diff --git "a/checkpoint-1152/trainer_state.json" "b/checkpoint-1152/trainer_state.json"
deleted file mode 100644--- "a/checkpoint-1152/trainer_state.json"
+++ /dev/null
@@ -1,7000 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.881203694988638,
-  "global_step": 1152,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.5625e-07,
-      "loss": 3.4895,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.125e-07,
-      "loss": 3.5428,
-      "step": 2
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.6875000000000006e-07,
-      "loss": 3.5726,
-      "step": 3
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.25e-07,
-      "loss": 3.6189,
-      "step": 4
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 7.8125e-07,
-      "loss": 3.511,
-      "step": 5
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 9.375000000000001e-07,
-      "loss": 3.4609,
-      "step": 6
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.0937500000000001e-06,
-      "loss": 3.4712,
-      "step": 7
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.25e-06,
-      "loss": 3.5399,
-      "step": 8
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.40625e-06,
-      "loss": 3.4956,
-      "step": 9
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.5625e-06,
-      "loss": 3.5031,
-      "step": 10
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.71875e-06,
-      "loss": 3.5565,
-      "step": 11
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.8750000000000003e-06,
-      "loss": 3.383,
-      "step": 12
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.0312500000000002e-06,
-      "loss": 3.3852,
-      "step": 13
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.1875000000000002e-06,
-      "loss": 3.3688,
-      "step": 14
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.3437500000000002e-06,
-      "loss": 3.3342,
-      "step": 15
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.5e-06,
-      "loss": 3.2689,
-      "step": 16
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.65625e-06,
-      "loss": 3.2455,
-      "step": 17
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.8125e-06,
-      "loss": 3.2564,
-      "step": 18
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.96875e-06,
-      "loss": 3.0662,
-      "step": 19
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3.125e-06,
-      "loss": 3.0245,
-      "step": 20
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3.28125e-06,
-      "loss": 2.9888,
-      "step": 21
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3.4375e-06,
-      "loss": 2.874,
-      "step": 22
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3.59375e-06,
-      "loss": 2.7424,
-      "step": 23
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3.7500000000000005e-06,
-      "loss": 2.7272,
-      "step": 24
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3.90625e-06,
-      "loss": 2.5789,
-      "step": 25
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 4.0625000000000005e-06,
-      "loss": 2.4029,
-      "step": 26
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 4.21875e-06,
-      "loss": 2.237,
-      "step": 27
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 4.3750000000000005e-06,
-      "loss": 2.1258,
-      "step": 28
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 4.53125e-06,
-      "loss": 1.9172,
-      "step": 29
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 4.6875000000000004e-06,
-      "loss": 1.8099,
-      "step": 30
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 4.84375e-06,
-      "loss": 1.7006,
-      "step": 31
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5e-06,
-      "loss": 1.5401,
-      "step": 32
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.156250000000001e-06,
-      "loss": 1.4624,
-      "step": 33
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.3125e-06,
-      "loss": 1.3664,
-      "step": 34
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.468750000000001e-06,
-      "loss": 1.2766,
-      "step": 35
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.625e-06,
-      "loss": 1.1971,
-      "step": 36
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.781250000000001e-06,
-      "loss": 1.1606,
-      "step": 37
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9375e-06,
-      "loss": 1.0934,
-      "step": 38
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 6.093750000000001e-06,
-      "loss": 1.0819,
-      "step": 39
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 6.25e-06,
-      "loss": 1.0678,
-      "step": 40
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 6.406250000000001e-06,
-      "loss": 1.0517,
-      "step": 41
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 6.5625e-06,
-      "loss": 1.0182,
-      "step": 42
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 6.718750000000001e-06,
-      "loss": 0.994,
-      "step": 43
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 6.875e-06,
-      "loss": 0.9798,
-      "step": 44
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 7.031250000000001e-06,
-      "loss": 0.9874,
-      "step": 45
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 7.1875e-06,
-      "loss": 0.9626,
-      "step": 46
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 7.343750000000001e-06,
-      "loss": 0.9631,
-      "step": 47
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 0.9222,
-      "step": 48
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 7.656250000000001e-06,
-      "loss": 0.9402,
-      "step": 49
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 7.8125e-06,
-      "loss": 0.9179,
-      "step": 50
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 7.96875e-06,
-      "loss": 0.9019,
-      "step": 51
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 8.125000000000001e-06,
-      "loss": 0.9063,
-      "step": 52
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 8.281250000000001e-06,
-      "loss": 0.9219,
-      "step": 53
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 8.4375e-06,
-      "loss": 0.9168,
-      "step": 54
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 8.59375e-06,
-      "loss": 0.8971,
-      "step": 55
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 8.750000000000001e-06,
-      "loss": 0.8797,
-      "step": 56
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 8.906250000000001e-06,
-      "loss": 0.9042,
-      "step": 57
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.0625e-06,
-      "loss": 0.8862,
-      "step": 58
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.21875e-06,
-      "loss": 0.8587,
-      "step": 59
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.375000000000001e-06,
-      "loss": 0.8852,
-      "step": 60
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.531250000000001e-06,
-      "loss": 0.8456,
-      "step": 61
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.6875e-06,
-      "loss": 0.8573,
-      "step": 62
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.84375e-06,
-      "loss": 0.8602,
-      "step": 63
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 1e-05,
-      "loss": 0.8527,
-      "step": 64
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.999984030281327e-06,
-      "loss": 0.8608,
-      "step": 65
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.999936121227315e-06,
-      "loss": 0.8444,
-      "step": 66
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.999856273144007e-06,
-      "loss": 0.8354,
-      "step": 67
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.99974448654146e-06,
-      "loss": 0.8175,
-      "step": 68
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.999600762133756e-06,
-      "loss": 0.8482,
-      "step": 69
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.99942510083899e-06,
-      "loss": 0.8259,
-      "step": 70
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.999217503779266e-06,
-      "loss": 0.8521,
-      "step": 71
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.998977972280691e-06,
-      "loss": 0.8328,
-      "step": 72
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.998706507873365e-06,
-      "loss": 0.8309,
-      "step": 73
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.998403112291374e-06,
-      "loss": 0.8383,
-      "step": 74
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.998067787472772e-06,
-      "loss": 0.8284,
-      "step": 75
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.997700535559575e-06,
-      "loss": 0.8049,
-      "step": 76
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.997301358897752e-06,
-      "loss": 0.8144,
-      "step": 77
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.996870260037196e-06,
-      "loss": 0.819,
-      "step": 78
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.996407241731718e-06,
-      "loss": 0.8077,
-      "step": 79
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.995912306939026e-06,
-      "loss": 0.813,
-      "step": 80
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.995385458820708e-06,
-      "loss": 0.8016,
-      "step": 81
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.994826700742211e-06,
-      "loss": 0.7992,
-      "step": 82
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.994236036272819e-06,
-      "loss": 0.8022,
-      "step": 83
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.993613469185631e-06,
-      "loss": 0.7922,
-      "step": 84
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.992959003457534e-06,
-      "loss": 0.8035,
-      "step": 85
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.992272643269181e-06,
-      "loss": 0.8092,
-      "step": 86
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.991554393004965e-06,
-      "loss": 0.8076,
-      "step": 87
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.990804257252987e-06,
-      "loss": 0.784,
-      "step": 88
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.99002224080503e-06,
-      "loss": 0.7834,
-      "step": 89
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.989208348656528e-06,
-      "loss": 0.8092,
-      "step": 90
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.988362586006531e-06,
-      "loss": 0.7941,
-      "step": 91
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.987484958257675e-06,
-      "loss": 0.7952,
-      "step": 92
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.986575471016152e-06,
-      "loss": 0.8039,
-      "step": 93
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.985634130091656e-06,
-      "loss": 0.8036,
-      "step": 94
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.984660941497375e-06,
-      "loss": 0.7878,
-      "step": 95
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.983655911449922e-06,
-      "loss": 0.7923,
-      "step": 96
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.982619046369321e-06,
-      "loss": 0.7793,
-      "step": 97
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.981550352878948e-06,
-      "loss": 0.7868,
-      "step": 98
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.980449837805495e-06,
-      "loss": 0.7756,
-      "step": 99
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.979317508178929e-06,
-      "loss": 0.7951,
-      "step": 100
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.978153371232444e-06,
-      "loss": 0.7923,
-      "step": 101
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.976957434402416e-06,
-      "loss": 0.7926,
-      "step": 102
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.975729705328356e-06,
-      "loss": 0.7874,
-      "step": 103
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.974470191852858e-06,
-      "loss": 0.7956,
-      "step": 104
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.973178902021555e-06,
-      "loss": 0.8077,
-      "step": 105
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.971855844083055e-06,
-      "loss": 0.8009,
-      "step": 106
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.97050102648891e-06,
-      "loss": 0.7801,
-      "step": 107
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.96911445789354e-06,
-      "loss": 0.7759,
-      "step": 108
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.967696147154187e-06,
-      "loss": 0.7591,
-      "step": 109
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.966246103330863e-06,
-      "loss": 0.766,
-      "step": 110
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.964764335686284e-06,
-      "loss": 0.8039,
-      "step": 111
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.963250853685813e-06,
-      "loss": 0.7847,
-      "step": 112
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.961705666997406e-06,
-      "loss": 0.7568,
-      "step": 113
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.96012878549154e-06,
-      "loss": 0.7781,
-      "step": 114
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.958520219241156e-06,
-      "loss": 0.7727,
-      "step": 115
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.956879978521596e-06,
-      "loss": 0.7913,
-      "step": 116
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.955208073810532e-06,
-      "loss": 0.7646,
-      "step": 117
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.953504515787902e-06,
-      "loss": 0.7628,
-      "step": 118
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.951769315335843e-06,
-      "loss": 0.7798,
-      "step": 119
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.950002483538626e-06,
-      "loss": 0.7794,
-      "step": 120
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.948204031682566e-06,
-      "loss": 0.7709,
-      "step": 121
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.946373971255975e-06,
-      "loss": 0.7729,
-      "step": 122
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.944512313949071e-06,
-      "loss": 0.7677,
-      "step": 123
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.942619071653914e-06,
-      "loss": 0.7687,
-      "step": 124
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.940694256464322e-06,
-      "loss": 0.794,
-      "step": 125
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.938737880675796e-06,
-      "loss": 0.7773,
-      "step": 126
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.936749956785446e-06,
-      "loss": 0.7839,
-      "step": 127
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.934730497491907e-06,
-      "loss": 0.774,
-      "step": 128
-    },
-    {
-      "epoch": 0.1,
-      "eval_loss": 1.3403148651123047,
-      "eval_runtime": 125.9471,
-      "eval_samples_per_second": 95.032,
-      "eval_steps_per_second": 23.764,
-      "step": 128
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.932679515695254e-06,
-      "loss": 0.7666,
-      "step": 129
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.930597024496933e-06,
-      "loss": 0.7804,
-      "step": 130
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.92848303719966e-06,
-      "loss": 0.7914,
-      "step": 131
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.926337567307346e-06,
-      "loss": 0.7687,
-      "step": 132
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.924160628525016e-06,
-      "loss": 0.738,
-      "step": 133
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.921952234758708e-06,
-      "loss": 0.7334,
-      "step": 134
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.919712400115393e-06,
-      "loss": 0.7717,
-      "step": 135
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.917441138902883e-06,
-      "loss": 0.7834,
-      "step": 136
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.915138465629737e-06,
-      "loss": 0.7625,
-      "step": 137
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.912804395005176e-06,
-      "loss": 0.7771,
-      "step": 138
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.910438941938977e-06,
-      "loss": 0.7398,
-      "step": 139
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.908042121541392e-06,
-      "loss": 0.7645,
-      "step": 140
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.905613949123036e-06,
-      "loss": 0.7697,
-      "step": 141
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.903154440194804e-06,
-      "loss": 0.7689,
-      "step": 142
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.90066361046776e-06,
-      "loss": 0.749,
-      "step": 143
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.898141475853047e-06,
-      "loss": 0.7712,
-      "step": 144
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.895588052461773e-06,
-      "loss": 0.7676,
-      "step": 145
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.893003356604923e-06,
-      "loss": 0.7765,
-      "step": 146
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.890387404793243e-06,
