diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6937 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 1125,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 1.0014,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995555555555556e-05,
+      "loss": 1.0007,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991111111111111e-05,
+      "loss": 0.9951,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986666666666667e-05,
+      "loss": 0.9766,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982222222222222e-05,
+      "loss": 0.9193,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977777777777778e-05,
+      "loss": 0.7835,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973333333333334e-05,
+      "loss": 0.6979,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968888888888889e-05,
+      "loss": 0.6342,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964444444444445e-05,
+      "loss": 0.5992,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96e-05,
+      "loss": 0.5719,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955555555555556e-05,
+      "loss": 0.5196,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951111111111112e-05,
+      "loss": 0.5081,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9466666666666665e-05,
+      "loss": 0.4953,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942222222222223e-05,
+      "loss": 0.4745,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9377777777777776e-05,
+      "loss": 0.4553,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933333333333334e-05,
+      "loss": 0.4292,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.928888888888889e-05,
+      "loss": 0.4179,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924444444444445e-05,
+      "loss": 0.4116,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92e-05,
+      "loss": 0.3949,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915555555555556e-05,
+      "loss": 0.3748,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9111111111111114e-05,
+      "loss": 0.39,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906666666666667e-05,
+      "loss": 0.3803,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9022222222222224e-05,
+      "loss": 0.3797,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897777777777778e-05,
+      "loss": 0.3697,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8933333333333335e-05,
+      "loss": 0.3669,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.3746,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8844444444444445e-05,
+      "loss": 0.3632,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88e-05,
+      "loss": 0.3648,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875555555555556e-05,
+      "loss": 0.3816,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871111111111111e-05,
+      "loss": 0.3511,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866666666666667e-05,
+      "loss": 0.3521,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.862222222222222e-05,
+      "loss": 0.3532,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8577777777777776e-05,
+      "loss": 0.3636,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853333333333334e-05,
+      "loss": 0.357,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848888888888889e-05,
+      "loss": 0.3544,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844444444444445e-05,
+      "loss": 0.3388,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 0.338,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835555555555556e-05,
+      "loss": 0.3496,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8311111111111115e-05,
+      "loss": 0.34,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826666666666667e-05,
+      "loss": 0.3584,
+      "step": 40
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8222222222222225e-05,
+      "loss": 0.3365,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817777777777778e-05,
+      "loss": 0.3421,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8133333333333336e-05,
+      "loss": 0.3492,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808888888888889e-05,
+      "loss": 0.3202,
+      "step": 44
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8044444444444446e-05,
+      "loss": 0.33,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8e-05,
+      "loss": 0.3309,
+      "step": 46
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7955555555555556e-05,
+      "loss": 0.34,
+      "step": 47
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791111111111111e-05,
+      "loss": 0.3343,
+      "step": 48
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7866666666666674e-05,
+      "loss": 0.3385,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782222222222222e-05,
+      "loss": 0.3351,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 0.3305,
+      "step": 51
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773333333333333e-05,
+      "loss": 0.3221,
+      "step": 52
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768888888888889e-05,
+      "loss": 0.3123,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764444444444445e-05,
+      "loss": 0.3175,
+      "step": 54
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.76e-05,
+      "loss": 0.3109,
+      "step": 55
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755555555555556e-05,
+      "loss": 0.3133,
+      "step": 56
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751111111111111e-05,
+      "loss": 0.3147,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746666666666667e-05,
+      "loss": 0.3168,
+      "step": 58
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7422222222222226e-05,
+      "loss": 0.317,
+      "step": 59
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737777777777778e-05,
+      "loss": 0.3149,
+      "step": 60
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.309814453125,
+      "eval_margin_accuracy": 0.8607598993638126,
+      "eval_runtime": 3842.3967,
+      "eval_samples_per_second": 194.888,
+      "eval_steps_per_second": 48.722,
+      "step": 60
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7333333333333336e-05,
+      "loss": 0.3053,
+      "step": 61
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.728888888888889e-05,
+      "loss": 0.3055,
+      "step": 62
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724444444444445e-05,
+      "loss": 0.3143,
+      "step": 63
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72e-05,
+      "loss": 0.3051,
+      "step": 64
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715555555555556e-05,
+      "loss": 0.3038,
+      "step": 65
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711111111111111e-05,
+      "loss": 0.3049,
+      "step": 66
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706666666666667e-05,
+      "loss": 0.3213,
+      "step": 67
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702222222222222e-05,
+      "loss": 0.3071,
+      "step": 68
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6977777777777785e-05,
+      "loss": 0.3021,
+      "step": 69
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6933333333333333e-05,
+      "loss": 0.2977,
+      "step": 70
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6888888888888895e-05,
+      "loss": 0.3229,
+      "step": 71
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6844444444444444e-05,
+      "loss": 0.3112,
+      "step": 72
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 0.3153,
+      "step": 73
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675555555555556e-05,
+      "loss": 0.3021,
+      "step": 74
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671111111111111e-05,
+      "loss": 0.304,
+      "step": 75
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.3057,
+      "step": 76
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662222222222222e-05,
+      "loss": 0.3139,
+      "step": 77
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657777777777778e-05,
+      "loss": 0.3009,
+      "step": 78
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653333333333334e-05,
+      "loss": 0.2947,
+      "step": 79
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648888888888889e-05,
+      "loss": 0.3039,
+      "step": 80
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644444444444445e-05,
+      "loss": 0.3034,
+      "step": 81
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64e-05,
+      "loss": 0.2906,
+      "step": 82
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635555555555556e-05,
+      "loss": 0.301,
+      "step": 83
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6311111111111113e-05,
+      "loss": 0.2929,
+      "step": 84
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626666666666667e-05,
+      "loss": 0.2991,
+      "step": 85
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6222222222222224e-05,
+      "loss": 0.3042,
+      "step": 86
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617777777777778e-05,
+      "loss": 0.2844,
+      "step": 87
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6133333333333334e-05,
+      "loss": 0.2935,
+      "step": 88
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608888888888889e-05,
+      "loss": 0.2949,
+      "step": 89
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6044444444444445e-05,
+      "loss": 0.2983,
+      "step": 90
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.2945,
+      "step": 91
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5955555555555555e-05,
+      "loss": 0.2931,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591111111111112e-05,
+      "loss": 0.2872,
+      "step": 93
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5866666666666666e-05,
+      "loss": 0.287,
+      "step": 94
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582222222222222e-05,
+      "loss": 0.2992,
+      "step": 95
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577777777777778e-05,
+      "loss": 0.2914,
+      "step": 96
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573333333333333e-05,
+      "loss": 0.3027,
+      "step": 97
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5688888888888893e-05,
+      "loss": 0.2986,
+      "step": 98
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564444444444444e-05,
+      "loss": 0.2964,
+      "step": 99
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.2899,
+      "step": 100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 0.2846,
+      "step": 101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5511111111111114e-05,
+      "loss": 0.2804,
+      "step": 102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546666666666667e-05,
+      "loss": 0.2874,
+      "step": 103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5422222222222225e-05,
+      "loss": 0.2922,
+      "step": 104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537777777777778e-05,
+      "loss": 0.302,
+      "step": 105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5333333333333335e-05,
+      "loss": 0.2864,
+      "step": 106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.528888888888889e-05,
+      "loss": 0.2923,
+      "step": 107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5244444444444446e-05,
+      "loss": 0.287,
+      "step": 108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52e-05,
+      "loss": 0.2852,
+      "step": 109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5155555555555556e-05,
+      "loss": 0.3036,
+      "step": 110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.511111111111112e-05,
+      "loss": 0.2751,
+      "step": 111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5066666666666667e-05,
+      "loss": 0.2961,
+      "step": 112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502222222222223e-05,
+      "loss": 0.2876,
+      "step": 113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497777777777778e-05,
+      "loss": 0.2939,
+      "step": 114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493333333333333e-05,
+      "loss": 0.2909,
+      "step": 115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4888888888888894e-05,
+      "loss": 0.2919,
+      "step": 116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484444444444444e-05,
+      "loss": 0.2774,
+      "step": 117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 0.2912,
+      "step": 118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475555555555555e-05,
+      "loss": 0.2785,
+      "step": 119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4711111111111115e-05,
+      "loss": 0.2754,
+      "step": 120
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.2724609375,
+      "eval_margin_accuracy": 0.8733073730429627,
+      "eval_runtime": 3811.133,
+      "eval_samples_per_second": 196.486,
+      "eval_steps_per_second": 49.122,
+      "step": 120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466666666666667e-05,
+      "loss": 0.2796,
+      "step": 121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4622222222222226e-05,
+      "loss": 0.289,
+      "step": 122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457777777777778e-05,
+      "loss": 0.281,
+      "step": 123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4533333333333336e-05,
+      "loss": 0.2881,
+      "step": 124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448888888888889e-05,
+      "loss": 0.2824,
+      "step": 125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.2848,
+      "step": 126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44e-05,
+      "loss": 0.2842,
+      "step": 127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435555555555556e-05,
+      "loss": 0.2924,
+      "step": 128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431111111111111e-05,
+      "loss": 0.2799,
+      "step": 129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426666666666667e-05,
+      "loss": 0.2782,
+      "step": 130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422222222222222e-05,
+      "loss": 0.2847,
+      "step": 131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417777777777778e-05,
+      "loss": 0.2851,
+      "step": 132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413333333333334e-05,
+      "loss": 0.2827,
+      "step": 133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408888888888889e-05,
+      "loss": 0.2713,
+      "step": 134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404444444444445e-05,
+      "loss": 0.2764,
+      "step": 135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.2828,
+      "step": 136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3955555555555554e-05,
+      "loss": 0.2773,
+      "step": 137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3911111111111116e-05,
+      "loss": 0.2758,
+      "step": 138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3866666666666665e-05,
+      "loss": 0.2803,
+      "step": 139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3822222222222227e-05,
+      "loss": 0.2703,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377777777777778e-05,
+      "loss": 0.2827,
+      "step": 141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373333333333334e-05,
+      "loss": 0.2759,
+      "step": 142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368888888888889e-05,
+      "loss": 0.2765,
+      "step": 143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364444444444445e-05,
+      "loss": 0.2825,
+      "step": 144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36e-05,
+      "loss": 0.2776,
+      "step": 145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355555555555556e-05,
