diff --git "a/checkpoint-2770/trainer_state.json" "b/checkpoint-2770/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2770/trainer_state.json"
@@ -0,0 +1,16726 @@
+{
+  "best_metric": 3.7452244758605957,
+  "best_model_checkpoint": "bird_sounds_classification/checkpoint-2770",
+  "epoch": 10.0,
+  "global_step": 2770,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.8957,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.921,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.9181,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.9141,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.9233,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.9152,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.9263,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.9305,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.9151,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.8992,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1e-06,
+      "loss": 3.9155,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.9264,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3e-06,
+      "loss": 3.8954,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.9049,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5e-06,
+      "loss": 3.9178,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.9094,
+      "step": 16
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.9141,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.8963,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.9003,
+      "step": 19
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.9276,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.9261,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2e-06,
+      "loss": 3.9129,
+      "step": 22
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.9343,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.987,
+      "step": 24
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5e-06,
+      "loss": 3.9166,
+      "step": 25
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6e-06,
+      "loss": 3.9105,
+      "step": 26
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.9047,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.9042,
+      "step": 28
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9e-06,
+      "loss": 3.9153,
+      "step": 29
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3e-06,
+      "loss": 3.9159,
+      "step": 30
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.9023,
+      "step": 31
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.8961,
+      "step": 32
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.8725,
+      "step": 33
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.8763,
+      "step": 34
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5e-06,
+      "loss": 3.9041,
+      "step": 35
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.8972,
+      "step": 36
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.7e-06,
+      "loss": 3.9123,
+      "step": 37
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.8862,
+      "step": 38
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.9151,
+      "step": 39
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.9086,
+      "step": 40
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.1e-06,
+      "loss": 3.8992,
+      "step": 41
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.8783,
+      "step": 42
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3e-06,
+      "loss": 3.9076,
+      "step": 43
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4e-06,
+      "loss": 3.9,
+      "step": 44
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5e-06,
+      "loss": 3.9286,
+      "step": 45
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.9123,
+      "step": 46
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7e-06,
+      "loss": 3.8866,
+      "step": 47
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.8716,
+      "step": 48
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.9158,
+      "step": 49
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 3.9134,
+      "step": 50
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.998161764705883e-06,
+      "loss": 3.9056,
+      "step": 51
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.996323529411765e-06,
+      "loss": 3.9211,
+      "step": 52
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.994485294117647e-06,
+      "loss": 3.9112,
+      "step": 53
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99264705882353e-06,
+      "loss": 3.9093,
+      "step": 54
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.990808823529413e-06,
+      "loss": 3.9111,
+      "step": 55
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.988970588235294e-06,
+      "loss": 3.9031,
+      "step": 56
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.987132352941177e-06,
+      "loss": 3.8199,
+      "step": 57
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9852941176470596e-06,
+      "loss": 3.9104,
+      "step": 58
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.983455882352941e-06,
+      "loss": 3.9171,
+      "step": 59
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.981617647058824e-06,
+      "loss": 3.9245,
+      "step": 60
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9797794117647065e-06,
+      "loss": 3.9053,
+      "step": 61
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.977941176470588e-06,
+      "loss": 3.9025,
+      "step": 62
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.976102941176471e-06,
+      "loss": 3.9206,
+      "step": 63
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9742647058823535e-06,
+      "loss": 3.9126,
+      "step": 64
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.972426470588236e-06,
+      "loss": 3.9147,
+      "step": 65
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.970588235294118e-06,
+      "loss": 3.9052,
+      "step": 66
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9687500000000005e-06,
+      "loss": 3.9048,
+      "step": 67
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.966911764705883e-06,
+      "loss": 3.8963,
+      "step": 68
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.965073529411765e-06,
+      "loss": 3.9229,
+      "step": 69
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9632352941176475e-06,
+      "loss": 3.9114,
+      "step": 70
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.96139705882353e-06,
+      "loss": 3.917,
+      "step": 71
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.959558823529413e-06,
+      "loss": 3.9059,
+      "step": 72
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9577205882352944e-06,
+      "loss": 3.9141,
+      "step": 73
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955882352941177e-06,
+      "loss": 3.9358,
+      "step": 74
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95404411764706e-06,
+      "loss": 3.9199,
+      "step": 75
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.952205882352941e-06,
+      "loss": 3.9629,
+      "step": 76
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.950367647058824e-06,
+      "loss": 3.9253,
+      "step": 77
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.948529411764707e-06,
+      "loss": 3.9178,
+      "step": 78
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.946691176470589e-06,
+      "loss": 3.8573,
+      "step": 79
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.944852941176471e-06,
+      "loss": 3.923,
+      "step": 80
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.943014705882354e-06,
+      "loss": 3.9113,
+      "step": 81
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.941176470588236e-06,
+      "loss": 3.9223,
+      "step": 82
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.939338235294118e-06,
+      "loss": 3.9211,
+      "step": 83
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.937500000000001e-06,
+      "loss": 3.9393,
+      "step": 84
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.935661764705883e-06,
+      "loss": 3.9066,
+      "step": 85
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.933823529411765e-06,
+      "loss": 3.9651,
+      "step": 86
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9319852941176476e-06,
+      "loss": 3.9318,
+      "step": 87
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.93014705882353e-06,
+      "loss": 3.8814,
+      "step": 88
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.928308823529413e-06,
+      "loss": 3.9313,
+      "step": 89
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9264705882352945e-06,
+      "loss": 3.9193,
+      "step": 90
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.924632352941177e-06,
+      "loss": 3.9002,
+      "step": 91
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.92279411764706e-06,
+      "loss": 3.9175,
+      "step": 92
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9209558823529415e-06,
+      "loss": 3.9267,
+      "step": 93
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.919117647058823e-06,
+      "loss": 3.9336,
+      "step": 94
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.917279411764706e-06,
+      "loss": 3.887,
+      "step": 95
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9154411764705885e-06,
+      "loss": 3.9319,
+      "step": 96
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.913602941176471e-06,
+      "loss": 3.8965,
+      "step": 97
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.911764705882353e-06,
+      "loss": 3.9333,
+      "step": 98
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9099264705882355e-06,
+      "loss": 3.8966,
+      "step": 99
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.908088235294118e-06,
+      "loss": 3.9095,
+      "step": 100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.90625e-06,
+      "loss": 3.9153,
+      "step": 101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9044117647058824e-06,
+      "loss": 3.8886,
+      "step": 102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.902573529411765e-06,
+      "loss": 3.925,
+      "step": 103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.900735294117648e-06,
+      "loss": 3.9202,
+      "step": 104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.898897058823529e-06,
+      "loss": 3.9166,
+      "step": 105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.897058823529412e-06,
+      "loss": 3.8953,
+      "step": 106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.895220588235295e-06,
+      "loss": 3.9142,
+      "step": 107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.893382352941176e-06,
+      "loss": 3.9326,
+      "step": 108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.891544117647059e-06,
+      "loss": 3.9234,
+      "step": 109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.889705882352942e-06,
+      "loss": 3.9074,
+      "step": 110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.887867647058823e-06,
+      "loss": 3.9124,
+      "step": 111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.886029411764706e-06,
+      "loss": 3.9224,
+      "step": 112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8841911764705886e-06,
+      "loss": 3.9149,
+      "step": 113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.882352941176471e-06,
+      "loss": 3.902,
+      "step": 114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.880514705882353e-06,
+      "loss": 3.8917,
+      "step": 115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8786764705882355e-06,
+      "loss": 3.9184,
+      "step": 116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.876838235294118e-06,
+      "loss": 3.9209,
+      "step": 117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.875e-06,
+      "loss": 3.9136,
+      "step": 118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8731617647058825e-06,
+      "loss": 3.9409,
+      "step": 119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.871323529411765e-06,
+      "loss": 3.9229,
+      "step": 120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.869485294117648e-06,
+      "loss": 3.9351,
+      "step": 121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8676470588235295e-06,
+      "loss": 3.9235,
+      "step": 122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.865808823529412e-06,
+      "loss": 3.9185,
+      "step": 123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.863970588235295e-06,
+      "loss": 3.9277,
+      "step": 124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8621323529411765e-06,
+      "loss": 3.9142,
+      "step": 125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.860294117647059e-06,
+      "loss": 3.9086,
+      "step": 126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.858455882352942e-06,
+      "loss": 3.9022,
+      "step": 127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.856617647058824e-06,
+      "loss": 3.8916,
+      "step": 128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.854779411764706e-06,
+      "loss": 3.8947,
+      "step": 129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.852941176470589e-06,
+      "loss": 3.9271,
+      "step": 130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.851102941176471e-06,
+      "loss": 3.9083,
+      "step": 131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.849264705882353e-06,
+      "loss": 3.8508,
+      "step": 132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.847426470588236e-06,
+      "loss": 3.8974,
+      "step": 133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.845588235294118e-06,
+      "loss": 3.9153,
+      "step": 134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.84375e-06,
+      "loss": 3.9164,
+      "step": 135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.841911764705883e-06,
+      "loss": 3.922,
+      "step": 136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.840073529411765e-06,
+      "loss": 3.9043,
+      "step": 137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.838235294117648e-06,
+      "loss": 3.9602,
+      "step": 138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.83639705882353e-06,
+      "loss": 3.9293,
+      "step": 139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.834558823529412e-06,
+      "loss": 3.9465,
+      "step": 140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.832720588235295e-06,
+      "loss": 3.9197,
+      "step": 141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8308823529411766e-06,
+      "loss": 3.9176,
+      "step": 142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.829044117647059e-06,
+      "loss": 3.9099,
+      "step": 143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.827205882352942e-06,
+      "loss": 3.9161,
+      "step": 144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.825367647058824e-06,
+      "loss": 3.9185,
+      "step": 145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823529411764706e-06,
+      "loss": 3.9115,
+      "step": 146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821691176470589e-06,
+      "loss": 3.9147,
+      "step": 147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.819852941176471e-06,
+      "loss": 3.9376,
+      "step": 148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818014705882353e-06,
+      "loss": 3.915,
+      "step": 149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.816176470588236e-06,
+      "loss": 3.9177,
+      "step": 150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.814338235294118e-06,
+      "loss": 3.9417,
+      "step": 151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8125e-06,
+      "loss": 3.8834,
+      "step": 152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.810661764705883e-06,
+      "loss": 3.912,
+      "step": 153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.808823529411765e-06,
+      "loss": 3.9026,
+      "step": 154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.806985294117648e-06,
+      "loss": 3.9035,
+      "step": 155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.80514705882353e-06,
+      "loss": 3.9025,
+      "step": 156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.803308823529412e-06,
+      "loss": 3.8928,
+      "step": 157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.801470588235295e-06,
+      "loss": 3.9144,
+      "step": 158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.799632352941177e-06,
+      "loss": 3.9156,
+      "step": 159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.797794117647059e-06,
+      "loss": 3.9111,
+      "step": 160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.795955882352942e-06,
+      "loss": 3.9151,
+      "step": 161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7941176470588245e-06,
+      "loss": 3.8999,
+      "step": 162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.792279411764706e-06,
+      "loss": 3.9154,
+      "step": 163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.790441176470589e-06,
+      "loss": 3.899,
+      "step": 164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7886029411764714e-06,
+      "loss": 3.9119,
+      "step": 165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.786764705882353e-06,
+      "loss": 3.8925,
+      "step": 166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.784926470588236e-06,
+      "loss": 3.9321,
+      "step": 167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.783088235294118e-06,
+      "loss": 3.909,
+      "step": 168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.781250000000001e-06,
+      "loss": 3.9066,
+      "step": 169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.779411764705883e-06,
+      "loss": 3.9001,
+      "step": 170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.777573529411765e-06,
+      "loss": 3.9022,
+      "step": 171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.775735294117648e-06,
+      "loss": 3.9078,
+      "step": 172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.77389705882353e-06,
+      "loss": 3.9196,
+      "step": 173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.772058823529412e-06,
+      "loss": 3.8968,
+      "step": 174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.770220588235295e-06,
+      "loss": 3.9011,
+      "step": 175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.768382352941177e-06,
+      "loss": 3.9063,
+      "step": 176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.766544117647059e-06,
+      "loss": 3.9207,
+      "step": 177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.764705882352941e-06,
+      "loss": 3.9207,
+      "step": 178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.762867647058824e-06,
+      "loss": 3.9065,
+      "step": 179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.761029411764706e-06,
+      "loss": 3.9008,
+      "step": 180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.759191176470588e-06,
+      "loss": 3.9283,
+      "step": 181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.757352941176471e-06,
+      "loss": 3.9035,
+      "step": 182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.755514705882353e-06,
+      "loss": 3.9002,
+      "step": 183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.753676470588235e-06,
+      "loss": 3.8927,
+      "step": 184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.751838235294118e-06,
+      "loss": 3.9271,
+      "step": 185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.75e-06,
+      "loss": 3.9176,
+      "step": 186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.748161764705883e-06,
+      "loss": 3.9195,
+      "step": 187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.746323529411765e-06,
+      "loss": 3.91,
+      "step": 188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.744485294117647e-06,
+      "loss": 3.8993,
+      "step": 189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74264705882353e-06,
+      "loss": 3.8949,
+      "step": 190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740808823529412e-06,
+      "loss": 3.8935,
+      "step": 191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.738970588235294e-06,
+      "loss": 3.8936,
+      "step": 192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.737132352941177e-06,
+      "loss": 3.8571,
+      "step": 193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7352941176470594e-06,
+      "loss": 3.9071,
+      "step": 194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.733455882352941e-06,
+      "loss": 3.9012,
+      "step": 195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.731617647058824e-06,
+      "loss": 3.9317,
+      "step": 196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.729779411764706e-06,
+      "loss": 3.8995,
+      "step": 197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.727941176470588e-06,
+      "loss": 3.9174,
+      "step": 198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.726102941176471e-06,
+      "loss": 3.8955,
+      "step": 199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.724264705882353e-06,
+      "loss": 3.9231,
+      "step": 200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.722426470588235e-06,
+      "loss": 3.936,
+      "step": 201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.720588235294118e-06,
+      "loss": 3.9145,
+      "step": 202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.71875e-06,
+      "loss": 3.9248,
+      "step": 203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.716911764705883e-06,
+      "loss": 3.9302,
+      "step": 204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.715073529411765e-06,
+      "loss": 3.9142,
+      "step": 205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.713235294117647e-06,
+      "loss": 3.9368,
+      "step": 206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.71139705882353e-06,
+      "loss": 3.9272,
+      "step": 207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.709558823529412e-06,
+      "loss": 3.9115,
+      "step": 208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.707720588235294e-06,
+      "loss": 3.9266,
+      "step": 209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 3.9063,
+      "step": 210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7040441176470595e-06,
+      "loss": 3.8973,
+      "step": 211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.702205882352941e-06,
+      "loss": 3.9274,
+      "step": 212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.700367647058824e-06,
+      "loss": 3.9189,
+      "step": 213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6985294117647065e-06,
+      "loss": 3.91,
+      "step": 214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.696691176470588e-06,
+      "loss": 3.9023,
+      "step": 215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.694852941176471e-06,
+      "loss": 3.9333,
+      "step": 216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6930147058823535e-06,
+      "loss": 3.9163,
+      "step": 217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.691176470588236e-06,
+      "loss": 3.9132,
+      "step": 218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.689338235294118e-06,
+      "loss": 3.8849,
+      "step": 219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 3.9059,
+      "step": 220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.685661764705883e-06,
+      "loss": 3.9018,
+      "step": 221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.683823529411765e-06,
+      "loss": 3.901,
+      "step": 222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.681985294117647e-06,
+      "loss": 3.9107,
+      "step": 223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.68014705882353e-06,
+      "loss": 3.8986,
+      "step": 224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.678308823529412e-06,
+      "loss": 3.9104,
+      "step": 225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.676470588235294e-06,
+      "loss": 3.9185,
+      "step": 226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.674632352941177e-06,
+      "loss": 3.8912,
+      "step": 227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.67279411764706e-06,
+      "loss": 3.9225,
+      "step": 228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.670955882352941e-06,
+      "loss": 3.9269,
+      "step": 229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.669117647058824e-06,
+      "loss": 3.9236,
+      "step": 230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.6672794117647066e-06,
+      "loss": 3.8892,
+      "step": 231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.665441176470588e-06,
+      "loss": 3.9083,
+      "step": 232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.663602941176471e-06,
+      "loss": 3.9088,
+      "step": 233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6617647058823535e-06,
+      "loss": 3.9078,
+      "step": 234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.659926470588236e-06,
+      "loss": 3.9213,
+      "step": 235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.658088235294118e-06,
+      "loss": 3.9141,
+      "step": 236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.6562500000000005e-06,
+      "loss": 3.8859,
+      "step": 237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.654411764705883e-06,
+      "loss": 3.906,
+      "step": 238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.652573529411765e-06,
+      "loss": 3.8916,
+      "step": 239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6507352941176475e-06,
+      "loss": 3.9085,
+      "step": 240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.64889705882353e-06,
+      "loss": 3.9039,
+      "step": 241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.647058823529412e-06,
+      "loss": 3.9172,
+      "step": 242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6452205882352945e-06,
+      "loss": 3.8915,
+      "step": 243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.643382352941177e-06,
+      "loss": 3.8961,
+      "step": 244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.64154411764706e-06,
+      "loss": 3.8966,
+      "step": 245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6397058823529414e-06,
+      "loss": 3.9654,
+      "step": 246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.637867647058824e-06,
+      "loss": 3.9111,
+      "step": 247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.636029411764707e-06,
+      "loss": 3.928,
+      "step": 248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.634191176470588e-06,
+      "loss": 3.9172,
+      "step": 249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.632352941176471e-06,
+      "loss": 3.9005,
+      "step": 250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630514705882354e-06,
+      "loss": 3.9194,
+      "step": 251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.628676470588236e-06,
+      "loss": 3.8748,
+      "step": 252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.626838235294118e-06,
+      "loss": 3.911,
+      "step": 253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.625000000000001e-06,
+      "loss": 3.9159,
+      "step": 254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.623161764705883e-06,
+      "loss": 3.8986,
+      "step": 255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.621323529411765e-06,
+      "loss": 3.9252,
+      "step": 256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.619485294117648e-06,
+      "loss": 3.9239,
+      "step": 257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.61764705882353e-06,
+      "loss": 3.8983,
+      "step": 258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.615808823529413e-06,
+      "loss": 3.9108,
+      "step": 259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6139705882352946e-06,
+      "loss": 3.9053,
+      "step": 260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.612132352941177e-06,
+      "loss": 3.9081,
+      "step": 261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.61029411764706e-06,
+      "loss": 3.9596,
+      "step": 262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6084558823529415e-06,
+      "loss": 3.8736,
+      "step": 263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.606617647058823e-06,
+      "loss": 3.9105,
+      "step": 264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.604779411764706e-06,
+      "loss": 3.8659,
+      "step": 265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6029411764705885e-06,
+      "loss": 3.8871,
+      "step": 266
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.601102941176471e-06,
+      "loss": 3.9245,
+      "step": 267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.599264705882353e-06,
+      "loss": 3.9141,
+      "step": 268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5974264705882355e-06,
+      "loss": 3.913,
+      "step": 269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.595588235294118e-06,
+      "loss": 3.9259,
+      "step": 270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.59375e-06,
+      "loss": 3.917,
+      "step": 271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5919117647058825e-06,
+      "loss": 3.9239,
+      "step": 272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.590073529411765e-06,
+      "loss": 3.9198,
+      "step": 273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.588235294117647e-06,
+      "loss": 3.9101,
+      "step": 274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5863970588235294e-06,
+      "loss": 3.9076,
+      "step": 275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.584558823529412e-06,
+      "loss": 3.9136,
+      "step": 276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.582720588235295e-06,
+      "loss": 3.9107,
+      "step": 277
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.02702702702702703,
+      "eval_loss": 3.9059979915618896,
+      "eval_runtime": 142.1807,
+      "eval_samples_per_second": 2.602,
+      "eval_steps_per_second": 0.654,
+      "step": 277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.580882352941176e-06,
+      "loss": 3.9043,
+      "step": 278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.579044117647059e-06,
+      "loss": 3.919,
+      "step": 279
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.577205882352942e-06,
+      "loss": 3.9638,
+      "step": 280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.575367647058823e-06,
+      "loss": 3.8705,
+      "step": 281
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.573529411764706e-06,
+      "loss": 3.8757,
+      "step": 282
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.571691176470589e-06,
+      "loss": 3.8841,
+      "step": 283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.569852941176471e-06,
+      "loss": 3.9085,
+      "step": 284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.568014705882353e-06,
+      "loss": 3.8842,
+      "step": 285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.5661764705882356e-06,
+      "loss": 3.9095,
+      "step": 286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.564338235294118e-06,
+      "loss": 3.9127,
+      "step": 287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5625e-06,
+      "loss": 3.8955,
+      "step": 288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5606617647058825e-06,
+      "loss": 3.9008,
+      "step": 289
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.558823529411765e-06,
+      "loss": 3.881,
+      "step": 290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.556985294117648e-06,
+      "loss": 3.8898,
+      "step": 291
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5551470588235295e-06,
+      "loss": 3.924,
+      "step": 292
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.553308823529412e-06,
+      "loss": 3.9063,
+      "step": 293
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.551470588235295e-06,
+      "loss": 3.9213,
+      "step": 294
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5496323529411765e-06,
+      "loss": 3.9123,
+      "step": 295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.547794117647059e-06,
+      "loss": 3.867,
+      "step": 296
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.545955882352942e-06,
+      "loss": 3.8986,
+      "step": 297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5441176470588235e-06,
+      "loss": 3.935,
+      "step": 298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.542279411764706e-06,
+      "loss": 3.8983,
+      "step": 299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.540441176470589e-06,
+      "loss": 3.9206,
+      "step": 300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.538602941176471e-06,
+      "loss": 3.9229,
+      "step": 301
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.536764705882353e-06,
+      "loss": 3.8912,
+      "step": 302
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.534926470588236e-06,
+      "loss": 3.9187,
+      "step": 303
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.533088235294118e-06,
+      "loss": 3.9085,
+      "step": 304
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.53125e-06,
+      "loss": 3.9166,
+      "step": 305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.529411764705883e-06,
+      "loss": 3.9341,
+      "step": 306
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.527573529411765e-06,
+      "loss": 3.9027,
+      "step": 307
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.525735294117648e-06,
+      "loss": 3.8721,
+      "step": 308
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.52389705882353e-06,
+      "loss": 3.9293,
+      "step": 309
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522058823529412e-06,
+      "loss": 3.9006,
+      "step": 310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520220588235295e-06,
+      "loss": 3.8877,
+      "step": 311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.518382352941177e-06,
+      "loss": 3.919,
+      "step": 312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516544117647059e-06,
+      "loss": 3.9037,
+      "step": 313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514705882352942e-06,
+      "loss": 3.9116,
+      "step": 314
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5128676470588236e-06,
+      "loss": 3.8164,
+      "step": 315
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.511029411764706e-06,
+      "loss": 3.9438,
+      "step": 316
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.509191176470589e-06,
+      "loss": 3.9129,
+      "step": 317
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.507352941176471e-06,
+      "loss": 3.9021,
+      "step": 318
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.505514705882353e-06,
+      "loss": 3.9306,
+      "step": 319
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.503676470588236e-06,
+      "loss": 3.9272,
+      "step": 320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.501838235294118e-06,
+      "loss": 3.9075,
+      "step": 321
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.5e-06,
+      "loss": 3.8948,
+      "step": 322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.498161764705883e-06,
+      "loss": 3.9115,
+      "step": 323
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.496323529411765e-06,
+      "loss": 3.9039,
+      "step": 324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.494485294117648e-06,
+      "loss": 3.8994,
+      "step": 325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.49264705882353e-06,
+      "loss": 3.9024,
+      "step": 326
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.490808823529412e-06,
+      "loss": 3.8941,
+      "step": 327
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.488970588235295e-06,
+      "loss": 3.9055,
+      "step": 328
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.487132352941177e-06,
+      "loss": 3.9013,
+      "step": 329
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.485294117647059e-06,
+      "loss": 3.8317,
+      "step": 330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.483455882352942e-06,
+      "loss": 3.9012,
+      "step": 331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4816176470588245e-06,
+      "loss": 3.9125,
+      "step": 332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.479779411764706e-06,
+      "loss": 3.9104,
+      "step": 333
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.477941176470589e-06,
+      "loss": 3.9224,
+      "step": 334
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4761029411764715e-06,
+      "loss": 3.9048,
+      "step": 335
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.474264705882353e-06,
+      "loss": 3.9096,
+      "step": 336
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.472426470588236e-06,
+      "loss": 3.9211,
+      "step": 337
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4705882352941184e-06,
+      "loss": 3.9193,
+      "step": 338
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.46875e-06,
+      "loss": 3.9223,
+      "step": 339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.466911764705883e-06,
+      "loss": 3.923,
+      "step": 340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.465073529411765e-06,
+      "loss": 3.8889,
+      "step": 341
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.463235294117648e-06,
+      "loss": 3.9045,
+      "step": 342
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.46139705882353e-06,
+      "loss": 3.8967,
+      "step": 343
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.459558823529412e-06,
+      "loss": 3.9626,
+      "step": 344
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.457720588235295e-06,
+      "loss": 3.9064,
+      "step": 345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.455882352941177e-06,
+      "loss": 3.877,
+      "step": 346
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.454044117647059e-06,
+      "loss": 3.8755,
+      "step": 347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.452205882352941e-06,
+      "loss": 3.9059,
+      "step": 348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.450367647058824e-06,
+      "loss": 3.8842,
+      "step": 349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.448529411764706e-06,
+      "loss": 3.9051,
+      "step": 350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.446691176470588e-06,
+      "loss": 3.9528,
+      "step": 351
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.444852941176471e-06,
+      "loss": 3.9053,
+      "step": 352
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.443014705882353e-06,
+      "loss": 3.9222,
+      "step": 353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.441176470588235e-06,
+      "loss": 3.8908,
+      "step": 354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.439338235294118e-06,
+      "loss": 3.8965,
+      "step": 355
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.4375e-06,
+      "loss": 3.8937,
+      "step": 356
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.435661764705883e-06,
+      "loss": 3.9123,
+      "step": 357
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.433823529411765e-06,
+      "loss": 3.8119,
+      "step": 358
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.431985294117647e-06,
+      "loss": 3.9067,
+      "step": 359
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.43014705882353e-06,
+      "loss": 3.8936,
+      "step": 360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.428308823529412e-06,
+      "loss": 3.9095,
+      "step": 361
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.426470588235294e-06,
+      "loss": 3.9131,
+      "step": 362
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.424632352941177e-06,
+      "loss": 3.9189,
+      "step": 363