-      "loss": 0.7854,
-      "step": 147
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.887740213737136e-06,
-      "loss": 0.7951,
-      "step": 148
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.885061800346563e-06,
-      "loss": 0.7646,
-      "step": 149
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.882352181730925e-06,
-      "loss": 0.7664,
-      "step": 150
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.879611375198963e-06,
-      "loss": 0.7477,
-      "step": 151
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.87683939825864e-06,
-      "loss": 0.7686,
-      "step": 152
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.874036268617034e-06,
-      "loss": 0.7611,
-      "step": 153
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.871202004180221e-06,
-      "loss": 0.7664,
-      "step": 154
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.86833662305316e-06,
-      "loss": 0.7451,
-      "step": 155
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.865440143539589e-06,
-      "loss": 0.774,
-      "step": 156
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.86251258414189e-06,
-      "loss": 0.7695,
-      "step": 157
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.859553963560982e-06,
-      "loss": 0.7446,
-      "step": 158
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.856564300696201e-06,
-      "loss": 0.7797,
-      "step": 159
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.85354361464518e-06,
-      "loss": 0.7668,
-      "step": 160
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.850491924703716e-06,
-      "loss": 0.7617,
-      "step": 161
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.847409250365666e-06,
-      "loss": 0.7408,
-      "step": 162
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.844295611322804e-06,
-      "loss": 0.7612,
-      "step": 163
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.841151027464707e-06,
-      "loss": 0.7597,
-      "step": 164
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.837975518878622e-06,
-      "loss": 0.7566,
-      "step": 165
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.834769105849341e-06,
-      "loss": 0.7747,
-      "step": 166
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.83153180885907e-06,
-      "loss": 0.7662,
-      "step": 167
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.828263648587298e-06,
-      "loss": 0.7692,
-      "step": 168
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.824964645910664e-06,
-      "loss": 0.781,
-      "step": 169
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.821634821902825e-06,
-      "loss": 0.7557,
-      "step": 170
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.818274197834326e-06,
-      "loss": 0.7287,
-      "step": 171
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.814882795172451e-06,
-      "loss": 0.7196,
-      "step": 172
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.811460635581102e-06,
-      "loss": 0.7687,
-      "step": 173
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.808007740920647e-06,
-      "loss": 0.7525,
-      "step": 174
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.804524133247788e-06,
-      "loss": 0.7662,
-      "step": 175
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.80100983481542e-06,
-      "loss": 0.7616,
-      "step": 176
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.797464868072489e-06,
-      "loss": 0.7547,
-      "step": 177
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.793889255663838e-06,
-      "loss": 0.7661,
-      "step": 178
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.79028302043008e-06,
-      "loss": 0.7624,
-      "step": 179
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.786646185407438e-06,
-      "loss": 0.7511,
-      "step": 180
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.782978773827607e-06,
-      "loss": 0.7449,
-      "step": 181
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.7792808091176e-06,
-      "loss": 0.744,
-      "step": 182
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.775552314899596e-06,
-      "loss": 0.7497,
-      "step": 183
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.7717933149908e-06,
-      "loss": 0.7512,
-      "step": 184
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.768003833403278e-06,
-      "loss": 0.7412,
-      "step": 185
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.764183894343812e-06,
-      "loss": 0.7194,
-      "step": 186
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.760333522213746e-06,
-      "loss": 0.7585,
-      "step": 187
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.75645274160882e-06,
-      "loss": 0.7492,
-      "step": 188
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.752541577319026e-06,
-      "loss": 0.7214,
-      "step": 189
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.74860005432844e-06,
-      "loss": 0.7406,
-      "step": 190
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.744628197815068e-06,
-      "loss": 0.7385,
-      "step": 191
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.740626033150683e-06,
-      "loss": 0.7513,
-      "step": 192
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.736593585900662e-06,
-      "loss": 0.736,
-      "step": 193
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.732530881823825e-06,
-      "loss": 0.7295,
-      "step": 194
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.728437946872266e-06,
-      "loss": 0.7546,
-      "step": 195
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.724314807191197e-06,
-      "loss": 0.7377,
-      "step": 196
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.720161489118766e-06,
-      "loss": 0.7383,
-      "step": 197
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.715978019185904e-06,
-      "loss": 0.735,
-      "step": 198
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.711764424116146e-06,
-      "loss": 0.7486,
-      "step": 199
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.707520730825461e-06,
-      "loss": 0.7599,
-      "step": 200
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.703246966422088e-06,
-      "loss": 0.7608,
-      "step": 201
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.698943158206351e-06,
-      "loss": 0.7457,
-      "step": 202
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.694609333670493e-06,
-      "loss": 0.7602,
-      "step": 203
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.690245520498496e-06,
-      "loss": 0.7467,
-      "step": 204
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.685851746565908e-06,
-      "loss": 0.7517,
-      "step": 205
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.681428039939664e-06,
-      "loss": 0.7466,
-      "step": 206
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.6769744288779e-06,
-      "loss": 0.7464,
-      "step": 207
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.67249094182979e-06,
-      "loss": 0.7468,
-      "step": 208
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.667977607435338e-06,
-      "loss": 0.7412,
-      "step": 209
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.663434454525218e-06,
-      "loss": 0.7558,
-      "step": 210
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.65886151212058e-06,
-      "loss": 0.7522,
-      "step": 211
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.654258809432865e-06,
-      "loss": 0.7825,
-      "step": 212
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.649626375863622e-06,
-      "loss": 0.7422,
-      "step": 213
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.644964241004312e-06,
-      "loss": 0.749,
-      "step": 214
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.64027243463613e-06,
-      "loss": 0.7458,
-      "step": 215
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.635550986729808e-06,
-      "loss": 0.7273,
-      "step": 216
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.63079992744542e-06,
-      "loss": 0.7418,
-      "step": 217
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.626019287132202e-06,
-      "loss": 0.7609,
-      "step": 218
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.621209096328344e-06,
-      "loss": 0.7546,
-      "step": 219
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.616369385760805e-06,
-      "loss": 0.7443,
-      "step": 220
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.611500186345112e-06,
-      "loss": 0.7614,
-      "step": 221
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.60660152918516e-06,
-      "loss": 0.7206,
-      "step": 222
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.601673445573021e-06,
-      "loss": 0.7414,
-      "step": 223
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.59671596698874e-06,
-      "loss": 0.7512,
-      "step": 224
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.591729125100133e-06,
-      "loss": 0.7415,
-      "step": 225
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.586712951762583e-06,
-      "loss": 0.736,
-      "step": 226
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.58166747901884e-06,
-      "loss": 0.7469,
-      "step": 227
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.576592739098816e-06,
-      "loss": 0.7419,
-      "step": 228
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.571488764419381e-06,
-      "loss": 0.732,
-      "step": 229
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.566355587584149e-06,
-      "loss": 0.7464,
-      "step": 230
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.561193241383277e-06,
-      "loss": 0.734,
-      "step": 231
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.556001758793252e-06,
-      "loss": 0.7277,
-      "step": 232
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.550781172976679e-06,
-      "loss": 0.7487,
-      "step": 233
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.545531517282074e-06,
-      "loss": 0.7384,
-      "step": 234
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.540252825243646e-06,
-      "loss": 0.7403,
-      "step": 235
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.534945130581087e-06,
-      "loss": 0.7224,
-      "step": 236
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.529608467199351e-06,
-      "loss": 0.7539,
-      "step": 237
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.524242869188445e-06,
-      "loss": 0.7423,
-      "step": 238
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.518848370823205e-06,
-      "loss": 0.727,
-      "step": 239
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.51342500656308e-06,
-      "loss": 0.7174,
-      "step": 240
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.507972811051909e-06,
-      "loss": 0.7391,
-      "step": 241
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.502491819117702e-06,
-      "loss": 0.7353,
-      "step": 242
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.496982065772425e-06,
-      "loss": 0.728,
-      "step": 243
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.491443586211756e-06,
-      "loss": 0.7333,
-      "step": 244
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.485876415814882e-06,
-      "loss": 0.749,
-      "step": 245
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.480280590144261e-06,
-      "loss": 0.7399,
-      "step": 246
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.474656144945397e-06,
-      "loss": 0.7333,
-      "step": 247
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.469003116146613e-06,
-      "loss": 0.7343,
-      "step": 248
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.463321539858821e-06,
-      "loss": 0.75,
-      "step": 249
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.45761145237529e-06,
-      "loss": 0.7316,
-      "step": 250
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.451872890171419e-06,
-      "loss": 0.7311,
-      "step": 251
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.446105889904496e-06,
-      "loss": 0.7395,
-      "step": 252
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.440310488413469e-06,
-      "loss": 0.7362,
-      "step": 253
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.434486722718712e-06,
-      "loss": 0.74,
-      "step": 254
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.428634630021783e-06,
-      "loss": 0.7308,
-      "step": 255
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.422754247705192e-06,
-      "loss": 0.7276,
-      "step": 256
-    },
-    {
-      "epoch": 0.2,
-      "eval_loss": 1.3016690015792847,
-      "eval_runtime": 124.5394,
-      "eval_samples_per_second": 96.106,
-      "eval_steps_per_second": 24.033,
-      "step": 256
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.416845613332162e-06,
-      "loss": 0.7373,
-      "step": 257
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.410908764646383e-06,
-      "loss": 0.7288,
-      "step": 258
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.404943739571774e-06,
-      "loss": 0.73,
-      "step": 259
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.398950576212249e-06,
-      "loss": 0.7479,
-      "step": 260
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.392929312851455e-06,
-      "loss": 0.7384,
-      "step": 261
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.386879987952549e-06,
-      "loss": 0.745,
-      "step": 262
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.380802640157937e-06,
-      "loss": 0.7251,
-      "step": 263
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.374697308289034e-06,
-      "loss": 0.7368,
-      "step": 264
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.36856403134601e-06,
-      "loss": 0.7431,
-      "step": 265
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.362402848507548e-06,
-      "loss": 0.7431,
-      "step": 266
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.356213799130594e-06,
-      "loss": 0.7181,
-      "step": 267
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.349996922750096e-06,
-      "loss": 0.7467,
-      "step": 268
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.343752259078761e-06,
-      "loss": 0.7489,
-      "step": 269
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.337479848006799e-06,
-      "loss": 0.7277,
-      "step": 270
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.331179729601665e-06,
-      "loss": 0.7474,
-      "step": 271
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.324851944107809e-06,
-      "loss": 0.7171,
-      "step": 272
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.318496531946411e-06,
-      "loss": 0.7664,
-      "step": 273
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.312113533715125e-06,
-      "loss": 0.7243,
-      "step": 274
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.305702990187831e-06,
-      "loss": 0.7242,
-      "step": 275
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.299264942314358e-06,
-      "loss": 0.7237,
-      "step": 276
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.292799431220229e-06,
-      "loss": 0.7108,
-      "step": 277
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.286306498206405e-06,
-      "loss": 0.7467,
-      "step": 278
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.27978618474901e-06,
-      "loss": 0.7445,
-      "step": 279
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.273238532499068e-06,
-      "loss": 0.7154,
-      "step": 280
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.266663583282254e-06,
-      "loss": 0.733,
-      "step": 281
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.260061379098596e-06,
-      "loss": 0.7206,
-      "step": 282
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.253431962122234e-06,
-      "loss": 0.7346,
-      "step": 283
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.246775374701139e-06,
-      "loss": 0.728,
-      "step": 284
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.24009165935684e-06,
-      "loss": 0.7297,
-      "step": 285
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.23338085878416e-06,
-      "loss": 0.7326,
-      "step": 286
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.226643015850938e-06,
-      "loss": 0.7304,
-      "step": 287
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.219878173597755e-06,
-      "loss": 0.7175,
-      "step": 288
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.213086375237662e-06,
-      "loss": 0.7307,
-      "step": 289
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.206267664155906e-06,
-      "loss": 0.7078,
-      "step": 290
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.199422083909644e-06,
-      "loss": 0.7404,
-      "step": 291
