diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13530 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 1.5025,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.2014,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 1.4268,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.4108,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 1.6487,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.3116,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 1.4292,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.6753,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 1.5403,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 1.4638,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 1.3273,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.4748,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 1.4437,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.3585,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 1.3907,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 1.3003,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 1.2613,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 1.3386,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.588235294117647e-05,
+      "loss": 1.4603,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 1.3109,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.176470588235295e-05,
+      "loss": 1.4869,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.470588235294118e-05,
+      "loss": 1.3065,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.764705882352942e-05,
+      "loss": 1.2601,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.058823529411765e-05,
+      "loss": 1.2743,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.352941176470589e-05,
+      "loss": 1.1704,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.647058823529411e-05,
+      "loss": 1.1551,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.941176470588235e-05,
+      "loss": 1.076,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.23529411764706e-05,
+      "loss": 1.3218,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.529411764705883e-05,
+      "loss": 1.3355,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.823529411764706e-05,
+      "loss": 1.2501,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.11764705882353e-05,
+      "loss": 1.3899,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.411764705882353e-05,
+      "loss": 1.1938,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.705882352941177e-05,
+      "loss": 1.2665,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001,
+      "loss": 1.271,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010294117647058823,
+      "loss": 1.2366,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010588235294117647,
+      "loss": 1.2666,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001088235294117647,
+      "loss": 1.4364,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011176470588235294,
+      "loss": 1.4345,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011470588235294118,
+      "loss": 1.3346,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011764705882352942,
+      "loss": 1.1571,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012058823529411765,
+      "loss": 1.1369,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001235294117647059,
+      "loss": 1.3395,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001264705882352941,
+      "loss": 1.2814,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012941176470588237,
+      "loss": 1.2829,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001323529411764706,
+      "loss": 1.4144,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013529411764705884,
+      "loss": 1.3002,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013823529411764707,
+      "loss": 1.2688,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001411764705882353,
+      "loss": 1.2941,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014411764705882354,
+      "loss": 1.2342,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014705882352941178,
+      "loss": 1.2505,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2876,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00015294117647058822,
+      "loss": 1.2052,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00015588235294117648,
+      "loss": 1.247,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001588235294117647,
+      "loss": 1.279,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016176470588235295,
+      "loss": 1.4225,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001647058823529412,
+      "loss": 1.2783,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016764705882352942,
+      "loss": 1.2701,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017058823529411766,
+      "loss": 1.2143,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001735294117647059,
+      "loss": 1.1219,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017647058823529413,
+      "loss": 1.3058,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017941176470588236,
+      "loss": 1.2263,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001823529411764706,
+      "loss": 1.1908,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018529411764705883,
+      "loss": 1.1254,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018823529411764707,
+      "loss": 1.2616,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001911764705882353,
+      "loss": 1.3775,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019411764705882354,
+      "loss": 1.2668,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019705882352941177,
+      "loss": 1.3207,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.2348,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999998963521776,
+      "loss": 1.1503,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999958540892524,
+      "loss": 1.3631,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999990671708875,
+      "loss": 1.0743,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999834163913867,
+      "loss": 1.2555,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999974088151827,
+      "loss": 1.3937,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999626870095333,
+      "loss": 1.3183,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199994921298814,
+      "loss": 1.3611,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999933666115578,
+      "loss": 1.1843,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999160464240746,
+      "loss": 1.2158,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998963539501557,
+      "loss": 1.1974,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999874588734642,
+      "loss": 1.2118,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998507508226524,
+      "loss": 1.5406,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998248402636017,
+      "loss": 1.2299,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997968571112013,
+      "loss": 1.2328,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997668014234594,
+      "loss": 1.2574,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997346732626795,
+      "loss": 1.4479,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997004726954624,
+      "loss": 1.1921,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996641997927043,
+      "loss": 1.2451,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019996258546295975,
+      "loss": 1.3283,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999585437285629,
+      "loss": 1.2803,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019995429478445834,
+      "loss": 1.0761,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994983863945388,
+      "loss": 1.3816,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994517530278686,
+      "loss": 1.5228,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999403047841243,
+      "loss": 1.2743,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999352270935625,
+      "loss": 1.2417,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019992994224162728,
+      "loss": 1.3346,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019992445023927396,
+      "loss": 1.125,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001999187510978872,
+      "loss": 1.3414,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199912844829281,
+      "loss": 1.0922,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019990673144569892,
+      "loss": 1.1706,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019990041095981363,
+      "loss": 1.2406,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001998938833847273,
+      "loss": 1.348,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019988714873397128,
+      "loss": 1.1514,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019988020702150618,
+      "loss": 1.2221,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019987305826172193,
+      "loss": 1.328,
+      "step": 103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019986570246943754,
+      "loss": 1.216,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001998581396599013,
+      "loss": 1.3349,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019985036984879052,
+      "loss": 1.5557,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001998423930522117,
+      "loss": 1.2588,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019983420928670044,
+      "loss": 1.1504,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019982581856922127,
+      "loss": 1.227,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019981722091716783,
+      "loss": 1.314,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019980841634836266,
+      "loss": 1.2199,
+      "step": 111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019979940488105722,
+      "loss": 1.4525,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019979018653393192,
+      "loss": 1.2992,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019978076132609599,
+      "loss": 1.126,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019977112927708744,
+      "loss": 1.0355,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019976129040687318,
+      "loss": 1.2737,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019975124473584863,
+      "loss": 1.3793,
+      "step": 117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001997409922848381,
+      "loss": 1.0846,
+      "step": 118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019973053307509452,
+      "loss": 1.1315,
+      "step": 119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019971986712829932,
+      "loss": 1.2724,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019970899446656253,
+      "loss": 1.0398,
+      "step": 121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001996979151124227,
+      "loss": 1.2935,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019968662908884693,
+      "loss": 1.1611,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019967513641923056,
+      "loss": 1.3912,
+      "step": 124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019966343712739742,
+      "loss": 1.2514,
+      "step": 125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019965153123759966,
+      "loss": 1.1017,
+      "step": 126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019963941877451763,
+      "loss": 1.1546,
+      "step": 127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019962709976326,
+      "loss": 1.2507,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019961457422936353,
+      "loss": 1.5428,
+      "step": 129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019960184219879303,
+      "loss": 1.3692,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001995889036979415,
+      "loss": 1.2622,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019957575875362993,
+      "loss": 1.4149,
+      "step": 132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019956240739310715,
+      "loss": 1.6812,
+      "step": 133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019954884964404996,
+      "loss": 1.1218,
+      "step": 134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019953508553456303,
+      "loss": 1.0271,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001995211150931787,
+      "loss": 1.0883,
+      "step": 136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001995069383488571,
+      "loss": 1.2352,
+      "step": 137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019949255533098604,
+      "loss": 1.091,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019947796606938084,
+      "loss": 1.1082,
+      "step": 139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019946317059428448,
+      "loss": 1.3496,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019944816893636725,
+      "loss": 1.1616,
+      "step": 141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019943296112672696,
+      "loss": 1.3368,
+      "step": 142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019941754719688873,
+      "loss": 1.4931,
+      "step": 143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019940192717880502,
+      "loss": 1.3244,
+      "step": 144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001993861011048554,
+      "loss": 1.343,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019937006900784665,
+      "loss": 1.2344,
+      "step": 146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019935383092101257,
+      "loss": 1.3948,
+      "step": 147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019933738687801403,
+      "loss": 1.2921,
+      "step": 148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019932073691293886,
+      "loss": 1.278,
+      "step": 149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019930388106030166,
+      "loss": 1.4108,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001992868193550439,
+      "loss": 1.2094,
+      "step": 151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019926955183253374,
+      "loss": 1.2681,
+      "step": 152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019925207852856597,
+      "loss": 1.374,
+      "step": 153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019923439947936204,
+      "loss": 1.1233,
+      "step": 154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019921651472156986,
+      "loss": 1.2079,
+      "step": 155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019919842429226372,
+      "loss": 1.3164,
+      "step": 156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019918012822894428,
+      "loss": 1.1898,
+      "step": 157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001991616265695385,
+      "loss": 1.2872,
+      "step": 158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019914291935239952,
+      "loss": 1.3129,
+      "step": 159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019912400661630658,
+      "loss": 1.4318,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019910488840046498,
+      "loss": 1.2104,
+      "step": 161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019908556474450593,
+      "loss": 1.6828,
+      "step": 162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001990660356884865,
+      "loss": 1.3609,
+      "step": 163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001990463012728896,
+      "loss": 1.3114,
+      "step": 164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019902636153862382,
+      "loss": 1.4224,
+      "step": 165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019900621652702336,
+      "loss": 1.0479,
+      "step": 166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019898586627984795,
+      "loss": 1.2699,
+      "step": 167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019896531083928273,
+      "loss": 1.3086,
+      "step": 168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001989445502479383,
+      "loss": 1.2605,
+      "step": 169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019892358454885042,
+      "loss": 1.1689,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019890241378548007,
+      "loss": 1.2177,
+      "step": 171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019888103800171333,
+      "loss": 1.184,
+      "step": 172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019885945724186125,
+      "loss": 1.2892,
+      "step": 173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019883767155065986,
+      "loss": 1.3962,
+      "step": 174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019881568097326988,
+      "loss": 1.2394,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019879348555527687,
+      "loss": 1.2743,
+      "step": 176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019877108534269093,
+      "loss": 1.1979,
+      "step": 177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019874848038194673,
+      "loss": 1.1101,
+      "step": 178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001987256707199034,
+      "loss": 1.2632,
+      "step": 179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019870265640384435,
+      "loss": 1.0411,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019867943748147726,
+      "loss": 1.1618,
+      "step": 181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019865601400093395,
+      "loss": 1.2277,
+      "step": 182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019863238601077024,
+      "loss": 1.222,
+      "step": 183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000198608553559966,
+      "loss": 1.2081,
+      "step": 184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019858451669792478,
+      "loss": 1.109,
+      "step": 185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000198560275474474,
+      "loss": 1.2588,
+      "step": 186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019853582993986467,
+      "loss": 1.3575,
+      "step": 187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019851118014477126,
+      "loss": 1.3443,
+      "step": 188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019848632614029174,
+      "loss": 1.2647,
+      "step": 189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019846126797794743,
+      "loss": 1.5478,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019843600570968275,
+      "loss": 1.6041,
+      "step": 191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019841053938786534,
+      "loss": 1.418,
+      "step": 192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019838486906528574,
+      "loss": 1.3145,
+      "step": 193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019835899479515737,
+      "loss": 1.2885,
+      "step": 194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019833291663111653,
+      "loss": 1.1609,
+      "step": 195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019830663462722207,
+      "loss": 1.203,
+      "step": 196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001982801488379555,
+      "loss": 1.2469,
+      "step": 197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019825345931822068,
+      "loss": 1.2714,
+      "step": 198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001982265661233438,
+      "loss": 1.2025,
+      "step": 199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019819946930907332,
+      "loss": 1.2784,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019817216893157968,
+      "loss": 1.2476,
+      "step": 201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019814466504745548,
+      "loss": 1.2354,
+      "step": 202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000198116957713715,
+      "loss": 1.2886,
+      "step": 203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019808904698779433,
+      "loss": 1.1855,
+      "step": 204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019806093292755125,
+      "loss": 1.4307,
+      "step": 205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019803261559126492,
+      "loss": 1.4193,
+      "step": 206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019800409503763602,
+      "loss": 1.191,
+      "step": 207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001979753713257863,
+      "loss": 1.1737,
+      "step": 208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001979464445152589,
+      "loss": 1.243,
+      "step": 209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019791731466601773,
+      "loss": 1.3641,
+      "step": 210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019788798183844775,
+      "loss": 1.2523,
+      "step": 211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001978584460933546,
+      "loss": 1.2865,
+      "step": 212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019782870749196466,
+      "loss": 1.2455,
+      "step": 213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019779876609592471,
+      "loss": 1.4786,
+      "step": 214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019776862196730193,
+      "loss": 1.1267,
+      "step": 215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019773827516858386,
+      "loss": 1.1323,
+      "step": 216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000197707725762678,
+      "loss": 1.2483,
+      "step": 217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019767697381291202,
+      "loss": 1.2256,
+      "step": 218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019764601938303334,
+      "loss": 1.3446,
+      "step": 219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019761486253720915,
+      "loss": 1.3073,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975835033400262,
+      "loss": 1.1838,
+      "step": 221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001975519418564908,
+      "loss": 1.3249,
+      "step": 222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019752017815202847,
+      "loss": 1.1298,
+      "step": 223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019748821229248405,
+      "loss": 1.2469,
+      "step": 224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001974560443441213,
+      "loss": 1.2255,
+      "step": 225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019742367437362306,
+      "loss": 1.258,
+      "step": 226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019739110244809083,
+      "loss": 1.2485,
+      "step": 227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019735832863504476,
+      "loss": 1.3442,
+      "step": 228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019732535300242358,
+      "loss": 1.2349,
+      "step": 229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019729217561858433,
+      "loss": 1.2588,
+      "step": 230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019725879655230227,
+      "loss": 1.4712,
+      "step": 231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019722521587277076,
+      "loss": 1.2458,
+      "step": 232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019719143364960105,
+      "loss": 1.1906,
+      "step": 233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019715744995282234,
+      "loss": 1.1116,
+      "step": 234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971232648528812,
+      "loss": 1.1613,
+      "step": 235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019708887842064194,
+      "loss": 1.0521,
+      "step": 236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001970542907273861,
+      "loss": 1.2548,
+      "step": 237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001970195018448125,
+      "loss": 1.2323,
+      "step": 238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019698451184503693,
+      "loss": 1.4259,
+      "step": 239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019694932080059217,
+      "loss": 1.1325,
+      "step": 240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019691392878442776,
+      "loss": 1.3215,
+      "step": 241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019687833586990973,
+      "loss": 1.2291,
+      "step": 242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019684254213082065,
+      "loss": 1.2551,
+      "step": 243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019680654764135945,
+      "loss": 1.292,
+      "step": 244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019677035247614105,
+      "loss": 1.2922,
+      "step": 245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019673395671019654,
+      "loss": 1.0648,
+      "step": 246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019669736041897268,
+      "loss": 1.2305,
+      "step": 247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019666056367833204,
+      "loss": 1.0777,
+      "step": 248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019662356656455264,
+      "loss": 1.1057,
+      "step": 249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019658636915432788,
+      "loss": 1.3594,
+      "step": 250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001965489715247664,
+      "loss": 1.2858,
+      "step": 251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965113737533918,
+      "loss": 1.2562,
+      "step": 252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019647357591814273,
+      "loss": 1.075,
+      "step": 253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964355780973723,
+      "loss": 1.1668,
+      "step": 254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001963973803698485,
+      "loss": 1.1547,
+      "step": 255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019635898281475346,
+      "loss": 1.1131,
+      "step": 256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001963203855116836,
+      "loss": 1.3103,
+      "step": 257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019628158854064956,
+      "loss": 1.2497,
+      "step": 258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962425919820757,
+      "loss": 1.15,
+      "step": 259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019620339591680023,
+      "loss": 1.2833,
+      "step": 260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019616400042607487,
+      "loss": 1.0163,
+      "step": 261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001961244055915647,
+      "loss": 1.2938,
+      "step": 262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019608461149534816,
+      "loss": 1.0977,
+      "step": 263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019604461821991667,
+      "loss": 0.9078,
+      "step": 264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019600442584817454,
+      "loss": 1.1471,
+      "step": 265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019596403446343877,
+      "loss": 1.5221,
+      "step": 266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019592344414943896,
+      "loss": 1.3637,
+      "step": 267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001958826549903171,
+      "loss": 1.1906,
+      "step": 268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001958416670706273,
+      "loss": 1.0934,
+      "step": 269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019580048047533578,
+      "loss": 1.2012,
+      "step": 270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019575909528982046,
+      "loss": 1.2908,
+      "step": 271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019571751159987115,
+      "loss": 1.2466,
+      "step": 272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019567572949168892,
+      "loss": 0.9926,
+      "step": 273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019563374905188637,
+      "loss": 1.2018,
+      "step": 274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000195591570367487,
+      "loss": 1.2056,
+      "step": 275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001955491935259255,
+      "loss": 1.2208,
+      "step": 276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019550661861504715,
+      "loss": 1.233,
+      "step": 277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001954638457231079,
+      "loss": 1.2352,
+      "step": 278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019542087493877412,
+      "loss": 1.2703,
+      "step": 279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001953777063511223,
+      "loss": 1.1301,
+      "step": 280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019533434004963913,
+      "loss": 1.3666,
+      "step": 281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019529077612422103,
+      "loss": 1.4975,
+      "step": 282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001952470146651741,
+      "loss": 1.2801,
+      "step": 283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019520305576321395,
+      "loss": 1.2487,
+      "step": 284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001951588995094655,
+      "loss": 1.2448,
+      "step": 285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019511454599546268,
+      "loss": 1.3375,
+      "step": 286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001950699953131484,
+      "loss": 1.4203,
+      "step": 287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001950252475548744,
+      "loss": 1.3922,
+      "step": 288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001949803028134007,
+      "loss": 1.0548,
+      "step": 289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019493516118189582,
+      "loss": 1.1612,
+      "step": 290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019488982275393642,
+      "loss": 1.2329,
+      "step": 291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019484428762350708,
+      "loss": 1.4126,
+      "step": 292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019479855588500015,
+      "loss": 1.2791,
+      "step": 293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019475262763321552,
+      "loss": 1.2292,
+      "step": 294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019470650296336047,
+      "loss": 1.3843,
+      "step": 295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019466018197104946,
+      "loss": 1.1819,
+      "step": 296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001946136647523038,
+      "loss": 1.2131,
+      "step": 297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019456695140355172,
+      "loss": 1.3227,
+      "step": 298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019452004202162796,
+      "loss": 1.1234,
+      "step": 299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001944729367037736,
+      "loss": 1.1919,
+      "step": 300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019442563554763594,
+      "loss": 1.1428,
+      "step": 301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001943781386512682,
+      "loss": 1.2858,
+      "step": 302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019433044611312937,
+      "loss": 1.1808,
+      "step": 303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000194282558032084,
+      "loss": 1.2501,
+      "step": 304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000194234474507402,
+      "loss": 1.6441,
+      "step": 305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001941861956387585,
+      "loss": 1.0559,
+      "step": 306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019413772152623336,
+      "loss": 1.1148,
+      "step": 307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019408905227031142,
+      "loss": 1.3489,
+      "step": 308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019404018797188183,
+      "loss": 1.2785,
+      "step": 309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019399112873223824,
+      "loss": 1.0916,
+      "step": 310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019394187465307828,
+      "loss": 1.0787,
+      "step": 311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001938924258365035,