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.422794117647059e-06,
+      "loss": 3.9028,
+      "step": 364
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.420955882352941e-06,
+      "loss": 3.911,
+      "step": 365
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.419117647058824e-06,
+      "loss": 3.8861,
+      "step": 366
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.4172794117647064e-06,
+      "loss": 3.9194,
+      "step": 367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.415441176470588e-06,
+      "loss": 3.8888,
+      "step": 368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.413602941176471e-06,
+      "loss": 3.9209,
+      "step": 369
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 3.8949,
+      "step": 370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.409926470588235e-06,
+      "loss": 3.8525,
+      "step": 371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.408088235294118e-06,
+      "loss": 3.9011,
+      "step": 372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.40625e-06,
+      "loss": 3.9155,
+      "step": 373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.404411764705883e-06,
+      "loss": 3.9018,
+      "step": 374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.402573529411765e-06,
+      "loss": 3.9371,
+      "step": 375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.400735294117647e-06,
+      "loss": 4.0396,
+      "step": 376
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.39889705882353e-06,
+      "loss": 3.9046,
+      "step": 377
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.397058823529412e-06,
+      "loss": 3.9383,
+      "step": 378
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.395220588235294e-06,
+      "loss": 3.9031,
+      "step": 379
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.393382352941177e-06,
+      "loss": 3.9034,
+      "step": 380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3915441176470595e-06,
+      "loss": 3.9138,
+      "step": 381
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.389705882352941e-06,
+      "loss": 3.8821,
+      "step": 382
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.387867647058824e-06,
+      "loss": 3.9014,
+      "step": 383
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3860294117647065e-06,
+      "loss": 3.9251,
+      "step": 384
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.384191176470588e-06,
+      "loss": 3.9485,
+      "step": 385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.382352941176471e-06,
+      "loss": 3.8887,
+      "step": 386
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3805147058823535e-06,
+      "loss": 3.8645,
+      "step": 387
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.378676470588235e-06,
+      "loss": 3.8943,
+      "step": 388
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.376838235294118e-06,
+      "loss": 3.9062,
+      "step": 389
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 3.894,
+      "step": 390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.373161764705883e-06,
+      "loss": 3.8975,
+      "step": 391
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.371323529411765e-06,
+      "loss": 3.9143,
+      "step": 392
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3694852941176474e-06,
+      "loss": 3.9158,
+      "step": 393
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.36764705882353e-06,
+      "loss": 3.914,
+      "step": 394
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.365808823529412e-06,
+      "loss": 3.8992,
+      "step": 395
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.363970588235294e-06,
+      "loss": 3.8928,
+      "step": 396
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.362132352941177e-06,
+      "loss": 3.9955,
+      "step": 397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.36029411764706e-06,
+      "loss": 3.8875,
+      "step": 398
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.358455882352941e-06,
+      "loss": 3.9393,
+      "step": 399
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.356617647058824e-06,
+      "loss": 3.9529,
+      "step": 400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.354779411764707e-06,
+      "loss": 3.8982,
+      "step": 401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.352941176470588e-06,
+      "loss": 3.8867,
+      "step": 402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.351102941176471e-06,
+      "loss": 3.921,
+      "step": 403
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.349264705882354e-06,
+      "loss": 3.9074,
+      "step": 404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.347426470588235e-06,
+      "loss": 3.9156,
+      "step": 405
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.345588235294118e-06,
+      "loss": 3.8489,
+      "step": 406
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.3437500000000006e-06,
+      "loss": 3.8925,
+      "step": 407
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.341911764705883e-06,
+      "loss": 3.8975,
+      "step": 408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.340073529411765e-06,
+      "loss": 3.9195,
+      "step": 409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.3382352941176475e-06,
+      "loss": 3.9293,
+      "step": 410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.33639705882353e-06,
+      "loss": 3.8872,
+      "step": 411
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.334558823529412e-06,
+      "loss": 3.9238,
+      "step": 412
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.3327205882352945e-06,
+      "loss": 3.8997,
+      "step": 413
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.330882352941177e-06,
+      "loss": 3.8918,
+      "step": 414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.32904411764706e-06,
+      "loss": 3.9261,
+      "step": 415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3272058823529415e-06,
+      "loss": 3.8922,
+      "step": 416
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.325367647058824e-06,
+      "loss": 3.8748,
+      "step": 417
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.323529411764707e-06,
+      "loss": 3.886,
+      "step": 418
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.3216911764705884e-06,
+      "loss": 3.906,
+      "step": 419
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.319852941176471e-06,
+      "loss": 3.8752,
+      "step": 420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.318014705882354e-06,
+      "loss": 3.8896,
+      "step": 421
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.316176470588236e-06,
+      "loss": 3.9088,
+      "step": 422
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.314338235294118e-06,
+      "loss": 3.8822,
+      "step": 423
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.312500000000001e-06,
+      "loss": 3.8848,
+      "step": 424
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.310661764705883e-06,
+      "loss": 3.8989,
+      "step": 425
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.308823529411765e-06,
+      "loss": 3.9137,
+      "step": 426
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.306985294117648e-06,
+      "loss": 3.9107,
+      "step": 427
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.30514705882353e-06,
+      "loss": 3.8984,
+      "step": 428
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.303308823529412e-06,
+      "loss": 3.9207,
+      "step": 429
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.301470588235295e-06,
+      "loss": 3.9043,
+      "step": 430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.299632352941177e-06,
+      "loss": 3.8906,
+      "step": 431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.29779411764706e-06,
+      "loss": 3.9199,
+      "step": 432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2959558823529416e-06,
+      "loss": 3.8849,
+      "step": 433
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.294117647058823e-06,
+      "loss": 3.9311,
+      "step": 434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.292279411764706e-06,
+      "loss": 3.9074,
+      "step": 435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2904411764705885e-06,
+      "loss": 3.8945,
+      "step": 436
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.28860294117647e-06,
+      "loss": 3.9328,
+      "step": 437
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.286764705882353e-06,
+      "loss": 3.8798,
+      "step": 438
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2849264705882355e-06,
+      "loss": 3.9068,
+      "step": 439
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.283088235294118e-06,
+      "loss": 3.9126,
+      "step": 440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.28125e-06,
+      "loss": 3.8877,
+      "step": 441
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2794117647058825e-06,
+      "loss": 3.9105,
+      "step": 442
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.277573529411765e-06,
+      "loss": 3.8941,
+      "step": 443
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.275735294117647e-06,
+      "loss": 3.9122,
+      "step": 444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2738970588235295e-06,
+      "loss": 3.8897,
+      "step": 445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.272058823529412e-06,
+      "loss": 3.9235,
+      "step": 446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.270220588235295e-06,
+      "loss": 3.9015,
+      "step": 447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2683823529411764e-06,
+      "loss": 3.9018,
+      "step": 448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.266544117647059e-06,
+      "loss": 3.8936,
+      "step": 449
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.264705882352942e-06,
+      "loss": 3.8925,
+      "step": 450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.262867647058823e-06,
+      "loss": 3.8914,
+      "step": 451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.261029411764706e-06,
+      "loss": 3.895,
+      "step": 452
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.259191176470589e-06,
+      "loss": 3.907,
+      "step": 453
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.25735294117647e-06,
+      "loss": 3.9336,
+      "step": 454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.255514705882353e-06,
+      "loss": 3.9143,
+      "step": 455
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.253676470588236e-06,
+      "loss": 3.9045,
+      "step": 456
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.251838235294118e-06,
+      "loss": 3.8867,
+      "step": 457
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.25e-06,
+      "loss": 3.8994,
+      "step": 458
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2481617647058826e-06,
+      "loss": 3.9215,
+      "step": 459
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.246323529411765e-06,
+      "loss": 3.9025,
+      "step": 460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.244485294117647e-06,
+      "loss": 3.9166,
+      "step": 461
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.2426470588235295e-06,
+      "loss": 3.9022,
+      "step": 462
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.240808823529412e-06,
+      "loss": 3.9368,
+      "step": 463
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.238970588235295e-06,
+      "loss": 3.9049,
+      "step": 464
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.2371323529411765e-06,
+      "loss": 3.8937,
+      "step": 465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.235294117647059e-06,
+      "loss": 3.9108,
+      "step": 466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.233455882352942e-06,
+      "loss": 3.8961,
+      "step": 467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2316176470588235e-06,
+      "loss": 3.8642,
+      "step": 468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.229779411764706e-06,
+      "loss": 3.9033,
+      "step": 469
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.227941176470589e-06,
+      "loss": 3.8794,
+      "step": 470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.226102941176471e-06,
+      "loss": 3.9012,
+      "step": 471
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.224264705882353e-06,
+      "loss": 3.9066,
+      "step": 472
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.222426470588236e-06,
+      "loss": 3.895,
+      "step": 473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.220588235294118e-06,
+      "loss": 3.9152,
+      "step": 474
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.21875e-06,
+      "loss": 3.8955,
+      "step": 475
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.216911764705883e-06,
+      "loss": 3.8859,
+      "step": 476
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.215073529411765e-06,
+      "loss": 3.9166,
+      "step": 477
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.213235294117647e-06,
+      "loss": 3.9213,
+      "step": 478
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.21139705882353e-06,
+      "loss": 3.8926,
+      "step": 479
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.209558823529412e-06,
+      "loss": 3.9371,
+      "step": 480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.207720588235295e-06,
+      "loss": 3.8835,
+      "step": 481
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.205882352941177e-06,
+      "loss": 3.9003,
+      "step": 482
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.204044117647059e-06,
+      "loss": 3.9181,
+      "step": 483
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.202205882352942e-06,
+      "loss": 3.892,
+      "step": 484
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.200367647058824e-06,
+      "loss": 3.8665,
+      "step": 485
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.198529411764706e-06,
+      "loss": 3.9085,
+      "step": 486
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.196691176470589e-06,
+      "loss": 3.8802,
+      "step": 487
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.194852941176471e-06,
+      "loss": 3.9198,
+      "step": 488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.193014705882353e-06,
+      "loss": 3.9018,
+      "step": 489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.191176470588236e-06,
+      "loss": 3.9065,
+      "step": 490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.189338235294118e-06,
+      "loss": 3.9043,
+      "step": 491
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1875e-06,
+      "loss": 3.9025,
+      "step": 492
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.185661764705883e-06,
+      "loss": 3.8515,
+      "step": 493
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.183823529411765e-06,
+      "loss": 3.8925,
+      "step": 494
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.181985294117647e-06,
+      "loss": 3.9417,
+      "step": 495
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.18014705882353e-06,
+      "loss": 3.9228,
+      "step": 496
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.178308823529412e-06,
+      "loss": 3.9237,
+      "step": 497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.176470588235295e-06,
+      "loss": 3.9108,
+      "step": 498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.174632352941177e-06,
+      "loss": 3.9109,
+      "step": 499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.172794117647059e-06,
+      "loss": 3.8823,
+      "step": 500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.170955882352942e-06,
+      "loss": 3.914,
+      "step": 501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.169117647058824e-06,
+      "loss": 3.8861,
+      "step": 502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.167279411764706e-06,
+      "loss": 3.9059,
+      "step": 503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.165441176470589e-06,
+      "loss": 3.8935,
+      "step": 504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1636029411764715e-06,
+      "loss": 3.9075,
+      "step": 505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.161764705882353e-06,
+      "loss": 3.9218,
+      "step": 506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.159926470588236e-06,
+      "loss": 3.9114,
+      "step": 507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.1580882352941185e-06,
+      "loss": 3.8665,
+      "step": 508
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.15625e-06,
+      "loss": 3.8972,
+      "step": 509
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.154411764705883e-06,
+      "loss": 3.9366,
+      "step": 510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.1525735294117654e-06,
+      "loss": 3.9204,
+      "step": 511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.150735294117648e-06,
+      "loss": 3.903,
+      "step": 512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.14889705882353e-06,
+      "loss": 3.9079,
+      "step": 513
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.147058823529412e-06,
+      "loss": 3.9246,
+      "step": 514
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.145220588235295e-06,
+      "loss": 3.9133,
+      "step": 515
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.143382352941177e-06,
+      "loss": 3.9275,
+      "step": 516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.141544117647059e-06,
+      "loss": 3.8852,
+      "step": 517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.139705882352941e-06,
+      "loss": 3.8748,
+      "step": 518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.137867647058824e-06,
+      "loss": 3.9109,
+      "step": 519
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.136029411764706e-06,
+      "loss": 3.9106,
+      "step": 520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.134191176470588e-06,
+      "loss": 3.8687,
+      "step": 521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.132352941176471e-06,
+      "loss": 3.9115,
+      "step": 522
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.130514705882353e-06,
+      "loss": 3.9291,
+      "step": 523
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.128676470588235e-06,
+      "loss": 3.891,
+      "step": 524
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.126838235294118e-06,
+      "loss": 3.9054,
+      "step": 525
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.125e-06,
+      "loss": 3.8861,
+      "step": 526
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.123161764705882e-06,
+      "loss": 3.883,
+      "step": 527
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.121323529411765e-06,
+      "loss": 3.9183,
+      "step": 528
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.119485294117647e-06,
+      "loss": 3.8538,
+      "step": 529
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 3.9181,
+      "step": 530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.115808823529412e-06,
+      "loss": 3.9074,
+      "step": 531
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.113970588235294e-06,
+      "loss": 3.859,
+      "step": 532
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.112132352941177e-06,
+      "loss": 3.8948,
+      "step": 533
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.110294117647059e-06,
+      "loss": 3.8735,
+      "step": 534
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.108455882352941e-06,
+      "loss": 3.9044,
+      "step": 535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.106617647058824e-06,
+      "loss": 3.9045,
+      "step": 536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1047794117647065e-06,
+      "loss": 3.8887,
+      "step": 537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.102941176470588e-06,
+      "loss": 3.8561,
+      "step": 538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.101102941176471e-06,
+      "loss": 3.8849,
+      "step": 539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0992647058823534e-06,
+      "loss": 3.9215,
+      "step": 540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.097426470588235e-06,
+      "loss": 3.916,
+      "step": 541
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.095588235294118e-06,
+      "loss": 3.8836,
+      "step": 542
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.09375e-06,
+      "loss": 3.881,
+      "step": 543
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.091911764705882e-06,
+      "loss": 3.8885,
+      "step": 544
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.090073529411765e-06,
+      "loss": 3.9141,
+      "step": 545
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.088235294117647e-06,
+      "loss": 3.9221,
+      "step": 546
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.08639705882353e-06,
+      "loss": 3.9106,
+      "step": 547
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.084558823529412e-06,
+      "loss": 3.8529,
+      "step": 548
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.082720588235294e-06,
+      "loss": 3.853,
+      "step": 549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.080882352941177e-06,
+      "loss": 3.9045,
+      "step": 550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.079044117647059e-06,
+      "loss": 3.9129,
+      "step": 551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.077205882352941e-06,
+      "loss": 3.8988,
+      "step": 552
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.075367647058824e-06,
+      "loss": 3.9316,
+      "step": 553
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0735294117647065e-06,
+      "loss": 3.9005,
+      "step": 554
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.04594594594594595,
+      "eval_loss": 3.8943464756011963,
+      "eval_runtime": 141.8971,
+      "eval_samples_per_second": 2.608,
+      "eval_steps_per_second": 0.655,
+      "step": 554
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.071691176470588e-06,
+      "loss": 3.8755,
+      "step": 555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.069852941176471e-06,
+      "loss": 3.9586,
+      "step": 556
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0680147058823535e-06,
+      "loss": 3.8969,
+      "step": 557
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.066176470588235e-06,
+      "loss": 3.9125,
+      "step": 558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.064338235294118e-06,
+      "loss": 3.8738,
+      "step": 559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 3.9892,
+      "step": 560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.060661764705883e-06,
+      "loss": 3.9066,
+      "step": 561
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.058823529411765e-06,
+      "loss": 3.8973,
+      "step": 562
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.0569852941176475e-06,
+      "loss": 3.9196,
+      "step": 563
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.05514705882353e-06,
+      "loss": 3.878,
+      "step": 564
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.053308823529412e-06,
+      "loss": 3.9476,
+      "step": 565
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.0514705882352944e-06,
+      "loss": 3.8901,
+      "step": 566
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.049632352941177e-06,
+      "loss": 3.927,
+      "step": 567
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.047794117647059e-06,
+      "loss": 3.8822,
+      "step": 568
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.045955882352941e-06,
+      "loss": 3.938,
+      "step": 569
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.044117647058824e-06,
+      "loss": 3.8705,
+      "step": 570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.042279411764707e-06,
+      "loss": 3.8969,
+      "step": 571
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.040441176470588e-06,
+      "loss": 3.8945,
+      "step": 572
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.038602941176471e-06,
+      "loss": 3.9055,
+      "step": 573
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.036764705882354e-06,
+      "loss": 3.8952,
+      "step": 574
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.034926470588235e-06,
+      "loss": 3.9069,
+      "step": 575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.033088235294118e-06,
+      "loss": 3.9034,
+      "step": 576
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.031250000000001e-06,
+      "loss": 3.8944,
+      "step": 577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.029411764705883e-06,
+      "loss": 3.8567,
+      "step": 578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.027573529411765e-06,
+      "loss": 3.883,
+      "step": 579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.0257352941176476e-06,
+      "loss": 3.8884,
+      "step": 580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.02389705882353e-06,
+      "loss": 3.9006,
+      "step": 581
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.022058823529412e-06,
+      "loss": 3.8631,
+      "step": 582
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.0202205882352945e-06,
+      "loss": 3.9281,
+      "step": 583
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.018382352941177e-06,
+      "loss": 3.9136,
+      "step": 584
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.016544117647059e-06,
+      "loss": 3.8971,
+      "step": 585
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.0147058823529415e-06,
+      "loss": 3.8815,
+      "step": 586
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.012867647058824e-06,
+      "loss": 3.8956,
+      "step": 587
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.011029411764707e-06,
+      "loss": 3.9125,
+      "step": 588
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0091911764705885e-06,
+      "loss": 3.9272,
+      "step": 589
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.007352941176471e-06,
+      "loss": 3.9132,
+      "step": 590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.005514705882354e-06,
+      "loss": 3.92,
+      "step": 591
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0036764705882355e-06,
+      "loss": 3.8869,
+      "step": 592
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001838235294118e-06,
+      "loss": 3.9242,
+      "step": 593
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.9256,
+      "step": 594
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.998161764705883e-06,
+      "loss": 3.9462,
+      "step": 595
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.996323529411765e-06,
+      "loss": 3.8908,
+      "step": 596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.994485294117648e-06,
+      "loss": 3.9261,
+      "step": 597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.99264705882353e-06,
+      "loss": 3.9065,
+      "step": 598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.990808823529412e-06,
+      "loss": 3.8781,
+      "step": 599
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.988970588235295e-06,
+      "loss": 3.8783,
+      "step": 600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.987132352941177e-06,
+      "loss": 3.8928,
+      "step": 601
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.98529411764706e-06,
+      "loss": 3.8868,
+      "step": 602
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.983455882352942e-06,
+      "loss": 3.8967,
+      "step": 603
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.981617647058823e-06,
+      "loss": 3.851,
+      "step": 604
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.979779411764706e-06,
+      "loss": 3.8998,
+      "step": 605
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9779411764705886e-06,
+      "loss": 3.9114,
+      "step": 606
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.97610294117647e-06,
+      "loss": 3.8974,
+      "step": 607
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.974264705882353e-06,
+      "loss": 3.914,
+      "step": 608
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9724264705882355e-06,
+      "loss": 3.9001,
+      "step": 609
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.970588235294118e-06,
+      "loss": 3.8892,
+      "step": 610
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.96875e-06,
+      "loss": 3.9047,
+      "step": 611
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9669117647058825e-06,
+      "loss": 3.8536,
+      "step": 612
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.965073529411765e-06,
+      "loss": 3.9175,
+      "step": 613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.963235294117647e-06,
+      "loss": 3.8781,
+      "step": 614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.9613970588235295e-06,
+      "loss": 3.8895,
+      "step": 615
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.959558823529412e-06,
+      "loss": 3.8261,
+      "step": 616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.957720588235294e-06,
+      "loss": 3.887,
+      "step": 617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.9558823529411765e-06,
+      "loss": 3.8918,
+      "step": 618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.954044117647059e-06,
+      "loss": 3.9152,
+      "step": 619
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.952205882352942e-06,
+      "loss": 3.8806,
+      "step": 620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.9503676470588234e-06,
+      "loss": 3.896,
+      "step": 621
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.948529411764706e-06,
+      "loss": 3.866,
+      "step": 622
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.946691176470589e-06,
+      "loss": 3.8914,
+      "step": 623
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.94485294117647e-06,
+      "loss": 3.9126,
+      "step": 624
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.943014705882353e-06,
+      "loss": 3.9088,
+      "step": 625
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.941176470588236e-06,
+      "loss": 3.8809,
+      "step": 626
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.939338235294118e-06,
+      "loss": 3.9812,
+      "step": 627
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.9375e-06,
+      "loss": 3.8903,
+      "step": 628
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.935661764705883e-06,
+      "loss": 3.8881,
+      "step": 629
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.933823529411765e-06,
+      "loss": 3.8948,
+      "step": 630
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.931985294117647e-06,
+      "loss": 3.9183,
+      "step": 631
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.9301470588235296e-06,
+      "loss": 3.8933,
+      "step": 632
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.928308823529412e-06,
+      "loss": 3.8459,
+      "step": 633
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.926470588235295e-06,
+      "loss": 3.8873,
+      "step": 634
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9246323529411766e-06,
+      "loss": 3.9084,
+      "step": 635
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.922794117647059e-06,
+      "loss": 3.9305,
+      "step": 636
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.920955882352942e-06,
+      "loss": 3.8834,
+      "step": 637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9191176470588235e-06,
+      "loss": 3.9081,
+      "step": 638
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.917279411764706e-06,
+      "loss": 3.8891,
+      "step": 639
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.915441176470589e-06,
+      "loss": 3.9156,
+      "step": 640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9136029411764705e-06,
+      "loss": 3.8956,
+      "step": 641
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.911764705882353e-06,
+      "loss": 3.9062,
+      "step": 642
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.909926470588236e-06,
+      "loss": 3.852,
+      "step": 643
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.908088235294118e-06,
+      "loss": 3.8811,
+      "step": 644
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.90625e-06,
+      "loss": 3.8982,
+      "step": 645
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.904411764705883e-06,
+      "loss": 3.8916,
+      "step": 646
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.902573529411765e-06,
+      "loss": 3.8889,
+      "step": 647
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.900735294117647e-06,
+      "loss": 3.8836,
+      "step": 648
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.89889705882353e-06,
+      "loss": 3.8858,
+      "step": 649
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.897058823529412e-06,
+      "loss": 3.9226,
+      "step": 650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.895220588235295e-06,
+      "loss": 3.88,
+      "step": 651
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.893382352941177e-06,
+      "loss": 3.9019,
+      "step": 652
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.891544117647059e-06,
+      "loss": 3.9161,
+      "step": 653
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.889705882352942e-06,
+      "loss": 3.8321,
+      "step": 654
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.887867647058824e-06,
+      "loss": 3.8858,
+      "step": 655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.886029411764706e-06,
+      "loss": 3.8875,
+      "step": 656
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.884191176470589e-06,
+      "loss": 3.9128,
+      "step": 657
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.882352941176471e-06,
+      "loss": 3.9009,
+      "step": 658
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.880514705882353e-06,
+      "loss": 3.8546,
+      "step": 659
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.878676470588236e-06,
+      "loss": 3.9097,
+      "step": 660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.876838235294118e-06,
+      "loss": 3.8824,
+      "step": 661
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.875e-06,
+      "loss": 3.8953,
+      "step": 662
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.873161764705883e-06,
+      "loss": 3.8983,
+      "step": 663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.871323529411765e-06,
+      "loss": 3.9264,
+      "step": 664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.869485294117647e-06,
+      "loss": 3.8979,
+      "step": 665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.86764705882353e-06,
+      "loss": 3.8801,
+      "step": 666
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.865808823529412e-06,
+      "loss": 3.8174,
+      "step": 667
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.863970588235295e-06,
+      "loss": 3.8782,
+      "step": 668
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.862132352941177e-06,
+      "loss": 3.8441,
+      "step": 669
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.860294117647059e-06,
+      "loss": 3.8826,
+      "step": 670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.858455882352942e-06,
+      "loss": 3.9378,
+      "step": 671
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.856617647058824e-06,
+      "loss": 3.8864,
+      "step": 672
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.854779411764706e-06,
+      "loss": 3.874,
+      "step": 673
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.852941176470589e-06,
+      "loss": 3.9054,
+      "step": 674
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8511029411764715e-06,
+      "loss": 3.8911,
+      "step": 675
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.849264705882353e-06,
+      "loss": 3.9021,
+      "step": 676
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.847426470588236e-06,
+      "loss": 3.9301,
+      "step": 677
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8455882352941185e-06,
+      "loss": 3.9065,
+      "step": 678
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.84375e-06,
+      "loss": 3.8838,
+      "step": 679
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.841911764705883e-06,
+      "loss": 3.9251,
+      "step": 680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8400735294117655e-06,
+      "loss": 3.8842,
+      "step": 681
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.838235294117647e-06,
+      "loss": 3.8571,
+      "step": 682
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.83639705882353e-06,
+      "loss": 3.8722,
+      "step": 683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8345588235294124e-06,
+      "loss": 3.8869,
+      "step": 684
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.832720588235295e-06,
+      "loss": 3.8968,
+      "step": 685
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.830882352941177e-06,
+      "loss": 3.9054,
+      "step": 686
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.829044117647059e-06,
+      "loss": 3.8092,
+      "step": 687
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.827205882352941e-06,
+      "loss": 3.8676,
+      "step": 688
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.825367647058824e-06,
+      "loss": 3.8615,
+      "step": 689
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8235294117647055e-06,
+      "loss": 3.8607,
+      "step": 690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.821691176470588e-06,
+      "loss": 3.8984,
+      "step": 691
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.819852941176471e-06,
+      "loss": 3.9044,
+      "step": 692
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.818014705882353e-06,
+      "loss": 3.9116,
+      "step": 693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.816176470588235e-06,
+      "loss": 3.8948,
+      "step": 694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8143382352941177e-06,
+      "loss": 3.888,
+      "step": 695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8125e-06,
+      "loss": 3.908,
+      "step": 696
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8106617647058825e-06,
+      "loss": 3.9121,
+      "step": 697
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8088235294117647e-06,
+      "loss": 3.883,
+      "step": 698
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8069852941176473e-06,
+      "loss": 3.918,
+      "step": 699
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8051470588235295e-06,
+      "loss": 3.8864,
+      "step": 700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8033088235294117e-06,
+      "loss": 3.8558,
+      "step": 701
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8014705882352943e-06,
+      "loss": 3.909,
+      "step": 702
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7996323529411765e-06,
+      "loss": 3.8228,
+      "step": 703
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.797794117647059e-06,
+      "loss": 3.9212,
+      "step": 704
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7959558823529413e-06,
+      "loss": 3.893,
+      "step": 705
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.794117647058824e-06,
+      "loss": 3.8535,
+      "step": 706
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.792279411764706e-06,
+      "loss": 3.8973,
+      "step": 707
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7904411764705882e-06,
+      "loss": 3.8784,
+      "step": 708
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.788602941176471e-06,
+      "loss": 3.8789,
+      "step": 709
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.786764705882353e-06,
+      "loss": 3.907,
+      "step": 710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7849264705882356e-06,
+      "loss": 3.9111,
+      "step": 711
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.783088235294118e-06,
+      "loss": 3.8925,
