diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12913 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9963382737576287,
+  "global_step": 2148,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 1.4717,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 1.4394,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 1.6015,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 1.5068,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.6053,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 1.5636,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 1.4664,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 1.2902,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 1.3183,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.4049,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 1.4321,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 1.3553,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.2785,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 1.2924,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.3955,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 1.3009,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 1.22,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 1.257,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 1.3151,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.2092,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 1.2724,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 1.2892,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 1.2776,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 1.3943,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.3105,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1544,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 1.1968,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 1.2995,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 1.2815,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.2486,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 1.2095,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 1.2916,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 1.2441,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 1.1681,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.3134,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 1.2234,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 1.193,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 1.1637,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 1.1679,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.1705,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 1.2658,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 1.2689,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 1.1513,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 1.2267,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2034,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 1.2554,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 1.3188,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 1.2369,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 1.2056,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1879,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 1.1966,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2614,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 1.2508,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 1.2303,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.183,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 1.215,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 1.1435,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 1.2562,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 1.1887,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2054,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 1.2011,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 1.1796,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 1.267,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 1.1658,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.2247,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999988626578683e-05,
+      "loss": 1.2006,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999954506340598e-05,
+      "loss": 1.219,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999989763936336e-05,
+      "loss": 1.1697,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999818025776324e-05,
+      "loss": 1.2861,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999715665760584e-05,
+      "loss": 1.2101,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959055954898e-05,
+      "loss": 1.1502,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999442707426083e-05,
+      "loss": 1.236,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999272109728216e-05,
+      "loss": 1.1992,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999907876684343e-05,
+      "loss": 1.1795,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998862679211524e-05,
+      "loss": 1.2148,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998623847324027e-05,
+      "loss": 1.2245,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998362271724203e-05,
+      "loss": 1.264,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999807795300706e-05,
+      "loss": 1.2176,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999777089181933e-05,
+      "loss": 1.2125,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997441088859477e-05,
+      "loss": 1.1754,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997088544877707e-05,
+      "loss": 1.234,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996713260675938e-05,
+      "loss": 1.1579,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996315237107823e-05,
+      "loss": 1.1571,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995894475078747e-05,
+      "loss": 1.2166,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995450975545808e-05,
+      "loss": 1.2772,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994984739517826e-05,
+      "loss": 1.1261,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999449576805534e-05,
+      "loss": 1.2533,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993984062270604e-05,
+      "loss": 1.276,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993449623327593e-05,
+      "loss": 1.268,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992892452441986e-05,
+      "loss": 1.1903,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992312550881167e-05,
+      "loss": 1.2145,
+      "step": 91
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991709919964227e-05,
+      "loss": 1.157,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991084561061965e-05,
+      "loss": 1.2749,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999043647559688e-05,
+      "loss": 1.2186,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989765665043152e-05,
+      "loss": 1.2154,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998907213092667e-05,
+      "loss": 1.1492,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988355874825003e-05,
+      "loss": 1.1904,
+      "step": 97
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987616898367405e-05,
+      "loss": 1.188,
+      "step": 98
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998685520323482e-05,
+      "loss": 1.2361,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986070791159855e-05,
+      "loss": 1.2308,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985263663926808e-05,
+      "loss": 1.2864,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984433823371636e-05,
+      "loss": 1.2753,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983581271381964e-05,
+      "loss": 1.1349,
+      "step": 103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982706009897076e-05,
+      "loss": 1.2002,
+      "step": 104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981808040907925e-05,
+      "loss": 1.2492,
+      "step": 105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980887366457093e-05,
+      "loss": 1.1327,
+      "step": 106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979943988638834e-05,
+      "loss": 1.2301,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978977909599033e-05,
+      "loss": 1.1957,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997798913153521e-05,
+      "loss": 1.2296,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997697765669653e-05,
+      "loss": 1.1441,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975943487383768e-05,
+      "loss": 1.2527,
+      "step": 111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974886625949344e-05,
+      "loss": 1.2105,
+      "step": 112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997380707479728e-05,
+      "loss": 1.1275,
+      "step": 113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972704836383216e-05,
+      "loss": 1.2446,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997157991321439e-05,
+      "loss": 1.1617,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970432307849654e-05,
+      "loss": 1.2486,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969262022899442e-05,
+      "loss": 1.1145,
+      "step": 117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996806906102579e-05,
+      "loss": 1.1679,
+      "step": 118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966853424942303e-05,
+      "loss": 1.2007,
+      "step": 119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996561511741417e-05,
+      "loss": 1.1321,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964354141258156e-05,
+      "loss": 1.2503,
+      "step": 121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963070499342575e-05,
+      "loss": 1.1545,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996176419458731e-05,
+      "loss": 1.181,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99604352299638e-05,
+      "loss": 1.2791,
+      "step": 124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995908360849501e-05,
+      "loss": 1.2134,
+      "step": 125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995770933325545e-05,
+      "loss": 1.1846,
+      "step": 126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995631240737117e-05,
+      "loss": 1.1478,
+      "step": 127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954892834019732e-05,
+      "loss": 1.2466,
+      "step": 128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953450616430217e-05,
+      "loss": 1.2402,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951985757883218e-05,
+      "loss": 1.2601,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950498261710822e-05,
+      "loss": 1.2187,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948988131296614e-05,
+      "loss": 1.1763,
+      "step": 132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947455370075663e-05,
+      "loss": 1.2212,
+      "step": 133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994589998153452e-05,
+      "loss": 1.1876,
+      "step": 134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944321969211198e-05,
+      "loss": 1.1339,
+      "step": 135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994272133669518e-05,
+      "loss": 1.172,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99410980876274e-05,
+      "loss": 1.1351,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939452225700236e-05,
+      "loss": 1.1818,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937783754657503e-05,
+      "loss": 1.1651,
+      "step": 139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993609267829445e-05,
+      "loss": 1.1706,
+      "step": 140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934379000457736e-05,
+      "loss": 1.2073,
+      "step": 141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932642725045442e-05,
+      "loss": 1.2186,
+      "step": 142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930883856007047e-05,
+      "loss": 1.2186,
+      "step": 143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929102397343417e-05,
+      "loss": 1.2422,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992729835310681e-05,
+      "loss": 1.2145,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992547172740086e-05,
+      "loss": 1.1659,
+      "step": 146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923622524380558e-05,
+      "loss": 1.2141,
+      "step": 147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921750748252266e-05,
+      "loss": 1.2255,
+      "step": 148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991985640327368e-05,
+      "loss": 1.1423,
+      "step": 149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991793949375383e-05,
+      "loss": 1.2387,
+      "step": 150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916000024053087e-05,
+      "loss": 1.1863,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991403799858313e-05,
+      "loss": 1.1563,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912053421806948e-05,
+      "loss": 1.2779,
+      "step": 153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910046298238827e-05,
+      "loss": 1.1193,
+      "step": 154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990801663244434e-05,
+      "loss": 1.2894,
+      "step": 155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905964429040335e-05,
+      "loss": 1.1416,
+      "step": 156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903889692694926e-05,
+      "loss": 1.1508,
+      "step": 157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990179242812748e-05,
+      "loss": 1.2213,
+      "step": 158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899672640108618e-05,
+      "loss": 1.2102,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897530333460187e-05,
+      "loss": 1.1932,
+      "step": 160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895365513055256e-05,
+      "loss": 1.2408,
+      "step": 161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989317818381811e-05,
+      "loss": 1.1277,
+      "step": 162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989096835072423e-05,
+      "loss": 1.1267,
+      "step": 163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988873601880029e-05,
+      "loss": 1.1168,
+      "step": 164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988648119312414e-05,
+      "loss": 1.2347,
+      "step": 165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884203878824798e-05,
+      "loss": 1.1614,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881904081082433e-05,
+      "loss": 1.1882,
+      "step": 167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879581805128357e-05,
+      "loss": 1.212,
+      "step": 168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987723705624502e-05,
+      "loss": 1.1359,
+      "step": 169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987486983976598e-05,
+      "loss": 1.1284,
+      "step": 170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987248016107591e-05,
+      "loss": 1.1582,
+      "step": 171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870068025610574e-05,
+      "loss": 1.1563,
+      "step": 172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867633438856823e-05,
+      "loss": 1.228,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865176406352562e-05,
+      "loss": 1.2773,
+      "step": 174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862696933686774e-05,
+      "loss": 1.1441,
+      "step": 175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860195026499473e-05,
+      "loss": 1.229,
+      "step": 176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857670690481708e-05,
+      "loss": 1.0738,
+      "step": 177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855123931375547e-05,
+      "loss": 1.2238,
+      "step": 178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985255475497406e-05,
+      "loss": 1.078,
+      "step": 179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984996316712132e-05,
+      "loss": 1.1612,
+      "step": 180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984734917371236e-05,
+      "loss": 1.216,
+      "step": 181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844712780693197e-05,
+      "loss": 1.2418,
+      "step": 182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842053994060793e-05,
+      "loss": 1.177,
+      "step": 183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839372819863046e-05,
+      "loss": 1.2282,
+      "step": 184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983666926419878e-05,
+      "loss": 1.2166,
+      "step": 185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983394333321773e-05,
+      "loss": 1.1608,
+      "step": 186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831195033120534e-05,
+      "loss": 1.2168,
+      "step": 187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98284243701587e-05,
+      "loss": 1.1598,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825631350634615e-05,
+      "loss": 1.2375,
+      "step": 189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822815980901518e-05,
+      "loss": 1.2102,
+      "step": 190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819978267363476e-05,
+      "loss": 1.1879,
+      "step": 191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817118216475407e-05,
+      "loss": 1.2183,
+      "step": 192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814235834743014e-05,
+      "loss": 1.2001,
+      "step": 193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811331128722807e-05,
+      "loss": 1.1948,
+      "step": 194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808404105022076e-05,
+      "loss": 1.2549,
+      "step": 195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805454770298877e-05,
+      "loss": 1.1271,
+      "step": 196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980248313126201e-05,
+      "loss": 1.1222,
+      "step": 197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799489194671023e-05,
+      "loss": 1.1922,
+      "step": 198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796472967336172e-05,
+      "loss": 1.202,
+      "step": 199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979343445611842e-05,
+      "loss": 1.1822,
+      "step": 200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790373667929422e-05,
+      "loss": 1.2069,
+      "step": 201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787290609731512e-05,
+      "loss": 1.1859,
+      "step": 202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784185288537662e-05,
+      "loss": 1.2879,
+      "step": 203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781057711411505e-05,
+      "loss": 1.0384,
+      "step": 204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777907885467293e-05,
+      "loss": 1.2297,
+      "step": 205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774735817869875e-05,
+      "loss": 1.2124,
+      "step": 206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771541515834714e-05,
+      "loss": 1.2308,
+      "step": 207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768324986627837e-05,
+      "loss": 1.1843,
+      "step": 208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765086237565826e-05,
+      "loss": 1.1966,
+      "step": 209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976182527601582e-05,
+      "loss": 1.2213,
+      "step": 210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975854210939547e-05,
+      "loss": 1.2288,
+      "step": 211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975523674517295e-05,
+      "loss": 1.2117,
+      "step": 212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751909190866916e-05,
+      "loss": 1.2378,
+      "step": 213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748559454046506e-05,
+      "loss": 1.2743,
+      "step": 214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974518754233131e-05,
+      "loss": 1.1426,
+      "step": 215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741793463391365e-05,
+      "loss": 1.2211,
+      "step": 216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973837722494713e-05,
+      "loss": 1.2085,
+      "step": 217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973493883476947e-05,
+      "loss": 1.1199,
+      "step": 218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973147830067963e-05,
+      "loss": 1.1606,
+      "step": 219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727995630549237e-05,
+      "loss": 1.1984,
+      "step": 220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724490832300266e-05,
+      "loss": 1.2662,
+      "step": 221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720963913905028e-05,
+      "loss": 1.2539,
+      "step": 222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717414883386148e-05,
+      "loss": 1.1718,
+      "step": 223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713843748816545e-05,
+      "loss": 1.189,
+      "step": 224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971025051831943e-05,
+      "loss": 1.1482,
+      "step": 225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970663520006826e-05,
+      "loss": 1.2271,
+      "step": 226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702997802286754e-05,
+      "loss": 1.159,
+      "step": 227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699338333248833e-05,
+      "loss": 1.2196,
+      "step": 228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695656801278635e-05,
+      "loss": 1.1728,
+      "step": 229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691953214750488e-05,
+      "loss": 1.2025,
+      "step": 230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688227582088878e-05,
+      "loss": 1.2473,
+      "step": 231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968447991176844e-05,
+      "loss": 1.1849,
+      "step": 232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968071021231395e-05,
+      "loss": 1.1599,
+      "step": 233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676918492300276e-05,
+      "loss": 1.2472,
+      "step": 234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673104760352387e-05,
+      "loss": 1.2208,
+      "step": 235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966926902514532e-05,
+      "loss": 1.2025,
+      "step": 236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665411295404156e-05,
+      "loss": 1.2443,
+      "step": 237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966153157990402e-05,
+      "loss": 1.2123,
+      "step": 238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657629887470038e-05,
+      "loss": 1.226,
+      "step": 239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653706226977323e-05,
+      "loss": 1.1827,
+      "step": 240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649760607350965e-05,
+      "loss": 1.1731,
+      "step": 241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645793037566008e-05,
+      "loss": 1.2196,
+      "step": 242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641803526647414e-05,
+      "loss": 1.2292,
+      "step": 243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637792083670072e-05,
+      "loss": 1.2433,
+      "step": 244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963375871775873e-05,
+      "loss": 1.1826,
+      "step": 245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962970343808804e-05,
+      "loss": 1.1897,
+      "step": 246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962562625388247e-05,
+      "loss": 1.1795,
+      "step": 247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962152717441634e-05,
+      "loss": 1.0895,
+      "step": 248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617406209013748e-05,
+      "loss": 1.1578,
+      "step": 249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613263367048595e-05,
+      "loss": 1.0988,
+      "step": 250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960909865794454e-05,
+      "loss": 1.1786,
+      "step": 251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604912091174978e-05,
+      "loss": 1.1984,
+      "step": 252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960070367626303e-05,
+      "loss": 1.2843,
+      "step": 253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596473422781507e-05,
+      "loss": 1.1384,
+      "step": 254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95922213403529e-05,
+      "loss": 1.1976,
+      "step": 255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587947438649362e-05,
+      "loss": 1.297,
+      "step": 256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958365172739266e-05,
+      "loss": 1.1631,
+      "step": 257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579334216354184e-05,
+      "loss": 1.0906,
+      "step": 258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957499491535491e-05,
+      "loss": 1.1082,
+      "step": 259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957063383426538e-05,
+      "loss": 1.1622,
+      "step": 260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956625098300567e-05,
+      "loss": 1.2316,
+      "step": 261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956184637154539e-05,
+      "loss": 1.2528,
+      "step": 262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557420009903636e-05,
+      "loss": 1.2346,
+      "step": 263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552971908148985e-05,
+      "loss": 1.1708,
+      "step": 264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548502076399458e-05,
+      "loss": 1.1591,
+      "step": 265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544010524822517e-05,
+      "loss": 1.2303,
+      "step": 266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539497263635027e-05,
+      "loss": 1.1443,
+      "step": 267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534962303103223e-05,
+      "loss": 1.19,
+      "step": 268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530405653542716e-05,
+      "loss": 1.1791,
+      "step": 269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525827325318442e-05,
+      "loss": 1.1365,
+      "step": 270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521227328844654e-05,
+      "loss": 1.1455,
+      "step": 271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951660567458489e-05,
+      "loss": 1.1289,
+      "step": 272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951196237305195e-05,
+      "loss": 1.1259,
+      "step": 273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507297434807888e-05,
+      "loss": 1.201,
+      "step": 274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502610870463955e-05,
+      "loss": 1.1428,
+      "step": 275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497902690680613e-05,
+      "loss": 1.1922,
+      "step": 276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493172906167484e-05,
+      "loss": 1.296,
+      "step": 277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948842152768333e-05,
+      "loss": 1.2057,
+      "step": 278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483648566036044e-05,
+      "loss": 1.1589,
+      "step": 279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478854032082598e-05,
+      "loss": 1.1562,
+      "step": 280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474037936729052e-05,
+      "loss": 1.2475,
+      "step": 281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469200290930494e-05,
+      "loss": 1.2004,
+      "step": 282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464341105691045e-05,
+      "loss": 1.134,
+      "step": 283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459460392063816e-05,
+      "loss": 1.1794,
+      "step": 284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454558161150895e-05,
+      "loss": 1.2167,
+      "step": 285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94496344241033e-05,
+      "loss": 1.1299,
+      "step": 286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944468919212098e-05,
+      "loss": 1.1845,
+      "step": 287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439722476452783e-05,
+      "loss": 1.1807,
+      "step": 288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434734288396418e-05,
+      "loss": 1.1416,
+      "step": 289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942972463929843e-05,
+      "loss": 1.1854,
+      "step": 290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424693540554195e-05,
+      "loss": 1.1516,
+      "step": 291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419641003607874e-05,
+      "loss": 1.09,
+      "step": 292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941456703995239e-05,
+      "loss": 1.2451,
+      "step": 293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409471661129414e-05,
+      "loss": 1.1746,
+      "step": 294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940435487872932e-05,
+      "loss": 1.1115,
+      "step": 295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399216704391172e-05,
+      "loss": 1.0982,
+      "step": 296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93940571498027e-05,
+      "loss": 1.2504,
+      "step": 297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938887622670025e-05,
+      "loss": 1.1533,
+      "step": 298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93836739468688e-05,
+      "loss": 1.1865,
+      "step": 299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937845032214188e-05,
+      "loss": 1.1739,
+      "step": 300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93732053644016e-05,
+      "loss": 1.1649,
+      "step": 301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367939085578576e-05,
+      "loss": 1.0735,
+      "step": 302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936265149765193e-05,
+      "loss": 1.2163,
+      "step": 303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357342612649254e-05,
+      "loss": 1.232,
+      "step": 304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935201244264659e-05,
+      "loss": 1.1653,
+      "step": 305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934666099976839e-05,
+      "loss": 1.2064,
+      "step": 306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341288296187494e-05,