+      "loss": 1.1702,
+      "step": 312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001938427823850191,
+      "loss": 1.2588,
+      "step": 313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019379294440153392,
+      "loss": 1.1058,
+      "step": 314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019374291198935982,
+      "loss": 1.2069,
+      "step": 315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019369268525221185,
+      "loss": 1.343,
+      "step": 316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019364226429420783,
+      "loss": 1.3341,
+      "step": 317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019359164921986825,
+      "loss": 1.0451,
+      "step": 318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001935408401341159,
+      "loss": 1.0893,
+      "step": 319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019348983714227583,
+      "loss": 1.2213,
+      "step": 320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019343864035007506,
+      "loss": 1.3557,
+      "step": 321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019338724986364222,
+      "loss": 1.1327,
+      "step": 322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019333566578950764,
+      "loss": 0.999,
+      "step": 323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019328388823460283,
+      "loss": 1.5289,
+      "step": 324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019323191730626036,
+      "loss": 1.1509,
+      "step": 325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019317975311221373,
+      "loss": 1.2142,
+      "step": 326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019312739576059711,
+      "loss": 1.2386,
+      "step": 327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019307484535994492,
+      "loss": 1.1913,
+      "step": 328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019302210201919186,
+      "loss": 1.2473,
+      "step": 329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019296916584767262,
+      "loss": 1.2853,
+      "step": 330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019291603695512156,
+      "loss": 1.309,
+      "step": 331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019286271545167254,
+      "loss": 1.1103,
+      "step": 332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019280920144785878,
+      "loss": 1.2968,
+      "step": 333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001927554950546124,
+      "loss": 1.1975,
+      "step": 334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019270159638326443,
+      "loss": 1.3768,
+      "step": 335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019264750554554453,
+      "loss": 1.3442,
+      "step": 336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019259322265358057,
+      "loss": 1.337,
+      "step": 337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019253874781989864,
+      "loss": 1.0822,
+      "step": 338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019248408115742275,
+      "loss": 1.2629,
+      "step": 339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019242922277947448,
+      "loss": 1.1279,
+      "step": 340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019237417279977285,
+      "loss": 1.3107,
+      "step": 341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019231893133243405,
+      "loss": 1.242,
+      "step": 342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001922634984919713,
+      "loss": 1.3852,
+      "step": 343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019220787439329438,
+      "loss": 1.0776,
+      "step": 344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001921520591517097,
+      "loss": 1.2565,
+      "step": 345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019209605288291978,
+      "loss": 1.2397,
+      "step": 346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001920398557030232,
+      "loss": 1.3005,
+      "step": 347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001919834677285142,
+      "loss": 1.1842,
+      "step": 348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019192688907628273,
+      "loss": 1.2918,
+      "step": 349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019187011986361374,
+      "loss": 1.1246,
+      "step": 350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019181316020818742,
+      "loss": 1.333,
+      "step": 351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001917560102280786,
+      "loss": 1.1095,
+      "step": 352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019169867004175675,
+      "loss": 1.1898,
+      "step": 353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001916411397680855,
+      "loss": 1.2508,
+      "step": 354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019158341952632272,
+      "loss": 1.0398,
+      "step": 355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019152550943611987,
+      "loss": 1.2314,
+      "step": 356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019146740961752202,
+      "loss": 1.2817,
+      "step": 357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019140912019096766,
+      "loss": 1.1854,
+      "step": 358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019135064127728818,
+      "loss": 1.0569,
+      "step": 359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001912919729977078,
+      "loss": 1.3705,
+      "step": 360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019123311547384331,
+      "loss": 1.1581,
+      "step": 361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019117406882770388,
+      "loss": 1.0548,
+      "step": 362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019111483318169054,
+      "loss": 1.3073,
+      "step": 363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019105540865859623,
+      "loss": 1.1966,
+      "step": 364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019099579538160542,
+      "loss": 1.1336,
+      "step": 365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001909359934742938,
+      "loss": 1.1724,
+      "step": 366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019087600306062817,
+      "loss": 1.239,
+      "step": 367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019081582426496598,
+      "loss": 1.0797,
+      "step": 368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001907554572120553,
+      "loss": 1.3036,
+      "step": 369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019069490202703438,
+      "loss": 1.2149,
+      "step": 370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001906341588354315,
+      "loss": 1.281,
+      "step": 371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019057322776316467,
+      "loss": 1.2726,
+      "step": 372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001905121089365413,
+      "loss": 1.2523,
+      "step": 373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019045080248225805,
+      "loss": 1.1704,
+      "step": 374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019038930852740058,
+      "loss": 1.1653,
+      "step": 375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019032762719944316,
+      "loss": 1.1893,
+      "step": 376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019026575862624844,
+      "loss": 1.2504,
+      "step": 377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001902037029360674,
+      "loss": 1.3984,
+      "step": 378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001901414602575387,
+      "loss": 1.2521,
+      "step": 379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019007903071968868,
+      "loss": 1.2099,
+      "step": 380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001900164144519311,
+      "loss": 1.0739,
+      "step": 381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018995361158406676,
+      "loss": 1.2888,
+      "step": 382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018989062224628324,
+      "loss": 1.2413,
+      "step": 383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018982744656915475,
+      "loss": 1.2772,
+      "step": 384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018976408468364162,
+      "loss": 1.0562,
+      "step": 385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018970053672109037,
+      "loss": 1.2799,
+      "step": 386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001896368028132331,
+      "loss": 1.2495,
+      "step": 387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018957288309218745,
+      "loss": 1.1946,
+      "step": 388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018950877769045624,
+      "loss": 1.1901,
+      "step": 389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018944448674092714,
+      "loss": 1.0915,
+      "step": 390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001893800103768725,
+      "loss": 1.3027,
+      "step": 391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000189315348731949,
+      "loss": 1.4335,
+      "step": 392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018925050194019746,
+      "loss": 1.094,
+      "step": 393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001891854701360424,
+      "loss": 1.26,
+      "step": 394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018912025345429196,
+      "loss": 1.1428,
+      "step": 395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018905485203013744,
+      "loss": 1.1153,
+      "step": 396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018898926599915318,
+      "loss": 1.39,
+      "step": 397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018892349549729615,
+      "loss": 1.3712,
+      "step": 398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018885754066090576,
+      "loss": 1.1251,
+      "step": 399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018879140162670347,
+      "loss": 1.5685,
+      "step": 400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018872507853179263,
+      "loss": 1.249,
+      "step": 401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018865857151365814,
+      "loss": 1.1222,
+      "step": 402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018859188071016617,
+      "loss": 1.346,
+      "step": 403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001885250062595638,
+      "loss": 1.2928,
+      "step": 404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018845794830047892,
+      "loss": 1.0523,
+      "step": 405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001883907069719197,
+      "loss": 1.124,
+      "step": 406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001883232824132745,
+      "loss": 1.1903,
+      "step": 407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001882556747643115,
+      "loss": 1.2598,
+      "step": 408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018818788416517844,
+      "loss": 1.1623,
+      "step": 409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018811991075640223,
+      "loss": 1.2127,
+      "step": 410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001880517546788888,
+      "loss": 1.4151,
+      "step": 411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018798341607392276,
+      "loss": 1.2056,
+      "step": 412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018791489508316703,
+      "loss": 1.0462,
+      "step": 413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018784619184866267,
+      "loss": 1.2483,
+      "step": 414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018777730651282847,
+      "loss": 1.3236,
+      "step": 415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001877082392184607,
+      "loss": 1.1879,
+      "step": 416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001876389901087329,
+      "loss": 1.1049,
+      "step": 417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018756955932719546,
+      "loss": 1.0995,
+      "step": 418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018749994701777536,
+      "loss": 1.063,
+      "step": 419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018743015332477588,
+      "loss": 1.1434,
+      "step": 420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018736017839287629,
+      "loss": 1.2464,
+      "step": 421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001872900223671316,
+      "loss": 1.1707,
+      "step": 422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018721968539297217,
+      "loss": 1.2639,
+      "step": 423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001871491676162035,
+      "loss": 1.2609,
+      "step": 424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001870784691830059,
+      "loss": 1.1206,
+      "step": 425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001870075902399341,
+      "loss": 1.1779,
+      "step": 426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018693653093391704,
+      "loss": 1.2995,
+      "step": 427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018686529141225767,
+      "loss": 1.0957,
+      "step": 428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001867938718226323,
+      "loss": 0.993,
+      "step": 429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018672227231309068,
+      "loss": 1.2656,
+      "step": 430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018665049303205549,
+      "loss": 1.3583,
+      "step": 431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018657853412832202,
+      "loss": 1.1753,
+      "step": 432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018650639575105798,
+      "loss": 1.0839,
+      "step": 433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018643407804980303,
+      "loss": 1.0579,
+      "step": 434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001863615811744687,
+      "loss": 1.1339,
+      "step": 435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018628890527533777,
+      "loss": 1.5151,
+      "step": 436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018621605050306422,
+      "loss": 1.1033,
+      "step": 437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018614301700867287,
+      "loss": 1.2376,
+      "step": 438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018606980494355896,
+      "loss": 1.2995,
+      "step": 439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001859964144594879,
+      "loss": 1.1819,
+      "step": 440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018592284570859496,
+      "loss": 1.2063,
+      "step": 441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000185849098843385,
+      "loss": 1.2484,
+      "step": 442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000185775174016732,
+      "loss": 1.0711,
+      "step": 443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018570107138187893,
+      "loss": 1.1875,
+      "step": 444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018562679109243731,
+      "loss": 1.4785,
+      "step": 445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000185552333302387,
+      "loss": 1.4187,
+      "step": 446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001854776981660757,
+      "loss": 1.1621,
+      "step": 447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018540288583821882,
+      "loss": 1.1524,
+      "step": 448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018532789647389904,
+      "loss": 1.2599,
+      "step": 449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018525273022856607,
+      "loss": 1.097,
+      "step": 450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018517738725803626,
+      "loss": 1.2165,
+      "step": 451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001851018677184923,
+      "loss": 1.2315,
+      "step": 452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001850261717664829,
+      "loss": 0.9369,
+      "step": 453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018495029955892248,
+      "loss": 1.2465,
+      "step": 454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018487425125309078,
+      "loss": 1.2831,
+      "step": 455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018479802700663268,
+      "loss": 1.2577,
+      "step": 456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018472162697755776,
+      "loss": 1.1233,
+      "step": 457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018464505132423983,
+      "loss": 1.203,
+      "step": 458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000184568300205417,
+      "loss": 1.1508,
+      "step": 459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018449137378019094,
+      "loss": 1.1357,
+      "step": 460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001844142722080268,
+      "loss": 1.2458,
+      "step": 461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018433699564875274,
+      "loss": 1.3916,
+      "step": 462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018425954426255978,
+      "loss": 1.1561,
+      "step": 463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001841819182100012,
+      "loss": 1.1037,
+      "step": 464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018410411765199243,
+      "loss": 1.3372,
+      "step": 465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018402614274981073,
+      "loss": 1.3155,
+      "step": 466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018394799366509452,
+      "loss": 1.2639,
+      "step": 467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001838696705598436,
+      "loss": 1.1655,
+      "step": 468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018379117359641824,
+      "loss": 1.0399,
+      "step": 469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001837125029375393,
+      "loss": 1.0648,
+      "step": 470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001836336587462876,
+      "loss": 1.4474,
+      "step": 471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018355464118610372,
+      "loss": 1.2198,
+      "step": 472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001834754504207876,
+      "loss": 1.2475,
+      "step": 473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018339608661449832,
+      "loss": 0.9374,
+      "step": 474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001833165499317535,
+      "loss": 1.1665,
+      "step": 475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018323684053742932,
+      "loss": 1.1142,
+      "step": 476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018315695859675977,
+      "loss": 1.277,
+      "step": 477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018307690427533672,
+      "loss": 1.216,
+      "step": 478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018299667773910924,
+      "loss": 1.5161,
+      "step": 479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018291627915438348,
+      "loss": 1.4463,
+      "step": 480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001828357086878222,
+      "loss": 1.154,
+      "step": 481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018275496650644444,
+      "loss": 1.0791,
+      "step": 482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018267405277762525,
+      "loss": 1.2681,
+      "step": 483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018259296766909527,
+      "loss": 1.0295,
+      "step": 484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001825117113489404,
+      "loss": 1.1922,
+      "step": 485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018243028398560142,
+      "loss": 0.9604,
+      "step": 486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018234868574787372,
+      "loss": 1.1175,
+      "step": 487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018226691680490694,
+      "loss": 1.0906,
+      "step": 488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001821849773262045,
+      "loss": 1.2605,
+      "step": 489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018210286748162336,
+      "loss": 1.1885,
+      "step": 490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018202058744137364,
+      "loss": 1.06,
+      "step": 491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018193813737601834,
+      "loss": 1.2855,
+      "step": 492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001818555174564728,
+      "loss": 1.1235,
+      "step": 493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018177272785400455,
+      "loss": 1.2737,
+      "step": 494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001816897687402328,
+      "loss": 1.1921,
+      "step": 495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001816066402871282,
+      "loss": 1.063,
+      "step": 496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001815233426670124,
+      "loss": 1.3188,
+      "step": 497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018143987605255774,
+      "loss": 1.1964,
+      "step": 498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001813562406167869,
+      "loss": 1.2531,
+      "step": 499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018127243653307248,
+      "loss": 1.118,
+      "step": 500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018118846397513665,
+      "loss": 1.0987,
+      "step": 501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018110432311705096,
+      "loss": 1.3884,
+      "step": 502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018102001413323565,
+      "loss": 1.4359,
+      "step": 503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018093553719845967,
+      "loss": 1.045,
+      "step": 504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001808508924878399,
+      "loss": 1.0859,
+      "step": 505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018076608017684128,
+      "loss": 1.2478,
+      "step": 506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018068110044127593,
+      "loss": 1.0507,
+      "step": 507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018059595345730318,
+      "loss": 1.2064,
+      "step": 508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000180510639401429,
+      "loss": 1.2447,
+      "step": 509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018042515845050576,
+      "loss": 1.4431,
+      "step": 510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001803395107817317,
+      "loss": 1.2119,
+      "step": 511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018025369657265075,
+      "loss": 1.0663,
+      "step": 512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000180167716001152,
+      "loss": 1.1376,
+      "step": 513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001800815692454694,
+      "loss": 1.0241,
+      "step": 514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001799952564841815,
+      "loss": 1.2796,
+      "step": 515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017990877789621082,
+      "loss": 1.2575,
+      "step": 516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017982213366082373,
+      "loss": 1.1562,
+      "step": 517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017973532395763,
+      "loss": 1.2806,
+      "step": 518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001796483489665823,
+      "loss": 1.1232,
+      "step": 519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017956120886797604,
+      "loss": 1.2081,
+      "step": 520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001794739038424488,
+      "loss": 1.361,
+      "step": 521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001793864340709802,
+      "loss": 1.0471,
+      "step": 522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017929879973489115,
+      "loss": 0.994,
+      "step": 523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017921100101584388,
+      "loss": 1.1898,
+      "step": 524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017912303809584126,
+      "loss": 1.2885,
+      "step": 525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017903491115722666,
+      "loss": 1.3112,
+      "step": 526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017894662038268333,
+      "loss": 1.115,
+      "step": 527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017885816595523423,
+      "loss": 1.3407,
+      "step": 528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001787695480582415,
+      "loss": 1.2152,
+      "step": 529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017868076687540624,
+      "loss": 1.1489,
+      "step": 530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017859182259076793,
+      "loss": 1.1711,
+      "step": 531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017850271538870423,
+      "loss": 1.1491,
+      "step": 532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017841344545393043,
+      "loss": 1.4032,
+      "step": 533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017832401297149928,
+      "loss": 1.2209,
+      "step": 534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001782344181268004,
+      "loss": 1.2372,
+      "step": 535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017814466110555998,
+      "loss": 1.1747,
+      "step": 536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017805474209384045,
+      "loss": 1.2833,
+      "step": 537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017796466127804,
+      "loss": 1.0638,
+      "step": 538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001778744188448922,
+      "loss": 1.3156,
+      "step": 539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001777840149814657,
+      "loss": 1.346,
+      "step": 540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017769344987516382,
+      "loss": 1.2409,
+      "step": 541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017760272371372402,
+      "loss": 1.0615,
+      "step": 542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001775118366852177,
+      "loss": 1.2862,
+      "step": 543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017742078897804974,
+      "loss": 1.0227,
+      "step": 544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017732958078095804,
+      "loss": 1.1639,
+      "step": 545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017723821228301322,
+      "loss": 1.4482,
+      "step": 546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017714668367361825,
+      "loss": 1.1363,
+      "step": 547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017705499514250784,
+      "loss": 1.3087,
+      "step": 548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017696314687974844,
+      "loss": 1.3111,
+      "step": 549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001768711390757374,
+      "loss": 1.2013,
+      "step": 550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000176778971921203,
+      "loss": 1.2325,
+      "step": 551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001766866456072036,
+      "loss": 0.998,
+      "step": 552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017659416032512776,
+      "loss": 1.3358,
+      "step": 553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017650151626669337,
+      "loss": 1.2797,
+      "step": 554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017640871362394753,
+      "loss": 1.1523,
+      "step": 555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001763157525892661,
+      "loss": 1.0524,
+      "step": 556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017622263335535324,
+      "loss": 1.1988,
+      "step": 557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001761293561152411,
+      "loss": 1.0297,
+      "step": 558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017603592106228927,
+      "loss": 1.2259,
+      "step": 559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001759423283901846,
+      "loss": 0.9118,
+      "step": 560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017584857829294062,
+      "loss": 1.2517,
+      "step": 561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017575467096489717,
+      "loss": 1.1667,
+      "step": 562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017566060660072008,
+      "loss": 1.1651,
+      "step": 563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017556638539540067,
+      "loss": 1.3498,
+      "step": 564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001754720075442554,
+      "loss": 1.2504,
+      "step": 565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017537747324292542,
+      "loss": 1.0148,
+      "step": 566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017528278268737625,
+      "loss": 1.2432,
+      "step": 567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017518793607389723,
+      "loss": 1.2598,
+      "step": 568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017509293359910136,
+      "loss": 1.2264,
+      "step": 569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017499777545992452,
+      "loss": 1.1781,
+      "step": 570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001749024618536255,
+      "loss": 1.1941,
+      "step": 571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017480699297778518,
+      "loss": 1.0549,
+      "step": 572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017471136903030637,
+      "loss": 1.2401,
+      "step": 573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017461559020941341,
+      "loss": 1.1683,
+      "step": 574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001745196567136516,
+      "loss": 1.281,
+      "step": 575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001744235687418869,
+      "loss": 1.1068,
+      "step": 576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017432732649330548,
+      "loss": 1.1492,
+      "step": 577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017423093016741333,
+      "loss": 1.3113,
+      "step": 578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017413437996403582,
+      "loss": 1.1857,
+      "step": 579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017403767608331733,
+      "loss": 1.1321,
+      "step": 580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017394081872572082,
+      "loss": 1.3185,
+      "step": 581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017384380809202737,
+      "loss": 1.2783,
+      "step": 582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017374664438333577,
+      "loss": 1.1044,
+      "step": 583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017364932780106212,
+      "loss": 1.1475,
+      "step": 584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017355185854693955,
+      "loss": 1.1435,
+      "step": 585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017345423682301755,
+      "loss": 1.3367,
+      "step": 586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017335646283166164,
+      "loss": 1.1303,
+      "step": 587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017325853677555312,
+      "loss": 0.9936,
+      "step": 588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001731604588576884,
+      "loss": 1.1361,
+      "step": 589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017306222928137875,
+      "loss": 1.375,
+      "step": 590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001729638482502498,
+      "loss": 1.0961,
+      "step": 591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017286531596824112,
+      "loss": 1.2328,
+      "step": 592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017276663263960586,
+      "loss": 1.2422,
+      "step": 593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017266779846891029,
+      "loss": 1.1101,