+      "step": 712
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.78125e-06,
+      "loss": 3.9401,
+      "step": 713
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7794117647058826e-06,
+      "loss": 3.8688,
+      "step": 714
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.777573529411765e-06,
+      "loss": 3.8858,
+      "step": 715
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7757352941176474e-06,
+      "loss": 3.9156,
+      "step": 716
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7738970588235296e-06,
+      "loss": 3.9049,
+      "step": 717
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.772058823529412e-06,
+      "loss": 3.883,
+      "step": 718
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7702205882352944e-06,
+      "loss": 3.8564,
+      "step": 719
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7683823529411766e-06,
+      "loss": 3.8927,
+      "step": 720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.766544117647059e-06,
+      "loss": 3.8769,
+      "step": 721
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7647058823529414e-06,
+      "loss": 3.9029,
+      "step": 722
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.762867647058824e-06,
+      "loss": 3.9064,
+      "step": 723
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.761029411764706e-06,
+      "loss": 3.9405,
+      "step": 724
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7591911764705883e-06,
+      "loss": 3.8377,
+      "step": 725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.757352941176471e-06,
+      "loss": 3.9144,
+      "step": 726
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.755514705882353e-06,
+      "loss": 3.9066,
+      "step": 727
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7536764705882357e-06,
+      "loss": 3.857,
+      "step": 728
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.751838235294118e-06,
+      "loss": 3.8759,
+      "step": 729
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 3.8411,
+      "step": 730
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7481617647058827e-06,
+      "loss": 3.9467,
+      "step": 731
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.746323529411765e-06,
+      "loss": 3.8648,
+      "step": 732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7444852941176475e-06,
+      "loss": 3.8636,
+      "step": 733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7426470588235297e-06,
+      "loss": 3.8303,
+      "step": 734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7408088235294123e-06,
+      "loss": 3.9108,
+      "step": 735
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7389705882352945e-06,
+      "loss": 3.903,
+      "step": 736
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7371323529411767e-06,
+      "loss": 3.8517,
+      "step": 737
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7352941176470593e-06,
+      "loss": 3.9169,
+      "step": 738
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7334558823529414e-06,
+      "loss": 3.8399,
+      "step": 739
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.731617647058824e-06,
+      "loss": 3.9108,
+      "step": 740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7297794117647062e-06,
+      "loss": 3.8493,
+      "step": 741
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7279411764705884e-06,
+      "loss": 3.9021,
+      "step": 742
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.726102941176471e-06,
+      "loss": 3.9044,
+      "step": 743
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.724264705882353e-06,
+      "loss": 3.9299,
+      "step": 744
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.722426470588236e-06,
+      "loss": 3.9097,
+      "step": 745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.720588235294118e-06,
+      "loss": 3.9244,
+      "step": 746
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7187500000000006e-06,
+      "loss": 3.9088,
+      "step": 747
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.716911764705883e-06,
+      "loss": 3.8867,
+      "step": 748
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.715073529411765e-06,
+      "loss": 3.9152,
+      "step": 749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7132352941176476e-06,
+      "loss": 3.912,
+      "step": 750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7113970588235298e-06,
+      "loss": 3.8801,
+      "step": 751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7095588235294124e-06,
+      "loss": 3.8928,
+      "step": 752
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7077205882352946e-06,
+      "loss": 3.9624,
+      "step": 753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7058823529411767e-06,
+      "loss": 3.9096,
+      "step": 754
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.7040441176470593e-06,
+      "loss": 3.8554,
+      "step": 755
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.7022058823529415e-06,
+      "loss": 3.9021,
+      "step": 756
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.700367647058824e-06,
+      "loss": 3.7969,
+      "step": 757
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6985294117647063e-06,
+      "loss": 3.796,
+      "step": 758
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.696691176470589e-06,
+      "loss": 3.9867,
+      "step": 759
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.694852941176471e-06,
+      "loss": 3.8547,
+      "step": 760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6930147058823533e-06,
+      "loss": 3.9183,
+      "step": 761
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.691176470588236e-06,
+      "loss": 3.9144,
+      "step": 762
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.689338235294118e-06,
+      "loss": 3.8977,
+      "step": 763
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6875000000000007e-06,
+      "loss": 3.8987,
+      "step": 764
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.685661764705883e-06,
+      "loss": 3.8994,
+      "step": 765
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.683823529411765e-06,
+      "loss": 3.8752,
+      "step": 766
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6819852941176477e-06,
+      "loss": 3.8946,
+      "step": 767
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.68014705882353e-06,
+      "loss": 3.8058,
+      "step": 768
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6783088235294125e-06,
+      "loss": 3.7745,
+      "step": 769
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6764705882352946e-06,
+      "loss": 3.9245,
+      "step": 770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6746323529411772e-06,
+      "loss": 3.8975,
+      "step": 771
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6727941176470594e-06,
+      "loss": 3.9134,
+      "step": 772
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.670955882352941e-06,
+      "loss": 3.8673,
+      "step": 773
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6691176470588234e-06,
+      "loss": 3.8821,
+      "step": 774
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.667279411764706e-06,
+      "loss": 3.9071,
+      "step": 775
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.665441176470588e-06,
+      "loss": 3.919,
+      "step": 776
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6636029411764708e-06,
+      "loss": 3.8902,
+      "step": 777
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.661764705882353e-06,
+      "loss": 3.9026,
+      "step": 778
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6599264705882356e-06,
+      "loss": 3.8472,
+      "step": 779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6580882352941178e-06,
+      "loss": 3.9874,
+      "step": 780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.65625e-06,
+      "loss": 3.8773,
+      "step": 781
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6544117647058825e-06,
+      "loss": 3.7648,
+      "step": 782
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6525735294117647e-06,
+      "loss": 3.8999,
+      "step": 783
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6507352941176473e-06,
+      "loss": 3.9231,
+      "step": 784
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6488970588235295e-06,
+      "loss": 3.8574,
+      "step": 785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6470588235294117e-06,
+      "loss": 3.8892,
+      "step": 786
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6452205882352943e-06,
+      "loss": 3.8979,
+      "step": 787
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6433823529411765e-06,
+      "loss": 3.8968,
+      "step": 788
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.641544117647059e-06,
+      "loss": 3.8651,
+      "step": 789
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6397058823529413e-06,
+      "loss": 3.8567,
+      "step": 790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6378676470588235e-06,
+      "loss": 3.9199,
+      "step": 791
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.636029411764706e-06,
+      "loss": 3.8299,
+      "step": 792
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6341911764705883e-06,
+      "loss": 3.9038,
+      "step": 793
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.632352941176471e-06,
+      "loss": 3.9121,
+      "step": 794
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.630514705882353e-06,
+      "loss": 3.8915,
+      "step": 795
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.6286764705882357e-06,
+      "loss": 3.8963,
+      "step": 796
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.626838235294118e-06,
+      "loss": 3.7507,
+      "step": 797
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.625e-06,
+      "loss": 3.8768,
+      "step": 798
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.6231617647058826e-06,
+      "loss": 3.9103,
+      "step": 799
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.621323529411765e-06,
+      "loss": 3.9213,
+      "step": 800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6194852941176474e-06,
+      "loss": 3.8851,
+      "step": 801
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6176470588235296e-06,
+      "loss": 3.9026,
+      "step": 802
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6158088235294118e-06,
+      "loss": 3.8877,
+      "step": 803
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6139705882352944e-06,
+      "loss": 3.8842,
+      "step": 804
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6121323529411766e-06,
+      "loss": 3.8703,
+      "step": 805
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.610294117647059e-06,
+      "loss": 3.8805,
+      "step": 806
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6084558823529414e-06,
+      "loss": 3.8648,
+      "step": 807
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.606617647058824e-06,
+      "loss": 3.8896,
+      "step": 808
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.604779411764706e-06,
+      "loss": 3.8869,
+      "step": 809
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6029411764705883e-06,
+      "loss": 3.8499,
+      "step": 810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.601102941176471e-06,
+      "loss": 3.8922,
+      "step": 811
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.599264705882353e-06,
+      "loss": 3.8819,
+      "step": 812
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5974264705882357e-06,
+      "loss": 3.8917,
+      "step": 813
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.595588235294118e-06,
+      "loss": 3.8994,
+      "step": 814
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.59375e-06,
+      "loss": 3.8913,
+      "step": 815
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5919117647058827e-06,
+      "loss": 3.9107,
+      "step": 816
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.590073529411765e-06,
+      "loss": 3.9041,
+      "step": 817
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5882352941176475e-06,
+      "loss": 3.8944,
+      "step": 818
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5863970588235297e-06,
+      "loss": 3.8676,
+      "step": 819
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5845588235294123e-06,
+      "loss": 3.9816,
+      "step": 820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5827205882352945e-06,
+      "loss": 3.8613,
+      "step": 821
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5808823529411767e-06,
+      "loss": 3.903,
+      "step": 822
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5790441176470593e-06,
+      "loss": 3.9405,
+      "step": 823
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5772058823529415e-06,
+      "loss": 3.8746,
+      "step": 824
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.575367647058824e-06,
+      "loss": 3.9041,
+      "step": 825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5735294117647062e-06,
+      "loss": 3.8438,
+      "step": 826
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5716911764705884e-06,
+      "loss": 3.8945,
+      "step": 827
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.569852941176471e-06,
+      "loss": 3.8888,
+      "step": 828
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5680147058823532e-06,
+      "loss": 3.8727,
+      "step": 829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.566176470588236e-06,
+      "loss": 3.7984,
+      "step": 830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.564338235294118e-06,
+      "loss": 3.8543,
+      "step": 831
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.07567567567567568,
+      "eval_loss": 3.8715946674346924,
+      "eval_runtime": 142.6452,
+      "eval_samples_per_second": 2.594,
+      "eval_steps_per_second": 0.652,
+      "step": 831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5625e-06,
+      "loss": 3.8845,
+      "step": 832
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.560661764705883e-06,
+      "loss": 3.8578,
+      "step": 833
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.558823529411765e-06,
+      "loss": 3.8816,
+      "step": 834
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5569852941176476e-06,
+      "loss": 3.8997,
+      "step": 835
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5551470588235298e-06,
+      "loss": 3.8693,
+      "step": 836
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5533088235294124e-06,
+      "loss": 3.8912,
+      "step": 837
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.5514705882352946e-06,
+      "loss": 3.8171,
+      "step": 838
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.5496323529411768e-06,
+      "loss": 3.9086,
+      "step": 839
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.5477941176470594e-06,
+      "loss": 3.906,
+      "step": 840
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.5459558823529415e-06,
+      "loss": 3.8868,
+      "step": 841
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.544117647058824e-06,
+      "loss": 3.8806,
+      "step": 842
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.5422794117647063e-06,
+      "loss": 3.8998,
+      "step": 843
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5404411764705885e-06,
+      "loss": 3.8707,
+      "step": 844
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.538602941176471e-06,
+      "loss": 3.8786,
+      "step": 845
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.5367647058823533e-06,
+      "loss": 3.8617,
+      "step": 846
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.534926470588236e-06,
+      "loss": 3.8815,
+      "step": 847
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.533088235294118e-06,
+      "loss": 3.917,
+      "step": 848
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.5312500000000007e-06,
+      "loss": 3.8588,
+      "step": 849
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 3.8235,
+      "step": 850
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.527573529411765e-06,
+      "loss": 3.8774,
+      "step": 851
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.5257352941176477e-06,
+      "loss": 3.9158,
+      "step": 852
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.52389705882353e-06,
+      "loss": 3.8705,
+      "step": 853
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.5220588235294125e-06,
+      "loss": 3.8952,
+      "step": 854
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.5202205882352947e-06,
+      "loss": 3.8585,
+      "step": 855
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.518382352941177e-06,
+      "loss": 3.9241,
+      "step": 856
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.5165441176470594e-06,
+      "loss": 3.8826,
+      "step": 857
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.514705882352941e-06,
+      "loss": 3.8563,
+      "step": 858
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.5128676470588234e-06,
+      "loss": 3.8963,
+      "step": 859
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.511029411764706e-06,
+      "loss": 3.8512,
+      "step": 860
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.509191176470588e-06,
+      "loss": 3.8837,
+      "step": 861
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.507352941176471e-06,
+      "loss": 3.8747,
+      "step": 862
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.505514705882353e-06,
+      "loss": 3.8878,
+      "step": 863
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.503676470588235e-06,
+      "loss": 3.8969,
+      "step": 864
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.5018382352941178e-06,
+      "loss": 3.8841,
+      "step": 865
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.5e-06,
+      "loss": 3.8761,
+      "step": 866
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4981617647058826e-06,
+      "loss": 3.8778,
+      "step": 867
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4963235294117647e-06,
+      "loss": 3.8743,
+      "step": 868
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4944852941176473e-06,
+      "loss": 3.9009,
+      "step": 869
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4926470588235295e-06,
+      "loss": 3.904,
+      "step": 870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4908088235294117e-06,
+      "loss": 3.8567,
+      "step": 871
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4889705882352943e-06,
+      "loss": 3.9257,
+      "step": 872
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4871323529411765e-06,
+      "loss": 3.8183,
+      "step": 873
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.485294117647059e-06,
+      "loss": 3.8525,
+      "step": 874
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4834558823529413e-06,
+      "loss": 3.7764,
+      "step": 875
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4816176470588235e-06,
+      "loss": 3.8376,
+      "step": 876
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.479779411764706e-06,
+      "loss": 3.8778,
+      "step": 877
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4779411764705883e-06,
+      "loss": 3.8971,
+      "step": 878
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.476102941176471e-06,
+      "loss": 3.9272,
+      "step": 879
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.474264705882353e-06,
+      "loss": 3.9807,
+      "step": 880
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4724264705882352e-06,
+      "loss": 3.8619,
+      "step": 881
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.470588235294118e-06,
+      "loss": 3.926,
+      "step": 882
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.46875e-06,
+      "loss": 3.907,
+      "step": 883
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4669117647058826e-06,
+      "loss": 3.786,
+      "step": 884
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.465073529411765e-06,
+      "loss": 3.8521,
+      "step": 885
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4632352941176474e-06,
+      "loss": 3.8393,
+      "step": 886
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4613970588235296e-06,
+      "loss": 3.9026,
+      "step": 887
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.459558823529412e-06,
+      "loss": 3.9556,
+      "step": 888
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4577205882352944e-06,
+      "loss": 3.8306,
+      "step": 889
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4558823529411766e-06,
+      "loss": 3.8948,
+      "step": 890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.454044117647059e-06,
+      "loss": 3.8971,
+      "step": 891
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.4522058823529414e-06,
+      "loss": 3.9087,
+      "step": 892
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.4503676470588236e-06,
+      "loss": 3.837,
+      "step": 893
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.448529411764706e-06,
+      "loss": 3.8944,
+      "step": 894
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.4466911764705884e-06,
+      "loss": 3.8641,
+      "step": 895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.444852941176471e-06,
+      "loss": 3.8963,
+      "step": 896
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.443014705882353e-06,
+      "loss": 3.8441,
+      "step": 897
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.4411764705882358e-06,
+      "loss": 3.874,
+      "step": 898
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.439338235294118e-06,
+      "loss": 3.9138,
+      "step": 899
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.4375e-06,
+      "loss": 3.7848,
+      "step": 900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.4356617647058827e-06,
+      "loss": 3.8444,
+      "step": 901
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.433823529411765e-06,
+      "loss": 3.9121,
+      "step": 902
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.4319852941176475e-06,
+      "loss": 3.845,
+      "step": 903
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.4301470588235297e-06,
+      "loss": 3.8382,
+      "step": 904
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.428308823529412e-06,
+      "loss": 3.8495,
+      "step": 905
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.4264705882352945e-06,
+      "loss": 3.9002,
+      "step": 906
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.4246323529411767e-06,
+      "loss": 3.8547,
+      "step": 907
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.4227941176470593e-06,
+      "loss": 3.8445,
+      "step": 908
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.4209558823529415e-06,
+      "loss": 3.7814,
+      "step": 909
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.419117647058824e-06,
+      "loss": 3.88,
+      "step": 910
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.4172794117647063e-06,
+      "loss": 3.8275,
+      "step": 911
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.4154411764705884e-06,
+      "loss": 3.8685,
+      "step": 912
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.413602941176471e-06,
+      "loss": 3.8847,
+      "step": 913
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.4117647058823532e-06,
+      "loss": 3.8547,
+      "step": 914
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.409926470588236e-06,
+      "loss": 3.79,
+      "step": 915
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.408088235294118e-06,
+      "loss": 3.8259,
+      "step": 916
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.40625e-06,
+      "loss": 3.9019,
+      "step": 917
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.404411764705883e-06,
+      "loss": 3.8579,
+      "step": 918
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.402573529411765e-06,
+      "loss": 3.9169,
+      "step": 919
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.4007352941176476e-06,
+      "loss": 3.8456,
+      "step": 920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.39889705882353e-06,
+      "loss": 3.892,
+      "step": 921
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.397058823529412e-06,
+      "loss": 3.8692,
+      "step": 922
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3952205882352946e-06,
+      "loss": 3.8589,
+      "step": 923
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3933823529411768e-06,
+      "loss": 3.9266,
+      "step": 924
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3915441176470594e-06,
+      "loss": 3.8824,
+      "step": 925
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3897058823529416e-06,
+      "loss": 3.8426,
+      "step": 926
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.387867647058824e-06,
+      "loss": 3.8007,
+      "step": 927
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3860294117647063e-06,
+      "loss": 3.8145,
+      "step": 928
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3841911764705885e-06,
+      "loss": 3.8853,
+      "step": 929
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.382352941176471e-06,
+      "loss": 3.8787,
+      "step": 930
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3805147058823533e-06,
+      "loss": 3.8604,
+      "step": 931
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.378676470588236e-06,
+      "loss": 3.8188,
+      "step": 932
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.376838235294118e-06,
+      "loss": 3.8472,
+      "step": 933
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3750000000000003e-06,
+      "loss": 3.749,
+      "step": 934
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.373161764705883e-06,
+      "loss": 3.9151,
+      "step": 935
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.371323529411765e-06,
+      "loss": 3.9068,
+      "step": 936
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3694852941176477e-06,
+      "loss": 3.9025,
+      "step": 937
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.36764705882353e-06,
+      "loss": 3.8105,
+      "step": 938
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3658088235294125e-06,
+      "loss": 3.8521,
+      "step": 939
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3639705882352947e-06,
+      "loss": 3.9107,
+      "step": 940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.362132352941177e-06,
+      "loss": 3.8213,
+      "step": 941
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3602941176470595e-06,
+      "loss": 3.8693,
+      "step": 942
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3584558823529412e-06,
+      "loss": 3.9231,
+      "step": 943
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3566176470588234e-06,
+      "loss": 3.9613,
+      "step": 944
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.354779411764706e-06,
+      "loss": 3.8031,
+      "step": 945
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.352941176470588e-06,
+      "loss": 3.9291,
+      "step": 946
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.351102941176471e-06,
+      "loss": 3.9175,
+      "step": 947
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.349264705882353e-06,
+      "loss": 3.9518,
+      "step": 948
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.347426470588235e-06,
+      "loss": 3.8479,
+      "step": 949
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.3455882352941178e-06,
+      "loss": 3.8762,
+      "step": 950
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.34375e-06,
+      "loss": 3.8534,
+      "step": 951
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.3419117647058826e-06,
+      "loss": 3.8771,
+      "step": 952
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.3400735294117648e-06,
+      "loss": 3.8623,
+      "step": 953
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.338235294117647e-06,
+      "loss": 3.8889,
+      "step": 954
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.3363970588235295e-06,
+      "loss": 3.8368,
+      "step": 955
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.3345588235294117e-06,
+      "loss": 3.884,
+      "step": 956
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.3327205882352943e-06,
+      "loss": 3.8734,
+      "step": 957
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.3308823529411765e-06,
+      "loss": 3.7955,
+      "step": 958
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.329044117647059e-06,
+      "loss": 3.8188,
+      "step": 959
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.3272058823529413e-06,
+      "loss": 3.849,
+      "step": 960
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.3253676470588235e-06,
+      "loss": 3.863,
+      "step": 961
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.323529411764706e-06,
+      "loss": 3.9277,
+      "step": 962
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.3216911764705883e-06,
+      "loss": 3.793,
+      "step": 963
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.319852941176471e-06,
+      "loss": 3.8677,
+      "step": 964
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.318014705882353e-06,
+      "loss": 3.8413,
+      "step": 965
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.3161764705882353e-06,
+      "loss": 3.8306,
+      "step": 966
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.314338235294118e-06,
+      "loss": 3.8914,
+      "step": 967
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.3125e-06,
+      "loss": 3.8469,
+      "step": 968
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.3106617647058827e-06,
+      "loss": 3.84,
+      "step": 969
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.308823529411765e-06,
+      "loss": 3.8828,
+      "step": 970
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.3069852941176474e-06,
+      "loss": 3.9993,
+      "step": 971
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.3051470588235296e-06,
+      "loss": 3.915,
+      "step": 972
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.303308823529412e-06,
+      "loss": 3.7709,
+      "step": 973
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.3014705882352944e-06,
+      "loss": 3.8683,
+      "step": 974
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2996323529411766e-06,
+      "loss": 3.9375,
+      "step": 975
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2977941176470592e-06,
+      "loss": 3.9044,
+      "step": 976
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2959558823529414e-06,
+      "loss": 3.9021,
+      "step": 977
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2941176470588236e-06,
+      "loss": 3.8651,
+      "step": 978
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.292279411764706e-06,
+      "loss": 3.8722,
+      "step": 979
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2904411764705884e-06,
+      "loss": 3.8855,
+      "step": 980
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.288602941176471e-06,
+      "loss": 3.861,
+      "step": 981
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.286764705882353e-06,
+      "loss": 3.8801,
+      "step": 982
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2849264705882353e-06,
+      "loss": 3.8604,
+      "step": 983
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.283088235294118e-06,
+      "loss": 3.842,
+      "step": 984
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.28125e-06,
+      "loss": 3.8221,
+      "step": 985
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2794117647058827e-06,
+      "loss": 3.8859,
+      "step": 986
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.277573529411765e-06,
+      "loss": 3.8611,
+      "step": 987
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2757352941176475e-06,
+      "loss": 3.8207,
+      "step": 988
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2738970588235297e-06,
+      "loss": 3.775,
+      "step": 989
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.272058823529412e-06,
+      "loss": 3.858,
+      "step": 990
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2702205882352945e-06,
+      "loss": 3.8059,
+      "step": 991
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2683823529411767e-06,
+      "loss": 3.8658,
+      "step": 992
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2665441176470593e-06,
+      "loss": 3.8327,
+      "step": 993
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2647058823529415e-06,
+      "loss": 3.897,
+      "step": 994
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2628676470588237e-06,
+      "loss": 3.8465,
+      "step": 995
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2610294117647063e-06,
+      "loss": 3.912,
+      "step": 996
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2591911764705885e-06,
+      "loss": 3.8713,
+      "step": 997
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.257352941176471e-06,
+      "loss": 3.8196,
+      "step": 998
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2555147058823532e-06,
+      "loss": 3.7875,
+      "step": 999
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.253676470588236e-06,
+      "loss": 3.8626,
+      "step": 1000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.251838235294118e-06,
+      "loss": 3.8419,
+      "step": 1001
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 3.7857,
+      "step": 1002
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.248161764705883e-06,
+      "loss": 3.8541,
+      "step": 1003
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.246323529411765e-06,
+      "loss": 3.8291,
+      "step": 1004
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.2444852941176476e-06,
+      "loss": 3.7617,
+      "step": 1005
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.24264705882353e-06,
+      "loss": 3.9104,
+      "step": 1006
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.240808823529412e-06,
+      "loss": 3.8978,
+      "step": 1007
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2389705882352946e-06,
+      "loss": 3.827,
+      "step": 1008
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2371323529411768e-06,
+      "loss": 3.8974,
+      "step": 1009
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.2352941176470594e-06,
+      "loss": 3.8966,
+      "step": 1010
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.2334558823529416e-06,
+      "loss": 3.8008,
+      "step": 1011
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.231617647058824e-06,
+      "loss": 3.8636,
+      "step": 1012
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.2297794117647064e-06,
+      "loss": 3.8003,
+      "step": 1013
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.2279411764705885e-06,
+      "loss": 3.8171,
+      "step": 1014
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.226102941176471e-06,
+      "loss": 3.8971,
+      "step": 1015
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.2242647058823533e-06,
+      "loss": 3.8496,
+      "step": 1016
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.222426470588236e-06,
+      "loss": 3.8715,
+      "step": 1017
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.220588235294118e-06,
+      "loss": 3.9174,
+      "step": 1018
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2187500000000003e-06,
+      "loss": 3.9274,
+      "step": 1019
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.216911764705883e-06,
+      "loss": 3.9359,
+      "step": 1020
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.215073529411765e-06,
+      "loss": 3.8354,
+      "step": 1021
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.2132352941176477e-06,
+      "loss": 3.9904,
+      "step": 1022
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.21139705882353e-06,
+      "loss": 3.8231,
+      "step": 1023
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.209558823529412e-06,
+      "loss": 3.8287,
+      "step": 1024
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.2077205882352947e-06,
+      "loss": 3.8522,
+      "step": 1025
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.205882352941177e-06,
+      "loss": 3.8832,
+      "step": 1026
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2040441176470595e-06,
+      "loss": 3.8262,
+      "step": 1027
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2022058823529412e-06,
+      "loss": 3.8892,
+      "step": 1028
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2003676470588234e-06,
+      "loss": 3.8721,
+      "step": 1029
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.198529411764706e-06,
+      "loss": 3.8922,
+      "step": 1030
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.196691176470588e-06,
+      "loss": 3.8586,
+      "step": 1031
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1948529411764704e-06,
+      "loss": 3.8284,
+      "step": 1032
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.193014705882353e-06,
+      "loss": 3.8622,
+      "step": 1033
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.191176470588235e-06,
+      "loss": 3.8174,
+      "step": 1034
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.189338235294118e-06,
+      "loss": 3.8811,
+      "step": 1035
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1875e-06,
+      "loss": 3.8267,
+      "step": 1036
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1856617647058826e-06,
+      "loss": 3.8518,
+      "step": 1037
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1838235294117648e-06,
+      "loss": 3.8098,
+      "step": 1038
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.181985294117647e-06,
+      "loss": 3.8925,
+      "step": 1039
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1801470588235296e-06,
+      "loss": 3.8589,
+      "step": 1040
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1783088235294117e-06,
+      "loss": 3.8627,
+      "step": 1041
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1764705882352943e-06,
+      "loss": 3.8578,
+      "step": 1042
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1746323529411765e-06,
+      "loss": 3.927,
+      "step": 1043
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1727941176470587e-06,
+      "loss": 3.8797,
+      "step": 1044
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1709558823529413e-06,
+      "loss": 3.775,
+      "step": 1045
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1691176470588235e-06,
+      "loss": 3.8828,
+      "step": 1046
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.167279411764706e-06,
+      "loss": 3.813,
+      "step": 1047
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1654411764705883e-06,
+      "loss": 3.9459,
+      "step": 1048
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.163602941176471e-06,
+      "loss": 3.8703,
+      "step": 1049
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.161764705882353e-06,
+      "loss": 3.8377,
+      "step": 1050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1599264705882353e-06,
+      "loss": 3.7724,
+      "step": 1051
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.158088235294118e-06,
+      "loss": 3.9384,
+      "step": 1052
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.15625e-06,
+      "loss": 3.9083,
+      "step": 1053
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1544117647058827e-06,
+      "loss": 3.858,
+      "step": 1054
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.152573529411765e-06,
+      "loss": 3.8304,
+      "step": 1055
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.150735294117647e-06,
+      "loss": 3.8262,
+      "step": 1056
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.1488970588235296e-06,
+      "loss": 3.7159,
+      "step": 1057
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.147058823529412e-06,
+      "loss": 3.8542,
+      "step": 1058