+      "loss": 1.1383,
+      "step": 307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335894344125105e-05,
+      "loss": 1.0994,
+      "step": 308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933047915585077e-05,
+      "loss": 1.1814,
+      "step": 309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932504274368232e-05,
+      "loss": 1.1784,
+      "step": 310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319585119985886e-05,
+      "loss": 1.1661,
+      "step": 311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314106297175833e-05,
+      "loss": 1.2625,
+      "step": 312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930860628771476e-05,
+      "loss": 1.2726,
+      "step": 313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930308510411344e-05,
+      "loss": 1.1634,
+      "step": 314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297542758930836e-05,
+      "loss": 1.1552,
+      "step": 315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929197926477403e-05,
+      "loss": 1.2052,
+      "step": 316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286394634298206e-05,
+      "loss": 1.1897,
+      "step": 317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928078888020664e-05,
+      "loss": 1.2657,
+      "step": 318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275162015250656e-05,
+      "loss": 1.1693,
+      "step": 319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926951405222959e-05,
+      "loss": 1.1583,
+      "step": 320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263845003990782e-05,
+      "loss": 1.1679,
+      "step": 321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925815488342952e-05,
+      "loss": 1.2153,
+      "step": 322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925244370348903e-05,
+      "loss": 1.171,
+      "step": 323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246711477160453e-05,
+      "loss": 1.1764,
+      "step": 324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240958217482783e-05,
+      "loss": 1.1394,
+      "step": 325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923518393754288e-05,
+      "loss": 1.1531,
+      "step": 326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229388650475397e-05,
+      "loss": 1.1347,
+      "step": 327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922357236946279e-05,
+      "loss": 1.1427,
+      "step": 328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217735107735255e-05,
+      "loss": 1.268,
+      "step": 329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211876878570727e-05,
+      "loss": 1.1485,
+      "step": 330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205997695294822e-05,
+      "loss": 1.1996,
+      "step": 331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920009757128083e-05,
+      "loss": 1.1876,
+      "step": 332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194176519949668e-05,
+      "loss": 1.2266,
+      "step": 333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188234554769857e-05,
+      "loss": 1.2082,
+      "step": 334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182271689257494e-05,
+      "loss": 1.2875,
+      "step": 335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176287936976214e-05,
+      "loss": 1.1865,
+      "step": 336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170283311537165e-05,
+      "loss": 1.1937,
+      "step": 337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164257826598973e-05,
+      "loss": 1.1441,
+      "step": 338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158211495867716e-05,
+      "loss": 1.1047,
+      "step": 339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152144333096885e-05,
+      "loss": 1.1542,
+      "step": 340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914605635208736e-05,
+      "loss": 1.181,
+      "step": 341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139947566687374e-05,
+      "loss": 1.2321,
+      "step": 342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913381799079249e-05,
+      "loss": 1.2395,
+      "step": 343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127667638345553e-05,
+      "loss": 1.1813,
+      "step": 344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121496523336676e-05,
+      "loss": 1.1771,
+      "step": 345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115304659803196e-05,
+      "loss": 1.1702,
+      "step": 346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910909206182965e-05,
+      "loss": 1.1314,
+      "step": 347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910285874354773e-05,
+      "loss": 1.219,
+      "step": 348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909660471913627e-05,
+      "loss": 1.2025,
+      "step": 349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090330002821207e-05,
+      "loss": 1.1078,
+      "step": 350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908403460887553e-05,
+      "loss": 1.2186,
+      "step": 351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907771855161928e-05,
+      "loss": 1.1887,
+      "step": 352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907138184541949e-05,
+      "loss": 1.1649,
+      "step": 353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065024504690172e-05,
+      "loss": 1.1401,
+      "step": 354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058646543892253e-05,
+      "loss": 1.1406,
+      "step": 355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052247977533598e-05,
+      "loss": 1.2044,
+      "step": 356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045828820168914e-05,
+      "loss": 1.1851,
+      "step": 357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903938908639976e-05,
+      "loss": 1.1922,
+      "step": 358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032928790874502e-05,
+      "loss": 1.1851,
+      "step": 359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026447948288265e-05,
+      "loss": 1.0981,
+      "step": 360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901994657338292e-05,
+      "loss": 1.1349,
+      "step": 361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901342468094705e-05,
+      "loss": 1.1336,
+      "step": 362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006882285815893e-05,
+      "loss": 1.2705,
+      "step": 363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900031940287134e-05,
+      "loss": 1.1983,
+      "step": 364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993736047041868e-05,
+      "loss": 1.1481,
+      "step": 365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987132233302538e-05,
+      "loss": 1.2455,
+      "step": 366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980507976674943e-05,
+      "loss": 1.3446,
+      "step": 367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973863292227173e-05,
+      "loss": 1.2166,
+      "step": 368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896719819507379e-05,
+      "loss": 1.1852,
+      "step": 369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896051270037578e-05,
+      "loss": 1.2208,
+      "step": 370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953806823340536e-05,
+      "loss": 1.1541,
+      "step": 371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947080579221813e-05,
+      "loss": 1.1566,
+      "step": 372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940333983319687e-05,
+      "loss": 1.235,
+      "step": 373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933567050980542e-05,
+      "loss": 1.2012,
+      "step": 374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926779797597003e-05,
+      "loss": 1.2131,
+      "step": 375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919972238607936e-05,
+      "loss": 1.1663,
+      "step": 376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913144389498384e-05,
+      "loss": 1.1273,
+      "step": 377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906296265799547e-05,
+      "loss": 1.1685,
+      "step": 378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889942788308875e-05,
+      "loss": 1.1065,
+      "step": 379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892539256989385e-05,
+      "loss": 1.1729,
+      "step": 380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885630403170915e-05,
+      "loss": 1.1507,
+      "step": 381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887870133734879e-05,
+      "loss": 1.2105,
+      "step": 382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871752075284454e-05,
+      "loss": 1.1375,
+      "step": 383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864782632785277e-05,
+      "loss": 1.2125,
+      "step": 384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857793025704545e-05,
+      "loss": 1.146,
+      "step": 385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850783269941407e-05,
+      "loss": 1.19,
+      "step": 386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843753381440846e-05,
+      "loss": 1.2291,
+      "step": 387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883670337619363e-05,
+      "loss": 1.1684,
+      "step": 388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829633270236308e-05,
+      "loss": 1.2472,
+      "step": 389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882254307965113e-05,
+      "loss": 1.1395,
+      "step": 390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881543282056604e-05,
+      "loss": 1.2327,
+      "step": 391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808302509154638e-05,
+      "loss": 1.2428,
+      "step": 392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880115216163613e-05,
+      "loss": 1.168,
+      "step": 393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793981794275295e-05,
+      "loss": 1.2491,
+      "step": 394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878679142338246e-05,
+      "loss": 1.2567,
+      "step": 395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877958106531344e-05,
+      "loss": 1.2501,
+      "step": 396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772350736469533e-05,
+      "loss": 1.1345,
+      "step": 397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876510045329745e-05,
+      "loss": 1.2642,
+      "step": 398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757830232289293e-05,
+      "loss": 1.2016,
+      "step": 399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750540089982528e-05,
+      "loss": 1.2061,
+      "step": 400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743230042959918e-05,
+      "loss": 1.1883,
+      "step": 401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735900107849515e-05,
+      "loss": 1.1378,
+      "step": 402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872855030132461e-05,
+      "loss": 1.1498,
+      "step": 403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721180640103688e-05,
+      "loss": 1.0477,
+      "step": 404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8713791140950405e-05,
+      "loss": 1.1917,
+      "step": 405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870638182067353e-05,
+      "loss": 1.2271,
+      "step": 406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869895269612694e-05,
+      "loss": 1.2007,
+      "step": 407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869150378420954e-05,
+      "loss": 1.2042,
+      "step": 408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684035101865255e-05,
+      "loss": 1.1675,
+      "step": 409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676546666082975e-05,
+      "loss": 1.2355,
+      "step": 410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669038493896534e-05,
+      "loss": 1.1827,
+      "step": 411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866151060238465e-05,
+      "loss": 1.1474,
+      "step": 412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653963008670897e-05,
+      "loss": 1.2114,
+      "step": 413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864639572992367e-05,
+      "loss": 1.2332,
+      "step": 414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638808783356138e-05,
+      "loss": 1.2373,
+      "step": 415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863120218622621e-05,
+      "loss": 1.1972,
+      "step": 416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862357595583649e-05,
+      "loss": 1.2143,
+      "step": 417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861593010953425e-05,
+      "loss": 1.1628,
+      "step": 418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860826466471137e-05,
+      "loss": 1.208,
+      "step": 419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860057963880432e-05,
+      "loss": 1.2277,
+      "step": 420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8592875049294107e-05,
+      "loss": 1.2415,
+      "step": 421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858515091370624e-05,
+      "loss": 1.19,
+      "step": 422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577407249610688e-05,
+      "loss": 1.1189,
+      "step": 423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856964407462184e-05,
+      "loss": 1.1291,
+      "step": 424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8561861406398472e-05,
+      "loss": 1.272,
+      "step": 425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554059262643694e-05,
+      "loss": 1.1438,
+      "step": 426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546237661104923e-05,
+      "loss": 1.1797,
+      "step": 427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538396619573827e-05,
+      "loss": 1.2159,
+      "step": 428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530536155886304e-05,
+      "loss": 1.2523,
+      "step": 429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852265628792243e-05,
+      "loss": 1.1934,
+      "step": 430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851475703360641e-05,
+      "loss": 1.2205,
+      "step": 431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506838410906556e-05,
+      "loss": 1.2039,
+      "step": 432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498900437835237e-05,
+      "loss": 1.2439,
+      "step": 433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849094313244883e-05,
+      "loss": 1.113,
+      "step": 434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482966512847698e-05,
+      "loss": 1.0863,
+      "step": 435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847497059717613e-05,
+      "loss": 1.2663,
+      "step": 436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466955403622307e-05,
+      "loss": 1.1951,
+      "step": 437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845892095041827e-05,
+      "loss": 1.1718,
+      "step": 438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450867255839854e-05,
+      "loss": 1.2259,
+      "step": 439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844279433820668e-05,
+      "loss": 1.1898,
+      "step": 440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434702215882085e-05,
+      "loss": 1.1148,
+      "step": 441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426590907273087e-05,
+      "loss": 1.1661,
+      "step": 442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418460430830353e-05,
+      "loss": 1.2616,
+      "step": 443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841031080504816e-05,
+      "loss": 1.1619,
+      "step": 444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840214204846432e-05,
+      "loss": 1.1855,
+      "step": 445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393954179660182e-05,
+      "loss": 1.2489,
+      "step": 446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385747217260564e-05,
+      "loss": 1.176,
+      "step": 447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8377521179933707e-05,
+      "loss": 1.1447,
+      "step": 448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369276086391257e-05,
+      "loss": 1.2419,
+      "step": 449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361011955388193e-05,
+      "loss": 1.1833,
+      "step": 450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835272880572281e-05,
+      "loss": 1.2339,
+      "step": 451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344426656236646e-05,
+      "loss": 1.2059,
+      "step": 452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833610552581448e-05,
+      "loss": 1.19,
+      "step": 453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832776543338426e-05,
+      "loss": 1.1525,
+      "step": 454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831940639791705e-05,
+      "loss": 1.1986,
+      "step": 455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311028438427026e-05,
+      "loss": 1.2028,
+      "step": 456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8302631573971397e-05,
+      "loss": 1.1734,
+      "step": 457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294215823650382e-05,
+      "loss": 1.1524,
+      "step": 458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8285781206607147e-05,
+      "loss": 1.2589,
+      "step": 459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277327742027797e-05,
+      "loss": 1.1144,
+      "step": 460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826885544914128e-05,
+      "loss": 1.2293,
+      "step": 461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82603643472194e-05,
+      "loss": 1.1522,
+      "step": 462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825185445557672e-05,
+      "loss": 1.1918,
+      "step": 463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824332579357057e-05,
+      "loss": 1.17,
+      "step": 464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234778380600953e-05,
+      "loss": 1.117,
+      "step": 465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226212236110537e-05,
+      "loss": 1.1722,
+      "step": 466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217627379584602e-05,
+      "loss": 1.1993,
+      "step": 467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209023830550978e-05,
+      "loss": 1.1941,
+      "step": 468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820040160858003e-05,
+      "loss": 1.168,
+      "step": 469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191760733284586e-05,
+      "loss": 1.2401,
+      "step": 470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8183101224319906e-05,
+      "loss": 1.2243,
+      "step": 471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174423101383646e-05,
+      "loss": 1.2346,
+      "step": 472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165726384215797e-05,
+      "loss": 1.1089,
+      "step": 473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157011092598634e-05,
+      "loss": 1.1745,
+      "step": 474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81482772463567e-05,
+      "loss": 1.1749,
+      "step": 475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813952486535674e-05,
+      "loss": 1.2021,
+      "step": 476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130753969507654e-05,
+      "loss": 1.2345,
+      "step": 477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812196457876046e-05,
+      "loss": 1.2219,
+      "step": 478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113156713108247e-05,
+      "loss": 1.17,
+      "step": 479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104330392586135e-05,
+      "loss": 1.1589,
+      "step": 480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809548563727121e-05,
+      "loss": 1.2126,
+      "step": 481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086622467282497e-05,
+      "loss": 1.2074,
+      "step": 482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077740902780913e-05,
+      "loss": 1.0972,
+      "step": 483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068840963969208e-05,
+      "loss": 1.2737,
+      "step": 484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805992267109194e-05,
+      "loss": 1.1111,
+      "step": 485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050986044435405e-05,
+      "loss": 1.1575,
+      "step": 486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80420311043276e-05,
+      "loss": 1.1572,
+      "step": 487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80330578711382e-05,
+      "loss": 1.2203,
+      "step": 488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024066365278475e-05,
+      "loss": 1.1889,
+      "step": 489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015056607201253e-05,
+      "loss": 1.0945,
+      "step": 490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8006028617400896e-05,
+      "loss": 1.0848,
+      "step": 491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996982416413223e-05,
+      "loss": 1.1379,
+      "step": 492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987918024815497e-05,
+      "loss": 1.1773,
+      "step": 493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978835463226338e-05,
+      "loss": 1.2081,
+      "step": 494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969734752305707e-05,
+      "loss": 1.1862,
+      "step": 495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796061591275485e-05,
+      "loss": 1.1603,
+      "step": 496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951478965316245e-05,
+      "loss": 1.1829,
+      "step": 497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7942323930773568e-05,
+      "loss": 1.274,
+      "step": 498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933150829951626e-05,
+      "loss": 1.2594,
+      "step": 499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792395968371633e-05,
+      "loss": 1.1782,
+      "step": 500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914750512974635e-05,
+      "loss": 1.1877,
+      "step": 501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79055233386745e-05,
+      "loss": 1.1477,
+      "step": 502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896278181804824e-05,
+      "loss": 1.073,
+      "step": 503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788701506339543e-05,
+      "loss": 1.1325,
+      "step": 504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877734004516986e-05,
+      "loss": 1.1092,
+      "step": 505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868435026280965e-05,
+      "loss": 1.1498,
+      "step": 506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785911814983961e-05,
+      "loss": 1.1628,
+      "step": 507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849783396385873e-05,
+      "loss": 1.0997,
+      "step": 508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784043078715337e-05,
+      "loss": 1.2472,
+      "step": 509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783106034341634e-05,
+      "loss": 1.1879,
+      "step": 510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782167208648957e-05,
+      "loss": 1.241,
+      "step": 511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812266037728395e-05,
+      "loss": 1.1788,
+      "step": 512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.78028422185286e-05,
+      "loss": 1.1961,
+      "step": 513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77934006503264e-05,
+      "loss": 1.138,
+      "step": 514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783941354598376e-05,
+      "loss": 1.2101,
+      "step": 515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774464352861446e-05,
+      "loss": 1.1435,
+      "step": 516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776496966667279e-05,
+      "loss": 1.1957,
+      "step": 517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775545731762983e-05,
+      "loss": 1.1467,
+      "step": 518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774592732737015e-05,
+      "loss": 1.2863,
+      "step": 519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736379717571467e-05,
+      "loss": 1.1845,
+      "step": 520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726814509951588e-05,
+      "loss": 1.1712,
+      "step": 521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771723172626833e-05,
+      "loss": 1.1627,
+      "step": 522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770763138831951e-05,
+      "loss": 1.1607,
+      "step": 523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698013517942857e-05,
+      "loss": 1.1639,
+      "step": 524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688378137015993e-05,
+      "loss": 1.1406,
+      "step": 525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678725267456368e-05,
+      "loss": 1.1745,
+      "step": 526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7669054931221213e-05,
+      "loss": 1.1929,
+      "step": 527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659367150307486e-05,
+      "loss": 1.2069,
+      "step": 528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649661946751834e-05,
+      "loss": 1.1996,
+      "step": 529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763993934263053e-05,
+      "loss": 1.1616,
+      "step": 530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630199360059425e-05,
+      "loss": 1.1172,
+      "step": 531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762044202119391e-05,
+      "loss": 1.2319,
+      "step": 532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610667348228845e-05,
+      "loss": 1.1339,
+      "step": 533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7600875363398528e-05,
+      "loss": 1.2145,
+      "step": 534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7591066088976632e-05,
+      "loss": 1.1435,
+      "step": 535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7581239547276156e-05,
+      "loss": 1.1795,
+      "step": 536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571395760649382e-05,
+      "loss": 1.1812,
+      "step": 537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756153475148782e-05,
+      "loss": 1.1643,
+      "step": 538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7551656542222147e-05,
+      "loss": 1.1671,
+      "step": 539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541761155322175e-05,
+      "loss": 1.129,
+      "step": 540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7531848613296782e-05,
+      "loss": 1.255,
+      "step": 541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752191893869387e-05,
+      "loss": 1.1061,
+      "step": 542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511972154100316e-05,
+      "loss": 1.1279,
+      "step": 543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502008282141916e-05,
+      "loss": 1.2521,
+      "step": 544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492027345483328e-05,
+      "loss": 1.2057,
+      "step": 545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7482029366828034e-05,
+      "loss": 1.1596,
+      "step": 546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747201436891828e-05,
+      "loss": 1.15,
+      "step": 547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461982374535022e-05,
+      "loss": 1.1808,
+      "step": 548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745193340649788e-05,
+      "loss": 1.1758,
+      "step": 549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441867487665084e-05,
+      "loss": 1.324,
+      "step": 550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431784640933423e-05,
+      "loss": 1.2155,
+      "step": 551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742168488923819e-05,
+      "loss": 1.1455,
+      "step": 552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411568255553124e-05,
+      "loss": 1.149,
+      "step": 553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740143476289038e-05,
+      "loss": 1.1563,
+      "step": 554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391284434300454e-05,
+      "loss": 1.1617,
+      "step": 555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381117292872133e-05,
+      "loss": 1.2069,
+      "step": 556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737093336173246e-05,
+      "loss": 1.1508,
+      "step": 557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7360732664046663e-05,
+      "loss": 1.2176,
+      "step": 558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73505152230181e-05,
+      "loss": 1.1715,
+      "step": 559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734028106188823e-05,
+      "loss": 1.2187,
+      "step": 560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330030203936542e-05,
+      "loss": 1.1669,
+      "step": 561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7319762672480496e-05,
+      "loss": 1.1905,
+      "step": 562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730947849087548e-05,
+      "loss": 1.1859,
+      "step": 563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7299177682514767e-05,
+      "loss": 1.2081,
+      "step": 564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728886027082944e-05,
+      "loss": 1.121,
+      "step": 565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727852627928836e-05,
+      "loss": 1.1692,
+      "step": 566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268175731398086e-05,
+      "loss": 1.0347,
+      "step": 567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725780865070285e-05,
+      "loss": 1.1838,
+      "step": 568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724742506078449e-05,
+      "loss": 1.1592,
+      "step": 569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723702498526239e-05,
+      "loss": 1.2058,
+      "step": 570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226608447793438e-05,
+      "loss": 1.1428,
+      "step": 571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721617547207197e-05,
+      "loss": 1.1725,
+      "step": 572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7205726081829713e-05,
+      "loss": 1.235,
+      "step": 573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195260300835733e-05,
+      "loss": 1.0361,