+      "loss": 0.278,
+      "step": 146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351111111111111e-05,
+      "loss": 0.2734,
+      "step": 147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346666666666667e-05,
+      "loss": 0.277,
+      "step": 148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3422222222222224e-05,
+      "loss": 0.2753,
+      "step": 149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337777777777778e-05,
+      "loss": 0.2728,
+      "step": 150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.2772,
+      "step": 151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.328888888888889e-05,
+      "loss": 0.2768,
+      "step": 152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324444444444445e-05,
+      "loss": 0.266,
+      "step": 153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32e-05,
+      "loss": 0.2659,
+      "step": 154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315555555555556e-05,
+      "loss": 0.2687,
+      "step": 155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.311111111111111e-05,
+      "loss": 0.2644,
+      "step": 156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3066666666666665e-05,
+      "loss": 0.268,
+      "step": 157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302222222222223e-05,
+      "loss": 0.2756,
+      "step": 158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2977777777777776e-05,
+      "loss": 0.2746,
+      "step": 159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293333333333334e-05,
+      "loss": 0.276,
+      "step": 160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2888888888888886e-05,
+      "loss": 0.2751,
+      "step": 161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284444444444445e-05,
+      "loss": 0.2705,
+      "step": 162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 0.2673,
+      "step": 163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275555555555556e-05,
+      "loss": 0.2586,
+      "step": 164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2711111111111114e-05,
+      "loss": 0.2787,
+      "step": 165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266666666666667e-05,
+      "loss": 0.258,
+      "step": 166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2622222222222224e-05,
+      "loss": 0.2541,
+      "step": 167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257777777777778e-05,
+      "loss": 0.2709,
+      "step": 168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2533333333333335e-05,
+      "loss": 0.2718,
+      "step": 169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248888888888889e-05,
+      "loss": 0.2574,
+      "step": 170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2444444444444445e-05,
+      "loss": 0.2715,
+      "step": 171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24e-05,
+      "loss": 0.2634,
+      "step": 172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235555555555556e-05,
+      "loss": 0.2746,
+      "step": 173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231111111111111e-05,
+      "loss": 0.2669,
+      "step": 174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226666666666667e-05,
+      "loss": 0.2602,
+      "step": 175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.2631,
+      "step": 176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217777777777778e-05,
+      "loss": 0.268,
+      "step": 177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213333333333334e-05,
+      "loss": 0.25,
+      "step": 178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208888888888889e-05,
+      "loss": 0.2849,
+      "step": 179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204444444444445e-05,
+      "loss": 0.2619,
+      "step": 180
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.251220703125,
+      "eval_margin_accuracy": 0.8872329855936413,
+      "eval_runtime": 3810.3489,
+      "eval_samples_per_second": 196.527,
+      "eval_steps_per_second": 49.132,
+      "step": 180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2e-05,
+      "loss": 0.2529,
+      "step": 181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.195555555555556e-05,
+      "loss": 0.2701,
+      "step": 182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1911111111111115e-05,
+      "loss": 0.269,
+      "step": 183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186666666666667e-05,
+      "loss": 0.2571,
+      "step": 184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1822222222222225e-05,
+      "loss": 0.2527,
+      "step": 185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177777777777778e-05,
+      "loss": 0.2581,
+      "step": 186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1733333333333336e-05,
+      "loss": 0.2594,
+      "step": 187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168888888888889e-05,
+      "loss": 0.2615,
+      "step": 188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1644444444444446e-05,
+      "loss": 0.2545,
+      "step": 189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.16e-05,
+      "loss": 0.2594,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155555555555556e-05,
+      "loss": 0.2544,
+      "step": 191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151111111111111e-05,
+      "loss": 0.2553,
+      "step": 192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146666666666667e-05,
+      "loss": 0.2527,
+      "step": 193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142222222222222e-05,
+      "loss": 0.2599,
+      "step": 194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1377777777777784e-05,
+      "loss": 0.2587,
+      "step": 195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133333333333333e-05,
+      "loss": 0.2441,
+      "step": 196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1288888888888895e-05,
+      "loss": 0.2589,
+      "step": 197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124444444444444e-05,
+      "loss": 0.2484,
+      "step": 198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12e-05,
+      "loss": 0.2704,
+      "step": 199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115555555555556e-05,
+      "loss": 0.2521,
+      "step": 200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 0.2618,
+      "step": 201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106666666666667e-05,
+      "loss": 0.2419,
+      "step": 202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1022222222222226e-05,
+      "loss": 0.2539,
+      "step": 203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097777777777778e-05,
+      "loss": 0.2658,
+      "step": 204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093333333333334e-05,
+      "loss": 0.2704,
+      "step": 205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088888888888889e-05,
+      "loss": 0.2539,
+      "step": 206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084444444444445e-05,
+      "loss": 0.2513,
+      "step": 207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08e-05,
+      "loss": 0.2649,
+      "step": 208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075555555555556e-05,
+      "loss": 0.2505,
+      "step": 209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071111111111111e-05,
+      "loss": 0.2551,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066666666666667e-05,
+      "loss": 0.2482,
+      "step": 211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062222222222222e-05,
+      "loss": 0.2481,
+      "step": 212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057777777777778e-05,
+      "loss": 0.2655,
+      "step": 213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0533333333333334e-05,
+      "loss": 0.2376,
+      "step": 214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0488888888888896e-05,
+      "loss": 0.2412,
+      "step": 215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0444444444444444e-05,
+      "loss": 0.2546,
+      "step": 216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 0.2533,
+      "step": 217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0355555555555555e-05,
+      "loss": 0.2577,
+      "step": 218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031111111111111e-05,
+      "loss": 0.2394,
+      "step": 219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026666666666667e-05,
+      "loss": 0.2394,
+      "step": 220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022222222222222e-05,
+      "loss": 0.2487,
+      "step": 221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017777777777778e-05,
+      "loss": 0.2565,
+      "step": 222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013333333333333e-05,
+      "loss": 0.2629,
+      "step": 223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008888888888889e-05,
+      "loss": 0.2441,
+      "step": 224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004444444444445e-05,
+      "loss": 0.2332,
+      "step": 225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4e-05,
+      "loss": 0.2485,
+      "step": 226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.995555555555556e-05,
+      "loss": 0.2471,
+      "step": 227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9911111111111114e-05,
+      "loss": 0.2441,
+      "step": 228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986666666666667e-05,
+      "loss": 0.2402,
+      "step": 229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9822222222222224e-05,
+      "loss": 0.242,
+      "step": 230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977777777777778e-05,
+      "loss": 0.2524,
+      "step": 231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9733333333333335e-05,
+      "loss": 0.2375,
+      "step": 232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968888888888889e-05,
+      "loss": 0.2437,
+      "step": 233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9644444444444445e-05,
+      "loss": 0.2569,
+      "step": 234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 0.2364,
+      "step": 235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9555555555555556e-05,
+      "loss": 0.2516,
+      "step": 236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951111111111112e-05,
+      "loss": 0.2441,
+      "step": 237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9466666666666666e-05,
+      "loss": 0.2524,
+      "step": 238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942222222222222e-05,
+      "loss": 0.2401,
+      "step": 239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937777777777778e-05,
+      "loss": 0.2378,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.2391357421875,
+      "eval_margin_accuracy": 0.8924624350325038,
+      "eval_runtime": 3808.9311,
+      "eval_samples_per_second": 196.6,
+      "eval_steps_per_second": 49.15,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933333333333333e-05,
+      "loss": 0.2567,
+      "step": 241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9288888888888894e-05,
+      "loss": 0.2362,
+      "step": 242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924444444444444e-05,
+      "loss": 0.2533,
+      "step": 243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 0.2402,
+      "step": 244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915555555555556e-05,
+      "loss": 0.2473,
+      "step": 245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9111111111111115e-05,
+      "loss": 0.2371,
+      "step": 246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906666666666667e-05,
+      "loss": 0.2406,
+      "step": 247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9022222222222225e-05,
+      "loss": 0.2428,
+      "step": 248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897777777777778e-05,
+      "loss": 0.2416,
+      "step": 249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8933333333333336e-05,
+      "loss": 0.2389,
+      "step": 250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.2497,
+      "step": 251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8844444444444446e-05,
+      "loss": 0.2462,
+      "step": 252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88e-05,
+      "loss": 0.2543,
+      "step": 253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8755555555555556e-05,
+      "loss": 0.2327,
+      "step": 254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871111111111111e-05,
+      "loss": 0.2336,
+      "step": 255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866666666666667e-05,
+      "loss": 0.2363,
+      "step": 256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.862222222222223e-05,
+      "loss": 0.2416,
+      "step": 257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857777777777778e-05,
+      "loss": 0.2454,
+      "step": 258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853333333333334e-05,
+      "loss": 0.2393,
+      "step": 259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848888888888889e-05,
+      "loss": 0.2451,
+      "step": 260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844444444444444e-05,
+      "loss": 0.2413,
+      "step": 261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.2412,
+      "step": 262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8355555555555553e-05,
+      "loss": 0.2378,
+      "step": 263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8311111111111115e-05,
+      "loss": 0.2509,
+      "step": 264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8266666666666664e-05,
+      "loss": 0.2453,
+      "step": 265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8222222222222226e-05,
+      "loss": 0.2354,
+      "step": 266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817777777777778e-05,
+      "loss": 0.249,
+      "step": 267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8133333333333336e-05,
+      "loss": 0.2522,
+      "step": 268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808888888888889e-05,
+      "loss": 0.2493,
+      "step": 269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804444444444445e-05,
+      "loss": 0.2531,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2567,
+      "step": 271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795555555555556e-05,
+      "loss": 0.2364,
+      "step": 272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791111111111111e-05,
+      "loss": 0.2195,
+      "step": 273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786666666666667e-05,
+      "loss": 0.2476,
+      "step": 274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782222222222222e-05,
+      "loss": 0.24,
+      "step": 275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.2414,
+      "step": 276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773333333333334e-05,
+      "loss": 0.2463,
+      "step": 277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768888888888889e-05,
+      "loss": 0.2409,
+      "step": 278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764444444444445e-05,
+      "loss": 0.2351,
+      "step": 279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.76e-05,
+      "loss": 0.2498,
+      "step": 280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7555555555555554e-05,
+      "loss": 0.2379,
+      "step": 281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7511111111111116e-05,
+      "loss": 0.2396,
+      "step": 282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7466666666666665e-05,
+      "loss": 0.2404,
+      "step": 283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742222222222223e-05,
+      "loss": 0.2322,
+      "step": 284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7377777777777775e-05,
