
      "eval_runtime": 14.3472,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 5.715,
      "step": 4900
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.9581616691707344e-07,
      "loss": 0.453,
      "step": 4901
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.928632801566767e-07,
      "loss": 0.5294,
      "step": 4902
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.899213767119469e-07,
      "loss": 0.5149,
      "step": 4903
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.869904576742345e-07,
      "loss": 0.4848,
      "step": 4904
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.840705241308129e-07,
      "loss": 0.4859,
      "step": 4905
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.811615771648957e-07,
      "loss": 0.4112,
      "step": 4906
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.7826361785560317e-07,
      "loss": 0.5105,
      "step": 4907
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.753766472779918e-07,
      "loss": 0.5511,
      "step": 4908
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.7250066650303504e-07,
      "loss": 0.5233,
      "step": 4909
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.6963567659762945e-07,
      "loss": 0.5611,
      "step": 4910
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.6678167862459813e-07,
      "loss": 0.5108,
      "step": 4911
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.639386736426842e-07,
      "loss": 0.4793,
      "step": 4912
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.61106662706554e-07,
      "loss": 0.4726,
      "step": 4913
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.5828564686679057e-07,
      "loss": 0.5313,
      "step": 4914
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.5547562716990334e-07,
      "loss": 0.463,
      "step": 4915
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.526766046583252e-07,
      "loss": 0.525,
      "step": 4916
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.4988858037040216e-07,
      "loss": 0.4904,
      "step": 4917
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.471115553404036e-07,
      "loss": 0.6137,
      "step": 4918
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.443455305985188e-07,
      "loss": 0.4569,
      "step": 4919
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.415905071708536e-07,
      "loss": 0.5043,
      "step": 4920
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.3884648607944047e-07,
      "loss": 0.5537,
      "step": 4921
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.3611346834221845e-07,
      "loss": 0.4525,
      "step": 4922
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.333914549730532e-07,
      "loss": 0.5171,
      "step": 4923
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.3068044698173037e-07,
      "loss": 0.5396,
      "step": 4924
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.2798044537394543e-07,
      "loss": 0.5699,
      "step": 4925
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.252914511513072e-07,
      "loss": 0.4931,
      "step": 4926
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.22613465311361e-07,
      "loss": 0.4959,
      "step": 4927
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.1994648884754564e-07,
      "loss": 0.512,
      "step": 4928
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.1729052274922954e-07,
      "loss": 0.4459,
      "step": 4929
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.146455680016913e-07,
      "loss": 0.5418,
      "step": 4930
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.1201162558612606e-07,
      "loss": 0.5341,
      "step": 4931
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.093886964796422e-07,
      "loss": 0.5147,
      "step": 4932
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.0677678165526466e-07,
      "loss": 0.4881,
      "step": 4933
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.041758820819318e-07,
      "loss": 0.5389,
      "step": 4934
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.0158599872449844e-07,
      "loss": 0.4763,
      "step": 4935
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.990071325437227e-07,
      "loss": 0.5002,
      "step": 4936
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.9643928449628933e-07,
      "loss": 0.5207,
      "step": 4937
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.9388245553478965e-07,
      "loss": 0.5018,
      "step": 4938
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.913366466077183e-07,
      "loss": 0.5254,
      "step": 4939
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.888018586594998e-07,
      "loss": 0.4835,
      "step": 4940
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.8627809263045536e-07,
      "loss": 0.5035,
      "step": 4941
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.837653494568226e-07,
      "loss": 0.4717,
      "step": 4942
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.8126363007075605e-07,
      "loss": 0.4763,
      "step": 4943
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.787729354003032e-07,
      "loss": 0.4942,
      "step": 4944
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.7629326636944505e-07,
      "loss": 0.5174,
      "step": 4945
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.7382462389805905e-07,