+      "step": 574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7184778152896367e-05,
+      "loss": 1.152,
+      "step": 575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7174279661855206e-05,
+      "loss": 1.1899,
+      "step": 576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716376485159299e-05,
+      "loss": 1.1411,
+      "step": 577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153233746027596e-05,
+      "loss": 1.1026,
+      "step": 578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714268636911397e-05,
+      "loss": 1.1394,
+      "step": 579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713212274484405e-05,
+      "loss": 1.2067,
+      "step": 580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121542897246765e-05,
+      "loss": 1.172,
+      "step": 581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7110946850387914e-05,
+      "loss": 1.2459,
+      "step": 582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100334628370165e-05,
+      "loss": 1.1609,
+      "step": 583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7089706255332966e-05,
+      "loss": 1.1553,
+      "step": 584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7079061755452518e-05,
+      "loss": 1.2186,
+      "step": 585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068401152941692e-05,
+      "loss": 1.1506,
+      "step": 586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705772447204999e-05,
+      "loss": 1.1598,
+      "step": 587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70470317370635e-05,
+      "loss": 1.2171,
+      "step": 588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036322972304812e-05,
+      "loss": 1.183,
+      "step": 589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025598202132978e-05,
+      "loss": 1.1113,
+      "step": 590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7014857450943476e-05,
+      "loss": 1.2317,
+      "step": 591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004100743168116e-05,
+      "loss": 1.1582,
+      "step": 592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6993328103275016e-05,
+      "loss": 1.2284,
+      "step": 593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982539555768527e-05,
+      "loss": 1.2533,
+      "step": 594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6971735125189187e-05,
+      "loss": 1.107,
+      "step": 595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696091483611367e-05,
+      "loss": 1.0996,
+      "step": 596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6950078713154716e-05,
+      "loss": 1.1809,
+      "step": 597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693922678096108e-05,
+      "loss": 1.1708,
+      "step": 598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6928359064217484e-05,
+      "loss": 1.1636,
+      "step": 599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691747558764455e-05,
+      "loss": 1.1048,
+      "step": 600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6906576375998755e-05,
+      "loss": 1.1933,
+      "step": 601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689566145407236e-05,
+      "loss": 1.2491,
+      "step": 602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884730846693368e-05,
+      "loss": 1.0875,
+      "step": 603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873784578725456e-05,
+      "loss": 1.153,
+      "step": 604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6862822675067933e-05,
+      "loss": 1.1399,
+      "step": 605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851845160655665e-05,
+      "loss": 1.2338,
+      "step": 606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840852060459035e-05,
+      "loss": 1.1503,
+      "step": 607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6829843399483872e-05,
+      "loss": 1.173,
+      "step": 608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818819202771405e-05,
+      "loss": 1.1897,
+      "step": 609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68077794953982e-05,
+      "loss": 1.1458,
+      "step": 610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796724302476106e-05,
+      "loss": 1.2346,
+      "step": 611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67856536491522e-05,
+      "loss": 1.1245,
+      "step": 612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774567560608712e-05,
+      "loss": 1.164,
+      "step": 613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6763466062063005e-05,
+      "loss": 1.1393,
+      "step": 614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675234917876748e-05,
+      "loss": 1.1972,
+      "step": 615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6741216936009532e-05,
+      "loss": 1.2736,
+      "step": 616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67300693591115e-05,
+      "loss": 1.177,
+      "step": 617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671890647343061e-05,
+      "loss": 1.1944,
+      "step": 618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670772830435889e-05,
+      "loss": 1.1668,
+      "step": 619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6696534877323155e-05,
+      "loss": 1.2268,
+      "step": 620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668532621778491e-05,
+      "loss": 1.1614,
+      "step": 621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674102351240317e-05,
+      "loss": 1.2074,
+      "step": 622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662863303220134e-05,
+      "loss": 1.2491,
+      "step": 623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651609099289645e-05,
+      "loss": 1.1345,
+      "step": 624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640339765048612e-05,
+      "loss": 1.28,
+      "step": 625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6629055326131205e-05,
+      "loss": 1.1877,
+      "step": 626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6617755808205966e-05,
+      "loss": 1.2251,
+      "step": 627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606441236975732e-05,
+      "loss": 1.0982,
+      "step": 628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6595111638177573e-05,
+      "loss": 1.0871,
+      "step": 629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583767037582755e-05,
+      "loss": 1.1261,
+      "step": 630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657240746099666e-05,
+      "loss": 1.1676,
+      "step": 631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561032934258744e-05,
+      "loss": 1.1646,
+      "step": 632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6549643483242453e-05,
+      "loss": 1.1672,
+      "step": 633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538239133855197e-05,
+      "loss": 1.2043,
+      "step": 634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526819912038275e-05,
+      "loss": 1.1851,
+      "step": 635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515385843766805e-05,
+      "loss": 1.1836,
+      "step": 636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6503936955049684e-05,
+      "loss": 1.1813,
+      "step": 637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649247327192952e-05,
+      "loss": 1.2147,
+      "step": 638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648099482048257e-05,
+      "loss": 1.1937,
+      "step": 639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469501626818694e-05,
+      "loss": 1.1912,
+      "step": 640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645799371708127e-05,
+      "loss": 1.1884,
+      "step": 641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6446471117447162e-05,
+      "loss": 1.2361,
+      "step": 642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434933854126646e-05,
+      "loss": 1.1961,
+      "step": 643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423381953363355e-05,
+      "loss": 1.1824,
+      "step": 644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6411815441434214e-05,
+      "loss": 1.2111,
+      "step": 645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640023434464939e-05,
+      "loss": 1.1995,
+      "step": 646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388638689352216e-05,
+      "loss": 1.2135,
+      "step": 647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637702850191915e-05,
+      "loss": 1.2705,
+      "step": 648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63654038087597e-05,
+      "loss": 1.1969,
+      "step": 649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353764636316375e-05,
+      "loss": 1.2371,
+      "step": 650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6342111011064616e-05,
+      "loss": 1.2054,
+      "step": 651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6330442959512743e-05,
+      "loss": 1.1821,
+      "step": 652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6318760508201887e-05,
+      "loss": 1.1781,
+      "step": 653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307063683705937e-05,
+      "loss": 1.1514,
+      "step": 654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295352512631474e-05,
+      "loss": 1.1559,
+      "step": 655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283627021617716e-05,
+      "loss": 1.2323,
+      "step": 656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271887237336454e-05,
+      "loss": 1.1605,
+      "step": 657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626013318649199e-05,
+      "loss": 1.1884,
+      "step": 658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624836489582108e-05,
+      "loss": 1.1566,
+      "step": 659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6236582392092863e-05,
+      "loss": 1.0873,
+      "step": 660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6224785702108822e-05,
+      "loss": 1.157,
+      "step": 661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6212974852702702e-05,
+      "loss": 1.1681,
+      "step": 662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6201149870740448e-05,
+      "loss": 1.1802,
+      "step": 663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189310783120173e-05,
+      "loss": 1.2086,
+      "step": 664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617745761677205e-05,
+      "loss": 1.2703,
+      "step": 665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61655903986583e-05,
+      "loss": 1.171,
+      "step": 666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615370915577309e-05,
+      "loss": 1.1751,
+      "step": 667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6141813915142505e-05,
+      "loss": 1.2857,
+      "step": 668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129904703824454e-05,
+      "loss": 1.1493,
+      "step": 669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6117981548908638e-05,
+      "loss": 1.103,
+      "step": 670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106044477516466e-05,
+      "loss": 1.1476,
+      "step": 671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094093516801005e-05,
+      "loss": 1.2008,
+      "step": 672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608212869394692e-05,
+      "loss": 1.1321,
+      "step": 673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070150036170408e-05,
+      "loss": 1.1623,
+      "step": 674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6058157570719125e-05,
+      "loss": 1.1381,
+      "step": 675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6046151324872153e-05,
+      "loss": 1.1885,
+      "step": 676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603413132593991e-05,
+      "loss": 1.1757,
+      "step": 677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6022097601264087e-05,
+      "loss": 1.1659,
+      "step": 678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010050178217624e-05,
+      "loss": 1.1627,
+      "step": 679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599798908420459e-05,
+      "loss": 1.2379,
+      "step": 680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985914346660177e-05,
+      "loss": 1.1305,
+      "step": 681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597382599305059e-05,
+      "loss": 1.1694,
+      "step": 682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961724050873025e-05,
+      "loss": 1.1812,
+      "step": 683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949608547655582e-05,
+      "loss": 1.2042,
+      "step": 684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937479510957196e-05,
+      "loss": 1.2699,
+      "step": 685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925336968367598e-05,
+      "loss": 1.0963,
+      "step": 686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5913180947507244e-05,
+      "loss": 1.1765,
+      "step": 687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590101147602724e-05,
+      "loss": 1.2056,
+      "step": 688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5888828581609294e-05,
+      "loss": 1.1901,
+      "step": 689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5876632291965637e-05,
+      "loss": 1.2822,
+      "step": 690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864422634838987e-05,
+      "loss": 1.127,
+      "step": 691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585219963800245e-05,
+      "loss": 1.2087,
+      "step": 692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583996332925949e-05,
+      "loss": 1.1694,
+      "step": 693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827713736443844e-05,
+      "loss": 1.1093,
+      "step": 694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581545088741947e-05,
+      "loss": 1.2218,
+      "step": 695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580317481008047e-05,
+      "loss": 1.1349,
+      "step": 696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579088553235106e-05,
+      "loss": 1.1091,
+      "step": 697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577858308218545e-05,
+      "loss": 1.2632,
+      "step": 698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5766267487567836e-05,
+      "loss": 1.1202,
+      "step": 699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575393877651231e-05,
+      "loss": 1.1587,
+      "step": 700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741596977062795e-05,
+      "loss": 1.15,
+      "step": 701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5729242117292986e-05,
+      "loss": 1.1889,
+      "step": 702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716874225306292e-05,
+      "loss": 1.0798,
+      "step": 703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704493329235756e-05,
+      "loss": 1.2704,
+      "step": 704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5692099457244017e-05,
+      "loss": 1.1857,
+      "step": 705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5679692637523213e-05,
+      "loss": 1.1825,
+      "step": 706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5667272898294942e-05,
+      "loss": 1.1883,
+      "step": 707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654840267810196e-05,
+      "loss": 1.1555,
+      "step": 708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5642394774349275e-05,
+      "loss": 1.1231,
+      "step": 709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562993644622175e-05,
+      "loss": 1.2109,
+      "step": 710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561746531176639e-05,
+      "loss": 1.1532,
+      "step": 711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560498139935108e-05,
+      "loss": 1.2574,
+      "step": 712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5592484737372782e-05,
+      "loss": 1.1988,
+      "step": 713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557997535425746e-05,
+      "loss": 1.1886,
+      "step": 714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5567453278460006e-05,
+      "loss": 1.1012,
+      "step": 715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5554918538464186e-05,
+      "loss": 1.1243,
+      "step": 716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5542371162782588e-05,
+      "loss": 1.0718,
+      "step": 717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552981117995652e-05,
+      "loss": 0.9951,
+      "step": 718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5517238618555976e-05,
+      "loss": 0.926,
+      "step": 719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550465350717957e-05,
+      "loss": 0.8722,
+      "step": 720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492055874454454e-05,
+      "loss": 0.8137,
+      "step": 721
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547944574903627e-05,
+      "loss": 0.885,
+      "step": 722
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546682315960906e-05,
+      "loss": 0.8681,
+      "step": 723
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545418813488524e-05,
+      "loss": 0.78,
+      "step": 724
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5441540703605496e-05,
+      "loss": 0.877,
+      "step": 725
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5428880894538745e-05,
+      "loss": 0.8386,
+      "step": 726
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5416208736482054e-05,
+      "loss": 0.9246,
+      "step": 727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540352425826058e-05,
+      "loss": 0.9193,
+      "step": 728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390827488727506e-05,
+      "loss": 0.8523,
+      "step": 729
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378118456763977e-05,
+      "loss": 0.903,
+      "step": 730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365397191279025e-05,
+      "loss": 0.8741,
+      "step": 731
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352663721209515e-05,
+      "loss": 0.8557,
+      "step": 732
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533991807552007e-05,
+      "loss": 0.9143,
+      "step": 733
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5327160283203008e-05,
+      "loss": 0.9255,
+      "step": 734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314390373278278e-05,
+      "loss": 0.9299,
+      "step": 735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5301608374793398e-05,
+      "loss": 0.8338,
+      "step": 736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288814316823374e-05,
+      "loss": 0.8034,
+      "step": 737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527600822847065e-05,
+      "loss": 0.8677,
+      "step": 738
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526319013886503e-05,
+      "loss": 0.8384,
+      "step": 739
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250360077163629e-05,
+      "loss": 0.8242,
+      "step": 740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237518072550781e-05,
+      "loss": 0.8574,
+      "step": 741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5224664154237991e-05,
+      "loss": 0.8175,
+      "step": 742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211798351463867e-05,
+      "loss": 0.9271,
+      "step": 743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198920693494049e-05,
+      "loss": 0.9223,
+      "step": 744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186031209621139e-05,
+      "loss": 0.8342,
+      "step": 745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173129929164645e-05,
+      "loss": 0.8286,
+      "step": 746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160216881470908e-05,
+      "loss": 0.8439,
+      "step": 747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5147292095913033e-05,
+      "loss": 0.8282,
+      "step": 748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5134355601890827e-05,
+      "loss": 0.9179,
+      "step": 749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512140742883073e-05,
+      "loss": 0.8347,
+      "step": 750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108447606185745e-05,
+      "loss": 0.9676,
+      "step": 751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509547616343538e-05,
+      "loss": 0.8379,
+      "step": 752
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082493130085571e-05,
+      "loss": 0.8115,
+      "step": 753
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069498535668618e-05,
+      "loss": 0.8005,
+      "step": 754
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056492409743117e-05,
+      "loss": 0.7604,
+      "step": 755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5043474781893906e-05,
+      "loss": 0.8077,
+      "step": 756
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030445681731975e-05,
+      "loss": 0.8192,
+      "step": 757
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017405138894413e-05,
+      "loss": 0.8141,
+      "step": 758
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004353183044331e-05,
+      "loss": 0.877,
+      "step": 759
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991289843870817e-05,
+      "loss": 0.9167,
+      "step": 760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978215151088842e-05,
+      "loss": 0.9478,
+      "step": 761
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965129134439199e-05,
+      "loss": 0.9403,
+      "step": 762
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4952031823688446e-05,
+      "loss": 0.9017,
+      "step": 763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938923248628832e-05,
+      "loss": 0.8269,
+      "step": 764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925803439078224e-05,
+      "loss": 0.866,
+      "step": 765
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912672424880046e-05,
+      "loss": 0.9083,
+      "step": 766
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899530235903213e-05,
+      "loss": 0.8554,
+      "step": 767
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488637690204205e-05,
+      "loss": 0.9207,
+      "step": 768
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873212453216244e-05,
+      "loss": 0.8745,
+      "step": 769
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4860036919370755e-05,
+      "loss": 0.8551,
+      "step": 770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846850330475766e-05,
+      "loss": 0.8603,
+      "step": 771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48336527165266e-05,
+      "loss": 0.8333,
+      "step": 772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820444107543665e-05,
+      "loss": 0.878,
+      "step": 773
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4807224533572376e-05,
+      "loss": 0.8964,
+      "step": 774
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4793994024683087e-05,
+      "loss": 0.9128,
+      "step": 775
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780752610971027e-05,
+      "loss": 0.9211,
+      "step": 776
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4767500322556237e-05,
+      "loss": 0.7794,
+      "step": 777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754237189583485e-05,
+      "loss": 0.895,
+      "step": 778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4740963242222213e-05,
+      "loss": 0.923,
+      "step": 779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727678510666458e-05,
+      "loss": 0.9196,
+      "step": 780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714383025134792e-05,
+      "loss": 0.8562,
+      "step": 781
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701076815870246e-05,
+      "loss": 0.9224,
+      "step": 782
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687759913140245e-05,
+      "loss": 0.8321,
+      "step": 783
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674432347236538e-05,
+      "loss": 0.8289,
+      "step": 784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661094148475126e-05,
+      "loss": 0.8618,
+      "step": 785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647745347196205e-05,
+      "loss": 0.8229,
+      "step": 786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4634385973764081e-05,
+      "loss": 0.7998,
+      "step": 787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462101605856711e-05,
+      "loss": 0.8487,
+      "step": 788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607635632017628e-05,
+      "loss": 0.9288,
+      "step": 789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459424472455188e-05,
+      "loss": 0.8488,
+      "step": 790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580843366629953e-05,
+      "loss": 0.8621,
+      "step": 791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4567431588735707e-05,
+      "loss": 0.883,
+      "step": 792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45540094213767e-05,
+      "loss": 0.9022,
+      "step": 793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4540576895084129e-05,
+      "loss": 0.8648,
+      "step": 794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4527134040412743e-05,
+      "loss": 0.8755,
+      "step": 795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513680887940798e-05,
+      "loss": 0.8138,
+      "step": 796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500217468269963e-05,
+      "loss": 0.9024,
+      "step": 797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448674381202527e-05,
+      "loss": 0.8732,
+      "step": 798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4473259949855034e-05,
+      "loss": 0.8694,
+      "step": 799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4459765912430783e-05,
+      "loss": 0.857,
+      "step": 800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4446261730447191e-05,
+      "loss": 0.883,
+      "step": 801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432747434622007e-05,
+      "loss": 0.8459,
+      "step": 802
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441922305569599e-05,
+      "loss": 0.9156,
+      "step": 803
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440568862443283e-05,
+      "loss": 0.8347,
+      "step": 804
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4392144171619085e-05,
+      "loss": 0.9418,
+      "step": 805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.437858972806411e-05,
+      "loss": 0.854,
+      "step": 806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436502532459998e-05,
+      "loss": 0.8598,
+      "step": 807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351450992081435e-05,
+      "loss": 0.9002,
+      "step": 808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337866761385798e-05,
+      "loss": 0.8616,
+      "step": 809
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324272663412897e-05,
+      "loss": 0.849,
+      "step": 810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310668729085016e-05,
+      "loss": 0.805,
+      "step": 811
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4297054989346812e-05,
+      "loss": 0.8386,
+      "step": 812
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4283431475165245e-05,
+      "loss": 0.9254,
+      "step": 813
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269798217529507e-05,
+      "loss": 0.7647,
+      "step": 814
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4256155247450951e-05,
+      "loss": 0.8474,
+      "step": 815
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4242502595963032e-05,
+      "loss": 0.8706,
+      "step": 816
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422884029412122e-05,
+      "loss": 0.8341,
+      "step": 817
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215168373002937e-05,
+      "loss": 0.8343,
+      "step": 818
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4201486863707485e-05,
+      "loss": 0.8659,
+      "step": 819
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418779579735598e-05,
+      "loss": 0.9398,
+      "step": 820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174095205091276e-05,
+      "loss": 0.8822,
+      "step": 821
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4160385118077895e-05,
+      "loss": 0.8781,
+      "step": 822
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146665567501957e-05,
+      "loss": 0.8397,
+      "step": 823
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41329365845711e-05,
+      "loss": 0.8572,
+      "step": 824
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4119198200514435e-05,
+      "loss": 0.8202,
+      "step": 825
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105450446582445e-05,
+      "loss": 0.817,
+      "step": 826
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091693354046928e-05,
+      "loss": 0.8989,
+      "step": 827
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077926954200925e-05,
+      "loss": 0.9239,
+      "step": 828
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064151278358651e-05,
+      "loss": 0.8314,
+      "step": 829
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405036635785542e-05,
+      "loss": 0.8113,
+      "step": 830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4036572224047574e-05,
+      "loss": 0.8797,
+      "step": 831
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022768908312408e-05,
+      "loss": 0.8048,
+      "step": 832
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400895644204811e-05,
+      "loss": 0.8954,
+      "step": 833
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995134856673677e-05,
+      "loss": 0.8406,
+      "step": 834
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.398130418362886e-05,
+      "loss": 0.9708,
+      "step": 835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967464454374063e-05,
+      "loss": 0.7799,
+      "step": 836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953615700390306e-05,
+      "loss": 0.8598,
+      "step": 837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3939757953179132e-05,
+      "loss": 0.7912,
+      "step": 838
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925891244262539e-05,
+      "loss": 0.8484,
+      "step": 839
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391201560518291e-05,
+      "loss": 0.8313,
+      "step": 840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898131067502947e-05,