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.19254967822767e-06,
-      "loss": 0.7197,
-      "step": 292
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.18565049101014e-06,
-      "loss": 0.7303,
-      "step": 293
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.178724566328288e-06,
-      "loss": 0.7415,
-      "step": 294
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.171771948424138e-06,
-      "loss": 0.7208,
-      "step": 295
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.164792681710231e-06,
-      "loss": 0.7306,
-      "step": 296
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.157786810769338e-06,
-      "loss": 0.7161,
-      "step": 297
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.150754380354176e-06,
-      "loss": 0.7241,
-      "step": 298
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.143695435387117e-06,
-      "loss": 0.7336,
-      "step": 299
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.136610020959909e-06,
-      "loss": 0.7359,
-      "step": 300
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.129498182333379e-06,
-      "loss": 0.7217,
-      "step": 301
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.122359964937154e-06,
-      "loss": 0.7304,
-      "step": 302
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.115195414369365e-06,
-      "loss": 0.7186,
-      "step": 303
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.108004576396352e-06,
-      "loss": 0.7263,
-      "step": 304
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.100787496952377e-06,
-      "loss": 0.7244,
-      "step": 305
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.093544222139338e-06,
-      "loss": 0.7333,
-      "step": 306
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.086274798226453e-06,
-      "loss": 0.7251,
-      "step": 307
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.078979271649988e-06,
-      "loss": 0.7352,
-      "step": 308
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.071657689012944e-06,
-      "loss": 0.7333,
-      "step": 309
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.064310097084766e-06,
-      "loss": 0.7597,
-      "step": 310
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.056936542801047e-06,
-      "loss": 0.7346,
-      "step": 311
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.049537073263218e-06,
-      "loss": 0.7216,
-      "step": 312
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.042111735738266e-06,
-      "loss": 0.7137,
-      "step": 313
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.0346605776584e-06,
-      "loss": 0.7367,
-      "step": 314
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.027183646620789e-06,
-      "loss": 0.7105,
-      "step": 315
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.01968099038722e-06,
-      "loss": 0.7175,
-      "step": 316
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.012152656883824e-06,
-      "loss": 0.7301,
-      "step": 317
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.00459869420074e-06,
-      "loss": 0.741,
-      "step": 318
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 8.99701915059184e-06,
-      "loss": 0.7401,
-      "step": 319
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 8.98941407447439e-06,
-      "loss": 0.7176,
-      "step": 320
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.981783514428762e-06,
-      "loss": 0.7192,
-      "step": 321
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.974127519198115e-06,
-      "loss": 0.7188,
-      "step": 322
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.966446137688086e-06,
-      "loss": 0.7165,
-      "step": 323
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.958739418966473e-06,
-      "loss": 0.735,
-      "step": 324
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.951007412262928e-06,
-      "loss": 0.7481,
-      "step": 325
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.943250166968645e-06,
-      "loss": 0.7261,
-      "step": 326
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.935467732636027e-06,
-      "loss": 0.7273,
-      "step": 327
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.927660158978392e-06,
-      "loss": 0.7476,
-      "step": 328
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.919827495869645e-06,
-      "loss": 0.7371,
-      "step": 329
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.911969793343953e-06,
-      "loss": 0.7171,
-      "step": 330
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.904087101595435e-06,
-      "loss": 0.7262,
-      "step": 331
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.89617947097784e-06,
-      "loss": 0.7197,
-      "step": 332
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.88824695200422e-06,
-      "loss": 0.7085,
-      "step": 333
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.88028959534662e-06,
-      "loss": 0.7312,
-      "step": 334
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.872307451835733e-06,
-      "loss": 0.711,
-      "step": 335
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.864300572460596e-06,
-      "loss": 0.728,
-      "step": 336
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.85626900836825e-06,
-      "loss": 0.7274,
-      "step": 337
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.84821281086343e-06,
-      "loss": 0.7384,
-      "step": 338
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.84013203140821e-06,
-      "loss": 0.7394,
-      "step": 339
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.832026721621709e-06,
-      "loss": 0.7384,
-      "step": 340
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.823896933279725e-06,
-      "loss": 0.7205,
-      "step": 341
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.815742718314438e-06,
-      "loss": 0.7282,
-      "step": 342
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.807564128814052e-06,
-      "loss": 0.7218,
-      "step": 343
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.799361217022478e-06,
-      "loss": 0.707,
-      "step": 344
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.791134035338991e-06,
-      "loss": 0.7184,
-      "step": 345
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.782882636317904e-06,
-      "loss": 0.7312,
-      "step": 346
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.774607072668225e-06,
-      "loss": 0.7238,
-      "step": 347
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.766307397253325e-06,
-      "loss": 0.7235,
-      "step": 348
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.757983663090593e-06,
-      "loss": 0.7208,
-      "step": 349
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.749635923351108e-06,
-      "loss": 0.7349,
-      "step": 350
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.741264231359293e-06,
-      "loss": 0.7305,
-      "step": 351
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.732868640592573e-06,
-      "loss": 0.7351,
-      "step": 352
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.724449204681036e-06,
-      "loss": 0.7337,
-      "step": 353
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.716005977407094e-06,
-      "loss": 0.7212,
-      "step": 354
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.707539012705133e-06,
-      "loss": 0.729,
-      "step": 355
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.699048364661167e-06,
-      "loss": 0.721,
-      "step": 356
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.690534087512502e-06,
-      "loss": 0.7261,
-      "step": 357
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.681996235647385e-06,
-      "loss": 0.7331,
-      "step": 358
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.67343486360465e-06,
-      "loss": 0.7045,
-      "step": 359
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.664850026073376e-06,
-      "loss": 0.7315,
-      "step": 360
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.656241777892544e-06,
-      "loss": 0.7326,
-      "step": 361
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.647610174050672e-06,
-      "loss": 0.7477,
-      "step": 362
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.638955269685475e-06,
-      "loss": 0.7128,
-      "step": 363
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.630277120083508e-06,
-      "loss": 0.7217,
-      "step": 364
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.621575780679814e-06,
-      "loss": 0.7161,
-      "step": 365
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.612851307057571e-06,
-      "loss": 0.7037,
-      "step": 366
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.604103754947731e-06,
-      "loss": 0.7301,
-      "step": 367
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.595333180228676e-06,
-      "loss": 0.724,
-      "step": 368
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.586539638925851e-06,
-      "loss": 0.7403,
-      "step": 369
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.577723187211404e-06,
-      "loss": 0.7171,
-      "step": 370
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.56888388140384e-06,
-      "loss": 0.7115,
-      "step": 371
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.56002177796765e-06,
-      "loss": 0.7171,
-      "step": 372
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.551136933512952e-06,
-      "loss": 0.7378,
-      "step": 373
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.542229404795133e-06,
-      "loss": 0.7304,
-      "step": 374
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.533299248714483e-06,
-      "loss": 0.722,
-      "step": 375
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.524346522315836e-06,
-      "loss": 0.7333,
-      "step": 376
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.515371282788201e-06,
-      "loss": 0.7387,
-      "step": 377
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.506373587464396e-06,
-      "loss": 0.7309,
-      "step": 378
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.497353493820688e-06,
-      "loss": 0.7183,
-      "step": 379
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.488311059476419e-06,
-      "loss": 0.7508,
-      "step": 380
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.479246342193643e-06,
-      "loss": 0.7139,
-      "step": 381
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.470159399876753e-06,
-      "loss": 0.7217,
-      "step": 382
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.461050290572114e-06,
-      "loss": 0.7212,
-      "step": 383
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.451919072467694e-06,
-      "loss": 0.7302,
-      "step": 384
-    },
-    {
-      "epoch": 0.29,
-      "eval_loss": 1.2867591381072998,
-      "eval_runtime": 125.743,
-      "eval_samples_per_second": 95.186,
-      "eval_steps_per_second": 23.803,
-      "step": 384
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.442765803892683e-06,
-      "loss": 0.7221,
-      "step": 385
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.433590543317132e-06,
-      "loss": 0.7119,
-      "step": 386
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.424393349351573e-06,
-      "loss": 0.7325,
-      "step": 387
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.415174280746645e-06,
-      "loss": 0.7297,
-      "step": 388
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.405933396392722e-06,
-      "loss": 0.7269,
-      "step": 389
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.396670755319537e-06,
-      "loss": 0.7146,
-      "step": 390
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.38738641669579e-06,
-      "loss": 0.6906,
-      "step": 391
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.378080439828799e-06,
-      "loss": 0.7537,
-      "step": 392
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.368752884164096e-06,
-      "loss": 0.7129,
-      "step": 393
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.359403809285054e-06,
-      "loss": 0.7295,
-      "step": 394
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.350033274912512e-06,
-      "loss": 0.7289,
-      "step": 395
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.340641340904391e-06,
-      "loss": 0.7383,
-      "step": 396
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.33122806725531e-06,
-      "loss": 0.724,
-      "step": 397
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.321793514096195e-06,
-      "loss": 0.7146,
-      "step": 398
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.312337741693917e-06,
-      "loss": 0.7437,
-      "step": 399
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.30286081045088e-06,
-      "loss": 0.7366,
-      "step": 400
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.293362780904662e-06,
-      "loss": 0.719,
-      "step": 401
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.2838437137276e-06,
-      "loss": 0.7085,
-      "step": 402
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.274303669726427e-06,
-      "loss": 0.7247,
-      "step": 403
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.264742709841869e-06,
-      "loss": 0.7299,
-      "step": 404
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.255160895148263e-06,
-      "loss": 0.7367,
-      "step": 405
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.245558286853164e-06,
-      "loss": 0.706,
-      "step": 406
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.23593494629695e-06,
-      "loss": 0.7351,
-      "step": 407
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.226290934952442e-06,
-      "loss": 0.7173,
-      "step": 408
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.216626314424496e-06,
-      "loss": 0.7385,
-      "step": 409
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.206941146449621e-06,
-      "loss": 0.7226,
-      "step": 410
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.19723549289558e-06,
-      "loss": 0.7345,
-      "step": 411
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.187509415760996e-06,
-      "loss": 0.7178,
-      "step": 412
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.177762977174956e-06,
-      "loss": 0.7214,
-      "step": 413
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.167996239396611e-06,
-      "loss": 0.7212,
-      "step": 414
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.158209264814785e-06,
-      "loss": 0.7257,
-      "step": 415
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.14840211594757e-06,
-      "loss": 0.6969,
-      "step": 416
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.138574855441929e-06,
-      "loss": 0.7235,
-      "step": 417
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.128727546073296e-06,
-      "loss": 0.7091,
-      "step": 418
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.118860250745173e-06,
-      "loss": 0.7068,
-      "step": 419
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.108973032488736e-06,
-      "loss": 0.7238,
-      "step": 420
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.099065954462422e-06,
-      "loss": 0.712,
-      "step": 421
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.089139079951526e-06,
-      "loss": 0.7036,
-      "step": 422
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.079192472367812e-06,
-      "loss": 0.7038,
-      "step": 423
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.069226195249087e-06,
-      "loss": 0.7143,
-      "step": 424
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.059240312258806e-06,
-      "loss": 0.7213,
-      "step": 425
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.049234887185667e-06,
-      "loss": 0.7113,
-      "step": 426
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.039209983943201e-06,
-      "loss": 0.7297,
-      "step": 427
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.029165666569361e-06,
-      "loss": 0.7215,
-      "step": 428
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.019101999226115e-06,
-      "loss": 0.7124,
-      "step": 429
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.009019046199036e-06,
-      "loss": 0.7121,
-      "step": 430
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.998916871896899e-06,
-      "loss": 0.738,
-      "step": 431
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.988795540851252e-06,
-      "loss": 0.7132,
-      "step": 432
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.978655117716018e-06,
-      "loss": 0.7056,
-      "step": 433
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.968495667267084e-06,
-      "loss": 0.7335,
-      "step": 434
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.958317254401871e-06,
-      "loss": 0.6961,
-      "step": 435
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.948119944138939e-06,
-      "loss": 0.7181,