      "loss": 0.5692,
      "step": 4946
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.71367008901926e-07,
      "loss": 0.6362,
      "step": 4947
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.6892042229274994e-07,
      "loss": 0.4959,
      "step": 4948
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.664848649781382e-07,
      "loss": 0.4993,
      "step": 4949
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.6406033786160467e-07,
      "loss": 0.4826,
      "step": 4950
    },
    {
      "epoch": 3.77,
      "eval_loss": 0.9463525414466858,
      "eval_runtime": 14.3415,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 5.718,
      "step": 4950
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.616468418425666e-07,
      "loss": 0.5063,
      "step": 4951
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.592443778163678e-07,
      "loss": 0.4653,
      "step": 4952
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.56852946674232e-07,
      "loss": 0.5062,
      "step": 4953
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.544725493033162e-07,
      "loss": 0.621,
      "step": 4954
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.521031865866674e-07,
      "loss": 0.5053,
      "step": 4955
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.497448594032492e-07,
      "loss": 0.5555,
      "step": 4956
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.473975686279184e-07,
      "loss": 0.5286,
      "step": 4957
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.450613151314551e-07,
      "loss": 0.5191,
      "step": 4958
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.427360997805295e-07,
      "loss": 0.4856,
      "step": 4959
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.4042192343772806e-07,
      "loss": 0.4916,
      "step": 4960
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.381187869615309e-07,
      "loss": 0.5827,
      "step": 4961
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.358266912063378e-07,
      "loss": 0.4667,
      "step": 4962
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.3354563702243536e-07,
      "loss": 0.4963,
      "step": 4963
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.3127562525602998e-07,
      "loss": 0.5162,
      "step": 4964
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2901665674922133e-07,
      "loss": 0.5348,
      "step": 4965
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2676873234001582e-07,
      "loss": 0.5396,
      "step": 4966
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2453185286232304e-07,
      "loss": 0.4839,
      "step": 4967
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.223060191459525e-07,
      "loss": 0.5571,
      "step": 4968
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.200912320166204e-07,
      "loss": 0.5014,
      "step": 4969
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.1788749229594617e-07,
      "loss": 0.5042,
      "step": 4970
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.1569480080144255e-07,
      "loss": 0.4625,
      "step": 4971
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.1351315834653217e-07,
      "loss": 0.5206,
      "step": 4972
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.1134256574053436e-07,
      "loss": 0.4996,
      "step": 4973
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.091830237886716e-07,
      "loss": 0.5113,
      "step": 4974
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.0703453329206313e-07,
      "loss": 0.4894,
      "step": 4975
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.0489709504773802e-07,
      "loss": 0.5167,
      "step": 4976
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.0277070984861201e-07,
      "loss": 0.4842,
      "step": 4977
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.0065537848350746e-07,
      "loss": 0.5118,
      "step": 4978
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.9855110173714997e-07,
      "loss": 0.5311,
      "step": 4979
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.9645788039015511e-07,
      "loss": 0.4974,
      "step": 4980
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.943757152190484e-07,
      "loss": 0.4762,
      "step": 4981
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.9230460699623865e-07,
      "loss": 0.4984,
      "step": 4982
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.9024455649005123e-07,
      "loss": 0.5048,
      "step": 4983
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.8819556446469488e-07,
      "loss": 0.5339,
      "step": 4984
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.8615763168028154e-07,
      "loss": 0.4943,
      "step": 4985
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.841307588928165e-07,
      "loss": 0.5836,
      "step": 4986
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.8211494685421492e-07,
      "loss": 0.4996,
      "step": 4987
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.80110196312272e-07,
      "loss": 0.5098,
      "step": 4988
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.7811650801069278e-07,
      "loss": 0.5075,
      "step": 4989
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.7613388268906573e-07,
      "loss": 0.6297,
      "step": 4990
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.7416232108288577e-07,