+      "step": 594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017256881366103328,
+      "loss": 1.1748,
+      "step": 595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017246967842116605,
+      "loss": 1.3677,
+      "step": 596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017237039295481168,
+      "loss": 0.9986,
+      "step": 597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017227095746778456,
+      "loss": 1.2788,
+      "step": 598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017217137216621016,
+      "loss": 1.1473,
+      "step": 599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017207163725652445,
+      "loss": 1.0596,
+      "step": 600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017197175294547353,
+      "loss": 1.176,
+      "step": 601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017187171944011329,
+      "loss": 1.1753,
+      "step": 602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017177153694780879,
+      "loss": 0.9697,
+      "step": 603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000171671205676234,
+      "loss": 1.431,
+      "step": 604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017157072583337123,
+      "loss": 1.0762,
+      "step": 605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017147009762751085,
+      "loss": 1.4879,
+      "step": 606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017136932126725078,
+      "loss": 1.1397,
+      "step": 607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017126839696149596,
+      "loss": 1.2918,
+      "step": 608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001711673249194581,
+      "loss": 1.27,
+      "step": 609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017106610535065517,
+      "loss": 1.2839,
+      "step": 610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001709647384649109,
+      "loss": 1.1277,
+      "step": 611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001708632244723545,
+      "loss": 1.2451,
+      "step": 612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001707615635834199,
+      "loss": 1.224,
+      "step": 613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017065975600884584,
+      "loss": 1.1693,
+      "step": 614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001705578019596749,
+      "loss": 1.1604,
+      "step": 615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017045570164725348,
+      "loss": 1.1424,
+      "step": 616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000170353455283231,
+      "loss": 1.1361,
+      "step": 617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017025106307955973,
+      "loss": 1.3433,
+      "step": 618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017014852524849426,
+      "loss": 1.198,
+      "step": 619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017004584200259107,
+      "loss": 1.3421,
+      "step": 620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000169943013554708,
+      "loss": 1.3464,
+      "step": 621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016984004011800404,
+      "loss": 1.1094,
+      "step": 622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001697369219059385,
+      "loss": 1.114,
+      "step": 623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001696336591322711,
+      "loss": 1.0978,
+      "step": 624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016953025201106093,
+      "loss": 1.2654,
+      "step": 625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016942670075666656,
+      "loss": 1.4481,
+      "step": 626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016932300558374514,
+      "loss": 1.13,
+      "step": 627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016921916670725233,
+      "loss": 1.1954,
+      "step": 628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016911518434244155,
+      "loss": 1.145,
+      "step": 629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016901105870486372,
+      "loss": 1.2081,
+      "step": 630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016890679001036677,
+      "loss": 1.2189,
+      "step": 631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016880237847509514,
+      "loss": 1.2548,
+      "step": 632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001686978243154894,
+      "loss": 1.0769,
+      "step": 633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016859312774828578,
+      "loss": 1.207,
+      "step": 634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016848828899051568,
+      "loss": 1.2526,
+      "step": 635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001683833082595053,
+      "loss": 1.3437,
+      "step": 636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016827818577287511,
+      "loss": 1.2119,
+      "step": 637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001681729217485395,
+      "loss": 0.9382,
+      "step": 638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001680675164047061,
+      "loss": 1.2919,
+      "step": 639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001679619699598757,
+      "loss": 1.2436,
+      "step": 640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016785628263284142,
+      "loss": 1.058,
+      "step": 641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016775045464268855,
+      "loss": 1.077,
+      "step": 642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016764448620879382,
+      "loss": 1.126,
+      "step": 643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016753837755082527,
+      "loss": 1.1716,
+      "step": 644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016743212888874144,
+      "loss": 0.9635,
+      "step": 645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016732574044279122,
+      "loss": 0.8972,
+      "step": 646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016721921243351324,
+      "loss": 1.2346,
+      "step": 647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001671125450817354,
+      "loss": 1.2805,
+      "step": 648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016700573860857452,
+      "loss": 1.1036,
+      "step": 649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016689879323543566,
+      "loss": 1.1667,
+      "step": 650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016679170918401204,
+      "loss": 1.2651,
+      "step": 651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016668448667628418,
+      "loss": 1.1998,
+      "step": 652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016657712593451965,
+      "loss": 1.0381,
+      "step": 653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016646962718127264,
+      "loss": 1.1996,
+      "step": 654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016636199063938337,
+      "loss": 1.1794,
+      "step": 655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016625421653197766,
+      "loss": 0.9635,
+      "step": 656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016614630508246658,
+      "loss": 0.9963,
+      "step": 657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016603825651454588,
+      "loss": 1.228,
+      "step": 658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001659300710521955,
+      "loss": 1.167,
+      "step": 659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001658217489196792,
+      "loss": 1.3202,
+      "step": 660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016571329034154405,
+      "loss": 1.0456,
+      "step": 661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016560469554262,
+      "loss": 1.2181,
+      "step": 662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016549596474801925,
+      "loss": 1.2705,
+      "step": 663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016538709818313604,
+      "loss": 1.1535,
+      "step": 664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016527809607364607,
+      "loss": 1.0159,
+      "step": 665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001651689586455059,
+      "loss": 1.1508,
+      "step": 666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001650596861249527,
+      "loss": 1.0439,
+      "step": 667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001649502787385036,
+      "loss": 1.3048,
+      "step": 668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016484073671295543,
+      "loss": 1.3086,
+      "step": 669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016473106027538393,
+      "loss": 1.1795,
+      "step": 670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016462124965314372,
+      "loss": 1.1655,
+      "step": 671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001645113050738673,
+      "loss": 1.1816,
+      "step": 672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016440122676546505,
+      "loss": 1.1037,
+      "step": 673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016429101495612453,
+      "loss": 1.3513,
+      "step": 674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016418066987430998,
+      "loss": 1.1256,
+      "step": 675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016407019174876196,
+      "loss": 1.0347,
+      "step": 676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016395958080849684,
+      "loss": 1.0716,
+      "step": 677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016384883728280626,
+      "loss": 1.1002,
+      "step": 678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016373796140125674,
+      "loss": 0.9785,
+      "step": 679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016362695339368913,
+      "loss": 1.107,
+      "step": 680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001635158134902182,
+      "loss": 1.3126,
+      "step": 681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016340454192123217,
+      "loss": 1.1347,
+      "step": 682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016329313891739209,
+      "loss": 1.1564,
+      "step": 683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016318160470963158,
+      "loss": 1.1214,
+      "step": 684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016306993952915617,
+      "loss": 1.0821,
+      "step": 685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016295814360744296,
+      "loss": 1.0957,
+      "step": 686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016284621717623998,
+      "loss": 1.178,
+      "step": 687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016273416046756585,
+      "loss": 1.1498,
+      "step": 688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016262197371370927,
+      "loss": 1.2957,
+      "step": 689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001625096571472285,
+      "loss": 0.9348,
+      "step": 690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016239721100095084,
+      "loss": 1.2325,
+      "step": 691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016228463550797234,
+      "loss": 1.0739,
+      "step": 692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016217193090165702,
+      "loss": 1.3574,
+      "step": 693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016205909741563663,
+      "loss": 1.1554,
+      "step": 694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001619461352838101,
+      "loss": 1.0253,
+      "step": 695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016183304474034303,
+      "loss": 1.0506,
+      "step": 696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001617198260196671,
+      "loss": 1.3249,
+      "step": 697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016160647935647987,
+      "loss": 1.1862,
+      "step": 698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016149300498574402,
+      "loss": 1.2605,
+      "step": 699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016137940314268695,
+      "loss": 1.0166,
+      "step": 700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016126567406280038,
+      "loss": 1.1251,
+      "step": 701
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016115181798183968,
+      "loss": 1.1132,
+      "step": 702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001610378351358236,
+      "loss": 1.0725,
+      "step": 703
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016092372576103362,
+      "loss": 1.038,
+      "step": 704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016080949009401345,
+      "loss": 1.0443,
+      "step": 705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016069512837156867,
+      "loss": 1.2413,
+      "step": 706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016058064083076618,
+      "loss": 1.2521,
+      "step": 707
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016046602770893365,
+      "loss": 1.0743,
+      "step": 708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016035128924365906,
+      "loss": 1.4178,
+      "step": 709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016023642567279033,
+      "loss": 1.211,
+      "step": 710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016012143723443452,
+      "loss": 1.1634,
+      "step": 711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016000632416695782,
+      "loss": 0.9715,
+      "step": 712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015989108670898443,
+      "loss": 1.1186,
+      "step": 713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001597757250993967,
+      "loss": 1.2126,
+      "step": 714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015966023957733415,
+      "loss": 1.2035,
+      "step": 715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001595446303821933,
+      "loss": 1.2056,
+      "step": 716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015942889775362695,
+      "loss": 1.2331,
+      "step": 717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015931304193154375,
+      "loss": 1.1868,
+      "step": 718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015919706315610787,
+      "loss": 1.2837,
+      "step": 719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015908096166773817,
+      "loss": 1.338,
+      "step": 720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015896473770710807,
+      "loss": 1.0693,
+      "step": 721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001588483915151447,
+      "loss": 1.4525,
+      "step": 722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001587319233330287,
+      "loss": 1.3511,
+      "step": 723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015861533340219347,
+      "loss": 1.1161,
+      "step": 724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015849862196432493,
+      "loss": 1.0558,
+      "step": 725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015838178926136078,
+      "loss": 1.0671,
+      "step": 726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015826483553549014,
+      "loss": 1.1497,
+      "step": 727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015814776102915295,
+      "loss": 1.2442,
+      "step": 728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001580305659850396,
+      "loss": 1.2528,
+      "step": 729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001579132506460903,
+      "loss": 1.1791,
+      "step": 730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015779581525549467,
+      "loss": 1.1496,
+      "step": 731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001576782600566911,
+      "loss": 1.0973,
+      "step": 732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015756058529336646,
+      "loss": 1.2501,
+      "step": 733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015744279120945534,
+      "loss": 1.236,
+      "step": 734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015732487804913977,
+      "loss": 0.9319,
+      "step": 735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015720684605684864,
+      "loss": 1.1089,
+      "step": 736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001570886954772571,
+      "loss": 1.2369,
+      "step": 737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015697042655528617,
+      "loss": 1.033,
+      "step": 738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015685203953610214,
+      "loss": 0.9563,
+      "step": 739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015673353466511618,
+      "loss": 1.1874,
+      "step": 740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001566149121879837,
+      "loss": 1.0652,
+      "step": 741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015649617235060395,
+      "loss": 0.9729,
+      "step": 742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001563773153991194,
+      "loss": 1.1644,
+      "step": 743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001562583415799154,
+      "loss": 1.1626,
+      "step": 744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015613925113961943,
+      "loss": 1.1397,
+      "step": 745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015602004432510084,
+      "loss": 1.2284,
+      "step": 746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015590072138347013,
+      "loss": 1.2414,
+      "step": 747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015578128256207857,
+      "loss": 1.0592,
+      "step": 748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015566172810851765,
+      "loss": 1.2382,
+      "step": 749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015554205827061855,
+      "loss": 1.2809,
+      "step": 750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015542227329645155,
+      "loss": 0.8668,
+      "step": 751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001553023734343258,
+      "loss": 0.6818,
+      "step": 752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015518235893278846,
+      "loss": 0.6771,
+      "step": 753
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015506223004062427,
+      "loss": 0.7074,
+      "step": 754
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015494198700685536,
+      "loss": 0.6281,
+      "step": 755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015482163008074016,
+      "loss": 0.6755,
+      "step": 756
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015470115951177343,
+      "loss": 0.6752,
+      "step": 757
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015458057554968533,
+      "loss": 0.5428,
+      "step": 758
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001544598784444412,
+      "loss": 0.7003,
+      "step": 759
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001543390684462409,
+      "loss": 0.506,
+      "step": 760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015421814580551826,
+      "loss": 0.5807,
+      "step": 761
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015409711077294066,
+      "loss": 0.7003,
+      "step": 762
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001539759635994085,
+      "loss": 0.6965,
+      "step": 763
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015385470453605456,
+      "loss": 0.7081,
+      "step": 764
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015373333383424353,
+      "loss": 0.717,
+      "step": 765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001536118517455717,
+      "loss": 0.6623,
+      "step": 766
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015349025852186605,
+      "loss": 0.7523,
+      "step": 767
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015336855441518414,
+      "loss": 0.8403,
+      "step": 768
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015324673967781317,
+      "loss": 0.5041,
+      "step": 769
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015312481456226986,
+      "loss": 0.659,
+      "step": 770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015300277932129967,
+      "loss": 0.6952,
+      "step": 771
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001528806342078763,
+      "loss": 0.591,
+      "step": 772
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001527583794752013,
+      "loss": 0.5676,
+      "step": 773
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015263601537670332,
+      "loss": 0.8915,
+      "step": 774
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015251354216603793,
+      "loss": 0.6137,
+      "step": 775
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015239096009708665,
+      "loss": 0.6979,
+      "step": 776
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001522682694239568,
+      "loss": 0.6997,
+      "step": 777
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015214547040098082,
+      "loss": 0.557,
+      "step": 778
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015202256328271575,
+      "loss": 0.7591,
+      "step": 779
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015189954832394266,
+      "loss": 0.7054,
+      "step": 780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015177642577966625,
+      "loss": 0.6023,
+      "step": 781
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015165319590511412,
+      "loss": 0.513,
+      "step": 782
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015152985895573647,
+      "loss": 0.5737,
+      "step": 783
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015140641518720545,
+      "loss": 0.6392,
+      "step": 784
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015128286485541459,
+      "loss": 0.5861,
+      "step": 785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015115920821647834,
+      "loss": 0.6899,
+      "step": 786
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015103544552673155,
+      "loss": 0.6122,
+      "step": 787
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015091157704272886,
+      "loss": 0.6617,
+      "step": 788
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015078760302124423,
+      "loss": 0.8937,
+      "step": 789
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015066352371927047,
+      "loss": 0.6895,
+      "step": 790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015053933939401852,
+      "loss": 0.5034,
+      "step": 791
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001504150503029171,
+      "loss": 0.5446,
+      "step": 792
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015029065670361203,
+      "loss": 0.6034,
+      "step": 793
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001501661588539659,
+      "loss": 0.6286,
+      "step": 794
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001500415570120573,
+      "loss": 0.6943,
+      "step": 795
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014991685143618042,
+      "loss": 0.6186,
+      "step": 796
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014979204238484446,
+      "loss": 0.5498,
+      "step": 797
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014966713011677318,
+      "loss": 0.5775,
+      "step": 798
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014954211489090426,
+      "loss": 0.6028,
+      "step": 799
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014941699696638887,
+      "loss": 0.6215,
+      "step": 800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014929177660259093,
+      "loss": 0.6362,
+      "step": 801
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014916645405908683,
+      "loss": 0.5457,
+      "step": 802
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014904102959566477,
+      "loss": 0.6612,
+      "step": 803
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001489155034723242,
+      "loss": 0.6341,
+      "step": 804
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014878987594927527,
+      "loss": 0.6023,
+      "step": 805
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014866414728693838,
+      "loss": 0.5799,
+      "step": 806
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014853831774594363,
+      "loss": 0.7384,
+      "step": 807
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014841238758713006,
+      "loss": 0.6398,
+      "step": 808
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014828635707154553,
+      "loss": 0.595,
+      "step": 809
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001481602264604457,
+      "loss": 0.65,
+      "step": 810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014803399601529395,
+      "loss": 0.6335,
+      "step": 811
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014790766599776038,
+      "loss": 0.6416,
+      "step": 812
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014778123666972173,
+      "loss": 0.6757,
+      "step": 813
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014765470829326033,
+      "loss": 0.6103,
+      "step": 814
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014752808113066413,
+      "loss": 0.7826,
+      "step": 815
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001474013554444257,
+      "loss": 0.8057,
+      "step": 816
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014727453149724183,
+      "loss": 0.7562,
+      "step": 817
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014714760955201306,
+      "loss": 0.6032,
+      "step": 818
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001470205898718431,
+      "loss": 0.9197,
+      "step": 819
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014689347272003813,
+      "loss": 0.639,
+      "step": 820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001467662583601065,
+      "loss": 0.6899,
+      "step": 821
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001466389470557581,
+      "loss": 0.4985,
+      "step": 822
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001465115390709036,
+      "loss": 0.5338,
+      "step": 823
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001463840346696543,
+      "loss": 0.6577,
+      "step": 824
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014625643411632125,
+      "loss": 0.5817,
+      "step": 825
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014612873767541485,
+      "loss": 0.508,
+      "step": 826
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001460009456116442,
+      "loss": 0.5341,
+      "step": 827
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014587305818991673,
+      "loss": 0.5713,
+      "step": 828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014574507567533752,
+      "loss": 0.5807,
+      "step": 829
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001456169983332087,
+      "loss": 0.6298,
+      "step": 830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014548882642902898,
+      "loss": 1.9463,
+      "step": 831
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014536056022849327,
+      "loss": 0.7911,
+      "step": 832
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014523219999749173,
+      "loss": 0.7806,
+      "step": 833
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014510374600210953,
+      "loss": 0.6927,
+      "step": 834
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014497519850862624,
+      "loss": 0.6821,
+      "step": 835
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001448465577835152,
+      "loss": 0.5841,
+      "step": 836
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000144717824093443,
+      "loss": 0.5784,
+      "step": 837
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000144588997705269,
+      "loss": 0.6407,
+      "step": 838
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001444600788860447,
+      "loss": 0.6555,
+      "step": 839
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001443310679030132,
+      "loss": 0.5664,
+      "step": 840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014420196502360863,
+      "loss": 0.5717,
+      "step": 841
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014407277051545564,
+      "loss": 0.5176,
+      "step": 842
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014394348464636883,
+      "loss": 0.6081,
+      "step": 843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014381410768435216,
+      "loss": 0.7215,
+      "step": 844
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014368463989759847,
+      "loss": 0.6465,
+      "step": 845
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001435550815544888,
+      "loss": 0.6049,
+      "step": 846
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014342543292359197,
+      "loss": 0.7839,
+      "step": 847
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014329569427366394,
+      "loss": 0.5071,
+      "step": 848
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014316586587364728,
+      "loss": 0.5441,
+      "step": 849
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014303594799267065,
+      "loss": 0.7512,
+      "step": 850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014290594090004812,
+      "loss": 0.7631,
+      "step": 851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014277584486527872,
+      "loss": 0.6088,
+      "step": 852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014264566015804593,
+      "loss": 0.6284,
+      "step": 853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001425153870482169,
+      "loss": 0.7256,
+      "step": 854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001423850258058421,
+      "loss": 0.7163,
+      "step": 855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014225457670115482,
+      "loss": 0.6409,
+      "step": 856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014212404000457028,
+      "loss": 0.6952,
+      "step": 857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014199341598668537,
+      "loss": 0.5267,
+      "step": 858
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014186270491827805,
+      "loss": 0.6278,
+      "step": 859
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001417319070703066,
+      "loss": 0.5684,
+      "step": 860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014160102271390933,
+      "loss": 0.5782,
+      "step": 861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014147005212040374,
+      "loss": 0.6022,
+      "step": 862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014133899556128624,
+      "loss": 0.7854,
+      "step": 863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014120785330823128,
+      "loss": 0.6847,
+      "step": 864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014107662563309114,
+      "loss": 0.6651,
+      "step": 865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000140945312807895,
+      "loss": 0.6497,
+      "step": 866
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014081391510484864,
+      "loss": 0.6519,
+      "step": 867
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001406824327963338,
+      "loss": 0.5676,
+      "step": 868
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001405508661549076,
+      "loss": 0.7146,
+      "step": 869
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014041921545330193,
+      "loss": 0.6473,
+      "step": 870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014028748096442292,
+      "loss": 0.5938,
+      "step": 871
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014015566296135047,
+      "loss": 0.6573,
+      "step": 872
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014002376171733759,
+      "loss": 0.5073,
+      "step": 873
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013989177750580974,
+      "loss": 0.8486,
+      "step": 874
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001397597106003645,
+      "loss": 0.6166,
+      "step": 875
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013962756127477077,
+      "loss": 0.5624,
+      "step": 876