-      "step": 436
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 7.937903801617555e-06,
-      "loss": 0.7185,
-      "step": 437
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.927668892097288e-06,
-      "loss": 0.6965,
-      "step": 438
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.917415280957592e-06,
-      "loss": 0.7143,
-      "step": 439
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.907143033697378e-06,
-      "loss": 0.6857,
-      "step": 440
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.896852215934605e-06,
-      "loss": 0.7419,
-      "step": 441
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.886542893405861e-06,
-      "loss": 0.7311,
-      "step": 442
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.876215131965938e-06,
-      "loss": 0.7175,
-      "step": 443
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.865868997587414e-06,
-      "loss": 0.7144,
-      "step": 444
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.855504556360227e-06,
-      "loss": 0.7097,
-      "step": 445
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.845121874491266e-06,
-      "loss": 0.7379,
-      "step": 446
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.834721018303935e-06,
-      "loss": 0.7051,
-      "step": 447
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.82430205423773e-06,
-      "loss": 0.7274,
-      "step": 448
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.81386504884782e-06,
-      "loss": 0.7252,
-      "step": 449
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.803410068804625e-06,
-      "loss": 0.7231,
-      "step": 450
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.792937180893377e-06,
-      "loss": 0.7079,
-      "step": 451
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.782446452013709e-06,
-      "loss": 0.7275,
-      "step": 452
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.771937949179214e-06,
-      "loss": 0.7096,
-      "step": 453
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.761411739517026e-06,
-      "loss": 0.7233,
-      "step": 454
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.750867890267391e-06,
-      "loss": 0.7055,
-      "step": 455
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.740306468783226e-06,
-      "loss": 0.7185,
-      "step": 456
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.729727542529707e-06,
-      "loss": 0.7363,
-      "step": 457
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.719131179083822e-06,
-      "loss": 0.7054,
-      "step": 458
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.708517446133951e-06,
-      "loss": 0.7074,
-      "step": 459
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.697886411479422e-06,
-      "loss": 0.7043,
-      "step": 460
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.687238143030094e-06,
-      "loss": 0.7194,
-      "step": 461
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.676572708805902e-06,
-      "loss": 0.707,
-      "step": 462
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.665890176936441e-06,
-      "loss": 0.7249,
-      "step": 463
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.655190615660524e-06,
-      "loss": 0.7069,
-      "step": 464
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.644474093325743e-06,
-      "loss": 0.7066,
-      "step": 465
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.633740678388036e-06,
-      "loss": 0.73,
-      "step": 466
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.622990439411251e-06,
-      "loss": 0.6964,
-      "step": 467
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.612223445066706e-06,
-      "loss": 0.7315,
-      "step": 468
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.6014397641327474e-06,
-      "loss": 0.7227,
-      "step": 469
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.590639465494316e-06,
-      "loss": 0.722,
-      "step": 470
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.579822618142505e-06,
-      "loss": 0.7132,
-      "step": 471
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.568989291174118e-06,
-      "loss": 0.7097,
-      "step": 472
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.558139553791227e-06,
-      "loss": 0.7008,
-      "step": 473
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.547273475300734e-06,
-      "loss": 0.6987,
-      "step": 474
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.536391125113926e-06,
-      "loss": 0.688,
-      "step": 475
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.5254925727460335e-06,
-      "loss": 0.7042,
-      "step": 476
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.514577887815779e-06,
-      "loss": 0.7144,
-      "step": 477
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.503647140044943e-06,
-      "loss": 0.7026,
-      "step": 478
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.492700399257913e-06,
-      "loss": 0.7263,
-      "step": 479
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.481737735381236e-06,
-      "loss": 0.7163,
-      "step": 480
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.470759218443177e-06,
-      "loss": 0.6961,
-      "step": 481
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.459764918573264e-06,
-      "loss": 0.6972,
-      "step": 482
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.44875490600185e-06,
-      "loss": 0.7343,
-      "step": 483
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.437729251059657e-06,
-      "loss": 0.7344,
-      "step": 484
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.426688024177326e-06,
-      "loss": 0.7145,
-      "step": 485
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.415631295884972e-06,
-      "loss": 0.6957,
-      "step": 486
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.40455913681173e-06,
-      "loss": 0.7303,
-      "step": 487
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.393471617685308e-06,
-      "loss": 0.712,
-      "step": 488
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.382368809331533e-06,
-      "loss": 0.6974,
-      "step": 489
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.37125078267389e-06,
-      "loss": 0.7112,
-      "step": 490
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.360117608733084e-06,
-      "loss": 0.7222,
-      "step": 491
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.3489693586265785e-06,
-      "loss": 0.7211,
-      "step": 492
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.3378061035681415e-06,
-      "loss": 0.7099,
-      "step": 493
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.326627914867388e-06,
-      "loss": 0.6936,
-      "step": 494
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.315434863929332e-06,
-      "loss": 0.7165,
-      "step": 495
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.304227022253918e-06,
-      "loss": 0.7313,
-      "step": 496
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.2930044614355844e-06,
-      "loss": 0.7184,
-      "step": 497
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.281767253162781e-06,
-      "loss": 0.721,
-      "step": 498
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.2705154692175335e-06,
-      "loss": 0.7114,
-      "step": 499
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.25924918147497e-06,
-      "loss": 0.7219,
-      "step": 500
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.24796846190287e-06,
-      "loss": 0.7179,
-      "step": 501
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.236673382561199e-06,
-      "loss": 0.7096,
-      "step": 502
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.225364015601655e-06,
-      "loss": 0.7087,
-      "step": 503
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.2140404332671986e-06,
-      "loss": 0.7304,
-      "step": 504
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.2027027078916e-06,
-      "loss": 0.7258,
-      "step": 505
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.191350911898975e-06,
-      "loss": 0.7023,
-      "step": 506
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.17998511780332e-06,
-      "loss": 0.678,
-      "step": 507
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.1686053982080454e-06,
-      "loss": 0.6853,
-      "step": 508
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.157211825805519e-06,
-      "loss": 0.7079,
-      "step": 509
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.1458044733766026e-06,
-      "loss": 0.7136,
-      "step": 510
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.134383413790178e-06,
-      "loss": 0.7125,
-      "step": 511
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.122948720002689e-06,
-      "loss": 0.6938,
-      "step": 512
-    },
-    {
-      "epoch": 0.39,
-      "eval_loss": 1.2771650552749634,
-      "eval_runtime": 125.1558,
-      "eval_samples_per_second": 95.633,
-      "eval_steps_per_second": 23.914,
-      "step": 512
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.111500465057673e-06,
-      "loss": 0.7068,
-      "step": 513
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.100038722085291e-06,
-      "loss": 0.7007,
-      "step": 514
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.088563564301874e-06,
-      "loss": 0.7191,
-      "step": 515
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.0770750650094335e-06,
-      "loss": 0.7086,
-      "step": 516
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.065573297595211e-06,
-      "loss": 0.7097,
-      "step": 517
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.054058335531203e-06,
-      "loss": 0.7117,
-      "step": 518
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.0425302523736906e-06,
-      "loss": 0.715,
-      "step": 519
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.030989121762773e-06,
-      "loss": 0.6838,
-      "step": 520
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.019435017421894e-06,
-      "loss": 0.7068,
-      "step": 521
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.0078680131573704e-06,
-      "loss": 0.7363,
-      "step": 522
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.996288182857924e-06,
-      "loss": 0.7056,
-      "step": 523
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.98469560049421e-06,
-      "loss": 0.717,
-      "step": 524
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.9730903401183384e-06,
-      "loss": 0.73,
-      "step": 525
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.961472475863406e-06,
-      "loss": 0.7257,
-      "step": 526
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.949842081943021e-06,
-      "loss": 0.7016,
-      "step": 527
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.938199232650834e-06,
-      "loss": 0.7329,
-      "step": 528
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 6.926544002360055e-06,
-      "loss": 0.7395,
-      "step": 529
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.914876465522984e-06,
-      "loss": 0.7116,
-      "step": 530
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.9031966966705326e-06,
-      "loss": 0.7127,
-      "step": 531
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.89150477041175e-06,
-      "loss": 0.7076,
-      "step": 532
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.879800761433345e-06,
-      "loss": 0.7042,
-      "step": 533
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.868084744499211e-06,
-      "loss": 0.7058,
-      "step": 534
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.856356794449945e-06,
-      "loss": 0.7024,
-      "step": 535
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.844616986202371e-06,
-      "loss": 0.7031,
-      "step": 536
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.832865394749065e-06,
-      "loss": 0.6902,
-      "step": 537
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.821102095157871e-06,
-      "loss": 0.6961,
-      "step": 538
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.809327162571422e-06,
-      "loss": 0.7104,
-      "step": 539
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.797540672206661e-06,
-      "loss": 0.7076,
-      "step": 540
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.785742699354365e-06,
-      "loss": 0.7202,
-      "step": 541
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 6.773933319378656e-06,
-      "loss": 0.6869,
-      "step": 542
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.7621126077165235e-06,
-      "loss": 0.7141,
-      "step": 543
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.750280639877343e-06,
-      "loss": 0.7161,
-      "step": 544
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.738437491442395e-06,
-      "loss": 0.6855,
-      "step": 545
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.726583238064379e-06,
-      "loss": 0.6893,
-      "step": 546
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.714717955466931e-06,
-      "loss": 0.7095,
-      "step": 547
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.702841719444141e-06,
-      "loss": 0.7148,
-      "step": 548
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.690954605860068e-06,
-      "loss": 0.7168,
-      "step": 549
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.679056690648256e-06,
-      "loss": 0.703,
-      "step": 550
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.667148049811249e-06,
-      "loss": 0.7157,
-      "step": 551
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.655228759420105e-06,
-      "loss": 0.6975,
-      "step": 552
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.643298895613909e-06,
-      "loss": 0.7071,
-      "step": 553
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.631358534599288e-06,
-      "loss": 0.699,
-      "step": 554
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.619407752649926e-06,
-      "loss": 0.7278,
-      "step": 555
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.607446626106071e-06,
-      "loss": 0.7217,
-      "step": 556
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.595475231374057e-06,
-      "loss": 0.7182,
-      "step": 557
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.583493644925803e-06,
-      "loss": 0.7345,
-      "step": 558
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.571501943298335e-06,
-      "loss": 0.7009,
-      "step": 559
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.559500203093296e-06,
-      "loss": 0.7243,
-      "step": 560
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.547488500976451e-06,
-      "loss": 0.722,
-      "step": 561
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.535466913677201e-06,
-      "loss": 0.7009,
-      "step": 562
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.52343551798809e-06,
-      "loss": 0.7066,
-      "step": 563
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.511394390764326e-06,
-      "loss": 0.7046,
-      "step": 564
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.4993436089232705e-06,
-      "loss": 0.6972,
-      "step": 565
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.487283249443963e-06,
-      "loss": 0.7082,
-      "step": 566
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.475213389366623e-06,
-      "loss": 0.7205,
-      "step": 567
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.463134105792155e-06,
-      "loss": 0.7044,
-      "step": 568
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.451045475881668e-06,
-      "loss": 0.6977,
-      "step": 569
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.4389475768559675e-06,
-      "loss": 0.697,
-      "step": 570
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.426840485995072e-06,
-      "loss": 0.6949,
-      "step": 571
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.414724280637713e-06,
-      "loss": 0.6855,
-      "step": 572
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.402599038180849e-06,
-      "loss": 0.709,
-      "step": 573
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.3904648360791645e-06,
-      "loss": 0.7135,
-      "step": 574
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.378321751844578e-06,
-      "loss": 0.7113,
-      "step": 575
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.36616986304574e-06,
-      "loss": 0.7191,
-      "step": 576
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.354009247307556e-06,
-      "loss": 0.7105,
-      "step": 577
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.341839982310665e-06,
-      "loss": 0.6997,
-      "step": 578
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.329662145790964e-06,
-      "loss": 0.703,
-      "step": 579
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.317475815539104e-06,
-      "loss": 0.6934,
-      "step": 580