      "loss": 0.4503,
      "step": 4991
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.7220182392354455e-07,
      "loss": 0.5105,
      "step": 4992
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.70252391938317e-07,
      "loss": 0.4984,
      "step": 4993
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.683140258503879e-07,
      "loss": 0.4885,
      "step": 4994
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.6638672637882546e-07,
      "loss": 0.4981,
      "step": 4995
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.6447049423859773e-07,
      "loss": 0.5205,
      "step": 4996
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.625653301405694e-07,
      "loss": 0.5632,
      "step": 4997
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.6067123479149514e-07,
      "loss": 0.4703,
      "step": 4998
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.587882088940229e-07,
      "loss": 0.5117,
      "step": 4999
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.5691625314669723e-07,
      "loss": 0.5134,
      "step": 5000
    },
    {
      "epoch": 3.81,
      "eval_loss": 0.9462335109710693,
      "eval_runtime": 14.3557,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 5000
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.5505536824395594e-07,
      "loss": 0.5564,
      "step": 5001
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.5320555487612686e-07,
      "loss": 0.5311,
      "step": 5002
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.5136681372943107e-07,
      "loss": 0.534,
      "step": 5003
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4953914548598624e-07,
      "loss": 0.5554,
      "step": 5004
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4772255082380337e-07,
      "loss": 0.5878,
      "step": 5005
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.459170304167734e-07,
      "loss": 0.5521,
      "step": 5006
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4412258493469387e-07,
      "loss": 0.4861,
      "step": 5007
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4233921504324566e-07,
      "loss": 0.5081,
      "step": 5008
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4056692140400284e-07,
      "loss": 0.5089,
      "step": 5009
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.3880570467442954e-07,
      "loss": 0.5305,
      "step": 5010
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.3705556550788312e-07,
      "loss": 0.5071,
      "step": 5011
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.3531650455361088e-07,
      "loss": 0.4539,
      "step": 5012
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.335885224567468e-07,
      "loss": 0.558,
      "step": 5013
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.318716198583214e-07,
      "loss": 0.5219,
      "step": 5014
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.3016579739524858e-07,
      "loss": 0.4971,
      "step": 5015
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.2847105570033547e-07,
      "loss": 0.5218,
      "step": 5016
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.2678739540227912e-07,
      "loss": 0.5178,
      "step": 5017
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.251148171256633e-07,
      "loss": 0.5372,
      "step": 5018
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.2345332149096167e-07,
      "loss": 0.4665,
      "step": 5019
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.2180290911453784e-07,
      "loss": 0.47,
      "step": 5020
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.2016358060864206e-07,
      "loss": 0.5102,
      "step": 5021
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1853533658141458e-07,
      "loss": 0.5369,
      "step": 5022
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1691817763688218e-07,
      "loss": 0.4888,
      "step": 5023
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1531210437496165e-07,
      "loss": 0.4439,
      "step": 5024
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1371711739144974e-07,
      "loss": 0.4904,
      "step": 5025
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1213321727804315e-07,
      "loss": 0.4445,
      "step": 5026
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.105604046223152e-07,
      "loss": 0.5068,
      "step": 5027
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.089986800077325e-07,
      "loss": 0.436,
      "step": 5028
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.0744804401364161e-07,
      "loss": 0.5062,
      "step": 5029
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.059084972152824e-07,
      "loss": 0.5331,
      "step": 5030
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.0438004018378133e-07,
      "loss": 0.5124,
      "step": 5031
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.0286267348614153e-07,
      "loss": 0.4518,
      "step": 5032
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.0135639768525939e-07,
      "loss": 0.5008,
      "step": 5033
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.986121333992126e-08,
      "loss": 0.5357,
      "step": 5034
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.837712100479013e-08,
      "loss": 0.4799,
      "step": 5035
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.690412123041559e-08,