+      "loss": 0.8837,
+      "step": 841
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3884237662805585e-05,
+      "loss": 0.8208,
+      "step": 842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870335422693938e-05,
+      "loss": 0.9286,
+      "step": 843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3856424378791205e-05,
+      "loss": 0.89,
+      "step": 844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3842504562740624e-05,
+      "loss": 0.8884,
+      "step": 845
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3828576006205383e-05,
+      "loss": 0.8353,
+      "step": 846
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814638740868549e-05,
+      "loss": 0.8545,
+      "step": 847
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3800692798432997e-05,
+      "loss": 0.8845,
+      "step": 848
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3786738210621346e-05,
+      "loss": 0.7829,
+      "step": 849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772775009175875e-05,
+      "loss": 0.8549,
+      "step": 850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758803225858463e-05,
+      "loss": 0.8659,
+      "step": 851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3744822892450504e-05,
+      "loss": 0.7929,
+      "step": 852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373083404075284e-05,
+      "loss": 0.929,
+      "step": 853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371683670258569e-05,
+      "loss": 0.8784,
+      "step": 854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702830909788584e-05,
+      "loss": 0.832,
+      "step": 855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3688816694220278e-05,
+      "loss": 0.8677,
+      "step": 856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674794087758685e-05,
+      "loss": 0.7896,
+      "step": 857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3660763122300806e-05,
+      "loss": 0.9225,
+      "step": 858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3646723829762663e-05,
+      "loss": 0.8672,
+      "step": 859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3632676242079208e-05,
+      "loss": 0.8587,
+      "step": 860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361862039120427e-05,
+      "loss": 0.8558,
+      "step": 861
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604556309110473e-05,
+      "loss": 0.9334,
+      "step": 862
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359048402778916e-05,
+      "loss": 0.8402,
+      "step": 863
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576403579250333e-05,
+      "loss": 0.9429,
+      "step": 864
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3562314995522563e-05,
+      "loss": 0.8501,
+      "step": 865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548218308652927e-05,
+      "loss": 0.9584,
+      "step": 866
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534113550706941e-05,
+      "loss": 0.8615,
+      "step": 867
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520000753768478e-05,
+      "loss": 0.809,
+      "step": 868
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3505879949939693e-05,
+      "loss": 0.8131,
+      "step": 869
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3491751171340955e-05,
+      "loss": 0.8055,
+      "step": 870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.347761445011077e-05,
+      "loss": 0.9068,
+      "step": 871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463469818405725e-05,
+      "loss": 0.8774,
+      "step": 872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3449317308400385e-05,
+      "loss": 0.8686,
+      "step": 873
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3435156952287243e-05,
+      "loss": 0.8621,
+      "step": 874
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3420988782276636e-05,
+      "loss": 0.902,
+      "step": 875
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406812830596682e-05,
+      "loss": 0.872,
+      "step": 876
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392629129493191e-05,
+      "loss": 0.8972,
+      "step": 877
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3378437711229608e-05,
+      "loss": 0.9022,
+      "step": 878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364238608086929e-05,
+      "loss": 0.8502,
+      "step": 879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350031852363624e-05,
+      "loss": 0.8394,
+      "step": 880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335817476375587e-05,
+      "loss": 0.8753,
+      "step": 881
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3321595512456028e-05,
+      "loss": 0.9206,
+      "step": 882
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307365992955428e-05,
+      "loss": 0.7924,
+      "step": 883
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3293128950241445e-05,
+      "loss": 0.913,
+      "step": 884
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3278884416698863e-05,
+      "loss": 0.9098,
+      "step": 885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264632424729499e-05,
+      "loss": 0.8652,
+      "step": 886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250373006752129e-05,
+      "loss": 0.905,
+      "step": 887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236106195202427e-05,
+      "loss": 0.8808,
+      "step": 888
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322183202253289e-05,
+      "loss": 0.8609,
+      "step": 889
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207550521212751e-05,
+      "loss": 0.8274,
+      "step": 890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193261723727914e-05,
+      "loss": 0.8378,
+      "step": 891
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3178965662580887e-05,
+      "loss": 0.8155,
+      "step": 892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316466237029069e-05,
+      "loss": 0.8696,
+      "step": 893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150351879392802e-05,
+      "loss": 0.8457,
+      "step": 894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136034222439067e-05,
+      "loss": 0.9545,
+      "step": 895
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121709431997638e-05,
+      "loss": 0.8966,
+      "step": 896
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107377540652883e-05,
+      "loss": 0.9064,
+      "step": 897
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093038581005342e-05,
+      "loss": 0.9158,
+      "step": 898
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3078692585671608e-05,
+      "loss": 0.7427,
+      "step": 899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30643395872843e-05,
+      "loss": 0.842,
+      "step": 900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049979618491953e-05,
+      "loss": 0.777,
+      "step": 901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035612711958962e-05,
+      "loss": 0.8333,
+      "step": 902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021238900365507e-05,
+      "loss": 0.7957,
+      "step": 903
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006858216407468e-05,
+      "loss": 0.8057,
+      "step": 904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299247069279636e-05,
+      "loss": 0.8589,
+      "step": 905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2978076362259256e-05,
+      "loss": 0.7948,
+      "step": 906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963675257538717e-05,
+      "loss": 0.7813,
+      "step": 907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2949267411392708e-05,
+      "loss": 0.8126,
+      "step": 908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934852856594524e-05,
+      "loss": 0.9242,
+      "step": 909
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920431625932734e-05,
+      "loss": 0.8683,
+      "step": 910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906003752211083e-05,
+      "loss": 0.807,
+      "step": 911
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891569268248425e-05,
+      "loss": 0.8524,
+      "step": 912
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877128206878655e-05,
+      "loss": 0.8659,
+      "step": 913
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2862680600950628e-05,
+      "loss": 0.8665,
+      "step": 914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848226483328085e-05,
+      "loss": 0.9283,
+      "step": 915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2833765886889584e-05,
+      "loss": 0.8938,
+      "step": 916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819298844528412e-05,
+      "loss": 0.9273,
+      "step": 917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280482538915252e-05,
+      "loss": 0.8961,
+      "step": 918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790345553684458e-05,
+      "loss": 0.8329,
+      "step": 919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277585937106127e-05,
+      "loss": 0.862,
+      "step": 920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2761366874234455e-05,
+      "loss": 0.8787,
+      "step": 921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2746868096169865e-05,
+      "loss": 0.8863,
+      "step": 922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732363069847642e-05,
+      "loss": 0.8825,
+      "step": 923
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717851828262141e-05,
+      "loss": 0.7991,
+      "step": 924
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703334404421856e-05,
+      "loss": 0.9907,
+      "step": 925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2688810831349341e-05,
+      "loss": 0.8619,
+      "step": 926
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267428114208114e-05,
+      "loss": 0.8511,
+      "step": 927
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2659745369667709e-05,
+      "loss": 0.8463,
+      "step": 928
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2645203547173338e-05,
+      "loss": 0.8928,
+      "step": 929
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263065570767609e-05,
+      "loss": 0.8789,
+      "step": 930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616101884267693e-05,
+      "loss": 0.8942,
+      "step": 931
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601542110053511e-05,
+      "loss": 0.829,
+      "step": 932
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258697641815243e-05,
+      "loss": 0.8311,
+      "step": 933
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.25724048416968e-05,
+      "loss": 0.9047,
+      "step": 934
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2557827413832355e-05,
+      "loss": 0.8134,
+      "step": 935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543244167718142e-05,
+      "loss": 0.883,
+      "step": 936
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528655136526447e-05,
+      "loss": 0.9176,
+      "step": 937
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514060353442702e-05,
+      "loss": 0.8332,
+      "step": 938
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499459851665432e-05,
+      "loss": 0.8568,
+      "step": 939
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2484853664406169e-05,
+      "loss": 0.8691,
+      "step": 940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247024182488938e-05,
+      "loss": 0.8411,
+      "step": 941
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2455624366352384e-05,
+      "loss": 0.8416,
+      "step": 942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441001322045285e-05,
+      "loss": 0.8551,
+      "step": 943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2426372725230889e-05,
+      "loss": 0.8329,
+      "step": 944
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411738609184638e-05,
+      "loss": 0.9153,
+      "step": 945
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397099007194524e-05,
+      "loss": 0.8764,
+      "step": 946
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2382453952561022e-05,
+      "loss": 0.813,
+      "step": 947
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2367803478597002e-05,
+      "loss": 0.8734,
+      "step": 948
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2353147618627673e-05,
+      "loss": 0.8858,
+      "step": 949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338486405990482e-05,
+      "loss": 0.8797,
+      "step": 950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323819874035064e-05,
+      "loss": 0.8251,
+      "step": 951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309148056123146e-05,
+      "loss": 0.8663,
+      "step": 952
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2294470985628483e-05,
+      "loss": 0.864,
+      "step": 953
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279788695936779e-05,
+      "loss": 0.8838,
+      "step": 954
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265101220445598e-05,
+      "loss": 0.8661,
+      "step": 955
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225040859256432e-05,
+      "loss": 0.956,
+      "step": 956
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2235710845714027e-05,
+      "loss": 0.8908,
+      "step": 957
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221008013327456e-05,
+      "loss": 0.8516,
+      "step": 958
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206300128848907e-05,
+      "loss": 0.8885,
+      "step": 959
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191587225734173e-05,
+      "loss": 0.9068,
+      "step": 960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2176869337450466e-05,
+      "loss": 0.8755,
+      "step": 961
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162146497476332e-05,
+      "loss": 0.892,
+      "step": 962
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147418739301582e-05,
+      "loss": 0.7795,
+      "step": 963
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2132686096427222e-05,
+      "loss": 0.823,
+      "step": 964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117948602365357e-05,
+      "loss": 0.8897,
+      "step": 965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103206290639137e-05,
+      "loss": 0.9289,
+      "step": 966
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088459194782662e-05,
+      "loss": 0.8261,
+      "step": 967
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073707348340922e-05,
+      "loss": 0.8258,
+      "step": 968
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205895078486971e-05,
+      "loss": 0.8952,
+      "step": 969
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044189537935546e-05,
+      "loss": 0.8629,
+      "step": 970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2029423641115609e-05,
+      "loss": 0.8786,
+      "step": 971
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014653127997653e-05,
+      "loss": 0.8536,
+      "step": 972
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1999878032179928e-05,
+      "loss": 0.8006,
+      "step": 973
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1985098387271115e-05,
+      "loss": 0.8544,
+      "step": 974
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970314226890238e-05,
+      "loss": 0.9043,
+      "step": 975
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1955525584666594e-05,
+      "loss": 0.8769,
+      "step": 976
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1940732494239676e-05,
+      "loss": 0.8914,
+      "step": 977
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1925934989259092e-05,
+      "loss": 0.8359,
+      "step": 978
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1911133103384493e-05,
+      "loss": 0.8901,
+      "step": 979
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1896326870285504e-05,
+      "loss": 0.8344,
+      "step": 980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881516323641621e-05,
+      "loss": 0.8418,
+      "step": 981
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1866701497142164e-05,
+      "loss": 0.8585,
+      "step": 982
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851882424486187e-05,
+      "loss": 0.9042,
+      "step": 983
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837059139382404e-05,
+      "loss": 0.8722,
+      "step": 984
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1822231675549101e-05,
+      "loss": 0.8987,
+      "step": 985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807400066714083e-05,
+      "loss": 0.8548,
+      "step": 986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792564346614572e-05,
+      "loss": 0.9085,
+      "step": 987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777724548997154e-05,
+      "loss": 0.8791,
+      "step": 988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176288070761768e-05,
+      "loss": 0.7954,
+      "step": 989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748032856241197e-05,
+      "loss": 0.8582,
+      "step": 990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733181028641885e-05,
+      "loss": 0.8487,
+      "step": 991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1718325258602963e-05,
+      "loss": 0.8326,
+      "step": 992
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703465579916611e-05,
+      "loss": 0.933,
+      "step": 993
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688602026383915e-05,
+      "loss": 0.8065,
+      "step": 994
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167373463181476e-05,
+      "loss": 0.8861,
+      "step": 995
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1658863430027775e-05,
+      "loss": 0.9331,
+      "step": 996
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643988454850255e-05,
+      "loss": 0.8435,
+      "step": 997
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629109740118063e-05,
+      "loss": 0.8088,
+      "step": 998
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1614227319675584e-05,
+      "loss": 0.9031,
+      "step": 999
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.159934122737562e-05,
+      "loss": 0.8964,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584451497079333e-05,
+      "loss": 0.8656,
+      "step": 1001
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1569558162656164e-05,
+      "loss": 0.8668,
+      "step": 1002
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155466125798374e-05,
+      "loss": 0.8452,
+      "step": 1003
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539760816947814e-05,
+      "loss": 0.8585,
+      "step": 1004
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152485687344219e-05,
+      "loss": 0.9309,
+      "step": 1005
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509949461368628e-05,
+      "loss": 0.8722,
+      "step": 1006
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1495038614636791e-05,
+      "loss": 0.8686,
+      "step": 1007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1480124367164144e-05,
+      "loss": 0.9718,
+      "step": 1008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465206752875889e-05,
+      "loss": 0.9608,
+      "step": 1009
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145028580570489e-05,
+      "loss": 0.853,
+      "step": 1010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435361559591594e-05,
+      "loss": 0.8634,
+      "step": 1011
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420434048483942e-05,
+      "loss": 0.9267,
+      "step": 1012
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405503306337313e-05,
+      "loss": 0.8512,
+      "step": 1013
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1390569367114433e-05,
+      "loss": 0.9281,
+      "step": 1014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1375632264785295e-05,
+      "loss": 0.9114,
+      "step": 1015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136069203332709e-05,
+      "loss": 0.7929,
+      "step": 1016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1345748706724128e-05,
+      "loss": 0.885,
+      "step": 1017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1330802318967762e-05,
+      "loss": 0.8147,
+      "step": 1018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315852904056303e-05,
+      "loss": 0.8718,
+      "step": 1019
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1300900495994949e-05,
+      "loss": 0.8223,
+      "step": 1020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.128594512879571e-05,
+      "loss": 0.8715,
+      "step": 1021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127098683647732e-05,
+      "loss": 0.9205,
+      "step": 1022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1256025653065174e-05,
+      "loss": 0.8265,
+      "step": 1023
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241061612591242e-05,
+      "loss": 0.7664,
+      "step": 1024
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226094749093988e-05,
+      "loss": 0.9004,
+      "step": 1025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211125096618302e-05,
+      "loss": 0.8588,
+      "step": 1026
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119615268921542e-05,
+      "loss": 0.873,
+      "step": 1027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1181177560942837e-05,
+      "loss": 0.8329,
+      "step": 1028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1166199745864246e-05,
+      "loss": 0.9059,
+      "step": 1029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115121927804944e-05,
+      "loss": 0.8643,
+      "step": 1030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136236191574262e-05,
+      "loss": 0.8863,
+      "step": 1031
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1121250520520499e-05,
+      "loss": 0.8305,
+      "step": 1032
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110626229897582e-05,
+      "loss": 0.8117,
+      "step": 1033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1091271561033696e-05,
+      "loss": 0.8782,
+      "step": 1034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.107627834079333e-05,
+      "loss": 0.856,
+      "step": 1035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061282672359552e-05,
+      "loss": 0.9187,
+      "step": 1036
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1046284589842785e-05,
+      "loss": 0.9023,
+      "step": 1037
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031284127358924e-05,
+      "loss": 0.877,
+      "step": 1038
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016281319029287e-05,
+      "loss": 0.796,
+      "step": 1039
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001276198980528e-05,
+      "loss": 0.914,
+      "step": 1040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986268801344555e-05,
+      "loss": 0.8536,
+      "step": 1041
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.097125916025846e-05,
+      "loss": 0.8817,
+      "step": 1042
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0956247309864434e-05,
+      "loss": 0.8602,
+      "step": 1043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0941233284309705e-05,
+      "loss": 0.8931,
+      "step": 1044
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0926217117746433e-05,
+      "loss": 0.986,
+      "step": 1045
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0911198844331659e-05,
+      "loss": 0.8645,
+      "step": 1046
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089617849822721e-05,
+      "loss": 0.883,
+      "step": 1047
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881156113599639e-05,
+      "loss": 0.8275,
+      "step": 1048
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866131724620117e-05,
+      "loss": 0.819,
+      "step": 1049
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0851105365464393e-05,
+      "loss": 0.9415,
+      "step": 1050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836077070312688e-05,
+      "loss": 0.8883,
+      "step": 1051
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0821046873349628e-05,
+      "loss": 0.9572,
+      "step": 1052
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0806014808764164e-05,
+      "loss": 0.8338,
+      "step": 1053
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.07909809107495e-05,
+      "loss": 0.882,
+      "step": 1054
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0775945213503004e-05,
+      "loss": 0.9092,
+      "step": 1055
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0760907751226141e-05,
+      "loss": 0.804,
+      "step": 1056
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0745868558124392e-05,
+      "loss": 0.8374,
+      "step": 1057
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730827668407171e-05,
+      "loss": 0.9351,
+      "step": 1058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0715785116287753e-05,
+      "loss": 0.8707,
+      "step": 1059
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700740935983194e-05,
+      "loss": 0.8461,
+      "step": 1060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685695161714256e-05,
+      "loss": 0.8286,
+      "step": 1061
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0670647827705323e-05,
+      "loss": 0.8877,
+      "step": 1062
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655598968184329e-05,
+      "loss": 0.8728,
+      "step": 1063
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.064054861738268e-05,
+      "loss": 0.87,
+      "step": 1064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0625496809535169e-05,
+      "loss": 0.9321,
+      "step": 1065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0610443578879913e-05,
+      "loss": 0.9134,
+      "step": 1066
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059538895965825e-05,
+      "loss": 0.824,
+      "step": 1067
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058033298611469e-05,
+      "loss": 0.842,
+      "step": 1068
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056527569249682e-05,
+      "loss": 0.9118,
+      "step": 1069
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0550217113055228e-05,
+      "loss": 0.842,
+      "step": 1070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0535157282043426e-05,
+      "loss": 0.8774,
+      "step": 1071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520096233717778e-05,
+      "loss": 0.9194,
+      "step": 1072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050503400233741e-05,
+      "loss": 0.8288,
+      "step": 1073
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489970622164146e-05,
+      "loss": 0.8307,
+      "step": 1074
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0474906127462416e-05,
+      "loss": 0.8888,
+      "step": 1075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0459840552499192e-05,
+      "loss": 0.8824,
+      "step": 1076
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444773931543898e-05,
+      "loss": 0.8971,
+      "step": 1077
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0429706298868343e-05,
+      "loss": 0.8819,
+      "step": 1078
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041463768874663e-05,
+      "loss": 0.8525,
+      "step": 1079
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0399568135455092e-05,
+      "loss": 0.8804,
+      "step": 1080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0384497673272202e-05,
+      "loss": 0.8205,
+      "step": 1081
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0369426336478508e-05,
+      "loss": 0.8546,
+      "step": 1082
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354354159356541e-05,
+      "loss": 0.8789,
+      "step": 1083
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339281176190743e-05,
+      "loss": 0.7981,
+      "step": 1084
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0324207421267392e-05,
+      "loss": 0.814,
+      "step": 1085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.030913292887452e-05,
+      "loss": 0.9397,
+      "step": 1086
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0294057733301841e-05,
+      "loss": 0.8631,
+      "step": 1087
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0278981868840663e-05,
+      "loss": 0.9086,
+      "step": 1088
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026390536978382e-05,
+      "loss": 0.8188,
+      "step": 1089
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0248828270425581e-05,
+      "loss": 0.8997,
+      "step": 1090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0233750605061594e-05,
+      "loss": 0.8182,
+      "step": 1091
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218672407988784e-05,
+      "loss": 0.8617,
+      "step": 1092
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0203593713505287e-05,
+      "loss": 0.8381,
+      "step": 1093
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188514555910374e-05,
+      "loss": 0.8852,
+      "step": 1094
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173434969504368e-05,
+      "loss": 0.9444,
+      "step": 1095
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158354988588568e-05,
+      "loss": 0.8405,
+      "step": 1096
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143274647465165e-05,
+      "loss": 0.8646,
+      "step": 1097
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128193980437175e-05,
+      "loss": 0.8423,
+      "step": 1098
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0113113021808356e-05,
+      "loss": 0.8758,
+      "step": 1099
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0098031805883128e-05,
+      "loss": 0.9255,
+      "step": 1100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0082950366966494e-05,
+      "loss": 0.8501,
+      "step": 1101
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0067868739363963e-05,
+      "loss": 0.8826,
+      "step": 1102
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0052786957381479e-05,
+      "loss": 0.8242,
+      "step": 1103
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0037705055325335e-05,
+      "loss": 0.7993,
+      "step": 1104
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022623067502096e-05,