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.305281069399989e-06,
-      "loss": 0.7149,
-      "step": 581
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.293077985272288e-06,
-      "loss": 0.6968,
-      "step": 582
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.280866641107926e-06,
-      "loss": 0.7212,
-      "step": 583
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.268647114911599e-06,
-      "loss": 0.7255,
-      "step": 584
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.25641948474026e-06,
-      "loss": 0.7096,
-      "step": 585
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.244183828702639e-06,
-      "loss": 0.6911,
-      "step": 586
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.23194022495873e-06,
-      "loss": 0.7146,
-      "step": 587
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.219688751719293e-06,
-      "loss": 0.7078,
-      "step": 588
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.2074294872453626e-06,
-      "loss": 0.7111,
-      "step": 589
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.195162509847738e-06,
-      "loss": 0.7068,
-      "step": 590
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.182887897886495e-06,
-      "loss": 0.7055,
-      "step": 591
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.17060572977047e-06,
-      "loss": 0.7064,
-      "step": 592
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.158316083956772e-06,
-      "loss": 0.6935,
-      "step": 593
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.146019038950275e-06,
-      "loss": 0.7179,
-      "step": 594
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.13371467330312e-06,
-      "loss": 0.7198,
-      "step": 595
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.121403065614207e-06,
-      "loss": 0.7302,
-      "step": 596
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.109084294528704e-06,
-      "loss": 0.7056,
-      "step": 597
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.096758438737534e-06,
-      "loss": 0.7166,
-      "step": 598
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.084425576976873e-06,
-      "loss": 0.7053,
-      "step": 599
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.072085788027659e-06,
-      "loss": 0.6997,
-      "step": 600
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.059739150715073e-06,
-      "loss": 0.7226,
-      "step": 601
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.047385743908045e-06,
-      "loss": 0.7111,
-      "step": 602
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.035025646518747e-06,
-      "loss": 0.7113,
-      "step": 603
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.02265893750209e-06,
-      "loss": 0.7229,
-      "step": 604
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.010285695855222e-06,
-      "loss": 0.7102,
-      "step": 605
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 5.9979060006170164e-06,
-      "loss": 0.7292,
-      "step": 606
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 5.9855199308675724e-06,
-      "loss": 0.7001,
-      "step": 607
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.973127565727711e-06,
-      "loss": 0.7106,
-      "step": 608
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.960728984358466e-06,
-      "loss": 0.7014,
-      "step": 609
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.948324265960581e-06,
-      "loss": 0.7062,
-      "step": 610
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.935913489774e-06,
-      "loss": 0.7019,
-      "step": 611
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.923496735077364e-06,
-      "loss": 0.6939,
-      "step": 612
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.9110740811875055e-06,
-      "loss": 0.7224,
-      "step": 613
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.898645607458941e-06,
-      "loss": 0.7202,
-      "step": 614
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.886211393283359e-06,
-      "loss": 0.7187,
-      "step": 615
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.873771518089124e-06,
-      "loss": 0.7084,
-      "step": 616
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.861326061340756e-06,
-      "loss": 0.7274,
-      "step": 617
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.848875102538435e-06,
-      "loss": 0.7115,
-      "step": 618
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.836418721217483e-06,
-      "loss": 0.6843,
-      "step": 619
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 5.823956996947861e-06,
-      "loss": 0.7063,
-      "step": 620
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.811490009333662e-06,
-      "loss": 0.7013,
-      "step": 621
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.7990178380126026e-06,
-      "loss": 0.7141,
-      "step": 622
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.786540562655507e-06,
-      "loss": 0.7127,
-      "step": 623
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.774058262965806e-06,
-      "loss": 0.7009,
-      "step": 624
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.761571018679025e-06,
-      "loss": 0.7071,
-      "step": 625
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.749078909562279e-06,
-      "loss": 0.7211,
-      "step": 626
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.736582015413752e-06,
-      "loss": 0.701,
-      "step": 627
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.724080416062198e-06,
-      "loss": 0.691,
-      "step": 628
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.711574191366427e-06,
-      "loss": 0.7211,
-      "step": 629
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.699063421214794e-06,
-      "loss": 0.688,
-      "step": 630
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.686548185524694e-06,
-      "loss": 0.7312,
-      "step": 631
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.674028564242042e-06,
-      "loss": 0.6939,
-      "step": 632
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.66150463734077e-06,
-      "loss": 0.7012,
-      "step": 633
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 5.648976484822313e-06,
-      "loss": 0.6963,
-      "step": 634
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.636444186715101e-06,
-      "loss": 0.7171,
-      "step": 635
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.623907823074044e-06,
-      "loss": 0.7037,
-      "step": 636
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.611367473980022e-06,
-      "loss": 0.713,
-      "step": 637
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.598823219539373e-06,
-      "loss": 0.7142,
-      "step": 638
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.5862751398833824e-06,
-      "loss": 0.7061,
-      "step": 639
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.573723315167773e-06,
-      "loss": 0.6989,
-      "step": 640
-    },
-    {
-      "epoch": 0.49,
-      "eval_loss": 1.2723302841186523,
-      "eval_runtime": 126.4363,
-      "eval_samples_per_second": 94.664,
-      "eval_steps_per_second": 23.672,
-      "step": 640
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.561167825572187e-06,
-      "loss": 0.6954,
-      "step": 641
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.54860875129968e-06,
-      "loss": 0.7098,
-      "step": 642
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.5360461725762045e-06,
-      "loss": 0.7096,
-      "step": 643
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.5234801696500996e-06,
-      "loss": 0.7219,
-      "step": 644
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.51091082279158e-06,
-      "loss": 0.7174,
-      "step": 645
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.498338212292216e-06,
-      "loss": 0.7046,
-      "step": 646
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.48576241846443e-06,
-      "loss": 0.6943,
-      "step": 647
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.473183521640978e-06,
-      "loss": 0.724,
-      "step": 648
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.460601602174437e-06,
-      "loss": 0.7001,
-      "step": 649
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.448016740436693e-06,
-      "loss": 0.682,
-      "step": 650
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.435429016818428e-06,
-      "loss": 0.6889,
-      "step": 651
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.422838511728601e-06,
-      "loss": 0.7084,
-      "step": 652
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.410245305593942e-06,
-      "loss": 0.7108,
-      "step": 653
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.397649478858437e-06,
-      "loss": 0.7307,
-      "step": 654
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.3850511119828095e-06,
-      "loss": 0.6986,
-      "step": 655
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.3724502854440085e-06,
-      "loss": 0.7042,
-      "step": 656
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.3598470797346945e-06,
-      "loss": 0.6955,
-      "step": 657
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.347241575362729e-06,
-      "loss": 0.7235,
-      "step": 658
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.334633852850657e-06,
-      "loss": 0.7169,
-      "step": 659
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.322023992735188e-06,
-      "loss": 0.7076,
-      "step": 660
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.309412075566692e-06,
-      "loss": 0.7115,
-      "step": 661
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.296798181908673e-06,
-      "loss": 0.7154,
-      "step": 662
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.284182392337269e-06,
-      "loss": 0.7133,
-      "step": 663
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.271564787440723e-06,
-      "loss": 0.6954,
-      "step": 664
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.258945447818872e-06,
-      "loss": 0.6985,
-      "step": 665
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.24632445408264e-06,
-      "loss": 0.6894,
-      "step": 666
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.233701886853515e-06,
-      "loss": 0.6973,
-      "step": 667
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.221077826763035e-06,
-      "loss": 0.719,
-      "step": 668
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.208452354452275e-06,
-      "loss": 0.7113,
-      "step": 669
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.1958255505713305e-06,
-      "loss": 0.6917,
-      "step": 670
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.1831974957788075e-06,
-      "loss": 0.7074,
-      "step": 671
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.1705682707412965e-06,
-      "loss": 0.727,
-      "step": 672
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.157937956132866e-06,
-      "loss": 0.7054,
-      "step": 673
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.145306632634544e-06,
-      "loss": 0.7132,
-      "step": 674
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.132674380933803e-06,
-      "loss": 0.7042,
-      "step": 675
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.120041281724048e-06,
-      "loss": 0.7062,
-      "step": 676
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.107407415704094e-06,
-      "loss": 0.7076,
-      "step": 677
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.094772863577655e-06,
-      "loss": 0.6934,
-      "step": 678
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.082137706052826e-06,
-      "loss": 0.6957,
-      "step": 679
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.069502023841576e-06,
-      "loss": 0.7104,
-      "step": 680
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.056865897659219e-06,
-      "loss": 0.7048,
-      "step": 681
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.044229408223906e-06,
-      "loss": 0.7057,
-      "step": 682
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.031592636256111e-06,
-      "loss": 0.7183,
-      "step": 683
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.018955662478109e-06,
-      "loss": 0.6956,
-      "step": 684
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.0063185676134695e-06,
-      "loss": 0.7088,
-      "step": 685
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 4.993681432386532e-06,
-      "loss": 0.7059,
-      "step": 686
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.981044337521892e-06,
-      "loss": 0.7231,
-      "step": 687
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.968407363743891e-06,
-      "loss": 0.7042,
-      "step": 688
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.955770591776095e-06,
-      "loss": 0.6963,
-      "step": 689
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.943134102340782e-06,
-      "loss": 0.697,
-      "step": 690
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.9304979761584256e-06,
-      "loss": 0.6979,
-      "step": 691
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.9178622939471746e-06,
-      "loss": 0.6939,
-      "step": 692
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.905227136422347e-06,
-      "loss": 0.7049,
-      "step": 693
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.892592584295907e-06,
-      "loss": 0.7023,
-      "step": 694
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.8799587182759526e-06,
-      "loss": 0.7277,
-      "step": 695
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.867325619066198e-06,
-      "loss": 0.7335,
-      "step": 696
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.854693367365457e-06,
-      "loss": 0.7116,
-      "step": 697
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.842062043867136e-06,
-      "loss": 0.704,
-      "step": 698
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 4.829431729258705e-06,
-      "loss": 0.7154,
-      "step": 699
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.816802504221194e-06,
-      "loss": 0.6967,
-      "step": 700
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.804174449428671e-06,
-      "loss": 0.7116,
-      "step": 701
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.791547645547727e-06,
-      "loss": 0.713,
-      "step": 702
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.778922173236968e-06,
-      "loss": 0.6949,
-      "step": 703
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.7662981131464866e-06,
-      "loss": 0.6969,
-      "step": 704
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.753675545917361e-06,
-      "loss": 0.6958,
-      "step": 705
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.74105455218113e-06,
-      "loss": 0.7099,
-      "step": 706
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.728435212559278e-06,
-      "loss": 0.6937,
-      "step": 707
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.715817607662731e-06,
-      "loss": 0.7165,
-      "step": 708
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.703201818091326e-06,
-      "loss": 0.711,
-      "step": 709
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.690587924433309e-06,
-      "loss": 0.6951,
-      "step": 710
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.677976007264812e-06,
-      "loss": 0.7308,
-      "step": 711
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 4.665366147149343e-06,
-      "loss": 0.693,
-      "step": 712
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.652758424637271e-06,
-      "loss": 0.6987,
-      "step": 713
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.640152920265306e-06,
-      "loss": 0.7145,
-      "step": 714
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.627549714555993e-06,
-      "loss": 0.7104,
-      "step": 715
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.614948888017191e-06,
-      "loss": 0.7041,
-      "step": 716
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.6023505211415635e-06,
-      "loss": 0.6999,
-      "step": 717
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.589754694406058e-06,
-      "loss": 0.7177,
-      "step": 718
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.577161488271401e-06,
-      "loss": 0.707,
-      "step": 719
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.564570983181573e-06,
-      "loss": 0.6847,
-      "step": 720
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.5519832595633076e-06,
-      "loss": 0.6904,
-      "step": 721
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.539398397825563e-06,
-      "loss": 0.7115,
-      "step": 722
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.526816478359023e-06,
-      "loss": 0.7155,
-      "step": 723
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.514237581535571e-06,
-      "loss": 0.7139,
-      "step": 724
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.501661787707784e-06,