      "loss": 0.5313,
      "step": 5036
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.54422145632372e-08,
      "loss": 0.4611,
      "step": 5037
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.399140154557451e-08,
      "loss": 0.5388,
      "step": 5038
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.255168271563363e-08,
      "loss": 0.4925,
      "step": 5039
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.112305860750403e-08,
      "loss": 0.5442,
      "step": 5040
    },
    {
      "epoch": 3.84,
      "learning_rate": 8.970552975116176e-08,
      "loss": 0.5035,
      "step": 5041
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.82990966724695e-08,
      "loss": 0.4822,
      "step": 5042
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.690375989315991e-08,
      "loss": 0.5422,
      "step": 5043
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.55195199308656e-08,
      "loss": 0.5554,
      "step": 5044
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.414637729909913e-08,
      "loss": 0.549,
      "step": 5045
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.278433250724304e-08,
      "loss": 0.468,
      "step": 5046
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.143338606058315e-08,
      "loss": 0.4889,
      "step": 5047
    },
    {
      "epoch": 3.85,
      "learning_rate": 8.00935384602719e-08,
      "loss": 0.5102,
      "step": 5048
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.87647902033517e-08,
      "loss": 0.5464,
      "step": 5049
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.744714178274826e-08,
      "loss": 0.4647,
      "step": 5050
    },
    {
      "epoch": 3.85,
      "eval_loss": 0.946294367313385,
      "eval_runtime": 14.3459,
      "eval_samples_per_second": 5.716,
      "eval_steps_per_second": 5.716,
      "step": 5050
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.614059368726722e-08,
      "loss": 0.5435,
      "step": 5051
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.484514640159423e-08,
      "loss": 0.4417,
      "step": 5052
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.356080040630487e-08,
      "loss": 0.5604,
      "step": 5053
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.228755617784466e-08,
      "loss": 0.5397,
      "step": 5054
    },
    {
      "epoch": 3.86,
      "learning_rate": 7.102541418854913e-08,
      "loss": 0.5569,
      "step": 5055
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.977437490663707e-08,
      "loss": 0.4378,
      "step": 5056
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.853443879620058e-08,
      "loss": 0.5352,
      "step": 5057
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.730560631721839e-08,
      "loss": 0.5578,
      "step": 5058
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.608787792554583e-08,
      "loss": 0.4901,
      "step": 5059
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.488125407292488e-08,
      "loss": 0.5367,
      "step": 5060
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.368573520697751e-08,
      "loss": 0.5718,
      "step": 5061
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.250132177119561e-08,
      "loss": 0.5188,
      "step": 5062
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.132801420497103e-08,
      "loss": 0.4497,
      "step": 5063
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.016581294355228e-08,
      "loss": 0.5027,
      "step": 5064
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.9014718418084477e-08,
      "loss": 0.515,
      "step": 5065
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.787473105558605e-08,
      "loss": 0.4979,
      "step": 5066
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.6745851278962036e-08,
      "loss": 0.4915,
      "step": 5067
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.5628079506980786e-08,
      "loss": 0.5121,
      "step": 5068
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.452141615431061e-08,
      "loss": 0.4862,
      "step": 5069
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.342586163148311e-08,
      "loss": 0.483,
      "step": 5070
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.2341416344916513e-08,
      "loss": 0.4594,
      "step": 5071
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.126808069690569e-08,
      "loss": 0.5171,
      "step": 5072
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.020585508562547e-08,
      "loss": 0.4875,
      "step": 5073
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.915473990512398e-08,
      "loss": 0.5757,
      "step": 5074
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.811473554533596e-08,
      "loss": 0.5128,
      "step": 5075
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.708584239206948e-08,
      "loss": 0.4817,
      "step": 5076
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.6068060827012535e-08,
      "loss": 0.4994,
      "step": 5077
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.5061391227723125e-08,
      "loss": 0.5847,
      "step": 5078
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.406583396765585e-08,
      "loss": 0.4928,
      "step": 5079
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.308138941611861e-08,
      "loss": 0.515,
      "step": 5080
    },
    {
      "epoch": 3.88,
      "learning_rate": 4.2108057938319287e-08,