+      "loss": 0.8892,
+      "step": 1105
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007541028218519e-05,
+      "loss": 0.9225,
+      "step": 1106
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.992458971781484e-06,
+      "loss": 0.9377,
+      "step": 1107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977376932497906e-06,
+      "loss": 0.8422,
+      "step": 1108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.962294944674667e-06,
+      "loss": 0.9221,
+      "step": 1109
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.947213042618523e-06,
+      "loss": 0.9194,
+      "step": 1110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.932131260636038e-06,
+      "loss": 0.8007,
+      "step": 1111
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917049633033511e-06,
+      "loss": 0.8146,
+      "step": 1112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901968194116875e-06,
+      "loss": 0.8568,
+      "step": 1113
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.886886978191644e-06,
+      "loss": 0.8056,
+      "step": 1114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.871806019562828e-06,
+      "loss": 0.8578,
+      "step": 1115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856725352534838e-06,
+      "loss": 0.8546,
+      "step": 1116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841645011411438e-06,
+      "loss": 0.8534,
+      "step": 1117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826565030495635e-06,
+      "loss": 0.8349,
+      "step": 1118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811485444089626e-06,
+      "loss": 0.816,
+      "step": 1119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.796406286494715e-06,
+      "loss": 0.8548,
+      "step": 1120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78132759201122e-06,
+      "loss": 0.8757,
+      "step": 1121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76624939493841e-06,
+      "loss": 0.8128,
+      "step": 1122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75117172957442e-06,
+      "loss": 0.806,
+      "step": 1123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.736094630216182e-06,
+      "loss": 0.892,
+      "step": 1124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.721018131159342e-06,
+      "loss": 0.9101,
+      "step": 1125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.705942266698162e-06,
+      "loss": 0.9487,
+      "step": 1126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690867071125482e-06,
+      "loss": 0.903,
+      "step": 1127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.675792578732613e-06,
+      "loss": 0.8535,
+      "step": 1128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.66071882380926e-06,
+      "loss": 0.8494,
+      "step": 1129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.645645840643465e-06,
+      "loss": 0.8087,
+      "step": 1130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.630573663521494e-06,
+      "loss": 0.8598,
+      "step": 1131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615502326727798e-06,
+      "loss": 0.8856,
+      "step": 1132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.600431864544913e-06,
+      "loss": 0.8823,
+      "step": 1133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.585362311253374e-06,
+      "loss": 0.8487,
+      "step": 1134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57029370113166e-06,
+      "loss": 0.8557,
+      "step": 1135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.555226068456106e-06,
+      "loss": 0.9581,
+      "step": 1136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.540159447500811e-06,
+      "loss": 0.8392,
+      "step": 1137
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52509387253759e-06,
+      "loss": 0.8386,
+      "step": 1138
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51002937783586e-06,
+      "loss": 0.9124,
+      "step": 1139
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494965997662592e-06,
+      "loss": 0.9015,
+      "step": 1140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479903766282227e-06,
+      "loss": 0.8449,
+      "step": 1141
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.464842717956576e-06,
+      "loss": 0.875,
+      "step": 1142
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.449782886944777e-06,
+      "loss": 0.8442,
+      "step": 1143
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434724307503182e-06,
+      "loss": 0.9113,
+      "step": 1144
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.419667013885312e-06,
+      "loss": 0.8238,
+      "step": 1145
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404611040341755e-06,
+      "loss": 0.8376,
+      "step": 1146
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389556421120092e-06,
+      "loss": 0.8134,
+      "step": 1147
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374503190464831e-06,
+      "loss": 0.8397,
+      "step": 1148
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359451382617323e-06,
+      "loss": 0.9778,
+      "step": 1149
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.344401031815675e-06,
+      "loss": 0.7893,
+      "step": 1150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.329352172294682e-06,
+      "loss": 0.8972,
+      "step": 1151
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314304838285747e-06,
+      "loss": 0.8443,
+      "step": 1152
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299259064016807e-06,
+      "loss": 0.8974,
+      "step": 1153
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28421488371225e-06,
+      "loss": 0.9403,
+      "step": 1154
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26917233159283e-06,
+      "loss": 0.8913,
+      "step": 1155
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254131441875606e-06,
+      "loss": 0.8716,
+      "step": 1156
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.23909224877386e-06,
+      "loss": 0.8959,
+      "step": 1157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.224054786496996e-06,
+      "loss": 0.849,
+      "step": 1158
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209019089250504e-06,
+      "loss": 0.8191,
+      "step": 1159
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.193985191235837e-06,
+      "loss": 0.846,
+      "step": 1160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178953126650373e-06,
+      "loss": 0.8792,
+      "step": 1161
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163922929687317e-06,
+      "loss": 0.9379,
+      "step": 1162
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148894634535608e-06,
+      "loss": 0.8598,
+      "step": 1163
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.133868275379886e-06,
+      "loss": 0.9306,
+      "step": 1164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118843886400365e-06,
+      "loss": 0.8881,
+      "step": 1165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103821501772789e-06,
+      "loss": 0.9151,
+      "step": 1166
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088801155668345e-06,
+      "loss": 0.9051,
+      "step": 1167
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07378288225357e-06,
+      "loss": 0.8664,
+      "step": 1168
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.058766715690296e-06,
+      "loss": 0.802,
+      "step": 1169
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043752690135567e-06,
+      "loss": 0.8957,
+      "step": 1170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.028740839741544e-06,
+      "loss": 0.9545,
+      "step": 1171
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.01373119865545e-06,
+      "loss": 0.8372,
+      "step": 1172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998723801019475e-06,
+      "loss": 0.9176,
+      "step": 1173
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983718680970713e-06,
+      "loss": 0.8157,
+      "step": 1174
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96871587264108e-06,
+      "loss": 0.8179,
+      "step": 1175
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.953715410157217e-06,
+      "loss": 0.864,
+      "step": 1176
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93871732764045e-06,
+      "loss": 0.851,
+      "step": 1177
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.923721659206674e-06,
+      "loss": 0.8463,
+      "step": 1178
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.908728438966305e-06,
+      "loss": 0.8808,
+      "step": 1179
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.893737701024185e-06,
+      "loss": 0.8413,
+      "step": 1180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878749479479504e-06,
+      "loss": 0.9245,
+      "step": 1181
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863763808425738e-06,
+      "loss": 0.8232,
+      "step": 1182
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848780721950563e-06,
+      "loss": 0.8894,
+      "step": 1183
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.833800254135758e-06,
+      "loss": 0.9169,
+      "step": 1184
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818822439057166e-06,
+      "loss": 0.8293,
+      "step": 1185
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.803847310784583e-06,
+      "loss": 0.848,
+      "step": 1186
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.788874903381698e-06,
+      "loss": 0.9037,
+      "step": 1187
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.773905250906015e-06,
+      "loss": 0.8824,
+      "step": 1188
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.758938387408761e-06,
+      "loss": 0.9233,
+      "step": 1189
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743974346934827e-06,
+      "loss": 0.9699,
+      "step": 1190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.729013163522683e-06,
+      "loss": 0.8693,
+      "step": 1191
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.714054871204294e-06,
+      "loss": 0.8735,
+      "step": 1192
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.699099504005054e-06,
+      "loss": 0.8191,
+      "step": 1193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684147095943699e-06,
+      "loss": 0.8708,
+      "step": 1194
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66919768103224e-06,
+      "loss": 0.8231,
+      "step": 1195
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.654251293275875e-06,
+      "loss": 0.8964,
+      "step": 1196
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639307966672913e-06,
+      "loss": 0.7958,
+      "step": 1197
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624367735214711e-06,
+      "loss": 0.8975,
+      "step": 1198
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60943063288557e-06,
+      "loss": 0.8934,
+      "step": 1199
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.594496693662688e-06,
+      "loss": 0.8553,
+      "step": 1200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.579565951516063e-06,
+      "loss": 0.9041,
+      "step": 1201
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564638440408411e-06,
+      "loss": 0.8352,
+      "step": 1202
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549714194295115e-06,
+      "loss": 0.8514,
+      "step": 1203
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534793247124115e-06,
+      "loss": 0.853,
+      "step": 1204
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.519875632835858e-06,
+      "loss": 0.8732,
+      "step": 1205
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504961385363212e-06,
+      "loss": 0.8377,
+      "step": 1206
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.490050538631375e-06,
+      "loss": 0.8896,
+      "step": 1207
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.475143126557814e-06,
+      "loss": 0.9242,
+      "step": 1208
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46023918305219e-06,
+      "loss": 0.8478,
+      "step": 1209
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.445338742016263e-06,
+      "loss": 0.8388,
+      "step": 1210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.430441837343841e-06,
+      "loss": 0.8257,
+      "step": 1211
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.415548502920668e-06,
+      "loss": 0.8666,
+      "step": 1212
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.400658772624382e-06,
+      "loss": 0.8762,
+      "step": 1213
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.38577268032442e-06,
+      "loss": 0.9543,
+      "step": 1214
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.370890259881938e-06,
+      "loss": 0.8631,
+      "step": 1215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356011545149752e-06,
+      "loss": 0.801,
+      "step": 1216
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341136569972227e-06,
+      "loss": 0.8904,
+      "step": 1217
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326265368185242e-06,
+      "loss": 0.8823,
+      "step": 1218
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31139797361609e-06,
+      "loss": 0.9037,
+      "step": 1219
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.29653442008339e-06,
+      "loss": 0.8477,
+      "step": 1220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.28167474139704e-06,
+      "loss": 0.8924,
+      "step": 1221
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.266818971358118e-06,
+      "loss": 0.8828,
+      "step": 1222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.251967143758806e-06,
+      "loss": 0.9561,
+      "step": 1223
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237119292382327e-06,
+      "loss": 0.8416,
+      "step": 1224
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22227545100285e-06,
+      "loss": 0.8815,
+      "step": 1225
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.207435653385428e-06,
+      "loss": 0.7973,
+      "step": 1226
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.192599933285922e-06,
+      "loss": 0.916,
+      "step": 1227
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.1777683244509e-06,
+      "loss": 0.8082,
+      "step": 1228
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.162940860617598e-06,
+      "loss": 0.8195,
+      "step": 1229
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148117575513816e-06,
+      "loss": 0.8559,
+      "step": 1230
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.133298502857836e-06,
+      "loss": 0.8052,
+      "step": 1231
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118483676358384e-06,
+      "loss": 0.8569,
+      "step": 1232
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1036731297145e-06,
+      "loss": 0.8101,
+      "step": 1233
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.088866896615507e-06,
+      "loss": 0.8717,
+      "step": 1234
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07406501074091e-06,
+      "loss": 0.8799,
+      "step": 1235
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.05926750576033e-06,
+      "loss": 0.8907,
+      "step": 1236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.044474415333412e-06,
+      "loss": 0.915,
+      "step": 1237
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029685773109766e-06,
+      "loss": 0.8735,
+      "step": 1238
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.014901612728887e-06,
+      "loss": 0.8799,
+      "step": 1239
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.000121967820077e-06,
+      "loss": 0.8005,
+      "step": 1240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985346872002349e-06,
+      "loss": 0.8986,
+      "step": 1241
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.970576358884391e-06,
+      "loss": 0.8006,
+      "step": 1242
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.955810462064457e-06,
+      "loss": 0.8613,
+      "step": 1243
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941049215130292e-06,
+      "loss": 0.8695,
+      "step": 1244
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926292651659083e-06,
+      "loss": 0.7912,
+      "step": 1245
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.911540805217341e-06,
+      "loss": 0.8759,
+      "step": 1246
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896793709360865e-06,
+      "loss": 0.9341,
+      "step": 1247
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882051397634646e-06,
+      "loss": 0.8267,
+      "step": 1248
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86731390357278e-06,
+      "loss": 0.9242,
+      "step": 1249
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.85258126069842e-06,
+      "loss": 0.9165,
+      "step": 1250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837853502523671e-06,
+      "loss": 0.8172,
+      "step": 1251
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823130662549537e-06,
+      "loss": 0.7805,
+      "step": 1252
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80841277426583e-06,
+      "loss": 0.8644,
+      "step": 1253
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.793699871151096e-06,
+      "loss": 0.7943,
+      "step": 1254
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778991986672546e-06,
+      "loss": 0.8663,
+      "step": 1255
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764289154285977e-06,
+      "loss": 0.9321,
+      "step": 1256
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.749591407435683e-06,
+      "loss": 0.8254,
+      "step": 1257
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.734898779554405e-06,
+      "loss": 0.8495,
+      "step": 1258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720211304063225e-06,
+      "loss": 0.7735,
+      "step": 1259
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.705529014371517e-06,
+      "loss": 0.8281,
+      "step": 1260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690851943876857e-06,
+      "loss": 0.8624,
+      "step": 1261
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676180125964937e-06,
+      "loss": 0.8576,
+      "step": 1262
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661513594009522e-06,
+      "loss": 0.8365,
+      "step": 1263
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.646852381372332e-06,
+      "loss": 0.8533,
+      "step": 1264
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.632196521403001e-06,
+      "loss": 0.888,
+      "step": 1265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.617546047438983e-06,
+      "loss": 0.8601,
+      "step": 1266
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602900992805477e-06,
+      "loss": 0.8137,
+      "step": 1267
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588261390815364e-06,
+      "loss": 0.752,
+      "step": 1268
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5736272747691155e-06,
+      "loss": 0.8993,
+      "step": 1269
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.558998677954719e-06,
+      "loss": 0.907,
+      "step": 1270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54437563364762e-06,
+      "loss": 0.8303,
+      "step": 1271
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.529758175110623e-06,
+      "loss": 0.9041,
+      "step": 1272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.515146335593831e-06,
+      "loss": 0.852,
+      "step": 1273
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500540148334573e-06,
+      "loss": 0.8235,
+      "step": 1274
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.485939646557302e-06,
+      "loss": 0.9119,
+      "step": 1275
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471344863473557e-06,
+      "loss": 0.8792,
+      "step": 1276
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456755832281859e-06,
+      "loss": 0.7845,
+      "step": 1277
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442172586167646e-06,
+      "loss": 0.8212,
+      "step": 1278
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.427595158303205e-06,
+      "loss": 0.89,
+      "step": 1279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4130235818475716e-06,
+      "loss": 0.9237,
+      "step": 1280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.398457889946491e-06,
+      "loss": 0.7663,
+      "step": 1281
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3838981157323106e-06,
+      "loss": 0.8394,
+      "step": 1282
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369344292323915e-06,
+      "loss": 0.9196,
+      "step": 1283
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354796452826665e-06,
+      "loss": 0.8249,
+      "step": 1284
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.340254630332294e-06,
+      "loss": 0.8618,
+      "step": 1285
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.325718857918861e-06,
+      "loss": 0.8295,
+      "step": 1286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.311189168650663e-06,
+      "loss": 0.8183,
+      "step": 1287
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296665595578147e-06,
+      "loss": 0.7794,
+      "step": 1288
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.282148171737864e-06,
+      "loss": 0.8062,
+      "step": 1289
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267636930152362e-06,
+      "loss": 0.924,
+      "step": 1290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2531319038301375e-06,
+      "loss": 0.834,
+      "step": 1291
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.238633125765547e-06,
+      "loss": 0.8682,
+      "step": 1292
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.224140628938733e-06,
+      "loss": 0.8873,
+      "step": 1293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209654446315546e-06,
+      "loss": 0.84,
+      "step": 1294
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.195174610847482e-06,
+      "loss": 0.9044,
+      "step": 1295
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180701155471592e-06,
+      "loss": 0.8749,
+      "step": 1296
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166234113110422e-06,
+      "loss": 0.8453,
+      "step": 1297
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151773516671917e-06,
+      "loss": 0.9285,
+      "step": 1298
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1373193990493736e-06,
+      "loss": 0.8518,
+      "step": 1299
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.12287179312135e-06,
+      "loss": 0.8322,
+      "step": 1300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1084307317515785e-06,
+      "loss": 0.7956,
+      "step": 1301
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.093996247788919e-06,
+      "loss": 0.9011,
+      "step": 1302
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079568374067269e-06,
+      "loss": 0.8109,
+      "step": 1303
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065147143405477e-06,
+      "loss": 0.873,
+      "step": 1304
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050732588607298e-06,
+      "loss": 0.8081,
+      "step": 1305
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0363247424612866e-06,
+      "loss": 0.8634,
+      "step": 1306
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.021923637740745e-06,
+      "loss": 0.8678,
+      "step": 1307
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0075293072036444e-06,
+      "loss": 0.8572,
+      "step": 1308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993141783592534e-06,
+      "loss": 0.9368,
+      "step": 1309
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978761099634497e-06,
+      "loss": 0.8942,
+      "step": 1310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964387288041039e-06,
+      "loss": 0.865,
+      "step": 1311
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.950020381508048e-06,
+      "loss": 0.8522,
+      "step": 1312
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.935660412715703e-06,
+      "loss": 0.9353,
+      "step": 1313
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921307414328393e-06,
+      "loss": 0.8812,
+      "step": 1314
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9069614189946595e-06,
+      "loss": 0.9038,
+      "step": 1315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.892622459347118e-06,
+      "loss": 0.8982,
+      "step": 1316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878290568002366e-06,
+      "loss": 0.8896,
+      "step": 1317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863965777560937e-06,
+      "loss": 0.8789,
+      "step": 1318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849648120607202e-06,
+      "loss": 0.8992,
+      "step": 1319
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.835337629709311e-06,
+      "loss": 0.8166,
+      "step": 1320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821034337419117e-06,
+      "loss": 0.8793,
+      "step": 1321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.806738276272089e-06,
+      "loss": 0.887,
+      "step": 1322
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.792449478787254e-06,
+      "loss": 0.8764,
+      "step": 1323
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778167977467112e-06,
+      "loss": 0.9078,
+      "step": 1324
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763893804797574e-06,
+      "loss": 0.8724,
+      "step": 1325
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749626993247876e-06,
+      "loss": 0.8296,
+      "step": 1326
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735367575270505e-06,
+      "loss": 0.8192,
+      "step": 1327
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7211155833011365e-06,
+      "loss": 0.8894,
+      "step": 1328
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.706871049758557e-06,
+      "loss": 0.8137,
+      "step": 1329
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692634007044576e-06,
+      "loss": 0.8575,
+      "step": 1330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.678404487543977e-06,
+      "loss": 0.8756,
+      "step": 1331
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6641825236244165e-06,
+      "loss": 0.8874,
+      "step": 1332
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6499681476363755e-06,
+      "loss": 0.7312,
+      "step": 1333
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6357613919130756e-06,
+      "loss": 0.8994,
+      "step": 1334
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621562288770393e-06,
+      "loss": 0.8871,
+      "step": 1335
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.607370870506812e-06,
+      "loss": 0.9007,
+      "step": 1336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.593187169403321e-06,
+      "loss": 0.8452,
+      "step": 1337
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5790112177233665e-06,
+      "loss": 0.7653,
+      "step": 1338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.564843047712762e-06,
+      "loss": 0.8619,
+      "step": 1339
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5506826915996195e-06,
+      "loss": 0.8752,
+      "step": 1340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536530181594276e-06,
+      "loss": 0.8493,
+      "step": 1341
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522385549889232e-06,
+      "loss": 0.8408,
+      "step": 1342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5082488286590495e-06,
+      "loss": 0.8361,
+      "step": 1343
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.494120050060312e-06,
+      "loss": 0.8988,
+      "step": 1344
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479999246231523e-06,
+      "loss": 0.8483,
+      "step": 1345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465886449293057e-06,
+      "loss": 0.812,
+      "step": 1346
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451781691347076e-06,
+      "loss": 0.8965,
+      "step": 1347
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437685004477441e-06,
+      "loss": 0.8683,
+      "step": 1348
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4235964207496715e-06,
+      "loss": 0.8534,
+      "step": 1349
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.409515972210843e-06,
+      "loss": 0.8882,
+      "step": 1350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.39544369088953e-06,
+      "loss": 0.8821,
+      "step": 1351
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3813796087957325e-06,
+      "loss": 0.8779,
+      "step": 1352
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.367323757920794e-06,
+      "loss": 0.8037,
+      "step": 1353
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353276170237339e-06,
+      "loss": 0.8293,
+      "step": 1354
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.339236877699196e-06,
+      "loss": 0.8385,
+      "step": 1355
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3252059122413174e-06,
+      "loss": 0.8543,
+      "step": 1356
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3111833057797266e-06,
+      "loss": 0.8408,
+      "step": 1357
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.297169090211418e-06,
+      "loss": 0.9593,
+      "step": 1358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28316329741431e-06,
+      "loss": 0.8642,
+      "step": 1359
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.269165959247166e-06,
+      "loss": 0.8369,
+      "step": 1360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.255177107549499e-06,
+      "loss": 0.8829,
+      "step": 1361
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.24119677414154e-06,
+      "loss": 0.8821,
+      "step": 1362
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227224990824126e-06,
+      "loss": 0.8428,
+      "step": 1363
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213261789378656e-06,
+      "loss": 0.8961,
+      "step": 1364
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.199307201567007e-06,
+      "loss": 0.9534,
+      "step": 1365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.185361259131454e-06,
+      "loss": 0.7416,
+      "step": 1366
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1714239937946186e-06,
+      "loss": 0.8136,
+      "step": 1367