+      "loss": 0.2333,
+      "step": 285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733333333333334e-05,
+      "loss": 0.2349,
+      "step": 286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.728888888888889e-05,
+      "loss": 0.235,
+      "step": 287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724444444444445e-05,
+      "loss": 0.2409,
+      "step": 288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72e-05,
+      "loss": 0.2653,
+      "step": 289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715555555555555e-05,
+      "loss": 0.2361,
+      "step": 290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7111111111111113e-05,
+      "loss": 0.2316,
+      "step": 291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706666666666667e-05,
+      "loss": 0.2391,
+      "step": 292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7022222222222224e-05,
+      "loss": 0.2414,
+      "step": 293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697777777777778e-05,
+      "loss": 0.2405,
+      "step": 294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6933333333333334e-05,
+      "loss": 0.2328,
+      "step": 295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688888888888889e-05,
+      "loss": 0.2368,
+      "step": 296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6844444444444445e-05,
+      "loss": 0.2394,
+      "step": 297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68e-05,
+      "loss": 0.2341,
+      "step": 298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675555555555556e-05,
+      "loss": 0.2385,
+      "step": 299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671111111111111e-05,
+      "loss": 0.2451,
+      "step": 300
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.23046875,
+      "eval_margin_accuracy": 0.8942919410925757,
+      "eval_runtime": 3815.2446,
+      "eval_samples_per_second": 196.275,
+      "eval_steps_per_second": 49.069,
+      "step": 300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.241,
+      "step": 301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.662222222222223e-05,
+      "loss": 0.2331,
+      "step": 302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6577777777777776e-05,
+      "loss": 0.2391,
+      "step": 303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653333333333334e-05,
+      "loss": 0.2446,
+      "step": 304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648888888888889e-05,
+      "loss": 0.232,
+      "step": 305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644444444444445e-05,
+      "loss": 0.2473,
+      "step": 306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 0.2473,
+      "step": 307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635555555555556e-05,
+      "loss": 0.2344,
+      "step": 308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6311111111111114e-05,
+      "loss": 0.2351,
+      "step": 309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626666666666667e-05,
+      "loss": 0.2415,
+      "step": 310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6222222222222225e-05,
+      "loss": 0.2387,
+      "step": 311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617777777777778e-05,
+      "loss": 0.2459,
+      "step": 312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6133333333333335e-05,
+      "loss": 0.2318,
+      "step": 313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608888888888889e-05,
+      "loss": 0.2405,
+      "step": 314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6044444444444446e-05,
+      "loss": 0.2286,
+      "step": 315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6e-05,
+      "loss": 0.2297,
+      "step": 316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5955555555555556e-05,
+      "loss": 0.2368,
+      "step": 317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591111111111111e-05,
+      "loss": 0.2333,
+      "step": 318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586666666666667e-05,
+      "loss": 0.2349,
+      "step": 319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582222222222222e-05,
+      "loss": 0.245,
+      "step": 320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577777777777778e-05,
+      "loss": 0.2405,
+      "step": 321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573333333333333e-05,
+      "loss": 0.2434,
+      "step": 322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568888888888889e-05,
+      "loss": 0.2335,
+      "step": 323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564444444444445e-05,
+      "loss": 0.2192,
+      "step": 324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.56e-05,
+      "loss": 0.2316,
+      "step": 325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.2307,
+      "step": 326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551111111111111e-05,
+      "loss": 0.2387,
+      "step": 327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546666666666667e-05,
+      "loss": 0.2347,
+      "step": 328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5422222222222226e-05,
+      "loss": 0.2296,
+      "step": 329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537777777777778e-05,
+      "loss": 0.233,
+      "step": 330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5333333333333336e-05,
+      "loss": 0.2287,
+      "step": 331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.528888888888889e-05,
+      "loss": 0.2398,
+      "step": 332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5244444444444447e-05,
+      "loss": 0.2425,
+      "step": 333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52e-05,
+      "loss": 0.2416,
+      "step": 334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515555555555556e-05,
+      "loss": 0.237,
+      "step": 335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.511111111111111e-05,
+      "loss": 0.2363,
+      "step": 336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506666666666667e-05,
+      "loss": 0.2514,
+      "step": 337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502222222222222e-05,
+      "loss": 0.2414,
+      "step": 338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4977777777777785e-05,
+      "loss": 0.2363,
+      "step": 339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493333333333333e-05,
+      "loss": 0.2423,
+      "step": 340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4888888888888895e-05,
+      "loss": 0.2299,
+      "step": 341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4844444444444444e-05,
+      "loss": 0.2386,
+      "step": 342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48e-05,
+      "loss": 0.2328,
+      "step": 343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475555555555556e-05,
+      "loss": 0.2373,
+      "step": 344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471111111111111e-05,
+      "loss": 0.2396,
+      "step": 345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466666666666667e-05,
+      "loss": 0.2332,
+      "step": 346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.462222222222222e-05,
+      "loss": 0.2374,
+      "step": 347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457777777777778e-05,
+      "loss": 0.2506,
+      "step": 348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453333333333334e-05,
+      "loss": 0.2246,
+      "step": 349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448888888888889e-05,
+      "loss": 0.2386,
+      "step": 350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.444444444444445e-05,
+      "loss": 0.2357,
+      "step": 351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 0.2372,
+      "step": 352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435555555555556e-05,
+      "loss": 0.2331,
+      "step": 353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431111111111111e-05,
+      "loss": 0.2207,
+      "step": 354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426666666666667e-05,
+      "loss": 0.2393,
+      "step": 355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4222222222222224e-05,
+      "loss": 0.2261,
+      "step": 356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417777777777778e-05,
+      "loss": 0.2447,
+      "step": 357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4133333333333334e-05,
+      "loss": 0.2437,
+      "step": 358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408888888888889e-05,
+      "loss": 0.2342,
+      "step": 359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4044444444444445e-05,
+      "loss": 0.2357,
+      "step": 360
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.229248046875,
+      "eval_margin_accuracy": 0.8949008861753441,
+      "eval_runtime": 3819.57,
+      "eval_samples_per_second": 196.052,
+      "eval_steps_per_second": 49.013,
+      "step": 360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.2385,
+      "step": 361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3955555555555555e-05,
+      "loss": 0.2471,
+      "step": 362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391111111111111e-05,
+      "loss": 0.2307,
+      "step": 363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3866666666666665e-05,
+      "loss": 0.2313,
+      "step": 364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382222222222222e-05,
+      "loss": 0.2306,
+      "step": 365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.377777777777778e-05,
+      "loss": 0.2339,
+      "step": 366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373333333333333e-05,
+      "loss": 0.2384,
+      "step": 367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368888888888889e-05,
+      "loss": 0.2338,
+      "step": 368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364444444444445e-05,
+      "loss": 0.2411,
+      "step": 369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.2393,
+      "step": 370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355555555555556e-05,
+      "loss": 0.2322,
+      "step": 371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3511111111111114e-05,
+      "loss": 0.2355,
+      "step": 372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346666666666667e-05,
+      "loss": 0.233,
+      "step": 373
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3422222222222224e-05,
+      "loss": 0.253,
+      "step": 374
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337777777777778e-05,
+      "loss": 0.2319,
+      "step": 375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.2362,
+      "step": 376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.328888888888889e-05,
+      "loss": 0.23,
+      "step": 377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3244444444444445e-05,
+      "loss": 0.225,
+      "step": 378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32e-05,
+      "loss": 0.2295,
+      "step": 379
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3155555555555556e-05,
+      "loss": 0.2264,
+      "step": 380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.311111111111112e-05,
+      "loss": 0.243,
+      "step": 381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3066666666666666e-05,
+      "loss": 0.2236,
+      "step": 382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.302222222222222e-05,
+      "loss": 0.2312,
+      "step": 383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.297777777777778e-05,
+      "loss": 0.238,
+      "step": 384
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.293333333333333e-05,
+      "loss": 0.2355,
+      "step": 385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2888888888888894e-05,
+      "loss": 0.2204,
+      "step": 386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.284444444444444e-05,
+      "loss": 0.2386,
+      "step": 387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 0.2263,
+      "step": 388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.275555555555555e-05,
+      "loss": 0.2286,
+      "step": 389
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2711111111111115e-05,
+      "loss": 0.2127,
+      "step": 390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.266666666666667e-05,
+      "loss": 0.2311,
+      "step": 391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2622222222222225e-05,
+      "loss": 0.2218,
+      "step": 392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.257777777777778e-05,
+      "loss": 0.2441,
+      "step": 393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253333333333333e-05,
+      "loss": 0.2337,
+      "step": 394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.248888888888889e-05,
+      "loss": 0.2371,
+      "step": 395
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2444444444444446e-05,
+      "loss": 0.234,
+      "step": 396
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.24e-05,
+      "loss": 0.2363,
+      "step": 397
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.235555555555556e-05,
+      "loss": 0.2385,
+      "step": 398
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.231111111111111e-05,
+      "loss": 0.2342,
+      "step": 399
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.226666666666667e-05,
+      "loss": 0.229,
+      "step": 400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222222222222223e-05,
+      "loss": 0.2453,
+      "step": 401
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.217777777777778e-05,
+      "loss": 0.2266,
+      "step": 402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.213333333333334e-05,
+      "loss": 0.2218,
+      "step": 403
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.208888888888889e-05,
+      "loss": 0.238,
+      "step": 404
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.204444444444444e-05,
+      "loss": 0.2243,
+      "step": 405
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.2228,
+      "step": 406
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1955555555555554e-05,
+      "loss": 0.2252,
+      "step": 407
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1911111111111116e-05,
+      "loss": 0.2368,
+      "step": 408
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1866666666666664e-05,
+      "loss": 0.2274,
+      "step": 409
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1822222222222226e-05,
+      "loss": 0.2243,
+      "step": 410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.177777777777778e-05,
+      "loss": 0.224,
+      "step": 411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.173333333333334e-05,
+      "loss": 0.2231,
+      "step": 412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.168888888888889e-05,
+      "loss": 0.2322,
+      "step": 413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.164444444444444e-05,
+      "loss": 0.2177,
+      "step": 414
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.16e-05,
+      "loss": 0.2355,
+      "step": 415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.155555555555556e-05,
+      "loss": 0.2292,
+      "step": 416
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.151111111111111e-05,
+      "loss": 0.2227,
+      "step": 417
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.146666666666667e-05,
+      "loss": 0.2362,
+      "step": 418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.142222222222222e-05,
+      "loss": 0.232,
+      "step": 419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137777777777778e-05,
+      "loss": 0.2335,
+      "step": 420
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.2269287109375,
+      "eval_margin_accuracy": 0.8952480917049929,
+      "eval_runtime": 3822.1229,
+      "eval_samples_per_second": 195.921,