-      "loss": 0.7214,
-      "step": 725
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.489089177208421e-06,
-      "loss": 0.7107,
-      "step": 726
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.4765198303499e-06,
-      "loss": 0.7114,
-      "step": 727
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.463953827423796e-06,
-      "loss": 0.7309,
-      "step": 728
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.45139124870032e-06,
-      "loss": 0.6868,
-      "step": 729
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.438832174427815e-06,
-      "loss": 0.7026,
-      "step": 730
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.426276684832229e-06,
-      "loss": 0.695,
-      "step": 731
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.413724860116619e-06,
-      "loss": 0.7076,
-      "step": 732
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.40117678046063e-06,
-      "loss": 0.7125,
-      "step": 733
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.388632526019981e-06,
-      "loss": 0.7007,
-      "step": 734
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.3760921769259585e-06,
-      "loss": 0.6994,
-      "step": 735
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.363555813284902e-06,
-      "loss": 0.7028,
-      "step": 736
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.351023515177688e-06,
-      "loss": 0.7047,
-      "step": 737
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.338495362659233e-06,
-      "loss": 0.7143,
-      "step": 738
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.325971435757961e-06,
-      "loss": 0.7092,
-      "step": 739
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.3134518144753085e-06,
-      "loss": 0.7156,
-      "step": 740
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.300936578785207e-06,
-      "loss": 0.7018,
-      "step": 741
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.2884258086335755e-06,
-      "loss": 0.6915,
-      "step": 742
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.275919583937805e-06,
-      "loss": 0.6891,
-      "step": 743
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.263417984586251e-06,
-      "loss": 0.7028,
-      "step": 744
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.250921090437723e-06,
-      "loss": 0.71,
-      "step": 745
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.2384289813209754e-06,
-      "loss": 0.6996,
-      "step": 746
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.225941737034196e-06,
-      "loss": 0.7116,
-      "step": 747
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.213459437344496e-06,
-      "loss": 0.7016,
-      "step": 748
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.2009821619874e-06,
-      "loss": 0.7103,
-      "step": 749
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.188509990666339e-06,
-      "loss": 0.6894,
-      "step": 750
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.1760430030521405e-06,
-      "loss": 0.7042,
-      "step": 751
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.163581278782519e-06,
-      "loss": 0.7124,
-      "step": 752
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.151124897461567e-06,
-      "loss": 0.6786,
-      "step": 753
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.138673938659245e-06,
-      "loss": 0.7049,
-      "step": 754
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.126228481910877e-06,
-      "loss": 0.7045,
-      "step": 755
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.113788606716641e-06,
-      "loss": 0.677,
-      "step": 756
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.101354392541061e-06,
-      "loss": 0.6974,
-      "step": 757
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.088925918812496e-06,
-      "loss": 0.7318,
-      "step": 758
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.076503264922637e-06,
-      "loss": 0.7107,
-      "step": 759
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.064086510226002e-06,
-      "loss": 0.6988,
-      "step": 760
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.05167573403942e-06,
-      "loss": 0.6958,
-      "step": 761
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.039271015641535e-06,
-      "loss": 0.6882,
-      "step": 762
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.026872434272291e-06,
-      "loss": 0.7055,
-      "step": 763
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.014480069132428e-06,
-      "loss": 0.7141,
-      "step": 764
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.002093999382985e-06,
-      "loss": 0.7035,
-      "step": 765
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.9897143041447794e-06,
-      "loss": 0.7042,
-      "step": 766
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.977341062497911e-06,
-      "loss": 0.6956,
-      "step": 767
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.964974353481254e-06,
-      "loss": 0.6955,
-      "step": 768
-    },
-    {
-      "epoch": 0.59,
-      "eval_loss": 1.2692409753799438,
-      "eval_runtime": 126.0431,
-      "eval_samples_per_second": 94.96,
-      "eval_steps_per_second": 23.746,
-      "step": 768
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.9526142560919566e-06,
-      "loss": 0.6964,
-      "step": 769
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.940260849284928e-06,
-      "loss": 0.6884,
-      "step": 770
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.927914211972342e-06,
-      "loss": 0.6978,
-      "step": 771
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.915574423023128e-06,
-      "loss": 0.7013,
-      "step": 772
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.903241561262468e-06,
-      "loss": 0.7124,
-      "step": 773
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.890915705471297e-06,
-      "loss": 0.7024,
-      "step": 774
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.878596934385794e-06,
-      "loss": 0.7031,
-      "step": 775
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.866285326696883e-06,
-      "loss": 0.7138,
-      "step": 776
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 3.853980961049727e-06,
-      "loss": 0.7161,
-      "step": 777
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.841683916043229e-06,
-      "loss": 0.7032,
-      "step": 778
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.829394270229531e-06,
-      "loss": 0.7102,
-      "step": 779
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.817112102113506e-06,
-      "loss": 0.6979,
-      "step": 780
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.804837490152262e-06,
-      "loss": 0.683,
-      "step": 781
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.7925705127546387e-06,
-      "loss": 0.6864,
-      "step": 782
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.780311248280708e-06,
-      "loss": 0.7354,
-      "step": 783
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.7680597750412715e-06,
-      "loss": 0.6988,
-      "step": 784
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.755816171297362e-06,
-      "loss": 0.689,
-      "step": 785
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.7435805152597416e-06,
-      "loss": 0.696,
-      "step": 786
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.7313528850884028e-06,
-      "loss": 0.7027,
-      "step": 787
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.7191333588920745e-06,
-      "loss": 0.687,
-      "step": 788
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.706922014727714e-06,
-      "loss": 0.7056,
-      "step": 789
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 3.694718930600012e-06,
-      "loss": 0.7121,
-      "step": 790
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.6825241844608973e-06,
-      "loss": 0.6855,
-      "step": 791
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.6703378542090373e-06,
-      "loss": 0.7134,
-      "step": 792
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.658160017689337e-06,
-      "loss": 0.7107,
-      "step": 793
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.6459907526924467e-06,
-      "loss": 0.709,
-      "step": 794
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.6338301369542604e-06,
-      "loss": 0.7103,
-      "step": 795
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.6216782481554236e-06,
-      "loss": 0.7255,
-      "step": 796
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.609535163920836e-06,
-      "loss": 0.703,
-      "step": 797
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.5974009618191512e-06,
-      "loss": 0.6917,
-      "step": 798
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.5852757193622878e-06,
-      "loss": 0.7056,
-      "step": 799
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.5731595140049296e-06,
-      "loss": 0.6945,
-      "step": 800
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.5610524231440324e-06,
-      "loss": 0.6958,
-      "step": 801
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.5489545241183322e-06,
-      "loss": 0.698,
-      "step": 802
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.536865894207845e-06,
-      "loss": 0.6973,
-      "step": 803
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.5247866106333795e-06,
-      "loss": 0.7151,
-      "step": 804
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.512716750556037e-06,
-      "loss": 0.7113,
-      "step": 805
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.50065639107673e-06,
-      "loss": 0.7088,
-      "step": 806
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.488605609235674e-06,
-      "loss": 0.6899,
-      "step": 807
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.4765644820119093e-06,
-      "loss": 0.7063,
-      "step": 808
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.464533086322801e-06,
-      "loss": 0.6992,
-      "step": 809
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.45251149902355e-06,
-      "loss": 0.7083,
-      "step": 810
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.4404997969067043e-06,
-      "loss": 0.7025,
-      "step": 811
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.428498056701665e-06,
-      "loss": 0.7017,
-      "step": 812
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.4165063550741985e-06,
-      "loss": 0.7221,
-      "step": 813
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.4045247686259436e-06,
-      "loss": 0.6963,
-      "step": 814
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.3925533738939286e-06,
-      "loss": 0.7003,
-      "step": 815
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.380592247350075e-06,
-      "loss": 0.6958,
-      "step": 816
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.368641465400713e-06,
-      "loss": 0.6989,
-      "step": 817
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.356701104386092e-06,
-      "loss": 0.7141,
-      "step": 818
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.344771240579897e-06,
-      "loss": 0.709,
-      "step": 819
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.3328519501887535e-06,
-      "loss": 0.6989,
-      "step": 820
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.320943309351746e-06,
-      "loss": 0.6851,
-      "step": 821
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.3090453941399346e-06,
-      "loss": 0.7048,
-      "step": 822
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.2971582805558622e-06,
-      "loss": 0.7083,
-      "step": 823
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.285282044533072e-06,
-      "loss": 0.7161,
-      "step": 824
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.2734167619356237e-06,
-      "loss": 0.6925,
-      "step": 825
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.2615625085576062e-06,
-      "loss": 0.6981,
-      "step": 826
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.249719360122659e-06,
-      "loss": 0.7107,
-      "step": 827
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.23788739228348e-06,
-      "loss": 0.7204,
-      "step": 828
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.226066680621347e-06,
-      "loss": 0.6901,
-      "step": 829
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.2142573006456377e-06,
-      "loss": 0.7023,
-      "step": 830
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.20245932779334e-06,
-      "loss": 0.709,
-      "step": 831
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.190672837428581e-06,
-      "loss": 0.7008,
-      "step": 832
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.178897904842132e-06,
-      "loss": 0.705,
-      "step": 833
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.167134605250938e-06,
-      "loss": 0.7034,
-      "step": 834
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.155383013797631e-06,
-      "loss": 0.6976,
-      "step": 835
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.1436432055500583e-06,
-      "loss": 0.7154,
-      "step": 836
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.131915255500792e-06,
-      "loss": 0.7226,
-      "step": 837
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.120199238566658e-06,
-      "loss": 0.7073,
-      "step": 838
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.1084952295882533e-06,
-      "loss": 0.7182,
-      "step": 839
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.096803303329469e-06,
-      "loss": 0.7134,
-      "step": 840
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.0851235344770176e-06,
-      "loss": 0.6802,
-      "step": 841
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.073455997639946e-06,
-      "loss": 0.6957,
-      "step": 842
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.0618007673491673e-06,
-      "loss": 0.7119,
-      "step": 843
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.0501579180569795e-06,
-      "loss": 0.7178,
-      "step": 844
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.0385275241365965e-06,
-      "loss": 0.6878,
-      "step": 845
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.0269096598816636e-06,
-      "loss": 0.6955,
-      "step": 846
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.015304399505791e-06,
-      "loss": 0.697,
-      "step": 847
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.0037118171420764e-06,
-      "loss": 0.7147,
-      "step": 848
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.9921319868426304e-06,
-      "loss": 0.6789,
-      "step": 849
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.9805649825781073e-06,
-      "loss": 0.6975,
-      "step": 850
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.9690108782372282e-06,
-      "loss": 0.6798,
-      "step": 851
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.957469747626311e-06,
-      "loss": 0.6772,
-      "step": 852
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.945941664468799e-06,
-      "loss": 0.7134,
-      "step": 853
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.9344267024047906e-06,
-      "loss": 0.6944,
-      "step": 854
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.9229249349905686e-06,
-      "loss": 0.6855,
-      "step": 855
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 2.9114364356981274e-06,
-      "loss": 0.7047,
-      "step": 856
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.8999612779147086e-06,
-      "loss": 0.7053,
-      "step": 857
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.8884995349423286e-06,
-      "loss": 0.6965,
-      "step": 858
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.877051279997312e-06,
-      "loss": 0.7024,
-      "step": 859
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.8656165862098227e-06,
-      "loss": 0.693,
-      "step": 860
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.854195526623399e-06,
-      "loss": 0.6933,
-      "step": 861
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.8427881741944824e-06,
-      "loss": 0.6949,
-      "step": 862
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.8313946017919575e-06,
-      "loss": 0.7189,
-      "step": 863
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.820014882196683e-06,
-      "loss": 0.696,
-      "step": 864
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.808649088101024e-06,
-      "loss": 0.7092,
-      "step": 865
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.7972972921083997e-06,
-      "loss": 0.7002,
-      "step": 866
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.7859595667328027e-06,
-      "loss": 0.709,
-      "step": 867
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.7746359843983466e-06,
-      "loss": 0.7134,
-      "step": 868
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 2.7633266174388018e-06,