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013949532980296841,
+      "loss": 0.6444,
+      "step": 877
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013936301645906744,
+      "loss": 0.6191,
+      "step": 878
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013923062151734769,
+      "loss": 0.4869,
+      "step": 879
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001390981452522581,
+      "loss": 0.6693,
+      "step": 880
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013896558793841622,
+      "loss": 0.608,
+      "step": 881
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013883294985060754,
+      "loss": 0.5686,
+      "step": 882
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001387002312637851,
+      "loss": 0.6524,
+      "step": 883
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013856743245306867,
+      "loss": 0.6816,
+      "step": 884
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013843455369374447,
+      "loss": 0.6014,
+      "step": 885
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001383015952612644,
+      "loss": 0.6389,
+      "step": 886
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001381685574312454,
+      "loss": 0.7432,
+      "step": 887
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013803544047946917,
+      "loss": 0.5692,
+      "step": 888
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013790224468188136,
+      "loss": 0.8574,
+      "step": 889
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013776897031459104,
+      "loss": 0.6702,
+      "step": 890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013763561765387015,
+      "loss": 0.7442,
+      "step": 891
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013750218697615297,
+      "loss": 0.652,
+      "step": 892
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001373686785580355,
+      "loss": 0.746,
+      "step": 893
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013723509267627479,
+      "loss": 0.7589,
+      "step": 894
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001371014296077887,
+      "loss": 0.5822,
+      "step": 895
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001369676896296548,
+      "loss": 0.5692,
+      "step": 896
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013683387301911037,
+      "loss": 0.6453,
+      "step": 897
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001366999800535513,
+      "loss": 0.5916,
+      "step": 898
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000136566011010532,
+      "loss": 0.7739,
+      "step": 899
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013643196616776432,
+      "loss": 0.5407,
+      "step": 900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001362978458031175,
+      "loss": 0.4962,
+      "step": 901
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013616365019461716,
+      "loss": 0.8785,
+      "step": 902
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013602937962044496,
+      "loss": 0.5731,
+      "step": 903
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013589503435893792,
+      "loss": 0.5532,
+      "step": 904
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000135760614688588,
+      "loss": 0.6762,
+      "step": 905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013562612088804127,
+      "loss": 0.6163,
+      "step": 906
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001354915532360975,
+      "loss": 0.5938,
+      "step": 907
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013535691201170958,
+      "loss": 0.7454,
+      "step": 908
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013522219749398295,
+      "loss": 0.5695,
+      "step": 909
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013508740996217493,
+      "loss": 0.7036,
+      "step": 910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013495254969569413,
+      "loss": 0.5777,
+      "step": 911
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013481761697410009,
+      "loss": 0.5765,
+      "step": 912
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013468261207710242,
+      "loss": 0.6484,
+      "step": 913
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013454753528456038,
+      "loss": 0.6945,
+      "step": 914
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013441238687648236,
+      "loss": 0.632,
+      "step": 915
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013427716713302506,
+      "loss": 0.7317,
+      "step": 916
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013414187633449312,
+      "loss": 0.8531,
+      "step": 917
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013400651476133844,
+      "loss": 0.5767,
+      "step": 918
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013387108269415977,
+      "loss": 0.5668,
+      "step": 919
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013373558041370178,
+      "loss": 0.6109,
+      "step": 920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013360000820085484,
+      "loss": 0.7598,
+      "step": 921
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013346436633665428,
+      "loss": 0.6683,
+      "step": 922
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001333286551022797,
+      "loss": 0.6422,
+      "step": 923
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013319287477905464,
+      "loss": 0.5122,
+      "step": 924
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013305702564844578,
+      "loss": 0.7089,
+      "step": 925
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013292110799206243,
+      "loss": 0.4886,
+      "step": 926
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013278512209165602,
+      "loss": 0.6161,
+      "step": 927
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001326490682291193,
+      "loss": 0.534,
+      "step": 928
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013251294668648611,
+      "loss": 0.5081,
+      "step": 929
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013237675774593045,
+      "loss": 0.6817,
+      "step": 930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013224050168976606,
+      "loss": 0.5695,
+      "step": 931
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001321041788004458,
+      "loss": 0.4673,
+      "step": 932
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013196778936056105,
+      "loss": 0.5672,
+      "step": 933
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013183133365284123,
+      "loss": 0.6797,
+      "step": 934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013169481196015305,
+      "loss": 0.6979,
+      "step": 935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013155822456550006,
+      "loss": 0.7084,
+      "step": 936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013142157175202198,
+      "loss": 0.6153,
+      "step": 937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013128485380299408,
+      "loss": 0.5626,
+      "step": 938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013114807100182677,
+      "loss": 0.7549,
+      "step": 939
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013101122363206488,
+      "loss": 0.7096,
+      "step": 940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013087431197738694,
+      "loss": 0.7002,
+      "step": 941
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013073733632160494,
+      "loss": 0.7507,
+      "step": 942
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013060029694866342,
+      "loss": 0.4978,
+      "step": 943
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013046319414263902,
+      "loss": 0.6454,
+      "step": 944
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013032602818773985,
+      "loss": 0.8764,
+      "step": 945
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013018879936830505,
+      "loss": 0.591,
+      "step": 946
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001300515079688039,
+      "loss": 0.587,
+      "step": 947
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012991415427383556,
+      "loss": 0.7213,
+      "step": 948
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001297767385681282,
+      "loss": 0.6097,
+      "step": 949
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012963926113653863,
+      "loss": 0.6996,
+      "step": 950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001295017222640516,
+      "loss": 0.7174,
+      "step": 951
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012936412223577915,
+      "loss": 0.7143,
+      "step": 952
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012922646133696017,
+      "loss": 0.5663,
+      "step": 953
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012908873985295968,
+      "loss": 0.7875,
+      "step": 954
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012895095806926836,
+      "loss": 0.4114,
+      "step": 955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012881311627150187,
+      "loss": 0.5148,
+      "step": 956
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012867521474540017,
+      "loss": 0.5049,
+      "step": 957
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012853725377682718,
+      "loss": 0.6965,
+      "step": 958
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012839923365176993,
+      "loss": 0.6105,
+      "step": 959
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001282611546563382,
+      "loss": 0.6577,
+      "step": 960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012812301707676367,
+      "loss": 0.6559,
+      "step": 961
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012798482119939956,
+      "loss": 0.6486,
+      "step": 962
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001278465673107199,
+      "loss": 0.7116,
+      "step": 963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012770825569731895,
+      "loss": 0.489,
+      "step": 964
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001275698866459107,
+      "loss": 0.4855,
+      "step": 965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012743146044332817,
+      "loss": 0.7555,
+      "step": 966
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012729297737652283,
+      "loss": 0.5122,
+      "step": 967
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012715443773256402,
+      "loss": 0.5268,
+      "step": 968
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012701584179863843,
+      "loss": 0.6619,
+      "step": 969
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001268771898620494,
+      "loss": 0.5818,
+      "step": 970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012673848221021633,
+      "loss": 0.6126,
+      "step": 971
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012659971913067414,
+      "loss": 0.6435,
+      "step": 972
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012646090091107266,
+      "loss": 0.6286,
+      "step": 973
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000126322027839176,
+      "loss": 0.4635,
+      "step": 974
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012618310020286202,
+      "loss": 0.6147,
+      "step": 975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012604411829012166,
+      "loss": 0.7019,
+      "step": 976
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012590508238905833,
+      "loss": 0.6354,
+      "step": 977
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012576599278788742,
+      "loss": 0.7266,
+      "step": 978
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001256268497749356,
+      "loss": 0.5729,
+      "step": 979
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012548765363864036,
+      "loss": 0.4397,
+      "step": 980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012534840466754912,
+      "loss": 0.656,
+      "step": 981
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012520910315031895,
+      "loss": 0.6567,
+      "step": 982
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001250697493757159,
+      "loss": 0.5867,
+      "step": 983
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001249303436326142,
+      "loss": 0.701,
+      "step": 984
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012479088620999593,
+      "loss": 0.6854,
+      "step": 985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001246513773969502,
+      "loss": 0.5732,
+      "step": 986
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001245118174826728,
+      "loss": 0.7276,
+      "step": 987
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012437220675646523,
+      "loss": 0.6758,
+      "step": 988
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012423254550773458,
+      "loss": 0.5307,
+      "step": 989
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012409283402599238,
+      "loss": 0.6631,
+      "step": 990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012395307260085458,
+      "loss": 0.5795,
+      "step": 991
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001238132615220405,
+      "loss": 0.477,
+      "step": 992
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012367340107937236,
+      "loss": 0.621,
+      "step": 993
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001235334915627748,
+      "loss": 0.6521,
+      "step": 994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001233935332622742,
+      "loss": 0.6035,
+      "step": 995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012325352646799795,
+      "loss": 0.5742,
+      "step": 996
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012311347147017407,
+      "loss": 0.638,
+      "step": 997
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012297336855913047,
+      "loss": 0.7119,
+      "step": 998
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012283321802529436,
+      "loss": 0.6114,
+      "step": 999
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012269302015919172,
+      "loss": 0.6364,
+      "step": 1000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012255277525144664,
+      "loss": 0.7205,
+      "step": 1001
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012241248359278064,
+      "loss": 0.5875,
+      "step": 1002
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001222721454740123,
+      "loss": 0.6314,
+      "step": 1003
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012213176118605637,
+      "loss": 0.6022,
+      "step": 1004
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012199133101992339,
+      "loss": 0.6624,
+      "step": 1005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012185085526671893,
+      "loss": 0.5933,
+      "step": 1006
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001217103342176432,
+      "loss": 0.6812,
+      "step": 1007
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012156976816399013,
+      "loss": 0.64,
+      "step": 1008
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012142915739714707,
+      "loss": 0.5847,
+      "step": 1009
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012128850220859397,
+      "loss": 0.4929,
+      "step": 1010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012114780288990297,
+      "loss": 0.6001,
+      "step": 1011
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012100705973273765,
+      "loss": 0.6169,
+      "step": 1012
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012086627302885234,
+      "loss": 0.6945,
+      "step": 1013
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012072544307009184,
+      "loss": 0.6537,
+      "step": 1014
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012058457014839046,
+      "loss": 0.626,
+      "step": 1015
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012044365455577171,
+      "loss": 0.6657,
+      "step": 1016
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012030269658434739,
+      "loss": 0.6808,
+      "step": 1017
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012016169652631726,
+      "loss": 0.5208,
+      "step": 1018
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012002065467396832,
+      "loss": 0.5952,
+      "step": 1019
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011987957131967418,
+      "loss": 0.6806,
+      "step": 1020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001197384467558945,
+      "loss": 0.6073,
+      "step": 1021
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011959728127517434,
+      "loss": 0.5898,
+      "step": 1022
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011945607517014356,
+      "loss": 0.6778,
+      "step": 1023
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011931482873351632,
+      "loss": 0.5788,
+      "step": 1024
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011917354225809034,
+      "loss": 0.7229,
+      "step": 1025
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011903221603674628,
+      "loss": 0.5356,
+      "step": 1026
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011889085036244723,
+      "loss": 0.5517,
+      "step": 1027
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011874944552823817,
+      "loss": 0.7379,
+      "step": 1028
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011860800182724507,
+      "loss": 0.4891,
+      "step": 1029
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011846651955267463,
+      "loss": 0.6063,
+      "step": 1030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011832499899781333,
+      "loss": 0.5974,
+      "step": 1031
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011818344045602727,
+      "loss": 0.7582,
+      "step": 1032
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011804184422076105,
+      "loss": 0.5445,
+      "step": 1033
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011790021058553751,
+      "loss": 0.6351,
+      "step": 1034
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011775853984395699,
+      "loss": 0.7757,
+      "step": 1035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011761683228969682,
+      "loss": 0.5035,
+      "step": 1036
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011747508821651059,
+      "loss": 0.6516,
+      "step": 1037
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011733330791822749,
+      "loss": 0.6328,
+      "step": 1038
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011719149168875203,
+      "loss": 0.5434,
+      "step": 1039
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011704963982206299,
+      "loss": 0.6575,
+      "step": 1040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011690775261221316,
+      "loss": 0.5159,
+      "step": 1041
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011676583035332853,
+      "loss": 0.641,
+      "step": 1042
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011662387333960774,
+      "loss": 0.6043,
+      "step": 1043
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011648188186532154,
+      "loss": 0.779,
+      "step": 1044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011633985622481202,
+      "loss": 0.7303,
+      "step": 1045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011619779671249223,
+      "loss": 0.7594,
+      "step": 1046
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011605570362284524,
+      "loss": 0.5406,
+      "step": 1047
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011591357725042393,
+      "loss": 0.591,
+      "step": 1048
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011577141788985004,
+      "loss": 0.7236,
+      "step": 1049
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011562922583581375,
+      "loss": 0.4595,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011548700138307294,
+      "loss": 0.6517,
+      "step": 1051
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001153447448264528,
+      "loss": 0.7733,
+      "step": 1052
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011520245646084491,
+      "loss": 0.793,
+      "step": 1053
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011506013658120687,
+      "loss": 0.6748,
+      "step": 1054
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011491778548256157,
+      "loss": 0.5877,
+      "step": 1055
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011477540345999669,
+      "loss": 0.7135,
+      "step": 1056
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011463299080866397,
+      "loss": 0.6537,
+      "step": 1057
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011449054782377855,
+      "loss": 0.5969,
+      "step": 1058
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011434807480061858,
+      "loss": 0.6718,
+      "step": 1059
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011420557203452444,
+      "loss": 0.6952,
+      "step": 1060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011406303982089816,
+      "loss": 0.6231,
+      "step": 1061
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011392047845520282,
+      "loss": 0.6124,
+      "step": 1062
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011377788823296186,
+      "loss": 0.5751,
+      "step": 1063
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011363526944975867,
+      "loss": 0.6537,
+      "step": 1064
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011349262240123575,
+      "loss": 0.698,
+      "step": 1065
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011334994738309423,
+      "loss": 0.732,
+      "step": 1066
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011320724469109321,
+      "loss": 0.6665,
+      "step": 1067
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011306451462104912,
+      "loss": 0.5394,
+      "step": 1068
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001129217574688352,
+      "loss": 0.5519,
+      "step": 1069
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011277897353038085,
+      "loss": 0.5069,
+      "step": 1070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011263616310167089,
+      "loss": 0.735,
+      "step": 1071
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011249332647874513,
+      "loss": 0.6731,
+      "step": 1072
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011235046395769772,
+      "loss": 0.4655,
+      "step": 1073
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011220757583467643,
+      "loss": 0.5297,
+      "step": 1074
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011206466240588208,
+      "loss": 0.641,
+      "step": 1075
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011192172396756797,
+      "loss": 0.5489,
+      "step": 1076
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011177876081603933,
+      "loss": 0.6024,
+      "step": 1077
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011163577324765248,
+      "loss": 0.598,
+      "step": 1078
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011149276155881445,
+      "loss": 0.5836,
+      "step": 1079
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011134972604598224,
+      "loss": 0.6395,
+      "step": 1080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011120666700566224,
+      "loss": 0.6359,
+      "step": 1081
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011106358473440963,
+      "loss": 0.5316,
+      "step": 1082
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011092047952882767,
+      "loss": 0.511,
+      "step": 1083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011077735168556729,
+      "loss": 0.6941,
+      "step": 1084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001106342015013262,
+      "loss": 0.429,
+      "step": 1085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011049102927284857,
+      "loss": 0.6474,
+      "step": 1086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011034783529692416,
+      "loss": 0.7696,
+      "step": 1087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011020461987038781,
+      "loss": 0.5363,
+      "step": 1088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011006138329011895,
+      "loss": 0.569,
+      "step": 1089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010991812585304069,
+      "loss": 0.6635,
+      "step": 1090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010977484785611951,
+      "loss": 0.6465,
+      "step": 1091
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010963154959636438,
+      "loss": 0.4752,
+      "step": 1092
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010948823137082649,
+      "loss": 0.5174,
+      "step": 1093
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010934489347659816,
+      "loss": 0.4826,
+      "step": 1094
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010920153621081264,
+      "loss": 0.6196,
+      "step": 1095
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010905815987064328,
+      "loss": 0.7292,
+      "step": 1096
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010891476475330302,
+      "loss": 0.8333,
+      "step": 1097
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010877135115604372,
+      "loss": 0.4909,
+      "step": 1098
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010862791937615544,
+      "loss": 0.6642,
+      "step": 1099
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010848446971096606,
+      "loss": 0.5732,
+      "step": 1100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010834100245784049,
+      "loss": 0.6713,
+      "step": 1101
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010819751791418011,
+      "loss": 0.6065,
+      "step": 1102
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010805401637742208,
+      "loss": 0.7532,
+      "step": 1103
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010791049814503888,
+      "loss": 0.6216,
+      "step": 1104
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010776696351453755,
+      "loss": 0.6085,
+      "step": 1105
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010762341278345909,
+      "loss": 0.6474,
+      "step": 1106
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010747984624937799,
+      "loss": 0.5874,
+      "step": 1107
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010733626420990134,
+      "loss": 0.7047,
+      "step": 1108
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001071926669626685,
+      "loss": 0.7304,
+      "step": 1109
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001070490548053503,
+      "loss": 0.6759,
+      "step": 1110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010690542803564852,
+      "loss": 0.6558,
+      "step": 1111
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010676178695129513,
+      "loss": 0.6651,
+      "step": 1112
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010661813185005187,
+      "loss": 0.6029,
+      "step": 1113
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010647446302970954,
+      "loss": 0.6244,
+      "step": 1114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010633078078808731,
+      "loss": 0.5587,
+      "step": 1115
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010618708542303226,
+      "loss": 0.6508,
+      "step": 1116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010604337723241853,
+      "loss": 0.5252,
+      "step": 1117
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010589965651414704,
+      "loss": 0.5962,
+      "step": 1118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010575592356614452,
+      "loss": 0.6492,
+      "step": 1119
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010561217868636313,
+      "loss": 0.5548,
+      "step": 1120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010546842217277974,
+      "loss": 0.5842,
+      "step": 1121
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010532465432339538,
+      "loss": 0.4654,
+      "step": 1122
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010518087543623447,
+      "loss": 0.7025,
+      "step": 1123
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010503708580934442,
+      "loss": 0.7613,
+      "step": 1124
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010489328574079486,
+      "loss": 0.6232,
+      "step": 1125
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010474947552867706,
+      "loss": 0.5128,
+      "step": 1126
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001046056554711034,
+      "loss": 0.6643,
+      "step": 1127
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010446182586620648,
+      "loss": 0.5394,
+      "step": 1128
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010431798701213886,
+      "loss": 0.6618,
+      "step": 1129
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010417413920707222,
+      "loss": 0.6952,
+      "step": 1130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010403028274919682,
+      "loss": 0.6442,
+      "step": 1131
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010388641793672078,
+      "loss": 0.624,
+      "step": 1132
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010374254506786962,
+      "loss": 0.5736,
+      "step": 1133
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010359866444088555,
+      "loss": 0.6099,
+      "step": 1134
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001034547763540268,
+      "loss": 0.5952,
+      "step": 1135
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010331088110556717,
+      "loss": 0.633,
+      "step": 1136
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010316697899379515,
+      "loss": 0.654,
+      "step": 1137
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010302307031701364,
+      "loss": 0.5691,
+      "step": 1138
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000102879155373539,
+      "loss": 0.6231,
+      "step": 1139
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001027352344617007,
+      "loss": 0.6383,
+      "step": 1140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010259130787984043,
+      "loss": 0.5586,
+      "step": 1141
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010244737592631181,
+      "loss": 0.5455,
+      "step": 1142
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001023034388994795,
+      "loss": 0.5995,
+      "step": 1143
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010215949709771866,
+      "loss": 0.7648,
+      "step": 1144
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010201555081941439,
+      "loss": 0.6018,
+      "step": 1145
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010187160036296103,
+      "loss": 0.5134,
+      "step": 1146
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010172764602676166,
+      "loss": 0.6937,
+      "step": 1147
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010158368810922729,
+      "loss": 0.5815,
+      "step": 1148
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010143972690877646,
+      "loss": 0.6674,
+      "step": 1149
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010129576272383445,
+      "loss": 0.6097,
+      "step": 1150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010115179585283273,
+      "loss": 0.8013,
+      "step": 1151
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001010078265942084,
+      "loss": 0.6637,
+      "step": 1152
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001008638552464034,
+      "loss": 0.617,
+      "step": 1153
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010071988210786411,
+      "loss": 0.7793,
+      "step": 1154
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010057590747704058,
+      "loss": 0.528,
+      "step": 1155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010043193165238591,
+      "loss": 0.5632,