+      "eval_steps_per_second": 48.98,
+      "step": 420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1333333333333334e-05,
+      "loss": 0.2174,
+      "step": 421
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.128888888888889e-05,
+      "loss": 0.2299,
+      "step": 422
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.124444444444445e-05,
+      "loss": 0.2404,
+      "step": 423
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.12e-05,
+      "loss": 0.2415,
+      "step": 424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1155555555555555e-05,
+      "loss": 0.2309,
+      "step": 425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.2376,
+      "step": 426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1066666666666665e-05,
+      "loss": 0.2383,
+      "step": 427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102222222222223e-05,
+      "loss": 0.2337,
+      "step": 428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0977777777777776e-05,
+      "loss": 0.219,
+      "step": 429
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.093333333333334e-05,
+      "loss": 0.2178,
+      "step": 430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.088888888888889e-05,
+      "loss": 0.2341,
+      "step": 431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.084444444444445e-05,
+      "loss": 0.2199,
+      "step": 432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.08e-05,
+      "loss": 0.214,
+      "step": 433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.075555555555556e-05,
+      "loss": 0.2352,
+      "step": 434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0711111111111114e-05,
+      "loss": 0.2348,
+      "step": 435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.066666666666667e-05,
+      "loss": 0.2312,
+      "step": 436
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0622222222222224e-05,
+      "loss": 0.2173,
+      "step": 437
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.057777777777778e-05,
+      "loss": 0.2327,
+      "step": 438
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0533333333333335e-05,
+      "loss": 0.2315,
+      "step": 439
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.048888888888889e-05,
+      "loss": 0.233,
+      "step": 440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.044444444444445e-05,
+      "loss": 0.237,
+      "step": 441
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.04e-05,
+      "loss": 0.2392,
+      "step": 442
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.035555555555556e-05,
+      "loss": 0.2225,
+      "step": 443
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.031111111111111e-05,
+      "loss": 0.2213,
+      "step": 444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0266666666666666e-05,
+      "loss": 0.2324,
+      "step": 445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0222222222222225e-05,
+      "loss": 0.232,
+      "step": 446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0177777777777776e-05,
+      "loss": 0.2321,
+      "step": 447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0133333333333335e-05,
+      "loss": 0.2325,
+      "step": 448
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.008888888888889e-05,
+      "loss": 0.2347,
+      "step": 449
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.004444444444445e-05,
+      "loss": 0.2196,
+      "step": 450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2272,
+      "step": 451
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.995555555555556e-05,
+      "loss": 0.2314,
+      "step": 452
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.991111111111111e-05,
+      "loss": 0.2294,
+      "step": 453
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.986666666666667e-05,
+      "loss": 0.2315,
+      "step": 454
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9822222222222225e-05,
+      "loss": 0.2377,
+      "step": 455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9777777777777777e-05,
+      "loss": 0.2254,
+      "step": 456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9733333333333336e-05,
+      "loss": 0.2146,
+      "step": 457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9688888888888887e-05,
+      "loss": 0.2211,
+      "step": 458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9644444444444446e-05,
+      "loss": 0.2318,
+      "step": 459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.96e-05,
+      "loss": 0.2314,
+      "step": 460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.955555555555556e-05,
+      "loss": 0.2296,
+      "step": 461
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.951111111111111e-05,
+      "loss": 0.2185,
+      "step": 462
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.946666666666667e-05,
+      "loss": 0.2305,
+      "step": 463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9422222222222222e-05,
+      "loss": 0.233,
+      "step": 464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.937777777777778e-05,
+      "loss": 0.2365,
+      "step": 465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9333333333333336e-05,
+      "loss": 0.2312,
+      "step": 466
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9288888888888888e-05,
+      "loss": 0.2326,
+      "step": 467
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9244444444444446e-05,
+      "loss": 0.2287,
+      "step": 468
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 0.2344,
+      "step": 469
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9155555555555557e-05,
+      "loss": 0.2197,
+      "step": 470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9111111111111112e-05,
+      "loss": 0.2278,
+      "step": 471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.906666666666667e-05,
+      "loss": 0.2302,
+      "step": 472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9022222222222223e-05,
+      "loss": 0.2341,
+      "step": 473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.897777777777778e-05,
+      "loss": 0.2168,
+      "step": 474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8933333333333333e-05,
+      "loss": 0.2291,
+      "step": 475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.2333,
+      "step": 476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8844444444444447e-05,
+      "loss": 0.2303,
+      "step": 477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.88e-05,
+      "loss": 0.2302,
+      "step": 478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8755555555555557e-05,
+      "loss": 0.2378,
+      "step": 479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8711111111111113e-05,
+      "loss": 0.2403,
+      "step": 480
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 0.2213134765625,
+      "eval_margin_accuracy": 0.8956593967170382,
+      "eval_runtime": 3813.375,
+      "eval_samples_per_second": 196.371,
+      "eval_steps_per_second": 49.093,
+      "step": 480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8666666666666668e-05,
+      "loss": 0.2172,
+      "step": 481
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8622222222222223e-05,
+      "loss": 0.2329,
+      "step": 482
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.857777777777778e-05,
+      "loss": 0.2254,
+      "step": 483
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8533333333333333e-05,
+      "loss": 0.232,
+      "step": 484
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8488888888888892e-05,
+      "loss": 0.2298,
+      "step": 485
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8444444444444447e-05,
+      "loss": 0.2358,
+      "step": 486
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.84e-05,
+      "loss": 0.224,
+      "step": 487
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8355555555555558e-05,
+      "loss": 0.2219,
+      "step": 488
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.831111111111111e-05,
+      "loss": 0.2412,
+      "step": 489
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8266666666666668e-05,
+      "loss": 0.2354,
+      "step": 490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8222222222222223e-05,
+      "loss": 0.2278,
+      "step": 491
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8177777777777782e-05,
+      "loss": 0.2367,
+      "step": 492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8133333333333334e-05,
+      "loss": 0.2275,
+      "step": 493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8088888888888893e-05,
+      "loss": 0.2215,
+      "step": 494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8044444444444444e-05,
+      "loss": 0.2226,
+      "step": 495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.229,
+      "step": 496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7955555555555558e-05,
+      "loss": 0.2331,
+      "step": 497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.791111111111111e-05,
+      "loss": 0.2257,
+      "step": 498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.786666666666667e-05,
+      "loss": 0.2374,
+      "step": 499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.782222222222222e-05,
+      "loss": 0.2202,
+      "step": 500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.2316,
+      "step": 501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7733333333333334e-05,
+      "loss": 0.2264,
+      "step": 502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7688888888888893e-05,
+      "loss": 0.2328,
+      "step": 503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7644444444444445e-05,
+      "loss": 0.2143,
+      "step": 504
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 0.2267,
+      "step": 505
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7555555555555555e-05,
+      "loss": 0.2138,
+      "step": 506
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.751111111111111e-05,
+      "loss": 0.2254,
+      "step": 507
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.746666666666667e-05,
+      "loss": 0.2241,
+      "step": 508
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.742222222222222e-05,
+      "loss": 0.2353,
+      "step": 509
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.737777777777778e-05,
+      "loss": 0.2186,
+      "step": 510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.733333333333333e-05,
+      "loss": 0.2221,
+      "step": 511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.728888888888889e-05,
+      "loss": 0.2276,
+      "step": 512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7244444444444445e-05,
+      "loss": 0.2253,
+      "step": 513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 0.2323,
+      "step": 514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7155555555555556e-05,
+      "loss": 0.2277,
+      "step": 515
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7111111111111114e-05,
+      "loss": 0.219,
+      "step": 516
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.706666666666667e-05,
+      "loss": 0.2167,
+      "step": 517
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.702222222222222e-05,
+      "loss": 0.2125,
+      "step": 518
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.697777777777778e-05,
+      "loss": 0.2319,
+      "step": 519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6933333333333332e-05,
+      "loss": 0.2287,
+      "step": 520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.688888888888889e-05,
+      "loss": 0.2251,
+      "step": 521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6844444444444446e-05,
+      "loss": 0.2332,
+      "step": 522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 0.2246,
+      "step": 523
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6755555555555556e-05,
+      "loss": 0.2337,
+      "step": 524
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6711111111111115e-05,
+      "loss": 0.2338,
+      "step": 525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.2248,
+      "step": 526
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6622222222222225e-05,
+      "loss": 0.234,
+      "step": 527
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.657777777777778e-05,
+      "loss": 0.2325,
+      "step": 528
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6533333333333332e-05,
+      "loss": 0.2309,
+      "step": 529
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.648888888888889e-05,
+      "loss": 0.2221,
+      "step": 530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6444444444444443e-05,
+      "loss": 0.2286,
+      "step": 531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.64e-05,
+      "loss": 0.227,
+      "step": 532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6355555555555557e-05,
+      "loss": 0.2196,
+      "step": 533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6311111111111115e-05,
+      "loss": 0.2339,
+      "step": 534
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6266666666666667e-05,
+      "loss": 0.2315,
+      "step": 535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6222222222222226e-05,
+      "loss": 0.2227,
+      "step": 536
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6177777777777777e-05,
+      "loss": 0.2175,
+      "step": 537
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6133333333333333e-05,
+      "loss": 0.2225,
+      "step": 538
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.608888888888889e-05,
+      "loss": 0.2165,
+      "step": 539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6044444444444443e-05,
+      "loss": 0.2302,
+      "step": 540
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 0.22265625,
+      "eval_margin_accuracy": 0.8962523169292075,
+      "eval_runtime": 3810.9828,
+      "eval_samples_per_second": 196.494,
+      "eval_steps_per_second": 49.124,
+      "step": 540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.2341,
+      "step": 541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5955555555555554e-05,
+      "loss": 0.2185,
+      "step": 542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5911111111111112e-05,
+      "loss": 0.2321,
+      "step": 543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5866666666666667e-05,
+      "loss": 0.2217,
+      "step": 544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5822222222222226e-05,
+      "loss": 0.2356,
+      "step": 545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5777777777777778e-05,
+      "loss": 0.2265,
+      "step": 546
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5733333333333337e-05,
+      "loss": 0.2228,
+      "step": 547
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5688888888888892e-05,
+      "loss": 0.2228,
+      "step": 548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5644444444444444e-05,
+      "loss": 0.2143,
+      "step": 549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.23,
+      "step": 550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 0.2325,
+      "step": 551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5511111111111113e-05,
+      "loss": 0.2321,
+      "step": 552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5466666666666668e-05,
+      "loss": 0.2315,
+      "step": 553
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5422222222222227e-05,
+      "loss": 0.2219,
+      "step": 554
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.537777777777778e-05,
+      "loss": 0.2338,
+      "step": 555
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5333333333333337e-05,
+      "loss": 0.2199,
+      "step": 556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.528888888888889e-05,