-      "loss": 0.6943,
-      "step": 869
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.7520315380971313e-06,
-      "loss": 0.6965,
-      "step": 870
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.740750818525031e-06,
-      "loss": 0.7196,
-      "step": 871
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.729484530782468e-06,
-      "loss": 0.7176,
-      "step": 872
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.7182327468372206e-06,
-      "loss": 0.6975,
-      "step": 873
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.7069955385644164e-06,
-      "loss": 0.7052,
-      "step": 874
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6957729777460817e-06,
-      "loss": 0.7142,
-      "step": 875
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6845651360706703e-06,
-      "loss": 0.7064,
-      "step": 876
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6733720851326127e-06,
-      "loss": 0.7057,
-      "step": 877
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6621938964318593e-06,
-      "loss": 0.715,
-      "step": 878
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6510306413734223e-06,
-      "loss": 0.6897,
-      "step": 879
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6398823912669175e-06,
-      "loss": 0.6922,
-      "step": 880
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6287492173261132e-06,
-      "loss": 0.7188,
-      "step": 881
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 2.6176311906684704e-06,
-      "loss": 0.706,
-      "step": 882
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.6065283823146916e-06,
-      "loss": 0.6942,
-      "step": 883
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.59544086318827e-06,
-      "loss": 0.7151,
-      "step": 884
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5843687041150294e-06,
-      "loss": 0.7047,
-      "step": 885
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5733119758226756e-06,
-      "loss": 0.6886,
-      "step": 886
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5622707489403436e-06,
-      "loss": 0.7075,
-      "step": 887
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5512450939981494e-06,
-      "loss": 0.7002,
-      "step": 888
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5402350814267364e-06,
-      "loss": 0.7007,
-      "step": 889
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.529240781556824e-06,
-      "loss": 0.7147,
-      "step": 890
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5182622646187648e-06,
-      "loss": 0.7103,
-      "step": 891
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.5072996007420868e-06,
-      "loss": 0.6987,
-      "step": 892
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.496352859955057e-06,
-      "loss": 0.7111,
-      "step": 893
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.4854221121842213e-06,
-      "loss": 0.7049,
-      "step": 894
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.4745074272539677e-06,
-      "loss": 0.683,
-      "step": 895
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.4636088748860738e-06,
-      "loss": 0.7179,
-      "step": 896
-    },
-    {
-      "epoch": 0.69,
-      "eval_loss": 1.2681297063827515,
-      "eval_runtime": 126.6345,
-      "eval_samples_per_second": 94.516,
-      "eval_steps_per_second": 23.635,
-      "step": 896
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.4527265246992667e-06,
-      "loss": 0.704,
-      "step": 897
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.4418604462087743e-06,
-      "loss": 0.6825,
-      "step": 898
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.4310107088258834e-06,
-      "loss": 0.7164,
-      "step": 899
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.4201773818574956e-06,
-      "loss": 0.6916,
-      "step": 900
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.4093605345056833e-06,
-      "loss": 0.7176,
-      "step": 901
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.398560235867252e-06,
-      "loss": 0.7117,
-      "step": 902
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.3877765549332936e-06,
-      "loss": 0.6892,
-      "step": 903
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.3770095605887487e-06,
-      "loss": 0.7112,
-      "step": 904
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.366259321611965e-06,
-      "loss": 0.7141,
-      "step": 905
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.3555259066742593e-06,
-      "loss": 0.6946,
-      "step": 906
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.3448093843394782e-06,
-      "loss": 0.6983,
-      "step": 907
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.334109823063562e-06,
-      "loss": 0.7059,
-      "step": 908
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.3234272911940998e-06,
-      "loss": 0.6979,
-      "step": 909
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.312761856969908e-06,
-      "loss": 0.694,
-      "step": 910
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.302113588520578e-06,
-      "loss": 0.6865,
-      "step": 911
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2914825538660507e-06,
-      "loss": 0.6993,
-      "step": 912
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2808688209161797e-06,
-      "loss": 0.697,
-      "step": 913
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2702724574702956e-06,
-      "loss": 0.7274,
-      "step": 914
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2596935312167766e-06,
-      "loss": 0.676,
-      "step": 915
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2491321097326125e-06,
-      "loss": 0.7063,
-      "step": 916
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2385882604829763e-06,
-      "loss": 0.6942,
-      "step": 917
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2280620508207873e-06,
-      "loss": 0.7099,
-      "step": 918
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.2175535479862927e-06,
-      "loss": 0.7418,
-      "step": 919
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.207062819106624e-06,
-      "loss": 0.7174,
-      "step": 920
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.1965899311953777e-06,
-      "loss": 0.7083,
-      "step": 921
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.1861349511521817e-06,
-      "loss": 0.684,
-      "step": 922
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.1756979457622736e-06,
-      "loss": 0.7072,
-      "step": 923
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.1652789816960677e-06,
-      "loss": 0.7316,
-      "step": 924
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.1548781255087353e-06,
-      "loss": 0.6925,
-      "step": 925
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.144495443639775e-06,
-      "loss": 0.7069,
-      "step": 926
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.134131002412588e-06,
-      "loss": 0.7018,
-      "step": 927
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.1237848680340635e-06,
-      "loss": 0.6931,
-      "step": 928
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.1134571065941396e-06,
-      "loss": 0.6855,
-      "step": 929
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.103147784065396e-06,
-      "loss": 0.7055,
-      "step": 930
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.092856966302624e-06,
-      "loss": 0.7017,
-      "step": 931
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0825847190424096e-06,
-      "loss": 0.7027,
-      "step": 932
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.072331107902713e-06,
-      "loss": 0.7046,
-      "step": 933
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.0620961983824478e-06,
-      "loss": 0.7141,
-      "step": 934
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.051880055861064e-06,
-      "loss": 0.6972,
-      "step": 935
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.041682745598129e-06,
-      "loss": 0.6954,
-      "step": 936
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.0315043327329164e-06,
-      "loss": 0.7176,
-      "step": 937
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.0213448822839813e-06,
-      "loss": 0.6966,
-      "step": 938
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.0112044591487497e-06,
-      "loss": 0.6961,
-      "step": 939
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.0010831281031025e-06,
-      "loss": 0.7034,
-      "step": 940
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9909809538009644e-06,
-      "loss": 0.6919,
-      "step": 941
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9808980007738877e-06,
-      "loss": 0.7084,
-      "step": 942
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.970834333430642e-06,
-      "loss": 0.6908,
-      "step": 943
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.960790016056801e-06,
-      "loss": 0.727,
-      "step": 944
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.950765112814333e-06,
-      "loss": 0.6916,
-      "step": 945
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.940759687741195e-06,
-      "loss": 0.713,
-      "step": 946
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.9307738047509147e-06,
-      "loss": 0.7095,
-      "step": 947
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.9208075276321887e-06,
-      "loss": 0.6981,
-      "step": 948
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.910860920048474e-06,
-      "loss": 0.7,
-      "step": 949
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.9009340455375803e-06,
-      "loss": 0.7072,
-      "step": 950
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8910269675112651e-06,
-      "loss": 0.7065,
-      "step": 951
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8811397492548283e-06,
-      "loss": 0.6972,
-      "step": 952
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8712724539267069e-06,
-      "loss": 0.7006,
-      "step": 953
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8614251445580717e-06,
-      "loss": 0.6966,
-      "step": 954
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8515978840524302e-06,
-      "loss": 0.6852,
-      "step": 955
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8417907351852148e-06,
-      "loss": 0.7103,
-      "step": 956
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.832003760603389e-06,
-      "loss": 0.7124,
-      "step": 957
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8222370228250452e-06,
-      "loss": 0.7167,
-      "step": 958
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8124905842390051e-06,
-      "loss": 0.7122,
-      "step": 959
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.8027645071044214e-06,
-      "loss": 0.7031,
-      "step": 960
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7930588535503812e-06,
-      "loss": 0.7198,
-      "step": 961
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7833736855755063e-06,
-      "loss": 0.6989,
-      "step": 962
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7737090650475586e-06,
-      "loss": 0.725,
-      "step": 963
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7640650537030496e-06,
-      "loss": 0.7158,
-      "step": 964
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7544417131468373e-06,
-      "loss": 0.7008,
-      "step": 965
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7448391048517378e-06,
-      "loss": 0.6996,
-      "step": 966
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7352572901581322e-06,
-      "loss": 0.7044,
-      "step": 967
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7256963302735752e-06,
-      "loss": 0.7096,
-      "step": 968
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.7161562862724018e-06,
-      "loss": 0.6957,
-      "step": 969
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.70663721909534e-06,
-      "loss": 0.7023,
-      "step": 970
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.6971391895491202e-06,
-      "loss": 0.6808,
-      "step": 971
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.6876622583060854e-06,
-      "loss": 0.7144,
-      "step": 972
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 1.678206485903805e-06,
-      "loss": 0.7106,
-      "step": 973
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.6687719327446917e-06,
-      "loss": 0.6884,
-      "step": 974
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.6593586590956084e-06,
-      "loss": 0.6937,
-      "step": 975
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.6499667250874873e-06,
-      "loss": 0.7006,
-      "step": 976
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.640596190714947e-06,
-      "loss": 0.7151,
-      "step": 977
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.631247115835905e-06,
-      "loss": 0.7036,
-      "step": 978
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.6219195601712011e-06,
-      "loss": 0.7016,
-      "step": 979
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.6126135833042107e-06,
-      "loss": 0.6869,
-      "step": 980
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.603329244680466e-06,
-      "loss": 0.7031,
-      "step": 981
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.594066603607277e-06,
-      "loss": 0.6865,
-      "step": 982
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.584825719253355e-06,
-      "loss": 0.7079,
-      "step": 983
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.5756066506484285e-06,
-      "loss": 0.6869,
-      "step": 984
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.5664094566828696e-06,
-      "loss": 0.6938,
-      "step": 985
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.5572341961073185e-06,
-      "loss": 0.6961,
-      "step": 986
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 1.5480809275323073e-06,
-      "loss": 0.6988,
-      "step": 987
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.5389497094278861e-06,
-      "loss": 0.7096,
-      "step": 988
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.529840600123248e-06,
-      "loss": 0.7149,
-      "step": 989
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.5207536578063587e-06,
-      "loss": 0.6976,
-      "step": 990
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.5116889405235807e-06,
-      "loss": 0.6968,
-      "step": 991
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.5026465061793122e-06,
-      "loss": 0.7032,
-      "step": 992
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4936264125356042e-06,
-      "loss": 0.6903,
-      "step": 993
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4846287172117996e-06,
-      "loss": 0.695,
-      "step": 994
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4756534776841642e-06,
-      "loss": 0.7042,
-      "step": 995
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4667007512855174e-06,
-      "loss": 0.7104,
-      "step": 996
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4577705952048704e-06,
-      "loss": 0.6902,
-      "step": 997
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4488630664870495e-06,
-      "loss": 0.7318,
-      "step": 998
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4399782220323515e-06,
-      "loss": 0.7082,
-      "step": 999
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.4311161185961614e-06,
-      "loss": 0.6879,
-      "step": 1000
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.4222768127885977e-06,
-      "loss": 0.6826,
-      "step": 1001
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.4134603610741526e-06,
-      "loss": 0.6975,
-      "step": 1002
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.4046668197713259e-06,
-      "loss": 0.7011,
-      "step": 1003
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3958962450522711e-06,
-      "loss": 0.6868,
-      "step": 1004
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3871486929424326e-06,
-      "loss": 0.7023,
-      "step": 1005
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3784242193201885e-06,
-      "loss": 0.6896,
-      "step": 1006
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.369722879916493e-06,
-      "loss": 0.724,
-      "step": 1007
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3610447303145258e-06,
-      "loss": 0.6863,
-      "step": 1008
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.352389825949329e-06,
-      "loss": 0.6964,
-      "step": 1009
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3437582221074574e-06,
-      "loss": 0.7075,
-      "step": 1010
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3351499739266255e-06,
-      "loss": 0.7074,
-      "step": 1011
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3265651363953536e-06,
-      "loss": 0.6977,
-      "step": 1012