      "loss": 0.4761,
      "step": 5081
    },
    {
      "epoch": 3.88,
      "learning_rate": 4.1145839895329054e-08,
      "loss": 0.4974,
      "step": 5082
    },
    {
      "epoch": 3.88,
      "learning_rate": 4.0194735644099035e-08,
      "loss": 0.5059,
      "step": 5083
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.9254745537460336e-08,
      "loss": 0.5311,
      "step": 5084
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.832586992412401e-08,
      "loss": 0.466,
      "step": 5085
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.740810914866444e-08,
      "loss": 0.4992,
      "step": 5086
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.6501463551549265e-08,
      "loss": 0.5407,
      "step": 5087
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.560593346911611e-08,
      "loss": 0.4554,
      "step": 5088
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.4721519233572585e-08,
      "loss": 0.5379,
      "step": 5089
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.384822117301289e-08,
      "loss": 0.4877,
      "step": 5090
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.298603961140123e-08,
      "loss": 0.5349,
      "step": 5091
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.2134974868581745e-08,
      "loss": 0.4976,
      "step": 5092
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.1295027260268563e-08,
      "loss": 0.5363,
      "step": 5093
    },
    {
      "epoch": 3.89,
      "learning_rate": 3.0466197098062423e-08,
      "loss": 0.5366,
      "step": 5094
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.964848468942738e-08,
      "loss": 0.5588,
      "step": 5095
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.8841890337710785e-08,
      "loss": 0.5311,
      "step": 5096
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.8046414342133285e-08,
      "loss": 0.5246,
      "step": 5097
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.7262056997795492e-08,
      "loss": 0.5057,
      "step": 5098
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.6488818595664655e-08,
      "loss": 0.5044,
      "step": 5099
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.5726699422587986e-08,
      "loss": 0.5407,
      "step": 5100
    },
    {
      "epoch": 3.89,
      "eval_loss": 0.94637131690979,
      "eval_runtime": 14.3561,
      "eval_samples_per_second": 5.712,
      "eval_steps_per_second": 5.712,
      "step": 5100
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.497569976128933e-08,
      "loss": 0.4589,
      "step": 5101
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.423581989036583e-08,
      "loss": 0.4623,
      "step": 5102
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.3507060084291265e-08,
      "loss": 0.5432,
      "step": 5103
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.2789420613409383e-08,
      "loss": 0.4553,
      "step": 5104
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.2082901743940565e-08,
      "loss": 0.5185,
      "step": 5105
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.1387503737985148e-08,
      "loss": 0.477,
      "step": 5106
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.0703226853513447e-08,
      "loss": 0.5407,
      "step": 5107
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.003007134436907e-08,
      "loss": 0.5137,
      "step": 5108
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.936803746027227e-08,
      "loss": 0.4935,
      "step": 5109
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.8717125446816585e-08,
      "loss": 0.4816,
      "step": 5110
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.8077335545468864e-08,
      "loss": 0.5214,
      "step": 5111
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.7448667993572588e-08,
      "loss": 0.5914,
      "step": 5112
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.68311230243412e-08,
      "loss": 0.5168,
      "step": 5113
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.6224700866868115e-08,
      "loss": 0.5294,
      "step": 5114
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.562940174611338e-08,
      "loss": 0.5744,
      "step": 5115
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.5045225882920343e-08,
      "loss": 0.5839,
      "step": 5116
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.4472173493992324e-08,
      "loss": 0.4105,
      "step": 5117
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.3910244791915938e-08,
      "loss": 0.4791,
      "step": 5118
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.3359439985151101e-08,
      "loss": 0.4569,
      "step": 5119
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.2819759278027698e-08,
      "loss": 0.5308,
      "step": 5120
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.2291202870748919e-08,
      "loss": 0.5646,
      "step": 5121
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.177377095939458e-08,
      "loss": 0.4216,
      "step": 5122
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.1267463735914473e-08,
      "loss": 0.5675,
      "step": 5123
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.077228138813502e-08,
      "loss": 0.5266,
      "step": 5124
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.0288224099749278e-08,
      "loss": 0.4873,
      "step": 5125
    },
    {