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.1452205882352944e-06,
+      "loss": 3.8638,
+      "step": 1059
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.1433823529411766e-06,
+      "loss": 3.9042,
+      "step": 1060
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.1415441176470592e-06,
+      "loss": 3.8701,
+      "step": 1061
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.1397058823529414e-06,
+      "loss": 3.7691,
+      "step": 1062
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.1378676470588236e-06,
+      "loss": 3.8731,
+      "step": 1063
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.136029411764706e-06,
+      "loss": 3.8722,
+      "step": 1064
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.1341911764705884e-06,
+      "loss": 3.8362,
+      "step": 1065
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.132352941176471e-06,
+      "loss": 3.9194,
+      "step": 1066
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.130514705882353e-06,
+      "loss": 3.8658,
+      "step": 1067
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.1286764705882354e-06,
+      "loss": 3.7822,
+      "step": 1068
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.126838235294118e-06,
+      "loss": 3.8979,
+      "step": 1069
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.125e-06,
+      "loss": 3.8638,
+      "step": 1070
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.1231617647058828e-06,
+      "loss": 3.8953,
+      "step": 1071
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.121323529411765e-06,
+      "loss": 3.8839,
+      "step": 1072
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.119485294117647e-06,
+      "loss": 3.9778,
+      "step": 1073
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.1176470588235297e-06,
+      "loss": 3.8903,
+      "step": 1074
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.115808823529412e-06,
+      "loss": 3.8579,
+      "step": 1075
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.1139705882352945e-06,
+      "loss": 3.7655,
+      "step": 1076
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.1121323529411767e-06,
+      "loss": 3.8569,
+      "step": 1077
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.1102941176470593e-06,
+      "loss": 3.7733,
+      "step": 1078
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1084558823529415e-06,
+      "loss": 3.8975,
+      "step": 1079
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1066176470588237e-06,
+      "loss": 3.9081,
+      "step": 1080
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1047794117647063e-06,
+      "loss": 3.8688,
+      "step": 1081
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.1029411764705885e-06,
+      "loss": 3.8985,
+      "step": 1082
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.101102941176471e-06,
+      "loss": 3.8815,
+      "step": 1083
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0992647058823533e-06,
+      "loss": 3.8982,
+      "step": 1084
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0974264705882354e-06,
+      "loss": 3.8199,
+      "step": 1085
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.095588235294118e-06,
+      "loss": 3.8913,
+      "step": 1086
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0937500000000002e-06,
+      "loss": 3.8682,
+      "step": 1087
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.091911764705883e-06,
+      "loss": 3.8735,
+      "step": 1088
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.090073529411765e-06,
+      "loss": 3.7854,
+      "step": 1089
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0882352941176476e-06,
+      "loss": 3.893,
+      "step": 1090
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.08639705882353e-06,
+      "loss": 3.8769,
+      "step": 1091
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.084558823529412e-06,
+      "loss": 3.9361,
+      "step": 1092
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0827205882352946e-06,
+      "loss": 3.7543,
+      "step": 1093
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.080882352941177e-06,
+      "loss": 3.8516,
+      "step": 1094
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0790441176470594e-06,
+      "loss": 3.9239,
+      "step": 1095
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0772058823529416e-06,
+      "loss": 3.8539,
+      "step": 1096
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0753676470588238e-06,
+      "loss": 3.8585,
+      "step": 1097
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0735294117647064e-06,
+      "loss": 3.9225,
+      "step": 1098
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0716911764705886e-06,
+      "loss": 3.8298,
+      "step": 1099
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.069852941176471e-06,
+      "loss": 3.841,
+      "step": 1100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0680147058823534e-06,
+      "loss": 3.8988,
+      "step": 1101
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.066176470588236e-06,
+      "loss": 3.8548,
+      "step": 1102
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.064338235294118e-06,
+      "loss": 3.9095,
+      "step": 1103
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0625000000000003e-06,
+      "loss": 3.8913,
+      "step": 1104
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.060661764705883e-06,
+      "loss": 3.9233,
+      "step": 1105
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.058823529411765e-06,
+      "loss": 3.8599,
+      "step": 1106
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0569852941176477e-06,
+      "loss": 3.8367,
+      "step": 1107
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.05514705882353e-06,
+      "loss": 3.7901,
+      "step": 1108
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.0945945945945946,
+      "eval_loss": 3.838702917098999,
+      "eval_runtime": 141.8389,
+      "eval_samples_per_second": 2.609,
+      "eval_steps_per_second": 0.656,
+      "step": 1108
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.053308823529412e-06,
+      "loss": 3.7794,
+      "step": 1109
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0514705882352947e-06,
+      "loss": 3.8637,
+      "step": 1110
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.049632352941177e-06,
+      "loss": 3.7863,
+      "step": 1111
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0477941176470595e-06,
+      "loss": 3.9245,
+      "step": 1112
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.0459558823529412e-06,
+      "loss": 3.8797,
+      "step": 1113
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.0441176470588234e-06,
+      "loss": 3.8391,
+      "step": 1114
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.042279411764706e-06,
+      "loss": 3.7441,
+      "step": 1115
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.0404411764705882e-06,
+      "loss": 3.8995,
+      "step": 1116
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.0386029411764704e-06,
+      "loss": 3.7882,
+      "step": 1117
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.036764705882353e-06,
+      "loss": 3.8309,
+      "step": 1118
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.034926470588235e-06,
+      "loss": 3.9182,
+      "step": 1119
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.033088235294118e-06,
+      "loss": 3.8266,
+      "step": 1120
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.03125e-06,
+      "loss": 3.833,
+      "step": 1121
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.0294117647058826e-06,
+      "loss": 3.8022,
+      "step": 1122
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.0275735294117648e-06,
+      "loss": 3.8981,
+      "step": 1123
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.025735294117647e-06,
+      "loss": 3.8455,
+      "step": 1124
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.0238970588235296e-06,
+      "loss": 3.9001,
+      "step": 1125
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.0220588235294118e-06,
+      "loss": 3.8753,
+      "step": 1126
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.0202205882352944e-06,
+      "loss": 3.8665,
+      "step": 1127
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.0183823529411765e-06,
+      "loss": 3.7868,
+      "step": 1128
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.0165441176470587e-06,
+      "loss": 3.8482,
+      "step": 1129
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.0147058823529413e-06,
+      "loss": 3.875,
+      "step": 1130
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.0128676470588235e-06,
+      "loss": 3.812,
+      "step": 1131
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.011029411764706e-06,
+      "loss": 3.926,
+      "step": 1132
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.0091911764705883e-06,
+      "loss": 3.805,
+      "step": 1133
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.0073529411764705e-06,
+      "loss": 3.86,
+      "step": 1134
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.005514705882353e-06,
+      "loss": 3.7903,
+      "step": 1135
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.0036764705882353e-06,
+      "loss": 3.8155,
+      "step": 1136
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.001838235294118e-06,
+      "loss": 3.8764,
+      "step": 1137
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3e-06,
+      "loss": 3.8134,
+      "step": 1138
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9981617647058827e-06,
+      "loss": 3.8877,
+      "step": 1139
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.996323529411765e-06,
+      "loss": 3.8153,
+      "step": 1140
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.994485294117647e-06,
+      "loss": 3.8429,
+      "step": 1141
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9926470588235297e-06,
+      "loss": 3.8923,
+      "step": 1142
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.990808823529412e-06,
+      "loss": 3.8527,
+      "step": 1143
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9889705882352944e-06,
+      "loss": 3.8033,
+      "step": 1144
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9871323529411766e-06,
+      "loss": 3.8288,
+      "step": 1145
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.985294117647059e-06,
+      "loss": 3.8506,
+      "step": 1146
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9834558823529414e-06,
+      "loss": 3.8286,
+      "step": 1147
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9816176470588236e-06,
+      "loss": 3.8395,
+      "step": 1148
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9797794117647062e-06,
+      "loss": 3.8023,
+      "step": 1149
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9779411764705884e-06,
+      "loss": 3.8523,
+      "step": 1150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.976102941176471e-06,
+      "loss": 3.7432,
+      "step": 1151
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.974264705882353e-06,
+      "loss": 3.8311,
+      "step": 1152
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9724264705882354e-06,
+      "loss": 3.8631,
+      "step": 1153
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.970588235294118e-06,
+      "loss": 3.8679,
+      "step": 1154
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.96875e-06,
+      "loss": 3.8725,
+      "step": 1155
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9669117647058828e-06,
+      "loss": 3.8268,
+      "step": 1156
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.965073529411765e-06,
+      "loss": 3.8411,
+      "step": 1157
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.963235294117647e-06,
+      "loss": 3.855,
+      "step": 1158
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9613970588235297e-06,
+      "loss": 3.8503,
+      "step": 1159
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.959558823529412e-06,
+      "loss": 3.9597,
+      "step": 1160
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9577205882352945e-06,
+      "loss": 3.9216,
+      "step": 1161
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9558823529411767e-06,
+      "loss": 3.8542,
+      "step": 1162
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9540441176470593e-06,
+      "loss": 3.8598,
+      "step": 1163
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9522058823529415e-06,
+      "loss": 3.7886,
+      "step": 1164
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9503676470588237e-06,
+      "loss": 3.871,
+      "step": 1165
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9485294117647063e-06,
+      "loss": 3.7283,
+      "step": 1166
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9466911764705885e-06,
+      "loss": 3.8953,
+      "step": 1167
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.944852941176471e-06,
+      "loss": 3.8464,
+      "step": 1168
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.9430147058823533e-06,
+      "loss": 3.9064,
+      "step": 1169
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 3.871,
+      "step": 1170
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.939338235294118e-06,
+      "loss": 3.9004,
+      "step": 1171
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.9375000000000003e-06,
+      "loss": 3.7616,
+      "step": 1172
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.935661764705883e-06,
+      "loss": 3.848,
+      "step": 1173
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.933823529411765e-06,
+      "loss": 3.8248,
+      "step": 1174
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9319852941176472e-06,
+      "loss": 3.903,
+      "step": 1175
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.93014705882353e-06,
+      "loss": 3.8677,
+      "step": 1176
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.928308823529412e-06,
+      "loss": 3.9226,
+      "step": 1177
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9264705882352946e-06,
+      "loss": 3.7422,
+      "step": 1178
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.924632352941177e-06,
+      "loss": 3.7978,
+      "step": 1179
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9227941176470594e-06,
+      "loss": 3.8876,
+      "step": 1180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9209558823529416e-06,
+      "loss": 3.8557,
+      "step": 1181
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9191176470588238e-06,
+      "loss": 3.8229,
+      "step": 1182
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9172794117647064e-06,
+      "loss": 3.9248,
+      "step": 1183
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9154411764705886e-06,
+      "loss": 3.9013,
+      "step": 1184
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.913602941176471e-06,
+      "loss": 3.851,
+      "step": 1185
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.9117647058823534e-06,
+      "loss": 3.9082,
+      "step": 1186
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.9099264705882355e-06,
+      "loss": 3.876,
+      "step": 1187
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.908088235294118e-06,
+      "loss": 3.8759,
+      "step": 1188
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.9062500000000003e-06,
+      "loss": 3.9206,
+      "step": 1189
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.904411764705883e-06,
+      "loss": 3.8957,
+      "step": 1190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.902573529411765e-06,
+      "loss": 3.9239,
+      "step": 1191
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.9007352941176477e-06,
+      "loss": 3.9166,
+      "step": 1192
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.89889705882353e-06,
+      "loss": 3.8544,
+      "step": 1193
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.897058823529412e-06,
+      "loss": 3.894,
+      "step": 1194
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8952205882352947e-06,
+      "loss": 3.8498,
+      "step": 1195
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.893382352941177e-06,
+      "loss": 3.8974,
+      "step": 1196
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8915441176470595e-06,
+      "loss": 3.7968,
+      "step": 1197
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8897058823529413e-06,
+      "loss": 3.7809,
+      "step": 1198
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8878676470588234e-06,
+      "loss": 3.871,
+      "step": 1199
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.886029411764706e-06,
+      "loss": 3.8878,
+      "step": 1200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8841911764705882e-06,
+      "loss": 3.8607,
+      "step": 1201
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8823529411764704e-06,
+      "loss": 3.8891,
+      "step": 1202
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.880514705882353e-06,
+      "loss": 3.8552,
+      "step": 1203
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.878676470588235e-06,
+      "loss": 3.834,
+      "step": 1204
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.876838235294118e-06,
+      "loss": 3.8468,
+      "step": 1205
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.875e-06,
+      "loss": 3.8964,
+      "step": 1206
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.873161764705882e-06,
+      "loss": 3.8691,
+      "step": 1207
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.871323529411765e-06,
+      "loss": 3.8125,
+      "step": 1208
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.869485294117647e-06,
+      "loss": 3.8795,
+      "step": 1209
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8676470588235296e-06,
+      "loss": 3.8424,
+      "step": 1210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8658088235294118e-06,
+      "loss": 3.8193,
+      "step": 1211
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8639705882352944e-06,
+      "loss": 3.8687,
+      "step": 1212
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8621323529411766e-06,
+      "loss": 3.861,
+      "step": 1213
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8602941176470587e-06,
+      "loss": 3.8656,
+      "step": 1214
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8584558823529413e-06,
+      "loss": 3.9022,
+      "step": 1215
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8566176470588235e-06,
+      "loss": 3.8321,
+      "step": 1216
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.854779411764706e-06,
+      "loss": 3.7633,
+      "step": 1217
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8529411764705883e-06,
+      "loss": 3.8294,
+      "step": 1218
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8511029411764705e-06,
+      "loss": 3.9509,
+      "step": 1219
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.849264705882353e-06,
+      "loss": 3.8795,
+      "step": 1220
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8474264705882353e-06,
+      "loss": 3.842,
+      "step": 1221
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.845588235294118e-06,
+      "loss": 3.9093,
+      "step": 1222
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.84375e-06,
+      "loss": 3.8812,
+      "step": 1223
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.8419117647058823e-06,
+      "loss": 3.8946,
+      "step": 1224
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.840073529411765e-06,
+      "loss": 3.7789,
+      "step": 1225
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.838235294117647e-06,
+      "loss": 3.9029,
+      "step": 1226
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.8363970588235297e-06,
+      "loss": 3.8211,
+      "step": 1227
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.834558823529412e-06,
+      "loss": 3.8598,
+      "step": 1228
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.8327205882352945e-06,
+      "loss": 3.9756,
+      "step": 1229
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.8308823529411766e-06,
+      "loss": 3.9057,
+      "step": 1230
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.829044117647059e-06,
+      "loss": 3.7876,
+      "step": 1231
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.8272058823529414e-06,
+      "loss": 3.7212,
+      "step": 1232
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.8253676470588236e-06,
+      "loss": 3.8179,
+      "step": 1233
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.8235294117647062e-06,
+      "loss": 3.8163,
+      "step": 1234
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.8216911764705884e-06,
+      "loss": 3.9242,
+      "step": 1235
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.8198529411764706e-06,
+      "loss": 3.873,
+      "step": 1236
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.818014705882353e-06,
+      "loss": 3.8443,
+      "step": 1237
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.8161764705882354e-06,
+      "loss": 3.8935,
+      "step": 1238
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.814338235294118e-06,
+      "loss": 3.7905,
+      "step": 1239
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8125e-06,
+      "loss": 3.8536,
+      "step": 1240
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8106617647058828e-06,
+      "loss": 3.808,
+      "step": 1241
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.808823529411765e-06,
+      "loss": 3.9448,
+      "step": 1242
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.806985294117647e-06,
+      "loss": 3.8521,
+      "step": 1243
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.8051470588235298e-06,
+      "loss": 3.751,
+      "step": 1244
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.803308823529412e-06,
+      "loss": 3.884,
+      "step": 1245
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.8014705882352945e-06,
+      "loss": 3.8477,
+      "step": 1246
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7996323529411767e-06,
+      "loss": 3.8389,
+      "step": 1247
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.797794117647059e-06,
+      "loss": 3.8672,
+      "step": 1248
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7959558823529415e-06,
+      "loss": 3.8047,
+      "step": 1249
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7941176470588237e-06,
+      "loss": 3.7926,
+      "step": 1250
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7922794117647063e-06,
+      "loss": 3.8726,
+      "step": 1251
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7904411764705885e-06,
+      "loss": 3.8393,
+      "step": 1252
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.788602941176471e-06,
+      "loss": 3.8778,
+      "step": 1253
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7867647058823533e-06,
+      "loss": 3.8925,
+      "step": 1254
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7849264705882355e-06,
+      "loss": 3.8154,
+      "step": 1255
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.783088235294118e-06,
+      "loss": 3.8656,
+      "step": 1256
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7812500000000003e-06,
+      "loss": 3.778,
+      "step": 1257
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.779411764705883e-06,
+      "loss": 3.7332,
+      "step": 1258
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.777573529411765e-06,
+      "loss": 3.8294,
+      "step": 1259
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7757352941176472e-06,
+      "loss": 3.7332,
+      "step": 1260
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.77389705882353e-06,
+      "loss": 3.9188,
+      "step": 1261
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.772058823529412e-06,
+      "loss": 3.968,
+      "step": 1262
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7702205882352946e-06,
+      "loss": 3.7484,
+      "step": 1263
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.768382352941177e-06,
+      "loss": 3.8176,
+      "step": 1264
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.766544117647059e-06,
+      "loss": 3.889,
+      "step": 1265
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7647058823529416e-06,
+      "loss": 3.7576,
+      "step": 1266
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.762867647058824e-06,
+      "loss": 3.9066,
+      "step": 1267
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7610294117647064e-06,
+      "loss": 3.9179,
+      "step": 1268
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7591911764705886e-06,
+      "loss": 3.8583,
+      "step": 1269
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.757352941176471e-06,
+      "loss": 3.9205,
+      "step": 1270
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7555147058823534e-06,
+      "loss": 3.9177,
+      "step": 1271
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7536764705882356e-06,
+      "loss": 3.7967,
+      "step": 1272
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.751838235294118e-06,
+      "loss": 3.8539,
+      "step": 1273
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 3.8079,
+      "step": 1274
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.748161764705883e-06,
+      "loss": 3.9461,
+      "step": 1275
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.746323529411765e-06,
+      "loss": 3.7719,
+      "step": 1276
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7444852941176473e-06,
+      "loss": 3.8149,
+      "step": 1277
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.74264705882353e-06,
+      "loss": 3.9015,
+      "step": 1278
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.740808823529412e-06,
+      "loss": 3.81,
+      "step": 1279
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.7389705882352947e-06,
+      "loss": 3.7756,
+      "step": 1280
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.737132352941177e-06,
+      "loss": 3.9043,
+      "step": 1281
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7352941176470595e-06,
+      "loss": 3.8091,
+      "step": 1282
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7334558823529413e-06,
+      "loss": 3.8349,
+      "step": 1283
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7316176470588235e-06,
+      "loss": 3.7652,
+      "step": 1284
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7297794117647056e-06,
+      "loss": 3.8246,
+      "step": 1285
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7279411764705883e-06,
+      "loss": 3.8528,
+      "step": 1286
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7261029411764704e-06,
+      "loss": 3.7847,
+      "step": 1287
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.724264705882353e-06,
+      "loss": 3.9511,
+      "step": 1288
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7224264705882352e-06,
+      "loss": 3.8301,
+      "step": 1289
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.720588235294118e-06,
+      "loss": 3.7516,
+      "step": 1290
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.71875e-06,
+      "loss": 3.8913,
+      "step": 1291
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.716911764705882e-06,
+      "loss": 3.9193,
+      "step": 1292
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.715073529411765e-06,
+      "loss": 3.9011,
+      "step": 1293
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.713235294117647e-06,
+      "loss": 3.8158,
+      "step": 1294
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.7113970588235296e-06,
+      "loss": 3.8093,
+      "step": 1295
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.7095588235294118e-06,
+      "loss": 3.8375,
+      "step": 1296
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.707720588235294e-06,
+      "loss": 3.904,
+      "step": 1297
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.7058823529411766e-06,
+      "loss": 3.7123,
+      "step": 1298
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.7040441176470588e-06,
+      "loss": 3.8847,
+      "step": 1299
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.7022058823529414e-06,
+      "loss": 3.7486,
+      "step": 1300
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7003676470588235e-06,
+      "loss": 3.868,
+      "step": 1301
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.698529411764706e-06,
+      "loss": 3.9272,
+      "step": 1302
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6966911764705883e-06,
+      "loss": 3.8874,
+      "step": 1303
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6948529411764705e-06,
+      "loss": 3.8191,
+      "step": 1304
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.693014705882353e-06,
+      "loss": 3.8455,
+      "step": 1305
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6911764705882353e-06,
+      "loss": 3.8703,
+      "step": 1306
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.689338235294118e-06,
+      "loss": 3.8898,
+      "step": 1307
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6875e-06,
+      "loss": 3.6968,
+      "step": 1308
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6856617647058823e-06,
+      "loss": 3.8408,
+      "step": 1309
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.683823529411765e-06,
+      "loss": 3.8895,
+      "step": 1310
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.681985294117647e-06,
+      "loss": 3.8222,
+      "step": 1311
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6801470588235297e-06,
+      "loss": 3.9387,
+      "step": 1312
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.678308823529412e-06,
+      "loss": 3.7954,
+      "step": 1313
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6764705882352945e-06,
+      "loss": 3.8852,
+      "step": 1314
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6746323529411767e-06,
+      "loss": 3.8554,
+      "step": 1315
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.672794117647059e-06,
+      "loss": 3.7885,
+      "step": 1316
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6709558823529415e-06,
+      "loss": 3.8982,
+      "step": 1317
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6691176470588236e-06,
+      "loss": 3.8127,
+      "step": 1318
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6672794117647062e-06,
+      "loss": 3.8841,
+      "step": 1319
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6654411764705884e-06,
+      "loss": 3.7996,
+      "step": 1320
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6636029411764706e-06,
+      "loss": 3.8403,
+      "step": 1321
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6617647058823532e-06,
+      "loss": 3.9211,
+      "step": 1322
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6599264705882354e-06,
+      "loss": 3.7752,
+      "step": 1323
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.658088235294118e-06,
+      "loss": 3.8452,
+      "step": 1324
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.65625e-06,
+      "loss": 3.8486,
+      "step": 1325
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6544117647058824e-06,
+      "loss": 3.8544,
+      "step": 1326
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.652573529411765e-06,
+      "loss": 3.8022,
+      "step": 1327
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.650735294117647e-06,
+      "loss": 3.867,
+      "step": 1328
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6488970588235298e-06,
+      "loss": 3.8698,
+      "step": 1329
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 3.795,
+      "step": 1330
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6452205882352946e-06,
+      "loss": 3.8937,
+      "step": 1331
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6433823529411767e-06,
+      "loss": 3.8101,
+      "step": 1332
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.641544117647059e-06,
+      "loss": 3.8737,
+      "step": 1333
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.6397058823529415e-06,
+      "loss": 3.8933,
+      "step": 1334
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.6378676470588237e-06,
+      "loss": 3.8787,
+      "step": 1335
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.6360294117647063e-06,
+      "loss": 3.8354,
+      "step": 1336
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.6341911764705885e-06,
+      "loss": 3.7723,
+      "step": 1337
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.6323529411764707e-06,
+      "loss": 3.8334,
+      "step": 1338
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.6305147058823533e-06,
+      "loss": 3.8109,
+      "step": 1339
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6286764705882355e-06,
+      "loss": 3.721,
+      "step": 1340
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.626838235294118e-06,
+      "loss": 3.9016,
+      "step": 1341
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6250000000000003e-06,
+      "loss": 3.8424,
+      "step": 1342
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.623161764705883e-06,
+      "loss": 3.8434,
+      "step": 1343
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.621323529411765e-06,
+      "loss": 3.8004,
+      "step": 1344
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6194852941176473e-06,
+      "loss": 3.8638,
+      "step": 1345
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.61764705882353e-06,
+      "loss": 3.8666,
+      "step": 1346
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.615808823529412e-06,
+      "loss": 3.8641,
+      "step": 1347
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.6139705882352947e-06,
+      "loss": 3.8302,
+      "step": 1348
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.612132352941177e-06,
+      "loss": 3.7778,
+      "step": 1349
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.610294117647059e-06,
+      "loss": 3.8042,
+      "step": 1350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.6084558823529416e-06,
+      "loss": 3.7827,
+      "step": 1351
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.606617647058824e-06,
+      "loss": 3.8356,
+      "step": 1352
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.6047794117647064e-06,
+      "loss": 3.7598,
+      "step": 1353
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6029411764705886e-06,
+      "loss": 3.8939,
+      "step": 1354
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.601102941176471e-06,
+      "loss": 3.8656,
+      "step": 1355
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5992647058823534e-06,
+      "loss": 3.8924,
+      "step": 1356
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5974264705882356e-06,
+      "loss": 3.8417,
+      "step": 1357
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.595588235294118e-06,
+      "loss": 3.8414,
+      "step": 1358
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5937500000000004e-06,
+      "loss": 3.7218,
+      "step": 1359
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.591911764705883e-06,
+      "loss": 3.7599,
+      "step": 1360
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.590073529411765e-06,
+      "loss": 3.9119,
+      "step": 1361
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5882352941176473e-06,
+      "loss": 3.8041,
+      "step": 1362
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.58639705882353e-06,
+      "loss": 3.777,
+      "step": 1363
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.584558823529412e-06,
+      "loss": 3.8667,
+      "step": 1364
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5827205882352947e-06,
+      "loss": 3.8421,
+      "step": 1365
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.580882352941177e-06,
+      "loss": 3.8412,
+      "step": 1366
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.579044117647059e-06,
+      "loss": 3.8712,
+      "step": 1367
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5772058823529413e-06,
+      "loss": 3.7757,
+      "step": 1368
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5753676470588235e-06,
+      "loss": 3.7206,
+      "step": 1369
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5735294117647057e-06,
+      "loss": 3.8781,
+      "step": 1370
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5716911764705883e-06,
+      "loss": 3.9022,
+      "step": 1371
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5698529411764704e-06,
+      "loss": 3.6748,
+      "step": 1372
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.568014705882353e-06,
+      "loss": 3.747,
+      "step": 1373
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5661764705882352e-06,
+      "loss": 3.8702,
+      "step": 1374
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5643382352941174e-06,
+      "loss": 3.7941,
+      "step": 1375
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5625e-06,
+      "loss": 3.8689,
+      "step": 1376
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5606617647058822e-06,
+      "loss": 3.7528,
+      "step": 1377
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.558823529411765e-06,
+      "loss": 3.7879,
+      "step": 1378
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.556985294117647e-06,
+      "loss": 3.842,
+      "step": 1379
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5551470588235296e-06,
+      "loss": 3.7839,
+      "step": 1380
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.553308823529412e-06,
+      "loss": 3.8471,
+      "step": 1381
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.551470588235294e-06,
+      "loss": 3.7944,
+      "step": 1382
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5496323529411766e-06,
+      "loss": 3.8677,
+      "step": 1383
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5477941176470588e-06,
+      "loss": 3.8319,
+      "step": 1384
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5459558823529414e-06,
+      "loss": 3.6766,
+      "step": 1385
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.12162162162162163,
+      "eval_loss": 3.8140854835510254,
+      "eval_runtime": 141.8941,
+      "eval_samples_per_second": 2.608,
+      "eval_steps_per_second": 0.655,
+      "step": 1385
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5441176470588236e-06,
+      "loss": 3.9093,
+      "step": 1386
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.5422794117647057e-06,
+      "loss": 3.931,
+      "step": 1387
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.5404411764705884e-06,
+      "loss": 3.7568,
+      "step": 1388
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.5386029411764705e-06,
+      "loss": 3.8262,
+      "step": 1389
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.536764705882353e-06,
+      "loss": 3.8426,
+      "step": 1390
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.5349264705882353e-06,
+      "loss": 3.7376,
+      "step": 1391
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.533088235294118e-06,
+      "loss": 3.8782,
+      "step": 1392
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.53125e-06,
+      "loss": 3.7147,
+      "step": 1393
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.5294117647058823e-06,
+      "loss": 3.7964,
+      "step": 1394
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.527573529411765e-06,
+      "loss": 3.783,
+      "step": 1395
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.525735294117647e-06,
+      "loss": 3.7448,
+      "step": 1396
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.5238970588235297e-06,
+      "loss": 3.7199,
+      "step": 1397
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.522058823529412e-06,
+      "loss": 3.7467,
+      "step": 1398
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.520220588235294e-06,
+      "loss": 3.7637,
+      "step": 1399