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.157495437259378e-06,
+      "loss": 0.892,
+      "step": 1368
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143575621208797e-06,
+      "loss": 0.8934,
+      "step": 1369
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129664577306069e-06,
+      "loss": 0.9142,
+      "step": 1370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.115762337194418e-06,
+      "loss": 0.9032,
+      "step": 1371
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1018689324970545e-06,
+      "loss": 0.8014,
+      "step": 1372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087984394817093e-06,
+      "loss": 0.7952,
+      "step": 1373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.074108755737463e-06,
+      "loss": 0.8896,
+      "step": 1374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0602420468208675e-06,
+      "loss": 0.886,
+      "step": 1375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046384299609695e-06,
+      "loss": 0.8737,
+      "step": 1376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.032535545625939e-06,
+      "loss": 0.8013,
+      "step": 1377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0186958163711425e-06,
+      "loss": 0.8038,
+      "step": 1378
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.004865143326324e-06,
+      "loss": 0.7449,
+      "step": 1379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9910435579518925e-06,
+      "loss": 0.8309,
+      "step": 1380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977231091687596e-06,
+      "loss": 0.7667,
+      "step": 1381
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.963427775952429e-06,
+      "loss": 0.8381,
+      "step": 1382
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.949633642144585e-06,
+      "loss": 0.8217,
+      "step": 1383
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9358487216413505e-06,
+      "loss": 0.8281,
+      "step": 1384
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922073045799077e-06,
+      "loss": 0.8996,
+      "step": 1385
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9083066459530765e-06,
+      "loss": 0.8479,
+      "step": 1386
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.894549553417559e-06,
+      "loss": 0.8805,
+      "step": 1387
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.880801799485565e-06,
+      "loss": 0.8423,
+      "step": 1388
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.867063415428902e-06,
+      "loss": 0.845,
+      "step": 1389
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.853334432498047e-06,
+      "loss": 0.8581,
+      "step": 1390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.839614881922109e-06,
+      "loss": 0.9121,
+      "step": 1391
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825904794908723e-06,
+      "loss": 0.8969,
+      "step": 1392
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.812204202644021e-06,
+      "loss": 0.8614,
+      "step": 1393
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7985131362925184e-06,
+      "loss": 0.8348,
+      "step": 1394
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784831626997067e-06,
+      "loss": 0.8754,
+      "step": 1395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7711597058787846e-06,
+      "loss": 0.8247,
+      "step": 1396
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.757497404036973e-06,
+      "loss": 0.8671,
+      "step": 1397
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.743844752549051e-06,
+      "loss": 0.8705,
+      "step": 1398
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7302017824705e-06,
+      "loss": 0.8406,
+      "step": 1399
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.716568524834758e-06,
+      "loss": 0.7667,
+      "step": 1400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.702945010653189e-06,
+      "loss": 0.789,
+      "step": 1401
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689331270914986e-06,
+      "loss": 0.9366,
+      "step": 1402
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675727336587104e-06,
+      "loss": 0.8864,
+      "step": 1403
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662133238614208e-06,
+      "loss": 0.8793,
+      "step": 1404
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648549007918565e-06,
+      "loss": 0.8797,
+      "step": 1405
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6349746754000215e-06,
+      "loss": 0.8187,
+      "step": 1406
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621410271935894e-06,
+      "loss": 0.8347,
+      "step": 1407
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.60785582838092e-06,
+      "loss": 0.8705,
+      "step": 1408
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.594311375567174e-06,
+      "loss": 0.9126,
+      "step": 1409
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.580776944304014e-06,
+      "loss": 0.827,
+      "step": 1410
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567252565377994e-06,
+      "loss": 0.8589,
+      "step": 1411
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.553738269552815e-06,
+      "loss": 0.8509,
+      "step": 1412
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5402340875692184e-06,
+      "loss": 0.8497,
+      "step": 1413
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526740050144967e-06,
+      "loss": 0.8232,
+      "step": 1414
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5132561879747314e-06,
+      "loss": 0.8533,
+      "step": 1415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4997825317300404e-06,
+      "loss": 0.8497,
+      "step": 1416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486319112059208e-06,
+      "loss": 0.876,
+      "step": 1417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.472865959587258e-06,
+      "loss": 0.8147,
+      "step": 1418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.459423104915873e-06,
+      "loss": 0.8146,
+      "step": 1419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.445990578623301e-06,
+      "loss": 0.8156,
+      "step": 1420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432568411264294e-06,
+      "loss": 0.8208,
+      "step": 1421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.419156633370049e-06,
+      "loss": 0.8915,
+      "step": 1422
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405755275448125e-06,
+      "loss": 0.7963,
+      "step": 1423
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392364367982375e-06,
+      "loss": 0.9692,
+      "step": 1424
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378983941432897e-06,
+      "loss": 0.8889,
+      "step": 1425
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365614026235924e-06,
+      "loss": 0.832,
+      "step": 1426
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3522546528037985e-06,
+      "loss": 0.8367,
+      "step": 1427
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.338905851524878e-06,
+      "loss": 0.8331,
+      "step": 1428
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.325567652763468e-06,
+      "loss": 0.8031,
+      "step": 1429
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.31224008685976e-06,
+      "loss": 0.8256,
+      "step": 1430
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298923184129754e-06,
+      "loss": 0.8388,
+      "step": 1431
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285616974865208e-06,
+      "loss": 0.8369,
+      "step": 1432
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.272321489333543e-06,
+      "loss": 0.8734,
+      "step": 1433
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.259036757777789e-06,
+      "loss": 0.7801,
+      "step": 1434
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.245762810416517e-06,
+      "loss": 0.6722,
+      "step": 1435
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.232499677443767e-06,
+      "loss": 0.5928,
+      "step": 1436
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.219247389028973e-06,
+      "loss": 0.6237,
+      "step": 1437
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20600597531692e-06,
+      "loss": 0.6309,
+      "step": 1438
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1927754664276265e-06,
+      "loss": 0.6686,
+      "step": 1439
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179555892456337e-06,
+      "loss": 0.6097,
+      "step": 1440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166347283473403e-06,
+      "loss": 0.5675,
+      "step": 1441
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1531496695242375e-06,
+      "loss": 0.6428,
+      "step": 1442
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.139963080629249e-06,
+      "loss": 0.5883,
+      "step": 1443
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.126787546783758e-06,
+      "loss": 0.6445,
+      "step": 1444
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113623097957952e-06,
+      "loss": 0.6252,
+      "step": 1445
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100469764096789e-06,
+      "loss": 0.6432,
+      "step": 1446
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.087327575119956e-06,
+      "loss": 0.6956,
+      "step": 1447
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074196560921776e-06,
+      "loss": 0.5548,
+      "step": 1448
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061076751371172e-06,
+      "loss": 0.6143,
+      "step": 1449
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.047968176311555e-06,
+      "loss": 0.6108,
+      "step": 1450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.034870865560807e-06,
+      "loss": 0.5721,
+      "step": 1451
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021784848911161e-06,
+      "loss": 0.6736,
+      "step": 1452
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008710156129184e-06,
+      "loss": 0.5748,
+      "step": 1453
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.995646816955671e-06,
+      "loss": 0.5915,
+      "step": 1454
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982594861105594e-06,
+      "loss": 0.6241,
+      "step": 1455
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9695543182680295e-06,
+      "loss": 0.6154,
+      "step": 1456
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956525218106097e-06,
+      "loss": 0.5598,
+      "step": 1457
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.943507590256883e-06,
+      "loss": 0.6015,
+      "step": 1458
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.930501464331385e-06,
+      "loss": 0.5638,
+      "step": 1459
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917506869914432e-06,
+      "loss": 0.5978,
+      "step": 1460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904523836564622e-06,
+      "loss": 0.5699,
+      "step": 1461
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891552393814257e-06,
+      "loss": 0.5989,
+      "step": 1462
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.87859257116927e-06,
+      "loss": 0.5648,
+      "step": 1463
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.865644398109177e-06,
+      "loss": 0.6132,
+      "step": 1464
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8527079040869674e-06,
+      "loss": 0.5491,
+      "step": 1465
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.839783118529095e-06,
+      "loss": 0.5401,
+      "step": 1466
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826870070835358e-06,
+      "loss": 0.6046,
+      "step": 1467
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813968790378866e-06,
+      "loss": 0.5798,
+      "step": 1468
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801079306505957e-06,
+      "loss": 0.6508,
+      "step": 1469
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788201648536137e-06,
+      "loss": 0.5828,
+      "step": 1470
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7753358457620105e-06,
+      "loss": 0.5699,
+      "step": 1471
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.762481927449222e-06,
+      "loss": 0.6081,
+      "step": 1472
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.749639922836373e-06,
+      "loss": 0.6003,
+      "step": 1473
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7368098611349715e-06,
+      "loss": 0.5853,
+      "step": 1474
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723991771529355e-06,
+      "loss": 0.5572,
+      "step": 1475
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711185683176628e-06,
+      "loss": 0.6312,
+      "step": 1476
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.698391625206608e-06,
+      "loss": 0.5657,
+      "step": 1477
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685609626721723e-06,
+      "loss": 0.623,
+      "step": 1478
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.672839716796995e-06,
+      "loss": 0.5699,
+      "step": 1479
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.660081924479934e-06,
+      "loss": 0.6357,
+      "step": 1480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647336278790488e-06,
+      "loss": 0.5201,
+      "step": 1481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.634602808720977e-06,
+      "loss": 0.6179,
+      "step": 1482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621881543236027e-06,
+      "loss": 0.6215,
+      "step": 1483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609172511272494e-06,
+      "loss": 0.5728,
+      "step": 1484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.596475741739426e-06,
+      "loss": 0.6094,
+      "step": 1485
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.583791263517949e-06,
+      "loss": 0.5571,
+      "step": 1486
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.571119105461258e-06,
+      "loss": 0.6018,
+      "step": 1487
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5584592963945075e-06,
+      "loss": 0.5924,
+      "step": 1488
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545811865114761e-06,
+      "loss": 0.6335,
+      "step": 1489
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.533176840390944e-06,
+      "loss": 0.5183,
+      "step": 1490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520554250963733e-06,
+      "loss": 0.6234,
+      "step": 1491
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5079441255455455e-06,
+      "loss": 0.6575,
+      "step": 1492
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495346492820431e-06,
+      "loss": 0.5458,
+      "step": 1493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.482761381444025e-06,
+      "loss": 0.6248,
+      "step": 1494
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4701888200434855e-06,
+      "loss": 0.5845,
+      "step": 1495
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457628837217417e-06,
+      "loss": 0.5405,
+      "step": 1496
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.445081461535813e-06,
+      "loss": 0.6076,
+      "step": 1497
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.432546721540002e-06,
+      "loss": 0.6205,
+      "step": 1498
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420024645742544e-06,
+      "loss": 0.7318,
+      "step": 1499
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.407515262627221e-06,
+      "loss": 0.5836,
+      "step": 1500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.395018600648925e-06,
+      "loss": 0.5871,
+      "step": 1501
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382534688233616e-06,
+      "loss": 0.5683,
+      "step": 1502
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.370063553778254e-06,
+      "loss": 0.5851,
+      "step": 1503
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.357605225650727e-06,
+      "loss": 0.5175,
+      "step": 1504
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345159732189809e-06,
+      "loss": 0.5629,
+      "step": 1505
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33272710170506e-06,
+      "loss": 0.535,
+      "step": 1506
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320307362476791e-06,
+      "loss": 0.6414,
+      "step": 1507
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.307900542755987e-06,
+      "loss": 0.552,
+      "step": 1508
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.295506670764245e-06,
+      "loss": 0.6079,
+      "step": 1509
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.283125774693709e-06,
+      "loss": 0.5609,
+      "step": 1510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2707578827070186e-06,
+      "loss": 0.6685,
+      "step": 1511
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258403022937207e-06,
+      "loss": 0.5232,
+      "step": 1512
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246061223487692e-06,
+      "loss": 0.6708,
+      "step": 1513
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233732512432166e-06,
+      "loss": 0.5504,
+      "step": 1514
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2214169178145545e-06,
+      "loss": 0.5605,
+      "step": 1515
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.209114467648946e-06,
+      "loss": 0.5665,
+      "step": 1516
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196825189919528e-06,
+      "loss": 0.6009,
+      "step": 1517
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1845491125805324e-06,
+      "loss": 0.552,
+      "step": 1518
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172286263556158e-06,
+      "loss": 0.5794,
+      "step": 1519
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.160036670740513e-06,
+      "loss": 0.5853,
+      "step": 1520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1478003619975484e-06,
+      "loss": 0.6261,
+      "step": 1521
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.135577365161017e-06,
+      "loss": 0.6306,
+      "step": 1522
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.123367708034361e-06,
+      "loss": 0.5868,
+      "step": 1523
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111171418390712e-06,
+      "loss": 0.6353,
+      "step": 1524
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.09898852397276e-06,
+      "loss": 0.6693,
+      "step": 1525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.086819052492757e-06,
+      "loss": 0.5605,
+      "step": 1526
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.074663031632403e-06,
+      "loss": 0.6312,
+      "step": 1527
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062520489042808e-06,
+      "loss": 0.6125,
+      "step": 1528
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.050391452344423e-06,
+      "loss": 0.5426,
+      "step": 1529
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.038275949126973e-06,
+      "loss": 0.6939,
+      "step": 1530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026174006949409e-06,
+      "loss": 0.6369,
+      "step": 1531
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.014085653339827e-06,
+      "loss": 0.6463,
+      "step": 1532
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.002010915795412e-06,
+      "loss": 0.6024,
+      "step": 1533
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.989949821782377e-06,
+      "loss": 0.6676,
+      "step": 1534
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.977902398735914e-06,
+      "loss": 0.6321,
+      "step": 1535
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.965868674060092e-06,
+      "loss": 0.6447,
+      "step": 1536
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.953848675127849e-06,
+      "loss": 0.6315,
+      "step": 1537
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.941842429280874e-06,
+      "loss": 0.5678,
+      "step": 1538
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929849963829595e-06,
+      "loss": 0.5989,
+      "step": 1539
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917871306053083e-06,
+      "loss": 0.6058,
+      "step": 1540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.905906483198999e-06,
+      "loss": 0.666,
+      "step": 1541
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89395552248354e-06,
+      "loss": 0.6044,
+      "step": 1542
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882018451091367e-06,
+      "loss": 0.6196,
+      "step": 1543
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.870095296175547e-06,
+      "loss": 0.6439,
+      "step": 1544
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.858186084857497e-06,
+      "loss": 0.6195,
+      "step": 1545
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.84629084422691e-06,
+      "loss": 0.6068,
+      "step": 1546
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.834409601341703e-06,
+      "loss": 0.6093,
+      "step": 1547
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822542383227953e-06,
+      "loss": 0.5937,
+      "step": 1548
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8106892168798305e-06,
+      "loss": 0.6464,
+      "step": 1549
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7988501292595546e-06,
+      "loss": 0.5303,
+      "step": 1550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.787025147297302e-06,
+      "loss": 0.566,
+      "step": 1551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7752142978911786e-06,
+      "loss": 0.6363,
+      "step": 1552
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7634176079071383e-06,
+      "loss": 0.5824,
+      "step": 1553
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7516351041789244e-06,
+      "loss": 0.572,
+      "step": 1554
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.739866813508013e-06,
+      "loss": 0.6178,
+      "step": 1555
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.728112762663548e-06,
+      "loss": 0.5124,
+      "step": 1556
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7163729783822835e-06,
+      "loss": 0.5652,
+      "step": 1557
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.704647487368528e-06,
+      "loss": 0.5709,
+      "step": 1558
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6929363162940655e-06,
+      "loss": 0.6014,
+      "step": 1559
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6812394917981166e-06,
+      "loss": 0.6475,
+      "step": 1560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6695570404872604e-06,
+      "loss": 0.5683,
+      "step": 1561
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.657888988935383e-06,
+      "loss": 0.5726,
+      "step": 1562
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.646235363683629e-06,
+      "loss": 0.5754,
+      "step": 1563
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6345961912403004e-06,
+      "loss": 0.5957,
+      "step": 1564
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.622971498080852e-06,
+      "loss": 0.6541,
+      "step": 1565
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6113613106477853e-06,
+      "loss": 0.6189,
+      "step": 1566
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.599765655350612e-06,
+      "loss": 0.6087,
+      "step": 1567
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.588184558565787e-06,
+      "loss": 0.5759,
+      "step": 1568
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5766180466366476e-06,
+      "loss": 0.5849,
+      "step": 1569
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.565066145873355e-06,
+      "loss": 0.5829,
+      "step": 1570
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5535288825528435e-06,
+      "loss": 0.6441,
+      "step": 1571
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.542006282918733e-06,
+      "loss": 0.6552,
+      "step": 1572
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.53049837318131e-06,
+      "loss": 0.6291,
+      "step": 1573
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5190051795174308e-06,
+      "loss": 0.6487,
+      "step": 1574
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5075267280704793e-06,
+      "loss": 0.5332,
+      "step": 1575
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4960630449503198e-06,
+      "loss": 0.634,
+      "step": 1576
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.484614156233196e-06,
+      "loss": 0.6352,
+      "step": 1577
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.473180087961727e-06,
+      "loss": 0.6311,
+      "step": 1578
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4617608661448043e-06,
+      "loss": 0.6034,
+      "step": 1579
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4503565167575505e-06,
+      "loss": 0.5944,
+      "step": 1580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.438967065741262e-06,
+      "loss": 0.5969,
+      "step": 1581
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4275925390033426e-06,
+      "loss": 0.5894,
+      "step": 1582
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4162329624172454e-06,
+      "loss": 0.65,
+      "step": 1583
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4048883618224315e-06,
+      "loss": 0.5009,
+      "step": 1584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3935587630242705e-06,
+      "loss": 0.5516,
+      "step": 1585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3822441917940353e-06,
+      "loss": 0.5567,
+      "step": 1586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3709446738687966e-06,
+      "loss": 0.6938,
+      "step": 1587
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359660234951393e-06,
+      "loss": 0.5418,
+      "step": 1588
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.348390900710358e-06,
+      "loss": 0.6426,
+      "step": 1589
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3371366967798657e-06,
+      "loss": 0.648,
+      "step": 1590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.325897648759684e-06,
+      "loss": 0.6037,
+      "step": 1591
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3146737822150945e-06,
+      "loss": 0.6202,
+      "step": 1592
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.303465122676851e-06,
+      "loss": 0.6421,
+      "step": 1593
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2922716956411105e-06,
+      "loss": 0.6042,
+      "step": 1594
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2810935265693957e-06,
+      "loss": 0.6287,
+      "step": 1595
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.269930640888499e-06,
+      "loss": 0.6013,
+      "step": 1596
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2587830639904726e-06,
+      "loss": 0.546,
+      "step": 1597
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.247650821232523e-06,
+      "loss": 0.5635,
+      "step": 1598
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2365339379369963e-06,
+      "loss": 0.6191,
+      "step": 1599
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2254324393912895e-06,
+      "loss": 0.5882,
+      "step": 1600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.214346350847807e-06,
+      "loss": 0.5522,
+      "step": 1601
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2032756975238976e-06,
+      "loss": 0.5134,
+      "step": 1602
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1922205046018008e-06,
+      "loss": 0.5926,
+      "step": 1603
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1811807972285967e-06,
+      "loss": 0.6242,
+      "step": 1604
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.17015660051613e-06,
+      "loss": 0.6733,
+      "step": 1605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1591479395409672e-06,
+      "loss": 0.5285,
+      "step": 1606
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.148154839344334e-06,
+      "loss": 0.5384,
+      "step": 1607
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1371773249320714e-06,
+      "loss": 0.6569,
+      "step": 1608
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1262154212745456e-06,
+      "loss": 0.6274,
+      "step": 1609
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1152691533066393e-06,
+      "loss": 0.6248,
+      "step": 1610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.104338545927643e-06,
+      "loss": 0.6544,
+      "step": 1611
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.093423624001248e-06,
+      "loss": 0.6987,
+      "step": 1612
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.082524412355452e-06,
+      "loss": 0.6701,
+      "step": 1613
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0716409357825194e-06,
+      "loss": 0.575,
+      "step": 1614
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0607732190389228e-06,
+      "loss": 0.6877,
+      "step": 1615
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0499212868452878e-06,
+      "loss": 0.6576,
+      "step": 1616
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0390851638863293e-06,
+      "loss": 0.5698,
+      "step": 1617
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0282648748108135e-06,
+      "loss": 0.6381,
+      "step": 1618
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.017460444231477e-06,
+      "loss": 0.6558,
+      "step": 1619
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0066718967249853e-06,
+      "loss": 0.581,
+      "step": 1620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.995899256831888e-06,
+      "loss": 0.637,
+      "step": 1621
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9851425490565256e-06,
+      "loss": 0.6435,
+      "step": 1622
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.974401797867025e-06,
+      "loss": 0.6494,
+      "step": 1623
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963677027695192e-06,
+      "loss": 0.6199,
+      "step": 1624
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.952968262936502e-06,
+      "loss": 0.6891,
+      "step": 1625
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9422755279500093e-06,
+      "loss": 0.6168,
+      "step": 1626
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.931598847058311e-06,
+      "loss": 0.6136,
+      "step": 1627
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9209382445474853e-06,
+      "loss": 0.6383,
+      "step": 1628
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9102937446670364e-06,
+      "loss": 0.6282,