+      "loss": 0.2229,
+      "step": 557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5244444444444447e-05,
+      "loss": 0.2338,
+      "step": 558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 0.2176,
+      "step": 559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5155555555555555e-05,
+      "loss": 0.2316,
+      "step": 560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5111111111111113e-05,
+      "loss": 0.2227,
+      "step": 561
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5066666666666665e-05,
+      "loss": 0.2191,
+      "step": 562
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5022222222222224e-05,
+      "loss": 0.2355,
+      "step": 563
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.497777777777778e-05,
+      "loss": 0.2236,
+      "step": 564
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4933333333333334e-05,
+      "loss": 0.2115,
+      "step": 565
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.488888888888889e-05,
+      "loss": 0.222,
+      "step": 566
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4844444444444444e-05,
+      "loss": 0.2227,
+      "step": 567
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.48e-05,
+      "loss": 0.2207,
+      "step": 568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.475555555555556e-05,
+      "loss": 0.2332,
+      "step": 569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4711111111111114e-05,
+      "loss": 0.2336,
+      "step": 570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.466666666666667e-05,
+      "loss": 0.223,
+      "step": 571
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4622222222222224e-05,
+      "loss": 0.2205,
+      "step": 572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.457777777777778e-05,
+      "loss": 0.2266,
+      "step": 573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4533333333333334e-05,
+      "loss": 0.2272,
+      "step": 574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.448888888888889e-05,
+      "loss": 0.2239,
+      "step": 575
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 0.2203,
+      "step": 576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.44e-05,
+      "loss": 0.2187,
+      "step": 577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4355555555555555e-05,
+      "loss": 0.2219,
+      "step": 578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.431111111111111e-05,
+      "loss": 0.214,
+      "step": 579
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.426666666666667e-05,
+      "loss": 0.2238,
+      "step": 580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4222222222222224e-05,
+      "loss": 0.2312,
+      "step": 581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.417777777777778e-05,
+      "loss": 0.2218,
+      "step": 582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4133333333333335e-05,
+      "loss": 0.2175,
+      "step": 583
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.408888888888889e-05,
+      "loss": 0.2341,
+      "step": 584
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4044444444444445e-05,
+      "loss": 0.209,
+      "step": 585
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4e-05,
+      "loss": 0.2241,
+      "step": 586
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3955555555555556e-05,
+      "loss": 0.2158,
+      "step": 587
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.391111111111111e-05,
+      "loss": 0.2348,
+      "step": 588
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3866666666666666e-05,
+      "loss": 0.2163,
+      "step": 589
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3822222222222225e-05,
+      "loss": 0.2112,
+      "step": 590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.377777777777778e-05,
+      "loss": 0.2287,
+      "step": 591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3733333333333335e-05,
+      "loss": 0.2223,
+      "step": 592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.368888888888889e-05,
+      "loss": 0.2222,
+      "step": 593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3644444444444446e-05,
+      "loss": 0.2232,
+      "step": 594
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.36e-05,
+      "loss": 0.2213,
+      "step": 595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3555555555555556e-05,
+      "loss": 0.2172,
+      "step": 596
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.351111111111111e-05,
+      "loss": 0.2245,
+      "step": 597
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3466666666666667e-05,
+      "loss": 0.2223,
+      "step": 598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3422222222222222e-05,
+      "loss": 0.231,
+      "step": 599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.337777777777778e-05,
+      "loss": 0.2353,
+      "step": 600
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 0.22216796875,
+      "eval_margin_accuracy": 0.8961107639055814,
+      "eval_runtime": 3812.0607,
+      "eval_samples_per_second": 196.439,
+      "eval_steps_per_second": 49.11,
+      "step": 600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.2321,
+      "step": 601
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.328888888888889e-05,
+      "loss": 0.2263,
+      "step": 602
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3244444444444446e-05,
+      "loss": 0.2324,
+      "step": 603
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.32e-05,
+      "loss": 0.2072,
+      "step": 604
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3155555555555557e-05,
+      "loss": 0.2318,
+      "step": 605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3111111111111112e-05,
+      "loss": 0.233,
+      "step": 606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3066666666666667e-05,
+      "loss": 0.2395,
+      "step": 607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3022222222222222e-05,
+      "loss": 0.2227,
+      "step": 608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2977777777777778e-05,
+      "loss": 0.2241,
+      "step": 609
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2933333333333333e-05,
+      "loss": 0.2241,
+      "step": 610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.288888888888889e-05,
+      "loss": 0.219,
+      "step": 611
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2844444444444447e-05,
+      "loss": 0.2317,
+      "step": 612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.2211,
+      "step": 613
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2755555555555557e-05,
+      "loss": 0.2164,
+      "step": 614
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2711111111111112e-05,
+      "loss": 0.2229,
+      "step": 615
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2666666666666668e-05,
+      "loss": 0.2239,
+      "step": 616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2622222222222223e-05,
+      "loss": 0.2224,
+      "step": 617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2577777777777778e-05,
+      "loss": 0.2241,
+      "step": 618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2533333333333333e-05,
+      "loss": 0.2265,
+      "step": 619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.248888888888889e-05,
+      "loss": 0.2228,
+      "step": 620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2444444444444447e-05,
+      "loss": 0.2177,
+      "step": 621
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 0.2116,
+      "step": 622
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2355555555555558e-05,
+      "loss": 0.2299,
+      "step": 623
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2311111111111113e-05,
+      "loss": 0.2235,
+      "step": 624
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2266666666666668e-05,
+      "loss": 0.2149,
+      "step": 625
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.2219,
+      "step": 626
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.217777777777778e-05,
+      "loss": 0.2247,
+      "step": 627
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2133333333333334e-05,
+      "loss": 0.2297,
+      "step": 628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.208888888888889e-05,
+      "loss": 0.2244,
+      "step": 629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2044444444444444e-05,
+      "loss": 0.2126,
+      "step": 630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.2137,
+      "step": 631
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1955555555555558e-05,
+      "loss": 0.2301,
+      "step": 632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1911111111111113e-05,
+      "loss": 0.2262,
+      "step": 633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.186666666666667e-05,
+      "loss": 0.2274,
+      "step": 634
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1822222222222224e-05,
+      "loss": 0.2241,
+      "step": 635
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.177777777777778e-05,
+      "loss": 0.2288,
+      "step": 636
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1733333333333334e-05,
+      "loss": 0.2288,
+      "step": 637
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.168888888888889e-05,
+      "loss": 0.2181,
+      "step": 638
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1644444444444445e-05,
+      "loss": 0.2337,
+      "step": 639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.16e-05,
+      "loss": 0.2294,
+      "step": 640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1555555555555555e-05,
+      "loss": 0.2214,
+      "step": 641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1511111111111114e-05,
+      "loss": 0.2225,
+      "step": 642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.146666666666667e-05,
+      "loss": 0.2249,
+      "step": 643
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1422222222222224e-05,
+      "loss": 0.2223,
+      "step": 644
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.137777777777778e-05,
+      "loss": 0.2258,
+      "step": 645
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 0.2226,
+      "step": 646
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.128888888888889e-05,
+      "loss": 0.232,
+      "step": 647
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1244444444444445e-05,
+      "loss": 0.2231,
+      "step": 648
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.12e-05,
+      "loss": 0.2162,
+      "step": 649
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1155555555555556e-05,
+      "loss": 0.2316,
+      "step": 650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.111111111111111e-05,
+      "loss": 0.2332,
+      "step": 651
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.106666666666667e-05,
+      "loss": 0.2334,
+      "step": 652
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1022222222222225e-05,
+      "loss": 0.2292,
+      "step": 653
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.097777777777778e-05,
+      "loss": 0.2137,
+      "step": 654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0933333333333335e-05,
+      "loss": 0.2338,
+      "step": 655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.088888888888889e-05,
+      "loss": 0.2299,
+      "step": 656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0844444444444446e-05,
+      "loss": 0.2127,
+      "step": 657
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.08e-05,
+      "loss": 0.2419,
+      "step": 658
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0755555555555556e-05,
+      "loss": 0.2203,
+      "step": 659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.071111111111111e-05,
+      "loss": 0.2271,
+      "step": 660
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 0.220703125,
+      "eval_margin_accuracy": 0.8963938699528334,
+      "eval_runtime": 3803.281,
+      "eval_samples_per_second": 196.892,
+      "eval_steps_per_second": 49.223,
+      "step": 660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0666666666666666e-05,
+      "loss": 0.2216,
+      "step": 661
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.062222222222222e-05,
+      "loss": 0.2302,
+      "step": 662
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.057777777777778e-05,
+      "loss": 0.2299,
+      "step": 663
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0533333333333336e-05,
+      "loss": 0.2209,
+      "step": 664
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.048888888888889e-05,
+      "loss": 0.2315,
+      "step": 665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0444444444444446e-05,
+      "loss": 0.2208,
+      "step": 666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.04e-05,
+      "loss": 0.2271,
+      "step": 667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0355555555555556e-05,
+      "loss": 0.2349,
+      "step": 668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.031111111111111e-05,
+      "loss": 0.2357,
+      "step": 669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0266666666666667e-05,
+      "loss": 0.2426,
+      "step": 670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0222222222222222e-05,
+      "loss": 0.2241,
+      "step": 671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0177777777777777e-05,
+      "loss": 0.2237,
+      "step": 672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0133333333333336e-05,
+      "loss": 0.2211,
+      "step": 673
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.008888888888889e-05,
+      "loss": 0.2216,
+      "step": 674
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0044444444444446e-05,
+      "loss": 0.2401,
+      "step": 675
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.2339,
+      "step": 676
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9955555555555557e-05,
+      "loss": 0.2178,
+      "step": 677
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9911111111111112e-05,
+      "loss": 0.2183,
+      "step": 678
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 0.2225,
+      "step": 679
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9822222222222223e-05,
+      "loss": 0.2307,
+      "step": 680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9777777777777778e-05,
+      "loss": 0.2373,
+      "step": 681
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9733333333333333e-05,
+      "loss": 0.2243,
+      "step": 682
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.968888888888889e-05,
+      "loss": 0.2241,
+      "step": 683
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9644444444444447e-05,
+      "loss": 0.2201,
+      "step": 684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.2171,
+      "step": 685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.2253,
+      "step": 686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9511111111111113e-05,
+      "loss": 0.2222,
+      "step": 687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 0.2348,
+      "step": 688
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9422222222222223e-05,
+      "loss": 0.2165,
+      "step": 689
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9377777777777778e-05,
+      "loss": 0.2287,
+      "step": 690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 0.2266,
+      "step": 691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.928888888888889e-05,
+      "loss": 0.2299,
+      "step": 692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9244444444444444e-05,
+      "loss": 0.2312,
+      "step": 693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.2241,
+      "step": 694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9155555555555558e-05,