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.5183823529411767e-06,
+      "loss": 3.8261,
+      "step": 1400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.516544117647059e-06,
+      "loss": 3.8239,
+      "step": 1401
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.5147058823529415e-06,
+      "loss": 3.816,
+      "step": 1402
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.5128676470588236e-06,
+      "loss": 3.8442,
+      "step": 1403
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.5110294117647063e-06,
+      "loss": 3.8801,
+      "step": 1404
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.5091911764705884e-06,
+      "loss": 3.8965,
+      "step": 1405
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.5073529411764706e-06,
+      "loss": 3.9093,
+      "step": 1406
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.5055147058823532e-06,
+      "loss": 3.8445,
+      "step": 1407
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.5036764705882354e-06,
+      "loss": 3.7602,
+      "step": 1408
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.501838235294118e-06,
+      "loss": 3.8217,
+      "step": 1409
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.5e-06,
+      "loss": 3.8104,
+      "step": 1410
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4981617647058824e-06,
+      "loss": 3.7757,
+      "step": 1411
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.496323529411765e-06,
+      "loss": 3.7992,
+      "step": 1412
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.494485294117647e-06,
+      "loss": 3.875,
+      "step": 1413
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4926470588235298e-06,
+      "loss": 3.927,
+      "step": 1414
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.490808823529412e-06,
+      "loss": 3.8832,
+      "step": 1415
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.488970588235294e-06,
+      "loss": 3.8145,
+      "step": 1416
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4871323529411768e-06,
+      "loss": 3.7506,
+      "step": 1417
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.485294117647059e-06,
+      "loss": 3.9218,
+      "step": 1418
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4834558823529416e-06,
+      "loss": 3.8855,
+      "step": 1419
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4816176470588237e-06,
+      "loss": 3.8391,
+      "step": 1420
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4797794117647063e-06,
+      "loss": 3.8401,
+      "step": 1421
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4779411764705885e-06,
+      "loss": 3.7426,
+      "step": 1422
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4761029411764707e-06,
+      "loss": 3.7924,
+      "step": 1423
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4742647058823533e-06,
+      "loss": 3.8432,
+      "step": 1424
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4724264705882355e-06,
+      "loss": 3.7874,
+      "step": 1425
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.470588235294118e-06,
+      "loss": 3.9072,
+      "step": 1426
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4687500000000003e-06,
+      "loss": 3.7645,
+      "step": 1427
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4669117647058825e-06,
+      "loss": 3.6841,
+      "step": 1428
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.465073529411765e-06,
+      "loss": 3.8278,
+      "step": 1429
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4632352941176473e-06,
+      "loss": 3.8375,
+      "step": 1430
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.46139705882353e-06,
+      "loss": 3.826,
+      "step": 1431
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4595588235294116e-06,
+      "loss": 3.8361,
+      "step": 1432
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4577205882352942e-06,
+      "loss": 3.7711,
+      "step": 1433
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4558823529411764e-06,
+      "loss": 3.8912,
+      "step": 1434
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.454044117647059e-06,
+      "loss": 3.7994,
+      "step": 1435
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4522058823529412e-06,
+      "loss": 3.861,
+      "step": 1436
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.450367647058824e-06,
+      "loss": 3.8326,
+      "step": 1437
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.448529411764706e-06,
+      "loss": 3.8962,
+      "step": 1438
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.446691176470588e-06,
+      "loss": 3.8763,
+      "step": 1439
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.444852941176471e-06,
+      "loss": 3.8224,
+      "step": 1440
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.443014705882353e-06,
+      "loss": 3.7163,
+      "step": 1441
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.4411764705882356e-06,
+      "loss": 3.663,
+      "step": 1442
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.4393382352941178e-06,
+      "loss": 3.7742,
+      "step": 1443
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.4375e-06,
+      "loss": 3.7776,
+      "step": 1444
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.4356617647058826e-06,
+      "loss": 3.7575,
+      "step": 1445
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.4338235294117647e-06,
+      "loss": 3.7288,
+      "step": 1446
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.4319852941176474e-06,
+      "loss": 3.7924,
+      "step": 1447
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.4301470588235295e-06,
+      "loss": 3.7177,
+      "step": 1448
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.428308823529412e-06,
+      "loss": 3.7735,
+      "step": 1449
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.4264705882352943e-06,
+      "loss": 3.7613,
+      "step": 1450
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.4246323529411765e-06,
+      "loss": 3.6932,
+      "step": 1451
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.422794117647059e-06,
+      "loss": 3.9268,
+      "step": 1452
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.4209558823529413e-06,
+      "loss": 3.8574,
+      "step": 1453
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.419117647058824e-06,
+      "loss": 3.8107,
+      "step": 1454
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.417279411764706e-06,
+      "loss": 3.8143,
+      "step": 1455
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.4154411764705883e-06,
+      "loss": 3.8233,
+      "step": 1456
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.413602941176471e-06,
+      "loss": 3.8765,
+      "step": 1457
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.411764705882353e-06,
+      "loss": 3.842,
+      "step": 1458
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.4099264705882357e-06,
+      "loss": 3.7893,
+      "step": 1459
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.408088235294118e-06,
+      "loss": 3.8639,
+      "step": 1460
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.40625e-06,
+      "loss": 3.837,
+      "step": 1461
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.4044117647058827e-06,
+      "loss": 3.8223,
+      "step": 1462
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.402573529411765e-06,
+      "loss": 3.7547,
+      "step": 1463
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.4007352941176474e-06,
+      "loss": 3.8397,
+      "step": 1464
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3988970588235296e-06,
+      "loss": 3.8184,
+      "step": 1465
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3970588235294122e-06,
+      "loss": 3.8426,
+      "step": 1466
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3952205882352944e-06,
+      "loss": 3.7062,
+      "step": 1467
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3933823529411766e-06,
+      "loss": 3.811,
+      "step": 1468
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.391544117647059e-06,
+      "loss": 3.8753,
+      "step": 1469
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3897058823529414e-06,
+      "loss": 3.7554,
+      "step": 1470
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.387867647058824e-06,
+      "loss": 3.8772,
+      "step": 1471
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.386029411764706e-06,
+      "loss": 3.7707,
+      "step": 1472
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3841911764705884e-06,
+      "loss": 3.7515,
+      "step": 1473
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3823529411764705e-06,
+      "loss": 3.8449,
+      "step": 1474
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.380514705882353e-06,
+      "loss": 3.7729,
+      "step": 1475
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3786764705882353e-06,
+      "loss": 3.8959,
+      "step": 1476
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3768382352941175e-06,
+      "loss": 3.8706,
+      "step": 1477
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.375e-06,
+      "loss": 3.8376,
+      "step": 1478
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3731617647058823e-06,
+      "loss": 3.8748,
+      "step": 1479
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.371323529411765e-06,
+      "loss": 3.8586,
+      "step": 1480
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.369485294117647e-06,
+      "loss": 3.7315,
+      "step": 1481
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3676470588235297e-06,
+      "loss": 3.8433,
+      "step": 1482
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.365808823529412e-06,
+      "loss": 3.8453,
+      "step": 1483
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.363970588235294e-06,
+      "loss": 3.9179,
+      "step": 1484
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3621323529411767e-06,
+      "loss": 3.7572,
+      "step": 1485
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.360294117647059e-06,
+      "loss": 3.8036,
+      "step": 1486
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3584558823529415e-06,
+      "loss": 3.9434,
+      "step": 1487
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3566176470588237e-06,
+      "loss": 3.7737,
+      "step": 1488
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.354779411764706e-06,
+      "loss": 3.8794,
+      "step": 1489
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 3.7338,
+      "step": 1490
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3511029411764706e-06,
+      "loss": 3.7767,
+      "step": 1491
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3492647058823532e-06,
+      "loss": 3.8121,
+      "step": 1492
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3474264705882354e-06,
+      "loss": 3.8633,
+      "step": 1493
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.345588235294118e-06,
+      "loss": 3.8826,
+      "step": 1494
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3437500000000002e-06,
+      "loss": 3.734,
+      "step": 1495
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3419117647058824e-06,
+      "loss": 3.7519,
+      "step": 1496
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.340073529411765e-06,
+      "loss": 3.8873,
+      "step": 1497
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.338235294117647e-06,
+      "loss": 3.8564,
+      "step": 1498
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.33639705882353e-06,
+      "loss": 3.7875,
+      "step": 1499
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.334558823529412e-06,
+      "loss": 3.8389,
+      "step": 1500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.332720588235294e-06,
+      "loss": 3.8298,
+      "step": 1501
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.3308823529411768e-06,
+      "loss": 3.8263,
+      "step": 1502
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.329044117647059e-06,
+      "loss": 3.8655,
+      "step": 1503
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.3272058823529416e-06,
+      "loss": 3.7766,
+      "step": 1504
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.3253676470588237e-06,
+      "loss": 3.7153,
+      "step": 1505
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.323529411764706e-06,
+      "loss": 3.753,
+      "step": 1506
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.3216911764705885e-06,
+      "loss": 3.8044,
+      "step": 1507
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.3198529411764707e-06,
+      "loss": 3.7462,
+      "step": 1508
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.3180147058823533e-06,
+      "loss": 3.7945,
+      "step": 1509
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.3161764705882355e-06,
+      "loss": 3.8137,
+      "step": 1510
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.314338235294118e-06,
+      "loss": 3.8258,
+      "step": 1511
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.3125000000000003e-06,
+      "loss": 3.851,
+      "step": 1512
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.3106617647058825e-06,
+      "loss": 3.8219,
+      "step": 1513
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.308823529411765e-06,
+      "loss": 3.7904,
+      "step": 1514
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.3069852941176473e-06,
+      "loss": 3.8711,
+      "step": 1515
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.30514705882353e-06,
+      "loss": 3.7317,
+      "step": 1516
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.3033088235294116e-06,
+      "loss": 3.8045,
+      "step": 1517
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.3014705882352943e-06,
+      "loss": 3.7574,
+      "step": 1518
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2996323529411764e-06,
+      "loss": 3.7033,
+      "step": 1519
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.297794117647059e-06,
+      "loss": 3.8548,
+      "step": 1520
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2959558823529412e-06,
+      "loss": 3.851,
+      "step": 1521
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2941176470588234e-06,
+      "loss": 3.8846,
+      "step": 1522
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.292279411764706e-06,
+      "loss": 3.8431,
+      "step": 1523
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.290441176470588e-06,
+      "loss": 3.7661,
+      "step": 1524
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.288602941176471e-06,
+      "loss": 3.8581,
+      "step": 1525
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.286764705882353e-06,
+      "loss": 3.7818,
+      "step": 1526
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2849264705882356e-06,
+      "loss": 3.7983,
+      "step": 1527
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2830882352941178e-06,
+      "loss": 3.838,
+      "step": 1528
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.28125e-06,
+      "loss": 3.709,
+      "step": 1529
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2794117647058826e-06,
+      "loss": 3.8734,
+      "step": 1530
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2775735294117648e-06,
+      "loss": 3.7633,
+      "step": 1531
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2757352941176474e-06,
+      "loss": 3.8374,
+      "step": 1532
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2738970588235296e-06,
+      "loss": 3.8527,
+      "step": 1533
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2720588235294117e-06,
+      "loss": 3.7979,
+      "step": 1534
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2702205882352943e-06,
+      "loss": 3.9243,
+      "step": 1535
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2683823529411765e-06,
+      "loss": 3.9123,
+      "step": 1536
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.266544117647059e-06,
+      "loss": 3.8441,
+      "step": 1537
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2647058823529413e-06,
+      "loss": 3.6624,
+      "step": 1538
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.262867647058824e-06,
+      "loss": 3.8357,
+      "step": 1539
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.261029411764706e-06,
+      "loss": 3.7374,
+      "step": 1540
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2591911764705883e-06,
+      "loss": 3.8683,
+      "step": 1541
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.257352941176471e-06,
+      "loss": 3.8525,
+      "step": 1542
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.255514705882353e-06,
+      "loss": 3.8723,
+      "step": 1543
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2536764705882357e-06,
+      "loss": 3.8028,
+      "step": 1544
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.251838235294118e-06,
+      "loss": 3.7724,
+      "step": 1545
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.25e-06,
+      "loss": 3.7953,
+      "step": 1546
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2481617647058827e-06,
+      "loss": 3.8992,
+      "step": 1547
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.246323529411765e-06,
+      "loss": 3.7464,
+      "step": 1548
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2444852941176475e-06,
+      "loss": 3.7829,
+      "step": 1549
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2426470588235296e-06,
+      "loss": 3.7783,
+      "step": 1550
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2408088235294122e-06,
+      "loss": 3.7639,
+      "step": 1551
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2389705882352944e-06,
+      "loss": 3.8702,
+      "step": 1552
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2371323529411766e-06,
+      "loss": 3.8081,
+      "step": 1553
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2352941176470592e-06,
+      "loss": 3.732,
+      "step": 1554
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2334558823529414e-06,
+      "loss": 3.8951,
+      "step": 1555
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.231617647058824e-06,
+      "loss": 3.8567,
+      "step": 1556
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.229779411764706e-06,
+      "loss": 3.9217,
+      "step": 1557
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.2279411764705884e-06,
+      "loss": 3.8078,
+      "step": 1558
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.2261029411764706e-06,
+      "loss": 3.8843,
+      "step": 1559
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.224264705882353e-06,
+      "loss": 3.7829,
+      "step": 1560
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.2224264705882354e-06,
+      "loss": 3.7859,
+      "step": 1561
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.2205882352941175e-06,
+      "loss": 3.7096,
+      "step": 1562
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.21875e-06,
+      "loss": 3.7558,
+      "step": 1563
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.2169117647058823e-06,
+      "loss": 3.7685,
+      "step": 1564
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.215073529411765e-06,
+      "loss": 3.8789,
+      "step": 1565
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.213235294117647e-06,
+      "loss": 3.8418,
+      "step": 1566
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.2113970588235293e-06,
+      "loss": 3.7614,
+      "step": 1567
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.209558823529412e-06,
+      "loss": 3.8484,
+      "step": 1568
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.207720588235294e-06,
+      "loss": 3.8733,
+      "step": 1569
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 3.91,
+      "step": 1570
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.204044117647059e-06,
+      "loss": 3.7435,
+      "step": 1571
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.2022058823529415e-06,
+      "loss": 3.7899,
+      "step": 1572
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.2003676470588237e-06,
+      "loss": 3.8131,
+      "step": 1573
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.198529411764706e-06,
+      "loss": 3.8623,
+      "step": 1574
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1966911764705885e-06,
+      "loss": 3.9623,
+      "step": 1575
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1948529411764707e-06,
+      "loss": 3.8614,
+      "step": 1576
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1930147058823533e-06,
+      "loss": 3.8395,
+      "step": 1577
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1911764705882354e-06,
+      "loss": 3.8705,
+      "step": 1578
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1893382352941176e-06,
+      "loss": 3.8814,
+      "step": 1579
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1875000000000002e-06,
+      "loss": 3.7294,
+      "step": 1580
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1856617647058824e-06,
+      "loss": 3.7987,
+      "step": 1581
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.183823529411765e-06,
+      "loss": 3.9202,
+      "step": 1582
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.181985294117647e-06,
+      "loss": 3.8419,
+      "step": 1583
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.18014705882353e-06,
+      "loss": 3.8676,
+      "step": 1584
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.178308823529412e-06,
+      "loss": 3.9422,
+      "step": 1585
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.176470588235294e-06,
+      "loss": 3.7685,
+      "step": 1586
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.174632352941177e-06,
+      "loss": 3.8497,
+      "step": 1587
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.172794117647059e-06,
+      "loss": 3.9639,
+      "step": 1588
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1709558823529416e-06,
+      "loss": 3.8535,
+      "step": 1589
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1691176470588238e-06,
+      "loss": 3.8169,
+      "step": 1590
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.167279411764706e-06,
+      "loss": 3.7593,
+      "step": 1591
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1654411764705886e-06,
+      "loss": 3.7814,
+      "step": 1592
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1636029411764707e-06,
+      "loss": 3.795,
+      "step": 1593
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1617647058823533e-06,
+      "loss": 3.8272,
+      "step": 1594
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1599264705882355e-06,
+      "loss": 3.8347,
+      "step": 1595
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.158088235294118e-06,
+      "loss": 3.8092,
+      "step": 1596
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1562500000000003e-06,
+      "loss": 3.881,
+      "step": 1597
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1544117647058825e-06,
+      "loss": 3.8513,
+      "step": 1598
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.152573529411765e-06,
+      "loss": 3.8728,
+      "step": 1599
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1507352941176473e-06,
+      "loss": 3.803,
+      "step": 1600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.14889705882353e-06,
+      "loss": 3.8366,
+      "step": 1601
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1470588235294117e-06,
+      "loss": 3.9716,
+      "step": 1602
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1452205882352943e-06,
+      "loss": 3.7949,
+      "step": 1603
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1433823529411765e-06,
+      "loss": 3.9464,
+      "step": 1604
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.141544117647059e-06,
+      "loss": 3.7243,
+      "step": 1605
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1397058823529412e-06,
+      "loss": 3.8275,
+      "step": 1606
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1378676470588234e-06,
+      "loss": 3.8628,
+      "step": 1607
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.136029411764706e-06,
+      "loss": 3.7648,
+      "step": 1608
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.1341911764705882e-06,
+      "loss": 3.8267,
+      "step": 1609
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.132352941176471e-06,
+      "loss": 3.9208,
+      "step": 1610
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.130514705882353e-06,
+      "loss": 3.741,
+      "step": 1611
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.128676470588235e-06,
+      "loss": 3.8741,
+      "step": 1612
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.126838235294118e-06,
+      "loss": 3.8353,
+      "step": 1613
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.125e-06,
+      "loss": 3.7605,
+      "step": 1614
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.1231617647058826e-06,
+      "loss": 3.8771,
+      "step": 1615
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.1213235294117648e-06,
+      "loss": 3.7695,
+      "step": 1616
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.1194852941176474e-06,
+      "loss": 3.7435,
+      "step": 1617
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.1176470588235296e-06,
+      "loss": 3.8416,
+      "step": 1618
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.1158088235294117e-06,
+      "loss": 3.7762,
+      "step": 1619
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.1139705882352944e-06,
+      "loss": 3.8753,
+      "step": 1620
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.1121323529411765e-06,
+      "loss": 3.8463,
+      "step": 1621
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.110294117647059e-06,
+      "loss": 3.7864,
+      "step": 1622
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.1084558823529413e-06,
+      "loss": 3.8474,
+      "step": 1623
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.1066176470588235e-06,
+      "loss": 3.8671,
+      "step": 1624
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.104779411764706e-06,
+      "loss": 3.8384,
+      "step": 1625
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.1029411764705883e-06,
+      "loss": 3.7705,
+      "step": 1626
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.101102941176471e-06,
+      "loss": 3.8266,
+      "step": 1627
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.099264705882353e-06,
+      "loss": 3.8695,
+      "step": 1628
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0974264705882357e-06,
+      "loss": 3.7394,
+      "step": 1629
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.095588235294118e-06,
+      "loss": 3.8757,
+      "step": 1630
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.09375e-06,
+      "loss": 3.8877,
+      "step": 1631
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0919117647058827e-06,
+      "loss": 3.9241,
+      "step": 1632
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.090073529411765e-06,
+      "loss": 3.9198,
+      "step": 1633
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0882352941176475e-06,
+      "loss": 3.8002,
+      "step": 1634
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0863970588235297e-06,
+      "loss": 3.8908,
+      "step": 1635
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.084558823529412e-06,
+      "loss": 3.7073,
+      "step": 1636
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0827205882352944e-06,
+      "loss": 3.8732,
+      "step": 1637
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0808823529411766e-06,
+      "loss": 3.8899,
+      "step": 1638
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0790441176470592e-06,
+      "loss": 3.8572,
+      "step": 1639
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0772058823529414e-06,
+      "loss": 3.8527,
+      "step": 1640
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.075367647058824e-06,
+      "loss": 3.8417,
+      "step": 1641
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.073529411764706e-06,
+      "loss": 3.8419,
+      "step": 1642
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0716911764705884e-06,
+      "loss": 3.907,
+      "step": 1643
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0698529411764706e-06,
+      "loss": 3.8896,
+      "step": 1644
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.068014705882353e-06,
+      "loss": 3.7503,
+      "step": 1645
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0661764705882354e-06,
+      "loss": 3.8586,
+      "step": 1646
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0643382352941176e-06,
+      "loss": 3.9173,
+      "step": 1647
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0625e-06,
+      "loss": 3.7068,
+      "step": 1648
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0606617647058823e-06,
+      "loss": 3.8009,
+      "step": 1649
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.058823529411765e-06,
+      "loss": 3.8202,
+      "step": 1650
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.056985294117647e-06,
+      "loss": 3.7903,
+      "step": 1651
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0551470588235293e-06,
+      "loss": 3.8178,
+      "step": 1652
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.053308823529412e-06,
+      "loss": 3.8199,
+      "step": 1653
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.051470588235294e-06,
+      "loss": 3.8116,
+      "step": 1654
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0496323529411767e-06,
+      "loss": 3.8428,
+      "step": 1655
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.047794117647059e-06,
+      "loss": 3.8184,
+      "step": 1656
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.045955882352941e-06,
+      "loss": 3.8504,
+      "step": 1657
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0441176470588237e-06,
+      "loss": 3.8529,
+      "step": 1658
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.042279411764706e-06,
+      "loss": 3.7947,
+      "step": 1659
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0404411764705885e-06,
+      "loss": 3.8466,
+      "step": 1660
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0386029411764707e-06,
+      "loss": 3.8571,
+      "step": 1661
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0367647058823533e-06,
+      "loss": 3.9425,
+      "step": 1662
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.14054054054054055,
+      "eval_loss": 3.786846876144409,
+      "eval_runtime": 142.725,
+      "eval_samples_per_second": 2.592,
+      "eval_steps_per_second": 0.652,
+      "step": 1662
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0349264705882355e-06,
+      "loss": 3.8003,
+      "step": 1663
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.0330882352941176e-06,
+      "loss": 3.7846,
+      "step": 1664
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.0312500000000002e-06,
+      "loss": 3.8741,
+      "step": 1665
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.0294117647058824e-06,
+      "loss": 3.8627,
+      "step": 1666
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.027573529411765e-06,
+      "loss": 3.8545,
+      "step": 1667
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.0257352941176472e-06,
+      "loss": 3.8277,
+      "step": 1668
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.0238970588235294e-06,
+      "loss": 3.8624,
+      "step": 1669
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.022058823529412e-06,
+      "loss": 3.8949,
+      "step": 1670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.020220588235294e-06,
+      "loss": 3.8579,
+      "step": 1671
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.018382352941177e-06,
+      "loss": 3.7085,
+      "step": 1672
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.016544117647059e-06,
+      "loss": 3.6744,
+      "step": 1673
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.0147058823529416e-06,
+      "loss": 3.8076,
+      "step": 1674
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.0128676470588238e-06,
+      "loss": 3.6488,
+      "step": 1675
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.011029411764706e-06,
+      "loss": 3.8435,
+      "step": 1676
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.0091911764705886e-06,
+      "loss": 3.787,
+      "step": 1677
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.0073529411764708e-06,
+      "loss": 3.8646,
+      "step": 1678
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.0055147058823534e-06,
+      "loss": 3.7707,
+      "step": 1679
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.0036764705882355e-06,
+      "loss": 3.7303,
+      "step": 1680
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.0018382352941177e-06,
+      "loss": 3.7749,
+      "step": 1681
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.8464,
+      "step": 1682
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9981617647058825e-06,
+      "loss": 3.8474,
+      "step": 1683
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.996323529411765e-06,
+      "loss": 3.8286,
+      "step": 1684
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9944852941176473e-06,
+      "loss": 3.8651,
+      "step": 1685
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.99264705882353e-06,
+      "loss": 3.88,
+      "step": 1686
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9908088235294117e-06,
+      "loss": 3.8216,
+      "step": 1687
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9889705882352943e-06,
+      "loss": 3.7991,
+      "step": 1688
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9871323529411765e-06,
+      "loss": 3.8404,
+      "step": 1689
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.985294117647059e-06,
+      "loss": 3.7851,
+      "step": 1690
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9834558823529413e-06,
+      "loss": 3.835,
+      "step": 1691
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9816176470588234e-06,
+      "loss": 3.836,
+      "step": 1692
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.979779411764706e-06,
+      "loss": 3.6747,
+      "step": 1693
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9779411764705882e-06,
+      "loss": 3.7738,
+      "step": 1694
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.976102941176471e-06,
+      "loss": 3.7974,
+      "step": 1695
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.974264705882353e-06,
+      "loss": 3.9257,
+      "step": 1696
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.972426470588235e-06,
+      "loss": 3.7793,
+      "step": 1697
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.970588235294118e-06,
+      "loss": 3.89,
+      "step": 1698
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.96875e-06,
+      "loss": 3.992,
+      "step": 1699
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9669117647058826e-06,
+      "loss": 3.8987,
+      "step": 1700
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9650735294117648e-06,
+      "loss": 3.894,
+      "step": 1701
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9632352941176474e-06,
+      "loss": 3.8464,
+      "step": 1702
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9613970588235296e-06,
+      "loss": 3.8553,
+      "step": 1703
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9595588235294118e-06,
+      "loss": 3.864,
+      "step": 1704
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9577205882352944e-06,
+      "loss": 3.858,
+      "step": 1705
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9558823529411766e-06,
+      "loss": 3.8453,
+      "step": 1706
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.954044117647059e-06,
+      "loss": 3.8767,
+      "step": 1707
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9522058823529413e-06,
+      "loss": 3.9418,
+      "step": 1708
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9503676470588235e-06,
+      "loss": 3.8135,
+      "step": 1709
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.948529411764706e-06,
+      "loss": 3.7664,
+      "step": 1710
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9466911764705883e-06,
+      "loss": 3.8857,
+      "step": 1711
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.944852941176471e-06,
+      "loss": 3.8334,
+      "step": 1712
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.943014705882353e-06,
+      "loss": 3.8144,
+      "step": 1713
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9411764705882353e-06,
+      "loss": 3.8831,
+      "step": 1714
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.939338235294118e-06,
+      "loss": 3.8263,
+      "step": 1715
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9375e-06,
+      "loss": 3.7624,
+      "step": 1716
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9356617647058827e-06,
+      "loss": 3.884,
+      "step": 1717
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.933823529411765e-06,
+      "loss": 3.6998,
+      "step": 1718
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.9319852941176475e-06,
+      "loss": 3.8494,
+      "step": 1719
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.9301470588235297e-06,
+      "loss": 3.8094,
+      "step": 1720
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.928308823529412e-06,
+      "loss": 3.9022,
+      "step": 1721
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.9264705882352945e-06,
+      "loss": 3.8381,
+      "step": 1722
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.9246323529411766e-06,
+      "loss": 3.751,
+      "step": 1723
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.9227941176470592e-06,
+      "loss": 3.8755,
+      "step": 1724
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.9209558823529414e-06,
+      "loss": 3.7505,
+      "step": 1725
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.9191176470588236e-06,
+      "loss": 3.7719,
+      "step": 1726
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.9172794117647062e-06,
+      "loss": 3.712,
+      "step": 1727
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.9154411764705884e-06,
+      "loss": 3.8366,
+      "step": 1728
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.9136029411764706e-06,
+      "loss": 3.8508,
+      "step": 1729
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.9117647058823528e-06,
+      "loss": 3.7747,
+      "step": 1730
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.9099264705882354e-06,
+      "loss": 3.8291,
+      "step": 1731
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.9080882352941176e-06,
+      "loss": 3.8345,
+      "step": 1732
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.90625e-06,
+      "loss": 3.8688,
+      "step": 1733
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.9044117647058824e-06,
+      "loss": 3.9339,
+      "step": 1734
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.9025735294117648e-06,
+      "loss": 3.8923,
+      "step": 1735
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.9007352941176471e-06,
+      "loss": 3.8169,
+      "step": 1736
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8988970588235295e-06,
+      "loss": 3.7106,
+      "step": 1737
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.897058823529412e-06,
+      "loss": 3.8349,
+      "step": 1738
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8952205882352941e-06,
+      "loss": 3.9044,
+      "step": 1739
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8933823529411765e-06,
+      "loss": 3.8336,
+      "step": 1740
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.891544117647059e-06,
+      "loss": 3.7836,
+      "step": 1741
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8897058823529413e-06,
+      "loss": 3.8564,
+      "step": 1742
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8878676470588237e-06,