+      "step": 1629
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8996653716298374e-06,
+      "loss": 0.548,
+      "step": 1630
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889053149612088e-06,
+      "loss": 0.6394,
+      "step": 1631
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.878457102753238e-06,
+      "loss": 0.5881,
+      "step": 1632
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.867877255155951e-06,
+      "loss": 0.5505,
+      "step": 1633
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8573136308860374e-06,
+      "loss": 0.5873,
+      "step": 1634
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.846766253972405e-06,
+      "loss": 0.6015,
+      "step": 1635
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.836235148407016e-06,
+      "loss": 0.5862,
+      "step": 1636
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.825720338144797e-06,
+      "loss": 0.5789,
+      "step": 1637
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.815221847103633e-06,
+      "loss": 0.6082,
+      "step": 1638
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8047396991642695e-06,
+      "loss": 0.6286,
+      "step": 1639
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.794273918170287e-06,
+      "loss": 0.6544,
+      "step": 1640
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7838245279280305e-06,
+      "loss": 0.6143,
+      "step": 1641
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7733915522065645e-06,
+      "loss": 0.624,
+      "step": 1642
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7629750147376122e-06,
+      "loss": 0.6589,
+      "step": 1643
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7525749392155122e-06,
+      "loss": 0.6181,
+      "step": 1644
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7421913492971506e-06,
+      "loss": 0.5982,
+      "step": 1645
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7318242686019158e-06,
+      "loss": 0.637,
+      "step": 1646
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7214737207116424e-06,
+      "loss": 0.538,
+      "step": 1647
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7111397291705566e-06,
+      "loss": 0.5164,
+      "step": 1648
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7008223174852355e-06,
+      "loss": 0.5496,
+      "step": 1649
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6905215091245205e-06,
+      "loss": 0.6428,
+      "step": 1650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6802373275195084e-06,
+      "loss": 0.6172,
+      "step": 1651
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6699697960634605e-06,
+      "loss": 0.5915,
+      "step": 1652
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6597189381117707e-06,
+      "loss": 0.571,
+      "step": 1653
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6494847769819022e-06,
+      "loss": 0.6209,
+      "step": 1654
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.639267335953343e-06,
+      "loss": 0.5402,
+      "step": 1655
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62906663826754e-06,
+      "loss": 0.5828,
+      "step": 1656
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.61888270712787e-06,
+      "loss": 0.6265,
+      "step": 1657
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.608715565699549e-06,
+      "loss": 0.5801,
+      "step": 1658
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5985652371096213e-06,
+      "loss": 0.5401,
+      "step": 1659
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5884317444468785e-06,
+      "loss": 0.5794,
+      "step": 1660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.578315110761812e-06,
+      "loss": 0.5339,
+      "step": 1661
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5682153590665803e-06,
+      "loss": 0.5677,
+      "step": 1662
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5581325123349155e-06,
+      "loss": 0.5647,
+      "step": 1663
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5480665935021208e-06,
+      "loss": 0.5936,
+      "step": 1664
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5380176254649804e-06,
+      "loss": 0.6572,
+      "step": 1665
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.527985631081723e-06,
+      "loss": 0.6287,
+      "step": 1666
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5179706331719654e-06,
+      "loss": 0.5791,
+      "step": 1667
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.507972654516676e-06,
+      "loss": 0.6003,
+      "step": 1668
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.497991717858086e-06,
+      "loss": 0.6339,
+      "step": 1669
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4880278458996877e-06,
+      "loss": 0.6081,
+      "step": 1670
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478081061306131e-06,
+      "loss": 0.5919,
+      "step": 1671
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4681513867032203e-06,
+      "loss": 0.5359,
+      "step": 1672
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.458238844677827e-06,
+      "loss": 0.6063,
+      "step": 1673
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4483434577778544e-06,
+      "loss": 0.6447,
+      "step": 1674
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.438465248512183e-06,
+      "loss": 0.6405,
+      "step": 1675
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.428604239350617e-06,
+      "loss": 0.6283,
+      "step": 1676
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.418760452723846e-06,
+      "loss": 0.6268,
+      "step": 1677
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.408933911023371e-06,
+      "loss": 0.611,
+      "step": 1678
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3991246366014743e-06,
+      "loss": 0.6231,
+      "step": 1679
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3893326517711546e-06,
+      "loss": 0.6179,
+      "step": 1680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3795579788060928e-06,
+      "loss": 0.5193,
+      "step": 1681
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.369800639940574e-06,
+      "loss": 0.5737,
+      "step": 1682
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.360060657369474e-06,
+      "loss": 0.7327,
+      "step": 1683
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3503380532481667e-06,
+      "loss": 0.6448,
+      "step": 1684
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.340632849692515e-06,
+      "loss": 0.5915,
+      "step": 1685
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.330945068778789e-06,
+      "loss": 0.6301,
+      "step": 1686
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3212747325436346e-06,
+      "loss": 0.616,
+      "step": 1687
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3116218629840104e-06,
+      "loss": 0.5693,
+      "step": 1688
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3019864820571447e-06,
+      "loss": 0.6084,
+      "step": 1689
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2923686116804923e-06,
+      "loss": 0.659,
+      "step": 1690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2827682737316704e-06,
+      "loss": 0.6292,
+      "step": 1691
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2731854900484153e-06,
+      "loss": 0.5643,
+      "step": 1692
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2636202824285312e-06,
+      "loss": 0.5578,
+      "step": 1693
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2540726726298547e-06,
+      "loss": 0.6557,
+      "step": 1694
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2445426823701723e-06,
+      "loss": 0.5901,
+      "step": 1695
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2350303333272127e-06,
+      "loss": 0.5043,
+      "step": 1696
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.225535647138557e-06,
+      "loss": 0.5864,
+      "step": 1697
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2160586454016262e-06,
+      "loss": 0.6668,
+      "step": 1698
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2065993496736036e-06,
+      "loss": 0.6162,
+      "step": 1699
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1971577814714017e-06,
+      "loss": 0.6151,
+      "step": 1700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1877339622716066e-06,
+      "loss": 0.6437,
+      "step": 1701
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.178327913510431e-06,
+      "loss": 0.6227,
+      "step": 1702
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.168939656583664e-06,
+      "loss": 0.6604,
+      "step": 1703
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1595692128466308e-06,
+      "loss": 0.5771,
+      "step": 1704
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1502166036141293e-06,
+      "loss": 0.5902,
+      "step": 1705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.14088185016039e-06,
+      "loss": 0.5705,
+      "step": 1706
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.131564973719039e-06,
+      "loss": 0.64,
+      "step": 1707
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.122265995483016e-06,
+      "loss": 0.5684,
+      "step": 1708
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.112984936604573e-06,
+      "loss": 0.5863,
+      "step": 1709
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1037218181951767e-06,
+      "loss": 0.6644,
+      "step": 1710
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.094476661325504e-06,
+      "loss": 0.6989,
+      "step": 1711
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085249487025367e-06,
+      "loss": 0.5805,
+      "step": 1712
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.076040316283673e-06,
+      "loss": 0.5917,
+      "step": 1713
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0668491700483784e-06,
+      "loss": 0.672,
+      "step": 1714
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.057676069226436e-06,
+      "loss": 0.5865,
+      "step": 1715
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0485210346837557e-06,
+      "loss": 0.7023,
+      "step": 1716
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0393840872451532e-06,
+      "loss": 0.5885,
+      "step": 1717
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0302652476942965e-06,
+      "loss": 0.5966,
+      "step": 1718
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0211645367736667e-06,
+      "loss": 0.6051,
+      "step": 1719
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0120819751845068e-06,
+      "loss": 0.557,
+      "step": 1720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.003017583586776e-06,
+      "loss": 0.658,
+      "step": 1721
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9939713825991103e-06,
+      "loss": 0.6416,
+      "step": 1722
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9849433927987492e-06,
+      "loss": 0.6232,
+      "step": 1723
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9759336347215287e-06,
+      "loss": 0.599,
+      "step": 1724
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9669421288617994e-06,
+      "loss": 0.5516,
+      "step": 1725
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9579688956724e-06,
+      "loss": 0.6748,
+      "step": 1726
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9490139555646015e-06,
+      "loss": 0.6145,
+      "step": 1727
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9400773289080644e-06,
+      "loss": 0.5702,
+      "step": 1728
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.931159036030793e-06,
+      "loss": 0.5839,
+      "step": 1729
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.92225909721909e-06,
+      "loss": 0.6089,
+      "step": 1730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.913377532717505e-06,
+      "loss": 0.583,
+      "step": 1731
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9045143627287932e-06,
+      "loss": 0.5174,
+      "step": 1732
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.895669607413867e-06,
+      "loss": 0.6501,
+      "step": 1733
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8868432868917497e-06,
+      "loss": 0.5511,
+      "step": 1734
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8780354212395423e-06,
+      "loss": 0.631,
+      "step": 1735
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8692460304923476e-06,
+      "loss": 0.6182,
+      "step": 1736
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8604751346432625e-06,
+      "loss": 0.6095,
+      "step": 1737
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8517227536433012e-06,
+      "loss": 0.6416,
+      "step": 1738
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8429889074013696e-06,
+      "loss": 0.5958,
+      "step": 1739
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8342736157842056e-06,
+      "loss": 0.5648,
+      "step": 1740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8255768986163558e-06,
+      "loss": 0.6107,
+      "step": 1741
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8168987756800938e-06,
+      "loss": 0.6224,
+      "step": 1742
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.808239266715419e-06,
+      "loss": 0.6145,
+      "step": 1743
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7995983914199722e-06,
+      "loss": 0.6027,
+      "step": 1744
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7909761694490235e-06,
+      "loss": 0.554,
+      "step": 1745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7823726204154013e-06,
+      "loss": 0.5545,
+      "step": 1746
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.773787763889463e-06,
+      "loss": 0.5029,
+      "step": 1747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7652216193990523e-06,
+      "loss": 0.6296,
+      "step": 1748
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.756674206429434e-06,
+      "loss": 0.6511,
+      "step": 1749
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.748145544423282e-06,
+      "loss": 0.615,
+      "step": 1750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7396356527806059e-06,
+      "loss": 0.5439,
+      "step": 1751
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.731144550858722e-06,
+      "loss": 0.5256,
+      "step": 1752
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7226722579722065e-06,
+      "loss": 0.6231,
+      "step": 1753
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7142187933928544e-06,
+      "loss": 0.5822,
+      "step": 1754
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.705784176349622e-06,
+      "loss": 0.625,
+      "step": 1755
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6973684260286071e-06,
+      "loss": 0.5854,
+      "step": 1756
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6889715615729762e-06,
+      "loss": 0.5954,
+      "step": 1757
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6805936020829528e-06,
+      "loss": 0.5608,
+      "step": 1758
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6722345666157448e-06,
+      "loss": 0.6612,
+      "step": 1759
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.66389447418552e-06,
+      "loss": 0.6153,
+      "step": 1760
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6555733437633559e-06,
+      "loss": 0.5863,
+      "step": 1761
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6472711942771934e-06,
+      "loss": 0.6367,
+      "step": 1762
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6389880446118078e-06,
+      "loss": 0.6685,
+      "step": 1763
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.630723913608745e-06,
+      "loss": 0.6,
+      "step": 1764
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6224788200662934e-06,
+      "loss": 0.5336,
+      "step": 1765
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6142527827394373e-06,
+      "loss": 0.589,
+      "step": 1766
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.60604582033982e-06,
+      "loss": 0.6192,
+      "step": 1767
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5978579515356819e-06,
+      "loss": 0.5383,
+      "step": 1768
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5896891949518456e-06,
+      "loss": 0.6125,
+      "step": 1769
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5815395691696466e-06,
+      "loss": 0.5441,
+      "step": 1770
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5734090927269163e-06,
+      "loss": 0.637,
+      "step": 1771
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5652977841179196e-06,
+      "loss": 0.6245,
+      "step": 1772
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5572056617933228e-06,
+      "loss": 0.567,
+      "step": 1773
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5491327441601478e-06,
+      "loss": 0.6124,
+      "step": 1774
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.541079049581733e-06,
+      "loss": 0.582,
+      "step": 1775
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.533044596377693e-06,
+      "loss": 0.5511,
+      "step": 1776
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.525029402823872e-06,
+      "loss": 0.5777,
+      "step": 1777
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.517033487152304e-06,
+      "loss": 0.61,
+      "step": 1778
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5090568675511696e-06,
+      "loss": 0.5619,
+      "step": 1779
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5010995621647672e-06,
+      "loss": 0.5499,
+      "step": 1780
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4931615890934447e-06,
+      "loss": 0.6256,
+      "step": 1781
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.485242966393593e-06,
+      "loss": 0.5546,
+      "step": 1782
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4773437120775713e-06,
+      "loss": 0.5737,
+      "step": 1783
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4694638441136955e-06,
+      "loss": 0.5849,
+      "step": 1784
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.461603380426173e-06,
+      "loss": 0.5586,
+      "step": 1785
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4537623388950795e-06,
+      "loss": 0.5464,
+      "step": 1786
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4459407373563072e-06,
+      "loss": 0.6119,
+      "step": 1787
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4381385936015302e-06,
+      "loss": 0.5728,
+      "step": 1788
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4303559253781595e-06,
+      "loss": 0.5854,
+      "step": 1789
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4225927503893145e-06,
+      "loss": 0.61,
+      "step": 1790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4148490862937624e-06,
+      "loss": 0.5226,
+      "step": 1791
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4071249507058926e-06,
+      "loss": 0.6071,
+      "step": 1792
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3994203611956826e-06,
+      "loss": 0.5532,
+      "step": 1793
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.391735335288632e-06,
+      "loss": 0.5519,
+      "step": 1794
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3840698904657547e-06,
+      "loss": 0.542,
+      "step": 1795
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3764240441635113e-06,
+      "loss": 0.6386,
+      "step": 1796
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3687978137737933e-06,
+      "loss": 0.5352,
+      "step": 1797
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3611912166438656e-06,
+      "loss": 0.5938,
+      "step": 1798
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.353604270076334e-06,
+      "loss": 0.604,
+      "step": 1799
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3460369913291072e-06,
+      "loss": 0.5674,
+      "step": 1800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3384893976153546e-06,
+      "loss": 0.5874,
+      "step": 1801
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3309615061034676e-06,
+      "loss": 0.5692,
+      "step": 1802
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3234533339170253e-06,
+      "loss": 0.594,
+      "step": 1803
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3159648981347485e-06,
+      "loss": 0.5667,
+      "step": 1804
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3084962157904623e-06,
+      "loss": 0.5883,
+      "step": 1805
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3010473038730632e-06,
+      "loss": 0.6043,
+      "step": 1806
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2936181793264691e-06,
+      "loss": 0.5586,
+      "step": 1807
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2862088590496013e-06,
+      "loss": 0.5683,
+      "step": 1808
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.278819359896315e-06,
+      "loss": 0.594,
+      "step": 1809
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2714496986753932e-06,
+      "loss": 0.6421,
+      "step": 1810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2640998921504865e-06,
+      "loss": 0.5996,
+      "step": 1811
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.256769957040085e-06,
+      "loss": 0.5551,
+      "step": 1812
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2494599100174753e-06,
+      "loss": 0.6013,
+      "step": 1813
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2421697677107081e-06,
+      "loss": 0.568,
+      "step": 1814
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2348995467025527e-06,
+      "loss": 0.5827,
+      "step": 1815
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2276492635304694e-06,
+      "loss": 0.6073,
+      "step": 1816
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2204189346865614e-06,
+      "loss": 0.5888,
+      "step": 1817
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.213208576617545e-06,
+      "loss": 0.5942,
+      "step": 1818
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2060182057247083e-06,
+      "loss": 0.6544,
+      "step": 1819
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1988478383638724e-06,
+      "loss": 0.6649,
+      "step": 1820
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1916974908453649e-06,
+      "loss": 0.6244,
+      "step": 1821
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1845671794339597e-06,
+      "loss": 0.6132,
+      "step": 1822
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.177456920348873e-06,
+      "loss": 0.6241,
+      "step": 1823
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1703667297636933e-06,
+      "loss": 0.5846,
+      "step": 1824
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1632966238063693e-06,
+      "loss": 0.5908,
+      "step": 1825
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1562466185591558e-06,
+      "loss": 0.6116,
+      "step": 1826
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1492167300585944e-06,
+      "loss": 0.6511,
+      "step": 1827
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1422069742954546e-06,
+      "loss": 0.5717,
+      "step": 1828
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.135217367214726e-06,
+      "loss": 0.5655,
+      "step": 1829
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1282479247155498e-06,
+      "loss": 0.6726,
+      "step": 1830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1212986626512112e-06,
+      "loss": 0.571,
+      "step": 1831
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.114369596829088e-06,
+      "loss": 0.6497,
+      "step": 1832
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1074607430106154e-06,
+      "loss": 0.5856,
+      "step": 1833
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100572116911255e-06,
+      "loss": 0.5736,
+      "step": 1834
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0937037342004542e-06,
+      "loss": 0.6158,
+      "step": 1835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0868556105016193e-06,
+      "loss": 0.5219,
+      "step": 1836
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0800277613920672e-06,
+      "loss": 0.6061,
+      "step": 1837
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0732202024029992e-06,
+      "loss": 0.5343,
+      "step": 1838
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0664329490194602e-06,
+      "loss": 0.5639,
+      "step": 1839
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0596660166803142e-06,
+      "loss": 0.5989,
+      "step": 1840
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0529194207781879e-06,
+      "loss": 0.6231,
+      "step": 1841
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0461931766594669e-06,
+      "loss": 0.5066,
+      "step": 1842
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.039487299624221e-06,
+      "loss": 0.619,
+      "step": 1843
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0328018049262134e-06,
+      "loss": 0.6217,
+      "step": 1844
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0261367077728279e-06,
+      "loss": 0.5551,
+      "step": 1845
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0194920233250594e-06,
+      "loss": 0.5696,
+      "step": 1846
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.012867766697464e-06,
+      "loss": 0.6388,
+      "step": 1847
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0062639529581341e-06,
+      "loss": 0.6662,
+      "step": 1848
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996805971286638e-07,
+      "loss": 0.6136,
+      "step": 1849
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.931177141841087e-07,
+      "loss": 0.6511,
+      "step": 1850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.865753190529537e-07,
+      "loss": 0.608,
+      "step": 1851
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.800534266170792e-07,
+      "loss": 0.6362,
+      "step": 1852
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.735520517117403e-07,
+      "loss": 0.6109,
+      "step": 1853
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.67071209125502e-07,
+      "loss": 0.5991,
+      "step": 1854
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.606109136002428e-07,
+      "loss": 0.5994,
+      "step": 1855
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.541711798310883e-07,
+      "loss": 0.5873,
+      "step": 1856
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.47752022466405e-07,
+      "loss": 0.645,
+      "step": 1857
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.413534561077464e-07,
+      "loss": 0.6277,
+      "step": 1858
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.349754953098333e-07,
+      "loss": 0.5621,
+      "step": 1859
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.286181545805096e-07,
+      "loss": 0.6431,
+      "step": 1860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.222814483807185e-07,
+      "loss": 0.5608,
+      "step": 1861
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.159653911244693e-07,
+      "loss": 0.5447,
+      "step": 1862
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.096699971787959e-07,
+      "loss": 0.6147,
+      "step": 1863
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.033952808637314e-07,
+      "loss": 0.6246,
+      "step": 1864
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.971412564522719e-07,
+      "loss": 0.5912,
+      "step": 1865
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.909079381703556e-07,
+      "loss": 0.5556,
+      "step": 1866
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.84695340196805e-07,
+      "loss": 0.6744,
+      "step": 1867
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.785034766633271e-07,
+      "loss": 0.5847,
+      "step": 1868
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.723323616544477e-07,
+      "loss": 0.6199,
+      "step": 1869
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.661820092075124e-07,
+      "loss": 0.5709,
+      "step": 1870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.60052433312627e-07,
+      "loss": 0.5375,
+      "step": 1871
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.539436479126429e-07,
+      "loss": 0.5739,
+      "step": 1872
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.478556669031179e-07,
+      "loss": 0.6239,
+      "step": 1873
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.417885041322871e-07,
+      "loss": 0.6661,
+      "step": 1874
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.357421734010274e-07,
+      "loss": 0.5943,
+      "step": 1875
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.29716688462836e-07,
+      "loss": 0.5838,
+      "step": 1876
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.237120630237871e-07,
+      "loss": 0.613,
+      "step": 1877
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.17728310742506e-07,
+      "loss": 0.62,
+      "step": 1878
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.117654452301449e-07,
+      "loss": 0.5753,
+      "step": 1879
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.058234800503328e-07,
+      "loss": 0.5743,
+      "step": 1880
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.999024287191726e-07,
+      "loss": 0.5618,
+      "step": 1881
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.940023047051782e-07,
+      "loss": 0.6197,
+      "step": 1882
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.881231214292751e-07,
+      "loss": 0.6447,
+      "step": 1883
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.822648922647458e-07,
+      "loss": 0.6421,
+      "step": 1884
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.764276305372143e-07,
+      "loss": 0.5481,
+      "step": 1885
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.706113495246059e-07,
+      "loss": 0.6203,
+      "step": 1886
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.648160624571244e-07,
+      "loss": 0.5512,
+      "step": 1887
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.59041782517217e-07,
+      "loss": 0.5436,
+      "step": 1888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.532885228395492e-07,
+      "loss": 0.6077,
+      "step": 1889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.475562965109695e-07,
+      "loss": 0.5781,