+      "step": 1156
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010028795493235575,
+      "loss": 0.5385,
+      "step": 1157
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010014397761540755,
+      "loss": 0.6662,
+      "step": 1158
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001,
+      "loss": 0.487,
+      "step": 1159
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985602238459247e-05,
+      "loss": 0.5168,
+      "step": 1160
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971204506764427e-05,
+      "loss": 0.5498,
+      "step": 1161
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956806834761411e-05,
+      "loss": 0.6681,
+      "step": 1162
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.942409252295946e-05,
+      "loss": 0.547,
+      "step": 1163
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92801178921359e-05,
+      "loss": 0.5374,
+      "step": 1164
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.913614475359662e-05,
+      "loss": 0.6526,
+      "step": 1165
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899217340579164e-05,
+      "loss": 0.7272,
+      "step": 1166
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884820414716728e-05,
+      "loss": 0.5324,
+      "step": 1167
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870423727616558e-05,
+      "loss": 0.8317,
+      "step": 1168
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856027309122355e-05,
+      "loss": 0.7425,
+      "step": 1169
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841631189077269e-05,
+      "loss": 0.5217,
+      "step": 1170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827235397323839e-05,
+      "loss": 0.5789,
+      "step": 1171
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812839963703899e-05,
+      "loss": 0.6686,
+      "step": 1172
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798444918058563e-05,
+      "loss": 0.6336,
+      "step": 1173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.784050290228134e-05,
+      "loss": 0.4526,
+      "step": 1174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769656110052053e-05,
+      "loss": 0.5258,
+      "step": 1175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75526240736882e-05,
+      "loss": 0.3994,
+      "step": 1176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.740869212015958e-05,
+      "loss": 0.6213,
+      "step": 1177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.726476553829932e-05,
+      "loss": 0.6058,
+      "step": 1178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712084462646098e-05,
+      "loss": 0.5478,
+      "step": 1179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69769296829864e-05,
+      "loss": 0.5626,
+      "step": 1180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.683302100620487e-05,
+      "loss": 0.5561,
+      "step": 1181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.668911889443285e-05,
+      "loss": 0.4083,
+      "step": 1182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654522364597319e-05,
+      "loss": 0.6723,
+      "step": 1183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640133555911449e-05,
+      "loss": 0.7471,
+      "step": 1184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62574549321304e-05,
+      "loss": 0.7165,
+      "step": 1185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611358206327923e-05,
+      "loss": 0.5743,
+      "step": 1186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.596971725080319e-05,
+      "loss": 0.6884,
+      "step": 1187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.582586079292778e-05,
+      "loss": 0.5175,
+      "step": 1188
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.568201298786119e-05,
+      "loss": 0.5781,
+      "step": 1189
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.553817413379356e-05,
+      "loss": 0.6548,
+      "step": 1190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.539434452889664e-05,
+      "loss": 0.6108,
+      "step": 1191
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.525052447132294e-05,
+      "loss": 0.609,
+      "step": 1192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510671425920516e-05,
+      "loss": 0.728,
+      "step": 1193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496291419065561e-05,
+      "loss": 0.5119,
+      "step": 1194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481912456376555e-05,
+      "loss": 0.5221,
+      "step": 1195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467534567660466e-05,
+      "loss": 0.6592,
+      "step": 1196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453157782722027e-05,
+      "loss": 0.5713,
+      "step": 1197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43878213136369e-05,
+      "loss": 0.7426,
+      "step": 1198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42440764338555e-05,
+      "loss": 0.6154,
+      "step": 1199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410034348585298e-05,
+      "loss": 0.6424,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395662276758147e-05,
+      "loss": 0.5176,
+      "step": 1201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381291457696779e-05,
+      "loss": 0.6709,
+      "step": 1202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36692192119127e-05,
+      "loss": 0.658,
+      "step": 1203
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.352553697029048e-05,
+      "loss": 0.5497,
+      "step": 1204
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338186814994814e-05,
+      "loss": 0.823,
+      "step": 1205
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323821304870489e-05,
+      "loss": 0.7548,
+      "step": 1206
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309457196435153e-05,
+      "loss": 0.5285,
+      "step": 1207
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295094519464972e-05,
+      "loss": 0.5574,
+      "step": 1208
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280733303733151e-05,
+      "loss": 0.6371,
+      "step": 1209
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266373579009867e-05,
+      "loss": 0.7519,
+      "step": 1210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252015375062206e-05,
+      "loss": 0.6185,
+      "step": 1211
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.237658721654092e-05,
+      "loss": 0.5116,
+      "step": 1212
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223303648546249e-05,
+      "loss": 0.5602,
+      "step": 1213
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208950185496114e-05,
+      "loss": 0.5772,
+      "step": 1214
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194598362257791e-05,
+      "loss": 0.6236,
+      "step": 1215
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180248208581994e-05,
+      "loss": 0.6236,
+      "step": 1216
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165899754215955e-05,
+      "loss": 0.6082,
+      "step": 1217
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.151553028903396e-05,
+      "loss": 0.728,
+      "step": 1218
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137208062384456e-05,
+      "loss": 0.6393,
+      "step": 1219
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122864884395633e-05,
+      "loss": 0.6315,
+      "step": 1220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.1085235246697e-05,
+      "loss": 0.6592,
+      "step": 1221
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094184012935674e-05,
+      "loss": 0.5673,
+      "step": 1222
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079846378918737e-05,
+      "loss": 0.5315,
+      "step": 1223
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065510652340185e-05,
+      "loss": 0.6109,
+      "step": 1224
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.051176862917356e-05,
+      "loss": 0.6455,
+      "step": 1225
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.036845040363562e-05,
+      "loss": 0.6155,
+      "step": 1226
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022515214388053e-05,
+      "loss": 0.7635,
+      "step": 1227
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008187414695932e-05,
+      "loss": 0.4373,
+      "step": 1228
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.993861670988111e-05,
+      "loss": 0.6805,
+      "step": 1229
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979538012961221e-05,
+      "loss": 0.6403,
+      "step": 1230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965216470307587e-05,
+      "loss": 0.731,
+      "step": 1231
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950897072715144e-05,
+      "loss": 0.6969,
+      "step": 1232
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93657984986738e-05,
+      "loss": 0.719,
+      "step": 1233
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.922264831443274e-05,
+      "loss": 0.585,
+      "step": 1234
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907952047117234e-05,
+      "loss": 0.7855,
+      "step": 1235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89364152655904e-05,
+      "loss": 0.648,
+      "step": 1236
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.879333299433777e-05,
+      "loss": 0.8423,
+      "step": 1237
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865027395401778e-05,
+      "loss": 0.6295,
+      "step": 1238
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850723844118556e-05,
+      "loss": 0.5245,
+      "step": 1239
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836422675234754e-05,
+      "loss": 0.5022,
+      "step": 1240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82212391839607e-05,
+      "loss": 0.5885,
+      "step": 1241
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807827603243204e-05,
+      "loss": 0.5466,
+      "step": 1242
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.793533759411796e-05,
+      "loss": 0.4989,
+      "step": 1243
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.779242416532361e-05,
+      "loss": 0.5479,
+      "step": 1244
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764953604230229e-05,
+      "loss": 0.6424,
+      "step": 1245
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.750667352125487e-05,
+      "loss": 0.5221,
+      "step": 1246
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736383689832914e-05,
+      "loss": 0.5971,
+      "step": 1247
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.722102646961919e-05,
+      "loss": 0.4959,
+      "step": 1248
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.707824253116481e-05,
+      "loss": 0.3366,
+      "step": 1249
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69354853789509e-05,
+      "loss": 0.3726,
+      "step": 1250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.679275530890681e-05,
+      "loss": 0.4428,
+      "step": 1251
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.665005261690579e-05,
+      "loss": 0.5772,
+      "step": 1252
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650737759876427e-05,
+      "loss": 0.6104,
+      "step": 1253
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636473055024134e-05,
+      "loss": 0.6651,
+      "step": 1254
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.622211176703815e-05,
+      "loss": 0.6323,
+      "step": 1255
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607952154479723e-05,
+      "loss": 0.6176,
+      "step": 1256
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.593696017910186e-05,
+      "loss": 0.5383,
+      "step": 1257
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.579442796547558e-05,
+      "loss": 0.6834,
+      "step": 1258
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565192519938145e-05,
+      "loss": 0.5084,
+      "step": 1259
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550945217622146e-05,
+      "loss": 0.4672,
+      "step": 1260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536700919133609e-05,
+      "loss": 0.5449,
+      "step": 1261
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522459654000332e-05,
+      "loss": 0.5759,
+      "step": 1262
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.508221451743844e-05,
+      "loss": 0.4519,
+      "step": 1263
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.493986341879314e-05,
+      "loss": 0.55,
+      "step": 1264
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479754353915514e-05,
+      "loss": 0.4691,
+      "step": 1265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465525517354724e-05,
+      "loss": 0.7719,
+      "step": 1266
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.451299861692707e-05,
+      "loss": 0.5472,
+      "step": 1267
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.437077416418627e-05,
+      "loss": 0.5773,
+      "step": 1268
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.422858211014997e-05,
+      "loss": 0.6816,
+      "step": 1269
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408642274957612e-05,
+      "loss": 0.6567,
+      "step": 1270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39442963771548e-05,
+      "loss": 0.4218,
+      "step": 1271
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380220328750781e-05,
+      "loss": 0.8275,
+      "step": 1272
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366014377518799e-05,
+      "loss": 0.52,
+      "step": 1273
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.351811813467851e-05,
+      "loss": 0.5299,
+      "step": 1274
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.337612666039227e-05,
+      "loss": 0.7501,
+      "step": 1275
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32341696466715e-05,
+      "loss": 0.5759,
+      "step": 1276
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.309224738778685e-05,
+      "loss": 0.6248,
+      "step": 1277
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295036017793702e-05,
+      "loss": 0.521,
+      "step": 1278
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2808508311248e-05,
+      "loss": 0.6641,
+      "step": 1279
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266669208177252e-05,
+      "loss": 0.6625,
+      "step": 1280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252491178348945e-05,
+      "loss": 0.6409,
+      "step": 1281
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.238316771030318e-05,
+      "loss": 0.6664,
+      "step": 1282
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224146015604302e-05,
+      "loss": 0.7011,
+      "step": 1283
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209978941446252e-05,
+      "loss": 0.756,
+      "step": 1284
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195815577923897e-05,
+      "loss": 0.6714,
+      "step": 1285
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181655954397275e-05,
+      "loss": 0.4277,
+      "step": 1286
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167500100218666e-05,
+      "loss": 0.5623,
+      "step": 1287
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.153348044732543e-05,
+      "loss": 0.6501,
+      "step": 1288
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139199817275496e-05,
+      "loss": 0.5333,
+      "step": 1289
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125055447176186e-05,
+      "loss": 0.505,
+      "step": 1290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.110914963755277e-05,
+      "loss": 0.6359,
+      "step": 1291
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096778396325377e-05,
+      "loss": 0.5734,
+      "step": 1292
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.082645774190971e-05,
+      "loss": 0.4936,
+      "step": 1293
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.068517126648369e-05,
+      "loss": 0.5992,
+      "step": 1294
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.054392482985646e-05,
+      "loss": 0.3389,
+      "step": 1295
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04027187248257e-05,
+      "loss": 0.4518,
+      "step": 1296
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026155324410555e-05,
+      "loss": 0.5322,
+      "step": 1297
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012042868032585e-05,
+      "loss": 0.5758,
+      "step": 1298
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99793453260317e-05,
+      "loss": 0.8122,
+      "step": 1299
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983830347368276e-05,
+      "loss": 0.5451,
+      "step": 1300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969730341565262e-05,
+      "loss": 0.7109,
+      "step": 1301
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955634544422834e-05,
+      "loss": 0.5986,
+      "step": 1302
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941542985160955e-05,
+      "loss": 0.5262,
+      "step": 1303
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927455692990818e-05,
+      "loss": 0.587,
+      "step": 1304
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913372697114764e-05,
+      "loss": 0.5661,
+      "step": 1305
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899294026726241e-05,
+      "loss": 0.6228,
+      "step": 1306
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885219711009704e-05,
+      "loss": 0.5837,
+      "step": 1307
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871149779140604e-05,
+      "loss": 0.6917,
+      "step": 1308
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857084260285294e-05,
+      "loss": 0.6235,
+      "step": 1309
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843023183600988e-05,
+      "loss": 0.5925,
+      "step": 1310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.828966578235685e-05,
+      "loss": 0.4964,
+      "step": 1311
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81491447332811e-05,
+      "loss": 0.5553,
+      "step": 1312
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800866898007663e-05,
+      "loss": 0.6669,
+      "step": 1313
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.786823881394364e-05,
+      "loss": 0.4613,
+      "step": 1314
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772785452598774e-05,
+      "loss": 0.6099,
+      "step": 1315
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.758751640721937e-05,
+      "loss": 0.6317,
+      "step": 1316
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744722474855336e-05,
+      "loss": 0.7409,
+      "step": 1317
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730697984080827e-05,
+      "loss": 0.6339,
+      "step": 1318
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716678197470564e-05,
+      "loss": 0.46,
+      "step": 1319
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702663144086957e-05,
+      "loss": 0.6124,
+      "step": 1320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688652852982595e-05,
+      "loss": 0.6611,
+      "step": 1321
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674647353200208e-05,
+      "loss": 0.5532,
+      "step": 1322
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660646673772582e-05,
+      "loss": 0.4256,
+      "step": 1323
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.64665084372252e-05,
+      "loss": 0.5391,
+      "step": 1324
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632659892062765e-05,
+      "loss": 0.2778,
+      "step": 1325
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.618673847795953e-05,
+      "loss": 0.6374,
+      "step": 1326
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.604692739914542e-05,
+      "loss": 0.589,
+      "step": 1327
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590716597400761e-05,
+      "loss": 0.5692,
+      "step": 1328
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.576745449226547e-05,
+      "loss": 0.6065,
+      "step": 1329
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.562779324353477e-05,
+      "loss": 0.588,
+      "step": 1330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.548818251732723e-05,
+      "loss": 0.732,
+      "step": 1331
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.53486226030498e-05,
+      "loss": 0.6655,
+      "step": 1332
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52091137900041e-05,
+      "loss": 0.4851,
+      "step": 1333
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506965636738583e-05,
+      "loss": 0.5523,
+      "step": 1334
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493025062428413e-05,
+      "loss": 0.4606,
+      "step": 1335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.479089684968106e-05,
+      "loss": 0.5971,
+      "step": 1336
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46515953324509e-05,
+      "loss": 0.3929,
+      "step": 1337
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451234636135969e-05,
+      "loss": 0.6812,
+      "step": 1338
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.437315022506439e-05,
+      "loss": 0.6789,
+      "step": 1339
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42340072121126e-05,
+      "loss": 0.4871,
+      "step": 1340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40949176109417e-05,
+      "loss": 0.4723,
+      "step": 1341
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.395588170987839e-05,
+      "loss": 0.5882,
+      "step": 1342
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.381689979713799e-05,
+      "loss": 0.6154,
+      "step": 1343
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367797216082402e-05,
+      "loss": 0.4924,
+      "step": 1344
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353909908892736e-05,
+      "loss": 0.6011,
+      "step": 1345
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.340028086932587e-05,
+      "loss": 0.619,
+      "step": 1346
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.326151778978372e-05,
+      "loss": 0.5211,
+      "step": 1347
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.312281013795064e-05,
+      "loss": 0.2716,
+      "step": 1348
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.298415820136159e-05,
+      "loss": 0.4895,
+      "step": 1349
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284556226743598e-05,
+      "loss": 0.5577,
+      "step": 1350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.270702262347722e-05,
+      "loss": 0.6368,
+      "step": 1351
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256853955667187e-05,
+      "loss": 0.4952,
+      "step": 1352
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243011335408931e-05,
+      "loss": 0.5676,
+      "step": 1353
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229174430268104e-05,
+      "loss": 0.5283,
+      "step": 1354
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215343268928012e-05,
+      "loss": 0.6,
+      "step": 1355
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201517880060049e-05,
+      "loss": 0.5529,
+      "step": 1356
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.187698292323638e-05,
+      "loss": 0.6129,
+      "step": 1357
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173884534366182e-05,
+      "loss": 0.6357,
+      "step": 1358
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160076634823007e-05,
+      "loss": 0.5877,
+      "step": 1359
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.146274622317288e-05,
+      "loss": 0.7829,
+      "step": 1360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.132478525459985e-05,
+      "loss": 0.3495,
+      "step": 1361
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.118688372849815e-05,
+      "loss": 0.6815,
+      "step": 1362
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.104904193073164e-05,
+      "loss": 0.6667,
+      "step": 1363
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091126014704032e-05,
+      "loss": 0.5822,
+      "step": 1364
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077353866303987e-05,
+      "loss": 0.5527,
+      "step": 1365
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063587776422088e-05,
+      "loss": 0.5849,
+      "step": 1366
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049827773594842e-05,
+      "loss": 0.4935,
+      "step": 1367
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036073886346138e-05,
+      "loss": 0.6335,
+      "step": 1368
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022326143187182e-05,
+      "loss": 0.6772,
+      "step": 1369
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008584572616448e-05,
+      "loss": 0.5518,
+      "step": 1370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994849203119612e-05,
+      "loss": 0.4049,
+      "step": 1371
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.981120063169499e-05,
+      "loss": 0.6957,
+      "step": 1372
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.967397181226016e-05,
+      "loss": 0.455,
+      "step": 1373
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.953680585736105e-05,
+      "loss": 0.5583,
+      "step": 1374
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.939970305133663e-05,
+      "loss": 0.5528,
+      "step": 1375
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926266367839508e-05,
+      "loss": 0.6318,
+      "step": 1376
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.912568802261307e-05,
+      "loss": 0.6801,
+      "step": 1377
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.898877636793517e-05,
+      "loss": 0.4815,
+      "step": 1378
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.885192899817324e-05,
+      "loss": 0.3488,
+      "step": 1379
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871514619700594e-05,
+      "loss": 0.6275,
+      "step": 1380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857842824797806e-05,
+      "loss": 0.6648,
+      "step": 1381
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844177543449997e-05,
+      "loss": 0.5768,
+      "step": 1382
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830518803984698e-05,
+      "loss": 0.6883,
+      "step": 1383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.816866634715881e-05,
+      "loss": 0.6354,
+      "step": 1384
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803221063943898e-05,
+      "loss": 0.5715,
+      "step": 1385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789582119955424e-05,
+      "loss": 0.6387,
+      "step": 1386
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.775949831023398e-05,
+      "loss": 0.5986,
+      "step": 1387
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.762324225406957e-05,
+      "loss": 0.5175,
+      "step": 1388
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74870533135139e-05,
+      "loss": 0.3076,
+      "step": 1389
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.73509317708807e-05,
+      "loss": 0.4861,
+      "step": 1390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7214877908344e-05,
+      "loss": 0.5329,
+      "step": 1391
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70788920079376e-05,
+      "loss": 0.6205,
+      "step": 1392
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694297435155426e-05,
+      "loss": 0.4511,
+      "step": 1393
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680712522094537e-05,
+      "loss": 0.6649,
+      "step": 1394
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667134489772029e-05,
+      "loss": 0.5394,
+      "step": 1395
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653563366334577e-05,
+      "loss": 0.5804,
+      "step": 1396
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639999179914518e-05,
+      "loss": 0.5859,
+      "step": 1397
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.626441958629826e-05,
+      "loss": 0.4855,
+      "step": 1398
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.612891730584026e-05,
+      "loss": 0.6424,
+      "step": 1399
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599348523866155e-05,
+      "loss": 0.4478,
+      "step": 1400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585812366550694e-05,
+      "loss": 0.508,
+      "step": 1401
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572283286697496e-05,
+      "loss": 0.6592,
+      "step": 1402
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.558761312351765e-05,
+      "loss": 0.5251,
+      "step": 1403
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54524647154396e-05,
+      "loss": 0.5836,
+      "step": 1404
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531738792289763e-05,
+      "loss": 0.4265,
+      "step": 1405
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518238302589994e-05,
+      "loss": 0.6757,
+      "step": 1406
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.504745030430588e-05,
+      "loss": 0.5816,
+      "step": 1407
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491259003782511e-05,
+      "loss": 0.6513,
+      "step": 1408
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477780250601704e-05,
+      "loss": 0.4129,
+      "step": 1409
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464308798829043e-05,
+      "loss": 0.5116,
+      "step": 1410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450844676390252e-05,
+      "loss": 0.5739,
+      "step": 1411
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437387911195875e-05,
+      "loss": 0.4302,
+      "step": 1412
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4239385311412e-05,
+      "loss": 0.4259,
+      "step": 1413
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.410496564106207e-05,
+      "loss": 0.4294,
+      "step": 1414
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397062037955507e-05,
+      "loss": 0.5428,
+      "step": 1415
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.383634980538286e-05,
+      "loss": 0.6312,
+      "step": 1416
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370215419688252e-05,
+      "loss": 0.5246,
+      "step": 1417
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356803383223569e-05,
+      "loss": 0.5685,
+      "step": 1418
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.343398898946805e-05,
+      "loss": 0.5887,
+      "step": 1419
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33000199464487e-05,
+      "loss": 0.6911,
+      "step": 1420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316612698088967e-05,
+      "loss": 0.6935,
+      "step": 1421
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30323103703452e-05,
+      "loss": 0.4703,
+      "step": 1422
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289857039221135e-05,
+      "loss": 0.6793,
+      "step": 1423
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276490732372522e-05,
+      "loss": 0.4813,
+      "step": 1424
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.263132144196455e-05,
+      "loss": 0.5421,
+      "step": 1425
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249781302384705e-05,
+      "loss": 0.6866,
+      "step": 1426
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236438234612987e-05,
+      "loss": 0.4021,
+      "step": 1427
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2231029685409e-05,
+      "loss": 0.48,
+      "step": 1428
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209775531811868e-05,
+      "loss": 0.5798,
+      "step": 1429
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196455952053084e-05,
+      "loss": 0.577,
+      "step": 1430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183144256875462e-05,
+      "loss": 0.3458,
+      "step": 1431
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.169840473873565e-05,
+      "loss": 0.5413,