+      "loss": 0.2277,
+      "step": 695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.2365,
+      "step": 696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 0.2163,
+      "step": 697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9022222222222223e-05,
+      "loss": 0.2301,
+      "step": 698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.897777777777778e-05,
+      "loss": 0.2119,
+      "step": 699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 0.2223,
+      "step": 700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.224,
+      "step": 701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8844444444444444e-05,
+      "loss": 0.2262,
+      "step": 702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.88e-05,
+      "loss": 0.2252,
+      "step": 703
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8755555555555558e-05,
+      "loss": 0.2195,
+      "step": 704
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8711111111111113e-05,
+      "loss": 0.2229,
+      "step": 705
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.2254,
+      "step": 706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8622222222222224e-05,
+      "loss": 0.2162,
+      "step": 707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8577777777777776e-05,
+      "loss": 0.2319,
+      "step": 708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 0.2299,
+      "step": 709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.848888888888889e-05,
+      "loss": 0.213,
+      "step": 710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8444444444444445e-05,
+      "loss": 0.2346,
+      "step": 711
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.84e-05,
+      "loss": 0.2286,
+      "step": 712
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8355555555555555e-05,
+      "loss": 0.2145,
+      "step": 713
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8311111111111114e-05,
+      "loss": 0.2266,
+      "step": 714
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 0.2322,
+      "step": 715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 0.2294,
+      "step": 716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.817777777777778e-05,
+      "loss": 0.2091,
+      "step": 717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 0.2234,
+      "step": 718
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.808888888888889e-05,
+      "loss": 0.2298,
+      "step": 719
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8044444444444445e-05,
+      "loss": 0.228,
+      "step": 720
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 0.2218017578125,
+      "eval_margin_accuracy": 0.8967036841177507,
+      "eval_runtime": 3811.0304,
+      "eval_samples_per_second": 196.492,
+      "eval_steps_per_second": 49.123,
+      "step": 720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8e-05,
+      "loss": 0.2205,
+      "step": 721
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7955555555555556e-05,
+      "loss": 0.2082,
+      "step": 722
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.791111111111111e-05,
+      "loss": 0.2236,
+      "step": 723
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 0.2287,
+      "step": 724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7822222222222225e-05,
+      "loss": 0.232,
+      "step": 725
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 0.2321,
+      "step": 726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 0.2284,
+      "step": 727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.768888888888889e-05,
+      "loss": 0.2272,
+      "step": 728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7644444444444446e-05,
+      "loss": 0.2255,
+      "step": 729
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.76e-05,
+      "loss": 0.2261,
+      "step": 730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7555555555555556e-05,
+      "loss": 0.2237,
+      "step": 731
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.751111111111111e-05,
+      "loss": 0.2197,
+      "step": 732
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 0.2274,
+      "step": 733
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7422222222222222e-05,
+      "loss": 0.2245,
+      "step": 734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.737777777777778e-05,
+      "loss": 0.2294,
+      "step": 735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.22,
+      "step": 736
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.728888888888889e-05,
+      "loss": 0.2128,
+      "step": 737
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7244444444444446e-05,
+      "loss": 0.2265,
+      "step": 738
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 0.224,
+      "step": 739
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7155555555555557e-05,
+      "loss": 0.2265,
+      "step": 740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7111111111111112e-05,
+      "loss": 0.2155,
+      "step": 741
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7066666666666667e-05,
+      "loss": 0.2215,
+      "step": 742
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7022222222222222e-05,
+      "loss": 0.2224,
+      "step": 743
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6977777777777777e-05,
+      "loss": 0.2167,
+      "step": 744
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6933333333333333e-05,
+      "loss": 0.2261,
+      "step": 745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.2198,
+      "step": 746
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6844444444444447e-05,
+      "loss": 0.2218,
+      "step": 747
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.2207,
+      "step": 748
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6755555555555557e-05,
+      "loss": 0.2241,
+      "step": 749
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6711111111111112e-05,
+      "loss": 0.2195,
+      "step": 750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2188,
+      "step": 751
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6622222222222223e-05,
+      "loss": 0.2285,
+      "step": 752
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6577777777777778e-05,
+      "loss": 0.2218,
+      "step": 753
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.23,
+      "step": 754
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.648888888888889e-05,
+      "loss": 0.2264,
+      "step": 755
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6444444444444447e-05,
+      "loss": 0.2107,
+      "step": 756
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 0.2156,
+      "step": 757
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6355555555555557e-05,
+      "loss": 0.2183,
+      "step": 758
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6311111111111113e-05,
+      "loss": 0.2266,
+      "step": 759
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6266666666666665e-05,
+      "loss": 0.2231,
+      "step": 760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6222222222222223e-05,
+      "loss": 0.2118,
+      "step": 761
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.617777777777778e-05,
+      "loss": 0.2221,
+      "step": 762
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 0.2072,
+      "step": 763
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.608888888888889e-05,
+      "loss": 0.2192,
+      "step": 764
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6044444444444444e-05,
+      "loss": 0.2222,
+      "step": 765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2131,
+      "step": 766
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5955555555555558e-05,
+      "loss": 0.2258,
+      "step": 767
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5911111111111113e-05,
+      "loss": 0.2196,
+      "step": 768
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 0.2219,
+      "step": 769
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.582222222222222e-05,
+      "loss": 0.2234,
+      "step": 770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.577777777777778e-05,
+      "loss": 0.226,
+      "step": 771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 0.2283,
+      "step": 772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.568888888888889e-05,
+      "loss": 0.22,
+      "step": 773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5644444444444444e-05,
+      "loss": 0.2071,
+      "step": 774
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.56e-05,
+      "loss": 0.2238,
+      "step": 775
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.2283,
+      "step": 776
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5511111111111114e-05,
+      "loss": 0.2156,
+      "step": 777
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.2285,
+      "step": 778
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5422222222222224e-05,
+      "loss": 0.2141,
+      "step": 779
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.537777777777778e-05,
+      "loss": 0.2231,
+      "step": 780
+    },
+    {
+      "epoch": 2.08,
+      "eval_loss": 0.2200927734375,
+      "eval_margin_accuracy": 0.8966662927530193,
+      "eval_runtime": 3812.3767,
+      "eval_samples_per_second": 196.422,
+      "eval_steps_per_second": 49.106,
+      "step": 780
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.2166,
+      "step": 781
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.528888888888889e-05,
+      "loss": 0.2162,
+      "step": 782
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5244444444444445e-05,
+      "loss": 0.2079,
+      "step": 783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.52e-05,
+      "loss": 0.2242,
+      "step": 784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5155555555555555e-05,
+      "loss": 0.2316,
+      "step": 785
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.2213,
+      "step": 786
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 0.226,
+      "step": 787
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5022222222222224e-05,
+      "loss": 0.2265,
+      "step": 788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.497777777777778e-05,
+      "loss": 0.2283,
+      "step": 789
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.2159,
+      "step": 790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4888888888888888e-05,
+      "loss": 0.2139,
+      "step": 791
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4844444444444444e-05,
+      "loss": 0.2174,
+      "step": 792
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.48e-05,
+      "loss": 0.2197,
+      "step": 793
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4755555555555556e-05,
+      "loss": 0.2133,
+      "step": 794
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4711111111111111e-05,
+      "loss": 0.2194,
+      "step": 795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4666666666666668e-05,
+      "loss": 0.2228,
+      "step": 796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4622222222222223e-05,
+      "loss": 0.2212,
+      "step": 797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4577777777777778e-05,
+      "loss": 0.2301,
+      "step": 798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 0.2226,
+      "step": 799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.448888888888889e-05,
+      "loss": 0.218,
+      "step": 800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 0.2239,
+      "step": 801
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.44e-05,
+      "loss": 0.2249,
+      "step": 802
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4355555555555556e-05,
+      "loss": 0.2191,
+      "step": 803
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4311111111111111e-05,
+      "loss": 0.2257,
+      "step": 804
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4266666666666667e-05,
+      "loss": 0.2228,
+      "step": 805
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.224,
+      "step": 806
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4177777777777779e-05,
+      "loss": 0.2239,
+      "step": 807
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 0.2242,
+      "step": 808
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4088888888888891e-05,
+      "loss": 0.2243,
+      "step": 809
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4044444444444446e-05,
+      "loss": 0.2169,
+      "step": 810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.226,
+      "step": 811
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3955555555555555e-05,
+      "loss": 0.2283,
+      "step": 812
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.391111111111111e-05,
+      "loss": 0.2146,
+      "step": 813
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3866666666666667e-05,
+      "loss": 0.2156,
+      "step": 814
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3822222222222222e-05,
+      "loss": 0.2223,
+      "step": 815
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3777777777777778e-05,
+      "loss": 0.216,
+      "step": 816
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 0.2226,
+      "step": 817
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.368888888888889e-05,
+      "loss": 0.2141,
+      "step": 818
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3644444444444445e-05,
+      "loss": 0.2263,
+      "step": 819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.2023,
+      "step": 820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3555555555555557e-05,
+      "loss": 0.2183,
+      "step": 821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.351111111111111e-05,
+      "loss": 0.2218,
+      "step": 822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3466666666666666e-05,
+      "loss": 0.213,
+      "step": 823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3422222222222223e-05,
+      "loss": 0.2213,
+      "step": 824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3377777777777778e-05,
+      "loss": 0.2083,
+      "step": 825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.2228,
+      "step": 826
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.328888888888889e-05,
+      "loss": 0.2136,
+      "step": 827
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3244444444444445e-05,
+      "loss": 0.2195,
+      "step": 828
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.32e-05,
+      "loss": 0.2187,
+      "step": 829
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3155555555555558e-05,
+      "loss": 0.222,
+      "step": 830
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3111111111111113e-05,
+      "loss": 0.2335,
+      "step": 831
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3066666666666666e-05,
+      "loss": 0.2168,
+      "step": 832
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3022222222222222e-05,
+      "loss": 0.2135,
+      "step": 833
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2977777777777777e-05,
+      "loss": 0.2224,
+      "step": 834
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 0.2148,
+      "step": 835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2888888888888889e-05,
+      "loss": 0.2278,
+      "step": 836
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2844444444444446e-05,
+      "loss": 0.2277,
+      "step": 837
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.2221,
+      "step": 838
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2755555555555556e-05,
+      "loss": 0.2155,
+      "step": 839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2711111111111113e-05,
+      "loss": 0.2128,
+      "step": 840
+    },
+    {
+      "epoch": 2.24,