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.3180037643526184e-06,
-      "loss": 0.7016,
-      "step": 1013
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.3094659124874998e-06,
-      "loss": 0.718,
-      "step": 1014
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.3009516353388357e-06,
-      "loss": 0.707,
-      "step": 1015
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.292460987294869e-06,
-      "loss": 0.7223,
-      "step": 1016
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.2839940225929066e-06,
-      "loss": 0.7001,
-      "step": 1017
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.275550795318965e-06,
-      "loss": 0.7024,
-      "step": 1018
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.2671313594074293e-06,
-      "loss": 0.6991,
-      "step": 1019
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.2587357686407093e-06,
-      "loss": 0.6992,
-      "step": 1020
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.250364076648894e-06,
-      "loss": 0.7028,
-      "step": 1021
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.242016336909409e-06,
-      "loss": 0.7003,
-      "step": 1022
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.2336926027466777e-06,
-      "loss": 0.7146,
-      "step": 1023
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.2253929273317767e-06,
-      "loss": 0.6865,
-      "step": 1024
-    },
-    {
-      "epoch": 0.78,
-      "eval_loss": 1.2678072452545166,
-      "eval_runtime": 126.3097,
-      "eval_samples_per_second": 94.759,
-      "eval_steps_per_second": 23.696,
-      "step": 1024
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.2171173636820965e-06,
-      "loss": 0.6948,
-      "step": 1025
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.20886596466101e-06,
-      "loss": 0.6955,
-      "step": 1026
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.200638782977524e-06,
-      "loss": 0.7001,
-      "step": 1027
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1924358711859491e-06,
-      "loss": 0.6819,
-      "step": 1028
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1842572816855635e-06,
-      "loss": 0.693,
-      "step": 1029
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1761030667202755e-06,
-      "loss": 0.6869,
-      "step": 1030
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1679732783782933e-06,
-      "loss": 0.7293,
-      "step": 1031
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1598679685917901e-06,
-      "loss": 0.694,
-      "step": 1032
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1517871891365728e-06,
-      "loss": 0.7244,
-      "step": 1033
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1437309916317507e-06,
-      "loss": 0.6934,
-      "step": 1034
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1356994275394057e-06,
-      "loss": 0.7087,
-      "step": 1035
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1276925481642682e-06,
-      "loss": 0.7021,
-      "step": 1036
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1197104046533814e-06,
-      "loss": 0.7045,
-      "step": 1037
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.11175304799578e-06,
-      "loss": 0.6913,
-      "step": 1038
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.1038205290221621e-06,
-      "loss": 0.7002,
-      "step": 1039
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0959128984045674e-06,
-      "loss": 0.6875,
-      "step": 1040
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0880302066560493e-06,
-      "loss": 0.6966,
-      "step": 1041
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0801725041303567e-06,
-      "loss": 0.6964,
-      "step": 1042
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0723398410216085e-06,
-      "loss": 0.7128,
-      "step": 1043
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0645322673639734e-06,
-      "loss": 0.6982,
-      "step": 1044
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.056749833031357e-06,
-      "loss": 0.7027,
-      "step": 1045
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0489925877370721e-06,
-      "loss": 0.6903,
-      "step": 1046
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.041260581033529e-06,
-      "loss": 0.711,
-      "step": 1047
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0335538623119168e-06,
-      "loss": 0.6961,
-      "step": 1048
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0258724808018866e-06,
-      "loss": 0.6975,
-      "step": 1049
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0182164855712395e-06,
-      "loss": 0.6867,
-      "step": 1050
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.0105859255256112e-06,
-      "loss": 0.6974,
-      "step": 1051
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.002980849408161e-06,
-      "loss": 0.7059,
-      "step": 1052
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.954013057992584e-07,
-      "loss": 0.696,
-      "step": 1053
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.878473431161767e-07,
-      "loss": 0.6894,
-      "step": 1054
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.80319009612779e-07,
-      "loss": 0.7057,
-      "step": 1055
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.728163533792124e-07,
-      "loss": 0.7003,
-      "step": 1056
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.653394223416e-07,
-      "loss": 0.6881,
-      "step": 1057
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.578882642617365e-07,
-      "loss": 0.7012,
-      "step": 1058
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.504629267367809e-07,
-      "loss": 0.7019,
-      "step": 1059
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.430634571989544e-07,
-      "loss": 0.7019,
-      "step": 1060
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.35689902915235e-07,
-      "loss": 0.7099,
-      "step": 1061
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.283423109870565e-07,
-      "loss": 0.7046,
-      "step": 1062
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.210207283500122e-07,
-      "loss": 0.7215,
-      "step": 1063
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.137252017735471e-07,
-      "loss": 0.7088,
-      "step": 1064
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.064557778606631e-07,
-      "loss": 0.7,
-      "step": 1065
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.992125030476229e-07,
-      "loss": 0.7037,
-      "step": 1066
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.919954236036499e-07,
-      "loss": 0.7102,
-      "step": 1067
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.84804585630637e-07,
-      "loss": 0.6973,
-      "step": 1068
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.776400350628467e-07,
-      "loss": 0.6911,
-      "step": 1069
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.705018176666224e-07,
-      "loss": 0.6951,
-      "step": 1070
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.633899790400923e-07,
-      "loss": 0.6801,
-      "step": 1071
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.563045646128832e-07,
-      "loss": 0.7069,
-      "step": 1072
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.492456196458249e-07,
-      "loss": 0.7039,
-      "step": 1073
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.422131892306623e-07,
-      "loss": 0.7186,
-      "step": 1074
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.352073182897702e-07,
-      "loss": 0.7145,
-      "step": 1075
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.282280515758639e-07,
-      "loss": 0.7132,
-      "step": 1076
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.212754336717138e-07,
-      "loss": 0.6908,
-      "step": 1077
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 8.143495089898607e-07,
-      "loss": 0.7016,
-      "step": 1078
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 8.074503217723312e-07,
-      "loss": 0.6983,
-      "step": 1079
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 8.005779160903571e-07,
-      "loss": 0.7035,
-      "step": 1080
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.937323358440935e-07,
-      "loss": 0.6988,
-      "step": 1081
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.869136247623365e-07,
-      "loss": 0.7013,
-      "step": 1082
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.801218264022453e-07,
-      "loss": 0.7088,
-      "step": 1083
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.733569841490629e-07,
-      "loss": 0.6972,
-      "step": 1084
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.666191412158403e-07,
-      "loss": 0.6858,
-      "step": 1085
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.599083406431617e-07,
-      "loss": 0.7078,
-      "step": 1086
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.532246252988617e-07,
-      "loss": 0.6828,
-      "step": 1087
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.465680378777662e-07,
-      "loss": 0.7005,
-      "step": 1088
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.399386209014054e-07,
-      "loss": 0.6927,
-      "step": 1089
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.333364167177481e-07,
-      "loss": 0.6869,
-      "step": 1090
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 7.267614675009322e-07,
-      "loss": 0.6914,
-      "step": 1091
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 7.202138152509929e-07,
-      "loss": 0.6992,
-      "step": 1092
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 7.13693501793597e-07,
-      "loss": 0.6901,
-      "step": 1093
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 7.072005687797723e-07,
-      "loss": 0.6728,
-      "step": 1094
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 7.007350576856442e-07,
-      "loss": 0.7052,
-      "step": 1095
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.942970098121704e-07,
-      "loss": 0.6781,
-      "step": 1096
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.878864662848755e-07,
-      "loss": 0.7024,
-      "step": 1097
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.815034680535915e-07,
-      "loss": 0.7083,
-      "step": 1098
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.751480558921919e-07,
-      "loss": 0.6986,
-      "step": 1099
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.688202703983348e-07,
-      "loss": 0.7016,
-      "step": 1100
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.625201519932018e-07,
-      "loss": 0.7088,
-      "step": 1101
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.562477409212404e-07,
-      "loss": 0.6875,
-      "step": 1102
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.500030772499061e-07,
-      "loss": 0.6862,
-      "step": 1103
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.437862008694085e-07,
-      "loss": 0.6797,
-      "step": 1104
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.375971514924539e-07,
-      "loss": 0.7171,
-      "step": 1105
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.314359686539917e-07,
-      "loss": 0.7086,
-      "step": 1106
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.253026917109678e-07,
-      "loss": 0.7074,
-      "step": 1107
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.191973598420631e-07,
-      "loss": 0.7049,
-      "step": 1108
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.131200120474512e-07,
-      "loss": 0.6803,
-      "step": 1109
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.07070687148546e-07,
-      "loss": 0.6985,
-      "step": 1110
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.010494237877534e-07,
-      "loss": 0.7024,
-      "step": 1111
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.950562604282267e-07,
-      "loss": 0.6908,
-      "step": 1112
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.890912353536188e-07,
-      "loss": 0.7023,
-      "step": 1113
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.831543866678391e-07,
-      "loss": 0.6944,
-      "step": 1114
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.772457522948077e-07,
-      "loss": 0.7271,
-      "step": 1115
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.713653699782179e-07,
-      "loss": 0.7157,
-      "step": 1116
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.655132772812899e-07,
-      "loss": 0.7093,
-      "step": 1117
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.59689511586532e-07,
-      "loss": 0.6889,
-      "step": 1118
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.538941100955053e-07,
-      "loss": 0.7112,
-      "step": 1119
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.481271098285818e-07,
-      "loss": 0.6972,
-      "step": 1120
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.423885476247104e-07,
-      "loss": 0.7056,
-      "step": 1121
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.366784601411807e-07,
-      "loss": 0.6812,
-      "step": 1122
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.309968838533891e-07,
-      "loss": 0.7162,
-      "step": 1123
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.253438550546042e-07,
-      "loss": 0.6811,
-      "step": 1124
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.197194098557401e-07,
-      "loss": 0.6966,
-      "step": 1125
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.141235841851189e-07,
-      "loss": 0.6986,
-      "step": 1126
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.085564137882448e-07,
-      "loss": 0.7063,
-      "step": 1127
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.030179342275765e-07,
-      "loss": 0.6933,
-      "step": 1128
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 4.975081808822979e-07,
-      "loss": 0.7102,
-      "step": 1129
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 4.920271889480933e-07,
-      "loss": 0.6942,
-      "step": 1130
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.865749934369224e-07,
-      "loss": 0.7335,
-      "step": 1131
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.811516291767964e-07,
-      "loss": 0.703,
-      "step": 1132
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.757571308115555e-07,
-      "loss": 0.7029,
-      "step": 1133
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.703915328006492e-07,
-      "loss": 0.7031,
-      "step": 1134
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.650548694189139e-07,
-      "loss": 0.7114,
-      "step": 1135
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.597471747563542e-07,
-      "loss": 0.7106,
-      "step": 1136
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.544684827179269e-07,
-      "loss": 0.7151,
-      "step": 1137
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.4921882702332165e-07,
-      "loss": 0.7114,
-      "step": 1138
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.4399824120674897e-07,
-      "loss": 0.6883,
-      "step": 1139
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.388067586167233e-07,
-      "loss": 0.7035,
-      "step": 1140
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.3364441241585175e-07,
-      "loss": 0.7296,
-      "step": 1141
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.2851123558061927e-07,
-      "loss": 0.7015,
-      "step": 1142
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.234072609011841e-07,
-      "loss": 0.6942,
-      "step": 1143
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 4.183325209811617e-07,
-      "loss": 0.6933,
-      "step": 1144
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 4.132870482374185e-07,
-      "loss": 0.6998,
-      "step": 1145
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 4.082708748998676e-07,
-      "loss": 0.7104,
-      "step": 1146
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 4.0328403301125973e-07,
-      "loss": 0.6964,
-      "step": 1147
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 3.9832655442697876e-07,
-      "loss": 0.7007,
-      "step": 1148
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 3.933984708148414e-07,
-      "loss": 0.7143,
-      "step": 1149
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 3.884998136548901e-07,
-      "loss": 0.6891,
-      "step": 1150
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 3.836306142391949e-07,
-      "loss": 0.712,
-      "step": 1151
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 3.7879090367165585e-07,
-      "loss": 0.711,
-      "step": 1152
-    },
-    {
-      "epoch": 0.88,
-      "eval_loss": 1.267896056175232,
-      "eval_runtime": 126.2323,
-      "eval_samples_per_second": 94.817,
-      "eval_steps_per_second": 23.71,
-      "step": 1152
-    }
-  ],
-  "max_steps": 1307,
-  "num_train_epochs": 1,
-  "total_flos": 4.611495605347287e+18,
-  "trial_name": null,
-  "trial_params": null
-}