+      "step": 1432
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156544630625555e-05,
+      "loss": 0.5556,
+      "step": 1433
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143256754693134e-05,
+      "loss": 0.5008,
+      "step": 1434
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129976873621494e-05,
+      "loss": 0.5195,
+      "step": 1435
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116705014939246e-05,
+      "loss": 0.6053,
+      "step": 1436
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.103441206158382e-05,
+      "loss": 0.5239,
+      "step": 1437
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090185474774192e-05,
+      "loss": 0.5045,
+      "step": 1438
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076937848265233e-05,
+      "loss": 0.6813,
+      "step": 1439
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063698354093255e-05,
+      "loss": 0.5776,
+      "step": 1440
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0504670197031635e-05,
+      "loss": 0.6314,
+      "step": 1441
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037243872522924e-05,
+      "loss": 0.4862,
+      "step": 1442
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024028939963552e-05,
+      "loss": 0.5236,
+      "step": 1443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.010822249419027e-05,
+      "loss": 0.6815,
+      "step": 1444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.997623828266243e-05,
+      "loss": 0.5533,
+      "step": 1445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.984433703864956e-05,
+      "loss": 0.5213,
+      "step": 1446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9712519035577105e-05,
+      "loss": 0.4664,
+      "step": 1447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.95807845466981e-05,
+      "loss": 0.5657,
+      "step": 1448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944913384509241e-05,
+      "loss": 0.53,
+      "step": 1449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931756720366621e-05,
+      "loss": 0.4287,
+      "step": 1450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.918608489515136e-05,
+      "loss": 0.6259,
+      "step": 1451
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9054687192105026e-05,
+      "loss": 0.5912,
+      "step": 1452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892337436690889e-05,
+      "loss": 0.5173,
+      "step": 1453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8792146691768726e-05,
+      "loss": 0.7255,
+      "step": 1454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866100443871381e-05,
+      "loss": 0.6497,
+      "step": 1455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852994787959628e-05,
+      "loss": 0.6155,
+      "step": 1456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.839897728609071e-05,
+      "loss": 0.43,
+      "step": 1457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8268092929693405e-05,
+      "loss": 0.703,
+      "step": 1458
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8137295081721986e-05,
+      "loss": 0.5811,
+      "step": 1459
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.800658401331467e-05,
+      "loss": 0.5184,
+      "step": 1460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.787595999542974e-05,
+      "loss": 0.6068,
+      "step": 1461
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774542329884518e-05,
+      "loss": 0.6412,
+      "step": 1462
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761497419415789e-05,
+      "loss": 0.6069,
+      "step": 1463
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748461295178315e-05,
+      "loss": 0.3996,
+      "step": 1464
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735433984195413e-05,
+      "loss": 0.4652,
+      "step": 1465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.722415513472128e-05,
+      "loss": 0.6558,
+      "step": 1466
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.709405909995187e-05,
+      "loss": 0.8196,
+      "step": 1467
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696405200732939e-05,
+      "loss": 0.5282,
+      "step": 1468
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683413412635272e-05,
+      "loss": 0.6799,
+      "step": 1469
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670430572633607e-05,
+      "loss": 0.6125,
+      "step": 1470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657456707640806e-05,
+      "loss": 0.4694,
+      "step": 1471
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.644491844551121e-05,
+      "loss": 0.6762,
+      "step": 1472
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.631536010240159e-05,
+      "loss": 0.5733,
+      "step": 1473
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6185892315647856e-05,
+      "loss": 0.528,
+      "step": 1474
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.605651535363117e-05,
+      "loss": 0.7113,
+      "step": 1475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592722948454437e-05,
+      "loss": 0.6064,
+      "step": 1476
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.57980349763914e-05,
+      "loss": 0.5652,
+      "step": 1477
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.56689320969868e-05,
+      "loss": 0.721,
+      "step": 1478
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.553992111395532e-05,
+      "loss": 0.6228,
+      "step": 1479
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5411002294730996e-05,
+      "loss": 0.444,
+      "step": 1480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528217590655702e-05,
+      "loss": 0.6434,
+      "step": 1481
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.515344221648484e-05,
+      "loss": 0.6018,
+      "step": 1482
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5024801491373765e-05,
+      "loss": 0.603,
+      "step": 1483
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489625399789048e-05,
+      "loss": 0.6296,
+      "step": 1484
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4767800002508276e-05,
+      "loss": 0.5766,
+      "step": 1485
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463943977150674e-05,
+      "loss": 0.5813,
+      "step": 1486
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451117357097103e-05,
+      "loss": 0.6493,
+      "step": 1487
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438300166679134e-05,
+      "loss": 0.6133,
+      "step": 1488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425492432466254e-05,
+      "loss": 0.5598,
+      "step": 1489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412694181008329e-05,
+      "loss": 0.532,
+      "step": 1490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.399905438835584e-05,
+      "loss": 0.4597,
+      "step": 1491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387126232458522e-05,
+      "loss": 0.5993,
+      "step": 1492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374356588367877e-05,
+      "loss": 0.7077,
+      "step": 1493
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3615965330345716e-05,
+      "loss": 0.6094,
+      "step": 1494
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.348846092909643e-05,
+      "loss": 0.5547,
+      "step": 1495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.336105294424194e-05,
+      "loss": 0.6686,
+      "step": 1496
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.323374163989352e-05,
+      "loss": 0.6638,
+      "step": 1497
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310652727996188e-05,
+      "loss": 0.7447,
+      "step": 1498
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.29794101281569e-05,
+      "loss": 0.4319,
+      "step": 1499
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285239044798695e-05,
+      "loss": 0.6549,
+      "step": 1500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.272546850275817e-05,
+      "loss": 0.2075,
+      "step": 1501
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2598644555574326e-05,
+      "loss": 0.2402,
+      "step": 1502
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.247191886933587e-05,
+      "loss": 0.2263,
+      "step": 1503
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2345291706739684e-05,
+      "loss": 0.1424,
+      "step": 1504
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2218763330278354e-05,
+      "loss": 0.1403,
+      "step": 1505
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209233400223963e-05,
+      "loss": 0.2413,
+      "step": 1506
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196600398470605e-05,
+      "loss": 0.2223,
+      "step": 1507
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.183977353955427e-05,
+      "loss": 0.232,
+      "step": 1508
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1713642928454485e-05,
+      "loss": 0.1117,
+      "step": 1509
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1587612412869954e-05,
+      "loss": 0.1926,
+      "step": 1510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.14616822540564e-05,
+      "loss": 0.2347,
+      "step": 1511
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1335852713061594e-05,
+      "loss": 0.3358,
+      "step": 1512
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121012405072478e-05,
+      "loss": 0.2393,
+      "step": 1513
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108449652767584e-05,
+      "loss": 0.2168,
+      "step": 1514
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.095897040433524e-05,
+      "loss": 0.1965,
+      "step": 1515
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0833545940913183e-05,
+      "loss": 0.2803,
+      "step": 1516
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070822339740908e-05,
+      "loss": 0.2326,
+      "step": 1517
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0583003033611163e-05,
+      "loss": 0.3001,
+      "step": 1518
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0457885109095745e-05,
+      "loss": 0.1457,
+      "step": 1519
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0332869883226817e-05,
+      "loss": 0.2022,
+      "step": 1520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.020795761515556e-05,
+      "loss": 0.2256,
+      "step": 1521
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008314856381963e-05,
+      "loss": 0.1564,
+      "step": 1522
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.995844298794271e-05,
+      "loss": 0.2224,
+      "step": 1523
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.983384114603412e-05,
+      "loss": 0.2162,
+      "step": 1524
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9709343296387954e-05,
+      "loss": 0.1737,
+      "step": 1525
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9584949697082936e-05,
+      "loss": 0.2902,
+      "step": 1526
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9460660605981515e-05,
+      "loss": 0.2317,
+      "step": 1527
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933647628072954e-05,
+      "loss": 0.1597,
+      "step": 1528
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.921239697875578e-05,
+      "loss": 0.1488,
+      "step": 1529
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9088422957271164e-05,
+      "loss": 0.1171,
+      "step": 1530
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896455447326849e-05,
+      "loss": 0.1866,
+      "step": 1531
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88407917835217e-05,
+      "loss": 0.2093,
+      "step": 1532
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8717135144585425e-05,
+      "loss": 0.1584,
+      "step": 1533
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.859358481279458e-05,
+      "loss": 0.2144,
+      "step": 1534
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847014104426353e-05,
+      "loss": 0.1768,
+      "step": 1535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83468040948859e-05,
+      "loss": 0.1824,
+      "step": 1536
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82235742203338e-05,
+      "loss": 0.1205,
+      "step": 1537
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8100451676057345e-05,
+      "loss": 0.0895,
+      "step": 1538
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7977436717284275e-05,
+      "loss": 0.2723,
+      "step": 1539
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7854529599019213e-05,
+      "loss": 0.1985,
+      "step": 1540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.773173057604322e-05,
+      "loss": 0.0592,
+      "step": 1541
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.76090399029134e-05,
+      "loss": 0.128,
+      "step": 1542
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.748645783396209e-05,
+      "loss": 0.1907,
+      "step": 1543
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.736398462329664e-05,
+      "loss": 0.2264,
+      "step": 1544
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7241620524798744e-05,
+      "loss": 0.1575,
+      "step": 1545
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711936579212369e-05,
+      "loss": 0.2068,
+      "step": 1546
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.699722067870035e-05,
+      "loss": 0.2262,
+      "step": 1547
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687518543773013e-05,
+      "loss": 0.1572,
+      "step": 1548
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675326032218684e-05,
+      "loss": 0.203,
+      "step": 1549
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6631445584815926e-05,
+      "loss": 0.1292,
+      "step": 1550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.650974147813396e-05,
+      "loss": 0.162,
+      "step": 1551
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.638814825442832e-05,
+      "loss": 0.1844,
+      "step": 1552
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6266666165756476e-05,
+      "loss": 0.1906,
+      "step": 1553
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614529546394549e-05,
+      "loss": 0.1858,
+      "step": 1554
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6024036400591544e-05,
+      "loss": 0.2224,
+      "step": 1555
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5902889227059345e-05,
+      "loss": 0.2398,
+      "step": 1556
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.578185419448173e-05,
+      "loss": 0.1705,
+      "step": 1557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.566093155375916e-05,
+      "loss": 0.1614,
+      "step": 1558
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554012155555883e-05,
+      "loss": 0.1217,
+      "step": 1559
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54194244503147e-05,
+      "loss": 0.1815,
+      "step": 1560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.529884048822661e-05,
+      "loss": 0.2684,
+      "step": 1561
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5178369919259836e-05,
+      "loss": 0.219,
+      "step": 1562
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.505801299314466e-05,
+      "loss": 0.184,
+      "step": 1563
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.493776995937573e-05,
+      "loss": 0.2357,
+      "step": 1564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481764106721157e-05,
+      "loss": 0.0447,
+      "step": 1565
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4697626565674225e-05,
+      "loss": 0.1686,
+      "step": 1566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457772670354847e-05,
+      "loss": 0.1212,
+      "step": 1567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.445794172938149e-05,
+      "loss": 0.209,
+      "step": 1568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4338271891482386e-05,
+      "loss": 0.1299,
+      "step": 1569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4218717437921445e-05,
+      "loss": 0.1193,
+      "step": 1570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4099278616529905e-05,
+      "loss": 0.1653,
+      "step": 1571
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397995567489921e-05,
+      "loss": 0.1601,
+      "step": 1572
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3860748860380584e-05,
+      "loss": 0.1412,
+      "step": 1573
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374165842008464e-05,
+      "loss": 0.1582,
+      "step": 1574
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.36226846008806e-05,
+      "loss": 0.1059,
+      "step": 1575
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350382764939609e-05,
+      "loss": 0.2172,
+      "step": 1576
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3385087812016344e-05,
+      "loss": 0.098,
+      "step": 1577
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326646533488384e-05,
+      "loss": 0.1376,
+      "step": 1578
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3147960463897887e-05,
+      "loss": 0.2369,
+      "step": 1579
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.302957344471383e-05,
+      "loss": 0.0564,
+      "step": 1580
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29113045227429e-05,
+      "loss": 0.1114,
+      "step": 1581
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.279315394315137e-05,
+      "loss": 0.2155,
+      "step": 1582
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267512195086022e-05,
+      "loss": 0.1469,
+      "step": 1583
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255720879054469e-05,
+      "loss": 0.2586,
+      "step": 1584
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2439414706633574e-05,
+      "loss": 0.1411,
+      "step": 1585
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2321739943308926e-05,
+      "loss": 0.2663,
+      "step": 1586
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.220418474450537e-05,
+      "loss": 0.2812,
+      "step": 1587
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2086749353909704e-05,
+      "loss": 0.2008,
+      "step": 1588
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196943401496038e-05,
+      "loss": 0.1116,
+      "step": 1589
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185223897084709e-05,
+      "loss": 0.2693,
+      "step": 1590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173516446450989e-05,
+      "loss": 0.1844,
+      "step": 1591
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1618210738639254e-05,
+      "loss": 0.2502,
+      "step": 1592
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1501378035675085e-05,
+      "loss": 0.1648,
+      "step": 1593
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1384666597806524e-05,
+      "loss": 0.2325,
+      "step": 1594
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.126807666697137e-05,
+      "loss": 0.1698,
+      "step": 1595
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.115160848485532e-05,
+      "loss": 0.2528,
+      "step": 1596
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103526229289193e-05,
+      "loss": 0.1592,
+      "step": 1597
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0919038332261816e-05,
+      "loss": 0.1795,
+      "step": 1598
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0802936843892156e-05,
+      "loss": 0.1912,
+      "step": 1599
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068695806845624e-05,
+      "loss": 0.0952,
+      "step": 1600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0571102246373086e-05,
+      "loss": 0.0548,
+      "step": 1601
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.04553696178067e-05,
+      "loss": 0.1872,
+      "step": 1602
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033976042266585e-05,
+      "loss": 0.1603,
+      "step": 1603
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0224274900603345e-05,
+      "loss": 0.2009,
+      "step": 1604
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.010891329101557e-05,
+      "loss": 0.211,
+      "step": 1605
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999367583304222e-05,
+      "loss": 0.1152,
+      "step": 1606
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9878562765565444e-05,
+      "loss": 0.1698,
+      "step": 1607
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9763574327209683e-05,
+      "loss": 0.1565,
+      "step": 1608
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9648710756340946e-05,
+      "loss": 0.1223,
+      "step": 1609
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.953397229106636e-05,
+      "loss": 0.1409,
+      "step": 1610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.941935916923384e-05,
+      "loss": 0.1594,
+      "step": 1611
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930487162843132e-05,
+      "loss": 0.1366,
+      "step": 1612
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.919050990598657e-05,
+      "loss": 0.2584,
+      "step": 1613
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907627423896641e-05,
+      "loss": 0.2114,
+      "step": 1614
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8962164864176385e-05,
+      "loss": 0.1434,
+      "step": 1615
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8848182018160314e-05,
+      "loss": 0.2167,
+      "step": 1616
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873432593719965e-05,
+      "loss": 0.2628,
+      "step": 1617
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8620596857313044e-05,
+      "loss": 0.2052,
+      "step": 1618
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8506995014256e-05,
+      "loss": 0.0812,
+      "step": 1619
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839352064352012e-05,
+      "loss": 0.1117,
+      "step": 1620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8280173980332914e-05,
+      "loss": 0.1139,
+      "step": 1621
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816695525965702e-05,
+      "loss": 0.1336,
+      "step": 1622
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8053864716189904e-05,
+      "loss": 0.2297,
+      "step": 1623
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.794090258436338e-05,
+      "loss": 0.1739,
+      "step": 1624
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.782806909834299e-05,
+      "loss": 0.1639,
+      "step": 1625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771536449202768e-05,
+      "loss": 0.1151,
+      "step": 1626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.760278899904916e-05,
+      "loss": 0.2025,
+      "step": 1627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.749034285277152e-05,
+      "loss": 0.1424,
+      "step": 1628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7378026286290744e-05,
+      "loss": 0.1672,
+      "step": 1629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7265839532434155e-05,
+      "loss": 0.1375,
+      "step": 1630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7153782823760054e-05,
+      "loss": 0.2604,
+      "step": 1631
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.704185639255708e-05,
+      "loss": 0.2279,
+      "step": 1632
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.693006047084383e-05,
+      "loss": 0.139,
+      "step": 1633
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.681839529036841e-05,
+      "loss": 0.1923,
+      "step": 1634
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.670686108260796e-05,
+      "loss": 0.2188,
+      "step": 1635
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.659545807876786e-05,
+      "loss": 0.0851,
+      "step": 1636
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6484186509781826e-05,
+      "loss": 0.1516,
+      "step": 1637
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.637304660631088e-05,
+      "loss": 0.1967,
+      "step": 1638
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6262038598743254e-05,
+      "loss": 0.1945,
+      "step": 1639
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.615116271719378e-05,
+      "loss": 0.2479,
+      "step": 1640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.604041919150317e-05,
+      "loss": 0.2543,
+      "step": 1641
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.592980825123803e-05,
+      "loss": 0.1208,
+      "step": 1642
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.581933012569004e-05,
+      "loss": 0.1305,
+      "step": 1643
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5708985043875506e-05,
+      "loss": 0.315,
+      "step": 1644
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5598773234534954e-05,
+      "loss": 0.0882,
+      "step": 1645
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.548869492613273e-05,
+      "loss": 0.218,
+      "step": 1646
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5378750346856296e-05,
+      "loss": 0.1408,
+      "step": 1647
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5268939724616056e-05,
+      "loss": 0.2152,
+      "step": 1648
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.51592632870446e-05,
+      "loss": 0.1806,
+      "step": 1649
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504972126149639e-05,
+      "loss": 0.1673,
+      "step": 1650
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4940313875047326e-05,
+      "loss": 0.1574,
+      "step": 1651
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48310413544941e-05,
+      "loss": 0.1591,
+      "step": 1652
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.472190392635395e-05,
+      "loss": 0.0993,
+      "step": 1653
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4612901816863976e-05,
+      "loss": 0.1947,
+      "step": 1654
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.450403525198076e-05,
+      "loss": 0.2988,
+      "step": 1655
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.439530445738004e-05,
+      "loss": 0.1762,
+      "step": 1656
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428670965845594e-05,
+      "loss": 0.1564,
+      "step": 1657
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.417825108032081e-05,
+      "loss": 0.1369,
+      "step": 1658
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.406992894780453e-05,
+      "loss": 0.2366,
+      "step": 1659
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.396174348545413e-05,
+      "loss": 0.1994,
+      "step": 1660
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.385369491753344e-05,
+      "loss": 0.105,
+      "step": 1661
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.374578346802239e-05,
+      "loss": 0.1239,
+      "step": 1662
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3638009360616664e-05,
+      "loss": 0.2729,
+      "step": 1663
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.353037281872739e-05,
+      "loss": 0.1813,
+      "step": 1664
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3422874065480346e-05,
+      "loss": 0.2376,
+      "step": 1665
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.331551332371585e-05,
+      "loss": 0.1415,
+      "step": 1666
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3208290815987995e-05,
+      "loss": 0.1443,
+      "step": 1667
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.310120676456434e-05,
+      "loss": 0.1476,
+      "step": 1668
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.299426139142553e-05,
+      "loss": 0.0924,
+      "step": 1669
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.288745491826459e-05,
+      "loss": 0.2135,
+      "step": 1670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.278078756648677e-05,
+      "loss": 0.1864,
+      "step": 1671
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.267425955720881e-05,
+      "loss": 0.1661,
+      "step": 1672
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2567871111258575e-05,
+      "loss": 0.1443,
+      "step": 1673
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2461622449174777e-05,
+      "loss": 0.2244,
+      "step": 1674
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.235551379120618e-05,
+      "loss": 0.2088,
+      "step": 1675
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2249545357311484e-05,
+      "loss": 0.2101,
+      "step": 1676
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2143717367158596e-05,
+      "loss": 0.1162,
+      "step": 1677
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2038030040124326e-05,
+      "loss": 0.084,
+      "step": 1678
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1932483595293895e-05,
+      "loss": 0.1421,
+      "step": 1679
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.182707825146056e-05,
+      "loss": 0.1702,
+      "step": 1680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.17218142271249e-05,
+      "loss": 0.1205,
+      "step": 1681
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.161669174049474e-05,
+      "loss": 0.1662,
+      "step": 1682
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1511711009484345e-05,
+      "loss": 0.113,
+      "step": 1683
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1406872251714234e-05,
+      "loss": 0.1627,
+      "step": 1684
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.130217568451065e-05,
+      "loss": 0.2975,
+      "step": 1685
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1197621524904886e-05,
+      "loss": 0.2312,
+      "step": 1686
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1093209989633235e-05,
+      "loss": 0.0597,
+      "step": 1687
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.098894129513629e-05,
+      "loss": 0.1278,
+      "step": 1688
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0884815657558474e-05,
+      "loss": 0.2307,
+      "step": 1689
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078083329274767e-05,
+      "loss": 0.1417,
+      "step": 1690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0676994416254855e-05,
+      "loss": 0.2378,
+      "step": 1691
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0573299243333455e-05,
+      "loss": 0.2388,
+      "step": 1692
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.046974798893908e-05,
+      "loss": 0.1793,
+      "step": 1693
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0366340867728947e-05,
+      "loss": 0.1946,
+      "step": 1694
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0263078094061492e-05,
+      "loss": 0.1877,
+      "step": 1695
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0159959881996e-05,
+      "loss": 0.1541,
+      "step": 1696
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005698644529199e-05,
+      "loss": 0.1258,
+      "step": 1697
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.995415799740896e-05,
+      "loss": 0.1686,
+      "step": 1698
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.985147475150577e-05,
+      "loss": 0.0873,
+      "step": 1699
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9748936920440286e-05,
+      "loss": 0.0979,
+      "step": 1700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9646544716769033e-05,
+      "loss": 0.2048,
+      "step": 1701
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9544298352746524e-05,
+      "loss": 0.0607,
+      "step": 1702
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9442198040325086e-05,
+      "loss": 0.2336,
+      "step": 1703
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9340243991154192e-05,
+      "loss": 0.1492,
+      "step": 1704
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9238436416580107e-05,
+      "loss": 0.2008,
+      "step": 1705
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9136775527645554e-05,
+      "loss": 0.189,
+      "step": 1706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.903526153508912e-05,