+      "step": 1890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.418451165704832e-07,
+      "loss": 0.6087,
+      "step": 1891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.361549960092206e-07,
+      "loss": 0.6487,
+      "step": 1892
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.304859477704096e-07,
+      "loss": 0.5781,
+      "step": 1893
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.248379847493469e-07,
+      "loss": 0.5511,
+      "step": 1894
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.192111197933616e-07,
+      "loss": 0.6226,
+      "step": 1895
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.136053657017983e-07,
+      "loss": 0.6152,
+      "step": 1896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08020735225976e-07,
+      "loss": 0.6032,
+      "step": 1897
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.024572410691655e-07,
+      "loss": 0.6595,
+      "step": 1898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.969148958865579e-07,
+      "loss": 0.6251,
+      "step": 1899
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.913937122852443e-07,
+      "loss": 0.5305,
+      "step": 1900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.858937028241675e-07,
+      "loss": 0.622,
+      "step": 1901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.804148800141164e-07,
+      "loss": 0.6042,
+      "step": 1902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.749572563176821e-07,
+      "loss": 0.6033,
+      "step": 1903
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.695208441492352e-07,
+      "loss": 0.5716,
+      "step": 1904
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.64105655874897e-07,
+      "loss": 0.5973,
+      "step": 1905
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.587117038125091e-07,
+      "loss": 0.5147,
+      "step": 1906
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.533390002316142e-07,
+      "loss": 0.5567,
+      "step": 1907
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.4798755735341e-07,
+      "loss": 0.5631,
+      "step": 1908
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.426573873507458e-07,
+      "loss": 0.604,
+      "step": 1909
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.373485023480719e-07,
+      "loss": 0.6302,
+      "step": 1910
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.320609144214263e-07,
+      "loss": 0.6032,
+      "step": 1911
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.267946355984001e-07,
+      "loss": 0.6038,
+      "step": 1912
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.215496778581209e-07,
+      "loss": 0.6092,
+      "step": 1913
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.163260531312032e-07,
+      "loss": 0.6793,
+      "step": 1914
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.11123773299751e-07,
+      "loss": 0.5852,
+      "step": 1915
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.05942850197303e-07,
+      "loss": 0.6085,
+      "step": 1916
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.007832956088278e-07,
+      "loss": 0.6221,
+      "step": 1917
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.956451212706804e-07,
+      "loss": 0.5289,
+      "step": 1918
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.905283388705863e-07,
+      "loss": 0.5876,
+      "step": 1919
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.854329600476105e-07,
+      "loss": 0.5728,
+      "step": 1920
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.803589963921274e-07,
+      "loss": 0.6524,
+      "step": 1921
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.753064594458058e-07,
+      "loss": 0.6747,
+      "step": 1922
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.70275360701572e-07,
+      "loss": 0.5611,
+      "step": 1923
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.652657116035865e-07,
+      "loss": 0.5761,
+      "step": 1924
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.602775235472169e-07,
+      "loss": 0.6158,
+      "step": 1925
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.553108078790215e-07,
+      "loss": 0.5551,
+      "step": 1926
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.503655758967031e-07,
+      "loss": 0.6172,
+      "step": 1927
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.454418388491112e-07,
+      "loss": 0.5876,
+      "step": 1928
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.405396079361847e-07,
+      "loss": 0.6036,
+      "step": 1929
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.356588943089569e-07,
+      "loss": 0.6317,
+      "step": 1930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.307997090695083e-07,
+      "loss": 0.6098,
+      "step": 1931
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.259620632709517e-07,
+      "loss": 0.5634,
+      "step": 1932
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.211459679174036e-07,
+      "loss": 0.5962,
+      "step": 1933
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.163514339639586e-07,
+      "loss": 0.6149,
+      "step": 1934
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.115784723166695e-07,
+      "loss": 0.5573,
+      "step": 1935
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.068270938325182e-07,
+      "loss": 0.57,
+      "step": 1936
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.020973093193882e-07,
+      "loss": 0.5516,
+      "step": 1937
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.973891295360467e-07,
+      "loss": 0.5932,
+      "step": 1938
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.92702565192118e-07,
+      "loss": 0.6374,
+      "step": 1939
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.880376269480525e-07,
+      "loss": 0.5794,
+      "step": 1940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833943254151152e-07,
+      "loss": 0.5405,
+      "step": 1941
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.787726711553486e-07,
+      "loss": 0.5969,
+      "step": 1942
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.741726746815589e-07,
+      "loss": 0.585,
+      "step": 1943
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.69594346457285e-07,
+      "loss": 0.653,
+      "step": 1944
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6503769689677733e-07,
+      "loss": 0.453,
+      "step": 1945
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.60502736364975e-07,
+      "loss": 0.4873,
+      "step": 1946
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5598947517748183e-07,
+      "loss": 0.5712,
+      "step": 1947
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5149792360054236e-07,
+      "loss": 0.5348,
+      "step": 1948
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4702809185101835e-07,
+      "loss": 0.6187,
+      "step": 1949
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4257999009636567e-07,
+      "loss": 0.6246,
+      "step": 1950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3815362845460954e-07,
+      "loss": 0.5524,
+      "step": 1951
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3374901699432923e-07,
+      "loss": 0.5961,
+      "step": 1952
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.293661657346204e-07,
+      "loss": 0.6077,
+      "step": 1953
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2500508464509037e-07,
+      "loss": 0.6162,
+      "step": 1954
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.206657836458161e-07,
+      "loss": 0.6546,
+      "step": 1955
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.16348272607342e-07,
+      "loss": 0.5925,
+      "step": 1956
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.12052561350641e-07,
+      "loss": 0.6192,
+      "step": 1957
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0777865964710005e-07,
+      "loss": 0.5992,
+      "step": 1958
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.03526577218496e-07,
+      "loss": 0.5559,
+      "step": 1959
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.99296323736974e-07,
+      "loss": 0.54,
+      "step": 1960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.950879088250226e-07,
+      "loss": 0.626,
+      "step": 1961
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9090134205546195e-07,
+      "loss": 0.5884,
+      "step": 1962
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.867366329514055e-07,
+      "loss": 0.7014,
+      "step": 1963
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8259379098625205e-07,
+      "loss": 0.6048,
+      "step": 1964
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.784728255836623e-07,
+      "loss": 0.6457,
+      "step": 1965
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7437374611752674e-07,
+      "loss": 0.5924,
+      "step": 1966
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.702965619119614e-07,
+      "loss": 0.6375,
+      "step": 1967
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.662412822412687e-07,
+      "loss": 0.6007,
+      "step": 1968
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.622079163299319e-07,
+      "loss": 0.5684,
+      "step": 1969
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5819647335258555e-07,
+      "loss": 0.5981,
+      "step": 1970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.54206962433995e-07,
+      "loss": 0.5097,
+      "step": 1971
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.502393926490355e-07,
+      "loss": 0.5963,
+      "step": 1972
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.462937730226812e-07,
+      "loss": 0.5903,
+      "step": 1973
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.42370112529965e-07,
+      "loss": 0.6562,
+      "step": 1974
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3846842009598093e-07,
+      "loss": 0.6475,
+      "step": 1975
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3458870459584294e-07,
+      "loss": 0.6146,
+      "step": 1976
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3073097485468275e-07,
+      "loss": 0.6771,
+      "step": 1977
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.268952396476144e-07,
+      "loss": 0.5776,
+      "step": 1978
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2308150769972404e-07,
+      "loss": 0.5338,
+      "step": 1979
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1928978768605234e-07,
+      "loss": 0.5561,
+      "step": 1980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.15520088231559e-07,
+      "loss": 0.65,
+      "step": 1981
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.117724179111248e-07,
+      "loss": 0.5808,
+      "step": 1982
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0804678524951504e-07,
+      "loss": 0.6247,
+      "step": 1983
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.043431987213663e-07,
+      "loss": 0.5457,
+      "step": 1984
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.006616667511697e-07,
+      "loss": 0.589,
+      "step": 1985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.970021977132487e-07,
+      "loss": 0.6079,
+      "step": 1986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9336479993173795e-07,
+      "loss": 0.6447,
+      "step": 1987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8974948168057237e-07,
+      "loss": 0.5963,
+      "step": 1988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.861562511834559e-07,
+      "loss": 0.6023,
+      "step": 1989
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.825851166138549e-07,
+      "loss": 0.5722,
+      "step": 1990
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7903608609497365e-07,
+      "loss": 0.5964,
+      "step": 1991
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7550916769973456e-07,
+      "loss": 0.6337,
+      "step": 1992
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.720043694507657e-07,
+      "loss": 0.5648,
+      "step": 1993
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.685216993203732e-07,
+      "loss": 0.6046,
+      "step": 1994
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.650611652305346e-07,
+      "loss": 0.683,
+      "step": 1995
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6162277505287216e-07,
+      "loss": 0.6424,
+      "step": 1996
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.582065366086362e-07,
+      "loss": 0.5782,
+      "step": 1997
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5481245766869165e-07,
+      "loss": 0.6178,
+      "step": 1998
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.514405459534974e-07,
+      "loss": 0.6102,
+      "step": 1999
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4809080913308583e-07,
+      "loss": 0.6055,
+      "step": 2000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.447632548270529e-07,
+      "loss": 0.5884,
+      "step": 2001
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.414578906045317e-07,
+      "loss": 0.5609,
+      "step": 2002
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3817472398418362e-07,
+      "loss": 0.5514,
+      "step": 2003
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3491376243417685e-07,
+      "loss": 0.632,
+      "step": 2004
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3167501337216792e-07,
+      "loss": 0.5757,
+      "step": 2005
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2845848416528816e-07,
+      "loss": 0.6259,
+      "step": 2006
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2526418213012602e-07,
+      "loss": 0.6463,
+      "step": 2007
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.220921145327115e-07,
+      "loss": 0.6112,
+      "step": 2008
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.189422885884962e-07,
+      "loss": 0.6257,
+      "step": 2009
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.158147114623399e-07,
+      "loss": 0.5695,
+      "step": 2010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127093902684907e-07,
+      "loss": 0.5345,
+      "step": 2011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0962633207057827e-07,
+      "loss": 0.5905,
+      "step": 2012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0656554388158168e-07,
+      "loss": 0.5831,
+      "step": 2013
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0352703266383166e-07,
+      "loss": 0.6712,
+      "step": 2014
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0051080532897838e-07,
+      "loss": 0.6089,
+      "step": 2015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9751686873798913e-07,
+      "loss": 0.6321,
+      "step": 2016
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.945452297011252e-07,
+      "loss": 0.6186,
+      "step": 2017
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9159589497792397e-07,
+      "loss": 0.5667,
+      "step": 2018
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8866887127719335e-07,
+      "loss": 0.6278,
+      "step": 2019
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8576416525698637e-07,
+      "loss": 0.6603,
+      "step": 2020
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.828817835245933e-07,
+      "loss": 0.6563,
+      "step": 2021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.800217326365228e-07,
+      "loss": 0.5662,
+      "step": 2022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7718401909848637e-07,
+      "loss": 0.5749,
+      "step": 2023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7436864936538623e-07,
+      "loss": 0.6096,
+      "step": 2024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7157562984130182e-07,
+      "loss": 0.5871,
+      "step": 2025
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.688049668794678e-07,
+      "loss": 0.542,
+      "step": 2026
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.660566667822705e-07,
+      "loss": 0.5384,
+      "step": 2027
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.633307358012215e-07,
+      "loss": 0.6558,
+      "step": 2028
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.606271801369552e-07,
+      "loss": 0.5642,
+      "step": 2029
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5794600593920794e-07,
+      "loss": 0.5989,
+      "step": 2030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5528721930680336e-07,
+      "loss": 0.5474,
+      "step": 2031
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.526508262876414e-07,
+      "loss": 0.614,
+      "step": 2032
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5003683287868276e-07,
+      "loss": 0.6025,
+      "step": 2033
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4744524502594004e-07,
+      "loss": 0.6591,
+      "step": 2034
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4487606862445547e-07,
+      "loss": 0.6509,
+      "step": 2035
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4232930951829426e-07,
+      "loss": 0.6323,
+      "step": 2036
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3980497350052914e-07,
+      "loss": 0.5563,
+      "step": 2037
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.37303066313228e-07,
+      "loss": 0.576,
+      "step": 2038
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3482359364743958e-07,
+      "loss": 0.6184,
+      "step": 2039
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3236656114318235e-07,
+      "loss": 0.6451,
+      "step": 2040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2993197438942672e-07,
+      "loss": 0.597,
+      "step": 2041
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.275198389240906e-07,
+      "loss": 0.5851,
+      "step": 2042
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2513016023402158e-07,
+      "loss": 0.5888,
+      "step": 2043
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2276294375498265e-07,
+      "loss": 0.5672,
+      "step": 2044
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.204181948716443e-07,
+      "loss": 0.596,
+      "step": 2045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1809591891757121e-07,
+      "loss": 0.5916,
+      "step": 2046
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1579612117520456e-07,
+      "loss": 0.5788,
+      "step": 2047
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1351880687586192e-07,
+      "loss": 0.5814,
+      "step": 2048
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1126398119971182e-07,
+      "loss": 0.5942,
+      "step": 2049
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0903164927577147e-07,
+      "loss": 0.6292,
+      "step": 2050
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0682181618189235e-07,
+      "loss": 0.5432,
+      "step": 2051
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0463448694474465e-07,
+      "loss": 0.5823,
+      "step": 2052
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0246966653981505e-07,
+      "loss": 0.6369,
+      "step": 2053
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0032735989138231e-07,
+      "loss": 0.5598,
+      "step": 2054
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.820757187252061e-08,
+      "loss": 0.5863,
+      "step": 2055
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.61103073050773e-08,
+      "loss": 0.7235,
+      "step": 2056
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.403557095966742e-08,
+      "loss": 0.6829,
+      "step": 2057
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.19833675556614e-08,
+      "loss": 0.6167,
+      "step": 2058
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.995370176117402e-08,
+      "loss": 0.6263,
+      "step": 2059
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.79465781930522e-08,
+      "loss": 0.5296,
+      "step": 2060
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.596200141687161e-08,
+      "loss": 0.5081,
+      "step": 2061
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.399997594691456e-08,
+      "loss": 0.6296,
+      "step": 2062
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.206050624617212e-08,
+      "loss": 0.569,
+      "step": 2063
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.014359672632421e-08,
+      "loss": 0.5367,
+      "step": 2064
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.824925174773513e-08,
+      "loss": 0.6022,
+      "step": 2065
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.637747561944131e-08,
+      "loss": 0.5909,
+      "step": 2066
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.452827259914252e-08,
+      "loss": 0.5282,
+      "step": 2067
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.270164689319181e-08,
+      "loss": 0.4961,
+      "step": 2068
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.089760265658663e-08,
+      "loss": 0.5329,
+      "step": 2069
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.911614399295663e-08,
+      "loss": 0.5838,
+      "step": 2070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.735727495455813e-08,
+      "loss": 0.6174,
+      "step": 2071
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.562099954226409e-08,
+      "loss": 0.598,
+      "step": 2072
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.390732170555191e-08,
+      "loss": 0.5295,
+      "step": 2073
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.221624534249681e-08,
+      "loss": 0.5787,
+      "step": 2074
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.054777429976621e-08,
+      "loss": 0.5858,
+      "step": 2075
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.890191237260201e-08,
+      "loss": 0.4701,
+      "step": 2076
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.727866330482168e-08,
+      "loss": 0.5512,
+      "step": 2077
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.567803078880385e-08,
+      "loss": 0.6166,
+      "step": 2078
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.410001846548385e-08,
+      "loss": 0.5578,
+      "step": 2079
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.254462992433817e-08,
+      "loss": 0.5572,
+      "step": 2080
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.101186870338781e-08,
+      "loss": 0.5613,
+      "step": 2081
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.950173828918048e-08,
+      "loss": 0.605,
+      "step": 2082
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.801424211678396e-08,
+      "loss": 0.6469,
+      "step": 2083
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.654938356978278e-08,
+      "loss": 0.65,
+      "step": 2084
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5107165980269315e-08,
+      "loss": 0.5996,
+      "step": 2085
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.36875926288316e-08,
+      "loss": 0.578,
+      "step": 2086
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2290666744552175e-08,
+      "loss": 0.575,
+      "step": 2087
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0916391504994826e-08,
+      "loss": 0.6433,
+      "step": 2088
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.95647700362034e-08,
+      "loss": 0.5881,
+      "step": 2089
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8235805412688567e-08,
+      "loss": 0.5424,
+      "step": 2090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.692950065742551e-08,
+      "loss": 0.5626,
+      "step": 2091
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.564585874184623e-08,
+      "loss": 0.6364,
+      "step": 2092
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.43848825858295e-08,
+      "loss": 0.566,
+      "step": 2093
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.314657505769758e-08,
+      "loss": 0.6096,
+      "step": 2094
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.193093897421173e-08,
+      "loss": 0.5292,
+      "step": 2095
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.0737977100557816e-08,
+      "loss": 0.6274,
+      "step": 2096
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9567692150347384e-08,
+      "loss": 0.6057,
+      "step": 2097
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8420086785611035e-08,
+      "loss": 0.6085,
+      "step": 2098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.729516361678619e-08,
+      "loss": 0.5818,
+      "step": 2099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6192925202720433e-08,
+      "loss": 0.6295,
+      "step": 2100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5113374050655947e-08,
+      "loss": 0.5245,
+      "step": 2101
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4056512616231764e-08,
+      "loss": 0.5504,
+      "step": 2102
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3022343303474858e-08,
+      "loss": 0.6062,
+      "step": 2103
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.201086846479239e-08,
+      "loss": 0.6174,
+      "step": 2104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1022090400970584e-08,
+      "loss": 0.6268,
+      "step": 2105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0056011361166972e-08,
+      "loss": 0.5806,
+      "step": 2106
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9112633542908156e-08,
+      "loss": 0.5393,
+      "step": 2107
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8191959092078714e-08,
+      "loss": 0.6258,
+      "step": 2108
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.729399010292343e-08,
+      "loss": 0.5988,
+      "step": 2109
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6418728618037285e-08,
+      "loss": 0.6076,
+      "step": 2110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5566176628365458e-08,
+      "loss": 0.5816,
+      "step": 2111
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4736336073193358e-08,
+      "loss": 0.5648,
+      "step": 2112
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3929208840146591e-08,
+      "loss": 0.6103,
+      "step": 2113
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.314479676518432e-08,
+      "loss": 0.5627,
+      "step": 2114
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2383101632597039e-08,
+      "loss": 0.6406,
+      "step": 2115
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.164412517499991e-08,
+      "loss": 0.6097,
+      "step": 2116
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0927869073332764e-08,
+      "loss": 0.5697,
+      "step": 2117
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0234334956849002e-08,
+      "loss": 0.6413,
+      "step": 2118
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.563524403122249e-09,
+      "loss": 0.5769,
+      "step": 2119
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.915438938034148e-09,
+      "loss": 0.5904,
+      "step": 2120
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.290080035774361e-09,
+      "loss": 0.5537,
+      "step": 2121
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.68744911883723e-09,
+      "loss": 0.5277,
+      "step": 2122
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.107547558017347e-09,
+      "loss": 0.5385,
+      "step": 2123
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.550376672407321e-09,
+      "loss": 0.5094,
+      "step": 2124
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.015937729396681e-09,
+      "loss": 0.5878,
+      "step": 2125
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.5042319446629855e-09,
+      "loss": 0.6393,
+      "step": 2126
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.0152604821773755e-09,
+      "loss": 0.6481,
+      "step": 2127
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5490244541934734e-09,
+      "loss": 0.5553,
+      "step": 2128
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.105524921252935e-09,
+      "loss": 0.5588,
+      "step": 2129
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.684762892176563e-09,
+      "loss": 0.5933,
+      "step": 2130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2867393240654244e-09,
+      "loss": 0.6107,
+      "step": 2131
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9114551222964027e-09,
+      "loss": 0.6566,
+      "step": 2132
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5589111405233124e-09,
+      "loss": 0.653,
+      "step": 2133
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.229108180672457e-09,
+      "loss": 0.644,
+      "step": 2134
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9220469929415176e-09,
+      "loss": 0.5518,
+      "step": 2135
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6377282757973344e-09,
+      "loss": 0.5531,
+      "step": 2136
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3761526759747956e-09,
+      "loss": 0.5921,
+      "step": 2137
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1373207884768366e-09,
+      "loss": 0.5876,
+      "step": 2138
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.212331565711108e-10,
+      "loss": 0.6151,
+      "step": 2139
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.278902717866576e-10,
+      "loss": 0.5587,
+      "step": 2140
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.572925739183443e-10,
+      "loss": 0.5808,
+      "step": 2141
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.09440451023535e-10,
+      "loss": 0.5247,
+      "step": 2142
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.843342394176496e-10,
+      "loss": 0.5328,
+      "step": 2143
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.819742236774946e-10,
+      "loss": 0.5915,
+      "step": 2144
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0236063664126328e-10,
+      "loss": 0.5768,
+      "step": 2145
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5493659402984224e-11,
+      "loss": 0.5871,
+      "step": 2146
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1373421318072731e-11,
+      "loss": 0.597,
+      "step": 2147
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.5972,
+      "step": 2148
+    },
+    {
+      "epoch": 3.0,
+      "step": 2148,
+      "total_flos": 3.773769285055283e+17,
+      "train_loss": 0.8857765065530381,
+      "train_runtime": 39271.2468,
+      "train_samples_per_second": 7.009,
+      "train_steps_per_second": 0.055
+    }
+  ],
+  "max_steps": 2148,
+  "num_train_epochs": 3,
+  "total_flos": 3.773769285055283e+17,
+  "trial_name": null,
+  "trial_params": null
+}