+      "loss": 3.7502,
+      "step": 1743
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.886029411764706e-06,
+      "loss": 3.7814,
+      "step": 1744
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8841911764705883e-06,
+      "loss": 3.8226,
+      "step": 1745
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8823529411764707e-06,
+      "loss": 3.7096,
+      "step": 1746
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.880514705882353e-06,
+      "loss": 3.8493,
+      "step": 1747
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8786764705882355e-06,
+      "loss": 3.7681,
+      "step": 1748
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8768382352941179e-06,
+      "loss": 3.8937,
+      "step": 1749
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 3.7546,
+      "step": 1750
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8731617647058824e-06,
+      "loss": 3.8569,
+      "step": 1751
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8713235294117648e-06,
+      "loss": 3.7643,
+      "step": 1752
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8694852941176472e-06,
+      "loss": 3.8164,
+      "step": 1753
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8676470588235296e-06,
+      "loss": 3.8028,
+      "step": 1754
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.865808823529412e-06,
+      "loss": 3.7994,
+      "step": 1755
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8639705882352942e-06,
+      "loss": 3.6638,
+      "step": 1756
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8621323529411766e-06,
+      "loss": 3.8002,
+      "step": 1757
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.860294117647059e-06,
+      "loss": 3.8703,
+      "step": 1758
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8584558823529414e-06,
+      "loss": 3.8682,
+      "step": 1759
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8566176470588238e-06,
+      "loss": 3.9009,
+      "step": 1760
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8547794117647062e-06,
+      "loss": 3.7888,
+      "step": 1761
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8529411764705884e-06,
+      "loss": 3.7294,
+      "step": 1762
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8511029411764708e-06,
+      "loss": 3.872,
+      "step": 1763
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8492647058823532e-06,
+      "loss": 3.861,
+      "step": 1764
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8474264705882356e-06,
+      "loss": 3.8371,
+      "step": 1765
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.845588235294118e-06,
+      "loss": 3.8096,
+      "step": 1766
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8437500000000003e-06,
+      "loss": 3.7628,
+      "step": 1767
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8419117647058825e-06,
+      "loss": 3.6311,
+      "step": 1768
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.840073529411765e-06,
+      "loss": 3.7765,
+      "step": 1769
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8382352941176473e-06,
+      "loss": 3.7715,
+      "step": 1770
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8363970588235297e-06,
+      "loss": 3.8914,
+      "step": 1771
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8345588235294117e-06,
+      "loss": 3.8217,
+      "step": 1772
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.832720588235294e-06,
+      "loss": 3.8074,
+      "step": 1773
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8308823529411765e-06,
+      "loss": 3.7807,
+      "step": 1774
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.8290441176470589e-06,
+      "loss": 3.7765,
+      "step": 1775
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.8272058823529413e-06,
+      "loss": 3.8895,
+      "step": 1776
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.8253676470588237e-06,
+      "loss": 3.9043,
+      "step": 1777
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.8235294117647058e-06,
+      "loss": 3.7651,
+      "step": 1778
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.8216911764705882e-06,
+      "loss": 3.8798,
+      "step": 1779
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.8198529411764706e-06,
+      "loss": 3.7807,
+      "step": 1780
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.818014705882353e-06,
+      "loss": 3.8245,
+      "step": 1781
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.8161764705882354e-06,
+      "loss": 3.8564,
+      "step": 1782
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.8143382352941178e-06,
+      "loss": 3.9185,
+      "step": 1783
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.8125e-06,
+      "loss": 3.9712,
+      "step": 1784
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.8106617647058824e-06,
+      "loss": 3.771,
+      "step": 1785
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.8088235294117648e-06,
+      "loss": 3.7597,
+      "step": 1786
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.8069852941176472e-06,
+      "loss": 3.949,
+      "step": 1787
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.8051470588235296e-06,
+      "loss": 3.746,
+      "step": 1788
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.803308823529412e-06,
+      "loss": 3.7631,
+      "step": 1789
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.8014705882352942e-06,
+      "loss": 3.6724,
+      "step": 1790
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7996323529411766e-06,
+      "loss": 3.8181,
+      "step": 1791
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.797794117647059e-06,
+      "loss": 3.6771,
+      "step": 1792
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7959558823529414e-06,
+      "loss": 3.8023,
+      "step": 1793
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7941176470588238e-06,
+      "loss": 3.6396,
+      "step": 1794
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7922794117647061e-06,
+      "loss": 3.8385,
+      "step": 1795
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7904411764705883e-06,
+      "loss": 3.859,
+      "step": 1796
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7886029411764707e-06,
+      "loss": 3.8541,
+      "step": 1797
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7867647058823531e-06,
+      "loss": 3.8321,
+      "step": 1798
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7849264705882355e-06,
+      "loss": 3.8423,
+      "step": 1799
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.783088235294118e-06,
+      "loss": 3.7461,
+      "step": 1800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.78125e-06,
+      "loss": 3.8433,
+      "step": 1801
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7794117647058825e-06,
+      "loss": 3.7245,
+      "step": 1802
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7775735294117649e-06,
+      "loss": 3.8481,
+      "step": 1803
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7757352941176473e-06,
+      "loss": 3.7244,
+      "step": 1804
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7738970588235297e-06,
+      "loss": 3.8453,
+      "step": 1805
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.772058823529412e-06,
+      "loss": 3.8145,
+      "step": 1806
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7702205882352943e-06,
+      "loss": 3.7138,
+      "step": 1807
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7683823529411767e-06,
+      "loss": 3.7176,
+      "step": 1808
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.766544117647059e-06,
+      "loss": 3.7195,
+      "step": 1809
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 3.8094,
+      "step": 1810
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7628676470588238e-06,
+      "loss": 3.7769,
+      "step": 1811
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7610294117647062e-06,
+      "loss": 3.8893,
+      "step": 1812
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7591911764705884e-06,
+      "loss": 3.8972,
+      "step": 1813
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7573529411764706e-06,
+      "loss": 3.7307,
+      "step": 1814
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.755514705882353e-06,
+      "loss": 3.8761,
+      "step": 1815
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7536764705882354e-06,
+      "loss": 3.8384,
+      "step": 1816
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7518382352941176e-06,
+      "loss": 3.8063,
+      "step": 1817
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.75e-06,
+      "loss": 3.8318,
+      "step": 1818
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7481617647058824e-06,
+      "loss": 3.7415,
+      "step": 1819
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7463235294117648e-06,
+      "loss": 3.7832,
+      "step": 1820
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7444852941176472e-06,
+      "loss": 3.7597,
+      "step": 1821
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7426470588235296e-06,
+      "loss": 3.7841,
+      "step": 1822
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7408088235294117e-06,
+      "loss": 3.8429,
+      "step": 1823
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7389705882352941e-06,
+      "loss": 3.7655,
+      "step": 1824
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7371323529411765e-06,
+      "loss": 3.8168,
+      "step": 1825
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.735294117647059e-06,
+      "loss": 3.7262,
+      "step": 1826
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7334558823529413e-06,
+      "loss": 3.8554,
+      "step": 1827
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7316176470588237e-06,
+      "loss": 3.8066,
+      "step": 1828
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.729779411764706e-06,
+      "loss": 3.6789,
+      "step": 1829
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7279411764705883e-06,
+      "loss": 3.7543,
+      "step": 1830
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7261029411764707e-06,
+      "loss": 3.7824,
+      "step": 1831
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.724264705882353e-06,
+      "loss": 3.6629,
+      "step": 1832
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.7224264705882355e-06,
+      "loss": 3.7854,
+      "step": 1833
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.7205882352941179e-06,
+      "loss": 3.7961,
+      "step": 1834
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.71875e-06,
+      "loss": 3.7546,
+      "step": 1835
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7169117647058825e-06,
+      "loss": 3.8099,
+      "step": 1836
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7150735294117649e-06,
+      "loss": 3.7741,
+      "step": 1837
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.7132352941176472e-06,
+      "loss": 3.693,
+      "step": 1838
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.7113970588235296e-06,
+      "loss": 3.7291,
+      "step": 1839
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.709558823529412e-06,
+      "loss": 3.903,
+      "step": 1840
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.7077205882352942e-06,
+      "loss": 3.8132,
+      "step": 1841
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.7058823529411766e-06,
+      "loss": 3.6792,
+      "step": 1842
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.704044117647059e-06,
+      "loss": 3.8166,
+      "step": 1843
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.7022058823529414e-06,
+      "loss": 3.8264,
+      "step": 1844
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.7003676470588238e-06,
+      "loss": 3.689,
+      "step": 1845
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.698529411764706e-06,
+      "loss": 3.7161,
+      "step": 1846
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6966911764705884e-06,
+      "loss": 3.7997,
+      "step": 1847
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6948529411764708e-06,
+      "loss": 3.6926,
+      "step": 1848
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6930147058823532e-06,
+      "loss": 3.7508,
+      "step": 1849
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6911764705882356e-06,
+      "loss": 3.7327,
+      "step": 1850
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.689338235294118e-06,
+      "loss": 3.6809,
+      "step": 1851
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6875000000000001e-06,
+      "loss": 3.794,
+      "step": 1852
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6856617647058825e-06,
+      "loss": 3.7229,
+      "step": 1853
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.683823529411765e-06,
+      "loss": 3.8143,
+      "step": 1854
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6819852941176473e-06,
+      "loss": 3.8186,
+      "step": 1855
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6801470588235297e-06,
+      "loss": 3.8055,
+      "step": 1856
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6783088235294117e-06,
+      "loss": 3.8658,
+      "step": 1857
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.676470588235294e-06,
+      "loss": 3.7594,
+      "step": 1858
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6746323529411765e-06,
+      "loss": 3.7909,
+      "step": 1859
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6727941176470589e-06,
+      "loss": 3.6839,
+      "step": 1860
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6709558823529413e-06,
+      "loss": 3.6973,
+      "step": 1861
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6691176470588235e-06,
+      "loss": 3.7935,
+      "step": 1862
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6672794117647059e-06,
+      "loss": 3.8034,
+      "step": 1863
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6654411764705883e-06,
+      "loss": 3.8299,
+      "step": 1864
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6636029411764707e-06,
+      "loss": 3.8731,
+      "step": 1865
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.661764705882353e-06,
+      "loss": 3.8029,
+      "step": 1866
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6599264705882354e-06,
+      "loss": 3.859,
+      "step": 1867
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6580882352941176e-06,
+      "loss": 3.7978,
+      "step": 1868
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.65625e-06,
+      "loss": 3.817,
+      "step": 1869
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6544117647058824e-06,
+      "loss": 3.7914,
+      "step": 1870
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6525735294117648e-06,
+      "loss": 3.742,
+      "step": 1871
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6507352941176472e-06,
+      "loss": 3.8804,
+      "step": 1872
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6488970588235296e-06,
+      "loss": 3.7164,
+      "step": 1873
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6470588235294118e-06,
+      "loss": 3.8845,
+      "step": 1874
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6452205882352942e-06,
+      "loss": 3.77,
+      "step": 1875
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6433823529411766e-06,
+      "loss": 3.7618,
+      "step": 1876
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.641544117647059e-06,
+      "loss": 3.8907,
+      "step": 1877
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6397058823529414e-06,
+      "loss": 3.7796,
+      "step": 1878
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6378676470588238e-06,
+      "loss": 3.8757,
+      "step": 1879
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.636029411764706e-06,
+      "loss": 3.8229,
+      "step": 1880
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6341911764705883e-06,
+      "loss": 3.7782,
+      "step": 1881
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6323529411764707e-06,
+      "loss": 3.7979,
+      "step": 1882
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6305147058823531e-06,
+      "loss": 3.6994,
+      "step": 1883
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6286764705882355e-06,
+      "loss": 3.7792,
+      "step": 1884
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.626838235294118e-06,
+      "loss": 3.8666,
+      "step": 1885
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.6250000000000001e-06,
+      "loss": 3.7292,
+      "step": 1886
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.6231617647058825e-06,
+      "loss": 3.7613,
+      "step": 1887
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.621323529411765e-06,
+      "loss": 3.7594,
+      "step": 1888
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.6194852941176473e-06,
+      "loss": 3.7984,
+      "step": 1889
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.6176470588235297e-06,
+      "loss": 3.7251,
+      "step": 1890
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.615808823529412e-06,
+      "loss": 3.7351,
+      "step": 1891
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.6139705882352943e-06,
+      "loss": 3.745,
+      "step": 1892
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.6121323529411767e-06,
+      "loss": 3.6524,
+      "step": 1893
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.610294117647059e-06,
+      "loss": 3.8404,
+      "step": 1894
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.6084558823529415e-06,
+      "loss": 3.7618,
+      "step": 1895
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.6066176470588239e-06,
+      "loss": 3.7798,
+      "step": 1896
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.604779411764706e-06,
+      "loss": 3.8866,
+      "step": 1897
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.6029411764705884e-06,
+      "loss": 3.7678,
+      "step": 1898
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.6011029411764706e-06,
+      "loss": 3.7194,
+      "step": 1899
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.599264705882353e-06,
+      "loss": 3.8635,
+      "step": 1900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5974264705882352e-06,
+      "loss": 3.8417,
+      "step": 1901
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5955882352941176e-06,
+      "loss": 3.7795,
+      "step": 1902
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.59375e-06,
+      "loss": 3.8634,
+      "step": 1903
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5919117647058824e-06,
+      "loss": 3.8006,
+      "step": 1904
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5900735294117648e-06,
+      "loss": 3.8424,
+      "step": 1905
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5882352941176472e-06,
+      "loss": 3.8695,
+      "step": 1906
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5863970588235294e-06,
+      "loss": 3.7717,
+      "step": 1907
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5845588235294118e-06,
+      "loss": 3.7287,
+      "step": 1908
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5827205882352941e-06,
+      "loss": 3.762,
+      "step": 1909
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5808823529411765e-06,
+      "loss": 3.7972,
+      "step": 1910
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.579044117647059e-06,
+      "loss": 3.8528,
+      "step": 1911
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5772058823529413e-06,
+      "loss": 3.8372,
+      "step": 1912
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5753676470588235e-06,
+      "loss": 3.7855,
+      "step": 1913
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.573529411764706e-06,
+      "loss": 3.8558,
+      "step": 1914
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5716911764705883e-06,
+      "loss": 3.8172,
+      "step": 1915
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5698529411764707e-06,
+      "loss": 3.7989,
+      "step": 1916
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.568014705882353e-06,
+      "loss": 3.7558,
+      "step": 1917
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5661764705882355e-06,
+      "loss": 3.7079,
+      "step": 1918
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5643382352941177e-06,
+      "loss": 3.9108,
+      "step": 1919
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5625e-06,
+      "loss": 3.7567,
+      "step": 1920
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5606617647058825e-06,
+      "loss": 3.6723,
+      "step": 1921
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5588235294117649e-06,
+      "loss": 3.7365,
+      "step": 1922
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5569852941176473e-06,
+      "loss": 3.9026,
+      "step": 1923
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5551470588235297e-06,
+      "loss": 3.7582,
+      "step": 1924
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5533088235294118e-06,
+      "loss": 3.7406,
+      "step": 1925
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5514705882352942e-06,
+      "loss": 3.8318,
+      "step": 1926
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5496323529411766e-06,
+      "loss": 3.8049,
+      "step": 1927
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.547794117647059e-06,
+      "loss": 3.7983,
+      "step": 1928
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5459558823529414e-06,
+      "loss": 3.8789,
+      "step": 1929
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5441176470588238e-06,
+      "loss": 3.8081,
+      "step": 1930
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.542279411764706e-06,
+      "loss": 3.7304,
+      "step": 1931
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5404411764705884e-06,
+      "loss": 3.8131,
+      "step": 1932
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5386029411764708e-06,
+      "loss": 3.7591,
+      "step": 1933
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5367647058823532e-06,
+      "loss": 3.8297,
+      "step": 1934
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5349264705882356e-06,
+      "loss": 3.7684,
+      "step": 1935
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.533088235294118e-06,
+      "loss": 3.8732,
+      "step": 1936
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5312500000000002e-06,
+      "loss": 3.7739,
+      "step": 1937
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5294117647058826e-06,
+      "loss": 3.8292,
+      "step": 1938
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.527573529411765e-06,
+      "loss": 3.7581,
+      "step": 1939
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.16216216216216217,
+      "eval_loss": 3.774104595184326,
+      "eval_runtime": 142.8751,
+      "eval_samples_per_second": 2.59,
+      "eval_steps_per_second": 0.651,
+      "step": 1939
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5257352941176473e-06,
+      "loss": 3.7403,
+      "step": 1940
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5238970588235297e-06,
+      "loss": 3.8248,
+      "step": 1941
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5220588235294117e-06,
+      "loss": 3.6871,
+      "step": 1942
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5202205882352941e-06,
+      "loss": 3.8357,
+      "step": 1943
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.5183823529411765e-06,
+      "loss": 3.8075,
+      "step": 1944
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.516544117647059e-06,
+      "loss": 3.7796,
+      "step": 1945
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5147058823529413e-06,
+      "loss": 3.7146,
+      "step": 1946
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5128676470588235e-06,
+      "loss": 3.8713,
+      "step": 1947
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5110294117647059e-06,
+      "loss": 3.876,
+      "step": 1948
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5091911764705883e-06,
+      "loss": 3.8634,
+      "step": 1949
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5073529411764707e-06,
+      "loss": 3.8102,
+      "step": 1950
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.505514705882353e-06,
+      "loss": 3.7924,
+      "step": 1951
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5036764705882352e-06,
+      "loss": 3.7546,
+      "step": 1952
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5018382352941176e-06,
+      "loss": 3.8252,
+      "step": 1953
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5e-06,
+      "loss": 3.8057,
+      "step": 1954
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4981617647058824e-06,
+      "loss": 3.8298,
+      "step": 1955
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4963235294117648e-06,
+      "loss": 3.7478,
+      "step": 1956
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4944852941176472e-06,
+      "loss": 3.737,
+      "step": 1957
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4926470588235294e-06,
+      "loss": 3.7556,
+      "step": 1958
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4908088235294118e-06,
+      "loss": 3.8264,
+      "step": 1959
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4889705882352942e-06,
+      "loss": 3.7312,
+      "step": 1960
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4871323529411766e-06,
+      "loss": 3.796,
+      "step": 1961
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.485294117647059e-06,
+      "loss": 3.8793,
+      "step": 1962
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4834558823529414e-06,
+      "loss": 3.7915,
+      "step": 1963
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4816176470588236e-06,
+      "loss": 3.754,
+      "step": 1964
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.479779411764706e-06,
+      "loss": 3.8069,
+      "step": 1965
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4779411764705884e-06,
+      "loss": 3.7583,
+      "step": 1966
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4761029411764708e-06,
+      "loss": 3.7031,
+      "step": 1967
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4742647058823532e-06,
+      "loss": 3.7798,
+      "step": 1968
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4724264705882355e-06,
+      "loss": 3.8637,
+      "step": 1969
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 3.7951,
+      "step": 1970
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4687500000000001e-06,
+      "loss": 3.8379,
+      "step": 1971
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4669117647058825e-06,
+      "loss": 3.6724,
+      "step": 1972
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.465073529411765e-06,
+      "loss": 3.7265,
+      "step": 1973
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4632352941176473e-06,
+      "loss": 3.5929,
+      "step": 1974
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4613970588235297e-06,
+      "loss": 3.791,
+      "step": 1975
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4595588235294119e-06,
+      "loss": 3.8131,
+      "step": 1976
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4577205882352943e-06,
+      "loss": 3.838,
+      "step": 1977
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4558823529411767e-06,
+      "loss": 3.857,
+      "step": 1978
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.454044117647059e-06,
+      "loss": 3.7491,
+      "step": 1979
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4522058823529415e-06,
+      "loss": 3.5938,
+      "step": 1980
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4503676470588239e-06,
+      "loss": 3.8188,
+      "step": 1981
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.448529411764706e-06,
+      "loss": 3.8125,
+      "step": 1982
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4466911764705884e-06,
+      "loss": 3.8759,
+      "step": 1983
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4448529411764706e-06,
+      "loss": 3.8634,
+      "step": 1984
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.443014705882353e-06,
+      "loss": 3.8508,
+      "step": 1985
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4411764705882352e-06,
+      "loss": 3.7074,
+      "step": 1986
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4393382352941176e-06,
+      "loss": 3.864,
+      "step": 1987
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4375e-06,
+      "loss": 3.7429,
+      "step": 1988
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4356617647058824e-06,
+      "loss": 3.7337,
+      "step": 1989
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4338235294117648e-06,
+      "loss": 3.8944,
+      "step": 1990
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4319852941176472e-06,
+      "loss": 3.7586,
+      "step": 1991
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4301470588235294e-06,
+      "loss": 3.8116,
+      "step": 1992
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4283088235294118e-06,
+      "loss": 3.8157,
+      "step": 1993
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4264705882352942e-06,
+      "loss": 3.8172,
+      "step": 1994
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4246323529411766e-06,
+      "loss": 3.7848,
+      "step": 1995
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.422794117647059e-06,
+      "loss": 3.9113,
+      "step": 1996
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.4209558823529411e-06,
+      "loss": 3.7809,
+      "step": 1997
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.4191176470588235e-06,
+      "loss": 3.7451,
+      "step": 1998
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.417279411764706e-06,
+      "loss": 3.8775,
+      "step": 1999
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.4154411764705883e-06,
+      "loss": 3.8819,
+      "step": 2000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.4136029411764707e-06,
+      "loss": 3.8487,
+      "step": 2001
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.4117647058823531e-06,
+      "loss": 3.72,
+      "step": 2002
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.4099264705882353e-06,
+      "loss": 3.8203,
+      "step": 2003
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.4080882352941177e-06,
+      "loss": 3.7654,
+      "step": 2004
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.40625e-06,
+      "loss": 3.8784,
+      "step": 2005
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.4044117647058825e-06,
+      "loss": 3.8024,
+      "step": 2006
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.4025735294117649e-06,
+      "loss": 3.7868,
+      "step": 2007
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.4007352941176473e-06,
+      "loss": 3.7785,
+      "step": 2008
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3988970588235295e-06,
+      "loss": 3.918,
+      "step": 2009
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3970588235294119e-06,
+      "loss": 3.6327,
+      "step": 2010
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3952205882352942e-06,
+      "loss": 3.7786,
+      "step": 2011
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3933823529411766e-06,
+      "loss": 3.7522,
+      "step": 2012
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.391544117647059e-06,
+      "loss": 3.8688,
+      "step": 2013
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3897058823529414e-06,
+      "loss": 3.8575,
+      "step": 2014
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3878676470588236e-06,
+      "loss": 3.8429,
+      "step": 2015
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.386029411764706e-06,
+      "loss": 3.8384,
+      "step": 2016
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3841911764705884e-06,
+      "loss": 3.8098,
+      "step": 2017
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3823529411764708e-06,
+      "loss": 3.8119,
+      "step": 2018
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3805147058823532e-06,
+      "loss": 3.8167,
+      "step": 2019
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3786764705882356e-06,
+      "loss": 3.7727,
+      "step": 2020
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3768382352941178e-06,
+      "loss": 3.8718,
+      "step": 2021
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3750000000000002e-06,
+      "loss": 3.798,
+      "step": 2022
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3731617647058826e-06,
+      "loss": 3.7045,
+      "step": 2023
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.371323529411765e-06,
+      "loss": 3.8145,
+      "step": 2024
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3694852941176474e-06,
+      "loss": 3.7804,
+      "step": 2025
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3676470588235298e-06,
+      "loss": 3.9408,
+      "step": 2026
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3658088235294117e-06,
+      "loss": 3.7253,
+      "step": 2027
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3639705882352941e-06,
+      "loss": 3.7543,
+      "step": 2028
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3621323529411765e-06,
+      "loss": 3.8434,
+      "step": 2029
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.360294117647059e-06,
+      "loss": 3.8461,
+      "step": 2030
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.358455882352941e-06,
+      "loss": 3.7705,
+      "step": 2031
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3566176470588235e-06,
+      "loss": 3.9134,
+      "step": 2032
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3547794117647059e-06,
+      "loss": 3.8107,
+      "step": 2033
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3529411764705883e-06,
+      "loss": 3.8778,
+      "step": 2034
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3511029411764707e-06,
+      "loss": 3.6885,
+      "step": 2035
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.349264705882353e-06,
+      "loss": 3.8994,
+      "step": 2036
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3474264705882353e-06,
+      "loss": 3.8841,
+      "step": 2037
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3455882352941177e-06,
+      "loss": 3.7173,
+      "step": 2038
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.34375e-06,
+      "loss": 3.8302,
+      "step": 2039
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3419117647058824e-06,
+      "loss": 3.9526,
+      "step": 2040
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3400735294117648e-06,
+      "loss": 3.7484,
+      "step": 2041
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3382352941176472e-06,
+      "loss": 3.8131,
+      "step": 2042
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3363970588235294e-06,
+      "loss": 3.8062,
+      "step": 2043
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3345588235294118e-06,
+      "loss": 3.7704,
+      "step": 2044
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3327205882352942e-06,
+      "loss": 3.7688,
+      "step": 2045
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3308823529411766e-06,
+      "loss": 3.6819,
+      "step": 2046
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.329044117647059e-06,
+      "loss": 3.7821,
+      "step": 2047
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3272058823529412e-06,
+      "loss": 3.8495,
+      "step": 2048
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3253676470588236e-06,
+      "loss": 3.7767,
+      "step": 2049
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.323529411764706e-06,
+      "loss": 3.8434,
+      "step": 2050
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3216911764705884e-06,
+      "loss": 3.714,
+      "step": 2051
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.3198529411764708e-06,
+      "loss": 3.7876,
+      "step": 2052
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.3180147058823532e-06,
+      "loss": 3.7688,
+      "step": 2053
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.3161764705882353e-06,
+      "loss": 3.8039,
+      "step": 2054
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.3143382352941177e-06,
+      "loss": 3.7814,
+      "step": 2055
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.3125000000000001e-06,
+      "loss": 3.8708,
+      "step": 2056
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.3106617647058825e-06,
+      "loss": 3.7506,
+      "step": 2057
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.308823529411765e-06,
+      "loss": 3.8229,
+      "step": 2058
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.3069852941176473e-06,
+      "loss": 3.84,
+      "step": 2059
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.3051470588235295e-06,
+      "loss": 3.8934,
+      "step": 2060
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.303308823529412e-06,
+      "loss": 3.7734,
+      "step": 2061
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.3014705882352943e-06,
+      "loss": 3.7292,
+      "step": 2062
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2996323529411767e-06,
+      "loss": 3.7871,
+      "step": 2063
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.297794117647059e-06,
+      "loss": 3.7663,
+      "step": 2064
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2959558823529415e-06,
+      "loss": 3.706,
+      "step": 2065
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2941176470588237e-06,
+      "loss": 3.8397,
+      "step": 2066
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.292279411764706e-06,
+      "loss": 3.8575,
+      "step": 2067
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2904411764705885e-06,
+      "loss": 3.6974,
+      "step": 2068
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2886029411764706e-06,
+      "loss": 3.8456,
+      "step": 2069
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2867647058823528e-06,
+      "loss": 3.8241,
+      "step": 2070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2849264705882352e-06,
+      "loss": 3.7675,
+      "step": 2071
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2830882352941176e-06,
+      "loss": 3.6908,
+      "step": 2072
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.28125e-06,
+      "loss": 3.8484,
+      "step": 2073
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2794117647058824e-06,
+      "loss": 3.7201,
+      "step": 2074
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2775735294117648e-06,
+      "loss": 3.7842,
+      "step": 2075
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.275735294117647e-06,
+      "loss": 3.702,
+      "step": 2076
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2738970588235294e-06,
+      "loss": 3.781,
+      "step": 2077
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2720588235294118e-06,
+      "loss": 3.8236,
+      "step": 2078
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2702205882352942e-06,
+      "loss": 3.8398,
+      "step": 2079
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2683823529411766e-06,
+      "loss": 3.8886,
+      "step": 2080
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.266544117647059e-06,
+      "loss": 3.6874,
+      "step": 2081
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2647058823529412e-06,
+      "loss": 3.7717,
+      "step": 2082
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2628676470588235e-06,
+      "loss": 3.7724,
+      "step": 2083
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.261029411764706e-06,
+      "loss": 3.7912,
+      "step": 2084
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2591911764705883e-06,
+      "loss": 3.8082,
+      "step": 2085
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2573529411764707e-06,
+      "loss": 3.7173,
+      "step": 2086