+      "loss": 0.1416,
+      "step": 1707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.893389464934484e-05,
+      "loss": 0.1223,
+      "step": 1708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8832675080541914e-05,
+      "loss": 0.1855,
+      "step": 1709
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.873160303850405e-05,
+      "loss": 0.1812,
+      "step": 1710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.863067873274925e-05,
+      "loss": 0.1614,
+      "step": 1711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.852990237248917e-05,
+      "loss": 0.1872,
+      "step": 1712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8429274166628783e-05,
+      "loss": 0.1898,
+      "step": 1713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8328794323766038e-05,
+      "loss": 0.2472,
+      "step": 1714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.822846305219121e-05,
+      "loss": 0.1822,
+      "step": 1715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8128280559886733e-05,
+      "loss": 0.1428,
+      "step": 1716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.80282470545265e-05,
+      "loss": 0.0876,
+      "step": 1717
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7928362743475588e-05,
+      "loss": 0.1814,
+      "step": 1718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.782862783378989e-05,
+      "loss": 0.201,
+      "step": 1719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7729042532215456e-05,
+      "loss": 0.1873,
+      "step": 1720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7629607045188343e-05,
+      "loss": 0.1411,
+      "step": 1721
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.753032157883397e-05,
+      "loss": 0.2571,
+      "step": 1722
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.743118633896674e-05,
+      "loss": 0.2074,
+      "step": 1723
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.733220153108973e-05,
+      "loss": 0.0418,
+      "step": 1724
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.723336736039418e-05,
+      "loss": 0.1441,
+      "step": 1725
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7134684031758907e-05,
+      "loss": 0.1844,
+      "step": 1726
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7036151749750227e-05,
+      "loss": 0.2149,
+      "step": 1727
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.693777071862128e-05,
+      "loss": 0.1859,
+      "step": 1728
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6839541142311596e-05,
+      "loss": 0.1287,
+      "step": 1729
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6741463224446926e-05,
+      "loss": 0.1693,
+      "step": 1730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6643537168338374e-05,
+      "loss": 0.19,
+      "step": 1731
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6545763176982474e-05,
+      "loss": 0.2565,
+      "step": 1732
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6448141453060458e-05,
+      "loss": 0.2097,
+      "step": 1733
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6350672198937887e-05,
+      "loss": 0.1507,
+      "step": 1734
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6253355616664253e-05,
+      "loss": 0.0695,
+      "step": 1735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.615619190797266e-05,
+      "loss": 0.0732,
+      "step": 1736
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6059181274279176e-05,
+      "loss": 0.2003,
+      "step": 1737
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5962323916682674e-05,
+      "loss": 0.2118,
+      "step": 1738
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5865620035964212e-05,
+      "loss": 0.1346,
+      "step": 1739
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.576906983258669e-05,
+      "loss": 0.2382,
+      "step": 1740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5672673506694555e-05,
+      "loss": 0.1172,
+      "step": 1741
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5576431258113097e-05,
+      "loss": 0.1653,
+      "step": 1742
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5480343286348407e-05,
+      "loss": 0.0919,
+      "step": 1743
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5384409790586595e-05,
+      "loss": 0.2602,
+      "step": 1744
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5288630969693617e-05,
+      "loss": 0.2001,
+      "step": 1745
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5193007022214855e-05,
+      "loss": 0.1939,
+      "step": 1746
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5097538146374512e-05,
+      "loss": 0.221,
+      "step": 1747
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.500222454007548e-05,
+      "loss": 0.1127,
+      "step": 1748
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4907066400898683e-05,
+      "loss": 0.182,
+      "step": 1749
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.481206392610278e-05,
+      "loss": 0.1913,
+      "step": 1750
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.47172173126238e-05,
+      "loss": 0.2412,
+      "step": 1751
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4622526757074628e-05,
+      "loss": 0.1501,
+      "step": 1752
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.452799245574463e-05,
+      "loss": 0.2613,
+      "step": 1753
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4433614604599366e-05,
+      "loss": 0.1139,
+      "step": 1754
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.433939339927992e-05,
+      "loss": 0.159,
+      "step": 1755
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4245329035102837e-05,
+      "loss": 0.2473,
+      "step": 1756
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415142170705941e-05,
+      "loss": 0.0628,
+      "step": 1757
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4057671609815403e-05,
+      "loss": 0.1261,
+      "step": 1758
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3964078937710744e-05,
+      "loss": 0.1136,
+      "step": 1759
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3870643884758913e-05,
+      "loss": 0.163,
+      "step": 1760
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3777366644646758e-05,
+      "loss": 0.2006,
+      "step": 1761
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3684247410733918e-05,
+      "loss": 0.2255,
+      "step": 1762
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3591286376052468e-05,
+      "loss": 0.1282,
+      "step": 1763
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3498483733306654e-05,
+      "loss": 0.2877,
+      "step": 1764
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3405839674872242e-05,
+      "loss": 0.1005,
+      "step": 1765
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3313354392796395e-05,
+      "loss": 0.1354,
+      "step": 1766
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3221028078797046e-05,
+      "loss": 0.139,
+      "step": 1767
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3128860924262598e-05,
+      "loss": 0.1864,
+      "step": 1768
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3036853120251568e-05,
+      "loss": 0.2124,
+      "step": 1769
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294500485749218e-05,
+      "loss": 0.1658,
+      "step": 1770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2853316326381792e-05,
+      "loss": 0.1196,
+      "step": 1771
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.276178771698676e-05,
+      "loss": 0.1507,
+      "step": 1772
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2670419219041962e-05,
+      "loss": 0.1981,
+      "step": 1773
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.257921102195024e-05,
+      "loss": 0.1559,
+      "step": 1774
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2488163314782317e-05,
+      "loss": 0.1011,
+      "step": 1775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2397276286275992e-05,
+      "loss": 0.1573,
+      "step": 1776
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.230655012483619e-05,
+      "loss": 0.2254,
+      "step": 1777
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22159850185343e-05,
+      "loss": 0.1251,
+      "step": 1778
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.212558115510783e-05,
+      "loss": 0.1653,
+      "step": 1779
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.203533872196003e-05,
+      "loss": 0.1458,
+      "step": 1780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1945257906159578e-05,
+      "loss": 0.1383,
+      "step": 1781
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1855338894440024e-05,
+      "loss": 0.1346,
+      "step": 1782
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1765581873199626e-05,
+      "loss": 0.0505,
+      "step": 1783
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1675987028500743e-05,
+      "loss": 0.2208,
+      "step": 1784
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.158655454606957e-05,
+      "loss": 0.1791,
+      "step": 1785
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.14972846112958e-05,
+      "loss": 0.1589,
+      "step": 1786
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1408177409232066e-05,
+      "loss": 0.218,
+      "step": 1787
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1319233124593764e-05,
+      "loss": 0.1795,
+      "step": 1788
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.12304519417585e-05,
+      "loss": 0.1443,
+      "step": 1789
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1141834044765774e-05,
+      "loss": 0.1797,
+      "step": 1790
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1053379617316684e-05,
+      "loss": 0.0949,
+      "step": 1791
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0965088842773338e-05,
+      "loss": 0.2004,
+      "step": 1792
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0876961904158743e-05,
+      "loss": 0.1335,
+      "step": 1793
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.078899898415616e-05,
+      "loss": 0.1295,
+      "step": 1794
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0701200265108865e-05,
+      "loss": 0.1275,
+      "step": 1795
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.061356592901984e-05,
+      "loss": 0.1496,
+      "step": 1796
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.052609615755121e-05,
+      "loss": 0.1381,
+      "step": 1797
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0438791132023983e-05,
+      "loss": 0.2132,
+      "step": 1798
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.035165103341773e-05,
+      "loss": 0.1589,
+      "step": 1799
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0264676042370025e-05,
+      "loss": 0.1602,
+      "step": 1800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0177866339176287e-05,
+      "loss": 0.1821,
+      "step": 1801
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009122210378922e-05,
+      "loss": 0.1271,
+      "step": 1802
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0004743515818538e-05,
+      "loss": 0.1783,
+      "step": 1803
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9918430754530625e-05,
+      "loss": 0.1455,
+      "step": 1804
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9832283998848035e-05,
+      "loss": 0.2656,
+      "step": 1805
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9746303427349277e-05,
+      "loss": 0.0839,
+      "step": 1806
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9660489218268318e-05,
+      "loss": 0.1602,
+      "step": 1807
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957484154949426e-05,
+      "loss": 0.1787,
+      "step": 1808
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9489360598571015e-05,
+      "loss": 0.1282,
+      "step": 1809
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.940404654269684e-05,
+      "loss": 0.1209,
+      "step": 1810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93188995587241e-05,
+      "loss": 0.2305,
+      "step": 1811
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9233919823158776e-05,
+      "loss": 0.0888,
+      "step": 1812
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9149107512160103e-05,
+      "loss": 0.1048,
+      "step": 1813
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9064462801540362e-05,
+      "loss": 0.2054,
+      "step": 1814
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8979985866764383e-05,
+      "loss": 0.19,
+      "step": 1815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8895676882949076e-05,
+      "loss": 0.2425,
+      "step": 1816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.881153602486335e-05,
+      "loss": 0.1186,
+      "step": 1817
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8727563466927568e-05,
+      "loss": 0.2041,
+      "step": 1818
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.864375938321311e-05,
+      "loss": 0.1984,
+      "step": 1819
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8560123947442298e-05,
+      "loss": 0.2162,
+      "step": 1820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8476657332987633e-05,
+      "loss": 0.1362,
+      "step": 1821
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8393359712871816e-05,
+      "loss": 0.1964,
+      "step": 1822
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.831023125976722e-05,
+      "loss": 0.1386,
+      "step": 1823
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8227272145995487e-05,
+      "loss": 0.1393,
+      "step": 1824
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8144482543527208e-05,
+      "loss": 0.0633,
+      "step": 1825
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8061862623981684e-05,
+      "loss": 0.1506,
+      "step": 1826
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7979412558626352e-05,
+      "loss": 0.2416,
+      "step": 1827
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.789713251837667e-05,
+      "loss": 0.1409,
+      "step": 1828
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7815022673795534e-05,
+      "loss": 0.1661,
+      "step": 1829
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7733083195093058e-05,
+      "loss": 0.2365,
+      "step": 1830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7651314252126282e-05,
+      "loss": 0.0904,
+      "step": 1831
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7569716014398584e-05,
+      "loss": 0.2067,
+      "step": 1832
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7488288651059627e-05,
+      "loss": 0.1365,
+      "step": 1833
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.740703233090475e-05,
+      "loss": 0.253,
+      "step": 1834
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.732594722237475e-05,
+      "loss": 0.2576,
+      "step": 1835
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7245033493555573e-05,
+      "loss": 0.0996,
+      "step": 1836
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7164291312177818e-05,
+      "loss": 0.1921,
+      "step": 1837
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7083720845616534e-05,
+      "loss": 0.2055,
+      "step": 1838
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7003322260890776e-05,
+      "loss": 0.1972,
+      "step": 1839
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6923095724663297e-05,
+      "loss": 0.2182,
+      "step": 1840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6843041403240246e-05,
+      "loss": 0.1489,
+      "step": 1841
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.67631594625707e-05,
+      "loss": 0.2591,
+      "step": 1842
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6683450068246485e-05,
+      "loss": 0.0749,
+      "step": 1843
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.66039133855017e-05,
+      "loss": 0.0718,
+      "step": 1844
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.652454957921238e-05,
+      "loss": 0.1761,
+      "step": 1845
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6445358813896295e-05,
+      "loss": 0.1137,
+      "step": 1846
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6366341253712438e-05,
+      "loss": 0.2232,
+      "step": 1847
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6287497062460722e-05,
+      "loss": 0.2448,
+      "step": 1848
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6208826403581788e-05,
+      "loss": 0.2006,
+      "step": 1849
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6130329440156432e-05,
+      "loss": 0.1753,
+      "step": 1850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6052006334905468e-05,
+      "loss": 0.1844,
+      "step": 1851
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5973857250189326e-05,
+      "loss": 0.158,
+      "step": 1852
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.589588234800756e-05,
+      "loss": 0.1319,
+      "step": 1853
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5818081789998805e-05,
+      "loss": 0.1131,
+      "step": 1854
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.574045573744023e-05,
+      "loss": 0.1311,
+      "step": 1855
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5663004351247257e-05,
+      "loss": 0.1514,
+      "step": 1856
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5585727791973237e-05,
+      "loss": 0.0816,
+      "step": 1857
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5508626219809065e-05,
+      "loss": 0.2285,
+      "step": 1858
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5431699794582997e-05,
+      "loss": 0.2171,
+      "step": 1859
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.535494867576016e-05,
+      "loss": 0.2423,
+      "step": 1860
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5278373022442272e-05,
+      "loss": 0.101,
+      "step": 1861
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5201972993367309e-05,
+      "loss": 0.243,
+      "step": 1862
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.512574874690923e-05,
+      "loss": 0.2244,
+      "step": 1863
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5049700441077541e-05,
+      "loss": 0.09,
+      "step": 1864
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4973828233517152e-05,
+      "loss": 0.1776,
+      "step": 1865
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4898132281507727e-05,
+      "loss": 0.2435,
+      "step": 1866
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4822612741963737e-05,
+      "loss": 0.1071,
+      "step": 1867
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4747269771433936e-05,
+      "loss": 0.1704,
+      "step": 1868
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4672103526100945e-05,
+      "loss": 0.2241,
+      "step": 1869
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4597114161781188e-05,
+      "loss": 0.1322,
+      "step": 1870
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4522301833924312e-05,
+      "loss": 0.1724,
+      "step": 1871
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4447666697613004e-05,
+      "loss": 0.2187,
+      "step": 1872
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4373208907562686e-05,
+      "loss": 0.2054,
+      "step": 1873
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4298928618121098e-05,
+      "loss": 0.0707,
+      "step": 1874
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.422482598326802e-05,
+      "loss": 0.1538,
+      "step": 1875
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4150901156615026e-05,
+      "loss": 0.1654,
+      "step": 1876
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4077154291405027e-05,
+      "loss": 0.0816,
+      "step": 1877
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.400358554051211e-05,
+      "loss": 0.1066,
+      "step": 1878
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3930195056441053e-05,
+      "loss": 0.1396,
+      "step": 1879
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3856982991327128e-05,
+      "loss": 0.1524,
+      "step": 1880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3783949496935788e-05,
+      "loss": 0.2058,
+      "step": 1881
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3711094724662255e-05,
+      "loss": 0.2274,
+      "step": 1882
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.363841882553134e-05,
+      "loss": 0.2599,
+      "step": 1883
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.356592195019698e-05,
+      "loss": 0.2293,
+      "step": 1884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3493604248942038e-05,
+      "loss": 0.1307,
+      "step": 1885
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3421465871677995e-05,
+      "loss": 0.1733,
+      "step": 1886
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3349506967944513e-05,
+      "loss": 0.1953,
+      "step": 1887
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3277727686909324e-05,
+      "loss": 0.2122,
+      "step": 1888
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3206128177367727e-05,
+      "loss": 0.2113,
+      "step": 1889
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3134708587742362e-05,
+      "loss": 0.114,
+      "step": 1890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3063469066082956e-05,
+      "loss": 0.0906,
+      "step": 1891
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2992409760065937e-05,
+      "loss": 0.1669,
+      "step": 1892
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.292153081699412e-05,
+      "loss": 0.1893,
+      "step": 1893
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2850832383796519e-05,
+      "loss": 0.2138,
+      "step": 1894
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2780314607027844e-05,
+      "loss": 0.228,
+      "step": 1895
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2709977632868409e-05,
+      "loss": 0.1707,
+      "step": 1896
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2639821607123748e-05,
+      "loss": 0.1182,
+      "step": 1897
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2569846675224139e-05,
+      "loss": 0.1208,
+      "step": 1898
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2500052982224641e-05,
+      "loss": 0.1804,
+      "step": 1899
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2430440672804545e-05,
+      "loss": 0.1045,
+      "step": 1900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2361009891267105e-05,
+      "loss": 0.1706,
+      "step": 1901
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.229176078153933e-05,
+      "loss": 0.1813,
+      "step": 1902
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2222693487171565e-05,
+      "loss": 0.1992,
+      "step": 1903
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2153808151337343e-05,
+      "loss": 0.143,
+      "step": 1904
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2085104916832978e-05,
+      "loss": 0.1989,
+      "step": 1905
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2016583926077264e-05,
+      "loss": 0.1973,
+      "step": 1906
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1948245321111207e-05,
+      "loss": 0.2126,
+      "step": 1907
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1880089243597792e-05,
+      "loss": 0.1621,
+      "step": 1908
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.181211583482158e-05,
+      "loss": 0.1857,
+      "step": 1909
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1744325235688536e-05,
+      "loss": 0.192,
+      "step": 1910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1676717586725527e-05,
+      "loss": 0.1756,
+      "step": 1911
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1609293028080326e-05,
+      "loss": 0.1627,
+      "step": 1912
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1542051699521118e-05,
+      "loss": 0.1832,
+      "step": 1913
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1474993740436202e-05,
+      "loss": 0.1893,
+      "step": 1914
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1408119289833851e-05,
+      "loss": 0.1302,
+      "step": 1915
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1341428486341876e-05,
+      "loss": 0.1583,
+      "step": 1916
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1274921468207389e-05,
+      "loss": 0.1764,
+      "step": 1917
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1208598373296564e-05,
+      "loss": 0.095,
+      "step": 1918
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.114245933909428e-05,
+      "loss": 0.127,
+      "step": 1919
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1076504502703867e-05,
+      "loss": 0.2093,
+      "step": 1920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1010734000846846e-05,
+      "loss": 0.2249,
+      "step": 1921
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0945147969862569e-05,
+      "loss": 0.2376,
+      "step": 1922
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0879746545708069e-05,
+      "loss": 0.1395,
+      "step": 1923
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0814529863957623e-05,
+      "loss": 0.2722,
+      "step": 1924
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0749498059802554e-05,
+      "loss": 0.1084,
+      "step": 1925
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0684651268051005e-05,
+      "loss": 0.2496,
+      "step": 1926
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0619989623127512e-05,
+      "loss": 0.2025,
+      "step": 1927
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0555513259072868e-05,
+      "loss": 0.1628,
+      "step": 1928
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.049122230954378e-05,
+      "loss": 0.1598,
+      "step": 1929
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042711690781254e-05,
+      "loss": 0.1731,
+      "step": 1930
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0363197186766904e-05,
+      "loss": 0.1869,
+      "step": 1931
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0299463278909637e-05,
+      "loss": 0.2014,
+      "step": 1932
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.023591531635838e-05,
+      "loss": 0.2335,
+      "step": 1933
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0172553430845289e-05,
+      "loss": 0.2102,
+      "step": 1934
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0109377753716753e-05,
+      "loss": 0.2466,
+      "step": 1935
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0046388415933228e-05,
+      "loss": 0.1863,
+      "step": 1936
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983585548068907e-06,
+      "loss": 0.1664,
+      "step": 1937
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.920969280311331e-06,
+      "loss": 0.1887,
+      "step": 1938
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.858539742461337e-06,
+      "loss": 0.192,
+      "step": 1939
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.796297063932614e-06,
+      "loss": 0.1731,
+      "step": 1940
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.734241373751541e-06,
+      "loss": 0.2196,
+      "step": 1941
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.672372800556884e-06,
+      "loss": 0.1072,
+      "step": 1942
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.610691472599442e-06,
+      "loss": 0.0923,
+      "step": 1943
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.549197517741959e-06,
+      "loss": 0.1198,
+      "step": 1944
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.487891063458743e-06,
+      "loss": 0.0631,
+      "step": 1945
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.426772236835368e-06,
+      "loss": 0.0938,
+      "step": 1946
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.365841164568523e-06,
+      "loss": 0.1634,
+      "step": 1947
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.305097972965627e-06,
+      "loss": 0.1544,
+      "step": 1948
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.24454278794471e-06,
+      "loss": 0.1718,
+      "step": 1949
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.18417573503404e-06,
+      "loss": 0.1294,
+      "step": 1950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.12399693937187e-06,
+      "loss": 0.1612,
+      "step": 1951
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.064006525706214e-06,
+      "loss": 0.1779,
+      "step": 1952
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.004204618394608e-06,
+      "loss": 0.2281,
+      "step": 1953
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.944591341403774e-06,
+      "loss": 0.2067,
+      "step": 1954
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.885166818309499e-06,
+      "loss": 0.1126,
+      "step": 1955
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.825931172296142e-06,
+      "loss": 0.0746,
+      "step": 1956
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.76688452615667e-06,
+      "loss": 0.171,
+      "step": 1957
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.708027002292218e-06,
+      "loss": 0.1874,
+      "step": 1958
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.649358722711832e-06,
+      "loss": 0.1327,
+      "step": 1959
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.590879809032349e-06,
+      "loss": 0.0973,
+      "step": 1960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.53259038247798e-06,
+      "loss": 0.0494,
+      "step": 1961
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.474490563880156e-06,
+      "loss": 0.1366,
+      "step": 1962
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.416580473677304e-06,
+      "loss": 0.1786,
+      "step": 1963
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.3588602319145e-06,
+      "loss": 0.2326,
+      "step": 1964
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.301329958243276e-06,
+      "loss": 0.1871,
+      "step": 1965
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.243989771921412e-06,
+      "loss": 0.1743,
+      "step": 1966
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.186839791812595e-06,
+      "loss": 0.1257,
+      "step": 1967
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.129880136386259e-06,
+      "loss": 0.0886,
+      "step": 1968
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.073110923717298e-06,
+      "loss": 0.0945,
+      "step": 1969
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.016532271485789e-06,
+      "loss": 0.1153,
+      "step": 1970
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.960144296976835e-06,
+      "loss": 0.0851,
+      "step": 1971
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.903947117080236e-06,
+      "loss": 0.1931,
+      "step": 1972
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.847940848290325e-06,
+      "loss": 0.1901,
+      "step": 1973
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.792125606705635e-06,
+      "loss": 0.1522,
+      "step": 1974
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.736501508028726e-06,
+      "loss": 0.1807,
+      "step": 1975
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.68106866756596e-06,
+      "loss": 0.0504,
+      "step": 1976
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.625827200227164e-06,
+      "loss": 0.2197,
+      "step": 1977
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.570777220525538e-06,
+      "loss": 0.2584,
+      "step": 1978
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.515918842577263e-06,
+      "loss": 0.1839,
+      "step": 1979
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.461252180101352e-06,
+      "loss": 0.1973,