+      "eval_loss": 0.221923828125,
+      "eval_margin_accuracy": 0.8967410754824822,
+      "eval_runtime": 3807.3544,
+      "eval_samples_per_second": 196.681,
+      "eval_steps_per_second": 49.17,
+      "step": 840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2666666666666668e-05,
+      "loss": 0.2166,
+      "step": 841
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2622222222222224e-05,
+      "loss": 0.2052,
+      "step": 842
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2577777777777777e-05,
+      "loss": 0.212,
+      "step": 843
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2533333333333332e-05,
+      "loss": 0.2114,
+      "step": 844
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.248888888888889e-05,
+      "loss": 0.2128,
+      "step": 845
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2444444444444445e-05,
+      "loss": 0.2167,
+      "step": 846
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.24e-05,
+      "loss": 0.2292,
+      "step": 847
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2355555555555557e-05,
+      "loss": 0.2188,
+      "step": 848
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2311111111111112e-05,
+      "loss": 0.2212,
+      "step": 849
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.22,
+      "step": 850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2222222222222222e-05,
+      "loss": 0.2188,
+      "step": 851
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2177777777777778e-05,
+      "loss": 0.2166,
+      "step": 852
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 0.208,
+      "step": 853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.208888888888889e-05,
+      "loss": 0.2252,
+      "step": 854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2044444444444445e-05,
+      "loss": 0.2166,
+      "step": 855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2188,
+      "step": 856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1955555555555556e-05,
+      "loss": 0.2187,
+      "step": 857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1911111111111112e-05,
+      "loss": 0.2175,
+      "step": 858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1866666666666668e-05,
+      "loss": 0.2258,
+      "step": 859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1822222222222223e-05,
+      "loss": 0.212,
+      "step": 860
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1777777777777778e-05,
+      "loss": 0.2272,
+      "step": 861
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1733333333333333e-05,
+      "loss": 0.2139,
+      "step": 862
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.168888888888889e-05,
+      "loss": 0.2219,
+      "step": 863
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1644444444444446e-05,
+      "loss": 0.2176,
+      "step": 864
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.16e-05,
+      "loss": 0.2322,
+      "step": 865
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.222,
+      "step": 866
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1511111111111111e-05,
+      "loss": 0.2193,
+      "step": 867
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1466666666666666e-05,
+      "loss": 0.2232,
+      "step": 868
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1422222222222223e-05,
+      "loss": 0.2203,
+      "step": 869
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1377777777777779e-05,
+      "loss": 0.2221,
+      "step": 870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.2216,
+      "step": 871
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1288888888888889e-05,
+      "loss": 0.2222,
+      "step": 872
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1244444444444444e-05,
+      "loss": 0.2191,
+      "step": 873
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.2255,
+      "step": 874
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1155555555555556e-05,
+      "loss": 0.2196,
+      "step": 875
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.2114,
+      "step": 876
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1066666666666667e-05,
+      "loss": 0.2165,
+      "step": 877
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1022222222222222e-05,
+      "loss": 0.2193,
+      "step": 878
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0977777777777779e-05,
+      "loss": 0.2346,
+      "step": 879
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 0.2211,
+      "step": 880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.088888888888889e-05,
+      "loss": 0.21,
+      "step": 881
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0844444444444445e-05,
+      "loss": 0.2235,
+      "step": 882
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.08e-05,
+      "loss": 0.2174,
+      "step": 883
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0755555555555557e-05,
+      "loss": 0.2135,
+      "step": 884
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0711111111111112e-05,
+      "loss": 0.2235,
+      "step": 885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.2203,
+      "step": 886
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0622222222222223e-05,
+      "loss": 0.2203,
+      "step": 887
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0577777777777778e-05,
+      "loss": 0.2138,
+      "step": 888
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0533333333333335e-05,
+      "loss": 0.216,
+      "step": 889
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.048888888888889e-05,
+      "loss": 0.2178,
+      "step": 890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0444444444444445e-05,
+      "loss": 0.2137,
+      "step": 891
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.04e-05,
+      "loss": 0.2287,
+      "step": 892
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0355555555555556e-05,
+      "loss": 0.2285,
+      "step": 893
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.031111111111111e-05,
+      "loss": 0.2129,
+      "step": 894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 0.2211,
+      "step": 895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.2201,
+      "step": 896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0177777777777778e-05,
+      "loss": 0.2208,
+      "step": 897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0133333333333333e-05,
+      "loss": 0.2201,
+      "step": 898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0088888888888889e-05,
+      "loss": 0.2285,
+      "step": 899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0044444444444446e-05,
+      "loss": 0.2186,
+      "step": 900
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 0.22021484375,
+      "eval_margin_accuracy": 0.8966636219412528,
+      "eval_runtime": 4823.8909,
+      "eval_samples_per_second": 155.235,
+      "eval_steps_per_second": 38.809,
+      "step": 900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1e-05,
+      "loss": 0.2134,
+      "step": 901
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 0.2198,
+      "step": 902
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.911111111111111e-06,
+      "loss": 0.2161,
+      "step": 903
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.866666666666667e-06,
+      "loss": 0.2325,
+      "step": 904
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.822222222222223e-06,
+      "loss": 0.2287,
+      "step": 905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.2172,
+      "step": 906
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 0.2196,
+      "step": 907
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 0.216,
+      "step": 908
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.644444444444444e-06,
+      "loss": 0.2146,
+      "step": 909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.2295,
+      "step": 910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.555555555555556e-06,
+      "loss": 0.2207,
+      "step": 911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 0.2176,
+      "step": 912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 0.2082,
+      "step": 913
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 0.2195,
+      "step": 914
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.377777777777779e-06,
+      "loss": 0.2082,
+      "step": 915
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.2269,
+      "step": 916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.288888888888888e-06,
+      "loss": 0.2267,
+      "step": 917
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 0.2248,
+      "step": 918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.2e-06,
+      "loss": 0.2286,
+      "step": 919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 0.218,
+      "step": 920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.111111111111112e-06,
+      "loss": 0.2274,
+      "step": 921
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.222,
+      "step": 922
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.022222222222223e-06,
+      "loss": 0.2159,
+      "step": 923
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 0.2259,
+      "step": 924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 0.2217,
+      "step": 925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.88888888888889e-06,
+      "loss": 0.2232,
+      "step": 926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.844444444444445e-06,
+      "loss": 0.2188,
+      "step": 927
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.8e-06,
+      "loss": 0.2272,
+      "step": 928
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.755555555555556e-06,
+      "loss": 0.2229,
+      "step": 929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 0.2153,
+      "step": 930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.2213,
+      "step": 931
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 0.2084,
+      "step": 932
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.577777777777778e-06,
+      "loss": 0.2244,
+      "step": 933
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.533333333333334e-06,
+      "loss": 0.2178,
+      "step": 934
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.488888888888889e-06,
+      "loss": 0.2209,
+      "step": 935
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 0.222,
+      "step": 936
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.2155,
+      "step": 937
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 0.2309,
+      "step": 938
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.311111111111111e-06,
+      "loss": 0.2246,
+      "step": 939
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 0.2187,
+      "step": 940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.222222222222223e-06,
+      "loss": 0.2237,
+      "step": 941
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 0.2117,
+      "step": 942
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.133333333333332e-06,
+      "loss": 0.2165,
+      "step": 943
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 0.2193,
+      "step": 944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.044444444444444e-06,
+      "loss": 0.2219,
+      "step": 945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2143,
+      "step": 946
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.955555555555557e-06,
+      "loss": 0.2224,
+      "step": 947
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.91111111111111e-06,
+      "loss": 0.2142,
+      "step": 948
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 0.2253,
+      "step": 949
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.822222222222222e-06,
+      "loss": 0.2276,
+      "step": 950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.777777777777777e-06,
+      "loss": 0.2141,
+      "step": 951
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 0.2149,
+      "step": 952
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.68888888888889e-06,
+      "loss": 0.2219,
+      "step": 953
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 0.2232,
+      "step": 954
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.6e-06,
+      "loss": 0.2189,
+      "step": 955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 0.2131,
+      "step": 956
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.511111111111112e-06,
+      "loss": 0.218,
+      "step": 957
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.2173,
+      "step": 958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.422222222222222e-06,
+      "loss": 0.2116,
+      "step": 959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 0.2245,
+      "step": 960
+    },
+    {
+      "epoch": 2.56,
+      "eval_loss": 0.220458984375,
+      "eval_margin_accuracy": 0.8969493988002714,
+      "eval_runtime": 3825.3088,
+      "eval_samples_per_second": 195.758,
+      "eval_steps_per_second": 48.94,
+      "step": 960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.333333333333334e-06,
+      "loss": 0.2275,
+      "step": 961
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.288888888888889e-06,
+      "loss": 0.2246,
+      "step": 962
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.244444444444445e-06,
+      "loss": 0.2148,
+      "step": 963
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2e-06,
+      "loss": 0.2203,
+      "step": 964
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.155555555555556e-06,
+      "loss": 0.2147,
+      "step": 965
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 0.2136,
+      "step": 966
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 0.2251,
+      "step": 967
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.022222222222223e-06,
+      "loss": 0.2163,
+      "step": 968
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9777777777777775e-06,
+      "loss": 0.2146,
+      "step": 969
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.933333333333334e-06,
+      "loss": 0.2167,
+      "step": 970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.888888888888889e-06,
+      "loss": 0.2271,
+      "step": 971
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.844444444444445e-06,
+      "loss": 0.2275,
+      "step": 972
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.2266,
+      "step": 973
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.755555555555555e-06,
+      "loss": 0.2285,
+      "step": 974
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.711111111111111e-06,
+      "loss": 0.2153,
+      "step": 975
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.2272,
+      "step": 976
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.622222222222223e-06,
+      "loss": 0.2253,
+      "step": 977
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.577777777777779e-06,
+      "loss": 0.2157,
+      "step": 978
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.533333333333333e-06,
+      "loss": 0.2214,
+      "step": 979
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.488888888888888e-06,
+      "loss": 0.2182,
+      "step": 980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4444444444444445e-06,
+      "loss": 0.2158,
+      "step": 981
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.2227,
+      "step": 982
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.355555555555557e-06,
+      "loss": 0.213,
+      "step": 983
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.311111111111112e-06,
+      "loss": 0.2097,
+      "step": 984
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.266666666666666e-06,
+      "loss": 0.2224,
+      "step": 985
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.222222222222222e-06,
+      "loss": 0.2095,