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2555147058823531e-06,
+      "loss": 3.761,
+      "step": 2087
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2536764705882353e-06,
+      "loss": 3.7925,
+      "step": 2088
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2518382352941177e-06,
+      "loss": 3.7401,
+      "step": 2089
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.25e-06,
+      "loss": 3.8769,
+      "step": 2090
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2481617647058825e-06,
+      "loss": 3.9441,
+      "step": 2091
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2463235294117649e-06,
+      "loss": 3.7706,
+      "step": 2092
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.244485294117647e-06,
+      "loss": 3.7737,
+      "step": 2093
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2426470588235295e-06,
+      "loss": 3.7552,
+      "step": 2094
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2408088235294119e-06,
+      "loss": 3.7735,
+      "step": 2095
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2389705882352943e-06,
+      "loss": 3.8378,
+      "step": 2096
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2371323529411767e-06,
+      "loss": 3.6474,
+      "step": 2097
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.235294117647059e-06,
+      "loss": 3.8589,
+      "step": 2098
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2334558823529412e-06,
+      "loss": 3.6263,
+      "step": 2099
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2316176470588236e-06,
+      "loss": 3.8027,
+      "step": 2100
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2297794117647058e-06,
+      "loss": 3.796,
+      "step": 2101
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2279411764705882e-06,
+      "loss": 3.7207,
+      "step": 2102
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2261029411764706e-06,
+      "loss": 3.8075,
+      "step": 2103
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.224264705882353e-06,
+      "loss": 3.8761,
+      "step": 2104
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2224264705882354e-06,
+      "loss": 3.7404,
+      "step": 2105
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2205882352941178e-06,
+      "loss": 3.8922,
+      "step": 2106
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.21875e-06,
+      "loss": 3.8768,
+      "step": 2107
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.2169117647058824e-06,
+      "loss": 3.6092,
+      "step": 2108
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.2150735294117648e-06,
+      "loss": 3.6752,
+      "step": 2109
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.2132352941176472e-06,
+      "loss": 3.8158,
+      "step": 2110
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.2113970588235296e-06,
+      "loss": 3.7398,
+      "step": 2111
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.209558823529412e-06,
+      "loss": 3.8378,
+      "step": 2112
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.2077205882352941e-06,
+      "loss": 3.708,
+      "step": 2113
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.2058823529411765e-06,
+      "loss": 3.7681,
+      "step": 2114
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.204044117647059e-06,
+      "loss": 3.8254,
+      "step": 2115
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.2022058823529413e-06,
+      "loss": 3.7685,
+      "step": 2116
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.2003676470588237e-06,
+      "loss": 3.6942,
+      "step": 2117
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1985294117647061e-06,
+      "loss": 3.92,
+      "step": 2118
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1966911764705883e-06,
+      "loss": 3.694,
+      "step": 2119
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1948529411764707e-06,
+      "loss": 3.7566,
+      "step": 2120
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.193014705882353e-06,
+      "loss": 3.7929,
+      "step": 2121
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1911764705882353e-06,
+      "loss": 3.809,
+      "step": 2122
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1893382352941177e-06,
+      "loss": 3.6985,
+      "step": 2123
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1875e-06,
+      "loss": 3.7969,
+      "step": 2124
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1856617647058825e-06,
+      "loss": 3.8082,
+      "step": 2125
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1838235294117649e-06,
+      "loss": 3.8358,
+      "step": 2126
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.181985294117647e-06,
+      "loss": 3.7409,
+      "step": 2127
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1801470588235294e-06,
+      "loss": 3.6596,
+      "step": 2128
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1783088235294118e-06,
+      "loss": 3.7785,
+      "step": 2129
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 3.814,
+      "step": 2130
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1746323529411766e-06,
+      "loss": 3.866,
+      "step": 2131
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.172794117647059e-06,
+      "loss": 3.7406,
+      "step": 2132
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1709558823529412e-06,
+      "loss": 3.6683,
+      "step": 2133
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1691176470588236e-06,
+      "loss": 3.6596,
+      "step": 2134
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.167279411764706e-06,
+      "loss": 3.7248,
+      "step": 2135
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1654411764705884e-06,
+      "loss": 3.8138,
+      "step": 2136
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1636029411764708e-06,
+      "loss": 3.8308,
+      "step": 2137
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.161764705882353e-06,
+      "loss": 3.8092,
+      "step": 2138
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1599264705882354e-06,
+      "loss": 3.7331,
+      "step": 2139
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1580882352941178e-06,
+      "loss": 3.7359,
+      "step": 2140
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1562500000000002e-06,
+      "loss": 3.8491,
+      "step": 2141
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1544117647058825e-06,
+      "loss": 3.8642,
+      "step": 2142
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.152573529411765e-06,
+      "loss": 3.6513,
+      "step": 2143
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1507352941176471e-06,
+      "loss": 3.9139,
+      "step": 2144
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1488970588235295e-06,
+      "loss": 3.6864,
+      "step": 2145
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1470588235294117e-06,
+      "loss": 3.7701,
+      "step": 2146
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.145220588235294e-06,
+      "loss": 3.689,
+      "step": 2147
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1433823529411765e-06,
+      "loss": 3.7826,
+      "step": 2148
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1415441176470589e-06,
+      "loss": 3.8122,
+      "step": 2149
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1397058823529413e-06,
+      "loss": 3.7883,
+      "step": 2150
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1378676470588237e-06,
+      "loss": 3.8785,
+      "step": 2151
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1360294117647059e-06,
+      "loss": 3.9005,
+      "step": 2152
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1341911764705883e-06,
+      "loss": 3.869,
+      "step": 2153
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1323529411764707e-06,
+      "loss": 3.88,
+      "step": 2154
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.130514705882353e-06,
+      "loss": 3.7647,
+      "step": 2155
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1286764705882354e-06,
+      "loss": 3.7608,
+      "step": 2156
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1268382352941178e-06,
+      "loss": 3.8569,
+      "step": 2157
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.125e-06,
+      "loss": 3.8418,
+      "step": 2158
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1231617647058824e-06,
+      "loss": 3.8406,
+      "step": 2159
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.1213235294117648e-06,
+      "loss": 3.7298,
+      "step": 2160
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.1194852941176472e-06,
+      "loss": 3.7855,
+      "step": 2161
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.1176470588235296e-06,
+      "loss": 3.7927,
+      "step": 2162
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.115808823529412e-06,
+      "loss": 3.6819,
+      "step": 2163
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.1139705882352942e-06,
+      "loss": 3.8876,
+      "step": 2164
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.1121323529411766e-06,
+      "loss": 3.9123,
+      "step": 2165
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.1102941176470588e-06,
+      "loss": 3.7854,
+      "step": 2166
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.1084558823529412e-06,
+      "loss": 3.883,
+      "step": 2167
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.1066176470588236e-06,
+      "loss": 3.71,
+      "step": 2168
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.104779411764706e-06,
+      "loss": 3.7341,
+      "step": 2169
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.1029411764705884e-06,
+      "loss": 3.7001,
+      "step": 2170
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.1011029411764707e-06,
+      "loss": 3.6824,
+      "step": 2171
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.099264705882353e-06,
+      "loss": 3.7421,
+      "step": 2172
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.0974264705882353e-06,
+      "loss": 3.7325,
+      "step": 2173
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0955882352941177e-06,
+      "loss": 3.851,
+      "step": 2174
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0937500000000001e-06,
+      "loss": 3.772,
+      "step": 2175
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0919117647058825e-06,
+      "loss": 3.8535,
+      "step": 2176
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.090073529411765e-06,
+      "loss": 3.8309,
+      "step": 2177
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.088235294117647e-06,
+      "loss": 3.6384,
+      "step": 2178
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0863970588235295e-06,
+      "loss": 3.6046,
+      "step": 2179
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0845588235294119e-06,
+      "loss": 3.8592,
+      "step": 2180
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0827205882352943e-06,
+      "loss": 3.6961,
+      "step": 2181
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0808823529411767e-06,
+      "loss": 3.7922,
+      "step": 2182
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.079044117647059e-06,
+      "loss": 3.7968,
+      "step": 2183
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0772058823529413e-06,
+      "loss": 3.8131,
+      "step": 2184
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0753676470588236e-06,
+      "loss": 3.801,
+      "step": 2185
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0735294117647058e-06,
+      "loss": 3.6458,
+      "step": 2186
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0716911764705882e-06,
+      "loss": 3.6855,
+      "step": 2187
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0698529411764706e-06,
+      "loss": 3.8189,
+      "step": 2188
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.068014705882353e-06,
+      "loss": 3.7964,
+      "step": 2189
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0661764705882354e-06,
+      "loss": 3.6956,
+      "step": 2190
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0643382352941176e-06,
+      "loss": 3.8703,
+      "step": 2191
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0625e-06,
+      "loss": 3.7828,
+      "step": 2192
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0606617647058824e-06,
+      "loss": 3.6857,
+      "step": 2193
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0588235294117648e-06,
+      "loss": 3.7364,
+      "step": 2194
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0569852941176472e-06,
+      "loss": 3.8068,
+      "step": 2195
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0551470588235296e-06,
+      "loss": 3.8961,
+      "step": 2196
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0533088235294118e-06,
+      "loss": 3.7794,
+      "step": 2197
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.0514705882352942e-06,
+      "loss": 3.7591,
+      "step": 2198
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.0496323529411765e-06,
+      "loss": 3.7285,
+      "step": 2199
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.047794117647059e-06,
+      "loss": 3.8174,
+      "step": 2200
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0459558823529413e-06,
+      "loss": 3.8515,
+      "step": 2201
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0441176470588237e-06,
+      "loss": 3.7553,
+      "step": 2202
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.042279411764706e-06,
+      "loss": 3.5999,
+      "step": 2203
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.0404411764705883e-06,
+      "loss": 3.7718,
+      "step": 2204
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.0386029411764707e-06,
+      "loss": 3.6309,
+      "step": 2205
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.036764705882353e-06,
+      "loss": 3.7269,
+      "step": 2206
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0349264705882353e-06,
+      "loss": 3.8262,
+      "step": 2207
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0330882352941177e-06,
+      "loss": 3.6814,
+      "step": 2208
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.03125e-06,
+      "loss": 3.8087,
+      "step": 2209
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0294117647058825e-06,
+      "loss": 3.6475,
+      "step": 2210
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0275735294117647e-06,
+      "loss": 3.8227,
+      "step": 2211
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.025735294117647e-06,
+      "loss": 3.757,
+      "step": 2212
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0238970588235294e-06,
+      "loss": 3.7494,
+      "step": 2213
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0220588235294118e-06,
+      "loss": 3.7782,
+      "step": 2214
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0202205882352942e-06,
+      "loss": 3.7041,
+      "step": 2215
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0183823529411766e-06,
+      "loss": 3.8817,
+      "step": 2216
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.15675675675675677,
+      "eval_loss": 3.7593867778778076,
+      "eval_runtime": 144.6016,
+      "eval_samples_per_second": 2.559,
+      "eval_steps_per_second": 0.643,
+      "step": 2216
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0165441176470588e-06,
+      "loss": 3.7311,
+      "step": 2217
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.0147058823529412e-06,
+      "loss": 3.8138,
+      "step": 2218
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.0128676470588236e-06,
+      "loss": 3.6592,
+      "step": 2219
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.011029411764706e-06,
+      "loss": 3.8356,
+      "step": 2220
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.0091911764705884e-06,
+      "loss": 3.7156,
+      "step": 2221
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.0073529411764708e-06,
+      "loss": 3.7526,
+      "step": 2222
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.005514705882353e-06,
+      "loss": 3.7126,
+      "step": 2223
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.0036764705882354e-06,
+      "loss": 3.723,
+      "step": 2224
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.0018382352941178e-06,
+      "loss": 3.8471,
+      "step": 2225
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.869,
+      "step": 2226
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.981617647058826e-07,
+      "loss": 3.7312,
+      "step": 2227
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.96323529411765e-07,
+      "loss": 3.7501,
+      "step": 2228
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.944852941176471e-07,
+      "loss": 3.7984,
+      "step": 2229
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.926470588235295e-07,
+      "loss": 3.793,
+      "step": 2230
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.908088235294117e-07,
+      "loss": 3.8679,
+      "step": 2231
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.889705882352941e-07,
+      "loss": 3.7919,
+      "step": 2232
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.871323529411765e-07,
+      "loss": 3.7891,
+      "step": 2233
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.85294117647059e-07,
+      "loss": 3.841,
+      "step": 2234
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.834558823529413e-07,
+      "loss": 3.8562,
+      "step": 2235
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.816176470588237e-07,
+      "loss": 3.9092,
+      "step": 2236
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.797794117647059e-07,
+      "loss": 3.9066,
+      "step": 2237
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.779411764705883e-07,
+      "loss": 3.8177,
+      "step": 2238
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.761029411764707e-07,
+      "loss": 3.8291,
+      "step": 2239
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.74264705882353e-07,
+      "loss": 3.7232,
+      "step": 2240
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.724264705882355e-07,
+      "loss": 3.7228,
+      "step": 2241
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.705882352941176e-07,
+      "loss": 3.8503,
+      "step": 2242
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.6875e-07,
+      "loss": 3.7716,
+      "step": 2243
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.669117647058824e-07,
+      "loss": 3.8207,
+      "step": 2244
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.650735294117648e-07,
+      "loss": 3.7668,
+      "step": 2245
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.632352941176472e-07,
+      "loss": 3.7285,
+      "step": 2246
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.613970588235296e-07,
+      "loss": 4.0027,
+      "step": 2247
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.595588235294118e-07,
+      "loss": 3.7481,
+      "step": 2248
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.577205882352942e-07,
+      "loss": 3.9156,
+      "step": 2249
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.558823529411764e-07,
+      "loss": 3.6292,
+      "step": 2250
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.540441176470588e-07,
+      "loss": 3.7208,
+      "step": 2251
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.522058823529412e-07,
+      "loss": 3.825,
+      "step": 2252
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.503676470588236e-07,
+      "loss": 3.8504,
+      "step": 2253
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.48529411764706e-07,
+      "loss": 3.7816,
+      "step": 2254
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.466911764705883e-07,
+      "loss": 3.7958,
+      "step": 2255
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.448529411764707e-07,
+      "loss": 3.7548,
+      "step": 2256
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.43014705882353e-07,
+      "loss": 3.6554,
+      "step": 2257
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.411764705882353e-07,
+      "loss": 3.8942,
+      "step": 2258
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.393382352941177e-07,
+      "loss": 3.647,
+      "step": 2259
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.375000000000001e-07,
+      "loss": 3.7699,
+      "step": 2260
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.356617647058824e-07,
+      "loss": 3.7737,
+      "step": 2261
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.338235294117648e-07,
+      "loss": 3.7753,
+      "step": 2262
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.319852941176471e-07,
+      "loss": 3.8769,
+      "step": 2263
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.301470588235295e-07,
+      "loss": 3.8848,
+      "step": 2264
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.283088235294119e-07,
+      "loss": 3.748,
+      "step": 2265
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.264705882352942e-07,
+      "loss": 3.8326,
+      "step": 2266
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.246323529411766e-07,
+      "loss": 3.8675,
+      "step": 2267
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.22794117647059e-07,
+      "loss": 3.8514,
+      "step": 2268
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.209558823529413e-07,
+      "loss": 3.7658,
+      "step": 2269
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.191176470588237e-07,
+      "loss": 3.7384,
+      "step": 2270
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.172794117647058e-07,
+      "loss": 3.7817,
+      "step": 2271
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.154411764705882e-07,
+      "loss": 3.76,
+      "step": 2272
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.136029411764706e-07,
+      "loss": 3.8039,
+      "step": 2273
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.117647058823529e-07,
+      "loss": 3.7519,
+      "step": 2274
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.099264705882353e-07,
+      "loss": 3.8345,
+      "step": 2275
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.080882352941177e-07,
+      "loss": 3.8411,
+      "step": 2276
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.0625e-07,
+      "loss": 3.6126,
+      "step": 2277
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.044117647058824e-07,
+      "loss": 3.851,
+      "step": 2278
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 9.025735294117648e-07,
+      "loss": 3.7382,
+      "step": 2279
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 9.007352941176471e-07,
+      "loss": 3.726,
+      "step": 2280
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.988970588235295e-07,
+      "loss": 3.6642,
+      "step": 2281
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.970588235294119e-07,
+      "loss": 3.7449,
+      "step": 2282
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.952205882352942e-07,
+      "loss": 3.6771,
+      "step": 2283
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.933823529411766e-07,
+      "loss": 3.749,
+      "step": 2284
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.91544117647059e-07,
+      "loss": 3.7423,
+      "step": 2285
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.897058823529412e-07,
+      "loss": 3.8984,
+      "step": 2286
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.878676470588236e-07,
+      "loss": 3.8241,
+      "step": 2287
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.86029411764706e-07,
+      "loss": 3.7171,
+      "step": 2288
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.841911764705883e-07,
+      "loss": 3.7772,
+      "step": 2289
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.823529411764707e-07,
+      "loss": 3.7561,
+      "step": 2290
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.805147058823531e-07,
+      "loss": 3.8011,
+      "step": 2291
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.786764705882353e-07,
+      "loss": 3.7904,
+      "step": 2292
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.768382352941177e-07,
+      "loss": 3.7924,
+      "step": 2293
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.75e-07,
+      "loss": 3.6614,
+      "step": 2294
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.731617647058824e-07,
+      "loss": 3.8004,
+      "step": 2295
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.713235294117648e-07,
+      "loss": 3.7227,
+      "step": 2296
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.694852941176471e-07,
+      "loss": 3.8628,
+      "step": 2297
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.676470588235295e-07,
+      "loss": 3.84,
+      "step": 2298
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.658088235294119e-07,
+      "loss": 3.8566,
+      "step": 2299
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.639705882352941e-07,
+      "loss": 3.8929,
+      "step": 2300
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.621323529411765e-07,
+      "loss": 3.8101,
+      "step": 2301
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.602941176470589e-07,
+      "loss": 3.7256,
+      "step": 2302
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.584558823529412e-07,
+      "loss": 3.9177,
+      "step": 2303
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.566176470588236e-07,
+      "loss": 3.8506,
+      "step": 2304
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.54779411764706e-07,
+      "loss": 3.8089,
+      "step": 2305
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.529411764705883e-07,
+      "loss": 3.8267,
+      "step": 2306
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.511029411764707e-07,
+      "loss": 3.8108,
+      "step": 2307
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.49264705882353e-07,
+      "loss": 3.8277,
+      "step": 2308
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.474264705882354e-07,
+      "loss": 3.7748,
+      "step": 2309
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.455882352941178e-07,
+      "loss": 3.7916,
+      "step": 2310
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.437500000000001e-07,
+      "loss": 3.7391,
+      "step": 2311
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.419117647058825e-07,
+      "loss": 3.8705,
+      "step": 2312
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.400735294117649e-07,
+      "loss": 3.7975,
+      "step": 2313
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.38235294117647e-07,
+      "loss": 3.6619,
+      "step": 2314
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.363970588235294e-07,
+      "loss": 3.7262,
+      "step": 2315
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.345588235294117e-07,
+      "loss": 3.7448,
+      "step": 2316
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.327205882352941e-07,
+      "loss": 3.6764,
+      "step": 2317
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.308823529411765e-07,
+      "loss": 3.8003,
+      "step": 2318
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.290441176470588e-07,
+      "loss": 3.658,
+      "step": 2319
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.272058823529412e-07,
+      "loss": 3.8074,
+      "step": 2320
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.253676470588236e-07,
+      "loss": 3.7665,
+      "step": 2321
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.235294117647059e-07,
+      "loss": 3.8429,
+      "step": 2322
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.216911764705883e-07,
+      "loss": 3.891,
+      "step": 2323
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.198529411764707e-07,
+      "loss": 3.8151,
+      "step": 2324
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.18014705882353e-07,
+      "loss": 3.8866,
+      "step": 2325
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.161764705882354e-07,
+      "loss": 3.7529,
+      "step": 2326
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.143382352941178e-07,
+      "loss": 3.8941,
+      "step": 2327
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.125000000000001e-07,
+      "loss": 3.8118,
+      "step": 2328
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.106617647058825e-07,
+      "loss": 3.7765,
+      "step": 2329
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.088235294117648e-07,
+      "loss": 3.8259,
+      "step": 2330
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.069852941176471e-07,
+      "loss": 3.8988,
+      "step": 2331
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.051470588235295e-07,
+      "loss": 3.7111,
+      "step": 2332
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.033088235294119e-07,
+      "loss": 3.7895,
+      "step": 2333
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.014705882352942e-07,
+      "loss": 3.8168,
+      "step": 2334
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.996323529411765e-07,
+      "loss": 3.6895,
+      "step": 2335
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.977941176470588e-07,
+      "loss": 3.705,
+      "step": 2336
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.959558823529412e-07,
+      "loss": 3.7424,
+      "step": 2337
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.941176470588236e-07,
+      "loss": 3.6825,
+      "step": 2338
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.922794117647059e-07,
+      "loss": 3.6312,
+      "step": 2339
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.904411764705883e-07,
+      "loss": 3.6848,
+      "step": 2340
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.886029411764707e-07,
+      "loss": 3.8568,
+      "step": 2341
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.86764705882353e-07,
+      "loss": 3.7696,
+      "step": 2342
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.849264705882354e-07,
+      "loss": 3.7791,
+      "step": 2343
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.830882352941177e-07,
+      "loss": 3.7782,
+      "step": 2344
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.8125e-07,
+      "loss": 3.6817,
+      "step": 2345
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.794117647058824e-07,
+      "loss": 3.7526,
+      "step": 2346
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.775735294117648e-07,
+      "loss": 3.6692,
+      "step": 2347
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.757352941176471e-07,
+      "loss": 3.7771,
+      "step": 2348
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.738970588235295e-07,
+      "loss": 3.7641,
+      "step": 2349
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.720588235294119e-07,
+      "loss": 3.7224,
+      "step": 2350
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.702205882352942e-07,
+      "loss": 3.9045,
+      "step": 2351
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.683823529411766e-07,
+      "loss": 3.6777,
+      "step": 2352
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.66544117647059e-07,
+      "loss": 3.7719,
+      "step": 2353
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.647058823529413e-07,
+      "loss": 3.7168,
+      "step": 2354
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.628676470588237e-07,
+      "loss": 3.8028,
+      "step": 2355
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.610294117647059e-07,
+      "loss": 3.8534,
+      "step": 2356
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.591911764705883e-07,
+      "loss": 3.6416,
+      "step": 2357
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.573529411764706e-07,
+      "loss": 3.7058,
+      "step": 2358
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.555147058823529e-07,
+      "loss": 3.9277,
+      "step": 2359
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.536764705882353e-07,
+      "loss": 3.8308,
+      "step": 2360
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.518382352941176e-07,
+      "loss": 3.7528,
+      "step": 2361
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.5e-07,
+      "loss": 3.6644,
+      "step": 2362
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.481617647058824e-07,
+      "loss": 3.8808,
+      "step": 2363
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.463235294117647e-07,
+      "loss": 3.8774,
+      "step": 2364
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.444852941176471e-07,
+      "loss": 3.8571,
+      "step": 2365
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.426470588235295e-07,
+      "loss": 3.6985,
+      "step": 2366
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.408088235294118e-07,
+      "loss": 3.7546,
+      "step": 2367
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.389705882352942e-07,
+      "loss": 3.7884,
+      "step": 2368
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.371323529411766e-07,
+      "loss": 3.8495,
+      "step": 2369
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 3.8278,
+      "step": 2370
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.334558823529413e-07,
+      "loss": 3.8158,
+      "step": 2371
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.316176470588237e-07,
+      "loss": 3.7087,
+      "step": 2372
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.297794117647059e-07,
+      "loss": 3.7061,
+      "step": 2373
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.279411764705883e-07,
+      "loss": 3.7877,
+      "step": 2374
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.261029411764707e-07,
+      "loss": 3.7624,
+      "step": 2375
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.24264705882353e-07,
+      "loss": 3.6838,
+      "step": 2376
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.224264705882353e-07,
+      "loss": 3.6878,
+      "step": 2377
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.205882352941176e-07,
+      "loss": 3.6837,
+      "step": 2378
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.1875e-07,
+      "loss": 3.6522,
+      "step": 2379
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.169117647058824e-07,
+      "loss": 3.8438,
+      "step": 2380
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.150735294117647e-07,
+      "loss": 3.7046,
+      "step": 2381
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.132352941176471e-07,
+      "loss": 3.8354,
+      "step": 2382
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.113970588235295e-07,
+      "loss": 3.7745,
+      "step": 2383
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 7.095588235294118e-07,
+      "loss": 3.879,
+      "step": 2384
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 7.077205882352942e-07,
+      "loss": 3.8688,
+      "step": 2385
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 7.058823529411766e-07,
+      "loss": 3.9224,
+      "step": 2386
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 7.040441176470588e-07,
+      "loss": 3.7479,
+      "step": 2387
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 7.022058823529412e-07,
+      "loss": 3.7001,
+      "step": 2388
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 7.003676470588236e-07,
+      "loss": 3.8452,
+      "step": 2389
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.985294117647059e-07,
+      "loss": 3.7145,
+      "step": 2390
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.966911764705883e-07,
+      "loss": 3.6996,
+      "step": 2391
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.948529411764707e-07,
+      "loss": 3.6727,
+      "step": 2392
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.93014705882353e-07,
+      "loss": 3.7747,
+      "step": 2393
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.911764705882354e-07,
+      "loss": 3.7433,
+      "step": 2394
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.893382352941178e-07,
+      "loss": 3.8303,
+      "step": 2395
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.875000000000001e-07,
+      "loss": 3.6821,
+      "step": 2396
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.856617647058825e-07,
+      "loss": 3.648,
+      "step": 2397
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.838235294117649e-07,
+      "loss": 3.7825,
+      "step": 2398
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.819852941176471e-07,
+      "loss": 3.7121,
+      "step": 2399
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.801470588235295e-07,
+      "loss": 3.7382,
+      "step": 2400
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.783088235294117e-07,
+      "loss": 3.7332,
+      "step": 2401
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.764705882352941e-07,
+      "loss": 3.8153,
+      "step": 2402
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.746323529411765e-07,
+      "loss": 3.9339,
+      "step": 2403
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.727941176470588e-07,
+      "loss": 3.6879,
+      "step": 2404
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.709558823529412e-07,
+      "loss": 3.7747,
+      "step": 2405
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.691176470588236e-07,
+      "loss": 3.795,
+      "step": 2406
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.672794117647059e-07,
+      "loss": 3.8234,
+      "step": 2407
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.654411764705883e-07,
+      "loss": 3.8175,
+      "step": 2408
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.636029411764706e-07,
+      "loss": 3.7169,
+      "step": 2409
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.61764705882353e-07,
+      "loss": 3.7274,
+      "step": 2410
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.599264705882354e-07,
+      "loss": 3.7801,
+      "step": 2411
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.580882352941177e-07,
+      "loss": 3.8364,
+      "step": 2412
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.562500000000001e-07,
+      "loss": 3.872,
+      "step": 2413
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.544117647058825e-07,
+      "loss": 3.8348,
+      "step": 2414
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.525735294117648e-07,
+      "loss": 3.8543,
+      "step": 2415
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.507352941176472e-07,
+      "loss": 3.8511,
+      "step": 2416
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.488970588235295e-07,
+      "loss": 3.82,
+      "step": 2417
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.470588235294118e-07,
+      "loss": 3.8348,
+      "step": 2418
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.452205882352942e-07,
+      "loss": 3.6814,
+      "step": 2419
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.433823529411764e-07,
+      "loss": 3.7721,
+      "step": 2420
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.415441176470588e-07,
+      "loss": 3.6613,
+      "step": 2421
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.397058823529412e-07,
+      "loss": 3.7546,
+      "step": 2422
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.378676470588235e-07,
+      "loss": 3.8473,
+      "step": 2423
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.360294117647059e-07,
+      "loss": 3.7451,
+      "step": 2424
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.341911764705883e-07,
+      "loss": 3.6373,
+      "step": 2425
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.323529411764706e-07,
+      "loss": 3.7275,
+      "step": 2426
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.30514705882353e-07,
+      "loss": 3.743,
+      "step": 2427
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.286764705882354e-07,
+      "loss": 3.8529,
+      "step": 2428
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.268382352941177e-07,
+      "loss": 3.7222,
+      "step": 2429
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.25e-07,
+      "loss": 3.7621,