+      "step": 1980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.406777346419436e-06,
+      "loss": 0.1304,
+      "step": 1981
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.3524944544555164e-06,
+      "loss": 0.2238,
+      "step": 1982
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.298403616735583e-06,
+      "loss": 0.1172,
+      "step": 1983
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.244504945387631e-06,
+      "loss": 0.1165,
+      "step": 1984
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.190798552141253e-06,
+      "loss": 0.186,
+      "step": 1985
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.13728454832745e-06,
+      "loss": 0.1662,
+      "step": 1986
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.083963044878472e-06,
+      "loss": 0.2221,
+      "step": 1987
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.0308341523273925e-06,
+      "loss": 0.1421,
+      "step": 1988
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.977897980808146e-06,
+      "loss": 0.174,
+      "step": 1989
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.92515464005511e-06,
+      "loss": 0.1804,
+      "step": 1990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.872604239402924e-06,
+      "loss": 0.2031,
+      "step": 1991
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.820246887786274e-06,
+      "loss": 0.1743,
+      "step": 1992
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.768082693739663e-06,
+      "loss": 0.2484,
+      "step": 1993
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.7161117653971926e-06,
+      "loss": 0.1365,
+      "step": 1994
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.6643342104923736e-06,
+      "loss": 0.1966,
+      "step": 1995
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.612750136357782e-06,
+      "loss": 0.1438,
+      "step": 1996
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.561359649924958e-06,
+      "loss": 0.2006,
+      "step": 1997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.510162857724167e-06,
+      "loss": 0.1715,
+      "step": 1998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.459159865884101e-06,
+      "loss": 0.1857,
+      "step": 1999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.408350780131778e-06,
+      "loss": 0.086,
+      "step": 2000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.35773570579219e-06,
+      "loss": 0.1806,
+      "step": 2001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.3073147477881624e-06,
+      "loss": 0.1882,
+      "step": 2002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.2570880106401995e-06,
+      "loss": 0.1298,
+      "step": 2003
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.207055598466094e-06,
+      "loss": 0.1974,
+      "step": 2004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.157217614980881e-06,
+      "loss": 0.1158,
+      "step": 2005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.1075741634965326e-06,
+      "loss": 0.2265,
+      "step": 2006
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.0581253469217395e-06,
+      "loss": 0.2109,
+      "step": 2007
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.008871267761762e-06,
+      "loss": 0.1994,
+      "step": 2008
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.959812028118172e-06,
+      "loss": 0.1711,
+      "step": 2009
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.910947729688598e-06,
+      "loss": 0.1619,
+      "step": 2010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.862278473766636e-06,
+      "loss": 0.0628,
+      "step": 2011
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.813804361241515e-06,
+      "loss": 0.2139,
+      "step": 2012
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.76552549259799e-06,
+      "loss": 0.1859,
+      "step": 2013
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.717441967916026e-06,
+      "loss": 0.0586,
+      "step": 2014
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.66955388687066e-06,
+      "loss": 0.1411,
+      "step": 2015
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6218613487318385e-06,
+      "loss": 0.2069,
+      "step": 2016
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.574364452364078e-06,
+      "loss": 0.1582,
+      "step": 2017
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.52706329622642e-06,
+      "loss": 0.2217,
+      "step": 2018
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.479957978372063e-06,
+      "loss": 0.1043,
+      "step": 2019
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.433048596448287e-06,
+      "loss": 0.1895,
+      "step": 2020
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.386335247696217e-06,
+      "loss": 0.0719,
+      "step": 2021
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.3398180289505606e-06,
+      "loss": 0.242,
+      "step": 2022
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.293497036639517e-06,
+      "loss": 0.1627,
+      "step": 2023
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.247372366784475e-06,
+      "loss": 0.201,
+      "step": 2024
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.201444114999842e-06,
+      "loss": 0.1653,
+      "step": 2025
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.155712376492905e-06,
+      "loss": 0.1695,
+      "step": 2026
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.1101772460635966e-06,
+      "loss": 0.194,
+      "step": 2027
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0648388181042005e-06,
+      "loss": 0.2193,
+      "step": 2028
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.019697186599337e-06,
+      "loss": 0.1011,
+      "step": 2029
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.97475244512563e-06,
+      "loss": 0.2071,
+      "step": 2030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.930004686851574e-06,
+      "loss": 0.1582,
+      "step": 2031
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.88545400453736e-06,
+      "loss": 0.167,
+      "step": 2032
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.841100490534545e-06,
+      "loss": 0.0922,
+      "step": 2033
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796944236786061e-06,
+      "loss": 0.2243,
+      "step": 2034
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.752985334825921e-06,
+      "loss": 0.0929,
+      "step": 2035
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.709223875779001e-06,
+      "loss": 0.1261,
+      "step": 2036
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.66565995036089e-06,
+      "loss": 0.099,
+      "step": 2037
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.622293648877696e-06,
+      "loss": 0.2375,
+      "step": 2038
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.579125061225897e-06,
+      "loss": 0.2126,
+      "step": 2039
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5361542768921015e-06,
+      "loss": 0.2117,
+      "step": 2040
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.493381384952855e-06,
+      "loss": 0.139,
+      "step": 2041
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4508064740744985e-06,
+      "loss": 0.2711,
+      "step": 2042
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.408429632512989e-06,
+      "loss": 0.118,
+      "step": 2043
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3662509481136485e-06,
+      "loss": 0.2073,
+      "step": 2044
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.324270508311068e-06,
+      "loss": 0.1873,
+      "step": 2045
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2824884001288705e-06,
+      "loss": 0.1212,
+      "step": 2046
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.240904710179539e-06,
+      "loss": 0.0597,
+      "step": 2047
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.199519524664253e-06,
+      "loss": 0.2244,
+      "step": 2048
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1583329293726996e-06,
+      "loss": 0.1768,
+      "step": 2049
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.117345009682916e-06,
+      "loss": 0.1463,
+      "step": 2050
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.076555850561048e-06,
+      "loss": 0.0881,
+      "step": 2051
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.035965536561248e-06,
+      "loss": 0.1154,
+      "step": 2052
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.995574151825498e-06,
+      "loss": 0.0852,
+      "step": 2053
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9553817800833555e-06,
+      "loss": 0.1523,
+      "step": 2054
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.91538850465184e-06,
+      "loss": 0.0451,
+      "step": 2055
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.875594408435312e-06,
+      "loss": 0.2029,
+      "step": 2056
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.835999573925153e-06,
+      "loss": 0.1682,
+      "step": 2057
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.796604083199784e-06,
+      "loss": 0.1837,
+      "step": 2058
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7574080179243022e-06,
+      "loss": 0.2174,
+      "step": 2059
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.718411459350446e-06,
+      "loss": 0.095,
+      "step": 2060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6796144883163964e-06,
+      "loss": 0.1086,
+      "step": 2061
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.641017185246576e-06,
+      "loss": 0.1637,
+      "step": 2062
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.602619630151527e-06,
+      "loss": 0.1904,
+      "step": 2063
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5644219026276905e-06,
+      "loss": 0.1992,
+      "step": 2064
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5264240818573047e-06,
+      "loss": 0.2215,
+      "step": 2065
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.488626246608184e-06,
+      "loss": 0.2001,
+      "step": 2066
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4510284752336084e-06,
+      "loss": 0.2316,
+      "step": 2067
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.413630845672122e-06,
+      "loss": 0.1056,
+      "step": 2068
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3764334354473793e-06,
+      "loss": 0.1036,
+      "step": 2069
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3394363216679656e-06,
+      "loss": 0.1307,
+      "step": 2070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3026395810273116e-06,
+      "loss": 0.1814,
+      "step": 2071
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2660432898034886e-06,
+      "loss": 0.1007,
+      "step": 2072
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2296475238589474e-06,
+      "loss": 0.1687,
+      "step": 2073
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1934523586405807e-06,
+      "loss": 0.114,
+      "step": 2074
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1574578691793476e-06,
+      "loss": 0.1167,
+      "step": 2075
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.121664130090285e-06,
+      "loss": 0.1974,
+      "step": 2076
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0860712155722617e-06,
+      "loss": 0.1641,
+      "step": 2077
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0506791994078132e-06,
+      "loss": 0.176,
+      "step": 2078
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.015488154963053e-06,
+      "loss": 0.1685,
+      "step": 2079
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9804981551875054e-06,
+      "loss": 0.1343,
+      "step": 2080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9457092726139055e-06,
+      "loss": 0.1998,
+      "step": 2081
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.911121579358089e-06,
+      "loss": 0.2121,
+      "step": 2082
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.876735147118814e-06,
+      "loss": 0.2184,
+      "step": 2083
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8425500471776832e-06,
+      "loss": 0.0813,
+      "step": 2084
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8085663503989333e-06,
+      "loss": 0.1482,
+      "step": 2085
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7747841272292683e-06,
+      "loss": 0.1727,
+      "step": 2086
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.741203447697749e-06,
+      "loss": 0.1709,
+      "step": 2087
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.707824381415702e-06,
+      "loss": 0.1735,
+      "step": 2088
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6746469975764355e-06,
+      "loss": 0.1088,
+      "step": 2089
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.641671364955256e-06,
+      "loss": 0.1853,
+      "step": 2090
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6088975519092063e-06,
+      "loss": 0.1165,
+      "step": 2091
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.576325626376952e-06,
+      "loss": 0.143,
+      "step": 2092
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5439556558786936e-06,
+      "loss": 0.204,
+      "step": 2093
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.511787707515967e-06,
+      "loss": 0.0883,
+      "step": 2094
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.479821847971542e-06,
+      "loss": 0.1628,
+      "step": 2095
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.448058143509235e-06,
+      "loss": 0.1388,
+      "step": 2096
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4164966599738104e-06,
+      "loss": 0.1333,
+      "step": 2097
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3851374627908874e-06,
+      "loss": 0.0304,
+      "step": 2098
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.353980616966689e-06,
+      "loss": 0.0816,
+      "step": 2099
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3230261870879956e-06,
+      "loss": 0.1857,
+      "step": 2100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.292274237322012e-06,
+      "loss": 0.1848,
+      "step": 2101
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2617248314161675e-06,
+      "loss": 0.1506,
+      "step": 2102
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2313780326980727e-06,
+      "loss": 0.1497,
+      "step": 2103
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2012339040753173e-06,
+      "loss": 0.1939,
+      "step": 2104
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1712925080353384e-06,
+      "loss": 0.1327,
+      "step": 2105
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.141553906645388e-06,
+      "loss": 0.1985,
+      "step": 2106
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.112018161552254e-06,
+      "loss": 0.2242,
+      "step": 2107
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0826853339822817e-06,
+      "loss": 0.2347,
+      "step": 2108
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.053555484741121e-06,
+      "loss": 0.2318,
+      "step": 2109
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0246286742136913e-06,
+      "loss": 0.1718,
+      "step": 2110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9959049623640148e-06,
+      "loss": 0.1051,
+      "step": 2111
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9673844087350736e-06,
+      "loss": 0.142,
+      "step": 2112
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9390670724487638e-06,
+      "loss": 0.0418,
+      "step": 2113
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9109530122056742e-06,
+      "loss": 0.0855,
+      "step": 2114
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8830422862850305e-06,
+      "loss": 0.1675,
+      "step": 2115
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8553349525445407e-06,
+      "loss": 0.1598,
+      "step": 2116
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8278310684203159e-06,
+      "loss": 0.1136,
+      "step": 2117
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8005306909267162e-06,
+      "loss": 0.1005,
+      "step": 2118
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7734338766562164e-06,
+      "loss": 0.2437,
+      "step": 2119
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7465406817793296e-06,
+      "loss": 0.1532,
+      "step": 2120
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7198511620444946e-06,
+      "loss": 0.212,
+      "step": 2121
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.693365372777922e-06,
+      "loss": 0.2266,
+      "step": 2122
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.667083368883482e-06,
+      "loss": 0.0443,
+      "step": 2123
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6410052048426384e-06,
+      "loss": 0.2466,
+      "step": 2124
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.615130934714304e-06,
+      "loss": 0.1958,
+      "step": 2125
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5894606121346745e-06,
+      "loss": 0.2137,
+      "step": 2126
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.56399429031725e-06,
+      "loss": 0.1757,
+      "step": 2127
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5387320220525803e-06,
+      "loss": 0.1702,
+      "step": 2128
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5136738597082645e-06,
+      "loss": 0.1383,
+      "step": 2129
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4888198552287624e-06,
+      "loss": 0.1478,
+      "step": 2130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4641700601353725e-06,
+      "loss": 0.1514,
+      "step": 2131
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4397245255260095e-06,
+      "loss": 0.1681,
+      "step": 2132
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4154833020752267e-06,
+      "loss": 0.1874,
+      "step": 2133
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3914464400340167e-06,
+      "loss": 0.1489,
+      "step": 2134
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3676139892297546e-06,
+      "loss": 0.205,
+      "step": 2135
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3439859990660776e-06,
+      "loss": 0.0865,
+      "step": 2136
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3205625185227499e-06,
+      "loss": 0.1647,
+      "step": 2137
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2973435961556646e-06,
+      "loss": 0.2638,
+      "step": 2138
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.274329280096609e-06,
+      "loss": 0.2224,
+      "step": 2139
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.251519618053265e-06,
+      "loss": 0.1054,
+      "step": 2140
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2289146573090882e-06,
+      "loss": 0.1336,
+      "step": 2141
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.206514444723139e-06,
+      "loss": 0.1323,
+      "step": 2142
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1843190267301297e-06,
+      "loss": 0.188,
+      "step": 2143
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1623284493401443e-06,
+      "loss": 0.1766,
+      "step": 2144
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1405427581387406e-06,
+      "loss": 0.0598,
+      "step": 2145
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1189619982866828e-06,
+      "loss": 0.1208,
+      "step": 2146
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.09758621451993e-06,
+      "loss": 0.1602,
+      "step": 2147
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0764154511495928e-06,
+      "loss": 0.1424,
+      "step": 2148
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.055449752061699e-06,
+      "loss": 0.1674,
+      "step": 2149
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0346891607172614e-06,
+      "loss": 0.1832,
+      "step": 2150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0141337201520662e-06,
+      "loss": 0.1935,
+      "step": 2151
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.937834729766392e-07,
+      "loss": 0.1571,
+      "step": 2152
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.736384613761806e-07,
+      "loss": 0.1727,
+      "step": 2153
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.53698727110408e-07,
+      "loss": 0.1787,
+      "step": 2154
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.339643115135133e-07,
+      "loss": 0.2011,
+      "step": 2155
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.144352554940949e-07,
+      "loss": 0.1262,
+      "step": 2156
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.951115995350368e-07,
+      "loss": 0.1385,
+      "step": 2157
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.759933836934186e-07,
+      "loss": 0.2013,
+      "step": 2158
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.570806476004945e-07,
+      "loss": 0.0793,
+      "step": 2159
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.383734304615143e-07,
+      "loss": 0.0962,
+      "step": 2160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.198717710557358e-07,
+      "loss": 0.1692,
+      "step": 2161
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.015757077363129e-07,
+      "loss": 0.1492,
+      "step": 2162
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.834852784301628e-07,
+      "loss": 0.1879,
+      "step": 2163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.656005206379657e-07,
+      "loss": 0.0905,
+      "step": 2164
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.479214714340432e-07,
+      "loss": 0.0722,
+      "step": 2165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.304481674662911e-07,
+      "loss": 0.2238,
+      "step": 2166
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.13180644956124e-07,
+      "loss": 0.1032,
+      "step": 2167
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.961189396983537e-07,
+      "loss": 0.1071,
+      "step": 2168
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.792630870611439e-07,
+      "loss": 0.1748,
+      "step": 2169
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.626131219859555e-07,
+      "loss": 0.0571,
+      "step": 2170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.461690789874352e-07,
+      "loss": 0.1636,
+      "step": 2171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.299309921533714e-07,
+      "loss": 0.1034,
+      "step": 2172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.138988951446045e-07,
+      "loss": 0.0938,
+      "step": 2173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.980728211949837e-07,
+      "loss": 0.1749,
+      "step": 2174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.824528031112553e-07,
+      "loss": 0.1304,
+      "step": 2175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.670388732730625e-07,
+      "loss": 0.154,
+      "step": 2176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.518310636327794e-07,
+      "loss": 0.104,
+      "step": 2177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.368294057155554e-07,
+      "loss": 0.1674,
+      "step": 2178
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.22033930619159e-07,
+      "loss": 0.1406,
+      "step": 2179
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.07444669013979e-07,
+      "loss": 0.0565,
+      "step": 2180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.93061651142912e-07,
+      "loss": 0.0616,
+      "step": 2181
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.788849068213197e-07,
+      "loss": 0.2057,
+      "step": 2182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6491446543700526e-07,
+      "loss": 0.2318,
+      "step": 2183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.511503559500474e-07,
+      "loss": 0.1345,
+      "step": 2184
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.375926068928671e-07,
+      "loss": 0.1608,
+      "step": 2185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.24241246370094e-07,
+      "loss": 0.0932,
+      "step": 2186
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1109630205848906e-07,
+      "loss": 0.1836,
+      "step": 2187
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9815780120697756e-07,
+      "loss": 0.2273,
+      "step": 2188
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.854257706364939e-07,
+      "loss": 0.1791,
+      "step": 2189
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.729002367399925e-07,
+      "loss": 0.2305,
+      "step": 2190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.605812254823593e-07,
+      "loss": 0.1529,
+      "step": 2191
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.484687624003447e-07,
+      "loss": 0.1679,
+      "step": 2192
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.36562872602586e-07,
+      "loss": 0.1778,
+      "step": 2193
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.248635807694744e-07,
+      "loss": 0.1135,
+      "step": 2194
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.133709111530991e-07,
+      "loss": 0.1459,
+      "step": 2195
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0208488757730304e-07,
+      "loss": 0.11,
+      "step": 2196
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9100553343749395e-07,
+      "loss": 0.2145,
+      "step": 2197
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.801328717007001e-07,
+      "loss": 0.2216,
+      "step": 2198
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.694669249054926e-07,
+      "loss": 0.1492,
+      "step": 2199
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5900771516188527e-07,
+      "loss": 0.1949,
+      "step": 2200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4875526415137907e-07,
+      "loss": 0.1104,
+      "step": 2201
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3870959312685123e-07,
+      "loss": 0.1878,
+      "step": 2202
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.288707229125553e-07,
+      "loss": 0.1504,
+      "step": 2203
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1923867390403197e-07,
+      "loss": 0.1732,
+      "step": 2204
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0981346606809838e-07,
+      "loss": 0.2556,
+      "step": 2205
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0059511894279236e-07,
+      "loss": 0.1782,
+      "step": 2206
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9158365163736146e-07,
+      "loss": 0.1403,
+      "step": 2207
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8277908283217404e-07,
+      "loss": 0.1491,
+      "step": 2208
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.741814307787304e-07,
+      "loss": 0.1186,
+      "step": 2209
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6579071329957395e-07,
+      "loss": 0.1553,
+      "step": 2210
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5760694778830242e-07,
+      "loss": 0.1199,
+      "step": 2211
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4963015120950107e-07,
+      "loss": 0.1342,
+      "step": 2212
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.418603400987317e-07,
+      "loss": 0.1508,
+      "step": 2213
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3429753056246608e-07,
+      "loss": 0.1853,
+      "step": 2214
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2694173827807465e-07,
+      "loss": 0.1363,
+      "step": 2215
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1979297849381565e-07,
+      "loss": 0.1357,
+      "step": 2216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1285126602873508e-07,
+      "loss": 0.1523,
+      "step": 2217
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.061166152727111e-07,
+      "loss": 0.2478,
+      "step": 2218
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.958904018637638e-08,
+      "loss": 0.2049,
+      "step": 2219
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.326855430110692e-08,
+      "loss": 0.1381,
+      "step": 2220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.715517071899992e-08,
+      "loss": 0.0901,
+      "step": 2221
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.124890211284042e-08,
+      "loss": 0.1667,
+      "step": 2222
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.554976072605691e-08,
+      "loss": 0.223,
+      "step": 2223
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.005775837272132e-08,
+      "loss": 0.1666,
+      "step": 2224
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.477290643751577e-08,
+      "loss": 0.1381,
+      "step": 2225
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.969521587571026e-08,
+      "loss": 0.2423,
+      "step": 2226
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.482469721312944e-08,
+      "loss": 0.1153,
+      "step": 2227
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.016136054616372e-08,
+      "loss": 0.1775,
+      "step": 2228
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.570521554168039e-08,
+      "loss": 0.1897,
+      "step": 2229
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.14562714371014e-08,
+      "loss": 0.1159,
+      "step": 2230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.74145370402812e-08,
+      "loss": 0.1498,
+      "step": 2231
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3580020729573333e-08,
+      "loss": 0.1637,
+      "step": 2232
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.99527304537639e-08,
+      "loss": 0.2599,
+      "step": 2233
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6532673732049262e-08,
+      "loss": 0.0656,
+      "step": 2234
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3319857654080513e-08,
+      "loss": 0.2193,
+      "step": 2235
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0314288879874633e-08,
+      "loss": 0.0867,
+      "step": 2236
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7515973639836702e-08,
+      "loss": 0.232,
+      "step": 2237
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4924917734771005e-08,
+      "loss": 0.0731,
+      "step": 2238
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2541126535814408e-08,
+      "loss": 0.1749,
+      "step": 2239
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.036460498445857e-08,
+      "loss": 0.1313,
+      "step": 2240
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.395357592538844e-09,
+      "loss": 0.1749,
+      "step": 2241
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.633388442223165e-09,
+      "loss": 0.0854,
+      "step": 2242
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.078701186000956e-09,
+      "loss": 0.1896,
+      "step": 2243
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.731299046660919e-09,
+      "loss": 0.1032,
+      "step": 2244
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5911848173132414e-09,
+      "loss": 0.0912,
+      "step": 2245
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6583608613562895e-09,
+      "loss": 0.1549,
+      "step": 2246
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.328291125099143e-10,
+      "loss": 0.2782,
+      "step": 2247
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.145910747710424e-10,
+      "loss": 0.1912,
+      "step": 2248
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0364782240257498e-10,
+      "loss": 0.1935,
+      "step": 2249
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1588,
+      "step": 2250
+    },
+    {
+      "epoch": 3.0,
+      "step": 2250,
+      "total_flos": 6.879030907659878e+17,
+      "train_loss": 0.667836711179879,
+      "train_runtime": 31107.2234,
+      "train_samples_per_second": 1.157,
+      "train_steps_per_second": 0.072
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2250,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 50000,
+  "total_flos": 6.879030907659878e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}