+      "step": 986
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.177777777777778e-06,
+      "loss": 0.2151,
+      "step": 987
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 0.2185,
+      "step": 988
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.088888888888889e-06,
+      "loss": 0.2205,
+      "step": 989
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.044444444444445e-06,
+      "loss": 0.2256,
+      "step": 990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6e-06,
+      "loss": 0.2148,
+      "step": 991
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.955555555555556e-06,
+      "loss": 0.2178,
+      "step": 992
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9111111111111115e-06,
+      "loss": 0.206,
+      "step": 993
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.866666666666667e-06,
+      "loss": 0.2258,
+      "step": 994
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.822222222222223e-06,
+      "loss": 0.2314,
+      "step": 995
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 0.2218,
+      "step": 996
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.733333333333333e-06,
+      "loss": 0.2156,
+      "step": 997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.688888888888889e-06,
+      "loss": 0.216,
+      "step": 998
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6444444444444445e-06,
+      "loss": 0.2192,
+      "step": 999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.2173,
+      "step": 1000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.2204,
+      "step": 1001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.511111111111111e-06,
+      "loss": 0.2052,
+      "step": 1002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 0.2193,
+      "step": 1003
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.422222222222222e-06,
+      "loss": 0.2342,
+      "step": 1004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3777777777777784e-06,
+      "loss": 0.2194,
+      "step": 1005
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.2208,
+      "step": 1006
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.288888888888889e-06,
+      "loss": 0.2155,
+      "step": 1007
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.244444444444445e-06,
+      "loss": 0.2157,
+      "step": 1008
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2e-06,
+      "loss": 0.2103,
+      "step": 1009
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.155555555555555e-06,
+      "loss": 0.2129,
+      "step": 1010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.1111111111111115e-06,
+      "loss": 0.216,
+      "step": 1011
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.066666666666667e-06,
+      "loss": 0.2109,
+      "step": 1012
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.022222222222223e-06,
+      "loss": 0.22,
+      "step": 1013
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.977777777777778e-06,
+      "loss": 0.2154,
+      "step": 1014
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.933333333333333e-06,
+      "loss": 0.2165,
+      "step": 1015
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.2179,
+      "step": 1016
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8444444444444446e-06,
+      "loss": 0.2198,
+      "step": 1017
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.2178,
+      "step": 1018
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.755555555555556e-06,
+      "loss": 0.2244,
+      "step": 1019
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.711111111111111e-06,
+      "loss": 0.2158,
+      "step": 1020
+    },
+    {
+      "epoch": 2.72,
+      "eval_loss": 0.2196044921875,
+      "eval_margin_accuracy": 0.8969307031179057,
+      "eval_runtime": 3801.4629,
+      "eval_samples_per_second": 196.986,
+      "eval_steps_per_second": 49.247,
+      "step": 1020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.2128,
+      "step": 1021
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.622222222222222e-06,
+      "loss": 0.2074,
+      "step": 1022
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5777777777777785e-06,
+      "loss": 0.2215,
+      "step": 1023
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 0.1993,
+      "step": 1024
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.488888888888889e-06,
+      "loss": 0.2148,
+      "step": 1025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.444444444444445e-06,
+      "loss": 0.2078,
+      "step": 1026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4e-06,
+      "loss": 0.2142,
+      "step": 1027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 0.2143,
+      "step": 1028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3111111111111115e-06,
+      "loss": 0.2266,
+      "step": 1029
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.266666666666667e-06,
+      "loss": 0.2171,
+      "step": 1030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.222222222222223e-06,
+      "loss": 0.223,
+      "step": 1031
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.177777777777778e-06,
+      "loss": 0.2186,
+      "step": 1032
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 0.2213,
+      "step": 1033
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.088888888888889e-06,
+      "loss": 0.2157,
+      "step": 1034
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.044444444444445e-06,
+      "loss": 0.2304,
+      "step": 1035
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.2118,
+      "step": 1036
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.955555555555555e-06,
+      "loss": 0.2153,
+      "step": 1037
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.911111111111111e-06,
+      "loss": 0.2121,
+      "step": 1038
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 0.2156,
+      "step": 1039
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8222222222222224e-06,
+      "loss": 0.217,
+      "step": 1040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.777777777777778e-06,
+      "loss": 0.2128,
+      "step": 1041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 0.2198,
+      "step": 1042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.688888888888889e-06,
+      "loss": 0.2224,
+      "step": 1043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6444444444444446e-06,
+      "loss": 0.2233,
+      "step": 1044
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6e-06,
+      "loss": 0.2222,
+      "step": 1045
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 0.2285,
+      "step": 1046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5111111111111116e-06,
+      "loss": 0.2101,
+      "step": 1047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.466666666666667e-06,
+      "loss": 0.2131,
+      "step": 1048
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4222222222222224e-06,
+      "loss": 0.2128,
+      "step": 1049
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3777777777777777e-06,
+      "loss": 0.2109,
+      "step": 1050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.2171,
+      "step": 1051
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2888888888888894e-06,
+      "loss": 0.2161,
+      "step": 1052
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.244444444444444e-06,
+      "loss": 0.2151,
+      "step": 1053
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.2206,
+      "step": 1054
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.155555555555556e-06,
+      "loss": 0.2227,
+      "step": 1055
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.111111111111111e-06,
+      "loss": 0.2221,
+      "step": 1056
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 0.2205,
+      "step": 1057
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0222222222222225e-06,
+      "loss": 0.2116,
+      "step": 1058
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.977777777777778e-06,
+      "loss": 0.2175,
+      "step": 1059
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9333333333333333e-06,
+      "loss": 0.2162,
+      "step": 1060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.888888888888889e-06,
+      "loss": 0.2196,
+      "step": 1061
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8444444444444446e-06,
+      "loss": 0.2169,
+      "step": 1062
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.2102,
+      "step": 1063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7555555555555555e-06,
+      "loss": 0.2119,
+      "step": 1064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.711111111111111e-06,
+      "loss": 0.2322,
+      "step": 1065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.2151,
+      "step": 1066
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6222222222222225e-06,
+      "loss": 0.2252,
+      "step": 1067
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5777777777777777e-06,
+      "loss": 0.2022,
+      "step": 1068
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5333333333333334e-06,
+      "loss": 0.2198,
+      "step": 1069
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.488888888888889e-06,
+      "loss": 0.2279,
+      "step": 1070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4444444444444447e-06,
+      "loss": 0.2259,
+      "step": 1071
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.2212,
+      "step": 1072
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3555555555555555e-06,
+      "loss": 0.2155,
+      "step": 1073
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.311111111111111e-06,
+      "loss": 0.2199,
+      "step": 1074
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 0.2224,
+      "step": 1075
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2222222222222225e-06,
+      "loss": 0.2171,
+      "step": 1076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1777777777777777e-06,
+      "loss": 0.219,
+      "step": 1077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1333333333333334e-06,
+      "loss": 0.224,
+      "step": 1078
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.088888888888889e-06,
+      "loss": 0.2294,
+      "step": 1079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0444444444444447e-06,
+      "loss": 0.2106,
+      "step": 1080
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 0.21923828125,
+      "eval_margin_accuracy": 0.8967864792825131,
+      "eval_runtime": 3809.3536,
+      "eval_samples_per_second": 196.578,
+      "eval_steps_per_second": 49.145,
+      "step": 1080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.2139,
+      "step": 1081
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9555555555555556e-06,
+      "loss": 0.2157,
+      "step": 1082
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9111111111111112e-06,
+      "loss": 0.233,
+      "step": 1083
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 0.2267,
+      "step": 1084
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8222222222222223e-06,
+      "loss": 0.205,
+      "step": 1085
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 0.2133,
+      "step": 1086
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7333333333333334e-06,
+      "loss": 0.2154,
+      "step": 1087
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6888888888888888e-06,
+      "loss": 0.2125,
+      "step": 1088
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6444444444444447e-06,
+      "loss": 0.2188,
+      "step": 1089
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.2309,
+      "step": 1090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5555555555555556e-06,
+      "loss": 0.2258,
+      "step": 1091
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5111111111111112e-06,
+      "loss": 0.2128,
+      "step": 1092
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4666666666666667e-06,
+      "loss": 0.2098,
+      "step": 1093
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4222222222222223e-06,
+      "loss": 0.2117,
+      "step": 1094
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3777777777777778e-06,
+      "loss": 0.2201,
+      "step": 1095
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.2182,
+      "step": 1096
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2888888888888889e-06,
+      "loss": 0.2223,
+      "step": 1097
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2444444444444445e-06,
+      "loss": 0.2046,
+      "step": 1098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.2165,
+      "step": 1099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1555555555555556e-06,
+      "loss": 0.2241,
+      "step": 1100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1111111111111112e-06,
+      "loss": 0.2204,
+      "step": 1101
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0666666666666667e-06,
+      "loss": 0.2129,
+      "step": 1102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0222222222222223e-06,
+      "loss": 0.2257,
+      "step": 1103
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.777777777777778e-07,
+      "loss": 0.2151,
+      "step": 1104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 0.2138,
+      "step": 1105
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 0.2147,
+      "step": 1106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.444444444444444e-07,
+      "loss": 0.2074,
+      "step": 1107
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.2122,
+      "step": 1108
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.555555555555556e-07,
+      "loss": 0.2031,
+      "step": 1109
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.111111111111112e-07,
+      "loss": 0.2189,
+      "step": 1110
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.2188,
+      "step": 1111
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.222222222222223e-07,
+      "loss": 0.219,
+      "step": 1112
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.777777777777778e-07,
+      "loss": 0.2225,
+      "step": 1113
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.333333333333333e-07,
+      "loss": 0.2218,
+      "step": 1114
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.888888888888889e-07,
+      "loss": 0.2126,
+      "step": 1115
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 0.2138,
+      "step": 1116
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.2238,
+      "step": 1117
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.555555555555556e-07,
+      "loss": 0.2092,
+      "step": 1118
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.111111111111111e-07,
+      "loss": 0.2263,
+      "step": 1119
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6666666666666667e-07,
+      "loss": 0.2151,
+      "step": 1120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2222222222222224e-07,
+      "loss": 0.2112,
+      "step": 1121
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.777777777777778e-07,
+      "loss": 0.2159,
+      "step": 1122
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3333333333333334e-07,
+      "loss": 0.2169,
+      "step": 1123
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 0.227,
+      "step": 1124
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.444444444444445e-08,
+      "loss": 0.2218,
+      "step": 1125
+    },
+    {
+      "epoch": 3.0,
+      "step": 1125,
+      "total_flos": 4.0255434320637235e+18,
+      "train_loss": 0.2445831168492635,
+      "train_runtime": 242015.9535,
+      "train_samples_per_second": 37.188,
+      "train_steps_per_second": 0.005
+    }
+  ],
+  "max_steps": 1125,
+  "num_train_epochs": 3,
+  "total_flos": 4.0255434320637235e+18,
+  "trial_name": null,
+  "trial_params": null
+}