      "epoch": 3.91,
      "learning_rate": 9.815292050333602e-09,
      "loss": 0.5079,
      "step": 5126
    },
    {
      "epoch": 3.91,
      "learning_rate": 9.353485415324325e-09,
      "loss": 0.5206,
      "step": 5127
    },
    {
      "epoch": 3.91,
      "learning_rate": 8.902804366037742e-09,
      "loss": 0.5026,
      "step": 5128
    },
    {
      "epoch": 3.91,
      "learning_rate": 8.463249069666778e-09,
      "loss": 0.4823,
      "step": 5129
    },
    {
      "epoch": 3.91,
      "learning_rate": 8.03481968926767e-09,
      "loss": 0.4944,
      "step": 5130
    },
    {
      "epoch": 3.91,
      "learning_rate": 7.617516383779943e-09,
      "loss": 0.5646,
      "step": 5131
    },
    {
      "epoch": 3.91,
      "learning_rate": 7.211339308003106e-09,
      "loss": 0.5026,
      "step": 5132
    },
    {
      "epoch": 3.92,
      "learning_rate": 6.81628861262329e-09,
      "loss": 0.4763,
      "step": 5133
    },
    {
      "epoch": 3.92,
      "learning_rate": 6.4323644441866e-09,
      "loss": 0.4716,
      "step": 5134
    },
    {
      "epoch": 3.92,
      "learning_rate": 6.05956694511911e-09,
      "loss": 0.5148,
      "step": 5135
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.6978962537135306e-09,
      "loss": 0.5238,
      "step": 5136
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.3473525041425334e-09,
      "loss": 0.5803,
      "step": 5137
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.007935826445431e-09,
      "loss": 0.5236,
      "step": 5138
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.679646346534838e-09,
      "loss": 0.4771,
      "step": 5139
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.362484186196669e-09,
      "loss": 0.5888,
      "step": 5140
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.056449463086809e-09,
      "loss": 0.4639,
      "step": 5141
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.761542290737774e-09,
      "loss": 0.5188,
      "step": 5142
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.4777627785453902e-09,
      "loss": 0.482,
      "step": 5143
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.205111031785446e-09,
      "loss": 0.543,
      "step": 5144
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.9435871516036994e-09,
      "loss": 0.5681,
      "step": 5145
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.69319123501921e-09,
      "loss": 0.4889,
      "step": 5146
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.4539233749143463e-09,
      "loss": 0.5191,
      "step": 5147
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.2257836600581007e-09,
      "loss": 0.5254,
      "step": 5148
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.008772175079443e-09,
      "loss": 0.5136,
      "step": 5149
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.8028890004806453e-09,
      "loss": 0.4612,
      "step": 5150
    },
    {
      "epoch": 3.93,
      "eval_loss": 0.9462961554527283,
      "eval_runtime": 14.3509,
      "eval_samples_per_second": 5.714,
      "eval_steps_per_second": 5.714,
      "step": 5150
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.6081342126406106e-09,
      "loss": 0.419,
      "step": 5151
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.4245078838082127e-09,
      "loss": 0.4477,
      "step": 5152
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.2520100820989644e-09,
      "loss": 0.4962,
      "step": 5153
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.0906408715083415e-09,
      "loss": 0.4574,
      "step": 5154
    },
    {
      "epoch": 3.93,
      "learning_rate": 9.404003118951288e-10,
      "loss": 0.553,
      "step": 5155
    },
    {
      "epoch": 3.93,
      "learning_rate": 8.012884589980729e-10,
      "loss": 0.5127,
      "step": 5156
    },
    {
      "epoch": 3.93,
      "learning_rate": 6.733053644225607e-10,
      "loss": 0.5174,
      "step": 5157
    },
    {
      "epoch": 3.93,
      "learning_rate": 5.564510756439489e-10,
      "loss": 0.535,
      "step": 5158
    },
    {
      "epoch": 3.94,
      "learning_rate": 4.5072563601422646e-10,
      "loss": 0.4557,
      "step": 5159
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.5612908475202154e-10,
      "loss": 0.4874,
      "step": 5160
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.726614569492636e-10,
      "loss": 0.5284,
      "step": 5161
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.0032278357118294e-10,
      "loss": 0.5237,
      "step": 5162
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.3911309145298034e-10,
      "loss": 0.5166,
      "step": 5163
    },
    {
      "epoch": 3.94,
      "learning_rate": 8.903240329982686e-11,
      "loss": 0.6528,
      "step": 5164
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.008073769352528e-11,
      "loss": 0.4902,
      "step": 5165
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.225810908251802e-11,
      "loss": 0.4511,
      "step": 5166
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.564527785217876e-12,
      "loss": 0.5146,
      "step": 5167
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.0,
      "loss": 0.5336,
      "step": 5168
    }
  ],
  "logging_steps": 1,
  "max_steps": 5168,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 1.5520882016885146e+18,
  "trial_name": null,
  "trial_params": null
}