+      "step": 2430
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.231617647058824e-07,
+      "loss": 3.755,
+      "step": 2431
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.213235294117647e-07,
+      "loss": 3.8091,
+      "step": 2432
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.194852941176471e-07,
+      "loss": 3.8066,
+      "step": 2433
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.176470588235295e-07,
+      "loss": 3.6961,
+      "step": 2434
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.158088235294118e-07,
+      "loss": 3.878,
+      "step": 2435
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.139705882352941e-07,
+      "loss": 3.7094,
+      "step": 2436
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.121323529411765e-07,
+      "loss": 3.8669,
+      "step": 2437
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.102941176470589e-07,
+      "loss": 3.7921,
+      "step": 2438
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.084558823529412e-07,
+      "loss": 3.7094,
+      "step": 2439
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.066176470588236e-07,
+      "loss": 3.622,
+      "step": 2440
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.04779411764706e-07,
+      "loss": 3.7612,
+      "step": 2441
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.029411764705883e-07,
+      "loss": 3.8429,
+      "step": 2442
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.011029411764707e-07,
+      "loss": 3.7919,
+      "step": 2443
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.992647058823531e-07,
+      "loss": 3.7504,
+      "step": 2444
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.974264705882353e-07,
+      "loss": 3.845,
+      "step": 2445
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.955882352941176e-07,
+      "loss": 3.9041,
+      "step": 2446
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.9375e-07,
+      "loss": 3.7744,
+      "step": 2447
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.919117647058824e-07,
+      "loss": 3.7116,
+      "step": 2448
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.900735294117647e-07,
+      "loss": 3.8122,
+      "step": 2449
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 3.8511,
+      "step": 2450
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.863970588235295e-07,
+      "loss": 3.7406,
+      "step": 2451
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.845588235294118e-07,
+      "loss": 3.9195,
+      "step": 2452
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.827205882352942e-07,
+      "loss": 3.9183,
+      "step": 2453
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.808823529411765e-07,
+      "loss": 3.8855,
+      "step": 2454
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.790441176470589e-07,
+      "loss": 3.7123,
+      "step": 2455
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.772058823529413e-07,
+      "loss": 3.7221,
+      "step": 2456
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.753676470588236e-07,
+      "loss": 3.794,
+      "step": 2457
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.735294117647059e-07,
+      "loss": 3.7516,
+      "step": 2458
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.716911764705882e-07,
+      "loss": 3.8284,
+      "step": 2459
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.698529411764706e-07,
+      "loss": 3.7762,
+      "step": 2460
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.680147058823529e-07,
+      "loss": 3.7962,
+      "step": 2461
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.661764705882353e-07,
+      "loss": 3.8181,
+      "step": 2462
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.643382352941177e-07,
+      "loss": 3.6142,
+      "step": 2463
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.625e-07,
+      "loss": 3.924,
+      "step": 2464
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.606617647058824e-07,
+      "loss": 3.7873,
+      "step": 2465
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.588235294117648e-07,
+      "loss": 3.6543,
+      "step": 2466
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.569852941176471e-07,
+      "loss": 3.696,
+      "step": 2467
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.551470588235294e-07,
+      "loss": 3.7413,
+      "step": 2468
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.533088235294118e-07,
+      "loss": 3.8391,
+      "step": 2469
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.514705882352942e-07,
+      "loss": 3.7809,
+      "step": 2470
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.496323529411765e-07,
+      "loss": 3.674,
+      "step": 2471
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.477941176470589e-07,
+      "loss": 3.7189,
+      "step": 2472
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.459558823529413e-07,
+      "loss": 3.7992,
+      "step": 2473
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.441176470588235e-07,
+      "loss": 3.7355,
+      "step": 2474
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.422794117647059e-07,
+      "loss": 3.8828,
+      "step": 2475
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.404411764705883e-07,
+      "loss": 3.6782,
+      "step": 2476
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.386029411764706e-07,
+      "loss": 3.6093,
+      "step": 2477
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.367647058823529e-07,
+      "loss": 3.691,
+      "step": 2478
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.349264705882353e-07,
+      "loss": 3.6619,
+      "step": 2479
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.330882352941177e-07,
+      "loss": 3.8944,
+      "step": 2480
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.3125e-07,
+      "loss": 3.8312,
+      "step": 2481
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.294117647058824e-07,
+      "loss": 3.7363,
+      "step": 2482
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.275735294117648e-07,
+      "loss": 3.5276,
+      "step": 2483
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.257352941176471e-07,
+      "loss": 3.8608,
+      "step": 2484
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.238970588235295e-07,
+      "loss": 3.8745,
+      "step": 2485
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.220588235294119e-07,
+      "loss": 3.7958,
+      "step": 2486
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.202205882352942e-07,
+      "loss": 3.8531,
+      "step": 2487
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.183823529411766e-07,
+      "loss": 3.7045,
+      "step": 2488
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.165441176470588e-07,
+      "loss": 3.8388,
+      "step": 2489
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.147058823529412e-07,
+      "loss": 3.8649,
+      "step": 2490
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.128676470588235e-07,
+      "loss": 3.9795,
+      "step": 2491
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.110294117647059e-07,
+      "loss": 3.6293,
+      "step": 2492
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.091911764705883e-07,
+      "loss": 3.8258,
+      "step": 2493
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.16756756756756758,
+      "eval_loss": 3.747398614883423,
+      "eval_runtime": 142.6563,
+      "eval_samples_per_second": 2.594,
+      "eval_steps_per_second": 0.652,
+      "step": 2493
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.073529411764706e-07,
+      "loss": 3.7742,
+      "step": 2494
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.05514705882353e-07,
+      "loss": 3.7896,
+      "step": 2495
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.036764705882354e-07,
+      "loss": 3.767,
+      "step": 2496
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.018382352941177e-07,
+      "loss": 3.6757,
+      "step": 2497
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.7333,
+      "step": 2498
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.981617647058825e-07,
+      "loss": 3.7456,
+      "step": 2499
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.963235294117648e-07,
+      "loss": 3.7664,
+      "step": 2500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.944852941176471e-07,
+      "loss": 3.6546,
+      "step": 2501
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.926470588235295e-07,
+      "loss": 3.7102,
+      "step": 2502
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.908088235294118e-07,
+      "loss": 3.8283,
+      "step": 2503
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.889705882352941e-07,
+      "loss": 3.7896,
+      "step": 2504
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.871323529411765e-07,
+      "loss": 3.8111,
+      "step": 2505
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.852941176470588e-07,
+      "loss": 3.6198,
+      "step": 2506
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.834558823529412e-07,
+      "loss": 3.7278,
+      "step": 2507
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.816176470588236e-07,
+      "loss": 3.7027,
+      "step": 2508
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.797794117647059e-07,
+      "loss": 3.8305,
+      "step": 2509
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.779411764705882e-07,
+      "loss": 3.7641,
+      "step": 2510
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.761029411764706e-07,
+      "loss": 3.6554,
+      "step": 2511
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.74264705882353e-07,
+      "loss": 3.7905,
+      "step": 2512
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.7242647058823533e-07,
+      "loss": 3.5754,
+      "step": 2513
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.7058823529411767e-07,
+      "loss": 3.8388,
+      "step": 2514
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.6875000000000006e-07,
+      "loss": 3.8282,
+      "step": 2515
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.669117647058824e-07,
+      "loss": 3.8828,
+      "step": 2516
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.6507352941176475e-07,
+      "loss": 3.7438,
+      "step": 2517
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.632352941176471e-07,
+      "loss": 3.8017,
+      "step": 2518
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.613970588235295e-07,
+      "loss": 3.7509,
+      "step": 2519
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5955882352941183e-07,
+      "loss": 3.8022,
+      "step": 2520
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.577205882352941e-07,
+      "loss": 3.7026,
+      "step": 2521
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5588235294117646e-07,
+      "loss": 3.7332,
+      "step": 2522
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5404411764705886e-07,
+      "loss": 3.7656,
+      "step": 2523
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.522058823529412e-07,
+      "loss": 3.8702,
+      "step": 2524
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5036764705882354e-07,
+      "loss": 3.8981,
+      "step": 2525
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.4852941176470594e-07,
+      "loss": 3.7603,
+      "step": 2526
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.466911764705883e-07,
+      "loss": 3.6499,
+      "step": 2527
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.448529411764706e-07,
+      "loss": 3.7807,
+      "step": 2528
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.43014705882353e-07,
+      "loss": 3.7718,
+      "step": 2529
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.4117647058823536e-07,
+      "loss": 3.7782,
+      "step": 2530
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.3933823529411765e-07,
+      "loss": 3.8509,
+      "step": 2531
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.375e-07,
+      "loss": 3.8103,
+      "step": 2532
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.356617647058824e-07,
+      "loss": 3.6662,
+      "step": 2533
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.3382352941176473e-07,
+      "loss": 3.6027,
+      "step": 2534
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.319852941176471e-07,
+      "loss": 3.6919,
+      "step": 2535
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.3014705882352947e-07,
+      "loss": 3.7634,
+      "step": 2536
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.283088235294118e-07,
+      "loss": 3.7289,
+      "step": 2537
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.2647058823529415e-07,
+      "loss": 3.7774,
+      "step": 2538
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.246323529411765e-07,
+      "loss": 3.7781,
+      "step": 2539
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.227941176470589e-07,
+      "loss": 3.8441,
+      "step": 2540
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.2095588235294123e-07,
+      "loss": 3.6877,
+      "step": 2541
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.191176470588235e-07,
+      "loss": 3.8197,
+      "step": 2542
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.1727941176470587e-07,
+      "loss": 3.7825,
+      "step": 2543
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.1544117647058826e-07,
+      "loss": 3.7872,
+      "step": 2544
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.136029411764706e-07,
+      "loss": 3.7787,
+      "step": 2545
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.1176470588235295e-07,
+      "loss": 3.7984,
+      "step": 2546
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.0992647058823534e-07,
+      "loss": 3.7131,
+      "step": 2547
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.080882352941177e-07,
+      "loss": 3.7869,
+      "step": 2548
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.0625000000000003e-07,
+      "loss": 3.8407,
+      "step": 2549
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.044117647058824e-07,
+      "loss": 3.8945,
+      "step": 2550
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.0257352941176477e-07,
+      "loss": 3.697,
+      "step": 2551
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.007352941176471e-07,
+      "loss": 3.7087,
+      "step": 2552
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.988970588235294e-07,
+      "loss": 3.747,
+      "step": 2553
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.970588235294118e-07,
+      "loss": 3.8015,
+      "step": 2554
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.9522058823529414e-07,
+      "loss": 3.6433,
+      "step": 2555
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.933823529411765e-07,
+      "loss": 3.8178,
+      "step": 2556
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.915441176470589e-07,
+      "loss": 3.8249,
+      "step": 2557
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.897058823529412e-07,
+      "loss": 3.8446,
+      "step": 2558
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.8786764705882356e-07,
+      "loss": 3.7414,
+      "step": 2559
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.8602941176470595e-07,
+      "loss": 3.8151,
+      "step": 2560
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.841911764705883e-07,
+      "loss": 3.688,
+      "step": 2561
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.8235294117647064e-07,
+      "loss": 3.7028,
+      "step": 2562
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.8051470588235293e-07,
+      "loss": 3.636,
+      "step": 2563
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.786764705882353e-07,
+      "loss": 3.652,
+      "step": 2564
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.7683823529411767e-07,
+      "loss": 3.7657,
+      "step": 2565
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.75e-07,
+      "loss": 3.7984,
+      "step": 2566
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.7316176470588235e-07,
+      "loss": 3.7719,
+      "step": 2567
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.7132352941176475e-07,
+      "loss": 3.7794,
+      "step": 2568
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.694852941176471e-07,
+      "loss": 3.8205,
+      "step": 2569
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.6764705882352943e-07,
+      "loss": 3.7515,
+      "step": 2570
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.6580882352941183e-07,
+      "loss": 3.7615,
+      "step": 2571
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.6397058823529417e-07,
+      "loss": 3.7461,
+      "step": 2572
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.621323529411765e-07,
+      "loss": 3.7666,
+      "step": 2573
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.602941176470588e-07,
+      "loss": 3.7117,
+      "step": 2574
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.584558823529412e-07,
+      "loss": 3.8768,
+      "step": 2575
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.5661764705882354e-07,
+      "loss": 3.722,
+      "step": 2576
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.547794117647059e-07,
+      "loss": 3.7573,
+      "step": 2577
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.529411764705883e-07,
+      "loss": 3.701,
+      "step": 2578
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.511029411764706e-07,
+      "loss": 3.7353,
+      "step": 2579
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4926470588235296e-07,
+      "loss": 3.8629,
+      "step": 2580
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4742647058823536e-07,
+      "loss": 3.8279,
+      "step": 2581
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.455882352941177e-07,
+      "loss": 3.8181,
+      "step": 2582
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4375000000000004e-07,
+      "loss": 3.8184,
+      "step": 2583
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4191176470588244e-07,
+      "loss": 3.7048,
+      "step": 2584
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4007352941176473e-07,
+      "loss": 3.7605,
+      "step": 2585
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.3823529411764707e-07,
+      "loss": 3.7882,
+      "step": 2586
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.363970588235294e-07,
+      "loss": 3.7615,
+      "step": 2587
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.345588235294118e-07,
+      "loss": 3.8021,
+      "step": 2588
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.3272058823529415e-07,
+      "loss": 3.7766,
+      "step": 2589
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.308823529411765e-07,
+      "loss": 3.883,
+      "step": 2590
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2904411764705884e-07,
+      "loss": 3.7975,
+      "step": 2591
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.2720588235294123e-07,
+      "loss": 3.7549,
+      "step": 2592
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.253676470588236e-07,
+      "loss": 3.7203,
+      "step": 2593
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.235294117647059e-07,
+      "loss": 3.6731,
+      "step": 2594
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.216911764705882e-07,
+      "loss": 3.7708,
+      "step": 2595
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.198529411764706e-07,
+      "loss": 3.8121,
+      "step": 2596
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.1801470588235295e-07,
+      "loss": 3.6584,
+      "step": 2597
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.161764705882353e-07,
+      "loss": 3.9055,
+      "step": 2598
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.143382352941177e-07,
+      "loss": 3.7132,
+      "step": 2599
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.125e-07,
+      "loss": 3.7559,
+      "step": 2600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.1066176470588237e-07,
+      "loss": 3.7324,
+      "step": 2601
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0882352941176476e-07,
+      "loss": 3.84,
+      "step": 2602
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.0698529411764705e-07,
+      "loss": 3.6149,
+      "step": 2603
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.0514705882352945e-07,
+      "loss": 3.8501,
+      "step": 2604
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.033088235294118e-07,
+      "loss": 3.8329,
+      "step": 2605
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.0147058823529413e-07,
+      "loss": 3.8633,
+      "step": 2606
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.9963235294117653e-07,
+      "loss": 3.7794,
+      "step": 2607
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.977941176470588e-07,
+      "loss": 3.7203,
+      "step": 2608
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.959558823529412e-07,
+      "loss": 3.7122,
+      "step": 2609
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.9411764705882356e-07,
+      "loss": 3.733,
+      "step": 2610
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.922794117647059e-07,
+      "loss": 3.7117,
+      "step": 2611
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.9044117647058824e-07,
+      "loss": 3.645,
+      "step": 2612
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8860294117647064e-07,
+      "loss": 3.8675,
+      "step": 2613
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8676470588235293e-07,
+      "loss": 3.6991,
+      "step": 2614
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.849264705882353e-07,
+      "loss": 3.8638,
+      "step": 2615
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8308823529411766e-07,
+      "loss": 3.6476,
+      "step": 2616
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.8125e-07,
+      "loss": 3.7259,
+      "step": 2617
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.794117647058824e-07,
+      "loss": 3.8833,
+      "step": 2618
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.775735294117647e-07,
+      "loss": 3.6329,
+      "step": 2619
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.757352941176471e-07,
+      "loss": 3.8142,
+      "step": 2620
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.7389705882352943e-07,
+      "loss": 3.7852,
+      "step": 2621
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.7205882352941177e-07,
+      "loss": 3.6707,
+      "step": 2622
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.7022058823529417e-07,
+      "loss": 3.8685,
+      "step": 2623
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6838235294117646e-07,
+      "loss": 3.8412,
+      "step": 2624
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6654411764705885e-07,
+      "loss": 3.8398,
+      "step": 2625
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.647058823529412e-07,
+      "loss": 3.8604,
+      "step": 2626
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6286764705882354e-07,
+      "loss": 3.8224,
+      "step": 2627
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.6102941176470593e-07,
+      "loss": 3.72,
+      "step": 2628
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.591911764705883e-07,
+      "loss": 3.7473,
+      "step": 2629
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.573529411764706e-07,
+      "loss": 3.7612,
+      "step": 2630
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.5551470588235296e-07,
+      "loss": 3.6752,
+      "step": 2631
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.536764705882353e-07,
+      "loss": 3.7414,
+      "step": 2632
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.518382352941177e-07,
+      "loss": 3.6937,
+      "step": 2633
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 3.8864,
+      "step": 2634
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.481617647058824e-07,
+      "loss": 3.6525,
+      "step": 2635
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.4632352941176473e-07,
+      "loss": 3.7045,
+      "step": 2636
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.4448529411764707e-07,
+      "loss": 3.7811,
+      "step": 2637
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.426470588235294e-07,
+      "loss": 3.7723,
+      "step": 2638
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.408088235294118e-07,
+      "loss": 3.9059,
+      "step": 2639
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.389705882352941e-07,
+      "loss": 3.8559,
+      "step": 2640
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.371323529411765e-07,
+      "loss": 3.798,
+      "step": 2641
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.3529411764705883e-07,
+      "loss": 3.7379,
+      "step": 2642
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.334558823529412e-07,
+      "loss": 3.7787,
+      "step": 2643
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.3161764705882355e-07,
+      "loss": 3.8253,
+      "step": 2644
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.2977941176470592e-07,
+      "loss": 3.7808,
+      "step": 2645
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.2794117647058823e-07,
+      "loss": 3.6646,
+      "step": 2646
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.261029411764706e-07,
+      "loss": 3.7457,
+      "step": 2647
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.2426470588235297e-07,
+      "loss": 3.8225,
+      "step": 2648
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.224264705882353e-07,
+      "loss": 3.7174,
+      "step": 2649
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.2058823529411768e-07,
+      "loss": 3.8852,
+      "step": 2650
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.1875e-07,
+      "loss": 3.7364,
+      "step": 2651
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.1691176470588237e-07,
+      "loss": 3.8263,
+      "step": 2652
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1507352941176473e-07,
+      "loss": 3.8587,
+      "step": 2653
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1323529411764708e-07,
+      "loss": 3.6841,
+      "step": 2654
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1139705882352945e-07,
+      "loss": 3.7906,
+      "step": 2655
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.0955882352941176e-07,
+      "loss": 3.6977,
+      "step": 2656
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.0772058823529413e-07,
+      "loss": 3.6885,
+      "step": 2657
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0588235294117647e-07,
+      "loss": 3.7529,
+      "step": 2658
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0404411764705884e-07,
+      "loss": 3.725,
+      "step": 2659
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.022058823529412e-07,
+      "loss": 3.7744,
+      "step": 2660
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.0036764705882355e-07,
+      "loss": 3.767,
+      "step": 2661
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.985294117647059e-07,
+      "loss": 3.9009,
+      "step": 2662
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.9669117647058824e-07,
+      "loss": 3.7754,
+      "step": 2663
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.948529411764706e-07,
+      "loss": 3.7756,
+      "step": 2664
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9301470588235298e-07,
+      "loss": 3.6718,
+      "step": 2665
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9117647058823532e-07,
+      "loss": 3.8339,
+      "step": 2666
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.8933823529411766e-07,
+      "loss": 3.8213,
+      "step": 2667
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.875e-07,
+      "loss": 3.653,
+      "step": 2668
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8566176470588237e-07,
+      "loss": 3.8195,
+      "step": 2669
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8382352941176472e-07,
+      "loss": 3.8466,
+      "step": 2670
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8198529411764709e-07,
+      "loss": 3.833,
+      "step": 2671
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.801470588235294e-07,
+      "loss": 3.7026,
+      "step": 2672
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7830882352941177e-07,
+      "loss": 3.7743,
+      "step": 2673
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7647058823529414e-07,
+      "loss": 3.7778,
+      "step": 2674
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7463235294117648e-07,
+      "loss": 3.8244,
+      "step": 2675
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7279411764705885e-07,
+      "loss": 3.747,
+      "step": 2676
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7095588235294122e-07,
+      "loss": 3.7837,
+      "step": 2677
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.6911764705882354e-07,
+      "loss": 3.8244,
+      "step": 2678
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.672794117647059e-07,
+      "loss": 3.7091,
+      "step": 2679
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6544117647058825e-07,
+      "loss": 3.7692,
+      "step": 2680
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6360294117647062e-07,
+      "loss": 3.6898,
+      "step": 2681
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6176470588235296e-07,
+      "loss": 3.7676,
+      "step": 2682
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.599264705882353e-07,
+      "loss": 3.7901,
+      "step": 2683
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5808823529411764e-07,
+      "loss": 3.6358,
+      "step": 2684
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5625e-07,
+      "loss": 3.7731,
+      "step": 2685
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.5441176470588238e-07,
+      "loss": 3.7159,
+      "step": 2686
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.5257352941176472e-07,
+      "loss": 3.73,
+      "step": 2687
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.5073529411764707e-07,
+      "loss": 3.9258,
+      "step": 2688
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.488970588235294e-07,
+      "loss": 3.7849,
+      "step": 2689
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4705882352941178e-07,
+      "loss": 3.7702,
+      "step": 2690
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4522058823529412e-07,
+      "loss": 3.5905,
+      "step": 2691
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4338235294117646e-07,
+      "loss": 3.8034,
+      "step": 2692
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4154411764705883e-07,
+      "loss": 3.8553,
+      "step": 2693
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.397058823529412e-07,
+      "loss": 3.7798,
+      "step": 2694
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3786764705882354e-07,
+      "loss": 3.806,
+      "step": 2695
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3602941176470589e-07,
+      "loss": 3.7418,
+      "step": 2696
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3419117647058823e-07,
+      "loss": 3.7032,
+      "step": 2697
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.323529411764706e-07,
+      "loss": 3.7004,
+      "step": 2698
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3051470588235297e-07,
+      "loss": 3.6563,
+      "step": 2699
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.286764705882353e-07,
+      "loss": 3.7654,
+      "step": 2700
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2683823529411765e-07,
+      "loss": 3.8808,
+      "step": 2701
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2500000000000002e-07,
+      "loss": 3.8361,
+      "step": 2702
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.2316176470588236e-07,
+      "loss": 3.7541,
+      "step": 2703
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.213235294117647e-07,
+      "loss": 3.8437,
+      "step": 2704
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1948529411764705e-07,
+      "loss": 3.7702,
+      "step": 2705
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1764705882352942e-07,
+      "loss": 3.8234,
+      "step": 2706
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1580882352941177e-07,
+      "loss": 3.7749,
+      "step": 2707
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1397058823529412e-07,
+      "loss": 3.7498,
+      "step": 2708
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1213235294117648e-07,
+      "loss": 3.719,
+      "step": 2709
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1029411764705884e-07,
+      "loss": 3.7922,
+      "step": 2710
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0845588235294118e-07,
+      "loss": 3.7657,
+      "step": 2711
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0661764705882354e-07,
+      "loss": 3.8531,
+      "step": 2712
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0477941176470588e-07,
+      "loss": 3.8902,
+      "step": 2713
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0294117647058824e-07,
+      "loss": 3.6529,
+      "step": 2714
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.011029411764706e-07,
+      "loss": 3.7144,
+      "step": 2715
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.926470588235295e-08,
+      "loss": 3.7093,
+      "step": 2716
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.74264705882353e-08,
+      "loss": 3.7385,
+      "step": 2717
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.558823529411766e-08,
+      "loss": 3.8658,
+      "step": 2718
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.375e-08,
+      "loss": 3.7143,
+      "step": 2719
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.191176470588236e-08,
+      "loss": 3.6873,
+      "step": 2720
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.00735294117647e-08,
+      "loss": 3.7374,
+      "step": 2721
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.823529411764707e-08,
+      "loss": 3.7236,
+      "step": 2722
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.639705882352943e-08,
+      "loss": 3.6656,
+      "step": 2723
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.455882352941177e-08,
+      "loss": 3.7009,
+      "step": 2724
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.272058823529412e-08,
+      "loss": 3.7995,
+      "step": 2725
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.088235294117648e-08,
+      "loss": 3.804,
+      "step": 2726
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.904411764705882e-08,
+      "loss": 3.7859,
+      "step": 2727
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.720588235294119e-08,
+      "loss": 3.8329,
+      "step": 2728
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.536764705882353e-08,
+      "loss": 3.8527,
+      "step": 2729
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.352941176470589e-08,
+      "loss": 3.8687,
+      "step": 2730
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.169117647058823e-08,
+      "loss": 3.8088,
+      "step": 2731
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.98529411764706e-08,
+      "loss": 3.7211,
+      "step": 2732
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.801470588235294e-08,
+      "loss": 3.826,
+      "step": 2733
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.61764705882353e-08,
+      "loss": 3.8344,
+      "step": 2734
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.433823529411765e-08,
+      "loss": 3.7468,
+      "step": 2735
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.250000000000001e-08,
+      "loss": 3.8694,
+      "step": 2736
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.066176470588235e-08,
+      "loss": 3.8458,
+      "step": 2737
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.882352941176471e-08,
+      "loss": 3.865,
+      "step": 2738
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.698529411764706e-08,
+      "loss": 3.7796,
+      "step": 2739
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.514705882352942e-08,
+      "loss": 3.8002,
+      "step": 2740
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.330882352941177e-08,
+      "loss": 3.6658,
+      "step": 2741
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.147058823529412e-08,
+      "loss": 3.777,
+      "step": 2742
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.9632352941176474e-08,
+      "loss": 3.673,
+      "step": 2743
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.779411764705883e-08,
+      "loss": 3.7164,
+      "step": 2744
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.595588235294118e-08,
+      "loss": 3.7202,
+      "step": 2745
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.4117647058823535e-08,
+      "loss": 3.7995,
+      "step": 2746
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.2279411764705884e-08,
+      "loss": 3.7512,
+      "step": 2747
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.044117647058824e-08,
+      "loss": 3.8668,
+      "step": 2748
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.8602941176470595e-08,
+      "loss": 3.875,
+      "step": 2749
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.6764705882352945e-08,
+      "loss": 3.837,
+      "step": 2750
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.49264705882353e-08,
+      "loss": 3.8107,
+      "step": 2751
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.308823529411765e-08,
+      "loss": 3.9115,
+      "step": 2752
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.1250000000000005e-08,
+      "loss": 3.869,
+      "step": 2753
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.9411764705882354e-08,
+      "loss": 3.8325,
+      "step": 2754
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.757352941176471e-08,
+      "loss": 3.824,
+      "step": 2755
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.573529411764706e-08,
+      "loss": 3.6754,
+      "step": 2756
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.3897058823529415e-08,
+      "loss": 3.8178,
+      "step": 2757
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.2058823529411767e-08,
+      "loss": 3.6439,
+      "step": 2758
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.022058823529412e-08,
+      "loss": 3.8224,
+      "step": 2759
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.8382352941176472e-08,
+      "loss": 3.6205,
+      "step": 2760
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6544117647058825e-08,
+      "loss": 3.8128,
+      "step": 2761
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.4705882352941177e-08,
+      "loss": 3.7489,
+      "step": 2762
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.286764705882353e-08,
+      "loss": 3.7814,
+      "step": 2763
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1029411764705884e-08,
+      "loss": 3.7514,
+      "step": 2764
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.191176470588236e-09,
+      "loss": 3.6821,
+      "step": 2765
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 7.3529411764705886e-09,
+      "loss": 3.8408,
+      "step": 2766
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.514705882352942e-09,
+      "loss": 3.7974,
+      "step": 2767
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.6764705882352943e-09,
+      "loss": 3.9221,
+      "step": 2768
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8382352941176471e-09,
+      "loss": 3.8255,
+      "step": 2769
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0,
+      "loss": 3.7908,
+      "step": 2770
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.15945945945945947,
+      "eval_loss": 3.7452244758605957,
+      "eval_runtime": 143.3771,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 0.649,
+      "step": 2770
+    }
+  ],
+  "max_steps": 2770,
+  "num_train_epochs": 10,
+  "total_flos": 3.006770814363167e+18,
+  "trial_name": null,
+  "trial_params": null
+}